Prostate cancer incidence, stage at diagnosis, treatment and survival in ethnic groups in South-East England

Abstract

Study Type – Prevalence (prospective cohort)
Level of Evidence 1b

OBJECTIVES

To use self-assigned ethnicity to examine patterns of incidence, stage, treatment and survival in patients with prostate cancer in South-east England.

PATIENTS AND METHODS

Data on 36 961 men resident in South-east England and diagnosed with prostate cancer between 1998 and 2003 were extracted from the Thames Cancer Registry. Ethnicity information was obtained from the Hospital Episode Statistics dataset, and matched to the cancer records. The ethnic groups examined were White (19 688), Black (1422) and Indian/Pakistani (397). Age-standardized incidence rate ratios were calculated overall and for narrower age groups, with White men as the baseline group. Logistic regression was used to assess whether patients had a stage of disease recorded at diagnosis, if so whether it was metastatic, and to examine treatment received. To assess overall and prostate cancer-specific survival (PCSS), Cox regression models were fitted, adjusting sequentially for age, socioeconomic status, treatment received and stage of disease at diagnosis.

RESULTS

Indian/Pakistani men had a lower age-standardized rate than White men (rate ratio 0.69, 95% confidence interval 0.63–0.75), while Black men had a higher rate ratio (2.51, 2.30–2.73). There was no difference in the proportion of men diagnosed with metastatic disease in each ethnic group. There was variation in recorded surgery and hormone treatment. Indian/Pakistani men had better PCSS than White men (fully adjusted hazard ratio 0.76, P = 0.024). There was no difference in PCSS between Black and White men (hazard ratio 0.93, P = 0.238).

CONCLUSIONS

Black men had the highest incidence of prostate cancer, followed by White, then Indian/Pakistani men. The relative excess of prostate cancer in Black vs White men was strongly age-dependent. Despite differences in recorded treatment, Indian/Pakistani men had better overall survival and PCSS. Black men also had better overall survival, and their PCSS was similar to that of White men. This might be due to access to the publicly funded National Health Service in the UK.

Abbreviations

ASRR

age-standardized incidence rate ratio

OS

overall survival

PCSS

prostate cancer-specific survival

TCR

Thames Cancer Registry

HES

Hospital Episode Statistics

ICD

International Classification of Diseases

SES

socioeconomic status.

INTRODUCTION

International differences in the incidence of prostate cancer are well established, with the highest rates found in the USA and the lowest rates in India and China [1]. Data from the USA [2,3] and the UK [4–7] have shown that Black men have higher incidence rates than White men within the same country. Population-based studies in the UK have shown that South Asians have a lower incidence of prostate cancer than non-South Asians, after adjusting for age [8,9] or age and socioeconomic status (SES) [10].

While to date there has been little research on patterns of prostate cancer stage and treatment in different ethnic groups in the UK, this has been more common in the USA. There have been conflicting results about whether there are differences in stage distributions between ethnic groups [11–13]. Black men have been shown to be less likely to receive surgery than White men [11,14–17], while there is less variation for other treatments [11,17]. Studies in the USA have found no difference in overall survival (OS) between ethnic groups, while Black men have worse prostate cancer-specific survival (PCSS) [18,19] and there was no difference between USA South Asian and White groups [13].

In the present study we aimed to use self-assigned ethnicity information for the first time in a large population in South-east England to examine the incidence, stage of disease at diagnosis, treatment received and survival in patients with prostate cancer.

PATIENTS AND METHODS

In the UK, cancer registries record the occurrence of cancer in their resident populations. During the period studied, the Thames Cancer Registry (TCR) covered a population of ≈14 million people in South-east England. This area includes London, which is the most ethnically diverse area of the UK [20]. Registration is initiated by clinical and pathological information received from hospitals, and by information about deaths provided by the NHS Central Register through the Office for National Statistics. Trained data-collection officers then extract further information on demographic details, disease stage and treatment in the first 6 months after diagnosis from the medical records. Data are quality assured, as they are added to the central database. Hospital Episode Statistics (HES) data come from hospital Patient Administration Systems and include patient, clinical, administrative and geographical details. Self-assigned ethnicity was introduced to the HES data in April 1995 using the codes from the England and Wales 1991 Census. In April 2001 the new 2001 Census ethnic codes were brought in, although the 1991 codes were still accepted until March 2003. Ethnicity information is more complete in HES than in the cancer registry dataset [21].

Data on all patients with prostate cancer (International Classification of Diseases, 10th revision, ICD-10, code C61) diagnosed between 1998 and 2003 were extracted from the TCR. For some patients information was only available from a death certificate, and these patients were excluded from analyses on stage, treatment and survival. The most recently recorded ethnicity information on patients admitted to NHS hospitals with cancer (or suspected cancer) who lived in the TCR area of South-east England was obtained from the HES dataset. These records were linked to the TCR records using first the NHS number and second the sex, date of birth and postcode. If no match to the HES data was found, the routinely recorded ethnicity in the TCR data was used, if available. Patients with no known ethnicity were included in the analysis as a separate ‘not known’ group.

Ethnicity codes from the 1991 and 2001 Census were both found in the HES dataset. To create groups sufficiently large to analyse stage, treatment and survival, three Black groups (Black Caribbean, Black African and Black Other) were merged and the Indian and Pakistani men were also combined (referred to as Indian/Pakistani). Age-standardized incidence rate ratios (ASRRs) were calculated for Indian/Pakistani and Black men, with White men as the baseline group, both overall and for narrower age groups (<60, 60–69, 70–79 and ≥80 years).

To assess variation in staging, two outcomes were explored: whether a stage at diagnosis was recorded, and if so whether the patient had metastases at the time of diagnosis. These outcomes were analysed using logistic regression, adjusting for age and SES.

The proportions of patients in each ethnic group receiving cancer surgery (defined as total removal of organ, or operation stated to be radical), hormone therapy, and radiotherapy were calculated. Logistic regression models were then fitted with adjustment for age, and separately with adjustment for age, SES and stage.

All patients who were not registered by death certificate only were followed up until 31 December 2006. Both OS and PCSS were examined. For the PCSS analyses, part 1 of the death certificate was examined and patients with prostate cancer (ICD-10 code C61, ICD-9 code 185) codes or text were regarded as dying from prostate cancer. Patients dying from all other causes of death were censored on the date of death. Cox regression was used, with all results adjusted for age. The effects of additionally adjusting for SES, stage and treatment received were also investigated.

RESULTS

There were 38 971 patients with prostate cancer diagnosed between 1998 and 2003 who were resident in the TCR area. Most patients were classified as White (20 465, 53%), 400 (1%) as Indian/Pakistani and 1435 (4%) as Black; 15 999 (41%) men had no known ethnicity. There were 2010 patients (5%) who were registered from death certificates only and were therefore excluded from stage, treatment and survival analyses. Of the 36 961 patients with complete registration records, ethnicity information was available for 22 157 (60%). The ethnic groups shown comprised the same proportions whether death certificate-only patients were included or excluded.

Indian/Pakistani men had a lower ASRR than White men (0.69, 95% CI 0.63–0.75), while Black men had a higher rate ratio (2.51, 2.30–2.73). ASRRs for specific age groups showed no variation for Indian/Pakistani men. However there was a strong gradient for Black men, whose ASRRs were more similar to White men as age increased (Fig. 1). In men aged <60 years the ASRR for Black vs White men was 3.77 (2.88–4.92), whereas in the oldest group the estimate was 1.71 (1.32–2.23).

Figure 1

ASRR with 95% CI, by ethnic group for specific age groups, compared with White men as baseline (ASRR = 1).

The adjusted proportions of patients having a stage at diagnosis recorded were 78% for White men, and 75% for both the Indian/Pakistani and Black men. Because the Indian/Pakistani group was small this difference was not statistically significant (P = 0.188), while the difference between White and Black men was significant (P = 0.018). Of those with a stage recorded, 22% of White men had metastases at the time of diagnosis, compared with 19% (P = 0.225) of Indian/Pakistani men and 23% (P = 0.192) of Black men. These proportions were adjusted for age and SES.

The proportions of each ethnic group with a record of receiving the different treatments, adjusted for age, SES and stage, are shown in Table 1. After adjusting for age, 10% of White men had cancer surgery, compared with 7% (P = 0.022) of Indian/Pakistani and 7% (P < 0.001) of Black men. These differences did not change materially and were still statistically significant after also adjusting for SES and stage.

Table 1. Adjusted proportions (%) patients with prostate cancer in South-east England diagnosed in 1998–2003 receiving cancer surgery, radiotherapy and hormone therapy treatments, and P values for the comparison with the White group

	%, P
	White*	Indian/Pakistani		Black
No. of patients	19 688	397		1422
Cancer surgery
Model 1	10	7	0.022	7	<0.001
Model 2	10	7	0.036	8	0.035
Hormone therapy
Model 1	44	36	0.004	35	<0.001
Model 2	44	37	0.012	38	<0.001
Radiotherapy
Model 1	15	15	0.841	12	0.004
Model 2	15	15	0.782	14	0.232

• * Baseline group. Model 1, adjusted for age; Model 2, adjusted for age, SES and stage.

Hormone therapy was the most common treatment; 44% of White men, 36% (P = 0.004) of Indian/Pakistani men and 35% (P < 0.001) of Black men were recorded as receiving hormone therapy, after adjusting for age. Additional adjustment for SES and stage slightly attenuated these differences, but they remained statistically significantly lower in the Indian/Pakistani (37%, P = 0.012) and Black (38%, P < 0.001) men compared with White men.

Radiotherapy was recorded in 15% of White and Indian/Pakistani (P = 0.841) men and 12% (P = 0.004) of Black men. After adjusting for SES and stage there was no longer any significant difference between receipt of radiotherapy in Black and White men (P = 0.232).

Survival analysis results, as estimated using deaths from any cause, for each ethnic group are shown in Table 2, adjusted for age, SES, stage of disease and treatment received. Adjusting for age, Indian/Pakistani men (hazard ratio 0.85, P = 0.055) and Black men (0.91, P = 0.048) had better OS than White men. After taking SES into account, this difference was strengthened, particularly for Black men (0.85, P = 0.001). Further adjustment for stage and treatment (cancer surgery, hormone therapy and radiotherapy) had little effect.

Table 2. Hazard ratios and 95% CI for OS and PCSS in patients diagnosed with prostate cancer in South-east England 1998–2003

Group	Hazard ratio (95% CI), P
	Model 1	Model 2	Model 3	Model 4
OS
White*	1.00	1.00	1.00	1.00
Indian/Pakistani	0.85 (0.72–1.00), 0.055	0.82 (0.70–0.97), 0.018	0.86 (0.73–1.01), 0.066	0.85 (0.72–1.00), 0.053
Black	0.91 (0.83–1.00), 0.048	0.85 (0.77–0.94), 0.001	0.83 (0.76–0.92), <0.001	0.82 (0.75–0.91), <0.001
SES quintile
1 (most affluent)*		1.00	1.00	1.00
2		1.05 (1.00–1.10), 0.066	1.02 (0.97–1.07), 0.378	1.03 (0.98–1.08), 0.314
3		1.08 (1.03–1.13), 0.003	1.07 (1.02–1.13), 0.005	1.07 (1.02–1.13), 0.005
4		1.16 (1.11–1.22), <0.001	1.11 (1.06–1.17), <0.001	1.10 (1.05–1.16), <0.001
5 (most deprived)		1.19 (1.13–1.25), <0.001	1.13 (1.08–1.19), <0.001	1.12 (1.06–1.18), <0.001
Treatment †
Cancer surgery				0.60 (0.56–0.66), <0.001
Radiotherapy				0.82 (0.78–0.86), <0.001
Hormone therapy				1.00 (0.96–1.03), 0.867
PCSS
White*	1.00	1.00	1.00	1.00
Indian/Pakistani	0.75 (0.59–0.95), 0.017	0.73 (0.58–0.92), 0.009	0.77 (0.61–0.97), 0.029	0.76 (0.60–0.97), 0.026
Black	1.01 (0.90–1.14), 0.839	0.97 (0.86–1.10), 0.674	0.93 (0.83–1.05), 0.250	0.93 (0.82–1.05), 0.242
SES quintile
1 (most affluent)*		1.00	1.00	1.00
2		1.07 (1.00–1.14), 0.062	1.02 (0.96–1.09), 0.496	1.03 (0.96–1.10), 0.431
3		1.08 (1.01–1.15), 0.028	1.06 (0.99–1.14), 0.077	1.07 (1.00–1.14), 0.068
4		1.17 (1.09–1.25), <0.001	1.07 (1.00–1.15), 0.039	1.07 (1.00–1.15), 0.046
5 (most deprived)		1.13 (1.05–1.21), 0.001	1.04 (0.97–1.11), 0.301	1.04 (0.97–1.11), 0.319
Treatment †
Cancer surgery				0.58 (0.51–0.65), <0.001
Radiotherapy				0.98 (0.92–1.04), 0.499
Hormone therapy				1.06 (1.01–1.11), 0.013

• * Baseline group;
• † †Each hazard ratio is for the contrast between a particular type of treatment and no record of this type of treatment. Model 1, adjusted for age; Model 2, adjusted for age and SES; Model 3, adjusted for age, SES and stage; Model 4, adjusted for age, SES, stage and treatment received (cancer surgery, radiotherapy and hormone therapy).

The PCSS results for each ethnic group are also shown in Table 2, adjusted for age, SES, stage of disease and treatment received. Black men had a different pattern of PCSS than for OS. There was no significant difference between Black and White men, after adjusting for age, SES, stage or treatment (hazard ratio 0.93, P = 0.242). Indian/Pakistani men had significantly better age-adjusted PCSS (0.75, P = 0.017). Adjustment for SES, stage and treatment did not affect this association (0.76, P = 0.026).

DISCUSSION

This study using cancer registration data and self-assigned ethnicity information for patients resident in South-east England found that Black men had the highest incidence of prostate cancer, followed by White, then Indian/Pakistani men. The relative excess of prostate cancer in Black compared to White men was strongly age-dependent, with younger Black men have higher incidence rates.

In the UK, the higher incidence of prostate cancer in Black men [7] and the lower incidence in South Asian men [22] compared to White men have recently been shown in other studies. Our analysis extends this observation with the examination of age-specific rates.

Studies of migrants have shown that rates of cancer incidence in migrant populations often converge towards the rate in the local population [9,23,24]. This might happen eventually in the Indian/Pakistani group in England, but the consistent and persistently high rate of prostate cancer in Black men in the USA and the high excess in young Black men shown in our study suggest that a high rate of prostate cancer is a characteristic of Black men which persists for generations after migration.

There was no variation in the proportion of men from different ethnic groups being diagnosed with metastatic disease. PSA information is not collected in the registry data, while analysis on the Gleason score data showed there was no variation between the ethnic groups (data not shown). In the USA, South Asian men have been shown to have a more advanced stage distribution than in with White men [13]. Mettlin et al.[11] found that Black men were more likely to have metastatic disease than White men, while Tewari et al.[12] found no significant difference in the stage distribution. The high proportion of patients with metastatic disease in South-east England compared with the USA is likely to be due to the lower levels of PSA testing in the UK. This will mean treatment patterns are different to those in the USA, although there should be no difference between ethnic groups in their treatment.

Black men were significantly less likely to receive radiotherapy than White men when adjusted for age, but this difference was attenuated when also adjusting for SES and stage. White men were significantly more likely to have cancer surgery and hormone therapy recorded than either Black or Indian/Pakistani men. Radical surgery is only used in men with localized prostate cancer. It is therefore used less than other treatments and the absolute differences seen in the proportions of men with surgery recorded were small (2–3%).

Several studies in the USA have shown that Black men are less likely to receive prostatectomy after adjusting for variables describing the tumour, patient and area of residence [12,15–17]. In another USA study, Desch et al.[18] compared patients receiving surgery with those receiving radiotherapy, and found that Black men were less likely to receive surgery, after adjusting for patient, area and health service characteristics. However, there was no difference in the likelihood of Black and White men receiving hormone therapy or orchidectomy instead of surgery and/or radiotherapy. Little difference was found in the proportions of Black and White patients receiving radiotherapy (28.5% and 28.8%, respectively) in a study that adjusted for stage [12].

The variation in recorded treatment in South-east England is not fully understood. The differences were attenuated in analyses that adjusted for area of residence (data not shown), suggesting either different treatment practices, or possibly ascertainment of treatment information varies geographically within the TCR area. The treatment decision of active surveillance is not recorded in the TCR data, and patients receiving no treatment could therefore not be distinguished between those with and without this explicit decision.

Black and Indian/Pakistani men had better OS than White men, whether adjusted for just age or also for SES, stage and treatment received. Indian/Pakistani men consistently had better PCSS than White men, regardless of adjustment for age, SES, stage or treatment, while there was no difference between Black and White men.

This analysis suggests that Black men with prostate cancer in South-east England have a case-fatality that is similar to, and not higher than, White men. This result is consistent with the recent large study from California [19], but different from most survival studies from the USA, which overall suggest that Black men had worse PCSS than White men [18].

By contrast with the study by Robbins et al.[13] in California, which found South Asians had statistically non-significant worse PCSS than White men, in the present study Indian/Pakistani men had better PCSS. This difference was insensitive to statistical adjustment for patient and disease characteristics. This might suggest an inherently better prognosis for prostate cancer in South Asian men, but the origin of such a difference is not known.

There were some limitations to the present study. There was a large proportion of patients whose ethnicity was not known. If these patients were not proportionally distributed between the ethnic groups this could bias the results. Due to the underlying population structure, most of the patients with an unknown ethnicity are likely to be White. Therefore re-coding these patients to White would lead to an attenuation of any results, as there will be some patients from other ethnic groups who are then misclassified as White.

Patients with no ethnicity information are a combination of those who had no link with the HES data or no ethnicity was recorded in HES, and no ethnicity information in the TCR record. Completeness of ethnicity is improving in both data sources; in the TCR database 7% of patients diagnosed in 1995 had a valid ethnic code compared with 39% in 2006.

Although the patients in the dataset were classified by the ethnic groups used in the Census, to have large enough numbers to analyse, several of the groups had to be combined. This could be masking important differences between ethnic groups, as well as possible differences within the ethnic groups used in the Census.

Studies examining more detailed biological information, including comorbidity and PSA levels, will be extremely useful in determining whether different ethnic groups are being diagnosed with different forms of prostate cancer. The continuing improvement in ethnicity data collection will enable future research to repeat these analyses without the limitation of a large proportion of patients whose ethnicity is unknown. International comparisons set up to examine and measure data in the same way would allow the apparent differences between countries to be examined more closely, and the reasons for any disparities determined. The apparent treatment differences seen need to be further explored, paying close attention to geographical factors and ascertainment.

In conclusion, this study from South-east England found that Black men had the highest incidence of prostate cancer, followed by White, then Indian/Pakistani men. The relative excess of prostate cancer in Black compared to White men was strongly age-dependent, and highest in the youngest groups. Despite differences in recorded treatment, Indian/Pakistani men had better OS and PCSS; Black men also had better OS, and their PCSS was similar to that of White men. This might be due to access to the publicly funded NHS in the UK.

CONFLICT OF INTEREST

None declared.