Distinguishing pulmonary sarcomatoid carcinoma from pleural sarcomatoid mesothelioma is challenging because of overlapping histology, immunophenotype, and clinical features. Reliable immunohistochemical markers to aid in this distinction would be very valuable. Recent studies have proposed that MUC4 expression is common in sarcomatoid carcinoma but not in sarcomatoid mesothelioma, with the converse pattern reported for GATA3.
To further explore the utility of MUC4 and GATA3 in distinguishing pulmonary sarcomatoid carcinoma from sarcomatoid mesothelioma.
Well-characterized cases of sarcomatoid carcinoma (n = 32) and sarcomatoid mesothelioma (n = 64) were included. Diagnoses were confirmed by thoracic pathologists with incorporation of immunophenotype, clinical, and radiographic features. Whole-tissue sections were stained for GATA3 and MUC4.
Patients with sarcomatoid carcinoma and sarcomatoid mesothelioma had similar mean age and male predominance. GATA3 was positive in 63 of 64 sarcomatoid mesotheliomas (98%; 42 diffuse, 16 patchy, 5 focal), and 15 of 32 sarcomatoid carcinomas (47%; 3 diffuse, 8 patchy, 4 focal). MUC4 was positive in 2 of 64 sarcomatoid mesotheliomas (3%; 1 patchy, 1 focal), and in 12 of 32 sarcomatoid carcinomas (38%; 5 diffuse, 6 patchy, 1 focal).
Diffuse GATA3 expression favors sarcomatoid mesothelioma over sarcomatoid carcinoma, which rarely shows diffuse expression (sensitivity and specificity of diffuse staining 66% and 94%, respectively). Focal and patchy GATA3 expression is observed in both tumor types, and therefore is not helpful in this distinction. Sensitivity of MUC4 for sarcomatoid carcinoma was low in our cohort, positive in only 38% with frequent patchy staining, but it was quite specific.
Pathologic diagnosis of sarcomatoid neoplasms in the lung and pleura is quite difficult, because the differential diagnosis includes a long list of primary and metastatic tumors. While many entities, such as metastatic melanoma and various sarcomas, may be easily distinguished using currently available immunohistochemistry and molecular tests, sarcomatoid carcinoma and sarcomatoid mesothelioma often have nonspecific features that make definitive diagnosis very challenging. Pulmonary sarcomatoid carcinomas are poorly differentiated nonsmall cell lung carcinomas that contain at least 10% sarcoma or sarcoma-like (spindle and/or giant cell) differentiation. Pleural sarcomatoid mesotheliomas are poorly differentiated malignancies that arise from mesothelial cells, and also show spindle cell or mesenchymal morphology.1 Both tumors often show high-grade histology without specific morphologic features. Thus, resolving this differential diagnosis may rely on clinical and radiographic findings, which can also be ambiguous in some cases. Differentiating these 2 tumors is important, given the differences in etiology and treatment.2–5 Smoking is the main risk factor for sarcomatoid carcinoma, whereas sarcomatoid mesothelioma is usually associated with occupational or paraoccupational exposure to asbestos.2–5 Making the distinction between these 2 tumors is also becoming more relevant with the routine use of targeted therapy. Sarcomatoid pulmonary carcinoma and sarcomatoid mesothelioma both have a high rate of PD-L1 overexpression, making immune checkpoint inhibitor therapy an attractive treatment option for both tumor types.6–10 However, there are additional potentially targetable mutations that may be found in pulmonary sarcomatoid carcinoma, including MET exon 14 skipping mutations11,12 and rare examples of mutations in genes that are more commonly observed in lung adenocarcinoma, such as EGFR and ALK.13–15
Unfortunately, the immunophenotypes of sarcomatoid mesothelioma and sarcomatoid carcinoma show significant overlap. Keratin is usually expressed by both tumor types. However, other mesothelial markers, such as calretinin, WT1, CK5/6, and D2-40, are often negative or only very focally expressed in sarcomatoid mesothelioma.16 Likewise, carcinoma markers, such as MOC31, BerEp4, polyclonal CEA, TTF1, and p40, are routinely negative in sarcomatoid carcinoma.17–19 Consequently, novel immunohistochemical markers that could reliably make this distinction would be of great value. Two recent studies have suggested the immunohistochemical markers MUC4 and GATA3 might be useful in this differential diagnosis. The first study found expression of MUC4 by immunohistochemistry in the majority of pulmonary sarcomatoid carcinomas, while no staining was present in sarcomatoid mesothelioma.20 Conversely, a different study described GATA3 expression as a common finding in sarcomatoid mesothelioma, but not in pulmonary sarcomatoid carcinoma.21 Given the intriguing findings of these small prior studies, the goal of this study was to build on these data and further explore the utility of these stains to distinguish sarcomatoid pulmonary carcinoma from sarcomatoid mesothelioma based on our own well-characterized patient cohorts.
MATERIALS AND METHODS
This study was approved by the institutional review board. The surgical pathology archives were searched for cases of sarcomatoid and desmoplastic mesothelioma for the period 1992 to 2012. Slides were reviewed and diagnoses were confirmed by 2 experienced thoracic pathologists based on morphology, immunophenotype, and clinical/radiographic features.
The immunohistochemical panel used to confirm the diagnosis of sarcomatoid mesothelioma was performed on whole-tissue sections, and included antibodies directed against keratin (clone AE1/AE3; Dako, Carpinteria, California), BerEp4 (Dako), WT1 (clone WT 49; Novocastra, Newcastle upon Tyne, United Kingdom), calretinin (clone CAL6; Novocastra), TTF1 (clone SPT24; Novocastra), and p40 (clone BC28; Biocare Medical, Pacheco, California). Any cases where the diagnosis was felt to be ambiguous based on morphology, immunophenotype, or clinical features were excluded. The final cohort of sarcomatoid mesothelioma included 64 cases; all had extensive pleural involvement with compatible clinical/radiographic presentation and histopathology. Thirteen (20%) had desmoplastic histology. All were diffusely positive for keratin AE1/AE3 and negative for BerEp4 and p40. Fifty-nine cases (88%) were at least focally positive for WT1, and 27 cases (40%) were positive for calretinin. Four cases (6%) of otherwise typical sarcomatoid mesothelioma had focal TTF1 expression, but this was judged to be nonspecific staining when considering the overall morphology and immunophenotype, and thus they were included in the cohort.
The 32 included cases of sarcomatoid carcinoma included 23 cases of pleomorphic carcinoma, 5 cases of spindle cell carcinoma, and 4 cases of carcinosarcoma. Most cases (20, 63%) had at least focal areas of recognizable nonsmall cell carcinoma morphology (15 adenocarcinoma, 3 squamous cell carcinoma, 2 large cell undifferentiated carcinoma). All had lung parenchymal involvement, and pleural involvement was also present in 2 cases. This cohort of pulmonary sarcomatoid carcinoma had been subjected to prior extensive immunohistochemical analysis on whole-tissue sections, including mesothelial markers, mesenchymal markers, multiple keratins, and TTF1, as previously reported,18 and p40 staining was added as well. In summary, all were at least focally positive for keratin, and most were at least focally positive for TTF1 (20 of 32; 63%), napsin (13 of 32; 41%), or p40 (10 of 32; 31%). They had also been subjected to genetic analysis using a cancer hot-spot panel, ALK and ROS1 rearrangement testing, as previously reported.13 The Ion AmpliSeq Cancer Hotspot Panel v2 Next Generation Sequencing panel was used, which tests for approximately 2800 individual mutations in 50 oncogenes and tumor suppressor genes, including EGFR, KRAS, NRAS, TP53, BRAF, ERBB2, JAK3, AKT1, ATM, MET, KIT, and PIK3CA.
A representative whole-tissue section from each case was stained for MUC4 (clone 8G7; Cell Marque, Rocklin, California) and GATA3 (clone L50-823; Biocare Medical, Concord, California). Staining was recorded in a semiquantitative fashion according to the number of tumor cells that were positive. Staining was considered focal if positivity was observed in less than 10% of tumor cells, patchy if 10% to 50% of tumor cells were positive, and diffuse if more than 50% of tumor cells were positive. When no appreciable staining was observed, the stain was considered negative.
Clinicopathologic features were extracted from the existing medical records and pathology reports.
Patients with sarcomatoid carcinoma (n = 32) and sarcomatoid mesothelioma (n = 64) had similar median age and age range (Table 1). Both tumor types showed a male predominance, but this was more pronounced in the sarcomatoid mesothelioma group, where 86% of patients were men.
Immunohistochemistry findings for MUC4 and GATA3 are summarized in Table 1. The rate of any degree of MUC4 positivity in sarcomatoid carcinoma was 38%, compared with 3% in sarcomatoid mesothelioma. GATA3 was positive to some degree in 47% of sarcomatoid carcinomas and 98% of sarcomatoid mesotheliomas. The combined immunohistochemistry results are shown on Table 2. Sarcomatoid mesotheliomas were GATA3 positive and MUC4 negative in 95.5% of cases. However, 35.5% of sarcomatoid carcinomas also showed this pattern. Sarcomatoid carcinomas were MUC4 positive and GATA3 negative in 25% of cases.
Considering the results of the immunohistochemistry in more detail, it was noted that 5 of 12 sarcomatoid carcinomas that were positive for MUC4 showed diffuse staining (Figure 1, A and B), while 6 had patchy staining, and 1 had only focal staining. Three of the MUC4-positive sarcomatoid carcinomas had only sarcomatoid areas in the stained sections (1 diffuse; 2 patchy). In the other 9 cases of MUC4-positive sarcomatoid carcinoma, both sarcomatoid morphology and more conventional areas of nonsmall cell carcinoma were present on stained sections. When considering staining in the sarcomatoid and nonsmall cell carcinoma components separately, 1 case showed similar diffuse positive staining pattern in both components; 4 were positive in both components but differentially stronger in nonsmall cell areas (diffuse  or patchy  in the epithelial component, and focal  or patchy  in the sarcomatoid component); and 4 were positive only in the epithelial component (1 diffuse, 2 patchy, 1 focal). One such example is illustrated in Figure 2, A through D, where MUC4 is strongly and diffusely positive in the nonsmall cell carcinoma component, while the expression of this marker was completely lost in the transition to the sarcomatoid areas. MUC4 was positive in only 2 of 64 sarcomatoid mesotheliomas, including 1 with patchy staining and 1 with focal staining, which is pictured in Figure 3, A through C. When any degree of positivity was considered, the sensitivity and specificity of MUC4 for the diagnosis of sarcomatoid carcinoma were 38% and 97%, respectively.
GATA3 was positive to some degree in the vast majority of sarcomatoid mesotheliomas (98%), with most cases showing strong and diffuse expression. Diffuse staining was observed in 42 cases (Figure 4, A and B), while 16 cases showed patchy staining and 5 had focal staining. GATA3 was positive in 15 of 32 sarcomatoid carcinomas (47%), and in most cases (12) the staining was patchy or focal. When any degree of positivity was considered, the sensitivity and specificity of GATA3 for the diagnosis of sarcomatoid mesothelioma were 97% and 53%, respectively.
The diagnosis of sarcomatoid carcinoma was well established in the 12 cases with patchy or weak GATA3 staining, based on a parenchymal lung location (12), the presence of a nonsmall cell carcinoma component (5), and expression of TTF1 (7), p40 (6), and/or MUC4 (3). Diffuse expression of GATA3 was present in 3 cases of sarcomatoid carcinoma. Given the high percentage of sarcomatoid mesothelioma that showed diffuse expression of GATA3, the diagnosis of sarcomatoid carcinoma was critically reconsidered in these 3 cases. One case showed definitive features of sarcomatoid carcinoma, including a parenchymal lung location, presence of mutations common in lung cancer including TP53 and KRAS, and a metastatic pattern highly suggestive of lung cancer with adrenal, liver, and lung metastases; additionally, there was no evidence of breast or urothelial primary site, which could also be expected to show diffuse GATA3 expression. The diagnosis in the second case was more difficult once the strong GATA3 expression was considered, because the patient did have parenchymal and pleural involvement, but there was strong and diffuse TTF1 expression (Figure 5, A through C), as well as pCEA expression, features that favor sarcomatoid carcinoma. The diagnosis in the third case was much more ambiguous once the strong GATA3 expression is considered. This tumor did have quite extensive pleural involvement in addition to parenchymal lung involvement, and only pankeratin was expressed without expression of other carcinoma or mesothelial markers. No mutations suggestive of lung cancer were detected by the genetic hotspot panel. While the morphologic features were felt to favor carcinoma, with large prominent nucleoli and marked pleomorphism, based on the GATA3 expression in conjunction with the disease distribution, we would consider this a probable sarcomatoid mesothelioma. After reclassification of this tumor as sarcomatoid mesothelioma, if only diffuse GATA3 staining is considered positive, the sensitivity and specificity of diffuse GATA3 staining for the diagnosis of sarcomatoid mesothelioma were 66% and 94%, respectively.
Recent data have indicated that MUC4 and GATA3 might be useful additions to the immunohistochemistry panel that is routinely used in evaluation of sarcomatoid thoracic malignancies.20,21 MUC4 is a membrane-bound glycoprotein that is an important component of the protective layer of mucin that covers many mucosal surfaces.22 A recent gene expression profiling study identified MUC4 overexpression in pulmonary sarcomatoid carcinomas compared with sarcomatoid mesothelioma.20 By immunohistochemistry, 72% of sarcomatoid carcinomas were positive for MUC4 protein, while no sarcomatoid mesotheliomas were positive. MUC4 had a superior sensitivity and specificity profile than other traditional mesothelial and carcinoma markers tested in that study.20 GATA3 is a transcription factor important in development and maintenance of several types of normal tissue, and has gained widespread use as a marker of breast and urothelial carcinomas.23,24 The study by Berg et al21 described GATA3 expression as a common finding in sarcomatoid mesothelioma, but not in pulmonary sarcomatoid carcinoma. Strong and diffuse GATA3 expression was observed in all 19 sarcomatoid and desmoplastic mesotheliomas tested in that study, compared with only 2 of 13 sarcomatoid carcinomas, which showed weak and patchy expression.21 While these data were promising, the studies were small, and we sought to explore the utility of these markers in a larger independent cohort.
Our study demonstrates that MUC4 expression is quite specific for sarcomatoid carcinoma when the differential diagnosis of sarcomatoid mesothelioma is considered, but unfortunately the sensitivity is quite low. We also noted that staining is frequently focal and patchy, and may be completely lost in sarcomatoid areas. This could pose issues regarding the utility of this stain in small biopsies, where only sarcomatoid areas may be represented. Therefore, we conclude that while MUC4 expression favors sarcomatoid carcinoma, a negative stain does not exclude the diagnosis. The rate of MUC4 positivity we observed is much lower than what was reported by Amatya et al,20 despite using similar methods with the same antibody clone and whole-tissue sections. They also reported that they evaluated MUC4 positivity in the spindled tumor cells alone20 ; therefore, the reason for this discrepancy in sensitivity is not entirely clear. In subsequent related correspondence, Amatya et al25 did report that there was no expression of MUC4 in all 5 pure spindled cell carcinomas they tested, but our study included mostly pleomorphic carcinomas, so this also does not explain the difference. The low MUC4 sensitivity we observed in sarcomatoid carcinoma was very similar to data recently reported in a letter to the editor by Berg et al,26 and they also reported a similar observation that MUC4 positivity may be diffuse in areas of nonsmall cell carcinoma morphology, but lost in sarcomatoid areas.
We found that GATA3 is a very sensitive maker of sarcomatoid mesothelioma, where its staining is usually strong and diffuse, similar to the study using the same antibody clone performed by Berg et al.21 However, we found some cases of sarcomatoid mesothelioma show focal or patchy staining, which is also observed in sarcomatoid carcinoma, and therefore is not specific. It seems that GATA 3 expression is most helpful when it is either completely negative, or strong and diffuse. A completely negative stain strongly favors sarcomatoid carcinoma over sarcomatoid mesothelioma. Robust strong and diffuse GATA3 expression favors sarcomatoid mesothelioma; when only diffuse GATA3 expression is considered, the sensitivity and specificity of this marker for sarcomatoid mesothelioma over sarcomatoid carcinoma are 66% and 94%, respectively. However, we found that rare examples of sarcomatoid pulmonary carcinoma have diffuse GATA3 expression as well, so it cannot be used to completely exclude that diagnosis. Given that focal and patchy staining is observed in both sarcomatoid mesothelioma and sarcomatoid carcinoma, this pattern of expression is not helpful to make the distinction between these 2 entities.
In summary, diffuse GATA3 expression favors a diagnosis of sarcomatoid mesothelioma over pulmonary sarcomatoid carcinoma, which only rarely shows diffuse GATA3 expression. Patchy or focal GATA3 staining is not helpful, because that type of staining can be seen in both sarcomatoid pulmonary carcinoma and sarcomatoid mesothelioma. Conversely, MUC4 expression is quite specific for sarcomatoid pulmonary carcinoma over sarcomatoid mesothelioma, but it was not a very sensitive marker in our cohort. The frequent patchy and focal expression pattern of MUC4 with loss or decreased expression in sarcomatoid areas may further limit the utility of this marker, particularly in the interpretation of small biopsies.
The authors have no relevant financial interest in the products or companies described in this article.