[CANCER RESEARCH 61, 5186–5192, July 1, 2001] Genetic Evidence for Early Divergence of Small Functioning and Nonfunctioning Endocrine Pancreatic Tumors: Gain of 9Q34 Is an Early Event in Insulinomas1

Ernst J. M. Speel,2 Alexander F. Scheidweiler, Jianming Zhao, Claudia Matter, Parvin Saremaslani, Ju¨rgen Roth, Philipp U. Heitz, and Paul Komminoth3 Department of Molecular Cell Biology, University of Maastricht, Research Institute Growth and Development, 6200 MD Maastricht, The Netherlands [E. J. M. S.], Department of Pathology [A. F. S., J. Z., C. M., P. S., P. U. H., P. K.], and Division of Cell and Molecular Pathology [J. R., P. K.], University of Zurich, 8091 Zurich, Switzerland

ABSTRACT fields), and/or exhibit Յ2% Ki-67 positive cells. Most insulinomas fall into this category, as do nonfunctioning (micro)adenomas that are The malignant potential among endocrine pancreatic tumors (EPTs) a rather common finding in carefully examined postmortem pancre- varies greatly and can frequently not be predicted using histopathological ases (4). EPTs that are Ͼ2 cm or show angioinvasion, higher numbers parameters. Thus, molecular markers that can predict the biological behavior of EPTs are required. In a previous comparative genomic hy- of mitoses, or percentages of Ki-67 positive cells are at risk of bridization study, we observed marked genetic differences between the malignancy. They include many of the functioning tumors other than various EPT subtypes and a correlation between losses of 3p and 6 and insulinomas. gains of 14q and Xq and metastatic disease. To search for genetic alter- Our understanding of the molecular mechanisms underlying the ations that play a role during early tumor development, we have studied tumorigenesis of EPTs is still scarce. A small percentage of EPTs 38 small (<2 cm) EPTs, including 24 insulinomas and 10 nonfunctioning occur in association with the dominantly inherited MEN 1 and, to a EPTs. Small EPTs are usually classified as clinically benign tumors in the lesser extent, VHL syndromes, the susceptibility genes of which have absence of histological signs of malignancy. Using comparative genomic been cloned and mapped to chromosome bands 11q13 and 3p25.5, hybridization, we identified chromosomal aberrations in 27 EPTs (mean, respectively (5, 6). The vast majority of EPTs, however, occur spo- 4.1). Interestingly, the number of gains differed strongly between non- radically, and only a subset (15–20%) of these tumors harbor somatic functioning and functioning EPTs (3.4 versus 1.5, respectively; and MEN 1 mutations (7). In contrast, VHL mutations appear not to be (30 ؍ as did the number of aberrations in the benign (n ,(0.0526 ؍ P In the relevant in the pathogenesis of sporadic EPTs (8), as are alterations of .(0.0022 ؍ tumors (3 versus 8.4, respectively; P (8 ؍ malignant (n insulinomas, 9q gain (common region of involvement: 9q34) was most the K-ras,N-ras, and TP53 genes (9–11). The involvement of chro- common (50%) and in nonfunctioning EPTs, gain of 4p was most common mosomal regions 9p21 (p16 gene) and 17q12-q21 (c-erbB-2 gene) in (40%). Most frequent losses in insulinomas involved 1p (20.8%), 1q, 4q, EPT tumorigenesis is controversial (10–14). 11q, Xq, and Y (all 16.7%) and in nonfunctioning EPTs, 6q. Losses of 3pq Recently, we have applied CGH to identify chromosomal alter- and 6q and gains of 17pq and 20q proved to be strongly associated with ations that are important in the pathogenesis of EPTs (14). Various malignant behavior in all of the small EPTs (P < 0.0219). Our results significant genetic changes could be identified, including losses of demonstrate marked genetic differences between small functioning and chromosomes 3, 6, 11, X, and Y and gains of chromosomes 5, 7, 9, 14, nonfunctioning EPTs, indicating that these subtypes evolve along different and 17. The total number of genetic changes per EPT appeared to be genetic pathways. In addition, our study endorses the importance of chromosomes 3 and 6q losses to discriminate EPTs with a malignant strongly correlated with both tumor size and disease stage. Moreover, behavior from benign ones. genetic differences could be detected in the various hormonal sub- types, indicating that these groups may evolve along genetically different pathways. The objective of the present study was to charac- INTRODUCTION terize the genetic alterations of EPTs with a diameter Յ2 cm, because

4 these lesions may provide information about the initiating genetic EPTs make up only 1–2% of all of the pancreatic tumors and can events in the pathogenesis of these tumors. We found marked genetic cause syndromes by uncontrolled expression of biologically active differences between small functioning and nonfunctioning EPTs, in- hormones. Such tumors are classified as functioning EPTs, whereas dicating that they evolve along different genetic pathways. In addi- EPTs that do not release any hormone or produce hormones that do tion, by correlating genetic alterations with tumor subtype and malig- not lead to clinical symptoms are called nonfunctioning tumors (1). nant potential, additional evidence was obtained showing that losses Because these clinical parameters do not provide information with of chromosomes 3 and 6q are related to malignancy in EPTs. respect to the biological behavior of EPTs, more indicative his- topathological classification schemes to predict prognosis have been established (2, 3). Accordingly, well-differentiated EPTs are classified MATERIALS AND METHODS Յ as benign if they are 2 cm in size, confined to the pancreas, Tumor Material and Patient Data. Thirty-eight EPTs with a maximum nonangioinvasive (with two or fewer mitoses per 10 high-power diameter of 2 cm (21 females, mean age 55.2 Ϯ 21.3 years and 17 males, mean age 57.1 Ϯ 15.1 years) were selected from the archives of the Departments of Received 12/22/00; accepted 5/2/01. Pathology, Universities of Zurich and Basel, Switzerland and University of The costs of publication of this article were defrayed in part by the payment of page Turin, Turin, Italy (Table 1). The samples included 15 frozen and 23 formalin- charges. This article must therefore be hereby marked advertisement in accordance with fixed, paraffin-embedded EPTs, which were all sporadic tumors and not 18 U.S.C. Section 1734 solely to indicate this fact. 1 Supported by Swiss Cancer League Grant SKL-649-2-1998 and Swiss National associated with the inherited MEN 1 or VHL syndromes. They were classified Science Foundation Grant 31-53625.98. according to the most recent classification (15) and consisted of 10 nonfunc- 2 To whom requests for reprints should be addressed, at Department of Molecular Cell tioning (6 benign and 4 malignant) and 28 functioning EPTs, including 24 Biology, Research Institute Growth and Development (GROW), University of Maastricht, insulinomas (22 benign and 2 malignant), 3 glucagonomas (2 benign and 1 P.O. Box 616, 6200 MD Maastricht, The Netherlands. Phone: 31-43-3881425; Fax: 31-43-3884151; E-mail: [email protected]. malignant), and 1 malignant gastrinoma. Thirty-six patients had localized 3 Present address: Institute for Pathology, Department of Medical Service, Kantons- disease, as defined by the absence of extrapancreatic spread of the tumor, spital Baden, CH-5404 Baden, Switzerland. whereas only 2 patients had advanced disease (patients 23 and 24 in Table 1), 4 The abbreviations used are: EPT, endocrine pancreatic tumor; CGH, comparative with tumor spread into the lymph nodes (one case) or the liver (one case). genomic hybridization; CRI, common region of involvement; TSC, tuberous sclerosis; FISH, fluorescence in situ hybridization; MEN 1, multiple endocrine neoplasia type 1; DNA Extraction. Genomic DNA from frozen samples was isolated by TSG, tumor suppressor gene; VHL, von Hippel Lindau. homogenizing ϳ5mm3 of each sample prior to DNA extraction using the 5186

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS

Table 1 Patient characteristics and genetic findings in small EPTs Number of changes FISH resultsb Size Nra Sex Age Diagnosis Ben/mal (cm) CGH results All Gains Losses 9c–9q34 6c–6q21 3c–4p16 Functioning EPTs 1 Female 57 Ins Ben 1.1 0 0 0 0 2 Female 73 Ins Ben 1 0 0 0 0 Disomy Disomy 3 Female 42 Ins Ben 1.5 0 0 0 0 Disomy Disomy 4 Female 59 Ins Ben 1.2 0 0 0 0 5 Male 62 Ins Ben 1.2 0 0 0 0 6 Female 26 Ins Ben 2 0 0 0 0 7 Female 22 Ins Ben 1.5 0 0 0 0 8 Male 52 Ins Ben 1.6 0 0 0 0 9 Female 72 Ins Ben 1.2 1pqϪ 202 10 Female Ins Ben 1.5 9q34ϩ 1 1 0 3–6 Trisomy 11 Female 80 Ins Ben 1 9q34ϩ 1 1 0 3–3/4–4 (37/37%) Trisomy 12 Male 40 Ins Ben 1 9q33–34ϩ 110 13 Female 75 Ins Ben 1.3 9qϩ 110 14 Female 32 Ins Ben 1.5 XpqϪ 202 15 Male 53 Ins Ben 1.2 5pqϩ, 9q33–34ϩ,Xpϩ 4 4 0 2–3 (51%) Disomy 16 Male 43 Ins Ben 1.2 9q22–34ϩ, 14q22-terϩ 220 17 Male 36 Ins Ben 1.9 16q21-terϪ, 22qϪ,Yϩ 312 18 Male 36 Ins Ben 1 11pqϪ, 13qϪ, 21qϩ 413 19 Male 37 Ins Ben 1.3 1p12–31Ϫ,4qϪ,9qϩ,YϪ 413 20 Male 72 Ins Ben 1.5 1pqϪ, 4pqϪ, 7pqϩ, 9pqϩ, 11pqϪ, 12 5 7 2–1 15qϩ,YϪ 21 Female 42 Ins Ben 1 4p15.2-qterϪ,5qϩ, 7pqϩ, 8pqϪ,9qϩ, 13 5 8 11p13–15.2Ϫ, 11q13-terϪ, 20qϩ, XpqϪ 22 Male 58 Ins Ben 2 1p12–31Ϫ, 1q23-terϪ, 2q22–33Ϫ, 10 3 6 9q31–34ϩ, 11q13-terϪ, 14qϩ, 15qϩ, 18q11.2–22Ϫ, Xq21.3-terϪ, Yp11.2– q11.2Ϫ 23 Female 26 Ins Mal 1.5 4p15.2-qterϪ, 6q12–22Ϫ,9qϩ, 13q21- 10 4 6 terϪ, 14q22-terϩ, 15qϩ, 20qϩ, Xp21-qterϪ 24 Male 56 Ins Mal 1.9 1p13–31Ϫ,1qϪ, 2q23–33Ϫ, 3p14.1- 13 6 6 4–4/4–8 (44/25%) 2–1 qterϪ,6qϪ,9qϩ, 12q23-terϩ, 17pqϩ, Xp11.4-qterϩ,YϪ 25 Female 64 Ga Mal 1.5 8q21.3-terϪ, 9q34ϩ 211 26 Female 67 Gluc Ben 0.5 0 0 0 0 27 Female 82 Glud Ben 11pqϪ, XpqϪ 404 28 Female 75 Glu Mal 1.9 7pqϩ,8pϪ, 17pqϩ, 22q13Ϫ, Xq22– 743 26Ϫ Nonfunctioning EPTs 29 Female 70 NF Ben 1 0 0 0 0 30 Female 51 NF Ben 1.2 0 0 0 0 31 Male 77 NF Ben 0.9 11pqϪ 202 32 Male 85 NF Ben 1.8 11pqϪ,YϪ 303 33 Male 61 NF Mal 1 6qϩ, 7q22-terϩ 220 34 Female 17 NF Mal 2 3pqϪ, 6pqϪ, 9pqϪ, 13qϪ, XpqϪ 9 0 9 1–2 35 Male 62 NF Ben 0.6 4pqϩ, 5p15ϩ,6qϪ, 7q11.2ϩ-q32–36ϩ, 11 10 1 12q24ϩ, 13qϩ, 14q23-terϩ, 17pqϩ, Yϩ 36 Male 70 NF Ben 1.1 4pqϩ, 6q12–23Ϫ, 7pqϩ, 10q21.3-terϩ, 10 8 2 2–4 13q21–32Ϫ, 20qϩ, 21qϩ,Yϩ 37 Female 72 NF Mal 2 1p21–31.1Ϫ, 4p13-q24ϩ, 6q12–22.1Ϫ, 11 8 3 8p21-terϩ, 12q13.3-terϩ, 14q21- terϩ, 18qϩ, 20pqϩ, 21qϪ 38 Male 71 NF Mal 2 1p21–31.1Ϫ, 2q14–35Ϫ, 3pqϪ, 4p15- 13 6 6 2–4 terϩ,6qϪ, 9q32-terϩ, 11p11.2–14.2Ϫ, 11q14-terϪ, 14q24.3- terϩ, 17pqϩ, 20qϩ a Nr, number; Ins, insulinoma; Ga, gastrinoma; Glu, glucagonoma; NF, nonfunctioning; Ben, benign; Mal, malignant; 9c,6c,3c, centromere 9,6,3. b FISH results: Disomy, Trisomy: 2 and 3 copies per nucleus of both probes; 2–1 (n%), 2 copies of the first probe and 1 copy of the second probe in n% of nuclei. c Incidental tumor discovered at autopsy, strongly positive for glucagon by immunohistochemistry. d Incidental tumor discovered at autopsy, strongly positive for glucagon by immunohistochemistry. No information available on exact (small) tumor size.

D-5000 Puregene DNA isolation kit (Gentra Systems, Minneapolis, MN). specimens were treated subsequently with 100% and 70% with cen- Genomic DNA from paraffin-embedded tumor material was isolated from 5 to trifugation steps in between (10 min, 14,000 rpm), after which the tissue 10 10-␮m thick sections as described previously (16, 17). Briefly, unstained material was dried in a speed-vac for 10 min at room temperature. Tissue was tissue sections were microdissected and scraped from the histological glass digested overnight with 1 mg/ml Proteinase K (Sigma Chemical Co., St. Louis, slides after identification of the appropriate tissue area using a stereomicro- MO) in digestion buffer [50 mM Tris (pH 8.0), 1 mM EDTA, and 0.5% SDS] scope (Stemi DV4, Zeiss, Germany) at ϫ10–40. The paraffin material was at 50°C. The DNA was purified by standard phenol-chloroform extraction and collected in a 1.5-ml Eppendorf microcentrifuge tube containing 100% ethanol ethanol precipitation and quantified by spectrophotometry. and centrifuged to the bottom of the tube (5 min, 14,000 rpm). After the CGH and Digital Image Analysis. CGH was performed essentially as ethanol was removed, 1 ml of xylol was added to the tissue material, and the described earlier (14). Briefly, 1 ␮g of tumor DNA was labeled with Spectrum tube was heated for 15 min at 55°C in a thermomixer. After centrifugation for Green-dUTPs (Vysis, Downers Grove, IL) by nick translation (BioNick kit; 10 min at 14,000 rpm, the xylol was discarded, and the tissue material was Life Technologies, Inc., Basel, Switzerland). Spectrum Red-labeled normal treated two additional times with fresh xylol, as described above. Next, the reference DNA (Vysis) was used for cohybridization. The hybridization mix- 5187

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS ture consisted of 200 ng of Spectrum Green-labeled tumor DNA, 200 ng of Spectrum-Red labeled normal reference DNA, and 10–20 ␮g of human Cot-1 DNA (Life Technologies, Inc.) dissolved in 10 ␮l of hybridization buffer [50% formamide and 2ϫ SSC (pH 7.0)]. Hybridization was carried out for 3 days at 37°C to denatured [5 min at 75°C in 70% formamide and 2ϫ SSC (pH 7.0)] normal male human metaphase spreads (Vysis). Slides were washed at 45°C three times for 10 min in 50% formamide/2ϫ SSC followed by two times for 5 min in 2ϫ SSC. The chromosomes were counterstained with 4,6-diamidino- 2-phenylindole for identification. Digital images were collected from six to seven metaphases using a Pho- tometrics cooled CCD camera (Microimager 1400; Xillix Technologies, Van- couver, Canada) attached to a Zeiss Axioskop microscope and a Power G3 Macintosh computer. The software program QUIPS (Vysis) was used to calculate average green:red ratio profiles for each chromosome. At least four observations per autosome and two observations per sex chromosome were included in each analysis. In a subset of cases, the fluorochrome labels for tumor and reference DNA were reversed in the CGH to confirm the reproduc- ibility of the detected chromosomal abnormalities. Positive, negative, and sex-mismatched controls were applied as described by Richter et al. (18). Gains and losses of DNA sequences were defined as chromosomal regions where both the mean green:red fluorescence ratio and its SD were Ͼ1.20 and Ͻ0.80, respectively. Over-representations were consid- ered amplifications when the fluorescence ratio values in a subregion of a chromosomal arm exceeded 1.5. In negative control hybridizations, the mean green:red ratio occasionally exceeded the fixed 1.2 cutoff level at the following chromosomal regions: 1p32-pter, 16p, 19, and 22. Gains of these known G-C-rich regions were therefore excluded from all of the analyses. Contingency table analysis was used to analyze the relationship between genomic alterations and malignancy, and hormonal subtype. Student’s t test and ANOVA were applied to compare the number of genomic alterations between the different EPT hormonal subtypes. Confirmation of CGH Data by FISH. To validate CGH data independ- ently, touch preparations of six EPTs were subjected to FISH using a combi- nation of a chromosome 9-specific centromere probe (pMR9␣) and a cosmid probe (cos-ABL-8) containing the cABL gene at 9q34 (19, 20), as well as a chromosome 6-specific centromere probe (p308; Ref. 21) and a PAC probe (66H14) mapping to the 6q21 region (kindly provided by Dr. P. Sinclair, Royal Fig. 1. Summary of all DNA copy number changes detected by CGH in 28 functioning Free and University College School of Medicine, London, United Kingdom). and 10 nonfunctioning EPTs. Vertical lines on the right of the chromosome ideograms indicate gains of the corresponding chromosomal regions; those on the left indicate losses. Cell processing, probe labeling, in situ hybridization, and detection and eval- Gains on 1p, 16p, 19, and 22 were not analyzed (see text). uation of the observed nuclear FISH signals were performed as described recently (22, 23). In addition, paraffin-embedded tissue sections (5 ␮m) of four EPTs, each showing different alterations for chromosomes 3 and 4 (Table 1), were used for independent FISH analysis with a combination of a chromosome were more frequent in female than in male patients (28.6% versus 3-specific centromere probe and a chromosome band 4p16-specific cosmid 5.8%, respectively), whereas chromosome Y loss was detected in 5 of probe (c5.5; Ref. 24, 25), as described previously (26). 17 (29.4%) males. Increased DNA sequence copy number (Ͼ15% of EPTs) most often involved chromosomes 7q, 9q, and 14q, with the RESULTS highest frequency of gains on chromosome 9q (36.8%; Fig. 1 and Table 3). A comparison between nonfunctioning and functioning CGH Findings in Association with Tumor Subtype and Malig- EPTs revealed marked differences in genetic alterations between these nant Outgrowth. Table 1 and Fig. 1 summarize all of the DNA two groups (Fig. 1 and Table 3). In particular, chromosome 6q losses sequence copy number changes detected by CGH in the 38 EPTs Յ2 (CRI, 6q12–22) and chromosome 4 gains were mainly detected in the cm, and representative CGH data are shown in Fig. 2A. Genetic nonfunctioning EPTs (P Յ 0.0027; Table 3), whereas 9q gains (CRI, aberrations were found in 27 of 38 EPTs (71%), and the average 9q34) were strongly associated with functioning EPTs (P ϭ 0.0404). number of chromosome arm aberrations per tumor was 4.1 Ϯ 4.6 The latter genetic aberration was detected in 12 of 24 (50%) insuli- (range, 0–13). Chromosomal losses (mean, 2.1; range, 0–9) were as nomas (P ϭ 0.0288) and was the sole alteration in 4 of these tumors. frequent as gains (mean, 2.0; range, 0–10), and no evident amplifi- Chromosomal losses of 3pq and 6q (P Յ 0.0005) as well as gains of cations could be detected (Table 1 and 2). The average number of 17pq and 20q (P Յ 0.0219) proved to be significantly correlated with aberrations was approximately two times higher in nonfunctioning malignant outgrowth of EPTs. Our data demonstrate that small EPTs EPTs than in functioning EPTs (6.1 versus 3.4, respectively), and of harbor fewer but similar genetic alterations as have been found in particular note, the difference between the average number of gains in larger EPTs by CGH (14) and indicate that already in small tumors an both groups (3.4 versus 1.5, respectively) almost reached statistical evident genetic divergence can be detected between nonfunctioning significance (P ϭ 0.0526; Table 2). In addition, the mean number of and functioning EPTs as well as between benign and malignant chromosomal changes, gains, and losses all were strongly associated tumors. with malignancy (P Յ 0.0242; Table 2). Confirmation of CGH Data by FISH. FISH analysis on touch Chromosomal regions that were most often lost (Ͼ15% of EPTs) preparations of six EPTs confirmed the CGH results of chromosome included 1p, 6q, 11pq, and Xq, with the highest frequency of losses on arms 9q and 6q (Table 1). Two benign insulinomas without chromo- chromosome 11q (21.1%; Fig. 1 and Table 3). In addition, Xq losses some imbalances (patients 2 and 3) presented two copies of the 5188

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS

Fig. 2. A, representative CGH results in small EPTs. Individual examples of fluorescent ratio profiles (right) and digital images (left) of chromosomes with recurrent gains and losses. Red vertical bar on left side of a chromosome ideogram (middle) indicates the region of loss, and green vertical bar on right side of an ideogram indicates the region of gain. B, examples of FISH analysis of EPT cases 24 (three images on the left; touch preparations) and 38 (right image; paraffin section). First two images on left, imbalances [4–4/4–8 and 2–12 (in some isolated large nuclei), respectively] between 9q34 (green) and centromere 9 (red). Third image from left, deletion of one copy of 6q21 (green) but retention of both centromeres 6(red). Image on right, duplication of the number of 4p16 loci (green) in comparison with the centromere 3 copy number (red). centromere and locus-specific probe per nucleus for both chromosome CGH could be detected in 51% of the nuclei by FISH, whereas no arms in the FISH analysis, indicating that these tumors have a diploid alterations for chromosome 6 were found by both methods. In the DNA content. Of two additional insulinomas showing 9q34 gain as malignant insulinoma of patient 24, FISH again detected different cell the sole alteration by CGH, one tumor (patient 10) displayed three populations with the chromosome 9 probes. Next to a major popula- copies of the centromere 9 probe and six copies of the 9q34 probe per tion with four copies of both centromere 9 and 9q34, 25% of nuclei nucleus in the major cell population, confirming the CGH finding by demonstrated a clear duplication of the cABL target at 9q34. In FISH. Also, a trisomy for both chromosome 6 probes was observed in addition, a number of very large nuclei was observed in between these this EPT, and this tumor appeared to be triploid. The other insulinoma nuclei containing two sometimes very large centromere spots together (patient 11) turned out to harbor two genetically heterogeneous cell with up to twelve 9q34 signals, suggesting that amplification of this populations exhibiting three and four copies of the chromosome 9 target had occurred (Fig. 2B). The frequency of these cells (5%) was, probes, respectively. Because of the blocking of repeated sequences however, much too low to identify this amplification as such by CGH. by Cot-1 DNA, the centromeric region cannot be analyzed by CGH, The loss of 6q found by CGH in this tumor was reflected by the loss which probably explains the detection of only the 9q34 gain. Discrim- of one copy of 6q21, whereas the centromere 6 target was still present ination of the two different cell populations was not possible with the as two copies per nucleus in the major cell population. chromosome 6 probes that demonstrated three copies per nucleus in Paraffin sections of four EPTs were subjected to FISH with a most cells. In the EPT of patient 15, the gain of 9q33–34 identified by combination of a centromere-3 probe and a 4p16-specific probe. As

Table 2 Genomic changes per case in small EPTs Aberrations per tumor

Tumor n All Pa Gains Pa Losses Pa All EPTs 38 4.1 Ϯ 4.6 2.0 Ϯ 2.7 2.1 Ϯ 2.6 Nonfunctioning EPTs 10 6.1 Ϯ 5.1 3.4 Ϯ 4.1 2.6 Ϯ 2.9 All functioning EPTs 28 3.4 Ϯ 4.3 NSb,c 1.5 Ϯ 1.9 0.0526 1.9 Ϯ 2.6 NS Insulinomas 24 3.5 Ϯ 4.5 NSd 1.5 Ϯ 1.9 0.0725 1.9 Ϯ 2.7 NS Benign EPTs 30 3 Ϯ 4 1.5 Ϯ 2.5 1.5 Ϯ 2.2 Malignant EPTs 8 8.4 Ϯ 4.4 0.0022 3.9 Ϯ 2.7 0.0242 4.3 Ϯ 3 0.0069 a ANOVA analysis. b NS ϭ not significant. c P for functioning versus nonfunctioning EPTs. d P for insulinomas versus nonfunctioning EPTs. 5189

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS

Table 3 Frequent genomic changes (%) in small EPTs and association with malignancy All EPTs Nonfuncta Funct Ins Ben Mal Locus n ϭ 38 n ϭ 10 n ϭ 28 Pb n ϭ 24 P n ϭ 30 n ϭ 8 P Losses 1p 18.4 20 17.9 NS 20.8 NS 13.3 37.5 NS 1q 10.5 0 14.3 NS 16.7 NS 10 12.5 NS 3p 7.9 20 3.6 NS 4.2 NS 0 37.5 0.0005 3q 7.9 20 3.6 NS 4.2 NS 0 37.5 0.0005 4q 10.5 0 14.3 NS 16.7 NS 10 12.5 NS 6q 18.4 50 7.1 0.0027 8.3 0.0062 6.7 62.5 0.0003 11p 18.4 30 14.3 NS 12.5 NS 20 12.5 NS 11q 21.1 30 17.9 NS 16.7 NS 23.3 12.5 NS Xq 18.4 10 21.4 NS 16.7 NS 13.3 37.5 NS Y 13.2 10 14.3 NS 16.7 NS 13.3 12.5 NS Gains 4p 10.5 40 0 0.0004 0 0.001 6.7 25 NS 4q 7.9 30 0 0.0025 0 0.005 6.7 12.5 NS 7q 15.8 30 10.7 NS 8.3 NS 13.3 25 NS 9q 36.8 10 46.4 0.0404 50 0.0288 33.3 50 NS 14q 15.8 30 10.7 NS 12.5 NS 10 37.5 NS 17p 10.5 20 7.1 NS 4.2 NS 3.3 37.5 0.0051 17q 10.5 20 7.1 NS 4.2 NS 3.3 37.5 0.0051 20q 13.2 30 7.1 NS 8.3 NS 6.7 37.5 0.0219 a Nonfunct, nonfunctioning EPTs; Funct, functioning EPTs; Ins, insulinomas; Ben, benign; Mal, malignant; NS, not significant. b Contingency table analysis. shown in Table 1 and Fig. 2B, FISH confirmed the chromosomal Four insulinomas exhibited 9q gain as the only detectable aberration, imbalances found for chromosomes 3 and 4p by CGH in all of the four and, thus, one might consider this aberration as an important genetic cases. event early in tumorigenesis. FISH analysis verified the 9q gains in four different tumors using a centromere-9 and 9q34-specific probe, DISCUSSION and moreover, showed the amplification of the cABL target at 9q34 in one malignant insulinoma. Gains and amplification of 9q34 have been With the exception of insulinomas, EPTs Յ2 cm in diameter are reported to occur in several other human neoplasms including squa- seldom, because most EPTs have a larger size at the time of diagnosis, mous cell carcinomas of the head and neck, adrenocortical carcino- present with local invasion or have already developed metastases in mas, pheochromocytomas, and the chronic myeloid leukemia-derived Յ regional lymph nodes or the liver in 50% of patients (1). Neverthe- cell line K562 (23, 27–29). The oncogenic significance of the cABL less, small EPTs can provide us with important information on genetic gene, however, has only been established in hematopoietic cells thus changes underlying early tumor development, because they mainly far, so that additional studies are needed to clarify its possible role in consist of benign tumors, including most insulinomas, which can the development of EPTs. Another candidate gene is the oncogene cause hyperinsulinemic hypoglycemia already at early tumor stages, VAV2 (30) located near the TSC1 gene at 9q34. Strangely enough, as well as nonfunctioning (micro)adenomas that are incidentally de- insulinomas are occasionally detected in TSC patients (31), and these tected at autopsy. It was the goal of this study to identify genetic tumors are rather expected to harbor a mutation in the TSC1 TSG alterations in small EPTs by CGH, to correlate the overall number of together with a loss of the other allele. We did not observe loss of genetic alterations with clinical and histopathological parameters, and heterozygosity using microsatellite marker D9S66 in ten of our spo- to evaluate if the identified chromosomal aberrations are similar to radic insulinoma cases. However, a TSC-related insulinoma may also those described previously in larger EPTs of different hormonal develop via an alternative genetic pathway including inactivation of subtypes (14). CGH detected chromosomal aberrations in 71% of the TSC2 gene at 16p13.3. Gains of chromosome 4p and 4q were EPTs, most commonly involving losses on chromosomes 1p, 6q, exclusively found in the nonfunctioning EPTs (in 40% and 30% of 11pq, Xq, and Y and gains on chromosomes 7q, 9q, and 14q. Hence, cases, respectively), a finding that is in accordance with the results of these chromosomal regions may harbor genes critically playing a role in the pathogenesis of small EPTs. In addition, marked genetic dif- Terris et al. (32) who found among other alterations gains of chro- ferences were observed between the functioning and nonfunctioning mosome 4 in 7 of 11 nonfunctioning EPTs. Nevertheless, this EPTs, and a strong correlation was found between the number of genomic change cannot be considered a potential initiating genetic genomic changes per tumor and malignant behavior. event, because chromosome 4 gains were only identified in neoplasms Ն The observed average number of genetic changes per tumor in this exhibiting 10 chromosome arm aberrations per tumor. This also study was in good agreement with the results described earlier for a accounts for the most frequently detected genetic alteration in non- limited number of EPTs Յ2 cm [Ref. 14; 4.1 versus 3.8, respectively] functioning EPTs, i.e., loss of 6q in 50% of cases. This alteration and proves that genetic instability already begins to emerge in small proved to be significantly more present in nonfunctioning than in EPTs. Interestingly, we noticed a strong tendency toward a difference functioning EPTs (P ϭ 0.0027), but caution should be taken into between the mean number of genetic aberrations in nonfunctioning account by interpreting these data, because three of five neoplasms and functioning tumors being most evident comparing the number of were malignant, and a strong association between 6q losses and gains in both tumor groups (3.4 versus 1.5, respectively; P ϭ 0.0526). malignant outgrowth have been observed by us in this (see below) as This difference has shown to be highly significant with increasing well as in previous studies accounting for EPTs in general, thus tumor size (14). Furthermore, a number of genomic changes could be including functioning EPTs. Thus, an early genetic event in nonfunc- characterized that were strongly correlated to either small functioning tioning EPTs is difficult to assign on the basis of our CGH results. The or nonfunctioning EPTs. One of the most striking findings in this difference in genetic makeup compared with functioning EPTs, how- study was the gain of chromosome 9q in 46.4% of functioning tumors ever, is evident, and the identification of a high number of aberrations and particularly in 50% of insulinomas, with the CRI being 9q34. already in rather small tumors strengthens our hypothesis that one or 5190

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS more genes playing a role in the guarding of genetic stability may be 5. Latif, F., Tory, K., Gnarra, J., Yao, M., Duh, F. M., Orcutt, M. L., Stackhouse, T., inactivated in nonfunctioning EPTs. Kuzmin, I., Modi, W., and Geil, L. Identification of the von Hippel-Lindau disease tumor suppressor gene. Science (Wash. DC), 260: 1317–1320, 1993. Genomic changes that were commonly encountered in both func- 6. Chandrasekharappa, S. C., Guru, S. C., Manickam, P., Olufemi, S-E., Collins, F. S., tioning and nonfunctioning tumors involved losses of 1p, 11pq, Xq, Emmert-Buck, M. R., Debelenko, L. V., Zhuang, Z., Lubensky, I. A., Liotta, L. A., Crabtree, J. S., Wang, Y., Roe, B. A., Weisemann, J., Boguski, M. S., Agarwal, S. K., and Y, as well as gains of 7q and 14q, alterations that have been Kester, M. B., Kim, Y. S., Heppner, C., Dong, Q., Spiegel, A. M., Burns, A. L., and observed previously in EPTs (14, 32). Chromosomal losses and loss Marx, S. Positional cloning of the gene for multiple endocrine neoplasia-Type 1. of heterozygosity at 11q13 are well-known findings in EPTs, because Science (Wash. DC), 276: 404–407, 1997. 7. Go¨rtz, B., Roth, J., Kra¨henmann, A., De Krijger, R. R., Muletta-Feurer, S., Ru¨timann, this locus harbors the MEN 1 gene (6) as well as an additional putative K., Saremaslani, P., Speel, E. J. M., Heitz, P. U., and Komminoth, P. Mutations and TSG more telomeric of the MEN 1 gene (33, 34). The significance of allelic deletions of the MEN 1 gene are associated with a subset of sporadic endocrine copy number decreases of the sex chromosomes (Xq in female and Y pancreatic and neuroendocrine tumors and not restricted to foregut neoplasms. Am. J. Pathol., 154: 429–436, 1999. in male patients) is unknown, and such losses are seen in many 8. Chung, D. C., Smith, A. P., Louis, D. N., Graeme-Cook, F., Warshaw, A. L., and malignancies, as well as in bone marrow cells of healthy elderly Arnold, A. A novel pancreatic endocrine tumor suppressor gene locus on chromo- people (35–37). Nevertheless, in our patient series, a number of EPTs some 3p with clinical prognostic implications. J. Clin. Investig., 100: 404–410, 1997. 9. Yashiro, T., Fulton, N., Hara, H., Yasuda, K., Montag, A., Yashiro, N., Straus, F., Ito, of relatively young patients proved to harbor these sex chromosome K., Aiyoshi, Y., and Kaplan, E. L. Comparison of mutations of ras oncogene in losses, suggesting that chromosome X may contain one or more human pancreatic exocrine and endocrine tumors. Surgery (St. Louis), 114: 758–764, TSGs, most likely at the pseudoautosomal region on Xp or at Xq22– 1993. 10. Evers, B. M., Rady, P. L., Sandoval, K., Arany, I., Tyring, S. K., Sanchez, R. L., 23. Gains of chromosome 7 have been reported to be present in Nealon, W. H., Townsend, C. M., and Thompson, J. C. Gastrinomas demonstrate 43–68% of EPTs by three independent CGH studies (14, 32, 38), and amplification of the HER-2/neu proto-oncogene. Ann. Surg., 219: 596–604, 1994. hepatocyte growth factor, epidermal growth factor receptor, and MET 11. Beghelli, S., Pelosi, G., Zamboni, G., Falconi, M., Iacono, C., Bordi, C., and Scarpa, A. Pancreatic endocrine tumours: evidence for a tumour suppressor pathogenesis and have been suggested as potential genes involved in the tumorigenesis for a tumour suppressor gene on chromosome 17p. J. Pathol., 186: 41–50, 1998. of EPTs. So far, however, none of these genes has been proven to be 12. Muscarella, P., Melvin, W. S., Fisher, W. E., Foor, J., Ellison, E. C., Herman, J. G., relevant. Schirmer, W. J., Hitchcock, C. L., DeYoung, B. R., and Weghorst, C. M. Genetic alterations in gastrinomas and nonfunctioning pancreatic neuroendocrine tumors: an Losses of 1p and gains of 14q have been associated with metastatic analysis of p16/MTS1 tumor suppressor gene inactivation. Cancer Res., 58: 237–240, disease in EPTs (14, 39), and these changes were, indeed, more 1998. 13. Chung, D. C., Brown, S. B., Graeme-Cook, F., Tillotson, L. G., Warshaw, A. L., frequently seen in the malignant than in the benign tumors in this Jensen, R. T., and Arnold, A. Localization of putative tumor suppressor loci by study. A statistically significant correlation, however, was observed genome-wide allelotyping in human pancreatic endocrine tumors. Cancer Res., 58: between the average number of genomic changes per tumor 3706–3711, 1998. ϭ 14. Speel, E. J. M., Richter, J., Moch, H., Egenter, C., Saremaslani, P., Ru¨timann, K., (P 0.0022), in particular losses of 3pq and 6q and gains of 17pq and Zhao, J., Barghorn, A., Roth, J., Heitz, P. U., and Komminoth, P. Genetic differences 20q (P Յ 0.0219), and malignant outgrowth. This implicates that in endocrine pancreatic tumor subtypes detected by comparative genomic hybridiza- these chromosomal regions may contain genes playing a role in tion. Am. J. Pathol., 155: 1787–1794, 1999. 15. Solcia, E., Klo¨ppel, G., and Sobin, L. H. Histological typing of endocrine tumors. In: malignant outgrowth, and we and others have previously found allelic WHO International Histological Classification of Tumors. Ed. 2. Berlin: Springer- losses, as well as genomic deletions at chromosomes 3 and 6q to be Verlag, 2000. associated with clinically malignant EPTs (8, 13, 14, 40–42). The 16. Perren, A., Roth, J., Muletta-Feurer, S., Saremaslani, P., Speel, E. J. M., Heitz, P. U., and Komminoth, P. Clonal analysis of sporadic pancreatic endocrine tumours. CRIs include 3p25, 3q22–26, and 6q21–22, but the respective genes J. Pathol., 186: 363–371, 1998. of interest still need to be identified. 17. Volante, M., Papotti, M., Roth, J., Saremaslani, P., Speel, E. J. M., Lloyd, R. V., In summary, our data show that EPTs Յ2 cm harbor chromosomal Carney, J. A., Heitz, P. U., Bussolati, G., and Komminoth, P. Mixed medullary- follicular thyroid carcinoma: molecular evidence for a dual origin of tumor compo- imbalances similar to larger tumors but at a lower frequency. Further- nents. Am. J. Pathol., 155: 1499–1509, 1999. more, small functioning EPTs already clearly differ in their pattern of 18. Richter, J., Jiang, F., Go¨ro¨g, J-P., Sartorius, G., Egenter, C., Gasser, T. C., Moch, H., genetic aberrations from small nonfunctioning tumors, indicating that Mihatsch, M. J., and Sauter, G. Marked genetic differences between stage pTa and stage pT1 papillary bladder cancer detected by comparative genomic hybridization. these EPT subgroups follow different genetic routes during tumori- Cancer Res., 57: 2860–2864, 1997. genesis. In particular, a copy number increase of chromosome 9q34 19. Heisterkamp, N., Groffen, J., and Stephenson, J. R. The human v-abl cellular homologue. J. Mol. Appl. Genet., 2: 57–68, 1983. represents a hitherto unrecognized but probably important early ge- 20. Rocchi, M., Archidiacono, N., Ward, D. C., and Baldini, A. A human chromosome netic event in insulinomas. In addition, our study provides additional 9-specific alphoid DNA repeat spatially resolvable from satellite 3 DNA by fluores- evidence for the association of chromosomes 3 and 6q losses and cent in situ hybridization. Genomics, 9: 517–523, 1991. 21. Jabs, E. W., Wolf, S. F., and Migeon, B. R. Characterization of a cloned DNA malignant outgrowth of EPTs. sequence that is present at centromeres of all human autosomes and the X chromo- some and shows polymorphic variation. Proc. Natl. Acad. Sci. USA, 81: 4884–4888, 1984. ACKNOWLEDGMENTS 22. Go¨rtz, B., Roth, J., Speel, E. J. M., Kra¨henmann, A., De Krijger, R. R., Matias-Guiu, X., Muletta-Feurer, S., Ru¨timann, K., Saremaslani, P., Heitz, P. U., and Komminoth, We thank M. Mihatsch, G. Sauter, M. Kasper (University of Basel, Basel, P. MEN 1 gene mutation analysis of sporadic adrenocortical lesions. Int. J. Cancer, Switzerland), and M. Volante (University of Turin, Turin, Italy) for providing 80: 373–379, 1999. 23. Dannenberg, H., Speel, E. J. M., Zhao, J., Saremaslani, P., Van der Harst, E., Roth, tissue samples and helpful discussions; P. Sinclair (Royal Free and University J., Heitz, P. U., Bonjer, H. J., Dinjens, W. N. M., Mooi, W. J., Komminoth, P., and College of Medicine, London, United Kingdom) for providing the 6q21 PAC De Krijger, R. R. Losses of chromosomes 1p and 3q are early genetic events in the probe; and N. Wey and I. Schneider for photographic reproductions. development of sporadic pheochromocytomas. Am. J. Pathol., 157: 353–359, 2000. 24. Landegent, J. E., Jansen in de Wal, N., Fisser-Groen, Y. M., Bakker, E., Van der Ploeg, M., and Pearson, P. L. Fine mapping of the Huntington disease linked D4S10 REFERENCES locus by non-radioactive in situ hybridization. Hum. Genet., 73: 354–357, 1986. 25. Waye, J. S., and Willard, H. F. Chromosome specificity of satellite : short- and 1. Klo¨ppel, G., In ’t Veld, P. A., Komminoth, P., and Heitz, P. U. The endocrine long-range organization of a diverged dimeric subset of human ␣ satellite from pancreas. In: K. Kovacs and S. L. Asa (eds.), Functional Endocrine Pathology, pp. chromosome 3. Chromosoma (Berl.), 97: 475–480, 1989. 415–487. Oxford, United Kingdom: Blackwell Scientific Publications, 1998. 26. Hopman, A. H. N., and Ramaekers, F. C. S. Processing and staining of cell and tissue 2. La Rosa, S., Sessa, F., Capelle, C., Riva, C., Leone, B. E., Klersy, C., Rindi, G., and material for interphase cytogenetics. In: J. P. Robinson (ed.), Current Protocols in Solcia, E. Prognostic criteria in non-functioning pancreatic endocrine tumours. Vir- Cytometry, pp. 8.5.1–8.5.22. New York: John Wiley & Sons, 1998. chows Arch., 429: 323–333, 1996. 27. Bockmuhl, U., Schwendel, A., Dietel, M., and Petersen, I. Distinct patterns of 3. Klo¨ppel, G. Classification of neuroendocrine tumors. Verh. Dtsch. Ges. Pathol., 81: chromosomal alterations in high- and low-grade head and neck squamous cell 111–117, 1997. carcinoma. Cancer Res., 56: 5325–5329, 1996. 4. Kimura, W., Kuroda, A., and Morioka, Y. Clinical pathology of endocrine tumors of 28. Zhao, J., Speel, E. J. M., Muletta-Feurer, S., Ru¨timann, K., Saremaslani, P., Roth, J., the pancreas. Analysis of autopsy cases. Dig. Dis. Sci., 36: 933–942, 1991. Heitz, P. U., and Komminoth, P. Analysis of genomic alterations in sporadic adre- 5191

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. GENETIC ALTERATIONS IN SMALL EPTS

nocortical lesions. Gain of chromosome 17 is an early event in adrenocortical 36. Knuutila, S., Bjo¨rkqvist, A-M., Autio, K., Tarkkanen, M., Wolf, M., Monni, O., tumorigenesis. Am. J. Pathol., 155: 1039–1045, 1999. Szymanska, J., Larramendy, M. L., Tapper, J., Pere, H., El-Rifai, W., Hemmer, S., 29. Collins, S. J., and Groudine, M. T. Rearrangement and amplification of c-abl se- Wasenius, V-M., Vidgren, V., and Zhu, Y. DNA copy number amplifications in quences in the human chronic myelogenous leukemia cell line K-562. Proc. Natl. human neoplasms. Am. J. Pathol., 152: 1107–1123, 1998. Acad. Sci. USA, 80: 4813–4817, 1983. 37. Guttenbach, M., Koschorz, B., Bernthaler, U., Grimm, T., and Schmid, M. Sex 30. Henske, E. P., Short, M. P., Jozwiak, S., Bovey, C. M., Ramlakhan, S., Haines, J. L., chromosome loss and aging: in situ hybridization studies on human interphase nuclei. and Kwiatkowski, D. J. Identification of VAV2 on 9q34 and its exclusion as the Am. J. Hum. Genet., 57: 1143–1150, 1995. tuberous sclerosis gene TSC1. Ann. Hum. Genet., 59: 25–37, 1995. 38. Stumpf, E., Aalto, Y., Ho¨o¨g, A., Kjellman, M., Otonkoski, T., Knuutila, S., and 31. Davoren, P. M., and Epstein, M. T. Insulinoma complicating tuberous sclerosis. Andersson, L. C. Chromosomal alterations in human pancreatic endocrine tumors. J. Neurol. Neurosurg. Psychiatry, 55: 1209–1212, 1992. Genes Chromosomes Cancer, 29: 83–87, 2000. 32. Terris, B., Meddeb, M., Marchio, A., Danglot, G., Fle´jou, J-F., Belghiti, J., 39. Ebrahimi, S. A., Wang, E. H., Wu, A., Schreck, R. R., Passaro, E., Jr., and Sawicki, Ruszniewski, P., and Bernheim, A. Comparative genomic hybridization analysis of M. P. Deletion of chromosome 1 predicts prognosis in pancreatic endocrine tumors. sporadic neuroendocrine tumors of the digestive system. Genes Chromosomes Can- Cancer Res., 59: 311–315, 1999. cer, 22: 50–56, 1998. 40. Hessman, O., Lindberg, D., Einarsson, A., Lillhager, P., Carling, T., Grimelius, L., 33. Chakrabarti, R., Srivatsan, E. S., Wood, T. F., Eubanks, P. J., Ebrahimi, S. A., Gatti, R. A., Passaro, E. Jr., and Sawicki, M. P. Deletion mapping of endocrine tumors Eriksson, B., Åkerstro¨m, G., Westin, G., and Skogseid, B. Genetic alterations on 3p, localizes a second tumor suppressor gene on chromosome band 11q13. Genes 11q13, and 18q in non-familial and MEN 1-associated pancreatic endocrine tumors. Chromosomes Cancer, 22: 130–137, 1998. Genes Chromosomes Cancer, 26: 258–264, 1999. 34. Speel, E. J. M., Meier, D., Matter, C., Muletta-Feurer, S., Van Asseldonk, M., Roth, 41. Barghorn, A., Komminoth, P., Bachmann, D., Ru¨timann, K., Saremaslani, P., J., Heitz, P. U., and Komminoth, P. Chromosome 11q losses in sporadic endocrine Muletta-Feurer, S., Roth, J., Heitz, P. U., and Speel, E. J. M. Chromosome 3p pancreatic tumors: another tumor suppressor gene telomeric of MEN1? Virchows alterations in endocrine pancreatic tumors: a common region of allelic loss at Arch., 435: 219, 1999. 3p23–p25 and association of 3p loss with advanced disease. J. Pathol., in press, 2001. 35. Jiang, F., Richter, J., Schraml, P., Bubendorf, L., Gasser, T., Sauter, G., Mihatsch, 42. Barghorn, A., Speel, E. J. M., Farspour, B., Saremaslani, P., Schmid, S., Perren, A., M. J., and Moch, H. Chromosomal imbalances in papillary renal cell carcinoma. Roth, J., Heitz, P. U., and Komminoth, P. Putative tumor suppressor loci at 6q22 and Genetic differences between histological subtypes. Am. J. Pathol., 153: 1467–1473, 6q23–q24 are involved in the malignant progression of sporadic endocrine pancreatic 1998. tumors. Am. J. Pathol., in press, 2001.

5192

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research. Genetic Evidence for Early Divergence of Small Functioning and Nonfunctioning Endocrine Pancreatic Tumors: Gain of 9Q34 Is an Early Event in Insulinomas

Ernst J. M. Speel, Alexander F. Scheidweiler, Jianming Zhao, et al.

Cancer Res 2001;61:5186-5192.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/61/13/5186

Cited articles This article cites 37 articles, 9 of which you can access for free at: http://cancerres.aacrjournals.org/content/61/13/5186.full#ref-list-1

Citing articles This article has been cited by 12 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/61/13/5186.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://cancerres.aacrjournals.org/content/61/13/5186. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from cancerres.aacrjournals.org on September 27, 2021. © 2001 American Association for Cancer Research.