Context.—Kaposi sarcoma (KS) is a vascular tumor frequently associated with advanced human immunodeficiency virus infection, advanced age, or iatrogenic immunosuppression. Immunohistochemistry for CD31 and CD34, and more recently for FLI1 and D2-40, has been used as ancillary diagnostic tests for KS, despite little information regarding the sensitivities and differential staining patterns of the latter 2 markers in the major clinical subtypes and histologic stages of KS.
Objective.—This retrospective study aims to assess the prevalence of the vascular markers D2-40 and FLI1 in the main clinical subgroups and tumor stages of KS.
Design.—Twenty-four cases of KS (12 acquired immunodeficiency syndrome [AIDS]–related cases and 12 non–AIDS-related cases; 11 nodular-stage and 13 patch/plaque–stage KS) were stained for CD34, CD31, D2-40, and FLI1 by immunohistochemistry. The distribution of immunoreactivity was compared between the clinical subtypes and tumor stages of KS using the Mann-Whitney test.
Results.—CD31, CD34, D2-40, and FLI1 strongly and diffusely stained tumor cells in 75%, 92%, 67%, and 92% of AIDS-related cases and 58%, 92%, 67%, and 75% of non–AIDS-related cases, respectively. Differences in the proportions of positive cases between AIDS-related and non–AIDS-related cases did not reach statistical significance. No significant staining differences were observed between nodular- and patch/plaque–stage KS either.
Conclusions.—There are no differences in the distribution of immunohistochemical reactivity for CD31, CD34, D2-40, or FLI1 between AIDS-related and non–AIDS-related KS or between nodular- and patch/plaque–stage KS. All of the markers studied demonstrated high sensitivity in both clinical settings and both stages of tumor progression.
Kaposi sarcoma (KS) is an uncommon vascular tumor caused by human herpesvirus 8 (HHV8).1 Although initially described in 1872 by Moritz Kaposi,2 KS became widely known as one of the defining illnesses of acquired immunodeficiency syndrome (AIDS) during the 1980s. It is now known that KS arises in 1 of 4 distinct clinical settings. The epidemic form, also referred to as the AIDS-related form, is the clinical subtype that arises in immunocompromised human immunodeficiency virus–infected patients. The iatrogenic form, also known as transplant-related KS, is a complication of immunosuppressive therapy and generally occurs in solid organ transplant patients. The sporadic or classic form typically arises in elderly patients, often in those of Eastern European descent. Finally, the endemic variant of KS is the form observed in human immunodeficiency virus–negative patients from sub-Saharan Africa.3 In the United States, KS is at least 20 000 times more common in persons with AIDS than in the general population and 300 times more common than in other immunosuppressed groups.4
The histologic differential diagnosis of KS includes other vascular tumors such as benign lymphangioendothelioma, hobnail hemangioma, spindle cell hemangioma, kaposiform hemangioendothelioma, cutaneous angiosarcoma, and acroangiodermatitis.5 When arising in the gastrointestinal tract, KS may be difficult to distinguish from gastrointestinal stromal tumor, as both are positive for c-kit (CD117).6,7 Immunohistochemical (IHC) studies are useful in confirming the vascular nature of the neoplasm and in documenting the presence of the pathogenic HHV8 virus, the presence of which, among other differential diagnostic considerations, distinguishes KS from other vascular tumors.
Monoclonal antibodies directed against CD31 (platelet/endothelial cell adhesion molecule, PECAM1) and CD34 (a hematopoietic progenitor cell surface protein) are relatively sensitive and specific markers of endothelial differentiation that are expressed in the majority of vascular tumors.8–10 Expression of CD34 in a variety of mesenchymal tumors and hematopoietic cells limits its specificity for endothelial cells, however.8 D2-40 is a monoclonal antibody directed against podoplanin, a cell surface sialoglycoprotein expressed in lymphatic endothelium and select epithelia.11–13 It is detectable by IHC in lymphatic malformations, a subset of angiosarcomas, and other vascular tumors such as kaposiform hemangioendothelioma.14,15 It is not specific for vascular neoplasms, however, as it is also expressed in mesothelioma and some carcinomas, particularly primary cutaneous carcinomas.16,17 The FLI1 (Friend leukemia virus integration 1) protein is a member of the ETS family of DNA-binding transcription factors. Besides staining normal endothelial cells and vascular tumors, nuclear expression is also seen in Ewing sarcoma/primitive neuroectodermal tumor and lymphoblastic lymphoma.18–20
In 2004, commercially available monoclonal antibodies directed against HHV8 became available for IHC on formalin-fixed, paraffin-embedded tissue. Subsequent studies have confirmed that one such clone directed against the virus latency–associated nuclear antigen shows high sensitivity and specificity for all clinical subtypes and tumor stages of KS.21,22 Similarly, previous studies have documented CD31 and CD34 expression in most cases of KS, regardless of clinical subtype or tumor stage.10 However, the proportion of nodular stage KS positive for CD31 is slightly less than that of patch/plaque–stage KS, suggesting that there may be subtle differences in the staining patterns for endothelial markers between the different histologic stages of KS.10 In particular, the prevalence of comparatively new endothelial markers D2-40 and FLI1 has not been thoroughly examined among the different clinical subtypes and tumor stages of KS. This retrospective study aims to assess the prevalence of the vascular markers D2-40 and FLI1 in the main clinical subgroups and tumor stages of KS.
MATERIALS AND METHODS
The surgical pathology archives at our institution were searched for cases diagnosed as KS between 1980 and 2009. Only cases with adequate tissue remaining in the paraffin tissue block for IHC were included. Archived hematoxylin–eosin glass slides were reviewed by 2 of the authors to confirm the diagnosis and select blocks for IHC. Medical records were reviewed to extract selected clinical information for clinical subtype classification (classic, AIDS-related, iatrogenic, or endemic). Human immunodeficiency virus status was not available for 6 non–AIDS-related cases; therefore, some cases were subclassified on clinical criteria alone. Because the number of non–AIDS-related cases was limiting (7 classic, 5 iatrogenic, and no endemic cases), we grouped these 12 cases together and randomly selected an equal number of cases from the AIDS-related group for this study. The Institutional Review Board at Vanderbilt University, Nashville, Tennessee, approved the study protocol.
Immunohistochemistry for HHV8, CD31, CD34, D2-40, and FLI1 was performed on formalin-fixed, paraffin-embedded tissue sections using the antibodies and protocols detailed in Table 1. Specific primary antibody binding was detected using EnVision+ (Dako North America, Carpinteria, California) with 3,3′-diaminobenzidine tetrahydrochloride as chromogen. Immunohistochemistry slides were evaluated at ×40 magnification and scored independently by 2 of the authors. For CD31, CD34, and D2-40, diffuse and strong membranous staining was scored as ++, diffuse but weak or equivocal staining was scored as +, and no staining above background was scored as −. For FLI1, positive staining in greater than or equal to 50% of tumor nuclei was scored as ++, less than 50% was scored as +, and negative staining was scored as −. There were no discrepancies in the interpretation of IHC scores between the 2 observers. Differences in the distribution of staining intensities between clinical groups and tumor stage were assessed using the Mann-Whitney test and differences in sensitivity were assessed using the Fisher exact test (GraphPad Prism v5.0, GraphPad Software, La Jolla, California).
In all 24 study cases, IHC for HHV8 was performed to confirm the diagnosis (data not shown). Of the 12 AIDS-related KS cases, 8 were cutaneous in origin and the remainder involved the oropharyngeal cavity. The 12 non–AIDS-related cases included 7 classic KS (age range, 64–84 years; median, 75 years), all of which were cutaneous. Iatrogenic KS comprised the remaining 5 non–AIDS-related cases. Of these, 2 were cutaneous, 2 involved soft tissue, and 1 arose in the colon of a patient on long-term corticosteroid therapy for ulcerative colitis. Four of the 5 iatrogenic cases were confirmed to be negative for human immunodeficiency virus infection. There was no association between clinical subgroup (AIDS-related and non–AIDS-related) and histologic tumor stage (Fisher exact test, P >.99).
IHC stains for CD34, CD31, D2-40, and FLI1 were positive in all 12 AIDS-related cases of KS. In the non–AIDS-related group, 2 cases were negative for CD31 and 1 case did not stain for D2-40 (Table 2). Strong and diffuse staining (++) of tumor cells for CD31, CD34, and D2-40 was observed in 75%, 92%, and 67% of AIDS-related cases and 58%, 92%, and 67% of non–AIDS-related cases, respectively (Figure 1, A through C). Nuclear FLI1 staining was diffuse (>50% of cells) in 11 cases (92%) of AIDS-related KS and 9 cases (75%) of non–AIDS-related KS (Figure 2, A through C). Overall, the sensitivity of each diagnostic marker examined was high (CD34, 100%; CD31, 92%; D2-40 96%; FLI1, 100%).
All diagnostic markers tested showed high sensitivity regardless of tumor stage or clinical subtype (Table 3). Although only 55% of nodular stage KS stained strongly and diffusely for CD31 and D2-40, these proportions were not significantly different from patch/plaque–stage KS (77% for both markers).
Statistical analyses disclosed no significant differences in the distributions of immunoreactivity or sensitivities between the clinical subgroups for any of the IHC markers evaluated. Similarly, no differences in staining distributions in any marker were observed between nodular- and patch/plaque–stage KS.
Immunohistochemical detection of HHV8, CD31, and CD34 in KS has been extensively studied.8–10 Fewer reports address the diagnostic utility of relatively new endothelial cell markers such as D2-40 and FLI1 in KS, particularly with regard to potential differences in immunoreactivity among the different clinical subtypes and tumor stages.11,12,18 We aimed to evaluate the prevalence of D2-40 and FLI1 expression in KS in comparison with more conventional endothelial markers (CD31 and CD34), as well as to ascertain whether there are differences in immunoreactivity among the different clinical subtypes and tumor stages of KS.
CD34 stained all 24 cases examined, whereas 2 cases of non–AIDS-related KS were negative for CD31, both the classic subtype. One case was at nodular stage and the other at plaque/patch stage. Despite a prior report of reduced CD31 expression with progression from early-stage to nodular-stage KS,10 we did not observe a decreased proportion of CD31-positive nodular KS. D2-40 also showed high sensitivity (overall, 96%), with only 1 negative non–AIDS-related tumor. This case was a plaque stage of KS, classic form. Previous studies have also found that most KSs of different histologic stages are positive for D2-40.11,12 Nuclear expression of FLI1 was observed in all study cases (sensitivity, 100%), most of which (83%) showed diffuse reactivity. In contrast to the other markers studied, FLI1 is a nuclear antigen and, as such, is less susceptible to histologic artifacts, especially in early patch-stage KS, when the number of tumor cells is minimal.18
In summary, our results suggest that the differences in the distribution of immunoreactivity for CD31, CD34, D2-40, and FLI1 between AIDS-related KS and non–AIDS-related KS and between patch/plaque- and nodular-stage KS are not significant. Most markers evaluated showed high sensitivity (>90%) regardless of the clinical setting and tumor stage; the lowest sensitivity of any subgroup was for CD31 in non–AIDS-related cases (83%). We conclude that D2-40 and FLI1 are useful diagnostic markers for both patch/plaque- and nodular-stage KS of AIDS-related, classic, and iatrogenic subtypes.
From the Department of Pathology and Laboratory Medicine, Vanderbilt University Medical Center, Nashville, Tennessee.
The authors have no relevant financial interest in the products or companies described in this article.
Presented at the XXVIIIth International Academy of Pathology meeting; October 12th, 2010; Sao Paulo, Brazil.