Increased body mass index is a risk factor for acute promyelocytic leukemia

Abstract Introduction Obesity has become increasingly prevalent worldwide and is a risk factor for many malignancies. We studied the correlation between body mass index (BMI) and the incidence of acute promyelocytic leukemia (APL), non‐APL acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), and control hospitalized patients without leukemia in the same community. Methods Multicenter, retrospective analysis of 71 196 patients: APL (n = 200), AML (n = 437), ALL (n = 103), nonleukemia hospitalized (n = 70 456) admitted to University of Maryland and Johns Hopkins Cancer Centers, and University of Maryland Medical Center. Results Patients with APL had a significantly higher unadjusted mean and median BMI (32.5 and 30.3 kg/m2) than those with AML (28.3 and 27.1 kg/m2), ALL (29.3 and 27.7 kg/m2), and others (29.3 and 27.7 kg/m2) (P < .001). Log‐transformed BMI multivariable models demonstrated that APL patients had a significantly higher adjusted mean BMI by 3.7 kg/m2 (P < .001) or approximately 10% (P < .01) compared to the other groups, when controlled for sex, race, and age. Conclusions This study confirms that when controlled for sex, age, and race there is an independent association of higher BMI among patients with APL compared to patients with ALL, AML, and hospitalized individuals without leukemia in the same community.


INTRODUCTION
The prevalence of overweight and obese individuals in developed western countries has been increasing at an alarming rate throughout the last few decades [1]. In addition to cardiovascular disease, kidney disease, diabetes, and musculoskeletal disorders, the International Agency for Research on Cancer (IARC) has acknowledged 13 cancers associated with overweight and obesity: gastrointestinal malignancies (esophageal adenocarcinoma, and cancers of gallbladder, liver, gastric cardia, pancreas, colon, and rectum), gynecological cancers (ovary, corpus uteri), breast cancer in postmenopausal women, meningioma, multiple myeloma, and thyroid and renal cell cancers [2][3][4][5][6][7][8].
The association between excess body fat and different types of leukemias is not well characterized. In experimental animal models, an inverse association between caloric or dietary restriction and leukemia was observed [8,9]. A meta-analysis of 21 prospective cohort studies demonstrated that obesity was associated with increased incidence (relative risk [RR] 1.26, 95% confidence interval [CI] 1.17-1.37, P < .001) and mortality (RR 1.29, 95% CI: 1.11-1.49, P = .001) of leukemia in adults among all subtypes including acute myeloid leukemia (AML), acute lymphoblastic leukemia (ALL), chronic myeloid leukemia (CML), and chronic lymphocytic leukemia (CLL) [10]. For AML, obesity was associated with an increased incidence, with RR of 1.53 (95% CI: 1.26-1.85, P < .001), compared to those with normal weight [10]. Once AML is categorized into acute promyelocytic leukemia (APL) versus non-APL, patients with APL had a significantly higher body mass index (BMI) than those with non-APL [7,11], with the BMI of patients with non-APL subtypes being similar to that of the general population when adjusted for age, sex, and race [11].
We performed a retrospective analysis of BMIs of patients with newly diagnosed APL, AML (non-APL subtypes), and ALL at the time of hospital admission, compared with patients without APL, AML, or

Study population and design
This two-center retrospective study was approved by the University To account for missing data, we multiply imputed missing data using the "mi" package [17]. All statistical tests were two-sided and P-values

Characteristics of patients subdivided by type
Baseline patient characteristics divided by type are summarized in  Figure 1). There was a significant difference in age and sex between the APL, AML, ALL, and other patients as well, with ALL and APL patients younger and APL patients being more frequently female (Table 1).

Multivariable models
We used multivariable linear regression to estimate the difference in BMI between the subgroups adjusted for gender, age, and race.  Abbreviations: AML, acute myeloid leukemia; APL, acute promyelocytic leukemia. a Caucasian was used as the reference for each race. b Acute lymphoblastic leukemia was used as the reference for comparison.

BMI and baseline characteristics among APL patients
In patients diagnosed with APL, we used regression models to check for statistical association between BMI (independent variable) and specific baseline characteristics (dependent variables). None of the univariable models showed statistically significant associations of higher BMI with the examined characteristics including risk status based on Sanz score or Breccia, WBC and platelet counts at diagnosis,  (Table 4).

DISCUSSION
The purpose of this study was to assess the association between elevated BMI and the diagnosis of APL when compared to the non-APL acute leukemias and nonleukemic population in the same community.
BMI of APL patients was significantly higher than those of patients is an independent association of higher BMI with APL compared with ALL, AML, and other diagnoses.
The data in the current literature on the association of body fat and leukemias are somewhat conflicting, and with this project we tried to clarify this more. Our observation aligns with previous literature describing higher BMI in APL patients compared to AML patients, with BMI of AML patients being similar to that of the general population averages [6,7,10,11]. Of note, other studies have utilized general population averages of BMI as a control; no study has directly used individual data of patients without leukemia admitted to the hospital in the same community as a control. On the other hand, in contrast to prior retrospective studies demonstrating an association between obesity and incidence of all leukemia subtypes (including both chronic and acute leukemias) [10,19], our study suggests that patients with AML and ALL are overweight but not obese, with BMIs similar to those of the general hospitalized population in the community. Of note, all subtypes of leukemia were included in one group in one of the studies [19], and APL patients were included among AML patients in the other study [10]. Additionally, these studies often included self-reported height and weight values, which can leave room for error, whereas our study used directly measured values [10,19].
It has been demonstrated that among the US population, there is a higher proportion of obesity among those 40 years and older compared with those 20-39 years old, with the greatest proportion of obese individuals among non-Hispanic African Americans [20]. We attempted to mitigate these variations by age and race by correcting BMI for age, race, and sex. Our data suggest that regardless of age/race/sex, APL patients have higher BMI than those with AML and ALL. One limi-tation of our study is that our data are based solely on hospitalized patients, which may lead to an inherent selection bias; however, based on National Health and Nutrition Examination Survey data, the mean BMI of adults in the United States is 29.1 kg/m 2 [21], which is similar to the BMI of 29.3 kg/m 2 found in hospitalized patients in our study.
The reason for the relationship between obesity and APL is unclear.
Chronically increased insulin levels have been associated with cancers of breast, colon, pancreas, and endometrium [22][23][24][25][26]. Obesity can lead to insulin resistance, in turn causing an increase in insulin secretion. Chronic hyperinsulinemia can have tumorigenic effects thought to be due to the direct action of insulin on the insulin receptors in the preneoplastic target cells, or perhaps because hyperinsulinemia causes changes in endogenous hormone metabolism such as the promotion of insulin-like growth factor-1 (IGF-1) [25,27]. One study in mice demonstrated that increased fat intake leading to weight gain can promote leukemogenesis likely through the IGF-1 pathway [28].
While some ALL and AML cell lines express insulin and IGF-1 receptors (IGF-1R) [29][30][31], almost all of APL cell lines such as HL-60, NB4, and PL-21 express abundant IGF-1R protein and proliferate with IGF and IGF analogs stimulation and their growth and basal DNA synthesis decrease with monoclonal antibodies directed against the IGF-1R and other IGF antagonists [32][33][34]. Additionally, the leptin receptor, which is proliferative and bears antiapoptotic effects when activated [35,36], is selectively upregulated in the APL cells, whereas the normal promyelocytes lack its expression, suggesting a possible link between the leptin-rich environment in obese individuals and development of APL [35,36].
In conclusion, we highlight that there is an independent association of higher BMI in patients with APL compared to patients with ALL, AML, and control nonleukemic patients admitted to the hospital in the same community. On the basis of our findings, supported by the prior literature reports, we suggest adding APL to the list of the cancers that are associated with overweight and obesity.

CONFLICT OF INTEREST
The authors declare that there is no conflict of interest. Smith, and Mohammad Imran assisted with acquisition of study data.

AUTHOR CONTRIBUTIONS
Ashkan Emadi conceived the idea, conceptualized the hypotheses, and designed and supervised all areas of the study. All the authors provided critical feedback, edited, and approved the typescript.