SEARCH

SEARCH BY CITATION

Keywords:

  • prostate cancer;
  • anthropometry;
  • growth

Abstract

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The role of growth from birth through puberty and through adult life has been the subject of epidemiologic investigation in regard to the risk of prostate cancer but the evidence remains weak and inconsistent. We investigated associations between prostate cancer risk and a number of markers of body growth, size and changes to size in a population-based, case-control study in Australia from 1994 to 1998. We analyzed data obtained in face-to-face interviews from 1,476 cases and 1,409 controls. The main outcomes of interest were the timing of the growth spurt in adolescence, the experience of acne and interviewer observation of facial acne scarring, body size at age 21, body size in reference year, maximum body weight and rate of body size change since age 21 years. Analysis was performed on all cases and also by tumour grade. We found no associations with measures of body size including body mass index and lean body mass at age 21 or later in adult life. Having a growth spurt later than friends reduced risk (odds ratio [OR] 0.79 [0.63–0.97]) and some measures of acne also gave odds ratios less than 1, for example, having facial acne scarring gave an OR of 0.67 (0.45–1.00). We conclude that markers of delayed androgen action, such as delayed growth spurt in puberty, and markers of other androgen-dependent activity in puberty, such as facial acne scarring, are associated with prostate cancer risk but we could detect no associations with markers of adult body size and growth including lean body mass. © 2002 Wiley-Liss, Inc.

Although prostate cancer (PCa) is a disease of older men, early physical growth and development might be important in influencing PCa risks that will only be manifested many years later.1 Measures of body mass in adult life might also be associated with PCa risk, but a recent review described the evidence relating various body dimensions such as height and body mass index (BMI) to PCa risk as weak and inconsistent.2 One contributory factor to this inconsistency is the wide variation in quality of studies, particularly in regard to their statistical power—the literature is replete with many small case-control studies that have also depended on self-reported estimates of body dimensions.3 Another factor affecting consistency of findings is poor disease specificity, since most epidemiologic studies accept all PCa as one biologic entity when it is probable that PCa is biologically heterogeneous not only in terms of grade and stage but also with respect to its clinical behaviour.4 The use of BMI as a measure of obesity also presents difficulties as BMI does not differentiate between lean mass and fat mass2, 5 but at certain ages may be more closely associated with lean mass, e.g., during adolescence. As lean body mass (LBM) is associated with androgen and other growth-factor activity6, 7 and fat mass (FM) is associated with increased oestrogenic activity,8 these 2 elements of body mass are likely to have different associations with a hormone-dependent malignancy such as PCa. BMI also combines height and weight, each of which has separate determinants.

We conducted a large case-control study of PCa to see if we could detect any associations with markers of early growth and adult body size, particularly with respect to relatively early age at onset (<70 years of age) and tumours of moderate to high grade. We hypothesised, reasoning by analogy with menarche and breast cancer, that early andrarche would be positively associated with PCa risk and late andrarche with reduced risk; the former being related to high and the latter with low androgen activity. We also hypothesised that LBM would be positively associated with PCa risk because of the known dependency of both LBM and PCa on androgens and other growth factors.6, 7 Similarly, we hypothesised that FM would be negatively associated with PCa risk because of the likely increased production of oestrogens by peripheral aromatisation in adipose tissue8 and a possible protective effect of a consequently increased ratio of oestrogens to androgens.9

MATERIAL AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We carried out a population-based, case-control study of PCa in Melbourne, Sydney and Perth, Australia, details of which have been published before.10 Subjects were usual residents of the 3 cities' metropolitan areas. The study protocol was approved by all relevant human research ethics committees in Victoria, New South Wales and Western Australia. Eligible cases comprised all male residents of Melbourne, Sydney and Perth diagnosed from 1994 to 1997 and notified to the population-based cancer registries with a histopathology-confirmed diagnosis of adenocarcinoma of the prostate (International Classification of Diseases 9th revision rubric 185), excluding well-differentiated tumours (defined as low grade, i.e., those with Gleason scores <5). Cases had to be aged <70 years at diagnosis and registered on the State Electoral Rolls (adult registration to vote is compulsory in Australia). All cases diagnosed before the age of 60 years were included. Initially, random samples of 50% of cases diagnosed aged 60–64 and 25% of cases diagnosed aged 65–69 were selected, with the proportions varying over time to fit interview quotas.

Controls were randomly selected from men on the current State Electoral Rolls and were frequency matched to the age distribution of the cases in a ratio of 1 control per case. Potential controls were matched against the cancer registries at the time of recruitment to exclude men with a known history of PCa. Three controls were subsequently diagnosed with PCa and were selected as eligible cases. These subjects are included as cases and as controls.

After seeking advice from the case subjects' urologists, from whom some clinical details were sought, cases were written to invite them to participate. Controls were written to directly. Face-to-face interviews were administered, usually at the man's home. The interview included questions on timing of growth spurt (relative to friends) and the nature of acne experienced in adolescence (during the interview the interviewer made an independent observation of facial acne scarring and baldness), height, weight and waist circumference at 21 years of age, weight and waist circumference in the reference year (1 year before diagnosis [cases] or interview [controls]) and maximum weight ever attained. In a subset of the main study, a self-completed questionnaire was completed about history of sexual activity including the age at first ejaculation. This was limited to 1,020 cases and 1,248 controls included in this analysis. The scope and content of the questions was arrived at after focus group work with samples of men in the target age group. It has not proved possible to independently validate these questions.

BMI was computed as weight (kg) / height2 (meters) and LBM was also computed using the algorithm (2.447 − 0.09516 × age + 0.1074 × height + 0.3362 × weight) / 0.732).11 The annual rate of change between age 21 and the reference year was computed for weight, waist, BMI and LBM.

Case-control analyses were conducted using unconditional logistic regression, adjusting for reference age (at diagnosis for cases and date of selection from electoral roll for controls), study centre (Sydney, Melbourne, Perth), calendar year, family history (none vs. any first-degree relative diagnosed with PCa) and country of birth (Australia vs. other). Study centre and calendar year were included because the zeal for PCa screening using PSA tests varied between centres and over time.12 To investigate whether effects differed by potential disease aggressiveness, we analysed men with moderate-grade and high-grade tumours separately. Tests for heterogeneity in the odds ratios (ORs) between high- and moderate-grade PCa were performed using polytomous logistic regression models.13

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Of the 2,328 cases and 3,125 controls that were considered eligible at the time of selection, 1,497 cases (65%) and 1,434 controls (46%) were interviewed. Doctors could not be identified for, or refused access to, 16% of cases. Controls were more likely to refuse (38% vs. 17%), to be unable to speak sufficient English to be interviewed (7% vs. 1%) or to have moved (8% vs. 2%). Because the Electoral Roll is never completely up to date, we consider that a more appropriate response rate in controls excludes those who had moved or died and was 50%. After excluding subjects with missing data on the variables to be controlled for in the analysis, there were 1,476 cases (1,111 with moderate-grade and 365 with high-grade tumours) and 1,409 controls available.

Table I shows the distributions of cases and controls in terms of selected demographic variables and the main anthropometric measures. Cases were slightly better educated and more likely to be Australian-born than controls. They were also more likely to have a family history of PCa. There was little difference between cases and controls with respect to smoking history or in regard to the mean values for body measures.

Table I. Description of the Sample by Case/Control Status and Main Measures of Body Size
 Controls (n = 1,409)Cases (n = 1,476)
n%n%
Age group    
 <5527619.623716.1
 55–5931622.441227.9
 60–6440829.036624.8
 65–6940929.046131.2
Education1    
 Primary14310.11067.2
 Secondary45232.146731.6
 Diploma59742.465044.0
 Higher degree21515.325317.1
Country of birth    
 Australia84960.31,01969.0
 Other56039.745731.0
Family history of prostate cancer    
 No first-degree relative affected1,32393.91,23883.9
 At least one first-degree affected866.123816.1
Smoking history    
 Never smoked48634.554036.6
 Ever-smoked92365.593663.4
Body measuresMeanSD2MeanSD2
  • 1

    Two controls had missing information on education.

  • 2

    SD, standard deviation.

 Height (cm)175.2(7.0)175.7(6.8)
 Weight (kg)80.6(12.7)81.0(12.8)
 Body mass index26.3(3.8)26.2(3.7)
 Waist (cm)94.7(9.2)94.8(8.9)
 LBM (kg)58.5(6.4)58.6(6.5)

Table II shows findings in terms of ORs and their 95% confidence intervals from logistic regression models for timing of growth spurt, acne and body size attained by 21 years of age for the total sample and for moderate- and high-grade tumours separately. A later growth spurt relative to peers reduced the risk of PCa by 21% (p = 0.02). There was also a 33% reduced risk of PCa associated with the interviewer's observation of facial acne scarring (p = 0.05). Similar ORs, but lacking statistical significance, were observed for having had acne, onset of acne after 15 years of age and having had more than 24 months of medical treatment for acne. Generally, the OR point estimates in the high-grade group had stronger effect sizes but were not significantly different from those for the moderate-grade group. Attained height, weight, waist circumference and BMI and LBM at age 21 essentially showed no associations with risk. There was only suggestive evidence of a trend with increasing quartiles of waist size at 18–21 years in the high-grade subgroup.

Table II. The Risk of Prostate Cancer Associated with Events Occurring in Puberty and Body Size at Age 21 Years: Odds Ratios (OR) and 95% Confidence Intervals by Tumour Grade
 Controls 1,409All subjectsp*Moderate gradep*High gradep*
Cases 1,476OR95% CICases 1,111OR95% CICases 365OR95% CI
Timing of growth spurt             
 Growth spurt earlier than friends1762021.09(0.86–1.39) 1481.03(0.80–1.34) 541.21(0.84–1.74) 
 Growth spurt same time as friends9591,0331.00 0.027761.00 0.112571.00 0.02
 Growth spurt later than friends2632260.79(0.63–0.97) 1800.81(0.65–1.03) 460.69(0.47–0.99) 
Observation of acne facial scarring             
 No acne scarring observed1,2791,3651.00 0.051,0241.00 0.233411.00 0.04
 Obvious acne scarring68500.67(0.45–1.00) 420.77(0.51–1.18) 80.46(0.21–1.00) 
Questions on acne             
 Never had acne (referent category)1,0691,1491.00 0.17 8531.000.40 2961.000.05
 Ever had acne3353250.85(0.70–1.04) 2570.90(0.74–1.09) 680.71(0.52–0.97) 
 Never had acne (referent category)1,0691,1491.00 0.368531.00 0.152961.00 0.28
 Acne onset before 15 years2162270.96(0.77–1.20) 1871.05(0.83–1.32) 400.67(0.45–0.99) 
 Acne onset after 15 years114970.75(0.55–1.02) 690.71(0.51–1.00) 280.90(0.56–1.44) 
 Never had acne (referent category)1,0691,1491.00 0.608531.00 0.802961.00 0.15
 Only face affected with acne2212110.88(0.70–1.10) 1700.94(0.74–1.19) 410.65(0.44–0.96) 
 Face and other areas affected with acne90880.90(0.64–1.26) 680.92(0.65–1.32) 200.85(0.49–1.47) 
 Only non-facial areas affected with acne21250.80(0.41–1.54) 180.75(0.37–1.55) 71.14(0.44–2.95) 
 Never had acne (referent category)1,0691,1491.00 0.408531.00 0.71 2961.000.15
 Not treated medically for acne2432380.88(0.70–1.09) 1890.92(0.73–1.16) 490.71(0.49–1.02) 
 Treated medically for acne90850.88(0.63–1.23) 660.91(0.64–1.31) 190.81(0.47–1.40) 
 Never had acne (referent category)1,0691,1491.00 0.548531.00 0.312961.00 0.73
 <24 months treatment for acne46561.09(0.71–1.69) 441.18(0.74–1.87) 120.84(0.42–1.71) 
 >24 months treatment for acne40260.63(0.37–1.09) 190.59(0.33–1.08) 70.81(0.35–1.90) 
Weight 21 years (kg)             
 <63.14834481.00 0.283351.00 0.441131.00 0.20
 63.1–67.52442611.02(0.80–1.29) 2021.03(0.80–1.32) 591.00(0.68–1.46) 
 67.6–74.53223821.20(0.97–1.49) 2851.20(0.95–1.51) 971.21(0.87–1.69) 
 >74.53483671.01(0.81–1.25) 2751.00(0.79–1.26) 921.04(0.74–1.46) 
Waist at 21 years (cm)             
 <793553021.00 0.172251.00 0.34771.00 0.06
 79–813484051.39(1.10–1.75) 3101.40(1.09–1.80) 951.43(0.98–2.07) 
 82–862732841.17(0.91–1.50) 2081.12(0.85–1.47) 761.36(0.92–2.01) 
 >863073371.20(0.95–1.53) 2521.19(0.92–1.55) 851.32(0.90–1.94) 
BMI at 21 years (kg/m2)             
 <20.53533481.00 0.362661.00 0.53821.00 0.25
 20.5–22.13723830.99(0.79–1.23) 2880.99(0.78–1.26) 951.02(0.71–1.46) 
 22.2–23.93373510.96(0.76–1.20) 2640.95(0.75–1.22) 870.97(0.67–1.41) 
 >23.93323761.10(0.88–1.39) 2791.09(0.85–1.40) 971.19(0.83–1.71) 
LBM at 21 years (kg)             
 <54.03182771.00 0.292071.00 0.43701.00 0.21
 54.1–57.03303621.14(0.89–1.44) 2751.13(0.88–1.47) 871.23(0.84–1.80) 
 57.1–61.04214731.20(0.96–1.51) 3591.21(0.95–1.54) 1141.24(0.86–1.78) 
 >61.03253461.05(0.82–1.34) 2561.02(0.78–1.33) 901.16(0.79–1.71) 
 *p for trend             

In the subset of 1,020 cases and 1,248 controls from whom we obtained a sexual history, we found weak evidence of an inverse association with age at first ejaculation. Comparing the referent category (<13 years of age) with first ejaculations occurring at ages 13–14, ages 15–16 and after age 16 gave ORs (95% confidence intervals) of 1.02 (0.80–1.3), 1.06 (0.80–1.39) and 0.77 (0.53–1.12), respectively, p trend >0.05. The ORs did not differ by tumour grade.

All associations between the measures of adult body size (i.e., in the reference year) and rates of change in weight, waist, BMI and LBM after the age of 21 years and with maximum weight were not statistically different from unity (data not shown).

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We have detected weak but statistically significant associations between PCa risk and the timing of the pubertal growth spurt relative to peers and with some aspects of acne in adolescence and markers of facial acne scarring in adult life but none with various measures of body size at age 21 or at reference age or with body size changes occurring in adult life.

In interpreting these findings, we have considered the extent to which they might be the result of bias. One possible weakness of our study is suboptimal response rates. However, a comparison of our controls with similarly aged men from the same Australian states in the 1995 National Nutrition Survey (NNS) shows only slight differences with respect to country of birth (Australian-born 62%; NNS 66%), BMI (acceptable weight 39%; NNS 37%) and the proportion of ex-smokers (50%; NNS 46%).14 Another possible weakness is that we conducted the study during a period of intense PSA testing and that this might have influenced the mix of PCa phenotypes that we ascertained. We have attempted to control for this by conducting subgroup analysis based on tumour grade and have essentially found little difference. Thus, we consider it unlikely that these potential weaknesses have affected our findings. A more problematic issue is the use of self-reported measurements of body sizes. The recall of physique when aged 21, by men aged up to 69 years, is prone to error that, if nondifferential, may have attenuated any association. Apart from the real difficulties in recalling the distant past, obese men tend to underreport their weight and vice versa,2 and this would also tend to attenuate any effect. The recall of body size in the reference year, however, is probably prone to differential error, cases being more likely to have experienced weight changes due to either the presence of the disease or to its treatment. For this reason, we have not reported the details of the essentially null findings in this regard.

In regard to adolescence and the timing of the pubertal growth spurt, we showed that later growth relative to peers was protective against PCa, analogous to delayed age at menarche being protective against breast cancer in women. This finding is consistent with other studies that have measured related markers. Hayes et al.16 showed a positive association with early height attainment. And others17 have shown a positive association with shaving commencement prior to age 15 but only in American blacks. These observations are also consistent with positive associations shown in some studies with early age at first sexual intercourse (18–20) and a negative association with age at first ejaculation.21 We also report weak evidence of an association with age at first ejaculation that is consistent with these data.

The Health Professionals Follow Up Study22 found a negative association between quintiles of obesity at age 10 and advanced PCa and metastatic PCa and also between height and advanced PCa, the OR comparing men 74 in. or taller with men 68 in. or shorter was 1.68 (95% CI 1.16–2.43). Their analysis lends support to the concept that the preadult hormonal milieu might be important to PCa risk. We did not measure childhood obesity nor did we have enough advanced cases to perform a separate analysis. Our analysis showed a null association with height irrespective of tumour grade.

In the Netherlands Cohort Study23 no association was found with adult height, BMI or LBM but there was a positive association with BMI calculated at age 20 and a negative association for gain in BMI since age 20 (no differences in regard to disease severity). BMI at age 20 probably reflects the effect of recent growth factor and androgen influences on attained lean mass, and the gain in BMI since age 20 probably reflects increasing fat mass as lean mass in adulthood tends to be stable, decreasing slowly with increasing age.7 The directions of the associations are therefore consistent with an early androgenic milieu increasing the risk of PCa and a later fat mass-influenced oestrogenic milieu decreasing risk.

A case-control study conducted in China24 found no association in a fairly lean population (mean BMI 21.9) with adult height or weight or with preadult or adult BMI, but found instead strong positive associations with waist hip ratio (WHR) and moderate negative associations with hip circumference. We found no associations in our study with waist circumference and did not collect hip circumference. The mean waist circumference in our study was about 95 cm compared to 82 cm in the Chinese study, and it might be that associations with abdominal obesity may only be detected in populations with a lower average girth.

Taken together, these findings suggest a role for androgens (and other growth factors) influencing PCa risk possibly by their effect on the timing and extent of prostatic development and growth and also on lean body mass growth generally. Further, height at puberty is related to levels of insulin-like growth factor 1 (IGF-1)25 and IGF-1 levels have also been associated with PCa risk in prospective studies.6

The negative associations that we find with markers of acne are intriguing. Acne is androgen dependent but is related to the action of a steroid reductase (5 alpha reductase 1) in the skin converting testosterone to its active form dihydrotestosterone (DHT), rather than the action of its isozyme (5 alpha reductase 2), which performs this function in the prostate.26 Androgens (as DHT) act through a single intracellular androgen receptor (AR), and genetic variation in the AR has been implicated in PCa risk.27 PCa has also been associated with the action of other growth factors such as IGF-1 that is known to modulate AR activity.28 Why 5 alpha reductase 1 activity leading to acne (particularly of sufficient severity to produce permanent facial scarring) should be protective against PCa is difficult to explain given our current understanding of the interaction of various androgen pathways with other molecules, such as IGF-1, in influencing androgen-dependent disorders. Presently, we can acknowledge facial acne scarring only as a possible marker of some form of androgen action during puberty and adolescence as we have no evidence that acne scarring is a marker of long-term androgen levels.

IGF-1 also provides a theoretical link between body size, physical activity and PCa risk.29 Our data, however, show no association between any measure of body size, or change over time in body size, and PCa risk. It is possible that the use of self-reported information rather than direct body measurements has attenuated the risk estimates and so the possibility of small risks cannot be excluded. Similarly, the range in BMI and other measures of body size in our population may be too narrow to detect a difference. Furthermore, relationships might exist only in subgroups based either on the aggressiveness of the tumour or on some familial or genetic trait.30 We have been unable to show any difference in risk by tumour grade but will be following cases indefinitely to capture details of tumour progression and death with a view to future reanalysis.

In conclusion, we have suggestive evidence that markers of androgen action in puberty may predict PCa risk in later life and that markers of severe acne are associated with reduced risk. With respect to our other hypotheses, we have no evidence that body size or changes to body size in adult life are important to PCa risk but would echo Nomura's recommendations2 that further work needs to be done within the context of prospective studies that have made direct measurements of anthropometric parameters, including lean body mass, and that have stored serum appropriately for the measurement of various hormones and growth factors such as IGF-1.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

The contribution of G.S. and P.B. was within the framework of support from the Italian Association of Cancer Research (Associazione Italiana per la Ricerca sul Cancro). We acknowledge the work of the study coordinator Ms. M. Staples and the research team, Ms. B. McCudden, Mr. J. Connal, Mr. R. Thorowgood, Ms. C. Costa, Ms. M. Kevan and Ms. S. Palmer. We thank the many urologists who kindly assisted us by providing information and access to their patients. We also express our gratitude to the many men who participated. Finally, we thank an anonymous reviewer for their very helpful comments.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIAL AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES
  • 1
    Andersson SO, Baron J, Wolk A, Lindgren C, Bergstrom R, Adami HO. Early life risk factors for prostate cancer: a population-based case-control study in Sweden. Cancer Epidemiol Biomarkers Prev 1995; 4: 18792.
  • 2
    Nomura AM. Body size and prostate cancer. Epidemiol Rev 2001; 23: 12631.
  • 3
    Giles GG, Ireland P. Diet, nutrition and prostate cancer. Int J Cancer 1997; 43: 15.
  • 4
    Miller GJ, Torkko KC. Natural history of prostate cancer—epidemiologic considerations. Epidemiol Rev 2001; 23: 148.
  • 5
    Garn SM, Leonard WR, Hawthorne VM. Three limitations of the body mass index. Am J Clin Nutr 1986; 44: 9967.
  • 6
    Pollak M. Insulin-like growth factors and prostate cancer. Epidemiol Rev 2001; 23: 5966.
  • 7
    Vermeulen A, Goemaere S, Kaufamn JM. Testosterone, body composition and aging. J Endocrinol Invest 1992; 22: 1106.
  • 8
    Pasquali R, Casimirri F, Cantobelli S, Melchionda N, Morselli Labate AM, Fabbri R, Capelli M, Bortoluzzi L. Effect of obesity and body fat distribution on sex hormones and insulin in men. Metabolism 1991; 40: 1014.
  • 9
    Gann P, Hennekens C, Ma J, Longcope C, Stampfer M. Prospective study of sex hormone levels and risk of prostate cancer. J Natl Cancer Inst 1996; 88: 111826.
  • 10
    Giles GG, Severi G, Sinclair R, English DR, McCredie MR, Johnson W, Boyle P, Hopper JL. Androgenetic alopecia and prostate cancer: findings from an Australian case-control study. Cancer Epidemiol Biomarkers Prev. 2002; 11: 54953.
  • 11
    Willett W. Anthropometric measures and body composition. In: WillettW, ed. Nutritional epidemiology. New York: Oxford University Press, 1990. 21744.
  • 12
    Smith DP, Armstrong BK. Prostate-specific antigen testing in Australia and association with prostate cancer incidence in New South Wales. Med J Australia 1998; 169: 1720.
  • 13
    Hosmer DW, Lemeshow S. Applied logistic regression. New York: Wiley, 1989.
  • 14
    Australian Bureau of Statistics. National Nutrition Survey: nutrient intakes and physical measurements. Cat No 48050. Canberra: Australian Bureau of Statistics, 1998.
  • 15
    Cameron R, Evers SE. Self-reported issues in obesity and weight management: state-of-the-art and future directions. Behav Assess 1990; 12: 91106.
  • 16
    Hayes RB, de Jong FH, Raatgever J, Bogdanovicz J, Schroeder FH, van der Maas P, Oishi K, Yoshida O. Physical characteristics and factors related to sexual development and behaviour and the risk for prostatic cancer. Eur J Cancer Prev 1992; 1: 23945.
  • 17
    Habel LA, Van Den Eeden SK, Friedman GD. Body size age at shaving initiation and prostate cancer in a large multiracial cohort. Prostate 2000; 43: 13643.
  • 18
    Andersson SO, Baron J, Bergstrom R, Lindgren C, Wolk A, Adami HO. Lifestyle factors and prostate cancer risk: a case-control study in Sweden. Cancer Epidemiol Biomarkers Prev 1996; 5: 50913.
  • 19
    Honda GD, Bernstein L, Ross RK, Greenland S, Gerkins V, Henderson BE. Vasectomy cigarette smoking and age at first sexual intercourse as risk factors for prostate cancer in middle-aged men. Br J Cancer 1988; 57: 32631.
  • 20
    Schuman LM, Mandel J, Blackard C, Bauer H, Scarlett J, McHugh R. Epidemiologic study of prostatic cancer: preliminary report. Cancer Treat Rep 1977; 61: 1816.
  • 21
    Rotkin ID. Studies in the epidemiology of prostatic cancer: expanded sampling. Cancer Treat Rep 1977; 61: 17380.
  • 22
    Giovannucci E, Rimm EB, Stampfer MJ, Colditz GA, Willett WC. Height, body weight and risk of prostate cancer. Cancer Epidemiol Biomarkers Prev 1997; 6: 55763.
  • 23
    Schuurman AG, Goldbohm RA, Dorant E, van den Brandt PA. Anthropometry in relation to prostate cancer risk in the Netherlands Cohort Study. Am J Epidemiol 2000; 151: 5419.
  • 24
    Hsing AW, Deng J, Sesterhen IA, Mostofi FK, Stanzyk FZ, Benichou J, Xie T, Gao Y-T Body size and prostate cancer: a population-based case-control study in China. Cancer Epidemiol Biomarkers Prev 2000; 9: 133541.
  • 25
    Juul A, Dalgaard P, Blum WF, Bang P, Hall K, Michaelsen KF, Muller J, Skakkebaek NE. Serum levels of insulin-like growth factor (IGF)-binding protein-3 (IGFBP-3) in healthy infants, children and adolescents: the relation to IGF-I, IGF-II, IGFBP-1, IGFBP-2, age, sex, body mass index and pubertal maturation. J Clin Endocrinol Metab 1995; 80: 253442.
  • 26
    Thigpen AE, Silver RI, Guileyardo JM, Casey ML, McConnell JD, Russell DW. Tissue distribution and ontogeny of steroid 5 alpha-reductase isoenzyme expression. J Clin Invest 1993; 92: 90310.
  • 27
    Makridakis NM, Reichardt JKV. Molecular epidemiology of hormone-metabolic loci in prostate cancer. Epidemiol Rev 2001; 23: 249.
  • 28
    Culig Z, Hobisch A, Cronauer MV, Radmayr C, Trapman J, Hittmair A, Bartsch G, Klocker H. Androgen receptor activation in prostatic tumour cell lines by insulin-like growth factor 1 keratinocyte growth factor and epidermal growth factor. Cancer Res 1994; 54: 54748.
  • 29
    Kaaks R, Lukanova A. Energy balance and cancer: the role of insulin and insulin-like growth factor-I. Proc Nutr Soc 2001; 60: 91106.
  • 30
    Norrish AE. Height-related risk factors for prostate cancer. Br J Cancer 2000; 82: 2415.