Estimates of the worldwide frequency of twelve major cancers.

Articles in the Update series Les articles de la rubrique give a concise, authoritative, Le point fournissent un bilan and up-to-date survey of the concis et fiable de la situa- present position in the se- tion actuelle dans le do- lected fields, and, over a maine considere. Des experts .19c a e period of years, will cover couvriront ainsi successive- many different aspects of ment de nombreux aspects des the biomedical sciences and sciences biomedicales et de la Le goint public health. Most of the sante publique. La plupart de articles will be written, by ces articles auront donc ete invitation, by acknowledged rediges sur demande par les experts on the subject. spcicialistes les plus autorises. Bulletin ofthe World Health Organization, 62 (2): 163-182 (1984) © World Health Organization 1984 Estimates of the worldwide frequency of twelve major cancers* D. M. PARKIN,' J. STJERNSWARD,2 & C. S. MUIR' By examination of incidence, mortality and relativefrequency data an estimate has been made of the number of cancer cases in twelve common sites and of all cancers that occurred in 1975 in the 24 areas of the world for which the United Nations publishes population data. While several cancers are of importance in localized areas or regions, e.g., cancer ofthe larynx, these are infrequent on the world level and are not included in this review. While the relative importance ofthe selected sites varies from one area to another, on a global basis the first six ranking cancer sites in males are lung, stomach, colon/rectum, mouth/pharynx, prostate, and oesophagus; in females, they are breast, cervix uteri, stomach, colon/rectum, lung, and mouth/pharynx. Cancers ofthese sites, together with the leukaemias and cancers of the liver, bladder, and lymphatic tissues, account for 75% of the estimated 5.9 million new cancers that occurred in 1975. When the two sexes are combined, stomach cancers are in first rank, followed closely by lung; it is suggested that, given current trends, their rank order will soon be reversed. There are clear opportunities for prevention of cancer by controlling tobacco smoking, reducing infection by hepatitis B virus, and curbing the excessive intake of alcohol. The increasing adoption of high fat diets may lead to more cancers of the large bowel, breast, and prostate. Cancer is increasingly recognized to be a global problem, and not one limited to the industrial nations. At present there is great variation in the patterns of cancer occurrence in the different regions of the world. Because it is predominantly a disease of the older age groups, its relative importance in any region will depend on the age composition of the inhabitants. The age structure of the population in developing countries is changing rapidly, largely as a result of a reduction in infectious disease mortality. The adoption of more "Western" life-styles also will probably increase the cancer risk. Since the developing countries currently contain some three-quarters of the global population, the size of the cancer problem and the investment needed for its control in all parts of the world are likely to increase considerably in the future. This article presents an estimate of the overall burden of the common sites of cancer; it describes both the present size of the problem and also the potential value of preventive * Requests for reprints should be sent to Dr D. M. Parkin, International Agency for Research on Cancer, 150 Cours Albert Thomas, 69372 Lyon Cedex 2, France. A French translation of this article will appear in a later issue of the Bulletin. ' Unit of Descriptive Epidemiology, International Agency for Research on Cancer, Lyon, France. 2 Cancer unit, World Health Organization, Geneva, Switzerland. 4390 -163- D. M. PARKIN ET AL. measures. However, preventive measures taken today will affect the cancer incidence only at some time in the future and a more reasonable assessment of the effects of preventive strategies requires that some projection of future cancer patterns is made: we thus speculate on how these patterns may change. None the less, it is clear that control of tobacco smoking alone would reduce the total burden by over a million cancers each year. Since we are concerned with preventing cancer, we concentrate in this paper on its incidence-the number of new cases occurring-rather than on mortality. Cancer incidence rates (number of new cases per 100 000 population) are very strongly related to age. For any given community, therefore, the crude rate (total cases per 100 000 population) will be highly dependent on the age structure of the population. Many developing countries have very young populations, with over 4007o of the total under 15 years old and less than 5% aged 65 years or more; in contrast, these percentages in Europe and North America are less than 25% under 15 years old, and more than 10% aged 65 or more. The crude incidence rates for cancer will therefore be lower in the developing countries than in developed countries, even if there is little difference in the age-specific rates. The actual number of cases occurring will also depend on the size of the population. Differences in the risk of developing cancer mrust be- studied using age-specific or age- standardized rates. We have not attempted to estimate these, partly because the available data are, for many areas of the world, insufficient, and partly because our main concern is to estimate the actual numbers of new cancer cases occurring each year. MATERIALS AND METHODS In order to build up the global picture, we chose to study cancer patterns in 24 geographical areas for which population estimates and projections are regularly produced by the United Nations. These areas are shown in Fig. 1, and the population estimates for Fig. 1. Map showing the 24 areas in the world. 164 WORLDWIDE FREQUENCY OF TWELVE CANCERS 165 1975 in Table 1. The year 1975 was chosen as this was the midpoint of much of the cancer incidence and mortality data available. For each of these areas, the cancer incidence rates for twelve common tumour sites have been estimated. The sites chosen were mouth and pharynx (140-149)," oesophagus (150), stomach (151), colon and rectum (153-154), liver (155), lung (162), breast (174), cervix uteri-invasive (180), prostate (185), bladder (188), lymphatic tissue (200-203), and leukaemia (204-207). These represent most of the common cancers for which data of reasonable quality can be obtained. Three of these (oral-pharyngeal, lymphatic tumours, and leukaemia) are groups of tumours of varying etiology and epidemiology; however, it is not always possible to study the subdivisions within the groups, either because the boundaries between them may not be clear (e.g., oral cavity) or because data are only available for the group as a whole. Several cancers that are common in many areas of the world have been excluded. Skin cancer is extremely common, but because of its (relatively) trivial nature, recording of Numbers in parentheses denote the rubrics of the Eighth Revision of the International Classification of Diseases (ICD). Table 1. Population estimates for different areas in the world, 1975° Population (millions) Age distribution (%) Area Total Males Females Under 1 5 65 years years or more 1. East Africa 115 57 58 45 2.9 2. Middle Africa 47 23 24 43 3.2 3. Northern Africa 94 47 47 44 3.5 4. Southern Africa 29 14 1 5 42 4.1 5. Western Africa 121 60 61 46 2.5 6. Caribbean 28 14 14 41 5.0 7. Middle America 80 40 40 46 3.4 8. Temperate S. America 38 19 19 30 7.4 9. Tropical S. America 176 88 88 42 3.5 10. Northern America 236 115 121 25 10.3 11. China 928 474 453 37 5.2 12. Japan 112 55 57 24 7.9 13. Other East Asia 57 29 28 38 3.7 14. Eastern South Asia 326 162 164 42 3.1 15. Middle South Asia 845 437 408 43 2.9 16. Western South Asia 85 43 42 43 4.0 17. Eastern Europe 106 51 55 23 11.4 18. Northern Europe 82 40 42 23 13.7 19. Southern Europe 134 65 69 26 10.7 20. Western Europe 1 52 74 78 23 13.6 21. Australia/New Zealand 17 8 9 28 8.7 22. Melanesia 3 1.5 1.5 43 3.1 23. Micronesia/Polynesia 1.3 0.7 0.6 43 2.8 24. USSR 254 118 136 26 9.0 Total 4066 2038 2028 37 5.7 a Source: Demographic indicators of countries: estimates and projections as assessed in 1980. New York, United Nations, 1982. 166 D. M. PARKIN ET AL. incidence is unlikely to be complete or accurate. Cancers of the pancreas and ovary pose considerable diagnostic problems, and it was considered that for many areas figures would be either unavailable or unreliable. Similar considerations led us to omit cancer of the uterine corpus; for this site there are further difficulties resulting from changes in classifi- cation. Some tentative estimates were prepared for laryngeal cancer which is relatively common in males (worldwide it is perhaps as frequent as leukaemia, for example) but rather rare in females, and so is not included in the tabulations. Certain cancers with localized areas of high frequency (e.g., penile cancer) have not been included since the over- all world totals are lower than any of the twelve cancers for which figures are presented. Finally, an estimate was made of the incidence of all malignant neoplasms (excluding skin (ICD 173)). Methods of estimation The data sources used for each of the 24 geographical areas are detailed below. The procedure employed was to estimate the crude incidence rates for all (or most) of the countries in an area, and then to calculate a weighted average, where the weights are the populations of the individual countries. Wherever they were available, incidence rates derived from population-based cancer registries have been used, notably those published in the series Cancer incidence in five continents (1-4). For many registries both the incidence and the mortality rates are available for the same populations, and for 35 centres (in 27 countries) the incidence was plotted against mortality for the tumour sites being studied. Not surprisingly there is a close correlation between the two, and a regression line can be fitted, the slope of which will be related to the lethality of the tumour. When no incidence data were available for a country, the rates have been estimated by using the mortality data (5) and these regression lines. For many countries and some entire areas, where neither incidence rates nor mortality rates were available, data on the relative frequency of different tumour types have been used, a hypothetical crude "all sites" rate being multiplied by the frequency of the site concerned. The "all sites" incidence was calculated from the regression lines of mortality vs incidence, derived as described above, using the estimates of cancer mortality (male and female) by world area, which were prepared by WHO (T. Nakada, personal communi- cation).? As much of the data on cancer incidence or mortality related to the mid-1970s, the estimates of rates and numbers of cancers have been prepared for 1975. 1. East Africa Population-based data have been available from several centres in the past: Kyadondo (Uganda) (1), Bulawayo (Southern Rhodesia, now Zimbabwe) (2, 3), Lourenco Marques (now Maputo, Mozambique) (1), and Moshi (United Republic of Tanzania). c Some recent relative frequency data from Kenya,d Madagascar (7) and Lusaka (Zambia)e have been used to estimate rates for these countries, using a hypothetical all-sites rate of 90 per 105 for males, and 100 per 105 for females. b The methodology involved was essentially that of Preston (6), in which information on mortality rates for 12 cause-of- death groupings from many different populations at various stages of development is used to derive a series of linear relationships of the form mi = ai + biM to estimate the cause-specific death rate (Mi) from the overall crude death rate (M). c LAUREN, K. Unpublished data from. Kilimanjaro Cancer Registry (Moshi, United Republic of Tanzania), 1980. d KUNGU, A. Kenya Cancer Registry: unpublished data for 1969-79. ' WATTS, T. Unpublished data from Zambia Cancer Registry, 1982. WORLDWIDE FREQUENCY OF TWELVE CANCERS 167 In all these centres, oesophageal cancer is more frequent than stomach cancer, with particularly high rates in Bulawayo. Liver cancer is very frequent; the rates appear to show a gradient from south to north, being very high in Mozambique and Bulawayo, and considerably lower in Uganda and Kenya (Madagascar, however, has a low relative frequency of liver cancer and high rates for colon cancer). Since the majority of the population inhabits the north of this area, the rates have been weighted accordingly. The frequency of lymphoma in young Africans is reflected by the relatively high crude incidence rate for these tumours. In all centres, cervix cancer appears to be approximately twice as frequent as breast cancer. 2. Middle Africa Cancer data from this area are particularly sparse. The figures reproduced are crude estimates based on relative frequency data from Angola!f Gabon (8), Cameroon (9) and the Congo (10), using a notional "all-sites" incidence of 90 per 105 (males) and 100 per 105 (females). Liver cancer is clearly the most commonly occurring tumour, and lymphoma is also frequent. 3. Northern Africa No population-based incidence data are available. However a comprehensive hospital registration system operates in Cairo (44) and good pathological data are available from three centres in Algeria (11). In addition, relative frequency data are available from Tunisiag and Sudan (45). An estimated "all-sites" incidence of 100 per l05 (males) and 110 per iOs (females) has been used to calculate incidence rates. Published mortality data from Egypt (5) seem to confirm the general pattern, but the rates are low. These sources suggest rather low rates for gastrointestinal tumours (including liver). Rates and frequencies for breast cancer exceed those for cervix in all countries except Algeria. The high estimated rates of bladder cancer reflect the frequency of this tumour in Egypt, and the elevated rates for oral and pharyngeal neoplasms are at least partly explained by cancer of the nasopharynx, which accounts for about half of such tumours in the Sudanese and Algerian data. 4. Southern Africa In this area, 90Wo of the population lives in the Republic (before 1961, the Union) of South Africa. Few recent data are available, but population-based rates, by racial group, are available for Johannesburg (1954) (1), Cape Province (1956-59) (2) and Natal (1964-66) (2). These allow approximate rates for this country to be estimated, given the racial composition of the country. In addition, there are some relative frequency data available for Botswana (12) and Lesotho (13) -estimated all-sites rates, 90 per 105 (males) and 100 per 105 (females). 5. Western Africa Data are available from three sources: population registration in Ibadan (Nigeria) (1-3) and Dakar (Senegal) (4), and the multi-centre hospital registry of Liberia (14). The combined rates are heavily weighted by the Nigerian data, since this country has almost 60% of the population in this area. The crude (all sites) rates from this source appear low (33 per 105 (males) and 45 per 105 (females)) in comparison with the rates based on estimated mortality. f LOPES, C. 0 cancro en Angola. Que perspectivas? (unpublished). g MOURALI, N. Unpublished data from Salah Azaiz Institute, Tunis, 1976-80. 168 D. M. PARKIN ET AL. Cervix cancer shows consistently higher rates in Ibadan than liver cancer (this is supported by relative frequency data from other Nigerian centres and from Liberia); hence the crude rates of liver cancer (the most frequent site in the Sahel countries) are similar to those for cervix cancer for the area as a whole. The high frequency of lymphomas in young persons results in high crude incidence rates. 6. Caribbean Recent population-based incidence data are available from four centres (Cuba, Puerto Rico, Jamaica, Netherlands Antilles), which together contain over half the population of the area (4). A combined rate weighting for the populations of these centres is given. 7. Middle America There are no population-based incidence data. The national pathology registry of Panama allows estimation of minimum incidence (crude rates for all sites are 58 per 105 (males), and 54 per 105 (females)) (15). Mortality rates are available for several countries, including Mexico, Costa Rica, Panama, Nicaragua, El Salvador and Honduras (5). The rates for the latter three seem particularly low and are likely to be underestimates; but the rates from the former three have been used together with the Panamanian incidence data to estimate rates of incidence for the area. 8. Temperate South America Incidence data have been published for La Plata (Argentina) for 1980(16) and mortality rates are available for Argentina (1977 only), Chile, and Uruguay (5). For liver and bladder cancer, incidence rates in Chile were estimated from registrations in 1959-62 (1) and mortality rates in 1980 (5). These different sets of data suggest a fairly uniform picture of tumour incidence in the three countries, the main exception being with regard to female cancers: cervix cancer is more frequent than breast cancer in Chile, the converse of the situation in Argentina and Uruguay. 9. Tropical South America Population-based incidence data are available from three contrasting areas of Brazil: Sao Paulo (4), Recife (3) and Fortaleza (46), from Cali (Colombia) (4), La Paz (Bolivia) (17), Asuncion (Paraguay)h and Lima (Peru) (18). Mortality rates from Ecuador and Ven- ezuela (5) were used to estimate the incidence in these countries. With appropriate weight- ings for the populations represented by these rates, estimates for the area can be produced. 10. North America Population-based incidence data are available from Canada (4) and from the SEER (Surveillance, Epidemiology, and End Results) programme in the USA (19), and the respective populations are in the ratio 1:9.3. The resulting weighted rates are as shown in Table 2. 11. China This is the largest of the 24 areas in this study so that the estimated incidence rates will be important in determining the global frequency of several tumour types. Incidence rates for hROLON, P. A. Unpublished data from Cancer Registry, Asunci6n, 1981. WORLDWIDE FREQUENCY OF TWELVE CANCERS 169 Table 2a. Estimated crude rates of cancer incidence (per 100 000), by sex, site and area Mouth/ Oesophagus Stomach Colon/ Liver Bronchus/ pharynx (150) (151) rectum (155) lung (140-149)a (153, 154) (162) Area M F M F M F M F M F M F 1. East Africa 2. Middle Africa 3. Northern Africa 4. Southern Africa 5. Western Africa 6. Caribbean 7. Middle America 8. Temperate S. America 9. Tropical S. America 10. North America 11. China 12. Japan 13. Other East Asia 14. Eastern South Asia 1 5. Middle South Asia 16. Western South Asia 17. Eastern Europe 18. Northern Europe 19. Southern Europe 20. Western Europe 21. Australia/New Zealand 22. Melanesia 23. Micronesia/ Polynesia 24. USSR 7 6 5 1 4.5 3.5 2.5 2 22 6 2 0.4 5 3 0.9 0.6 3.5 2 4 3 18 6.5 2.5 1 11.5 5.5 2 1.5 4.5 1.5 5.5 3.5 3 2 2.5 0.6 8.5 4 13 3 13 8 4.5 7.5 13 2.5 19 3 2 1.5 0.4 0.2 2.5 2 1.5 1.5 9 3 0.9 0.5 12 3.5 7.5 3 15.5 7.5 10 11 4.5 3 35 12 2.5 2 2 1 9 6.5 4.5 6 2 1.5 7.5 3 18 6.5 11.5 3.5 27 17 23 27 5.5 3.5 39 7 9 4 4 1 19 11 7 7.5 3 3 11 3 16 6.5 5 2 12 7 48 46 3 8 4.5 23 13 29 15 8.5 8 17 3.5 2 8 2.5 85 50 17 15 14 7 3.5 3.5 0.8 28 14 8 6 18 9 4.5 3.5 1.5 8 4.5 6.5 5.5 15 18 9.5 6 4.5 4.5 2.5 3.5 2.5 1.5 1.5 68 21 6.5 11 5 7 24 9 4 13 5 4 17 6 1 6 1.5 4 2 2.5 2 6 3.5 5 4 2 1 11 2.5 12 3 3.5 1 41 25 22 22 7 5.5 59 8.5 9 5 7 5 31 21 43 46 2.5 1.5 93 23 14.5 3 8 2.5 44 30 30 30 11 5 62 13 20 4 10 1.5 34 22 43 45 4.5 2.5 87 11 15.5 4 5 2.5 16 10 38 39 1.5 0.8 57 12 19 14 2 0.3 4 2.5 5 2.5 14 5 4 1 4 1.5 3 0.4 7 3 6.5 5 13 4 8 5.5 47 36 11 15 5 2.5 8 2 7 5.5 43 10 a Numbers in parentheses indicate the ICD codes according to the Eighth Revision of the International Classification of Diseases. Chinese populations are available from population-based cancer registries in Shanghai, Hong Kong and Singapore (4). In addition, mortality rates for the whole of China for 1975-78 are presented in the National Cancer Atlas (20). We have used the rates in this latter volume to calculate crude mortality for 1975, and estimated the incidence from these mortality rates. The rates for nasopharyngeal cancer have been multiplied by 1.5 to approximate the rate of all oropharyngeal cancer. No national mortality data are available for prostatic cancer, and the estimate produced is a weighted average of the incidence observed in the Chinese populations of Shanghai, Hong Kong and Singapore. 170 D. M. PARKIN ET AL. Table 2b. Estimated crude rates' of cancer incidence (per 100 000), by sex, site and area Lymphatic All sites Breast Cervix Prostate Bladder tissue Leukaemia (excl. skin) (174) (180) (185) (188) (200-203) (204-207) (140-207, excl. 173) Area F F M M F M F M F M F 1. East Africa 2. Middle Africa 3. Northern Africa 4. Southern Africa 5. Western Africa 6. Caribbean 7. Middle America 8. Temperate S. America 9. Tropical S. America 10. North America 11. China 12. Japan 13. Other East Asia 14. Eastern South Asia 1 5. Middle South Asia 16. Western South Asia 1 7. Eastern Europe 18. Northern Europe 19. Southern Europe 20. Western Europe 21. Australia/New Zealand 22. Melanesia 23. Micronesia/ Polynesia 24. USSR 9 12 25 20 7 30 14 79 21 21 19 27 1 1 20 21 24 5.5 6 2 8 3 19 7.5 22 27 32 1 0 3.5 1 10 5.5 1.5 1 86 83 3 1.5 12 7.5 1 0.6 90 100 16 6 11 5.5 5 3.5 100 110 3.5 1.5 6 3.5 3.5 2.5 121 116 2 1 9 5.5 2 1.5 42 52 7.5 2 8 5 5 3.5 146 123 3.5 1 4 3.5 2.5 2.5 73 94 1 6 6 6 4.5 7 5 232 256 7 4.5 4.5 3.5 104 135 87 13 55 22 7.5 16.5 14 1 1 8 12 20 16 14 29 17 26 21 2 4 3 2.5 15 24 1.5 323 315 2 0.6 3 2.5 4.5 3.5 138 123 4.5 2 5.5 4 4.5 4 199 171 3.5 1 4 2 4 3 128 144 2 0.6 3.5 2 4.5 3 91 89 1.5 0.4 4.5 2 1 5 4.5 3.5 5 0.9 6.5 4 36 82 61 84 66 24 17 13 23 13 17 37 26 42 34 3 2 72 79 4 3 82 72 1 1 2.5 1 1 8.5 7.5 5.5 234 216 22 7.5 14 1 1 9 7.5 334 313 21 3.5 1 2 9.5 1 0 6.5 310 253 25 5.5 1 2 9.5 9.5 7.5 365 325 1 8 5.5 1 5 1 2 9 6.5 273 254 13 20 1 1 0.3 8 4 4 2 107 114 7 20 1 2 0.8 0.9 6 2.5 3.5 1.5 80 90 23 23 7 9 3 4 3 5.5 4.5 175 168 e Values > 10 rounded to nearest whole number Values < 10 rounded to nearest 0.5 Values < 1 rounded to nearest 0.1 12. Japan The numbers of cancer cases for the whole of Japan have been estimated from the data of seven population-based cancer registries (21). Figures for cervix cancer are not presented separately in this publication, and for this site incidence rates have been calculated for the four registries in Cancer incidence in five continents (4) (Fukuoka, Miyagi, Nagasaki, and Osaka) combined. WORLDWIDE FREQUENCY OF TWELVE CANCERS 171 13. Other East Asia This area includes Hong Kong where incidence rates are available (4) and the Republic of Korea, where multi-centre hospital registration provides information on relative frequency from which incidence rates have been estimated' (crude all-sites rates: 120 per 105 (males) and 140 per 105 (females)). These data suggest a rate of cervix cancer almost double that of breast cancer, and the frequency of liver cancer may well approach that seen in Hong Kong Chinese. Stomach cancer appears to be some 10 times more frequent than oesophageal cancer. 14. Eastern South Asia This large area has a population-based registry in Singapore which publishes the inci- dence rates for Chinese and Malay populations separately (4), hence some idea of incidence rates in neighbouring countries (Malaysia, Indonesia, Vietnam) can be obtained. Incidence rates are also produced by cancer registries located in Rangoon (Burma),J and Manila (Philippines) (47). There is a multi-hospital registry in operation in Thailand (48), and relative frequency figures have been published from Indonesia (22, 23) and Malaysia (24). Approximate rates can be estimated for these countries based on hypothetical all-sites rates of 80 per 105 (males) and 105 per 105 (females), and very tentative estimates produced for the area as a whole. Relatively high rates of respiratory cancer and liver cancer are present in all of the above countries, and liver cancer is uniformly more frequent than stomach cancer (except for Singapore Chinese). 15. Middle South Asia Population-based incidence rates are available for the Indian centres of Bombay (4), Pune (4), Ahmedabad,k Madras, (49) and Bangalore (50). These together with the exten- sive relative frequency data from Indian cities published by Jussawala et al. (25) allow fair estimates of the probable order of rates for India as a whole. Relative frequency data are available from a multi-centre study in Pakistan (51), from hospital registries in Bangladesh (26) and Sri Lanka (27), and from pathology series in Iran (28) and Afghanistan (29). Crude all-sites rates of 100 per 105 (males) and 100 per 105 (females) have been assumed in the estimation of site-specific rates. Oral cancer is very common in all these countries, and the overall estimated rates suggest that oesophageal cancer incidence exceeds that of stomach cancer. 16. Western South Asia This area includes the Arab countries of Asia plus Turkey and Israel. Rates for Arab populations are available from the Kuwait cancer registry (30) and Israel cancer registry (4), and relative frequency data are available from Saudi Arabia (31). For Turkey, mortality data collected by Firat (52) suggest all-sites rates of 88 per 105 (males) and 72 per 105 (females). These mortality rates and relative frequency from a national pathology series (32) have been used to estimate rates for Turkey. Weightings for the populations are Arabs 42: Turks 40: Jews 3. The estimated rates for "all sites" from these data appear to be low; in particular, cervix cancer seems to be extremely rare in these populations, whilst lymphoma and lung cancer appear to be relatively frequent. i Woo, Z. H. ET AL. Unpublished interim report, Cancer Register Programme in the Republic of Korea. Seoul, National Medical Centre, 1982. ' Annual report, Rangoon Cancer Registry, 1978-80. k PATEL, T. B. Unpublished data from Ahmedabad Cancer Registry, 1978-81. 172 D. M. PARKIN ET AL. 17. Eastern Europe Data from six cancer registries in Poland, and two others in Hungary and Romania appear in Cancer incidence in five continents (4). National incidence rates have been published for the German Democratic Republic (4), Czechoslovakia (33) and Bulgaria (34). The rates for the area as a whole are appropriately weighted averages. 18. Northern Europe The whole of the population of this area (except for Ireland) is subject to national cancer registration (3, 4, 35), so the figures for this area are likely to be the most accurate of those presented in Table 2. 19. Southern Europe Cancer registry data for centres in Spain, Italy, and Yugoslavia permit estimation of the rates in these countries (4). These can be checked against estimates derived from mortality rates (5) to ensure that rates from single centres are reasonably representative. For Portugal and Greece, incidence rates have been estimated from the mortality data alone. 20. Western Europe There are several cancer registries in France, Switzerland, and the Federal Republic of Germany (4) and mortality rates are available for these countries as well as for Austria, Belgium, Luxembourg, and the Netherlands (5). Reasonable estimates of incidence rates can be made. 21. Australia/New Zealand Cancer registries for the states of New South Wales and South Australia (4) provide data for half the population of Australia, and there is a National Registry in New Zealand (4). Incidence estimates should be fairly accurate for the area as a whole. 22. Melanesia Incidence rates are available for New Caledonia,/ and these have been used together with estimates of incidence for Papua New Guinea based on recent relative frequency data (36) and all-sites rates of 100 per 105 (males) and 110 per 105 (females). As the area is small, any crudity of estimated rates will have little effect on global estimates. 23. Micronesia/Polynesia This is a very small area. Rates have been estimated using relative frequency data in Fiji (37) and for the Polynesian population of Hawaii (4), with estimated all-sites rates of 80 per 105 (males) and 90 per 10 (females). 24. USSR Incidence rates for the USSR are available for a restricted series of sites (38) including the entire group of lymphatic and haematopoietic neoplasms (ICD 200-209). The incidence rates of prostatic and bladder cancer have been estimated from mortality rates, as have the relative proportions of lymphomas and leukaemias (39). There are no data on the probable level of liver cancer, and the rates estimated for Eastern Europe have been used. 1 THEVENOT, H. Unpublished data from Cancer Registry of Nouvelle Caledonie, 1982. WORLDWIDE FREQUENCY OF TWELVE CANCERS RESULTS AND DISCUSSION The estimated annual numbers of new cases of the twelve selected cancers in each of the 24 areas are shown in Table 3. The accuracy of these figures will depend on the validity of the estimates of incidence rates and on the population size. Thus inaccurate estimates of incidence in small areas such as Melanesia or Micronesia will be of little consequence, compared with errors in data from China or Middle South Asia, for example. The effect of some variation in the estimated incidence rates for these latter areas on the world totals for individual tumours can readily be tested; in most instances, the relative ranking of the cancer sites in question is unaffected. The results incidate that globally the most common neoplasm is almost certainly cancer of the stomach, incidence rates of which are high in Europe, East Asia and South America. However, it is interesting to note that the annual number of cases of lung cancer is not far behind, and this appears to be the most common tumour among males. The results presented refer to a single point in time. However, the incidence rates for stomach cancer are declining throughout the world, whilst with a few exceptions those for lung cancer are rising rapidly -the USSR registered a 55'Yo increase in lung cancer cases between 1970 and 1980, for example (40). Thus, the relative position of these two cancers is likely to change in the near future. Indeed, as a result of the energetic expansion of sales of high tar and high nicotine cigarettes in developing countries, very large numbers of lung cancers are now inevitable. For females, lung cancer is now in fifth rank and, given the present trends in female smoking, is likely to move up. Colo-rectal cancer is frequent in Western populations. As these cancers show a steep increase in incidence in relation to age, the crude rates will be especially high where there are many elderly. For all areas, the incidence rates in males and females are very similar. This site probably ranks third or fourth in the world total. Liver cancer seems to give rise to less than half as many tumour cases as lung cancer despite the fact that it is more common in Africa and Eastern Asia (except Japan). Since liver cancer is one of the more difficult tumours to diagnose clinically, its incidence is probably underestimated, although many series will include liver metastases of other cancers along with true liver cancers; so it is questionable whether its incidence has been significantly more underestimated than that of, say, lung cancer. Liver cancer is primarily a disease of developing countries with very young populations so that, although the age- specific rates may be very high, the crude rates (Table 2) are not. China accounts for over 40%o of the global total; Africa, where liver cancer is almost certainly the most common tumour, accounts for only 13 07o. The total numbers of cases of cancer of the oral cavity and pharynx disguise very large differences in the individual sites throughout the world. In Chinese populations, most such tumours are nasopharyngeal, whereas in the Indian subcontinent the great majority are cancers of the oral cavity and related to the chewing of tobacco and/or betel nut. Lip cancer is very frequent in some western populations, and where laryngeal cancer is common, there is also an apparently high rate of cancers classified as "hypopharynx". Perhaps the most striking feature of the results is the frequency, among females, of breast cancer and cervix cancer which, taking account of the fact that these are restricted to a single sex, now present the biggest challenge in the field of cancer prevention. It is only very approximately true that cervix cancer occurs more frequently in developing countries. Breast cancer would appear to be the more frequent cancer overall; it occurs in older age groups compared with cervix cancer, and is more frequent in "Western" populations; thus one half (50%) of the estimated cases are from North America and Europe (excluding USSR) where there are only about 18%7o of the female population of the world. 173 D. M. PARKIN ET AL. LO CV) 00 (0' C; O~O~(I (_N (N, ()0 66- Ur-C) 14 (NO) '-. N (0O 14 - cD co it 0 LC) 0) (N (N CC) C1 - C" - * LO N. 0( -O(N 0 (N (0 Cs 1- 4 0 0D' '- - co lq N CD C5 1D 0 00 0 I 6 rl -, 0 0 Cr-r '-0'- 't- L- 4Lfl0) '- N c) LOC) s00 CN rls a7 CN 0 CD (D N _-0 r- (D CV) 04DO 0 CC) (0( (_N N.. (N (00) > c.i.'. 0)LDn N-CC)'- E_ E 2 _ _ 2 2_ E EE E E E EE E-E E-E E-E E- CN lq (D(N. 0 r( N- (V _: Lfi 00 (C) M cli 06 C(N 0n it 04 0a)C) E -E NNr- 0) -(N M (N(NC CC Cn CN0 )C C It un r a U')0U (D LO -MOLDO It M60c ( rC)(06 (Nr- (N om0)~ 0 )00) E E (D0Cu < u.nY.O o n @~~E oE -c E ' 0Q0 6r '- 174 -(N CC) -01'- co 0 CD (0.00 _N O UC) LO(0 ( NCNs en 'a M U) QQ 0 .r.. E a) E in (D r-0 CY) r- (D 0 - 0 1D Ct) C)-(N'- C 666a (Nr.-0 -0O- (N CV) LO t0)CY) -ON ) (N 07 `. (I Int co 0 0~ - 0 c _ co 0 - OCCu 0 0 cn 0 Ut _C X .C- 0C x 0 O QC X- D (N Q(C 0 - - -OCN 0NO ((0On '-0'- 0 'a 0 -c 0 .C 0 U) M aD M -c 4_ c p (0 ii 0) C C Cu 0 c M Cl) 0 4- 0 0 0 M 0 0 (D U-E ~0 .0 E C') 0 Cu I- .C w1 ._ M 0 :2 0) O M C C o 2 Dn0 Z (n CC .C 4- n 0 0 (0 WORLDWIDE FREQUENCY OF TWELVE CANCERS 00 "t N U) CO CY) 04 _ C r' _ 00 - r co C- r CD Cn o Cn CD)O -D_. w-_ _ _ r, CD CY _- ui r air6 6O 4 -LC 6O 4 r aic6 6ao a i -L 616 6O 6 6OcOON m C1 cn M C) Cv 'I * 1* CD r- U')-CD 04-XCY Y (D CY C - r- CY) q* CY 0 o CD O (D (D C LO r. cn O Cn --N C%4 , C Ul) CO u _-6 64 ( O tD 'O O O- cX (D -O -_ 6c 6 66 6 66 6 ooro N ) _ _ _ M r, o u) o u') M N u') c r, a) N _ M N M ) LO r, c 00 _r a) o u' u) _ _ _ o it I- c- _ 4 6 ai ai 6 6 t6 cN _' c --: csc- ai CD ai 6 6 _' ui tD c- vi c6 cO O6 O6 i6 _- _- _- _ cN _ _ cN _- -N 14 CD CD -N n _ a) o o qt 14 r, N a) CD u) _ a) 00 r, q 0 N to r, c N N lo MO) N _- _ _ u') o U 06 t- cs c6 P 6 ai 6 o) cs-: 6 ei 4 c oo _ 0o66 aiu rW ci _: 6 cN 6 6 6 6 ai _0 0 -_ N C ' _)r- _ NC qt N _ IN -U) O0 0N N U) N N CD -00 a) oo ctM0 _ NUC r&qt N 6 66N C I - . . . . .6 .~.66.6 . .~ .:.: . .~.~ .~4u . ..6r.-.:.6 . .1a _O_U)NX(DOD ItN -Crs-°° 00 0C Cu) f0U)-O c.C- --C - n '- O-)ONCCOODO C v)O eC-c1 vi r e4 0505 r' -6e.j(CD- P ci c' i0;6o :c1- jc'irP -6 -: 666 I 1I u N- l -C - ' - (N E E E E E E E E E E E E +--0~~~~ 0ac/ 0 c, C) C .c )< < 0 cv C0o w U) w CL CL (oL0 0 U ui LU Z C C C C c 0) ac o) -a t _ CM : 0 6 v- Ir- C(4 w .a c W C- (N - M .-a, )Ca C > 0- C(N Cc C) (N _ ) '- CD (D NC a _ CN(D ( U) (N (D - 0 CN(0 0Cs 04 CD4 0 0 ('0 N CD 14 cs 0) tD 1- - 0CD c (6 N - w a, w) 0 .~~~~~~~~~~~~~~. 175 (N 0 0- u) cN a) ul) N q CD 0 u) z COcs 0) CD U) cDCD a, -i Lq CY) cBT 0 C -i:2 c a)U- D. M. PARKIN ET AL. 0 _--_ o r. 0o rN r a) rc Nr a) -NC4 (D N t .00-O r- U O Cn LOrC a)c 0 r-' or- L6 -0 6r oo rCDa -DO 4 oNc -- 6 --'0N ,t t a) N C4 ,* ,* in CD - - CO C14 Cl) LO 04 - M N Cn tO It It a) C) - - P)ICL _- C1 Cl) M r _ C' U) LO Lo CDV MI 4 r- - cn r-s rb Pb o r LO ML- 00)0 C')n(DO CN '- (N c)(OLO CN-C_ wCo LO,*M) N rNLC N 000 mom NN,o')0)N c0-r- CN 0o C; .4.~.~.~.6.~ .~ .; .~ .: .i r. .l .~00'- oo N'-w o o -o N 00'- '-'-N '--(N N (n0)N _-Lr' N CN~ '-_ oN CN CN - M r NC )0 OR OR eo C c _ a)O0 _ ) o OCo N 0o )0 ) (N C -NC CN CDO C'0'-* LNOC bJ 00-L_ )c) '-'O- --C "-O N (D(O LOL_ NLON 'o- OW N O O- n O OO -O-r -O--OM _C1*00 O- * m M LO r NW-r- -O- _ N MC) _ I CD0 - rN 0 CN 00 C') c' '- '- - N C) 4 0 (N 00 t (D C 0 01C - N (N D) C' (N CD OR nL N _- C) I I LO LO CM - - N - CY) u) L c') 0) a) CN CL rN C' ) - CY) LL E E E E E E E E E E E E E Cu M Cu ._ -_ < C < a Mu 0 wU 2 Z I (NC C') Cu 4- C -Cu 0 Cu .i= - E n6 C.)~ ~ ~~ ~ ~ ~ ~~ C'<CE O Cu 0 0 C .0. . 0. 0 Cu- CD 176 z a) C M -C C a) ._) CD Ul) (n C a) _'l tU) = L) < XQAC .T E 0 M C) -C .2 = a3 -j 'a E,M cn X m u in M cn M 9) 'D 0 Cua -cF 0 cu CO a M c .l C L() n 0) 7n C Cu U 0 a,) Cu c (CO C.) U CD 0CO Cu 4i) w Cu -o Cu M- WORLDWIDE FREQUENCY OF TWELVE CANCERS 1 o * CDC a)) CV) N M co-_ Lof- LO - - DCD OJaO) - c Dr-c CODDU U) 0 r-6ci 6 L 0 ) 0 0 -L -w1 D OC~~n _ --c' 0- C 00 t IO )c) _ N C') -- CO CD o C C It o CY) _- _ )') CO _ C_N _ - N _- CIO 04LNU) r-i0C'4 CO COC- -N-nCCv CCo) -O uCIM D _U)0 oI- 6 6 6 6 _c'a'aou 6 uic 6 c_ 6 6_ _ _ _ N _) _ _ 4 0 04 CD CD 4 04 t (D CD 'q 0 CO N O r- t - LO Cn 00 -*u 0) CD _ r- ¢ (D - 00 - CN 0 Cf) t 0) 0) N - 0 CV CY) 0' 0) c a; 4 c -ON CO C') 0 0) cN C) - CY) 0,) co CY) E E E E E E E E E E E .C ci~ 0 0. 4-1 C 0. 00 0 0/ 0 O o - co _ _ U)Cj C C WC C C cocco Cc>6- c U) < U) () ui LLI~~ z C/) < 10) 0) co 0 0) 0) 04 6 in U) U) 0- r- U) C') CY0) N a) 0 _11( N i - d _ _ C') CY)- r---- r *U) - o) P- a a) (N _ -N -U, U- '0 0 C 0 177 D. M. PARKIN ET AL. The figures for cancer of the prostate are probably the least reliable of the twelve sites considered. It is a tumour of elderly males, and is often asymptomatic and found on autopsy; the incidence rates will thus depend on the level of diagnostic services available. It is the most common tumour of males in Jamaica (4) and ranks second in the Black population of the USA (19). Given their young populations, the figures from some African centres (e.g., Liberia (14)) suggest that it is by no means uncommon in that continent also. Bladder cancer is predominantly a tumour of males (ratio 3:1 overall), and is more frequent in the developed countries of Europe and North America than elsewhere, with the exception of those areas where schistosomiasis is endemic, especially Egypt. Incidence rates for tumours ofIymphatic tissue and for leukaemias show relatively small variations between the different areas. Lymphomas are particularly common in African populations where these tumours are only slightly less frequent than cancers of the liver and cervix. The technique used in this analysis is patently crude, and the results for several large portions of the globe are only informed estimates. Very large contributions to the world totals are made by the populous areas of China and South Asia for most of the tumour sites. By considering large areas of the world, half of which have populations in excess of a hundred million (Table 1), it is evident that localized areas of high risk for a given cancer (e.g., of the oesophagus -in Brittany and Normandy in France or around Lake Victoria in Central Africa) will have little influence on the figures in Tables 2 and 3. Such areas are, however, of great interest for the population concerned as well as for the analytical epidemiologist seeking determinants. The rank order of the twelve selected cancers discussed above is given in Table 3 and summarized in Table 4. The consolidated totals at the foot of Table 3 and the rankings given in Table 4 conceal substantial differences between the 24 areas. Five of these areas are considered in Table 5. For females, despite large differences in the crude incidence rates in different areas, breast or cervix cancers are always in first Table 4. Rank order by frequency of occurrence place, the other often being in second or of the twelve selected cancers in males, females, third rank. For males there is somewhat and both sexes more variety and preventive measures will therefore be different in each region. Rank Males Females Both sexes The figures presented are cross-sectional and are poor prognosticators in that they 1 Lung Breast Stomach take into account neither the evolution of 2 Stomach Cervix Lung population structure nor the increases and 3 Colon/ Stomach Breast decreases of certain cancers, trends that are rectum to some extent inevitable owing to the expo- 4 MouthJ Colon/ Colon/ sures already experienced by existing birth pharynx rectum rectum cohorts (41). 5 Prostate Lung Cervix Mention was made above of time-trends 6 Oesophagus Mouth/ Mouth/ in stomach and lung cancer. The evolution pharynx pharynx of several other sites can also be forecast. 7 Liver Oesophagus Oesophagus Rising trends in breast cancer in certain 8 Bladder Lymphatic Liver populations, e.g., in China and Japan 9 Lymphatic Liver Lymphatic where this cancer has hitherto been very 10 Leukaemia Leukaemia Prostate rare, and a fall in uterine cervix cancer in 1 _ Bladder Bladder many populations are likely to strengthen 12 - _ Leukaemia the first position occupied by breast cancer. 178 WORLDWIDE FREQUENCY OF TWELVE CANCERS 179 E - a) c E E CD 0 LL (I 0 ) (+ -j>w -0) E oa U) ~C Cu 0 0 00. -C -J a~~~~~~~~~nC-) 0 U) ~~~~~~~~~~~~~~U) (D m .5 0~~~Eu. > Cu .- ) o co~ 0c >) Cu C -o*0~~~~~~~~~oO r.C 0) 0 E 0 -~0 ci) -i 0 C) ~~~~~~~~~~~~U) .C CuE a) ~~ ~~~XCu a +- -c -i ~~~~~~~~~ ~ E 0 CuE 0 a) 2? 00) a) C Cu LL 0 ( 0 U o CE :C Cu C-) 0) E *2~m C * .C~~~~~~~CuU)C)~ ~~~~~~ .C m ~ Cu 0. C) a) 4-- aaa) 0~~~~~~~~~~~~4-' C)0.~~~~4-3 0 C~~~~ Cu C) C u U CC) 0) 0)*C E C C E a) LI. m J - C 'P E C < a) a) C C) .0 (D E E 00 Cu) C~0 0 Z C E a) CC - _X C~~~~~raCuCl X Cu Cu U)~~~~~~.C CDm a) C) E > Cu a)~~~~~~~~mE U) C a) ~~ ~~~~>? > a C C a) ) CC a) m C C 0 cc CuLo-~ 180 D. M. PARKIN ET AL. In the USA and Northern Europe, cancers of the ovary and corpus uteri are now more frequent than carcinoma of the cervix. With improvements in diagnosis, the number of cancers of the prostate and pancreas will probably assume more importance in global terms. Cancer of the prostate has a high and increasing incidence in the Black populations of North America and the Caribbean and, when looked for, is fairly frequent in Africa. The rapid rise in prostate cancer in the low-risk populations of China and Japan bodes ill. Yet, it is worth emphasizing that many of these cancers are preventable today. Lung cancer and much oropharyngeal, oesophageal and bladder cancer are determined by tobacco smoking and to a lesser extent alcohol consumption. Interventions to control these determinants can yield results within a decade. On the basis of the figures in Table 3, over 1 million cancers which at present occur each year could be avoided by this single step (42). It may be ihat with improved personal hygiene and a rise in the age of marriage, cancers of the cervix uteri will show the same decline in developing countries as already experienced in most of Europe and North America. Cancers of the breast in women and of the large bowel in both sexes deserve the closest attention. Current etiological hypotheses suggest that these (and prostate cancer) are linked with a high fat diet and low intake of dietary fibre. Yet in many areas of the world there is a trend towards the consumption of such diets. Cancer of the stomach occurs most frequently in persons with a low intake o fresh food and vegetables and it is probable that in those areas where this cancer is still common, increased availability of these commodities would accelerate the fall in incidence. In the case of primary liver cancer, the long-term goal of prevention by means of immunization to reduce chronic carriage of hepatitis B virus presents an exciting prospect (43); in the meantime, the reduction of aflatoxin intake may be rewarding. While the causes of malignant lymphoma and leukaemia remain unknown, it is difficult to forecast their evolution. There is no doubt that many developing countries have been complacent about the cancer problem because their priorities lay in the fields of infectious, parasitic and nutritional disease. Elimination of certain infectious diseases is now technically feasible even if the control of several parasitic diseases and of protein-energy malnutrition often calls for large investments. None the less, the demographic changes following reduction of infant mortality are resulting in increasing numbers of persons in the middle and older age groups, many of whom are being encouraged to smoke large quantities of the most dangerous types of cigarettes. The consequences of this one habit are going to require an investment in diagnostic and treatment facilities for cancer as well as heart and lung diseases that may far outweigh the revenue to be obtained from the taxation of tobacco. The adoption of a high fat and low fibre diet, which seems to be conducive to cancers in the large bowel, breast, and prostate, needs critical assessment. For several numerically important cancers, particularly in males, it is still not too late to implement the appropriate preventive measures. By the year 2000 the battle may be lost. REFERENCES 1. DOLL, R. ET AL. Cancer incidence in five continents, Vol. 1. Berlin, Springer-Verlag, 1966. 2. DOLL, R. ET AL. Cancer incidence in five continents, Vol. 2. Berlin, Springer-Verlag, 1970. 3. WATERHOUSE, J. ET AL. Cancer incidence in five continents, Vol. 3. Lyon, IARC, 1976. 4. WATERHOUSE, J. ET AL. Cancer incidence in five continents, Vol. 4. Lyon, IARC, 1982. 5. WORLD HEALTH ORGANIZATION. World Health StatisticsAnnual, 1978-1982. Vitalstatisticsand causes of death. Geneva, 1978-82. 6. PRESTON, S. H. Mortality patterns in. national populations. New York, Academic Press, 1976. WORLDWIDE FREQUENCY OF TWELVE CANCERS 181 7. COULANGES, P. ET AL. Le cancer a Madagascar. Archives de l'Institut Pasteur de Madagascar, 48: 171-212 (1979). 8. DENUES, A. R. T. & MUNZ, W. Malignancies at the hospital of Doctor Albert Schweitzer, Lambarene, Gabon, 1950-65. International journal of cancer, 2: 406-411 (1967). 9. JENSEN, 0. M. ET AL. Cancer in Cameroon: a relative frequency study. Revue dk'pidemiologie et de sante publique, 26: 147-159 (1978). 10. TUYNS, A. J. & RAVISSE, P. Cancer in Brazzaville, the Congo. Journal of the National Cancer Institute, 44: 1121-1127 (1970). 11. YAKER, A. Profil de la morbidite cancereuse en Algerie (1966-1975). Algiers, SNED, 1980. 12. MACRAE, S. M. & COOK, B. V. A retrospective study of the cancer patterns among hospital in-patients in Botswana 1960-72. British journal of cancer, 32: 121-133 (1975). 13. MARTIN, P. M. D. ET AL. The cancer spectrum in Lesotho, 1950-1969. S. African journal of science, 72: 168-171 (1976). 14. SOBO, A. 0. Cancer in Liberia: a review of cases registered from the Liberian Cancer Registry, 1973-1977. Cancer, 49: 1945-1951 (1982). 15. REEVES, W. C. ET AL. Cancer incidence in the Republic of Panama. Journal of the National Cancer Institute, 68: 219-225 (1982). 16. BIANCO, M. Cancer characteristics in Argentina. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 17. Rios DALENZ, J. ET AL. Morbidity from cancer in La Paz, Bolivia. International journal of cancer, 28: 307-314 (1981). 18. OLIVARES, L. Cancer incidence in Peru. In: Aoki, K. et al., ed., Cancer prevention in developing countries, Nagoya, University Press, 1982. 19. SEER: Incidence and mortality data 1973-1977. National Cancer Institute Monograph, 57, (1981). 20. Atlas of cancer mortality in the People's Republic of China. Shanghai, China Map Press, 1981. 21. HANAI, A. & FUJIMOTO, I. Cancer incidence in Japan in 1975 and changes of epidemiological features for cancer in Osaka. National Cancer Institute Monograph, 62: 3-7 (1982). 22. SOERIPTO ET AL. Cancer in Yogyakarta, Indonesia: relative frequencies. British journal of cancer, 36: 141-148 (1977). 23. SUKARD.A, I. D. (1982) Cancer statistics at Dr Sutomo Hospital, Surabaya, Indonesia. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 24. ARMSTRONG, R. W. & AHLUWALIA, H. S. Cancer incidence in Malaysia. National Cancer Institute Monograph, 53: 53-58 (1979). 25. JUSSAWALLA, D. J. & GANGADHARAN, P. Epidemiology of cancer in the Indian subcontinent. Indian journal of cancer, Supplements to Vol. 10-12 (1973-75). 26. RAHIM, M. A. & SARMA, S. K. Cancer statistics of Dhaka Medical College Hospital, 1979. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 27. SIVAYOHAM, S. Cancer in relation to health problems in Sri Lanka. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 28. HABIBI, A. Cancer in Iran. Pathologia et microbiologia., 35: 181-183 (1970). 29. SOBIN, L. H. Cancer in Afghanistan. Cancer, 23: 679-688 (1969). 30. OMAR, Y. T. Cancer incidence in Kuwait. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 31. EL-AKKAD, S. Pattern of cancer in Saudi Arabia. Data published by Shobokshi, 0. In: Aoki, K. et al., ed., Cancer prevention in developing countries. Nagoya, University Press, 1982. 32. BILIR, N. Cancer frequency in Turkey. Kanser, 11: 93-97 (1981). 33. DURKOVSKY, J. ET AL. Cancer control in Czechoslovakia. In: Napalkov, N. P. & Eckhardt, S., ed., Cancer control in the countries of the Councilfor Mutual Economic Assistance. Budapest, Akademiai Kiado, 1982. 34. PoPov, I. ET AL. Cancer control in Bulgaria. In: Napalkov, N. P. & Eckhardt, S., ed., Cancer control in the countries of the Councilfor Mutual Economic Assistance. Budapest, Akademiai Kiado, 1982. 182 D. M. PARKIN ET AL. 35. OFFICE OF POPULATION, CENSUSES AND SURVEYS. Cancer statistics 1975, registrations. London, HMSO, 1981. 36. ATKINSON, L. ET AL. Cancer reporting in Papua New Guinea: 1958-70 and 1971-78. National Cancer Institute Monograph, 62: 65-71 (1983). 37. BOYD, J. T. ET AL. Cancer incidence in Fiji. International journal of epidemiology, 2: 177-187 (1973). 38. NAPALKOV, N. P. ET AL. Cancer incidence in the USSR in 1975. Voprosy onkologii, 24(6): 8-37 (1978). 39. NAPALKOV, N. P. ET AL. Mortality rates from cancer in the USSR. Voprosy onkologii, 23: 3-12 (1977). 40. NAPALKOV, N. P. ET AL. Incidence of malignant tumours in the USSR between 1970 and 1980. Voprosy onkologii, 28(10): 26-71 (1982). 41. HAKAMA, M. Projection of cancer incidence: experience and some results in Finland. World Health Statistics Quarterly, 33: 228-240 (1980). 42. DOLL, R. & PETO, R. The causes of cancer. Oxford, Oxford University Press, 1981. 43. WHO Technical Report Series, No. 691, 1983 (Prevention of liver cancer: report of a WHO meeting). 44. ABOUL NASR, A. L. ET AL. Cancer registration in 1975-78. Cairo, The Cancer Registry for the Metropolitan Cairo Area. 45. Sudan Cancer Registry data. In: Annual statistical report, 1978. Khartoum, Sudan Ministry of Health, 1978. 46. DA SILVA, M. G. C., ed., Cancer in Forta.eza, 1978-80. Fortaleza, Cancer Institute and Registry of Ceara, 1982. 47. Cancer registration in Metro Manila. Statistical report 1975-77. Manila, Philippine Cancer Society. 48. Cancer statistics, 1975, 1976. Bangkok, National Cancer Institute, Ministry of Public Health. 49. MADRAS METROPOLITAN TUMOUR REGISTRY. Annual reportfor 1982. Madras, Cancer Institute, 1982. 50. NATIONAL CANCER REGISTRY PROJECT. Population-based cancer registry: annual report, 1982. Bangalore, Kidwai Memorial Institute of Oncology. 51. PAKISTAN, MEDiCAI. RESEARCH COUNCIL. Malignant tumours: report of a multicentre study, 1977-80. Karachi, Saad Publications, 1982. 52. FIRAT, D. Cancer statistics in Turkey. Ankara, Turkish Association for Cancer Research and Control, 1982.