Effect of obesity on prostate-specific antigen recurrence after radiation therapy for localized prostate cancer as measured by the 2006 Radiation Therapy Oncology Group-American Society for Therapeutic Radiation and Oncology (RTOG-ASTRO) Phoenix consensus definition


  • The views expressed in this article are those of the authors and do not reflect the official policy or position of the Department of the Navy, Department of Defense, or the United States Government.

  • This article is a US Government work and, as such, is in the public domain in the United States of America.



Given the limited data regarding the impact of obesity on treatment outcomes after external beam radiation therapy (EBRT) for the definitive treatment of prostate cancer, the authors sought to evaluate the effect of obesity as measured by body mass index (BMI) on biochemical disease recurrence (BCR) using the most current 2006 Radiation Therapy Oncology Group-American Society for Therapeutic Radiation and Oncology (RTOG-ASTRO) Phoenix consensus definition (prostate-specific antigen [PSA] nadir + 2 ng/mL).


A retrospective cohort study identified men who underwent primary EBRT for localized prostate cancer between 1989 and 2003 using the Center for Prostate Disease Research (CPDR) Multi-center National Database. BMI was calculated (in kg/m2) and the data were analyzed. Univariate and multivariate Cox proportional hazards regression analyses were used to determine whether BMI significantly predicted BCR.


Of the 1868 eligible patients, 399 (21%) were obese. The median age of the patients and pretreatment PSA level were 70.2 years and 8.2 ng/mL, respectively. Of 1320 patients for whom data were available with which to calculate PSA recurrence (PSA nadir + 2 ng/mL), a total of 554 men (42.0%) experienced BCR. On univariate analysis, BMI was found to be an independent predictor of PSA recurrence (P = .02), as was race, pretreatment PSA level, EBRT dose, clinical T classification, Gleason score, PSA nadir, and the use of androgen-deprivation therapy (ADT). On multivariate analysis, BMI remained a significant predictor of BCR (P = .008).


To the authors' knowledge, this is the first study to report the association between obesity and BCR after EBRT for localized prostate cancer as measured by the updated 2006 RTOG-ASTRO definition. A higher BMI is associated with greater odds of BCR after undergoing definitive EBRT. Cancer 2007. Published 2007 by the American Cancer Society.

Obesity is an epidemic in the U.S., affecting >30% of adults, which reflects a prevalence that has doubled in the past 20 years.1, 2 Obesity is associated with multiple chronic disorders, including increased overall mortality from cardiovascular disease, diabetes, and several types of cancer. Previous studies have demonstrated associations between obesity and an increased risk of endometrial, breast, kidney, esophageal, and colon cancer. Data regarding the correlation between obesity and the diagnosis of prostate cancer are less clear, because several investigators have reported conflicting results.3–5 Despite inconsistent findings correlating the incidence of prostate cancer with obesity, increased prostate cancer mortality in obese men appears to be more convincing.6 The Cancer Prevention Study II suggests that men with a body mass index (BMI) ≥30 kg/m2 have a 20% to 34% increased risk of prostate cancer death compared with their normal weight counterparts (BMI < 25 kg/m2).3 A prospective cohort study of >400,000 men followed for 13 years similarly suggests that obese men are at an increased risk for death from prostate cancer.2

Several surgical series have demonstrated the impact of obesity on prostate cancer outcomes. Recently, several multicenter studies have presented data demonstrating that men with prostate cancer and an increased BMI who are surgically treated with radical retropubic prostatectomy were at an increased risk for worse pathologic features, a higher pathologic grade, and higher rates of biochemical recurrence (BCR) compared with normal weight patients.7–9

A recent study by Strom et al. found that BMI was a predictor of biochemical failure among patients treated with radiation therapy using the original American Society for Therapeutic Radiology and Oncology (ASTRO) definition of 3 consecutive increases in prostate-specific antigen (PSA).10, 11 In 2006, the Radiation Therapy Oncology Group (RTOG) -ASTRO Phoenix consensus conference recommended a new definition of BCR after external beam radiation therapy (EBRT), defined as a PSA nadir + 2 ng/mL.12 This definition was considered more generalizable to all risk groups, and more appropriate for patients treated with hormonal therapy. Therefore, we sought to evaluate the effect of obesity on BCR using this new “Phoenix” definition for BCR after EBRT utilizing a large, multiinstitutional database.


Subjects who consented to participate in the Center for Prostate Disease Research (CPDR) Multi-center National Database were examined. This database is governed under an Institutional Review Board (IRB) -approved protocol at the Uniformed Services University of the Health Sciences. Men aged ≥18 years who have undergone a transrectal ultrasound-guided biopsy or transurethral resection of the prostate are eligible for consent to this database. Nine clinical sites contributed to the CPDR database for this study, including: Walter Reed Army Medical Center; National Naval Medical Center; Naval Medical Center, San Diego; Brooke Army Medical Center; Wilford Hall Medical Center; Madigan Army Medical Center; Eisenhower Army Medical Center; Malcom Grow Medical Center; and Naval Medical Center, Portsmouth. This database has been described previously.8

The CPDR database identified 1874 prostate cancer patients for whom height and weight data were recorded who had undergone EBRT for localized prostate cancer between January 1, 1989 and December 31, 2003. Inclusion criteria consisted of patients with known height and weight information, clinical T1-2 prostate cancer, a nadir PSA <4 ng/mL, and the availability of post-nadir PSA values to determine BCR. Patients with an EBRT total dose of >8000 grays (Gy) were excluded (n = 6), as were patients undergoing salvage treatment (prostatectomy, brachytherapy, or cryotherapy).

Clinical, pathologic, and follow-up data were retrieved for each patient, including age, race, height, weight, serum PSA level, biopsy cancer grade (Gleason score), and cancer stage as determined by digital rectal examination according to the 2002 TNM staging system. Additional characteristics recorded included family history and the presence of comorbid disease (diabetes, cardiac disease, hypertension, or hypercholesteremia). Complications from RT were evaluated by recording the presence or absence of symptoms. Urinary symptoms included the presence of urgency, frequency, dysuria, and nocturia, and gastrointestinal symptoms recorded were diarrhea, rectal bleeding, rectal pain, or incontinence of stool. These symptoms were dichotomized by patient BMI group and evaluated by the presence of either 1, 2, or more symptoms. The previous or concurrent usage of androgen-deprivation therapy (ADT) was also evaluated. After EBRT, patients were generally followed with PSA measurements and examinations for evidence of disease recurrence every 3 months for the first year, semiannually the second year, and annually thereafter or until death.

BMI was analyzed as both a continuous and categoric variable using the National Heart, Lung, and Blood Institute guidelines, and was calculated according to the following formula: BMI = weight in kilograms divided by height in meters squared. BMI was examined as a categoric variable using the National Institutes of Health definitions of “normal weight” (<25 kg/m2), “overweight” (25–30 kg/m2), and “obese” (>30 kg/m2). Normal and overweight patients (BMI ≤30 kg/m2) were combined and compared with obese patients (BMI >30 kg/m2). PSA data were log-transformed due to nonnormality in the PSA distribution. BCR, or PSA recurrence, was defined by the 2006 RTOG-ASTRO Phoenix consensus conference definition and is the most recently recommended criteria for PSA recurrence in the EBRT patient population. It is defined as a PSA increase ≥2 ng/mL above the nadir PSA (lowest PSA achieved) after EBRT with or without short-term hormonal therapy. The EBRT completion date was defined as the EBRT start date plus 51.7 days (average duration of EBRT) and was assigned to patients whose completion date was missing (n = 18 patients).

Univariate and multivariate Cox proportional hazards regression analyses were used to determine whether BMI significantly predicted BCR as defined by ASTRO criteria. The multivariate model included all covariates of interest that were found to be independently associated with the study outcome (time to PSA recurrence according to the RTOG-ASTRO Phoenix definition) at the significance level of P ≤ .15.


Table 1 demonstrates the clinical and pathologic characteristics of the study population segregated by pre-EBRT BMI grouping. Of 1868 patients, 399 (21%) were obese, and 1469 (79%) were not obese. Those with moderate to severe obesity (BMI ≥35 kg/m2) represented 5.9% of the cohort. African-American patients represented 24.4% of the group and as the BMI increased, the percentage of African-American patients comprising each BMI group also increased (P < .001). The median age of the patients at diagnosis was 70.7 years (range, 40–93 years), and an inverse correlation was noted between decreasing age and increase in BMI. The median pretreatment PSA was 8.2 ng/mL (range, 0.5–400 ng/mL) for the cohort, and there was a trend toward decreasing PSA in those with a higher BMI. The median EBRT dose was 68.4 Gy (range, 60–80 Gy), and those patients with a higher BMI had a statistically, but not clinically, significantly increased EBRT dose. The median follow-up time was 42.6 months (range, 24–174 months). The biopsy Gleason score was between 2 and 6 for 817 of the patients (56.3%), whereas 419 patients (28.9%) had a Gleason score of 7, and 215 patients (14.8%) had a Gleason score of 8 to 10. All patients had clinically localized prostate cancer, with 377 patients (20.2%) having T2c or higher disease. The use of neoadjuvant or adjuvant hormonal therapy was found in 27% of the cohort; there was a trend toward its use in those patients with a higher BMI, but this trend did not meet statistical significance (P = .07). The median PSA nadir was 0.4 ng/mL. Clinical stage and Gleason score were not found to be associated with increasing BMI.

Table 1. Study Cohort Characteristics by Obesity (N = 1868)
VariableBMI (kg/m2)P*
Normal (<25)Overweight (25–29)Mildly obese (30–34)Moderately/Severely obese (≥35)
  • BMI indicates body mass index; SD, standard deviation; PSA, prostate-specific antigen; Gy, grays; ADT, androgen-deprivation therapy.

  • *

    P was derived from the Student t test.

  • PSA data were logarithmically transformed to produce a normal data distribution but are presented as actual values.

Total659 (35.3%)810 (43.4%)288 (15.4%)111 (5.9%) 
Age at diagnosis, y    <.0001
 Mean ± SD71.0 ± 6.468.8 ± 6.767.6 ± 6.665.7 ± 6.1 
 Median (range)72.1 (48.0–87.6)69.5 (40.2–84.9)68.2 (49.8–93.5)65.6 (47.5–78.5) 
PSA (ng/mL)    .2599
 Mean ± SD16.0 ± 29.413.8 ± 17.214.9 ± 28.210.8 ± 11.8 
 Median (range)8.8 (0.5–400.1)8.0 (0.5–180.0)7.9 (0.5–370.3)8.1 (1.3–85.4) 
Follow-up time, mo    .0416
 Mean ± SD51.5 ± 36.450.8 ± 37.444.8 ± 33.646.1 ± 32.6 
 Median (range)44.9 (0–162.6)43.5 (0.6–174.3)36.7 (0–167.1)42.8 (0.8–133.8) 
Dose, Gy/100    <.0001
 Mean ± SD69.1 ± 24.169.7 ± 29.469.8 ± 26.370.1 ± 28.0 
 <70431 (65.4%)416 (51.4%)130 (45.1%)48 (43.2%)<.0001
 >=70228 (34.6%)394 (48.6%)158 (54.9%)63 (56.8%) 
Ethnicity    <.0001
 White and other508 (78.9%)610 (77.2%)201 (70.5%)64 (58.2%) 
 African American136 (21.1%)180 (22.8%)084 (29.5%)46 (41.8%) 
Clinical stage    .1405
 T1c228 (35.0%)307 (38.2%)120 (41.7%)54 (50.0%) 
 T2a179 (27.4%)195 (24.2%)067 (23.3%)23 (21.3%) 
 T2b101 (15.5%)141 (17.5%)045 (15.6%)15 (13.9%) 
 ≥T2c144 (22.1%)161 (20.0%)056 (19.4%)16 (14.8%) 
Gleason score    .6237
 2–6253 (57.2%)380 (57.1%)133 (54.7%)51 (51.0%) 
 7130 (29.4%)187 (28.1%)074 (30.5%)28 (28.0%) 
 8–10059 (13.4%)099 (14.9%)036 (14.8%)21 (21.0%) 
Neoadjuvant ADT    .0757
 No500 (75.9%)582 (71.9%)196 (68.1%)79 (71.2%) 
 Yes159 (24.1%)228 (28.1%)092 (31.9%)32 (28.8%) 

Clinical and pathologic factors were then compared to determine their correlation with PSA recurrence after EBRT. Of the 1320 patients for whom data were available with which to calculate PSA recurrence (PSA nadir + 2 ng/mL), BCR occurred in 554 patients (42%) (Table 2). Baseline PSA, total EBRT dose, clinical stage, and Gleason score were found to be statistically significant predictors of PSA recurrence. Table 3 delineates the association between race and increased BMI in the setting of BCR. These data failed to demonstrate that the African-American cohort, with a significantly higher BMI in general, is independently correlatassociated with an increased risk of PSA recurrence.

Table 2. Study Cohort Characteristics by PSA Recurrence (n = 1320)
VariablePSA recurrenceP*
  • PSA indicates prostate-specific antigen; SD, standard deviation; BMI, body mass index; Gy, grays; ADT, androgen-deprivation therapy; EBRT, external beam radiation therapy.

  • *

    P was derived from the Student t test.

  • PSA data were logarithmically transformed to produce a normal data distribution but are presented as actual values.

Total766 (58.0%)554 (42.0%) 
Age at diagnosis, y  .9723
 Mean ± SD69.4 ± 6.669.4 ± 6.5 
 Median (range)70.2 (40.2–87.6)70.2 (47.3–86.8) 
PSA, ng/mL  <.0001
 Mean ± SD11.7 ± 23.019.9 ± 26.2 
 Median (range)7.2 (0.5–400.1)11.8 (1.1–300.0) 
BMI, kg/m2  .2324
 Mean ± SD26.9 ± 4.727.2 ± 4.8 
 <25252 (59.3%)173 (40.7%).8998
 25–29353 (57.9%)257 (42.1%) 
 30–34122 (56.7%)93 (43.3%) 
 ≥3539 (55.7%)31 (44.3%) 
Dose, Gy/100
 Mean ± SD69.5 ± 25.468.8 ± 24.3<.0001
 <70392 (51.1%)375 (48.9%)<.0001
 ≥70374 (67.6%)179 (32.4%) 
Ethnicity  .1100
 White and other589 (59.2%)406 (40.8%) 
 African American162 (54.0%)138 (46.0%) 
Clinical stage  <.0001
 T1c310 (65.8%)161 (34.2%) 
 T2a224 (63.8%)127 (36.2%) 
 T2b116 (53.2%)102 (46.8%) 
 ≥T2c113 (41.4%)160 (58.6%) 
Gleason score  <.0001
 2–6371 (45.4%)196 (34.6%) 
 7178 (59.9%)119 (40.1%) 
 8–1072 (41.1%)103 (58.9%) 
Neoadjuvant ADT  .5211
 No521 (57.4%)386 (42.6%) 
 Yes245 (59.3%)168 (40.7%) 
PSA nadir after EBRT, ng/mL  <.0001
 <0.5531 (66.8%)264 (33.2%) 
 0.5–0.9159 (57.2%)119 (42.8%) 
 ≥176 (30.8%)171 (69.2%) 
Year of diagnosis (median)19961994<.0001
Table 3. PSA Recurrence Stratified by Race
VariablePSA recurrenceP
  1. PSA indicates prostate-specific antigen; BMI, body mass index.

Total766 (58.0%)554 (42.0%) 
White and other589 (59.2%)406 (40.8%) 
BMI, kg/m2  .5533
 <25207 (61.4%)130 (38.6%)
 25–29275 (59.3%)189 (40.7%)
 30–3483 (54.6%)69 (45.4%)
 ≥3524 (57.1%)18 (42.9%)
African American162 (54.0%)138 (46.0%) 
BMI, kg/m2  .4751
 <2540 (51.3%)38 (48.7%)
 25–2969 (51.9%)64 (48.1%)
 30–3439 (62.9%)23 (37.1%)
 ≥3514 (51.8%)13 (48.2%)

Univariate Cox proportional hazards models of PSA recurrence were used to evaluate clinical and pathologic factors (Table 4). On univariate analysis, PSA, BMI, and EBRT dose, as continuous variables, as well as clinical T classification, Gleason score, the use of hormonal therapy, and degree of PSA nadir were found to be significant predictors of BCR.

These parameters were factored into the multivariate analysis model that demonstrated that BMI, examined as both a continuous and categoric variable, was a significant predictor of PSA recurrence. In addition, PSA, clinical stage, Gleason score, and PSA nadir also were found to be statistically significant.

Table 4. Cox Proportional Hazards Regression Analysis of Time to PSA Recurrence
VariableUnivariate analysis (n = 1300)Multivariate analysis (n = 1210)
  • PSA indicates prostate-specific antigen; HR, hazards ratio; 95% CL, 95% confidence limits; BMI, body mass index; Gy, grays; ADT, androgen-deprivation therapy; EBRT, external beam radiation therapy.

  • *

    PSA data were logarithmically transformed to produce a normal data distribution but are presented as actual values.

Age at diagnosis, y0.9960.9831.009.5589
Natural log of PSA, ng/mL*1.5341.4021.678<.00011.3611.2231.514<.0001
BMI, kg/m2 (continuous)1.0201.0031.037.02341.0251.0061.045.0085
BMI, kg/m2   .1204   .0323
 <251.000   1.000   
Radiation dose, Gy (continuous)1.0011.0011.002<.00011.0011.0001.001.0556
Ethnicity   .0384   .8387
 White and other1.000   1.000   
 African American1.2271.0111.488 0.9780.7921.208 
Clinical stage   <.0001   .0058
 T1c1.000   1.000   
Gleason score   <.0001   <.0001
 2–61.000   1.000   
Neoadjuvant ADT   <.0001    
 No1.000   1.000   
 Yes1.9651.6302.368 1.8291.4212.355<.0001
PSA nadir after EBRT, ng/mL   <.0001   <.0001
 <0.51.000   1.000   
Year of diagnosis (continuous)1.0701.0381.102<.00011.0450.9971.095.0677

Toxicity arising from EBRT was also evaluated based on physician reporting of patient symptoms. When dichotomized by BMI (<30 kg/m2 vs ≥30 kg/m2), there appeared to be no difference with regard to rectal symptoms, but urinary symptoms were significantly more common in the obese patient cohort (P = .01) (Table 5). Approximately 4% of patients had some interruption of EBRT in the cohort evaluated.

Table 5. Reported Symptoms After EBRT by BMI Group
  1. EBRT indicates external beam radiation therapy; BMI, body mass index.

Gastrointestinal symptoms    .3994
Urinary symptoms    .0109
Interruptions to EBRT    .6707

BCR-free survival was estimated using the Kaplan-Meier method and compared among several patient groups. Obese patients (BMI ≥ 30 kg/m2) were found to have significantly higher rates of PSA recurrence over time than nonobese patients (BMI <30 kg/m2) (P = .03) (Fig. 1).

Figure 1.

Prostate-specific antigen recurrence-free survival by BMI group (n = 1320). 95% CL indicates 95% confidence limits.


In addition to the established surgical literature, the results of the current study support the association of obesity and worse outcomes in men with prostate cancer who are treated with EBRT. This association was found to be independently significant when utilizing a more generalizable and robust definition of PSA failure after RT despite other known prognostic factors being considered.12, 13 To our knowledge, the explanations for this association remain unknown, but may be technical, biologic, or a combination of both.

Treatment volumes for EBRT typically include a margin around the prostate to account for daily setup variability, prostate organ motion, and subclinical disease extension beyond the prostate. EBRT is technically more difficult in obese patients, with a greater chance of daily setup errors. Obese men have increased daily prostate movement, and daily localization with implanted fiducial markers has been shown to improve the precision of EBRT in the treatment of morbidly obese men with prostate cancer.14, 15 This type of image guidance was not utilized in the current study population. The increased use of image-guided RT, particularly in obese men, would mitigate the effect of daily variability. Obese patients in the current study also appeared to have increased reports of urinary symptoms after RT, with similar rates of rectal symptoms.

Obesity has been shown to have minimal impact on the outcomes of brachytherapy. Investigators demonstrated BMI to have no statistically significant impact on BCR-free survival over an 8-year follow-up period for men treated with permanent prostate brachytherapy with the implantation of palladium-103 (103-Pd) or iodine-125 (125-I) seeds with and without supplemental EBRT or ADT.16 In addition, obese men (BMI >35 kg/m2) who undergo brachytherapy reportedly demonstrate no significant difference in quality-of-ife parameters, including bowel and sexual function, compared with nonobese patients.17 These findings may be the result of the significant dose escalation achieved with prostate brachytherapy, which may overcome any adverse biologic factors, patient selection bias, or radiation dose distributions within the prostate achieved with brachytherapy.

Body habitus may impact surgical results and BCR rates, because obese men are more likely to have positive surgical margins.18 However, even in men with negative surgical margins, BMI reportedly remained a significant predictor of biochemical failure.9 Furthermore, an association has been suggested between increased BMI and higher grade, more aggressive prostate cancer.

Biologic explanations for the association between obesity and more aggressive prostate cancer may involve metabolic factors. For example, excessive body mass has been shown to influence serum androgen levels and other growth factors.12 Obese men have decreased testosterone and sex hormone-binding globulin with increased estrogenic compounds.19 In addition, obesity is associated with higher insulin and insulin-like growth factor-1 levels, both of which are mitogenic.20 These factors may induce the progression of subclinical tumors to more aggressive prostate cancer.21

Limitations of the current study include the relatively short follow-up period of 42 months for EBRT outcomes and those inherent in retrospective studies. Other anthropomorphic measures of obesity such as abdominal circumference or waist-to-hip ratio may prove to be more predictive of both comorbid disease and prostate cancer outcomes, but were not available for patients in the current study. Finally, inherent limitations to definitions of PSA recurrence after EBRT are present, particularly when patients who were treated with hormonal therapy are included. Despite these limitations, the findings of the current study reaffirm those from previous studies that implicate the detrimental effect of obesity on prostate cancer outcomes. It is also interesting to note that despite the statistical significance, the absolute difference in PSA recurrence was only 3.6%, which is consistent with existing data, and a difference in overall survival has not been demonstrated to date. The existing data should be taken in that context.

In summary, higher BMI (≥30 mg/kg2) appears to be associated with a statistically higher risk of PSA recurrence after definitive EBRT. However, the magnitude of this increased risk appears be clinically small. This may be a result of more aggressive cancer or due to technical difficulties, such as setup variability during RT. However, taken in context with published surgical series, these data suggest that the worse outcomes noted in obese men are at least in part due to more biologically aggressive disease, thereby explaining epidemiologic associations with increased prostate cancer mortality. Improving treatment outcomes in these patients may require improving surgical techniques, more accurate EBRT targeting (such as image-guided RT), recommendations for brachytherapy, or more extensive use of hormonal therapy. In addition, consideration should be given to dietary modification and exercise regimens that may complement traditional therapies in the management of prostate cancer in obese patients, although the question of whether obesity effects are reversible remains to be determined.


To our knowledge, this is the first study to report the association between obesity and BCR after EBRT for the treatment of prostate cancer as measured by the updated RTOG-ASTRO Phoenix definition (PSA nadir + 2 ng/mL). A higher BMI appears to be correlated with a statistically significant but clinically small increased risk of PSA recurrence after definitive EBRT. Decreased PSA-free survival in obese patients may be the result of more aggressive cancer or setup difficulties during RT. Further research should be conducted to improve outcomes in these patients.


Supported by the Center for Prostate Disease Research, a Department of Defense program of the Uniformed Services University of the Health Sciences funded by the U.S. Army Medical Research and Material Command.