Biological Aging and Chemotoxicity in Patients with Colorectal Cancer: A Secondary Data Analysis Using EHR Data

1. Introduction

Colorectal cancer (CRC) remains a significant public health burden in the United States, being the third most commonly diagnosed cancer and the second leading cause of cancer-related deaths [1]. While advancements in screening, advanced treatment modalities, and systemic therapies have improved survival rates, leading to a growing population of CRC survivors, these individuals often face substantial long-term challenges due to chemotoxicity and related side effects [1]. Over the past two decades, the 5-year survival rate for CRC has increased, with approximately 65% of patients surviving beyond this milestone [2]. This rise in survivorship is attributed to earlier detection through colonoscopy screening and more effective treatment regimens, including the first-line chemotherapy (i.e., 5-Fluorouracil [5-FU]-based chemotherapy) [3]. However, a critical consequence of improved survival is the increasing number of patients experiencing persistent chemotoxicity and adverse symptoms following treatment.

A major challenge for CRC survivors is the substantial burden of chemotherapy-induced side effects, with approximately 50% of patients reporting such effects. These toxicities typically emerge during the first six months of first-line treatment and commonly include gastrointestinal (GI) complications and hematological disorders [4]. These adverse effects not only impair quality of life but may also result in early treatment discontinuation, emergency room visits, hospitalizations, dose reductions, cancer-related mortality, disease recurrence, and long-term functional impairments [4-6]. Current evidence indicates that between 40% and 70% of CRC survivors continue to experience persistent chemotherapy-related toxicity even after treatment completion, significantly compromising their daily functioning and overall well-being [6].

The growing population of CRC survivors and the high prevalence of treatment-related morbidity underscore the urgent need to further understand the mechanisms underlying chemotoxicity and develop targeted interventions to alleviate these complications [7,8]. Sensitive biomarkers of chemotoxicity remain limited due to inconclusive findings (e.g., variations by cancer stage), non-modifiable factors (e.g., race), or factors difficult to modify at the individual level (e.g., socioeconomic status, geographic disparities) [9]. However, emerging evidence suggests that biological aging processes may contribute to chemotherapy-induced toxicity in cancer patients [9]. Specifically, several modifiable (e.g., lifestyles, stress, and pro-inflammatory status) and non-modifiable (e.g., race, area-level socioeconomic status) factors are shown to be associated with chemotoxicity in cancer patients [4,10]. These risk factors are closely associated with measures of biological aging; thus, biological aging markers may potentially be used as a sensitive risk factor for chemotoxicity.

Cancer patients frequently exhibit accelerated biological aging, wherein their physiological state appears significantly older than their chronological age would suggest, compared to healthy individuals [11-13]. This phenomenon is more evident in cancer patients than healthy individuals, who may maintain robust physical and cognitive function well into older age [11,14]. The accelerated aging observed in cancer patients manifests through multiple clinical dimensions, including premature frailty, increased vulnerability to age-related diseases, and functional decline in daily activities [15,16]. These effects stem largely from the unintended consequences of cancer treatments, which can trigger fundamental aging processes at the cellular and molecular level [11,17,18]. For example, therapy-induced DNA damage and epigenetic alterations, such as aberrant DNA methylation patterns, further contribute to the accelerated aging phenotype by disrupting normal cellular function and gene regulation [19,20]. Epidemiological studies demonstrate substantially higher rates of frailty among cancer survivors compared to age-matched controls [21], with hematopoietic cell transplant recipients showing particularly pronounced effects [22]. The risk of developing secondary malignancies increases dramatically in survivors, with childhood cancer patients facing a 20% cumulative incidence of subsequent neoplasms over 30 years [23]. Cardiovascular disease, cognitive decline, and other age-related conditions also emerge earlier in cancer survivors [24,25]. To our knowledge, only a few studies in other cancer types [26-28] showed potential links between biological aging markers and overall chemotoxicity incidences and cancer survival. The link between biological aging markers and chemotoxicity in CRC is unknown.

Biological age represents an individual’s functional and physiological condition compared to their actual chronological age, serving as an indicator of the biological aging processes [29]. Recent advances in aging research have introduced multiple biomarker-based approaches to quantify biological aging more precisely. Among these, epigenetic clocks such as the Horvath, Hannum, and GrimAge models analyze DNA methylation patterns and exhibit robust correlations with lifespan and age-related health outcomes [30]. However, these epigenetic techniques typically involve complex laboratory analyses and substantial costs, making them impractical for widespread epidemiological research [31]. As an alternative, Levine Phenotypic Age offers a more practical solution by combining routine peripheral blood tests – including albumin, creatinine, glucose, hepatic renal functions, and C-reactive protein levels using routine circulatory blood samples– to evaluate biological aging comprehensively, without requiring further blood assays [32]. This approach proves particularly valuable for large-scale studies due to its cost-effectiveness and accessibility. Levine Phenotypic Age acceleration quantifies the difference between an individual’s biological and chronological ages, where positive values suggest age acceleration [32]. Existing research has established significant relationships between Phenotypic Age acceleration and mortality risks in various chronic conditions, including cardiovascular disease and diabetes [24,25]. However, the utility of Levine Phenotypic Age for measuring longitudinal changes in biological aging (both absolute values and age acceleration) during chemotherapy remains unexplored, along with the associated risk factors and their impact on chemotoxicity in CRC patients receiving 5-FU-based chemotherapy.

To address this knowledge gap, this study utilized electronic health record (EHR) data from The Ohio State University Comprehensive Cancer Center to investigate: (1) the change of biological aging as measured by Levine Phenotypic Age (without requiring further blood assays) over six months of 5-FU-based chemotherapy, (2) potential risk factors for accelerated aging, and (3) the relationship between biological aging and chemotoxicity in patients with CRC. We hypothesized that elevated biological age and age acceleration at baseline (i.e., <30 days before chemotherapy initiation) and the relative increase in these measures during 6-month chemotherapy would be associated with increased chemotoxicity. These findings could improve our understanding of aging dynamics in CRC patients undergoing chemotherapy and inform the development of personalized tools for screening chemotoxicity risk and optimizing long-term patient management strategies.

2. Materials and Methods

2.1. Data Collection

Our study employed a retrospective cohort design using de-identified EHR data from The Ohio State University Comprehensive Cancer Center, a major academic medical institution serving Ohio and surrounding Midwestern regions. Data spanned January 2013 through December 2019, with patient identification and data extraction performed by the institution's Honest Broker Operations Committee in full compliance with HIPAA regulations. The analysis focused on 1,129 adult patients (age >18 years) with stage II-III CRC who completed 8-12 cycles of 5-FU-based chemotherapy (including 5-FU monotherapy, Xeloda, or FOLFOX regimens) following tumor resection without stoma. Inclusion criteria included patients with a single primary CRC diagnosis, available routine blood test results at baseline (< 30 days before starting chemotherapy) and 6-month post-treatment initiation, documented sociodemographic characteristics, and residential zip code information. Exclusion criteria included patients with active ostomies, chronic GI conditions, steroid/immunosuppressant use, a history of neoadjuvant therapy, and/or concurrent radiation/immunotherapy, pregnancy status, or incomplete medical records at either baseline or 6-month follow-up.

2.2. Measurements

Patient Characteristics

The dataset incorporated comprehensive demographic (race, ethnicity, age, sex, marital status) and clinical variables (cancer sites, types of chemotherapy, history of other cancer treatments, cancer stages, body mass index, comorbidities, and lifestyle factors).

Biological Age Assessment

Levine's Phenotypic Age formula [32] was implemented using nine routinely measured biomarkers: albumin, alkaline phosphatase, creatinine, glucose, C-reactive protein, lymphocyte percentage, mean cell volume, red cell distribution width, and white blood cell count at baseline (<30 days before starting a chemotherapy) and 6 months after chemotherapy initiation.

The calculation formula for Levine Phenotypic Age

$$Levine Pheno Age=141.50+{\displaystyle \frac{Ln[-0.00553\times Ln(exp\left(\frac{-1.51714 \times \mathrm{e}\mathrm{x}\mathrm{p}\left(xb\right)}{0.0076927}\right)\left)\right]}{0.09165}}$$

where xb = − 19.907 − 0.0336 × Albumin (g/L) + 0.0095 × Creatinine (μmol/L) + 0.1953 × Glucose (mmol/L) + 0.0954 × LnCRP (mg/dL) − 0.0120 × Lymphocyte Percent (%) + 0.0268 × Mean Cell Volume (fL) + 0.3306 × Red Cell Distribution Width (%) + 0.00188 × Alkaline Phosphatase (U/L) + 0.0554 × White Blood Cell Count (1000 cells/ μL) + 0.0804 × chronological age (years).

Biological Age Acceleration. We computed the biological age acceleration as Levine Phenotypic Age – Chronological Age at the time of biomarker collection [32].

Social Determinants of Health (SDOH)

We included available SDOH in the EHR data, including insurance type, marital status and employment information. Geographic disparities were quantified using the Area Deprivation Index (ADI) [33,34], a composite metric (scale 1-100) incorporating 17 neighborhood-level socioeconomic indicators. Patient zip codes were matched to 2015 ADI data, with scores categorized into tertiles denoting relative disadvantages.

Chemotherapy Toxicity Evaluation

Global, GI, and Hematological toxicities. Treatment-related adverse events were tracked for six months post-initiation, combining incidences reported as: (1) clinician-reported Common Terminology Criteria for Adverse Events (CTCAE), v5.0 graded events (mild to life-threatening), and (2) relevant ICD-9/10 codes. In cancer clinical studies, adverse events (AEs) are documented and categorized according to the U.S. National Cancer Institute's (NCI) CTCAE. It is validated in cancer patients (reliability = 0.95, and high validity [35]. The positive toxicity cases required at least one positive documentation for 6 months of chemotherapy [35].

2.3. Statistical Methods

Descriptive statistics characterized the study population. We utilized IBM SPSS Statistics (version 28.0; IBM Corp., Armonk, NY, USA). Descriptive statistics were computed for all variables, including means ± standard deviations or standard errors for continuous variables and frequencies (percentages) for categorical variables. First, we examined changes in biological aging (“original values of biological age” and “age acceleration: biological age-chronological age”) from baseline to 6 months after chemotherapy initiation, using the paired t-tests. Second, we examined the associations of potential risk factors with biological aging at baseline, 6 months after chemotherapy initiation, and changes in biological aging over time for 6 months, and also examined the associations of biological aging measures with chemotoxicity using a one-way analysis of variance (ANOVA). All ANOVA models met the assumptions of normality and homogeneity of variance. Lastly, we performed both unadjusted and adjusted logistic regression analyses (with Odds Ratio [OR] and 95% Confidence Interval [CI]) to examine the impact of biological aging measures on chemotoxicity outcomes. Adjusted models controlled for covariates and other significant risk factors of chemotoxicity. In our study, we identified covariates based on the literature and Table 2 showing some relationships with biological aging and chemotoxicity outcomes, including age groups, employment status, cancer stages, comorbidities, marital status, employment status, and ADI measured at baseline. The variance inflation factor (VIF) among independent variables ranged from 1.1 to 3.7, indicating no multicollinearity in our analyses. Analyses were conducted using SPSS Version 28 (Chicago, IL) with a two-sided significance level of 0.05. Multiple comparison corrections were not applied as our study was a hypothesis-driven study [36].

3. Results

3.1. Participants Characteristics

Table 1 provides a summary of study participant characteristics, biological aging and chemotoxicity data in our study. The study population consisted of 1,129 CRC patients with a mean chronological age of 57.3 years (standard deviation 13.7, range 18-89). The cohort comprised 53.4% female participants, with racial/ethnic distribution showing 68.1% Non-Hispanic White, 29.0% Non-Hispanic Black, and 2.9% other racial/ethnic groups. No Hispanic patients were identified in our dataset. 55.8% of patients had colon cancer only, 34.5% had rectal cancer only, and 9.7% had cancer involving both sites. The majority (62.2%) were stage II, while 37.8% were stage III. Nearly 60% (59.7%) had a modified comorbidity index score above 2, indicating frequent comorbidities. The majority received GI surgery (79.7%), radiation (17.2%), and immunotherapy (5.1%). Health behaviors included 17.1% current smokers, 21.6% heavy alcohol users, and 42.9% reporting regular physical activity. Insurance coverage showed 61.5% with private insurance and 38.5% with Medicare/Medicaid. 56.0% of patients were married/partnered, and 44.0% were divorced/widowed/single. The majority of participants (42.0%) lived in moderately disadvantaged areas (ADI 34–66), while 33.0% resided in the least disadvantaged neighborhoods (ADI 0–33) and 25.0% in the most disadvantaged (ADI 67–100). Nearly half (47.1%) were currently employed, 27.6% were unemployed, and a quarter (25.4%) were retired.

Biological age. The biological age, as measured by Levine Phenotypic Age, showed an average increase from 59.1 years before chemotherapy to 61.8 years after six months of treatment, representing a mean increase of 2.7 years (Table 1). The age acceleration (i.e., difference between biological and chronological age) widened from 1.2 years at baseline to 2.8 years post-treatment, indicating accelerated aging worsened during chemotherapy.

Chemotoxicity. Chemotoxicity was frequently observed, with 56.3% experiencing global chemotoxicity, 40.8% GI toxicity, and 22.8% hematological toxicity. These findings collectively demonstrate significant biological aging effects and substantial treatment-related morbidity in this colorectal cancer population undergoing chemotherapy.

3.2. Risk Factors of Levine Phenotypic Age

Table 2 describes the risk factors of biological aging as measured by Levine Phenotypic Age, at baseline, 6 months, and changes from baseline to 6 months. As expected, younger patients (age <50) showed significantly lower Levine Phenotypic Age at baseline (48.7 vs 69.9, p<0.001) and 6 months (51.7 vs 72.3, p<0.001). Notably, stage II patients experienced a greater Phenotypic Age increase over time as compared to stage III patients (2.3 vs. 1.8 years, p=0.013). Patients with higher comorbidity burdens showed greater Levine Phenotypic Age increases over time (3.0 vs. 2.4 years, p = 0.013). Socioeconomic factors significantly impacted aging trajectories: divorced/widowed/single, residents of disadvantaged neighborhoods (ADI 67-100), and unemployed/retired study participants, all demonstrated accelerated biological aging. These findings highlight how clinical and socioeconomic factors distinctly influence biological aging during cancer treatment.

Table 2. Factors related to the Biological Age (Levine Phenotypic Age) Over the Course of Chemotherapy.

	Biological Age (Levine Phenotypic Age mean ± SE)
	At Baseline	F, pa	6 months after chemotherapyc	F, pa	Changes overtime	paired t, pb Time effects	F, pa group effects
Age group: Young Adult (18 <age <50, n=332) Older Adult (age > 50, n =797)	48.7(0.4) 69.9(0.9)	38.6,<.001	51.7 (0.4) 72.3 (0.3)	38.8,<.001	3.0 (0.06) 2.4 (0.05)	28.75, <.001 47.94, <.001	7.98, .005
Sex: Male Female	58.2(0.4) 60.0(0.3)	3.35, .067	60.5(0.4) 63.0(0.5)	3.35, .067	2.3 (0.02) 3.0 (0.03)	35.43, <.001 43.71, <.001	0.08, .769
Race/Ethnicity: Non-Hispanic White Non-Hispanic Black Non-Hispanic Other	60.1 (0.4) 59.6 (0.3) 57.5 (0.4)	2.97, .051	63.3 (0.4) 62.1 (0.5) 56.9 (0.4)	3.23, .040	3.2 (0.03) 2.5 (0.02) 2.4 (0.01)	52.01, <.001 18.73, <.001 8.25, <.001	1.12, .328
Cancer Site: Colon only Rectal only Colon and Rectal	58.0 (0.5) 60.0 (0.4) 59.3 (0.3)	2.31, .101	60.8 (0.5) 62.7 (0.3) 61.6 (0.6)	3.32, .112	2.8 (0.02) 2.7 (0.03) 2.3 (0.02)	10.39, .135 13.44, .112 16.51, .101	1.18, .313
Cancer stages: II III	58.5(0.4) 59.7(0.5)	1.58, .162	61.5 (0.5) 62.1 (0.6)	1.79, .112	3.0 (0.02) 2.4 (0.03)	21.85, <.001 49.33, <.001	4.35, .013
5-FU (fluorouracil)-based chemotherapy FOLFOX (infusion) FOLFIRI (infusion) CAPEOX (oral) Single-Agen 5-FU (infusion)	57.6 (0.4) 58.4 (0.9) 60.8 (1.3) 58.6 (1.7)	1.11, .523	60.6 (0.5) 61.0 (0.8) 63.2 (1.5) 61.2 (1.8)	0.91, .941	3.0 (0.4) 2.6 (0.8) 2.4 (1.3) 2.6 (1.5)	5.6, .542 6.3, .481 2.1, .994 6.1, .501	1.18, .313
Body Mass Index (BMI) Obese (>30) Overweight(>25, <30) Normal (>21, <25) Underweight (<21)	58.8 (0.9) 59.4 (0.8) 59.1 (0.6) 59.1 (1.3)	1.65, .209	61.8 (0.9) 62.0 (0.8) 61.7 (0.6) 61.7 (1.3)	1.11, .123	3.0 (0.9) 2.6 (0.8) 2.6 (0.5) 2.6 (1.2)	5.4, .499 6.1, .312 5.5, .561 6.5, .209	2.01, .129
Modified Comorbidity Index: >2 < 2	58.9(0.4) 58.1(0.5)	1.61, .203	61.7 (0.4) 60.6(0.5)	3.08, .080	2.9 (0.04) 2.5 (0.04)	40.23, <.001 39.15, <.001	20.3, <.001
Radiation: Yes No	58.5(0.4) 59.7(0.6)	1.77, .233	61.2 (0.5) 62.5 (0.5)	1.66, .129	2.7 (0.5) 2.8 (0.4)	9.15, .121 5.55, .312	1.87, .133
Immunotherapy: Yes No	58.1(0.4) 60.1(0.5)	1.56, .122	60.8 (0.4) 62.8(0.6)	1.82, .132	2.7 (0.02) 2.7 (0.03)	14.23, .121 11.15, .212	2.01, .122
GI surgery: Yes No	58.0(0.5) 60.2(0.6)	1.89, .132	60.7 (0.5) 62.9 (0.5)	1.55, .122	2.7 (0.03) 2.7 (0.04)	39.45, <.001 29.15, <.001	2.22, .159
Current Smoking Status: Yes No	58.9 (0.4) 59.3 (0.5)	1.96, .195	61.3 (0.5) 62.3 (0.4)	2.71, .231	2.4 (0.04) 3.0 (0.01)	51.45, <.001 43.21, <.001	2.56, .222
Current Heavy Alcohol Use: Yes No	58.5 (0.5) 59.7(0.6)	1.75, .133	61.1 (0.4) 62.6(0.5)	1.88, .195	2.6(0.05) 2.9(0.04)	49.31, <.001 41.21, <.001	2.11, .298
Routine Physical Activity: Yes No	57.8(0.5) 60.4(0.5)	1.79, .195	60.2(0.4) 63.4(0.4)	2.09, .185	2.4(0.05) 3.0(0.04)	51.41, <.001 39.41, .004	2.41, .187
Primary Insurance Types: Private Medicare/Medicaid	58.9 (0.4) 59.3 (0.5)	1.72, .194	61.3(0.5) 62.5 (0.4)	2.11, .132	2.4 (0.4) 3.2 (0.5)	12.41, .496 19.31, .312	2.41, .195
Marital status: Married/Partnered ivorced/Widowed/Single	58.5 (0.5) 59.7 (0.5)	1.74, .175	60.0 (0.5) 62.8(0.5)	2.21, .110	2.5 (0.05) 2.8 (0.05)	44.07, <.001 36.91, <.001	8.9, <.001
ADI, Tertile: 0-33 34-66 67-100	58.2(0.7) 58.7 (0.6) 60.4 (0.5)	1.98,.138	60.7(0.7) 61.4(0.5) 63.4 (0.3)	3.19, .041	2.5 (0.08) 2.7 (0.06) 3.0 (0.06)	22.8, <.001 32.46, <.001 40.21, <.001	12.99,<.001
Employment Status: Employed Unemployed/Retired	57.6 (0.5) 60.6(0.7)	7.21,<.001	60.4 (0.5) 63.4 (0.7)	7.60,<.001	2.6 (0.05) 2.8 (0.07)	35.12, <.001 28.36, <.001	4.88,<.001

Note. Characteristics of variables were described using means with standard error (SE). P-values in bold if they are <.05, as this is considered the statistical significance level, based on either using ^aANOVA for group differences or ^bpaired t-tests for time effects.

Table 2. Factors related to the Biological Age (Levine Phenotypic Age) Over the Course of Chemotherapy.

	Biological Age (Levine Phenotypic Age mean ± SE)
	At Baseline	F, pa	6 months after chemotherapyc	F, pa	Changes overtime	paired t, pb Time effects	F, pa group effects
Age group: Young Adult (18 <age <50, n=332) Older Adult (age > 50, n =797)	48.7(0.4) 69.9(0.9)	38.6,<.001	51.7 (0.4) 72.3 (0.3)	38.8,<.001	3.0 (0.06) 2.4 (0.05)	28.75, <.001 47.94, <.001	7.98, .005
Sex: Male Female	58.2(0.4) 60.0(0.3)	3.35, .067	60.5(0.4) 63.0(0.5)	3.35, .067	2.3 (0.02) 3.0 (0.03)	35.43, <.001 43.71, <.001	0.08, .769
Race/Ethnicity: Non-Hispanic White Non-Hispanic Black Non-Hispanic Other	60.1 (0.4) 59.6 (0.3) 57.5 (0.4)	2.97, .051	63.3 (0.4) 62.1 (0.5) 56.9 (0.4)	3.23, .040	3.2 (0.03) 2.5 (0.02) 2.4 (0.01)	52.01, <.001 18.73, <.001 8.25, <.001	1.12, .328
Cancer Site: Colon only Rectal only Colon and Rectal	58.0 (0.5) 60.0 (0.4) 59.3 (0.3)	2.31, .101	60.8 (0.5) 62.7 (0.3) 61.6 (0.6)	3.32, .112	2.8 (0.02) 2.7 (0.03) 2.3 (0.02)	10.39, .135 13.44, .112 16.51, .101	1.18, .313
Cancer stages: II III	58.5(0.4) 59.7(0.5)	1.58, .162	61.5 (0.5) 62.1 (0.6)	1.79, .112	3.0 (0.02) 2.4 (0.03)	21.85, <.001 49.33, <.001	4.35, .013
5-FU (fluorouracil)-based chemotherapy FOLFOX (infusion) FOLFIRI (infusion) CAPEOX (oral) Single-Agen 5-FU (infusion)	57.6 (0.4) 58.4 (0.9) 60.8 (1.3) 58.6 (1.7)	1.11, .523	60.6 (0.5) 61.0 (0.8) 63.2 (1.5) 61.2 (1.8)	0.91, .941	3.0 (0.4) 2.6 (0.8) 2.4 (1.3) 2.6 (1.5)	5.6, .542 6.3, .481 2.1, .994 6.1, .501	1.18, .313
Body Mass Index (BMI) Obese (>30) Overweight(>25, <30) Normal (>21, <25) Underweight (<21)	58.8 (0.9) 59.4 (0.8) 59.1 (0.6) 59.1 (1.3)	1.65, .209	61.8 (0.9) 62.0 (0.8) 61.7 (0.6) 61.7 (1.3)	1.11, .123	3.0 (0.9) 2.6 (0.8) 2.6 (0.5) 2.6 (1.2)	5.4, .499 6.1, .312 5.5, .561 6.5, .209	2.01, .129
Modified Comorbidity Index: >2 < 2	58.9(0.4) 58.1(0.5)	1.61, .203	61.7 (0.4) 60.6(0.5)	3.08, .080	2.9 (0.04) 2.5 (0.04)	40.23, <.001 39.15, <.001	20.3, <.001
Radiation: Yes No	58.5(0.4) 59.7(0.6)	1.77, .233	61.2 (0.5) 62.5 (0.5)	1.66, .129	2.7 (0.5) 2.8 (0.4)	9.15, .121 5.55, .312	1.87, .133
Immunotherapy: Yes No	58.1(0.4) 60.1(0.5)	1.56, .122	60.8 (0.4) 62.8(0.6)	1.82, .132	2.7 (0.02) 2.7 (0.03)	14.23, .121 11.15, .212	2.01, .122
GI surgery: Yes No	58.0(0.5) 60.2(0.6)	1.89, .132	60.7 (0.5) 62.9 (0.5)	1.55, .122	2.7 (0.03) 2.7 (0.04)	39.45, <.001 29.15, <.001	2.22, .159
Current Smoking Status: Yes No	58.9 (0.4) 59.3 (0.5)	1.96, .195	61.3 (0.5) 62.3 (0.4)	2.71, .231	2.4 (0.04) 3.0 (0.01)	51.45, <.001 43.21, <.001	2.56, .222
Current Heavy Alcohol Use: Yes No	58.5 (0.5) 59.7(0.6)	1.75, .133	61.1 (0.4) 62.6(0.5)	1.88, .195	2.6(0.05) 2.9(0.04)	49.31, <.001 41.21, <.001	2.11, .298
Routine Physical Activity: Yes No	57.8(0.5) 60.4(0.5)	1.79, .195	60.2(0.4) 63.4(0.4)	2.09, .185	2.4(0.05) 3.0(0.04)	51.41, <.001 39.41, .004	2.41, .187
Primary Insurance Types: Private Medicare/Medicaid	58.9 (0.4) 59.3 (0.5)	1.72, .194	61.3(0.5) 62.5 (0.4)	2.11, .132	2.4 (0.4) 3.2 (0.5)	12.41, .496 19.31, .312	2.41, .195
Marital status: Married/Partnered ivorced/Widowed/Single	58.5 (0.5) 59.7 (0.5)	1.74, .175	60.0 (0.5) 62.8(0.5)	2.21, .110	2.5 (0.05) 2.8 (0.05)	44.07, <.001 36.91, <.001	8.9, <.001
ADI, Tertile: 0-33 34-66 67-100	58.2(0.7) 58.7 (0.6) 60.4 (0.5)	1.98,.138	60.7(0.7) 61.4(0.5) 63.4 (0.3)	3.19, .041	2.5 (0.08) 2.7 (0.06) 3.0 (0.06)	22.8, <.001 32.46, <.001 40.21, <.001	12.99,<.001
Employment Status: Employed Unemployed/Retired	57.6 (0.5) 60.6(0.7)	7.21,<.001	60.4 (0.5) 63.4 (0.7)	7.60,<.001	2.6 (0.05) 2.8 (0.07)	35.12, <.001 28.36, <.001	4.88,<.001

Note. Characteristics of variables were described using means with standard error (SE). P-values in bold if they are <.05, as this is considered the statistical significance level, based on either using ^aANOVA for group differences or ^bpaired t-tests for time effects.

3.3. Risk Factors of Age Acceleration

Table 3 provides the risk factors for age acceleration (the age gap between biological age measured by Levine Phenotypic Age and chronological age during chemotherapy. The younger adult group showed greater age acceleration compared to the older adult group (with differences of 2.3 vs. 0.9 years at baseline and 4.1 vs. 2.3 years after treatment, Ps <0.001). Both age groups showed similar increases in age acceleration over the treatment period, with younger patients gaining 1.8 years and older patients gaining 1.4 years. Clinical characteristics showed important associations with age acceleration. Patients with stage III cancer presented with lower baseline age acceleration than stage II patients (1.3 vs 1.9 years, p<0.001), and experienced more pronounced increases in aging during treatment (3.8 vs 2.6 years, p<0.001). Patients with more comorbidities experienced a more substantial increase in age acceleration during chemotherapy (1.8 vs 1.4 years, p<0.001). Socioeconomic factors revealed disparities in biological aging patterns. Unmarried patients maintained higher age acceleration values than their married counterparts at both baseline (2.2 vs 1.0 years, p=0.006) and post-treatment (3.8 vs 2.6 years, p<0.001). Neighborhood disadvantage showed strong associations with age acceleration, as residents of the most disadvantaged areas (ADI: 67-100) had higher age acceleration values of 2.1 years at baseline and 4.1 years post-treatment, compared to residents of the least disadvantaged areas (p=0.008 and p<0.001, respectively). Employment status significantly influenced age acceleration over time, with unemployed or retired patients experiencing greater increases than employed patients (1.7 vs 1.5 years, p<0.001).

3.4. Associations of Biological Age with Chemotoxicity

Table 4 shows the results of associations of biological age (baseline and changes over time) with chemotoxicity outcomes. Patients who experienced global chemotoxicity showed markedly higher biological age (Levine Phenotypic Age) compared to those without toxicity, with values of 60.5 versus 57.7 years at baseline (p<0.001) and 63.8 versus 59.8 years after treatment (p<0.001). These patients also demonstrated greater increases in Levine Phenotypic Age over time (3.3 vs. 2.1 years, p<0.001). Patients reported GI and hematological toxicities had higher baseline Levine Phenotypic Age, and greater increases in Levine Phenotypic Age over time, as compared to the patients without these adverse events.

Similar patterns emerged for age acceleration, where global chemotoxicity cases had substantially higher age acceleration at baseline (2.1 vs. 0.3 years, p<0.001), post-treatment (4.3 vs. 1.3 years, p<0.001), and over time (2.2 vs. 1.1 years, p<0.001). Similar results were found in GI and hematological toxicities. Despite this, age acceleration rates (i.e., age acceleration changes over time) were similar among groups with GI (1.9 years accelerated) and hematological toxicities (1.9 years accelerated). The age acceleration differences were even more pronounced, with hematological toxicity cases showing higher baseline (2.3 vs. 0.1 years, p<0.001) and post-treatment (4.2 vs. 1.4 years, p<0.001) values than those without hematological toxicity.

3.5. Impact of Biological Age on Chemotoxicity

Given the significant replationship between biological age and chemotoxicity outcomes shown in Table 4, we performed further logistic regression analyses to examine the impact of biological age on chemotoxicity (i.e., strengths of the associations) in Table 5. The analysis revealed significant associations between biological aging measures and chemotoxicity risk in both unadjusted and adjusted models. For global chemotoxicity, baseline Levine Phenotypic Age showed strong predictive value with an adjusted odds ratio (aOR) of 1.27 (95% CI: 1.22-1.32, p<0.001), indicating a 27% increased risk per year of higher biological age. Changes in Phenotypic Age over time demonstrated an even greater association with risk (aOR=2.74, 95% CI: 2.45-3.04, p<0.001). Similar patterns emerged for age acceleration measures, with baseline values (aOR=1.27, p<0.001) and changes over time (aOR=2.74, p<0.001) both showing significant links to global toxicity risk.

For GI and hematological toxicities, results were similar showing significant associations between biological measures and chemotoxicity. Overall, compared to Levine Phenotypic Age and its age acceleration at baseline, changes in these biological aging measures were more strongly associated with chemotoxicity outcomes. Most of the biological aging measures at baseline and changes over time (Levine Phenotypic Age and age acceleration) in unadjusted models reached significance in the adjusted models.

4. Discussion

This study provides compelling evidence that chemotherapy in CRC patients is associated with measurable increases in biological aging, as captured by Levine’s Phenotypic Age using EHR data. Our findings demonstrate that biological aging progresses significantly over six months of treatment, with Phenotypic Age increasing by an average of 2.7 years, while age acceleration increased from 1.2 to 2.8 years. Our study also demonstrates that biological aging during chemotherapy is influenced by a combination of clinical factors, including comorbidity burden, along with socioeconomic determinants such as marital status, neighborhood characteristics, and employment status. The findings identify specific patient subgroups who may be particularly vulnerable to accelerated biological aging during cancer treatment, suggesting potential targets for more personalized treatment approaches and supportive care interventions. Below, we interpret our findings in the context of prior research, discuss clinical implications, and explore potential biological mechanisms.

Biological aging during chemotherapy. Our study provides evidence that 5-FU-based chemotherapy accelerates biological aging in CRC patients, aligning with findings from breast cancer research and preclinical models [9,37]. Although cancer therapies—including chemotherapy, radiotherapy, and immunotherapy—are designed to induce apoptosis or senescence in malignant cells, they also inadvertently promote senescence in healthy tissues, contributing to systemic biological aging [9,37]. Chemotherapy-induced biological aging likely arises through multiple interconnected pathways [9,13,27,29,38,39]. First, chemotherapy agents cause DNA damage and cellular senescence, as demonstrated by elevated expression of senescence markers such as p16^INK4A in breast cancer patients following treatment [40]. Second, chemotherapy is associated with epigenetic aging and telomere attrition. For instance, large-scale studies reveal that breast cancer survivors treated with chemotherapy exhibit 1-2 years of accelerated epigenetic aging (measured by the Horvath and GrimAge clocks) compared to non-chemotherapy controls [38]. Additionally, chemotherapy patients show shortened leukocyte telomere length in breast cancer [41] and increased inflammatory markers (e.g., IL-6, CRP) in CRC [42], which correlate with reduced survival and worse cancer health outcomes. Lastly, agents such as 5-FU and oxaliplatin generate reactive oxygen species (ROS), overwhelming endogenous antioxidant defenses and accelerating cellular aging [43]. Our findings highlight the potential utility of Levine Phenotypic Age, a biomarker derived from EHR-based inflammatory and clinical measures, in predicting chemotoxicity. Several mechanisms may explain its predictive power. Chemotherapy induces a senescence-associated secretory phenotype (SASP), leading to the release of pro-inflammatory cytokines (e.g., IL-6, CRP) that exacerbate tissue damage and systemic aging [44]. Patients with elevated baseline Levine Phenotypic Age may have diminished stress response capacity, rendering them more vulnerable to treatment toxicity [32]. The integration of biological aging biomarkers, such as Levine Phenotypic Age, into clinical practice could improve risk stratification and guide therapeutic decisions. Further research is needed to elucidate the precise mechanisms linking chemotherapy-induced aging and adverse treatment outcomes

Risk factors for biological aging during chemotherapy. Our findings align with prior work showing that cancer treatment accelerates aging biomarkers [45], but we extend this by identifying key risk factors for biological aging. The elevated biological aging in stage III versus stage II patients in our study may reflect combined effects of more aggressive therapies (e.g., chemotherapy-induced senescence) and greater tumor burden (e.g., pro-inflammatory signaling), both known to accelerate aging-related processes [4]. In our study, older adults exhibited higher baseline biological age, while younger adults showed greater age acceleration. Unlike chronological age, biological age may be influenced by factors such as lifestyle and stressors [46]. The younger group may be aging more quickly than anticipated due to more aggressive treatment plans, and chronic exposure to various risk factors of aging, such as higher work-and-life stress levels, unhealthy lifestyles, and low resilience [47]. This suggests that younger individuals are also vulnerable to chemotherapy-induced premature aging, possibly due to stronger treatment regimens or less physiological reserve. Greater comorbidities, advanced cancer stages, higher ADI (i.e., higher area-level SDOH disparities), non-employment status, and marital status (divorced, widowed, or single) were also associated with biological age and age acceleration. Potential mechanisms can be considered to explain our findings. Older adults’ elevated baseline biological age likely reflects the lifelong accumulation of comorbidities and chronic inflammation [48]. In contrast, younger adults’ accelerated aging may stem from acute stressors, unhealthy lifestyles, and more aggressive cancer treatments [49]. Social factors associated with biological aging in our study are supported by other studies: poor socioeconomic conditions (e.g., poor area-level SDOH, unemployment status, and family support) support heightened chronic stress, dysregulating epigenetic repair [50]. Poor income status, loneliness, unhealthy lifestyle, and poor environmental conditions are also associated with biological aging [50]. These findings suggest that chemotherapy may amplify biological aging, which various mechanisms can influence. This underscores the need for age-tailored interventions, such as comorbidity management, socioeconomic support, to mitigate chemotherapy-associated aging risks. Further investigations are needed to determine how biological aging differs in various chronological age groups.

Biological age is a risk factor for chemotoxicity. Our study is also the first to examine associations between biological aging and 5-FU-based chemotoxicity outcomes in CRC using Levine Phenotypic Age. Patients with higher baseline Phenotypic Age and greater increases over time had elevated risk of global, GI, and hematological toxicities. Several studies support our findings showing associations between biological aging and chemotoxicity in other cancer types. In patients with breast cancer, T cell p16^INK4a expression was positively associated with peripheral neuropathy after 1.5 years of treatment with paclitaxel and docetaxel [51], and fatigue after one month of taxane-based chemotherapy [52]. Multiple studies have shown a link between baseline telomere length and chemotherapy-induced toxicities in breast cancer and lymphoma patients [53-55].

Additionally, recent research identified a connection between p16^INK4a expression and epigenetic aging markers such as the Hannum and PhenoAge DNA methylation clocks, along with various mRNA signatures of T cell senescence. This study suggests that aging-related biomarkers, particularly p16^INK4a and epigenetic clocks, may help predict physical and cognitive frailty in older adults with blood cancers [56]. Despite the current literature on biological aging and chemotoxicity mentioned above, indicators of epigenetic age (e.g., DNA methylation clocks) and T cell expression profiles remain limited for routine clinical use due to insufficient clinical validation, high costs, and the technical complexity of measuring these markers [9]. Therefore, our study enhances the existing literature by demonstrating the potential utility of the blood-based Levine Phenotypic Age algorithm to predict various types of chemotoxicity (GI and hematological) commonly observed in patients with CRC. These results demonstrate that easy-to-use biological aging measures (i.e., Levine PhenoAge) can serve as robust predictors of various chemotoxicities in CRC. This suggests that biological aging assessments could help identify high-risk patients before treatment and monitor toxicity risk during chemotherapy. The maintained significance in adjusted models in our study underscores the independent predictive value of these aging measures beyond sociodemographic and clinical factors [15].

Clinical Implications

Our findings suggest that biological aging assessments could improve risk stratification before chemotherapy. For example, clinicians can consider identifying patients at high risk of chemotoxicity by utilizing biological aging and related factors. Furthermore, researchers can consider developing risk screening tools by integrating biological aging markers and relevant clinical and social factors to guide treatment decisions. Second, patients with high Levine Phenotypic Age and age acceleration may benefit from dose modifications, comorbidity management, enhanced supportive care, or anti-aging interventions such as antioxidants (e.g., coenzyme Q10) or lifestyle interventions to mitigate toxicity [57]. Third, Levine Phenotypic Age can be used to monitor long-term survivorship care in patients with biological aging. For example, the high-risk groups of aging may need earlier surveillance and rehabilitation for aging-related chronic diseases, such as cognitive impairment and cardiovascular diseases.

Limitations and Future Directions

This is a secondary data analysis, so several factors relevant to our study were not available, such as education levels, annual household income, which can better reflect individual socioeconomic status. Furthermore, our study was not able to address some confounders such as stress levels and diet quality, in our analyses. Since clinicians do not use the CTCAE tool or patient-reported (PRO) CTCAE on a routine basis to report chemotoxicity at the many tertiary cancer centers, this could limit the accurate capture of chemotoxicity data. Therefore, routine use of the clinician-reported CTCAE tool as well as PRO data should be considered in future studies. Future validation of the Levine Phenotypic Age in a separate cohort of CRC patients is needed to confirm our findings. Using only two timepoints of follow-up data has limitations in drawing clear causal relationships between biological aging and chemotoxicity outcomes. Addressing long-term follow-up is required to determine if accelerated aging persists post-treatment and impacts long-term survivorship outcomes (e.g., frailty, quality of life, function, fatigue and secondary cancers). Furthermore, before considering anti-aging interventions for CRC patients, future research should prioritize evaluating existing aging metrics, including molecular biological markers and clinical frailty, to determine the optimal tool or develop novel aging-based screening and chemotoxicity management strategies for use in CRC. Lastly, current knowledge about aging in cancer survivors is largely based on studies of older adults with non-CRC, underscoring the need to expand investigations to both young and older adults with CRC.

5. Conclusions

This study demonstrates that chemotherapy accelerates biological aging in CRC patients. Levine's Phenotypic Age, calculated using routine blood test results from EHR, served as a strong predictor of treatment toxicity (global, GI, and hematological). Both young and older adults, those with greater comorbidities, more advanced cancer stages, and socioeconomic disadvantaged (including area-level disparities) individuals are at higher risk for chemotoxicity. Future research should validate the current findings in different cohorts of patients with CRC to facilitate the use of blood-based biological aging markers in routine clinical use, develop aging-based chemotoxicity screening tools, and explore whether targeting aging mechanisms (e.g., anti-inflammatories) can improve treatment tolerance and survivorship outcomes.