A systematic review and meta-analysis of eight retrospective cohorts covering 311,402 ACDF patients found that perioperative GLP-1 receptor agonist use was associated with significantly lower odds of pseudarthrosis at 6, 12, and 24 months after surgery. The first review to examine this question…
Perioperative use of glucagon-like peptide-1 receptor agonists GLP-1RAs was associated with significantly lower odds of pseudarthrosis after anterior cervical discectomy and fusion ACDF , according to a systematic review and meta-analysis published in the European Spine Journal, the official journal of the European Spine Society, the European Spinal Deformity Society, and the European Section of the Cervical Spine Research Society. The review pooled eight retrospective cohort studies covering 311,402 ACDF patients, including 8,182 GLP-1RA users matched to non-users. It is the first systematic review to specifically investigate how GLP-1RAs influence outcomes after this procedure.
The association held at every time point examined. GLP-1RA users had lower odds of pseudarthrosis at 6 months OR 0.60, 95% CI 0.52-0.71; I² = 16.6% , at 12 months OR 0.62, 95% CI 0.53-0.73; I² = 39.1% , and at 24 months OR 0.67, 95% CI 0.57-0.79; I² = 34.9% . The pooled analysis was rated Level III, therapeutic study , reflecting its reliance on retrospective observational cohorts rather than randomized assignment.
The finding matters beyond the spine clinic. GLP-1RAs are among the most widely prescribed peptide drugs in the world, used for type 2 diabetes mellitus and obesity, and the review adds clinical weight to evidence that these peptides act on bone beyond their canonical metabolic functions. If confirmed prospectively, the result could inform perioperative decision-making for one of the most common spinal surgeries performed.
Pseudarthrosis is failure of the bony fusion that ACDF is designed to achieve. It is a leading cause of late revision surgery after cervical arthrodesis, and surgical management means reoperation, additional hospitalization, and uncertain pain relief. In this analysis, the odds of pseudarthrosis among GLP-1RA users were approximately 40% lower at 6 months, 38% lower at 12 months, and 33% lower at 24 months relative to matched non-users.
The effect sizes were stable across follow-up intervals, and the heterogeneity statistics were modest. The I² value of 16.6% at 6 months indicates low inconsistency among the pooled studies; the 12-month figure of 39.1% and the 24-month figure of 34.9% indicate low to moderate inconsistency. That consistency makes it less likely the association rests on a single dominant study, though it does not by itself address causation.
The secondary endpoints did not behave as cleanly. Findings for reoperation, dysphagia, deep vein thrombosis, and readmission were heterogeneous and inconsistent across the included studies. For dysphagia, the most common complication after ACDF, mixed results may reflect differences in how the complication was defined and captured across administrative and registry data. For deep vein thrombosis and readmission, event rates are low, and retrospective cohorts are poorly positioned to adjudicate them.
The investigators followed PRISMA guidelines for systematic reviews and meta-analyses. They searched MEDLINE, Embase, PubMed, and the Cochrane Library from database inception through February 2026, and two reviewers extracted data independently. Eligible studies were limited to human observational cohort studies comparing ACDF patients with perioperative GLP-1RA exposure against matched non-users. Eight studies met the criteria, producing a pooled population of 311,402 ACDF patients.
Outcomes were assessed at 6, 12, and 24 months after surgery. The prespecified endpoints were pseudarthrosis, reoperation, dysphagia, deep vein thrombosis, and readmission. The pseudarthrosis association was consistent across all three time points. The secondary endpoints were pooled but produced heterogeneous and inconsistent results, and the review flags the evidence on postoperative complications other than pseudarthrosis as inconclusive.
The design imposes real limits. Retrospective cohort studies can adjust for measured confounders, and the included studies matched GLP-1RA users to non-users, but matching cannot eliminate residual confounding. Patients prescribed GLP-1RAs for diabetes or obesity differ from other patients in duration of metabolic disease, adherence behavior, frequency of medical contact, and weight trajectory, all of which may be imperfectly captured in records. The Level III therapeutic study rating is the formal acknowledgment of that: the evidence base supports an association, not a causal claim, and treatment decisions built on it should be made accordingly.
GLP-1 is an incretin hormone secreted by intestinal L cells in response to nutrient intake. Its receptor, the GLP-1 receptor , is a G protein-coupled receptor expressed not only on pancreatic beta cells but also on osteoblasts, osteocytes, and, in some reports, osteoclast precursors. GLP-1RAs are peptide analogs engineered to resist degradation by dipeptidyl peptidase-4 , extending the activity of native GLP-1 and providing sustained receptor activation. Their established effects are glycemic control, appetite suppression, and weight loss, which is why the class is approved for type 2 diabetes mellitus and obesity.
The proposed skeletal effects are both direct and indirect. Preclinical work has suggested that GLP-1 receptor signaling can promote osteoblast differentiation and activity, restrain osteoclast-mediated resorption, and dampen the local and systemic inflammatory milieu that suppresses bone formation. Chronic hyperglycemia and inflammation both impair the cellular machinery of fracture repair, and GLP-1RAs counter both by lowering glucose, reducing body weight, and attenuating inflammatory cytokine production. For a fusion mass to form after ACDF, osteoprogenitor cells in the decorticated endplates and the interbody graft must proliferate, differentiate, and lay down new bone across the disc space. A drug that shifts that balance toward osteogenesis, directly or by improving the metabolic environment, could plausibly reduce nonunion risk.
The meta-analysis cannot separate those pathways, and it was not designed to. But the consistency of the signal across three time points, in a pooled population of more than 300,000 patients, gives the mechanistic literature a clinical anchor. It also frames the next question: whether the effect is specific to GLP-1 receptor activation or a downstream consequence of improved metabolic control. That distinction matters for choosing agent, dose, and duration in any future study.
For spine surgeons, the immediate takeaway is not a change in prescribing. Many patients hold GLP-1RAs for diabetes and weight management, and this review provides no basis to start, stop, or alter those drugs around cervical fusion. What it does provide is a signal for risk stratification and perioperative counseling. Patients on GLP-1RAs may carry a lower baseline risk of nonunion, and surgeons tracking fusion outcomes should record GLP-1RA exposure as a covariate in their own registries.
For researchers, the review is a road map. The clearest gap is prospective confirmation: no prospective studies of GLP-1RA use and ACDF outcomes exist, and the authors state explicitly that future prospective research is required. A prospective cohort with verified drug exposure, standardized imaging, and CT-based fusion assessment at fixed intervals would test the 6-, 12-, and 24-month associations directly. Mechanistic work should ask whether the effect tracks with glucose control, weight loss, or circulating inflammatory markers, and whether specific agents or doses change the signal.
For the peptide field, the clinical signal is a reminder that approved peptide therapeutics can carry pleiotropic effects that surface years after launch. That has supply chain implications. GLP-1RA demand is already high and rising, and the class carries established cold chain and stability requirements. If the bone healing signal survives prospective testing, the perioperative population could add a distinct demand stream, and clinicians will need assurance that the peptide product they administer is consistent from lot to lot. Peptide manufacturers should monitor this evidence base closely: a confirmed skeletal indication would place GLP-1RA quality, purity, and supply continuity under a different kind of scrutiny, and preparation requires lead times measured in years.
Every included study was retrospective, and the authors rate the overall body of evidence as Level III. That is the central limitation. Matching controls for measured variables, but not for the unmeasured differences that accompany a GLP-1RA prescription: disease severity, lifestyle, healthcare engagement, and the metabolic history that led to the drug in the first place. Reverse causation is possible, though the likely direction of such bias would attenuate the observed association if patients with more aggressive metabolic disease are both more likely to receive GLP-1RAs and less likely to fuse. The size and direction of residual confounding cannot be determined from pooled retrospective data.
The secondary endpoints remain unresolved. Findings for reoperation, dysphagia, deep vein thrombosis, and readmission were heterogeneous and inconsistent, which the authors attribute to study heterogeneity rather than a demonstrated absence of effect. The evidence on postoperative complications other than pseudarthrosis is therefore inconclusive. Even for the primary endpoint, the I² values, while low to moderate, indicate that the included studies were not perfectly interchangeable.
The search window is another limitation. The literature search ran from database inception to February 2026, so later publications are not captured. That is routine for a meta-analysis, but in a fast-moving area such as GLP-1RA research, the evidence base will need periodic updating. The open questions are specific and testable: What mechanism links GLP-1 receptor activation to spinal fusion? Does the effect vary by specific GLP-1RA agent, dose, or duration of therapy? Are the inconsistent secondary findings true effects or artifacts of heterogeneous design? The study that would settle them is now well defined: a prospective cohort or randomized trial with verified exposure, standardized fusion assessment, and collection of metabolic and inflammatory biomarkers. Until that work is done, the association stands as a striking clinical signal in search of a mechanism.
Peptides referenced: Glucagon, GLP-1.
Related reading: GLP-1 Initiation Tied to Fewer Coded GI Symptoms in IBS, Peptide self-assembly death pathways classified into five themes, Semaglutide Solid Lipid Nanoparticles Boost Oral Permeability Sixfold, Semaglutide Slows Proteomic Dementia Risk Signature in SELECT Analysis.