Cardiorespiratory fitness and mitochondrial oxidative capacity are associated with reduced walking speed in older adults, but their impact on walking speed in older adults with diabetes has not been clearly defined. We examined differences in cardiorespiratory fitness and skeletal muscle mitochondrial oxidative capacity between older adults with and without diabetes, as well as determined their relative contribution to slower walking speed in older adults with diabetes. Participants with diabetes (n = 159) had lower cardiorespiratory fitness and mitochondrial respiration in permeabilized fiber bundles compared with those without diabetes (n = 717), following adjustments for covariates including BMI, chronic comorbid health conditions, and physical activity. Four-meter and 400-m walking speeds were slower in those with diabetes. Mitochondrial oxidative capacity alone or combined with cardiorespiratory fitness mediated ∼20–70% of the difference in walking speed between older adults with and without diabetes. Additional adjustments for BMI and comorbidities further explained the group differences in walking speed. Cardiorespiratory fitness and skeletal muscle mitochondrial oxidative capacity contribute to slower walking speeds in older adults with diabetes.
The contributors to slower walking speed in older adults with diabetes remain unclear.
This study was conducted to answer the question of how mitochondrial oxidative capacity and cardiorespiratory fitness impact walking speed in older adults with diabetes.
We found that mitochondrial oxidative capacity, cardiorespiratory fitness, and walking speed were lower in older adults with diabetes compared with those without diabetes.
In addition, mitochondrial oxidative capacity and cardiorespiratory fitness contributed to slower walking speed in those with diabetes.
Introduction
The U.S. population is rapidly aging, and in 2022, the percentage of Americans aged ≥65 years diagnosed with diabetes remained high at 30% (1). This represents a health care challenge as older adults with diabetes have a greater risk of cardiovascular disease and mobility disability (2,3). The contributors to impaired mobility in older individuals with diabetes include the presence of comorbidities, increased BMI (4), and lower muscle strength and quality (5–7). However, the relationship between diabetes and loss of mobility measured by 4-m and 400-m walking speeds remains only partially explained by muscle variables, such as quality and strength (5). Skeletal muscle mitochondrial oxidative capacity is an integral contributor to cardiorespiratory fitness, along with the ability of the cardiopulmonary system to supply oxygen to contracting muscles. Our group and others reported that cardiorespiratory fitness and skeletal muscle mitochondrial oxidative capacity are significantly associated and that each link to reduced walking speed in aging, which is a predictor of mobility disability (8–10). For example, mitochondrial oxidative capacity assessed using 31P-magnetic resonance spectroscopy (MRS) is highly correlated with walking speed and partially explains age-related poorer performance in short (6-m) and long (400-m) walking tasks (10). Several studies have found a close relationship between cardiorespiratory fitness and walking speed in older adults (11), which suggests that the decline in aerobic capacity also contributes to slower walking speed and mobility disability with age.
Muscle biopsies from individuals with diabetes have been shown to have lower oxidative phosphorylation (OXPHOS) capacity compared with muscle from healthy control individuals (12–14). However, some studies (15,16), but not all (12,13,17), indicated no differences between patients with diabetes and BMI-matched control individuals, indicating that obesity per se may underlie the lower muscle oxidative capacity. Indeed, other studies in people without diabetes have shown that BMI is strongly related to muscle oxidative capacity (18). In addition, reports assessing mitochondrial oxidative capacity in older adults with diabetes tend to have small sample sizes (n = 8–12) and lack adjustment for critical confounding variables, including physical activity and adiposity (13,14,17), and there are few reports that focused on older adults with diabetes specifically. Whether the presence of diabetes in a population of older adults further exacerbates declines in mitochondrial oxidative capacity and cardiorespiratory fitness remains unclear. Lower fitness is a predictor of insulin resistance and diabetes (19), and prevalence of diabetes is greater in individuals with low fitness (20). Skeletal muscle mitochondrial oxidative capacity is an integral contributor to cardiorespiratory fitness, along with pulmonary function (ventilation and alveolar-capillary gas exchange), cardiac stroke volume, oxygen transport and delivery (hemoglobin concentration, arteries and arterioles, capillary density, and diffusion capacity system) to supply oxygen to contracting muscles. While each of these factors could contribute to walking speed, it is unknown whether diabetes-associated changes in cardiorespiratory fitness and/or muscle energetics explain slower walking speed seen in older adults with diabetes, independent of other covariates.
In this analysis, we leveraged data from the Study of Muscle, Mobility and Aging (SOMMA) (21) to investigate the association of diabetes with cardiorespiratory fitness and mitochondrial oxidative capacity in a cohort of 879 older adults who were well phenotyped in terms of the free-living physical activity behaviors and body composition. We assessed mitochondrial oxidative capacity using two approaches: 31P-MRS to measure muscle in vivo (ATPmax) and high-resolution respirometry to measure mitochondrial respiration in muscle biopsies, a complementary approach that interrogates mitochondrial OXPHOS capacity at the myocellular level. Here, we tested the hypothesis that diabetes would be associated with lower cardiorespiratory fitness and muscle oxidative capacity in older adults independent of adiposity, physical activity levels, chronic comorbid health conditions, and oral hypoglycemic medication use. We also tested the extent to which cardiorespiratory fitness and mitochondrial respiration mediate slower walking speed of older adults with diabetes.
Research Design and Methods
Participant Recruitment
SOMMA (https://sommaonline.ucsf.edu) is a prospective longitudinal cohort study of older adults designed to understand the biological basis of muscle aging (21). Men and women (≥70 years old) were recruited between April 2019 and December 2021. Participants were enrolled at the University of Pittsburgh and Wake Forest University School of Medicine if they were willing and able to complete a skeletal muscle biopsy and undergo MRI and MRS. Exclusion criteria included an inability to walk one quarter of a mile or climb a flight of stairs, having a BMI >40 kg/m2, having an active malignancy or dementia, or having any medical contraindication to biopsy or magnetic resonance. Finally, participants must have been able to complete the 400-m walking test; those who appeared as they might not be able to complete the 400-m walk at the in-person screening visit completed a short-distance walk (4-m) to ensure their walking speed was ≥0.6 m/s. Individuals with diabetes were identified by self-report, use of prescribed hypoglycemic medication, or an HbA1c ≥6.5%. The WIRB-Copernicus Group institutional review board (study no. 20180764) approved the study, and all participants provided written informed consent.
Baseline Assessments
Details on study design and methodology have been published elsewhere (21). The SOMMA baseline visit generally consisted of 3 days of assessments over several weeks. On “Day 1,” the 400-m walk plus most other in-person assessments were performed. On “Day 2,” cardiopulmonary exercise testing (CPET), MRI, and MRS were assessed. On “Day 3,” a muscle biopsy specimen was collected. Age, sex, and race were collected by self-report. Self-reported medical conditions (heart disease, stroke, kidney disease/renal failure, and peripheral vascular disease) were assessed to determine whether participants had those conditions. Participants were asked to bring all prescription medications they had taken in the 30 days prior to their “Day 1” clinic visit. If a participant forgot to bring one or more medications, clinic staff obtained this information over the telephone or at the return visit. Prescription medication use was reviewed and updated at the additional baseline visit days where CPET and tissue sampling were done. Body weight, height, and physical function assessments were also measured, and accelerometry devices were provided. Most SOMMA participants completed CPET, MRS and provided a percutaneous biopsy of the vastus lateralis muscle, which was performed under fasted conditions. On the day of the muscle biopsy, participants prescribed oral hypoglycemic medication were asked to take their medications after tissue sampling.
Mobility Assessments
Mobility was assessed by 400-m and 4-m walking tests. For the 400-m walking test, participants were instructed to walk 10 complete laps around a set course at their usual pace and without overexerting themselves. The 4-m walking speed was measured in two trials, and the participant’s best time was used to calculate walking speed in m/s.
Physical Activity Assessments
Two devices were used to assess activity: the thigh-worn activPal that recorded total daily step count and the wrist-worn ActiGraph GT9X. Devices were placed on the participants at the baseline “Day 1” visit with a data collection period of 7 full days. Data were included only if the device was worn over a 24-h period and had ≥17 h/day wear time. Activity levels collected from the ActiGraph device were determined by cut points described by Montoye et al. (22), and included total daily physical activity time and time spent in sedentary physical activity, light physical activity, and moderate to vigorous physical activity (MVPA).
CPET
Participants walked for 5 min at a preferred walking speed, and a progressive symptom-limited exercise protocol ensued with increases in speed (0.5 mph) and incline (2.5%) in 2-min increments using a modified Balke protocol or a manual protocol (23,24). Cardiorespiratory fitness was determined as VO2peak (mL/min) which was identified as the highest 30-s average of VO2 (mL/min) achieved.
MRS and MRI
31P-MRS was used to assess in vivo mitochondrial adenosine triphosphate (ATP) generation in the quadriceps muscle during an acute bout of knee isometric extension exercise (21,24). A detailed description of the 31P-MRS protocol can be found in Mau et al. (24). A 3-T magnetic resonance magnet (MAGNETOM Prisma [Pittsburgh] or MAGNETOM Skyra [Wake Forest]; Siemens Healthineers) with a 12-cm 31P/1H dual-tuned, surface radiofrequency coil (PulseTeq) was placed over the quadriceps with the shell of the coil consistently placed 10 cm superior to the patella over the distal vastus lateralis muscle. Dynamic quantification of phosphocreatine (PCr), inorganic phosphate, phosphodiesterase, and ATP peak areas were acquired using 100-μs block pulse targeting a pulse angle of 45° with 2,048 points and a sweep width of 5,000 Hz. Seventy-five dynamic spectra were acquired, averaging four acquisitions with a 1.5-s recycle time each for a spectral time resolution of 6-s throughout the rest, exercise, and recovery periods. Data were immediately processed to determine PCr breakdown and acidosis and then used to adjust exercise duration for a second or third trial to satisfy breakdown and acidosis criteria. Raw data were used to quantify peaks representing phosphorus metabolites using jMRUI version 6.0 software. Temporal response of PCr area was used to calculate ATPmax (aka, Qmax) from PCr recovery rates (25). PCr area time course was referenced to the last 10 spectra of a 76-spectrum series. ATPmax was calculated using the recovery time constant as the inverse of the rate constant as described in Meyerspeer et al. (26) and assuming a resting/recovered PCr concentration of 24.5 mmol/L. Inorganic phosphate chemical shifts were followed relative to PCr = −2.54 parts/million to calculate intracellular pH to monitor muscle acidosis. Quality control criteria for inclusion of dynamic 31P-MRS data in analyses were 1) PCr breakdown between 33 and 50%, 2) pH throughout dynamic experiment >6.8, and 3) monoexponential fit of PCr recovery data. Participant data not meeting all criteria were excluded from the analysis.
An AMRA scanning protocol with the Dixon water-fat imaging method was used to assess abdominal subcutaneous adipose tissue (ASAT), visceral adipose tissue (VAT), the VAT/ASAT ratio, and quadricep muscle fat infiltration (QFI). The entire body was scanned, and images were analyzed using AMRA Researcher (AMARA Medical AB, Linköping Sweden) (27).
Muscle Biopsy and Mitochondrial Respiration
Muscle biopsy was completed with a Bergström trocar (5 or 6 mm) as previously described (24). Approximately 10 mg of muscle tissue was allocated for respirometry assays and were immediately submerged in ice cold BIOPS buffer. Fiber bundles of ∼2–3 mg wet weight were gently teased apart and permeabilized as previously described (24), and weight was measured using an analytical balance (Mettler Toledo, Columbus, OH). Permeabilized fiber bundle (PmFB) oxygen consumption was measured using a high-resolution respirometer (Oxygraph-2k; Oroboros Instruments, Innsbruck, Austria). PmFBs were placed into chambers containing MiR05 buffer supplemented with 25 μmol/L blebbistatin in the absence of light. Two protocols were completed in duplicate measuring carbohydrate-supported (protocol 1: 5 mmol/L pyruvate and 2 mmol/L malate) and fatty acid oxidation (FAO)–supported (protocol 2: 25 μmol/L palmitoyl-carnitine and 2 mmol/L malate) respiration. For protocol 1, state 3 respiration was stimulated with the addition of 4.2 mmol/L ADP (OXPHOSCHO) followed by subsequent additions of 10 mmol/L glutamate and 10 mmol/L succinate sequentially (maxOXPHOSCHO). Maximal uncoupled respiration was stimulated with consecutive 0.5 μmol/L titrations of 2-[2-[4-(trifluoromethoxy)phenyl]-hydrazinylidene]propanedinitrile until maximum electron transport system (ETS) capacity was attained (maxETSCHO). For protocol 2, state 3 respiration was stimulated with the addition of 4 mmol/L ADP (OXPHOSFAO) followed by subsequent additions of 10 mmol/L glutamate and 10 mmol/L succinate to further stimulate mitochondrial complexes I and II (maxOXPHOSFAO). PmFBs eliciting an oxygen consumption response to cytochrome c oxidase >15% were not included in the final analysis. Respiratory control ratios for protocol 1 (7.11 ± 2.86) and protocol 2 (3.47 ± 1.74) were assessed for quality control. Respiration was analyzed using DatLab 7.4.0.4 software. In SOMMA, not all participants completed both protocols because of time constraints. Protocol 1 (carbohydrates) was prioritized over protocol 2 (fatty acids) when limited technician time was available, which left fewer participants with complete data for protocol 2.
Statistical Analysis
Differences in participant characteristics between older adults with and without diabetes were compared using the t test for continuous variables (or Kruskal-Wallis nonparametric test for skewed variables) and χ2 test for dichotomized variables. Linear regressions without adjustment were performed to determine differences in cardiorespiratory fitness and mitochondrial oxidative capacity between those with and without diabetes, and additional adjustments were made for the following confounders: age, race, sex, site/technician, BMI, and chronic conditions (model 1). Model 1 was further adjusted for steps per day from activPal or volume of the adipose tissue depots ASAT, VAT, VAT/ASAT ratio, and QFI. Finally, to determine the impact of hypoglycemic medication use within the group of older adults with diabetes, multivariable linear regression was performed adjusting for age, race, and site/technician. Those analyses were completed using JMP version 16 software.
Linear regression was also used to determine the association of diabetes with walking speed. We evaluated the potential mediating influence of VO2peak and/or mitochondrial respiration. We did this by comparing the β-coefficient for the association of diabetes status on walking speed from the 400-m and 4-m walking tests. The base model included the following confounding variables: sex, age, and race. Model 2 included the base model plus technician and maxOXPHOSCHO; model 3 included the base model plus technician and maxOXPHOSFAO; model 4 included the base model plus clinical site and VO2peak; model 5 included the base model plus technician, VO2peak, and maxOXPHOSCHO; and model 6 included the base model plus technician, VO2peak, and maxOXPHOSFAO. Fully adjusted models also included the potential confounders/mediators BMI and chronic conditions. Finally, we compared the percent difference in the β-coefficient between the base model and subsequent models to understand how each adjustment impacted group differences in walking speed. Those analyses were performed using SAS 9.4 software (SAS Institute Inc., Cary, NC).
Data and Resource Availability
All SOMMA data are publicly available via a web portal. Updated data sets are released approximately every 6 months (https://sommaonline.ucsf.edu/).
Results
Participant Characteristics
A total of 879 participants (59.2% women) with an average age of 76.5 ± 5.0 years were enrolled in SOMMA (21). A summary of participants with available data is included in Supplementary Fig. 1. Participant characteristics are summarized in Table 1. Eighteen percent of the SOMMA cohort was classified as having diabetes. The proportion of women to men was greater in the group without diabetes (P = 0.02). The proportion of White participants compared with non-White participants was greater in the group without diabetes (P < 0.001). Older adults with diabetes had a higher body weight, waist circumference, BMI, and HbA1c (all P < 0.001). Participants with diabetes tended to have a higher prevalence of heart disease and hypertension, whereas those without diabetes more often had a history of cancer (P < 0.001). Approximately 75% of participants with diabetes reported use of a hypoglycemic agent, where 83% used metformin, 19% used insulin, and 2% used thiazolidinediones. Total daily steps and time in MVPA was significantly lower (P < 0.001 for both) and total sedentary time significantly higher in participants with diabetes (P < 0.001). QFI was higher in those with diabetes (P = 0.002). In addition, both ASAT and VAT volume and VAT/ASAT ratio were significantly higher in older adults with diabetes (P = 0.005 for ASAT and P < 0.001 for both VAT and VAT/ASAT ratio) (Table 1).
. | Without diabetes . | With diabetes . | P . |
---|---|---|---|
Basic characteristics | |||
Participants, n | 717 | 159 | |
Age (years) | 76.47 ± 5.05 | 75.79 ± 4.84 | 0.12 |
Female | 61 (438) | 51 (81) | 0.02 |
Race (White) | 88 (631) | 70 (111) | <0.001 |
Hispanic ethnicity | 1 (7) | 1 (2) | 0.76 |
Weight (kg) | 74.74 ± 14.75 | 82.41 ± 15.58 | <0.001 |
Height (m) | 1.66 ± 0.10 | 1.67 ± 0.10 | 0.16 |
Waist circumference (cm) | 92.69 ± 12.98 | 100.90 ± 12.57 | <0.001 |
BMI (kg/m2) | 27.17 ± 4.44 | 29.55 ± 4.63 | <0.001 |
HbA1c (%) | 5.51 ± 0.31 | 6.73 ± 0.89 | <0.001 |
Systolic blood pressure (mmHg) | 129.71 ± 15.80 | 133.72 ± 16.08 | 0.004 |
Comorbidities | |||
Lung disease | 13 (93) | 14 (22) | 0.77 |
Arthritis | 57 (409) | 52 (83) | 0.25 |
Stroke | 2 (16) | 3 (5) | 0.51 |
Heart disease | 6 (42) | 11 (18) | 0.02 |
Cancer | 43 (308) | 24 (38) | <0.001 |
Hypertension | 26 (187) | 37 (59) | 0.006 |
Kidney disease/renal failure | 3 (24) | 6 (9) | 0.19 |
Peripheral vascular disease | 1 (4) | 1 (2) | 0.37 |
Hypothyroidism | 19 (133) | 17 (27) | 0.64 |
Glaucoma | 9 (65) | 11 (18) | 0.39 |
Macular degeneration | 9 (61) | 5 (8) | 0.12 |
Cataracts | 66 (472) | 62 (99) | 0.37 |
Hypoglycemic medications* | |||
All | — | 75 (118) | |
Insulin | — | 19 (23) | |
Metformin | — | 84 (99) | |
Thiazolidinediones | — | 3 (3) | |
Physical activity | |||
Total daily steps | 7,066.68 ± 3,233.95 | 5,848.70 ± 2,791.12 | <0.001 |
Total sedentary time (min) | 610.10 ± 112.28 | 659.40 ± 108.65 | <0.001 |
Time in light physical activity (min) | 98.20 ± 26.54 | 92.84 ± 29.16 | 0.027 |
Time in MVPA (min) | 192.95 ± 86.40 | 155.66 ± 76.74 | <0.001 |
Body adiposity | |||
ASAT (L) | 7.50 ± 3.14 | 8.30 ± 3.31 | 0.005 |
VAT (L) | 3.96 ± 2.21 | 5.36 ± 2.31 | <0.001 |
VAT/ASAT | 0.57 ± 0.34 | 0.72 ± 0.38 | <0.001 |
QFI (%) | 0.066 ± 0.022 | 0.072 ± 0.020 | 0.002 |
. | Without diabetes . | With diabetes . | P . |
---|---|---|---|
Basic characteristics | |||
Participants, n | 717 | 159 | |
Age (years) | 76.47 ± 5.05 | 75.79 ± 4.84 | 0.12 |
Female | 61 (438) | 51 (81) | 0.02 |
Race (White) | 88 (631) | 70 (111) | <0.001 |
Hispanic ethnicity | 1 (7) | 1 (2) | 0.76 |
Weight (kg) | 74.74 ± 14.75 | 82.41 ± 15.58 | <0.001 |
Height (m) | 1.66 ± 0.10 | 1.67 ± 0.10 | 0.16 |
Waist circumference (cm) | 92.69 ± 12.98 | 100.90 ± 12.57 | <0.001 |
BMI (kg/m2) | 27.17 ± 4.44 | 29.55 ± 4.63 | <0.001 |
HbA1c (%) | 5.51 ± 0.31 | 6.73 ± 0.89 | <0.001 |
Systolic blood pressure (mmHg) | 129.71 ± 15.80 | 133.72 ± 16.08 | 0.004 |
Comorbidities | |||
Lung disease | 13 (93) | 14 (22) | 0.77 |
Arthritis | 57 (409) | 52 (83) | 0.25 |
Stroke | 2 (16) | 3 (5) | 0.51 |
Heart disease | 6 (42) | 11 (18) | 0.02 |
Cancer | 43 (308) | 24 (38) | <0.001 |
Hypertension | 26 (187) | 37 (59) | 0.006 |
Kidney disease/renal failure | 3 (24) | 6 (9) | 0.19 |
Peripheral vascular disease | 1 (4) | 1 (2) | 0.37 |
Hypothyroidism | 19 (133) | 17 (27) | 0.64 |
Glaucoma | 9 (65) | 11 (18) | 0.39 |
Macular degeneration | 9 (61) | 5 (8) | 0.12 |
Cataracts | 66 (472) | 62 (99) | 0.37 |
Hypoglycemic medications* | |||
All | — | 75 (118) | |
Insulin | — | 19 (23) | |
Metformin | — | 84 (99) | |
Thiazolidinediones | — | 3 (3) | |
Physical activity | |||
Total daily steps | 7,066.68 ± 3,233.95 | 5,848.70 ± 2,791.12 | <0.001 |
Total sedentary time (min) | 610.10 ± 112.28 | 659.40 ± 108.65 | <0.001 |
Time in light physical activity (min) | 98.20 ± 26.54 | 92.84 ± 29.16 | 0.027 |
Time in MVPA (min) | 192.95 ± 86.40 | 155.66 ± 76.74 | <0.001 |
Body adiposity | |||
ASAT (L) | 7.50 ± 3.14 | 8.30 ± 3.31 | 0.005 |
VAT (L) | 3.96 ± 2.21 | 5.36 ± 2.31 | <0.001 |
VAT/ASAT | 0.57 ± 0.34 | 0.72 ± 0.38 | <0.001 |
QFI (%) | 0.066 ± 0.022 | 0.072 ± 0.020 | 0.002 |
Data are mean ± SD or % (n). Boldface indicates significance accepted at P < 0.05.
Calculated only for participants taking medications.
Cardiorespiratory Fitness and Muscle Oxidative Capacity in Diabetes
In the adjusted model (Table 2, model 1 plus physical activity), VO2peak was significantly lower in older adults with diabetes compared with those without diabetes (P = 0.004). OXPHOSCHO (P = 0.12) and maxOXPHOSCHO (P = 0.10) were similar between groups. maxETSCHO was significantly lower in participants with diabetes (P = 0.03) (Table 2). OXPHOSFAO was similar between participants with and without diabetes (P = 0.10), and maxOXPHOSFAO remained significantly lower in older adults with diabetes in model 1 plus physical activity (P = 0.006). Sensitivity analyses completed in participants with data for both protocol 1 and protocol 2 had similar results (data not shown) to ensure that differences between protocols were not due to differences in sample size.
Model . | Diabetes status . | VO2peak (mL/min) . | OXPHOSCHO (pmol/mg/ww) . | maxOXPHOSCHO (pmol/mg/ww) . | maxETSCHO (pmol/mg/ww) . | OXPHOSFAO (pmol/mg/ww) . | maxOXPHOSFAO (pmol/mg/ww) . |
---|---|---|---|---|---|---|---|
Unadjusted | WOD | 1,532.10 ± 16.67 | 31.08 ± 0.45 | 60.63 ± 0.74 | 81.43 ± 0.97 | 12.94 ± 0.22 | 58.22 ± 0.84 |
WD | 1,513.75 ± 36.19 | 28.10 ± 0.97 | 55.11 ± 1.61 | 73.80 ± 2.11 | 11.93 ± 0.47 | 53.31 ± 1.78 | |
Model 1 | WOD | 1,486.68 ± 19.07 | 30.14 ± 0.91 | 59.02 ± 1.45 | 77.53 ± 1.90 | 12.58 ± 0.47 | 59.25 ± 1.64 |
WD | 1,386.69 ± 25.66* | 27.88 ± 1.11 | 54.98 ± 1.80 | 71.25 ± 2.37 | 11.49 ± 0.55 | 54.12 ± 1.93 | |
Model 1 + PA (steps/day) | WOD | 1,487.30 ± 19.53 | 30.15 ± 0.91 | 58.65 ± 1.47 | 77.02 ± 1.98 | 12.60 ± 0.50 | 59.21 ± 1.61 |
WD | 1,412.11 ± 26.03 | 28.51 ± 1.11 | 55.89 ± 1.78 | 71.83 ± 2.44 | 11.70 ± 0.59 | 54.29 ± 1.92 | |
Model 1 + ASAT (L) | WOD | 1,492.37 ± 19.52 | 30.34 ± 0.94 | 59.50 ± 1.49 | 77.64 ± 1.95 | 12.67 ± 0.48 | 59.83 ± 1.67 |
WD | 1,376.07 ± 26.56* | 27.83 ± 1.15 | 55.07 ± 1.82 | 70.92 ± 2.45 | 11.37 ± 0.57 | 53.85 ± 1.98 | |
Model 1 + VAT (L) | WOD | 1,488.92 ± 19.56 | 30.16 ± 0.94 | 59.16 ± 1.49 | 77.33 ± 1.96 | 12.62 ± 0.49 | 59.57 ± 1.68 |
WD | 1,386.86 ± 26.60* | 28.04 ± 1.15 | 55.46 ± 1.81 | 71.29 ± 2.45 | 11.50 ± 0.57 | 54.66 ± 1.98 | |
Model 1 + VAT/ASAT | WOD | 1,491.43 ± 19.47 | 30.29 ± 0.94 | 59.39 ± 1.49 | 77.53 ± 1.95 | 12.67 ± 0.49 | 59.90 ± 1.68 |
WD | 1,391.58 ± 26.76* | 28.02 ± 1.16 | 55.44 ± 1.83 | 71.33 ± 2.46 | 11.44 ± 0.57 | 54.01 ± 2.00 | |
Model 1 + QFI (%) | WOD | 1,486.37 ± 19.00 | 30.28 ± 0.93 | 59.38 ± 1.48 | 77.78 ± 1.94 | 12.68 ± 0.48 | 59.74 ± 1.68 |
WD | 1,386.10 ± 25.75* | 27.97 ± 1.14 | 55.33 ± 1.81 | 71.06 ± 2.43 | 11.37 ± 0.57 | 54.22 ± 1.98 |
Model . | Diabetes status . | VO2peak (mL/min) . | OXPHOSCHO (pmol/mg/ww) . | maxOXPHOSCHO (pmol/mg/ww) . | maxETSCHO (pmol/mg/ww) . | OXPHOSFAO (pmol/mg/ww) . | maxOXPHOSFAO (pmol/mg/ww) . |
---|---|---|---|---|---|---|---|
Unadjusted | WOD | 1,532.10 ± 16.67 | 31.08 ± 0.45 | 60.63 ± 0.74 | 81.43 ± 0.97 | 12.94 ± 0.22 | 58.22 ± 0.84 |
WD | 1,513.75 ± 36.19 | 28.10 ± 0.97 | 55.11 ± 1.61 | 73.80 ± 2.11 | 11.93 ± 0.47 | 53.31 ± 1.78 | |
Model 1 | WOD | 1,486.68 ± 19.07 | 30.14 ± 0.91 | 59.02 ± 1.45 | 77.53 ± 1.90 | 12.58 ± 0.47 | 59.25 ± 1.64 |
WD | 1,386.69 ± 25.66* | 27.88 ± 1.11 | 54.98 ± 1.80 | 71.25 ± 2.37 | 11.49 ± 0.55 | 54.12 ± 1.93 | |
Model 1 + PA (steps/day) | WOD | 1,487.30 ± 19.53 | 30.15 ± 0.91 | 58.65 ± 1.47 | 77.02 ± 1.98 | 12.60 ± 0.50 | 59.21 ± 1.61 |
WD | 1,412.11 ± 26.03 | 28.51 ± 1.11 | 55.89 ± 1.78 | 71.83 ± 2.44 | 11.70 ± 0.59 | 54.29 ± 1.92 | |
Model 1 + ASAT (L) | WOD | 1,492.37 ± 19.52 | 30.34 ± 0.94 | 59.50 ± 1.49 | 77.64 ± 1.95 | 12.67 ± 0.48 | 59.83 ± 1.67 |
WD | 1,376.07 ± 26.56* | 27.83 ± 1.15 | 55.07 ± 1.82 | 70.92 ± 2.45 | 11.37 ± 0.57 | 53.85 ± 1.98 | |
Model 1 + VAT (L) | WOD | 1,488.92 ± 19.56 | 30.16 ± 0.94 | 59.16 ± 1.49 | 77.33 ± 1.96 | 12.62 ± 0.49 | 59.57 ± 1.68 |
WD | 1,386.86 ± 26.60* | 28.04 ± 1.15 | 55.46 ± 1.81 | 71.29 ± 2.45 | 11.50 ± 0.57 | 54.66 ± 1.98 | |
Model 1 + VAT/ASAT | WOD | 1,491.43 ± 19.47 | 30.29 ± 0.94 | 59.39 ± 1.49 | 77.53 ± 1.95 | 12.67 ± 0.49 | 59.90 ± 1.68 |
WD | 1,391.58 ± 26.76* | 28.02 ± 1.16 | 55.44 ± 1.83 | 71.33 ± 2.46 | 11.44 ± 0.57 | 54.01 ± 2.00 | |
Model 1 + QFI (%) | WOD | 1,486.37 ± 19.00 | 30.28 ± 0.93 | 59.38 ± 1.48 | 77.78 ± 1.94 | 12.68 ± 0.48 | 59.74 ± 1.68 |
WD | 1,386.10 ± 25.75* | 27.97 ± 1.14 | 55.33 ± 1.81 | 71.06 ± 2.43 | 11.37 ± 0.57 | 54.22 ± 1.98 |
Data are mean ± SE. Boldface indicates significance accepted at P < 0.05 and an asterisk indicates significance at 0.001. Model 1 is adjusted for age, race, sex, site/technician, BMI, and comorbidities. PA, physical activity; WD, with diabetes; WOD, without diabetes; ww, wet weight.
Adjustment for levels of ASAT, VAT, VAT/ASAT ratio, and QFI did not impact group differences in mitochondrial respiration and VO2peak measurements, with an exception for OXPHOSCHO (Table 2). Adjustment for VAT tended to explain some variance in OXPHOSCHO between participants with and without diabetes (P = 0.06) (Table 2). Within the diabetes group only, ATPmax was lower in older adults taking any hypoglycemic medication (P = 0.05) or metformin alone (P = 0.01) (Supplementary Table 1). These findings make it difficult to distinguish the impact of medications versus diabetes on ATPmax, and therefore, further analysis of mitochondrial oxidative capacity focused on respiration measurements. VO2peak and mitochondrial oxidative capacity assessed by respiration were similar with the use of all hypoglycemic agents and metformin (Supplementary Table 1).
VO2peak and Mitochondrial Oxidative Capacity Mediate the Relationship Between Diabetes and Walking Speed
To determine the extent to which cardiorespiratory fitness and mitochondrial oxidative capacity were linked to clinically meaningful outcomes in older adults with and without diabetes, we next evaluated their relationship with 400-m and 4-m walking speeds. Both were significantly lower in participants with diabetes compared with those without diabetes following adjustments for sex, age, and race (base model) (Table 3). To assess whether VO2peak and mitochondrial respiration mediated the relationship between diabetes and walking speed, we compared the β-coefficient for mean 400-m and 4-m walking speed from the base linear regression model with the β-coefficient of models that included VO2peak and/or mitochondrial respiration variables. Additional individual adjustments for maxOXPHOSCHO (400-m, P = 0.02; model 2), maxOXPHOSFAO (400-m, P = 0.02; model 3), and VO2peak (400-m, P = 0.01; model 4) resulted in no significant change in group differences in 400-m walking speed between participants with and without diabetes (Table 3). Alternatively, for 4-m walking speed, further adjustments for maxOXPHOSCHO (4-m, P = 0.12; model 2) and maxOXPHOSFAO (4-m, P = 0.51; model 3) significantly mediated differences between groups, whereas VO2peak (4-m, P = 0.04; model 4) did not. Examining the impact of combinations of VO2peak and mitochondrial respiration revealed that VO2peak with maxOXPHOSCHO (400-m, P = 0.07; 4-m, P = 0.27; model 5) or maxOXPHOSFAO (400-m, P = 0.20; 4-m, P = 0.99; model 6) mediated differences between those with and without diabetes. Full adjustments including potential confounders/mediators BMI and chronic conditions explained the remaining variance in walking speed between groups (Table 3). Comparing the β-coefficients for mean 400-m and 4-m walking speed from the base linear regression model with the those of models that included both VO2peak and mitochondrial respiration variables revealed that VO2peak and mitochondrial respiration explained an additional ∼46–100% of the variance in 400-m and 4-m walking speeds between groups (Fig. 1).
. | Without diabetes . | With diabetes . | |
---|---|---|---|
Multivariable model . | Mean ± SE . | β ± SE . | P . |
400-m walking speed (m/s) | |||
Base model: adjusted for sex, age and race | 1.06 ± 0.01 | −0.05 ± 0.01 | 0.00 |
Model 2: base model plus maxOXPHOSCHO | 1.06 ± 0.01 | −0.04 ± 0.02 | 0.02 |
Model 3: base model plus maxOXPHOSFAO | 1.06 ± 0.01 | −0.04 ± 0.02 | 0.02 |
Model 4: base model plus VO2peak | 1.07 ± 0.01 | −0.04 ± 0.01 | 0.01 |
Model 5: base model plus VO2peak and maxOXPHOSCHO | 1.07 ± 0.01 | −0.03 ± 0.02 | 0.07 |
Model 6: base model plus VO2peak and maxOXPHOSFAO | 1.07 ± 0.01 | −0.02 ± 0.02 | 0.20 |
Fully adjusted: model 5 plus BMI and comorbidities | 1.06 ± 0.01 | 0.003 ± 0.01 | 0.87 |
Fully adjusted: model 6 plus BMI and comorbidities | 1.06 ± 0.01 | −0.005 ± 0.02 | 0.78 |
4-m walking speed (m/s) | |||
Base model: adjusted for sex, age, and race | 1.05 ± 0.01 | −0.04 ± 0.02 | 0.01 |
Model 2: base model plus maxOXPHOSCHO | 1.04 ± 0.01 | −0.03 ± 0.02 | 0.12 |
Model 3: base model plus maxOXPHOSFAO | 1.05 ± 0.01 | −0.01 ± 0.02 | 0.51 |
Model 4: base model plus VO2peak | 1.05 ± 0.01 | −0.04 ± 0.02 | 0.04 |
Model 5: base model plus VO2peak and maxOXPHOSCHO | 1.05 ± 0.01 | −0.02 ± 0.02 | 0.27 |
Model 6: base model plus VO2peak and maxOXPHOSFAO | 1.05 ± 0.01 | 0.00 ± 0.02 | 0.99 |
Fully adjusted: model 5 plus BMI and comorbidities | 1.05 ± 0.01 | 0.003 ± 0.02 | 0.84 |
Fully adjusted: model 6 plus BMI and comorbidities | 1.05 ± 0.01 | 0.01 ± 0.02 | 0.50 |
. | Without diabetes . | With diabetes . | |
---|---|---|---|
Multivariable model . | Mean ± SE . | β ± SE . | P . |
400-m walking speed (m/s) | |||
Base model: adjusted for sex, age and race | 1.06 ± 0.01 | −0.05 ± 0.01 | 0.00 |
Model 2: base model plus maxOXPHOSCHO | 1.06 ± 0.01 | −0.04 ± 0.02 | 0.02 |
Model 3: base model plus maxOXPHOSFAO | 1.06 ± 0.01 | −0.04 ± 0.02 | 0.02 |
Model 4: base model plus VO2peak | 1.07 ± 0.01 | −0.04 ± 0.01 | 0.01 |
Model 5: base model plus VO2peak and maxOXPHOSCHO | 1.07 ± 0.01 | −0.03 ± 0.02 | 0.07 |
Model 6: base model plus VO2peak and maxOXPHOSFAO | 1.07 ± 0.01 | −0.02 ± 0.02 | 0.20 |
Fully adjusted: model 5 plus BMI and comorbidities | 1.06 ± 0.01 | 0.003 ± 0.01 | 0.87 |
Fully adjusted: model 6 plus BMI and comorbidities | 1.06 ± 0.01 | −0.005 ± 0.02 | 0.78 |
4-m walking speed (m/s) | |||
Base model: adjusted for sex, age, and race | 1.05 ± 0.01 | −0.04 ± 0.02 | 0.01 |
Model 2: base model plus maxOXPHOSCHO | 1.04 ± 0.01 | −0.03 ± 0.02 | 0.12 |
Model 3: base model plus maxOXPHOSFAO | 1.05 ± 0.01 | −0.01 ± 0.02 | 0.51 |
Model 4: base model plus VO2peak | 1.05 ± 0.01 | −0.04 ± 0.02 | 0.04 |
Model 5: base model plus VO2peak and maxOXPHOSCHO | 1.05 ± 0.01 | −0.02 ± 0.02 | 0.27 |
Model 6: base model plus VO2peak and maxOXPHOSFAO | 1.05 ± 0.01 | 0.00 ± 0.02 | 0.99 |
Fully adjusted: model 5 plus BMI and comorbidities | 1.05 ± 0.01 | 0.003 ± 0.02 | 0.84 |
Fully adjusted: model 6 plus BMI and comorbidities | 1.05 ± 0.01 | 0.01 ± 0.02 | 0.50 |
Boldface indicates significance accepted at P < 0.05. Models including maxOXPHOSCHO and maxOXPHOSFAO are adjusted for technician, and models with VO2peak are adjusted for site.
Discussion
This study is the first to investigate whether cardiorespiratory fitness and skeletal muscle oxidative capacity contribute to slow walking speed in older adults with diabetes in a unique, well-phenotyped cohort of older adults that included 159 participants with diabetes. The main novel findings of the study are that 1) the association of diabetes status with slow walking speed was mediated by VO2peak and skeletal muscle mitochondrial respiration, 2) VO2peak and mitochondrial respiration were lower in older adults with diabetes while controlling for objectively assessed physical activity and adiposity, and 3) older adults taking hypoglycemic medications (insulin, thiazolidinediones, and/or metformin) or metformin alone had significantly lower ATPmax compared with older adults with diabetes but not taking medications, an observation that aligns with reports in the literature that metformin can impair mitochondrial oxidative capacity. Slower walking speed is indicative of poor health and has been shown to associate with survival in older adults (28). Together, these findings are impactful as they are the first to link a biological quality of muscle (oxidative capacity) to explain slower walking speed in a particularly vulnerable patient population.
The well-phenotyped SOMMA cohort provided an opportunity to assess the impact of diabetes status on VO2peak and mitochondrial oxidative capacity in older adults while controlling for objectively measured indices of physical activity, adiposity, and chronic conditions in addition to demographic variables. Many of the previous studies reporting lower muscle oxidative capacity measured in PmFBs from patients with diabetes have not rigorously assessed and controlled for the participants’ physical activity (12–14,17). This is important because physical activity interventions (e.g., walking) can improve mitochondrial energetics in type 2 diabetes (29), during weight loss in obesity (30), and in older adults (31). Here, we report that lower levels of objectively assessed physical activity partially explain lower respiration in older adults with diabetes. Interestingly, comparing our respirometry protocols with distinct substrate combinations, we found that maxOXPHOSFAO respiration remains lower in the diabetes group, while maxOXPHOSCHO did not differ following adjustments for physical activity. This suggests that carbohydrate-supported coupled respiration may be more sensitive to levels of physical activity compared with FAO-supported respiration. This is also in line with evidence indicating reduced rates of lipid oxidation in skeletal muscle from individuals with diabetes, potentially because of mitochondrial overload and incomplete FAO (32). We also acknowledge that reduced mitochondrial oxidative capacity may be due to reduced mitochondria content. Future analysis will investigate this further.
The influence of adiposity on mitochondrial energetics has been considered in prior studies of diabetes, typically by matching control subjects for BMI (14). Some studies (15,16), but not all (12,13,17), indicated no differences between patients with diabetes and BMI-matched control subjects, suggesting that obesity per se may underlie the lower muscle oxidative capacity. Here, we examined the influence that BMI and individual adipose depots (ASAT, VAT, and QFI) and VAT/ASAT ratio had on muscle mitochondrial energetics in diabetes. However, regardless of whether adiposity was adjusted for BMI alone or combined with individually objectively measured adipose depots, mitochondrial respiration generally remained significantly lower in participants with diabetes. Taken together, these findings indicate that adiposity and physical activity largely (there were some exceptions) do not entirely explain lower mitochondrial respiration in older adults with diabetes.
Metformin is one of the most commonly prescribed medications for diabetes, and although its mechanism of action remains highly debated, there is evidence of an inhibitory effect on complex I of the electron transport chain (33,34). Both preclinical and clinical studies reported maladaptation of mitochondrial function following metformin treatment, and blunting of exercise induced increases in muscle hypertrophy and cardiorespiratory fitness, specifically in older adults (33,35,36). Conversely, others have reported that metformin can improve mitochondrial function via effects on mitophagy, autophagy (37), or AMPK activation (38). Our current understanding of how metformin impacts skeletal muscle mitochondrial function of older adults is reviewed in detail elsewhere (39,40). Here, we reveal lower ATPmax in older adults with diabetes taking hypoglycemic medication or metformin alone compared with older adults with diabetes not taking medications. Interestingly, we did not observe an association of metformin with diabetes status in the respiration assays, suggesting different sensitivities to metformin based on whether mitochondrial energetics are assessed in vivo (ATPmax) compared with PmFB preparations ex vivo. However, an important caveat is that participants were asked to withhold medication prior to the muscle biopsy and respirometry assays but not on the day of 31P-MRS assessments. In addition, others have reported no effect on mitochondrial respiration after a 2-week metformin treatment compared with control subjects (41) or patients with long-term diabetes (42). Given that metformin is water soluble, it is also plausible that the inhibitory effect of metformin is lost when the muscle fiber bundles are washed in preparation for the respirometry assay. Further work is needed to decipher the impact metformin has on skeletal muscle mitochondrial energetics in older adults with diabetes.
Zhao et al. (43) reported that adults with diabetes were 31–34% less likely to participate in physical activity, which in turn may contribute to lower VO2peak. In line with these reports, SOMMA participants with diabetes were significantly less active and significantly more sedentary. Additional adjustments for physical activity did not resolve differences in VO2peak between groups, suggesting that lower VO2peak in older adults with diabetes is independent of physical activity in this cohort. In addition, independent of adiposity measured by BMI or combined BMI with ASAT, VAT, VAT/ASAT, and QFI, VO2peak remained significantly lower in participants with diabetes compared with those without. Taken together, our work suggests that physical activity and adiposity may not be implicated in lower cardiorespiratory fitness in older adults with diabetes and that other factors, including perhaps genetics and heritability, are more important determinants of low cardiorespiratory fitness in diabetes.
Previous work from our laboratory and others revealed a significant relationship between walking speed and both VO2peak and mitochondrial energetics in older adults (9,28,44). Here, we explored the contribution of both mitochondrial respiration and cardiorespiratory fitness to mobility in older adults with diabetes. We report that mitochondria respiration, both independently (∼40–70%) and combined with VO2peak (∼55–100%), mediated the variance in 4-m walking speed in participants with diabetes. Differences between groups are further explained by additional adjustments for potential confounders/mediators, including BMI and comorbidities. Reports have indicated that muscle mass is lost at an accelerated rate in older adults with diabetes (7), contributing to reductions in muscle strength and quality (6) and ultimately reducing physical function and mobility (5,45). We extend these findings by highlighting how mitochondrial respiration may independently and combined with cardiorespiratory fitness also contribute to lower walking speed in older adults with diabetes. This finding is of particular interest because slower walking speed is an important indicator of health and has been shown to associate with survival in older adults (28). Further assessments of specific aspects of the cardiorespiratory system (e.g., pulmonary function, oxygen transportation, muscle capillarization) and mitochondrial function (e.g., H2O2 emission, calcium retention capacity, membrane potential) may reveal novel targets to prevent mobility loss in older adults with diabetes.
Our study has several strengths but also a few notable weaknesses that we are well positioned to address in the future. First, the duration of diabetes status and the length of time that participants have been taking oral hypoglycemic agents were not recorded. However, this cohort is being followed longitudinally, so there is opportunity to capture that data in the future and for powerful analysis of changes over time in the same individuals. In addition, participants were predominantly (85.6%) non-Hispanic White, limiting our ability to generalize to other racial/ethnic groups where mobility disability is more prevalent (46). The participants recruited were able to complete a 4-m walking test at a gait speed of ≥0.6 m/s, which may be faster than those with more advanced diabetes who are unable to complete mobility-related tasks (2). However, longitudinal assessments will provide an opportunity to study decline in mobility and incident mobility disability in those with diabetes in the future. The strengths of our study design include the collection of muscle biopsies and use of two measurements of mitochondrial oxidative capacity in a large cohort of older individuals combined with rigorous assessment of objectively measured physical activity, fitness, and body composition. This unique study design allows us to link a fundamental biological process in muscle biopsies to a key clinical outcome. Furthermore, the collection of these biological resources, and the foundational work completed in the present study, will support future analysis focused on further understanding the biological qualities of muscle, including measures of mitochondrial content, that contribute to slower walking speed in older adults with diabetes.
In summary, our findings highlight that mitochondrial respiration independently and combined with cardiorespiratory fitness contribute to slower 400-m and 4-m walking speeds in older adults with diabetes. Additionally, future work should aim to decipher the impact of diabetes medication on mitochondrial function, as this remains a gap in the literature.
This article contains supplementary material online at https://doi.org/10.2337/figshare.25492276.
Article Information
Acknowledgments. The authors thank all participants who participated in this study at the University of Pittsburgh and Wake Forest University. They also acknowledge the dedicated staff and investigators at both clinical sites and at the San Francisco Coordinating Center.
Funding. National Institute on Aging (NIA) grant R01AG059416 funded SOMMA. Infrastructure support for SOMMA was funded, in part, by NIA Claude D. Pepper Older American Independence Centers at the University of Pittsburgh grant P30AG024827 and Wake Forest University School of Medicine grant P30AG021332 and by National Center for Advancing Translational Science at Wake Forest University grant UL1TR001420 and grants R01AG060153 and R01AG060542 (to P.M.Co.). G.D. was supported by the American Diabetes Association grant 1-19-PDF-006 during data collection and analysis.
Duality of Interest. S.R.C. and P.M.Ca. are consultants to Bioage Laboratories. P.M.Ca. is a consultant to and owns stock in MyoCorps. No other potential conflicts of interest relevant to this article were reported.
Author Contributions. S.V.R. and L.-Y.L. completed the statistical analysis, with final review from the statistical team at the coordinating center. S.R.C., P.M.Ca., A.B.N., R.T.H., S.B.K., and B.H.G. enabled the study with either funding acquisition, project administration, and/or conceptualization of the study. S.V.R. and P.M.Co. conceived the idea and cowrote the manuscript. G.D., L.-Y.L., P.K., I.J.S., F.M.B., D.J.M., E.S., and P.M.Co. completed experiments or quality control, validation, and interpretation of data. G.D., L.-Y.L., P.M.Ca., P.K., T.M., M.J.J., A.J.M., E.E.K., D.J.M., F.G.S.T., A.B.N., R.T.H., S.B.K., B.H.G., and S.R.C. edited the manuscript. P.M.Co. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Prior Presentation. Parts of this study were presented in abstract form at the 82nd Scientific Sessions of the American Diabetes Association, New Orleans, LA, 3–7 June 2022, and the American College of Sports Medicine Annual Meeting and World Congress, Denver, CO, 30 May–2 June 2023. Following submission of the manuscript, some data from the manuscript were also presented at the Gerontological Society of America (GSA) Annual Scientific Meeting, Tampa, FL, 8–12 November 2023 (symposium presentation).