Changing patterns of Hodgkin lymphoma incidence in Singapore
Version of Record online: 9 MAY 2008
Copyright © 2008 Wiley-Liss, Inc.
International Journal of Cancer
Volume 123, Issue 3, pages 716–719, 1 August 2008
How to Cite
Hjalgrim, H., Seow, A., Rostgaard, K. and Friborg, J. (2008), Changing patterns of Hodgkin lymphoma incidence in Singapore. Int. J. Cancer, 123: 716–719. doi: 10.1002/ijc.23504
- Issue online: 28 MAY 2008
- Version of Record online: 9 MAY 2008
- Manuscript Accepted: 24 JAN 2008
- Manuscript Received: 19 MAY 2007
- Hodgkin lymphoma;
A bimodal age-specific incidence pattern with a relatively high proportion of cases occurring in adolescents and young adults is a hallmark of Hodgkin lymphoma (HL) epidemiology in Western industrialized countries. The young adult incidence peak is believed to reflect the association between HL risk in young adults and an affluent childhood socioeconomic environment. However, the dynamic development of the young adult incidence peak following socioeconomic development implied by this interpretation has scarcely been demonstrated in a single population over time. We, therefore, analyzed incidence patterns of HL in Singapore between 1968 and 2004, during which time period a socioeconomic transition towards Western World lifestyles took place. As hypothesized a HL incidence peak emerged among adolescents and young adults in Singapore. Thus, in the age groups 15–19 and 20–24 years incidence rates increased annually by 7.0% (95% confidence interval 3.4%–10.7%) and 3.4% (0.1%–6.8%), respectively, in men and by 13.7% (9.1%–18.6%) and 12.2% (7.8–16.8%), respectively, in women between 1968 and 2004. However, the incidence peak remained considerably lower than what can be observed in young adults in the Western World. It remains to be determined to what extent the current lower incidence of HL in young Asian adults should be attributed to birth cohort phenomena, as would be suggested by continued increase in incidence, and to ethnic variation in HL susceptibility between Asian and non-Asian populations, respectively. © 2008 Wiley-Liss, Inc.
It is well-established that the occurrence of Hodgkin lymphoma (HL) varies considerably between populations.1–5 This is particularly true for children and young adults. In industrialized countries in the Western World, for example, HL incidence is low in childhood but increases during adolescence to peak in young adulthood.1, 2 In comparison, socioeconomically deprived populations reports have suggested higher incidence of HL in children, notably boys, combined with no young adult incidence peak.3, 4
To a large extent, these differences in HL incidence patterns in children and young adults are believed to reflect the relationship of an affluent socioeconomic childhood environment with the risk of HL in childhood (decreased risk) and in early adulthood (increased risk).3, 6 Consistent with this interpretation, incidence patterns displaying features characteristic of both socioeconomically developed and deprived settings have been observed in populations of intermediate socioeconomic levels.3, 5, 6
It is remarkable, however, that a dynamic transition between the different archetypical incidence patterns in conjunction to socioeconomic development implied by the above interpretation has scarcely been demonstrated in a single population over time. Most likely this has historical reasons as, in most settings, the emergence of the young adult incidence peak particularly characteristic for Western World populations have predated valid cancer registration. Accordingly, the most tangible evidence to suggest development of a young adult incidence peak comprises changing HL-specific mortality rates during the first half of the twentieth century in the U.S. and U.K.,2, 3, 6 analyses of HL incidence trends in Connecticut, US, e.g., between 1938 and 1992,7 and a comparison of 1947 with 1969–80 U.S. HL incidence rates.8
In Asian populations, HL incidence has historically displayed a low incidence throughout the first 4 decades of life with no peaks in childhood or young adulthood.1, 2, 5, 9 This distinguishing incidence pattern most likely reflects a balance between prevailing levels of socioeconomic development and genetic susceptibility to HL in these populations.9 From the perspectives of both the latter genetic susceptibility to HL and of the presumed dynamic nature of the young adult HL incidence peak it is therefore of interest that a distinct young adult peak, albeit considerably lower than in young adult U.S. Caucasians, has become visible among Asian immigrants in recent U.S. HL incidence data.9
This finding prompted us to assess whether HL incidence trends similar to those observed for U.S. Asian immigrants could be observed in recent data from Singapore. Thus, since its independence in 1965, Singapore has undergone rapid economic development to achieve one of the highest standards of living in Asia,10 and among the highest in the world. These changes have been accompanied by an equally rapid social and demographic transition towards more Westernized lifestyles and family structure.11, 12 We speculated that the secular changes in Singapore would be reflected in changing HL incidence patterns and in particular by the appearance of an incidence peak among young adults.
Material and methods
Since its inception in 1968, the Singapore Cancer Registry has been part of the International Agency for Research on Cancer serial publication Cancer Incidence in 5 Continents, attesting to its level of completeness and rigor in cancer reporting.13 In the Register we identified all HL patients (193 women, 337 men) diagnosed between 1968 and 2004. Annual Singapore population statistics (residents only) were obtained from the same source.
We first estimated sex- and 5-year age group-specific incidence rates of HL in 4 successive calendar periods of approximately equal length (1968–77, 1978–86, 1987–95 and 1996–2004). Next, we estimated incidence trends in specific age bands (0–14, 15–19, 20–24, 25–29, 30–49, 50+ years) by Poisson regression analyses of a data set cross-classifying events and time at risk by age (0–4, 5–9,…, 80+ years), calendar year (1968, 1969,…, 2004), and sex. We estimated sex- and age group-specific log-linear incidence trends in a model adjusted for sex and 5-year age categories. Confidence intervals were based on Wald tests, and all p-values were based on two-sided likelihood-ratio tests.
The age-specific incidence pattern for HL changed markedly between 1968 and 2004 (Fig. 1; Table I). In particular, HL incidence increased significantly in adolescents and young adults to produce an incidence peak, whereas in older adults HL incidence remained constant (in women) or decreased (in men) (Fig. 1, Table I). No changes in HL incidence were observed in children under the age of 15 years. Statistically, incidence trends among persons born before and after 1980, respectively, could be assumed to be similar (p = 0.31) (data not shown), as were incidence trends for older adults (50+ years) identical before and after 1980 (p = 0.50; data not shown).
|No. of cases 1968–2004||Crude incidence rate per 100.00 persons per year with 95% CI||Annual incidence rate change (%) with 95% CI||p-Value for trend|
|Men (age group)|
|≤14 years||36||0.34 (0.20–0.58)||0.14 (0.05–0.36)||0.34 (0.18–0.63)||0.28 (0.14–0.53)||−0.3 (−3.4, 2.8)||0.84|
|15–19 years||41||0.31 (0.12–0.84)||0.59 (0.28–1.23)||1.26 (0.73–2.16)||1.73 (1.08–2.79)||7.0 (3.4, 10.7)||0.0001|
|20–24 years||37||0.74 (0.37–1.48)||0.67 (0.33–1.34)||0.38 (0.14–1.01)||1.77 (1.10–2.85)||3.4 (0.1, 6.8)||0.04|
|25–29 years||25||0.67 (0.28–1.61)||0.46 (0.19–1.10)||0.32 (0.12–0.85)||0.94 (0.52–1.70)||2.2 (−1.4, 5.9)||0.24|
|30–49 years||74||0.87 (0.56–1.35)||0.33 (0.17–0.63)||0.54 (0.36–0.82)||0.43 (0.28–0.64)||−1.9 (−3.8, 0.1)||0.07|
|≥50 years||124||2.10 (1.45–3.04)||2.24 (1.61–3.13)||1.24 (0.84–1.81)||1.18 (0.85–1.64)||−2.4 (−4.0, −0.7)||0.006|
|Women (age group)|
|≤14 years||11||0.08 (0.03–0.26)||0.07 (0.02–0.29)||0.14 (0.05–0.39)||0.07 (0.02–0.26)||0.6 (−3.9, 5.2)||0.80|
|15–19 years||19||–||0.09 (0.01–0.63)||0.21 (0.05–0.83)||1.74 (1.06–2.84)||13.7 (9.1, 18.6)||<0.0001|
|20–24 years||27||0.10 (0.01–0.68)||0.35 (0.13–0.94)||0.39 (0.15–1.04)||1.88 (1.19–2.99)||12.2 (7.8, 16.8)||<0.0001|
|25–29 years||24||0.68 (0.28–1.64)||0.38 (0.14–1.00)||0.40 (0.17–0.97)||0.80 (0.43–1.50)||2.6 (−2.3, 7.9)||0.30|
|30–49 years||54||0.28 (0.12–0.62)||0.30 (0.15–0.59)||0.45 (0.29–0.72)||0.41 (0.27–0.62)||2.0 (−0.9, 5.0)||0.19|
|≥50 years||58||0.82 (0.45–1.47)||0.80 (0.46–1.37)||0.49 (0.27–0.88)||0.72 (0.48–1.08)||−0.9 (−3.2, 1.4)||0.43|
The ethnic composition of the Singapore population was stable in the study period, comprising ∼77% Chinese, ∼14% Malays, ∼7% Indians and ∼1% individuals of other ethnicities. This distribution was also apparent among the HL patients, 355 (67%) being Chinese, 106 (20%) Malays, 58 (11%) Indians and 11 (2%) of other ethnicity. The small number of non-Chinese patients prevented meaningful comparisons of ethnicity-specific incidence trends. However, the described incidence changes in adolescents and young adults were observed both in the Chinese population and the combined group of all non-Chinese populations, while the declining incidence in 30+ year old men seemed to be restricted to the Chinese population (data not shown).
Histological information was reported to the Register for 366 cases (69%), of which the nodular sclerosis and mixed cellularity HL subtypes made up 66% (n = 241) and 20% (n = 73), respectively. The proportion of unspecified HL cases decreased over time, from 66% in 1968–1977 to 19% in the period 1996–2004, rendering evaluations of temporal changes in the subtype composition difficult. Still, it is noteworthy that 76% of HLs in the age group 15–29 years were of the nodular sclerosis subtype in the period 1996–2004.
Overall, the present analyses demonstrated that HL incidence patterns in Singapore changed between 1968 and 2004. In particular, we found evidence that a young adult incidence peak emerged during a time period characterized by considerable socioeconomic and demographic transition. At the same time, in contrast, the incidence of HL among older adults tended to decrease, whereas with the limitations inherent in statistical analyses of only 47 observed cases, the low incidence of HL for children under the age of 15 years appeared to have been stable over the 37 year study period.
The tendency for decreasing HL incidence in the oldest age group parallels observations in other geographical settings.14, 15 There, the decrease has in part be attributed to improving diagnostic precision accompanied by reduced misclassification of non-Hodgkin lymphomas.14, 16, 17 This phenomenon presumably also contributed to the decreasing HL incidence in older adults in Singapore, however, it is hardly a plausible explanation for the continued decreasing HL incidence in older adults observed also in more recent years. Accordingly, the decreasing incidence of HL in older adults presumably is a true phenomenon both in Singapore as well as elsewhere.15, 17
Diagnostic misclassification of non-Hodgkin lymphomas as HL has been less pronounced in younger than in older adults.14, 16 In addition, the reduction of such misclassification could not explain the observed increasing HL incidence in young adults but would rather render the observed trend estimates conservative. Although we cannot rule out a contribution from reduced diagnostic misclassification in the opposite direction, i.e. of HL previously registered as non-Hodgkin lymphoma (the incidence of which increased during the study period18), we consider it unlikely to entirely explain our observation.14
Genetic traits, the prevalence of which may be linked to ethnicity, most certainly influence the risk of or more specifically the susceptibility to HL19 upon relevant environmental stimuli. The emergence of a young adult HL incidence peak could therefore theoretically be explained by changes in the ethnic composition of the Singapore population. Given the short study period in our investigation, this could only be due to immigration of susceptible individuals, or alternatively by the emigration of non-susceptible individuals. However, the ethnic composition of the Singapore population was stable during the study period, and moreover, the incidence increase in young adults was observed both in the Chinese and non-Chinese populations in Singapore.
We therefore speculate that the development of a young adult incidence peak in Singapore should be attributed to changes in the prevalence of environmental risk factors for HL. In earlier epidemiological studies, risk of HL in young adults has been associated with correlates of socioeconomic affluence such as small sib-ships, early birth order, advanced parental education and low housing density.20 During the study period, a demographic and socioeconomic transition occurred in Singapore,11, 12 as evidenced by a wide variety of official statistics. The gross national product increased to become the highest in the region and among the highest in the world,10 with standards of living improving to become comparable with those of many developed countries.12 Also, the average size of households changed during the study period from 5.3 persons in the 1970 census to 4.2 persons in 1990.21, 22 Among other things, this resulted from fertility levels falling, e.g. from more than 3.0 per 1,000 women in 1970 to replacement levels in 1975, from whence a less steep reduction continued until the 1980s. Accordingly, fourth order births made up 33% of all births in 1970 but less than 10% after 1980.23 Moreover, a public rehousing program with the construction of large numbers of low-cost public housing units to replace crowded tenements in the city changed housing density. Hence, the number of rooms per household increased from 0.76 in 1954 to 2.15 in 1970, and the average number of persons per room fell from 4.84 to 2.52.24 On the basis of the current understanding of HL epidemiology, it is therefore entirely conceivable that these changes in living conditions and family structure would be accompanied by changes in HL incidence in young adults to approach patterns seen in Western World countries.
Further support for this interpretation comes from the histological subtype composition of the observed young adult HL cases, which also varies by socioeconomic level and ethnicity. While mixed cellularity and lymphocyte depletion HL appear to dominate in developing regions and in Asia, nodular sclerosis HL dominates in the young adult incidence peak in socioeconomic affluent settings.3, 9, 15, 25 Although the large number of unclassified cases of HL in the Singapore Cancer Registry early in the study period prevented meaningful subtype-specific analyses, we note that nodular sclerosis HL ultimately constituted 76% of the young adult cases, consistent with the suspected explanation for the increasing HL incidence.
Analogous to our observations, a young adult HL incidence peak is also visible in recent data from Hong Kong.13 Our observations are compatible with those of Glaser and Hsu among Asian U.S. immigrants9 and with their suggestion that HL risk factors may operate similarly in both Asian and non-Asian populations. Irrespective of the increase, however, young adult HL incidence in these Asian populations remain considerably lower than in Caucasians of the same age in Western countries.13, 15 Because of the association with childhood socioeconomical environment,20 incidence of HL in young adults may reflect circumstances prevailing decades earlier. Therefore, it remains to be determined to what extent the current lower incidence of HL in Asian young adults should be attributed to birth cohort phenomena, as would be suggested by continued increase in incidence, and to ethnic variation in HL susceptibility between Asian and non-Asian populations, respectively.
In conclusion, we observed the emergence of a young adult HL incidence peak in Singapore in conjunction to changes in the country's socioeconomic profile. The coincidence of these 2 phenomena would be consistent with the established association between young adult HL risk and correlation of childhood socioeconomical environment. The lower young adult peak incidence in Singapore than in Western World countries is likely to reflect a balance between environmental risk factors, whose full impact on incidence rates may not yet be evident, and a lower prevalence of genetic susceptibility to HL in Asians. Future studies should aim at identifying the components of this gene–environment interaction.
We thank the National Disease Registries Office of the Health Promotion Board, Singapore, for approval to conduct this study, and the Singapore Cancer Registry for providing nonidentifiable, aggregate incidence data. The authors declare no conflicts of interest.
- 4Hodgkin's and non-Hodgkin's. Lymphomas Cancer Surveys 1994; 19/20: 423–53., , .
- 10World in Figures. London: The Economist and Profile Books, 2006.
- 13Cancer incidence in five continents, vol. VIII. Lyon: IARC, 2002., , , , .
- 18Trends in cancer incidence in Singapore 1968–2002, Report No. 6. Singapore Cancer Registry: Singapore, 2004., , , , , .
- 20Hodgkin lymphoma. In: SchottenfeldD,FraumeniJ,Jr, eds. Cancer epidemiology and prevention,3rd edn. Oxford: Oxford University Press, 2006. 872–97., .
- 21Report on the census of population 1970. Singapore: Government Printing Office, 1973..
- 22Census of population 1990. Singapore: Department of Statistics, 1992..
- 23Singapore's current population trends. Singapore: Singapore Department of Statistics, 2002..
- 242007. Available at http://countrystudies.us/singapore/14.htm. Singapore: a country study. Washington: GPO for the Library of Congress,