DETERMINING BETWEEN T-CELL AND ANTIBODY-MEDIATED KIDNEY TRANSPLANT REJECTION NON-INVASIVELY BY URINARY EXFOLIATED PROXIMAL TUBULE CELL MULTISPECTRAL AUTOFLUORESCENCE

Differentiation between T-cell and antibody-mediated kidney transplant rejection is important, as the two main rejection subtypes have different mechanisms and responses to treatment. Precisely diagnosing the type of graft rejection allows for targeted immunosuppressive therapy and maximizes graft survival whilst minimizing risks of graft loss. There is currently a critical unmet need to develop non-invasive yet accurate methods for early diagnosis of post-transplant complications to allow for timely intervention and optimization of clinical outcomes. Our group previously demonstrated proof-of-concept that measurement of multispectral autofluorescence signals emitted from proximal tubule cells (PTCs) exfoliated into urine in kidney transplant recipients could differentiate between acute tubular necrosis, graft rejection and non-rejection interstitial fibrosis and tubular atrophy to an excellent degree of accuracy. Building on from this initial study, we have applied our novel non-invasive ‘liquid biopsy’ methodology (IP Australia Patent Number: AU2024904153) to determine its ability in reliably diagnosing between T-cell and antibody-mediated kidney transplant rejection. 

Adult patients over the age of 18 years who have undergone kidney transplantation and are subsequently diagnosed with either T-cell or antibody-mediated rejection following transplant kidney biopsy were recruited. Spot urine samples were collected on the same day just prior to the biopsy procedure. On assessment, urinary exfoliated PTCs were extracted using a validated immunomagnetic separation method based on anti-CD13 and anti-SGLT2 antibodies. Imaging of samples were performed on a customized multispectral Olympus IX83 microscope with a cooled, low-noise Nüvü™ EMCCD camera and a multi-LED light source. Image analysis involved data preprocessing followed by binary classification of the study groups based on cell multispectral autofluorescence features. Binary classification was performed through a random forest classifier and the AutoGluon machine-learning software developed from Amazon Web Services. Results were validated by 10-fold cross-validation.

10 patients with T-cell rejection and 10 patients with antibody-mediated rejection were included in this study (20 patients total). No statistically significant differences in key demographic and clinical characteristics between the two study groups (i.e. age, sex, number of days from transplant to biopsy, estimated glomerular filtration rate, urine albumin-creatinine ratio and serum creatinine) were observed. Assessment of cell autofluorescence features differentiated between the exfoliated PTCs of these two groups with an area under the receiver operating characteristic curve (AUC) value of 0.82 ± 0.07 (Figure 1).

These results provide early evidence that measurement of urinary exfoliated PTCs multispectral autofluorescence may differentiate between T-cell and antibody-mediated transplant graft rejection to a good degree of accuracy. Validation of our findings with larger multi-centre datasets are now required. Given the retrieval of human urine is non-invasive, convenient to obtain, and given our novel diagnostic methodology is robust, it is foreseeable that this approach could potentially be implemented within the acute transplant rejection scenario to guide clinical decision-making in the future.