8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

1/12

Original Contribution

Physical Activity, Obesity, Energy Intake, and the Risk of Non-Hodgkins

Lymphoma: A Population-based Case-Control Study

Sai Yi Pan 1 , Yang Mao 1 , Anne-Marie Ugnat 1 , and the Canadian Cancer RegistriesEpidemiology Research Group

1 Surveillance and Risk Assessment Division, Centre for Chronic Disease Prevention and Control,Public Health Agency of Canada, Ottawa, Ontario, Canada.

Received for publication February 24, 2005; accepted for publication August 1, 2005.

The authors conducted a population-based case-control study of 1,030 cases with histologically conrmed,incident non-Hodgkins lymphoma (NHL) and 3,106 controls to assess the impact of recreational physical activity,obesity, and energy intake on NHL risk in Canada from 1994 to 1997. Compared with those for subjects in thelowest quartiles of total recreational physical activity, multivariable-adjusted odds ratios for subjects in the highestquartile were 0.79 (95% condence interval (CI): 0.59, 1.05) for men and 0.59 (95% CI: 0.42, 0.81) for women.Obesity (body mass index 30 kg/m 2 ) was associated with odds ratios of 1.59 (95% CI: 1.18, 2.12) for men and1.36 (95% CI: 1.00, 1.84) for women. For men and women with a lifetime maximum body mass index of 30 kg/m 2 ,respective odds ratios were 1.55 (95% CI: 1.16, 2.06) and 1.10 (95% CI: 0.83, 1.46). For men and women in thehighest quartiles of calorie intake, respective odds ratios were 1.95 (95% CI: 1.45, 2.62) and 1.13 (95% CI: 0.84,1.52). Some differences were found between histologic subtypes of NHL for these associations. This studysuggests that recreational physical activity decreases NHL risk, while obesity and excess calorie intake increase

it. More studies are needed to conrm these results, especially the differences between histologic subtypes.

case-control studies; energy intake; exercise; lymphoma, non-Hodgkin; obesity; recreation

Abbreviations: BMI, body mass index; CI, condence interval; MET, metabolic equivalent task; NHL, non-Hodgkins lymphoma.

The incidence and mortality of non-Hodgkins lymphoma(NHL) have steadily increased in Canada (1) and in theUnited States since the 1970s but have stabilized in recentyears in the United States (2). Over the last 30 years, the

number of new cases of NHL has more than tripled amongboth men and women while the rates have doubled; and thenumber of annual deaths from NHL also has almost tripled,although the mortality rate has increased more modestly (1).This trend of increase appears worldwide (3), and the in-cidence rates in Canada and the United States are among thehighest in theworld (2). However, the reason for the increaseis largely unknown.

The etiology of NHL is poorly understood. Suspected risk factorsincludeimmunodeciency; infections of humanT-cell

lymphotrophic virus (types I and II), Epstein-Barr virus, and Helicobacter pylori ; family history; and agricultural andpesticide exposure (3). Immunodeciency is the strongestrisk factor known to increase NHL risk (4), supported by

evidence of substantially increased risk of NHL for patientstreated with immunosuppressive drugs (5), people infectedwith human immunodeciency virus (6), and young peoplewith ataxia-telangiectasia or the Wiskott-Aldrich syndromeas well as children with X-linked lymphoproliferative syn-drome or combined immunodeciency (7).

Physical activity has been associated with reduced risksof some types of cancer (810), and improving immunefunction has been hypothesized to be one of the underlyingmechanisms (10, 11); thus, it may decrease the risk of NHL.

Correspondence to Dr. Yang Mao, Surveillance and Risk Assessment Division, Centre for Chronic Disease Prevention and Control, Public HealthAgency of Canada, 120 Colonnade Road, Locator 6702A, Ottawa, Ontario, Canada K1A 0K9 (e-mail: Yang_Mao@phac-aspc.gc.ca).

1162 Am J Epidemiol 2005;162:11621173

American Journal of EpidemiologyCopyright 2005 by the Johns Hopkins Bloomberg School of Public HealthAll rights reserved; printed in U.S.A.

Vol. 162, No. 12DOI: 10.1093/aje/kwi342

Advance Access publication November 3, 2005

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

2/12

Obesity has been associated with signicant metabolic ab-normalities, including insulin resistance, glucose intoler-ance, and diabetes mellitus (12) as well as impairedimmune function (13). Excess energy intake is also relatedto obesity. However, limited data are available, and the nd-ings from previous epidemiologic research on the associa-tion of physical activity and obesity with NHL risk havebeen conicting (1423). Since there are very few knownrisk factors for NHL, and physical activity and obesity aremodiable lifestyle factors, if proven to be related to the risk of NHL, adoption of a healthy lifestyle would be a meaning-ful strategy for combating this tumor. Therefore, we con-ducted this study to examine the impact of recreationalphysical activity, obesity, and energy intake on the risk of NHL, using data from a large, population-based case-controlstudy in Canada.

MATERIALS AND METHODS

Study population

This study was part of the National Enhanced CancerSurveillance System, a multicomponent, collaborative proj-ect of Health Canada and the provincial cancer registries.The case-control component included individual data from21,020 Canadians with one of 19 types of cancers and 5,039population controls aged 2076 years; these data were col-lected between 1994 and 1997 in eight of the 10 Canadianprovinces (Alberta, British Columbia, Manitoba, Newfound-land, Nova Scotia, Ontario, Prince Edward Island, and Sas-katchewan). The respective ethics review boards of eachprovince reviewed and approved the study proposal. Thecurrent analysis was based on 1,030 incident cases of NHL

and 3,106 controls from all eight provinces except Ontario.The population-based provincial cancer registries identi-ed NHL cases through a review of pathology reports. Allcases were patients with histologically conrmed incidentNHL, newly diagnosed between 1994 and 1997 in the sevenparticipating provinces. The cancer registries tried to iden-tify cases as soon as possible after diagnosis to reduce theloss of subjects caused by severe illness and death. Theregistries identied 1,678 NHL cases. Physicians refusedconsent to contact 109 cases (6.5 percent), and 147 cases(8.8 percent) died before they could be sent questionnaires.Questionnaires were mailed to 1,422 cases; 1,030 casescompleted and returned the questionnaires, representing72.4 percent of cases who were sent questionnaires and61.4 percent of ascertained cases.

The morphologic data were derived from pathologyreports and were coded by using the International Classi-cation of Diseases for Oncology , Second Edition. The his-tologic subtypes of NHL were grouped on the basis of thiscoding by using the method developed by Groves et al. (24).However, because of the small number of cases for thecategories of high-grade and peripheral T cell, the histologicsubtypes were grouped into the following four broader cat-egories: diffuse, follicular, small lymphocytic, and others.

In the National Enhanced Cancer Surveillance System,frequency matching to the overall case group (19 types of cancers) was used to selectpopulation controls with a similar

ageand sex distributionso there would be at least onecontrolfor every case within each sex and 5-year age group for anyspecic cancer site in each province. The sampling strategyfor control selection varied by province, depending on dataavailability, data quality (completeness and timeliness), andthe condentiality restrictions of provincial databases.

Prince Edward Island, Nova Scotia, Manitoba, Saskatch-ewan, and British Columbia used provincial health insur-ance plans to obtain a random sample of the provincialpopulation stratied by age group and sex. More than 95percent of Canadians are covered by these public plans, andindividuals are excluded only if they are covered throughother federal plans. Newfoundland and Alberta used similarrandom digit dialing protocols to obtain population samples.In Alberta, the University of Albertas Population ResearchLaboratory generated a random sample of provincial tele-phone numbers, including unlisted numbers. Of the numberscalled, 4 percent were not in service or were assigned tobusinesses, 3.6 percent involved a communication barrier,and, for 11.5 percent, there was no answer after attempting

to call eight times. Of those households contacted, 91.3 per-cent agreed to a census of residents, and 90.1 percent of theeligible individuals agreed to have a questionnaire sent.Ninety-nine percent of Albertan households have tele-phones, and the Population Research Laboratory estimatesthat 9297 percent of people in the province are reachable.Newfoundland Telephone Company provided the local can-cer registry with a random sample of Newfoundland phonenumbers, including unlisted numbers. Exact contact andeligibility rates are unavailable; however, study personnelestimated that 85 percent of the phone numbers werereached. Cooperation levels were similar to those in Alberta.Of the controls who were sent questionnaires, 83 percent

and 75 percent completed and returned them in Alberta andNewfoundland, respectively.The provincial cancer registries recruited 5,107 subjects

without cancer in the seven participating provinces studiedand mailed these subjects the same questionnaires as thosesent to cases. Questionnaires were returned for 81 controls(1.6 percent) because of a wrong or old address, and noupdated address could be found. A total of 3,106 controlscompleted and returned questionnaires, representing 60.8percent of the ascertained controls.

Data collection

Theprovincialregistriescollecteddata by self-administeredquestionnaires, with telephone follow-up when necessaryfor clarication and completeness. The registries used thesame protocol to collect data for both cases and controls.

The questionnaires were designed to obtain detailed dataon risk factors for cancers. Information was collected oneducation, average family income over the last 5 years, mar-ital status, ethnic group, height, weight, physical activity,alcohol consumption, diet (69-item food frequency ques-tions), and vitamin and mineral supplement use 2 years be-fore interview. Questionnaires also gathered informationabout smoking history, menstrual and reproductive history,employment history, residential history, and history of occu-pational exposure to some specic chemicals.

Physical Activity, Obesity, Energy Intake, and Non-Hodgkins Lymphoma 1163

Am J Epidemiol 2005;162:11621173

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

3/12

Assessment of physical activity

The questionnaire elicited information on recreationalphysical activity 2 years before interview. Respondents wereasked in which seasons, how often, and how long per ses-sion, on average, they participated in each of the 12 mostcommon types of leisure-time physical activity in Canada.

Individual activities included walking for exercise, joggingor running, gardening or yard work, home exercise or exer-cise class, golf, racquet sports, bowling or curling, swim-ming or water exercise, skiing or skating, bicycling, socialdancing, and other strenuous exercise. Respondents indi-cated their usual frequency of participating in each of theseactivities by choosing one of the following categories:never, less than once per month, 13 times per month, 12times per week, 36 times per week, or every day. Time persession was recorded as less than 15 minutes, 1530 min-utes, 3160 minutes, and more than 60 minutes. We esti-mated the intensity of each activity by assigning a specicmetabolic equivalent task (MET) value to each reported

activity. The MET values used here were abstracted fromthe Compendium of Physical Activities (25, 26). A MET isdened as the ratio of the associated metabolic rate for aspecic activity compared with the resting metabolic rate(27). One MET is the average seated resting energy cost foran adult and is set at 3.5 ml/kg of body weight per minute of oxygen.

A weekly number of MET-hours was derived for eachactivity by combining the frequency, duration, and METvalue (intensity) of each activity. We categorized levels of recreational activity as moderate (MET 3 6), vigorous(MET > 6), and total (moderate plus vigorous) (28). Thevariable used in the analysis was the sum of each categoryof moderate, vigorous, and total physical activity.

Assessment of obesity and energy intake

Participants in the study reported their adult height andweight 2 years before interview as well as their lifetimemaximum weight (except during pregnancy). As a measureof overweight and obesity, body mass index (BMI) wascalculated as the reference weight in kilograms divided byheight in meters squared. On the basis of World HealthOrganization standards, obesity was dened as a BMI of 30 kg/m 2 or more, and overweight was dened as a BMI of between 25 and less than 30 kg/m

2for both sexes (29).

The questionnaire asked subjects the usual frequency andportion size for each of the 69 food items consumed 2 yearsbefore interview. We calculated weekly intake of caloriesfor each item by multiplying the quantity of each item perweek by the associated calorie value, which is determinedfrom food composition data by using the Canadian NutrientGuide (30). We summed the weekly calorie intake for all 69items to obtain total calorie intake.

Statistical analysis

We estimated the risk of NHL associated with recrea-tional physical activity and obesity based on odds ratiosand corresponding 95 percent condence intervals, using

unconditional logistic regression with the software packageSAS (version 8; SAS Institute, Inc., Cary, North Carolina).Variables were categorized into quartiles based on the dis-tribution of the variables in the control population.

Because cases and controls were not directly matched, themethods for identifying cases and controls varied by prov-ince;and, because ageis associated with NHLrisk,all logisticregression analyses were controlled for province of resi-dence and age to remove the impact of any uneven distribu-tion of these factors between cases and controls. We used thechange-in-point-estimate approach to assess the potentialconfounding effect of a wide range of factors, including age,educational level, family income adequacy, marital status,alcohol consumption, smoking, BMI, total calorie intake,menopausal status, and number of livebirths. We retainedvariables in the nal models that are considered biologicallyimportant if their inclusion changed the odds ratio estimateappreciably, regardless of the statistical signicance. We ad- justed the nal multivariate models for age (years, continu-ous), province of residence, education (years completed:< 10, 1012, > 12), alcohol consumption (servings per week,continuous), pack-years of smoking (continuous), total cal-orie intake (kilocalories per week, continuous), self-reportedexposure to some chemicals, and ever employment in someoccupations. We conducted tests for trends for all models of categorized data by treating the different categories as a sin-gle ordinal variable.

The literature suggests that risk factors for NHL maydiffer by histologic subtype; therefore, we performed strat-ied analysis by histologic subtype of NHL. Because BMI isrelated to levels of insulin and insulin-like growth factors,which may also be affected by physical activity (1012), weassessed possible effect modication by BMI.

RESULTS

Data from 1,030 NHL cases and 3,106 controls wereavailable for analysis. Diffuse NHL was the most commonhistologic subtype among cases ( n 419, 40.7 percent of allcases), followed by follicular ( n 242, 23.5 percent) andsmall lymphocytic ( n 100, 9.7 percent). All other typesaccounted for 26.1 percent ( n 269) of all cases of NHL.

Table 1 shows the distribution of some selected character-istics of NHL cases and controls. For both sexes combined,compared with controls, cases tended to have higher totalcalorie intake and were more likely to be obese and to beexposed to several chemicals (pesticides, herbicides, vinylchloride, benzidine, benzene, mineral or cutting oil, anddyestuffs). There were no clear differences between casesand controls regarding other variables. For men and womenseparately, the distribution of these variables among casesand controls was similar to that for both sexes combined,except that female cases drank less alcohol than did femalecontrols.

Table 2 presents the NHL risks associated with differentlevels of recreational physical activity (moderate, vigorous,and total), BMI, and calorie intake, by sex. Compared withthose for subjects in the lowest quartiles of total activity,multivariable-adjusted odds ratios for those in the highest

1164 Pan et al.

Am J Epidemiol 2005;162:11621173

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

4/12

TABLE 1. Selected characteristics of non-Hodgkins lymphoma cases and controls, National Enhanced Cancer Surveillance System, Canada, 19941997

CharacteristicMen Women

Cases(n 569)

Controls(n 1,642) p value

Cases(n 461)

Controls(n 1,464) p value

Ca(n

Age (years) 56.7 (12.5) 57.9 (14.6) 0.061 57.1 (12.6) 56.2 (12.2) 0.172 56.9 (12.6) 2024 0.7 2.2 0.9 1.3 0.8 2529 1.9 3.8 1.9 1.6 1.9 3034 2.8 4.9 3.0 2.5 2.9

3539 7.0 5.9 5.6 4.1 6.4 4044 5.6 4.2 5.6 8.1 5.6 4549 9.3 4.1 10.6 11.5 9.9 5054 11.8 5.9 10.0 11.6 11.0 5559 12.7 8.8 11.5 13.4 12.2 6064 15.3 14.2 15.2 15.4 15.3 6669 15.6 21.3 18.7 16.2 17.0 7076 17.1 24.6 16.9 14.2 17.0

Province of residenceNewfoundland 6.0 7.6 0.9 1.3 6.3 Prince Edward Island 2.6 5.0 1.9 1.6 2.7 Nova Scotia 8.3 20.5 3.0 2.5 9.0 Manitoba 9.7 9.5 5.6 4.1 10.4 Saskatchewan 9.3 9.0 5.6 8.1 10.1 Alberta 25.7 19.9 10.6 11.5 25.4 British Columbia 38.5 28.5 10.0 11.6 36.0

Educational level (years) 12.2 (3.8) 11.8 (4.0) 0.038 11.8 (3.5) 11.9 (3.2) 0.537 12.0 (3.6) Alcohol drinking

(servings/week) 6.4 (10.4) 6.4 (11.1) 0.962 1.7 (3.6) 2.4 (5.4) 0.002 4.3 (8.4) Total calorie intake

(kcal/week) 14,944 (5,322) 14,042 (5,745) 0.0007 13,002 (5,097) 12,892 (7,174) 0.714 14,075 (5,309) Vegetable consumption

(servings/week) 19.9 (12.4) 19.4 (12.4) 0.414 21.2 (13.2) 21.3 (15.8) 0.933 20.5 (12.8) Pack-years of smoking 20.7 (23.5) 21.4 (24.3) 0.541 10.4 (15.5) 9.6 (15.3) 0.336 16.1 (20.9) Smoking status

Never smoked 27.6 25.6 46.4 48.6 36 Former smoker 45.2 49.7 32.8 31.5 39.6 Current smoker 27.2 24.7 20.8 19.9 24.4

Chemical exposure yYes 39.5 34.2 13.7 11.5 28.0 No 60.5 65.8 86.3 88.5 72.0

Occupational exposure zYes 39.2 36.9 9.5 5.7 25.9 No 60.8 63.1 90.5 94.3 74.1

* Values are expressed as mean (standard deviation) or as percentage.y Self-reported exposure to pesticides, herbicides, vinyl chloride, benzidine, benzene, mineral or cutting oil, and dyestuffs.z Ever employed in farming, horticultural, and animal husbandry occupations; forestry and logging occupations; chemicals, petroleum, rubber, and plastic pro

machining, shaping, and forming occupations; fabricating, assembling, and repairing occupations for metal, rubber, and plastic products; mechanics and repaire

A m J E

pi d emi ol 2 0 0 5 ; 1 6 2 : 1 1 6 2 1 1 7 3

b y g u e s t o n M a y 1 5 , 2 0 1 4 h t t p : / / a j e . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d f r o m

http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

5/12

http://aje.oxfordjournals.org/

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

6/12

quartiles were 0.79 (95 percent condence interval (CI):0.59, 1.05) for men and 0.59 (95 percent CI: 0.42, 0.81)for women. The risk reduction was attributed to both mod-erate and vigorous physical activity but was more profoundfor women than for men. For obese (BMI 30 kg/m

2)

men and women compared with those whose BMI was18.5 < 25 kg/m

2, the respective multivariable-adjusted odds

ratios were 1.59 (95 percent CI: 1.18, 2.12) and 1.36 (95 per-cent CI: 1.00, 1.84). For subjects with a lifetime maximumBMI of 30 kg/m 2 , the odds ratios were 1.55 (95 percentCI: 1.16, 2.06) for men and 1.10 (95 percent CI: 0.83, 1.46)for women. We found an increased odds ratio of 1.95 for menin the highest quartile of calorie intake, but not for women(odds ratio 1.13).

We also examined the risks of NHL associated with rec-reational physical activity (moderate, vigorous, and total),BMI, and calorie intake by NHL histologic subtype (table 3).We observed various degrees of reduction in NHL risk re-lated to total activity forall four subtypes, with a largereffectfor follicular, small lymphocytic, and others than for the dif-

fuse type. Theobesity-related increase in NHL risk wasmoreprofound for diffuse and others types than for follicular andsmall lymphocytic type. The risk increase associated with ex-cess calorie intake was observed for follicular, small lympho-cytic,andotherstypes,with no apparent effect fordiffuse type.

We then assessed the effect of total recreational physicalactivity on NHL risk stratied by BMI, by sex (table 4).A decreased NHL risk associated with total activity wasrestricted to women whose BMI was 18.5 < 25 and 25< 30 kg/m

2and to men whose BMI was 18.5 < 25 and

BMI 30 kg/m2.

DISCUSSIONOur results indicated that higher levels of recreational

physical activity were associated with a reduced risk of NHL for both sexes, with a larger effect for women. Thisactivity-related decrease in risk was attributed to both mod-erate and vigorous activity. We observed greater reductionsin NHL risk associated with recreational physical activity forfollicular, small lymphocytic, and others subtypes of NHLtumors. This study also suggested that obese men andwomen had a signicantly increased risk of NHL, and themagnitude of the obesity-related risk increase was greaterfor diffuse and others subtypes. We also found that excessenergy intake increased the risk of NHL for men, and thisincrease was observed for all four subtypes except diffuse.

Only a few studies have been published on the associationbetween physical activity and NHL risk. To our knowledge,our study is the second that has assessed the relation of physical activity with NHL risk by histologic subtype.The prospective study of Iowa women (14) found no age-adjusted association between level of recreational physicalactivity and NHL risk, overall and by subtype, but a sugges-tive (nonsignicant) inverse association for follicular sub-types. However, the Iowa study did not assess duration of physical activity. In another cohort study on college alumni(15), those who played sports at least 5 hours per week attheir colleges had a (nonsignicant) relative risk of 0.67 of

developing NHL compared with those who played less ornot at all. Two other studies (16, 17) found no evidence of anassociation between occupational physical activity andNHL risk; however, as the authors of one of the studiespointed out, the occupation-based measures of physical ac-tivity might have introduced nondifferential misclassica-tion, which tends to bias the risk estimates toward unity. Thediscrepancy between our study and others could be due tothe different parameters (e.g., duration, frequency, intensity,and type) of physical activity assessed by different studies,different proles of the populations, different strategies of adjustment for confounders, or different proportions of his-tologic subtypes.

Research on the relation of obesity to NHL risk is alsoscarce, and the results are conicting. The increased risk of NHL associated with obesity observed in our study is con-sistent with ndings from four other studies (1821). Apopulation-based case-control study with 725 cases and1,566 controls showed elevated risks associated with obesityfor NHL and the two major subtypes, diffuse large cell and

follicular lymphoma (18). In a Swedish hospital-based co-hort study of 29,129 persons, women (but not men) with ahospital diagnosis of obesity had an excess risk of NHLwhen compared with the incidence in the general population(19). An elevated risk of NHL associated with a higherponderal index was also reported in an early cohort study(20). Furthermore, in the largest cohort study (404,576 menand 495,477 women), BMI was signicantly associated withhigher rates of death due to NHL among both men andwomen (21). However, three studies found no associationbetween obesity and NHL risk. The cohort study of Iowawomen reported no relation, either overall or by subtype(14), and neither did an Italian case-control study (22) nor

a Danish record-linkage study (23).The underlying mechanisms operative in the associationbetween physical activity and NHL have not been estab-lished. One of the plausible mechanisms hypothesized isthe exercise-induced increase in antitumor immune defenses(10, 11). Moderate habitual physical activity may enhanceimmune function by increasing the number and activity of macrophages, natural killer cells, lymphokine-activatedkiller cells, and regulating cytokines (3133). Experimentalstudies have demonstrated greater activity of natural killercells and of lymphokine-activated killer cells in trained orphysically active mice, resulting in greater clearance of tu-mor cells and incidence of tumors (3436). Studies in hu-mans have suggested that moderate exercise training hasbeen associated with increases in natural killer cell activityor cell count (3740), alterations in lymphocyte subpopu-lations (41, 42), changes in interleukin-2 production andinterleukin-2 receptor expression (38, 39), and elevation inimmunoglobulin levels (43). Because immunodeciency isa strong risk factor for NHL, physical-activity-related im-provement in immune function may play a role in the pro-tective effect of physical activity against developing NHL.Immunotherapy with high-dose interleukin-2 has beenshown to successfully treat tumors in animal models (44)and cause dramatic tumor regressions in some patients withmetastatic melanoma, renal cell carcinoma, and NHL (4547). Other hypothesized mechanisms include improving

Physical Activity, Obesity, Energy Intake, and Non-Hodgkins Lymphoma 1167

Am J Epidemiol 2005;162:11621173

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

7/12

http://aje.oxfordjournals.org/

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

8/12

Vigorous z0 1,383 43 1.00 Ref 1.00 Ref 119 1.00 < 2.3 578 14 0.84 0.45, 1.56 0.87 0.46, 1.62 60 1.18 2.3 < 11.8 576 28 1.64 0.95, 2.74 1.66 0.94, 2.75 47 0.86

11.8 569 15 0.98 0.53, 1.82 1.00 0.53, 1.90 43 0.74 p for trend 0.35 0.35 0

Total< 6.3 778 27 1.00 Ref 1.00 Ref 81 1.00 6.3 < 17.0 776 26 0.85 0.49, 1.49 0.89 0.50, 1.57 65 0.71 17.0 < 34.4 776 23 0.72 0.41, 1.27 0.75 0.42, 1.35 57 0.58

34.4 776 24 0.74 0.42, 1.32 0.74 0.41, 1.33 66 0.63

p for trend 0.26 0.27 0Body mass index

18.5 < 25 1,399 35 1.00 Ref 1.00 Ref 108 1.00 25 < 30 1,178 48 1.61 1.03, 2.52 1.64 1.04, 2.59 96 1.20

30 450 13 1.23 0.64, 2.35 1.27 0.66, 2.44 58 2.08 p for trend 0.21 0.18 0

Lifetime maximum body mass index18.5 < 25 880 29 1.00 Ref 1.00 Ref 69 1.00 25 < 30 1,307 40 0.88 0.54, 1.44 0.88 0.54, 1.45 111 1.16

30 875 31 1.09 0.65, 1.84 1.08 0.63, 1.84 85 1.59 p for trend 0.74 0.78 0

Total calorie intake (kcal/week)< 10,166 774 19 1.00 Ref 1.00 Ref 57 1.00 10,166 < 12,762 776 15 0.77 0.39, 1.54 0.76 0.38, 1.51 57 1.08 12,762 < 15,895 775 35 1.87 1.06, 3.31 1.76 0.98, 3.16 70 1.36

15,895 774 31 1.72 0.96, 3.08 1.76 0.96, 3.21 85 1.66

p for trend 0.008 0.009 0* Odds ratios (ORs) were adjusted for age, province, sex, education, pack-years of smoking, alcohol drinking, exposure to some chemicals, and occupational exposure. Physic

(or lifetime maximum body mass index), and total calorie intake were adjusted for each other.y MET, metabolic equivalent task; CI, condence interval; Ref, referent.z Moderate and vigorous activity were adjusted for each other. Weight (kg)/height (m) 2 .

A m J E

pi d emi ol 2 0 0 5 ; 1 6 2 : 1 1 6 2 1 1 7 3

b y g u e s t o n M a y 1 5 , 2 0 1 4 h t t p : / / a j e . o x f o r d j o u r n a l s . o r g / D o w n l o a d e d f r o m

http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/http://aje.oxfordjournals.org/

8/12/2019 Am. J. Epidemiol.-2005-Pan-1162-73

9/12

antioxidant defense systems, increasing insulin sensitivity,and decreasing levels of insulin and insulin-like growth fac-tors (10, 11). Physical activity may also decrease NHL risk through its inuence on obesity. The greater risk reductionassociated with physical activity among obese men (but notwomen) observed in our study supports a possible role of physical activity in affecting NHL risk for men through itsinuence on obesity, and it also suggests that differentmechanisms might be involved for men and women regard-ing the association of recreational physical activity withNHL risk.

The mechanism for the link between NHL risk and obesityis not clear. One proposed hypothesis is the decreased im-mune response associated with obesity. Studies of immuno-logic function in obese humans and experimental animalsindicate that the excess adiposity is associated with impair-ments in host defense mechanisms (13, 4851). Differentanimal models of obesity have shown a decrease in allT-lymphocyte subsets and the B-cell population as well aslower lymphocyte responsiveness to different mitogens in

obese animals compared with lean ones (5255). Investiga-tions in humans also suggested a reduced number of subsetsof T cells and their functions in obese humans (56) or a lowercapacity of lymphocytes to proliferate in response to mito-gen activation (57). Research showed that energy restrictionor adequate weight reduction could restore the impaired im-mune response in overweight rats (58) or obese humans (56).

One recent population-based case-control studysuggestedthat the involvement of leptin and its receptor in the regula-tion of immune function might be one of the mechanismsunderlying the association between obesity and NHL (18).Leptin is a hormone primarily derived from adipocytes thatplays an important role in the regulation of food intake,

energy expenditure, and the control of body weight (59).Leptins weight-regulating effects are mediated throughthe binding and activation of its receptor (60). The case-control study by Skibola et al. (18) found that genetic poly-morphisms in the leptin and its receptor genes associatedwith an obese phenotype were associated with increasedrisks of NHL, and the authors suggested that genetic inter-actions between leptin and its receptor might promote theimmune dysfunction associated with the pathogenesis of lymphoma.

Another possible mechanism explaining the relation of obesity with NHL risk is the hypothesis that obesity cancause changes in the metabolism of endogenous hormones,including sex steroids, insulin, and insulin-like growth fac-tors, which could distort the normal balance between cellproliferation, differentiation, and apoptosis (61).

The stratied analyses by NHL subtype were exploratory,and we cannot explain the difference in association withphysical activity, obesity, and energy intake between sub-types. It could be due to subtype misclassication sincesubtyping was based on cancer registry pathology reportrather than review by a panel of pathologists.

Our study used the Working Formulation classicationsystem with a modication for NHL subtyping. In our study,the provincial cancer registry coded NHL by using Interna-tional Classication of Diseases for Oncology , Second Edi-tion, morphologic classication. We did not use the revised T

A B L E 4

. R i s k o f n o n - H o d g k i n s l y m

p h o m a a s s o c i a t e d w i t h t o t a l r e c r e a t i o n a l p h y s i c a l a c t i v i t y

, b y b o d y m a s s i n d e x a n d s e x

, N a t i o n a l E n h a n c e d C a n c e r S u r v e i l l a n c e S y s t e m

,

C a n a d a

, 1 9 9 4

1 9 9 7

P h y s i c a l

a c t i v i t y l e v e l

( M E T * - h o u r s /

w e e k )

B o d y m a s s i n d e x y 1 8

. 5