Zhou Yulai, Huang Zixuan, Ren Jiabin, Liu Chunxiao, Yu Tiantian, Wu Weibin, Fan Jianxia
School of Medicine, International Peace Maternity and Child Health Hospital, Shanghai Jiao Tong University, Shanghai, China.
Shanghai Key Laboratory of Embryo Original Diseases, Shanghai, China.
BMC Pregnancy Childbirth. 2025 Aug 6;25(1):821. doi: 10.1186/s12884-025-07931-3.
This study aimed to identify risk factors for adverse pregnancy outcomes (APO) in women with isolated maternal hypothyroxinemia (IMH) and to develop a nomogram for predicting APO risk during routine antenatal visits.
Data from 1254 IMH pregnancies, collected between January 2016 and December 2018 at the International Peace Maternal and Child Health Hospital (IPMCH) in Shanghai, China, were analyzed. APO, the primary outcome, included preterm birth (PTB), macrosomia, gestational diabetes mellitus (GDM), and hypertensive disorders of pregnancy (HDP). Multivariable logistic regression analyses identified risk factors for APO in IMH, and the least absolute shrinkage and selection operator (LASSO) regression algorithm was applied for feature selection, with cross-validation determining the optimal tuning parameter (λ). A nomogram based on the multivariable logistic regression model was developed to estimate APO risk, validated using 500 bootstrap resampling and a 2019 cohort of 745 women from the same institution. Model performance was assessed using discrimination and calibration metrics.
Among 57 features, maternal age, elevated diastolic blood pressure, a family history of hypertension or diabetes, pre-conception BMI, fasting plasma triglycerides, fasting plasma glucose, HbA1c, and monocyte counts in the first trimester were identified via LASSO regression as significant risk factors, with pre-conception BMI being the most prominent. The area under the receiver operating characteristic curve (ROC-AUC), sensitivity, specificity, PPV, NPV and F1 score were 0.73 (95% CI: 0.70-0.76), 62.09%, 72.11%, 59.03%, 74.60% and 60.49% respectively in the training cohort, and 0.72 (95% CI: 0.68-0.76), 60.01%, 72.65%, 56.15%, 75.12% and 58.01% respectively in the validation cohort. Calibration plots demonstrated good consistency between predicted and observed APO incidences, and decision curve analyses indicated a positive net clinical benefit across cohorts.
This study presented a robust early-pregnancy predictive model for APO in IMH women, offering clinicians a valuable tool for risk assessment and the early identification of high-risk pregnancies.
本研究旨在确定单纯孕妇甲状腺素水平低下（IMH）女性不良妊娠结局（APO）的风险因素，并制定一种列线图，用于在常规产前检查期间预测APO风险。
分析了2016年1月至2018年12月在中国上海国际和平妇幼保健院（IPMCH）收集的1254例IMH妊娠的数据。主要结局APO包括早产（PTB）、巨大儿、妊娠期糖尿病（GDM）和妊娠高血压疾病（HDP）。多变量逻辑回归分析确定了IMH中APO的风险因素，并应用最小绝对收缩和选择算子（LASSO）回归算法进行特征选择，通过交叉验证确定最佳调整参数（λ）。基于多变量逻辑回归模型开发了一种列线图，用于估计APO风险，并使用500次自助重采样和来自同一机构的2019年745名女性队列进行验证。使用区分度和校准指标评估模型性能。
在57个特征中，通过LASSO回归确定母亲年龄、舒张压升高、高血压或糖尿病家族史、孕前体重指数、空腹血浆甘油三酯、空腹血糖、糖化血红蛋白和孕早期单核细胞计数为显著风险因素，其中孕前体重指数最为突出。在训练队列中，受试者操作特征曲线下面积（ROC-AUC）、灵敏度、特异度、阳性预测值、阴性预测值和F1分数分别为0.73（95%CI：0.70-0.76）、62.09%、72.11%、59.03%、74.60%和60.49%，在验证队列中分别为0.72（95%CI：0.68-0.76）、60.01%、72.65%、56.15%、75.12%和58.01%。校准图显示预测的和观察到的APO发生率之间具有良好的一致性，决策曲线分析表明各队列的净临床效益为正。
本研究提出了一种针对IMH女性APO的强大的早期妊娠预测模型，为临床医生提供了一个用于风险评估和早期识别高危妊娠的有价值工具。
