The Pace of Biological Aging Predicts Non-Specific Chronic Low Back Pain Severity

Abstract

This study aimed to determine if and how the pace of biological aging was associated with nonspecific cLBP and to compare what measure of epigenetic age acceleration most strongly predicts cLBP outcomes. We used the Dunedin Pace of Aging from the Epigenome (DunedinPACE), Horvath’s, Hannum’s, and PhenoAge clocks to determine the pace of biological aging in 69 cLBP, and 49 pain-free controls (PFCs) adults, ages 18 to 85 years. On average, participants with cLBP had higher DunedinPACE (p < 0.001) but lower Horvath (p = 0.04) and Hannum (p = 0.02) accelerated epigenetic age than PFCs. There was no significant difference in PhenoAge acceleration between the cLBP and PFC groups (p = 0.97). DunedinPACE had the largest effect size (Cohen’s d = 0.78) on group differences. In univariate regressions, a unit increase in DunedinPACE score was associated with 265.98 times higher odds of cLBP than the PFC group (p < 0.001). After controlling for sex, race, and BMI, the odds ratio of cLBP to PFC group was 149.62 (p < 0.001). Furthermore, among participants with cLBP, DunedinPACE scores positively correlated with pain severity (rs = 0.385, p = 0.001) and interference (rs = 0.338, p = 0.005). Epigenetic age acceleration from Horvath, Hannum, and PhenoAge clocks were not significant predictors of cLBP. The odds of a faster pace of biological aging are higher among adults with cLBP, and this was associated with greater pain severity and disability. Future interventions to slow the pace of biological aging may improve cLBP outcomes.

Non-specific chronic low back pain (cLBP) is a prevalent and debilitating musculoskeletal problem affecting a substantial proportion of adults worldwide.¹ The exact etiology of non-specific cLBP is unknown but has been associated with advancing age. The prevalence of cLBP increases linearly from the 3^rd to 6^th decades of life,² and about 21 to 68 percent of adults 60 years and older have cLBP.³ A 2022 systematic review and meta-analysis found that non-specific cLBP is highly prevalent in older adults, leading to significant disability, decreased work productivity and reduced quality of life.⁴ Given the aging population, there has been much interest in investigating psychosocial as well as biological factors associated with higher prevalence and comorbid factors of non-specific cLBP in adults.^5–7 In 2019, the National Health Interview Survey found that about 6 in 10 adults had pain in the past 3 months, with cLBP being the most common type of pain. The percentage of adults with cLBP increased from about 28% in those 18 to 24 years to about 46% in those 65 years and older.⁸ Also, a research consortium study of “Localized and Generalized Musculoskeletal Pain: Psychobiological Mechanisms and Implications for Treatments” found a paradoxical relationship between chronological age and pain-related outcomes (depression, cognitive function, disability, pain intensity, and perceived control over life) among adults with cLBP.⁵ However, most of these studies have focused on chronological age (the number of years an individual has been alive), which may not be the best indicator of how an individual ages or the risks of morbidity and mortality.⁹ An objective biological measure of the aging process that can reliably differentiate individuals with nonspecific cLBP from pain-free controls (PFCs) can shed more light on the underlying mechanisms.

Unlike chronological aging, which is the same rate for everyone, biological aging can vary substantially between individuals, leading to disease risk, morbidity, and mortality variations.^10,11 Biological aging refers to the graduate multi-system decline in physiological function due to the accumulation of damage to the body in response to genetic predisposition, lifestyle, and environmental exposures, leading to increased disease risk.¹⁰ While several approaches have been developed to measure biological markers of aging, DNA methylation (DNAm) is a common type of epigenetic modification that predominantly occurs at cytosines in a CpG (Cytosine Phosphate Guanine) dinucleotide context in differentiated mammalian cells in response to environmental exposures.¹² The stability of DNAm provides a reliable approach to measure biological age and aging – DNAm clocks or epigenetic clocks.¹³ DNAm clocks use the methylation levels in specific CpG sites to predict age, health, lifestyle, and disease risk.

Emerging evidence suggests that epigenetic age acceleration correlates with chronic pain and psychosocial predictors of chronic pain.^14–16 Cruz-Almeida and colleagues reported a higher epigenetic age acceleration in community-dwelling adults with chronic musculoskeletal pain,¹⁵ and worse pain outcomes.¹⁶ At the same time, Kwiatkowska et al. found no relationship between DNAm age and heat pain sensitivity as well as fibromyalgia.¹⁷ Most epigenetic age and chronic pain studies have used first or second-generation epigenetic clocks. This is concerning because first-generation epigenetic, Hannum¹⁸ and Horvath,¹⁹ clocks were developed to predict an individual’s chronological age based on DNAm modifications at 71 and 353 CpG sites’, respectively. While these clocks highly correlate with chronological age, their prediction of morbidity and mortality was fairly weak.^10,13 Second-generation, PhenoAge,²⁰ and GrimAge²¹, clocks have also been criticized for their focus on morbidity and time to all-cause mortality/lifespan, which limits their ability to predict biological aging.

Recently, researchers have developed a novel method to quantify the pace of biological aging. The Dunedin Pace of Aging from the Epigenome (DunedinPACE) was trained as a third-generation epigenetic clock to predict individual differences in the rate of physiological decline over time.²² Given that the DunedinPACE was trained to predict the pace of aging, it may be a more robust predictor of an age-related chronic condition such as non-specific cLBP. To our knowledge, no study has investigated the association between biological pace of aging and non-specific cLBP using a robust measure such as DunedinPACE.

In the present investigation, we utilized DunedinPACE to investigate if and how the pace of biological aging is associated with non-specific cLBP in community-dwelling adults. We hypothesized that after controlling for chronological age, individuals with non-specific cLBP would have higher DunedinPACE scores (faster rate of biological aging) compared to pain-free controls (PFCs). For comparative analysis, we used first- and second-generation epigenetic clocks to determine which measure of epigenetic age acceleration most strongly predicts non-specific cLBP. We hypothesized that DunedinPACE would have a stronger association with cLBP outcomes than epigenetic aging accelerates from first- and second-generation epigenetic clocks.

Methods

Overview

We conducted this study in accordance with the ethical guidelines of the Helsinki Declaration. All study participants provided written informed consent, and the Institutional Review Board (IRB) at the University of Alabama at Birmingham approved the study protocol. Data for this cross-sectional, secondary data analysis were collected as part of an ongoing study examining epigenomic as well as racial and socioeconomic status disparities in cLBP. Phenotypic data were collected as part of an R01 study (R01MD010441), and epigenomic data was collected as part of an ancillary R01 study (R01AR079178).

Participants and Procedures

The current study included 118 participants (69 were diagnosed with cLBP, and 49 were PFCs), ages 18 to 85 years. We previously published details of the study protocol.^23–25 Briefly, all participants were recruited to take part in the study through flyers posted at pain clinics and communities around the University of Alabama at Birmingham. Eligibility was assessed over the telephone and confirmed with electronic health records before the assessment visit. Besides the electronic health records, the diagnosis of non-specific cLBP was confirmed using the joint clinical practice guidelines from the American College of Physicians and the American Pain Society.²⁶ All participants with nonspecific cLBP reported persistent pain for at least three consecutive months and at least half the days in the past six months.²⁷ Participants were excluded if they had any of the following: 1) cLBP attributable to infection, trauma, malignancy, or ankylosing spondylitis, 2) systemic infection, 3) chronic inflammatory disease, 4) poorly controlled diabetes, 5) systemic rheumatic disease (e.g., rheumatoid arthritis, systemic lupus, erythematosus, fibromyalgia), 6) uncontrolled high blood pressure (SBP > 150mmHg and/or DBP > 90mmHg), and 7) neurological diseases. The same inclusion and exclusion criteria applied to PFCs except for the diagnosis of non-specific cLBP.

On the assessment day, a research staff explained the consent form to the participants with the opportunity to read and ask questions. All participants supplied verbal and written consent before starting the study. Only measures relevant to this study are reported.

Sociodemographic Information

All participants self-identified their assigned sex at birth (male, female, or intersex) and “race” (American Indian or Alaska Native, Asian, Black/African American, Native Hawaiian or Other Pacific Islander, and White). Given that the primary study was focused on disparities between Blacks and Whites, all non-Black, non-White, and multi-racial participants were excluded from the study. Socioeconomic status was self-reported as household income after taxes: < $24,999; $25,000 to $49,999; $50,000 to $74,999; $75,000 to $99,999; and >$100,000.

Pain Severity and Pain Interference

Pain severity and interference were assessed using the Brief Pain Inventory Short Form (BPI-SF).²⁸ The BPI-SF is a widely used and validated self-report questionnaire that measures the intensity and impact of pain on various aspects of an individual’s daily life. Using a numeric scale of 0 to 10, participants rated the worst, least, and average in the last 24 hours as well as current pain. The average score measures pain severity, where 0 represents no pain, and 10 illustrates the most severe pain. Participants also rated how pain interferes with daily activities (general activity, mood, walking ability, work, relations with others, sleep, and enjoyment of life), and the average score represents pain interference (0 = no interference and 10 = maximum interference). For this study, BPI-SF had excellent internal consistency reliability with Cronbach’s alpha of 0.92.

EPIC Array Processing

Genomic DNA was extracted from peripheral blood samples using the Gentra Puregene DNA Purification Protocol (Qiagen, Valencia, CA, United States) and quantified using a Nanodrop 2000 spectrophotometer (Thermo Fisher Scientific). Extracted DNA was shipped on dry ice to the University of Minnesota Genomic Center for sequencing. Following the manufacturer’s guidelines, genome-wide DNA methylation was assessed with the Infinium EPIC (850K) array.

Probe-level intensity datasets from Illumina’s iScan system were exported as raw methylome (*.idat) data objects and other appropriate sample-level datasheets were created for downstream qualification of methylation estimates. Data file formats adopted in our bioinformatics efforts are suitable options for widely used methylation data analysis modules like minfi²⁹ and wateRmelon R packages³⁰. Given the advancements in statistical methodologies and evaluating the influence of normalization techniques on epigenome data, we explored 4 major normalization schemes namely, SWAN,³¹ Illumina,³² Noob,³³ and Quantile³⁴. Raw IDAT files was imported using Minfi R package to visualize the raw intensity data and identify sex mismatches and outlier presence. Principal variance component analysis (PVCA) module of minfi was employed to quantify the variance explained as cell proportions for CD8 T lymphocytes, CD4 T lymphocytes, natural killer cells, B cells, monocytes, and granulocytes estimation. A combination metrics of percentage of sample outliers, detection rates based on p-value for each sample at each probe and individual probes are considered to ensure positive detection of methylation signals.³⁵ All 4 normalization techniques yielded similar methylation beta estimates. methyAge module of ENmix was used to estimate epigenetic age. Popular epigenetic clock calculators like Hovath, Hannum or PhenoAge methods and pace of aging DunedinPACE are implemented as a one-stop solution in the methyAge module of Enmix.³⁶

Epigenetic Age Acceleration and Biological Aging

Epigenetic age was calculated with the procedure developed by Horvath¹⁹, Hannum¹⁸, and Levine (PhenoAge)²⁰ for blood samples. Horvath’s age was calculated using the 353 CpGs, which have been associated with cell intrinsic aging.¹⁹ Hannum’s epigenetic age was calculated from the 71 CpGs, which have been associated with immunological aging,¹⁸ and Levine’s epigenetic age (PhenoAge) was calculated from 513 CpGs.²⁰ The one-stop solution in the methyAge module of Enmix was used to calculate epigenetic age estimates.³⁶ For each epigenetic age, we calculated epigenetic age acceleration as the difference between epigenetic age and chronological age, as described initially by Horvath.¹⁹ The pace of biological aging was calculated as the DunedinPACE score.²²

Statistical Analysis

All data were analyzed using IBM SPSS for Windows version 29 (SPSS Inc., Chicago, IL, United States). We checked data for missing values, normality, and outliers. No outliers were found, and only samples with complete phenotype and epigenetic ages were included in the analysis. Descriptive analyses compared chronological age, epigenetic age, and general sociodemographic variables between non-specific cLBP and PFC groups. Spearman rho’s rank correlations were used to assess the relationship between variables since several vital variables were not normally distributed. Given that the t-test is robust, it was used to determine average group differences.

To determine what measure of biological aging was a stronger predictor of cLBP, we used binary logistic regression to estimate the odds ratio of having cLBP versus the PFC group. The dependent variable was pain status (cLBP vs. PFC), and the DunedinPACE score was the predictor in various models. Model 1 was the univariate unadjusted model; Model 2 adjusted for sex; Model 3 adjusted for sex and race; and Model 4 adjusted for sex, race, and BMI. A similar analysis was performed using epigenetic age acceleration derived from Horvath, Hannum, and PhenoAge clocks for comparative analyses. As part of SPSS Process Macro, we used a bootstrapping technique to resample the dataset with 5000 bootstrap samples. Statistical significance was set at two-tailed p < 0.05. We did not control for chronological age in the models because of the collinearity between chronological age and biological aging.

Results

The final sample for this study consisted of 69 adults with non-specific cLBP and 49 PFCs. Half of the participants identified as male, and slightly more than half (58.5%, n = 69) self-identified as African American. The average chronological age of the participants was 43.57 (SD = 14.15) years at the time of DNA sampling. The average epigenetic age of the participants as measured with Horvath, Hannum, and PhenoAge clocks were 53.15 (SD = 11.63), 50.94 (SD = 12.22), and 36.93 (SD = 13.72) years, respectively. As shown in Figure 1, chronological age strongly correlates with epigenetic age: Horvath (r_s = 0.89, p < 0.001), Hannum (r_s = 0.89, p < 0.001), PhenoAge (r_s = 0.90, p < 0.001) and DunedinPACE (r_s = 0.41, p < 0.001).

Figure 1.

Scatterplots of chronological vs. epigenetic age by Horvath’s, Hannum’s, and PhenoAge clocks. The diagonal green line corresponds to the predicted age equal to the chronological age, and the blue straight lines correspond to the linear regressions.

Table 1 shows no significant differences between the non-specific cLBP and PFC groups regarding sex, race, and household income (p > 0.05). The non-specific cLBP group was significantly older in chronological age than the PFC group, 46.14 (SD = 13.39) vs. 39.94 (SD = 14.52). We did not deem this difference to be clinically significant and controlled for chronological age in partial correlation analyses. The cLBP group had higher average DunedinPACE scores and lower Horvath and Hannum accelerated epigenetic age scores (i.e., less epigenetic age acceleration) than the PFC group. The cLBP and PFC groups had comparable PhenoAge accelerated epigenetic ages. Notably, the DunedinPACE score had a moderate effect size (Cohen’s d = 0.78), suggesting a moderate to strong difference between cLBP and PFC groups.

Table 1:

Distribution of baseline covariates among cases of nonspecific chronic low back pain (cLBP) and pain free controls (PFCs)

Covariate	cLBP (n = 69)	PFC (n = 49)	p-Values	Cohen’s d	95% CI
Sex (n, %)			0.852
Male	34 (49.3)	25 (51)
Female	35 (50.7)	24 (49)
Race (n, %)			0.078
Non-Hispanic Black	45 (65.2)	24 (49)
Non-Hispanic White	24 (34.8)	25 (51)
Income (n, %)			0.056
< 24,999	26 (38.2)	6 (13)
25k to 49,999	18 (26.5)	16 (34.8)
50k to 74,999	10 (14.7)	11 (23.9)
75k to 99,999	6 (8.8)	4 (8.7)
> 100,000	8 (11.8)	9 (19.6)
Chronological age (mean, SD)	46.14 (13.39)	39.94 (14.52)	0.018	0.45	0.08, 0.82
DunedinPACE (mean, SD)	1.10 (0.15)	0.99 (0.11)	< 0.001	0.78	0.40, 1.16
Accelerated Epigenetic (mean, SD)
ΔHorvath	8.53 (7.66)	11.06 (4.61)	0.04	−0.38	−0.7, −0.01
ΔHannum	6.60 (9.82)	8.45 (4.90)	0.023	−0.23	−.059, −0.14
ΔPhenoAge	~6.62 (8.14)	~6.66 (4.41)	0.97	0.007	−0.36, 0.37

Notes: SD = standard deviation; ΔHorvath = difference between epigenetic age by Horvath’s and chronological age; ΔHannum = difference between epigenetic age by Hannum’s and chronological age; ΔPhenoAge = difference between phenotypic epigenetic age and chronological age; DunedinPACE = Pace of biological aging by Dunedin’s;

Spearman’s Rho Bivariate Correlations Between Variables

As seen in Table 2, for the total sample (cLBP and PFC combined), the DunedinPACE scores were significantly correlated with race (r_s = −0.364, p < 0.01), income (r_s = −0.328, p < 0.01), BMI (r_s = 0.513, p < 0.01), and chronological age (r_s = 0.409, p < 0.01). Horvath epigenetic age acceleration, DunedinPACE score, and chronological age variables significantly correlated with pain severity and interference. Interestingly, pain severity and interference negatively correlated with Horvath’s epigenetic age acceleration but positively correlated with chronological age and DunedinPACE scores. Also, Hannum and PhenoAge epigenetic age acceleration scores did not significantly correlate with pain outcomes.

Table 2.

Spearman Rho’s correlations among outcomes and covariates

	Pain status	Sex	Race	Income	BMI	Horvath EAA	Hannum EAA	PhenoAge EAA	Chrono Age	Dunedin PACE	Pain Severity
Pain status	1
Sex	−0.017	1
Race	0.162	0.017	1
Income	.234*	0.003	.229*	1
BMI	−.237*	0.081	−.332**	−0.158	1
Horvath EAA	0.181	0.007	0.074	−0.005	−0.048	1
Hannum EAA	0.034	0.028	.210*	0.048	−0.087	.674**	1
PhenoAge EAA	−0.035	0.1	0.153	−0.051	0.071	.420**	.417**	1
Chrono Age	−.221*	−0.056	−0.084	0.049	0.106	−.565**	−.594**	−.207*	1
Dunedin PACE	−.359**	0.083	−.364**	−.328**	.513**	−.264**	−.240**	0.119	.409**	1
Pain Severity	−.849**	0.049	−.247**	−.377**	.257**	−.232*	−0.088	−0.053	.298**	.453**	1
Pain Interference	−.819**	0.065	−.193*	−.342**	.302**	−.265**	−0.13	−0.002	.287**	.476**	.885**

^*Correlation is significant at the 0.05 level

^**Correlation is significant at the 0.01 level

BMI = body mass index; EAA = Epigenetic age acceleration measured as the difference between epigenetic age and chronological age.

Predicting Pain Status from Pace of Biological Aging

Binary logistic regression models were used to predict the outcome of non-specific cLBP versus PFC group, using DunedinPACE scores. Odd ratios of having cLBP per unit change in the pace of biological aging, modeled as a continuous variable, are summarized in Table 3. A unit increase in DunedinPACE score was associated with 265 times higher odds of having cLBP compared to the PFC group (95% CI: 12.97, 5455.23). This model was statistically significant (p < 0.001). In multiple logistic regression models adjusted for sex, race, and BMI, the odds of having cLBP per unit increase in DunedinPACE scores were 149 times compared to PFC groups (95% CI: 4.71, 4751.99). This adjusted model was statistically significant (p = 0.005).

Table 3.

Binary Logistic regression predicting cLBP by epigenetic age acceleration and biological aging

	Model 1		Model 2		Model 3		Model 4
Predictor	OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
DunedinPACE	265.98	12.97, 5455.23	271.43	13.06, 5639.25	232.63	9.52, 5686.25	149.62	4.71, 4751.99
ΔHorvath	0.93	0.86, 0.99	0.93	0.87, 0.99	0.93	0.87, 1.001	0.93	0.86, 0.99
ΔHannum	0.96	0.91, 1.02	0.96	0.91, 1.02	0.97	0.92, 1.03	0.97	0.91, 1.03
ΔPhenoAge	1	0.95, 1.05	1	0.95, 1.05	1.008	0.95, 1.06	1.004	0.95, 1.06

Notes: ΔHorvath = difference between epigenetic age by Horvath’s and chronological age; ΔHannum = difference between epigenetic age by Hannum’s and chronological age; ΔPhenoAge = difference between phenotypic epigenetic age and chronological age; DunedinPACE = Pace of biological aging by Dunedin’s; OR = odds ratio; Boots CI = Boots strap confidence interval; Model 1: unadjusted; Model 2: adjusted for sex; Model 3: adjusted for sex and race; Model 4: adjusted for sex, race, and BMI.

For comparative analysis, we also predicted the outcome of cLBP versus the PFC group using Horvath, Hannum, and PhenoAge epigenetic age acceleration. Odd ratios of having cLBP compared to the PFC group per unit increase in epigenetic age acceleration measures were 0.93 (95% CI: 0.87, 0.99) for Horvath’s; 0.96 (95% CI: 0.91, 1.02) for Hannum’s; and 1 (95% CI: 0.95, 1.05) for PhenoAge clocks. In multiple logistic regression models adjusted for sex, race, and BMI, the odd ratios of having cLBP per year increase in epigenetic age acceleration were 0.93 (95% CI: −0.86, 0.99) for Horvath’s; 0.97 (95% CI: 0.91, 1.03) for Hannum’s; and 1.004 (95% CI: 0.95, 1.06) for PhenoAge. Of note, only the binary and multiple logistic regression models using accelerated epigenetic age from Horvath’s clock were statistically significant. Epigenetic age acceleration from Horvath, Hannum, and PhenoAge clocks showed slightly lower odd ratios for cLBP compared to the PFC group.

Associations of Biological Aging and Pain Outcomes

For participants with cLBP, DunedinPACE scores positively correlated with pain severity (r_s = 0.385, p = 0.001) and interference (r_s = 0.338, p = 0.005) in daily living (Figure 2). Partial correlations accounting for chronological age, sex, and race also suggested that a faster pace of biological aging (higher DunedinPACE scores) was significantly associated with greater pain severity and interference, r_s = 0.247, p < 0.046 and r_s = 0.292, p < 0.017, respectively. Since accelerated epigenetic aging using Horvath, Hannum, and PhenoAge clocks, did not significantly predict non-specific cLBP, we did not examine their correlation with pain severity and interference.

Figure 2.

Correlations of pace of biological aging with pain severity and interference in adults with non-specific chronic low back pain. Pain severity and interference assessed using the Brief Pain Inventory.

Discussion

Aging increases the risk for chronic pain, and most non-specific cLBP cases occur later in life.⁶ The potential to capture the individual dynamics that define the risk of non-specific cLBP attributable to biological aging is of significant interest to pain researchers and clinicians alike. To our knowledge, this is the first study to report a significant association between non-specific cLBP in adults and the pace of biological aging, as measured by DunedinPACE. Our findings suggest that the odds of faster biological aging are about 149 times in adults with nonspecific cLBP compared to PFCs. The average DunedinPACE score reflected a pace of biological aging that is faster for adults with cLBP than PFC. Furthermore, we observed that biological aging has a larger effect size (more meaningful or significant differences) than chronological age in terms of the differences between adults with non-specific cLBP and PFCs. These findings suggest that the pace of biological aging may be more relevant than chronological age in defining non-specific cLBP as an age-related chronic condition.

Our observation of an accelerated pace of biological aging in individuals with cLBP aligns with previous research showing that chronic musculoskeletal pain conditions are associated with accelerated epigenetic aging.^14,15 The strong association between DunedinPACE scores and cLBP status suggests that cLBP may be associated with greater physiological decline. Thus, the current study provides further evidence that non-specific cLBP is not only associated with localized musculoskeletal condition but may also have broader systemic implications, influencing the overall pace of aging in affected individuals.

One potential explanation for the observed association between cLBP and the faster pace of biological aging is the epigenetically induced changes in gene expression and physiological processes. Emerging evidence suggest that differential methylation of genes in key nociceptive and neuroinflammatory pathways are associated with cLBP.^37,38 Inflammation is an essential driver of the aging process, and proinflammatory cytokines are elevated in individuals with cLBP.^39,40 It is possible that persistent stress and chronic inflammation associated with cLBP contribute to the accumulation of DNA damage and cell senescence, resulting in accelerated aging at the molecular level.⁴¹ However, it remains unclear whether an accelerated pace of aging results in non-specific cLBP or non-specific cLBP causes accelerated aging. Longitudinal studies are needed to clarify the exact mechanism underlying the relationship between non-specific cLBP and the pace of biological aging.

Among adults with non-specific cLBP, we found that DunedinPACE scores positively correlated with pain severity and interference. While these findings should not be confused with causality, they suggest that the subjective experience of pain may be associated with an accelerated aging process. Chronic pain and chronic stress have been described as two sides of the same coin,⁴² and nonspecific cLBP has been associated with epigenetic changes in stress dysregulation pathways.²⁵ It is possible that the experience of living with cLBP and associated stress accelerates the aging process by promoting oxidative stress, cell senescence, and DNA damage, which results in a faster physiological decline.

Despite the robust relationship between DunedinPACE and cLBP, we did not observe a significant relationship between accelerated epigenetic age from Horvath, Hannum, or PhenoAge clocks and pain severity or interference. Also, our comparative analyses revealed that epigenetic ages derived from DNAm with Horvath’s, Hannum’s, and PhenoAge were closely related to the chronological age. Still, the epigenetic age acceleration from these clocks were not significant predictors of non-specific cLBP in adjusted regression models. This discrepancy between DNAm clocks supports the idea that the clocks may be measuring different constructs,^11,22 and the DunedinPACE is a more robust measure of biological aging.²² Our findings support the observation that first and second-generation epigenetic clocks may not be good predictors of age-related diseases, functional decline, or disease risk.¹³ Although DunedinPACE has not been extensively validated, especially in chronic pain, it offers several advantages over prior epigenetic clocks in capturing age-related physiological changes. Thus, it has the potential to provide insights into the underlying mechanisms of the relationship between non-specific cLBP and accelerated aging processes. Future studies should explore these potential mediators to understand the intricate interplay between pain and biological aging.

We acknowledge that this study has some limitations. First, we used a cross-sectional design that included DNAm data from a single time point. This design precludes a test of how changes in cLBP trajectory might affect DNAm and the pace of biological aging. Ultimately, longitudinal repeated-measures studies, preferably in a natural setting, will be needed to establish causality. Second, despite being the first study to examine the relationship between DunedinPACE and cLBP, our sample size was relatively small. Using bootstrapping for regression analysis increased the statistical power of our analyses. Third, our study did not include other epigenetic clocks (e.g., GrimAge), epigenetic modifiers (e.g., histone modifications), and gene expression or physiological biomarkers (e.g., proteins). Previous studies by Cruz-Almeida were specifically interested in predicting mortality because they were focused in middle-to-older aged adults (on average around 60 years of age), which maybe why they focused on DNAGrimAge,^15,16 but the present study included individuals as young as 18 years of age (average age 30–40s), thus mortality is not as relevant. Fourth, we estimated DNA concentration and purity using NanoDrop Spectrophotometry. While this approach safe, reliable, and commonly used,⁴³ it cannot differentiate DNA from RNA and protein which also absorb at 260 nm, requiring a method that is fluorometric-based. Finally, the biological underpinning of DunedinPACE and the epigenetic clocks remain to be well validated. Epigenetic medications are at the core of biological aging, yet there is significant variability in the predictive ability of the various clocks.

Within the bounds of these limitations, our findings have implications for clinical practice and research. Our results suggest that more extensive studies are needed to examine the relationship between biological aging and non-specific cLBP. At the same time, our results contribute evidence towards proof of concept for the hypothesis that cLBP is associated with accelerated biological aging. Thus, interventions that reverse or slow the aging process may also be effective in pain management.

Conclusion

Nonspecific cLBP is a significant public health problem with an unknown cause. Despite being labeled as an age-related chronic condition, this is the first study to demonstrate that the odds of an accelerated pace of biological aging as significantly higher in adults with cLBP compared to PFCs. Also, the pace of biological aging, as measured with DunedinPACE, correlates with pain severity and interference. Longitudinal studies are needed to elucidate further the causal mechanisms of the correlation between biological aging and cLBP.

Highlights.

• A cross-sectional study of adults with non-specific chronic low back pain.

• The pace of biological aging predicts cLBP better than chronological age.

• DunedinPACE predicts cLBP better than epigenetic age acceleration from other DNA methylation clocks that predict phenotypic age.

• The pace of biological aging correlates with pain severity and interference.

Perspectives:

Accelerated epigenetic aging is common among adults with non-specific chronic low back pain (cLBP). Higher DunedinPACE scores positively correlate with pain severity and interference, and better predict cLBP than other DNA methylation clocks. Interventions to slow the pace of biological aging may be viable targets for improving pain outcomes.