Natural language processing is emerging as a key application of artificial intelligence in oncology. This systematic review aims to evaluate the performance and methodological frameworks of natural language processing systems in gynecologic oncology. We conducted a systematic review following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis 2020 guidelines. MEDLINE, EMBASE, and Web of Science were searched for studies published between January 2015 and February 2025. Outcomes were synthesized across 3 research questions: the accuracy of natural language processing systems used as consultation service platforms; the accuracy of natural language processing systems used as clinical decision support systems; and the benchmarking methodologies applied, including their associated methodological outcomes. Consultation service platforms deliver general medical information, whereas clinical decision support systems provide recommendations that are integrated into the patient's clinical workflow. This review analyzed 12 retrospective studies. Consultation service platforms were less accurate than clinicians (60% vs 86.7%) and rated lower in response quality (2.96/5 vs 4.2/5) but outperformed guideline-based answers (1.54/2 vs 1.38/2). In cervical cancer, ChatGPT surpassed experts (7.0 vs 6.1). ChatGPT-4 showed a concordance of 70% with the National Comprehensive Cancer Network and 60% with the European Society of Gynaecological Oncology guidelines in clinical decision support tasks, with an overall recommendation accuracy of 75%. IBM Watson achieved a 72.8% concordance with guidelines. Prompting was applied from 100% to 37.5% across studies. Qualitative benchmarking varied across studies: 83.3% used clinical guidelines and 37.5% of consultation service platforms studies used expert answers. Four- or 5-point scales and binary scoring were used to assess consultation service platforms and clinical decision support systems, respectively. Clinicians remain superior in complex reasoning, but natural language processing systems demonstrate robust performance in guideline-driven tasks, with advantages in speed, readability, and reproducibility. However, performance declined in nuanced scenarios and among under-represented patient sub-groups. Large language models currently play a supportive rather than substitutive role in gynecologic oncology.

Maternal smoking (MS) during pregnancy is linked to well-documented adverse health effects for the mother and foetus, however the role of fetal sex was largely overlooked. Primary cultures of male and female human umbilical vein endothelial cells (MHUVECs and FHUVECs, respectively) were used. IL-6, IL-8, and TNF-α levels were measured in HUVECs supernatant. The expression of genes and proteins (oestrogen receptors (ERs), Hsp90, Beclin-1, p62, LC3, LAMP-1 and Parkin), as well as the expression of miR-29a-3p, miR-29b-3p, miR-126-3p, miR-133a-3p, and miR-146a-5p were analysed in cells obtained from foetuses born to non-smoking and smoking mothers. In HUVECs from foetuses born to non-smoking mothers, Beclin-1 protein was higher in MHUVECs (1.8 fold increase), while Parkin, Hsp90 proteins, and miR-146a-5p were elevated in FHUVECs (2.2, 2.6, and 2.1 fold increase, respectively), with no other significant differences. MS amplified these sex differences, with specific effects based on foetus sex. FHUVECs obtained from foetus born to smoking mothers showed higher levels of IL-8 (1399.36 ± 123.96 pg/ml for FHUVECs vs 655.11 ± 215.94; pg/ml for MHUVECs; P < 0.001), Hsp90 gene and protein (3.3 and 2.6 fold increase), and ERβ protein and Beclin-1 gene (2.1, and 4.9 fold increase), and lower levels of miR-29b-3p, miR-133a-3p, and miR-146a-5p than MHUVECs (0.27, 0.68, and 0.1 fold change, respectively). This study shows that primary HUVECs from fetuses born to smoking mothers retain a memory of smoking effects, with sex differences in gene expression, miRNA profiles, and autophagic responses, suggesting that maternal smoking impacts endothelial cell physiology in a sex-dependent manner.

Background and Objectives: Ectopic pregnancy (EP) is a life-threatening medical and surgical condition. Tubal stump EPs and heterotopic pregnancies can occur after in vitro fertilization-embryo transfer (IVF-ET), even after salpingectomy. The purpose of this study is to investigate the risk factors, diagnosis, and treatment of tubal stump EPs after IVF-ET in patients with prior salpingectomy or adnexectomy. We also aim to evaluate the intrauterine pregnancy (IUP) outcome in cases of heterotopic pregnancy in this population. Materials and Methods: This systematic review (PROSPERO CRD42023352959) followed PRISMA guidelines. A literature search of MEDLINE^®, Scopus, Web of Science, and clinicaltrials.gov was conducted on 30 April 2024. We included studies on tubal stump EP after IVF-ET in patients with previous salpingectomy or adnexectomy and created a qualitative summary. Results: We included 40 studies reporting on 57 patients (58 EP episodes). Most patients (69.0%) had prior bilateral salpingectomy. Tubal rupture occurred in 69.6% of cases, with 69.0% of these cases reporting hemoperitoneum. Abdominal pain was the most frequent symptom (71.7%). Heterotopic pregnancy occurred in 60.0% of cases (82.7% singletons). The IUP outcome was delivery in 81.9% of cases, with 95.5% of singletons delivering at term, compared with 40.0% of twins. The surgical approach (laparoscopy vs. laparotomy) did not change the IUP outcome. Tubal stump excision (74.1%) was the most common treatment. Overall, the certainty of the evidence was judged as moderate to very low according to the GRADE-CERQual approach, mainly due to small sample sizes, observational designs, and heterogeneity among studies. Conclusions: This review, the first on this topic, provides key data for counselling patients with a tubal stump heterotopic pregnancy. Despite its rarity, close follow-up until 8–10 weeks is recommended for IVF-ET patients with positive β-hCG, monitoring for abdominal pain. Successful management (expectant, medical, or surgical) should be guided by β-hCG levels and ultrasound findings (e.g., absence of heartbeat). Medical treatment shows encouraging obstetric outcomes and warrants further research.

To assess the accuracy of ultrasound in identifying residual cervical tumors following conization, using histology as the gold standard. This retrospective, single-center study included patients with early-stage cervical cancer (International Federation of Gynecology and Obstetrics IA1-IB1 and early IB2) who underwent conization followed by secondary surgical treatment (re-conization, hysterectomy, or trachelectomy). Patients with locally advanced disease were excluded. All patients underwent trans-vaginal or trans-rectal ultrasound after conization and before definitive surgery between January 2015 and May 2025. Ultrasound findings were classified as follows: a) absence of residual tumor, b) presence of residual tumor, or c) uncertain (suggestive of either residual disease or post-conization artifacts). For statistical analysis, uncertain cases were conservatively considered positive for residual disease. Additionally, a secondary analysis was performed in which "uncertain" cases were classified as negative. Ultrasound results were compared with final histopathology, and diagnostic performance metrics with 95% confidence intervals were calculated. A total of 166 patients were included (mean age, 44.4 ± 9.2 years). Residual tumor was found at final histology in 75 patients (45.2%), while 91 (54.8%) had no residual disease. The median tumor diameter at conization was 9.8 ± 6.2 mm, and the mean interval between conization and ultrasound was 61.2 ± 45.5 days. Ultrasound correctly identified residual tumor in 40/75 cases (53.3%), while false-positive findings occurred in 16/91 patients (17.6%). Uncertain ultrasound findings were reported in 43 cases (25.9%), of which 16 (37.2%) had residual disease at histology. Residual tumors were predominantly hypoechoic and showed moderate (45.0%) or rich (35.0%) vascularization on color Doppler imaging. Ultrasound demonstrated a sensitivity of 74.7% (95% confidence interval 64.8 to 84.5), specificity of 52.7% (95% confidence interval 42.5 to 63.0), and overall accuracy of 62.7% (95% confidence interval 53.1 to 72.2). This study shows that ultrasound has sub-optimal performance in detecting residual tumor after conization for cervical cancer, indicating that optimal timing and accurate assessment of residual disease remain clinically relevant challenges.

The growing capabilities of Large Language Models (LLMs) in understanding and generating clinical text are transforming the processing of unstructured medical data. This study presents a prompt-based framework for extracting structured information from Multidisciplinary Tumor Board (MTB) reports on ovarian cancer, written in Italian. Focusing on the anamnestic section, the main features (about Demographic, Scores, Comorbidities, Treatments and Gynecological History) are extracted through in-context learning prompts, and evaluated with the LLM-as-a-judge majority-voting scheme. Results show high accuracy across macroareas and strong agreement among judge models, confirming the robustness of the approach. The occurrence-based analysis further highlights how variable frequency influences extraction reliability, ensuring a more balanced interpretation of model performance. Overall, the framework enables reliable LLM-based information extraction and supports the creation of structured clinical knowledge for oncology applications.