Effect of Glucagon-like Peptide-1 Receptor Agonists on Cardio-Metabolic Risk Factors among Obese/Overweight Individuals Treated with Antipsychotic Drug Classes: An Updated Systematic Review and Meta-Analysis of Randomized Controlled Trials

Round 1

Reviewer 1 Report

In this manuscript, the authors update a meta-analysis of RCTs using GLP-1RAs in overweight/obese patients with psychotic disorders being treated with antipsychotics and report on various cardiometabolic risk factors. They find that GLP-1RAs were effective in reducing body weight and BMI and had positive impacts on fasting plasma glucose and lipid profile. They did not identify an impact on waist circumference or blood pressure. On the whole, this was a reasonably well done analysis though there are some concerns that need to be addressed.

1.       In the introduction, it may be worth discussing how psychotic disorders are associated with a greatly increased mortality with around 15 years of life lost (Hjorthoj et al. Lancet Psychaitry. 2017. https://doi.org/10.1016/s2215-0366(17)30078-0) and that cardiovascular disease accounts for the highest mortality in this population (Olfson et al. JAMA Psychiatry. 2017. https://doi.org/10.1001/jamapsychiatry.2015.1737). This, I think, will help to put into perspective why it is so important to identify treatments that can improve cardiometabolic risk factors in this population.

2.       You make the claim here that liraglutide is superior to exenatide based on your subgroup analysis. However, this is based only on two studies per subgroup, and a significant difference between these subgroups was only found for waist circumference. Furthermore, this difference seems to be driven entirely by a single study (Ishoy et al. 2017) in which there may be multiple other factors contributing to the lack of efficacy other than the specific GLP-1RA used. For instance, this study had the shortest duration of treatment. Additionally, Siskind et al (2018) demonstrated in their previous individual participant data meta-analysis of GLP-1RAs in this population that the specific antipsychotic the patients were being treated with seemed to greatly modify the efficacy of the GLP-1RAs, with GLP-1RAs having much greater effect in those being treated with clozapine and olanzapine, both well known for inducing weight gain, compared to those on all other antipsychotics. Of note, two of the studies included in this meta-analysis that demonstrated a significant effect of GLP-1RAs on cardiometabolic risk factors (Larsen et al. 2017 and Siskind et al. 2017) were both done exclusively in patients on either clozapine or olanzapine whereas the study showing the least efficacy (Ishoy et al. 2017) included patients on a variety of antipsychotics. The last study included in the meta-analysis also had a mix of antipsychotics, but it is worth noting that around 20-26% of the patients in this study were either on clozapine or olanzapine. At any rate, I think there may be numerous other factors contributing to this apparent subgroup difference and that there is not enough evidence here to make the claim that there is really a difference in efficacy between liraglutide and exenatide. It may also be worth doing some further sensitivity analyses utilizing some of these other possible factors. For instance, you could do meta-regressions looking at efficacy versus % of patients in study on olanzapine/clozapine and/or versus duration of treatment to see if a relationship exists between these things.

3.       You mentioned that safety profiles for GLP-1RAs are well established, which is why they were not assessed in this study. Still, it may be worth considering an assessment of tolerability by comparing the proportion of subjects leaving the study due to any adverse event in the GLP-1RA-treated group versus the placebo group. This measure was used by a recent network meta-analysis that looked at GLP-1RAs among numerous other agents for their efficacy in improving cardiometabolic risk factors (Wang et al. J Psychopharmacol. 2021. https://doi.org/10.1177/02698811211035391). In that study, the tolerability of GLP-1RAs was quite good. Of note, this study also included an additional RCT of exenatide that does not appear to have been included in the current meta-analysis (NCT00845507, https://clinicaltrials.gov/ct2/show/NCT00845507). While I didn’t find any publications related to this clinical trial, it appears that the data is readily available, and its inclusion in the current meta-analysis would likely further bolster your findings and help you ascertain if exenatide really is less efficacious than liraglutide. It would also provide further justification for an updated meta-analysis as opposed to just including one additional study not in the meta-analysis by Siskind et al. (2018).

4.       There are some stylistic as well as what appears to be mislabeling of some of the figures. For instance, the order of the studies in Figure 3 appears to be different from all the other figures. Also, Figure 4 appears to be mislabeled as “fasting plasma glucose” when the text would suggest that it is actually waist circumference. The titles for Supplementary Figures 4 and 5 appear to be switched. Also, For Supplementary Figure 4, the x-axis labels “Favours GlP-1RA” and “Favours control” likely need to be reversed.

5.       You make the statement that the funnel plot reveals asymmetry, but I’m not sure this is a valid statement with only four studies plotted. Plus, though you bring up the possibility of publication bias, this and its potential implications are never addressed any further in the discussion.

6.       On the whole, the paper is well-written and easy to read. Still, I would suggest a few edits. In the results section, I would replace the word “rest” with the word “other.” On line 242, I believe “in” should be “is.” On line 246, I believe “weight” should be “BMI.” References 30 and 31 appear to be the same reference repeated.

7.       In the discussion, lines 243-249, you draw comparisons between this meta-analysis and a prior meta-analysis of metformin, which seems to have very similar efficacy. The similarity between these two treatments was also demonstrated in the above-mentioned network meta-analysis (Wang et al. J Psychopharmacol. 2021. https://doi.org/10.1177/02698811211035391). If these two drugs are so similar, then it begs the question why do we need drugs other than metformin, which is a well-established treatment. Thus, I think a brief discussion of why it is good to have additional treatment options may be beneficial. It may also be worth pointing out that exenatide can be dosed once weekly, which may help with adherence over a once or twice daily medication.

Author Response

Reviewer #1

In this manuscript, the authors update a meta-analysis of RCTs using GLP-1RAs in overweight/obese patients with psychotic disorders being treated with antipsychotics and report on various cardiometabolic risk factors. They find that GLP-1RAs were effective in reducing body weight and BMI and had positive impacts on fasting plasma glucose and lipid profile. They did not identify an impact on waist circumference or blood pressure. On the whole, this was a reasonably well done analysis though there are some concerns that need to be addressed.

1. In the introduction, it may be worth discussing how psychotic disorders are associated with a greatly increased mortality with around 15 years of life lost (Hjorthoj et al. Lancet Psychaitry. 2017. https://doi.org/10.1016/s2215-0366(17)30078-0) and that cardiovascular disease accounts for the highest mortality in this population (Olfson et al. JAMA Psychiatry. 2017. https://doi.org/10.1001/jamapsychiatry.2015.1737). This, I think, will help to put into perspective why it is so important to identify treatments that can improve cardiometabolic risk factors in this population.

Answer: We cordially thank the reviewer for this important comment. A short paragraph has now been added in the “Introduction” section, commenting on the shorter life expectancy of subjects with psychotic disorders (especially schizophrenia) and the impact of cardiovascular disease on their prognosis. Thank you once again for this significant comment.

2. You make the claim here that liraglutide is superior to exenatide based on your subgroup analysis. However, this is based only on two studies per subgroup, and a significant difference between these subgroups was only found for waist circumference. Furthermore, this difference seems to be driven entirely by a single study (Ishoy et al. 2017) in which there may be multiple other factors contributing to the lack of efficacy other than the specific GLP-1RA used. For instance, this study had the shortest duration of treatment. Additionally, Siskind et al (2018) demonstrated in their previous individual participant data meta-analysis of GLP-1RAs in this population that the specific antipsychotic the patients were being treated with seemed to greatly modify the efficacy of the GLP-1RAs, with GLP-1RAs having much greater effect in those being treated with clozapine and olanzapine, both well known for inducing weight gain, compared to those on all other antipsychotics. Of note, two of the studies included in this meta-analysis that demonstrated a significant effect of GLP-1RAs on cardiometabolic risk factors (Larsen et al. 2017 and Siskind et al. 2017) were both done exclusively in patients on either clozapine or olanzapine whereas the study showing the least efficacy (Ishoy et al. 2017) included patients on a variety of antipsychotics. The last study included in the meta-analysis also had a mix of antipsychotics, but it is worth noting that around 20-26% of the patients in this study were either on clozapine or olanzapine. At any rate, I think there may be numerous other factors contributing to this apparent subgroup difference and that there is not enough evidence here to make the claim that there is really a difference in efficacy between liraglutide and exenatide. It may also be worth doing some further sensitivity analyses utilizing some of these other possible factors. For instance, you could do meta-regressions looking at efficacy versus % of patients in study on olanzapine/clozapine and/or versus duration of treatment to see if a relationship exists between these things.

Answer: We cordially thank the reviewer for this targeted and significant comment. Indeed, with the exception of waist circumference, no subgroup difference between liraglutide and exenatide for the main outcomes of our meta-analysis was shown. This is now clearly stated in the last sentence of the first paragraph in the “Discussion” section, while the erroneous conclusion “However, liraglutide has a better efficacy profile compared to exenatide, and should be preferred for use in those subjects.” has now been deleted from our revised manuscript. Regarding the assessment of confounding factors that could have influenced the generated results, we totally agree that a sensitivity analysis according to baseline characteristics of interest would be of value; however, as stated now in the limitations section of our revised manuscript, lack of access to individual participants’ data did not permit us to conduct such analyses.

3. You mentioned that safety profiles for GLP-1RAs are well established, which is why they were not assessed in this study. Still, it may be worth considering an assessment of tolerability by comparing the proportion of subjects leaving the study due to any adverse event in the GLP-1RA-treated group versus the placebo group. This measure was used by a recent network meta-analysis that looked at GLP-1RAs among numerous other agents for their efficacy in improving cardiometabolic risk factors (Wang et al. J Psychopharmacol. 2021. https://doi.org/10.1177/02698811211035391). In that study, the tolerability of GLP-1RAs was quite good.

Answer: We cordially thank the reviewer for this important and targeted comment. Indeed, we have added in our revised manuscript a safety outcome regarding the risk for treatment discontinuation with GLP-1RAs versus control, demonstrating that use of GLP-1RAs compared to control did not increase the risk for discontinuation of treatment. Corresponding forest plot is provided as supplementary figure 7 in the revised manuscript.

Of note, this study also included an additional RCT of exenatide that does not appear to have been included in the current meta-analysis (NCT00845507, https://clinicaltrials.gov/ct2/show/NCT00845507). While I didn’t find any publications related to this clinical trial, it appears that the data is readily available, and its inclusion in the current meta-analysis would likely further bolster your findings and help you ascertain if exenatide really is less efficacious than liraglutide. It would also provide further justification for an updated meta-analysis as opposed to just including one additional study not in the meta-analysis by Siskind et al. (2018).

Answer: We heartily thank the reviewer for this notice and suggestion to include the trial registered as NCT00845507. However, we preferred to exclude unpublished data, since details on research protocol are not provided, whereas it appears rather conflicting that this trial was completed 8 years ago, with results, but there are no relevant publications in major databases. Thank you once again for the great help.

4. There are some stylistic as well as what appears to be mislabeling of some of the figures. For instance, the order of the studies in Figure 3 appears to be different from all the other figures. Also, Figure 4 appears to be mislabeled as “fasting plasma glucose” when the text would suggest that it is actually waist circumference. The titles for Supplementary Figures 4 and 5 appear to be switched. Also, For Supplementary Figure 4, the x-axis labels “Favours GlP-1RA” and “Favours control” likely need to be reversed.

Answer: We would like to heartily thank the reviewer for this important comments. All comments have been addressed in the revised manuscript accordingly.

5. You make the statement that the funnel plot reveals asymmetry, but I’m not sure this is a valid statement with only four studies plotted. Plus, though you bring up the possibility of publication bias, this and its potential implications are never addressed any further in the discussion.

Answer: We would like to thank the reviewer for this major comment. Indeed, we have expanded corresponding section in the revised manuscript, while we have also added a short comment in the “Limitations” section. In general, according to Cochrane recommendations, it is true that “As a rule of thumb, tests for funnel plot asymmetry should be used only when there are at least 10 studies included in the meta-analysis, because when there are fewer studies the power of the tests is too low to distinguish chance from real asymmetry.”. Thank you once again for this important comment.

6. On the whole, the paper is well-written and easy to read. Still, I would suggest a few edits. In the results section, I would replace the word “rest” with the word “other.” On line 242, I believe “in” should be “is.” On line 246, I believe “weight” should be “BMI.” References 30 and 31 appear to be the same reference repeated.

Answer: We would like to thank the reviewer for those comments. Manuscript has been revised accordingly.

7. In the discussion, lines 243-249, you draw comparisons between this meta-analysis and a prior meta-analysis of metformin, which seems to have very similar efficacy. The similarity between these two treatments was also demonstrated in the above-mentioned network meta-analysis (Wang et al. J Psychopharmacol. 2021. https://doi.org/10.1177/02698811211035391). If these two drugs are so similar, then it begs the question why do we need drugs other than metformin, which is a well-established treatment. Thus, I think a brief discussion of why it is good to have additional treatment options may be beneficial. It may also be worth pointing out that exenatide can be dosed once weekly, which may help with adherence over a once or twice daily medication.

Answer: We cordially thank the reviewer for this targeted and crucial comment. A short paragraph commenting on this has been added in the “Discussion” section of the revised manuscript, please see: “Regarding the comparison between GLP-1RAs and other treatment options with similar efficacy in terms of weight loss in this population, such as metformin [29], it seems that use of GLP-1RAs should be prioritized, due to their pleiotropic effects, including cardio- and reno-protective actions [36, 37], if tolerated and, of course, if being affordable. In addition, once-weekly dosing regiments of some GLP-1RAs (such as exenatide, but also dulaglutide and semaglutide) might also be more efficacious as a treatment strategy, since it may enhance compliance of patients to treatment, compared to once- or twice-daily administered pharmacologic interventions.”. Thank you once again.

Reviewer 2 Report

This review  addresses an interesting and important topic. The manuscript  is very well written. The background is clearly presented in the Introduction. Methods including the search strategy  are timely and appropriate. According to the appropriate search strategy the number of studies included in this review is small. Nevertheless novel, interesting results are provided clearly  in text, figures  and tables.  The discussion is fully appropriate. Limitations are approprately considered. Clear conclusions are presented.  In my view no corrections or clarifications of this manuscript are are required. I wish to congratulate the authors to an excellent job.

Author Response

Reviewer #2

This review  addresses an interesting and important topic. The manuscript  is very well written. The background is clearly presented in the Introduction. Methods including the search strategy  are timely and appropriate. According to the appropriate search strategy the number of studies included in this review is small. Nevertheless novel, interesting results are provided clearly  in text, figures  and tables.  The discussion is fully appropriate. Limitations are approprately considered. Clear conclusions are presented.  In my view no corrections or clarifications of this manuscript are are required. I wish to congratulate the authors to an excellent job.

Answer: We cordially thank the reviewer for his/her comments on our manuscript and the acceptance of our work.

Reviewer 3 Report

Overall the paper is well written and analysis well conducted. However, there is only limited originality and inovativeness, as the glucose-related effects of this treatment class are well known in general populations. The paper deals with a specific subpopulation, while there is no relevant concern, why this population should react differently to the treatment. 

This limits the innovativeness and overall impact, otherwise the manuscript is fine. 

Author Response

Reviewer #3

Overall the paper is well written and analysis well conducted. However, there is only limited originality and inovativeness, as the glucose-related effects of this treatment class are well known in general populations. The paper deals with a specific subpopulation, while there is no relevant concern, why this population should react differently to the treatment. 

This limits the innovativeness and overall impact, otherwise the manuscript is fine. 

Answer: We cordially thank the reviewer for his/her comments on our manuscript. Indeed, eligible studies are only a few, while overall sample size is relatively small; however, as stated in the “Introduction” section of our revised manuscript, subjects with psychotic disorders experience a significantly shorter life span compared to the general population, with cardiovascular disease representing the leading cause of death. Therefore, adequate control of cardio-metabolic risk factors is of outmost importance for those individuals, in order to substantially decrease morbidity and mortality burden.

Round 2

Reviewer 1 Report

On the whole, the authors have done a reasonable job of addressing the concerns raised in the first review, and I respect their reasoning for not wanting to include unpublished data in this study. I have just one minor issue regarding the author's use of the phrase "continuous variables" on line 133 since aren't all the measures meta-analyzed in this study technically continuous?

Author Response

We cordially thank the reviewer for his/her contribution to the improvement of our manuscript's overall quality.

Since we added the secondary outcome "risk for treatment discontinuation" after reviewer's suggestion during the previous round of revision, we now have numeric (continuous) variables and one dichotomous variable, as well. The typographical error in the sentence has been corrected. Thank you once again for the meticulous assessment of our manuscript and your contribution to its optimization.