Hemos enviado hoy mismo esta nota a la EMA:
Trastuzumab for gastric adenocarcinoma: mistake in the survival analysis by subgroups
We have reviewed the ToGA study of trastuzumab in adenocarcinoma of the gastric or gastroesophageal junction, in order to determine which patients benefited from treatment. This is a phase III randomized trial, published in The Lancet on August 28, 20101. In reviewing the overall survival analysis by subgroups in the forest plot (figure 3 of referred article), we have seen that 95% confidence intervals of HR for patients with measurable and not measurable disease do not overlap at all, which would probably mean a statistically significant difference. In contrast, 95% confidence interval of HR for patients with performance status (ECOG) 2 is completely contained within the interval calculated for patients with performance status 0-1, and that would mean that there is no statistically significant difference. The forest plot shows the same results as outlined in the European Public Assessment Report (EPAR) for extension of indication to gastric adenocarcinoma2 (p. 22). Accordingly, patients with measurable disease would not benefit from treatment, and there wouldn´t be any difference between patients with better or worse performance status.
However, the EPAR report includes statistical analysis, and the table “Summary of multiple Cox regression for overall survival” (p. 23) shows p = 0.62 and HR = 1.12 for measurable vs. not measurable disease, and p <0.0001, HR = 0.40 for performance status 0-1 vs. 2. Accordingly, differences were found only among patients with better or worse performance status. This seems a contradiction in a relevant matter, since it influences the selection of patients eligible for treatment. We could not check what is wrong with the published data, as the article from The Lancet did not include an analysis of statistical significance by subgroups. The documentation issued by the FDA doesn’t show a similar analysis that allows us to solve the question. It would be necessary to confirm the statistical significance of differences for overall survival in these subgroups, to undo this mistake and properly address the treatment with trastuzumab in patients with metastatic gastric adenocarcinoma.
- Bang YJ, Van Cutsem E, Feyereislova A, et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for HER2-positive advanced gastric or gastrooesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet 2010; 376:687-97.
- European Medicines Agency. Variation Assessment Report for Herceptin (trastuzumab). Procedure No. EMEA/H/C/278/II/0047. http://www.ema.europa.eu/docs/en_GB/document_library/EPAR_-_Assessment_Report_-_Variation/human/000278/WC500074921.pdf [accesed: 14/7/2011].