Patient-reported endometriosis symptom change during GLP-1 receptor agonist use, and its relationship to weight loss: an international cross-sectional survey (GLEAM)

Technical Summary: GLEAM Survey on GLP-1 RAs and Endometriosis Symptoms

Problem Statement
Patient communities have reported anecdotal improvements in endometriosis and adenomyosis symptoms while using glucagon-like peptide-1 receptor agonists (GLP-1 RAs), such as semaglutide and tirzepatide. A prior international survey (Gholiof et al.) documented this signal but could not determine whether the reported improvement was driven by the drugs' non-weight mechanisms or simply by the degree of weight loss, as weight reduction itself reduces pelvic inflammation and estrogen production. The central research gap was whether symptom improvement is independent of the magnitude of weight loss and baseline body mass index (BMI).

Methodology
The GLEAM study is an international, anonymous, cross-sectional online survey conducted via a self-hosted LimeSurvey instance.

• Design & Registration: The study was pre-registered on the Open Science Framework (OSF) prior to data collection. It is observational, uncontrolled, and hypothesis-generating rather than confirmatory.
• Participants: Eligible respondents were adults assigned female at birth with a self-reported diagnosis of endometriosis and/or adenomyosis who had ever used a GLP-1 RA. Recruitment occurred through 40+ international patient communities and social media channels using neutral wording to invite all outcomes (improvement, no change, or worsening).
• Data Collection: The survey collected demographics, body metrics (height, start weight, current weight), exposure details (agent, duration, reason for use), and outcomes. The primary outcome was a 7-point Global Rating of Change (GRoC) for overall symptoms and nine specific domains (pelvic pain, dysmenorrhoea, bloating, etc.).
• Analysis Plan: The primary endpoint was the proportion of respondents reporting improvement (overall GRoC ≥ "slightly better" AND ≥1 domain improved). The analysis tested the association between improvement and weight change/baseline BMI using logistic regression, adjusting for agent family, duration, and reason for use. The study explicitly addressed collinearity between weight loss and baseline BMI. Sensitivity analyses included confirmed-diagnosis-only subsets, exclusion of weight-motivated starters, and complete-case analyses.
• Sample: Recruitment closed at the pre-registered cap of 250 completed responses. The final analytic sample consisted of 227 respondents with usable weight data.

Key Results

• Prevalence of Improvement: 63% (95% CI 57–69) of the analytic sample met the strict primary improvement endpoint. A looser "any-domain" improvement rate was 84% (95% CI 78–88).
• Weight Loss Association: The adjusted odds ratio (OR) for improvement per percentage point of weight loss was near null at 1.02 (95% CI 0.98–1.05). The across-band trend test was non-significant (p = 0.219). Improvement was reported across all weight-loss bands, including those with minimal or no weight loss.
• Baseline BMI and Indication: In contrast to the weight-loss degree, baseline BMI showed a statistically significant inverse association with improvement (adjusted OR 0.94 per BMI unit, p = 0.009); respondents with lower baseline BMI reported more improvement. Furthermore, starting the medication specifically for weight loss was strongly inversely associated with improvement (adjusted OR 0.24, p < 0.001).
• Subgroup Analysis: The pre-specified key subgroup (normal BMI and <5% weight loss) contained only 14 respondents (12 improved, 86%), a size too small for inference and far below the expected sample proportion, reflecting a selection bias toward higher-BMI, weight-motivated users.
• Sensitivity: The near-null weight-loss OR remained stable across sensitivity analyses (range 1.01–1.03), whereas the headline proportion of improvement varied significantly (from 54% in complete-case to 81% when excluding weight-motivated starters).

Key Contributions

1. First Pre-Registered Test: This is the first pre-registered study to explicitly test whether the GLP-1 RA symptom-improvement signal in endometriosis is independent of the degree of weight loss.
2. Hypothesis Generation: The study converts anecdotal community observations into a testable, pre-registered hypothesis. It provides a rationale for a future controlled trial by demonstrating that reported improvement does not scale linearly with weight loss.
3. Confounding Identification: The analysis highlights that while weight loss degree may not predict improvement, baseline BMI and the reason for starting the drug (weight loss vs. other) are significant predictors, suggesting strong confounding by indication and selection bias.
4. Methodological Transparency: The paper provides a fully reproducible, macro-driven analysis pipeline with open code, detailing how self-reported data, unit errors, and missingness were handled.

Significance and Claims
The authors explicitly state that this study does not establish efficacy or causation. The near-null association between weight loss and improvement is consistent with a weight-independent pharmacological effect, but it is equally consistent with non-drug artifacts such as regression to the mean, expectation effects, self-selection, and recall bias.

The paper concludes that the observed pattern—improvement occurring across weight-change groups without scaling to the amount lost—warrants a controlled randomized trial to definitively separate drug effects from weight loss and expectation effects. The authors emphasize that GLP-1 RAs are not approved for endometriosis and carry reproductive safety caveats (e.g., pre-conception washout requirements), so these findings do not support clinical use. The study serves as a "building block, not a verdict," narrowing the question for future prospective research.