External validation and clinical utility of prognostic prediction models for gestational diabetes mellitus: a prospective cohort study.
- 作者列表："Meertens LJE","Scheepers HCJ","van Kuijk SMJ","Roeleveld N","Aardenburg R","van Dooren IMA","Langenveld J","Zwaan IM","Spaanderman MEA","van Gelder MMHJ","Smits LJ
INTRODUCTION:We performed an independent validation study of all published first trimester prediction models, containing non-invasive predictors, for the risk of gestational diabetes mellitus (GDM). Furthermore, the clinical potential of the best performing models was evaluated. MATERIAL AND METHODS:Systemically selected prediction models from the literature were validated in a Dutch prospective cohort using data from Expect Study I and PRIDE Study. The predictive performance of the models was evaluated by discrimination and calibration. Clinical utility was assessed using decision curve analysis. Screening performance measures were calculated at different risk thresholds for the best model and compared with current selective screening strategies. RESULTS:The validation cohort included 5260 women. GDM was diagnosed in 127 women (2.4%). The discriminative performance of the 12 included models ranged from 68 to 75%. Nearly all models overestimated the risk. After recalibration, agreement between the observed outcomes and predicted probabilities improved for most models. CONCLUSIONS:The best performing prediction models showed acceptable performance measures and may enable more personalized based antenatal care for women at risk of developing GDM compared to current applied strategies.
简介: 我们对所有已发表的妊娠早期预测模型进行了一项独立的验证研究，其中包含妊娠期糖尿病 (GDM) 风险的非侵入性预测因子。此外，评估了最佳表现模型的临床潜力。 材料和方法: 使用预期研究 I 和 PRIDE 研究的数据，在荷兰前瞻性队列中验证了来自文献的系统选择的预测模型。通过判别和校准评价模型的预测性能。使用决策曲线分析评估临床效用。在最佳模型的不同风险阈值下计算筛选性能指标，并与当前的选择性筛选策略进行比较。 结果: 验证队列包括 5260 名女性。127 例 (2.4%) 诊断为 GDM。12 个纳入模型的判别性能范围为 68 ~ 75%。几乎所有的模型都高估了风险。重新校准后，大多数模型的观察结果和预测概率之间的一致性得到改善。 结论: 与当前应用的策略相比，表现最好的预测模型显示出可接受的性能指标，并可能使有发生 GDM 风险的妇女能够获得更个性化的产前护理。
METHODS:Aims We aimed to develop a prediction model based on clinical and biochemical variables for gestational diabetes mellitus (GDM) based on the 2013 World Health Organization (WHO) criteria. Methods A total of 1843 women from a Belgian multi-centric prospective cohort study underwent universal screening for GDM. Using multivariable logistic regression analysis, a model to predict GDM was developed based on variables from early pregnancy. The performance of the model was assessed by receiver-operating characteristic (AUC) analysis. To account for over-optimism, an eightfold cross-validation was performed. The accuracy was compared with two validated models (van Leeuwen and Teede). Results A history with a first degree relative with diabetes, a history of smoking before pregnancy, a history of GDM, Asian origin, age, height and BMI were independent predictors for GDM with an AUC of 0.72 [95% confidence interval (CI) 0.69–0.76)]; after cross-validation, the AUC was 0.68 (95% CI 0.64–0.72). Adding biochemical variables, a history of a first degree relative with diabetes, a history of GDM, non-Caucasian origin, age, height, weight, fasting plasma glucose, triglycerides and HbA_1c were independent predictors for GDM, with an AUC of the model of 0.76 (95% CI 0.72–0.79); after cross-validation, the AUC was 0.72 (95% CI 0.66–0.78), compared to an AUC of 0.67 (95% CI 0.63–0.71) using the van Leeuwen model and an AUC of 0.66 (95% CI 0.62–0.70) using the Teede model. Conclusions A model based on easy to use variables in early pregnancy has a moderate accuracy to predict GDM based on the 2013 WHO criteria.
METHODS:Leveraging the availability of nationwide electronic health records from over 500,000 pregnancies in Israel, a machine-learning approach offers an alternative means of predicting gestational diabetes at high accuracy in the early stages of pregnancy. Gestational diabetes mellitus (GDM) poses increased risk of short- and long-term complications for mother and offspring^ 1 – 4 . GDM is typically diagnosed at 24–28 weeks of gestation, but earlier detection is desirable as this may prevent or considerably reduce the risk of adverse pregnancy outcomes^ 5 , 6 . Here we used a machine-learning approach to predict GDM on retrospective data of 588,622 pregnancies in Israel for which comprehensive electronic health records were available. Our models predict GDM with high accuracy even at pregnancy initiation (area under the receiver operating curve (auROC) = 0.85), substantially outperforming a baseline risk score (auROC = 0.68). We validated our results on both a future validation set and a geographical validation set from the most populated city in Israel, Jerusalem, thereby emulating real-world performance. Interrogating our model, we uncovered previously unreported risk factors, including results of previous pregnancy glucose challenge tests. Finally, we devised a simpler model based on just nine questions that a patient could answer, with only a modest reduction in accuracy (auROC = 0.80). Overall, our models may allow early-stage intervention in high-risk women, as well as a cost-effective screening approach that could avoid the need for glucose tolerance tests by identifying low-risk women. Future prospective studies and studies on additional populations are needed to assess the real-world clinical utility of the model.
METHODS::Repurposing of currently approved medications is an attractive option for the development of novel treatment strategies against physiological and infectious diseases. The antidiabetic sulfonylurea glyburide has demonstrated off-target capacity to inhibit activation of the NLRP3 inflammasome in a variety of disease models, including vaginal candidiasis, caused primarily by the fungal pathogen Candida albicans Therefore, we sought to determine which of the currently approved sulfonylurea drugs prevent the release of interleukin 1β (IL-1β), a major inflammasome effector, during C. albicans challenge of the human macrophage-like THP1 cell line. Findings revealed that the second-generation antidiabetics (glyburide, glisoxepide, gliquidone, and glimepiride), which exhibit greater antidiabetic efficacy than prior iterations, demonstrated anti-inflammatory effects with various degrees of potency as determined by calculation of 50% inhibitory concentrations (IC50s). These same compounds were also effective in reducing IL-1β release during noninfectious inflammasome activation (e.g., induced by lipopolysaccharide [LPS] plus ATP), suggesting that their anti-inflammatory activity is not specific to C. albicans challenge. Moreover, treatment with sulfonylurea drugs did not impact C. albicans growth and filamentation or THP1 viability. Finally, the use of ECE1 and Candidalysin deletion mutants, along with isogenic NLRP3-/- cells, demonstrated that both Candidalysin and NLRP3 are required for IL-1β secretion, further confirming that sulfonylureas suppress inflammasome signaling. Moreover, challenge of THP1 cells with synthetic Candidalysin peptide demonstrated that this toxin is sufficient to activate the inflammasome. Treatment with the experimental inflammasome inhibitor MCC950 led to similar blockade of IL-1β release, suggesting that Candidalysin-mediated inflammasome activation can be inhibited independently of potassium efflux. Together, these results demonstrate that the second-generation antidiabetic sulfonylureas retain anti-inflammatory activity and may be considered for repurposing against immunopathological diseases, including vaginal candidiasis.