Associations between Complete Remissions (CRs) with 7+3 Induction Chemotherapy for Acute Myeloid Leukemia and 2-3 Year Survival ("Potential Cure") over the Past Four Decades: Analysis of SWOG Trial Data

Background: CRs are known to be associated with improved overall survival (OS) for acute myeloid leukemia (AML) patients (pts) treated with curative-intent, induction therapy such as 7+3. Here we extend these findings by quantifying the relationship between CR and 2-3 year survival ("potential cure") and evaluate if and how any relationship has changed over the past four decades.

Patients and Methods: We analyzed 1247 patients randomized to 7+3 arms from 5 SWOG studies and restricted to patients age 65 or younger: S8600 (n=530), S9031 (n=98), S9333 (n=57), S0106 (n=301), S1203 (n=261). S8600 enrolled patients in the 1980s, S9031 and S9333 in the 1990s, S0106 in the 2000s, and S1203 in the 2010s. S9031 and S9333 were analyzed together. All 5 protocols gave 7+3 per contemporary standard, which changed over time. In S8600, S9031, and S9033 the ara-C and daunorubin doses were 200mg/m² and 45mg/m², in S0106 100mg/m² and 60mg/m², and in S1203 200mg/m² and 90mg/m². CR was defined morphologically. We estimated the hazard rates for death over the first 5 years from treatment in S0106 and S1203, and these findings led us to use logistic regression models to evaluate associations with the endpoint of alive at 2 or 3 years. Patients followed for less than 2 and 3 years were excluded. Importance of predictors in the logistic regression models was evaluated with the Wald Chi-squared test minus degrees of freedom.

Results: Overall survival has improved dramatically over the 4 decades (Figure 1). The hazard for death in the two most recent studies, S0106 and S1203, was stable for the first two years and then fell by half after year 2 and further fell in year 4 (Table 1). In further analyses we used both year 2 and year 3 (with similar results, year 2 data presented below) as a date to denote patients as long-term survivors and potentially cured, although their subsequent death rate was still considerably higher than expected for an age-matched population. Across the 5 studies the median time to CR was 34 days, (1^st and 3^rd quartiles 27 and 47 days, min and max 12 and 164 days). The effects of being in CR by day 60 or day 100 were similar (results presented for day 100 below). In multivariable analysis for the outcome alive at year 2 or 3 we evaluated CR by day 100, age, gender, performance status (PS), pre-study white blood cell count (WBC), platelets, and blasts, de novo versus secondary AML, cytogenetic risk, and decade of study protocol (Table 2). Study protocol was the most important variable, with CR by day 100 the second most important variable; the 3^rd and 4^th most important variables were cytogenetics and age, respectively (Figure 2). FLT3 and NPM1 data were available for patients on S0106 and S1203. Inclusion of those covariates in a multivariable model restricted to the two studies had FLT3 as the 4^th most important predictor in lieu of age.

Conclusion: Protocol/decade, which captures all the other factors not included in our multivariable model, was the most important predictor of being alive at year 2 or 3. The next most important factor was achievement of first morphologic CR by day 100. In addition to serving as an (imperfect) surrogate for OS, achievement of CR also provides important information regarding 2- and 3-year survival and potential cure.

Support: NIH/NCI grants CA180888 and CA180819

Acknowledgment: The authors wish to gratefully acknowledge the important contributions of the late Dr. Stephen H. Petersdorf to SWOG and to study S0106.

View largeDownload slide

View largeDownload slide

Close modal