Asthma and chronic obstructive pulmonary disease are usually treated as separate conditions, but a meaningful share of patients carry features of both at the same time. Researchers call this asthma-COPD overlap syndrome, or ACOS. According to a 2026 network meta-analysis published in EClinicalMedicine, ACOS accounts for roughly 15 to 25 percent of all chronic obstructive airway disease and is tied to more frequent flare-ups and higher mortality than either condition alone.
The same research group noticed something worth investigating: a wave of newer glucose-lowering drugs has transformed diabetes care over the past decade, and several of those agents work through pathways that could plausibly touch lung biology. Their question was straightforward. Do these drugs move the needle on ACOS risk, and if so, which ones, at what doses, and in which patients?
To answer that, the team conducted a dose-stratified network meta-analysis, a method that lets researchers compare many treatments against each other simultaneously by weaving together evidence from trials that never directly competed head-to-head. The final dataset covered 316,832 adults enrolled across 128 randomized controlled trials.
What ACOS is and why it matters
Asthma inflames and narrows the airways in a reversible, often allergic pattern. COPD damages the air sacs and small airways in a largely irreversible, inflammation-driven pattern linked to long-term irritant exposure. When a patient shows hallmarks of both, the diagnosis becomes ACOS.
ACOS is clinically frustrating because standard treatment guidelines are written for one condition or the other, not the combination. Patients tend to experience more exacerbations, more hospitalizations, and higher overall mortality than people with asthma or COPD alone. Finding drugs that reduce ACOS-related events would represent a meaningful advance, even if those drugs were originally developed for an entirely different purpose.
Drug classes under the microscope
The analysis focused on three modern classes of glucose-lowering therapy. The first class, glucagon-like peptide-1 (GLP-1) receptor agonists, works by mimicking a natural hormone that stimulates insulin release, suppresses glucagon, and slows gastric emptying. Injectable semaglutide belongs to this class. The second class, SGLT2 inhibitors, lowers blood sugar by blocking a kidney transporter so glucose is excreted in urine rather than reabsorbed. The agents canagliflozin, empagliflozin, and dapagliflozin are examples. The third class, DPP-4 inhibitors, prevents the breakdown of the same GLP-1 hormone and includes saxagliptin.
Each class acts through distinct molecular machinery. That distinction turned out to matter a great deal once the respiratory data were examined.
Findings for ACOS risk
The headline result from the meta-analysis was that several specific agents were associated with meaningfully lower ACOS risk compared to placebo or standard care. Injectable semaglutide showed a risk ratio of 0.64 with a 95 percent confidence interval of 0.49 to 0.84, meaning the observed ACOS event rate was roughly 36 percent lower in the semaglutide arms than in control arms across trials. Among SGLT2 inhibitors, canagliflozin came in at a risk ratio of 0.62, empagliflozin at 0.70, and dapagliflozin at 0.76, all with confidence intervals that did not cross 1.0.
Risk ratios below 1.0 indicate fewer events in the treated group relative to the control group. A ratio of 0.62 does not mean the drug cured ACOS; it means participants receiving that agent in these trial populations had statistically fewer ACOS-related respiratory events recorded during follow-up than participants in the comparison arms.
The dose-stratified portion of the analysis added nuance. Certain regimens showed stronger signals at particular dose levels, suggesting that the relationship between dose and respiratory outcome may not be linear and deserves further prospective study.
Asthma and COPD signals examined separately
The researchers also broke ACOS apart into its component conditions to ask whether the associations were driven more by asthma outcomes, COPD outcomes, or both.
Dapagliflozin was specifically associated with lower asthma risk. Canagliflozin, empagliflozin, and injectable semaglutide were each associated with lower COPD risk. This suggests the agents are not interchangeable even within the same drug class, and the respiratory fingerprint of each compound may reflect distinct mechanisms at the airway level.
Injectable semaglutide showed stronger associations in participants who had diabetes, which the authors flag as a relevant subgroup finding. Whether the signal in non-diabetic populations would replicate is a question the current data cannot fully answer.
The cautionary finding around saxagliptin
Not every association pointed in a favorable direction. Saxagliptin, a DPP-4 inhibitor, was associated with higher asthma risk, with a risk ratio of 2.09 and a confidence interval of 1.01 to 4.33. That confidence interval barely clears 1.0 on the lower bound, which means the finding carries uncertainty, but the direction of the association stands out clearly against the rest of the results.
The authors conclude that saxagliptin may warrant caution in people prone to asthma, and they call for prospective evaluation rather than treating this as a definitive contraindication. It is worth noting that DPP-4 inhibition as a mechanism is not new to respiratory concerns. DPP-4 degrades more than just GLP-1, and some substrates it processes have roles in airway inflammation, which provides at least a plausible biological backdrop for the observed signal.
The divergence between saxagliptin and the rest of the DPP-4 class also reinforces a theme running through the entire analysis: agent-level differences within the same drug class can be substantial, and class-level generalizations may mislead.
Methodological considerations
A network meta-analysis of this scale has real strengths. Pooling 128 randomized trials and 316,832 participants provides statistical power that no single trial could match, and the network structure allows indirect comparisons that would otherwise be unavailable. The investigators searched eight databases and preregistered the study with PROSPERO, lending additional credibility to the design.
Heterogeneity, the variation in results across trials, was assessed with tau-squared and related statistics. The authors report no major heterogeneity, inconsistency, or small-study effects, which reduces some common concerns about bias in large meta-analyses. Comparison-adjusted funnel plots and Egger regression were used to check for publication bias.
Still, important limitations apply. Network meta-analyses inherit the limitations of the trials they pool. If the original trials were not designed to measure ACOS as a primary endpoint, the ACOS event counts may be incomplete or variably defined across studies. The subgroup signals, particularly those in diabetic populations, require confirmation in prospective studies built specifically to test respiratory outcomes. The authors explicitly call for that next step.
Why these findings interest the research community
ACOS sits at an intersection where diabetes research, pulmonology, and inflammatory biology all converge. GLP-1 receptor agonists and SGLT2 inhibitors have already accumulated evidence of cardiovascular and renal protection beyond their glucose-lowering effects. The current meta-analysis adds respiratory outcomes to that evolving picture.
Mechanistically, several threads are plausible. GLP-1 receptors are expressed in lung tissue. SGLT2 inhibitors reduce systemic inflammation and fluid overload, both of which can worsen airway disease. None of these pathways have been proven to drive the ACOS associations seen here, but the literature suggests they provide a biologically coherent framework for further investigation.
For researchers studying peptides that interact with metabolic and inflammatory pathways, the specificity of the findings is notable. Two agents from the same drug class can produce opposite directional signals on the same respiratory outcome. That level of granularity points toward mechanisms that deserve investigation at the molecular level, not just the population level.
