Main

Pancreatic ductal adenocarcinoma is the fourth leading cause of cancer death in the US and about 32 300 patients will die from this disease in 2006.1 The vast majority of patients with pancreatic ductal adenocarcinoma present with advanced non-resectable disease because of the lack of sensitive and specific tools to screen for early disease. Unfortunately non-surgical treatments such as chemotherapy, radiation therapy, and immunotherapy are minimally effective for these patients.2 The development of early detection and novel treatment strategies are the keys to improve prognosis. A better understanding of pathogenesis of this cancer at the molecular level including identification and characterization of novel genes involved in the pathogenesis are essential to achieve these goals.

Several recent studies have shown that a novel gene, maspin, is overexpressed in the vast majority of pancreatic ductal adenocarcinomas but not in normal ductal epithelium.3, 4 Maspin (SERPINB5), a serine proteinase inhibitor, was first identified as a potential tumor suppressor on the basis of its differential expression between normal mammary epithelial cells and human breast carcinoma cell lines.5 Several studies have demonstrated that maspin might be a prognostic tumor marker. For example, high expression of maspin has been reported to be associated with a better prognosis in oral squamous cell carcinoma,6 thyroid carcinoma,7 colon adenocarcinoma8 and prostate adenocarcinoma,9 whereas in breast carcinoma,10 maspin expression was associated with a poor prognosis. The data on lung carcinomas11, 12, 13 and ovarian epithelial carcinomas14, 15, 16 have been conflicting. Although it has been demonstrated that human pancreatic ductal adenocarcinoma acquired maspin expression through hypomethylation of the maspin promoter,17, 18 little is known about the prognostic significance of maspin protein expression in human pancreatic ductal adenocarcinoma.

In this study, we investigated maspin protein expression in a large series of 223 surgically resected pancreatic ductal adenocarcinomas using immunohistochemical staining and high throughput tissue microarrays (TMAs). All these 223 patients underwent resection (pancreatoduodenectomy, distal pancreatectomy, or total pancreatectomy) at The Johns Hopkins Hospital between 1998 and 2003. We correlated maspin expression with postoperative survival as well as other clinicopathologic factors such as patient age, tumor size, tumor differentiation, pathologic stage, lymph node status, vascular invasion, and perineural invasion.

Materials and methods

Permission to perform this study was obtained from The Johns Hopkins Institutional Review Board.

Patient Selection

A total of 223 patients who had surgery (Whipple procedure, distal pancreatectomy, or total pancreatectomy) performed at The Johns Hopkins Hospital for primary invasive pancreatic ductal adenocarcinoma between 1998 and 2003 (five cases in 2003) were included in this study. Clinicopathologic factors were obtained from patients' electronic medical record. The follow-up information was obtained either from the electronic medical record and/or hospital tumor registry. Of the 223 patients, 116 were men and 107 were women. They had a mean age of 66 years at the time of surgery (range 32–89 years).

TMA Construction

TMAs were generated from archival resected cancer tissue. Each patient's tumor was represented by multiple3, 4 cores (each core is 1.5 mm in size) on each array to exclude effects of heterogeneous antigen expression. Normal pancreatic tissue and non-neoplastic pancreatic tissue from these patients were also included in these TMAs.

Immunohistochemistry

Unstained 4-μm sections were cut from each TMA and deparaffinized by routine techniques before placing in 200 ml Target Retrieval Solution, pH 6.0 (Dako, Envision Plus Detection Kit, Carpinteria, CA, USA) for 20 min at 100°C. After cooling for 20 min, slides were quenched with 3% H2O2 for 5 min before incubating with monoclonal antibody to maspin (dilution 1:50, BD Biosciences PharMingen, San Diego, CA, USA) for 60 min using the Dako Autostainer. Labeling was detected with the Dako Envision system as per the manufacturer's protocol. All sections were counterstained with Giles' hematoxylin. Cytoplasmic and/or nuclear maspin labeling was reported as negative (<5% cells expression), focally positive (5–50% cells), or diffusely positive (>50% cells). Only epithelial labeling was scored.

Statistical Analysis

Postoperative survival time was calculated as the time in months from the date of surgery to death (for non-censored events) or to the most recent contact/visit (for censored events) as of August 1 or September 1, 2006. Survival curves were analyzed by the Kaplan–Meier method, and the differences in survival curves were compared by the log-rank test using GraphPad Prism 4.0 software (GraphPad Software Inc., San Diego, CA, USA). Other statistical analyses were performed with SAS software version 9.1 (SAS Institute Inc., Carry, NC, USA). Range and frequency distributions of all continuous and categorized variables were examined. The t-test was used to compare the mean difference in age, tumor size, survival time, and positive nodes between groups. Fisher exact or χ2 tests were applied to compare gender, tumor differentiation, tumor stage, node status, vascular invasion, and perineural invasion between groups. Multivariate survival analysis was performed using the Cox proportional-hazard model. For all these analyses, a P-value of less than 0.05 was considered statistically significant.

Results

Patient's Follow-Up

A total of 223 patients with primary invasive pancreatic ductal adenocarcinoma who underwent resection (Whipple procedure, distal pancreatectomy, or total pancreatectomy) at The Johns Hopkins Hospital between 1998 and 2003 were included in this study. All patients had follow-up until their death or until their most recent contact or visit (1 August or 1 September 2006). At the time of the most recent contact, 39 of 223 patients were still alive, whereas 184 patients had succumbed to their disease.

Expression of Maspin was Associated with a Shorter Postoperative Survival

Of 223 pancreatic ductal adenocarcinomas, 209 (93.7%) showed expression of maspin including 39 focal (5–50% neoplastic cells) and 170 diffuse (50–100% neoplastic cells) expression. Expression was strong in the vast majority of the cancers and moderate in the remaining cases (Figure 1). Fourteen of 223 (6.3%) pancreatic ductal adenocarcinomas showed no expression of maspin. Normal ductal epithelium, acinar cells, and islet cells did not label for maspin (Figure 1). Analysis of the survival data revealed that patients whose carcinoma showed no maspin expression had an overall better survival than those with maspin expression (median survival: 29.5 months vs 17 months). The difference between the survival curves was almost reaching statistical significance (P=0.0533) (Figure 2). Maspin expression was not associated with patient age, gender, tumor size, tumor pathologic stage, lymph node status, vascular invasion, or perineural invasion (Table 1).

Figure 1
figure 1

Immunohistochemical staining of maspin in pancreatic ductal adenocarcinomas and PanINs. About 94% pancreatic ductal adenocarcinomas are immunohistochemically positive for maspin, the vast majority of them are diffusely positive (a) and a small percentage of them are focally positive (b). Normal ductal epithelium is negative for maspin (c). A small percentage of carcinomas also had nuclear staining of maspin in addition to cytoplasmic labeling (d). About 50% low-grade PanINs (PanIN1a e and PanIN1b f) and about 80% high-grade PanINs (PanIN2 g and PanIN3 h) are also immunohistochemically positive for maspin.

Figure 2
figure 2

Kaplan–Meier analysis of postoperative survival in patients whose tumors showed no maspin expression (n=14) vs those did (n=209). Patients whose tumors showed expression of maspin had shorter survival than those whose tumor did not (P=0.053). Median postoperative survival was 29.5 months in the former group of patients and 17 months in the latter group of patients.

Table 1 Comparison of clinicopathologic factors between patients whose tumors showed no expression and patients whose tumors showed expression of maspin

Patients whose tumors showed maspin expression by immunostaining were further divided into those with focal expression (5–50% cells, n=39) and those with diffuse expression (>50% cells, n=170). Patients whose carcinomas did not express maspin survived longer than patients whose carcinomas were diffusely positive for maspin (Figure 3, P=0.0454). Median survival time for these 14 patients whose tumor showed no maspin expression was 29.5 months in contrast to only 17 months for the latter group of 170 patients (P=0.045, Table 2). With a minimal follow-up of 40 months after surgery (the shortest follow-up time for those still alive in the latter group) (the longest follow-up for patients who were still alive as of August 1 or September 1 2006 is 102 months postoperatively), six of 14 patients (42%) whose tumor showed no maspin expression were still alive, whereas only 26 of 170 (15%) whose tumor showed diffuse expression were alive (P<0.05, Table 2). Between these two groups (no maspin expression vs diffuse maspin expression), there was no significance difference in patient age, gender, tumor size, tumor pathologic stage, lymph node status, vascular invasion, or perineural invasion (Table 2).

Figure 3
figure 3

Kaplan–Meier analysis of postoperative survival in patients whose tumors showed no expression (n=14) vs those showed diffuse expression (n=170) of maspin. Patients whose tumors showed diffuse expression of maspin had shorter survival and the difference is statistically significant (P=0.0454). Median postoperative survival was 29.5 months in the former group of patients and 17 months in the latter group of patients.

Table 2 Comparison of clinicopathologic factors between patients whose tumors exhibited no expression of maspin and patients whose tumors exhibited diffuse expression of maspin

The difference in overall survival between patients whose carcinomas demonstrated focal expression of maspin and those whose tumor demonstrated no maspin expression (Figure 4, P=0.1601) and that between those with focal expression and those with diffuse expression (Figure 5, P=0.5567) were not statistically significant.

Figure 4
figure 4

Kaplan–Meier analysis of postoperative survival in patients whose tumors showed no expression (n=14) vs those showed focal expression (n=39) maspin. Although patients whose tumors showed focal expression had shorter survival than those whose tumor did not, the difference is not statistically significant (P=0.1601). Median postoperative survival was 29.5 months in the former group of patients and 16 months in the latter group of patients.

Figure 5
figure 5

Kaplan–Meier analysis of postoperative survival in patients whose tumors showed focal expression (n=39) vs those showed diffuse expression of maspin (n=170). There is no difference in the survival between these two groups of patients (P=0.5567). Median postoperative survival was 16 months in the former group of patients and 17 months in the latter group of patients.

Positive Maspin Stain Remains an Independent Prognostic Marker after Adjusting Other Clinicopathologic Factors

We have shown that patients whose tumors showed maspin expression especially those with diffuse expression have worse postoperative survival than those whose tumor did not. In order to analyze whether maspin expression is an independent prognostic factor for postoperative survival, we performed multivariate survival analysis using the Cox-proportional hazards model. A stepwise selection process with five explanatory variables (number of positive nodes, tumor differentiation, maspin expression, patient age, and tumor size) was performed to conduct the multivariate survival analysis. As expected, our analysis showed that age, tumor size, tumor differentiation, and number of positive nodes were independent prognostic markers (hazard ratio=1.016, 1.115, 1.667, 1.068, respectively; P=0.0254, 0.03, 0.001, 0.001, respectively). After adjusting these prognostic markers, maspin expression is still an independent adverse prognostic marker for postoperative survival (hazard ration 2.351, P=0.0146, patients whose tumors showed diffuse expression; hazard ration 2.433, P=0.0106, patients whose tumors showed either focal or diffuse expression).

Nuclear Expression of Maspin is Associated with Better Tumor Differentiation

Several studies have reported that the intracellular location of maspin might also have prognostic significance.16, 19, 20, 21 Among the 209 cases showing maspin expression, 27 cases had both cytoplasmic and nuclear labeling (the percentage of cells with nuclear labeling ranged from 1 to 50% of the positive cells with a mean of 8%), and the remaining 182 cases had cytoplasmic expression only. Although patients whose tumors showed nuclear staining of maspin in addition to cytoplasmic staining have longer mean postoperative survival (24 months) than those with only cytoplasmic staining (16 months), the difference is not statistically significant (Figure 6, P=0.1149). The only clinicopathologic variable strongly associated with nuclear labeling of maspin is tumor differentiation. Twenty-one of the 27 (78%) carcinomas with nuclear expression in addition to cytoplasmic labeling were well to moderately differentiated. In contrast, only 92 of 182 carcinomas (50%) with cytoplasmic maspin labeling only were well to moderately differentiated (P=0.0072, Table 3). Intracellular location of maspin (with or without nuclear labeling) was not significantly associated with gender, mean age at the time of surgery, tumor size, pathologic stage pT, lymph node status, vascular invasion or perineural invasion (P>0.05 for all these parameters, Table 3).

Figure 6
figure 6

Kaplan–Meier analysis of postoperative survival in patients whose tumors showed nuclear in addition to cytoplasmic expression of maspin (n=27) vs those showed only cytoplasmic expression of maspin (n=182). Median postoperative survival was 24 months in the former group of patients and 16 months in the latter group of patients. The survival between these two groups of patients is not statistically significant (P=0.1149).

Table 3 Comparison of clinicopathologic factors between patients whose tumors demonstrated nuclear in addition to cytoplasmic expression of maspin and patients whose tumors demonstrated only cytoplasmic expression of maspin

Maspin Overexpression is Present in the Majority of Pancreatic Cancer Precursors-Pancreatic Intraepithelial Neoplasia

Invasive pancreatic ductal adenocarcinoma develops through a multi-step progression through precancerous lesions called pancreatic intraepithelial neoplasia (PanIN).22 Maspin expression in the multi-stage progression of pancreatic ductal adenocarcinoma has not been systemically examined. Therefore, in this study, we also studied 41 PanIN lesions (23 low-grade PanIN1, 18 high-grade PanIN2 and PanIN3). Among them, 25 (60%) expressed maspin, and 16 (40%) did not. By histologic grade, 11/23 (48%) low-grade PanIN (1a and 1b) (seven focal and four diffuse) and 14/18 (78%) high-grade PanIN (2 and 3) lesions expressed maspin (six focal and eight diffuse) (Figure 1).

Discussion

We investigated maspin expression in a large series of 223 pancreatic ductal adenocarcinomas. Our results show that maspin overexpression is observed in the vast majority of pancreatic ductal adenocarcinomas (209 of 223 tumors or 94%) but not in any normal pancreatic element. The percentage of pancreatic ductal adenocarcinomas with positive immunohistochemical staining for maspin in our study is similar to those reported in previous studies.3, 4, 23 In addition, in our study 14 pancreatic ductal adenocarcinomas did not show maspin expression and these patients demonstrated longer postoperative survival than those whose tumor did exhibit maspin expression (Figure 2). More specifically, those with no maspin expression had significant better survival than those with diffuse expression of maspin (Figure 3). Our multivariate survival analysis using the Cox proportional hazard model showed that maspin expression is an independent adverse prognostic marker for postoperative survival after adjusting other clinicopathologic factors. Expression of maspin was not associated with other clinicopathologic factors such as patient age, gender, mean tumor size, lymph node status, vascular invasion, or perineural invasion.

Maspin is a member of the serpin family of protease inhibitors.5 Maspin was originally thought to be a tumor suppressor based on its ability to inhibit invasion and motility of prostate and breast cancer cells in vitro5, 24 and its strong association with tumor progression and/or metastasis in some cancers such as breast cancer,25 prostate cancer,9 and oral squamous cell carcinoma26 and thyroid cancer,7 and colorectal cancer.8 In ovarian cancer, the data have been conflicting: two studies14, 15 showed that maspin overexpression is associated with shorter overall survival, whereas another study16 showed that only cytoplasmic labeling of maspin is associated with a worse prognosis, and high nuclear labeling was associated with better survival. In lung cancer, maspin expression has been associated with a better prognosis,13 worse prognosis,11 or no association at all.12 In this study, we showed that maspin overexpression was associated with decreased postoperative survival, especially in those patients whose carcinoma shows diffuse expression of maspin. Therefore, our finding and those of others7, 8, 9, 12, 13, 14, 15, 16, 25, 26 indicate that maspin is a prognostic factor but that its prognostic significance differs among different types of tumor.

There is some evidence that the subcellular location of maspin might be important for maspin function. For example, the presence of nuclear labeling has been associated with a good prognosis in breast cancer,19 ovarian cancer,16 and squamous cell carcinoma of the larynx.20 In contrast, in stage III colonic carcinomas, nuclear maspin expression was an adverse prognostic factor for overall survival.21 In this study, we did observe that patients whose pancreatic ductal adenocarcinomas showed nuclear expression of maspin in addition to cytoplasmic labeling of maspin had longer survival than those whose tumors demonstrated only cytoplasmic expression of maspin but the difference does not reach statistical significance. However, we did observe that pancreatic ductal adenocarcinomas with nuclear staining of maspin are more likely to be better differentiated (well to moderately differentiated) than those without (Table 3). Some have proposed that nuclear maspin might be the biologic active form and that cytoplasmic maspin may be inactive,11 but subcellular localization studies have indicated that maspin is predominantly a soluble cytoplasmic protein.20 More research is warranted to investigate the role of nuclear maspin in tumor pathogenesis.

It has been shown that human pancreatic ductal adenocarcinomas acquired maspin expression through hypomethylation of the maspin promoter.17, 18 Similar mechanism has been reported in cholangiocarcinoma,27 lung cancer,28 and melanoma.29 In contrast, in breast cancer,30 thyroid cancer31 and oral cancer,32 the maspin promoter is reportedly hypermethylated. The variable prognostic significance of maspin and the different molecular mechanisms regulating its expression in different types of tumors suggest that the role of maspin in tumorigenesis is likely to be tissue/organ-specific, and that the exact signaling pathway(s) regulated by maspin might depend on the genetic and epigenetic background of a specific neoplasm and its specific microenvironment.33, 34

It is known that pancreatic ductal adenocarcinoma develops through a multi-step progression model from low-grade PanIN to high-grade PanIN to invasive cancer.22 Only one previous study has addressed maspin expression in a small number of pre-invasive lesions. In that study,4 none of the eight PanIN-1 and two PanIN-2 lesions showed expression of maspin but all seven PanIN-3 lesions did express maspin. In this study, we investigated maspin expression in a larger series of 41 PanINs. We observed that maspin overexpression was not only in the vast majority of high-grade (78%) PanIN lesions (PanIN2 and PanIN3) but also in a significant number (48%) of low-grade PanIN lesions (PanIN1a and PanIN1b). Molecular studies have shown that maspin overexpression in pancreatic ductal adenocarcinoma is acquired through hypomethylation of the maspin promoter.17, 18 Our result suggests that hypomethylation of the maspin promoter probably occurs early in the multi-step progression model of pancreatic ductal adenocarcinomas.

In summary, in this study we investigated maspin overexpression in a large series of 223 patients with surgically resected pancreatic ductal adenocarcinomas. We found that overexpression of maspin in pancreatic ductal adenocarcinoma is associated with worse postoperative survival especially in patients whose tumors exhibited diffuse expression of maspin. After adjusting other clinicopathologic factors, maspin express is still an independent adverse prognosticator for postoperative survival. Nuclear labeling of maspin is associated with better tumor differentiation although this staining pattern is not associated with a better prognosis. Maspin overexpression is also observed in a significant percentage of precancerous lesions-PanINs, suggesting that maspin upregulation occurs early during the multi-step progression model of pancreatic ductal adenocarcinoma.