[CANCER RESEARCH 52, 1451-1456, March 15, 1992]

Chromosome 3p Deletions in Head and Neck Carcinomas: StatisticalAscertainment of Allelic Loss1

Farida I.at it, Matthew Fivash, Gladys Glenn, Kaiman Tory, Mary Lou Orcutt, Krista Hampsch, John Delisio,Michael I .crinan,2 Janet Cowan, Michael Beckett, and Ralph Weichselbaum

Laboratory of ¡mmunobiology [F.L., G. G., M.L.O., M.L.), Data Management Services, Inc. fM.F.J, and Program Resources, Inc./DynCorp, National Cancer Institute-Frederick Cancer Research and Development Center, Frederick, Maryland 21702-1201 [K.T., K.H., J.D.J; and the Department of Radiation Oncology, University ofChicago Medical Center, Chicago, Illinois 60637[J.C., M.B., R.W.]

ABSTRACT

Loss of function of tumor suppressor genes is important in the originand progression of common adult tumors. Loss of heterozygosity indicating allelic loss has been used to detect chromosomal regions that harborthese genes. Using over 20 restriction fragment length polymorphismmarkers spaced throughout the entire length of chromosome 3p, we havegenerated 3p allelotypes for 18-26 head and neck squamous cell carcinoma cell lines. We then estimated the average heterozygosity over 19loci for a random sample drawn from natural populations to be 7.80 andthat for the tumor lines to be 1.65, indicating a gross reduction ofheterozygosity, presumably due to allelic loss. Further comparison of perlocus heterozygosity in normal and tumor DNAs showed which locicontributed to the general loss of heterozygosity. We showed that thecommonly deleted region of 3p probably lies telomeric to D3S3 (3pl4)and centramene to RAH (3p25). This large region includes severalputative tumor suppressor genes involved in multiple common tumortypes of lung, breast, kidney, ovary, and cervix. The data demonstratethat chromosome 3p allelic loss is a common event in head and neckcancers and suggest that chromosome 3p tumor suppressor genes contribute to the pathogenesis of these tumors.

INTRODUCTION

Tumors of the head and neck are a heterogeneous group ofneoplasms that manifest a wide range of clinical behaviors (1,2). In Western countries they constitute up to 5% of all cancersand 90-95% of them are squamous cell carcinomas (1-3).Tobacco and alcohol, well established as attributable risk factors, and some industrial carcinogens are thought to be amongetiological factors for head and neck cancers (1-4). Geneticpredisposition has also been described in some families withmultiple affected individuals (5). However, despite emphasis onearly diagnosis, survival rates for patients with head and necktumors have not improved dramatically (1).

Lack of progress in head and neck oncology emphasizes theimportance of molecular studies of genetic changes associatedwith the origin, development, and malignant progression ofthese neoplasms. It is assumed that knowledge of such alterations may establish molecular correlates of tumor behavior.This will, in turn, benefit the patient with diagnostic andprognostic information, regarding the clinical course andoutcome.

Human carcinogenesis is thought to be a multistep process

Received 9/16/91; accepted 1/8/92.The costs of publication of this article were defrayed in part by the payment

of page charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1This project has been funded at least in part with Federal funds from theDepartment of Health and Human Services under Contracts NO1-CO-74102(K.T., K.H., J.D.), NO1-CO-74103 (M.F.), and CA 41 068 (R.W.). The contentof this publication does not necessarily reflect the views or policies of theDepartment of Health and Human Services, nor does mention of trade names,commercial products, or organizations imply endorsement by the United StatesGovernment.

2To whom requests for reprints should be addressed, at the Laboratory of

Immunobiology, National Cancer Institute, Frederick Cancer Research and Development Center, Building 560, Room 12-71, Frederick, MD 21702-1201.

driven by genetic changes that include inactivation of TSGs3

(3) on several chromosomes (6, 7). Inactivation of both copiesof TSGs as required by the "two-hit" mutation model (6) occurs

over a period of time. The first mutation is proposed to occureither in the germline or in the somatic cells, whereas thesecond hit occurs in somatic cells. This second event oftenresults in reduction to homozygosity of a segment of the chromosome region that harbors the particular TSG. Allelic loss asdetected by RFLP deletion analysis has been successfully usedto identify TSG loci in a growing number of human cancers(7). A high density set of ordered RFLP markers is required todelineate further the candidate region for a particular TSG.

Deletions in the short arm of chromosome 3 are common inmany human cancers (7, 8), including sporadic and hereditaryrenal carcinomas, small cell lung carcinoma, nonsmall cell lungcarcinoma, and carcinomas of the ovary, breast, and cervix. Thechromosome 3p region involved in these malignancies havebeen mapped with sufficient precision: the von Hippel-LindauTSG to a small region between RAF1 (3p25) and D3S18 (3p26)(9); the renal carcinomas and small cell lung carcinoma genesto 3pl4-p26 (8); the non-small cell lung carcinoma gene to3pl4-pter (8); and the uterine cervix gene to 3pl4-p21 (10).

Previously, cytogenetic studies have been performed in anumber of head and neck tumor cell lines (11-13). These studiesshowed complex karyotypes with frequent abnormalities ofchromosomes 1, 3, 6, 9, and 11. Common findings for chromosome 3 included deletions in the short arm with breakpointsat 3pl 1, 3pl2, and 3pl4 (11-13).

To date there have been no reports describing allelic loss inHNSCC. The lack of such studies prompted us to initiatemolecular mapping of chromosome 3p loci involved in headand neck cancers. This study was greatly facilitated by a highdensity set of RFLP markers for which order and locationinformation was obtained (14, 15).

Herein we present an extensive allelotype analysis of theshort arm of chromosome 3 using 18-26 head and neck tumorcell lines and over 20 RFLP markers spaced throughout theentire length of the short arm. Because of the absence ofmatching normal tissues, we ascertained allelic loss by a novelstatistical method. We tentatively mapped the commonly deleted region distal to D3S3 (3pl4) and proximal to D3S18(3p26).

MATERIALS AND METHODS

HNSCC Cell Lines. The establishment of the HNSCC cell lines wasreported earlier (16, 17). They were derived from patient tumors priorto treatment and were free of any nontumorous contaminants thatcould affect alíeledeletion analyses. To avoid changes in long termcultures early passage cells were used for RFLP studies. DNA finger-

3The abbreviations used are: TSGs, tumor suppressor genes; HNSCC, head

and neck squamous cell carcinoma; FOH, frequencies of heterozygosity; RFLP,restriction fragment length polymorphism.

1451

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

3p DELETION IN HEAD AND NECK CANCER

printing with a chromosome 3 hypervariable minisatellite probe (14)was performed to confirm that no cross-contamination occurred in ourcell cultures and that all cell lines were unique isolates (not shown).The HNSCC cell lines and selected clinical characteristics of thepatients from whom they were derived were described earlier (16). Over80% of them belonged to the Caucasian population.

DNA Extraction and Southern Blot Analysis. DNA was preparedfrom monolayer cell cultures as described elsewhere (17). DNA (5-10>ig)was digested with a 10-fold excess of restriction enzymes (BethesdaResearch Laboratories, Inc., Bethesda, MD) according to the manufacturer's specifications in buffers supplemented with S mM spermidine.

Fractionation of digested DNA was performed by elect rophoresis in1% agarose gel in re-circulating 20 mM Tris HC1, pH 7.3-10 IHMacelate2 mM EDTA buffer for approximately 10-20 h at 2 V/cm. Followingelectrophoresis DNA was transferred to nylon membranes (MicronSeparations, Inc., Westboro, MA) by using the capillary transfermethod (18); afterwards the membranes were baked for 3 h at 80°Cin

a vacuum oven.Radioactive DNA probes were prepared by random primer extension

method (19) using [a-32P]dCTP (specific activity, >3000 ¿iCi/mmo!;

Amersham, Arlington Heights, IL). Hybridization of DNA was doneat 65"C for 12-18 h in a 5x standard saline-citrate solution containing

0.75 M sodium chloride, 0.075 M sodium citrate (pH 7.0), 1% sodiumdodecyl sulfate, denatured salmon sperm DNA (100 //«/nil),and ifnecessary varying amount of human DNA and denatured labeled DNAprobe. Following hybridization, filters were washed at 60'C in 0.1 x

standard saline-citrate and 0.1% sodium dodecyl sulfate.Radioactive bands were visualized by autoradiography at -HOT with

X-Omat AR (Eastman Kodak, Rochester, NY) X-ray film and a CronexLightning Plus (Dupont Chemical Co., Newton, CT) intensifyingscreen. After the results were obtained with a particular probe, radioactive DNA probe was stripped off the filter by incubating in 0.2 MNaOH at room temperature for 15 min and the filter was re-hybridized

to a new probe.DNA Probes. All probes used in this study recognize RFLP in human

DNA derived from chromosome 3p except probe EFD 64.1 that recognizes an RFLP on 3q. A complete description of the probes, includingtheir chromosomal location (map position), alíelefrequencies andRFLP revealing enzymes are detailed in Table 1.

Cytogenetic Analysis. Standard cytogenetic procedures were used inthese studies. Briefly, cultured cell lines were harvested for analyses as

described earlier (20) with 1 h Colcemid treatment and hypotonietreatment with warmed potassium chloride/sodium citrate solution(0.06 M KC1/0.6% sodium citrate mixed 1:1 before use) for 14 min.Fifteen G-banded cells were karyotyped from each cell line. Cytogeneticanalyses were done by one of us (J. C.) on the 14 of the cell lines.

AlíeleFrequency Determinations. The frequency of alíelesfor each ofthe 18-20 loci analyzed in the present study was determined usinginformation generated from DNA samples obtained from Centred'Etude du Polymorphisme Humain and from the literature. The

observed frequency of heterozygosity in the HNSCC cell lines wasdetermined using information generated from corresponding DNAsamples.

Statistical Methods. The method used to calculate the expectedheterozygosity in a randomly chosen individual is provided in the"Appendix." In general, the x1 goodness of fit test and tests assuming

binomial or multinominal distributions (21) were used to calculatesignificance of differences between experimental groups of data (e.g.,estimated and observed heterozygosity at 3p loci).

RESULTS

In the commonly used method, allelic loss is determined bycomparing constitutional and tumor genotypes for RFLPmarker loci for a specific chromosome or chromosome regionor throughout the whole genome (recently termed allelotype orallelotyping) (22).

In the present work we use a novel method to ascertain allelicloss on chromosome 3p in HNSCC. First, we generate anallelotype of 3p using a number of HNSCC cell lines and set of3p RFLP marker loci for which order and map position wasestablished (Table 1). We then compared the average heterozygosity computed for these loci in a randomly chosen individual from natural populations with observed heterozygosity intumor cell DNAs. Further, after concluding a general loss ofalíelesin the tumor population we estimate the statistical significance of loss for each individual locus which allows us todelimit the commonly deleted region on 3p. Finally, to delineatefurther the deleted region, we again compare per locus heterozygosity in the natural population and tumor cell DNA.

Table 1 Chromosome 3p polymorphic locf

LocusD3S215D3S238D3S22SD3S197D3SJ9ID3S18D3S18D3S732D3S57IRAF\D3S588THRBD3S734D3S232D3F1SS2D3S32D3S2D3S642D3S3D3S4D3S216D3S733D3SS73D3S42Location3p26-pter3p26-pter3p26-pter3p26-pter3p26-pter3p26.13p26.13p263p253p253p24-p253p243p21-p243p21-p243p213pl4-p213pl4.2-p213pl33pl33pll-pl23pl33pl33pl33qllProbeProbe(s)XLIB11-31XLIB48-95XLIB42-26XLIB21-14XLIB38-96CRIL162cL

162-1XCH4A-52XLIB

2-42p627XLIB

12-69pHB302XLIB7-78XLIB45-82pH3H2pEFDI45.1pHF12-32XCh

4A-7pMS1-37B67XLIB

32-90XLIB17-78XLIB3-23pEFD64.1Restriction

enzymeMsplTaqlHindmHindMsplOral««millHindlllMsplBgllHindlllHindmTaqlMsplHindlllTaqlMsplMsplMsplMsplEcoRlHindlllEcoRlHindSizeof alíeles

(kilobases)10.5;

12.59.4;3.89.8;5.52.2;1.7

+0.56.8;4.73.8;1.52.8;2.58.7;4.79.0;

7.5+1.52.9;2.654.0;3.37.2;

6.5; 5.9;5.07.0;5.511.

5; 8.9;7.64.5;3.52.0;2.39.0;4.02.9;1.3+1.63.8;

3.4;2.84.9;3.7+1.24.8;

1.63.5;3.012.0;7.57.3;1.41.08-5.5PolymorphismAllelic

frequencynormalpopulations0.88:0.120.39:0.610.82:0.180.20:0.800.27:0.730.25:0.750.1:0.90.69:0.310.78:0.220.75:0.250.78:0.220.07:0.1;

0.46:0.360.35;0.650.02:0.75;0.230.61:0.390.5:0.50.47:0.530.74:0.260.34:0.58:0.080.06:0.940.16:0.840.68:0.320.20:0.800.38:0.620.006-0.42Allelic

proportions inhead and necktumors0.83:0.170.23:0.770.73:0.270.48:0.520.41:0.590.17:0.830.05:0.950.56:0.440.74:0.260.67:0.330.71:0.290.04:0.04:0.56:0.360.26:0.740.04:0.88:0.080.47:0.530.52:0.480.37:0.630.59:0.410.44:0.44:0.120.12:0.880.18:0.820.70:0.300.16:0.840.43:0.57

•¿�Polymorphic probes and the loci they represent are listed in the order they occur on the chromosome. The order was established by physical mapping and genetic

linkage (Refs. 9, 14, 15; K. Tory, F. Latif, W. Modi, L. Schmidt, M. H. Wei, H. Li, P. Cobler, M. L. Orcutt, J. Delisio, L. Geil, B. Zbar, and M. I. Lerman. A geneticlinkage map of 96 loci on the short arm of human chromosome 3, Genomics, in press, 1992.)

1452

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

3p DELETION IN HEAD AND NECK CANCER

2S i w, r* *o —¿�< u- ^o —¿�•¿�v v i/ —¿� iVif^t/l

—¿�INfNfH

sa

8a

i

ÕO

—¿�_ _ r l r i

—¿�rNfSÕN —¿�—¿�—¿�(N(N»Nt

—¿�ÕN*-— »SÕN.- —¿�(N —¿�—¿�—¿�rlfN —¿�<H —¿�ÕH —¿�

ri —¿�—¿�—¿�rinn —¿�—¿�ri —¿� fsfs —¿�fN —¿�t

SíjJ£

u.<K

—¿�•ÕN —¿�fSÕN —¿�fN— f

—¿�fN —¿�ÕN '

—¿�r I —¿�ri . r I —¿�r I —¿�r l —¿�

SQ

—¿�rl —¿�—¿�—¿�rin —¿�—¿�~— rt —¿�ri— —¿�nr<

—¿�—¿�«Nr-tíN —¿�—¿�IN—

ÕNÕN —¿�—¿�rSÕN—.«N —¿�fN(N—.CNïS —¿�ÕNÕNÕN

¡I

ï8

*V ""

Û Ú

r- «/iÕNw»f*i ««n—¿�'i1 T T *?

«WS-íi-iW

SriO »N

HHH">¡\3SSl>óü0 i

—¿�-t oo O Ov «rir l —¿�u-, —¿�—¿�̂.

Ei £ibis Ãs̄ ï

Table 2 presents results of allelotyping of 18 HNSCC celllines tested with 19-22 markers and an additional 8 lines testedwith only 7 critical marker loci. A brief empirical inspection ofthe data clearly showed very low levels of heterozygosity for alltested loci suggesting possible allelic losses. Every locus distalto D3S2 retained heterozygosity at least in one or more tumorgenotypes, suggesting that the critical region must be proximalto this locus. Interestingly, a variable number of tandom repeatstype 3q marker retained heterozygosity in 18 of the 26 cell linesindicating that mostly the short arm suffered allelic losses andthat a total loss of one of the chromosome 3s did not occur.

To precisely ascertain loss of heterozygosity on 3p the datawere subjected to statistical analysis. Fig. 1 presents the expected distribution of the number of heterozygous loci in arandom sample of individuals assayed for the 22 loci withvarying per locus heterozygosities (Table 1). Assuming thatboth the normal population (unrelated individuals provided bythe Centre d'Etude du Polymorphisme Humain reference fam

ilies) and the tumor-bearing subpopulation are in Hardy-Weinberg equilibrium, these distributions can be calculated from perlocus FOH, (see "Appendix" for the derivation of the probabil

ity function). FOH for the normal population was generatedusing Centre d'Etude du Polymorphisme Humain-providedDNAs and, for some loci, the Centre d'Etude du Polymor

phisme Humain database. FOH for the tumor-bearing subpopulation was derived from an estimate based on the tumorallelotypes (Table 2) using alíelerepresentation at each locusin the tumor DNA. A x2 goodness of fit test of the hypothesisthat both distributions are the same yielded P = 0.99. Sincethese two distributions are statistically the same, the normalpopulation may be used as a reference for comparison with thetumor samples.

The tumor subpopulation may be examined for an overallloss of heterozygosity by comparing the observed number ofheterozygous loci per individual with the expected number ofheterozygous loci. This comparison may be made in three

0.10 -

Tumor Subpopulation

Normal Population

4 6 8 10 12 U 16 18 20

Number of Heterozygous Loci per Individual

22

Fig. 1. Calculated distribution of the number of heterozygous loci in a randomindividual with a sample of 22 scored chromosome 3p loci. •¿�.heterozygosity forindividuals drawn from the healthy natural population; í"¡,heterozygosity for

individuals from the tumor-bearing subpopulation. The per locus levels of heterozygosity were obtained for the unrelated individuals of Centre d'Etude du

Polymorphisme Humain families, and those for the tumor subpopulation werefrom alíelerepresentation in tumor cells (see "Materials and Methods"). Thecomputations use Equations A-C (see "Appendix").

1453

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

3p DELETION IN HEAD AND NECK CANCER

Table 3 Comparison of expected and observed number of heterozygous loci per individual

Expected no. ofheterozygous loci/individual

Observed no. ofheterozygous loci/individual

Probability of as few orfewer heterozygous loci/

individual

10 cell lines on22 loci

17 cell lines on19 loci

27 cell lines on5 loci

8.78

7.80

2.40

3.44

1.65

0.50

1.1 x IO"2»

6.1 x IO"56

4.7 x 10~28

U.DU*

0.55o5

0.50'•öC

0.45|

0,00

0.35

i0.30t

0.25

Õ§•0.205

0.100.05

n nn^

Expected Distribution

•¿�ObservedDistribution--i

il\.Lili

lili.

Table 4 Probability that 3p loci in tumor DNA have the same heterozygosity asthat of the normal population

O 2 4 6 8 10 12 14 16 18 20

Number of Heterozygous Loci per Individual

Fig. 2. Distribution of individual heterozygosity for chromosome 3p loci innormal and tumor cells with a sample of 19 scored loci. •¿�distribution in tumorcells; Q. expected heterozygosity for the healthy normal population.

different ways: for 10 cell lines on 22 loci; 17 cell lines on 19loci; and 28 cell lines on 5 loci. For each comparison a distribution for the number of heterozygous loci per individual wasdetermined (see "Appendix"), using the per locus FOH for the

normal population and the tumors. These distributions werecompared with a x2 goodness of fit test to ensure the validity

of using the normal population as a reference. Table 3 showsthe results of comparing the expected and observed number ofheterozygous loci per individual for the three ways outlinedabove. In each case it is clear that the tumors show a significantloss of heterozygosity. Fig. 2 illustrates this conclusion for thecase 17 cell lines on 19 loci.

The per locus alíeleproportions in both the normal population and the tumors were found to be statistically the same(Table 1). This fact allows per locus comparisons of the expected and observed number of hétérozygotesin the tumorpanel. Individual loci which show significant loss of heterozygosity are identified at the 0.0023 significance level (to controltype 1 errors). Table 4 shows which loci have lost significantheterozygosity. Nine of the 19-22 loci in this study showed

significant loss of heterozygosity. These nine loci are the majorcontributors to the overall loss of heterozygosity observed whenall 19-22 loci are considered together. Because these loci areordered they roughly define the common deleted region as distalto 3pl4 (D3S3) and proximal to 3p25 (RAF1). The contributionof individual loci to the overall loss of heterozygosity could alsobe determined by comparing the expected and observed FOHfor each locus. Fig. 3 presents these comparisons. Again, itcould be concluded that loci distal to D3S3 (3pl4) and proximalto D3S18 (3p26) show a significant loss of heterozygosity.

No. of3p loci hétérozygotesProbabilityD3S21S

20.6438D3S23830.2145D3S22510.1569D3SI9740.2691D3S19140.1363D3S1870.0039D3S732D3S571RAPÃŒD3SS88THRBD3S734D3S232D3F15S2D3S320.00540.00380.00094

x10-'°0.00034

xIO'50.0022I

xIO"78xIO'5D3S2

0 9 x10-*D3S6421 4 xIO'5D3S3

00.1307D3S400.0049D3S21620.0036D3S73320.0419D3SS733 0.0075Significant

(+/-r—————-——+++++++++————-

°+/-, +, significance at the 0.0023 level.

DISCUSSION

Loss of heterozygosity at TSG loci occurs frequently inhuman cancers and correlates with tumor stage and clinicalbehavior (7). Previously, cytogenetic analyses of HNSCC haverevealed complex karyotypes suggestive of consistent deletionsat TSG sites including chromosome 3p. However, to date therehave been no molecular studies describing allelic loss at TSGloci in head and neck cancers. The present paper examinesallelic loss at chromosome 3p TSG sites in HNSCC. We used19-22 HNSCC cell lines and a high density set of RFLPmarkers (over 20) on chromosome 3p to generate 3p allelotypesof tumor cells. The high-resolution RFLP allelotypes weresubjected to a novel statistical method to ascertain allelic lossand to define the commonly deleted region. This paper describesthe experimental data and the empirical and statistical analyses,while the "Appendix" provides the derivation of the probability

function for distribution of heterozygous loci in a randomindividual. Using these novel approaches, we demonstrated forthe first time allelic losses from a large area of chromosome 3pin 18-28 HNSCC cell lines. We defined this area broadly asdistal to 3pl4 and proximal to 3p25 in agreement with our (notshown) and published cytogenetic observations (11-13). Thislarge portion of chromosome 3p is known to harbor TSGsimportant in the development of common human cancers ofthe kidney, lung, breast, ovary, and cervix (7, 8), and the vonHippel Lindau TSG (9). The dramatic loss of heterozygosityon 3p from this large region first shown in the present worksuggests that TSGs located in this region of 3p may be important in the pathogenesis of HNSCC.

In relation to tumor stage, it should be emphasized that the1454

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

3p DELETION IN HEAD AND NECK CANCER

07

0.6 h

0.5

0.3

02

0.1

00

LOCUSFig. 3. Comparisons of per locus heterozygosity for normal and tumor cells

for 22 scored chromosome 3p loci. TUMOR, observed frequency of heterozygosityfound in tumor cells; NORMAL, FOH derived from natural populations; **, locithat showed significant reduction of heterozygosity. Table 4 shows the P valuesassociated with each comparison. •¿�.tumor, ES.normal.

HNSCC cell lines used in our study were derived from latestage I and early stage II disease at which point there are nodiscernible nodal (N) or distant métastases(M) (Ti_2, N0, M0).This would suggest that 3p TSGs might be most important tothe initiation of the tumorigenic process and/or early development of HNSCC. Obviously, analysis of a greater number ofearly (and late stage) forms of HNSCC is required to confirm3p involvement in initiation and progression of head and neckcancers.

The present paper outlines a novel statistical method toascertain alleile losses in tumor DNA samples when the absenceof normal matching tissues precludes the use of previous methods that utilized comparisons of normal and tumor genotypesof RFLP marker loci. Chromosome-specific allelic loss for a

high density marker allelotype can easily be computed usingthis approach (see "Appendix"). We believe that this method

will prove important in analyzing allelotypes of tumor cell linesthat have been established earlier. However, further work isnecessary to examine the robustness of the method, especiallywith regard to high resolution mapping.

In summary, we have detected 3p allelic loss in a large sample(18-26) of HNSCC cell lines using a novel statistical approachto ascertain loss of heterozygosity. A high density set of 3pRFLP markers allows us to define the commonly deleted regionas 3pl4 to p26. Further work will be aimed at mapping thedeleted TSG sites with better precision and trying to findcorrelates with tumor stage and behavior.

ACKNOWLEDGMENTS

We are grateful to Drs. B. Zbar and G. Alvord for support and Dr.J. Rowley's critical reading of the manuscript.

APPENDIX: A Probability Function for the Expected Numberof Heterozygous Loci per Individual

M. Fivash

locus heterozygosity with the assumptions that all the loci are in HardyWeinberg equilibrium and there is random association between the loci.This problem has been examined by Mitton and Pierce (23) through asimulation study. The results of their simulation study were confirmedanalytically by Chakraborty (24). Chakraborty presented a recursionfor calculating the distribution function for the expected number ofheterozygous loci per individual. The distribution function then yieldsthe probability function through successive differences. Calculation ofthe distribution function and extraction of successive differences may,however, be avoided by a recursion which directly yields the probabilityfunction.

Denote by h¡frequency of heterozygosity of the genotype at the ithlocus. Then the probability that none of the loci express a heterozygousgenotype is

- *<) (A)

Denote the total number of loci by m and let .v be the number ofheterozygous loci. Define the following quantities used to calculate therecursion.

Pi = Pr(x = i) r0, = Po (B)

Then the probability that some number, /, of the loci express heterozygous genotypes in an individual may be found through the recursionrelations (21).

The probability function for the number of heterozygous loci perindividual (of those used in this analysis) is calculated from the per

(C)

The first sum in Equations C need not be actually calculated becauseall but one term of this sum is calculated for the first component of thevector ;•,,,-A PC program is available from the author which imple

ments this recursion.When all the //, are equal the probability function reduces to the

familiar binomial probability function.The distribution of the number of heterozygous loci per individual

in a random sample was calculated for the normal population from theper locus FOH and for the tumor subpopulation from an estimate ofthe per locus FOH based on the alíelerepresentation at each locus intumor DNA. Ax2 goodness of fit test showed that both distributions

may be considered the same.The expected number, x, of heterozygous loci per individual and the

variance of .v may be determined from the probability function for thenormal population by Equations D and C.

(D)

The expected and observed number of heterozygous loci per individual may be compared by finding the probability that as few or fewerheterozygous loci per individual will be observed in the tumor panel bychance if it is assumed that the tumor panel has the same distributionof heterozygous loci per individual as the normal population. Thisprobability can be calculated based on the distribution of the numberof heterozygous loci per individual in the normal population and isgiven by

(E)

where g¡is the number of heterozygous loci in the ith individual. ThisP value is 5.4 x IO"58for the tumor subpopulation (17 cell lines and

19 loci) considered here. Such a small probability for this overall testsuggests that many tumor subpopulation loci have moved in the direction of reduced heterozygosity.

1455

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

3p DELETION IN HEAD AND NECK CANCER

REFERENCES

1. Jacobs, C. (ed.). Carcinomas of the Head and neck: Evaluation and Management. London: Kluwer Academic Publishers, 1990.

2. Snow, G. B. (ed.). Head and Neck Carcinomas. Current Opinions in Oncology, Vol. 2, pp. 537-589, 1990.

3. Jacobs, C, Goffinet, D. R., and Fee, W. E. Head and Neck SquamousCancers. Current Problems in Cancer, Vol. 14, Nl, Chicago: Year BookMedical Publishers, Inc., January/February, 1990.

4. Franceschi, S., Talamini, R., Barra, S., Barón,A. E., Negri, E., Bidoli, E.,Serraino, D., and La Vecchia, C. Smoking and drinking in relation to cancersof the oral cavity, pharynx, larynx, and esophagus in northern Italy. CancerRes., 50: 6502-6507, 1990.

5. Gencik, K . Wey, W., and Muller, H. High incidence of laryngeal carcinomain a Swiss family. J. Otorhinol. Laryngol., 48: 162-166, 1986.

6. Knudson, A. Genetics and etiology of human cancer. Adv. Hum. Genet., 8:1-66, 1977.

7. Geiser, A. G., and Stanbridge, E. J. A review of the evidence for tumorsuppressor genes. Crit. Rev. Oncogenesis, /: 261-276, 1989.

8. Zbar, B. Chromosomal deletions in lung cancer and renal cancer. In: V. T.DeVita, Jr., S. 1Miman, and S. A. Rosenberg (als.). Important Advances inOncology 1989, pp. 41-60. Philadelphia: J. B. Lippincott Co., 1989.

9. Hosoe, S., Brauch, H., Latif, F., Glenn, G., Daniel, L., Bale, S., Choyke, P.,Gorin, M-. Oldfield, E., Herman, A., Goodman, J., Orcutt, M. I... Hampsch,K., Delisio, J., Modi, W., McBride, W., Anglard, P., Weiss, G., Walther, M.M., Linchan, W. M., 1.crinan, M. I., and Zbar, B. Localization of the vonHippel-Lindau disease gene to a small region of chromosome 3. Genomics,«.•634-640,1990.

10. Yokota, J., Tsukada, Y., Nakajima, T., Gotoh, M., Shimosato, Y., Mori, N.,Tsunokawa, Y., Sugimura, T., and Terada, M. Loss of heterozygosity on theshort arm of chromosome 3 in carcinoma of the uterine cervix. Cancer Res.,«.•3598-3601,1989.

11. Hauser-Urfer, I. F., and Stauffer, J. Comparative chromosome analysis ofnine squamous cell carcinoma cell lines from tumors of the head and neck.Cytogenet. Cell Genet., 39: 35-39, 1985.

12. Sacks, P. G., Parnés,S. M., Gallick, G. E., Lichtner, R., Pathak, S., andParsons, D. F. Establishment and characterization of two new squamous cellcarcinomas cell lines derived from tumors of the head and neck. Cancer Res.,48: 2858-2866, 1988.

13. Heo, D. S., Snyderman, C., Gollin, S. M., Pan, S., Walker, E., Deka, R.,

Barnes, E. L., Johnson, J. T., Herberman, R. B., and Whiteside, T. L.Biology, cytogenetics, and sensitivity to immunological effector cells of newhead and neck squamous cell carcinoma lines. Cancer Res., 49: 5167-5175,1989.

14. 1.orinan, M. I., Latif, F., Glenn, G. M., Daniel, L. N., Brauch, H., Hosoe,S., Hampsch, K., Delisio, J., Orcutt, M. L., McBride, O. W., Grzeschik, K-H., Takahashi, T., Minna, J., Anglard, P., Linehan, W. M., and Zbar, B.Isolation and regional localization of a large collection (2000) of single copyDNA fragments on human chromosome 3 for the cloning of the renal cellcarcinoma and small cell lung carcinoma tumor suppressor genes. Hum.Genet., 86: 567-577, 1991.

15. Naylor, S. L., and Bishop, D. T. Report of the committee on the geneticconstitution of chromosome 3. Cytogenet. Cell Genet., 5/: 106-120, 1989.

16. Weichselbaum, R. R., Dahlberg, W., Beckett, M., Karrison, T., Miller, D.,Clark, J., and Ervin, T. J. Radiation-resistant and repair-proficient tumorcells may be associated with radiotherapy. Failure in head-and-neck cancerpatients. Proc. Nati. Acad. Sci. USA, 83: 2684-2688, 1986.

17. Weichselbaum, R. R., Dunphy, E. J., Beckett, M. A., Tybor, A. G., Moran,W. Y., Goldman, M. E., Vokes, E. E., and Pamyl, W. R. Epidermal growthfactor receptor gene amplification and expression in head and neck cancercell lines. Head & Neck; September/October, 437-442, 1989.

18. Southern, E. Detection of specific sequences among DNA fragments separated by gel electrophoresis. J. Mol. Biol., 98: 503-517, 1975.

19. Feinberg, A. P., and Vogelstein, B. A technique for radiolabelling DNArestriction endonuclease fragments to high specific activity. Anal. Biochem.,132:6-13, 1983.

20. Cowan, J. M., Beckett, M. A., Tarbell, N., and Weichselbaum, R. R. Symmetrical chromosome rearrangements in cell lines established from humanradiation-induced sarcomas. Cancer Genet. Cytogenet., SO: 125-137, 1990.

21. Larson, H. J. Introduction to Probability Theory and Statistical Inference.New York: John Wiley & Sons, 1969.

22. Vogelstein, B., Fearon, E. R., Kern, S. E., Hamilton, S. R., Preisinger, A.C., Nakamura, Y., and White, R. Allelotype of coloréela!carcinomas. Science(Washington DC), 244: 207-211, 1989.

23. Milton, J. B., and Pierce, B. A. The distribution of individual heterozygosityin natural populations. Genetics, 95: 1043-1054, 1980.

24. Chakraborty, R. Distribution of the number of heterozygous loci in anindividual in natural populations. Genetics, 98:461-466, 1981.

1456

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from

1992;52:1451-1456. Cancer Res   Farida Latif, Matthew Fivash, Gladys Glenn, et al.   Statistical Ascertainment of Allelic LossChromosome 3p Deletions in Head and Neck Carcinomas:

  Updated version

  http://cancerres.aacrjournals.org/content/52/6/1451

Access the most recent version of this article at:

   

   

   

  E-mail alerts related to this article or journal.Sign up to receive free email-alerts

  Subscriptions

Reprints and

  .pubs@aacr.orgDepartment at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

  Permissions

  Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/52/6/1451To request permission to re-use all or part of this article, use this link

Research. on January 29, 2018. © 1992 American Association for Cancercancerres.aacrjournals.org Downloaded from