We studied 982 adult patients who underwent kidney transplant alone (ABO compatible and DSA absent, flow crossmatch (FC) negative at the time of transplant based on last serum within <30 days of transplant) from January 1, 2014 to December 31, 2018 at the Thomas E. Starzl Transplantation Institute—University of Pittsburgh. Repeat kidney and kidney after other solid organ transplant patients were included. We excluded those with early graft loss or death (<90 days, n = 23) or an unstable 1st year course (defined as those requiring for-cause biopsy in 1st year, n = 223) to limit for-cause DSA testing that often accompanies for-cause biopsies and graft dysfunction. The remaining 736 patients served as our primary study cohort (Figure 1).

DSA was tested within the 1st year (1, 3, 6, 9, 12 months) per our center’s screening protocol, at time of any biopsy, and annually until 5 years. DSA was considered newly detected as last serum sample available at time of transplant was DSA negative (prior serum was not analyzed). DSA was measured using One Lambda LABScreen™ single antigen bead assay and considered positive if adjusted mean fluorescent intensity (MFI) was ≥1,000 units based on our HLA lab’s designation. A single positive DSA reading (for either class) was considered as a single positive and multiple positive DSA tests for the same class separated in time were considered as multiple positive.

Induction was mainly with thymoglobulin and rarely with basiliximab (if 0% calculated panel reactive antibody [cPRA], 0 antigen mismatch, and a living donor [LD] transplant recipient). For maintenance, majority were on mycophenolate mofetil and calcineurin inhibitor (mainly Tacrolimus) with a minority also on prednisone (those with cPRA ≥ 90% or those on prednisone prior, 5 mg daily or their dose prior to transplant). Prednisone (5 mg daily) was subsequently added to maintenance regimen for any rejection episodes (clinical or subclinical). There was no systematic center protocol for adjusting maintenance immunosuppression based on DSA detection alone.

Protocol biopsies were recommended to all patients at ∼3 and 12 months post-transplant unless contraindicated. Potential contraindications included those patients on systemic anticoagulation, those on dual anti-platelet therapy, those with intrabdominal kidney location, those who received en bloc kidneys, those with active malignancy or serious infection at time of scheduled protocol biopsy, or those lacking transportation. Additionally, as with any medical procedure, patients had the option to decline recommendation to undergo a protocol biopsy after risks and benefits were thoroughly discussed. Biopsies were scored using Banff 2013 and later 2017 classification (24, 25). For-cause biopsies were done for renal dysfunction (rise in serum creatinine >25% from baseline and/or new or worsening proteinuria [>1 g/day and/or >1 g/g urine protein to creatinine ratio]), but not for isolated DSA detection alone.

Protocol biopsy findings were defined as no inflammation (NI, Banff t score 0 + i/ti score 0), subclinical inflammation (SCI, minimal inflammation [MI] Banff t score <0 + i/ti score ≥0 or Banff Borderline Changes [BBC] Banff t score >0 + i/ti score ≥0 and <1A TCMR), and subclinical TCMR (SC-TCMR, ≥1A TCMR). Those with subclinical ABMR (SC-ABMR) were included within these three groups using associated findings (NI, SCI, or SC-TCMR) and were also analyzed separately. Protocol biopsies were also grouped based on timing and maximum grade (highest grade noted on any 1st year protocol biopsy). For-cause biopsies beyond the 1st year were defined as negative (no pathologic findings), inflammation (MI or BBC), rejection (≥1A TCMR and/or ABMR), and non-alloimmune events (urinary tract infection, BK virus nephropathy, acute tubular injury, glomerulonephritis, secondary oxalate nephropathy).

The median follow up was ∼3.3 years (Table 1).

Primary outcomes were kidney function (estimated glomerular filtration rate [eGFR] using CKD-EPI formula) and survival (patient, combined patient and graft, death censored graft survival [DCGS], and graft failure [defined as graft loss with return to dialysis or re-transplant, death, or eGFR < 20 ml/min at last follow up] free survival). Secondary outcomes were subclinical events (SCI, SC-TCMR, SC-ABMR, mean cumulative acute scores [defined as sum of Banff i/ti, t, g, ptc, and v scores], and mean IFTA score) within 1st year and clinical events (rejection, inflammation, non-alloimmune events) beyond 1st year.

Patient information was obtained through specified personnel at Thomas E. Starzl Transplantation Institute as regulated by the institutional review board (IRB) at the University of Pittsburgh. The institution maintains a prospectively collected electronic database of all kidney transplant patients. The studies involving human participants were reviewed and approved by the University of Pittsburgh IRB and the patients were not required to provide written consent for this study per the University of Pittsburgh IRB. We collected data under IRB number PRO-13060220. The activities reported are consistent with the Principles of the Declaration of Istanbul as outlined in the “Declaration of Istanbul on organ trafficking and Transplant Tourism” and Declaration of Helsinki.

Analysis was completed using SAS 9.4 (SAS Institute Inc., Cary NC). Continuous variables are presented as mean ± standard deviation for normally distributed data and median with interquartile range for nonparametric data. Differences in baseline and transplant variables were assessed using analysis of variance and chi-square tests. We evaluated differences in recipient and donor demographics (age, race, gender), and other variables [body mass index (BMI) at transplant, preemptive transplant, End Stage Kidney Disease (ESKD) cause, cold ischemia time (CIT), Kidney Donor Prognostic Index (KDPI), donor type [deceased donor (DD) vs. LD], PRA (panel reactive antibody) Class I/II, cPRA, HLA mismatch (A, B, DR), Cytomegalovirus (CMV)/Epstein Barr Virus (EBV) serostatus, delayed graft function (DGF), biopsy accrual rates] between groups (Table 1). Linear mixed model was used to assess eGFR with a serum creatinine value of 8 mg/dl assigned for graft loss. Covariates for multivariate analysis were identified and evaluated for inclusion before modeling. All multivariate analysis included recipient age, donor type (LD vs. DD), PRA I/II ≥ 90%, cPRA ≥ 90%, DGF, and SCI/SC-TCMR using a backward selection Cox Regression Model with variables with p < 0.2 included in the model. Survival (patient, graft, graft failure free) was examined by Kaplan Meier method with survival curves compared by Log rank test. Adjusted Bonferroni p-values were used with multiple log-rank comparisons. We examined relationship between DSA and 1st year protocol biopsy findings on eGFR in an exploratory analysis. A p-value <0.05 was considered statistically significant.

The DSA+ cohort included 131 patients (18%) with at least one positive screening DSA test (class I and/or II) during the 1st year and the remaining 605 patients (82%) were the DSA- cohort. DSA+ patients had less Caucasians (68% vs. 78%, p = 0.04), fewer were pre-emptive (12% vs. 20%, p = 0.04), more were deceased donor (76% vs. 67%, p = 0.03), more were anti-HLA sensitized (% cPRA ≥ 90%, 24% vs. 13%, p = 0.002), more were DR mismatched (1 [1–2] vs. 1 [1–2], p = 0.02), and more had DSA detected (persisting from 1st year or new) beyond 1 year (39% vs. 10%, p < 0.001) (Table 1). Other donor and recipient variables as well as protocol biopsy accrual rates were all similar. DSA detection was comparable for those with and without protocol biopsy (19% with vs. 16% without, p = 0.36), but those without protocol biopsy were more likely to have diabetic ESKD, prior transplant, increased anti-HLA sensitization, received DD transplant, longer CIT, and less likely to have had a preemptive transplant (Supplementary Table S1).

DSA were primarily detected ∼1–4 months post-transplant (class I at 40 days [30–108] and class II at 38 days [31–135]) with 77% of first DSA detected within 100 days (Table 2). Additionally, 60% of DSA+ patients had at least one multiple positive DSA for the same class and 18% of DSA+ patients had both class I and II DSA detected within 1st year.

Protocol biopsy results for those with at least one protocol biopsy (n = 515, 70%) are shown in Table 3. DSA+ patients had similar protocol biopsy rates vs. DSA- patients (73% [96 patients, 159 biopsies] vs. 69% [419 patients, 696 biopsies], p = 0.36). Mean cumulative acute and IFTA scores were similar at 3 and 12 months. Frequency of protocol biopsies with NI, SCI, and SC-TCMR were comparable among groups based on 3-months, 12-months, and maximum 1st year grade (Table 3). There was an increased incidence of SC-ABMR during the 1st year in DSA+ vs. DSA- patients (4% vs. 0%, p < 0.001), though overall occurrence was rare (0.8%) (Table 3). There were seven cases of SC-ABMR in six recipients (1 SC-ABMR alone, 3 with concurrent SCI, 3 with concurrent SC-TCMR).

DSA+ patients had similar proportion of for-cause biopsies beyond 1st year vs. DSA- patients (23% [30 patients, 36 biopsies] vs. 25% [151 patients, 189 biopsies], p = 0.62, Table 4). The distribution of biopsy findings was similar between DSA+ and DSA- cohorts (p = 0.35, Figures 2A,B), including rates of overall clinical rejection (56% vs. 42%, p = 0.14). Clinical TCMR was lower among DSA+ patients (28% vs. 36%, p < 0.001), though severity of TCMR was similar (Table 4). The distribution of type rejection was different (p < 0.001) and favored more mixed rejection (ABMR + TCMR) in DSA+ patients (Figures 2C,D). Lastly, distribution of non-alloimmune events on for-cause biopsies was similar (16% vs. 14%, p = 0.71, Table 4).

Using a linear mixed model, DSA+ and DSA- groups had similar eGFR over study period (54.8 vs. 53.8 ml/min/1.73 m², p = 0.56, Figure 3). Subgroup exploratory analysis in those patients with protocol biopsy revealed eGFR was similar among DSA+ vs. DSA- patients when stratified by 1st year subclinical events (NI, SCI, SC-TCMR), albeit there was slightly increased eGFR for DSA+ with SCI vs. DSA-with SCI patients (p = 0.02, 61 ml/min vs. 54 ml/min, Supplementary Figure S1).

Overall, DSA+ patients had similar patient survival (83% vs. 90%, p = 0.15, Figure 4A), combined patient and graft survival (78% vs. 85%, p = 0.09, Figure 4B), DCGS (91% vs. 94%, p = 0.30, Figure 4C), and graft failure free survival (76% vs. 82%, p = 0.11, Figure 4D) vs. DSA- patients. Among DSA+ patients, survival was similar whether based on timing of detection (Figure 5) or DSA class (Figure 6). We also assessed survival stratified by protocol biopsy status. First, those with protocol biopsy had better patient survival (p = 0.004), combined patient and graft survival (p = 0.02), and graft failure free survival compared to those without protocol biopsy (p = 0.045), but not DCGS (p = 0.68) (Supplementary Figure S2). Among those without protocol biopsy, DSA+ patients had similar survival vs. DSA- patients (Supplementary Figure S3). Conversely, among those with at least one protocol biopsy, DSA+ patients had decreased patient (p = 0.04), patient and graft (p = 0.02), graft failure free survival (p = 0.05), but not DCGS (p = 0.13) compared to DSA-patients (Supplementary Figure S4).

Given a trend towards worse graft failure free survival in DSA+ patients, specifically those who had a protocol biopsy, we performed a backwards cox regression multivariate analysis that found recipient age (1.03, 95% CI 1.01–1.05, p = 0.01) and DSA detection within 1st year (1.91, 95% CI 1.03–3.55, p = 0.04) as independent predictors for graft failure among those who had a protocol biopsy (Table 5).

Though excluded from our primary cohort analysis, we did note increased DSA within 1st year in unstable vs. DSA+ stable patients (30% vs. 18%, p < 0.001). We explored demographic differences among stable and unstable cohorts based on DSA status (Supplementary Tables S2, S3) and DSA characteristics among DSA+ patients (unstable vs. stable) (Supplementary Table S4). Interestingly, DSA- unstable patients received kidney transplants with higher KDPI and had more DGF than DSA- stable patients (Supplementary Table S2). Further, DSA+ unstable patients were younger (46 vs. 51, p = 0.02) with increased overall number of DSA tests, overall number of positive class II tests, and a trend towards more combined Class I and Class II DSA detection vs. DSA+ stable patients, though the timing of 1st positive test (Class I/II) was similar for DSA+ unstable vs. DSA+ stable patients (Supplementary Tables S3, S4). Patient survival, patient and graft survival, DCGS, and graft failure free survival were similar among DSA+ vs. DSA- unstable patients (Supplementary Figure S5). However, when all four groups were included, there was significant differences in survival among the four groups as the unstable cohort had inferior survival overall, particularly the DSA+ unstable group (Figure 7). Interestingly, when including entire population (both stable and unstable), DSA+ stable patients did have inferior patient/graft survival (p = 0.001), DCGS (p = 0.03), and graft failure free survival (p = 0.001) vs. DSA-stable patients (Figure 7).