NCI Common Toxicity Criteria and Mortality After Chemotherapy for Acute Myeloid Leukemia (AML)

Abstract 1479

Abnormal values for creatinine, bilirubin, AST, and ALT following chemotherapy, in particular in phase 1 trials, are interpreted using NCI common toxicity criteria. By convention, grade 3 or 4 toxicity leads to dose reduction while grade 1 or 2 toxicity does not. However, grade 3 – 4 toxicity is typically asymptomatic, calling into question the relevance of these criteria. In an attempt to provide a more empirical basis for decisions we analyzed the relation between (a) grades of NCI toxicity for bilirubin, creatinine, ALT, and AST following chemotherapy for AML and (b) death. Specifically we examined 752 courses of chemotherapy in 533 patients delivered at the University of Washington or Fred Hutchinson Cancer Research Center for newly- diagnosed or relapsed/refractory AML between 2002 and 2012. Patients had at least 2 sets of renal and liver function test values but were excluded if either pretreatment bilirubin or creatinine was greater than 2.0 to better approximate a population in an early phase trial. We compared initial and maximal lab values in the first 28 days in the 40 courses that resulted in death before 28 days, and in the 712 courses that did not (6% death rate; the death rate per patient was 7%). Maximum lab values for creatinine were strongly related to mortality, with grade O having a 1.0% mortality rate and with each increasing grade having a statistically significant increase in mortality over the immediately preceding level: grade I: 5%, grade II: 22%, grade III or IV: 50%. Bilirubin had a significant increase in mortality at grade II (10.0%, p=.009 relative to grade O) and grade III-IV (15.4%, p=.0007 relative to grade O). AST was also associated with increased mortality at grade II (9.5%, p=.04 relative to grade O) and grade III-IV (15.6%, p =.0006 relative to grade O). ALT was not correlated to mortality, even at grade III-IV (9.5% versus 5.3% at grade 0, p=0.23). Although these correlations obviously do not imply causation, it might be argued that grade II toxicity for creatinine, associated with a 4-fold increase in mortality, should result in dose reduction. In contrast, it might be argued that dose reduction should not occur for grade III/IV ALT toxicity, which correlated poorly with mortality. In any event our data suggest that if the principal purpose of toxicity grading is to guide subsequent dose reduction to prevent death, the NCI toxicity criteria should be revisited.