From January 2005 through December 2009, 134 patients were initially diagnosed as having mesenchymal gastrointestinal tumors at Jilin University First Hospital. Thirty-three patients were excluded from the study due to recurred tumors or the tumors being partially resected. Thus, 101 patients underwent successful surgical operations for complete resection of tumors. Following the surgery, patients with high-risk and intermediate risk were treated with imatinib (Glivec^®, Novartis Pharmaceuticals, Basel, Switzerland) at a dose of 400 mg/d for 3 years. No imatinib treatment was given before the surgery. Five cases were lost to follow-up. Ninety-six patients were retrospectively evaluated in the study. Informed consents were obtained from all patients and the study was approved by the local human ethical committee of Jilin University First Hospital. Original hematoxylin and eosin-stained sections were reviewed in each case by two pathologists (Jin MS and Wang YP) according to GIST characteristics described by Miettinen[14]. All tumors from 96 patients were confirmed to be GISTs based on a combination of histological evaluations (highly cellular spindles/epithelioids/mixed cell tumors), and c-kit, DOG1, CD34 positive staining. The clinical information regarding the patients is summarized in Table 1.

Histological sections (4 μm) of 10% formalin-fixed, paraffin-embedded material were used for immunohistochemical staining. Prior to a primary antibody staining, the slide was pretreated with citric acid or ethylenediaminetetraacetic acid buffer in a pressure cooker for antigen retrieval. Endogenous peroxidase activity was quenched by 3% H₂O₂ blocking reagent for 10 min. The slide was incubated with a primary antibody at 4 °C overnight, and then immunostained with the avidin-biotin peroxidase complex (DAKO, CA). Finally, the slide was stained with diaminobenzidine according to the manufacturer’s protocol (DAKO, CA). The slide was rinsed three times with phosphate buffered saline after each step of staining. The sections were stained with primary antibodies against c-kit (Clone: YR145, dilution: 1/50, Cell Marque Corporation, CA), CD34 (QBEnd/10, dilution: 1/100, Neomarkers, CA), Ki-67 (MIB-1, dilution: 1/100, DAKO, Carpinteria, CA), DOG1 (SP31, dilution: 1/100, Spring Bioscience, Pleasanton), SMA (IA4, dilution: 1/200, Cell Marque Corporation, CA), p27 (1B4, dilution: 1/20, Novocastra), p53 (SP5, dilution: 1/100, Zymed Laboratories, San Francisco), S-100 (6E6, dilution: 1/100, Neomarkers, CA), Desmin (D33, dilution:1/50, Cell Marque Corporation, CA), EGFR (EGFR.113, dilution: 1/200, Novocastra Laboratories Ltd, Newcastle, United Kingdom), and COX-2 (SP21, dilution: 1/50, Neomarkers, CA, United States), respectively. All primary antibodies used in the study were biotinylated monoclonal antibodies. The stained slides were evaluated quantitatively or semi-quantitatively by two independent pathologists who were blinded from clinical data. Percentages of positive cells stained with a special antibody observed by two pathologists were consistent and the mean values were determined.

The nuclear staining for Ki-67 and p53, and cytoplasmic immunostaining for EGFR and COX-2, were considered as positive cells of the reaction. According to previous studies[15,16], the following scoring assessments for Ki-67 and p53 were used. The score 0 was assigned for < 5%, 1 for > 5% and < 10%, 2 for > 10% of Ki-67 staining positive cells. The p53 scoring system was 0 assigned for < 5%, 1 for > 5% and < 25%, 2 for > 25% of p53 staining positive cells. EGFR scoring system was 0 assigned for < 10%, 1 for > 10% and < 60%, 2 for > 60% of EGFR positive cells based on the systems described by Nakagawa et al[17] and Gumurdulu et al[18]. The COX-2 scoring system was 0 assigned for no positive cells; 1 for < 25% and score 2 for > 25% of COX-2 staining positive cells according to Fux et al[19].

The Chi-square test and Fisher’s exact test were used to analyze relationships between clinicopathological features and expression levels of biomarkers. Kaplan-Meier test was applied with a log-rank test to study associations between categorical variables and the mean values of survival among groups. Cox proportional hazards regression analysis was used to estimate a hazard risk for survival and 95% CI was applied. The SPSS program (version 18.0) was used for statistical analysis. A P value of < 0.05 was considered statistically significant.

Clinicopathological features of the patients are summarized in Table 1. The median age of 96 patients was 55 years (range, 26-82 years). Histomorphology showed that the neoplastic cells were predominantly spindle-shaped (83/96, 86.5%). Based on the modified NIH risk consensus system, 45 (46.9%), 24 (25.0%), 24 (25.0%) and 3 (3.1%) cases were classified as high-risk, intermediate-risk, low risk and very low risk categories, respectively. Fifty-three cases (55.2%) had mild nuclear atypia; 32 cases (33.4%) showed severe nuclear atypia, but 11 patients (11.4%) had no nuclear atypia. Tumor necrosis was found in 39 cases of the patients (40.6%).

At the time of study, the mean or the median duration of the follow-up period was 31 mo or 29 mo, respectively. Medical charts were available for 96 of 101 patients (95%). Sixty-nine patients (54.2%) received the imatinib treatment at a dose of 400 mg/d for 13 mo to 36 mo (median, 26 mo). Thirty-seven patients (82%) from the high risk group and 15 patients (62.5%) from the intermediate group required the imatinib treatment. Disease specific 1, 2, 3 and 4 year survival probabilities were 0.97, 0.89, 0.79, and 0.77 (0.65-0.87), respectively. Of the 96 cases, 19 patients (19.8%) died from GISTs and 6 patients (6.3%) died from unrelated causes.

Eighty-eight (91.3%) tumor specimens were stained positive for c-kit. The tumors isolated from 8 patients (8.7%) were negative for c-kit but positive for DOG1 and/or CD34 staining. Reactivity with Desmin was found in 3 (3.1%) cases. Positive SMA and S-100 staining were also noted in 46 (47.9%) and 12 (12.5%) cases, respectively. Based on the Ki-67 index, 53.1% of tumors (n = 51) scored 0; 34.4% (n = 33) scored 1; and 12.5% (n = 12) scored 2. 34.4% of tumor specimens (n = 33) were p53 staining positive in the nuclei of over 25% of the cells. EGFR staining was found in most cases. Forty-two (43.8%) cases scored 2, and 27 (28.1%) cases scored 1 for EGFR staining. COX-2 overexpressed in 36 (37.5%) cases (Table 2 and Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Images of gastrointestinal stromal tumor using hematoxylin-eosin staining and immunohistochemical staining. A: Hematoxylin-eosin stain; B: DOG1 stain; C: Ki-67 stain; D: P53 stain; E: Epidermal growth factor receptor stain; F: Cyclooxygenase-2 stain.

Clinicopathological features and GIST grades categorized by the staining of Ki-67, p53, EGFR or COX-2 are established in Table 3. The expression of Ki-67 was significantly associated with tumor size (P = 0.02), mitotic rate (P < 0.001) and the risk of malignancy (P < 0.001). The p53 expression was also correlated with mitotic rate (P = 0.04), tumor site (P = 0.02) and the risk of malignancy (P = 0.04). The levels of COX-2 protein were significantly higher in gastric tumors and spindle cell-like tumors (P < 0.001 and P = 0.05, respectively). In contrast, no correlation was found between the EGFR expression and clinicopathological factors or the risk of malignancy.

The 3-year survival rates for disease specific survival (DSS) were 100%, 89%, 79% and 67% for groups at very low-risk, low-risk, intermediate-risk and high risk by the modified NIH risk categories, respectively. Associations between DSS and different protein biomarkers were analyzed using a multivariate analysis (Table 3 and Figure 2). The survival rates were strongly associated with tumor size (P = 0.004), mitotic count (P = 0.001), tumor location (P = 0.018), the NIH modified risk criteria (P < 0.001, Figure 2A), Ki-67 amplification (P < 0.001, Figure 2B), and adjuvant imatinib therapy (median survival period 52 mo vs 37 mo, χ² = 7.618, P = 0.006, Figure 2C). No significance was found when comparing survival rates with the EGFR expression, or the COX-2 expression. In the high-risk group, Ki-67 overexpression was significantly associated with poor survival (χ² = 10.44, P = 0.006), but no statistical significance was found between the p53 expression and survival (χ² = 4.744, P = 0.089). Using a multivariate analysis, a poor survival was observed in the high risk category [relative risk (RR) = 12.23; 95% CI: 1.61-92.81] graded using the modified NIH risk consensus system or in category 2 scored by the Ki-67 expression (RR = 15.78; 95% CI: 4.25-59.37) (Table 3).

Open in New Tab   Full Size Figure   Download Figure

Figure 2 Kaplan-Meier plots for predicting disease specific survival based on the modified National Institutes of Health consensus system (A), Ki-67 expression (B) and adjuvant imatinib therapy (C).

Gastrointestinal stromal tumors (GISTs) are known to have a wide variability in malignancy and prognosis. Risk grading based on tumor size, location and mitotic counts has been proposed to predict adverse outcomes of GIST in the literature.

Recent molecular studies have found that the deregulations of Ki-67, cyclin A, B1, D1, E, cdc2 and other cell-cycle regulators were significantly associated with a shorter disease-free survival in GISTs.

In this study, expressions of Ki-67, p53, epidermal growth factor receptor (EGFR) and COX-2 were investigated in a large cohort of GIST patients and their roles as prognostic values for GISTs were also evaluated. To our knowledge, this is the first assessment of the prognostic value of EGFR in patients with GISTs.

The immunohistochemical staining of these tumorigenetic and cell proliferative proteins provides an alternative approach for follow-up and clinical decisions regarding the treatment for GISTs.

This paper describes a retrospective clinicopathological study of GIST.