To compare 5-year biochemical control, toxicity, and patient-reported quality of life (QOL) outcomes for African American and White patients treated with proton therapy (PT) for prostate cancer.
We reviewed the medical records of 1,066 men with clinically localized prostate cancer. Patients were treated with definitive PT between 2006 and 2010. Patients received a median radiation dose of 78 Gy (RBE) with conventional fractionation (1.8- 2 Gy [RBE] per fraction). Sixty-eight (6.4%) men self-identified as African American and 998 (93.6%) self-identified as White. Five-year rates of biochemical control, grade 3 genitourinary and gastrointestinal toxicity, and patient-reported QOL are reported and compared between African American and White patients.
Median biochemical follow-up was 5.0 years for both African American and White patients. Median follow-up for toxicity was 5.0 and 5.2 years, respectively. On multivariate analysis, race was not a significant predictor for 5-year freedom from biochemical failure (HR 0.8, p=0.55). No significant association was found between race and grade 3 genitourinary toxicity on multivariate analysis at 5 years (HR 2.5, p=0.10). Patient-reported QOL using median EPIC bowel, urinary incontinence, and irritative summaries scores were not significantly different between the groups. African Americans had higher median sexual summary scores at 2 years than White patients (75 vs. 54, p=0.01) but by 5+ years, the sexual summary scores were no longer significantly different (63 vs. 53, p=0.35).
With a median follow-up of 5 years, there were no racial disparities in biochemical control, grade 3 toxicity, or patient-reported QOL after PT for prostate cancer.
In 2016, over 180,000 men in the United States are expected to be diagnosed with prostate cancer . For many years, disparities have existed between African American and White men in the incidence of prostate cancer and prostate cancer-specific mortality. Although these racial disparities have narrowed in recent years, between 2008 and 2012 African American men still had a 70% higher incidence of prostate cancer than White men, and prostate cancer-specific mortality was 138% higher than that for White men . Such differences in the incidence and mortality of African American men may relate to genetic mutations that promote prostate cancer progression ; geographic, socioeconomic, or educational barriers to medical care [4, 5]; and/or the efficacy of available treatments, such as radical prostatectomy or radiation therapy, which may not be as effective for African Americans as for White patients. In support of the last potential cause, several studies have shown race to be an independent risk factor for recurrence after definitive prostate cancer treatment [6–11]. Similarly, some studies have shown that African Americans may have a higher risk for toxicity and worse quality of life (QOL) after definitive prostate cancer treatment [12, 13].
In the last 20 years, the efficacy and safety of radiation therapy has greatly improved. The likelihood of biochemical control with modern high-dose radiation has increased by 10% to 30%, and the risk for toxicity continues to decrease [14–16]. In the case of proton therapy (PT), a highly conformal type of advanced radiation therapy, early outcomes for patients with prostate cancer have shown biochemical control rates over 90% at 5 years and grade 3 toxicity rates below 3% . In light of such technological advancements, an important question becomes, do racial disparities still exist in tumor response to high dose conformal radiation therapy? The purpose of our study was to determine whether race influences tumor control, the risk for treatment-related toxicity, or patient-reported QOL following definitive PT.
Materials and Methods
This study was an institutional review board-approved review of 1,538 men with biopsy-proven localized prostate cancer enrolled in an outcome tracking protocol and treated with PT at our institution between 2006 and 2010. Patients were excluded from this analysis if they had nodal metastasis present before treatment (n=13), prior local treatment for prostate cancer (n=14), did not complete PT (n=5), refused, were ineligible for, or withdrew from the outcomes tracking protocol (n=19), or if there was no clinical follow-up information available (n=11). Patients were also excluded if they were treated on a prospective protocol (n=209), treated with hypofractionated PT (n=141), or if they had less than 2 years of biochemical follow up for reasons other than death (n=60). A total of 1,066 men were included in this analysis and each patient had biochemical, clinical, and QOL data collected prospectively at regular intervals. A total of 68 men (6.4%) self-identified as African American and 998 (93.6%) men self-identified as White.
Patient histories were reviewed and potential risk factors for treatment failure and toxicity were recorded, including maximum prostatic specific antigen (PSA) level, clinical stage, the results of pretreatment staging studies, prostate size per transrectal ultrasound findings at the time of fiducial marker placement, maximum Gleason score, maximum percentage of involvement in any biopsy core, percentage of zones involved on prostate biopsy, and the use of androgen deprivation therapy (ADT). A 10-zone or more prostate biopsy was recommended within 6 months of beginning PT. Pretreatment work-up for all patients included medical history, digital rectal examination, and in-house pathology review of prostate biopsies to verify the diagnosis and Gleason score. Work-up also included PSA, 1.5- to 3.0-Tesla magnetic resonance prostate imaging (MRI) and computed tomography (CT) scans of the pelvis, and bone scans in patients with intermediate- and high-risk disease. Baseline patient and treatment characteristics are provided in Table 1.
Treatment Simulation, Planning, and Delivery
Treatment planning procedures have also been previously described . The clinical target volume (CTV) for low-risk patients included only the prostate as visualized on fused MRI and CT images; the CTV in intermediate- and high-risk patients also included the proximal 2 cm of seminal vesicles. A reduction off of the seminal vesicles was allowed in patients with intermediate-risk disease per physician discretion. For high-risk patients with a risk for pelvic node involvement >15%, intensity-modulated radiation therapy was delivered to the initial CTV, which included the pelvic nodes, the proximal seminal vesicles, and the prostate. Initially, the CTV was expanded by 5 mm axially and 8 mm in the superior and inferior axes to create a planning target volume (PTV), but in 2008 the parameters for PTV expansion were reduced to 4 mm axially and 6 mm in the superior and inferior axes in response to an in-house intrafraction motion analysis. Lateral or lateral anterior oblique fields were employed with PT. Treatment was planned with an Eclipse system (Varian Medical Systems, Palo Alto, CA) using a CT Hounsfield conversion algorithm . Distal and proximal range uncertainty margins of 5 mm and a smearing value of 19 mm were used.
Treatment planning guidelines included goals for both target coverage and dose constraints for organs at risk, including the bladder, bladder wall, rectum, rectal wall, and femoral heads. For target coverage, 95% of the PTV received 100% of the prescribed dose and 100% of the PTV received at least 95% of the prescribed dose. ADT was recommended for patients with high-risk prostate cancer for a duration of 6 to 24 months. Most patients with low- or intermediate-risk disease who had ADT had it prescribed by outside physicians before consultation for PT. One hundred seventy patients (15.9%) received neoadjuvant, concurrent, or adjuvant ADT. Treatment characteristics for the patient cohort are listed in Table 1.
Follow-Up and Observed Outcome
Patients were seen at 6-month intervals after treatment to prospectively assess toxicity using Common Terminology Criteria for Adverse Events, version 3.0 (CTCAEv3) . Follow-up also included a medical history review and physical examinations at 6-month intervals and PSA tests performed every 3 months. Biochemical failure was determined using the Phoenix definition (nadir + 2 ng/ml) . The time to reported biochemical outcomes was calculated from the radiation start date. Clinical failure (local, regional, or distant) was based on available clinical, histologic, or radiographic evidence of disease recurrence. In the event of biochemical failure, patients had bone scans, pelvic MRI, and occasionally positron emission tomography-CT. Biopsy of the prostate was not performed unless pelvic MRI or a digital rectal examination suggested possible local disease progression. Patients also prospectively completed International Prostate Symptom Score and QOL assessments (such as the Expanded Prostate Cancer Index Composite; EPIC) at 6-month intervals after treatment for 5 years, then annually thereafter. The EPIC summary and subscales were calculated and reported using a scale of 0 to 100, with higher scores indicating better outcomes. Toxicities occurring ≥6 months after PT were scored as “late” and those occurring during treatment or <6 months after treatment were scored as “acute.” CTCAE, version 4.0 (CTCAEv4)  was published in 2009; all patients with CTCAEv3 grade 3 toxicities were retrospectively categorized using CTCAEv4 criteria.
Grade 3 genitourinary (GU) toxicities included urinary frequency or urgency resulting in urination ≥1 time per hour or necessitating a catheter, urinary retention requiring more than daily catheterization or surgical intervention such as a transurethral resection of the prostate, transurethral needle ablation, or indwelling suprapubic catheter, hematuria requiring blood transfusion(s), hyperbaric oxygen treatment, or a surgical intervention such as a cystoscopy with a procedure (either a biopsy showing necrosis or cauterization). Cystoscopies revealing normal bladder and not requiring an intervention (such as biopsy, cautery, or resection) were not scored as a grade 3 toxicity. Grade 3 gastrointestinal (GI) toxicities included episodes of rectal bleeding or ulceration requiring transfusion, hyperbaric oxygen, surgical intervention, or those affecting instrumental self-care activities of daily living.
Simulation, Planning, and Treatment
SAS and JMP software were utilized for all statistical analysis (SAS Institute, Cary, NC). The extent of correlation between race and a series of nominal covariates was assessed with Fisher's exact test. Correlations between race and baseline domain scores from the EPIC survey were assessed with the Wilcoxon test, a nonparametric analog to the independent t-test. The Kaplan-Meier product limit method provided estimates of freedom from biochemical progression (FFBP) and genitourinary (GU) and GI toxicity; the log-rank test statistic provided an estimate of the ability of race to predict for each of these endpoints. Proportional hazards regression of these time-dependent endpoints using race and other selected prognostic factors allowed detection of any independently predictive prognostic factors; forward selection was utilized to render the most parsimonious final model.
The median follow-up for biochemical outcomes was 5.0 years (range, 0.5 to 7.0 years) for African American patients and 5.0 years (range, 0.5 to 8.2 years) for White patients. For patients who did not receive ADT, the median PSA nadir by race was 0.49 ng/ml for African Americans and 0.3 ng/ml for Whites, a statistically significant difference (p<0.01). The median time to PSA nadir was 3 years for both African American (range, 0 to 6.1 years) and White (range, 0 to 7.4 years) patients. Biochemical failure occurred in 6 (8.8%) African American patients at a median of 2.8 years (range, 0.8 to 5.4) and 81 (8.1%) White patients at a median of 3.4 years (range, 0.3 to 7.1 years). As shown in Figure 1, 5-year FFBP rates were 92.1% for African Americans and 92.4% for White patients, and the values were not significantly different (p=0.65). The FFBP rates by risk group for African American and White patients were as follows: 95.5% vs. 99.1%, p=0.08, for low-risk; 96.4% vs. 92.6%, p=0.73, for intermediate-risk; and 78.3% vs. 74.6%, p=0.87, for high-risk prostate cancer.
The risk for clinical failure and survival was also not affected by race. Five-year distant metastasis-free survival was 96.9% for African American and 92.6% for White patients (p=0.96). Similarly, overall survival rates (93.9% vs. 96.4%; p=0.12) and cause-specific survival rates (97.0% vs. 97.3%; p=0.81) were not significantly different for African American and White patients, respectively. The risk for nodal failure was 0 (0%) vs. 14 (1.4%) for African Americans and White patients, respectively (p=0.99). A total of 2 (2.9%) African American patients experienced biopsy-proven local failure versus 13 (1.3%) White patients, and the difference was not significantly different (p=0.24).
Predictors of Biochemical Failure
Results from the univariate and multivariate analyses of factors potentially predictive of 5-year FFBP are shown in Table 2. On multivariate analysis, risk group (low- vs. intermediate- vs. high risk; p<0.01), the presence of perineural invasion (no vs. yes; p<0.01), and the percentage of positive zones on biopsy (<50% vs. ≥50; p=0.02) were significant predictors of biochemical failure. Race was not a predictor of 5-year FFBP (HR 0.8, p=0.55).
Univariate and multivariate analyses of clinical and treatment factors potentially associated with late CTCAEv4 grade 3+ GU toxicities in all patients are shown in Table 3. African Americans had a higher risk for CTCAEv4 grade 3+ GU toxicity when compared to White patients (6.4% vs. 2.1%), but the difference was not statistically significant (p=0.06). On multivariate analysis, significant associations were found between late CTCAEv4 grade 3+ GU toxicities and pretreatment use of alpha-blockers (p<0.01) and pretreatment transurethral resection of the prostate (p<0.01). Race was not a predictor of CTCAEv4 grade 3+ GU toxicity (HR 2.5; p=0.10). Late CTCAEv3 and v4 grade 3 GI symptoms occurred in 7 patients. The 5-year actuarial incidence of late grade 3 GI toxicity was 0.8% in White patients and 0% in African Americans, and this difference was not statistically significant (p=0.5).
As shown in Figure 2, patient-reported QOL was similar between African American patients and White patients in follow-up. The median EPIC summary scores for bowel, urinary irritative/obstructive, and urinary incontinence domains were not significantly different for African American patients and White patients in follow-up. African Americans had higher median EPIC sexual summary scores at 2 years than White patients (75 vs. 54; p=0.01), but by 5+ years of follow up, the EPIC sexual summary scores were no longer significantly different (63 vs. 53, p=0.35). Median International Prostate Symptom Scores did not differ at baseline (6 vs. 8; p=0.06) or 5+ years (7 vs. 7; p=0.69) of follow-up after PT for African American and White patients.
This study evaluates the effect of race on tumor control, toxicity, and QOL at a high-volume PT center. In our population, race did not appear to play a role in predicting biochemical control or GI toxicity. Although QOL was very similar between the groups, African Americans at 2 years maintained better sexual health than White patients. This difference was short-lived, however, with no difference seen by 5+ years in EPIC sexual summary scores. Grade 3 GU toxicity was more common among African Americans but the difference was not statistically significant. With greater patient numbers and longer follow-up the differences may achieve statistical significance. The apparent increase in the rate of grade 3 GU toxicity among African Americans may be attributable to pretreatment prognostic factors. Compared to White patients, African Americans were more likely to have diabetes mellitus, which is a known risk factor for GU toxicity after radiation therapy . Although this factor was included in the multivariate analysis, other unaccounted-for differences in treatment or pretreatment characteristics may have played a role in the observed difference in GU toxicity.
Results have been mixed when accounting for the relationship between race and GU toxicity after prostate cancer treatment. Sanda et al published a prospective cohort study of 1,201 patients with prostate cancer treated with brachytherapy, radical prostatectomy, or external-beam radiation therapy . Patient-reported QOL was assessed prospectively using the EPIC QOL form. With a median follow-up of 30 months, the authors found that African American race was associated with worse GU toxicity following prostate cancer treatment. Specifically, African Americans developed urinary incontinence more often than White patients after radical prostatectomy. Similarly, Rice et al published a prospective cohort study including prostate cancer patients treated with external-beam radiation therapy or radical prostatectomy . At a median follow up of 12 months, African Americans had a greater decline in EPIC urinary function scores after treatment regardless of treatment choice. Michalski et al published results from the prospective Radiation Therapy Oncology Group trial 0126, which suggested that disparities may not be present when conformal radiation therapies like intensity-modulated radiation therapy are delivered. With a cohort of 743 men, it was shown that the rate of physician-reported GU toxicity was not affected by race . Furthermore, in a large study including over 1,700 men treated with radical prostatectomy at a tertiary care center, race was not associated with outcome. In the publication by Prabhu et al, urinary function was scored prospectively using the University of California, Los Angeles–Prostate Cancer Index–Urinary Function Index and patients were followed for a median of 120 months . The authors found that race was not associated with urinary function following radical prostatectomy.
Our study shows that race has no effect on the risk of disease recurrence after PT. This finding underscores the robust effectiveness of high-dose PT in the management of prostate cancer. Our findings, when combined with the broader literature, also support the position that when patients with prostate cancer are treated on prospective protocols, treated at well-established cancer centers, and/or equal-access treatment centers, tumor control outcomes are very similar between White patients and African American patients. For example, Lee et al published results of in a series of 246 men with prostate cancer treated with brachytherapy at Memorial-Sloan Kettering Cancer Center (New York, NY) between 1992 and 1997 . Patients were treated to a median dose of 140 Gy and, when controlling for pretreatment Gleason score, PSA, and tumor stage, race was not a predictor of 5-year biochemical recurrence-free survival. Similarly, Kupelian et al published results from a series of men treated for prostate cancer at the Cleveland Clinic (Cleveland, OH) with external-beam photon radiation therapy . Patients were treated to a median of 70 Gy using conventionally fractionated photon radiation. When 5-year overall survival or biochemical progression-free survival were examined on multivariate analysis, race was not predictive of either outcome. Roach et al evaluated results from 3 combined Radiation Therapy Oncology Group protocols: 7506, 7706, and 8307 . Over 1,200 patients were included and 120 patients (9%) were African American. All patients were treated with external-beam radiation between 1976 and 1985. On multivariate analysis, race was not found to be predictive of survival without evidence of disease in any of the protocols evaluated. Finally, Graham-Steed evaluated the results of 1,270 men treated for prostate cancer with surgery or radiation therapy in the Veteran's Administration Health System . With a minimum of 11 years follow-up, African American race was not independently associated with prostate cancer-specific mortality.
Conversely, population-based studies evaluating the effect of race in response to radiation therapy, especially those using the Surveillance, Epidemiology, and End Results Program (SEER) database, often show that race is an independent risk factor for recurrence following treatment for localized prostate cancer. For example, Austin et al published a retrospective review of data from the SEER database . It included 914 patients treated with external-beam radiation therapy from 1973 to 1987. African Americans presented at younger ages than White patients and, on multivariate analysis, when controlling for stage and tumor grade, African Americans had worse overall survival following radiation therapy than White patients. Cohen et al published data from a SEER-Medicare linked database that included over 28,000 men treated for prostate cancer between 1986 and 1996 in the U.S. . Patients were treated with either prostatectomy or radiation therapy. At 10 years of follow-up, disease-free survival was 58% and 65% for African American and White patients, respectively, and the difference was statistically significant. Furthermore, on multivariate analysis, African American race was independently predictive for disease recurrence after surgery. Because patients treated at established cancer centers appear to have fewer racial disparities in treatment outcomes for prostate cancer, the disparities seen among the unselected patients in SEER-based studies suggest that the racial disparities in outcomes in the U.S. may be linked to differences in access to quality care. African Americans may have worse disease at the time of diagnosis because of a delay in diagnosis or a delay in treatment. Alternatively, racial disparities may exist because African Americans receive inappropriate treatments more often than White patients. Indeed, Stokes et al, also using data from a SEER-Medicare linked database, showed that the time from diagnosis to treatment was longer for African American men with prostate cancer when compared to White patients in the U.S. . The disparity in treatment delays was greatest for patients with high-risk prostate cancer and the authors suggested that these delays were related to poor access to care, which could contribute to disease outcome disparities. Furthermore, Mahal et al in a similar study found that, between 2007 and 2010, African Americans with localized prostate cancer were less likely to receive definitive treatment for prostate cancer than White patients in the U.S. . The authors found that insurance status affected this risk since African Americans with insurance were equally likely as Whites to be treated for prostate cancer. This study, among others, suggests that access to care remains a major issue in the U.S. among patients at risk for prostate cancer, which could contribute to the racial disparities seen in prostate cancer-specific mortality.
African Americans may have biologically more aggressive prostate cancer at the time of diagnosis as well. In our study, African Americans presented with a median age that was 3 years younger than White patients and a median PSA level that was significantly higher than White patients. This is similar to several studies showing that African Americans present with prostate cancer at younger ages, have higher grade tumors, and have higher PSA levels than White patients . Molecular and genetic differences have been reported as possible explanations to these disparities. For example, Gaston et al found that androgen receptor expression on prostate cancer and benign prostate cells was more common among African Americans than White patients . The study suggested that androgen receptor activity among African Americans was higher than among White patients, potentially leading to more aggressive prostate cancer at the time of diagnosis. Genetic mutations may account for racial differences in disease aggressiveness. Although the specific genes that are involved are still being investigated, Freedman et al found that mutations within the 8q24 locus increase the risk for prostate cancer particularly for men younger than 72 years of age, and that mutations in this region are more common among African Americans than White patients . Epigenetic changes have recently been observed that can promote and maintain the malignant phenotype of prostate cancer. Some of these epigenetic changes have been found to be more common among African Americans with prostate cancer. For example, Devaney et al demonstrated that African American patients with prostate cancer have higher levels of promoter methylation at the SNRPN and ABCG5 genes when compared to White patients . Because these genes regulate tumor invasiveness and proliferation, and methylation is thought to be correlated with more aggressive tumors, the differences seen in methylation may be responsible for the development of more aggressive disease among African Americans.
Our study has several limitations, including the potential for selection bias related to clinical factors and access to care. The patients who chose to be treated with PT may not be directly comparable to patients who receive treatment at other facilities or regions. Patients who choose to be treated with PT often have the ability to travel long distances for treatment and have extensively educated themselves on the treatment options for prostate cancer in the U.S., potentially making them a unique population, not comparable to other prostate cancer populations. Additionally, race is increasingly being recognized as a poor surrogate for other more important factors including culture, genetics, and geography. We were not able to include data representing these other important factors in this study. The strengths of this analysis are its large cohort size, short accrual times, and consistent patient management at a single-institution high-volume treatment center with consistent guidelines for radiation delivery, follow-up, and toxicity management. Another important strength is the simultaneous and prospective assessment of outcomes, such as disease control, patient-reported QOL, and physician-reported toxicity assessment.
Race was not an independent risk factor for recurrence after treatment with high-dose PT. With a median follow-up of 5 years, there were no racial disparities in biochemical control, grade 3 toxicity, or patient-reported QOL after PT for prostate cancer. If these results are maintained with longer follow-up and with greater patient numbers, it would suggest that racial differences in prostate cancer-specific mortality may not be related to response to high-dose radiation therapy when controlling for other factors like access to care.
ADDITIONAL INFORMATION AND DECLARATIONS
Conflicts of Interest: The authors have no conflicts of interest to disclose.