T2D Islet Biomarkers
Huang et al. 2022 identified four diagnostic biomarkers for T2D using LASSO regression on 111 DEGs from GSE41762 (islet bulk RNA-seq):
Biomarker Panel
| Gene | AUC | Known Role in Diabetes |
|---|---|---|
| SLC2A2 (GLUT2) | 0.9 | Glucose transporter; biallelic mutations cause neonatal diabetes and Fanconi-Bickel syndrome. |
| SERPINF1 (PEDF) | 0.83 | Serpin family member; associated with obesity, insulin resistance, and leptin levels. |
| RASGRP1 | 0.83 | Ras guanyl nucleotide exchange factor; involved in VEGF-mediated angiogenesis and metformin pro-angiogenic effects. |
| CHL1 | 0.83 | Neural cell adhesion molecule; regulates β-cell proliferation via ERK and p53 pathways. |
Nomogram Performance
- Combined model (SLC2A2 + SERPINF1 + RASGRP1 + CHL1): AUC = 0.902.
- Calibration and decision-curve analysis confirmed nomogram validity.
Tissue Context
These biomarkers were identified from pancreatic islet tissue, not blood or PBMCs. Their utility as circulating or PBMC biomarkers in T2D has not been tested.
Tang et al. 2026 Biomarker Panel
Tang et al. 2026 identified a complementary set of four T2DM signature genes using scRNA-seq (GSE221156, pancreatic islet) with LASSO regression, validated by ROC on an independent muscle microarray (GSE29221) and qRT-PCR on peripheral blood:
| Gene | AUC (GSE29221) | qRT-PCR Direction | Known Function |
|---|---|---|---|
| BUB1 | 0.931 | Up in T2DM | Mitotic checkpoint kinase; cell cycle regulation |
| CTSB | 0.882 | Up in T2DM | Cathepsin B; lysosomal cysteine protease |
| PNLIP | 0.819 | Up in T2DM | Pancreatic lipase; triglyceride digestion |
| NAMPT | 0.694 | Down in T2DM | NAD+ biosynthesis; also pro-inflammatory eNAMPT cytokine |
Comparison to Huang et al. 2022 Panel
- Tang et al. used scRNA-seq → DEGs → LASSO whereas Huang et al. used bulk RNA-seq → DEGs → LASSO — different starting data types but convergent methodology.
- Tang et al. validated via cross-tissue (islet→muscle→blood) translation; Huang et al. validated via nomogram on islet data alone.
- Neither panel shares individual gene members, suggesting that islet-derived biomarker candidates are sensitive to the specific tissue compartment, cohort, and analytical pipeline used.
- Combined, both panels support the feasibility of transcriptomic biomarker discovery from islet tissue but highlight that reproducible candidates across studies remain limited.