Plasma proteomic signatures of a direct measure of insulin sensitivity in two population cohorts
Authors' conclusion
A plasma proteomic signature identified using a standard LASSO approach improves the cross-sectional estimation of the M value over routine clinical variables. However, a small subset of these proteins identified using a stability selection algorithm affords much of this improvement, especially when considering cross-cohort analyses. Our approach provides opportunities to improve the identification of insulin-resistant individuals at risk of insulin resistance-related adverse health consequences.
Study context
- Population
- Participants of the RISC study and Uppsala Longitudinal Study of Adult Men (ULSAM) with fasting plasma proteomics (828 proteins) and euglycaemic-hyperinsulinaemic clamp966 RISC + 745 ULSAM · Participants of the RISC study 966 · ULSAM 745
- Exposure
- Clinical variables plus plasma proteins vs routine clinical variables alone for estimating the EIC M value
- Outcome
- Proportion of M value variance explained (R2)
Methodological detailstwo population cohorts, high-throughput proximity extension assay, least absolute shrinkage and selection operator (LASSO) models tested within and across cohorts
- Two population cohorts
- High-throughput proximity extension assay
- LASSO models tested within and across cohorts
Main finding
As reported in the abstract
RISC: R2 with clinical variables
0.237
95% CI 0.178, 0.303
RISC: R2 with clinical variables + 53 proteins
0.456
0.372, 0.536
ULSAM: R2 with clinical variables
0.443
0.360, 0.530
ULSAM: R2 with clinical variables + 61 proteins
0.632
0.569, 0.698
Observational data: associations, not effects.
Further findings
Cross-cohort M value R2 (standard LASSO)
- Trained RISC, tested ULSAM (51 proteins)
- 0.491 [0.433, 0.539]
- Trained ULSAM, tested RISC (67 proteins)
- 0.369 [0.331, 0.416]
M value R2 with randomised LASSO and stability selection (2 proteins per cohort, 3 unique)
- RISC
- 0.352 (0.266, 0.439)
- ULSAM
- 0.495 (0.404, 0.585)
- RISC to ULSAM
- 0.444 [0.391, 0.497]
- ULSAM to RISC
- 0.348 [0.300, 0.396]
Protein-only models
- Models of proteins alone were as effective as models with clinical variables and proteins (standard or randomised LASSO)
- Most consistently selected protein: IGF-binding protein 2
Original publication
Composed by AI from the published abstract and checked against it. How Briefs are made →