- childhood acute leukaemia;
- infectious aetiology;
- childhood acute lymphoblastic leukaemia
- Top of page
There are three current hypotheses concerning infectious mechanisms in the aetiology of childhood leukaemia: exposure in utero or around the time of birth, delayed exposure beyond the first year of life to common infections and unusual population mixing. No specific virus has been definitively linked with childhood leukaemia and there is no evidence to date of viral genomic inclusions within leukaemic cells. The case–control and cohort studies have revealed equivocal results. Maternal infection during pregnancy has been linked with increased risk whilst breast feeding and day care attendance in the first year of life appear to be protective. There is inconclusive evidence from studies on early childhood infectious exposures, vaccination and social mixing. Some supportive evidence for an infectious aetiology is provided by the findings of space-time clustering and seasonal variation. Spatial clustering suggests that higher incidence is confined to specific areas with increased levels of population mixing, particularly in previously isolated populations. Ecological studies have also shown excess incidence with higher population mixing. The marked childhood peak in resource-rich countries and an increased incidence of the childhood peak in acute lymphoblastic leukaemia (ALL) (occurring at ages 2–6 years predominantly with precursor B-cell ALL) is supportive of the concept that reduced early infection may play a role. Genetically determined individual response to infection may be critical in the proliferation of preleukaemic clones as evidenced by the human leucocyte antigen class II polymorphic variant association with precursor B-cell and T-cell ALL.
The aetiology of childhood leukaemia remains uncertain, but recent research has provided new clues. Both genetic susceptibility and environmental exposures are likely to be involved. It is possible that both prenatal and postnatal environmental exposures may play a crucial role in triggering the onset of leukaemia. The process leading to the onset of childhood leukaemia is likely to involve at least two events (Knudson, 1987; Greaves, 1988). Whilst the first event may be either germline (although this appears to be rare in childhood leukaemia) or somatic because of endogenous or environmental factors, the final ‘critical’ event may always involve an environmental factor. Both events would lead to cellular genetic changes and/or the proliferation of premalignant clones. Infections have been considered a prime candidate for such environmental aetiological agents that promote the onset of leukaemia. A causal link has been established between certain viruses and the development of leukaemia in animals (Penrose, 1970).
The aim of this review is to critically consider the evidence for the role of infectious exposures in the aetiology of childhood leukaemia, especially for acute lymphoblastic leukaemia (ALL). Both epidemiological and direct evidence was taken into account. A critical review of the literature has been undertaken specifically for this exercise using PubMed (http://www.ncbi.nlm.nih.gov/PubMed) from 1970 to date and the extensive review of Little (1999) for earlier references and for references up to and including 1997. A small number of pre-1998 references were found using PubMed that were not included in Little (1999). The present review reports in detail on references from 1998 to date, but only cites selected references that show marked results from the earlier period. The main findings of the remaining references from the earlier period (up to 1997) are summarised. Because of limitations of space it is not possible to report all of the literature in detail, especially non-significant results. Indeed, non-significant results may reflect either a real lack of association or lack of sufficient statistical power to detect an association. It is also possible that statistically significant results can sometimes arise by chance and thus are expected to occur from time to time. However, non-significant findings are summarised. It is also not possible to consider other aetiological factors in detail. Such other agents may indeed interact with genetic and infectious aetiological components. For a general review of the aetiology of childhood cancer including leukaemia, the reader is referred to Little (1999).
We first reviewed the current hypotheses concerning infectious mechanisms. Secondly, case–control and cohort epidemiological studies were reviewed together. These cover direct infectious exposures and vaccinations, breast feeding, early day care attendance and unusual population mixing. The results are interpreted in the light of the mechanistic hypotheses. Descriptive studies are then reviewed, encompassing incidence trends, space-time clustering, spatial clustering and seasonal variation. Again, these are critically interpreted as providing support for or against the aetiological hypotheses. Finally, the evidence of actual infectious involvement is analysed.
- Top of page
Kinlen (1988, 1995) proposed that an excess of childhood leukaemia would be seen in locations that had an unusual type of population mixing. The predisposing agent would be an, as yet unrecognised but relatively mild, infection and leukaemia would be a rare consequence of exposure to this infectious agent. However, in relatively isolated communities, a higher proportion of the population would have previously been unexposed to such an infection. A rapid influx of newcomers to such a previously isolated community would lead to a much greater than normal level of contact between the infected newcomers and the susceptible original residents. The leukaemia risk might be increased in both the original residents and the newcomers, who would reflect the degree of immune naivety in both populations.
- Top of page
Greaves (1988) proposed that the precursor B-cell subtype of ALL occurs as a consequence of at least two independent mutations. The first mutation occurs in utero or shortly after birth and creates a preleukaemic clone of cells. From twin studies and with the use of Guthrie cards, the majority of childhood cases of ALL that occur between the ages of 2 and 6 years have been shown to have in utero initiation. However, older children with TEL/AML1 and hyperdiploid ALL in particular have also been shown to have in utero initiation and a long latency. The second mutation occurs later (at a mean of 3–4 years after the first event) and precipitates the onset of overt leukaemia. Greaves (1988) suggested that common infections promote the second mutation and/or proliferation. Furthermore, Greaves (1988) inferred that delay in the normal pattern of exposure of the immune system to infection might lead to an increase of the number of susceptible preleukaemic cells and thus the chance of the second critical mutation occurring, leading to overt leukaemia.
- Top of page
Smith (1997) proposed that childhood peak ALL, which classically occurs between 2 and 6 years of age in populations of children in socio-economically developed communities, is due to in utero exposure to infection and that the childhood peak mainly consists of cases of precursor B-cell ALL.
Genetic predisposition and susceptibility to leukaemia
- Top of page
It is well-known that children with certain genetic syndromes involving obvious cytogenetic or genetic disruption are associated with an increased risk of developing childhood leukaemia. These include, for example, Down's syndrome, neurofibromatosis type I, ataxia telangiectasia, Fanconi anaemia and hereditary immunodeficiency. These predisposing disorders account for <5% of cases. Several recent studies suggest that how the individual recognizes foreign antigens, responds to them, recognizes and repairs DNA damage and metabolises chemicals may increase the likelihood of leukaemia developing. The susceptible individual with polymorphic ‘altered function’ genes may be at greater risk from known, or as yet unknown, environmental factors (Wiemels et al, 1999, 2001; Baria et al, 2002; Taylor et al, 2002; Barber et al, 2003).
The individual's response to infection is mediated by multiple cellular and humoral factors. Human leucocyte antigen (HLA) class II alleles influence the handling of foreign antigens. Taylor et al (2002) reported that carriage of the HLA class II DPB1*0201 allele (and other HLA class II DPB alleles with the same peptide pockets) increased susceptibility to the development of precursor B-cell ALL [relative risk (RR) = 1·76; 95% confidence interval (CI): 1·20–2·56] and T-cell ALL (RR = 1·93; 95% CI: 1·01–3·80). Other alleles, e.g. DPB1*0101, appeared to afford protection. The peptide-binding pockets determine the body's TH1/TH2 response to antigen exposure. They had also shown an excess frequency of the haplotype DQA1*0101/*0104 and DQB1*0501 in boys with precursor B-cell ALL, but not girls (there is a male excess of childhood peak ALL) (Taylor et al, 1998). Other, as yet unidentified, class II alleles and immune response genes may also play a role. The report by Kerr et al (2003) of a persisting presence of Parvovirus B19 infection in some leukaemic cases is of interest. This infection is associated with a significant cytokine cascade and consequent disturbance of marrow function. Suppression of normal marrow activity and/or proliferation of preleukaemic clones by such a cascade could precipitate the onset of overt leukaemia. What is not clear is whether these responses to infection(s) promote proliferation of malignant clones, suppress normal ‘protective’ immune responses to such abnormal cells, or do both.
Infectious episodes and vaccinations and the risk of childhood leukaemia
- Top of page
A number of recent case–control studies have reported on maternal infections, childhood infections and vaccinations and the risk of childhood leukaemia (Tables I–III).
Table I. Maternal Infections and risk of childhood leukaemia: recent studies (articles published post 1998).
|Study||Disease||Place||Study design||Infection||Risk estimate (95% CI)|
|Lehtinen et al (2003)||ALL||Finland/Iceland||Case–control||EBV in mother||OR = 2·9 (1·5–5·8)|
|Naumberg et al (2002)||Leukaemia||Sweden||Case–control||Maternal lower genital tract infection during pregnancy||OR = 1·78 (1·17–2·72)|
|Infante-Rivard et al (2000)||ALL||Quebec, Canada||Case–control||Recurrent maternal infections||OR = 1·09 (0·65–1·84)|
|McKinney et al (1999)||Leukaemia||Scotland||Case–control||Any infection in pregnancy||OR = 1·43 (0·85–2·42)|
|ALL||OR = 1·44 (0·81–2·55)|
Table II. Childhood Infections and risk of childhood leukaemia: recent studies (Post, 1998).
|Study||Disease||Place||Study design||Infection||Risk estimate (95% CI)|
|Jourdan-Da Silva et al (2004)||ALL||France||Case–control||Early infections||OR = 0·8 (0·6–1·0)|
|Nyari et al (2003)||L & NHL||UK||Case–control||Measles||Age 0–2 months: OR = 1·0 (0·8–1·2)|
|Other infections|| OR = 1·0 (0·8–1·2)|
|Respiratory infections|| OR = 1·1 (0·9–1·3)|
|Influenza|| OR = 0·9 (0·7–1·2)|
|Measles||Age 3–5 months: OR = 1·0 (0·8–1·2)|
|Other infections|| OR = 1·0 (0·9–1·2)|
|Respiratory infections|| OR = 1·0 (0·9–1·2)|
|Influenza|| OR = 0·8 (0·6–1·1)|
|Chan et al (2002)||ALL||Hong Kong||Case–control||Roseola/fever and rash in first year of life||OR = 0·33 (0·16–0·68)|
|Tonsillitis 3/12 months before diagnosis||OR = 2·56 (1·22–5·38)|
|Perrillat et al (2002a)||AL||France||Case–control||Repeated early common infections||OR = 0·6 (0·4–1·0)|
|Petridou et al (2001)||Leukaemia||Greece||Case–control||Seropositivity for:|| |
| EBV||OR = 0·4 (0·2–0·8)|
| HHV-6||OR = 0·5 (0·3–0·9)|
| Parainfluenza 1, 2 and 3||OR = 1·9 (1·1–3·2)|
| Mycoplasma||OR = 0·4 (0·1–1·2)|
|Neglia et al (2000)||ALL||USA||Case–control||Increasing number of ear infections||P-value for trend = 0·03|
|Dockerty et al (1999a)||Leukaemia||New Zealand||Case–control||Influenza infection in first year of life||OR = 6·8 (1·8–25·7)|
|McKinney et al (1999)||ALL||Scotland||Case–control||Neonatal infection||OR = 0·49 (0·26–0·95)|
|Schuz et al (1999a)||AL||Germany||Case–control||Chickenpox||OR = 0·8 (0·7–1·0)|
Table III. Vaccinations and risk of childhood leukaemia.
|Study||Disease||Place||Study design||Vaccination||Risk estimate (95% CI)|
|Statistically significant protective effect|
| Groves et al (1999)||ALL||USA||Case–control||Conjugate Hib-vaccine||OR = 0·57 (0·36–0·89)|
| Schuz et al (1999a)||AL||Germany||Case–control||≥4 routine immunizations||OR = 0·3 (P < 0·05)|
| Nishi and Miyake (1989)||Non-T-cell||Hokkaido,||Case–control||BCG vaccination||RR = 0·1 (0·0–1·0)|
|ALL||Japan||Measles or measles vaccination||RR = 0·2 (0·1–0·7)|
| McKinney et al (1987)||Leukaemia||UK||Case–control||Immunizations||OR = 0·2 (0·1–0·9)|
|Statistically significant increased risk|
| Dockerty et al (1999a)||Leukaemia||New Zealand||Case–control||Measles vaccination||OR = 1·87 (1·00–3·48)|
| Buckley et al (1994)||Common ALL||USA and Canada||Case–control||MMR vaccination||OR = 1·7 (P < 0·01)|
| Groves et al (2002)||Leukaemia||Finland||Clinical trial||Hib-vaccine|| HbOC arm: OR = 1·14 (0·63–2·08)|
| Auvinen et al (2000)||Leukaemia||France||Clinical trial||Hib-vaccine||Early vaccination arm:|
|RR = 0·72 (0·46–1·13)|
| Dockerty et al (1999a)||Leukaemia||New Zealand||Case–control||Any vaccinations||OR = 0·71 (0·36–1·38)|
| Petridou et al (1997a)||Leukaemia||Greece||Case–control||Viral vaccinations||OR = 1·23 (0·91–1·66)|
Maternal infections and the risk of childhood leukaemia.
- Top of page
There have been four recent case–control studies (see Table I). Two of these studies (Naumberg et al, 2002; Lehtinen et al, 2003) have reported a significantly increased risk (for childhood leukaemia) associated with maternal infection during pregnancy (Table I). Specifically, Epstein–Barr virus (EBV) infection had an odds ratio (OR) of 2·9 (95% CI: 1·5–5·8; Lehtinen et al, 2003) and lower genital tract infection had an OR of 1·8 (95% CI: 1·2–2·7; Naumberg et al, 2002). Additionally, McKinney et al (1999) reported a non-significantly raised OR for leukaemias (OR = 1·43; 95% CI: 0·85–2·42) and for ALL (OR = 1·44; 95% CI: 0·81–2·55) for any infection during pregnancy. However, Infante-Rivard et al (2000) showed little effect for recurrent maternal infections (OR = 1·09; 95% CI: 0·65–1·84). Studies from 1997 or earlier have been previously reviewed by Little (1999). Amongst them, Roman et al (1997) reported a significantly increased risk of leukaemia with non-specific viral infection (OR = 6·0; 95% CI: 1·2–29·7) and a non-significantly raised OR of 4·0 for non-specific viral infection and subsequent childhood ALL. Another study looked at leukaemia subtype and the increased risk was particularly apparent for precursor B-cell ALL, with an OR of 1·5 (P < 0·05) for non-specific maternal infection (Buckley et al, 1994). Six other reported case–control studies looked at maternal infection and risk of childhood leukaemia (reviewed by Little, 1999). One reported a non-significantly raised OR for varicella during pregnancy and subsequent childhood leukaemia, another reported a non-significantly raised OR for influenza during pregnancy and subsequent childhood ALL, whilst four showed ORs very close to 1.
Childhood infections and the risk of childhood leukaemia.
- Top of page
There have been nine recent case–control studies (see Table II). Three of these studies (Dockerty et al, 1999a; Petridou et al, 2001; Chan et al, 2002) showed a statistically significant increased risk associated with childhood exposure to some infections, with one study (Dockerty et al, 1999a) emphasizing exposure during the first year of life as a risk factor (Table II). However, seven of these recent studies (McKinney et al, 1999; Schuz et al, 1999a; Neglia et al, 2000; Petridou et al, 2001; Chan et al, 2002; Perrillat et al, 2002a; Jourdan-Da Silva et al, 2004) showed a statistically significant protective effect. Chan et al (2002) showed a statistically significant protective effect of infection with roseola (fever plus rash) against ALL when exposure occurred during the first year of life, but a statistically significant increased risk of developing ALL for those with tonsillitis 3–12 months before diagnosis. Also, Petridou et al (2001) showed a statistically significant protective effect for seropositivity to EBV and human herpes virus (HHV)-6, but a statistically significant increased risk for seropositivity to parainfluenza 1, 2 and 3. One study showed no effect, but this was for leukaemia and non-Hodgkin's lymphoma (NHL) analysed together (Nyari et al, 2003). Older studies have been reviewed by Little (1999). Of these, three case–control studies showed a statistically significant increased risk, whilst five showed a statistically significant protective effect. Two of these studies (McKinney et al, 1987; Hartley et al, 1988) emphasized exposure during the first year of life as a risk factor, whilst one study (Van-Steensel-Moll et al, 1986) showed a protective effect associated with exposure during the first year of life.
There were seven other case–control studies (reviewed by Little, 1999) that did not show any statistically significant relationships between childhood infections and subsequent risk of developing leukaemia, ALL or acute non-lymphocytic leukaemia (ANLL). Additionally, there were no consistently high or low ORs that were associated with specific infections, including chickenpox, mumps, measles, whooping cough and dysentery.
Vaccinations and the risk of childhood leukaemia.
- Top of page
Four studies showed a protective effect for immunization, including Bacillus Calmette-Guerin (BCG), measles and conjugate Haemophilus influenzae type b (Hib)-vaccines (McKinney et al, 1987; Nishi & Miyake, 1989; Groves et al, 1999; Schuz et al, 1999a), with ORs ranging from 0·1 to 0·6, whilst one study showed an increase following measles, mumps and rubella (MMR) vaccination (Buckley et al, 1994), with an OR of 1·7 (P < 0·01) (Table III). A non-significant protective effect was found in a clinical trial comparing administration of Hib-vaccine at age 3 months versus administration at age 2 years (Auvinen et al, 2000). The incidence of subsequent leukaemia was lower in the early vaccination arm (RR = 0·72; 95% CI: 0·46–1·13). Another clinical trial failed to show any difference between two Hib-vaccination formulations and risk of childhood leukaemia (Groves et al, 2002). A non-significant increased risk was found to be associated with viral vaccinations in a case–control study from Greece (Petridou et al, 1997a) and a non-significant decreased risk was found with any vaccinations from a case–control study from New Zealand, whilst an increased risk for measles vaccination was found from the same study (Dockerty et al, 1999a) (Table III). Earlier studies on vaccinations and risk of childhood leukaemia (reviewed by Little, 1999) were essentially inconclusive.
Whilst, it is difficult to draw any firm conclusions from these data, there is a suggestion that maternal infection during pregnancy may be linked with an increased risk of childhood leukaemia development. However, the results of all the studies on childhood exposure to infections (six increased, 12 decreased risks, seven no significant effect) and childhood immunizations (four protective, two increased risk, four non-significant effect) are inconclusive.
Breast feeding and the risk of childhood leukaemia
- Top of page
Breast feeding has been taken as a surrogate for both infectious exposure and protection. Fifteen case–control studies have been reviewed in the paper by the UK Childhood Cancer Study (UKCCS) (UKCCS Investigators, 2001; Table IV). Overall, considering ‘ever breast-fed’versus‘never’, there was a protective effect (OR = 0·86; 95% CI: 0·81–0·92), although there was an error in the transcription of numbers from Davis et al (1988) (which, if taken into account favours protection). Five of the studies showed statistically significant decreased risks, five found non-significantly increased risks and five found non-significantly decreased risks. Eleven of these case–control studies looked at duration of breast feeding. For duration of 6 months or less, overall there was a protective effect (OR = 0·91; 95% CI: 0·85–0·99). One of the studies showed a statistically significant increased risk, one showed a statistically significant decreased risk, four found non-significantly increased risks and five found non-significantly decreased risks. For duration of more than 6 months, overall there was a protective effect (OR = 0·78; 95% CI: 0·71–0·85). Four of the studies showed statistically significant decreased risks, three found non-significantly increased risks and four found non-significantly decreased risks. Overall there was a tendency for longer duration of breast feeding to decrease the risk of childhood leukaemia developing. More recently Perrillat et al (2002b) conducted a case–control study from France covering cases aged over 2 years and diagnosed between 1995 and 1999. They found a protective effect of ‘ever having breast-fed’ for all leukaemia (OR = 0·8; 95% CI: 0·6–1·2) and a marked dose–response effect for ALL, but not ANLL. For ALL and breast feeding more than 12 months the OR was 0·5 (95% CI: 0·1–2·5). However, three other recent studies (Murray et al, 2002; Lancashire & Sorahan, 2003; Jourdan-Da Silva et al, 2004) have found no association with breast feeding, irrespective of its duration (Table IV).
Table IV. Breast feeding and risk of childhood leukaemia.
|Study||Disease||Period of diagnosis||Place||Study design||Risk estimate (95% CI)|
|Jourdan-Da Silva et al (2004)||AML||1995–98||France||Case–control||Ever breast-fed: OR = 1·1 (0·9–1·5)|
|AML|| OR = 1·4 (0·8–2·5)|
|Lancashire and Sorahan (2003)||ALL||1972–81||UK||Case–control||Ever breast-fed: OR = 1·04 (0·86–1·26)|
|Perrillat et al (2002b)||AL||1995–99||France||Case–control||Duration ≤6 months: OR = 0·5 (0·2–1·0)|
|Murray et al (2002)||ALL||1971–86||Northern Ireland||Cohort||Never breast-fed: RR = 0·98 (0·68–1·42)|
|UKCCS Investigators (2001)||Leukaemia||1991–96||UK||Case–control||Ever breast-fed: OR = 0·86 (0·81–0·92)|
|Duration ≤6 months: OR = 0·91 (0·85–0·99)|
|Duration ≥6 months: OR = 0·78 (0·71–0·85)|
Breast feeding may provide early exposure to infectious agents, by close maternal–infant contact but also passive antibody transfer early on, both of which may contribute to a degree of protection. However, we must express caution because of the well-documented different breast feeding rates by socio-economic status. In case–control studies, controls are very consistently from higher socio-economic groups than cases. Therefore, the findings may be due to confounding factors and not represent a true causal relationship.
Day care attendance and risk of childhood leukaemia
- Top of page
Early-in-life social contact/day care attendance has been taken as a surrogate for infectious exposure. No studies have shown an increased risk of childhood leukaemia associated with day care attendance. However, five case–control studies (Table V; Petridou et al, 1993; Infante-Rivard et al, 2000; Ma et al, 2002; Perrillat et al, 2002a; Jourdan-Da Silva et al, 2004) showed a protective effect of attendance. For three of these studies (Petridou et al, 1993; Infante-Rivard et al, 2000; Jourdan-Da Silva et al, 2004) the effect was associated with day care attendance in the first 2 years of life, whilst for one (Ma et al, 2002) the effect was associated with prolonged day care attendance. A further five case–control studies showed non-significant protective effects or no association with day care (Table V; Roman et al, 1994; Petridou et al, 1997a; Neglia et al, 2000; Rosenbaum et al, 2000; Chan et al, 2002). Day care attendance would provide increased opportunity for exposure to infectious agents.
Table V. Day care attendance and risk of childhood leukaemia.
|Study||Disease||Period of diagnosis||Place||Study design||Risk estimate (95% CI)|
|Statistically significant protective effect|
| Jourdan-Da Silva et al (2004)||AL||1995–98||France||Case–control||Day care before 3 months: OR = 0·6 (0·4–0·8)|
| Perrillat et al (2002a)||AL||1995–99||France||Case–control||Ever day care attendance: OR = 0·6 (0·4–1·0)|
| Ma et al (2002)||ALL||1995–99||Northern California, USA||Case–control||OR/1000 child hours: 0·991 (0·984–0·999)|
|OR/50 000 child hours: 0·64 (0·45–0·95)|
| Infante-Rivard et al (2000)||ALL||1980–93||Quebec, Canada||Case–control||First year of life day care attendance: OR = 0·49 (0·31–0·77)|
| Petridou et al (1993)||Leukaemia||1987–91||Athens region, Greece||Case–control||Day care attendance during first 2 years of life: OR = 0·28 (0·09–0·88)|
|Non-significant effects/no association|
| Chan et al (2002)||ALL||1994–97||Hong Kong||Case–control||Day care in first year of life: OR = 0·96 (0·70–1·32)|
|Day care in last year before diagnosis: OR = 0·88 (0·34–2·12)|
| Rosenbaum et al (2000)||ALL||1980–91||New York State||Case–control||Stayed at home compared with day care for >36 months: OR = 1·32 (0·70–2·52)|
| Neglia et al (2000)||ALL||1989–93||USA||Case–control||Any day care: OR = 0·96 (0·82–1·12)|
| Petridou et al (1997a)||Leukaemia||1993–94||Greece||Case–control||Any day care: OR = 0·83 (0·51–1·37)|
| Roman et al (1994)||ALL||1972–89||Southern England||Case–control||Preschool playgroup for more than 3 months in the year before diagnosis: OR = 0·6 (0·2–1·8)|
Caution should be exercised in interpreting these findings. The possibility exists that there may be social class bias between the case and control groups regarding use of day care. Because of costs involved, early and prolonged day care attendance is likely to be more prevalent amongst higher socio-economic groups. Controls were very consistently of higher socio-economic status than cases in most such case–control studies. It should also be noted that ALL tends to have a higher incidence in groups of higher socio-economic status and thus unadjusted risk estimates would tend to reduce the effect of this bias. Four studies (Infante-Rivard et al, 2000; Neglia et al, 2000; Ma et al, 2002; Perrillat et al, 2002a) have attempted to take socio-economic status or its surrogates into account. All four of these studies found a protective effect of day care attendance. Overall, we still have to conclude that the findings may be due to confounding factors and may not represent a true causal effect. The large UKCCS study results are eagerly awaited.
Individual social-mixing and risk of childhood leukaemia
- Top of page
There are various ways in which any child may be exposed to infectious agents. These include exposures within the family, from older siblings and parents, and also exposures from the wider community. In order to assess these effects, studies on birth order of the index case, parental occupational contact levels and individual migration are considered. It must be stressed that all of these studies relate to individuals, as distinct from ecological level studies, which are considered later.
Birth order and risk of childhood leukaemia.
- Top of page
High birth order may be taken as a surrogate for early exposure to infection from siblings. There have been 13 recent case–control studies and one cohort study examining the effect of birth order of the index case on subsequent risk of childhood leukaemia (Table VI). Five of these studies showed that a statistically significant increased risk was associated with high birth order (Infante-Rivard et al, 2000; Bener et al, 2001; Reynolds et al, 2002; Shu et al, 2002; Jourdan-Da Silva et al, 2004), including one that showed a specific association for ANLL in children aged 0–2 years (Reynolds et al, 2002) and another that showed a specific association for ALL aged 0–4 years (Infante-Rivard et al, 2000). The latter study (Infante-Rivard et al, 2000) also showed a statistically significant protective effect for children who had older siblings in the first year of life and were aged 4 or more years at the time of diagnosis. One other large study showed a statistically significant protective effect for higher birth order (Dockerty et al, 2001). Eight of the recent studies (Petridou et al, 1997a; McKinney et al, 1999; Schuz et al, 1999b; Neglia et al, 2000; Murray et al, 2002; Okcu et al, 2002; Perrillat et al, 2002a; Paltiel et al, 2004) found no statistically significant relationships with birth order. Older studies have been reviewed by Little (1999). Most of these studies did not show an increased risk for first-born individuals. However, in one of the largest earlier studies, there was a non-significant trend for decreasing risk with increasing birth order for cases of ALL, aged 0–4 years (Westergaard et al, 1997).
Table VI. Birth order and risk of childhood leukaemia: recent studies (articles published post 1996).
|Study||Disease||Period of diagnosis||Place||Study design||Risk estimate (95% CI)|
|Jourdan-Da Silva et al (2004)||ALL||1995–98||France||Case–control||1: OR = 1 (reference)|
|2: OR = 0·9 (0·7–1·2)|
|3: OR = 1·1 (0·8–1·7)|
|4+: OR = 2·0 (1·1–3·7)|
|Paltiel et al (2004)||Leukaemia||1964–76||Israel||Cohort||1: OR = 1|
|2+: OR = 1·07 (0·6–1·8)|
|Murray et al (2002)||ALL||1971–86||Northern Ireland||Cohort||1: OR = 1 (reference)|
|2+: OR = 0·98 (0·71–1·36)|
|Okcu et al (2002)||Leukaemia||1995||Texas||Case–control||1: OR = 1|
|2: OR = 1·1 (0·7–1·9)|
|3: OR = 1·0 (0·5–1·9)|
|4: OR = 1·0 (0·4–2·3)|
|5: OR = 1·2 (0·3–5·2)|
|6+: OR = 0·5 (0·1–4·3)|
|Shu et al (2002)||ALL||1989–93||USA||Case–control||1: OR = 1 (reference)|
|2: OR = 1·3 (1·1–1·6)|
|3: OR = 1·5 (1·2–2·0)|
|4+: OR = 2·0 (1·3–3·0)|
|Perrillat et al (2002a)||AL||1995–99||France||Case–control||1: OR = 1 (reference)|
|2: OR = 1·2 (0·8–1·9)|
|3: OR = 0·9 (0·5–1·5)|
|4+: OR = 1·5 (0·8–3·0)|
|Reynolds et al (2002)||AL||1988–97||California||Case–control||ANLL in 0–2 year olds 3+: OR = 1·59 (1·00–2·53)|
|Bener et al (2001)||L & L||1983–97||United Arab Emirates||Case–control||Higher birth order: OR = 1·80 (1·13–2·86)|
|Dockerty et al (2001)||Leukaemia||1968–86||England and Wales||Case–control||ALL age 1–5:|
|1: OR = 1|
|2: OR = 0·85 (0·73–0·98)|
|3: OR = 0·74 (0·60–0·91)|
|4: OR = 0·64 (0·47–0·87)|
|5: OR = 0·61 (0·36–1·03)|
|6+: OR = 0·43 (0·26–0·73)|
|Infante-Rivard et al (2000)||ALL||1980–93||Quebec, Canada||Case–control||Having older siblings at time of diagnosis in children aged <4 years: OR = 4·54 (2·27–9·07)|
|Having older siblings in first year of life in children aged ≥4 years: OR = 0·46 (0·22–0·97)|
|Neglia et al (2000)||ALL||1989–93||USA||Case–control||1: OR = 1|
|2: OR = 1·08 (0·93–1·26)|
|3+: OR = 1·05 (0·88–1·26)|
|Schuz et al (1999b)||Leukaemia||1992–94||Germany||Case–control||1: OR = 1·1 (0·9–1·3)|
|2+: OR = 1|
|McKinney et al (1999)||Leukaemia||1991–94||Scotland||Case–control||1OR = 1|
|2+: OR = 0·82 (0·55–1·23)|
|Petridou et al (1997a)||Leukaemia||1993–94||Greece||Case–control||Higher birth order: OR = 0·74 (0·48–1·15)|
Parental occupational contact levels and risk of childhood leukaemia.
- Top of page
Three case–control studies (Table VII; Kinlen, 1997; Kinlen et al, 2002; Pearce et al, 2004) showed an increased risk for childhood leukaemia associated with high levels of paternal occupational contact (number of social contacts with the father whilst at work), one case–control study (Kinlen & Bramald, 2001) showed a decreasing trend with increasing levels of paternal occupational contact, whilst one death-certificate study (Table VII; Fear et al, 1999) showed no association. The increase in one UK study was associated with ALL in 2–5 year olds (Pearce et al, 2004) and the increase in a study from Sweden was associated with leukaemia at 0–4 years of age in children from the most rural counties (Kinlen et al, 2002). These findings are interpreted as a direct result of lack of early immune stimulation. The indication from these studies is that previous immune isolation is particularly a risk factor for childhood leukaemia in those aged up to 5 years. In contrast, a case–control study from Scotland found that the decreasing trend for increasing paternal occupational contact was found for 5–14 year olds and in rural areas in periods of higher population mixing (Kinlen & Bramald, 2001). This protective effect for older children is interpreted as due to immunity acquired at an earlier age or by an immunizing effect of low doses of an infective agent at a later age.
Table VII. Parental occupational contact and risk of childhood leukaemia
|Study||Disease||Period of diagnosis||Place||Study design||Risk estimate (95% CI)|
|Pearce et al (2004)||Leukaemia||1968–97||UK||Case–control||Fathers’ occupational contact very high: ALL aged 2–5 years: OR = 1·5 (1·1–2·1)|
|Kinlen et al (2002)||Leukaemia||1958–98||Sweden||Case–control||0–4 year olds: using two classifications of paternal contact level in most rural counties:|
|(1) OR = 3·47 (1·54–7·85)|
|(2) OR = 1·59 (1·07–2·38)|
|Kinlen and Bramald (2001)||Leukaemia||1950–89||Scotland||Case–control||5–14 year olds: decreasing trend for those in rural areas in period of higher population mixing with increasing paternal occupational contact (P < 0·05)|
|Fear et al (1999)||Leukaemia||1959–63||England and Wales||Population-based Death Certificate Study||Fathers’ occupational contact:|
|1970–90||High social contact: PCMR = 94 (87–102)|
|Medium social contact: PCMR = 101 (95–106)|
|Kinlen (1997)||Leukaemia||1950s onwards||Five studies of unusual population mixing in the UK||Case–control||Paternal occupational contact:|
|Medium/low: OR = 1|
|High: OR = 1·75|
|Very high: OR = 2·17|
Migration and risk of childhood leukaemia.
- Top of page
There have been two case–control studies examining the effect of migration on the risk of childhood leukaemia (Alexander et al, 1993; Chan et al, 2002). Chan et al (2002), in a study from Hong Kong, found a statistically significant protective effect associated with change of area of residence in the first year of life (OR = 0·47; 95% CI: 0·23–0·98) and a statistically significant increased risk associated with change of area of residence in the year before diagnosis (OR = 3·92; 95% CI: 1·47–10·46). Alexander et al (1993) examined migration patterns of children with leukaemia and NHL in Northern England. The study found few case–control differences. There was a suggestion that case children were more mobile in the first few years of life and particularly for cases diagnosed shortly after the move in residence. There were relatively few changes in residence for mothers in the years prior to the case children's birth, but it was found that 18 case mothers had moved compared with eight control mothers (P < 0·05) after birth.
Community population-mixing and risk of childhood leukaemia
- Top of page
Community level population mixing is a direct method for testing Kinlen's (1988, 1995) hypothesis. A cohort study (Kinlen & Balkwill, 2001) showed increased risks for increased levels of population mixing. This study considered mortality from leukaemia and found that, during World War II, when there was a vastly increased population in the isolated Scottish islands, mortality from childhood leukaemia was significantly increased for the war time period, but not for the postwar period (RR = 3·64; 95% CI: 1·67–6·92 for 1941–45 and RR = 1·06; 95% CI: 0·22–3·09 for 1946–55) (see also under Ecological Studies).
Geographical incidence of childhood leukaemia
- Top of page
The incidence of childhood leukaemia varies greatly between countries. A number of registries from the International Agency for Research on Cancer (IARC) publication (Parkin et al, 1998) were chosen to provide a representative range from the more resource-rich to the more deprived countries and populations. These registries attempt to be truly population-based, trying not to be selective regarding cases included. Incidence rates, per 100 000 person-years, for lymphoid leukaemia, obtained from these registries, are presented in Table VIII, ordered by magnitude (from highest to lowest) of the childhood peak (aged 0–4 years). There is a tendency for higher rates to be seen in the more resource-rich countries and amongst more affluent populations within countries, whilst lower rates are observed in resource-limited countries and more deprived populations within countries.
Table VIII. Incidence of lymphoid leukaemia: international comparisons.
|Registry||0–4 years||5–9 years||10–14 years|
|USA, SEER, White (1988–92)||6·0||2·9||1·9|
|UK, England and Wales (1981–90)||5·2||2·6||1·6|
|Cali, Colombia (1982–91)||3·8||3·6||1·9|
|USA, SEER, Black (1983–92)||2·7||1·9||1·4|
|Israel, Jews (1980–89)||2·5||1·5||1·4|
|India, Bombay (1980–92)||1·9||1·6||1·2|
|Israel, non-Jews (1980–89)||1·6||1·7||1·5|
|Ibadan, Nigeria (1985–92)||1·1||5·6||1·3|
Table VIII shows that a marked childhood peak (ages 0–4 years) is evident in Denmark, Australia, Sweden, US whites [from the Surveillance, Epidemiology and End Results (SEER) study] and England and Wales (UK). For these registries, the childhood peak (ages 0–4 years), rates were >5 per 100 000 person-years. Lower rates at ages 0–4 were observed for children from Cali (Colombia), Japan, Estonia, African-Americans in the USA compared with whites (from the SEER study), both Israeli Jews and Israeli non-Jews, Bulgaria, Bombay (India) and Estonia. The lowest incidence rates were seen in Ibadan (Nigeria).
There is generally at least a twofold difference in childhood peak incidence rates between the most and least affluent populations. These incidence data would suggest that either social behaviour or lifestyle factors in more affluent populations leads to a much greater chance of children developing leukaemia. One possible explanation has been proposed that deprivation increases the chance of greater infectious exposure early in life and affluence reduces such exposure.
Evidence from temporal trends in the incidence of childhood leukaemia
- Top of page
The study of incidence trends requires high quality population-based incidence and corresponding population denominator data over a sustained time period. It also requires sufficiently large numbers so that the possibility of chance variation may be excluded. There are a number of studies of incidence trends in childhood leukaemia (Little, 1999). ALL is known to have increased in incidence in the United States and Europe from the 1920s to the 1980s by about a factor of four (Doll, 1989). Whilst there was inconsistency between studies, the three largest studies from Europe (Parkin et al, 1996), England and Wales, UK (Draper et al, 1994) and the USA (Gurney et al, 1996) were in general agreement. European data for the period 1980–86 showed an increase in the incidence of all leukaemias of 0·6% per annum and 0·4% per annum for the period 1987–91. The rise was most apparent for 1–4 year olds (Parkin et al, 1996). Data for the period 1953–91 for England and Wales, UK showed an increase in the incidence of ALL of about 1·2% per annum. Again, this increase was most marked for 1–4 year olds (Draper et al, 1994). Data for the period 1974–91 from the United States SEER study showed an increase in the incidence of childhood ALL of 1·6% per annum, stable rates for acute myeloid leukaemia (AML) amongst whites, but an increase in AML of 2·7% per annum amongst African-Americans (Gurney et al, 1996). Table IX presents the six most recent studies from Sweden, Spain, Italy, USA, Hungary and the UK (McNally et al, 2000, 2001; Jakab et al, 2002; McNeil et al, 2002; Magnani et al, 2003; Dreifaldt et al, 2004; Gonzalez et al, 2004). Statistically significant increases were found in five of the six studies. Two of these studies showed a marked increase for precursor B-cell ALL in the childhood peak (McNally et al, 2000; Magnani et al, 2003). Furthermore, in one study the childhood peak increase was found to be greater for girls than for boys, although the difference was not statistically significant (McNally et al, 2000).
The continued upward trend, particularly in the incidence of childhood peak ALL, is of interest. Most cases are of the precursor B-cell subtype. Either more children are being exposed to an, as yet unknown, environmental, directly transforming aetiological agent or more children are subject to a more complex mechanism, such as decreasing exposure to infectious agents, early in life. A general increase in living conditions would lead to the latter situation. There is evidence to support this, in that the childhood peak in incidence emerged earlier in more affluent countries or populations, but only more recently in less affluent countries or populations (Little, 1999).
Differences in the reported rates for childhood cancer between boys and girls have been noted in developing countries (Pearce & Parker, 2001). However, the recent difference noted from North West England, where the increase for females was more marked than for males, cannot be explained as an artefact because of gender differences in reporting rates (McNally et al, 2000). Rather, this may reflect upon change in behaviour, allowing girls as well as boys to have increased social contact and hence a lessening of differences in opportunity for early exposure to infectious agents.
Space-time clustering in the incidence of childhood leukaemia
- Top of page
Space-time clustering is said to occur when excess numbers of cases are observed within small geographical locations at limited points in time which cannot be explained in terms of general excesses in those locations or at those times. Hence, space-time clustering may be described as ‘the irregular grouping of cases of any disease simultaneously in space and time’. These irregularities might result from any number of the following situations:
A small number of ‘locations’ with greatly increased incidence at short (but distinct) periods of time.
A large number of ‘locations’ with moderately increased incidence at limited periods of time.
A small number of short (but distinct) ‘time periods’ with greatly increased incidence at limited locations.
A large number of limited ‘time periods’ with moderately increased incidence at limited locations.
If infections are involved in the aetiology of childhood leukaemia then the distribution of cases may exhibit space-time clustering. However, this would only happen if the infection occurred in ‘mini-epidemics’ or if it only affected a limited number of susceptible people. An infection that was ubiquitous or endemic would be predicted to lead to a homogeneous case distribution and not lead to space-time clustering. In studying incidence patterns, the distribution of times and places may be taken with respect to times and residential addresses at birth, diagnosis and/or onset of symptoms. Consistent and reliable population-based data are required over a relatively wide geographical area and time span. If the initiating exposure was an infection that occurred at or around the time of birth or in utero, then the distribution of cases may be predicted to exhibit space-time clustering based on time and place of birth. If the initiating exposure was an infection that occurred at or around the time of diagnosis (or onset), then the distribution of cases may be predicted to exhibit space-time clustering based on time and place of diagnosis (or onset). The latter two situations would also arise if there is a relatively constant lag-time from exposure to diagnosis (or onset). One other situation is relevant, where there is substantial migration between birth and diagnosis and space-time clustering is based on time of diagnosis and place of birth. This would only occur if exposure occurred around the place of birth and there was a constant lag-time between exposure and diagnosis.
A number of statistical methods have been applied to test for space-time clustering (Knox, 1964; Pike & Smith, 1974; Diggle et al, 1995; and methods reviewed in Little, 1999). Some of the earlier methods have been criticized because they rely on an arbitrary choice of boundaries to define ‘close in time’ and ‘close in space’. The more recent method of Diggle et al (1995) overcomes this problem. Nevertheless significant space-time clustering would be suggestive of a role for infections in the aetiology of childhood leukaemia.
Recent studies of space-time clustering are presented in Table X. Two studies from the UK and Europe (Alexander et al, 1998a; Birch et al, 2000) exhibited space-time clustering based on time and place of diagnosis (i.e. infection close to diagnosis). Two studies from the UK and Sweden (Gustafsson & Carstensen, 2000; McNally et al, 2002) found space-time clustering based on time and place of birth (infection at some time distant from onset including in utero). One study from the UK (Birch et al, 2000) revealed space-time clustering based on time of diagnosis and place of birth (infection around the place of birth and a constant lag-time between exposure and diagnosis). Twenty-eight older studies of space-time clustering are reviewed in Little (1999). Five of these older studies, from the UK, Greece, Australia and the USA, exhibited space-time clustering based on time and place of diagnosis. Five of the older studies, from the UK, USA and New Zealand, showed space-time clustering based on time and place of onset (i.e. infection close to presentation with leukaemia). The two studies from New Zealand used different methods on the same data set. Three older studies found space-time clustering based on time and place of birth. Fifteen older studies found little or no evidence of space-time clustering. However, many of the older analyses have used small data sets and inappropriate methods that have employed an arbitrary choice of ‘close in time’ and ‘close in space’. Two of the most recent analyses (Birch et al, 2000; McNally et al, 2002) have addressed the arbitrary boundary problem by using the method of Diggle et al (1995).
Table X. Space-time clustering of childhood leukaemia: recent studies (articles published post 1997).
|Study||Country||Disease||Age (years)||Time period|
|Statistically significant based on time of diagnosis and place of diagnosis|
|Birch et al (2000)||North West England||Leukaemia||0–14||1954–85|
|Alexander et al (1998a)||Europe and Australia||ALL||<1||1980–89|
|Statistically significant based on time of birth and place of birth|
|McNally et al (2002)||North West England||ALL||0–14||1980–2001|
|Gustafsson and Carstensen (2000)||Sweden||ALL||4–14||1973–96|
|Statistically significant based on time of diagnosis and place of birth|
|Birch et al (2000)||North West England||Leukaemia||0–14||1954–85|
The two studies which analysed the most recent data from the UK and Sweden (Gustafsson & Carstensen, 2000; McNally et al, 2002) only found space-time clustering based on time and place of birth and not based on time and place of diagnosis, more suggestive of an early life event effect of exposure.
If taken together, all of these data implies that most frequently infection may play a role at the time of second or later events, i.e. postbirth when there is already a preleukaemic clone in existence. The infection either induces further mutations leading to overt leukaemia or triggers proliferation. However, the data showing space-time clustering based on time and place of birth implies involvement in the initial ‘hit’ or event inducing the preleukaemic clone. Thus, it is feasible that infection may play a role in both early events leading to a preleukaemic clone as well as at a later proliferative stage.
Spatial clustering in the incidence of childhood leukaemia
- Top of page
Spatial clustering is defined as ‘the irregular grouping of cases of any disease in space’. Such an irregular distribution is a general phenomenon and is not confined to one specific small area. These irregularities could arise either because there are a small number of areas with greatly increased incidence or a large number of areas with moderately increased incidence. The presence of spatial clustering may indicate the involvement of any environmental factor including infections. It is certainly more appropriate than the analysis of space-time clustering if the infectious agents are associated with sustained and variable latent periods. A number of different statistical methods have been applied to test for spatial clustering related to the diagnostic residence address (including Potthoff & Whittinghill, 1966; Muirhead & Ball, 1989; Cuzick & Edwards, 1990; and methods reviewed in Alexander & Boyle, 1996 and Little, 1999). These studies depend on accurate denominator (population) counts by age and sex at small area levels. Any inaccuracy in the denominator data could lead to either a dilution or exaggeration of a clustering effect. For example, the underestimation of a population count in a small area may lead to an artificial excess, whilst the overestimation of such a population count may lead to an artificial deficit. Recent studies of spatial clustering are presented in Table XI. Four studies, from Europe, Hong Kong, Greece and the UK (Knox & Gilman, 1996; Alexander et al, 1997, 1998b; Petridou et al, 1997b), demonstrated statistically significant evidence of spatial clustering. One study, from New Zealand (Dockerty et al, 1999b), showed evidence of spatial clustering for ALL for one age group only (10–14 year olds). Two studies from North West England and Sweden showed no evidence of spatial clustering (Hjalmars et al, 1996; McNally et al, 2003). Older studies on clustering of leukaemia are reviewed in Little (1999) and Alexander and Boyle (1996). In summary, spatial clustering has been found in the UK, Greece and Hong Kong, but not in Sweden or in metropolitan areas of the USA. However, the geographical area units in the USA were much larger than in other countries, which may have diluted small-scale clustering effects. Significant spatial clustering was strongest for cases in young children. In the UK, spatial clustering was confined to sparsely populated areas, whereas in Greece it was confined to more urban areas. The findings of spatial clustering are, in general, consistent with a role for infections or other localized environmental factors, especially in childhood peak ALL.
Table XI. Spatial clustering of childhood leukaemia: recent studies (articles published post 1995).
|Study||Country||Disease||Age (years)||Time period||Clustering method||Results|
|McNally et al (2003)||North West England||Acute leukaemia||0–14||1976–2000||Potthoff-Whittinghill||No evidence of clustering|
|Dockerty et al (1999b)||New Zealand||Leukaemia||0–14||1976–87||Cuzick-Edwards||No evidence of clustering overall but significant clustering for ALL, ages 0–14 (P = 0·003)|
|Alexander et al (1998b)||Europe: 17 countries||Leukaemia||0–14||1980–89||Potthoff-Whittinghill||Overall significant evidence of clustering for total leukaemia within small census areas (P = 0·03)|
|Alexander et al (1997)||Hong Kong||Leukaemia||2–6||1984–90||Potthoff-Whittinghill||Evidence of clustering of ALL at ages 0–4 (P = 0·09) and in the childhood peak (P < 0·05)|
|Petridou et al (1997b)||Greece||Leukaemia||0–14||1980–89||Potthoff-Whittinghill||Significant evidence of clustering|
|Hjalmars et al (1996)||Sweden||Leukaemia||0–14||1973–93||Spatial scan statistic||No significant clusters found|
|Knox and Gilman (1996)||UK||Leukaemia||0–14||1953–83||Close pairs method||Short range spatial clustering for leukaemia at place of registration, birth and death addresses|
Seasonal variation in the incidence of childhood leukaemia
- Top of page
Seasonal variation may provide evidence of an infectious aetiology for childhood leukaemia, or indeed, it may indicate the involvement of other environmental factors, such as pesticides that are applied consistently at one time of the year. If infections are involved, then greater population mixing at certain times of year is likely to lead to an increased occurrence of ‘mini-epidemics’ of infection. As a rare consequence of infection, a limited number of cases will develop leukaemia. Several different statistical methods have been used to test for seasonal variation (for example, Edwards, 1961; Walter & Elwood, 1975). Seasonal variation related to time of birth would indicate exposure to an infection in utero or around the time of birth. Seasonal variation related to time of diagnosis or time of first symptom would indicate exposure close to diagnosis with a short lag-time between date of infectious exposure and date of diagnosis/date of first symptom, or a longer but constant lag-time. Table XII presents the results of studies from Iran, UK, Denmark, USA and Greece (Zannos-Mariolea et al, 1975; Badrinath et al, 1997; Westerbeek et al, 1998; Ross et al, 1999; Feltbower et al, 2001; Higgins et al, 2001; Sorensen et al, 2001; Karimi & Yarmohammadi, 2003) that have found significant seasonal variability related to date of birth, date of first symptom and/or date of diagnosis. There was a lot of variability in the time of peak incidence, but four of six studies showed summer as the peak time for diagnosis. There were two studies (from Hungary and the USA) of seasonal variation related to date of birth, which showed no effect (Meltzer et al, 1996; Kajtar et al, 2003). In general, the results are more consistent with an infectious exposure than with any other environmental exposure. Variability between- and within-countries would be consistent with different patterns of population mixing, leading to ‘mini-epidemics’ of infections. Variation in ‘peak’ seasons would also be expected with climate differences, which clearly influence the timing of particular infectious outbreaks.
- Top of page
Geographical ecological studies on population mixing are presented in Table XIII (for other studies on population mixing see ‘Parental occupational contact levels and risk of childhood leukaemia’ and Table VII). The earlier population mixing studies, which relate to areas of population growth, wartime evacuation, national service, construction work, commuting and mass tourism are extensively reviewed by Little (1999). They present a complex picture. Five of Kinlen's studies, from the UK, showed significant excesses in areas of ‘unusual’ rural population mixing (reviewed in Little, 1999). These ‘unusual’ rural situations included: areas with an unusually high proportion of residents in the highest social class; areas with the highest ratios of wartime evacuee children compared with local children; areas with the largest proportion of oil workers; rural new towns with a greater density of children and greater diversity of origin than overspill towns; and in areas which had the highest proportions of servicemen.
Table XIII. Recent (Post, 1995) ecological studies: population mixing.
|Labar et al (2004)||Croatia||ALL||0–14||Incidence of ALL increased in four counties where population mixing had occurred during the war period (P < 0·05)|
|Law et al (2003)||UK||ALL||0–14||Elevated risks in areas with a low diversity of origins of migrants (P < 0·05)|
|Boutou et al (2002)||France||ALL||1–6||Highest tertile of population mixing|
| IRR = 5·5 (1·4–23·3)|
|Dickinson et al (2002)||UK||Leukaemia||0–14||For urban areas only increased risk with higher levels of inward migration, particularly from outside the region|
|RR = 1·9 (1·2–2·9)|
|Parslow et al (2002)||UK||Leukaemia||0–14||Lower incidence in areas of higher population mixing: top decile IRR = 0·72 (0·54–0·97)|
|Dickinson and Parker (1999)||UK||ALL and NHL||0–14||Higher among children born in areas with highest levels of population mixing RR = 11·7 (3·2–43)|
|Alexander et al (1997)||Hong Kong||ALL and precursor||0–4||For small census areas with extreme population mixing, overall incidence raised and significant evidence of spatial clustering (P < 0·05)|
|Stiller and Boyle (1996)||UK||ALL||0–9||Trends for higher incidence at ages 0–4 and 5–9 with the proportion of children in a district who had recently entered the district (P < 0·05)|
|The combination of higher migration with greater diversity of origins or distance moved was associated with higher incidence in both age groups|
More recently, five studies (see Table XIII), from Croatia, France and the UK, showed significant excesses associated with general increased levels of population mixing (Stiller & Boyle, 1996; Dickinson & Parker, 1999; Boutou et al, 2002; Dickinson et al, 2002; Labar et al, 2004). These also included war situations and urban areas. However, two studies, from the UK (Parslow et al, 2002; Law et al, 2003) reported lower incidence associated with population mixing. A recent study from Canada showed a statistically significant excess for rural areas that had experienced population growth exceeding 20% (Koushik et al, 2001). One study from Hong Kong found spatial clustering in areas of extreme population mixing (Alexander et al, 1997).
- Top of page
Three ecological studies have found links between marked population increase within a short period of time and increased risk of childhood leukaemia. An ecological study from Canada (Koushik et al, 2001) found a statistically significant increase in the incidence of ALL in rural areas that experienced >20% population growth (RR = 1·9; 95% CI: 1·1–2·8). Another ecological study from the UK (Langford, 1991) found a statistically significant increase in childhood leukaemia mortality in areas that experienced more than a 50% increase in population over the decade 1961–71 (RR = 1·41; 95% CI: 1·13–1·76). The most recent study from the USA examined SEER data. Wartenberg et al (2004) found that changes in rural county population sizes from 1980 to 1989 were associated with increased rates for childhood ALL. The associations were strongest among 0–4 year olds, born in the same state as diagnosis, in extremely rural counties and when counties adjacent to non-rural counties were excluded.
- Top of page
Five studies (see Table XIV), from the UK, Sweden, Taiwan and the USA (Table XIII), showed increased incidence (Muirhead, 1995; Li et al, 1998; Hjalmars & Gustafsson, 1999; McNally et al, 2003) or mortality (Gilman & Knox, 1998) in more densely populated or urban areas. However, one study from Europe and Australia (Alexander et al, 1999) found a curvilinear association with population density. The highest rates were found in areas that had an ‘intermediate’ population density (500–750 persons/km2), with incidence rate ratio relative to areas with ≥1000 persons/km2 = 1·16 (95% CI: 1·07–1·26).
Table XIV. Recent (articles published post 1994) ecological studies: population density
|McNally et al (2003)||North West England||ALL||0–14||Higher incidence in areas of greater population density (P < 0·05)|
|Alexander et al (1999)||Europe and Australia||Leukaemia||0–14||Curvilinear association with population density: highest rates in areas which had ‘intermediate’ population density (500–750 persons/km2), with incidence rate ratio relative to areas with ≥1000 persons/km2 = 1·16 (1·07–1·26)|
|Hjalmars and Gustafsson (1999)||Sweden||ALL||0–14||In population centres, excess of ALL: OR = 1·68 (1·44–1·95), but not ANLL: OR = 1·13 (0·98–1·32)|
|Li et al (1998)||Taiwan||Leukaemia||0–4||Urban areas showed a higher incidence rate RR = 1·3 (1·1–1·6)|
|Gilman and Knox (1998)||UK||Leukaemia||0–15||High mortality associated with areas having high population densities|
|Muirhead (1995)||USA||L & NHL||0–14||RR for highest population density areas relative to lowest = 1·4 (1·0–2·0)|
- Top of page
A number of ecological studies have examined the relationship between incidence of leukaemia and area-based measures of socio-economic status or deprivation (reviewed by Little, 1999). Whilst these studies have used various ecological measures of socio-economic status or deprivation, most have found a weak positive association between leukaemia and high socio-economic status. One other study (Alexander et al, 1996) showed that the incidence of ALL in children aged 1–7 was increased in areas with least car ownership (RR = 1·28; 95% CI: 1·12–4·64). In this context, the area-based level of car ownership may be taken as a measure of deprivation. Thus, the finding from this study is not in agreement with the general trend from the majority of other studies. Additionally, McNally et al (2003) found no association between leukaemia incidence and area-based measures of deprivation.
Specific viral exposures
- Top of page
Given the extensive literature regarding infections and, in particular, viral genomic inclusion seen in animals with leukaemia (cats, chickens, etc.), it is not surprising that some researchers have postulated that childhood leukaemia may be initiated by infection (Kinlen, 1995; Smith, 1997; Smith et al, 1998). The case–control studies summarised earlier that suggested an association with maternal infection during pregnancy lend some weight to such an hypothesis, particularly following the convincing evidence of in utero initiation of most ALL cases (Wiemels et al, 1999). The identity of the specific infectious agents that might be responsible remains in doubt and indeed, the timing of such exposure and leukaemic initiation are also unknown (Knox et al, 1983; Van-Steensel-Moll et al, 1986; Alexander, 1993).
A limited number of studies have been performed where advancing technology has facilitated reliable searches of the genome. Smith et al (1999) reported on an investigation of leukaemic cells from 25 children with precursor B-cell ALL looking for genomic sequences of the JC, BK and Simian 40 polyoma viruses. They did not detect any such inclusion and concluded that the majority of precursor B-cell ALL could not be due to these particular viruses. MacKenzie et al (1999) conducted a similar study and also found no evidence of genomic inclusion of these particular viruses. Priftakis et al (2003) tested 54 Guthrie cards from newborns who subsequently developed ALL and also tested 37 from healthy controls. They found no evidence for the presence of JC or BK viruses. MacKenzie et al (2001) also conducted a study of herpes viruses based on the known transforming capabilities of some herpes viruses, e.g. EBV and HHV-8. There had been a previous report of HHV-6 sequences having been detected in T-cell ALL (Luka et al, 1991) and in an adult patient with precursor B-cell ALL (Daibata et al, 1998). In the latter, the genomic inclusion was both in normal tissue as well as the leukaemic cells and thought to be a bystander phenomenon. Furthermore, Ma et al (2000) reported on the presence of HHV-6 genomic inclusion detected by polymerase chain reaction (PCR) in a significant percentage of the bone marrow samples from Chinese adults and children with leukaemia. Shiramizu et al (2002) tested 28 paediatric ALL patients for the TT virus, but all of the diagnostic specimens were negative for the virus, although seven patients had follow-up specimens that were positive for the TT virus. They concluded that the TT virus is unlikely to be causally related to ALL. MacKenzie et al (2001) carried out two studies, one on a panel of 20 pretreatment leukaemic samples (bone marrow/blood) looking for the presence of lymphotropic herpes viruses using conventional molecular techniques (Southern blot for EBV, PCR for HHV-6, -7 and -8) and then on a second, independent cohort of 27 leukaemia samples, which were analysed by TaqMan real-time, quantitative-PCR. Childhood solid tumour patients provided control samples (glioma, Ewing's sarcoma, rhabdomyosarcoma) since normal healthy childhood samples were not available. Eighteen of the leukaemic samples (where adequate material was available) were examined for herpes virus genome inclusion using a degenerate PCR assay capable of detecting genome from HHVs but also putative new members of the family. Low levels of herpes virus genomes were detected in patient and control samples at similar levels. No novel herpes virus genomes were detected. They concluded that this group of herpes viruses were unlikely to be aetiologically involved as transforming agents in precursor B-cell ALL.
High levels of HHV-6 antibodies were reported in ALL patients compared with healthy controls by Ablashi et al (1988) but subsequent studies by Levine et al (1992) and Schlehofer et al (1996) have found no such association. The study by Chan et al (2002) showing a protective effect of roseola infection adds weight to the evidence that HHV-6 antibodies are coincidental and not causally related to ALL development. Schlehofer et al (1996) also reported that children under 6 years with acute leukaemia were more likely to be EBV seropositive than age-matched controls, but there has been no further confirmation of that.
There have been limited studies to date trying to define the organism(s) to which abnormal immune responses might lead to the conversion of a preleukaemic clone into overt leukaemia [time-point two in Greaves’ (1988) hypothesis]. Alexander (1997) looked at disease surveillance data for England and Wales regarding mycoplasma pneumonia and reported that, for cases of childhood leukaemia, there was a trend for the population incidence of that infection to be at its lowest levels when they were young. Petridou et al (2001) studied 94 incident cases of ALL and 94 matched controls in Greece. They found that, whilst there was little evidence for an association of ALL with specific infectious agents amongst children aged 0–4 years, for children aged 5 years or more, leukaemia was inversely associated with seropositivity to EBV, HHV-6, Mycoplasma pneumoniae and Parvovirus B19.
A report by Kerr et al (2003) found evidence of the persistent presence of Parvovirus B19 infection in cerebrospinal fluid coincidentally examined in four of 16 childhood cases of ALL (three) and AML (one). This finding is of interest in that, previously, parvovirus infection has been reported in association with ALL but considered to be a consequence of the immunosuppression and not causally related to the ALL. Acute parvovirus infection is associated with a significant cytokine cascade, which is associated with a degree of disturbed haematopoiesis and/or suppression of normal marrow function, which may allow ‘release’ of low level malignant clones or indeed, induce proliferation.
- Top of page
Largely because of the extensive series of studies led by Greaves (1999), we now know that, certainly for infant ALL with MLL rearrangements, most precursor B-cell ALL (certainly TEL-AML1 and hyperdiploid ALL) and childhood AML with t(8;21) translocations, the first genetic event in childhood leukaemia involves production of a preleukaemic clone in utero. Greaves (1999) clearly hypothesized that whatever the cause of the gene rearrangements in utero, postnatal events including infection are almost certainly required to promote the development of clinical leukaemia. He has clearly defined the need for at least two and possibly multiple postnatal events. Therefore, when considering aetiology we have to think of the initiating event occurring most frequently in utero and subsequent events occurring postnatally. Whether infection plays a part at either or both time-points is unclear.
Smith (1997) proposed that the first ‘genetic’ event might well be initiated by infection. All the studies to date have shown no evidence of viral genomic inclusion in leukaemic cells. However, the suggestion that maternal infection during pregnancy may be associated with an increased risk of childhood leukaemia would be supportive of such a relationship. Kinlen's (1988, 1995) hypothesis includes the possibility that infection might occur in utero or near to birth. The space-time clustering data based on time and place of birth would clearly be consistent with events in utero. The findings from the two most recent space-time clustering studies from the UK and Sweden (Gustafsson & Carstensen, 2000; McNally et al, 2002) would lend support to in utero exposure, certainly in the most recent time period.
Clearly there is quite a lot of supportive data to suggest that the later postnatal events are indeed related to infections and/or the body's response to them. Evidence from the geographical incidence and temporal increases in the incidence of childhood leukaemia is consistent with lack of immune stimulation during the first few years of life, particularly in more affluent communities and societies. The limited and protective effect of breast feeding and more significantly, the apparent benefit of increased social contact because of nursery attendance in the first year of life would again be supportive of the concept that leukaemia is more likely if there is an absence of infectious exposure early in life and that leukaemia results from some form of delayed and possibly abnormal response to infection at a later age. The finding of the HLA class II allele linkage to precursor B-cell ALL is further corroborative evidence. Space-time clustering studies centred on time and place of diagnosis are clearly also consistent with the delayed infection hypotheses, as are those based on time of diagnosis and place of birth and also Kinlen's studies on unusual population mixing. It is very important to realize that the Greaves (1988) and Kinlen population mixing hypotheses are not mutually exclusive. Elements of both may be involved in individual cases. Infection may initiate a preleukaemic clone but that may not lead to overt leukaemia in the absence of delayed later infection and/or abnormal response to such infection. Population mixing increases the chance of infectious exposure in susceptible people at any stage.
At the present time, in the absence of definitive evidence regarding either specific single or multiple infectious culpable agents at either time-point, we have to continue to investigate events and exposures around the two time-points and the molecular events that result from such exposures. Greater clarification as to likely aetiological agents may emerge. For example, in infant leukaemia, where very characteristic rearrangement of the MNLL gene occurs, there has been both epidemiological and molecular evidence to suggest that exposure to naturally occurring topoisomerase II inhibitors and the inability by the mother and fetus to metabolize them rapidly significantly increases the risk of developing leukaemia. Topoisomerase inhibitors reduce DNA damage that is inappropriately repaired. A combination of exposure, DNA damage and inappropriate repair plus failure to metabolize all contributes to the initiation of a rare leukaemia. We cannot fail to conclude that, ultimately, there may be a number of pathways in the development of any childhood leukaemia that nevertheless will always involve initial breakage and inaccurate repair of DNA in response to infection, chemicals, low level irradiation or other, as yet unknown, environmental exposures. Then one or more subsequent further events convert a preleukaemic clone into an overt malignant population. Two genetic events and possibly also proliferative stimuli and/or suppression of bone marrow are all required to produce an overt leukaemia. Genetic susceptibility increasingly looks likely in terms of not only the response to infection, but also the recognition and repair of the DNA.
The study of the molecular events associated with leukaemic transformation may help us to better understand the changes. However, at present we are still some time away from measures that would enable us to more subtly target therapy or indeed apply preventative measures based on a clear understanding of causation, including immune modulation of response to infection.
- Top of page
Authors thank Cancer Research UK for financial support. Tim O.B. Eden is in Cancer Research UK, Professor of Paediatric Oncology at the University of Manchester.
- Top of page
- 1988) Human B-lymphotropic virus (human herpesvirus-6). Journal of Virological Methods, 21, 29–48. , , , , , , , , , , , , , , , , , & (
- 1993) Viruses, clusters and clustering of childhood leukaemia: a new perspective? European Journal of Cancer, 29A, 1424–1443. (
- 1997) Is Mycoplasma pneumoniae associated with childhood acute lymphoblastic leukaemia? Cancer Causes and Control, 8, 803–811. (
- 1996) (eds) Methods for Investigating Localized Clustering of Disease. IARC Scientific Publications, No. 135, IARC, Lyon. & (
- 1990) Community lifestyle characteristics and risk of acute lymphoblastic leukaemia in children. The Lancet, 336, 1461–1465. , , & (
- 1993) Migration patterns of children with leukaemia and non-Hodgkin's lymphoma in three areas of Northern England. Journal of Public Health Medicine, 15, 9–15. , & (
- 1996) Isolation, car ownership, and small area variation in incidence of acute lymphoblastic leukaemia in children. Paediatric and Perinatal Epidemiology, 10, 411–417. , & (
- 1997) Clustering of childhood leukaemia in Hong Kong: association with the childhood peak and common acute lymphoblastic leukaemia and with population mixing. British Journal of Cancer, 75, 457–463. , , , , , , , , & (
- The EUROCLUS Project (1998a) Spatial temporal patterns in childhood leukaemia: further evidence for an infectious origin. EUROCLUS project. British Journal of Cancer, 77, 812–817. , , , , , , , , , , , , , , & on behalf of
- The EUROCLUS Project (1998b) Spatial clustering of childhood leukaemia: summary results from the EUROCLUS project. British Journal of Cancer, 77, 818–824. , , , , , , , , , , , , , , , , , , & on behalf of
- 1999) Population density and childhood leukaemia: results of the EUROCLUS Study. European Journal of Cancer, 35, 439–444. , , , , , , , , , , , , , , , , , , & (
- 2000) Haemophilus influenzae type B vaccination and risk of childhood leukaemia in a vaccine trial in Finland. British Journal of Cancer, 83, 956–958. , & (
- 1997) Seasonality in the diagnosis of acute lymphocytic leukaemia. British Journal of Cancer, 75, 1711–1713. , & (
- 2003) Constitutional sequence variation in the Fanconi anaemia group C (FANCC) gene in childhood acute myeloid leukaemia. British Journal of Haematology, 121, 57–62. , , , , , & (
- 2002) Chromosomal radiosensitivity in young cancer patients: possible evidence of genetic predisposition. International Journal of Radiation Biology, 78, 341–346. , , , , & (
- 2001) Longer breast-feeding and protection against childhood leukaemia and lymphomas. European Journal of Cancer, 37, 234–238. , & (
- 2000) Space-time clustering patterns in childhood leukaemia support a role for infection. British Journal of Cancer, 82, 1571–1576. , , , , & (
- 2002) Population mixing and leukaemia in young people around the La Hague nuclear waste reprocessing plant. British Journal of Cancer, 87, 740–745. , , , & (
- 1994) Epidemiological characteristics of childhood acute lymphocytic leukemia. Analysis by immunophenotype. The Childrens Cancer Group. Leukemia, 8, 856–864. , , , , , & (
- 2002) Is the timing of exposure to infection a major determinant of acute lymphoblastic leukaemia in Hong Kong? Paediatric and Perinatal Epidemiology, 16, 154–165. , , , , , , , , , , , & (
- 1990) Spatial clustering for inhomogeneous populations (with discussion). Journal of the Royal Statistical Society, Series B, 52, 73–104. & (
- 1998) Identification of integrated human herpes virus 6 DNA in early pre-B cell acute lymphoblastic leukemia. Leukemia, 12, 1002–1004. , , , , & (
- 1988) Infant feeding and childhood cancer. The Lancet, ii, 365–368. , & (
- 1999) Quantifying the effect of population mixing on childhood leukaemia risk: the Seascale cluster. British Journal of Cancer, 81, 144–151. & (
- 2002) Population mixing and childhood leukaemia and non-Hodgkin's lymphoma in census wards in England and Wales, 1966–87. British Journal of Cancer, 86, 1411–1413. , , & (
- 1995) Second-order analysis of space-time clustering. Statistical Methods in Medical Research, 4, 124–136. , , & (
- 1999a) Infections, vaccinations, and the risk of childhood leukaemia. British Journal of Cancer, 80, 1483–1489. , , , , & (
- 1999b) An assessment of spatial clustering of leukaemias and lymphomas among young people in New Zealand. Journal of Epidemiology and Community Health, 53, 154–158. , & (
- 2001) Case-control study of parental age, parity and socioeconomic level in relation to childhood cancers. International Journal of Epidemiology, 30, 1428–1437. , , , & (
- 1989) The epidemiology of childhood leukaemia. Journal of the Royal Statistical Society, Series A, 152, 341–351. (
- 1994) Childhood cancer. Cancer Surveys, 19/20, 493–517. , & (
- 2004) Increasing incidence rates of childhood malignant diseases in Sweden during the period 1960–1998. European Journal of Cancer, 40, 1351–1360. , & (
- 1961) The recognition and estimation of cyclic trends. Annals of Human Genetics, 25, 83–86. (
- 1999) Are the children of fathers whose jobs involve contact with many people at an increased risk of leukaemia? Occupational and Environmental Medicine, 56, 438–442. , , & (
- 2001) Seasonality of birth for cancer in Northern England, UK. Paediatric and Perinatal Epidemiology, 15, 338–345. , , , & (
- 1998) Geographical distribution of birth places of children with cancer in the UK. British Journal of Cancer, 77, 842–849. & (
- 2004) Trends in childhood cancer incidence and mortality in Catalonia, Spain, 1975–1998. European Journal of Cancer Prevention, 13, 47–51. , , , , , & (
- 1988) Speculations on the cause of childhood acute lymphoblastic leukaemia. Leukemia, 2, 120–125. (
- 1999) Molecular genetics, natural history and the demise of childhood leukaemia. European Journal of Cancer, 35, 1941–1953. (
- 1999) Infant vaccinations and risk of childhood acute lymphoblastic leukaemia in the USA. British Journal of Cancer, 81, 175–178. , , , , , & (
- 2002) Haemophilus influenzae type b vaccine formulation and risk of childhood leukaemia. British Journal of Cancer, 87, 511–512. , & (
- 1996) Trends in cancer incidence among children in the US. Cancer, 78, 532–541. , , , , & (
- 2000) Space-time clustering of childhood lymphatic leukaemias and non-Hodgkin's lymphomas in Sweden. European Journal of Epidemiology, 16, 1111–1116. & (
- 1988) The Inter-regional Epidemiological Study of Childhood Cancer (IRESCC): past medical history in children with cancer. Journal of Epidemiology and Community Health, 42, 235–242. , , , , , , , , , , & (
- 2001) Season of birth and diagnosis of children with leukaemia: an analysis of over 15000 UK cases occurring from 1953–95. British Journal of Cancer, 84, 406–412. , , & (
- 1999) Higher risk for acute childhood lymphobastic leukaemia in Swedish population centres 1973–94. Swedish Child Leukaemia Group. British Journal of Cancer, 79, 30–33. & (
- 1996) Childhood leukaemia in Sweden: using GIS and a spatial scan statistic for cluster detection. Statistics in Medicine, 15, 707–715. , , & (
- 2000) Markers of infection, breast-feeding and childhood acute lymphoblastic leukaemia. British Journal of Cancer, 83, 1559–1564. , & (
- 2002) Epidemiologic studies in a population-based childhood cancer registry in Northeast Hungary. Medical and Pediatric Oncology, 38, 338–344. , , & (
- 2004) Infectious diseases in the first year of life, perinatal characteristics and childhood acute leukaemia. British Journal of Cancer, 90, 139–145. , , , , , , , , , , , , , , , , , , & (
- 2003) Month of birth in childhood acute lymphoblastic leukemia [Hungarian]. Orvosi Hetilap, 144, 1869–1871. , & (
- 2003) Seasonal variations in the onset of childhood leukaemia/lymphoma: April 1996 to March 2000, Shiraz, Iran. Haematological Oncology, 21, 51–55. & (
- 2003) Association of acute parvovirus B19 infection with new onset of acute lymphoblastic and myeloblastic leukaemia. Journal of Clinical Pathology, 56, 873–875. , , , , , , , , , & (
- 1988) Evidence for an infective cause of childhood leukaemia: comparison of a Scottish New Town with nuclear reprocessing sites in Britain. The Lancet, ii, 1323–1327. (
- 1995) Epidemiological evidence for an infective basis in childhood leukaemia. British Journal of Cancer, 71, 1–5. (
- 1997) High-contact paternal occupations, infection and childhood leukaemia: five studies of unusual population-mixing of adults. British Journal of Cancer, 76, 1539–1545. (
- 2001) Infective cause of childhood leukaemia and wartime population mixing in Orkney and Shetland, UK. The Lancet, 357, 858. & (
- 2001) Paternal occupational contact level and childhood leukaemia in rural Scotland: a case-control study. British Journal of Cancer, 84, 1002–1007. & (
- 1991) Contacts between adults as evidence for an infective origin of childhood leukaemia: an explanation for the excess near nuclear establishments in West Berkshire? British Journal of Cancer, 64, 549–554. , & (
- 2002) A case-control study of childhood leukaemia and paternal occupational contact level in rural Sweden. British Journal of Cancer, 86, 732–737. , & (
- 1964) The detection of space-time interactions. Applied Statistics, 13, 25–29. (
- 1996) Spatial clustering of childhood cancers in Great Britain. Journal of Epidemiology and Community Health, 50, 313–319. & (
- 1983) Foetal infection, childhood leukaemia and cancer. British Journal of Cancer, 48, 849–852. , & (
- 1987) A two-mutation model for human cancer. Advances in Viral Oncology, 7, 1–17. (
- 2001) An ecologic study of childhood leukemia and population mixing in Ontario, Canada. Cancer Causes and Control, 12, 483–490. , & (
- 2004) Haematological malignancies in Croatia: investigating the theories of depleted uranium, chemical plant damage and ‘population mixing’. European Journal of Epidemiology, 19, 55–60. , , , , , , , , & (
- 2003) Breastfeeding and childhood cancer risks: OSCC data. British Journal of Cancer, 88, 1035–1037. & (
- 1991) Childhood leukaemia mortality and population change in England and Wales 1969–73. Social Science and Medicine, 33, 435–440. (
- United Kingdom Childhood Cancer Study Investigators (2003) Childhood cancer and population mixing. American Journal of Epidemiology, 158, 328–336. , , &
- 2003) Maternal herpes virus infections and risk of acute lymphoblastic leukemia in the offspring. American Journal of Epidemiology, 158, 207–213. , , , , , , , , , , & (
- 1992) Antibodies to human herpes virus-6 in patients with acute lymphocytic leukemia. Leukemia, 6, 1229–1231. , , & (
- 1998) Urbanization and childhood leukaemia in Taiwan. International Journal of Epidemiology, 27, 587–591. , & (
- 1999) Epidemiology of Childhood Cancer. IARC Scientific Publications, No. 149, IARC, Lyon. (
- 1991) HHV-6 genome in T-cell acute lymphoblastic leukaemia. The Lancet, 16, 1277–1278. , & (
- 2000) Human herpes virus 6 in hematologic diseases in China. Haematologica, 85, 458–463. , , , , , , & (
- 2002) Day-care attendance and risk of childhood acute lymphoblastic leukaemia. British Journal of Cancer, 86, 1419–1424. , , , , , & (
- 1999) JC and BK virus sequences are not detectable in leukaemic samples from children with common acute lymphoblastic leukaemia. British Journal of Cancer, 81, 898–899. , , , & (
- 2001) Screening for herpes virus genomes in common acute lymphoblastic leukemia. Leukemia, 15, 415–421. , , , , , , & (
- 2003) Increasing incidence of childhood leukemia in Northwest Italy, 1975–98. International Journal of Cancer, 105, 552–557. , , , , , & (
- 1987) The inter-regional epidemiological study of childhood cancer (IRESCC): a case–control study of aetiological factors in leukaemia and lymphoma. Archives of Disease in Childhood, 62, 279–287. , , , , , , , , , & (
- 1999) Pre- and perinatal risk factors for childhood leukaemia and other malignancies: a Scottish case control study. British Journal of Cancer, 80, 1844–1851. , , , & (
- 2000) Temporal increase in the incidence of childhood peak precursor B-cell acute lymphoblastic leukaemia seen in North West England. The Lancet, 356, 485–486. , , & (
- 2001) Examination of temporal trends in the incidence of childhood leukaemias and lymphomas provides aetiological clues. Leukemia, 15, 1612–1618. , , , , & (
- 2002) Space-time clustering analyses of childhood acute lymphoblastic leukaemia by immunophenotype. British Journal of Cancer, 87, 513–515. , & (
- 2003) Geographical and ecological analyses of childhood acute leukaemias and lymphomas in north-west England. British Journal of Haematology, 123, 60–65. , , , & (
- 2002) SEER update of incidence and trends in pediatric malignancies: acute lymphoblastic leukemia. Medical and Pediatric Oncology, 39, 554–557. , , & (
- 1996) Month-of-birth and incidence of acute lymphoblastic leukaemia in children. Leukemia and Lymphoma, 23, 85–92. , & (
- 1995) Childhood leukemia in metropolitan regions in the United States: a possible relation to population density? Cancer Causes and Control, 6, 383–388. (
- 1989) Contribution to the discussion at the Royal Statistical Society meeting on cancer near nuclear establishments. Journal of the Royal Statistical Society, Series A, 152, 376. & (
- 2002) Association of early life factors and acute lymphoblastic leukaemia in childhood: historical cohort study. British Journal of Cancer, 86, 356–361. , , , , , , & (
- 2002) Perinatal exposure to infection and risk of childhood leukaemia. Medical and Pediatric Oncology, 38, 391–397. , , , & (
- 2000) Patterns of infection and day care utilization and risk of childhood acute lymphoblastic leukaemia. British Journal of Cancer, 82, 234–240. , , , , , , , & (
- 1989) A case-control study of non-T cell acute lymphoblastic leukaemia of children in Hokkaido, Japan. Journal of Epidemiology and Community Health, 43, 352–355. & (
- 2003) Childhood cancer in relation to infections in the community during pregnancy and around the time of birth. International Journal of Cancer, 104, 772–777. , & (
- 2002) Birth weight, ethnicity, and occurrence of cancer in children: a population-based, incident case-control study in the state of Texas, USA. Cancer Causes and Control, 13, 595–602. , , , , , & (
- 2004) Birth weight and other risk factors for acute leukemia in the Jerusalem perinatal study cohort. Cancer Epidemiology Biomarkers and Prevention, 13, 1057–1064. , , , , , , & (
- 1996) Childhood leukaemia in Europe after Chernobyl: 5 year follow-up. British Journal of Cancer, 73, 1006–1012. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , & (
- 1998) (eds) International Incidence of Childhood Cancer, Vol. II. IARC Scientific Publications, No. 144, IARC, Lyon. , , , , , , & (
- 2002) Population mixing, childhood leukaemia, CNS tumours and other childhood cancers in Yorkshire. European Journal of Cancer, 38, 2033–2040. , , , & (
- 2001) Childhood cancer registrations in the developing world: still more boys than girls. International Journal of Cancer, 91, 402–406. & (
- 2004) Fathers’ occupational contacts and risk of childhood leukemia and non-Hodgkin lymphoma. Epidemiology, 15, 352–356. , & (
- 1970) Cat leukaemia. British Medical Journal, i, 755. (
- 2002a) Day-care, early common infections and childhood acute leukaemia: a multicentre French case-control study. British Journal of Cancer, 86, 1064–1069. , , , , , , , , , & (
- 2002b) Breast-feeding, fetal loss and childhood acute leukaemia. European Journal of Pediatrics, 161, 235–237. , , , , , , , , , & (
- 1993) Age of exposure to infections and risk of childhood leukaemia. British Medical Journal, 307, 774. , , , , , , & (
- 1997a) The risk profile of childhood leukaemia in Greece: a nationwide case-control study. British Journal of Cancer, 76, 1241–1247. , , , , , , , , , & (
- 1997b) Aggregation of childhood leukemia in geographic areas of Greece. Cancer Causes and Control, 8, 239–245. , , , , , , , , , & (
- The Childhood Haematologists-Oncologists Group (2001) Evidence on the infectious etiology of childhood leukemia: the role of low herd immunity (Greece). Cancer Causes and Control, 12, 645–652. , , , , , & for
- 1974) A case-control approach to examine diseases for evidence of contagion, including disease with long latent periods. Biometrics, 30, 263–279. & (
- 1966) Testing for homogeneity: II. The Poisson distribution. Biometrika, 53, 183–190. & (
- 2003) Human polyomavirus DNA is not detected in Guthrie cards (dried blood spots) from children who developed acute lymphoblastic leukemia. Medical and Pediatric Oncology, 40, 219–223. , , , , , , & (
- 2002) Birth characteristics and leukemia in young children. American Journal of Epidemiology, 155, 603–613. , & (
- 1994) Leukaemia risk and social contact in children aged 0–4 years in southern England. Journal of Epidemiology and Community Health, 48, 601–602. , , & (
- 1997) Leukaemia and non-Hodgkin's lymphoma in children and young adults: are prenatal and neonatal factors important determinants of disease? British Journal of Cancer, 76, 406–415. , & (
- 2000) Early child-care and preschool experiences and the risk of childhood acute lymphoblastic leukaemia. American Journal of Epidemiology, 152, 1136–1144. , & (
- 1999) Seasonal variations in the diagnosis of childhood cancer in the United States. British Journal of Cancer, 81, 549–553. , , , , & (
- 1996) Sero-epidemiological analysis of the risk of virus infections for childhood leukaemia. International Journal of Cancer, 65, 584–590. , , , , , , , , , & (
- 1999a) Association of childhood leukaemia with factors related to the immune system. British Journal of Cancer, 80, 585–590. , , , & (
- 1999b) Association of childhood cancer with factors related to pregnancy and birth. International Journal of Epidemiology, 28, 631–639. , , , & (
- 2002) Investigation of TT virus in the etiology of pediatric acute lymphoblastic leukemia. Pediatric Hematology and Oncology, 19, 543–551. , , , & (
- 2002) Birth characteristics, maternal reproductive history, hormone use during pregnancy, and risk of childhood acute lymphocytic leukemia by immunophenotype (United States). Cancer Causes and Control, 13, 15–25. , , , , , , & (
- 1997) Considerations on a possible viral etiology for B-precursor acute lymphoblastic leukemia of childhood. Journal of Immunotherapy, 20, 89–100. (
- 1998) Evidence that childhood acute lymphoblastic leukemia is associated with an infectious agent linked to hygiene conditions. Cancer Causes and Control, 9, 285–298. , , , , & (
- 1999) Investigation of leukemia cells from children with common acute lymphoblastic leukemia for genomic sequences of the primate polyomaviruses JC virus, BK virus, and simian virus 40. Medical and Pediatric Oncology, 33, 441–443. , , , , , & (
- 2001) Seasonal variation in month of birth and diagnosis of early childhood acute lymphoblastic leukaemia. Journal of the American Medical Association, 285, 168–169. , , & (
- 1996) Effect of population mixing and socioeconomic status in England and Wales, 1979–85, on lymphoblastic leukaemia in children. British Medical Journal, 313, 1297–1300. & (
- 1998) Evidence that an HLA-DQA1-DQB1 haplotype influences susceptibility to childhood common acute lymphoblastic leukaemia in boys provides further support for an infection-related aetiology. British Journal of Cancer, 78, 561–565. , , , , , , , , & (
- UKCCS Investigators (2002) United Kingdom Childhood Cancer Study. Genetic susceptibility to childhood common acute lymphoblastic leukaemia is associated with polymorphic peptide-binding pocket profiles in HLA-DPB1*0201. Human Molecular Genetics, 11, 1585–1597. , , , , , , , , &
- UKCCS Investigators (2001) Breastfeeding and childhood cancer. British Journal of Cancer, 85, 1685–1694.
- 1986) Childhood leukaemia and infectious diseases in the first year of life: a register based case-control study. American Journal of Epidemiology, 124, 590–594. , & (
- 1975) A test for seasonality of events with a variable population at risk. British Journal of Preventive and Social Medicine, 29, 18–21. & (
- 2004) Childhood leukaemia incidence and the population mixing hypothesis in US SEER data. British Journal of Cancer, 90, 1771–1776. , & (
- 1998) Seasonal variations in the onset of childhood leukaemia and lymphoma. British Journal of Cancer, 78, 119–124. , , , , , , & (
- 1997) Birth characteristics, sibling patterns and acute leukemia risk in childhood: a population-based cohort study. Journal of the National Cancer Institute, 89, 939–947. , , , , , , & (
- 1999) Prenatal origin of acute lymphoblastic leukaemia in children. The Lancet, 354, 1499–1503. , , , , , , , & (
- United Kingdom Childhood Cancer Study investigators (2001) Methylenetetrahydrofolate reductase (MTHFR) polymorphisms and risk of molecularly defined subtypes of childhood acute leukemia. Proceedings of the National Academy of Sciences of the United States of America, 98, 4004–4009. , , , , , &
- 1975) Epidemiology of acute leukemia of childhood in Greece. Nouvelle Revue Francaise D Hematolgie, 15, 649–655. , , , & (