Using Polygenic Risk Scores to Aid Diagnosis of Patients With Early Inflammatory Arthritis: Results From the Norfolk Arthritis Register

OBJECTIVE

There is growing evidence that genetic data are of benefit in the rheumatology outpatient setting by aiding early diagnosis. A genetic probability tool (G-PROB) has been developed to aid diagnosis has not yet been tested in a real-world setting. Our aim was to assess whether G-PROB could aid diagnosis in the rheumatology outpatient setting using data from the Norfolk Arthritis Register (NOAR), a prospective observational cohort of patients presenting with early inflammatory arthritis.

METHODS

Genotypes and clinician diagnoses were obtained from patients from NOAR. Six G-probabilities (0%-100%) were created for each patient based on known disease-associated odds ratios of published genetic risk variants, each corresponding to one disease of rheumatoid arthritis, systemic lupus erythematosus, psoriatic arthritis, spondyloarthropathy, gout, or "other diseases." Performance of the G-probabilities compared with clinician diagnosis was assessed.

RESULTS

We tested G-PROB on 1,047 patients. Calibration of G-probabilities with clinician diagnosis was high, with regression coefficients of 1.047, where 1.00 is ideal. G-probabilities discriminated clinician diagnosis with pooled areas under the curve (95% confidence interval) of 0.85 (0.84-0.86). G-probabilities <5% corresponded to a negative predictive value of 96.0%, for which it was possible to suggest >2 unlikely diseases for 94% of patients and >3 for 53.7% of patients. G-probabilities >50% corresponded to a positive predictive value of 70.4%. In 55.7% of patients, the disease with the highest G-probability corresponded to clinician diagnosis.

CONCLUSION

G-PROB converts complex genetic information into meaningful and interpretable conditional probabilities, which may be especially helpful at eliminating unlikely diagnoses in the rheumatology outpatient setting.

Genetic Studies Investigating Susceptibility to Psoriatic Arthritis: A Narrative Review

PURPOSE

Approximately 30% of patients with psoriasis will develop psoriatic arthritis (PsA), leading to a decreased quality of life for the patient caused by increasing disability and additional health complications. The identification of risk factors for the development of PsA would facilitate the development of risk prediction models in which patients with psoriasis at high risk of developing PsA could be targeted in a stratified medicine approach, enabling early intervention and treatment. PsA is known to have a genetic contribution to susceptibility, and the identification of genetic risk factors that differentiate PsA from cutaneous-only psoriasis is a key area of research. This narrative review summarizes the discovery of genetic risk factors and, with the aid of a primer on risk prediction models, discusses their potential role for the classification of PsA risk and diagnosis.

METHODS

All relevant research articles were identified through searches of the PubMed database for literature published up until December 2022. Search terms included psoriatic arthritis, genetic susceptibility, genetic association, genome-wide association study, GWAS, prediction, and polygenic risk score.

FINDINGS

The current literature reveals considerable overlap between the genetic susceptibility loci for PsA and psoriasis. Several PsA-specific genetic risk factors have been reported, and most notably these implicate the HLA-B and IL23R genes. Efforts to include genetic risk factors in prediction models for the development of PsA have reported good discrimination.

IMPLICATIONS

Key messages emerging from this narrative are as follows: the limited number of PsA-specific susceptibility loci reported to date suggest larger studies are required, facilitated by international collaboration, to achieve the power to detect further genetic factors; the early promising results for genetic-based risk prediction require further validation in independent datasets; and risk prediction models combining clinical and genetic risk factors have yet to be explored.

POS0395 EXPLORING THE POTENTIAL OF GENOMIC RISK PREDICTION FOR CORONARY ARTERY DISEASE IN PATIENTS WITH RHEUMATOID ARTHRITIS

Background

Patients with rheumatoid arthritis (RA) have a higher prevalence of coronary artery disease (CAD) than the general population which contributes to early mortality. However, CAD screeing tools developed in the general population are less effective for estimating CAD risk in RA patients. This is mainly due to the differing contribution from traditional risk factors and the contribution from disease-specific factors. Understanding of the genetic basis of CAD has improved over recent years and shows promise for improving risk prediction in the form of genetic risk scores (GRs), in particular with the development of the metaGRS approach, which combines multiple polygenic risk scores.

Objectives

This study hypothesise that the metaGRS approach can help us improve CAD risk prediction in patients with RA.

Methods

Patients were recruited from the Norfolk Arthritis Register (NOAR), a longitudinal observational study focused on the cause and outcome of inflammatory polyarthritis. Analysis was restricted to patients who satisfied the 2010 ACR criteria cumulatively over five years and had detailed clinical history at baseline and follow-up for ten years. We developed a prediction model based on traditional risk factors[1], and explored the inclusion of a metaGRS. We used a meta-analytic approach to calculate a new metaGRS for CAD, using the effect-sizes from three large-scale, genome-wide, and targeted GRs derived from 1,745,179 [2], 6,630,150 [3], and 40,079 SNPs [4]. We tested the metaGRS in combination with available data on traditional risk factors in a subset of patients with available genetic data. Cox proportional hazards models were used to derive risk equations for evaluation of 10-year risk of CAD. We applied multiple imputations with chained equations to replace missing values. Calibration and discrimination were determined in a separate cohort of 423 individuals.

Results

A total of 2123 patients were included in the analysis with 136 incident cases of self-reported CAD (defined as a composite outcome of myocardial infarction, angina, heart attack, arrhythmia, angioplasty, and coronary artery bypass grafting).The model using only traditional risk factors achieved an AUC of 0.81 (95% CI 0.80, 0.82), with a calibration slope of 1.10, and explained approximately 71% (95% CI 69, 72%) of the variance of the outcome. The hazard ratio for age was found to be 1.00 (95% CI 0.99, 1.01) indicating risk remains the same across all age groups. Inclusion of a CAD metaGRS improves the AUC to 0.82 (95% CI 0.80, 0.83), explains more of the variance at 81% (95% CI 79, 82%) but worsens calibration slope to 0.93. A likelihood ratio test indicates that the integrated model is a better fit (p = 0.04).

Conclusion

An integrated risk score, that combines traditional risk factors with a metaGRS, improves CAD prediction in patients with RA. Further research is required to better understand the role of heritable components contributing to CAD risk in RA patients. By refining the underlying GRS, we hope to further improve risk prediction, through this integrated approach.

References

[1]Hippisley-Cox, Julia, Carol Coupland, and Peter Brindle. “Development and validation of QRISK3 risk prediction algorithms to estimate future risk of cardiovascular disease: prospective cohort study.” bmj 357 (2017).

[2]Inouye M, Abraham G, Nelson CP, Wood AM, Sweeting MJ, Dudbridge F, et al. Genomic Risk Prediction of Coronary Artery Disease in 480,000 Adults: Implications for Primary Prevention. J Am Coll Cardiol. 2018;72(16):1883–93.

[3]Khera A V., Chaffin M, Aragam KG, Haas ME, Roselli C, Choi SH, et al. Genome-wide polygenic scores for common diseases identify individuals with risk equivalent to monogenic mutations. Nat Genet [Internet]. 2018;50(9):1219–24. Available from: http://dx.doi.org/10.1038/s41588-018-0183-z

[4]Elliott J, Bodinier B, Bond TA, Chadeau-Hyam M, Evangelou E, Moons KGM, et al. Predictive Accuracy of a Polygenic Risk Score-Enhanced Prediction Model vs a Clinical Risk Score for Coronary Artery Disease. JAMA - J Am Med Assoc. 2020;323(7):636–45.

Disclosure of Interests

None declared.

Comparative genetic analysis of psoriatic arthritis and psoriasis for the discovery of genetic risk factors and risk prediction modelling

Objectives

Psoriatic arthritis (PsA) has a strong genetic component, and the identification of genetic risk factors could help identify the ~30% of psoriasis patients at high risk of developing PsA. Our objectives were to identify genetic risk factors and pathways that differentiate PsA from cutaneous‐only psoriasis (PsC) and to evaluate the performance of PsA risk prediction models.

Methods

Genome‐wide meta‐analyses were conducted separately for 5,065 patients with PsA and 21,286 healthy controls and separately for 4,340 patients with PsA and 6,431 patients with PsC. The heritability of PsA was calculated as a single‐nucleotide polymorphism (SNP)–based heritability estimate (h²_SNP) and biologic pathways that differentiate PsA from PsC were identified using Priority Index software. The generalizability of previously published PsA risk prediction pipelines was explored, and a risk prediction model was developed with external validation.

Results

We identified a novel genome‐wide significant susceptibility locus for the development of PsA on chromosome 22q11 (rs5754467; P = 1.61 × 10⁻⁹), and key pathways that differentiate PsA from PsC, including NF‐κB signaling (adjusted P = 1.4 × 10⁻⁴⁵) and Wnt signaling (adjusted P = 9.5 × 10⁻⁵⁸). The heritability of PsA in this cohort was found to be moderate (h²_SNP = 0.63), which was similar to the heritability of PsC (h²_SNP = 0.61). We observed modest performance of published classification pipelines (maximum area under the curve 0.61), with similar performance of a risk model derived using the current data.

Conclusion

Key biologic pathways associated with the development of PsA were identified, but the investigation of risk classification revealed modest utility in the available data sets, possibly because many of the PsC patients included in the present study were receiving treatments that are also effective in PsA. Future predictive models of PsA should be tested in PsC patients recruited from primary care.

OA28 Exploring the potential of polygenic risk scores for predicting coronary artery disease in patients with rheumatoid arthritis

Background/Aims

Patients with rheumatoid arthritis (RA) have a higher prevalence of coronary artery disease (CAD) than the general population which contributes to early mortality. Current screening tools for CAD, which are developed in the general population, are less effective for estimating CAD risk in patients with RA. This reduced performance is mainly due to the differing contribution from traditional risk factors and the contribution from disease-specific factors. Our understanding of the genetic basis of CAD has improved over recent years and shows promise for improving risk prediction in the form of polygenic risk scores (PRS). We hypothesise that PRS can help us improve CAD risk prediction in patients with RA by providing more accurate models of risk.

Methods

Patients were recruited from the Norfolk Arthritis Register (NOAR), a detailed community-based longitudinal observational study focused on the cause and outcome of inflammatory polyarthritis, between 1990 and 2017. Analysis was restricted to patients who satisfied the 2010 ACR criteria cumulatively over five years and had detailed clinical history at baseline and follow-up. We developed a prediction model based on traditional risk factors and explored the inclusion of a PRS (49K SNPs) in a subset of patients with available genetic data. Cox proportional hazards models were used to derive risk equations for evaluation of 10-year risk of CAD. We applied multiple imputations with chained equations using the Random Forest algorithm to replace missing values. Measures of calibration and discrimination were determined in the validation cohort of 423 individuals.

Results

A total of 2123 patients were included in the analysis with 136 incident cases of self-reported CAD. The model using only traditional risk factors achieved an AUC of 0.72 (95% CI 0.71, 0.73), with a calibration slope of 1.03, and explained approximately 50% (95% CI 47, 52%) of the variance of the outcome. We found that being male reduces the risk by a factor of 0.82 (95% CI 0.68, 1.00). The hazard ratio for age was found to be 1.00 (95% CI 0.99, 1.01) indicating risk remains the same across all age groups. Inclusion of a CAD PRS increased the performance with an AUC of 0.76 (95% CI 0.75, 0.77), explained variance of 53% (95% CI 49, 56%) but with a slightly worse calibration slope of 0.91.

Conclusion

An integrated risk score, that combines traditional risk factors with a PRS, improves CAD prediction in patients with RA. Further research is required to better understand the role of heritable components contributing to CAD risk in RA patients. By refining the underlying PRS, we hope to further improve CAD risk prediction in RA patients, through this integrated approach.

Disclosure

M. Soomro: None. M. Stadler: None. S. Viatte: None. A. MacGregor: None. S. Verstappen: None. A. Barton: None. J. Bowes: None.

Application of information theoretic feature selection and machine learning methods for the development of genetic risk prediction models

In view of the growth of clinical risk prediction models using genetic data, there is an increasing need for studies that use appropriate methods to select the optimum number of features from a large number of genetic variants with a high degree of redundancy between features due to linkage disequilibrium (LD). Filter feature selection methods based on information theoretic criteria, are well suited to this challenge and will identify a subset of the original variables that should result in more accurate prediction. However, data collected from cohort studies are often high-dimensional genetic data with potential confounders presenting challenges to feature selection and risk prediction machine learning models. Patients with psoriasis are at high risk of developing a chronic arthritis known as psoriatic arthritis (PsA). The prevalence of PsA in this patient group can be up to 30% and the identification of high risk patients represents an important clinical research which would allow early intervention and a reduction of disability. This also provides us with an ideal scenario for the development of clinical risk prediction models and an opportunity to explore the application of information theoretic criteria methods. In this study, we developed the feature selection and psoriatic arthritis (PsA) risk prediction models that were applied to a cross-sectional genetic dataset of 1462 PsA cases and 1132 cutaneous-only psoriasis (PsC) cases using 2-digit HLA alleles imputed using the SNP2HLA algorithm. We also developed stratification method to mitigate the impact of potential confounder features and illustrate that confounding features impact the feature selection. The mitigated dataset was used in training of seven supervised algorithms. 80% of data was randomly used for training of seven supervised machine learning methods using stratified nested cross validation and 20% was selected randomly as a holdout set for internal validation. The risk prediction models were then further validated in UK Biobank dataset containing data on 1187 participants and a set of features overlapping with the training dataset.Performance of these methods has been evaluated using the area under the curve (AUC), accuracy, precision, recall, F1 score and decision curve analysis(net benefit). The best model is selected based on three criteria: the ‘lowest number of feature subset’ with the ‘maximal average AUC over the nested cross validation’ and good generalisability to the UK Biobank dataset. In the original dataset, with over 100 different bootstraps and seven feature selection (FS) methods, HLA_C_*06 was selected as the most informative genetic variant. When the dataset is mitigated the single most important genetic features based on rank was identified as HLA_B_*27 by the seven different feature selection methods, consistent with previous analyses of this data using regression based methods. However, the predictive accuracy of these single features in post mitigation was found to be moderate (AUC= 0.54 (internal cross validation), AUC=0.53 (internal hold out set), AUC=0.55(external data set)). Sequentially adding additional HLA features based on rank improved the performance of the Random Forest classification model where 20 2-digit features selected by Interaction Capping (ICAP) demonstrated (AUC= 0.61 (internal cross validation), AUC=0.57 (internal hold out set), AUC=0.58 (external dataset)). The stratification method for mitigation of confounding features and filter information theoretic feature selection can be applied to a high dimensional dataset with the potential confounders.

Correlation of the density of yeast Malassezia with the clinical severity of seborrhoeic dermatitis

OBJECTIVE

To correlate the density of the yeast Malassezia with the clinical seventy of seborrhoeic dermatitis.

METHOD

Fifty patients and twenty control subjects were selected for the study. The patients were evaluated both clinically for the severity of seborrhoeic dermatitis and microscopically for the presence of the yeast Malassezia.

RESULTS

Out of 50 patients Malassezia was present in 41 patients (82%). On microscopic evaluation it was found that patients with mild seborrhoeic dermatiis had a density of 2+ (more than 5 but less than 10 yeast cells per high power field (hpf). Patients with moderate seborrhoeic dermatitis had a density of 3+ (more than 10 but less than 20 yeast cells per hpf) and patients with severe seborrhoeic dermatitis had a density of 4+ (more than 20 yeast cells per hpf). Of the 20 normal subjects only 8 (40%) had Malassezia and they had a density of 1+ (5 or fewer yeast cells per hpf). The results show a strong correlation of the yeast Malassezia to the severity of seborrhoeic dermatitis (p value < 0.05).

CONCLUSION

Malassezia increases in proportion with the severity of seborrhoeic dermatitis; an antifungal agent should therefore be used in the treatment of seborrhoeic dermatitis.