Long-Term Childhood Cancer Survival and Risk of Second Neoplasms in Children with Cancer in Spain

1. Introduction

Childhood cancer survival rates have increased substantially in recent decades thanks to improvements in treatment [1,2]. Currently, five-year survival rates exceed 80% in almost all European countries, as well as in the United States, Canada, Australia, New Zealand, and several Asian countries [2]. However, childhood cancer survivors experience adverse health effects attributable to either their cancer or its treatment [1,2,3]. Between them, childhood cancer survivors are more likely to develop cancer in adulthood than people of similar age in the general population, due to both host factors and cancer treatments [4,5,6,7]. Several studies have suggested that the risk of developing a second malignant neoplasm (SMN) in survivors at 3 or 5 years is 3 to 15 times higher than the incidence in the general population, and that the cumulative risk 20 years after the original diagnosis ranges from 2% to 12% [8,9]. However, due to changes in treatments, this risk may vary over time. On the other hand, although the main risk factor for a second cancer is treatment (especially high-dose radiotherapy and some chemotherapeutic agents), the risk can also vary depending on the patient's sex, age at diagnosis of the first cancer, and genetic predisposition to cancer [8,9,10]. In the survivor population, the incidence of second cancers does not stabilize over time. [11,12,13,14,15]

Since most children have had limited exposure to known environmental carcinogens and are expected to live for many years after cancer treatment, they offer an opportunity to gain knowledge about the mechanisms by which certain chemotherapeutic agents and radiation can induce neoplastic changes [16]. Furthermore, given all this evidence, it is necessary to emphasize the importance of ensuring continuous follow-up throughout the life of a childhood cancer survivor. The available information has largely derived from studies limited to second cancers diagnosed at least 5 years after the first diagnosis [17].

To our knowledge, this study provides, for the first time, a comprehensive population-based assessment of the risk of developing a second malignancy among childhood cancer survivors in Spain. In this article we present the incidence of childhood cancer in the period 2015-2019, the 5 and 10-year survival for childhood cancer between 1990 and 2019, and the incidence of second cancers in individuals diagnosed with childhood cancer between 1990 and 2009 in Spain. We also report on the most common second cancers that occur in this cohort twenty years after the diagnosis of childhood cancer.

2. Materials and Methods

Study design and data sources

In this retrospective cohort population-based cancer registry study, all Spanish population-based cancer registries pertaining to the Spanish Network of Cancer Registries (REDECAN) were invited to participate. Twelve registries participated, covering 14 provinces and two islands. Three independent analyses were performed: One on the incidence of childhood cancer (0 to 14 years) for the period 2015-2019, another on 5- and 10-year survival of childhood cancer for the periods 1990-1999 and 2000-2009, and finally, another on the risk of developing SMN in children with incident cancer during the period 1990-2009 and the sub-periods 1990-1999 and 2000-2009. Table A1 shows the registries included in each of the analyses as well as the periods covered for each of them. In 2000, the proportion of the Spanish child population (ages 0-14) covered by these registries was 31.5% and in 2017 it was 32.2%. The 0-14 age group represented 14.8% of all ages in 2000 and 14.9% in 2017.

All registries used the 3rd edition of the International Classification of Diseases for Oncology [18] for coding the tumor site, morphology, behavior, multiple primary tumors, and basis of diagnosis. Tumors were classified centrally according to the 3rd edition of the International Classification of Childhood Cancer (ICCC-3) [19]. This study includes all malignant tumors of any site, as well as intracranial and intraspinal tumors of benign and uncertain behavior diagnosed before the age of 15 in the study populations.

SMNs were defined as neoplasms on a new location, that were not direct spread or metastases of the primary neoplasm, or the neoplasms on the same location as the primary ones but of different histological type and they were selected based on the International Rules for Multiple Primary Cancers of 2004 [20]. Cancers diagnosed after the diagnosis of a SMN were not included in the analysis. SMN diagnosed within two months of the first cancer were not excluded, assuming that, virtually, all childhood cancers detected shortly after a first childhood cancer would have been diagnosed later anyway. Cancer data included individual records of cases with codes for the following variables: sex, age, date of incidence, and tumor sequence (i.e., the numerical order of occurrence of the neoplasm), site, morphology, behavior, most valid basis of diagnosis, data of last follow-up, and vital status.

Population estimates as of July 1 of each year during the study period for each province or island were obtained from the National Institute of Statistics (INE) [21].

Table 1 shows the participating registries, as well as the territory (province/s or island/s) covered by each of them, the population covered in 2000 and in 2017, the number of childhood cancers diagnosed by period, and the percentage of cases diagnosed only by death certificate (%DCO) and of cases with morphological verification (%MV) during the period 2015-2019. Data were obtained and analyzed from de-identified cases of the Spanish Network of Cancer Registries (REDECAN) joint database.

Procedures

All cases were extracted from the REDECAN joint database, which is systematically evaluated for data quality. All questionable cases and issues raised were referred to the contributing registries with a request for correction or a response. This iterative process results in an improvement in the overall quality of the data included in the analyses. Table 1 shows two quality indicators for each registry of the period 2015-2019: the proportion of morphologically verified cases and the proportion of cases registered solely from the death certificate.

Statistical Analysis

Incidence

To find out the recent incidence of childhood cancer in Spain, we calculated the absolute number of incident cases (cases) and the age-standardized incidence rates (ASIRw) using the World standard population [22] for children 0–14 years (Table A2) by sex, expressed per million child-years, for the period 2015-2019. We calculated the incidence sex ratios by dividing the number of incident cases in boys by the number of incident cases in girls.

Survival

We calculated 5-year and 10-year age-adjusted observed survival and their confidence intervals (95% CI), which in children corresponds very closely to relative survival since competing risks of death are negligible. Survival for the periods 1990-1999 and 2000–2009 was estimated from all cases diagnosed during these periods, irrespective of the potential follow-up, using the complete survival approach because almost all children had been followed up for at least 10 years by December 31, 2019. Survival estimates were calculated with the Kaplan-Meier method. In the survival analyses, cases registered solely from the death certificate and cases diagnosed incidentally by autopsy were excluded.

To ensure comparability between sexes and time periods, and with European figures for survival time analyses for a given cancer, we standardized by the age distribution of all European children diagnosed with cancer using the same weights as those used by Botta et al. [23] using four age classes (<1 year, 1–4 years, 5–9 years and 10–14 years) (Table A3).

We calculated survival for all tumors as a whole and for 15 specific categories of ICCC-3: Acute lymphoid leukaemias (ICCC category Ia), Acute myeloid leukaemias (Ib), Hodgkin lymphoma (IIa), Non-Hodgkin lymphoma except Burkitt lymphoma (IIb), Burkitt lymphoma (IIc), CNS and miscellaneous intracranial and intraspinal neoplasms (III), Ependymomas and choroid plexus tumor (IIIa), Astrocytomas (IIIb), Intracranial and intraspinal embryonal tumors (IIIc), Neuroblastoma and ganglioneuroblastoma (IVa), Retinoblastoma (V), Nephroblastoma and other non-epithelial renal tumors (VIa), Osteosarcomas (VIIIa), Ewing tumor and related sarcomas of bone (VIIIc), and Rhabdomyosarcomas (IXa). Non-melanoma skin cancers were not included because most cancer registries do not register these malignant neoplasms.

Risk of developing a second cancer

The standardized incidence ratio (SIR) was calculated as the ratio of the observed number of SMNs to the number expected if patients in the cohort had the same cancer rates as the general reference population. The observed number of cases included all SMNs diagnosed in each cohort (by sex and cancer type) during each defined time period. The expected number of cases was calculated by multiplying the cumulative observed person-years by the incidence rates for cancer site, sex, five-year age group, and calendar year in the general population. The rates of the general population were obtained from the population-based cancer registries that participated in the study.

In each patient, person-years at risk were defined as the period between the first childhood cancer diagnosis and the date of the second cancer diagnosis, the date of death, or the date of end of follow-up, whichever occurred first.

The SIR was calculated by sex, by period of the first diagnosis of childhood cancer (1990-1999, 2000-2009 and 1990-2009) and by time between the first and second cancer (<1 year, 1-4 years, 5-9 years and 10-19 years) for all tumors as a whole. The calculation for two different periods (1990-1999 and 2000-2009) was performed to observe possible differences in risk over time.

The SIR was also calculated based on the type of the first tumor (12 groups of the ICCC-3) for the 20-year period between the first and second cancer cases, by sex.

Finally, the SIR was also calculated based on the type of SMN according to 25 specific ICD-10 categories (Lip, oral cavity and pharynx (C00-14), esophagus (C15), stomach (C16), colon (C18), rectum (C19-C21), liver (C22), gallbladder and biliary tree (C23-C24), pancreas (C25), larynx (C32), trachea, lung and bronchus (C33-C34), skin melanoma (C43), breast (C50), cervix uteri (C53), corpus uteri (C54), ovary (C56), prostate (C61), testis (C62), kidney (C64), urinary bladder (C67, D09.0, D41.4), brain and central nervous system (C70-C72), thyroid (C73), Hodgkin’s lymphoma (C81), non-Hodgkin lymphoma (C82-C86,C88), myeloma (C90) and leukemia (C91-C95)), and a category called “Other” which covers all the rest of the types of malignant tumors, for the 20-year period between the first and second cancer and by sex. Non-melanoma skin cancers were not included because most cancer registries do not register these malignant neoplasms.

We included all secondary cancers, regardless of the time elapsed between the first and second cancers.

The assumption that the observed number of SMN followed a Poisson distribution was used to calculate 95% confidence intervals (95% CI). Results are considered statistically significant if 95%CI does not include 1.

The Excess Absolute Risk (EAR) was calculated by subtracting the expected number of SMNs from the observed number of SMNs and dividing the difference by the person-years at risk, expressing the number of cases in excess or deficit by 10,000 person-years at risk.

Finally, we calculated the observed age-specific incidence rates of SMNs in the childhood cancer cohort of the study and compared them with the age-specific incidence rates of the general population covered by the cancer registries in Spain.

All the analyses were computed using R software (version 4.5) [24].

3. Results

3.1. Incidence

31.56% of the childhood Spanish population aged 0–14 years (contributing 10,942,347 person-years) was covered by the registries included in the incidence calculations for the period 2015-2019 (Table 1 and Table A1). During this quinquennium, 1,910 childhood tumors were diagnosed in the population covered by this study. The ASIRw of childhood cancer in Spain during 2015–2019 was 181.3 cases per million person-years, with higher rates observed in boys (192,5) than in girls (170.1). The sex ratio of the number of cases was 1.20. Leukaemias were the ICCC-3 group with the highest incidence (53.4), followed by CNS neoplasms (45.2), lymphomas (23.7), and neuroblastoma (13.1). Table 2 shows the incidence of the 12 ICCC-3 groups by sex.

3.2. Survival

The 5-year age-standardized observed survival rate for all cancers and both sexes combined increased significantly from the period 1990-1999 (74.1; 95% CI: 72.5-75.8) to the period 2000-2009 (77.8; 95% CI: 76.4-79.3). In boys these values were 72.3 and 75.9, and in girls 76.5 and 80.4. The 10-year survival rates also increased between the two periods: 71.3 and 75.5 in both sexes combined, 69.4 and 73.6 in boys, and 73.9 and 78.0 in girls (Table 3 and Figure 1 and Figure 2).

For both sexes combined and the period 2000-2009, the highest 10-year survival rate was for Retinoblastoma (98.9), followed by Hodgkin lymphomas (95.0), Nephroblastoma and other non-epithelial renal tumors (90.3), Burkitt lymphomas (83.9), Non-Hodgkin lymphomas (83.5), Lymphoid leukaemias (81.0) and Astrocytomas (80.5). Below 80% were Neuroblastoma and ganglioneuroblastoma (69.7), CNS and miscellaneous intracranial and intraspinal neoplasms (63.5), Rhabdomyosarcomas (62.3), Acute myeloid leukaemias (60.2). The lowest rates corresponded to Osteosarcomas (54.3), Ewing tumor and related sarcomas of bone (51.7), Ependymomas and choroid plexus tumor (51.1), and Intracranial and intraspinal embryonal tumors (44.0) (Table 3).

Between the periods 1990-1999 and 2000-2009, 5-year survival rates increased significantly by an average of 3.7 percentage points across all cancers combined. By tumor type, only acute lymphoblastic leukemia showed a significant increase, but all other cancer types showed non-significant increases between 0.6% and 10.0%, except for three: Burkitt lymphoma (-0.5%), osteosarcomas (-7.8%), and rhabdomyosarcomas (-6.8%). Between the same two decades 1990-1999 and 2000-2009, 10-year survival rates increased by an average of 4.2% across all cancer types combined. Acute lymphoblastic leukemias, Hodgkin lymphomas and non-Hodgkin lymphomas showed significant increases. Across the three most common ICCC-3 groups combined (leukaemias, lymphomas, and CNS tumors) survival rates increased by 5.2% (Table 3).

3.2. Risk of developing a second malignant neoplasm

Table A4 shows the characteristics of childhood cancers included in the second neoplasms risk study, by period. Of the 3,834 children under 15 years of age diagnosed with cancer during 1990–2009, 62 (1.6%) were diagnosed with a SMN during 48,964.01 person-years of follow-up (median follow-up, 15.15 years; interquartile range [IQR], 4.51–20.00 years). The mean time between first and second diagnosis was 12.29 years (IQR, 7.70–18.25 years) for childhood patients diagnosed of first cancer between 1990 and 1999 and 7.85 years (IQR, 4.11–11.29 years) for childhood patients diagnosed of first cancer between 2000 and 2009. The mean time between first and second diagnosis in the second decade was shorter because childhood cancers in the second decade were followed for a shorter time (Table A4).

For both sexes combined, the SIR of a SMN in patients with any first childhood cancer diagnosed between 1990 and 2009 was 5.67 at 20 years after the diagnosis of the first cancer. The relative risk was 12.44 during the first year, 5.99 between the first and fourth years, 4.05 between the fifth and nineth years, and 5.60 between the tenth and nineteenth years after the diagnosis of the first cancer. These values were higher, although not statistically significant, for patients with their first childhood cancer in the period 2000–2009 than for patients with their first childhood cancer in the period 1990–1999 (Table 4).

Likewise, these risk values were higher, although not statistically significant, in women than in men 20 years after the diagnosis of the first cancer (6.34 versus 5.15), although they were lower in women during the first year after the diagnosis of the first cancer (4.51 vs 17.60). Similar patterns were observed in the ten-year periods of first diagnosis 1990-1999 and 2000-2009 (Table 4).

The EAR 20 years after the first diagnosis of childhood cancer was 10.43 per 10,000 person-years at risk. The excess was higher, although not statistically significant, in women (11.44) than in men (9.60), and in the 2000-2009 period group (12.04) than in the 1990-1999 period group (9.03) (Table 4).

According to age at first cancer diagnosis, all age groups showed significantly elevated SIRs: 6.30 for those aged 0 to 4 years, 5.19 for those aged 5 to 9 years, and 5.47 for those aged 10 to 14 years (Table 5).

The risk of developing a SMN of any type 20 years after the diagnosis of a childhood cancer varied according to the ICCC-3 group of the first childhood cancer (Table 6). Seven ICCC-3 groups presented significantly elevated SIRs: Other malignant epithelial neoplasms and malignant melanomas (7.52), CNS tumors (SIR=7.35), Bone tumors (6.87), Lymphomas (6.13), Soft tissue sarcomas and other extraosseous sarcomas (6.12), Neuroblastoma (6.02) and Leukaemias (4.91). The ICCC-3 groups with the highest EAR per 10,000 person-years were: Other malignant epithelial neoplasms and malignant melanomas (18.46), Bone tumors (16.63), CNS tumors (14.02), Lymphomas (13.56), and Soft tissue sarcomas and other extraosseous sarcomas (11.84). By sex, the statistically significant SIRs were higher in women for Leukaemias (6.28 vs 3.93), Lymphomas (7.87 vs 5.25), CNS tumors (9.05 vs 5.74) and Other malignant epithelial neoplasms and malignant melanomas, and in men for Bone tumors and Soft tissue sarcomas and other extraosseous sarcomas (Table 6).

The most common types of SMN with a statistically significant elevated SIR at 20 years of diagnosis were thyroid carcinomas (n=12, 19%, SIR=14.14), leukaemias (n=11, 18%, SIR=7.07), bone sarcomas (n=10, 16%, SIR=16.25), soft tissues sarcomas (n=6, 10%, SIR=13.98), brain and central nervous system tumors (n=5, 8%, SIR=5.14) and breast carcinomas (n=3, 5%, SIR=8.02). By sex, thyroid (33.37), soft tissue (11.73), bone (10.20) and leukaemias (8.27) in men, and soft tissue (17.29), breast (8.02), thyroid (7.82) and bone (6.86) showed significant increased SIRs (Table 7).

Finally, Figure 3 shows the observed age-specific incidence rates of SMN in the study cohort and the age-specific incidence rates of the general population covered by cancer registries in Spain. In all five-year age groups, the observed rates were clearly higher than expected.

4. Discussion

4.1. Incidence

Registration of non-malignant CNS tumors may have varied by registry, which may lead to a slight underestimation of the incidence of all cancers combined. The ASIRw of childhood cancer for both sexes combined in the period 2015-2019 in Spain was 181.3 per million person-years, a figure slightly higher than that obtained by Steliarova-Foucher et al. for Southern Europe during the 2001-2010 period (170.8) [25]. In our study, ASIRw were higher among boys than among girls, with a sex ratio (SR=1.13) similar to that observed in all regions of the world (SR=1.17) [25]. The sex-specific incidence varied according to the diagnostic group, with neuroblastomas, germ cell and gonadal tumors, and epithelial tumors being more common in girls. In Spain, leukaemias, followed by CNS neoplasms, and lymphomas, were the tumors with the highest ASIRw, following the same ranking observed in most regions of the world, with the exception of Africa, where lymphomas are more common than CNS tumors [25]. The remaining groups follow the same order as in Southern Europe [25] except for bone tumors (group VIII) and other malignant epithelial neoplasms (group XI) which rank 7th and 8th in Southern Europe, and 8th and 7th in Spain.

4.2. Survival

Five-year survival rate for all childhood cancer combined in Spain in 2000–2009 was 77.8% (95% CI 76.4–79.3), showing a significant increase of almost four percentage points compared with that of the period 1990–1999 (74.1%; 95% CI:72.5-75.8). Significant progress was observed over time in the most common type of cancer, lymphoid leukaemias. Survival remained stable for the other types of tumors although non-significant increases were observed in most of them (acute myeloid leukaemias, Hodgkin lymphomas, non-Hodgkin lymphomas (except Burkitt lymphoma), CNS tumors, neuroblastoma, retinoblastoma, nephroblastoma, and Ewing sarcoma). Only Burkitt lymphoma, osteosarcomas, and rhabdomyosarcomas showed decreases, although these were not statistically significant.

The 5-year survival rate for childhood cancer in Spain during the period 2000-2009 was 3.5 points lower than the estimated rate for Europe as a whole and 3.3 points lower than the estimated rate for Spain for the following five-year period 2010-2014 [23]. This would indicate that survival rates are still increasing. The increase of 5-year survival rate for all childhood cancer combined in Spain between 1990s and 2000s is similar than that observed in Europe between the periods 2004-2006 and 2010–14 where rates increased from 78% to 81%. En Europe, significant progress of survival over time was observed for almost all cancers, although remained stable for Burkitt lymphoma, non-Hodgkin lymphomas, osteosarcomas, Ewing sarcoma, and rhabdomyosarcomas.

Ten-year survival rates also showed a significant increase between the 1990s and the 2000s. For all tumors combined, the rate rose from 71.3% to 75.5%. The types of cancer with the most significant increases were lymphoid leukemias (from 72.8% to 81.0%), Hodgkin lymphomas (from 88.7% to 95.0%), and non-Hodgkin lymphomas (except Burkitt lymphoma) (from 72.3% to 83.5%). Although the rate for acute myeloid leukemias increased by 10 percentage points (from 50.3% to 60.2%), this increase was not statistically significant.

4.3. Risk of developing a second malignant neoplasm

To the best of our knowledge, this is the first study to analyze the risk of developing a SMN after childhood cancer in Spain. This study examines the risk of SMN in a Spanish population-based cohort of 3,834 childhood cancer survivors diagnosed between 1990 and 2009, with long-term follow-up of up to 20 years.

Most studies on secondary neoplasms exclude cancers diagnosed within the first two months [15,26], within the first six months [27,28], within the first three years [29,30,31], or within the first five years [16,17,32,33,34,35,36] after the original diagnosis to minimize selection bias, i.e., the misclassification of progression or recurrence of childhood cancer as a SMN. This is a subjective decision. We included these cancers because we wanted to estimate the overall risk of developing an SMN, and a significant proportion of second cancers occur during the first few years following the first cancer. Excluding these cancers would significantly underestimate the absolute risk estimate. Furthermore, the high quality of the registries allowed us to rely on the data regarding SMN. Therefore, in our study, the follow-up period for SMN began from the date of diagnosis of the first malignancy as Olsen and colleagues did in in the Nordic countries [7,14]. This allows us to understand the true risk of developing a SMN, regardless of the underlying causes.

The results of this study show an overall SIR for an SMN after 19 years from the first cancer of 5.67, being similar for both sexes. It is difficult to compare the SIR values with those of other studies, as they are influenced not only by patient characteristics, environmental factors and treatment of the first cancer, but also by the time period of the initial diagnosis, the inclusion or not of all tumors diagnosed after the first cancer, the age range included for the first cancer (0-14 or 0-19), the follow-up time of each study or the attained age during follow-up and the type of study. These factors partly explain why SIRs reported for SMN in people who have had childhood cancer vary so widely, ranging from 3 to 20 [11,14,17,27,31,32,36,37].

The SIRs for a SMN changed slightly with the calendar period of first childhood cancer diagnosis, with the highest risks observed for children diagnosed during the most recent diagnosis period (2000-2009) compared with the 1990-1999 period. However, although differences in risk were observed across all follow-up periods (0–, 1–4, 5–9, and 10–19 years), these differences in risk between the two time periods were not statistically significant, probably due to the small number of cases. This increase in the SIR occurred despite advances in radiation treatment during the 1990s and 2000s, and Olsen et al. had already suggested that these differences were likely due to the role of chemotherapeutic agents in the etiology of second cancers [14]. In fact, since the late 1990s, chemotherapy treatment was greatly intensified, and radiotherapy became more aggressive in some cancers. In the 1990s, treatments were not as aggressive, and patients who relapsed died quickly. Subsequently, relapses also began to be treated, and the intensity of treatments, and therefore the risk of SMNs, increased even further. However, we cannot rule out the possibility that a small portion of the excess SMTs is also due to improvements in diagnostic and recording procedures (e.g., neuroimaging studies).

Previous studies showed that relative risk of SMN decreases over time from the first cancer diagnosis, although it remains high decades after the diagnosis of childhood cancer [7,12,15,30,31,32,38]. In our study, the SIR decreases up to 9 years after the first cancer; however, the SIR between 10 and 19 years (5.60) is slightly higher, although not statistically significant, than between 5 and 9 years after the diagnosis of the first cancer. This is consistent with the fact that the relative risk remains elevated during decades, suggesting that the carcinogenic effects of childhood cancer treatment persist throughout life [14). Olsen and collaborators comment that the reduction of the relative risk as patients became older appeared to be a consequence of the age-dependent increase in background rates (unrelated to radiation treatment or chemotherapy), rather than a moderation of the carcinogenic effect associated with treatment for childhood cancer.

Conversely, in large and very long follow-up studies, the EAR attributable to the status of former childhood cancer patient increases throughout life (as the years pass since the diagnosis of the first cancer). Therefore, the number of patients with SMNs after childhood cancer continues to rise not only because of a growing number of long-term survivors but also because the average age of the childhood cancer survivor population will increase (Olsen, 2009). In our study, although the EAR decreases in the first few years after childhood cancer, the EAR during the first nine years after childhood cancer (7.85; 95% CI:4.65-12.09) is lower than the EAR between 10 and 19 years after childhood cancer (14.35; 95% CI:8.97-21.31).

A younger age at the time of diagnosis of the first cancer has been associated with a higher relative risk (RR) of developing SMNs [7,12,15,26,28,31,32,39]. In our study, although we observed elevated SIRs in all three age groups, the SIR for children aged 0 to 4 years (6.30; 95% CI: 3.94–9.55) was only slightly higher —and not statistically significant— compared with those diagnosed at ages 5 to 9 and 10 to 14 years.

Despite the small number of cases, the SIR for SMNs in our cohort was statistically elevated in seven of the twelve childhood cancer ICCC-3 groups as their first cancer (ranging from 4.91 for leukaemias to 7.52 for Other malignant epithelial neoplasms and malignant melanomas) and greater than 1.0—although not statistically significant—in three additional groups. ICCC-3 groups I to IV (leukaemias, lymphomas, CNS neoplasms and neuroblastomas) accounted for 69% of the SMNs.

In absolute numbers, the most SMNs common tumors according to the ICD-10 were thyroid (n=12), leukaemias (n=11), bone (n=10), soft tissue (n=6), central nervous system (n=5), and breast (n=3). The highest relative risk of second cancers was observed for bone sarcomas (SIR=16.74; 95% CI:7.74–30.0), thyroid carcinomas (SIR=14.14; 95% CI:7.27-24.77), and soft tissue sarcomas (SIR=13.98; 95% CI:5.03-30.63). These three types of second cancers are also among those that show higher SIRs in multiple studies [14]. Significantly elevations were also observed in breast cancer, other and unspecified malignancies, leukaemias, and CNS and miscellaneous intracranial and intraspinal neoplasms.

The main strength of this study lies in the fact that it is based on the most recent and high-quality population data available on the incidence and survival of childhood cancer, as well as on the risk of developing a second cancer after having suffered from childhood cancer in Spain. The use of data from high-quality cancer registries to identify and verify SMNs resulted in a virtually complete and unbiased report, avoiding potential selection biases due to nonparticipation, loss to follow-up, or differential reporting due to referral patterns among treatment centers. Follow-up through data linkage reduced the loss of subjects over time.

One limitation of this study is the relatively small numbers in some risk categories, which prevents a meaningful examination of some types of second cancers diagnoses.

Biases due to potentially better ascertainment of patients with a SMN compared with those with a single tumor, which constitutes a possible surveillance bias, could have resulted in an overestimation of relative risks. However, screening programs for asymptomatic adult childhood cancer survivors have not been common in Spain; therefore, this type of bias would be of minor importance.

Finally, for more than 50 years, SMNs in childhood cancer survivors has been recognized as a late sequela of treatment [40], and some studies have shown significant heterogeneity in relative risks according to treatments, with the highest-risk therapeutic group being those exposed to both, radiotherapy and chemotherapy [31]. A major limitation of this study is the lack of detailed treatment information, which precludes direct assessment of treatment-related risks. Nevertheless, the use of high-quality population-based registries ensures robust and generalizable estimates.

5. Conclusions

In the 2000s, the five-year survival rate for childhood cancer in Spain was almost 80%, but survivors had a high risk of developing a second cancer for at least 20 years after their first diagnosis. These findings highlight the need to balance treatment efficacy with long-term safety and to strengthen long-term follow-up strategies. Reducing treatment-related toxicity while maintaining survival gains should be a key priority in pediatric oncology.