Our findings suggest evidence of temporal clustering of neuroblastic tumours in children and young people over more prolonged periods of time (between years). There was no evidence of EPV over shorter time periods (i.e. fortnights or months). Temporal clustering was only apparent between quarters for cases aged < 18 months, and, in contrast, only between years for cases aged 18 months to 19 years. Overall, temporal clustering was exhibited only for neuroblastic tumours in girls and not boys.
There is a paucity of literature concerning the aetiology of neuroblastic tumours. Hereditary neuroblastoma occurs in 1-2% of cases, most notably due to germline mutations in ALK [19,20,21]. One study found that aetiological risk factors related to the prenatal and perinatal period may be associated with the age of diagnosis . Another study showed that congenital anomalies and pre-eclampsia were associated with increased risk of neuroblastic tumours at age < 18 months . A pooled analysis of French studies demonstrated that congenital malformations and fetal growth anomalies conferred increased risk of a neuroblastic tumour .
To our knowledge the present study is only the second study to demonstrate temporal clustering amongst cases of neuroblastic tumours, although a specific temporal cluster was identified using different methodology in a study from the province of Cordoba, Argentina . However, the findings of the present study from Canada contrast with those of a previous study from northern England . The Canadian data demonstrated temporal clustering between years, whereas the data from northern England showed that clustering was apparent between fortnights or between months. In addition, the present study found that temporal clustering was only present for females and not for males. This contrasts with the findings from northern England where clustering was exhibited amongst both males and females, but was far more pronounced amongst males. The reasons for these differences between the two studies are not clear. However, we might speculate that different demographics and socioeconomic conditions might lead to markedly distinct patterns of exposure to an unknown aetiological agent or agents. There are similarities in the overall levels of deprivation between Ontario and the whole of the UK. However, there are distinct differences in the nature of deprivation between the two locations [26, 27]. The population of Ontario is approximately 14.7 million, is ethnically diverse and covers a large geographical area of 917,741 km2. Although the province comprises large rural parts, approximately 11.3 million live in metropolitan areas . In contrast, the northern region of England has a total population of approximately 3.1 million and covers an area of 15,337 km2 with a mixture of urban and rural areas. Population density varies widely. Northumberland and Cumbria are the two most sparsely populated counties in England, while Tyne & Wear is one of the most densely populated. Almost 1.4 million people live in the more urban areas of Newcastle upon Tyne, Gateshead, Sunderland, Middlesbrough, North and South Tyneside. The population of northern England is predominantly Caucasian and ethnic minorities accounted for under 2% during the study period. The northern region is one of the most deprived in England [29,30,31].
The lengthier temporal periods seen in the present study suggest more prolonged lag times for the spread of a geographically widespread environmental agent. Alternatively, it is possible that there may have been more variability in the length of the time between development and diagnosis of a neuroblastic tumour in Ontario compared to northern England. The temporal period was shorter for the younger age group (< 18 months), where clustering occurred between quarters, than for the older age group (18 months – 19 years), where clustering occurred between years. This suggests shorter lag times for the younger age group, which is consistent with the previous study from northern England . It is also possible that this pattern of occurrence might reflect a long-term increase in incidence. The differences between the studies in the findings based on gender should be interpreted with caution, as there is no readily apparent explanation. However, we might speculate that this may be related to differences in lag time from exposure to onset, or differences in patterns of exposure to putative environmental risk factors, between males and females. Gender differences in the pattern of occurrence of neuroblastic tumours have been noted previously . It should be acknowledged that chance may play a role in the differences found specifically for temporal clustering between the two studies. Further research is needed to provide a clearer explanation.
The findings from both the present study and the previous study from northern England  provide support for the involvement of a temporally varying environmental agent. Other descriptive studies have analysed space-time clustering and spatial clustering amongst cases of neuroblastic tumours [33,34,35,36,37,38]. However, there was inconsistency between the findings from these studies. This suggests that the environmental agent or agents involved only rarely lead to the initiation of a neuroblastic tumour, or that the studies had varying sensitivity to identify the impacts of these agents.
Some methodological issues should be noted. The method used in this study is based on the idea that the form that any temporal clustering might take is unknown. Consequently, the approach taken does not base the analysis on a particular model of temporal clustering. Muirhead and Potthoff and Whittinghill have considered the power of this approach to detect EPV [5, 6, 18]. Muirhead, with reference to a comparative study organised by the International Agency for Research on Cancer, concluded that this methodology has reasonable power to detect EPV under a range of models for over-dispersion, including those for which the constant variance: incidence assumptions does not hold . If there were seasonal variation that cut across quarters, then the approach used here might well detect EPV. Methods that look specifically for seasonal variation would have greater power to detect such variation. However, other studies provide little evidence of seasonal variation in neuroblastic tumours and, in any case, the focus of the current analysis is on variation that need not be seasonal [40, 41]. The time periods used are arbitrary, but, given that the analysis covers such a long period (more than 30 years), the particular choice of fortnights, months and years should not affect the ability to detect short-term and/or longer-term variation. We recognise that some of the results exhibit large uncertainties in the estimates due to the small number of events, and thus should not be over interpreted.
In conclusion, this study of neuroblastic tumours from Canada has found evidence of temporal clustering. However, the scale of the temporal clustering differs from a previous study from northern England. In contrast to the findings from that study, the Canadian data have demonstrated that the clustering was confined to more prolonged temporal intervals (principally years, rather than shorter periods). The temporal clustering found in the present study may be either characterised by ‘peaks and troughs’ or by a long-term increase in the number of cases. Both of these scenarios are consistent with the involvement of one or more widespread environmental agents in aetiology. It is possible that different environmental agents are involved in Canada and northern England, and this could explain the differences in patterns that has been seen. Further research is needed to identify putative aetiological agents. In addition, larger studies of temporal clustering could be undertaken (e.g. by the recording of date of diagnosis in the International Neuroblastoma Risk Group database, which would require information on the geographical region or country for each case).