Journal Logo

Original Clinical Science—General

Does a Useful Test Exist to Properly Evaluate the Pathogenicity of Donor-specific Antibodies? Lessons From a Comprehensive Analysis in a Well-studied Single-center Kidney Transplant Cohort

Gautier Vargas, Gabriela MD1; Olagne, Jérome MD1,2; Parissiadis, Anne MD3; Joly, Mélanie MD1; Cognard, Noelle MD1; Perrin, Peggy MD1; Froelich, Nadine3; Guntz, Philippe MD3; Gachet, Christian MD, PhD3; Moulin, Bruno MD, PhD1,4; Caillard, Sophie MD, PhD1,4

Author Information
doi: 10.1097/TP.0000000000003080



Antibody-mediated rejection (AMR) was defined for the first time at the 1997 Banff conference.1 Its diagnosis currently relies on evidence of microcirculation injury in allograft biopsies, the presence of circulating antidonor antibodies, and signs of interactions between antibodies and the kidney endothelium.2 Donor-specific antibody (DSA) detected in the serum (sDSA) has been improved by the Luminex assay, which is significantly more sensitive than both complement-dependent cytotoxicity tests3 and enzyme-linked immunosorbent assays. Luminex technology has provided an improved understanding of the pathogenesis of alloimmunity in kidney allograft recipients, although the clinical relevance of all antibodies detectable by Luminex technology has not yet been fully elucidated. In this regard, the detection of sDSA by sensitive assays does not necessarily imply a positive crossmatch or a threat for kidney allograft function.4-6 Consequently, a precise characterization of harmful antibodies remains challenging. Complement assays detecting sDSA that can bind C1q or C3d—which reflect 2 different steps in complement activation—might enable the identification of sDSA that have the potential to activate the complement cascade.7 Several studies8-11 have shown that complement-binding sDSA are associated with AMR and poor graft survival. However, there is also evidence that sDSA levels—as expressed by the mean fluorescence intensity (MFI)—are well correlated with complement fixation and graft outcomes—being ultimately sufficient to predict the risk associated with DSA.12 Moreover, additional tools have recently been investigated to characterize the pathogenic potential of DSA, including the detection of intragraft DSA (gDSA). Bachelet et al13 have also shown that the presence of gDSA might reflect an aggressive pathogenic humoral process.

Using a single-center cohort, the purpose of this study was to provide a detailed characterization of DSA in kidney recipients, in which an allograft biopsy was performed, to (1) define the relevance of both C1q and C3d assays and intragraft fixation of DSA, (2) evaluate the pathogenic implications of DSA characteristics in biopsy-proven AMR, and (3) describe their effect on graft survival.


Study Population

From the entire cohort, we identified 127 patients with a functioning graft who had at least 1 DSA in 2011. Of them, 94 had a “for-cause” biopsy following the detection of DSA. Patients who had a desensitization therapy before biopsy (n = 5) and patients in whom DSA were no longer present on the day of biopsy (n = 3) were excluded. A study flowchart is provided in Figure S1, SDC, Therefore, the current single-center retrospective study was conducted in a sample of 86 kidney transplant recipients who had DSA and underwent a kidney biopsy.

The characteristics of the donors and recipients, as well as transplantation and graft survival data, were collected from our local database and completed with the French registry database (Cristal). All patients received an induction therapy with rabbit antithymocyte globulin or anti-interleukin2 receptor antibody (anti-IL2R). Maintenance therapy consisted of a calcineurin inhibitor with mycophenolate mofetil or azathioprine—the latter drug being used for patients transplanted before the introduction of mycophenolate mofetil (n = 8). All patients received corticosteroids for at least 3−6 months. In the event of AMR, patients were treated with steroid pulses, intravenous immunoglobulin, plasmapheresis, and 1 or 2 doses of rituximab.

The study protocol was approved by the local Institutional Review Board (DC-2013-1990). All participants provided written consent in accordance with the Declaration of Helsinki.

Allograft Pathology

Allograft biopsies (n = 86) were performed between 2003 and 2014 for cause in cases of increased serum creatinine (ie, >20% compared to previous measurements) or detection of proteinuria, and/or after sDSA occurrence (in the absence of graft dysfunction). We analyzed 1 biopsy per patient. For the purpose of the study, the 86 biopsies were retrospectively reviewed and regraded according to the 2013 Banff classification.2 C4d was detected by immunofluorescence (polyclonal anti-C4d antibody; Ventana 760-4436). The group of patients with AMR consisted of cases with acute or chronic AMR, including those associated with T cell-mediated rejection (TCMR)—defined as mixed rejections. The group of patients without AMR comprised biopsies with features of TCMR rejection only, normal histology, glomerulopathy recurrence, polyomavirus viral nephropathy, calcineurin inhibitor nephrotoxicity, and tubulo-interstitial chronic nephropathy.

HLA Methods

All patients were negative for T and B complement-dependent cytotoxicity crossmatch at the time of transplantation, and all of them were ABO compatible. Patients were screened annually following transplantation and after immunological risk events (eg, blood transfusion, pregnancy, minimization of immunosuppression) and/or according to the physician’s request. A serum sample for anti-HLA antibody detection was also collected systematically on the day a biopsy was performed. The detection of HLA antibodies was prospectively accomplished using the Luminex technology as of 2008, being retrospectively performed on stored sera for all patients in whom a biopsy was obtained before 2008 (n = 3).

HLA Typing

All patients and donors were molecularly typed for HLA-A*, B*, C*, DRB1*, and DQB1* antigens. Donors were further typed for loci corresponding to recipient antibodies, and all DSA were defined according to a known typing. Molecular typing was performed by reverse SSO (LABtype SSO; One Lambda, Canoga Park, CA) or SSP (Olerup SSP, West Chester, PA).

Detection and Characterization of Anti-HLA Antibodies and Complement-binding Assays

Patient sera were tested for the presence of IgG anti-HLA antibodies by performing LABScreen single antigen bead (SAB) (LS1A04 and LS2A01; One Lambda, Inc., Austin, TX, USA), C1qScreen (One Lambda), and C3d-binding (Immucor Lifecode Transplant diagnostics, Nijlen, Belgium) assays. All tests were conducted according to the manufacturer’s instructions (Supplemental Materials and Methods). Before IgG-SAB testing, all sera were treated with ethylenediaminetetraacetic acid after pretreatment according to a previously described methodology.14 Serum titration was performed in 21 sera when a prozone effect was suspected.12 For SAB analysis, all beads showing a normalized MFI greater than 1000 were considered positive. In complex cases, the cutoff was defined by an HLA MatchMaker analysis, with positivity being assigned according to the cutoff established by the MatchMaker program.15 The highest sDSA (hsDSA) was defined as the sDSA with the highest MFI value on the day of allograft biopsy.

The complement fixation capacity of the highest sDSA (hsDSA) was analyzed. C1q tests were conducted on heat-inactivated sera (30 minutes at 56°C, decomplementation). The cutoff was defined by an HLA MatchMaker analysis, and positivity was assigned according to a cutoff established with the MatchMaker program.15 The program identifies the MFI values for recipient self-HLA alleles and calculates a mean self-count and SD. The cutoff was defined as the value above the mean + 3 SD.

Detection of gDSA

Allograft elution was performed in 80 biopsies. In 6 cases, the specimen size was insufficient for the elution assay. The elution of anti-HLA antibodies was performed using unfixed frozen material from kidney biopsies in accordance with the published literature13,16 and using the kit Gamma ELU-KIT II (lots 426 and 450; Immucor Inc., Norcross, GA, USA; see Supplemental Materials and Methods). Identification of anti-HLA antibodies in the eluate was systematically performed for class I and/or II—depending on the sDSA type. gDSA positivity was assigned a cutoff identified by the MatchMaker program.15

Statistical Analysis

Descriptive statistics are presented as percentages, mean, and SD. Student’s t test, χ2, and Fisher exact test were used for quantitative and qualitative variables, as appropriate. Statistical significance was defined as a P < 0.05 (2-tailed). Receiver operating characteristic (ROC) curves were plotted for (1) hsDSA MFI based on SAB tests and DSA–C1q-binding capacity (One Lambda), (2) hsDSA MFI of IgG-Ab and C3d-binding capacity (Immucor), (3) hsDSA MFI of SAB tests based on the Luminex assay or the hsDSA MFI of IgG-Ab based on Immucor assays and the detection of gDSA, (4) the hsDSA MFI of SAB tests based on Luminex assays or the hsDSA MFI of IgG-Ab based on Immucor assays and the occurrence of AMR. Sensitivity and specificity of hsDSA MFI, C1q, and C3d fixation ability, as well as the presence of gDSA for AMR prediction, were calculated. Kaplan–Meier analyses were performed to evaluate graft survival after biopsy according to AMR, Banff scores, characteristics of the DSA, their C1q- and C3d-binding capacities, and the presence of gDSA. Factors were compared using the log-rank test. The date of event onset was assumed to be the date of graft loss, patient death, or the end of the study (February 2018). Multivariate analyses for graft survival were conducted using Cox regression. The following variables were entered in the model: MFI value, complement fixation capacity, and gDSA presence. All statistical analyses were performed using SPSS 20.0 Software (SPSS, Inc., Chicago, IL).


Cohort Characteristics

The demographic characteristics of the 86 patients are summarized in Table 1. Among patients who had sensitizing events before transplantation (ie, previous transplantation, and/or pregnancy, and/or blood transfusion, n = 71), 30 (35%) were sensitized against HLA in the pretransplant period. The mean follow-up time was 11.7 ± 5.5 years after transplantation (3–25 y).

TABLE 1. - Characteristics of the 86 kidney transplant recipients
Characteristic N (%) or mean ± SD
Age at transplantation (y) 41.7 ± 14.7 [4–74]
Sex (F/M) 42 (49) / 44 (51)
Primary kidney disease: GN 39 (45)
 TICN 18 (21)
 APKD 11 (13)
 Other 18 (21)
Sensitization before transplantation
 CPRA >80% 11/27 (41)
 CPRA 20–80% 7/27 (26)
 CPRA <20% 9/27 (33)
Pregnancies (among F) 28 (66)
Blood transfusion 31 (36)
Previous graft 12 (14)
Deceased donor / living donor 82 (95) / 4 (5)
SCD/ECD 70 (81) / 16 (19)
Blood transfusions after transplantation 35 (41)
Mismatch HLA class I
34 (40)
 3–4 52 (60)
Mismatch HLA class II
53 (62)
 2 33 (38)
ATG / anti-IL2-R induction 71 (83) / 15 (17)
Maintenance therapy
 CSA/Tac 55 (64) /21 (24)
 MMF/AZA 70 (81) /8 (9)
Data are presented as mean ± SD or N (%) as appropriate.
Anti-IL2-R, anti-interleukin 2 receptor antibodies; APKD, autosomal polycystic kidney disease; ATG, antithymocyte globulin; AZA, azathioprine; CPRA, calculated panel reactive antibodies; CSA, cyclosporine A; ECD, expanded criteria donors; GN, glomerulonephritis; MMF, mycophenolate mofetil; SCD, standard criteria donors; Tac, tacrolimus; TICN, tubulo-interstitial chronic nephropathy.

Serum DSA Characteristics Based on One Lambda SAB and C1q Assay

The mean delay between transplantation and sDSA appearance was 3.6 ± 4.9 years (0–22 y). sDSA were identified before transplantation (and persisting thereafter) in 27 patients (31%), whereas 59 cases (69%) had de novo sDSA. A comparison of these 2 patient groups is provided in Table S1, SDC, sDSA were mainly class II; 64 patients had only class II sDSA (74.4%), 14 patients had only class I sDSA (16.3%), and 8 patients had a combination of class I and class II sDSA (9.3%) (Table S2, SDC, The MFI of the highest sDSA value was >5000 in 55 patients (64%).

Forty-five of the 86 patients (52%) had C1q+ hsDSA, whereas 40 patients (46%) had C1q hsDSA (1 test was uninterpretable, Table S3, SDC, ROC curves indicated that an MFI cutoff of 5000 for the One Lambda SAB test could predict positivity in the C1q assay with a sensitivity of 98% and a specificity of 70% (area under curve [AUC] = 0.972; P < 0.001; Figure 1A). C1q+ DSA were mostly class II DSA and had a higher MFI based on the SAB test than C1q DSA (Table S4, SDC,

ROC curves of C1q and C3d positivity according to hsDSA MFI. A, ROC curve of C1q positivity according to hsDSA MFI. An MFI cutoff of 5000 had a sensitivity of 98% and a specificity of 70% to predict C1q assay positivity; AUC = 0.972 and P < 0.001. MFI SA in A is MFI based on single antigen testing. B, ROC curve of C3d positivity according to hsDSA MFI. An MFI cutoff of 5000 had a sensitivity of 95% and a specificity of 78% to predict C3d assay positivity; AUC = 0.956 and P < 0.001. MFI SA in B is MFI based on the Immucor test. AUC, area under curve; DSA, donor-specific antibody; hsDSA, highest MFI of DSA in the serum; MFI, mean fluorescence intensity; ROC, receiver operating characteristic.

Serum DSA Characteristics Based on Immucor Assays: IgG-Ab and C3d Assay

DSA were identified by Immucor assays in 83 of 86 patients (96.5%). The 3 DSA not recognized by Immucor assay were as follows: B55 (One Lambda MFI = 5471), B37 (One Lambda MFI = 2679), and DQA05 (One Lambda MFI = 15277). DSA detected by the Immucor assay were C3d positive in 42 cases (50.6%). The concordance between C1q and C3d positivity was 93% (Table S5, SDC, The ROC curve indicated that an MFI cutoff of 5000 for the Immucor IgG-Ab test could predict DSA positivity in the C3d assay with a sensitivity of 95% and a specificity of 78% (AUC = 0.956; P < 0.001; Figure 1B).

Allograft Pathology

The mean delay between sDSA appearance and graft biopsy was 4.6 ± 5.7 years. Among the 86 patients, biopsy revealed AMR lesions in 63 cases (73%), including acute lesions (acute AMR) in 27 and chronic lesions (chronic AMR) in 36 patients. Features of TCMR were associated with AMR lesions (mixed AMR) in 24 of these 63 patients. Three patients developed an AMR in the first posttransplant year, whereas the other cases were “late AMR”—regardless of the rejection being acute or chronic. Twenty patients (23%) exhibited no features of rejection, and 3 had isolated TCMR lesions (although they all showed DSA on the day of the biopsy). Patients with normal histology and TCMR were combined in a unique control group and compared to patients with (n = 63) and without (n = 23) AMR. The characteristics of both groups are given in Table 2 for the entire cohort and in Table S6, SDC, (for patients with stable function) and Table S7, SDC, (for patients with kidney dysfunction) according to graft function at the time of biopsy.

TABLE 2. - Characteristics of patients according to the presence of AMR, N = 86, mean ± SD or N (%)
Patients with AMR
n = 63
Patients without AMR
n = 23
Age at transplantation (y) 40.3 ± 15.5 45.3 ± 11.8 0.161
Sex (F/M) 27/36 15/8 0.065
Sensitized before transplantation 16 (25.4) 14 (60.9) 0.003
 Pregnancies 18/27 (66.6) 11/15 (73)
 Blood transfusions 23 (36.5) 8 (34.8) 0.883
 Previous transplantation 5 (7.9) 7 (30.4) 0.012
Blood transfusions after transplantation 27 (42.9) 8 (34.8) 0.497
ATG (vs anti-IL2R) 50 (79.4) 21 (91.3) 0.171
Tacrolimus (vs cyclosporine) 11 (18) 10 (43.5) 0.020
MPA (vs azathioprine) 50 (80.6) 18 (81.8) 0.904
De novo DSA (vs preformed) 48 (76.2) 11 (47.8) 0.014
Class II hsDSA (vs class I) 55 (87.3) 17 (73.9) 0.152
hsDSA MFI 11656 ± 7128 6184 ± 5823 <0.001
hsDSA >5000 (vs <5000) 46 (73) 9 (39) 0.004
hsDSA C1q+a 39 (62.9) 6 (26) 0.002
hsDSA C3d+ 37 (58.7) 5 (21.7) 0.003
gDSA+ 48 (76.2) 11(47.8) 0.012
Data are presented as mean ± SD or N (%) as appropriate.
aOne result of C1q fixation was uninterpretable.
AMR, antibody-mediated rejection; anti-IL2R, antiinterleukin 2 receptor antibodies; ATG, antithymocyte globulin; DSA, donor-specific antibody; gDSA: intragraft DSA; hsDSA, highest MFI of DSA in the serum; hsDSA C1q+, highest MFI of DSA in the serum able to fix C1q; hsDSA C3d+, highest MFI DSA in the serum able to fix C3d; MPA, mycophenolic acid; MFI, mean fluorescence intensity.

ROC curve analysis predicted AMR with an AUC of 0.722 (P = 0.002) for One Lambda SAB hsDSA MFI values, whereas the AUC for the Immucor IgG-Ab hsDSA MFI assay was 0.640 (P = 0.055; Figure 2).

ROC curve of AMR according to DSA MFI based on SA beads from One Lambda and Immucor. An MFI threshold of 5000 based on SA One Lambda predicted AMR with a sensitivity of 75% and a specificity of 60%; AUC = 0.722, P = 0.02. An MFI threshold of 5000 based on IgG-Ab Immucor predicted AMR with a sensitivity of 63% and a specificity of 57%; AUC = 0.640, P = 0.055. AMR, antibody-mediated rejection; AUC, area under curve; DSA, donor-specific antibody; MFI, mean fluorescence intensity; ROC, receiver operating characteristic; SA, single antigen.

Histological lesions were significantly more severe in patients harboring DSA with a higher MFI (>5000 versus <5000), C1q+ hsDSA (versus C1q), or C3d+ hsDSA (versus C3d) based on glomerulitis and peritubular capillaritis scoring (Table S8, SDC, When these analyses were restricted to patients diagnosed with AMR (n = 63), there was no difference in Banff scoring according to hsDSA MFI, C1q, or C3d positivity (Table S9, SDC,


gDSA were detected in 59 biopsies (73.7%) as follows: anti-HLA class I was identified in 7 cases, and anti-HLA class II was noted in 52 cases.

Table 3 shows the concordance between the characteristics of the sDSA and the presence of gDSA. ROC curve analysis indicated that One Lambda SAB hsDSA MFI values predicted the detection of DSA in the graft with an AUC of 0.856 (P < 0.01), whereas the AUC of Immucor IgG-Ab hsDSA MFI values was 0.833 (P < 0.01; Figure 3). gDSA were detected in 48 (81.3%) biopsies with AMR. Eleven patients with gDSA did not show features of AMR in corresponding biopsies. Of note, 8 of these patients underwent a subsequent biopsy in the subsequent 4–104 months (mean 54.6 months). In 5 of these individuals (62.5%), features of AMR appeared on the subsequent biopsy (2 acute AMR and 3 chronic AMR).

TABLE 3. - Concordance between the presence of gDSA and the characteristics of DSA found in the serum on the day of biopsy, N = 80 biopsies
Preformed DSA
n = 25
De novo DSA
n = 55
MFI >5000
n = 55
n = 43a
n = 40
n = 28
gDSA+ n = 59 15 44 48 42 38 25
gDSA n = 21 10 11 7 1 2 3
aOne result was uninterpretable for C1q fixation.
DSA, donor-specific antibody; gDSA, intragraft DSA; MFI, mean fluorescence intensity.

ROC curve of the presence of gDSA according to DSA MFI based on SA beads by One Lambda and Immucor. An MFI threshold of 5000 for SA One Lambda predicted intragraft DSA with a sensitivity of 79% and a specificity of 71%; AUC = 0.856, P < 0.01. An MFI threshold of 5000 for IgG-Ab Immucor predicted intragraft DSA with a sensitivity of 74% and a specificity of 86%; AUC = 0.833, P < 0.01. AUC, area under the curve; DSA, donor-specific antibody; gDSA, intragraft DSA; MFI, mean fluorescence intensity; ROC, receiver operating characteristic; SA, single antigen.

Factors Associated With Allograft Survival

The following factors were associated with a lower graft survival rate in the entire cohort: male sex (P = 0.020), the presence of AMR (P < 0.001), particularly in cases of mixed rejection (P < 0.001), and higher values (2–3 versus 0–1) of interstitial inflammation (P = 0.001), glomerulitis (P < 0.001), peritubular capillaritis (P < 0.001), chronic interstitial inflammation (P = 0.001), chronic glomerulitis (P = 0.001), chronic vasculitis (P = 0.002), mesangial matrix increase (P < 0.001), arteriolar hyalinosis (P = 0.009), interstitial fibrosis/tubular atrophy (P = 0.002), and C4d positivity (P = 0.005). One Lambda SAB hsDSA MFI values >5000 were also associated with decreased graft survival (P = 0.010). We did not observe an effect on graft survival based on de novo versus preformed sDSA (P = 0.256), sDSA class (P = 0.075), C1q+ sDSA (P = 0.064), C3d+ sDSA (P = 0.304), and the presence of gDSA (P = 0.21)—some of these parameters approached statistical significance (Figure 4). In multivariate analysis, MFI values >5000 remained independently associated with graft survival (Table 4).

TABLE 4. - Allograft risk loss according to hDSA characteristics
P OR 95% CI
C1q test 0.861 1.088 0.42-2.78
MFI >5000 0.035 3.402 1.09-10.59
gDSA 0.885 0.921 0.30-2.78
Italics indicate a statistically significant result.CI, confidence interval; gDSA, intragraft DSA; MFI, mean fluorescence intensity; OR, odds ratio.

Kaplan–Meier analysis of graft survival in 86 patients according to hsDSA MFI, C1q and C3d positivity, and gDSA. A, hsDSA MFI >5000 (green curve, n = 55) vs hsDSA MFI <5000 (blue curve, n = 31), P = 0.010. B, C1q+ DSA (green curve, n = 45) vs C1q DSA (blue curve, n = 40), P = 0.064. C, C3d+ DSA (green curve, n = 42) vs C3d DSA (blue curve, n = 43), P = 0.304. D, Presence of gDSA (green curve, n = 59) vs absence of gDSA (blue curve, n = 21), P = 0.210. C1q+ DSA, DSA able to fix C1q; C1q DSA, DSA unable to fix C1q; C3d+ DSA, DSA able to fix C3d; C3d DSA, DSA unable to fix C3d; DSA, donor-specific antibody; gDSA, intragraft DSA; hsDSA MFI, highest MFI of DSA; MFI, mean fluorescence intensity.

When these analyses were restricted to patients with AMR (n = 63), Banff scoring of i (interstitial infiltration; P < 0.001), g (glomerulitis; P = 0.011), cg (chronic glomerulitis; P = 0.016), ci (chronic interstitial infiltration; P = 0.007), cv (chronic vascularitis; P = 0.007), mm (mesangial matrix increase; P < 0.001), and interstitial fibrosis/tubular atrophy (P = 0.01) remained associated with graft loss, whereas sex, C4d staining, sDSA MFI, C1q or C3d positivity, and the presence of gDSA were not (Figure 5).

Kaplan–Meier analysis of graft survival in 63 patients with AMR according to hsDSA MFI, C1q and C3d positivity, and gDSA. A. hsDSA MFI >5000 (green curve, n = 46) vs hsDSA MFI <5000 (blue curve, n = 17), P = 0.087. B, C1q+ DSA (green curve, n = 39) vs C1q DSA (blue curve, n = 23), P = 0.454. C, C3d+ DSA (green curve, n = 37) vs C3d DSA (blue curve, n = 26), P = 0.794. D, Presence of gDSA (green curve, n = 48) vs absence of gDSA (blue curve, n = 11), P = 0.618. AMR, antibody-mediated rejection; C1q+ DSA, DSA able to fix C1q; C1q DSA, DSA unable to fix C1q; C3d+ DSA, DSA able to fix C3d; C3d DSA, DSA unable to fix C3d; DSA, donor-specific antibody; gDSA, intragraft DSA; hsDSA MFI, highest MFI of DSA; MFI, mean fluorescence intensity.

When graft survival analyses were conducted in the subgroups of patients with preformed or de novo DSA, the results were largely similar to those observed in the entire cohort, particularly with regard to the DSA characteristics. The only differences were related to a more relevant important impact of interstitial inflammation and Banff chronic scores on graft survival in the subgroup of patients with de novo DSA.

Patients who had graft biopsy taken “for cause” and those with stable function had a very similar long-term graft survival—although graft loss occurred more rapidly in the former group (Figure S2, SDC, Finally, in the subgroup of patients with stable renal function at the time of biopsy, DSA MFI >5000 (P = 0.039) and gDSA positivity (P = 0.027) were strongly associated with a decreased graft survival (Figure 6), whereas C3d (P = 0.427), C1q (P = 0.153), and DSA class (P = 0.482) were not. In the subgroup of patient with graft dysfunction at the time of biopsy, C1q and C3d positivity as well as gDSA presence did not show an association with graft survival (Figure 7).

Kaplan–Meier analysis of graft survival in 28 patients with stable allograft function at the time of biopsy according to gDSA positivity (A) and DSA MFI (B): gDSA positive (green curve, n = 17) vs negative (blue curve, n = 9); P = 0.027; hsDSA MFI >5000 (green curve, n = 16) vs hsDSA MFI <5000 (blue curve, n = 12), P = 0.039. DSA, donor-specific antibody; gDSA, intragraft DSA; hsDSA MFI, highest MFI of DSA; MFI, mean fluorescence intensity.
Kaplan–Meier analysis of graft survival in 58 patients with allograft dysfunction at the time of biopsy according to gDSA (A) and C3d positivity (B): gDSA positive, (green curve, n = 42) vs negative (blue curve, n = 12); P = 0.909; C3d+ DSA (green curve, n = 31) vs C3d DSA (blue curve, n = 26), P = 0.668. DSA, donor-specific antibody; gDSA, intragraft DSA; C3d+ DSA, DSA able to fix C3d; C3d DSA, DSA unable to fix C3d.


In our cohort, approximately half of the donor-specific anti-HLA antibodies (DSA) were able to fix complement fractions C1q and/or C3d. Furthermore, we showed that this capacity was strongly correlated with DSA fluorescence intensity, which directly reflects antibody “levels” in the sera. Similar observations regarding C1q positivity were previously reported for preformed DSA17-19 and de novo DSA.12,20-22 However, the lack of a correlation between SAB MFI values and C1q positivity in other studies7,10 could be attributed to a prozone effect,14 particularly when sera were neither pretreated nor diluted before the SAB assay.

We also demonstrate that the Immucor assay is slightly less sensitive than the One Lambda assay, because 3 samples with DSA were missed by the former assay in our patient cohort. Few data are currently available on the correlation between Immucor IgG-Ab MFI values and C3d positivity for DSA. Claisse et al23 demonstrated a correlation between MFI values and C3d positivity in 7 patients before and after plasmapheresis. We also detected a convincing association between IgG-Ab MFI values based on the Immucor assay and C3d positivity. Moreover, in our cases, C1q and C3d tests yielded very similar results (93% concordance rate), supporting the relative equivalence between the 2 assays.

In our series, only 75% of patients with DSA on the day of the biopsy showed histological features of AMR. Of note, this ratio was not significantly different between patients with stable renal function and those with graft dysfunction (64% versus 77%, respectively), suggesting that kidney recipients with de novo DSA appearance should benefit from an early allograft biopsy. We thus aimed to investigate whether some characteristics of their serum DSA could predict the presence of AMR lesions. In our study, DSA that appeared de novo and with a higher MFI were more frequently associated with AMR occurrence. DSA that were able to fix C1q or C3d were more frequently associated with AMR and C4d-positive staining. This association is probably related to MFI levels, as shown by Yell et al,21 who observed a significant relationship between DSA with an MFI >7000 and biopsy-proven AMR in a cohort of 34 patients. An association between C1q+ DSA and AMR was also described by Loupy et al10 and Yamamoto et al24 upon studying routine biopsies, and more recently in other studies evaluating for-cause biopsies.8,25-27 Nevertheless, we observed that a substantial number of patients developed biopsy-proven AMR even in the presence of low MFI values for DSA (28%) or no DSA complement fixation (37% for C1q and 41% for C3d). This was also demonstrated in a more recent report, in which DSA complement fixation was not found to improve AMR prediction.27,28 It is possible that—in these cases—DSA pathogenicity is not determined by its capacity to activate complement but is rather related to complement-independent mechanisms such as antibody-dependent cellular cytotoxicity mediated by natural killer cells.29,30 It can also be related to the presence of unidentified non-HLA antibodies, which were found to activate endothelial cells and induce microcirculation inflammation.31-35 In our study, we ruled out the presence of donor-specific major histocompatibility-complex class I-related chain A antibodies in patients with AMR features and low DSA MFI values. Another simple explanation for the lack of correlation may rely in the fact that complement-binding tests are not functional assays but only binding assays. We are unable to ensure that in vitro complement activity is equivalent to the in vivo setting. For example, multiple low-level MFI DSA—which cannot be detected by complement-binding techniques—could be sufficient in vivo to bind complement on cell surfaces.

In contrast, approximately 15% of patients harboring C1q or C3d fixing DSA did not develop rejection. In this regard, we hypothesize that antibodies could be capable of fixing C1q but may be unable to activate the complement cascade. This possibility is in line with the observations by Duquesnoy et al,36 who reported that the activation of the complement cascade depends of the type and the chemical/physical properties of the epitope. Moreover, among the subgroup of patients who developed AMR, the presence of DSA with high-level MFI values or complement-fixing capacity was not associated with more severe histological microcirculatory inflammatory lesions compared with that in patients with low or non–complement-fixing DSA. These observations do not support a systematic association between the presence of a complement-fixing DSA and the severity of humoral features.

In our series, we found gDSA in 74% of biopsies. This corroborates previous findings from 48 pediatric kidney recipients.37 In our cohort, the identification of DSA in the kidney correlated well with MFI levels, as it was previously described in a French study.13 However, the presence of gDSA was not associated with AMR severity, as reported in an Italian pediatric series.37 These results also corroborate the findings of a recent French report focusing on adult patients.28 Nevertheless, the detection of gDSA in the absence of AMR lesions, which occurred in 11 patients of our cohort, could be used as an early predictor of AMR. Interestingly, 8 of these patients had subsequent biopsies showing the appearance of acute or chronic AMR lesions in 5 of them. Finally, gDSA deposition was a good predictor of graft failure in patients with subclinical AMR in our cohort. We therefore suggest that gDSA could precede the emergence of aggressive humoral lesions and may serve as a useful predictor of AMR and graft survival.

In keeping with the published literature, we observed a strong effect of AMR on graft survival. We also found an association between DSA MFI levels and graft survival. However, neither DSA class nor the ability to activate the complement fractions was associated with poorer graft survival in our entire cohort. Our results are in contrast to those of Loupy et al or Sicard et al10,11 but corroborate some more recent literature20,28,38 in which DSA were found to alter graft survival regardless of whether they could fix complement.

The main limitation of our study lies in its retrospective, single-center nature. Owing to the caveats inherent in the study design, most of the study patients displayed late AMR. Consequently, our results might not be generalizable to patients with early AMR and acute renal failure.

Conversely, an important strength of our study is the extensive analysis of DSA characteristics—which were precisely studied with the MatchMaker Program and evaluated by several sensitive assays (not only in sera but also in the allograft itself). Finally, they were correlated with the histological features of the biopsy performed on the same day—which was a homogeneous factor in the selected group.

In conclusion, we show that—even with a very thorough analysis of DSA—we were not entirely able to accurately evaluate the risk of developing AMR in patients with DSA. Moreover, graft survival at midterm does not seem to be affected by the characteristics of serum DSA assessed in our study—with the exception of simple quantification of MFI values. Altogether, these data indicate that the field of AMR is still relatively naive and that further work is required to both completely understand the pathogenicity of DSA and improve their characterization. Finally, the detection of gDSA could have a predictive value for AMR and/or graft survival in early stages.


We gratefully acknowledge the Centre de Ressources Biologiques for logistic help with frozen biopsy material.


1. Racusen LC, Solez K, Colvin RB, et al. The Banff 97 working classification of renal allograft pathology. Kidney Int. 1999; 552713–723
2. Haas M, Sis B, Racusen LC, et al.; Banff Meeting Report Writing Committee. Banff 2013 meeting report: inclusion of c4d-negative antibody-mediated rejection and antibody-associated arterial lesions. Am J Transplant. 2014; 142272–283
3. Tait BD, Süsal C, Gebel HM, et al. Consensus guidelines on the testing and clinical management issues associated with HLA and non-HLA antibodies in transplantation. Transplantation. 2013; 95119–47
4. Aubert V, Venetz JP, Pantaleo G, et al. Low levels of human leukocyte antigen donor-specific antibodies detected by solid phase assay before transplantation are frequently clinically irrelevant. Hum Immunol. 2009; 708580–583
5. Bartel G, Regele H, Wahrmann M, et al. Posttransplant HLA alloreactivity in stable kidney transplant recipients-incidences and impact on long-term allograft outcomes. Am J Transplant. 2008; 8122652–2660
6. Terasaki PI. A personal perspective: 100-year history of the humoral theory of transplantation. Transplantation. 2012; 938751–756
7. Chen G, Sequeira F, Tyan DB. Novel C1Q assay reveals a clinically relevant subset of human leukocyte antigen antibodies independent of immunoglobulin G strength on single antigen beads. Hum Immunol. 2011; 7210849–858
8. Sutherland SM, Chen G, Sequeira FA, et al. Complement-fixing donor-specific antibodies identified by a novel C1Q assay are associated with allograft loss. Pediatr Transplant. 2012; 16112–17
9. Chin C, Chen G, Sequeria F, et al. Clinical usefulness of a novel C1Q assay to detect immunoglobulin G antibodies capable of fixing complement in sensitized pediatric heart transplant patients. J Heart Lung Transplant. 2011; 302158–163
10. Loupy A, Lefaucheur C, Vernerey D, et al. Complement-binding anti-HLA antibodies and kidney-allograft survival. N Engl J Med. 2013; 369131215–1226
11. Sicard A, Ducreux S, Rabeyrin M, et al. Detection of c3d-binding donor-specific anti-HLA antibodies at diagnosis of humoral rejection predicts renal graft loss. J Am Soc Nephrol. 2015; 262457–467
12. Tambur AR, Herrera ND, Haarberg KM, et al. Assessing antibody strength: comparison of MFI, c1q, and titer information. Am J Transplant. 2015; 1592421–2430
13. Bachelet T, Couzi L, Lepreux S, et al. Kidney intragraft donor-specific antibodies as determinant of antibody-mediated lesions and poor graft outcome. Am J Transplant. 2013; 13112855–2864
14. Schnaidt M, Weinstock C, Jurisic M, et al. HLA antibody specification using single-antigen beads–a technical solution for the prozone effect. Transplantation. 2011; 925510–515
15. Duquesnoy RJ. Antibody-reactive epitope determination with HLAMatchmaker and its clinical applications. Tissue Antigens. 2011; 776525–534
16. Martin L, Guignier F, Bocrie O, et al. Detection of anti-HLA antibodies with flow cytometry in needle core biopsies of renal transplants recipients with chronic allograft nephropathy. Transplantation. 2005; 79101459–1461
17. Otten HG, Verhaar MC, Borst HP, et al. Pretransplant donor-specific HLA class-I and -II antibodies are associated with an increased risk for kidney graft failure. Am J Transplant. 2012; 1261618–1623
18. Crespo M, Torio A, Mas V, et al. Clinical relevance of pretransplant anti-HLA donor-specific antibodies: does c1q-fixation matter? Transpl Immunol. 2013; 291–428–33
19. Schaub S, Hönger G, Koller MT, et al. Determinants of C1Q binding in the single antigen bead assay. Transplantation. 2014; 984387–393
20. Guidicelli G, Guerville F, Lepreux S, et al. Non-complement-binding de novo donor-specific anti-HLA antibodies and kidney allograft survival. J Am Soc Nephrol. 2016; 272615–625
21. Yell M, Muth BL, Kaufman DB, et al. C1Q binding activity of de novo donor-specific HLA antibodies in renal transplant recipients with and without antibody-mediated rejection. Transplantation. 2015; 9961151–1155
22. Moreno Gonzales MA, Mitema DG, Smith BH, et al. Comparison between total IgG, C1q, and C3d single antigen bead assays in detecting class I complement-binding anti-HLA antibodies. Transplant Proc. 2017; 4992031–2035
23. Claisse G, Absi L, Cognasse F, et al. Relationship between mean fluorescence intensity and c1q/c3d-fixing capacities of anti-HLA antibodies. Hum Immunol. 2017; 784336–341
24. Yamamoto T, Watarai Y, Takeda A, et al. De novo anti-HLA DSA characteristics and subclinical antibody-mediated kidney allograft injury. Transplantation. 2016; 100102194–2202
25. Malheiro J, Tafulo S, Dias L, et al. Determining donor-specific antibody c1q-binding ability improves the prediction of antibody-mediated rejection in human leucocyte antigen-incompatible kidney transplantation. Transpl Int. 2017; 304347–359
26. Bamoulid J, Roodenburg A, Staeck O, et al. Clinical outcome of patients with de novo c1q-binding donor-specific HLA antibodies after renal transplantation. Transplantation. 2017; 10192165–2174
27. Eskandary F, Bond G, Kozakowski N, et al. Diagnostic contribution of donor-specific antibody characteristics to uncover late silent antibody-mediated rejection-results of a cross-sectional screening study. Transplantation. 2017; 1013631–641
28. Courant M, Visentin J, Linares G, et al. The disappointing contribution of anti-human leukocyte antigen donor-specific antibodies characteristics for predicting allograft loss. Nephrol Dial Transplant. 2018; 33101853–1863
29. Zhang Q, Reed EF. Non-MHC antigenic targets of the humoral immune response in transplantation. Curr Opin Immunol. 2010; 225682–688
30. Hidalgo LG, Sis B, Sellares J, et al. NK cell transcripts and NK cells in kidney biopsies from patients with donor-specific antibodies: evidence for NK cell involvement in antibody-mediated rejection. Am J Transplant. 2010; 1081812–1822
31. Morales-Buenrostro LE, Alberú J. Anti-major histocompatibility complex class I-related chain A antibodies in organ transplantation. Transplant Rev (Orlando). 2008; 22127–38
32. Regele H. Non-HLA antibodies in kidney allograft rejection: convincing concept in need of further evidence. Kidney Int. 2011; 796583–586
33. Dragun D, Müller DN, Bräsen JH, et al. Angiotensin II type 1-receptor activating antibodies in renal-allograft rejection. N Engl J Med. 2005; 3526558–569
34. Taniguchi M, Rebellato LM, Cai J, et al. Higher risk of kidney graft failure in the presence of anti-angiotensin II type-1 receptor antibodies. Am J Transplant. 2013; 13102577–2589
35. Zhang X, Reed EF. Effect of antibodies on endothelium. Am J Transplant. 2009; 9112459–2465
36. Duquesnoy RJ, Marrari M, Jelenik L, et al. Structural aspects of HLA class I epitopes reacting with human monoclonal antibodies in Ig-binding, C1q-binding and lymphocytotoxicity assays. Hum Immunol. 2013; 74101271–1279
37. Nocera A, Tagliamacco A, Cioni M, et al. Kidney intragraft homing of de novo donor-specific HLA antibodies is an essential step of antibody-mediated damage but not per se predictive of graft loss. Am J Transplant. 2017; 173692–702
38. Kauke T, Oberhauser C, Lin V, et al. De novo donor-specific anti-HLA antibodies after kidney transplantation are associated with impaired graft outcome independently of their C1q-binding ability. Transpl Int. 2017; 304360–370

Supplemental Digital Content

Copyright © 2019 Wolters Kluwer Health, Inc. All rights reserved.