Successful application of a previously derived prognostic index in the analysis of a randomised trial of 281 patients with high grade non-Hodgkin's lymphoma (HIGNHL)

Abstract

The selection of patients for experimental therapies for high grade non-Hodgkin's lymphoma (NHL) is now recognised to be very influential in affecting results. We showed previously that simple clinical indices could be used to create an index of risk of death in a series of 972 patients. We wished to test this prognostic index in a subsequent randomised treatment trial of CHOP-based chemotherapy. 281 patients with high grade NHL randomised between two chemotherapy designs were followed up from 1-6 years and survival analysed to develop a Cox model which was then compared against the previously described prognostic index. The previous index based on age, sex, performance status, stage and symptoms was similar to the data-derived model. Three year survivals for the best (109), intermediate (93) and worst (79) cohorts were 69%, 50% and 24% respectively, similar to the survivals in the previous series. CHOP-based chemotherapy is probably adequate for around 40% of cases of high grade NHL. The prospective test of a prognostic index shows that we can be confident about selecting poor-risk patients for trials of novel therapies, including dose intensification.