550 Differential Retinoblastoma and p16INK4A Protein Expression in Neuroendocrine Tumors of the Lung Hirotoshi Dosaka-Akita, M.D., Ph.D.1 Philip T. Cagle, M.D.2 Hiromitsu Hiroumi, M.D.1 Masahiro Fujita, M.D., Ph.D.3 Motoyuki Yamashita, M.D., Ph.D.4 Anupama Sharma, M.D.5 Yoshikazu Kawakami, M.D., Ph.D.1 William F. Benedict, M.D.4 1 First Department of Medicine, Hokkaido University School of Medicine, Sapporo, Japan. 2 Department of Pathology, Baylor College of Medicine, Houston, Texas. 3 Department of Pathology, National Sapporo Hospital, Sapporo, Japan. 4 Department of Genitourinary Medical Oncology, University of Texas M. D. Anderson Cancer Center, Houston, Texas. 5 Department of Pathology, Duke University Medical Center, Durham, North Carolina. Supported in part by National Institute of Health Grant CA-54672 to William F. Benedict and by a Grant-in-Aid (No. 11670558) from the Minister of Education, Science and Culture of Japan to Hirotoshi Dosaka-Akita. The authors thank Michio Shimizu, Department of Pathology, Hokkaido University Medical Hospital; Kenzo Okamoto, Department of Pathology, Hokkaido Kin-ikyou Chuo Hospital; Takehito Nakabayashi, Department of Respirology, National Sapporo Hospital; Tamotsu Hirata, Department of Surgery, National Sapporo Minami Hospital; Yoshikazu Araya, Department of Respirology, National Hakodate Hospital; Tetsuo Shimizu and Toshiaki Fujikane, Department of Medicine, National Dohoku Hospital; Masato Hashimoto, Department of Surgery, Iwamizawa Rousai Hospital; Takashi Yoshikawa, First Department of Medicine, Obihiro Kousei Hospital; and Kazuo Takaoka, Department of Respirology, Nikkou Memorial Hospital; for resected tumors and clinical data. Address for reprints: Hirotoshi Dosaka-Akita, M.D., Ph.D., First Department of Medicine, Hokkaido University School of Medicine, North 15, West 7, Kita-ku, Sapporo, 060-8638, Japan. Received June 18, 1999; revision received October 6, 1999; accepted October 6, 1999. © 2000 American Cancer Society BACKGROUND. Neuroendocrine neoplasms of the lung represent a wide spectrum of phenotypically and biologically distinct entities. Their histopathologic diagnosis, which carries therapeutic and prognostic significance, may sometimes be difficult because of their overlapping features. We previously demonstrated that large cell neuroendocrine carcinomas (LCNECs) and small cell lung carcinomas (SCLCs) failed to show positive nuclear staining of RB protein (RB⫺), whereas typical and atypical carcinoids (TCs and ACs) showed nuclear RB immunostaining (RB⫹). METHODS. In the current study, a series of 58 surgically resected lung tumors, of which 33 tumors were initially diagnosed as SCLCs and 25 as TCs or ACs, were studied for RB and p16 protein expression by immunohistochemistry. They were also reviewed for their pathologic diagnosis; the reviewers were blinded to the RB and p16 protein status. RESULTS. Nineteen tumors were diagnosed as TCs, 5 as ACs, 7 as LCNECs, and 27 as SCLCs. Three of seven LCNECs were RB⫹, whereas the other four were RB⫺. In contrast, all 19 TCs were RB⫹ and all 27 SCLCs were RB⫺. In addition, two of five ACs were RB⫹, whereas the other three were RB⫺. Interestingly, all 3 RB⫹ LCNECs and the 1 RB⫹ AC tested failed to show nuclear staining of p16 protein in any tumor cells (p16⫺), although some normal stromal cells showed nuclear staining of p16 protein (p16⫹) as positive internal controls, indicating loss of p16 function in these tumors. It is also noteworthy that the three RB⫹ LCNECs were initially diagnosed as SCLCs and one of the RB⫺ ACs was initially considered a TC. With the exception of TCs, tumors were significantly more prevalent among heavy smokers with ⬎20 pack-years compared with nonsmokers and light smokers with ⱕ20 pack-years (P ⬍ 0.01). CONCLUSIONS. These findings suggest that all SCLCs and LCNECs have abnormalities in the p16:RB pathway, as do at least certain ACs, whereas the p16:RB pathway is normal in TCs. Cancer 2000;88:550 –56. © 2000 American Cancer Society. KEYWORDS: neuroendocrine tumors of the lung, retinoblastoma protein, p16INK4A protein, diagnostic marker, immunohistochemistry. N euroendocrine (NE) neoplasms of the lung represent a wide spectrum of pathologically and biologically distinct entities.1– 4 Of these, atypical carcinoids (ACs) and large cell neuroendocrine carcinomas (LCNECs) are much less commonly encountered neoplasms than the other histologic types of NE tumors, including typical carcinoids (TCs) and small cell lung carcinomas (SCLCs). In the past, LCNECs may have been classified as either atypical carcinoids or SCLCs because of their overlapping features.5,6 It has been recently indicated that 1) classification of NE tumors of the lung is most reproducible for classification of TC and SCLC but less reproducible for AC and LCNEC, and 2) a need for more careful definition and application of criteria for TC versus AC and for SCLC versus LCNEC.7 Moreover, the correct histopathologic diagnosis of specific NE tumors may be important for the consideration of different therapeutic interventions and may have potential prognostic implications.6,8 RB and 16NK4A Proteins in Neuroendocrine Lung Tumors/Dosaka-Akita et al. It has been reported that the frequency of tumor suppressor gene alterations, including those of the p53 and retinoblastoma (RB) genes, varies among categories of NE tumors.9 –22 Provided that the identification of a specific alteration is consistently and reproducibly distinctive among specific NE tumors, such alterations might be useful for the differential diagnosis of NE tumors of the lung. The RB gene is a prototypical tumor suppressor gene encoding a nuclear phosphoprotein with a molecular weight of 105–110 kD.23 The functional loss of RB protein is believed to be a key event in the development of a variety of neoplasms, including SCLCs.24 Most RB gene alterations result in the loss of RB protein expression or in a truncated RB protein, which does not enter the nucleus. Heterogeneous positive nuclear RB immunostaining is, in general, indicative of normal RB function, whereas negative intranuclear RB immunostaining in all tumor cells reflects functional loss of the RB gene.25,26 Moreover, to date, the presence or absence of normal RB protein expression determined by immunohistochemistry has been found to be the most sensitive and specific method for determining RB status in a given tumor.27 We have previously demonstrated that SCLCs and LCNECs failed to show positive nuclear staining of RB protein by immunohistochemistry; in contrast, TCs and ACs showed nuclear RB immunostaining, although the latter had a stronger and more homogeneous staining pattern than the former.28 p16INK4A (p16) protein, the product of the CDKN2 gene, which has been found to be deleted in a variety of tumor cells,29,30 inhibits CDK4- and CDK6-mediated phosphorylation of RB protein.31,32 Therefore, p16 and RB proteins are suggested to function in a single regulatory pathway of the cell cycle and tumor suppression,31,32 which is supported by observation of an inverse correlation between alterations of both proteins in primary lung carcinomas and lung carcinoma cell lines.33–37 Immunohistochemical detection of p16 protein is a sensitive and specific method of screening for p16 alterations resulting from both homozygous deletion and DNA hypermethylation.37 We have recently reported that RB-negative bladder tumors exhibited strong nuclear p16 staining, whereas each tumor showing strong, homogeneous nuclear RB staining (overexpression) was p16 negative.38 These findings support the idea that loss of p16 protein function could be related to RB overexpression, since p16 can induce transcriptional down-regulation of RB and its loss may lead to aberrant regulation of RB expression. Conversely, loss of RB function has been associated with high p16 protein expression in several types of tumors, including nonsmall cell lung carcinomas.37 551 In the current study, we investigated expression of RB and p16 proteins in surgically resected NE tumors of the lung, and showed the potential implication of both proteins in their diagnosis. MATERIALS AND METHODS Tumor Specimens A total of 58 lung tumor specimens, which had been surgically resected and initially diagnosed at surgery as carcinoids (25 tumors) or SCLCs (33 tumors) between 1979 and 1996, were collected from a cohort of surgically resected lung tumor specimens and analyzed in the current study. For all 33 tumors diagnosed as SCLC, biopsies and cytologic examination before surgery were not diagnostic, and the diagnosis of SCLC was made after surgery only by subsequent pathologic analysis. The tumors were obtained at Hokkaido University Medical Hospital, National Sapporo Hospital, National Sapporo Minami Hospital, National Hakodate Hospital, National Dohoku Hospital, Iwamizawa Rosai Hospital, Obihiro Kosei Hospital, Hokkaido Kin-ikyo Chuo Hospital, and Nikko Memorial Hospital, all located in Hokkaido, Japan, and evaluated by different pathologists at the time of surgery as routine pathologic practice in each hospital. In this cohort, 19 tumors were diagnosed as TCs, 6 as ACs, and 33 as SCLCs. Subsequently, the 58 tumors were analyzed for their RB and p16 protein expression and were also independently reclassified by one of us (P.T.C.) according to established histopathologic criteria28 and blinded to the RB and p16 protein status. Briefly, the tumors were evaluated histologically for a spectrum of features, including organoid pattern, mitoses, necrosis, nuclear pleomorphism, nucleoli, nuclear-to-cytoplasmic ratio, nuclear chromatin, and evidence of neuroendocrine differentiation. Clinical and clinicopathologic data at surgery were available on all 58 tumors. Immunohistochemical Analysis Immunohistochemical staining for RB and p16 proteins was performed on formalin fixed, paraffin embedded tissue sections. The methods for staining RB nuclear protein have been described previously.26,39 Briefly, after deparaffinization and hydrogen peroxide treatment of 5 m sections from formalin fixed, paraffin embedded tissue blocks, they were processed for immunostaining for RB protein with an affinity-purified polyclonal anti-RB antibody, RB-WL-1,40 at a dilution of 1:628. The RB immunostaining pattern for each case was independently evaluated by two investigators (H.D.-A., W.F.B.). Tumors were scored for the presence or absence of nuclear RB staining using criteria previously established.41 The entire specimens 552 CANCER February 1, 2000 / Volume 88 / Number 3 were examined at ⫻100 and ⫻400 magnifications. To be considered adequate, a tumor section had to contain numerous normal RB positive stromal cells as internal controls. To be scored RB negative (RB⫺), all tumor cells in the section had to show no nuclear RB staining when there were adjacent RB positive normal stromal cells in the same tumor section. Tumors were scored as RB positive (RB⫹) if several tumor nuclei had RB staining, although in all RB⫹ cases the majority of tumor cells had some nuclear staining in this cohort. Results of the RB immunostaining patterns were then compared with the histopathologic diagnosis category and with the clinical and clinicopathologic data. Immunohistochemical staining of p16 nuclear protein was performed as previously described.38 Briefly, after deparaffinization, hydrogen peroxide treatment, and antigen retrieval procedure of 5 m sections from formalin fixed, paraffin embedded tissue blocks, they were processed for immunostaining for p16 protein with a mouse monoclonal antibody against human p16 protein, NCL-p16, clone DCS-50 (Vector Laboratories, Burlingame, CA) at a 1:25 dilution. The p16 immunostaining pattern for each case was independently evaluated by two investigators (H.D.-A., W.F.B.). The criteria for the evaluation of the presence or absence of nuclear p16 staining (p16⫹ or p16⫺) were the same as those described previously for RB protein. Statistical Analysis The associations between histologic types and various characteristics were analyzed by the chi-square test or the Fisher exact test as appropriate.42 The association between histologic types and age was analyzed by the Student t test. The significance level chosen was P ⬍ 0.05, and all tests were two-sided. RESULTS A series of surgically resected 58 lung tumors were reevaluated for their pathologic diagnosis. Nineteen tumors initially diagnosed as TCs at surgery were reevaluated and determined to be 18 TCs and 1 AC; 6 tumors initially diagnosed as ACs were reevaluated and found to be 1 TC, 4 ACs, and 1 LCNEC; and 33 tumors initially diagnosed as SCLCs were reevaluated and found to be 6 LCNECs and 27 SCLCs. Therefore, 19 tumors were subsequently diagnosed as TCs, 5 as ACs, 7 as LCNECs, and 27 as SCLCs by pathologic reevaluation in this study (Table 1). These tumors were studied for their RB and p16 protein expression by immunohistochemistry. Three of seven LCNECs were RB⫹ (Fig. 1C) and the other four were RB⫺, whereas numerous RB⫹ normal stro- TABLE 1 Pathologic Diagnosis and RB/p16 Protein Status in Neuroendocrine Tumors of the Lung Diagnosis at surgery TC 19 AC 6 SCLC 33 Diagnosis in this study TC 18 ACa 1 TCa 1 AC 4 LCNECa 1 LCNECa 6 SCLC 27 RB/p16 status ⫹/⫹ or NE 18 ⫺/⫹b 1 ⫹/⫹ 1 ⫺/⫹ or NE 2 ⫹/⫺ or NE 2 ⫺/⫹ 1 ⫹/⫺b 3 ⫺/⫹ 3 ⫺/⫹ 27 TC: typical carcinoid; AC: atypical carcinoid; SCLC: small cell lung carcinoma; NE: not evaluable because of the loss of adequate p16 protein expression in positive internal control cells (normal stromal cells). a Diagnosis reclassified by P.T.C. b Initial diagnosis at surgery was inconsistent with RB/p16 status. mal cells were present in every section as internal controls (Fig. 1A). Two of five ACs were RB⫹ and the other three, one of which contained several cells with a polygonal shape and more vesicular nuclei, reminiscent of LCNEC, were RB⫺, although again numerous RB⫹ stromal cells were present in the sections. In contrast, all 19 TCs were RB⫹ and all 27 SCLCs were RB⫺. This difference in the loss of RB protein among specific NE tumors was statistically significant (TCs vs. ACs, P ⬍ 0.01; TCs vs. LCNECs, P ⬍ 0.01; TCs vs. SCLCs, P ⬍ 0.01; ACs vs. SCLCs, P ⬍ 0.01; LCNECs vs. SCLCs, P ⬍ 0.01) (Table 2). No differences in the histologic features were found between RB⫹ and RB⫺ LCNECs and between RB⫹ and RB⫺ ACs. The reciprocal loss of p16 and RB proteins has been reported in primary lung carcinomas and lung carcinoma cell lines,33–37 to support the hypothesis that both proteins function in a single regulatory pathway of the cell cycle and tumor suppression.31,32 Interestingly, all three RB⫹ LCNECs and the one RB⫹ AC tested failed to show nuclear staining of p16 protein in any tumor cells, although some normal stromal cells were p16⫹ as positive internal controls (Fig. 1D; Table 3), indicating loss of p16 function in these tumors. All four RB⫺ LCNECs and the two RB⫺ ACs examined manifested strong p16 immunostaining (Fig. 1B; Table 3), which was also observed in SCLCs. This difference in the loss of p16 protein among NE tumors was statistically significant (ACs vs. SCLCs, P ⬍ 0.01; LCNECs vs. SCLCs, P ⬍ 0.01) (Table 2). These findings suggest that all SCLCs and LCNECs have abnormalities in the p16:RB pathway, as do at least certain ACs, by loss of either RB or p16 protein in LCNECs and ACs and by loss of RB protein in SCLCs (Tables 2 RB and 16NK4A Proteins in Neuroendocrine Lung Tumors/Dosaka-Akita et al. 553 FIGURE 1. Immunohistochemical staining of RB (A and C) and p16 proteins (B and D) in large cell neuroendocrine carcinomas is shown. The staining pattern of an RB⫺/p16⫹ tumor is shown in A and B, and staining for an RB⫹/p16⫺ tumor is shown in C and D. A representative RB⫹ normal stromal cell is shown in A (large arrow) and two p16⫹ stromal cells are seen in D (small arrows) as internal controls. and 3). Of note, one RB⫹/p16⫹ tumor initially diagnosed as AC at surgery was found to be a TC. Moreover, one RB⫺/p16⫹ tumor initially diagnosed as TC was found to be an AC, and three RB⫹/p16⫺ tumors initially diagnosed as SCLCs were found to be LCNECs by pathologic reevaluation blinded to the RB and p16 protein status (italic in Table 1). The associations between histologic types of NE tumors and clinical and clinicopathologic characteristics were next analyzed (Table 2). Patients with LCNECs and SCLCs were significantly older than those with TCs (P ⫽ 0.03 and P ⬍ 0.01, respectively), whereas patients with ACs were older but not significantly older than those with TCs. All three patients with ACs, 6 of 7 with LCNECs, and 22 of 26 with SCLCs whose smoking habits could be identified were heavy smokers with more than 20 pack-years, whereas only 2 of 13 patients with TCs were heavy smokers (P ⬍ 0.01). DISCUSSION In the current study, we showed that all SCLCs and LCNECs had abnormalities in the p16:RB pathway, as did at least certain ACs, by loss of RB protein in SCLCs and by reciprocal loss of either RB or p16 protein in LCNECs and ACs, whereas the p16:RB pathway was normal in TCs. In the previous study,28 using resected specimens, wedge biopsies, and bronchoscopic biopsies, we showed the distinct differences in RB expression in SCLCs and LCNECs compared with TCs and ACs: 40 SCLCs and 6 LCNECs failed to show RB staining, whereas 44 TCs and 15 ACs manifested RB staining, although the latter showed a stronger and more homogeneous RB staining pattern than the former. Based on the recent findings in bladder carcinomas in which strong and homogeneous staining of RB protein was associated with loss of p16 protein function,38 many of the previously reported ACs may have been 554 CANCER February 1, 2000 / Volume 88 / Number 3 TABLE 2 Clinical and Clinicopathologic Characteristics of RB-Stained and p16-Stained Neuroendocrine Tumors of the Lung Characteristics TCs ACs LCNECs SCLCs P valuea Age (yrs, mean ⫾ SD) (Range) Gender Male Female Smoking (pack-years)b 0–20 ⬎20 pStagec I II IIIa IIIb IV RB (⫹) (⫺) p16d (⫹) (⫺) 50.2 ⫾ 14.0 21–67 60.4 ⫾ 12.1 45–72 63.7 ⫾ 10.9 46–82 62.1 ⫾ 8.9 36–74 ⬍0.01 (TCs/SCLCs) 0.03 (TCs/LCNECs) 11 8 5 0 6 1 21 6 NS 11 2 0 3 1 6 4 22 ⬍0.01 12 1 1 0 0 3 0 0 0 1 2 2 1 0 1 9 7 8 1 2 NS 19 0 2 3 3 4 0 27 ⬍0.01 7 0 2 1 4 3 27 0 ⬍0.01 a P values determined by the Student t test for age, and P values by the chi-square test or the Fisher exact test as appropriate for others. NS: not statistically significant. b Data on smoking habits were available on patients with 13 TCs, 3 ACs, 7 LCNECs, and 26 SCLCs. c Data on pStage were available in 14 TCs, 4 ACs, 6 LCNECs, and 27 SCLCs. d Not evaluable for 12 TCs and 2 ACs because of the loss of adequate p16 protein expression in positive internal control cells (normal stromal cells). TABLE 3 RB and p16 Protein Expression in ACs and LCNECs Tumors ACs # 13-4 # 13-6 # 17-2 # 18-3 # 18-4 LCNECs # 2-3 # 3-5 # 7-1 # 13-5 # 1-2 # 3-3 # 3-4 RB p16 ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ NEa ⫹ NE ⫺ ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ RB: retinoblastoma; ACs: atypical carcinoids; LCNECs: large cell neuroendocrine carcinomas. a Not evaluable because of the loss of adequate p16 protein expression in positive internal control cells (normal stromal cells). p16⫺ in retrospect and need to be evaluated for their p16 status. The current study indicates that LCNECs could have intact RB protein, but all such RB⫹ LCNECs had lost p16 function. In addition, AC may show reciprocal loss of RB and p16 proteins as men- tioned above, whereas all of the SCLCs studied by us to date have lost RB function as reported in the previous28 and current studies. Accurate diagnosis of specific NE tumors may influence the therapy given and the clinical outcome.6,8 In this sense, different patterns of RB and p16 protein expression between TCs and ACs and between LCNECs and SCLCs determined by immunohistochemistry may be helpful in the differential diagnosis of specific NE tumors. It is sometimes difficult to distinguish AC from TC and LCNEC from SCLC even in surgically resected specimens,7 and these tumors are often diagnosed based on bronchoscopic biopsies, fine-needle aspiration biopsies, or biopsies of lymph node metastases, which are smaller and more crushed compared with surgically resected specimens. The separation of AC from TC is based on numbers of mitotic figures and necrosis as well as disorganized architecture, cellularity, and cytologic atypia.43 It may remain difficult, therefore, in certain cases, to distinguish AC from TC in a given tumor, even though new criteria for separation of AC from TC has been recently proposed.8 LCNECs generally have more cytoplasm than SCLCs and are unlikely to be spindle cells but show considerable overlap in other features with RB and 16NK4A Proteins in Neuroendocrine Lung Tumors/Dosaka-Akita et al. SCLCs, including organoid pattern, a high frequency of mitosis, and areas of necrosis. Consequently, the RB and p16 protein status determined by immunohistochemistry may on occasion be a diagnostic marker for ACs versus TCs and for LCNECs versus SCLCs. Discrepant RB and p16 immunostaining patterns for TCs, such as either negative RB or p16 immunostaining, and for SCLCs, such as positive RB and negative p16 immunostaining, may be a basis for reevaluation of the histopathologic diagnosis of a given tumor. In fact, in the current study, one RB⫺/p16⫹ tumor initially diagnosed as TC was found to be an AC, and three RB⫹/p16⫺ tumors initially diagnosed as SCLC were found to be LCNECs by pathologic reevaluation (Table 1). In addition, one RB⫹/p16⫹ tumor initially diagnosed as AC was found to be a TC (Table 1). If RB⫹ ACs are found to be negative for p16 expression in future studies, this may also allow a distinction between TCs and ACs to be made using this criterion. In any case, more studies with larger cohorts of NE tumors of the lung will be needed to conclude that RB and p16 protein status can be of definite use in histologic classification, because we analyzed small numbers of ACs and LCNECs in the current study. Although NE tumors of the lung share common phenotypic features, suggesting a genotypic relation, they differ remarkably in their molecular genetic and cytogenetic characteristics, including loss of RB and p16 proteins,9,11,12,17,20,28 p53 mutations,9,10,12,14,16,17,22 MEN1 gene mutations,18 loss of heterozygosity (LOH) of 3p alleles,15,19,22 and 11q deletion by comparative genomic hybridization,21 highlighting an early fundamental molecular genetic divergence during the development of these tumors. The significant association between specific NE tumor types, including AC, LCNEC, and SCLC, and heavy smoking with more than 20 pack-years; and the finding that patients with ACs, LCNECs, and SCLCs were older than those with TCs in this study suggest that development of these types of NE lung tumors may be caused by alterations of the genes, including the RB gene damaged by carcinogens in tobacco smoke in older individuals after certain periods of carcinogen exposure. In conclusion, the current study indicates that the p16:RB pathway of tumor suppression can be altered by loss of either RB or p16 protein in ACs and LCNECs and by loss of RB protein in SCLCs, whereas the p16:RB pathway is normal in TCs. The RB and p16 protein status determined by immunohistochemistry may also be useful as a diagnostic marker for certain cases of LCNECs versus SCLCs and for ACs versus TCs. 555 REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. Warren WH, Faber LP, Gould VE. Neuroendocrine neoplasms of the lung. J Thorac Cardiovasc Surg 1989;98:321–32. Colby TV, Koss MN, Travis WD. Carcinoid and other neuroendocrine tumors. In: Rosai J and Sobin LH, editors. Atlas of tumor pathology. 3rd series, Fascicle 13. Tumors of the lower respiratory tract. Washington, DC: Armed Forces Institute of Pathology, 1995;287–317. Cagle PT. Tumors of the lung (excluding lymphoid tumors). In: Thurlbeck WM, Churg AM, editors. Pathology of the lung. 2nd edition. New York: Thieme Medical Publishers, Inc., 1995;437–552. Linnoila RI. Spectrum of neuroendocrine differentiation in lung cancer cell line featured by cytomorphology, markers, and their corresponding tumors. J Cell Biochem (Suppl) 1996;24:9 –106. Travis WD, Linnoila RI, Tsokos MG, Hitchcock CL, Cutler GB Jr., Nieman L, et al. Neuroendocrine tumors of the lung with proposed criteria for large-cell neuroendocrine carcinoma. Am J Surg Pathol 1991;15:529 –53. Colby TV, Koss MN, Travis WD. Small cell carcinoma and large cell neuroendocrine carcinoma. In: Rosai J and Sobin LH, editors. Atlas of tumor pathology. 3rd series, Fascicle 13. Tumors of the lower respiratory tract. Washington, DC: Armed Forces Institute of Pathology, 1995:235–57. Travis WD, Gal AA, Colby TV, Klimstra DS, Falk R, Koss MN. Reproducibility of neuroendocrine lung tumor classification. Hum Pathol 1998;29:272–9. Travis WD, Rush W, Flieder DB, Falk R, Fleming MV, Gal AA, et al. Survival analysis of 200 pulmonary neuroendocrine tumors with clarification of criteria for atypical carcinoid and its separation from typical carcinoid. Am J Surg Pathol 1998;22:934 – 44. Barbareschi M, Girlando S, Mauri FA, Arrigoni G, Laurino L, Palma PD, et al. Tumor suppressor gene products, proliferation, and differentiation markers in lung neuroendocrine neoplasms. J Pathol 1992;166:343–50. Lohmann DR, Fesseler B, Putz B, Reich U, Bohm J, Prauer H, et al. Infrequent mutations of the p53 gene in pulmonary carcinoid tumors. Cancer Res 1993;53:5797– 801. Gouyer V, Gazzeri S, Brambilla E, Bolon I, Moro D, Perron P, et al. Loss of heterozygosity at the RB locus correlates with loss of RB protein in primary malignant neuroendocrine lung carcinomas. Int J Cancer 1994;58:818 –24. Lai S, Brauch H, Knutsen T, Johnson BE, Nau MM, Mitsudomi T, et al. Molecular genetic characterization of neuroendocrine lung cancer cell lines. Anticancer Res 1995;15: 225–32. Jiang S, Kameya T, Sato Y, Yanase N, Yoshimura H, Kodama T. Bcl-2 protein expression in lung cancer and close correlation with neuroendocrine differentiation. Am J Pathol 1996;148:837– 46. Przygodzki RM, Finkelstein SD, Langer JC, Swalsky PA, Fishback N, Bakker A, et al. Analysis of p53, K-ras, and c-raf-1 in pulmonary neuroendocrine tumors. Am J Pathol 1996;148: 1531– 41. Hurr K, Kemp B, Silver SA, El-Naggar AK. Microsatellite alteration at chromosome 3p loci in neuroendocrine and nonneuroendocrine lung tumors. Am J Pathol 1996;149:613– 20. Brambilla E, Negoescu A, Gazzeri S, Lantuejoul S, Moro D, Brambilla C, et al. Apoptosis-related factors p53, Bc12, and Bax in neuroendocrine lung tumors. Am J Pathol 1996;149: 1941–52. 556 CANCER February 1, 2000 / Volume 88 / Number 3 17. Rusch VW, Klimstra DS, Venkatraman ES. Molecular markers help characterize neuroendocrine lung tumors. Ann Thorac Surg 1996;62:798 – 810. 18. Debelenko LV, Brambilla E, Agarwal SK, Swalwell JI, Kester MB, Lubensky IA, et al. Identification of MEN1 gene mutations in sporadic carcinoid tumors of the lung. Hum Mol Genet 1997;6:2285–90. 19. Kovatich AK, Friedland DM, Druck T, Hadaczek P, Huebner K, Comis RL, et al. Molecular alterations to human chromosome 3p loci in neuroendocrine lung tumors. Cancer 1998; 83:1109 –17. 20. Gazzeri S, Valle VD, Chaussade L, Brambilla C, Larsen CJ, Brambilla E. The human p19ARF protein encoded by the b transcript of the p16INK4a gene is frequently lost in small cell lung cancer. Cancer Res 1998;58:3926 –31. 21. Walch AK, Zitzelsberger HF, Aubele MM, Mattis AE, Bauchinger M, Candidus S, et al. Typical and atypical carcinoid tumors of the lung are characterized by 11q deletions as detected by comparative genomic hybridization. Am J Pathol 1998;153:1089 –98. 22. Onuki N, Wistuba I, Travis WD, Virmani AK, Yashima K, Brambilla E, et al. Genetic changes in the spectrum of neuroendocrine lung tumors. Cancer 1999;85:600 –7. 23. Weinberg RA. The retinoblastoma protein and cell cycle control. Cell 1995;81:323–30. 24. Benedict WF, Xu H, Hu S, Takahashi R. Role of the retinoblastoma gene in the initiation and progression of human cancer. J Clin Invest 1990;85:988 –93. 25. Xu H, Hu S, Benedict WF. Lack of nuclear RB protein staining in G0/middle G1 cells: correlation to changes in total RB protein level. Oncogene 1991;6:1139 – 46. 26. Xu H, Hu S, Cagle PT, Moore GE, Benedict WF. Absence of retinoblastoma protein expression in primary non-small cell lung carcinomas. Cancer Res 1991;51:2735–9. 27. Zhang X, Xu H, Murakami Y, Sachse R, Yashima K, Hirohashi S, et al. Deletion of chromosome 1q, mutations in retinoblastoma 1, and retinoblastoma protein state in human hepatocellular carcinoma. Cancer Res 1994;54:4177– 82. 28. Cagle PT, El-Naggar AK, Xu H, Hu S, Benedict WF. Differential retinoblastoma protein expression in neuroendocrine tumors of the lung. Am J Pathol 1997;150:393– 400. 29. Kamb A, Gruis NA, Weaver-Feldhaus J, Liu Q, Harshman K, Tavtigian SV, et al. A cell cycle regulator potentially involved in genesis of many tumor types. Science 1994;264:436 – 40. 30. Nobori T, Miura K, Wu DJ, Lois A, Takabayashi K, Carson DA. Deletions of the cyclin-dependent kinase-4 inhibitor gene in multiple human cancers. Nature 1994;368:753– 6. 31. Serrano M, Hannon GJ, Beach D. A new regulatory motif in cell-cycle control causing specific inhibition of cyclin D/CDK4. Nature 1993;366:704 –7. 32. Lukas J, Parry D, Aagaard L, Mann DJ, Bartkova J, Strauss M, et al. Retinoblastoma protein-dependent cell-cycle inhibition by the tumor suppressor p16. Nature 1995;375:503– 6. 33. Otterson GA, Kratzke RA, Coxon A, Kim YW, Kaye FJ. Absence of p16INK4 protein is restricted to the subset of lung cancer cell line that retains wild type RB. Oncogene 1994;9: 3375– 8. 34. Shapiro GI, Edwards CD, Kobzik L, Godleski J, Richards W, Sugarbaker DJ, et al. Reciprocal Rb inactivation and p16INK4 expression in primary lung cancers and cell lines. Cancer Res 1995;55:505–9. 35. Kelley MJ, Nakagawa K, Steinberg SM, Mulshine JL, Kamb A, Johnson BE. Differential inactivation of CDKN2 and Rb protein in non-small-cell and small-cell lung cancer cell lines. J Natl Cancer Inst 1995;87:756 – 61. 36. Geradts J, Kratzke RA, Niehans GA, Lincoln CE. Immunohistochemical detection of the cyclin-dependent kinase inhibitor 2/multiple tumor suppressor gene 1 (CDKN2/MTS1) product p16INK4A in archival human solid tumors: correlation with retinoblastoma protein expression. Cancer Res 1995;55:6006 –11. 37. Kinoshita I, Dosaka-Akita H, Mishina T, Akie K, Nishi M, Hiroumi H, et al. Altered p16INK4 and retinoblastoma protein status in non-small cell lung cancer: potential synergistic effect with altered p53 protein on proliferative activity. Cancer Res 1996;56:5557– 62. 38. Benedict WF, Lerner SP, Zhou J, Shen X, Tokunaga H, Czerniak B. Level of retinoblastoma protein expression correlates with p16 (MTS-1/INK4A/CDKN2) status in bladder cancer. Oncogene 1999;18:1197–203. 39. Logothetis CJ, Xu H, Ro JY, Hu S, Sahin A, Ordonez N, et al. Altered expression of retinoblastoma protein and known prognostic variables in locally advanced bladder cancer. J Natl Cancer Inst 1992;84:1256 – 61. 40. Xu H, Hu S, Hashimoto T, Takahashi R, Benedict WF. The retinoblastoma susceptibility gene product: a characteristic pattern in normal cells and abnormal expression in malignant cells. Oncogene 1989;4:807–21. 41. Xu H, Quinlan DC, Davidson AG, Hu S, Summers CL, Li J, et al. Altered retinoblastoma protein expression and prognosis in early-stage non-small-cell lung carcinoma. J Natl Cancer Inst 1994;86:695–9. 42. Mehta CR, Patel NR. A network algorithm for performing Fisher’s exact test in r ⫻ c contingency tables. J Am Stat Assoc 1983;78:427–34. 43. Arrigoni MG, Woolner LB, Bernatz PE. Atypical carcinoid tumors of the lung. J Thorac Cardiovasc Surg 1972;64: 413–21.