Differences in prognostic factors and survival among white and Asian men with prostate cancer, California, 1995–2004

Authors


  • The collection of cancer incidence data used in this study was supported by the California Department of Health Services as part of the statewide cancer reporting program mandated by California Health and Safety Code Section 103885, the National Cancer Institute's Surveillance, Epidemiology, and End Results Program under contract N01-PC-35136 awarded to the Northern California Cancer Center; contract N01-PC-35139 awarded to the University of Southern California; and contract N02-PC-15105 awarded to the Public Health Institute; and the Centers for Disease Control and Prevention's National Program of Cancer Registries, under agreement U55/CCR921930-02 awarded to the Public Health Institute.

  • The ideas and opinions expressed herein are those of the authors and endorsement by the State of California, Department of Health Services, the National Cancer Institute, and the Centers for Disease Control and Prevention or their contractors and subcontractors is not intended nor should be inferred.

Abstract

BACKGROUND.

There are very limited data concerning survival from prostate cancer among Asian subgroups living in the U.S., a large proportion of whom reside in California. There do not appear to be any published data on prostate cancer survival for the more recently immigrated Asian subgroups (Korean, South Asian [SA], and Vietnamese).

METHODS.

A study of prognostic factors and survival from prostate cancer was conducted in non-Hispanic whites and 6 Asian subgroups (Chinese, Filipino, Japanese, Korean, SA, and Vietnamese), using data from all men in California diagnosed with incident prostate cancer during 1995–2004 and followed through 2004 (n = 116,916). Survival was analyzed using Cox proportional hazards models.

RESULTS.

Whites and Asians demonstrated significant racial differences in all prognostic factors: age, summary stage, primary treatment, histologic grade, socioeconomic status, and year of diagnosis. Every Asian subgroup had a risk factor profile that put them at a survival disadvantage compared with whites. Overall, the 10-year risk of death from prostate cancer was 11.9%. However, in unadjusted analyses Japanese men had significantly better survival than whites; Chinese, Filipino, Korean, and Vietnamese men had statistically equal survival; and SA men had significantly lower survival. On multivariate analyses adjusting for all prognostic factors, all subgroups except SA and Vietnamese men had significantly better survival than whites; the latter 2 groups had statistically equal survival.

CONCLUSIONS.

Traditional prognostic factors for survival from prostate cancer do not explain why most Asian men have better survival compared with whites, but they do explain the poorer survival of SA men compared with whites. Cancer 2007. © 2007 American Cancer Society.

Studies conducted in the U.S. have consistently found that the incidence of prostate cancer among Asian men is much lower than among nearly all other racial/ethnic groups.1 In contrast to the data available on incidence, to our knowledge there are few published data regarding survival from prostate cancer among Asians in the U.S.2–5 A national study of survival from the most common cancers reported that, in the aggregate, Asians had better prostate cancer survival than non-Hispanic whites.2 However, grouping all Asians into 1 category is inadvisable, because it may mask differences in incidence and survival among various subgroups.6 For example, some studies have found that, compared with whites, Chinese and Japanese men may have higher survival from prostate cancer, whereas Filipino men may have lower survival.3, 4 Lin et al.3 noted that the survival advantage for Chinese and Japanese men occurred despite a higher prevalence of distant stage at diagnosis. An analysis of racial survival differences using national cancer registry data from the 1970s found that Hawaiian, Japanese, Chinese, and Filipino men all had higher survival from prostate cancer than non-Hispanic white men.5 Because the determinants of prostate cancer survival are not well understood, analysis of racial differences may be an important source of hypotheses for future research.

Although there are limited published data regarding prostate cancer survival in the more well-established Asian subgroups (Chinese, Filipino, and Japanese), to our knowledge there are no published data concerning prostate cancer survival for the more recently immigrated Asian subgroups, such as Koreans, South Asians, and Vietnamese. ‘South Asian’ refers to persons tracing their ancestry to India, Pakistan, Bangladesh, Sri Lanka, Nepal, and Bhutan. In the 2000 Census, South Asians represented the third largest Asian subgroup in the U.S., with a population of nearly 1.9 million. Despite the substantial numbers of South Asians in the U.S., the only data regarding cancer survival for this group appear to be for breast cancer. Parikh-Patel et al.7 found that compared with non-Hispanic whites (hereafter referred to simply as ‘whites’), South Asians in California had lower survival from breast cancer, a higher prevalence of distant stage at diagnosis, and higher socioeconomic status (SES).

Because of the small number of studies examining prostate cancer survival in U.S. Asians and the absence of data for more recent immigrants, we conducted a study of survival from prostate cancer using data from all newly diagnosed cases of the disease occurring in California between 1995 and 2004. We compared survival for whites and the 6 largest Asian subgroups (Chinese, Filipino, Japanese, Korean, South Asian, and Vietnamese, hereafter referred to simply as ‘Asians’). Because of the large and diverse population of the state, data were available for 116,916 men with prostate cancer (108,076 whites and 8,840 Asians).

MATERIALS AND METHODS

Study Population

Prostate cancer cases were identified through the California Cancer Registry (CCR), a population-based registry that has collected cancer incidence data for the entire population of California since 1988 through a system of regional registries. Because California has a population of 33.9 million (as of the 2000 U.S. Census8), nearly 140,000 cases of invasive cancer are added to the registry each year; data regarding more than 2.2 million invasive cases have been collected since 1988. Since 2000, all CCR regions have participated in the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program, which requires the highest standards of data quality, as judged by completeness, accuracy, and timeliness.

CCR routinely classifies race into 29 mutually exclusive groups, using information in the medical record. There are 12 Asian subgroups, with 6 of these (Chinese, Filipino, Japanese, Korean, South Asian, and Vietnamese) constituting approximately 90% of all Asians in the registry database. Separately, Hispanic ethnicity is determined using race, Spanish origin, place of birth, and surname. The 6 largest Asian subgroups above were included in the current study, whereas the others were excluded because of inadequate numbers for analysis; persons of Hispanic ethnicity were also excluded.

We restricted eligibility to men diagnosed between 1995 and 2004 because of significant changes in the rules for registering new prostate cancer cases beginning January 1, 1995, and because 2004 was the latest year with complete case ascertainment. Cases were followed through the end of 2004, the latest year with complete cause of death ascertainment. Between January 1, 1995, and December 31, 2004, 143,752 white and Asian men were newly diagnosed with primary, histologically confirmed, invasive prostate cancer (International Classification of Diseases for Oncology, Second Edition [ICD-O-2] site code C61.9). We restricted study eligibility to men who were actively followed, with adenocarcinoma (ICD-O-2 histology codes 8010, 8140–8570), and for whom prostate cancer was the only known cancer (n = 117,560). After excluding 644 patients (0.6% of eligible patients) with missing cause of death, the final sample included 116,916 men.

Measures

In a previous study,9 block group-level data from the U.S. Census were used to create a composite area-based measure of SES. Using principal component analysis, each residential block group in California was assigned an SES score. These residential block groups were also ranked into quintiles based on their SES scores. Using residential address at diagnosis, cases accessed into CCR are routinely geocoded and assigned the SES score for their block group. Cases missing a block group because of incomplete address at the time of diagnosis are randomly allocated to block groups within their county of residence.

To determine stage at diagnosis, we used the SEER summary stage (or simply, summary stage), a combined clinical and operative/pathologic assessment that includes all information available within 4 months of diagnosis in the absence of disease progression, or through completion of the first course of surgery. For prostate cancer, summary stage is derived from 6 extent of disease fields: tumor size, clinical extension, pathologic extension (for patients undergoing prostatectomy), level of lymph node involvement, number of regional lymph nodes examined, and number of regional lymph nodes involved by tumor. During the study period, 2 schemes for summary staging were in effect: 1 for diagnosis years 1995–2000 and the other for diagnosis years 2001–2004. For prostate, there were only minor changes in the staging rules. For example, a case with tumor extension to the levator muscles would have been coded as distant stage during 1995 through 2000, but as regional stage during 2001 through 2004. The ICD-O grade levels used in the present study correspond to the following ranges in the Gleason score: well differentiated, 2 to 4; moderately differentiated, 5 to 7; and poorly differentiated/undifferentiated, 8 to 10. (Note that we did not have the raw Gleason score, only the ICD-O grade level.) Because men could receive more than 1 type of treatment for their prostate cancer, we determined each patient's primary treatment through a hierarchical scheme used by SEER investigators.1 Patients receiving surgery, whether alone or with radiation, hormone therapy, chemotherapy, or other treatment, were coded as surgery. In this scheme, ‘surgery’ denotes radical prostatectomy.

Date of death and cause of death were available through linkages to death certificate data. Using rules developed by SEER, survival time was calculated as the number of completed months between the date of diagnosis and whichever of the following was earliest: date of death; date last known to be alive; or December 31, 2004 (for patients with follow-up after this date). Patients' date of last known vital status is regularly updated through various follow-up activities, including contacting hospitals and physicians and linking patients' data to administrative databases. The survival endpoint for the current study was death from prostate cancer (International Classification of Diseases, Ninth Revision [ICD-9] code 185 or ICD-10 code C61).

Statistical Analysis

We used Student t-tests and chi-square tests to compare prognostic factors among white and Asian men. To confirm the proportional hazards assumption, we examined Kaplan-Meier plots of survival (S) versus time (T) and log(-log(S)) versus log(T) for whites and Asians. The Cox proportional hazards model was then used to assess the effect of race on survival, before and after adjustment for other prognostic factors. In these models we estimated the hazards ratio (HR) and 95% confidence interval (95% CI) for prostate cancer death. Using product terms, we tested for and did not find any statistically significant interactions between race and any of the following covariates: age, summary stage, primary treatment, histologic grade, SES, and year of diagnosis. Age and SES were modeled as quintiles rather than as continuous variables. Patients with missing stage and/or grade were not dropped from the analysis, but were assigned a value of ‘unknown’ for stage and/or grade. These analyses were conducted using PROC LIFETEST and PROC TPHREG in SAS 9.1 software for Windows (SAS Institute Inc, Cary, NC).

To account for racial differences in competing causes of death, we used the cause-specific survival methodology described by Marubini and Valsecchi.10 Cause-specific survival is a net survival measure representing survival of a specified cause of death in the absence of other causes. We specified prostate cancer as the cause of death; men who did not die or died from other causes were coded as censored observations.

RESULTS

The men contributed a total of 485,433 person-years of follow-up. During this time a total of 17,802 patients (15.2%) died; prostate cancer was listed as the underlying cause of death for 6157 patients (34.6%), and 11,645 patients (65.4%) were listed as deaths from other causes. Thus, the proportion of men who died from prostate cancer was 5.3%. Overall, the proportion of men with complete follow-up for vital status through 2004 was 99.1% (99.2% for whites and 97.8% for Asians).

Table 1 shows differences in prognostic factors among whites and the 6 Asian subgroups. Compared with whites, all Asian subgroups had unfavorable distributions of summary stage, histologic grade, and primary treatment. For example, in multivariate models, distant and unknown stage were associated with substantially worse prognosis, and every Asian subgroup had a higher proportion of men with distant or unknown stage compared with whites. The results for age were mixed, with Chinese, Filipino, Japanese, Korean, and Vietnamese men diagnosed at older ages, but South Asian men were diagnosed at younger ages than whites. Similarly, Chinese, Filipino, Japanese, Korean, and Vietnamese men had lower SES, whereas South Asian men had higher SES than whites. All Asian subgroups had a later year of diagnosis (ie, closer to 2004) than whites. Thus, Chinese, Filipino, Japanese, Korean, and Vietnamese men had unfavorable distributions for 5 of the 6 prognostic factors, and South Asian men had unfavorable distributions for 3 of the 6 prognostic factors.

Table 1. Differences in Prognostic Factors in White and Asian Men* With Prostate Cancer, California, 1995–2004 (n = 116,916)
CharacteristicRace (no. of patients)
White (108,076)Chinese (2637)Filipino (3235)Japanese (1383)Korean (380)South Asian (657)Vietnamese (548)
  • *

    Persons of Hispanic ethnicity were excluded from all these race groups. The South Asian group includes Asian Indians, Pakistanis, Bangladeshis, Bhutanese, Nepalese, Sikkimese, and Sri Lankans.

  • See text for details regarding summary stage, histologic grade, primary treatment, and socioeconomic status.

Mean age at diagnosis, y67.870.968.771.569.166.768.8
 (P for comparison with whites) (<.001)(<.001)(<.001)(.001)(.004)(.007)
Quintile of age (y), %
 1 (<60)20.29.416.78.211.323.012.8
 2 (60–65)20.114.818.013.521.819.223.9
 3 (66–70)20.221.921.121.026.623.722.8
 4 (71–75)18.425.420.620.620.018.419.0
 5 (≥76)21.228.623.723.720.315.721.5
 (P for comparison with whites) (<.001)(<.001)(<.001)(<.001)(.003)(<.001)
Summary stage, %
 Localized84.382.880.285.078.480.187.0
 Regional, extension only7.56.78.25.89.58.76.0
 Regional, lymph nodes only1.00.91.20.81.61.50.2
 Regional, extension and lymph nodes0.40.50.30.41.10.50.0
 Distant4.06.05.83.84.05.64.4
 Unknown2.93.14.44.35.53.72.4
 (P for comparison with whites) (<.001)(<.001)(.008)(.003)(.06)(.13)
Histologic grade, %
 Well differentiated5.74.66.84.64.24.16.4
 Moderately differentiated68.261.261.563.757.169.067.9
 Poorly differentiated22.930.927.628.634.523.722.6
 Undifferentiated0.30.50.30.10.30.20.6
 Unknown3.23.34.13.14.23.23.1
 (P for comparison with whites) (<.001)(<.001)(<.001)(<.001)(.45)(.84)
Primary treatment, %
 Surgery35.728.229.932.335.536.226.8
 Radiation therapy32.736.331.834.132.132.933.0
 Hormone therapy10.916.212.812.715.311.711.1
 Other treatments9.27.611.07.26.68.413.3
 No treatment listed11.511.714.513.810.510.815.7
 (P for comparison with whites) (<.001)(<.001)(<.001)(.05)(.87)(<.001)
Quintile of socioeconomic status scale, %
 1 (lowest)6.89.913.97.518.48.815.0
 214.614.119.814.318.411.922.3
 320.515.624.619.215.818.625.0
 424.524.324.230.418.220.718.1
 5 (highest)33.736.117.528.729.240.019.7
 (P for comparison with whites) (<.001)(<.001)(<.001)(<.001)(<.001)(<.001)
Year of diagnosis
 1995–199617.412.316.815.110.511.312.8
 1997–199819.015.216.419.118.219.211.0
 1999–200020.520.920.520.717.418.719.0
 2001–200221.725.823.023.127.623.928.3
 2003–200421.425.923.322.026.326.929.0
 (P for comparison with whites) (<.001)(<.001)(.23)(<.001)(<.001)(<.001)

Despite their poorer risk factor profiles, Figure 1 shows that in unadjusted descriptive analyses, all Asian subgroups except South Asian men had lower 10-year risk of death from prostate cancer than whites. Japanese men had the lowest risk, followed by Vietnamese, Chinese, Korean, and Filipino men. South Asian men were the only subgroup whose unadjusted risk was higher than whites. Overall, the 10-year risk of death from prostate cancer was 11.9%, but the variation in risk across the race groups was large, ranging from 8.1% in Japanese men to 16.4% in South Asian men. Table 2 shows that in unadjusted statistical comparisons, only Japanese men had significantly better survival than whites, whereas Chinese, Filipino, Korean, and Vietnamese men had statistically equal survival, and South Asian men had significantly poorer survival. Japanese men had a much lower risk of prostate cancer death than whites (44% lower), whereas the groups with statistically equal survival had, on average, only 10% lower risk than whites. Due to the small numbers of Asians relative to whites, we did not have adequate statistical power to detect these smaller differences at the α = 0.05 level.

Figure 1.

Unadjusted differences in risk of death from prostate cancer, white and Asian men, California, 1995–2004 (n = 116,916).

Table 2. Unadjusted Differences in Survival From Prostate Cancer, White and Asian Men, California, 1995–2004 (n = 116,916)
Race*No. of patientsNo. of deaths from prostate cancerUnadjusted HR for death from prostate cancer (95% CI)
  • HR indicates hazards ratio; 95% CI, 95% confidence interval.

  • *

    Persons of Hispanic ethnicity were excluded from all these race groups. The South Asian group includes Asian Indians, Pakistanis, Bangladeshis, Bhutanese, Nepalese, Sikkimese, and Sri Lankans.

  • Determined using the Cox regression model that included only race.

Chinese26371120.88 (0.73–1.06)
Filipino32351530.94 (0.80–1.10)
Japanese1383490.66 (0.50–0.88)
Korean380160.89 (0.54–1.45)
South Asian657431.40 (1.04–1.89)
Vietnamese548200.81 (0.52–1.25)
White108,07657641.00 (referent)

Table 3 shows the results for race and all other factors when present together in a multivariate model. As expected from their poorer risk factor profiles, every Asian subgroup demonstrated an improvement in their survival vis-a-vis white men after multivariate adjustment. Specifically, after adjustment for all prognostic factors, the protective HRs for Chinese, Filipino, Japanese, Korean, and Vietnamese men became more protective, and the adverse HR for South Asian men was markedly attenuated. On the multivariate analysis, Chinese, Filipino, Japanese, and Korean men all had significantly better survival compared with whites, whereas South Asian and Vietnamese men had statistically equal survival. The factors other than race are listed in the table in the order they entered the model in stepwise Cox regression, reflecting their importance as prognostic factors. The most important factor was summary stage, followed (in order) by histologic grade, primary treatment, age, SES, and year of diagnosis.

Table 3. Multivariate Model of Survival From Prostate Cancer, White and Asian Men, California, 1995–2004 (n = 116,916)
Factor*Multivariate HR for death from prostate cancer (95% CI)
  • HR indicates hazards ratio; 95% CI, 95% confidence interval.

  • *

    Factors other than race are listed inthe order in which they were entered into the model in stepwise Cox regression. See text for details regarding summary stage, histologic grade, primary treatment, and socioeconomic status.

  • Determined using the Cox proportional hazards model including all factors shown in the table.

  • Persons of Hispanic ethnicity were excluded from all these race groups. The South Asian group includes Asian Indians, Pakistanis, Bangladeshis, Bhutanese, Nepalese, Sikkimese, and SriLankans.

Race
 Chinese0.51 (0.43–0.62)
 Filipino0.59 (0.51–0.70)
 Japanese0.49 (0.37–0.65)
 Korean0.60 (0.37–0.98)
 South Asian1.10 (0.81–1.49)
 Vietnamese0.68 (0.44–1.05)
 White1.00 (referent)
Summary stage
 Localized1.00 (referent)
 Regional, extension only2.75 (2.49–3.04)
 Regional, lymph nodes only3.92 (3.33–4.61)
 Regional, extension and lymph nodes4.38 (3.34–5.75)
 Distant12.39 (11.58–13.25)
 Unknown2.92 (2.65–3.21)
Histologic grade
 Well differentiated1.00 (referent)
 Moderately differentiated1.89 (1.59–2.26)
 Poorly differentiated6.38 (5.35–7.63)
 Undifferentiated7.68 (5.93–9.95)
 Unknown5.72 (4.74–6.90)
Primary treatment
 Surgery0.25 (0.22–0.29)
 Radiation therapy0.76 (0.69–0.83)
 Hormone therapy1.36 (1.25–1.47)
 Other treatments1.18 (1.07–1.30)
 No treatment listed1.00 (referent)
Quintile of age (y)
 1 (<60)1.00 (referent)
 2 (60–65)0.93 (0.83–1.03)
 3 (66–70)1.05 (0.95–1.16)
 4 (71–75)1.11 (1.01–1.22)
 5 (≥76)1.68 (1.53–1.84)
Quintile of socioeconomic status
 1 (lowest)1.00 (referent)
 20.90 (0.82–0.99)
 30.86 (0.79–0.94)
 40.75 (0.68–0.82)
 5 (highest)0.66 (0.60–0.72)
Year of diagnosis
 1995–19961.00 (referent)
 1997–19980.96 (0.90–1.03)
 1999–20000.84 (0.78–0.91)
 2001–20020.81 (0.74–0.88)
 2003–20040.87 (0.75–1.00)

To place the results for Asian subgroups in a larger context, we also performed analyses comparing Hispanic and black men with non-Hispanic whites. The unadjusted HRs were 1.11 (95% CI, 1.05–1.19) and 1.35 (95% CI, 1.26–1.44) for Hispanic and black men, respectively.

Because the role of treatment in prostate cancer survival is controversial, we performed additional analyses adjusting for all prognostic factors in Table 3 except primary treatment. In these analyses, the multivariate HRs (Asian subgroups vs whites) for death from prostate cancer were: Chinese, 0.51 (95% CI, 0.42–0.62); Filipino, 0.58 (95% CI, 0.49–0.68); Japanese, 0.46 (95% CI, 0.35–0.61); Korean, 0.63 (95% CI, 0.39–1.03); South Asian, 1.19 (95% CI, 0.88–1.60); and Vietnamese, 0.71 (95% CI, 0.46–1.11).

We also performed analyses using a different approach to remove confounding by key prognostic factors. In these analyses we compared prostate cancer survival between non-South Asian men (Chinese, Filipino, Japanese, Korean, and Vietnamese) and whites within each stratum of histologic grade, summary stage, primary treatment, and SES. For all 4 of these factors the Asian men had better survival than whites within every stratum. For histologic grade, the stratum-specific HRs (Asian vs white) for death from prostate cancer were: well differentiated, 0.88; moderately differentiated, 0.83; poorly differentiated/undifferentiated, 0.71; and unknown, 0.64. For summary stage, the stratum-specific HRs were: localized, 0.66; regional, 0.76; distant, 0.64; and unknown, 0.69. For primary treatment, the stratum-specific HRs were: surgery, 0.94; radiation therapy, 0.81; hormone therapy, 0.71; other treatments, 0.73; and no treatment listed, 0.66. For quintiles (Q) of SES, the stratum-specific HRs were: Q1, 0.75; Q2, 0.73; Q3, 0.93; Q4, 0.86; and Q5, 0.76.

We performed other analyses to evaluate several potential biases. In each case, the results were essentially identical to those in Table 3. In the first, we excluded study subjects whose address could not be fully geocoded and so had block group randomly assigned from within their county. In the second, we excluded cases diagnosed during 2001–2004, when a new summary staging system was used. In the third, radical prostatectomy cases coded as regional disease were ‘downcoded’ to localized stage.

DISCUSSION

In this study of 116,916 white and Asian California men followed for up to 10 years, we found that all Asian subgroups except South Asian men had better unadjusted survival from prostate cancer than whites. Unadjusted risk of death from prostate cancer was lowest for Japanese men (34% lower than whites) and highest for South Asian men (40% higher than whites). The survival advantage for 5 of the 6 Asian subgroups was observed despite the fact that all Asian groups had risk factor profiles that put them at a survival disadvantage compared with whites. All 6 Asian subgroups had a greater frequency of distant stage, high-grade tumors, and a higher frequency of noncurative treatments. To our knowledge, these are the first published data regarding prostate survival for Korean, South Asian, and Vietnamese men living in the U.S., and are based on population-based statewide cancer registry data that include all newly diagnosed prostate cancer cases among men in California between 1995 and 2004. Patient inclusion for the study was not restricted by stage at diagnosis, age, region of California, enrollment in a specific health care system, or type of treatment. Completeness of follow-up was excellent, with approximately 99% of the men having follow-up for vital status through 2004. Because cause of death was available for all study participants who died, we were able to account for racial differences in competing causes of death in the analyses. Studies of prostate cancer patients have indicated excellent agreement between underlying cause of death listed on death certificates and clinician review of medical records.11

Our most notable finding is the more favorable survival among nearly all Asian subgroups, despite poorer prognostic profiles. Although the reasons for better unadjusted survival in 5 of the 6 Asian subgroups are not clear, they likely reflect unmeasured prognostic factors and may be the same reasons for the lower incidence of prostate cancer among Asians. Several hypotheses have been advanced for the lower Asian incidence rate, including racial differences in endogenous hormones, dietary factors, and body composition. Steroid hormones, particularly androgens, are suspected to play a major role in prostate carcinogenesis. Risk of prostate cancer appears to be strongly related to levels of bioavailable testosterone (ie, that converted to dihydrotestosterone [DHT] by the enzyme 5α-reductase). Ross et al.12 studied a healthy young population of U.S. whites and Japanese men and found evidence of lower testosterone metabolism in the Japanese men, most likely due to a lower activity of 5α-reductase. In a similar study of healthy young U.S. whites and Chinese men in Hong Kong, Lookingbill et al.13 also found evidence of lower 5α-reductase activity in the Chinese men. De Jong et al.14 directly measured circulating total testosterone levels in older Dutch whites and Japanese men, and found 71% higher levels in whites. Considerable consistency across studies indicates that a high intake of fat, particularly animal fat, is a risk factor for prostate cancer.15–24 Conversely, diets high in vegetables are associated with reduced risk of prostate cancer.25, 26 Phytoestrogens, such as those found in soy foods, have also been associated with reduced prostate cancer risk.27–32 Two recent studies have also suggested that obesity may increase the risk of high-grade or fatal prostate cancer.33, 34 Based on self-reported height and weight data from the 2005 California Health Interview Survey (CHIS), the prevalence of obesity (body mass index ≥30 kg/m2) among Asians in California was only 7.1%, compared with 19.6% among whites.35

The discordant survival results for South Asian men are consistent with the findings of Parikh-Patel et al.,7 who found that in California, South Asian women with breast cancer had more distant stage disease and lower unadjusted survival when compared with white women. The findings are also consistent with those of Jain et al.,36 who compared incidence rates of prostate cancer across racial groups in California and found the rate was substantially higher in South Asian men than in other Asian subgroups. The heterogeneity of South Asian relative to the other Asian subgroups may partially explain their disparate results. Linguistic and religious variation within the subgroup contribute to differences in lifestyle factors such as diet, smoking, and alcohol use that affect cancer incidence and survival.

Lack of data regarding Gleason scores and prostate-specific antigen (PSA) levels are notable limitations of the current study. Both are important prognostic factors and, in their absence, we used histologic grade as a measure of tumor aggressiveness. However, the ICD-O grade levels corresponded to different groups of Gleason scores than many clinicians would use. For example, although the ICD-O grades combine Gleason scores of 5 to 7 in a single category, in contemporary urologic practice cases with Gleason scores of 6 and 7 are treated quite differently because they differ entirely in aggressiveness. PSA level at diagnosis is an independent prognostic factor, and it may well be that the survival differences observed in the current study are due to racial/ethnic PSA differences. In addition, our data almost certainly suffer from some degree of race misclassification. It is possible that some South Asians may be misclassified as whites because of their surnames, but it is unclear whether racial misclassification is greater for South Asians compared with any of the other groups in the current study.

Later stage at diagnosis appeared to be the most important reason for the poorer survival from prostate cancer noted among South Asian men. Prostate tumors can be detected at early stages through screening, and data from the 2005 CHIS indicate there are large racial differences in the use of PSA testing in California. Only 27.4% of Asian men reported ever having a PSA test, versus 50.1% of white men.35 These differences could imply disparities in prostate cancer screening awareness, access to screening programs, health insurance status, and other factors. We used summary staging rather than other, more clinically oriented staging schemes such as that developed by the American Joint Committee on Cancer (AJCC). Whereas AJCC staging is useful for making treatment decisions, summary staging is adequate for population-based research. One study that assessed the accuracy of prostate cancer staging in a population-based tumor registry found that use of more accurate clinical stage data obtained through medical record review did not appear to have an appreciable impact on racial stage differences.37 South Asian men also had a higher prevalence of tumors with Gleason score ≥8, as well as greater use of hormone therapy, both of which are independent risk factors for poorer survival.

Thus, the findings for 5 of the 6 Asian subgroups were paradoxical, demonstrating better unadjusted survival than whites, despite risk factor profiles that put them at a survival disadvantage. The survival advantage for these groups may be due to factors we did not measure in our study, such as endogenous hormone levels, dietary factors, or body composition. Further research is needed to identify which, if any, of these factors may explain these paradoxical findings for Asian men. All these factors are particularly difficult to study because it is unclear, for example, which of the dozens of hormones involved in testosterone metabolism should be measured, and at what age. Understanding the potential interactions of each of these factors with age will be critically important for the development of effective prevention strategies. Only the findings for South Asian men were intuitive, demonstrating both a poorer risk factor profile and poorer survival compared with whites. The sharp differences in outcomes noted between South Asian men and the other Asian subgroups may be helpful in guiding future research. In summary, these results argue that traditional prognostic factors for survival from prostate cancer, stage, grade, treatment, age, year of diagnosis, and SES do not explain why the majority of Asian men have better survival compared with whites, but they do explain the poorer survival of South Asian men compared with whites.

Ancillary