Abdominal and pelvic adipose tissue distribution and risk of prostate cancer recurrence after radiation therapy

Fat distribution varies between individuals of similar body mass index (BMI). We hypothesized that visceral obesity is more strongly associated with poor prostate cancer outcomes than overall obesity defined by BMI.


| INTRODUCTION
Prostate cancer is the most frequently diagnosed non-skin cancer among men in the United States. 1 Obesity is a risk factor linked with increased prostate cancer aggressiveness and a poorer prognosis. Specifically, obesity has been associated with an increased risk of biochemical recurrence following radical prostatectomy and external beam radiotherapy, higher frequency of metabolic complications and treatment failure after androgen deprivation therapy (ADT), and increased likelihood of prostate cancer-specific mortality. 2 Though studies in radiationtreated patients are relatively few, a secondary analysis of data from a randomized trial testing adjuvant vs salvage ADT in men treated with external beam radiation therapy found that larger body mass index (BMI) was associated with higher prostate cancer-specific mortality. 3 Understanding the association between obesity and aggressive prostate cancer has important public health implications as lifestyle interventions can influence this potentially modifiable risk factor.
Most studies to date have focused on the relationship between BMI, a measure of overall adiposity, and prostate cancer outcomes.
Few have considered adipose tissue distribution measures, including visceral adipose tissue, a fat depot known to be more metabolically active than subcutaneous areas. 4 Also, most studies have examined associations between obesity and prostate cancer outcomes in white men. Black men have a lower quantity of visceral adipose tissue compared with other races, 5 but a higher prevalence of obesityrelated metabolic disease. [6][7][8] Therefore, it is important to have a diverse study cohort to understand and account for potential racial differences in associations with prostate cancer outcomes.
We previously examined the relationship between adipose tissue distribution and tumor aggressiveness at diagnosis in a retrospective, racially diverse cohort of prostate cancer patients treated with primary radiation at the Durham North Carolina Veterans Administration Hospital. 9 To extend our previous analysis, the objective of this current study was to examine the association between obesity, adipose tissue distribution, and risk of prostate cancer recurrence, overall and stratified by race and by receipt of ADT. We hypothesized that excess visceral adiposity would be more strongly associated with poor prostate cancer outcomes than overall obesity, as defined by high BMI. We also examined periprostatic adipose tissue (PPAT), a type of visceral fat enveloping the prostate. 10 2 | MATERIALS AND METHODS

| Study design
We identified a cohort of 407 men as a consecutive series of patients with biopsy-confirmed prostate cancer that were treated with primary external beam radiotherapy (XRT) or brachytherapy from 2005 to 2011 at the Durham Veterans Affairs (VA) Medical Center. We excluded 1 patient missing BMI data, 4 patients missing pre-biopsy prostate-specific antigen (PSA), and 1 patient missing follow-up data, resulting in a study cohort of 401 men.

| Adipose tissue measurement
Visceral fat area (VFA), subcutaneous fat area (SFA), and PPAT area were measured using radiotherapy planning computed tomography (CT) scans, as previously described. 9 Briefly, VFA and SFA were calculated using a single CT slice at the level of the L4/ L5 vertebrae, a validated method used by other groups. 11-13 PPAT area was calculated using a single CT slice at the first anterior point of the pubic symphysis, as we and others have done previously. 9

| Obesity definitions
Height and weight were measured at the closest medical visit before radiation therapy and were abstracted from patient medical records.
These values were used to calculate BMI, which was categorized as greater than or equal to 30 vs less than 30 kg/m 2 . According to the World Health Organization definitions, a BMI ≥ 30 is considered obese. 16 There are no clearly defined categories for VFA, SFA, WC, or PPAT. Therefore, adipose tissue measures were categorized as greater than or equal to median vs less than median for this cohort.
We also considered continuous measures of each fat type, measured per unit standard deviation.

| Prostate cancer treatment and outcome definitions
According to the Phoenix definition, 17 PSA recurrence was defined as at least 2 ng/mL above the post-radiation PSA nadir.
Recurrence was recorded on the date of the first PSA > 2 ng/mL above the nadir. Recurrence due to radiographic progression was defined as evidence of prostate cancer metastases on a scan in DI BELLA ET AL.
| 1245 the absence of PSA recurrence. Recurrence was recorded on the date of the scan. Recurrence due to treatment was defined as receiving either a radical prostatectomy after initial radiation, additional radiation more than 16 weeks after the start of initial radiation, or chemotherapy more than 6 months after the start of initial radiation. Recurrence was recorded on the date of initiation of salvage therapy. When determining the date of recurrence, definitions based on PSA recurrence or recurrence due to radiographic progression were prioritized over definitions based on recurrence due to treatment.
Among patients who received ADT as part of their initial radiation treatment, we categorized ADT use as either short-duration or long-duration. ADT was generally initiated 6 to 8 weeks before the start of radiation in both short-duration and long-duration users.
ADT use was considered short-duration if it lasted less than or equal to 6 months. Long-duration ADT consisted of ADT use that was longer than 6 months (commonly 18-36 months total). We did not collect information regarding the type of ADT.
All-cause mortality was abstracted from the electronic health records. Prostate cancer-specific mortality was defined as dying with progressive, metastatic castration-resistant prostate cancer and no other obvious cause of death.

| Statistical analysis
Patient demographic and clinical characteristics were summarized among patients overall and stratified by ADT use. Cox proportional hazards models were used to test the association between each measure of adiposity (overall: BMI; visceral: WC and VFA; subcutaneous: SFA; and pelvic: PPAT area and PPAT density) and risk of recurrence, prostate cancer-specific mortality, and all-cause mortality. We treated recurrence as our primary outcome, with prostate cancer-specific mortality and all-cause mortality treated as secondary outcomes. Each adiposity measure was examined in separate models due to collinearity and was categorized as described above.
Age-adjusted models were run as well as models adjusted for age (continuous), race (black and non-black), year of radiation (continuous), biopsy grade group (1, 2-3, 4-5), PSA (continuous, logtransformed), clinical stage (T1, T2/T3), and ADT (none, shortduration, and long-duration). Due to a low number of events, only an age-adjusted analysis was run for the prostate cancer-specific mortality outcome.
In a secondary analysis, models for recurrence were re-run stratified by race and the interaction between race and each adiposity measure was tested using a Wald test. Similarly, we stratified T A B L E 1 Demographic and clinical characteristics overall and by ADT use Age, mean ± SD 63.9 (6.7) 63. Abbreviations: ADT, androgen deprivation therapy; BMI, body mass index; PSA, prostate-specific antigen; TRUS, transrectal ultrasonography.
analyses for the recurrence outcome by receipt of ADT and the interactions between ADT use and each adiposity measure were tested. SAS 9.4 (SAS Institute Inc., Cary, NC) was used for statistical analyses. P < .05 was considered statistically significant.

| RESULTS
The mean (SD) age of our study cohort at diagnosis was 63.9 (6.7) years and 238 (59%) men were black (  Table 2). When stratified by race, similar, nonsignificant results for the association between adiposity measures and risk of recurrence were seen in black men compared with nonblack men (Table 3). When adiposity measures were modeled as continuous variables per standard deviation unit, again we found no associations between adiposity measures and any outcomes (data not shown).
Despite these null findings overall, we found statistically significant interactions between ADT use and WC (P = .025), VFA (P = .002), SFA (P = .010), PPAT area (P = .002), PPAT density (P = .002), but not BMI (P = .12), in predicting risk of recurrence. Thus, given this evidence for ADT acting as an effect modifier of the association between adiposity measures and recurrence risk, models were stratified by ADT use. Among men who did not receive ADT, multivariable-adjusted hazard ratios (HRs) for the association between WC, VFA, SFA, PPAT area, and PPAT density and recurrence ranged from 1.17 to 1.79, indicating a positive association between higher adiposity measures and increased risk of recurrence, although none reached statistical significance (Table 4). In contrast, among men who did receive ADT, the multivariable-adjusted HRs for associations between WC, VFA, SFA, PPAT area, and PPAT density and recurrence ranged from 0.46 to 0.83, indicating an inverse association between higher adiposity and risk of recurrence, although again these did not reach statistical significance, with the exception of PPAT area (

| DISCUSSION
Visceral obesity may be a better measure of a metabolically unhealthy obesity phenotype than BMI, 18  has been linked with prostate cancer risk in some [23][24][25] but not all 26,27 studies. As such, measuring adipose tissue distribution and, in particular, visceral adiposity may be key to improving our understanding of the relationship between obesity and prostate cancer outcomes.
One of our hypotheses focused on PPAT due to previous in vitro research that showed that coculture with PPAT produced an aggressive phenotype in prostate cancer cells. 28,29 One study used magnetic resonance spectroscopy and found that the fatty acid composition was altered in PPAT of patients with aggressive prostate cancer. 30 When observing secretions from PPAT explants, PPATconditioned media from more obese patients caused significantly more proliferation of prostate cancer and endothelial cells in vitro than PPAT-conditioned media obtained from leaner men. 31 Other research using gene expression of PPAT found that PPAT of obese men had higher metalloproteinase activity, which contributes to immunoinflammatory responses and ultimately promotes oncogenesis. 32 Although some epidemiological data support a relationship between PPAT and prostate cancer aggressiveness, 14,33 only one study before the current study, to our knowledge, examined the association of PPAT with prostate cancer outcomes. This study, among men receiving primary ADT, found that PPAT volume was significantly higher in patients who developed castration-resistant prostate cancer. 34 In contrast, although we did not study castrationresistant prostate cancer, we found no association between PPAT and prostate cancer recurrence or mortality. Given the extremely few studies in this area, more are needed to determine if knowledge of PPAT area could inform prostate cancer prognosis.
Our findings from prespecified secondary analysis suggest that abdominal and pelvic obesity may have varying effects on prostate cancer progression depending on the receipt of ADT. Specifically, T A B L E 3 Hazard ratios and 95% confidence intervals for associations between measures of adiposity and risk of recurrence, stratified by race  Declining testosterone levels, achieved with ADT, correlate with increasing body fat accumulation and decreasing lean body mass. 36 ADT is also related to metabolic changes such as decreasing insulin sensitivity and increasing low-density lipoprotein cholesterol, highdensity lipoprotein cholesterol, and triglycerides. 37 Another factor that may have contributed to significant differences in associations between obesity and outcomes by ADT is that tumor characteristics varied by ADT. Men receiving ADT had more aggressive tumors than those who received radiation alone and the effect of obesity on outcomes in these men may be masked by the more aggressive clinical course of their disease.
Our observation that associations between obesity and prostate cancer outcomes varied significantly by ADT use could have a biological explanation. To our knowledge, no studies have investigated associations between obesity and prostate cancer outcomes and how they may vary by receipt of hormone therapy. However, in a metaanalysis that looked at obesity and breast cancer by hormone therapy, there was a stronger association between obesity and postmenopausal breast cancer risk in women not using estrogenprogestin therapy. 38 It has been proposed that exogenous hormones provided by estrogen-progestin therapy may mask the effect of altered levels of endogenous hormones in obesity, thereby blocking the effect of obesity on breast cancer risk. Among women using estrogen-progestin therapy, estrogen levels are elevated.
Therefore, the relative impact of adipose tissue estrogen production on the tumor is expected to be reduced. 38  ADT. However, given that this is the first study to test this hypothesis in prostate cancer, more research is necessary to determine if hormone therapy could modify associations between obesity and outcomes in prostate cancer patients.
This study has several limitations and strengths that should be considered. For example, despite the relatively long follow-up (median of 9.3 years) there was a relatively low number of events, specifically for outcomes such as prostate cancer-specific mortality, which led us to use recurrence as the primary outcome.
Also, the cohort was composed of patients treated with radiation due to the availability of planning CT scans for determining adipose tissue measurements. Therefore, this may limit generalizability of our results when looking at all patients with prostate cancer undergoing various treatment regimens. Future studies are necessary to determine how outcomes in different treatment groups are affected by adipose tissue distribution. One major strength of this study is the racially diverse population. The cohort was 59% black, which allowed for more generalizability and unique contrasts to previous studies with cohorts that were predominantly made up of white men.
In conclusion, most findings in this study of radiotherapy-treated patients were null, such as associations between adiposity measures and risk of recurrence. However, in secondary analyses, we found