A Multicenter Randomized Controlled Trial Of Urine CXCL10 Monitoring Post-Kidney Transplant

Introduction: Elevated urine CXCL10 detects subclinical and early clinical rejection, decreases with rejection therapy and has a worse long-term graft prognosis. We hypothesized the early treatment of rejection, as detected by urine CXCL10, would improve important 12-mo transplant outcomes. 
Method: This biomarker-enriched RCT enrolled incident adult kidney patients for CXCL10 monitoring at 2, 4, 6, 8, 12, 16, 20, 24, 36 wks post-transplant [Fig 1A]. Patients with confirmed elevated CXCL10 were randomized to CXCL10-biopsy vs routine monitoring. The primary composite outcome was death-censored graft loss, clinical BPAR, 12-mo tubulitis, 12-mo iIFTA and dnDSA. The principal secondary outcome was eGFR. The trial was stopped early due to recruitment feasibility and new data on the limited efficacy TCMR therapy. 
Results: 260 patients on TAC/MPA/prednisone underwent CXCL10 monitoring and 72 were randomized to the Intervention (n=32) vs Control (n=40) Arms [Fig 1B].  Randomized patients were older (58 (IQR 50-65)) vs 51 (41-60) yr, p=0.0001), more males (78% vs 58%, p=0.003), with more diabetes (39% vs 26%, p=0.003). There was no difference in pre-transplant immunological risk between randomized and non-randomized patients (HLA eplet MM p=0.81, repeat transplant p>0.99, PRA p=0.70, pre-transplant DSA p>0.99). 
CXCL10 biopsies occurred at 4.37 (IQR 2.46-6.87) months post-transplant and detected subclinical Borderline (n=2); Banff 1B (n=1); Banff 2A (n=2); Banff 3 (n=1); chronic active AMR (n=1); BKVN (n=5); and pyelonephritis (n=1). There were numerically fewer primary endpoints in Intervention vs Control (8/32, 25% vs 13/40, 33%), but the primary composite outcome was not significant (RR 0.77, 95%CI 0.34-1.59, p=0.49). The Intervention vs Control Arms had less clinical BPAR (6.3% (2/32) vs 18% (7/40)); 12-mo tubulitis (9.4% vs 18%) and 12-mo iIFTA (3.1 vs 5.0%); but more dnDSA (6.3% vs 0%). One Intervention patient was not randomized per protocol timeframes and admitted with a clinical Banff 2A before CXCL10 biopsy occurred, leading to 1 of 2 clinical BPAR in Intervention Arm. Urine ACR was lower in the Intervention Arm, which is due in part to decreasing urine CXCL10 protein following treatment. There was no difference in serum creatinine or tacrolimus levels [Fig 2]. Graft loss (1) and death (4) occurred in non-randomized patients only.
Conclusion: Urine CXCL10 is a post-transplant marker that detects rejection early. There were numerically fewer primary endpoints in the Intervention vs Control Arms, but the overall primary composite outcome was not significant. This is an underpowered trial with significant type II error risk (target randomization 250). CXCL10-guided biopsy and treatment resulted in reduced 12-mo urine ACR compared to Control and Not Randomized patients. The effectiveness of the early detection of rejection by urine CXCL10 may be partly limited by incomplete response rates to standard TCMR therapy.