Antibody-mediated rejection as a contributor to previously unexplained early liver allograft loss

Authors


Abstract

We analyzed 60 patients with idiopathic early allograft loss (defined as death or retransplantation at <90 days) to determine the relative contribution of preformed donor-specific human leukocyte antigen alloantibodies (DSAs) to this endpoint, and we defined strict criteria for the diagnosis of antibody-mediated rejection (AMR) in liver allografts. The inclusion criteria encompassed the availability of a pretransplant serum sample and both postreperfusion and follow-up tissue specimens for a blinded, retrospective re-review of histology and complement component 4d (C4d) staining. AMR was diagnosed on the basis of the presence of all 4 of the following strict criteria: (1) DSAs in serum, (2) histopathological evidence of diffuse microvascular injury/microvasculitis consistent with antibody-mediated injury, (3) diffuse C4d staining in the portal microvasculature with or without staining in the sinusoids or central veins in at least 1 sample, and (4) the exclusion of other causes of a similar type of injury. Patients thought to be experiencing definite AMR on the basis of routine histopathology alone showed the highest levels of DSA sensitization. Forty percent of patients with pretransplant DSAs with a pattern of bead saturation after serial dilutions developed AMR. Another multiparous female developed what appeared to be a strong recall response, which resulted in combined AMR and acute cellular rejection (ACR) causing graft failure. A contribution of DSAs to allograft failure could not be excluded for 3 additional patients who received marginal grafts. In conclusion, liver allograft recipients with preformed DSAs with a high mean fluorescence intensity despite dilution seem to be at risk for clinically significant allograft injury and possibly for loss from AMR, often in combination with ACR. Liver Transpl 20:218-227, 2014. © 2013 AASLD.

Abbreviations
ACR

acute cellular rejection

AIH

autoimmune hepatitis

AMR

antibody-mediated rejection

AST

aspartate aminotransferase

C4d

complement component 4d

CC

cryptogenic cirrhosis

DSA

donor-specific human leukocyte antigen alloantibody

HLA

human leukocyte antigen

IgG

immunoglobulin G

LT

liver transplantation

MELD

Model for End-Stage Liver Disease

MFI

mean fluorescence intensity

MFIsum

sum of the mean fluorescence intensities of each individual, undiluted donor-specific antibody

NASH

nonalcoholic steatohepatitis

OR

operating room

PBC

primary biliary cirrhosis

POD

postoperative day

PSC

primary sclerosing cholangitis.

Although antibody-mediated rejection (AMR) can cause or contribute to early allograft loss in ABO-compatible liver transplantation (LT), most programs do not routinely monitor for this potential cause of allograft dysfunction and loss. The most convincing case series of hyperacute liver allograft AMR highlighted 3 simultaneous liver-kidney transplants in which hyperacute rejection of the kidney was abruptly followed by precipitous liver allograft loss, which developed over a period of days instead of hours.[1] Liver histology demonstrated diffuse coagulative necrosis with immunoglobulin M– and C1q-positive staining consistent with hyperacute rejection. However, the perceived infrequency of these findings resulted in a lack of standardized diagnostic criteria for liver AMR.

Unlike hyperacute rejection, early liver allograft injury and loss have been found more frequently in crossmatch-positive patients,[2, 3] but not all crossmatch-positive patients experience graft dysfunction or loss. Those who do more commonly have a persistently positive posttransplant crossmatch, falling complement levels, circulating immune complexes, and refractory thrombocytopenia4-6; such a clinical profile supports the diagnosis of AMR in patients with appropriate histology and complement component 4d (C4d) staining. This phenomenon has been reaffirmed as a rare but ongoing issue in our current era of immunosuppression.[7-13] Early in the course of liver allograft injury due to donor-specific human leukocyte antigen alloantibodies (DSAs), the histology mimics preservation injury,[5, 14] which often evolves into acute cellular rejection (ACR) and later can progress into vanishing bile duct syndrome or chronic rejection.[12, 13, 15-20] Patients are often diagnosed with ACR when they have AMR or combined AMR and ACR for 2 reasons: (1) a broad spectrum of findings, including those appearing to be attributable to DSAs, have traditionally been attributed to cellular rejection, and (2) AMR of the liver is often quickly followed by cellular rejection and an accompanying cellular infiltrate.[10] However, more study is needed in this area.

Recent data have enabled the identification and characterization of DSAs that increase the risk of acute and chronic rejection in sensitized recipients.[7, 8, 12, 18, 19, 21, 22] Unique morphological features, many of which were previously attributed to ACR, aid in the early diagnosis of AMR in LT patients with circulating DSAs, but more precision, reliability, and reproducibility are needed.[8, 10, 12] Most of these features can be seen in animal models of AMR, and they pathognomonically show diffuse microvascular injury that can evolve into hemorrhagic necrosis.[23-26] As part of this diagnostic effort, which includes the aforementioned clinical profile and standard histology, C4d staining in formalin-fixed, paraffin-embedded tissue can point toward AMR as a cause of injury,[10, 11] but sensitivity is improved with fresh tissue.[8, 9] Despite these advances, only when AMR is accurately diagnosed early and successfully treated can graft outcomes improve because a delay in the diagnosis of AMR usually results in substantial allograft injury or failure.[27]

Potential complications from preformed DSAs resulted in some centers recommending the utilization of crossmatch-negative donors when possible,[28] but the poor predictive power in combination with the illness severity of recipients precluded implementation. The goals of this study, therefore, were to (1) define strict criteria for the diagnosis of acute AMR in liver allograft recipients, (2) identify the prevalence of preformed DSAs and subsequent AMR in patients with previously unexplained liver allograft loss within 90 days of LT, and (3) identify preformed DSAs and early histopathological characteristics in at-risk sensitized recipients.

PATIENTS AND METHODS

Case Selection and Analysis

From January 1989 to July 2010, ABO-compatible LT without another organ was performed 3137 times, and 337 patients (10.7%) experienced allograft loss within 90 days. Patients were excluded from the study for the following reasons: a lack of a pre-LT sample (n = 51); death with a functioning graft (n = 86); known causes of graft loss or death in our database or upon a re-review of the clinical history (n = 123); and a lack of back-table, postreperfusion, or follow-up biopsies (n = 17). All remaining 60 patients experienced unexplained graft loss and were included in the study.

Serum samples and liver tissues were prospectively collected on the day before transplantation or on the day of transplantation and were obtained from the Annette C. and Harold C. Simmons Transplant Institute biorepository after institutional review board approval. Blinded pretransplant serum samples were analyzed for all patients. In addition, for all patients, at least 1 back-table or postperfusion liver biopsy sample and at least 1 follow-up liver biopsy, explant, or autopsy tissue sample were stained and re-reviewed. All serum and tissue samples were linked with prospectively collected clinical and laboratory data locked into the Liver Transplant Research Database System.

Histopathological Analysis

Twelve patients underwent back-table liver biopsy, 59 patients underwent postreperfusion liver biopsy, 30 patients underwent follow-up indication liver biopsy, 48 patients had explant samples, and 2 patients had autopsy tissue samples. All these samples were re-evaluated initially without knowledge of DSA or C4d staining results by pathologists experienced in liver and transplantation pathology (S.M.S. and A.J.D.). All formalin-fixed, paraffin-embedded tissue samples, with the exception of all back-table biopsy samples and 1 postreperfusion biopsy sample (insufficient tissue), were stained for C4d. The evaluation of C4d was based on endothelial cell staining of the portal microvascular, sinusoidal, and central vein endothelium, as previously described.[10] AMR was determined to be a substantial contributor to allograft loss only when the following 4 criteria were all met: (1) DSAs in serum, (2) diffuse microvascular injury seen on hematoxylin and eosin staining consistent with antibody-mediated injury,[5, 10] (3) diffuse C4d staining in the portal microvasculature with or without staining present in the sinusoids or central veins in at least 1 sample, and (4) a clinical profile consistent with AMR and reasonably excluding other insults that might cause similar injury. Diffuse C4d staining was defined as strong staining of portal vein branches and portal capillaries in a majority of the portal tracts. In cases with evidence of 2 or 3 criteria without a precise etiology for allograft loss, DSAs could not be excluded as a contributor to allograft failure.

Human Leukocyte Antigen (HLA) Tissue Typing

All recipients and donors were typed for HLA-A, HLA-B, HLA-DRB1, HLA-DRB345, and HLA-DQ with commercially available serological typing trays (Terasaki HLA tissue typing trays) or molecular methods (and Micro SSP or LabType SSO; One Lambda, Inc., Canoga Park, CA). HLA-Cw typing for donors was available for half of the cases. Before 1998, serology was used exclusively. Since 1998, all donor class I and II HLA typing and patient class II typing have been performed with molecular methods, whereas patient class I typing has been performed with serology.

HLA Immunoglobulin G (IgG) Antibody Determination

All sera were anonymized and sent to the Terasaki Foundation Laboratory for DSA evaluation. The detection of HLA IgG antibodies was performed with LABScreen single-antigen class I (lot 6) and II beads (lot 8; One Lambda) according to the manufacturer's protocol. A normalized trimmed mean fluorescence intensity (MFI) > 1000 was considered positive; when the MFI value exceeded 10,000 for a single DSA, serial dilutions (neat, 1:3, 1:9, 1:27, and 1:81) were performed. Bead saturation was defined as an antibody that maintained a similar MFI (with at least an MFI >15,000) regardless of dilution (≥1:27).

IgG Subclass DSAs and C1q Evaluation

All DSA-positive sera were further evaluated for IgG subclasses and C1q-binding antibodies. The evaluation of IgG subclass DSAs was performed with the same LABScreen single-antigen beads that were used for the initial HLA IgG antibody determination, but the secondary antibody was replaced with an antibody specific for each of the IgG subclasses, as previously described.[19] A normalized trimmed MFI value > 500 was defined as positive for IgG subclasses. For C1q-fixing antibody determinations, the C1qScreen kit (One Lambda) was used according to the manufacturer's protocol. A normalized trimmed mean MFI > 1000 was considered positive for C1q-fixing antibodies.

Statistical Analysis

Patient characteristics are reported as medians and interquartile ranges as appropriate and are compared with the likelihood ratio chi-square test for categorical values and with the Kruskal-Wallis test for continuous variables in the tables as appropriate. Statistical significance is defined as P < 0.05. SAS 9.1 was used for all statistical analyses.

RESULTS

Table 1 details characteristics of the 60 patients with unexplained early allograft loss, who were divided into 3 groups: (1) patients who experienced AMR as the probable primary cause of early allograft failure (n = 3), (2) patients for whom DSAs could not be excluded as a contributor to early allograft failure (n = 3), and (3) patients without definitive histopathological features of AMR (n = 54). Only female sex was statistically significantly more common in patients with AMR (which was expected); there were no other statistically significant differences between the groups' demographics.

Table 1. Patient and Donor Characteristics for the 60 Recipients With Early Graft Loss (<90 Days After LT)
 AMR Changes Present (n = 3)Possible Contribution From DSAs (n = 3)No Definitive Evidence of AMR (n = 54)
  1. a

    The data are presented as medians and interquartile ranges.

  2. b

    P = 0.02. No other statistically significant differences were found between the groups.

  3. c

    Day 1 immunosuppression.

Age (years)a56 (47-58)52 (40-55)50 (41-58)
MELD scorea27 (12-40)16 (13-33)17 (12-26)
Preoperative dialysis [n (%)]1 (33)03 (6)
Cold ischemia time (hours)a11.8 (7.5-12.2)8.4 (7.8-8.7)9.6 (8.2-11.7)
Warm ischemia time (minutes)a62 (52-84)52 (50-55)56 (46-68)
OR time (hours)a7.8 (5.1-7.9)6.3 (4.8-8.2)6.8 (5.8-7.8)
Time to graft loss (days)a19 (19-40)3 (0-54)5.5 (3-19)
Male sex [n (%)]b01 (33)38 (70)
Race [n (%)]   
African American001 (2)
Hispanic01 (33)8 (15)
Other001 (2)
Caucasian3 (100)2 (67)44 (81)
Liver disease [n (%)]   
Hepatitis C02 (67)14 (26)
PSC/PBC/AIH2 (67)09 (17)
Alcohol1 (33)08 (15)
NASH/CC01 (33)6 (11)
Metabolic/congenital003 (6)
Other0014 (26)
Hepatocellular carcinoma [n (%)]005 (9)
Donor age (years)a42 (37-45)56 (55-59)50 (36-59)
Donor race [n (%)]   
Caucasian3 (100)2 (67)37 (69)
African American01 (33)10 (19)
Other007 (13)
Induction [n (%)]1 (33)1 (33)16 (30)
Calcineurin inhibitor (%)c   
Tacrolimus663329
Cyclosporine0052
None336619
Steroids (%)c1006660
Sirolimus (%)c0012
Mycophenolate (%)c06635

Preformed DSAs were found in 53% of the total patient population, but there were differences in the DSA characteristics between the 3 groups (Table 2). The candidate cases were identified on the basis of hematoxylin and eosin staining alone, but the results correlated with the median MFIs of patients' single highest class I DSAs [20,266, 20,109, and 2783 for patients with AMR, patients for whom DSAs could not be excluded as a contributor to allograft loss, and patients without AMR, respectively (P = 0.03)] and class II DSAs [11,794, 5414, and 2385, respectively (P = 0.07)]. DSA-positive patients had 1 to 8 distinct DSAs detected. In addition, an effort to capture the potential detrimental effects of multiple DSAs through the determination of the sum of the mean fluorescence intensities of each individual, undiluted donor-specific antibody (MFIsum) showed an even stronger correlation with the histopathological findings. The median MFIsum was considerably higher for patients with AMR (95,988) versus patients for whom DSAs could not be excluded as a contributing factor to allograft loss (40,752) and patients without AMR (3643, P = 0.002). Because MFIsum does not capture the true amount of DSAs (on account of bead saturation), we performed dilutional analyses. Eighty percent of the patients with at least 1 DSA with bead saturation at a 1:27 or 1:81 dilution had AMR as a probable primary reason for allograft loss or as a nonexcludable contributor to allograft loss, and 2 of the 3 AMR cases showed focal or diffuse C4d staining of microvascular endothelial cells on postreperfusion biopsy.

Table 2. DSAs in Patients With Early Graft Loss (<90 Days): Classes, Median MFIs, and Numbers of Patients With Newly Identified Causes of Graft Loss Categorized by the Presence or Absence of DSAs
ClassAMRPossible Contribution From DSAsNo AMRP Value
  1. a

    The data are presented as medians and interquartile ranges for positive patients.

  2. b

    At least 1 DSA maintained a similar MFI (>15,000) regardless of the dilution (≥1:27).

  3. c

    Two patients with DSA saturation and 1 patient without bead saturation had DSAs as nonexcludable contributors to allograft failure.

I [n (%)]02 (67)7 (13)0.002
II [n (%)]0015 (28)
I and II [n (%)]3 (100)1 (33)4 (7)
None [n (%)]0028 (52)
MFIAMRPossible Contribution From DSAsNo AMRP Value
Class Ia20,266 (4815-26,234)20,109 (16,804-22,717)2783 (1436-11,527)0.03
Class IIa11,794 (9429-24,865)44162385 (1287-4944)0.07
MFIsuma95,988 (26,336-131,135)40,752 (39,691-63,284)3643 (1436-12,805)0.002
Causes of Graft LossDSA SaturationbDSA-Positive (No Saturation)DSA-NegativeTotal
AMR ± ACR [n (%)]2 (40)1 (4)03 (5)
ACR/ischemia [n (%)] 2 (7)5 (18)7 (12)
Fat/ischemia [n (%)] 11 (41)6 (21)17 (28)
Vascular [n (%)] 4 (15)8 (29)12 (20)
Unknown [n (%)]3 (60)c9 (33)c9 (32)21 (35)

Details for the DSAs, laboratory parameters, and histology for each of the 3 AMR cases are shown in Figs. 1 to 3. The first 2 cases (Figs. 1 and 2) shared multiple high-MFI class I and II DSAs (some with bead saturation at dilutions ≥ 1:27, several of the IgG3 subclass, and many C1q-positive). These 2 patients also shared a slightly delayed, high aminotransferase peak, platelet consumption, and persistent bilirubin elevation. The third case was a multiparous female with moderate-MFI DSAs of the IgG1 subclass that were C1q-negative who experienced an apparent recall AMR response characterized by a delay in aminotransferase elevations and platelet consumption.

Figure 1.

Patient 1 was a 56-year-old female with a gynecological history significant for 1 still birth and with a MELD score of 12 who underwent primary LT for AIH with a 37-year-old male donor after 12.2 hours of cold ischemia in 1990. She had (A) saturation of single-antigen beads of numerous DSAs at a 1:81 dilution against class I before transplantation, (B) 7 distinct DSAs (MFIs reported at a 1:3 dilution) in all, and (C) a clinical course characterized by early postoperative platelet consumption and recurrent rejection. (D) The postreperfusion biopsy showed endothelial reactivity and diffuse C4d positivity (top panels), and this was followed by evidence of combined AMR and ACR on day 16 characterized by diffuse microvascular endothelial cell injury (bottom panels).

Figure 2.

Patient 2 was a 47-year-old primiparous female with a MELD score of 40 who underwent primary LT for alcohol-induced cirrhosis with a 45-year-old female donor after 7.6 hours of cold ischemia in 2005. (A) She had high mean class I and II single-antigen beads at a 1:81 dilution of numerous class I and II DSAs, (B) 8 distinct DSAs were found, and (C) her clinical course was characterized by early postoperative platelet consumption. (D) The postreperfusion biopsy showed endothelial reactivity and C4d positivity (top panels), and this evolved into combined AMR and ACR in her only liver biopsy performed on day 12, which was characterized by very severe endothelial cell reactivity and damage (bottom panels).

Figure 3.

Patient 3 was a 58-year-old multiparous female with a MELD score of 27 who underwent primary LT for PBC with a 58-year-old male donor after 11.8 hours of cold ischemia in 2004. She had (A,B) 5 distinct preformed lower MFI class I and II DSAs (MFIs reported at a 1:3 dilution) and (C) a clinical course characterized by an apparent recall response. (D) Although the postreperfusion biopsy was unremarkable (not shown), the day 7 indication liver biopsy showed combined AMR and ACR (top panels). The autopsy sample on day 20 showed severe combined AMR and ACR with active lymphocytic intimal inflammation, intimal foam cells (arrow), and continued C4d positivity despite the postmortem nature of the tissue sample (bottom panels).

For all 3 cases with definitive evidence of AMR, the histology was characterized primarily by portal microvascular endothelial cell hypertrophy or near hobnailing, eosinophilic and neutrophilic portal capillaritis/inlet venulitis, portal edema, and a ductular reaction with diffuse portal microvascular C4d staining. All 3 cases for whom DSAs could not be excluded as a contributor to allograft loss had lower C1q values, even though 2 had DSAs that reached bead saturation at a dilution ≥ 1:27. The full case report details for all 6 cases can be found in the supporting information, and Table 3 outlines the patient, donor, and DSA characteristics for all DSA-positive cases. The remaining cases, except as indicated, had postperfusion and follow-up tissue samples that were C4d-negative.

Table 3. Patient, Donor, and Antibody Characteristics for the 32 Patients With DSAs Present Before LT
Patient NumberACRFat (%)aPreservation Injury on Postreperfusion Liver BiopsyMFIsumC1qsumIgG3sumDays to Graft LossPatient Age (Years)Donor Age (Years)Cold Ischemia Time (Hours)MELD ScoreCause
  1. NOTE: Although all patients originally had unexplained graft loss, an etiology was able to be determined for many retrospectively. A single-antigen bead test was used to detect C1q-fixing antibodies.

  2. a

    Microvesicular fat is described as mild or moderate; macrovesicular steatosis is indicated as a percentage.

  3. b

    Preservation injury on postreperfusion liver biopsy was based on a combination of the severity and location of hepatocyte necrosis and netrophilic inflammation (Kakizoe S, et al. Hepatology 11(6) 932; 1990).

  4. c

    Diffuse C4d staining was found on a postreperfusion liver biopsy sample.

  5. d

    Diffuse C4d staining was found on an indication liver biopsy sample.

  6. e

    Focal C4d staining was found on a postreperfusion liver biopsy sample.

10101115,79158,195280540563712.212AMRcd
22Moderate/403131,59670,20939,0371947457.640AMR/ACRed
31Mild/20120,9750019584211.827AMR/ACRd
40Moderate/40070,39611,3310052598.416Fat/DSA?
530140,75233,688790340567.813ACR/DSA?
605139,691282405455558.733DSA?
705331,13010,93811,9853432113.220Unknown
80Mild/5130,86321,0498185450516.316Unknown
90Mild/5112,8050019585013.822Unknown
100538076001242287.513Unknown
110Moderate/15378510063728N/A28Unknown
120536711010241513814.114Unknown
13102494406248317346.432Unknown
140002679001038166.436Unknown
150532650003454210.817Unknown
16101212100252454.38Unknown
170Moderate121,88000138239.921Ischemia
180Moderate316,565003862748.922Ischemia/fat
1900115,915003333381130Vascular
200Mild314,83901153359409.615Ischemia
210Mild/52695600563637.613Vascular
22030337490830462538.319Ischemia/fat
2317033538001165211.79Ischemia/fat
240603333800386058N/A35Ischemia/fat
253032863006544713.38Ischemia/ACR
26003144300258408.917Vascular
2705021436003673210.320Ischemia/fat
28065312670017505311.510Ischemia/fat
290Moderate/511287001160545.734Vascular
30103127917749293412114.317Ischemia/ACR
310302116600252429.614Ischemia/fat
320Severe3104000367108.712Ischemia/fat

Although the primary evaluation of the clinicopathological parameters did not reveal a reason for allograft loss in all cases, a re-evaluation of the clinical profile and reperfusion, post-LT, and explant pathology resulted in the identification of the cause of allograft failure for 65% of the patients (Table 2): severe ischemia compounded by either underlying severe donor macrovesicular steatosis (28%) or ACR without evidence of AMR (11%) and unrecognized hepatic artery and portal vein thrombosis or suboptimal liver perfusion (20%) were common.

DISCUSSION

Liver allografts are more resistant to preformed DSAs causing early or acute AMR than other solid organ transplants, but accumulating evidence favors those who have recognized AMR as a potential cause of liver injury.[1-3, 8, 14, 28] Recognized risk factors that were identified in the era before single-antigen bead technology allowed accurate HLA DSA determinations included high-titer preformed antibodies, the persistence of anti-donor antibodies after transplantation, and otherwise unexplained thrombocytopenia and hypocomplementemia.[4-6, 14, 29]

In an effort to further refine risk factors and establish an early AMR diagnosis, we analyzed 60 cases of idiopathic early (<90 days after LT) allograft loss in patients with precisely defined preformed HLA alloantibodies on the basis of the following stringent criteria: (1) DSAs in serum, (2) histopathological evidence of diffuse microvascular endothelial cell injury/microvasculitis, (3) diffuse and strong C4d positivity in at least 1 of the tissue samples, and (4) the reasonable exclusion of other causes of injury that might result in similar findings. Importantly, this study shows that strict adherence to routine histopathological findings of diffuse microvascular endothelial cell reactivity, which approaches hobnailing and microvasculitis, reliably point toward recipients with the highest levels of sensitization. When these findings are combined with diffuse C4d positivity, evidence of serum sensitization, and the exclusion of other insults that might cause a similar injury pattern, one can confidently establish an acute AMR diagnosis. These observations should serve as an impetus for further study of the correlation between routine histopathology, C4d deposition, and DSAs in liver allograft recipients.

Different C4d staining patterns have been described, and although a consensus agreement regarding a diagnostic pattern is not available, diffuse portal microvascular positivity in formalin-fixed, paraffin-embedded samples is emerging as most strongly correlated with DSA-induced injury.[8-12] In our opinion, however, when one is attempting to establish a diagnosis of AMR in this new era of sensitive DSA detection, stringent criteria should be used to avoid overdiagnosis. C4d-negative AMR has been identified in renal allografts and likely also occurs in the liver, but until more is learned about liver AMR, we favor taking this conservative approach.[30]

Interestingly, 53% of all patients with early allograft failure had preformed DSAs with MFIs > 1000. This is a substantially higher percentage than that found in our recently reported population of 1270 consecutive LT recipients (36%).[13] Moreover, dilutional studies of strong pretransplant DSAs found 5 patients with at least 1 DSA with a pattern of bead saturation at a dilution ≥ 1:27. Two of these 5 cases (40%) showed convincing evidence of AMR in tissue specimens, as defined previously. In another 2 patients (40%), antibody-mediated injury could not be excluded as a contributor to early allograft loss. These observations are consistent with previous, less specific dilutional analyses using conventional lymphocytotoxic crossmatches.[2, 4]

The rarity (2/59) of diffuse C4d positivity in our formalin-fixed, paraffin-embedded archival tissue samples from postreperfusion liver biopsies might be related to the high level of sensitization in our recipients who developed definitive evidence of AMR, the low sensitivity of our C4d stain, or both, but both of these recipients subsequently developed AMR resulting in allograft loss. This study, however, was biased by the inclusion of only patients who experienced allograft failure. Nevertheless, one might consider performing postreperfusion liver biopsy and staining the sample for C4d in recipients positive for preformed, high-MFI DSAs as an early warning sign for AMR. Regardless, because of the intercenter variability of sensitivity for C4d staining, this finding will need confirmation.[8, 9, 11, 12, 31]

Notably, 1 of the 3 patients who developed AMR had a moderate MFI for several non–complement-fixing DSAs and benign reperfusion biopsy findings. She was, however, a multiparous female and developed what appeared to be a very strong recall response with dramatic AMR and ACR: diffuse microvascular endothelial cell hypertrophy and damage, diffuse C4d positivity on follow-up biopsies, and lymphocytic arteritis with rare intimal foam cells within 27 days of transplantation. The combination of arteritis and diffuse C4d positivity is accepted as evidence of AMR for other solid organ allografts.[32-36] Unfortunately, a posttransplant follow-up serum sample was not available for analysis for any of our AMR cases.

This more detailed analysis confirms previous studies showing that severe AMR is a relatively rare event after transplantation and quickly evolves into combined AMR and ACR.[5, 8, 28] It is unknown whether current immunosuppression and surgical advances would have altered the outcomes of some of our earlier patients.[7, 8, 10, 12] With the inclusion of only convincing cases in this series, AMR was thought to be the primary cause of early allograft loss for only 1% of all early allograft failures and a possible contributing factor to early allograft loss for another 1% of cases; the figures rise to 5% for historical cases with a previously undetermined cause of failure and 10% for those with preformed DSAs.

Although this is somewhat speculative, a combination of a diffuse pattern of C4d staining and serology suggests that high-MFI class I DSAs that override potential innate protective mechanisms might have the most potential for immediate damage resulting in rapid allograft failure. This contention is based on the diffuse expression of class I HLA throughout the liver microvasculature, in contrast to the limited expression of class II HLA in the portal capillaries in a subpopulation of donor livers, unless the liver is affected by an inflammatory disorder.[23, 37, 38] Because of their limited expression, class II DSAs may cause slower graft injury, and they are associated with an increased risk of ACR and chronic rejection.[12, 13, 18, 21] However, because all definite AMR cases had class I and II DSAs present before the transplant, it is possible that class I DSA–mediated injury leads to the up-regulation of class II HLA in the liver and enables the pathological potential of class II DSAs. Conversely, no definitive proof of AMR was detected in 80% of the DSA-positive recipients. However, most of these had low-MFI DSAs. Stringent criteria were used to identify convincing cases, but this does not exclude the possibility of more subtle forms of antibody-mediated injuries or longer term consequences if the allografts had survived.

Fortunately, in confirmation of previous studies, preformed DSAs with lower MFIs and, by extension, most low-titer crossmatch results did not result in identifiable graft damage or loss when standard low-risk donors were used.[2, 7, 8, 31, 39] The role that DSAs play in allograft loss when marginal organs are transplanted, however, remains unclear. This consideration was evident from our 3 cases for whom it was determined that DSAs or AMR was not the primary cause of liver allograft loss but DSAs could not be excluded as a contributor to liver allograft failure. Therefore, caution may be needed when high-risk donors are being matched to patients with preformed, high-MFI DSAs. Although we find this hypothesis interesting, further study is needed to define the precise donor and antibody characteristics that should not be mixed. Therefore, the testing of all patients before transplantation and follow-up testing to determine persistence after transplantation are needed to define the precise donor and antibody characteristics that cannot be matched.

Finally, this study suffered from several weaknesses. First, it was a retrospective, single-center study. Second, we relied entirely on formalin-fixed, paraffin-embedded tissue samples (some decades old), which are known to have inferior C4d staining sensitivity. Third, follow-up serum samples for determining the postoperative presence of DSAs would have been ideal for substantiating the AMR diagnoses because in the majority of cases, DSAs disappear, yet those for which they have been shown to persist have an increased risk of inferior outcomes. Fourth, we cannot exclude the possibility that current standards of care would have improved the outcomes seen in these historical patients. We attempted to mitigate these shortcomings by first reviewing the cases for a characteristic pattern of microvascular injury without knowledge of the DSA status or C4d staining and then rigorously excluding other causes of allograft injury that might have shown a similar pattern of histopathology.

In conclusion, preformed, high-MFI DSAs can result in substantial early graft injury from AMR, which usually combines with ACR to cause allograft failure. Although this was a rare event [1% of all early graft losses (3/337) and 5% of previously unexplained early graft losses (3/60)], graft loss due to AMR occurred in 40% of the patients with at least 1 preformed DSA with a bead saturation at a dilution ≥ 1:27; this was especially true for those with a postreperfusion formalin-fixed, paraffin-embedded biopsy sample diffusely positive for C4d. Testing all LT candidates for DSAs before transplantation and conducting protocol postreperfusion biopsy and follow-up single-antigen bead testing would be ideal for determining which highly sensitized patients might be at risk and should be placed at centers dedicated to studying the phenomenon. Broader recommendations for carrying out similar testing in low-risk patients at all centers must be proven to be cost-effective before their widespread implementation.

Ancillary