Peritumoral/vascular expression of PSMA as a diagnostic marker in hepatic lesions

Cholangiocarcinomas arise from the epithelial cells of the bile ducts. Although rare in US, these cancers are highly lethal because most are locally advanced at the time of initial diagnosis. According to the SEER database, approximately 8000 new cases were reported each year in US [1]. Despite multidisciplinary interventions, including surgery, chemotherapy and radiotherapy, there is limited improvement in survival. Only 10% of patients with extrahepatic cholangiocarcinoma and 8% of patients with intrahepatic cholangiocarcinoma survive for 5 years or longer [2]. Additionally, the actual number of cases is likely to be higher because these cancers can be hard to diagnose and may be misclassified. Intrahepatic cholangiocarcinoma considered a primary liver tumor, shares histological and immunohistochemical features with many other primary and metastatic tumors, especially pancreatic ductal adenocarcinoma. The differential diagnosis between primary cholangiocarcinoma and metastatic pancreatic ductal adenocarcinoma is extremely important for clinical management but there are no reliable molecular markers to differentiate these 2 morphological similar entities.

PSMA is a 100 kDa type II-transmembrane glycoprotein with folate hydrolase and neurocarboxypeptidase activity and is expressed in prostate epithelium and upregulated on the surface of prostatic adenocarcinoma cells [3]. PSMA has diagnostic and therapeutic importance and PSMA-based imaging technology (PSMA PET-CT) and targeted therapy have been used for prostate cancer management [4, 5]. Studies have shown that PSMA enzymatic activity is involved malignancy-driven neoangiogenesis and is expressed in the endothelium of tumor-associated neovasculature of breast, lung, thyroid, hepatocellular carcinoma (HCC), transitional cell carcinoma of the urinary bladder, and a small portion of prostate adenocarcinoma [6, 7]. However, these studies focused on malignancy at the primary site and none of them explored the metastasis loci in liver.

Recently, the presence of PSMA in primary cholangiocarcinoma was detected by PSMA PET-CT [8]. However, no similar findings were reported for metastatic pancreatic ductal adenocarcinoma in liver. Our study evaluates the histology and imaging correlation of PSMA for intrahepatic cholangiocarcinoma and metastatic pancreatic ductal adenocarcinoma. We also investigated PSMA expression in other hepatic masses that are commonly seen in daily practice, including primary and metastatic lesions that may potentially interfere with cholangiocarcinoma diagnosis. We speculated that PSMA expression is different between primary cholangiocarcinoma and metastatic pancreatic ductal adenocarcinoma in the liver.

We showed novel evidence that significantly enhanced tumor-associated neovascular PSMA expression is present in primary cholangiocarcinoma but not in metastatic pancreatic ductal adenocarcinoma, although the two are morphologically indistinguishable. Our finding is consistent with recent report that primary cholangiocarcinoma could be visualized in liver with PSMA traced PET-CT scan [8]. Although the case number was relatively limited, this finding could provide a potential marker for differentiating diagnosis.

Tumor cell PSMA expression was mainly reported in both benign and malignant prostate tissue, while neovascular expression was found in a wider spectrum of malignant neoplasms [6, 7, 9, 10]. Chang et al. identified high frequency of neovascular expression in epithelial tumors (such as carcinomas, including renal cell carcinoma, transitional cell carcinoma, pancreatic ductal adenocarcinoma, breast carcinoma and colonic adenocarcinoma), neuroendocrine tumors, melanoma, and non-vascular soft tissue sarcomas [6]. In contrast to previous studies, we focused on metastasis loci instead of primary lesions. Metastatic carcinomas, such as pancreatic ductal and colonic adenocarcinoma, showed reduced PSMA neovascular expression compared to their primary sites.

The exact role of PSMA in tumor-associated neoangiogenesis is not fully understood. An animal model study showed that angiogenesis was severely impaired in PSMA-null mice. This angiogenic defect occurred at the level of endothelial cell invasion through the extracellular matrix barrier [11]. The author also revealed PSMA as a principal component of a regulatory loop that modulates laminin-specific integrin signaling and GTPase-dependent, p21-activated kinase 1 (PAK-1) activity [11]. On the other hand, tumor desmoplasia and hypoxia were also associated with neoangiogenesis [12, 13]. A more recent study showed that prominent stromal fibrosis was common in pancreatic ductal adenocarcinoma and played a role in regulation of angiogenesis [13]. Based on the above findings, we can propose two possible reasons for the different PSMA profiles between primary and metastatic sites. First, PMSA expression is related to differences in the severity of hypoxia between the primary and metastatic site. Generally, hypoxia is more common at the primary site and more severe with large tumors. In contrast, metastatic sites have more available blood supply and thus less risk of hypoxia stress. Therefore, hypoxia-induced angiogenesis would be less prominent in metastatic site, and PSMA, as a part of neoangiogenesis, would be negatively regulated. Second, desmoplasia appears more frequently in primary sites compared to metastatic loci, at least in the case of pancreatic ductal adenocarcinoma. Therefore, desmoplasia-related neoangiogenesis, as well as PSMA, would be down-regulated in metastatic lesions, leading to PSMA loss in metastatic malignant tissues.

To better encompass the scope of lesions that are frequently seen in liver, we also explored PSMA expression in HCC, the most common primary malignancy in liver. In our study showed, 68% of the cases of HCC had neovascular PSMA expression, consistent with previous studies. However, the positive rates among previous reports showed disagreement [7, 10, 14]. Tolkach et al. reported more than 80% of tumors showed vascular expression in 158 cases of HCC [10]. However, in another analysis, the author found weak PSMA expression in only 3 out of 44 HCC cases (6.8%) [7]. The inconsistent observation might reflect the heterogeneous nature in PSMA expression in HCCs. In fact, we found that PSMA peritumoral/vascular expression was significantly higher in tumors arising from cirrhotic background with hepatic virus infection. Per our data, 90% of cases of HCC with cirrhosis were PMSA peritumoral/vascular positive, much higher than expression in non-cirrhotic HCC (6/12, 50%). Further analysis showed 73.3% PSMA positive cases had Hepatitis C or Hepatitis B infection, significantly higher than the non-viral infected cohort. Since previous studies did not provide cirrhotic information, a definitive conclusion cannot yet be made. Regardless, our findings shed light on the role of PSMA as an important ancillary marker to help differentiate HCCs from cirrhotic nodules in morphologically challenging cases, since cirrhotic liver background has shown to be consistently negative for PSMA peritumoral/vascular expression in our study and many previous reports [15‐18]. Other clinicopathological factors that might interfere PSMA expression included tumor differentiation. Recently, Jiao et al. reported that in 103 cases of HCC cohort, patients with moderate/poorly differentiated tumors were more likely to have high PSMA expression, compared to patients with well-differentiated tumors. Patients with high tumor stage and lymph node metastasis tended to have high PSMA expression [14]. Our study also found a similar trend, as more than 80% patients with Grade 3 HCCs had PMSA peritumoral/vascular expression compared to only two-thirds of patients with Grade 2 HCCs. However, no statistical significance was reached due to limited sample volume. Additionally, the mechanism behind higher frequency of PSMA expression with cirrhotic background is not well characterized yet. Whether PSMA played a role in poorly differentiated HCC carcinogenesis needs further investigation as well.

To the best of our knowledge, we are the first group to report lack of PMSA expression in hepatic adenoma regardless of subtypes. Based on the known mechanism of PSMA in neoangiogenesis, our finding was expected and consistent with results from other benign tumors. Mhawech-Fauceglia et al. showed PSMA was negative in all 846 benign tumors they tested, which included neurofibroma, lipoma, leiomyoma, colon polyps, thyroid adenoma and others [7].

We only noticed PMSA expression in tumor cells of metastatic prostate cancer. Additionally, it was not surprising that only 62.5% tumors were PMSA positive in our study as a very similar expression rate was reported previously in a larger cohort [7]. Overall, 93 out of 141 (66.0%) prostate adenocarcinoma cases were cellularly positive for PSMA. Treatment history and hormonal secretion status may affect the expression of PSMA. However, we also found a low rate of peritumoral/vascular expression in metastatic prostate cancer. Morphological and clinical correlation were need in diagnosis.

In summary, our study profiled PSMA expression in common hepatic masses. We identified significantly enhanced tumor-associated neovascular PSMA expression in primary cholangiocarcinoma and not identified in metastatic pancreatic ductal adenocarcinoma. We also showed higher frequency of neovascular PSMA expression in HCCs arising from cirrhotic liver background and provided a potential marker for differentiating tumor nodules from cirrhotic nodules in difficult cases. Although the mechanism was not fully uncovered, the unique role of PSMA in tumor neoangiogenesis might extend the diagnostic and therapeutic value of PMSA analysis from prostate cancer to more solid malignancies.