• Allograft monitoring;
  • antibody-mediated rejection;
  • complement C4d;
  • crossmatching;
  • liver allograft


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

C4d-assisted recognition of antibody-mediated rejection (AMR) in formalin-fixed paraffin-embedded tissues (FFPE) from donor-specific antibody-positive (DSA+) renal allograft recipients prompted study of DSA+ liver allograft recipients as measured by lymphocytotoxic crossmatch (XM) and/or Luminex. XM results did not influence patient or allograft survival, or cellular rejection rates, but XM+ recipients received significantly more prophylactic steroids. Endothelial C4d staining strongly correlates with XM+ (<3 weeks posttransplantation) and DSA+ status and cellular rejection, but not with worse Banff grading or treatment response. Diffuse C4d staining, XM+, DSA+ and ABO– incompatibility status, histopathology and clinical–serologic profile helped establish an isolated AMR diagnosis in 5 of 100 (5%) XM+ and one ABO-incompatible, recipients. C4d staining later after transplantation was associated with rejection and nonrejection-related causes of allograft dysfunction in DSA– and DSA+ recipients, some of whom had good outcomes without additional therapy. Liver allograft FFPE C4d staining: (a) can help classify liver allograft dysfunction; (b) substantiates antibody contribution to rejection; (c) probably represents nonalloantibody insults and/or complete absorption in DSA– recipients and (d) alone, is an imperfect AMR marker needing correlation with routine histopathology, clinical and serologic profiles. Further study in late biopsies and other tissue markers of liver AMR with simultaneous DSA measurements are needed.


acute cellular rejection


anti-human globulin


autoimmune hepatitis


alkaline phosphatase


alanine transaminase


antibody-mediated rejection


aspartate transaminase


complement-dependent cytotoxicity


Cox Proportional Hazard


chronic rejection


donor-specific antibodies


formalin-fixed paraffin-embedded


gamma glutamyl transpeptidase


hepatitis B


hepatitis C


panel-reactive antibody


primary sclerosing cholangitis


posttransplant lymphoproliferative disorder


rejection activity index


total bilirubin




XM positive


XM negative


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

Antibody-mediated rejection (AMR) can contribute to severe acute and chronic liver allograft injury and failure in humans (1,2). This was first conclusively shown in experimental animals (3) and then in humans when crossing ABO blood-group barriers (1) followed by those harboring preformed lymphocytotoxic antibodies (2). As in other solid organ allografts, microvascular endothelial cell antibody binding, triggers complement fixation followed by intrahepatic platelet-fibrin thrombi and leukocyte margination (2,4). The most severe form of acute injury culminates in consumptive coagulopathy and thrombocytopenia, microvascular thrombosis and arterial vasospasm, recognized histopathologically as platelet aggregates in the portal and/or central veins (CVs), neutrophilic exudation and rarely lymphocytic arteritis (2,4). Microvascular injury, in turn, can cause patchy hepatocyte necrosis and centrilobular swelling, cholangiolar proliferation and cholestasis. AMR is often accompanied or quickly followed by acute cellular rejection (ACR), and it rarely leads to chronic rejection (CR; Refs. 2,5–7) because immunoglobulin and complement binding to the peribiliary vascular plexus likely contributes to biliary damage (8).

Compared to other solid organ allografts, however, a diagnosis of isolated liver allograft AMR is difficult to establish with certainty: It requires careful biopsy review, clinicopathologic correlation and exclusion of other insults (1,2). Difficulties are encountered because (1) liver allografts are large, able to absorb high antibody loads and resistant to AMR-related damage (2,4,6,8,9); (2) immune deposits pointing toward an underlying injury cause are (a) ephemeral (1,2), (b) can be associated with other insults (see later) and (c) more easily detected in frozen sections (10) and (3) clinicopathologic similarities exist between AMR and preservation injury, sepsis and biliary/vascular complications (1,2).

Recognition that C4d persists for several days, can be detected in formalin-fixed paraffin-embedded (FFPE) tissues, and correlates with circulating donor-specific antibodies (DSA), made possible the recognition of AMR in kidney allografts (11,12). Interpretation and practical utility of C4d staining in liver allografts, however, is less clear than in other allografts (13–21).

Bellamy recently reviewed the literature on C4d immunohistochemistry in liver allografts (13–21) concluding that the practical utility of C4d is limited, but might identify a small subgroup of individuals with chronic humoral microvascular injury (13). Normal livers and liver allograft biopsies are usually C4d–. Portal venous, arterial and portal capillary, and sinusoidal endothelial C4d has been detected in crossmatch (XM)-positive (XM+) recipients more often than XM-negative (XM–) controls (19) and in those who developed isolated AMR (10,21). Endothelial cell C4d staining might be most specific for AMR, but “portal stromal” C4d staining has also been described in ABO-incompatible AMR (16), ACR (19) and CR (7,22).

C4d deposits are also often accompanied by ACR (7,13–21) and in some studies are directly proportional to the Banff grade (13–21). Necrotic hepatocytes can also show nonspecific C4d staining. Sakashita et al. (19) concluded that C4d staining was more common and strong in XM+ recipients, but focal and weak staining was not diagnostically useful.

Portal microvascular C4d deposits can also be detected when other insults, including biliary obstruction (13), recurrent hepatitis B (HBV; Ref. 14) and recurrent hepatitis C (HCV; Ref. 17), are thought to be the cause of allograft dysfunction. C4d staining has also been described in portal venous and capillary; sinusoidal, CV and arterial endothelial cells; in lymphoid nodules; and in periductal and portal stromal cells in native pediatric livers with HBV, HCV, autoimmune hepatitis (AIH) and overlap syndromes between AIH and primary sclerosing cholangitis (PSC; Ref. 23). Endothelial C4d deposits in nonrejection-related allograft disorders are reportedly less widespread than in severe ACR or AMR. In one study, diffuse sinusoidal C4d, in addition to portal vascular staining, distinguished moderate-to-severe ACR from other complications (20). Similar to kidney and heart allografts, liver C4d deposits have also been associated with macrophage and plasma cell infiltrates (15).

We previously showed that XM+ liver allograft recipients can develop antibody-mediated abnormalities shortly after transplantation, including thrombocytopenia, hypocomplementemia, hepatocyte swelling, hepatocanalicular cholestasis and cholangiolar proliferation often accompanied by ACR (2,24–26), which was related to XM titer. This study revisited the significance of a positive lymphocytotoxic XM and determined whether C4d staining could help better monitor and classify such patients by improving recognition of AMR-related syndromes.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

Patient population and antibody testing

Primary liver allograft recipients (n = 809) at the University of Pittsburgh Medical Center, between October 23, 2003 and June 30, 2009 (Table 1), were divided into two groups on the basis of anti-human globulin (AHG) T-cell complement-dependent cytotoxicity (CDC) XM (XM+: 100/809; including 84 XM– and 11 XM+ living donor recipients). In 62 of 809 of the recipients, DSA was measured by Luminex single antigen beads in serum samples paired within 2 weeks of late (>21 days) liver allograft biopsies (see section “Supplemental Methods” of Supporting Information for XM and DSA testing details).

Table 1.  General demographic data for the XM+ and XM– recipients
Variablesn = 709%n = 100%n = 809%p-Value
  1. XM–= crossmatch negative; XM+= crossmatch positive; bx = biopsies.

Recipient gender      0.0001
Donor gender      0.15
Recipient age (yrs)      0.19
 Mean (SD)54.3 (11.8)55.8 (11.5)54.5 (11.7) 
Donor age (yrs)      0.92
 Mean (SD)46.5 (18.7)46.4 (21.6)46.5 (19.1) 
Cold ischemia time (min)      0.14
 Mean (SD)569.1 (219.4)544.8 (218.2)566 (219) 
Total number of bx posttransplantation      0.44
 Mean (SD)2.3 (1.7)2.2 (1.7)2.3 (1.7) 

Clinical, histopathological and laboratory data were obtained from the Electronic Data Interface for Transplantation (EDIT) database according to University of Pittsburgh Institutional Review Board (IRB) protocols, PRO10110160 and PRO10110393. Tacrolimus monotherapy, without induction therapy or steroids, was initiated immediately posttransplantation in the vast majority of patients, with levels maintained around 8–10 ng/mL. Patients with renal impairment (on renal replacement therapy, Creatinine >2.0 or urine output <30 cc/h) received Basilixumab (20 mg IV) on postoperative day (POD) 1 and 5, mycophenolate mofetil (1000 mg bid) and prednisone (20 mg daily) on POD 1 and tacrolimus was started on POD 7. Subjects with tacrolimus neurotoxicity (27) were converted to cyclosporine. Alemtuzumab (CamPath, 30 mg, Genzyme) induction was utilized in 47 patients and rabbit antithymocyte globulin (Thymoglobulin, ∼5 mg/kg) in 56 patients (Table 2; Ref. 28). Both antibody-induction regimens included prophylactic 1 gm intravenous dexamethasone as coverage against a cytokine release syndrome and followed by tacrolimus monotherapy, as described earlier. XM+ recipients also received an additional steroid recycle (200, 160, 120, 80, 40 and 20 mg) and maintained on 20 mg/day of Prednisone for the first several weeks after transplantation.

Table 2.  Average medication exposure during the first 21 days after transplantation
  1. XM–= crossmatch negative; XM+= crossmatch positive. n = number of patient in each group; *values for each medication represents area under the curve (AUC) for exposure to that specific medication during the first 21 days after transplantation.

Sirolimus*n = 18n = 4N = 220.64
 Mean (SD)255.3 (191.4)291.4 (59.4)261.9 (174.2) 
Tacrolimus*n = 617n = 84N = 701  1
 Mean (SD)3.2 (1.9)3.4 (2.4)3.3 (1.9) 
Cyclosporine*n = 63n = 9N = 720.62
 Mean (SD)133.1 (97.5)105.7 (70.2)129.7 (94.5) 
Prednisone*n = 50n = 37N = 870.03
 Mean (SD)8.7 (6.2)11.5 (6.9)9.9 (6.6) 
Solu-Medrol bolus*n = 212n = 57N = 2690.02
 Mean (SD)1047.6 (657)831.2 (471.8)1001.8 (627.8) 
Alemtuzumab (CamPath)      0.18
Thymoglobulin       0.4
 No66293.4%9191.0% 753 93.1% 

Pathology studies

All posttransplant liver allograft biopsies (n = 1170), obtained on indication or by protocol at 1-year posttransplant (unless the patient refused) from 668 patients (range 1–12 per patient; mean 2.3), were divided into three time periods: early (1–21 days; n = 324), intermediate (22–365 days; n = 502) and late (>1 year; n = 344). Prospectively scored histopathological features are detailed in the “Supplemental Methods” section of Supporting Information.

Immunohistochemistry was retrospectively performed on FFPE tissue sections following standard protocols with rabbit anti-human C4d polyclonal antibody (Alpco Diagnostics, Windham, NH, USA). C4d-stained liver biopsy samples were divided into two groups for correlation with: (1) XM (n = 230; XM+: 34; XM–: 196) and (2) DSA (n = 67; DSA+: 17; DSA–: 50). C4d staining was scored on two different occasions using two different systems: (1) separately scored portal tract (PT) stroma and portal endothelial compartments (hepatic artery, portal vein and capillary), sinusoids and CVs, as negative, minimal (<10% staining of the overall sampled tissue), focal (10–50%) and diffuse (>50%); (2) minimal/negative: portal tract staining in an occasional compartment; focal: distinct endothelial cell staining of at least two portal compartments (portal vein, hepatic artery and portal capillary) in a minority of PTs; diffuse: as above for focal in a majority of PTs.

Statistical analysis

Descriptive statistics (mean, median and standard deviation) and comparisons between XM+/– and C4d+/– groups were made using chi-square, Student's t-test or Wilcoxon rank-sum tests. Exact methods were applied where appropriate. Analyses were performed using SAS version 9.2 (SAS Institute Inc., Cary, NC, USA). Further statistical analysis descriptions are in the “Supplemental Methods” section of Supporting Information.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

Influence of XM on survival, laboratory values, histopathology findings and diagnoses

After adjusting for gender and original disease, because most XM+ patients are female and many suffered from autoimmune disorders, allograft (p = 0.52) and patient (p = 0.42) survival were similar for XM– and XM+ recipients, but the latter were slightly lower (Figure 1). XM+ patients, however, received more prednisone and Solu-Medrol boluses early after transplantation (Table 2) based on our previous observation that AMR has the potential to injure XM+ liver allografts (2,29).


Figure 1. Allograft and patient survival adjusted for gender and pretransplant disease diagnosis in XM+ and XM– liver allograft recipients.

Download figure to PowerPoint

Causes of allograft failure did not differ between XM– (n = 125) and XM+ (n = 17) groups (p = 0.7): patient death from sepsis or infection (17 vs. 3), recurrent original disease, including malignancy (35 vs. 2), vascular accident or ischemia (8 vs. 3), primary dysfunction (17 vs. 2), rejection (7 vs. 1) and others (unknown cause, technical complications, posttransplant lymphoproliferative disorder [PTLD]; 58 vs. 6). One of eleven XM+ living donor allografts failed; this was attributed to patient death because of sepsis and cardiovascular arrest, but the allograft never functioned properly.

Liver function tests were examined during two phases: early, <3 weeks posttransplant and late, >1 year posttransplant. There were no differences between XM+ and XM– recipients for alkaline phosphatase (ALKP), gamma glutamyl transpeptidase (GGTP) and total bilirubin (TB) during the first 3 weeks after transplantation, but XM+ recipients experienced a more significant drop in platelet counts, as seen before (Refs. 2,4; Figure 2). XM+ recipients, however, showed significantly lower aspartate transaminase (AST) and alanine transaminase (ALT) levels during the first 3 weeks after transplantation (Figure 2), which is probably attributable to the “prophylactic” increase in immunosuppression and slightly shorter cold ischemic time. More than 1 year after transplantation, however, XM+ patients showed significantly higher ALT, GGTP and ALKP (Table S1), suggestive of biliary injury.


Figure 2. Comparison of liver injury test results and peripheral blood platelet counts from XM+ and XM– liver allograft recipients during the first 3 weeks after transplantation. The p values represent the comparison over the entire 3-week time period.

Download figure to PowerPoint

There were no significant differences in biopsy findings from XM+ versus XM– recipients within the first year after transplantation except for increased perivenular fibrosis in XM+ patients (p = 0.02), appearing between 22 and 365 days. Significant histopathological findings observed in XM+ recipients late (>1 year) after transplantation included: bile duct inflammation and damage (p = 0.01); central necrosis (p = 0.02); interface necro-inflammatory activity (p = 0.006) and rejection activity index (RAI) scores for: portal inflammation (p = 0.02), bile duct damage (p = 0.003) and venous subendothelial inflammation (p = 0.006). Differences in these prospectively scored histopathologic findings independently substantiate the differences in liver injury tests >1 year after transplantation (Table S1). XM+ recipients experienced a higher rate of either ACR or CR (33% vs. 25%), but the difference did not reach statistical significance (p = 0.2). Discrepancy between histopathologic parameters and diagnoses of ACR was attributable to scoring of histopathologic parameters that were felt to be rejection-related, but below the threshold, needed for a definitive diagnosis.

C4d staining in liver biopsies

We next examined C4d staining in biopsies from XM+ and XM– recipients to determine if potential intragraft antibody deposition correlates with XM status, histopathologic findings and clinical diagnoses. Biopsies were grouped and analyzed according to time after transplant (early ≤21 days and late >21 days).

Focal or diffuse C4d staining of hepatic artery, portal vein, portal capillary or sinusoidal endothelium was detected significantly more often in early biopsies from XM+ than XM– recipients within the first 3 weeks after transplantation (Table 3). Arterial scoring was based on endothelial, and not on internal elastic lamina or mural, staining. The total C4d score, calculated by adding the staining score for each compartment (negative = 0, minimal = 1, focal = 2 and diffuse = 3) strongly correlated with the pretransplant XM status (p = 0.0001) and was directly proportional to the semiquantitative XM scoring (data not shown). Primary routine histopathologic diagnoses, in cases with diffuse C4d staining in any compartment, were as follows: biliary stricturing (n = 3), ischemic injury (n = 3), ACR (n = 6) and AMR (n = 1; Figure 3). Sensitivity and specificity analyses for each of the compartments, however, were less than desirable (Table S2).

Table 3.  C4d staining in the various vascular endothelial cell compartments within liver biopsies obtained within the first 3 weeks after transplantation from XM+ and XM– recipients. Only biopsies that were considered adequate were included in the analysis
Variablesn = 196(%)n = 34(%)p-Value*
  1. *p-Values obtained using regression models adjusted for gender. XM–, crossmatch negative; XM+= crossmatch positive; N = none; M = minimal; F = focal; D = diffuse.

Artery     0.004
 N, M171872162 
 F, D 25131338 
Portal vein    0.0001
 N, M159811647 
 F, D 37191853 
Central vein     0.950
 N, M1961003088 
 F, D  00 412 
Portal capillaries     0.004
 N, M158811956 
 F, D 38191544 
Sinusoidal      0.03
 N, M192983088 
 F, D  42 412 
Total stain    0.0001
 Mean (SD)2.0 (2.1)4.4 (4.0) 

Figure 3. Liver allograft from a XM+ recipient obtained 10 days after transplantation for primary biliary cirrhosis. The patient developed otherwise unexplained hyperbilirubinemia and showed a significant fall in platelet counts. Comparison of the H&-E-stained slide (top) and C4d-stained slide (bottom). The portal tracts (PT) from each slide are shown at higher magnification in the left inset. Note the positive C4d staining in the portal vein, portal capillaries and central vein (CV). The sinusoids were weakly stained or negative in this case. Note also the low-grade ductular reaction and mild acute cholangiolitis (arrow in top panel shown at higher magnification in inset) with mild hepatocyte swelling and hepatocanalicular cholestasis.

Download figure to PowerPoint

C4d staining also showed correlations with “prospectively” scored routine histopathology findings (Table 4). Biopsies with either focal or diffuse portal capillary C4d deposits were more likely to show higher RAI subendothelial scores; sinusoidal C4d deposits correlated with RAI for portal inflammation and total C4d score correlated with all three RAI parameters (Table 4). C4d+ biopsies also showed insignificant trends toward higher incidences of ACR and cholestatic liver injury tests (Table 5). Within the subgroup of 34 recipients with ACR, there was no difference between C4d+ (n = 21) and C4d– (n = 13) biopsies with respect to the severity of Banff ACR grade or the liver injury test profile during the first 3 weeks posttransplantation (data not shown), although the cohort tested was relatively small. When XM+/C4d+ were compared to XM–/C4d– recipients, there was slightly more ACR (47% vs. 38%; p = 0.6), a trend toward lower AST (p = 0.06) and ALT (p = 0.14) levels, but significantly higher ALKP levels (p = 0.04). Neither the pattern nor intensity of C4d staining differed between living and deceased donor allografts.

Table 4.  Correlation between C4d staining in various hepatic vascular endothelial cell compartments and routine histopathologic findings for biopsies obtained within the first 3 weeks after transplantation
Histological featuresARTPVCVCapSinusTotal score1
  1. 1The total C4d scores was calculated by adding together the C4d score for each compartment, with minimal = 1, focal = 2 and diffuse = 3 for each compartment. 2Negative correlation. 3Positive correlation.

Ductular reaction0.950.650.460.190.290.99
Central necrosis0.550.680.070.480.760.61
Central fibrosis0.320.500.290.990.780.51
Dis. Ballooning0.00620.610.270.310.740.37
Arch distal fibrosis0.150.420.900.220.320.62
Portal inflammatory intensity0.520.080.0220.230.590.98
Lobular inflammatory intensity0.590.400.480.970.270.74
Necrosis infarct ischemia0.520.910.200.230.410.56
Portal fibrosis0.380.930.840.130.780.82
RAI–bile duct damage0.360.680.870.280.230.053
RAI–portal inflammation0.130.920.590.140.0330.053
RAI–venous endothelial inflammation0.070.980.920.060.200.08
Table 5.  Liver injury test profile of patients whose biopsies showed focal or diffuse C4d staining, irrespective of the XM results, versus those with minimal or negative staining during the first 3 weeks after transplantation
  1. AST = aspartate aminotransferase; ALT = alanine aminotransferase; GGTP = gamma glutamyl transpeptide; ALKP = alkaline phosphatase.

AST (IU/L)    0.06
 Mean (SD)417.7 (1085)169.1 (329.5) 
ALT (IU/L)    0.80
 Mean (SD)324.2 (613.6)177.7 (167) 
GGTP (IU/L)    0.44
 Mean (SD)275.2 (309.1)268.3 (216.1) 
ALKP (IU/L)    0.13
 Mean (SD)124.6 (60.5)200.6 (216.1) 
Total bilirubin (mg/dL)    0.31
 Mean (SD)8.7 (5.4)9.5 (5.7) 
Rejection    0.26

Focal or diffuse C4d staining, irrespective of XM results, decreased with the distance of that compartment from the afferent blood supply: portal vein > portal capillaries > hepatic artery branches > sinusoids > CVs (Table 3). Portal structures (portal vein, capillaries and hepatic arteries) more often showed concordant C4d staining with each other than combinations that included either the sinusoids or CVs (data not shown).

In late biopsies, C4d staining was compared to Luminex-based HLA DSA status. A total of 67 biopsies from 62 patients taken >21 days after transplantation (range 22–1049 days) and within 2 weeks of DSA testing were analyzed (Table 6). Biopsies from DSA+ recipients (8/17; 47%) more often showed histopathologic evidence of ACR compared to 6 of 50 (12%) of DSA− recipients (p = 0.005). DSA+ recipients more frequently showed portal vein endothelial C4d staining and a trend toward more portal capillary staining and a higher total C4d score (Table 6). However, biopsies from DSA+/C4d+ recipients (n = 10) failed to show higher liver injury test profiles or more severe Banff grade of ACR (data not shown) compared to DSA–/C4d– recipients (n = 29).

Table 6.  C4d staining in the various liver endothelial cell compartments according to the DSA status in biopsies obtained within 2 weeks of the DSA test
Variablesn = 50(%)n = 17(%)p-Value
  1. p-Values obtained using regression models adjusted for gender. DSA = donor-specific antibody negative; DSA+= donor-specific antibody positive; N = none; M = minimal; F = focal; D = diffuse.

Arteries    0.950
 N, M33661271 
 F, D1734529 
Portal vein    0.030
 N, M3876847 
 F, D1224953 
Central vein    0.960
 N, M479417100 
 F, D3600 
Portal capillaries    0.130
 N, M3060741 
 F, D20401059 
Sinusoidal    0.880
 N, M47941694 
 F, D3616 
Total stain   0.150
 Mean (SD)2.9 (2.9)3.9 (2.9) 

Primary histopathologic diagnoses in DSA+/C4d+ (defined as a total C4d score >3) recipients (n = 9) included: ACR

(n = 3; occurring at 181d, 186d, 425d); ACR with biliary strictures in a XM+ patient previous diagnosed with AMR early after transplantation (n = 1; 373d); sinusoidal congestion and nodular regenerative hyperplasia (n = 1; 372d); changes suggestive of biliary strictures (n = 1; 30d); ischemic cholangitis associated with an arterial problem (n = 1; 31d); ischemic cholangitis, changes suspicious for AMR (n = 1; 23d), followed by allograft failure attributed to severe ACR and CR; and nonspecific changes/low-grade ductular reaction (n = 1; 218d). Two of these recipients experienced allograft failure in which rejection was a major or contributing factor; one died of sepsis with low-grade ductopenia, one died of allograft failure because of noncompliance; and five are alive with functioning allografts.

Conversely, primary histopathologic diagnoses in DSA–/C4d+ (total C4d score >3) recipients (n = 14) were: recurrent HCV (n = 4; 235d, 350d, 347d, 369d); nonspecific changes/minimal inflammation (n = 4; 368d, 379d, 353d, 348d); indeterminate to mild ACR (n = 2; 31d, 70d); steatohepatitis/other (n = 2; 29d and 857d); changes suggestive of biliary strictures (n = 1; 46d); and changes suspicious for AMR (n = 1 95d). All patients are alive with functioning allografts, except one that died of simultaneous lung allograft rejection and one patient required retransplantation for strongly suspected AMR even with negative DSA testing.

Does C4d staining assist the histopathological diagnosis?

To address this question, all early biopsies were blindly and retrospectively re-reviewed knowing only that biopsies were obtained <21 days after transplantation. A retrospective diagnosis was rendered on the basis of the H&E stain alone. Then, C4d staining was re-evaluated semiquantitatively (minimal/negative, focal and diffuse; see “Patients and Methods”), to determine if it added information that could help better classify the case.

Results showed excellent correlation between the total numerical C4d score and semiquantitative C4d scoring (p < 0.0001). And diffuse C4d staining (n = 22/220; 10%) helped more specifically classify cases early after transplantation; most also showed periportal sinusoidal staining. Diffuse C4d+/XM+ recipients (n = 5), after detailed retrospective clinicopathologic review, likely suffered from relatively isolated AMR. Two of these AMR cases also showed mild portal eosinophilia—a frequent manifestation of early cellular rejection. AMR was originally suspected in three, but repeat XM or DSA testing was not carried out. Three of five are alive 4–7 years after transplantation with functioning grafts: two have recurrent HCV and one developed high-grade biliary strictures in a living donor allograft; two died later after transplantation with functioning grafts. Another AMR case was identified “blindly” by diffuse C4d staining (5d) in an ABO-incompatible allograft that eventually failed from AMR-induced hemorrhagic necrosis (34d).

A component of AMR was suspected in recipients with active or resolving ACR (n = 15) on the basis of diffuse C4d+ staining, but no differences were seen between diffuse C4d+ and C4d– rejection with respect to Banff grade of or liver injury tests (data not shown). Prophylactic steroid therapy in XM+ recipients, however, complicated the analysis. One diffuse C4d+/XM– case had severe hepatic artery stenosis; graft failure eventually caused patient death at 3 months. Altogether 21 of 22 patients with strong and diffuse C4d staining had coexistent ACR or predominantly AMR.

Focal C4d (n = 35/220; 16%) was helpful in 16 of 35 of the biopsies (44%), but resulted in fewer (29%) more specific diagnoses. Most biopsies (163/220; 74%) showed minimal (n = 84) or negative (n = 79) C4d staining, but results helped clarify the diagnosis in 40% and 56% biopsies, respectively, by excluding that AMR might be contributing to histologic patterns that mimic AMR.

A similar analysis was done on biopsies obtained >21 days from DSA+ recipients (n = 67), but diffuse C4d staining 6 of 67 (9%) was less intense than in the early XM+ group. Diffuse C4d+/DSA+ (n = 3) recipients showed coexistent mild ACR (n = 1; 181d) and experienced graft failure because of noncompliance; moderate to severe ACR (n = 1; 425d) treated successfully after which recurrent HCV reappeared; and mild CV fibrosis (n = 1; 218d) and normal liver injury tests and has a functioning allograft without further biopsies. Diagnosis and follow-up in the three diffuse C4d+/DSA– recipients showed: allograft failure felt to be consistent with AMR (n = 1; 95d), suspected AIH (n = 1; 28d) without further biopsies and ischemic cholangitis allograft failure (n = 1; 39 months).

Focal C4d staining was seen in nine patients, five of whom were DSA+ and showed: AMR (23d)—allograft eventually failed from severe acute and chronic rejection; failed allograft with combined ACR and AMR (46d); failed allograft with biliary cast syndrome (186d) preceded by severe ACR; nonspecific changes (218d) with functioning graft 2 years later; biliary problem (373d) that continues for several years. Those with focal C4d+/DSA– recipients showed: ACR (n = 2), biliary (n = 1) and recurrent HCV (n = 1). All these allografts remain functional after several years follow-up, but two have biliary strictures. Altogether, 5 of 8 focal or diffuse C4d+/DSA+ and 3 of 7 focal or diffuse C4d+/DSA– recipients showed significant rejection-related activity.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

This study confirms that liver allografts are very resistant to AMR from preformed and de novo alloantibodies, as detected by XM and HLA DSA testing, respectively. Possible mechanisms have been reviewed elsewhere (6). Liver allografts, however, are not completely spared from the damaging effects of AMR (1,2,8,24,29), which has adversely affected patient and allograft survival in some (25,26,30,31), but not all studies (reviewed in Ref. 32). XM+ recipients, in this study, experienced statistically insignificant decreased graft and patient survival rates. Studies showing poorer graft and/or patient survival rates in XM+ recipients from our center (2,29) were based on patient subpopulations that were treated differently, without the prophylactic steroids reported herein.

C4d deposits are detected significantly more often in biopsies from XM+ recipients within the first 3 weeks after transplantation and late biopsies from DSA+ recipients—similar to other studies (13,19). Most C4d deposits were detected in the proximal liver microvasculature: portal veins, portal capillaries and hepatic artery branches, which are first exposed to the circulation, and less in the sinusoids and CVs. This is likely attributable to the large liver volume, ability to absorb antibodies and use of FFPE tissue staining, which is less sensitive than assays on frozen tissue. Interestingly, C4d staining showed a much stronger correlation with XM than with DSA testing. This might be due to the lower sensitivity (higher antibody concentrations), greater cytotoxic functional significance, and the measurement of a broader array of antibody specificities in XM testing; allo-antibodies are not alone in their ability to cause allograft damage, particularly late after transplantation.

A consensus conference on AMR in solid organ allografts (33), which excluded consideration of livers, required: intragraft C4d deposits, allograft dysfunction and circulating antidonor antibodies for a diagnosis of AMR. Until there is a better understanding of subtle and/or chronic antibody-mediated liver allograft injury, it is probably best to construct a restrictive AMR diagnosis that would also require “tissue injury pattern consistent with AMR”. Findings that should trigger further investigation early after transplantation include an otherwise unexplained ductular reaction, variable neutrophilic, eosinophilic and histiocytic portal inflammation, cholangiolitis and variable centrilobular hepatocyte swelling and hepatocanalicular cholestasis—identical to those described nearly 20 years ago (2) and a more recent case report of isolated AMR (32).

Once other differential diagnostic possibilities, such as preservation/reperfusion injury, sepsis and biliary stricturing, have been reasonably excluded, “diffuse” C4d (strong portal vein, capillary and often periportal sinusoidal staining in a majority of PTs) and affirmation of XM+, DSA+ or ABO incompatibility status can help to further substantiate an AMR diagnosis, but only when C4d staining is obvious and interpreted in conjunction with routine histopathologic, serologic and clinical profiles. This algorithm led to a 5% incidence rate of relatively isolated AMR in XM+ patients early after transplantation, although many more have combined ACR/AMR, as in previous studies (2) and in kidney (34) and lung (35) allografts. Diffuse C4d staining, however, was also seen in several XM– and DSA– recipients with biliary strictures, autoimmune and HCV hepatitis.

Focal C4d staining alone, either without DSA or without tissue damage suggestive of AMR, was less discriminating or helpful—at least after short-term follow-up. Further study and long-term follow-up are needed. We know, however, that stable pediatric liver allograft recipients with tissue C4d deposits on frozen sections and normal or near normal liver injury tests and biopsies can harbor DSA and their potentially damaging effects are manifested clinically only after weaning of immunosuppression (36,37). Conversely, minimal/negative C4d staining can help exclude AMR in most cases where routine histopathologic findings might suggest and AMR component. The exception is rare cases that have very strong histopathologic evidence of an AMR component such as arteritis, but are C4d–.

Interpretation of C4d stains (especially focal C4d), late (>1 year) can be more complex than early posttransplantation because: (a) necroinflammatory diseases that commonly recur late after transplantation (e.g. HCV, biliary strictures) are also associated with C4d deposits even when patients are DSA–; (b) XM+ recipients carry a histopathologic legacy of enhanced immunologic reactivity, but DSA might not be detected in the serum because of the large absorptive capacity of the liver; and (c) C4d staining tended to be less intense than in XM+ patients in this study.

Bouron-Dal Soglio et al. (23) reported C4d deposits in 83% native liver biopsies with AIH, in 40% HCV biopsies and in 89% HBV biopsies, but no expression in four noninflammatory liver specimens used as negative controls (23). Perihepatocyte C4d deposits have also been observed with steatohepatitis (38). These observations suggest that C4d deposition is not restricted to alloantibody-mediated injury. C4d deposits have also been reported in graft-versus-host disease (39) and CR (7,22), which are likely alloantibody related.

It is uncertain whether other insults or non-HLA antibodies contribute to C4d deposition or if the liver absorbs all DSA preventing detection in the peripheral circulation. Further studies, long-term follow-up and perhaps a better tissue marker(s) of AMR are needed. Ideally, this would include routine crossmatching, even retrospectively, protocol biopsy analysis and pre- and posttransplant solid phase DSA measurements, such as, Luminex single antigen testing, combined with donor and recipient HLA typing. It should be remembered, however, that HLA DSA are only a relatively narrow range of possible antibodies that could potentially trigger C4d deposition.

In summary, C4d staining in liver allografts can be useful, particularly when obvious and diffuse, but absolutely needs to be correlated with routine histopathologic findings and clinical and serologic/immunologic profiles, as suggested for renal allografts (34). The significance and pitfalls of C4d staining interpretation and the pathophysiology of AMR in the liver, particularly late after transplantation, is less mature than for kidney allografts (12) and more studies, with longer term follow-up, are needed to detect the entire range of injury and response to injury.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

This manuscript was not prepared or funded, in any part, by a commercial organization.

The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.


  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Patients and Methods
  5. Results
  6. Discussion
  7. Disclosure
  8. References
  9. Supporting Information

Table S1: Liver injury test profiles of XM+ and XM- liver allograft recipients >1 year after transplantation

Table S2: Sensitivity and specificity analysis of C4d staining for the crossmatch status

AJT_3786_sm_Methods.docx59KSupporting info item
AJT_3786_sm_Tables.docx97KSupporting info item

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.