Loss of P16 Protein Defines High-Risk Patients with Gastrointestinal Stromal Tumors: a Tissue Microarray Study

638 Vol. 11, 638–645, January 15, 2005 Clinical Cancer Research

Loss of p16 Protein Defines High-Risk Patients with Gastrointestinal Stromal Tumors: A Tissue Microarray Study

Regine Schneider-Stock,1 Carsten Boltze,4 expected frequency [i.e., marker combinations: p16 positive, Jerzy Lasota,5 Brigitte Peters,2 Chris L. Corless,6 no metastases, and death of disease and p16 loss, metastases, and still alive]. The ‘‘type’’ whose observed frequency was Petra Ruemmele,8 Luigi Terracciano,10 3 11 significantly higher than the expected frequency consisted of Matthias Pross, Luigi Insabato, the following marker pattern: p16 loss, necrosis, and death of 11 9 Dolores Di Vizio, Igor Iesalnieks, disease (P < 0.001). In the multivariate Cox regression Stefan Dirnhofer,10 Arndt Hartmann,8 analysis, p16 loss, necrosis, and metastases each had Michel Heinrich,7 Markku Miettinen,5 independent prognostic value. P16 loss is a common molecular Albert Roessner,1 and Luigi Tornillo10 abnormality in GISTs and might be used in routine diagnosis to identify patients with high-risk tumors. Departments of 1Pathology, 2Biometrics, and 3General Surgery, Otto-von-Guericke University Magdeburg, Magdeburg, Germany; 4Department of Pathology, University of Rostock, Rostock, Germany; 5Department of Soft Tissue Pathology, Armed Forces Institute, INTRODUCTION Washington, District of Columbia; 6Department of Pathology and Gastrointestinal stromal tumors (GIST) are the most 7 Division of Hematology/Oncology, Oregon Health and Science common mesenchymal tumors of the gastrointestinal tract. University Cancer Institute and Portland Virginia Medical Center, Portland, Oregon; Departments of 8Pathology and 9General Surgery, They are characterized by the expression of the KIT (CD117, University of Regensburg, Regensburg, Germany; 10Institute of stem cell factor receptor) protein (1, 2). The mechanism Pathology, University of Basel, Basel, Switzerland; and 11Department responsible for constitutive activation of KIT protein in most of Biomorphological Sciences, Section of Pathology, University GISTs is a gain-of-function mutation in the KIT gene (3). ‘‘Federico II,’’ Naples, Italy Subsets of 5% to 7% of GISTs have activating mutations in the closely related PDGFRA gene (4). Interestingly, KIT and ABSTRACT PDGFRA mutation status in GISTs is predictive of clinical response to the tyrosine kinase inhibitor imatinib (5). Despite clearly defined histologic criteria, the prediction GISTs have a wide clinical spectrum that ranges from of tumor behavior for patients with gastrointestinal stromal benign to malignant behavior. Small tumors (V5 cm) that tumors (GIST) still poses a challenge to pathologists. show low mitotic frequency (V5 mitoses per 50 high-power Therefore, searching for alternative markers that allow for field) usually pursue a benign clinical course; however, a better prognostic evaluation is an important task. To small subset of mitotically inactive tumors do subsequently determine the practicability of immunohistochemical staining metastasize (1). Moreover, gastric tumors seem to behave less for p16 in clinical cases, we examined p16 protein expression aggressively than small intestinal tumors of similar size and in a group of 284 GISTs, a subset of which had long-term mitotic activity (6). Additional adverse prognostic factors, follow-up (median, 45 months; range, 1-204 months). P16 tested in large series of GISTs, include aneuploidy in DNA protein expression was ascertained on tissue microarrays as flow cytometry (7) as well as the presence of tumor necrosis well as on standard sections. Survival analyses were carried (8). No consensus has been reached regarding the prognostic out in 157 patients. P16 loss was found in 50% of GISTs, there value of a high Ki-67 score (1). being no correlation with age, sex, histologic subtype, signs of There are several indications that molecular alterations necrosis, or metastases. Patients having p16-negative tumors may predict the clinical outcome in GISTs, such as telomerase had a worse prognosis than those with p16-positive tumors activity (9, 10), DNA copy number changes (11, 12), KIT (P = 0.012) with a 2.3-fold relative increased risk of dying of mutation type (13, 14), hypermethylation of the E-cadherin disease. P16 loss identified a subgroup of gastric tumors with promoter (15), and expression of a set of six genes: CCNB1, a worse prognosis (P = 0.03). The multivariate configural CENP-F, FAK , HMG2, TSG101, and ezrin (16). Abnormal- frequency analysis identified two ‘‘antitypes,’’ whose observed ities in cell cycle regulators have been related to proliferation frequency was found to be significantly lower than the and progression of cancer cells. In a recent study, we reported on the prognostic significance of p16 gene alterations in 39 GISTs 17. The p16 tumor suppressor gene at 9p21 inhibits cell cycling by arresting cells in G1-S phase of the cell cycle Received 7/9/04; revised 9/2/04; accepted 9/17/04. (18). P16 operates through inhibition of cyclin-dependent The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked kinase 4, which in turn inhibits phosphorylation of pRb to advertisement in accordance with 18 U.S.C. Section 1734 solely to prevent cells from progressing into S phase (19). P16 loss has indicate this fact. been reported to predict poor clinical outcome in several Requests for reprints: Regine Schneider-Stock, Department of human tumors (20–24). Pathology, Otto-von-Guericke University Magdeburg, Leipziger Str. 44, 39120 Magdeburg, Germany. Phone: 49-391-6715060; Fax: 49-391- To confirm our previously reported data in a larger group 6715060; E-mail: [email protected]. of GISTs and to evaluate the reliability of immunohistochemical D2005 American Association for Cancer Research. p16 staining in the clinical setting, we determined p16 protein

expression in a group of 284 GISTs using tissue microarrays Table 2 Criteria for the definition of malignancy risk in GISTs (6) (TMA) and sections of conventionally cut tumor blocks. Risk scale Size (cm) Mitotic index Very low risk <2 <5 per 50 high-power field MATERIALS AND METHODS Low risk 2-5 <5 per 50 high-power field Clinical and Morphologic Features Intermediate risk <5 6-10 per 50 high-power field Patients. An overview of the clinical and morphologic data 5-10 <5 per 50 high-power field High risk >5 >5 per 50 high-power field is given in Table 1. We analyzed 284 primary GISTs from 284 >10 Any mitotic rate patients. The tumors were collected from the institutes of Any size >10 per 50 high-power field pathology at the University of Basel (Basel, Switzerland), University ‘‘Federico II’’ (Naples, Italy), Oregon Health and Science University (Portland, OR), University of Regensburg (Regensburg, Germany), and Armed Forces Institute of Pathology activity and tumor size (Table 2), GISTs were morphologically (Washington, DC). The study was approved by the investigator’s classified as very low risk, low-risk, intermediate-risk, or high- institutional review board. Our study included 160 men (56.3%) risk tumors. According to this histologic classification, we and 124 women (43.7%), their ages ranging from 9 to 91 years, investigated 28 very low risk, 55 low-risk, 43 intermediate-risk, with a mean (SE) of 60.9 (15.1) years. The primary tumors and 158 high-risk GISTs. (localization was known for 262 tumors) originated in the stomach Construction of TMA. For TMA construction, we used (n = 159), small intestine (n = 73), colon (n = 9), mesentery (n = a H&E-stained slide from each block to define representative 7), esophagus (n = 7), and rectum (n = 7). Tumors (data were tumor regions. Tissue cylinders with a diameter of 0.6 mm were available for 190 GIST) were histologically classified as punched from the tumor areas of each block and brought into predominantly spindled (n = 82), epithelioid (n = 49), or mixed a recipient paraffin block using a precision instrument (Beecher spindled-epithelioid (n = 59). Tumor size ranged from 0.4 to 45 cm Instruments, Sun Prairie, WI; ref. 25). Three different TMAs in greatest dimension, with a mean (SE) of 8.3 (6.6) cm. The were constructed, each containing tumors of the above- occurrence of necrosis was recorded in 100 cases: 38 cases (38%) mentioned institutes. To overcome the problem of tissue showed at least some tumor necrosis. Status of metastases was microheterogeneity and to increase the number of evaluable recorded for 195 patients, of whom 49 (25.1%) had metastases. cases, we replicated the TMAs by punching the donor tissue Morphologic Classification. The diagnosis of GISTs was blocks at least thrice. Thus, the analysis of p16 expression based on previously published criteria (6). Based on mitotic represents the mean of the staining of three different slides containing a single core per slide. Immunohistochemistry Table 1 Association between p16 loss and various clinicopathologic factors in GISTs KIT Protein. All GISTs were immunohistochemically positive for the KIT protein (antibody CD117, DAKO Corp., p16 loss, Carpinteria, CA). Factors n n (%) P P16 Protein. P16 immunohistochemistry was done as Morphologic classification described previously by Schneider-Stock et al. (17). Briefly, Very low risk 28 14 (50) 0.421 Low risk 55 22 (40) a monoclonal mouse antibody to p16 (1:100 dilution, Quartett, Intermediate risk 43 23 (53.5) Berlin, Germany) and antigen retrieval using microwave heating High risk 158 83 (52.5) [thrice for 10 minutes; 10 mmol/L citrate buffer (pH 6.0)] were Sex used after inhibition of endogeneous peroxidase activity. The Male 160 85 (53.1) 0.282 primary antibody was incubated for 1 hour at 37jC. The slides Female 124 57 (46) Age (mean, t test) were subsequently incubated with a 1:10 dilution of normal p16 loss 142 62.4 y 0.087 swine serum (Vector, Burlingame, CA). After washing in PBS Positive p16 protein 142 59.4 y (pH 7.4), the samples were incubated with a 1:200 dilution of expression biotinylated goat anti-mouse secondary antibody (Vector) for 30 Localization Stomach 159 89 (56) 0.062 minutes at room temperature. The detection of bound antibody Small intestine 73 25 (34.2) was accomplished using the avidin-biotin complex method Colon 9 4 (44.4) (Dianova UniTect ABC System XHC1, Berlin, Germany). A 5% Esophagus 7 6 (85.7) solution of New Fuchsine (red) was used as a chromogen. Mesenterium 7 2 (28.6) Specificity for immunostaining was checked by omitting single Rectum 7 5 (71.4) Tumor subtype steps in the immunohistochemical protocol and by replacing Spindle 82 38 (46.3) 0.167 primary antibody with nonimmune serum. Epithelioid 49 31 (63.3) Immunohistochemical staining for p16 was evaluated by Spindle/epithelioid 59 30 (50.8) estimating 10 high-power fields. Tumor nuclei with or without Necrosis Yes 38 16 (42.1) 0.309 cytoplasmic staining were counted according to a four-point No 62 33 (53.2) semiquantitative scale [no staining, 0-10% (0); weak, 11-20% (1); Metastases moderate, 21-50% (2); strong, >50% (3)]. A cutoff at 20% Yes 49 24 (49) 0.511 positivity was used for prognostic analysis. Nontumorous stromal No 146 80 (54.8) cells showing nuclear reactivity served as an internal control.

Surgery and Follow-up method and compared using the log-rank test. Deaths from For follow-up analysis, we used the time of the last clinical unrelated causes were censored. Multivariate analysis was based appointment or the date of death. The median follow-up time on Cox regression analysis. P < 0.05 was considered statistically was 45 months (range, 1-204 months). Survival analysis was significant. Calculations were carried out by SPSS version 9.0 carried out in 157 patients. All tumors were treated solely by software package (SPSS, Inc., Chicago, IL). surgical resection. All cases were R0 resected (local or multivisceral) with histologically confirmed margins. RESULTS Statistical Analysis p16 Immunostaining Frequency of p16 Loss All statistical analyses were carried out using the v2 test or Half of our GIST tumors showed p16 loss (142 cases), Fisher’s exact test in cross tables and one-way ANOVA (for whereas the other 50% of GISTs expressed p16 protein comparison of means) to assess the relationship between p16INK4 immunohistochemically (Fig. 1). Considering normal tissue loss and clinicopathologic factors. All statistical tests were two specimens, we found p16 protein expression exclusively in the sided. We used a log-linear model and the related v2 test for Cajal cells, whereas the other parts of the submucosa, including analysis in multivariate contingency tables as well as the myenteric plexuses, were immunonegative. configuration frequency analysis for the identification of types Validation of TMA Immunostaining. To validate TMA (observed frequency is higher than the expected frequency) and data, we compared the immunohistochemical staining results of antitypes (observed frequency is lower than expected; ref. 26). p16 protein on whole tissue sections of 40 cases with the results Total survival curves were drawn according to the Kaplan-Meier achieved on core biopsies represented on the TMA (Fig. 1) and

Fig. 1 Representative p16 immunohistochemistry shows reproducibility and quality of staining using TMA technique on GISTs: 1, original tissue sections after manual tissue punching with homogeneous staining or loss of p16 (A, high-risk GIST; B, high-risk GIST; C, low-risk GIST; magnification, Â6.4); 2, window of 12 punched and stained tissue samples (magnification, Â12.5); 3 and 4, higher magnification (Â25 and Â100) shows positive staining of p16 protein in the cytoplasm and nuclei of GISTs (B and C) and complete loss of p16 protein expression (A).

Table 3 Prognostic significance of p16 loss, histologically determined 3.1-fold increased risk for each upward step in the risk category malignancy risk, necrosis, metastases, and age in GISTs using Cox (P < 0.001; Table 3). regression model (univariate analysis) It is noteworthy that the six very low risk cases and three low- Overall survival risk cases with loss of p16 behaved aggressively. Considering Variable n Relative risk P both groups together (very low risk and low risk), the difference between p16-negative and p16-positive GISTs reached borderline Histologically defined risk 157 3.107 <0.001 Necrosis 71 5.900 0.008 significance in the Kaplan-Meier analysis (P = 0.059). Metastases 157 2.090 0.013 Prognostic Value of Metastases. The occurrence of p16 loss 157 2.310 0.016 metastases was significantly associated with a worse prognosis Age 157 1.027 0.026 in the group of 157 GISTs for which full clinical follow-up data Subtype 157 1.594 0.126 were available (P = 0.011, log-rank; Fig. 4A). For patients Localization 157 1.101 0.219 Sex 157 0.818 0.498 developing metastases, univariate Cox analysis revealed a 2-fold increased risk of dying from disease; P = 0.013; Table 3). As metastases constituted a predictor for worse prognosis, we found discordance in only one case. Here, the whole tissue assessed the value of p16 loss in the prognostically favorable section was considered p16 negative, whereas the core biopsy group of GISTs without metastases separately and found that p16 was taken from a small area (<15%) with moderately positive loss defined a subset of GISTs with a worse prognosis (Fig. 4B). p16 staining. Thus, there was only one false-positive case among The multivariate contingency table revealed that loss of 40 microarrayed tumors (2.5%). These results supported the use p16, metastases, and death of disease were highly associated of TMAs, and we therefore pooled the data obtained from the variables (P = 0.005; Table 4). The two antitypes reflected the TMAs and, where available, full tissue sections. underrepresented combinations for the three variables. For patients affected by tumors with p16 loss and metastases, there Correlation between p16 Protein Loss and Clinicopath- b ologic Data. Gastric tumors showed p16 loss at higher fre- is a high probability that they will die of disease (antitype ). Conversely, patients with p16-positive GISTs and obviously no quency than did intestinal GISTs (56% for GISTs in the stomach c versus 34.2% for GIST in the small intestine), although the dif- signs of metastases have a more favorable outcome (antitype ). ference failed to reach statistical significance (P = 0.28). GISTs Prognostic Value of Necrosis. Necrosis clearly indicated in the esophagus were almost exclusively p16-negative tumors malignant potential in our group of GISTs (P = 0.002, log- (6 of 7 cases). There did not exist any relationship between loss of rank; Fig. 5). For patients showing necrosis, univariate Cox p16 protein and sex, age, histologic subtype, morphologic classi- analysis revealed a 5.9-fold increased risk of dying (P = 0.008; fication, signs of necrosis, or occurrence of metastases (Table 1). Table 3). Analyzing the group of GISTs without necrosis separately, we found no additional value for loss of p16 that Survival Analyses allowed us to select a subgroup with a worse clinical outcome Prognostic Value of p16 Loss. Patients who had tumors (P = 0.163). Nevertheless, the multivariate contingency table with p16 loss had a worse prognosis than those having tumors without p16 loss. After 5 years, patients affected by tumors with p16 loss and those without had overall survivals of 55.7% and 76.8%, respectively. There was a high predictive value for p16 loss on the clinical outcome of these patients (P = 0.012, log-rank; Fig. 2). Thirty-five of 88 (39.8%) patients with p16-negative tumors but only 14 of 69 (20.3%) patients with p16-positive tumors died of disease. Thus, p16INK4 loss seems to be useful for defining a subgroup with unfavorable prognosis in GISTs. Univariate Cox regression analysis revealed a correlation between p16 loss and overall survival (Table 3). For patients affected by tumors with loss of p16, there was a 2.3-fold increased risk of dying of disease. Prognostic Value of the Histologically Determined Risk of Malignancy. The concordance between the histologically determined risk of malignancy and clinical behavior was evaluated in 178 cases, for which data on the clinical course were available. Only 6 of 18 (33.4%) tumors of the very low risk group and 6 of 27 (22.2%) tumors of the low-risk group developed metastases and/or died of disease. There was a significant correlation between the morphologically determined risk and the clinical behavior of the tumors (P < 0.001). Furthermore, the Kaplan-Meier curve reflected the prognostic Fig. 2 Kaplan-Meier plot for survival on 157 GIST patients as a value of the classic morphologic diagnosis (P < 0.001, log- function of p16 status. Patients having tumors with p16 loss (n = 88) rank; Fig. 3), also demonstrating the accuracy of sampling of have a significantly poorer prognosis than patients having tumors with our patient group. Univariate Cox regression analysis revealed a positive p16 immunostaining (n = 69).

Prognostic Value of Localization, Histologic Subtype, Age, and Sex. Using the Kaplan Meier curves, we did not detect any significant relationship between tumor localization, histologic subtype, sex, and age and overall survival. In univariate Cox analysis, only age revealed a statistical significance (Table 3) but without any statement about the risk for both sexes. As we found that gastric GISTs tended to show p16 loss more frequently than GISTs of the small intestine, we separately analyzed the prognostic value of p16 loss in both groups. Whereas loss of p16 seemed to identify a subgroup of gastric tumors with a worse prognosis (P = 0.037; Fig. 6), it had no additional value for predicting the prognosis in the GISTs of small intestinal origin (P = 0.74).

Multivariate Cox Regression Analysis When those factors showing a statistically significant (P < 0.05) relationship to outcome in univariate analysis were included in multivariate Cox regression analysis, only loss of p16, necrosis, and metastases retained their strong independent prognostic value (Table 6).

Fig. 3 Kaplan-Meier plot for survival on the 284 GIST patients as a DISCUSSION function of morphologically determined risk of malignancy. Whereas the There is an increasing necessity to better characterize intermediate-risk group (2) tended to show a poorer prognosis than the pooled very low risk and low-risk groups (1), the high-risk group molecular changes in GISTs. Consequently, one might obtain differed significantly from both (1 and 2). predictive information allowing for better patient care. In this context, we studied the clinical relevance of p16 inactivation in GISTs. Our results suggest that loss of p16 protein expression at showed a significant association between p16 loss and necrosis the immunohistochemical level can be considered an indepen- in relation to death of disease. We identified an overrepresented dent, reliable prognostic variable in patients with GIST. Besides, ‘‘type’’ consisting of the combination of the following p16 loss, necrosis, and metastases were confirmed to be variables: p16 loss, necrosis, and death of disease (Table 5). independent variables predicting outcome in GIST patients.

Fig. 4 Kaplan-Meier plot for survival on the 157 GIST patients as a function of metastasis status (A) and on the 119 GIST patients without metastases as a function of p16 status (B). Patients developing metastases (n = 38) have a significantly poorer prognosis than those affected by tumors without metastases. In the group of metastasis-free patients, p16 loss (n = 49) displayed the worst clinical outcome.

Table 4 Multivariate configuration analysis, including the three subgroups of GISTs as far as their localization is concerned. variables: p16 expression, metastases, and death of disease In a recent study, Lasota et al. (13) identified one type of KIT p16 Observed Expected mutation (internal tandem duplication in the 3V KIT-juxta- Expression Metastasis n n membrane domain) to be associated with gastric GISTs with a Loss No Alive 46 45.88 more favorable outcome. In contrast, the KIT 1530ins6 Loss No DOD 24 20.82 mutation defines GISTs of intestinal origin with a more Loss Yes Alive 7 14.65 Antitype aggressive potential (14). Interestingly, in a recent study by (p16 loss, metastases, Lasota et al. of GISTs lacking KIT gene mutations, it was and alive) found that nearly one third of the gastric tumors had an exon Loss Yes DOD 11 6.65 18 mutation in the PDGFRA gene, whereas only 2.7% of Positive No Alive 42 35.98 intestinal GISTs had such mutations. The majority of the Positive No DOD 7 16.32 Antitype PDGFRA mutant GISTs pursued a benign course (29). It was (positive p16 reported that KIT and PDGFRA mutant GISTs show an expression, activation of the mitogen-activated protein kinase pathway (4), no which in turn seems to be a possible stimulus of p16 metastases, regulation (30). If p16 is an effector downstream of KIT and and died of disease) PDGFRA, p16 loss might be dependent on a specific type of Positive Yes Alive 13 11.49 KIT mutation, but this has to be proven in future studies. Positive Yes DOD 7 5.21 P16 is involved in cell cycle regulation through the Total n 157 157 P = 0.005 retinoblastoma pathway and acts at the G1-S interface (19). Abbreviation: DOD, died of disease. Loss of p16 protein causes loss of the capacity to inactivate the cyclin-dependent kinase 4. This eliminates the ability of the cell to maintain an active retinoblastoma protein that ultimately leads Separation of malignant from benign GISTs seems to to cell cycle progression. Inactivation of the p16 function has be practically impossible at present (6). In a recent study of been reported as one of the most frequent events in human another, much smaller group of GISTs (17), p16 alterations cancers (31), its frequency ranging from f20% in sporadic defined a subgroup with unfavorable prognosis in the benign melanoma, non–small cell lung cancer, head and neck cancer, and borderline groups with GISTs. In that study, in contrast to and esophageal cancer to 30% in transitional cell cancer of the our TMA-based study, we subdivided GISTs into only three bladder, 35% in gliomas, and 50% in pancreatic cancer and different groups (benign, borderline, and malignant), because squamous cell carcinoma of the bladder (32–35). We found p16 the number of GIST cases available was too small to obtain inactivation in 50% of GISTs. This p16 protein loss can be a group number sufficient for statistical evaluation. With the greater number of GISTs available for this study, we were able to use the criteria of the GISTs consensus meeting (6) to divide the tumors into four different risk groups. An additional advantage of this study is that many cases with long-term follow-up were included, because recurrences and metastases in GISTs may occur years after primary surgery. Evaluating the prognostic impact of p16 loss in the groups with very low risk and low-risk GISTs, we found that all very low risk GISTs and three of six low-risk GISTs that followed an unfavorable clinical course were p16-negative tumors. The significance in the Kaplan-Meier analysis reached borderline level. Thus, in both studies, p16 loss was superior to the classic morphologic classification variables. There was an excellent concordance of p16 evaluation between conventional sections and TMAs. This indicates that the TMA technology can be reliably applied to high-throughput analyses of immunohistochemical p16 protein expression. It is noteworthy that p16 loss occurred more frequently in gastric GISTs in comparison with GISTs of the small intestine, although this difference reached only borderline significance. P16 loss seems to define a subgroup of gastric GISTs with a poor clinical outcome, whereas there was no prognostic impact of p16 loss on the GISTs originating from the small intestine. Gastric GISTs comprise nearly two thirds of all GISTs and have been reported to generally show a better survival than do Fig. 5 Kaplan-Meier plot for survival on the 71 GIST patients as a intestinal GISTs of similar size and mitotic activity (27, 28). function of necrosis status. Patients having GISTs with necrosis (n = 28) Almost nothing is known about prognostically different are characterized by a more aggressive clinical course.

Table 5 Multivariate configuration analysis, including the three Table 6 Prognostic significance of p16 loss, necrosis, and metastases in variables: p16 expression, necrosis, and death of disease GISTs using Cox regression model (multivariate analysis with reduction using all eight variables from the univariate p16 Observed Expected Cox regression analysis) Expression Necrosis n n Overall survival (n = 71) Loss No Alive 24 20.13 Loss No DOD 3 4.09 Variable Relative risk P Loss Yes Alive 7 13.11 Metastases 4.969 0.019 Loss Yes DOD 6 2.67 Type p16 loss 4.992 0.022 (p16 loss, Necrosis 4.422 0.038 necrosis, and died of disease) Our results indicate that p16 alterations constitute a major Positive No Alive 16 15.60 Positive No DOD 0 3.17 molecular abnormality in GISTs with a considerable prognostic Positive Yes Alive 12 10.16 impact. We consider the immunohistochemical detection of p16 Positive Yes DOD 3 2.07 protein expression in GISTs as a reliable marker for daily routine Total n 71 71 P = 0.017 diagnosis and evaluation of the malignancy risk, particularly in Abbreviation: DOD, died of disease. small and mitotically inactive tumors.

REFERENCES caused by diverse mechanisms as shown recently by Schneider- 1. Miettinen M, El-Rifai W, Sobin L H, Lasota J. Evaluation of ma- Stock et al. (17). In that study, diverse genetic p16 alterations lignancy and prognosis of gastrointestinal stromal tumors: a review. Hum correlated significantly with loss of p16 protein expression. The Pathol 2002;33:478–83. inactivation mechanisms included loss of heterozygosity at 9p21 2. Rubin BP, Singer S, Tsao C, et al. KIT activation is ubiquitous feature gene mutations, eliminating the interaction of the p16 protein of gastrointestinal stromal tumors. Cancer Res 2001;81:8118–21. with cyclin-dependent kinase 4 or causing the generation of 3. Heinrich MC, Rubin BP, Longley BJ, Fletcher JA. Biology and a truncated protein, homozygous deletions, and methylation of genetic aspects of gastrointestinal stromal tumors: kit activation and cytogenetic alterations. Hum Pathol 2002;33:484–95. the p16 gene promoter region as also shown in other studies 4. Heinrich MC, Corless CL, Duensing A, et al. PDGFRA activating (36–38). There is accumulating evidence that expression of p16 mutations in gastrointestinal stromal tumors. Science 2003;299:708–10. and retinoblastoma is mutually counterbalanced to maintain 5. Heinrich MC, Corless CL, Demetri GD, et al. Kinase mutations and growth inhibitory activity. Furthermore, p16 loss seems to link imatinib response in patients with metastatic gastrointestinal stromal cancer and aging by interacting with p53 inactivation and tumors. J Clin Oncol 2003;21:4342–9. telomerase activation (30). 6. Fletcher CDM, Berman JJ, Corless C, et al. Diagnosis of gastrointestinal stromal tumors: a consensus approach. Hum Pathol 2002;33:459–65. 7. Rudolph P, Chiaravalli AM, Pauser U, et al. Gastrointestinal mesenchymal tumors—immunophenotypic classification and survival analysis. Virchows Arch 2002;441:238–48. 8. Lerma E, Oliva E, Tugues D, Prat J. Stromal tumours of the gastrointestinal tract: a clinicopathological and ploidy analysis of 33 cases. Virchows Arch 1994;424:19–24. 9. Sakurai S, Fukayama M, Kaizaki Y, et al. Telomerase activity in gastrointestinal stromal tumors. Cancer 1998;83:2060–6. 10. Gu¨nther T, Schneider-Stock R, Ha¨ckel C, et al. Telomerase activity and expression of hTRT and hTR in gastrointestinal stromal tumors in comparison with extragastrointestinal sarcomas. Clin Cancer Res 2000;6:1811–8. 11. El-Rifai W, Sarlomo-Rikala M, Andersson LC, Knuutila S, Miettinen M. DNA sequence copy number changes in gastrointestinal stromal tumors: tumor progression and prognostic significance. Cancer Res 2000;60:3899–903. 12. Gunawan B, Bergmann F, Hoer J, et al. Biological and clinical significance of cytogenetic abnormalities in low-risk and high-risk gastrointestinal stromal tumors. Hum Pathol 2003;33:316–21. 13. Lasota J, Dansonka-Mieszkowska A, Stachura T, et al. Gastrointes- tinal stromal tumors (GIST) with internal tandem duplications in 3Vend of kit juxtamembrane domain occur predominantly in stomach and generally have a favorable course. Mod Pathol 2003;16:1257–64. 14. Lasota J, Kopczynski J, Sarlomo-Rikala M, et al. KIT 1530ins6 mutation defines subset of predominantly malignant gastrointestinal stromal tumors (GIST) of intestinal origin. Hum Pathol 2003;34: 1306–12. Fig. 6 Kaplan-Meier plot for survival on the 93 gastric GIST patients 15. House MG, Guo MZ, Efron DT, et al. Tumor suppressor gene as a function of p16 status. P16 loss defines a subgroup of GISTs in the hypermethylation as a predictor of gastric stromal tumor behavior. stomach (n = 37) characterized by worse prognosis. J Gastrointest Surg 2003;7:1004–14.

16. Koon N, Schneider-Stock R, Sarlomo-Rikala M, et al. Molecular 27. Ueyama T, Guo KJ, Hashimoto H, Daimaru Y, Enjoji MA. targets for tumour progression in gastrointestinal stromal tumors. Gut Clinicopathologic and immunohistochemical study of gastrointestinal 2004;53:235–40. stromal tumors. Cancer 1992;69:947–55. 17. Schneider-Stock R, Boltze C, Lasota J, et al. High prognostic value 28. Emory TS, Sobin LH, Lukes I, Lee DH, O’Leary TJ. Prognosis of of p16INK4 alterations in gastrointestinal stromal tumors. J Clin Oncol gastrointestinal smooth-muscle (stromal) tumors: dependence on ana- 2003;21:1688–97. tomic site. Am J Surg Pathol 1999;23:82–7. 18. Okamoto A, Demetrick DJ, Spillare EA, et al. Mutations and altered 29. Lasota J, Dansonka-Mieszkowska A, Sobin LH, Miettinen M. A expression of p16INK4 in human cancer. Proc Natl Acad Sci U S A great majority of GISTs with PDGFRA mutations represents gastric 1994;91:11045–9. tumors of low or no malignant potential. Lab Invest 2004;84:874–83. 19. Serrano M. The tumor suppressor protein p16INK4a. Exp Cell Res 30. Sharpless NE, De Pinho RA. Telomeres, stem cells, senescence, and 1997;237:7–13. cancer. J Clin Invest 2004;113:160–8. 20. Kawaguchi K, Oda Y, Saito T, et al. Mechanisms of inactivation of 31. Serrano M, Lee H, Chin L, Cordon-Cardo C, Beach D, DePinho RA. the p16INK4a gene in leiomyosarcoma of soft tissue: decreased p16 Role of the INK4a locus in tumour suppression and cell mortality. Cell expression correlates with promotor methylation and poor prognosis. 1996;85:27–37. J Pathol 2003;201:487–95. 32. Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K, Carson DA. 21. Gessner C, Liebers U, Kuhn H, et al. INK4A are independent Deletions of the cyclin-dependent kinase-4 inhibitor gene in multiple positive prognostic markers for advanced tumour stage of nonsmall cell human cancers. Nature 1994;368:753–6. lung cancer. Eur Respir J 2002;19:134–40. 33. Reed AL, Califano J, Cairns P, et al. High frequency of p16 22. Hilton DA, Penney M, Evans B, Sanders H, Love S. Evaluation of (CDKN2/MTS-1/INK4A) inactivation in head and neck squamous cell molecular markers in low-grade diffuse astrocytomas: loss of p16 and carcinoma. Cancer Res 1996;56:3630–3. retinoblastoma protein expression is associated with short survival. Am 34. Caldas C, Hahn SA, da Costa LT, et al. Frequent somatic mutations J Surg Pathol 2002;26:472–8. and homozygous deletions of the p16 (MTS1) gene in pancreatic 23. Ma¨kitie AA, MacMillan C, Ho J, et al. Loss of p16 expression has adenocarcinoma. Nat Genet 1994;8:27–32. prognostic significance in human nasophrayngeal carcinoma. Clin 35. Schutte M, Hruban RH, Geradts J, et al. Abrogation of the Rb/p16 Cancer Res 2003;9:2177–84. tumor-suppressive pathway in virtually all pancreatic carcinomas. Cancer 24. Esteller M, Gonzalez S, Risques RA, et al. K-ras and p16 Res 1997;57:3126–30. aberrations confer poor prognosis in human colorectal cancer. J Clin 36. Liggett WH, Sidransky D. Role of the p16 tumor suppressor gene in Oncol 2001;19:299–304. cancer. J Clin Oncol 1998;16:1197–206. 25. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays 37. Rocco JW, Sidransky D. p16 (MTS-1/CDKN2/INK4a) in cancer for high-throughput molecular profiling of hundreds of specimens. Nat progression. Exp Cell Res 2001;264:42–55. Med 1998;4:844–7. 38.MerloA,HermanJG,MaoL,etal.5VCpG island 26. Lautsch E, von Weber S. Methoden und Anwendungen der methylation is associated with transcriptional silencing of the Konfigurationsfrequenzanalyse. Weinheim: Psychologie Verlags Union; tumor suppressor p16(CDKN2/MTS1 in human cancer. Nat Med 1995. 1995;1:685–92.

Downloaded from clincancerres.aacrjournals.org on September 29, 2021. © 2005 American Association for Cancer Research. Loss of p16 Protein Defines High-Risk Patients with Gastrointestinal Stromal Tumors: A Tissue Microarray Study

Regine Schneider-Stock, Carsten Boltze, Jerzy Lasota, et al.

Clin Cancer Res 2005;11:638-645.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/11/2/638

Cited articles This article cites 37 articles, 13 of which you can access for free at: http://clincancerres.aacrjournals.org/content/11/2/638.full#ref-list-1

Citing articles This article has been cited by 11 HighWire-hosted articles. Access the articles at: http://clincancerres.aacrjournals.org/content/11/2/638.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/11/2/638. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.