Rethinking Success in Pediatric Oncology: Beyond 5-Year Survival

Success in 5-Year Survival After Childhood and Adolescent Cancer

Remarkable advances in the treatment of childhood and adolescent cancer have resulted in substantial improvements in 5- and 10-year survival for the vast majority of cancer diagnoses.^1,2 Basic and clinical research is currently focused on designing novel therapies for the remaining 15% of patients who do not survive beyond 5 years from diagnosis and designing new treatment regimens to decrease the risk of acute and long-term toxicity.^3,4 Although 5-year survival has served as the primary benchmark of success for these efforts, patients with childhood and adolescent cancer represent a unique population with multiple decades of extended life accompanied by risk of late occurring sequelae of cancer treatment. The follow-up of 5-year survivors of childhood and adolescent cancer has documented substantially increased risk, cumulative burden, and severity of therapy-related chronic health conditions experienced by the majority of long-term survivors.⁵ Despite risk-stratified therapy, premature mortality risk from causes such as subsequent neoplasms, cardiac disease, pulmonary disease, and other chronic health conditions remains significantly elevated in long-term survivors.^6-8 Therefore, using 5-year overall survival as the primary benchmark of success does not adequately reflect the increased risk of morbidity and mortality experienced throughout the lifespan of childhood and adolescent survivors.

Reduction in Annual Excess Deaths as the Measure of Success

The National Cancer Institute's SEER publishes Cancer Statistics Review (CSR) monographs annually. Traditionally, the report includes a matrix of survival probabilities at multiple time points by cancer type. These data inform health policy decisions and critically frame the success of the nation's cancer response. To more accurately represent the lifelong burden of childhood and adolescent cancer and time trends of its associated long-term morbidity and mortality, measures beyond 5-year survival are needed. As part of the Childhood Cancer Data Initiative (CCDI), the House Appropriations Committee recently recognized the need for a comprehensive survival metric that considers the long-term survival of patients with childhood and adolescent cancer.

To begin a national dialogue on this topic, we propose SEER monitor and incorporate annual trends in the number of annual excess deaths among childhood and adolescent cancer survivors in the United States into the Annual Report to the Nation on the Status of Cancer and CSR. Excess deaths are calculated by subtracting the expected number of deaths in an age, sex, race, and calendar-year matched US population from the observed deaths among childhood and adolescent cancer survivors. Therefore, measuring excess death provides an estimate of deaths, directly (eg, progression) and indirectly (eg, treatment-related cardiovascular disease), attributable to cancer over the lifetime of individuals with childhood and adolescent cancer.⁹ If we are truly making progress in reducing the risk of morbidity and mortality experienced throughout the lifespan of patients with childhood and adolescent cancer, including the 5 years after cancer diagnosis, we ought to observe reduced total counts of excess deaths over time, given that the incidence of childhood and adolescent cancer has increased only slightly since 1975 (0.6% per year).¹⁰

To conceptualize output for the proposed metric, we reviewed data from SEER-9 registries, which covers 9.4% of the US population since 1975, for an estimated 445,647 US individuals under 20 years of age diagnosed with cancer between 1975 and 2016.¹¹ Our proposed report is presented in Figure 1. We describe three categories of excess deaths: (1) excess deaths within the first 5 years of diagnosis, as the majority of these deaths are likely related to progressive disease or acute toxicity or infection; (2) excess deaths between 5 and 10 years, which would reflect a mixture of recurrence or progressive disease and deaths associated with treatment exposures; and (3) excess deaths 10 or more years from diagnosis, which would be mostly related to treatment exposures rather than recurrence or progressive disease.⁹ In this patient cohort, an estimated 126,952 excess deaths occurred including 101,674 (80.1% of excess deaths) < 5 years, 10,999 (8.7%) between 5.0 and 9.9 years, and 14,279 (11.2%) ≥ 10 years after diagnosis (Fig 1A).

FIG 1.

(A) Estimated numbers and proportions of excess deaths among childhood and adolescent cancer cases diagnosed at 0-19 years of age in the United States between 1975 and 2016, according to years from diagnosis by diagnosis, sex, and race (≤ 5.0 years after diagnosis [majority of excess deaths would be related to progressive disease, infection, or acute toxicity]; 5.0-9.9 years [excess deaths would reflect a mixture of recurrence or progressive disease plus deaths associated with treatment exposures]; ≥ 10.0 years post diagnosis [majority of excess deaths would be related to treatment exposures, with few from recurrence or progressive disease]); (B) estimated numbers of excess deaths among all childhood and adolescent cancer diagnosed at 0-19 years of age in the United States between 1975 and 2016, according to calendar year of death and years from diagnosis; (C) estimated numbers of excess deaths; and (D) absolute excess deaths per 1,000 patients or survivors at risk among childhood and adolescent patients with ALL, CNS malignancy, and HL, diagnosed at 0-19 years of age in the United States between 1975 and 2016, according to calendar year of death and years from diagnosis. ALL, acute lymphoblastic leukemia; AML, acute myeloid leukemia; ES, Ewing sarcoma; GCT, germ cell tumor; HL, Hodgkin lymphoma; Liver, liver carcinomas; NBL, neuroblastoma; NHL, non-Hodgkin lymphoma; OS, osteosarcoma; RB, retinoblastoma; Renal, renal carcinomas; Rhabdo, rhabdomyosarcoma; STS, soft-tissue sarcoma.

Reflecting the remarkable improvement in 5-year survival from 63% in 1975¹⁰ to over 84% in 2010,² the annual excess deaths because of childhood and adolescent cancer occurring < 5 years after diagnosis have steadily declined from 1985 to 2016 (Fig 1B). However, the total annual number of excess deaths declined only modestly in the same period, after adding excess deaths occurring 5.0-9.9 years postdiagnosis and those occurring ≥ 10 years postdiagnosis. Excess deaths occurring ≥ 10 years after diagnosis increased over time, estimated at 29% of all excess deaths in 2016, nearly canceling the reduction of annual excess deaths in the first 5 years postdiagnosis.

Patterns of Annual Excess Deaths Differ Across Diagnoses

Considerable heterogeneity exists in the number of excess deaths across diagnoses, in part because of their incidence and 5-year survival rates but also because of differences in treatment regimens or intensities and risk for life-threatening late effects. Reflecting their incidence and 5-year survival patterns over the past four decades,¹⁰ CNS malignancies and acute lymphoblastic leukemia (ALL) had the highest numbers of total excess deaths, followed by acute myeloid leukemia (Fig 1A). The total number of excess deaths in ALL declined from 1985 to 2016, whereas the total number of excess deaths for CNS tumors increased (Fig 1C): the proportion of excess deaths within the first 5 years from diagnosis are similar, however (79% of ALL and 78% of CNS excess deaths). This juxtaposition is in part because of marked advances in the treatment of ALL and a growing number of CNS tumor survivors at an increased risk for severe and life-threatening late effects.^12,13 This becomes clear as we observe the absolute excess risk (the number of excess deaths per 1,000 patients or survivors at risk) in CNS survivors plateau over time compared with a continued decline in ALL survivors (Fig 1D).

Specifically, for CNS tumors, 5-year survival has increased dramatically over the past 5 decades because of advances in diagnostic imaging, neurosurgery, and radiation therapy. Yet, these improvements are accompanied by steep increases in excess deaths ≥ 10 years from diagnosis (33% of excess deaths in 2016, Fig 1C). This mirrors the trajectory where an increase in excess late mortality follows advances in 5-year survival achieved by modern treatment protocols. High excess late mortality is also seen in several other tumor groups including Hodgkin lymphoma (HL) where 47.6% of total excess deaths occurred ≥ 10 years postdiagnosis (1975-2016). HL has had high 5-year survival rates for decades (87% in 1970¹⁰), in part because of its sensitivity to radiation therapy, and annual excess deaths within 5 years from diagnosis have been declining (Fig 1C). However, the increasing long-term survivor population of HL encounters a wide range of late effects including cardiovascular diseases and subsequent neoplasms, leading to an appreciable increase in the number of excess deaths occurring ≥ 10 years postdiagnosis (87.7% of excess deaths in 2016, Fig 1C). This increase, driven largely by the increase of the survivor population of HL (Figs 1C and 1D), persists despite the implementation of risk- and response-based treatment approaches designed to reduce therapeutic exposures including lowering cumulative doses of radiation, anthracyclines, and bleomycin in an effort to decrease cardiovascular and pulmonary morbidity.¹⁴ Given the growing population of 5-year survivors, a pressing issue for CNS, HL, and other tumor groups with high late excess mortality is the prevention of premature morbidity and mortality.

Adoption of Risk-Adapted Therapy to Decrease Acute and Late Toxicity

As our understanding of the molecular biology of cancer has increased, so too has our ability to design risk-stratified treatment protocols to minimize exposures in standard risk patients. For ALL, risk adaptation of therapy improved disease-free survival¹² greatly, reflected clearly in the aforementioned steep decline of excess deaths occurring within 5 years of diagnosis (Fig 1C). Recent data suggest the risk of premature morbidity and mortality after survival has also been decreased,¹⁵ keeping the counts of annual excess deaths in 5.0-9.9 years and ≥ 10 years after ALL diagnosis from increasing, despite the growing population of ALL survivors (Figs 1C and 1D). Most notably, annual excess deaths 5.0-9.9 years postdiagnosis decreased from 2010 to 2016, which may reflect the long-term impact of contemporary protocols. The current ALL therapy may, therefore, represent a model for long-term mortality implications of successfully risk-stratifying therapy.

Measuring the Long-Term Impact of Novel Therapies

Modifying treatment to reduce adverse late-effects while maintaining equipoise with current 5-year survival remains a challenge for many tumor groups. Even as new immunotherapies, local control modalities such as proton beam radiotherapy, and targeted small molecule inhibitors are introduced as first-line therapies improving disease-free 5-year survival, it remains unclear whether novel therapies will extend survival past 5 years without late relapse and premature mortality because of disease progression. Notably, excess deaths occurring between 5.0 and 9.9 years have remained constant over time (Fig 1B). As clinical trials test immunotherapies and targeted therapies as front-line treatments, there is a need for long-term end points and clinical trial infrastructure to follow trial participants for long-term outcomes. In the absence of such an infrastructure, excess deaths will be able to inform about the long-term impact of such therapies. Recently, the National Cancer Institute proposed the development of an integrated or federated pediatric data infrastructure for clinical trials and observational cohort studies as part of their CCDI. Specifically, the National Childhood Cancer Registry, to be funded by the CCDI, has the potential to generate detailed population level disease and outcome data on all survivors to improve late outcome surveillance as long-term risk assessment.

Impact of Assessing Long-Term Excess Mortality Trends

Our review of long-term excess mortality trends among childhood and adolescent cancer survivors illustrates important trade-offs that occur with refinement and advancement of cancer therapy that produces increasing numbers of survivors at risk for late morbidity and premature mortality.^5-7 As the survivor population grows, so too does the number of excess deaths experienced by long-term survivors. Without effective implementation of therapeutic strategies that maintain efficacy while avoiding treatment-related late morbidity, accompanied by advancements in survivorship care, this trend is likely to continue. Raising the profile of long-term excess mortality trends in conjunction with traditional 5-year survival measures can inform national health policy decision making, international guideline development, investments in novel treatments, and future research priorities.

With respect to health policy implications, excess deaths because of childhood and adolescent cancer is a valuable metric to guide reimbursement policy for survivorship care to increase access to health and psychosocial services. Given the disproportionately high late excess deaths seen for several malignancies, there is a pressing need for a comprehensive regular assessment of survivors by providers with expert knowledge. However, there is no established reimbursement strategy for cancer survivorship visits. Instead, providers rely on a variety of billing codes that reimburse for managing comorbidities and health promotion (eg, smoking cessation) and not all services are reimbursable. Incorporation of excess deaths into the SEER CSR would provide advocates with data demonstrating the need and importance of survivorship care.

However, the diversity of survivors and treatment exposures in combination with the development of new treatments limit our ability to design a single comprehensive survivorship-care protocol that would satisfy the needs of all survivors of childhood and adolescent cancer. The recent Childhood Cancer Survivorship, Treatment, Access, and Research Act provides funding for a comprehensive framework for improving physical, psychosocial, and behavioral outcomes and improvement of healthcare delivery in survivors of childhood and adolescent cancer. Continued emphasis on long-term excess mortality is important to reinforce these funding priorities, build research programs designed to identify high-risk patients or survivors and preventable deaths, and design or test intervention strategies for prevention and early detection of chronic or late-onset morbidity.

In summary, research priorities and investments appropriately continue to focus on developing efficacious treatment strategies for children with incurable cancers, including integration of immunotherapy and molecularly targeted agents. However, success is more accurately evaluated by a measure that considers excess risk of death 5 or more years postdiagnosis of childhood and adolescent cancer. An expanded focus is needed on research to reduce acute, chronic, and late-onset toxicities, particularly morbidity that can directly or indirectly affect the risk of subsequent premature mortality. This will require implementing precision medicine approaches to identify high-risk patients and preventable deaths and designing or testing of interventions for prevention and early detection of chronic or late-onset morbidity associated with premature mortality.

DATA SHARING STATEMENT

The data underlying this article are available from the US National Cancer Institute's SEER 9-registries which can be accessed at: https://seer.cancer.gov/.

AUTHOR CONTRIBUTIONS

Administrative support: Leslie L. Robison

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Rethinking Success in Pediatric Oncology: Beyond 5-Year Survival

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).