血浆蛋白质组与2型糖尿病及相关特征的关联:来自纵向KORA S4/F4/FF4研究的结果
基于KORA S4队列1653名基线参与者的233种血浆蛋白,中位随访13.5年,横断面分析n=1300、纵向分析n=1143。分别识别出14、24和4种蛋白与糖尿病前期、新诊断2型糖尿病和发病2型糖尿病相关,孟德尔随机化提示脂蛋白脂肪酶对2型糖尿病和空腹胰岛素有因果效应,12种生物标志物联合使预测性能改善(ΔAUC 0.0219;95%CI 0.0052,0.0624)。
要点
- 在KORA S4队列中识别出与糖尿病前期、新诊断及发病2型糖尿病相关的多组血浆蛋白。
- 孟德尔随机化分析提示脂蛋白脂肪酶对2型糖尿病和空腹胰岛素存在因果效应。
- 12种优先Lasso筛选的生物标志物联合可改善预测性能(ΔAUC 0.0219;95%CI 0.0052,0.0624)。
核心结论
- 研究识别出参与糖代谢紊乱和2型糖尿病发生的新候选蛋白,并证实了既往报道的蛋白。
- 作者认为这些发现强调蛋白在2型糖尿病发病机制中的重要性。
- 作者认为所识别的推定蛋白可作为糖尿病治疗和预防的潜在药理靶点。
研究设计与样本
- 研究为纵向KORA S4/F4/FF4队列研究,基线测量1653名参与者的233种血浆蛋白,中位随访时间13.5年。
- 横断面分析采用逻辑回归(n=1300),纵向分析采用校正区间删失数据的Cox回归(n=1143)。
- 应用两水平增长模型分析蛋白与重复测量的空腹血糖、2小时血糖、空腹胰岛素、HOMA-B、HOMA-IR、HbA1c的关联。
- 采用两样本孟德尔随机化分析探讨因果关联。
- 在Framingham-Offspring风险评分组分基础上用priority-Lasso构建预测模型,并以AUC评价预测准确性。
主要发现
- 分别有14、24和4种蛋白与现患糖尿病前期(即糖耐量受损和/或空腹血糖受损)、现患新诊断2型糖尿病和发病2型糖尿病相关,其中28种蛋白重叠。
- IL-17D、IL-18受体1、碳酸酐酶-5A、IL-1受体2型(IL-1RT2)和基质细胞外磷酸糖蛋白为新候选蛋白。
- 胰岛素样生长因子结合蛋白2(IGFBP2)、脂蛋白脂肪酶(LPL)和对氧磷酶3(PON3)与发病2型糖尿病呈反向关联,成纤维细胞生长因子21与发病2型糖尿病呈正向关联。
- LPL与葡萄糖相关特征的纵向变化相关,IGFBP2和PON3与胰岛素相关和葡萄糖相关特征的纵向变化均相关。
- 孟德尔随机化分析提示LPL对2型糖尿病和空腹胰岛素存在因果效应。
关键数值与推荐
- 与现患糖尿病前期相关的蛋白为14种。
- 与现患新诊断2型糖尿病相关的蛋白为24种。
- 与发病2型糖尿病相关的蛋白为4种,三组间重叠蛋白28种。
- 加入12种priority-Lasso筛选的生物标志物后预测性能改善:ΔAUC 0.0219;95%CI 0.0052,0.0624。
临床怎么用
- 原文未据此给出推荐。
局限性
- 原文未报告局限性。
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
AIMS/HYPOTHESIS: This study aimed to elucidate the aetiological role of plasma proteins in glucose metabolism and type 2 diabetes development. METHODS: We measured 233 proteins at baseline in 1653 participants from the Cooperative Health Research in the Region of Augsburg (KORA) S4 cohort study (median follow-up time: 13.5 years). We used logistic regression in the cross-sectional analysis (n=1300), and Cox regression accounting for interval-censored data in the longitudinal analysis (n=1143). We further applied two-level growth models to investigate associations with repeatedly measured traits (fasting glucose, 2 h glucose, fasting insulin, HOMA-B, HOMA-IR, HbA1c), and two-sample Mendelian randomisation analysis to investigate causal associations. Moreover, we built prediction models using priority-Lasso on top of Framingham-Offspring Risk Score components and evaluated the prediction accuracy through AUC. RESULTS: We identified 14, 24 and four proteins associated with prevalent prediabetes (i.e. impaired glucose tolerance and/or impaired fasting glucose), prevalent newly diagnosed type 2 diabetes and incident type 2 diabetes, respectively (28 overlapping proteins). Of these, IL-17D, IL-18 receptor 1, carbonic anhydrase-5A, IL-1 receptor type 2 (IL-1RT2) and matrix extracellular phosphoglycoprotein were novel candidates. IGF binding protein 2 (IGFBP2), lipoprotein lipase (LPL) and paraoxonase 3 (PON3) were inversely associated while fibroblast growth factor 21 was positively associated with incident type 2 diabetes. LPL was longitudinally linked with change in glucose-related traits, while IGFBP2 and PON3 were linked with changes in both insulin- and glucose-related traits. Mendelian randomisation analysis suggested causal effects of LPL on type 2 diabetes and fasting insulin. The simultaneous addition of 12 priority-Lasso-selected biomarkers (IGFBP2, IL-18, IL-17D, complement component C1q receptor, V-set and immunoglobulin domain-containing protein 2, IL-1RT2, LPL, CUB domain-containing protein 1, vascular endothelial growth factor D, PON3, C-C motif chemokine 4 and tartrate-resistant acid phosphatase type 5) significantly improved the predictive performance (ΔAUC 0.0219; 95% CI 0.0052, 0.0624). CONCLUSIONS/INTERPRETATION: We identified new candidates involved in the development of derangements in glucose metabolism and type 2 diabetes and confirmed previously reported proteins. Our findings underscore the importance of proteins in the pathogenesis of type 2 diabetes and the identified putative proteins can function as potential pharmacological targets for diabetes treatment and prevention.
引用
- Luo H, Bauer A, Nano J, et al. Associations of plasma proteomics with type 2 diabetes and related traits: results from the longitudinal KORA S4/F4/FF4 Study. Diabetologia. 2023. PMID: 37308750. doi:10.1007/s00125-023-05943-2
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。