Daily Endocrinology Research Analysis

BACKGROUND: Artificial intelligence (AI) models analyzing embryo time-lapse images have been developed to predict the likelihood of pregnancy following in vitro fertilization (IVF). However, limited research exists on methods ensuring AI consistency and reliability in clinical settings during its development and validation process. We present a methodology for developing and validating an AI model across multiple datasets to demonstrate reliable performance in evaluating blastocyst-stage embryos. METHODS: This multicenter analysis utilizes time-lapse images, pregnancy outcomes, and morphologic annotations from embryos collected at 10 IVF clinics across 9 countries between 2018 and 2022. The four-step methodology for developing and evaluating the AI model include: (I) curating annotated datasets that represent the intended clinical use case; (II) developing and optimizing the AI model; (III) evaluating the AI's performance by assessing its discriminative power and associations with pregnancy probability across variable data; and (IV) ensuring interpretability and explainability by correlating AI scores with relevant morphologic features of embryo quality. Three datasets were used: the training and validation dataset (n = 16,935 embryos), the blind test dataset (n = 1,708 embryos; 3 clinics), and the independent dataset (n = 7,445 embryos; 7 clinics) derived from previously unseen clinic cohorts. RESULTS: The AI was designed as a deep learning classifier ranking embryos by score according to their likelihood of clinical pregnancy. Higher AI score brackets were associated with increased fetal heartbeat (FH) likelihood across all evaluated datasets, showing a trend of increasing odds ratios (OR). The highest OR was observed in the top G4 bracket (test dataset G4 score ≥ 7.5: OR 3.84; independent dataset G4 score ≥ 7.5: OR 4.01), while the lowest was in the G1 bracket (test dataset G1 score < 4.0: OR 0.40; independent dataset G1 score < 4.0: OR 0.45). AI score brackets G2, G3, and G4 displayed OR values above 1.0 (P < 0.05), indicating linear associations with FH likelihood. Average AI scores were consistently higher for FH-positive than for FH-negative embryos within each age subgroup. Positive correlations were also observed between AI scores and key morphologic parameters used to predict embryo quality. CONCLUSIONS: Strong AI performance across multiple datasets demonstrates the value of our four-step methodology in developing and validating the AI as a reliable adjunct to embryo evaluation.

Distinguishing benign from malignant adrenocortical tumors (ACT) is not always easy, particularly for tumors with unclear malignant potential based on the histopathological features comprised of the Weiss score. Previous studies reported the potential utility of immunohistochemistry (IHC) markers to recognize malignancy, in particular the Insulin-like growth factor 2 (IGF2) and the proliferation marker, Ki-67. However, this information was not compiled before. Therefore, this review aimed to collect the evidence on the potential diagnosis utility of IGF2 and Ki-67 IHC staining. Additionally, a meta-analysis was performed to assess the Ki-67 accuracy to identify adrenocortical carcinoma. The systematic review and meta-analysis were conducted according to the PRISMA guidelines. From the 26 articles included in the systematic review, 21 articles provided individual data for IGF2 (n = 2) or for Ki-67 (n = 19), while 5 studies assessed both markers. IGF2 staining was positive in most carcinomas, in contrast to adenomas. However, the different immunostaining evaluation methods adopted among the studies impeded to perform a meta-analysis to assess IGF2 diagnostic accuracy. In contrast, for the most commonly used cut-off value of 5% stained cells, Ki-67 showed pooled specificity, sensitivity and log diagnostic odds ratio of 0.98 (95% CI 0.95 to 0.99), 0.82 (95% CI 0.65 to 0.92) and 4.26 (95% CI 3.40 to 5.12), respectively. At the 5% cut-off, Ki-67 demonstrated an excellent specificity to recognize malignant ACT. However. the moderate sensitivity observed indicates the need for further studies exploring alternative threshold values. Additionally, more studies using similar approaches are needed to assess the diagnostic accuracy of IGF2.Registration code in PROSPERO: CRD42022370389.

AIM: Transient hypothyroxinemia of prematurity (THOP) has been associated with suboptimal neurodevelopment. We aimed to assess neurodevelopment in very preterm infants with treated and untreated THOP. METHODS: This study was a multicentre, cohort study, based on prospectively collected data in four French level III neonatal intensive care units. Infants born before 32 weeks of gestation between 2009 and 2020 who underwent a thyroid function test were included. THOP was defined as low free thyroxine and unelevated thyroid stimulating hormone. Infants were classified as no THOP, treated THOP, and untreated THOP. The primary outcome was suboptimal neurodevelopment at 2 years of age evaluated by clinical examination. RESULTS: Three hundred and seventy-three infants (54% male) born at a median gestational age of 28 weeks of gestation were included. There was no significant difference in neurodevelopment at 2 years of age when comparing the no THOP to the THOP group (Odds Ratio (OR) 1.4, 95% confident Interval (CI) 0.8-2.3) nor when comparing the treated with the untreated THOP group (OR 0.8, 95% CI 0.3-1.9). Results remained unchanged after adjusting for confounding factors. CONCLUSION: In very preterm infants treated THOP was not associated with improved neurodevelopment compared to untreated THOP. Numerous biases could have limited treatment effect.