Integrating polygenic risk scores and constructing a predictive model to assess the risk of knee osteoarthritis in Taiwanese population

Integrating polygenic risk scores and constructing a predictive model to assess the risk of knee osteoarthritis in Taiwanese population | 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 Integrating polygenic risk scores and constructing a predictive model to assess the risk of knee osteoarthritis in Taiwanese population Kai-Xuan Lim, Yun-Jer Shieh, Hsiu-Jung Liao, Tzu-Hung Hsiao, Chi-Ying F. Huang, and 3 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-5391580/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 Objective: Knee osteoarthritis (OA) is a prevalent degenerative joint disease that can lead to severe limitations in daily activities and function. Genetic components contribute significantly to knee OA and it always coexists with other comorbid conditions. Our goal is to identify the most effective PGS panel that can be applied in the Taiwanese population. We also aim to develop a predictive model for the risk of knee OA by integrating the polygenic risk scores(PGS) and comorbidity information. Method We analyzed a total of 43,686 patients from the Taiwan Precision Medicine Initiative (TPMI) database and 1,363 cases of knee OA were identified. We validated which of the four published PGS panels was the most appropriate for predicting the risk of knee OA in the Taiwanese population. We then integrate the validated PGS panel and common comorbidities with knee OA to evaluate the predictive performance of the model. Results We found that PGS002767 showed the strongest association with our knee OA patients (OR: 1.59 at rank 5, OR: 1.46 at rank 4 compared to rank 1). GERD (OR:2.17), insomnia (OR:1.57), dyslipidemia (OR:1.49), and hypertension (OR:1.47) were retained in the final model. The AUC of our final model which included PGS002767, BMI and comorbidities was 0.72. Conclusion: To the best of our knowledge, this is the first study to validate the applicability of PGS panels in knee OA patient of the Taiwanese population. We create a predictive model for knee OA allowing clinicians to quickly and cost-effectively screen patients at risk for developing knee OA. Knee Osteoarthritis Osteoarthritis Polygenic risk scores osteoarthritis risk Figures Figure 1 Figure 2 Figure 3 Introduction Osteoarthritis(OA) is a chronic degenerative joint disease that affects more than 250 million people worldwide, with more than 60 million cases in East Asia [1-2]. Of all the joints affected, the knee is the most commonly affected joint by OA. Knee OA can lead to severe joint pain and limited functional activities, and can have a serious impact on quality of life and livelihood [3]. Knee OA is known to be a multifactorial disease and many studies have shown that the risk factors for knee OA include heredity, gender, aging, obesity and others [4]. These factors contribute to the heterogeneity in the onset and progression of knee OA [5,6]. Among all these risk factors, genetic factors undoubtedly play an important role in knee OA. Previous twin and family studies suggest that the influence of genetic factors is approximately 40 percent for knee OA [7]. However, the development of OA is not solely attributed to the action of a single gene. Rather, it is more likely due to the interaction of multiple genes. In fact, many associated genes have demonstrated significant genetic influence in both genome-wide association studies(GWAS) and familial association studies [8-10]. Previous meta-GWAS study has identified hundreds of single nucleotide polymorphism(SNP) loci associated with knee OA [10]. As the genetic architecture of knee OA is likely to involve many loci, each with small effect sizes, polygenic score(PGS) or polygenic risk score (PRS) has been proposed to be used as a risk prediction tool in clinical practice or in a screening program by using information from multiple small effect SNPs to calculate the disease risk in individuals based on their genetic makeup[11,12]. PRS provided information that can be used to predict the risk of knee OA and help to develop personalized prevention and management plans for patients. PGS have demonstrated their predictive power in several diseases [13,14]. However, it should be noted that due to the limited sample sizes in Asian genetic databases, the development and validation of most PGS, including those related to knee OA, have primarily been conducted using populations of European ancestry [15]. In the largest PGS database, the PGS catalog, there are four PGS panels for knee OA, and they were primarily developed using the European population [16]. In situations where sample sizes remain limited, validating the applicability of published PGS panels to different ethnic groups appears to be a faster, more efficient, and cost-effective approach. Early identification of patients with knee OA is crucial for appropriate treatment and improved outcomes. On the other hand, knee OA is a common health problem among older adults, which often occurs alongside other health conditions or comorbidities. These comorbidities include back pain [17], metabolic disorders [18], mental disorders [19], digestive disorders [20], endocrine disorders [21], cardiovascular disorders [22], and respiratory disorders [23]. Evidence suggests that these comorbidities may share underlying physiological and inflammatory mechanisms with knee OA [24]. Furthermore, some of these comorbidities may also share similar environmental risk factors such as occupation and living environment [25]. Additionally, they could be caused by similar biomechanical factors [26].Despite these observations, it is still unclear whether these comorbidities represent or share factors that influence knee OA. It is needed to better understand the relationship between these conditions and knee OA, which could lead to new approaches for preventing and treating this condition. This is the first time we have the opportunity to integrate the genetic profiles of a disease through the largest non-European cohort genetic database, namely the Taiwan Precision Medicine Initiative (TPMI). To the best of our knowledge, this is the first study to validate the most efficacious PGS panel from an external cohort for identifying knee OA patients in the Taiwanese population, utilizing data from the TPMI. Additionally, we sought to investigate the prevalent comorbidities associated with knee OA and integrate this information with PGS panels to establish a superior predictive model for knee OA. Methods Study populations We analyzed 43,686 participants of TPMI who were recruited from Taichung Veterans General Hospital (TCVGH) from 2009 to December 2022. The aim of TPMI is to establish a genetic database for the Taiwanese population through a collaborative effort between Academia Sinica and 16 medical centers in Taiwan. All subjects in the study provided informed consent prior to participation. The research protocol was approved by the Institutional Review Boards of TCVGH (IRB number: SE23188B). Patients with knee OA were identified by International Classification of Diseases, Ninth Revision (ICD-9) codes of 715.00(Osteoarthrosis, generalized, unspecified site), 715.10(Osteoarthrosis, localized, primary, unspecified site), 715.16(Osteoarthrosis, localized, primary, lower leg), 715.18(Osteoarthrosis, localized, primary, other specified site), 715.30(Osteoarthrosis, localized, not specified whether primary or secondary involving unspecified site), 715.36(Osteoarthrosis, localized, not specified whether primary or secondary involving lower leg), 715.38(Osteoarthrosis localized not specified whether primary or secondary involving other specified sites), 715.90(Osteoarthrosis, unspecified whether generalized or localized involving unspecified site), 715.96(Osteoarthrosis, unspecified whether generalized or localized involving lower leg), 715.98(Osteoarthrosis unspecified whether generalized or localized involving other specified site) in Taiwan's National Health Insurance Research Database (NHIRD). As there has always been concern about over-coding in the NHIRD, we also used an additional knee replacement surgery code (64164B) and an articular injection code (39005C) to reconfirm that the patient actually had knee OA. Considering that some patients with OA of other joints (e.g. shoulder arthritis) or other types of knee arthritis (e.g. rheumatoid arthritis) were also treated with joint injections and replacements, our first author, KX,LIM, reviewed their outpatient medical records to confirm the diagnosis of knee OA. Controls were identified as participants without osteoarthritis (ICD-9 715.XX). We used the MatchIt package (method = ‘nearest’) in R to match the cases to the controls in a 1:4 ratio based on sex and age [27]. We then randomly divided the sex- and age-matched participants into two groups, the training group and the test group, in an 8:2 ratio (Figure1). Comorbidities identification We select 13 common diseases associated with knee OA that have been reported, including: diabetes (ICD-9: 250.X), hypertension (ICD-9: 401.X, 405.X), dyslipidemia (ICD-9: 272.X), chronic obstructive pulmonary disease (COPD) (ICD-9: 490.X–496.X), stroke (ICD-9: 433.X–434.X), chronic kidney disease (CKD) (ICD-9: 585.X), backpain (ICD-9: 724.5, 724.9), gastroesophageal reflux disease (GERD) (ICD-9: 530.X), insomnia (ICD-9: 307.4, 327.X, 780.5), depression (ICD-9: 296.X), thyroid disorder (ICD-9: 240.X–246.X), duodenal ulcer (ICD-9: 532), and colonic diverticula (ICD-9: 562). Genotyping and calculating polygenic scores The genome data of TPMI participants were collected by Axiom Genome-Wide TPM 2.0 Array with 686,463 SNPs and imputed to the Genome Reference Consortium Human Build 37 (GRCh37) by Michigan Imputation Server [28]. We downloaded 4 published knee OA PGS panels (PGS001192, PGS002729, PGS002749 and PGS002767) from the PGS catalog database. We calculated the polygenic score of each participant by the information of the SNPs and the weight of the SNPs provided by the PGS panel separately using PLINK 1.9 [29]. Statistical analysis The main statistical analyses were performed using R 4.2. We examined the distribution of the 4 PGS panels using the Shapiro-Wilk test, skewness and kurtosis. Next, we divided all our participants into 5 equal parts based on their PGS score, ranging from low to high: rank 1 (0-20%), rank 2 (20-40%), rank 3 (40-60%), rank 4 (60-80%), and rank 5 (80-100%). We used the independent t-test and chi-squared test to compare the continuous or categorical variables between the case and control groups. Univariate logistic regression was used to determine which PGS panel was the most appropriate for discriminating knee OA patients in our cohort and to analyze the association between knee OA and the most appropriate PGS panel. Multivariate logistic regression was used for two purposes in the training group. First, to analyze the association between knee OA and PRS with adjustment for age, gender, and all comorbidities. Second, to establish the predictive equation for the test group. Statistical significance was set at p<0.05. Constructing the prediction equation for knee OA To construct the prediction equation and to account for the genetic influence of other comorbidities, genetic correlation was used to analyze the association between knee OA and the comorbidities that showed significance in the multivariate logistic regression analysis. Genetic correlation was performed using LDSC software [30]. The GWAS summary statistics, which are required inputs for LDSC, were analyzed using the TPMI cohort with a case-to-control ratio of 1:1–4 for each disease. The GWAS analyses were subject to consistent quality control (QC) using PLINK 1.9, including sample QC and SNP QC. The LD score of each SNP was calculated using LDSC in a 1 Mb window. The statistical significance for genetic correlation was set at p < 0.05/(n+1), where "n" depended on the number of remaining comorbidities. The receiver operating characteristic curve (ROC curve) and the area under the curve (AUC) were calculated to examine the predictive effect in the test group with or without BMI and comorbidities, respectively. The statistical significance was set at p<0.05. Results Characteristics of study population From our database, there were 1,363 participants defined as the case group with knee OA and 42,323 participants defined as the control group without knee OA. The mean age of our studied cases is 73 years old and the male to female ratio is approximately 1:2.5. After matching for sex and age with a case to control ratio 1:4, the matching group consisted of 1,363 cases and 5,452 controls. Individuals in the matching group were then allocated to the training and test groups in an 8:2 ratio. The training group had 1,123 cases and 4,329 controls, while the test group had 240 cases and 1,123 controls. The distribution of 13 previously reported common comorbidities between groups are shown in Supplementary Table 1. PGS002767 is the most appropriate PGS panel to assess the genetic risk of knee OA The SNPs retention rates for the 4 PGS panels after imputation were as follows: 88% for PGS001192, 100% for PGS002729, 100% for PGS002749, and 91% for PGS002767. The distributions of the 4 PGS panels were approximately symmetric based on their skewness and kurtosis, although PGS002749 did not show a normal distribution according to the Shapiro-Wilk test . PGS002767 had the strongest association with our knee OA patients among the 4 PGS panels, as it showed a progressive increase in the disease odd ratio(OR) and a progressive decrease in p-value with increasing rank. PGS002749 had a secondary association because its OR at rank 5 showed significance, but its ORs at ranks 2 to 4 were not significantly different. The associations for knee OA between PGS001192 and PGS002729 were weak because its OR did not show a clear trend or was not significant. Therefore, PGS002767 was the most appropriate PGS panel for genetic risk assessment of knee OA in Taiwanese population (Supplementary Table 2). BMI, GERD, insomnia, hypertension, and dyslipidemia are positively correlation with PGS002767 Univariable and multivariable logistic regression analysis were used to evaluate the relationship between PGS002767, BMI and comorbidities. Among the 13 previously reported common comorbidities, most variables including diabetes, hypertension, dyslipidemia, COPD, stroke, CKD, back pain, GERD, insomnia, duodenal ulcer, and colonic diverticula showed a significant association with knee OA. On the other hand, depression and thyroid disorders did not show a significant association. Further multivariable logistic regression analysis demonstrated that the associations of comorbidities were dominated by GERD [OR:2.17], insomnia [OR:1.57], dyslipidemia [OR:1.49] and hypertension [OR:1.47]. BMI also showed a significant positive correlation with knee OA [OR:1.12] in multivariable logistic regression analysis. As a result, BMI, PGS002767 and four most important comorbidities were retained in our model for predicting knee OA, indicating that our model preserved and integrated the unique information from these variables. Building the knee OA predictive model To avoid the inherent genetic effects of the 4 comorbidities mentioned above, we examined the genetic correlation of GERD, insomnia, hypertension, and dyslipidemia to assess the impact of including them in the predictive model. Table 2 shows that there was no significant genetic correlation between 4 comorbidities and knee OA. The area under curve(AUC) of our predictive model was 0.72, as shown in Figure 2a, when considering PGS002767, BMI, and comorbidities simultaneously. At the same time, we also observed the predictive effects of different variables. The AUC when considering only PGS002767 was 0.58, BMI was 0.66 and comorbidities was 0.64 (Figure 2b). Male patients with knee OA are more predictable in model Figure 3 showed that PGS002767 has better performance in predicting the risk of developing knee OA in male patients. In the analysis of gender subgroups, the ORs for the male group are more precise than those for the female group, with the confidence intervals(CI) observed to be narrower in the male group. Discussion Understanding the genetic etiology of knee OA is challenging because knee OA is a multifactorial disease with a polygenic genetic architecture. Many knee OA related GWAS and PRS studies have been conducted to evaluate the genetic influence, but most of them are limited to the European population. Recently, a study was published to evaluate the association between PRS and knee OA in the Japanese population[31]. To the best of our knowledge and literature review, this is the first study to validate the appropriateness of PGS from external cohort for the knee OA in the Taiwanese population through the largest non-European cohort genetic database, Taiwan Precision Medicine Initiative (TPMI PGS002767 showed the strongest association with knee OA in the Taiwanese population among the other PGS panels in our study. In the context of genetic differences in knee OA susceptibility between European and Asian populations, PGS002767 contains information from 1,052,275 SNPs, which means that PGS002767 has effectively captured information from almost the entire genome [16]. This may be the main reason why PGS002767 showed the best performance in our study. Second, the SNPs used to construct PGS002729 and PGS002749 were obtained from the same meta-analysis GWAS, which included 826,690 individuals from 9 different populations [10]. However, the representation of East Asian populations in the meta-analysis GWAS study was minority, which may explain the relatively low correlation between PGS002729, PGS002749, and our population [11]. The sparse PRS model used to construct PGS001192 eliminates the small effect size of SNPs. This method may not be consistent with the characteristics of knee OA which is caused by a lot of small effect loci [32]. Therefore, here are the reasons that may explain why PGS002767 is more suitable for our population compared to other PGS panels. On the other hand, using PGS002767 to assess the risk of knee OA in the Taiwanese population gives similar OR/AUC to studies using their own PGS panels [11]. This suggests that genes only influence only a portion of knee OA incidence, highlighting the need for further consideration of acquired factors. Knee OA involves both genetic and environmental factors and is associated with multiple chronic diseases [4]. In our hospital-based cohort, we investigated the impact of comorbidities and BMI on the development of knee OA. Most previously reported comorbidities show a significant association with knee OA in univariate analysis, except for depression and thyroid disorder. The meta-analysis study showed that the prevalence ratio of comorbidities in the upper gastrointestinal, psychological, cardiovascular, musculoskeletal, respiratory, and endocrine comorbidities associated with knee OA is consistent with our findings [33]. The lack of a significant association between depression and knee OA in our cohort is mainly due to the limited sample size. Fewer patients with mental disorders were recruited in the TPMI. Without professional assistance, it is difficult for patients with mental disorders to provide informed consent for research. Besides, a population-based cohort study conducted in Nord-Trøndelag also suggests that there is no correlation between thyroid-stimulating hormone and knee OA incidence [34]. Furthermore, a cohort study further investigated the relationship between two thyroid antibodies, anti-thyroid peroxidase antibody(TPOAb) and anti-thyroglobulin antibody(TgAb), and knee OA as well as chondrocalcinosis. The study found that only TPOAb above a certain concentration was associated with chondrocalcinosis, while neither TPOAb nor TgAb was associated with knee OA [35]. In multivariate analysis, GERD, insomnia, hypertension, and dyslipidemia overshadowed the effects of other comorbidities. Knee OA, once considered a primarily non-inflammatory condition, is now recognized to be associated with chronic low-grade inflammation. This inflammatory component, although subtle, has the potential to trigger various systemic effects. Notably, it has been suggested that this chronic inflammation may contribute to GERD, insomnia , and hypertension. However, it is questionable whether patients with knee OA have an increased risk of gastrointestinal complications due to the frequent use of anti-inflammatory drugs. A population-based cohort study showed that gastrointestinal complications remained significantly associated with knee OA even after excluding patients using anti-inflammatory drugs [36]. Another study also indicates that GERD has a significant presence in two of five comorbidity clusters in OA [37]. Insomnia is the only related comorbidity in our study that is associated with psychological factors. Insomnia can arise as a consequence of worsening OA pain, which tends to intensify during nighttime hours when native cortisol production reaches its lowest point. In 2018, a comprehensive study investigated the sleep patterns of OA patients, revealing that approximately 53% of them exhibited symptoms of insomnia [38]. A meta-analysis reveals that hypertension causes knee OA not only through inflammation but also through significant structural damage to the knee joint, and this effect remains even after adjustment for sex and BMI [39]. The relationship between dyslipidemia and knee OA remains uncertain. Two recent systematic reviews found no association between dyslipidemia and knee OA due to the high heterogeneity of the studies they reviewed [40]. However, in our study, dyslipidemia may represent a partial characteristic of cardiovascular disease and metabolic syndrome. On the other hand, BMI maintained its consistent effect in both univariate and multivariate analyses, independent of the effects of comorbidities. Furthermore, extensive research has firmly established the influence of BMI on knee OA. From our findings, we ultimately incorporated BMI, four specific comorbidities and combined the effects of PGS002767 to construct the knee OA prediction model. As a culmination of these factors, the resulting knee OA prediction model demonstrated an AUC value of 0.72. This value indicates a respectable level of predictive ability for identifying knee OA cases. Previous research has utilized various factors to develop predictive models for the incidence of knee OA, including knee injury, knee surgery, knee joint space narrowing, family history, occupation, cartilage-derived biomarkers, education level, physical activity, hand OA, and pain [11,41,42]. These models achieved AUC values ranging from 0.59 to 0.83. Among these models, they require detailed medical history, family history, additional imaging studies or blood tests. As a result, they are not readily applicable to the general population. The advantage of our model is that it achieves relatively good predictive performance using readily available clinical information without requiring extensive additional testing or population surveys. Once a patient in your hospital is enrolled in TPMI and their genetic data is available, you can use the genetic information, BMI, and comorbidities to screen the patient. This individual's risk of disease can help the patient and physician decide whether or not to intervene early in treatment. To our knowledge, our model is the first to incorporate comorbidities as a factor in the prediction of knee OA. Furthermore, in order to validate the independence of these comorbidities, we performed genetic correlation to confirm that these comorbidities were not genetically correlated. The AUC of the model considering only PGS002767 was 0.58, which is greater than or equal to previous studies considering genetics as a predictive factor, with AUC values ranging from 0.51 to 0.58 [11,41,42]. This result demonstrates that PGS002767 has excellent performance in the Taiwanese population, rivaling its efficacy in European populations. In addition, we found that male patients were particularly suitable for the use of PGS002767. Based on our results, we have provided a simple and convenient model for clinical practitioners to predict knee OA patients. Our study encountered several limitations. First, our model’s development relied on a case-control study design instead of a cohort study, which may affect its generalizability. Second, we did not fully impute the PGS002767 as its SNPs retention rate is 91% compared to our array, which may have an impact on its true effect. Third, our study could be susceptible to selection bias as we only included patients from a single tertiary medical center, raising the possibility that our findings primarily pertain to individuals with more severe symptoms. Fourth, our study lacks certain important factors for predicting knee OA, such as occupational stress or activity frequency, we hope that over time, as the TMPI program is more widely promoted and the integration of information is more fully consolidated, we will be able to incorporate these factors into our model. Finally, the sample size in the training group was relatively small, with only 1,123 cases, which may affect the precision of the model parameters. Furthermore, when assessing the relationship between knee OA and prevalent comorbidities, the time of initial diagnosis of knee OA preceded the establishment of the TPMI database, making it challenging to accurately determine the tine of disease onset. Conclusion In this study, we confirm the use of PGS002767 in predicting knee OA in the Taiwanese population. In addition, we develop a predictive model for knee OA that includes PGS002767, BMI and four specific comorbidities. Our results show that this model has a moderate level of predictive efficacy, making it a practical tool for clinical application. Declarations Acknowledgements We thank all the participants and investigators from Taiwan Precision Medicine Initiative which was funded by Academia Sinica (40-05-GMM, AS-GC-110-MD02 and 236e-1100202) and National Development Fund, Executive Yuan(NSTC111-3114-Y-001-001) Contributions Conception and design of the study: CHL, CHC, CYH and YLH. Acquisition of data: KXL, YCH, THH. Analysis and interpretation of data: KXL, YCH, THH. Drafting or revising the manuscript: KXL, YCH, THH, and HJL. All authors have approved the final article. Role of the funding source This research was funded by the Ministry of Science and Technology (MOST 111-2314-B-418-004, MOST 111-2321-B-418-002-, NSTC 112-2321-B-418-004- and NSTC 112-2314-B-418-010-MY3), Far Eastern Memorial Hospital (FEMH; 2022-C-042, 2022-C-123, 2023-C-124, 2023-C-064 and 2024-C-105), National Yang Ming Chiao Tung University–FEMH Cooperation (111DN12, 111DN22, 112DN13, 112DN22, 113DN09 and 113DN16), and National Taiwan University Hospital–FEMH Cooperation (111-FTN0012). Competing interests The authors declare no competing interests. References Safiri, S., et al., Global, regional and national burden of osteoarthritis 1990-2017: a systematic analysis of the Global Burden of Disease Study 2017. Annals of the rheumatic diseases, 2020. 79 (6): p. 819-828. Berenbaum, F., Osteoarthritis as an inflammatory disease (osteoarthritis is not osteoarthrosis!). Osteoarthritis and cartilage, 2013. 21 (1): p. 16-21. Vitaloni, M., et al., Global management of patients with knee osteoarthritis begins with quality of life assessment: a systematic review. BMC musculoskeletal disorders, 2019. 20 (1): p. 1-12. Suri, P., D.C. Morgenroth, and D.J. Hunter, Epidemiology of osteoarthritis and associated comorbidities. PM&R, 2012. 4 : p. S10-S19. Neogi, T., The epidemiology and impact of pain in osteoarthritis. Osteoarthritis and cartilage, 2013. 21 (9): p. 1145-1153. Wei, J., et al., Risk factor heterogeneity for medial and lateral compartment knee osteoarthritis: analysis of two prospective cohorts. Osteoarthritis and cartilage, 2019. 27 (4): p. 603-610. Spector, T.D. and A.J. MacGregor, Risk factors for osteoarthritis: genetics. Osteoarthritis and cartilage, 2004. 12 : p. 39-44. Karlsson, C., et al., Genome-wide expression profiling reveals new candidate genes associated with osteoarthritis. Osteoarthritis and cartilage, 2010. 18 (4): p. 581-592. Neame, R., et al., Genetic risk of knee osteoarthritis: a sibling study. Annals of the rheumatic diseases, 2004. 63 (9): p. 1022-1027. Boer, C.G., et al., Deciphering osteoarthritis genetics across 826,690 individuals from 9 populations. Cell, 2021. 184 (18): p. 4784-4818. e17. Sedaghati-Khayat B, Boer CG, Runhaar J, Bierma-Zeinstra SMA, Broer L, Ikram MA, Zeggini E, Uitterlinden AG, van Rooij JGJ, van Meurs JBJ. Risk Assessment for Hip and Knee Osteoarthritis Using Polygenic Risk Scores. Arthritis Rheumatol. 2022 Sep;74(9):1488-1496. Choi, S.W., T.S.-H. Mak, and P.F. O’Reilly, Tutorial: a guide to performing polygenic risk score analyses. Nature protocols, 2020. 15 (9): p. 2759-2772. Li, Z., et al., Polygenic Risk Scores have high diagnostic capacity in ankylosing spondylitis. Annals of the rheumatic diseases, 2021. 80 (9): p. 1168-1174. Mavaddat, N., et al., Polygenic risk scores for prediction of breast cancer and breast cancer subtypes. The American Journal of Human Genetics, 2019. 104 (1): p. 21-34. Lambert, S.A., et al., The Polygenic Score Catalog as an open database for reproducibility and systematic evaluation. Nature Genetics, 2021. 53 (4): p. 420-425. Mars, N., et al., Systematic comparison of family history and polygenic risk across 24 common diseases. The American Journal of Human Genetics, 2022. 109 (12): p. 2152-2162. Burnett, D.R., et al., A retrospective study of the relationship between back pain and unilateral knee osteoarthritis in candidates for total knee arthroplasty. Orthopaedic Nursing, 2012. 31 (6): p. 336-343. Gandhi, R., et al., Asian ethnicity and the prevalence of metabolic syndrome in the osteoarthritic total knee arthroplasty population. The Journal of arthroplasty, 2010. 25 (3): p. 416-419. Parmelee, P.A., C.A. Tighe, and N.D. Dautovich, Sleep disturbance in osteoarthritis: linkages with pain, disability, and depressive symptoms. Arthritis care & research, 2015. 67 (3): p. 358-365. King, L., et al., Are medical comorbidities contributing to the use of opioid analgesics in patients with knee osteoarthritis? Osteoarthritis and Cartilage, 2020. 28 (8): p. 1030-1037. van Dijk, G.M., et al., Comorbidity, limitations in activities and pain in patients with osteoarthritis of the hip or knee. BMC musculoskeletal disorders, 2008. 9 (1): p. 1-10. Kim, H.S., et al., Association between knee osteoarthritis, cardiovascular risk factors, and the Framingham Risk Score in South Koreans: a cross-sectional study. PloS one, 2016. 11 (10): p. e0165325. Wshah, A., et al., Prevalence of osteoarthritis in individuals with COPD: a systematic review. International journal of chronic obstructive pulmonary disease, 2018: p. 1207-1216. Valdes, A., Metabolic syndrome and osteoarthritis pain: common molecular mechanisms and potential therapeutic implications. 2020, Elsevier. p. 7-9. Callahan, L.F., et al., Associations of educational attainment, occupation and community poverty with knee osteoarthritis in the Johnston County (North Carolina) osteoarthritis project. Arthritis research & therapy, 2011. 13 : p. 1-9. Liikavainio, T., et al., Loading and gait symmetry during level and stair walking in asymptomatic subjects with knee osteoarthritis: importance of quadriceps femoris in reducing impact force during heel strike? The knee, 2007. 14 (3): p. 231-238. Stuart, E.A., et al., MatchIt: nonparametric preprocessing for parametric causal inference. Journal of statistical software, 2011. Das, S., et al., Next-generation genotype imputation service and methods. Nature genetics, 2016. 48 (10): p. 1284-1287. Chang, C.C., et al., Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience, 2015. 4 (1): p. s13742-015-0047-8. Bulik-Sullivan, B., et al., An atlas of genetic correlations across human diseases and traits. Nature genetics, 2015. 47 (11): p. 1236-1241. Morita Y, et al., Improved genetic prediction of the risk of knee osteoarthritis using the risk factor-based polygenic score. Arthritis Res Ther. 2023 Jun 12;25(1):103. Valdes, A.M. and T.D. Spector, Genetic epidemiology of hip and knee osteoarthritis. Nature Reviews Rheumatology, 2011. 7 (1): p. 23-32. Swain, S., et al., Comorbidities in osteoarthritis: a systematic review and meta‐analysis of observational studies. Arthritis care & research, 2020. 72 (7): p. 991-1000. Hellevik, A.I., et al., Incidence of total hip or knee replacement due to osteoarthritis in relation to thyroid function: a prospective cohort study (The Nord-Trøndelag Health Study). BMC Musculoskeletal Disorders, 2017. 18 (1): p. 1-9. Tagoe, C.E., et al., Association of anti-thyroid antibodies with radiographic knee osteoarthritis and chondrocalcinosis: a NHANES III study. Therapeutic Advances in Musculoskeletal Disease, 2021. 13 : p. 1759720X211035199. Gustafsson, K., et al., Health status of individuals referred to first-line intervention for hip and knee osteoarthritis compared with the general population: an observational register-based study. BMJ open, 2021. 11 (9): p. e049476. Kovari, E., et al. Comorbidity clusters in generalized osteoarthritis among female patients: A cross-sectional study. in Seminars in Arthritis and Rheumatism. 2020. Elsevier. Taylor SS et al., Prevalence of and characteristics associated with insomnia and obstructive sleep apnea among veterans with knee and hip osteoarthritis. BMC Musculoskelet Disord 2018;19:79 Lo, K., et al., Association between hypertension and osteoarthritis: a systematic review and meta-analysis of observational studies. Journal of Orthopaedic Translation, 2022. 32 : p. 12-20. Xie, Y., et al., Metabolic syndrome, hypertension, and hyperglycemia were positively associated with knee osteoarthritis, while dyslipidemia showed no association with knee osteoarthritis. Clinical Rheumatology, 2021. 40 : p. 711-724. Valdes, A.M., et al., Large scale meta-analysis of urinary C-terminal telopeptide, serum cartilage oligomeric protein and matrix metalloprotease degraded type II collagen and their role in prevalence, incidence and progression of osteoarthritis. Osteoarthritis and cartilage, 2014. 22 (5): p. 683-689. 42. Sharma, L., et al., Knee tissue lesions and prediction of incident knee osteoarthritis over 7 years in a cohort of persons at higher risk. Osteoarthritis and cartilage, 2017. 25 (7): p. 1068-1075. Tables Table 1 and 2 are available in the Supplementary Files section. Additional Declarations No competing interests reported. Supplementary Files Table.docx SupplmentaryTable.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-5391580","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":376958018,"identity":"696d3abe-d1ce-4b95-8789-7d1a00ce10ed","order_by":0,"name":"Kai-Xuan Lim","email":"","orcid":"","institution":"Far Eastern Memorial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Kai-Xuan","middleName":"","lastName":"Lim","suffix":""},{"id":376958020,"identity":"6f5a3dac-0096-4703-8e77-e9a5628bd970","order_by":1,"name":"Yun-Jer Shieh","email":"","orcid":"","institution":"Far Eastern Memorial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yun-Jer","middleName":"","lastName":"Shieh","suffix":""},{"id":376958021,"identity":"d1fd4386-2955-4b1e-b3f9-e6860e970853","order_by":2,"name":"Hsiu-Jung Liao","email":"","orcid":"","institution":"Far Eastern Memorial Hospital","correspondingAuthor":false,"prefix":"","firstName":"Hsiu-Jung","middleName":"","lastName":"Liao","suffix":""},{"id":376958022,"identity":"e804dc66-3faa-4203-93e7-eb8e8b610f31","order_by":3,"name":"Tzu-Hung Hsiao","email":"","orcid":"","institution":"Taichung Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Tzu-Hung","middleName":"","lastName":"Hsiao","suffix":""},{"id":376958024,"identity":"7cdf3753-cd84-4c9c-ab2a-bc7549b1ef4f","order_by":4,"name":"Chi-Ying F. Huang","email":"","orcid":"","institution":"National Yang Ming Chiao Tung University","correspondingAuthor":false,"prefix":"","firstName":"Chi-Ying","middleName":"F.","lastName":"Huang","suffix":""},{"id":376958026,"identity":"50c6382b-6e32-47fe-a31a-659aa7a52922","order_by":5,"name":"Yi-Ling Hsieh","email":"","orcid":"","institution":"Taichung Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Yi-Ling","middleName":"","lastName":"Hsieh","suffix":""},{"id":376958027,"identity":"edbac8a0-27d3-4ef5-adcb-04ff16a64c7e","order_by":6,"name":"Chih-Hung Chang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAt0lEQVRIiWNgGAWjYHACNhCRwM/MwHiAgYGZGB3MEC2SzQwMJGoxOECsFn7p88cefKixyzM+zvzgwMc2awb+9u4EvFok+5LZDWccSy42O8xmcHBmWzqDxJmzG/BqMTjDzCbN23AgcdthHobDvG2HGQwkconUsrmZZC0bmInVItnDbCYJ9EviDJBfZpxL5yHoF34exmcSwBBL7O8//PDBhzJrOf72XvxaMAAPacpHwSgYBaNgFGAFAPtGQ2k2GoMxAAAAAElFTkSuQmCC","orcid":"","institution":"Far Eastern Memorial Hospital","correspondingAuthor":true,"prefix":"","firstName":"Chih-Hung","middleName":"","lastName":"Chang","suffix":""},{"id":376958030,"identity":"38d3b60d-760d-49ed-91a1-24f9cd65156a","order_by":7,"name":"Cheng-Hung Lee","email":"","orcid":"","institution":"Taichung Veterans General Hospital","correspondingAuthor":false,"prefix":"","firstName":"Cheng-Hung","middleName":"","lastName":"Lee","suffix":""}],"badges":[],"createdAt":"2024-11-05 02:53:12","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-5391580/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-5391580/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":70405534,"identity":"8cb3ae12-1046-466b-a633-58015e8d5723","added_by":"auto","created_at":"2024-12-02 23:13:48","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":283147,"visible":true,"origin":"","legend":"\u003cp\u003eFlowchart of the study design\u003c/p\u003e","description":"","filename":"Picture1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/ad7f1f4d8a765304674d5e08.jpg"},{"id":70405533,"identity":"2d17b726-8e2e-4a6b-a13c-9ec0b5c61478","added_by":"auto","created_at":"2024-12-02 23:13:48","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":49041,"visible":true,"origin":"","legend":"\u003cp\u003eThe ROC curve plot displays various combinations of variables as follows: (a) shows the accuracy of predicting knee OA using PGS002767, BMI, and comorbidities. (b) shows the accuracy of predicting knee OA using only the PG002767, BMI or comorbidities.\u003c/p\u003e","description":"","filename":"Picture2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/bb25e153f9e9599dd49e208c.jpg"},{"id":70404838,"identity":"81b79698-b060-4a72-9594-ea63513b3592","added_by":"auto","created_at":"2024-12-02 23:05:48","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":15113,"visible":true,"origin":"","legend":"\u003cp\u003eThe forest plot for the OR of PGS002767 on gender subgroups analysis in TPMI.\u003c/p\u003e","description":"","filename":"Picture3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/e87f5b384c062d35462d67bf.jpg"},{"id":70405980,"identity":"ae544fcb-a50a-41e8-a3e2-7b479e227c41","added_by":"auto","created_at":"2024-12-02 23:21:51","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":802160,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/52dcdadf-6e0f-4901-9c3c-3884c4f995d5.pdf"},{"id":70404840,"identity":"b6190162-36c0-47d0-84f1-05ed3f9acbd4","added_by":"auto","created_at":"2024-12-02 23:05:48","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":18341,"visible":true,"origin":"","legend":"","description":"","filename":"Table.docx","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/d5bed6f417a9be59d6d8a519.docx"},{"id":70404841,"identity":"d0978748-d231-4fa1-b589-5be8ea7ef9cd","added_by":"auto","created_at":"2024-12-02 23:05:48","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":22356,"visible":true,"origin":"","legend":"","description":"","filename":"SupplmentaryTable.docx","url":"https://assets-eu.researchsquare.com/files/rs-5391580/v1/7c7ded34ee0e6ef1011f6429.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrating polygenic risk scores and constructing a predictive model to assess the risk of knee osteoarthritis in Taiwanese population","fulltext":[{"header":"Introduction","content":"\u003cp\u003eOsteoarthritis(OA) is\u0026nbsp;a chronic degenerative joint disease that affects more than 250 million\u0026nbsp;people worldwide, with more than 60 million cases in East Asia\u0026nbsp;[1-2].\u0026nbsp;Of all the joints affected, the knee is the most commonly affected joint by OA. Knee OA can lead to severe joint pain and limited functional activities, and can have a serious impact on quality of life and livelihood [3].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eKnee OA is known to be a multifactorial disease and many studies have shown that the risk factors for knee OA include heredity,\u0026nbsp;gender, aging, obesity and others [4]. These factors contribute to the heterogeneity in the onset and progression of\u0026nbsp;knee OA [5,6]. Among all these risk factors, genetic factors undoubtedly play an important role in knee OA. Previous twin and family studies suggest that the influence of genetic factors is approximately 40 percent for knee OA [7]. However, the development of OA is not solely attributed to the action of a single gene. Rather, it is more likely due to the interaction of multiple genes. In fact, many associated genes have demonstrated significant genetic influence in both genome-wide association studies(GWAS) and familial association studies [8-10].\u0026nbsp;Previous\u0026nbsp;meta-GWAS study\u0026nbsp;has identified\u0026nbsp;hundreds\u0026nbsp;of\u0026nbsp;single nucleotide polymorphism(SNP) loci associated with\u0026nbsp;knee OA [10]. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eAs the genetic architecture of knee OA is likely to involve many loci, each with small effect sizes,\u0026nbsp;polygenic score(PGS) or polygenic risk score (PRS) has been proposed to\u0026nbsp;be used as a risk prediction tool in clinical practice or in a screening program\u0026nbsp;by using information from multiple small effect SNPs to calculate the disease risk in individuals based on their genetic makeup[11,12]. PRS provided information that can be used to predict the risk of knee OA and help to develop personalized prevention and management plans for patients.\u0026nbsp;PGS have\u0026nbsp;demonstrated their predictive power in several diseases [13,14]. However, it should be noted that due to the limited sample sizes in Asian genetic databases, the development and validation of most PGS, including those related to\u0026nbsp;knee OA, have primarily been conducted using populations of European ancestry [15]. In the largest PGS\u0026nbsp;database, the PGS catalog, there are four PGS panels for knee\u0026nbsp;OA, and they\u0026nbsp;were primarily developed using the European population [16]. In situations where sample sizes remain limited, validating the applicability of published PGS panels to different ethnic groups appears to be a faster, more efficient, and cost-effective approach.\u0026nbsp;Early identification of patients with knee OA is crucial for appropriate treatment and improved outcomes.\u003c/p\u003e\n\u003cp\u003eOn the other hand, knee OA is a common health problem among older adults, which often occurs alongside other health conditions or comorbidities. These comorbidities include back pain [17], metabolic disorders [18], mental disorders [19], digestive disorders [20], endocrine disorders [21], cardiovascular disorders [22], and respiratory disorders [23]. Evidence suggests that these comorbidities may share underlying physiological and inflammatory mechanisms with knee OA [24]. Furthermore, some of these comorbidities may also share similar environmental risk factors such as occupation and living environment [25]. Additionally, they could be caused by similar biomechanical factors [26].Despite these observations, it is still unclear whether these comorbidities represent or share factors that influence knee OA. It is needed to better understand the relationship between these conditions and knee OA, which could lead to new approaches for preventing and treating this condition.\u003c/p\u003e\n\u003cp\u003eThis is the first time we have the opportunity to integrate the genetic profiles of a disease through the largest non-European cohort genetic database, namely the Taiwan Precision Medicine Initiative (TPMI). To the best of our knowledge, this is the first study to validate the most efficacious PGS panel from an external cohort for identifying knee OA patients in the Taiwanese population, utilizing data from the TPMI. Additionally, we sought to investigate the prevalent comorbidities associated with knee OA and integrate this information with PGS panels to establish a superior predictive model for knee OA.\u003c/p\u003e"},{"header":"Methods","content":"\u003cp\u003e\u003cstrong\u003eStudy populations\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe analyzed 43,686 participants of TPMI who were recruited from Taichung Veterans General Hospital (TCVGH) from 2009 to December 2022. The aim of TPMI is to establish a genetic database for the Taiwanese population through a collaborative effort between Academia Sinica and 16 medical centers in Taiwan. All subjects in the study provided informed consent prior to participation. The research protocol was approved by the Institutional Review Boards of TCVGH (IRB number: SE23188B). Patients with knee OA were identified by International Classification of Diseases, Ninth Revision (ICD-9) codes of 715.00(Osteoarthrosis, generalized, unspecified site), 715.10(Osteoarthrosis, localized, primary, unspecified site), 715.16(Osteoarthrosis, localized, primary, lower leg), 715.18(Osteoarthrosis, localized, primary, other specified site), 715.30(Osteoarthrosis, localized, not specified whether primary or secondary involving unspecified site), 715.36(Osteoarthrosis, localized, not specified whether primary or secondary involving lower leg), 715.38(Osteoarthrosis localized not specified whether primary or secondary involving other specified sites), 715.90(Osteoarthrosis, unspecified whether generalized or localized involving unspecified site), 715.96(Osteoarthrosis, unspecified whether generalized or localized involving lower leg), 715.98(Osteoarthrosis unspecified whether generalized or localized involving other specified site)\u0026nbsp;in\u0026nbsp;Taiwan\u0026apos;s National Health Insurance Research Database (NHIRD). As there has always been concern about over-coding in the NHIRD, we also used an additional knee replacement surgery code (64164B) and an articular injection code (39005C) to reconfirm that the patient actually had knee OA. Considering that some patients with OA of other joints (e.g. shoulder arthritis) or other types of knee arthritis (e.g. rheumatoid arthritis) were also treated with joint injections and replacements, our first author, KX,LIM, reviewed their outpatient medical records to confirm the diagnosis of knee OA.\u0026nbsp;Controls\u0026nbsp;were identified as participants without osteoarthritis (ICD-9 715.XX). We used the MatchIt package (method = \u0026lsquo;nearest\u0026rsquo;) in R to match the cases to the controls in a 1:4 ratio based on sex and age [27]. We then randomly divided the sex- and age-matched participants into two groups, the training group and the test group, in an 8:2 ratio (Figure1).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eComorbidities identification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe select 13 common diseases associated with knee OA that have been reported, including: diabetes (ICD-9: 250.X), hypertension (ICD-9: 401.X, 405.X), \u0026nbsp; dyslipidemia (ICD-9: 272.X), chronic obstructive pulmonary disease (COPD) (ICD-9: 490.X\u0026ndash;496.X), stroke (ICD-9: 433.X\u0026ndash;434.X), chronic kidney disease (CKD) (ICD-9: 585.X), backpain (ICD-9: 724.5, 724.9), gastroesophageal reflux disease (GERD) (ICD-9: 530.X), insomnia (ICD-9: 307.4, 327.X, 780.5), depression (ICD-9: 296.X), thyroid disorder (ICD-9: 240.X\u0026ndash;246.X), duodenal ulcer (ICD-9: 532), and colonic diverticula (ICD-9: 562).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGenotyping and calculating polygenic scores\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe genome data of TPMI participants were collected by Axiom Genome-Wide TPM 2.0 Array with 686,463 SNPs and imputed to the Genome Reference Consortium Human Build 37 (GRCh37) by Michigan Imputation Server [28]. We downloaded 4 published \u0026nbsp;knee OA PGS panels (PGS001192, PGS002729, PGS002749 and PGS002767) from the PGS catalog database. We calculated the polygenic score of each participant by the information of the SNPs and the weight of the SNPs provided by the PGS panel separately using PLINK 1.9 [29].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStatistical analysis\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe main statistical analyses were performed using R 4.2. We examined the distribution of the 4 PGS panels using the Shapiro-Wilk test, skewness and kurtosis. Next, we divided all our participants into 5 equal parts based on their PGS score, ranging from low to high: rank 1 (0-20%), rank 2 (20-40%), rank 3 (40-60%), rank 4 (60-80%), and rank 5 (80-100%). We used the independent t-test and chi-squared test to compare the continuous or categorical variables between the case and control groups. Univariate logistic regression was used to determine which PGS panel was the most appropriate for discriminating knee OA patients in our cohort and to analyze the association between knee OA and the most appropriate PGS panel. Multivariate logistic regression was used for two purposes in the training group. First, to analyze the association between knee OA and PRS with adjustment for age, gender, and all comorbidities. Second, to establish the predictive equation for the test group. Statistical significance was set at p\u0026lt;0.05.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConstructing the prediction equation for knee OA\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo construct the prediction equation and to account for the genetic influence of \u0026nbsp; other comorbidities, genetic correlation was used to analyze the association between knee OA and the comorbidities that showed significance in the multivariate logistic regression analysis. Genetic correlation was performed using LDSC software [30]. The GWAS summary statistics, which are required inputs for LDSC, were analyzed using the TPMI cohort with a case-to-control ratio of 1:1\u0026ndash;4 for each disease. The GWAS analyses were subject to consistent quality control (QC) using PLINK 1.9, including sample QC and SNP QC. The LD score of each SNP was calculated using LDSC in a 1 Mb window. The statistical significance for genetic correlation was set at p \u0026lt; 0.05/(n+1), where \u0026quot;n\u0026quot; depended on the number of remaining comorbidities.\u003c/p\u003e\n\u003cp\u003eThe receiver operating characteristic curve (ROC curve) and the area under the curve (AUC) were calculated to examine the predictive effect in the test group with or without BMI and comorbidities, respectively. The statistical significance was set at p\u0026lt;0.05.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eCharacteristics of study population\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFrom our database, there were 1,363 participants defined as the case group with knee OA and 42,323 participants defined as the control group without knee OA. The mean age of our studied cases is 73 years old and the male to female ratio is approximately 1:2.5. After matching for sex and age with a case to control ratio 1:4, the matching group consisted of 1,363 cases and 5,452 controls. Individuals in the matching group were then allocated to the training and test groups in an 8:2 ratio. The training group had 1,123 cases and 4,329 controls, while the test group had 240 cases and 1,123 controls. The distribution of 13 previously reported common comorbidities between groups are shown in Supplementary Table 1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePGS002767 is the most appropriate PGS panel to assess the genetic risk of knee OA\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe SNPs retention rates for the 4 PGS panels after imputation were as follows: 88% for PGS001192, 100% for PGS002729, 100% for PGS002749, and 91% for PGS002767. The distributions of the 4 PGS panels were approximately symmetric based on their skewness and kurtosis, although PGS002749 did not show a normal distribution according to the Shapiro-Wilk test . PGS002767 had the strongest association with our knee OA patients among the 4 PGS panels, as it showed a progressive increase in the disease odd ratio(OR) and a progressive decrease in p-value with increasing rank. PGS002749 had a secondary association because its OR at rank 5 showed significance, but its ORs at ranks 2 to 4 were not significantly different. The associations for knee OA between PGS001192 and PGS002729 were weak because its OR did not show a clear trend or was not significant. Therefore, PGS002767 was the most appropriate PGS panel for genetic risk assessment of knee OA in Taiwanese population (Supplementary Table 2).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBMI, GERD, insomnia, hypertension, and dyslipidemia are positively correlation with PGS002767\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUnivariable and multivariable logistic regression analysis were used to evaluate the relationship between PGS002767, BMI and comorbidities. Among the 13 previously reported common comorbidities, most variables including diabetes, hypertension, \u0026nbsp;dyslipidemia, COPD, stroke, CKD, back pain, GERD, insomnia, duodenal ulcer, and colonic diverticula \u0026nbsp;showed a significant association with knee OA. On the other hand, depression and thyroid disorders did not show a significant association. Further multivariable logistic regression analysis demonstrated that the associations of comorbidities were dominated by GERD [OR:2.17], insomnia [OR:1.57], dyslipidemia [OR:1.49] and hypertension [OR:1.47]. BMI also showed a significant \u0026nbsp; positive correlation with knee OA [OR:1.12] in multivariable logistic regression analysis. As a result, BMI, PGS002767 and four most important comorbidities were retained in our model for predicting knee OA, indicating that our model preserved and integrated the unique information from these variables.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBuilding the knee OA predictive model \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo avoid the inherent genetic effects of the 4 comorbidities mentioned above, we examined the genetic correlation of GERD, insomnia, hypertension, and dyslipidemia to assess the impact of including them in the predictive model. Table 2 shows that there was no significant genetic correlation between 4 comorbidities and knee OA. The area under curve(AUC) of our predictive model was 0.72, as shown in Figure 2a, when considering PGS002767, BMI, and comorbidities simultaneously. At the same time, we also observed the predictive effects of different variables. The AUC when considering only PGS002767 was 0.58, BMI was 0.66 and comorbidities was 0.64 (Figure 2b).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMale patients with knee OA are more predictable in model \u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFigure 3 showed that PGS002767 has better performance in predicting the risk of developing knee OA in male patients. In the analysis of gender subgroups, the ORs for the male group are more precise than those for the female group, with the confidence intervals(CI) observed to be narrower in the male group.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eUnderstanding the genetic etiology of knee OA is challenging because knee OA is a multifactorial disease with a polygenic genetic architecture. Many knee OA related GWAS and PRS studies have been conducted to evaluate the genetic influence, but most of them are limited to the European population. Recently, a study was published to evaluate the association between PRS and knee OA in the Japanese population[31]. To the best of our knowledge and literature review, this is the first study to validate the appropriateness of PGS from external cohort for the knee OA in the Taiwanese population\u0026nbsp;through the largest non-European cohort genetic database, Taiwan Precision Medicine Initiative (TPMI\u003c/p\u003e\n\u003cp\u003ePGS002767 showed the strongest association with knee OA in the Taiwanese population among the other PGS panels in our study. In the context of genetic differences in knee OA susceptibility between European and Asian populations, PGS002767 contains information from 1,052,275 SNPs, which means that PGS002767 has effectively captured information from almost the entire genome [16]. This may be the main reason why PGS002767 showed the best performance in our study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSecond, the SNPs used to construct PGS002729 and PGS002749 were obtained from the same meta-analysis GWAS, which included 826,690 individuals from 9 different populations [10]. However, the representation of East Asian populations in the meta-analysis GWAS study was minority, which may explain the relatively low correlation between PGS002729, PGS002749, and our population [11]. The sparse PRS model used to construct PGS001192 eliminates the small effect size of SNPs. This method may not be consistent with the characteristics of knee OA which is caused by a lot of small effect loci [32]. Therefore, here are the reasons that may explain why PGS002767 is more suitable for our population compared to other PGS panels. On the other hand, using PGS002767 to assess the risk of knee OA in the Taiwanese population gives similar OR/AUC to studies using their own PGS panels [11]. This suggests that genes only influence only a portion of knee OA incidence, highlighting the need for further consideration of acquired factors.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eKnee OA involves both genetic and environmental factors and is associated with multiple chronic diseases [4]. In our hospital-based cohort, we investigated the impact of comorbidities and BMI on the development of knee OA. Most previously reported comorbidities show a significant association with knee OA in univariate analysis, except for depression and thyroid disorder. The meta-analysis study showed that the prevalence ratio of comorbidities in the upper gastrointestinal, psychological, cardiovascular, musculoskeletal, respiratory, and endocrine comorbidities associated with knee OA is consistent with our findings [33]. The lack of a significant association between depression and knee OA in our cohort \u0026nbsp;is mainly due to the limited sample size. Fewer patients with mental disorders were recruited in the TPMI. \u0026nbsp;Without professional assistance, it is difficult for patients with mental disorders to provide informed consent for research. \u0026nbsp;Besides, a population-based cohort study conducted in Nord-Tr\u0026oslash;ndelag also suggests that there is no correlation between thyroid-stimulating hormone and knee OA incidence [34]. Furthermore, a cohort study further investigated the relationship between two thyroid antibodies, anti-thyroid peroxidase antibody(TPOAb) and anti-thyroglobulin antibody(TgAb), and knee OA as well as chondrocalcinosis. The study found that only TPOAb above a certain concentration was associated with chondrocalcinosis, while neither TPOAb nor TgAb was associated with knee OA [35].\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eIn multivariate analysis, GERD, insomnia, hypertension, and dyslipidemia overshadowed the effects of other comorbidities. Knee OA, once considered a primarily non-inflammatory condition, is now recognized to be associated with chronic low-grade inflammation. This inflammatory component, although subtle, has the potential to trigger various systemic effects. Notably, it has been suggested that this chronic inflammation may contribute to GERD, insomnia , and hypertension. However, it is questionable whether patients with knee OA have an increased risk of gastrointestinal complications due to the frequent use of anti-inflammatory drugs. A population-based cohort study showed that gastrointestinal complications remained significantly associated with knee OA even after excluding patients using anti-inflammatory drugs [36]. Another study also indicates that GERD has a significant presence in two of five comorbidity clusters in OA [37].\u0026nbsp;Insomnia is the only related comorbidity in our study that is associated with psychological factors. Insomnia can arise as a consequence of worsening OA pain, which tends to intensify during nighttime hours when native cortisol production reaches its lowest point. In 2018, a comprehensive study investigated the sleep patterns of OA patients, revealing that approximately 53% of them exhibited symptoms of insomnia [38]. A meta-analysis reveals that hypertension causes knee OA not only through inflammation but also through significant structural damage to the knee joint, and this effect remains even after adjustment for sex and BMI [39]. The relationship between dyslipidemia and knee OA remains uncertain. Two recent systematic reviews found no association between dyslipidemia and knee OA due to the high heterogeneity of the studies they reviewed [40]. However, in our study, dyslipidemia \u0026nbsp;may represent a partial characteristic of cardiovascular disease and metabolic syndrome. On the other hand, BMI maintained its consistent effect in both univariate and multivariate analyses, independent of the effects of comorbidities. Furthermore, extensive research has firmly established the influence of BMI on knee OA. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFrom our findings, we ultimately incorporated BMI, four specific comorbidities and combined the effects of PGS002767 to construct the knee OA prediction model. As a culmination of these factors, the resulting knee OA prediction model demonstrated an AUC value of 0.72. This value indicates a respectable level of predictive ability for identifying knee OA cases. Previous research has utilized various factors to develop predictive models for the incidence of knee OA, including knee injury, knee surgery, knee joint space narrowing, family history, occupation, cartilage-derived biomarkers, education level, physical activity, hand OA, and pain [11,41,42]. These models achieved AUC values ranging from 0.59 to 0.83. Among these models, they require detailed medical history, family history, additional imaging studies or blood tests. As a result, they are not readily applicable to the general population. The advantage of our model is that it achieves relatively good predictive performance using readily available clinical information without requiring extensive additional testing or population surveys. Once a patient in your hospital is enrolled in TPMI and their genetic data is available, you can use the genetic information, BMI, and comorbidities to screen the patient. This individual\u0026apos;s risk of disease can help the patient and physician decide whether or not to intervene early in treatment.\u003c/p\u003e\n\u003cp\u003eTo our knowledge, our model is the first to incorporate comorbidities as a factor in the prediction of knee OA. Furthermore, in order to validate the independence of these comorbidities, we performed genetic correlation to confirm that these comorbidities were not genetically correlated. The AUC of the model considering only PGS002767 was 0.58, which is greater than or equal to previous studies considering genetics as a predictive factor, with AUC values ranging from 0.51 to 0.58 [11,41,42]. This result demonstrates that PGS002767 has excellent performance in the Taiwanese population, rivaling its efficacy in European populations. In addition, we found that male patients were particularly suitable for the use of \u0026nbsp; PGS002767. Based on our results, we have provided a simple and convenient model for clinical practitioners to predict knee OA patients.\u003c/p\u003e\n\u003cp\u003eOur study encountered several limitations. First, our model\u0026rsquo;s development relied on a case-control study design instead of a cohort study, which may affect its generalizability. Second, we did not fully impute the PGS002767 as its SNPs retention rate is 91% compared to our array, which may have an impact on its true effect. Third, our study could be susceptible to selection bias as we only included patients from a single tertiary medical center, raising the possibility that our findings primarily pertain to individuals with more severe symptoms. Fourth, our study lacks certain important factors for predicting knee OA, such as occupational stress or activity frequency, we hope that over time, as the TMPI program is more widely promoted and the integration of information is more fully consolidated, we will be able to incorporate these factors into our model. Finally, the sample size in the training group was relatively small, with only 1,123 cases, which may affect the precision of the model parameters. Furthermore, when assessing the relationship between knee OA and prevalent comorbidities, the time of initial diagnosis of knee OA preceded the establishment of the TPMI database, making it challenging to accurately determine the tine of disease onset.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eIn this study, we confirm the use of PGS002767 in predicting knee OA in the Taiwanese population. In addition, we develop a predictive model for knee OA that includes PGS002767, BMI and four specific comorbidities. Our results show that this model has a moderate level of predictive efficacy, making it a practical tool for clinical application.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe thank all the participants and investigators from Taiwan Precision Medicine Initiative which was funded by Academia Sinica (40-05-GMM, AS-GC-110-MD02 and 236e-1100202) and National Development Fund, Executive Yuan(NSTC111-3114-Y-001-001)\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eContributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConception and design of the study: CHL, CHC, CYH\u0026nbsp;and\u0026nbsp;YLH. Acquisition of data: KXL, YCH, THH. Analysis and interpretation of data: KXL, YCH, THH. Drafting or revising the manuscript: KXL, YCH, THH, and HJL. All authors have approved the final article.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eRole of the funding source\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis research was funded by the Ministry of Science and Technology (MOST 111-2314-B-418-004, MOST 111-2321-B-418-002-, NSTC 112-2321-B-418-004- and NSTC 112-2314-B-418-010-MY3), Far Eastern Memorial Hospital (FEMH; 2022-C-042, 2022-C-123, 2023-C-124, 2023-C-064 and 2024-C-105), National Yang Ming Chiao Tung University\u0026ndash;FEMH Cooperation (111DN12, 111DN22, 112DN13, 112DN22, 113DN09 and 113DN16), and National Taiwan University Hospital\u0026ndash;FEMH Cooperation (111-FTN0012).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interests\u003c/strong\u003e\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe authors declare no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSafiri, S., et al., Global, regional and national burden of osteoarthritis 1990-2017: a systematic analysis of the Global Burden of Disease Study 2017. Annals of the rheumatic diseases, 2020. \u003cstrong\u003e79\u003c/strong\u003e(6): p. 819-828.\u003c/li\u003e\n\u003cli\u003eBerenbaum, F., Osteoarthritis as an inflammatory disease (osteoarthritis is not osteoarthrosis!). Osteoarthritis and cartilage, 2013. \u003cstrong\u003e21\u003c/strong\u003e(1): p. 16-21.\u003c/li\u003e\n\u003cli\u003eVitaloni, M., et al., Global management of patients with knee osteoarthritis begins with quality of life assessment: a systematic review. BMC musculoskeletal disorders, 2019. \u003cstrong\u003e20\u003c/strong\u003e(1): p. 1-12.\u003c/li\u003e\n\u003cli\u003eSuri, P., D.C. Morgenroth, and D.J. Hunter, Epidemiology of osteoarthritis and associated comorbidities. PM\u0026amp;R, 2012. \u003cstrong\u003e4\u003c/strong\u003e: p. S10-S19.\u003c/li\u003e\n\u003cli\u003eNeogi, T., The epidemiology and impact of pain in osteoarthritis. Osteoarthritis and cartilage, 2013. \u003cstrong\u003e21\u003c/strong\u003e(9): p. 1145-1153.\u003c/li\u003e\n\u003cli\u003eWei, J., et al., Risk factor heterogeneity for medial and lateral compartment knee osteoarthritis: analysis of two prospective cohorts. Osteoarthritis and cartilage, 2019. \u003cstrong\u003e27\u003c/strong\u003e(4): p. 603-610.\u003c/li\u003e\n\u003cli\u003eSpector, T.D. and A.J. MacGregor, Risk factors for osteoarthritis: genetics. Osteoarthritis and cartilage, 2004. \u003cstrong\u003e12\u003c/strong\u003e: p. 39-44.\u003c/li\u003e\n\u003cli\u003eKarlsson, C., et al., Genome-wide expression profiling reveals new candidate genes associated with osteoarthritis. Osteoarthritis and cartilage, 2010. \u003cstrong\u003e18\u003c/strong\u003e(4): p. 581-592.\u003c/li\u003e\n\u003cli\u003eNeame, R., et al., Genetic risk of knee osteoarthritis: a sibling study. Annals of the rheumatic diseases, 2004. \u003cstrong\u003e63\u003c/strong\u003e(9): p. 1022-1027.\u003c/li\u003e\n\u003cli\u003eBoer, C.G., et al., Deciphering osteoarthritis genetics across 826,690 individuals from 9 populations. Cell, 2021. \u003cstrong\u003e184\u003c/strong\u003e(18): p. 4784-4818. e17.\u003c/li\u003e\n\u003cli\u003eSedaghati-Khayat B, Boer CG, Runhaar J, Bierma-Zeinstra SMA, Broer L, Ikram MA, Zeggini E, Uitterlinden AG, van Rooij JGJ, van Meurs JBJ. Risk Assessment for Hip and Knee Osteoarthritis Using Polygenic Risk Scores. Arthritis Rheumatol. 2022 Sep;74(9):1488-1496.\u003c/li\u003e\n\u003cli\u003eChoi, S.W., T.S.-H. Mak, and P.F. O\u0026rsquo;Reilly, Tutorial: a guide to performing polygenic risk score analyses. Nature protocols, 2020. \u003cstrong\u003e15\u003c/strong\u003e(9): p. 2759-2772.\u003c/li\u003e\n\u003cli\u003eLi, Z., et al., Polygenic Risk Scores have high diagnostic capacity in ankylosing spondylitis. Annals of the rheumatic diseases, 2021. \u003cstrong\u003e80\u003c/strong\u003e(9): p. 1168-1174.\u003c/li\u003e\n\u003cli\u003eMavaddat, N., et al., Polygenic risk scores for prediction of breast cancer and breast cancer subtypes. The American Journal of Human Genetics, 2019. \u003cstrong\u003e104\u003c/strong\u003e(1): p. 21-34.\u003c/li\u003e\n\u003cli\u003eLambert, S.A., et al., The Polygenic Score Catalog as an open database for reproducibility and systematic evaluation. Nature Genetics, 2021. \u003cstrong\u003e53\u003c/strong\u003e(4): p. 420-425.\u003c/li\u003e\n\u003cli\u003eMars, N., et al., Systematic comparison of family history and polygenic risk across 24 common diseases. The American Journal of Human Genetics, 2022. \u003cstrong\u003e109\u003c/strong\u003e(12): p. 2152-2162.\u003c/li\u003e\n\u003cli\u003eBurnett, D.R., et al., A retrospective study of the relationship between back pain and unilateral knee osteoarthritis in candidates for total knee arthroplasty. Orthopaedic Nursing, 2012. \u003cstrong\u003e31\u003c/strong\u003e(6): p. 336-343.\u003c/li\u003e\n\u003cli\u003eGandhi, R., et al., Asian ethnicity and the prevalence of metabolic syndrome in the osteoarthritic total knee arthroplasty population. The Journal of arthroplasty, 2010. \u003cstrong\u003e25\u003c/strong\u003e(3): p. 416-419.\u003c/li\u003e\n\u003cli\u003eParmelee, P.A., C.A. Tighe, and N.D. Dautovich, Sleep disturbance in osteoarthritis: linkages with pain, disability, and depressive symptoms. Arthritis care \u0026amp; research, 2015. \u003cstrong\u003e67\u003c/strong\u003e(3): p. 358-365.\u003c/li\u003e\n\u003cli\u003eKing, L., et al., Are medical comorbidities contributing to the use of opioid analgesics in patients with knee osteoarthritis? Osteoarthritis and Cartilage, 2020. \u003cstrong\u003e28\u003c/strong\u003e(8): p. 1030-1037.\u003c/li\u003e\n\u003cli\u003evan Dijk, G.M., et al., Comorbidity, limitations in activities and pain in patients with osteoarthritis of the hip or knee. BMC musculoskeletal disorders, 2008. \u003cstrong\u003e9\u003c/strong\u003e(1): p. 1-10.\u003c/li\u003e\n\u003cli\u003eKim, H.S., et al., Association between knee osteoarthritis, cardiovascular risk factors, and the Framingham Risk Score in South Koreans: a cross-sectional study. PloS one, 2016. \u003cstrong\u003e11\u003c/strong\u003e(10): p. e0165325.\u003c/li\u003e\n\u003cli\u003eWshah, A., et al., Prevalence of osteoarthritis in individuals with COPD: a systematic review. International journal of chronic obstructive pulmonary disease, 2018: p. 1207-1216.\u003c/li\u003e\n\u003cli\u003eValdes, A., Metabolic syndrome and osteoarthritis pain: common molecular mechanisms and potential therapeutic implications. 2020, Elsevier. p. 7-9.\u003c/li\u003e\n\u003cli\u003eCallahan, L.F., et al., Associations of educational attainment, occupation and community poverty with knee osteoarthritis in the Johnston County (North Carolina) osteoarthritis project. Arthritis research \u0026amp; therapy, 2011. \u003cstrong\u003e13\u003c/strong\u003e: p. 1-9.\u003c/li\u003e\n\u003cli\u003eLiikavainio, T., et al., Loading and gait symmetry during level and stair walking in asymptomatic subjects with knee osteoarthritis: importance of quadriceps femoris in reducing impact force during heel strike? The knee, 2007. \u003cstrong\u003e14\u003c/strong\u003e(3): p. 231-238.\u003c/li\u003e\n\u003cli\u003eStuart, E.A., et al., MatchIt: nonparametric preprocessing for parametric causal inference. Journal of statistical software, 2011.\u003c/li\u003e\n\u003cli\u003eDas, S., et al., Next-generation genotype imputation service and methods. Nature genetics, 2016. \u003cstrong\u003e48\u003c/strong\u003e(10): p. 1284-1287.\u003c/li\u003e\n\u003cli\u003eChang, C.C., et al., Second-generation PLINK: rising to the challenge of larger and richer datasets. Gigascience, 2015. \u003cstrong\u003e4\u003c/strong\u003e(1): p. s13742-015-0047-8.\u003c/li\u003e\n\u003cli\u003eBulik-Sullivan, B., et al., An atlas of genetic correlations across human diseases and traits. Nature genetics, 2015. \u003cstrong\u003e47\u003c/strong\u003e(11): p. 1236-1241.\u003c/li\u003e\n\u003cli\u003eMorita Y, et al., Improved genetic prediction of the risk of knee osteoarthritis using the risk factor-based polygenic score. Arthritis Res Ther. 2023 Jun 12;25(1):103.\u003c/li\u003e\n\u003cli\u003eValdes, A.M. and T.D. Spector, Genetic epidemiology of hip and knee osteoarthritis. Nature Reviews Rheumatology, 2011. \u003cstrong\u003e7\u003c/strong\u003e(1): p. 23-32.\u003c/li\u003e\n\u003cli\u003eSwain, S., et al., Comorbidities in osteoarthritis: a systematic review and meta‐analysis of observational studies. Arthritis care \u0026amp; research, 2020. \u003cstrong\u003e72\u003c/strong\u003e(7): p. 991-1000.\u003c/li\u003e\n\u003cli\u003eHellevik, A.I., et al., Incidence of total hip or knee replacement due to osteoarthritis in relation to thyroid function: a prospective cohort study (The Nord-Tr\u0026oslash;ndelag Health Study). BMC Musculoskeletal Disorders, 2017. \u003cstrong\u003e18\u003c/strong\u003e(1): p. 1-9.\u003c/li\u003e\n\u003cli\u003eTagoe, C.E., et al., Association of anti-thyroid antibodies with radiographic knee osteoarthritis and chondrocalcinosis: a NHANES III study. Therapeutic Advances in Musculoskeletal Disease, 2021. \u003cstrong\u003e13\u003c/strong\u003e: p. 1759720X211035199.\u003c/li\u003e\n\u003cli\u003eGustafsson, K., et al., Health status of individuals referred to first-line intervention for hip and knee osteoarthritis compared with the general population: an observational register-based study. BMJ open, 2021. \u003cstrong\u003e11\u003c/strong\u003e(9): p. e049476.\u003c/li\u003e\n\u003cli\u003eKovari, E., et al. Comorbidity clusters in generalized osteoarthritis among female patients: A cross-sectional study. in Seminars in Arthritis and Rheumatism. 2020. Elsevier.\u003c/li\u003e\n\u003cli\u003eTaylor SS et al., Prevalence of and characteristics associated with insomnia and obstructive sleep apnea among veterans with knee and hip osteoarthritis. BMC Musculoskelet Disord 2018;19:79\u003c/li\u003e\n\u003cli\u003eLo, K., et al., Association between hypertension and osteoarthritis: a systematic review and meta-analysis of observational studies. Journal of Orthopaedic Translation, 2022. \u003cstrong\u003e32\u003c/strong\u003e: p. 12-20.\u003c/li\u003e\n\u003cli\u003eXie, Y., et al., Metabolic syndrome, hypertension, and hyperglycemia were positively associated with knee osteoarthritis, while dyslipidemia showed no association with knee osteoarthritis. Clinical Rheumatology, 2021. \u003cstrong\u003e40\u003c/strong\u003e: p. 711-724.\u003c/li\u003e\n\u003cli\u003eValdes, A.M., et al., \u003cem\u003eLarge scale meta-analysis of urinary C-terminal telopeptide, serum cartilage oligomeric protein and matrix metalloprotease degraded type II collagen and their role in prevalence, incidence and progression of osteoarthritis.\u003c/em\u003e Osteoarthritis and cartilage, 2014. \u003cstrong\u003e22\u003c/strong\u003e(5): p. 683-689.\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003e42. Sharma, L., et al., \u003cem\u003eKnee tissue lesions and prediction of incident knee osteoarthritis over 7 years in a cohort of persons at higher risk.\u003c/em\u003e Osteoarthritis and cartilage, 2017. \u003cstrong\u003e25\u003c/strong\u003e(7): p. 1068-1075.\u003c/p\u003e"},{"header":"Tables","content":"\u003cp\u003eTable 1 and 2 are available in the Supplementary Files section.\u003c/p\u003e"}],"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":"Knee Osteoarthritis, Osteoarthritis, Polygenic risk scores, osteoarthritis risk","lastPublishedDoi":"10.21203/rs.3.rs-5391580/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-5391580/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eObjective:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eKnee osteoarthritis (OA) is a prevalent degenerative joint disease that can lead to severe limitations in daily activities and function. Genetic components contribute significantly to knee OA and it always coexists with other comorbid conditions. Our goal is to identify the most effective PGS panel that can be applied in the Taiwanese population. We also aim to develop a predictive model for the risk of knee OA by integrating the polygenic risk scores(PGS) and comorbidity information.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethod\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe analyzed a total of 43,686 patients from the Taiwan Precision Medicine Initiative (TPMI) database and 1,363 cases of knee OA were identified. We validated which of the four published PGS panels was the most appropriate for predicting the risk of knee OA in the Taiwanese population. We then integrate the validated PGS panel and common comorbidities with knee OA to evaluate the predictive performance of the model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe found that PGS002767 showed the strongest association with our knee OA patients (OR: 1.59 at rank 5, OR: 1.46 at rank 4 compared to rank 1). GERD (OR:2.17), insomnia (OR:1.57), dyslipidemia (OR:1.49), and hypertension (OR:1.47) were retained in the final model. The AUC of our final model which included PGS002767, BMI and comorbidities was 0.72.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusion:\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo the best of our knowledge, this is the first study to validate the applicability of PGS panels in knee OA patient of the Taiwanese population. We create a predictive model for knee OA allowing clinicians to quickly and cost-effectively screen patients at risk for developing knee OA.\u003c/p\u003e","manuscriptTitle":"Integrating polygenic risk scores and constructing a predictive model to assess the risk of knee osteoarthritis in Taiwanese population","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-12-02 23:05:44","doi":"10.21203/rs.3.rs-5391580/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":"919b74bb-ef13-44af-b774-ed602b7f05e9","owner":[],"postedDate":"December 2nd, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-12-02T23:05:46+00:00","versionOfRecord":[],"versionCreatedAt":"2024-12-02 23:05:44","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-5391580","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-5391580","identity":"rs-5391580","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}