The review presents the techniques as potential non-invasive adjuncts to existing transplant assessment rather than established clinical tests. The analytical basis for this approach lies in the characteristic patterns produced when molecules scatter laser light. A Raman spectrum can represent the combined molecular composition of proteins, nucleic acids, lipids, metabolites and other substances within a sample.

In kidney transplantation, investigators have examined signals associated with collagen, creatinine, amino acids, extracellular-matrix remodeling, immune-complex deposition and oxidative stress, all of which are connected in the source material with acute or chronic rejection. Some of these processes may become detectable before changes appear in a standard biopsy. The method does not necessarily identify a single disease-specific molecule, however; much of its discriminatory power comes from mathematical analysis of the combined spectral pattern.

That distinction was evident in a study of serum collected from kidney-transplant recipients classified by contemporary biopsy, Banff scoring and immunohistochemistry. The sample included seven recipients with antibody-mediated rejection, three with T-cell-mediated rejection and seven with no major abnormality, with the last group serving as the control. A sensing chip with gold-coated zinc-oxide nanorods was used to amplify the SERS signal from 5 microliters of serum.

Although the raw spectra appeared broadly similar and did not show distinct group-specific peaks, two machine-learning approaches separated the groups with reported accuracies of 93.53 percent and 98.82 percent. The findings indicate that computational classification can extract information from complex differences in molecular signals, even when individual spectral features are not visually obvious. Urine has attracted attention because it can be obtained without a needle or incision and can contain both urinary and blood-derived indicators of graft status.

In one study, investigators collected 60 samples from 20 kidney-transplant recipients: one day before transplantation, on the first postoperative day and on postoperative day seven. The researchers used two SERS approaches, one based on a colloidal silver film and another incorporating a 4-mercaptopyridine probe to improve sensitivity for small metabolic structures. The principal metabolites examined were choline, creatinine and urea.

Classification models reported accuracies of 96 to 100 percent between transplantation stages, while postoperative samples showed increased signals associated with amino acids and lipids and decreased signals associated with creatinine and nucleic acids relative to preoperative samples. Because the comparison included only transplantation time points and no healthy or chronic-kidney-disease groups, the results demonstrate stage-dependent spectral differences but do not establish specificity for rejection. Animal experiments offer additional evidence that urine SERS can reflect disease-associated and treatment-associated biology.

In one rat model, renal-allograft recipients received either anti-TGF-beta antibody or a nonspecific control antibody rather than standard pharmaceutical immunosuppression, with 10 animals per group. A silver-nanoparticle substrate was used for SERS, and principal component-linear discriminant analysis distinguished the antibody-treated and control groups with 90 percent sensitivity and 70 percent specificity. A second model used obese, genetically diabetic ZSF1 rats, with 10 animals receiving control chow and 11 receiving chow supplemented with grape powder.

SERS analysis distinguished the diet groups with 72.7 percent sensitivity and 60 percent specificity. Receiver operating characteristic areas under the curve were approximately 0.85 for the transplant model and 0.80 for the diabetic model. These findings show that reproducible urine signatures occurred in experimentally induced renal disease, but they are animal-model observations and do not directly establish human transplant benefit.

The connection between spectral information and routine biochemical measures was examined in 110 transplant recipients whose urine and blood were collected within three days after transplantation. For SERS analysis, urine was mixed with silver nanoparticles, air-dried for one hour at room temperature and scanned. Partial least-squares regression then related the spectra to laboratory measurements.

Urine spectra correlated with urinary creatinine at R² = 0.8106, urinary urea at R² = 0.7808 and urinary protein at R² = 0.4660. Spectral analysis also predicted serum creatinine at R² = 0.7628 and blood urea nitrogen at R² = 0.6539, with all five relationships reported as statistically significant at p less than 0.01. These results show that urine-derived spectra can contain information about established biochemical indicators of kidney function, although the study did not report the detection of rejection itself.

Donor kidneys may present another potential application because Raman methods can examine molecular features before or after implantation. In a study of urine from 30 deceased donors, participants were divided into three groups of 10: donors with biopsy-proven acute tubular necrosis, donors whose recipients experienced delayed graft function, and controls with neither condition. SERS with iodide-modified silver nanoparticles generated distinct spectra for the groups, with sensitivity greater than 90 percent for distinguishing acute tubular necrosis from both delayed graft function and controls.

Differences were also observed between the delayed-graft-function and control groups. Many kidneys with acute tubular necrosis nevertheless produced acceptable transplant outcomes. The findings suggest that a urinary spectral profile may provide information beyond traditional donor-risk categories, but the study did not test whether applying those profiles to organ selection would change clinical decisions.

A separate label-free Raman-imaging study investigated hydroxyethyl starch, an expander used during donor resuscitation, in renal-allograft biopsies obtained three months after transplantation. The study included 26 recipients: 20 grafts came from donors exposed to hydroxyethyl starch within 72 hours before procurement, while six control grafts came from donors who received no colloid. A vibrational band near 480 inverse centimeters was associated with hydroxyethyl starch and was absent in the controls, although positivity was determined through a fitting algorithm that incorporated the full spectral profile rather than this band alone.

Approximately 40 percent of exposed biopsies had signals above the positivity threshold, and none of the controls did. Stronger relative signals correlated with higher estimated glomerular filtration rate at three months, with r = 0.88 and p less than 0.001, and with a lower kidney donor risk index, with r = −0.72 and p less than 0.001. The researchers indicated that the relationship between hydroxyethyl starch signal and graft function probably reflected underlying graft quality rather than a beneficial effect of the expander.

One possible explanation was that higher-functioning kidneys might reabsorb and retain more hydroxyethyl starch within tubular cells. Raman imaging is invasive in this setting because it was performed on biopsy tissue, despite the broader potential for non-invasive sampling through fluids. The donor-assessment evidence therefore spans pre-transplant urine analysis and post-transplant imaging of excised tissue rather than a single non-invasive donor-testing procedure.

Beyond organ chemistry, Raman spectroscopy can detect the biochemical remodeling that accompanies T-cell activation, a process involved in acute rejection. In one in-vitro experiment, human T lymphocytes isolated from healthy participants were divided among resting, nonspecifically activated and alloantigen-activated conditions. Nonspecific activation used CD3/CD28-coated beads and interleukin-2, whereas alloantigen activation was produced through a two-way mixed lymphocyte culture involving cells from nonrelated individuals and antigen-presenting cells.

The researchers measured 75 resting-cell samples, 75 CD3/CD28-activated samples and 40 alloantigen-activated samples. At selected wavelengths assigned to nucleic acids, amino acids and mixed protein–nucleic-acid contributions, nonspecifically activated cells had peak magnitudes approximately 4.3 to 32 percent below those of resting cells. Classification between nonspecific and alloantigen activation reported 100 percent sensitivity and specificity under the experimental conditions.

A related study compared resting, mitomycin-C-inactivated and alloimmune-activated human T cells, yielding 35 spectra for resting cells, 28 for inactivated cells and 51 for alloimmune-activated cells. Significant differences appeared at 1182 and 1195 inverse centimeters, wavelengths associated with amino-acid vibrations and protein structures. One statistical analysis reported 95.7 percent sensitivity and 100 percent specificity for distinguishing the three conditions, while discriminant-function analysis reported 89.3 percent sensitivity and 93.8 percent specificity.

Together, the two experiments indicate that Raman spectra contain measurable differences associated with immune-cell activation. They were performed with isolated cells manipulated in vitro, however, rather than demonstrating clinical detection of acute rejection in transplant recipients. The review identifies sample standardization and computational interpretation as central requirements for translating Raman methods into clinical use.

Subtle differences may require advanced methods such as machine learning, while the reviewed studies differ in their samples, endpoints and statistical approaches. The reported applications span three areas: monitoring graft-related changes in urine, blood or serum; assessing donor-associated molecular features and potential risk markers; and identifying biochemical changes in activated T lymphocytes. Within the evidence presented, the strongest conclusion is that Raman spectroscopy and SERS can reveal transplantation-associated molecular patterns with non-invasive sampling in some settings and may serve as useful additions to biopsy, laboratory testing and clinical assessment.

Whether those capabilities can improve graft outcomes, reduce biopsy use or expand safe use of donor organs remains an unanswered clinical question rather than an established result of the studies reviewed.

Subject of Research: Raman spectroscopy and SERS for kidney-transplant monitoring, donor assessment and immune-activation analysis

Article Title: From Spectra to Clinical Insight: Raman Spectroscopy in Kidney Transplant Monitoring and Donor Assessment

Article References: Peveri, E., Asthana, A., Byers, L. N., Senger, R., Issa, S., Robertson, J., Dehoff, D. M., & Orlando, G. (2026). From Spectra to Clinical Insight: Raman Spectroscopy in Kidney Transplant Monitoring and Donor Assessment. Current Transplantation Reports, 13(1), Article 34. https://doi.org/10.1007/s40472-026-00531-w

Image Credits: AI Generated

DOI: 10.1007/s40472-026-00531-w

Keywords: Raman spectroscopy, surface-enhanced Raman scattering, kidney transplantation, graft rejection, donor assessment, urine biomarkers, T-cell activation, graft dysfunction, non-invasive diagnostics, Spectra, Clinical, Insight