Esophageal granular cell tumor associated with pseudoepitheliomatous hyperplasia.
Granular cell tumors (GCT) of the esophagus are usually detected during endoscopic procedures as incidental findings, although they can be associated with dysphagia. They present as mucosal nodules, usually less than 1.0 cm, and can present as multiple lesions.1 GCTs are more common in females in the fourth and fifth decades and are more common in African Americans. Although most frequently described in the head and neck region, GCTs of the digestive system have been described with the most common location being the esophagus accounting for approximately 1-2% of lesions. Other common digestive tract locations include the colon and perianal region. The tumor is thought to be of neural origin, possibly arising from Schwann cells, and consists of nests of polygonal cells with coarsely granular cytoplasm and with small hyperchromatic nuclei. GCTs can sometime have features of extensive hyalinization and/or dystrophic calcifications. Being of neural origin, the tumor cells stain positive for S100 immunohistochemistry. Other positive immunohistochemical markers described for GCTs include CD68, inhibin, vimentin, neuron-specific enolase and myelin basic protein. The tumor is characteristically negative for cytokeratins and muscle markers. The cytoplasmic granules stain positive for periodic acid-schiff and are diastase resistant. When identified in association with squamous mucosa, it is frequently associated with pseudoepitheliomatous hyperplasia, which should not be mistaken for squamous cell carcinoma or dysplasia.2,10 While GCTs are frequently found in the esophagus and head and neck region, GCTs at other sites including the appendix, breast and vulva have been described.3,4,5,10
GCTs have a characteristic histologic appearance, and when present in the esophagus alternative diagnoses which may be considered at other sites are less likely. For example, as previously mentioned the most common location for GCTs is the oral cavity/head and neck region. In such locations, where skeletal muscle could be in close proximity, a rhabdomyoma might be considered in the differential; however, the cytoplasm would not have the characteristic granular appearance. Additionally, when present in deeper soft tissues and the oral cavity, the cells of alveolar soft part sarcoma (which will also be positive for PAS-D) can resemble the cell population of GCTs.2 There is also case reports of amelanotic melanomas being confused for granular cell tumors in the oral cavity 6.
While generally considered to be benign lesions, malignant potential has been described in 1-2% of cases; however, this characterization is considered controversial. Clinical features pointing to malignant GCTs include large size, invasion of adjacent structures and rapid growth.10 Potential criteria for classifying GCTs has been outlined based on histologic features, including: spindling of tumor cells, presence of vesicular nuclei with larger nucleoli, increased mitotic rate (defined as 2 mitoses per 10 hpf), a high nuclear to cytoplasmic ratio, pleomorphism and necrosis. It is reported that neoplasms which meet 3 or more of these criteria should be considered malignant, those that meet 1 or 2 of the criteria should be classified as atypical, and those with no features or only focal pleomorphism should be classified as benign. In a case study of 73 GCTs that were classified as either malignant, metastasizing, multicentric or atypical, 39% of patients with a GCT classified as malignant based on these criteria died of disease at a median of three years. Metastases from histologically malignant tumors occurred in 50% of patients and were identified most commonly in the lymph nodes, lungs and bone and less frequently in the liver, bowel, thyroid, heart and brain. The study also looked at potential staining differences and in their study only one histologically malignant GCT did not stain for S100 (but did stain for CD57 and vimentin). CD68 was identified in 100% of benign lesions, 56% of atypical lesions and 27% of malignant lesions. As expected, Ki67 staining increased in the spectrum between benign and malignant lesions. Additionally, p53 staining was also observed in as many as 50% of malignant GCTs as compared to negative staining in 100% of benign GCTs.7 In a separate study of 26 colonic GCTs, 40% of the cases had diffuse nuclear pleomorphism and were characterized as atypical. Subsequently, 25% of these cases had local recurrence at follow up (6 months and 1 year) thought to be due to incomplete excision.10
Some attention has been given to potential genetic alterations in GCTs. In the previously mentioned study of 73 GCTs, DNA ploidy analysis was performed. In eleven benign GCTs, nine were diploid while two were hyperdiploid. An atypical granular cell tumor was aneuploid and of six malignant granular cell tumors analyzed, two were aneuploid, two were hyperdiploid and two were diploid.7 A 2004 study utilizing microarray analysis of 19,200 genes in attempts to identify up-regulated and down-regulated genes localized numerous genes involved in signal transduction (e.g. receptor tyrosine kinase up-regulation), cell cycle regulation (e.g. Rho GTPases up-regulation), and cytoskeleton organization (e.g. CD44 up-regulation, MACF1 down-regulation).8 In a subsequent 20013 article looking at loss of heterozygosity in granular cell tumors using PCR analysis, identified loss of heterozygosity at chromosome 17p13.1 and/or chromosome 9p22-p13 in five of eight oral GCTs and specifically involved tumor suppressor genes TP53 and P16. The study also raises the possibility of other gene involvement at these loci and, although a limited number of cases, clearly defines itself as a pilot study.9
Contributed by Brian K. Theisen MD and Aatur D. Singhi MD PhD