Research on Lipidomic Profiling and Biomarker Identification for Osteonecrosis of the Femoral Head

Abstract

Objectives: Abnormal lipid metabolism is increasingly recognized as a contributing factor to the development of osteonecrosis of the femoral head (ONFH). This study aimed to explore the lipidomic profiles of ONFH patients, focusing on distinguishing between traumatic ONFH (TONFH) and non-traumatic ONFH (NONFH) subtypes and identifying potential biomarkers for diagnosis and understanding pathogenesis. Methods: Plasma samples were collected from 92 ONFH patients (divided into TONFH and NONFH subtypes) and 33 healthy normal control (NC) participants. Lipidomic profiling was performed using ultra-high performance liquid chromatography-tandem mass spectrometry (UHPLC-MS/MS). Data analysis incorporated a machine learning-based feature selection method, least absolute shrinkage and selection operator (LASSO) regression, to identify significant lipid biomarkers. Results: Distinct lipidomic signatures were observed in both TONFH and NONFH groups compared to the NC group. LASSO regression identified 11 common lipid biomarkers that signify shared metabolic disruptions in both ONFH subtypes, several of which exhibited strong diagnostic performance with areas under the curve (AUCs) > 0.7. Additionally, subtype-specific lipid markers unique to TONFH and NONFH were identified, providing insights into the differential pathophysiological mechanisms underlying these subtypes. Conclusions: This study highlights the importance of lipidomic profiling in understanding ONFH-associated metabolic disorders and demonstrates the utility of machine learning approaches, such as LASSO regression, in high-dimensional data analysis. These findings not only improve disease characterization but also facilitate the discovery of diagnostic and mechanistic biomarkers, paving the way for more personalized therapeutic strategies in ONFH.

Keywords: LASSO; diagnostic biomarker; feature selection; lipidomic profile; osteonecrosis of the femoral head.

Figure 1

The Pearson correlation analysis among the QC samples based on the LC-MS/MS lipidomic profiling data. The plot illustrates the Pearson correlation coefficients for each pair of quality control (QC) samples in both the positive (pos) and negative (neg) ion modes, providing a measure of the reproducibility and reliability of the lipidomic data across the QC samples.

Figure 2

Lipid subclass analysis. A total of 664 lipid compounds were annotated in the positive mode, and 359 lipid compounds in the negative mode, each classified into their respective subclasses. The horizontal axis quantifies the number of lipid compounds, while the vertical axis lists the names of each lipid subclass. TAG: triacylglyceride; PC: phosphatidylcholine; SM: sphingomyelin; ACar: acylcarnitine; DAG: diacylglycerol; PE: phosphatidylethanolamines; Cer: ceramide; CE: cholesteryl ester; PA: phosphatidic acid; GlcCer: glucosylceramide; PG: phosphatidylglycerol; PS: phosphatidylserine; PI: phosphatidylinositol; GM3: ganglioside; GlcADG: glycosyldiacylglycerols; MGDG: monogalactosyldiacylglycerol; FAHFA: fatty acid esters of hydroxy fatty acids.

Figure 3

PLS-DA plots based on the lipidomic profiling in the positive (Pos) and negative (Neg) ion modes. The analysis was performed using the “mixOmics” package in the R platform, based on the lipidomic profiling data of the samples. (A) PLS-DA plots for TONFH and NC samples; (B) PLS-DA plots for NONFH and NC samples; and (C) PLS-DA plots for TONFH and NONFH samples.

Figure 4

Using a LASSO regression for lipid feature selection across the different groups. The data processing was performed using a LASSO function, and the graph plotting was carried out with the lassoPlot function in MATLAB. The parameter Lambda (λ) was used to determine the optimal set of lipid features. (A,B) Cross-validated mean squared error (MSE) of LASSO fit and trace plot of coefficients fit by LASSO for TONFH versus NC comparison; (C,D) cross-validated MSE of LASSO fit and trace plot of coefficients fit by LASSO for NONFH versus NC comparison; and (E,F) cross-validated MSE of LASSO fit and trace plot of coefficients fit by LASSO for TONFH versus NONFH comparison.

Figure 5

T-SNE plots based on lipidomic profiling. The analysis was performed using the “Rtsne” package in the R platform, based on the lipidomic profiling data of the samples. (A) T-SNE plots for TONFH versus NC samples utilizing the entire set of 1358 lipid features; (B) T-SNE plots for TONFH versus NC samples utilizing the 56 lipid features selected by LASSO; (C) T-SNE plots for NONFH versus NC samples utilizing the entire set of 1358 lipid features; (D) T-SNE plots for NONFH versus NC samples utilizing the 43 lipid features selected by LASSO; (E) T-SNE plots for NONFH versus NONFH samples utilizing the entire set of 1358 lipid features; and (F) T-SNE plots for TONFH versus NONFH samples utilizing the selected 67 lipid features selected by LASSO. Following the LASSO-based feature selection, the discrimination between the groups significantly improved, with a clearer separation of the samples for both the TONFH vs. NC and NONFH vs. NC comparisons.

Figure 6

Analysis of relative abundances of the 11 common lipid features. (A) Eleven lipid features were identified as overlapping in both TONFH and NONFH compared to the NC; of these, six (indicated in red) were elevated and five (indicated in green) were decreased in the disease conditions. (B) Relative abundances of the 11 common lipids across the three groups. Statistical analysis was performed using the Wilcoxon rank-sum test to compare the different groups. * p < 0.05, ^ p < 0.01, # p < 0.001.