One of the major issues in the clinical management of kidney recipients with antibody-mediated rejection (ABMR) is the lack of early and reliable markers that surrogate long-term allograft outcomes. We investigated the role of C1q-binding donor-specific anti-HLA antibodies (DSAs) for decision making in kidney recipients with ABMR receiving standard of care (SOC) treatment.
Among 1196 kidney recipients transplanted between 2008 and 2011, we prospectively enrolled all patients with biopsy-proven active ABMR according to the most recent Banff criteria, who received standardized treatment, incuding plasma exchange, high-dose intravenous immunoglobulin, rituximab and steroids. Patients were systematically assessed at the time of ABMR diagnosis and 3-month post-treatment initiation for clinical data, histological characteristics and DSA characteristics (specificity, mean fluorescence intensity [MFI] and C1q-binding capacity).
We included 139 kidney recipients with ABMR diagnosed at a median time of 15.5 months (IQR, 5.8-36.8) post-transplant receiving SOC treatment. At ABMR diagnosis, patients with C1q-binding DSAs (N=91, 65%) showed similar 6-year allograft survival compared to patients with non-C1q-binding DSAs (N=48, 35%) (66% vs 83%, p=0.07). At the time of post-treatment evaluation, patients with C1q-binding DSAs (N=38, 27%) had decreased 6-year graft survival compared to patients with non-C1q-binding DSAs (N=101, 73%) (44% vs 83%, p<0.001). After adjusting for GFR, proteinuria, all histological Banff scores and DSA characteristics, post-treatment C1q-binding DSAs were independent predictors of allograft loss (adjusted HR=2.6, p=0.007) and outperformed DSA MFI level to predict allograft loss (increase in C-statistic from 0.63 to 0.68; cNRI: 0.83, P<0.001; IDI: 0.08, P=0.002). We built a conditional inference tree for evidence-based clinical decision making including the most informative variables after treatment (cross-validated accuracy: 0.77): GFR, presence of cg lesion and presence of C1q-binding DSA. The decision tree segregated 5 groups of patients with distinct outcomes; for instance, patients with GFR >33 mL/min and without C1q-binding DSA showed a 5-year allograft survival of 93% vs 35% in patients with GFR ≤33 mL/min, without cg lesion and with C1q-binding DSA.
C1q-binding anti-HLA DSAs are strong and independent predictors of long-term kidney allograft loss after SOC treatment of ABMR and allowed to build a clinical decision tree showing high accuracy in risk stratifying allograft outcomes.