Semaglutide Treatment of Antipsychotic-Treated Patients With Schizophrenia, Prediabetes, and Obesity

Key Points

Question

Can the glucagon-like peptide-1 receptor agonist semaglutide safely reduce glucose levels and body weight in patients with schizophrenia, prediabetes, and obesity?

Findings

In this placebo-controlled randomized clinical trial of 154 patients with schizophrenia, overweight or obesity, and prediabetes, 30 weeks of semaglutide, 1.0 mg/week, significantly lowered glycosylated hemoglobin A_1c. Moreover, semaglutide reduced body weight by a mean value of 9.21 kg and improved physical quality of life without exerting negative mental health consequences.

Meaning

Semaglutide is safe and effective for lowering blood glucose and weight in patients with schizophrenia who also experience overweight or obesity and prediabetes.

Abstract

Importance

Patients with schizophrenia have reduced life expectancy due to cardiovascular disease and obesity-related type 2 diabetes, exacerbated by second-generation antipsychotic (SGA) medication. Existing interventions have shown limited effect.

Objectives

To assess the effect of the once-weekly glucagon-like peptide-1 receptor agonist semaglutide in SGA-treated adults (aged 18-60 years) with schizophrenia, prediabetes (glycosylated hemoglobin A_1c [HbA_1c], 5.7%-6.4% of total hemoglobin) (to convert HbA_1c from percentage of total hemoglobin to mmol/mol, use the following formula: (HbA_1c % − 2.152)/0.09148), and overweight or obesity (body mass index [BMI], calculated as weight in kilograms divided by height in meters squared, ≥27).

Design, Setting, and Participants

This placebo-controlled, double-blinded randomized clinical trial was conducted from January 2022 to May 2024, with 30 weeks of follow-up, among regional community-based mental health services in 2 regions of Denmark (Region of Southern Denmark and Region of Zealand). SGA-treated patients with schizophrenia, prediabetes, and overweight or obesity were randomized to semaglutide or placebo. Data analysis was completed from May 2024 to January 2025.

Intervention

Once-weekly subcutaneous semaglutide or placebo for 30 weeks; semaglutide was titrated up to 1.0 mg/week over 8 weeks.

Main Outcomes and Measures

The primary outcome was change in HbA_1c. Secondary end points included changes in body weight, schizophrenia symptoms based on Positive and Negative Syndrome Scale 6 (PANSS-6) score, and physical and mental quality of life (QoL) (assessed via the 36-item Short Form Survey, version 2 [SF-36v2]).

Results

A total of 154 patients were recruited and randomized 1:1 to semaglutide or placebo (87 female participants (56.5%); mean [SD] age, 38.3 [10.7] years). Of 154 randomized patients, 141 (91.5%) completed the trial—74 of 77 patients randomized to semaglutide (96%) and 67 of 77 randomized to placebo (87%). Semaglutide reduced HbA_1c by 0.46% of total hemoglobin (95% CI, −0.56% to −0.36%) and body weight by 9.21 kg (95% CI, −11.68 to −6.75). An HbA_1c less than 5.7% of total hemoglobin was achieved in 81% vs 19% of patients treated with semaglutide and placebo, respectively (P < .001); improvements in high-density cholesterol by 10.81 mg/dL (95% CI, 2.70-18.53; P = .007) and triglycerides by −29.20 mg/dL (95% CI, −55.75 to 2.65; P = .03) (to convert to millimoles per liter, multiply by 0.0113) were also observed. Finally, semaglutide improved physical QoL by 3.75 points on the SF-36v2 (95% CI, 1.52-5.98; P = .001) but had no significant effect on mental QoL scores or PANSS-6 score. Gastrointestinal symptoms were more frequent in semaglutide-treated patients. A few semaglutide-treated patients were hospitalized more frequently than observed in the placebo-treated group, but the number of serious adverse effects did not differ between groups.

Conclusions and Relevance

In this multicenter, double-blinded randomized clinical trial, 30 weeks of administration of semaglutide, up to 1.0 mg/week, was safe, lowered blood glucose (as measured by HbA_1c) and weight, and improved physical QoL in SGA-treated patients with schizophrenia, prediabetes, and obesity without worsening mental health.

Trial Registration

ClinicalTrials.gov Identifier: NCT05193578

This multicenter randomized clinical trial assesses the effect of once-weekly administration of the glucagon-like peptide-1 receptor agonist semaglutide in adults treated with second-generation antipsychotics with schizophrenia, prediabetes, and overweight or obesity.

Introduction

Patients with schizophrenia spectrum disorders (SZ) have a life expectancy reduced by 15 years compared to the general population.¹ This may be attributed to modifiable causes, such as unhealthy diet, physical inactivity, smoking, and excessive use of alcohol and drugs, which significantly elevate the risk of metabolic and cardiovascular comorbidities. In addition, nonmodifiable risk factors, such as genetics, may also play a role, as there is evidence of shared genetic vulnerability between SZ and type 2 diabetes (T2D).^1,2,3 Additionally, patients with SZ face significant barriers in accessing the somatic health care system, leading to delayed diagnoses and inadequate care.^2,3,4 Consequently, all-cause mortality rates are more than 3-fold higher in patients with SZ compared to the general population, with a 3-fold increased risk of death from T2D and a 2-fold increased risk of death from cardiovascular disease (CVD).^3,5

The metabolic consequences of an unhealthy lifestyle can be worsened by the prescription of second-generation antipsychotic (SGA) medication, which rapidly increases appetite and weight.^4,6,7,8,9 This weight gain may compromise quality of life (QoL) and lead to discontinuation of SGA treatment.^10,11 Additionally, SGAs may cause impaired glucose tolerance, T2D, hyperlipidemia, and CVD.^4,12

Weight loss interventions may improve adherence to SGA treatment. In this context, glucagon-like peptide-1 receptor agonists (GLP-1 RAs) are promising, as this class of drug can suppress appetite and correct several of the pathophysiological effects of prediabetes.¹³ Three randomized clinical trials (RCTs) have tested GLP-1 RAs in nondiabetic SZ.^14,15,16 The largest trial found that 16 weeks of treatment with liraglutide improved glucose tolerance and caused a small weight reduction.¹⁶ This observation aligns with the much larger SCALE trial, where 56 weeks of liraglutide reduced T2D incidence and improved diabetes markers, especially in those with prediabetes.¹⁷ However, the SCALE trial excluded patients with severe psychiatric disorders and patients using medications causing clinically significant weight gain.¹⁷

Given the metabolic challenges of patients with SZ, we conducted an RCT comparing 30 weeks of semaglutide (Ozempic [Novo Nordisk]) treatment, up to 1.0 mg/week, to placebo. The study assessed feasibility and effects on glycosylated hemoglobin A_1c (HbA_1c, primary outcome), as well as weight, blood pressure, metabolic variables, QoL, and Positive and Negative Syndrome Scale 6 (PANSS-6) score in SGA-treated patients with SZ with prediabetes and overweight or obesity. Of note, our trial was designed prior to the approval of semaglutide in doses up to 2.4 mg weekly (Wegovy [Novo Nordisk]) for obesity treatment, and therefore, we did not aim for a higher dose than the one approved for T2D management at that time. Furthermore, patients with T2D were excluded, as they were already eligible to receive semaglutide through routine clinical care without trial participation.

Methods

Study Design

From January 2022 to May 2024, we conducted an investigator-initiated, double-blinded, parallel-group, superiority, multicenter, longitudinal RCT comparing 30 weeks of semaglutide injections vs placebo in the Danish Regions of Southern Denmark and Zealand. The trial received approval from the Danish Medicines Agency (2020110216) and The Regional Committees on Health Research Ethics for Southern Denmark (S-20200182) and followed Consolidated Standards of Reporting Trials (CONSORT) reporting guidelines. The trial protocol is available in Supplement 1.

Participants

The study enrolled patients diagnosed with either schizophrenia, schizotypal disorder, or schizoaffective disorder according to the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10; diagnosis codes F20.X, F21.X, or F25.X). All patients received SGA treatment and were between 18 and 60 years of age. Patients were enrolled from publicly funded, tax-financed, regional community-based mental health services, private psychiatry practices, or general practitioners. Recruited patients lived in 2 geographically distinct areas in Denmark and lived in cities, villages, and in rural areas. Eligible patients were required to be stable receiving SGA treatment for at least 6 months and without introduction of new antipsychotic drugs. Furthermore, they were required to have prediabetes (HbA_1c level between 5.7% and 6.4% of total hemoglobin, equivalent to 39-47 mmol/mol [to convert HbA_1c from percentage of total hemoglobin to mmol/mol, use the following formula: (HbA_1c % − 2.152)/0.09148]) and a body mass index (BMI, calculated as weight in kilograms divided by height in meters squared) of 27 or higher. It should be stressed that Ozempic (ie, semaglutide up to 1.0 mg/weekly) does not have an approved indication in this patient population experiencing SZ, prediabetes, and overweight or obesity.

Key exclusion criteria included an HbA_1c of 6.5% or higher, current treatment with antidiabetic medication other than metformin, or the presence of serious somatic illness. A comprehensive list of all relevant inclusion and exclusion criteria is listed in eTable 1 in Supplement 2. Sex was self-reported, with the options of male and female. Most participants were White (144 participants), with 10 participants categorized as being of other ethnic origins by the assessor during the inclusion procedure. Of these, 7 were of Asian descent and 3 of African origin. Participants gave oral informed consent prior to giving written informed consent.

We included participants receiving metformin treatment in accordance with the UK National Institute for Health and Care Excellence (NICE) guidelines, which recommend metformin as a first-line pharmacological treatment for managing weight gain and metabolic disturbances in individuals with SZ, particularly when lifestyle interventions alone are insufficient. NICE guidelines further advise that metformin should be considered early, especially in patients at high risk, such as those who have experienced significant weight gain associated with antipsychotic treatment. We did not provide any specific lifestyle intervention to the participants, as such measures are routinely addressed during their visits to general practitioners and psychiatric care units.

Randomization and Masking

Block randomization with blocks of 4 or 6 randomization numbers was used. The randomization list was provided by the manufacturer of semaglutide and placebo (Novo Nordisk A/S). Participants were assigned a number upon entering the study. The process was carried out using REDCap and the pharmacy at Odense University Hospital, with randomization executed in a continuous fashion without the use of any stratification variables. Each randomization number corresponded to a prepackaged set of study medication or placebo for the entire study period. All participants and study personnel were blinded throughout the study; only the pharmacy had access to the randomization list. Blinding was maintained through the following measures: the trial drug and placebo were visually indistinguishable; medication packs were labeled with participant numbers rather than names; unblinded staff responsible for drug allocation had no contact with blinded study personnel; and all staff received thorough training on blinding procedures. Unblinding occurred only after all primary and secondary end points had been assessed, except in the case of a female participant, who died of natural causes at home. Unblinding revealed that this participant had received placebo.

Procedures

Apart from the initial enrollment, the study was designed as a fully home-based intervention. Administration or training of eligible individuals in self-injection of study drug, as well as all data collection, was conducted in participants’ homes by mobile study nurses, ensuring optimal adherence and minimizing discomfort. As some patients demonstrated the ability to adhere to treatment and self-inject, the frequency of home visits was adjusted on an individual basis, with weekly visits reduced to monthly visits for data collection when appropriate. Research personnel transported all necessary equipment to participants’ homes for each visit. Participants continued to receive management as usual at the psychiatric clinics from which they were recruited.

Participants received semaglutide (1.34 mg/mL) or placebo for 30 weeks, with subcutaneous injections once weekly. The titration period was 8 weeks, starting with 0.25 mg weekly for 4 weeks, increasing to 0.50 mg weekly from week 5 to week 8, and reaching 1.0 mg or the highest tolerated weekly dose at week 9 until the end of the intervention. Placebo-treated patients were uptitrated similarly.

Overnight fasting blood samples were collected at baseline and after 30 weeks of treatment and used for assessment of HbA_1c, triglycerides, cholesterol, (ie, high-density lipoprotein [HDL] and low-density lipoprotein [LDL]), as well as glucose, C-peptide, and insulin. In addition, we registered BMI, waist circumference, blood pressure, psychotic symptoms, and QoL. At week 15, we collected nonfasting samples for determination of HbA_1c and safety parameters (eg, hemoglobin, liver and kidney function tests, and pancreatic amylase).

Psychiatric symptoms were assessed using the PANSS-6.¹⁸ To optimize interrater reliability and validity, all staff conducting PANSS-6 interviews underwent comprehensive group-based training together with the certified PANSS instructor (S.D.). Initially, interviews were conducted in pairs to ensure consistency. Random interviews were video-recorded and subsequently co-rated under the auspices of the PANSS instructor (S.D.). Team ratings were systematically logged, and a semistructured interview guide, the Simplified Negative and Positive Symptoms Interview, was used to standardize the assessment process. Regular supervision ensured adherence to interview protocols and scoring reliability.¹⁹

QoL was assessed using the 36-item Short Form Survey, version 2 (SF-36v2).²⁰ The questionnaire rates 8 distinct health domains, which can be grouped into 2 main summary components. The Physical Component Summary (PCS) evaluates physical health, including limitations in physical activities, bodily pain, and overall physical functioning. The Mental Component Summary (MCS) evaluates mental and emotional health, addressing areas like emotional well-being and social interactions. The minimal important difference is 2 for PC and 3 for MCS.²¹ All patients followed their clinical psychiatric treatment plan with individually adjusted follow-up visits at their usual clinician or treatment facility. Some patients experienced psychotic exacerbations requiring hospitalization. In such cases, the continuation of the study drug was assessed based on the patient’s condition and required approval from both the patient and the study physicians.

Outcomes

The primary outcome, HbA_1c, was measured at Departments of Biochemistry at the local hospitals according to International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) standard. Secondary outcomes were psychotic symptoms (PANSS-6), QoL (SF-36v2), blood pressure, body weight, waist circumference, and fasting concentrations of triglyceride, cholesterol (HDL and LDL), fasting glucose, insulin, and C-peptide. All outcomes except fasting blood measures were assessed at baseline, week 15, and week 30. Fasting blood samples were collected only at baseline and week 30. The participants were contacted every week for assessment of adverse events, and data were drawn from the hospital records in case of admission.

Statistical Analysis

The sample size estimation was based on previous studies.^16,22 With a power of 90% and a 2-sided P value and α level of .05, we needed 65 participants in each arm.

Primary and secondary outcomes were analyzed using a mixed-effect model with random intercept and random slope, which was preferred over G-computation, as no time-varying exposure and confounding were present. To test whether there was a treatment effect at weeks 15 and 30, we fitted a mixed-effect model by restricted maximum likelihood estimation and used a t test based on the Kenward-Roger method for computing the degrees of freedom. All mixed-effect models included a covariate of the baseline measurements of the given outcome to handle the difference at baseline and an interaction term with time. The model fit of all mixed-effect models was assessed graphically by the normality of the random effects and residuals. If a mixed effect yielded an unsatisfactory fit, bootstrapping was used. Primary and secondary end points are presented as margins plots. For all estimates, 95% confidence intervals and corresponding 2-sided P values were presented. P < .05 was considered statistically significant. All analyses were conducted either in Stata version 18 (StataCorp) or in R using the lme4, boot, and emmeans package (R Foundation). The statistical analysis plan can be accessed at ClinicalTrials.gov identifier: NCT05193578.

Results

Trial Population and Baseline Characteristics

We assessed 402 potential participants for eligibility and randomized 154 participants—77 to the semaglutide group and 77 to placebo (Figure 1). Of 154 randomized patients, 87 (56.5%) were female, and mean (SD) age was 38.3 (10.7) years. Three participants withdrew their consent to participate after randomization but before any follow-up visits and were not included in the efficacy analysis. A further 9 participants withdrew their consent during the study: 5 due to lack of weight loss, 3 without giving cause of withdrawal, and 1 due to worsening of psychotic symptoms. One participant was lost to follow-up due to death of natural causes. Hence, 141 participants (91.5%) completed the trial (Figure 1), including 74 of the 77 participants randomized to semaglutide (3 discontinued) and 67 of the 77 randomized to placebo (10 discontinued). Participants receiving metformin treatment continued their medication during the trial. In total, 11 participants received metformin—4 were randomized to semaglutide and 7 to placebo. The randomization resulted in 2 comparable groups (Table 1), albeit the placebo group by chance contained a lower fraction of women and had statistically but hardly clinically relevant lower concentrations of creatinine and hemoglobin.

Figure 1. CONSORT Diagram, Describing Assessment for Eligibility, Inclusion, and Randomization.

Table 1. Baseline Characteristics.

Statistic	No. (%)		P value
	Active	Placebo
Age, mean (SD), y	39.1 (10.9)	37.6 (10.6)	.40
Sex
Female	35 (45)	52 (68)	.004b
Male	42 (55)	24 (31)
Smoking	48 (62)	46 (61)	.66
Alcohol consumption >14 units weekly	1 (1)	0	>.99
Schizophrenia	77 (100)	76 (100)	>.99
Duration of diagnosis, mo
6-<12	2 (3)	1 (1)	.67
12-24	5 (6)	3 (4)
>24	70 (91)	71 (93)
Missing	0	1 (1)
Antipsychotic treatment
Quetiapine	34 (44)	36 (47)	.69
Olanzapine	13 (17)	9 (12)	.37
Risperidone	2 (3)	8 (11)	.047
Ziprasidone	3 (4)	3 (4)	.99
Paliperidone	5 (6)	4 (5)	.75
Aripiprazole	23 (30)	33 (43)	.08
Clozapine	22 (29)	19 (25)	.62
Other	8 (10)	13 (17)	.23
No. of drugs
1	46 (60)	33 (43)	.08
2	29 (38)	37 (49)
≥3	2 (3)	6 (8)
Psychiatric evaluation, mean (SD)
PANSS-6 score	20.05 (4.93)	21.71 (5.75)	.06
Positive score	9.96 (3.21)	10.92 (3.53)	.08
Negative score	10.16 (3.08)	10.79 (3.43)	.24
SF-36 score, mean (SD)
Physical component	46.52 (8.79)	44.78 (8.97)	.24
Mental component	39.47 (11.49)	37.37 (10.92)	.26
Clinical characteristics, mean (SD)
Body weight, kg	122.28 (25.92)	119.66 (27.68)	.55
WC, cm	120.78 (15.54)	116.86 (14.22)	.11
HC, cm	127.26 (16.26)	126.96 (17.5)	.91
BMIa	40.66 (8.14)	40.16 (8.47)	.71
Diastolic BP, mm Hg	85.17 (7.74)	82.69 (8.61)	.07
Systolic BP, mm Hg	130.04 (12.98)	129.66 (12.55)	.86
Glucose metabolism, mean (SD)
HbA1c, %	5.90 (0.24)	5.95 (0.22)	.17
Fasting blood glucose, mg/dL	113.15 (16.94)	110.45 (17.12)	.34
Fasting C-peptide, ng/mL	4.94 (1.79)	4.51 (1.45)	.23
Fasting insulin level, µIU/mL	32.38 (19.28)	29.05 (17.79)	.42
Organ function, mean (SD)
Creatinine, mg/dL	0.84 (0.17)	0.76 (0.16)	.004b
Cholesterol levels, mg/dL
Total	192.28 (33.22)	191.12 (43.24)	.86
LDL	116.22 (35.52)	113.51 (35.14)	.68
HDL	40.54 (9.65)	42.86 (10.81)	.17
Triglycerides, mg/dL	205.31 (124.78)	196.46 (115.93)	.65
Other biochemical variables, mean (SD)
Urate, mg/dL	6.22 (1.18)	5.88 (1.34)	.23
CRP, mg/dL	1.03 (1.00)	1.03 (1.09)	.97
Hemoglobin, g/dL	14.68 (1.64)	13.78 (1.61	<.001b
Amylase, U/L	31.82 (17.32)	32.79 (17.31)	.73

Abbreviations: BMI, body mass index; BP, blood pressure; CRP, C-reactive protein; HbA_1c, glycosylated hemoglobin A_1c; HC, hip circumference; HDL, high-density lipoprotein; LDL, low-density lipoprotein; PANSS-6, Positive and Negative Syndrome Scale 6; SF-36, 36-item Short Form Survey; WC, waist circumference.

SI conversion factors: To convert HbA_1c from percentage of total hemoglobin to mmol/mol, use the following formula: (HbA_1c % − 2.152)/0.09148; glucose, from mg/dL to mmol/L, multiply by 0.0555; C-peptide, from ng/mL to nmol/L, multiply by 0.331; insulin, from µIU/mL to pmol/L, multiply by 6.945; creatinine, from mg/dL to µmol/L, multiply by 88.4; total cholesterol, LDL, and HDL, from mg/dL to mmol/L, multiply by 0.0259; triglycerides, from mg/dL to mmol/L, multiply by 0.0113; urate, from mg/dL to mmol/L, multiply by 0.0595; CRP, from mg/dL to mg/L, multiply by 10; hemoglobin, from g/dL to g/L, multiply by 10; amylase, from U/L to µkat/L, multiply by 0.0167.

^a Calculated as weight in kilograms divided by height in meters squared.

^b Significant P value.

Glycemic Control

HbA_1c demonstrated a significant treatment effect across the whole study period. The treatment effects were −0.32% (95% CI, −0.40% to −0.23%; P = .001) and −0.46% (95% CI, −0.56% to −0.36%; P < .001) at weeks 15 and 30, respectively, both favoring semaglutide. From the margins plot (Figure 2), it was evident that the semaglutide-treated group experienced a decrease in Hba_1c between baseline and week 15, then remained stable between weeks 15 and 30. Concomitant with the reduction in HbA_1c, the semaglutide-treated group showed significant reductions in fasting plasma blood glucose by 15.68 mg/dL (95% CI, −20.72 to −10.63) (to convert to millimoles per liter, multiply by 0.0555) (Figure 2) and C-peptide levels after 30 weeks compared to the placebo-treated group (Table 2). All patients had prediabetes at the time of recruitment (as per protocol). After 30 weeks of treatment, 60 of 74 participants in the semaglutide group (81%) had obtained an HbA_1c below 5.7%, whereas this was obtained by only 13 of 69 participants in the placebo group (19%) (P < .001).

Figure 2. Hemoglobin A_1c (HbA_1c), Body Weight, Waist Circumference, and Fasting Blood Glucose.

Changes in HbA_1c (A), body weight (B), waist circumference (C), and fasting blood glucose (D). Body mass index calculated as weight in kilograms divided by height in meters squared. Patients received 30 weeks of treatment with either placebo or semaglutide, up to 1.0 mg/week. At baseline, the descriptive values are presented as mean values. At week 15 and at week 30, the mean changes from baseline values are presented with 95% confidence intervals. Significant changes in HbA_1c between groups were present at week 15 and week 30. Changes in fasting blood glucose levels between groups were significant at week 30. Changes in weight between groups were significant at week 15 and week 30. Changes in waist circumference between groups were significant at week 30.

SI conversion factors: To convert HbA_1c from percentage of total hemoglobin to mmol/mol, use the following formula: (HbA_1c % − 2.152)/0.09148; glucose, from mg/dL to mmol/L, multiply by 0.0555.

Table 2. Main Results.

Clinical characteristic	Mean (95% CI)			P value
	Semaglutide	Placebo	Difference between active treatment and placeboa
Body weight, kg	−9.04 (−10.75 to −7.34)	0.17 (−1.65 to 1.99)	−9.21 (−11.68 to −6.75)	<.001b
WC, cm	−5.58 (−5.21 to −7.70)	0.61 (−1.75 to 2.97)	−6.19 (−9.26 to −3.12)	<.001b
HC, cm	−5.82 (−7.35 to −4.29)	−1.57 (−2.96 to −0.18)	−4.25(−6.21 to −2.3)	<.001b
BMIc	−2.98 (−3.53 to −2.42)	−0.12 (−0.72 to 0.48)	−2.86 (−3.71 to −2.01)	<.001b
Systolic BP, mm Hg	−2.91 (−5.75 to −0.08)	−0.66 (−3.7 to 2.38)	−2.26 (−6.42 to 1.91)	.29
Diastolic BP, mm Hg	−1.61 (−3.57 to .35)	−0.23 (−3.24 to 2.79)	−1.38 (−5.03 to 2.26)	.46
Glucose metabolism
HbA1c, %	−0.38 (−0.45 to −0.30)	0.08 (0.01 to 0.16)	−0.46 (−0.56 to −0.36)	<.001b
Fasting blood glucose, mg/dL	−12.43 (−15.32 to −9.73)	3.24 (0.72 to 7.03)	−15.68 (−20.72 to −10.63)	<.001b
Fasting C-peptide, ng/mL	−0.41 (−0.71 to −0.11)	−0.05 (−0.40 to 0.31)	−0.36 (−0.84 to 0.12)	.14
Fasting insulin level, µIU/mL	−6.72 (−11.43 to −2.02)	2.48 (−5.54 to 10.51)	−9.21 (−18.81 to 0.39)	.06
Other biochemical tests
Creatinine, mg/dL	0.03 (0.00 to 0.05)	−0019 (−0.03 to 0.01)	0.04 (0.01 to 0.07)	.01b
Cholesterol, mg/dL
Total	3.47 (−15.44 to 22.39)	−9.27 (−27.03 to 8.88)	12.74 (−13.90 to 39.38)	.35
LDL	−3.47 (−7.72 to 0.77	−3.47 (−8.11 to 1.54)	0.00 (−6.56 to 6.18)	.98
HDL	11.20 (3.47 to 18.92)	0.39 (3.09 to 3.86)	10.81 (2.70 to 18.53)	.007b
Triglycerides, mg/dL	−34.51 (−53.10 to 15.93)	−5.31 (−25.66 to 15.04)	−29.20 (−55.75 to 2.65)	.03b
CRP, mg/dL	−0.11 (−0.35 to 0.12)	−0.00 (−0.21 to 0.21)	−0.11 (−0.39 to 0.18)	.46
Urate, mg/dL	−0.50 (−0.67 to 0.34)	0.00 (−0.17 to 0.17)	−0.50 (−0.67 to −0.17)	.002b
Hemoglobin, g/dL	0.03 (0.15 to 0.21)	−0.06 (−0.27 to 0.15)	0.10 (0.18 to 0.37)	.49
Amylase, U/L	4.97 (2.87 to 7.07)	0.01 (−2.69 to 2.68)	4.98 (1.72 to 8.23)	.003b

Abbreviations: BMI, body mass index; BP, blood pressure; CRP, C-reactive protein; HbA_1c, glycosylated hemoglobin A_1c; HC, hip circumference; HDL, high-density lipoprotein; LDL, low-density lipoprotein; WC, waist circumference.

SI conversion factors: To convert HbA_1c from percentage of total hemoglobin to mmol/mol, divide by 0.09148; glucose, from mg/dL to mmol/L, multiply by 0.0555; C-peptide, from ng/mL to nmol/L, multiply by 0.331; insulin, from µIU/mL to pmol/L, multiply by 6.945; creatinine, from mg/dL to µmol/L, multiply by 88.4; total cholesterol, LDL, and HDL, from mg/dL to mmol/L, multiply by 0.0259; triglycerides, from mg/dL to mmol/L, multiply by 0.0113; CRP, from mg/dL to mg/L, multiply by 10; urate, from mg/dL to mmol/L, multiply by 0.0595; hemoglobin, from g/dL to g/L, multiply by 10; amylase, from U/L to µkat/L, multiply by 0.0167.

^a Change from baseline to week 30 in the semaglutide-treated group and the placebo-treated group.

^b Significant P value.

^c Calculated as weight in kilograms divided by height in meters squared.

Body Weight

At week 30, the mean weight loss difference between semaglutide and placebo treatment was −9.21 kg (95% CI, −11.68 to −6.75; P < .001) (Figure 2). For waist circumference, the mean difference was −6.19 cm (95% CI, −9.26 to −3.12; P < .001) (Figure 2), and for hip circumference, −4.25 cm (95% CI, −6.21 to −2.30; P < .001) compared to placebo treatment (Table 2).

Other Metabolic Parameters

HDL levels increased significantly by 10.81 mg/dL (95% CI, 2.70-18.53; P = .007) (to convert to millimoles per liter, multiply by 0.0259), whereas triglycerides decreased by 29.20 mg/dL (95% CI, −55.75 to 2.65; P = .03) (to convert to millimoles per liter, multiply by 0.0113). No changes were observed in mean differences for total cholesterol, LDL, and systolic and diastolic blood pressure comparing semaglutide to placebo (Table 2).

PANSS-6

No significant group differences in PANSS-6 scores were observed either at 15 weeks or at 30 weeks (Figure 3).

Figure 3. Quality of Life and Psychiatric Symptoms.

Changes in 36-item Short Form Survey (SF-36) Physical Component Summary (PCS) (A), SF-36 Metal Component Summary (MCS) (B), Positive and Negative Syndrome Scale (PANSS) positive symptoms (C), and PANSS negative symptoms (D) in patients with schizophrenia, body mass index (calculated as weight in kilograms divided by height in meters squared) of ≥27, and prediabetes (hemoglobin A_1c between 5.6% and 6.4%). Patients received 30 weeks of treatment with either placebo or semaglutide, up to 1.0 mg/week. At baseline, the descriptive values are presented as mean values. At week 15 and at week 30, the mean changes from baseline values are presented with 95% confidence intervals. Group differences were significant for SF-36 PCS at week 15 and week 30, whereas no other significant differences were observed.

SF-36v2

After 15 weeks and 30 weeks of treatment, patients receiving semaglutide scored significantly higher than those receiving placebo in PCS, showing a mean difference of 3.75 points (95% CI, 1.52-5.98; P = .001) at week 30. This finding is clinically relevant. In contrast, no significant changes were observed for the MCS, showing a mean difference of −0.78 points (95% CI, −4.18 to 2.62; P = .65) at week 30 (Figure 3).

Adverse Events

The most common adverse events associated with semaglutide were gastrointestinal symptoms, but high rates were observed in both groups, particularly at study start. The incidence of nausea, abdominal pain, and constipation was higher in the semaglutide group, but decreased over time (eTables 2 and 3 in Supplement 2). Overall, no changes were observed in liver parameters, although a small and hardly clinical relevant increase was observed in serum amylase in the semaglutide-treated group (Table 2). Two participants who developed adverse effects at the maximum dose had their dosage reduced to 0.50 mg weekly of semaglutide or placebo for the remainder of the study. Serious adverse events were numerically but not statistically lower in the semaglutide group compared to the placebo group (11.7% vs 18.2%) (eTable 3 in Supplement 2). In the placebo group, there were 47 hospitalizations among 24 patients, and in the semaglutide group, 68 admissions among 15 patients. This difference was driven by a small number of participants who accounted for a disproportionately high number of hospitalizations. The occurrence of somatic serious adverse events did not differ between groups.

The most common serious adverse event was exacerbation of psychiatric illness, leading to voluntary admission to a psychiatric hospital. The frequency of admission for suicidal ideation was not different between groups (eTable 4 in Supplement 2).

Discussion

This RCT in patients with overweight or obesity, schizophrenia, and prediabetes demonstrated that 30 weeks of semaglutide treatment was able to reduce HbA_1c levels and body weight and improve physical well-being compared to placebo. Furthermore, semaglutide was well tolerated and did not cause any mental deterioration as judged from PANSS-6 and the SF36-v2. Finally, compared to other medications previously studied in this patient population—for example, metformin, topiramate, and liraglutide—semaglutide was associated with a tolerable adverse event profile, with no treatment-related discontinuations.^16,23

Thirty weeks of semaglutide treatment caused a mean reduction in HbA_1c of 0.46% compared to placebo. This effect was larger than the reductions previously reported in the short-term studies using exenatide¹⁴ and liraglutide¹⁶ in psychiatric populations, and the reduction was also larger than in the SCALE study, which investigated 56 weeks of treatment with liraglutide in obese individuals without T2D.¹⁷ Notably, our results are comparable to the recent SELECT study,²⁴ which compared semaglutide, 2.4 mg/week, vs placebo over an average of 104 weeks in obese individuals without diabetes. In that study, the baseline HbA_1c level was 5.78% of total hemoglobin (39.7 mmol/mol) (ie, comparable with the baseline HbA_1c of the present study), and the reduction in HbA_1c was also comparable, being just above 0.32% in the SELECT study compared to just above 0.46% in our trial. Thus, it appears that in SGA-treated patients with SZ, 30 weeks of treatment with semaglutide, 1 mg/week, is comparable to semaglutide, 2.4 mg/week, for 104 weeks, when it comes to improvement in HbA_1c. On this basis, we speculate that overweight or obese patients without diabetes with SZ and SGA treatment have an increased responsiveness to semaglutide, but whether this is caused by reductions in the adverse effects of SGA (eg, appetite) or the psychiatric disease per se remains unknown.²⁵

Alongside reductions in HbA_1c, semaglutide reduced the fraction of patients with prediabetes by 4-fold—a finding not observed in earlier studies using exenatide, whereas similar results have been reported in a liraglutide study by Larsen and colleagues.^14,16 This is clinically meaningful and should be taken into consideration when managing high-risk patients with SZ aiming to prevent CVD.²⁶

Thirty weeks of semaglutide treatment caused a significant weight reduction of 9.21 kg (7.5%) compared to placebo (ie, slightly less than the 8.5% weight loss observed in the SELECT study). The difference may be attributed to the higher semaglutide dose and longer treatment duration in the SELECT trial.²⁴ Since the cardiovascular risk reduction in the SELECT study was primarily driven by weight loss, we believe weight management—for instance, via treatment with a GLP-1 RA—should be part of the preventive intervention for CVD in patients with SZ with antipsychotics-induced prediabetes.

HDL increased and triglyceride decreased in the semaglutide-treated group, and this aligns with the SELECT study.²⁴ In contrast, we did not observe changes in LDL or blood pressure as seen in larger and longer-running studies with GLP-1 RA treatment. In SCALE¹⁷ using liraglutide, in STEP²⁷ and SELECT²⁴ using semaglutide, and in SURMOUNT^28,29 using the dual GLP-1 RA/glucose-dependent insulinotropic polypeptide receptor agonist (GIP RA) tirzepatide, significant reductions in these important risk factors for CVD were seen with active treatment. The reason for this observation deserves further investigation.

This study did not reveal group differences in psychotic symptoms as measured by the PANSS-6. This is consistent with other GLP-1 RA studies.^15,16 When considered alongside the absence of change in the mental component of QoL (MCS), these findings suggest that semaglutide does not negatively affect psychiatric symptoms in this study population. Conversely, semaglutide demonstrated a clinically meaningful positive effect on the physical component of QoL (PCS), which also has been documented in other populations treated with GLP-1 RAs.^27,28,29,30 Such improvements suggest a multifaceted benefit of GLP-1 RAs beyond glycemic control, potentially offering enhancements in physical well-being. However, as the improvement in physical QoL was a secondary finding, this finding requires additional confirmation.

Strength and Limitations

The placebo-controlled, double-blinded design of this RCT ensures high methodological rigor. The low dropout rate (<10%), likely due to home-based interventions, strengthens the validity of the findings. The study included a diverse patient population from multiple treatment centers across urban and rural areas, enhancing generalizability. Additionally and reassuringly, semaglutide was well tolerated and showed no adverse effects on mental health. Finally, the home-based visit design, which involved assessments, study drug administration, and data collection, including most blood samples, proved effective in minimizing dropout rates, which were so low that we refrained from performing a detailed analysis of dropout characteristics. However, we acknowledge that the home-based approach is resource intensive and may not be practical in larger clinical settings. Therefore, it is reassuring that many patients with SZ can self-administer semaglutide after brief training, supporting its practicality in real-world use. Regarding mental QoL and psychiatric symptoms, semaglutide was neutral, but our study may have been underpowered to detect such changes, highlighting the need for long-term studies in this patient population.

Conclusions

In this RCT, in patients with schizophrenia, prediabetes, and obesity, once-weekly semaglutide injections, up to 1.0 mg/week, were safe and effective. Compared to placebo, semaglutide significantly reduced HbA_1c, body weight, and metabolic variables and improved physical QoL without affecting mental status. Given these benefits, semaglutide should be considered for patients with SZ, prediabetes, and a BMI of 27 or higher, as the potential for weight loss and prevention of T2D may justify the economic cost of treatment.

Supplement 1.

Trial Protocol

Supplement 2.

eTable 1. Inclusion and Exclusion Criteria

eTable 2. Probability of Side Effects

eTable 3. Adverse Events and Severe Adverse Events

eTable 4. Hospitalizations

Supplement 3.

Data Sharing Statement