A lipidome-wide association study of the lipoprotein insulin resistance index

Minoo Bagheri, Hemant K. Tiwari, Anarina L. Murillo, Rafet Al-Tobasei, Donna K. Arnett, Tobias Kind, Dinesh Kumar Barupal, Sili Fan, Oliver Fiehn, Jeff O’Connell, May E. Montasser, Stella Aslibekyan, Marguerite R. IrvinLipids 2020 · DOI

Abstract: Non-storage lipidomics species show a significant inverse association with the LPIR metabolic dysfunction score and present a promising focus for future therapeutic and prevention studies.

Introduction

The Lipoprotein Insulin Resistance (LPIR) score is a powerful tool for predicting Type 2 Diabetes (T2D) and insulin resistance. Calculated via Nuclear Magnetic Resonance (NMR), it measures the size and concentration of lipoprotein particles like VLDL, LDL, and HDL. While the clinical utility of the LPIR score is well-established, the underlying molecular machinery—specifically the individual lipid species driving these associations—remains largely unexplored. Identifying these specific biomarkers could offer new avenues for early detection and therapeutic intervention.

Methods

Researchers employed a lipidome-wide association study approach using two distinct cohorts: the GOLDN discovery cohort (980 participants of European American descent) and the HAPI Heart replication cohort (590 participants). Lipidomic profiling was performed using Ultra-Performance Liquid Chromatography coupled to Mass Spectrometry (UPLC–QTOFMS), identifying 413 unique lipid species.

To manage the high dimensionality of the lipidomic data, the team used Principal Component Factor Analysis (PCFA) to group lipids into three functional factors: Storage Lipids (Factor 1), Non-storage Lipids (Factor 2), and Mixed Lipids (Factor 3).

PCA scree plot indicating the explained variance of the first 10 components

Linear mixed-effects models, adjusted for age, sex, BMI, and clinical center, were used to assess associations. In the replication cohort, HOMA-IR was used as a proxy for the LPIR score due to data availability.

Key Results

The analysis revealed that 319 out of 413 lipids were significantly associated with the LPIR score. The factor analysis demonstrated a clear divergence in metabolic impact:

  • Factor 1 (Storage Lipids): Primarily triglycerides and diglycerides, positively associated with LPIR (β = 0.025, p = 4.52 × 10⁻⁷¹).
  • Factor 2 (Non-storage Lipids): Primarily phospholipids, inversely associated with LPIR (β = -0.013, p = 2.78 × 10⁻¹⁸).
  • Factor 3 (Mixed Lipids): Positively associated with LPIR (β = 0.021, p = 5.84 × 10⁻⁴¹).
Adjusted correlation between selected lipids and LPIR score and its components

These patterns were robustly replicated in the HAPI Heart study, confirming the consistency of this lipidomic signature across different populations.

Discussion and Insights

The study highlights a fundamental "Storage vs. Structure" divide in the lipidome. Storage lipids (TGs and DGs), particularly those with shorter, saturated fatty acid chains, are strong indicators of metabolic dysfunction. Conversely, non-storage phospholipids like phosphatidylcholines show a protective inverse association. This suggests that metabolic health depends not just on total fat levels, but on the specific ratio of storage to structural lipids.

While the cross-sectional nature of the study prevents establishing causality, these findings provide a detailed molecular map of insulin resistance. Future work should explore whether targeting these specific lipid species can prevent the progression of metabolic disease.

Limitations

The study's primary limitations include its cross-sectional design, the focus on populations of European descent, and the use of proxy measures in the replication phase. Further research in more diverse cohorts is necessary to generalize these findings.

Summary updated

Abstract

Non-storage lipidomics species show a significant inverse association with the LPIR metabolic dysfunction score and present a promising focus for future therapeutic and prevention studies.
BibTeX
@inproceedings{bagheri2020lipidome,
  author = {Minoo Bagheri and Hemant K. Tiwari and Anarina L. Murillo and Rafet Al-Tobasei and Donna K. Arnett and Tobias Kind and Dinesh Kumar Barupal and Sili Fan and Oliver Fiehn and Jeff O’Connell and May E. Montasser and Stella Aslibekyan and Marguerite R. Irvin},
  title = {{A lipidome-wide association study of the lipoprotein insulin resistance index}},
  booktitle = {Lipids},
  year = {2020},
  doi = {10.1186/s12944-020-01321-8},
}
Figure