Combination of GLP‐1 receptor agonists and behavioural treatment in type 2 diabetes elicits synergistic effects on body weight: A retrospective cohort study

Abstract Aims Intensification of type 2 diabetes (T2DM) treatment with GLP‐1 receptor agonists (GLP‐1RAs) promotes weight loss. We aimed to determine the synergistic effect of behavioural programmes on body weight on top of GLP‐1RA treatment. Materials and methods We retrospectively analysed the time course of 328 individuals with T2DM starting GLP‐1RA treatment because of insufficient metabolic control. In 29, a structured programme of elementary nutritional counselling was also implemented (elementary nutritional education [ENE]‐5 group sessions), whereas 53 entered a programme of cognitive‐behavioural treatment (CBT‐12 group sessions). Both programmes were completed within 6 months of switching to GLP‐1RAs. Data of body weight and metabolic control were followed up to 2 years as part of regular follow‐up. Weight loss targets (≥10% and ≥5%) and metabolic target (HbA1c < 7%) were analysed by Cox regression model in comparison with standard care (SC, N = 244). Results Body weight remarkably decreased following both behavioural programmes, with significant differences compared with SC at 2 years (CBT, 8.5 ± 5.9% vs 6.3 ± 6.9 in ENE and only 3.1 ± 5.7 in SC; P < 0.001 and P = 0.045 vs CBT and ENE, respectively). The 10% weight loss was achieved and maintained in approximately 30% of cases during follow‐up, and an additional 35% of cases lost between 5% and 10%. Data were consistent between behavioural programmes, after adjustment for confounders, including initial body weight (logreg Mantel‐Cox: ENE vs SC, P < 0.01; CBT vs SC, P < 0.001). No differences in metabolic control were detected between groups. Conclusions Initiation of GLP‐1RA treatment provides an opportunity for addressing patients' needs of weight control. By producing initial weight loss, patients' motivation and self‐efficacy are expected to increase and adherence to long‐term lifestyle changes might be more easily attained.


| INTRODUC TI ON
The treatment of type 2 diabetes mellitus (T2DM) is facing totally new paradigms compared with the standards of care published only a few years ago. 1 While metformin is maintained as first-line therapy, second line has moved from basal insulin or insulin-secreting agents (sulphonylureas and glinides) to the three classes of dipeptidyl-peptidase-4 inhibitors (DPP4-Is), sodium-glucose cotransporter-2 inhibitors (SGLT2-Is) and glucagon-like peptide-1 receptor agonists (GLP-1RAs). 2 The new treatments have several advantages compared with the old classes; in particular, they do not increase the risk of hypoglycaemia, 3 and, limited to SGLT2-Is and GLP-1RAs, reduce cardiovascular risk and body weight. Cardioprotection has been demonstrated both in prospective trials 4,5 and in real-world observational studies, [6][7][8] and makes SGLT2-Is and GLP-1RAs. The loss of body weight is particularly large during GLP-1RA treatment, 9 and these drugs are the treatment of choice in the presence of obesity. 10 A drug of the class has been approved for treatment in obesity independently of diabetes (liraglutide) 11 and a novel drug (semaglutide) might be a likely candidate too, considering the impressive results on body weight. 12 The treatment of obesity remains an area of intense research, but very few drugs are approved by international agencies. 13 Weight loss remains a pivotal issue for an effective control of T2DM, 14 but the results achieved by behavioural treatment, both in the presence and in the absence of diabetes, are generally difficult to maintain, 15 unless patients are engaged into intense programmes aimed at calorie restriction and physical activity. 16 Long-term results are driven by motivation, a construct which is maintained by self-efficacy, that is the evidence that some results may be effectively achieved, and any amount of initial weight loss may strengthen adherence to lifestyle changes. 17 Thus, the beneficial effects of novel glucose-lowering drugs on body weight might start a virtuous circle, comparable to the synergistic effect reported by Wadden et al 18 for sibutramine, when combined with cognitive-behavioural therapy (CBT) in the treatment of obesity.
In our unit, two graded programmes of lifestyle changes are operative in patients with obesity and/or T2DM. 19 These programmes are routinely offered to patients with T2DM, either at diagnosis or at times of treatment intensification for insufficient metabolic control. We retrospectively analysed the effects of GLP-1RAs on body weight and metabolic control in T2DM subjects, according to their participation in programmes aimed at lifestyle changes at times of treatment switch. The underlying hypothesis was that GLP-1RA treatment might produce a much larger improvement whenever associated with behavioural support.

| Patients
We retrospectively analysed all T2DM cases who were prescribed a GLP-1RA agonist in our outpatient unit since its initial approval in the Italian market (February 2008). According to initial rules dictated by the Italian Medicines Agency (AIFA), they had to fulfil definite criteria, to undergo an online monitoring system (closed on August 2010) and to adhere to follow-up visits, initially planned after 3-4 months and after 6-8 months, and every 6 months thereafter. After exclusion of cases who moved from our unit to other diabetes centres (N = 23), cases who stopped treatment for adverse events (N = 12), cases who had started treatment by <6 months (N = 48) and cases who initiated insulin treatment during follow-up-to avoid the counteracting effect on insulin on body weight (N = 6)-we could retrieve 328 cases; in this cohort, 1-year and 2-year follow-up data were available for 264 and 162 cases, respectively. Their baseline values are reported in Table 1. According to our procedures, following motivational interviewing, all patients first seen inside the centre are invited to take part in structured behavioural programmes, modulated according to the severity of their weight excess and unhealthy eating pattern (see below). 19 At any follow-up visit, patients receive motivational reinforcement on lifestyle changes and adherence to healthy diet and habitual physical activity, and the proposal to enter a structured programme is repeated in the presence of insufficient metabolic control. The final goal of behavioural treatment is weight loss, and two different targets were considered: weight loss ≥5% and ≥10% initial body weight.

| Behavioural programmes
In addition to nutritional counselling by the physicians or dieticians

| Measures
At baseline and at any control visits, body weight and height, waist circumference and blood pressure were available, 22

| Statistical analysis
All data were initially anonymised, implemented on an Excel database and analysed using StatView 5.0™ statistical package (SAS Institute Inc) and SPSS v.20 (IBM Company). A descriptive analysis was initially carried out by computing mean and standard deviation, or median and interquartile range for non-normally distributed data, TA B L E 1 Baseline characteristics of the population with type 2 diabetes at first prescription of a GLP-1RA, grouped according to treatment programmes (means ± SD, prevalence [95% confidence interval] or median [interquartile range])

| Effects of treatment on body weight and target reach
GLP-1RA treatment was associated with a systematic reduction in body weight in all cohorts (Figure 1). Body weight loss was much larger in subjects entering the superimposed behavioural treatment, and particularly in the CBT cohort, where it was on average 6.5 ± 5.8% in subjects reaching the 1-year observation period and 8.5 ± 5.9% after 2 years (vs 6.3 ± 6.9% in ENE and only 3.1 ± 5.7% in SC after 2 years; P < 0.001 and P = 0.045 vs CBT and ENE, respectively).
At 1-year follow-up, the 10% weight loss target was achieved in only 6% of cases in SC, vs 14% and 26% in the ENE and CBT groups, respectively (P < 0.001); after 2 years, only 8% of cases treated by SC achieved the 10% weight loss, vs 27% in ENE and 30% in CBT (P < 0.05 for both) (Figure 2).
In a multivariate Cox proportional-hazards model, the probability to achieve the 5% weight loss was significantly increased neither by adding behavioural treatment to the initiation of GLP-1RA treatment, nor by any of the two treatment programmes (Table 2). On the contrary, any behavioural treatment more than doubled the probability to achieve 10% weight loss (

| Effects on metabolic control
HbA1c decreased rapidly in response to GLP-1RA treatment, on average by 0.98 ± 1.22% and 0.94 ± 1.27% after 1 and 2 years, respectively, without differences between treatment programmes. After 1 year, the standard HbA1c target of 7% was achieved in 36% of cases treated by GLP-1RA + SC, in 38% following GLP-1RA + ENE and 40% following GLP-1RA + CBT (not different) and no differences were also observed after 2 years (on average, 53%).
Changes in HbA1c were also driven by the different use of other glucose-lowering drugs. The use of sulphonylureas/repaglinide declined remarkably in the three cohorts, as was the case of basal insulin and pioglitazone, with minor differences between groups (Table 3).
The effects on metabolic control are more clearly demonstrated in the cohort observed during the whole period of study (Table S1).
In these patients, a systematic decline in HbA1c was observed, without differences in relation to behavioural treatment, accompanied by minor changes in lipid levels. Systolic blood pressure did not vary, whereas diastolic pressure decreased significantly in all groups.

| D ISCUSS I ON
The report shows how much a behavioural programme may help The present study was addressed to test the opposite mechanism, that is, the effectiveness of weight loss programmes to exploit the effects of GLP-1RA on body weight. Behaviour-induced weight loss strongly depends on adherence to and persistence in healthy, calorie-restricted diet and physical activity, 31 in turn favoured by self-efficacy. 32 The initial weight loss induced by GLP-1RA treatment is likely to strengthen self-efficacy, enhancing the effects of behavioural programmes. In keeping with this hypothesis, body weight continued to decline on average in patients participating in behavioural programmes, and the percentage of cases reaching the challenging 10% weight loss target increased progressively in the 2-year observation period to 29%, with an additional 36% achieving the 5% target. Notably, the probability to reach the weight loss target of 5% was rather high also in subjects on SC (24%), which explain the lack of statistical difference when compared to BT. Per protocol, two behavioural programmes of different intensity are offered to patients with T2DM in poor metabolic control at our department; they were reported to facilitate metabolic control as well as weight loss and to retard insulin treatment. 19 A reanalysis of those historical data showed that ENE produced a weight loss of 4% at 1 year (vs 14% in the present series, P = 0.051) and 9% at 2 years (vs 27%, P < 0.05), whereas CBT resulted in a weight loss of 18% and 20%, respectively. 19 These figures are remarkably lower than the percentage of cases at target by the simultaneous GLP-1RA initiation, in keeping with an additive effect.
The study has limits, which should be discussed. First, the report is an audit of patients receiving treatment in a single department, not a randomized controlled study, and is prone to several biases in spite of multiple adjustments. Second, the per-protocol analysis was based on patients' adherence to the behavioural programmes, proposed to patients on the basis of their poor metabolic control and the degree of obesity.
These cohorts might include subjects more motivated to weight loss, thus biasing the comparison with SC. However, the analysis of the historical cohort indicates that also in comparison with equally motivated patients GLP-1RA treatment may significantly enhance the beneficial effects on body weight. Several studies identified early weight loss as predictor of long-term outcome 33,34 ; accordingly, GLP-1RA-associated  35,36 Accordingly, the use of other glucose-lowering drugs was also modified, possibly blurring the final effects in the retrospective analysis.
Finally, we did not systematically measure food intake and physical activity to associate weight loss with lifestyle changes.
In conclusion, the initiation of GLP-1RA treatment might be a pivotal starting point to accompany motivated subjects with T2DM towards lifestyle changes. Both gastrointestinal symptoms and the possible effects of these drugs on the central mechanism(s) of satiety might start and enhance a virtuous circle leading to large and persistent weight loss when accompanied by behavioural programmes. A randomized controlled trial has been recently set up to support these retrospective data.

CO N FLI C T O F I NTE R E S T S
All authors declare no conflict of interest in relation to the material presented in this study.

AUTH O R CO NTR I B UTI O N S
GM, LM and MLP conceived the study; LM, SC, MTC, AM collected the data; MLP, SC and GM were involved in statistical analyses and drafted the manuscript; all authors contributed substantially to its revision and agreed to be accountable for all the aspects of the work; and GM takes responsibility for the paper as a whole.

E TH I C A L A PPROVA L
The anonymous analysis represents an internal audit, and does not require signed consent by patients.

DATA AVA I L A B I L I T Y
The whole set of data is available upon request. TA B L E 3 Treatment with glucose-lowering drug before switching to GLP-1RAs (Pre) and in addition to GLP-1RAs during the observation study (basal, 1 y, 2 y) The analysis indicates heterogeneity of use among groups (Chi 2 , P value <0.05).