Postoperative Crohn's Disease Maintenance
The Cochrane collaboration review of medical treatment in post-operative Crohn's disease performed in 2009 elegantly demonstrates the heterogeneity of the studies to date. In contrast to the Peyrin-Biroulet meta-analysis below the 5-ASA and placebo treatment arms were not considered equivalent. There are therefore two sets of comparison of clinical and endoscopic recurrence: thiopurine vs. placebo and thiopurine vs. 5-ASA. Thiopurines were nonsuperior against 5-ASA in predicting clinical recurrence at 12 or 24 months (12 months RR 1.43; 95% CI 0.95–2.16; 24 months RR 1.31; 95% CI 0.95–1.81). They were however superior in any endoscopic recurrence (RR 1.45; 95% CI 1.03–2.06) but not severe recurrence, at 12 months. Combining the only two studies with a formal placebo arm (studies from D'Haens and Hanauer et al.) reveal thiopurine use reduces clinical and severe endoscopic recurrence at 12 months (clinical recurrence RR 0.59; 95% CI 0.38–0.92, NNT = 7, endoscopic recurrence RR 0.64; 95% CI 0.44–0.92, NNT = 4).
The Peyrin-Biroulet et al. meta-analysis however assumed parity between placebo and 5-ASA and therefore produces slightly different outcomes in their analysis, despite examining the same data as the Cochrane review. 198 patients treated with purine analogues (151 AZA, 47 MP) were assessed against 235 'control arms' (154 5-ASA, 40 placebo, 41 metronidazole). Thiopurine treatment was more effective in reducing clinical recurrence at both 12 months (8% mean difference, 95% CI 1–15%, P = 0.021) and 24 months (only Hanauer and Ardizzonne's studies remain in the analysis – 13% mean difference, 96% CI 2–24%, P = 0.018). When removing all but the 'true' placebo arms, they unsurprisingly replicate the Cochrane collaboration's findings.
This trend continues in the endoscopic analysis with subtle differences in the inclusion and outcome criteria: Peyrin-Biroulet et al. assess Rutgeerts' i2–4 'severe' and i3–4 'very severe' recurrence at 12 months. Coupled to the inclusion of clinical trials with 'control arms' as defined above this produces differing results. Their assessment concludes that of 293 patients evaluated at 12 months (82 AZA, 47 MP, 83 5-ASA, 40 placebo and 41 metronidazole) thiopurines were more effective than control arms in preventing severe (i2–4) but not very severe endoscopic recurrence (i3–4) (mean difference 15% 95% CI 1.8–29%). Interestingly, a further post-operative recurrence metanalysis by Shen et al. comparing purine analogues to 5-ASA has in addition included the Reinisch et al. severe endoscopic recurrence intervention study (and excluded the D'Haens metronidazole-AZA cross-over study), and reported further differing results of thiopurine efficacy. This perhaps demonstrates the difficulties clinicians face when presented with three subtly different meta-analyses in the absence of any new controlled data.
Although grouping heterogeneous studies in such a manner permits greater numbers, and therefore power, to detect smaller observed effects, caution must be exercised in applying translation to clinical practice – given the variety in follow-up, drug usage and assessment methods.
Meta-analyses
The Cochrane collaboration review of medical treatment in post-operative Crohn's disease performed in 2009 elegantly demonstrates the heterogeneity of the studies to date. In contrast to the Peyrin-Biroulet meta-analysis below the 5-ASA and placebo treatment arms were not considered equivalent. There are therefore two sets of comparison of clinical and endoscopic recurrence: thiopurine vs. placebo and thiopurine vs. 5-ASA. Thiopurines were nonsuperior against 5-ASA in predicting clinical recurrence at 12 or 24 months (12 months RR 1.43; 95% CI 0.95–2.16; 24 months RR 1.31; 95% CI 0.95–1.81). They were however superior in any endoscopic recurrence (RR 1.45; 95% CI 1.03–2.06) but not severe recurrence, at 12 months. Combining the only two studies with a formal placebo arm (studies from D'Haens and Hanauer et al.) reveal thiopurine use reduces clinical and severe endoscopic recurrence at 12 months (clinical recurrence RR 0.59; 95% CI 0.38–0.92, NNT = 7, endoscopic recurrence RR 0.64; 95% CI 0.44–0.92, NNT = 4).
The Peyrin-Biroulet et al. meta-analysis however assumed parity between placebo and 5-ASA and therefore produces slightly different outcomes in their analysis, despite examining the same data as the Cochrane review. 198 patients treated with purine analogues (151 AZA, 47 MP) were assessed against 235 'control arms' (154 5-ASA, 40 placebo, 41 metronidazole). Thiopurine treatment was more effective in reducing clinical recurrence at both 12 months (8% mean difference, 95% CI 1–15%, P = 0.021) and 24 months (only Hanauer and Ardizzonne's studies remain in the analysis – 13% mean difference, 96% CI 2–24%, P = 0.018). When removing all but the 'true' placebo arms, they unsurprisingly replicate the Cochrane collaboration's findings.
This trend continues in the endoscopic analysis with subtle differences in the inclusion and outcome criteria: Peyrin-Biroulet et al. assess Rutgeerts' i2–4 'severe' and i3–4 'very severe' recurrence at 12 months. Coupled to the inclusion of clinical trials with 'control arms' as defined above this produces differing results. Their assessment concludes that of 293 patients evaluated at 12 months (82 AZA, 47 MP, 83 5-ASA, 40 placebo and 41 metronidazole) thiopurines were more effective than control arms in preventing severe (i2–4) but not very severe endoscopic recurrence (i3–4) (mean difference 15% 95% CI 1.8–29%). Interestingly, a further post-operative recurrence metanalysis by Shen et al. comparing purine analogues to 5-ASA has in addition included the Reinisch et al. severe endoscopic recurrence intervention study (and excluded the D'Haens metronidazole-AZA cross-over study), and reported further differing results of thiopurine efficacy. This perhaps demonstrates the difficulties clinicians face when presented with three subtly different meta-analyses in the absence of any new controlled data.
Although grouping heterogeneous studies in such a manner permits greater numbers, and therefore power, to detect smaller observed effects, caution must be exercised in applying translation to clinical practice – given the variety in follow-up, drug usage and assessment methods.
Source...