Influence of obesity on biochemical and clinical failure after external-beam radiotherapy for localized prostate cancer


  • Results from the study were presented at the 2005 Genitourinary SPORE meeting (February 1, 2005) and seminars at the Mayo Clinic (November 7, 2004) and Fred Hutchinson Cancer Research Center (April 20, 2005).



Several reports have shown that obesity is associated with increased risk of biochemical failure after radical prostatectomy. However, limited information is available regarding the impact of obesity on prostate cancer progression after radiotherapy. The current study sought to determine whether obesity was an independent predictor of biochemical failure (BF) and clinical recurrence (CF) among patients treated with external-beam radiotherapy (EBRT).


A retrospective analysis was performed on 873 patients receiving EBRT as the sole treatment for localized prostate cancer between 1988 and 2001. The Kaplan–Meier method, log-rank test, and Cox proportional hazards analyses were performed.


Of the 873 patients, 18% were mildly obese and 5% were moderately to severely obese. Obesity was related to younger age at diagnosis (P < .001), more recent year of diagnosis (P = .03), and race (P = .03), with African-American men having the highest obesity rates. During a mean follow-up of 96 months, 295 patients experienced BF and 127 had CF. On multivariate analysis, controlling for clinical and treatment characteristics, increased body mass index (BMI) significantly predicted BF (hazards ratio [HR] = 1.04; 95% confidence interval [95% CI], 1.02–1.07) with a positive trend by BMI category (P = .001). Similar results were found when the outcome was CF; BMI remained an independent predictor of progression (HR = 1.05; 95% CI, 1.01–1.09), with a statistically significant trend by increased BMI category (P = .03).


The current findings validate the important role of obesity, not only on BF but also on CF, and suggest a link to the biologic basis of tumor progression that can be therapeutically exploited. Cancer 2006. © 2006 American Cancer Society.

Obesity has become one of the major public health concerns in the U.S.1 It has been linked to risk of several types of cancer, but to our knowledge, the correlation between obesity and the incidence of prostate cancer remains unclear.2–5 There are several lines of evidence to suggest that diet and weight gain may be important environmental factors implicated in prostate carcinogenesis. Obesity has been associated with more aggressive variants of prostate cancer6–8 and with adverse outcome, including mortality.6–8 Recently, in a series of 526 prostatectomy patients, we reported that obesity and weight gained after age 25 years are associated with biochemical failure (BF).9 Other studies have reported similar results,10, 11 suggesting an association with a more aggressive form of prostate cancer. However, a new retrospective study reported that BMI had no impact on BF after permanent prostate brachytherapy.12 Experimental observations suggest that obesity affects sex steroid, insulin, and insulin-like growth factor (IGF)-1 pathways that in turn modulate prostate cancer progression.13, 14

To our knowledge, no studies to date have examined the correlation between obesity and prostate cancer progression after primary therapy with external-beam radiotherapy (EBRT). We therefore sought to determine whether obesity was an independent predictor of BF and clinical recurrence among patients treated with radiotherapy at The University of Texas M. D. Anderson Cancer Center.


Patient Characteristics

The use of institutional patient data was approved by the institutional review board of the University of Texas M. D. Anderson Cancer Center. The study subjects included patients with histologically confirmed adenocarcinoma of the prostate who received radiation therapy at M. D. Anderson between 1988 and 2001. From the institutional database, we identified 1376 consecutive patients who underwent full-dose EBRT for localized disease (T1c to T3, NX, M0), who had not received neoadjuvant or adjuvant therapy. Of these, 956 had height and weight documented in the chart at the time of initial evaluation for EBRT. We excluded 83 patients with <12 months of follow-up, leaving 873 subjects in the analysis. The 503 patients excluded had similar clinical characteristics (e.g., clinical stage, age at diagnosis, Gleason score, pretreatment serum prostate-specific antigen [PSA] and year of diagnosis) to the group with anthropometric measurements included in the analysis (data not shown).

Among the 873 patients included in the analysis, 766 were non-Hispanic white, 22 were Hispanic, 71 were African-American, and 14 were Asian. Clinical and pathologic information had been recorded during the initial visit. Data were available regarding medical history, physical examination (including digital rectal examination [DRE]), PSA, and determination of Gleason score on the basis of either needle biopsy of the prostate or transurethral resection. All pathology specimens were reviewed by M. D. Anderson pathologists. For this study, each patient's disease stage was recalculated on the basis of the 1997 American Joint Committee on Cancer staging system. Further workup with bone scan, computed tomography of the abdomen and pelvis, magnetic resonance imaging of the abdomen and pelvis, chest radiography, and various blood studies were performed at the discretion of the treating physician. Because clinicopathologic features were similar between Hispanic and non-Hispanic whites, these groups were combined in the analyses.

During the study period, EBRT was administered using either a conventional 4-field approach in 85% of the patients or a conformal approach with 3-dimensional treatment planning (3D-CRT) in 15% of patients. The median dose was 68 grays (Gy) for the conventionally treated patients and 78 Gy for those treated by 3D-CRT. Details regarding both the conventional and 3D-CRT techniques have been published previously.15–17

Clinical Follow-Up

Patients were followed up with DRE and PSA measurements at 3 months after the completion of EBRT and then every 3 to 6 months for the first 2 years, and every 6 to 12 months thereafter. Follow-up information was obtained from patients' hospital records or by contacting outside physicians or hospitals. Weight and height were measured and recorded by a nurse at the first clinical visit.

The first endpoint of this study was BF using the American Society for Therapeutic Radiology and Oncology (ASTRO) definition18: 3 consecutive increases in posttreatment PSA concentration after achievement of a nadir. The date of BF was defined as the point midway between the PSA nadir and the first PSA increase, and the time to biochemical recurrence as the interval between the date of completion of EBRT and the ASTRO failure date.

The second endpoint was clinical recurrence, including local recurrence and distant metastasis determined by physical examination, biopsy, or radiologic studies. The time to clinical recurrence was calculated from the date of completion of radiation to the date of confirmed failure.

Statistical Analysis

Body mass index (BMI) (weight in kilograms divided by height in meters squared [kg/m2]) was analyzed as both a continuous and categorical variable using the National Heart, Lung, and Blood Institute guidelines. Four BMI categories were used: normal weight (18.5–24.9 kg/m2), overweight (25–29.9 kg/m2), mildly obese (30.0–34.9 kg/m2), and moderately and severely obese (≥35.0 kg/m2). Because there were only a few patients who were underweight (n = 3), they were included in the normal weight group. Clinical stage (T1c, T2, and T3) and Gleason score (≤6, 7, and ≥8) were analyzed as categorical variables. Radiation dose was analyzed both as a continuous and a categorical variable (<70 Gy and ≥70 Gy) and year of diagnosis as a continuous variable. Due to the skewed distribution, pretreatment PSA was log-transformed and analyzed continuously. Clinical characteristics were compared across the BMI groups and by progression status using analysis of variance (ANOVA) for continuous variables or chi-square test for categorical variables. The Kaplan–Meier survival plots and log-rank test were used to estimate progression-free survival rates. Univariate Cox proportional hazards analyses evaluating each of the potential risk factors individually were conducted to evaluate the crude effect of each variable on risk of BF and clinical failure (CF). To estimate the independent effects of variables of interest, we fit multivariate Cox proportional hazard models incorporating significant clinicopathologic parameters and BMI using a backward-stepwise selection procedure. All analyses were performed using the statistical software package Splus 2000 (MathSoft, Seattle, WA).


Table 1 shows patient characteristics by pretreatment BMI categories. The mean age at diagnosis for the 873 patients was 68 years (range, 41–84 years) and follow- up time averaged 96 months (range, 14–201 months). Normal weight was recorded for 234 men (27%), 434 (50%) were overweight, 157 (18%) were mildly obese, and 48 (5%) were moderately to severely obese. Age at the time of EBRT was found to be inversely correlated with increasing BMI category (P < .001). African-American patients had the highest frequency of obesity (35%, median BMI: 27.9 kg/m2), followed by whites (23%, median BMI: 27.0 kg/m2) and Asians (14%, median BMI: 25.4 kg/m2). There were more patients who never smoked (57%) in the normal weight group. A higher frequency of diabetes was predictably found with increasing BMI category (P = .001). Although a statistically significant difference was found among the distributions of clinical stage by BMI categories (P = .03), there did not appear to be a trend across categories. There was no association noted between Gleason score or PSA level and obesity. The total prescribed dose to the prostate and seminal vesicle was similar for all the groups. There were no differences in the treatment technique and nadir PSA levels (data not shown) across BMI groups. Using the Spearman correlation coefficient (r), we found that higher BMI was correlated with more recent year of EBRT (r = 0.073; P = .03). The mean follow-up time did not vary significantly by BMI category (97 months, 95 months, 93 months, and 93 months from normal weight to severely obese, respectively, P = .81). The mean follow-up among patients who did not experience disease progression was 83 months (range, 14–201 months).

Table 1. Characteristics by Obesity
Normal (<25)Overweight (25–29)Mildly obese (30–34)Moderately to severely obese (≥35)
  1. BMI indicates body mass index; SD, standard deviation; PSA, prostate-specific antigen; Gy, grays.

Total234 (27%)434 (50%)157 (18%)48 (5%) 
Age at diagnosis, y
 Mean ± SD69.3 ± 6.6968.0 ± 6.2367.1 ± 6.4765.0 ± 6.50<.001
 White209 (89%)401 (92%)139 (89%)39 (81%).03
 African-American18 (8%)28 (7%)16 (10%)9 (19%) 
 Asian7 (3%)5 (1%)2 (1%) 
Clinical stage
 T1c74 (32%)147 (34%)73 (47%)20 (42%).03
 T2109 (46%)174 (40%)52 (33%)18 (37%) 
 T351 (22%)113 (26%)32 (20%)10 (21%) 
Gleason score
 2–6136 (58%)266 (61%)96 (61%)24 (50%).59
 777 (33%)123 (28%)46 (29%)20 (42%) 
 8–1021 (9%)45 (11%)15 (10%)4 (8%) 
PSA (ng/mL)
 Mean ± SD11.6 ± 13.010.5 ± 9.69.4 ± 6.512.1 ± 10.9.19
Dose (Gy)
 Mean ± SD70.6 ± 4.970.7 ± 4.970.9 ± 4.770.5 ± 4.6.90
 <7087 (37%)170 (39%)54 (34%)16 (33%).68
 ≥70147 (63%)264 (61%)103 (66%)32 (67%) 
 No186 (96%)342 (92%)117 (86%)39 (83%).001
 Yes7 (4%)28 (8%)19 (14%)8 (17%) 
 Never98 (57%)168 (51%)53 (43%)20 (51%).14
 Ever73 (43%)162 (49%)69 (57%)19 (49%) 

Table 2 compares patient characteristics by progression status (BF and CF). Of the 873 patients in this study, 295 (34%) had BF and 127 (15%) progressed to CF. Patients who progressed (BF and CF) were diagnosed at a younger age and were more likely to have presented with more advanced disease (higher clinical stage, Gleason score, pretreatment PSA >10 ng/mL) than those who did not progress. Those who progressed also received a significantly lower radiation dose and were diagnosed in earlier years than those men whose disease did not progress. No differences were found by ethnicity, history of diabetes, smoking, or BMI for either BF or CF.

Table 2. Characteristics by Biochemical and Clinical Failure
VariableBiochemical/Clinical failure
NoneBF onlyBF and CFP
  1. BF indicates biochemical failure; CF, clinical failure; SD, standard deviation; PSA, prostate-specific antigen; Gy, grays; BMI, body mass index.

Total578 (66%)168 (19%)127 (15%) 
Age at diagnosis, y
 Mean ± SD68.1 ± 6.469.5 ± 6.265.7 ± 6.8<.001
Year of diagnosis (median)199519921991<.001
 White522 (91%)147 (87%)119 (94%).35
 African-American48 (8%)16 (10%)7 (5%) 
 Asian8 (1%)5 (3%)1 (1%) 
Clinical stage
 T1c258 (45%)43 (25%)13 (10%)<.001
 T2231 (40%)70 (42%)52 (41%) 
 T389 (15%)55 (33%)62 (49%) 
Gleason score
 2–6367 (64%)84 (50%)71 (56%).001
 7170 (29%)60 (36%)36 (28%) 
 8–1041 (7%)24 (14%)20 (16%) 
PSA (ng/mL)
 Mean ± SD8.5 ± 5.814.5 ± 12.715.3 ± 17.2<.001
Dose (Gy)
 Mean ± SD71.9 ± 4.868.4 ± 4.267.8 ± 3.8<.001
 <70150 (26%)96 (57%)81 (64%)<.001
 ≥70428 (74%)72 (43%)46 (36%) 
BMI (kg/m2)
 Mean ± SD27.5 ± 4.327.7 ± 4.628.4 ± 4.80.23
 <25158 (27%)46 (27%)30 (24%) 
 25–29293 (51%)82 (49%)59 (47%).47
 30–35101 (18%)28 (17%)28 (22%) 
 ≥3526 (4%)12 (7%)10 (8%) 
 No461 (92%)117 (90%)106 (95%).41
 Yes43 (8%)13 (10%)6 (5%) 
 Never230 (51%)54 (46%)55 (60%).13
 Ever222 (49%)64 (54%)37 (40%) 

Biochemical Failure

The overall 5- and 10-year ASTRO-defined biochemical-free survival rates were 66% and 63%, respectively, and the mean time to failure was 28 months. Figure 1 shows the survival curves stratified by BMI categories. There were no significant differences in BF-free survival (P = .30) among the different BMI groups. However, men in the most obese group (BMI ≥35 kg/m2) had the lowest BF-free survival compared with those with normal weight BMI <25 kg/m2 (P = .06). Time to BF tended to decrease by BMI categories (normal weight: 30 months; overweight: 28 months; mildly obese: 28 months; and moderately to severely obese: 26 months).

Figure 1.

Biochemical failure-free survival by body mass index (BMI) groups among 939 patients.

Using Cox proportional hazards models, risk factors were evaluated as potential prognostic indicators by univariate analysis. As expected, we found the traditional clinicopathologic variables pre-EBRT PSA (hazards ratio [HR] = 7.73; P < .001), stage T2 (HR = 2.04; P < .001), T3 (HR = 4.01; P < .001), and Gleason score (Gleason 7, HR = 1.32 [P < .03]; Gleason 8–10, HR = 2.13 [P < .001]) to be strong predictors of BF (Table 3). BMI, as a continuous variable, was also associated with a borderline significant increased risk of BF (HR = 1.02; P = .09). There was an increased risk of BF with ascending BMI category, although the trend was not statistically significant (P = .13). Higher radiation dose (HR = 0.87; P < .001) and later year of diagnosis (HR = 0.81; P < .001) were associated with reduced risk. A multivariate model, after simultaneous adjustment for all variables (Table 3), was used to determine the significant predictors of BF. With all clinical characteristics included, BMI, as a continuous variable, remained an independent risk factor of BF (HR = 1.04; P = .001). In a separate model including BMI as a categorical variable, we found a statistically significant trend (P = .001) by BMI categories. Because obese patients were more likely to be diabetic, we ran the same multivariate analysis including diabetes as a covariate in a subset of 746 patients for whom we had diabetes information. Adding diabetes did not change the strength of association between obesity and BF as BMI remained statistically significant when analyzed continuously (HR = 1.04; P = .01) and categorically (P-trend = 0.02).

Table 3. Cox Regression Analysis of Biochemical Failure
HR (95% CI)PHR (95% CI)P
  • HR indicates hazards ratio; 95% CI, 95% confidence interval; Log10, PSA, prostate-specific antigen; BMI, body mass index.

  • *

    An alternative multivariate model using BMI as a categoril variable as opposed to a continuous variable.

Age (continuous)0.99 (0.98–1.01).53 
 African-American1.01 (0.66–1.54).98 
 Asian vs. white1.36 (0.61–3.06).46 
Clinical stage
 T1c1.00 1.00 
 T22.04 (1.49–2.80)<.0011.51 (1.08–2.10).02
 T34.01 (2.91–5.51)<.0011.95 (1.35–2.81)<.001
Gleason score
 2–61.00 1.00 
 71.32 (1.02–1.70).031.83 (1.40–2.40)<.001
 8–102.13 (1.53–2.98)<.0011.81 (1.28–2.57).001
Log10 PSA (continuous)7.73 (5.19–11.5)<.0014.00 (2.74–5.83)<.001
Dose (continuous)0.87 (0.85–0.90)<.0010.93 (0.89–0.97).001
BMI (continuous)1.02 (1.00–1.05).091.04 (1.02–1.07).001
 <251.00 1.00.16
 25–291.05 (0.79–1.38).761.23 (0.92–1.63) 
 30–351.11 (0.79–1.58).541.55 (1.09–2.21).02
 ≥351.56 (0.97–2.51).071.99 (1.23–3.20).005
 Trend testP = 0.13 P = 0.001 
Diagnosis year (continuous)0.81 (0.78–0.84)<.0010.89 (0.83–0.95).001

Clinical Failure

The second endpoint considered was CF-free survival. At time of last follow-up, of the 295 patients who experienced BF, 127 (43%) had documented CF (local in 91 patients, distant metastasis in 20 patients, or both in 16 patients). Only 13% of patients with CF received salvage treatment before failure. The use of salvage therapy was similar between BMI categories (P = .91).

The mean time from EBRT to CF was 54 months and the time from BF to CF was 29 months. The overall CF-free survival was 89% at 5 years and 82% at 10 years. Figure 2A shows the CF-free survival curves among the 873 patients by BMI category. There were no statistically significant differences in CF-survival noted between the 4 different BMI groups (P = .19). However, when comparing obese (BMI ≥30 kg/m2) with nonobese patients, obese men had a significantly higher rate of CF (P = .04) (Fig. 2B).

Figure 2.

(A) Clinical failure-free survival by body mass index (BMI) groups among 939 patients. (B) Clinical failure-free survival by obesity among 939 patients.

Univariate and multivariate analysis results are shown in Table 4. Multivariate analysis results showed that, similar to the BF results, clinicopathologic features were strong predictors of CF (Stage III, HR = 4.70 [P < .001]), Gleason score 8–10 (HR = 1.97; P = .01), and pre-EBRT PSA (HR = 2.81; P < .001). Increased radiation dose (HR = 0.90; P = .01) and use of salvage treatment after BF (HR = 0.48; P = .01) were associated with reduced risk. BMI, as a continuous variable, remained an independent predictor of CF (HR = 1.05; P = .01). In an alternative multivariate model with BMI as a categorical variable, we found a statistically significant trend (P = .03) with increasing BMI category. The addition of diabetes to the multivariate analysis did not change the results: BMI, both as a continuous (P = .02) or categorical variable (P = .04), remained an independent predictor of CF.

Table 4. Cox Regression Analysis of Clinical Failure
HR (95% CI)PHR (95% CI)P
  • HR indicates hazards ratio; 95% CI, 95% confidence interval; Log10, PSA, prostate-specific antigen; BMI, body mass index.

  • *

    An alternative multivariate model using BMI as a categoric variable as opposed to a continuous variable.

Age (continuous)0.95 (0.92–0.97)<.0010.95 (0.92–0.97)<.001
 African-American0.65 (0.30–1.40).27 
 Asian0.51 (0.07–3.63).50 
Clinical stage
 T1c1.00 1.00 
 T23.17 (1.73–5.82)<.0012.93 (1.57–5.48).001
 T36.36 (3.49–11.6)<.0014.70 (2.43–9.09)<.001
Gleason score
 2–61.00 1.00 
 71.14 (0.76–1.70).531.40 (0.92–2.12).12
 8–101.88 (1.14–3.09).011.97 (1.16–3.34).01
Log10 PSA (continuous)3.96 (2.27–6.89)<.0012.81 (1.63–4.84)<.001
Dose (continuous)0.90 (0.86–0.94)<.0010.90 (0.84–0.97).01
BMI (continuous)1.04 (1.01–1.08).021.05 (1.01–1.09).01
 <251.00 1.00 
 25–291.09 (0.71–1.70).690.97 (0.62–1.52).89
 30–351.53 (0.91–2.56).111.65 (0.79–3.43).18
 ≥351.78 (0.87–3.64).111.66 (0.98–2.83).06
 Trend testP = .04 P = .03 
Year of diagnosis (continuous)0.84 (0.79–0.89)<.0010.98 (0.88–1.09).98
Salvage therapy (yes vs. no)0.84 (0.48–1.47).840.48 (0.27–0.84).01

Finally, to determine whether obesity increases risk for distant failure separately from overall CF, we analyzed time to distant failure among our cases (36 cases with distant failure and 837 without distant failure). In a multivariate Cox regression analysis, BMI analyzed continuously was associated with an increased risk of distant failure, but did not reach statistical significance (HR = 1.04; 95% confidence interval [95% CI], 0.97–1.12 [P = .22]). However, when we categorized BMI by obesity (BMI <30 vs. ≥30 kg/m2), we found that obese men had a 2-fold risk of developing metastases compared with nonobese men after adjustment for the same factors included in Table 4 (HR = 2.01; 95% CI, 1.00–4.05 [P = .05]). Our data therefore suggest that obesity is associated not only with local failure but also with metastatic disease.


In the last year, several reports have suggested that obesity, measured by increased BMI, is associated with increased BF among patients treated with radical prostatectomy.9–11

To our knowledge, no previous study has investigated the correlation between obesity and prostate cancer progression among patients treated with EBRT as definitive therapy for localized prostate cancer. Using a database of a large series of patients irradiated at 1 institution, we found that for both outcomes (BF and CF), there was a significant trend of higher risk of recurrence with increasing BMI category after adjusting for all clinical and tumor characteristics and treatment factors. Moderately and severely obese men had a 99% greater risk of developing BF (P = .005). Furthermore, in a similar analysis, BMI remained an independent predictor of CF (P = .01), with a significant trend by BMI category (P = .03). These findings support the hypothesis that the degree of obesity is an important independent prognostic factor for patients with clinically localized prostate cancer.

The association between obesity and prostate cancer progression/mortality is supported by several lines of independent evidence. Large studies investigating obesity and prostate cancer mortality in large populations have demonstrated a consistent association between BMI and death from prostate cancer.3, 19 Recently, several studies have reported an association between obesity and BF.6, 10, 11 In addition to being obese at diagnosis, we also found that obesity at age 40 years and weight gain over time were strong predictors of BF.9 In a study by Mallah et al.,20 increased BMI was found to be only weakly associated with disease progression. However, the definition of disease progression used, as well as the statistical approach, makes their results hard to compare with other reports. In the only study that included radiotherapy patients, no impact of BMI on BF was reported among patients treated with permanent brachytherapy.12 The results are difficult to interpret, however, because 81% of the patients received supplemental therapies that could have masked differences in outcome.

Radiation patients comprise a heterogeneous group that includes patients who choose this therapy as well as patients who, because of comorbidities including obesity-related conditions, are not candidates for prostatectomy. In our study, the frequency of obesity, the correlations between BMI and year of treatment, and obesity with younger age at diagnosis were similar to those reported in previous surgical and radiotherapy series.6, 11, 12 As expected, our population was older and had a larger percentage of men diagnosed with a higher Gleason score and PSA compared with past reports. However, adverse clinical features did not differ significantly by BMI category, similar to data from the brachytherapy series.12 A history of diabetes was not associated with outcome, and its inclusion in the multivariate analysis had little impact on the results.

In addition, to corroborate the role of obesity in a cohort of patients treated with EBRT, we studied CF as a subsequent endpoint after BF. Our finding that increased BMI is an independent predictor of local and distant metastasis further supports our hypothesis that obesity may affect prostate cancer survival in men diagnosed with localized disease. Technical difficulties in performing prostatectomy and even EBRT in obese men could account, at least in part, for this difference in outcome. However, the fact that the same association was found among patients with different risk profiles, treated with different therapies, suggests that poorer outcomes in obese men are not related to differences in treatment as much as to differences in tumor behavior between obese and nonobese patients.

We examined the potential for differences in radiation technique in obese and nonobese men. Of note is that during the era in which patients in this study were treated, a more simple, conventional radiation technique was used in the majority of cases, including those in the most obese category. More recently, highly conformal techniques with smaller margins necessitate daily prostate localization techniques such as ultrasound or fiducial markers with portal imaging so that the target is not missed. These targeting techniques are especially important in obese patients, as it has been suggested that EBRT could be more difficult to deliver accurately in these patients because of a greater chance of daily setup error and excessive intraabdominal adipose tissue, which may increase internal organ motion.21, 22 Millender et al.,23 using gold markers and portal imaging, showed that positioning error magnitude had a mean of 11.4 mm in the right-left direction, 7.2 mm in the superior-inferior direction, and 2.6 mm in the anterior-posterior direction in morbidly obese men treated with external radiotherapy. Of note, however, is that patients in this study population were treated in the era when margins were generally large enough to account for the setup error described, which is especially relevant for obese patients, who might be more prone to setup errors. Because studies to date show that the most significant therapeutic factor affecting BF after EBRT is radiation dose rather than technique,24 we also included dose in the multivariate analysis, although as shown in Table 1, there was no significant difference in dose across BMI categories. Therefore, it appears unlikely that radiation technique alone can explain our results, but surely analysis of the effect of BMI on outcome while controlling for treatment technique, setup inaccuracy, and internal organ motion is warranted. Radiographic tools are now available by which this type of study can be done.

Sex steroids, leptin, insulin, IGF-1, and IGF binding proteins, as well as diet and physical activity, have been hypothesized to influence prostate carcinogenesis and may explain the observed worse outcomes among obese men. Obesity has been associated with lower testosterone levels25 and more advanced prostate cancer.26, 27 In addition, obesity results in higher insulin and IGF-1 levels,28, 29 both of which are mitogenic and antiapoptotic, and have been associated with advanced prostate cancer.30, 31 The interrelations between these hormones and obesity are extremely complex, and well-designed studies will be necessary to clarify the underlying mechanisms between obesity and prostate cancer progression.

Our study included a clinically homogeneous series of patients from a single institution treated with EBRT only. There are the usual inherent limitations, of course. This study is hospital-based and the patient population of M. D. Anderson is subject to the vagaries of referral patterns. BMI is the most widely used anthropometric measurement; however, its utility is limited as it does not truly distinguish between adiposity and lean body mass. Because relatively long follow-up is necessary to assess BF, and especially CF, this study, similar to all similar analyses of the era, must control for patient characteristics and treatment factors dissimilar to those seen more currently.

In summary, the findings of the current study validate the importance of obesity in prostate cancer progression and suggest a link to the biology of this tumor. Future studies should evaluate the relation of obesity with dietary factors, genetic modifiers of steroid androgen metabolism, insulin, and a detailed investigation of the insulin growth factor pathway to explore the underlying mechanisms of action in prostate carcinogenesis. Understanding the mechanisms by which weight gain contributes to prostate cancer progression will lead to rationally designed preventive strategies.


We thank Alicia Arciniega and Martha Reyes for data collection, and our patients whose cooperation made this study possible.