Mesonephric-like endometrial carcinoma is a rare but frequently misclassified and aggressive malignancy. KRAS mutations, limited estrogen receptor (ER) expression, and TTF-1, GATA3, and luminal CD10 expression are described in these tumors, but an immunohistochemistry-based screening approach has not been studied. We assessed 300 endometrial carcinomas/carcinosarcomas to ascertain the specificity of TTF-1/GATA3/luminal CD10 expression with or without ER staining for this diagnosis. Next-generation sequencing and morphologic review were performed on screen-positive cases. In all, 3% (9/300) were TTF-1+; 2 coexpressed GATA3. No cases expressed luminal CD10 or GATA3 in isolation. Two TTF-1+/ER− cases, one of which was also GATA3+, were reclassified as mesonephric-like based on morphology and molecular results (KRAS mutations without mismatch repair deficiency, TP53 mutations, or PTEN mutations): these represented 0.7% of all cases (2/300). The reclassified cases were originally diagnosed as grade 1 and 2 endometrioid carcinoma, and the latter had pulmonary metastases and pelvic recurrences. Six TTF-1+ cases retained their original serous (3) and endometrioid (3) diagnoses; 1 was reclassified as dedifferentiated. All had negative or low ER. KRAS mutations were identified in 4 TTF-1+ non–mesonephric-like cases, including 1 serous and 1 grade 3 endometrioid with p53 abnormalities, 1 mismatch repair–deficient endometrioid with a complex molecular profile, and 1 endometrioid with mucinous differentiation. These findings suggest that TTF-1 and ER are good first-line screens for mesonephric-like carcinoma, but caution that a TTF-1+/ER− immunoprofile is not specific, even in the setting of KRAS mutations. A final diagnosis of mesonephric-like carcinoma requires integration of morphologic and immunohistochemical features, with molecular support when relevant.

Immunohistochemistry (IHC) for mismatch repair (MMR) proteins is recommended in endometrial carcinomas as a screening test for Lynch syndrome, and mismatch repair deficiency (MMRd) is reported in ∼30% of cases. However, few studies have evaluated the rate of MMR loss in uterine carcinosarcomas. A 5-year retrospective database search of uterine carcinosarcomas was performed at 3 academic institutions. The histologic diagnoses, type of carcinoma present, and MMR IHC interpretations were confirmed by a gynecologic pathologist. One hundred three cases of uterine carcinosarcomas with available MMR IHC results were identified. Ninety-nine cases (96%) showed intact expression and 4 cases (4%) showed loss of MLH1/PMS2. All MMRd carcinosarcomas identified in this series had an endometrioid carcinomatous component and wild-type p53 expression. In contrast, the majority of MMR intact carcinosarcomas had a serous morphology and aberrant p53 expression. Three additional cases initially diagnosed as carcinosarcoma also revealed MMRd; however, given the lack of clear mesenchymal differentiation, these cases were reclassified as dedifferentiated endometrial carcinomas and were subsequently excluded from the carcinosarcoma category. No cases of Lynch syndrome were identified among carcinosarcoma patients, as all 4 MMRd cases were due to somatic MLH1 hypermethylation. In summary, we found that the rate of MMRd is markedly lower in uterine carcinosarcoma when compared with endometrial carcinoma. In the setting of MMR loss, a diagnosis of dedifferentiated carcinoma should be considered as almost half of the MMRd tumors which were called carcinosarcomas initially were reclassified as dedifferentiated on review. However, given the interobserver variability in the classification of carcinosarcoma versus dedifferentiated carcinoma a universal screening approach that includes uterine carcinosarcoma is still recommended.

The central nervous system (CNS) is composed of interacting compartments, including neurons, glia, vasculature, and cerebral spinal fluid. The extracellular matrix (ECM) links these compartments together through ligand-receptor interactions including integrins, cell adhesion molecules, and cell-surface glycoproteins. The ECM constitutes approximately 20% of the mature brain volume and is comprised of an intricate interconnected matrix of a variety of glycoproteins, proteoglycans, and glycosaminoglycans that stabilize connections between and among CNS compartments (Nicholson &

Uterine carcinosarcomas have few adjuvant treatment options. Programmed cell death ligand-1 (PD-L1) expression in these tumors may predict response to checkpoint inhibitor therapies. An increase in PD-L1 expression has been shown in endometrial carcinomas with mismatch repair (MMR) deficiencies; however, few studies have evaluated PD-L1 expression in uterine carcinosarcomas. We examined PD-L1 expression in 41 cases of uterine carcinosarcoma using combined positive scores (CPS) and tumor proportion scores (TPS), and correlated with MMR status, p53 expression, and epithelial histotype. In addition to confirming the diagnosis of carcinosarcoma, the epithelial components were stratified based on endometrioid versus serous histology. Thirty-three cases (80%) were positive for PD-L1, defined as a CPS score of ≥ 1 or a TPS score of ≥ 1%. Twelve cases (29%) showed high expression of PD-L1, defined as a CPS score of ≥ 10 or a TPS score of ≥ 10%. The majority of the morphologically adjudicated carcinosarcomas had a serous epithelial component (83%) rather than endometrioid (17%), which was reinforced by aberrant p53 staining predominantly within cases with serous morphology. The majority of carcinosarcomas showed at least focal PD-L1 expression, predominantly in tumor-associated immune cells. Carcinosarcomas with endometrioid morphology were significantly more likely to have high-level PD-L1 (5/7 vs. 7/34; P = 0.015). MMR-deficient carcinosarcomas were also more likely to have high-level PD-L1 (2/3 vs. 10/28); however, this did not reach statistical significance (P = 0.2) and overall MMR-deficiency was uncommon (3 cases, 7%). These findings suggest that PD-L1 may be additive to MMR testing as a predictive biomarker for checkpoint inhibitor vulnerability in carcinosarcomas.

Tumor cell expression of major histocompatibility complex (MHC) class I is required for antigen presentation and adaptive immune recognition. Absent or diminished MHC class I expression is thought to contribute to immunotherapeutic resistance in some epithelial tumors but has not been previously studied in cervical and vulvar carcinoma. Given that anti–programmed cell death 1 (PD-1) checkpoint inhibition is deployed for programmed cell death ligand 1 (PD-L1)-positive recurrent and metastatic cervical squamous carcinomas, identifying tumors with loss of MHC class I is of clinical interest to optimize the selection of immunotherapeutic candidates. Immunohistochemistry for PD-L1 and MHC class I combined A, B, and C heavy chains (MHC class I) was assessed in 58 human papillomavirus–associated cervical and vulvar lesions, including 27 squamous intraepithelial lesions (SILs) and 31 invasive squamous cell carcinoma (SCC). Although 84% of SCC and 22% of SIL were PD-L1-positive, 35.5% (11/31) of SCC and 18.5% (5/27) of SIL also showed clonal or complete loss of MHC class I. Loss of MHC class I expression was more common in PD-L1-positive (10/26, 38%) versus PD-L1-negative SCC (1/5, 20%). In summary, over one third of human papillomavirus–associated cervical and vulvar SCC show clonal or complete loss of MHC class I expression, including many PD-L1-positive cases. This suggests that the efficacy of checkpoint inhibitors targeting the PD-1/PD-L1 axis may be limited in a subset of cervical and vulvar squamous neoplasms due to an impaired ability to engage with the adaptive immune system related to loss of MHC class I expression.