Background The effectiveness of treatment with anti-TNF inhibitors in rheumatoid arthritis (RA) is variable with a significant minority of patients experiencing poor response. Ideally clinicians would like to identify which patients are likely to respond prior to commencement on treatment, however no reliable marker of response exists. Investigators have hypothesised that a proportion of the variance in response is underpinned by genetics; however genetic association studies conducted to date have added little evidence to this proposed explanation. Potential reasons for the lack of genetic evidence include (i) treatment response has no/very little genetic basis (ii) previous studies have lacked power to detect modest genetic effects (iii) the measure of treatment response, i.e. the 28 joint count disease activity score (DAS28), is inappropriate.
Objectives First, to estimate heritability of treatment response using genome-wide genetic variation data in a large cohort of TNF-treated RA patients from the UK and, second, to investigate heritability of the individual components of the DAS28 composite score.
Methods 1,152 RA patients, recruited into the BRAGGSS cohort prior to commencement on anti-TNF therapy were genotyped using affymetrix or illumina genome-wide single nucleotide polymorphism (SNP) arrays. Array datasets were subsequently imputed using IMPUTE2 and, after quality control (QC) procedures, were merged using PLINK. Genome-wide Complex Trait Analysis (GCTA) software was used to estimate heritability. Related individuals (2nd – 3rd cousin or above) were removed from the genetic relationship matrix (GRM). In a second analysis, low frequency SNPs (i.e. MAF<5%) were excluded from the GRM (results reported in parentheses). The DAS28 was calculated just prior to and at six months following treatment. Outcome included change in: DAS28, swollen joint count (SJC), tender joint count (TJC), erythrocyte sedimentation rate (ESR), and patient global wellbeing visual analogue scale (VAS). Statistical models were adjusted for clinical and/or biochemical covariates where appropriate.
Results Following QC, 2,400,691 imputed SNPs were available for analysis in 1,150 patients. A further 56 individuals were removed due to relatedness. The resultant cohort was 78% female, 80% were concurrently treated with disease modifying anti-rheumatic drugs (DMARDs), the mean health assessment questionnaire (HAQ) score was 1.63 and the mean DAS28 at inclusion, and following six months of treatment, was 6.40 and 3.95, respectively. The GRM explained approximately 5% (10%) of the variance in treatment response when the composite DAS28 score was investigated. The variance explained rose to approximately 10% (25%) and 30% (48%) when change in SJC and ESR was investigated, respectively. The explained variance in change in TJC and VAS was 0.1% (<0.1%) and 7% (<0.01%).
Conclusions The amount of variability in treatment response explained by genetic variation was modest. The lack of consistent genetic signal for this phenotype is likely to be further compounded by the use of a composite measure i.e. DAS28. Future, well powered genetic association studies should prioritise objective components of DAS28 i.e. ESR and SJC.
Disclosure of Interest None Declared