Background: Response Evaluation Criteria in Solid Tumors (RECIST) are widely used to assess the effect of chemotherapy in patients with cancer. We hypothesised that the change in unidimensional tumour size handled as a continuous variable was more reliable than RECIST in predicting overall survival (OS).
Methods: The prospective Pharmacogenoscan study enrolled consecutive patients with non-small-cell lung cancer (NSCLC) at any stage seen between 2005 and 2010 at six hospitals in France, given chemotherapy. After exclusion of patients without RECIST or continuous-scale tumour size data and of those with early death, 464 patients were left for the survival analyses. Cox models were built to assess relationships between RECIST 1.1 categories or change in continuous-scale tumour size and OS. The best model was defined as the model minimising the Akaike Information Criterion (AIC).
Results: OS was 14.2 months (IQR, 7.3-28.9 months). According to RECIST 1.1, 146 (31%) patients had a partial or complete response, 245 (53%) stable disease, and 73 (16%) disease progression. RECIST 1.1 predicted better OS than continuous-scale tumour in early (<6 months) predicted survival analyses (p = 0.03) but the accuracy of the two response evaluation methods was similar in late (≥6 months) predicted survival analyses (p = 0.15).
Conclusion: In this large observational study, change in continuous-scale tumour size did not perform better than RECIST 1.1 in predicting survival of patients given chemotherapy to treat NSCLC.
Trial registration: NCT00222404.