Association between class III obesity and overall survival in previously untreated younger patients with acute myeloid leukemia enrolled on SWOG S1203

There has been ongoing debate on the association between obesity and outcomes in acute myeloid leukemia (AML). Currently there are few studies that have stratified outcomes by class I obesity, class II obesity, and class III obesity; and a more nuanced understanding is becoming increasingly important with the rising prevalence of obesity. We examined the association between body mass index (BMI) and outcomes in previously untreated AML in younger patients (age ≤60) enrolled in SWOG S1203 (n=729). Class III obesity was associated with an increased rate of early death (p=0.004) and worse overall survival (OS) in multivariate analysis (hazard ratio (HR) 2.48, 95% confidence interval (CI) 1.62–3.80 versus normal weight). Class III obesity was also associated with worse OS after allogeneic hematopoietic cell transplant (HR 2.37, 95% CI 1.24–4.54 versus normal weight). These findings highlight the unique risk of class III obesity in AML, and the importance of further investigation to better characterize this patient population.


Introduction
The prevalence of obesity is rapidly increasing in the United States, rising from 31% to 42% from 1999 to 2017. 1 This is re ected in the rates of obesity in patients with AML, which increased from 36% in 1997 to 49% in 2010. 2 Obesity is associated with increased risk of solid organ and hematologic malignancies, including myelodysplastic syndrome (MDS) and acute myeloid leukemia (AML). 3,4][7][8][9] Given the increasing prevalence of obesity, it is increasingly important to understand the association between obesity and outcomes in patients with AML.
Studies looking at the impact of obesity in patients with AML commonly de ne obesity as body mass index (BMI) ≥ 30 kg/m 2 .Obesity can be further subdivided into class I obesity (BMI 30 to <35 kg/m 2 ), class II obesity (BMI 35 to <40 kg/m 2 ) and class III obesity (BMI ≥40 kg/m 2 ).1][12] A recent study found that class I obesity is associated with a 25% increase in health care expenditure compared with normal BMI, while class III obesity is associated with a 100% increase. 13We set out to examine the association between classes of obesity and outcomes in patients with AML treated with induction chemotherapy and allogeneic hematopoietic cell transplant (HCT).Further strati cation of at-risk patient populations could aid in design of future interventions.

Study Population
We retrospectively reviewed a group of 738 patients with previously untreated AML, as de ned by World Health Organization (WHO) criteria, as part of clinical trial SWOG S1203.This trial was conducted by the SWOG Cancer Research Network that enrolled patients between 2012 and 2015. 14Inclusion criteria were age 18-60 years old, no severe comorbidities precluding intensive chemotherapy, and normal cardiac function.SWOG criteria previously described by Guillermo et al. were used to classify cytogenetic risk.Nine patients who did not receive protocol therapy or had missing BMI were excluded from analysis, leaving a nal study analysis cohort of 729 patients.
Subjects were randomized to one of three induction chemotherapy arms: daunorubicin and cytarabine (DA), idarubicin and cytarabine (IA), and IA with vorinostat.Treatment for the DA arm included: cytarabine 100 mg/m 2 /day intravenous (IV) on days 1-7 and daunorubicin 90 mg/m 2 /day IV on days 1-3.Treatment for the IA arm included: cytarabine 1500 mg/m 2 /day IV on days 1-4 and idarubicin IV 12 mg/m 2 /day days 1-3.
Treatment for the IA and vorinostat arm included: oral vorinostat 500 mg three times per day on days 1-3, cytarabine 1500 mg/m 2 /day IV on days 4-7 and idarubicin 12 mg/m 2 /day IV days 4-6.Dosing was based on body surface area (BSA) for cytarabine, but a BSA cap of 2.5 m 2 was used for daunorubicin and idarubicin.Subjects with adverse risk cytogenetics were recommended to receive HCT during rst complete remission (CR1). 14 used body mass index (BMI) cutoffs per the WHO to de ne underweight (BMI <18.

Discussion
8]15 Here, we made use of a uniform prospectively enrolled group of newly diagnosed younger patients with AML in SWOG trial 1203, where we could show class III obesity was associated with decreased OS both overall and after HCT in CR1.Our results suggest that the degree of obesity may be negatively correlated with outcomes.The pathophysiology of the effect is unknown, but notably while class III obesity was associated with increased early death rate, it was not associated with rate of composite CR.Similarly, for the subset who received HCT, class III obesity was correlated with decreased OS after HCT after adjusting for covariates.
This study evaluated the ndings in one clinical trial in adult patients age 60 or younger that excluded patients with cardiac co-morbidities such as heart failure, new or unstable angina, and prolonged QTc.
Class III obesity is known to be independently associated with increased mortality from heart disease, cancer, and diabetes, and overall reduction in life expectancy. 16,17No comorbidity data were available, and it would have been interesting to examine if comorbidities, such as diabetes, were associated with infection risk or other complications.In addition to comorbidity burden, there have been other proposed mechanisms for worse outcomes in class III versus class I obesity.The "obesity paradox" has been identi ed in cardiovascular disease, where patients with class I obesity have a more favorable prognosis compared to individuals who are normal or underweight.Some patients with BMI ≥ 30 kg/m 2 may have an athletic build and thus increased lean mass. 18Sarcopenic obesity, on the other hand, has been associated with worse prognosis and functional capacity in heart failure. 19,20Future studies investigating the intersection of sarcopenia and obesity in patients with AML may help identify opportunities for intervention to improve outcomes.
Chemotherapy dosing strategy may also impact outcomes in obese patients, and patients in the S1203 study had a BSA cap for anthracycline dose.This study preceded the 2021 American Society of Clinical Oncology (ASCO) guideline recommending full weight-based cytotoxic chemotherapy in obese adults with cancer. 21Weight based chemotherapy dosing is generally not associated with increased toxicity in obese patients, and under-dosing can be associated with worse outcomes.Other studies indicate that different dosage strategies for AML (including total body weight, dosage capped, idealized body weight, and adjusted body weight) were not associated with changes in OS, CR or increased chemotoxicity. 22,23Many AML patients ultimately undergo allogeneic HCT, and a study by Doney et al. found that patients with BMI ≥ 35.0 kg/m 2 had worse non-relapse mortality associated with increased conditioning intensity. 246][27] A multicenter study of 4215 patients with AML did not nd any differences in transplant-related mortality or OS in obese patients (BMI > 30 to 34 kg/m 2 ) and morbidly obese patients (BMI ≥ 35 kg/m 2 ). 28In contrast, another study of patients with AML undergoing allogeneic HCT found that obese patients (BMI ≥ 30 kg/m 2 ) had higher rates of non-relapse mortality and shorter OS after HCT. 29 The subjects included in our analyses were relatively homogenous, which may have reduced confounding factors present in prior studies.Our study also analyzed class III obesity (BMI ≥ 40 kg/m 2 ) separately, and found that Class III obesity seemed to portend worse outcomes than class I and II obesity after HCT.It is unlikely that delayed transplantation in obese patients in our analyses contributed to the worse outcomes, since there was no signi cant difference in the time to transplant between different BMI groups.Interestingly, receipt of HCT was not associated with increased OS in multivariate analysis.This could be due to the study design: all adverse risk patients were recommended for HCT, and intermediate risk patients were recommended for HCT per local institutional standards.
Our study was limited by its retrospective nature and susceptibility to selection bias.A large majority of patients in our study (88%) had good performance status.Patients with the highest burden of comorbidities from obesity may have been disproportionally excluded due to the strict inclusion criteria.While this limits generalizability of our ndings to patients with poor performance status, inclusion of more patients with poor performance status may have introduced more confounding factors in multiple BMI groups.There were also fewer patients with class II obesity than class III obesity in our study, and the sample size of patients with class I and II obesity may have limited our ability to identify a signi cant effect.