Abstract
OBJECTIVE
The aim of this study was to describe cross-sectional and longitudinal associations between glycated hemoglobin (HbA1c) levels and strategies to control type 2 diabetes with baseline levels and 8-year changes in a deficit accumulation frailty index (FI), a commonly used marker of biological aging.
RESEARCH DESIGN AND METHODS
We conducted exploratory analyses from 4,169 participants, aged 45–76 years, who were followed in the Action for Health in Diabetes (Look AHEAD) randomized controlled clinical trial, pooling data across intervention groups. We related baseline and 8-year levels of HbA1c with FI scores using analyses of variance and covariance. Associations between 8-year changes in FI and the use of diabetes medication classes and weight changes were assessed with control for HbA1c levels. Inverse probability weighting was used to assess bias associated with differential follow-up.
RESULTS
Baseline and average HbA1c levels over time of <7%, as compared with ≥8%, were associated with less increase in FI scores over 8 years (both P ≤ 0.002). After adjustment for HbA1c, use of metformin and weight loss >5% were independently associated with slower increases in frailty.
CONCLUSIONS
Lower HbA1c levels among individuals with diabetes are associated with slower biological aging as captured by a deficit accumulation FI. Strategies to control diabetes through weight loss or metformin use may also slow aging.
Introduction
Type 2 diabetes is reported to accelerate biological aging (1,2) and is associated with both higher levels and more rapid increases in deficit accumulation frailty indices (FIs) (3). These indices combine age-related measures of individual physical and cognitive limitations with comorbidities and are used as indicators of the aging process (4). Among older individuals with type 2 diabetes, maintaining a lower FI is associated with fewer adverse health events (5,6), and increases in FIs are associated with subsequent increased risk for mortality and poorer trajectories of physical and cognitive function (7).
It is unclear whether better control of type 2 diabetes may slow increases in FI. While some have reported that lower glycated hemoglobin (HbA1c) among individuals with type 2 diabetes is associated with lower FI scores (3,8,9), others have reported that low HbA1c may be associated with higher FI scores and worse trajectories of FI over time (10,11). Associations may be complicated by clinical factors, including treatment with high hypoglycemia risk medications that may lower HbA1c but can cause serious adverse effects.
To further explore this topic, we examined the association between glycemic control and trajectories of FI among adults with type 2 diabetes over 8 years of follow-up. We also examined whether associations varied depending on the strategies for achieving control (weight loss, use of metformin as a first-line strategy, and other pharmacological therapies) because others have reported evidence that both intentional weight loss and metformin use may slow biological aging (e.g., 12,13). We examined whether weight loss and use of metformin (compared with the use of other oral medication and insulin) are independently associated with better 8-year trajectories of FI scores. We analyzed data from the large, diverse, and well-characterized cohort assembled by the Look AHEAD (Action for Health in Diabetes) randomized controlled clinical trial of a multidomain intensive lifestyle intervention (ILI) (14). We previously reported that this intervention, relative to a control condition of diabetes support and education (DSE), resulted in smaller 8-year FI increases (9).
Research Design and Methods
The Look AHEAD protocol and Consolidated Standards of Reporting Trials (CONSORT) diagram have been published (14,15). It was a multisite, single-masked randomized controlled clinical trial that recruited 5,145 individuals (during 2001 to 2004) from 16 U.S. centers. All individuals had type 2 diabetes and met the following criteria: 45–76 years of age, BMI >25 kg/m2 (>27 kg/m2 if on insulin), HbA1c <97 mmol/mol (11%), systolic/diastolic blood pressure <160/<100 mmHg, triglycerides <600 mg/dL, and a successful maximum graded exercise test. Protocols and consent forms were approved by local Institutional Review Boards.
Our descriptive analyses are based on data from the public use databases from Look AHEAD that are housed in the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) Central Repository (https://repository.niddk.nih.gov/studies/look-ahead/).
These include data on 4,906 of 5,145 Look AHEAD participants who agreed for their data to be shared. Of this subset, 4,174 (85%) had sufficient data to compute FIs at year 8 of follow-up and comprise our analysis data set. Supplementary Exhibit S5 provides a description of how these 4,147 participants differ from the 732 other repository participants who were excluded.
Interventions
Look AHEAD randomly assigned participants to ILI or DSE. The ILI targeted reducing caloric intake and increasing physical activity to induce weight loss ≥7% and maintaining this over time (16). Caloric consumption goals of 1,200 to 1,800 kcal/day were based on initial weight. Physical activity of >175 min/week through activities similar in intensity to brisk walking was targeted, as was improved diet (<30% calories from fat, <10% calories from saturated fat, and >15% calories from protein). During the first 6 months, ILI participants attended three group meetings and one individual session per month. For the remainder of the first year, they were provided two group meetings and one individual meeting per month. The intensity of the intervention gradually decreased thereafter.
DSE participants were invited to attend group sessions focused on diet, physical activity, and social support (17). Four meetings were offered during year 1, three per year during years 2–4, and one annually thereafter. Participants did not receive specific diet, activity, or weight goals or information on behavioral strategies. Both groups received feedback on cardiometabolic risk factors (lipids, HbA1c, and blood pressure).
Interventions ended September 2012, when follow-up ranged from 8.4 to 11.1 years among participants. We used data from the first 8 years of follow-up to span the full time of intervention delivery for all participants.
Diabetes Control and Treatment
HbA1c was assessed at baseline, annually through years 1–4, and at years 6 and 8 using standard protocols (14) by a central laboratory. We used the HbA1c at baseline and the average HbA1c across all assessments to characterize diabetes control for each individual and grouped individuals according to <7.0%, 7.0–7.9%, and ≥8%. The cut point of HbA1c <7% was based on clinical recommendations during the administration of the Look AHEAD intervention (18). The cut point of 8% provided a reasonable number of participants with higher HbA1c levels and has been used elsewhere in epidemiological research (e.g., O’Sullivan et al. [19]).
Participants were weighed annually by centrally trained and certified staff who were masked to intervention assignment (14). Participants were grouped into three categories according to weight changes from baseline: >5% decrease, no change in either direction of >5%, and >5% increase. A >5% weight loss has been defined as clinically significant (20), and cut points of >5% weight gains are often used in epidemiological studies (21).
Participants were asked to bring prescription medications to annual clinic visits for review. Diabetes medication use (grouped as insulins, metformin, and other oral medications) was summarized as the percent of these assessments (baseline through year 8) at which use of these medications was recorded and grouped as no use, <50% use, and ≥50% use.
Deficit Accumulation FI
We previously constructed a 38-item FI with components based on annual medical histories, clinic-based assessments, behaviors, functions, and abilities (9). Our original FI included two markers directly related to diabetes control (insulin use and fasting glucose): we eliminated these for the current analyses, resulting in a 36-component FI (see Supplementary Exhibit S7). Individual component scores range between 0 and 1, with higher scores reflecting health deficits. We calculated the total FI as the ratio of the sum of the individual component scores divided by the number of components, expressed as a percentage ranging from 0 to 100.
Statistical Analysis
Differences in baseline characteristics across HbA1c categories were assessed using χ2 tests and analyses of variance. A scatterplot with an overlaid spline regression line was used to portray the association between baseline HbA1c and FI. Associations between baseline HbA1c and 8-year changes in FI were assessed using analyses of covariance, with adjustment for age (continuous), sex, diabetes duration, and intervention assignment. Associations between mean 8-year HbA1c and 8-year FI changes were also based on ANCOVA with the same covariates. To examine the association between diabetes medication use (insulin, metformin, and other medications) and 8-year FI change, we used ANCOVA adjusted for mean HbA1c. Additional adjustment for baseline age, sex, and diabetes duration had little impact on results and are not reported. We used the same approach to examine the association between BMI and FI change. This latter analysis was repeated for participants grouped according to baseline age to assess whether associations extended to relatively older participants. The potential impact of differential attrition was gauged using inverse probability of attrition weighting (22), with weights based on logistic regression with the following predictors at baseline: age, BMI, years of education, sex, smoking status, history of cardiovascular disease, history of diabetes among first-degree relatives, intervention assignment, general health and mental health component scores from the SF-36, and interactions between age and each of these factors.
Data and Resource Availability
As noted above, data from the Look AHEAD program, for participants who provided consent, are publicly available at the NIDDK Central Repository.
Results
Table 1 describes characteristics of the cohort at baseline and cross-sectional associations with baseline HbA1c. Overall, those in the group with the highest baseline HbA1c levels were more likely to be younger with higher BMI and longer durations of diabetes and were more likely to use insulin. Random assignment to the Look AHEAD interventions was independent of baseline HbA1c. Mean FI was greater among those with HbA1c levels ≥8% (53 mmol/mol) compared with HbA1c levels <7% (64 mmol/mol). Overall, the correlation between baseline HbA1c and FI was modest (r = 0.08) and, based on spline regression, was approximately linear (Supplementary Exhibit S1).
Table 1.
Baseline characteristics at Look AHEAD enrollment by baseline HbA1c categories***
| Baseline characteristic | Baseline HbA1c (%) | |||
|---|---|---|---|---|
| <7.0 (53 mmol/mol) (N = 1,944) | 7.0–7.9 (N = 1,310) | ≥8.0 (64 mmol/mol) (N = 915) | P value* | |
| Age, years (%) | ||||
| 45–54 (23.1) | 390 (20.1) | 288 (22.0) | 291 (31.8) | <0.001 |
| 55–64 (56.9) | 1,138 (58.5) | 737 (56.3) | 499 (54.5) | |
| 65–76 (19.8) | 416 (21.4) | 285 (21.8) | 125 (13.7) | |
| Sex (%) | ||||
| Female (59.1) | 1,131 (58.2) | 785 (59.9) | 549 (60.0) | 0.51 |
| Male (40.9) | 813 (41.8) | 525 (40.1) | 366 (40.0) | |
| BMI, kg/m2 (%) | ||||
| 25–29 (15.2) | 330 (17.0) | 183 (14.0) | 121 (13.2) | 0.01 |
| 30–39 (62.6) | 1,215 (62.5) | 824 (62.9) | 571 (62.4) | |
| ≥40 (22.2) | 399 (20.5) | 303 (23.1) | 223 (24.4) | |
| Diabetes duration, years (%) (missing = 27) | ||||
| 0–4 (46.6) | 1,149 (59.5) | 513 (39.5) | 267 (29.2) | <0.001 |
| ≥5 (53.4) | 782 (40.5) | 785 (60.5) | 646 (70.8) | |
| Diabetes medication (%) (missing = 41) | ||||
| Insulin (14.9) | 132 (6.9) | 225 (17.4) | 257 (28.2) | <0.001 |
| Metformin/no insulin (52.4) | 991 (51.5) | 704 (54.4) | 468 (51.4) | |
| Other medications/no insulin (18.8) | 403 (21.0) | 231 (17.8) | 144 (15.8) | |
| No medication (13.9) | 397 (20.6) | 135 (10.4) | 41 (4.5) | |
| History of CVD (%)** | ||||
| No (86.9) | 1,714 (88.2) | 1,123 (85.7) | 785 (85.8) | 0.07 |
| Yes (13.1) | 230 (11.8) | 187 (14.3) | 130 (14.2) | |
| Intervention group (%) | ||||
| DSE (49.3) | 947 (48.7) | 656 (50.1) | 453 (49.5) | 0.74 |
| ILI (50.7) | 997 (51.3) | 654 (49.9) | 462 (50.5) | |
| FI | 19.93 (0.15) | 20.82 (0.19) | 21.53 (0.22) | <0.001 |
Data are N (% of column) or mean (SE) unless otherwise indicated.
χ2 test or ANOVA.
History of cardiovascular disease: self-report of prior myocardial infarction, coronary artery bypass, angioplasty/stent procedures, peripheral vascular disease, stroke, stable angina, or class I/II heart failure.
N = 5 participants included in our analysis files did not have records of a baseline HbA1c and did not contribute to this table.
Over 8 years of follow-up, those with greater mean HbA1c were more likely to be younger and female and to have shorter diabetes durations, to not use insulin, to be assigned to ILI, and to have a lower baseline FI (Table 2).
Table 2.
Baseline characteristics at Look AHEAD enrollment by mean HbA1c over follow-up
| Baseline characteristic | Mean HbA1c (%) over 8 years | |||
|---|---|---|---|---|
| <7.0 (53 mmol/mol) (N = 2,181) | 7.0–7.9 (N = 1,296) | ≥8.0 (64 mmol/mol) (N = 697) | P value | |
| Age, years (%) | ||||
| 45–54 (23.1) | 390 (17.9) | 319 (24.6) | 259 (37.2) | <0.001 |
| 55–64 (56.9) | 1,268 (58.1) | 737 (56.9) | 375 (53.8) | |
| 65–76 (19.8) | 523 (24.0) | 240 (18.5) | 63 (9.0) | |
| Sex (%) | ||||
| Female (59.1) | 1,264 (58.0) | 760 (58.6) | 445 (63.8) | 0.02 |
| Male (40.9) | 917 (42.0) | 536 (41.4) | 252 (36.2) | |
| BMI, kg/m2 (%) | ||||
| 25–29 (15.2) | 355 (16,3) | 182 (14.0) | 98 (14.1) | 0.30 |
| 30–39 (62.6) | 1,342 (61.5) | 834 (64.4) | 439 (63.0) | |
| ≥40 (22.2) | 484 (22.2) | 280 (21.6) | 160 (23.0) | |
| Diabetes duration, years (%) (missing = 27) | ||||
| 0–4 (46.6) | 1,216 (56.2) | 487 (37.8) | 229 (32.9) | <0.001 |
| ≥5 (53.4) | 946 (34.8) | 802 (62.2) | 467 (67.1) | |
| Diabetes medication (%) (missing = 41) | ||||
| Insulin (14.9) | 162 (7.5) | 248 (19.3) | 204 (29.4) | <0.001 |
| Metformin/no insulin (52.4) | 1,106 (51.3) | 720 (56.0) | 341 (49.2) | |
| Other medications/no insulin (18.8) | 445 (20.6) | 216 (16.8) | 117 (16.9) | |
| No medication (13.9) | 442 (20.5) | 101 (7.9) | 31 (4.5) | |
| History of CVD (%) | ||||
| No (86.9) | 1,921 (88.1) | 1,099 (84.8) | 607 (87.1) | 0.02 |
| Yes (13.1) | 260 (11.9) | 197 (15.2) | 90 (12.9) | |
| Intervention group (%) | ||||
| DSE (49.3) | 984 (45.1) | 697 (53.8) | 379 (54.4) | <0.001 |
| ILI (50.7) | 1,197 (54.9) | 599 (46.2) | 318 (45.6) | |
| FI | 20.22 (0.16) | 20.60 (0.19) | 21.49 (0.26) | <0.001 |
Data are N (% of column) or mean (SD) unless otherwise indicated.
As seen in Table 3, mean 8-year FI increased in a graded fashion with greater baseline HbA1c levels (P = 0.002). FI increases were also larger among those with greater average HbA1c levels over follow-up (P < 0.001), and this relationship was similar across participants grouped by baseline FI (interaction P = 0.63, data not shown). As seen in Supplementary Exhibit S2, associations were similar for participants grouped by baseline age (interactions P > 0.20).
Table 3.
Eight-year changes in frailty scores by HbA1c groupings
| Mean (SE) | P value | |
|---|---|---|
| Baseline HbA1c (%) | 0.002 | |
| <7.0 | 2.43 (0.29) | |
| 7.0–7.9 | 2.88 (0.31) | |
| ≥8.0 | 3.54 (0.34) | |
| Mean HbA1c over time (%) | <0.001 | |
| <7.0 | 1.99 (0.28) | |
| 7.0–7.9 | 3.23 (0.31) | |
| ≥8.0 | 4.47 (0.37) |
Mean HbA1c levels over time varied among participants grouped by diabetes medication use and BMI change (Supplementary Exhibit S3). Participants with mean HbA1c levels ≥8% (64 mmol/mol) over time, as compared with those averaging lower HbA1c, were less likely to have been assigned to ILI and more likely to use insulin. They were also less likely to use metformin or other oral diabetes medications or not use any medications. They also were less likely to have average weight losses of >5%.
As seen in Table 4, after adjustment for mean HbA1c levels over follow-up, FI increases were smaller among individuals who were using metformin or no medications to treat their type 2 diabetes and were greater among individuals who were using insulin. There was little association between the use of other oral diabetes medications and FI changes. There was a graded relationship between weight loss and mean (SE) FI progression: 1.41 (0.32) for those with average weight losses >5%, 2.90 (0.27) for those with stable weight, and 6.05 (0.46) for those with weight gains >5%. As seen in Supplementary Exhibit S4, the relationship was fairly linear. This association was seen in both intervention groups. Compared with those with >5% average gains in weight, those with >5% weight losses among ILI participants had mean (SE) 5.19 (0.80) less FI accumulation (N = 99 >5% gainers vs. N = 923 >5% losers). Among DSE participants, the mean difference was 4.26 (0.64) (N = 238 >5% gainers vs. N = 385 >5% losers). Weight loss remained associated with less FI accumulation after adjustment for medication use (P < 0.001), and metformin use remained associated with less FI accumulation after adjustment for weight loss (P = 0.008).
Table 4.
Eight-year changes in FI for participants grouped by percent use of diabetes medications across annual visits and by mean percent change in BMI over time, with adjustment for mean HbA1c across follow-up
| Diabetes control strategy | 8-year changes in FI | |
|---|---|---|
| Mean (SE) | Mean (95% CI) difference from reference | |
| Insulin | ||
| Never (N = 2,616) | 1.93 (0.27) | Reference |
| <50% (N = 749) | 3.85 (0.37) | 1.93 (1.29, 2.56) |
| ≥50% (N = 809) | 4.70 (0.37) | 2.77 (2.11, 3.44) |
| Metformin | ||
| Never (N = 866) | 3.21 (0.35) | Reference |
| <50% (N = 1,111) | 3.32 (0.32) | 0.11 (0.55, 1.63) |
| ≥50% (N = 2,197) | 2.23 (0.28) | −0.98 (−1.57, −0.40) |
| Other oral medications | ||
| Never (N = 269) | 2.77 (0.28) | Reference |
| <50% (N = 1,114) | 2.67 (0.32) | −0.10 (−0.62, 0.42) |
| ≥50% (N = 361) | 2.80 (0.45) | 0.04 (−0.79, 0.86) |
| No medications | ||
| Never (N = 2,940) | 2.79 (0.27) | Reference |
| <50% (N = 816) | 2.97 (0.62) | 0.18 (−0.41, 0.77) |
| ≥50% (N = 418) | 1.83 (0.45) | −0.96 (−1.78, −0.15) |
| Mean percent change in BMI over time | ||
| >5% gain (N = 338) | 6.05 (0.46) | Reference |
| −5% to 5% (N = 2,507) | 2.90 (0.27) | −3.14 (2.31, 3.78) |
| >5% loss (N = 1,329) | 1.41 (0.32) | −4.63 (3.73, 5.53) |
Supplementary Exhibit S5 describes differences in baseline characteristics between individuals in the Look AHEAD public use database (i.e., those who consented to have their data shared) who were included in our analyses versus those who were excluded because of missing 8-year data. Supplementary Exhibit S6 presents results from analyses parallel to those used for Table 4, but with inverse probability weighting to provide some protection against differential missing data. As can be seen, although there are minor differences in fitted means, the overall interpretation of these results closely mirrors those for Table 4.
Conclusions
Our descriptive analyses support that better diabetes control and specific approaches to diabetes treatment (weight loss and metformin use) might slow biological aging as captured by deficit accumulation FIs.
Others have reported that lower HbA1c is associated with lower incidence of frailty and less progression of FIs (3). However, some have raised concerns that among individuals meeting criteria for frailty, low HbA1c levels may accelerate its progression by increasing risks for functional deficits, falls, and mortality (11,23). This has led to recommendations that HbA1c targets should be adjusted upwards as frailty develops (24,25). This drew us to examine whether relatively lower HbA1c levels were associated with greater FI increases among Look AHEAD participants with higher baseline FI (i.e., those in the highest baseline tertile [FI >21]). Among these individuals, there was a graded relationship between 8-year subsequent increases in FI and average HbA1c levels over time (P = 0.01). For those with average HbA1c levels <7% (53 mmol/mol) over follow-up, mean (SE) increases in FI were 0.79 (0.47). For those with mean HbA1c levels 7–7.9%, mean FI increases were 1.93 (0.49) and 3.04 (0.50) among those with mean HbA1c levels ≥8% (64 mmol/mol). Thus, this suggests that among Look AHEAD participants with the greatest baseline FI, higher HbA1c levels both at baseline and averaged over follow-up were associated with accelerated aging. It may be that Look AHEAD inclusion criteria (e.g., successful graded exercise test) and regular monitoring of HbA1c and other metabolic risk factors altered relationships. In addition, there are reports that there may be different subtypes of diabetes related to frailty and HbA1c, which may introduce heterogeneity in findings (26). We also note there is some evidence that among Look AHEAD participants with higher baseline FI, assignment to ILI was associated with a (nonsignificant but troubling) increased incidence of major cardiovascular events, with a hazard ratio of 1.29 (95% CI 0.94, 1.42) (6). This is despite ILI being equally successful in producing weight losses and fitness gains among participants with both relatively low and high baseline FI.
Look AHEAD participants who did not use diabetes medications were almost exclusively those with well-controlled HbA1c (Supplementary Exhibit S3). These included some participants for whom diabetes was reversed (i.e., who maintained HbA1c levels <6.5% [48 mmol/mol] with no medications during follow-up after meeting criteria for diabetes at baseline) (27). As we note, FI tended to increase relatively slowly among individuals who were most often not using diabetes medications
Increases in FI tended also to be slower among individuals using metformin. There is a growing appreciation that metformin use may slow both increases in frailty (28–30) and, underlying this, biological aging (31–34). Metformin may target multiple pathways in aging (e.g., inflammation, senescence, and immunity), both intra- and extracellularly (13,34,35). In some mouse models, long-term treatment with metformin starting at middle age extended both health span and life span (36). However, in a study of adults in the Diabetes Prevention Program clinical trial, an average of 2.8 years of randomization to metformin versus placebo was found to be ineffective in reducing frailty (37) or reducing mortality (38). Perhaps a longer duration of use is necessary for benefits, or differences may be due to the random assignment to metformin in the Diabetes Prevention Program. Clinical trials to assess the association between frailty and metformin are currently underway, including one studying metformin use in older adults with obesity (28). Others are planned (34,39). The potential benefits we see for metformin do not appear to extend to other diabetes medications, including insulin. Of note, this does not apply to sodium–glucose cotransporter 2 inhibitors and glucagon-like peptide 1 receptor agonists, which were not approved and available to participants over the time course of this study.
As noted in the introduction, we have previously examined the association that 8-year changes in FI have with subsequent (over up to 10 years) trajectories of walking speed, cognitive function, and mortality in the Look AHEAD cohort (7). Compared with participants with essentially no 8-year progression of FI (lower tertile), those with FI increases of ≥5.4 units (upper quartile) had significantly poorer profiles of subsequent walking speed and cognitive function and greater rates of mortality: hazard ratio 2.32 (95% CI 1.84, 2.94). Individuals with intermediate increases in FI (mid-tertile: midrange = 2.61) also had worse profiles of functions and increased mortality compared with those in the lower tertile of FI increases. Thus, the strength of associations seen in Table 4 may have important clinical implications.
Limitations
We report data from eligible volunteers for a randomized clinical trial of weight loss who may not represent general populations. We cannot rule out causation by indication in that medication use was not randomly assigned: a randomized clinical trial would be required to rule out reverse causation. In addition, our analyses of associations with weight loss are not based on intention to treat but instead are based on achieved weight loss. We have documented that individuals excluded from our analyses due to missing data differed from those who were included according to many baseline characteristics; however, based on our inverse probability weighting analyses, this did not appear to materially bias our findings.
Summary
Among adults with type 2 diabetes, poorer glycemic control is associated with greater frailty and worse profiles of frailty over time. Frailty progressed more slowly among individuals achieving weight loss and those treated with metformin, which may reflect slower biological aging.
Article Information
Funding. This research was funded by two diversity supplements to the Action for Health in Diabetes Extension Study Biostatistics Research Center (NIDDK grants 3U01DK057136-19S1 and 3U01DK057136-19S2) and by the parent award U01DK057136. Support was also provided by National Institute on Aging grants R01AG058571, U01AG073697, and P30AG021332. Additional funding sources for the Look AHEAD study group are listed in Supplementary Exhibit S8.
Duality of Interest. No potential conflicts of interest relevant to this article were reported.
Author Contributions. F.R.S. conducted statistical analyses and collaborated on drafting the manuscript. J.N.J., S.J.P., S.B.K., E.J.B., M.N.M., and C.K.F. reviewed and edited the manuscript and contributed to the discussion. M.A.E. obtained funding, conceived the research, and drafted the manuscript. All authors approved the final version of the manuscript. F.R.S. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Footnotes
This article contains supplementary material online at https://doi.org/10.2337/figshare.21657221.
This article is featured in a podcast available at diabetesjournals.org/journals/pages/diabetes-core-update-podcasts.
A complete list of the members of the Look AHEAD Research Group can be found within the publicly available resources at https://www.lookaheadtrial.org.
References
- 1. Morley JE. Diabetes and aging: epidemiologic overview. Clin Geriatr Med 2008;24:395–405 [DOI] [PubMed] [Google Scholar]
- 2. Monickaraj F, Aravind S, Gokulakrishnan K, et al. Accelerated aging as evidenced by increased telomere shortening and mitochondrial DNA depletion in patients with type 2 diabetes. Mol Cell Biochem 2012;365:343–350 [DOI] [PubMed] [Google Scholar]
- 3. Aguayo GA, Hulman A, Vaillant MT, et al. Prospective association among diabetes diagnosis, HbA1c, glycemia, and frailty trajectories in an elderly population. Diabetes Care 2019;42:1903–1911 [DOI] [PubMed] [Google Scholar]
- 4. Hanlon P, Butterly E, Lewsey J, Siebert S, Mair FS, McAllister DA. Identifying frailty in trials: an analysis of individual participant data from trials of novel pharmacological interventions. BMC Med 2020;18:309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Hanlon P, Jani BD, Butterly E, et al. An analysis of frailty and multimorbidity in 20,566 UK Biobank participants with type 2 diabetes. Commun Med (Lond) 2021;1:28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Simpson FR, Pajewski NM, Beavers KM, et al. Does the impact of intensive lifestyle intervention on cardiovascular disease risk vary according to frailty as measured via deficit accumulation? J Gerontol A Biol Sci Med Sci 2021;18:339–345 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Espeland MA, Justice JN, Bahnson J, et al. Eight-year changes in multimorbidity and frailty in adults with type 2 diabetes mellitus: associations with cognitive and physical function and mortality. J Gerontol A Biol Sci Med Sci 2022;77:1691–1698 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Yanagita I, Fujihara Y, Eda T, et al. Low glycated hemoglobin level is associated with severity of frailty in Japanese elderly diabetes patients. J Diabetes Investig 2018;9:419–425 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Simpson FR, Pajewski NM, Nicklas B, et al.; Indices for Accelerated Aging in Obesity and Diabetes Ancillary Study of the Action for Health in Diabetes (Look AHEAD) Trial . Impact of multidomain lifestyle intervention on frailty through the lens of deficit accumulation in adults with type 2 diabetes mellitus. J Gerontol A Biol Sci Med Sci 2020;75:1921–1927 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Zaslavsky O, Walker RL, Crane PK, Gray SL, Larson EB. Glucose levels and risk of frailty. J Gerontol A Biol Sci Med Sci 2016;71:1223–1229 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Abdelhafiz AH, Peters S, Sinclair AJ. Low glycaemic state increases risk of frailty and functional decline in older people with type 2 diabetes mellitus - evidence from a systematic review. Diabetes Res Clin Pract 2021;181:109085. [DOI] [PubMed] [Google Scholar]
- 12. Yaskolka Meir A, Keller M, Bernhart SH, et al. Lifestyle weight-loss intervention may attenuate methylation aging: the CENTRAL MRI randomized controlled trial. Clin Epigenetics 2021;13:48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Justice JN, Gubbi S, Kulkarni AS, Bartley JM, Kuchel GA, Barzilai N. A geroscience perspective on immune resilience and infectious diseases: a potential case for metformin. Geroscience 2021;43:1093–1112 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Ryan DH, Espeland MA, Foster GD, et al.; Look AHEAD Research Group . Look AHEAD (Action for Health in Diabetes): design and methods for a clinical trial of weight loss for the prevention of cardiovascular disease in type 2 diabetes. Control Clin Trials 2003;24:610–628 [DOI] [PubMed] [Google Scholar]
- 15. Wing RR, Bolin P, Brancati FL, et al.; Look AHEAD Research Group . Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med 2013;369:145–154 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Wadden TA, West DS, Delahanty L, et al.; Look AHEAD Research Group . The Look AHEAD study: a description of the lifestyle intervention and the evidence supporting it [published correction appears in Obesity (Silver Spring) 2007;15:1339]. Obesity (Silver Spring) 2006;14:737–752 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Wesche-Thobaben JA. The development and description of the comparison group in the Look AHEAD Trial. Clin Trials 2011;8:320–329 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Skyler JS, Bergenstal R, Bonow RO, et al.; American Diabetes Association; American College of Cardiology Foundation; American Heart Association . Intensive glycemic control and the prevention of cardiovascular events: implications of the ACCORD, ADVANCE, and VA Diabetes Trials: a position statement of the American Diabetes Association and a Scientific Statement of the American College of Cardiology Foundation and the American Heart Association. J Am Coll Cardiol 2009;53:298–304 [DOI] [PubMed] [Google Scholar]
- 19. O’Sullivan CJN, Hynes N, Mahendran B, et al. Haemoglobin A1c (HbA1C) in non-diabetic and diabetic vascular patients. Is HbA1C an independent risk factor and predictor of adverse outcome? Eur J Vasc Endovasc Surg 2006;32:188–197 [DOI] [PubMed] [Google Scholar]
- 20. Williamson DA, Bray GA, Ryan DH. Is 5% weight loss a satisfactory criterion to define clinically significant weight loss? Obesity (Silver Spring) 2015;23:2319–2320 [DOI] [PubMed] [Google Scholar]
- 21. Newman AB, Yanez D, Harris T, Duxbury A, Enright PL; Cardiovascular Study Research Group . Weight change in old age and its association with mortality. J Am Geriatr Soc 2001;49:1309–1318 [DOI] [PubMed] [Google Scholar]
- 22. Weuve J, Tchetgen Tchetgen EJ, Glymour MM, et al. Accounting for bias due to selective attrition: the example of smoking and cognitive decline. Epidemiology 2012;23:119–128 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Abdelhafiz AH, Sinclair AJ. Low HbA1c and increased mortality risk-is frailty a confounding factor? Aging Dis 2015;6:262–270 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Strain WD, Down S, Brown P, Puttanna A, Sinclair A. Diabetes and frailty: an expert consensus statement on the management of older adults with type 2 diabetes. Diabetes Ther 2021;12:1227–1247 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Westall SJ, Narayanan RP, Watmough S, et al. The individualisation of glycaemic targets in response to patient characteristics in type 2 diabetes: a scoping review. Clin Med (Lond) 2022;22:257–265 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Abdelhafiz AH, Sinclair AJ. Metabolic phenotypes explain the relationship between dysglycaemia and frailty in older people with type 2 diabetes. J Diabetes Complications 2022;36:108144. [DOI] [PubMed] [Google Scholar]
- 27. Gregg EW, Chen H, Wagenknecht LE, et al.; Look AHEAD Research Group . Association of an intensive lifestyle intervention with remission of type 2 diabetes. JAMA 2012;308:2489–2496 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Espinoza SE, Musi N, Wang C-P, et al. Rationale and study design of a randomized clinical trial of metformin to prevent frailty in older adults with prediabetes. J Gerontol A Biol Sci Med Sci 2020;75:102–109 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Baskaran D, Aparicio-Ugarriza R, Ferri-Guerra J, Milyani R, Florez H, Ruiz JG. Is there an association between metformin exposure and frailty? Gerontol Geriatr Med 2020;6:2333721420924956. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Piskovatska V, Stefanyshyn N, Storey KB, Vaiserman AM, Lushchak O. Metformin as a geroprotector: experimental and clinical evidence. Biogerontology 2019;20:33–48 [DOI] [PubMed] [Google Scholar]
- 31. Torres W, Nava M, Galbán N, et al. Anti-aging effect of metformin: a molecular and therapeutical perspective. Curr Pharm Des 2020;26:4496–4508 [DOI] [PubMed] [Google Scholar]
- 32. Glossmann HH, Lutz OMD. Metformin and aging: a review. Gerontology 2019;65:581–590 [DOI] [PubMed] [Google Scholar]
- 33. Campbell JM, Bellman SM, Stephenson MD, Lisy K. Metformin reduces all-cause mortality and diseases of ageing independent of its effect on diabetes control: a systematic review and meta-analysis. Ageing Res Rev 2017;40:31–44 [DOI] [PubMed] [Google Scholar]
- 34. Barzilai N, Crandall JP, Kritchevsky SB, Espeland MA. Metformin as a tool to target aging. Cell Metab 2016;23:1060–1065 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Kulkarni AS, Gubbi S, Barzilai N. Benefits of metformin in attenuating the hallmarks of aging. Cell Metab 2020;22:15–30 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Martin-Montalvo A, Mercken EM, Mitchell SJ, et al. Metformin improves healthspan and lifespan in mice. Nat Commun 2013;4:2192. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37. Hazuda HP, Pan Q, Florez H, et al. Association of intensive lifestyle and metformin interventions with frailty in the Diabetes Prevention Program Outcomes Study. J Gerontol A Biol Sci Med Sci 2021;76:929–936 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Lee CG, Heckman-Stoddard B, Dabelea D, et al.; Diabetes Prevention Program Research Group . Effect of metformin and lifestyle interventions on mortality in the Diabetes Prevention Program and Diabetes Prevention Program Outcomes Study. Diabetes Care 2021;44:2775–2782 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39. Justice JN, Niedernhofer L, Robbins PD, et al. Development of clinical trials to extend healthy lifespan. Cardiovasc Endocrinol Metab 2018;7:80–83 [DOI] [PMC free article] [PubMed] [Google Scholar]