7
J. Pathol. 185: 382–388 (1998) COMPARATIVE GENOMIC HYBRIDIZATION REVEALS FREQUENT CHROMOSOME 13q AND 4q LOSSES IN RENAL CARCINOMAS WITH SARCOMATOID TRANSFORMATION 1 , 1 *, 1 , 1 , 2 , 1 , 3 , 1 . 1 1 Institute of Pathology, University of Basel, 4003 Basel, Switzerland 2 Urologic Clinics, University of Basel, 4003 Basel, Switzerland 3 Division of Scientific Photography, University of Basel, Switzerland SUMMARY Renal cell carcinomas (RCCs) with sarcomatoid transformation show the most malignant behaviour of all renal carcinoma types. In this study, comparative genomic hybridization was used to screen for losses and gains of DNA sequences along all chromosome arms in 12 sarcomatoid (S) RCCs. On average, there were 8·6 aberrations per tumour. DNA sequence losses (5·2 &4·4) were slightly more frequent than gains (3·4 &2·6). DNA gains most often involved chromosomes 17 (33 per cent), 7, and 8q (25 per cent each). High-level co-amplification involving 11q22–23 and 7p21–22 in one SRCC was not present in adjacent non-sarcomatous tumour areas, raising the possibility of oncogene involvement at these loci for sarcomatoid transformation. DNA losses were most prevalent at 13q (75 per cent) and 4q (50 per cent), suggesting that inactivation of tumour suppressor genes at chromosomes 13q and 4q may be linked to sarcomatoid growth of RCC. It is concluded that SRCCs are genetically highly complex. Chromosomes 13q, 4q, 7p21–22, and 11q22–23 may carry genes with relevance for sarcomatoid growth in RCC. ? 1998 John Wiley & Sons, Ltd. KEY WORDS—sarcomatoid renal cell carcinoma; molecular cytogenetics; comparative genomic hybridization; chromosome 13q; chromosome 4q INTRODUCTION Renal cell carcinomas (RCCs) sometimes show sarco- matoid transformation, resulting in heterogeneous RCCs with carcinomatous and sarcomatous compo- nents. Sarcomatoid renal cell carcinomas (SRCCs) are highly aggressive neoplasms with an extremely poor prognosis. Virtually all patients with SRCC die within 5 years, and the 2-year survival rate is less than 20 per cent. 1–3 Genetic alterations present in SRCC might therefore pinpoint loci carrying genes of which a dis- turbed function can contribute to an aggressive pheno- type in RCC patients. Little is known, however, about the genetic aberrations in SRCC. Recent cytogenetic and molecular studies of RCC have demonstrated a number of chromosomal loci as being important in cancer devel- opment and progression, including 3p, 6q, 8p, 9pq, and 14q. 4–8 Recently, mutations of the p53 gene on chromo- some 17p were found in 11 of 14 SRCCs, 9 suggesting that p53 alterations are associated with sarcomatoid transformation in RCC. So far, there have been no systematic cytogenetic or allelotype studies on the genetics of SRCC. In the present study, to characterize the genetic changes in SRCC, comparative genomic hybridization (CGH) was applied to a set of 12 SRCC cases. CGH is based on hybridization of dierentially labelled DNAs, one from the tumour and another from the normal reference to normal metaphase spreads, allowing one to screen for DNA sequence losses and gains along all chromosome arms. 10 In a previous study, we have analysed 41 clear cell RCCs by CGH. 11 DNA losses were most frequently observed on chromosomes 3p (56 per cent), 6q (22 per cent), 13q (24 per cent), and 8p, 14q, and Xq (all 20 per cent) and gains at chromosomes 5q (17 per cent) and 7 (15 per cent). No high-level DNA amplifications were seen. MATERIALS AND METHODS Tumour samples All slides of 623 RCC specimens obtained from radical nephrectomies at the Institute of Pathology, University of Basel, between 1970 and 1994 were reviewed for the presence of sarcomatoid areas. Sarco- matoid areas were identified morphologically. For inclu- sion in this study, the sarcomatous component had to involve an area greater than 80 per cent of the tumour. Tumour tissue blocks that contained pure sarcomatoid *Correspondence to: Holger Moch, MD, Institut für Pathologie der Universität Basel, Schönbeinstrasse 40, CH-4003 Basel, Switzerland. Contract grant sponsor: Swiss National Science Foundation; Contract grant number: 3200-043969.95. Contract grant sponsor: Krebsforschung Schweiz; Contract grant number: 367-9-1996. CCC 0022–3417/98/040382–07 $17.50 ? 1998 John Wiley & Sons, Ltd. Received 12 May 1997 Accepted 20 March 1998

Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

Embed Size (px)

Citation preview

Page 1: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

J. Pathol. 185: 382–388 (1998)

COMPARATIVE GENOMIC HYBRIDIZATION REVEALSFREQUENT CHROMOSOME 13q AND 4q LOSSES IN

RENAL CARCINOMAS WITH SARCOMATOIDTRANSFORMATION

1, 1*, 1, 1, 2, 1, 3, 1 . 1

1Institute of Pathology, University of Basel, 4003 Basel, Switzerland2Urologic Clinics, University of Basel, 4003 Basel, Switzerland

3Division of Scientific Photography, University of Basel, Switzerland

SUMMARY

Renal cell carcinomas (RCCs) with sarcomatoid transformation show the most malignant behaviour of all renal carcinoma types. Inthis study, comparative genomic hybridization was used to screen for losses and gains of DNA sequences along all chromosome arms in12 sarcomatoid (S) RCCs. On average, there were 8·6 aberrations per tumour. DNA sequence losses (5·2&4·4) were slightly morefrequent than gains (3·4&2·6). DNA gains most often involved chromosomes 17 (33 per cent), 7, and 8q (25 per cent each). High-levelco-amplification involving 11q22–23 and 7p21–22 in one SRCC was not present in adjacent non-sarcomatous tumour areas, raising thepossibility of oncogene involvement at these loci for sarcomatoid transformation. DNA losses were most prevalent at 13q (75 per cent)and 4q (50 per cent), suggesting that inactivation of tumour suppressor genes at chromosomes 13q and 4q may be linked to sarcomatoidgrowth of RCC. It is concluded that SRCCs are genetically highly complex. Chromosomes 13q, 4q, 7p21–22, and 11q22–23 may carrygenes with relevance for sarcomatoid growth in RCC. ? 1998 John Wiley & Sons, Ltd.

KEY WORDS—sarcomatoid renal cell carcinoma; molecular cytogenetics; comparative genomic hybridization; chromosome 13q;chromosome 4q

INTRODUCTION

Renal cell carcinomas (RCCs) sometimes show sarco-matoid transformation, resulting in heterogeneousRCCs with carcinomatous and sarcomatous compo-nents. Sarcomatoid renal cell carcinomas (SRCCs) arehighly aggressive neoplasms with an extremely poorprognosis. Virtually all patients with SRCC die within 5years, and the 2-year survival rate is less than 20 percent.1–3 Genetic alterations present in SRCC mighttherefore pinpoint loci carrying genes of which a dis-turbed function can contribute to an aggressive pheno-type in RCC patients. Little is known, however, aboutthe genetic aberrations in SRCC. Recent cytogenetic andmolecular studies of RCC have demonstrated a numberof chromosomal loci as being important in cancer devel-opment and progression, including 3p, 6q, 8p, 9pq, and14q.4–8 Recently, mutations of the p53 gene on chromo-some 17p were found in 11 of 14 SRCCs,9 suggestingthat p53 alterations are associated with sarcomatoidtransformation in RCC. So far, there have been no

*Correspondence to: Holger Moch, MD, Institut für Pathologie derUniversität Basel, Schönbeinstrasse 40, CH-4003 Basel, Switzerland.

Contract grant sponsor: Swiss National Science Foundation;Contract grant number: 3200-043969.95.

Contract grant sponsor: Krebsforschung Schweiz; Contract grantnumber: 367-9-1996.

CCC 0022–3417/98/040382–07 $17.50? 1998 John Wiley & Sons, Ltd.

systematic cytogenetic or allelotype studies on thegenetics of SRCC.

In the present study, to characterize the geneticchanges in SRCC, comparative genomic hybridization(CGH) was applied to a set of 12 SRCC cases. CGH isbased on hybridization of differentially labelled DNAs,one from the tumour and another from the normalreference to normal metaphase spreads, allowing one toscreen for DNA sequence losses and gains along allchromosome arms.10 In a previous study, we haveanalysed 41 clear cell RCCs by CGH.11 DNA losseswere most frequently observed on chromosomes 3p (56per cent), 6q (22 per cent), 13q (24 per cent), and 8p,14q, and Xq (all 20 per cent) and gains at chromosomes5q (17 per cent) and 7 (15 per cent). No high-level DNAamplifications were seen.

MATERIALS AND METHODS

Tumour samples

All slides of 623 RCC specimens obtained fromradical nephrectomies at the Institute of Pathology,University of Basel, between 1970 and 1994 werereviewed for the presence of sarcomatoid areas. Sarco-matoid areas were identified morphologically. For inclu-sion in this study, the sarcomatous component had toinvolve an area greater than 80 per cent of the tumour.Tumour tissue blocks that contained pure sarcomatoid

Received 12 May 1997Accepted 20 March 1998

Page 2: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

383CGH IN SARCOMATOID RENAL CARCINOMA

tumour were selected for DNA extraction. To enrich fortumour, all tumour blocks were trimmed as necessary bycutting normal tissue away with a scalpel.

DNA preparation

Five-micrometer sections were cut from tumourblocks and stained with haematoxylin and eosin (H&E)to ensure that the average tumour cell content of thesections was greater than 75 per cent. DNA was isolatedfrom 25 15 ìm thick paraffin sections. Sections weredeparaffinized and suspended in DNA extraction buffercontaining 0·5 mg/ml proteinase K. Additional protein-ase K was added 24 and 48 h later, for a total incubationtime of 72 h. DNA extraction and labelling were doneessentially as described before.11 Briefly, 1 ìg of tumourDNA was nick-translated using a commercially avail-able kit (BioNick kit; Life Technologies). Fluorescein-12-labelled dUTP and Texas-Red-5-labelled dUTP(DuPont, Boston, MA, U.S.A.) were used for directlabelling of tumour and normal DNA (extracted frommononuclear cells of healthy volunteers).

CGH and digital image analysis

Fluorescein-labelled tumour DNA (200 ng), 200 ng ofTexas Red-labelled reference DNA, and 20 ìg ofunlabelled human Cot-1 DNA (GIBCO BRL, LifeTechnologies; Gaithersburg, MD, U.S.A.) were hybrid-ized to normal metaphase spreads (VYSIS Inc.,Downers Grove, IL, U.S.A.). Hybridization was allowedto proceed for 3 days at 37)C in a moist chamber.Post-hybridization washes were as previously de-scribed.11 Digital images (4,6-diamidino-2-phenylindol,FITC, and Texas Red) were collected from six to sevenmetaphases using a Photometrics cooled CCD camera(Microimager 1400; Xillix Technologies, Vancouver,British Columbia, Canada) and a Sun workstation. theVYSIS software program (VYSIS Inc., Downers Grove,IL, U.S.A.) was used to calculate average green to redfluorescence ratio profiles for each chromosome. At leastfour observations per autosome and two observationsper sex chromosome were included in the analysis.

Controls and threshold definition

Each CGH experiment included a normal cell linewith known aberrations (positive control) and a hybrid-ization of two differentially labelled sex-mismatchednormal DNAs (negative control). The cut-off valuesused for definition of the DNA sequence copy numbergains and losses were based on a series of ten controlexperiments comparing two differentially labelled nor-mal DNA samples of formalin-fixed tissue. Gains ofDNA sequences were defined as chromosomal regionswhere the mean green to red fluorescence ratio wasabove 1·15, whereas losses were defined as regions wherethe mean was below 0·85. To define an aberration, it wasadditionally required that the first standard deviationwas above (gain) or below (loss) 1·0. Telomeric andheterochromatic regions were excluded from the analy-sis. Overrepresentations were considered as high-level

? 1998 John Wiley & Sons, Ltd.

amplifications when the fluorescence ratio valuesexceeded 1·5 in a subregion of a chromosome arm.12 Itshould be emphasized that the green to red ratio profilesdo not provide information on absolute copy numbers,such as the level of gene amplification. In controlhybridizations of normal tissue, the mean green to redratio occasionally exceeded the fixed 1·15 cut-off level atG–C-rich chromosomal regions known to produce false-positive results by CGH, including 1p32-pter, 16p, 19,and 22. These regions were also excluded from allanalyses.

RESULTS

A total of 12 SRCCs were included in this study.According to the TNM staging system (InternationalUnion Against Cancer), three tumours were stage 4,seven were stage 3, and two were stage 2. The diametersof the tumours ranged from 4 to 16 cm (median 10 cm).The patient ages ranged from 34 to 84 years (median70·5 years). A detailed description of histological find-ings is given in Table I. The sarcomatoid componentshowed marked nuclear atypia in all tumours. Consider-able morphological heterogeneity has been observedwithin the tumours. Carcinomatous components weresometimes distinctly separated from the sarcomatoidcomponent. Most cases contained sarcomatoid areaswith pleomorphic cells. The sarcomatoid componentwas characterized by interlacing fascicles of pleomorphicand spindle cells in three cases.

Follow-up was available in 11 of 12 patients. All 11patients died of disease within 16 months (mean survival5·5&4·0 months).

A summary of all DNA sequence copy number aber-rations detected in this series of 12 SRCCs by CGH isillustrated in Fig. 1. One tumour revealed no DNAsequence copy number alterations by CGH. On average,there were 8·6 aberrations per tumour (range 0–20): 5·2losses (range 0–15) and 3·4 gains (range 0–9).

Chromosomal regions that were most often lostincluded 3p (33 per cent), 4q (50 per cent), 6q, 9p, 11q,18q, Xq (25 per cent each), as well as 13q (75 per cent).Losses affecting chromosome 3 were detected in fourtumours (33 per cent). Three of these were losses of thep-arm, whereas one tumour also showed loss of parts ofthe q-arm. Two had a loss of the entire chromosome 4.Four tumours lost parts of the 4q arm. The minimalcommon region of loss was narrowed down to 4q28.DNA sequence losses involving chromosome 9 werepresent in 5 of 12 tumours (42 per cent), involving 9p inthree and 9q in two. Chromosome 13q was involved innine tumours (75 per cent). Loss of the entire 13q armwas found in six. Eight of nine tumours with 13qloss included the 13q12–14 region, the locations ofthe retinoblastoma (Rb) and the BRCA-2 genes,respectively.

Increased DNA sequence copy number was mostoften detected at chromosomes 7, 8q, 12q (25 per centeach), and 17 (33 per cent). Almost always, the entirelong arms of the chromosomes were gained. Twotumours showed regional gains at 11q13. The histo-pathological data and the chromosomal gains and

J. Pathol. 185: 382–388 (1998)

Page 3: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

384 F. JIANG ET AL.

losses are shown in Table I. The number of cases wastoo small to examine genetic differences betweenthe spindle and pleomorphic cells of sarcomatoidcomponents.

High-level amplifications (green to red ratio above1·5) were found in one tumour (case 6). This showed aco-amplification of 11q22–23 and 7p21–22. These ampli-fications were confirmed in two control hybridizations.From this tumour, a tissue block containing non-sarcomatous carcinoma areas had been retained andcould be investigated by CGH. This unique tumour areashowed six identical aberrations (DNA losses of 5p, 5q,13q, Xq and gains of 7p and 8q), but lacked the two geneamplifications as well as DNA losses at chromosomes2q, 3p, 10, and Xp (Fig. 2).

DISCUSSION

In this study, CGH was used to characterize genomicalterations in SRCC. It showed that SRCCs are geneti-cally highly complex. The result of this study suggeststhat 13q deletions are the most prevalent alteration inSRCC.

Target tumour suppressor genes that could be inacti-vated through 13q deletion include the retinoblastomagene (Rb) and the BRCA-2 gene, on 13q14 and 13q12,respectively. Involvement of BRCA-2 has been impli-cated in various tumours, but alterations of the BRCA-2gene have not been investigated in RCC. Altered Rbprotein expression were found to be rare in primaryRCC.13 Interestingly, deletions and rearrangements ofthe Rb gene have been detected in human sarcomas andaltered expression levels of the Rb gene product havebeen shown in up to 70 per cent of primary sarcomas.14

It has therefore been suggested that alterations of the Rbgene may be associated with the development of humansarcomas. One might speculate that 13q loss is also

? 1998 John Wiley & Sons, Ltd.

linked to sarcomatoid transformation in RCC, but thisassumption is not supported by our findings in one case,in which both the sarcomatoid and the carcinomatousarea displayed a 13q deletion. Future studies on SRCCwith a small sarcomatous component (<50 per cent), inwhich the carcinomatous and sarcomatous parts can bebetter separated than in our tumour set, must showwhether 13q deletions are consistently present in thecarcinomatous and sarcomatous components.

Previous studies examining loss of heterozygosity(LOH) have shown that 13q allelic imbalances can occurin RCC. However, the rates of LOH (6–33 per cent) inRCC without sarcomatous parts were comparablylow.13,15–17 Recently, we observed chromosome 13qlosses in 24 per cent in a different independent tumourset of 41 clear cell RCCs by CGH.11 The high prevalenceof 13q losses in SRCC is consistent with a role of 13qdeletions in RCC progression. Interestingly, LOH at 13qwas also observed in 50–100 per cent of collecting ductcarcinomas (CDCs) of the kidney,18,19 another rarehistological subtype of RCC with a particularly aggres-sive behaviour. This raises the possibility that inactiva-tion of a gene at 13q might be relevant for RCCprogression. This hypothesis would also be consistentwith the higher frequency of 13q deletions in high-stageRCC than in low-stage tumours, as found by Anglardet al.15

Whereas the frequency of 6q, 9p, and 11q deletionswas in the range of previous RFLP and CGHstudies,11,20,21 chromosome 4q has not been reported tobe frequently deleted in renal cancer. Losses of 4q wereobserved in 5–8 per cent by LOH studies17,22 and in 10per cent by CGH.11 Although the number of tumoursexamined in this study was low, the frequent finding of4q losses (6 of 12 tumours) would be consistent with arole of a tumour suppressor gene on 4q for tumourprogression in RCC. Studies of colon, cervix, andhepatocellular carcinoma have already suggested the

Table I—Sumary of the gains, losses, and amplifications of DNA sequence copy number detected with CGH in 12 SRCCs

CaseNo.

Cell type insarcomatoid

area*

Cell type incarcinoma

area† Copy number gains Copy number losses Amplifications

1 pleo mix 5p, 7q, 8q, 4q, 8p, 13q, 17p2 pleo/sp mix, pap 3q, Xp, Xq 9q, 13q3 pleo mix 1q, 5q, 7q 3p, 9p, 13q, 14q4 sp chromo 14q, 16q Xp, Xq5 pleo eos 10p, 17q, 20q 3p, 4q, 6q, 13q6 sp chromo 2p, 7p, 8q, 11q, 12q, 17p, 17q 3p, 4p, 4q, 5p, 5q, 10p, 10q, 13q,

18q, 20p, Xq7p21–22, 11q22–23

7 pleo mix 10p, 11q, 12q, 17p, 17q, 20q 1p, 3p, 4q, 6q, 9p, 11q, 13q, 18q8 sp cl 9q, 12q, 16p, 17p, 17q, Y 1p, 1q-, 2q, 5q, 6q, 8q, 9p, 11p, 11q,

12q, 13q, 14q, 18q, Xq9 pleo/sp mix 3q, 8q, 16q, 17p, Xq 4p, 4q, 8p, 11q, 13q, Xp, Y

10 pleo eos 2q, 13q11 pleo eos12 sp/pleo mix 7p, 7q 9q, Y

*sp=spindle cells; pleo=pleomorphic cells.†cl=clear cell type; eos=eosinophilic cell type; mix=clear and eosinophilic cell types; chromo=chromophobe cell type; pap=papillary growth

pattern.

J. Pathol. 185: 382–388 (1998)

Page 4: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

idization. Losses are shown on the left and gains on the right of

385C

GH

INS

AR

CO

MA

TO

IDR

EN

AL

CA

RC

INO

MA

?1998

JohnW

iley&

Sons,L

td.J.

Pathol.

185:382–388

(1998)

Fig. 1—Summary of the gains and losses of DNA sequences observed in 12 SRCCs by comparative genomic hybrthe chromosomes. High-level amplifications are indicated by thick lines

Page 5: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

386 F. JIANG ET AL.

Fig

.2—

Exa

mpl

esof

the

gree

nto

red

rati

opr

ofile

sfo

ra

sarc

omat

oid

area

(A)

and

aca

rcin

omat

ous

area

(B)

ofon

eSR

CC

(tum

our

6;H

&E

stai

ning

,#

280)

.Id

enti

cal

chro

mos

omal

chan

ges

inbo

thtu

mou

rar

eas

are

show

nw

ithi

nbo

xes.

Arr

ows

indi

cate

high

-lev

elam

plifi

cati

ons.

The

mea

ngr

een

tore

dflu

ores

cenc

era

tio

profi

le(t

hick

line)

and

its

1SD

(thi

nlin

e)ar

esh

own

for

chro

mos

omes

from

pter

toqt

er.

The

chro

mos

ome

iden

tific

atio

nis

show

nbe

low

the

profi

le.

Bar

son

the

left

ofth

ech

rom

osom

eid

eogr

ams

indi

cate

DN

Alo

sses

and

bars

onth

eri

ght

indi

cate

gain

s

? 1998 John Wiley & Sons, Ltd. J. Pathol. 185: 382–388 (1998)

Page 6: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

387CGH IN SARCOMATOID RENAL CARCINOMA

presence of tumour suppressor genes on 4q.23–25 Theminimal common region of loss (4q28) did not appear toinclude the autosomal dominant polycystic kidneydisease (ADPKD 2) gene locus at 4q13–23.26

Bennington and Beckwith27 considered the sarcoma-toid portion to be derived from the stroma of RCC. Thefact that small sarcomatoid tumours which could beinterpreted as early SRCC cannot be found arguesagainst a de novo development of SRCC. Most authorstherefore regard SRCC as a metaplastic transformationof other carcinoma types (e.g., clear cell, papillary,chromophobe or collecting duct RCC).28,29 It has beenproposed that molecular analyses may help to assignSRCC to a special renal tumour subtype, when nocarcinomatous areas are detected.29 One candidate for agenetic lesion that can help to distinguish RCC subtypesare deletions of chromosome 3p, since LOH for chro-mosome 3p is detectable in the majority of clear cellRCCs,30 in some chromophobe RCCs,31 but not inpapillary RCC.32 It is therefore possible that some ofour cases without 3p losses might be derived frompapillary RCC. It cannot be excluded, however, thatsome 3p deletions were not detected in this study, sincedeletions less than 10 MB are usually missed by CGH.

DNA gains most often involved chromosome 17 (33per cent). Previous studies have shown that amplifi-cation and overexpression of the HER-2/neu gene onchromosome 17q21–22 play a minor role in the onco-genesis of RCC.33–35 In most tumours, relative copynumber gains detected by CGH affected large areas ofthe genome, in general at least a chromosome arm. Thebiological meaning of such large overrepresentedgenome fragments is unclear. While it is possible thatthey lead to an overexpression of one or several genes, arelationship between such large genome overrepresen-tations and gene overexpression has not yet beendemonstrated. Interestingly, two tumours had smallsubregional gains at 11q13, where the PRAD1/(CCND1)/cyclin D1 gene is located. Future studies willhave to evaluate whether cyclin D1, which is frequentlyamplified in many other tumour types,36,37 is involved inSRCC.

High-level amplifications, even when detected infre-quently, may highlight locations of dominant oncogenesinvolved in tumour progression. In this set of SRCC,high-level co-amplification was found in 1 of 12 tumoursat 7p21–22 and 11q22–23. Since these amplificationswere not detectable in adjacent tumour areas withnon-sarcomatous components, it is possible that theactivation of oncogenes at these loci might have contrib-uted to sarcomatoid transformation in this tumour.While there are no strong candidate oncogenes on7p21–22, several potential oncogenes exist on 11q22–23,including the ETS1 gene.38 Interestingly, ETS1 proteinsoccur in fibroblasts and endothelial cells of invasivetumours and can activate collagenase 1 genes. It hasbeen shown that ETS1 is expressed in sarcomas.38

In summary, our data indicate that SRCCs are geneti-cally complex, with a high average number of aberra-tions. The inactivation of a tumour suppressor geneat chromosomes 13q and 4q may be important forsarcomatoid growth of RCC.

? 1998 John Wiley & Sons, Ltd.

ACKNOWLEDGEMENTS

We thank Hedvika Novotny, Martina Storz, and thestaff of the Institute for Pathology, University of Basel,for their technical support. We are grateful to LucasRosenthaler and Armin Wittmann (Division of Scien-tific Photography, University of Basel) for assistancewith the computer systems. We also thank R. B.McGandy for comments on this manuscript. Thisstudy was supported by the Swiss National ScienceFoundation (3200-043969.95) and KrebsforschungSchweiz (367-9-1996).

REFERENCES1. Ro J, Ayala A, Sella A, Samuels M, Swanson D. Sarcomatoid renal cell

carcinoma: clinicopathologic. Cancer 1987; 59: 516–526.2. Bertoni F, Ferri C, Benati A, Bacchini P, Corrado F. Sarcomatoid

carcinoma of the kidney. J Urol 1987; 137: 25–28.3. Tomera KM, Farrow GM, Lieber MM. Sarcomatoid renal carcinoma.

J Urol 1983; 130: 657–659.4. Presti J, Rao H, Chen Q, et al. Histopathological, cytogenetic, and

molecular characterization of renal cortical tumors. Cancer Res 1991; 51:1544–1552.

5. Presti J, Reuter V, Cordon-Cardo C, Mazumdar M, Fair W, Jhanwar S.Allelic deletions in renal tumors: histopathological correlations. Cancer Res1993; 53: 5780–5783.

6. Trash-Bingham CA, Salazar H, Freed JJ, Greenberg RE, Tartof KD.Genomic alterations and instabilities in renal cell carcinomas and theirrelationship to tumor pathology. Cancer Res 1995; 55: 6189–6195.

7. van den Berg E, van der Hout AH, Oosterhuis JW, et al. Cytogeneticanalysis of epithelial renal-cell tumors: relationship with a new histopatho-logical classification. Int J Cancer 1993; 55: 223–227.

8. Bugert P, Kovacs G. Molecular differential diagnosis of renal cell carci-nomas by microsatellite analysis. Am J Pathol 1996; 149: 2081–2088.

9. Oda H, Nakatsuru Y, Ishikawa T. Mutations of the p53 gene and p53overexpression are associated with sarcomatoid transformation in renal cellcarcinoma. Cancer Res 1995; 55: 658–662.

10. Kallioniemi A, Kallioniemi O, Sudar D, et al. Comparative genomichybridization for molecular cytogenetic analysis of solid tumors. Science1992; 258: 818–821.

11. Moch H, Presti JC 6Jr, Sauter G, et al. Genetic aberrations detected bycomparative genomic hybridization are associated with clinical outcome inrenal cell carcinoma. Cancer Res 1996; 56: 27–30.

12. Kallioniemi A, Kallioniemi OP, Citro G, et al. Identification of gains andlosses of DNA sequences in primary bladder cancer by comparativegenomic hybridization. Genes Chromosomes Cancer 1995; 12: 213–219.

13. Ishikawa J, Xu HJ, Hu SX, et al. Inactivation of the retinoblastoma gene inhuman bladder and renal cell carcinomas. Cancer Res 1991; 51: 5736–5743.

14. Cance W, Brennan M, Dudas M, Huang C, Cordon-Cardo C. Alteredexpression of the retinoblastoma gene product in human sarcomas. N EnglJ Med 1990; 323: 1457.

15. Anglard P, Tory K, Brauch H, et al. Molecular analysis of genetic changesin the origin and development of renal cell carcinoma. Cancer Res 1991; 51:1071–1077.

16. Bergerheim U, Nordenskjöld M, Collins V. Deletion mapping in humanrenal cell carcinoma. Cancer Res 1989; 49: 1390–1398.

17. Morita R, Ishikawa J, Tsutsumi M, et al. Allelotype of renal cell carcinoma.Cancer Res 1991; 51: 820–823.

18. Polascik TJ, Cairns P, Epstein JI, et al. Distal nephron renal tumors:microsatellite allelotype. Cancer Res 1996; 56: 1892–1895.

19. Schoenberg M, Cairns P, Brooks JD, et al. Frequent loss of chromosomearms 8p and 13q in collecting duct carcinoma (CDC) of the kidney. GenesChromosomes Cancer 1995; 12: 76–80.

20. Morita R, Saito S, Ishikawa J, et al. Common regions of deletion onchromosomes 5q, 6q and 10q in renal cell carcinoma. Cancer Res 1991; 51:5817–5820.

21. Cairns P, Tokino K, Eby Y, Sidransky D. Localization of tumor suppressorloci on chromosome 9 in primary human renal cell carcinomas. Cancer Res1995; 55: 224–227.

22. Thrash Bingham CA, Greenberg RE, Howard S, et al. Comprehensiveallelotyping of human renal cell carcinomas using microsatellite DNAprobes. Proc Natl Acad Sci USA 1995; 92: 2854–2858.

23. Ried T, Knutzen R, Steinbeck R, et al. Comparative genomic hybridizationreveals a specific pattern of chromosomal gains and losses during the genesisof colorectal tumors. Genes Chromosomes Cancer 1996; 15: 234–245.

24. Mitra AB, Murty VV, Li RG, Pratap M, Luthra UK, Chaganti RS.Allelotype analysis of cervical carcinoma. Cancer Res 1994; 54: 4481–4487.

J. Pathol. 185: 382–388 (1998)

Page 7: Comparative genomic hybridization reveals frequent chromosome 13q and 4q losses in renal carcinomas with sarcomatoid transformation

388 F. JIANG ET AL.

25. Yeh SH, Chen PJ, Lai MY, Chen DS. Allelic loss on chromosomes 4q and16q in hepatocellular carcinoma: association with elevated alpha-fetoprotein production. Gastroenterology 1996; 110: 184–192.

26. San Millan JL, Viribay M, Peral B, Martinez I, Weissenbach J, Moreno F.Refining the localization of the PKD2 locus on chromosome 4q by linkageanalysis in Spanish families with autosomal dominant polycystic kidneydisease type 2. Am J Hum Genet 1995; 56: 248–253.

27. Bennington J, Beckwith J. Tumors of the Kidney, Renal Pelvis and Ureter.Washington, DC: Armed Forces Institute of Pathology, 1975.

28. Zollinger H. Niere und ableitende Harnwege. In: Doerr W, Uehlinger E,eds. Spezielle Pathologische Anatomie. Berlin: Springer-Verlag, 1966; 693.

29. Kovacs G. Molecular differential pathology of renal tumours. Histopathol-ogy 1993; 22: 1–8.

30. Zbar B, Brauch H, Talmadge C, Linehan M. Loss of alleles of loci on theshort arm of chromosome 3 in renal cell carcinoma. Nature 1987; 327:721–727.

31. Kovacs A, Kovacs G. Low chromosome number in chromophobe renalcarcinoma. Genes Chromosomes Cancer 1992; 4: 267–268.

32. Kovacs G. Papillary renal cell carcinoma. A morphologic and cytogeneticstudy of 11 cases. Am J Pathol 1989; 134: 27–34.

? 1998 John Wiley & Sons, Ltd.

33. Rotter M, Block T, Busch R, Thanner S, Hofler H. Expression ofHER-2/neu in renal-cell carcinoma. Correlation with histologic subtypesand differentiation. Int J Cancer 1992; 52: 213–217.

34. Weidner U, Peter S, Strohmeyer T, Hussnatter R, Ackermann R, Sies H.Inverse relationship of epidermal growth factor receptor and HER2/neugene expression in human renal cell carcinoma. Cancer Res 1990; 50:4504–4509.

35. Lipponen P, Eskelinen M, Hietala K, Syrjanen K, Gambetta R. Expressionof proliferating cell nuclear antigen (PC10), p53 protein and c-erbB-2 inrenal adenocarcinoma. Int J Cancer 1994; 57: 275–280.

36. Speicher MR, Howe C, Crotty P, du Manoir S, Costa J, Ward DC.Comparative genomic hybridization detects novel deletions and amplifica-tions in head and neck squamous cell carcinomas. Cancer Res 1995; 55:1010–1013.

37. Foulkes WD, Campbell IG, Stamp GW, Trowsdale J. Loss of hetero-zygosity and amplification on chromosome 11q in human ovarian cancer.Br J Cancer 1993; 67: 268–273.

38. Sacchi N, Wendtner CM, Thiele CJ. Single-cell detection of ets-1 transcriptsin human neuroectodermal cells. Oncogene 1991; 6: 2149–2154.

J. Pathol. 185: 382–388 (1998)