Physical Activity, Sitting Time, and Risk of Myelodysplastic Syndromes, Acute Myeloid Leukemia, and Other Myeloid Malignancies

Myeloid neoplasms are a heterogeneous group of rare tumors associated with irregular hematopoiesis. Among this group of neoplasms, there are two major types of myeloid leukemia, acute myeloid leukemia (AML) and chronic myeloid leukemia (CML), and another set of malignancies called myelodysplastic syndromes (MDS; ref. 1). MDS was previously thought to be a premalignant condition, but as of 2001, became reportable to tumor registries (2). Because their incidence is low, there are few epidemiologic studies of lifestyle factors related to the risk of myeloid neoplasms (https://seer.cancer.gov/statfacts/html/). However, many myeloid neoplasms have poor 5-year relative survival rates; therefore, the identification and promotion of lifestyle behaviors amenable to prevention could have a marked impact on mortality.

There is growing evidence suggesting that physical activity plays an important role in cancer etiology and prevention. For example, there is convincing evidence that regular moderate-to-vigorous physical activity (MVPA) decreases the risk of colorectal cancer, and probable evidence that regular MVPA decreases the risk of breast and endometrial cancers (3). In addition, findings from a recent large pooled analysis of prospective cohort studies found an association between high levels of physical activity and a lower incidence of eight additional types of cancer, including all myeloid leukemia [Hazard ratio (HR), 0.80, confidence intervals (95% CI), 0.70–0.92, 90^th vs. 10^th percentile of leisure-time physical activity; ref. 4]. Importantly, this study only compared risk of myeloid leukemia in the extremely active (90^th percentile) versus the extremely inactive (10^th percentile), creating the need for a more detailed analysis. Overall, studies of MVPA and myeloid neoplasms, myeloid leukemia, or subtypes are rare, making it difficult to draw conclusions about these relationships. Furthermore, it is unclear if sitting time is independently associated with risk of myeloid neoplasms, as the relationship between sitting time and this group of diseases has not yet been studied.

Physical inactivity, if found to be associated with the risk of myeloid neoplasms, would be one of the few known modifiable behaviors associated with the risk of myeloid neoplasms. The primary aims of this study were to evaluate the associations of MVPA and total sitting time with risk of myeloid neoplasms overall and myeloid leukemia specifically. We also explored associations of MVPA and sitting time with risk of AML and MDS.

The Cancer Prevention Study-II (CPS-II) is a prospective study of cancer mortality initiated by the American Cancer Society in 1982, and is described in detail elsewhere (5). In 1992, a subset of the CPS-II participants from 21 states were invited to join the CPS-II Nutrition Cohort (CPS-II NC; ref. 6). The CPS-II NC, which includes over 184,000 participants aged 40 years and older at baseline, was established to update demographic and lifestyle information and capture cancer incidence. CPS-II NC participants completed a 10-page questionnaire at home and received subsequent questionnaires every two years beginning in 1997. The 1999 CPS-II NC survey was used as baseline in the current analysis to most closely reflect the time at which MDS first became reportable to tumor registries. All aspects of the CPS-II are approved by the Emory University Institutional Review Board.

Among the 151,343 men and women who completed the 1999 survey, participants were excluded if they: were lost to follow-up (n = 5,742), had a personal history of cancer other than non-melanoma skin cancer (n = 27,787), were missing information on MVPA or sitting time (n = 5,524), or had a missing or implausible body mass index (BMI; n = 3,260). The remaining 109,030 participants were eligible for this analysis. Sociodemographic characteristics of excluded participants were not different from those in the final analytic cohort. Person-time was calculated as the number of years between the 1999 survey return date and date of myeloid neoplasm diagnosis, date of censoring, date of death, or the end of follow-up on June 30, 2013, whichever came first. Reasons for censoring included a reported hematologic cancer that could not be verified, diagnosis of another type of cancer, or failure to return subsequent follow-up surveys (unless deceased).

Information regarding leisure-time MVPA was collected with the question, “During the past year, what was your average total time per week spent at each of the following activities?” with items including walking, jogging/running, dancing, lap swimming, playing tennis or racquetball, bicycling/exercise machines, and engaging in aerobics/calisthenics. The midpoint values from responses [none, 1–19 min/wk, 20–59 min/wk, 1 h/wk, 1–1.5 h/wk, 2–3 h/wk, 4–6 h/wk, 7–10 h/wk, and 11+ h/wk (11 h/wk used instead of midpoint)] were multiplied by the corresponding metabolic equivalent (MET) value of each activity. Given the mean age of the cohort, conservative MET values for each activity were assigned on the basis of the 2011 Compendium of Physical Activities to calculate metabolic equivalent hours/week (MET-h/wk) of each activity (7). Values were summed and re-categorized as none, >0–<7.5, 7.5–<15, 15-<22.5, or ≥22.5 MET-h/wk of MVPA. These cut-points are consistent with 2018 U.S. Federal MVPA recommendations as: none, less than guidelines, meeting guidelines, double to triple guidelines, and triple guidelines or more. The second MVPA category (>0–<7.5) served as the referent for all analyses as: (i) assigning the lowest MVPA category as the referent may lead to biased risk estimates if inactivity is a result of disability or illnesses, and (ii) this group was larger than the lowest activity group.

Time spent sitting was assessed with the question, “During the past year, what was your average total time per week spent at each of the following activities?” with activities including: sitting at work, sitting or driving in a car/bus/train, sitting or lying watching TV, sitting at home reading, and other sitting. Responses included: none, 1–39 min/wk, 40–89 min/wk, 1.5 h/wk, 2–3 h/wk, 4–6 h/wk, 7–10 h/wk, 11–20 h/wk, 21–30 h/wk, 31–40 h/wk, or 40+ h/wk. The midpoint value from each sitting category [i.e., 20 min, 65 min, 1.5, 2.5, 5, 8.5, 15.5, 35.5, and 40 hours (40 h/wk used instead of midpoint)] was summed and re-categorized as 0 ≤ 3 (referent), 3–5, or ≥6 h/d of total sitting (8). As time spent watching television is a common leisure-time sedentary behavior that may be reported more accurately than total sedentary time (9), time spent sitting or lying watching TV was separately assessed in the following categories: <1 (referent), 1–2, and ≥3 h/d.

Myeloid neoplasms were defined using the 2008 World Health Organization Classification Scheme (1). Subtypes were categorized on the basis of the International Classification of Diseases for Oncology, second and third editions (AML ICD-O: 9840,9861,9866–9867, 9870, 9871–9874, 9891, 9895–9897, 9910, 9920, 9930, 9931; CML ICD-O: 9863, 9875, 9876; other myeloid leukemias ICD-O: 9860,9864,9868; MDS ICD-O: 9980, 9982, 9983, 9984, 9985, 9986, 9987, 9989, 9991, 9992).

We identified 255 participants diagnosed with a myeloid leukemia (n = 155 AML, n = 63 CML, and n = 37 other myeloid leukemia) between 1999 and June 2013. Between January 1, 2001 (when MDS became reportable to tumor registries) and the end of follow-up, we identified 154 cases of MDS. Collectively, all myeloid leukemia and MDS cases comprise the 409 myeloid neoplasms included in this analysis.

Most cases were identified through biennial linkage of the CPS-II cohort with the National Death Index and subsequently verified by linkage with state tumor registries (n = 206). An additional 147 cases were identified by self-report on CPS-II NC surveys and subsequently verified by linkage with state tumor registries. Finally, 56 cases were identified as deaths, but did not link with a tumor registry, and therefore date of death was used as a proxy.

Cox proportional hazards regression models were used to compute multivariable adjusted HRs and 95% CIs for the associations of MVPA and sitting time with risk of: (i) all myeloid neoplasms (all myeloid leukemias and MDS), (ii) all myeloid leukemias (AML, CML, and all other myeloid leukemias), (iii) AML, and (iv) MDS; there were too few cases to assess associations with CML alone. Models adjusted for age (continuous), sex, and other potential covariates, including: race/ethnicity (white, other), education (high school or some college, college graduate or higher, unknown), smoking status (never, current, former, unknown), years since quitting among former smokers (<10, 10–19, ≥ 20 years), cigarette frequency and smoking duration among current smokers (<20 cigarettes/d and smoking ≤35 years, <20 cigarettes/d and smoking >35 years, 20+ cigarettes/d and smoking ≤35 years, 20+ cigarettes/d and smoking >35 years), alcohol use (non-drinker, <1, 1, >1 drinks/d, unknown), pesticide exposure (yes, no, unknown), formaldehyde exposure (yes, no, unknown), and BMI (continuous). Mutual adjustment for sitting time (in models for associations of MVPA) or MVPA (in models for associations of sitting time) did not significantly change results. Two sensitivity analyses were conducted: (i) among participants who were life-long non-smokers or former smokers of more than 20 years at baseline to account for potential residual confounding and (ii) excluding myeloid malignancies identified during the first year of follow-up to address the possibility of reverse causality. Restricted cubic splines were used to characterize the dose–response relationships with continuous MET-h/wk of MVPA (10). Interaction terms between MVPA and follow-up time were created to test the Cox proportional hazards assumption, which was not violated for any association.

Participants were followed for 10.0 years on average (maximum 13.8 years). The mean (SD) age at the start of follow-up was 69 (6.1) years, and the mean age at diagnosis was 75 (5.4) years. Overall, 39.5% of participants reported an insufficient amount of MVPA (less than 7.5 MET-h/wk) and 12% of participants reported 6+ hours of sitting per day. The most common type of MVPA was walking (90.5% of active participants reported walking), followed by biking (29.8% reported biking) and dancing (10.6% reported dancing). Compared with more active participants, less active participants were more likely to be current smokers, had less education, and had a higher average BMI at baseline (Table 1).

Associations of MVPA with risk of myeloid malignancies are shown in Table 2. Compared with insufficient MVPA (>0–<7.5 MET-h/wk), the HR (95% CI) for 7.5–<15 MET-h/wk MVPA and risk of all myeloid neoplasms was 0.74 (95% CI, 0.56–0.98) and for 15–<22.5 MET-h/wk was 0.75 (95% CI, 0.53–1.07); however, there was no statistically significant association with higher amounts of MVPA (22.5+ MET-h/wk, HR, 0.93; 95% CI, 0.73–1.20). Restricted cubic splines confirmed the presence of a non-linear relationship between MVPA and risk of myeloid neoplasms (P = 0.0058; Supplementary Fig. S1). MVPA was not associated with the risk of myeloid leukemia. Total sitting time and sitting while watching TV (Supplementary Table S1) were similarly not associated with the risk of myeloid neoplasms or myeloid leukemia.

Results stratified by the myeloid neoplasm subtypes AML and MDS varied. For MDS (n = 154 cases), the HRs for 7.5–<15 and 15–<22.5 MET-h/wk were 0.57 (95% CI, 0.35–0.92) and 0.51 (95% CI, 0.27–0.98), respectively, and there was no association among the most active (HR 0.93; 95% CI, 0.63–1.37; Supplementary Fig. S2). MVPA was not associated with the risk of AML (n = 155 cases). Total sitting time and sitting while watching TV were not associated with the risk of MDS or AML.

Sensitivity analyses (i) among non-smokers and long-term former smokers and (ii) excluding the first year of cases showed similar results for all analyses (Table 3).

Results from this large, prospective cohort study suggest there may be an inverse association between MVPA and the risk of all myeloid neoplasms, more specifically MDS, with no evidence of reduced risk among the most active group. Sitting time was not associated with risk of myeloid neoplasms in the current study, which to the best of our knowledge is the first to examine this association.

Our finding of a possible inverse association between MVPA and risk of myeloid neoplasms is consistent with results from the VITAL prospective cohort study, which found that any, compared with no, recreational physical activity was associated with a lower risk of all myeloid neoplasms (RR, 0.59; 95% CI, 0.40–0.88; ref. 11); however, the difference in physical activity measures between the two studies makes direct comparison difficult. Reasons for the lack of a linear dose–response association between MVPA and risk of myeloid neoplasms observed in our study are unclear. It is possible that the older age of participants at baseline (age 69 years on average) could have led to misclassification of leisurely, light intensity walking as moderate intensity, thereby leading to over-reporting of MVPA. Given the relatively small number of cases, it is also possible that this finding is due to chance. Similarly, it is possible that the relationship observed reflects a true non-linear dose–response association between MVPA and myeloid neoplasms.

In our study, the lack of association between MVPA and risk of myeloid leukemia was unexpected in light of findings from the recent large pooled analysis (4) which showed a significant inverse association between leisure-time physical activity and risk of myeloid leukemia (HR, 0.80; 95% CI, 0.70–0.92; 90^th vs. 10^th percentile of leisure-time physical activity). Although the pooled analysis included CPS-II NC data, all analyses compared only the 90^th and the 10^th percentiles of physical activity. The pooled analysis only included myeloid leukemia and therefore did not include MDS (ICD-O: 9987 and 9989 not included), which was the outcome most strongly associated with MVPA in the current study.

We found no association between MVPA and risk of the myeloid leukemia subtype AML. These findings are similar to a case–control study that reported no association between MVPA and the risk of AML (OR, 0.93; 95% CI, 0.54–1.60) or CML (OR, 0.99; 95% CI, 0.46–2.12, 1^st vs. 4^th quartile; ref. 12). Similarly, in the NIH-AARP prospective cohort study, there was no association between vigorous-intensity physical activity (VPA) and the risk of AML (HR, 1.09; 95% CI, 0.84–1.41), but there was a statistically significant inverse association with CML (HR, 0.70; 95% CI, 0.49–0.99, ≥3 times/wk vs. <1 time/wk; refs. 13–15).

Disentangling the respective etiology of myeloid neoplasm subtypes, if they do in fact differ, is difficult. There have been evolving definitions of the subtypes, especially MDS, limiting the identification of possible risk factors (2, 16). It is likely that before 2001, some neoplasms defined as AML may have actually been MDS, complicating comparisons with older studies. Furthermore, AML arises secondary to MDS in about 1/3 of patients (17). It is believed that MDS may be underreported as the diagnosis can be clinically difficult given the various presentations and diagnostic preferences (2). Despite these difficulties, further exploration of the potential role of modifiable behaviors in the prevention of myeloid neoplasms is important, as only 31% of MDS and 27.4% of AML patients survive past five years (2).

The strengths of this study include the prospective design with 14 years of follow-up (1999–2013) and the ability to control for several potential confounders. This is also the first prospective study on the associations between physical activity, sitting time, and MDS. Limitations of this study include the use of self-reported MVPA and sitting time, which may lead to nondifferential misclassification. However, the CPS-II physical activity and sitting time items are very similar to the items validated in the Nurses' Health Study II (18). These measures have also been associated with various cancers in this cohort (19). Another limitation is the relatively small number of CML and AML cases. Given the rarity of myeloid neoplasms, pooled prospective analyses are necessary to confirm the association with MVPA and to further explore the possibility of a non-linear dose–response, especially among myeloid neoplasm subtypes. Pooled analyses would also allow for the exploration of effect modification by sex, which is especially relevant as males have higher age-adjusted incidence rates of MDS (20), AML (https://seer.cancer.gov/statfacts/html/amyl.html), and CML (https://seer.cancer.gov/statfacts/html/clyl.html), and effect modification by BMI, which has been shown to be associated with myeloid malignancies in some studies (21–23). Analyses were further limited as most of the MVPA reported by CPS-II NC participants includes moderate-intensity physical activity, so relationships with VPA alone could not be explored. Similarly, although we did not see an association with sitting time, relationships with risk of myeloid neoplasms should be further explored, as it is not possible to draw conclusions about the associations from this study alone. Finally, the participants in the CPS-II NC are predominately white and highly educated older adults, and therefore may not represent the general U.S. population.

This study suggests that there may be an association between MVPA and the risk of myeloid neoplasms, particularly MDS, a highly fatal and understudied cancer. This study adds to the sparse literature on MVPA, sitting time, and the risk of myeloid neoplasms. As the U.S. population continues to age, the number of people diagnosed with myeloid neoplasms will increase. Therefore, more studies, particularly pooled analyses, are necessary to better understand if and how MVPA may decrease the risk of this group of fatal diseases.

No potential conflicts of interest were disclosed.

The views expressed here are those of the authors and do not necessarily represent the American Cancer Society or the American Cancer Society—Cancer Action Network.

Conception and design: E. Rees-Punia

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): A.V. Patel, S.M. Gapstur

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): E. Rees-Punia, A.V. Patel, E.A. Fallon, L.R. Teras

Writing, review, and/or revision of the manuscript: E. Rees-Punia, A.V. Patel, E.A. Fallon, S.M. Gapstur, L.R. Teras

Study supervision: L.R. Teras

The authors express sincere appreciation to all Cancer Prevention Study-II participants, and to each member of the study and biospecimen management group. The authors would also like to acknowledge the contribution from central cancer registries supported through the Centers for Disease Control and Prevention's National Program of Cancer Registries and cancer registries supported by the National Cancer Institute's Surveillance Epidemiology and End Results Program. The American Cancer Society funds the creation, maintenance, and updating of the Cancer Prevention Study-II cohort. The American Cancer Society funds the creation, maintenance, and updating of the Cancer Prevention Study-II.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.