Introduction
As populations age, there are growing concerns about age‐related health conditions such as frailty, which is defined as a decline in function in many physiological systems, accompanied by increased risk of adverse outcomes due to stressors [1]. Given its clinical relevance, there is a pressing need to identify the biological mechanisms underlying frailty in older adults.
Circadian rhythms allow organisms to adapt their behavior and physiology to 24‐h environmental cycles. These endogenous oscillations play a central role in the time‐of‐day‐dependent regulation of multiple processes, including sleep–wake cycles, metabolic homeostasis, and immune function [2]. At the molecular level, the circadian clock is sustained by negative feedback loops of transcription and translation of core “clock genes”.
The transcription factor heterodimer BMAL1/CLOCK promotes the expression of PERIOD (PER1/2/3) and CRYPTOCHROME (CRY1/2) genes, whose protein products translocate back to the nucleus to inhibit BMAL1/CLOCK activity. Additional regulatory loops involving nuclear receptors such as REV‐ERBs and RORs contribute to the robustness of this molecular system [3]. The molecular clock is expressed throughout the organism, including peripheral tissues.
Chronotype refers to the individual differences in preferred timing of sleep and wakefulness. It is considered an output of the endogenous circadian system and has been shown to shift toward a morning preference with advancing age [4]. However, a systematic characterization of chronotype in the Mexican older adult population has not yet been reported.
There is a bidirectional relationship between circadian rhythms and health, whereby disruptions of circadian organization (e.g., shift work) adversely affect multiple health domains, while several health conditions can, in turn, disturb daily rhythms [5]. Moreover, aging itself is associated with alterations in circadian rhythms, including sleep timing, maintenance, and quality, which may impair physiological recovery from stressors [6]. In Mexican older adults, the prevalence of insomnia has been estimated to range between 30% and 36% [7, 8].
Notably, insomnia and poor sleep quality have been associated with an increased risk of frailty, particularly among older women [8].
Recent studies have described the relationship between circadian disruption and frailty. A longitudinal study showed that lower daily rhythm strength and stability, as determined through actigraphy, are associated with a higher risk of developing frailty and a faster progression of this syndrome [9]. Furthermore, a bidirectional relationship was described between frailty and sleep disturbances using data from genome‐wide association studies [10].
Given the importance of circadian regulation for health and aging, a better understanding of the interplay between daily rhythms, metabolic health, and frailty may contribute to strategies aimed at improving quality of life in older adults. Therefore, the aim of this study was to investigate the association between circadian biology and frailty in older adults by integrating chronotype, insomnia, metabolic parameters, and peripheral clock gene expression. Specifically, we examined (i) the association between chronotype, frailty, and other health‐related variables, including insomnia; (ii) the relationship between chronotype and metabolic parameters; and (iii) the association between PBMC clock gene expression and frailty status.
Methods
Study Design and Cohort Description
This was a cross‐sectional study from the 6th round of the Cohort of Obesity, Sarcopenia and Frailty of Older Mexican Adults (COSFAMM). COSFAMM is a prospective study that started in 2014 (described in [11]), where beneficiaries of the Mexican Social Security Institute (IMSS in Spanish) living in Mexico City, aged 60 or more, were randomly selected and invited to participate. Yearly personal interviews and geriatric assessments are performed by trained health workers.
In 2022, 751 participants were included in round six of COSFAMM. The study was approved by the IMSS National Commission for Scientific Research and Ethics Committee (CONBIOETICA‐09‐CEI‐009‐20,160,601) with registration number R‐2023‐785‐073. All participants were informed of the procedures and use of the data collected in the study and signed an informed consent form.
Clinical and Sociodemographic Data Collection
Trained healthcare professionals collected and digitally recorded questionnaire data from participants during a structured interview, including demographic characteristics, health and anthropometric information. In 2022, 751 participants of COSFAMM were interviewed, 748 completed all questionnaires, and 663 participants consented to a blood draw. Abdominal obesity was defined as a waist circumference > 88 cm for women and > 102 cm for men.
Physical activity level was categorized by 748 participants as “none”, “low” (15–120 min weekly), “moderate” (3–4 h weekly), and “high” (5 or more hours weekly). Reporting of two or more chronic illnesses was registered as comorbidity, and polypharmacy was defined as taking 5 or more medications.
Chronotype Assessment
Chronotype was evaluated using the Spanish language version of the MEQ [12], completed by 750 participants. The MEQ consists of 19 items and a possible score between 16 and 86. Participants were classified as either extreme evening (16–30), moderate evening (31–41), intermediate (42–58), moderate morning (59–69), or extreme morning (70–86).
The Cronbach's alpha for the MEQ in this population was 0.634.
Insomnia Assessment
Insomnia was evaluated with the Athens Insomnia Scale (AIS) [13], which consists of 8 questions about sleep difficulties, sleep quality, and the impact of insomnia during the day. Participants rate items on a scale from 0 to 3 according to the severity of the aspect of insomnia, and a cutoff score of 6 was considered to classify participants as presenting with insomnia [14].
Frailty Assessment
Frailty status was determined using the five criteria established by Fried [15]: Exhaustion, low grip strength (adjusted by sex and body mass index), involuntary weight loss, slow walking speed (time to walk 4.5 m, adjusted for sex and height), and low physical activity. One point is assigned for each criterion; participants with 0 points are considered robust, with 1–2 points considered prefrail, and with ≥ 3 points considered frail.
Metabolic Parameters and Serologic Exam
Serum levels of fasting blood glucose, creatinine, total cholesterol, HDL cholesterol, triglycerides, albumin, aspartate aminotransferase (AST), and alanine transaminase (ALT) were measured. From 663 participants who consented to a blood draw, a complete panel was obtained from 635. Body mass index (BMI) was estimated by dividing weight in kilograms by the square of height in meters.
Body fat percentage was measured by bioelectrical impedance.
PBMCs Isolation
Blood samples were drawn from COSFAMM participants at 8–10 am. For a subset of 82 participants, 4 mL of total blood supplemented with heparin was mixed with 4 mL of phosphate‐buffered saline. The cell suspension was layered onto 2 mL of Ficoll and centrifuged at 500 × g for 40 min.
The white blood cell fraction was then resuspended in PBS and washed twice at 400 × g for 15 min. The pellet was resuspended in TRIzol reagent (ABP Biosciences).
RNA Isolation
Total RNA was isolated from PBMCs with the TRIzol method following the manufacturer's instructions. RNA was quantified using a NanoDrop instrument (Thermo Fisher Scientific), and 500 ng of cDNA was synthesized with the GoScript Reverse Transcriptase kit (Promega).
RT‐qPCR. RT‐qPCR was performed using the Chai Green qPCR Master mix (Chai), in a CFX96 Real‐Time PCR Detection System (Bio‐Rad), using primers with the following sequences: ACTB Fwd GAAGATCAAGATCATTGCTCCTC, ACTB Rev. ATCCACATCTGCTGGAAGG, BMAL1 Fwd TCCAATGTGGAATCCTGGGC, BMAL1 Rev.
CACACAGGAAGCCCTCTAGC, CLOCK Fwd TCTGCCTCTTCTCGGAGTTC, CLOCK Rev. GTATCCGTCGGGATCTTGGT, DBP Fwd CTGGCCACGAGACCTTTGAC, DBP Rev. GGCTCCAGTATTTCTCATCCTTC, CRY1 Fwd CTGCGTCTACATCCTGGACC, CRY1 Rev.
ACTGAAGCAAAAATCGCCACC, PER1 Fwd TCTGTGCTGAAGCAGGATCG, PER1 Rev. GCTGGTCCTCAGAAAACCGA, PER2 Fwd CAGCCAAGTGAACGAACTGC, PER2 Rev. CCGGTGCTCTCAGATGAGTC, PER3 Fwd ACAGCAGCCACAGTGAACAG, PER3 Rev.
CGTCTCTCCGAGGGGAAGTA, REVERBA Fwd GAAGATCAAGATCATTGCTCCTC, REVERBA Rev. ATCCACATCTGCTGGAAGG. Ct values were normalized to Actin expression with the ΔΔCt method to the Control group, which included participants classified as Robust.
Statistical Analyses
For descriptive data, continuous variables are presented as mean ± SEM, and for categorical variables, proportion of the total (%). Differences between frailty and insomnia groups were compared using the chi‐square and Fisher's exact test, as appropriate. Associations between chronotype and insomnia were examined using binary logistic regression models, consolidating chronotype categories into 3: Intermediate, moderate morning, and extreme morning.
Both crude and adjusted models were performed, adjusting for sociodemographic and health‐related covariates including age, sex, physical activity, BMI, frailty status, insomnia, and comorbidity burden. Multiple linear regression analyses were conducted to evaluate associations between chronotype (MEQ) score and metabolic parameters, including glucose, creatinine, total cholesterol, HDL cholesterol, triglycerides, albumin, AST, ALT, BMI, and body fat percentage. Both crude and adjusted models were assessed.
Because multiple metabolic outcomes were analyzed, p‐values from adjusted models were corrected for multiple comparisons using the Benjamini–Hochberg false discovery rate (BH‐FDR) method. For gene expression analyses, a PCA was performed using centered and scaled relative expression values of eight clock genes. The first two principal components (PC1 and PC2) explained approximately 80% of the total variance and were retained for downstream analyses.
Gene loadings were examined to identify the contribution of individual genes to each component. Differences in PC1 scores across frailty groups were evaluated using one‐way analysis of variance (ANOVA), followed by Tukey's post hoc tests for pairwise comparisons. Additionally, gene‐specific linear regression models were performed to evaluate associations between frailty and individual clock gene expression levels in both crude and adjusted models, controlling for age, sex, BMI, insomnia, and chronotype score.
Multiple‐testing correction for gene expression analyses was performed using the BH‐FDR method. Statistical analyses were performed using SPSS, R (v3.5.2), and RStudio (v12.0), including packages for data visualization, multivariate analysis, and regression modeling. Statistical significance tests were two‐sided and considered at p < 0.05.
Results
Demographic Data
For this study, a total of 751 participants were included: 362 men (48.2%) and 389 women (51.8%) between the ages of 62 and 97 (average 69.89 ± 0.3 for women and 69.8 ± 0.3 for men). Marital status was “Partnered” for 84.2% of the men and 75.6% of the women, and employment status was “Active” for 20.8% of the men and 28.3% of the women.
Insomnia by Chronotype, Frailty and Health Variables
The median score for the MEQ was 62 points for men and 64 for women. Interestingly, no participants were classified as having an “extreme evening” chronotype, and only one man and one woman self‐reported as “moderate evening”. Also, 83 men (23%) and 116 women (29.8%) had an “intermediate” chronotype, while 228 men (63.2%) and 226 women (58.1%) were classified as “moderate morning”, and finally, 49 men (13.6%) and 95 women (11.8%) were identified as having an “extreme morning” chronotype (Table 1).
For frailty, only 31 men (8.6%) and 24 women (6.2%) were considered “robust”, while 272 men (75.1%) and 277 women (71.2%) were categorized as “prefrail”. Finally, 59 men (16.3%) and 88 women (22.6%) met the criteria for “frailty”.
Next, we analyzed the association between chronotype, frailty, and other health variables and insomnia (Table 1). We observed an effect of chronotype and frailty on the proportion of women with insomnia. For chronotype (p = 0.017), women with an intermediate chronotype were more likely to have insomnia.
Interestingly, for the frailty phenotype (p = 0.019), prefrail women were less likely to have insomnia, and frail women were more likely to report this sleep difficulty. To further investigate the relationship between chronotype and metabolic health, multivariable linear regression analyses were performed using MEQ score and metabolic parameters, including glucose, creatinine, total cholesterol, HDL cholesterol, triglycerides, albumin, AST, ALT, BMI, and body fat percentage. Although some nominal associations were observed in unadjusted analyses, none remained statistically significant after adjustment for covariates and correction for multiple comparisons (Table S1).
To further examine the association between insomnia and chronotype, we performed binary logistic regression analyses; however, due to the low frequency of evening chronotypes in this older adult cohort, categories were consolidated into three groups: Intermediate, moderate morning, and extreme morning. We found that individuals with an extreme morning chronotype had lower odds of insomnia compared with those with an intermediate chronotype (OR = 0.391, 95% CI 0.172–0.808, p = 0.016). After adjusting for age, sex, BMI, abdominal obesity, frailty status, physical activity level, and comorbidities, the association between extreme morning chronotype and insomnia remained statistically significant (OR = 0.393, 95% CI 0.169–0.827, p = 0.020).
Consistent with previous analyses (Table 1), women independently showed higher odds of reporting insomnia (OR = 1.922, 95% CI 1.270–2.949, p = 0.002).
Association Between Clock Gene Expression and Frailty Status
Given the central role of regulation by the molecular circadian clock in aging‐related processes, we next explored the association between clock gene expression and frailty status in this older adult cohort. PCA was used to summarize the expression of 8 circadian clock genes at a morning time point in PBMCs of 82 participants. PC1 and PC2 explained a total of 79.9% of the variance, with PC1 accounting for 67.1% (SD 2.32), and PC2 an additional 12.8% (SD 1.01).
All genes showed a positive loading in PC1 (Table S2), indicating a coordinated pattern of overall expression of circadian clock genes. In contrast, PC2 was characterized by opposing loadings of two subsets of genes within the molecular clock: Strong, positive loading for DBP and CLOCK, along with PER2 and CRY1, contrasting with strong negative loading by BMAL1, PER1, and PER2 (Table S2). A PCA biplot of circadian gene expression displaying frailty status suggests a partial separation of participants, where individuals classified as Robust tend to cluster toward higher PC1 scores.
Moreover, prefrail and frail older adults tend to cluster around lower PC1 values (Figure 1). To further evaluate differences in global clock gene expression patterns across frailty groups, PC1 scores were compared using one‐way ANOVA. PC1 scores differed significantly according to frailty status [(F2,79) = 3.39, p = 0.039]; specifically, frail (p = 0.039) and prefrail (p = 0.040) participants exhibited significantly lower PC1 scores compared to robust individuals (Figure 2A).
Based on PCA results, genes with the highest absolute loadings on PC1 were subsequently analyzed individually using multivariable linear regression models (Table S3). In unadjusted linear regression models, both prefrailty and frailty were associated with lower expression of CLOCK, CRY1, DBP, and PER3. After adjustment for insomnia status, chronotype score, age, sex, and BMI, lower CLOCK expression remained significantly associated with prefrailty (Figure 2B and Table S3).
Discussion
In this study, we investigated the relationship between chronotype, insomnia, frailty, metabolic health, and peripheral clock gene expression in a large cohort of community‐dwelling older adults. Our findings indicate that morning chronotype was associated with lower odds of insomnia, while frailty status was associated with differences in the expression of several clock genes measured in peripheral blood mononuclear cells. Together, these findings support a link between circadian‐related characteristics and health outcomes in older adults.
Consistent with previous reports indicating a shift toward morningness with aging [4, 16], the majority of participants in this cohort were classified as moderate or extreme morning types, with virtually no representation of evening chronotypes. While the distribution of chronotype in this cohort limits comparisons across the full morningness–eveningness spectrum, it provides an opportunity to examine whether variability within predominantly morning‐oriented older adults remains associated with sleep health outcomes. We observed a significant association between chronotype and insomnia, specifically among women.
Individuals with an extreme morning chronotype showed lower odds of reporting insomnia compared with those with an intermediate chronotype. These findings are consistent with prior evidence linking earlier sleep timing with measures of higher sleep quality, such as more time in Rapid Eye Movement (REM) sleep and shorter REM onset [17, 18]. Notably, female sex emerged as an independent predictor of insomnia, in agreement with previous studies reporting a higher prevalence of sleep disturbances among Mexican older women [19].
Additionally, the observed interaction between frailty status and insomnia in women supports prior observations [8, 20], indicating that sleep disturbances and frailty may reinforce each other.
In contrast, we found limited evidence supporting an association between chronotype and metabolic health. Although several nominal associations were observed in unadjusted analyses, none remained significant after adjustment for covariates and correction for multiple testing. This may indicate that chronotype is not a major determinant of metabolic variability in this cohort of older adults.
Alternatively, the predominance of morning chronotypes and the near absence of evening‐type individuals may have reduced our ability to detect associations previously reported in populations with a broader chronotype distribution [21, 22].
A second major finding of this study was the association between frailty status and peripheral clock gene expression. Principal component analysis identified a coordinated pattern of expression across eight clock genes, with robust participants exhibiting higher PC1 scores than both pre‐frail and frail individuals. Notably, pre‐frail and frail participants did not differ from one another, suggesting that differences in clock gene expression may be represented regardless of the progression of frailty.
Consistent with the PCA results, gene‐specific analyses demonstrated lower expression of CLOCK, CRY1, DBP, and PER3 among pre‐frail and frail participants in unadjusted models, while lower CLOCK expression remained significantly associated with pre‐frailty after adjustment for chronotype, insomnia, age, sex, and BMI. Clock gene expression was measured at a single morning time point, and therefore the present study cannot assess circadian rhythmicity, phase, amplitude, synchronization, or rhythm robustness. Consequently, the observed differences should be interpreted as differences in morning expression levels and coordinated expression patterns of peripheral clock genes, rather than direct evidence of circadian dysfunction or altered circadian rhythmicity.
Nevertheless, peripheral clock gene expression provides a useful molecular readout of clock‐related biological processes and has been widely used to investigate links between circadian biology and human health [23, 24, 25, 26], which is especially valuable in studies with community‐dwelling older adults, where sample collection opportunities are limited, and minimally invasive methods are required [27].
The biological mechanisms underlying the association between frailty and peripheral clock gene expression remain to be determined. Aging is accompanied by changes in molecular clock function across multiple tissues [28], such as reduced amplitude in circadian temperature and melatonin secretion rhythms [29, 30]. Alterations in clock gene expression have been linked to inflammatory signaling [31], metabolic regulation [32], and cellular stress responses [33].
Frailty is similarly characterized by multisystem physiological dysregulation involving metabolic, immune, and musculoskeletal pathways [34]. It is therefore plausible that the lower expression of clock genes observed in frail individuals reflects biological processes associated with aging and physiological vulnerability. However, longitudinal studies incorporating repeated sampling across the day will be required to determine whether these expression differences correspond to alterations in circadian timing or rhythm characteristics.
Taken together, our findings indicate that chronotype, sleep health, and frailty are associated with distinct aspects of circadian‐related biology in older adults. While morning chronotype was associated with lower odds of insomnia, frailty was associated with lower morning expression of several peripheral clock genes and with differences in global gene expression patterns summarized by PCA. Future studies combining molecular, behavioral, and physiological circadian measures will be necessary to clarify the mechanisms linking circadian biology and frailty during aging.
Strengths and Limitations
Strengths of this study include the large, well‐characterized cohort, the integration of multiple circadian‐related measures, and the combined use of behavioral, metabolic, and molecular data. However, limitations should be acknowledged. The predominance of morning chronotypes in this older adult cohort limits this study's ability to evaluate differences between morning and evening chronotypes.
Therefore, the observed associations should be interpreted as differences across the spectrum of morning preference rather than as comparisons between morning and evening preference. The cross‐sectional design precludes causal inference between frailty, clock gene expression, and insomnia. While differences in chronotype, insomnia, and peripheral clock gene expression were associated with frailty status, it is plausible that frailty‐related physiological changes influence sleep characteristics, metabolic regulation, and molecular gene expression patterns.
Additionally, clock gene expression was assessed at a single morning time point, which does not capture circadian dynamics across the full 24‐h cycle.
Author Contributions
C.E.O.‐V. contributed to experimental procedures and data analysis. P.G‐dlT. contributed to conceptualization, funding acquisition, and writing (reviewing and editing). S.S.‐G. contributed to conceptualization, funding acquisition, and writing (reviewing and editing).
L.A.‐A. contributed to resources and writing (reviewing and editing). L.M.‐V. contributed to conceptualization, funding acquisition, data analysis, and writing (original draft).
Funding
This work was supported by the Alzheimer's Association/National Academy of Neuropsychology grant ALZ‐NAN‐22‐941933 to P.G‐dlT. Additional funds were provided by the 2019 Mexican L'Oréal‐UNESCO‐AMC Programme for Women in Science to L.M.‐V.
Ethics Statement
The study was approved by the IMSS National Commission for Scientific Research and Ethics Committee (CONBIOETICA‐09‐CEI‐009‐20,160,601) with registration number R‐2023‐785‐073. All participants were informed of the procedures and use of the data collected in the study and signed an informed consent form.
Conflicts of Interest
The authors declare no conflicts of interest.
Supporting information