The expression of cytokeratin 19 in lymph nodes was a poor prognostic factor for hepatocellular carcinoma after hepatic resection

Hepatocellular carcinoma (HCC) is the most common primary malignancy of the liver with an estimated annual death incidence of approximately 600,000 worldwide. In Taiwan, it is the second most common cause of cancer death and causes more than 7,500 deaths each year[1]. Surgical resection remains the most effective therapy in selected patients, but approximately 75% of patients with HCC have advanced unresectable diseases upon presentation. Compared with other malignancies such as lung cancer, esophageal cancer, renal cancer, gastric cancer, and intra-hepatic cholangiocarcinoma, the incidence of lymph node metastasis (LNM) in primary HCC is very low, and the prognosis is poor when LNM occurs[2‐11]. Nevertheless, despite this poor outcome, previous study has documented that the 5-year overall survival rate of HCC with LNM after hepatectomy and lymph node dissection was 13.6%, and 5 out of 22 patients had survival of more than two years[12]. Due to LNM’s dismal role in prognosis, it is imperative to identify those patients at risk for LNM for more aggressive treatment and detailed examination.

Cytokeratin19 (CK19) is a biliary epithelial cell marker and is generally expressed in intrahepatic cholangiocarcinoma (ICC) cells[13]. Studies have shown that the expression of CK19 in primary HCC is associated with poorer outcome[14]. There have also been studies demonstrating that CK19 expression in primary HCC is a significant risk factor for developing LNM[15‐17]. However, few studies have investigated the expression profile of CK19 in regional lymph nodes (LN) of HCC after hepatic resection. HCC with CK19 expression, speculatively, may be a special subtype of HCC with distinct clinical behavior from HCC without CK19 expression. Therefore, the purpose of this study was to examine the expression of CK19 in primary liver tumor and regional LN of HCC with and without lymph node metastasis. The correlation between CK19 expression in LN and LNM was determined. Clinico-pathological variables were also investigated to explore the impact of CK19 expression on survival outcome. Finally, the significance of HCC categorization according to CK19 expression was elaborated.

There were 34 (80.95%) males and 8 (19.05%) females. Twenty-eight (66.67%) patients were HBV carriers and 9 (21.42%) patients had chronic HCV infection. As for the severity of liver cirrhosis, 41 (97.62%) patients were Child-Pugh classification A whereas only one patient (2.38%) had Child-Pugh classification B liver cirrhosis. In terms of tumor T stage, there were 12 (28.57%) T1, 6 (14.29%) T2, 18 (42.86%) T3, and 6 (14.29%) T4 patients.

Table 1 summarizes the relationship between clinicopathological variables and CK19 expression in HCC. CK19 expression was positive in nine (21.43%) primary liver tumor specimens and five (11.90%) regional LNs; immunopositivity of CK19 in primary liver tumor was significantly correlated with CK19 expression in LN (Spearman correlation coefficient 0.704, P <0.001). Statistical analysis showed that LNM (Spearman correlation coefficient 0.427, P = 0.005) and tumor non-encapsulation (Spearman correlation coefficient 0.382, P = 0.013) were significantly related to CK19 expression in primary liver tumor. On the other hand, patient age, gender, cigarette smoking, alcohol consumption, hepatitis viral status, preoperative total bilirubin, preoperative AFP, preoperative alkaline phosphatase (ALK-P), Child-Pugh classification, tumor size, T stage, tumor rupture, vascular invasion, daughter nodules, and histologic grade were not related to CK19 expression in primary liver tumor.

Of 42 patients, 26 (61.9%) patients died during the follow-up period, including one patient who died from acute respiratory failure unrelated to HCC or liver cirrhosis. The 5-year OS rate was 32.5%, with median survival time of 22.08 months. Patients with CK19 (+) primary liver tumor or CK 19(+) LN had significantly poorer overall survival than those without CK19 expression (for primary liver tumor, 9.37 ± 3.28 (2.942 to 15.798) months versus 28.96 ± 20.96 (0 to 70.039) months, P = 0.026, and for LN, 9.37 ± 3.17 (3.165 to 15.575) months versus 28.96 ± 12.92 (3.642 to 54.278) months, P = 0.007, respectively) (Figures 1 and2). The 5-year OS rates for patients with CK19(+) primary liver tumor or CK19(+) LN were 11.1% and 0%, respectively, while those for CK19(−) primary liver tumor or CK19(−) LN were 45.8% and 43.3%, respectively. Vascular invasion (P = 0.005), daughter nodules (P = 0.009), positive resection margin (P = 0.029), and AJCC tumor stage (P = 0.020) significantly decreased overall survival for HCC (Table 2). In patients with N1 disease, although not statistically significant yet, CK19 expression in metastatic LN decreased the OS months (9.37 ± 3.17 versus 27.88 ± 8.25, P = 0.068). After classifying these N1 patients into three groups based on the expression of CK19 in primary liver tumor, we found that patients with CK19(+) LNM and CK19(+) primary tumor had a significantly poorer OS than patients with CK19(−) LNM and CK19(−) primary tumor (P = 0.042) (Figure 3). Cox regression analysis demonstrated that CK19 expression in regional LN (HR: 5.695 (1.85 to 17.52), P = 0.002) and daughter nodules (HR: 2.573 (1.50 to 4.40), P = 0.01) were independent poor prognostic factors for overall survival. CK19 expression in primary liver tumor, on the other hand, was not a significant independent prognostic factor after Cox regression analysis (Table 3). Figure 4 is the immunohistochemical microphotograph of primary liver tumor and regional LN.

During development of human liver, hematopoietic stem cells first differentiate into hepatic stem cells (hepatoblasts), then hepatic progenitor cells (HPC), and finally hepatocytes or biliary epithelial cells[19‐21]. Cytokeratins, on the other hand, are typical epithelial cell markers that are expressed in a tissue-specific and differentiation-dependent manner[22]. Differences in cytokeratin expression are routinely used to determine the primary cells of origin of malignant tumor[23, 24]. In normal human liver, hepatocytes express CK8 and CK18, while biliary epithelial cells express CK7 and CK19[25]. Hepatic progenitor cells, on the other hand, express markers specific for both hepatocytes and biliary epithelial cells. As progenitor cells differentiate into different cell lineages, they lose specific phenotypic characteristics. For example, CK19 is first lost when hepatic progenitor cells differentiate into hepatocyte lineage[14]. In other words, tumor cells of HCC should not express CK19 if they originate from hepatocytes. However, the current study and several previous reports showed that some HCC can express CK19, the marker specific for biliary epithelial cells[14, 15, 26‐30]. These consistent findings suggested that some HCC may develop from, instead of hepatocytes directly, but hepatic progenitor cells which express CK19[14, 25, 31, 32]. An earlier study that focused on immunohistologic expression of HCC patients who received hepatectomy indicated that about 10% of HCC with typical microscopic histologic features had cholangiocellular characteristics[14]. The study employed CK19 expression as the cholangiocellular marker and concluded that the cholangiocellular-marker positive HCC may originate from hepatic progenitor cells, or they may acquire the characteristics of cholangiocellular epithelium by metaplasia. Further studies are warranted to investigate the carcinogenesis of this special HCC subtype.

Our study found that CK19 expression in primary liver tumor was associated with a higher incidence of LNM and tumor non-encapsulation. An earlier published work reported that HCC with LNM, though occurring in a very low incidence, tended to be an infiltrating type of HCC and carried an extremely poor prognosis[12]. A study from our group also indicated that HCC encapsulation was associated with less vascular invasion and was a significant prognostic factor for better outcome in larger HCC. These studies suggested that LNM and tumor non-encapsulation were significant indicators for HCC invasiveness and poor outcome. In other words, CK19 expression of primary liver tumor may possibly indicate the worse prognosis for the patient. The survival analysis from the current study further strengthened this implication. Reports done by groups in China and by others stated that increased expression of CK19 in HCC was significantly correlated with LNM and a poorer outcome[14, 16, 17]. CK19 expression by HCC was also suggested to be a predictor of early postoperative recurrence and was associated with increased invasiveness and lymph node metastasis[16]. Studies focused on molecular classification and diagnosis of HCC also indicated that CK19 expression can be used as a marker of aggressive and advanced HCC[33, 34]. Our findings were consistent with these reports and we concluded that: (1) CK19 expression in primary liver tumor was an indicator of HCC invasiveness including LNM and non-encapsulation, and (2) CK 19 expression in primary liver tumor was a poor prognostic factor for overall survival.

Several studies, including the aforementioned ones, demonstrated the impact of CK19 expression by primary tumor on HCC outcome[14‐17, 25, 33, 34]. Nevertheless, few investigated the molecular signature of metastatic LN and the influence of these specific molecules on HCC outcome. Our study was the first one in the literature to examine the expression of CK19 by regional LN of HCC. The expression of CK19 in regional LN, which usually did not have epithelial cells, may indicate that the CK19(+) primary liver tumor cell had metastasized from liver to regional LN. We found that seven out of nine (77.8%) CK19 (+) HCC had LNM, and all patients who had positive CK19 expression in their LN (five patients) were also positive for CK19 expression in primary liver tumor. Therefore, we suggested that HCC with immunopositivity of CK19, meaning presence of hepatic progenitor cells, had a high propensity for LNM.

Our study shows that CK19 expression in primary liver tumor or regional LN was associated with poorer overall survival. However, after Cox regression multivariate analysis, only CK19 expression in LN (HR: 5.695 (1.85 to 17.52), P = 0.002) and the presence of daughter nodules (HR: 2.573 (1.50 to 4.40), P = 0.01) were independent poor prognostic factors for overall survival. In addition, in patients with LNM, CK19 expression in metastatic LN significantly decreased the overall survival, when compared with CK19(−) LNM in CK19(−) primary liver tumor. Subgroup analysis found that there was no significant difference between CK19 (+) LNM and CK19(−) LNM regarding most major clinicopathological factors. This indicated that the worst prognosis of CK19 (+) LNM was not attributed to other prognostic factors, but only to CK19 expression in metastatic LN. Our study, therefore, was the first report in the English literature to demonstrate the worst prognostic impact of CK19 (+) LNM on HCC after hepatectomy. The nature of CK19 (+) primary liver tumor with CK19(−) metastatic LN remains uncertain and mandates further investigation. As shown above, HCC with LNM carried a poor outcome[12]. Our study further explored the possibility that HCC with CK19(+) metastatic LN may be an HCC subtype with even worse prognosis. However, with appropriate treatment and follow-up, this subset of patients still can achieve a median overall survival of 9.37 ± 3.17 months after hepatectomy. Our earlier study demonstrated that LNM may not be a contraindication for curative resection for HCC. A more aggressive surgical treatment including lymph node dissection was suggested when LNM was suspected[12]. The present study echoed that finding and we further suggested that CK19 immunohistochemical staining of primary liver tumor and dissected LN should be performed after hepatectomy to determine LN status and to predict outcome.

This study had some drawbacks. First, the rarity of HCC with LNM resulted in a small sample size and rendered statistical analysis difficult. We believe that with a larger sample size, our findings will become even more significant and persuasive. Second, since this study was a retrospective hospital-based analysis, incomplete data collection were inevitable when reviewing records from a very long time ago. Selection bias might also exist when enrolling patients into this study. In addition, there is fundamental problem in using immunohistochemical staining in evaluating the expression profile of CK19. It requires experienced histopathologists for staining and interpreting the immunohistochemical examination result. Other techniques, including polymerase chain reaction, could be more objective and sensitive. Therefore, a well-designed prospective study with long-term follow-up is required to further validate our study.

In conclusion, our study showed that (1) CK19 expression in primary liver tumor is an indicator of HCC invasiveness including LNM and non-encapsulation, (2) CK19 expression in metastatic LN of primary HCC iss associated with an extremely poor outcome after operation, and (3) CK19 immunohistochemical staining of primary liver tumor and regional LN should be performed after hepatic resection to determine LN status and to predict outcome. Since CK19 was considered a marker of hepatic progenitor cells and biliary epithelial cells, the subset of HCC expressing CK19 in metastatic LN may represent an HCC category that originates from HPC or by metaplasia and carries an extremely poor prognosis. It is of clinical significance to identify these patients at risk for more aggressive surgical treatment and postoperative follow-up. A recent study also indicated that the activation of the epidermal growth factor (EGF)-EGF receptor signaling pathway is associated with the development of CK19 (+) HCC, and this pathway may account for the poor prognosis of patients[35]. The mechanism by which EGFR signaling pathway induces CK19 expression remains unknown. Further studies are warranted to elucidate the carcinogenesis of this subset of HCC.

Written informed consent was obtained from the patient for publication of this report and any accompanying images.