Immunoreactivity for Calretinin and Keratins in Desmoid Fibromatosis and Other Myofibroblastic Tumors

Abstract

Calretinin is an intracellular calcium-binding EF-hand protein of the calmodulin superfamily. It plays a role in diverse cellular functions, including message targeting and intracellular calcium signaling. It is expressed in the mesothelium, mast cells, some neural cells, and fat cells, among others. Because of its relative specificity for mesothelial neoplasms, calretinin is widely used as one of the primary immunohistochemical markers for malignant mesothelioma and in differentiating it from adenocarcinoma. On the basis of our sporadic observation on calretinin immunoreactivity in desmoid fibromatosis, we systematically evaluated calretinin, keratin cocktail (AE1/AE3), and WT1 immunoreactivity in 268 fibroblastic/myofibroblastic neoplasms. Calretinin was observed in 75% (44/58) of desmoid fibromatosis, 50% (21/42) of proliferative fasciitis, 23% (8/35) of nodular fasciitis, 33% (13/40) of benign fibrous histiocytoma, 35% (22/62) of malignant fibrous histiocytoma, and 13% (4/31) of solitary fibrous tumors but not in normal connective tissue fibroblasts at various sites. Keratin AE1/AE3 immunoreactivity was also commonly (6/13) present in the large ganglion-like cells of proliferative fasciitis and sometimes in nodular fasciitis (3/35), solitary fibrous tumor (3/27), and malignant fibrous histiocytoma (9/62). Nuclear immunoreactivity for WT1 or keratin 5 positivity was not detected in myofibroblastic tumors. On the basis of these observations, it can be concluded that calretinin and focal keratin immunoreactivity is fairly common in benign and malignant fibroblastic and myofibroblastic lesions. Calretinin-positive and keratin-positive spindle cells in desmoid and nodular fasciitis or calretinin-positive ganglion-like cells in proliferative fasciitis should not be confused with elements of epithelioid or sarcomatoid mesothelioma. These diagnostic pitfalls can be avoided with careful observation of morphology, quantitative differences in keratin expression, and use of additional immunohistochemical markers such keratin 5 and WT1 to verify true epithelial and mesothelial differentiation typical of mesothelioma.

Calretinin is an intracellular calcium-binding EF-hand protein of the calmodulin superfamily that plays a role in diverse cellular functions, including message targeting and intracellular calcium buffering and signaling.^2,21 In 2003, a microarray study of 5233 samples showed that calretinin is expressed in normal tissues, such as in neurons in the cerebrum and cerebellum; pleural mesothelia; testicular Leydig cells; endometrial stromal cells; luminal cells of the breast; adrenal cortex and medullary sustentacular cells; and neural elements of the urogenital and gastrointestinal tracts.⁹ It was also expressed, at least weakly, in a variety of neoplasms of the skin, mesothelia, respiratory tract, genitourinary tract (male and female), brain, soft tissues, and in some neuroendocrine tumors.⁹

Calretinin is widely used as one of the primary positive immunohistochemical markers for mesothelial cells and malignant mesothelioma on account of its relative specificity and sensitivity in the detection of mesotheliomas.^2,3,10,19 In particular, it is commonly used as part of a panel to identify malignant mesothelioma and distinguish it from pulmonary adenocarcinoma. In addition, calretinin is also a marker for sex cord stromal tumors¹¹ and a potential marker for adipose tissue tumors.¹ It has also been suggested as being useful in the separation of neurofibroma from schwannoma, as only the latter is typically positive.⁴

We have serendipitously encountered patients with chest wall desmoid fibromatosis presented as mesothelioma, on the basis of immunohistochemical calretinin positivity. This study was conducted to systematically examine this diagnostic pitfall: the potential calretinin expression in myofibroblastic neoplasms that can be confused with mesothelioma. Other markers of mesothelioma, such as keratin cocktail AE1/AE3, keratin 5, selected other keratins, and WT1, were also examined.

MATERIALS AND METHODS

Study Material

A total of 268 fibroblastic and myofibroblastic neoplasms were examined histologically and immunohistochemically, including 58 cases of desmoid fibromatosis (later abbreviated as “desmoid”), 42 proliferative fasciitis, 35 nodular fasciitis, 40 benign fibrous histiocytomas, 31 solitary fibrous tumors, and 62 malignant fibrous histiocytomas/myxofibrosarcomas (including cases also designated as pleomorphic undifferentiated sarcomas). All diagnoses were verified by histology and extensively supported by immunohistochemistry. In addition, selected normal tissues were examined for calretinin expression.

Immunohistochemical studies were conducted on 4-μm-thick sections derived from multitissue blocks. These blocks were prepared using a previously described manual method.¹⁶ In brief, cuboidal or rectangular tumor samples were dissected from formaldehyde-fixed and paraffin-embedded tissue pieces or embedded blocks using a 1-sided razor blade (Electron Microscopy Sciences, Hatfield, PA). The pieces were then embedded in a paraffin block in a desired order in multiple rows of 5 to 10 pieces per row.

Antibodies and Immunostaining

Immunohistochemical analysis was performed using a Leica Bond-Max automated immunostainer (Leica Biosystems, Bannockburn, IL). Heat-induced epitope retrieval with a high-pH epitope retrieval buffer (Leica) was performed for 25 minutes. Primary antibody (polyclonal calretinin-specific antibody from Zymed/Invitrogen, Camarillo, CA) was used at a dilution of 1:100 and applied for 30 minutes, followed by Bond-Max polymer for 15 minutes. Diaminobenzidine was used as the chromogen, followed by a light hematoxylin counterstain. Mast cells, fat, and mesothelia were used as internal positive controls, and malignant epithelial mesothelioma was used as an external control.

Most tumors were also examined using antibodies to AE1/AE3 keratin cocktail (Chemicon # MAB3412, Temecula, CA; diluted 1:200), keratin 5 (Novocastra/Leica; diluted 1:100), and WT1 (Dako Cytomation, Carpinteria, CA; diluted 1:200) using an immunostaining protocol as described above. In selected cases, calretinin immunoreactivity was demonstrated using the same antibody in the Ventana Benchmark system, with concordant results. Selected cases were evaluated for expression of keratins 7 (clone OV-TL 12/30; Dako Cytomation; diluted 1:500), 8 (clone TS1; Labvision/Neomarkers; diluted 1:100), and 18 (clone DC10; Novocastra/Leica; diluted 1:100). In some cases, keratin AE1/AE3 immunoreactivity was confirmed using alternative enzymatic epitope retrieval (Sigma protease type II for 5 min), with the same antibody dilution as above.

RESULTS

In normal human tissue, calretinin was detected in the mesothelia, mast cells, intestinal and some central nervous system neural elements, and variably in adipose tissue but not in normal fibroblastic stromal elements of various organs, including breast, dermis of the skin, gastrointestinal tract, omentum, synovia, urinary bladder, and uterine cervix. The expression of calretinin, keratins AE1/AE3, and WT1 in myofibroblastic neoplasms has been summarized in Table 1 and illustrated in Figures 1–3.

TABLE 1.

Expression of Calretinin, Keratins (AE1/AE3), WT1, and CK5 in Desmoid Fibromatosis and Other Myofibroblastic Tumors

Diagnosis	Calretinin	Keratins (AE1/AE3)	WT1 and CK5
Desmoid fibromatosis	44/58 (75%)	0/37	0/37
Proliferative fasciitis	21/42 (50%)	6/13 (46%)	0/13
Nodular fasciitis	8/35 (23%)	3/35 (9%)	0/35
Benign fibrous histiocytoma	13/40 (33%)	0/32	0/32
Malignant fibrous histiocytoma	22/62 (35%)	9/62 (15%)	0/62
Solitary fibrous tumor	4/31 (13%)	3/27 (11%)	0/27

FIGURE 1.

Calretinin positivity in desmoid fibromatosis (C and D). Cytoplasmic and nuclear calretinin positivity is seen in both spindled and epithelioid myofibroblasts, here illustrated in 2 different cases with matching hematoxylin and eosin sections (A and B).

FIGURE 3.

A, Histologic appearance of malignant fibrous histiocytoma. B, This tumor shows focal nuclear and cytoplasmic immunoreactivity for calretinin. C, Focal keratin AE1/AE3 positivity in malignant fibrous histiocytoma. D, Calretinin positivity in perivascular spindled cells in solitary fibrous tumor.

Desmoid Fibromatosis

Desmoids contained calretinin-positive spindled or epithelioid/dendritic-shaped tumor cells in 75% of the cases (44/58) (Fig. 1). Nuclear and cytoplasmic immunoreactivity was present focally or diffusely, with an estimated mean proportion of positive cells of 33% (range, 10% to 100%). Positivity was observed in desmoids from the abdominal wall, peripheral soft tissues, and mesenteries. Desmoids were keratin AE1/AE3 negative except for some endothelial or pericytic cells. None of the 37 desmoids studied contained nuclear WT1 or keratin 5 immunoreactivity, but variable cytoplasmic positivity was common.

Proliferative and Nodular Fasciitis

Proliferative fasciitis cases (most of them from extremity locations) contained varying numbers of large ganglion-like cells with abundant mildly eosinophilic cytoplasm (Fig. 2A). Half (21/42) of these cases showed calretinin positivity in the ganglion-like cells (Figs. 2B and C). These cells typically showed both cytoplasmic and nuclear staining, either focally or more often diffusely (mean proportion of positive cells: 69%; range, 1% to 100%). In nodular fasciitis (most of them from the extremities), calretinin positivity was detected in a minor portion of spindled myofibroblasts in 23% (8/35) of the cases. Staining was focal in most of the cases, with a mean proportion of positive cells of 32% (from 5% to 90%).

FIGURE 2.

A, Histologic appearance of ganglion-like cells in proliferative fasciitis. B and C, Cytoplasmic and nuclear calretinin positivity in ganglion-like cells of proliferative fasciitis. D, Focal keratin AE1/AE3-positive ganglion-like cells, in which the immunostain highlights a somewhat dendritic appearance.

Keratin AE1/AE3 positivity (usually focal) was seen in 46% (6/13) of cases of proliferative fasciitis, mainly in the ganglion cell–like elements, some of which showed dendritic morphology in the keratin immunostaining (Fig. 2D). Comparable AE1/AE3 keratin positivity was also seen after alternative epitope retrieval by means of protease digestion. Focal keratin 7 was also similarly detected in the ganglion-like cells. However, in those tumors with focal AE1/AE3 positivity, there was no keratin 5, 8, or 18 immunoreactivity or nuclear WT1 immunoreactivity.

Rare keratin-positive spindled cells, possibly related to pericytes of the complex capillary network, were detected in 3/35 cases of nodular fasciitis. In addition, endothelial cells were variably, usually focally, keratin AE1/AE3 positive. No positivity was detected with antibody to keratin 5, and no nuclear WT1 immunoreactivity was observed.

Other Fibroblastic/Undifferentiated Tumors

Spindle cells of benign fibrous histiocytoma (Fig. 3A) showed calretinin-positive tumor cells in 33% (13/40) of cases examined, but no AE1/AE3 keratin-positive tumor cells were detected. These cases showed focal areas of calretinin-positive spindle cells (mean proportion of positive cells, 31%; range, 20% to 50%).

Malignant fibrous histiocytomas/myxofibrosarcomas (Fig. 3A) often (22/62, 35%) contained calretinin-positive cells (Fig. 3B), and 9/62 cases (15%) also showed small numbers of keratin AE1/AE3-positive tumor cells (Fig. 3C). Of the AE1/AE3-positive cases, 4 were also positive for calretinin, 7 were positive for keratin 18, 2 for keratin 8, and none for keratin 7. None of these tumors showed keratin 5 positivity or nuclear WT1 immunoreactivity.

Only few solitary fibrous tumors contained calretinin-positive tumor cells (13%; 4/31) (Fig. 3D). Focal keratin AE1/AE3 and keratin 18 positivity beyond endothelia and pericytes was detected in 3 cases, all of which were negative for calretinin. Of those keratin-positive cases, 1 was also positive for keratins 7 and 8.

DISCUSSION

In this study, we observed unexpected, rather widespread, immunohistochemical expression of calretinin in the neoplastic cells of desmoid fibromatosis (75% of cases) and other myofibroblastic neoplasms, including nodular and proliferative fasciitis (50%), benign fibrous histiocytoma, and myxofibrosarcoma/malignant fibrous histiocytoma/pleomorphic undifferentiated sarcoma. Even solitary fibrous tumor, although mostly calretinin negative, showed positive spindle cells in 10% of cases. In these tumors, both nuclear and cytoplasmic immunoreactivity was detected, as typical of calretinin. We conclude that neither the morphology nor distribution of calretinin-positive cells is consistent with normal calretinin-positive cellular elements, such as fat or mast cells.

Calretinin has been one of the most commonly used positive markers in the diagnosis of mesothelioma, as it is expressed in different types of mesothelioma including epithelial and sarcomatoid ones.^2,3,10,19 In addition, calretinin is expressed in a variety of other normal cell types and tumors. Lugli and colleagues investigated calretinin expression using tissue microarray in 5233 tissue samples from 128 different tumor categories and 76 different normal tissue types. At least 1 case with weak expression could be found in 74 of 128 (58%) different tumor types, and 46 entities (36%) had at least 1 tumor with strong positivity.⁹ With regard to normal tissues, a particularly strong expression was found in Leydig cells of the testis, neurons of the brain, theca lutein and theca interna cells of the ovary, and mesothelium. With regard to tumors other than mesothelioma, strong expression was identified in mesothelial adenomatoid tumors, testicular Leydig cell tumors, and adrenocortical adenomas.⁹

With regard to soft tissue tumors, marked cytoplasmic and nuclear calretinin positivity has been detected in different types of synovial sarcoma as well, including biphasic, monophasic, and poorly differentiated ones.¹⁶ Similar to that seen in mesothelioma, both cytoplasmic and nuclear calretinin positivity has been reported in the spindle and epithelial components.¹⁶ Normal adipose tissue and lipomatous tumors may also exhibit calretinin immunoreactivity. Calretinin-positive nuclear and cytoplasmic staining was reported in lipomas and their variants, and in well-differentiated liposarcomas, at least focally.¹ Other studies demonstrate calretinin positivity in thymoma and thymic carcinoma,²⁰ schwannoma,⁴ and high-grade breast carcinomas⁹ among others.

Little attention has been paid to calretinin immunoreactivity in fibroblastic/myofibroblastic neoplasms, such as desmoid. Because these tumors can occur in a wide variety of body sites, including the chest wall and serous cavities, they can potentially simulate mesothelioma, especially localized, nondiffuse forms of mesothelioma and sarcomatoid mesothelioma.

Calretinin positivity in desmoid and other myofibroblastic tumors, if interpreted out of context in a small sample (cytologic or needle biopsy specimen), may mislead one to suspect a mesothelioma. However, lack of cytologic atypia and the overall morphologic myofibroblastic appearance, as well lack of other positive mesothelioma markers (such as Wilms tumor protein 1 and keratins 5/6), immunohistochemically helps to distinguish desmoid from epithelial and sarcomatoid mesotheliomas. The same is true for other myofibroblastic neoplasms with calretinin positivity. Genetic studies such as gene copy number evaluation may also be helpful in the differential diagnosis of sarcomatoid mesothelioma and other spindle cell tumors.⁷

Adding significance to the differential diagnosis from mesothelioma is the fact that various fibroblastic and myofibroblastic tumors can involve the chest wall, abdominal wall, and even abdominal cavity. Surveys at the Armed Forces Institute of Pathology (AFIP) showed that 12% of nodular fasciitis, 14% of proliferative fasciitis, and 30% of extrapleural solitary fibrous tumors involved the chest/abdominal wall. In addition, desmoid fibromatosis is notorious for involving the abdominal wall, especially in women. AFIP experience showed involvement of the chest wall in 19% of cases and sometimes also of the pleural cavity or mediastinum.^12,13

Presence of keratin immunoreactivity in some myofibroblastic neoplasms is an additional feature that can mislead one to interpret such a finding as proof of epithelial differentiation and strengthen the suspicion of mesothelioma when facing a calretinin-positive tumor. Keratin positivity in the calretinin-positive epithelioid, ganglion cell–like elements in proliferative fasciitis especially might be misinterpreted as evidence for epithelial differentiation and supporting mesothelioma. However, in our experience it is differentially diagnostically helpful that sarcomatoid mesotheliomas are profusely and diffusely keratin positive, in contrast to focal, limited positivity observed in most keratin-positive myofibroblastic tumors. Furthermore, consistent lack of nuclear WT1 and keratin 5/6 immunoreactivity is also helpful. However, these 2 markers may show reduced or even no expression in some mesotheliomas, especially the sarcomatoid ones.^10,19

The fact that keratin positivity in proliferative fasciitis could be demonstrated with different epitope retrieval modes and different antibodies with concordant results (including one for keratin 7) supports true keratin expression, as opposed to technical or other artifacts. Some older studies also support the presence of keratin positivity in fibroblastic and myofibroblastic lesions. Transformed cultured fibroblasts have been long known to acquire expression of keratins, especially keratins 8 and 18.^6,22 In addition, benign fibroblastic neoplasms, such as collagenous fibroma, have been demonstrated to occasionally contain keratin AE1/AE3 immunoreactive lesional cells.¹⁵ Inflammatory myofibroblastic tumor is also well known for significant, even striking, keratin expression in spindled myofibroblasts.^5,8 One older report identified keratin-positive cells in 1 of 15 cases of ischemic fasciitis/atypical decubital fibroplasia, a reactive myofibroblastic proliferation related to proliferative fasciitis.¹⁷ Another study from 1991 from AFIP did not detect keratin AE1/AE3 positivity in nodular fasciitis, in contrast to our findings.¹⁸ This might be explained with the use of older, less sensitive, methods of detection and protocols in the past manual immunohistochemistry protocols, compared with the exquisitely sensitive detection systems utilized in the most modern automated immunohistochemistry.

In summary, we show that calretinin immunoreactivity is commonly detectable in benign myofibroblastic lesions, especially in desmoid fibromatosis, nodular fasciitis, and proliferative fasciitis. Also some malignant fibroblastic tumors, such as malignant fibrous histiocytoma/myxofibrosarcoma, show calretinin positivity. Such calretinin immunoreactivity should be understood as a part of the immunophenotypic features of myofibroblastic tumors and not confused with malignant mesothelioma. Furthermore, focal keratin AE1/AE3 immunoreactivity also occurs in some benign and malignant myofibroblastic lesions, further increasing the risk of misinterpretation of these elements as evidence of a mesothelial proliferation. Careful observation of morphology and quantitative differences in keratin expression, and application of additional immunohistochemical markers for mesothelial differentiation, such as WT1 and keratin 5, helps to avoid these diagnostic pitfalls.