Correlation of hK6 expression with tumor recurrence and prognosis in advanced gastric cancer

A total of 129 samples were obtained from patients from January 2007 to November 2011 who underwent curative surgery for advanced gastric cancer in HuiZhou Municipal Central Hospital of GuangDong. Resection samples were confirmed to be primary advanced gastric cancer by clinical pathology, none of the patients received any chemotherapy, radiotherapy and other adjuvant therapy prior to operation. 52 blocks from adjacent noncancerous gastric tissues (at least 5 cm away from the cancer margin) were obtained from the patients. In addition, 36 specimens obtained from patients who underwent surgery and were confirmed to be gastric ulcer used as controls in the same period.

Patients^, clinicopathological characteristics such as gender and age, tumor location, size, differentiation, T stage, lymph node metastasis, TNM stage and whether recurrence were retrospectively reviewed. The mean age was 60 years (range: 28 ~ 80 years)with 56 women and 73 men. All the patients were staged based on the TNM classification of the American Joint Committee on Cancer (AJCC, 7th Edition criteria, 2010) [7].Tumor location in gastric fundus/cardia was 21 cases, gastric body 32 cases, gastric antrum 76 cases. Tumor size was greater than 5 cm in 70 cases, and less than 5 cm in 59 cases, 16 cases classified as well or moderately differentiated(including tubular and papillary adenocarcinoma), 113 cases as poorly differentiated (poorly differentiated adenocarcinoma and other cell types such as mucinous adenocarcinoma and signet ring cell carcinoma and so on), 19 cases were at T2, 55 cases were at T3 and T4 respectively. N0/N1 was found in 49 cases, N2/N3 in 80 cases, 43 cases withI/II stage, 86 cases with III/IV stage, written informed consent was obtained from all patients or their families. The study was approved by the Institute^,s Ethics Committee of HuiZhou Municipal Central Hospital. All the specimens were fixed in 10% neutralized maldehyde, embedded in paraffin, cut into 3-μm thick sections, and mounted on glass slides for immunohistochemistry analysis.

Immunohistochemistry was performed using the Horseradish Peroxidase two-step method, which avoids the interference of endogenous biotin, thus reducing the nonspecific false-positive background staining. Immunohistochemistry was carried out in eight steps according to the instructions: 1, The paraffin sections were deparaffinized in xylene, rehydrated through gradient ethanol. 2, Antigen retrieval was performed in Tris-EDTA buffer (pH 9.0) with a pressure cooker. 3, Block the endogenous peroxidase with 3% hydrogen peroxide. 4, Then incubated with the purified rabbit polyclonal antibody(AP6325a, Abgent, San Diego, USA), antigen retrieval and dilution were determined by preliminary experiments. 5, Detection of hK6 immunocomplex was incubated with the PV9000 kits (Golden bridge Biotechnology Co, Ltd, Beijing, China). 6, Stained with diaminobenzidine substrate chromogen solution (Goldenbridge Biotechnology Co, Ltd) until the optimal staining was achieved. 7, The sections were counterstained with hematoxylin, differentiated, returned blue. 8, Finally, dehydrated through gradient alcohol, cleared in xylene, neutral gum mounted. Then observed under the optical microscope.

All slides were assessed separately by two senior pathologists with no prior knowledge of clinicopathological parameters, the final score for their average. For evaluation of hK6 immunohistochemical staining, the percentage of positive cells was evaluated quantitatively 10 fields within the tumor were selected randomly, expression in 1000 cancer cells (100 cells per field) was evaluated using a high-power magnifcation (400×) and an average was evaluated quantitatively and scored as: 0 for the percentage of positive cells ≤ 10%, 1 for staining of 11 ~ 25%, 2 for staining of 26 ~ 50%, and 3 for staining of 51 ~ 75%, 4 for staining of > 75% of the cells examined [6], the staining intensity was respectively graded semiquantitatively as follows: 0 no signal, 1 weak, 2 moderate, and 3 strong according to the previous literatures [8], a total staining score of 4 or more was graded as immunohistochemical positive.

All these patients received fluorouracil based adjuvant chemotherapy after surgery. As a regular follow-up, all the patients were checked every 3 months during the first 2 years and every 6 months during the third to the fifth year, once a year thereafter. Medical work-up consisted of history and physical examination, hematology and biochemical tests including serum CEA, CA19-9 and CA72-4, and imaging including chest radiography, barium meal, abdominal ultrasonography (US), computed tomography (CT), and endoscopy according to the clinical situation. The follow-up time had the day of surgery as a starting point, the time of tumor recurrence and death were recorded, these two points were evaluated for prognostic analysis. If the follow-ups were incomplete, patients or their families were contacted by telephone, the closing date for follow-up was May 2012, the median follow-up period after surgery was 25 months (range 4–64 months).

Statistical analysis was carried out using the SPSS 13.0, the relationships between hK6 expression and patient clinicopathological characteristics were analyzed with the Chi-square (χ²) and Fisher’s exact test. For survival analysis, two end points were examined:recurrence-free survival (RFS) and overall survival (OS) respectively. RFS is defined as the time interval between the date of surgery and the date of first detected recurrence or metastasis, and OS is defined as the time interval between the date of surgery and the date of death or the date of last follow-up for those who were alive at the end of the study. The Kaplan–Meier model was used to examine survival between the patients positive and negative hK6 expression. The significance of any difference in the survival curves was determined with the log-rank test. Variables achieving a significance level in the univariate analysis were subsequently introduced in a forward LR proportional hazard analysis (Cox’s model)to identify those variables independently associated with survival, the p value of 0.05 (or less)was considered as statistically significant.

Different proteins were expressed in different location of the cell, such as cytoplasm, membrane, nuclear [9, 10].hK6 positive staining was mainly expressed in the cytoplasm of gastric cancer cells as brown or dark brown granules (Figure 1) also was observed in the adjacent noncancerous and gastric ulcer epithelial cells by immunohistochemistry. The positive rate of KLK6 expression in gastric cancer tissues was 77.5% (100/129), significantly higher than that in adjacent noncancerous tissues (36.5%) and ulcer (33.3%) tissues respectively (P <0.001; Table 1).

The correlation between hK6 expression levels and clinicopathological characteristics is as summarized in Table 2, hK6 expression significantly correlated with TNM stage (P = 0.005), vascular invasion (P = 0.037) and perineural invasion (P = 0.035). hK6 positive tumors more frequently had later stage, an increased incidence of vascular and perineural invasion, on the other hand, no statistical significant difference was observed between its expression and gender, age, tumor site, size, differentiation, T stage, lymph lodes metastasis or margin (P > 0.05).

As the follow-up ended, 68 (52.7%) patients had died, A total of 61 patients died because of tumor progression and 7 due to other causes. In all, there were 69 (53.5%) patients developed tumor recurrence, 31 of whom developed as loco-regional relapse, 19 as distant metastases (11 in liver, 3 in lung, and 5 in bone), 14 as peritoneal recurrence and 5 as remnant; All but eight patients with tumor recurrence died from it.

Kaplan–Meier survival curves indicated patients with positive hK6 expression were more likely to have both a shorter RFS (P = 0.002, Figure 2A) compared to patients with negative hK6 expression. Univariate analysis expression with regard to RFS, using the Cox proportional hazard regression model, showed that hK6 expression, TNM stage, tumor size, T stage, lymph lodes, margin and were influencing factors for RFS, When hK6 expression along with other factors were examined in a multivariate model, TNM stage, hK6 expression, tumor size, margin were independent predictors of tumor recurrence (Table 3), indicating hK6 positive patients are twice more likely to have a recurrence than hK6 negative ones.

According to the survival analysis, Kaplan–Meier survival analysis showed that patients with positive hK6 expression have lower OS (P = 0.002, Figure 2B), univariate analysis revealed that positive hK6 expression, TNM stage, tumor size, lymph node, T stage, and margin are important factors influencing the OS rates, when these factors were involved in a multivariate analysis (Table 4), TNM stage and hK6 were independent prognostic factors of OS.

Multivariate analysis indicated that hK6 expression was one of the independent prognostic factors of overall survival for the patients with gastric cancer next to TNM stage (P = 0.011).