Haemangiosarcoma in the Golden Retriever Lifetime Study: frequency, risk factors and post-diagnosis survival

Haemangiosarcoma in the Golden Retriever Lifetime Study: frequency, risk factors and post-diagnosis survival | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Haemangiosarcoma in the Golden Retriever Lifetime Study: frequency, risk factors and post-diagnosis survival Collette Taylor, Georgina J. Barry, Dan G. O’Neill, Alexandra Guillén, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7064010/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background The Golden Retriever Lifetime Study (GRLS) is a prospective, longitudinal cohort study launched in the United States in 2012 to investigate genetic, environmental, and lifestyle factors contributing to disease. As the cohort of dogs now approaches end of life, it presents unique opportunities to explore haemangiosarcoma—a highly aggressive malignancy commonly reported in Golden Retrievers and the most commonly diagnosed tumour in the GRLS cohort. This study aimed to calculate incidence, explore potential risk factors for diagnosis, and describe tumour-specific survival. Methods Dogs diagnosed with haemangiosarcoma before December 31st, 2024, were classified as cases and all remaining dogs as non-cases. Data from owner and veterinarian surveys, along with necropsy reports, were processed to derive variables on demographics, lifestyle, environmental exposures, comorbidities, and medications. Analyses of the non-demographic, potentially modifiable factors were considered exploratory given the limited prior evidence and potential for residual confounding. Risk factor analysis for haemangiosarcoma diagnosis used multivariable binary logistic regression modelling. Median survival times (MST) were calculated. Results Overall, 490 of 3,044 dogs (16.1%) were diagnosed with haemangiosarcoma, giving an incidence rate of 1.69 (95% CI: 1.54–1.85) cases per 100 dog-years. Dogs aged < 7 or ≥ 10 years had lower odds of diagnosis than those aged 7–9 years. Male sex, higher median adult bodyweight, gastrointestinal comorbidities and NSAID use were associated to higher odds of diagnosis, Previous cancer history and higher owner-reported activity levels were associated with reduced odds of diagnosis. MST from diagnosis across all body-locations was 0 days (95% CI: 0–0), but cutaneous haemangiosarcoma showed a significantly longer median survival (log-rank test p < 0.01) of 19 days (95% CI: 0–479). Conclusions This study provides a longitudinal epidemiological profile of haemangiosarcoma in Golden Retrievers. The relatively high incidence rate and poor outcomes observed highlight the ongoing clinical challenges of this disease. Risk factor modelling identified both established (age, sex, bodyweight) and novel associations (NSAID use, gastrointestinal comorbidities, prior cancer, activity level). While these novel associations are hypothesis-generating and warrant cautious interpretation, they provide direction for future research to validate findings and elucidate underlying mechanisms. Haemangiosarcoma Golden Retriever Epidemiology Cancer Risk factors diagnosis canine Figures Figure 1 Figure 2 Figure 3 Introduction The Golden Retriever Lifetime Study (GRLS) is a prospective, longitudinal, lifetime cohort study designed to allow for exploration of demographic, environmental and lifestyle risk factors and their associations with outcomes of interest such as neoplasia 1 , 2 . Evidence pyramids place cohort studies above designs like case–control studies or tumour-registry case series, because cohorts yield more reliable estimates of disease frequency and associated risk factors 3 , 4 . Haemangiosarcoma is a malignant neoplasm originating from vascular endothelial or bone marrow progenitor cells that has been most commonly reported in visceral organs (typically spleen, heart, liver) and cutaneous or subcutaneous tissues 5 , 6 . The estimated incidence in across a range of breeds varies, from 0.007 to 0.4% based on veterinary clinic and tumour registry populations, respectively 7 – 11 . A previous study on the GRLS population reported a high incidence of haemangiosarcoma (1.1 cases per 100 dog years) 12 . As the GRLS cohort is now nearing the end of their lives, an update on haemangiosarcoma incidence and exploration of factors potentially associated with its development in this population is warranted. The most consistently reported risk factors associated with canine haemangiosarcoma previously have been breed and age 8 . A general increased risk in cancer diagnoses for Golden Retrievers has been reported previously 7 , 13 , 14 and this has also been seen specifically for haemangiosarcoma 7 , 8 . Consistent with most other canine cancers, increasing age has been associated with an increased odds of a haemangiosarcoma diagnosis, although this hasn’t been examined specifically in Golden Retrievers 8 , 10 . For other demographic factors such as sex, neutering or bodyweight, associations with haemangiosarcoma diagnosis are less clear. The association between neutering and development of haemangiosarcoma is not fully understood. Increased odds of haemangiosarcoma in neutered dogs have been reported in some previous studies 7 , 8 . In Golden Retrievers specifically, neutering associations varied with one study finding no difference between neuter statuses 15 and one study reporting increased odds for neutered females (versus entire) but no effect of neutering on male risk 14 . Age at neutering and type of neutering (spaying versus castration) may also impact this association 14 . Age at neutering does not appear to have been explored specifically for haemangiosarcoma development in Golden Retrievers, however recommendations for general cancer development have indicated that age of neutering influences cancer development overall 16 . Sex has not been consistently associated with haemangiosarcoma diagnosis for any dog breeds. However, an increased risk for haemangiosarcoma in male dogs across populations with a range of dog breeds studied has been seen in two previous studies 7 , 10 . Whilst environmental exposures, health and lifestyle factors are commonly explored in human cancer epidemiology studies, these are only beginning to be studied in veterinary oncology 17 , 18 . A better understanding of these factors is important as these are modifiable risk factors during a dog’s life, unlike fixed factors such as breed, sex or age. The GRLS design allows for exploration of these modifiable factors. Specific to haemangiosarcoma, one canine study reported higher socioeconomic status and urban areas to be associated with increased odds of diagnosis 8 . In human oncology, there is evidence that a wide range of drugs can impact the risk of developing different cancers, with the strongest evidence base appearing to exist for non-steroidal anti-inflammatory (NSAID) drugs having a chemopreventive effect 19 – 21 . However, medication as a protective or promoting factor in cancer development in dogs is not well explored. Certain disorders such as chronic inflammatory conditions have been shown to affect the development and progression of certain cancers in humans and animals, but this has not been reported specifically for haemangiosarcoma in animals 21 – 23 . Recent work by Ronzani & Hooper (2025) has shown physical activity intensity and frequency to be important predictors in overall cancer development in the GRLS cohort, although due to the modelling type used the directionality of these associations is not clear 24 . Survival times for haemangiosarcoma cases in dogs have been previously reported as very short, particularly for visceral forms, with one primary-care study in the UK reporting a median survival time (MST) of 9 days 25 . And even though studies in specialist centres of the subset of dogs referred for extended care report longer MSTs (ranging from 48–172 days), the overall outcome remains poor in most cases 26 – 30 . The current cohort study aimed to investigate the frequency, risk factors, and post-diagnosis survival associated with haemangiosarcoma in a large group of Golden Retrievers followed from puppyhood in the GRLS cohort. The study hypothesised that: Older age and neutering status are associated with increased odds of haemangiosarcoma diagnosis. Dogs with visceral forms of haemangiosarcoma have a significantly shorter median survival time than cutaneous forms The specific objectives of the study were to (i) estimate the frequency of haemangiosarcoma diagnoses, (ii) identify demographic, lifestyle, environmental and clinical variables associated with haemangiosarcoma diagnosis, and (iii) determine post-diagnosis survival times for the overall population and for the different anatomical locations. Methods Study population and data preparation Between 2012 and 2015, 3,044 pedigree Golden Retrievers aged between six months and two years at the time of application were enrolled into the GRLS from across the contiguous United States 1 , 2 , 31 . Further details on enrolment requirements are described elsewhere 1 . Dogs were excluded from enrolment if they had a prior malignancy diagnosis or a chronic life-threatening condition 31 The entire cohort was included in the current study. A censor date of 31st December 2024 was used to classify each dog on whether they had developed haemangiosarcoma by this point. Dogs diagnosed with haemangiosarcoma by this date were assigned as cases. All remaining dogs that had either died or been withdrawn from the study without being diagnosed with haemangiosarcoma, or dogs that were still alive and in the cohort but had not developed haemangiosarcoma by 31st December 2024, were classified as non-cases. The 31st December 2024 date was also used as their endpoint for calculation of variables with an endpoint required. The haemangiosarcoma case definition required either a) a laboratory-confirmed diagnosis of haemangiosarcoma supported by histological analysis by a board-certified pathologist of samples obtained for diagnostic purposes or during post-mortem examination and b) a non-laboratory confirmed diagnosis based on the clinical opinion of the attending veterinarian that may or may not have been supported by diagnostic imaging, direct mass visualisation or through macroscopic post-mortem examination 2 , 31 . Location of mass was determined by either the Morris Animal Foundation GRLS operations team or study veterinarian using clear criteria, with adjudication by a staff veteterinarian 1 , 2 . Enrolment time was defined from date of enrolment into the study until the endpoint diagnosis for all dogs. Data used in this study were generated from anonymised annual owner and veterinarian questionnaires and necropsy reports 31 . A detailed explanation of the data captured for dogs in the GRLS cohort is explained more fully by Labadie and colleagues (2022). In brief, demographic, lifestyle, environmental exposures and medication usage were captured in annual questionnaires throughout each dog’s life 2 . Given that some dogs died or become unenrolled between questionnaires, the most recent study year questionnaire was matched to the final study year of enrolment for that dog. The years of the study were grouped as follows: ‘early life’ was defined as records from study years (SY) 0–2, ‘rest of life’ was SY 3 until end of life and ‘whole life’ was all records available from all study years. As the potential most important carcinogen exposure window/s in a dog’s life have not been determined, we created a further timepoint for variables where values were calculated for the five years prior to endpoint diagnosis (‘5y prior to endpoint’). The data transformation processes and the variables created are explained in Table 1 and with further detail in the appendices (Table A1). Table 1 Available data and transformation process for haemangiosarcoma diagnosis risk factor exploration. Table indicates what data was available and the transformation process to generate variables for exploration of factors associated with a haemangiosarcoma diagnosis in the Golden Retriever Lifetime Study cohort (n = 3044). SY = study years. Variable type Data transformation process/es Variables created Demographics Age Cases: age at first diagnosis with haemangiosarcoma. Non-cases: age at earliest of 31/12/24 or date of death/unenrolment. Age was categorised according to a recent haemangiosarcoma study by Barry and colleagues 8 , biologically relevant categories and into quartiles Age at final date, age at final date biological, age at final date quartiles Sex and neutering Age at neutering calculated if neutering date and birth date available. For dogs where neutering date was missing this variable was not calculated Sex, neuter, sex neuter, age at neutering Reproduction Reproductive history variables were calculated for dogs prior to neutering dates (if available) or endpoint date. For certain variables, two variations were made – one that excluded certain groups (entire, neutered, male or female dogs), depending on the variable and one that included all dogs and had these excluded categories retained Mismating ever, reason for neutering, heats before neutering, number of times bred from before neutering, Height, weight and body condition score (BCS) Veterinarian physical examination data (height, bodyweight, Purina 9-point Body Condition Score (BCS) 63 from the preceding 5 years from endpoint of haemangiosarcoma diagnosis or final record was either averaged across the 5 years or the most common (mode) BCS was reported. BCS was also classified into underweight (0–3), ideal (4–5) and overweight (6–9). The weight and height records of dogs whose contribution to the cohort finished before 18 months old were excluded from these variables. Heights and weights were converted from inches and pounds to centimetres and kilograms and outliers more than 1.5 times outside of the interquartile range were removed. Mean and median adult weight, height and BCS 5y prior, mode and classified BCS 5y prior Lifestyle Activity Owner annual questionnaire data on frequency, duration and intensity of walks and aerobic activities. Questionnaires were modified between study years (SY) 2 and 3 so not all information was captured across all years of study. Owner free-text responses about duration of activity were initially categorised then converted to a numeric category (out of 100) as below to allow for calculations. Activity level was captured consistently across all years as an ordinal variable 24 Early life mean activity duration– daily USDA advised, daily activity 1h, daily activity KC 2h Early, rest, whole and 5y prior to endpoint – mean, median, mode, minimum and maximum activity levels Duration free text was classified: = 120 min = 120 min = 100 Activity level: None, little, moderate, very active None = 0 Little = 1 Moderate = 2 Very active = 3 Early life exercise duration was averaged across study years and classified into United States Department for Agriculture (USDA) > 30 min/day recommendation for dogs and the UK Kennel Club (KC) > 2h/day activity recommendation for Golden Retrievers. Lifestyle Owner annual questionnaire response on dog’s main lifestyle were grouped into: companion/pet, competitive, breed/showing, working, miscellaneous. The mode lifestyle group across all study years and the 5 years prior to endpoint were also determined Main lifestyle category whole life and 5y prior to endpoint Home/environment exposures Socioeconomic status Clinic and owner zip codes linked to county-level Federal Information Processing Standards (FIPS) codes and 2019 county-level Multideprivation Index (MDI) values 64 MDI quintiles for veterinary clinic and owner home House features Owner annual questionnaire data on house type, house location (region and urban/rural classification) and household exposures/usage (fuel and water sources, pipe type). Annual owner responses then averaged (if binary exposure) or mode value determined for: - Early life (from SY1-2) - Adult life/rest of life (SY3 and beyond) - Whole life (all SY records) - 5 years prior to endpoint House type, urban rural status, region, water source, metal pipes, plastic pipes, heating fuel source, cooking fuel source for early life, rest of life, whole life and 5 y prior to endpoint Exposures Owner annual questionnaire on environmental exposures: aerosols, air cleaners, HEPA filters, moth balls, incense/candles, smoke, hours of smoke exposure, weed killer, insecticide and fertiliser. Questionnaire data were also captured on where dogs spent their time or slept (house, garage, outside). Annual owner responses on exposures were then aggregated for varying time points to create binary exposure variables for each environmental exposure across these time points: - Early life (from SY1-2) - Adult life/rest of life (SY3 and beyond) - Whole life (all SY records) - 5 years prior to endpoint The most common sleep location was also calculated for these time points. Aerosol, air cleaner, HEPA filter, mothballs, incense/candles, weed treatment, insect treatment, fertilizer treatment, exposed to smoke, smoke exposure hours, dog sleep location (garage, house or outside), mode sleep location for early life, rest of life, whole life and 5 y prior to endpoint Comorbidities An extensive list of conditions grouped by body system from annual veterinarian questionnaires were grouped in several ways for either 5 years prior to the endpoint or the dog’s whole life. For certain potential chronic inflammatory conditions any single record of the condition was considered to indicate it was a chronic/lifetime condition even if no further mention of it in subsequent study years e.g. Cushing’s disease. For other conditions, the condition must have been recorded for > 1 study year to be considered chronic e.g. heartworm infection. Presence of cancer (any, benign or malignant) 5 years prior to endpoint was assessed for using annual questionnaires and the malignancy and cause of death dataset. Full list of the conditions corresponding to each grouping in Supplementary Information Table A1. Chronic inflammation presence, number of comorbidities, type of comorbidities present (parasitism, infectious, immune mediated, cardiovascular, gastrointestinal, orthopaedic, other), presence of cancer (all cancers, benign, malignant), count of cancers (all cancers, benign, malignant) for whole life and 5 y prior to endpoint Intestinal Parasitism Tick-Borne Parasitism Other Parasitism Infectious Orthopaedic Immune Mediated Cardiovascular Gastrointestinal Inflammatory Other Chronic Inflammatory All cancers Malignant cancers Benign cancers Medications Medications Medications prescribed from annual veterinarian and owner questionnaires were grouped into therapeutic indication according to the National Office of Animal health (NOAH )compendium datasheets 65 Drug class usage was then split into quartiles (or halves where there was not enough variability in prescription numbers between dogs) for lifetime and 5 years (‘5y’) prior to endpoint usage. Usage of drugs of additional interest (NSAIDs, fenbendazole, steroids) were explored separately beyond drug classes and usage in lifetime and 5 years prior to endpoint were calculated. Lifetime usage and 5y prior to endpoint for each therapeutic indication according to the NOAH compendium and also specifically NSAIDs, steroids and fenbendazole Anti-histamines Anti-inflammatory Anti-microbial Anti-parasitics Anti-septics Anti-virals Cardio-respiratory Dietary supplements Diuretics Enteric Fluid and metabolites Hormones and related Immunologicals Miscellaneous Survival time was calculated in days from date of diagnosis to date of death, date of withdrawal or to date of censorship of data (31st December 2024). Kaplan–Meier survival ( i.e. , time-to-event) curves were constructed to describe time to death from haemangiosarcoma diagnosis for all cases and then separately for each tumour location group. Haemangiosarcoma locations were classified as: splenic, cardiac, visceral, other or cutaneous. Visceral haemangiosarcomas were those with haemangiosarcoma simultaneously identified at multiple different viscera (primarily cardiac and splenic). A log-rank test was used to verify if the survival time for different haemangiosarcoma locations differed significantly (p < 0.05). Ethics and consent to participate Ethical approval for the usage of GRLS data for this study was granted by the Royal Veterinary College Social Science Research Ethical Review Board (URN SR2023 - 0175). Ethical approval and consent to participate in the GRLS study was described by Guy and colleagues (2015) 1 . Dog owners signed a consent form to participate in the study and the overall study design was reviewed by the Animal Welfare Advisory Board of Morris Animal Foundation 1 . Data analysis Power calculations using the previous proportions of haemangiosarcoma diagnosed in neutered and entire dogs in this cohort (8% and 5% respectively 31 ) and an anticipated odds ratio of 1.5, indicated that our study would provide 78% power to detect a difference in haemangiosarcoma incidence between the groups 32 . Assuming a two-sided significance level (α) of 0.05, we calculated that a sample of 1,059 dogs per group would be required to provide 80% power to detect an odds ratio of 1.5 or greater using a two-sample test of proportions 30 . Lifetime prevalence of haemangiosarcoma was calculated as the number of haemangiosarcoma cases divided by the number of dogs enrolled in the GRLS cohort (n=3044). Incidence rate was calculated per 100 dog-years by dividing the number of haemangiosarcoma cases by the total years of participation of all 3,044 dogs. Years of participation is defined for cases as up until the date of diagnosis and for non-cases until the endpoint of December 31 st 2024. The ninety five percent confidence interval (95% CI) was estimated using exact methods 33 . Multivariable binary logistic regression modelling was used to explore associations between the risk factors and a haemangiosarcoma diagnosis. Variables with liberal association (p<0.2) in initial univariable binary logistic regression modelling were carried forward for exploration in multivariable binary logistic regression modelling 32 . A correlation matrix was built, and Variance Inflation Factor (VIF) was used to assess for collinearity between any liberally significant variables (p 5 32 . For collinear variables, each of the variables were assessed individually in the multivariable model and the variable with smallest model Akaike Information Criteria (AIC) value was retained for further modelling. The multivariable model was built using a manual stepwise forwards construction approach, where variables with the smallest p-value from the likelihood ratio test (LRT) and/or smallest AIC value were included first. Variables were retained in the model if a significant likelihood ratio test (LRT) result indicated the model was improved with variable inclusion. AIC values were used to compare non-nested models that were built with different collinear variables to determine which model balanced goodness of fit and complexity best. Pairwise interactions were assessed for all variables in the final model and interactions were included in the final model where significant (p 0.05 indicated no evidence that the model poorly fitted the data 32 . Statistical significance was set at the 5% level 32 . All data cleaning and analyses were performed in R Studio 34 using the following packages: survminer, dynpred, ggsurvfit, ggplot2, finalfit and pROC. Results Study population descriptives The overall proportion of dogs diagnosed with haemangiosarcoma in the GRLS cohort by December 31st 2024 was 16.1% (95% CI 14.8–17.4%). The overall incidence rate was 1.69 (95%CI 1.54–1.85) cases per 100 dog years. There were 490 dogs (both laboratory confirmed and veterinarian suspected) diagnosed with haemangiosarcoma in the GRLS cohort (n = 3044) by December 31st, 2024 or at any earlier date of censorship. The median duration enrolled for the overall cohort was 9.66 years (IQR = 7.09–10.35 years, range = 0.03–12.27 years), for the haemangiosarcoma cases was 8.40 years (IQR = 7.31–9.41 years, range = 2.92–12.12 years) and for the non-cases was 9.88 years (IQR = 6.99–10.45 years, range = 0.03–12.27 years). By December 31st, 2024, 1110 dogs (36.4%) were still alive and enrolled in GRLS. The demographics of the GRLS cohort at enrolment have been described in previous studies 2 , 12 . At enrolment, the cohort consisted of 1504 females (49.4%) and 1540 males (50.6%) For cases, 213 were female (43.5%) and 277 were male (56.5%). Of the non-cases, 1291 were female (50.5%) and 1263 were male (49.4%). The majority of haemangiosarcoma cases (n = 415, 84.7%) and non-cases (n = 2027, 79.3%) were neutered by the study cut-off date. The median age of the cohort at endpoint was 10.68 years (IQR = 8.47–11.65, range = 0.73–14.51 years). The median age of cases at their haemangiosarcoma diagnosis was 9.73 years (IQR = 8.68–10.86 years, range 3.60-13.32 years). The median age of the non-cases at their endpoint (either death from non-haemangiosarcoma causes, study withdrawal or still alive by 31st December 2024) was 10.83 years (IQR = 8.28–11.74, range 0.73–14.51 years). Haemangiosarcoma case descriptives Of the 490 dogs diagnosed with haemangiosarcoma, 9 dogs had two separate diagnoses of haemangiosarcoma recorded giving 499 separate haemangiosarcoma diagnosis events. Of these 9 dogs, 6 (66.7%) of the initial haemangiosarcoma were reported as cutaneous. The majority of haemangiosarcoma diagnoses overall were definitive as confirmed by a pathologist (n = 370, 74.1%), while the remaining had their clinical diagnosis supported through either direct visualisation or imaging of tumour without microscopy (n = 74, 14.7%) or on clinical suspicion only (n = 56, 11.2%) Haemangiosarcoma diagnosis was classified into the following main locations for each dog: spleen (n = 176, 35.2%), cardiac (n = 146, 29.2%), visceral (n = 91, 18.2%), cutaneous (n = 17,3.4%) and other (n = 70,14.0%). Due to the aggressive nature of haemangiosarcoma, half of the haemangiosarcoma diagnoses had tumour lesions in multiple organs (n = 252, 50.1%). The frequency of all haemangiosarcoma masses identified (n = 1053) is reported below in Fig. 1. The main locations were spleen (n = 252, 23.9%), cardiac (n = 240, 22.8%), lung (n = 170, 16.1%) and liver (n = 155, 14.7%). Risk factors associated with haemangiosarcoma diagnosis Univariable logistic regression Univariable binary logistic regression modelling identified 123 variables from the 172 variables as liberally significantly (LRT p-value < 0.20) associated with a haemangiosarcoma diagnosis (Table 2 and full univariable output in Supplementary Material Table A2). Due to many of these variables being variations of the same data, such as different ways to divide age at diagnosis, there were 179 pairs of correlated variables from 97 variables. Correlated variables were assessed in non-nested models to decide which variable/s to include in a final model. Multivariable risk factor analysis The final model retained 7 variables: age at endpoint, sex, median adult weight, any other cancer diagnosed, a chronic gastrointestinal comorbidity, number of NSAIDs administered and the mode of reported activity level, as recorded 5 years prior to diagnosis (Fig. 2). No significant pairwise interactions were identified between any final model variables. The final model showed no evidence of poor fit to the date (Hosmer and Lemeshow test p = 0.09) and showed good discriminatory ability with an ROC value of 0.80. All age groups had significantly reduced odds of haemangiosarcoma diagnosis compared to dogs aged > 7-<=9 years of age at endpoint. The odds of diagnosis were lowest in the two extremes of age group: dogs 13 years old (OR = 0.05, 95%CI 0.01–0.14). Male dogs had increased odds of haemangiosarcoma diagnosis compared to females (OR = 1.27, 95%CI 1.02–1.57). When assessing median adult weight in the 5 years prior to endpoint, dogs weighing 25–35 kg had over twice the odds of a haemangiosarcoma diagnosis compared to dogs weighing 15– 35 kg also had increased odds (OR = 1.63, 95% CI: 1.05–2.49). Dogs with a recorded non-haemangiosarcoma cancer in the 5 years prior to endpoint had a nearly four-fold reduction in odds of haemangiosarcoma (OR = 0.27, 95% CI: 0.21–0.36), while dogs with a chronic gastrointestinal comorbidity had increased odds (OR = 1.79, 95% CI: 1.21–2.61). Dogs with NSAID administration within the 5 years before endpoint had a higher odds of haemangiosarcoma. This was significant across most quartiles of NSAID usage, with the strongest association seen in dogs receiving two presciptions (OR = 2.16, 95% CI: 1.56–2.99). Dogs in the highest usage quartile (4–11 doses) also had increased odds (OR = 1.71, 95% CI: 1.08–2.65). The most frequently reported (mode) activity level five years prior to endpoint were also associated with diagnosis. Compared to dogs described as having little activity, those reported as moderately active (OR = 0.67, 95% CI: 0.46–0.99) or very active (OR = 0.49, 95% CI: 0.31–0.76) had significantly reduced odds of haemangiosarcoma. Survival analysis The median survival time (MST) from first diagnosis for haemangiosarcoma cases was 0 days (95% CI 0–0, IQR 0–10 days, range 0-1693 days) (Table 3 and Fig. 3). The 1-year survival rate was 2.6% (1.5–4.4%). Only cutaneous haemangiosarcoma had an MST longer than 0 days (MST = 19 days, 95%CI 0-479). The survival time of cutaneous haemangiosarcoma cases was significantly longer than visceral cases (Log-rank test p = 0.006). Additionally, between specific visceral locations there was a significant difference in survival time between splenic and cardiac (p = < 0.001) and splenic only and other visceral locations (p = 0.04) Table 3 Summary table of case numbers, median survival time (MST), interquartile range (IQR) and 1-year survival rate for 490 haemangiosarcoma cases (with 502 separate haemangiosarcoma diagnoses) identified from 3044 Golden Retrievers in the Golden Retriever Lifetime Study cohort . Haemangiosarcoma (n = 502) diagnoses are grouped by location: all, cardiac, splenic, other visceral and cutaneous. Location All Splenic Cardiac Visceral Cutaneous Number of haemangiosarcoma diagnoses (%) 499 176 (35.2) 146(29.2) 91 (18.1) 17 (3.4) MST (95% CI) 0 (0–0) 0 (0–3) 0 (0–0) 0 (0–0) 19 (0-479) IQR days (range, days) 0–10 (0-1693) 0–54 (0-774) 0–0 (0-1302) 0–0 (0-365) 0-165 (0-1543) 1-year survival rate (%) (95%CI) 2.6% (1.5–4.4) 2.3% (0.9–6.1) 1.3% (0.3–5.2) 1.01% (0.18–5.5%) 22% (9.4–53) Discussion Haemangiosarcoma is a highly aggressive cancer with poor prognosis that commonly affects middle-aged to older dogs, with Golden Retrievers being at increased risk 12 . Despite its clinical importance, the underlying risk factors for haemangiosarcoma development in dogs remain poorly understood. Leveraging the depth and longitudinal design of the Golden Retriever Lifetime Study (GRLS), this study aimed to determine the frequency of haemangiosarcoma diagnoses, identify associated risk factors, and characterise post-diagnosis survival outcomes in Golden Retrievers in the US. We hypothesised that older age and neutering status would be associated with increased odds of haemangiosarcoma diagnosis, and that dogs with visceral forms of haemangiosarcoma would have significantly shorter median survival times than those with cutaneous forms. Our findings confirm a high burden of haemangiosarcoma within the breed, with the disease diagnosed in 16.1% of the cohort. Multivariable modelling identified older age, male status, higher median adult weight, a history of chronic gastrointestinal disease, a non-haemangiosarcoma cancer diagnosis, NSAID administration, and reduced activity levels as significant risk factors for haemangiosarcoma diagnosis. For the non-demographic variables it was considered that any associations detected were exploratory and hypothesis generating for future work. Additionally, although cutaneous haemangiosarcoma was associated with longer survival than visceral forms, prognosis was poor across all haemangiosarcoma locations, with very short median survival times (MST 0 days, 95% CI 0–0) following diagnosis regardless of anatomical site. Factors associated with a haemangiosarcoma diagnosis Age Consistent with previous studies, the current study found a peak diagnosis of haemangiosarcoma in middle aged to older dogs. Dogs aged between 7–9 years had the highest odds of diagnosis with haemangiosarcoma in the current study. In comparison, dogs under five years and dogs aged over 13 years were 20 times less likely to be diagnosed with haemangiosarcoma. The median age at first diagnosis was 9.73 years, which aligns with previous reports in the literature across various breeds that range from 7.9–10.7 years) 8 , 10 , 31 , 35 – 37 . Previous studies examining age across a range of breeds identified an older, but comparable peak to that seen in this present study 8 , 10 . This difference could be explained by the fact that these studies included all dog breeds in their analyses whereas the current study focussed on one, large breed (Golden Retrievers). Large breed dogs have been shown to have shorter lifespans than most small and medium breeds, which means that their peak risk of haemangiosarcoma development and diagnosis should predictably be chronologically earlier than smaller dogs 38 , 39 . In addition, as this is an enrolled, optional, lifetime study, this cohort might select for more motivated owners who access veterinary care more frequently which could lead to an earlier diagnosis. Sex and neutering In the present study, males had 1.3 times increased odds of haemangiosarcoma diagnosis compared to females). An association between male sex and haemangiosarcoma diagnosis has been reported previously 7 . Additionally, one study has shown male sex to be associated with increased hazard of death from haemangiosarcoma, across a multi- breed population 30 . In the current study, it was hypothesised that neutering and/or age at neutering would be important risk factors for the diagnosis of haemangiosarcoma. Several other studies have explored sex and neutering associations with haemangiosarcoma in conjunction, with most studies finding both neutered and entire males to be at increased odds of haemangiosarcoma diagnosis versus entire females 8 , 10 . This same association was seen in univariable analyses in this study, but the sex variable was deemed to have better fit for the final model than sex-neuter status, age at neutering or neutering variables so these were not included in the final regression model 38 , 39 . The reasons behind the sex association identified here are unclear and likely multifactorial and encompass the effects of androgens on cancer biology, immunologic differences and lifestyle factors 40 , 41 . While angiosarcoma in people is not typically considered a hormone-dependent cancer, emerging evidence suggests that sex hormones can influence the tumour microenvironment and potentially affect immune surveillance 42 . Additionally, due to the high frequency of neutering in most dog populations (80% in the GRLS cohort) there may be other, sex-linked genetic factors associated to the increased odds of diagnosis seen in male dogs here 42 . Weight Higher median adult weight was associated with increased odds of haemangiosarcoma diagnosis in this study. Overweight and obese status have been linked to increased cancer risk in both human and veterinary studies, potentially due to the low-grade chronic inflammation associated with excess body fat 43 . However, interpreting weight in isolation is challenging without concurrent body condition score (BCS) and sex data, as these are important for determining whether a dog is truly overweight. According to American Kennel Club breed standards, the normal weight range for adult Golden Retrievers is 25–29 kg for females and 29–32 kg for males 44 , meaning that many dogs in the 25–35 kg group could fall within a normal weight range depending on sex. An interaction between weight and sex was tested but was not statistically significant. Although BCS may offer better insight into adiposity status, it was not retained in the final multivariable model. Previous studies have found higher weight to be associated with increased odds of haemangiosarcoma and when considered in conjunction with these results in a single breed might indicate increased risk in larger animals rather than an effect of overweight/obesity status 8 , 10 . Comorbidities Having had a previous separate cancer diagnosis within the five years prior to endpoint was associated with reduced odds of a haemangiosarcoma diagnosis. We postulate that this could reflect survival bias, where those dogs diagnosed with an earlier cancer are more likely to have died or be euthanised from that cancer and therefore not live long enough to develop a haemangiosarcoma. Alternatively, this may just reflect that the cases were younger and had less prior cancers, as the five years prior to haemangiosarcoma diagnosis were likely earlier years in their life, than non-cases where the five years was largely prior to the end of our censoring point. In humans, treatment with chemotherapy and radiotherapy in cancer survivors may increase the risk of second cancers during their lifetime 45 , 46 , it is also possible that this could have a protective or immunomodulatory effect on the development of haemangiosarcoma in the short team, given the shorter life-span seen in dogs, although there is no evidence for this phenomenon in veterinary oncology. Chronic inflammatory conditions have been suggested to play a role in the development of certain cancers due to continued, low-grade inflammation creating favourable conditions for carcinogenesis and tumour metastasis 22 , 23 . However, when all chronic inflammatory conditions in the five years prior to diagnosis were grouped together, this variable showed no significant effect in the current study. One reason for this variable not being significant is that potentially not all chronic inflammatory conditions are associated with cancer development. For example, chronic inflammatory conditions associated with lymphoedema development are risk factors for angiosarcoma development in humans 47 . The only comorbidity retained in our final multivariable model was the presence of a gastrointestinal comorbidity which was associated with increased odds of a haemangiosarcoma diagnosis. Chronic gastrointestinal conditions have not previously been associated with haemangiosarcoma specifically; however, a range of gastrointestinal conditions have been associated with other cancers in veterinary and human studies 23 . Medications An association between increased NSAID usage and higher odds of haemangiosarcoma diagnosis was seen in the current study. This association was potentially unexpected as some human studies have reported that prior NSAID use may reduce the risk of the development of some cancers and postulated that this could be mediated by enhancing DNA damage repair, or inhibition of tumour proliferation and invasion via COX-dependent or/and -independent pathways 19 , 48 . In veterinary oncology, certain NSAIDs have been shown to have anti-tumour properties, for example the use of piroxicam in dogs with transitional cell carcinoma of the bladder 21 , 49 . Deracoxib has been suggested as beneficial in dogs with stage III haemangiosarcoma in one study, although a control group was not included 50 . There are several possible reasons why the opposite NSAID relationship was seen in the current study. Firstly, veterinary studies have looked at NSAIDs as a therapy once a cancer is diagnosed rather than a protective effect, whereas the current study examined factors associated specifically with the development of haemangiosarcoma 21 .There is also evidence to suggest that COX-2 expression may differ between haemangiosarcoma locations; while expression was not seen in splenic forms, frequent but weak expression is present in cutaneous forms 51 , 52 . Furthermore, the type of cyclo-oxygenase (COX) inhibition by different types of NSAIDs varies and could have varying effects on tumour initiation or development 21 . Additionally, looking further into NSAID prescribing timings is warranted, for example NSAID prescription within a few months of haemangiosarcoma diagnosis could have been for presenting signs associated with an undiagnosed haemangiosarcoma mass. The NSAID association seen here should be explored further to determine if different NSAIDs have differing associations and whether dosage, duration of course/s and frequency of usage affects outcomes. Activity level Dogs with higher owner-reported activity levels (moderate or very active) had reduced odds of haemangiosarcoma diagnosis in the current study. An association between more physical activity and reduced risk of the development of certain cancers has been previously reported in human oncology 53 . A recent GRLS cohort study identified activity as an important predictor in cancer diagnosis 24 . Ronzani and colleagues (2025) found that frequency and intensity of exercise had a greater impact on diagnosis, than other metrics such as exercise type. Whilst our study did not examine the full range of activity variables that Ronzani and colleagues (2025) explored; it is of interest to see an activity association with a specific cancer (haemangiosarcoma) for this cohort. However, this relationship warrants further investigation, as it remains unclear whether the protective effect of higher activity levels is due to physical activity itself or related factors such as lower bodyweight or reduced chronic inflammation 45 ,4647, 48 . Survival time after haemangiosarcoma diagnosis Median survival time for Golden Retrievers with haemangiosarcoma was short (overall MST 0 days, 95%CI 0–0 days), likely underscoring the aggressive nature of this malignancy. However, it is unclear how much effect the perceived grave prognosis of haemangiosarcoma affects euthanasia choices and whether this becomes a self-fulfilling prophecy. Even though cutaneous haemangiosarcoma can have a more favourable prognosis and the potential for curative resection, the MST was still very short (19 days, 95% CI 0-479) 5 , although survival time was significantly increased versus visceral haemangiosarcomas. The current study reported a much shorter MST compared to many of the previous cutaneous haemangiosarcoma studies reported in the literature (MST: 172–1189 days) (Hargis et al., 1992; Nóbrega et al., 2019; Shiu et al., 2011; Ward et al., 1994). In addition, previous studies of various visceral haemangiosarcoma locations (cardiac and splenic), reported MST ranged from 7-259 days which is longer than the MST of 0 days seen for each visceral locations in the current study. 26 , 28 , 30 , 36 , 57 . The longer MSTs previously reported may reflect referral bias whereby most of these earlier studies included referral cases which could exhibit selection bias for more clinically stable case populations and early disease stage or towards owners highly committed to trialling treatment 58 . A recent, primary-care study of UK dogs reported survival times more in line with what was seen in this study (splenic MST: 4 days, cardiac MST: 0 days) 30 . Given that the GRLS cohort dogs are largely primary-care practice based, the aforementioned primary-care population study may be a more appropriate population to compare survival times to. However, it is also possible that the very short MSTs in this population were due to Golden Retrievers being more prone to develop more aggressive haemangiosarcoma phenotype than the wider dog population, and the higher prevalence of cardiac forms which are typically associated with a poor prognosis regardless of the treatment pursued 5 , 30 , 36 . Therefore, the current results derived just from Golden Retrievers should be extrapolated to other dog brees with caution. Strengths and limitations As these analyses were undertaken within the Golden Retriever Lifetime Study (GRLS), a prospective longitudinal cohort, there are several inherent strengths and limitations to consider. A key strength is the repeated, annual collection of data across a wide range of variables such as clinical examination findings, activity levels, comorbidities, and medication usage 1 , 2 . This facilitates the exploration of time-specific (‘time window’) associations with haemangiosarcoma, while also reducing recall bias compared to study designs reliant on lifetime recall at the point of diagnosis. Additionally, given that the influence of environment and lifestyle on cancer development remains poorly understood in veterinary medicine, this study provides a unique opportunity for hypothesis generation in this area. However, there are several important limitations. Much of the data was manually collected via owner and veterinarian questionnaires and only at annual time points, which introduces the potential for recall inaccuracies or misreporting across each one-year window. For example, estimating the frequency of routine medication use or characterising a dog's activity level over a year may be imprecise. Additionally, one criterion for enrolment was pedigree status, which might mean these Golden Retrievers are more inbred than non-pedigree Golden Retrievers. Inbreeding could potentially affect cancer risk; although this elevated risk hasn’t been demonstrated in Golden Retrievers, it has been observed in some other breeds 59 . Furthermore, the cohort size (~ 3,000 dogs) may be underpowered to detect environmental associations of small magnitude but potentially significant biological relevance. For example, a large electronic health record study identified an association between herbicide exposure and canine lymphoma diagnosis (Schofield et al., 2019), an association not replicated in smaller studies. This discrepancy is likely at least partly attributable to limited statistical power; Schofield and colleagues estimated that approximately 200 cases derived from a population of 400,000 were required to detect such an effect with sufficient power 17 , 60 . Finally, the optimal exposure window for carcinogen-related cancer development remains uncertain. While this study examined multiple windows — including early life, lifetime, and the five years preceding diagnosis/end-point — it is possible that key exposures occurred outside these periods. For instance, in utero or neonatal exposures (prior to enrolment in GRLS from six months of age) may be relevant but were not captured in this dataset. One older study reported perinatal radiation exposure in Beagles increased the risk of fatal malignancies 61 . This evidence is lacking for other potential carcinogens and other cancers. Future work Whilst several of the factors associated with haemangiosarcoma diagnosis in this study were anticipated (age and weight), others, namely the associations between haemangiosarcoma diagnosis and NSAID usage and previous cancer diagnosis, were novel findings. As this was an exploratory study, no adjustment was made for multiple comparisons. The analyses were intended to identify potential associations and generate hypotheses for future confirmatory studies. As such, findings should be interpreted with caution given the increased risk of Type I error due to multiple testing. Future studies should be undertaken to explore these novel associations further. For example, causal study methods such as target trial emulations could utilise the observational data in the GRLS and attempt to explore whether there is any causality between NSAID usage and haemangiosarcoma development 62 . Additionally, further exploration of the association between gastrointestinal comorbidities and cancer development should be undertaken. Other future studies should explore whether the associations identified here are unique to haemangiosarcoma diagnosis or are also recognised in other cancers in this cohort. Conclusions This study provides valuable preliminary insights into the epidemiology of haemangiosarcoma within a large, longitudinal cohort of Golden Retrievers, identifying a range of demographic, lifestyle, and health-related factors potentially associated with diagnosis. The very poor survival outcomes observed highlight the ongoing clinical challenges posed by this disease. While the study confirmed some anticipated associations, such as those with age, it also revealed several unanticipated findings, including associations with NSAID usage and prior cancer diagnoses. These novel associations should be interpreted with caution and considered hypothesis-generating, warranting further investigation in independent datasets to explore potential underlying mechanisms and confirm their validity. Declarations Funding The Golden Retriever Lifetime Study and this manuscript were made possible through financial support provided by the Morris Family Foundation, the Mark & Bette Morris Family Foundation, VCA, the V Foundation, Blue Buffalo Company, Petco Love, Zoetis, Antech Inc., Elanco, the Purina Institute, Orvis, the Golden Retriever Foundation, the Hadley and Marion Stuart Foundation, Mars Veterinary, generous private donors and the Flint Animal Cancer Center at Colorado State University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. References Guy MK, Page RL, Jensen WA, et al. The golden retriever lifetime study: Establishing an observational cohort study with translational relevance for human health. Philosophical Transactions of the Royal Society B: Biological Sciences . 2015;370(1673). doi:10.1098/rstb.2014.0230 Labadie J, Swafford B, DePena M, Tietje K, Page R, Patterson-Kane J. Cohort profile: The Golden Retriever Lifetime Study (GRLS). PLoS One . 2022;17(6 June). doi:10.1371/journal.pone.0269425 Pugh CA, Bronsvoort BM de C, Handel IG, Summers KM, Clements DN. What can cohort studies in the dog tell us? Canine Genet Epidemiol . 2014;1(1):5. doi:10.1186/2052-6687-1-5 Burns PB, Rohrich RJ, Chung KC. The levels of evidence and their role in evidence-based medicine. Plast Reconstr Surg . 2011;128(1):305-310. doi:10.1097/PRS.0b013e318219c171 De Nardi AB, de Oliveira Massoco Salles Gomes C, Fonseca-Alves CE, et al. Diagnosis, Prognosis, and Treatment of Canine Hemangiosarcoma: A Review Based on a Consensus Organized by the Brazilian Association of Veterinary Oncology, ABROVET. Cancers (Basel) . 2023;15(7). doi:10.3390/cancers15072025 Withrow SJ, Vail DM, Page RL. Withrow & MacEwen’s Small Animal Clinical Oncology. 5th ed. Elsevier Saunders; 2013. Robinson KL, Bryan ME, Atkinson ES, Keeler MR, Hahn AW, Bryan JN. Neutering Is Associated with Developing Hemangiosarcoma in Dogs in the Veterinary Medical Database: An Age and Time-Period Matched Case-Control Study (1964-2003) . Vol 61.; 2020. Dorea F,, Brodbelt D. Society for Veterinary Epidemiology and Preventive Medicine Conference Proceedings. In: Dorea F,, Brodbelt D, eds. Society for Veterinary Epidemiology and Preventive Medicine Conference Proceedings . ; 2025:149-160. Dobson JM, Samuel S, Milstein H, Rogers K, Wood JLN. Canine neoplasia in the UK: Estimates of incidence rates from a population of insured dogs. Journal of Small Animal Practice . 2002;43(6):240-246. doi:10.1111/j.1748-5827.2002.tb00066.x Carnio A, Eleni C, Cocumelli C, et al. Evaluation of intrinsic and extrinsic risk factors for dog visceral hemangiosarcoma: A retrospective case-control study register-based in Lazio region, Italy. Prev Vet Med . 2020;181. doi:10.1016/j.prevetmed.2020.105074 Srebernik N, Appleby E. Breed prevalence and sites of haemangioma and haemangiosarcoma in dogs. Veterinary Record . 1991;129(18):408-409. doi:10.1136/vr.129.18.408 Hillman A, Swafford B, Delavenne C, Fieten H, Boerkamp K, Tietje K. Descriptive analysis of haemangiosarcoma occurrence in dogs enrolled in the Golden Retriever lifetime study. Vet Comp Oncol . Published online August 27, 2023. doi:10.1111/vco.12933 Fleming JM, Creevy KE, Promislow DEL. Mortality in North American Dogs from 1984 to 2004: An Investigation into Age-, Size-, and Breed-Related Causes of Death. J Vet Intern Med . 2011;25(2):187-198. doi:10.1111/j.1939-1676.2011.0695.x Torres de la Riva G, Hart BL, Farver TB, et al. Neutering Dogs: Effects on Joint Disorders and Cancers in Golden Retrievers. PLoS One . 2013;8(2). doi:10.1371/journal.pone.0055937 Kent MS, Burton JH, Dank G, Bannasch DL, Rebhun RB. Association of cancer-related mortality, age and gonadectomy in golden retriever dogs at a veterinary academic center (1989-2016). PLoS One . 2018;13(2). doi:10.1371/journal.pone.0192578 Hart BL, Hart LA, Thigpen AP, Willits NH. Assisting Decision-Making on Age of Neutering for 35 Breeds of Dogs: Associated Joint Disorders, Cancers, and Urinary Incontinence. Front Vet Sci . 2020;7. doi:10.3389/fvets.2020.00388 Wise CF, Breen M, Stapleton HM. Canine on the Couch: The New Canary in the Coal Mine for Environmental Health Research. Environment and Health . 2024;2(8):517-529. doi:10.1021/envhealth.4c00029 Owada K, Nicholls E, Soares Magalhães RJ, Palmieri C. Environmental exposure and cancer occurrence in dogs: a critical appraisal of evidence. Res Vet Sci . 2025;184. doi:10.1016/j.rvsc.2024.105517 Harris R, Beebe-Donk J, Doss H, Doss D. Aspirin, ibuprofen, and other non-steroidal anti-inflammatory drugs in cancer prevention: A critical review of non-selective COX-2 blockade (Review). Oncol Rep . Published online April 1, 2005. doi:10.3892/or.13.4.559 Pottegård A, Friis S, Christensen R de P, Habel LA, Gagne JJ, Hallas J. Identification of Associations Between Prescribed Medications and Cancer: A Nationwide Screening Study. EBioMedicine . 2016;7:73-79. doi:10.1016/j.ebiom.2016.03.018 Raposo TP, Beirão BCB, Pang LY, Queiroga FL, Argyle DJ. Inflammation and cancer: Till death tears them apart. Veterinary Journal . 2015;205(2):161-174. doi:10.1016/j.tvjl.2015.04.015 Multhoff G, Molls M, Radons J. Chronic inflammation in cancer development. Front Immunol . 2012;2(JAN). doi:10.3389/fimmu.2011.00098 Morrison WB. Inflammation and cancer: A comparative view. J Vet Intern Med . 2012;26(1):18-31. doi:10.1111/j.1939-1676.2011.00836.x Ronzani D, Hooper SE. Physical activity predictors of cancer in Golden Retrievers: it’s about frequency and intensity, not type. Veterinary Oncology . 2025;2(1):11. doi:10.1186/s44356-025-00024-5 Taylor C, Barry GJ, O’Neill DG, et al. Survival time and prognostic factors in dogs clinically diagnosed with haemangiosarcoma in UK first opinion practice. Clegg, S, ed. PLoS One . 2025;20(6):e0316066. doi:10.1371/journal.pone.0316066 Wendelburg KM, Price LL, Burgess KE, Lyons JA, Lew FH,, Berg J. Survival time of dogs with splenic hemangiosarcoma treated by splenectomy with or without adjuvant chemotherapy: 208 cases (2001–2012). JAVMA . 2015;247(4):393. Bray JP, Orbell G, Cave N, Munday JS. Does thalidomide prolong survival in dogs with splenic haemangiosarcoma? Journal of Small Animal Practice . 2018;59(2):85-91. doi:10.1111/jsap.12796 Mullin CM, Arkans MA, Sammarco CD, et al. Doxorubicin chemotherapy for presumptive cardiac hemangiosarcoma in dogs. Vet Comp Oncol . 2016;14(4):e171-e183. doi:10.1111/vco.12131 Story AL, Wavreille V, Abrams B, Egan A, Cray M, Selmic LE. Outcomes of 43 small breed dogs treated for splenic hemangiosarcoma. Veterinary Surgery . 2020;49(6):1154-1163. doi:10.1111/vsu.13470 Taylor C, Barry GJ, O’Neill DG, et al. Survival time and prognostic factors in dogs clinically diagnosed with haemangiosarcoma in UK first opinion practice. Clegg, S, ed. PLoS One . 2025;20(6):e0316066. doi:10.1371/journal.pone.0316066 Hillman A, Swafford B, Delavenne C, Fieten H, Boerkamp K, Tietje K. Descriptive analysis of haemangiosarcoma occurrence in dogs enrolled in the Golden Retriever lifetime study. Vet Comp Oncol . 2023;21(4):700-708. doi:10.1111/vco.12933 Dohoo I, Martin W, Stryhn H. Veterinary Epidemiological Research . VER Inc.; 2009. Kirkwood BR, Sterne JAC. Essential Medical Statistics . 2nd ed. Blackwell; 2003. doi:10.1192/bjp.112.483.211-a R Core Team. RStudio. Published online 2019. Davies O, Taylor AJ. Refining the “double two-thirds” rule: Genotype-based breed grouping and clinical presentation help predict the diagnosis of canine splenic mass lesions in 288 dogs. Vet Comp Oncol . 2020;18(4):548-558. doi:10.1111/vco.12574 Yamamoto S, Hoshi K, Hirakawa A, Chimura S, Kobayashi M, Machida N. Epidemiological, clinical and pathological features of primary cardiac hemangiosarcoma in dogs: A review of 51 cases. Journal of Veterinary Medical Science . 2013;75(11):1433-1441. doi:10.1292/jvms.13-0064 Szivek A, Burns RE, Gericota B, et al. Clinical outcome in 94 cases of dermal haemangiosarcoma in dogs treated with surgical excision: 1993-2007. Vet Comp Oncol . 2012;10(1):65-73. doi:10.1111/j.1476-5829.2011.00282.x McMillan KM, Bielby J, Williams CL, Upjohn MM, Casey RA, Christley RM. Longevity of companion dog breeds: those at risk from early death. Sci Rep . 2024;14(1). doi:10.1038/s41598-023-50458-w Lewis TW, Wiles BM, Llewellyn-Zaidi AM, Evans KM, O’Neill DG. Longevity and mortality in Kennel Club registered dog breeds in the UK in 2014. Canine Genet Epidemiol . 2018;5(1). doi:10.1186/s40575-018-0066-8 Cook MB, McGlynn KA, Devesa SS, Freedman ND, Anderson WF. Sex disparities in cancer mortality and survival. Cancer Epidemiology Biomarkers and Prevention . 2011;20(8):1629-1637. doi:10.1158/1055-9965.EPI-11-0246 Radkiewicz C, Johansson ALV, Dickman PW, Lambe M, Edgren G. Sex differences in cancer risk and survival: A Swedish cohort study. Eur J Cancer . 2017;84:130-140. doi:10.1016/j.ejca.2017.07.013 Özdemir BC, Dotto GP. Sex Hormones and Anticancer Immunity. Clinical Cancer Research . 2019;25(15):4603-4610. doi:10.1158/1078-0432.CCR-19-0137 Marchi PH, Vendramini THA, Perini MP, et al. Obesity, inflammation, and cancer in dogs: Review and perspectives. Front Vet Sci . 2022;9. American Kennel Club. Official Standard for the Golden Retriever. 1981. Accessed June 3, 2025. https://images.akc.org/pdf/breeds/standards/GoldenRetriever.pdf Dracham CB, Shankar A, Madan R. Radiation induced secondary malignancies: A review article. Radiat Oncol J . 2018;36(2):85-94. doi:10.3857/roj.2018.00290 Morton LM, Dores GM, Schonfeld SJ, et al. Association of Chemotherapy for Solid Tumors With Development of Therapy-Related Myelodysplastic Syndrome or Acute Myeloid Leukemia in the Modern Era. JAMA Oncol . 2019;5(3):318. doi:10.1001/jamaoncol.2018.5625 Cao J, Wang J, He C, Fang M. Angiosarcoma: A Review of Diagnosis and Current Treatment . Vol 9.; 2019. www.ajcr.us/ Cuzick J, Otto F, Baron JA, et al. Aspirin and non-steroidal anti-inflammatory drugs for cancer prevention: an international consensus statement. Lancet Oncol . 2009;10(5):501-507. doi:10.1016/S1470-2045(09)70035-X Knapp DW, Richardson RC, Chan TCK, et al. Piroxicam Therapy in 34 Dogs With Transitional Cell Carcinoma of the Urinary Bladder. J Vet Intern Med . 1994;8(4):273-278. doi:10.1111/j.1939-1676.1994.tb03232.x Kahn SA, Mullin CM, de Lorimier LP, et al. Doxorubicin and deracoxib adjuvant therapy for canine splenic hemangiosarcoma: a pilot study. Can Vet J . 2013;54(3):237-242. Heller DA, Clifford CA, Goldschmidt MH, Holt DE, Manfredi MJ, Sorenmo KU. Assessment of Cyclooxygenase-2 Expression in Canine Hemangiosarcoma, Histiocytic Sarcoma, and Mast Cell Tumor. Vet Pathol . 2005;42(3):350-353. doi:10.1354/vp.42-3-350 Nóbrega DF, Sehaber VF, Madureira R, Bracarense APFRL. Canine Cutaneous Haemangiosarcoma: Biomarkers and Survival. J Comp Pathol . 2019;166:87-96. doi:10.1016/j.jcpa.2018.10.181 Mctiernan A, Friedenreich CM, Katzmarzyk PT, et al. Physical Activity in Cancer Prevention and Survival: A Systematic Review. Med Sci Sports Exerc . 2019;51(6):1252-1261. doi:10.1249/MSS.0000000000001937 Ward H, Fox LE, Calderwood‐Mays MB, Hammer AS, Couto CG. Cutaneous Hemangiosarcoma in 25 Dogs: A Retrospective Study. J Vet Intern Med . 1994;8(5):345-348. doi:10.1111/j.1939-1676.1994.tb03248.x Shiu KB, Flory AB, Anderson CL, et al. Predictors of outcome in dogs with subcutaneous or intramuscular hemangiosarcoma. J Am Vet Med Assoc . 2011;238(4):472-479. doi:10.2460/javma.238.4.472 Hargis AM, Ihrke PJ, Spangler WL, Stannard AA. A Retrospective Clinicopathologic Study of 212 Dogs with Cutaneous Hemangiomas and Hemangiosarcomas. Journal of Veterinary Pathology . 1992;29(4). Schultheiss PC. A retrospective study of visceral and nonvisceral hemangiosarcoma and hemangiomas in domestic animals. Journal of veterinary diagnostic investigation . 2004;16:522-526. Bartlett PC, Van Buren JW, Neterer M, Zhou C. Disease surveillance and referral bias in the veterinary medical database. Prev Vet Med . 2010;94(3-4):264-271. doi:10.1016/j.prevetmed.2010.01.007 Nunney L. The effect of body size and inbreeding on cancer mortality in breeds of the domestic dog: a test of the multi-stage model of carcinogenesis. R Soc Open Sci . 2024;11(1). doi:10.1098/rsos.231356 Schofield I, Stevens KB, Pittaway C, et al. Geographic distribution and environmental risk factors of lymphoma in dogs under primary-care in the UK. Journal of Small Animal Practice . 2019;60(12):746-754. doi:10.1111/jsap.13075 Benjamin SA, Saunders WJ, Angleton GM,, Lee AC. Radiation Carcinogenesis in Dogs Irradiated During Prenatal and Postnatal Development. J Radiat Res . Published online 1991:86-103. Hernán MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. Am J Epidemiol . 2016;183(8):758-764. doi:10.1093/aje/kwv254 Purina Institute. The Purina 9-point body condition score system. Accessed January 24, 2025. https://www.purinainstitute.com/centresquare/nutritional-and-clinical-assessment-tools/the-purina-body-condition-system#:~:text=The%20Purina%209-Point%20Body%20Condition%20System%2C%20available%20for,a%20pet %20for%20excess%20or%20inadequate%20body%20fat. U.S Census Bureau. Poverty data sources: Poverty rates. Accessed January 17, 2025. https://www.census.gov/topics/income-poverty/poverty/about/related-sites/rates.html National Office of Animal Health. NOAH Compendium of Data Sheets for Animal Medicines . Accessed January 28, 2025. https://www.noahcompendium.co.uk/datasheets Table 2 Table 2 is available in the Supplementary Files section. Additional Declarations Competing interest reported. J.L and B.S are employees of the Morris Animal Foundation and authors C.T and G.B’s work is funded by a grant from the Morris Animal Foundation. Supplementary Files 250630SupplementarytableA1.docx 25063SuppmaterialtableA2.docx Table2.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-7064010","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":483691181,"identity":"a8421b9e-a7ad-4cbd-922c-58d324da4d6c","order_by":0,"name":"Collette Taylor","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA3UlEQVRIiWNgGAWjYLCCBwwMPPzsDTAuGwHlIPkEBgYZyZ4DJGqxMbiRQKQW/vnNzyQSKu7xSM58+0ziB4OdPINEWgJeLRLH2MwkEs4U8/BLp5tJ9jAkGzZIpB3Aq8WAjcFMIrEtgUdydhqzMQMDcwKDRHoDAS3s38BaDG4eA2mpJ0YLD8QWgxtsjI8ZGA4DtRBwmMSxnGKLhDNAh/WkMT7sMThu2MbzLAGvFv7m4xtvfKhIsOdnP8Zw4EdFtTw/e5oBXi1AwCKB5E7CEQkCzB+IUDQKRsEoGAUjGQAADAo5qA4iAQoAAAAASUVORK5CYII=","orcid":"","institution":"The Royal Veterinary College","correspondingAuthor":true,"prefix":"","firstName":"Collette","middleName":"","lastName":"Taylor","suffix":""},{"id":483691182,"identity":"c11721c2-7da4-4ce3-9fad-3ad88dbca596","order_by":1,"name":"Georgina J. Barry","email":"","orcid":"","institution":"The Royal Veterinary College","correspondingAuthor":false,"prefix":"","firstName":"Georgina","middleName":"J.","lastName":"Barry","suffix":""},{"id":483691183,"identity":"f6068ef7-a8cb-4feb-9274-a33814efc886","order_by":2,"name":"Dan G. O’Neill","email":"","orcid":"","institution":"The Royal Veterinary College","correspondingAuthor":false,"prefix":"","firstName":"Dan","middleName":"G.","lastName":"O’Neill","suffix":""},{"id":483691184,"identity":"beb95296-a0a0-4d25-8332-eaa1f3bea227","order_by":3,"name":"Alexandra Guillén","email":"","orcid":"","institution":"The Royal Veterinary College","correspondingAuthor":false,"prefix":"","firstName":"Alexandra","middleName":"","lastName":"Guillén","suffix":""},{"id":483691185,"identity":"d56f178b-7c54-4979-abc4-0f6362c92310","order_by":4,"name":"Julia Labadie","email":"","orcid":"","institution":"Morris Animal Foundation","correspondingAuthor":false,"prefix":"","firstName":"Julia","middleName":"","lastName":"Labadie","suffix":""},{"id":483691186,"identity":"5f913e4a-ddb4-457e-a0c8-4e4493ff2f67","order_by":5,"name":"Brenna Swafford","email":"","orcid":"","institution":"Morris Animal Foundation","correspondingAuthor":false,"prefix":"","firstName":"Brenna","middleName":"","lastName":"Swafford","suffix":""},{"id":483691187,"identity":"e42c567f-4137-4ee3-aa33-ba6eba9dcb61","order_by":6,"name":"Dave C. Brodbelt","email":"","orcid":"","institution":"The Royal Veterinary College","correspondingAuthor":false,"prefix":"","firstName":"Dave","middleName":"C.","lastName":"Brodbelt","suffix":""}],"badges":[],"createdAt":"2025-07-07 09:53:29","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7064010/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7064010/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":86665558,"identity":"e58de4f4-9709-4fb2-86bf-906c31c1afd3","added_by":"auto","created_at":"2025-07-14 11:03:59","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":1715053,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eTumour locations (n=1053) identified at haemangiosarcoma diagnosis in 490 Golden Retriever dogs diagnosed with haemangiosarcoma in the Golden Retriever Lifetime Study. \u003c/strong\u003eMore than one location could be recorded per dog.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/21003e7258c3bb3393e1b664.png"},{"id":86664872,"identity":"b4f685cc-b469-410a-90ef-a2b63ee3cb1c","added_by":"auto","created_at":"2025-07-14 10:55:59","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2054073,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eForest plot showing factors associated with haemangiosarcoma diagnosis (n=490) in the Golden Retriever Lifetime Study cohort (n=3044) in the final multivariable logistic regression model.\u003c/strong\u003e Third and fourth columns represent the number of non-cases and cases, respectively and percentages for each category in the model. The fifth column indicates the odds ratio for each category and 95% confidence interval (CI) and p-value.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/1db8077df972b3c1ef567a2a.png"},{"id":86665559,"identity":"12699295-adda-4661-9a13-e67fa1bdac0f","added_by":"auto","created_at":"2025-07-14 11:03:59","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":165771,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eKaplan-Meier survival curve showing survival times for haemangiosarcoma cases diagnosed in Golden Retrievers dogs enrolled in the Golden Retriever Lifetime Study cohort (n=490) separated by location affected: cardiac, cutaneous, other visceral, splenic and other. \u003c/strong\u003eDay 0 designates day of first diagnosis. Shaded areas indicate the 95% confidence interval for each curve.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/6926148a9b2adaaa9f0667d5.png"},{"id":101296896,"identity":"9a4c862d-4099-4390-8ee5-6da802dd2086","added_by":"auto","created_at":"2026-01-28 09:22:47","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":6250939,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/68555856-cd7c-4e8e-946e-5aab7c8aa242.pdf"},{"id":86664866,"identity":"cf1f5c7e-197b-4de9-b0d2-19bcf66720a7","added_by":"auto","created_at":"2025-07-14 10:55:59","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":58138,"visible":true,"origin":"","legend":"","description":"","filename":"250630SupplementarytableA1.docx","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/93c9da1778a8fd6dd17d9ee7.docx"},{"id":86664870,"identity":"ba7e4e10-1e2e-4595-a43f-8a76d6c9f67d","added_by":"auto","created_at":"2025-07-14 10:55:59","extension":"docx","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":195379,"visible":true,"origin":"","legend":"","description":"","filename":"25063SuppmaterialtableA2.docx","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/73a32e724e53206c9095a23d.docx"},{"id":86664869,"identity":"c24d05a0-5631-49dd-b5d9-a0092173c571","added_by":"auto","created_at":"2025-07-14 10:55:59","extension":"docx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":22947,"visible":true,"origin":"","legend":"","description":"","filename":"Table2.docx","url":"https://assets-eu.researchsquare.com/files/rs-7064010/v1/379de5942b613e214ad9ce0c.docx"}],"financialInterests":"Competing interest reported. J.L and B.S are employees of the Morris Animal Foundation and authors C.T and G.B’s work is funded by a grant from the Morris Animal Foundation.","formattedTitle":"Haemangiosarcoma in the Golden Retriever Lifetime Study: frequency, risk factors and post-diagnosis survival","fulltext":[{"header":"Introduction","content":"\u003cp\u003eThe Golden Retriever Lifetime Study (GRLS) is a prospective, longitudinal, lifetime cohort study designed to allow for exploration of demographic, environmental and lifestyle risk factors and their associations with outcomes of interest such as neoplasia \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Evidence pyramids place cohort studies above designs like case–control studies or tumour-registry case series, because cohorts yield more reliable estimates of disease frequency and associated risk factors \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e,\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eHaemangiosarcoma is a malignant neoplasm originating from vascular endothelial or bone marrow progenitor cells that has been most commonly reported in visceral organs (typically spleen, heart, liver) and cutaneous or subcutaneous tissues \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u003c/sup\u003e. The estimated incidence in across a range of breeds varies, from 0.007 to 0.4% based on veterinary clinic and tumour registry populations, respectively \u003csup\u003e\u003cspan additionalcitationids=\"CR8 CR9 CR10\" citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e–\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e. A previous study on the GRLS population reported a high incidence of haemangiosarcoma (1.1 cases per 100 dog years)\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. As the GRLS cohort is now nearing the end of their lives, an update on haemangiosarcoma incidence and exploration of factors potentially associated with its development in this population is warranted.\u003c/p\u003e\u003cp\u003eThe most consistently reported risk factors associated with canine haemangiosarcoma previously have been breed and age\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. A general increased risk in cancer diagnoses for Golden Retrievers has been reported previously \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e,\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e and this has also been seen specifically for haemangiosarcoma \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Consistent with most other canine cancers, increasing age has been associated with an increased odds of a haemangiosarcoma diagnosis, although this hasn’t been examined specifically in Golden Retrievers\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. For other demographic factors such as sex, neutering or bodyweight, associations with haemangiosarcoma diagnosis are less clear.\u003c/p\u003e\u003cp\u003eThe association between neutering and development of haemangiosarcoma is not fully understood. Increased odds of haemangiosarcoma in neutered dogs have been reported in some previous studies \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. In Golden Retrievers specifically, neutering associations varied with one study finding no difference between neuter statuses \u003csup\u003e\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e and one study reporting increased odds for neutered females (versus entire) but no effect of neutering on male risk \u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Age at neutering and type of neutering (spaying versus castration) may also impact this association\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u003c/sup\u003e. Age at neutering does not appear to have been explored specifically for haemangiosarcoma development in Golden Retrievers, however recommendations for general cancer development have indicated that age of neutering influences cancer development overall \u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Sex has not been consistently associated with haemangiosarcoma diagnosis for any dog breeds. However, an increased risk for haemangiosarcoma in male dogs across populations with a range of dog breeds studied has been seen in two previous studies \u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eWhilst environmental exposures, health and lifestyle factors are commonly explored in human cancer epidemiology studies, these are only beginning to be studied in veterinary oncology \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e. A better understanding of these factors is important as these are modifiable risk factors during a dog’s life, unlike fixed factors such as breed, sex or age. The GRLS design allows for exploration of these modifiable factors. Specific to haemangiosarcoma, one canine study reported higher socioeconomic status and urban areas to be associated with increased odds of diagnosis \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. In human oncology, there is evidence that a wide range of drugs can impact the risk of developing different cancers, with the strongest evidence base appearing to exist for non-steroidal anti-inflammatory (NSAID) drugs having a chemopreventive effect \u003csup\u003e\u003cspan additionalcitationids=\"CR20\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e–\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. However, medication as a protective or promoting factor in cancer development in dogs is not well explored. Certain disorders such as chronic inflammatory conditions have been shown to affect the development and progression of certain cancers in humans and animals, but this has not been reported specifically for haemangiosarcoma in animals \u003csup\u003e\u003cspan additionalcitationids=\"CR22\" citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e–\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Recent work by Ronzani \u0026amp; Hooper (2025) has shown physical activity intensity and frequency to be important predictors in overall cancer development in the GRLS cohort, although due to the modelling type used the directionality of these associations is not clear \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSurvival times for haemangiosarcoma cases in dogs have been previously reported as very short, particularly for visceral forms, with one primary-care study in the UK reporting a median survival time (MST) of 9 days \u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. And even though studies in specialist centres of the subset of dogs referred for extended care report longer MSTs (ranging from 48–172 days), the overall outcome remains poor in most cases \u003csup\u003e\u003cspan additionalcitationids=\"CR27 CR28 CR29\" citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e–\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe current cohort study aimed to investigate the frequency, risk factors, and post-diagnosis survival associated with haemangiosarcoma in a large group of Golden Retrievers followed from puppyhood in the GRLS cohort.\u003c/p\u003e\u003cp\u003eThe study hypothesised that:\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003col\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eOlder age and neutering status are associated with increased odds of haemangiosarcoma diagnosis.\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003cspan\u003e\u003cli\u003e\u003cp\u003eDogs with visceral forms of haemangiosarcoma have a significantly shorter median survival time than cutaneous forms\u003c/p\u003e\u003c/li\u003e\u003c/span\u003e\u003c/ol\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eThe specific objectives of the study were to (i) estimate the frequency of haemangiosarcoma diagnoses, (ii) identify demographic, lifestyle, environmental and clinical variables associated with haemangiosarcoma diagnosis, and (iii) determine post-diagnosis survival times for the overall population and for the different anatomical locations.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003eStudy population and data preparation\u003c/p\u003e\u003cp\u003eBetween 2012 and 2015, 3,044 pedigree Golden Retrievers aged between six months and two years at the time of application were enrolled into the GRLS from across the contiguous United States\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. Further details on enrolment requirements are described elsewhere \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Dogs were excluded from enrolment if they had a prior malignancy diagnosis or a chronic life-threatening condition\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e The entire cohort was included in the current study. A censor date of 31st December 2024 was used to classify each dog on whether they had developed haemangiosarcoma by this point. Dogs diagnosed with haemangiosarcoma by this date were assigned as cases. All remaining dogs that had either died or been withdrawn from the study without being diagnosed with haemangiosarcoma, or dogs that were still alive and in the cohort but had not developed haemangiosarcoma by 31st December 2024, were classified as non-cases. The 31st December 2024 date was also used as their endpoint for calculation of variables with an endpoint required. The haemangiosarcoma case definition required either a) a laboratory-confirmed diagnosis of haemangiosarcoma supported by histological analysis by a board-certified pathologist of samples obtained for diagnostic purposes or during post-mortem examination and b) a non-laboratory confirmed diagnosis based on the clinical opinion of the attending veterinarian that may or may not have been supported by diagnostic imaging, direct mass visualisation or through macroscopic post-mortem examination\u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. Location of mass was determined by either the Morris Animal Foundation GRLS operations team or study veterinarian using clear criteria, with adjudication by a staff veteterinarian\u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Enrolment time was defined from date of enrolment into the study until the endpoint diagnosis for all dogs.\u003c/p\u003e\u003cp\u003eData used in this study were generated from anonymised annual owner and veterinarian questionnaires and necropsy reports \u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. A detailed explanation of the data captured for dogs in the GRLS cohort is explained more fully by Labadie and colleagues (2022). In brief, demographic, lifestyle, environmental exposures and medication usage were captured in annual questionnaires throughout each dog’s life \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. Given that some dogs died or become unenrolled between questionnaires, the most recent study year questionnaire was matched to the final study year of enrolment for that dog. The years of the study were grouped as follows: ‘early life’ was defined as records from study years (SY) 0–2, ‘rest of life’ was SY 3 until end of life and ‘whole life’ was all records available from all study years. As the potential most important carcinogen exposure window/s in a dog’s life have not been determined, we created a further timepoint for variables where values were calculated for the five years prior to endpoint diagnosis (‘5y prior to endpoint’). The data transformation processes and the variables created are explained in Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e and with further detail in the appendices (Table A1).\u003c/p\u003e\u003cdiv class=\"gridtable\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003e\u003cb\u003eAvailable data and transformation process for haemangiosarcoma diagnosis risk factor exploration.\u003c/b\u003e Table indicates what data was available and the transformation process to generate variables for exploration of factors associated with a haemangiosarcoma diagnosis in the Golden Retriever Lifetime Study cohort (n = 3044). SY = study years.\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"4\"\u003e\u003c/colgroup\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eVariable type\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eData transformation process/es\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eVariables created\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003eDemographics\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eAge\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eCases: age at first diagnosis with haemangiosarcoma. Non-cases: age at earliest of 31/12/24 or date of death/unenrolment. Age was categorised according to a recent haemangiosarcoma study by Barry and colleagues \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e, biologically relevant categories and into quartiles\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eAge at final date, age at final date biological, age at final date quartiles\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSex and neutering\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eAge at neutering calculated if neutering date and birth date available. For dogs where neutering date was missing this variable was not calculated\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSex, neuter, sex neuter, age at neutering\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eReproduction\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eReproductive history variables were calculated for dogs prior to neutering dates (if available) or endpoint date. For certain variables, two variations were made – one that excluded certain groups (entire, neutered, male or female dogs), depending on the variable and one that included all dogs and had these excluded categories retained\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMismating ever, reason for neutering,\u003c/p\u003e\u003cp\u003eheats before neutering, number of times bred from before neutering,\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHeight, weight and body condition score (BCS)\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eVeterinarian physical examination data (height, bodyweight, Purina 9-point Body Condition Score (BCS)\u003csup\u003e\u003cspan citationid=\"CR63\" class=\"CitationRef\"\u003e63\u003c/span\u003e\u003c/sup\u003e from the preceding 5 years from endpoint of haemangiosarcoma diagnosis or final record was either averaged across the 5 years or the most common (mode) BCS was reported. BCS was also classified into underweight (0–3), ideal (4–5) and overweight (6–9).\u003c/p\u003e\u003cp\u003eThe weight and height records of dogs whose contribution to the cohort finished before 18 months old were excluded from these variables. Heights and weights were converted from inches and pounds to centimetres and kilograms and outliers more than 1.5 times outside of the interquartile range were removed.\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMean and median adult weight, height and BCS 5y prior, mode and classified BCS 5y prior\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eLifestyle\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003eActivity\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eOwner annual questionnaire data on frequency, duration and intensity of walks and aerobic activities. Questionnaires were modified between study years (SY) 2 and 3 so not all information was captured across all years of study. Owner free-text responses about duration of activity were initially categorised then converted to a numeric category (out of 100) as below to allow for calculations. Activity level was captured consistently across all years as an ordinal variable\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"3\" rowspan=\"4\"\u003e\u003cp\u003eEarly life mean activity duration– daily USDA advised, daily activity 1h, daily activity KC 2h\u003c/p\u003e\u003cp\u003eEarly, rest, whole and 5y prior to endpoint – mean, median, mode, minimum and maximum activity levels\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eDuration free text was classified:\u003c/p\u003e\u003cp\u003e=\u0026lt;30 min, 31-60min, 61-90min, 91-120min, \u0026gt; 120 min\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e=\u0026lt;30 min = 20\u003c/p\u003e\u003cp\u003e31-60min = 40\u003c/p\u003e\u003cp\u003e61-90min = 60\u003c/p\u003e\u003cp\u003e91-120min = 80\u003c/p\u003e\u003cp\u003e\u0026gt; 120 min = 100\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eActivity level: None, little, moderate, very active\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eNone = 0\u003c/p\u003e\u003cp\u003eLittle = 1\u003c/p\u003e\u003cp\u003eModerate = 2\u003c/p\u003e\u003cp\u003eVery active = 3\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eEarly life exercise duration was averaged across study years and classified into United States Department for Agriculture (USDA) \u0026gt; 30 min/day recommendation for dogs and the UK Kennel Club (KC) \u0026gt; 2h/day activity recommendation for Golden Retrievers.\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLifestyle\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eOwner annual questionnaire response on dog’s main lifestyle were grouped into: companion/pet, competitive, breed/showing, working, miscellaneous.\u003c/p\u003e\u003cp\u003eThe mode lifestyle group across all study years and the 5 years prior to endpoint were also determined\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMain lifestyle category whole life and 5y prior to endpoint\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eHome/environment exposures\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSocioeconomic status\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eClinic and owner zip codes linked to county-level Federal Information Processing Standards (FIPS) codes and 2019 county-level Multideprivation Index (MDI) values \u003csup\u003e\u003cspan citationid=\"CR64\" class=\"CitationRef\"\u003e64\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eMDI quintiles for veterinary clinic and owner home\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eHouse features\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eOwner annual questionnaire data on house type, house location (region and urban/rural classification) and household exposures/usage (fuel and water sources, pipe type).\u003c/p\u003e\u003cp\u003eAnnual owner responses then averaged (if binary exposure) or mode value determined for:\u003c/p\u003e\u003cp\u003e- Early life (from SY1-2)\u003c/p\u003e\u003cp\u003e- Adult life/rest of life (SY3 and beyond)\u003c/p\u003e\u003cp\u003e- Whole life (all SY records)\u003c/p\u003e\u003cp\u003e- 5 years prior to endpoint\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eHouse type, urban rural status, region, water source, metal pipes, plastic pipes, heating fuel source, cooking fuel source for early life, rest of life, whole life and 5 y prior to endpoint\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eExposures\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eOwner annual questionnaire on environmental exposures: aerosols, air cleaners, HEPA filters, moth balls, incense/candles, smoke, hours of smoke exposure, weed killer, insecticide and fertiliser. Questionnaire data were also captured on where dogs spent their time or slept (house, garage, outside).\u003c/p\u003e\u003cp\u003eAnnual owner responses on exposures were then aggregated for varying time points to create binary exposure variables for each environmental exposure across these time points:\u003c/p\u003e\u003cp\u003e- Early life (from SY1-2)\u003c/p\u003e\u003cp\u003e- Adult life/rest of life (SY3 and beyond)\u003c/p\u003e\u003cp\u003e- Whole life (all SY records)\u003c/p\u003e\u003cp\u003e- 5 years prior to endpoint\u003c/p\u003e\u003cp\u003eThe most common sleep location was also calculated for these time points.\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eAerosol, air cleaner, HEPA filter, mothballs, incense/candles, weed treatment, insect treatment, fertilizer treatment, exposed to smoke, smoke exposure hours, dog sleep location (garage, house or outside), mode sleep location for early life, rest of life, whole life and 5 y prior to endpoint\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eComorbidities\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"13\" rowspan=\"14\"\u003e\u0026nbsp;\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eAn extensive list of conditions grouped by body system from annual veterinarian questionnaires were grouped in several ways for either 5 years prior to the endpoint or the dog’s whole life. For certain potential chronic inflammatory conditions any single record of the condition was considered to indicate it was a chronic/lifetime condition even if no further mention of it in subsequent study years e.g. Cushing’s disease. For other conditions, the condition must have been recorded for \u0026gt; 1 study year to be considered chronic e.g. heartworm infection.\u003c/p\u003e\u003cp\u003ePresence of cancer (any, benign or malignant) 5 years prior to endpoint was assessed for using annual questionnaires and the malignancy and cause of death dataset. Full list of the conditions corresponding to each grouping in Supplementary Information Table A1.\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"13\" rowspan=\"14\"\u003e\u003cp\u003eChronic inflammation presence, number of comorbidities, type of comorbidities present (parasitism, infectious, immune mediated, cardiovascular, gastrointestinal, orthopaedic, other), presence of cancer (all cancers, benign, malignant), count of cancers (all cancers, benign, malignant) for whole life and 5 y prior to endpoint\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eIntestinal Parasitism\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eTick-Borne Parasitism\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eOther Parasitism\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInfectious\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eOrthopaedic\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eImmune Mediated\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eCardiovascular\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eGastrointestinal\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eInflammatory Other\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eChronic Inflammatory\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAll cancers\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eMalignant cancers\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eBenign cancers\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"4\" nameend=\"c4\" namest=\"c1\"\u003e\u003cp\u003e\u003cb\u003eMedications\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\" morerows=\"14\" rowspan=\"15\"\u003e\u003cp\u003eMedications\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003eMedications prescribed from annual veterinarian and owner questionnaires were grouped into therapeutic indication according to the National Office of Animal health (NOAH )compendium datasheets \u003csup\u003e\u003cspan citationid=\"CR65\" class=\"CitationRef\"\u003e65\u003c/span\u003e\u003c/sup\u003e\u003c/p\u003e\u003cp\u003eDrug class usage was then split into quartiles (or halves where there was not enough variability in prescription numbers between dogs) for lifetime and 5 years (‘5y’) prior to endpoint usage.\u003c/p\u003e\u003cp\u003eUsage of drugs of additional interest (NSAIDs, fenbendazole, steroids) were explored separately beyond drug classes and usage in lifetime and 5 years prior to endpoint were calculated.\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\" morerows=\"14\" rowspan=\"15\"\u003e\u003cp\u003eLifetime usage and 5y prior to endpoint for each therapeutic indication according to the NOAH compendium and also specifically NSAIDs, steroids and fenbendazole\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-histamines\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-inflammatory\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-microbial\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-parasitics\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-septics\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eAnti-virals\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eCardio-respiratory\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eDietary supplements\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eDiuretics\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eEnteric\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eFluid and metabolites\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eHormones and related\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eImmunologicals\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e\u003cp\u003e\u003cb\u003eMiscellaneous\u003c/b\u003e\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/table\u003e\u003c/div\u003e\u003cp\u003eSurvival time was calculated in days from date of diagnosis to date of death, date of withdrawal or to date of censorship of data (31st December 2024). Kaplan–Meier survival (\u003cem\u003ei.e.\u003c/em\u003e, time-to-event) curves were constructed to describe time to death from haemangiosarcoma diagnosis for all cases and then separately for each tumour location group. Haemangiosarcoma locations were classified as: splenic, cardiac, visceral, other or cutaneous. Visceral haemangiosarcomas were those with haemangiosarcoma simultaneously identified at multiple different viscera (primarily cardiac and splenic). A log-rank test was used to verify if the survival time for different haemangiosarcoma locations differed significantly (p \u0026lt; 0.05).\u003c/p\u003e\u003ch2\u003eEthics and consent to participate\u003c/h2\u003e\n\u003cp\u003eEthical approval for the usage of GRLS data for this study was granted by the Royal Veterinary College Social Science Research Ethical Review Board (URN SR2023 - 0175). Ethical approval and consent to participate in the GRLS study was described by Guy and colleagues (2015) \u003csup\u003e1\u003c/sup\u003e. Dog owners signed a consent form to participate in the study and the overall study design was reviewed by the Animal Welfare Advisory Board of Morris Animal Foundation \u003csup\u003e1\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Data analysis\u003c/p\u003e\n\u003cp\u003ePower calculations using the previous proportions of haemangiosarcoma diagnosed in neutered and entire dogs in this cohort (8% and 5% respectively \u003csup\u003e31\u0026nbsp;\u003c/sup\u003e) and an anticipated odds ratio of 1.5, indicated that our study would provide 78% power to detect a difference in haemangiosarcoma incidence between the groups \u003csup\u003e32\u003c/sup\u003e. \u0026nbsp;Assuming a two-sided significance level (\u0026alpha;) of 0.05, we calculated that a sample of 1,059 dogs per group would be required to provide 80% power to detect an odds ratio of 1.5 or greater using a two-sample test of proportions \u003csup\u003e30\u003c/sup\u003e. Lifetime prevalence of haemangiosarcoma was calculated as the number of haemangiosarcoma cases divided by the number of dogs enrolled in the GRLS cohort (n=3044). Incidence rate was calculated per 100 dog-years by dividing the number of haemangiosarcoma cases by the total years of participation of all 3,044 dogs. Years of participation is defined for cases as up until the date of diagnosis and for non-cases until the endpoint of December 31\u003csup\u003est\u003c/sup\u003e 2024. The ninety five percent confidence interval (95% CI) was estimated using exact methods\u003csup\u003e33\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eMultivariable binary logistic regression modelling was used to explore associations between the risk factors and a haemangiosarcoma diagnosis. Variables with liberal association (p\u0026lt;0.2) in initial univariable binary logistic regression modelling were carried forward for exploration in multivariable binary logistic regression modelling \u003csup\u003e32\u003c/sup\u003e. A correlation matrix was built, and Variance Inflation Factor (VIF) was used to assess for collinearity between any liberally significant variables (p\u0026lt;0.2). Substantial collinearity was deemed present where correlation was \u0026ge;0.7 and/or VIF\u0026gt; 5 \u003csup\u003e32\u003c/sup\u003e. For collinear variables, each of the variables were assessed individually in the multivariable model and the variable with smallest model Akaike Information Criteria (AIC) value was retained for further modelling. The multivariable model was built using a manual stepwise forwards construction approach, where variables with the smallest p-value from the likelihood ratio test (LRT) and/or smallest AIC value were included first. Variables were retained in the model if a significant likelihood ratio test (LRT) result indicated the model was improved with variable inclusion. AIC values were used to compare non-nested models that were built with different collinear variables to determine which model balanced goodness of fit and complexity best. Pairwise interactions were assessed for all variables in the final model and interactions were included in the final model where significant (p \u0026lt;0.05). Performance of the final model was assessed through Area Under the Curve (AUC). Model fit to data was assessed by a Hosmer-Lemeshow test, where p \u0026gt;0.05 indicated no evidence that the model poorly fitted the data \u003csup\u003e32\u003c/sup\u003e. Statistical significance was set at the 5% level\u003csup\u003e32\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eAll data cleaning and analyses were performed in R Studio\u003csup\u003e34\u003c/sup\u003e using the following packages: survminer, dynpred, ggsurvfit, ggplot2, finalfit and pROC.\u0026nbsp;\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003eStudy population descriptives\u003c/p\u003e\n\u003cp\u003eThe overall proportion of dogs diagnosed with haemangiosarcoma in the GRLS cohort by December 31st 2024 was 16.1% (95% CI 14.8\u0026ndash;17.4%). The overall incidence rate was 1.69 (95%CI 1.54\u0026ndash;1.85) cases per 100 dog years. There were 490 dogs (both laboratory confirmed and veterinarian suspected) diagnosed with haemangiosarcoma in the GRLS cohort (n\u0026thinsp;=\u0026thinsp;3044) by December 31st, 2024 or at any earlier date of censorship. The median duration enrolled for the overall cohort was 9.66 years (IQR\u0026thinsp;=\u0026thinsp;7.09\u0026ndash;10.35 years, range\u0026thinsp;=\u0026thinsp;0.03\u0026ndash;12.27 years), for the haemangiosarcoma cases was 8.40 years (IQR\u0026thinsp;=\u0026thinsp;7.31\u0026ndash;9.41 years, range\u0026thinsp;=\u0026thinsp;2.92\u0026ndash;12.12 years) and for the non-cases was 9.88 years (IQR\u0026thinsp;=\u0026thinsp;6.99\u0026ndash;10.45 years, range\u0026thinsp;=\u0026thinsp;0.03\u0026ndash;12.27 years). By December 31st, 2024, 1110 dogs (36.4%) were still alive and enrolled in GRLS.\u003c/p\u003e\n\u003cp\u003eThe demographics of the GRLS cohort at enrolment have been described in previous studies\u003csup\u003e\u003cspan class=\"CitationRef\"\u003e2\u003c/span\u003e,\u003cspan class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. At enrolment, the cohort consisted of 1504 females (49.4%) and 1540 males (50.6%) For cases, 213 were female (43.5%) and 277 were male (56.5%). Of the non-cases, 1291 were female (50.5%) and 1263 were male (49.4%). The majority of haemangiosarcoma cases (n\u0026thinsp;=\u0026thinsp;415, 84.7%) and non-cases (n\u0026thinsp;=\u0026thinsp;2027, 79.3%) were neutered by the study cut-off date.\u003c/p\u003e\n\u003cp\u003eThe median age of the cohort at endpoint was 10.68 years (IQR\u0026thinsp;=\u0026thinsp;8.47\u0026ndash;11.65, range\u0026thinsp;=\u0026thinsp;0.73\u0026ndash;14.51 years). The median age of cases at their haemangiosarcoma diagnosis was 9.73 years (IQR\u0026thinsp;=\u0026thinsp;8.68\u0026ndash;10.86 years, range 3.60-13.32 years). The median age of the non-cases at their endpoint (either death from non-haemangiosarcoma causes, study withdrawal or still alive by 31st December 2024) was 10.83 years (IQR\u0026thinsp;=\u0026thinsp;8.28\u0026ndash;11.74, range 0.73\u0026ndash;14.51 years).\u003c/p\u003e\n\u003cp\u003eHaemangiosarcoma case descriptives\u003c/p\u003e\n\u003cp\u003eOf the 490 dogs diagnosed with haemangiosarcoma, 9 dogs had two separate diagnoses of haemangiosarcoma recorded giving 499 separate haemangiosarcoma diagnosis events. Of these 9 dogs, 6 (66.7%) of the initial haemangiosarcoma were reported as cutaneous. The majority of haemangiosarcoma diagnoses overall were definitive as confirmed by a pathologist (n\u0026thinsp;=\u0026thinsp;370, 74.1%), while the remaining had their clinical diagnosis supported through either direct visualisation or imaging of tumour without microscopy (n\u0026thinsp;=\u0026thinsp;74, 14.7%) or on clinical suspicion only (n\u0026thinsp;=\u0026thinsp;56, 11.2%)\u003c/p\u003e\n\u003cp\u003eHaemangiosarcoma diagnosis was classified into the following main locations for each dog: spleen (n\u0026thinsp;=\u0026thinsp;176, 35.2%), cardiac (n\u0026thinsp;=\u0026thinsp;146, 29.2%), visceral (n\u0026thinsp;=\u0026thinsp;91, 18.2%), cutaneous (n\u0026thinsp;=\u0026thinsp;17,3.4%) and other (n\u0026thinsp;=\u0026thinsp;70,14.0%). Due to the aggressive nature of haemangiosarcoma, half of the haemangiosarcoma diagnoses had tumour lesions in multiple organs (n\u0026thinsp;=\u0026thinsp;252, 50.1%). The frequency of all haemangiosarcoma masses identified (n\u0026thinsp;=\u0026thinsp;1053) is reported below in Fig. 1. The main locations were spleen (n\u0026thinsp;=\u0026thinsp;252, 23.9%), cardiac (n\u0026thinsp;=\u0026thinsp;240, 22.8%), lung (n\u0026thinsp;=\u0026thinsp;170, 16.1%) and liver (n\u0026thinsp;=\u0026thinsp;155, 14.7%).\u003c/p\u003e\n\u003cp\u003eRisk factors associated with haemangiosarcoma diagnosis\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eUnivariable logistic regression\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUnivariable binary logistic regression modelling identified 123 variables from the 172 variables as liberally significantly (LRT p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.20) associated with a haemangiosarcoma diagnosis (Table \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e and full univariable output in Supplementary Material Table A2). Due to many of these variables being variations of the same data, such as different ways to divide age at diagnosis, there were 179 pairs of correlated variables from 97 variables. Correlated variables were assessed in non-nested models to decide which variable/s to include in a final model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMultivariable risk factor analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe final model retained 7 variables: age at endpoint, sex, median adult weight, any other cancer diagnosed, a chronic gastrointestinal comorbidity, number of NSAIDs administered and the mode of reported activity level, as recorded 5 years prior to diagnosis (Fig.\u0026nbsp;2).\u003c/p\u003e\n\u003cp\u003eNo significant pairwise interactions were identified between any final model variables. The final model showed no evidence of poor fit to the date (Hosmer and Lemeshow test p\u0026thinsp;=\u0026thinsp;0.09) and showed good discriminatory ability with an ROC value of 0.80. All age groups had significantly reduced odds of haemangiosarcoma diagnosis compared to dogs aged\u0026thinsp;\u0026gt;\u0026thinsp;7-\u0026lt;=9 years of age at endpoint. The odds of diagnosis were lowest in the two extremes of age group: dogs\u0026thinsp;\u0026lt;\u0026thinsp;5 years old (OR\u0026thinsp;=\u0026thinsp;0.02, 95%CI 0.01\u0026ndash;0.05) and dogs\u0026thinsp;\u0026gt;\u0026thinsp;13 years old (OR\u0026thinsp;=\u0026thinsp;0.05, 95%CI 0.01\u0026ndash;0.14). Male dogs had increased odds of haemangiosarcoma diagnosis compared to females (OR\u0026thinsp;=\u0026thinsp;1.27, 95%CI 1.02\u0026ndash;1.57).\u003c/p\u003e\n\u003cp\u003eWhen assessing median adult weight in the 5 years prior to endpoint, dogs weighing 25\u0026ndash;35 kg had over twice the odds of a haemangiosarcoma diagnosis compared to dogs weighing 15\u0026ndash;\u0026lt;25 kg (OR\u0026thinsp;=\u0026thinsp;2.35, 95% CI: 1.79\u0026ndash;3.09). Dogs weighing\u0026thinsp;\u0026gt;\u0026thinsp;35 kg also had increased odds (OR\u0026thinsp;=\u0026thinsp;1.63, 95% CI: 1.05\u0026ndash;2.49).\u003c/p\u003e\n\u003cp\u003eDogs with a recorded non-haemangiosarcoma cancer in the 5 years prior to endpoint had a nearly four-fold reduction in odds of haemangiosarcoma (OR\u0026thinsp;=\u0026thinsp;0.27, 95% CI: 0.21\u0026ndash;0.36), while dogs with a chronic gastrointestinal comorbidity had increased odds (OR\u0026thinsp;=\u0026thinsp;1.79, 95% CI: 1.21\u0026ndash;2.61).\u003c/p\u003e\n\u003cp\u003eDogs with NSAID administration within the 5 years before endpoint had a higher odds of haemangiosarcoma. This was significant across most quartiles of NSAID usage, with the strongest association seen in dogs receiving two presciptions (OR\u0026thinsp;=\u0026thinsp;2.16, 95% CI: 1.56\u0026ndash;2.99). Dogs in the highest usage quartile (4\u0026ndash;11 doses) also had increased odds (OR\u0026thinsp;=\u0026thinsp;1.71, 95% CI: 1.08\u0026ndash;2.65).\u003c/p\u003e\n\u003cp\u003eThe most frequently reported (mode) activity level five years prior to endpoint were also associated with diagnosis. Compared to dogs described as having little activity, those reported as moderately active (OR\u0026thinsp;=\u0026thinsp;0.67, 95% CI: 0.46\u0026ndash;0.99) or very active (OR\u0026thinsp;=\u0026thinsp;0.49, 95% CI: 0.31\u0026ndash;0.76) had significantly reduced odds of haemangiosarcoma.\u003c/p\u003e\n\u003cp\u003eSurvival analysis\u003c/p\u003e\n\u003cp\u003eThe median survival time (MST) from first diagnosis for haemangiosarcoma cases was 0 days (95% CI 0\u0026ndash;0, IQR 0\u0026ndash;10 days, range 0-1693 days) (Table \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e and Fig. 3). The 1-year survival rate was 2.6% (1.5\u0026ndash;4.4%). Only cutaneous haemangiosarcoma had an MST longer than 0 days (MST\u0026thinsp;=\u0026thinsp;19 days, 95%CI 0-479). The survival time of cutaneous haemangiosarcoma cases was significantly longer than visceral cases (Log-rank test p\u0026thinsp;=\u0026thinsp;0.006). Additionally, between specific visceral locations there was a significant difference in survival time between splenic and cardiac (p\u0026thinsp;=\u0026thinsp;\u0026lt;\u0026thinsp;0.001) and splenic only and other visceral locations (p\u0026thinsp;=\u0026thinsp;0.04)\u003c/p\u003e\n\u003ctable id=\"Tab3\" border=\"1\"\u003e\n \u003ccaption language=\"En\"\u003e\n \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\n \u003cdiv class=\"CaptionContent\"\u003e\n \u003cp\u003e\u003cstrong\u003eSummary table of case numbers, median survival time (MST), interquartile range (IQR) and 1-year survival rate for 490 haemangiosarcoma cases (with 502 separate haemangiosarcoma diagnoses) identified from 3044 Golden Retrievers in the Golden Retriever Lifetime Study cohort\u003c/strong\u003e. Haemangiosarcoma (n\u0026thinsp;=\u0026thinsp;502) diagnoses are grouped by location: all, cardiac, splenic, other visceral and cutaneous.\u003c/p\u003e\n \u003c/div\u003e\n \u003c/caption\u003e\n \u003cthead\u003e\n \u003ctr\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eLocation\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eAll\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eSplenic\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCardiac\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eVisceral\u003c/p\u003e\n \u003c/th\u003e\n \u003cth align=\"left\"\u003e\n \u003cp\u003eCutaneous\u003c/p\u003e\n \u003c/th\u003e\n \u003c/tr\u003e\n \u003c/thead\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eNumber of haemangiosarcoma diagnoses (%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e499\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e176 (35.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e146(29.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e91 (18.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e17 (3.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eMST (95% CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0\u0026ndash;0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0\u0026ndash;3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0\u0026ndash;0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0 (0\u0026ndash;0)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e19 (0-479)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003eIQR days (range, days)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u0026ndash;10 (0-1693)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u0026ndash;54 (0-774)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u0026ndash;0 (0-1302)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0\u0026ndash;0 (0-365)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e0-165 (0-1543)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1-year survival rate (%) (95%CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.6% (1.5\u0026ndash;4.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e2.3% (0.9\u0026ndash;6.1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.3% (0.3\u0026ndash;5.2)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e1.01% (0.18\u0026ndash;5.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd align=\"left\"\u003e\n \u003cp\u003e22% (9.4\u0026ndash;53)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e"},{"header":"Discussion","content":"\u003cp\u003eHaemangiosarcoma is a highly aggressive cancer with poor prognosis that commonly affects middle-aged to older dogs, with Golden Retrievers being at increased risk\u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. Despite its clinical importance, the underlying risk factors for haemangiosarcoma development in dogs remain poorly understood. Leveraging the depth and longitudinal design of the Golden Retriever Lifetime Study (GRLS), this study aimed to determine the frequency of haemangiosarcoma diagnoses, identify associated risk factors, and characterise post-diagnosis survival outcomes in Golden Retrievers in the US. We hypothesised that older age and neutering status would be associated with increased odds of haemangiosarcoma diagnosis, and that dogs with visceral forms of haemangiosarcoma would have significantly shorter median survival times than those with cutaneous forms. Our findings confirm a high burden of haemangiosarcoma within the breed, with the disease diagnosed in 16.1% of the cohort. Multivariable modelling identified older age, male status, higher median adult weight, a history of chronic gastrointestinal disease, a non-haemangiosarcoma cancer diagnosis, NSAID administration, and reduced activity levels as significant risk factors for haemangiosarcoma diagnosis. For the non-demographic variables it was considered that any associations detected were exploratory and hypothesis generating for future work. Additionally, although cutaneous haemangiosarcoma was associated with longer survival than visceral forms, prognosis was poor across all haemangiosarcoma locations, with very short median survival times (MST 0 days, 95% CI 0\u0026ndash;0) following diagnosis regardless of anatomical site.\u003c/p\u003e\u003cp\u003eFactors associated with a haemangiosarcoma diagnosis\u003c/p\u003e\u003cp\u003e\u003cb\u003eAge\u003c/b\u003e\u003c/p\u003e\u003cp\u003eConsistent with previous studies, the current study found a peak diagnosis of haemangiosarcoma in middle aged to older dogs. Dogs aged between 7\u0026ndash;9 years had the highest odds of diagnosis with haemangiosarcoma in the current study. In comparison, dogs under five years and dogs aged over 13 years were 20 times less likely to be diagnosed with haemangiosarcoma. The median age at first diagnosis was 9.73 years, which aligns with previous reports in the literature across various breeds that range from 7.9\u0026ndash;10.7 years) \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e,\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e,\u003cspan additionalcitationids=\"CR36\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. Previous studies examining age across a range of breeds identified an older, but comparable peak to that seen in this present study \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. This difference could be explained by the fact that these studies included all dog breeds in their analyses whereas the current study focussed on one, large breed (Golden Retrievers). Large breed dogs have been shown to have shorter lifespans than most small and medium breeds, which means that their peak risk of haemangiosarcoma development and diagnosis should predictably be chronologically earlier than smaller dogs \u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. In addition, as this is an enrolled, optional, lifetime study, this cohort might select for more motivated owners who access veterinary care more frequently which could lead to an earlier diagnosis.\u003c/p\u003e\u003cp\u003e\u003cb\u003eSex and neutering\u003c/b\u003e\u003c/p\u003e\u003cp\u003eIn the present study, males had 1.3 times increased odds of haemangiosarcoma diagnosis compared to females). An association between male sex and haemangiosarcoma diagnosis has been reported previously\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. Additionally, one study has shown male sex to be associated with increased hazard of death from haemangiosarcoma, across a multi- breed population\u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. In the current study, it was hypothesised that neutering and/or age at neutering would be important risk factors for the diagnosis of haemangiosarcoma. Several other studies have explored sex and neutering associations with haemangiosarcoma in conjunction, with most studies finding both neutered and entire males to be at increased odds of haemangiosarcoma diagnosis versus entire females \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e. This same association was seen in univariable analyses in this study, but the sex variable was deemed to have better fit for the final model than sex-neuter status, age at neutering or neutering variables so these were not included in the final regression model \u003csup\u003e\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e,\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. The reasons behind the sex association identified here are unclear and likely multifactorial and encompass the effects of androgens on cancer biology, immunologic differences and lifestyle factors \u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e,\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e. While angiosarcoma in people is not typically considered a hormone-dependent cancer, emerging evidence suggests that sex hormones can influence the tumour microenvironment and potentially affect immune surveillance\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e. Additionally, due to the high frequency of neutering in most dog populations (80% in the GRLS cohort) there may be other, sex-linked genetic factors associated to the increased odds of diagnosis seen in male dogs here \u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eWeight\u003c/b\u003e\u003c/p\u003e\u003cp\u003eHigher median adult weight was associated with increased odds of haemangiosarcoma diagnosis in this study. Overweight and obese status have been linked to increased cancer risk in both human and veterinary studies, potentially due to the low-grade chronic inflammation associated with excess body fat \u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e. However, interpreting weight in isolation is challenging without concurrent body condition score (BCS) and sex data, as these are important for determining whether a dog is truly overweight. According to American Kennel Club breed standards, the normal weight range for adult Golden Retrievers is 25\u0026ndash;29 kg for females and 29\u0026ndash;32 kg for males \u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e, meaning that many dogs in the 25\u0026ndash;35 kg group could fall within a normal weight range depending on sex. An interaction between weight and sex was tested but was not statistically significant. Although BCS may offer better insight into adiposity status, it was not retained in the final multivariable model. Previous studies have found higher weight to be associated with increased odds of haemangiosarcoma and when considered in conjunction with these results in a single breed might indicate increased risk in larger animals rather than an effect of overweight/obesity status \u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eComorbidities\u003c/b\u003e\u003c/p\u003e\u003cp\u003eHaving had a previous separate cancer diagnosis within the five years prior to endpoint was associated with reduced odds of a haemangiosarcoma diagnosis. We postulate that this could reflect survival bias, where those dogs diagnosed with an earlier cancer are more likely to have died or be euthanised from that cancer and therefore not live long enough to develop a haemangiosarcoma. Alternatively, this may just reflect that the cases were younger and had less prior cancers, as the five years prior to haemangiosarcoma diagnosis were likely earlier years in their life, than non-cases where the five years was largely prior to the end of our censoring point. In humans, treatment with chemotherapy and radiotherapy in cancer survivors may increase the risk of second cancers during their lifetime \u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e,\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e\u003c/sup\u003e, it is also possible that this could have a protective or immunomodulatory effect on the development of haemangiosarcoma in the short team, given the shorter life-span seen in dogs, although there is no evidence for this phenomenon in veterinary oncology.\u003c/p\u003e\u003cp\u003eChronic inflammatory conditions have been suggested to play a role in the development of certain cancers due to continued, low-grade inflammation creating favourable conditions for carcinogenesis and tumour metastasis \u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e,\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. However, when all chronic inflammatory conditions in the five years prior to diagnosis were grouped together, this variable showed no significant effect in the current study. One reason for this variable not being significant is that potentially not all chronic inflammatory conditions are associated with cancer development. For example, chronic inflammatory conditions associated with lymphoedema development are risk factors for angiosarcoma development in humans\u003csup\u003e\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e\u003c/sup\u003e. The only comorbidity retained in our final multivariable model was the presence of a gastrointestinal comorbidity which was associated with increased odds of a haemangiosarcoma diagnosis. Chronic gastrointestinal conditions have not previously been associated with haemangiosarcoma specifically; however, a range of gastrointestinal conditions have been associated with other cancers in veterinary and human studies \u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMedications\u003c/b\u003e\u003c/p\u003e\u003cp\u003eAn association between increased NSAID usage and higher odds of haemangiosarcoma diagnosis was seen in the current study. This association was potentially unexpected as some human studies have reported that prior NSAID use may reduce the risk of the development of some cancers and postulated that this could be mediated by enhancing DNA damage repair, or inhibition of tumour proliferation and invasion via COX-dependent or/and -independent pathways\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e. In veterinary oncology, certain NSAIDs have been shown to have anti-tumour properties, for example the use of piroxicam in dogs with transitional cell carcinoma of the bladder \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e,\u003cspan citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u003c/sup\u003e. Deracoxib has been suggested as beneficial in dogs with stage III haemangiosarcoma in one study, although a control group was not included \u003csup\u003e\u003cspan citationid=\"CR50\" class=\"CitationRef\"\u003e50\u003c/span\u003e\u003c/sup\u003e. There are several possible reasons why the opposite NSAID relationship was seen in the current study. Firstly, veterinary studies have looked at NSAIDs as a therapy once a cancer is diagnosed rather than a protective effect, whereas the current study examined factors associated specifically with the development of haemangiosarcoma \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e.There is also evidence to suggest that COX-2 expression may differ between haemangiosarcoma locations; while expression was not seen in splenic forms, frequent but weak expression is present in cutaneous forms \u003csup\u003e\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e,\u003cspan citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u003c/sup\u003e. Furthermore, the type of cyclo-oxygenase (COX) inhibition by different types of NSAIDs varies and could have varying effects on tumour initiation or development \u003csup\u003e\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e. Additionally, looking further into NSAID prescribing timings is warranted, for example NSAID prescription within a few months of haemangiosarcoma diagnosis could have been for presenting signs associated with an undiagnosed haemangiosarcoma mass. The NSAID association seen here should be explored further to determine if different NSAIDs have differing associations and whether dosage, duration of course/s and frequency of usage affects outcomes.\u003c/p\u003e\u003cp\u003e\u003cb\u003eActivity level\u003c/b\u003e\u003c/p\u003e\u003cp\u003eDogs with higher owner-reported activity levels (moderate or very active) had reduced odds of haemangiosarcoma diagnosis in the current study. An association between more physical activity and reduced risk of the development of certain cancers has been previously reported in human oncology \u003csup\u003e\u003cspan citationid=\"CR53\" class=\"CitationRef\"\u003e53\u003c/span\u003e\u003c/sup\u003e. A recent GRLS cohort study identified activity as an important predictor in cancer diagnosis \u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Ronzani and colleagues (2025) found that frequency and intensity of exercise had a greater impact on diagnosis, than other metrics such as exercise type. Whilst our study did not examine the full range of activity variables that Ronzani and colleagues (2025) explored; it is of interest to see an activity association with a specific cancer (haemangiosarcoma) for this cohort. However, this relationship warrants further investigation, as it remains unclear whether the protective effect of higher activity levels is due to physical activity itself or related factors such as lower bodyweight or reduced chronic inflammation\u003csup\u003e\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e,4647,\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eSurvival time after haemangiosarcoma diagnosis\u003c/p\u003e\u003cp\u003eMedian survival time for Golden Retrievers with haemangiosarcoma was short (overall MST 0 days, 95%CI 0\u0026ndash;0 days), likely underscoring the aggressive nature of this malignancy. However, it is unclear how much effect the perceived grave prognosis of haemangiosarcoma affects euthanasia choices and whether this becomes a self-fulfilling prophecy. Even though cutaneous haemangiosarcoma can have a more favourable prognosis and the potential for curative resection, the MST was still very short (19 days, 95% CI 0-479) \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, although survival time was significantly increased versus visceral haemangiosarcomas. The current study reported a much shorter MST compared to many of the previous cutaneous haemangiosarcoma studies reported in the literature (MST: 172\u0026ndash;1189 days) (Hargis et al., 1992; N\u0026oacute;brega et al., 2019; Shiu et al., 2011; Ward et al., 1994). In addition, previous studies of various visceral haemangiosarcoma locations (cardiac and splenic), reported MST ranged from 7-259 days which is longer than the MST of 0 days seen for each visceral locations in the current study. \u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e,\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e\u003c/sup\u003e. The longer MSTs previously reported may reflect referral bias whereby most of these earlier studies included referral cases which could exhibit selection bias for more clinically stable case populations and early disease stage or towards owners highly committed to trialling treatment \u003csup\u003e\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e\u003c/sup\u003e. A recent, primary-care study of UK dogs reported survival times more in line with what was seen in this study (splenic MST: 4 days, cardiac MST: 0 days) \u003csup\u003e\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Given that the GRLS cohort dogs are largely primary-care practice based, the aforementioned primary-care population study may be a more appropriate population to compare survival times to. However, it is also possible that the very short MSTs in this population were due to Golden Retrievers being more prone to develop more aggressive haemangiosarcoma phenotype than the wider dog population, and the higher prevalence of cardiac forms which are typically associated with a poor prognosis regardless of the treatment pursued\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e,\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e. Therefore, the current results derived just from Golden Retrievers should be extrapolated to other dog brees with caution.\u003c/p\u003e\u003cp\u003eStrengths and limitations\u003c/p\u003e\u003cp\u003eAs these analyses were undertaken within the Golden Retriever Lifetime Study (GRLS), a prospective longitudinal cohort, there are several inherent strengths and limitations to consider. A key strength is the repeated, annual collection of data across a wide range of variables such as clinical examination findings, activity levels, comorbidities, and medication usage \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e. This facilitates the exploration of time-specific (\u0026lsquo;time window\u0026rsquo;) associations with haemangiosarcoma, while also reducing recall bias compared to study designs reliant on lifetime recall at the point of diagnosis. Additionally, given that the influence of environment and lifestyle on cancer development remains poorly understood in veterinary medicine, this study provides a unique opportunity for hypothesis generation in this area.\u003c/p\u003e\u003cp\u003eHowever, there are several important limitations. Much of the data was manually collected via owner and veterinarian questionnaires and only at annual time points, which introduces the potential for recall inaccuracies or misreporting across each one-year window. For example, estimating the frequency of routine medication use or characterising a dog's activity level over a year may be imprecise. Additionally, one criterion for enrolment was pedigree status, which might mean these Golden Retrievers are more inbred than non-pedigree Golden Retrievers. Inbreeding could potentially affect cancer risk; although this elevated risk hasn\u0026rsquo;t been demonstrated in Golden Retrievers, it has been observed in some other breeds \u003csup\u003e\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e\u003c/sup\u003e. Furthermore, the cohort size (~\u0026thinsp;3,000 dogs) may be underpowered to detect environmental associations of small magnitude but potentially significant biological relevance. For example, a large electronic health record study identified an association between herbicide exposure and canine lymphoma diagnosis (Schofield et al., 2019), an association not replicated in smaller studies. This discrepancy is likely at least partly attributable to limited statistical power; Schofield and colleagues estimated that approximately 200 cases derived from a population of 400,000 were required to detect such an effect with sufficient power \u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e,\u003cspan citationid=\"CR60\" class=\"CitationRef\"\u003e60\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFinally, the optimal exposure window for carcinogen-related cancer development remains uncertain. While this study examined multiple windows \u0026mdash; including early life, lifetime, and the five years preceding diagnosis/end-point \u0026mdash; it is possible that key exposures occurred outside these periods. For instance, in utero or neonatal exposures (prior to enrolment in GRLS from six months of age) may be relevant but were not captured in this dataset. One older study reported perinatal radiation exposure in Beagles increased the risk of fatal malignancies \u003csup\u003e\u003cspan citationid=\"CR61\" class=\"CitationRef\"\u003e61\u003c/span\u003e\u003c/sup\u003e. This evidence is lacking for other potential carcinogens and other cancers.\u003c/p\u003e\u003cp\u003eFuture work\u003c/p\u003e\u003cp\u003eWhilst several of the factors associated with haemangiosarcoma diagnosis in this study were anticipated (age and weight), others, namely the associations between haemangiosarcoma diagnosis and NSAID usage and previous cancer diagnosis, were novel findings. As this was an exploratory study, no adjustment was made for multiple comparisons. The analyses were intended to identify potential associations and generate hypotheses for future confirmatory studies. As such, findings should be interpreted with caution given the increased risk of Type I error due to multiple testing. Future studies should be undertaken to explore these novel associations further. For example, causal study methods such as target trial emulations could utilise the observational data in the GRLS and attempt to explore whether there is any causality between NSAID usage and haemangiosarcoma development \u003csup\u003e\u003cspan citationid=\"CR62\" class=\"CitationRef\"\u003e62\u003c/span\u003e\u003c/sup\u003e. Additionally, further exploration of the association between gastrointestinal comorbidities and cancer development should be undertaken. Other future studies should explore whether the associations identified here are unique to haemangiosarcoma diagnosis or are also recognised in other cancers in this cohort.\u003c/p\u003e"},{"header":"Conclusions","content":"\u003cp\u003eThis study provides valuable preliminary insights into the epidemiology of haemangiosarcoma within a large, longitudinal cohort of Golden Retrievers, identifying a range of demographic, lifestyle, and health-related factors potentially associated with diagnosis. The very poor survival outcomes observed highlight the ongoing clinical challenges posed by this disease. While the study confirmed some anticipated associations, such as those with age, it also revealed several unanticipated findings, including associations with NSAID usage and prior cancer diagnoses. These novel associations should be interpreted with caution and considered hypothesis-generating, warranting further investigation in independent datasets to explore potential underlying mechanisms and confirm their validity.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eFunding\u003c/p\u003e\n\u003cp\u003eThe Golden Retriever Lifetime Study and this manuscript were made possible through financial support provided by the Morris Family Foundation, the Mark \u0026amp; Bette Morris Family Foundation, VCA, the V Foundation, Blue Buffalo Company, Petco Love, Zoetis, Antech Inc., Elanco, the Purina Institute, Orvis, the Golden Retriever Foundation, the Hadley and Marion Stuart Foundation, Mars Veterinary, generous private donors and the Flint Animal Cancer Center at Colorado State University. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eGuy MK, Page RL, Jensen WA, et al. The golden retriever lifetime study: Establishing an observational cohort study with translational relevance for human health. \u003cem\u003ePhilosophical Transactions of the Royal Society B: Biological Sciences\u003c/em\u003e. 2015;370(1673). doi:10.1098/rstb.2014.0230\u003c/li\u003e\n\u003cli\u003eLabadie J, Swafford B, DePena M, Tietje K, Page R, Patterson-Kane J. Cohort profile: The Golden Retriever Lifetime Study (GRLS). \u003cem\u003ePLoS One\u003c/em\u003e. 2022;17(6 June). doi:10.1371/journal.pone.0269425\u003c/li\u003e\n\u003cli\u003ePugh CA, Bronsvoort BM de C, Handel IG, Summers KM, Clements DN. What can cohort studies in the dog tell us? \u003cem\u003eCanine Genet Epidemiol\u003c/em\u003e. 2014;1(1):5. doi:10.1186/2052-6687-1-5\u003c/li\u003e\n\u003cli\u003eBurns PB, Rohrich RJ, Chung KC. The levels of evidence and their role in evidence-based medicine. \u003cem\u003ePlast Reconstr Surg\u003c/em\u003e. 2011;128(1):305-310. doi:10.1097/PRS.0b013e318219c171\u003c/li\u003e\n\u003cli\u003eDe Nardi AB, de Oliveira Massoco Salles Gomes C, Fonseca-Alves CE, et al. Diagnosis, Prognosis, and Treatment of Canine Hemangiosarcoma: A Review Based on a Consensus Organized by the Brazilian Association of Veterinary Oncology, ABROVET. \u003cem\u003eCancers (Basel)\u003c/em\u003e. 2023;15(7). doi:10.3390/cancers15072025\u003c/li\u003e\n\u003cli\u003eWithrow SJ, Vail DM, Page RL. \u003cem\u003eWithrow \u0026amp; MacEwen\u0026rsquo;s Small Animal Clinical Oncology.\u003c/em\u003e 5th ed. Elsevier Saunders; 2013.\u003c/li\u003e\n\u003cli\u003eRobinson KL, Bryan ME, Atkinson ES, Keeler MR, Hahn AW, Bryan JN. \u003cem\u003eNeutering Is Associated with Developing Hemangiosarcoma in Dogs in the Veterinary Medical Database: An Age and Time-Period Matched Case-Control Study (1964-2003)\u003c/em\u003e. Vol 61.; 2020.\u003c/li\u003e\n\u003cli\u003eDorea F,, Brodbelt D. Society for Veterinary Epidemiology and Preventive Medicine Conference Proceedings. In: Dorea F,, Brodbelt D, eds. \u003cem\u003eSociety for Veterinary Epidemiology and Preventive Medicine Conference Proceedings\u003c/em\u003e. ; 2025:149-160.\u003c/li\u003e\n\u003cli\u003eDobson JM, Samuel S, Milstein H, Rogers K, Wood JLN. Canine neoplasia in the UK: Estimates of incidence rates from a population of insured dogs. \u003cem\u003eJournal of Small Animal Practice\u003c/em\u003e. 2002;43(6):240-246. doi:10.1111/j.1748-5827.2002.tb00066.x\u003c/li\u003e\n\u003cli\u003eCarnio A, Eleni C, Cocumelli C, et al. Evaluation of intrinsic and extrinsic risk factors for dog visceral hemangiosarcoma: A retrospective case-control study register-based in Lazio region, Italy. \u003cem\u003ePrev Vet Med\u003c/em\u003e. 2020;181. doi:10.1016/j.prevetmed.2020.105074\u003c/li\u003e\n\u003cli\u003eSrebernik N, Appleby E. Breed prevalence and sites of haemangioma and haemangiosarcoma in dogs. \u003cem\u003eVeterinary Record\u003c/em\u003e. 1991;129(18):408-409. doi:10.1136/vr.129.18.408\u003c/li\u003e\n\u003cli\u003eHillman A, Swafford B, Delavenne C, Fieten H, Boerkamp K, Tietje K. Descriptive analysis of haemangiosarcoma occurrence in dogs enrolled in the Golden Retriever lifetime study. \u003cem\u003eVet Comp Oncol\u003c/em\u003e. Published online August 27, 2023. doi:10.1111/vco.12933\u003c/li\u003e\n\u003cli\u003eFleming JM, Creevy KE, Promislow DEL. Mortality in North American Dogs from 1984 to 2004: An Investigation into Age-, Size-, and Breed-Related Causes of Death. \u003cem\u003eJ Vet Intern Med\u003c/em\u003e. 2011;25(2):187-198. doi:10.1111/j.1939-1676.2011.0695.x\u003c/li\u003e\n\u003cli\u003eTorres de la Riva G, Hart BL, Farver TB, et al. Neutering Dogs: Effects on Joint Disorders and Cancers in Golden Retrievers. \u003cem\u003ePLoS One\u003c/em\u003e. 2013;8(2). doi:10.1371/journal.pone.0055937\u003c/li\u003e\n\u003cli\u003eKent MS, Burton JH, Dank G, Bannasch DL, Rebhun RB. Association of cancer-related mortality, age and gonadectomy in golden retriever dogs at a veterinary academic center (1989-2016). \u003cem\u003ePLoS One\u003c/em\u003e. 2018;13(2). doi:10.1371/journal.pone.0192578\u003c/li\u003e\n\u003cli\u003eHart BL, Hart LA, Thigpen AP, Willits NH. Assisting Decision-Making on Age of Neutering for 35 Breeds of Dogs: Associated Joint Disorders, Cancers, and Urinary Incontinence. \u003cem\u003eFront Vet Sci\u003c/em\u003e. 2020;7. doi:10.3389/fvets.2020.00388\u003c/li\u003e\n\u003cli\u003eWise CF, Breen M, Stapleton HM. Canine on the Couch: The New Canary in the Coal Mine for Environmental Health Research. \u003cem\u003eEnvironment and Health\u003c/em\u003e. 2024;2(8):517-529. doi:10.1021/envhealth.4c00029\u003c/li\u003e\n\u003cli\u003eOwada K, Nicholls E, Soares Magalh\u0026atilde;es RJ, Palmieri C. Environmental exposure and cancer occurrence in dogs: a critical appraisal of evidence. \u003cem\u003eRes Vet Sci\u003c/em\u003e. 2025;184. doi:10.1016/j.rvsc.2024.105517\u003c/li\u003e\n\u003cli\u003eHarris R, Beebe-Donk J, Doss H, Doss D. Aspirin, ibuprofen, and other non-steroidal anti-inflammatory drugs in cancer prevention: A critical review of non-selective COX-2 blockade (Review). \u003cem\u003eOncol Rep\u003c/em\u003e. Published online April 1, 2005. doi:10.3892/or.13.4.559\u003c/li\u003e\n\u003cli\u003ePotteg\u0026aring;rd A, Friis S, Christensen R de P, Habel LA, Gagne JJ, Hallas J. Identification of Associations Between Prescribed Medications and Cancer: A Nationwide Screening Study. \u003cem\u003eEBioMedicine\u003c/em\u003e. 2016;7:73-79. doi:10.1016/j.ebiom.2016.03.018\u003c/li\u003e\n\u003cli\u003eRaposo TP, Beir\u0026atilde;o BCB, Pang LY, Queiroga FL, Argyle DJ. Inflammation and cancer: Till death tears them apart. \u003cem\u003eVeterinary Journal\u003c/em\u003e. 2015;205(2):161-174. doi:10.1016/j.tvjl.2015.04.015\u003c/li\u003e\n\u003cli\u003eMulthoff G, Molls M, Radons J. Chronic inflammation in cancer development. \u003cem\u003eFront Immunol\u003c/em\u003e. 2012;2(JAN). doi:10.3389/fimmu.2011.00098\u003c/li\u003e\n\u003cli\u003eMorrison WB. Inflammation and cancer: A comparative view. \u003cem\u003eJ Vet Intern Med\u003c/em\u003e. 2012;26(1):18-31. doi:10.1111/j.1939-1676.2011.00836.x\u003c/li\u003e\n\u003cli\u003eRonzani D, Hooper SE. Physical activity predictors of cancer in Golden Retrievers: it\u0026rsquo;s about frequency and intensity, not type. \u003cem\u003eVeterinary Oncology\u003c/em\u003e. 2025;2(1):11. doi:10.1186/s44356-025-00024-5\u003c/li\u003e\n\u003cli\u003eTaylor C, Barry GJ, O\u0026rsquo;Neill DG, et al. Survival time and prognostic factors in dogs clinically diagnosed with haemangiosarcoma in UK first opinion practice. Clegg, S, ed. \u003cem\u003ePLoS One\u003c/em\u003e. 2025;20(6):e0316066. doi:10.1371/journal.pone.0316066\u003c/li\u003e\n\u003cli\u003eWendelburg KM, Price LL, Burgess KE, Lyons JA, Lew FH,, Berg J. Survival time of dogs with splenic hemangiosarcoma treated by splenectomy with or without adjuvant chemotherapy: 208 cases (2001\u0026ndash;2012). \u003cem\u003eJAVMA\u003c/em\u003e. 2015;247(4):393.\u003c/li\u003e\n\u003cli\u003eBray JP, Orbell G, Cave N, Munday JS. Does thalidomide prolong survival in dogs with splenic haemangiosarcoma? \u003cem\u003eJournal of Small Animal Practice\u003c/em\u003e. 2018;59(2):85-91. doi:10.1111/jsap.12796\u003c/li\u003e\n\u003cli\u003eMullin CM, Arkans MA, Sammarco CD, et al. Doxorubicin chemotherapy for presumptive cardiac hemangiosarcoma in dogs. \u003cem\u003eVet Comp Oncol\u003c/em\u003e. 2016;14(4):e171-e183. doi:10.1111/vco.12131\u003c/li\u003e\n\u003cli\u003eStory AL, Wavreille V, Abrams B, Egan A, Cray M, Selmic LE. Outcomes of 43 small breed dogs treated for splenic hemangiosarcoma. \u003cem\u003eVeterinary Surgery\u003c/em\u003e. 2020;49(6):1154-1163. doi:10.1111/vsu.13470\u003c/li\u003e\n\u003cli\u003eTaylor C, Barry GJ, O\u0026rsquo;Neill DG, et al. Survival time and prognostic factors in dogs clinically diagnosed with haemangiosarcoma in UK first opinion practice. Clegg, S, ed. \u003cem\u003ePLoS One\u003c/em\u003e. 2025;20(6):e0316066. doi:10.1371/journal.pone.0316066\u003c/li\u003e\n\u003cli\u003eHillman A, Swafford B, Delavenne C, Fieten H, Boerkamp K, Tietje K. Descriptive analysis of haemangiosarcoma occurrence in dogs enrolled in the Golden Retriever lifetime study. \u003cem\u003eVet Comp Oncol\u003c/em\u003e. 2023;21(4):700-708. doi:10.1111/vco.12933\u003c/li\u003e\n\u003cli\u003eDohoo I, Martin W, Stryhn H. \u003cem\u003eVeterinary Epidemiological Research\u003c/em\u003e. VER Inc.; 2009.\u003c/li\u003e\n\u003cli\u003eKirkwood BR, Sterne JAC. \u003cem\u003eEssential Medical Statistics\u003c/em\u003e. 2nd ed. Blackwell; 2003. doi:10.1192/bjp.112.483.211-a\u003c/li\u003e\n\u003cli\u003eR Core Team. RStudio. Published online 2019.\u003c/li\u003e\n\u003cli\u003eDavies O, Taylor AJ. Refining the \u0026ldquo;double two-thirds\u0026rdquo; rule: Genotype-based breed grouping and clinical presentation help predict the diagnosis of canine splenic mass lesions in 288 dogs. \u003cem\u003eVet Comp Oncol\u003c/em\u003e. 2020;18(4):548-558. doi:10.1111/vco.12574\u003c/li\u003e\n\u003cli\u003eYamamoto S, Hoshi K, Hirakawa A, Chimura S, Kobayashi M, Machida N. Epidemiological, clinical and pathological features of primary cardiac hemangiosarcoma in dogs: A review of 51 cases. \u003cem\u003eJournal of Veterinary Medical Science\u003c/em\u003e. 2013;75(11):1433-1441. doi:10.1292/jvms.13-0064\u003c/li\u003e\n\u003cli\u003eSzivek A, Burns RE, Gericota B, et al. Clinical outcome in 94 cases of dermal haemangiosarcoma in dogs treated with surgical excision: 1993-2007. \u003cem\u003eVet Comp Oncol\u003c/em\u003e. 2012;10(1):65-73. doi:10.1111/j.1476-5829.2011.00282.x\u003c/li\u003e\n\u003cli\u003eMcMillan KM, Bielby J, Williams CL, Upjohn MM, Casey RA, Christley RM. Longevity of companion dog breeds: those at risk from early death. \u003cem\u003eSci Rep\u003c/em\u003e. 2024;14(1). doi:10.1038/s41598-023-50458-w\u003c/li\u003e\n\u003cli\u003eLewis TW, Wiles BM, Llewellyn-Zaidi AM, Evans KM, O\u0026rsquo;Neill DG. Longevity and mortality in Kennel Club registered dog breeds in the UK in 2014. \u003cem\u003eCanine Genet Epidemiol\u003c/em\u003e. 2018;5(1). doi:10.1186/s40575-018-0066-8\u003c/li\u003e\n\u003cli\u003eCook MB, McGlynn KA, Devesa SS, Freedman ND, Anderson WF. Sex disparities in cancer mortality and survival. \u003cem\u003eCancer Epidemiology Biomarkers and Prevention\u003c/em\u003e. 2011;20(8):1629-1637. doi:10.1158/1055-9965.EPI-11-0246\u003c/li\u003e\n\u003cli\u003eRadkiewicz C, Johansson ALV, Dickman PW, Lambe M, Edgren G. Sex differences in cancer risk and survival: A Swedish cohort study. \u003cem\u003eEur J Cancer\u003c/em\u003e. 2017;84:130-140. doi:10.1016/j.ejca.2017.07.013\u003c/li\u003e\n\u003cli\u003e\u0026Ouml;zdemir BC, Dotto GP. Sex Hormones and Anticancer Immunity. \u003cem\u003eClinical Cancer Research\u003c/em\u003e. 2019;25(15):4603-4610. doi:10.1158/1078-0432.CCR-19-0137\u003c/li\u003e\n\u003cli\u003eMarchi PH, Vendramini THA, Perini MP, et al. Obesity, inflammation, and cancer in dogs: Review and perspectives. \u003cem\u003eFront Vet Sci\u003c/em\u003e. 2022;9.\u003c/li\u003e\n\u003cli\u003eAmerican Kennel Club. Official Standard for the Golden Retriever. 1981. Accessed June 3, 2025. https://images.akc.org/pdf/breeds/standards/GoldenRetriever.pdf\u003c/li\u003e\n\u003cli\u003eDracham CB, Shankar A, Madan R. Radiation induced secondary malignancies: A review article. \u003cem\u003eRadiat Oncol J\u003c/em\u003e. 2018;36(2):85-94. doi:10.3857/roj.2018.00290\u003c/li\u003e\n\u003cli\u003eMorton LM, Dores GM, Schonfeld SJ, et al. Association of Chemotherapy for Solid Tumors With Development of Therapy-Related Myelodysplastic Syndrome or Acute Myeloid Leukemia in the Modern Era. \u003cem\u003eJAMA Oncol\u003c/em\u003e. 2019;5(3):318. doi:10.1001/jamaoncol.2018.5625\u003c/li\u003e\n\u003cli\u003eCao J, Wang J, He C, Fang M. \u003cem\u003eAngiosarcoma: A Review of Diagnosis and Current Treatment\u003c/em\u003e. Vol 9.; 2019. www.ajcr.us/\u003c/li\u003e\n\u003cli\u003eCuzick J, Otto F, Baron JA, et al. Aspirin and non-steroidal anti-inflammatory drugs for cancer prevention: an international consensus statement. \u003cem\u003eLancet Oncol\u003c/em\u003e. 2009;10(5):501-507. doi:10.1016/S1470-2045(09)70035-X\u003c/li\u003e\n\u003cli\u003eKnapp DW, Richardson RC, Chan TCK, et al. Piroxicam Therapy in 34 Dogs With Transitional Cell Carcinoma of the Urinary Bladder. \u003cem\u003eJ Vet Intern Med\u003c/em\u003e. 1994;8(4):273-278. doi:10.1111/j.1939-1676.1994.tb03232.x\u003c/li\u003e\n\u003cli\u003eKahn SA, Mullin CM, de Lorimier LP, et al. Doxorubicin and deracoxib adjuvant therapy for canine splenic hemangiosarcoma: a pilot study. \u003cem\u003eCan Vet J\u003c/em\u003e. 2013;54(3):237-242.\u003c/li\u003e\n\u003cli\u003eHeller DA, Clifford CA, Goldschmidt MH, Holt DE, Manfredi MJ, Sorenmo KU. Assessment of Cyclooxygenase-2 Expression in Canine Hemangiosarcoma, Histiocytic Sarcoma, and Mast Cell Tumor. \u003cem\u003eVet Pathol\u003c/em\u003e. 2005;42(3):350-353. doi:10.1354/vp.42-3-350\u003c/li\u003e\n\u003cli\u003eN\u0026oacute;brega DF, Sehaber VF, Madureira R, Bracarense APFRL. Canine Cutaneous Haemangiosarcoma: Biomarkers and Survival. \u003cem\u003eJ Comp Pathol\u003c/em\u003e. 2019;166:87-96. doi:10.1016/j.jcpa.2018.10.181\u003c/li\u003e\n\u003cli\u003eMctiernan A, Friedenreich CM, Katzmarzyk PT, et al. Physical Activity in Cancer Prevention and Survival: A Systematic Review. \u003cem\u003eMed Sci Sports Exerc\u003c/em\u003e. 2019;51(6):1252-1261. doi:10.1249/MSS.0000000000001937\u003c/li\u003e\n\u003cli\u003eWard H, Fox LE, Calderwood‐Mays MB, Hammer AS, Couto CG. Cutaneous Hemangiosarcoma in 25 Dogs: A Retrospective Study. \u003cem\u003eJ Vet Intern Med\u003c/em\u003e. 1994;8(5):345-348. doi:10.1111/j.1939-1676.1994.tb03248.x\u003c/li\u003e\n\u003cli\u003eShiu KB, Flory AB, Anderson CL, et al. Predictors of outcome in dogs with subcutaneous or intramuscular hemangiosarcoma. \u003cem\u003eJ Am Vet Med Assoc\u003c/em\u003e. 2011;238(4):472-479. doi:10.2460/javma.238.4.472\u003c/li\u003e\n\u003cli\u003eHargis AM, Ihrke PJ, Spangler WL, Stannard AA. A Retrospective Clinicopathologic Study of 212 Dogs with Cutaneous Hemangiomas and Hemangiosarcomas. \u003cem\u003eJournal of Veterinary Pathology\u003c/em\u003e. 1992;29(4).\u003c/li\u003e\n\u003cli\u003eSchultheiss PC. A retrospective study of visceral and nonvisceral hemangiosarcoma and hemangiomas in domestic animals. \u003cem\u003eJournal of veterinary diagnostic investigation\u003c/em\u003e. 2004;16:522-526.\u003c/li\u003e\n\u003cli\u003eBartlett PC, Van Buren JW, Neterer M, Zhou C. Disease surveillance and referral bias in the veterinary medical database. \u003cem\u003ePrev Vet Med\u003c/em\u003e. 2010;94(3-4):264-271. doi:10.1016/j.prevetmed.2010.01.007\u003c/li\u003e\n\u003cli\u003eNunney L. The effect of body size and inbreeding on cancer mortality in breeds of the domestic dog: a test of the multi-stage model of carcinogenesis. \u003cem\u003eR Soc Open Sci\u003c/em\u003e. 2024;11(1). doi:10.1098/rsos.231356\u003c/li\u003e\n\u003cli\u003eSchofield I, Stevens KB, Pittaway C, et al. Geographic distribution and environmental risk factors of lymphoma in dogs under primary-care in the UK. \u003cem\u003eJournal of Small Animal Practice\u003c/em\u003e. 2019;60(12):746-754. doi:10.1111/jsap.13075\u003c/li\u003e\n\u003cli\u003eBenjamin SA, Saunders WJ, Angleton GM,, Lee AC. Radiation Carcinogenesis in Dogs Irradiated During Prenatal and Postnatal Development. \u003cem\u003eJ Radiat Res\u003c/em\u003e. Published online 1991:86-103.\u003c/li\u003e\n\u003cli\u003eHern\u0026aacute;n MA, Robins JM. Using Big Data to Emulate a Target Trial When a Randomized Trial Is Not Available. \u003cem\u003eAm J Epidemiol\u003c/em\u003e. 2016;183(8):758-764. doi:10.1093/aje/kwv254\u003c/li\u003e\n\u003cli\u003ePurina Institute. The Purina 9-point body condition score system. Accessed January 24, 2025. https://www.purinainstitute.com/centresquare/nutritional-and-clinical-assessment-tools/the-purina-body-condition-system#:~:text=The%20Purina%209-Point%20Body%20Condition%20System%2C%20available%20for,a%20pet\n%20for%20excess%20or%20inadequate%20body%20fat.\u003c/li\u003e\n\u003cli\u003eU.S Census Bureau. Poverty data sources: Poverty rates. Accessed January 17, 2025. https://www.census.gov/topics/income-poverty/poverty/about/related-sites/rates.html\u003c/li\u003e\n\u003cli\u003eNational Office of Animal Health. NOAH Compendium of Data Sheets for Animal Medicines . Accessed January 28, 2025. https://www.noahcompendium.co.uk/datasheets\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Table 2","content":"\u003cp\u003eTable 2 is available in the Supplementary Files section.\u003c/p\u003e\n"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Haemangiosarcoma, Golden Retriever, Epidemiology, Cancer, Risk factors, diagnosis, canine","lastPublishedDoi":"10.21203/rs.3.rs-7064010/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7064010/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003ch2\u003eBackground\u003c/h2\u003e\u003cp\u003eThe Golden Retriever Lifetime Study (GRLS) is a prospective, longitudinal cohort study launched in the United States in 2012 to investigate genetic, environmental, and lifestyle factors contributing to disease. As the cohort of dogs now approaches end of life, it presents unique opportunities to explore haemangiosarcoma\u0026mdash;a highly aggressive malignancy commonly reported in Golden Retrievers and the most commonly diagnosed tumour in the GRLS cohort. This study aimed to calculate incidence, explore potential risk factors for diagnosis, and describe tumour-specific survival.\u003c/p\u003e\u003ch2\u003eMethods\u003c/h2\u003e\u003cp\u003eDogs diagnosed with haemangiosarcoma before December 31st, 2024, were classified as cases and all remaining dogs as non-cases. Data from owner and veterinarian surveys, along with necropsy reports, were processed to derive variables on demographics, lifestyle, environmental exposures, comorbidities, and medications. Analyses of the non-demographic, potentially modifiable factors were considered exploratory given the limited prior evidence and potential for residual confounding. Risk factor analysis for haemangiosarcoma diagnosis used multivariable binary logistic regression modelling. Median survival times (MST) were calculated.\u003c/p\u003e\u003ch2\u003eResults\u003c/h2\u003e\u003cp\u003eOverall, 490 of 3,044 dogs (16.1%) were diagnosed with haemangiosarcoma, giving an incidence rate of 1.69 (95% CI: 1.54\u0026ndash;1.85) cases per 100 dog-years. Dogs aged\u0026thinsp;\u0026lt;\u0026thinsp;7 or \u0026ge;\u0026thinsp;10 years had lower odds of diagnosis than those aged 7\u0026ndash;9 years. Male sex, higher median adult bodyweight, gastrointestinal comorbidities and NSAID use were associated to higher odds of diagnosis, Previous cancer history and higher owner-reported activity levels were associated with reduced odds of diagnosis. MST from diagnosis across all body-locations was 0 days (95% CI: 0\u0026ndash;0), but cutaneous haemangiosarcoma showed a significantly longer median survival (log-rank test p\u0026thinsp;\u0026lt;\u0026thinsp;0.01) of 19 days (95% CI: 0\u0026ndash;479).\u003c/p\u003e\u003ch2\u003eConclusions\u003c/h2\u003e\u003cp\u003eThis study provides a longitudinal epidemiological profile of haemangiosarcoma in Golden Retrievers. The relatively high incidence rate and poor outcomes observed highlight the ongoing clinical challenges of this disease. Risk factor modelling identified both established (age, sex, bodyweight) and novel associations (NSAID use, gastrointestinal comorbidities, prior cancer, activity level). While these novel associations are hypothesis-generating and warrant cautious interpretation, they provide direction for future research to validate findings and elucidate underlying mechanisms.\u003c/p\u003e","manuscriptTitle":"Haemangiosarcoma in the Golden Retriever Lifetime Study: frequency, risk factors and post-diagnosis survival","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-07-14 10:55:54","doi":"10.21203/rs.3.rs-7064010/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7f5013c7-27b1-42c5-bd88-81652e77081c","owner":[],"postedDate":"July 14th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-01-28T00:38:52+00:00","versionOfRecord":[],"versionCreatedAt":"2025-07-14 10:55:54","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7064010","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7064010","identity":"rs-7064010","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}