The descriptive epidemiology of Kaposi's sarcoma (KS) presents peculiar features. Until the onset of the AIDS epidemic, KS was recognized as a disease of elderly Jews of Eastern European origin, based on case-series [1–3], population-based incidence data from the USA  and crude rates from Israel . In sub-Saharan Africa, there are important variations among regions and tribes that strongly suggest a transmissible infectious agent, without excluding a genetic component of disease susceptibility [6–10]. Reported data from Europe provide evidence of a hundredfold difference between the south and the north, and of an increased incidence prior to the 1980s [11–17].
The role of induced immunodeficiency in modulating the occurrence of KS lesions is shown by their regression upon withdrawal of immunosuppressive agents [18, 19]. The appearance of the AIDS epidemic has offered the opportunity to investigate the relationship between immune stimulation and immune dysregulation, lifestyle factors and KS tumour cell proliferation. KS incidence is higher in HIV-seropositive subjects infected through homosexual contract than in other seropositive groups such as haemophiliacs and drug users [20,21]. Recently, a novel virus belonging to the Gammaherpesvirinae [termed human herpesvirus 8 (HHV-8)] was identified in almost all clinical presentations of KS [22,23]; this strongly points towards its involvement in KS pathogenesis. However, its seroprevalence data have been limited up to now becuase of the lack of a reliable assay and so far there have been limited data from the general population.
The pronounced gradient reported for classic Kaposi's sarcoma (CKS) incidence from northern to southern Europe may reflect either variation in the seroprevalence of the virus or host factors or both. It can, therefore, be expected that the distribution of CKS in Jewish migrants from different regions to Israel varies accordingly. In this paper, we present the results of a population-based analysis of time trends in the incidence of CKS in Jews living in Israel and of differences in the incidence according to area of origin.
The study population comprises cases of KS registered since 1960 in the population-based Israel Cancer Registry (ICR). More than 90% of all cancers are estimated to be reported to the ICR [24,25].
For the period between 1961 and 1989, cases were retrieved from the ICR file of all subjects notified and recorded as KS (International Classification of Diseases for Oncology (ICD-O) morphology code: 9140) , independently of topographical and behavioural codes. To increase completeness of case ascertainment, we reviewed cases registered as a primary tumour with uncertain or low appraisal of diagnosis, such as sarcomas not otherwise specified, which were localized in the skin, soft tissues and blood vessel (n = 1509). The criterium for KS was histological confirmation of a clinically definitive diagnosis based on a review of available medical records. The ascertainment survey did not reveal false-positive KS, while a total of 28 additional diagnoses (1.8%) were established only on the basis of medical records either as morphology misclassification (n = 21) or mis-registration of cancer site (n = 7).
The information retrieved from the ICR included date of birth, gender, country of birth or origin and date of immigration for Jewish immigrants, origin of parents for Israel-born Jews, date of KS diagnosis, topographical ICD-O codes, basis and appraisal of diagnosis and medical documentation. We excluded cases among non-residents, prevalent diagnoses among immigrants before their date of arrival in Israel and incidence diagnosed up to 6 months after date of immigration, the last to solve ‘ill-migration effect’. Incidence cases from population living outside the pre-1967 boundaries (West Bank and Gaza Strip) were also excluded, because of uncertainties regarding the accurate of notification. To distinguish CKS from AIDS-associated KS, the ICR file was linked with the national HIV-seropositive and AIDS registries. The HIV and AIDS registries started to record diagnoses from 1982 and receive notifications from multisources. Both registries use identifiers for the purpose of quality control and follow-up based on routine record linkages with different files, including population and death registers. A unique number identifies each Israeli citizen, of whatever origin, which allows full match-pairs to be found through record linkage procedures using dynamic key-identifiers and validators. Both registries were linked using different matching values to report anonymous match-pairs (for example gender, date and place of birth, father's name, residence, date of arrival if migrants, date of diagnosis). The files were linked twice after an interval of 6 months, to increase sensitivity and specificity. The linkage resulted in the exclusion of 32 HIV-seropositive KS cases (31 cases on the first linkage, an additional case on the second linkage).
Incidence rates were based on the number of CKS patients by 5-year age group, country of birth, gender and 5-year periods of diagnosis. The corresponding person-years at risk were calculated based on population census and inter-census figures, the latter being obtained after complete reporting of births, migration and death to the up-to-date Population Registry . The analysis of immigration status focused on continent of origin (Africa, Asia and Europe, Americas and Oceania combined) and on selected countries or groups of countries. The direct method of standardization was used, with the world standard population as reference.
We also conducted a multivariate Poisson regression analysis, adjusting for age at diagnosis, gender, country or continent of birth and calendar period or birth cohort, to assess the incidence rate among the different populations . All the analyses were repeated for the period 1972–1989 to examine the trend through a period of more stable incidence rates. Regression models were fitted using the STATA 4.0 statistical package program .
During the period 1961–1989, a total of 1098 cases of CKS occurred in the total Jewish population: 778 among men and 320 among women. Age-specific rates of CKS cases increased greatly with age; from 8.8 per million in men and 2.5 per million in women in those aged 35–44 to 250.7 per million in men and 88.4 per million in women in those subjects aged 75 or more. Among the population aged 5–14 the rates per million were 0.2 and 0.1 in males and females, respectively, with no cases at age below 5 (Table 1).
In the total Jewish population, the age-standardized rate was 16.9 per million in men and 6.3 per million in women during the period 1961–1989. There was a strong increase in the incidence of CKS in both genders during the late 1960s and the early 1970s (Fig. 1). The incidence during the periods preceding and following the steep increase, however, did not show a consistent trend in either gender. The ratio between the incidence in men and in women was 2.7 and was rather constant during the study period (range in quinquennia: 2.4–3.6).
The incidence was analysed by quinquennium (Table 2). The rates during 1965–1969 were 8.0 per million in men and 2.2 per million in women; corresponding figures in the following quinquennium were 17.9 and 6.7. Rates during the period 1970–1989 were 19.3 (standard error 0.2) per million in men and 7.3 (standard error 0.1) per million in women. In both genders, the incidence rates were not consistently different in Israel-born Jews and migrants across time, and the strong increase in incidence during the late 1960s and the early 1970s was present in both groups.
When we analysed CKS incidence rates in immigrants by continent of origin (Table 3), Jews born in Europe, and America and Oceania experienced rates similar to those of Israel-born Jews. While no patterns emerged for immigrants from Eastern Europe and countries of the Balkan, immigrants from Central European countries showed a statistically significant lower incidence rate in both genders compared with Jews born in Israel. Jews born in Africa displayed the highest rates in both genders, consistently across all calendar periods (data not shown). Most of these immigrants were from Northern Africa, and the incidence was particularly high among immigrants from Morocco, Algeria and Tunisia (rates per million 30.3 in men and 13.2 in women). Similarly, immigrants from Iraq (rates per million 41.2 in men and 21.7 in women) and Turkey (40.6 in men and 12.3 in women) showed high rates. Again, this pattern was present during the whole study period.
Risk of CKS was strongly dependent on age, while no trend was found according to calendar period. A strong birth cohort effect was found in models including age at diagnosis and birth cohort for the period 1961–1989. During the whole study period, immigrants as a whole had a relative risk (RR) of 1.17 [95% confidence interval (CI) 0.90–1.52] compared with Jews born in Israel. Relative risk varied from 0.47 (95% CI 0.34–0.65) for the birth cohort 1890–1899 to 2.31 (95% CI 1.33–4.00) for the birth cohort 1940–1949, taking the 1920–1929 birth cohort as reference. However, this trend mainly resulted from the younger cohorts contributed more than older cohorts to the post-1971 incidence. When the analysis was limited to the period 1970–1989, no residual cohort effect was detected: the RRs in the different decennial cohorts of birth varied between 0.90 and 1.03, taking again, the 1920–1929 birth cohort as the reference.
This study, based on 1098 registered CKS cases from 1961 to 1989 confirmed the findings of hospital-based reports and crude incidence rates, which indicated a high rate of CKS among Jews. The rates during 1970–1989 were 19.3 per million in men and 7.3 per million in women. These rates are 40 times those reported in comparable periods for England and Wales , New South Wales  and Denmark . In these countries the incidence is below 0.5 per million in men and below 0.25 per million in women. Incidence rates in Jews were up to 10 times higher than those reported from Sweden , Greece  and from the United States in the pre-AIDS era  (rates in the order of 3 per million in men and 1 per million in women), and were similar to those reported from Sardinia  and Sicily .
An interesting result is the increased risk among immigrants from North Africa, a result that agrees with incidence studies of KS before the AIDS era in Mediterranean countries. In Italy, a South to North increasing gradient in CKS incidence has been described . Only one case of CKS occurred among Italian immigrants in the present study. Similarly, KS incidence was reported to be high in Greece in the pre-AIDS era . The small number of cases from Greece (n = 8) and the lack of populations figures for the country precluded an analysis of Greek immigrants. However, the incidence among immigrants from Bulgaria and Greece was not high compared with other Mediterranean countries. An excess of CKS has been observed among immigrants to New South Wales (Australia) from Southern and Eastern Europe, the Middle East and Eastern European countries , and among immigrants to England and Wales (UK) from the Middle East and North Africa .
The relatively low incidence of CKS among immigrants from Central European countries is noteworthy and consistent with sparse reports from this area [1,2,33]. Immigrants from the former USSR, Poland and Romania had CKS rates higher than those of immigrants from other European countries but lower than those of immigrants from North Africa. This is consistent with previous reports and studies of immigrants to Australia , England and Wales  and Los Angeles .
It is worth noting that the pattern of CKS among the immigrant group differs from that of other neoplasms which have been suggested to share aetiological and mechanistic factors with KS . In particular, the incidence of non-Hodkin's lymphoma during 1961–1981 was elevated in migrants from Eastern European countries and Turkey, but not among immigrants from Northern Africa and Iraq .
Descriptive studies of CKS from a variety of countries reported an increase in incidence rates in the late 1960s and the early 1970s [12,16]. The same pattern was observed in the present study. The increase may be real or may result from improved diagnosis or registration. Under-registration of KS during the 1960s is likely to have played some role; however, it is unlikely to be the only explanation for the observed trend, as there is no reason why completeness of registration for KS would differ very much from that of other neoplasms. In addition, the increase is limited to the period 1968–1972, with no further increase in either gender after 1972. The fact that in Israel during the same time period similar increases were seen for cancers of the lip and cutaneous malignant melanoma [35,36] supports, in principle, the hypothesis that improved diagnosis during this period may have played a role. However, it is unlikely that it will entirely explain the observed increase, since the incidence of CKS has always been higher by several times in both native Jews born in Israel and migrants compared with other countries, even before 1970 [12, 16]. The improved diagnosis explanation, therefore, could be both unlikely and unsatisfying. The next simple answer is to note that the incidence increase of cutaneous malignant melanoma during the same period is likely to be real and to reflect increased sunlight exposure and not changes of diagnosis criteria . In addition, Israeli dermatologists and pathologists were well aware of the clinical and pathological features of KS for more than 50 years and the population of Israel had universal access to qualified medical care throughout the study period.
An alternative explanation for the increase could be the contribution from migrants from areas at different risk, who were infected with different HHV-8 strains that may have evoled in small, isolated populations : such different strains may interact in a type of ‘commute effect’ . In particular, there were large immigration waves to Israel from Iraq (during 1950–1951) and from North African countries (1951–1968) . No population-based incidence of KS is recorded for these countries. However, reports from the 1950s and 1960s suggest a relatively high occurrence of KS in North African countries, in particular in local Jewish communities [41,42].
The similarity in CKS rates between immigrants and Israel-born Jews has several possible explanations: (i) an important role of genetic factors, (ii) a similar predominant role of environmental factors in immigrants and their offspring (e.g., a vertically acquired infection), and (iii) the effect of a genetic factor modified by environmental conditions. We could not formally approach the third hypothesis because of the lack of detailed population figures by country of birth among parents of Israel-born Jews aged 45 and above. However, we retrieved the country of birth of the parents of 69 (77%) of the 90 Israel-born CKS patients that were notified during the period 1970–1989. For 45 patients (65%), at least one parent was born in Europe; for 14 (20%) at least one parent was born in Africa or Asia, and for 10 (14%) both parents were born in Israel. These proportions roughly correspond to the figures among immigrants . The first two explanations remain valid hypotheses and are likely to contribute to the pattern of our results.
On the one hand, the evidence of a strong difference in incidence by country of origin gives weight to an important role of geographical origin in the risk of CKS. On the other hand, the relatively high incidence rates in Jews born in Israel does not support a purely geographical hypothesis. In general, our data support the hypothesis of a genetic component of CKS affecting the response to an infectious or other environmental agent .
In conclusion, this study confirms, at the population level, the presence of elevated rates of CKS among Israeli Jews. The results also suggest that factors related to geographical origin and ancestry, or exposure early in life, contribute to the increased risk. In particular, our results confirm the notion of a high-risk area in Mediterranean countries.
The authors acknowledge the contribution of the staff of the Israel Cancer Registry and thank Dr S. Franceschi for helpful comments.
1. Dörffel J: Histogenesis of multiple idiopathic hemorrhagic sarcoma of Kaposi
. Arch Dermatol Syph
2. Rothman S: Some clinical aspects of Kaposi's sarcoma in the European and North American populations
. Acta UICC
3. Laor T, Schwartz RA: Epidemiologic aspect of American Kaposi's sarcoma
. J Surg Oncol
4. Ross RK, Casagrande JT, Dworsky RL, Levine A, Mack T: Kaposi's sarcoma in Los Angeles, California
. J Natl Cancer Inst
5. Landman R, Karsenty E. Katz L, Saimot AG: Le sarcome de Kaposi en Isrîl: donnes épidémiologiques 1960–1980
. Bull Soc Path Ex
6. Oettlé AG: Geographical and racial differences in the frequency of Kaposi's sarcoma as evidence of environmental or genetic causes
. Acta UICC
7. Davies JNP, Elmes S, Hutt MSR, Mtimavalye LAR, Owor R, Shaper L: Cancer in an African community, 1897–1956
. Br Med J
8. Slavin G, Cameron H, Singh H: Kaposi's sarcoma in mainland Tanzania: a report of 117 cases
. Br J Cancer
9. Taylor JF, Smith PG, Bull D, Pike MC: Kaposi's sarcoma in Uganda: geographic and ethnic distribution
. Br J Cancer
10. Wabinga HR, Parkin DM, Wabwire-Mangen F, Mugerwa JW: Cancer in Kampala, Uganda, in 1989–91: changes in incidence in the era of AIDS
. Int J Cancer
11. Gafa L, Gafa R, Dardanoni L: II sarcoma di Kaposi a Ragusa e in Sicilia. In: I Réunion du groupe pour I'pidémiologie et l'enregistrement du cancer dans les pays de langue latine (IARC Internal Report)
. Lyon: International Agency for Research on Cancer; 1984.
12. Dictor M, Attewell R: Epidemiology of Kaposi's sarcoma in Sweden prior to the acquired immunodeficiency syndrome
. Int J Cancer
13. Grulich A, Beral V, Swerdlow AJ: Kaposi's sarcoma in England and Wales before the AIDS epidemic
. Br J Cancer
14. Franceschi S, Geddes M: Epidemiology of classic Kaposi's sarcoma, with special reference to Mediterranean population
15. Cottoni F, De Marco R, Montesu MA: Classical Kaposi's sarcoma in north-east Sardinia: an overview from 1977 to 1991
. Br J Cancer
16. Hjalgrim H, Melbye M, Pukkala E, et al.
: Epidemiology of Kaposi's sarcoma in the Nordic countries before the AIDS epidemic
. Br J Cancer
17. Touloumi G, Kaklamanis L, Potouridou, et al
.: The epidemiologic profile of Kaposi's sarcoma in Greece prior to and during the AIDS era
. Int J Cancer
18. Penn I: Overview. Kaposi's sarcoma in transplant recipients
19. Kinlen LJ, Skeil AGR, Peto J, Doll R: Collaborative United Kingdom-Australasian study of cancer in patients treated with immunosuppressive drugs
. Br Med J
20. Rutherford GW, Payne SS, Lemp GF: The epidemiology of AIDS related Kaposi's sarcoma in San Francisco
. J AIDS
21. Beral V: Epidemiology of Kaposi's sarcoma
. Cancer Surv
22. Chang Y, Cesarman E, Pessin MS, et al.
: Identification of herpervirus-like sequences in AIDS-associated Kaposi's sarcoma
23. Moore PS, Chang Y: Detection of herpesvirus-loke DNA sequences in Kaposi's sarcoma in patients with and without HIV infections
. N Engl J Med
24. Steinitz R, Parkin D, Young J, Bieber C, Katz L, (Eds): Cancer Incidence in Jewish Migrants to Israel, 1961–1981 (IARC Scientific Publications 98)
. Lyon: International Agency for Research on Cancer; 1989.
25. arkin DM, Muir CS, Whelan SL, Gao Y-T, Ferlay J, Powell J, (Eds): Cancer Incidence in Five Continents
, Vol VI. (IARC Scientific Publications 120)
. Lyon: International Agency for Research on Cancer; 1992.
26. Word Health Organization: International Classification of Disease in Oncology
. Geneva: Word Health Organization; 1979.
27. Central Bureau of Statistics: Annual Statistics Abstract of Israel, Series
13–45, 1962–1994. Jerusalem: Central Bureau of Statistics.
28. Breslow N, Day N: Statistical methods in Cancer Research
, Vol II, The Design and Analysis of Cohort Studies. (IARC Scientific Publication 82)
. Lyon: International Agency for Reserch on Cancer; 1987.
29. StataCorp: Stata Statistical Software: Release
5.0. College Station, TX: Stata Corporation; 1997.
30. Kaldor JM, Coates M, Vettom L, Taylor R: Epidemiological characteristics of Kaposi's sarcoma prior to the AIDS epidemic
. Br J Cancer
31. Biggar RJ, Horm J, Fraumeni JF Jr, Greene MH, Goedert JJ: Incidence of Kaposi's sarcoma and mycosis fungoides in the United States including Puerto Rico, 1973–1981
. J Natl Cancer Inst
32. Geddes M, Franceschi S, Barchielli A, et al.
: Kaposi's sarcoma in Italy before and after the AIDS epidemic
. Br J Cancer
33. Hutt MSR: Kaposi's sarcoma
. Br Med Bull
34. IARC: Human immunodeficiency viruses and human T-cell lymphotropic viruses
. In: IARC Monographs on the Evaluation of Carcinogenic Risks to Humans
, Vol 67. Lyon: International Agency for Research on Cancer; 1996.
35. Doll R, Muir C, Waterhouse J, (Eds): Cancer Incidence in Five Continents
, Vol II. Geneva: International Union against Cancer. Berlin: Springer-Verlag; 1970.
36. Waterhouse JAH, Muir C, Correa P, Powell J: Cancer Incidence in Five Continents
, Vol III. Lyon: International Agency for Research on Cancer; 1976.
37. van der Esch EP, Muir CS, Nectoux J, et al.
: Temporal change in diagnostic criteria as a cause of the increase of malignant melanoma over time is unlikely
. Int J Cancer
38. Zong JCH, Metroka C, Reitz MS, Nicholas J, Hayward GS: Strain variability among Kaposi sarcoma associated herpesvirus (human herpesvirus 8) genomes: evidence that a large cohort of United States AIDS patients may have been infected by a single common isolate
. J Virol
39. Kinlen LJ: Epidemiology evidence for an infective basis in childhood leukaemia
. Br J Cancer
40. Davies AM: Migrants and their children in Israel: identification and change
. Isr J Med Sci
41. Chadli A, Rethers L, Landreat A, Habanec B: La physionomie du cancer en Tunisie: étude de 7959 cancers primitifs
. Arch Inst Pasteur Tunis
42. Marill FG, Sayag J, Hadida E: L'acro-sarcomatose de Kaposi en Algérie
. Méd Trop
43. Dalgleish AG: Kaposi's sarcoma
. Br J Cancer