Round Spermatid Injection (ROSI) as a Last Resort in High-Risk Azoospermia: Chain of Outcome Metrics and Real-World Safety Signals

Background/Objectives: Round spermatid injection (ROSI) is considered an experimental “last resort” option for couples with severe male factor infertility when mature spermatozoa cannot be obtained. We aimed to identify which stage of the clinical chain most strongly constrains overall success in routine practice and to describe the observed safety signal. Methods: We conducted a retrospective single-center cohort study of 221 consecutive ROSI-evaluated cycles (2021–2024). Outcomes were analyzed using a chain-of-outcome framework with explicit denominators: cycle-level feasibility (≥1 injected oocyte), two pronuclei (2PN) formation per injected oocyte, blastocyst development per 2PN, transfer per blastocyst cycle, and clinical pregnancy per transfer and per initiated cycle. Exact (Clopper–Pearson) 95% confidence intervals (CIs) were reported. Results: ROSI feasibility was observed in 5 of 221 initiated cycles (2.3%; exact 95% CI 0.7–5.2). Among the five transfer procedures performed after successful progression through upstream stages, clinical pregnancy occurred in four (80.0%; exact 95% CI 28.4–99.5). At the initiated-cycle level, overall clinical pregnancy was 4 of 221 cycles (1.8%; exact 95% CI 0.5–4.6). Conclusions: The overall effectiveness of ROSI remained low at the initiated-cycle level because very few cycles reached procedural feasibility and early attrition remained substantial. Conditional downstream outcomes may appear favorable only among the rare cycles reaching fertilization and transfer, while safety inference remains highly imprecise due to small denominators. Because only five cycles reached feasibility, all downstream conditional estimates remained highly unstable and sensitive to single-case variation.