Data mining and clinical observational study on the association between smoking and premature ovarian insufficiency

This exploratory study investigates the association between smoking and premature ovarian insufficiency (POI) to inform smoking cessation strategies and support women's reproductive health.

Using literature data mining and association rule analysis, we constructed a protein-protein interaction (PPI) network to identify potential core targets associated with POI. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment analyses were performed to explore biological functions and signaling pathways potentially involved in POI pathophysiology. In a parallel clinical study, we measured urinary cotinine levels in Chinese patients presenting with infrequent menstruation (May 2020-May 2025). Participants were categorized according to established clinical criteria for the presence or absence of POI, and statistical analyses were conducted to evaluate the association between cotinine exposure and POI risk.

Data mining identified six cigarette smoke components associated with POI-related ovarian dysfunction: nicotine, benzo[a]pyrene (BaP), nicotine-derived nitrosamine ketone (NNK), acrolein, 1, 3-butadiene, and cotinine. Topological network analysis suggested a potential central role for these compounds in tobacco-related POI mechanisms. However, degree, betweenness, and closeness centrality only reflect topological position within the constructed network and do not confirm dominant biological function. Moreover, more extensively studied compounds tend to have more documented molecular targets in public databases, which may artificially inflate their network centrality. GO enrichment indicated that these compounds may interfere with granulosa cell proliferation by modulating the cell cycle, while KEGG analysis linked them to pathways involved in progesterone-mediated oocyte maturation, cell cycle regulation, and oocyte meiosis. In the clinical cohort, higher urinary cotinine levels were associated with increased POI prevalence (quartile analysis) and a nonlinearly elevated odds risk (restricted cubic spline), particularly above the median. Urinary cotinine demonstrated weak positive associations with follicle-stimulating hormone (FSH) and luteinizing hormone (LH), but no significant initial association with anti-Müllerian hormone (AMH). Logistic regression suggested a moderate predictive value for POI, which persisted after multivariable adjustment and subgroup analyses. Notably, cotinine exhibited a nonlinear association with AMH (minimal change below the median, followed by a sharper increase above it), while maintaining linear relationships with FSH and LH. Multivariable linear regression indicated consistent positive associations between cotinine and all three hormonal markers.

These findings suggest that smoking may be associated with POI through plausible biological pathways identified via network pharmacology, and that urinary cotinine levels correlate with both POI risk and key ovarian reserve markers. Given the observational and exploratory nature of this study, the results raise hypotheses regarding smoke-associated ovarian dysfunction rather than establishing definitive causality, and they support further prospective research alongside targeted clinical and public health interventions.