Predictors of Early Mortality, Response, and Survival in Newly Diagnosed Acute Myeloid Leukemia (AML) Using a Contemporary Academic Cohort

Background

Recent advances have led to higher response rates and improved survival in patients with newly diagnosed AML. However, early death, lack of response, and long term leukemia-free survival remain important challenges. Early mortality in AML has been reported to be as high as 20-30%. Similar to relapse risk, disease related factors (along with patient and treatment factors) may also contribute significantly to early mortality. Here we investigated predictors of early (4- and 8-week) mortality, response, relapse-free survival, and overall survival in a contemporary cohort from a single large academic medical center.

Methods

We analyzed all newly diagnosed patients with AML presenting to and treated at our institution from January 2012 to January 2020. For 4- and 8-week mortality and overall response rate (CR/CRi), logistic regression models were generated to identify factors associated with these outcomes. Kaplan-Meier methods were used to determine the median of the time to event outcomes. Cox proportional hazards models were used to identify any association with each of the variables and survival outcomes. Multivariate models were performed for all the outcomes, the models included variables with p<0.25 in the univariate analysis. Backward elimination methods were used to determine a final multivariate models for each outcome.

Results

Our cohort included 1576 consecutive patients (pts) with newly diagnosed AML with a median follow-up of 11 months (range 0 - 96). The median age is 66 (range: 17 - 94), with 66% of pts being ≥ 60 years (yrs) old. By 2017 ELN risk classification, 44% were classified as adverse, 31% intermediate, and 25% were favorable risk. To simplify induction regimens, we classified patients as receiving intensive (high dose Ara-C based), low intensity, and low intensity with venetoclax. 44% received intensive, 42% received low intensity, and 14% received low intensity with venetoclax therapy.

The 4- and 8-week mortality rates were 5% and 11%, respectively. Among pts < 60 yrs the 4- and 8 week mortality rates were 3% and 5%, respectively; while those rates in pts ≥ 60 yrs were 6% and 14%, respectively. In the final reduced models the following factors are associated with both increased 4- and 8-week mortality: elevated baseline total bilirubin, age ≥ 60, and ELN intermediate and adverse risk classifications, when compared to ELN favorable. Elevated baseline WBC and bone marrow blasts were associated with 8-week but not 4-week mortality. Baseline platelet count ≥ 8k/µL was associated with reduced 4 and 8-week mortality. Among 97 pts who died between 4 and 8 weeks 87 patients (90%) had active disease or no response to their initial therapy, suggesting disease refractoriness as an important contributor to early death.

The factors associated with decreased CR/CRi were increased baseline peripheral blasts, total bilirubin, serum creatinine, age ≥ 60, and ELN risk classifications when compared to favorable risk disease. Treatment with low intensity plus venetoclax and intensive therapy were associated with increased CR/CRi rates as was baseline platelet count ≥ 8k/µL.

Median relapse-free survival for the cohort is 8 months (95% CI: 7- 9). Median overall survival for the cohort is 13 months (95% CI: 12- 14), with median OS for those ≥ 60 yrs: 10months (95%CI: 9– 11) and those < 60 yrs: 37months (95%CI: 24– 67). Factors associated with increased hazard of both relapse and death were increased baseline total bilirubin, creatinine, age ≥ 60, and ELN risk classifications when compared favorable disease. Factors associated with decreased hazard of both relapse and death were treatment with low intensity plus venetoclax and intensive therapy when compared to low intensity, and baseline hemoglobin ≥ 9 g/dl and platelet count ≥ 8k/µL.

Conclusion

In this study, using a contemporary cohort of consecutive, unselected AML patients, we identified parameters associated with early mortality, response, relapse-free, and overall survival. Compared to published historical data, patients at a high-volume academic center experienced lower early mortality and favorable long term outcomes. We identify both patient (i.e. baseline organ dysfunction) and disease (i.e. ELN risk) intrinsic characteristics that have a significant impact clinical outcomes in patients with AML. Further analysis including the impact mutational groups outside of ELN and treatment on or off investigational trials will be presented.