Incidence of prostate cancer in Sri Lanka using cancer registry data and comparisons with the incidence in South Asian men in England

Authors


Weranja K. B. Ranasinghe, Department of Urology, Austin Hospital, Melbourne, Australia. e-mail: weranja@gmail.com

Abstract

Study Type – Prevalence (retrospective cohort)

Level of Evidence 2b

What’s known on the subject? and What does the study add?

The incidence of prostate cancer in South Asia is low but the incidence of prostate cancer is Sri Lanka is unknown. This study examines the latter and compares these rates to migrant South Asian population to the UK and attempts to examine reasons for the differences.

OBJECTIVE

• To investigate the incidence of carcinoma of the prostate (CaP) in Sri Lanka and compare the rates with the migrant population in the UK.

METHODS

• The Sri Lanka cancer registry data were used to determine the rates of CaP in Sri Lanka from 2001 to 2005.

• The incidence of CaP in 8 426 000 men, aged ≥30 years, was analysed using 5-year age bands and age-standardized rates were calculated using European standard population data.

• The relative risk was calculated by comparison with the South Asian migrant population in the UK using the Prostate Cancer in Ethnic Subgroups (PROCESS) study data, a population-based retrospective cohort study of 2140 men carried out over a 5-year period in four predefined areas of southern England.

• Data from incidental findings of CaP in Sri Lanka on transurethral resection of the prostate (TURP) specimens were also analysed.

RESULTS

• In all, 1378 new cases of CaP were diagnosed during the 5-year period with a mean age of 69.2 years at diagnosis.

• Compared with the previous 5 years, the incidence of CaP had doubled to 5.7 per 100 000, but was significantly lower than in the PROCESS study (relative risk 0.25).

• Districts with a higher population density had higher adjusted rates of CaP in Sri Lanka (5.8–12.4 per 100 000).

• For TURP specimens, 16.8–18.75% had incidental diagnoses of CaP in Sri Lanka, higher than other published studies.

CONCLUSIONS

• The Sri Lankan cancer registry data showed a low rate of CaP, similar to other South Asian countries, but the actual incidence of CaP in Sri Lanka is probably higher than reported, as seen in the densely populated districts and the high rate of incidental diagnosis of CaP in TURP specimens.

• The incidence of CaP in migrant South Asians in the UK was much higher than in Sri Lanka.

• Further studies are required to examine the environmental and genetic components which may be responsible for the low incidence of CaP in Sri Lanka.

Abbreviations
CaP

carcinoma of the prostate

PROCESS

Prostate Cancer in Ethnic Subgroups

NCCP

National Cancer Control Programme

NHSL

National Hospital of Sri Lanka

MI ratio

mortality : incidence ratio

INTRODUCTION

Carcinoma of the prostate (CaP) is the fifth most common cancer in the world (the second most common cancer in men) accounting for 679 000 new cases annually and responsible for 5.8% of cancer-related deaths in men in 2002 [1]. The low mortality rate resulted in CaP being the most prevalent cancer in men in the year 2000 with an estimated 1.5 million men living with the diagnosis [2]. About three-quarters of this occurred in men aged ≥65 years [1].

There is international variation in the incidence of CaP with the USA, northern and western Europe and Australia/New Zealand all having higher numbers [1]. The highest age-adjusted incidence of CaP is seen in the USA (124.8 per 100 000) [1], whereas the lowest is seen in the East Asian region, namely China [1] (2.3 per 100 000 person-years), Hong Kong (7.9 per 100 000), Japan (9.0 per 100 000) and Singapore (9.8 per 100 000) [3].

The South Asian countries are shown to have a low incidence of CaP [1]. India is amongst the countries with a low incidence of CaP (4.55–7.9 per 100 000) [3,4], approximately 18-fold less than that of the USA [5]. The incidence of CaP in Pakistan is shown to be as low as 5.58 [4].

Interestingly the data on migrant populations also vary significantly from their original country’s data [1,6,7]. In the USA the incidence of CaP is high for migrant populations of Southeast Asian origin, where even in the Chinese population the incidence reaches 80.4 per 100 000 [7]. The South Asian population living in the USA (Indian, Pakistani, Bangladeshi and Sri Lankan) has 110.7 per 100 000 incidence of CaP, a 15-fold increase compared with Indian males living in India [6].

The incidence of CaP in South Asian men (Indian, Pakistani, Bangladeshi) living in southern England was 49.55 per 100 000, much higher than their counterparts living in the Asian subcontinent, as shown by the Prostate Cancer in Ethnic Subgroups (PROCESS) study, a population-based retrospective cohort study [8]. This has been attributed to a variety of diet and environmental factors as well as early detection, reporting bias, lead time and case identification [6].

To our knowledge there have been no previous published studies on the incidence of CaP in Sri Lanka. Therefore, we investigated the incidence of CaP in Sri Lanka and compared this with the migrant South Asian population in the PROCESS cohort study to look for any significant difference [8].

METHODS

Sri Lanka, although a small island in the Indian subcontinent, has a population of 19 886 000 with a high population density in its main cities of Colombo, Gampaha, Kalutara and Kandy [9]. Sri Lanka has a free public healthcare system, with the government spending 7.8% of its annual budget on health in 2005 [10]. Parallel to this, there is also a private healthcare system, but this services only a small number of the population in the bigger cities. The WHO statistics show that Sri Lanka had 10 479 physicians in 2004, accounting for a physician density of six physicians and 29 hospital beds per 10 000 population [10]. This healthcare system is quite comparable to the health systems of India (eight physicians per 10 000) and Pakistan (six physicians per 10 000) [10].

The National Cancer Control Programme (NCCP) in Sri Lanka is the central collection and storage centre of cancer data. Patients with new diagnoses of CaP in the outpatient and inpatient setting in peripheral hospitals are diagnosed mainly on clinical findings and the majority proceed to have a histological diagnosis such as transrectal prostate biopsies. (PSA testing is not done routinely in Sri Lanka to aid diagnosis.) The patient data initially collected and stored in an allocated central hospital in each region are then verified and analysed by the NCCP by follow-up for at least 2–3 years. During this process, if any dubious data are noted, the data are re-examined, followed up and re-validated by the NCCP to ensure reliability. The national cancer registry is built up from these data and published every 5 years in a bulletin. The CaP incidence data (ICD-10 C61) supplied for our use are twice verified by the cancer registry and immediately prior to official publication. Approximately 80% of the diagnoses of CaP (2001–2005) are based on histological diagnosis (internal communications). Population denominator data, for the whole country and for specific cities, came from the national census data (2001) [9,11].

As Sri Lanka during this period had an ongoing war in the northern and eastern areas of the island, the data collected from these regions would have been suboptimal. Therefore we decided to analyse the national rate but put more emphasis on data from only the main cities, where patients would have better access to healthcare. Consultation in the private sector is more frequent in these larger cities, but a large group of the private sector patients are also followed up in the public sector because of the availability of good cancer treatment centres. Hence these patients are likely to be included in the data provided to us.

We initially calculated age-specific incidence across a 5-year period 2001–2005 in 5-year age groups from 30 years to ≥75 years. We then used direct standardization to calculate an age-standardized rate using the European hypothetical population [11]. To help interpret these results we used the same methods to calculate an age-standardized rate for incidence data from South Asian populations based in England and Mumbai, India. We calculated the age-standardized relative risk rate for the Sri Lankan rate compared with the English and Indian rates.

To ensure reliability of the Sri Lankan cancer registry data we also included data from the records of the histopathological laboratories of the Faculty of Medicine and the National Hospital of Sri Lanka (NHSL) on the rates of incidental diagnosis of CaP in TURP specimens within a 2-year period (2007 and 2008) and compared these with the rates from the cancer registry. The histopathological laboratories of the Faculty of Medicine and NHSL perform routine analyses, reported by consultant pathologists, of TURP specimens from the two urology wards at NHSL consisting of patients mainly from the western province.

RESULTS

Across the 5-year period there were 1378 cases (between 250 and 300 per year) and the mean age of the cases was 69.2 years (inter-quartile range 64–75 years). District of residence was available for 91.4% of cases. Not surprisingly the more densely populated districts of Colombo (27.9%), Gampaha (11.9%) and Kandy (8.9%) had the most cases although Kurunegala has a greater population than Kandy. Only 32 cases of CaP (2.3%) were ascertained from seven districts from the north and northeast coastal areas (Jaffna, Mannar, Vavuniya, Mullaitivu, Killinochchi, Batticaloa, Trincomalee) where there are no accurate population census data due to unstable security conditions, so we decided to drop these observations and base all further analyses on the remaining 18 districts with a population of 8 426 000 men.

Figure 1 shows the age-specific rates for men aged ≥30 years stratified by the three districts with the largest number of cases and the remaining 15 districts of Sri Lanka combined. Colombo, with the best healthcare system, not surprisingly has the highest rates, which rise exponentially with age, followed by Kandy, Gampaha and then the other districts. The age-specific rates for these areas plus the combined Sri Lankan rates for all 18 districts are shown in Table 1. The age-specific rates for Colombo were between 61% and 87% lower than the rates reported for South Asians residing in southern England. After age standardization to the world hypothetical population, the incidence rate was 5.7 per 100 000 person-years. This was much lower than the South Asian rates from England [8] (age-standardized relative risk 0.25).

Figure 1.

Age-specific incidence rates of CaP in Sri Lanka by district and for South Asian migrants to England.

Table 1.  Age-specific and standardized rates for districts in Sri Lanka and South Asians in England
Age groupColomboGampahaKandyOther districtsSri LankaEngland*Relative risk**
  1. *South Asians residing in southern England. **Relative risk of Colombo vs South Asians in southern England. ***Standardized to European hypothetical population. N/A, not applicable.

30–34000000N/A
35–390.40000.10N/A
40–440.50.31.80.20.33.80.13
45–491.20.300.50.55.30.22
50–543.13.74.01.92.416.40.19
55–59 11.55.24.74.85.888.20.13
60–6429.022.510.012.715.9 1100.26
65–6952.925.134.119.826.01360.39
70–7499.444.762.235.246.93410.29
≥75120.943.559.232.247.25120.24
Age-standardized rates (per 100 000)***12.45.86.84.25.749.60.25

On analysis of the NHSL data, of the 557 routine TURPs performed on males between the ages of 46 and 89 (mean age 68.82) in the years 2007 and 2008, 99 (16.8–18.75%) had incidental diagnoses of CaP. The majority of diagnoses of CaP (19.4%) were seen in the 66–70 age group while the 71–75 and 76–80 age groups had 12.2%. Only one male was diagnosed with CaP in the 46–50 age group. Over 90% of cases were from the western province (Gampaha, Colombo and Kalutara districts) with a predominance in the Colombo district.

DISCUSSION

CaP was the tenth most common cancer in Sri Lankan males in 2000 but the second most common cancer in the group aged ≥65 years [12]. The age-adjusted rate for a standard world population was 2.0 per 100 000 from 1995 to 2000 [12]. This rate doubles, as seen in the present study, for the next 5 years, where the incidence reaches 5.7 per 100 000 in 2005 (Fig. 2). This is unlikely to be a true increase in the incidence, as there is no significant increase in the rate per year over the 5-year period from 2000 to 2005 (Fig. 2). One of the present authors previously showed that the CaP incidence was higher than the published cancer registry data at the time in 1999, due to imperfections in data collection [13]. Therefore, these increases may be due to improvements in data collection, use of better detection methods and improved awareness and early presentation to healthcare services. This is reflected in the data from Colombo, which suggest that the incidence could even be as much as 12.5 per 100 000. Despite this, the overall incidence of CaP continues to be low.

Figure 2.

Incidence of CaP in Sri Lanka (adapted from the data provided by NCCP).

A previous publication from Mumbai has reported rates for 2000 of 13.6 per 100 000 for men aged 50–69 years and 107 per 100 000 for men >70 years [14]. These results are also similar to the Colombo rates which provide some weak evidence that there may not be major differences in the risk of CaP across the Indian subcontinent. This is in keeping with the data from other studies showing a low rate in the South Asian population in the Indian subcontinent [1,3–5].

TURPs are done in Sri Lanka for mainly obstructive symptoms and the histology is routinely examined for CaP. We have also included these data as a tool to observe for incidental diagnoses of CaP. A comparison with the literature (Table 2[15–18]) shows that the incidental finding of CaP in TURP specimens in Sri Lanka was higher than the data recorded in other studies. This suggests that the actual incidence of CaP may be higher than the data from the cancer registry, in keeping with the data from Colombo.

Table 2.  Percentage of incidental findings of CaP on TURP specimens
AuthorsSourceYearAge (years)Total no. of TURPs (no. of CaP detected on TURP)% TURP biopsies malignantReported national incidence (per 100 000)
  • *

    Age incidence rate per 100 000 for men >40 years.

  • **

    **Age-adjusted incidence rate from SEER data for the USA.

Mai et al. [15]Ottawa Hospital – Civic Campus1989–1990 (pre PSA era)Mean 71 (±6) to 73 (±9)533 (69)12.9237 (1987–1990*)
1997–1999 (PSA era)Mean 70 (±8) to 72 (±8)449 (36)8.0311 (1996–1998*)
Merrill and Wiggins [16]Utah cancer registry1992–199945+642610.5 (CI9.8–11.2)178.9**
Argyropoulos et al. [17]Clinic Athens General Hospital1999–2003Mean 69.7786 (34)4.3N/A
Jones et al. [18]Tertiary care hospital, USA1986–87 (pre PSA era)All but one >70228 (34)14.9 119.02–133.66**
1994–2000 (PSA era)501 (26)5.2180.04–183.45**
Sri Lanka National Hospital dataSri Lanka National Hospital, Colombo200746–89272 (51)18.75N/A
200846–89285 (48)16.8N/A

An alternative method of detecting under-diagnosis of CaP is by the mortality : incidence (MI) ratio. The MI standardized ratio in Sri Lanka for the year 1997 was 0.2 (mortality due to CaP 0.4 per 100 000 and incidence of CaP 2.0 per 100 000 Age Standardised Radio, ASR) [12]. In comparison with the data from 1988 to 1992, the Sri Lankan MI ratio is still lower than for low risk countries – Singapore (0.38), Japan (0.42) and Hong Kong (0.35) [3]. The MI ratio data can also be affected by under-assessment of both the mortality and the incidence data. Notably, the lowest MI ratio was in the US population, due to increased incidence reporting and improved survival data and treatment methods [3].

Age, family history and ethnicity are all known risk factors for CaP [19]. Therefore, numerous studies have been carried out investigating the reasons for a low incidence of CaP in South Asian populations. Numerous genes, diet, smoking, physical activity, antioxidant lycopenes, androgen levels and 5-alpha-reductase levels have all been associated with CaP [19,20].

The majority of the Sri Lankan population (Sinhalese and Tamils) have a common ancestry with the Indian population suggesting a similar genetic component, along with similar dietary, environmental and social factors. Several increases in dietary substances such as tomatoes, nutrients such as selenium and vitamin E are shown to lower the rates of CaP [20]. Phyto-oestrogens are shown to be protective for CaP, as seen in Japanese men with a high intake of soy products [21]. Phyto-oestrogens, isoflavonoids and lignans, are also found in legumes, especially lentils and beans, which are often eaten along with rice, which is the staple diet in Sri Lanka. The typical Sri Lankan diet consists of a high intake of carbohydrates, comparatively low amounts of fat, vegetables and fruits, which also is a feature seen throughout South Asia. All the above features in the South Asian diet may have a protective effect for CaP [19,20].

Remarkably, there seems to be an increased incidence of CaP in the migratory population of South Asian men in the developed countries, as seen in the UK (49.55 per 100 000 [8]) and California (69.9 per 100 000 [6]). Compared with the PROCESS study, our results also show a significant increase in the incidence of CaP in the migratory group to the UK. Interestingly a similar pattern is seen in Southeast Asian men [1,3] and their migratory population to developed countries [7].

The differences could be attributed to a higher rate of detection of CaP in the migratory population. This may be due to early cancer detection methods, better awareness programmes and low threshold for screening for CaP, such as the routine use of PSA. There is also some evidence to show that in fact the migratory population seek healthcare more readily [8].

The multiethnic cohort study [19] further shows these features in the Japanese population who migrated to Hawaii. It is seen that in the second generation of the Japanese population the incidence of many cancers was more similar to that in the native Hawaiian population than in the first generation Japanese [19]. This implies that the changes in the risk of cancer in adulthood were attributable to childhood exposure or that the environmental and behavioural changes were not completed until after the first generation [19]. However, the incidence of CaP was higher in the first generation than the second [19]. A possible explanation was under diagnosis as a result of the low rates of prostate screening in Japan at the time compared with the routine prostate screening system of Hawaii [19].

The above evidence strongly suggests that there may be protective genetic components associated with the South Asian/Southeast Asian population, which could be altered due to new environmental and behavioural factors. Several genes of high penetrance and low penetrance with polymorphism have been identified and are thought to be potentially associated with CaP [20]. However, no conclusive evidence for any genetic protective effect from CaP in South Asian males has been published.

STRENGTHS AND LIMITATIONS

Although all necessary measures are taken by the NCCP to minimize the extraneous variables affecting the data (personal communication), the incidence of CaP may be higher than reported [13]. As discussed previously, errors are mainly due to data from the war areas and financial and logistical problems in data collection. We have tried to counteract these errors by removing the data from the war areas and also using the incidental findings of CaP during TURP to gain better validity of the results.

The majority of the data from the cancer registry are from histological diagnoses but there is room for further improvement in data collection methods such as histological data from post mortem findings and data from the private sector. Therefore diagnosis of CaP in the data provided minimizes false positives, which may be seen with PSA testing.

The incidence rates of CaP in any country are partly dependent on the level of investigative procedures used. PSA is shown to be a useful test in the diagnosis of CaP [15], but in countries where PSA testing is low or rarely used as a screening or diagnostic test, the incidence of prostate cancer is likely to be low. PSA has a limited use in Sri Lanka due to financial constraints and therefore some of the patients may present late and may not be diagnosed.

The routine TURP data may have a selection bias as the majority of the patients admitted to the NHSL are from the western province. The data collection was also limited to a period of 2 years. Therefore these rates may not be a true reflection of the entire Sri Lankan population.

The cancer registry data show a low incidence of CaP in Sri Lanka at 5.7 per 100 000 person-years. However, the actual rates may be in the region of 12.4 per 100 000 as seen in the Colombo data. The incidence rate seems to be higher than in year 2000, which is probably due to better data collection/healthcare improvements rather than an actual increase in rate as seen by the yearly data over the 5-year period. However, there is always room for improvement in data collection methods in the cancer registry data and increasing active awareness. The trend in CaP seems to be similar to that of other South Asian countries and is significantly less than for the migratory population in the UK.

Remarkably, the rate of incidental diagnoses of CaP in TURP specimens in Sri Lanka was higher than the published data, but conversion to clinically detectable CaP seems to be low. Based on our findings, as well as the PROCESS study results [8], it seems likely that factors such as environment, diet and behaviour are responsible for this conversion. Therefore, further studies are required to examine the environmental and genetic components (such as genetic polymorphism or cytochrome activity) which affect the incidence of CaP in the Sri Lankan population.

CONFLICT OF INTEREST

None declared.

ACKNOWLEDGEMENTS

We would like to thank Dr M. A. Y. Ariyaratne, Director of NCCP, for granting us access and for providing the NCCP data. We are extremely grateful to Professor Yoav Ben Shlomo of the Department of Epidemiology at the University of Bristol, UK, for the statistical analysis and Dr Mala Tudawe of the Faculty of Medicine Sri Lanka for all the guidance and support with the project. We would also like to thank Dr Jayantha Balawardena and Mr B. J. Ranasinghe, who were extremely helpful in the data acquisition.

Ancillary