Methods: Research on early detection of Cytomegalovirus (CMV) mediated sub-clinical rejection (sAR) is in its nascence. This study explores the efficacy of urine-based biomarkers for sAR risk prognostication. Urine samples were serially collected (1 week- 2yrs) across 125 unique renal allograft recipients from UCSF and UCLA. These CMV+ (blood CMV-PCR) patients received valganciclovir therapy until CMV- for 3 consecutive months.

Analysis: We computed a machine-learning based rejection prediction score incorporating urine cell-free DNA (cfDNA), methylated cfDNA, total protein, CXCL10, clusterin and creatinine. Rejection was determined between scores of 32-100; higher score associated with biopsy severity (Yang et al, STM 2021). In a cross-sectional analysis, Kruskal-Wallis test was performed on the urine-score distribution, to determine sAR during CMV+. Subsequently UCSF samples were analyzed longitudinally to establish the relationship between CMV and sAR; this was validated independently in UCLA samples.

Results: Significant differences were observed in scores >32 between biopsy-confirmed AR vs. stable (p: 0.0034) and CMV vs. stable (p=0.003). Longitudinal analysis yielded 3 primary sAR patterns following CMV infection; (i)persistence of sAR leading to new development of chronic allograft injury; (ii)early sAR (score >32) followed by a further increase in score (>55) with biopsy confirmed cAR; (iii)sAR (>32) followed by recovery of injury with CMV resolution and a decline in score <32).

Conclusions: We provide a urine-based method for detecting sAR at the time of CMV infection, and its potential clinical utility for titrating use of anti-viral therapy and recipient immunosuppression load.