Rationale and Study Design of a Randomized Clinical Trial of Metformin to Prevent Frailty in Older Adults With Prediabetes

Abstract

Background

Frailty is a geriatric syndrome that leads to poor health outcomes with aging. Previous studies have demonstrated that insulin resistance and inflammation predict frailty onset. Metformin is a widely used, well-tolerated drug that improves insulin sensitivity and displays anti-inflammatory properties. It is also known to prevent diabetes onset in adults with prediabetes. We hypothesize that metformin in older adults with prediabetes will promote healthy aging and prevent frailty. Here we describe an ongoing placebo-controlled, double-blinded clinical trial of metformin for the prevention of frailty in older adults with prediabetes.

Methods

Older adults aged more than 65 years are randomized to metformin or placebo and are followed for 2 years. Prediabetes, required for inclusion, is assessed by 2-hour oral glucose tolerance test. Exclusion criteria are baseline frailty (Fried criteria), diabetes, dementia, untreated depression, active malignancy, or severe cardiovascular, pulmonary, and neurologic diseases. Primary outcome is frailty; secondary outcomes are physical function (Short Physical Performance Battery), systemic and skeletal muscle tissue inflammation, muscle insulin signaling, insulin sensitivity (insulin clamp), glucose tolerance (oral glucose tolerance test), and body composition (dual-energy x-ray absorptiometry). Subjects are followed every 3 months for safety assessments and every 6 months for frailty assessment (Fried criteria) and oral glucose tolerance test, and every 12 or 24 months for secondary outcomes. Enrollment of 120 subjects (completers) will take place over a 2-year period.

Conclusion

Metformin is being examined in this study as a potential therapeutic agent to prevent frailty in older adults with prediabetes. Findings from this trial may have future implications for the screening and potential treatment of prediabetes in older patients with metformin for the prevention of frailty.

Frailty is a geriatric syndrome that leads to poor health outcomes such as falls, disability, institutionalization, and death (1). The prevalence of frailty is estimated at 7%–15% among community-dwelling older U.S. adults (1–3). The cost of frailty in the United States is estimated to be more than $18 billion annually (4), the result of increased health care costs associated with increased comorbidity, health care utilization and loss of independence (5,6). Although the importance of frailty and its impact on an aging U.S. society have been widely recognized, to date there is no effective intervention to prevent or treat frailty.

Approximately 25%–30% of older adults in the United States have diabetes and another 25%–40% have prediabetes (7,8). There is a progressive decline in glucose tolerance with aging (9). Although the exact etiology is unknown, age-dependent decreases in pancreatic β cell function and insulin sensitivity are believed to play important roles in the deterioration of glucose homeostasis that occurs with advancing age (10). These are also likely related to increased adiposity, sarcopenia, and reduced physical activity with aging (11). Aging is also known to be associated with decreased β cell proliferation (12) and increased β cell susceptibility to apoptosis (13), which is pronounced in those with impaired glucose tolerance (14). Several studies have demonstrated a strong association between diabetes, insulin resistance, and frailty (15–17). Diabetes is one of the most significant predictors of frailty onset and worsening over time (18). Therefore, both insulin resistance and diabetes are associated with frailty and predict its onset.

Metformin, a biguanide, is the most widely used oral antidiabetic drug and is generally recommended for first-line treatment of type 2 diabetes. Early initiation at the time of diagnosis, when the hemoglobin A1c (HbA1c) is not significantly elevated, has been associated with improved glycemic control over time and decreased long-term complications (19). Metformin has been shown to be highly effective at reducing the onset of diabetes in the Diabetes Prevention Program study (20), and at reducing systemic inflammation (21). More recently, animal studies suggest that metformin has life extension properties and improves health span (22).

This research will test metformin as a pharmacological intervention that could potentially be used to prevent or delay the development of frailty in older adults. We focus on glucose intolerant subjects, a population that encompasses more than 40% of older adults without diagnosed diabetes (8), and is most likely to benefit from metformin, as they are at increased risk for frailty onset (16). Because of the enormous costs associated with frailty (both personal and economic), an intervention that prevents or delays frailty, even in a subpopulation of older adults such as those with prediabetes, would have a major positive impact on our society. To the best of our knowledge, this is the first study to determine a potential intervention targeted toward central pathogenic processes (inflammation and insulin resistance) involved in the etiology of frailty in humans (Figure 1).

Figure 1.

Aging and prediabetes are strongly linked with inflammation and insulin resistance, two major predictors of frailty. Metformin may prevent frailty by improving inflammation and insulin resistance in older adults with prediabetes. Inflammation will be measured systemically and in skeletal muscle. Insulin resistance is measured via insulin clamping and oral glucose tolerance testing. Frailty will be measured using the Fried criteria and the frailty index.

Methods

Design Overview

This is a double-blinded, randomized, placebo-controlled trial designed to assess whether treatment with metformin (oral maximum dose of 1,000 mg twice a day) for 2 years will prevent or delay the onset of frailty in participants 65 years and older. The protocol for this study has been approved by the institutional review board at The University of Texas Health Science Center at San Antonio and the National Institute on Aging-Appointed Data Safety Monitoring Board (DSMB). The DSMB also approved the study manual of procedures prior to study initiation. The DSMB performs ongoing review and evaluation of the accumulated study data for participant safety, study conduct, and progress, and makes recommendations to the National Institute on Aging concerning the continuation, modification, or termination of the trial. The study is registered with clinicaltrials.gov (NCT02570672).

Study Population

Participants are community-dwelling older adults with prediabetes who are either non-frail or pre-frail (frail are excluded). Individuals who meet eligibility criteria are randomized and followed for 2 years (Figure 2). Recruitment methods include media advertisements and appearances, community events, and query of electronic medical records for potentially eligible subjects. Inclusion and exclusion criteria are the following.

Figure 2.

Study design includes initial screening for eligibility, baseline assessments of glucose metabolism and inflammation, randomization to metformin or placebo, and follow-up for 2 years. OGTT = oral glucose tolerance test.

Inclusion criteria

1. Men and women

2. All ethnic groups

3. Age 65 years and older

4. Community-dwelling

5. Prediabetic based on oral glucose tolerance test (OGTT) with 2 hour blood glucose values of 140–199 mg/dL after a 75-g oral glucose load, and no clinical diagnosis of diabetes in the past 12 months

6. Characterized as non-frail or pre-frail (Fried criteria), defined as the presence of 0 (non-frail) or 1 or 2 (pre-frail) of 5 frailty characteristics: (i) weak hand grip strength, (ii) slow walking speed, (iii) low physical activity, (iv) unintentional weight loss of at least 10 lb over the past year, and (v) self-reported exhaustion (1,3)

7. Laboratory values: Hematocrit at least 33%; aspartate aminotransferase less than two times the upper limit of normal; alanine aminotransferase less than two times the upper limit of normal; alkaline phosphatase less than two times the upper limit of normal; normal urinalysis; normal electrolytes; normal platelets; normal PT and PTT; glomerular filtration rate (GFR) of at least 45 mL/min; and urine protein ≤ 100 mg/dL by urinalysis.

Exclusion criteria

1. Characterized as frail (Fried criteria), defined as the presence of at least three of five frailty characteristics as described earlier (1,3)

2. Subjects with OGTT 2-hour glucose of at least 200 mg/dL, fasting glucose of at least 126 mg/dL, or currently taking glucose lowering medications or any drugs known to affect glucose homeostasis

3. Untreated depression or score more than 7 on the Geriatric Depression Scale

4. Diagnosis of any disabling neurologic disease such as Parkinson’s disease, amyotrophic lateral sclerosis, multiple sclerosis, cerebrovascular accident with residual deficits (muscle weakness or gait disorder), severe neuropathy, diagnosis of dementia or Mini-Mental State Exam score less than 24, cognitive impairment due to any reason such that the patient is unable to provide informed consent

5. History of moderate to severe heart disease (New York Heart Classification greater than grade II) or pulmonary disease (dyspnea on exertion upon climbing one flight of stairs or less)

6. Poorly controlled hypertension (systolic more than 160 mm Hg, diastolic more than 100 mm Hg)

7. Subjects who have been treated with long-term (>30 days) systemic steroids, anabolic steroids, growth hormone, or immunosuppressants within the last 6 months. Males with a medical history of testosterone deficiency who are on a stable dose of testosterone replacement (for ≥3 months) are allowed.

8. Subjects who have been treated with short-term (<30 days) systemic steroids, anabolic steroids, growth hormone, or immunosuppressants within the last 1 month

9. Chronic inflammatory condition, autoimmune disease, or infectious processes (eg, active tuberculosis, HIV, rheumatoid arthritis, systemic lupus erythematosus, acute or chronic hepatitis B or C)

10. Active tobacco use (within 6 months)

11. Active malignancy, non-skin

12. Hypersensitivity or allergy to metformin or pioglitazone

13. Donated blood within the last 2 months.

Study Procedures

Telephone screening

Potential participants undergo telephone screening to rule out clear excluding medical history prior to in-person screening. Individuals known to have no clear exclusionary criteria at the time of telephone screening are invited to an initial screening visit (Visit 1).

Screening visit

At the screening visit and after informed consent, the history and physical examination are performed to document medical history and physical exam findings according to key inclusion and exclusion criteria. This includes measurement of vital signs, height, body weight, and waist circumference. Medication history is also performed. Treatment with anticoagulants, aspirin (up to 325 mg), and clopidogrel are permitted if these can be held for 7 days prior to muscle biopsy procedures.

A frailty assessment is performed using Fried criteria (1), which have been standardized for a South Texas older adult population (see frailty definition in the Inclusion Criteria and Exclusion Criteria section) (3). Standardized cut points to establish low physical activity, weakness, and slowness have been published previously (3).

Glucose tolerance is assessed with a 2-hour OGTT. Baseline, fasting samples for determination of glucose, insulin, and free fatty acid concentrations are drawn at −30, −15, and 0 minutes. At time zero, subjects ingest 75 g of glucose. Glucose, insulin, and free fatty acid are determined every 30 minutes for 2 hours following glucose ingestion. Subjects qualify as prediabetic if the 2-hour glucose is 140–199 mg/dL; participants with 2-hour glucose outside the prediabetic range are excluded. Clinical laboratory tests include complete blood count, comprehensive metabolic panel with liver function, lipid panel, HbA1c, coagulation tests (PT and PTT), Vitamin B12 level, and urinalysis. As measures of lower extremity function, the Short Physical Performance Battery, the 6-minute walk, and lower extremity isometric strength are also performed following the OGTT for those who meet all inclusion and exclusion criteria.

Subjects who meet study eligibility at the screening visit return for Visit 2 (7–21 days after Visit 1) and undergo assessment of body composition, insulin sensitivity, and muscle biopsies. Whole body dual-energy x-ray absorptiometry is performed to measure lean and fat body mass. Insulin sensitivity is measured after an overnight fast using a 120-minute euglycemic hyperinsulinemic clamp with tritiated glucose as described (23). Indirect calorimetry is performed during the 45 minutes prior to initiating the clamp and 30 minutes before the end of the insulin clamp to measure fat and carbohydrate oxidation and oxidative and nonoxidative glucose disposal. Thirty minutes before starting the clamp, a biopsy of the vastus lateralis muscle is performed for measurements of local (muscle) inflammation as described (24). A second muscle biopsy is performed in the contralateral leg before the end of the clamp for measurements of insulin receptor signaling. At the end of this visit, subjects receive standard diet and exercise advice using resources available from the National Institute on Aging, including “Exercise & Physical Activity: Your Everyday Guide from the National Institute on Aging” (25), and “What’s On Your Plate? Smart Food Choices for Healthy Aging” (26). At the end of the visit, subjects are randomized to study drug and study medication is dispensed.

Administration of study drug

Participants are randomized to metformin or placebo in 1:1 design. All study staff are blinded to the study drug assignment except for the study pharmacist and the study statistician. Study drug is initiated at 500 mg daily and if tolerating study drug are instructed to increase to 500 mg twice daily after 2 weeks. Phone calls are made weekly to titrate medications as tolerated until the patient is at maximum dose tolerated with goal dose of 1,000 mg twice daily. The titration of study medication occurs over a run-in period of up to 3 months depending upon individual patient tolerability to the study drug. With a longer study titration period we do not anticipate differential attrition by study group assignment. During the pilot phase of this study, which occurred from August 2016 to September 2017, no participants had compliance less than 80% as measured by pill counts. Further, no subjects needed to permanently discontinue the medication due to intolerable side effects. On the basis of this pilot information, we did not feel that a run-in period prior to randomization was necessary. The administration of metformin in adults with normal renal function is safe; the inclusion and exclusion criteria have been tailored so as to reduce the possibility of adverse events. The most severe adverse event associated with metformin use is lactic acidosis, which is very rare (occurs in <1%) and is more common in patients with impaired renal function. Therefore, individuals with creatinine clearance less than 45 mL/min are excluded from this study based on current U.S. Food and Drug Administration dosing recommendations. More common adverse reactions (occur in >10%) are diarrhea, nausea, and vomiting. Subjects are provided with the common side effects as part of the informed consent process and are monitored closely throughout the study.

Follow-up study procedures

Subjects are seen in-person for biopsy site check at 1 week, and then repeat history and physical examination at 3 months, 6 months, and then every 6 months thereafter. OGTT is performed every 6 months to monitor for diabetes conversion. Frailty is assessed every 6 months and is operationalized as frailty category (non-frail, pre-frail, or frail) and score (0–5). At baseline the criterion of weight loss is assessed by self-report of unintentional weight loss as described earlier. However, at the follow-up exams weight loss is assessed by percent weight loss. Subjects who lose at least 5% of their baseline weight are considered to have met the criteria for weight loss. Measurement of insulin sensitivity and muscle biopsies is repeated at 12 and 24 months, such that the treatment effect on tissue inflammation and insulin signaling may be determined. Safety laboratory measurements (complete blood count, chemistry with renal function) are performed every 3 months. The full study schedule is shown in Table 1.

Table 1.

Study Visits and Procedures

Visit	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
Year	1									2
Month	Screening visit	0	1	3	6	9	12	12	12	15	18	21	24	24	24
Time interval		3–21 d	5–7 d	2.5 mo	3 mo	3 mo	3 mo	3–7 d	5–7 d	2.5 mo	3 mo	3 mo	3 mo	3–7 d	5–7 d
H and P, with assessment of affect, cognition, and frailty	X			X	X		X				X		X
ECG	X						X						X
Urinalysis	X						X
OGTT	X				X		X				X		X
CBC, complete chemistry, lipid panel	X			X	X	X	X			X	X	X	X
HbA1c	X			X	X	X	X			X	X	X	X
Coagulation tests	X						X						X
Systemic inflammation		X			X			X			X			X
DXA		X						X						X
Insulin clamp and muscle biopsies		X						X						X
Examination of biopsy site			X						X						X

Note: CBC = complete blood count; DXA = dual-energy x-ray absorptiometry; ECG = electrocardiogram; H and P = history and physical; HbA1c = hemoglobin A1c; OGTT = oral glucose tolerance test.

Management of diabetes conversions

Any one of the following prompts 2-hour OGTT to confirm possible diagnosis of diabetes: fasting glucose of ≥126 mg/dL, HbA1c of ≥6.5%, or 2-hour OGTT glucose of ≥200 mg/dL. If any of these occur, OGTT is conducted within 6 weeks. Fasting plasma glucose and HbA1c are measured for clinical signs and symptoms of diabetes; and, if fasting glucose is ≥126 mg/dL or HbA1c is ≥6.5%, OGTT is conducted within 6 weeks. If the 2-hour OGTT glucose is ≥200 mg/dL, diabetes conversion is confirmed (based on Diabetes Prevention Program study protocol) (27). If diabetes conversion is confirmed, measurements of frailty, insulin sensitivity (clamp), body composition (dual-energy x-ray absorptiometry), and inflammation (muscle and systemic) are performed. After these repeat measurements are completed, study allocation is unblinded. Subjects who are randomized to placebo are initiated on metformin and titrated as earlier. All study staff remain blinded to the prior treatment allocation. Subjects who convert to diabetes continue in study follow-up, and if the HbA1c reaches ≥7.5, subjects are initiated on 15-mg pioglitazone in addition to metformin unless there is a contraindication to pioglitazone. Pioglitazone is not initiated in any subject with hematuria, history of bladder cancer, lower extremity edema more than one, or known history of osteopenia or osteoporosis. The rationale for adding pioglitazone versus other antidiabetic agents is that the low-dose metformin–thiazolidinedione combination has been shown to have a potent effect to prevent diabetes (28). Pioglitazone is increased to 30 mg if HbA1c remains ≥7.5. If maximal metformin (1,000 mg BID) and pioglitazone (30 mg daily) doses are reached (or if the patient cannot tolerate metformin and/or pioglitazone) and HbA1c is ≥8, subjects are referred to their primary care physicians for further treatment, but remain in the study for frailty assessment and all other follow-up studies.

Side Effects and Safety Concerns Related to Use of Metformin

Gastrointestinal side effects include diarrhea, abdominal pain, vomiting, nausea, a metallic taste, bloating, flatulence, and decreased appetite. If these symptoms are mild and tolerable, the study medication is continued. If they are moderate or difficult to tolerate, study medication is withheld or dose reduced. In the event that diarrhea, abdominal pain, or vomiting becomes severe enough to cause dehydration or volume depletion, the study medication is discontinued immediately and the participant is evaluated and treated appropriately.

Metformin is not known to cause renal insufficiency; however, it is associated with an increased risk for lactic acidosis if used in persons with GFR or creatinine clearance rate below 30 mL/min. Thus, metformin use is contraindicated in patients with GFR less than 30 mL/min, and should not be initiated in patients with GFR less than 45 mL/min (29). On the basis of U.S. Food and Drug Administration dosing guidelines, subjects with GFR less than 45 mL/min are not enrolled in the study. If the GFR is less than 45 mL/min at any point in the study, the study drug is discontinued and the serum creatinine and GFR is rechecked in 2 weeks. Study drug is restarted at the previously tolerated dose if the repeat GFR is more than 45 mL/min. If the GFR upon repeat assessment is less than 30 mL/min, the study drug is discontinued permanently. If the GFR falls between 30 and 45 mL/min, study drug is initiated at reduced dosage (maximum of 1,000 mg/day). Metformin may rarely be associated with the development of lactic acidosis, defined as a metabolic acidosis with lactate at least 5.0 mM. If unexplained metabolic acidosis occurs, the study drug is discontinued immediately and not restarted. Because of the potential danger of contrast induced renal insufficiency and lactic acidosis associated with metformin, subjects are informed at study initiation and a warning letter is sent to all primary care providers to alert them that the study drug should be discontinued prior to any radiological studies involving contrast dyes. Serum creatinine level is checked 48 hours after dye administration. The study drug is reinitiated if the GFR is in acceptable range as described earlier.

In all cases, if the study drug is discontinued, subjects remain in the study for continued follow-up and assessment of outcomes.

Study Endpoints and Statistical Analysis

Analysis of primary outcomes

The primary endpoint for this trial is progression of frailty status, considered both as a category (non-frail, pre-frail, and frail) and score (0–5). Frailty progression is defined as worsening of frailty category and/or score. Primary inference is to be drawn based on intention to treat analyses, which will include all subjects randomized in the study. The impact of metformin treatment on the progression of frailty status (non-frail, pre-frail, frail) over time will be assessed with a repeated measures mixed-effects multinomial logistic model that models the log-odds of adjacent frailty statuses in terms of an indicator of metformin use, time, and the metformin use by time interaction. The possibility of nonlinear trends of the log-odds of frailty statuses will be assessed by including an additional time factor (eg, time squared) and its interaction with metformin use. The best fitting model will be used for inference. The impact of metformin on the log-odds of frailty status progression will be revealed by a significant time by treatment interaction, indicating that the temporal change in the likelihood of frailty progression differs between the metformin arm and the placebo arm. Given that the randomization was not based on all covariates due to feasibility, there is potential to improve estimation efficiency of the treatment effect by adjusting for age, sex, body mass index, sarcopenia, blood pressure, fasting glucose, HbA1c, insulin sensitivity, lipids, inflammation, insulin signaling, and diabetes conversion. Thus, both unadjusted and adjusted analyses will be conducted. All analyses are to be carried out using SAS PROC NLMIXED. No interim monitoring of study outcomes will occur unless prompted by the reviewing DSMB.

Secondary Causal Modeling

Average complier effect will be calculated to inform the metformin effect under full compliance to treatment assignments. A per protocol analysis will additionally be conducted as a secondary analysis to inform the metformin effect among subjects who complete the study. Contrasting the primary intention to treat analysis with the per protocol analysis (under no unmeasured confounding and ignorability assumptions) will allow the assessment of the impact of missing data and noncompliance on the estimation of the treatment effect. Separately, we will also consider marginal structural modeling to obtain the overall metformin effect, accounting for time-varying covariates (eg, diabetes status and treatment change due to noncompliance) and the history of intermediate outcomes (eg, frailty, physical function).

Analysis of secondary outcomes

Secondary outcomes include gait speed, grip strength, 6-minute walk, Short Physical Performance Battery, body composition, and frailty as defined by a deficit accumulation index (30). Analysis for secondary outcomes will be carried out in similar manner to the analysis for the primary outcome as described earlier.

Mediators and mediation analysis

Mediators are systemic inflammation, muscle inflammation, and insulin signaling. To address the question as to whether an anti-frailty effect of metformin is mediated by its insulin-sensitizing and anti-inflammation effects, we will use the unified mediation analysis framework by Vanderweele (31). Separate mixed-effects model analyses will be conducted to assess the effect of metformin treatment on repeated measures of each insulin sensitivity and inflammation measure. In each of these analyses, the outcome (or its normal transformation) will be modeled in terms of an indicator of metformin treatment, time, treatment by time interaction, and covariates. The coefficient associated with the treatment by time interaction will inform the effect of metformin on the temporal change of each biomarker. Mixed-effects multinomial logistic model will be conducted to assess whether the odds of an increase in frailty score (or category) is associated with the change in insulin sensitivity from baseline to 24 months (ΔIS), and change in inflammation from baseline to 24 months (ΔI) by modeling the log-odds of frailty worsening in terms of metformin treatment, time, ΔIS, ΔI, (metformin) Treatment × Time, and covariates. The Treatment ×Time × ΔIS and Treatment × Time × ΔI interactions will be examined to assess whether metformin attenuates the effect of ΔIS and ΔI: this effect will be revealed by a significant coefficient associated with Treatment × Time × ΔIS or Treatment × Time × ΔI. If some ΔIS and ΔI measures are collinear, then we will model the odds of frailty worsening in terms of ΔIS and ΔI separately.

Power analysis

A prior study has shown that 44%–53% of nondisabled, community-dwelling older adults will worsen in frailty status over an 18-month period (32). However, the baseline prevalence of frailty in this cohort was somewhat higher (25.7%) than what has been observed in other cohort studies (33). We have unpublished data from community-dwelling older adults in the San Antonio Longitudinal Study of Aging demonstrating that 53% of non-frail individuals worsened in the frailty score by 1 point or more over 18 months (S. E. Espinoza, M.D. and H. P. Hazuda, Ph.D., unpublished data, 2019). The baseline frailty prevalence in this cohort was 9.2%. Therefore, 2 years is likely sufficient follow-up time for this study, particularly in a cohort of individuals with prediabetes who are expected to have accelerated progression of sarcopenia and frailty (16,34).

With n = 60 subjects completing the study per group and α = 0.05, this study will attain a power of 82.2% for testing H_o:p_M = p_C versus H₁:p_M ≠ p_C, where p_M and p_C are the probabilities of worsening in frailty among subjects randomized to metformin and control, if p_M = 0.28 and p_C = 0.53 [PASS, version 15; NCSS, Kaysville, UT]. These parameter specifications are consistent with prior data that 53% of non-frail older adults were observed to worsen in frailty in the Sacramento Area Latino Study on Aging (SALSA) study and prior work showing that that 44%–53% will worsen in frailty status over an 18-month period (30). In addition, the overall effect size was projected by incorporating the possibility of noncompliance as well as treatment group switching due to diabetes conversion. Assuming 5% lost to follow-up based on the pilot phase of the study, 64 subjects per group will be enrolled to achieve n = 60 subjects completing the study.

A month clinical trial of insulin sensitizers versus placebo in patients with impaired glucose fasting glucose or diabetes found decreased tumor necrosis factor α in treated patients (effect size = 1.5) (35). Thus we assume the same or larger decrease in tumor necrosis factor α associated with metformin treatment. Our calculation showed that less than 60 completers in each group, 6 repeated measures, autocorrelation 0.5, and overall Type I error of 0.05, we will have at least 95% power to detect a group difference with regard to mean of tumor necrosis factor α, and at least 95% power to detect an increase in tumor necrosis factor α and frailty worsening in the combined cohort. The Diabetes Prevention Program demonstrated improved fasting glucose in metformin-treated adults (decrease of 4.1 mg/dL) with prediabetes and increased fasting glucose (by 0.4 mg/dL) in placebo group (36). Assuming the same metformin effect on fasting glucose in this study and an autocorrelation of 0.5 among 6 repeated measures of fasting glucose, this study (60 completers per arm) will attain a power of 80% for detecting a 4.5 mg/dL decrease of fasting glucose by metformin treatment conditioned on Type I error 0.05 under two-sided testing.

Conclusion

Frailty is a major health care problem with devastating consequences on the U.S. aging population. A recent systematic review revealed 14 prior studies that have examined the effectiveness of exercise, nutritional interventions, and geriatric assessment for frailty prevention (37). The Lifestyle Interventions and Independence for Elders (LIFE) study examined the effect of physical activity on physical performance in older sedentary adults with physical limitations and who were at risk for future disability (38,39). They found that a moderate-intensity program comprised aerobic, resistance, and flexibility training resulted in an almost 20% reduced risk of mobility disability (hazard ratio = 0.82, 95% confidence interval: 0.69–0.98, p = .03) compared to health education as a control group. Analyses examining frailty as an outcome from this study’s intervention gave mixed results. Initial data from the pilot phase of the LIFE study (LIFE-P) suggested that frailty measured by Fried criteria was reduced in the intervention group (40). In the Phase 3 LIFE study, which was conducted later, frailty was assessed using an abbreviated frailty measurement, which did not show an effect of the intervention on frailty (41). Thus, although it is acknowledged that exercise can beneficially affect frailty measures and is important recommendation for older adults, unfortunately the majority of older adults in the United States remain sedentary (42), and questions remain about the feasibility of implementing these interventions broadly in a clinical setting, and whether the effects of exercise are sustainable beyond the period of training (43).

Although aging itself is well known to be a leading risk factor for several chronic diseases including frailty, aging per se has not been generally considered a risk factor that is modifiable. Recent major advances in aging biology research have led to interventions that have the potential to modify life span and health span. The finding that the mammalian target of rapamycin inhibitor extends life span in rodents (44) has spurred a rapidly growing interest in developing pharmacologic interventions to improve health span and to prevent or delay geriatric syndromes such as frailty (45). Metformin has also been identified as a potential pharmacologic intervention to modify aging, in part because of its potential modifying effects on aging pathways (46) and observational evidence suggesting that it reduces all-cause mortality and frailty (47,48).

To our knowledge, this is one of the first randomized trials designed to examine the effect of a pharmacologic intervention to prevent frailty in older adults (49,50). We propose that insulin resistance and inflammation are major contributors to frailty, and that the use of metformin to modulate insulin resistance and inflammation will prevent the onset and progression of frailty. Other ongoing studies are evaluating related outcomes. For example, ENRGISE (Enabling Reduction of Low-Grade Inflammation in Seniors) is a clinical trial currently underway that is studying whether modulating inflammation with angiotensin receptor blocker and omega-3 polyunsaturated acids improves major mobility disability. A Phase 2b trial is testing the effects of inflammation reduction via allogeneic mesenchymal cell transfusion on frail older adults (51,52). Finally, the Targeting Aging with Metformin trial, currently in planning phase, will examine whether metformin can modify aging, which will be measured via reduction in the onset of a composite endpoint of multiple age-related conditions, including cardiovascular events, cancer, dementia, and mortality (46).

Metformin is thought to lower blood glucose in part by activating AMP-activated protein kinase, a major cellular regulator of lipid and glucose metabolism (53). Although it is not entirely clear whether the potential life extension effect of metformin is directly related to its AMP-activated protein kinase activating properties, AMP-activated protein kinase can interact with the mammalian target of rapamycin pathway, and as stated earlier inhibition of mammalian target of rapamycin with rapamycin has been shown to extend life span in mice (44). Epidemiological studies indicate that diabetic subjects who took metformin had lower cancer- and all cause-related mortality rates compared with those who took other antidiabetic agents (54). We are in an exciting time in translational approaches to human aging and at the dawn of developing clinical interventions that can prevent geriatric conditions such as frailty. The results of this study may have implications for the prevention of frailty in prediabetic older adults who are at high risk for becoming frail.

Funding

This work was supported by the National Institute on Aging (NIA/NIH R01AG052697), the San Antonio Claude D. Pepper Older Americans Independence Center (NIA/NIH P30 AG044271), and the Veterans Affairs San Antonio Geriatric Research, Education, and Clinical Center.

Conflict of interest statement

The authors whose names are listed immediately below certify that they have no affiliations with or involvement in any organization or entity with any financial interest (such as honoraria; educational grants; participation in speakers’ bureaus; membership, employment, consultancies, stock ownership, or other equity interest; and expert testimony or patent-licensing arrangements) or nonfinancial interest (such as personal or professional relationships, affiliations, knowledge, or beliefs) in the subject matter or materials discussed in this manuscript (S.E.E., N.M., C.W., J.M., B.O., T.R., B.P., A.C., M.M., D.K., Y.L., V.G., T.E.J., L.C.K., and R.J.).