Alterations in the p16INK4a/Cyclin D1/RB Pathway in Gastrointestinal ...

4 downloads 49 Views 956KB Size Report
Key Words: Endocrine carcinoma; Carcinoid; Gastrointestinal tract; p16INK4a; Cyclin D1; Retinoblastoma protein; pRB. DOI: 10.1309/TLLVXK9HVA89CHPE.
Anatomic Pathology / RB Signaling in GI Endocrine Tumors

Alterations in the p16INK4a/Cyclin D1/RB Pathway in Gastrointestinal Tract Endocrine Tumors Anna Fen-Yau Li, MD, PhD,1,2 Alice Chia-Heng Li,3 Shyh-How Tsay, MD,1 Wing-Yin Li, MD,1 Wen-Yih Liang, MD,1 and Jeou-Yuan Chen, PhD4 Key Words: Endocrine carcinoma; Carcinoid; Gastrointestinal tract; p16INK4a; Cyclin D1; Retinoblastoma protein; pRB DOI: 10.1309/TLLVXK9HVA89CHPE

Abstract A series of 76 cases of gastrointestinal tract endocrine tumors, including 21 poorly differentiated endocrine carcinomas (PDECs, small cell carcinomas) and 55 well-differentiated endocrine neoplasms (WDENs, carcinoids), 18 metastatic and 37 nonmetastatic rectal carcinoids, were examined by immunohistochemical analysis for p16INK4a, cyclin D1, and retinoblastoma protein (pRB) expression. Overexpression of p16INK4a was noted in 16 (76%) of the PDECs and none of the WDENs (P < .0001). Loss of pRB expression was demonstrated in 14 (67%) of the PDECs and 17 (31%) of the WDENs (P = .004). Overexpression of cyclin D1 was noted in 49 (89%) of the WDENs and 3 (14%) of the PDECs (P < .0001). Loss of pRB expression was noted in 11 (61%) of 18 metastatic WDENs and only 6 (16%) of 37 nonmetastatic WDENs (P = .001). The p16INK4a/cyclin D1/pRB pathway was altered in gastrointestinal tract endocrine tumors, and the loss of expression of pRB may be helpful in identifying patients at high risk of metastasis in rectal WDENs.

© American Society for Clinical Pathology

Endocrine cells are distributed throughout the body, and endocrine tumors are characterized by a marked diversity, which results from the large functional, structural, and embryological heterogeneity of normal endocrine cells.1-3 Among endocrine tumors, the respiratory system, gastrointestinal (GI) tract, and head and neck regions are the most common locations involved.1,2,4 Most of the studies on neuroendocrine tumors have focused on small cell lung carcinomas (SCLCs) and carcinoids of the respiratory system, and relatively little information is available on the pathogenesis of extrapulmonary neuroendocrine neoplasms. GI endocrine tumors constitute a heterogeneous group of tumors with differing clinical features, functional properties, and clinical courses according to their site of origin, secreting products, and histologic subtypes.1,5-10 The staging system and treatment protocols for GI tract endocrine tumors are mostly derived from experience with similar lesions in the lung, based on the presumption that these diseases are similar in many ways. The 2000 World Health Organization classification divides gastrointestinal endocrine tumors into well-differentiated endocrine neoplasms (WDENs, which include typical and atypical carcinoids) and poorly differentiated endocrine carcinomas (PDECs, small cell carcinomas).11 Clinically, WDENs are, in general, indolent and resistant to chemotherapy, whereas PDECs are more aggressive but may respond transiently to cisplatin-based chemotherapy.6,12 Retinoblastoma protein (pRB), p16INK4a, and cyclin D1 are major components of the RB pathway, which controls G1/S transition in the cell cycle, and are altered in many cancers, including SCLCs. Deregulation of the RB pathway through genetic and epigenetic alterations of its components is a frequent event in the vast majority of various types of human

535

Am J Clin Pathol 2008;130:535-542 DOI: 10.1309/TLLVXK9HVA89CHPE

535 535

Li et al / RB Signaling in GI Endocrine Tumors

cancer, such as SCLCs and gallbladder and gastric adenocarcinomas.13-18 The mechanisms involved include activating (CDK4) and inactivating (p16INK4a) mutations, deletions (RB and p16INK4a), amplifications (CCND1 and CDK4), transcription silencing by promoter methylation (p16INK4a and RB), and hyperphosphorylation (RB). In small cell carcinomas of the lung, the inactivation of both RB alleles at chromosomal region 13q14 is common, and more than 90% show RB protein abnormalities.13,19-21 The p16INK4a tumor suppressor protein functions as a G1-specific cell cycle inhibitor that prevents the association of cyclin-dependent kinase (CDK) 4 and CDK6 with D-type cyclin. Expression of p16INK4a was demonstrated in more than 95% of SCLCs.20 In carcinoid tumors of the lung, gain-of-function mutation in cyclin D1 is a common cause leading to an aberrant RB pathway.15,20 To better understand the possible involvement of the RB pathway in the pathogenesis of endocrine tumors in the GI tract, we previously demonstrated that overexpression of p16INK4a is found in more than 70% of PDECs of the GI tract but is rarely found in WDENs (carcinoids).22 An inverse relationship between the expression of pRB and p16INK4a has been found in many human cancers,20,23 and the expression of p16INK4a, pRB, and cyclin D1 in primary GI endocrine tumors is not yet well characterized. With the aim of investigating whether the RB signaling axis is involved in the pathogenesis of endocrine tumors in the GI tract, we determined the expression of p16INK4a, cyclin D1, and pRB in the spectrum of GI tract endocrine tumors, including PDECs and WDENs, by immunohistochemical analysis. Our findings showed that the components of the RB signaling pathway are frequently involved in GI tract endocrine tumors, and the loss of pRB expression may serve as a marker for predicting metastatic behavior in rectal WDENs.

Materials and Methods Cases Cases of primary endocrine tumors in the GI tract treated in Taipei Veterans General Hospital, Taipei, Taiwan, from January 1985 to December 2006 were reviewed for the availability of pathologic materials and medical records. Neuroendocrine differentiation of the lesions can be demonstrated by immunohistochemical studies of neuroendocrine markers: chromogranin A (LK2H10, BioGenex, San Ramon, CA), synaptophysin (Snm88, BioGenex), CD56 (NCAM, Novocastra Laboratories, Newcastle upon Tyne, England), and neuron-specific enolase (BBS/NC/VI-H14, DAKO, Glostrup, Denmark). The majority of tumors were diffusely and strongly positive for at least 2 of the 4 neuroendocrine markers. PDECs were distinguished from WDENs on morphologic grounds by following the 2000 World Health 536 536

Am J Clin Pathol 2008;130:535-542 DOI: 10.1309/TLLVXK9HVA89CHPE

Organization guidelines on the classification of endocrine tumors of the GI tract and lung.11,24 Briefly, a typical carcinoid is characterized by cells uniform in size with round or oval nuclei, inconspicuous nucleoli, and eosinophilic cytoplasm. The cells are arranged in trabecular anastomosing structures, tubular structures, and solid nests. Mitosis is almost absent, and angioinvasion is infrequent. Typical carcinoids have fewer than 2 mitoses per 2 mm2 and lack necrosis. Slight nuclear atypia and increased mitoses, 2 to 10 mitoses per 2 mm2 and/or foci of necrosis, can be found in atypical carcinoids. Typical carcinoids and atypical carcinoids are grouped together as WDENs; small cell carcinomas are considered PDECs, in which mitoses are usually greater than 10 per 2 mm2 and necrosis is more commonly found. Small cell carcinomas consist of small cells with scant cytoplasm, ill-defined cell borders, finely granular nuclear chromatin, and absent or inconspicuous nucleoli. Medical records and histopathologic sections were reviewed. Clinicopathologic data, including location of the primary neoplasm, histologic type, staging, metastasis, recurrence, type of treatment, and overall survival, were traced. To evaluate the involvement of metastasis, we included barium studies; computed tomography of the chest, abdomen, and brain; ultrasonography of the abdomen; and whole body bone scans. For patients with nonmetastatic carcinoid, more than 5 years of follow-up without evidence of metastasis was required. Specimens were obtained through radical surgery, excisional biopsy, or tumor debulking. Small specimens obtained by endoscopic biopsy and needle biopsy were insufficient for this study and were excluded. This study was carried out according to the guidelines of the Committee of Human Subjects, Taipei Veterans General Hospital, and consent from each patient was obtained. Immunohistochemical Staining Immunohistochemical staining was performed as previously described22 to examine the expression of proteins, including p16INK4a (JC8, Biocare Medical, Walnut Creek, CA), cyclin D1 (Sp4, Neomarkers, Fremont, CA), and pRB (RB1, DAKO). Endogenous peroxidase was quenched by hydrogen peroxide, and nonspecific adsorption was reduced by blocking against swine serum. The bound antibodies were detected by using the Liquid DAB substrate pack from BioGenex. Immunohistochemical analysis for chromogranin A (LK2H10), synaptophysin (Snm88), neuron-specific enolase (BBS/NC/VI-H14), and CD56 (NCAM) were performed to confirm the neuroendocrine differentiation and diagnosis. The immunopositivity of p16INK4a, cyclin D1, and pRB was interpreted without knowledge of the diagnosis. Only nuclear staining was considered specific and was scored as positive. Nonneoplastic fibroblasts, endothelial cells, mucosal glands, and lymphoid tissue of the GI tract served as internal © American Society for Clinical Pathology

Anatomic Pathology / Original Article

positive control samples for pRB and cyclin D1 stains. Cervical squamous cell carcinomas were included to serve as positive control samples for p16INK4a. A diffusely strong stain, usually more than 70% of the tumor cells, for p16INK4a was defined as overexpression of positive signals, and a focal or weak stain for p16INK4a was interpreted as negative for overexpression.22 For cyclin D1, nuclear staining in more than 5% of the nuclei within the tumor was read as overexpression; otherwise it was considered negative for overexpression. Staining of pRB was scored as negative if fewer than 1% of the tumor cells exhibited nuclear staining in contrast with the adequate nuclear staining of the surrounding nontumor tissue cells as an internal positive control. Staining of pRB was scored as positive if more than 1% of the tumor cells showed nuclear staining. Statistical Analysis The association between immunohistochemical staining and histopathologic data was estimated by using the Pearson χ2 test and the Fisher exact test, as appropriate. The association between histologic types and age was analyzed by using the Student t test. Computation was performed by using the Stata software program (Stata, College Station, TX). The significance level chosen was a P value lower than .05, and all tests were 2-sided.

Results Patients We studied 76 cases of GI tract endocrine tumors. Among them, 21 were cases of GI tract PDEC, including 5 cases in the esophagus, 9 gastric lesions, 1 small intestinal lesion, and 6 colorectal lesions; and 55 were typical carcinoids (WDENs), including 18 carcinoid tumors of the rectum with lymph node and/or liver metastasis, and 37 rectal carcinoid tumors without evidence of metastasis or recurrence after at least 5 years of follow-up. The mitotic counts of all 55 carcinoids were not greater than 2 mitoses per 2 mm2 and lacked necrosis, and the cases were, therefore, diagnosed as typical carcinoids; there were no atypical carcinoids in our study. The patients with PDEC were 17 men and 4 women with a mean ± SD age of 66.8 ± 15.3 years (range, 32-88 years). The subgroup of WDEN with metastasis included 13 men and 5 women with a mean ± SD age of 60.0 ± 12.0 years (range, 41-76 years). In the group of WDEN without metastasis, there were 20 men and 17 women with a mean ± SD age of 63.6 ± 12.5 years (range, 20-78 years). There was no statistically significant age difference among these 3 groups. Rectal carcinoids, which constituted 86.3% (138/160) of total carcinoids in the GI tract, were the most common WDENs in the local patients.25 Carcinoids in other parts of the GI tract were rare and were not included in this study. © American Society for Clinical Pathology

Immunohistochemical Staining for p16INK4a, Cyclin D1, and pRB By immunohistochemical staining, we examined the expression status of p16INK4a, cyclin D1, and pRB in the 21 PDECs of various origins and 55 WDENs, 18 metastatic and 37 nonmetastatic rectal carcinoids. Representative images of routine H&E staining and immunohistochemical studies of p16INK4a, cyclin D1, and pRB staining in PDECs, WDENs with metastasis, and WDENs without metastasis are shown in zImage 1z. The detailed phenotypes of p16INK4a and cyclin D1 overexpression and pRB expression are summarized in zTable 1z. The most common phenotype of PDEC was p16INK4a+/ CCND1–/pRB– (13/21 [62%]), the most common phenotype of metastatic carcinoid was p16INK4a–/CCND1+/pRB– (8/18 [44%]), and p16INK4a–/CCND1+/pRB+ was the most common phenotype in nonmetastatic carcinoids (30/37 [81%]). Our data demonstrate that overexpression of p16INK4a was frequently observed in PDECs (16/21 [76%]), and none of the 55 WDENs were found to overexpress p16INK4a (P < .0001). Overexpression of cyclin D1 was more commonly observed in WDENs than in PDECs (49/55 [89%] vs 3/21 [14%], respectively; P < .0001). A similar approach revealed that pRB expression was detected in 7 (33%) of 21 PDEC and 38 (69%) of 55 WDEN cases (P = .004). We concluded that overexpression of p16INK4a (P < .0001) and loss of pRB expression (P = .004) were significantly associated with PDECs and not with WDENs zTable 2z; in contrast, overexpression of cyclin D1 was significantly associated with WDENs (P < .0001). Regional differences in morphologic features, protein expression, and behavior of endocrine cell tumors in different parts of the GI tract are well recognized. We therefore further compared this series of endocrine tumors in different settings. The 5 esophagus PDECs and 9 stomach PDECs were grouped together as PDECs in the upper GI tract (foregut) and compared with 6 colon PDECs (hindgut). The only case of a small intestinal PDEC (terminal ileum, midgut) was p16INK4a+/CCND1–/pRB– and was not included in the comparison. Overexpression of p16 and loss of pRB expression were confirmed as being common in foregut and hindgut PDECs, but significantly increased cyclin D1 overexpression was noted in hindgut PDECs (P = .0175) zTable 3z. PDECs of the colon and WDENs in the rectum were further compared, and significant p16 overexpression in PDECs (P = .0005) and cyclin D1 overexpression in WDENs (P = .03) was confirmed zTable 4z. In WDENs, a loss of pRB expression was found in 11 (61%) of 18 metastatic carcinoids, whereas only 6 (16%) of 37 carcinoids with no signs of metastasis had lost pRB expression zTable 5z. There was a close association of the loss of RB with metastatic carcinoids (P = .001); there was also the trend that a higher percentage of nonmetastatic carcinoids were positive for cyclin D1 overexpression than

537

Am J Clin Pathol 2008;130:535-542 DOI: 10.1309/TLLVXK9HVA89CHPE

537 537

Li et al / RB Signaling in GI Endocrine Tumors

A

B

C

D

E

F

zImage 1z The most common phenotypes are demonstrated: p16INK4a+/CCND1–/retinoblastoma protein (pRB)– in poorly differentiated endocrine carcinoma (of esophagus, A, D, G, and J), p16INK4a–/CCND1+/pRB– in metastatic well-differentiated endocrine neoplasm (WDEN; metastatic rectal carcinoid, B, E, H, and K), and p16INK4a–/CCND1+/pRB+ in nonmetastatic WDEN (nonmetastatic rectal carcinoid, C, F, I, and L). A-C, H&E, ×400. D, Immunostaining showing strong nuclear staining on most tumor cells positive for p16INK4a overexpression (×400). E, Immunostaining showing scattered nuclear staining, negative for p16INK4a overexpression (×400). F, Immunostaining negative for p16INK4a overexpression (×400). 538 538

Am J Clin Pathol 2008;130:535-542 DOI: 10.1309/TLLVXK9HVA89CHPE

© American Society for Clinical Pathology

Anatomic Pathology / Original Article

G

H

I

J

K

L

G, Negative cyclin D1 immunostaining, no cyclin D1 overexpression (×400). H, Positive cyclin D1 immunostaining with more than 5% positive cells, positive for cyclin D1 overexpression (×400). I, Positive cyclin D1 immunostaining with more than 5% positive cells, positive for cyclin D1 overexpression (×400). J, Negative pRB immunostaining (×400). K, Negative pRB immunostaining. Note the presence of endothelial cells with strong nuclear staining. The weak background cytoplasmic staining was not considered as specific (×400). L, pRB immunostaining showing strongly stained nuclei on most tumor cells, no loss of pRB (×400). © American Society for Clinical Pathology

539

Am J Clin Pathol 2008;130:535-542 DOI: 10.1309/TLLVXK9HVA89CHPE

539 539

Li et al / RB Signaling in GI Endocrine Tumors

zTable 1z p16INK4a/Cyclin D1/pRB Phenotypes in 76 Gastrointestinal Tract Endocrine Tumors*



Phenotype

PDEC (n = 21)



p16INK4a+/CCND1–/pRB– p16INK4a+/CCND1–/pRB+ p16INK4a+/CCND1+/pRB+ p16INK4a–/CCND1–/pRB– p16INK4a–/CCND1–/pRB+ p16INK4a–/CCND1+/pRB– p16INK4a–/CCND1+/pRB+





Metastatic Carcinoid (n = 18)





13 (62) 2 (10) 1 (5) 0 (0) 3 (14) 1 (5) 1 (5)



0 (0) 0 (0) 0 (0) 3 (17) 1 (6) 8 (44) 6 (33)



Nonmetastatic Carcinoid (n = 37) 0 (0) 0 (0) 0 (0) 1 (3) 1 (3) 5 (14) 30 (81)

CCND1+, CCND1–, positive or negative for cyclin D1 overexpression, respectively; p16INK4a+, p16INK4a–, positive or negative for p16INK4a overexpression, respectively; PDEC, poorly differentiated endocrine carcinoma; pRB, retinoblastoma protein; pRB+, pRB–, positive or negative for pRB expression, respectively. * Data are given as number (percentage).

zTable 2z Immunohistochemical Profiles of p16INK4a, Cyclin D1, and pRB in PDEC and WDEN of the Gastrointestinal Tract*

Positive



p16INK4A overexpression PDEC (n = 21) WDEN (n = 55) Cyclin D1 overexpression PDEC (n = 21) WDEN (n = 55) pRB expression PDEC (n = 21) WDEN (n = 55)









Negative

16 5 0 55 3 18 49 6 7 14 38 17

P