Current research is inconclusive regarding the effect of obesity on outcomes after a prostate cancer diagnosis. The objective of this study was to examine associations between obesity and the risks of developing metastasis or prostate cancer-specific mortality in a population-based cohort of men with prostate cancer.
Seven hundred fifty-two middle-aged men with prostate cancer who were enrolled in a case-control study and remain under long-term follow-up for disease progression and mortality formed the study cohort. Body mass index (BMI) in the year before diagnosis was obtained at the time of initial interview. Cox proportional hazards models were used to estimate hazard ratios (HRs) and 95% confidence intervals (95% CIs) of prostate cancer metastasis and mortality associated with obesity, controlling for age, race, smoking status, Gleason score, stage at diagnosis, diagnostic prostate-specific antigen level, and primary treatment.
Obesity (BMI ≥30 kg/m2) was associated with a significant increase in prostate cancer mortality (HR, 2.64; 95% CI, 1.18–5.92). Among men who were diagnosed with local- or regional-stage disease, obesity also was associated with an increased risk of developing metastasis (HR, 3.61; 95% CI, 1.73–7.51). Associations generally were consistent across strata defined by Gleason score (2–6 or 7 [3 + 4] vs 7 [4 + 3] or 8–10), stage (local vs regional/distant for mortality), and primary treatment (androgen-deprivation therapy use: yes vs no).
Little is known about whether lifestyle modifications, such as weight reduction, could affect the outcomes of prostate cancer patients after diagnosis and treatment. Epidemiologic data consistently have indicated that obesity is associated with a modestly increased risk of prostate cancer-specific death,1, 2 but those studies did not address whether the increased mortality was because of the association of obesity with prognostic factors (eg, Gleason score, stage) or because of the effects of obesity on disease progression after treatment. Several studies have indicated that obesity is associated with poor prognostic factors at diagnosis, such as high-grade or nonlocalized prostate cancer,3–5 and we recently reported findings from the Prostate Cancer Prevention Trial in which obesity increased the risk of developing high-grade prostate cancer (Gleason scores of 8–10) by 78%.6 Furthermore, several clinical studies have indicated that obesity is associated with an increased risk of biochemical recurrence or disease progression after radical prostatectomy (RP).7–11 Inferences from those studies are limited, because biochemical recurrence is correlated only weakly with risk of death,12, 13 and obesity simply may affect the success of surgery rather than the biology of cancer. A single study has examined the risk of biochemical recurrence, prostate cancer metastasis, and death after RP, and no associations were observed after controlling for clinical prognostic factors at baseline.14 Current research is too limited to support strong conclusions regarding the role of obesity in outcomes after prostate cancer diagnosis.
In this study, we address whether obesity affects the risk of prostate cancer metastasis or death in a population-based cohort after an initial diagnosis of prostate cancer. Unlike other clinical outcome studies, our sample of patients with prostate cancer included all men without consideration of treatment, stage, or grade. Because obesity is a modifiable risk factor, results from this study may help address whether weight reduction should be a specific recommendation after primary treatment of prostate cancer.
MATERIALS AND METHODS
Study Participants and Data Collection
Study participants were 752 men with newly diagnosed prostate cancer who participated in a population-based, case-control study. Details of selecting the study participants and data collection for this study have been described previously.15 Briefly, eligible patients were Caucasian and African-American men who were residents of King County (Seattle, WA), ages 40–64 years, who were newly diagnosed with histologically confirmed prostate cancer from January 1, 1993 through December 31, 1996, and who we identified through the Seattle-Puget Sound Surveillance, Epidemiology, and End Results (SEER) cancer registry. Because the initial study was designed to examine the risk of prostate cancer in middle-aged men, all patients aged 40 years to 59 years and a random 75% sample of patients aged 60 years to 64 years at diagnosis were recruited for the study. Of 915 men who were eligible for participation, 752 men (82.1%) were interviewed. Reasons for nonresponse were physician's refusal to allow contact (2.6%), patient refusal (12.5%), inability to locate (1.5%), illness (0.4%), and death (0.2%).
After giving their informed consent, participants completed a structured, in-person interview that was conducted by men who were trained interviewers. Detailed information was collected on demographic and lifestyle characteristics, family history of prostate cancer, and medical and prostate cancer screening histories. Height and weight (1 year prior to prostate cancer diagnosis) were self-reported by all participants, and body mass index (BMI) (in kgμm2) was classified into normal weight (BMI <25 kg/m2), overweight (BMI 25–29.9 kg/m2), and obese (BMI ≥30 kg/m2) categories. Tumors were graded using the Gleason scoring system.16 Gleason score was obtained from biopsy reports (30%) or from surgical pathology reports (70%) for men undergoing RP, and was classified into scores from 2 to 6, a score of 7 with a primary pattern of 3 (3 + 4), and a score of 7 with a primary pattern of 4 (4 + 3) or scores from 8 to 10. The 7 (3 + 4) category included 22 patients who had Gleason scores of 7 but had unknown predominant cancer patterns. Cancer stage was defined according to SEER criteria as localized (stages A and B; confined to the prostate), regional (stage C; regional spread outside the prostatic capsule), and distant (stage D; metastases) using pathologic data for men who underwent RP and clinical information for all others.17 When primary treatment was used as a covariate, the categories were RP without androgen-deprivation therapy (ADT), RP with ADT, radiation without ADT, radiation with ADT, watchful waiting/other, and ADT only. The SEER database is updated quarterly to ascertain vital status and underlying causes of death if patients are deceased.
In January 2004, a follow-up survey was sent to all living men (as recorded in the SEER database) who had provided consent for future contact to collect information on quality of life, secondary therapies, prostate-specific antigen (PSA) results, physicians' diagnoses of prostate cancer recurrence or progression, and diagnostic procedures (ie, bone scan, magnetic resonance image [MRI] or computed tomography [CT] scan, biopsy), the date of the most recent procedures, and whether the test result was positive and showed cancer recurrence or progression. Of 631 men who were contacted, 31 men (5%) refused to participate, 70 men (11%) did not return their questionnaires, and 10 men (2%) had died, leaving 520 men (82%) who completed the survey. Men who did not complete a follow-up survey were more likely to be younger (<50 years) at diagnosis, African American, and less educated compared with respondents (P < .05), but they did not differ by BMI or clinical characteristics, such as tumor stage, Gleason score, or primary treatment.
The mortality endpoints for this study were prostate cancer-specific mortality and other-cause mortality. The underlying causes of death were determined through the SEER registry, which routinely links with the Washington State computerized death database, and were verified by a review of death certificates. Of the 752 men in the cohort, 50 died of prostate cancer, and 64 died of other causes.
The disease progression endpoint was defined as development of metastases among men with local- or regional-stage disease at diagnosis. These analyses included 514 men who completed the survey, 36 of whom developed metastases based on a bone scan, an MRI or CT scan that showed prostate cancer, or a positive bone or lymph node biopsy. We also included an additional 30 men who had died of prostate cancer before the follow-up survey, because we assumed that these men developed metastases at some time before death.
We examined means and distributions of demographic and clinical characteristics across BMI categories and tested for statistically significant associations using t tests and chi-square tests. Unadjusted prostate cancer-specific and other-cause mortality rates were calculated by dividing the number of deaths by the corresponding number of person-years of follow-up. Multiple contrasts in statistical tests of differences in mortality rates across demographic and clinical characteristics were adjusted using the Bonferroni method.18
We used Cox proportional hazards models to estimate hazard ratios (HRs) and 95% confidence intervals (95% CIs) for prostate cancer-specific and other-cause mortality.19 Survival, which was defined as the time from diagnosis to death or censoring for men who remained alive, was the time-dependent variable used in the analysis. The censoring date for patients who remained alive was December 31, 2005, which was the most recent date that participants were matched with the cancer registry database. Kaplan-Meier survival plots and log-rank tests were used to estimate mortality-free survival by BMI categories.
For the analysis of disease progression, survival was defined as the time from diagnosis to the first self-reported evidence of development of metastasis. The censoring date for patients without a metastasis event was the date that their follow-up questionnaire was received. The time of first metastasis in the 30 men who had died of prostate cancer before the follow-up survey was unknown; therefore, we used a method that was developed by Robins et al20 to estimate the risk of progression with incomplete data. A logistic regression model was fit to predict the probability of having a date of prostate cancer metastasis using baseline data regressors; then, weighted Cox proportional hazards models were used to calculate HRs and 95% CIs for prostate cancer progression in which the weights were the inverse of the probability of having a metastasis date for men with a metastatic event, and the weight was 1.0 for the men who were censored.20
Table 1 provides the distributions of demographic and clinical characteristics of study patients and corresponding unadjusted mortality rates for prostate cancer-specific and other-cause mortality. Most patients were Caucasians, 60% were aged <60 years at diagnosis, 16% were current smokers, 27% were diagnosed with regional- or distant-stage disease, 14% had Gleason scores ≥7 (4 + 3), 63% underwent radial prostatectomy as their primary treatment, and 72% had diagnostic serum PSA levels >4.0 ng/mL. The mean BMI (± standard deviation) was 26.7 ± 3.9 kg/m2, and 17.0% of patients were classified as obese (BMI ≥30 kg/m2). Unadjusted prostate cancer-specific mortality rates were higher for men who were obese, current smokers, diagnosed with regional- or distant-stage cancer, treated with radiation and ADT or ADT alone, had serum PSA levels ≥20 ng/mL, or had tumors with Gleason scores ≥7 (3 + 4). Other-cause mortality did not differ by Gleason score, cancer stage, or serum PSA level.
Table 1. Demographics, Clinical Characteristics, and Prostate Cancer-specific and Other-cause Mortality Rates (per 1000 Person-years)
There were 50 prostate cancer-specific deaths and 64 deaths from other causes during an average follow-up of 9.5 years (range, 1.2–12.6 years). Figure 1A shows the Kaplan-Meier survival curves for prostate cancer-specific mortality stratified by BMI categories. Obese men had significantly shorter prostate cancer-free survival (P = .03). Figure 1B shows the Kaplan-Meier survival curves for other-cause mortality stratified by BMI categories. There were no significant differences in other-cause mortality among different BMI categories (P = .40).
Table 2 provides the distributions of demographic and clinical characteristics by BMI categories. There were no significant associations between obesity and prognostic factors, including Gleason score, cancer stage, and serum PSA level. However, compared with men of normal weight, obese men were somewhat less likely to undergo RP (51% vs 64%) and were somewhat more likely to have ADT added to RP or radiation (8% vs 5% and 10% vs 5%, respectively).
Table 2. Demographic and Clinical Characteristics of 752 Men With Prostate Cancer by Body Mass Index
BMI indicates body mass index; SD, standard deviation; PSA, prostate-specific antigen; ADT, androgen-deprivation therapy; RP, radical prostatectomy.
P value (chi-square test for categorical variables or t test for continuous variables).
Mean ± SD age at diagnosis, y
57.6 ± 4.6
57.6 ± 4.8
57.4 ± 5
Stage at diagnosis
Gleason score at diagnosis
7 (3 + 4)
7(4 + 3) or 8–10
PSA level at diagnosis, ng/mL
RP without ADT
RP with ADT
Radiation without ADT
Radiation with ADT
Table 3 provides the adjusted HRs for associations of demographic and clinical characteristics with the risk of prostate cancer-specific and other-cause mortality. The highest relative HRs, as expected, were among men with higher Gleason scores and regional- or distant-stage disease and among men who received radiation and ADT either alone or combined. The serum PSA level at diagnosis did not predict outcome after controlling for other prognostic factors.
Table 3. Associations of Demographic and Clinical Characteristics With the Risk of Prostate Cancer-specific and Other-cause Mortality
Table 4 provides adjusted HRs for associations of BMI with the risk of prostate cancer-specific and other-cause mortality. After controlling for age, race, smoking status, and clinical prognostic factors at baseline, obesity was associated with a hazard of prostate cancer mortality of 2.64 (95% CI, 1.18–5.92) relative to men with normal BMI. Associations of obesity with prostate cancer-specific death differed modestly across strata defined by Gleason score and disease stage, with stronger positive associations observed in obese men with higher Gleason scores or localized stage at diagnosis. Results stratified by treatment and within treatment subgroups, including men who received only ADT, were consistent with the overall HR of 2.64. There was some suggestion that the HR was greater among men who did not receive ADT (HR, 15.92), but the 95% CI was wide (1.37–85.18), and the interaction testing ADT versus no ADT was not statistically significant (P = .32). We also examined this association among men who were undergoing RP alone and found similar results (BMI 25–29.9 kg/m2: HR, 10.8; 95% CI, 1.1–106.2; BMI 30 kg/m2: HR, 7.4; 98% CI, 0.4–148.2).
Table 4. Associations of Body Mass Index With the Risk of Prostate Cancer-specific and Other-cause Mortality Stratified by Gleason Score, Stage of Disease at Diagnosis, and Primary Treatment
Table 5 provides adjusted HRs for associations of demographic and clinical characteristics with the risk of prostate cancer metastasis. Similar to findings for prostate cancer-specific mortality, both higher Gleason scores and regional stage were associated with increased risks of disease metastasis. The association with Gleason score was particularly strong: Compared with patients who had Gleason scores from 2 to 6, patients who had a Gleason score of 7 (3 + 4) were at 2.43-fold increased risk of metastasis, and patients who had Gleason scores ≥7 (4 + 3) were at 10.27-fold increased risk of metastasis. Current smoking and high serum PSA levels (≥20 ng/mL) at diagnosis also were associated significantly with an increased risk of prostate cancer metastasis.
Table 5. Associations of Demographic and Clinical Characteristics With the Risk of Prostate Cancer Metastasis Among Men With Local- or Regional-stage Disease at Diagnosis
HR indicates hazard ratio; 95% CI 95% confidence interval; BMI, body mass index; PSA, prostate-specific antigen; RP, radical prostatectomy; ADT, androgen-deprivation therapy.
This category included patients who had a positive bone scan, bone/lymph node biopsy, or magnetic resonance image/computed tomography scan that showed cancer after primary treatment and men who were diagnosed with localized or regional-stage disease who also died of prostate cancer prior to the follow-up survey.
HRs were adjusted for age at diagnosis.
HRs were adjusted for age at diagnosis, race, smoking status, Gleason score, stage at diagnosis, serum PSA levels, and primary treatment.
Men with missing PSA data were excluded from the trend test.
Table 6 provides adjusted HRs for associations of BMI with the risk of prostate cancer metastasis. Overall, the risk of metastasis increased with higher BMI (P for trend = .0006), and obese men had a 3.61 relative hazard of metastases compared with men of normal weight. In stratified analyses, results did not differ significantly across strata defined by Gleason score, cancer stage, or primary treatment. However, HR point estimates were higher among men with higher Gleason scores or regional stage at diagnosis and among men who received ADT.
Table 6. Associations of Body Mass Index With the Risk of Prostate Cancer Metastasis Among Men With Local- or Regional-stage Disease at Diagnosis Stratified by Gleason Score, Disease Stage, and Primary Treatment
HR indicates hazard ratio; 95% CI 95% confidence interval; BMI, body mass index; ADT, androgen-deprivation therapy.
This category included men who had a positive bone scan, bone/lymph node biopsy, or magnetic resonance image/computed tomographic scan that showed cancer after primary treatment and men who were diagnosed with local or regional stage disease who also died of prostate cancer prior to the follow-up survey.
HRs were adjusted for age at diagnosis.
HRs were adjusted for age at diagnosis, race, smoking status, Gleason score, serum PSA levels, stage at diagnosis, and primary treatment.
HRs were adjusted for age, race, smoking status, Gleason score, and stage at diagnosis. Serum PSA levels and other treatments could not be entered into the model because of the small numbers of events in certain subgroups.
In this cohort of 752 patients with prostate cancer who were diagnosed in midlife, obesity was associated with a 2.6-fold increased risk of prostate cancer-specific mortality and a 3.6-fold increase in risk of disease metastasis among men who were diagnosed with local- or regional-stage cancer. Associations of obesity with prostate cancer-specific mortality or metastasis did not differ significantly across strata defined by Gleason score, stage at diagnosis, or primary treatment. However, the increased risks for both prostate cancer-specific death and metastasis associated with obesity were greater among men with higher Gleason scores. There was a suggestion that the association of obesity with disease progression was stronger among men who had regional-stage disease at diagnosis compared with men who had local-stage disease.
Our results are somewhat consistent with epidemiologic cohort studies that have examined the risk of prostate cancer-specific mortality in initially healthy men.1, 2, 21–23 Among those studies, in the Swedish construction worker study, a 40% increase was observed in the risk of prostate cancer death in obese men compared with men of normal weight22; and, in the 2 large American Cancer Society cohort studies, it was observed that obese men had a 27% and 21% increased risk of prostate cancer death compared with men of normal weight.1 Our results also are consistent with many studies that have reported an association between obesity and an increased risk of biochemical recurrence among men who undergo RP.9, 11, 24–26
The results of the current study were consistent with a recently reported finding that obesity was associated with a 2.4 relative hazard for prostate cancer-specific death from a cohort of 2367 men with prostate cancer who were followed for up to 23 years.27 However, our results were not consistent with a recent publication by Siddiqui et al.,14 who observed no associations between obesity and the risk of biochemical recurrence, the development of metastases, or prostate cancer death. There are several differences between our study and the study by Siddiqui et al that may explain, at least in part, the different findings. The study by Siddiqui et al. included patients who were diagnosed between 1990 and 1999, whereas our study included incident cases identified from a population-based cancer registry over 4 years beginning in 1993. It is possible that the increases in the prevalence of obesity, the use of adjuvant therapy, and PSA screening between 1990 and 1999 may obscure associations of obesity with outcomes in a study that accrued patients over this long period. The study by Siddiqui et al included only men who underwent RP at a single hospital, whereas our study included men regardless of primary treatment(s). Finally, our study patients were younger (age range, 40–64 years; median age, 58 years) than the patients reported by Siddiqui et al (median age, 65 years). It is possible that obesity is associated with poor prostate cancer outcomes in younger men only, although a mechanism for such an effect is difficult to formulate.
We carefully examined whether treatment with ADT could bias our results, because men who receive ADT have a poorer prognosis and are likely to gain weight because of their treatment. In the current study, we conducted baseline interviews as soon as practical after diagnosis (mean interval, 9 months) and asked men to recall their weight 1 year prior to their diagnosis. We used this approach to avoid any bias caused by the effects of diagnosis or treatment on reported weight; however, we recognize that a large weight gain caused by ADT could bias results nonetheless. We conducted 2 analyses to address this possibility. In the first analysis, we stratified by ADT use (yes vs no) and controlled for other treatments (prostatectomy, radiation, none), and we observed that the relative hazards for prostate cancer-specific mortality and for disease progression associated with obesity in both groups (ie, with and without ADT use) were consistent with the overall findings. In a second analysis, we examined the subgroup of men who received ADT treatment only, in which obesity was associated with a 3.2 relative hazard for prostate cancer-specific death with a stepwise increase in risk across categories of increasing BMI. Based on these results, we believe it is unlikely that our findings could be solely because of biased recall of weight after treatment with ADT.
Several mechanisms could explain associations of obesity with adverse prostate cancer outcomes. First, obesity in men is associated with altered steroid hormone concentrations, including lower levels of testosterone and sex hormone-binding globulin and higher levels of estrogens.28, 29 In recent studies, higher serum testosterone levels have been associated with a reduced risk of high-grade disease and an increased risk of low-grade disease,30–32 and higher serum estradiol levels have been associated with a decreased risk of nonaggressive cancer, but not aggressive cancer.33 Thus, the net effect of obesity-related changes in sex-hormone concentrations may be an increased risk of more advanced and poorly differentiated tumors and progression of androgen-independent tumors. Second, high levels of leptin and other adipokines in obese men also may affect prostate cancer outcomes.34 In vitro and in vivo experiments have demonstrated that leptin can promote angiogenesis, which is an important factor in the growth and spread of many cancers, including prostate cancer.35 There is also some evidence that high leptin levels are associated with prostate tumor progression and more advanced disease.36, 37 Finally, obesity may affect prostate cancer outcomes through an inflammation-related pathway. Adipose tissue produces inflammatory cytokines, such as tumor necrosis factor α and interleukins (IL) (eg, IL6, IL8), and obesity is associated with high levels of circulating inflammatory cytokines.38 Thus, chronic inflammation in obese men may play an important role in prostate tumor growth and progression.39, 40 The interrelations between obesity, steroid hormones, cytokines, and adipokines are complex, and it may not be possible to distinguish among these mechanisms in human research.
The current study has several limitations. The small number of African-Americans limits our ability to examine whether these findings differ between Caucasians and African-Americans. Height and weight 1 year prior to diagnosis were self-reported, which may introduce error. Multiple direct measures of obesity before and after diagnosis would be important to better understand the association in future studies. Another limitation is that cancer stage was defined by SEER using pathologic data for men undergoing RP and clinical information for all others. However, our findings were consistent across men who were treated with prostatectomy, radiation, or ADT, suggesting that this is not introducing an important bias. In addition, in stratified analyses, the numbers of deaths and metastasis events in each stratum were small, limiting the power to detect differences in associations across strata.
This study also has several strengths. First, we evaluated outcomes in a population-based sample regardless of treatment. Second, we used development of metastases as an endpoint rather than biochemical failure, which may not predict prostate cancer-specific death.12 Third, we had almost 1 decade of follow-up (median duration, 9.7 years) for a total of 752 prostate cancer patients. Thus, we are able to observe prostate cancer metastases and deaths over an extended period in the study cohort. Finally, causes of death were obtained from the SEER cancer registry as coded by the state nosologist and were verified by reviewing copies of death certificates.
In conclusion, our results support the hypothesis that obesity at the time of prostate cancer diagnosis is associated with an increased risk of developing metastases and prostate cancer-specific death. Associations were apparent regardless of stage or primary treatment, and there was some evidence that the effects of obesity on prostate cancer outcomes are stronger among men with higher Gleason scores at diagnosis. Although a randomized clinical trial would be needed to definitively determine whether weight reduction would be an effective adjunct treatment for men diagnosed with prostate cancer, these results provide yet 1 more important reason for men to adopt healthful patterns of diet and physical activity to achieve and maintain a normal weight.
Supported by National Institutes of Health (NIH) grants P50 CA97186 and RO1-CA56678, NIH contract N01-CN-05230 from the National Cancer Institute, and additional support from the Fred Hutchinson Cancer Research Center.