p27 in Nonfunctioning Pancreatic Endocrine Tumors

229

Clinical Research

Prognostic Significance of p27, Ki-67, and Topoisomerase IIα Expression in Clinically Nonfunctioning Pancreatic Endocrine Tumors Hee Jin Chang, MD,1 Kenneth P. Batts, MD,2 Ricardo V. Lloyd, MD,2 Thomas J. Sebo, MD,2 Geoffrey B. Thompson, MD,3 Christine M. Lohse, BS,4 and Shane V. Pankratz, PHD4 Abstract Nonfunctioning islet cell tumors or pancreatic endocrine tumors are the most common type of malignant islet cell tumor. Although previously detected usually at an advanced stage because of mass effect, the early detection rate of small localized disease has been increasing. To date it has been difficult to predict the clinical behavior in localized regional nonfunctioning tumors. To investigate potential markers predicting malignancy and poor prognosis in nonfunctioning pancreatic endocrine tumors, we analyzed the expression of Ki-67, topoisomerase IIα (TopoIIα), and p27, as well as a variety of clinicopathologic parameters in 76 cases of nonfunctioning islet cell tumors (23 benign cases and 53 malignant cases). Ki-67, TopoIIα, and p27 labeling indices were significantly different between benign and malignant tumors. Expression of Ki-67, TopoIIα, and p27 were associated with survival in patients with a malignant tumor in a univariate setting. However, only p27 and TopoIIα were jointly associated with survival in multivariate analysis. Immunohistochemical staining for p27, TopoIIα, and Ki-67 can be helpful in the diagnosis of nonfunctioning pancreatic endocrine tumor. Analysis of p27 and TopoIIα may also have potential utility as prognostic factors for malignant tumors. Key Words: Nonfunctioning pancreatic endocrine tumor; prognosis; p27; Ki-67; topoisomerase IIα. 1

Department of Pathology, National Medical Center, Seoul, Korea; Department of 2 Laboratory Medicine and Pathology, 3Surgery and 4 Biostatistics, Mayo Clinic, Rochester, MN Address correspondence and reprint requests to: Dr. Hee Jin Chang Department of Pathology National Medical Center 18-79, Ulchiro 6-Ka, Joong-ku Seoul 100-196 Korea E-mail: [email protected] Endocrine Pathology, vol. 11, no. 3, 229–241, Fall 2000 © Copyright 2000 by Humana Press Inc. All rights of any nature whatsoever reserved. 1046–3976/00/11:229–241/$13.25

Introduction It is often difficult to predict clinical outcome by histologic analysis in pancreatic endocrine tumors [1,2]. The most reliable criteria for malignancy are gross infiltration of adjacent organs, metastasis, and/or blood vessel invasion [1–4]. Many patients with pancreatic endocrine tumors have clinical evidence of hormonal hypersecretion but 15–53% of pancreatic tumors are nonfunctioning or clinically silent [1,5,6]. They are usually diagnosed at an advanced stage,

because of the lack of clinical symptoms related to hormonal excess [6,7]. The majority of nonfunctioning tumors, 64– 92%, are malignant [1,4,5] and their clinical manifestations are similar to pancreatic ductal adenocarcinoma [4–6]. With the advances of diagnostic imaging techniques and the widespread use of abdominal computed tomography and ultrasonography, the incidental detection rate of a nonfunctioning tumor in its early stage has increased [8,9]. However, it is still 229

230

Endocrine Pathology

Volume 11, Number 3

difficult to predict preoperatively the malignant potential of localized nonfunctioning tumor in its early stages because the most reliable diagnostic criteria of malignancy—local invasion and metastasis—usually develop late. Generally, a localized pancreatic endocrine tumor of less than 2 cm diameter is regarded as a benign tumor [10] that can be cured by resection or enucleation. However, size alone cannot be an absolute criterion for diagnosing malignancy. Although microvascular or perineural microinvasion may be valuable markers for diagnosis of malignancy [2], these features are often difficult to detect radiographically and are also difficult to demonstrate in a small biopsy specimen. Recently, cell-cycle-related proteins including p27 and Ki-67 have become useful adjuncts in the diagnosis of malignant endocrine tumors [11–13]. p27 is a cyclindependent kinase inhibitor and has an inhibitory effect at the G1-to-S phase transition in the cell cycle through its negative effect on cyclin E-CDK2, cyclin D-CDK4/6, and other cyclins [14]. Decreased p27 expression has been shown to correlate with tumor aggressiveness and decreased survival in various epithelial malignancies [14]. Ki-67 labeling index (LI) is considered to be a useful prognostic marker for pancreatic endocrine tumors [15], but there are some differences in the interpretation of what constitutes a high Ki-67 LI [16, 17]. One of the difficulties in interpreting these findings is that the studies included both functioning and nonfunctioning pancreatic endocrine tumors [16,17]. Topoisomerase IIα (TopoIIα) is another cell-cycle-related protein that is expressed in the S phase and G2/M phase [18]. It also has prognostic relevance not only as a marker of cell proliferation but also as a potential predictor of response to chemo-

Fall 2000

therapy involving topoisomerase-inhibiting drugs [18]. Given the significance of combination chemotherapy as first-line treatment in advanced malignant pancreatic endocrine tumors [19], an analysis of TopoIIα labeling index in nonfunctioning pancreatic endocrine tumor may provide important information not only on the proliferation status of tumor cells, but also on the potential efficacy of chemotherapeutic agents. The aim of this study was to evaluate the expression of p27, Ki-67, and TopoIIα in benign and malignant nonfunctioning endocrine tumors of pancreas with additional examination of a variety of clinicopathologic parameters in an attempt to determine their prognostic significance.

Materials and Methods Clinical Cases

Patients with a diagnosis of pancreatic endocrine tumor who were treated at the Mayo Clinic between 1979 to 1994 were reviewed. Among a total of 203 patients, 117 cases of functioning tumor with a clinical syndrome associated with hormonal hypersecretion and 10 cases of nonfunctioning tumors, without available tissues, were excluded. Nonfunctioning pancreatic endocrine tumors were defined as pancreatic neoplasms that have endocrine-type histology and express endocrine markers, but lack any association with a clinical syndrome caused by hormonal hypersecretion [2]. The study group consisted of 76 patients with clinically nonfunctioning tumors, including 61 cases from resection and 15 cases from needle biopsies. Malignancy was based on the presence of invasion into adjacent tissue or organs, the presence of regional or distant metastasis, or the presence of angioinvasion

p27 in Nonfunctioning Pancreatic Endocrine Tumors

[3,4]. Among the 76 cases, 53 were malignant and 23 were benign. Four patients had the multiple endocrine neoplasia (MEN) type I syndrome; one of these tumors was benign and three were malignant. Clinical data including age, sex, tumor site and size, evidence of local invasion, metastasis or recurrence, and type of treatment were obtained by chart review. Follow-up information was available in all patients, with an average duration of follow-up of 73 mo. Histologic Review

All H&E stained slides of formalinfixed, paraffin-embedded tissue sections were reviewed for histologic pattern, nuclear atypia, mitosis, capsular penetration, and angio- or perineural invasion. The histologic pattern was recorded according to Soga’s classification: type A (solid nest), type B (trabecular), type C (tubuloacinar), and type D (solid–undifferentiated) [20]. The nuclear atypia was graded as grade 0 if the nuclei were small and uniform similar to normal islet cells, grade 1 if they showed minimal to mild variation of their size with fine chromatin and inconspicuous nucleoli, grade 2 if they were moderately irregular in size and shape with an increase of nuclear/cytoplasmic ratio, coarsely granular chromatin pattern, and small to medium sized nucleoli, and grade 3 if they were markedly hyperchromatic, irregular in size and shape [2]. Mitotic index was assessed by counting the numbers of mitoses per 10 high-power fields (HPF) (×400) in the most mitotically active areas. Immunohistochemistry

Representative sections for immunostaining were cut at 5 µm. Antibodies and dilutions used included anti-p27 (Trans-

231

duction Laboratory, Lexington, KY, 1:1000), MIB-1 (Ki-67) (AMAC, Westbrook, ME, 1:50), and antitopoisomerase IIα (DAKO, Carpinteria, CA, 1:200). For diagnostic confirmation of pancreatic endocrine tumor, a panel of immunostaining was performed and the antibodies were as follows: antichromogranin (Roche, Indianapolis, IN, 1:1000), antiinsulin (Novocastra, New Castle, UK, 1:100), antiglucagon (DAKO, 1:1500), antigastrin (DAKO, 1:1000), and antisomatostatin (DAKO, 1:750). Sections were treated with 0.1 mol/L citrate, pH 6.0, in an 800-W microwave oven for 15 min for antigen retrieval before immunostaining of p27, Ki-67, and TopoIIα. Immunostaining was done with the Elite avidin–biotin–peroxidase kit (Vector Laboratories, Burlingame, CA) according to the manufacturer’s specifications. Slides were counterstained with hematoxylin for 1 s. Positive controls for immunostaining consisted of tonsil tissues for p27, Ki-67, and TopoIIα. Negative controls consisted of substituting normal mouse serum for the primary antibodies. Immunoreactivity of each antibody was analyzed by quantifying nuclear staining without knowledge of the diagnosis or outcome. The percentages of positive cells were calculated after counting of at least 1000 tumor cells. Labeling indices (LI) of Ki-67 and TopoIIα were enumerated by two observers in two different ways: (a) microscopic counting in areas of maximum staining intensity (visual counting) and (b) random counting using the CAS 200 digital image analyzer (DIA) and proliferation index software program (Becton Dickinson, Cellular Imaging Systems, San Jose, CA). p27 immunoreactive cells were randomly counted with the microscope for each slide. A minimum of 10 HPF were counted. When 10% of the cases were

232

Endocrine Pathology

Volume 11, Number 3

blindly recounted, interobserver and intraobserver differences were within 5% of the original count. Statistical Analysis

Comparisons of the clinical and pathologic findings including the Ki-67, p27, and TopoIIα LI between benign and malignant tumors were conducted using Wilcoxon rank sum, chi-square, and Fisher’s exact tests. Correlations among the labeling indices were assessed using Spearman’s rank correlation coefficients. Survival rates of patients with malignant tumors were calculated using Kaplan– Meier methods. Univariate associations between survival and the clinicopathological characteristics were evaluated using Cox proportional hazards modeling. Multivariate Cox modeling using a backward selection procedure was used to develop a joint model to predict survival of patients with malignant tumors. Cutoff values for the labeling indices that best distinguished between patients with good or poor survival were determined using recursive partitioning techniques. In all tests, a p value < 0.05 was considered statistically significant. Results Clinicopathological Findings

Clinicopathological features are summarized in Table 1. In comparison to malignant tumors, benign tumors were more common in women (78% vs 43%, p = 0.005) and smaller (mean 4.5 cm vs 7.6 cm, p = 0.013). However, four malignant tumors were less than 2 cm in diameter, and one benign tumor was 10 cm in diameter. Tumors in the head of the pancreas were more likely to be benign, whereas tumors in the tail were more likely to be malignant (p = 0.011). Patients with

Fall 2000

malignant tumors were less likely to have curative operation (p < 0.001). All patients with benign tumors except one patient who died within 1 mo after surgery from postoperative complications were alive with an average follow-up of 90 mo. Thirty-six of the 53 patients with malignant tumors were deceased at followup. Among deceased patients, two had died perioperatively because of sepsis and the remaining patients died of disease. Average follow-up duration for malignant cases was 65 mo. Survival rates at 5 and 10 yr were 48.8% and 30.1%, respectively. Among the patients with malignant tumors, 44 cases had evidence of metastasis to liver (35), lymph nodes (25), omentum (1), stomach (1), and bone (1), at the time of diagnosis or postoperatively. Angioinvasion, perineural invasion, and capsular penetration were identified in 30, 11, and 45 cases, respectively. Malignant tumors showed a significantly higher mitotic rate compared to benign tumors (mean 6.1/10 HPF vs 0.2/10 HPF, p < 0.001). There was some evidence that malignant tumors were more likely to have more nuclear atypia compared to benign tumors (p = 0.047). However, analysis of the histologic patterns did not show a statistically significant difference between these two groups. All tumors were positive for chromogranin A. Immunohistochemical staining for insulin, glucagon, gastrin, and somatostatin was performed in 30 cases and in 13 of these (43%) there was focal positivity for one or more of the hormone markers. p27 Expression

Immunohistochemical staining for p27 in normal islet cells showed diffuse strong nuclear staining with variable weak cytoplasmic positivity (Fig. 1). Most tumor

233

p27 in Nonfunctioning Pancreatic Endocrine Tumors

Table 1. Comparison of Clinicopathological Findings in Benign and Malignant Nonfunctioning Islet Cell Tumors

Gender (Females)a Mean ISD Age (yr) Siteb Head Neck Body Tail Mean ± SD Sizeb (cm) Operationa Curative Palliative Non a

Follow-up Duration (months) Histologic Pattern

A B C D

Mitotic Figurea,b Nuclear Atypiaa

0 1 2 3

Ki-67 LIa(%) Visual analysis Image analysis Topoisomerase IIα LIa(%) Visual analysis Image analysis P27 LIa,b(%)b

Benign (n = 23)

Malignant (n = 53)

Total (n = 76)

18 (78%) 58 ± 14.7

23 (43%) 52 ± 16.2

41 (54%) 54 ± 15.9

8 (35%) 6 (26%) 3 (13%) 6 (26%) 4.5 ± 3.2

19 (36%) 1 (2%) 8 (15%) 25 (47%) 7.6 ± 5.9

27 (36%) 7 (9%) 11 (14%) 31 (41%) 6.6 ± 5.3

22 (96%) 0 (0%)

13 (25%) 26 (49%)

35 (46%) 26 (34%)

1 (4%)

14 (26%)

15 (20%)

90 ± 47

65 ± 61

73 ± 58

8 (35%) 10 (44%) 4 (17%) 1 (4%)

24 (45%) 17 (32%) 3 (6%) 9 (17%)

32 (42%) 27 (36%) 7 (9%) 10 (13%)

0.2 ± 0.4

6.1 ± 10.5

4.3 ± 9.2

2 (9%) 12 (52%) 8 (35%) 1 (4%)

2 (4%) 13 (24%) 29 (55%) 9 (17%)

4 (5%) 25 (32%) 37 (49%) 10 (13%)

1.5 ± 1.2 0.4 ± 0.4

10.0 ± 12.0 2.8 ± 4.5

7.5 ± 10.7 2.1 ± 3.9

1.2 ± 0.5 0.4 ± 0.4

9.2 ± 9.6 4.1 ± 5.7

6.8 ± 8.8 3.0 ± 5.1

69.3 ± 16.8

44.4 ± 21.4

51.9 ± 23.0

a

- p < 0.05. Mean ± SD.

b

cells revealed nuclear staining only, but focal cytoplasmic staining was present in 29 cases. However, there was no significant difference in cytoplasmic expression rate between benign and malignant tumors (10 of 23 cases, 43% vs 19 of 53 cases, 36%) and no correlation with the p27 LI. The p27 LIs were significantly lower in malignant tumors than in benign tumors (mean ± SD 44.4% ± 21.4 vs 69.3% ±

16.8, p < 0.0001) (Figs. 2 and 3), but there was considerable overlap in the p27 indices between the two groups (Fig. 4). Ki-67 and Topoisomerase IIa Expression

The immunohistochemical staining of Ki-67 and TopoIIα were significantly different between benign and malignant tumors (Figs. 2 and 3). The LIs of Ki-67

234

Endocrine Pathology

Volume 11, Number 3

Fall 2000

Fig. 1. Immunohistochemical staining for p27 in normal pancreatic tissue: Islet cells show diffuse nuclear staining with weak cytoplasmic positivity, whereas acinar and ductal cells show occasional weak staining.

p27 in Nonfunctioning Pancreatic Endocrine Tumors

235

Fig. 3. Immunohistochemical staining for p27 (A), Ki-67 (B) and topoisomerase IIα (C) in malignant nonfunctioning pancreatic endocrine tumor: p27 was not expressed in tumor cells but endothelial cells were positive, whereas Ki-67 and topoisomerase IIα were frequently positive for tumor cells.

and TopoIIα tended to be higher by visual analysis compared to image analysis. However, there was strong correlation between the visual and DIA assessments of both Ki-67 (r = 0.73, p < 0.001) and TopoIIα LI (r = 0.83, p < 0.001). On visual counting, the mean ± SD percentage of Ki-67 LI was 10.0% ± 12.0 for malignant tumors and 1.5% ± 1.2 for benign tumors

(p < 0.001) mean ± SD. TopoIIα LI was 9.2% ± 9.6 and 1.2% ± 0.5 for malignant and benign tumors, respectively (p < 0.001). On image analysis, mean ± SD Ki-67 LI was 2.8% ± 4.5 and 0.4% ± 0.4 for malignant tumors (p = 0.001) mean ± SD TopoIIα LI was 4.1% ± 5.7 and 0.4% ± 0.4, for malignant and benign tumors (p < 0.001).

Fig. 2. (Opposite page) Immunohistochemical staining for p27 (A), Ki-67 (B) and topoisomerase IIα (C) in benign nonfunctioning pancreatic endocrine tumor: p27 was diffusely expressed in the tumor cells but Ki-67 or topoisomerase IIα was expressed in only in a few cells.

236

Endocrine Pathology

Volume 11, Number 3

Fall 2000

Fig. 4. Labeling indices of Ki-67, topoisomerase IIα and p27 in benign (open circle and square) and malignant (shaded circle and square) nonfunctioning pancreatic endocrine tumor.

Correlation among p27 Expression with Ki-67 and Topoisomerase IIα Expression

The Ki-67 LIs were highly correlated with TopoIIα index on both visual (r = 0.92, p < 0.001) and DIA assessments (r = 0.74, p < 0.001). The p27 LIs were negatively correlated with visual Ki-67 (r = –0.43, p < 0.001), visual TopoIIα (r = –0.44, p < 0.001), DIA Ki-67 (r = –0.31, p = 0.007), and DIA TopoIIα indexes (r = –0.34, p = 0.002). Relationship between Survival and Clinicopathologic Variables for Malignant Tumors

Hazard ratios and 95% confidence intervals (CI) for univariate and multivariate associations between survival in patients with malignant tumors and the clinicopathologic characteristics are summarized in Table 2. In univariate analysis, male sex, type of operation, mitosis, and LI of Ki-67, TopoIIα, and p27 LI were significantly associated with survival. In multivariate analysis, p27 LI and DIA index of TopoIIα were jointly significantly associated with survival (Table 2).

Cutoff values of the LI identifying patients with poor survival were p27 ≥ 47%, visual Ki-67 index ≥ 9%, DIA Ki-67 index ≥ 3.95%, visual TopoIIα index ≥ 9.3 %, and DIA TopoIIα index ≥ 9% (Table 3). Figure 5 illustrates the survival curves for patients with p27 ≤ 47% vs those with p27 > 47%. Discussion In this study we examined the diagnostic utility and prognostic significance of p27, Ki-67, and TopoIIα in clinically nonfunctioning pancreatic endocrine tumors. Although the tumors were clinically nonfunctioning, focal immunoreactivity for various pancreatic hormones was found in 43% of the cases examined, confirming that focal expression of specific hormones does not necessarily correlate with the clinical function of the tumors. Among those markers, p27 expression proved to be the most reliable prognostic indicator. As in carcinomas of colon [21], ovary [22], urinary bladder [23], and exocrine pancreas [24], loss of p27 expres-

237

p27 in Nonfunctioning Pancreatic Endocrine Tumors

Table 2. Clinicopathologic Characteristics Associated with Survival in Malignant Nonfunctioning Islet Cell Tumors in Univariate and Multivariate Analysis Univariate analysis Hazard ratio (95% CI) p value Gender Female1.00 Male Agea Operation Curative Palliative None Mitotic Figurea Ki-67 LIa Visual Analysis Image Analysis

Multivariate analysis Hazard ratio (95% CI) p value

NS 2.23 NS

(1.09–4.58)

0.029 1.16

1.04 –1.29)

0.006

(1.48–27.96) (1.78–38.35)

0.013 0.007

(1.05–1.83) (0.81–0.96)

0.019 0.003

1.00 7.02 10.68 1.16

(1.63–30.18) (2.38–48.00) (1.03–1.30)

0.009 0.002 0.016

1.00 6.43 8.26 NS

1.20 1.42

(1.06–1.36) (1.04–1.92)

0.005 0.026

NS NS

1.17 1.44 0.89

(1.02–1.36) (1.08–1.91) (0.82–0.96)

0.030 0.013 0.002

NS 1.39 0.88

Topoisomerase IIα LIa Visual Analysis Image Analysis P27 LI* a

Risk ratio (95% CI) represents a five unit increase in the characteristics listed. p ≥ 0.05.

Table 3. Univarate Associations with Survival in Patients with Malignant Nonfunctioning Islet Cell Tumors for Cutoff Values of Labeling Indices

Ki-67 LI Visual analysis (≥ 9.0%) Image analysis (≥ 3.95%) Topoisomerase IIα LI Visual analysis (≥ 9.3%) Image analysis (≥ 9.0%) p27 LI (≤ 47.0%)

sion was significantly correlated with poor prognosis in nonfunctioning pancreatic endocrine tumors. Patients with tumors having a p27 LI lower than 47% showed significantly poorer survival (Fig. 5). The loss of p27 did not seem to be as striking in nonfunctioning pancreatic endocrine tumor, compared to some other tumors. In our study, 22 of 53 malignant cases showed more than 50% of p27 LI. This is in contrast to the data for exocrine pan-

Hazard ratio (95% CI)

p value

2.82 (1.42-5.61) 4.70 (2.12-10.42)

0.003 < 0.001

3.26 (1.62-6.57) 3.11 (1.38-6.99) 3.02 (1.48-6.16)

0.001 0.006 0.002

creatic carcinomas. Lu et al. [24] reported that 31 of 38 pancreatic carcinomas showed 0–10% of cells expressing p27. The higher p27 LI in pancreatic endocrine tumors would be consistent with the less aggressive behavior of this tumor. Unlike pancreatic ductal carcinoma, which is often fatal within 1 to 2 yr of the diagnosis, nonfunctioning pancreatic endocrine tumors grow slowly and metastatic disease does not exclude the possibility of extended

238

Endocrine Pathology

Volume 11, Number 3

Fall 2000

Fig. 5. Kaplan–Meier survival curve illustrates significantly different survival rates for malignant tumors with p27 ≤ 47% versus p27 > 47%.

survival [5]. Tissue specificity of p27 protein may also be related to higher expression in pancreatic endocrine tumor. In the normal pancreas, p27 expression is higher in islet cells than in acinar or ductal cells [24]. Given the diversity of p27 expression in normal pancreatic tissue, the difference of p27 LI between pancreatic endocrine and exocrine tumor may not be so significant. p27 is an essential component of the pathway that connects mitogenic signals to the cell cycle [14]. p27 expression has been reported to be inversely correlated to the proliferative activity in some endocrine tumors such as those in the parathyroid [13]. In our study, a weakly negative relationship between p27 and Ki-67 and between p27 and TopoIIα was observed. Some cases had a low proliferative activity, in spite of low p27 LI or vice versa. A possible explanation for the former could be involvement of other CDK inhibitors, such as p15, p16, p18, p19, p21, or p57 in regulating the cell cycle activity [25,26]. Rb and p53 tumor suppressor genes and c-myc oncogene could overcome p27-induced growth arrest [14,28]. In pancreatic endocrine tumors, alteration of Rb or p53 has been reported to be rare [29,30], whereas

overexpression of c-myc oncogene has been observed [30,31]. Thus, this oncogene may allow proliferation of tumor cells despite high p27 level in the latter. Recent studies have shown cytoplasmic accumulation of p27 in tumor cells [32,33]. Although the biologic significance of this observation is still unknown, it suggests that various factors including ubiquination and proteasome degradation, overexpression of cyclin D3, loss of tumor suppressor genes, or binding to a transcriptional activator may be related to changes in the level of nuclear p27 expression [33]. Although only nuclear localization of p27 was analyzed, we observed cytoplasmic staining of p27 in 29 cases of nonfunctioning pancreatic endocrine tumors (10 benign and 19 malignant tumors), but the frequency or intensity of cytoplasmic labeling was not related to malignancy or to nuclear p27 labeling. Many authors have emphasized the significance of Ki-67 LI in the diagnosis of pancreatic endocrine tumor as a useful marker for malignancy and poor prognosis [2,3,15–17,34]. However, the published cutoff values for malignancy in pancreatic endocrine tumors has varied from 2% to

p27 in Nonfunctioning Pancreatic Endocrine Tumors

10% [2,16,17,34]. Pelosi et al. [16] and Clarke et al. [34] showed that the Ki-67 LI of nonfunctioning tumor was 1.5- to 4-fold higher than that of functioning tumors. In this study, we excluded functioning tumors and analyzed the LI in two different ways. The indices were significantly higher in malignant tumors and were univariately associated with poor prognosis. The result from image analysis was much lower than that of visual counting. This may be related to the differences in the area counted on the slides and the method of counting. Visual counting was done in the areas with the highest number of labeled cells, whereas image analysis was done randomly. Although both methods were positively correlated with each other, the variability of indices by counting method may be a limiting factor for the use of these markers in diagnostic pathology unless a rapid reliable semiquantitative method can be used. TopoIIα is a newly recognized proliferation marker [18,35,37] and has been used as an indicator of malignant potential in uterine cervical squamous lesions [18] and adrenal cortical neoplasms [35]. DNA topoisomerase plays an important catalytic role in regulating the topologic status of DNA [38]. It causes a transient break of a double-strand of DNA through which another DNA molecule is passed [18]. In mammalian cells, TopoII protein exists in two different isoforms, a and b [35,38]. The a isoform predominates in proliferating cells, whereas the b isoform predominates in resting cells [35,38].Our study indicates that TopoIIa is also helpful for the diagnosis of malignant nonfunctioning pancreatic endocrine tumor and for predicting the prognosis of these tumors. TopoIIα can be substituted for Ki-67, but it has some of the same drawbacks, as

239

shown in this study, it can also differ according to counting methods. However, the use of TopoIIα LI as a proliferating marker may have several advantages over Ki-67. TopoIIα can give a better estimate of the number of actively cycling cells than Ki-67 because TopoIIα is present during the late S and G1 phases of the cycle, whereas Ki-67 is present in all phases of the cell cycle except G0 [36]. TopoIIα is also known to be a molecular target for a number of clinically useful antitum or drugs, such as etoposide, doxorubicin, daunorubicin, amsacrine, teniposide, mitoxantrone, and actinomycin D [18,37, 38]. Although the mainstay of treatment for pancreatic endocrine tumor is surgical resection, combination chemotherapy such as streptozotocin plus doxorubicin or 5-fluorouracil is still used for the treatment of advanced malignant pancreatic endocrine tumors [19]. Therefore, pancreatic endocrine tumors with a high TopoIIα LI may show better response to doxorubicin. Several studies have shown that the expression of TopoIIα correlates with the response to TopoIIα-inhibiting drugs [39,40]. Capella et al. proposed several criteria for determining malignancy in nonfunctioning pancreatic endocrine tumor, in which tumors ≥3 cm and/or with angioinvasion would be regarded as malignant [10]. La Rosa et al. also suggested that a mitotic index ≥ 2, size ≥ 4 cm, angio- or perineural invasion, capsular penetration, nuclear atypia, lack of progesterone receptors, presence of calcitonin, and Ki-67 LI > 2% were correlated with malignancy [2]. We examined the prognostic significance of histological variables with tumor size. Among these variables, only mitosis was univariately correlated with prognosis, but low mitotic index does not exclude malignancy because 14 of 53

240

Endocrine Pathology

Volume 11, Number 3

malignant cases (26%) did not show increased mitotic figures. This result confirms that routine histologic parameters for malignancy are not as helpful in the diagnosis of nonfunctioning pancreatic endocrine tumor. In summary, immunohistochemical staining of p27, Ki-67, and TopoIIα can be helpful in the diagnosis of malignancy in nonfunctioning pancreatic endocrine tumor and the LI of p27 and TopoIIα may have potential as new prognostic factors for these tumors. References 1. Klöppel G, Heitz PU. Pancreatic endocrine tumors. Path Res Pract 183:155–68, 1988. 2. La Rosa S, Sessa F, Capella C, Riva C, Leone BE, Klersy C, Rindi G, Solcia E. Prognostic criteria in nonfunctioning pancreatic endocrine tumors. Virchows Arch 429:323–333, 1996. 3. Capella C, La Rosa S, Solcia E. Criteria for malignancy in pancreatic endocrine tumors. Endocr Pathol 8:87–90, 1997. 4. Eckhauser FE, Cheung PS, Vinik AI, Strodel WE, Lloyd RV, Thompson NW. Nonfunctioning malignant neuroendocrine tumors of the pancreas. Surgery 100:978–988, 1986. 5. Kent III RB, van Heerden JA, Weiland LH. Nonfunctioning islet cell tumors. Ann Surg 193:185–190, 1981. 6. Lo CH, van Heerden JA, Thompson GB, Grant CS, Soreide JA, Harmsen WS. Islet cell carcinoma of the pancreas. World J Surg 20:878–884, 1996. 7. Liu TT, Zhu Y, Cui QC, Cai LX, Ye SF, Zhong SX, Jia HP. Nonfunctioning pancreatic endocrine tumors. An immunohistochemical and electron microscopic analysis of 26 cases. Path Res Pract 188:191–198, 1992. 8. Furukawa H, Mukai K, Kosuge T, Kanai Y, Shimada K, Yamamoto J, Mizuguchi Y, Ushio K. Nonfunctioning islet cell tumors of the pancreas: clinical, imaging and pathological aspects of 16 patients. Jpn J Clin Oncol 28:255–261, 1998.

Fall 2000

9. Thompson GB, van Heerden JA, Grant CS, Carney A, Ilstrup DM. Islet cell carcinomas of the pancreas: a twenty-year experience. Surgery 104:1011–1017, 1988. 10. Capella C, Heitz PU, Höfler H, Solcia E, Klöppel G. Revised classification of neuroendocrine tumors of the lung, pancreas and gut. Virchows Arch 425:547–560, 1995. 11. Erickson LA, Jin L, Wollan PC, Thompson GB, van Heerden JA, Lloyd RV. Expression of p27kip1 and Ki-67 in benign and malignant thyroid tumors. Mod Pathol 11:169–174, 1998. 12. Erickson LA, Jin L, Wollan PC, Thompson GB, van Heerden JA, Lloyd RV. Parathyroid hyperplasia, adenomas, and carcinomas. Differential expression of p27kip1 protein. Am J Surg Pathol 23:288–295, 1999. 13. Lloyd RV, Jin L, Qian X, Kulig E. Aberrant p27 kip1 expression in endocrine and other tumors. Am J Pathol 150:401-407, 1997. 14. Lloyd RV, Erickson LA, Jin L, Kulig E, Qian X, Cheville JC, Scheithauer BW. p27kip1: a multifunctional cyclin-dependent kinase inhibitor with prognostic significance in human cancers. Am J Pathol 154:313–323, 1999. 15. Lloyd RV. Utility of Ki-67 as a prognostic marker in pancreatic endocrine neoplasms. Am J Clin Pathol 109:245–247, 1998. 16. Pelosi G, Bresaola E, Bogina G, Pasini F, Rodella S, Castelli P, Calogero I, Serio G, Zamboni G. Endocrine tumors of the pancreas: Ki-67 immunoreactivity on paraffin sections is an independent predictor for malignancy: a comparative study with proliferating-cell nuclear antigen and progesterone receptor protein immunostaining, mitotic index, and other clinicopathologic variables. Hum Pathol 27:1124–1134, 1996. 17. Perret AG, Mosnier JF, Buono JP, Berthelot P, Chipponi J, Balique JG, et al. The relationship between MIB-1 proliferation index and outcome in pancreatic neuroendocrine tumors. Am J Clin Pathol 109:286–293, 1998. 18. Gibbons D, Fogt F, Kasznica J, Holden J, Nikulasson S. Comparison of Topoisomerase IIa and MIB-1 expression in uterine cervical squamous lesions. Mod Pathol 10:409–413, 1997. 19. Oberg K. Neuroendocrine gastrointestinal tumors–a condensed overview of diagnosis and treatment. Ann Oncol 10:S3–8, 1999.

p27 in Nonfunctioning Pancreatic Endocrine Tumors

20. Soga J, Tazawa K. Pathologic analysis of carcinoids. Histologic revaluation of 62 cases. Cancer 28:990–998, 1971. 21. Loda M, Cukor B, Tam SW, Lavin P, Fiorentino M, Draetta G, Jessup M, Pagano M. Increased proteasome-dependent kinase inhibitor p27 in aggressive colorectal carcinomas. Nature Med 3:231–234, 1997. 22. Newcomb EW, Sosnow M, Demopoulos RI, Zeleniuch-Jacquotte A, Sorich J, Speyer JL. Expression of the cell cycle inhibitor is a new prognostic marker associated with survival in epithelial ovarian tumors. Am J Pathol 154:119–125, 1999. 23. Del Pizzo JJ, Borkowski A, Jacobs SC, Kyprianou N. Loss of cell cycle regulators p27Kip1 and cyclin E in transitional cell carcinoma of the bladder correlates with tumor grade and patient survival. Am J Pathol 155:1129–1136, 1999. 24. Lu CD, Morita S, Ishibashi T, Hara H, Isozaki H, Tanigawa N. Loss of p27Kip1 expression independently predicts poor prognosis for patients with resectable pancreatic adenocarcinoma. Cancer 85:1250-1260, 1999. 25. Hannon GJ, Beach D. p15INK4B is a potential effector of TGF-β -induced cell cycle arrest. Nature 54:535-540, 1994. 26. Sherr CJ. Cancer cell cycles. Science 274:16721677, 1996. 27. Vlach J, Hennecke S, Alevizopoulos K, Conti D, Amati B. Growth arrest by the cyclindependent kinase inhibitor p27Kip1 is abrogated by c-myc. EMBO J 15:6569–6604, 1996. 28. Beghelli S, Pelosi G, Zamboni G, Falconi M, Iacono C, Bordi C, Scarpa A. Pancreatic endocrine tumors: Evidence for a tumor suppressor pathogenesis and for a tumor suppressor gene on chromososme 17p. J Pathol 186:41–™50, 1998. 29. Chung DC, Smith AP, Louis DN, GraemeCook F, Warshaw AL, Arnold A. Analysis of the retinoblastoma tumor suppressor gene in pancreatic endocrine tumors. Clin Endocr 47:523–528, 1997. 30. Höfler H, Ruhri C, Putz B, Wirnsberger G, Hauser H. Oncogene expression in endocrine pancreatic tumors. Virchows Arch B Cell Pathol 1988;55:355–361. 31. Pavelic K, Hrascan R, Kapitanovic S, Vranes Z, Cabrizan T, Spaventi S, Molecular genet-

32.

33.

34.

35.

36.

37.

38.

39.

40.

241

ics of malignant insulinoma. Anticancer Res 16:1707–1718, 1996. Masciullo V, Sgambato A, Pacilio C, Pucci B, Ferrandina G, Palazzo J, et al. Frequent loss of expression of the cyclin-dependent kinase inhibitor p27 in epithelial ovarian cancer. Cancer Res 59:3790–3794, 1999. Baldassarre G, Belletti B, Bruni P, Boccia A, Trapasso F, Pentimalli F, et al. Overexpressed cyclin D3 contributes to retaining the growth inhibitor p27 in the cytoplasm of thyroid tumor cells. J Clin Invest 104:865–874, 1999. Clarke MR, Baker EE, Weyant RJ, Hill L, Carty SE. Proliferative activity in pancreatic endocrine tumors: Association with function, metastases, and survival. Endocr Pathol 8:181–187, 1997. Iino K, Sasano H, Yabuki N, Oki Y, Kikuchi A, Yoshimi T, Nagura H. DNA topoisomerase IIα and Ki-67 in human adrenocortical neoplasms: a possible marker of differentiation between adenomas and carcinomas. Mod Pathol 10:901–907, 1997. Lynch BJ, Guinee DG, Holden JA. Human DNA topoisomerase II-alpha: a new marker of cell proliferation in invasive breast cancer. Hum Pathol 28:1180–1188, 1997. Rudolph P, MacGrogan G, Bonichon F, Frahm SO, de Mascarel I, Trojani M, et al. Prognostic significance of Ki-67 and topoisomerase IIα expression in infiltrating ductal carcinoma of the breast. Breast Cancer Res Treat 55:61– 71, 1999. Tanoguchi K, Sasano H, Yabuki N, Kikuchi A, Ito K, Sato S, Yajima A. Immunohistochemical and two-parameter flow cytometric studies of DNA topoisomerase II α in human epithelial ovarian carcinoma and germ cell tumor. Mod Pathol 11:186–193, 1998. Guinee Jr DG, Holden JA, Benfield JR, Woodward ML, Przygodzki RM, Fishback NF, Koss MN, Travis WD. Comparison of DNA topoisomerase IIα expression in small cell and nonsmall cell carcinoma of the lung. In search of a mechanism of chemotherapeutic response. Cancer 78:729–735, 1996. Jarvinen TAH, Kononen J, Pelto-Huikko M, Isola J. Expression of topoisomerase II alpha is associated with rapid cell proliferation, aneuploidy, and c-erbB2 overexpression in breast cancer. Am J Pathol 148:2073–2082, 1996.