Background: Despite the existence of numerous biologic pathways potentially linking increased physical activity to decreased risk of hematologic cancers, the associations between physical activity and subtype-specific hematologic cancers have not been comprehensively quantified.

Methods: We conducted a systematic review and meta-analysis of physical activity in relation to subtype-specific hematologic cancers. We summarized the data from 23 eligible studies (15 cohort and eight case–control studies) and estimated summary relative risks (RR) and 95% confidence intervals (CI) using random-effects models.

Results: When comparing high versus low physical activity levels, the RR for non-Hodgkin lymphoma was 0.91 (95% CI, 0.82–1.00), for Hodgkin lymphoma it was 0.86 (95% CI, 0.58–1.26), for leukemia it was 0.97 (95% CI, 0.84–1.13), and for multiple myeloma it was 0.86 (95% CI, 0.68–1.09). When focusing on subtypes of non-Hodgkin lymphoma, the RR for diffuse large B-cell lymphoma was 0.95 (95% CI, 0.80–1.14) and for follicular lymphoma it was 1.01 (95% CI, 0.83–1.22). In an exploratory analysis combining all hematologic cancers, high versus low physical activity levels yielded a statistically significant RR of 0.93 (95% CI, 0.88–0.99).

Conclusions: Physical activity showed statistically nonsignificant associations with risks of non-Hodgkin lymphoma, Hodgkin lymphoma, multiple myeloma, and leukemia. These findings may not represent a true lack of associations given the variation in high versus low physical activity definitions, the quality of physical activity assessments, and the variability in hematologic cancer classification schemes in individual studies.

Impact: Physical activity is unrelated to risks of subtype-specific hematologic cancers. Cancer Epidemiol Biomarkers Prev; 23(5); 833–46. ©2014 AACR.

Hematologic cancers represent a heterogeneous group of malignant neoplasms of the hematopoietic and lymphoid tissues and they comprise non-Hodgkin lymphoma, Hodgkin lymphoma, leukemia, and multiple myeloma. In 2008, 6.9% of cancer incidence and 7.3% of cancer mortality worldwide were due to hematologic cancers (1). Non-Hodgkin lymphoma and leukemia rank among the top 10 cancer sites globally, and in developed countries, non-Hodgkin lymphoma is the seventh most frequently diagnosed cancer (1). Among non-Hodgkin lymphoma, diffuse large B-cell lymphoma is the most common subtype.

Obesity has been suggested to be a risk factor for non-Hodgkin lymphoma, Hodgkin lymphoma, leukemia, and multiple myeloma (2–4). Similarly, a deficient or suppressed immune system is a recognized risk factor common to non-Hodgkin lymphoma, Hodgkin lymphoma, multiple myeloma, and leukemia (5–8). Physical activity is inversely linked to adiposity (9) and is positively associated with immune function (10) and thus, may play a role in the prevention of hematologic cancer. Physical activity may also prevent the development of hematologic cancer independently of body weight control through enhancements of insulin sensitivity, anti-inflammatory mechanisms, and antioxidant defense systems (11–16).

Despite the existence of numerous biologic pathways potentially linking increased physical activity to decreased risk of hematologic cancer, the epidemiologic evidence relating physical activity to hematologic cancer has been mixed, and the results of a recent meta-analysis on physical activity and lymphoma were not statistically significant (17). Thus, the relation between physical activity and hematologic cancer remains inconclusive.

The purpose of the current systematic review and meta-analysis was to quantify the association between physical activity and hematologic cancer subtypes. To our knowledge, the current investigation is the first meta-analysis to include a comprehensive investigation of physical activity in relation to hematologic cancer subtypes.

Systematic search strategy and study selection

Studies were identified through a systematic search of PubMed from inception to June 2013 using the following search terms for physical activity: physical activity, exercise, cardiorespiratory fitness, cardiovascular fitness, resistance training, endurance training, aerobic, sport, athletes, players, lifestyle, and anthropometric. Those terms were combined with the following (truncated) terms for hematologic cancers using an AND operator: hematologic malignancy, hematologic neoplasm, hematopoietic malignancy, hematopoietic neoplasm, lymphoid, leukemia, lymphoma, multiple myeloma, and cancer. Cancer outcomes other than hematologic cancers were excluded. The search was limited to studies in humans and no language restrictions were imposed.

Using that search, 949 citations were identified (Supplementary Fig. S1). After screening titles and abstracts, 911 citations were excluded; the remaining 38 full articles were given a more detailed assessment. Studies that met the following criteria were considered further: observational cohort or case–control studies of physical activity and hematologic cancers that provided an effect estimate [relative risk (RR), HR, OR, or standardized incidence ratio (SIR)], with 95% confidence intervals (CI) or sufficient information to calculate those values. Fifteen citations were excluded after full-text screening and two additional articles were found by a manual search. Of the 25 studies (18–42) on physical activity and hematologic cancers, two studies (41, 42) were removed because they did not provide a risk estimate for a hematologic cancer subtype. The remaining 23 studies (18–40) were included in the meta-analysis.

Data extraction

For each study, the following data were extracted by the principal reviewer (C. Jochem): first author's last name, year of publication, sample size, number of cases, study geographic region, type of physical activity assessment, domains and intensities of physical activity, timing in life of physical activity, definition of the highest and the lowest categories of physical activity, reported physical activity effect estimates with corresponding 95% CI, adjustment variables, and information needed to assess the methodological quality of each study. All data extraction was double-checked by a second reviewer (G. Behrens).

Quality assessment

Quality assessment was independently conducted by all reviewers and any disagreement was resolved by consensus. A quality score developed by Monninkhof and colleagues (43) and previously used in several meta-analyses of physical activity and specific cancers (17, 43–46) was applied to assess the methodological quality of the selected studies (supplementary Materials and Methods).

Statistical analysis

Effect estimates were interpreted as estimates of the RRi. The natural logarithms of those risk estimates [log(RRi)] with their corresponding standard errors si = [log(upper 95% CI bound or RR) − log(RR)]/1.96 were calculated using a random-effects model to determine the weighted average of those log(RRi)s while allowing for effect measure heterogeneity. The log(RRi)s were weighted by wi = 1/(si2 + t2), where si denoted the standard error of log(RRi) and t2 denoted the restricted maximum likelihood estimate of the overall variance (47).

The main meta-analysis included one risk estimate per hematologic cancer subtype (non-Hodgkin lymphoma, diffuse large B-cell lymphoma, follicular lymphoma, Hodgkin lymphoma, leukemia, and multiple myeloma), per gender, and per physical activity domain for each study. If risk estimates were available separately for each sex as well as for men and women combined, the latter were not included in the main meta-analysis. If risk estimates were provided for different numbers of adjustment factors, the maximally adjusted risk estimate was used. If a study presented several risk estimates for physical activity at various time periods in life, the risk estimate for the most recent time period was chosen (21, 22, 34) as most studies provided risk estimates for recent physical activity.

The 23 underlying studies used varying definitions of hematologic cancer subtypes, sometimes without providing information about the specific classification scheme used. In our meta-analysis, the hematologic cancer groups of non-Hodgkin lymphoma, Hodgkin lymphoma, diffuse large B-cell lymphoma, follicular lymphoma, chronic lymphocytic lymphoma/small lymphocytic lymphoma, and multiple myeloma included all risk estimates that were available for those specific subtypes. Our group of non-Hodgkin lymphoma did not include histologic subtypes of non-Hodgkin lymphoma. No study provided information on the specific subtype of Hodgkin lymphoma. Our group of leukemia included risk estimates that were either classified as all leukemia or specific leukemia subtypes, such as acute or chronic myeloid leukemia. Our groupings are consistent to the extent possible across studies included in the meta-analyses.

Publication bias was assessed using a funnel plot and Begg's rank correlation test (48). Statistical heterogeneity between risk estimates was estimated using the Q- and I2 statistics (47).

For each subsite-specific hematologic cancer, a priori determined subanalyses were performed that were stratified by study design, gender, study quality score, physical activity domain, type of physical activity assessment, timing in life of physical activity, number of adjustment factors, adjustment for smoking, adjustment for alcohol consumption, adjustment for adiposity, and study geographic region. Potential heterogeneity of the physical activity and hematologic cancer subtype associations was assessed according to those factors using random-effects meta-regression, where the model that included the current factor of interest as a single explanatory variable was compared with the null model that included no explanatory variable.

An exploratory dose–response meta-analysis was conducted using fractional polynomials based on a modified version (Supplementary Materials and Methods) of the method proposed by Rota and colleagues (49). For that analysis, we considered seven studies (19, 20, 22, 23, 29, 39, 40) that investigated recreational physical activity expressed as metabolic equivalent task (MET)-hours or MET-minutes per week in relation to hematologic cancer subtypes. The midpoint of each physical activity category was used as the dose associated with the RR for that category. Because the highest categories of physical activity in the underlying studies were open-ended, the dose corresponding to the highest category was defined as 1.5 times the value of the lower bound of that category. The reference level (lowest category) was set to 0 MET-hours. On the basis of the gender-specific RR estimates from the seven included studies, a total of 27 individual dose–response relationships were summarized using multivariate random-effects meta-analysis.

All statistical analyses were performed with R version 2.12.2 (50) using the packages “metafor” (51) and “mvmeta” (52). All risk estimates are reported with 95% CI and P < 0.05 were considered statistically significant.

The current meta-analysis comprised a total of 1,648,601 subjects and 19,334 hematologic cancer cases. Table 1 presents the main characteristics and results of the 15 cohort studies (26–40) and eight case–control studies (18–25) included. Three studies investigated more than one physical activity domain (19, 24, 40) and nine studies provided results stratified by gender (20, 22, 23, 25, 26, 29, 32, 39, 40).

Table 1.

Characteristics of the 15 cohort studies and eight case–control studies of physical activity and hematologic cancer risk included in the meta-analysis

Authors, year, genderStudy regionSubjectsCasesHematologic cancer subtypePA: domain, timing in lifeRR (95% CI), high vs. low PALow PA defined asHigh PA defined asAdjustment factors (excluding sex)QS (%)
Cohort studies (N = 15) 
 Birmann et al., 2007 
  Men North America 46,960 86 MM RPA, consistent 0.80 (0.50–1.50) <2 h/wk PA ≥7 h/wk PA Updated age, BMI 73 
  Women  89,663 129   0.50 (0.20–1.40)    73 
 Blair et al., 2005 
  Women North America 37,083 95 MM RPA, recent 0.88 (0.51–1.49) Low PA High PA — 48 
 Cerhan et al., 2002 Women North America 37,932 252 NHL RPA, past 0.83 (0.59–1.11) Low PA High PA Age 60 
   61 Leukemia  0.91 (0.45–1.67)    60 
 Hofmann et al., 2013 
  Men North America 291,471 319 MM RPA, recent 1.10 (0.82–1.48) <16.25 MET-h/wk ≥50 MET-h/wk Age (at baseline), race, age-specific BMI 65 
  Women  193,578 152 MM  1.43 (0.90–2.26)    65 
 Kabat et al., 2012 
  Women North America 157,852 285 CLL/SLL RPA, recent 1.03 (0.70–1.51) 0 h/wk strenuous PA ≥2 h/wk strenuous PA Age, smoking, servings of alcohol per week, education, ethnicity, BMI enrollment in the observational study, treatment arm assignment in the clinical trials 82 
   1,071 NHL  0.97 (0.79–1.20)    82 
   286 DLBCL  0.89 (0.59–1.36)    82 
   205 FL  1.25 (0.81–1.94)    82 
 Kabat et al., 2013 
  Men and women combined North America 493,188 178 Leukemia TPA, recent 0.70 (0.49–0.99) <1 time/wk VPA ≥3 times/wk VPA Age, BMI, smoking intensity, years of education 62 
 Khan et al., 2006 
  Men Asia 46,157 35 MM RPA, recent 0.44 (0.21–0.93) <30 min walking/d ≥1 hour walking/day Age 64 
  Women  63,541 31 MM  0.58 (0.26–1.27)    64 
 Lim et al., 2007 
  Men and women combined North America 465,858 234 CLL/SLL TPA, recent 0.80 (0.51–1.25) PA <1 time/wk PA ≥5 times/wk Age, race, education, BMI, caloric intake 65 
   1,340 NHL  0.97 (0.81–1.16)    65 
   343 DLBCL  0.87 (0.61–1.25)    65 
   257 FL  0.96 (0.63–1.46)    65 
   56 HL  1.60 (0.72–3.54) PA <1 time/wk PA 3–4 times/wk  65 
 Lu et al., 2009 
  Women North America 121,216 124 CLL/SLL RPA, recent 1.50 (0.86–2.63) 0–0.50 h/wk/y strenuous PA plus MPA ≥4 h/wk/y strenuous PA plus MPA Age, height, weight at cohort entry, age at menarche long-term strenuous PA plus MPA 61 
   574 NHL  1.11 (0.86–1.44)    61 
   155 DLBCL  1.00 (0.62–1.62)    61 
   121 FL  1.01 (0.57–1.79)    61 
 Ma et al., 2010 
  Men and women combined North America 491,163 338 Leukemia RPA, recent 1.09 (0.84–1.41) <3 times/mo VPA ≥3 times/wk VPA Age, smoking status, total energy intake 69 
 Paffenbarger et al., 1992 
  Men and women combined North America 56,683 86 NHL RPA, past 0.67 (0.40–1.13) <5 hours of VPA per week ≥5 hours of VPA per week Age 57 
   52 HL  0.73 (0.38–1.39)    57 
   81 Leukemia  0.84 (0.45–1.58)    57 
 Pukkala et al., 2000 
  Men Europe 2,269 NHL OPA, past 0.78 (0.29–1.70) General population Inclusion in athletic group 5-year age groups 57 
   HL  1.33 (0.27–3.88)    57 
   Leukemia  0.51 (0.17–1.19)    57 
   MM  1.39 (0.56–2.85)    57 
 Soll-Johanning et al., 2004 
  Men Europe 14,568 44 Leukemia OPA, past 1.08 (0.78–1.45) Copenhagen population Mail carrier Age, calendar period 45 
 Teras et al., 2012 
  Men North America 69,849 271 CLL/SLL RPA, recent 0.95 (0.62–1.46) ≤0 MET-h/wk ≥17.5 MET-h/wk Age at baseline, family history of hematopoietic cancer, education, smoking status, alcohol intake, BMI, height, sitting time 65 
   1,139 NHL  1.02 (0.82–1.26)    65 
   238 DLBCL  1.14 (0.70–1.88)    65 
   143 FL  0.86 (0.49–1.51)    65 
   211 MM  0.99 (0.60–1.64)    65 
  Women  77,001 211 CLL/SLL  0.73 (0.44–1.20)    65 
   863 NHL  0.69 (0.54–0.89)    65 
   197 DLBCL  0.71 (0.44–1.16)    65 
   137 FL  0.77 (0.41–1.43)    65 
   132 MM  0.54 (0.28–1.04)    65 
 van Veldhoven et al., 2011 
  Men Europe 127,353 373 NHL OPA, recent 1.05 (0.46–2.41) Sedentary activities Manual/heavy manual activities Age, center, hypertension, hyperlipidemia, education, diabetes 67 
    NHL RPA, recent 1.11 (0.49–2.52) <14.25 MET-h/wk ≥45.75 MET-h/wk  71 
   45 DLBCL TPA, recent 2.03 (0.38–10.79) Inactive Active  75 
   45 FL  2.82 (0.52–15.23)    75 
   77 MM  0.34 (0.05–2.14)    75 
  Women  216,403 402 NHL RPA, recent 1.25 (0.49–3.16) <14.25 MET-h/wk ≥45.75 MET-h/wk  71 
    NHL OPA, recent 2.06 (0.65–6.51) Sedentary activities Manual/heavy manual activities  67 
   49 DLBCL TPA, recent 5.03 (1.19–21.33) Inactive Active  75 
   49 FL  2.59 (0.54–12.38)    75 
   88 MM  0.65 (0.14–3.16)    75 
Case–control studies (N = 8) 
 Brownson et al., 1991 
  Men North America 17,147 462 NHL OPA, recent 0.80 (0.60–1.30) PA required <20% of the time PA required >80% of the time Age, smoking 45 
   376 Leukemia  0.90 (0.60–1.40)    45 
 Cerhan et al., 2005 
  Men and women combined North America 864 458 NHL RPA, recent 0.83 (0.50–1.38) <30 METs/wk >1,080 METs/wk Age, race, study center, alcohol use, height, BMI 58 
   458 NHL OPA, recent 0.98 (0.55–1.74) Mostly sitting Mostly exercise  49 
 Kasim et al., 2009 
  Men North America 2,032 390 Leukemia RPA, recent 1.31 (0.93–1.85) <5.9 MET-h/wk ≥38.8 MET-h/wk 10-year age group, BMI, educational years, total fruits and vegetables consumption, occupational exposure to benzene and ionizing radiation 58 
  Women  1,727 263 Leukemia  0.97 (0.66–1.42) <5.6 MET-h/wk ≥38.8 MET-h/wk  58 
 Keegan et al., 2006 
  Women North America 441 187 HL RPA, recent 0.56 (0.37–0.86) Nonparticipation in ≥2 times/wk strenuous PA for ≥1 mo ≥2 times/wk strenuous PA for ≥1 mo Age, race/ethnicity, education, Jewish upbringing, having a single room at age 11 years, living in a single family home at age 8 years, living in a rented family home at age 8 years, having ever nursed children, number of miscarriages, smoking status 1 year before diagnosis/interview, history of first or second degree relative with lymphoma 63 
 Kelly et al., 2012 
  Men North America 758 184 CLL/SLL RPA, recent 1.71 (1.08–2.70) <615 METs/wk ≥2,701 METs/wk Age, country of residence 58 
   516 NHL  1.37 (0.98–1.92)    58 
   88 DLBCL  1.19 (0.64–2.21)    58 
   113 FL  0.96 (0.53–1.72)    58 
  Women  591 94 CLL/SLL  0.53 (0.27–1.06)    58 
   358 NHL  0.71 (0.47–1.07)    58 
   79 DLBCL  0.80 (0.38–1.69)    58 
   109 FL  0.90 (0.49–1.67)    58 
 Pan et al., 2005 
  Men North America 2,211 569 NHL RPA, recent 0.79 (0.59–1.05) <6.4 MET-h/wk ≥37.4 MET-h/wk Age, province, education, smoking, alcohol drinking, exposure to certain chemicals, occupational exposure, BMI, total caloric intake, MPA 67 
  Women  1,925 461 NHL  0.59 (0.42–0.81) <6.1 MET-h/wk ≥31.4 MET-h/wk  67 
 Parent et al., 2011 
  Men North America 748 215 NHL RPA, past 1.17 (0.81–1.69) Sports or outdoor activities <1 time/wk Sports or outdoor activities ≥1 time/wk Age, SES, educational level, ethnicity, respondent status, smoking, BMI, sports, outdoor activities 75 
   54 HL  0.97 (0.45–2.08)    75 
   215 NHL OPA, consistent 0.56 (0.26–1.20) ≤1.5 MET (at least 75% of work years in sedentary jobs) ≥4 METs (at least 75% of work years was spent in very active jobs)  66 
   54 HL  0.84 (0.17–4.23)    66 
 Zahm et al., 1999 
  Men North America 3,856 985 NHL OPA, recent 1.00 (0.70–1.30) Energy expenditure <8 kJ/min Energy expenditure >12 kJ/min Age, state of residence 51 
  Women  854 180 NHL  1.70 (0.20–11.50) Energy expenditure <5 kJ/min Energy expenditure >8 kJ/min  51 
Authors, year, genderStudy regionSubjectsCasesHematologic cancer subtypePA: domain, timing in lifeRR (95% CI), high vs. low PALow PA defined asHigh PA defined asAdjustment factors (excluding sex)QS (%)
Cohort studies (N = 15) 
 Birmann et al., 2007 
  Men North America 46,960 86 MM RPA, consistent 0.80 (0.50–1.50) <2 h/wk PA ≥7 h/wk PA Updated age, BMI 73 
  Women  89,663 129   0.50 (0.20–1.40)    73 
 Blair et al., 2005 
  Women North America 37,083 95 MM RPA, recent 0.88 (0.51–1.49) Low PA High PA — 48 
 Cerhan et al., 2002 Women North America 37,932 252 NHL RPA, past 0.83 (0.59–1.11) Low PA High PA Age 60 
   61 Leukemia  0.91 (0.45–1.67)    60 
 Hofmann et al., 2013 
  Men North America 291,471 319 MM RPA, recent 1.10 (0.82–1.48) <16.25 MET-h/wk ≥50 MET-h/wk Age (at baseline), race, age-specific BMI 65 
  Women  193,578 152 MM  1.43 (0.90–2.26)    65 
 Kabat et al., 2012 
  Women North America 157,852 285 CLL/SLL RPA, recent 1.03 (0.70–1.51) 0 h/wk strenuous PA ≥2 h/wk strenuous PA Age, smoking, servings of alcohol per week, education, ethnicity, BMI enrollment in the observational study, treatment arm assignment in the clinical trials 82 
   1,071 NHL  0.97 (0.79–1.20)    82 
   286 DLBCL  0.89 (0.59–1.36)    82 
   205 FL  1.25 (0.81–1.94)    82 
 Kabat et al., 2013 
  Men and women combined North America 493,188 178 Leukemia TPA, recent 0.70 (0.49–0.99) <1 time/wk VPA ≥3 times/wk VPA Age, BMI, smoking intensity, years of education 62 
 Khan et al., 2006 
  Men Asia 46,157 35 MM RPA, recent 0.44 (0.21–0.93) <30 min walking/d ≥1 hour walking/day Age 64 
  Women  63,541 31 MM  0.58 (0.26–1.27)    64 
 Lim et al., 2007 
  Men and women combined North America 465,858 234 CLL/SLL TPA, recent 0.80 (0.51–1.25) PA <1 time/wk PA ≥5 times/wk Age, race, education, BMI, caloric intake 65 
   1,340 NHL  0.97 (0.81–1.16)    65 
   343 DLBCL  0.87 (0.61–1.25)    65 
   257 FL  0.96 (0.63–1.46)    65 
   56 HL  1.60 (0.72–3.54) PA <1 time/wk PA 3–4 times/wk  65 
 Lu et al., 2009 
  Women North America 121,216 124 CLL/SLL RPA, recent 1.50 (0.86–2.63) 0–0.50 h/wk/y strenuous PA plus MPA ≥4 h/wk/y strenuous PA plus MPA Age, height, weight at cohort entry, age at menarche long-term strenuous PA plus MPA 61 
   574 NHL  1.11 (0.86–1.44)    61 
   155 DLBCL  1.00 (0.62–1.62)    61 
   121 FL  1.01 (0.57–1.79)    61 
 Ma et al., 2010 
  Men and women combined North America 491,163 338 Leukemia RPA, recent 1.09 (0.84–1.41) <3 times/mo VPA ≥3 times/wk VPA Age, smoking status, total energy intake 69 
 Paffenbarger et al., 1992 
  Men and women combined North America 56,683 86 NHL RPA, past 0.67 (0.40–1.13) <5 hours of VPA per week ≥5 hours of VPA per week Age 57 
   52 HL  0.73 (0.38–1.39)    57 
   81 Leukemia  0.84 (0.45–1.58)    57 
 Pukkala et al., 2000 
  Men Europe 2,269 NHL OPA, past 0.78 (0.29–1.70) General population Inclusion in athletic group 5-year age groups 57 
   HL  1.33 (0.27–3.88)    57 
   Leukemia  0.51 (0.17–1.19)    57 
   MM  1.39 (0.56–2.85)    57 
 Soll-Johanning et al., 2004 
  Men Europe 14,568 44 Leukemia OPA, past 1.08 (0.78–1.45) Copenhagen population Mail carrier Age, calendar period 45 
 Teras et al., 2012 
  Men North America 69,849 271 CLL/SLL RPA, recent 0.95 (0.62–1.46) ≤0 MET-h/wk ≥17.5 MET-h/wk Age at baseline, family history of hematopoietic cancer, education, smoking status, alcohol intake, BMI, height, sitting time 65 
   1,139 NHL  1.02 (0.82–1.26)    65 
   238 DLBCL  1.14 (0.70–1.88)    65 
   143 FL  0.86 (0.49–1.51)    65 
   211 MM  0.99 (0.60–1.64)    65 
  Women  77,001 211 CLL/SLL  0.73 (0.44–1.20)    65 
   863 NHL  0.69 (0.54–0.89)    65 
   197 DLBCL  0.71 (0.44–1.16)    65 
   137 FL  0.77 (0.41–1.43)    65 
   132 MM  0.54 (0.28–1.04)    65 
 van Veldhoven et al., 2011 
  Men Europe 127,353 373 NHL OPA, recent 1.05 (0.46–2.41) Sedentary activities Manual/heavy manual activities Age, center, hypertension, hyperlipidemia, education, diabetes 67 
    NHL RPA, recent 1.11 (0.49–2.52) <14.25 MET-h/wk ≥45.75 MET-h/wk  71 
   45 DLBCL TPA, recent 2.03 (0.38–10.79) Inactive Active  75 
   45 FL  2.82 (0.52–15.23)    75 
   77 MM  0.34 (0.05–2.14)    75 
  Women  216,403 402 NHL RPA, recent 1.25 (0.49–3.16) <14.25 MET-h/wk ≥45.75 MET-h/wk  71 
    NHL OPA, recent 2.06 (0.65–6.51) Sedentary activities Manual/heavy manual activities  67 
   49 DLBCL TPA, recent 5.03 (1.19–21.33) Inactive Active  75 
   49 FL  2.59 (0.54–12.38)    75 
   88 MM  0.65 (0.14–3.16)    75 
Case–control studies (N = 8) 
 Brownson et al., 1991 
  Men North America 17,147 462 NHL OPA, recent 0.80 (0.60–1.30) PA required <20% of the time PA required >80% of the time Age, smoking 45 
   376 Leukemia  0.90 (0.60–1.40)    45 
 Cerhan et al., 2005 
  Men and women combined North America 864 458 NHL RPA, recent 0.83 (0.50–1.38) <30 METs/wk >1,080 METs/wk Age, race, study center, alcohol use, height, BMI 58 
   458 NHL OPA, recent 0.98 (0.55–1.74) Mostly sitting Mostly exercise  49 
 Kasim et al., 2009 
  Men North America 2,032 390 Leukemia RPA, recent 1.31 (0.93–1.85) <5.9 MET-h/wk ≥38.8 MET-h/wk 10-year age group, BMI, educational years, total fruits and vegetables consumption, occupational exposure to benzene and ionizing radiation 58 
  Women  1,727 263 Leukemia  0.97 (0.66–1.42) <5.6 MET-h/wk ≥38.8 MET-h/wk  58 
 Keegan et al., 2006 
  Women North America 441 187 HL RPA, recent 0.56 (0.37–0.86) Nonparticipation in ≥2 times/wk strenuous PA for ≥1 mo ≥2 times/wk strenuous PA for ≥1 mo Age, race/ethnicity, education, Jewish upbringing, having a single room at age 11 years, living in a single family home at age 8 years, living in a rented family home at age 8 years, having ever nursed children, number of miscarriages, smoking status 1 year before diagnosis/interview, history of first or second degree relative with lymphoma 63 
 Kelly et al., 2012 
  Men North America 758 184 CLL/SLL RPA, recent 1.71 (1.08–2.70) <615 METs/wk ≥2,701 METs/wk Age, country of residence 58 
   516 NHL  1.37 (0.98–1.92)    58 
   88 DLBCL  1.19 (0.64–2.21)    58 
   113 FL  0.96 (0.53–1.72)    58 
  Women  591 94 CLL/SLL  0.53 (0.27–1.06)    58 
   358 NHL  0.71 (0.47–1.07)    58 
   79 DLBCL  0.80 (0.38–1.69)    58 
   109 FL  0.90 (0.49–1.67)    58 
 Pan et al., 2005 
  Men North America 2,211 569 NHL RPA, recent 0.79 (0.59–1.05) <6.4 MET-h/wk ≥37.4 MET-h/wk Age, province, education, smoking, alcohol drinking, exposure to certain chemicals, occupational exposure, BMI, total caloric intake, MPA 67 
  Women  1,925 461 NHL  0.59 (0.42–0.81) <6.1 MET-h/wk ≥31.4 MET-h/wk  67 
 Parent et al., 2011 
  Men North America 748 215 NHL RPA, past 1.17 (0.81–1.69) Sports or outdoor activities <1 time/wk Sports or outdoor activities ≥1 time/wk Age, SES, educational level, ethnicity, respondent status, smoking, BMI, sports, outdoor activities 75 
   54 HL  0.97 (0.45–2.08)    75 
   215 NHL OPA, consistent 0.56 (0.26–1.20) ≤1.5 MET (at least 75% of work years in sedentary jobs) ≥4 METs (at least 75% of work years was spent in very active jobs)  66 
   54 HL  0.84 (0.17–4.23)    66 
 Zahm et al., 1999 
  Men North America 3,856 985 NHL OPA, recent 1.00 (0.70–1.30) Energy expenditure <8 kJ/min Energy expenditure >12 kJ/min Age, state of residence 51 
  Women  854 180 NHL  1.70 (0.20–11.50) Energy expenditure <5 kJ/min Energy expenditure >8 kJ/min  51 

Abbreviations: CLL/SLL, chronic lymphocytic lymphoma/small lymphocytic lymphoma; DLBCL, diffuse large B-cell lymphoma; FL, follicular lymphoma; HL, Hodgkin lymphoma; MM, multiple myeloma; MPA, moderate physical activity; NHL, non-Hodgkin lymphoma; OPA, occupational physical activity; PA, physical activity; QS, quality score; RPA, recreational physical activity; SES, socioeconomic status; TPA, total physical activity; VPA, vigorous physical activity.

Comparing the highest versus the lowest level of physical activity, the 23 risk estimates for non-Hodgkin lymphoma were consistent with a borderline statistically significant inverse association (RR, 0.91; 95% CI, 0.82–1.00; I2 = 39%; Fig. 1). When focusing on subtypes of non-Hodgkin lymphoma, the nine risk estimates for diffuse large B-cell lymphoma yielded a statistically nonsignificant relation with high versus low physical activity (RR, 0.95; 95% CI, 0.80–1.14; I2 = 0%; Fig. 2). Similarly, the summary risk estimate of the nine available risk estimates for physical activity and follicular lymphoma was null at 1.01 (95% CI, 0.83–1.22; I2 = 2%; Fig. 2). The six risk estimates available for Hodgkin lymphoma yielded a statistically nonsignificant risk reduction with high versus low physical activity levels (RR, 0.86; 95% CI, 0.58–1.26; I2 = 35%; Fig. 1). An analysis of all risk estimates available for non-Hodgkin lymphoma and Hodgkin lymphoma combined (N = 29) showed a statistically significant reduction in lymphoma risk with high versus low physical activity level (RR, 0.90; 95% CI, 0.81–0.99; Fig. 1).

Figure 1.

Forest plot of a random-effects meta-analysis including risk estimates of non-Hodgkin lymphoma and Hodgkin lymphoma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for non-Hodgkin lymphoma and Hodgkin lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Figure 1.

Forest plot of a random-effects meta-analysis including risk estimates of non-Hodgkin lymphoma and Hodgkin lymphoma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for non-Hodgkin lymphoma and Hodgkin lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Close modal
Figure 2.

Forest plot of a random-effects meta-analysis including risk estimates of diffuse large B-cell lymphoma and follicular lymphoma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for diffuse large B-cell lymphoma and follicular lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Figure 2.

Forest plot of a random-effects meta-analysis including risk estimates of diffuse large B-cell lymphoma and follicular lymphoma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for diffuse large B-cell lymphoma and follicular lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Close modal

Nine risk estimates were available for high versus low physical activity and leukemia and resulted in a summary risk estimate of 0.97 (95% CI, 0.84–1.13; I2 = 23%; Fig. 3). Similarly, seven risk estimates for physical activity and chronic lymphocytic lymphoma/small lymphocytic lymphoma yielded a null summary risk estimate of 0.99 (95% CI, 0.75–1.29; I2 = 54%; Fig. 3). By comparison, the 12 available risk estimates of high versus low physical activity and multiple myeloma showed a statistically nonsignificant inverse relation (RR, 0.86; 95% CI, 0.68–1.09; I2 = 38%; Fig. 3).

Figure 3.

Forest plot of a random-effects meta-analysis including risk estimates of leukemia, chronic lymphocytic lymphoma/small lymphocytic lymphoma, and multiple myeloma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for leukemia and chronic lymphocytic lymphoma/small lymphocytic lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Figure 3.

Forest plot of a random-effects meta-analysis including risk estimates of leukemia, chronic lymphocytic lymphoma/small lymphocytic lymphoma, and multiple myeloma. RRs quantify the association between physical activity (PA) and hematologic cancer subtype risk for the highest versus lowest levels of physical activity, grouped by men, women, and men and women combined for leukemia and chronic lymphocytic lymphoma/small lymphocytic lymphoma. The size of each box is proportional to the weight the risk estimate contributed to the summary risk estimate. OPA, occupational physical activity; RPA, recreational physical activity; TPA, total physical activity.

Close modal

We next investigated potential effect modification of the associations between physical activity and hematologic cancer subtypes (Supplementary Tables S1A–G). No statistically significant effect modification was observed for any of the variables studied for any of the hematologic cancer subtypes, apart from apparent effect modification of the physical activity and diffuse large B-cell lymphoma relation by study geographic region. Specifically, physical activity was positively associated with diffuse large B-cell lymphoma in studies from Europe (RR, 3.41; 95% CI, 1.14–10.18), whereas it was unrelated to diffuse large B-cell lymphoma in studies from North America (RR, 0.92; 95% CI, 0.77–1.10; Pdifference = 0.02).

In a subanalysis, we combined 75 risk estimates and noted a statistically significant 7% reduced risk of total hematologic cancer when comparing the highest versus the lowest level of physical activity (RR, 0.93; 95% CI, 0.88–0.99; Supplementary Table S2). No evidence for publication bias was found in a funnel plot (Supplementary Fig. S2). Begg's rank correlation test also indicated no publication bias (P = 0.82). In contrast, the Q-statistic indicated heterogeneity of study results (P = 0.04; I2 = 24%).

In further subanalyses, we found that the summary risk estimate for total hematologic cancer did not differ by study design (Pdifference = 0.72). In contrast, the relation of physical activity to total hematologic cancer varied according to gender, showing an inverse relation with physical activity in women (RR, 0.88; 95% CI, 0.79–0.98), but not in men (RR, 1.03; 95% CI, 0.94–1.01; Pdifference = 0.03). Furthermore, studies with a large number of adjustment factors yielded a more pronounced inverse relation of physical activity to total hematologic cancer (RR, 0.85; 95% CI, 0.76–0.95) than studies with an intermediate (RR, 1.03; 95% CI, 0.94–1.12) or low number of adjustment factors (RR, 0.92; 95% CI, 0.82–1.04; Pdifference = 0.04). Also, studies that adjusted for smoking showed a stronger inverse association between physical activity and total hematologic cancer (RR, 0.86; 95% CI, 0.78–0.95) than studies that did not adjust for smoking (RR, 1.00; 95% CI, 0.93–1.07; Pdifference = 0.02).

We conducted an exploratory dose–response meta-analysis of recreational physical activity expressed as energy-expenditure in MET-hours/week in relation to total hematologic cancer risk. The following dose–response curve was identified as the one with the best model fit: RR = exp(a1 dose2 + a2 dose3), where a1 = −0.000271, a2 = −0.000004, var(a1) = 1 × 10−8, cov(a1, a2) = −2 × 10−10, and var(a2) = 3 × 10−12 (Fig. 4). That model did not indicate heterogeneity between studies (I2 = 9%; Pheterogeneity = 0.29). It showed an inverse association between physical activity and total hematologic cancer (Fig. 4), suggesting that a risk reduction of 10% was reached at 26 MET-hours/week (RR, 0.90; 95% CI, 0.83–0.97). The maximum risk reduction was detected at 41 MET-hours/week (RR, 0.86; 95% CI, 0.76–0.97). A linear dose–response meta-analysis including all physical activity doses up to 41 MET-hours indicated that each 12 MET-hour/week increase in recreational physical activity was associated with a 5% reduced total hematologic cancer risk (RR, 0.95; 95% CI, 0.91–0.99). Energy expenditure values exceeding 48 MET-hours/week were no longer related to a statistically significant risk reduction.

Figure 4.

Dose–response analysis of MET-hours/week spent in moderate to vigorous physical activity in relation to total hematologic cancer risk.

Figure 4.

Dose–response analysis of MET-hours/week spent in moderate to vigorous physical activity in relation to total hematologic cancer risk.

Close modal

This is the first meta-analysis to comprehensively examine physical activity in relation to hematologic cancer subtypes. We found that high versus low physical activity showed statistically nonsignificant relations to non-Hodgkin lymphoma, Hodgkin lymphoma, multiple myeloma, and leukemia. The relations of physical activity to hematologic cancer subtypes were not modified by study design, gender, study quality score, physical activity domain, type of physical activity assessment, timing in life of physical activity, number of adjustment factors, adjustment for smoking, adjustment for alcohol consumption, and adjustment for adiposity.

A recent meta-analysis (17) that focused on non-Hodgkin lymphoma and Hodgkin lymphoma pooled the data from 12 studies (16 risk estimates for non-Hodgkin lymphoma; four risk estimates for Hodgkin lymphoma) and reported a statistically nonsignificant reduction in lymphoma risk with high versus low levels of physical activity (RR, 0.90; 95% CI, 0.79–1.02), with observed heterogeneity by study design (Pdifference = 0.04). Our meta-analysis included nine additional risk estimates for lymphoma and yielded an identical point estimate that was statistically significant (RR, 0.90; 95% CI, 0.81–0.99) and showed no effect modification by study design (Pdifference = 0.72).

In contrast to the previous meta-analysis (17), the present meta-analysis considered all main hematologic cancer subtypes (non-Hodgkin lymphoma, Hodgkin lymphoma, multiple myeloma, and leukemia). In addition, it incorporated one additional study by Paffenbarger and colleagues (36) that provided risk estimates for physical activity and both non-Hodgkin lymphoma and Hodgkin lymphoma, in addition to leukemia. The current meta-analysis also examined the potential for publication bias.

Our exploratory dose–response meta-analysis indicated that a 5% risk reduction of total hematologic cancer could be achieved for each 12 MET-hour/week increase in recreational physical activity (equivalent to walking at a moderate pace for 30 minutes daily; ref. 53). Our observation of an upward trend of the dose–response curve for values exceeding 41 MET-hours/week indicates that there may be an upper level of physical activity that no longer provides protection against total hematologic cancer risk. In this context, it is interesting to note that extremely high levels of physical activity are thought to impair the immune system (54). Particular caution is needed in interpreting our findings for total hematologic cancer because of etiologic heterogeneity for individual hematologic cancer subtypes.

Two case–control studies on physical activity and hematologic cancer (20, 23) presented results stratified by body mass index (BMI). In one study (20), physical activity showed a more pronounced inverse association with adult leukemia among obese than overweight or normal weight men and women, although the test for interaction was not statistically significant. In the other study (23), an inverse relation of physical activity to non-Hodgkin lymphoma was seen only among obese or normal weight men, whereas no association with physical activity was noted among overweight men. In contrast, physical activity displayed inverse associations with non-Hodgkin lymphoma among nonobese women, whereas the relation of physical activity to non-Hodgkin lymphoma among obese women was null (23). Those inconsistent gradients in hematologic cancer risk associated with physical activity across categories of BMI suggest that physical activity and adiposity affect hematologic cancer risk through distinct etiologic mechanisms. In support of the notion that the inverse association between physical activity and hematologic cancer is not mediated by healthy body mass among physically active persons, in our meta-analysis the magnitude of the summary risk estimates of physical activity and hematologic cancer subtypes that were adjusted for BMI were similar to the risk estimates that were unadjusted for BMI.

In our study, the associations between physical activity and hematologic cancer subtypes were not modified by gender. However, in a subanalysis of total hematologic cancer, we noted that the relation with physical activity was statistically significantly inverse in women but null in men. Possible explanations for the observed difference between women and men include reproductive factors and exogenous estrogens, but relevant biologic pathways of hematologic cancer risk as they interact with physical activity are not well investigated.

Although the relations of physical activity to hematologic cancer subtypes did not vary according to smoking, in a subanalysis of total hematologic cancer, we found that studies that adjusted for smoking showed a stronger inverse association with physical activity than studies that did not adjust for smoking. Smoking is inversely related to physical activity (55) and positively associated with all investigated hematologic cancer subtypes (56–58) except multiple myeloma (59). Thus, adjustment for smoking would be expected to produce a less pronounced inverse association between physical activity and total hematologic cancer than not adjusting for smoking. However, because studies that adjusted for smoking also tended to include a large number of additional adjustment factors, we were unable to discern the effects of controlling for smoking from those controlling for a large number of other adjustment factors, the latter variable of which also represented an effect modifier of the physical activity and total hematologic cancer relation.

Our observation that European studies yielded a positive association between physical activity and diffuse large B-cell lymphoma was based on only two RR estimates. Multiple comparisons could have yielded such a finding by chance because we evaluated a large number of potential effect modifiers.

The etiologic pathways linking increased physical activity to decreased risk of hematologic cancer are not established, but several plausible mechanisms exist. Immunodeficiency is a risk factor for non-Hodgkin lymphoma and Hodgkin lymphoma (5, 8), and physical activity enhances the immune system by increasing the number and activity of natural killer cells, macrophages, lymphokine-activated killer cells, and regulating cytokines (16). Physical activity may also decrease hematologic cancer risk by reducing insulin resistance (12) and decreasing levels of insulin and insulin-like growth factors (11, 60, 61). A further potential mechanism is an exercise-induced reduction of proinflammatory mediators such as interleukin-6 (IL-6), C-reactive protein (CRP), and TNF-α (16), factors that have been positively associated with hematologic cancer (62, 63). Physical activity may also increase the production of antioxidant enzymes, thereby improving DNA repair capacity (15, 16). Finally, physical activity may indirectly decrease hematologic cancer risk by preventing adiposity and reducing body fat (13) because obesity represents a risk factor for non-Hodgkin lymphoma, Hodgkin lymphoma, leukemia, and multiple myeloma (2–4).

Strengths of the present meta-analysis are its large sample size, which provided substantial statistical power and permitted extensive subanalyses. A major asset is the heterogeneity of hematologic cancer subtypes included in the analysis, which enabled us to examine whether physical activity differentially affects specific hematologic cancer subtypes. Also, we comprehensively assessed effect modification of the relation of physical activity to subtype-specific hematologic cancers. In addition, we applied a study quality score that addressed potential selection bias, misclassification, and confounding. Reassuringly, we found no heterogeneity of the risk estimates between higher and lower quality studies.

Limitations of the current meta-analysis are the heterogeneous study designs and the variability in the definitions of high and low levels of physical activity in the underlying studies. However, we addressed the heterogeneity in high and low levels of physical activity in an exploratory dose–response meta-analysis in which we combined RRs of hematologic cancer associated with comparable doses of physical activity. It is possible that preclinical symptoms of hematologic cancer influenced physical activity habits at baseline. Our analyses were limited in that we were unable to remove cases diagnosed in the time period proximal to physical activity assessment. Thus, we may have overestimated any possible inverse association between physical activity and hematologic cancers in our study.

A further potential shortcoming of our study is measurement error in physical activity assessment because studies used self-reports or interview methods to assess physical activity rather than objective instruments such as accelerometers. However, nondifferential misclassification of physical activity levels would have tended to underestimate but not overstate risk estimates. Notwithstanding, the common use of accelerometers might have increased comparability between studies.

Comparability of studies would also have been enhanced if all studies had used a uniform hematologic cancer classification scheme. Some studies did not specify the classification scheme used, whereas other studies provided detailed information about the use of common classification schemes such as the second or third editions of the International Classification of Diseases for Oncology (ICD-O-2/-3) or the InterLymph classification scheme (64). We were unable to distinguish between classical and nonclassical histologic types of Hodgkin lymphoma or between younger adult and older adult Hodgkin lymphoma because of lack of such data.

In summary, our meta-analysis indicates that physical activity is unrelated to risk of subtypes of hematologic cancer. Future studies should include standardized, high-quality physical activity assessments to refine our knowledge regarding whether specific intensities, durations, and frequencies of physical activity may be potentially relevant for protection from hematologic cancers. In addition, mechanistic research is needed to delineate the biologic mechanisms underlying the possible relations of physical activity to hematologic cancer subtypes.

No potential conflicts of interest were disclosed.

Conception and design: C. Jochem, M.F. Leitzmann, G. Behrens

Development of methodology: C. Jochem, M.F. Leitzmann, G. Behrens

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): C. Jochem, M.F. Leitzmann, G. Behrens

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): C. Jochem, M.F. Leitzmann, G. Behrens

Writing, review, and/or revision of the manuscript: C. Jochem, M.F. Leitzmann, M. Keimling, D. Schmid, G. Behrens

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C. Jochem

Study supervision: M.F. Leitzmann, G. Behrens

The authors are indebted to Sylvia Pietsch for expert help.

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.

1.
Ferlay
J
,
Shin
HR
,
Bray
F
,
Forman
D
,
Mathers
C
,
Parkin
DM
. 
Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008
.
Int J Cancer
2010
;
127
:
2893
917
.
2.
Larsson
SC
,
Wolk
A
. 
Body mass index and risk of multiple myeloma: a meta-analysis
.
Int J Cancer
2007
;
121
:
2512
6
.
3.
Larsson
SC
,
Wolk
A
. 
Overweight and obesity and incidence of leukemia: a meta-analysis of cohort studies
.
Int J Cancer
2008
;
122
:
1418
21
.
4.
Larsson
SC
,
Wolk
A
. 
Body mass index and risk of non-Hodgkin's and Hodgkin's lymphoma: a meta-analysis of prospective studies
.
Eur J Cancer
2011
;
47
:
2422
30
.
5.
Hartge
P
,
Wang
SP
,
Bracci
PM
,
Devesa
SS
,
Holly
EA
. 
Non-Hodgkin lymphoma
. In:
Schottenfeld
D
,
Fraumeni
JFJ
,
editors
. 
Cancer epidemiology and prevention
. 3rd ed.
New York
:
Oxford University Press
; 
2006
.
p.
898
918
.
6.
De Roos
AJ
,
Baris
D
,
Weiss
NS
,
Herrinton
LJ
. 
Multiple myeloma
. In:
Schottenfeld
D
,
Fraumeni
JFJ
,
editors
. 
Cancer epidemiology and prevention
. 3rd ed.
New York
:
Oxford University Press
; 
2006
.
p.
919
45
.
7.
Linet
MS
,
Devesa
SS
,
Morgan
GJ
. 
The leukemias
. In:
Schottenfeld
D
,
Fraumeni
JFJ
,
editors
. 
Cancer epidemiology and prevention
. 3rd ed.
New York
:
Oxford University Press
; 
2006
.
p.
841
71
.
8.
Mueller
NE
,
Grufferman
S
. 
Hodgkin lymphoma
. In:
Schottenfeld
D
,
Fraumeni
JFJ
,
editors
. 
Cancer epidemiology and prevention
. 3rd ed.
New York
:
Oxford University Press
; 
2006
.
p.
872
97
.
9.
Wing
RR
. 
Physical activity in the treatment of the adulthood overweight and obesity: current evidence and research issues
.
Med Sci Sports Exerc
1999
;
3
:
547
52
.
10.
Woods
JA
. 
Physical activity, exercise, and immune function
.
Brain Behav Immun
2005
;
19
:
369
70
.
11.
McTiernan
A
. 
Mechanisms linking physical activity with cancer
.
Nat Rev Cancer
2008
;
8
:
205
11
.
12.
Hawley
JA
,
Lessard
SJ
. 
Exercise training-induced improvements in insulin action
.
Acta Physiol
2008
;
192
:
127
35
.
13.
McTiernan
A
,
Sorensen
B
,
Irwin
ML
,
Morgan
A
,
Yasui
Y
,
Rudolph
RE
, et al
Exercise effect on weight and body fat in men and women
.
Obesity
2007
;
15
:
1496
512
.
14.
Shephard
RJ
,
Shek
PN
. 
Cancer, immune function, and physical activity
.
Can J Appl Physiol
1995
;
20
:
1
25
.
15.
Gomez-Cabrera
MC
,
Domenech
E
,
Vina
J
. 
Moderate exercise is an antioxidant: upregulation of antioxidant genes by training
.
Free Radic Biol Med
2008
;
44
:
126
31
.
16.
Pan
SY
,
Morrison
H
. 
Physical activity and hematologic cancer prevention
.
Recent Results Cancer Res
2011
;
186
:
135
58
.
17.
Vermaete
NV
,
Wolter
P
,
Verhoef
GE
,
Kollen
BJ
,
Kwakkel
G
,
Schepers
L
, et al
Physical activity and risk of lymphoma: a meta-analysis
.
Cancer Epidemiol Biomarkers Prev
2013
;
22
:
1173
84
.
18.
Brownson
RC
,
Chang
JC
,
Davis
JR
,
Smith
CA
. 
Physical activity on the job and cancer in Missouri
.
Am J Public Health
1991
;
81
:
639
42
.
19.
Cerhan
JR
,
Bernstein
L
,
Severson
RK
,
Davis
S
,
Colt
JS
,
Blair
A
, et al
Anthropometrics, physical activity, related medical conditions, and the risk of non-hodgkin lymphoma
.
Cancer Causes Control
2005
;
16
:
1203
14
.
20.
Kasim
K
,
Johnson
KC
,
Levallois
P
,
Abdous
B
,
Auger
P
Canadian Cancer Registries Epidemiology Research G
. 
Recreational physical activity and the risk of adult leukemia in Canada
.
Cancer Causes Control
2009
;
20
:
1377
86
.
21.
Keegan
TH
,
Glaser
SL
,
Clarke
CA
,
Dorfman
RF
,
Mann
RB
,
DiGiuseppe
JA
, et al
Body size, physical activity, and risk of Hodgkin's lymphoma in women
.
Cancer Epidemiol Biomarkers Prev
2006
;
15
:
1095
101
.
22.
Kelly
JL
,
Fredericksen
ZS
,
Liebow
M
,
Shanafelt
TD
,
Thompson
CA
,
Call
TG
, et al
The association between early life and adult body mass index and physical activity with risk of non-Hodgkin lymphoma: impact of gender
.
Ann Epidemiol
2012
;
22
:
855
62
.
23.
Pan
SY
,
Mao
Y
,
Ugnat
AM
Canadian Cancer Registries Epidemiology Research G
. 
Physical activity, obesity, energy intake, and the risk of non-Hodgkin's lymphoma: a population-based case-control study
.
Am J Epidemiol
2005
;
162
:
1162
73
.
24.
Parent
ME
,
Rousseau
MC
,
El-Zein
M
,
Latreille
B
,
Desy
M
,
Siemiatycki
J
. 
Occupational and recreational physical activity during adult life and the risk of cancer among men
.
Cancer Epidemiol
2011
;
35
:
151
9
.
25.
Zahm
SH
,
Hoffman-Goetz
L
,
Dosemeci
M
,
Cantor
K
,
Blair
A
. 
Occupational physical activity and non-Hodgkin's lymphoma
.
Med Sci Sports Exerc
1999
;
31
:
566
71
.
26.
Birmann
BM
,
Giovannucci
E
,
Rosner
B
,
Anderson
KC
,
Colditz
GA
. 
Body mass index, physical activity, and risk of multiple myeloma
.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
1474
8
.
27.
Blair
CK
,
Cerhan
JR
,
Folsom
AR
,
Ross
JA
. 
Anthropometric characteristics and risk of multiple myeloma
.
Epidemiology
2005
;
16
:
691
4
.
28.
Cerhan
JR
. 
Anthropometric characteristics, physical activity, and risk of non-Hodgkin's lymphoma subtypes and B-cell chronic lymphocytic leukemia: a prospective study
.
Am J Epidemiol
2002
;
156
:
527
35
.
29.
Hofmann
JN
,
Moore
SC
,
Lim
U
,
Park
Y
,
Baris
D
,
Hollenbeck
AR
, et al
Body mass index and physical activity at different ages and risk of multiple myeloma in the NIH-AARP diet and health study
.
Am J Epidemiol
2013
;
177
:
776
86
.
30.
Kabat
GC
,
Wu
JW
,
Moore
SC
,
Morton
LM
,
Park
Y
,
Hollenbeck
AR
, et al
Lifestyle and dietary factors in relation to risk of chronic myeloid leukemia in the NIH-AARP diet and health study
.
Cancer Epidemiol Biomarkers Prev
2013
;
22
:
848
54
.
31.
Kabat
GC
,
Kim
MY
,
Jean Wactawski
W
,
Bea
JW
,
Edlefsen
KL
,
Adams-Campbell
LL
, et al
Anthropometric factors, physical activity, and risk of non-Hodgkin's lymphoma in the Women's Health Initiative
.
Cancer Epidemiol
2012
;
36
:
52
9
.
32.
Khan
MMH
,
Mori
M
,
Sakauchi
F
,
Matsuo
K
,
Ozasa
K
,
Tamakoshi
A
. 
Risk factors for multiple myeloma: evidence from the Japan Collaborative Cohort (JACC) study
.
Asian Pac J Cancer Prev
2006
;
7
:
575
81
.
33.
Lim
U
,
Morton
LM
,
Subar
AF
,
Baris
D
,
Stolzenberg-Solomon
R
,
Leitzmann
M
, et al
Alcohol, smoking, and body size in relation to incident Hodgkin's and non-Hodgkin's lymphoma risk
.
Am J Epidemiol
2007
;
166
:
697
708
.
34.
Lu
Y
,
Prescott
J
,
Sullivan-Halley
J
,
Henderson
KD
,
Ma
H
,
Chang
ET
, et al
Body size, recreational physical activity, and B-cell non-Hodgkin lymphoma risk among women in the California teachers study
.
Am J Epidemiol
2009
;
170
:
1231
40
.
35.
Ma
X
,
Park
Y
,
Mayne
ST
,
Wang
R
,
Sinha
R
,
Hollenbeck
AR
, et al
Diet, lifestyle, and acute myeloid leukemia in the NIH-AARP cohort
.
Am J Epidemiol
2010
;
171
:
312
22
.
36.
Paffenbarger
RS
,
Lee
I-M
,
Wing
AL
. 
The influence of physical activity on the incidence of site-specific cancers in college alumni
.
Adv Exp Med Biol
1992
;
322
:
7
15
.
37.
Pukkala
E
,
Kaprio
J
,
Koskenvuo
M
,
Kujala
U
,
Sarna
S
. 
Cancer incidence among FinnishWorld class male athletes
.
Int J Sports Med
2000
;
21
:
216
20
.
38.
Soll-Johanning
H
,
Bach
E
. 
Occupational exposure to air pollution and cancer risk among Danish urban mail carriers
.
Int Arch Occup Environ Health
2004
;
77
:
351
6
.
39.
Teras
LR
,
Gapstur
SM
,
Diver
WR
,
Birmann
BM
,
Patel
AV
. 
Recreational physical activity, leisure sitting time and risk of non-Hodgkin lymphoid neoplasms in the American Cancer Society Cancer Prevention Study II Cohort
.
Int J Cancer
2012
;
131
:
1912
20
.
40.
van Veldhoven
CM
,
Khan
AE
,
Teucher
B
,
Rohrmann
S
,
Raaschou-Nielsen
O
,
Tjonneland
A
, et al
Physical activity and lymphoid neoplasms in the European Prospective Investigation into Cancer and nutrition (EPIC)
.
Eur J Cancer
2011
;
47
:
748
60
.
41.
Robsahm
TE
,
Hestvik
UE
,
Veierod
MB
,
Fagerlie
A
,
Nystad
W
,
Engebretsen
L
, et al
Cancer risk in Norwegian world class athletes
.
Cancer Causes Control
2010
;
21
:
1711
9
.
42.
Wannamethee
SG
,
Shaper
AG
,
Walker
M
. 
Physical activity and risk of cancer in middle-aged men
.
Br J Cancer
2001
;
85
:
1311
6
.
43.
Monninkhof
EM
,
Elias
SG
,
Vlems
FA
,
van der Tweel
I
,
Schuit
AJ
,
Voskuil
DW
, et al
Physical activity and breast cancer: a systematic review
.
Epidemiology
2007
;
18
:
137
57
.
44.
Behrens
G
,
Leitzmann
MF
. 
The association between physical activity and renal cancer: systematic review and meta-analysis
.
Br J Cancer
2013
;
108
:
798
811
.
45.
Liu
Y
,
Hu
F
,
Li
D
,
Wang
F
,
Zhu
L
,
Chen
W
, et al
Does physical activity reduce the risk of prostate cancer? A systematic review and meta-analysis
.
Eur Urol
2011
;
60
:
1029
44
.
46.
Voskuil
DW
,
Monninkhof
EM
,
Elias
SG
,
Vlems
FA
,
van Leeuwen
FE
Task Force Physical A
. 
Physical activity and endometrial cancer risk, a systematic review of current evidence
.
Cancer Epidemiol Biomarkers Prev
2007
;
16
:
639
48
.
47.
Higgins
JP
,
Thompson
SG
. 
Quantifying heterogeneity in a meta-analysis
.
Stat Med
2002
;
21
:
1539
58
.
48.
Begg
CB
,
Mazumdar
M
. 
Operating characteristics of a rank correlation test for publication bias
.
Biometrics
1994
;
50
:
1088
101
.
49.
Rota
M
,
Bellocco
R
,
Scotti
L
,
Tramacere
I
,
Jenab
M
,
Corrao
G
, et al
Random-effects meta-regression models for studying nonlinear dose-response relationship, with an application to alcohol and esophageal squamous cell carcinoma
.
Stat Med
2010
;
29
:
2679
87
.
50.
R Core Team
. 
R: a language and environment for statistical computing
.
Vienna, Austria
:
R Core Team
; 
2013
.
51.
Viechtbauer
W
. 
Conducting meta-analyses in R with the metafor package
.
J Stat Softw
2010
;
36
:
1
48
.
52.
Gasparrini
A
,
Armstrong
B
,
Kenward
MG
. 
Multivariate meta-analysis for non-linear and other multi-parameter associations
.
Stat Med
2012
;
31
:
3821
39
.
53.
Ainsworth
BE
,
Haskell
WL
,
Herrmann
SD
,
Meckes
N
,
Bassett
DR
 Jr
,
Tudor-Locke
C
, et al
2011 Compendium of physical activities: a second update of codes and MET values
.
Med Sci Sports Exerc
2011
;
43
:
1575
81
.
54.
Walsh
NP
,
Gleeson
M
,
Pyne
DB
,
Nieman
DC
,
Dhabhar
FS
,
Shephard
RJ
, et al
Position statement. Part two: maintaining immune health
.
Exerc Immunol Rev
2011
;
17
:
64
103
.
55.
Leitzmann
MF
,
Koebnick
C
,
Abnet
CC
,
Freedman
ND
,
Park
Y
,
Hollenbeck
A
, et al
Prospective study of physical activity and lung cancer by histologic type in current, former, and never smokers
.
Am J Epidemiol
2009
;
169
:
542
53
.
56.
Castillo
JJ
,
Dalia
S
. 
Cigarette smoking is associated with a small increase in the incidence of non-Hodgkin lymphoma: a meta-analysis of 24 observational studies
.
Leuk Lymphoma
2012
;
53
:
1911
9
.
57.
Castillo
JJ
,
Dalia
S
,
Shum
H
. 
Meta-analysis of the association between cigarette smoking and incidence of Hodgkin's Lymphoma
.
J Clin Oncol
2011
;
29
:
3900
6
.
58.
Brownson
RC
,
Novotny
TE
,
Perry
MC
. 
Cigarette smoking and adult leukemia. A meta-analysis
.
Arch Intern Med
1993
;
153
:
469
75
.
59.
Psaltopoulou
T
,
Sergentanis
TN
,
Kanellias
N
,
Kanavidis
P
,
Terpos
E
,
Dimopoulos
MA
. 
Tobacco smoking and risk of multiple myeloma: a meta-analysis of 40 observational studies
.
Int J Cancer
2013
;
132
:
2413
31
.
60.
Khandwala
HM
,
McCutcheon
IE
,
Flyvbjerg
A
,
Friend
KE
. 
The effects of insulin-like growth factors on tumorigenesis and neoplastic growth
.
Endocr Rev
2000
;
21
:
215
44
.
61.
Gupta
K
,
Krishnaswamy
G
,
Karnad
A
,
Peiris
AN
. 
Insulin: a novel factor in carcinogenesis
.
Am J Med Sci
2002
;
323
:
140
5
.
62.
Sugiyama
H
,
Inoue
K
,
Ogawa
H
,
Yamagami
T
,
Soma
T
,
Miyake
S
, et al
The expression of IL-6 and its related genes in acute leukemia
.
Leuk Lymphoma
1996
;
21
:
49
52
.
63.
Mischke
R
,
Waterston
M
,
Eckersall
PD
. 
Changes in C-reactive protein and haptoglobin in dogs with lymphatic neoplasia
.
Vet J
2007
;
174
:
188
92
.
64.
Turner
JJ
,
Morton
LM
,
Linet
MS
,
Clarke
CA
,
Kadin
ME
,
Vajdic
CM
, et al
InterLymph hierarchical classification of lymphoid neoplasms for epidemiologic research based on the WHO classification (2008): update and future directions
.
Blood
2010
;
116
:
e90
8
.

Supplementary data