Reflections on estimands for patient-reported outcomes in cancer clinical trials.

It is common and important to include the patient's perspective of the impact of treatment on health-related quality of life (HRQoL) outcomes. In this commentary, we focus on applying the new addendum to ICH E9 guideline E9 (R1) relating to the estimand framework to Patient Reported Outcomes (PROs) collected in cancer clinical trials, from a statistician's viewpoint. Currently, common practice for statistical analysis of PRO endpoints of published cancer clinical trials demonstrates ambiguity, leaving critical questions unspecified, hindering conclusions about the effect of treatment on PRO endpoints as well as comparability between clinical trials. To avoid this scenario, we advocate the systematic use of the estimand framework which requires the prospective definition of clear PRO research questions. Among the five attributes of the estimands framework, the definition of the endpoint (what is the right PRO measure and timeframe to target and why?), the intercurrent event identification and management (what happens with PRO data post-disease progression, what is the impact of death?) and the population-level summary (what is an acceptable statistical summary for PRO data?) require the most attention for PRO estimands. We identify good practice and highlight discussion points including the challenges of statistical analysis in the presence of missing and/or unobservable data and in relation to death. Through this discussion we highlight that there is no "statistical magic", but that the estimand framework will help you find out what you really want to know when quantifying the benefit of treatments from the patients' perspective.

What matters most? Different stakeholder perspectives on estimands for an invented case study in COPD.

In drug development, we ask ourselves which population, endpoint and treatment comparison should be investigated. In this context, we also debate what matters most to the different stakeholders that are involved in clinical drug development, for example, patients, physicians, regulators and payers. With the publication of draft ICH E9 addendum on estimands in 2017, we now have a common framework and language to discuss such questions in an informed and transparent way. This has led to the estimand discussion being a key element in study development, including design, analysis and interpretation of a treatment effect. At an invited session at the 2018 PSI annual conference, PSI hosted a role-play debate where the aim of the session was to mimic a regulatory and payer scientific advice discussion for a COPD drug. Including role-play views from an industry sponsor, a patient, a regulator and a payer. This paper presents the invented COPD case-study design and considerations relating to appropriate estimands are discussed by each of the stakeholders from their differing viewpoints with the additional inclusion of a technical (academic) perspective. The rationale for each perspective on approaches for handling intercurrent events is presented, with a key emphasis on the application of while-on-treatment and treatment policy estimands in this context. It is increasingly recognised that the treatment effect estimated by the treatment policy approach may not always be of primary clinical interest and may not appropriately communicate to patients the efficacy they can expect if they take the treatment as directed.

Modelling Health State Utilities as a Transformation of Time to Death in Patients with Non-Small Cell Lung Cancer.

BACKGROUND: When utilities are analyzed by time to death (TTD), this has historically been implemented by 'grouping' observations as discrete time periods to create health state utilities. We extended the approach to use continuous functions, avoiding assumptions around groupings. The resulting models were used to test the concept with data from different regions and different country tariffs. METHODS: Five-year follow-up in advanced non-small cell lung cancer (NSCLC) was used to fit six continuous TTD models using generalized estimating equations, which were compared with progression-based utilities and previously published TTD groupings. Sensitivity analyses were performed using only patients with a confirmed death, the last year of life only, and artificially censoring data at 24 months. The statistically best-fitting model was then applied to data subsets by region and different EQ-5D-3L country tariffs. RESULTS: Continuous (natural) [Formula: see text] and [Formula: see text] models outperformed other continuous models, grouped TTD, and progression-based models in statistical fit (mean absolute error and Quasi Information Criterion). This held through sensitivity and scenario analyses. The pattern of reduced utility as a patient approaches death was consistent across regions and EQ-5D tariffs using the preferred [Formula: see text] model. CONCLUSIONS: The use of continuous models provides a statistically better fit than TTD groupings, without the need for strong assumptions about the health states experienced by patients. Where a TTD approach is merited for use in modelling, continuous functions should be considered, with the scope for further improvements in statistical fit by both widening the number of candidate models tested and the therapeutic areas investigated.

Health-related quality of life of patients with resistant/intolerant chronic phase chronic myeloid leukemia treated with asciminib or bosutinib in the phase 3 ASCEMBL trial.

In ASCEMBL, an open-label, randomized Phase 3 study, asciminib demonstrated superior efficacy and better safety profile compared with bosutinib in patients with chronic myeloid leukemia in chronic phase (CML-CP) previously treated with ≥2 tyrosine kinase inhibitors. Health-related quality of life (HRQOL) reported by patients is key to understanding the benefit and impact of treatment on patients' lives, and is becoming increasingly important as the life expectancy of CML-CP patients increases and patients require long-term treatment. In ASCEMBL, patients completed questionnaires to assess CML symptoms and interference with daily life (M.D. Anderson Symptom Inventory - CML [MDASI-CML]), general HRQOL (five-level EQ-5D [EQ-5D-5L], Patient Global Impression of Change - CML [PGIC-CML]), and impact of CML on working life and activity (Work Productivity and Activity Impairment questionnaire - CML [WPAI-CML]). Patients' CML symptoms and HRQOL remained stable during 48 weeks of treatment with asciminib, with a general trend for decreased CML symptom severity, particularly for fatigue, and improvement in HRQOL. A clinically meaningful increase in diarrhea severity was observed in patients treated with bosutinib compared to asciminib. These data provide better understanding of the patient perspective and treatment impact on HRQOL in a later-line setting, where little information has been published to date.

First-line nivolumab plus ipilimumab with two cycles of chemotherapy versus chemotherapy alone (four cycles) in metastatic non-small cell lung cancer: CheckMate 9LA 2-year patient-reported outcomes.

BACKGROUND: In CheckMate 9LA (NCT03215706), first-line nivolumab plus ipilimumab with chemotherapy (2 cycles) significantly improved overall survival versus chemotherapy (4 cycles) in patients with metastatic non-small cell lung cancer and no known sensitising epidermal growth factor receptor/anaplastic lymphoma kinase alterations. We present exploratory patient-reported outcomes (PROs; minimum follow-up, 2 years). METHODS: In patients (N = 719) randomised 1:1 to nivolumab plus ipilimumab with chemotherapy or chemotherapy alone, disease-related symptom burden and health-related quality of life were assessed using the Lung Cancer Symptom Scale (LCSS) and 3-level EQ-5D (EQ-5D-3L). Treatment-phase changes in LCSS average symptom burden index (ASBI), LCSS three-item global index (3-IGI) and EQ-5D-3L visual analogue scale (VAS) and utility index (UI) over time were analysed descriptively and using mixed-effect model repeated measures. Time-to-deterioration/improvement analyses were conducted. RESULTS: Treatment-phase PRO questionnaire completion rates were >80%. Mean treatment-phase changes showed no deterioration from baseline in both arms for LCSS ASBI/3-IGI and EQ-5D-3L VAS/UI; however, minimally important differences were not met. Mixed-effect model repeated measures analyses showed overall reduction in symptom burden from baseline for both arms; changes from baseline for LCSS 3-IGI and EQ-5D-3L VAS/UI were numerically improved with nivolumab plus ipilimumab with chemotherapy versus chemotherapy, but minimally important differences were not met. Nivolumab plus ipilimumab with chemotherapy delayed time-to-definitive-deterioration versus chemotherapy (LCSS ASBI: hazard ratio, 0.62 [95% confidence interval, 0.45-0.87]); results were similar across PRO measures. CONCLUSIONS: At 2-year minimum follow-up, first-line nivolumab plus ipilimumab with chemotherapy reduced the risk of definitive deterioration in disease-related symptom burden and health-related quality of life versus chemotherapy and maintained QoL in patients with metastatic non-small cell lung cancer. CLINICAL TRIAL REGISTRATION: ClinicalTrials.gov Identifier, NCT03215706.

Effect of ß2-adrenergic receptor gene (ADRB2) 3' untranslated region polymorphisms on inhaled corticosteroid/long-acting ß2-adrenergic agonist response.

BACKGROUND: Evidence suggests that variation in the length of the poly-C repeat in the 3' untranslated region (3'UTR) of the ß2-adrenergic receptor gene (ADRB2) may contribute to interindividual variation in ß-agonist response. However, methodology in previous studies limited the assessment of the effect of sequence variation in the context of poly-C repeat length. The objectives of this study were to design a novel genotyping method to fully characterize sequence variation in the ADRB2 3'UTR poly-C repeat in asthma patients treated with inhaled corticosteroid and long-acting ß2-adrenergic agonist (ICS/LABA) combination therapy, and to analyze the effect of the poly-C repeat polymorphism on clinical response. METHODS: In 2,250 asthma patients randomized to treatment with budesonide/formoterol or fluticasone/salmeterol in a six-month study (AstraZeneca study code: SD-039-0735), sequence diversity in the ADRB2 poly-C repeat region was determined using a novel sequencing-based genotyping method. The relationship between the poly-C repeat polymorphism and the incidence of severe asthma exacerbations, and changes in pulmonary function and asthma symptoms from baseline to the average during the treatment period, were analyzed. RESULTS: Poly-C repeat genotypes were assigned in 97% (2,192/2,250) of patients. Of the 13 different poly-C repeat alleles identified, six alleles occurred at a frequency of >5% in one or more population in this study. The repeat length of these six common alleles ranged from 10 to 14 nucleotides. Twelve poly-C repeat genotypes were observed at a frequency of >1%. No evidence of an association between poly-C repeat genotype and the incidence of severe asthma exacerbations was observed. Patients' pulmonary function measurements improved and asthma symptoms declined when treated with ICS/LABA combination therapy regardless of poly-C repeat genotype. CONCLUSIONS: The extensive sequence diversity present in the poly-C repeat region of the ADRB2 3'UTR did not predict therapeutic response to ICS/LABA therapy.

First-line nivolumab plus ipilimumab versus chemotherapy for the treatment of unresectable malignant pleural mesothelioma: patient-reported outcomes in CheckMate 743.

OBJECTIVE: In CheckMate 743 (NCT02899299), nivolumab + ipilimumab significantly prolonged overall survival in patients with unresectable malignant pleural mesothelioma (MPM). We present patient-reported outcomes (PROs). MATERIALS AND METHODS: Patients (N = 605) were randomized to nivolumab + ipilimumab or chemotherapy. Changes in disease-related symptom burden and health-related quality of life (HRQoL) were evaluated descriptively using the Lung Cancer Symptom Scale (LCSS)-Mesothelioma (Meso) average symptom burden index (ASBI), LCSS-Meso 3-item global index (3-IGI), 3-level EuroQol 5-dimensional (EQ-5D-3L) visual analog score (VAS), and EQ-5D-3L utility index. PROs were assessed at baseline and every 2 (nivolumab + ipilimumab) or 3 weeks (chemotherapy) through 12 weeks, every 6 weeks through 12 months, every 12 weeks thereafter, and at specified follow-ups. Mixed-effect model repeated measures (MMRM) and time to deterioration analyses were conducted. RESULTS: Completion rates were generally >80%. LCSS-Meso ASBI mean changes from baseline trended to improve over time with nivolumab + ipilimumab and deteriorate with chemotherapy, but did not meet clinically important difference thresholds [±10 score change]. EQ-5D-3L VAS mean scores improved over time with nivolumab + ipilimumab; by week 60, patients had scores consistent with United Kingdom normal population values. MMRM analyses favored nivolumab + ipilimumab for all individual symptoms except cough. Nivolumab + ipilimumab delayed time to definitive deterioration in HRQoL (hazard ratio 0.52 [95% confidence interval 0.36-0.74]) and showed a trend in symptom delay versus chemotherapy. CONCLUSIONS: Nivolumab + ipilimumab decreased the risk of deterioration in disease-related symptoms and HRQoL versus chemotherapy and maintained QoL in patients with unresectable MPM.

Three-year follow-up and patient-reported outcomes from CheckMate 078: Nivolumab versus docetaxel in a predominantly Chinese patient population with previously treated advanced non-small cell lung cancer.

OBJECTIVES: In the phase 3 CheckMate 078 study, nivolumab prolonged overall survival (OS) and showed a favorable safety profile versus docetaxel in a predominantly Chinese patient population with previously treated advanced non-small cell lung cancer (aNSCLC). However, long-term efficacy, safety, and health-related quality of life findings with second-line nivolumab are very limited in Asian patients with previously treated aNSCLC. Here, we report updated clinical data and patient-reported outcomes (PROs) from the phase 3 CheckMate 078 trial with a 3-year minimum follow-up. MATERIALS AND METHODS: Patients with aNSCLC and disease progression after platinum-doublet chemotherapy were randomized 2:1 to nivolumab (3 mg/kg every 2 weeks) or docetaxel (75 mg/m2 every 3 weeks) until progression or unacceptable toxicity. The primary endpoint was OS; secondary endpoints included objective response rate, progression-free survival, safety, and disease-related symptom deterioration assessed using the Lung Cancer Symptom Scale (LCSS) by Week 12. Additional PRO assessments were exploratory endpoints. RESULTS: At ≥ 37.3 months follow-up, 3-year OS rates were 19% with nivolumab and 12% with docetaxel; 30% and 0% of responders remained in response for ≥ 3 years, respectively. Incidence of treatment-related adverse events occurring after 2 years was lower than during the first 2 years. No new treatment-related deaths were reported. By Week 12 of treatment, rates of disease-related symptom deterioration were 32% with nivolumab and 47% with docetaxel. Completion rates for PRO questionnaires were ≥ 80% in both arms. Clinically meaningful and sustained improvements in LCSS Average Symptom Burden Index scores and delayed time to first symptom deterioration were observed with nivolumab against docetaxel. CONCLUSIONS: At 3 years, nivolumab continued to demonstrate survival benefit versus docetaxel, exhibiting improvements in disease-related symptoms and overall health status in a predominantly Chinese patient population with previously treated aNSCLC. No new safety signals were observed. These findings are similar to the global population.

Improving patient-clinician communication following nephrectomy in renal cell carcinoma: Development, content validation and pilot testing of a conversation aid tool.

OBJECTIVES: This study developed, and established the content validity, of a conversation aid tool (CAT) for use in clinical practice with renal cell carcinoma (RCC) patients who receive a curative nephrectomy and are at high-risk of recurrence. The CAT was pilot tested in a sample of RCC patients to establish whether the CAT increases knowledge of RCC, treatment options (such as adjuvant therapy), and care options. METHODS: A cross-sectional, mixed methods design was used involving initial, exploratory interviews with RCC patients, RCC specialists and a steering group. Further content validation interviews were conducted with RCC patients and specialists. A web-based survey was conducted with RCC patients (N = 60), to compare the CAT versus a standard of care (SOC) consultation comparator tool on patient knowledge. RESULTS: Findings from exploratory interviews were used to develop the CAT. Content validation interviews demonstrated that the CAT was well understood and relevant to RCC patients. The web-based survey demonstrated that viewing the CAT significantly improved participants knowledge of RCC, and care options, when compared to the SOC. CONCLUSION: The findings highlight that the CAT is a relevant, comprehensive and well-understood tool for use in the post-nephrectomy consultation. PRACTICE IMPLICATIONS: Use of the CAT may increase patient knowledge of RCC and care options.

What is an estimand & how does it relate to quantifying the effect of treatment on patient-reported quality of life outcomes in clinical trials?

BACKGROUND: Published in 2019, a new addendum to the ICH E9 guideline presents the estimand framework as a systematic approach to ensure alignment among clinical trial objectives, trial execution/conduct, statistical analyses, and interpretation of results. The use of the estimand framework for describing clinical trial objectives has yet to be extensively considered in the context of patient-reported outcomes (PROs). We discuss the application of the estimand framework to PRO objectives when designing clinical trials in the future, with a focus on PRO outcomes in oncology trial settings as our example. MAIN: We describe the components of an estimand and take a naïve PRO trial objective to illustrate how to apply attributes described in the estimand framework to inform construction of a detailed clinical trial objective and its related estimand. We discuss identifying potential post-randomization events that alter the interpretation of the endpoint or render its observation impossible (also defined as intercurrent events) in the context of PRO endpoints, and the implications of how to handle intercurrent events in the construction of the PRO objective. Using a simple objective statement, "What is the effect of treatment X on patient's quality of life?", we build up an example estimand statement and also use a previously published phase III oncology clinical trial to illustrate how an estimand for a PRO objective could have been written to align to the estimate framework. CONCLUSION: The use of the estimand framework, as described in the new ICH E9 (R1) addendum guideline will become a key common framework for developing clinical trial objectives for evaluating effects of treatment. In the context of considering PROs, the framework provides an opportunity to more precisely specify and build the rationale for patient-focused objectives. This will help to ensure that clinical trials used for registration are designed and analysed appropriately, enabling all stakeholders to accurately interpret conclusions about the treatment effects for patient-focused outcomes.

Ticagrelor yields consistent dose-dependent inhibition of ADP-induced platelet aggregation in patients with atherosclerotic disease regardless of genotypic variations in P2RY12, P2RY1, and ITGB3.

The platelet P2Y(12) receptor is the target of clopidogrel therapy, which has been shown to reduce thromboembolic complications of atherosclerotic disease but has limitations in terms of variability of response and irreversibility of effect. This receptor is also a target for ticagrelor (AZD6140), the first reversibly binding oral P2Y(12) receptor antagonist that does not require metabolic activation and yields more consistent inhibition of platelet aggregation than clopidogrel therapy. Single nucleotide polymorphisms (SNPs) have been described in the gene for this receptor (P2RY12), some of which have been associated with variability in platelet reactivity. SNPs in P2RY1 and ITGB3 have also been reported by some groups to affect platelet reactivity to adenosine diphosphate (ADP). We assessed whether SNPs in these genes influenced the pharmacodynamic response to ticagrelor in patients enrolled in both the DISPERSE study (stable atherosclerotic disease) and the DISPERSE2 study (non-ST-segment elevation acute coronary syndromes). Platelet aggregation data (at baseline and 4 weeks) and DNA samples from clopidogrel-naive Caucasian patients treated with ticagrelor were available for 151 patients. Seventy four SNPs within three genes were genotyped. After adjustment for multiple comparisons, none of these SNPs were found to significantly influence inhibition of ADP-induced platelet aggregation by ticagrelor.

Nivolumab plus ipilimumab versus chemotherapy as first-line treatment in advanced non-small-cell lung cancer with high tumour mutational burden: patient-reported outcomes results from the randomised, open-label, phase III CheckMate 227 trial.

BACKGROUND: In the phase III CheckMate 227 study, first-line nivolumab + ipilimumab significantly prolonged progression-free survival (co-primary end-point) versus chemotherapy in patients with advanced non-small-cell lung cancer (NSCLC) and high tumour mutational burden (TMB; ≥10 mutations/megabase). AIM: To evaluate patient-reported outcomes (PROs) in this population. METHODS: Disease-related symptoms and general health status were assessed using the validated PRO questionnaires Lung Cancer Symptom Scale (LCSS) and EQ-5D, respectively. LCSS average symptom burden index (ASBI) and three-item global index (3-IGI) and EQ-5D visual analogue scale (VAS) and utility index (UI) scores and changes from baseline were analysed descriptively. Longitudinal changes were assessed by mixed-effect model repeated measures (MMRMs) and time to first deterioration/improvement analyses. RESULTS: In the high TMB population, PRO questionnaire completion rates were ∼90% at baseline and >80% for most on-treatment assessments. During treatment, mean changes from baseline with nivolumab + ipilimumab showed early, clinically meaningful improvements in LCSS ASBI/3-IGI and EQ-5D VAS/UI; with chemotherapy, symptoms and health-related quality of life remained stable (LCSS ASBI/3-IGI, EQ-5D UI) or improved following induction (EQ-5D VAS). MMRM-assessed changes in symptom burden were improved with nivolumab + ipilimumab versus chemotherapy. Symptom deterioration by week 12 was lower with nivolumab + ipilimumab versus chemotherapy (22.3% versus 35.0%; absolute risk reduction: 12.7% [95% confidence interval 2.4-22.5]), irrespective of discontinuation. Time to first deterioration was delayed with nivolumab + ipilimumab versus chemotherapy across LCSS and EQ-5D summary measures. CONCLUSION: First-line nivolumab + ipilimumab demonstrated early, sustained improvements in PROs versus chemotherapy in patients with advanced NSCLC and high TMB. CLINICAL TRIAL REGISTRATION: NCT02477826.

Effect of ADRB2 polymorphisms on response to longacting beta2-agonist therapy: a pharmacogenetic analysis of two randomised studies.

BACKGROUND: New evidence has suggested that people with asthma who are homozygous for arginine at aminoacid 16 of the beta2-adrenergic receptor (ADRB2) might not benefit from longacting beta2-agonist therapy. We, therefore, investigated whether ADRB2 polymorphisms affect response to longacting beta2-agonists in combination with inhaled corticosteroids. METHODS: Asthmatics were stratified by ADRB2 genotype in two studies to assess the effects of inhaled corticosteroids plus longacting beta2-agonists on asthma exacerbations. In study 1 (double-blind), 2250 asthmatics were randomly assigned to budesonide plus formoterol maintenance and reliever therapy, fixed-dose budesonide plus formoterol, or fixed-dose fluticasone plus salmeterol for 6 months. Study 2 (open-label) consisted of 405 asthmatics and compared an adjustable regimen of budesonide plus formoterol with fixed-dose budesonide plus formoterol and fixed-dose fluticasone plus salmeterol for 7 months. The relation between ADRB2 polymorphism, severe asthma exacerbations, and other asthma outcomes was analysed. Primary endpoints for studies 1 and 2 were severe asthma exacerbation and asthma control as assessed by measures of exacerbations, respectively. FINDINGS: In study 1, Gly16Arg genotype had no effect on the percentage of participants with severe exacerbations across all treatment groups (99 [12%] of 833 Gly/Gly, 110 [11%] of 1028 Gly/Arg, and 32 [9%] of 361 Arg/Arg participants). Secondary endpoints, including forced expiratory volume in 1 s, peak expiratory flow, use of as-needed medication, and number of nights with awakenings were similar between genotype groups. No relation was recorded between ADRB2 haplotype and primary and secondary endpoints. In study 2, the frequency of asthma exacerbations (15 [9%] of 168 Gly/Gly, 13 [8%] of 169 Gly/Arg, and 6 [9%] of 67 Arg/Arg participants) and other study endpoints were closely similar for all ADRB2 genotypes. INTERPRETATION: Since we showed no pharmacogenetic effect of ADRB2 variation on therapeutic response in asthma, patients, irrespective of their genotype, can continue to receive inhaled corticosteroids plus longacting beta2-agonists.

Association Between Plasma Genotyping and Outcomes of Treatment With Osimertinib (AZD9291) in Advanced Non-Small-Cell Lung Cancer.

PURPOSE: Third-generation epidermal growth factor receptor (EGFR) tyrosine kinase inhibitors (TKIs) have demonstrated potent activity against TKI resistance mediated by EGFR T790M. We studied whether noninvasive genotyping of cell-free plasma DNA (cfDNA) is a useful biomarker for prediction of outcome from a third-generation EGFR-TKI, osimertinib. METHODS: Plasma was collected from all patients in the first-in-man study of osimertinib. Patients who were included had acquired EGFR-TKI resistance and evidence of a common EGFR-sensitizing mutation. Genotyping of cell-free plasma DNA was performed by using BEAMing. Plasma genotyping accuracy was assessed by using tumor genotyping from a central laboratory as reference. Objective response rate (ORR) and progression-free survival (PFS) were analyzed in all T790M-positive or T790M-negative patients. RESULTS: Sensitivity of plasma genotyping for detection of T790M was 70%. Of 58 patients with T790M-negative tumors, T790M was detected in plasma of 18 (31%). ORR and median PFS were similar in patients with T790M-positive plasma (ORR, 63%; PFS, 9.7 months) or T790M-positive tumor (ORR, 62%; PFS, 9.7 months) results. Although patients with T790M-negative plasma had overall favorable outcomes (ORR, 46%; median PFS, 8.2 months), tumor genotyping distinguished a subset of patients positive for T790M who had better outcomes (ORR, 69%; PFS, 16.5 months) as well as a subset of patients negative for T790M with poor outcomes (ORR, 25%; PFS, 2.8 months). CONCLUSION: In this retrospective analysis, patients positive for T790M in plasma have outcomes with osimertinib that are equivalent to patients positive by a tissue-based assay. This study suggests that, upon availability of validated plasma T790M assays, some patients could avoid a tumor biopsy for T790M genotyping. As a result of the 30% false-negative rate of plasma genotyping, those with T790M-negative plasma results still need a tumor biopsy to determine presence or absence of T790M.

Classification of rheumatoid arthritis status with candidate gene and genome-wide single-nucleotide polymorphisms using random forests.

Using the North American Rheumatoid Arthritis Consortium (NARAC) candidate gene and genome-wide single-nucleotide polymorphism (SNP) data sets, we applied regression methods and tree-based random forests to identify genetic associations with rheumatoid arthritis (RA) and to predict RA disease status. Several genes were consistently identified as weakly associated with RA without a significant interaction or combinatorial effect with other candidate genes. Using random forests, the tested candidate gene SNPs were not sufficient to predict RA patients and normal subjects with high accuracy. However, using the top 500 SNPs, ranked by the importance score, from the genome-wide linkage panel of 5742 SNPs, we were able to accurately predict RA patients and normal subjects with sensitivity of approximately 90% and specificity of approximately 80%, which was confirmed by five-fold cross-validation. However, in a complete training-testing framework, replication of genetic predictors was less satisfactory; thus, further evaluation of existing methodology and development of new methods are warranted.