Initiation of New Glucose-Lowering Therapies May Act to Reduce Physical Activity Levels: Pooled Analysis From Three Randomized Trials

Recent innovations in diabetes management include the licensing of sodium–glucose cotransporter 2 inhibitors (SGLT2i) and glucagon-like peptide 1 receptor agonists (GLP-1RA) (1), which are associated with weight loss and wide-ranging cardiometabolic benefits. However, effects on physical activity are not well characterized. Change in weight is a predictor of change in physical activity (2,3), supporting a hypothesis that weight loss may cause an increase in physical activity. Conversely, the less-than-predicted weight loss with SGLT2i, the reduction in resting metabolic rate with GLP-1RA, and experimental animal studies suggest an alternative hypothesis of reduced physical activity (4–6).

We pool data from three randomized trials to undertake an exploratory analysis investigating the impact of SGLT2i or GLP-1RA therapy on accelerometer-assessed physical activity and how this compares with a weight-neutral (dipeptidyl peptidase 4 inhibitor [DPP4i]) or dietary therapy.

We pool data from three randomized trials, Diabetes Interventional Assessment of Slimming or Training to Lessen Inconspicuous Cardiovascular Dysfunction (the DIASTOLIC study) (7), the effects of Liraglutide in Young adults with type 2 Diabetes (LYDIA) study (8), and SGLT-2 Inhibitor Empagliflozin Effects on Appetite and Weight Regulation (SEESAW) (9), in people with type 2 diabetes that were conducted in the same research center with similar inclusion criteria (detailed in Supplementary Table 1). DIASTOLIC was a three-armed 12 week randomized controlled trial testing the efficacy of a low-energy meal replacement plan (∼810 kcal/day) or exercise training (excluded from this analysis) on cardiac function compared with usual care in working age adults (18–65 years) (7). LYDIA was a 26 week randomized trial investigating the effects of liraglutide (GLP-1RA) (1.8 mg, once daily) compared with sitagliptin (100 mg, once daily) on cardiac function in younger adults (18–50 years) (8). SEESAW was a 24 week four-arm placebo-controlled randomized trial investigating the effect of empagliflozin (SGLT2i) (25 mg, once daily), an energy-restricted diet (360 kcal/day reduction to match the estimated energy deficit with empagliflozin), or their combination in adults (ages 30–75 years) (9).

All groups received general lifestyle advice including recommendations to become more active. No additional structured support for physical activity promotion was provided.

Physical activity was measured by accelerometer (ActiGraph GT3X+) worn on a waistband (in the right anterior axillary line) for seven consecutive days during waking hours. Acceleration data were integrated into 60 s epochs. At least 3 days’ valid wear (≥10 h of data per day) were required. Nonwear time was determined by ≥1 h of consecutive zero counts. ActiGraph accelerometers provide a valid measure of steps taken (10,11), which was included as a measure of total physical activity. Freedson cut points were used to classified time spent sedentary and in moderate- to vigorous-intensity physical activity (MVPA) (12).

Across all studies, body weight and height (for BMI) were captured according to the same standard operating procedures. HbA_1c was analyzed in the same quality-controlled clinical laboratory. Sex, ethnicity, and age were captured by self-report.

Data were included from control (DIASTOLIC, SEESAW), diet (DIASTOLIC, SEESAW), liraglutide (LYDIA), sitagliptin (LYDIA), empagliflozin (SEESAW), and combined (SEESAW) arms, covering 198 participants. Of these, 148 (75%) participants had valid baseline and follow-up physical activity data.

Data were pooled and analyzed with generalized estimating equations, taking account of potential clustering by study. Change from baseline in measures of physical activity were investigated by treatment group, with adjustment for change in wear time, baseline value, baseline BMI, diabetes duration, age, sex, and ethnicity. Change values were compared with the pooled effect observed in control groups. Models were further adjusted for change in body weight to investigate whether any reported differences were independent of weight loss. Data are reported as marginal means (95% CI) with P < 0.05 denoting significance.

This analysis included 148 individuals (median age 52 years, BMI 34.0 kg/m², and ambulatory activity 6,075 steps/day; 41% women). Participant characteristics are presented by study (Supplementary Table 2) and by inclusion/exclusion status (Supplementary Table 3).

The pooled control (mean weight change −1.4 kg; 95% CI −2.0 to −0.9), energy-restricted diet (−2.1 kg; −2.7 to −1.5), meal replacement plan (−13.4 kg; −14.6 to −11.9), empagliflozin (−1.8 kg; −3.2 to −0.4), liraglutide (−4.1 kg; −4.8 to −3.5), and combined empagliflozin and diet intervention (−5.1 kg; −6.5 to −3.6) groups lost weight, with weight loss of the meal replacement plan, liraglutide, and combined empagliflozin and diet intervention groups different from that of control (Supplementary Fig. 1). There was no change in body weight with sitagliptin. The meal replacement plan and all drug therapies resulted in reduced HbA_1c (Supplementary Fig. 1).

Participants in the liraglutide (mean change −1,144 steps/day; 95% CI −2,069 to −220 [17.3% reduction from baseline]), empagliflozin (−1,132 steps/day; −1,739 to −524 [19.0% reduction]), and sitagliptin (−852 steps/day; −1,625 to −78 [12.0% reduction]) groups all had reduced total physical activity, with reductions different from those under control conditions (Fig. 1). There was no change in the combined empagliflozin and diet intervention. The meal replacement plan increased physical activity (654 steps/day; 117–1,192 [10.3% increase]). The distribution of change values is presented by group in Supplementary Fig. 2.

Results for MVPA followed a similar pattern (Fig. 1), with decreases of 8–11 min/day reported across liraglutide, empagliflozin, and sitagliptin groups. The same therapies resulted in increased sedentary time, with the increase of the liraglutide group (mean increase 37.9 min/day; 95% CI 28.6, 47.2) different from that for the control group (Fig. 1).

The differences in comparisons with the control group for decreased total physical activity and MVPA with liraglutide, empagliflozin, and sitagliptin were maintained after adjustment for change in weight. In contrast, the increase in total physical activity with the meal replacement plan was attenuated.

The initiation of GLP-1RA, SGLT2i, or DPP-4i was associated with reductions in physical activity of ∼1,000 steps/day or 10 min/day of MVPA. This was in contrast to dietary interventions, which resulted in no difference in or increased physical activity.

Physical activity is a vital component of overall health (13), with 500 steps/day or 5 min of moderate-intensity activity suggested to provide the minimum clinical important difference (14). Therefore, the findings of this study are clinically relevant and warrant further investigation.

Of note, the reductions in physical activity were similar across GLP-1RA, SGLT2i, and DPP-4i. Further, the reductions to physical activity were independent of any co-occurring weight loss. These findings suggest mechanisms other than weight loss or the specific targets of the therapies. It is possible that conscious or unconscious behavioral responses to the initiation of any glucose-lowering therapy may lead to reduced physical activity. This is consistent with recent research showing that the initiation of antihypertensive or lipid-lowering medication is associated with reduced physical activity levels (15), which supports a hypothesis that the prescription of new preventative mediation leads to false reassurance and substitution with less healthy lifestyle practices in some individuals. That decreases in physical activity were attenuated when SGLT2i therapy was combined with a dietary intervention further supports this behavioral hypothesis.

These findings have implications for understanding the less-than-predicted weight loss shown to occur with SGLT2i therapies (4). Although some biological mechanisms have been postulated (4), our data suggest that reductions in physical activity may also need to be considered.

This is a post hoc analysis of studies that were not designed to address the research question and should be interpreted accordingly. Nevertheless, the aim was to highlight a novel hypothesis using data from randomized trials, which limited the potential for bias.

Further research is needed to confirm these findings. One way of achieving this could be through embedding physical activity as a core outcome within phase 3 clinical trials to more widely assess the impact of initiating glucose-lowering therapies across different settings and therapies. If findings are replicated, targeted behavioral interventions designed to reverse the reduction in physical activity will be required.

Funding. This study was in part supported by the National Institute for Health Research (NIHR) Leicester Biomedical Research Centre. DIASTOLIC was funded by a grant from NIHR (CDF 2014-07-045).

The views expressed are those of the authors and not necessarily those of the National Health Service, the NIHR, or the Department of Health and Social Care.

Duality of Interest. SEESAW was funded by an investigator-initiated grant from Boehringer Ingelheim. LYDIA was funded by an investigator-initiated grant from Novo Nordisk. T.Y. and G.P.M. report funding in the form of an investigator-initiated trial from AstraZeneca and have acted as co-investigators on obesity-related research supported by Novo Nordisk. J.A.S. and E.A. report funding in the form of an investigator-initiated trial from AstraZeneca. D.R.W. has received honoraria as a speaker for AstraZeneca, Sanofi, and Lilly and has received research funding from Novo Nordisk. K.K. has acted as consultant, advisory board member, or speaker for Abbott, Amgen, AstraZeneca, Bayer, Napp Pharmaceuticals, Lilly, Merck Sharp & Dohme, Novartis, Novo Nordisk, Roche, Berlin-Chemie AG/Menarini Group, Sanofi, Servier, Boehringer Ingelheim, EACME grants from Boehringer Ingelheim, AstraZeneca, Novartis, Novo Nordisk, Sanofi, Lilly, and Merck Sharp & Dohme. M.J.D. has acted as consultant, advisory board member, and speaker for Novo Nordisk, Sanofi, Lilly, and Boehringer Ingelheim; advisory board member and speaker for AstraZeneca; advisory board member for Janssen, Lexicon, Servier, and Gilead Sciences; and speaker for Napp Pharmaceuticals, Mitsubishi Tanabe Pharma, and Takeda Pharmaceuticals International. She has received grants in support of investigator and investigator-initiated trials from Novo Nordisk, Sanofi, Lilly, Boehringer Ingelheim, AstraZeneca, and Janssen. D.J.S. has received research funding from Boehringer Ingelheim. No other potential conflicts of interest relevant to this article were reported.

The funders of SEESAW, LYDIA, and DIASTOLIC had no role in collection, analysis, or interpretation of data or writing of the manuscript.

Author Contributions. T.Y. drafted the manuscript and analyzed study data. T.Y., J.A.S., and J.A.K. contributed to the study hypothesis and analysis plan. J.H. and C.L.E. contributed to the collection of accelerometer data. C.L.E. processed the accelerometer data, and both contributed to data interpretation. All authors contributed to the design of the original studies. All authors revised the manuscript for important intellectual content. T.Y. 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.