Stronger Association between Obesity and Biochemical Progression after Radical Prostatectomy among Men Treated in the Last 10 Years

Obesity is a major public health concern that is rapidly reaching epidemic proportions. The prevalence of obesity among adults in the United States has doubled from around 15% to over 30% in the last 20 years (1). Obesity has been associated with several types of cancer including postmenopausal breast and colon cancer (2). Moreover, obesity has been associated with increased risk of prostate cancer death in multiple prospective cohort studies, suggesting more aggressive disease among obese men (3–5).

The last 20 years has seen dramatic changes in prostate cancer detection and management including the increased popularity, safety, and reduced morbidity of radical retropubic prostatectomy (RRP; refs. 6, 7), the introduction of prostate-specific antigen (PSA)–based staging (8), and ultimately screening (9) for early-stage asymptomatic prostate cancer leading to a dramatic stage migration (10). These changes have had a major impact of the typical patient undergoing RRP. However, whether these temporal changes, along with the increasing prevalence and extent of obesity, have affected the association between obesity and aggressive prostate cancer is unknown.

We sought to examine whether the association between obesity and outcomes after RRP had changed as a result of the introduction of PSA-based screening and the increasing prevalence and extent of obesity. To accomplish this, we examined temporal changes in the association between obesity and pathologic grade, stage, and biochemical progression after RRP among men treated over a 20-year time span by a single high-volume surgeon at a tertiary care referral center.

Study population and assessment of body mass index and other clinicopathologic variables. After obtaining Institutional Review Board approval and informed consent when appropriate, 3,341 consecutive patients treated with anatomic RRP for prostate adenocarcinoma by one surgeon (P.C.W.) from 1985 to 2004 at the Johns Hopkins Hospital were identified. Men treated with preoperative hormonal (luteinizing hormone releasing hormone agonist, antiandrogen, or 5-α reductase type II inhibitor) therapy (n = 31), chemotherapy (n = 2), or radiation therapy (n = 4) were excluded. Men diagnosed from a transurethral resection specimen (clinical stage T_1a/T_1b) were excluded (n = 167). To ensure that all men had equal opportunity for PSA follow-up, men without a preoperative PSA (n = 132 of which 117 were treated in 1985 or 1986), suggesting poor PSA utilization, were excluded. Preoperative body mass index (BMI; weight in kilograms divided by height in meters squared), abstracted from the medical records, was available on 2,832 men and they constitute the study population. Men who received adjuvant radiation (n = 10) before biochemical progression (a single PSA ≥0.2 ng/mL; ref. 11) were censored at the time of adjuvant therapy.

The prostatectomy specimens were sectioned as previously described (12). Prostate weight was determined by measurement of the gross RRP specimen weight, including seminal vesicles and tips of the vasa. Follow-up included PSA measurements quarterly for year 1, semiannually for year 2, and yearly thereafter.

Statistical analysis. We explored differences in the distribution of clinical characteristics among the BMI (kg/m²) groups of normal weight (<25), overweight (25-29.9), obese (30-34.9), and moderately and severely obese (≥35). Due to the limited number of men in the moderately and severely obese group (n = 25), these men were combined with the obese group. No statistically significant differences were found between normal weight and overweight men in the distribution of any clinicopathologic features (all P ≥ 0.08) or risk of biochemical progression (log-rank, P = 0.86). Therefore, the normal weight and overweight groups were combined and BMI was evaluated as a binary variable of non-obese (BMI <30 kg/m²) versus obese (BMI ≥30 kg/m²).

Differences in the distribution of clinicopathologic characteristics between the BMI groups were compared using the Wilcoxon rank sum test for continuous variables and the χ² test for categorical variables. The odds ratio of the following binary pathologic outcomes was estimated for obesity using logistic regression: high-grade disease (Gleason sum ≥4 + 3), positive surgical margins, and extraprostatic extension. There were few men with seminal vesicle invasion or lymph node metastases. We adjusted for preoperative PSA (continuous variable), age at RRP (continuous), race (black, other, versus white), height (tertiles: ≤69, 69.1-71.9, ≥72 in.), year of surgery (continuous variable), and prostate weight (continuous variable). In the model predicting high-grade disease, we also adjusted for clinical stage (T_2a, T_2b, T_2c, T_3a versus T_1c), surgical margin status, extraprostatic extension, seminal vesicle invasion, and lymph node metastasis. For predicting the other binary pathologic outcomes, we also adjusted for pathologic Gleason sum (3 + 4, ≥4 + 3 versus 2-6). Because the data for preoperative PSA and prostate weight were not normally distributed, we examined the data after logarithmic transformation. To examine whether the association between BMI and pathologic findings had changed over time, patients were divided into groups based on 5-year periods of time: 1985 to 1989 (n = 394), 1990 to 1994 (n = 653), 1995 to 1999 (n = 985), and 2000 to 2004 (n = 800).

Time to biochemical progression was compared between the BMI categories using Kaplan-Meier plots and the log-rank test. To estimate the relative risk (RR) of progression associated with obesity, we used a Cox proportional hazards regression model. We ran two separate multivariable Cox models to examine whether obesity was associated with biochemical progression and whether this association was independent of any association between obesity and pathologic findings. The first model adjusted for only preoperative clinical characteristics (age, race, height, year of surgery, clinical stage, biopsy Gleason sum, and preoperative PSA), whereas the second model adjusted for preoperative clinical characteristics as well as pathologic variables (prostate weight, pathologic Gleason sum, positive surgical margins, extraprostatic extension, seminal vesicle invasion, and lymph node metastasis). Because men treated from 2000 to 2004 had limited follow-up (mean and median: 1.8 and 1.0 years, respectively) and a low number of recurrences (n = 16 or 2%), these men were combined with the group of men treated between 1995 and 1999 for statistical analyses examining the association between obesity and progression. The point estimates for all crude and age-adjusted models for predicting both the binary pathologic outcomes and time to progression were similar to the multivariable adjusted point estimates and therefore only results for the multivariable models are shown. All statistical analyses were done using STATA 8.0 (Stata Corp., College Station, TX).

Patient demographics. Mean ± SD and median BMI were 26.1 ± 2.9 and 25.8 kg/m², respectively. A total of 1,023 men (36%) were normal weight, 1,536 (54%) were overweight, and 273 (10%) were obese. There was a weak positive association between more recent year of surgery and BMI (Spearman r = 0.05, P = 0.02). Obesity was associated with black race (P = 0.003), younger age (P < 0.001), advanced clinical stage (P = 0.02), and positive surgical margins (P = 0.01; Table 1). Given that preoperative PSA and prostate weight were related to age (Spearman r = 0.16, P < 0.001 and r = 0.33, P < 0.001, respectively) and obese men were younger, we examined the association between obesity and preoperative PSA and prostate weight using linear regression adjusting for age. After adjustment for age, there was a statistically significant association between obesity and larger prostate weight (P = 0.002) but not preoperative PSA (P = 0.34).

Pathologic stage and grade. After multivariable adjustment for clinicopathologic characteristics, there was a statistically significant positive association between obesity and positive surgical margins and extraprostatic extension as well as a modest, but not statistically significant, association with high-grade disease (Table 2). Over time, the strength of the association between obesity and high-grade disease increased and approached statistical significance among men treated between 2000 and 2004. On the contrary, the association between obesity and positive surgical margins decreased over time; obesity was only statistically significantly associated with positive surgical margins among men treated before 1995. A less clear pattern was noted for changes over time in the association between obesity and extraprostatic extension, although the strongest association and only time point for which the association was statistically significant was among men treated from 2000 to 2004.

Biochemical progression. Mean ± SD and median follow-up among men without biochemical progression were 6.0 ± 4.3 and 5 years (range 1-19), respectively. During this time, 374 patients (14%) progressed. After multivariable adjustment for preoperative clinical characteristics, obesity was associated with an increased risk of biochemical progression, although the association was not statistically significant (RR, 1.36; 95% confidence interval (95% CI), 0.98-1.89; P = 0.07; Fig. 1; Table 3). The association between obesity and progression remained unchanged when pathologic characteristics were also considered in the multivariable model (RR, 1.36; 95% CI, 0.98-1.90; P = 0.07; Table 3).

There were no statistically significant associations between obesity and biochemical progression among men treated between 1985 and 1989 or among men treated between 1990 and 1994 (Table 3). Among men treated since 1995, obesity was associated with a statistically significant increased risk of biochemical progression when adjusting for preoperative clinical characteristics (RR, 1.90; 95% CI, 1.90-3.30; P = 0.02; Fig. 2). After further adjustment for postoperative pathologic features, obesity remained associated with an increased risk of progression among men treated since 1995, although the risk was attenuated and no longer statistically significant (RR, 1.55; 95% CI, 0.88-2.73; P = 0.13).

The influence of the introduction of PSA-based screening on the association between obesity and aggressive prostate cancer is not known. To examine this, we analyzed outcomes among men treated by RRP by a single surgeon over the past 20 years. With the exception of surgical margin status, the positive association between obesity and high-grade disease, advanced stage, and biochemical progression after RRP was generally strongest among men treated in the last 10 years. This suggests that factors possibly related to PSA-based screening have strengthened the positive association between obesity and aggressive prostate cancer among men undergoing RRP.

Prior studies have found that obese men are at increased risk of prostate cancer death (4, 5, 13). Among men undergoing RRP, a prior report on a subset of men in the current data set (14), as well other prior reports (15, 16), found that obese men had higher-grade cancers at the time of RRP. Moreover, prior studies have also found a positive association between BMI and biochemical progression (15, 16). Together these studies suggest that obese men in general and specifically men undergoing RRP may have more aggressive disease. Potential reasons for more aggressive prostate cancer among obese men include alterations in the balance of serum hormonal concentrations (estrogen, testosterone, insulin, and leptin; refs. 17–19), as well as dietary and lifestyle factors (20–22).

We hypothesized that temporal changes in prostate cancer diagnosis as a result of PSA-based screening, along with the dramatic increase in the extent of obesity (1), may have changed the strength of the association between obesity and aggressive disease over time. To address this we examined outcomes among men treated by RRP over a 20-year time by dividing patients into 5-year blocks. The earliest time period studied, 1985 to 1989, represents a time when routine PSA screening was uncommon. Although all men in this study had preoperative PSA concentrations available, the lack of men with T_1c disease during this time period (10) attests to the lack of PSA screening. In contrast, the most recent time period, 2000 to 2004, represents a time when nearly all men in our study underwent PSA-based screening and many had likely been serially screened for years before diagnosis. When patients were broken into these 5-year intervals, in general, we found a stronger association between obesity and high-grade cancer, advanced stage, and biochemical progression among men treated more recently. Moreover, after adjusting for the increased odds of high-grade cancer and advanced stage among obese men treated in the last 10 years, there remained a positive, albeit slightly attenuated, association between obesity and biochemical progression (RR, 1.90 down to 1.58). Thus, although the positive association between obesity and high-grade and advanced cancer may explain some, it cannot explain all of the increased risk of biochemical progression observed among obese men in the last 10 years.

There are several possible nonmutually exclusive reasons for the stronger association between obesity and aggressive prostate cancer among men undergoing RRP over time. First, before the introduction of PSA screening, screening for prostate cancer was uncommon and most cancers were detected due to a work-up resulting from urinary symptoms, which generated concern for an underlying cancer. However, the symptoms were often related to benign prostatic hyperplasia. Given that obese men have larger-sized prostates (23, 24) and greater urinary symptoms than normal weight men (25), it is entirely possibly that obese men were more likely to consult their physicians for urinary symptoms and undergo a work-up for possible prostate cancer. These greater urinary symptoms among obese men would have generated more opportunities for prostate cancer detection and would result in cancers being detected earlier, attenuating the potential biological association between obesity and aggressive disease. Indeed, with the exception of positive surgical margins, we found no statistically significant associations between obesity and stage, grade, or progression among men treated in the earlier time periods.

Whereas it is possible there was a bias for increased detection of prostate cancers among obese men in earlier years, in more recent years we suggest that there may also be a bias against detection of cancers among obese men. This detection bias in more recent years may result from increasing evidence that obese men have lower PSA concentrations (26–29), presumably the result of lower androgenicity among obese men resulting in less PSA production (30), given that PSA production is under direct androgenic control (31). Lower PSA concentrations would make obese men less likely to have an abnormal PSA test and undergo biopsy resulting in fewer cancers detected. Another possible detection bias results from the fact that multiple studies, including the current, found that obese men have larger prostates (23, 24, 27). Prostatic enlargement would make detection of an existent cancer less likely, given an equal-sized tumor and an equal number of biopsy cores were obtained (32, 33). Combining lower PSA concentrations and prostatic enlargement would represent an inherent bias against detecting cancers among obese men. Consequently, if cancers are harder to detect among obese men, this may lead to delayed diagnosis and subsequently later-stage disease at the time diagnosis. Contrary to this hypothesis, in the current study obese men had similar PSA concentrations as non-obese men. However, it must be kept in mind that patients undergoing RRP represent only a subset of men diagnosed with cancer, which in turn represents only a subset of the general population. As such, the characteristics of men undergoing RRP may not reflect the general population. Therefore, similar PSA concentrations between obese and non-obese men in the current study do not necessarily imply that at the population level the PSA concentrations would also be similar between obese and non-obese men. For example, several population-based studies have indeed found lower PSA concentrations among obese men (26, 28, 29), although others have found no statistically significant differences in PSA concentration by BMI (34). Therefore, the argument that PSA-based screening is inherently biased against detecting cancers in obese men remains a hypothesis that requires further study.

An additional explanation for the current findings relates not to detection bias, but rather a differential association between obesity and the development of early-stage versus advanced prostate cancer. It has been suggested that independent of detection bias, obesity may be “protective” for developing prostate cancer in certain population subsets (35). In this prior study, obesity provided “protection” for early-stage disease, but not advanced-stage disease in those who were under the age of 60 years or had a family history of prostate cancer. The mechanism proposed for this protection was lower androgenic activity among obese men (30). Regardless of the mechanism, this raises the interesting hypothesis that obesity may influence the development of advanced-stage and early-stage prostate cancer differently: obesity may “protect” against early-stage disease, but not advanced disease. If true, on a population basis the overall number of obese men with early-stage disease may be reduced without any effect on the number of obese men with advanced disease. Therefore, when studying only men with prostate cancer (as in the current study), the net effect would be an increased relative percentage of obese men with advanced-stage disease due to decreased numbers of men with early-stage disease (and thus reduced overall numbers of men with prostate cancer) but an equal number of men with advanced-stage disease. In an era when only a minority of men undergoing RRP had early-stage disease (1985-1989; ref. 10), the protection provided by obesity for early-stage disease would be less appreciable, resulting in only a slight increase in the percentage of obese men with advanced disease. However, in an era when the vast majority of men undergoing RRP have early-stage disease (e.g., 1995-2004; ref. 10), the protection provided by obesity for early-stage disease would be more dramatic, resulting in a large relative increase in the percentage of obese men with advanced disease. Therefore, the overall stage migration in prostate cancer coupled with obesity being protective for early-stage disease could result in a stronger association between obesity and advanced disease over time. It is important to note that although in this prior study (35) the risk of early-stage prostate cancer was decreased only among a subset of men (<60 years old or positive family history), the majority of men in the current study were under the age of 60 (median age of 58) and therefore the findings of this prior study may be particularly applicable to the current cohort.

Although in general the association between obesity and aggressive disease was stronger over time, the association between obesity and positive surgical margins was weaker over time. Alternative hypotheses are needed to explain this finding. We also examined biochemical progression as an end point. Several studies found that among men treated with RRP, a shorter time to biochemical progression is associated with increased risk for developing metastatic disease (36, 37). Although, recent studies suggested that postoperative PSA doubling time may be a better surrogate than biochemical progression for prostate cancer–specific mortality, PSA doubling times were not available on all our patients (38, 39). Men who undergo RRP represent only a subset of all men diagnosed with prostate cancer. As such, there are many patient-based and surgeon-based selection factors in who receives which treatment option. Therefore, differences in baseline disease characteristics between obese and non-obese men may reflect this selection. The current cohort was younger, had a lower BMI, lower preoperative PSA, and less racial diversity than in other studies that examined the association between BMI and outcomes after RRP (15, 16). Whether the current findings are applicable to men treated in different practice settings needs to be assessed. The narrow range of BMI in the current study limited our ability to study men with a very high BMI (≥35 kg/m²), in whom it has previously been suggested there is a particularly high risk of progression (16). It is possible that obese men who altered their diet and lifestyle had a positive effect on tumor growth delaying their recurrences. It is anticipated that this bias, if present, would attenuate the association between obesity and progression. Therefore, the current findings may be a slight underestimate of the magnitude of the association between obesity and progression. Despite the overall large number of men in the current study (nearly 3,000), the relatively small number of obese men reduced our statistical power to detect differences between obese and non-obese men. Although obesity was associated with race, the study was underpowered to address the important question of whether the association between obesity and progression is different between black and white men. Ultimately, we do not know the reason for the apparent stronger association between obesity and aggressive prostate cancer over time nor for the weaker association with positive surgical margins over time. The possible explanations presented above are not mutually exclusive and certainly plausible, but are at best hypotheses. Further studies are needed, first to confirm the current findings and second to explore various explanations for these findings.

In the current study, over the last 20 years the positive association between obesity and high-grade cancer, advanced stage, and biochemical progression after RRP has in general become stronger. However, the association with positive surgical margins has become weaker. These findings need to be confirmed in other studies. The reasons for the stronger association between obesity and aggressive prostate cancer among men undergoing RRP over time in the current study are not clear, although factors possibly related to PSA-based screening and/or other temporal changes in prostate cancer diagnosis may play a role.