Age-modulated association between recombinant LH supplementation and ongoing pregnancy and miscarriage: a real-world overlap-weighted IVF cohort study

Introduction Recombinant luteinizing hormone (r-hLH) supplementation during controlled ovarian stimulation remains controversial and criteria for patient selection are not established. Methods In this retrospective cohort study, 1,983 ICSI cycles contributed by 1,491 patients were classified as stimulated with recombinant FSH alone (414 cycles) or with recombinant FSH plus r-hLH (1,569 cycles). Images from 12,092 oocytes in the time-lapse-imaged subset were captured immediately post-ICSI and scored retrospectively with an artificial intelligence platform generating the MAGENTA Score; embryos were cultured in a time-lapse system for morphokinetic assessment. Propensity-score overlap weighting on maternal age was applied overall and within age strata (<36, 36-38 and >38 years); restricted cubic splines modelled age-dependent associations with treatment. Results Fertilization computed over all injected oocytes recorded in the clinical database was essentially identical between protocols (73.7% r-hFSH only vs 72.9% r-hFSH + r-hLH). When restricted to the time-lapse-imaged subset, fertilization was higher with r-hLH (88.1% vs 85.7%, p<0.001); this difference may partly reflect selection into time-lapse culture and should be interpreted with that caveat. In the overall cohort, LH supplementation was further associated with faster morphokinetic events, higher blastulation (55.0% vs 51.3%, p=0.033), lower miscarriage (7.0% vs 12.2%, p=0.024), and higher ongoing pregnancy (54.5% vs 48.5%, p=0.049). In women <36 (95.2% vs 90.9%, p=0.030) and 36-38 years (86.6% vs 77.9%, p<0.001), fertilization in the time-lapse subset was higher with LH supplementation. In women >38 years, LH was associated with shorter morphokinetic timings, higher MAGENTA Score (5.91 vs 5.71, p=0.044) and KIDScore (4.29 vs 3.94, p=0.027), higher fertilization in the subset (95.1% vs 86.5%, p=0.016) and blastulation (50.8% vs 43.0%, p<0.001), lower miscarriage (6.8% vs 14.8%, p=0.045), and higher ongoing pregnancy (45.7% vs 34.1%, p=0.015). Because ongoing pregnancy was derived as clinical pregnancy minus miscarriage, these two outcomes are not independent. Discussion r-hLH supplementation was associated with an age-dependent pattern of laboratory and clinical outcomes, negligible in younger women and progressively more favourable at advanced reproductive age. Because adjustment was limited to maternal age and the design is observational, these associations should be regarded as hypothesis-generating.