7
Vol. 6, 557-563, August 1997 Cancer Epidemiology, Biomarkers & Prevention 557 3 The abbreviations used are: BMI, body mass index; IGF-I, insulin-like growth factor-I; RE, relative risk; PSA, prostatic specific antigen; CI, confidence interval. Height, Body Weight, and Risk of Prostate Cancer’ Edward Giovannucci,2 Eric B. Rimm, Meir J. Stampfer, Graham A. Colditz, and Walter C. Willett Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital [E. G., E. B. R., M. J. S., G. A. C., W. C. W.], and Departments of Nutrition [E. G., E. B. R., M. J. S., W. C. W.] and Epidemiology [E. B. R., M. J. S., G. A. C., W. C. WI, Harvard School of Public Health, Boston, Massachusetts 02115 Abstract Using data from the Health Professionals Follow-Up Study, we prospectively examined the relationships between height, body mass index, waist and hip circumferences, and risk of total and advanced (extraprostatic and metastatic) prostate cancer. In addition, we assessed adiposity during childhood, adolescence, and early, middle, and late adulthood using pictograms in relation to prostate cancer risk. Between 1986 and 1994, 1,369 cases of prostate cancer (excluding stage Al) were confirmed in 47,781 men. Adult body mass index and waist and hip circumferences were not appreciably related to risk of total prostate cancer or advanced prostate cancer. In contrast, preadult (age 10) obesity assessed in 33,336 men in 1988 was prospectively related to lower risk of advanced [relative risk (RR) = 0.72 with 95% confidence interval (CI) = 0.47-1.10, between high and low quintiles; = 0.06] and metastatic prostate cancer (HR 0.38 with 95% CI 0.19-0.77; ‘trend 0.004). For the advanced lesions, an association was observed with height (RR = 1.68 with 95% CI = 1.16-2.43 for men 74 inches or taller, relative to men 68 inches or shorter; nd 0.01). In an analysis limited to particularly aggressive forms of prostate cancer, i.e., cases found to be metastatic at time of diagnosis between 1988 and 1994 after a negative digital rectal examination in 1988, we found that obesity at ages 5 and 10 had a strong inverse association (RR = 0.16 with 95% CI = 0.05-0.54, between high and low quintiles at age 10) and that tallness had a strong direct association with risk of metastatic disease (RR = 2.29 with 95% CI = 1.04-5.05, for height 74 inches versus 68 inches). Our findings suggest that the preadult hormonal milieu, as reflected in attained height and childhood obesity, may have a strong influence on prostate carcinogenesis. Introduction In men, obesity is associated with various endocrine aberra- tions, including higher serum estrogen and lower testosterone levels (1-6). On the basis that this hormonal pattern may decrease risk of prostate cancer (7), a lower rate of this malig- nancy may be predicted among heavier men. On the other hand, increased body weight has been linked to higher risk of several malignancies. Studies of obesity in relation to prostate cancer risk have yielded inconsistent results. Case-control studies have not generally supported an association between BMI3 and risk of prostate cancer (8-14), but one study found a direct associ- ation (15). Prospective data on BMI and risk of prostate cancer have been conflicting, with some studies supporting a direct association (16, 17) and others finding no relationship (18, 19). BMI is correlated with both adiposity and lean body mass (20). The results from one prospective study (21) indicated that a direct association between BMI and risk of prostate cancer was due to lean body mass rather than adiposity, suggesting that BMI may not be an unconfounded measure of obesity in this circumstance. The studies examining obesity and prostate cancer have focused on adult BMI. The prostate gland is essentially dormant until puberty, when hormonal changes stimulate its develop- ment. Recent studies demonstrating that prostatic intraepithelial neoplasia, a cancer precursor lesion, is already common among men in their twenties (22, 23) suggest that prostatic carcino- genesis begins early. The influence of hormones may be par- ticularly strong during puberty and adolescence, when the pros- tate achieves full development. Attained height and preadult obesity may reflect events that occur during this period, char- acterized by profound changes in levels of sex hormones, growth hormone, and IGF-I. Taller individuals are at higher risk of several malignancies, including colon and breast cancer (24, 25), but the available data examining height as a risk factor for prostate cancer are conflicting; some studies support an asso- ciation (26-28) whereas others do not (14, 19). Preadult adi- posity has not been studied. Using data from the Health Professionals Follow-Up Study, we examined the relationships between height, BMI, and risk of prostate cancer occurrence and progression. In addition, we examined the RR of prostate cancer associated with waist and hip circumferences, indicators of fat distribution, and meas- ures of adiposity less confounded with lean body mass. We also used adiposity level during childhood, adolescence, and early adulthood to examine the relationship between body mass at the period of prostate development and prostate cancer risk. Received 2/14/97; revised 4/28/97; accepted 5/3/97. 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 This work was supported by NIH Grants CA 55075 and HL 35464, Special Institution Grant 18 from the American Cancer Society, and a Faculty Research Award (FRA-398) from the American Cancer Society (to 0. A. C.). 2 To whom requests for reprints should be addressed, at Channing Laboratory, Department of Medicine, Harvard Medical School and Brigham and Women’s Hospital, 181 Longwood Avenue, Boston, MA 02115. Patients and Methods Study Population. The Health Professionals Follow-Up Study is an ongoing prospective cohort study of the causes of cancer Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 American https://bloodcancerdiscov.aacrjournals.org Downloaded from

Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

  • Upload
    others

  • View
    1

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

Vol. 6, 557-563, August 1997 Cancer Epidemiology, Biomarkers & Prevention 557

3 The abbreviations used are: BMI, body mass index; IGF-I, insulin-like growthfactor-I; RE, relative risk; PSA, prostatic specific antigen; CI, confidence interval.

Height, Body Weight, and Risk of Prostate Cancer’

Edward Giovannucci,2 Eric B. Rimm, Meir J. Stampfer,Graham A. Colditz, and Walter C. Willett

Channing Laboratory, Department of Medicine, Harvard Medical School and

Brigham and Women’s Hospital [�E. G., E. B. R., M. J. S., G. A. C., W. C. W.],and Departments of Nutrition [E. G., E. B. R., M. J. S., W. C. W.] andEpidemiology [E. B. R., M. J. S., G. A. C., W. C. WI, Harvard School ofPublic Health, Boston, Massachusetts 02115

Abstract

Using data from the Health Professionals Follow-UpStudy, we prospectively examined the relationshipsbetween height, body mass index, waist and hip

circumferences, and risk of total and advanced(extraprostatic and metastatic) prostate cancer. Inaddition, we assessed adiposity during childhood,

adolescence, and early, middle, and late adulthood usingpictograms in relation to prostate cancer risk. Between1986 and 1994, 1,369 cases of prostate cancer (excluding

stage Al) were confirmed in 47,781 men. Adult bodymass index and waist and hip circumferences were notappreciably related to risk of total prostate cancer oradvanced prostate cancer. In contrast, preadult (age 10)obesity assessed in 33,336 men in 1988 was prospectively

related to lower risk of advanced [relative risk (RR) =

0.72 with 95% confidence interval (CI) = 0.47-1.10,between high and low quintiles; � = 0.06] andmetastatic prostate cancer (HR 0.38 with 95% CI0.19-0.77; ‘�trend 0.004). For the advanced lesions, anassociation was observed with height (RR = 1.68 with95% CI = 1.16-2.43 for men 74 inches or taller, relativeto men 68 inches or shorter; ��nd 0.01). In ananalysis limited to particularly aggressive forms ofprostate cancer, i.e., cases found to be metastatic at time

of diagnosis between 1988 and 1994 after a negativedigital rectal examination in 1988, we found that obesityat ages 5 and 10 had a strong inverse association (RR =

0.16 with 95% CI = 0.05-0.54, between high and lowquintiles at age 10) and that tallness had a strong directassociation with risk of metastatic disease (RR = 2.29with 95% CI = 1.04-5.05, for height �74 inches versus

�68 inches). Our findings suggest that the preadulthormonal milieu, as reflected in attained height andchildhood obesity, may have a strong influence onprostate carcinogenesis.

Introduction

In men, obesity is associated with various endocrine aberra-tions, including higher serum estrogen and lower testosterone

levels (1-6). On the basis that this hormonal pattern maydecrease risk of prostate cancer (7), a lower rate of this malig-nancy may be predicted among heavier men. On the other hand,increased body weight has been linked to higher risk of several

malignancies. Studies of obesity in relation to prostate cancerrisk have yielded inconsistent results. Case-control studies havenot generally supported an association between BMI3 and risk

of prostate cancer (8-14), but one study found a direct associ-ation (15). Prospective data on BMI and risk of prostate cancerhave been conflicting, with some studies supporting a direct

association (16, 17) and others finding no relationship (18, 19).BMI is correlated with both adiposity and lean body mass (20).

The results from one prospective study (21) indicated that adirect association between BMI and risk of prostate cancer wasdue to lean body mass rather than adiposity, suggesting thatBMI may not be an unconfounded measure of obesity in thiscircumstance.

The studies examining obesity and prostate cancer havefocused on adult BMI. The prostate gland is essentially dormant

until puberty, when hormonal changes stimulate its develop-ment. Recent studies demonstrating that prostatic intraepithelialneoplasia, a cancer precursor lesion, is already common amongmen in their twenties (22, 23) suggest that prostatic carcino-genesis begins early. The influence of hormones may be par-ticularly strong during puberty and adolescence, when the pros-

tate achieves full development. Attained height and preadultobesity may reflect events that occur during this period, char-acterized by profound changes in levels of sex hormones,growth hormone, and IGF-I. Taller individuals are at higher risk

of several malignancies, including colon and breast cancer (24,25), but the available data examining height as a risk factor forprostate cancer are conflicting; some studies support an asso-ciation (26-28) whereas others do not (14, 19). Preadult adi-posity has not been studied.

Using data from the Health Professionals Follow-UpStudy, we examined the relationships between height, BMI, andrisk of prostate cancer occurrence and progression. In addition,we examined the RR of prostate cancer associated with waist

and hip circumferences, indicators of fat distribution, and meas-ures of adiposity less confounded with lean body mass. We alsoused adiposity level during childhood, adolescence, and earlyadulthood to examine the relationship between body mass at the

period of prostate development and prostate cancer risk.

Received 2/14/97; revised 4/28/97; accepted 5/3/97.

The costs of publication of this article were defrayed in part by the payment ofpage charges. This article must therefore be hereby marked advertisement in

accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1 This work was supported by NIH Grants CA 55075 and HL 35464, Special

Institution Grant 18 from the American Cancer Society, and a Faculty Research

Award (FRA-398) from the American Cancer Society (to 0. A. C.).2 To whom requests for reprints should be addressed, at Channing Laboratory,

Department of Medicine, Harvard Medical School and Brigham and Women’sHospital, 181 Longwood Avenue, Boston, MA 02115.

Patients and Methods

Study Population. The Health Professionals Follow-Up Studyis an ongoing prospective cohort study of the causes of cancer

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 2: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

558 Height, Body Weight, and Prostate Cancer

and heart disease in men (29). The cohort consists of 51,529

American male dentists, optometrists, osteopaths, pharmacists,podiatrists, and veterinarians who were 40-75 years old andresponded to a mailed questionnaire in 1986. We elicited in-

formation on age, marital status, height and weight, ancestry,medications, disease history, physical activity, and diet. The

assessment of anthropometric measures (30), diet (3 1), andphysical activity (32) have been validated in substudies withinthe cohort. To form the analytic cohort, we excluded 2218 menwho reported cancer at baseline (other than nonmelanoma skin

cancer). To be able to control for dietary factors, we also

excluded 1,530 men who did not adequately complete thedietary questionnaire [they left 70 or more items blank orreported intake of more than 17,600 joules (4,200 kcal) or less

than 3,350 joules (800 kcal) per day]. After these baselineexclusions, 47,781 participants remained.

Follow-Up of the Cohort. Follow-up questionnaires were sentin 1987, 1988, 1990, 1992, and 1994 to ascertain new cases ofa variety of diseases and to update and expand exposure infor-mation. In 1994, the participant was asked whether he hadreceived a PSA examination. Most of the deaths in the cohortwere reported by family members or by the postal system in

response to the follow-up questionnaires, which included cer-

tified mail to nonresponders. In addition, we searched theNational Death Index, a highly sensitive method (33), to iden-

tify deaths among nonresponders. Because the age-adjustedincident rate of prostate cancer in our cohort was quite com-

parable to United States rates (actually 6% higher in our co-hort), based on the Surveillance, Epidemiology, and End Re-

sults (SEER) Program data from 1986 to 1989 beforewidespread PSA screening, we are likely to have detected the

vast majority of diagnosed prostate cancers in our cohort.

Assessment of Anthropometric Variables. Men reported

their current weight, height, and weight at age 21 on the 1986baseline questionnaire. In 1987, we mailed a supplementaryquestionnaire to all cohort members to obtain additional expo-sure information, including circumference measures, not in-cluded on the 1986 baseline questionnaire. We instructed the

participant to measure (to the nearest quarter of an inch) hiswaist at the umbilicus and his hips at the largest circumference

between the waist and thighs while standing and avoiding bulky

clothing (30). Along with the questionnaire, we provided a tapemeasure and an illustration to standardize the measurements.We used BMI (kg/m2) as a measure of total adiposity, waist:hip

ratio as a measure of relative distribution of fat, and waist

circumference to estimate total abdominal fat (34). We alsoevaluated weight change during adulthood, assessed as weightin 1986 minus weight at age 21.

We evaluated the precision of self-reported anthropomet-

tic measures among a sample of the cohort members who were

part of a dietary validation study (30). Briefly, trained techni-cians visited randomly selected Boston-area cohort members of

similar age distribution as the entire cohort twice, approxi-mately 6 months apart, to measure current weight and waist andhip circumferences. The Pearson correlation between self-re-

port and the average of the technicians’ two measurements was

0.97 for weight, 0.95 for waist circumference, 0.88 for hipcircumference, and 0.69 for waist:hip ratio. Relative to the

technicians’ measurements, the men slightly overestimated

their waist (0.36 in) and underestimated their hip circumference

(0.78 in) and weight (2.3 pounds).To estimate level of adiposity during different periods

throughout life, we included a section on the 1988 question-naire that depicted nine illustrations of body builds or figures

ranging from thin to obese and asked the participant to select

the figure that best depicted his body build at each of the

following ages: 5, 10, 20, 30, 40, and current age (35). Theremote recall of childhood weight and body build by elderly

subjects using these pictograms has been previously shown inanother population to provide useful information in ranking

individuals (36). For example, in 71-76-year-old individuals,the correlations between their recalled preadult body build and

direct measures of the BMI that had been taken at the corre-sponding time period were mostly in the range of 0.5-0.7 (36).

Identification of Cases of Prostate Cancer. On the 1988,1990, 1992, and 1994 questionnaires, we inquired about a

diagnosis of prostate cancer during the prior 2 years. Upon a

report, we asked the man for written permission to obtainhospital records and pathology reports. For deceased men, we

contacted next of kin. After repeated mailings, we received

questionnaires or confirmed deaths of more than 96% of eligi-

ble participants in 1988 and in 1990, 94% in 1992, and 90% in1994, thus accounting for 94% of the total possible person-

years of follow-up. We estimate having ascertained over 98%of the deaths in this cohort. From 1986 to the end of this studyperiod (January 31, 1994), we identified 141 1 incident cases of

prostate cancer. Of these, 1226 cases (87%) were confirmed bymedical records, 150 (11%) additional men provided informa-

tion regarding the basis of diagnosis and subsequent treatment,

and 35 (2%) could not be recontacted. When we obtainedmedical records and pathology reports, a diagnosis of adeno-

carcinoma of the prostate was confirmed in 99% of the cases.Because of the high accuracy of reporting prostate cancer, we

analyzed all men who reported a diagnosis of prostate cancer,

even if we could not review medical records.

A study physician staged prostate cancers according toinformation received from medical reports [stage A, occult or

incidental finding (Al, focal; A2, diffuse); stage B, confined toprostate gland; stage C, localized to periprostatic area; stageDl, metastatic disease involving only regional lymph nodes;and stage D2, those that have metastasized to other organs].

This classification criteria included information from anywork-up during the initial diagnosis, including staging prosta-

tectomy and bone scans. Prostatectomy could result in upstag-ing of apparent stage B cancers (tumors with penetration

through the capsule move to stage C, and those with lymphnode metastases move to stage Dl). Because approximately

60% of the men received a prostatectomy, some advancedcancers may have been missed.

From 1986 to 1989, the age-specific incidence rates ofprostate cancer were constant, and the vast majority of caseswere initially detected via routine rectal examination, due to

surgery for benign prostatic hyperplasia or due to symptoms.

The age-specific incidence rate approximately doubled in the

subsequent 4-year period (1990-1993), presumably because ofwider use of semm PSA level for screening. By January 1994,

almost 80% of men older than 60 years had received a PSAexamination.

Data Analysis. Each of the participants was followed begin-

ning on the month of return of the baseline questionnaire and

ending on the month of diagnosis of prostate cancer, month ofdeath from other causes, or the end of the study period. Forexposures assessed initially in follow-up questionnaires (cir-

cumferences in 1987, and preadult pictograms in 1988), fol-low-up time accrued beginning on the month of return of thesequestionnaires. We calculated incidence rates of prostate cancer

for men in a specific category (of body mass, height, etc.) bydividing the number of incident cases by the number of person-

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 3: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

Cancer Epidemiology, Bloinarkers & Prevention 559

Table 1 RR and 95 % CI of prostate cancer by level o f BMI in 1986 in h ealth professio nals (1986-1994)

BMI (kg/m2)

�trend<23#{176} 23-23.9 24-24.9 25-25.9 26-26.9 27-28.9 �29

Person-years 62,592 50,390 46,836 58,987 34,156 50,165 40,712

Total prostate cancer (n) 230 223 202 229 123 205 126

Age-adjusted RR 1.0 1.21 1.15 1.00 0.92 1.06 0.86 0.05

Multivariate p,p�b 1 .0 1 .25 1 .20 1.05 0.94 1 . I 1 0.90 0.46

95% CI 1 .03-1 .5 1 0.98-1 .46 0.87-1 .28 0.74-1 . 1 8 0.90-1 .36 0.71-1.15

Advanced prostate cancer (n) 74 69 59 66 32 64 43

Age-adjusted RE 1.0 1.15 1.03 0.88 0.73 1.03 0.90 0.23

Multivariate p�p�b 1.23 1.18 1.01 0.89 1.22 1.19 0.16

95% CI 0.88-1.72 0.83-1.68 0.71-1.43 0.58-1.36 0.85-1.76 0.79-1.82

Metastatic prostate cancer (n) 43 32 21 33 18 28 19

Age-adjusted RR 1.0 0.93 0.64 0.77 0.72 0.79 0.69 0.14

Multivariate pj�b 1 .0 0.99 0.7 1 0.84 0.88 0.91 0.95 0.29

95% CI 0.62-1.57 0.42-1.21 0.52-1.36 0.50-1.55 0.55-1.53 0.52-1.73

a Reference group.b Adjusted for age, height, and BMI at age 21 by multiple logistic regression.

years. The RR was computed as the rate among men in aspecific category divided by the rate among men in a specifiedreference category. We used the Mantel-Haenszel summaryestimator (37) to adjust for age (across 2.5-year categories), and

multivariate logistic regression with repeated measures (38)biennially to estimate RRs when controlling simultaneously formore than one risk factor. We tested for linear trends using theMantel-Haenszel extension test for trend in the age-adjusted

analysis (37) and, when controffing for multiple covariates, bymodeling the specific exposure as a continuous variable in a

logistic model which included the covariates. All reported Psare two-tailed.

Stage Al lesions are relatively innocuous and are espe-cially prone to detection bias because they are found inciden-tally in prostatectomy specimens for benign prostatic disease;we excluded these (45 cases, or 3% of the total) and limited our

analysis to the 1369 non-stage Al cases. Previous studies found

that some factors may be related primarily to risk of “aggres-sive” or advanced prostate cancers, such as cancers that have

already metastasized when diagnosed. We thus conducted ad-ditional analyses limited to the 423 advanced cases, defined asthose diagnosed at stage C or D (extraprostatic), and 201metastatic cases (stage D) only. Of the remaining cases, 668were organ-confined (stage A2 or B) and 278 (19.7% of total)were of undetermined stage.

The entire analytic cohort consisted of 47,781 men. Therewere missing data for adult BMIs for 1087 men, and 2138 men

did not provide a BMI at age 21 years. Among the 47,781 men,30,966 provided data on waist and hip circumferences in 1987,and 33,336 provided data on pictograms in 1988. Because the

information on circumference measures and pictograms wasalways used prospectively in relation to diagnosis of prostate

cancer, these analyses should not be prone to selection or recallbiases. Multivariate models included only men who had com-plete data for all covariates in the model.

Results

BMI in 1986 was unrelated to risk of prostate cancer (Table 1).A slightly lower risk among the heaviest men, suggested in theage-adjusted analysis, was attenuated when we controlled ad-ditionally for height and BMI at age 21. The attenuation wascaused primarily by controlling for BMI at age 21. No associ-

ation was noted between BMI and extraprostatic or metastaticcases. To eliminate the possibility that prediagnostic weight

loss among cases obscured an association, we excluded the first

2 years of follow-up; these analyses yielded similar results asthose for the entire population for total prostate cancer (RR =

0.82, 95% CI = 0.65-1.04 for BMI �29, relative to <23,

adjusted for age; ‘�trcnd 0.06) or advanced prostate cancer(RR = 0.79, 95% CI = 0.52-1.21; P� = 0.19). As for theoverall analysis, the slight suggestion of an inverse association

did not persist when we controlled for BMI at age 21. Adultweight gain (weight in 1986 minus weight at age 21) was not

associated with total prostate cancer (RR = 0.89, 95% CI =

0.69-1.16 for a weight gain of 40 pounds or more relative to

weight maintenance within 5 pounds, controlling for age and

BMI at age 21; 1�trend 0.35) nor with advanced prostate

cancer (RR = 0.98, 95% CI = 0.64-1.49; ��nd 0.79). Thenull relation with adult weight gain did not vary by age.

In contrast to BMI in 1986, a higher BMI at age 21 was

strongly related with a decreased risk of advanced and meta-

static prostate cancer, adjusting for height and BMI in 1986

(Table 2). Further adjustment for dietary variables, vasectomy,

smoking, alcohol, and race did not alter the RRs. Total prostate

cancer was not appreciably related to BMI at age 21.We next examined height in 1986 in relation to risk of

prostate cancer (Table 3). For total prostate cancer, we found a

slight direct association between height and risk of prostate

cancer, although the association was not statistically significant(P = 0. 12) after adjusting for BMI in 1986 and at age 21 . For

the advanced or metastatic lesions, a stronger positive associ-

ation was noted: men who were �74 inches tall were at 60-70% greater risk of being diagnosed with the more advanced

forms of the disease than were men who were �68 inches tall.We examined waist and hip circumference measures as-

sessed in 1987 in relation to prostate cancer risk. Among menwho met inclusion criteria, 30,966 provided data on this in

1987. No appreciable association was noted for waist circum-

ference, but an inverse association was noted between hip

circumference and total, advanced, and metastatic prostate can-

cer (Table 4). The inverse association between hip circumfer-

ence and risk of metastatic prostate cancers persisted when weadded BMI in 1986 to the model (RR = 0.47 with 95% CI =

0.22-1.01, between high and low quintiles; 1�trend 0.05), butit became attenuated and nonsignificant when we controlled for

BMI at age 21 years (RR = 0.70 with 95% CI = 0.39-1.25;

‘�trend 0.23).We explored further the inverse association between obe-

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 4: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

560 Height, Body Weight, and Prostate Cancer

Table 2 RE and 95% CI of prostate cancer by BMI at age 21 in health professionals (1986-1994)

BMI (kg/m2) at age 21��rcnd

<21Y’ 20-21 .9 22-22.9 23-23.9 24-25.9 �26

Person-years 42,419 80,683 48,028 56,825 68,620 39,597

Total prostate cancer (n) 229 353 188 200 223 104

Age-adjusted RR 1 .0 0.98 1 .00 1 .03 0.96 0.82 0.12

Multivariate p,p�b 1 .0 0.98 1 .00 1 .03 1 .00 0.87 0.60

95% CI 0.83-1.16 0.82-1.22 0.84-1.26 0.82-1.22 0.67-1.12

Advanced prostate cancer (n) 8 1 1 17 59 56 60 26

Age-adjusted RE 1.0 0.91 0.88 0.79 0.72 0.58 0.003

Multivariate RRb 1 .0 0.91 0.88 0.77 0.7 1 0.53 0.006

95% Cl 0.69-1.22 0.62-1.24 0.54-1.10 0.50-1.02 0.33-0.86

Metastatic prostate cancer (n) 45 53 32 27 20 14

Age-adjusted RR 1 .0 0.74 0.87 0.69 0.45 0.59 0.004

Multivariate RRb 1.0 0.78 0.92 0.72 0.46 0.57 0.01

95% CI 0.52-1.16 0.58-1.47 0.44-1.20 0.27-0.81 0.29-1.10

a Reference group.b Adjusted for age, height, and BMI in 1986 by multiple logistic regression.

Table 3 RR and 95% CI of prostate cancer by adult height among health professionals (1986-1994)

Height (inches)

‘�trend�68” 69 70 71 72 73 �74

Person-years 93,050 41,560 60,660 47,690 47,944 25,702 34,918

Total prostate cancer (n) 417 179 241 180 149 91 112

Age-adjusted RR 1.0 1.09 1.09 1.12 1.02 1.23 1.34 0.01

Multivariate pp�b 1.0 1.09 1.07 1.08 0.98 1.22 1.37 0.12

95% CI 0.91-1.30 0.91-1.27 0.90-1.30 0.80-1.19 0.97-1.55 1.10-1.70

Advanced prostate cancer (n) I 17 54 67 61 50 34 40

Age-adjusted RE I .0 1 . 16 1 .06 1 .32 1 . 19 1 .56 1 .62 0.002

Multivariate pj�b 1.0 1.14 1.06 1.29 1.1 1 1.63 1.68 0.01

95% Cl 0.81-1.59 0.77-1.45 0.93-1.79 0.78-1.58 1.10-2.41 1.16-2.43

Metastatic prostate cancer (n) 57 28 36 27 22 14 17

Age-adjusted RE 1.0 1.25 1.18 1.24 1.09 1.37 1.48 0.16

Multivariate �p�b 1.0 1.18 1.15 1.14 0.95 1.42 1.48 0.32

95% CI 0.74-1.87 0.75-1.78 0.70-1.84 0.56-1.61 0.79-2.57 0.85-2.56

a Reference group.

F, Adjusted for age and BMI in 1986 and at age 21 by multiple logistic regression.

sity at a younger age and risk of prostate cancer by using theself-assessment of body adiposity. In 1988, the participantsreported their body build at age 5, 10, 20, 30, 40, and age in1988 using a series of nine pictograms, which ranged subjec-tively from very lean to very obese. These self-reported assess-ments were used to form approximate quintiles or quartiles(when the distribution was narrow) for the various ages. Higher

reported level of adiposity at ages 5, 10, and 20 years wasassociated with a lower risk of prostate cancer, particularly for

metastatic lesions (Table 5). Specifically, men who in 1988reported greater adiposity at ages 10 and 20 had approximately

one-third of the risk of advanced prostate cancer as the leanestmen. Because of high correlations among the body build as-sessments at ages 5, 10, and 20, it is difficult to separate theindependent effect of obesity for each age (Spearman correla-tion = 0.92 between adiposity at ages 5 and 10, 0.80 between

5 and 20, and 0.87 between 10 and 20). Nonetheless, when wemodeled various combinations of adiposity at the different

ages, obesity at age 10 was the strongest predictor of reducedrisk (RR = 0.41, with 95% CI = 0.14-1.25, when adjusted foradiposity at age 5, and RR = 0.18, with 95% CI = 0.18-0.96,when adjusted for adiposity at age 20). There were suggestiveinverse associations also with measures of adiposity at ages 30

and 40. However, when controlled for adiposity at age 10, the

relationships with adiposity at ages 30 and 40 did not persist,whereas the inverse association at age 10 remained significant.

We conducted an analysis using the 1988-1994 follow-upamong men who had no prostate cancer at baseline in 1988 butreported having had a recent digital rectal examination on the1988 questionnaire, the only common screening test for pros-tate cancer at the time. We would expect that newly diagnosedcases of metastatic prostate cancer would represent particularly

aggressive forms of the disease because they did not have

clinically detectable disease in 1988. In this analysis, we foundthat obesity at age 10 had a strong inverse association (RR =

0.16 with 95% CI = 0.05-0.54, between high and low quin-tiles), and tallness had a strong direct association with risk of

metastatic disease (RR = 2.29 with 95% CI = 1.04-5.05, forheight of �74 inches versus �68 inches). Neither obesity norheight predicted organ-confined (stage A or B) disease (RR =

0.89 with 95% CI = 0.63-1.26, and RR = 1.16 with 95% CI =

0.77-1.76, for obesity at age 10 and height, respectively).

Discussion

Tallness was a marker of increased risk, whereas greater adi-posity between the ages of 5 to 20 years was associated with a

lower risk of prostate cancer, particularly for advanced disease.

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 5: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

Cancer Epidemiology, Biomarkers & Prevention 561

Table 4 RRU of prostate cancer by assessment of waist and hip circumference in the Health Professional Follow-Up Study (1987-1994)”

Quintiles 95% CI’

�I 2 3 4 5 QSvs.Ql

Waist (inches) �34.25 34.5-36 36.25-37.75 38-40 �40.25

Total prostate cancer 1.0 1.20 0.99 1.01 0.99 (0.78-1.26) 0.26

Advanced prostate cancer 1 .0 1 . 17 1 .03 1 .07 1 . 12 (0.73-1 .7 1 ) 0.96

Metastatic prostate cancer I .0 0.89 0.79 1 .03 0.87 (0.47-1 .63) 0.92

Hips (inches) �37.5 37.75-38.75 39-40 40.25-41.75 �42

Total prostate cancer 1.0 0.98 0.91 0.82 0.85 (0.68-1.06) 0.04

Advanced prostate cancer 1.0 1.10 0.92 0.84 0.82 (0.54-1.24) 0.13

Metastatic prostate cancer 1.0 0.70 0.74 0.42 0.62 (0.34-1.12) 0.04

Waist:hip ratio �0.89 0.90-0.92 0.93-0.95 0.96-0.98 �0.98

Total prostate cancer 1.0 0.90 1.03 0.97 0.96 (0.77-1.20) 0.97

Advanced prostate cancer I .0 0.90 1 .02 1 .05 1 .05 (0.7 1-1 .56) 0.48

Metastatic prostate cancer I .0 0.98 1 .03 1 .21 1 .58 (0.87-2.89) 0.05

a j�j� adjusted for age and height.

b Among 30,966 men who completed questionnaire in 1987 (948 total cases).

‘ 95% CI for quintile 5 vs. quintile 1.d Mantel extension test for trend across quintiles.

Table 5 Age-adjusted RR and 95% CI for high category of obesity relative to low category in the Health Professional Follow-Up Study (1988-1994)”

Age (years)In 1988

5 10 20 30 40

Total prostate cancer

RR (95% CI)’ 0.80 (0.64-1.01) 0.81 (0.64-1.02) 0.78 (0.61-1.01) 0.84 (0.67-1.06) 0.79 (0.62-1.02) 1.1 1 (0.88-1.39)

���trend 0.02 0.01 0.06 0.04 0.08 0.73Advanced prostate cancer

RR (95% CI)’ 0.77 (0.52-1.16) 0.72(0.47-1.10) 0.62 (0.38-1.02) 0.89(0.58-1.38) 0.75 (0.46-1.24) 1.27 (0.82-1.97)

pdd 0.08 0.06 0.12 0.47 0.84 0.32Metastatic prostate cancer

RR (95% CI)’ 0.52 (0.26-1.02) 0.38 (0.19-0.77) 0.31 (0.12 0.76) 0.44 (0.23-0.84) 0.46 (0.21-1.02) 0.96(0.50-1.84)p�d 0.02 0.004 0.01 0.05 0.16 0.52

a Based on self-reported pictograms of body size at various ages as indicated, assessed in 1988.

b Among 33,336 men free of cancer who completed questionnaire in 1988 (922 total cases).

‘ RR for high quintile or quartile of level of adiposity relative to low quintile or quartile.d Mantel extension test for trend across increasing level of adiposity.

In contrast, anthropometric measures of adult obesity were notindependently related to risk of this malignancy. The reported

measures of adult BMI, height, and waist and hip circumferencewere quite precise (30). The self-reported anthropometric meas-

ures of earlier periods in life using the body build illustrationshave been found to provide useful information, but they havesome degree of measurement error, with correlations with di-rect measures ranging mostly between 0.5 and 0.7 (36). Be-

cause of the prospective design, measurement error is likely tobe nondifferential with regard to outcome, and thus, the rela-

tionships we observed with preadult obesity probably were

attenuated by measurement error.BMI, which is correlated with lean body mass as well as

fat mass (39), is a less than ideal measure of adiposity to study

the androgen-sensitive prostate because muscle mass is related

to androgens (21). However, in our study, adult circumferencemeasures, which may be more strongly related to endocrineabnormalities typically associated with obesity (1), were notappreciably related to prostate cancer risk. A lower risk ofprostate cancer was suggested only with greater hip circumfer-

ence, although this relationship became attenuated and was notstatistically significant when adjusted for BMI at age 21. Ofnote, in one small study, an inverse correlation with free tes-

tosterone was stronger for hip circumference (r = -0.78) thanfor waist circumference (r = -0.64), waist:hip ratio (r =

-0.05), or BMI (r = -0.47; Ref. 1). Previous studies haveinconsistently found BMI to be associated with a higher risk ofprostate cancer, but in at least one study, this was due to the leanmass component of BMI. The overall data, including our re-sults, do not suggest adult adiposity correlates with higher risk

of prostate cancer. Possibly, the variation in results could resultin part to population differences (e.g., due to age range of thespecific population) that determine whether BMI correlatesmore strongly with lean body mass or with adiposity.

The relationship between height and prostate cancer risk issupported by most, but not all, studies. Four recent studies,including ours (26-28), have found a moderately higher risk ofprostate cancer among taller men. Two case-control studies

conducted in the early l970s reported no association betweenheight and risk of this cancer (14, 19). However, inspection of

one of these studies (from Fig. 8 in Ref. 14), which did notpresent odds ratios, showed an excess of cases relative to

controls among men S feet, 8 inches tall (the median) or taller.Data examining other preadult exposures and risk of

prostate cancer are sparse. A case-control study in Sweden(27) found that men who were less than 14 years of age atonset of beard growth were at higher risk of prostate cancer,and the authors also found a direct association between

height and prostate cancer risk. These findings on earlypubertal development and height are in general concordance

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 6: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

562 Height, Body Weight, and Prostate Cancer

with our results. The Swedish study found only a weak

inverse association between BMI at age 20 and prostatecancer risk, an association that did not persist in multivariate

analysis. However, that study had limited range in BMIbecause only 3 kg/m2 separated upper and lower categories,in comparison to 6 kg/m2 in our study, and it did not report

on the risk for more advanced lesions. This limited range inBMI may have reflected less overall obesity, which wouldrender BMI more a measure of lean body mass than ofadiposity. One case-control that assessed various physical

characteristics believed to be related to sex hormones duringphysical examinations in college men did not find any ob-vious relation with prostate cancer risk (19). However, be-

cause these men were of college age, events around the timeof puberty may not have been adequately assessed.

Our findings regarding anthropometric measures may berelated to the profound hormonal changes that begin at puberty.Height is determined by genetic and nutritional factors, medi-

ated largely through the endocrine system. At puberty, levels ofboth testosterone and IGF-I rise dramatically. Although thedeterminants of height are complex, men who attain a greaterheight may have been exposed to higher levels of IGF-I andtestosterone during a critical growth period in general and for

the prostate gland specifically. Obesity in males is associatedwith endocrine aberrations, including higher estrogen and lower

testosterone levels (1-6). Lower level of growth hormone, amajor determinant of IGF-I levels, has also been associatedwith both preadult and adult obesity (40-42). These hormonal

changes may reflect either a cause or a consequence of obesity,

but some of these changes are at least partially reversiblethrough weight loss (42).

The prostate gland is essentially dormant until puberty,

when a complex interaction between sex and other growthhormones, as well as their binding proteins, induces its devel-opment. Prostatic growth is controlled by serum testosterone,which is converted intracellularly by 5-a-reductase to dihy-

drotestosterone (43). Androgens may sensitize hormone-depen-

dent tissue to the effects of IGF-I and may increase IGF-Iproduction at the time of puberty (44). Prostate epithelial cellshave IGF-I receptors, and IGF-I stimulates mitogenicity in

prostate cell lines (45-48) and increases activity of 5-a-reduc-tase (49). By stimulating prostatic epithelial cell division, thesesame hormonal factors may be critical in carcinogenesis. Thus,prostate cancer occurs rarely in castrated men (50), and theprolonged administration of high levels of testosterone induces

prostate cancer in rats (5 1, 52). Moreover, higher testosteronelevels increase risk of prostate cancer, whereas higher levels of

sex hormone binding globulin, which renders testosterone in-accessible, may be protective (7).

Relationships among preadult obesity, testosterone levels,

and development of gonads and secondary sex organs aresupported in animal studies. Rats that are homozygous for thecorpulent gene (cp/cp), characterized by marked hyperphagia,obesity, and hyperinsulinemia (53), may be a relevant model.Male cp/cp rats demonstrate delayed puberty, marked reductionin the characteristic rise of testosterone levels at puberty, anddecreased organ weights for testes and seminal vesicles. Bysexual maturity, the weight of the testes catches up to normal,but the levels of testosterone remain about only one-third thatof normal levels. Moreover, seminal vesicle size also remainsabnormally small.

Inadequate nutritional intake before adulthood will stuntoverall growth and organ cellularity (54). The members of

populations that experience some degree of energy restrictionduring the preadult period would tend to be shorter on average

and to experience lower stimulation from growth-enhancinghormones, such as IGF-I (55). On the basis of internationalcorrelational data, these populations are at lower risk of prostate

cancer mortality (56). Because our population of health profes-sionals is unlikely to have experienced substantial undemour-ishment, the lower rate of advanced prostate cancer among

shorter men may reflect lower levels of growth-related hor-mones more closely related to genetic factors than to severeenergy restriction. Because obesity is more prevalent in devel-

o_ countries, it may appear paradoxical that preadult obesityprotects against the occurrence of prostate cancer. However,

within a uniformly adequately nourished population, obesity in

individuals may be related to a hormonal milieu, including lowtestosterone and IGF-I and high estrogen, which may lower riskof this malignancy.

The hormonal milieu associated with linear growth andpreadult obesity may influence risk of advanced prostate can-

cer, but the mechanism at the cellular or genetic level is un-known. Perhaps a higher level of androgens and other growthfactors present during the development of the prostate glandincreases the pool of stem cells at risk, or increases the mitotic

rate, which may enhance the likelihood of the occurrence andpropagation of a genetic mistake. Our findings were limited toaggressive forms of prostate cancer, suggesting that some pre-

adult hormonal factors influence only a subset of prostaticcancers, which have the potential for rapid progression, or thatsome of the factors which determine aggressive behavior are“programmed” early during carcinogenesis. At this point, thesepossibilities are only theoretical, with little supporting cvi-dence.

Our findings suggest that the preadult hormonal milieu, asreflected in attained height and obesity, may have a profound

influence on prostate carcinogenesis. Although these results areof interest in understanding prostate carcinogenesis, the impactof adolescent obesity on overall health status is adverse (57).The concept that aggressive behavior of a tumor is strongly

determined by hormonal stimulation during organ development

deserves further study.

Acknowledgments

We are indebted to Mira Koyfman, Mildred Wolff, Elizabeth Frost-Hawes,Kathleen Markam, Kerry Demers, and Jill Arnold for expert help.

References

1. Pasquali, R., Casimirri, F., Cantobelli, S., Melchionda, N., Morselli Labate, A.

M., Fabbn, R., Capelli, M., and Bortoluzzi, L Effect of obesity and body

fat distribution on sex hormones and insulin in men. Metabolism, 40: 101-104,

1991.

2. Amatrauda, J. M., Harman, S. M., Pourmotabbed, G., and Lockwood, D. H.

Decreased plasma testosterone and fractional binding of testosterone in obesemales. J. Clin. Endocrinol. & Metab., 47: 268-271, 1978.

3. Barrett-Connor, E., and Khaw, K. T. Cigarette smoking and increased endog-

enous estrogen levels in men. Am. J. Epidemiol., 126: 187-192, 1987.

4. Dai, W. S., Gutain, J. P., Kuller, L. H.. and Cauley, J. A. Cigarette smokingand serum sex hormones in men. Am. J. Epidemiol., 128: 796-705, 1988.

5. Zumoff, B. Hormonal abnormalities in obesity. Acts. Med. Scand., 723

(Suppi.): 152-160, 1988.

6. Field, A. E., Colditz, G. A., Willeti, W. C., Longcope, C., and McKinlay, J. B.The relation of smoking, age, relative weight, and dietary intake to serum adrenalsteroids, sex hormones, and sex hormone-binding globulin in middle aged men.

J. Clin. Endocrinol. Metab., 79: 1310-1316, 1994.

7. Gann, P. H., Hennekens, C. H., Ma, J., Longcope, C., and Stampfer, M. J. A

prospective study of sex hormone levels and risk of prostate cancer. J. Nail.Cancer Inst. (Bethesda), 88: 1 1 18-1 126, 1996.

8. Kolonel, L. N., Yoshizawa, C. N., Hankin, and J. H. Diet and prostatic cancec

a case-control study in Hawaii. Am. J. Epidemiol., 127: 999-1012, 1988.

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from

Page 7: Height, Body Weight, and Risk of Prostate Cancer’ · 558 Height, Body Weight, and Prostate Cancer and heart disease in men (29). The cohort consists of 51,529 American male dentists,

Cancer Epidemiology, Blomarkers & Prevention 563

9. Graham, S., Haughey, B., Marshall, J., Priore, R., Byers, T., Rzepka, T.,Mettlin, C., and Pontes, J. E. Diet in the epidemiology of carcinoma of the

prostate gland. J. Nail. Cancer Inst. (Bethesda), 70: 687-692, 1983.

10. Ross, R. K., Shimizu, H., Paganini-Hill, A., Honda, G., and Henderson, B. E.Case-control studies of prostate cancer in blacks and whites in southern Califor-ma. J. Nat!. Cancer Inst. (Bethesda), 78: 86-74, 1987.

I 1. West, D. W., Slattery, M. L., Robison, L. M., French, T. K., and Mahoney,

A. W. Adult dietary intake and prostate cancer risk in Utah: a case-control studywith special emphasis on aggressive tumors. Cancer Causes & Control, 2:85-94,

1991.

12. Whittemore, A. S., Kolonel, L N., Wu, A. H., John, E. M., Gallagher, R. P.,

Howe, G. R., Burch, 3. D., Hankin, J., Dreon, D. M., West, D. W., Teh, C-Z., and

Paffenbarger, R. S., Jr. Prostate cancer in relation to diet, physical activity, and

body size in blacks, whites, and Asians in the United States and Canada. J. Nail.

Cancer Inst. (Bethesda), 87: 652-661, 1995.

13. Talamini, R., La Vecchia, C., Decarli, A., Negri, E., and Franceschi, S.Nutrition, social factors, and prostatic cancer in a Northern Italian population.

Br. J. Cancer, 53: 817-821, 1986.

14. Wynder, E. L., Mabuchi, K., and Whitmore, W. F., Jr. Epidemiology ofcancer of the prostate. Cancer (Phila.), 28: 344-360, 1971.

15. LaVecchia, C., Negn, E., Parazzini, F., Boyle, P., D’Avanzo, B., Levi, F.,Gentile, A., and Franceschi, S. Height and cancer risk in a network of case-control

studies from northern Italy. mt. J. Cancer, 45: 275-279, 1990.

16. Law, E. A., and Garfinkel, L. Variations in mortality by weight among

750,000 men and women. J. Chronic Dis., 32: 563-576, 1979.

17. Snowdon, D. A., Phillips, R. L., and Choi, W. Diet, obesity, and risk of fatal

prostate cancer. Am. J. Epidemiol., 120: 244-250, 1984.

18. Whittemore, A. S., Paffenbarger, R. S., Jr., Anderson, K., and Lee, J. E. Early

precursors of site-specific cancers in college men and women. I. Nail. Cancer Inst.

(Bethesda), 74: 43-51, 1985.

19. Greenwald, P., Damon, A., Kirmiss, V., and Poland, A. K. Physical activity

and demographic features of men before developing cancer of the prostate.J. NatI. Cancer Inst. (Bethesda), 53: 341-346, 1974.

20. Garn, S. M., Leonard, W. R., and Hawthorne, V. M. Three limitations of the

body mass index. Am. J. Clin. Nutr., 44: 996-997, 1986.

21. Severson, R. K., Grove, J. S., and Nomura, A. M. Body mass and prostatic

cancer: a prospective study. Br. Med. J., 297: 713-715, 1988.

22. Sakr, W. A., Haas, G. P., Cassin, B. F., Pontes, F. H., and Crissman, J. D. The

frequency ofcarcinoma and intraepithelial neoplasia of the prostate in young male

patients. J. Urol., 150: 379-385, 1993.

23. Cheville, J. C., and Bostwick, D. G. Postatrophic hyperplasia of the prostate.

A histologic mimic of prostatic adenocarcinoma. Am. J. Surg. Pathol., 19:

1068-1076, 1995.

24. Albanes, D., and Taylor, P. R. International differences in body height andweight and their relationship to cancer incidence. Nutr. Cancer, 14: 69-77, 1990.

25. Albanes, D., Jones, D. Y., Schatzkin, A., Micozzi, M. S., and Taylor, P. R.

Adult stature and risk of cancer. Cancer Res., 48: 1658-1662, 1988.

26. La Marchand, L., Kolonel, L. N., Wilkins, L. R., Myers, B. C., and Hirohata,

T. Animal fat consumption and prostate cancer: a prospective study in Hawaii.Epidemiology, 5: 276-282, 1994.

27. Andersson, 5-0., Baron, J., Wolk, A., Lindgren, C., Bergstrom, R., and

Adami, H-O. Early life risk factors for prostate cancer a population-based

case-control study in Sweden. Cancer Epidemiol., Biomarkers & Prey., 4: 187-192, 1995.

28. Herbert, P. R., Ajani, U., Cook, N., Lee, I. M., Chan, K. S., and Hennekens,

C. H. Adult height and incidence of total malignant neoplasms and prostatecancer: The Physicians’ Health Study. Am. J. Epidemiol., 1435: 78, 309, 1996.

29. Giovannucci, E., Rimm, E. B., Colditz, G. A., Stampfer, M. J., Ascherio, A.,

Chute, C. C., and Willett, W. C. A prospective study of dietary fat and risk of

prostate cancer. J. Nail. Cancer Inst. (Bethesda), 85: 157 1-1579, 1993.

30. Rimm, E. B., Stampfer, M. J., Colditz, G. A., Chute, C. G., Litin, L., andWillett, W. C. Validity of self-reported waist and hip circumferences in men and

women. Epidemiology, 1: 466-473, 1990.

31. Rimm, E. B., Giovannucci, E. L, Stampfer, M. J., Colditz, G. A., Litin, L. B.,Willett, W. C. Reproducibility and validity of an expanded self-administered food

frequency questionnaire among male health professionals. Am. J. Epidemiol.,135: 114-126, 1992.

32. Chasan-Taber, S., Rimm, E. B., Stampfer, M. J., Spiegelman, D., Colditz,G. A., Giovannucci, E., Ascherio, A., and Willett, W. C. Reproducibility andvalidity of a self-administered physical activity questionnaire for male health

professionals. Epidemiology, 7: 81-86, 1996.

33. Stampfer, M. J., Willett, W. C., Speizer, F. E., Dysert, D. C., Lipnick, R.,

Rosner, B., and Hennekens, C. H. Test of the national death index. Am. J.Epidemiol., 119: 837-839, 1984.

34. Willeu, W. C. Nutritional Epidemiology, pp. 217-44. New York: Oxford

University Press, 1990.

35. Stunkard, A. J., Sorensen, T., and Schulsinger, F. Use of the Danish adoption

register for the study of obesity and thinness. In: S. S. Kety. L P. Rowland, R.LSidman, and S. W. Matthysse (eds.), Genetics of Neurological and Psychiatric

Disorders, pp. 115-120. New York: Raven Press, 1983.

36. Must, A., Willeti, W. C., and Dicta, W. H. Remote recall ofchildhood height,

weight, and body build by elderly subjects. Am. J. Epidemiol., 138: 56-64, 1993.

37. Rothman, K. J. Modern Epidemiology. Boston: Little, Brown, & Co., 1986.

38. Cupples, L A., D’Agostino, R., Anderson, K., and Kannel, W. B. Compar-

ison of baseline and repeated measure covariate techniques in the Framingham

heart study. Stat. Med., 7: 205-218, 1988.

39. Garn, S. M., Leonard, W. R., and Hawthorne, V. M. Three limitations of thebody mass index. Am. J. Clin. Nutr., 44: 996-997, 1986.

40. Martha, P. M., Gorman, K. M., Blizzard, R. M., Rogol, A. D., and Veldhuis,J. D. Endogenous growth hormone secretion and clearance rates in normal boys,

as determined by deconvolution analysis: relationship to age, pubertal status, and

body mass. J. Chin. Endocrinol. & Metab., 74: 336-344, 1992.

41. Minuto, F., Barreca, A., Del Monte, P., Fortini, P., Resentini, M., Morabito,F., and Giordano, G. Spontaneous growth hormone and somatomedin-C/insulin-

like growth factor-I secretion in obese subjects during puberty. J. Endocrinol.

Invest., 11: 489-495, 1988.

42. Williams, T., Berelowitz, M., Joffe, S. N., Thorner, M. 0., Rivier, J., Vale,W., and Frohman, L A. Impaired growth hormone responses to growth hormone-

releasing factor in obesity. A pituitary defect reversed with weight reduction.N. Engl. J. Med., 311: 1403-1407, 1984.

43. Coffey, D. S. Physiological control of prostatic growth: an overview. In:

Prostate Cancer. UICC Technical Report Series, Vol. 48, pp. 4-23. Geneva:International Union Against Cancer, 1979.

44. Keenam, B. S., Richards, G. E., Ponder, S. W., Dakkas, J. S., Nagamani, M.,and Smithe, ER. Androgen-stimulated pubertal growth: the effects of testosteroneand dihydrotestosterone on growth hormone and insulin-like growth factor-I in

the treatment of short stature and delayed puberty. 3. Clin. Endocrinol. & Metab.,

76: 996-1001, 1993.

45. Cohen, P., Peehl, D. M., Lamson, G., and Rosenfeld, R. G. Insulin-like

growth factors (IGFs), IGF receptors and IGF binding proteins in primary cultures

of prostate epithelial cells. J. Clin. Endocrinol. & Metab., 73: 401-407, 1991.

46. Cohen, P., Peehl, D. M., Graves, H. C. B., and Rosenfeld, R. G. Biologicaleffects of prostate specific antigen (PSA) as an IGF binding protein-3 (IGFBP-3)

protease. J. Endocrinol., 142: 407-415, 1994.

47. Iwamura, M., Sluss, P. M., Casamento, J. B., and Cockett, A. T. Insulin-like

growth factor I: action and receptor characterization in human prostate cancer celllines. Prostate, 22: 243-252, 1993.

48. Peehl, D. M. Prostate specific antigen role and function. Cancer (Phila.), 75:

2021-2026, 1995.

49. Horton, R., Pasupuletti, V., and Antonipillai, I. Androgen induction of steroid5 alpha-reductase may be mediated via insulin-like growth factor-I. Endocrinol-

ogy, 133: 447-451, 1993.

50. Hovenian, M. S., and Deming, C. L. The heterologous growth of cancer of

the human prostate. Surg. Gynecol. & Obstet., 86: 29-35, 1948.

51. Noble, R. L. The development of prostatic adenocarcinoma in NB rats

following prolonged sex hormone administration. Cancer Res., 37: 1929-1933,1977.

52. Pollard, M., Luckert, P. H., and Schmidt, M. A. Induction of prostate

adenocarcinomas in Lobund Wistar rats by testosterone. Prostate, 4: 563-568,

1982.

53. Hoversland, R. C. Onset of obesity and puberty in genetically obese SHHF/

Mcc-cp rats. hit. J. Obes., 16: 977-984, 1992.

54. Winick, M., and Nobel, A. Cellular response in rats during malnutrition at

various ages. J. Nutr., 89: 300-306, 1966.

55. Smith, W. J., Underwood, L E., and Clemmons, D. R. Effects of caloric orprotein restriction on insulin-like growth factor-l (IGF-l) and IGF-binding pro-

teins in children and adults. J. Clin. Endocrinol. Metab., 80: 443-449, 1995.

56. Zaridze, D. G., Boyle, P., and Smans, M. International trends in prostatic

cancer. tat. J. Cancer, 33: 223-230, 1984.

57. Must, A., Jacques, P. F., Dallal, G. E., Bajema, C. J., and Dietz, W. H.

Long-term morbidity and mortality of overweight adolescents. A follow-up of theHarvard Growth Study of 1922 to 1935. N. Engl. J. Med., 37: 1350-1355, 1992.

Association for Cancer Research. by guest on August 21, 2020. Copyright 1997 Americanhttps://bloodcancerdiscov.aacrjournals.orgDownloaded from