Article Text
Statistics from Altmetric.com
Editor,—I have read with interest the article by Cuffari et al(Gut1996;39:401–6) and do not feel that the conclusions of their study are supported by the results. The major aim of their study was to determine whether erythrocyte 6 mercaptopurine metabolite (6-TG) concentrations correlated with disease activity in patients with Crohn’s disease. The authors state that their results show an inverse correlation (r = −0.457, p < 0.05) between 6-mercaptopurine nucleotide concentrations and the Harvey-Bradshaw index of disease activity. They conclude that these results support the immunosuppressive role of 6-TG metabolites and that measurement of concentrations are useful in the treatment of patients with inflammatory bowel disease (IBD). However, using the same statistical test as the authors, I could find no such correlation (0.322, p > 0.05) when I analysed the data presented in figure 4. The original data may not be represented accurately in graphical form and I would welcome the authors’ comments clarifying this. I do not think that the study provides any evidence that measurement of 6-TG metabolites are useful in the management of patients with IBD.
Reply
Editor,—We appreciate Dr Ballinger’s interest in our study, and for her astutely pointing out that figure 4 from our paper inaccurately displayed our original data. Hence any results calculated directly from the graph would be erroneous. A corrected version of the graph is shown. The significant inverse correlation (r = −0.457, p < 0.05) between erythrocyte 6-TG concentrations and the patients’ Harvey-Bradshaw index (HBI) of disease was obtained using a matrix calculation of the Spearman correlation coefficient, with Systat software. The parameters included in our matrix analysis were neutrophil, lymphocyte and total leucocyte counts, haemoglobin, serum albumin, erythrocyte 6-TG and 6-methylmercaptopurine concentrations. The only significant correlation to the HBI was observed for 6-TG concentrations. The level of significance (p < 0.05) was maintained even when the analysis did not compensate for the ties in HBI values.
Furthermore, we have re-calculated the Spearman correlation coefficient between HBI values and erythrocyte 6-TG values alone, using SPSS, Sigma Stat and Systat software programs. Ther value obtained was −0.409, which yields a p value of 0.04 using a two tailed analysis. Ther value obtained was the same when manually calculated for ties using the formula ∑y2 = (N3 − N)/12 − ∑Ty, as described in a reference textbook on non-parametric statistics.1-1
Therefore, despite the limitation of interpreting data from 25 paediatric patients, our initial conclusion that 6-TG metabolite measurements are useful in the management of patients with IBD is firmly held. This conclusion is furthermore sustained by our analysis of almost 200 samples collected prospectively from adult and paediatric patients with IBD on long term 6-mercaptopurine. Using these metabolite studies, we have been able to identify clearly non-complying patients as well as those whose concentrations were in the toxic range, resulting in haematological and biochemical abnormalities.
References
- 1-1.↵