Childhood leukaemias and CNS tumours: Correlation of international incidence rates

https://doi.org/10.1016/j.ejca.2005.07.036Get rights and content

Abstract

Childhood leukaemia has a potential infectious aetiology whilst infections may also be linked to paediatric central nervous system (CNS) tumours. Using data from 29 countries we investigated the correlation between international incidence rates of childhood leukaemia and CNS tumours, focusing on acute lymphoblastic leukaemia (ALL), astrocytoma and ependymoma-subtypes that are hypothesised to have an infectious aetiology. Relationships between incidence rates and national demographic factors were also examined using Pearson’s correlation coefficient to quantify associations. Comparing two diagnostic categories of leukaemia with four groups of CNS tumours, a highly significant positive correlation was found between ALL and astrocytoma (r = 0.57, P = 0.002). Higher rates of ALL and CNS tumours were associated with increased affluence, with the strongest correlation for Gross Domestic Product per capita and CNS tumours (r = 0.70, P < 0.001). National incidence rates of childhood ALL and astrocytomas were highly correlated and this may reflect a common environmental cause whose origin may be infectious in nature. International incidence of ALL and CNS tumours were also correlated with economic related factors. Variation in levels of ascertainment may partially explain this, although childhood environmental exposures related to infections will also be affected by levels of affluence.

Introduction

The aetiology of childhood cancer is likely to be multi-factorial, given the heterogeneity in incidence, mortality and pathology between the diagnostic groups. The proportion of childhood cancers attributed to known genetic or inherited susceptibility is very small and therefore environmental exposures are likely to have an important part to play. For certain tumours in children, exposure to infections have been suggested as possible aetiological agents. Evidence is strongest for acute lymphoblastic leukaemia (ALL), the most common subtype (80%) of childhood leukaemia.1, 2, 3, 4, 5 More recently it has been suggested that childhood central nervous system (CNS) tumours may also have links with infections,6, 7, 8 with the strongest support being for the subtypes of astrocytoma and ependymoma.6, 9 However, the mechanism describing the role infections may play in the development of these conditions, whether it is through direct contact with a specific environmental contagion that may damage DNA or a rare autoimmune response to infection in general, is still unclear.

In the context of a common aetiology of specific tumours, we have investigated whether this was reflected in correlation between incidence rates. A previous international analysis suggested that ALL was highly correlated with diabetes, and that this observation might be explained by factors associated with affluence.10 We have extended this approach by comparing the incidence rates of leukaemia and CNS tumours to investigate whether the hypothesis of an infectious aetiology for ALL, ependymoma and astrocytoma was reflected in positive correlations between these subtypes. We also explored whether there was any correlation between the incidence rates of leukaemias, CNS tumours and their subtypes and demographic characteristics including indicators of affluence.

Section snippets

Patients and methods

World standardised international incidence rates of childhood cancer were extracted for the analysis.11 Subtypes were classified according to the International Classification for Childhood Cancer (ICCC)12 and correlations were calculated using Pearson’s correlation coefficient. Due to small numbers all analyses were repeated using Spearman’s rank correlation coefficient. As multiple tests were being carried out, P-values of 0.01 or less were chosen as being significant, this value was chosen to

Results

A summary of the cancer incidence rates for the 29 countries are given in Table 1.

Correlations were calculated between total leukaemia and CNS tumour incidence and then between subtypes (Table 2). Correlation between CNS tumours and all leukaemias was positive but not statistically significant (r = 0.25, P = 0.20). A highly significant positive correlation was found between ALL and astrocytoma (r = 0.57, P < 0.01). A positive association was also found between incidence rates of ALL and ependymoma but

Discussion

This first analysis of the correlation between international rates of leukaemias and CNS tumours indicated a highly significant positive association between international incidence rates of ALL and astrocytoma. The likelihood that this reflects a real effect is strengthened by the lack of such a strong correlation between the larger groupings of all leukaemias and all CNS tumours. The less significant correlations between ALL and all CNS tumours and astrocytomas and all leukaemias is likely to

Conflict of interest statement

None declared.

Acknowledgements

Richard Feltbower is supported by funding from the Candlelighter’s Trust. We thank Adam Wingfield for his assistance with collecting the demographic data.

References (18)

There are more references available in the full text version of this article.

Cited by (2)

View full text