Sinusoidal C4d deposits in liver allografts indicate an antibody-mediated response: Diagnostic considerations in the evaluation of liver allografts


  • Kenneth Andreoni was formerly an associate and adjunct professor of surgery at the University of North Carolina at Chapel Hill. This study was performed at the University of North Carolina.


There is a paucity of data concerning the correlation of complement component 4d (C4d) staining in liver allografts and antibody-mediated rejection. Data about the location and character of C4d deposits in native and allograft liver tissues are inconsistent. We performed C4d immunofluorescence (IF) on 141 fresh-frozen liver allograft biopsy samples and native livers, documented the pattern of C4d IF staining, and correlated the findings with the presence of donor-specific alloantibodies (DSAs). A linear/granular sinusoidal pattern of C4d IF was noted in 18 of 28 biopsy samples obtained after transplantation from patients with positive crossmatch and detectable donor-specific alloantibody (pos-XM/DSA) findings. None of the 59 tested biopsy samples from patients with negative crossmatch and detectable donor-specific alloantibody (neg-XM/DSA) findings were C4d-positive (P < 0.001). No significant association was found between pos-XM/DSA and C4d IF staining in other nonsinusoidal liver compartments. To compare the results of sinusoidal C4d staining with IF and 2 immunohistochemistry (IHC) techniques, C4d IHC was performed on 19 liver allograft biopsy samples in which a sinusoidal pattern of C4d IF had been noted. Sinusoidal C4d IHC findings were negative for 17 of the 19 biopsy samples; 2 showed weak and focal staining, and both patients had pos-XM/DSA findings. Portal vein endothelium staining was present in only 1 IF-stained biopsy sample (pos-XM/DSA) but in 11 IHC-stained biopsy samples (2 of the 11 samples had neg-XM/DSA findings). We conclude that sinusoidal C4d deposits detected by IF in frozen tissue samples from liver allograft recipients correlate with the presence of DSAs and an antibody-mediated alloresponse. These observations are similar to findings reported for other solid organ transplants and can provide relevant information for patient management. Further validation of IHC techniques for C4d detection in liver allograft tissue is required. Liver Transpl 18:641–658, 2012. © 2012 AASLD.

Historically, antibody-mediated rejection (AMR) in ABO-compatible liver transplantation has been considered to be only a minor clinical concern because of the rare occurrence of graft failure in this setting. Approximately 10% to 30% of liver transplant candidates are sensitized to human leukocyte antigens (HLAs) according to the calculated panel reactive antibody test,1, 2 but less than 15% of sensitized patients develop severe AMR.3 However, it became evident in 2 independent studies from 19954 and 20113 that a persistent positive crossmatch (XM) in the first weeks after transplantation is a predictor for the development of acute AMR and graft failure. Histological criteria for establishing a diagnosis of AMR in liver allografts are only slowly evolving, and in contrast to other solid organ transplants, the routine evaluation of complement component 4d (C4d) is not a standard procedure in the pathological processing of liver graft biopsy samples.

The presence of C4d staining along capillaries in kidney, pancreas, heart, and lung transplants in the setting of graft dysfunction is strong evidence for the consideration of AMR.5-8 The most experience with the evaluation of C4d has been gained in kidney transplantation. There are more then 300 publications describing the characteristics of C4d deposits in kidney transplantation.9 In contrast, there are few publications and substantial confusion about the characteristics of C4d staining in liver allograft tissue.10-12 Some studies have attempted to use C4d as a marker for differentiating acute cellular rejection (ACR) from recurrent hepatitis C.11, 13-15 Other publications have attempted to describe C4d in the context of complex immunological events that involve cellular and antibody-mediated injury/rejection and are difficult to separate.16 Only 2 studies have correlated the presence of C4d on liver sinusoidal endothelial cells with the presence of circulating donor-specific alloantibodies (DSAs) directed against HLA molecules3, 17; a case report on late ACR/AMR identified DSAs and retrospectively detected C4d in small portal vessels by immunohistochemistry (IHC), but the technique was not described.18 Most previously published works are case reports and/or small case series, and they often lack documentation of pretransplant and posttransplant XM results and/or DSA levels, as summarized in Table 1.3, 11, 14-17, 19-30 In a recent work based on a large number of patients, Lunz et al.30 provided only vague information about XMs and DSAs at the time of biopsy, and they focused mostly on ACR and/or chronic rejection (CR). They also studied tissue stored in paraffin. In the literature, there are inconsistencies in the description of the intensity of C4d deposits and their locations in various compartments of liver tissue. The vast majority of published works are based on tissues fixed in formalin and embedded in paraffin and then retrospectively evaluated with various IHC techniques (Table 1). Surprisingly, the gold standard evaluation of C4d, which involves immunofluorescence (IF) performed on frozen tissue, has scarcely been used for liver tissue.31 From other reports, it is well known that C4d staining can be found on various structures without direct diagnostic significance. This led us to use IF as the gold standard for C4d detection on fresh-frozen tissue and to carefully look for areas of diagnostic staining similar to capillary C4d findings in other organs.

Table 1. C4d Reports With Various Tissue Storage and Staining Techniques
StudyDSA/XM ReportTechniqueStorageDiagnosisC4d
Dankof et al.16 (2005)UnknownIHC/IFParaffinACRPositiveReported as positiveNegative
Schmeding et al.11 (2006)UnknownIHCParaffinACRPositivePositiveNegative
Bu et al.19 (2006)UnknownIHCParaffinACRPositive Positive
Sakashita et al.20 (2007)Positive XMIHCParaffinACR/CR + moreVariable stromal Perivenular
 Negative XMIHCParaffinACRStromalSome internal laminaNot reported
Bellamy et al.21 (2007)CDC/incompleteIHCParaffinACRPositive Positive
Martelius et al.22 (2009)Positive (1/7)IHCFrozenCRPositive Weak
Barth et al.23 (2010)Developed DSAsIHCParaffinNo liver rejectionNot reported Negative
Schmeding et al.24 (2010)UnknownELISAFrozenACRDid not show ELISA value in C4d determination
Troxell et al.25 (2006)UnknownIHCParaffinAMR?/ACRPositivePositiveMinimal
Jain et al.14 (2006)UnknownIHCParaffinACRPositiveNegativeNegative
Lorho et al.15 (2006)UnknownIFFrozenACR/CRPositive, location not specified
Kamar et al.26 (2009)PresentIFNot reportedACR/AMRPositive, location not specified
Rostron et al.27 (2005)PresentNot reportedParaffinACR/AMREquivocal, location not specified
Musat at al.28 (2011)PresentIHCParaffinACRPredominantFewSome
Kozlowski et al.3 (2011)PresentIFFrozenAMRNegativeOccasional internal laminaPositive
 PresentIFFrozenAMR/ACRPositiveOccasional internal laminaPositive
Watson et al.17 (2006)PresentIFFrozenAMRNegativeNegativePositive
Aguilera et al.29 (2011)UnknownIHCParaffinCRNegativeNegativePositive
Lunz et al.30 (2012)Vague reportIHCParaffinCR/ACRPositive and negativePositive and negativePositive and negative

We previously published a well-documented case of isolated AMR without a cellular component of acute rejection; a positive XM, circulating DSAs, and graft dysfunction were found after transplantation. In this case, C4d deposits were found by IF along sinusoids.17 We further confirmed the presence of linear C4d deposits in liver sinusoids in additional cases of AMR when IF was used on frozen biopsy specimens.3 The current study was designed to further evaluate whether complement-binding HLA antibodies predominantly target the liver sinusoidal endothelium in a fashion similar to that observed for other solid organs in which capillaries are targeted by alloantibodies and complement. We also suspect that inconsistencies reported in the literature concerning the distribution of C4d staining in liver allografts are related to the techniques (eg, the less sensitive IHC method for formalin-fixed tissue samples) and the lack of detailed qualitative descriptions of the staining patterns. Most studies have also failed to use control tissues to differentiate specific findings from nonspecific findings.

To the best of our knowledge, this is the only study to date that has both evaluated liver tissues collected in a prospective fashion from native livers and allografts for C4d deposits and determined the status of HLA antibodies/DSAs in the recipient circulation before and after transplantation. We also conducted a comparative study of liver tissue C4d staining examined by IF and IHC methodologies.


ACR, acute cellular rejection; AMR, antibody-mediated rejection; BT, before transplantation; C4d, complement component 4d; CDC, complement-dependent cytotoxicity; CR, chronic rejection; DSA, donor-specific alloantibody; EDTA, ethylene diamine tetraacetic acid; ELISA, enzyme-linked immunosorbent assay; FU, follow-up; HCC, hepatocellular carcinoma; H&E, hematoxylin and eosin; HLA, human leukocyte antigen; IEL, internal elastic lamina; IF, immunofluorescence; IHC, immunohistochemistry; NA, not available; neg-XM/DSA, negative crossmatch and detectable donor-specific alloantibody; NS, not significant; PNF, primary nonfunction; POD, postoperative day; POM, postoperative month; pos-XM/DSA, positive crossmatch and detectable donor-specific alloantibody; NT, not tested; PP, post perfusion; PTLD, posttransplant lymphoproliferative disorder; RAI, rejection activity index; THV, hepatic vein; Tris, tris(hydroxymethyl)aminomethane; UHB, University Hospital Basel; UNC, University of North Carolina at Chapel Hill; XM, crossmatch.


Collected liver tissue was submitted for standard formalin fixation and histological examinations and for freezing in an optimal cutting temperature compound with subsequent C4d staining by IF. The liver tissue was obtained during organ preparation on the back table before transplantation (BT) and/or approximately 2 hours after the completion of liver allograft revascularization [ie, post perfusion (PP)]. Biopsy samples were also collected during the postoperative follow-up (FU) for diagnostic purposes. Control biopsy tissue samples were obtained from 7 nontransplant patients who underwent partial resection for various noncancerous liver lesions [2 males and 5 females; mean age = 46 years (range = 16-74 years, median = 47 years)]. BT specimens collected during backbench procurement and before exposure to recipients' blood also served as controls. Cases were followed prospectively from 2004 to 2011. Only patients with complete data were incorporated into the study (pretransplant XM status, DSA levels before and after transplantation and at the time of biopsy, biopsy specimens, and clinical FU information). The institutional review board of the University of North Carolina at Chapel Hill (UNC) approved this study.

C4d IF Staining.

Staining was performed on 4-μm-thick sections cut from tissue embedded in an optimal cutting temperature compound and frozen in precooled isopentane. A mouse monoclonal antibody (Quidel, San Diego, CA) was used for the detection of C4d as described in a previously published protocol.8 The results of the C4d IF staining involving sinusoids were recorded semiquantitatively. Diffuse staining was defined as staining involving >50% of the sinusoidal compartment, whereas focal staining involved <50% of the sinusoidal compartment. The pattern of C4d deposition within sinusoids was described as either linear (Fig. 1A) or granular (Fig. 2). The assessment followed general criteria used in kidney allograft pathology. In the evaluation of C4d deposits in kidney allografts, staining patterns along the peritubular capillary beds in the renal cortex and medulla are evaluated. In liver tissue, linear staining (mild to marked) shows a continuous staining pattern with fine, linear decoration of sinusoidal spaces in a wallpaper-like fashion. Granular staining shows fine, closely clustered dots arranged in a stringlike pattern in the sinusoidal space (a pattern very similar to the pattern observed in membranous glomerulopathies and the staining of glomerular basement membranes). Chunky staining refers to relatively short, thick, and irregular cords in the sinusoidal compartment showing an intense, positive staining reaction. The intensity of staining was recorded as follows: (0) negative, (1+) weak, (2+) moderate, and (3+) strong. The assessment followed criteria used in nephropathology. The C4d staining intensity was semiquantitatively graded from 0 (negative) to 3+ (strongly positive). The staining of more than 10% of the sinusoidal compartment in the liver (intensity of 1+ or greater) was considered positive (focal, 10%-50%; diffuse, >50%). C4d deposition in the portal stroma, the endothelium of portal vessels or hepatic veins, and the arterial internal elastic lamina (IEL) was also documented. IF staining results were scored by 2 observers. Consensus data were used for statistical analysis.

Figure 1.

(A) Linear sinusoidal endothelial C4d deposits, (B) negative C4d IHC findings for the same specimen, and (C) focal sinusoidal C4d deposits by IHC techniques. With the IF technique, the deposits were linear.

Figure 2.

Granular sinusoidal C4d IF staining.

C4d IHC Staining

C4d IHC was performed on matched formalin-fixed, paraffin-embedded specimens for a subset of cases; this revealed positive sinusoidal IF staining for C4d as well as 2 negative C4d specimens (a control group). The staining method was similar to that used for renal allograft biopsy specimens. Antigen retrieval was performed via pressure cooking at 125°C for 60 seconds and then at 90°C for 10 seconds. The slides were incubated for 30 minutes with the primary antibody (polyclonal anti-human C4d; catalog number 004-BIRC4d, ALPCO Diagnostics), which was diluted 1:20 in a common antibody diluent (catalog number HK-156-5K, Biogenex). After washing with a wash buffer (×10) containing tris(hydroxymethyl)aminomethane (Tris)/hydrochloric acid and sodium chloride (catalog number S3006, Dako) and peroxidase blocking, the tissue was incubated with an Envision+ System–labeled polymer/horseradish peroxidase anti-rabbit antibody (catalog number SH25-500D, Dako).

The antibody dilution was optimized with a known C4d-positive renal allograft biopsy sample. Renal allograft tissue known to contain C4d deposits was used as a positive control.

Evaluation of Interlaboratory C4d IHC Staining Variability

To evaluate the interlaboratory staining variability of IHC procedures that might influence the staining results for C4d, unstained slides of 7 liver tissue specimens were selected that demonstrated negative (n = 2), weak (n = 2), or strong (n = 3) staining of the sinusoidal endothelium. They were sent to an outside institution for C4d IHC staining. Two different routinely used IHC staining protocols were used at the Department of Pathology of the University Hospital Basel (UHB; Basel, Switzerland). The first protocol used at UHB was identical to our staining protocol at UNC. The second protocol employed an ethylene diamine tetraacetic acid (EDTA)/Tris buffer instead of the citrate buffer. Sections of kidneys with known C4d deposits were used as positive controls. Two pathologists who were blinded to all clinical and laboratory data scored the slides independently.


Liver core biopsy tissues were fixed in 10% buffered formalin, and 5-μm sections were cut from the paraffin-embedded tissue for hematoxylin and eosin (H&E) and trichrome stains for morphological evaluations. The slides were evaluated and scored on the basis of previously described pathological criteria for liver AMR.32

Detection of Alloantibodies and DSAs in Transplant Recipients

Recipient sera were routinely screened for the presence of class I and II HLA antibodies with FlowPRA screening beads (One Lambda, Inc., Canoga Park, CA). Sera demonstrating positive FlowPRA screens were tested with FlowPRA class I and/or II single antigen beads (One Lambda). The reactivity to individual HLA antigens was considered positive in FlowPRA single antigen assays if >50% of the bead cluster shifted to the right of the negative control discriminator. The relative strength of the alloantibody in the patient's serum was characterized by the determination of the ratio of the mean fluorescent intensity of the bead cluster to the mean fluorescent intensity of the same bead cluster with the negative control serum. More recently, sera demonstrating positive screening assays were further tested with LABScreen class I and/or II single antigen beads on the Luminex platform (One Lambda). The reactivity to individual HLA antigens was considered positive if a median fluorescent intensity value > 1000 was obtained.

The donor specificity of HLA antibodies was confirmed by a flow cytometry XM. A 3-color technique [CD19-phycoerythrin, CD3–peridinin chlorophyll protein, and anti-immunoglobulin G F(ab)2–fluorescein isothiocyanate] was used to detect immunoglobulin G antibodies binding to pronase-treated donor T and/or B lymphocytes. Flow XMs were acquired and analyzed with a FACSCalibur flow cytometer (BD Biosciences). The positive cutoff channel shift for flow XM results was established as follows: the negative control serum was run against 20 donor cells, and then the mean fluorescent intensity values plus or minus 2 standard deviations were determined. In some cases, complement-dependent cytotoxicity (CDC)–based XMs were performed for additional information. The Amos modified method was used for T and B cell CDC XMs along with the anti-human globulin–augmented XM for T lymphocytes. Testing was performed on serum samples collected before and after liver transplantation (Tables 2 and 3) on postoperative day (POD) 2 to postoperative month (POM) 27, with the majority of the testing done in the first month after transplantation.

Table 2. Correlation of IF Findings in Native Liver Tissues and Pos-XM/DSA Patients
  XM IFCourse at Time of Biopsy 
PatientBiopsy TypeT CellB CellDSAs: Classes I and IISinusoids*IELPortal VeinBile DuctTHVNecrotic CellsInterstitiumAMRACRLong-Term Outcome
  • *

    The intensity of staining was recorded as follows: (0) negative, (1+) weak, (2+) moderate, and (3+) strong.

  • Good liver function.

  • Cholestasis.

  • §


  • Although the RAI was reported to be 2 or 3, ACR was considered to be undetermined.

3ControlntntNADiffuse, thick cords (3+)NANANA
7ControlntntNAFocal, moderate, chunky (2+)+NANANA
8BT/PP7689B39(2041); C12(1204)PP: diffuse, linear (2+)PP: +BT: +38 months†
9BT/PPntntDR53 (3271)BT/PP: granular/chunkyPP: +PP: +25 months†
10BT/PP211DR51(15,337); DR53(15,506); DR1(12,999)+PP: +BT: +27 months†
11PP270393A25(1920); B61(16,627); DR11(10,848)+++27 months†
12BT/PP153DQ8(8094); DQ7(7628)+PP: +PP: +30 months†
13BT/PP487479A29(5138); B44(16,476); B62(14,652); C10(7081)++30 months†
14BT/PP177165A2(9422); B8(2758); C10(1839)PP: focal (1+)Weak++34 months†
 11 daysntntDR53(4859)+ 
 21 daysntntDR53(5963); DQ8(3006)++ 
15BT/PPntntB53(5539)BT: focal/granular PP: diffuse (2+)BT: +PP: +26 months†
1617 months378370A2(12,704); A24(9591)Diffuse, linear (1+)50 months‡
17BT/PP79109B35(1806)PP: +BT: +34 months†
18BT/PP96178A23(15,333); B81(4903)++14 months§
19BT/PP473466ntPP: diffuse, linear/granular (2+)++Retransplantation at 13 months
3 days273270ntDiffuse, linear/granular (2+)++
24 days106393Ratios: A30(51); A71(25); B72(59); B35(6)DR11(69); DR11(69); DR15(75); DRw51(61)DQ7(566)Diffuse, linear (2+)++
14 days273270ntDiffuse, granular (2+/3+)++
2 monthsntntRatios: A30(60); A74(19); B72(4); B35(6)DR11(9); DR11(9); DR15(11); DRw51(36)DQ7(132)Diffuse, coarse, granular (3+)++++
4 monthsntntRatios: A30(19); A74(7); B72(4); B35(5)DR11(3); DR11(1.3); DR15(18); DRw51(50)DQ7(200)Diffuse, granular (2+/3+)+++
8 monthsntntRatios: A30(6); A74(2); B72(2); B35(4)DR11(0.8); DR11(2.8); DR15(2); DRw51(8)DQ7(172)Diffuse, granular (2+)++
10 daysntntA68(7993); B44(11,516); B72(10,725)Diffuse, linear (2+/3+)PP: +33 months
2033 days355355A2(1430); A66(13,361); B57(4253); B58(3004)Diffuse, linear (3+)+RAI 337 months
6 monthsntntA2(793); A66(390); B57(124); B58(11)Diffuse, linear (2+)RAI 3
21BT/PP521443B18(3146); B44(12,223); C5(3464); DR13(1276); DR17(1995); DQ6(3084)PP: diffuse, linear (3+)+40 days (failure)§
7 days215348B18(10,025); B44(2940); C5(78); DR13(7184); DR17(8987); DQ6(2040)Diffuse, linear (2+/3+)++
28 daysntntB18(4080); B44(244); C5(69); DR13(6416); DR17(7039); DQ6(9556)Focal, linear/granular (2+)+++
22PP343B negativeB7(12,771)Focal, linear/granular (2+)+112 days (failure)§
40 daysntntRatio: B7(2)Diffuse, strong, linear (3+)+RAI 2
Table 3. Correlation of IF Findings in Neg-XM/DSA Patients
PatientBiopsy TypeXMDSAs: Classes I and IIIFClinical Course
T CellB CellSinusoids*IELPortal VeinBile DuctTHVNecrotic CellsInterstitiumAMRACROutcome
  • *

    The intensity of staining was recorded as follows: (0) negative, (1+) weak, (2+) moderate, and (3+) strong.

23BT/PP80No DSA+ 
3 daysntntNo DSA+ 
4 monthsntntNo DSA+30 months
24PP ++37 months
25PP +38 months
2612 months Died at 19 months (HCC)
271 year 9 months +Died at 46 months (liver failure)
28BT/PP +++ 
6 days +++36 months
29BT/PP PP: focal, weakPP: +PP: + 
70 days +Died at 150 days (PTLD?)
30BT/PP ++33 months
31BT/PP BT: +PP: +36 months
32BT/PP ++PP: + 
4 days ++Died at 5 days (PNF)
33BT/PP +++   
8 days Focal, faint (1+)+RAI 335 months
34BT/PP BT: +   
2 days ++  
2 months +39 months
35PP ++  Died at 12 months (sepsis)
36BT/PP PP: +PP: +24 months
37PP Focal, granular (1+)+24 months
38BT/PP PP: +PP: ++   
39BT/PP ++29 months
40BT/PP BT/PP: focal, chunky
     PP: faint, granularBT: +PP: +PP: +RAI 4/6 × 236 months
41BT/PP BT/PP: focal, thick, strong, cordlike, chunkyPP: +31 months
42BT/PP 32 months
43BT/PP BT: +BT: +29 months
44BT/PP  BT: +29 months
45BT/PP +PP: + 
 3 months +Died at 5 months (sepsis)
46PP +39 months
4726 months ++ 
 27 months ++54 months
48BT/PP +28 months
49BT/PP +29 months
50BT/PP ++27 months
3 days +   
40 days +   
51BT/PP PP: thick chicken wire (2+)PP: ++26 months
52BT/PP PP: +BT: +RAI 4 at 24 months26 months
53PP ++RAI at 24 months24 months
54BT/PP60No DSA+ 35 months
55BT/PP ++Died at 5 days (PNF)
56BT/PP PP: +RAI 5 at 8 days31 months
57BT/PP BT: +PP: +35 months
58BT/PP BT: +++RAI 6 at 8 days31 months
59BT/PP PP: +PP: +34 months
60BT/PP PP: +30 months
61BT/PP PP: +PP: +Died at 18 months (car accident)
62BT/PP PP: ++RAI 4 at 19 months26 months
63BT/PP BT: +BT: +Died at 22 months (HCC)
64BT/PP BT/PP: focal/ granular++29 months
65BT/PP +36 months
42 days ++ 
70 days  
66BT/PP PP: +PP: +25 months
67PP +++25 months
68BT/PP PP: +PP: ++25 months

Statistical Analysis

Differences between nominal variables were determined with Fisher's exact test (GraphPad Software, La Jolla, CA). A 2-sided P value < 0.05 was considered statistically significant.


C4d IF staining was performed for 134 liver allograft biopsy samples and 7 native liver biopsy samples from 68 patients [44 males and 24 females; mean age = 53 years (range = 15-73 years, median = 54 years); Table 4].

Table 4. Results of C4d Staining in All Groups and Tissue Compartments
Tested Patients (n = 68)Number of Biopsy Samples (n = 141)Sinusoids: Diffuse and Linear or Granular (n)Sinusoids: Any Signal (n)IEL (n)Portal Vein (n)Bile Duct (n)Hepatic Vein (n)Necrotic Cells (n)Interstitium (n)
Control native livers (n = 7)70725340707070716
Controls BT (n = 47)4704764124230470704712352027
Pos-XM/DSA (n = 15)PP: 135876850130130138558
FU: 1513213278015015015015510
Neg-XM/DSA (n = 46)PP: 42042636192304204204226162121
FU: 1701711698017017116314710
Fisher's TestSinusoids: Diffuse and Linear or Granular (P)Sinusoids: Any Signal (P)IEL (P)Portal Vein (P)Bile Duct (P)Hepatic Vein (P)Necrotic Cells (P)Interstitium (P)
PP pos-XM/DSA versus neg-XM/DSA<0.001<0.01NSNSNSNSNSNS
FU pos-XM/DSA versus neg-XM/DSA<0.001<0.001NSNSNSNSNSNS
Control versus PP + FU pos-XM/DSA<0.001<0.001NSNSNSNSNSNS
Control versus PP + FU neg-XM/DSANSNSNSNSNSNS<0.01NS

Positive-XM and/or DSA Recipients

From 15 patients with a positive pretransplant XM, we collected 39 biopsy specimens; in 11 cases, we collected both BT and PP samples from the same allograft (Tables 2 and 4). Two specimens were collected PP only. FU specimens (n = 15) were collected from POD 3 to POM 17. We collected FU biopsy samples from 6 patients. Three patients had a complete panel of BT and PP samples (including FU samples; Table 2).

Negative-XM and/or DSA Recipients

Ninety-six biopsy samples were collected from 46 liver allograft recipients in this group (36 BT samples, 43 PP samples, and 17 FU samples). Both BT and PP biopsy samples were collected from 36 patients; from 9 patients, all 3 types of biopsy samples were obtained (Table 3).

C4d IF Staining of Various Compartments

We evaluated the characteristics of C4d staining in several liver tissue compartments. Portal veins and bile ducts never displayed positive C4d IF in specimens collected from patients, regardless of the XM results or DSA testing results. We observed IF staining in a hepatic vein in only 1 specimen collected during FU biopsy from a recipient without HLA antibodies. We also identified staining of necrotic hepatocytes (Fig. 3) and the interstitium of allografts in both groups: the positive crossmatch and detectable donor-specific alloantibody (pos-XM/DSA) group and the negative crossmatch and detectable donor-specific alloantibody (neg-XM/DSA) group. There were no statistical differences between the C4d IF staining in the aforementioned tissue compartments of the 2 groups of allograft recipients (including BT, PP, and FU subgroups; Table 4). There were no observable qualitative or quantitative differences in C4d staining in the aforementioned compartments. However, in liver sinusoids, a variety of C4d IF staining patterns were observed. The sinusoids displayed diffuse and linear staining, focal and granular staining, and thick cords.

Figure 3.

Necrotic hepatocytes.

The IEL of the liver arteries was positively stained by IF (Fig. 4) in many BT, PP, and FU biopsy specimens of allografts in the pos-XM/DSA and neg-XM/DSA groups. In all, 15 of 28 specimens from the pos-XM/DSA group had positive C4d IF staining of the IEL, and 28 of 51 cases were positive in the neg-XM/DSA group. The specific numbers for each subgroup are indicated in Table 4. The differences between groups were not significant (NS) with respect to IEL staining (Table 4).

Figure 4.

Arterial IEL.

IF Allograft PP Study (Tables 2-4)

Seven of 13 pos-XM/DSA patients had any sinusoidal IF signal in PP biopsy samples, and 5 had diffuse, linear or granular staining. Six neg-XM/DSA patients had any IF signal, and none of the 42 specimens had diffuse, linear or granular sinusoidal patterns of C4d deposits. This difference was significant between the pos-XM/DSA and neg-XM/DSA groups (P < 0.01; Table 4). C4d staining for necrotic hepatocytes was greater in the reperfused allografts (PP) versus the native livers, and this most likely reflected ischemia/reperfusion injury. The greater proportion of reperfused allografts in the neg-XM/DSA group versus the native liver group led to the greater proportion of C4d staining.

IF Allograft FU Biopsy Study (Tables 2-4)

Fifteen of the 134 liver allograft biopsy samples were obtained from patients with pos-XM/DSA findings in the postoperative FU period. Thirteen samples had IF staining, and all 13 had a diffuse and linear pattern (Fig. 1A) or a granular pattern in sinusoids (Fig. 2). Only 1 of the 17 biopsy samples with neg-XM/DSA findings had any positive signal, and that sample did not stain in a diffuse, linear or granular IF pattern. The differences between the pos-XM/DSA and neg-XM/DSA groups were significant (P < 0.001).

The combined results for all specimens (BT, PP, and FU) are presented in Table 4. All differences are statistically significant (P < 0.001).

IF Control Study

IF Allograft BT Study (Tables 2-4)

We collected 47 BT specimens from the donor livers of the pos-XM/DSA and neg-XM/DSA groups. A positive signal in the sinusoids was detected in only 2 of the 11 patients in the pos-XM/DSA group. For the neg-XM/DSA group, only 4 of the 36 biopsy samples showed any positive sinusoidal signal. A diffuse, linear or granular pattern was not observed in either group.

Additional Negative Controls (Tables 2 and 4)

Tissue was obtained from 7 nontransplant patients who had undergone partial liver resection for the removal of benign lesions. No specimen showed linear or granular sinusoidal C4d staining. In 2 control patients, diffuse, thick, and cordlike (Fig. 5) or focal and chunky sinusoidal C4d deposits (Fig. 6) were seen, and they differed from the more linear deposits seen in allografts (Fig. 1A). None were stained with a diffuse, linear or granular pattern on IF. Interstitial staining was seen in only 1 specimen. The IEL was stained positively in 3 cases. Other liver compartments were also negative for C4d according to IF.

Figure 5.

C4d IF staining in liver sinusoids: thick cords.

Figure 6.

C4d IF staining in liver sinusoids: focal and chunky.

Sinusoidal C4d IF staining was not statistically different for BT and native liver specimens, nor was C4d IF staining in other liver compartments (including the interstitium).

C4d IHC Staining Results

C4d IHC was performed for 19 specimens collected from 15 liver transplant patients for whom a sinusoidal pattern of C4d staining was noted by IF in a previous evaluation. Four of these patients had neg-XM/DSA findings. Sinusoidal C4d IF was diffuse in 12 biopsy samples from 8 patients and was focal in samples from 7 patients. The neg-XM/DSA patients demonstrated focal and weak staining intensity mostly. At our institution, C4d IHC findings were negative in the sinusoids for 17 of 19 biopsy samples. In patients 2 and 15 (Table 5), the IHC sinusoidal staining intensity was weaker, and the pattern changed from diffuse positivity to focal positivity in comparison with the corresponding IF results. Positive IHC staining was observed for the portal vein endothelium in 11 of 19 specimens from 10 of 15 patients; 2 patients had neg-XM/DSA findings. Six of the biopsy samples were moderately stained for 5 patients, and the remaining 5 samples were weakly stained. Only in 1 case (case 4, Table 5) was C4d staining positive with both IF and IHC methods, and this was the only patient who had a portal vein endothelium that was positive according to the IF technique. In the remainder of the cases, the 2 methods did not correlate.

Table 5. Summary of Comparative IF and IHC Staining Results
CaseStainPODSinusoidsCentral VeinPortal FieldsNecrotic HepatocytesXM and/ or DSA
1IF33Diffuse, strong, linear00000NAPositive
IHC 000Moderate00NA 
IF180Diffuse, moderate, linear00000NA 
IHC 0 0Moderate00NA 
2IF14Diffuse, moderate, linear0Positive0Positive0NAPositive
IHC 00Positive0Positive0NA 
IF24Diffuse, moderate, linear00000NA 
IHC Focal, faintModerate0ModeratePositive0NA 
IF3Diffuse, strong, linear/granular00000NA 
IHC 000000NA 
IF60Diffuse, strong, linear0Positive00PositiveNA 
IHC 000000NA 
3IF0Diffuse, faint0Positive00PositivePositiveNegative
IHC 0Weak0Weak00NA 
4IF0Focal, weak, linear00Weak00NAPositive
IHC 000Weak00NA 
5IF13Diffuse, moderate, granular0000PositiveNAPositive
IHC 000Moderate00NA 
6IF8Focal, weak, linear00000NANegative
IHC 000000NA 
7IF0Focal, moderate, granular00000NAPositive
IHC 000Weak00NA 
8IF0Diffuse, weak granular0000PositivePositivePositive
IHC 000000NA 
9IF0Focal, weak granular0000PositiveNANegative
IHC 000000NA 
10IF0Focal, weak granular0000PositivePositivePositive
IHC 000000NA 
11IF515Diffuse, weak, linear0Positive000NAPositive
IHC 000Weak00NA 
12IF12Focal, weak, linear0000PositiveNANegative
IHC 000Weak00NA 
13IF0Focal, faint0Positive00PositivePositivePositive
IHC 000000NA 
14IF0Diffuse, strong, linear0Positive000NAPositive
IHC 000Moderate00NA 
15IF40Diffuse, strong, linear00000NAPositive
IHC Focal, moderate, linearModerate0Moderate00NA 

Evaluation of Interlaboratory C4d IHC Staining Variability

Unstained formalin-fixed, paraffin-embedded liver tissue slides from 7 patients were sent to UHB for C4d staining by IHC. Five were C4d-positive by IF at UNC. All 7 samples had also been stained at UNC with IF and IHC techniques. IF and IHC were compared at UNC, and 2 different IHC protocols were compared at UHB. At UHB, 6 of 6 IHC-stained sections were negative with the standard staining protocol (including all 5 with C4d staining by IF). The modification of the standard staining protocol to include an EDTA/Tris buffer instead of a citrate buffer increased the detection of C4d deposits along sinusoids in 4 of 4 cases; however, the staining intensity was much weaker and was very focal (<5% of sinusoids) in all 4 cases in comparison with the IF staining results (Table 6 and Fig. 1A-C).

Table 6. Interlaboratory Variability: Selected Cases From Table 5
  IHC Protocols
CaseIF Protocol: UNCUNCFirstSecond
2Diffuse, strong, linear/granular00Focal, faint
5Diffuse, moderate, granular00Focal, weak, linear/granular
4Focal, weak, linear00Focal, faint
13Focal, faint00NA
14Diffuse, strong, linear00Focal, moderate, linear/granular

H&E Staining

Biopsy samples for histological and C4d IF evaluations were collected at the same time. The timing of each collection is specified in Tables 2 and 3. H&E-stained biopsy samples were evaluated for features of AMR or chronic injury (inflammation), and the findings were correlated with the IF results for C4d and the XM/DSA results (Table 7). Scores ≥ 2 for sinusoidal neutrophil deposits, cholestasis, hepatocyte ballooning, and cholangiolar proliferation were found for 18% to 36% of the specimens positive for XM/DSA/C4d IF staining. For patients with specimens negative for XM/DSA/C4d IF staining, corresponding histological AMR features were seen in only 0% to 10% of biopsy samples. A comparison of the deposits of sinusoidal neutrophils with scores ≥ 2 in the group positive for XM/DSA/C4d by IF and in the group negative for these features indicated a correlation of sinusoiditis with C4d IF deposits (P < 0.01). Indication biopsy was performed for 6 patients with pos-XM/DSA findings (Table 2); 3 of these patients later developed AMR and progressed to liver failure. In 9 of the 10 biopsy samples for these 3 patients from POD 3 to POM 4 (patients 19, 21, and 22; Table 2), we observed sinusoiditis ranging from grade 1 to grade 3 with a mean grade of 1.9 ± 0.9 and a median grade of 2. The rest of the pos-XM/DSA patients without clinical evidence of AMR and the patients in the neg-XM/DSA group did not show neutrophil deposits in sinusoids according to their indication biopsy samples (Table 3).

Table 7. H&E Staining in XM/C4d-Positive or -Negative Patients and Tissues
XM/DSAIF: C4dSamples (n)GradeAMR–Ischemia/ReperfusionInflammation
Sinusoidal NeutrophilsCholestasisBallooningCholangiolar ProliferationPortalLobularVenous SubendothelialBile DuctArteritis

Chronic changes with periportal, lobular, venous subendothelial, and bile duct inflammation (score ≥ 2) were seen in 0% to 15% of the samples from the group positive for XM/DSA/C4d IF and in 0% to 7% of the samples from the group negative for the respective features (Table 7).

Clinical Course

Three patients with a positive XM had DSAs and H&E-stained liver injuries characteristic of AMR after transplantation, were nonresponsive to desensitization therapy, and subsequently developed liver failure and were diagnosed with clinical AMR. Two died from complications of graft dysfunction and infection, and 1 required retransplantation and is now doing well. These 3 patients demonstrated positive diffuse and linear sinusoidal endothelial IF staining. Details concerning the patient outcomes and FU duration are listed in Tables 2 and 3.


In this study, we focused on the demonstration of C4d on the liver graft endothelium as an indirect sign of the interaction of preformed anti-HLA antibodies with the graft and resultant local complement activation.

Details of clinical courses and outcomes, results of AMR treatment, and criteria necessary for establishing AMR in ABO-compatible liver allotransplantation were published by our group previously.3 The presence of preformed antibodies alone does not often result in a clinical course of AMR.3 The activation of complement and deposits of C4d infrequently leads to liver dysfunction and rejection, despite antibody-antigen-complement interactions. Here we narrowed our analysis to describe the localization of C4d deposits in liver allografts in the presence of preformed anti-HLA antibodies and positive XMs.

Liver sinusoids parallel the capillary compartments in other organs in which the blood flow is slowed and is no longer pulsatile (with resulting pressures in the range of 2-5 mm Hg).33 We can only speculate that these conditions may foster the interaction of anti-HLA alloantibodies with endothelial cell antigens in liver sinusoids and allow deposits of complement from the blood stream similar to C4d deposits seen in peritubular capillaries of kidney allografts with AMR. We clearly found a correlation between the presence of HLA antibodies with donor specificity in ABO-compatible liver allograft recipients and the presence of C4d deposits detected by the IF technique in the sinusoids.

We detected other nonlinear sinusoidal staining patterns in control specimens that did not appear specific (eg, the cordlike staining pattern). We can only speculate that portal vein blood with its possible exposure to pathogens and toxins draining from the intestine may contribute to the activation of complement and nonspecific binding downstream within the liver. Another speculation involves the peculiar blood supply to the hepatic sinusoids, which receive blood from both hepatic artery and portal vein components. These components may not mix in the sinusoids but instead may drain to separate sections of the sinusoidal reservoir. This sinusoidal vascular supply is regulated by the oxygen supply to the liver33 and may theoretically affect the distribution of anti-HLA antibodies to the sinusoidal endothelium and result in nonuniformly focal and less linear C4d deposition in cases of native liver samples (our control group).

C4d IF also highlighted the IEL in a significant number of liver allograft tissue specimens in all groups of specimens. This nonspecific finding is also common in kidney allografts, in which C4d staining of the arterial IEL is considered a nonspecific IF event and serves as an internal positive control. On the basis of our findings, arterial IF staining in frozen tissue cannot be considered specific for AMR in the transplantation of ABO-compatible liver allografts.

In a kidney transplant study involving a pretransplant desensitization plasmapheresis protocol for a positive XM, specimens collected within an hour after perfusion did not show evidence of C4d deposits in a majority of the cases (frozen specimens were tested and evaluated with IF). In our liver study, PP biopsy samples were collected several hours after reperfusion near the end of the surgical case. This is much longer after reperfusion in comparison with kidney transplant surgery. Also, in contrast to the renal transplant study reported by Haas et al.,34 none of our recipients were treated with pretransplant plasmapheresis to reduce levels of HLA alloantibodies. Nevertheless, in this kidney study, C4d IF deposits were seen as early as an hour after perfusion in a few cases.

Although IF is the gold standard for the detection of C4d deposits, most investigators who have characterized C4d staining in liver allografts have done so on formalin-fixed tissue with IHC. IHC, as it is performed at UNC and UHB, appears to be difficult to interpret reliably because the staining appears weak. Our study comparing IF on frozen tissue and IHC on formalin-fixed, paraffin-embedded tissue for the detection of sinusoidal C4d deposits has revealed that the IF technique is more sensitive. Only 11% of the cases with C4d positivity by IF showed any sinusoidal staining by IHC. In the 2 cases with sinusoidal staining for C4d by IHC, the intensity of the staining was weaker and very focal in comparison with the corresponding IF result. Seemayer et al.31 reported similar findings from a comparative study of C4d IF and IHC in renal allograft tissue. In clinical practice, the interpretation of focal and faint sinusoidal staining would undoubtedly be challenging and subject to increased intraobserver variability. Other groups have made an effort to determine the significance of C4d deposits in liver tissue as well. Bellamy et al.21 reported 12 patients with positive cytotoxic XMs. Because of the way in which the report is constructed, we can only assume that T lymphocytotoxicity tests were performed BT but not necessarily at the time of biopsy. There is evidence that many positive XMs of liver allograft recipients convert into negative XMs, and DSAs are eliminated within several days of transplantation when alloantibodies are tested with more sensitive methods such as T/B cell flow cytometry and solid phase testing (single antigen beads).3 In Bellamy et al.'s report, there was a case highly suspicious for accelerated humoral rejection with IHC positivity for C4d in sinusoids, but the patient's XM was not determined. Further confusion about C4d deposits in sinusoids and/or periportal areas is created by the fact that some of the cases in Bellamy et al.'s work were diagnosed with ACR and/or CR, and a cellular component was prominent in periportal areas. These cells are also capable of producing complement. This was discussed at length previously3 and by Bellamy et al. An appropriate correlation of IHC findings and DSAs was applied in work by Musat et al.28 These authors focused their attention on ACR, with the majority of the biopsies performed 253 to 3550 days after transplantation. They described mostly mild/moderate ACR and ductopenia; therefore, this publication was more focused on CR rather than AMR. In addition, even in cases evaluated early after transplantation, the information on the presensitization status of liver transplant candidates was missing, and control tissue from nontransplant livers or liver allografts before exposure to recipients' blood was missing. This group reported that C4d is deposited by IHC mostly but not only in periportal areas. The C4d localization correlated with the cellular components of the host immune response concentrated in the periportal regions of the liver allograft tissue.

When IF was used, no correlation was found between C4d deposits in the portal stroma, IEL, and necrotic hepatocytes and the presence of circulating DSAs and/or a positive XM in liver transplant recipients. We interpret these nonsinusoidal IF staining patterns as not characteristic for AMR. Although C4d deposits were noted in the portal vessel and central vein endothelia by IHC, this was rarely observed with the IF technique. A review of C4d IF staining in BT biopsy samples showed an absence of significant sinusoidal C4d staining, and this supports the concept that the presence of C4d deposits in the sinusoids is a result of graft and recipient interactions and is not a result of an underlying donor disease, organ procurement, or the preservation process. Our results did show a weak or moderate portal vein IHC staining pattern in several biopsy samples with strongly positive sinusoidal IF staining, which previous authors have reported with the IHC technique in liver allografts11, 14-16, 20, 21, 25, 35, 36 (Table 5). The pattern does suggest to us that portal vein endothelium IHC staining is not as sensitive as sinusoidal IF findings.

The number of collected FU samples was smaller than the number of BT and PP samples. We performed FU biopsy only for clinical indications. However, even with this small number of FU biopsy samples, the analysis indicated statistical significance for our findings (Table 4).

Chunky or thick cords of sinusoidal C4d deposits were seen in 2 nontransplant negative control tissue samples and in 5 BT specimens and 2 PP specimens from the transplant groups with pretransplant pos-XM/DSA or neg-XM/DSA findings (Tables 2 and 3). C4d deposits would not be expected in either of these clinical scenarios, and these 2 patterns were not observed in patients with DSAs and clinical evidence of AMR.3 The underlying etiology of this nonspecific pattern of staining has yet to be elucidated.

The results of our interlaboratory staining findings raise concerns about the variability in the quality and sensitivity of C4d IHC staining from laboratory to laboratory. Many variables, including the length of tissue fixation, the titer of the primary antibody, the antigen retrieval method, the buffers, and other technical variables, affect the quality of IHC staining.27 Our findings may explain why there is no consensus on C4d findings in studies published to date. Recently, Aguilera et al.29 described details of C4d IHC staining of formalin-fixed, paraffin-embedded biopsy samples of liver tissue with evidence of complement deposits on the sinusoidal endothelium in patients diagnosed with CR. This study was performed as an investigation of de novo immune hepatitis developing in liver allografts. The authors did not provide information concerning the presence of DSAs of an HLA origin or correlations with the IF staining technique. Until a C4d IHC staining protocol can be validated through a comparison with C4d IF results, the IF technique should be routinely used for the detection of sinusoidal endothelial C4d deposits.

The histology (H&E) of AMR has been well described in the past.32 Interestingly, we found a correlation between C4d IF staining in transplant patients with pos-XM/DSA findings and sinusoidal neutrophil deposits with a grade ≥ 2. We believe that this finding may correlate with similar findings in the field of kidney transplantation, in which peritubular capillaritis is seen in kidney AMR. In order to be confident that these findings parallel those for kidney AMR, more work needs to be done.

Although there is much that the liver transplant community can apply from what is already understood about AMR in other allograft tissues, future prospective studies are needed to correlate C4d IF findings in both PP and FU biopsy samples with corresponding measurements of circulating DSAs and the clinical course/graft function. Because AMR is a dynamic process, the results of C4d staining and the antibody status should be followed over time with the integration of clinical findings, long-term outcomes, histological features, and effects of therapeutic interventions. It appears that IHC on fresh-frozen or formalin-fixed tissue for the detection of sinusoidal C4d deposits should not be exclusively used for C4d diagnostic purposes until validation and standardization with the IF technique on fresh-frozen specimens are adequate. Because pretransplant positive XMs are relatively rare occurrences, well-designed multicenter, prospective studies using a detailed C4d staining methodology for both IF on frozen tissue and IHC on frozen and fixed tissues will be necessary to clarify C4d staining for AMR in liver transplantation. We need further IHC studies to select IHC protocols that better correlate with clinical, histological, and C4d IF findings. We likely will need to first perform IHC in parallel studies on fresh tissue along with IF and then determine whether this IHC method is appropriate with paraffin-embedded tissue for providing reproducible results.


The authors are thankful to Professor M. J. Mihatsch and his colleagues from the Department of Pathology of the University Hospital Basel (Basel, Switzerland) for their generous support of parts of the immunohistochemistry study. They also thank Dr. Tara Rubinas and Dr. John Woosley from the Department of Pathology of the University of North Carolina at Chapel Hill for their expert support.