Childhood cancer incidence is increasing worldwide; developing countries have higher incidence and mortality as compared to developed countries [1]. Globally top-five childhood cancer sites are leukaemia, lymphoma, central nervous system (CNS), kidney and liver, with boys showing a higher proportion of cancer than girls [2,3]. Age standard incidence rate (ASIRs) in India for childhood cancer were 91 per one million and 65 per one million in boys and girls respectively [3].

According to a recent report based on 28 population based cancer registries (PBCRs), the proportion of new childhood cancer to total cancer varied from 0.8% - 4.7% in boys and 0.5% - 2.6% in girls during 2012-2016. This may be due to variation in environmental exposures or biological susceptibility within Indian regions [4]. Delhi showed the highest proportion of childhood cancer for both boys (4.7%) and girls (2.6%) compared to other region of the country. Delhi observed highest ASIRs among boys (203.1 per one million) as well as among girls (125.3 per one million) based on 2012-2014 data. The various studies have been conducted to study the temporal trend of different cancer sites other than childhood cancer in Delhi [5,6]. We examined the patterns and temporal trend of childhood cancer in urban Delhi from 1990-2014.

The study is based on data extracted from Delhi PBCR, one of the oldest cancer registries of India established in 1986, for all new childhood cancer cases diagnosed between 1990 and 2014. The new cases of top five childhood cancers – Leukaemia (C91-C95), Lymphoma (C81-C85, C96), Central Nervous System (C70-C72), Bone (C40-C41), and Retinoblastoma (C69) were summarized according to gender into five 5-year period (1990-1994, 1995-1999, 2000-04, 2005-2010, 2010-2014). This registry fulfilled the IARC data quality standards and the data was published in cancer incidence in five continents volumes IX and volume X [7,8]. The international classification of disease for oncology (ICD-O) code 9th revision was used for the period 1988-2000 and 10th revision was utilized for the period 2001-2014.

According to the 2011 census, the total population of Delhi was 1,67,53,235 with 97.5% of people living in urban areas. Out of these, 0-14 years population was 45,65,319 and about 97.1% of 0-14 them live in urban areas of Delhi.

The target population for 0-14 years used in determining the incidence rates of respective years was estimated from the 1991, 2001, 2011 census reports of India using difference distribution method [9,10]. These estimates were not adjusted for immigration and fertility changes over the period due to non-availability of adequate data. ASIRs were calculated by the direct method using WHO World standard population distribution, year 2000 [11].

Joinpoint regression model using joinpoint regression programme [12] was applied to assess the magnitude of time trends on the ASIRs of top five childhood cancers and total childhood cancer during 1990-2014 according to gender. ASIR was determined using the formula: Number of new childhood cancer cases in a specific age-group during a period*1000,000/Estimated population of a specific age group during the period.

Simple linear regression was performed to assess the trend of relative proportion of new childhood cancer cases to all age-group cancers. The percentage increase in top-five new childhood cancer cases was calculated by: (childhood cancer cases during 2010-2014 minus childhood cancer cases during 1990-1994) *100 /Childhood cancer cases 1990-1994. The median age was determined using the median formula for group data on combined 25-years data.

The estimated annual percentage changes (EAPCs) in cancer incidence rates was calculated by fitting a jointpoint regression model, assuming a constant rate of change in the logarithm of the annual ASIRs in each segment. The significance of EAPC was tested using asymptotic t-test and considered significant at 5% if 95% Confidence interval (CI) of EAPC does not include zero. The connecting points of the jointpont segment was treated as break. For 25-data points, joinpoint software recommended a maximum 4 joinpoints. This started with a minimum zero i.e straight line. Monte Carlo permutation test with 4499 randomly permuted data sets was applied to test the additional requirement of joinpoint, the obtained P-value was adjusted according to Bonferroni correction due to multiple comparisons [13]. ASIRs show a rising trend when the point estimate and lower limit of 95% CI are >0. In contrast, a decreasing trend can be seen when the point estimate and upper 95% CI are <0. Else, the ASIRs are deemed to be stable over the time period. The multiple comparisons, asymptotic t-test and Monte Carlo permutation test are part of joinpoint software [12].

The percentage of new childhood cancer cases to total cancer cases were significantly decreasing in both genders. This may due be to the falling trend of fertility rate and growing trend of expectancy of age in Delhi as well as in India [14]. The total fertility rate decreased from 4.83 in 1980 to 2.3 in 2015 as per United Nation websites [15]. In Indian urban areas, life expectancy at birth increased from 65.4 during 1990-94 to 71.5 during 2010-2014. The life expectancy at birth in Delhi during 2010-14 was 73.5 [16].

ASIRs of childhood cancer showed an increasing trend during 1990-2014 in boys and in girls during 2005-2014. Our results of rising trend of new childhood cancer cases was akin to Canada [17], Australia, and Taiwan [18,19] (Suppl. Table I). To the best of our knowledge, no Indian study has assessed the time trend analysis of new childhood cancer cases, albeit one study reported rising trend in childhood cancer among the Indian PBCRs comparing three time periods [20]. Leukemia contributed to 36.1% of total childhood cancer cases as also seen in Canada [32.4%; 1992-2010], Australia [32.4%; 1983-2006], Taiwan [33.9%; 1996-2010], Estonia [31.1%;1970-2016], Thailand [36.1%: 1990-2011]. [17-19,21-22]. Globally, the next two commonest childhood cancers are CNS tumors with range 13.9%-22.7%, followed by lymphoma with range 10%-11.2%, albeit in urban Delhi lymphoma and CNS tumors account for with a percentage of 15.0% and 11.0% of all the childhood cancer cases respectively [17-20]. However, in Chennai lymphoma and CNS tumors account for 20% and 11% of all childhood cases respectively [23].

ASIR for childhood cancer in Delhi was 129.8 per one million [95% CI: 127.6-132.1] from 1990-2014 which is closer to the Taiwan (125.0 per one million from 1996-2010) [19] and lower than United States (172.8 per one million between 2007 and 2011) [24] and Australia (157.5 per one million from 1997-2006) [18].

The etiology of childhood cancer is still limited, some of the factors like environmental exposure, genetic, parental smoking, higher birth weight, and high maternal age are associated with most childhood cancers [25,26]. A large US case-control study reported an increase of 8% in overall childhood cancer risk for each 5-year increase in maternal age [27]. Likewise, an increasing trend in median marriage age was observed in India [28]. The main reasons for increasing median age are rise in female workforce and higher education enrolment of females especially in urban India [28]. Indirectly, the median age of first-time mother showed an increasing trend but exact magnitude cannot be estimated.

The change in trends could also be influenced due to shift in coding or registration practices and improvement in diagnostic advancement during the period. The immigrants might also contribute to increase in new childhood cancer cases. Immigration in Delhi steadily increased from 6.34 lakhs during 1961-1971 to 22.2 lakhs during 2001-2011 [29]. However, effect of this increase on childhood cancer cannot be ascertained due to non-availability of data.

The main strength of Delhi PBCR is almost complete coverage and collection high-quality data. The Delhi PBCR data is also included in the International Agency for Research on Cancer scientific publications [7,8]. Although Delhi PBCR collects the data of only those patients who have been residing for at least one year in Delhi but veracity cannot be confirmed. The projected population of each year to calculate the ASIRs does not adjust for immigration and fertility changes over the years which may over-estimate the incidence rate. The subgroups of leukemia and lymphoma trend could not be observed due to a small number of cases. Age-period-cohort (APC) model that assess the effect of age, effect of period, and effect of cohort on the incidence, was not performed in the present study. The join point analysis developed by National cancer Institute, USA, is a robust method to assess the trend analysis especially for cancer incidence and mortality data [12,13].

The trend of overall new childhood cancer cases showed a significant increase especially in the latest decade in Delhi for boys and girls. The contribution of childhood cancer to total cases showed a decreasing trend during this period. Compared to girls, boys had doubled childhood cancer cases during 1990-2014. Leukemia is the most common cancer site and contributed to one-third of total childhood cancers during the 25-year period. The trend and knowledge of present status of childhood cancer helps the public health policy makers as a baseline for future planning and allocation resources.

Note: Supplementary material related to this study is available with the online version at www.indianpediatrics.net

Contributors: All authors approved the final version of manuscript, and are accountable for all aspects related to the study.

Funding: None; Competing interest: None stated.