Journal Logo

Original Clinical Science—Liver

Peripheral Donor-specific Antibodies Are Associated With Histology and Cellular Subtypes in Protocol Liver Biopsies of Pediatric Recipients

Cousin, Vladimir L.1; Rougemont, Anne-Laure MD2,3; Rubbia-Brandt, Laura MD, PhD2,3; Wildhaber, Barbara E. MD1,4; Villard, Jean MD, PhD3,5; Ferrari-Lacraz, Sylvie MD, PhD3,5; McLin, Valérie A. MD1,4

Author Information
doi: 10.1097/TP.0000000000003099



Protocol liver biopsies (PB) have shown that inflammation and fibrosis are present to varying degrees in the clinically quiescent allograft of the pediatric recipient.1,2 The long-term significance of these findings is still open to debate, but warrants attention to ensure extended allograft longevity.3 Recently, there has been much focus on graft fibrosis, but the molecular and cellular events upstream of fibrosis are still poorly understood.4-6 It follows that a detailed description of the cellular actors in the quiescent liver allograft is paramount to begin to unravel the immune mechanisms underlying both fibrosis and graft tolerance.

The role of donor-specific antibody (DSA) is increasingly understood in other solid organ transplants,7,8 but their significance in LT is still unclear, probably owing to the liver’s inherent immunomodulatory function.9,10 In children, DSA have been linked to chronic rejection and fibrosis11,12 and possibly to a higher rate of graft abnormalities.4,13-15 In contrast to these reports, the presence of DSA has also been shown to fluctuate over time in living donor pediatric recipients who were successfully weaned off of immunosuppression (IS) with minimal postweaning histological modification, thereby questioning the biological and histological significance of circulating DSA.16 Understanding whether circulating DSA are the hallmark of a cellular process in situ may be clinically useful: should circulating DSA be associated with a histological phenotype, they could ultimately serve as biomarkers to inform management decisions.

We hypothesized that DSA may be associated with the cellular effectors of injury in the quiescent allograft. Therefore, the primary aim of our study was to describe and quantify precisely immune cells in the portal tracts and lobules of PB after pediatric liver transplantation (pLT). The secondary aim was to analyze the relationship between peripheral DSA profile at the time of PB, histological phenotype and cellular subtypes identified in PB.


We performed a retrospective review of patients having undergone pLT in our center from 1989 to identify those with at least 1 PB. Protocol biopsy was defined as a biopsy performed as a surveillance biopsy, otherwise not indicated by any clinical or laboratory anomalies. Exclusion criteria were the following: ALT or GGT >50 IU/L,4,6,17 biopsy performed within 12 months of a previous one (protocol or clinically indicated). Patient assent or caregiver consent was obtained at the time of listing for transplant for the retrospective use of samples. Consent for the procedure was obtained at the time of PB. All biopsies were performed using the Menghini technique with a 1.6 mm needle, following current recommendations.18 Serum obtained at time of PB or the closest available was analyzed for concentration, number, specificity, and C1q binding activity of DSA. All patients were on a calcineurin inhibitor (CNI) based regimen at the time of biopsy. Type of CNI and trough level depended on transplant era and clinical needs.

Histology and Immunohistochemistry

Two to four 3 μm-thick sections of the paraffin-embedded liver tissue were stained with Hematoxylin and Eosin, Masson’s trichrome and reticulin stain. All were reviewed by 1 pediatric pathologist (A.-L.R.). Histology review and immunohistochemistry evaluation were performed blinded to clinical data and DSA status. Forty-two (42/44) biopsies were recut and immunostained for immune cell markers (2 samples were too small for immunohistochemistry after histological analysis). The following markers were used to characterize immune cells: CD3, CD4, and CD8 for T lymphocytes, CD20 for B lymphocytes, MUM1 for plasma cell, CD68 for Küpffer cell, and FoxP3 for Treg.9 Detailed immunohistochemistry methods for are summarized in SDC1,

Biopsies were separated into 4 different phenotypes upon histological assessment, before immunohistochemical evaluation: biopsies with no relevant histological anomalies (normal), biopsies with fibrosis only (absence of inflammation), biopsies with inflammation only (absence of fibrosis), and biopsies with both inflammation and fibrosis. Inflammation varied from minimal (few inflammatory cells in a minority of the portal tracts and/or lobules) to moderate or severe. Moderate or severe inflammation in the portal tracts was defined by nodular infiltrates or presence of interface activity. In the lobules and centrilobular areas, dense or atypical lymphohistiocytic aggregates and/or centrilobular vein endotheliitis with or without hepatocyte drop-out, defined moderate or severe inflammation. Macrophages were not considered as part of the inflammatory signature. Allograft fibrosis was evaluated using the Liver Allograft Fibrosis Score (LAFS).6 In short, the semiquantitative LAFS fibrosis scoring system was applied as previously described.6 The Masson’s trichrome stain, and to a lesser extent the reticulin stain, were used to assess fibrosis deposition in portal tracts, along sinusoids, and centrilobular veins. In each of these 3 areas, fibrosis was staged from 0 to 3, leading to a final LAFS score of 0–9. Rejection was evaluated according to the latest definitions.19

Immunohistochemistry was analyzed simultaneously by A.-L.R. and V.L.C. For each cellular marker, positive cells were quantified manually, and portal tract surface was determined using NIS-Element Documentation software (Nikon AG, Zurich, Switzerland). Positive cells were counted in all available portal tracts, and density reported as the median per surface unit (1000 µm2). Lobular inflammatory cells were counted semiquantitatively on 3 selected complete lobules per biopsy,20 and expressed as a median per PB (Figure S1, SDC,

Patient records were queried for donor, recipient and surgical characteristics. History of previous T cell-mediated rejection (TCMR) was recorded.19 History of vascular or biliary complication was defined as previous vascular or biliary problem requiring intervention (surgical or by interventional radiology). Technical variant stands for split graft or reduced graft. Blood group mismatch was defined as compatible but not identical. Laboratory values were all obtained in the 24 hours preceding PB. Positive autoantibodies were defined as presence of at least 1 positive autoantibody at the time of PB. Methods for autoantibody detection are summarized in the SDC (

Recipient- and donor-HLA typing was performed at the time of transplant, and determination of anti-HLA antibody was performed on retrospective samples for the purpose of this study. Serum samples collected at time of PB in 40 patients were retrospectively analyzed. For those patients who did not have serum at time of PB, 1 was obtained 1 year before PB, two 6 months after, and 1 sample was missing. HLA-A, -B, -DR and DQ typing for both recipients and donors were performed by SSO-Luminex method (One Lambda Inc., Los Angeles, CA). Screening for anti-HLA antibody at the time of LT using an ELISA assay before 200821 and by LABScreen Mixed (One Lambda Inc.) thereafter.22 Anti-HLA identification was performed using LABscreen single antigen.

DSA were considered positive above a Mean Fluorescence Index (MFI) of 1000.11,14,23-30 De novo DSA were defined as DSA present only after transplant in patients either without or with different anti-HLA antibodies before LT. C1q binding affinity of DSA was measured using a C1q-binding Luminex assay. MFI above 300 was considered positive for C1q+ DSA.

Statistical Analysis

Continuous variables were defined using median and interquartile range (IQR). Cell density was reported as median and IQR. For categorical variables, we used proportion in percent. A Wilcoxon rank-sum test was used to compare continuous variables and a Chi-squared test or Fisher’s exact test for categorical variables. P < 0.05 was considered statistically significant. Statistical analyses were carried out using Stata software, version 14.1 (StataCorp, College Station, TX).


Population Characteristics

One hundred thirty-six patients were transplanted in our center between 1989 and 2014. Among these, 44 patients underwent at least 1 PB. Patient characteristics are summarized in Table 1. None of the patients included underwent retransplantation. and there were not ABO incompatible grafts.

Patient characteristics

The primary IS protocol was calcineurin-inhibitor based, with preferential use of tacrolimus from the year 2000. At the time of PB, patients were on varying, clinically indicated IS. For those on CNI, 20 (46.5%) were in their target trough range, 8 (18.6%) below target, 15 (34.9%) above target. Steroids were used in 4 patients at PB, and 10 had a combination CNI with mycophenolate or azathioprine. One patient had mycophenolate mofetil and steroids at biopsy because of renal impairment.

Preformed DSA were present in 22.7% (10/44) of patients, most against class II antigens (8/10 preformed DSA) with DQ DSA being the most frequent (5/10). De novo DSA were present in 27/43 children (62.8%). All patients with de novo DSA had at least 1 class II DSA. De novo DSA were mainly DQ (24/27). One patient had persistent and no de novo DSA following LT. C1q binding activity was identified in 55.6% of patients with de novo DSA (Table 1). Furthermore, C1q binding DSA were more frequent in patients with high de novo DSA MFI (P < 0.001). The complete DSA results are summarized in Table S1 (SDC, Two C1q+ class I DSA were identified only through the use of the C1q-binding assays. According to recent definitions and using a C4d marker (data not shown), 7 patients met criteria for possible chronic antibody-mediated rejection and 3 for probable chronic active antibody-mediated rejection.19 One patient presented with plasma cell-rich rejection and 2 with Banff 3 rejection on PB.

Cellular Characteristics of Portal Tract Infiltrate

The numbers of portal tracts available for analysis per PB varied slightly due to tissue sectioning. Mean and median numbers of portal tracts assessed were: CD3 (12;13), CD4 (14;15), CD8 (16;15), CD20 (14;15), MUM1 (13;13), CD68 (13;14), and FoxP3 (15;16). Overall characteristics of the portal infiltrate are summarized in Figure S2 (SDC, In normal PB, portal infiltrate was mostly composed of CD3+ T cells (0.84 cells/1000 µm2 [IQR 0.81–0.85]) and CD68+ macrophages (0.84 cells/1000 µm2 [IQR 0.81–0.92]). Compared to all biopsies with histological abnormalities, normal biopsies were remarkable for fewer portal CD3+ cells (0.84 cells/1000 µm2 [IQR 0.81–0.85] versus 1.46 [IQR 0.97–1.84], P = 0.03). FoxP3+ cells were rare in portal tracts regardless of histological phenotype.

Histological Phenotype: Association With Cellular and Serum Markers

Biopsies were classified into 4 phenotypes: normal (n = 5), fibrosis only (n = 6), inflammation only (n = 15), and inflammation and fibrosis (n = 18). The biopsies read as normal (11%, 5/44) served as internal controls. Median time at PB was as follows: 49 months (IQR 48–61) for normal, 35 months (IQR 24–70) for inflammation only, 72 months (IQR 59–108) for inflammation and fibrosis, and 84.5 months (IQR 76–96) for fibrosis only. PB displaying inflammation were performed significantly earlier than those with inflammation and fibrosis (P = 0.009). Histological findings are detailed in Table 2. Figure 1 describes the portal tract infiltrate according to phenotype. We found a significantly denser infiltrate of CD3+ cells in patients with inflammation and fibrosis compared to either normal biopsies or biopsies with fibrosis. PB with fibrosis had a very low CD20+ cell count, significantly lower than those with inflammation only or inflammation and fibrosis. The CD68+ infiltrate was denser in PB with fibrosis in comparison to those with inflammation only. Representative histology is illustrated in Figure 2. The presence of inflammation (any) was not associated with fibrosis (any) (P = 1), portal fibrosis (P = 0.48), or centrolobular fibrosis (P = 0.49) (Table S2, SDC, The proportion of patients within their CNI-trough target range did not differ significantly between phenotypes (Figure S3, SDC,

Histological characteristics of protocol biopsies
Portal infiltrate and histological subtype. Y-axis: cellular density in the portal tract expressed as cells/1000 µm2. X-axis: cellular subtypes. Blue: normal PB. Green: fibrosis only. Yellow: inflammation only. Red: inflammation with fibrosis. *P < 0.05; **P < 0.01. DSA, donor-specific antibody; PB, protocol liver biopsies.
Comparison of selected immunohistochemical findings (CD3, CD20, and CD68) in the 4 histological categories (normal, inflammation only, fibrosis only, inflammation with fibrosis). A–D, “Normal” biopsy (no significant inflammation or fibrosis on Hematoxylin and Eosin (H&E) evaluation, (A) performed in a 6-y-old girl who underwent liver transplantation for cirrhosis of unknown etiology at age 1 y and 4 mo. CD3+ T cells (B) and CD20+ B cells (C) are sparse, whereas CD68 (D) highlights a predominant population of lobular macrophages. E–H, Protocol biopsy performed in a 14-y-old boy 5 y after liver transplantation for lysosomal acid lipase deficiency, included in the “inflammation only” phenotype. Portal tracts contained a moderate inflammatory component on H&E evaluation (E), mainly composed of CD3+ T cells (F). CD20+ B cells (G) and CD68+ macrophages (H) are comparatively sparse. I–L, “Fibrosis only” biopsy (no significant inflammation on H&E evaluation), performed 10 y after liver transplantation for BA performed at 9 mo of age (male patient). Mild periportal fibrosis, scored as F1, is seen on Masson’s trichrome and is accompanied by focal perisinusoidal and pericentrilobular fibrosis (I). CD3+ (J) and CD20+ (K) are few, whereas moderate numbers of CD68+ macrophages are seen (L). M–P, Protocol liver biopsy displaying both “inflammation and fibrosis.” PB performed 5 y after liver transplantation at 14 mo of age in a female patient with BA. Fibrosis was classified as F3 (Masson’s trichrome, M). CD3+ cells (N) comprise the main portal tract inflammatory cell population, whereas CD20+ cells are rare (O). Mild to moderate numbers of CD68+ macrophages populate the lobules (P). Original magnification ×100. PB, protocol liver biopsies.

Associations between de novo DSA and histological phenotypes are summarized in Figure 3A and B. C1q binding DSA were significantly more prevalent in patients whose PB showed inflammation and fibrosis (52.9%) in comparison to those with fibrosis only (0%) (P = 0.04). There was a higher prevalence of C1q binding DSA in PB with inflammation only (46.7%) compared to fibrosis only (0%) (P = 0.06), but this did not reach statistical significance. Pretransplant and transplant variables did not differ between phenotypes. Likewise, there was no significant difference in posttransplant factors between phenotypes.

DSA and histological phenotypes. Panel A: de novo DSA according to histological phenotype. Significantly higher proportion of DSA in inflammation and fibrosis compared to normal (88% vs 20%, P = 0.009) or fibrosis (88% vs 33%, P = 0.02). Panel B: number of de novo DSA according to histological phenotype. The number of de novo DSA was significantly higher in patients with inflammation and fibrosis in comparison to all other phenotypes. *P < 0.05; **P < 0.01. DSA, donor-specific antibody.

Portal Infiltrate, Anti-HLA Antibodies and Autoantibodies

De novo DSA were associated respectively with a higher number of portal CD3+, CD4+, CD8+, and CD20+ cells (Figure 4A). There was also a significantly higher density of CD3+ and CD8+ cells in the portal tracts or lobules of patients with C1q binding DSA (Figure 4B). Autoantibodies were identified in 22 patients at the time of PB. The presence of autoantibodies was associated with CD8+ cell density in portal tracts (0.51 [IQR 0.34–0.69] cells/1000 µm2 versus 0.25 [IQR 0.18–0.51] cells/1000 µm2, P = 0.008) (Figure 4C).

Cell type and antibody subtype. Y-axis: cellular density in the portal tract expressed as cells/1000 µm2. X-axis: cellular subtypes. Panel A: cellular characteristics of the infiltrate according to the presence of DSA (27 patients with de novo DSA). Panel B: characteristics of the infiltrate according to the presence of C1q binding DSA (17 patients had presence of de novo DSA), Panel C: characteristics of the infiltrate according to the presence of auto-antibodies (22 patients had autoantibodies present in serum). Auto-antibodies were anti-neutrophil-cytoplasmic-antigen (ANCA), anti-nuclear antibody (ANA), anti-smooth muscle (SMA), anti-mitochondrial, anti-liver-kidney-microsome (LKM), and anti soluble liver antigen (SLA). *P < 0.05; **P < 0.01; ***P = <0.001. DSA, donor-specific antibody.

The following parameters were not associated with cellular density or predominant cell type in the portal tracts: pre-LT DSA characteristics, type of CNI at biopsy, donor-recipient gender mismatch, donor-recipient blood group mismatch, history of TCMR, history of biliary or vascular complication and indication for LT.

Cellular Characteristics of Lobular Infiltrate

Overall, the main lobular cells were CD68+ cells (macrophages) and CD3+ cells (T cells). CD20+ B cells, MUM1+ B cells, and plasma cells were rare (Figure S4A, SDC, Furthermore, normal PB were characterized by fewer CD8+ cytotoxic T cells in acinar zone 1 (P = 0.05) compared to all other histological phenotypes.

Upon closer examination, lobules of PB with fibrosis, either with or without inflammation, showed significantly higher number of CD8+ cells in zone 3 in comparison to PB with inflammation only (Table 3A). Biopsies with fibrosis showed a lower number of CD20+ cells in zone 3 in comparison to PB with inflammation, either with or without fibrosis. Moreover, CD68+ cells were in higher number in zone 3 in case of fibrosis only when compared to PB with inflammation and fibrosis (Figure 5).

Lobular infiltrate and histological features
Comparison of CD8+ cytotoxic T cell infiltrate, between categories “Inflammation only” and “Fibrosis only.” A–C, Biopsy performed 5 y after transplantation for biliary atresia in a 10-y-old girl, considered as “inflammation only” (no fibrosis). Despite a dense inflammatory infiltrate in the portal tracts on H&E evaluation (A), partially highlighted by CD8 (B), CD8+ cytotoxic T cells are sparse in acinar periportal zone 1, and in pericentrilobular zone 3 (B and C). D–F, Same biopsy as Figure 3I–L. Unlike the previous patient, no significant inflammation is seen and periportal fibrosis is only mild on Masson’s trichrome (A), yet increased dispersed CD8+ cytotoxic T cells are seen throughout the lobule (E, acinar periportal and midlobular zones 1 and 2; F, midlobular and pericentrilobular zones 2 and 3). Original magnification ×200. H&E, hematoxylin and eosin.

CD8+ cells were present in significantly higher numbers in zone 1 (P = 0.02) and zone 3 (P = 0.003) of biopsies displaying fibrosis (with or without inflammation) in comparison to PB without fibrosis. These cells were more numerous in zones 1 (P = 0.01), 2 (P = 0.03), and 3 (P = 0.002) of biopsies displaying portal fibrosis. Finally, their presence in zone 3 was associated with centrolobular fibrosis (P = 0.008 in zone 3) (Table 3B).

Fibrosis and Associated Factors

Median/mean LAFS was 0/0.5, 0/0.5, and 0/0.75, respectively, in portal, sinusoidal, and centrolobular areas. The overall median/mean LAFS was 1.5/1.8. Presence of fibrosis (with or without inflammation) was associated with time from LT to biopsy: biopsies without fibrosis were obtained at a median 36 months (IQR 24–65.5) versus 80 months (IQR 58.5–102) for biopsies with fibrosis (P = 0.002). The presence of centrolobular fibrosis and portal fibrosis were also associated with time after pLT. PB with portal fibrosis were obtained at a median time of 73 months (IQR 60–120) and those without at 49 months (IQR 26–85) (P = 0.02). PB with centrolobular fibrosis were obtained at a median 80 months from LT (IQR 59.5–108), while those without were obtained at 48.5 months (IQR 28.5–83) (P = 0.01). Inflammatory cell density in the portal tracts did not differ between biopsies with fibrosis and PB without fibrosis (Figure S4B, SDC,, and this held true when specifically looking at centrolobular fibrosis or portal fibrosis.

The following factors were not associated with a significant difference in fibrosis, portal fibrosis or centrolobular fibrosis: diagnosis, age at transplantation, cold ischemia time, gender- or blood group-mismatch, pretransplantation DSA status, type of CNI at biopsy, history of vascular or biliary complications, or history of TCMR. Overall, DSA class, titer and number were not significantly different between patients with or without fibrosis, whether portal or centrolobular.


In a representative sample of PB in stable pLT recipients, inflammation and fibrosis were frequent, often overlapping findings, confirming the findings of others.1,2,4,31-33 Similar to other reports, a few patients (11%) displayed normal histology on PB. Biopsies with inflammation were common and characterized by a portal infiltrate enriched in CD3+ and CD20+ cells with comparatively low CD68+ cells. The presence of inflammation was associated with the presence of class II de novo DSA and number of DSA. The presence of fibrosis was associated with time since transplant. At the tissue level, it was associated with the presence of lobular CD8+ inflammation. All in all, these findings confirm our hypothesis that peripheral DSA may be a molecular signature of inflammation in the graft, both portal and lobular. Although this study was not designed to test the causality of DSA in graft injury, it does provide evidence that in the seemingly quiescent allograft, inflammation characterized by a rich T cell and B cell infiltrate is often associated with circulating DSA at the time of PB, and that the presence of CD68+ cells may be protective of lymphocytic inflammation.

Lobular and Portal Inflammation

Late posttransplant inflammation in the allograft is a commonly accepted problem affecting as many as 75% of recipients.1,2 Evidence suggests that inflammation leads to fibrosis although recent reports are contradictory.4 In the present study, CD3+ and CD20+ cells were significantly more numerous in the portal tracts of PB with inflammation, suggesting that both of these cell types are active in the seemingly quiescent allograft. In fact, the presence of portal CD20+ cells was associated with the presence of DSA. >60% of our patients presented with de novo DSA, mostly class II, a proportion in the higher range of literature reports.3,11-14,23,25

Given that in the present study portal inflammation was associated with de novo DSA, class II de novo DSA, number of de novo DSA, and C1q binding DSA, it is tempting to speculate that CD20+ cells in the portal tracts of PBs with inflammation are the histological signature of measurable peripheral DSA titers. Moreover, an association between DSA and indeterminate rejection or chronic rejection was previously described by others.11,13,34 Although our findings suggest that DSA and inflammation are linked, the link remains speculative.19,35,36 Indeed, in the presence of inflammation, hepatocytes upregulate class II HLA expression which in turn may increase production of de novo class II DSA, but the primum movens remains unclear.9 Therefore, DSA may serve as a biomarker of allograft injury.

The presence of portal CD8+ cells was associated with the presence of measurable serum autoantibodies. In autoimmune liver disease, CD8+ cytotoxic T cells recognize antibodies thereby leading to injury.37,38 The presence of autoantibodies has been described in association with allograft inflammation.1,4,39 Moreover, a recent report suggests that antibodies other than DSA contribute to graft hepatitis.26 Therefore, akin to DSA, it is unclear whether these antibodies are pathogenic or a signature of immune activity in the allograft. But it is intriguing that portal CD8+ cells are associated with both de novo DSA and autoantibodies. This finding suggests that T-cells are more than just the effectors of T-cell mediated rejection; rather they may also play a direct or indirect effector role in the humoral response.

Importantly, CD68+ cells were less numerous in biopsies with inflammation. CD68+ cells are intriguing in that their number varies according to histological phenotype. While they are rarer in inflammation, they are more numerous in normal biopsies, yet in larger numbers, they seem to be associated with the presence of fibrosis. It is probable that most of the CD68+ cells are Kupffer cells9 which have the ability to clear immune complexes thereby protecting the graft from the potential deleterious effects of antibodies directed against the donor.9,10,40-42 It was recently shown that Kupffer cells were abundant in the biopsies of strictly selected, operationally tolerant pediatric liver transplant recipients, raising the question of their protective role.16 Along the same lines, a recent report in adult liver transplantation suggested that a lower Kupffer cell count was associated with worse graft outcomes.43 It also is known that recipient macrophages repopulate the liver in the first-month post-LT,9 and it has been suggested that such chimerism may be an important mechanism of tolerance and immunomodulation. Together with previous reports, our data suggests that CD68+ cells may confer a degree of protection against inflammation in the long-term liver allograft. Further characterization of this cell population may offer mechanistic insight into tolerance.


Although studying fibrosis was not a primary aim of our study, it is an important part of any long-term allograft study, as it has been reported by many to be progressive.1,2,4 To this end, we used the LAFS to confirm the findings of others that in our series fibrosis, portal fibrosis and centrolobular fibrosis were indeed associated with time after transplantation.1,2,32 Although overall inflammation (any) was not associated with the presence of fibrosis (any), the presence of lobular CD8+ cells was associated with portal and centrolobular fibrosis, raising the question of the role of CD8+ cells in the fibrogenic cascade. Given that others have described an association between DSA and fibrosis,4,11,12,15,23,34 it is appealing to consider that in the lobule, CD8+ cells partner with DSA in the fibrosis process, although postulating about the mechanism based on present data is premature.

Normal Biopsies Following Pediatric LT

Five biopsies in our cohort were described as normal and served as internal controls. These biopsies were characterized by minimal portal infiltrate by a senior histopathologist and only 1 (1/5) paired serum sample had minimal measurable de novo DSA. This is in contrast to biopsies with inflammation which were characterized by comparatively higher numbers of CD3+ and CD20+ cells and a higher prevalence of DSA. PB with fibrosis resembled normal ones in that they also displayed low CD20+ counts and minimal peripheral DSA. Where the 2 differed was in the comparatively high number of macrophages present in PB with fibrosis. The role of these macrophages may be where future research needs to focus, and their role in the balance between ‘mopping up’ circulating DSA and generating fibrosis further defined.

Taken together, the bird’s-eye view of our cross-sectional data suggests that earlier grafts may be subject to inflammation characterized by the presence of CD3+-, CD8+-, and CD20+-cells, possibly linked to measurable peripheral de novo DSA titers, and that with time inflammation may progress to fibrosis. Fibrosis is associated with the presence of CD8+ cells and some CD68+ cells in tissue, while at the serum level autoantibodies are associated with the presence of CD8+ cells, but fewer DSA than in inflammation. The association of lobular CD8+ cells with fibrosis underscores the importance of immunohistochemistry in analyzing the cellular infiltrate of PB. Indeed, while at a given time-point the presence of overall inflammation was not associated with concurrent fibrosis (any type), the presence of minimally increased numbers CD8+ T cells in the lobules (compared to normal biopsies) was associated with portal and centrolobular fibrosis. Without this detailed analysis of cellular subtypes, the significance of this very subtle CD8+ lobular infiltrate might have been overlooked and potential mechanistic implications missed. Finally, at the clinical level, the presence of CD8+ cells in the lobules of PB with fibrosis raises the question of the effectiveness of lower CNI trough levels in protecting the late allograft from cytotoxic injury.


Although to the best of our knowledge this is the first study to analyze the cellular infiltrate of PB in pediatric liver transplant recipients, it does present some limitations. First, clinical and surgical variables were retrieved in a retrospective manner. Second, cases spanning an extended time period were included, suggesting that changes in patient management may have occurred over time. Next, each cell type was analyzed and counted on a different slide leading to inherent variation in cell number and distribution. Macrophage subpopulations were not analyzed, hindering a more functional interpretation. Moreover, sample size may have impacted our ability to confirm associations observed in other reports, especially association of DSA and fibrosis or association between inflammation and fibrosis.4,31 For this reason, we did not perform extensive statistical analysis which can be plagued by a lack of power. Unfortunately, our sample size is too small to draw conclusions about the possible influence of CNI trough level on histological phenotype, but it is intriguing that PB with inflammation were taken at an earlier time point than biopsies with fibrosis, and that CNI trough levels are typically higher at earlier time points post-pLT, raising the question of the impact of CNI trough levels on inflammation. Finally, we did not include non-DSA anti-HLA or DSA IgG subclasses in the analysis.

In conclusion, PB of pediatric recipients displayed characteristic cellular signatures according to their histological phenotype. The majority of biopsies displayed inflammation with fibrosis and were characterized by abundant CD3+ and CD20+ cells, while CD68+ cells were more numerous in biopsies histologically considered as devoid of significant inflammation. Moreover, the presence of peripheral de novo DSA was associated with a denser portal infiltrate composed of CD3+, CD4+, CD8+, and CD20+ cells. In the lobule, the presence of lobular CD8+ cells was associated with fibrosis and the presence of circulating autoantibodies. How the cellular and humoral actors interact in inflammation and fibrosis is still unclear. What is increasingly clear, however, is that peripheral DSA and autoantibodies are associated with active liver injury in the quiescent allograft, whether inflammation or fibrosis, and that inflammation and fibrosis have different cellular signatures in PB of pediatric LT recipients. How the different cell types contribute to the chronic changes observed in quiescent allografts and how this may impact therapy remains to be elucidated. The routine use of immunohistochemistry to analyze cellular actors in PB should be considered to expand our understanding of histopathological findings.


We thank Simona Korff, PhD for data collection and Patrizia Bodignon and all the staff of the Division of Clinical Pathology and the Transplant Immunology Unit for their help in tissue processing and serum sample analysis.


1. Evans HM, Kelly DA, McKiernan PJ, et al. Progressive histological damage in liver allografts following pediatric liver transplantation. Hepatology. 2006; 43:1109–1117
2. Scheenstra R, Peeters PM, Verkade HJ, et al. Graft fibrosis after pediatric liver transplantation: ten years of follow-up. Hepatology. 2009; 49:880–886
3. Kelly D, Verkade HJ, Rajanayagam J, et al. Late graft hepatitis and fibrosis in pediatric liver allograft recipients: current concepts and future developments. Liver Transpl. 2016; 22:1593–1602
4. Varma S, Ambroise J, Komuta M, et al. Progressive fibrosis is driven by genetic predisposition, allo-immunity, and inflammation in pediatric liver transplant recipients. EBioMedicine. 2016; 9:346–355
5. Venturi C, Reding R, Quinones JA, et al. Relevance of activated hepatic stellate cells in predicting the development of pediatric liver allograft fibrosis. Liver Transpl. 2016; 22:822–829
6. Venturi C, Sempoux C, Bueno J, et al. Novel histologic scoring system for long-term allograft fibrosis after liver transplantation in children. Am J Transplant. 2012; 12:2986–2996
7. Campbell P. Clinical relevance of human leukocyte antigen antibodies in liver, heart, lung and intestine transplantation. Curr Opin Organ Transplant. 2013; 18:463–469
8. Chinen J, Buckley RH. Transplantation immunology: solid organ and bone marrow. J Allergy Clin Immunol. 2010; 1252 Suppl 2S324–S335
9. Demetris AJ, Bellamy CO, Gandhi CR, et al. Functional immune anatomy of the liver-as an allograft. Am J Transplant. 2016; 16:1653–1680
10. Taner T, Stegall MD, Heimbach JK. Antibody-mediated rejection in liver transplantation: current controversies and future directions. Liver Transpl. 2014; 20:514–527
11. Miyagawa-Hayashino A, Yoshizawa A, Uchida Y, et al. Progressive graft fibrosis and donor-specific human leukocyte antigen antibodies in pediatric late liver allografts. Liver Transpl. 2012; 18:1333–1342
12. Ohe H, Uchida Y, Yoshizawa A, et al. Association of anti-human leukocyte antigen and anti-angiotensin II type 1 receptor antibodies with liver allograft fibrosis after immunosuppression withdrawal. Transplantation. 2014; 98:1105–1111
13. Grabhorn E, Binder TM, Obrecht D, et al. Long-term clinical relevance of de novo donor-specific antibodies after pediatric liver transplantation. Transplantation. 2015; 99:1876–1881
14. Wozniak LJ, Hickey MJ, Venick RS, et al. Donor-specific HLA antibodies are associated with late allograft dysfunction after pediatric liver transplantation. Transplantation. 2015; 99:1416–1422
15. Tokodai K, Miyagi S, Nakanishi C, et al. Association of post-transplant donor-specific HLA antibody with liver graft fibrosis during long-term follow-up after pediatric liver transplantation. Pediatr Transplant. 2018; 22:e13169
16. Feng S, Demetris AJ, Spain KM, et al. Five-year histological and serological follow-up of operationally tolerant pediatric liver transplant recipients enrolled in WISP-R. Hepatology. 2017; 65:647–660
17. Feng S, Ekong UD, Lobritto SJ, et al. Complete immunosuppression withdrawal and subsequent allograft function among pediatric recipients of parental living donor liver transplants. JAMA. 2012; 307:283–293
18. Dezsőfi A, Baumann U, Dhawan A, et al.; ESPGHAN Hepatology CommitteeLiver biopsy in children: position paper of the ESPGHAN Hepatology Committee. J Pediatr Gastroenterol Nutr. 2015; 60:408–420
19. Demetris AJ, Bellamy C, Hübscher SG, et al. 2016 comprehensive update of the Banff Working Group on liver allograft pathology: introduction of antibody-mediated rejection. Am J Transplant. 2016; 16:2816–2835
20. Rappaport AM, Borowy ZJ, Lougheed WM, et al. Subdivision of hexagonal liver lobules into a structural and functional unit; role in hepatic physiology and pathology. Anat Rec. 1954; 119:11–33
21. Worthington JE, Robson AJ, Sheldon S, et al. A comparison of enzyme-linked immunoabsorbent assays and flow cytometry techniques for the detection of HLA specific antibodies. Hum Immunol. 2001; 62:1178–1184
22. Gebel HM, Harris SB, Zibari G, et al. Conundrums with flowPRA beads. Clin Transplant. 2002; 16Suppl 724–29
23. Kivelä JM, Kosola S, Peräsaari J, et al. Donor-specific antibodies after pediatric liver transplantation: a cross-sectional study of 50 patients. Transpl Int. 2016; 29:494–505
24. O’Leary JG, Kaneku H, Susskind BM, et al. High mean fluorescence intensity donor-specific anti-HLA antibodies associated with chronic rejection postliver transplant. Am J Transplant. 2011; 11:1868–1876
25. Tokodai K, Kawagishi N, Miyagi S, et al. The significance of screening for HLA antibodies in the long-term follow-up of pediatric liver transplant recipients. Transplant Proc. 2016; 48:1139–1141
26. Hirata Y, Yoshizawa A, Egawa H, et al. Impact of antibodies that react with liver tissue and donor-specific anti-HLA antibodies in pediatric idiopathic posttransplantation hepatitis. Transplantation. 2017; 101:1074–1083
27. Del Bello A, Congy-Jolivet N, Muscari F, et al. Prevalence, incidence and risk factors for donor-specific anti-HLA antibodies in maintenance liver transplant patients. Am J Transplant. 2014; 14:867–875
28. Levitsky J, Kaneku H, Jie C, et al. Donor-specific HLA antibodies in living versus deceased donor liver transplant recipients. Am J Transplant. 2016; 16:2437–2444
29. Tokodai K, Miyagi S, Nakanishi C, et al. Impact of the trough level of calcineurin inhibitor on the prevalence of donor-specific human leukocyte antigen antibodies during long-term follow-up after pediatric liver transplantation: antibody strength and complement-binding ability. Transplant Direct. 2017; 3:e196
30. Tokodai K, Miyagi S, Nakanishi C, et al. Effect of recipient age at liver transplantation on prevalence of post-transplant donor-specific HLA antibody. Ann Transplant. 2017; 22:333–337
31. Ekong UD, Melin-Aldana H, Seshadri R, et al. Graft histology characteristics in long-term survivors of pediatric liver transplantation. Liver Transpl. 2008; 14:1582–1587
32. Venturi C, Sempoux C, Quinones JA, et al. Dynamics of allograft fibrosis in pediatric liver transplantation. Am J Transplant. 2014; 14:1648–1656
33. Feng S, Bucuvalas JC, Demetris AJ, et al. Evidence of chronic allograft injury in liver biopsies from long-term pediatric recipients of liver transplants. Gastroenterology. 2018; 155:1838–1851.e7
34. Dao M, Habès D, Taupin JL, et al. Morphological characterization of chronic antibody-mediated rejection in ABO-identical or ABO-compatible pediatric liver graft recipients. Liver Transpl. 2018; 24:897–907
35. Kim PT, Demetris AJ, O’Leary JG. Prevention and treatment of liver allograft antibody-mediated rejection and the role of the ‘two-hit hypothesis’. Curr Opin Organ Transplant. 2016; 21:209–218
36. O’Leary JG, Demetris AJ, Friedman LS, et al. The role of donor-specific HLA alloantibodies in liver transplantation. Am J Transplant. 2014; 14:779–787
37. Heneghan MA, Yeoman AD, Verma S, et al. Autoimmune hepatitis. Lancet. 2013; 382:1433–1444
38. Liberal R, Mieli-Vergani G, Vergani D. Clinical significance of autoantibodies in autoimmune hepatitis. J Autoimmun. 2013; 46:17–24
39. Rajanayagam JK, Haller W, Debray D, et al. Serum autoantibodies are associated with chronic hepatitis and graft fibrosis after pediatric liver transplantation. Transplantation. 2016; 100Suppl 5SS79
40. Adams DH, Sanchez-Fueyo A, Samuel D. From immunosuppression to tolerance. J Hepatol. 2015; 621 SupplS170–S185
41. Demetris AJ, Isse K. Tissue biopsy monitoring of operational tolerance in liver allograft recipients. Curr Opin Organ Transplant. 2013; 18:345–353
42. Demetris AJ, Zeevi A, O’Leary JG. ABO-compatible liver allograft antibody-mediated rejection: an update. Curr Opin Organ Transplant. 2015; 20:314–324
43. Hidaka M, Eguchi S, Takatsuki M, et al. The Kupffer cell number affects the outcome of living donor liver transplantation from elderly donors. Transplant Direct. 2016; 2:e94

Supplemental Digital Content

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