Risk and presenting features of prostate cancer amongst African-Caribbean, South Asian and European men in North-east London

Authors


Frank Chinegwundoh, Department of Urology, Barts and the London NHS Trust, West Smithfield, London, EC1A 7BE, UK.
e-mail: Frank.Chinegwundoh@nhs.net

Abstract

OBJECTIVE

To determine whether there are ethnic differences in the incidence and presenting features of all patients with prostate cancer presenting in North-east London, UK.

PATIENTS AND METHODS

All newly diagnosed men with prostate cancer between 1999 and 2000 who were resident in the East London and City Health Authority were identified from various sources. Key clinical features were extracted from hospital records. The age-adjusted incidence rates for European, South Asian and African-Caribbean patients were calculated using census denominator data.

RESULTS

For men aged >50 years the annual age-adjusted incidence rates (95% confidence interval) were 213 (186–240), 647 (504–789) and 199 (85–310) for the European, African-Caribbean and South Asian patients, respectively. African-Caribbean men had a three times greater risk (risk ratio 3.07, 2.40–3.93, P < 0.001) than European men. South Asian men had a lower risk of prostate cancer but this could have been compatible with chance. There was no evidence of marked ethnic differences for prostate-specific antigen levels, clinical staging and Gleason scores.

CONCLUSION

The greater risk of prostate cancer for African-Caribbean men in South-east England is consistent with data from the USA and the Caribbean. Future work needs to determine whether this risk differs according to country of origin, and which genetic and/or environmental risk factors might be important in explaining these observations.

Abbreviations
ELCHA

East London and City Health Authority.

INTRODUCTION

There is a wealth of information from the USA showing that African-American men have a greater risk of developing prostate cancer than White men. The degree of relative increase varies from as little as a 20% relative increase, unadjusted for socio-economic position [1], to up to 66%[2]. Black men in the USA also present when younger, with higher PSA levels and more advanced clinical stage [3–5]. Similarly, very high prostate cancer incidence rates were reported from the Caribbean [6]. It is unclear why there are such differences, although environmental factors, e.g. differential access to healthcare, dietary and genetic factors [7–13], have all been suggested as possible explanations. By contrast, prostate cancer rates for men from the Indian subcontinent are much lower (30% relative reduction) than in England and Wales [14], although this difference might be artefactual due to lower detection rates in the Indian population.

There are few data on prostate cancer in ethnic groups within the UK; it seems reasonable to assume that the greater relative risk for prostate cancer amongst African-American and Caribbean men will also be present amongst African-Caribbean men in the UK, but good empirical data are required to confirm this assumption. One study reported greater mortality among immigrants born in Africa and the Caribbean [15], but mortality rates are influenced both by incidence and case fatality, so it is unclear whether this reflects true differences in incidence and/or worse case fatality rates. The Longitudinal Study [16] found a standardized incidence ratio of 2.2 (95% CI 1.0–4.1) amongst 1082 Black immigrants in England and Wales followed up for 18 years, but this was based on only nine cases and the wide 95% CIs cannot exclude that rates are almost identical to those in White men. South Asians had a slightly lower risk (standardized incidence ratio 0.7, 95% CI 0.2–1.8) based on four cases. No study, to the best of our knowledge, has described the presenting features of prostate cancer in African-Caribbean and South Asian communities in the UK.

The first Black communities in East London were established in the mid-18th century, but the greatest population expansion in ethnic groups was after immigration from the Commonwealth in the 1950s [17]. Greater London contains 80% of the total UK Black African, 60% of the Black Caribbean, and 30% of the South Asian population [18]. East London and City Health Authority (ELCHA), which incorporates the East End of London, therefore includes an ideal population for investigating variations in prostate cancer risk by ethnicity. The aim of the present study was to compare incidence rates and presenting features of prostate cancer in African-Caribbean and South Asian communities of North-east London, compared with the European population.

PATIENTS AND METHODS

We undertook a population-based study by auditing all men diagnosed with prostate cancer between 1999 and 2000 and living in ELCHA, a region that includes the London Boroughs of Tower Hamlets, Hackney, Newham and the City of London, and has a population of just over 600 000.

Cases were identified using the following sources of information; (a) a centralized histology database, which has pathology from the three health trusts (Barts and the London, Newham and Homerton) that serve this catchment population; (b) urology and oncology department clinical databases; and (c) the Thames Cancer Registry. Only incident patients with newly diagnosed histologically confirmed adenocarcinoma of the prostate between 1 January 1999 until 31 December 2000 were included. Potential duplicate cases were excluded by checking against patient name, date of birth and hospital number. Eligible cases had to be resident within the ELCHA to be included, based on their address postcode. Clinical information on presenting features was extracted by relevant medical staff from medical records, and included age, PSA level, TNM stage and presence or absence of metastases on bone scan at the time of diagnosis. Denominator data for ELCHA were obtained from the Greater London Authority mid-year estimates of the male population by age and ethnic group for 1999.

We defined three ethnic groups using the 2001 census ethnic group categories: White (including British, Irish, ‘other White’); Asian (Indian, Pakistani, Bangladeshi, other Asian, White and Asian mixed); and Black (Black Caribbean, Black African, Other Black, White and Black African or Caribbean mixed). Most of the ethnicity coding was taken from the clinical databases based on two of the authors (A.L. and F.C.) who routinely ask patients their self-reported ethnic origin. Additional sources of information came from the patient administration system, hospital records, and occasionally from patients’ GPs.

We calculated crude and age-adjusted incidence rates using direct standardization with the European Standard population (Nunnally 1996). Risk ratios and 95% CIs were calculated for African-Caribbean and South Asian men, relative to European men, by Poisson regression modelling adjusting for age in 5-year age bands (50–54, 55–59 . . . ≥ 85). We repeated the analysis for men aged 50–69 years, as this reduces the risk of under-ascertaining older men, who are more likely to be under the care of geriatricians, or not have a histological diagnosis, and hence might have been missed by the present methods of case identification. Differences in clinical features were tested using the chi-square statistic and logistic regression modelling for dichotomous outcomes, or linear regression modelling for continuous variables. As PSA level was highly skewed to the right, it was log-transformed and geometric means are presented. PSA levels were also grouped into an ordinal variable as 0–9.9, 10–19.9, 20–99.9, and >100 ng/mL. Gleason score was categorized into four groups as 2–4, 5–6, 7 and 8–10.

RESULTS

We identified 359 newly diagnosed patients in a 2-year period, i.e. 248 European, 91 African-Caribbean, 20 South Asian men. The mean age of the men was 74.1 years (median 74) but there were marked differences across the ethnic groups, i.e. Europeans 75.6, African-Caribbean 70.4 (difference 5.2, 95% CI 3.3–7.1, P < 0.001), South Asian 71.7 (difference 4.0, 95% CI 0.3–7.7, P = 0.03). There was no a priori sample size calculation, as the study was based on a pragmatic review of current patients. The number of cases and age-specific incidence rates are shown in Table 1. For every age group, the age-specific incidence rates were higher amongst African-Caribbean than among European men. For South Asians, rates were generally lower, but were statistically unstable as there were too few events. Adjusting for differences in the age structures of the populations, as African-Caribbean and South Asian populations were younger, increased the differences in crude incidence rates (Table 1), as the rate for European men was slightly diminished, whilst those for African-Caribbean and South Asian men were increased. This is quantified in Table 2, which presents the risk ratios and 95% CI for both crude and age-adjusted rates. The doubled risk for African-Caribbean men was increased to more than a threefold risk, whilst the apparent halving of risk for South Asians was less marked, with a 30% reduction, after accounting for their younger population. When this analysis was restricted to cases age <70 years, the greater risk for African-Caribbean men was further increased, whilst the risk ratio for South Asian was more markedly decreased.

Table 1.  Crude, age-adjusted and age-specific incidence rates per 100 000 for European, African Caribbean and South Asian men
Age group, yearsEuropeanAfrican CaribbeanSouth Asian
PopulationCancer casesIncidence rate/100 000PopulationCancer casesIncidence rate/100 000PopulationCancer casesIncidence rate/100 000
50–54 9 778  1  51365 2  731526 0   0
55–59 7 352  4 271623 8 2461788 3  84
60–64 6 552 14107180414 3882076 4  96
65–69 6 464 35271 129116 6201477 1  34
70–74 5 960 53445 856211227 829 5 302
75–79 4 798 66688 420182143 333 3 450
80–84 2 502 43859 170 92647  116 1 431
≥85 2 026 32790  87 31724  62 32419
Total45 432248 761691 820720 
Crude rate  273   597   122
Adjusted rate  213   647   199
95% CI  186–240   504–789    89–310
Table 2.  Crude and age-adjusted risk ratios by ethnicity
VariableEuropeanAfrican-CaribbeanSouth Asian
RR (95% CI)PRR (95% CI)P
Crude1.02.19 (1.72–2.78)<0.0010.45 (0.28–0.70)0.001
Age-adjusted1.03.07 (2.40–3.93)<0.0010.70 (0.44–1.11)0.13
Age-adjusted1.03.34 (2.21–5.03)<0.0010.56 (0.28–1.23)0.16
for men <70 years

Table 3 shows the PSA levels, clinical stage and Gleason scores at presentation. The crude proportions and geometric mean for PSA level do not appear to be markedly different by any ethnic group. A linear regression model of log PSA, adjusting for age group, stage and Gleason score, showed only a modest increase in PSA levels for Black men (22% relative increase), which could have occurred by chance (P = 0.17). Similarly, differences for clinical staging were more marked after age adjustment (odds ratio for advanced stage 1.55, 95% CI 0.86–2.80, P = 0.14), but were compatible with chance. There were no differences for Gleason score. South Asians had a lower risk of advanced stage or higher Gleason score, but the few patients meant that we could not exclude chance variations.

Table 3.  Clinical features at presentation by ethnicity
VariableN (%)
EuropeanAfrican-CaribbeanSouth Asian
  • *

    Geometric mean.

PSA group, ng/mL
 0–9.9 42 (16.9)19 (20.9) 3 (15)
 10–19.9 56 (22.6)16 (17.6) 6 (30)
 20–99.9 97 (39.1)36 (39.6) 8 (40)
 ≥100 53 (21.4)20 (22.0) 3 (15)
P  0.17 0.95
Mean PSA level* 37.642.432.8
P  0.54 0.72
Stage
 T1-T3, M0 191 (77.0)67 (73.6)17 (85)
 T4 and/or M1 and/or N1 57 (23.0)24 (26.4) 3 (15)
P  0.12 0.5
Gleason score
 2–4 39 (15.7)17 (18.7) 6 (30)
 5–6 84 (33.9)32 (35.2) 7 (35)
 7 48 (19.4)15 (16.5) 2 (10)
 8–10 59 (23.8)23 (25.3) 5 (25)
 Missing 18 (7.3) 4 (4.4) 0
P  0.87 0.4

DISCUSSION

This study confirms the higher rates of prostate cancer amongst African-Caribbean men and is the first to examine whether there are any ethnic differences in clinical presentation. The higher incidence of prostate cancer in this Black population than in Europeans mirrors the ethnic differences found in the USA. However, the threefold relative risk identified in the present study is much higher than that found in the USA [2,19], although a direct comparison is complicated, as a much larger proportion of USA cases, both amongst Black and White men, are screen-detected, resulting in over-diagnosis [20], unlike the situation in the UK.

Another possible explanation is that most Black men in the present population, compared with the USA, are first-generation migrants, who were born in the Caribbean and migrated to the UK in the 1950s. The prostate cancer incidence in first-generation migrants, e.g. Asian Americans born outside the USA, tends to be similar to that of their birthplace [21]. Jamaica has one of the highest prostate cancer incidence rates in the world [6], and therefore the first-generation migrants from the Caribbean to the UK might be expected to have a high incidence of prostate cancer.

A life-course approach [22] to aetiological risk factors might be helpful in speculating about these ethnic differences. Conventional epidemiology has tended to focus on adult lifestyle risk factors, such as diet. For example, the high saturated-fat diet of African-American men has been suggested to explain their greater risk of prostate cancer [7,8] but this has not been detected in all studies [23]. This risk factor is unlikely to contribute to the ethnic differences identified in the present study, as first-generation African-Caribbean men in England consume and purchase less fat than Caucasians [24,25]. Risk factors in young adulthood, e.g. exposure to sexually transmitted infections, might also increase prostate cancer [26], and there is a higher incidence of infections such as gonorrhoea in Black men in London [27]. Finally, it is important to consider early-life risk factors, e.g. growth and development during puberty, which would have been influenced by the host environment before migration, as a reason why British Black men have greater relative risks than those reported in the USA. Such environmental risk factors might interact with a potentially greater susceptibility due to genetic variation.

Both the African-Caribbean and South Asian men with prostate cancer presented when younger than the Europeans. One possible explanation for this discrepancy would be if men in an ethnic minority had a greater probability of being screen-detected and hence be younger at diagnosis. This is unlikely, as screening is still relatively rare in the UK and is more likely to occur amongst affluent men. In addition, we found no difference in the proportion of men with localized T1 disease by ethnicity. A more plausible explanation is that this difference is a demographic artefact, as these populations are much younger, with fewer men aged >75 years, than their European counterparts. Therefore the mean age at presentation, even if the age-specific incidence rates were identical, would have to be lower for the ethnic minority groups.

We found no evidence of any marked differences in clinical presentation, unlike in data from the USA. Some have cited poor access to healthcare in the USA as a possible reason why Black men present with more advanced disease. Although the Greater London population of Black men has three times the unemployment rate, and a 30% lower income, than the White population [28], the UK NHS provides access to free healthcare, and this might explain why we found no differences.

South Asian men had the lowest age-adjusted incidence rate of prostate cancer, similar to those found in India [14], although given the statistical imprecision this observation must be interpreted cautiously. There were too few cases to identify any potential differences in presenting features. These differences are also of potential interest in terms of both genetic and/or environmental clues to potential aetiological risk factors.

There are several limitations to the present study: these include; (i) the possible inclusion of patients who were not permanent residents; (ii) exclusion of private patients; (iii) incomplete data on pathological staging. The Gleason score errors were minimized by processing all histology in a central laboratory, but the reports were produced by more than one pathologist and hence are not standardized. Population numbers were based on mid-year estimates, and therefore there might be errors due to under-ascertainment of the population. This would be a particular concern if it was worse for Black than White men, and would result in artefactually higher rates for Black men. However, this problem would be less marked for older men, as they are less likely to be unregistered, and is unlikely to explain such large differences in the observed rates. There was no regular prostate cancer screening in the catchment area, although it is possible that some cases were screen-detected. However, we found no difference in T stage at presentation between the ethnic groups, which might have occurred if one group had more screen-detected cancers.

It is important that our results are confirmed and that more precise estimates for different subgroups are produced. A multicentre retrospective cohort study (PROCESS), funded by the UK Department of Health, is currently investigating ethnic differences in prostate cancer risk in Greater London and Bristol. This study will be able to compare risks for Black Caribbean and Black African men, as well as examine in far greater detail any differences in clinical presentation, disease management, natural history and survival patterns.

Our preliminary findings will be of particular interest to urologists, GPs and public health practitioners working in areas with large ethnic minority communities, and to Black men themselves. Raising awareness amongst the medical profession of the greater risk of prostate cancer in younger African-Caribbean men is important, and might lower the threshold for specialist referral and increase the positive predictive value of diagnostic tests in this population of men [29].

CONFLICT OF INTEREST

None declared.

ACKNOWLEDGEMENTS

The authors acknowledge the Urology Consultants at Barts and The London NHS Trust, Newham University Hospital NHS Trust, and Homerton University NHS Trust for their inclusion of patients in the study. We also thank the R&D at Newham and the audit departments at Newham and Barts for their help and support and the Thames Cancer registry.

Ancillary