Survival advantage of Asian metastatic prostate cancer patients treated with external beam radiotherapy over other races/ethnicities

Survival in metastatic prostate cancer (mPCa), despite promising new systemic therapies, remains low [1]. Due to its heterogeneity at presentation, no “one fits all” treatment strategy for mPCa is available [2]. Data from the multi-arm, multi-stage “STAMPEDE” trial investigating androgen-deprivation therapy (ADT) versus ADT plus external beam radiotherapy (EBRT) to the prostate revealed added survival benefit of EBRT to the prostate in a subgroup of patients with low-volume metastatic burden [3]. Since then, EBRT to the prostate in low-volume mPCa became standard of care and complements the multitude of systemic treatment options that are available [4‐6]. Recently, the efficacy of EBRT to the prostate in low-volume mPCa was also validated in North American patients [7].

However, no stratification according to race/ethnicity was applied in neither the original STAMPEDE trial nor in the subsequent North American validation study. Based on evidence suggesting racial/ethnic differences in mPCa survival outcomes, it is of interest to test whether some racial/ethnic groups may benefit of EBRT more than others [8‐10]. We addressed this void and tested for survival differences between Caucasian, African-American, Hispanic/Latino and Asian mPCa patients who received EBRT to the prostate within the SEER database. Although SEER does not provide information on applied radiation doses, types or fields, to the best of our knowledge, there is no basis for substantial race/ethnicity-related differences in this regard. For this reason, we relied on the SEER-derived EBRT variable in mPCa patients as surrogate for the type of treatment that was performed within the STAMPEDE trial. We hypothesized, that after adjustment for other-cause mortality (OCM), differences may exist in cancer-specific mortality (CSM) between different racial/ethnic groups.

Within the SEER database, we identified newly diagnosed mPCa Caucasian, African-American, Hispanic/Latino and Asian patients who were treated with EBRT to the prostate between 2004 and 2016. These selection criteria resulted in a cohort of 4282 assessable patients: 2737 Caucasians, 797 African-Americans, 469 Hispanic/Latinos and 252 Asians. Patient age at diagnosis, year of diagnosis, prostate-specific antigen (PSA) at diagnosis, clinical T and N stage (cT, cN) as well as biopsy Gleason Grade Group (GGG) and M1 substages (M1a, M1b and M1c according to the seventh Edition of the American Joint Committee on Cancer Staging Manual, AJCC [11]), represented covariates. The outcome variable of interest was CSM. OCM was also quantified and used for adjustment in competing risks analyses focusing on CSM.

We hypothesized that important survival differences may exist between race/ethnicity groups in newly diagnosed mPCa treated with EBRT to the prostate. To test our hypothesis, we relied on CSM and OCM data recorded in the SEER database between 2004 and 2016. Our analyses yielded several noteworthy findings.

First, considering patient and tumor characteristics, we generally recorded no meaningful differences between race/ethnicity at initial presentation. Two exceptions were identified. The first one consisted of oldest age at presentation in Asians (median 70 years), followed by Caucasians (median 68 years), Hispanic/Latinos (median 66 years) and African-Americans (median 63 years), in that order. Additionally, Asians also exhibited the second highest rates of visceral metastases (21.8%) after Hispanic/Latinos (22.4%), compared to African-Americans (19.3%) and Caucasians (16.3%). These observations indicate that Asian patients are disadvantaged with respect to survival probability based on more advanced age and based on highest rate of visceral metastasis. In consequence, Asian patients would be expected to exhibit worse survival characteristics than other race/ethnic groups, which was not the case in the current analyses, since Asians exhibited most favorable survival outcomes.

In the second part of the analyses, we focused on CSM. Here, we relied on competing risks regression to account for the confounding effect of OCM. This methodology was applied in univariable, as well as in multivariable analyses. In multivariable analyses, besides adjustment for OCM, our modeling strategy also adjusted for residual differences in patient and prostate cancer characteristics that may still exist between race/ethnic groups despite absence of statistically significant findings identified in descriptive analyses (Table 1). Here, Asians exhibited lower CSM (47.8% at 60 months and 59.5% at 120 months) than Caucasians (61.1% at 60 months and 71.5% at 120 months), Hispanic/Latinos (61.0% at 60 months and 67.8% at 120 months) and African-Americans (60.6% at 60 months and 72.0% at 120 months). Conversely, no statistically significant differences were recorded between African-Americans and Caucasians, as well as between Hispanic/Latinos and Asians. These observations suggest that despite multivariable adjustment for residual differences in patient and prostate cancer characteristics, as well as after adjustment for OCM, Asian mPCa patients treated with EBRT to the prostate exhibit better CSM outcomes than the three other examined racial/ethnic groups.

To the best of our knowledge, we are the first to examine racial/ethnic group differences in mPCa treated with EBRT. However, Asian prostate cancer patients have also been found to have more favorable prognosis than other races/ethnicities in other contexts. For example, in a recent report by Deuker et al., lower CSM of Asian patients compared to Caucasian patients in both localized PCa as well as mPCa has been observed [14]. However, in their analyses, Deuker et al. did not focus on EBRT-treated mPCa patients. Furthermore, our findings also contrast with other reports that stated worse survival rates in African-American prostate cancer patients or more favorable survival in Hispanic/Latino prostate cancer patients [15‐19]. However, also none of these reports focused on a subgroup of mPCa patients treated by EBRT. In consequence, our results cannot be directly compared to any other previous study. Apart from potentially other unmeasured prognostic factors within the SEER database, it is a matter of discussion whether differences in modifiable factors (e.g. lifestyle and dietary, access to health-care, socioeconomic status) and/or non-modifiable factors (e.g. genetic factors or differences in testosterone levels, radiosensitivity of the prostate, response to systemic treatment), might play a role in the lower CSM of Asian mPCa patients [8, 9].

Taken together, we performed a detailed analysis of the effect of race/ethnicity on CSM in newly diagnosed mPCa treated with EBRT to the prostate. We observed lower CSM in Asians, relative to Caucasians. Conversely, no differences in CSM were recorded between African-Americans and Caucasians as well as Hispanic/Latinos and Caucasians. The observed CSM advantage of Asian patients persisted despite detailed multivariable adjustment, adjustment for OCM as well as propensity score matching for all available patient and prostate cancer characteristics. In consequence, our observations provide robust, albeit retrospective evidence that the prognosis of newly diagnosed Asian mPCa patients treated with EBRT to the prostate is better than that of other racial/ethnic groups. This observation could not have been made in the original STAMPEDE [3] trial, since stratification for race/ethnicity was neither planned nor possible. Our findings should ideally be validated within other large scaled prospective or retrospective databases. Unfortunately, the nature of the SEER database does not allow to test for patient or disease characteristics that could explain why Asian patients displayed better survival.

It is of note that we could not strictly apply the CHAARTED [6] study definition of low-volume newly diagnosed mPCa according to available fields within the SEER database. However, EBRT to the prostate is only guideline-recommended in low-volume mPCa [2]. Nevertheless, no evidence existed to support EBRT to the prostate in patients other than those with low-volume mPCa. In consequence, it is highly unlikely, that a significant proportion of our individuals harbored high volume disease. Moreover, no evidence suggests that clinically meaningful differences in mPCa existed between racial/ethnic groups, as evidenced by absence of significant differences in clinical cancer characteristics such as PSA, GGG or cT/N stages. Despite some minor differences in M1 substage distributions between races/ethnicities, we accounted for these confounders by the means of propensity score matching. In consequence, differential bias in EBRT use due to differences in PCa characteristics between racial/ethnic groups is unlikely operational in our cohort. However, SEER neither accounts for the type or dose of applied systemic treatment nor for the applied radiation dose or type. Nonetheless, there is no basis for differences in rates or types of treatments administered according to racial/ethnic groups. In consequence, it is unlikely that those limitations represented sources of differential bias or confounding.

In the setting of EBRT-treated mPCa patients, Asian race/ethnicity exhibits more favorable CSM than Caucasian, African-American and Hispanic/Latino race/ethnicity. This observation may warrant consideration in prognostic stratification schemes for newly diagnosed mPCa patients.