Dr. Cooper questions the statistical validity of our conclusions regarding the effect of folic acid supplementation on the efficacy of methotrexate. Our sample size and power calculations were based on the toxicity data, which were the primary outcome of the study. We could not show any difference in the efficacy of methotrexate among the three treatment groups. Although Dr. Cooper is correct that we might not have had adequate power to show the existence of such differences, no trend suggested a dose response in patients receiving the higher dose of folic acid and experiencing less efficacy from methotrexate compared with those receiving the lower dose or those receiving no supplementation. The data presented by Dr. Cooper show random variation in efficacy measures, a factor in full agreement with the clinical experience that some patients respond to methotrexate and others do not. Given these random figures for the three treatment groups, increasing the sample size would probably not have altered the efficacy data in a clinically (not only a statistically) meaningful way.

That only some patients develop methotrexate toxicity has been documented in other studies. We have previously shown that folate status declines during low-dose methotrexate administration [1] and that in patients receiving placebo, future toxicity correlates with initial folate status [2]. Therefore, it may not be sufficient to reserve folate supplementation for patients with initially low levels, given that folate status deteriorates during methotrexate administration. Because folic acid is inexpensive, prescribing at least 400 to 1000 µg of folic acid daily (and more if suspected methotrexate toxicity occurs) is prudent.

In response to the Drs. Kavanaugh, it is a good idea to evaluate RDW as well as the mean corpuscular volume as early markers of increased risk for methotrexate toxicity. Our report clearly warns that, because of the possibility of a dimorphic anemia, the mean corpuscular volume cannot replace blood folate levels as an indicator of risk.

Regarding Dr. Shiroky's questions, it is true that the initial reports (abstracts) of the trial indicated that the high-dose folic acid level was 50 mg/wk. The folic acid content of the capsules was analyzed and found to contain 1.0 ± 0.15 and 5.5 ± 0.3 mg of folic acid per capsule, even though the manufacturer was requested to supply 1.0 and 10 mg of folic acid. We did not analyze the supplements until the end of the trial because we were blinded to the randomization protocol. Therefore, we believe that we honestly reported the doses used in the final publication, and we did not rely on the manufacturer's values. No problems were associated with the original dose, and the dosage was not changed.

We reported that in many patients, joint swelling and tenderness did not markedly or moderately improve. We defined moderate improvement as a decrease in these variables of 30% to 49% (the American College of Rheumatology core set of criteria for rheumatoid arthritis trials requires only 20% improvement) [3]. Failure to achieve moderate improvement, as defined, is not the same as having no clinical improvement. If the rheumatologists "were hesitant" about increasing the methotrexate dose, they were equally hesitant with the patients from all three study arms. The mean methotrexate dose in our patients has been reported previously and is similar to the dose used in the trial [4]. Our rheumatologists and research assistants were blinded to the study group to which the patients were randomly assigned and thus used clinical judgment to manage their patients. Toxicities in both the low- and high-dose folic acid supplement groups were mild, of minor clinical importance, and of short duration. The observation of a difference in frequency of toxicity in these two groups is not clinically important.

Average methotrexate doses of 9 mg/wk and 12 mg/wk are not significantly different. We made no attempt to interfere with the rheumatologist's selection of the appropriate methotrexate dose. The designs of Shiroky and colleagues' study [5] and our study do not differ greatly. In our previous study [2], uninterrupted, daily folic acid (1 mg) did not alter the efficacy of methotrexate. Thus, the timing of the folic acid dose appears to be unimportant to the efficacy of the methotrexate. We pointed out the need for more specific investigations regarding the timing of the folate relative to the antifolate dose.

Finally, we agree with Dr. Shiroky that a trial comparing folic acid and leucovorin would be useful, and we would be pleased to collaborate in such a trial. We have tried unsuccessfully to tap various sources to fund such a study. Such a trial, if correctly conducted, should definitively determine which is the best supplement to use in patients with rheumatoid arthritis treated with methotrexate.