Abstract
BACKGROUND: Aberrant vascular endothelial growth factor (VEGF)-driven angiogenesis is central to the establishment and persistence of endometriosis. Although numerous anti-angiogenic compounds have been tested, evidence remains fragmented, and no comparative framework guides the selection of agents compatible with fertility preservation. OBJECTIVE: To identify and rank vascular-targeted pharmacotherapies that most effectively regress endometriotic lesions and could be prioritised for fertility-sparing clinical translation. METHODS: We conducted a systematic review and frequentist network meta-analysis (PROSPERO CRD420251082905) of controlled studies evaluating VEGF-directed agents in endometriosis. Six databases and two trial registries were searched from inception to 28 July 2025. Thirty-one studies met inclusion criteria (five early-phase human trials, six patient-derived cell models, two baboon experiments, and 18 rodent experiments) investigating 23 pharmacological agents. Primary outcomes were lesion area, lesion number, and VEGF expression; secondary outcomes included microvascular density, endometrial cell proliferation, and apoptosis. Random-effects pairwise and network models (R 4.3 "netmeta") generated standardised mean differences (SMDs) with 95% CIs and SUCRA rankings. Transitivity, heterogeneity (τ (2), I (2)), and inconsistency were formally assessed. Drug-gene target intersections across six cheminformatic databases mapped mechanistic convergence. RESULTS: Curcumin achieved the greatest lesion-size reduction (SMD = -1.08, 95% CI = -1.38 to -0.79) and the steepest fall in microvascular density (-16%), while cetrorelix most effectively reduced lesion number (SMD = -0.78, 95% CI = -1.36 to -0.20). Retinoic acid and bevacizumab halved VEGF expression. Global inconsistency was non-significant (P > 0.18) and heterogeneity moderate (I (2) ≤ 65%). A network analysis identified a 16-gene hub (CASP3, MAPK1/3, AKT1, STAT3, etc.) underpinning effective drugs; curcumin targeted 14 of these nodes. LIMITATIONS: Fifty-eight per cent of the data derived from rodent models revealed that these models do not menstruate and incompletely recapitulate human endometriosis. Most experiments reported short-term surrogate outcomes and rarely measured pain, fertility, or quality of life. Risk of bias was frequently moderate to high, and therefore, small-study or publication bias cannot be excluded. CONCLUSION: This study provides the first, exploratory comparative synthesis of VEGF-directed pharmacotherapies for endometriosis across pre-clinical and early-phase human models. Apparent efficacy rankings, including the high placement of curcumin, should be interpreted as relative signals within a limited and biased dataset rather than as evidence of inherent biological superiority or clinical readiness. These findings are best viewed as hypothesis-generating and may assist in prioritising agents for more rigorous translational and clinical research, rather than supporting specific treatment recommendations. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD420251082905.