Rheumatology 1999; 38: 789-790
© 1999 British Society for Rheumatology
Letters to the Editor |
Combination therapy in rheumatoid arthritis
Department of Clinical Epidemiology, VU University Hospital, Room VE 9-78, PO Box 7057, 1007 MB Amsterdam, The Netherlands
Correspondence to:
M. Boers.
SIR, We thank Drs O'Dell, Moore and Klassen for their reaction to our systematic review on combined treatment with second-line anti-rheumatic drugs in rheumatoid arthritis [1]. This allows us to discuss their study [2] in more detail than was possible in our original review [3]. Any review contains arbitrary elements. Therefore, it is essential that all choices in the process of data aggregation be explicit. In this way, the final evaluation is up to the readers. With Dr O'Dell c.s. we recognize the importance of long-term follow-up. Considering the small number of studies with extended follow-up, we chose to compare data on outcomes from the selected high-quality studies after the most extended period of treatment, with (1) patients in the intervention group(s) still on their intended regimen and (2) before any selection of patients with regard to therapy efficacy or other criteria had taken place (intention-to-treat principle).
We agonized over the best interpretation of the results of O'Dell et al.'s study, but finally decided that we did not agree with their own assessment. The key difference in interpretation lies at the 9 month point. O'Dell et al. made a legitimate but arbitrary choice to regard any patient not meeting the 50% Paulus criterion for improvement at 9 months as a therapy failure. Unfortunately, these patients were not followed up further and no data are available on them. Notably, until 9 months, a suboptimal response could be handled by increases in methotrexate dose, so that most drop-outs during this period would be for other causes, e.g. toxicity. In contrast, the majority of drop-outs at or after 9 months would be for lack of efficacy.
In essence, the procedure results in two studies, reported as one: (1) a 9 month randomized controlled trial with the 50% criterion as endpoint (overall results according to intention-to-treat, P=0.12, reported in our review as trend); (2) a further follow-up study of the responders. This has the potential to magnify the (trend) differences at 9 months, e.g. when the mean level of Paulus response in other groups would be slightly below 50%. In addition, after selection of responders at the 9 month time point, the distribution of prognostic factors in the three groups (other than treatment) cannot be assumed to be comparable in the way they were at baseline, immediately after randomization. In our view, any comparison of the three groups in such a follow-up study should be viewed with caution, and certainly not pooled with the primary results under (1).
To support O'Dell et al.'s conclusion that triple therapy is more effective than the other modalities tested, we would need 2-yr data of all randomized patients.
For instance, we would have liked to see `sensitivity analyses' of the full dataset (i.e. all randomized patients) using different cut-off points for response. This is pertinent in view of the proven greater discriminative capacity of the ACR20 criterion [4]. It would also be nice to know which patients actually started alternative therapy after they were judged to be treatment failures according to the trial definition. Also, survival analyses are more suited for `one-off' events such as recurrence of cancer, and less for disease activity, which is essentially a continuous variable. Thus, an analysis of the area under the curve of response (or another index) of all randomized patients might have added valuable information.
With O'Dell, we do believe that their results indicate that responders on triple therapy are more likely to maintain response on follow-up than patients in the other groups.
In conclusion, the final interpretation depends on whether one accepts the 50% Paulus criterion as sole judge of efficacy and, in addition, whether one is willing to assume that the selection of patients for follow-up after 9 months based on response did not influence the distribution of prognostic factors in the treatment groups (other than treatment) present at baseline.
We thank Dr Hurst for his reaction. Pharmacokinetic explanations of differences in the efficacy of various drug combinations are outside the scope of our review. Nevertheless, we agree with Dr Hurst that investigators should try to understand benefits that are due to altered pharmacokinetics. On the other hand, we do not think that complete understanding is a `conditio sine qua non' for testing combinations, let alone a review on combination therapy. A recent study shows there is no interaction between methotrexate and sulphasalazine with the usual doses for treatment of rheumatoid arthritis [5]. With regard to an interaction between methotrexate and cyclosporin, data presented at the FDA Arthritis Advisory Panel Hearing show that although some metabolites of methotrexate are slightly increased, these are not likely to have any anti-inflammatory activity.
References
- O'Dell JR, Moore GF, Klassen LW. Combination therapy in rheumatoid arthritis: a comment. Rheumatology 1999;38:577.
-
O'Dell J, Haire C, Erikson N et al. Treatment of rheumatoid arthritis with methotrexate alone, sulfasalazine and hydroxychloroquine, or a combination of all three medications. N Engl J Med 1996;334:1287–91.
[Abstract/Free Full Text] -
Verhoeven AC, Boers M, Tugwell P. Combination therapy in rheumatoid arthritis: updated systematic review. Br J Rheumatol 1998;37:612–9.
[Abstract/Free Full Text] - Felson DT, Anderson JJ, Lange ML, Wells G, LaValley MP. Should improvement in rheumatoid arthritis clinical trials be defined as fifty percent or seventy percent improvement in core set measures, rather than twenty percent? Arthritis Rheum 1998;41:1564–70.[Web of Science][Medline]
- Haagsma CJ, Russel FGM, Vree T, van Riel PLCM, van de Putte LBA. Combination of methotrexate and sulphasalazine in patients with rheumatoid arthritis: Pharmacokinetic analysis and relationship to clinical response. Br J Clin Pharmacol 1996;42:195–200.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||