A Clinical Prognostic Index for Patients Treated with Erlotinib in National Cancer Institute of Canada Clinical Trials Group Study BR.21

    loading  Checking for direct PDF access through Ovid

Abstract

Introduction:

BR.21 demonstrated significant survival benefit for non-small cell lung cancer patients receiving erlotinib compared with placebo. We undertook to characterize, by exploratory subset analysis, patients less likely to benefit from erlotinib.

Methods:

Using stratification and potential prognostic factors, Cox regression with stepwise selection with minimum Akaike Information Criteria was used to separate erlotinib patients into risk categories based on 10th, 50th, and 90th percentiles of prognostic index scores. The hypothesis was that characteristics of treated patients in the highest risk group would be predictive of lack of benefit from erlotinib when comparing erlotinib to placebo patients in the same risk group.

Results:

Ten factors (smoking history, performance status, weight loss, anemia, lactic dehydrogenase, response to prior chemotherapy, time from diagnosis, number of prior regimens, epidermal growth factor receptor copy, and ethnicity) were predictive of overall survival for erlotinib-treated patients and were used in the final model. Four risk groups were derived from the index score of the Prognostic Model: Low Risk (HR = 0.34, p < 0.001), Intermediate Low and Intermediate High Risk (HR 0.76, p = 0.05; HR 0.92; p = 0.51) and High Risk (HR 1.07; p = 0.78). Median survivals for erlotinib (placebo) patients in each group were 20.6 (8.9), 10.4 (7.6), 4.0 (4.1), 1.9 (2.3) months. The trend test showed that higher risk was associated with shorter survival (p < 0.001) and less treatment effect (p = 0.03).

Conclusions:

By establishing a prognostic model, we identified a small group of patients who did not seem to benefit from erlotinib in this study. This model requires prospective validation to confirm that it is both prognostic and predictive of outcome.

Related Topics

    loading  Loading Related Articles