AMC researchers find way to detect kidney transplant rejection with a blood drop

Korean researchers have developed a method to detect kidney transplant rejection early with a drop of serum from a patient's blood.

Although kidney transplants greatly enhance survival and quality of life for end-stage renal disease patients, the risk of rejection remains. Diagnosing rejection caused by antibodies and T-cells attacking the transplanted kidney also requires invasive biopsies.

Led by Professor Kim Jun-ki from the Department of Convergence Medicine at Asan Medical Center (AMC) and Professor Shin Sung from the Department of Kidney and Pancreas Transplantation at the University of Ulsan College of Medicine, the team succeeded in using surface-enhanced Raman spectroscopy (SERS) and an AI-based algorithm to detect transplant rejection in kidney transplant patients. 

Their findings, published in the journal Biosensors and Bioelectronics in June, suggest that a minimally invasive, AI-based, high-sensitivity technique could provide a more precise diagnosis of transplant rejection.

Typically, after a kidney transplant, patients undergo a biopsy using a needle with a gauge of 16 to 18 (about 1.5mm in diameter and nine to 12 cm in length) to detect rejection. The tissue is then chemically analyzed and graded according to the Banff classification system used for the pathological classification of kidney transplant rejection. However, the invasiveness of biopsies limits their frequent use and poses a high risk of complications such as bleeding.

Besides, kidney function is monitored through blood tests measuring creatinine and blood urea nitrogen (BUN) levels. Still, these methods lack the sensitivity needed for early detection of kidney damage due to rejection. 

The research team focused on SERS, which enhances the sensitivity of detecting low-concentration analytes through localized-surface plasmon resonance (LSPR) modes on metal substrates. 

The researchers developed gold-zinc oxide (Au-ZnO) nanoparticle-based SERS, which demonstrated high reliability and sensitivity in experiments for diagnosing atherosclerosis and cancer. The high-sensitivity diagnostic results were achieved by analyzing the spectrum patterns generated by various nanotechnology-based biomarkers with machine learning algorithms.

The team hypothesized that analyzing the Raman patterns produced by various biomarkers in the serum with AI technology could lead to a more precise diagnosis of transplant rejection, based on the multifaceted elements of the Banff classification.

Professor Shin's team classified patient samples into three groups through prognostic analysis of transplant rejection: no rejection, antibody-mediated rejection (ABMR), and T-cell-mediated rejection (TCMR). By evaluating kidney damage and function post-transplant, they validated the Raman signal analysis process and assessed the diagnostic accuracy of the Raman signals relative to kidney damage.

The SERS and AI-based discriminant analysis achieved accuracy rates of 93.53 percent and 98.82 percent for each type of rejection, using principal component analysis (PCA) to reduce variables through dimensionality reduction before performing discriminant analysis. 

The team also confirmed that this AI-based analysis technique could monitor patients with mixed types of rejection.

"Our technology, combining SERS chips and AI algorithms, successfully identified rejection patterns in clinical samples, which is very promising," Professor Kim said.

Professor Shin said, "With further research and validation, this minimally invasive method could allow kidney transplant patients to be diagnosed with rejection through a simple blood test."

The full study detailing these findings is published in Biosensors and Bioelectronics (IF 10.7).