Abstract Study question Does trophoblasts of PCOS patients with aberrant iron metabolism, and how does metformin ameliorate androgen-induced ferroptosis in trophoblasts in vivo and in vitro? Summary answer Metformin ameliorates placental development and attenuates spontaneous miscarriage rate in early gestation by inhibiting ferritinophagy in trophoblasts of PCOS patients with hyperandrogenism. What is known already PCOS is a common reproductive endocrine disease in women of reproductive age, and a large proportion of patients present with hyperandrogenism. Studies have found that the incidences of spontaneous miscarriage and adverse pregnancy outcomes in PCOS patients are higher than those in healthy controls, which placental dysfunction is an important factor. It is related to mitochondrial dysfunction, oxidative stress and metabolic disorders. Ferroptosis is an iron-dependent and phospholipid peroxidation-mediated cell death. Studies have shown that iron ions participate in a variety of biochemical reactions in humans, and its abnormal metabolism may lead to some diseases especially in the reproductive system. Study design, size, duration Villi at gestational weeks of 6-8 were collected from 18 healthy controls and 18 women with PCOS. Human trophoblast stem cells (hTSCs) were isolated from fresh villi (6-8 weeks, n = 3) and cultured with testosterone, ferroptosis or autophagy inhibitors, metformin, or NCOA4 small interfering RNA. A dihydrotestosterone (DHT)-induced PCOS C57BL/6 mouse model was treated with or without metformin (3 months before mating) (n = 20 per group). Mouse placentas were obtained at embryonic days 11.5 (n = 50 per group). Participants/materials, setting, methods The levels of iron ions and lipid peroxidation were measured by PGSK staining, BODIPYC11 staining and MDA assay kit. The expressions of ferroptosis-related genes and proteins were assessed by qPCR, western blot and immunofluorescence. The binding of androgen receptor (AR) and NCOA4 to the putative androgen response element (ARE) was examined by ChIP, EMSA and luciferase assay. The influence of metformin on ferritinophagy in placentas of PCOS mice was confirmed by western blot and immunofluorescence. Main results and the role of chance The levels of iron ions and lipid peroxidation in villi at gestational weeks of 6-8 were higher in PCOS patients than in healthy controls. In vitro experiments, testosterone-induced ferroptosis was ameliorated by ferroptosis inhibitors, and ferritinophagy was mitigated by autophagy inhibitors or NCOA4 knockdown in primary cultured hTSCs. Testosterone induced ARs into the nucleus, where the recruitment of AR homodimers along with NCOA4 to the positive ARE on the SQSTM1 promoter, leading to autophagy induction and degradation of ferritin heavy chain 1 (FTH1) in autophagosomes, resulting in the release of ferritin bound irons as free ferrous irons. Metformin ameliorated testosterone-induced ferroptosis in hTSCs to a certain extent by inhibiting ferritinophagy. Compared to those of control mice, the placentas of DHT-treated PCOS mice showed increased levels of iron ions and lipid peroxidation, and altered expressions of ferroptosis-related proteins. Hyperandrogenism also induced ferritinophagy in labyrinthine zone of placentas in PCOS mice. Metformin ameliorated placental development was associated with decreased iron ions accumulation by inhibiting ferritinophagy in PCOS mouse placentas, which observably attenuated embryo absorption rate and spontaneous miscarriage incidence in the early gestation of PCOS mice. Limitations, reasons for caution The numbers of PCOS patients and mouse placentas in this study were small. The placental labyrinthine zone was not separated from the endocrine junctional zone and maternal decidua for molecular analyses. In addition, it would be better to isolated and cultured mouse TSCs to explore mechanisms on ferritinophagy in further. Wider implications of the findings This study provides further evidence for metformin as a PCOS treatment, which ameliorates placental development and attenuates spontaneous miscarriage rate in early gestation. Trial registration number not applicable