Lack of correlation between objective response and death rate in multiple myeloma patients treated with oral melphalan and prednisone.

An analysis of survival by response category was performed on 76 patients with stage II and III multiple myeloma, who were treated with oral melphalan and prednisone. The analysis demonstrated a survival advantage for responders over non-responders only in stage III patients (32.3 months for responders vs. 15.6 months for non-responders, p = 0.03). However, two possible sources of error must be considered: a) the poor prognosis of early responders that may adversely affect the survival of all responders, and b) the bias introduced by the 'guarantee time' of responders (i.e., the time on-study required to detect the response). Exclusion from the analysis of the unfavourable subgroup of 'early' responders (median survival 14.7 mos.) provided an improvement of the difference in survival between the remaining 'slow' responders and non-responders in stage III (p = 0.005) as well as in the series as a whole (p = 0.025). Because of the consistent 'guarantee time' of slow responders, the Mantel-Byar test (which credits survival to responders only after the response has been obtained) was then applied. The survival advantage of slow responders over non-responders, previously observed in all patients, particularly those with stage III, was not confirmed by the Mantel-Byar test (chi-square 0.831 and 1.457, respectively), thus supporting the hypothesis of an equal death rate over time in each response category. It therefore appears that the usual response criteria (which require at least a 50% reduction of the myeloma protein) should perhaps be reassessed, as they seem to be an inadequate parameter for evaluation of treatment effectiveness in multiple myeloma.