Childhood and Adolescence Cancers in the Palermo Province (Southern Italy): Ten Years (2003–2012) of Epidemiological Surveillance

Italy has one of the highest paediatric cancer incidence rates in Europe. We compared cancer incidence and survival rates in children (0–14 years) and adolescents (15–19 years) residing in Palermo Province (PP) with statistics derived from Italian and European surveillance systems. We included all incident cancer cases, malignant tumours and non-malignant neoplasm of central nervous system (benign and uncertain whether malignant or benign), detected in children and adolescents by the Palermo Province Cancer Registry (PPCR) between 2003 and 2012. A jointpoint regression model was applied. Annual Average Percentage Changes were calculated. The Besag–York-Mollie model was used to detect any cluster. The 5-year survival analysis was computed using Kaplan-Meier and actuarial methods. We identified 555 paediatric cancer incident cases (90% “malignant tumours”). No difference in incidence rates was highlighted between PPCR and Italy 26 registries and between PPCR and Southern Europe. No jointpoint or significant trend was identified and no cluster was detected. The 5-year overall survival didn’t differ between PP and the Italian AIRTUM pool. A borderline higher statistically significant survival was observed in age-group 1–4 when comparing PPCR to EUROCARE-5. The epidemiological surveillance documented in the PP was a paediatric cancer burden in line with Italy and southern Europe. The study supports the supplementary role of general population-based cancer registries to provide paediatric cancer surveillance of local communities.


Introduction
Even if childhood cancer is a rare event [1], it is one of the major causes of death in the younger ages of life [2]. Childhood cancer has a worldwide impact of about 100,000 deaths every year before the age of 15, more than 90% of which occur in resource-limited countries [3,4]. Despite important A further comparison of cancer's epidemiological impact on childhood and adolescence between PP and data derived from the Italian and European surveillance systems was performed.

Materials and Methods
According to standards and guidelines for cancer registration in Europe provided by European Network cancer registries [26] and the International Agency for Research on Cancer (IARC) [27], we have selected all incidents of tumour cases ("all tumours"), including malignant tumours, as defined according to the fifth digit of the morphology code (behaviour/3), and non-malignant neoplasm of the central nervous system (benign and uncertain whether malignant or benign), coded by using the ICD-O classification III edition (ICD-O-3) [28], which occurred between 2003 and 2012 in resident children (0-14 years-old) and adolescents (15-19 years-old) in the PP. Then, all tumour cases were manually trans-coded by using the third revision of International Classification of Childhood Cancer (ICCC-3) [29].
Furthermore, according to the recommendations for a standard dataset for the European Network of Cancer Registries [26], we have explored the following quality indicators of registration: (a) percentage of microscopic verification (MV%): diagnosis based on cytology and/or histology of a metastasis and/or histology of a primary tumour; (b) percentage of Death Certificate Only (DCO%): diagnosis made from death certificate (no information available except the cause and the date of death); c) percentage of non-microscopic verification (NMV%): diagnosis made before death and based only on clinical-instrumental investigation, except histology. There was no case with an "unknown" basis of diagnosis. Cases microscopically verified ranged between 81.8% and 92.3% for "all tumours" (Table 1a) and between 86.7% and 94.5% for "malignant tumours" (Table 1b), respectively, while only one malignant case was reported as DCO.   We have identified n. 555 neoplasms, n. 500 of which were "malignant tumours" ( Table 1). The tumour cases were further categorized in the 12 tumours' main groups (Table 2).  Absolute and relative frequencies, and male to female ratio when appropriate, were calculated. The cancer incidence rate per million persons-years was standardized on European population (Std-EU) and was computed for the 0-14 age group using the following weights per each age group: 0-4 = 0.8, 5-9 = 0.7, 10-14 = 0.7. Age-specific incidence rates per million persons-years were calculated for the 15-19 age group per year each year within the study period [13]. The rates were calculated using the annual inter-census estimates between 2002-2012 of the resident population, males and females aged 0-19 and by a single year of age, provided by the Italian Statistics Institute (ISTAT) [25]. Further, a comparison between age-specific incidence rates (0, 1-4, 5-9, 10-14, [15][16][17][18][19] in the PP and in the pool of the 26 Italian registries providing childhood cancer data for the period between 1992-2013 (26 Italian registries) [10] was performed using Fisher's test with a statistical significance set at p-value = 0.05. Moreover, the comparison of cancer incidence in children and adolescents between PPCR and the "southern European region" was performed using world standardized population rates (WSR) [11].
The time trends of cancer incidence rates were fitted under a log-linear model by using the Jointpoint Regression Program software [30]. The jointpoint regression model was applied to identify any modification in time trends for Std-EU rates in the 0-14 age-groups and age-specific incidence rates for the 15-19 age-group. The permutation test obtained by the grid search method suggested by Lerman [31] was used in order to determine the number of significant jointpoints between 2003-2012. Given the size of the available time series, the maximum detectable jointpoints was equal to 1. Significance was set at α = 0.05 and the test was adjusted by using the Bonferroni correction.
Any differences in Annual Percentage Changes (APCs) between the two time periods in consideration were calculated.
Annual Average Percentage Changes (AAPCs) were calculated by fitting the squares' regression line to the natural logarithm of the rates. When the regression coefficient β is equal to zero then AAPC equals zero.
The incident cancer cases were localized using a space-time combination of address and date of diagnosis [32]. Standardized incidence ratio (SIR) for the ten-year study period was calculated by indirect method for each PP municipality. To perform the spatial analysis, SIRs were corrected using Besag-York-Mollie (BYM) Bayesian method [33]. SIR-BYM and the post-probability (PProb) maps were generated. PProb higher than 0.95 was considered statistically significant.
The 5-year survival analysis was restricted to "malignant tumours" only. For multiple tumours we followed the IARC-IACR rules [27]. End of follow-up was set at 31 December 2017. Cases lost to follow-up were 0.8%.
The 5-year observed survival and 95% Confidence Intervals (CIs) in the age groups 0-14 and 15-19 were computed applying the Kaplan-Meier method to compare the results with AIRTUM pool, available only for the time period between 2003-2008 [14], while the 5-year observed survival and 95%CIs in the age groups <1, 1-4, 5-9 and 10-14 were calculated by using the actuarial method to make comparisons with the Eurocare-5 study, available only for the period between 2000-2007 [14]. The computed 95%CIs were used to visually compare the survival data observed in the PP with the selected published data [13,14].
The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the Ethics Committee of Palermo "Palermo 1" (N. 05/2017).

Results
We identified 555 incidents of cancer cases (312 males and 243 females), defined as "all tumours" including malignant tumours and non-malignant neoplasm of the central nervous system (benign and uncertain whether malignant or benign), coded by using the ICD-O-3 [21], and registered in children and adolescents by the PPCR between 2003 and 2012 (Table 1a). Within the study cohort, 500 cases (279 males and 221 females) were "malignant tumours" (Table 1b). Moreover, the distribution of incident cases was comparable for genders and age-groups in both "all tumours" and "malignant tumours" categories, the observed frequency being higher in males than in females and the male to female ratio being higher than 1 in all age groups. Table 2 reports the frequency distribution of the incidents of cancer cases according to the 12 main ICCC-3 groups for both 0-14 and 15-19 age categories. Within the age class, 0-14 Leukaemia was the most frequent (28%), followed by CNS neoplasm (25%), Lymphoma & related (11%), Neuroblastoma (10%), Soft tissue sarcoma (6%), Renal and Bone tumours (5% respectively), Germ cell and Carcinoma & Melanoma (3% respectively), Others & unspecified tumours and Retinoblastoma (2% respectively), and Hepatic tumours (1%).
No difference in children was reported when we compared incidences between PPCR (WSR = 185.6; 95%CI: 166.3-204.9) and "Southern European Region" (WSR = 170.9). In the same direction, incidences of cancer in adolescents has not shown any difference between PPCR (ASR = 249.8; 95%CI: 214.7-284.8), "Italy 26 registries" (ASR = 275.4; 95%CI: 262.4-288.4) and "Southern European Region" (ASR = 243.7) (data not shown). Figure 2 represents the results of the jointpoint regression model and the values of the AAPCs for both "all tumours" (a) and "malignant cancers only" (b) in the study period. No jointpoint has been identified in incidence time-trends for childhood and adolescent cancer groups for "all tumours" nor for "malignant cancers only".
Further, no statistically significant trend was detected in any considered age-group, neither for "all tumours" nor "malignant tumours". Figure 3 shows the SIRs by age-group and by municipality of (a) "all tumours", the (b) BYM estimates and their (c) Pprobs. Despite the significance of many municipality SIRs (results not shown), as the maximum Pprob value was of 88%, no significant BYM estimate was highlighted. Therefore, no cluster was identified. There was no difference when comparing incidence rates between PPCR (Std-EU = 182.3; 95%CI: 163.3-201.2) and "Italy 26 registries" (Std-EU = 180.0; 95%CI: 174.0-186.0) for the 0-14 age-group. No difference in children was reported when we compared incidences between PPCR (WSR = 185.6; 95%CI: 166.3-204.9) and "Southern European Region" (WSR = 170.9). In the same direction, incidences of cancer in adolescents has not shown any difference between PPCR (ASR = 249.8; 95%CI: 214.7-284.8), "Italy 26 registries" (ASR = 275.4; 95%CI: 262.4-288.4) and "Southern European Region" (ASR = 243.7) (data not shown). Figure 2 represents the results of the jointpoint regression model and the values of the AAPCs for both "all tumours" (a) and "malignant cancers only" (b) in the study period. No jointpoint has been identified in incidence time-trends for childhood and adolescent cancer groups for "all tumours" nor for "malignant cancers only".
Further, no statistically significant trend was detected in any considered age-group, neither for "all tumours" nor "malignant tumours". Figure 3 shows the SIRs by age-group and by municipality of (a) "all tumours", the (b) BYM estimates and their (c) Pprobs. Despite the significance of many municipality SIRs (results not shown), as the maximum Pprob value was of 88%, no significant BYM estimate was highlighted. Therefore, no cluster was identified.     Table 3 shows the comparison of the 5-year survival rate restricted to "malignant tumours" between PPCR and Italian AIRTUM pool [13] and between PPCR and EUROCARE-5 study [12]. A 5-year overall survival rate of 82% (95%CI: 78-87) for the childhood group and of 86% (95%CI: 81-91) for the adolescent group was documented in the PP during the study period, but no statistically significant difference was reported with regard to the Italian AIRTUM pool (Table 3a). Table 3. Comparison of five-year observed survival (%) and its 95%CI for "malignant tumours": Palermo Province Cancer Registry versus Italian AIRTUM pool (a) and versus Eurocare-5 study (b).   Table 3 shows the comparison of the 5-year survival rate restricted to "malignant tumours" between PPCR and Italian AIRTUM pool [13] and between PPCR and EUROCARE-5 study [12]. A 5-year overall survival rate of 82% (95%CI: 78-87) for the childhood group and of 86% (95%CI: 81-91) for the adolescent group was documented in the PP during the study period, but no statistically significant difference was reported with regard to the Italian AIRTUM pool (Table 3a). A 5-year survival of 85% (95%CI: 74-98) for the <1 age-group, of 86% (95%CI: 80-93) for the 1-4 age-group, of 84% (95%CI: 76-93) for the 5-9 age-group and of 85% (95%CI: 78-93) for the 10-14 age-group was documented in the PP, with a borderline higher statistically significant survival in the age-group 1-4 when comparing PPCR to EUROCARE-5 (Table 3b). age-group was documented in the PP, with a borderline higher statistically significant survival in the age-group 1-4 when comparing PPCR to EUROCARE-5 (Table 3b).

Discussion
This study reports incidences and survival statistics of tumour cases which occurred in childhood (0-14 years) and adolescence (15-19 years) over a ten-year period in the PP. Our results summarize the epidemiological impact of childhood and adolescence tumours on the fifth Italian administrative most populated area within a country documenting one of the highest incidences of cancer rates in Europe [13], allowing us at the same time to make comparisons between both national and European frameworks, particularly the southern European region, by using data derived from a general population-based cancer registry.

Discussion
This study reports incidences and survival statistics of tumour cases which occurred in childhood (0-14 years) and adolescence (15-19 years) over a ten-year period in the PP. Our results summarize the epidemiological impact of childhood and adolescence tumours on the fifth Italian administrative most populated area within a country documenting one of the highest incidences of cancer rates in Europe [13], allowing us at the same time to make comparisons between both national and European frameworks, particularly the southern European region, by using data derived from a general population-based cancer registry.
We preferred to analyse data considering the 0-14 and 15-19 age categories, instead of the commonly used 0-14 and 0-19 age-groups, according to the different characteristics of the tumours occurring within the two chosen age-groups (distribution of neoplasm with heterogeneous prognosis, treatment protocols and diagnostic-therapeutic strategies and host biology) [14] and the well-known difficulties of adolescents to be diagnosed and properly cured. This choice was also related to the availability of data deriving from a GPCR: as its registration rules, unlike paediatric-specialized cancer registries, are not exclusively designed to collect data on paediatric cancers [34,35], GPCRs codify childhood and adolescence incident tumour cases by using the third revision of the ICD-O-3 [28] needing to trans-codify into ICCC-3 [29].
Cancers in childhood and adolescence in the PP were more frequent in males than females both for "all tumours" and "malignant tumours", as expected according to literature [36,37].
Uniformity with national statistics has been highlighted with regard to the relative frequency of tumour types and the burden of disease. In fact, according to the 12 main ICCC-3 groups, the results provided by PPCR showed the same ranking reported by literature [10], the Leukaemia group being the most represented in the 0-14 age category, followed by CNS neoplasm and Lymphoma & related neoplasm groups, while Lymphoma & related neoplasm groups ranked first followed by Carcinoma & Melanoma and Leukaemia groups in the 15-19 age category.
Further, as compared to "Italy 26 registries" [13], the incidence rates per million persons-years in specific age classes <1, 1-4, 5-9, 10-14 and 15-19 reported by the PPCR resulted in a substantial overlapping. Also, the graphical representation of the age-specific incidence rates showed the classical "U" shape documented by literature [36,37].
In the same direction, there was no difference in standardised incidence rates for all the considered age-groups when comparing statistics provided by PPCR to "Italy 26 registries" or to data available for the southern European region, depicting a local and national epidemiological context in line with a wider area documenting the highest incidence rates in children and adolescents as compared to the entire European region where, in the 1990s, an age-standardized cancer incidence rate of 130 per million persons-years was documented among children aged between 0 to 14 years old, whilst, over the period between 1995-2002, the annual overall incidence rate in adolescents (15-19 years) was of 157 per million persons-years [10,12].
Nevertheless, the evidences provided by jointpoint regression model and spatial analysis, taken together, make us confident to state that-despite the relatively short observational time window available -the surveillance system documented the absence of environmental risk areas as well as the occurrence of any specifically relevant event in the PP during the study period.
Within the study limitations, the documented evidences of an improvement over time in registration quality and population coverage of cancer registries [11,38], particularly, immediately after the start-up period, should be taken into account when interpreting our data.
Moreover, despite the fact that clustering for childhood tumours has been reported in literature, particularly for leukaemia, intracranial and intraspinal embryonal tumours, and Hodgkin lymphomas [18,39,40], the scanty number of specific types of incident cancers registered by the PPCR in a relatively short period made us hesitant to explore clusters for leukaemia or other specific tumour sites.
The aforementioned limitations together with the documented evidence of an increase in the overall rates of childhood neoplasm in the last decay [11,12] are crucial points to consider, since GPCRs are currently involved in the communication of risk related to environment and cancers and their mission includes the translation and dissemination of evidence to enable informed decision-making and to empower the general population or other stakeholders, while at the same time preserving a rigorous methodological approach [41]. Therefore, according to the available body of evidence, greater attention should be paid with regard to childhood and adolescence tumours in order to explore the implementation of a complementary methodological approach to the commonly used graphical and visual formats (choropleth maps, tables, etc.) also in the risk communication of cancer paediatric burden to stakeholders, particularly to communities and local authorities [41].
Lastly, the comparison of the 5-year overall survival rate, restricted to "malignant tumours" only, between data provided by PPCR and the Italian AIRTUM pool [13] has documented overlapping values of point estimates and their CIs for both children and adolescents. Moreover, the 5-year survival point estimates are higher in the PP when comparing the PPCR data with Eurocare-5 study [14], representing survival data on a European scale, even if only a borderline higher statistically significant survival observed in favour of the PP for age-group 1-4.
The previous findings can be explained by the presence of a high quality network of specialized centers for the treatment of paediatric cancers in Italy, ensuring specific healthcare pathways and protocols; also, in relation to the patients' migration across the country [13].

Conclusions
The epidemiological surveillance conducted by the PPCR in the PP has highlighted a burden of childhood cancer in line with the Italian and the southern European frameworks.
As authors agree with the recent proposal to "move towards pan-European coverage" by implementing the institution of childhood specialized cancer registries on a national level [35], potential perspectives for the future are to intensify the collaboration between PPCR and the European Network for Cancer Research in Children and Adolescents [42], as well as with clinicians, as already documented by the previous joint experience conducted by the Italian Association of Paediatric Haematology and Oncology and the AIRTUM network [13]. In this context, the progressive adoption of the new guidelines and recommendations, on which staging systems should be adopted by population-based cancer registries for the major childhood cancers, will make it easy to perform comparative studies on incidences and other outcomes of interest, particularly survival [43]. Nevertheless, if it seems appropriate to think globally, then it is necessary to act locally in order to answer the health demands of each specific community. To this end, the experience of the PPCR supports a supplementary role of general population-based cancer registries, as compared to paediatric specialized cancer registries, to carry out childhood and adolescence cancer surveillance with the aim to assess and control the impact of cancerous diseases on local communities [44]. Acknowledgments: Authors didn't receive any grant or funding in support of the research work or for covering the costs to publish in open access. The data that supports the findings of this study are available from Palermo Province Cancer Registry, but restrictions apply to the availability of this data, which was used under license for the current study, and so is not publicly available. Data is however available on demand from the authors upon reasonable request and with permission of Palermo Province Cancer Registry, registrotumoripalermo@unipa.it.

Conflicts of Interest:
The authors declare no conflict of interest.