STK11 (LKB1) immunohistochemistry is a sensitive and specific marker for STK11 adnexal tumours.

The clinicopathologic features of the STK11 adnexal tumors (n=17) are presented in Table 1 . Five cases were previously reported in the literature as either STK11 adnexal tumors (n=3) 1 , 3 or unclassified sex cord-stromal tumors (n=2). 8 Patient age ranged from 16 to 74 (median 43) years. Ten patients presented with pelvic masses, and other presentations included weight gain, suspicion for ectopic pregnancy, symptoms of urinary tract infection, vaginal discharge and acute abdominal pain. Tumors frequently involved the ovaries, fallopian tubes, para-adnexal soft tissue, and peritoneal surfaces. The tumors ranged in size from 5 to 20 (median 10.8) cm. All but one patient were treated with surgery, five received chemotherapy, three received an aromatase inhibitor, and one was treated with the mTOR inhibitor everolimus. Follow-up data was available for 12 patients and ranged from 5 to 172 (mean 59.3) months. All but two patients experienced a recurrence/metastasis and one of the patients without recurrence had only 5 months of available follow up. All patients with recurrences had abdominopelvic disease, and one also developed a pulmonary metastasis. Late recurrences were common, and occurred from 10 to 108 months after the initial diagnosis. Two patients died of disease. Three patients had a history of PJS, one had germline mutations in both PALB2 and MUTYH , and three had negative germline mutation testing for STK11 . There was significant heterogeneity of histologic patterns both within and among STK11 adnexal tumors ( Figure 1 ). Microscopically, they showed a spectrum of growth patterns, often characterized by interconnecting cords and trabeculae of tumor cells within a myxoid, edematous, or hyalinized stroma. Other architectural patterns included cystic, microcystic, follicular, tubular, reticular, cribriform, papillary, nodular, solid, and nested. Cytologically the tumor cells were typically cuboidal or columnar, but were sometimes spindled with variable amounts of eosinophilic to clear cytoplasm; occasionally signet ring-like cells were present ( Figure 2 ). Nuclei were round to ovoid, with vesicular chromatin and prominent nucleoli. Mitotic activity varied from rare to 20 mitoses per 10 high power fields. One tumor in a PJS patient (case 16) was associated with a sex cord tumor with annular tubules (SCTAT). The immunohistochemical profile of the STK11 adnexal tumors included in this study is summarized in Table 2 . STK11 adnexal tumors showed a non-specific immunophenotype, including frequent, but variable, positivity for AE1/AE3 (12/13), CAM5.2 (6/7), CK7 (6/10), EMA (5/14), calretinin (16/16), inhibin (14/17), WT1 (12/13), estrogen receptor (14/15), progesterone receptor (7/12), androgen receptor (5/5), CD10 (4/9), D2–40 (4/7), desmin (9/10), and CD56 (3/5). Tumors showed limited expression of SF1 (patchy in 4/11 cases), CD99 (patchy in 2/4), GATA3 (patchy in 2/11 cases), CK5/6 (rare cells in 2/5), p16 (patchy in 2/4), CD117 (patchy in 2/2), glypican-3 (patchy in 1/2), AFP (patchy in 1/3), PAX8 (patchy in 1/11), BEREP4 (focal in 1/3), and p63 (rare cells in 1/3). They were negative for CK20 (n=6), CDX2 (n=3), caldesmon (n=2), panTRK (n=2), synaptophysin (n=3), chromogranin (n=2), HMB45 (n=2), S100 (n=3), SALL 4 (n=4), TTF-1 (n=6), and claudin-4 (n=1). β-catenin showed membranous expression (n=4) and p53 was wild-type (n=6) in all stained tumors. BAP1 (n=6), SMARCA4/BRG1 (n=6), and SMARCB1/INI-1 (n=3) all showed retained/intact nuclear expression. PD-L1 and FOLR1 were negative in one tumor (case 14). DNA sequencing was performed on 13/17 (76%) STK11 adnexal tumors using a targeted panel that includes the STK11 gene ( Table 3 ). The four tumors (cases 2, 12, 16, and 17) which were not sequenced for STK11 were diagnosed based on morphologic and immunohistochemical features which were consistent with STK11 adnexal tumors, including loss of STK11 expression. One case without sequencing was from a patient with known PJS. Of the 13 sequenced tumors, STK11 alterations were identified in 12/13 (92%). Mechanisms of STK11 inactivation included loss of function point mutations (nonsense, frameshift, or splice site variants), single copy loss, and a single case with a large structural rearrangement involving STK11 ; in 7 of the 8 cases which were sequenced on a platform which reported copy number variants and loss of heterozygosity, there was evidence of biallelic inactivation, either through loss of the second copy or copy-neutral loss of heterozygosity. Case 1 showed single copy deletion of STK11 , without a loss of function point mutation. One tumor (case 13) had a structural rearrangement which was likely truncating. In one tumor (case 5), no STK11 alteration was identified, but notably the molecular assay used in this tumor is not validated to report copy number variations, and a diagnosis of STK11 adnexal tumor was made based on the morphology and immunophenotype. The two patients with known PJS whose tumors were sequenced had pathogenic point mutations in STK11 . Recurrent copy number changes included gain of chromosome 1q and loss of 1p and 11q. In total, 122 tumors were stained with STK11, including 17 STK11 adnexal tumors, 10 FATWO, 22 adult granulosa cell tumors (AGCT), 10 juvenile granulosa cell tumors, 4 Sertoli-Leydig cell tumors, 2 Leydig cell tumors, 1 Sertoli cell tumor, 1 steroid cell tumor, 4 extra-ovarian sex cord-stromal tumors, 16 ovarian endometrioid carcinomas, 8 tubo-ovarian high-grade serous carcinomas, 5 peritoneal malignant mesotheliomas, 5 ovarian mesonephric-like adenocarcinomas, 14 ovarian carcinosarcomas, 2 pelvic plexiform leiomyomata, and 1 ovarian solid pseudopapillary tumor. All 17 STK11 adnexal tumors showed complete loss of cytoplasmic staining for STK11 in tumor cells, with retained staining in non-neoplastic stromal and inflammatory cells as a positive internal control ( Figure 3 ). 103/105 (98%) other tumor types showed intact cytoplasmic staining for STK11 in all tumor cells; one endometrioid carcinoma with mucinous differentiation included on a TMA showed complete loss of staining, and one high-grade serous carcinoma showed subclonal loss of expression ( Figure 4 ). Sequencing was performed on 42/105 (40%) of the potential morphologic mimics (5 AGCT, 8 ovarian endometrioid carcinomas, 8 tubo-ovarian high-grade serous carcinomas, 2 FATWO, 5 peritoneal malignant mesotheliomas, 5 ovarian mesonephric-like adenocarcinomas, 8 ovarian carcinosarcomas, and 1 solid pseudopapillary tumor). None showed STK11 inactivating alterations.

STK11 adnexal tumors and other tumors in the differential diagnosis were retrieved from the files of five institutions: Brigham and Women’s Hospital (Boston, MA), Boston Children’s Hospital (Boston, MA), Stanford University Medical Center (Stanford, CA), King Edward Memorial Hospital (Perth, Australia), and The Belfast Health and Social Care Trust (Belfast, UK). This study was approved by the institutional review board at Brigham and Women’s Hospital. Because many patients were either deceased or lost to follow up, obtaining informed consent was not possible. Representative hematoxylin and eosin-stained slides were reviewed. STK11 adnexal tumors were mostly identified based on the morphologic and immunophenotypic appearance with molecular confirmation, but in a small number of cases molecular profiling was not performed and the diagnosis was based on the morphology and immunophenotype (see later). Cases without molecular profiling performed as part of their initial workup were sequenced using a tumor-only, hybrid-capture based DNA sequencing panel at Brigham and Women’s Hospital, as previously described. 7 IHC for STK11 (LKB1) was performed on 4 μm thick whole tissue sections (n=85) or tissue microarrays (TMA, n=37) cut from formalin-fixed paraffin-embedded tumor blocks using an anti-STK11 monoclonal rabbit antibody (clone D60C5F10; 1:1000 dilution, Cell Signaling, Danvers, MA) and pressure cooker antigen retrieval (0.01 M citrate buffer; pH, 6.0). Tissue microarrays were composed of 1.5 mm cores, in duplicate for each tumor. Appropriate positive and negative controls were assessed. STK11 staining was scored as either positive/intact (positive cytoplasmic staining in tumor cells) or negative/lost (complete loss of cytoplasmic staining in tumor cells and intact staining in internal controls). Other immunohistochemical stains were performed using standard protocols at referring institutions during routine clinical workup of the tumors.

STK11 IHC is a highly sensitive and specific marker for STK11 adnexal tumors and can help distinguish them from histologic mimics. Compared to sequencing, STK11 IHC is more cost-effective, faster, and possibly more sensitive, particularly if copy number alterations are not assessed on a molecular assay. Loss of expression of STK11 appears to be a reliable surrogate for molecular studies to differentiate STK11 adnexal tumors from morphologic mimics. Our data suggest that sequencing is no longer required to diagnose STK11 adnexal tumors if STK11 IHC supports the diagnosis in the appropriate morphologic context.

STK11 adnexal tumor is a rare neoplasm, but accurate recognition is important given its aggressive clinical behavior and association with PJS. However, the diagnosis may be challenging due to its morphologic and immunophenotypic heterogeneity which results in a broad differential diagnosis. In this study, we showed that all STK11 adnexal tumors exhibited complete loss of STK11 cytoplasmic expression, and of the 105 morphologic mimics, all but 2 exhibited retained staining, indicating that STK11 IHC is both a sensitive and specific marker for this entity. All sequenced tumors in our series that were morphologically and immunohistochemically consistent with STK11 adnexal tumor demonstrated inactivating alterations in STK11 , except for one tumor. For case 5, no STK11 single nucleotide variants or small insertion-deletion variants were identified. However, the assay used for that case was not validated to report copy number changes in STK11 . It is therefore possible that STK11 was inactivated in that tumor through an alternate mechanism such as a deep deletion, a large structural variant, or promoter methylation. In both the initial series of these neoplasms 1 and in our current study, sequencing identified only a single altered STK11 allele in several cases (e.g., our case 1). However, these tumors showed complete loss of STK11 staining, suggesting that both alleles were inactivated even though only a single allelic alteration was identified by sequencing. Intact STK11 staining was seen in non-neoplastic (internal control) cells in patients with PJS, indicating that inactivation of a single allele alone would not result in loss of STK11 expression. These cases highlight the important point that STK11 IHC staining may be more sensitive than some molecular assays (i.e., those that do not report copy number changes in STK11 ), and can help to contextualize sequencing results (e.g., when only single copy inactivation is found). In addition to identifying STK11 as a very useful diagnostic immunohistochemical marker of STK11 adnexal tumors, this series has significantly expanded the number of reported cases 1 , 3 , 4 from 25 to 39. Five of the tumors reported in the current series were previously published: cases 3, 8 and 9 as STK11 adnexal tumors, 1 , 3 and cases 12 and 16 as sex cord-stromal tumors. 8 While the main findings of the seminal series 1 have been reinforced herein, we identified three novel aspects of these tumors: 1) frequent desmin expression (90%), further emphasizing the polyphenotypic immunoprofile of STK11 adnexal tumor, 2) STK11 rearrangement as an uncommon mechanism of inactivation, and 3) rare pulmonary metastases in addition to more common peritoneal dissemination. An interesting point is that both STK11 adnexal tumor and uterine tumor resembling ovarian sex cord tumor (UTROSCT) 9 – 12 share a polyphenoptypic immunophenotype with expression of epithelial, smooth muscle and sex cord markers and hormone receptors, although the two tumors are associated with different molecular events. Combining prior reports and the cases described in this series, STK11 adnexal tumor has an aggressive behavior with 81% of patients experiencing a recurrence and 19% dying of disease. Recurrences occurred up to 108 months after initial surgery. 41% of reported patients have PJS; this may be an underestimate given that some patients diagnosed with this neoplasm may have PJS but have not been investigated for it. Given the frequent and late recurrences and the possibility of STK11 germline mutations, extended clinical follow up and genetic counselling is recommended. The product of the STK11 tumor suppressor gene, serine/threonine kinase 11 (STK11, also known as LKB1) is part of a protein complex that regulates cell metabolism, growth, and apoptosis through activation of adenosine monophosphate-activated protein kinase (AMPK). 13 AMPK then regulates energy homeostasis by phosphorylation of targets including mTOR complex 1 (mTORC1) and HIF-1α. 14 , 15 Germline STK11 loss-of-function mutations underlie PJS, which is characterized by distinctive intestinal hamartomatous polyps and mucocutaneous pigmented lesions. There is an increased incidence of pancreatic carcinoma, breast carcinoma, gastric-type cervical adenocarcinoma, and SCTAT. 16 STK11 inactivating alterations have been identified in several tumor types, including lung and pancreatic adenocarcinoma, and cholangiocarcinoma. 13 In the gynecologic tract, STK11 alterations have been identified in a small percentage of cervical gastric-type adenocarcinomas, 17 , 18 invasive stratified mucin-producing adenocarcinomas, 18 and ovarian mucinous neoplasms as well as SCTAT. 19 Importantly, these tumors, most of which show mucinous differentiation, are not in the morphologic differential diagnosis of an STK11 adnexal tumor. In a prior study of 30 ovarian carcinomas, two tumors had loss of heterozygosity at the STK11 locus and one had a missense mutation in STK11 , but none demonstrated biallelic inactivation of STK11 . 20 The single case in our series with subclonal loss of expression, a high-grade serous carcinoma, possibly had a clone with a deep deletion of the STK11 locus secondary to the genomic instability which is common in that tumor type. Loss of expression was also identified in an ovarian endometrioid carcinoma with mucinous differentiation, which is not a feature of STK11 adnexal tumors. Therefore, although ovarian carcinomas may occasionally show STK11 inactivation, it appears to occur infrequently in tumors that are in the morphologic differential diagnosis of STK11 adnexal tumors, exemplified by the retained STK11 staining in the other tumors in our series. Histologically, STK11 adnexal tumors show considerable variability in architecture and cytomorphology, and therefore have a broad differential diagnosis, including endometrioid carcinoma, FATWO, sex cord-stromal tumors (particularly AGCT), peritoneal mesothelioma, mesonephric-like adenocarcinoma, high-grade serous carcinoma, and UTROSCT. In younger patients, co-expression of keratin and desmin can also raise the possibility of a desmoplastic small round cell tumor. 3 FATWO is perhaps the most challenging entity in the differential diagnosis for STK11 adnexal tumor, since both typically have a paratubal localization, variable histomorphology, and a nonspecific immunoprofile. Although FATWOs, in contrast to STK11 adnexal tumors, do not usually show a myxoid matrix and often have a characteristic sieve-like architecture, it can be challenging to distinguish these two entities based on morphology alone. 1 , 2 While STK11 mutations had been reported in FATWO prior to the initial description of STK11 adnexal tumor, 21 , 22 there is an emerging consensus that these “ STK11 -mutant FATWO” represent STK11 adnexal tumors. 1 , 2 It is also likely that some prior reports of malignant FATWOs represent STK11 adnexal tumor. In the past, sequencing for STK11 variants was the most reliable way to differentiate these entities. However, the data presented herein show that immunohistochemistry for STK11 can act as a reliable surrogate. Due to their tubular or gland-like formation, spindled cells, and keratin and hormone receptor expression, STK11 adnexal tumors can be misdiagnosed as Mullerian adenocarcinomas, particularly endometrioid carcinoma. Extensive sampling can be helpful to identify features typical of endometrioid carcinoma, including overt gland formation, squamous or mucinous differentiation, or endometriosis. Positive staining for claudin-4 and PAX8 (almost always negative in STK11 adnexal tumor) 1 also supports a diagnosis of endometrioid carcinoma, although some cases are negative. 23 STK11 adnexal tumor may also morphologically mimic mesonephric-like adenocarcinoma due to the variety of architectural patterns and an overlapping immunophenotype. In contrast to mesonephric-like adenocarcinoma, STK11 adnexal tumors are usually positive for ER and PR and negative for PAX8, TTF1 and GATA3. In problematic cases, molecular profiling may be helpful to differentiate endometrioid carcinoma and mesonephric-like adenocarcinoma from STK11 adnexal tumors, as the latter harbors STK11 mutations, but not KRAS, NRAS , PTEN , CTNNB1 , or ARID1A mutations which are characteristic of endometrioid and mesonephric-like adenocarcinomas. 24 – 26 Sex cord-stromal tumors and STK11 adnexal tumors can both display cords, trabeculae, nests, tubules, and solid growth. Although sex cord-stromal tumors usually arise within the ovary, they may occasionally be extra-ovarian in origin, 27 like STK11 adnexal tumors. Furthermore, both tumors share positivity for markers such as calretinin, inhibin, and WT1. However, sex cord-stromal tumors frequently have estrogenic or androgenic manifestations, which are rare in STK11 adnexal tumors. 2 Furthermore, molecular studies can also be helpful, as sex cord-stromal tumors often have FOXL2 or DICER1 mutations. 28 In a similar vein, STK11 adnexal tumors in close proximity to the uterus can be misdiagnosed as a UTROSCT, as these tumors share morphologic and immunohistochemical features, but UTROSCT harbor recurrent fusions involving ESR1 or GREB1 and NCOA1–3 , but do not have STK11 mutations. 12 While PJS-associated SCTAT have STK11 mutations, 29 they are morphologically easily distinguishable from STK11 adnexal tumors. Mesothelioma also shares architectural patterns and cytomorphology with STK11 adnexal tumor and has a similar immunoprofile, including common positivity for calretinin, D2–40, and WT1, and negative staining for claudin-4. 30 However, mesothelioma typically forms confluent sheets over much of the peritoneal surface, unlike STK11 adnexal tumors which form larger localized masses. 31 Also, although mesotheliomas may rarely exhibit STK11 mutations, 32 other alterations seen in mesothelioma, such as BAP1 mutations and homozygous deletions of MTAP have not been identified in STK11 adnexal tumors. Finally, ER, PR, and desmin are usually negative in mesotheliomas, 30 , 33 but most STK11 adnexal tumors are positive for these markers. This study has several limitations. Although STK11 IHC appears to have excellent sensitivity and specificity, the STK11 antibody is not currently available in most pathology laboratories and only a relatively small number of STK11 adnexal tumors and mimics were included in this study. Also, about a third of the morphologic mimics were examined as part of a TMA, which potentially limited detection of cases with patchy expression or subclonal loss. Additionally, molecular confirmation of the diagnosis of STK11 adnexal tumor was not possible in all cases. Finally, due to the fact that many of the cases were consults, only representative slides for each case were available for review, precluding a rigorous examination of morphologic features.

STK11 adnexal tumor is a rare, recently described neoplasm which is associated with Peutz–Jeghers syndrome (PJS); in the original report of this neoplasm, 47% of patients were known to have PJS with STK11 germline mutations. 1 – 4 Based on the 25 previously reported cases, tumors present at a mean age of 39 (range 16 – 66) years, and predominately originate in the para-adnexal soft tissues and can secondarily involve the fallopian tube and ovary. Tumors follow an aggressive course, with frequent metastases (50%), recurrences (82%), and a mortality rate of 18%. 1 , 3 , 4 STK11 adnexal tumors are remarkable for their morphologic heterogeneity. Architecturally, they can show solid, papillary, cribriform, corded, and glandular growth. Furthermore, their immunohistochemical profile is non-specific: they have a “polyphenotypic” immunohistochemical profile with positivity for keratins and variable expression of hormone receptors, mesothelial and sex-cord markers. Due to their heterogeneous morphology and immunoprofile, STK11 adnexal tumors have a broad differential diagnosis, 1 – 3 including Mullerian carcinomas, mesothelioma, female adnexal tumor of Wolffian origin (FATWO), and sex cord-stromal tumors. Currently, the diagnosis of an STK11 adnexal tumor requires molecular confirmation of STK11 inactivation/alterations. STK11 (also known as liver kinase B1 or LKB1) immunohistochemistry (IHC) has been studied as a surrogate for STK11 mutations in several tumor types, including pulmonary, 5 pancreatic, and biliary neoplasms. 6 However, it has not been previously examined in the context of STK11 adnexal tumors. Herein, we investigate the diagnostic utility of STK11 IHC in a cohort of STK11 adnexal tumors and their morphologic mimics in the hope that this may represent an immunohistochemical marker that can act as a surrogate for molecular alterations in STK11 . As STK11 adnexal tumor is a recently described entity, we also sought to expand the number of reported cases and describe novel clinicopathologic features of this tumor.