Survival Rates of Childhood Cancer Patients in Osaka, Japan

Jpn J Clin Oncol 2004;34(1)50 54 Epidemiology Note Survival Rates of Childhood Cancer Patients in Osaka, Japan Wakiko Ajiki, Hideaki Tsukuma and Akira Oshima Department of Cancer Control and Statistics, Osaka Medical Center for Cancer and Cardiovascular Diseases, Osaka, Japan Received October 28, 2003; accepted November 27, 2003 Background: Survival of childhood cancer is often described in specialized hospitals and/or institutions, but not in the general population. Methods: The trends and patterns of childhood cancer survival during 20 years were investigated using the Osaka Cancer Registry s data. The study subjects were 3443 reported cases diagnosed as having the first primary cancer between 1975 and 1994 at the ages of 0 14 years and who lived in Osaka Prefecture, excluding Osaka City. All childhood cancers were classified into 12 diagnostic groups according to the International Classification of Childhood Cancer based on the histology of the cancer. The 5 year relative survival was calculated by the Ederer II method. Results: The 5-year relative survival was 58% for all cancers. The survivals of retinoblastoma and kidney tumor were as high as 89 and 72%, respectively, whereas those of leukemia and liver tumor were as low as 45 and 47%, respectively. When divided into the first and second half-periods, the survival for all cancers was markedly improved from 48 to 68%. The survivals, especially for with leukemia, lymphoma, neuroblastoma, hepatoma and gonadal or embryonal tumors, were enhanced by >20% between the first and second half-periods. When these results were compared with those reported in England and Wales and the USA, however, the survivals for many diagnostic groups in Osaka were lower, except that for neuroblastoma. Conclusions: The survivals of childhood cancer have been markedly improved in Japan, but this finding suggested that the introduction and practice of chemotherapy was insufficient. Key words: childhood cancer survival population-based cancer registry INTRODUCTION Childhood cancers, defined as malignant tumors diagnosed before the age of 15 years, have been significant causes of death in children. Death due to cancer in childhood has shown a tendency to decrease, as judged from both the number of deaths and the mortality rate. For example, when we analyzed data for Osaka for 24 years, from 1972 to 1995, after dividing these years into four periods of 6 years each, the mortality rates per one million population of children decreased, that is, the rate for the first period was 48.5 and that for the last period was as low as 32.6. In contrast, the crude incidence rates for cancer For reprints and all correspondence: Wakiko Ajiki, Department of Cancer Control and Statistics, Osaka Medical Center for Cancer and Cardiovascular Diseases, 3 3 Nakamichi 1-chome, Higashinari-ku, Osaka 537-8511, Japan. E-mail: ajiki-mi@umin.ac.jp during these four periods increased, i.e. from 109.7 to 134.0. Therefore, the decrease in the death rate of childhood cancer in these periods has been suggested to be due to the effects of improvements in diagnosis and therapy. Results of clinical trials or reports from certain institutions have shown that the survival rates of childhood cancer were markedly enhanced with improvements in therapeutic methods. However, it is possible that results obtained from subjected to clinical trials or those hospitalized in specific institutions may not represent a trend in the general population. Nearly 30% underwent treatments in hospitals other than nine specific institutions (five medical universities and four specific hospitals) in Osaka during 1989 1999 (1). In addition, since the incidence rates for childhood cancers are expected to be low, variation by chance will greatly affect a trend when data in small population-based cancer registries are used. To analyze chronological changes in survival 2004 Foundation for Promotion of Cancer Research

Jpn J Clin Oncol 2004;34(1) 51 rates of childhood cancer, it will be necessary to use large population-based cancer registries. In 1995, the authors (2) reported survival of childhood cancer based on data from the Osaka Cancer Registry (population: 8.8 million at the 2000 census), one of the largest population-based cancer registries in the world, in which the study subjects were diagnosed as having cancer between 1975 and 1984. In this study, we extended the observation period from 1975 to 1994 and analyzed trends in survival rates of childhood cancer during 20 years. PATIENTS AND METHODS PATIENTS The number of study subjects was 3443; 3503 who were living in Osaka Prefecture, excluding Osaka City, were diagnosed as having a first primary cancer between 1975 and 1994 at the ages of 0 14 years, but 60 of these cases (1.7%) were excluded because only death certificates were available (called death certificate only cases or DCO). The Osaka Cancer Registry included information obtained from (i) reports from medical institutions in Osaka Prefecture, (ii) death records of inhabitants of Osaka Prefecture referring to neoplasms and (iii) autopsy records of medical institutions in Osaka Prefecture (originally compiled in the Autopsy Records of the Japanese Society of Pathology). Concerning childhood cancers, we further used (iv) information about cancer cases in Osaka Prefecture extracted from the Nationwide Registry of Childhood Cancers of the Society for Protection of Children with Cancer (Kansai branch), (v) records of cancer in Osaka Prefecture extracted from the Childhood Cancer Registry of the committee for malignant tumors of the Japanese Society for Pediatric Surgery and (vi) information from application forms used in the Research Project for Pediatric Chronic Severe Diseases in order to have accurate information on. PROGNOS INVESTIGATION The Osaka Cancer Registry tried to trace the prognoses of in three ways as follows. 1. Collation of cancer death information in Osaka Prefecture: death information described as cancer, including cancersuspected cases or neoplasms with unclear properties, was checked against the registered cases, based on a combination of five indicators as follows: date of birth, gender, first Chinese character of family name, address (city/ward/town/village and street/district number) and primary site. 2. Collation of total death information in Osaka Prefecture: a magnetic death file of vital statistics was checked against the registered cases without a death record. In cases showing complete concordance of three indicators, i.e. date of birth, address (city, town or village) and gender, the death record form was examined as to whether or not it was a true match. 3. Check by referring to resident cards: this check was done for subjects diagnosed after 1975 in collaboration with health centers in Higashi Osaka City, Sakai City and Osaka Prefecture. Subjects who had been living in Osaka City were excluded from the survival analysis, because collaboration with Osaka City started with diagnosed after 1993. When we obtained information on moving the home address of subjects, we asked city, ward, town or village offices all over Japan for information on resident cards. The prognoses of were confirmed and recorded as of 5 years after diagnoses. Throughout the observation periods, the vital status of 111 cases (3.2%) was unknown as of 5 years after diagnoses. The proportion of prognosis-unknown cases decreased from 4.6% for the first half-period to 1.7% in the second half-period, hence the accuracy of the prognosis investigation was considered to have improved during this period. CLASSIFICATION OF CHILDHOOD CANCER It is difficult to clarify characteristics of childhood cancer if we use ICD-10, which has been used for usual statistical analyses for cancer, as childhood cancer will develop at various sites and show various types of histology. Therefore, the introduction of the International Classification of Childhood Cancer 1996 has been proposed (3) and thus we classified all childhood cancers into 12 diagnostic groups using this classification. MEASUREMENT OF SURVIVAL RATES Five-year cumulative survival rates were calculated using the Kaplan Meier method and expected survival rates were calculated using the cohort survival table by the Ederer II method (4). The former was divided by the latter and 5-year relative survival rates were obtained. The total observation period of 20 years was divided into the first 10 years (1779 cases) and the second 10 years (1664 cases) and the survival rates were compared between these two periods. Diagnostic groups consisting of many cases of were divided into four periods (I IV) of 5 years each. COMPARISON WITH STATISTICS FOR ENGLAND AND WALES OR THE USA Five-year relative survival rates in the second 10 year period were compared with those for England and Wales (5) and ER in the USA (6). RESULTS Fig. 1 shows the relative distribution of diagnostic groups for the total observation period. Among childhood cancers, leukemia (I) was the most prevalent, followed by neoplasms in the central nervous system and intracranial and intraspinal neoplasms (III), lymphoma and tumors of reticuloendothelial

52 Survival of childhood cancers in Osaka Figure 1. Distribution according to diagnostic groups of childhood cancer. The were living in Osaka Prefecture and diagnosed with a first primary cancer between 1975 and 1994. Table 1. Five-year relative survival rates (%) of childhood cancer in Osaka, Japan, and their standard errors (), according to observation periods Diagnosis group Total period First period Second period Z-value 1975 94 1975 84 1985 94 Total childhood cancers 3443 57.9 0.9 1779 48.0 1.2 1664 68.2 1.2 11.9** I. Leukemia 1034 45.2 1.6 569 32.4 2.0 465 60.4 2.3 9.2** Lymphatic leukemia 665 53.1 2.0 351 39.9 2.7 314 67.6 2.7 7.3** Acute non-lymphatic leukemia 227 26.4 3.0 138 16.0 3.2 89 42.2 5.3 4.2** II. Lymphoma and reticuloendothelial system 379 55.7 2.6 202 45.7 3.5 177 67.0 3.6 4.2** neoplasm Non-Hodgkin lymphoma 279 53.7 3.0 151 43.2 4.1 128 66.0 4.2 3.9** III. Central nervous system tumor and intracranial 709 60.4 1.9 391 54.6 2.6 318 67.5 2.7 3.4 and intraspinal neoplasm IV. Sympathetic nervous system tumor 295 61.9 2.9 100 35.7 5.0 195 74.6 3.2 6.6** Neuroblastoma 293 61.8 2.9 99 35.7 5.0 194 74.5 3.2 6.5** V. Retinoblastoma 122 89.2 2.9 77 86.8 4.0 45 93.3 3.8 1.2 VI. Kidney tumor 122 71.8 4.2 63 68.6 6.0 59 75.2 5.8 0.8 VII. Liver tumor 87 46.7 5.4 41 35.2 7.6 46 56.7 7.3 2.0* VIII. Malignant bone tumor 127 52.8 4.4 55 47.3 6.7 72 57.0 5.8 1.1 IX. Soft tissue sarcoma 190 62.1 3.6 88 61.2 5.4 102 62.9 4.8 0.2 X. Embryonal and gonadal tumor 258 57.4 2.6 127 44.6 4.2 131 75.3 3.0 5.9** XI. Carcinoma and other malignant epithelial tumor 63 56.1 5.9 32 42.3 8.8 31 75.3 7.5 2.9** XII. Others and unclassified tumor 57 56.1 6.1 34 42.3 8.7 23 75.4 5.9 3.1** *P < 0.05; **P < 0.01.

Jpn J Clin Oncol 2004;34(1) 53 Table 2. Comparison of 5-year relative survival rates (%) of childhood cancer in Osaka, Japan, with those in England and Wales and the USA Diagnosis group Osaka* England and Wales ER, USA 1985 94 1986 90 1985 97 Total childhood cancers 68.2 (65.8 70.6) 74.0 Leukemia 60.4 (55.9 64.9) 69.0 73.3 Non-Hodgkin lymphoma 66.0 (57.8 74.2) 76.0 75.6 Neuroblastoma 74.5 (68.2 80.8) 67.0 Retinoblastoma 93.3 (85.9 100.7) 93.0 93.8 Kidney tumor 75.2 (63.8 86.6) 84.0 90.5 Liver tumor 56.7 (42.4 71.0) 43.0 58.5 Malignant bone tumor 57.0 (45.6 68.4) 51.0 66.9 Soft tissue sarcoma 62.9 (53.5 72.3) 66.0 73.8 Embryonal and gonadal neoplasm 75.3 (69.4 81.2) 83.0 85.2 *95% confidence intervals in parentheses. From Ref. 5, but the survival of neuroblastoma was from the Nationa1 Registry of Childhood Tumors in Great Britain, 1992 96. Dashes indicate data not given. From Ref. 6. system (II) and then tumors of the sympathetic nervous system (neuroblastoma) (IV). As for changes during the observation period, marked changes were not detected in the distribution of diagnostic groups, except an increase in neuroblastoma (from 5.6 to 11.7%). Table 1 shows the numbers and 5-year relative survival rates with standard errors observed in the total period or during different observation periods. Diagnostic groups in which significant increases in the survival rates were observed are marked with asterisks. Analysis for the total period showed that the 5-year relative survival rate of all childhood cancer was 58%. Calculated according to different diagnostic groups, the rates for retinoblastoma and kidney tumors were as high as 89 and 72%, respectively, whereas those of leukemia and liver tumors were as low as 45 and 47%, respectively. When the first and second 10 year periods were compared, the 5 year relative survival rate of total childhood cancer was markedly improved during these periods, i.e. from 48 to 68%. When analyzed according to different diagnostic groups, the survival rates were significantly improved in many groups. Especially, more than 20% increases in the 5-year relative survival rates were noticed for leukemia, lymphoma, neural tumors, hepatoma and gonadal or embryonal tumors. When we analyzed the survival rates of five diagnostic groups during four periods of 5 years each, from period I to IV, there was >30% improvement (from 36 to 69%) in lymphoma, >40% improvement in leukemia, neuroblastoma and hepatoma (from 22 to 66%, from 41 to 83% and from 29 to 75%, respectively) and nearly a 60% improvement in gonadal or embryonal tumors (from 29% to 88%). Table 2 shows comparatively the survival rates of childhood cancer in Osaka Prefecture, England and Wales and the USA. The survival rates in Osaka were in general lower than those reported from England and Wales or the USA, with the exception of neuroblastoma. DISCUSSION We analyzed trends and changes in the survival rates of childhood cancer for 20 years using the Osaka Cancer Registry. Since information on childhood cancer has been obtained from various sources in the Osaka Cancer Registry, the registration is considered highly accurate. In this study, the proportion of DCO cases was as low as 1.7%, so the effects not included in the statistical analyses were considered negligible. Information on prognoses was also highly accurate and the proportion whose prognoses was not known after 5 years was as low as 3.2%. Therefore, this study describes the clinical conditions of childhood cancer in Osaka Prefecture as accurately as possible. The survival rates of not only with childhood cancer who had undergone specified treatment or were treated in hospitals specialized in childhood cancer, but also who had been registered in the Osaka Cancer Registry were markedly improved during the 20-year study period. The improvement in survival rates was observed in many diagnostic groups and this is thought to be due to the improvement of diagnostic techniques and the development of interdisciplinary therapy. Clinical trials have reported the progress in treatment, mainly chemotherapy, hence the improvements in treatment and its practice were thought to affect greatly the increases in the survival rates. In addition, early detection of neuroblastoma, due to introduction of a mass screening system for neuroblastoma in infants in 1984 and its practice on a large scale, and improvements in examination methods may have greatly affected the survival rates. Still, the survival rates of many diagnostic groups excluding neuroblastoma in Osaka Prefecture were lower than those

54 Survival of childhood cancers in Osaka reported in England and Wales or the USA. In addition, the 5- year survival rate of testicular tumors, which has been reported to be >90% in population-based cancer registries in Europe and the USA owing to the development of chemotherapy, was <80%, although this rate was similar to those in Europe and USA (7) for cancer-specialized hospitals. Similarly, it was suggested that the introduction and implementation of chemotherapy were not sufficient even in diagnostic groups of cancers that are expected to be cured by chemotherapy. The improvement in the survival rate with childhood cancer, especially leukemia, in the UK was reported to be due to centralization of treatment hospitals in specialized medical institutions. It is also important in Osaka to examine and compare trends in the therapy with childhood cancer and their effects on survival rates according to different types of treatment institutions and numbers treated in them. The survival rate of neuroblastoma in Osaka was markedly higher than that in England and Wales or the USA. In Japan, the mass screening of infants for neuroblastoma started in 1984 as a national project. Over-diagnoses of neuroblastoma due to this project have been thought to greatly affect an apparent increase in its incidence and survival (8). The survival rate of neuroblastoma as compared with that of England and Wales or the USA should be evaluated carefully, taking these overdiagnoses into consideration. It may be possible that the distribution of the histology pattern is greatly different from that of England and Wales or the USA even in the same diagnostic group. In this study, the 12 major diagnostic groups were compared; in the future it will be necessary to analyze the survival rate of each group based on the precise disease entity and using larger numbers of cases. In this study, as many as possible comparable data from England and Wales and the USA were selected from published reports. For international comparison of survival rates, it is desirable to use data obtained in the same diagnostic years and by the same criteria for diagnostic groups. Currently an international collaborative study on survival of cancer, the CONCORD study of the European Union and North America, has started and the Osaka Cancer Registry will take part in this study. Through this type of collaborative study, the differences in the survival rates and their causes will be clarified. References 1. Ajiki W, Tsukuma H, Oshima A. Step toward the realization of a population-based national childhood cancer registry in Japan: recommendation from population-based cancer registry. Jpn J Pediatr Oncol 2002;39:150 8 (in Japanese). 2. Ajiki W, Hanai A, Tsukuma H, Hiyama T, Fujimoto I. s of childhood cancer in Osaka, Japan, 1975 1984. Jpn J Cancer Res 1995;86:13 20. 3. Karamárová E, Stiller CA, Ferlay J, Parkin DM, Draper G.J, Michaelis J, et al., editors. International Classification of Childhood Cancer 1996. IARC Technical Report No. 29. Lyon: IARC 1996. 4. Ajiki W, Matsuda T, Sato Y, Fujita M, Yamazaki S, Murakami R, et al. A standard method of calculating relative survival rates in population-based cancer registries. Jpn J Cancer Clin 1998;44:981 93 (in Japanese with English abstract). 5. Coleman MP, Babb P, Sloggett A, Quinn M, De Stavola B. Socioeconomic inequalities in cancer survival in England and Wales. Cancer 2001; 91(Suppl 1):208 16. 6. ER Cancer Statistics Review 1973 1998. Bethesda, MD: National Cancer Institute (http://seer.cancer.gov/csr/). 7. Oshima A, Kitagawa T, Ajiki W, Tsukuma H, Takenaka S, Iura A. Survival of testicular cancer in Osaka, Japan. Jpn J Clin Oncol 2001; 31:438 43. 8. Ajiki W, Tsukuma H, Oshima A, Kawa K. Effects of mass screening for neuroblastoma on incidence, mortality and survival rates in Osaka, Japan. Cancer Causes Control 1998;9:631 6.