LOWER BODY MASS INDEX IS ASSOCIATED WITH A HIGHERPROSTATE CANCER DETECTION RATE AND LESS FAVORABLE

PATHOLOGICAL FEATURES IN A BIOPSY POPULATION

JOSEPH C. PRESTI, JR.,*,† UNA LEE, JAMES D. BROOKS AND MARTHA K. TERRISFrom the Division of Urologic Oncology and Department of Urology, Stanford University School of Medicine (JCP, JDB), Stanford,

California, and Division of Urology, Medical College of Georgia (UL, MKT), Augusta, Georgia

ABSTRACT

Purpose: Body mass index (BMI), calculated as weight in kg divided by the square of height inm, is used as an indicator of obesity. We assessed the relationship between BMI, and prostatecancer detection rates and biopsy features in a referral based biopsy population.

Materials and Methods: A total of 787 consecutive patients referred for abnormal digital rectalexamination and/or prostate specific antigen (PSA) greater than 4 ng/ml underwent systematicprostate biopsy. Three standard categories of BMI were considered, namely normal—less than25, overweight—25 to 29.9 and obese—30 or greater kg/m2. The presence or absence of cancer,percent of core involvement and tumor grade were correlated with BMI. Additional analysescontrolled for patient age, PSA and prostate volume.

Results: For the entire population detection rates were highest in the normal BMI groupcompared to the overweight or obese group (52% vs 37% vs 42%, p � 0.0026). When stratified byage, this observation was true for men younger than 70 years (49% vs 32% vs 37%, p � 0.0042)but not for men 70 years or older. When only patients with PSA 10 ng/ml or less were considered,detection rates were highest in the normal BMI group (44% vs 28% vs 36%, p � 0.0061). Thisobservation also persisted in patients younger than 70 years with PSA 10 ng/ml or less, or whenonly patients younger than 70 years with a total prostate volume of less than 50 cc were included.Of patients with cancer those with a normal BMI had a greater length of needle core involvementon biopsy.

Conclusions: Normal BMI correlates with a higher cancer detection rate and larger cancers inmen undergoing prostate biopsy.

KEY WORDS: prostate, prostatic neoplasms, biopsy, obesity, body mass index

Higher body mass index (BMI), calculated as weight in kgdivided by the square of the height in m, as an indicator ofobesity has been shown to be a risk factor in breast and coloncancer. The role of obesity in prostate cancer remains contro-versial. Existing literature focuses on obesity or BMI as apotential risk factor for prostate cancer and/or mortality.Numerous case control and cohort studies of adult body massor obesity as a risk factor have yielded inconsistent results.Additionally, the 5 largest prospective studies are not con-clusive. The study of Andersson et al of 135,000 Swedishconstruction workers had one of the largest number of cases(2,368) and one of the longest followups (18 years).1 Thestudy showed a stronger relationship to mortality ratherthan incidence. In the Health Professionals Follow Up Studyof 47,781 men adult measures of obesity were not related tototal, advanced or metastatic prostate cancer.2 However, obe-sity at ages 5 and 10 years showed an inverse relationship.Higher body mass at age 21 years was also strongly related toa lower risk of advanced or metastatic cancer. More recentlyan update of this study demonstrated that a BMI of greaterthan 30 kg/m2 was associated with a lower risk of prostatecancer only in men under age 60 years or those with a family

history positive for prostate cancer.3 The Netherlands cohortstudy of 58,279 men 55 to 69 years old showed a moderateassociation between BMI at age 20 years and prostate can-cer.4 In the Alpha-Tocopherol Beta-Carotene Cancer Preven-tion Trial men with the highest BMI were at 40% increasedrisk for prostate cancer.5 In the Cancer Prevention Study II adose response type of outcome was observed for prostatecancer mortality and BMI with the relative risk increasing bya factor of 5%, 20% and 34% in men who were grades 1 to 3overweight, respectively.6

Why obesity might alter the risk of prostate cancer isspeculative. Abundant adipose tissue is associated withhigher plasma estrogen and lower testosterone. A prospec-tive study showed a strong trend of increasing prostate can-cer risk with greater concentrations of plasma testosteroneafter adjusting for sex hormone binding globulin.7 Racialvariations in hormone levels have also been suggested tocontribute to the observed difference in the race related in-cidence.8 We assessed the relationship between BMI, and theprostate cancer detection rate and biopsy pathological fea-tures in a referral based biopsy population.

MATERIALS AND METHODS

A retrospective analysis of prospectively collected datafrom a referral based biopsy population was performed. Atotal of 825 consecutive patients referred due to abnormaldigital rectal examination and/or prostate specific antigen(PSA) 4 ng/ml or greater underwent an 8 to 12 core biopsyscheme between February 1, 1998 and August 7, 2002. Clin-

Accepted for publication January 2, 2004.Study received human subjects committee approval.* Correspondence: Department of Urology, S-283, Stanford Uni-

versity School of Medicine, 300 Pasteur Dr., Stanford, California94305-5118 (telephone: 650-725-5544; FAX: 650-723-0765; e-mail:jpresti@stanford.edu).

† Financial interest and/or other relationship with InternationalCenter for Post Graduate Medical Education and TAP Pharmaceu-ticals.

0022-5347/04/1716-2199/0 Vol. 171, 2199–2202, June 2004THE JOURNAL OF UROLOGY® Printed in U.S.A.Copyright © 2004 by AMERICAN UROLOGICAL ASSOCIATION DOI: 10.1097/01.ju.0000124847.82541.60

2199

ical information included patient age, height and weightmeasured at biopsy, PSA, digital rectal examination findingsand prostate size on transrectal ultrasound. Pathological in-formation included the presence or absence of cancer, tumorgrade and percent of total core involvement, as calculated bythe sum of the measured length of cancer involvement oneach core in mm divided by the total length of all coresobtained in mm. Complete data were available on 787 pa-tients and they were included in this study.

For analyses 3 standard categories of BMI were consid-ered, namely normal—less than 25, overweight—25 to 29.9and obese—30 or greater kg/m2. The presence or absence ofcancer was correlated with BMI using the chi-square orFisher exact test. In patients with a positive biopsy thepercent of core involvement and tumor grade were comparedwith the Mann-Whitney U and chi-square or Fisher exacttest, respectively. Additional analyses controlled for age(younger than 70 vs 70 years or older), PSA (10 or less vsgreater than 10 ng/ml), digital rectal examination (normal vsabnormal) and prostate volume (50 or less vs greater than 50cc). Logistic regression analysis was performed, treating BMIas a continuous variable, to study the interrelationships be-tween BMI, age and PSA for predicting a positive biopsy.

RESULTS

In the entire population of 787 patients median age was66.0 years (IQR 60.0 to 70.0), median PSA was 7.2 ng/ml(IQR 5.2 to 10.7), median prostate volume was 45 cc (IQR 33to 67) and median BMI was 27.9 (IQR 24.9 to 31.4). Theoverall cancer detection rate was 42%. The breakdown withrespect to BMI category was normal in 204 patients (26%),overweight in 329 (42%) and obese in 254 (32%). The cancerdetection rate in each of the BMI categories was 52%, 37%and 42% for the normal, overweight and obese groups, re-spectively. Table 1 shows a comparison of the positive andnegative biopsy groups. Patients with a positive biopsy wereolder and had higher PSA and a smaller prostate. There wasa trend toward lower BMI in patients with a positive biopsyand yet it did not achieve statistical significance.

Table 2 shows the relationship between BMI and the can-cer detection rate. When the entire population was consid-ered, the cancer detection rate was highest in the normalBMI group. When stratified by age, this observation heldtrue for men younger than 70 years but not for older men. Asimilar observation was noted when we stratified for PSAless than 10 ng/ml or prostate size less than 50 cc. When onlypatients with a normal digital rectal examination were ana-lyzed, the same observations held true (data not shown).

When BMI was considered a continuous variable, logisticregression was possible. When the entire population wasconsidered, multivariate analysis demonstrated that PSAand age were the only significant predictors of a positivebiopsy (OR 1.07, 95% CI 1.05 to 1.09, p � 0.0001 and OR 1.04,95% CI 1.01 to 1.06, p � 0.0009, respectively). When onlymen younger than 70 years were considered, only PSA was asignificant predictor of a positive biopsy (OR 1.12, 95% CI1.08 to 1.66, p � 0.0001).

Table 3 shows the relationship between BMI and biopsypathological features of the cancer detected. With respect tothe cumulative length of core involvement, patients with a

normal BMI had a greater percent of total core length in-volvement, as detected by biopsy. This difference attainedstatistical significance when comparing the normal BMIgroup with the overweight group for almost all stratifica-tions, but not when comparing the normal to the obese group.No statistical difference was seen among any of the groupswhen comparing tumor grade (table 4).

DISCUSSION

Previous studies of BMI and prostate cancer have looked atthe association of the risk of the disease or outcomes oftherapy with BMI. It should be emphasized that in the cur-rent study we did not attempt to address BMI and risk sincewe analyzed a referral based biopsy population. To ourknowledge why BMI could influence the risk or mortality ofprostate cancer is unknown. One might speculate that hor-monal mechanisms are in effect. High adipose tissue is asso-ciated with higher serum estrogen and lower serum testos-terone. If this mechanism is in effect, the duration of obesitymight also be important. The possible influence of hormonallevels on prostate cancer development are well known be-cause prostate cancer rarely occurs in men who are castratedat an early age or in chronic alcoholics. The results of theProstate Cancer Prevention Trial suggest that cancer ratesare lower in men on finasteride.9 If a high BMI is indeedassociated with lower androgens, our study supports thisfinding. Our observations differ from those of the latter trialbecause men on finasteride, which lowers dihydrotestoster-one, had a higher prevalence of high grade cancer. In thecurrent study we found no impact of BMI on tumor grade.

Regarding the detection rate, when all patients were con-sidered, we noted a significantly higher cancer detection ratein those with a normal BMI. When stratified by age, it wastrue of men younger than 70 years but not in older men. Apossible explanation of this finding is that perhaps a highBMI with higher estrogen can be protective in earlier onsetprostate cancer. Alternatively, as mentioned, the duration ofobesity might be important. It is possible that the higher BMIin patients older than 70 years occurred later in life, ieincreasing weight associated with older age, perhaps due to amore sedentary lifestyle and age related changes in metabo-lism. If the latter is true, the duration of altered hormonalexposures might be shorter in patients older than 70 years.

It is well established that in a referral based biopsy popu-lation detection rates are inversely related to prostate size.To minimize the possibility of sampling error as an explana-tion for the observation of differing detection rates and BMI,we stratified the data on PSA and prostate volume. Whenconsidering men younger than 70 years with PSA less than10 ng/ml or a prostate volume of less than 50 cc, a normalBMI was associated with a higher detection rate.

In addition to detection rate differences, we also wished toaddress a possible association between BMI and the patho-logical features on biopsy of the cancers detected. To do it wecompared the cumulative length of the cancers on needlecores as well as the grade of the cancers. We recognize thatneedle biopsies of the prostate typically underestimate thevolume and grade of the cancers seen at subsequent radicalprostatectomy, although such errors should have beenequally likely in all groups of our subgroup analyses. Since aprevious study has demonstrated that in a biopsy populationincreasing age is associated with a higher detection rate,larger cancers and a greater likelihood of high grade diseaseon biopsy, we believed that it was important to stratify ouranalyses for age.10 Normal BMI was associated with a largercancer on the biopsy irrespective of age (table 3). BMI did notalter the grade of the cancers detected. These latter findingsconflict with a prior radical prostatectomy series showingthat men with a higher BMI were more likely to have non-organ confined and higher grade prostate cancer.11 However,

TABLE 1. Patients with positive and negative biopsy

Median Pos (IQR) Median Neg (IQR) p Value (Mann-Whitney U test)

No. pts 334 453Age 67.0 (62.0–72.0) 65.0 (59.0–69.0) 0.0000014PSA (ng/ml) 8.75 (6.1–14.4) 6.5 (4.8–9.0) 0.0000001Prostate vol

(cc)38 (30–56) 52 (36–77) 0.000001

BMI (kg/m2) 27.5 (24.6–31.4) 28.0 (25.5–31.5) 0.13

BODY MASS INDEX AND PROSTATE CANCER DETECTION2200

it should be noted that it was a surgical series, which wouldadd additional selection biases, while ours was a biopsy pop-ulation series.

Our study is strengthened by the large number of obesepatients. Also, BMI measurements were made by the medicalstaff and not by patient reporting, and the use of extendedcore biopsy schemes minimized the number of false-negativeresults. The limitations of this study include the lack of dataregarding the duration of patient exposure to the measuredBMI. Although our measurements of BMI were made byclinic staff, it is well recognized that BMI may not be theoptimal means of assessing obesity. Future studies shoulduse other anthromorphometric parameters, such as waist-to-hip ratios. In addition, we did not capture other cofoundingvariables, such as family history, diet or other variables, thatmight interact with cancer detection rates. Additional stud-ies are needed to assess better the possible link between BMIand hormone levels in patients undergoing prostate biopsy.

CONCLUSIONS

Normal BMI is associated with a higher prostate cancerdetection rate as well as larger cancers, as assessed by bi-opsy. The mechanisms of this association need further study.

REFERENCES

1. Andersson, S. O., Wolk, A., Bergstrom, R., Adami, H. O.,Engholm, G., Englund, A. et al: Body size and prostate cancer:a 20-year follow-up study among 135006 Swedish constructionworkers. J Natl Cancer Inst, 89: 385, 1997

2. Giovannucci, E., Rimm, E. B., Stampfer, M. J., Colditz, G. A. andWillett, W. C.: Height, body weight, and risk of prostate can-cer. Cancer Epidemiol Biomarkers Prev, 6: 557, 1997

3. Giovannucci, E., Rimm, E. B., Liu, Y., Lietzmann, M., Wu, K.,Stampfer, M. J. et al: Body mass index and risk of prostatecancer in U.S. health professionals. J Natl Cancer Inst, 95:1240, 2003

4. Schuurman, A. G., Goldbohm, R. A., Dorant, E. and van den

TABLE 2. BMI and detection rate

% Normal % Overweight % Obesep Value (chi-square test)

All Normal vsOverweight

Normal vsObese

All pts 52 37 42 0.0026 0.006 0.036Age:

Younger than 70 49 32 37 0.0042 0.0001 0.003370 or Older 57 51 57 0.70 0.91 0.46

PSA (ng/ml):10 or Less 44 28 36 0.0061 0.0001 0.0015Greater than 10 67 62 62 0.75 0.12 0.17

Prostate vol (cc):Less than 50 52 45 50 0.016 0.006 0.10850 or Greater 52 27 34 0.44 0.99 0.58

Age younger than 70 �:PSA 10 ng/ml or less 42 23 32 0.0033 0.0001 0.001Prostate vol less than 50 cc 56 39 47 0.034 0.018 0.37

For entire population sample size provided at least an 80% power to identify a 10% difference among the 3 groups (p �0.05).

TABLE 3. BMI and cancer extent on biopsy

Median % Total Core Length Involvement (IQR) p Value (Mann-Whitney U test)

Normal Overweight Obese Normal vsOverweight

Normal vsObese

All pts 11 (3.6–24) 5.6 (1.9–15) 8.0 (3.8–15) 0.0014 0.11Age:

Younger than 70 10 (3.5–23) 6.3 (1.8–14) 7.8 (3.5–15) 0.027 0.2370 or Older 12 (4.3–28) 3.8 (2.2–19) 8.7 (5.0–18) 0.028 0.33

PSA (ng/ml):10 or Less 6.8 (2.2–14) 3.2 (1.5–9.1) 7.5 (2.9–15) 0.045 0.61Greater than 10 18 (8.2–40) 10 (3.3–22) 8.9 (5.8–22) 0.021 0.029

Prostate vol (cc):Less than 50 11 (3.6–24) 6.6 (2.1–16) 8.5 (4.1–21) 0.011 0.2650 or Greater 10 (3.8–19) 3.2 (1.2–12) 7.4 (3.8–13) 0.23 0.68

Age younger than 70 �:PSA 10 ng/ml or less 8.3 (3.3–14) 2.4 (1.3–8.4) 6.6 (3.2–14) 0.015 0.72Prostate vol less than 50 cc 11 (3.8–23) 7.8 (2.0–15) 7.4 (3.2–15) 0.063 0.10

TABLE 4. BMI and cancer grade

% Pts With Biopsy Gleason Grade 4 or 5 p Value (chi-square or Fisher exact test)

Normal Overweight Obese All Normal vsOverweight

Normal vsObese

All pts 59 58 60 0.93 0.87 0.93Age:

Younger than 70 56 49 57 0.54 0.47 0.9370 or Older 64 74 66 0.61 0.48 0.99

PSA (ng/ml):10 or Less 42 42 52 0.42 0.88 0.34Greater than 10 80 76 75 0.86 0.83 0.79

Prostate vol (cc):Less than 50 60 57 60 0.91 0.79 0.9150 or Greater 53 58 59 0.92 0.77 0.76

Age younger than 70 �:PSA 10 or less 45 34 50 0.29 0.37 0.67Prostate vol less than 50 cc 56 50 55 0.78 0.65 0.89

BODY MASS INDEX AND PROSTATE CANCER DETECTION 2201

Brandt, P. A.: Anthropometery in relation to prostate cancerrisk in the Netherlands Cohort Study. Am J Epidemiol, 151:541, 2000

5. Aziz, N. M., Hartman, T. and Barrett, M.: Weight and prostatecancer in the Alpha-Tocopherol Beta-Carotene Cancer Preven-tion (ATBC) Trial. In: Proceedings of the American Society ofClinical Oncology, vol. 19, abstract 647a, 2000

6. Calle, E. E., Rodriguez, C., Walker-Thurmond, K. and Thun,M. J.: Overweight, obesity, and mortality from cancer in aprospectively studied cohort of U.S. adults. N Engl J Med, 348:1625, 2003

7. Gann, P. H., Hennekens, C. H., Ma, J., Longcope, C. andStampfer, M. J.: Prospective study of sex hormone levels andrisk of prostate cancer. J Natl Cancer Inst, 88: 1118, 1996

8. Gapstur, S. M., Gann, P. H., Kopp, P., Colangelo, L., Longcope,C. and Liu, K.: Serum androgen concentrations in young men:a longitudinal analysis of associations with age, obesity, and

race. The CARDIA male hormone study. Cancer EpidemiolBiomarkers Prev, 11: 1041, 2002

9. Thompson, I. M., Goodman, P. J., Tangen, C. M., Lucia, M. S.,Miller, G. J., Ford, L. G. et al: The influence of finasteride onthe development of prostate cancer. N Engl J Med, 349: 215,2003

10. Presti, J. C., Jr., O’Dowd, G. J., Miller, M. C., Mattu, R. andVeltri, R. W.: Extended peripheral zone biopsy schemes in-crease cancer detection rates and minimize variance in pros-tate specific antigen and age related cancer rates: results of acommunity multi-practice study. J Urol, 169: 125, 2003

11. Amling, C. L., Kane, C. J., Riffenburgh, R. H., Ward, J. F.,Roberts, J. L., Lance, R. S. et al: Relationship betweenobesity and race in predicting adverse pathologic variablesin patients undergoing radical prostatectomy. Urology, 58:723, 2001

BODY MASS INDEX AND PROSTATE CANCER DETECTION2202