Approach for Classification and Severity Grading of Long-term and Late-Onset Health Events among Childhood Cancer Survivors in the St. Jude Lifetime Cohort

This article is featured in Highlights of This Issue, p. 661

Investigators, having achieved remarkable progress in developing curative therapy for pediatric malignancies, now have a responsibility to evaluate cancer-related morbidity and its impact on long-term survivor health and quality of life (1, 2). Previous research has established that childhood cancer survivors commonly experience long-term (persistent) health problems following diagnosis and treatment and are at risk for late-onset health events occurring at rates exceeding those of sibling and population comparison groups (3–9). The morbidity associated with childhood cancer survival is multifactorial, with patient, treatment, and health care circumstances influencing outcomes (2). The reported prevalence estimates of specific complications vary by data collection methods (e.g., patient report, registry/administrative data, clinical assessment) as well as time (e.g., from diagnosis, attained age) of assessment. These disparities complicate comparison of research outcomes across studies and challenge the characterization of high-risk survivors who may benefit from alternate treatment strategies, heightened surveillance, and preventive or remedial interventions.

Essential to the characterization of high-risk morbidity profiles associated with cancer treatment is the use of a common rubric for classifying and grading adverse outcomes. The NCI Common Terminology Criteria for Adverse Events (CTCAE) provides a descriptive terminology that is widely used for grading severity of adverse events observed in clinical trials (10–12). However, despite significant revisions over time, the current CTCAE v4.03 (10) is still not fully representative of the spectrum of outcomes experienced by pediatric and aging adult survivors of childhood cancer. Moreover, CTCAE v4.03 does not routinely integrate detailed patient-reported and medical outcomes data available from clinically assessed cohorts, which may increase the likelihood of inconsistent assessments among research investigators in long-term follow-up settings. To address these deficiencies, we adopted a standardized severity grading of long-term and late-onset health events to utilize in the St. Jude Lifetime Cohort (SJLIFE) Study population. Specifically, we developed an approach that is applicable to childhood cancer survivors across the lifespan by modifying the existing CTCAE v4.03 criteria and aligning grading rubrics from other sources for conditions not included or optimally addressed in the CTCAE v4.03. The purpose of this article is to describe the methods of long-term and late-onset adverse event assessment used in the SJLIFE study, where data from multiple diagnostic modalities and patient-reported outcomes are ascertained.

The ongoing institutional review board (IRB)–approved SJLIFE study was initiated in late 2007, with the aim of facilitating longitudinal evaluation of health outcomes among individuals surviving pediatric cancer (13). Eligibility criteria for participation in SJLIFE initially included diagnosis of pediatric cancer treated or followed at St. Jude Children's Research Hospital (SJCRH, Memphis, TN), attained age of 18 years or older, and survival of 10 or more years from diagnosis. In 2015, eligibility criteria were expanded to include 5-year survivors of any age. The SJLIFE study design involves a retrospective cohort with prospective follow-up and ongoing accrual (Fig. 1). The retrospective component of SJLIFE utilizes (3–9) data from surviving cancer patients treated at SJCRH since its opening in 1962. During and following treatment of pediatric malignancy, cancer remission status and treatment-related toxicities are routinely monitored by the primary oncology team and/or the long-term follow-up (After Completion of Therapy) clinic until the survivor is 10 years from diagnosis and at least 18 years of age. Data obtained from medical record review of all participants include demographic details, the cumulative doses of specific chemotherapeutic agents, the fields and doses of radiation, information on surgical interventions, primary cancer recurrences and subsequent neoplasms, and acute and late organ-specific toxicity.

In addition to longitudinal evaluations undertaken as part of SJLIFE, all oncology patients transitioned from SJCRH long-term follow-up care to community providers are followed by the IRB-approved St. Jude Long-Term Follow-Up (SJLTFU) study. All SJCRH patients are invited to participate in the SJLTFU study at diagnosis. Health and vital status of SJLTFU participants are monitored by the St. Jude Cancer Registry and supplemented by periodic National Death Index searches.

Following provision of informed consent, participants in the SJLIFE cohort are invited to return to SJCRH at least once every 5 years for follow-up using protocol-based medical evaluations and assessments of patient-reported outcomes, neurocognitive function, and physical performance status. Permission for release of medical records is requested at each evaluation to validate interim, survivor-reported medical events. Data available through both retrospective health record review and prospective, standardized clinical assessment provide detailed information about symptoms, physical findings, laboratory/diagnostic study results, and clinical interventions to consider in the severity grading of chronic and late health events experienced by cohort members.

A large and diverse multidisciplinary team reviewed data regarding health events routinely collected as part of the SJLIFE and SJLTFU studies, focusing on persistent health conditions present from diagnosis or developing during or shortly after therapy (long term) and those developing 5 or more years after diagnosis (late onset); congenital conditions and acute cancer- and treatment-related toxicities that subsequently resolved were excluded. The compiled health events were then compared with those in CTCAE v4.03.

The grading criteria for each late effect featured in CTCAE v4.03 were reviewed by the multidisciplinary team. Minor modifications were made to the CTCAE grading schema for some conditions to integrate specific diagnostic findings, clinical management, surgical interventions, and patient-reported outcomes, with the goal of creating a more transparent and uniformly replicable grading rubric (Table 1). Clinical management was incorporated into the grading criteria to account for the treatment burden and intervention risks among survivors whose adherence to clinical management resulted in normal laboratory and diagnostic testing results.

In addition, pediatric-specific criteria (e.g., bone mineral density deficit; ref. 14) and more conservative diagnostic ranges were used to revise definitions of certain CTCAE v4.03 conditions (e.g., bradycardia and tachycardia) to avoid overdiagnosis based on assessments that fell marginally outside the standard reference ranges. Grading criteria for CTCAE v4.03 events originally designed to capture acute toxicities (e.g., seizures) were modified to facilitate chronic event grading that coincided with the traditional categories [mild (grade 1), moderate (grade 2), severe/disabling (grade 3), life-threatening (grade 4), or death (grade 5)].

Chronic and late health events perceived to be relevant to pediatric cancer survivors that were not included or optimally addressed in CTCAE v4.03 were also identified (e.g., liver fibrosis/cirrhosis; Table 2). Metrics for severity grading of newly identified events were derived from established standards (e.g., body mass index for overweight and underweight pediatric survivors) or developed by multidisciplinary team consensus using a rubric similar to that of the CTCAE. Detailed grading criteria for neuropsychologic outcomes were outlined by psychologists, incorporating patient-reported outcomes and the results of validated cognitive and psychologic measures and comprehensive psychosocial evaluations by study social workers (Supplementary Table S1). Proxy parent report was used when patient self-report was not appropriate (i.e., young age of participant, severe cognitive impairment). Novel (compared with CTCAE) grading procedures were outlined for the spectrum of benign and malignant subsequent neoplasms experienced by childhood cancer survivors and mapped using histology-based International Classification of Diseases for Oncology, Third Edition (ICD-O-3; ref. 15), in combination with lesion site and surgical International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM; ref.16) codes (Table 3). With the exception of amputation, surgical interventions were not graded as a chronic health condition; instead, the clinical or functional consequences of the procedure were graded (e.g., chronic kidney disease following nephrectomy).

Using organ system–based categories, 190 medical and 18 neuropsychologic conditions were selected for late effects grading (Fig. 2; Supplementary Tables S1 and S2). In all, categories were used as published in CTCAE v4.03 for 91 conditions/events (44%) and modified from those of CTCAE v4.03 categories in 94 (45%). Another 23 (11%) required development of new grading criteria for late effects not included in CTCAE v4.03 or for events with CTCAE v4.03 grading not suitable for pediatric or chronic (vs. acute) health conditions.

The majority of individuals treated for cancer during childhood, adolescence, and young adulthood will experience extended survival after reaching the 5-year milestone from diagnosis (1, 2). An accurate characterization of cancer-related morbidity is essential to optimize therapeutic approaches for newly diagnosed patients and guide health surveillance recommendations for long-term survivors. The ability to compare outcomes from multiple cohorts requires the use of a common language for the assessment of adverse health events. Historically, CTCAE has provided comprehensive guidelines that enable consistent evaluations of treatment-related toxicity, but its application to cancer survivor cohorts has been limited by a primary focus on acute toxicities and lack of consideration of pediatric-specific reference ranges and developmental health risks (17).

Challenged with defining the long-term impact of cancer and its treatment in a large cohort of clinically assessed cancer survivors who developed health events across an age spectrum, we modified the CTCAE v4.03 to facilitate consistent and transparent late effects assessment by research team members. Age-appropriate reference ranges were incorporated in the grading criteria for a variety of conditions. Rather than relying on the organ system–specific “other” category for many events, clinically relevant data were added in an effort to augment the grading criteria. Our approach to grading the severity of subsequent neoplasms illustrates how histologic subtype and clinical management were integrated into the assessment of the generic category of “neoplasms, benign, malignant, and unspecified” (Table 3). Inclusion of details about conditions represented within a generic category, diagnostic parameters, and surgical and medical management in grading criteria was perceived by research staff as particularly helpful in improving accuracy and uniformity of assessments. In this regard, we noted that several categories in the proposed CTCAE v5.0 include similar specifications.

As highlighted by previous investigators, guidelines for evaluating adverse events impacting physical and intellectual growth and development in pediatric cancer survivors are not adequately represented in CTCAE v4.03 (17). This deficiency is particularly problematic in the long-term follow-up setting given the high prevalence of endocrine and cognitive late effects associated with specific pediatric cancer therapies (18–25). Children also experience emotional and psychosocial challenges that are unique from those of adults, necessitating addition of novel categories of pediatric-focused neuropsychiatric outcomes (20, 23, 26). Incorporating developmentally sensitive patient-reported outcomes into the grading criteria for many outcomes, especially neuropsychologic outcomes (Supplementary Table S1), enhanced our ability to assure that toxicity assessment considered the patient's perspective and chronic symptoms, which has been reported to be lacking in clinician-based assessments (27, 28).

Our efforts to standardize late effects toxicity assessments for the SJLIFE study should be considered in the context of several limitations. We focused on the assessment of late health outcomes and recognize a more thorough consideration of acute toxicity grading criteria in children is also needed. Although comprehensive in our attempts to be inclusive of the wide range of cancer- and treatment-related late effects, it is possible that we have overlooked other adverse events experienced by childhood cancer survivors. Finally, the modifications and additions to the CTCAE v4.03 reflect the opinions of investigators from a single institution. Broader, multi-institutional collaboration will be required to achieve the goal of a common language for the assessment of late effects of pediatric cancer and its treatment across an age spectrum.

Standardized measures for assessing the severity of long-term and late-occurring health conditions in childhood cancer survivors are needed. We believe that the approach adopted for the SJLIFE cohort augments the existing CTCAE rubric to allow uniform assessment and grading of toxicities across a wide spectrum of clinical and research environments. This mechanism provides a platform upon which to further develop and harmonize a system that facilitates future collaborative investigations.

No potential conflicts of interest were disclosed.

Conception and design: M.M. Hudson, M.J. Ehrhardt, M. Baassiri, D.K. Srivastava, M. Wilson, K.K. Ness, L.L. Robison

Development of methodology: M.M. Hudson, M.J. Ehrhardt, N. Bhakta, M. Baassiri, H. Eissa, W. Chemaitilly, D.M. Green, G.T. Armstrong, T.M. Brinkman, J.L. Klosky, K.R. Krull, R.B. Khan, D.K. Srivastava, V.M. Joshi, D. Stokes, M.E. Hoehn, M. Wilson, K.K. Ness, L.L. Robison

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M.M. Hudson, M.J. Ehrhardt, N. Bhakta, M. Baassiri, H. Eissa, D.M. Green, D.A. Mulrooney, J.L. Klosky, K.R. Krull, N.D. Sabin, J.K. Bass, S. Kaste, V.M. Joshi, M.E. Hoehn, M. Wilson

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M.M. Hudson, M.J. Ehrhardt, N. Bhakta, W. Chemaitilly, D.A. Mulrooney, J.L. Klosky, K.R. Krull, D.K. Srivastava, Y. Yasui, D. Stokes, M. Wilson

Writing, review, and/or revision of the manuscript: M.M. Hudson, M.J. Ehrhardt, N. Bhakta, M. Baassiri, H. Eissa, W. Chemaitilly, D.M. Green, D.A. Mulrooney, G.T. Armstrong, T.M. Brinkman, J.L. Klosky, K.R. Krull, N.D. Sabin, C.L. Wilson, I.-C. Huang, J.K. Bass, S. Kaste, R.B. Khan, D.K. Srivastava, Y. Yasui, S. Srinivasan, D. Stokes, M.E. Hoehn, M. Wilson, K.K. Ness, L.L. Robison

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M.M. Hudson, H. Eissa, D.A. Mulrooney, K. Hale, M.E. Hoehn, K.K. Ness, L.L. Robison

Study supervision: M.M. Hudson

Other (data analysis): S. Kaste

Other [reviewed data discrepancies to go back to primary source (echocardiograms)]: V.M. Joshi

The authors thank Kathy Laub for technical support with manuscript preparation.

This work was supported by St. Jude Children's Research Hospital Cancer Center support grant number 5P30CA021765-33 (to C. Roberts), the St. Jude Lifetime Cohort Study grant number U01 CA195547 (to M.M. Hudson and L.L. Robison), and American Lebanese Syrian Associated Charities (to all authors).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.