Nonclassic Histologic Variants of Kaposi Sarcoma With Nonspecific Endoscopic Inflammatory Patterns: A Near-Miss Diagnosis Lookout in Gastrointestinal Tract Biopsies

Kaposi sarcoma (KS), a low-grade angioproliferative neoplasm caused by human herpesvirus 8 (HHV8) infection, is known to affect patients who are immunosuppressed, with AIDS-associated KS being the most common form of KS in the United States.^1–3 The less prevalent non–AIDS-associated epidemiologic subtypes include classic, African (endemic), and iatrogenic (or transplant-associated) KS.^2,4 The most widely recognized and reported characteristic histopathologic features of KS include fascicles of spindle cells with minimal atypia, slitlike spaces containing extravasated erythrocytes, hyaline globules, and hemosiderin deposition.^2,4 The histologic spectra of cutaneous KS has, however, widened over time to include other nonclassic variants that have been variably described as telangiectatic, intravascular, angiosarcoma-like, lymphangioma-like, micronodular, keloidal, pyogenic granuloma–like, etc.^5–9

The gastrointestinal (GI) tract is the most common extracutaneous site for KS. A GI-based diagnosis of KS frequently precedes a cutaneous KS diagnosis and can be observed in approximately half of patients with AIDS.^10–16 Additionally, patients with GI-KS may be asymptomatic or present with vague nonspecific symptoms such as abdominal pain, nausea, vomiting, diarrhea, and hemorrhage, with the GI endoscopic examination and biopsy reported to have a high false-negative rate (potentially reported to be about 35%–85%) for KS.^{10,11,17–19} Despite this, there are very limited data characterizing histologic variants at various GI sites, and the histomorphologic spectrum of GI-KS is not nearly as well understood when compared with its cutaneous counterparts. A recent study by Zheng et al²⁰ was potentially the first to document the diverse histologic presentations of KS in the GI tract. Notably, they identified 7 previously unreported variants of GI-KS, namely lymphangioma/lymphangiectatic-like, mucosal hemorrhage/telangiectatic-like, mucosal inflammation–like, granulation tissue–like, mucosal prolapse–like, GI stromal tumor (GIST)–like, and inflammatory myofibroblastic tumor (IMT)–like.²⁰

Given the paucity of studies discussing the histologic variants of KS in the GI tract coupled with reportedly high false-negative rates of endoscopic biopsies for the detection of GI-KS, we aimed to present our single-institution large case analysis to validate and further elucidate nonclassic histologic variants of KS in the GI tract. We also correlated each histologic variant with endoscopic findings to explore potential correlates between endoscopic patterns and histologic subtypes.

MATERIALS AND METHODS

This study was approved by our medical center’s institutional review board. A retrospective pathology database search was performed from 2016 to 2024 to retrieve cases of luminal GI biopsies diagnosed as KS. Resection specimens were excluded for the purpose of this study. Archival tissue, clinical history, and endoscopic findings were extracted from electronic medical records in accordance with the Health Insurance Portability and Accountability Act.

A total of 44 luminal GI biopsies of KS from 28 patients were retrieved and reviewed. Hematoxylin-eosin (H&E)–stained sections, HHV-8 immunostains, and all other available ancillary stains were retrieved and rereviewed by 2 pathologists for all cases. Any coexisting pathology was also reviewed and recorded.

RESULTS

Histologic Features and Histologic-Endoscopic Correlation

We classified the histologic features seen in our series of 44 biopsies broadly into classic (conventional) pattern and nonclassic (nonconventional) histologic patterns.

Classic (Conventional) Histologic Pattern

The classic histology of KS (16 of 44 biopsies; 36%) was characterized by sheets and fascicles of spindled cells that showed minimal to mild atypia along with slitlike spaces containing extravasated erythrocytes (Figure 1, A). Biopsies that showed any proportion of the aforementioned histology was grouped into this classic/conventional pattern given the widespread recognition of this pattern among pathologists. The majority of the cases with this classic (conventional) histologic pattern presented endoscopically as a nodule (9 of 16; 56%) (Table 1). This pattern was seen as an exclusive pattern in only 5 of 16 biopsies (31%) and was seen in association with one or more nonclassic (nonconventional) histology patterns in the remaining 11 cases (69%) (Table 2).

View Figure

• Download as Powerpoint
• Download Figure

Table 1.Endoscopic Findings Correlating With Classic (Conventional) and Nonclassic (Nonconventional) Histology Patterns and Sites of Involvement in 44 Gastrointestinal Mucosal Biopsies of Kaposi Sarcoma

View Fullscreen

Table 2.Primary (Predominant) and Secondary Histologic Patterns in 44 Gastrointestinal Mucosal Biopsies of Kaposi Sarcoma

View Fullscreen

Nonclassic (Nonconventional) Histologic Patterns

The majority of KS in our luminal GI biopsies presented exclusively as nonclassic histologic pattern(s) (28 of 44 biopsies; 64%). These cases did not show any classic (conventional) histology pattern on these superficial mucosal biopsies. The most common endoscopic correlate for the biopsies with nonclassic histologic patterns was a nonspecific inflammatory pattern seen on endoscopic examination (ie, erythema, ulcer, edema, or friability) in 36% (10 of 28) of biopsies raising clinical concern for an infectious etiology. The other varied endoscopic findings seen in association with nonclassic histologic patterns are listed in Table 1. Of note, the majority (91%; 10 of 11) of cases presenting as a nonspecific inflammatory pattern on endoscopy showed nonclassic (nonconventional) patterns on histologic examination.

We divided the nonclassic (nonconventional) histologic patterns seen in our biopsies into 5 different histologic patterns, paralleling the study by Zheng et al.²⁰ The biopsies were subcategorized into these 5 patterns based on the predominant subtype of nonclassic histologic pattern identified in that biopsy (Table 3). As mentioned above, any case with any proportion of classic histology was included under the classic (conventional) histology cases. The 5 nonclassic (nonconventional) histologic patterns encountered in our study were as follows.

Table 3.Summary of Characteristic Histologic Features and Histologic Differential Diagnoses for Nonclassic (Nonconventional) Histomorphologic Variants Seen in Gastrointestinal Mucosal Biopsies With Kaposi Sarcoma

View Fullscreen

Mucosal Hemorrhage–Like Pattern

This variant constituted 36% (10 of 28) of cases with a nonclassic histology pattern and 23% (10 of 44) of overall cases. The histologic features of this variant were prominent mucosal hemorrhage with variably associated conspicuous extravasated red blood cells and/or blood lakes (Figure 1, B). Additionally, there was evidence of associated mucosal injury and reactive epithelial changes in the adjacent mucosa. The prominent hemorrhage obscured the lesional cells, raising considerations for procedural hemorrhage or nonspecific hemorrhage and a potential for missed diagnosis. However, the lesional cells were readily apparent on HHV8 immunostain in all cases. Half of the biopsies with this histology exhibited it as the exclusive histologic pattern, and the remaining biopsies were associated with other nonclassic histologic forms presenting as secondary patterns, detailed in Table 2.

Mucosal Prolapse–Like Pattern

This variant constituted 21% (6 of 28) of cases with nonclassic histology and 14% (6 of 44) of overall cases. This pattern was almost always seen in combination with other nonclassic histologic patterns, listed in Table 2. The observed pattern displayed bland stromal or smooth muscle–like cells that extended into the superficial lamina propria, infiltrating the spaces between and encasing the glands or crypts. A subtle, swirling arrangement of these cells, suggestive of fibromuscular hyperplasia, produced a prolapse-type appearance, highlighting a potential diagnostic pitfall (Figure 1, C). However, these cells were readily apparent on HHV8 immunostain in all cases.

Mucosal Inflammation–Like Pattern

This variant constituted 18% (5 of 28) of cases with nonclassic histology and 11% (5 of 44) of overall cases. Histologically, this pattern presented as a busy and cellular-appearing lamina propria secondary to brisk lymphoplasmacytic infiltrates in the lamina propria. Interspersed lesional cells were masked by the prominent inflammatory type infiltrates and/or in part mimicked the inflammatory cells, and were seen on the HHV8 immunostain (Figure 1, D). This histologic pattern raised a consideration for infectious or other inflammatory etiologies and a potential for misdiagnosis. This pattern was also almost always seen in combination with other nonclassic histology patterns (Table 2).

Granulation Tissue–Like Pattern

This variant constituted 18% (5 of 28) of cases with nonclassic histology and 11% (5 of 44) of overall cases. Histologically, this pattern recapitulated granulation tissue with inflammatory cells, fibroblasts, inconspicuous thin-walled capillaries, and reactive endothelial cells. The interspersed lesional cells mimicking one or more of the aforementioned cell populations were readily identified on the HHV8 immunostain (Figure 1, E). This was present as an exclusive nonclassic histology pattern in 2 of the 5 biopsies (40%), and it was seen in association with other nonclassic histology pattern in the remaining 3 (60%) (Table 2).

Dilated Vascular- or Lymphatic-Like Pattern

This variant was the least common nonclassic histology pattern in our study and constituted only 7% (2 of 28) of cases with nonclassic histology and 5% (2 of 44) of overall cases. This variant featured irregular dilated-appearing and focally blood-filled vascular- or lymphatic-like spaces lined by flattened endothelial-like cells but showed conspicuous immunoreactivity on HHV-8 immunostain (Figure 1, F). This was seen as the only pattern in one biopsy, and in another biopsy it was seen in association with other nonclassic histology patterns (Table 2).

Other Histologic Features

The most common unifying feature of the biopsies showing the nonclassic (nonconventional) histologic patterns was the presence of non–spindle cell morphology or subtle proliferation of lesional cells mimicking other benign cells such as smooth muscle cells, stromal cells, endothelial cells, or inflammatory cells on routine H&E slides, raising a differential diagnosis with other benign reactive or inflammatory lesions (Table 3). Many cases were so subtle and near normal that the lesional cells were highlighted only on HHV8 immunostain (Figure 2, A through D). The majority (71%; 20 of 28) of biopsies with nonclassic histology showed more than one nonclassic pattern on biopsy, that is, the primary (most predominant) pattern and secondary (less common) pattern(s) (Table 2; Figure 3, A through F). HHV8 immunostain was performed on all 44 biopsies and was positive. The lesional tissue in the biopsy cases varied from 1 to 3.5 mm in greatest extent, with a mean size of 2.1 mm as measured on HHV8 immunostain in all cases. All cases (including all cases with nonclassic histology) showed at least some extravasated red blood cells in the lesional tissue on reevaluation of cases. The other ancillary stains performed were CD31 immunostain (in 14% of biopsies; 6 of 44) and CD34 immunostain (in 16% of biopsies; 7 of 44) and these were positive in all cases performed.

View Figure

• Download as Powerpoint
• Download Figure

View Figure

• Download as Powerpoint
• Download Figure

DISCUSSION

The diagnosis of GI-KS is important to make because therapy depends upon the extent of disease, and an early diagnosis can lower morbidity and mortality.^17,21,22 Existing data suggest that GI tract involvement by KS is asymptomatic in more than 80% of patients, can occur in the absence of cutaneous disease, and hence is probably underdiagnosed.^{10,21,23–25} Patients with GI involvement have a 2 to 5 times higher risk of death and benefit from chemotherapy to improve survival.^23,26,27 Dual treatment with highly active retroviral therapy and chemotherapy (liposomal doxorubicin therapy) has been associated with improved mortality and morbidity in AIDS patients with KS involving the GI tract.^21,22,24,28 Patients with CD4⁺ T-cell counts less than 100 cells/μL, men who have sex with men, high HIV ribonucleic acid viral load, and no highly active retroviral therapy treatment have been reported to be strong predictors of GI-KS.^21,29 Hence, performing screening endoscopies with biopsy in select patients has been encouraged for early detection and treatment of GI-KS to improve patient outcomes, even in the absence of cutaneous disease, and in fact is a standard practice in Japan.^21,29,30 A variety of presentations for KS on GI endoscopy has been described in the literature, such as flat macules, patches and papules, nodules, masses, small polypoid excrescences, ulcerations, and large bulky tumors.^{10,13,18,30–32} Based on our endoscopy reports, we similarly divided the endoscopic patterns in our cohort into nodules, papules, masses, polyps, and plaquelike lesions. Additionally, endoscopic findings in many of our cases revealed descriptions such as erythema, friability, edema, or nonspecific ulcer, many times raising a possibility of infectious etiologies on endoscopic reports. We combined these patterns under the category of nonspecific inflammatory endoscopic appearance, and these constituted a quarter of our cases.

Biopsy remains the gold standard for the diagnosis of KS.^10,17,18,33 We found that the majority (64%; n = 28) of KS cases in luminal GI mucosal biopsies in our study presented exclusively as nonclassic (nonconventional) pattern(s) on histologic evaluation. None of these 28 biopsies showed any coexisting classic (conventional) histology pattern. The remaining 36% of cases did show classic histology; however, only about 30% of those cases showed an exclusive classic histology pattern, with the remaining 70% depicting a nonclassic histology pattern coexisting with the classic histology pattern. Overall, the cases with a pure or exclusive classic histology pattern constituted just 11% of the total cases. The various nonclassic or nonconventional histologic patterns have been widely described in the literature at cutaneous sites. However, the literature on histologic variability of KS in the GI tract is relatively scarce, and these nonclassic histologic variants were not in fact formally described in the literature until recently by Zheng et al²⁰ in a multi-institutional study wherein they divided their cohort of 46 cases (43 superficial mucosal biopsies and 3 resections) into varied histologic variants of KS. The 7 nonconventional histologic variants in their study included lymphangioma/lymphangiectatic-like, mucosal hemorrhage/telangiectatic-like, mucosal inflammation–like, granulation tissue–like, mucosal prolapse–like, GIST-like, and IMT-like.²⁰ Somewhat similarly to the study by Zheng et al,²⁰ we divided our nonclassic histology patterns into 5 different morphologic subtypes, all showing the presence of non–spindle cell morphology or subtle proliferation of lesional cells mimicking other benign cells such as smooth muscle cells, stromal cells, endothelial cells, or inflammatory cells, raising a differential diagnosis with other benign reactive or inflammatory lesions. Our most common nonclassic histologic pattern was mucosal hemorrhage–like pattern, wherein the presence of hemorrhage in the mucosa masked the possible lesional cells, making it difficult, if not impossible, to diagnose them on H&E stain. Most cases of this histologic subtype presented as nonspecific inflammatory changes on endoscopy. The other variants in our study included mucosal prolapse–like pattern (giving an appearance reminiscent of benign mucosal prolapse), mucosal inflammation–like and granulation tissue–like patterns (raising a consideration for an inflammatory, infectious, or drug-associated etiology), and dilated vascular- or lymphatic-like pattern (giving an appearance of nonspecific lymphovascular dilation and/or congestion). Extravasated red blood cells were found in association with lesional tissue in all cases on careful rereview.

Overall, our study was in keeping with the study by Zheng et al²⁰ depicting that GI-KS lesions show a broad histopathologic spectrum similar to the wide spectra described at cutaneous sites. However, the GI mucosal biopsies show different and unique histologic patterns when compared with the histologic patterns described in skin counterparts, and this is likely secondary to variations in tissue microenvironment. Zheng et al²⁰ found classic histology in 24% of overall cases, whereas we found classic histology in 36% (but 70% of these had coexisting nonclassic histology). We also had some differences from the study by Zheng et al.²⁰ Their study included both resections and biopsies. In our personal experience, resection specimens almost always show classic histology and hence we did not include any resection specimens in our study. Of note, all 3 resections in the study by Zheng et al²⁰ also showed classic histology. Also, we studied only the mucosal biopsies as there are potentially more chances of missing this diagnosis in limited biopsy material. Second, we did not characterize any cases as mimicking GIST or IMT-like histology, as any cases that had significantly obvious lesional spindle cells were grouped with the classic (conventional) histology pattern in our study. It would have been interesting to see how Zheng et al classified these 2 variants and how much they differed from classic histology; however, the absence of photomicrographs of these 2 specific patterns in their paper makes that assessment limited. Third, it was challenging for us to distinguish between lymphatic-like and vascular-like spaces in our cases (especially given the extravasated red blood cells), hence we combined these 2 into the same variant, namely dilated vascular- or lymphatic-like pattern. Currently, there are not enough data to discern whether there are any significant prognostic or clinical differences between these varied histologic subtypes; however, we believe the primary importance of this subtyping lies in alerting pathologists to the diverse nonclassic patterns, thereby preventing misdiagnosis. This is also supported by the observation that most of these nonclassic histologic subtypes coexist in the same biopsy and 70% of our biopsies showed a mixture of various histologic patterns.

Additionally, we correlated endoscopic findings with histologic findings and found that more than 90% of endoscopic nonspecific inflammatory lesions had nonclassic patterns on histologic examination. The existing data suggest that the endoscopic biopsy is associated with high false-negative rates (reported to be potentially about 35%–85%) for GI-KS.^{11,17–19,34} Some authors suggest that the low yield for a KS diagnosis on a GI biopsy is probably due to the prevalent submucosal location of the tumor.^11,18 The mucosal biopsies are mostly superficial and rarely sample any submucosa, wherein the classic or conventional histology may be more readily appreciable. We hypothesize that KS involving the GI tract probably tends to show a more nonclassic histology when evaluating only the superficial mucosa/lamina propria for lesional cells, given the differing ways that the lesional neoplastic cells colonize and percolate through the preexisting glands/crypts in the lamina propria, and hence we likely see mostly nonclassic histology in superficial mucosal biopsies. This may in part explain the high false-negative rates reported in the literature for GI-KS on endoscopic examination even when the biopsy is taken from a focus that is clinically suspicious for KS.¹⁷ Nagata et al¹⁷ found that the biopsy site of esophagus and endoscopic early stage (small size or patch appearance) were significantly associated with false-negative rate on endoscopic biopsy, whereas the findings of submucosal tumor with ulceration were significantly associated with true-positive results. Our study had many subtle cases, one of which was a superficially sampled esophageal biopsy with only squamous epithelium; the lesional cells were visualized only on the HHV8 stain and were seen exclusively in squamous peripapillary hemorrhagic areas, as also depicted in Figure 2.

One could also contend whether the designation of sarcoma is justifiable for cases that do not fit the clinical picture, have a nonspecific inflammatory pattern endoscopically, and have lesional cells identifiable only on HHV8 immunostain. Given that these patients are immunosuppressed, could it be the virus just colonizing the tissue and acting as a bystander, somewhat similar to occasional noncytomegalic immunoreactive CMV-positive cells seen in GI mucosal biopsies, which are deemed to represent innocent bystanders in the background of inflammation by some versus true pathogens by others.^35,36 Future studies would be helpful to further look into this possibility.

Nonetheless, the awareness of varied nonclassic histology patterns mimicking inflammatory or reactive histology on a limited biopsy material in the setting of immunosuppression should prompt the pathologist to actively consider this diagnosis and have an extremely low threshold to order the HHV8 stain. The importance of HHV8 as well as vascular endothelial marker immunostaining has been well established in the diagnosis of GI-KS.^4,37 Antibody to HHV8 latent nuclear antigen 1 is 99% to 100% sensitive and 100% specific for KS and is thus useful for distinguishing it from the mimickers.^38,39 All our biopsies (100%) showed positivity for HHV8 stain. Vascular endothelial markers (CD31 and CD34) were also positive in the selected cases in which they were performed in our study.

A coexisting infectious pathology was seen in 16% of our cases. Zheng et al²⁰ found a coexisting infection in a similar number (17%) of their cases. This is not surprising, as this patient population is prone to dual and/or multiple pathologies. The histologic changes that were present secondary to coexisting infectious or drug-induced etiologies (such as ulcer, inflammation, vascular changes, or granulation tissue) in these biopsies can further confound the assessment of KS-associated nonclassic (nonconventional) histology patterns. Hence, pathologists should have an extremely low threshold to perform the confirmatory HHV8 stain, despite finding a coexisting alternate etiology that may explain the inflammatory-type nonclassic histologic patterns on the biopsy.

Although our study was a retrospective single-institutional study, our patient population and case mix constitute a fair number of immunosuppressed patients who are at risk of developing KS, as seen by the number of GI-KS biopsies retrieved from our single-institution study. Our clinical experience in reviewing these cases, coupled with our high level of suspicion and low threshold of ordering the ancillary HHV8 stain even on cases in which the lesional cells were not readily appreciated on H&E stain, gives us this unique opportunity to comprehensively review the variable spectra of histopathologic findings that can be seen in these cases.

To summarize, conventional histology on GI biopsy and endoscopic correlate of a mass lesion in immunosuppressed patient often alert a pathologist to a diagnosis of KS. However, pathologists need to be aware that GI-KS can often present endoscopically as nonspecific inflammatory pattern, and unfortunately these are the cases that are more likely to show nonclassic and subtle patterns on GI biopsies and are easier to miss. The clinical history or clinical suspicion of KS may not be available to the pathologist on the requisition slip. Pathologists should be aware of the subtle and nonclassic histomorphologic diversity of KS on mucosal GI biopsy that can mimic varied benign reactive and inflammatory conditions and should be a prompt to actively consider this diagnosis, even in the absence of known cutaneous KS. Pathologists should also be aware that these patients are prone to dual or multiple pathologies, and the coexisting pathology can confound the histologic assessment and may overshadow KS-associated histologic changes.