The disparity and trend of childhood cancer burden in survival perspective: a systematic analysis for the global burden of disease 2021.

Introduction
Among diseases affecting children and adolescents, cancer is one of the few that cannot be prevented[1]. Although most childhood cancers have no identifiable cause, they can be cured through early diagnosis and high-quality multidisciplinary treatment. However, childhood cancer remains a leading cause of death and poor health, resulting in a significant loss of life expectancy. It was calculated that cancer was the ninth leading cause of disability-adjusted life years (DALYs) in children and adolescents worldwide, mainly consisting of years of life lost (YLL)[2]. The survival rate of childhood cancer varies widely by regions, and the disparity is extremely evident, for example, the 5-year survival rate ranged from 8.1% (4.4–13.7) in eastern Africa to 83.0% (81.6–84.4) in North America[3]. While in Europe, the 5-years survival rate has reached 81% for all childhood cancers over the past 10 years ago[4]. In order to transform the current health unfairness, the World Health Organization (WHO) developed the CureAll Framework with a target of 5-years survival rate of at least 60% by 2030, promoting to more lives saving and life quality of children with cancer.
Indicators for measuring population survival include survival rate, DALYs, mortality-to-incidence ratio (MIR), and death. Previous studies have analyzed the global disease burden of childhood cancer from multiple perspectives. The GBD 2017 study, which used DALYs for the first time, demonstrated that the global burden of childhood cancer was substantial and disproportionately affected populations in resource-limited settings[2]. The study noted the importance of childhood cancer to global cancer and child health and provided a baseline burden reference for management. Therefore, it did not analyze the regional distribution or temporal trends for specific cancers. Another study reported cancer burden in children under 15 years from 1990 to 2019, calling for more health attention in low Sociodemographic Index (SDI) region[5]. In addition, a study used CONCORD’s childhood cancer registry records from 1995 to 2014 to estimate 5-year survival rates across countries and regions and the improved effects of different policy interventions[4]. The study suggested that future investments at the health system and facility level should be strengthened to improve global childhood cancer outcome. However, the CONCORD study only included the three most common types of cancer and did not analyze regional differences of specific cancer. Clearly, more evidence is needed to support the implementation of the WHO CureAll framework.
CureAll framework stated that the childhood cancer care pathway was from symptom onset through diagnosis and treatment and ultimately to survivorship or end-of-life care. The healthcare system must not only prevent premature death in children from cancer but also strive to prolong survival and improve life quality. The 5-year survival directly reflect the survival possibility, but it cannot be obtained in a timely and open manner on a global scale. MIR is a proxy measurement for survival rates, which is defined as the number (or rate) ratio of the death to incidence in a given year[6]. This metric is often used to evaluate healthcare system efficacy and fairness, such as adult tumors[7], HIV[8], stroke[9], etc., but it has seldom been described in childhood cancer. Notably, the MIR cannot reflect the long-term survival impact of cancer on children and adolescents. There is a time lag between incidence and death, and its accuracy is influenced by the quality of cancer diagnosis, reporting and registration. Generally, the DALY measures lifespan loss due to premature death and disability, thereby indicating the severity of the threat posed by death to life. This study proposes a multi-dimensional evaluation of childhood cancer survival status using MIR, DALYs, and mortality.
Based on the GBD2021, this study systematically presents the burden of childhood cancer and temporal trends, focusing on DALY, MIR, and mortality. It is firmly believed that comprehensive and updated evidence about childhood cancer survival demonstration in the study could contribute to the credible resource allocation planning and tailored public health measures for childhood cancer.
HIGHLIGHTS
In high Sociodemographic Index (SDI) region, the disease burden of the brain and central nervous system cancer has ranked first.

The largest survival disparity may occur in eye cancer, according to the mortality-to-incidence ratio between low and high SDI region.

In low SDI region, the overall absolute value of disability-adjusted life years (DALYs) increased, and a rise absolute value, period and cohort risks for nervous system cancer was observed.

Breast cancer and drug-related liver cancer in adolescents showed continuous rises in both DALY and death.

Methods

Data sources and study design
The data used in this study were retrieved from the GBD 2021, which quantifies health loss for 371 diseases or injuries and 88 risk factors with estimated incidence, death rate, point prevalence (hereafter referred to as prevalence), YLLs, years lived with disability (YLDs), and DALYs in 204 countries and territories[10,11].
The 25 childhood cancers included in this cross-sectional study consisted of an all level 3 grade of neoplasm in GBD (see Table 1). The childhood age group encompasses children and adolescents defined as ages 0–19 years, divided into <5 years old, 5–9 years old, 10–14 years old, and 15–19 years old at 5-year intervals. Neoplasm in GBD2021 is defined based on the ICD-10 classification system and divided into three levels. The higher the level, the more detailed the classification. This study analyzed all level 3 grade of neoplasm, excluding benign tumors or cancers without data. Compared with the GBD2017 database, several cancers occurring in children and adolescents are newly added to the GBD 2021 database, including Burkitt lymphoma, other non-Hodgkin lymphoma, eye cancer (including retinoblastoma and other eye cancers as distinct causes), soft tissue and other extraosseous sarcomas, malignant neoplasm of bone and articular cartilage, neuroblastoma, and other peripheral nervous cell tumors. Compared with the GBD2017 study, the present study further analyzed leukemia, neuroblastoma, and other peripheral nervous cell tumors; myelodysplastic, myeloproliferative, and other hematopoietic neoplasms; lip and oral cavity cancer; stomach cancer; pancreatic cancer; colon and rectum cancer; bladder cancer; cervical cancer; ovarian cancer; testicular cancer; nasopharyngeal cancer; tracheal, bronchus, and lung cancer; malignant neoplasm of bone and articular cartilage; soft tissue and other extraosseous sarcomas; breast cancer; thyroid cancer; eye cancer; and malignant skin melanoma.

We conducted a study mainly examining DALYs, MIR, and mortality, as well as the incidence, prevalence, YLLs, and YLDs. MIR is the ratio of death and incidence figures (either actual absolute case numbers or population rates) registered in the same period of time for a specific cancer. It related to the disease malignancy, approximately equivalent to 1–5 years survival probability and is a useful measure when considering the quality of cancer care and cancer reporting in a country[6]. The global DALYs, MIR, and death of childhood cancer in 2021 are described. Notably, subgroup analysis for childhood cancer revealed distribution characteristics. Furthermore, a trend analysis of childhood cancer was conducted to assess long-term trends from 1990 to 2021, as well as the effects of age-period-cohort (APC) analysis. Moreover, among adolescents, the burden of cancer attributable to risk factors was also evaluated. More details about the interpretation of the risk factor-cancer pairs can be found in the Appendix. Finally, a sensitivity analysis of childhood cancer grouped according to the ICCC-3, a commonly used classification system for childhood cancer based on ICD-O-3, was also provided.
The GBD study followed the Guidelines for Accurate and Transparent Health Assessment Reporting (GATHER) statement and was approved by the ethics committees of the University of Washington’s Institutional Review Board (Study 9060). This cross-sectional study has been reported in line with the STROCSS guidelines[12].

Statistical analysis
The Joinpoint model was chosen to estimate the long-term trends by average of the annual percent changes (AAPC), which present the values and confidence intervals of total and local changes using the grid search method to fit the log-linear regression function with unknown connection points[13]. AAPC provided a total change of these trends over a specific period as a weighted average of the annual percent changes (APCs). In addition, age, period, and cohort influence the time trend and mutually affect each other. Thus, we used the APC model to find the real relationships[14]. The APC model only analyzed cancers with two or more age groups. Using the 1971–1976 birth cohort and 1990–1991 period as references in the APC model. The sensitivity analysis of different reference point was conducted.
Autoregressive Integrated Moving Average model (ARIMA) was used to predict the burden of childhood cancer. As this study did not involve seasonal data, only the simple ARIMA (p, d, and q) model was utilized for the statistical analysis, where p, d, and q are the continuous and seasonal orders of autoregression, degree of difference, and order of moving average, respectively[15]. To avoid overfitting, the study employed a rolling prediction, with the initial training set from 1990 to 2010 and the initial validation set from 2011 to 2015. A new year is added on a rolling basis for each modeling, with the training set timeline ending in 2017 and the validation set ending in 2018–2021. A total of seven model training and validation sessions were conducted, and the optimal parameters of the seven models were selected based on AIC and prediction error. Then, the ARIMA model with optimal parameters was chosen to predict the DALY burden in 2022–2030 based on the data from 1990 to 2021, as shown in Table 4. Besides, the sensitivity analysis of ARIMIA was added according to its statistical condition, such as parameter tuning and breakpoint modeling, as shown in Table 5. The model selection method, forecast performance, and residuals of the ARIMA model were tested.

This study analyzed the age-standardized rates and numbers of all indicators for the 0–19 age group. For absolute numbers (e.g., number of deaths, number of DALYs, etc.), the study used the number of “<20 year” (which is equivalent to summing up the numbers for each age group). For age-standardized rate, such as the age-standardized death rate (ASDR) and the age-standardized DALYs (AS-DALYs), the rate of “<5 years,” “5–9 years,” “10–14 years,” and “15–19 years” were extracted, respectively, and age-standardized rate of 0–19 age group was calculated. Age-standardized rate was used in the AAPC, ARIMA, and APC models. As for the disease burden attributable to risk factors, data are available only for the “15–19 years” age group, and calculation was age specific. The direct age standardization method was used to calculate the age-standardized rate with reference to the WHO standard age composition[16], and the confidence intervals were also calculated by Frey’s method[17]. The Monte Carlo simulation method was used to calculate the confidence intervals of MIR and total percentage change, routinely assuming that original data followed a Poisson distribution. The normal distribution was as sensitivity analysis for MIR.
Generally, the age-standardized rates (uncertainty interval) are commonly used to compare different cancers, countries, and regions or to calculate the MIR, sex ratio, and AAPC. The global MIR for each cancer is expressed as the median and quartiles of the MIR for 204 countries, and the sex ratio is shown in the same way. MIR is calculated by dividing the age-standardized mortality rate per 100 000 population by the age-standardized incidence rate per 100 000 population for each country/region in a given year. Total percentage change and AAPC from 1990 to 2021were used to examine the long-term trend.
We used R (version 4.2.1) and Joinpoint Software (version 5.1.0) for statistical analysis and data visualization. APC model used R package “apc” to analyze. ARIMA uses the R packages “tseries” and “forecast.” The hypotheses were two-sided with a test level of 0.05. The Methods section in the Appendix provides details of the statistical analysis.

Result

Overview of global childhood cancer burden in 2021
Table 2 and Supplemental Digital Content Table S1, available at: http://links.lww.com/JS9/G703 present the overall burden of childhood cancers worldwide. In 2021, there were 114.50 thousand (95% UI 113.84–115.17) cancer-related death and 275.99 thousand (95% UI 274.97–277.03) new cases of childhood cancer occurred globally. A total of 9343.21 thousand (95% UI 8546.65–10 174.74) DALYs attributed to cancer for children and adolescents, including 9203.39 (98.50%) thousand (95% UI 8401.48–10 005.30) YLL. The top five contributors to childhood cancer were leukemia, brain and central nervous system cancer, other malignant neoplasms, non-Hodgkin lymphoma, and malignant neoplasm of bone and articular cartilage, which together accounted for roughly 76.84% of AS-DALYs and 72.94% percent of ASDR. In terms of MIR, the total childhood cancer was 0.41 (IQR 0.26–0.58), and the most serious cancers were pancreatic cancer, tracheal, bronchus, and lung cancer, liver cancer, stomach cancer, as well as leukemia with MIR values of 0.95 (IQR 0.91–0.97), 0.95 (IQR 0.87–0.98), 0.76 (IQR 0.57–0.79), 0.73 (IQR 0.58–0.83), and 0.72 (IQR 0.44–0.88), respectively.

Supplemental Digital Content Table S2, available at: http://links.lww.com/JS9/G703 shows the demographic subgroups for childhood cancer. The population aged <5 years had the highest AS-DALY and ASDR, and the MIR peaked in the 10–14 years age group at 0.49 (95% UI 0.31–0.62). Compared to females, male children face a more adverse cancer burden. Notably, there is a large gap in the cancer burden between different GBD world regions. The highest absolute burden rested in South Asia, and the highest age-standardized rate was in Eastern Sub-Saharan Africa. Region with the lowest absolute burden and age-standardized rate both was Australasia. Regarding childhood cancer MIR, Central Sub-Saharan Africa had the highest MIR, and the lowest MIR was in high-income Asia Pacific.

Subgroups of childhood cancer burden
Figure 1 and Supplemental Digital Content Figure S1, available at: http://links.lww.com/JS9/G703 show the childhood cancer burden grouped by SDI regions and age groups in 2021. Generally, among all childhood cancers, AS-DALYs and ASDR decreased with increasing SDI levels, but not for nervous system cancers with completely inverse trends. The AS-DALYs and ASDR in High SDI were 13.36 per 100 000 population and 0.17 per 100 000 population higher than those in the low SDI region, respectively. Notably, in the high SDI region, the disease spectrum changed such that the brain and central nervous system cancer surpassed leukemia in AS-DALY and ASDR, and it is estimated that the AS-DALYs of nervous system cancer would be greater than the hematologic malignancy (Fig. 2). The sensitivity analysis of the ARIMA model showed that the main analysis was quite robust, as detailed in the Appendix.

There was a significant difference of overall MIR between SDI regions (P < 0.05). The global median MIR and the ratio of MIR between the low SDI regions to high SDI regions are shown in Figure 3. Supplemental Digital Content Figure S2 shows the confidence intervals of MIR. The MIR for most cancers were below 10. Although the value of MIR outweighed 0.7 across all SDI regions, pancreatic cancer [ratio = 1.24 (95% CI 0.44–3.74)], tracheal, bronchus, and lung cancer [ratio = 1.39 (95% CI 0.81–2.39)] had a low difference. Cancers with large gaps between SDI regions had relatively low MIR values. The MIR of eye cancer had the greatest disparity among the SDI regions, which was estimated 31.83 (95% UI 20.38–53.14) times between the low and high SDI regions. In addition to testicular cancer, Hodgkin lymphoma and malignant skin melanoma had ratios of 16.74 (95% UI 10.49–27.81), 15.93 (95%UI 11.74–23.08), and 11.52 (95% UI 6.9–20.52), respectively. The confidence interval for MIR is as shown in the Supplemental Digital Content Figure S2, available at: http://links.lww.com/JS9/G703.

Figure 4 and its Supplemental Digital Content Figures S3–S5, available at: http://links.lww.com/JS9/G703 illustrate the global distribution of MIR for various cancers among children and adolescents in 2021. The global distribution across all childhood cancers and the top five malignancies by MIR revealed a similar geospatial disparity, with African nations exhibiting high MIR values, while low MIR values predominantly persisted in Western Europe, North America, and Oceania nations. The global range in childhood cancer MIR spanned 0.66, from the lowest observed value of 0.10 (Estonia) to the highest of 0.76 (Central African Republic). The highest MIR for pancreatic cancer, tracheal, bronchial, and lung cancers, liver cancer, leukemia, and stomach cancer were 0.99 (Somalia), 0.99 (Afghanistan), 0.93 (Marshall Islands), 0.99 (Central African Republic), and 0.95 (Afghanistan), while the lowest MIR appeared in France (MIR = 0.61), France (MIR = 0.54), Japan (MIR = 0.27), San Marino (MIR = 0.08), and Japan (MIR = 0.26), respectively.

Temporal trends and APC effects of childhood cancer burden
Table 3 and Supplemental Digital Content Table S3, available at: http://links.lww.com/JS9/G703 present the global percentage change and AAPC of the childhood cancer burden from 1990 to 2021. Over the past 30 years, the global childhood cancer burden was on a downward trend, with a decreased rate of 42.17% (95% CI −42.63 to −41.76) and 42.69% (95% UI −42.73 to −42.64) for death and DALY, respectively, as well as an AAPC of −1.77% (−1.80 to −1.74) and −1.80% (95% UI −1.83 to −1.77). Stomach cancer and leukemia had the largest decline, with a rate percentage change of −62.56% (95% UI −66.75 to −58.01) and −56.43% (95% UI −66.06 to −42.32) for AS-DALY, and stomach cancer exhibited the most rapid decline speed with an AAPC of −3.22% (95% UI −3.25 to −3.19). Conversely, breast cancer was the only cancer for which both the percentage change and the trend escalated; its number and rate percentage change were 86.90% (95% UI 42.23–141.86) and 60.15% (95% UI 21.87–107.24) for DALYs, with an AAPC of 1.43% (95% UI 1.39–1.47). As shown in Supplemental Digital Content Table S6, the DALY and death rate of breast cancer mainly keeps growing in low SDI, low-middle and middle SDI region. Especially in low-middle SDI region, the breast cancer possessed the largest baseline number of 22, 266 (95% UI 15, 913.06 to 29, 632.43) and the most rapid growth rate of 1.69 (95% UI 1.65–1.73) of DALY.

Supplemental Digital Content Figure S6 and Supplemental Digital Content S7 show the temporal trends of cancer accumulation and composition grouped by the SDI regions. In 1990, the high-middle SDI region had the highest childhood cancer AS-DALY and ASDR, but the low-SDI region has shouldered the greatest burden since the early 2000s. Over the past three decades, larger reductions in disease burden have been observed in the high-middle and middle SDI regions.
Figure 5 and Supplemental Digital Content Figure S8, available at: http://links.lww.com/JS9/G703 show the APC effects on childhood cancer DALY in global and SDI regions. The effects of age on the DALY rate indicated a decline with age across all SDI regions. Also, period and cohort effects on childhood cancer DALYs declined over time. However, in low SDI and low-middle SDI region, the period and cohort effects continued to elevate DALY rate for brain and central nervous system cancer, neuroblastoma and other peripheral nervous cell tumors, and kidney cancer. As shown in Supplemental Digital Content Table S4, the total percentage change in the overall childhood cancer DALY rate and number showed a declining trend across all SDI regions; however, the absolute number increased by 35.92% (95% UI 11.64–67.70) in low SDI region. Moreover, the absolute DALY for brain and central nervous system cancer and for neuroblastoma and other peripheral nervous cell tumors in low SDI region increased by 97.91% (95% UI 16.83–187.83) and 141.95% (95% UI 33.91–295.11), respectively.

The attributable burden of adolescent cancer
Supplemental Digital Content Table S5 presents the attributable DALYs for adolescent cancer in global and SDI regions. In 2021, the global attributable DALYs rate was 10.21 (95% UI 8.6412.21) per 100 000 population in adolescents. Unsafe sex was the primary cause of death. Overall, the attributable DALY rate showed a declining trend [AAPC: −0.90 (−0.96 to −0.85)] in global and all SDI regions, and the attributable DALY rate decreased with the increasing SDI region. However, the DALY attributable to drug use escalated across SDI regions and exhibited a sustained increasing trend worldwide [AAPC: 3.61 (95% CI 3.56–3.66)] and in all SDI regions. The outcome pairs with drug use in adolescent was liver cancer in GBD (as shown in Figure 6), which comprised two subtypes (liver cancer due to hepatitis B and liver cancer due to hepatitis C). In high SDI region, the attributable DALY rate of liver cancer had the both highest burden and AAPC (as shown in Supplemental Digital Content Figure S9), with the value of 0.63 (95% UI 0.46–0.83) per 100 000 and 4.81% (95% UI 4.67–4.93).

Sensitivity analysis of childhood cancer grouped according to the ICCC-3 was shown in Supplemental Digital Content Table S7.

Discussion
To the best of our knowledge, this study is the first to systematically analyze DALYs, MIR, and mortality in childhood cancer. These indicators can complement each other to provide a more complete picture of the healthcare systems’ capacity and different insights into medical resource allocation and guide resource allocation. We found that the overall burden of childhood cancer remained heavy, with changes in the disease spectrum observed in high SDI region. There are several patterns of regional variation in MIR for childhood cancer. While the AS-DALY and ASDR of childhood cancer declined, breast cancer showed an increasing trend. In low SDI region, the overall absolute value of DALYs increased, and a rise in absolute value, period, and cohort risks for nervous system cancer was observed. The DALY of liver cancer caused by drug use also increased, particularly in regions with a high SDI. This study highlights specific regions or cancer types that require enhanced attention, reflecting diverse challenges in the burden of childhood cancer, which warrants further research for validation.
Globally, childhood cancer contributed over 0.1 million deaths and 9.3 million DALYs, low and low-middle SDI regions were the primary sources of DALY for childhood cancer, which was consistent with a previous study1, highlighting the significance of reducing the burden of improving global childhood cancer survival rate and lifespan. Similarly, the top ranked five cancers, including leukemia, brain and central nervous system cancer, other malignant neoplasms, non-Hodgkin lymphoma, and malignant neoplasms of bone and articular cartilage, collectively accounted for 76.84% of AS-DALYs and 72.94% of ASDR.
However, the predominant disease burden of DALY and death in the high SDI region has shifted from leukemia to brain and central nervous system cancer, and the DALY of nervous system cancer will be greater than that of hematologic malignancies by 2024. One study based on GBD2019 reported that the spectrum of cancers is changing in children aged 0–9 years[18]. Brain and other nervous system tumors are the leading cause of cancer death among children and adolescents younger than 20 years in America[19]. The transition may be caused by following reasons. Over the past decades, there have been substantial improvements in leukemia diagnosis and management, especially with advances in laboratory techniques and new technologies, leading to marked reductions in global DALYs and mortality for leukemia[20,21]. Compared with leukemia, the lower incidence of brain and central nervous system cancer could make it difficult to recruit eligible subjects and sufficient samples to complete various clinical trials, limiting opportunities to refine treatment regimens and validate new therapies to some extent[22].
Furthermore, it is reasonable to believe that nervous system cancer was severely underdiagnosed in low SDI regions. First, brain and central nervous system cancers are asymptomatic diseases in which early symptoms are usually hidden or easy to miss, and thus, the majority of patients may be first diagnosed in the late stage of cancer or completely not be detected[23]. Second, the proportions of AS-DALYs, ASDR, and ASIR of nervous system cancer increased with increasing levels of SDI, a totally reverse trend compared to other childhood cancers. Therefore, the limited diagnostic capacity, misdiagnosis, and incomplete cancer registration systems further intensified the burden gap and unfairness of this cancer across SDI settings[24]. In the future, more comprehensive and integrated measures mainly for early detection, diagnosis, treatment, and prognosis of brain and central nervous system cancer are imperative to curb this disease burden trend.
The MIR reflects childhood cancer survival outcomes from a prognostic perspective. Top ranked five cancers of MIR, including pancreatic cancer, tracheal, bronchus, and lung cancer, liver cancer, stomach cancer, and leukemia, all have a median MIR among 204 countries greater than 0.7. In particular, the MIR of pancreatic, tracheal, bronchus, and lung cancer outweighed 0.5 across all countries, but the MIR was only 1.24 and 1.39, respectively. In recent study reported that the overall 5-year relative survival rate was 13% of pancreatic cancer and 25% of lung and bronchus cancer in US[19]. Due to nonspecific symptoms and limited diagnostics, pancreatic and lung cancers are often detected late and even with metastatic spread, resulting in poor survival[25,26]. Except for the top two cancers, we observed large global disparities in MIR (as shown in Figure 3), with a similar distribution in liver cancer, stomach cancer, and leukemia, where high MIR (MIR > 0.7) countries were primarily distributed in Africa, South America, and South Asia, and low MIR (MIR < 0.3) countries mostly occurred in Western Europe, North America, and Australasia. The MIR difference around the world reflects inequities for children in access to diagnosis and treatment, quality of care throughout the cancer care continuum, and outcomes, which are socially, politically, and economically produced and systematic[27]. The management of childhood cancer often requires surgical intervention for diagnosis and definitive treatment, but children living in low- and middle-income countries lack access to essential surgical care, such as surgical infrastructure and pediatric surgeon[28,29]. Surgical accessibility in these countries may be improved through strengthening the regional referral network, upgrading the diagnosis and treatment level of imaging, pathology, anesthesia, and pediatric intensive care, and strengthening the construction of surgical talent echelon or children’s oncology centers.
Several huge disparities in the MIR were observed across different SDI levels, particularly for eye cancer, testicular cancer, Hodgkin lymphoma, and malignant skin melanoma, with the ratio of the low SDI region to the high SDI region more than 10 times. In childhood eye cancer, retinoblastoma accounted for 97.20% of DALYs, and the MIR between low and high SDI regions was as high as 43.5 (95% UI 27.93–88.61). In the US, the 5-year survival rate of retinoblastoma is reported to reach 96%[14]. However, retinoblastoma has a poor survival rate (30% survival rate in low-income countries) in developing countries due to delayed diagnosis, which is determined by socioeconomic and cultural factors that also influence treatment adherence[30]. International Society of Paediatric Oncology—Paediatric Oncology in Developing Countries (SIOP-PODC) has generated guidelines for the clinical management of retinoblastoma in developing countries. They have developed a retinoblastoma classification system based on available resources, which enables medical institutions of different levels to conduct the best diagnosis and treatment within their capacity, provided the staging and treatment of retinoblastoma, and counseling of families for whom compliance is an issue[31]. The effectiveness and feasibility of the SIOP-PODC’s proposed management guidelines for retinoblastoma in lower-middle-income settings has been proven[32]. In settings with limited resources, we recommend increasing the investment in medical resources for cancers with a high potential for survival improvement, such as retinoblastoma, while simultaneously considering the actual cancer burden. On the other hand, governments should manage retinoblastoma in accordance with the SIOP-PODC guidelines and continuously improve the quality of medical services. International cooperation and assistance should be strengthened to provide technical and equipment support and to help cultivate medical talent. Similarly, for testicular cancer, Hodgkin lymphoma, and malignant skin melanoma, the SIOP-PODC retinoblastoma guidelines are drawn upon, which stratify care by resource, but guidelines adapted to resource-limited areas have yet to be explored and validated.
Over the past 30 years, there has been a global downward trend in AS-DALYs and ASDR for childhood cancer, with an overall average annual decline of −1.80% (95% UI: −1.83 to −1.77) and −1.77% (95% UI: −1.80 to −1.74), respectively. However, more attention should be paid to several exceptions and worsening situations. First, adolescent breast cancer is the only cancer with an increased disease burden. Generally, younger breast cancer patients are relatively special, with a faster rate of incidence growth but lower survival rates than older patients, requiring targeted treatment strategies to improve their survival and overall health[33,34]. Second, in low SDI regions, the absolute terms of DALY and death increased by 35.88% (95% UI 0.44–102.49) and 38.2% (95% UI 2.91–104.24), respectively, but their rates were both declined. Given that the rates were relative numbers and could be influenced by changes in the numerator and denominator, the contradictory trends between absolute terms and rates in the low SDI region seem to be dominated by the decreasing competitor risk of death from infectious diseases[35]. The APC effect predicted a future decline of childhood cancer DALY rate in all SDI regions, similar to previous finding[5]. The specific causes and associated mechanisms of the APC effect leading to an increased kidney cancer and nervous system cancer burden in low and low-middle SDI regions remain to be identified.
In general, almost all childhood cancers cannot be prevented[2]. In GBD 2021, the risk factors for cancers were only reported in adolescents and contributed to minor part (2.77%) of DALY rate. Despite a global decline in the total DALYs for liver cancer, an increasing trend of liver cancer (due to hepatitis B and hepatitis C) DALY rate attributable to drug use (injection drug use, IDU) was observed across all SDI regions. IDU through injection equipment facilitates the transmission of hepatitis C virus and hepatitis B virus in the bloodstream, further leading to hepatitis and even liver cancer. The increasing trend is a component of the broader rise in total burden related to hepatitis caused by IDU (as shown in GBD 2021), which may be linked to the low global coverage of major harm reduction and treatment interventions for people who inject drugs[36]. When implementing intervention measures such as needle and syringe exchange programs and opioid agonist treatment among adolescent IDUs in different countries, it is recommended to consider their specific risk environments[37]. Before IDU, adolescents are likely to smoke cigarette early, use inhalants and marijuana, leave school early, and engage in delinquent or criminal activities and inability to access addiction treatment before injection[38,39]. Moreover, the period from adolescence to early adulthood is critical for developing a healthy awareness and behavior. More measures are needed to help adolescents stay away from IDU, with policy, schools, and families potentially playing a key role in developing healthy habits[40].
This study had several limitations. First, the current GBD classification system lacks certain finer types of childhood cancer; for example, there is no evidence of low-grade gliomas and Wilms tumors, which are tracer cancers required by the WHO guideline of the CureAll Framework. Second, the potential underestimation bias due to registration and misclassification in low SDI regions should be improved in future studies. Third, although a high MIR suggests poor survival, MIR is a cross-sectional measure, and the exact quantitative relationship is not linear. Interpretation of the MIR is limited by its registration biases, diagnostic access, and the accuracy of death reporting. According to DALYs and deaths, while the MIR may be reliable for reflecting healthcare services in different regions, it should be carefully considered, and more evidence relating to survival rates is required. Fourth, it is impossible to eliminate the variations caused by measurement errors and inaccurate reporting of the original data, and the quality and collection of the original data also influences the estimation of the disease burden. Fifth, the ARIMA model compared disease burden trends for hematologic cancer and nervous system cancer, without estimating effects from policy, demographic, or healthcare changes. Furthermore, this study only examined the impact of specific risk factors on children aged 15–19.
In summary, the study reveals complex differences in childhood cancer burden that are intertwined over time across the three dimensions of cancer type, different indicators, and geography. We are calling for paying more attention to the first rank on DALY of brain and central nervous system cancer in high SDI region; setting the priority to narrow the burden gap on retinoblastoma, testicular cancer, Hodgkin lymphoma, and malignant cutaneous melanoma through conducting more economical and inclusive interventions worldwide. Also, breast cancer and drug use induced liver cancer should not be ignored.