[Skip to Content]
Access to paid content on this site is currently suspended due to excessive activity being detected from your IP address 54.159.189.139. Please contact the publisher to request reinstatement.
[Skip to Content Landing]
Download PDF
Figure 1.
Fine mapping of loss of gene architecture at the cyclin-dependent kinase 2A (CDKN2A) (p14ARF, p16INK4) and methylthioadenosine phosphorylase (MTAP) loci in squamous cell carcinomas of the head and neck. A, Genomic organization of the CDKN2B, CDKN2A, and MTAP genes at 9p21 illustrating 3 multiplex ligation-dependent probe amplification assay probes for CDKN2B, 7 for CDKN2A, and 3 for MTAP. B, Loss of gene segments is illustrated by dashed lines and retention by solid lines. The UMSCC-11B cell line retained all 9p21 loci tested. Note the homozygous loss of CDKN2B, CDKN2A, and the 3′ end of MTAP in UMSCC-11A. The UMSCC-81A cell line had homozygous loss of CDKN2B and CDKN2A but retained MTAP. The UMSCC-81B cell line indicated loss in only exon 1α of p16INK4a.

Fine mapping of loss of gene architecture at the cyclin-dependent kinase 2A (CDKN2A) (p14ARF, p16INK4) and methylthioadenosine phosphorylase (MTAP) loci in squamous cell carcinomas of the head and neck. A, Genomic organization of the CDKN2B, CDKN2A, and MTAP genes at 9p21 illustrating 3 multiplex ligation-dependent probe amplification assay probes for CDKN2B, 7 for CDKN2A, and 3 for MTAP. B, Loss of gene segments is illustrated by dashed lines and retention by solid lines. The UMSCC-11B cell line retained all 9p21 loci tested. Note the homozygous loss of CDKN2B, CDKN2A, and the 3′ end of MTAP in UMSCC-11A. The UMSCC-81A cell line had homozygous loss of CDKN2B and CDKN2A but retained MTAP. The UMSCC-81B cell line indicated loss in only exon 1α of p16INK4a.

Figure 2.
Fine-mapping multiplex ligation-dependent probe amplification assay illustrated for cell lines UMSCC-81A/B. A, Note 38 individual peaks ranging from 130 to 472 base pairs in the male control sample (Table). The 3 cyclin-dependent kinase 2B (CDKN2B) probes are depicted in orange (probes 1, 2, and 3 in Figure 1); the 2 CDKN2A p14 probes are in blue (probes 4 and 5 in Figure 1); the 2 CDKN2A probes between p14 and p16 are in purple (probes 6 and 7 in Figure 1); the 3 CDKN2A p16 probes are in green (probes 8, 9, and 10 in Figure 1); and the 3 methylthioadenosine phosphorylase (MTAP) probes are in red (probes 11, 12, and 13 in Figure 1). B, Note the complete loss of the CDKN2B/A genes (a total of 10 probes) in UMSCC-81A compared with the normal male genomic DNA. Note the retention of all 3 MTAP probes. C, The UMSCC-81B cell line indicated retention of all 9p21 loci except for exon 1α (p16INK4a), making the latter the smallest region of overlap for loss in the CDKN2AARF, INK4a region. Note the retention of all 3 MTAP probes. Asterisk indicates the minimal region of loss (CDKN2A exon 1α).

Fine-mapping multiplex ligation-dependent probe amplification assay illustrated for cell lines UMSCC-81A/B. A, Note 38 individual peaks ranging from 130 to 472 base pairs in the male control sample (Table). The 3 cyclin-dependent kinase 2B (CDKN2B) probes are depicted in orange (probes 1, 2, and 3 in Figure 1); the 2 CDKN2A p14 probes are in blue (probes 4 and 5 in Figure 1); the 2 CDKN2A probes between p14 and p16 are in purple (probes 6 and 7 in Figure 1); the 3 CDKN2A p16 probes are in green (probes 8, 9, and 10 in Figure 1); and the 3 methylthioadenosine phosphorylase (MTAP) probes are in red (probes 11, 12, and 13 in Figure 1). B, Note the complete loss of the CDKN2B/A genes (a total of 10 probes) in UMSCC-81A compared with the normal male genomic DNA. Note the retention of all 3 MTAP probes. C, The UMSCC-81B cell line indicated retention of all 9p21 loci except for exon 1α (p16INK4a), making the latter the smallest region of overlap for loss in the CDKN2AARF, INK4a region. Note the retention of all 3 MTAP probes. Asterisk indicates the minimal region of loss (CDKN2A exon 1α).

Figure 3.
DNA amplification by real-time polymerase chain reaction for UMSCC-11A and UMSCC-11B with β-globin (control) and cyclin-dependent kinase 2A (CDKN2A) primers run in separate reactions but in the same run. Note that the β-globin control gene is equally amplified in UMSCC-11A and UMSCC-11B (short arrows). Note that the CKDN2A gene is also amplified to the same extent in UMSCC-11B (long arrows). In UMSCC-11A, homozygous loss of CDKN2A was observed, as indicated by a delayed threshold cycle.

DNA amplification by real-time polymerase chain reaction for UMSCC-11A and UMSCC-11B with β-globin (control) and cyclin-dependent kinase 2A (CDKN2A) primers run in separate reactions but in the same run. Note that the β-globin control gene is equally amplified in UMSCC-11A and UMSCC-11B (short arrows). Note that the CKDN2A gene is also amplified to the same extent in UMSCC-11B (long arrows). In UMSCC-11A, homozygous loss of CDKN2A was observed, as indicated by a delayed threshold cycle.

Table. 
9p21 Fine-Mapping Multiplex Ligation-Dependent Probe Amplification Gene Probe Panel
9p21 Fine-Mapping Multiplex Ligation-Dependent Probe Amplification Gene Probe Panel
1.
Van Dyke  DLWorsham  MJBenninger  MS  et al.  Recurrent cytogenetic abnormalities in squamous cell carcinomas of the head and neck. Genes Chromosomes Cancer 1994;9192- 206
PubMedArticle
2.
Danahey  DGTobin  EJSchuller  DEBier-Laning  CMWeghorst  CMLang  JC p16 Mutation frequency and clinical correlation in head and neck cancer. Acta Otolaryngol 1999;119285- 288
PubMedArticle
3.
Lydiatt  WMDavidson  BJSchantz  SPCaruana  SCChaganti  RS 9p21 Deletion correlates with recurrence in head and neck cancer. Head Neck 1998;20113- 118
PubMedArticle
4.
Sherr  CJ Cancer cell cycles. Science 1996;2741672- 1677
PubMedArticle
5.
Johnson  DGWalker  CL Cyclins and cell cycle checkpoints. Annu Rev Pharmacol Toxicol 1999;39295- 312
PubMedArticle
6.
Haber  DA Splicing into senescence: the curious case of p16 and p19ARF. Cell 1997;91555- 558
PubMedArticle
7.
Xing  EPNie  YSong  Y  et al.  Mechanisms of inactivation of p14ARF, p15INK4b and p16INK4a genes in human esophageal squamous cell carcinoma. Clin Cancer Res 1999;52704- 2713
PubMed
8.
Sherr  CJ The Pezcoller Lecture: cancer cell cycles revisited. Cancer Res 2000;603689- 3695
PubMed
9.
Hannon  GJBeach  D p15INK4B Is a potential effector of TGF-β–induced cell cycle arrest. Nature 1994;371257- 261
PubMedArticle
10.
Shintani  SNakahara  YMihara  MUeyama  YMatsumura  T Inactivation of the p14(ARF), p15(INK4B) and p16(INK4A) genes is a frequent event in human oral squamous cell carcinomas. Oral Oncol 2001;37498- 504
PubMedArticle
11.
Rocco  JWSidransky  D p16 (MTS-1/CDKN2/INK4a) in cancer progression. Exp Cell Res 2001;26442- 55
PubMedArticle
12.
Borg  ASandberg  TNilsson  K  et al.  High frequency of multiple melanomas and breast and pancreas carcinomas in CDKN2A mutation-positive melanoma families. J Natl Cancer Inst 2000;921260- 1266
PubMedArticle
13.
Rocco  JWLi  DLiggett  WH  et al.  p16INK4A adenovirus-mediated gene therapy for human head and neck squamous cell cancer. Clin Cancer Res 1998;41697- 1704
PubMed
14.
Christopher  SADiegelman  PPorter  CWKruger  WD Methylthioadenosine phosphorylase, a gene frequently codeleted with p16(cdkN2a/ARF), acts as a tumor suppressor in a breast cancer cell line. Cancer Res 2002;626639- 6644
PubMed
15.
Harasawa  HYamada  YKudoh  M  et al.  Chemotherapy targeting methylthioadenosine phosphorylase (MTAP) deficiency in adult T cell leukemia (ATL). Leukemia 2002;161799- 1807
PubMedArticle
16.
Efferth  TMiyachi  HDrexler  HGGebhart  E Methylthioadenosine phosphorylase as target for chemoselective treatment of T-cell acute lymphoblastic leukemic cells. Blood Cells Mol Dis 2002;2847- 56
PubMedArticle
17.
Chen  YJLin  SCKao  T  et al.  Genome-wide profiling of oral squamous cell carcinoma. J Pathol 2004;204326- 332
PubMedArticle
18.
Nobori  TTakabayashi  KTran  P  et al.  Genomic cloning of methylthioadenosine phosphorylase: a purine metabolic enzyme deficient in multiple different cancers. Proc Natl Acad Sci U S A 1996;936203- 6208
PubMedArticle
19.
Carey  TEVan Dyke  DLWorsham  MJ  et al.  Characterization of human laryngeal primary and metastatic squamous cell carcinoma cell lines UM-SCC-17A and UM-SCC-17B. Cancer Res 1989;496098- 6107
PubMed
20.
Bradford  CRKimmel  KAVan Dyke  DL  et al.  11p deletions and breakpoints in squamous cell carcinoma: association with altered reactivity with the UM-E7 antibody. Genes Chromosomes Cancer 1991;3272- 282
PubMedArticle
21.
Worsham  MJPals  GSchouten  J  et al.  Delineating genetic pathways of disease progression in head and neck squamous cell carcinoma. Arch Otolaryngol Head Neck Surg 2003;129702- 708
PubMedArticle
22.
Schouten  JPMcElgunn  CJWaaijer  RZwijnenburg  DDiepvens  FPals  G Relative quantification of 40 nucleic acid sequences by multiplex ligation-dependent probe amplification. Nucleic Acids Res 2002;30e57
PubMedArticle
23.
Liu  TJZhang  WWTaylor  DL  et al.  Growth suppression of tumor cells of human head and neck cancer cells by the introduction of wild-type p53 gene via a recombinant adenovirus. Cancer Res 1994;543662- 3667
PubMed
24.
Sumantran  VNEalovega  MWNunez  G  et al.  Overexpression of Bcl-XS sensitizes MCF-7 cells to chemotherapy-induced apoptosis. Cancer Res 1995;552507- 2510
PubMed
25.
Hoffmann  THafner  DBallo  H  et al.  Antitumor activity of anti-epidermal growth factor receptor monoclonal antibodies and cisplatin in ten human head and neck squamous cell carcinoma lines. Anticancer Res 1997;174419- 4425
PubMed
26.
Shin  DMDonato  NJPerez-Soler  R  et al.  Epidermal growth factor receptor–targeted therapy with C225 and cisplatin in patients with head and neck cancer. Clin Cancer Res 2001;71204- 1213
PubMed
27.
Bonner  JARaisch  KPTrummell  HQ  et al.  Enhanced apoptosis with combination C225/radiation treatment serves as the impetus for clinical investigation in head and neck cancers. J Clin Oncol 2000;1847S- 53S
PubMed
28.
Toohey  JI Methylthioadenosine nucleoside phosphorylase deficiency in methylthio-dependent cancer cells. Biochem Biophys Res Commun 1978;8327- 35
PubMedArticle
29.
Trackman  PCAbeles  RH Methionine synthesis from 5′-S-methylthioadenosine: resolution of enzyme activities and identification of 1-phospho-5-S methylthioribulose. J Biol Chem 1983;2586717- 6720
PubMed
30.
Kamatani  NNelson-Rees  WACarson  DA Selective killing of human malignant cell lines deficient in methylthioadenosine phosphorylase, a purine metabolic enzyme. Proc Natl Acad Sci U S A 1981;781219- 1223
PubMedArticle
31.
Olopade  OIPomykala  HMHagos  F  et al.  Construction of a 2.8-megabase yeast artificial chromosome contig and cloning of the human methylthioadenosine phosphorylase gene from the tumor suppressor region on 9p21. Proc Natl Acad Sci U S A 1995;926489- 6493
PubMedArticle
32.
Schmid  MMalicki  DNobori  T  et al.  Homozygous deletions of methylthioadenosine phosphorylase (MTAP) are more frequent than p16INK4A (CDKN2) homozygous deletions in primary non-small cell lung cancers (NSCLC). Oncogene 1998;172669- 2675
PubMedArticle
33.
Della Ragione  FRusso  GOliva  A  et al.  5′-Deoxy-5′-methylthioadenosine phosphorylase and p16INK4 deficiency in multiple tumor cell lines. Oncogene 1995;10827- 833
PubMed
34.
Stadler  WMSherman  JBohlander  SK  et al.  Homozygous deletions within chromosomal bands 9p21–22 in bladder cancer. Cancer Res 1994;542060- 2063
PubMed
35.
Garcia-Castellano  JMVillanueva  AHealey  JH  et al.  Methylthioadenosine phosphorylase gene deletions are common in osteosarcoma. Clin Cancer Res 2002;8782- 787
PubMed
36.
Cairns  PPolascik  TJEby  Y  et al.  Frequency of homozygous deletion at p16/CDKN2 in primary human tumours. Nat Genet 1995;11210- 212
PubMedArticle
37.
Brat  DJJames  CDJedlicka  AE  et al.  Molecular genetic alterations in radiation-induced astrocytomas. Am J Pathol 1999;1541431- 1438
PubMedArticle
38.
Chen  ZHOlopade  OISavarese  TM Expression of methylthioadenosine phosphorylase cDNA in p16−, MTAP− malignant cells: restoration of methylthioadenosine phosphorylase-dependent salvage pathways and alterations of sensitivity to inhibitors of purine de novo synthesis. Mol Pharmacol 1997;52903- 911
PubMedArticle
39.
Myers  RWAbeles  RH Conversion of 5-S-methyl-5-thio-D-ribose to methionine in Klebsiella pneumoniae: stable isotope incorporation studies of the terminal enzymatic reactions in the pathway. J Biol Chem 1990;26516913- 16921
PubMed
40.
O’Shaughnessy  JADemers  LMJones  SE  et al.  α-Difluoromethylornithine as treatment for metastatic breast cancer patients. Clin Cancer Res 1999;53438- 3444
PubMed
41.
Levin  VAUhm  JHJaeckle  KA  et al.  Phase III randomized study of postradiotherapy chemotherapy with α-difluoromethylornithine-procarbazine: N-(2-chloroethyl)-N'-cyclohexyl-N-nitrosurea, vincristine (DFMO-PCV) versus PCV for glioblastoma multiforme. Clin Cancer Res 2000;63878- 3884
PubMed
Original Article
April 2006

Fine-Mapping Loss of Gene Architecture at the CDKN2B (p15INK4b), CDKN2A (p14ARF, p16INK4a), and MTAP Genes in Head and Neck Squamous Cell Carcinoma

Author Affiliations

Author Affiliations: Department of Otolaryngology–Head and Neck Surgery and Research Division, Henry Ford Hospital, Detroit, Mich (Drs Worsham, Chen, Sethi, and Benninger and Ms Tiwari); and Department of Clinical Genetics, VU Medical Center (Dr Pals), and MRC-Holland (Dr Schouten), Amsterdam, the Netherlands.

Arch Otolaryngol Head Neck Surg. 2006;132(4):409-415. doi:10.1001/archotol.132.4.409
Abstract

Objective  To identify the extent and the smallest region of loss for CDKN2BINK4b, CDKN2AARF,INK4a, and MTAP. Homozygous deletions of human chromosome 9p21 occur frequently in malignant cell lines and are common in squamous cell carcinoma of the head and neck (HNSCC). This complex region encodes the tumor suppressor genes cyclin-dependent kinase 2B (CDKN2B) (p15INK4b) and CDKN2A (p14ARF, p16INK4a) and the housekeeping gene methylthioadenosine phosphorylase (MTAP).

Design  A targeted probe panel designed to finely map the region of 9p21 loss comprised 3 probes for CDKN2BINK4b, 7 for CDKN2AARF, INK4a, and 3 for MTAP and was interrogated using the multiplex ligation-dependent probe amplification assay (MLPA). The MLPA genomic copy number alterations for CDKN2A were validated using real-time polymerase chain reaction.

Subjects  Six HNSCC primary (A) and recurrent or metastatic (B) cell lines were examined: UMSCC-11A/11B, UMSCC-17A/17B, and UMSCC-81A/81B.

Results  Cell line UMSCC-11B retained all 9p loci tested in the region. Cell lines UMSCC-17A/B indicated homozygous deletion of CDKN2AARF, INK4a starting at p16INK4 exon 1α to include exons 2 and 3. Homozygous loss was indicated for CDKN2BINK4b and CDKN2AARF,INK4a in UMSCC-11A, and UMSCC-81A. Cell line UMSCC-81B indicated retention of all 9p loci except for exon 1α (p16INK4a). Selective loss of the 3′ end of MTAP was observed in UMSCC-11A. Genomic alterations by fine-mapping MLPA were validated at the DNA level for CDKN2A.

Conclusions  We identified exon 1α (p16INK4a) as the smallest region of loss in the CDKN2AARF, INK4a gene. The frequency and precise loss of CDKN2BINK4b, CDKN2AARF, INK4a, and MTAP in the prognosis of 9p21-deleted HNSCC may provide impetus for use of these targets as therapeutic biomarkers in head and neck cancer.

The most common chromosome rearrangements associated with squamous cell carcinomas of the head and neck (HNSCCs) are unbalanced translocations leading to chromosomal deletions.1 Chromosome 9p21 has been reported to be a critical region of loss in various cancers. Genetic alterations at the 9p21 locus have been linked to malignant progression in HNSCC.2,3 The cyclin-dependent kinase 2A (CDKN2A) and CDKN2B genes map to 9p21 and are in tandem, spanning a region of approximately 80 kilobases (kb), with CDKN2B located 25 kb centromeric to CDKN2A4,5 (Figure 1). The CDKN2A locus controls the Rb pathway (which regulates G1/S-phase transition) and the p53 pathway (which induces growth arrest or apoptosis in response to either DNA damage or inappropriate mitogenic stimuli by generating 2 gene products).6,7 This regulatory function is achieved through the p16 protein product, which functions upstream of Rb, and the p14 protein, which blocks MDM2 inhibition of p53 activity, preventing p53 degradation, and thus permitting p53-induced apoptosis or growth arrest.8 The CDKN2D/p14 gene has a unique first exon 1β that, under the control of its own promoter, splices onto exon 2 of CDKN2A/p16 in an alternative reading frame, allowing production of the 2 totally unrelated proteins, p14 and p16, respectively (Figure 1).8

Whereas CDKN2A/p16 mutations selectively inactivate the Rb pathway, deletion of the CDKN2A locus impairs the Rb and p53 pathways.68 Deletion of the CDKN2A locus also frequently affects the CDKN2B locus, which encodes p15, an important mediator of the antiproliferative effect of transforming growth factor β.9 Inactivation of the CDKN2B (p15), CDKN2A (p14), and CDKN2A (p16) genes is a frequent event in human oral SCCs.10 The main modes of p16INK4a inactivation in HNSCC are known to include homozygous deletions (40%-60%), mutations (15%-20%), and gene hypermethylation events (5%).10,11 Somatic alterations in the CDKN2A gene occur in many cancer types, and germline mutation carriers are predisposed to a high risk of pancreatic and breast cancers.12

Although inactivation of p16 has been reported as the most common genetic alteration in HNSCC, making it an ideal target for gene replacement,13 the extent to which p14 is independently altered, or the frequency with which p14 and p16 are affected, has not been examined thoroughly. Because these proteins are completely unrelated, bearing no amino acid homology to one another, each of these targets has the potential to independently serve as therapeutic interventions in HNSCC.

The human methylthioadenosine phosphorylase (MTAP) gene at 9p21, approximately 100-kb telomeric to CKDN2A, is an essential enzyme for the salvage of adenine and methionine (Figure 1). Cells that lack this enzyme become sensitive to purine synthesis inhibitors or methionine starvation and can be therapeutically exploited for selective therapy.1416 Deletion of MTAP was the most frequent alteration in genomewide profiling studies of oral squamous cell carcinoma.17 Because deletion of CDKN2AINK4a (p16) and CDKN2AARF (p14) causes dysregulation of the 2 pathways, Rb and p53, important in most cancers, loss of MTAP activity is thought to be incidental and not of pathogenic consequence.18

The goal of this study was to fine map precise architectural loss of CDKN2A (p14, p16), CDKN2B (p15), and MTAP to identify the extent to which this locus is affected by deletion, offering additional therapeutic targets of consideration in HNSCC.

METHODS
HNSCC CELL LINES AND DNA EXTRACTION

Tumor sample acquisition, tissue culture, karyotype analysis, and genomewide mapping for altered individual gene loci methods have been detailed elsewhere.1921 Cell lines UMSCC-11A and UMSCC-11B were derived from tumor tissue obtained from the primary tumor site (larynx) before and after chemotherapy, respectively. Cell lines UMSCC-17A (supraglottis) and UMSCC-17B (neck soft tissue) were derived from tumor tissue obtained simultaneously from primary (A) and metastatic (B) sites. Cell lines UMSCC-81A (larynx) and UMSCC-81B (tonsillar pillar) were derived from tumor tissue obtained from primary (A) and second primary (B) sites.

DNA from the 6 cell lines was extracted using the QIAamp Kit (Qiagen Inc, Chatsworth, Calif) at passages 83 and 90 for UMSCC-11A and UMSCC-11B, respectively; 138 and 184 for UMSCC-17A and UMSCC-17B, respectively; and 24 and 129 for UMSCC-81A and UMSCC-81B, respectively.

FINE-MAPPING LOSS AT THE 9p21 LOCUS

The multiplex ligation-dependent probe amplification assay (MLPA)21,22 provides relative quantification of loss and gain of gene loci using the candidate gene approach. Our group21 recently showed that for HNSCC cell lines UMSCC-11A/B, UMSCC-17A/B, and UMSCC-81A/B, loss and gain of genetic loci using a genomewide probe panel concurred with tumor karyotypes. The 112–gene probe panel with single probes for the CDKN2A and CDKN2B genes indicated loss of CDKN2A in all except UMSCC-11B, with concomitant loss of CDKN2B in UMSCC-11A, UMSCC-17B, and UMSCC-81A.

To examine the extent of deletion at the CDKN2B/CDKN2A/MTAP locus, a targeted MLPA probe panel was designed to dissect out the region of 9p21 loss. This panel of 38 probes comprises 17 control non–chromosome 9p probes and 21 chromosome 9p21 probes (Table). The latter start 3349 kb from the centromeric region and extend up to 600 kb to the 9p telomere to include 3 probes for CDKN2BINK4b, 7 for CDKN2AARF, INK4a, and 3 for MTAP (Table and Figure 1A).

INTERPRETATION

Normal tissue from each cancer patient serves as an internal reference when available. For cell lines, when normal DNA samples are not available, control (normal) male and female DNA samples are run with each probe set. Quantification—loss or gain of gene loci—is determined through a process of normalization. This process addresses variations in the surface area of a peak (intensity) encountered due to fluctuations in the assay run, such as amount of DNA, ploidy variations, and polymerase chain reaction (PCR) conditions. Briefly, the peak area for each probe is expressed as a percentage of the total surface area of all peaks of a sample in an assay run (Figure 2).21 Relative copy number for each probe is obtained as a ratio of the normalized value for each locus (peak) of the sample to that of the control. A difference is significant only if the ratio is less than 0.7 (loss) or higher than 1.3 (gain). A relative copy number of 2 is considered normal, 1 or 0 is considered loss, and 3 or more is considered gain. Loss of a gene copy is indicated by a reduction in peak height for that particular gene probe, homozygous loss is indicated by the absence of a peak (illustrated for CDKN2A and CDKN2B probes in UMSCC-81A) (Figure 2B), and gain of a copy number is correspondingly denoted by an increase in peak height.

Cell line DNA, starting with approximately 20 to 50 ng, is interrogated for the 38 genes in 1 reaction tube. Briefly, DNA is diluted with water to a total volume of 5 μL and then is denatured by heating for 5 minutes at 98°C in a thermocycler. Binary MLPA probes are added and allowed to hybridize to their targets during 16 hours of incubation at 60°C. Dilution buffer and a special Ligase-65 enzyme (MRC-Holland) are added to the vial. During 15 minutes of incubation at 60°C, the 2 parts of a probe can be ligated to each other and become an amplifiable molecule provided that the complementary sequence is present in the sample. This is followed by the addition of PCR primers, deoxynucleotide triphosphates, and Taq polymerase, followed by the following cycles: 1 cycle for 1 minute at 95°C; 10 cycles for 30 seconds at 95°C, 30 seconds at 70°C, and 1 minute at 72°C; and 30 cycles for 30 seconds at 95°C, 30 seconds at 60°C, and 1 minute at 72°C. All (ligated) probes are amplified by the same primer pair, one of which is tagged with a fluorescent dye. Amplification products (38 individual probe peaks in the control [Figure 2A]) are analyzed using an ABI 310/ABI 3100 DNA sequencer, and the products are quantified, normalized, and interpreted.

REAL-TIME PCR

Real-time PCR was used to confirm MLPA DNA copy number loss and retention of CDKN2A in cell lines UMSCC-11A/B, UMSCC-17A/B, and UMSCC-81A/B. In real-time PCR, the amount of product formed (detected by binding of the fluorescent dye SYBR green I) is plotted as a function of the number of cycles. Each cycle represents a doubling of DNA (ie, as measured by fluorescence of SYBR green I bound to double-stranded DNA), so the accumulation of DNA product is exponential across consecutive cycles (ie, 2n). It is the ability to identify the exponential phase of product formation “in real time” that makes the method quantitative because the amount of product is exponentially related to the amount of template (the unknown quantity) during this phase. Therefore, during the exponential phase of PCR product formation, the amount of product doubles during each cycle. The beginning of the exponential phase of amplification (so-called crossing points or threshold cycle [Ct]) is considered the most reliable point of the PCR related to sample concentration. Two samples that reach the threshold within 1 cycle of each other (eg, Ct1 = 10, Ct2 = 11) differ by 2-fold in the amount of amplified product (target gene).

Melting curve analysis is an exact and fast method for checking PCR specificity. Every DNA fragment melts at a characteristic temperature, called the melting temperature, defined as the temperature at which 50% of the DNA is single stranded. The most important criteria that determine the melting temperature are the G+C content and the length of the fragment. The LightCycler Instrument (Roche Diagnostic Corp, Indianapolis, Ind) monitors the fluorescence continuously while raising the temperature gradually. When the temperature in the capillary reaches the melting temperature of the fragment under study, there is a sharp decrease in fluorescence because SYBR green I dye is released from the amplicon. Plotting fluorescence vs temperature generates melting curves. When melting curves are displayed as the first negative derivative of fluorescence vs temperature, a peak is generated at the melting temperature. DNA melting curves were acquired using the LightCycler by measuring the fluorescence of SYBR green I during a linear temperature transition from 65°C to 97°C at 0.1°C/s. Fluorescence data were converted into melting peaks by using the LightCycler software (version 5.32) to plot the negative derivative of fluorescence over temperature vs temperature. For PCR products generated on the LightCycler, melting curve analysis was performed immediately after amplification.

Primers were designed in intron 1α of CDKN2A between nucleotides 187602 and 187697 using Oligo 6 (Molecular Biology Insights Inc, Cascade, Colo). The primer pairs were as follows: 5′ nucleotide 187602: 5′ GGC AAG GAG GAC CAT AAT TC 3′; 3′ nucleotide 187697: 5′ GGA CCA AGA CTT CGC TGA C 3′. Detection and quantification of target DNA was performed using the double-stranded DNA specific dye SYBR green I, which is similar to ethidium bromide in that it binds preferentially to double-stranded DNA, emitting a fluorescent signal that is proportional to the amplified target concentration. The signal is measured in channel 1 (at 530 nm) at the end of the elongation phase once per cycle and is monitored in real time. The ubiquitous β-globin gene was used as an internal control (housekeeping gene).

RESULTS
FINE MAPPING

The UMSCC-11B cell line retained all 9p loci tested in the region. Homozygous loss was indicated for CDKN2BINK4b and CDKN2AARF, INK4a in UMSCC-81A and UMSCC-11A (Figure 1B and Figure 2, UMSCC-81A). Cell lines UMSCC-17A/B indicated homozygous deletion of CDKN2AARF, INK4a starting at p16 exon 1α to include exons 2 and 3. Cell line UMSCC-81B indicated retention of all 9p loci except for exon 1α (p16INK4a, probe 8) (Figure 1A), making the latter the smallest region of overlap for loss in the CDKN2AARF, INK4a region (Figure 1B and Figure 2).

In UMSCC-11A, of the 3 MTAP MLPA probes—1 at the 3′ end (probe 13, exon 7) (Figure 1A), 1 in exon 6 (probe 12) (Figure 1A), and 1 at the 5′ end (probe 11)—homozygous loss was indicated only for the 3′ end of the MTAP gene.

VALIDATION OF GENOMIC LOSS OF THE CDKN2A LOCUS

Genomic loss and retention of CDKN2A copy number with real-time PCR for UMSCC-11A/B, UMSCC-17A/B, and UMSCC-81A/B was concordant with confirming retention of the CDKN2A intron 1α sequences in UMSCC-11B and UMSCC-81B and loss in UMSCC-11A, UMSCC-17A/B, and UMSCC-81A (illustrated for UMSCC-11A/B in Figure 3). Here, normal copy number was indicated for β-globin in UMSCC-11A/B and for CDKN2A in UMSCC-11B. In UMSCC-11A, however, loss of CDKN2A is indicated by a delayed Ct. Melting curve analysis confirmed lack of a specific melting peak for CDKN2A. The MLPA fine mapping confirmed homozygous loss of this region (Figure 1B).

COMMENT

Genetic alterations provide a means of identifying tumor cells and defining changes that presumably determine biological differences from their normal counterparts. The underlying hypothesis is that behavior of tumor cells is determined by genetic changes that alter cell growth, cell differentiation, programmed cell death, and cell migration. Knowledge of the genetic mechanisms that drive cancer growth and development can provide better diagnostic and prognostic information and more appropriate selection of therapy.

The precise role of the clinical application of molecular prognostic markers in HNSCC remains elusive. Studies of consistent genetic changes in HNSCC have been instrumental in the initiation of gene therapy trials in vitro and in experimental animals. The transfer of wild-type p53 into HNSCC tumor cells with mutant p53 was shown to induce growth arrest and tumor regression.23 Other genes being targeted for gene therapy include the B-cell leukemia/lymphoma 2 (BCL2) gene homologue BCL2XS, which inhibits BCL2 function and has been shown to have altered the sensitivity of tumor cells to chemotherapeutic agents by restoring sensitivity to apoptosis.24 Several strategies directed against epidermal growth factor receptor overexpression have included the use of monoclonal antibodies targeted against epidermal growth factor receptor.25 A chimeric anti–epidermal growth factor receptor monoclonal antibody (C225) has been tested in combination with either cisplatin26 or radiation,27 and the results are promising.

Genetic alterations at the 9p21 locus have been linked to malignant progression in HNSCC.2,3 Currently, recombinant adenovirus capable of directing a high level of p16 protein expression (Ad5-p16) demonstrated a significant antitumor effect of Ad5-p16 against human HNSCC in vivo.13 Other proteins produced at this 9p21 locus—p15 (CDKN2B), p14 (CDKN2A), and MTAP—each have the potential to independently serve as therapeutic targets in HNSCC.

In this study, 7 HNSCC cell lines, including 6 that indicated loss and retention of the CDKN2A/B locus by an MLPA genomewide 112-probe panel,21 were examined using a fine-mapping 9p21 probe set. We found homozygous loss at the CKDN2A locus in 5 of 6 HNSCC cell lines: UMSCC-11A, UMSCC-17A/B, UMSCC-81A, and UMSCC-81B. The UMSCC-81B cell line indicated retention of all 9p loci except for exon 1α (p16INK4a), making the latter the smallest region of overlap for loss in the CDKN2AARF, INK4a region (Figure 1B and Figure 2). Studies reporting homozygous deletions of CDKN2A in HNSCC may, therefore, underestimate the frequency of this region depending on the extent to which deletion mapping was performed.

Homozygous loss for CDKN2BINK4b and CDKN2AARF, INK4a was observed in UMSCC-81A (Figure 1B). For UMSCC-11A, homozygous loss of 9p21 observed by MLPA fine-mapping analysis was also supported by real-time quantitative PCR for the CDKN2A gene. In addition, the inherent sensitivity of real-time PCR revealed the late appearance of a product (Figure 3), where this late product seems to represent a minor population of cells (many logs fewer than the predominant population) from the original tumor that may correspond to the population that survived chemotherapy to predominate in the UMSCC-11B cell line. The latter would explain the observation of an intact 9p21 region in UMSCC-11B and offers a rationale, presumably, for selection of a tumor subpopulation that escapes chemotherapy likely aided by an intact 9p21 DNA repair and survival repertoire.

In UMSCC-11A, of the 3 MTAP MLPA probes (1 at the 3′ end, 1 in intron 6, and 1 at the 5′ end), homozygous loss was indicated for only the 3′ end of the MTAP gene, pointing to selective loss of the carboxy terminal region of the gene (Figure 1B).

Toohey,28 a quarter of a century ago, first recognized that certain murine malignant hematopoietic cell lines lacked MTAP29 activity. The function of MTAP is to cleave methylthioadenosine, a by-product of polyamine metabolism, to adenine and 5′-methylthioribose1-phosphate, which are recycled to adenine nucleotides and methionine, respectively. The MTAP, the first enzyme in the pathway, seems to be expressed in all normal human tissues,30,31 which suggests that the entire salvage pathway is present in all of the cells in the human body, designating MTAP an important housekeeping gene in malignancy.

Genetic studies show high rates of MTAP loss in non–small cell lung cancer, melanoma, bladder cancer, pancreatic cancer, osteosarcoma, and endometrial cancer.3235 Tumors that lack MTAP are expected to be sensitive to inhibitors of purine synthesis or methionine starvation. Genomewide profiling studies of oral SCC17 found frequent deletion of MTAP; however, mapping of precise loss of MTAP and its association with concomitant or independent loss with CDKN2A/B genes was not addressed.

Because deletion of the CDKN2AINK4a (p16) and CDKN2AARF (p14) genes causes dysregulation of the 2 pathways important in most cancers (Rb and p53), loss of MTAP activity is thought to be incidental and not of pathogenic consequence.18 However, there are several reasons proposed to suggest that this may not be the case.14 First, homozygous deletion is an unusual mechanism for inactivation of a tumor suppressor gene. Most tumor suppressor genes are inactivated by point mutation of one allele followed by loss of the other allele (loss of heterozygosity). This is rarely observed for CDKN2A.36 A likely hypothesis for this observation is that homozygous deletion can remove more than 1 gene from the region, whereas point mutation followed by loss of heterozygosity cannot.14 Second, in certain cancers, loss of MTAP has been observed in cells that retain p16. Schmid et al32 found in a study of non–small cell lung cancer that homozygous deletion of MTAP occurred in 38% (19 of 50) of the samples compared with only 18% (9 of 50) for p16. In another study37 it was found that 3 of 7 primary astrocytomas were deleted for MTAP, but only 2 of 7 were deleted for p16. The fact that MTAP is lost independently of p16 hints that loss of MTAP may have some functional basis in tumor biology.14 In UMSCC-11A, despite homozygous loss of CDKN2A/B genes, only the 3′ end of MTAP was lost. The latter may point to the 3′ end as a selective target for gene deletion, suggesting the carboxy terminal of the gene as a critical region.

Tumors that lack MTAP are expected to be sensitive to inhibitors of purine synthesis or methionine starvation. It has been demonstrated that the antipurine-related growth-inhibitory action of the antifolate agent methotrexate was more pronounced in subsets of pancreatic carcinoma cell lines that were p16−, MTAP− than in pancreatic carcinomas or normal keratinocyte epithelial cells that were p16+, MTAP+.38 Furthermore, it was shown that the co-addition of an inhibitor of MTAP enhanced the potency and efficacy of the antipurine-related growth-inhibitory actions of methotrexate in MTAP− but not MTAP+ cell lines.38 Thus, MTAP+ cells are less sensitive to the inhibitory effects that antifolates such as methotrexate have on purine de novo synthesis. In contrast, because MTAP-deficient malignant cells cannot recycle the purine moiety of methylthioadenosine, tumor cells are more dependent on purine de novo biosynthesis and are more sensitive than MTAP-containing cells to the antipurine actions of antifolates.

To examine the effects of MTAP in tumorigenesis, Christopher et al14 reintroduced MTAP into MCF-7 breast adenocarcinoma cells. Although MTAP expression does not affect the growth rate of cells in standard tissue culture conditions, it severely inhibits their ability to form colonies in soft agar or collagen. In addition, the study showed that MTAP-expressing cells are suppressed for tumor formation when implanted into severe combined immunodeficient (SCID) mice. Because MTAP expression causes a significant decrease in intracellular polyamine levels and alters the ratio of putrescine to total polyamines, consistent with this observation, the study also showed that the polyamine biosynthesis inhibitor α-difluoromethylornithine inhibited the ability of MTAP-deficient cells to form colonies in soft agar, whereas addition of the polyamine putrescine-stimulated colony formation in MTAP-expressing cells, suggesting that MTAP-negative tumors may be particularly sensitive to α-difluoromethylornithine. The results of this study indicate that MTAP has tumor suppressor activity and suggest that its effects may be mediated by altering intracellular polyamine pools.

There are a variety of potential implications of these findings for clinical cancer chemotherapy, including HNSCC. If normal cells, which are uniformly MTAP+,39 are intrinsically less sensitive to the antipurine actions of antifolates such as methotrexate, then this might in part be the basis for the relatively selective action that these agents have against certain malignancies. If the corollary is correct that MTAP-deficient malignant cells are especially sensitive to certain antifolates that act in whole or in part by inhibiting purine de novo synthesis, one would theoretically obtain an improved therapeutic index by identifying the subset of patients with MTAP-deficient malignancies (including HNSCC) and treating them with such agents. α-Difluoromethylornithine is currently undergoing clinical trials for use as adjunct therapy for a variety of different cancers.40,41 Genomewide profiling studies of oral squamous cell carcinoma,17 which found frequent deletions of MTAP and loss of MTAP in 1 of 6 cell lines in this study, provide a strengthening rationale for a role of MTAP in HNSCC. Stratifying the prognosis of 9p21-deleted HNSCC for MTAP status may provide a rationale to examine this locus in larger studies to assess the feasibility of polyamine biosynthesis inhibitors as novel treatment interventions in MTAP-deleted HNSCC tumors.

There is a large gap in our knowledge and understanding of the frequency and precise loss of p15, p14, p16, and MTAP in HNSCC tumorigenesis. Better understanding of the contribution of these 4 gene products in the prognosis of 9p21-altered HNSCC can provide impetus for exploitation of these targets as therapeutic biomarkers in head and neck cancer.

Back to top
Article Information

Correspondence: Maria J. Worsham, PhD, Department of Otolaryngology–Head and Neck Surgery, Henry Ford Hospital, 1 Ford Pl, 1D, Detroit, MI 48202 (mworsha1@hfhs.org).

Submitted for Publication: April 30, 2005; final revision received September 1, 2005; accepted October 10, 2005.

Author Contributions: Dr Worsham had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Financial Disclosure: None.

Funding/Support: This study was supported by grant 1 R01 DE15990 from the National Institutes of Health, Bethesda, Md (Dr Worsham).

Previous Presentation: This study was presented as a late-breaking abstract at the annual meeting of the American Society of Investigative Pathology; April 15, 2003; San Diego, Calif.

References
1.
Van Dyke  DLWorsham  MJBenninger  MS  et al.  Recurrent cytogenetic abnormalities in squamous cell carcinomas of the head and neck. Genes Chromosomes Cancer 1994;9192- 206
PubMedArticle
2.
Danahey  DGTobin  EJSchuller  DEBier-Laning  CMWeghorst  CMLang  JC p16 Mutation frequency and clinical correlation in head and neck cancer. Acta Otolaryngol 1999;119285- 288
PubMedArticle
3.
Lydiatt  WMDavidson  BJSchantz  SPCaruana  SCChaganti  RS 9p21 Deletion correlates with recurrence in head and neck cancer. Head Neck 1998;20113- 118
PubMedArticle
4.
Sherr  CJ Cancer cell cycles. Science 1996;2741672- 1677
PubMedArticle
5.
Johnson  DGWalker  CL Cyclins and cell cycle checkpoints. Annu Rev Pharmacol Toxicol 1999;39295- 312
PubMedArticle
6.
Haber  DA Splicing into senescence: the curious case of p16 and p19ARF. Cell 1997;91555- 558
PubMedArticle
7.
Xing  EPNie  YSong  Y  et al.  Mechanisms of inactivation of p14ARF, p15INK4b and p16INK4a genes in human esophageal squamous cell carcinoma. Clin Cancer Res 1999;52704- 2713
PubMed
8.
Sherr  CJ The Pezcoller Lecture: cancer cell cycles revisited. Cancer Res 2000;603689- 3695
PubMed
9.
Hannon  GJBeach  D p15INK4B Is a potential effector of TGF-β–induced cell cycle arrest. Nature 1994;371257- 261
PubMedArticle
10.
Shintani  SNakahara  YMihara  MUeyama  YMatsumura  T Inactivation of the p14(ARF), p15(INK4B) and p16(INK4A) genes is a frequent event in human oral squamous cell carcinomas. Oral Oncol 2001;37498- 504
PubMedArticle
11.
Rocco  JWSidransky  D p16 (MTS-1/CDKN2/INK4a) in cancer progression. Exp Cell Res 2001;26442- 55
PubMedArticle
12.
Borg  ASandberg  TNilsson  K  et al.  High frequency of multiple melanomas and breast and pancreas carcinomas in CDKN2A mutation-positive melanoma families. J Natl Cancer Inst 2000;921260- 1266
PubMedArticle
13.
Rocco  JWLi  DLiggett  WH  et al.  p16INK4A adenovirus-mediated gene therapy for human head and neck squamous cell cancer. Clin Cancer Res 1998;41697- 1704
PubMed
14.
Christopher  SADiegelman  PPorter  CWKruger  WD Methylthioadenosine phosphorylase, a gene frequently codeleted with p16(cdkN2a/ARF), acts as a tumor suppressor in a breast cancer cell line. Cancer Res 2002;626639- 6644
PubMed
15.
Harasawa  HYamada  YKudoh  M  et al.  Chemotherapy targeting methylthioadenosine phosphorylase (MTAP) deficiency in adult T cell leukemia (ATL). Leukemia 2002;161799- 1807
PubMedArticle
16.
Efferth  TMiyachi  HDrexler  HGGebhart  E Methylthioadenosine phosphorylase as target for chemoselective treatment of T-cell acute lymphoblastic leukemic cells. Blood Cells Mol Dis 2002;2847- 56
PubMedArticle
17.
Chen  YJLin  SCKao  T  et al.  Genome-wide profiling of oral squamous cell carcinoma. J Pathol 2004;204326- 332
PubMedArticle
18.
Nobori  TTakabayashi  KTran  P  et al.  Genomic cloning of methylthioadenosine phosphorylase: a purine metabolic enzyme deficient in multiple different cancers. Proc Natl Acad Sci U S A 1996;936203- 6208
PubMedArticle
19.
Carey  TEVan Dyke  DLWorsham  MJ  et al.  Characterization of human laryngeal primary and metastatic squamous cell carcinoma cell lines UM-SCC-17A and UM-SCC-17B. Cancer Res 1989;496098- 6107
PubMed
20.
Bradford  CRKimmel  KAVan Dyke  DL  et al.  11p deletions and breakpoints in squamous cell carcinoma: association with altered reactivity with the UM-E7 antibody. Genes Chromosomes Cancer 1991;3272- 282
PubMedArticle
21.
Worsham  MJPals  GSchouten  J  et al.  Delineating genetic pathways of disease progression in head and neck squamous cell carcinoma. Arch Otolaryngol Head Neck Surg 2003;129702- 708
PubMedArticle
22.
Schouten  JPMcElgunn  CJWaaijer  RZwijnenburg  DDiepvens  FPals  G Relative quantification of 40 nucleic acid sequences by multiplex ligation-dependent probe amplification. Nucleic Acids Res 2002;30e57
PubMedArticle
23.
Liu  TJZhang  WWTaylor  DL  et al.  Growth suppression of tumor cells of human head and neck cancer cells by the introduction of wild-type p53 gene via a recombinant adenovirus. Cancer Res 1994;543662- 3667
PubMed
24.
Sumantran  VNEalovega  MWNunez  G  et al.  Overexpression of Bcl-XS sensitizes MCF-7 cells to chemotherapy-induced apoptosis. Cancer Res 1995;552507- 2510
PubMed
25.
Hoffmann  THafner  DBallo  H  et al.  Antitumor activity of anti-epidermal growth factor receptor monoclonal antibodies and cisplatin in ten human head and neck squamous cell carcinoma lines. Anticancer Res 1997;174419- 4425
PubMed
26.
Shin  DMDonato  NJPerez-Soler  R  et al.  Epidermal growth factor receptor–targeted therapy with C225 and cisplatin in patients with head and neck cancer. Clin Cancer Res 2001;71204- 1213
PubMed
27.
Bonner  JARaisch  KPTrummell  HQ  et al.  Enhanced apoptosis with combination C225/radiation treatment serves as the impetus for clinical investigation in head and neck cancers. J Clin Oncol 2000;1847S- 53S
PubMed
28.
Toohey  JI Methylthioadenosine nucleoside phosphorylase deficiency in methylthio-dependent cancer cells. Biochem Biophys Res Commun 1978;8327- 35
PubMedArticle
29.
Trackman  PCAbeles  RH Methionine synthesis from 5′-S-methylthioadenosine: resolution of enzyme activities and identification of 1-phospho-5-S methylthioribulose. J Biol Chem 1983;2586717- 6720
PubMed
30.
Kamatani  NNelson-Rees  WACarson  DA Selective killing of human malignant cell lines deficient in methylthioadenosine phosphorylase, a purine metabolic enzyme. Proc Natl Acad Sci U S A 1981;781219- 1223
PubMedArticle
31.
Olopade  OIPomykala  HMHagos  F  et al.  Construction of a 2.8-megabase yeast artificial chromosome contig and cloning of the human methylthioadenosine phosphorylase gene from the tumor suppressor region on 9p21. Proc Natl Acad Sci U S A 1995;926489- 6493
PubMedArticle
32.
Schmid  MMalicki  DNobori  T  et al.  Homozygous deletions of methylthioadenosine phosphorylase (MTAP) are more frequent than p16INK4A (CDKN2) homozygous deletions in primary non-small cell lung cancers (NSCLC). Oncogene 1998;172669- 2675
PubMedArticle
33.
Della Ragione  FRusso  GOliva  A  et al.  5′-Deoxy-5′-methylthioadenosine phosphorylase and p16INK4 deficiency in multiple tumor cell lines. Oncogene 1995;10827- 833
PubMed
34.
Stadler  WMSherman  JBohlander  SK  et al.  Homozygous deletions within chromosomal bands 9p21–22 in bladder cancer. Cancer Res 1994;542060- 2063
PubMed
35.
Garcia-Castellano  JMVillanueva  AHealey  JH  et al.  Methylthioadenosine phosphorylase gene deletions are common in osteosarcoma. Clin Cancer Res 2002;8782- 787
PubMed
36.
Cairns  PPolascik  TJEby  Y  et al.  Frequency of homozygous deletion at p16/CDKN2 in primary human tumours. Nat Genet 1995;11210- 212
PubMedArticle
37.
Brat  DJJames  CDJedlicka  AE  et al.  Molecular genetic alterations in radiation-induced astrocytomas. Am J Pathol 1999;1541431- 1438
PubMedArticle
38.
Chen  ZHOlopade  OISavarese  TM Expression of methylthioadenosine phosphorylase cDNA in p16−, MTAP− malignant cells: restoration of methylthioadenosine phosphorylase-dependent salvage pathways and alterations of sensitivity to inhibitors of purine de novo synthesis. Mol Pharmacol 1997;52903- 911
PubMedArticle
39.
Myers  RWAbeles  RH Conversion of 5-S-methyl-5-thio-D-ribose to methionine in Klebsiella pneumoniae: stable isotope incorporation studies of the terminal enzymatic reactions in the pathway. J Biol Chem 1990;26516913- 16921
PubMed
40.
O’Shaughnessy  JADemers  LMJones  SE  et al.  α-Difluoromethylornithine as treatment for metastatic breast cancer patients. Clin Cancer Res 1999;53438- 3444
PubMed
41.
Levin  VAUhm  JHJaeckle  KA  et al.  Phase III randomized study of postradiotherapy chemotherapy with α-difluoromethylornithine-procarbazine: N-(2-chloroethyl)-N'-cyclohexyl-N-nitrosurea, vincristine (DFMO-PCV) versus PCV for glioblastoma multiforme. Clin Cancer Res 2000;63878- 3884
PubMed
×