In vitro adherence of lymphocytes to dermal endothelium under shear stress: implications in pathobiology and steroid therapy of acute cutaneous GVHD

Allogeneic hematopoietic stem cell (allo-HSC) transplantation is curative therapy for aplastic anemia and a variety of hematologic malignancies; however, significant treatment-related morbidity and mortality result from the development of acute graft-versus-host disease (aGVHD). Three tissues are principally involved in aGVHD reactions, the skin, liver, and gut,1,2with acute cutaneous GVHD (acGVHD) being the most frequent manifestation of this entity.3 The conventional initial therapy for aGVHD reactions is corticosteroids,3 4 but use of these broad immunosuppressive agents results in significant morbidity and mortality because of development of secondary infections. Thus, despite decades of clinical observations and of research on the pathobiology of aGVHD, more information is needed on the effector process(es) leading to tissue injury to develop more specific therapies to treat this condition.

Perivascular lymphocytic infiltrates are a hallmark of acGVHD and are the earliest feature of this entity.5 The infiltrating lymphocytes are of donor origin,6 and the composition reflects that of cells in circulation,7 indicating that the accumulation of effector cells results from emigration from the vasculature. The cascade of events leading to extravasation of all leukocytes is initiated by the attachment of circulating cells to endothelium with sufficient strength to overcome the physical forces (shear stress) of blood flow.8 These initial “rolling” contacts serve as a molecular braking mechanism, allowing the leukocytes to sample the local concentration(s) of chemokines and other inflammatory mediators, thereby leading to endothelial firm attachment and transmigration. This process is tightly regulated, and extensive in vivo and in vitro data indicate that tissue-specific migration of lymphocytes is regulated in part by up-regulation of surface adhesive structures on endothelium of target tissues.9-12 In aGVHD, lymphocytic infiltrates typically develop during periods of profound lymphopenia in the peri-engraftment period, indicating that the endothelial adhesive mechanism(s) mediating tissue-specific lymphocyte recruitment must be highly effective and efficient. Given this fact, it is surprising that virtually nothing is known about the endothelial change(s) directing the tissue tropisms of lymphocytes in aGVHD.

In this study, we performed Stamper-Woodruff assays on skin biopsy specimens of cutaneous eruptions occurring after auto- and allo-HSC transplantation to analyze the capacity of dermal endothelium to support lymphocyte binding under shear stress. We show here that dermal vessels in acGVHD reactions, but not in most other post-HSC transplantation skin eruptions, are specialized to support shear-resistant adherence of lymphocytes. This endothelial adhesive capability is modulated by steroid therapy for aGVHD. Moreover, antibody-blocking studies indicate that the shear-resistant adhesion molecules E-selectin, P-selectin, and vascular cell adhesion molecule–1 (VCAM-1), which are expressed on dermal endothelium and mediate recruitment of lymphocytes to skin,13 are not the primary effectors of this binding interaction. The results of these studies offer new insights into the effector limb of acGVHD reactions and provide a novel perspective on the pharmacology of steroid therapy in aGVHD reactions.

All patients gave written informed consent to participate in this study according to a protocol reviewed and approved by the Scientific Review Committee of the H. Lee Moffitt Cancer Center and the Institutional Review Board of the University of South Florida. Skin biopsy samples were obtained from 25 allogeneic hematopoietic stem cell transplant (allo-HSCT) recipients (16 men and 9 women, aged 20-61 years [median, 41 years]) and 22 auto-HSCT recipients (4 men and 18 women, aged 24-66 years [median, 45 years]). All auto-HSCT recipients were biopsied only once, but several allo-HSCT recipients underwent more than one biopsy (total of 40 evaluable allo-HSCT biopsies). For the allogeneic population, the transplantation preparative regimens used were busulfan/cyclophosphamide14 (BuCy2; n = 19), busulfan/cyclophosphamide/thiotepa15 (n = 1), cyclophosphamide/total body irradiation (TBI)16 (n = 1), TBI/etoposide17 (n = 3), and BuCy2 with antithymocyte globulin (30 mg/kg every day for 3 days; n = 1). All allo-HSCT recipients received related donor hematopoietic stem cells: 19 were HLA-matched and 6 had one locus mismatch at either A, B, or DR loci (4 mismatched at A locus, 1 mismatched at DR locus, and 1 at B locus). For acute GVHD prophylaxis, 24 patients received cyclosporin A/methotrexate,18 and 1 patient received cyclosporin A/methylprednisolone.19 For the auto-HSCT group, the transplantation preparative regimens were busulfan/cyclophosphamide/etoposide20 (n = 1), thiotepa/mitoxantrone/paclitaxel21 (n = 9), cyclophosphamide/thiotepa/carboplatin22 (n = 1), mitoxantrone/thiotepa21 (n = 3), ifosfamide/carboplatin/etoposide21 (n = 4), and BuCy2 (n = 4).

Punch biopsies (6 mm) were obtained from skin lesions developing within 100 days of HSC transplantation. All allo-HSCT recipients received stem cells collected from bone marrow. The stem cells used for auto-HSCT recipients were from peripheral blood or bone marrow, obtained using a variety of priming regimens as previously described21 23: granulocyte colony-stimulating factor (G-CSF)–primed bone marrow (n = 2), G-CSF–primed peripheral blood stem cells (PBSCs) (n = 3), G-CSF–primed bone marrow and PBSCs (n = 9), cyclophosphamide/etoposide and G-CSF–mobilized PBSCs (n = 7), and cyclophosphamide/paclitaxel and G-CSF–mobilized PBSCs (n = 1). Skin lesions developing after transplantation were typically generalized erythematous papular eruptions that covered less than 50% surface area at time of initial biopsy. All biopsies were obtained at sites of maximal involvement (with exception of face, palms, or soles), within 4 days of presentation. Specimens were divided in 2: one part was snap frozen in liquid nitrogen for adherence assays, and one part was placed in buffered formalin for histologic analysis. If an infectious process was suspected, a separate 2-mm punch biopsy was submitted for culture and sensitivity.

Histopathologic criteria for acGVHD on formalin-fixed hematoxylin/eosin-stained sections were based on those established by Lerner et al24 with minor modification.25 All histologic grades consist of superficial dermal perivascular mononuclear infiltrates and, additionally, are defined by each of the following: vacuolar alteration of the basal layer of the epidermis (histologic grade 1); vacuolization, spongiosis, and necrosis/dyskeratosis of basilar keratinocytes and/or follicular epithelium (histologic grade 2); formation of subepidermal clefts (histologic grade 3); and complete epidermal loss (histologic grade 4). For allo-HSCT recipients developing skin rashes early in the posttransplantation period (ie, within 20 days of receiving stem cells), a clinicohistologic diagnosis of acGVHD required that the skin eruption was progressing, despite discontinuation of potentially offending medications, in a pattern of distribution characteristic of acGVHD and, further, that either one of the following were also evident: (1) histopathologic changes (especially histologic grade 3) consistent with acGVHD or (2) concomitant clinical signs/symptoms of liver and/or intestinal aGVHD with histopathologic changes consistent with aGVHD involvement in either of these organs. If clinical criteria for aGVHD were not met on patients from whom skin biopsy at any time within 100 days after transplantation showed basilar vacuolization and/or keratinocyte necrosis, a clinicohistologic diagnosis of interface dermatitis was rendered, whereas a clinicohistopathologic diagnosis of morbilliform eruption was made if superficial dermal perivascular mononuclear cell infiltrates were evident without vacuolar changes in the basal layer. For diagnosis of G-CSF reaction, an increase in size and number of dermal macrophages was the predominant histopathologic finding.26 For all patients with acGVHD, the overall “clinicohistologic grade” was defined based on the conventional staging criteria of the extent of cutaneous surface area involvement: less than 25% involvement, grade I; 25% to 50% involvement, grade II; more than 50% involvement, grade III; bullae formation or desquamation, grade IV.

For both auto- and allo-HSCT recipients, medications that were temporally associated and deemed potentially causative of skin eruptions were discontinued at the time of skin biopsy. Repeat skin biopsies were performed only if the clinical distribution and/or appearance of the eruption changed after 3 days following initial biopsy, or if the patient was started on steroids and consented to rebiopsy. Thus, as determined by their clinical course, all auto-HSCT recipients underwent only 1 biopsy; however, 12 of the allo-HSCT recipients had repeat biopsies (3 of these patients had 3 biopsies). All biopsies were reviewed for histologic changes by 2 staff dermatopathologists, independently, without knowledge of the adherence assay results. The pertinent patient histories and histopathology findings are summarized in Tables 1, 2, and 3.

For allo-HSCT recipients with skin eruptions clinically consistent with acGVHD and covering more than 50% of total surface area, methylprednisolone therapy was initiated at 2 mg/kg per day (divided dose, every 6 hours). Biopsies were obtained from 6 patients before and after initiation of steroid therapy; in each instance, consecutive biopsies were obtained from involved skin adjacent to the initial biopsy.

Lymphocyte suspensions for lymphocyte-skin adherence assays consisted of human peripheral blood mononuclear cells (PBMCs) or rat thoracic duct lymphocytes (TDLs).27-29 Human PBMCs were isolated by Ficoll-Hypaque (Pharmacia Fine Chemicals, Piscataway, NJ) density gradient centrifugation of venous blood collected in sodium citrate. The interface cells were collected, washed 4 times in RPMI-1640 medium (Gibco-BRL, Grand Island, NY), and suspended at 10⁷ cells/mL in RPMI 1640 containing 5% fetal calf serum (FCS) for adherence assays. In some studies, PBMCs were separated into B-cell and T-cell (CD4⁺ and CD8⁺ subsets) fractions by negative selection using magnetic bead technology, as per manufacturer's recommendations (Miltenyi Biotec, Auburn, CA).

To obtain TDLs, the thoracic duct of Sprague-Dawley rats was cannulated, and draining lymph was collected at room temperature in phosphate-buffered saline (PBS) containing 5 U/mL heparin over 4-hour periods, beginning 12 hours following cannulation. Collected TDLs were washed 3 times in RPMI-1640 and resuspended at 10⁷ cells/mL in RPMI containing 5% FCS for adherence assays. Human PBMCs and rat TDLs both adhered to dermal endothelium with equal efficacy. For antibody-blocking studies, both sources of lymphocytes were used.

All patient samples were registered by using coded identifiers, assays were performed, and results were interpreted without knowledge of the patient's clinical history, condition, or pathology findings. The lymphocyte-skin adherence assay uses Stamper-Woodruff assay conditions originally described for investigating lymphocyte binding to lymph node high endothelial venules,30 with substitution of skin frozen sections for lymph node sections.10 For all adherence assays, frozen skin punch biopsy specimens were embedded in Tissue-Tek OCT (Miles, Elkhart, IN) compound for cryostat sectioning (8-μm sections). Sections were placed on glass slides and fixed in 3% glutaraldehyde in PBS. Following fixation, slides were rinsed in PBS and incubated in 0.2 M lysine monohydrochloride to block residual reactive carbonyl groups. Slides were then washed in RPMI 1640 medium containing 1% FCS and placed on trays. Cell suspensions (PBMCs or TDLs) were deposited as 0.2-mL aliquots (at a concentration of 10⁷ cells/mL) onto sections, and trays were placed on a rotating platform (80 rpm on a flat orbital shaker) at 4°C. After 30 minutes, sections were rinsed to remove nonadherent cells, fixed in 3% glutaraldehyde in PBS, and stained with methyl green-thionin. Lymphocyte binding to sections was analyzed by light microscopy. A minimum of 4 frozen sections from each patient was analyzed for adherence; sections that contained lymphocytes adherent as chains and clusters of more than 3 cells within discrete areas of the papillary dermis were scored as positive for adherence. With every assay performed on skin specimens, sections of rat lymph node served as control to confirm the capacity of input lymphocytes to bind vascular structures under Stamper-Woodruff assay conditions.

For all adherence assays, frozen tissue sections (8 μm) were cut sequentially, and alternating sections were either fixed in glutaraldehyde for adherence assays or fixed in acetone for immunohistochemical staining of endothelial structures. The location on tissue sections of bound cells following the in vitro adherence assay was compared with the immunohistochemical staining pattern of sequential sections to determine whether cells were adhering to endothelial structures. Immunohistochemistry was performed by using a 3-step alkaline phosphatase method (LSAB+ Kit; Dako, Carpinteria, CA). Primary antibody consisted of antifactor VIII (Dako) or anti-CD34 monoclonal antibodies (antibody HPCA-2; BD Pharmingen, San Jose, CA), followed by biotinylated secondary antibody and streptavidin-conjugated alkaline phosphatase according to manufacturer's recommendations (Dako). New Fuchsin was used as the chromagen, and sections were lightly counterstained with hematoxylin.

For antibody-blocking studies, adherence assays were performed on alternating sequential sections in the presence or absence of antibody. Results of antibody incubations were examined semiquantitatively, using an ocular grid to count bound lymphocytes within comparable consecutive areas of the dermis. Monoclonal antibodies known to block functional leukocyte–endothelial adhesion mediated by leukocyte proteins L-selectin (LAM1-3, human; HRL-1, rat) and leukocyte function–associated antigen–1 (LFA-1) and by endothelial proteins E-selectin, P-selectin, and VCAM-1 were obtained commercially; with the exception of antihuman L-selectin antibody LAM1-3 (Beckman-Coulter, Brea, CA), all other antibodies were from Becton Dickinson Pharmingen (San Jose, CA). Antibodies were added to cell suspensions and tissue sections in concentrations as high as 50 μg/mL. Controls to verify the blocking activity of input antibodies consisted of Stamper-Woodruff assays and parallel plate flow chamber studies of relevant adhesion molecules expressed on native cells or on stable transfectants as previously described.27-29 

Comparisons of groups were performed by using Chi-square and Fisher exact tests. All values of P presented are 2-tailed.

Human PBMCs obtained from healthy volunteers and rat TDLs adhered to discrete areas of the papillary/upper reticular dermis in punch biopsy specimens of a subset of cutaneous eruptions occurring following HSCT. A typical result from the adherence assay demonstrating the binding of lymphocytes is shown in Figure1. Lymphocyte adherence was evident within 15 minutes of incubation of cells with skin sections and maximal by 30 minutes. In each case in which adherence was observed, lymphocytes did not attach to epidermis or to structures within the deep reticular dermis. At input concentrations of 10⁷lymphocytes/mL for each cell source, the number of cells adherent and the patterns of adherence were similar among both rat TDLs and human PBMCs, and there were no differences observed in the capacity of isolated human B cells or CD4⁺ or CD8⁺ T cells to bind (results not shown); the adherent lymphocytes appeared as clumps or chains (Figure 1A), and the localization of the adherent lymphocytes within papillary/upper reticular dermal areas was traceable through consecutive sections of skin, indicating that adherence was to specific structures. Analysis of sequential sections comparing the location of adhering cells with the pattern of immunohistochemical staining for CD34 or factor VIII antigen revealed that, in all cases, human PBMCs and rat TDLs adhered only to endothelial structures in the papillary/upper reticular dermis (Figure 1B).

Results of lymphocyte adherence assays performed on skin specimens from patients undergoing single biopsies are shown in Table1. Within the group of patients receiving auto-HSCT (n = 22), binding was observed in only 2 specimens (UPIN 303, 352), each with the histologic diagnosis of morbilliform eruption. These specimens were from patients who had clinically extensive eruptions involving more than 75% of their body surface area, in each case resolving gradually (over 2 weeks) after discontinuation of antibiotics. Four other specimens with histopathology of morbilliform eruption were from auto-HSCT patients with localized eruptions, each of which resolved after several days after medication alterations. The remaining 16 specimens from auto-HSCT recipients demonstrated a variety of other histologic abnormalities as shown in Table 1.

The data for all 25 allo-HSCT recipients are summarized in Tables 1 and2, separated into those patients undergoing one biopsy only (Table 1) and patients undergoing multiple biopsies (27 biopsies obtained from 12 allo-HSCT recipients; Table 2). As shown in Tables 1 and 2, dermal papillary vascular structures from skin specimens from all patients with clinicohistologic diagnosis of acGVHD supported lymphocyte adherence under shear conditions, whereas adherence was not observed in biopsies from all allo-HSCT recipients (UPIN 55, 79, 85, 103, 119, 132, and first biopsies of 88 and 118) whose skin eruptions were not due to acGVHD. Collectively, the data in Tables 1 and 2 show that of 26 skin specimens from allo-HSCT recipients that supported lymphocyte adherence, 24 were obtained from eruptions of acGVHD; notably, the 2 patients whose skin biopsies showed lymphocyte adherence to dermal vessels with initial histopathology not consistent with GVHD (morbilliform eruption, UPIN 86 and 100) were diagnosed with acGVHD on rebiopsy (Table 2).

Within the group of patients undergoing repeat biopsies (Table 2), several biopsies were from patients who were receiving steroid therapy for aGVHD (n = 6, total 7 biopsies). Although the rash of acGVHD had not resolved at time of rebiopsy in any of these patients, 3 specimens (from 2 patients: UPIN 96 and UPIN 98) showed normal histology following initiation of steroids, whereas another 4 specimens (from 4 patients: UPIN 86, 114, 123, 126) had persistent histologic changes consistent with GVHD (Table 2). In 5 of the patients (UPIN 86, 96, 98, 114, 126), steroids were effective in treating acGVHD, and, in each case, lymphocyte binding to dermal vessels was abrogated prior to evidence of the clinical response (rash resolution). Conversely, the adherence assay of lesional skin from UPIN 123 obtained during steroid therapy showed lymphocyte binding to dermal papillary endothelium, and, notably, this patient had clinical progression of acGVHD while on steroids. Table 3 summarizes the results of adherence assays of all skin specimens obtained from all patients.

Blocking studies using monoclonal antibodies directed against adhesion proteins were performed on all sections that demonstrated lymphocyte adherence. Although the blocking monoclonal antibody used for these studies inhibited binding of respective receptor-ligand interactions in parallel plate flow chamber studies at input concentrations of 1 μg/mL (as previously described28 29), preincubation of lymphocytes with functional blocking anti–LFA-1 monoclonal antibody (clone G25.2; BD Pharmingen) and anti–L-selectin blocking antibodies (LAM1-3 for PBMCs and HRL-1 for rat TDLs) had no effect on adherence (Table4). Moreover, antibodies directed against the endothelial proteins E-selectin, P-selectin, and VCAM-1 had no effect on binding of lymphocytes to the dermal endothelium (Table4), suggesting that these proteins are not the primary effectors of the observed adhesive interaction.

The migration of lymphocytes from vascular to tissue compartments is a tightly regulated physiologic process that is fundamental to the development of both cellular and humoral immunologic responses. An essential prerequisite for the tissue-specific recruitment of lymphocytes from the vasculature is the initial shear-resistant adhesion of circulating cells to target tissue endothelium.8,9 The paradigm of this process is the migration of lymphocytes into lymph nodes, which requires the attachment of lymphocytes to specialized nodal vascular structures known as high endothelial venules (HEVs).9,30 The adhesive interaction between lymphocytes and HEVs is mediated principally by lymphocyte L-selectin (a membrane glycoprotein also known as the lymph node homing receptor) and specific ligands for L-selectin (known as addressins) that are constitutively expressed on HEVs.8,9The expression of addressins on HEVs allows leukocytes (including neutrophils) that express L-selectin to undergo shear-resistant “rolling” interactions on HEVs.31 However, selectivity in the firm adherence and transmigration of lymphocytes into the nodal parenchyma is achieved by unique chemokine receptor expression (CCR7) on lymphocytes that recognize the chemokine SLC that is constitutively expressed on HEVs.32 Neutrophils lack CCR7 and, therefore, cannot undergo G protein–coupled LFA-1 activation with subsequent firm adherence. Thus, in this model, selectivity in recruitment occurs beyond the expression of the addressins that mediate initial shear-resistant rolling adhesive interactions, but these rolling interactions are indispensable for subsequent extravasation. The extensive evidence that discrete endothelial membrane adhesion molecules regulate lymphocyte migration to lymphoid tissues has led to the notion that selectivity in lymphocyte trafficking to inflammatory sites results, in part, from expression of endothelial adhesion molecules that are inducible in an inflammation-specific manner.

The Stamper-Woodruff adherence assay used in our studies has been a useful tool for examining the molecular basis of lymphocyte migration and, in fact, allowed for the identification of the lymphocyte and endothelial structures that regulate lymphocyte migration into lymph node.30 This assay approximates physiologic interactions in that it measures, under shear conditions that mimic blood flow, the functional interaction of adhesion molecules within respective cellular lipid bilayers, ie, in native states on the lymphocyte and endothelium. This approach has provided evidence that specialized vascular changes promote lymphocyte migration into synovium in rheumatoid arthritis,12 and we have used it to examine the dermal endothelium in psoriasis, an autoimmune disease typified by extensive dermal lymphocytic infiltrates. In our previous investigations,10 we observed that dermal endothelium of normal skin does not support lymphocyte adherence, but dermal endothelium at sites of active psoriasis does. These studies suggested that lymphocyte migration to psoriatic skin is regulated by an inflammation-associated, tissue-specific receptor/ligand adhesive interaction.10 Although the precise receptor-ligand interactions regulating lymphocyte trafficking to psoriatic skin have yet to be identified, the data suggest that L-selectin–ligand interactions are not involved, but the integrin LFA-1 contributes to adhesion.11 33 

The data presented here provide evidence of a papillary dermal endothelium recognition system that promotes lymphocyte migration into skin in aGVHD. The finding that anti–L-selectin and anti–LFA-1 blocking monoclonal antibodies did not affect PBMC adherence to dermal endothelium in acGVHD suggests that the receptor/ligand system promoting this interaction is separate from that directing lymphocyte trafficking to lymphoid tissues and into psoriatic skin, respectively. Beyond this fact, increased expression of vascular adhesion proteins E-selectin and VCAM-1 is observed in cutaneous inflammatory processes13,34,35 and, in particular, in acGVHD.36 Indeed, rather than the inducible pattern of expression characteristic for E- and P-selectin, dermal microvascular cells constitutively express these adhesion molecules.13 37 However, neither E-selectin, P-selectin, nor VCAM-1 appear to play central roles in the observed PBMC binding to dermal endothelium of acGVHD under Stamper-Woodruff assay conditions.

In our studies, PBMCs from transplant recipients themselves were not used in adherence assays to their own skin or to that of other patients, as patients were typically profoundly anemic and lymphopenic at the time of biopsy. However, it should be noted that B cells and CD4⁺ and CD8⁺ T cells obtained from peripheral blood of healthy volunteers each bound with equal efficacy. The finding that B cells bind equivalently to T cells excludes the possibility that the observed adhesive interaction requires allorecognition of endothelial major histocompatibility complex (MHC) antigens as has been described in nonshear (static) adhesion assays using primed T cells and interferon-treated endothelial cells in culture.38 39 Moreover, lymphocytes from rat thoracic duct bind reproducibly to the dermal endothelium, indicating that lymphocyte membrane determinants that recognize the adhesive ligand(s) expressed on endothelium of acGVHD lesions are phylogenetically conserved. Thus, lymphocytes capable of recognizing these ligand(s) appear to circulate naturally in both humans and rodents, suggesting that PBMCs circulating in the posttransplantation setting have similar capabilities. Taken together, these data indicate that changes in the endothelium of the tissue confer the specificity for lymphocyte adherence.

The capacity of dermal endothelium to support lymphocyte binding does not appear to be a consequence of the type of transplantation conditioning or of the stem cells received, as assay results were independent of the various conditioning and stem cell mobilization regimens used. The endothelial change(s) instead appear to be related intimately with the development of acGVHD (Table 3) and correlate closely with clinical outcome following initiation of steroids for therapy of GVHD (Table 2). Because both patients initially without acGVHD whose skin biopsies demonstrated adherence subsequently developed acGVHD, a positive adherence assay result may be predictive of acGVHD. Although these results do not supplant the use of other available in vitro assays such as the skin explant model40,41 that can predict the occurrence of aGVHD, the capability to perform ongoing, dynamic assessment(s) after transplantation may offer insights on the course of GVHD and thereby assist clinical decision making. The finding that a negative adherence assay result following initiation of steroids correlates with clinical resolution of acGVHD suggests that the therapeutic effect of steroids may be mediated, at least in part, by down-regulation of relevant endothelial adhesion molecules promoting the migration of effector cells into the target tissue. In fact, there is evidence from in vitro studies that expression of inflammatory adhesion proteins on human endothelial cells is decreased by steroids.42 

Within the group that had auto-HSC transplantation, biopsies of cutaneous lesions in 2 of 22 patients showed binding in the adherence assay. Each of these specimens had histopathologic diagnosis of morbilliform eruption. Both patients from whom these specimens were obtained had intense erythematous papular eruptions that ultimately covered more than 75% of body surface, yet these skin changes gradually resolved without steroid therapy. The patients had different transplantation conditioning regimens (busulfan/cyclophosphamide [UPIN 303] and thiotepa/mitoxantrone/paclitaxel [UPIN 352]). A syndrome of autologous GVHD is recognized in both animal models and human HSCT recipients, which is potentiated by the use of cyclosporin A,43 44 although neither of these patients received cyclosporin A. Further studies will be needed to determine whether lymphocyte adherence to skin eruptions in auto-HSCT recipients is related to the syndrome of autologous acGVHD or represents a heightened endothelial reaction to the conditioning regimen or to drugs administered after transplantation. In any case, identification of the precise molecular mechanism(s) mediating the enhanced endothelial capacity to support lymphocyte binding under shear may provide an approach to aid in diagnosing acGVHD and to monitor response to therapy.

Although the relevant endothelial molecule(s) still remains to be fully clarified, the results of this study contribute important new insights into the pathobiology of acGVHD: our findings provide first evidence that up-regulated expression of shear-resistant adhesion molecules within discrete endothelial sites mediate or facilitate the tissue tropism of aGVHD effector cells. The pathogenesis of aGVHD involves the migration of both alloreactive lymphocytes and of natural killer (NK) cells into target tissues,45-49 and each has been implicated as etiologic in the inflammatory changes. The mononuclear cell infiltrates of aGVHD are composed of donor-derived cells,6,50,51 indicating that cellular infiltrates arise from recruitment of circulating cells. These observations suggest that there are specific PBMC-endothelial adhesive mechanisms that promote directed migration of effector cells, including lymphocytes and NK cells, into sites of aGVHD. The heightened migration of such cells into the dermis may represent an augmented, pathobiologic response of a physiologic mononuclear cell migration pattern that contributes to the development of the skin-associated lymphoid tissues.52 It is also possible that endothelial changes in the skin, liver, and gastrointestinal tract may be the common thread for the tissue selectivity observed in aGVHD reactions; specific endothelial ligands may be expressed in these tissues disproportionately that promote entry and accumulation of effector cells into these sites, leading to pathologic damage. It is tempting to propose that similar endothelial ligands are expressed in all 3 target tissues in aGVHD, and future studies will examine whether endothelial structures of gut or liver involved in aGVHD reactions support lymphocyte adherence as is demonstrated here for skin. Further characterization of the lymphocyte and endothelial adhesion molecules that mediate selective entry of effector cells into organs affected by aGVHD should allow an inflammatory-specific, molecular approach to treating aGVHD, thereby avoiding use of broad immunosuppressives that increase the risk of opportunistic infections and of manipulations such as T-cell depletion that result in higher graft failure rates and diminish the effectiveness of the graft-versus-leukemia effect.4 53 

We thank Dr Ling Fu for expert technical assistance essential to the execution of this study.