Deficit Accumulation Index and Biological Markers of Aging in Survivors of Childhood Cancer

This cross-sectional study assesses the use of a deficit accumulation index to measure aging-related conditions, such as epigenetic age and telomere length, in adult survivors of acute lymphoblastic leukemia, Hodgkin lymphoma, or central nervous system tumors.


Introduction
Survivors of childhood cancer are exposed to therapies that may engender substantial molecular damages at a young age, playing a role in an accelerated aging trajectory during their life span. 1 A recent study found that such survivors experience substantial premature aging compared with community controls according to a measure of physiological aging called the deficit accumulation index (DAI), 2 which can be constructed using readily available clinical information from medical records and routine patient reports.The DAI comprises 30 or more aging-related items, such as chronic health conditions, activities of daily living, and psychosocial well-being.The underlying hypothesis is that molecular damage over time is associated with physiological deficits that accumulate and can be counted. 33][4][5][6][7][8][9] However, little data exist to demonstrate how the DAI is associated with markers of underlying biological aging among survivors of childhood cancer.
2][13] Various DNA methylation-based epigenetic clocks have been proposed to estimate the epigenetic age of an individual. 10Epigenetic age acceleration (EAA), defined as the difference between epigenetic age and chronological age, is associated with an increased risk of multimorbidity and mortality. 14Telomeres protect the ends of chromosomes during mitosis and undergo attrition during normal cellular division, ultimately resulting in replicative senescence or apoptosis.Thus, telomere shortening serves as another molecular marker for biological aging. 11Survivors of childhood cancer have shorter mean leukocyte telomere length (LTL)   and greater EAA compared with community controls, suggesting accelerated biological aging. 15,16th of these aging-related biomarkers are associated with chronic health conditions in survivors of childhood cancer 15,16 but have yet to be examined in the context of a comprehensive measure of aging, such as the DAI.
We designed this study to examine the associations between the DAI and EAA or mean LTL, 2 highly cited molecular biomarkers of aging. 17We hypothesized that childhood cancer survivors with a higher DAI would have greater EAA and shorter mean LTL.Understanding the association between these markers could inform the underlying mechanisms of aging phenotypes in survivors of childhood cancer and identify potential intervention targets.These data may further validate the DAI as a marker of biological aging in a population of young adult survivors. designed an aging-related DAI using 44 items available from the SJLIFE Cohort questionnaires, comprehensive clinical and physical examinations, and medical records (eTable 1 in Supplement 1). 2,3,19This DAI included a diverse set of aging-related items across multiple systems, including chronic health conditions and functional, psychosocial, and mental well-being that were all assessed concurrently.Each item was weighted from 0 to 1 (with 1 indicating presence of deficit or most severe deficit).Item weights were summed and divided by the total items available, resulting in a ratio ranging from 0 to 1.A higher DAI indicated greater deficit accumulation.1][22] Any participant with more than 10% of items missing from the DAI was excluded, consistent with prior studies.The DAI was calculated at the same time as the blood draw for the genetic or epigenetic profiling.

Methods
Sex; time since diagnosis; and treatment history, including chemotherapy, surgical procedures, and radiotherapy (RT), were abstracted from medical records.Age at the time of DAI and blood draw was recorded.
Genome-wide methylation profiling data were generated for the SJLIFE Cohort using the Infinium MethylationEPIC BeadChip (Illumina) for 850K cytosine-guanine dinucleotide (CpG) sites. 16 hypothesized that the DAI was a measure of physiological aging; therefore, we wanted to find its association with an epigenetic age that was derived from an epigenetic clock that was trained to identify CpG sites associated with morbidity.Epigenetic age based on the Levine clock (ie, DNAm PhenoAge) was calculated for each individual using 513 informative CpG sites, which were previously shown to predict a composite phenotypic age encompassing 9 blood-based clinical biomarkers. 23Am PhenoAge was chosen because it outperforms other epigenetic clocks in estimating all-cause mortality, time to death, and age-related health outcomes in the general population. 10Additionally, because the DAI is deemed to represent physiological aging, we picked a clock most likely to relate to physiological aging rather than intrinsic aging (eg, Horvath or Hannum clocks).
A previous study reported that EAA based on DNAm PhenoAge was associated with specific cancer treatment exposures and chronic health conditions 16 and found different cross-sectional patterns of DNAm PhenoAge across chronological age-defined groups. 24We derived the EAA by calculating the residual for each individual based on the least-squares regression of estimated DNAm PhenoAge against chronological age at blood draw for DNA sampling.The WGS data were generated using the HiSeq X Ten System (Illumina). 25 The TelSeq 26 software (TelSeq) was used to estimate mean LTL from the WGS data. 15The TelSeq estimates were highly correlated with Southern blot

JAMA Network Open | Oncology
Deficit Accumulation Index and Biological Markers of Aging in Cancer Survivors measurements based on 93 samples from the SJLIFE Cohort (r = 0.64). 15Mean LTL was regressed on chronological age at DNA sampling, with additional adjustment for DNA extraction methods to obtain the residual for subsequent analysis.The WGS and methylation data were generated from DNA extracted from the same time point.

Statistical Analysis
All analyses were done separately for the 2 primary exposures: EAA and mean LTL residual.Linear abdominal RT, and alkylating agents) and may be potential confounders of these associations. 15,16rst, a linear regression model examined the associations between treatments and mean LTL residual or EAA adjusted for sex and time since diagnosis.Second, the DAI was added to this model, and change in R 2 was examined using an F test. 27 All analyses were conducted in SAS, version 9.4 (SAS Institute Inc), between January 2022 and January 2023.Two-sided P = .05was considered statistically significant.

Results
We

Discussion
Survivors of childhood cancer were exposed to curative treatments associated with cellular and molecular damages to normal tissues during key developmental stages, potentially altering their physiological and biological aging trajectory compared with individuals without a history of  childhood cancer.Ultimately, this accelerated aging can be a factor in increased burden of age-related chronic health conditions in this population.A previous study demonstrated that the DAI was associated with aging-related outcomes in survivors of childhood cancer, suggesting that it may serve as a marker of physiological aging. 2 The present study further found that the DAI was associated with EAA, a marker of biological aging.The EAA has been widely used in molecular aging research in the general population and more recently among survivors of childhood cancer. 16,28ecifically, we found that survivors with a high DAI had 3.7 years greater EAA than those with a low important evidence for the clinical utility of the DAI as a marker of aging.Generally, there is substantial heterogeneity in aging that is not clinically apparent, especially in younger cancer survivors.Therefore, tools such as the DAI that help clinicians to identify those patients most at risk for accumulation of deficits beyond the deficits expected based on chronological age may be used to personalize care toward treating the underlying biological mechanisms rather than managing the symptoms inventoried in the DAI.Information about deficit accumulation may motivate patients to engage in various health behaviors that prevent or decrease existing deficits (eg, smoking cessation or increasing physical activity).For example, eating nutritious foods or exercising regularly may affect multiple hallmarks of aging, reduce deficits, and improve aging-related 31 While the present study had a cross-sectional design, the results, especially the dose-response association between DAI and EAA, are consistent with a mechanistic association between deficit accumulation and biological aging.In future work, examining longitudinal associations to support causal inferences will be critical.
The association between DAI and EAA in a young adult population of cancer survivors allows for flexibility in the choice of a primary marker of aging in the research setting.The DAI can be ascertained in an office visit, via self-reported questionnaire data, and by medical record data.The items assessed in the DAI can vary across studies, provided that each item is associated with aging and there are at least 30 items.Therefore, compared with EAA, the DAI may be a more convenient, feasible, and inexpensive surrogate for biological aging.The DAI may be particularly useful in large population-based studies when bioassays are cost-prohibitive.Alternatively, measuring EAA may be useful in identifying molecular mechanisms or detecting subtle changes in aging, such as after an intervention (prognostic value) or before the onset of health conditions (predictive value).The finding that DAI was associated with EAA in survivors of childhood cancers suggested that either the DAI or EAA may be used to measure aging, depending on the goals and feasibility constraints of the proposed research.
Interestingly, the data found that the DAI was associated with EAA, which measures biological aging through epigenetic dysregulation, but not with mean LTL, which is highly relevant to cellular senescence.3][34][35] While both EAA and mean LTL were considered to be biomarkers of aging, EAA was reflective of a mechanism unrelated to mitotic age (telomere shortening). 36Previous studies reported that EAA followed an accelerated aging trajectory and mean LTL followed an accentuated aging trajectory in survivors of childhood cancer. 15,16The DAI also followed an accelerated aging trajectory in survivors of childhood cancer, which may make it more similar to the EAA.Additionally, telomere attrition did not have marked association with cell physiological function until a critical telomere length was reached, at which point the cell became senescent. 37,38e results suggested heterogeneity in the associations between DAI and EAA by diagnosis, with the most consistent associations noted among survivors of Hodgkin lymphoma.This finding is consistent with previous works that demonstrated Hodgkin lymphoma survivors were more likely to have a medium or high DAI and higher EAA compared with community controls. 2,16Among CNS tumor survivors, however, only those with a high DAI had an association with EAA.Previous studies could not identify an association between cranial RT and EAA, 16 but cranial RT was associated with an increased DAI. 2 Therefore, this pattern may be indicative of a threshold for the cumulative indirect implications over time of CNS-directed therapy for aging.These data emphasized the heterogeneity in aging among survivors of childhood cancers and supported the need for future research on how to use markers of aging to personalize care.

Strengths and Limitations
This study has several strengths, including the large number of well-characterized survivors of  32,33 This study also has limitations.The findings were limited by the cross-sectional nature of the data, and we cannot infer that changes to DNA methylation patterns resulted in a higher deficit accumulation or vice versa.Future longitudinal studies are needed to delineate temporal associations between EAA and the DAI.We estimated telomere length from WGS; therefore, the finding that there was no association between the DAI and mean LTL may be biased by measurement error. 26ditionally, the analyses were restricted to survivors with European ancestry, which limits the generalizability of these findings.It remains unclear whether the DAI is associated with EAA among survivors with non-European ancestry, although the literature suggested that survivors with Hispanic and Black race and ethnicity may be at higher risk for EAA and deficit accumulation. 2,14Expanded molecular profiling in survivors and controls from underrepresented racial and ethnic groups as well as new enrollment of individuals with non-European ancestry are being implemented to answer these questions.Future research will be able to confirm or refute the generalizability of the current findings.Our ability to detect differences across diagnostic groups was limited by the sample size.

Conclusions
The

Figure 1 .
Figure 1.Association Between the Deficit Accumulation Index (DAI) and Biomarkers of Aging

Figure 2 .
Figure 2. Association Between the Deficit Accumulation Index (DAI) and Mean Leukocyte Telomere Length (LTL) or Epigenetic Age Acceleration (EAA) by Diagnosis

JAMA Network Open | Oncology Deficit
18is cross-sectional study analyzed data from the St Jude Lifetime (SJLIFE) Cohort, which was established in 2007 to facilitate prospective clinical assessments of survivors of childhood cancer Accumulation Index and Biological Markers of Aging in Cancer Survivors Research Hospital in Memphis, Tennessee.18TheSt Jude Children's Research Hospital Institutional Review Board approved the SJLIFE study protocol, including the present analysis.All participants provided written informed consent for use of their data and biospecimens.We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.
JAMA Network Open.2023;6(11):e2344015. doi:10.1001/jamanetworkopen.2023.44015(Reprinted) November 20, 2023 2/13 Downloaded from jamanetwork.comby guest on 11/24/2023 who were treated at St Jude Children's regression estimated the adjusted least-squares mean of EAA or mean LTL residual for each DAI category (low, medium, high) adjusted for sex and time since diagnosis, which were considered potential confounders.Models were not adjusted for chronic health conditions or health behaviors because these items or concepts were included in the DAI.A priori, we hypothesized that the association between the DAI and EAA or mean LTL residual may differ by diagnosis group or by age group (eg, age and diagnosis may be effect modifiers).Therefore, models were repeated for each of the common diagnosis groups (acute lymphoblastic leukemia, central nervous system [CNS] tumors, and Hodgkin lymphoma) and again by age groups (<30, 30 to <40, and Ն40 years).To inform how the DAI played a role in the estimation of EAA and mean LTL residual, we ran 2 additional linear regression models accounting for treatments previously found to be associated with these biomarkers (mean LTL model: chest RT, abdominal RT, and vincristine; EAA model: chest RT, However, the adjusted least-squares mean of EAA increased across the 3 DAI groups (Figure 1B), with a linear dose response (greater values of EAA reflected greater aging).Compared with survivors in the low DAI group, those in the high DAI group Deficit Accumulation Index and Biological Markers of Aging in Cancer SurvivorsTable.Demographic and Clinical Characteristics of the Study Sample included 2101 childhood cancer survivors with European ancestry, of whom 979 were females (46.6%) and 1122 were males (53.4%) with a mean (SD) age of 33.9 (9.1) years.The median (IQR) time since diagnosis was 25.1 (18.7-31.9)years(Table).The most common diagnoses were acute lymphoblastic leukemia (717 [34.1%]),Hodgkin lymphoma (260 [12.4%]), and CNS tumors (227[10.8%]).Among the survivors, 237 (11.3%) had a high DAI, 428 (20.4%) had a medium DAI, and 1436 (68.3%) had a low DAI.Participants with biomarker and DAI data had similar demographic and clinical characteristics to those of participants without these data excluded from this analysis (eTable 2 in Supplement 1).After adjusting for sex and time since diagnosis, there was no association between mean LTL residual and DAI groups (Figure1A).10% of the total variance in EAA in the treatment-adjusted model.Mean LTL residual did not appear to be associated with the DAI, and results were similar regardless of diagnosis (Figure2).Survivors of CNS tumor had a greater adjusted least-squares mean of LTL than survivors of acute lymphoblastic leukemia and Hodgkin lymphoma.In contrast, there was an association between EAA and DAI.Specifically, among acute lymphoblastic leukemia JAMA Network Open | Oncology JAMA Network Open.2023;6(11):e2344015. doi:10.1001/jamanetworkopen.2023.44015(Reprinted) November 20, 2023 5/13 Downloaded from jamanetwork.comby guest on 11/24/2023 survivors, those in the medium DAI group experienced greater EAA compared with those in the low DAI group (β = 2.27 [95% CI, 0.78-3.76;P = .001])(eTable 3 in Supplement 1).Although no differences were seen between DAI groups, we found a dose-response association qualitatively, with EAA increasing across DAI groups among Hodgkin lymphoma survivors.Among CNS tumor survivors, only those in the high DAI group had a significantly elevated adjusted least-squares mean of EAA compared with those in the low DAI group (β = 4.34; 95% CI, 0.82-7.86;P = .01)

Table .
Demographic and Clinical Characteristics of the Study Sample (continued) Based on 2018 Children's Oncology Group Long-Term Follow-Up Guidelines.
a A total of 180 participants were missing educational level data.bAtotal of 36 participants were missing smoking status.cHigh dose was 1 g/m 2 per treatment.d Deficit Accumulation Index and Biological Markers of Aging in Cancer Survivors childhood cancer and robust measurement of physiological aging based on DNA methylation profiling.Because we hypothesized that the DAI was a measure of physiological aging, we associated it with epigenetic age from an epigenetic clock such as DNAm PhenoAge.Furthermore, this JAMA Network Open | Oncology JAMA Network Open.2023;6(11):e2344015. doi:10.1001/jamanetworkopen.2023.44015(Reprinted) November 20, 2023 9/13 Downloaded from jamanetwork.comby guest on 11/24/2023 epigenetic clock was most consistently associated with deficit accumulation in aging adults without cancer.
DAI was significantly associated with EAA but not with mean LTL among survivors of childhood cancers.Both the DAI and EAA were effective at identifying aging phenotypes in this study, and either may be used alone or with mean LTL to measure aging and response to interventions targeting aging pathways, depending on the proposed aims of future clinical and population-based research.37. Blackburn EH.Telomere states and cell fates.Nature.2000;408(6808):53-56. doi:10.1038/3504050038.Sanders JL, Newman AB.Telomere length in epidemiology: a biomarker of aging, age-related disease, both, or neither?Epidemiol Rev. 2013;35(1):112-131.doi:10.1093/epirev/mxs008SUPPLEMENT 1. eFigure.Participant Flow Chart eTable 1.St. Jude Lifetime Cohort Aging-Related Deficits Accumulation Index eTable 2. Characteristics of Those With and Without Complete Data eTable 3. Specific Results From Linear Regression Models Examining the Associations of mean LTL and EAA Among Survivors of Childhood Cancer eTable 4. Distribution of DAI Groups by Age Categories