To determine if chemotherapy dose intensity (DI) influences treatment outcome, 60 published studies in limited- and extensive-stage small-cell carcinoma of the lung (SCCL) were retrospectively analyzed for relationship between intended Dl and response (complete response [CR] or partial response [PR]) or median survival (MS). Agents used in the regimens included cyclophosphamide (C), doxorubicin (A), vincristine (V), etoposide (E), and cisplatin (P). Relative DI (RDI) of each study regimen was calculated against a reference regimen, and weighted regression analysis was used. Additionally, analysis of individual drug RDI within combinations was performed. For CAV, increasing RDI of the regimen showed no correlation with outcome. For the individual drugs, C RDI correlated positively, while A RDI correlated negatively with attainment of CR in limited disease, but both only after unduly influential observations were eliminated. In extensive-stage disease, A RDI correlated positively with CR +, PR but only in randomized trials, and this correlation lost statistical significance after unduly influential observations were eliminated. For CAE and CAVE, the RDI of the regimens correlated positively with MS in extensive-stage disease as did the C RDI. In limited disease, the C RDI correlated negatively with MS. For EP, no significant correlations were seen. We conclude that DI-outcome correlations are not consistent for these chemotherapy regimens in SCCL. Meta-analysis of retrospective data can generate hypotheses for testing in prospective clinical trials, but study sample and method of analysis can appreciably affect conclusions.
