Statistics from Altmetric.com
I read with interest the recent article by Crilly and colleagues1 on the potential use of interleukin 1 (IL1) gene polymorphisms in predicting the need for joint surgery in patients with rheumatoid arthritis. Although the authors' findings are certainly interesting and worthy of further investigation, I would like to raise a couple of points as to how these were arrived at in the first place.
When adopting the “candidate gene” approach in any disease association study, it pays to consider carefully the physiological context of a hypothesis. In this case the authors worked on the premise that one of the IL1 gene family polymorphisms investigated might correlate with disease outcome. These included polymorphisms both in the IL1 genes and the IL1 receptor antagonist (IL1ra) locus. The IL1ra is unique in being the only known endogenous cytokine receptor antagonist. The genes encoding IL1α, β and the IL1ra have co-evolved as a cytokine control mechanism2: agents eliciting IL1 production generally induce IL1ra gene transcription as well, either directly or through a feedback loop triggered by IL1 itself. IL1ra competes with IL1 for binding to the cognate receptor, but produces no known signalling events upon binding. Additionally, one of the two IL1 receptors, type II, also lacks any known signalling capacity and is thought to function as a decoy binding site. If it is assumed that polymorphisms found in various members of the IL1 gene family modify gene function then the implication is that they should be studied collectively, not individually. In support of this argument comes recent experimental evidence showing that plasma IL1ra levels are determined by allelic combinations within the IL1 gene family,3 as well as evidence for linkage disequilibrium between polymorphic markers spread across the IL1 gene family locus.4 It is fair to say that Crilly and colleagues did allude in their paper to the possible existence of an extended IL1 gene family haplotype. However, there is already sufficient published evidence3-6 to justify examination of haplotypic combinations of the investigated alleles in population studies.
On a different note, although the authors' argument about the validity of surgery as a study end point is convincing, their selection of a 15 year disease duration free from surgery as the cut off time point for this study seems dubious. A previous study by the same group had calculated that the median disease duration before surgery was 14.6 years, implying that there are a substantial number of patients in the “no surgery” group who will require surgery shortly after the 15 year time point, thereby potentially confounding any significant statistical findings. One might argue that the aim of this study was not to differentiate between patients that will or will not require surgery but, rather, to differentiate between rapid (<15 years) versus delayed (>15 years to surgery) disease progression. Even so, a more valid statistical approach might have been to analyse all patients as a single group and try to correlate genotype (or haplotype) with disease duration to surgery.
I thank Dr Vamvakopoulos for his interest in our article1-1 and would like to make the following response.
We agree that testing for haplotype combinations of the interleukin 1 (IL1) gene family allele would be useful. However, for a large number of alleles to be tested a considerable sample size would be needed to correct for multiple allele testing. This would require a multicentre study; this is limited by funding constraints.
Secondly, for the statistical analysis we predefined early and late surgery. Analysis was undertaken after discussion with a qualified medical statistician. Should the correspondent wish to analyse our data, we would be happy to allow access if guarantees can be provided.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.