Influence of obesity on tumour volume in patients with prostate cancer

Authors


  • Take Home Message: Body mass index is independently associated with prostate cancer volume at radical prostatectomy.

Umberto Capitanio, University Vita-Salute San Raffaele, Via Olgettina 60, 20132, Milan, Italy. e-mail: capitanio.umberto@hsr.it

Abstract

Study Type – Prognosis (individual cohort)

Level of Evidence 2b

What's known on the subject? and What does the study add?

Obesity is associated with more aggressive prostate cancer.

Prostate cancer tumour volume is affected by excess weight, after adjustment for all possible clinical and pathological confounders.

OBJECTIVE

  • • To investigate the association between body mass index and tumour volume at radical prostatectomy in a large European population.

PATIENTS AND METHODS

  • • Recent data support the hypothesis that the hormonal environment in overweight and obese men may alter androgen-dependent prostate growth. Body mass index (BMI) has been implicated in prostate cancer pathophysiology.
  • • We analysed 1275 patients with prostate cancer who underwent radical prostatectomy at a single tertiary care institution. Mean tumour volume (TV) was evaluated according to BMI WHO categories (normal <25 kg/m2 vs overweight 25–30 kg/m2 vs obese 30–35 kg/m2 vs severely obese >35 kg/m2).
  • • Univariable linear regression analyses targeted the association between BMI and TV at radical prostatectomy. Multivariable analyses were adjusted for age, prostate-specific antigen value, biopsy Gleason sum, clinical stage and prostate volume.

RESULTS

  • • Mean BMI was 26.3 kg/m2 (median 26; range 16.7–42.0). Mean TV was 5.6 mL (median 3.3; range 0.1–61.2). The mean prostate-specific antigen value was 10.3 ng/dL (median 6.6; range 0.3–327).
  • • The mean TV was 5.0, 5.8, 6.3 and 9.2 mL in normal, overweight, obese and severely obese patients, respectively (P= 0.03). TVs in men with a normal BMI were 84% smaller than in severely obese men (5.0 vs 9.2 mL).
  • • On univariable analysis, BMI was correlated with TV at radical prostatectomy (P < 0.001). On multivariable analysis, BMI reached the independent predictor status after adjustment for age, prostate-specific antigen value, biopsy Gleason score, clinical stage and prostate volume (P= 0.03).

CONCLUSION

  • • We showed that BMI is independently associated with prostate cancer volume at radical prostatectomy. The present results confirm that obesity may play a key role in prostate cancer pathophysiology.
Abbreviations
PCa

prostate cancer

BMI

body mass index

RP

radical prostatectomy

TV

tumour volume

SVI

seminal vesicle invasion

LNI

lymph node invasion

PSMs

positive surgical margins

TZ

transitional zone.

INTRODUCTION

Obesity is increasingly recognized as a crucial risk factor for several cancers [1]. Overweight individuals are more likely to have changes in plasmatic levels of insulin-like growth factor 1, oestrogens, sex hormone binding globulin and testosterone [2–5]. These biological alterations may be involved in carcinogenesis and be the cause of increased mortality from cancer [2–9].

With regard to prostate cancer (PCa), the mechanisms underlying the link between obesity and tumour progression are not fully understood [6–8,10], and the evidence that body mass index (BMI) is connected with PCa is inconsistent. In a recent meta-analysis, Renehan et al.[1] showed that BMI is weakly connected with prostate carcinogenesis relative to other tumours – namely gastroenteric tract neoplasms. Conversely, several clinical studies have suggested that an increased BMI predisposes individuals to more unfavourable PCa grade and stage and/or to a greater biochemical recurrence rate after radical prostatectomy (RP) [11–15].

To date, only one American retrospective study tested the association between obesity and tumour volume (TV), showing that higher BMI is associated with increased cancer size [16]. However, country-specific lifestyle habits, dietary considerations and the prevalence of obesity in the US population may limit such results to the North American population. Our underlying hypothesis is that excess weight is associated with larger prostate tumours independent of obesity prevalence, race and continent of origin. Consequently, we decided to test the relationship between obesity and cancer volume in a large European population.

METHODS

After obtaining institutional review board approval, data from 1275 consecutive Caucasian patients treated with RP and pelvic lymphadenectomy between February 2006 and August 2009 at Vita-Salute San Raffaele were included in the present analyses. Clinical stage was assigned by the attending urologist according to the 2009 TNM classification [17]. The pretreatment PSA level (AxSYM PSA assay; Abbott Laboratories, Abbott Park, IL, USA) was measured before DRE and TRUS.

All RP specimens were inked, mounted whole, and step-sectioned at 3-mm intervals according to the Stanford protocol [18]. All sections were examined entirely by microscopic evaluation by the same dedicated uropathologist (M.F.). The cancer-bearing areas were manually mapped on each section. Paraffin-embedded haematoxylin and eosin stained sections were prepared. According to the College of American Pathologists guidelines (http://www.cap.org/apps/docs/committees/cancer/cancer_protocols/2009/Prostate_09protocol.doc) TV was calculated by visual inspection. In brief, the percentage involvement of each slide was visually estimated and the assessment of TV for the entire prostate was accomplished by summing and averaging the area on each slide and then multiplying by the specimen weight. Eyeball measurement allowed the overall TV to be calculated, also considering multifocality and irregular shapes. The RP specimens were assessed for the presence of extracapsular extension, which was diagnosed if cancer was evident outside the prostatic capsule and the seminal vesicles and lymph nodes were free of tumour. Seminal vesicle invasion (SVI) was diagnosed if the tumour invaded the muscular wall of one or both seminal vesicles without evidence of lymph node invasion (LNI). LNI was assigned if one or several pelvic lymph nodes were involved with cancer. Non-organ-confined disease included extracapsular extension and/or SVI and/or LNI.

BMI, abstracted from the perioperative anaesthesia records and based on patient report, was calculated from weight/height (kg/m2). The effect of BMI on TV was examined, categorized and continuously coded. First, we relied on BMI categorized according to the WHO divisions (normal <25 kg/m2 vs overweight 25–30 kg/m2 vs obese 30–35 kg/m2 vs severely obese >35 kg/m2). Clinical and pathological characteristics among BMI groups were compared using anova or Kruskall–Wallis for continuous variables, depending on whether they were normally distributed. The chi-squared test was used for categorical variables.

Second, we relied on continuously coded BMI. Specifically, univariable linear regression analyses targeted the association between BMI and TV at RP. Multivariable linear regression analyses were adjusted for age, PSA level, biopsy Gleason sum, clinical stage and prostate volume. We used scatter plots with the intent of providing a graphical display of the association between TV and BMI. Finally, univariable logistic regression analyses targeted the association between BMI and positive surgical margins (PSMs), SVI and LNI. Multivariable analyses were adjusted for other possible confounders at diagnosis, namely PSA level, biopsy Gleason sum and clinical stage. Statistical Package for the Social Sciences (SPSS, Chicago, IL, USA) version 15.0 was used for all analyses. All tests were two-sided, with a significance level set at P < 0.05.

RESULTS

Table 1 shows the descriptive characteristics of the total cohort of 1275 men. Mean age was 64.7 years (median 65, range 39–85). Mean BMI was 26.3 (median 26, range 16.7–42.0). The mean TV was 5.6 mL (median 3.3, range 0.1–61.2). The mean TV was 5.0, 5.8, 6.3 and 9.2 mL in normal, overweight, obese and severely obese patients, respectively (P= 0.03). TV in men with a normal BMI was 84% smaller than in severely obese men (5.0 vs 9.2 mL) (Fig. 1). Mean prostate and transitional zone (TZ) volumes were 47.2 and 25.4 mL, 55.7 and 33.4 mL (P < 0.001), 59.0 and 35.1 mL and 53.8 and 30.0 mL (P= 0.001) in normal, overweight, obese and severely obese men, respectively. The mean percentage of the prostate involved with cancer was 12.6%, 12.6%, 14.5% and 15.2% in normal, overweight, obese and severely obese patients, respectively (P= 0.8). The prevalence of pathological Gleason sum 8–10 was 10.9%, 12.3%, 14.8% and 20.0% in normal, overweight, obese and severely obese patients, respectively (P= 0.2). The prevalence of SVI and LNI was lower in normal-weight patients relative to their overweight, obese and severely obese counterparts (Table 1, both P= 0.04). PSA was higher in obese men (Table 1). However, PSA levels were not significantly different (P= 0.6) across BMI categories after adjustment for age, prostate volume, pathological Gleason sum, extracapsular extension, PSMs and SVI.

Table 1.  Descriptive characteristics of the study population
 OverallNormal(<25 kg/m2)Overweight(25–30 kg/m2)Obese(30–35 kg/m2)Severely obese(>35 kg/m2)P
  1. Results were stratified according to BMI WHO categories (normal <25 kg/m2 vs overweight 25–30 kg/m2 vs obese 30–35 kg/m2 vs severely obese >35 kg/m2). n= 1275.

Number of patients (%)1275 (100.0)468 (36.7)657 (51.5)135 (10.6)15 (1.2)NA
Age (median), range, years64.7 (65), 39–8564.7 (65), 39–8364.6 (65), 46–8564.9 (65), 41–7763.0 (63.2), 45–720.8
PSA (median), range, ng/mL10.3 (6.6), 0.3–3279.5 (6.3), 0.7–2739.8 (6.7), 0.3–15214.7 (7.0), 2.0–32714.2 (6.4), 0.5–810.02
Biopsy Gleason sum     0.4
 6796 (62.4)296 (63.2)413 (62.9)79 (58.5)8 (53.3)
 7341 (26.7)120 (25.6)180 (27.4)35 (25.9)6 (40.0)
 8–10119 (9.3)45 (9.6)54 (8.2)19 (14.1)1 (6.7)
 Missing data19 (1.5)7 (1.5)10 (1.5)2 (1.5)0 (0.0)
Prostate volume (median), range, mL52.8 (46), 5–23347.2 (42.3), 5–14755.7 (49.0), 10–23359.0 (53.0), 10–16053.8 (48.0), 33–86<0.001
TZ volume (median), range, mL30.7 (26.0), 1–15025.4 (22), 1–10733.4 (30), 1–15035.1 (27), 1–13030.0 (30), 11–490.001
Clinical stage     0.02
 T1c759 (357)263 (56.2)404 (61.5)81 (60.0)11 (73.3)
 T2357 (28.0)160 (34.2)161 (24.5)34 (25.2)2 (13.3)
 T388 (6.9)25 (5.3)50 (7.6)12 (8.9)1 (6.7)
 Missing data71 (5.6)20 (4.3)42 (6.4)8 (5.9)1 (6.7)
Pathological Gleason sum     0.2
 6476 (37.6)181 (38.7)252 (38.4)39 (28.9)7 (46.7)
 7573 (44.9)220 (47.0)294 (44.7)57 (42.2)2 (13.3)
 8–10155 (12.2)51 (10.9)81 (12.3)20 (14.8)3 (20.0)
 Missing data68 (5.3)16 (3.4)30 (4.6)19 (14.1)3 (20.0)
Organ confined disease911 (71.5)341 (73.0)472 (71.8)89 (65.9)9 (60.0)0.3
SVI171 (13.4)60 (12.8)80 (12.2)27 (20.0)4 (26.7)0.04
LNI133 (10.4)41 (8.8)70 (10.7)17 (12.6)5 (33.3)0.04
PSMs326 (25.6)113 (24.1)163 (24.8)42 (31.1)8 (53.3)0.02
TV (median), range, mL5.6 (3.3), 0.1–61.25.0 (3.0), 0.2–50.45.8 (3.3), 0.1–61.26.3 (3.8), 0.4–48.89.2 (6.0), 0.5–35.00.03
Figure 1.

Scatterplot illustrating the correlation between BMI and TV in 1275 patients with PCa treated with RP.

Because patients with higher BMI also showed higher PSA levels, larger prostates and larger TZ volumes, we examined the correlation among these variables (Table 2). We found that PSA correlated with larger prostates (Spearman: 0.16; P < 0.001) and, more interestingly, with larger cancers (Spearman: 0.38; P < 0.001) and a higher percentage of tumour involvement (Spearman: 0.24; P < 0.001). We found that larger prostate size correlated with smaller tumours (Spearman: −0.12; P= 0.001) and smaller percentage tumour involvement (Spearman: −0.48; P < 0.001).

Table 2.  Overall correlation between BMI, age, PSA, per cent of cancer involvement, prostate size, TZ amount and TV within the study population
 BMIAgePSAProstate volumeTZ volumePer cent of cancer involvementTV
SpearmancoefficientPSpearmancoefficientPSpearmancoefficientPSpearmancoefficientPSpearmancoefficientPSpearmancoefficientPSpearmancoefficientP
  1. n= 1275.

BMINANA−0.010.80.080.0050.110.0020.110.010.060.10.11<0.001
Age−0.010.8NANA0.020.40.090.0090.090.060.040.30.050.07
PSA0.080.0050.020.4NANA0.16<0.0010.150.0010.24<0.0010.38<0.001
Prostate volume0.110.0020.090.0090.16<0.001NANA0.95<0.001−0.48<0.001−0.120.001
TZ volume0.110.010.090.060.150.0010.95<0.001NANA−0.45<0.001−0.110.02
Per cent of cancer involvement0.060.10.040.30.24<0.001−0.48<0.001−0.45<0.001NANA0.91<0.001
TV0.11<0.0010.050.070.38<0.001−0.120.001−0.110.020.91<0.001NANA

Table 3 shows univariable and multivariable analyses predicting the presence of PSMs, SVI and LNI at RP. After adjustment for all available clinical characteristics, BMI was an independent predictor of unfavourable characteristics at RP (PSMs, SVI and LNI).

Table 3.  Univariable and multivariable logistic regression analyses predicting SVI, LNI and PSMs relying on clinical characteristics in 1275 patients treated with RP and pelvic lymphadenectomy
PredictorsUnivariable analysisMultivariable analysis
SVILNIPSMsSVILNIPSMs
ORPORPORPORPORPORP
  1. OR, odds ratio.

BMI1.060.011.070.011.060.0031.050.051.070.041.060.01
PSA1.05<0.0011.04<0.0011.02<0.0011.03<0.0011.03<0.0011.020.003
Biopsy Gleason score <0.001 <0.001 <0.001 <0.001 <0.001 <0.001
7 vs 65.37<0.0016.09<0.0012.30<0.0014.06<0.0014.15<0.0011.83<0.001
8–10 vs 616.64<0.00118.96<0.0012.71<0.00111.41<0.00112.81<0.0011.99<0.001
Clinical stage <0.001 <0.001 <0.001 0.003 0.001 0.03
T2 vs T1c2.13<0.0012.53<0.0011.56<0.0011.400.11.730.0031.380.04
T3 vs T1c8.92<0.0019.97<0.0013.03<0.0012.750.0013.37<0.0011.790.03

Table 4 shows univariable and multivariable linear regression analyses predicting TV at RP. With univariable analyses (Table 4), BMI results correlated with TV at RP (B= 0.2; P= 0.001). Moreover, PSA level, biopsy Gleason score and clinical stage were associated with cancer size at RP as well (all P < 0.001). Conversely, patient age and prostate volume did not predict TV at RP (P > 0.4). On multivariable analysis, after adjustment for age, PSA level, biopsy Gleason score, clinical stage and prostate volume, BMI reached the independent predictor status (P= 0.04) (Table 4).

Table 4.  Univariable and multivariable linear regression analyses predicting PCa volume in 1275 patients treated with RP and pelvic lymphadenectomy
PredictorsUnivariable analysesMultivariable analyses
B coefficientPB coefficientP
BMI0.20.0010.140.04
Age0.020.4−0.070.8
Diabetes1.410.050.870.3
PSA0.11<0.0010.10<0.001
Biopsy Gleason score3.30<0.0012.49<0.001
Clinical stage3.06<0.0011.58<0.001
Prostate volume−0.010.50.070.4

Table 5 shows the descriptive characteristics of overweight, obese and severely obese patients after stratification for low (≤3.4 mL, below the median value) and high (>3.4 mL, above the median value) TV. At diagnosis, overweight patients with high TV showed smaller prostate (50.8 vs 61.2 mL), higher PSA levels (13.5 vs 8.0 ng/mL) and higher biopsy Gleason 8–10 prevalence (12.7% vs 5.7%) relative to their counterparts with low TV at RP (all P < 0.001; Table 5). No difference regarding patient age was recorded between overweight patients with low or high TV (P= 0.2; Table 5).

Table 5.  Descriptive characteristics of overweight, obese and severely obese patients (n= 807)
 Overall (BMI >25 kg/m2)Low TV (≤3.4 mL)High TV (>3.4 mL)P
  1. Results were stratified according to PCa volume (dichotomized according to median TV).

Number of patients (%)807 (100.0)406 (50.3)401 (49.7)NA
Age (median), range, years64.6 (65), 40–8564.4 (65), 45–8564.9 (66), 40–810.2
PSA (median), range, ng/mL10.7 (6.7), 0.3–3278.0 (5.9), 0.3–18013.5 (7.9), 0.9–327<0.001
Biopsy Gleason sum   <0.001
 6500 (62.0)284 (70.0)216 (53.9)
 7221 (27.4)91 (22.4)130 (32.4)
 8–1074 (9.2)23 (5.7)51 (12.7)
 Missing data12 (1.4)8 (2.0)4 (1.0)
Prostate volume (median), range, mL56.2 (49), 10–23361.2 (56), 10–23350.8 (42), 13–171<0.001
TZ volume (median), range, mL33.6 (29), 1–15038.9 (33), 1–15029.5 (22), 1–1130.007
Clinical stage   <0.001
 T1c496 (61.5)279 (68.7)217 (54.1)
 T2197 (24.4)95 (23.4)102 (25.4)
 T363 (7.8)18 (4.4)45 (11.2)
 Missing data51 (6.3)14 (3.4)37 (9.2)
Pathological Gleason sum   <0.001
 6298 (36.9)209 (51.5)89 (22.2)
 7353 (43.7)147 (36.2)206 (51.3)
 8–10104 (12.9)16 (3.9)88 (21.9)
 Missing data52 (6.4)34 (8.4)18 (4.5)
Organ-confined disease570 (70.6)350 (86.2)220 (54.9)<0.001
SVI111 (13.8)16 (3.9)95 (23.7)<0.001
LNI92 (11.4)13 (3.2)79 (19.7)<0.001
PSMs213 (26.4)51 (12.6)162 (40.4)<0.001
TV (median), range, mL5.9 (3.4), 0.1–61.21.7 (1.7), 0.1–3.410.2 (6.7), 3.42–61.2<0.001

DISCUSSION

To date, the association between PCa and obesity has remained a controversial issue [19–21]. However, there is an emerging consensus that obesity may be associated with more aggressive PCa, as evidenced by higher-grade disease in obese patients undergoing RP, greater biochemical recurrence risk after surgery [11–13,22,23] and greater risk for PCa-specific mortality [14,15,24].

In the event that metabolic and hormonal changes represent the basis for prostate carcinogenesis, we hypothesized that TV – a proxy of cancer growth – may be affected by excess weight after adjustment for all possible clinical and pathological confounders. Several RP series have addressed the link between obesity and treatment outcomes [11–15,22–24], but only one American retrospective study has previously specifically tested the association between obesity and PCa volume, showing that higher BMI is associated with larger cancers [16]. However, such results were limited to the North American population because of country-specific lifestyle, race prevalence and specific dietary habits [22,25,26]. Indeed, the correlation between obesity and cancer volume that was shown might be biased by the environment, genetic factors and racial characteristics found in the US population. In conclusion, no previous study tested the hypothesis that excess weight is associated with larger prostate tumours independent of environment, race and continent of origin.

First, we confirmed that BMI, PSA levels, prostate volume, TZ volume, PCa volume and per cent tumour involvement are robustly correlated with each other (Table 2). Consequently, multivariable analyses accounting for all possible confounders are mandatory for testing the association between obesity and TV. Second, we confirmed that BMI is an independent predictor of unfavourable characteristics at RP (e.g. PSMs, SVI and LNI) as well.

More importantly, our multivariable adjusted results confirm a strong correlation between obesity and TV, also after correction for all possible significant confounders. Interestingly, we found very similar PCa volumes across all the WHO BMI categories between our European results and the American data reported by Freedland et al.[16] (5.0, 5.8, 6.3 and 9.2 mL vs 5.6, 6.1, 6.2 and 7.6 mL in normal, overweight, obese and severely obese patients, respectively).

Finally, a correlation between obesity and tumour grade is also suggested in the present study, although a significant difference was not achieved. Indeed, the prevalence of Gleason 8–10 PCa was 10.9%, 12.3%, 14.8% and 20.0% in normal, overweight, obese and severely obese patients, respectively. Although a trend could be evidenced, a significant difference was not achieved probably because of the relatively small number of patients included in the obese and severely obese groups.

In our population, in descriptive analyses we found higher PSA levels among obese patients (Table 1). This facet may appear incongruent with previously published data showing lower PSA levels among obese patients [27]. However, in multivariable analyses, after adjusting for age, prostate volume, pathological Gleason sum, extracapsular extension, PSMs and SVI, PSA concentration was not significantly different across BMI categories (P= 0.6).

Our results are of particular interest considering recent basic science data regarding the relationship between adiposity and tumour progression [28–30]. Several possible underlying mechanisms for an association between PCa and BMI were reported. Indeed, several in vivo and in vitro studies suggested that the adiposity-related changes in metabolism and endogenous hormone levels may affect PCa growth [28,29]. More specifically, a significant metabolic consequence of obesity is a reduced tissue response to insulin, especially in terms of reduced uptake of glucose [30]. Insulin resistance leads to chronically elevated blood levels of insulin; insulin is a growth-enhancing hormone and thus a biologically plausible risk factor for cancer development and growth. We implemented these results, confirming that obesity is statistically and clinically linked with final TV at RP.

Finally, our study showed that patient age does not correlate with TV and percentage of cancer involvement (Table 2). Moreover, in univariable and multivariable analyses (Table 4), age did not predict TV at RP, excluding any possible biological mechanism between aging and final PCa volume.

The present study presents some limitations, however. First, we restricted our study to patients undergoing RP. However, an accurate TV assessment requires the complete removal of the gland. In addition, BMI was abstracted from the perioperative anaesthesia records and based on patient report. A measurement by healthcare personnel prior to surgery would have removed a potential bias. Anthropomorphic measurements, such as lean body mass or the waist-to-hip ratio, were not available. Finally, the inclusion of patients treated in the past 5 years precludes any attempt to report on oncological outcomes.

We have shown that BMI is independently associated with PCa volume at RP, and the present results confirm that hormonal metabolism may play a key role in PCa pathophysiology. Besides all available confounders, obesity might provide an intrinsic microenvironment that favours cancer growth, independent of lifestyle, dietary factors and country of origin.

CONFLICT OF INTEREST

None declared.

Ancillary