Cohort Profile: The St. Jude Lifetime Cohort Study (SJLIFE) for paediatric cancer survivors.

Cohort Profile: The St. Jude Lifetime Cohort Study (SJLIFE) for paediatric cancer survivors Carrie R Howell , Kari L Bjornard, Kirsten K Ness, Nicole Alberts, Gregory T Armstrong, Nickhill Bhakta, Tara Brinkman, Eric Caron, Wassim Chemaitilly, Daniel M Green, Tim Folse, I-Chan Huang, John L. Jefferies, Sue Kaste, Kevin R Krull, Jennifer Q Lanctot, Daniel A Mulrooney, Geoffrey Neale, Kim E Nichols, Noah D Sabin, Kyla Shelton, Deo Kumar Srivastava, Zhaoming Wang, Carmen Wilson, Yutaka Yasui, Alia Zaidi, Jinghui Zhang, Leslie L Robison, Melissa M Hudson and Matthew J Ehrhardt * Department of Epidemiology and Cancer Control, St. Jude Children’s Research Hospital, Memphis, TN, USA, Department of Oncology, St. Jude Children’s Research Hospital, Memphis, TN, USA, Department of Psychology, St. Jude Children’s Research Hospital, Memphis, TN, USA, Department of Global Pediatric Medicine, St. Jude Children’s Research Hospital, Memphis, TN, USA, Department of Pediatric Medicine, St. Jude Children’s Research Hospital, Memphis, TN, USA, Division of Adult Cardiovascular Disease, The University of Tennessee Health Science Center, Memphis, TN, USA, Department of Diagnostic Imaging, St. Jude Children’s Research Hospital, Memphis, TN, USA, Hartwell Center, St. Jude Children’s Research Hospital, Memphis, TN, USA, Department of Biostatistics, St. Jude Children’s Research Hospital, Memphis, TN, USA and Department of Computational Biology, St. Jude Children’s Research Hospital, Memphis, TN, USA

However, in the early years of these studies, many were limited by dependence upon self-reported outcomes or registry data, smaller population sizes and restriction to single disease subsets and/or limited follow-up duration. To address these limitations, some have evolved to include periodic, prospective physical assessments extending beyond the period of cancer centre follow-up for single disease groups, 11 nested subsets within the larger cohort 14 , and more recently for all survivors, 12,17 in an effort to gain a better understanding of the true prevalence of disease and contributing risk factors. To this end, St. Jude Children's Research Hospital (SJCRH) established in 2007 the prospective St. Jude Lifetime (SJLIFE) cohort, open to survivors of all paediatric cancer subtypes representing the spectrum of paediatric, adolescent and young adult cancers, as well as frequencymatched community controls, including collection of comprehensive treatment data on all participants, provision of protocol-based medical assessments, assessment of patient-reported outcomes, validation of self-reported medical events, performance of periodic longitudinal evaluations and collection of biologic specimens. 18 Longitudinal, systematic medical assessments facilitate elucidation of the pathophysiology of cancer treatmentrelated morbidity, identification of biomarkers of subclinical organ dysfunction and characterization of highrisk survivor groups who may benefit from interventions to preserve health.
At its inception in 2007, SJLIFE was an institutionally funded resource designed to study adults previously treated for childhood cancer at SJCRH who had survived 10 years from cancer diagnosis. In 2015, SJLIFE received extramural support from the National Institutes of Health (CA195547) to expand enrollment to include individuals of any age treated at SJCRH who had survived 5 years from cancer diagnosis in order to facilitate: (i) detection of asymptomatic and premorbid conditions; (ii) discovery of early predictors of adverse outcomes; and (iii) to collectively provide the basis for impactful and targeted interventions. These changes significantly increased the number of eligible survivors and aligned eligibility requirements with existing observational cohorts of childhood cancer survivors, expanding the opportunity for novel research initiatives, including replication and validation of previous findings. The study was reviewed by the SJCRH Institutional Review Board and ethical approval was obtained on 25 April 2007. Details regarding study design, recruitment and participant characteristics of the original cohort have been previously published. 18 This study is registered at ClinicalTrials.gov under the identifier NCT00760656.
Who is in the cohort?
The SJLIFE study utilizes a retrospective cohort study design with prospective follow-up and ongoing accrual of patients diagnosed and treated at SJCRH over five decades . Additional information regarding the SJLIFE cohort is available at https://sjlife.stjude.org/. St. Jude provides financial support for transportation, medical evaluations, meals, domiciliary care and monetary compensation for days of study participation. The original inclusion criteria, as established in 2007, were limited to adults 18 years of age at the time of evaluation who were treated at SJCRH and had lived 10 years from initial cancer diagnosis. In 2015, the criteria were expanded to include survivors of any age at the time of evaluation who had lived 5 years post diagnosis and who had been diagnosed through 30 June 2012. Table 1 shows the distribution of characteristics of the pool of survivors eligible for recruitment into the cohort prior to expansion and after expansion.
To be eligible for participation in SJLIFE, individuals must have: (i) a history of childhood cancer diagnosed and have been treated at SJCRH between 1962 and 2012 and (ii) have survived 5 years from diagnosis. Whereas SJLIFE does not specify age at cancer diagnosis, SJCRH generally restricts acceptance to children <25 years of age at the time of cancer diagnosis. Recruitment utilizes a centralized process: (i) eligible survivors (through 30 June 2012) were identified via predetermined queries of medical record data in the SJCRH Clinical Research Informatics System; (ii) personalized invitation letters that provide mechanisms to respond electronically or via mail are sent to the survivor's last known address; (iii) if there is no response after 2 weeks, a study interviewer calls to determine the survivor's level of interest in participating; (iv) those interested in participating are transferred to a visit coordinator who describes the study and schedules the survivor for an on-campus visit (average duration of 3-4 days). If participants cannot be reached via voice call, electronic communications (i.e. email and text messaging) are attempted. To minimize loss to follow-up (both for original recruitment and for subsequent visits), the study team uses a combination of tracing procedures (e.g. patient and family contact information, postal forwarding addresses and internet queries) to locate participants, and verifies participant addresses at each point of contact (e.g. at the time of on-campus visit and follow-up phone calls).
To facilitate estimates of risk associated with cancer and its treatment, SJLIFE is recruiting age-, sex-and racefrequency matched individuals without a history of childhood cancer to complete the same clinical assessment as survivors. We refer to these individuals as 'community controls' since they are being recruited through a variety of approaches from the same general geographic area as the SJLIFE survivor population ( Figure 1). To achieve the desired geographic distribution, we offer the SJLIFE survivor participant the opportunity to identify a friend or non-first degree relative to accompany them to their on-campus visit and enrol as a community control. Additional community controls are recruited from the Memphis area through advertisements. As of 12 March 2020, 736 (target accrual of 1250) community controls have completed on-campus evaluations. The geographical distribution of eligible survivor and control participants is shown in Figure 1.
The following categories are used to classify participants: (i) 'eligible participants' are individuals meeting eligibility criteria and alive at the time of recruitment; (ii) 'enrolled participants' are those who have signed consent to participate; (iii) 'confirmed interest participants' have been contacted and are interested, but have not yet signed consent to participate; (iv) 'non-participants' have declined participation, failed to respond after initially indicating interest or were not able to be contacted. As of 12 March 2020, 73.3% (6005/8192) of all living, eligible survivors have enrolled and 80.1% have enrolled or confirmed their interest in participating. Among those approached for study participation, 80.3% (6005/7471) have been successfully enrolled, 87.8% have enrolled or have confirmed interest in participating and 94.9% of those enrolled have elected to complete an on-campus assessment ( Figure 2). Differences between participants and non-participants are detailed in Supplementary  How often have they been followed up?
The timing and sequence of participant data acquisition for events occurring prior to and after study enrollment are reflected in Figure 3. During and following treatment of paediatric 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 18 years of age, whichever occurs later. Data abstracted from medical records for all participants include demographics, cumulative chemotherapy doses, radiation fields and cumulative doses, information on surgical interventions, primary cancer recurrences and subsequent neoplasms, and acute and late organ-specific toxicity.
The health and vital status of potential participants are monitored by the St. Jude Cancer Registry and supplemented by National Death Index (NDI) searches. The NDI provides a centralized database of death record information ascertained by state vital statistics offices. The SJLIFE study team periodically submits a request to NDI for vital statistics on study participants known to have died or without known contact since the last query. The NDI then returns a list of probabilistic matches to the SJLIFE team using a standardized algorithm. 19 The results are then processed by the SJLIFE team via the Center for Disease Control and Prevention's National Program of Cancer Registries (NPCR) data linkage process. 20 The combined findings of both NDI and NPCR are then centrally reviewed and a final match status assigned. The Cancer Registry is an established SJCRH resource designed to maintain systematic, life-time annual follow-up for all   in the original and expanded cohorts) are invited to return to SJCRH at least once every 5 years for protocol-based medical evaluations and assessments of neurocognitive function, physical performance status and patient-reported outcomes. Permission for release of medical records for designated facilities up to 1 year from the campus visit is requested at each evaluation to validate interim, survivorreported medical events, in particular those relating to the most recent campus visit (e.g. follow-up biopsy and breast cancer diagnosis after a screening imaging abnormality identified during a SJLIFE visit). In addition, two studyteam members have the sole responsibility of following-up on health events identified during SJLIFE evaluations and those self-reported between campus visits, and their contact supplements the annual contact by the Cancer Registry. When new interval events are reported, additional permission to seek medical records is obtained on a case-by-case basis. Records are routinely requested for events involving imaging, tissue biopsies, surgical and diagnostic procedures, or interventions specifically impacting gradable conditions (e.g. skin biopsies, outside breast imaging) and/or clinical management (e.g. suspected cancer recurrence, chronic kidney disease, chronic hepatitis C). With the exception of vital status, interval events not coinciding with a campus visit (e.g. a self-reported myocardial infarction occurring 3 years after the last SJLIFE campus assessment) are censored from research analyses until the next subsequent campus visit occurs for uniform crosssectional validation of events.
As of 12 March 2020, 51.3% (2680/5223) of survivors who had completed a baseline clinical assessment have returned for one or more subsequent follow-up assessments. Supplementary Table 2, available as Supplementary data at IJE online, describes baseline characteristics as well as providing a comparison between those who have returned for a subsequent visit and those who declined additional visits. Table 2 describes grade 3-4 chronic conditions by organ system prevalent at baseline.

What has been measured?
Two levels of participation are offered in SJLIFE and have been applied to both the original and expanded cohorts: (i) an on-campus, comprehensive health evaluation (including health questionnaires and direct assessment) or (ii) comprehensive health questionnaires only, that are completed by mail or phone interview (Table 3). Detailed methodology on data collection, including medical record abstraction, questionnaires and medical assessments have been previously published for the original cohort. 18 On campus study participation involves travel to the SJCRH campus for a comprehensive assessment averaging 3-4 days, during which biologic specimens (e.g. blood, urine) are collected; metabolic, cognitive and neuromuscular functional status are systematically evaluated; and screening of organ function is performed (Supplementary Table 3, available as Supplementary data at IJE online). These uniform assessments (Table 3) include identification and grading of chronic conditions using modified Common Terminology Criteria for Adverse Events (CTCAE) criteria, 21 assessment of patient-reported outcomes, whole genome and exome sequencing, and linkage to the NDI for mortality followup. Relevant findings are summarized on a survivorship care plan and mailed to the survivor after the on-campus assessment, with direct phone contact by a study nurse or healthcare provider for critical findings that should be followed up by the participant's community providers (e.g. suspicious lesions found on breast mammography). Oncampus assessments are funded by institutional and extramural support. No study-related costs are submitted to healthcare insurance agencies.

What has it found? Key findings and publications
Since the inception of the SJLIFE study, 120 manuscripts have been published featuring results of late health outcomes. These publications have yielded a number of important findings based on analyses of on-campus participants, including clinically ascertained and validated prevalence 6 and cumulative burden estimates 4 for cancer treatment-related organ dysfunction, and documentation of novel late health outcomes. [43][44][45][46][47] Prior to SJLIFE, prospective clinical ascertainment of a large population of childhood cancer survivors to determine prevalence of chronic conditions had not been performed. Hudson et al. reported the prevalence of chronic conditions and the proportion associated with treatment exposures in the first 1713 adult SJLIFE survivors who had   completed a baseline assessment. Pulmonary, auditory, endocrine, reproductive, cardiac and neurocognitive adverse health outcomes were common, with the cumulative prevalence of chronic conditions by organ-specific outcomes by age 50. 6 Bhatka et al. 4 subsequently expanded upon these findings by applying the mean cumulative count method, 48 which estimates the mean number of recurrent or multiple health events occurring in a cohort over time in the presence of competing risk events, in order to describe the unique patterns and excess cumulative burden of chronic health conditions experienced by childhood cancer survivors compared with community controls. By age 50 years, survivors experienced, on average, 17.1 grade 1-5 and 4.7 grade 3-5 chronic health conditions compared with 9.2 and 2.3, respectively, among community controls.
Most recently, Wang et al. 49 utilized whole genome and exome sequencing to assess the contribution of germline genetic abnormalities to subsequent neoplasm risk in survivors assessed for mutations in cancer predisposition genes. Pathogenic/likely pathogenic mutations were identified in 5.8% of survivors and were associated with increased risk of breast cancer and sarcoma in irradiated survivors and for developing any subsequent neoplasm, breast cancer, non-melanoma skin cancer and two or more histologically distinct subsequent neoplasms among non-irradiated survivors.
What are the main strengths and weaknesses?
Through systematic, prospective medical assessments and longitudinal follow-up, the SJLIFE cohort provides the opportunity to accurately characterize contributions of treatment, behavioural, genetic and social factors to the health status of childhood cancer survivors. A major strength of SJLIFE is the ability to perform direct and longitudinal medical assessments on a large cohort of prospectively monitored childhood cancer survivors. Robust characterization of health outcomes and cancer-, cancer treatmentand health behaviour-related exposures provide an unparalleled opportunity to identify novel associations and to refine our understanding of the effects of cancer and its treatment on the health of aging survivors.
The SJLIFE study must be interpreted in the context of a number of limitations. Notably, SJLIFE is a singleinstitution study, therefore late effects and detection may be biased by institutional practices, whereas population demographics and ethnicity may not be generalizable to the remainder of the USA. In addition, because it was not established until 2007, many survivors treated in earlier decades were no longer living at study onset, introducing the potential for survival bias. Furthermore, while participation rates in survivorship research are typically high, it remains possible that survivors with the highest burden of late effects are more likely to decline participation. However, access to treatment records for eligible nonparticipants and regular linkage to NDI records wellpositions SJLIFE to better characterize these biases moving forward. Lastly, as the SJLIFE demographics are reflective of the surrounding population at large, the study is currently limited in its ability to perform robust analyses on minority populations, therefore application of the findings should be considered in the context of one's surrounding demographics.
Can I get hold of the data? Where can I find out more? SJCRH has established the St. Jude Cloud (http://www. stjude.cloud/) to provide secure sharing and collaborative analysis of large and complex datasets to facilitate research addressing paediatric cancer and other rare diseases. The St. Jude Cloud, developed as a partnership between SJCRH, DNAnexus and Microsoft, is a secure cloud-based data-sharing and collaboration environment. Data are aggregated at the patient level, providing researchers access to an extensive public repository of paediatric cancer genomics data, accelerated data mining, analysis and visualization capabilities. Data requests can be made on the website and require a standard data-use agreement. The St. Jude Cloud provides genomic sequencing data to the global research community, while making complex computational analysis pipelines available through a collection of bioinformatics tools designed to help both experts and nonspecialists interrogate genomic data. SJLIFE genomic data consisting of 5020 (4382 survivor and 638 community controls) whole genomes (30x) and whole exomes (100x) are currently available to the scientific community on the St. Jude Cloud.
Non-genomic SJLIFE data are posted to the Survivorship Portal on the St. Jude Cloud and include detailed information on: (i) cancer-related variables (diagnosis ICD-O-3 and ICCC-3), age at diagnosis, length of follow-up, and treatment including region and organspecific radiation dosimetry, surgical procedures and medications including cumulative doses for individual chemotherapeutic agents, plus selected antibiotics and immunesuppressants; (ii) sociodemographic data (sex, race, ethnicity, education, income, employment) and health behaviours (e.g. smoking/tobacco use, drug use, alcohol consumption, physical activity and diet); and (iii) outcomes (modified CTCAE classification and severity grading for 190 medical and 18 neuropsychologic conditions, laboratory-based values (e.g. CBC, renal, liver, kidney, endocrine function) and procedural assessments (e.g. echocardiography, pulmonary function, dual-energy X-ray absorptiometry, etc.), data collected through physical function and neurocognitive assessments, and survivorreported data obtained via the portfolio of SJLIFE questionnaires.

Supplementary data
Supplementary data are available at IJE online. • Study details and data are available at http://sjlife. stjude.org/ and http://www.stjude.cloud/