1477
EDITORIALS
Childhood leukaemia: an infectious disease? known causes of leukaemia in children--exposure to X rays in utero and ionising radiation-and these are unlikely to account for more than 8% of cases.! The disease is also associated with genetic abnormalities, of which Down’s syndrome is best known, but this applies to only 3% of children with leukaemia. Over 80% of childhood leukaemia is therefore unexplained. Might viral infections be an important causal factor? In adults, human T-lymphotropic virus type I (HTLV-1) is the one known viral cause of human leukaemia and the epidemiological features of this disease are not fully characterised. HTLV-1 infection clusters geographically, being rare outside the Caribbean and Japan; the prevalence of serological markers increases with age and the risk of leukaemia is low (perhaps only 1 in 1000 persistently infected individuals eventually manifest leukaemia2). Mode of transmission is unclear; early infection may occur without antibodies, as has been described for murine leukaemia virus.3 Although HTLV-1 causes considerably less than 1 % of all leukaemia, the little we know about the association already illustrates that simple models may not be sufficient for studying viral
There
are
only
two
leukaemogenesis. Many viruses, mostly retroviruses, cause leukaemia in animals; the best studied are those of mice, cats, and cattle. In mice, the pattern of leukaemia virus transmission differs in laboratory and wild populations. Thus laboratory animals seem to have high levels of vertical transmission, with integration of viral DNA into the host genome, whereas in wild mice transmission is mainly perinatal via infected maternal milk.3 Wild mice show clustering of infection by region and by family. The viruses do not produce overt disease during the acute infection, but persist in the host and induce disease many months or years later. Bovine leukaemia virus is strongly associated with lymphocytes. Cell transfer is needed for transmission and may occur
transplacentally, iatrogenically, or by arthropods (flies can transfer enough cells to infect).4 In cats, age at infection is an important determinant of persistence: virtually all persistently infected cats die within 5 years.s Dose of infecting organism is also important in determining persistence, and crowding of animals in a household may enhance the risk of infection.6,7 Some of the animal leukaemia viruses may also cause non-neoplastic diseaseseg, feline leukaemia virus may cause glomerulonephritis.’ Features of animal leukaemia viruses can be used as a model of a putative childhood leukaemia virus (CLV): (a) acute infection is insignificant clinically; (b) disease is related to persistent infection; and (c) persistence is determined by
and/or by a large dose of infectious of This model viral leukaemia-from infection to organism. to disease-resembles the association between persistence B virus and hepatitis hepatocellular carcinoma, in which various risk factors may act at different stages of the process. For hepatitis B and hepatocellular carcinoma, child-to-child transmission seems to be related to crowding of children;8 persistence of infection is closely related to age at infection, especially with infection during the first year of life.9 For leukaemia, the probability of contact between infectious and susceptible individuals might likewise be related to household crowding of young children, analogous to the observations in cats.7 Family structure and use of day centres for young children could contribute to crowding. Degree of contact may play a part in determining persistence, as in feline leukaemia. Thus crowded animals or children may be more likely to become infected and to receive a large dose of infectious organism. young age at infection
step is to apply this model to the that most leukaemia in children is hypothesis attributable to a common childhood infection. On this basis one would predict that high levels of child contact are a powerful predictor of infection at a young age. Urban populations may be large enough to maintain infection in an endemic state, with a young median age at infections whereas in rural areas infections may be present only intermittently, with higher median age at infection. Age at infection would tend to be lower in children of lower socioeconomic status and in those who have older siblings. However, these predictions are not borne out by the facts. Childhood leukaemia is associated with geographical areas of high socioeconomic status," as confirmed by Alexander and colleagues in this issue (p 1461), and several studies have shown that the socioeconomic status of affected individuals is higher than that of the general population.12-14 This socioeconomic correlation is not apparent in mortality figures for England and Wales, but with the improved treatment of the disease there are survival biases in these data. The disease seems to be more frequent in firstbom16,17 and single children,t8 although this finding is not universal. 14 The suggestion of Alexander et al that leukaemia is more common in isolated communities of higher socioeconomic status contrasts with the observations of Cook-Mozaffari and co-workers," who did not find an association with rural areas. Perhaps the discrepancy can be explained by the way in which a rural area was defined. Knox reported an excess incidence of leukaemia in urban areas for children over 6 years ;19 Alexander et al show that the effect of rural isolation is much stronger in those aged 1-7. The
next
1478
What about antenatal transmission of CLV? Knox2o pointed out that "mothers of affected children will be characterised by relative affluence, small sibships, good housing, and uncrowded neighbourhoods". When these predictions were investigated in a study of childhood leukaemia21 they were found not to fit. However, the researchers studied children whose birth dates extended over a long period and had to take account of the changes in family structure over time. Both chickenpox and influenza in pregnancy have been associated with an increased risk of leukaemia in the child. 2223 Since many common viral infections show seasonal variation one might expect such variation in dates of birth of affected children. A small summer excess in births has been reported2425 (which might be consistent with in-utero exposure during the winter or spring months), but these results are not persuasive. Co-twins with leukaemia have a single cell clone as the source of their disease,’ which implies that the disease arose in utero and spread from one twin to the other by mixing of placental circulation. The Knox model may not be appropriate to test for an infectious aetiology if the maternal infections were persistent; in bovine leukaemia, for example, 15% of infections occur transplacentally from persistently infected cows.
and co-workerS26-27 have examined geographical variation in leukaemia with the added dimension of change over time. They suggest that some leukaemia clusters may be associated with small rural communities that are expanding rapidly as a result of in-migration. This proposal was tested first in the new town of Glenrothes in Scotland and subsequently in fourteen new towns throughout Britain. The findings were consistent with the hypothesis, in that there was an excess of leukaemia in the rural towns. From these findings one would expect such communities to have low rates of leukaemia before growth whereas the results of Alexander et al suggest that these communities have higher than average background rates of disease. The short time lag between population growth and increase in leukaemia suggested by Kinlen’s data implies that the leukaemia must be caused by an effect on the child or mother during pregnancy and not by something that the mother carries with her from childhood. Periodic outbreaks of infection in rural populations usually lead Kinlen
to higher age at infection, so the postnatal virus model would not be satisfied.
persistent
Another way of looking for evidence of infection is to search for time-space clustering, but many noninfectious risk factors also aggregate in space and time. Application of space-time cluster methods to the proposed CLV is hampered by the fact that one does not know at what age exposure might have occurred and therefore at what time and place in an affected child’s life one should look for clustering. Moreover, if a large proportion of CLV infections were subclinical and persistent, but still infectious, clustering might be
very weak. Four studies have found such clustering of young children with leukaemia when onset of
used as the reference point.1128-30 Clustering was strongest in children under the age of 6 years and there was no evidence for clustering of dates of birth. However, many studies have not shown an association. Smith 31 concluded that the clustering found could have arisen by chance. Thus, there are paradoxes in the data, and dilemmas in interpretation, but little good evidence of
symptoms
was
infectious aetiology for an appreciable proportion of childhood leukaemia. It will be important to examine separately the observed differences in the epidemiological patterns in children under and over 7 years, since they may represent different aetiologies. That the major cell type of leukaemia under 7 years is the B cell whereas over this age there is more heterogeneity of cell types accords with this view. Moreover, childhood leukaemia peaks at 3 years, and the peak has appeared in this century. One possibility would be to look at morbidity-eg, glomerulonephritis-in siblings of affected children and controls, and it might be valuable to re-examine the characteristics of affected children and controls for evidence of prenatal infection. an
1. Doll R. The epidemiology of childhood 1989; 152: 341-51.
leukaemia. J R Statist Soc Ser A
2. Tajima K, Tominga S, Suchi T. Malignant lymphomas in Japan: epidemiological analysis on adult T-cell leukaemia/lymphoma. Haematol Oncol 1986; 4: 31-44. 3. Gardner MB. Naturally occurring leukaemia viruses in wild mice: how good a model for humans? Cancer Surv 1987; 6: 55-71. 4. Burny A, Cleuter Y, Kettman R, et al. Bovine leukaemia: facts and hypotheses derived from the study of an infectious cancer. Cancer Surv 1987; 6: 139-59. 5. McClelland AJ,
Hardy WD, Zuckerman EE. Prognosis of healthy feline leukemia virus infected cats. In: Hardy WD Jr, Essex M, McClelland AJ, eds. Feline leukaemia virus. New York: Elsevier, 1980: 121-31. 6. Hoover EA, Olsen RG, Hardy WD Jr, Schaller JP, Mathes LE. Feline leukemia virus infection: age-related variation in response of cats to experimental infection. JNCI 1976; 57: 365-69. 7. Francis DP, Essex M, Jakowski RM, Cotter SM, Lerer TJ, Hardy WD Jr. Increased risk for lymphoma and glomerulonephritis in a closed population of cats exposed to feline leukemia virus. Am J Epidemiol 1980; 111: 337-46. 8. Whittle HC, Bradley AK, McLauchlan, et al. Hepatitis B virus infection in two Gambian villages. Lancet 1983; i: 1203-06. 9. Coursaget P, Yvonnet B, Chotard J, et al. Age and sex related study of hepatitis B virus chronic carrier state in infants from an endemic area (Senegal). J Med Virol 1987; 22: 1-5. 10. Anderson RM, May RM. Directly transmitted infectious diseases: control by vaccination. Science 1982; 215: 1053-60. 11. Cook-Mozaffari P, Darby SC, Doll R, et al. Geographical variation in mortality from leukaemia and other cancers in England and Wales in relation to proximity to nuclear installations, 1969-78. Br J Cancer 1989; 59: 476-85. 12. Sanders BM, White GC, Draper GJ. Occupations of fathers of children dying from neoplasms. J Epidemiol Commun Health 1981; 35: 245-50. 13. McWhirter WR. The relationship of incidence of childhood lymphoblastic leukaemia to social class. Br J Cancer 1982; 46: 640-45. 14. Fasal E, Jackson EW, Klauber MR. Birth characteristics and leukemia in childhood JNCI 1971; 47: 510-09. 15. Office of Population, Censuses and Surveys. Occupational mortality, 1979-80, 1982-83. Childhood supplement series DS no 8. London: HM Stationery Office, 1988. 16. Stewart A, Webb J, Hewitt D. A survey of childhood malignancies. Br Med J 1958; i: 1495-508. 17. Stark CR, Mantel N. Effects of maternal age and birth order on the risk of mongolism and leukemia. JNCI 1966; 37: 687-98. 18. van Steensel-Moll HA, Valkenberg HA, van Zanen GE. Childhood
1479
leukemia and infectious diseases in the first year of life: a register based control study. Am J Epidemiol 1986; 124: 590-94. 19. Knox EG. Epidemiology of childhood leukaemia in Northumberland and Durham. Br J Prev Soc Med 1964; 18: 17-24. 20. Knox EG. Epidemiology of prenatal infections: an extension of the congenital rubella model. Stat Med 1983; 2: 1-12. 21. Knox EG, Stewart AM, Kneale GW. Foetal infection, childhood leukaemia and cancer. Br J Cancer 1983; 48: 849-52. 22. Frederick J, Alberman EA. Reported influenza in pregnancy and subsequent cancer in the child. Br Med J 1972; ii: 485-88. 23. Fine PEM, Adelstein AM, Snowman J, Clarkson JA, Evans SM. Long term effects of exposure to viral infections in utero. Br Med J 1985; 290: 509-11. 24. Bailar JC III, Gurian JM. Month of birth and cancer mortality. JNCI case
1964; 33: 237-42.
CR, Mantel N. Temporal-spatial distribution of birth dates for Michigan children with leukemia. Cancer Res 1967; 27: 1749-55.
25. Stark
26. Kinlen L. Evidence for an infective cause of childhood leukaemia: comparison of a Scottish new town with nuclear reprocessing sites in Britain. Lancet 1988; ii: 1323-27. 27. Kinlen LJ, Clarke K, Hudson C. Evidence from population mixing in British New Towns 1946-85 of an infective basis for childhood leukaemia. Lancet 1990; 336: 577-82. 28. Meighan SS, Knox G. Leukemia in childhood: epidemiology in Oregon. Cancer 1965; 18: 811-14. 29. Till MM, Hardisty RM, Pike MC, et al. Childhood leukaemia in Greater London: a search for evidence of clustering. Br Med J 1967; iii: 755-58. 30. Gunz FW, Spears GFS. Distribution of acute leukaemia in time and space: studies in New Zealand. Br Med J 1968; 4: 604-08. 31. Smith PG. Spatial and temporal clustering. In: Schottenfeld P, Fraumeni JF, eds. Cancer epidemiology and prevention. Philadelphia: Saunders, 1982: 391-408.
Perinatal prophylaxis of tuberculosis There is general agreement in many countries about the way in which tuberculosis should be treated. Much credit for this accord must go to the UK Medical Research Council for their efforts in coordinating prospective clinical trials in a various locations.1 By contrast, there is no such agreement about perinatal prophylaxis against the disease. Numerous management protocols have been developed in countries with different prevalence rates of pulmonary tuberculosis, but few studies have successfully examined the efficacy of these regimens. In areas of high prevalence, prevention of neonatal disease begins with maternal screening. In the USA, mothers from high-risk groups undergo tuberculin testing and chest radiographs are taken of those with reactions greater than Heaf grade 2 or of more than 5 mm induration with the Mantoux test. In some countries routine chest radiographs are done, with the necessary shielding. Active disease is then treated with rifampicin, isoniazid, and pyrazinamide.2 For sensitive strains of Mycobacterium tuberculosis the patient is deemed to be non-infectious within 2 weeks of starting therapy, although numbers of viable organisms are greatly reduced after only 24 hours. Some experts recommend that individuals with evidence of previous disease but no history of adequate therapy should start prophylactic isoniazid from the second trimester for 9-12 months.That this form of secondary prophylaxis may not be accepted as adequate therapy by North American immigration officials has encouraged doctors in some countries to use full therapeutic regimens for such patients.
Isolation of newborn babies is often advised, despite there being no clear evidence of benefit.3 In developing countries, depriving a child of breast milk may diminish its chances of survival. If maternal compliance with treatment cannot be guaranteed and there is concern about protection before neonatal BCG vaccination becomes effective, the baby can be given isoniazid for as long as the mother remains smear-positive, although there is little scientific support for this use of the drug. There is no need for patients to discontinue breastfeeding when they are receiving antituberculous treatment since only small amounts of the drugs pass into the milk.s Moreover, breast milk may enhance the immunogenicity of BCG given neonatally.6 The immediate family should be screened to detect any additional sources of infection. Management of babies born to smear-positive mothers differs in the UK and the USA. Both American and British thoracic societies recommend 3 months of prophylactic isoniazid, but opinion is divided about the usefulness of BCG vaccination after this period. BCG is favoured in the UK, but in the USA the only recipients are those whose mothers have multiresistant M tuberculosis or may not have complied with therapy. In both countries infants are reviewed at 3 months, when a tuberculin test is done and a chest radiograph is taken. Isoniazid is discontinued in children with clear chest radiographs and negative tuberculin tests. Positive tuberculin reactors are investigated for active disease and treated if necessary;.1 otherwise isoniazid prophylaxis is continued for 9-12 months. If the mother remains smear-positive at the 3-month review the child’s isoniazid prophylaxis is continued until the mother’s sputum becomes negative.7,8 In countries that practice neonatal BCG vaccination, regular review is appropriate with treatment if active disease develops. Isoniazid is undoubtedly effective in primary prophylaxis-protection generally varies from 75 to 93 % and may be complete if the drug is given under supervision. Isoniazid therapy requires regular monitoring; to be effective, good compliance is required. Hepatotoxicity is seen infrequently. Some Asian countries have isoniazid primary resistance rates of 11 % of more, so reliance on isoniazid alone for prevention of tuberculosis may mean that some babies are left unprotected. Rifampicin may be added to the regimen in areas where isoniazid-resistant M tuberculosis occurs, but this agent also has sideeffects. Single-agent prophylaxis is associated with the emergence of secondary isoniazid resistance, with rates of up to 16%. It is difficult to determine whether the original organism was resistant or whether resistance emerged after therapy. 10 Neonatal BCG vaccine is usually effective in preventing tuberculosis in childhood; its efficacy is only 38 to 75 % for pulmonary disease, but protection is higher for meningitis and miliary spread.11,12 Such differences are related to age at the time of vaccination,
EDITORIALS
Childhood leukaemia: an infectious disease? known causes of leukaemia in children--exposure to X rays in utero and ionising radiation-and these are unlikely to account for more than 8% of cases.! The disease is also associated with genetic abnormalities, of which Down’s syndrome is best known, but this applies to only 3% of children with leukaemia. Over 80% of childhood leukaemia is therefore unexplained. Might viral infections be an important causal factor? In adults, human T-lymphotropic virus type I (HTLV-1) is the one known viral cause of human leukaemia and the epidemiological features of this disease are not fully characterised. HTLV-1 infection clusters geographically, being rare outside the Caribbean and Japan; the prevalence of serological markers increases with age and the risk of leukaemia is low (perhaps only 1 in 1000 persistently infected individuals eventually manifest leukaemia2). Mode of transmission is unclear; early infection may occur without antibodies, as has been described for murine leukaemia virus.3 Although HTLV-1 causes considerably less than 1 % of all leukaemia, the little we know about the association already illustrates that simple models may not be sufficient for studying viral
There
are
only
two
leukaemogenesis. Many viruses, mostly retroviruses, cause leukaemia in animals; the best studied are those of mice, cats, and cattle. In mice, the pattern of leukaemia virus transmission differs in laboratory and wild populations. Thus laboratory animals seem to have high levels of vertical transmission, with integration of viral DNA into the host genome, whereas in wild mice transmission is mainly perinatal via infected maternal milk.3 Wild mice show clustering of infection by region and by family. The viruses do not produce overt disease during the acute infection, but persist in the host and induce disease many months or years later. Bovine leukaemia virus is strongly associated with lymphocytes. Cell transfer is needed for transmission and may occur
transplacentally, iatrogenically, or by arthropods (flies can transfer enough cells to infect).4 In cats, age at infection is an important determinant of persistence: virtually all persistently infected cats die within 5 years.s Dose of infecting organism is also important in determining persistence, and crowding of animals in a household may enhance the risk of infection.6,7 Some of the animal leukaemia viruses may also cause non-neoplastic diseaseseg, feline leukaemia virus may cause glomerulonephritis.’ Features of animal leukaemia viruses can be used as a model of a putative childhood leukaemia virus (CLV): (a) acute infection is insignificant clinically; (b) disease is related to persistent infection; and (c) persistence is determined by
and/or by a large dose of infectious of This model viral leukaemia-from infection to organism. to disease-resembles the association between persistence B virus and hepatitis hepatocellular carcinoma, in which various risk factors may act at different stages of the process. For hepatitis B and hepatocellular carcinoma, child-to-child transmission seems to be related to crowding of children;8 persistence of infection is closely related to age at infection, especially with infection during the first year of life.9 For leukaemia, the probability of contact between infectious and susceptible individuals might likewise be related to household crowding of young children, analogous to the observations in cats.7 Family structure and use of day centres for young children could contribute to crowding. Degree of contact may play a part in determining persistence, as in feline leukaemia. Thus crowded animals or children may be more likely to become infected and to receive a large dose of infectious organism. young age at infection
step is to apply this model to the that most leukaemia in children is hypothesis attributable to a common childhood infection. On this basis one would predict that high levels of child contact are a powerful predictor of infection at a young age. Urban populations may be large enough to maintain infection in an endemic state, with a young median age at infections whereas in rural areas infections may be present only intermittently, with higher median age at infection. Age at infection would tend to be lower in children of lower socioeconomic status and in those who have older siblings. However, these predictions are not borne out by the facts. Childhood leukaemia is associated with geographical areas of high socioeconomic status," as confirmed by Alexander and colleagues in this issue (p 1461), and several studies have shown that the socioeconomic status of affected individuals is higher than that of the general population.12-14 This socioeconomic correlation is not apparent in mortality figures for England and Wales, but with the improved treatment of the disease there are survival biases in these data. The disease seems to be more frequent in firstbom16,17 and single children,t8 although this finding is not universal. 14 The suggestion of Alexander et al that leukaemia is more common in isolated communities of higher socioeconomic status contrasts with the observations of Cook-Mozaffari and co-workers," who did not find an association with rural areas. Perhaps the discrepancy can be explained by the way in which a rural area was defined. Knox reported an excess incidence of leukaemia in urban areas for children over 6 years ;19 Alexander et al show that the effect of rural isolation is much stronger in those aged 1-7. The
next
1478
What about antenatal transmission of CLV? Knox2o pointed out that "mothers of affected children will be characterised by relative affluence, small sibships, good housing, and uncrowded neighbourhoods". When these predictions were investigated in a study of childhood leukaemia21 they were found not to fit. However, the researchers studied children whose birth dates extended over a long period and had to take account of the changes in family structure over time. Both chickenpox and influenza in pregnancy have been associated with an increased risk of leukaemia in the child. 2223 Since many common viral infections show seasonal variation one might expect such variation in dates of birth of affected children. A small summer excess in births has been reported2425 (which might be consistent with in-utero exposure during the winter or spring months), but these results are not persuasive. Co-twins with leukaemia have a single cell clone as the source of their disease,’ which implies that the disease arose in utero and spread from one twin to the other by mixing of placental circulation. The Knox model may not be appropriate to test for an infectious aetiology if the maternal infections were persistent; in bovine leukaemia, for example, 15% of infections occur transplacentally from persistently infected cows.
and co-workerS26-27 have examined geographical variation in leukaemia with the added dimension of change over time. They suggest that some leukaemia clusters may be associated with small rural communities that are expanding rapidly as a result of in-migration. This proposal was tested first in the new town of Glenrothes in Scotland and subsequently in fourteen new towns throughout Britain. The findings were consistent with the hypothesis, in that there was an excess of leukaemia in the rural towns. From these findings one would expect such communities to have low rates of leukaemia before growth whereas the results of Alexander et al suggest that these communities have higher than average background rates of disease. The short time lag between population growth and increase in leukaemia suggested by Kinlen’s data implies that the leukaemia must be caused by an effect on the child or mother during pregnancy and not by something that the mother carries with her from childhood. Periodic outbreaks of infection in rural populations usually lead Kinlen
to higher age at infection, so the postnatal virus model would not be satisfied.
persistent
Another way of looking for evidence of infection is to search for time-space clustering, but many noninfectious risk factors also aggregate in space and time. Application of space-time cluster methods to the proposed CLV is hampered by the fact that one does not know at what age exposure might have occurred and therefore at what time and place in an affected child’s life one should look for clustering. Moreover, if a large proportion of CLV infections were subclinical and persistent, but still infectious, clustering might be
very weak. Four studies have found such clustering of young children with leukaemia when onset of
used as the reference point.1128-30 Clustering was strongest in children under the age of 6 years and there was no evidence for clustering of dates of birth. However, many studies have not shown an association. Smith 31 concluded that the clustering found could have arisen by chance. Thus, there are paradoxes in the data, and dilemmas in interpretation, but little good evidence of
symptoms
was
infectious aetiology for an appreciable proportion of childhood leukaemia. It will be important to examine separately the observed differences in the epidemiological patterns in children under and over 7 years, since they may represent different aetiologies. That the major cell type of leukaemia under 7 years is the B cell whereas over this age there is more heterogeneity of cell types accords with this view. Moreover, childhood leukaemia peaks at 3 years, and the peak has appeared in this century. One possibility would be to look at morbidity-eg, glomerulonephritis-in siblings of affected children and controls, and it might be valuable to re-examine the characteristics of affected children and controls for evidence of prenatal infection. an
1. Doll R. The epidemiology of childhood 1989; 152: 341-51.
leukaemia. J R Statist Soc Ser A
2. Tajima K, Tominga S, Suchi T. Malignant lymphomas in Japan: epidemiological analysis on adult T-cell leukaemia/lymphoma. Haematol Oncol 1986; 4: 31-44. 3. Gardner MB. Naturally occurring leukaemia viruses in wild mice: how good a model for humans? Cancer Surv 1987; 6: 55-71. 4. Burny A, Cleuter Y, Kettman R, et al. Bovine leukaemia: facts and hypotheses derived from the study of an infectious cancer. Cancer Surv 1987; 6: 139-59. 5. McClelland AJ,
Hardy WD, Zuckerman EE. Prognosis of healthy feline leukemia virus infected cats. In: Hardy WD Jr, Essex M, McClelland AJ, eds. Feline leukaemia virus. New York: Elsevier, 1980: 121-31. 6. Hoover EA, Olsen RG, Hardy WD Jr, Schaller JP, Mathes LE. Feline leukemia virus infection: age-related variation in response of cats to experimental infection. JNCI 1976; 57: 365-69. 7. Francis DP, Essex M, Jakowski RM, Cotter SM, Lerer TJ, Hardy WD Jr. Increased risk for lymphoma and glomerulonephritis in a closed population of cats exposed to feline leukemia virus. Am J Epidemiol 1980; 111: 337-46. 8. Whittle HC, Bradley AK, McLauchlan, et al. Hepatitis B virus infection in two Gambian villages. Lancet 1983; i: 1203-06. 9. Coursaget P, Yvonnet B, Chotard J, et al. Age and sex related study of hepatitis B virus chronic carrier state in infants from an endemic area (Senegal). J Med Virol 1987; 22: 1-5. 10. Anderson RM, May RM. Directly transmitted infectious diseases: control by vaccination. Science 1982; 215: 1053-60. 11. Cook-Mozaffari P, Darby SC, Doll R, et al. Geographical variation in mortality from leukaemia and other cancers in England and Wales in relation to proximity to nuclear installations, 1969-78. Br J Cancer 1989; 59: 476-85. 12. Sanders BM, White GC, Draper GJ. Occupations of fathers of children dying from neoplasms. J Epidemiol Commun Health 1981; 35: 245-50. 13. McWhirter WR. The relationship of incidence of childhood lymphoblastic leukaemia to social class. Br J Cancer 1982; 46: 640-45. 14. Fasal E, Jackson EW, Klauber MR. Birth characteristics and leukemia in childhood JNCI 1971; 47: 510-09. 15. Office of Population, Censuses and Surveys. Occupational mortality, 1979-80, 1982-83. Childhood supplement series DS no 8. London: HM Stationery Office, 1988. 16. Stewart A, Webb J, Hewitt D. A survey of childhood malignancies. Br Med J 1958; i: 1495-508. 17. Stark CR, Mantel N. Effects of maternal age and birth order on the risk of mongolism and leukemia. JNCI 1966; 37: 687-98. 18. van Steensel-Moll HA, Valkenberg HA, van Zanen GE. Childhood
1479
leukemia and infectious diseases in the first year of life: a register based control study. Am J Epidemiol 1986; 124: 590-94. 19. Knox EG. Epidemiology of childhood leukaemia in Northumberland and Durham. Br J Prev Soc Med 1964; 18: 17-24. 20. Knox EG. Epidemiology of prenatal infections: an extension of the congenital rubella model. Stat Med 1983; 2: 1-12. 21. Knox EG, Stewart AM, Kneale GW. Foetal infection, childhood leukaemia and cancer. Br J Cancer 1983; 48: 849-52. 22. Frederick J, Alberman EA. Reported influenza in pregnancy and subsequent cancer in the child. Br Med J 1972; ii: 485-88. 23. Fine PEM, Adelstein AM, Snowman J, Clarkson JA, Evans SM. Long term effects of exposure to viral infections in utero. Br Med J 1985; 290: 509-11. 24. Bailar JC III, Gurian JM. Month of birth and cancer mortality. JNCI case
1964; 33: 237-42.
CR, Mantel N. Temporal-spatial distribution of birth dates for Michigan children with leukemia. Cancer Res 1967; 27: 1749-55.
25. Stark
26. Kinlen L. Evidence for an infective cause of childhood leukaemia: comparison of a Scottish new town with nuclear reprocessing sites in Britain. Lancet 1988; ii: 1323-27. 27. Kinlen LJ, Clarke K, Hudson C. Evidence from population mixing in British New Towns 1946-85 of an infective basis for childhood leukaemia. Lancet 1990; 336: 577-82. 28. Meighan SS, Knox G. Leukemia in childhood: epidemiology in Oregon. Cancer 1965; 18: 811-14. 29. Till MM, Hardisty RM, Pike MC, et al. Childhood leukaemia in Greater London: a search for evidence of clustering. Br Med J 1967; iii: 755-58. 30. Gunz FW, Spears GFS. Distribution of acute leukaemia in time and space: studies in New Zealand. Br Med J 1968; 4: 604-08. 31. Smith PG. Spatial and temporal clustering. In: Schottenfeld P, Fraumeni JF, eds. Cancer epidemiology and prevention. Philadelphia: Saunders, 1982: 391-408.
Perinatal prophylaxis of tuberculosis There is general agreement in many countries about the way in which tuberculosis should be treated. Much credit for this accord must go to the UK Medical Research Council for their efforts in coordinating prospective clinical trials in a various locations.1 By contrast, there is no such agreement about perinatal prophylaxis against the disease. Numerous management protocols have been developed in countries with different prevalence rates of pulmonary tuberculosis, but few studies have successfully examined the efficacy of these regimens. In areas of high prevalence, prevention of neonatal disease begins with maternal screening. In the USA, mothers from high-risk groups undergo tuberculin testing and chest radiographs are taken of those with reactions greater than Heaf grade 2 or of more than 5 mm induration with the Mantoux test. In some countries routine chest radiographs are done, with the necessary shielding. Active disease is then treated with rifampicin, isoniazid, and pyrazinamide.2 For sensitive strains of Mycobacterium tuberculosis the patient is deemed to be non-infectious within 2 weeks of starting therapy, although numbers of viable organisms are greatly reduced after only 24 hours. Some experts recommend that individuals with evidence of previous disease but no history of adequate therapy should start prophylactic isoniazid from the second trimester for 9-12 months.That this form of secondary prophylaxis may not be accepted as adequate therapy by North American immigration officials has encouraged doctors in some countries to use full therapeutic regimens for such patients.
Isolation of newborn babies is often advised, despite there being no clear evidence of benefit.3 In developing countries, depriving a child of breast milk may diminish its chances of survival. If maternal compliance with treatment cannot be guaranteed and there is concern about protection before neonatal BCG vaccination becomes effective, the baby can be given isoniazid for as long as the mother remains smear-positive, although there is little scientific support for this use of the drug. There is no need for patients to discontinue breastfeeding when they are receiving antituberculous treatment since only small amounts of the drugs pass into the milk.s Moreover, breast milk may enhance the immunogenicity of BCG given neonatally.6 The immediate family should be screened to detect any additional sources of infection. Management of babies born to smear-positive mothers differs in the UK and the USA. Both American and British thoracic societies recommend 3 months of prophylactic isoniazid, but opinion is divided about the usefulness of BCG vaccination after this period. BCG is favoured in the UK, but in the USA the only recipients are those whose mothers have multiresistant M tuberculosis or may not have complied with therapy. In both countries infants are reviewed at 3 months, when a tuberculin test is done and a chest radiograph is taken. Isoniazid is discontinued in children with clear chest radiographs and negative tuberculin tests. Positive tuberculin reactors are investigated for active disease and treated if necessary;.1 otherwise isoniazid prophylaxis is continued for 9-12 months. If the mother remains smear-positive at the 3-month review the child’s isoniazid prophylaxis is continued until the mother’s sputum becomes negative.7,8 In countries that practice neonatal BCG vaccination, regular review is appropriate with treatment if active disease develops. Isoniazid is undoubtedly effective in primary prophylaxis-protection generally varies from 75 to 93 % and may be complete if the drug is given under supervision. Isoniazid therapy requires regular monitoring; to be effective, good compliance is required. Hepatotoxicity is seen infrequently. Some Asian countries have isoniazid primary resistance rates of 11 % of more, so reliance on isoniazid alone for prevention of tuberculosis may mean that some babies are left unprotected. Rifampicin may be added to the regimen in areas where isoniazid-resistant M tuberculosis occurs, but this agent also has sideeffects. Single-agent prophylaxis is associated with the emergence of secondary isoniazid resistance, with rates of up to 16%. It is difficult to determine whether the original organism was resistant or whether resistance emerged after therapy. 10 Neonatal BCG vaccine is usually effective in preventing tuberculosis in childhood; its efficacy is only 38 to 75 % for pulmonary disease, but protection is higher for meningitis and miliary spread.11,12 Such differences are related to age at the time of vaccination,
