A new method for classifying different phenotypes of kidney transplantation

Area of Science:

• Nephrology
• Immunology
• Biomarker Discovery

Background:

• Kidney transplantation is the optimal treatment for end-stage renal disease.
• Allograft failure is often caused by inflammation-induced rejection.
• Identifying inflammation biomarkers can predict renal allograft rejection.

Purpose of the Study:

• To characterize inflammation factors for predicting kidney transplant rejection.
• To develop a classifier for different kidney transplant phenotypes.
• To investigate the role of inflammation in chronic rejection.

Main Methods:

• Collected serum from kidney transplant patients with stable function, impaired function, acute rejection, and chronic rejection.
• Measured expression profiles of 40 inflammatory proteins using quantitative protein microarrays.
• Reduced data dimensionality with partial least squares (PLS) and classified phenotypes using support vector machines (SVMs).
• Constructed a protein-protein interaction (PPI) network.

Main Results:

• Identified significant differences in inflammation proteins between chronic rejection and other groups (30 between CR and ST, 16 between CR and AR, 13 between CR and UNST).
• Revealed a PPI network involving 33 inflammatory proteins, highlighting a potential role for ICAM-1 in chronic rejection.
• Developed a PLS-SVMs model using two principal components, achieving 77.5% accuracy in classifying chronic rejection.
• Created GPS-CKT software for phenotype classification.

Conclusions:

• A strong correlation exists between inflammation and kidney transplantation outcomes.
• Network biomarkers, rather than single factors, show potential for classifying different kidney transplant phenotypes.
• The developed classifier and software offer a promising tool for predicting chronic rejection.