Risk of ovarian cancer in women with pelvic inflammatory disease and homologous recombination repair gene mutations under 55: a population-based cohort study.

Ovarian cancer (OC) stands as a preeminent menace among female malignancies, characterized by its pernicious nature and enigmatic etiology, often diagnosed at advanced stages, consequently rendering it a formidable adversary [ 1 ]. According to the statistics of Cancer Statistics, 2022, the incidence of OC ranks second place but occupies first in mortality in gynecological cancers [ 2 ]. The dearth of early screening modalities and diagnostic complexities accentuates its lethal proclivity [ 3 ]. Pelvic inflammatory disease (PID), a heterogeneous inflammatory spectrum encompassing salpingitis, oophoritis, non-cervical uterine inflammatory maladies, and other pelvic inflammatory disorders, emerges as a plausible nexus in the etiological panorama of OC [ 4 ]. While prior investigations by Risch and Howe [ 5 ] first posited PID as an escalatory precursor to epithelial OC, the subsequent wealth of literature, albeit prodigious, remained bedeviled by conspicuous inconsistencies [ 6 7 8 ]. Remarkably, the literature is notably bereft of large-scale prospective inquiries into the association between PID and OC with a long follow-up time. Concurrently, germline homologous recombination repair (gHR) pathogenic variant, an acknowledged risk factor in the OC milieu, underpins 15%–20% genetic heritability [ 9 ]. The therapeutic advent of poly (ADP-ribose) polymerase (PARP) inhibitors added a salient dimension, particularly in gHR-mutated patients [ 3 10 11 12 ]. However, the enigmatic interplay between genetic susceptibility of gHR mutations and other exposures remains uncharted. To illuminate this intricate nexus, we conducted a large prospective study within the UK Biobank, unraveling the putative linkages among PID history, OC risk, and the interactive effects with the genetic risk of OC. No statistically significant association was observed between PID history and OC risk both in the whole group and during the tumor subtyping analysis. However, our investigations of subgroup analyses on age, unveiled a discernible positive association between PID history and OC risk in the subset aged younger than 55 years (hazard ratio [HR]=1.92; 95% confidence interval [CI]=1.02, 3.63). Regrettably, no statistically significant interactive effect between gHR mutation and PID history was discerned. Thus, our study bridged this lacuna, providing new insights into the association between PID history and OC risk through the expansive lens of the UK Biobank while concurrently scrutinizing the interactive interplay between PID history and genetic susceptibility.

We conducted a large prospective study in the UK Biobank to assess the association between PID history and risk of OC, as well as delving into the nuanced interactive dynamics between PID history and gHR pathogenic variant ( Fig. 1 ). The baseline characteristics of all participants were presented in Table 1 . Among the 261,624 participants, the median follow-up time was 12.9 years (interquartile range, 12.1–13.6; overall range, 0–15.8). We documented 901 incident cases of OC (18 in PID group and 883 in the non-PID group, corresponding to incidence rates 3.51 and 2.72 per 10,000 person-years, respectively). The mean ± standard deviation age of participants at enrollment was 56.8±8.0 years and 83.6% were Caucasian. Compared with the non-exposures, participants with PID were younger, non-Caucasian, current smokers, had a higher BMI, higher Townsend deprivation index, received non-college education and accompanied with comorbid endometriosis (p<0.05). gHR, germline homologous recombination repair; ICD-10, International Classification of Diseases tenth revision; OC, ovarian cancer; PID, pelvic inflammatory disease. Values are presented as mean ± standard deviation or number (%). For covariates with missing data, the missing indicator method for categorial variables was utilized. BMI, body mass index; OCP, oral contraceptive; PID, pelvic inflammatory disease. * p<0.05 compared with non-PID. Utilizing 3 Cox regression models, we scrutinized the putative association between PID and the risk of OC ( Table 2 ). In the first model, a crude HR of 1.29 (95% CI=0.81, 2.06) was carried out. Transitioning to the second model, a compendium of variables encompassing age, population, BMI, Townsend deprivation index, smoking and drinking statuses, and educational attainment were considered, yielding an adjusted HR of 1.46 (95% CI=0.91, 2.33). In the third model, an additional layer of refinement was incorporated, incorporating covariates inclusive of endometriosis, OCP, and parity, thereby eliciting an adjusted HR of 1.45 (95% CI=0.90, 2.32). These outcomes uniformly underscored the absence of statistically significant associations between a history of PID and the occurrence for OC across 3 Cox regression models (all p-value >0.05). This result of the overall analysis may be ascribed to the older age profile characterizing our study cohort, diverging notably from the demographic milieu of other investigations [ 6 ]. Additionally, we conducted a subgroup analysis by OC subtypes, where no significant associations were found between PID history and the risk of OC ( Table S1 ). Model 1 is crude, adjusted for nothing. Model 2 is adjusted for age, body mass index, population, smoking status, drinking status, education degree and Townsend deprivation index. Model 3 is adjusted for endometriosis, parity and oral contraceptive in addition to model 2. CI, confidence interval; HR, hazard ratio; OC, ovarian cancer; PID, pelvic inflammatory disease. Previously, gHR ( BRCA1 , BRCA2 , RAD51C , RAD51D , BRIP1 ) was reported to contribute to 15%–20% hereditary OC [ 9 ]. Thus, we extracted gHR pathogenic variant carriers from paired peripheral blood WES data and assessed whether the 5-gene set was related to a higher risk of OC. There was a prevalence of 0.024% (2,466/102,874) carriers in the younger age group and 0.022% (3,479/158,570) in the other group. As expected, individuals with gHR pathogenic variants were significantly related with the higher risk of OC than those non-carriers (HR=3.60; 95% CI=2.25, 5.75), even after adjustment (3.57; 2.23, 5.71) ( Table S2 ). Subsequently, we tried to explore the potential interaction between PID and genetic vulnerability. However, we did not find a significant interactive relationship between gHR mutation and PID history ( Table S3 ), which may result from that PID is a highly heterogeneous disease and affected by many other factors such as age other than the genetic effects. The relative excess risk due to interaction (RERI) is −0.57 (95% CI=−7.08, 5.93), and the attributable proportion due to interaction (AP) is 0.17 (95% CI=−2.50, 2.15). We also explored the joint effect of gHR mutation and PID history compared with those non-PID and non-carriers, the results remained unsignificant (HR=3.27; 95% CI=0.46, 23.28) ( Fig. 2 ). CI, confidence interval; HR, hazard ratio; OC, ovarian cancer; PID, pelvic inflammatory disease. In the age subgroup analysis using the fully adjusted model, we found a positive association between a history of PID and the risk of OC in participants aged younger than 55 years (HR=1.92; 95% CI=1.02, 3.63) ( Table 3 ). Intriguingly, within the elder cohort, no statistically significant association between a history of PID and OC risk was manifest in the adjusted model (HR=1.10; 95% CI=0.52, 2.32), intimating an age-dependent predilection wherein the younger demographic with PID bears an augmented risk of OC. Values are presented as hazard ratio (95% confidence interval). Model 2 was adjusted for age, body mass index, population, smoking status, drinking status, education degree, Townsend deprivation index, endometriosis, parity and oral contraceptive. Model 3 was adjusted for germline homologous recombination repair mutation status in addition to model 2. OC, ovarian cancer; PID, pelvic inflammatory disease. Additionally, we delved the joint effect between gHR mutations and PID in the younger subgroup. Participants endowed with both gHR mutations and a history of PID at an early age (<55 years old) exhibited the highest risk of HR 7.40 (95% CI=1.03, 53.10) ( Fig. 3 ). However, our evaluation of the interactive effect between gHR mutations and PID within this subgroup yielded non-significant outcomes (RERI=2.96; 95% CI=−11.71, 17.63, and AP=0.40; 95% CI=−0.81, 1.61; Table S3 ). This underscored the complexity of relationship between PID history, gHR mutations, and OC risk in the younger demographic. CI, confidence interval; HR, hazard ratio; OC, ovarian cancer; PID, pelvic inflammatory disease.

In this large prospective investigation involving 261,624 participants and a mean observational duration of 12.6 years, adjustments for potential confounding variables, encompassing age, BMI, smoking and drinking status, education degree, Townsend deprivation index, parity, OCP use, and endometriosis, were undertaken. Regrettably, no discernible statistically significant association was ascertained between a history of PID and the risk of OC. Nevertheless, a suggestive association was observed within the subset aged younger than 55 years. Curiously, participants harboring gHR mutations and a history of PID at an early age emerged as having the pinnacle risk of developing OC. This intriguing finding portends potential implications for the connections between genetic susceptibility and inflammatory antecedents in the etiological trajectory of OC. Remarkably, this investigation, to the best of our knowledge, represents the large-scale, nationwide prospective study discovering the interplay between PID history, genetic susceptibility and the risk of OC within the UK Biobank. Risch and Howe [ 5 ] reported a case-control study that the history of PID would increase the risk of subsequent epithelial OC with an odds ratio (OR) 1.53 (95% CI=1.10, 2.13). Noteworthily, Lin et al. [ 6 ] conducted a prospective cohort study in the Longitudinal Health Insurance Database 2005 in Taiwan, with 67,936 PID cases and 135,872 controls. During its 3-year follow-up period, 90 had developed OC (42 cases with PID and 48 cases in control group). The adjusted HR for OC in patients with PID was 1.95 (95% CI=1.27, 2.92) and for women ages 35 years or younger with PID, the adjusted HR was slightly higher than in older women (2.23; 95% CI=1.02, 4.79 vs. 1.82; 95% CI=1.10, 3.04). This study was strong evidence for the association between PID history and risk of OC. However, after this, another similar prospective study was published in 2017. Rasmussen et al. [ 7 ] conducted a study in the Danish Civil Registration System with 81,281 PID cases and 1,237,648 controls. Five thousand three hundred and thirty-six individuals had developed OC (246 cases with PID and 5,110 cases in non-PID group). They did not find a significant association between a history of PID and risk of OC no matter how many episodes of PID. Analyzing by tumor subtyping histology, the results remained unsignificant. Rasmussen et al. [ 15 ] published a pooled study that the association between PID status and OC among participants in Australia, Europe, and North America from 1989 to 2009 was not significant, and by contrast, it was significant when it comes to the association between PID status and the risk of borderline OC. Lately, Jonsson et al. [ 8 ] reported that PID was associated with the risk of OC with a OR of 1.35 (95% CI=1.19, 1.53) and recurrent PID increased the risk with an OR of 2.50 (95% CI=1.44, 4.35) in Sweden National Cancer Register. The reviewed literature presents incongruent findings regarding the correlation between PID history and OC risk. Many studies adopt a case-study or retrospective approach, lacking consistent outcomes. Notably, prospective investigations utilizing extensive national databases with prolonged follow-up periods are scare with this field. Given this discrepancy, we posit that augmenting this research field with data from UK Biobank could offer substantial supplementation. Its vast dataset and longitudinal design could provide invaluable insights, potentially bridging gaps and enhancing the understanding of the association between PID history and OC risk. Hillis et al. [ 16 ] showed that among adolescents and women younger than 30 years the strongest predictor of recurrence of PID was infection with Chlamydia trachomatis; risk was raised 2 to 8 times compared with that in women aged 30–44 years. Kelly et al. [ 17 ] also reported that among adolescent girls 47% had recurrent PID, probably because of high rates of sexual activity, poor use of protection and treatment, and vulnerable immune status. Risch and Howe [ 5 ] also mentioned that among women younger than 60 years of age at interview, the risk was particularly elevated (OR=1.60; 95% CI=1.09, 2.35). Even some women who reported a history of PID before age 20 had a higher OR of 3.08 (95% CI=1.17, 8.13). The outcomes indicate a potential elevation in OC risk linked to PID history at a younger age. It is worth noting that our cohort’s higher median age of 58.0 years than other studies [ 6 ]. This age contrast might partly contribute to our observed positive association within the subgroup aged younger than 55 years, while not evident across the entire cohort. The discrepancy underscores the potential impact of age variation on the relationship between PID history and OC risk. OC epitomizes genomic instability within the homologous recombination repair (HRR) pathway, notably the high-grade serous OC. Among the pivotal genes, BRCA1/2 , RAD41C/D , and BRIP1 stand as the germline top 5 highly-mutated, crucial genes within the HRR mechanism [ 9 18 ]. Deficiency in HRR serve as a focal point for PARP inhibitors, functioning as predictive markers for their efficacy in OC therapy [ 18 ]. The pivotal role of gHR in OC pathogenesis prompts inquiry into the interplay dynamics between PID history and gHR. Nevertheless, no statistically significant interactive effects were revealed in our study, potentially attributed to the multifaceted nature of PID characterized by diverse types and etiologies. The incidence of OC in individuals with HRR deficiency presents a notable age-related trend [ 19 ]. Initial diagnoses predominantly occur at a younger age, with a diminishing disparity in incidence as age increases. This age-dependent pattern suggests potential variations in OC etiologies across distinct age cohorts. However, a noteworthy observation emerged: participants harboring both gHR mutations and a history of PID at an early age (<55 years old) manifested the highest susceptibility to OC. This result implied us that take an in-time prevention for those people with gHR and PID history at a younger age, even preventive surgery of bilateral ovaries. This correlation underscores the necessity for proactive measures targeting this demographic, advocating for timely preventive interventions, potentially including prophylactic bilateral oophorectomy. Inflammation resulting from bacterial or viral infections is extensively documented as a prominent precursor to human carcinogenesis [ 20 21 22 23 24 ]. For example, Helicobacter pylori is linked to gastric cancer, human papilloma virus is connected to cervical cancer, and hepatitis B and hepatitis C viruses are correlated with hepatocellular carcinoma [ 24 ]. In addition, Nene et al found that having a community type O cervicovaginal microbiota was significantly associated with the presence of OC or factors those would affect the risk of the disease, such as age and BRCA1 germline variants [ 25 ]. Whether infection affects the risk of developing OC in patients with PID should be assessed. Even though the precursors of cancer vary by case, in previous studies inflammation has been proposed as a potential biomarker of carcinogenesis. The observations suggest a potential interplay between perturbations in the microenvironment of the genital tract and distinct genetic mutations, possibly influencing the pathogenesis of tumors within the genital system. Owing to the limitations of UK Biobank, the data were collected from the self-administrated touch-screen questionnaires, we cannot avoid the possibility of errors and omissions in the reporting of personal medical history and the absence of detailed histological subtyping of OC, notably the characterization of high-grade serous OC, which poses a further limitation in delineating precise tumor subtypes within the dataset. Additionally, due to the small sample size of OC cases with PID in the UK Biobank, it is undeniable that our conclusion requires further research and validation in other real-world cohorts. In summary, our study underscores the potential pathogenic role of PID in OC, particularly when concomitant with gHR mutations during early age onset. Given the burgeoning array of targeted approaches for gHR-associated OC, these patients demand heightened vigilance and specialized, individualized follow-up strategies.

The UK Biobank is a population-based prospective cohort study of over 500,000 participants. They were 39–71 years old when recruited in 22 assessment centers between 2006 and 2010 throughout the UK. At the first visit, participants were required to sign consent and completed the self-administrated touch-screen questionnaire, including sociodemographic factors, family history, early exposures, physical factors, environmental factors, lifestyle and health status, and importantly, most of them had undergone the whole-exome sequencing (WES) in the peripheral blood [ 13 ]. Therefore, the UK Biobank has other than abundant epidemiological data but matched genetic data. The application number of UKB in our study was 88,159. PID was classified with the International Statistical Classification of Diseases and Related Health Problems (ICD-10) codes N70, N71, N73, and N74 (tenth revision, clinical modification) [ 8 ]. Diagnoses of PID were made on the basis of the patient’s self-report via answering the questionnaire on the touchscreen and reports from medical history. We excluded participants who was assigned male at birth or genetic sex were not female (n=237,897), who had prevalent OC (n=604, ICD-10 code C56), and who reported PID after recruitment (n=2,276, ICD-10 codes N70, N71, N73, and N74), leaving 261,624 participants in our study. Comprehensive assessment of participant characteristics involved an examination of sociodemographic facets, age, ethnicity, and educational attainment, along with lifestyle factors such as smoking and drinking status at baseline. Additionally, personal details, including body mass index (BMI), parity, and oral contraceptive (OCP) usage, were gathered through self-administered questionnaires. Each participant's socioeconomic status was delineated by the assignment of a Townsend deprivation index, a metric reflective of the socio-economic milieu linked to the residential postcode. We also collected the physician’s diagnosis of endometriosis from the self-completed baseline assessment questionnaire. Genetic susceptibility was determined through an analysis of gHR mutations—specifically, BRCA1 , BRCA2 , RAD51C , RAD51D , and BRIP1 [ 9 ]. Under the American College of Medical Genetics and Genomics guideline, we used the snpeff software to annotate the variants and identified pathogenic or likely-pathogenic variants within any of these quintessential genes, categorizing individuals as gHR mutation carriers [ 14 ]. This methodological rigor ensured a nuanced exploration of the genetic landscape underpinning our cohort [ 14 ]. Date of death was obtained from death certificates held within the National Health Service Information Centre (England and Wales) and the National Health Service Central Register (Scotland). Prevalent and incident cancer cases within the UK Biobank cohort were identified by national cancer registries. The primary outcome was defined as the first diagnosis of OC with ICD-10 code C56. Tumor histology data was extracted under the Field ID 40011 in the UK Biobank. Baseline characteristics were described across the entire cohort, stratified into PID and non-PID cohorts. The person-time contributed by participants was calculated from the date they attended the assessment center until the earliest of the following: the occurrence of OC, loss to follow-up, death, or the end of the study's observation period (October 2022). We implemented Cox's proportional hazard models adjusted for confounding factors—age, BMI, genetic ethnic grouping, Townsend deprivation index, smoking and drinking status, educational attainment, endometriosis, OCP history, and parity. We calculated the HRs and corresponding 95% CIs between exposures and outcomes. In adhering to the proportional hazard assumption, each dichotomous variable underwent rigorous scrutiny for proportionality through diagnostic univariate Cox regression. For covariates with missing data, we used the missing indicator method for categorial variables and used the mean imputation for continuous variables. All data analyses were done with R (version 4.3.1; R Foundation for Statistical Computing, Vienna, Austria). Significance was set at p<0.05 except indicated otherwise.