Functional Outcomes and Social Attainment in Asian/Pacific Islander Childhood Cancer Survivors in the United States: A Report from the Childhood Cancer Survivor Study

Contemporary treatment strategies have contributed to a decline in mortality among survivors of childhood cancer (1). However, survivorship comes at a cost of being at elevated risk for a variety of chronic health conditions associated with cancer diagnosis and/or treatment, compromised health status that can last a lifetime. Long-term survivors of childhood cancer are also at risk of developing neurocognitive deficits and psychologic distress (2–7). Reports from the Childhood Cancer Survivor Study (CCSS) have stated that up to a third of survivors report problems with task efficiency, memory, organization, or emotional regulation (6, 8–10). Survivors are also more likely than their siblings to report symptoms of anxiety, depression, somatization, and suicidal ideation (2, 5, 10–12). Consequently, neurocognitive deficits and psychosocial distress can have negative effects on survivors' health-related qualify of life (HRQoL), social attainment, and functional independence (8, 13–15). The risk factors for poor functional outcomes in survivors of childhood cancer are well documented in the literature, and include central nervous system (CNS) cancers and CNS-directed therapies, such as cranial radiation (CRT), intrathecal chemotherapy, and high-dose systemic methotrexate (16, 17). As survivors age, frailty and chronic health conditions such as cardiopulmonary and endocrine complications can also lead to worse functional outcomes (7, 12, 17–20).

In addition to clinical and treatment predictors, differences by race and ethnicity in these outcomes have become the focus of emerging research. One report by the CCSS showed that compared with non-Hispanic White (NHW) survivors, non-Hispanic Black (NHB), and Hispanic survivors are three times more likely to report moderate-to-severe endocrine complications even after adjusting for socioeconomic status and obesity (21). Another recent CCSS report did not identify differences in neurocognitive function between Hispanic, NHB, and NHW survivors, but the minority survivors reported poorer HRQoL outcomes than NHW survivors, and this disparity became more evident as the survivors aged (22). These studies suggest that the burden of morbidity borne by childhood cancer survivors differs by race and ethnicity. However, it is important to highlight that neither of these previous reports examine functional outcomes specifically in survivors of Asian or Pacific Islander (API) descent, members of the most populous group in the world.

The API population in the United States is a unique and heterogeneous group, due to broad historical immigration patterns. Differences in cancer incidence and survival patterns have been reported between API and other racial/ethnic groups (23, 24). Unfortunately, given the relatively small population of API individuals in the United States, studies describing the health and functional outcomes in childhood cancer survivors in this population are limited. One recently published study revealed that compared with NHW children and adolescents, non-Hispanic API patients exhibited a higher risk of death from high- and low-amenability cancers (25). Genetic variations may account for the observed variations in treatment outcomes and risk profiles for exposure-related late effects (26). External factors such as the socioeconomic status, cultural values, and family functioning may also contribute to differences in mental health and HRQoL outcomes between patients with cancer of NHW and API descents (23, 26). These observations highlight the need to comprehensively assess the psychosocial and functional outcomes of API survivors of childhood cancer in the United States.

The objectives of this study were to (i) compare neurocognitive function, emotional distress, HRQoL, and social attainment outcomes between API survivors of childhood cancer and sibling controls, (ii) compare differences in outcomes between API and NHW survivors, and (iii) identify the factors associated with these outcomes in API survivors.

The CCSS is a multi-institutional retrospectively constructed cohort with prospective follow-up of 5-year survivors of childhood cancer. It includes survivors diagnosed with cancer between 1970 and 1999 at an age younger than 21 years (27, 28). Multiple cancer diagnoses were included. Institutional Review Boards at the 31 participating institutions approved the CCSS study protocol, and all participants provided written informed consent in accordance with the Declaration of Helsinki. At cohort entry, survivors identified a living sibling nearest to them in age. A random sample of siblings was contacted to participate, and they served as the comparison population for functional outcomes of this study (27, 28). With the exception of cancer-specific topics, information collected from the sibling cohort is identical to that obtained on the survivor population.

All participants completed a baseline questionnaire assessing their demographic factors. Information on socioeconomic factors and health-related outcomes were obtained during follow-up assessment in 2003 or 2015. Race/ethnicity information was obtained using self-reported race categories of White, Black, American Indian or Alaska Native (AIAN), API, or Other (with the option to write in their race). Hispanic ethnicity was reported through a separate binary (yes/no) question. For this study, 14,498 eligible survivors and 3,282 siblings completed both the baseline and follow-up questionnaires on functional outcomes in 2003, 2007, or 2015. We excluded participants in the AIAN (n = 82), Black (n = 654), Hispanic (n = 587), and Other racial categories (n = 786), leaving two mutually exclusive race/ethnicity populations, NHW and API (Supplementary Fig. S1).

Neurocognitive outcomes were measured using the CCSS-Neurocognitive Questionnaire (CCSS-NCQ), a 25-item instrument that has previously been validated in CCSS survivors and siblings (29). It consists of four domains: task efficiency, emotional regulation, organization, and memory. Raw scores were converted to T-scores with reference to sibling norms (mean = 50, SD = 10 in siblings), with higher scores indicating more neurocognitive problems. Consistent with other CCSS publications, neurocognitive impairment in this study was defined as a T-score ≥90th percentile of the sibling T-score distribution for each domain.

Emotional distress was measured using the Brief-Symptom Inventory-18 (BSI-18), which includes subscales for anxiety, depression, and somatization (30). Sex-specific T-scores were calculated on the basis of standardized normative values (30). A higher score is indicative of a higher level of emotional distress. Participants with T-scores ≥90th percentile were classified as presenting with clinically significant levels of emotional distress.

The HRQoL was measured using the 36-Item Short Form Health Survey (SF-36) from the Medical Outcomes Survey, which includes eight domains: general health, physical role, physical function, bodily pain, vitality, mental health, social function, and emotional role (31). Scores for each domain were converted into T-scores based on age- and sex-specific norms (32), with a higher score indicating better HRQoL. Impairment was defined as a T-score falling at least one SD below the normative mean of 50.

Data on social attainment were examined for those participants who were older than 25 years at the most recent available follow-up timepoint. These data included highest educational attainment (college graduate and above vs. below), employment status (categorized as full-time employment vs. others), annual household income (≤20,000 vs. >20,000 U.S. dollars), current marital status (ever-married vs. never-married), and independent living status.

The predictor of interest in this study is race (API vs. NHW). The other covariates included sex, race, age at completion of survey, age at diagnosis and original cancer diagnosis. Treatment variables included intrathecal methotrexate (yes vs. no), intravenous methotrexate (yes vs. no) and CRT dose. All treatment exposures (radiation and chemotherapy) were abstracted from the medical records of treating institutions. CRT maximum target dose was determined on the basis of a detailed review of radiotherapy records and taken as the sum of the prescribed dose from all overlapping brain fields.

Chronic health conditions were graded for severity according to the NCI's Common Terminology Criteria for Adverse Events (CTCAE), version 4.03 (33), using previously reported methods (6, 7, 18, 20). Survivors reported their organ-specific chronic health conditions and age at onset at baseline and at subsequent follow-up evaluations. For this study, the health conditions of interest were limited to cardiovascular, pulmonary, endocrine, and neurologic systems, in accordance with evidence from literature and previous CCSS reports supporting an association of these chronic health conditions with functional outcomes in survivors (6, 7, 18, 20).

Descriptive analyses were performed to summarize the distributions of the outcome variables and covariates according to groupings consistent with previous CCSS publications. Univariate analyses (χ² test for categorical variables and t test for continuous variables) were conducted to compare baseline characteristics between API survivors versus NHW survivors.

Proportions of survivors and siblings with functional impairment within the API and NHW are presented for descriptive purposes. All subsequent comparison of the neurocognitive, emotional, and HRQoL outcomes was conducted in the form of continuous T-scores. To ascertain the effect of cancer and treatment on functional outcomes in the API group, neurocognitive function (CCSS-NCQ), emotional distress (BSI-18), and HRQoL (SF-36) scores were analyzed within the API survivors using multiple linear regression, adjusted for sex, age at diagnosis, and age at evaluation as a basic model. Then cancer diagnosis groups and treatment variables (intravenous methotrexate, intrathecal methotrexate, and CRT dose) were added to the basic model in two separate steps, to examine the effects of cancer diagnosis and treatment, respectively. These two sets of cancer-related variables are highly correlated and could not be added together in a single model. The assumptions of linear models were checked and no violation was detected (34). Comparison between API survivors and siblings were done in a single model adjusted for sex and age and evaluation. Modifications of the linear models by generalized estimating equations (GEE) were used to account for possible within-family correlation between survivors and siblings from the same family. Estimates of regression coefficients and 95% confidence intervals (CI) were reported. The social attainment outcomes (employment, marital status, education attainment, household income, and independent living) were compared between the survivors and siblings of API using logistic regression with the GEE modifications, after adjusting for sex and the age at evaluation. Estimates of ORs and 95% CIs are reported. Comparison of functional outcomes was not conducted among NHW survivors versus NHW siblings as this data was presented in a previous publication (22).

To evaluate the effect of race on functional outcomes, multiple linear regression was used to compare the T-score outcomes between the survivors of API and NHW, adjusting for sex, age at evaluation, and treatment variables (intravenous methotrexate, intrathecal methotrexate, and CRT dose).

Finally, the same regression methods were used to identify clinical and treatment factors associated with the T-score outcomes among the API survivors, after adjusting for sex and the age at evaluation. Clinical variables included cancer diagnosis groups and age at diagnosis, and treatment variables included the intravenous and intrathecal methotrexate and CRT doses. CRT dose was categorized as five levels (none, <18 Gy, 18 to 23.9 Gy, 24 to 49.9 Gy, and ≥50 Gy) based on its distribution. These variables were identified a priori because previous studies have suggested that the factors associated with poor functional outcomes in childhood cancer survivors include female sex, CNS tumor diagnosis, younger age at diagnosis, intravenous and intrathecal methotrexate and higher CRT doses (7, 15–20). All analyses were conducted using SAS 9.4 software (SAS Institute). A P value of < 0.05 was considered statistically significant, and all tests were two sided.

Raw data for this study were generated at the data repository of the Childhood Cancer Survivor Study (https://ccss.stjude.org/). Derived data supporting the findings of this study are available from the corresponding author upon request.

A total of 203 API survivors and 31 API siblings and 12,186 NHW survivors and 2,908 NHW siblings completed the baseline and follow-up questionnaires (Table 1). The mean age at diagnosis was 7.0 (SD = 5.4) years for API survivors and 7.9 (SD = 5.8) years for NHW survivors (P = 0.036). The mean age at assessment for API survivors [29.2 (SD = 6.3) years] was slightly younger than that of NHW survivors [31.5 (SD = 7.3) years] (P < 0.001).

The proportion of leukemia survivors was higher in the API group (46.3%) than the NHW group (37.0%; P = 0.009). The proportions of survivors diagnosed with CNS tumor were similar, 14.6% and 15.5% in API and NHW survivors, respectively. A higher proportion of API survivors received intravenous methotrexate (36.5% vs. 25.8%; P < 0.001) and intrathecal methotrexate (46.8% vs. 38.4%; P = 0.008), as compared with NHW survivors. API survivors were also treated with marginally higher mean doses of intravenous methotrexate [42.4 (SD = 70.2) g/m²], as compared with NHW survivors [27.3 (SD = 63.9) g/m²] (P = 0.09). API survivors also received higher doses of intrathecal methotrexate than NHW survivors [216.1 (SD = 134.6) mg/m² vs. 166.6 (SD = 145.2) mg/m²; P = 0.006]. API and NHW survivors received similar doses of CRT (P = 0.89).

The most commonly reported moderate-to-severe chronic health condition was endocrine (API: 15.9% and NHW: 20.7%; P = 0.07) complications. There was significantly higher proportion of NHW survivors who reported moderate-to-severe cardiovascular conditions, as compared with API survivors (18.1% vs. 10.2%; P < 0.001). None of the API and NHW siblings reported any moderate-to-severe chronic health conditions. As the proportion of API survivors and siblings who reported chronic health conditions was small, we did not include chronic health conditions as a predictor in subsequent analyses.

After adjusting for sex and the age at evaluation, no difference was observed in neurocognitive outcomes between API survivors and siblings (Supplementary Table S1), as well as between API and NHW survivors (Table 2). Descriptively, more than a quarter of the survivors reported impairments in task efficiency (API: 30.6%, NHW: 24.7%) and memory (API: 32.1%, NHW: 25.3%; Table 2).

Within the API group, survivors of CNS tumors reported more problems than did survivors of non-CNS solid tumors with task efficiency [difference estimate (Est) = 10.23; 95% CI = 3.23–17.24; P = 0.004] and memory (Est = 11.64; 95% CI = 5.40–17.88; P < 0.001), after adjusting for sex and the age at evaluation (Table 3). Higher CRT doses were associated with poorer task efficiency and memory, and intravenous methotrexate was associated with poorer task efficiency (Table 3). Female survivors reported more emotional dysregulation than male survivors (Est = 4.40; 95% CI = 0.27–8.52; P = 0.037). Survivors older than 45 years reported more emotional dysregulation (Est = 12.10; 95% CI = 4.94–19.27; P = 0.001) and memory problems (Est = 9.91; 95% CI = 2.01–17.81; P = 0.014) than younger survivors.

No difference was observed for emotional distress between API survivors and siblings (Supplementary Table S1). The proportion of API survivors and siblings reporting depressive symptoms was 10.9% and 3.3%, respectively. We also did not identify differences in the level of emotional distress between API and NHW survivors.

Among API survivors, female survivors reported more somatization problems than male survivors (Est = 2.93; 95% CI = 0.26–5.61; P = 0.031), after adjusting for sex and the age at evaluation (Supplementary Table S2).

API survivors reported worse physical function than their sibling controls [52.13 (SD = 8.32) vs. 55.06 (SD = 2.10); P = 0.016] (Supplementary Table S1). Interestingly, siblings [51.99 (SD = 9.45)] reported more pain than survivors [54.11 (SD = 8.98)] (P = 0.034). No difference was observed for other HRQoL outcomes between the API survivors and siblings.

API survivors reported, on average, less bodily pain than NHW survivors [54.11 (SD = 8.98) vs. 51.32 (SD = 10.12); P < 0.001] (Table 2). Up to a fifth of the survivors reported impaired general health (API: 21.8%, NHW: 22.6%), poor vitality (API: 17.1%, NHW: 22.1%), and pain (API: 10.2%, NHW: 16.1%).

An older age (>45 years) was associated with a worse HRQoL in terms of performance function, physical role, bodily pain, vitality, social functioning, and mental health in API survivors (all, P < 0.05; Table 4). Intravenous methotrexate was associated with poor physical functioning (Est = −2.84; 95% CI = −4.95 to −0.74 = 1.07; P = 0.008), vitality (Est = −3.36; 95% CI = −6.66 to −0.05; P = 0.046) and role emotional functioning (Est = −5.46; 95% CI = −9.79 to −1.13; P = 0.014; Table 4). Interestingly, survivors treated with intrathecal methotrexate had better physical functioning (Est = 4.12; 95% CI = 2.01–6.22; P = 0.001), vitality (Est = 4.40; 95% CI = 0.85–7.96; P = 0.015), and less bodily pain (Est = 4.06; 95% CI = 1.09–7.03; P = 0.007) than survivors who were not treated with intrathecal methotrexate.

No difference was observed for social attainment between API survivors and siblings (Supplementary Table S1). The proportion of survivors (63.5%) who attained at least a college degree was lower than the sibling group with marginal statistical significance (77.5%; P = 0.06).

NHW survivors were less likely than API survivors to attain at least a college degree (OR = 0.50; 95% CI = 0.34–0.73; P < 0.001; Fig. 1). API survivors (58.5%) were more likely than NHW survivors (35.8%) to be never-married (OR = 2.83; 95% CI = 1.93–4.13; P < 0.001). API survivors (39.2%) were three times more likely than NHW survivors (17.9%) to live dependently (OR = 3.10; 95% CI = 2.02–4.74; P < 0.001).

API survivors who were diagnosed at a younger age (0–4 years) were more likely to be never-married and less likely to live independently than older survivors (all, P < 0.05; Table 5). Compared with survivors of non-CNS solid tumors, survivors of CNS tumors were at elevated risk of not completing a college education (OR = 2.69; 95% CI = 1.12–6.42; P = 0.026) and living dependently (OR = 3.01; 95% CI = 1.09–8.28; P = 0.033).

Our study identified no overall statistically significant differences in neurocognitive problems, emotional distress, and HRQoL outcomes between API and NHW survivors. The clinical and treatment factors associated with functional outcomes in this study were largely similar to those that are well established in the literature, suggesting that these functional outcomes were primarily driven by the cancer and treatment exposures. After adjusting for age, sex, and treatment, API survivors were two times more likely than NHW survivors to complete a college education, but were less likely to be married or live independently. Consistent with observations across the general population in the United States, the differences in education achievement and social attainment between API and NHW survivors may likely be influenced by latent environmental and sociocultural factors.

We did not observe statistically significant differences in the neurocognitive functioning scores between API and NHW survivors. However, it is worth mentioning that close to a third of the API survivors reported impairments in task efficiency and memory, approximately 20% higher than the proportion of impairment in NHW survivors. Although the current analysis lacked power to detect statistical differences, such a substantial effect size found between API and NHW survivors suggests a potential clinically relevant disparity in neurocognitive outcomes and hence warrant a more comprehensive investigation using a combination of self-report and objective testing. We also identified an older age, CNS tumor diagnosis, and higher doses of CRT as associated with worse neurocognitive function and HRQoL in API survivors. These findings are consistent with those of worse health and functional outcomes in previous CCSS reports and in the literature (16, 17), indicating that clinical and treatment exposures are strong determinants of long-term functional outcomes. These subgroups of high-risk survivors, regardless of race and ethnicity, warrant closer monitoring and targeted interventions to improve their functional outcomes.

API survivors had higher levels of education attainment and were two times more likely than NHW survivors to complete a college education. This observation is consistent with studies in the general population, which have shown that regardless of cognitive abilities or sociodemographics factors, Asian Americans attain college education more than NHW due to higher expectations of academic achievement (35). Similarly, cultural factors might explain why a higher proportion of NHW survivors were married and lived independently, as compared with API survivors. In the United States, adult Asian Americans are more likely than Whites to reside with family members within multigenerational households because they tend to assume responsibility in providing care for elderly parents (36). A recent report also showed that the proportion of never-married Asian Americans had increased from 13% in 1980 to 19% in 2012 for various reasons, including the higher prioritization of education, career development, or life choices (37). Therefore, researchers should note that such indicators are substantially confounded by sociocultural factors and may not be accurate markers of differential functional independence in the API population. Collectively, these observations demonstrate the complex relationships between racial and sociocultural factors when examining functional attainment in childhood cancer survivors.

The influence of sociocultural factors should be considered when interpreting the results. For example, we found that API survivors reported less bodily pain than NHW survivors. This finding should not instinctively lead to the conclusion that API survivors suffer less pain than NHW survivors because the expression of distressing emotions, including cancer pain and psychologic stress, tends to be suppressed in many Asian cultures (38). Although the current analysis does not take into account the influence of sociocultural factors, future studies should examine other predictors of functional outcomes that might be influenced by race and ethnicity, including risky health behaviors, healthcare utilization patterns, and family functioning (26, 39–42). For example, psychologic stress has been associated with risky health behaviors (e.g., tobacco use, risky alcohol use, physical inactivity) in survivors of childhood cancer, and non-White race/ethnicity has been identified as a risk factor (43). However, given the small proportion of API survivors in most studies, non-Hispanic black survivors most likely dominated the minority group. In the general U.S. population, the prevalence of cigarette smoking is lower among API adults than among other racial ethnic groups (44). Notably, there are differential patterns of health behaviors even among Asian Americans; for example, tobacco use is highest among those of Korean and Vietnamese ethnicity, compared to those of Chinese and Indian ethnicity (45, 46). Consequently, the presentation of functional impairment and psychosocial outcomes may differ among API survivors. This possibility emphasizes the need to understand ethnic differences and consider sociocultural factors in long-term survivorship experiences among Asian-American subgroups. Future studies should include collection of data on health behaviors, sociocultural environment and health care system factors (medical insurance coverage and accessibility to quality survivorship care) so that we can further identify the relevant levels of intervention that are required to reduce cancer health disparities.

Our findings may lay a foundation for future multinational and multiracial research in the growing population of Asian childhood cancer survivors. We acknowledge that the outcomes evaluated in Asian-American survivors cannot be extrapolated to survivors on the Asian continent, due to potential differences in treatment regimens, health care systems, and lifestyle factors (47). However, our findings from Asian Americans may justify the need for the investigation of such outcomes in Asia, which is projected to face an emerging population of childhood cancer survivors in the next decades. Currently, the few studies that have evaluated functional outcomes in native Asian survivors have yielded inconclusive results (47, 48). Compared with healthy controls, childhood cancer survivors in South Korea reported significantly lower physical and psychosocial HRQoL scores (49). However, HRQoL and emotional distress did not differ considerably between Chinese survivors and sibling controls in Hong Kong (50). Another Japanese study also revealed no significant differences in depression and anxiety between survivors and their siblings, though survivors exhibited higher levels of post-traumatic stress symptoms and post-traumatic growth than non-cancer unrelated controls (51). A recently published systematic review reported that 10.0%–42.8% of Asian survivors of childhood cancer in Japan, South Korea, Taiwan, Hong Kong, and Thailand demonstrated mild-to-moderate impairments in intelligence (overall IQ; ref. 48). These differences in neurocognitive and psychologic functioning between Asian ethnic groups emphasize the need to tailor behavioral and psychologic interventions in multi-ethnic communities. The comparison of outcomes between API survivors in the United States and in Asia would also yield novel information about the interacting effects of genetic, region-specific environmental, and sociocultural factors on late effects and functional outcomes in cancer survivors.

Our study has the following limitations. Given the small sample of API siblings, our analysis might not be sufficiently powered to detect statistical differences in outcomes between survivors and siblings as observed in other published reports on the overall CCSS population (6, 7, 12, 52), though descriptively, the prevalence of impairment in task efficiency, memory and depression seemed higher in API survivors than API siblings. Consistent with other CCSS publications (22, 28), there is a higher proportion of siblings is in the NHW group (19.3%) than the API group (13.2%), leading to potential selection bias. The sample size was also too small to allow other potentially meaningful analyses, such as an evaluation of differences in chronic health conditions between API and NHW survivors and a subgroup analysis stratified by treatment exposure (e.g., CRT). However, to the best of our knowledge, this is the largest study to evaluate functional outcomes in API survivors of childhood cancer in the United States and to include the detailed cancer therapy history and non-cancer siblings as comparators. The association analysis of functional outcomes might be confounded by cancer diagnoses and treatment variables. For example, intrathecal methotrexate chemotherapy was associated with better task efficiency and multiple HRQoL measures, possibly because intrathecal methotrexate chemotherapy is inversely related to use and dose of CRT therapy in survivors of acute lymphoblastic leukemia (ALL) and CNS tumors, that is, those ALL survivors who did not receive intrathecal methotrexate likely received CRT and CNS tumor survivors typically do not receive intrathecal methotrexate (17, 53).

Investigations of racial and ethnic disparities in cancer survivorship have become increasingly important. In conclusion, we found differences in social attainment between API and NHW survivors, although statistically significant disparity in long-term neurocognitive and emotional outcomes were not identified. Consistent with the literature, API survivors of CNS tumors and survivors treated with higher CRT doses had a higher risk of poorer neurocognitive function and a worse HRQoL. Future studies should evaluate racial/ethnic differences in environmental and sociocultural factors among API survivors and how these may have differential effects on health and functional outcomes. From a clinical perspective, unveiling the contributors to racial/ethnic differences in functional outcomes will pave the way for the development of targeted interventions to mitigate these differences. The eventual goal is to develop or expand ongoing culturally and linguistically tailored interventions to address modifiable factors in at-risk groups, such as programs to improve social functioning and functional independence in API survivors of CNS tumors or those diagnosed with cancer at a younger age. Efforts should be made to promote health equity by race and ethnicity among all survivors of childhood cancer in the United States.

W.M. Leisenring reports grants from NIH during the conduct of the study. G.T. Armstrong reports grants from NIH during the conduct of the study. K.R. Krull reports grants from National Cancer Institute during the conduct of the study. No disclosures were reported by the other authors.

S. Sato: Conceptualization, investigation, writing–original draft. N. Li: Data curation, formal analysis, investigation, writing–review and editing. S.B. Dixon: Investigation, writing–review and editing. M. Kato: Investigation, writing–review and editing. H. Zhang: Investigation, writing–review and editing. C.K. Li: Investigation, writing–review and editing. R.M. Howell: Investigation, writing–review and editing. W.M. Leisenring: Investigation, writing–review and editing. S. Bhatia: Investigation, writing–review and editing. K.C. Oeffinger: Investigation, writing–review and editing. G.T. Armstrong: Conceptualization, funding acquisition, investigation, methodology, writing–review and editing. Y. Yasui: Conceptualization, formal analysis, supervision, investigation, methodology, writing–review and editing. K.R. Krull: Conceptualization, formal analysis, investigation, writing–review and editing. Y.T. Cheung: Conceptualization, data curation, formal analysis, investigation, methodology, writing–original draft, project administration.

This work was supported by the NCI (CA55727, G.T. Armstrong, Principal Investigator). Support to St. Jude Children's Research Hospital was also provided by the Cancer Center Support (CORE) grant (CA21765, C. Roberts, Principal Investigator) and the American Lebanese-Syrian Associated Charities (ALSAC).

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.