- Email: [email protected]
Differential Chromosome Allelic Imbalance in the Progression of Human Prostate Cancer
0022-5347/96/1566-2079$03.00/0
THE JOURNAL OF U m o L o w Copyright 0 1996 by kwERI(.AN UROLOCICAL ASSOCIATION,
Vol. 156.2079-2083,December 1996 Printed in U.S.A.
INC.
DIFFERENTIAL CHROMOSOME ALLELIC IMBALANCE IN THE PROGRESSION OF HUMAN PROSTATE CANCER ALAIN LATIL, GEORGES FOURNIER, O L M E R CUSSENOT
AND
ROSETTE LIDEREAU*
From the Laboratoire d’Oncogenetique, Centre Rene Huguenin, St-Cloud, the Departement d’llrologie, H6pital & la Cavale Blanche, Brest, and the Departement d’Urologie, CHU Saint-Louis, Paris, France
ABSTRACT
Purpose: I t is widely accepted that a n accumulation of genetic alterations plays a n important role in the genesis of human cancers. We wished to obtain a comprehensive view of the role of genetic changes in prostate cancer. Materials and Methods: We screened 42 primary prostate tumors for allelic imbalance (AI) on 8 autosomal chromosome arms of interest (5q, 7q, 8p, lOq, 13q, 16q, 17q, 18q) by using 2 DNA probes for restriction fragment length polymorphism (RFLP) and 19 microsatellite markers (CA repeats). Results: The most frequent allelic imbalances were observed on 8p (58%)and 16q (53%).AI exceeding 20% was also observed a t sites on chromosome arms 7q (46%), 1Oq (23%), 13q (26%), 17q (34%)and 18q (39%),whereas AI was infrequent on 5q (10%). Conclusions: The data indicate that a relatively large number of chromosome loci play a part in the etiology and progression of this tumor type. Moreover, our findings suggest that inactivation of a putative tumor suppressor gene on 7q and 13q is a n early event in prostate tumorigenesis. In contrast, the close link between a n invasive phenotype and AI on 1Oq and 18q suggests that these genetic alterations occur late in prostate tumorigenesis. KEY WOROS:prostate cancer; allelic influence; loss of heterozygosity;tumor suppressor gene
cally localized prostate carcinomas underwent radical prostatectomy with lymph node excision. Thirteen patients had confined prostate tumors and twelve patients had local extracapsular extension. A radical prostatectomy specimen was first sliced thickly and samples of tissue were cut out of suspect areas. One part of the specimen was immediately placed in liquid nitrogen for high-molecular-weight DNA extraction, while adjacent sections of the initial specimen removed for molecular analysis were stained with hematoxylin and eosin to determine the proportion of tumor cells in each sample. A sample was considered suitable for DNA analysis if the proportion of tumor cells was 60% or more. Transurethral resection. The 12 patients who underwent this procedure had regional lymph node involvement (6 cases) or bone metastases (6 cases). Between 6 and 12 chips were obtained during resection. Specimens were selected for DNA extraction as described above. Needle biopsy. Samples were obtained in this way from 5 patients who had clinically localized prostate carcinomas with local extracapsular extention. The biopsy needle was introduced into suspect areas by the transrectal route using an automatic gun and a needle guide with digital control. Three 14-gauge biopsies were taken from the same site, placed in liquid nitrogen and checked histologically. Samples were selected for DNA extraction as described above. MATERIALS AND METHODS The histological diagnosis, Gleason scoreg and pathologic Patients and samples. Forty-two samples of prostate tu- tumor stage according to the tumor-nodes-metastasis classimors were obtained from patients undergoing surgery a t St fication of prostate cancerlo were determined in each case h i s Hospital (Paris)and La Cavale Blanche Hospital (Bmst). during a routine workup after surgery. The Gleason score for the 42 primary tumors ranged from The samples were obtained in 3 ways, as follows: Prostatectomy specimens. Twenty-five patients with clini- 5 to 9. By combining the Gleason score with the pathologic stage, we confirmed the well-established positive link be9 p t e d for publication April 30, 1996. Requests for reprints: Laboratoire d‘oncogenetique,Centre Rene tween grade and stage. The tumors were subdivided into Humenin, 35 rue Dailly, F-92211, St-Cloud, France. three groups corresponding to the TNM staging system as This work was supported by the Ligue Nationale de Lutte Contre described by Bostwick et al.” Group A patients had disease le Cancer; the Cornit& Regionaux des Hauts de Seine, du Val d’oise et des Yvelines; and the Association pour la Recherche sur les Tu- limited to the prostate (n = 13; 31%); group B patients had local extracapsular extension (n = 17; 40%); and group C meurs Prostatiques.
Tumorigenesis results from an accumulation of genetic abnormalities such as functional inactivation of tumor suppressor genes (TSGs) and activation of oncogenes.’ In prostate cancer the most frequent type of tumor-associated mutation is somatic loss of heterozygosity (LOH) at specific regions on chromosome arms 7q, 8p, lOq, 16q, 17q and 18q.2-7 These are believed to unmask recessive mutations inactivating tumor suppressor genes which otherwise regulate normal cellular growth and suppress abnormal cell proliferation. With the aim of elucidating the role of genetic alterations in the development of prostate cancer, we focused on these regions which have previously been shown to be deleted in prostate cancer, and also on 5q and 13q, which are regions of particular interest harboring known TSGs. We compared normal and tumor DNA from 42 patients, by using two DNA probes for restriction fragment length polymorphism (RFLP) and 19 microsatellite markers (CA repeats). We used RFLP and microsatellite markersS because they make it possible to monitor any numerical imbalance that may arise between the two parental chromosome homologues during tumorigenesis. Moreover, microsatellite repeat polymorphism is highly informative and can be identified rapidly by means of polymerase chain reaction (PCR) amplification of minimal tumor DNA.
2079
2080
CHROMOSOME ALLELIC IMBALANCE IN PROSTATE CANCER
Statistical analysis. Differences in the distribution bepatients had regional lymph node involvement or distant ,ween the different populations were tested by the chi-square metastases (n = 12; 29%). Microscopic lymph node metasta,est with Yates’ correction for adjustement of the continuity ses were seen in 6 patients in group C, and distant metastases )f the chi-square distribution. were found in the other 6 cases. We analyzed allelic imbalance (AI)on 8 autosomal chroRESULTS mosome arms (5q, 7q, 8p, lOq, 13q, 16q, 17q, 18q)by using 19 variable numbers of tandem repeat probes (CA repeats) and Eight autosomal chromosome arms were assessed a t 2 or two DNA probes for RFLP (pMetH on 7q and p15-65 on 18q) more loci for allelic imbalance (AI)in 42 cases of prostate to screen the 42 prostate tumor samples. At least two mark- Icancer. Informativity was a t least 90% (38/42),except for 16q ers were used for each region of interest in each chromosome (76%;32/42). Allelic imbalance on a t least 1of these chromoarm, to obtain a maximum of informative results. Table 1 some arms was found in all tumor DNAs. The frequency of A1 gives details of the loci investigated and their corresponding in informative cases varied considerably. Allelic imbalance locations.’z-1“ Peripheral blood leukocytes were used as a was frequent on chromosome arms 8p (58%),16q (53%) and source of normal DNA for each patient. 7q ( 4 6 8 ) and relatively frequent on arms 18q (39%), 17q Southern blot analysis. Frozen tissue samples were ground (34%), 13q (2657-1, and 1Oq (238). In contrast, the frequency of in liquid nitrogen to a fine powder using a mortar and pestle. AI was only 10% on 5q. The frequency of A1 on all the loci High-molecular-weight DNA was prepared by proteinase K tested is shown in table 1. Representative results are shown digestion and phenoVchloroform extraction, from both blood in the figure. and tissue samples. For each sample, 10 pg. of genomic DNA Allelic imbalance on the chromosome arms studied was was digested with Taql (pmetH) or Pstl (p15-65) and frac- analyzed according to the tumor staging system of Bostwick tionated by electrophoresis on 0.8% agarose gel. Leukocyte et al.11 Allelic imbalance for loci on chromosome arm 7q and and tumor DNA from each patient were analyzed in adjacent 13q was observed in respectively 62% (8/13) and 36% (4/11) of tracks. DNA was then transferred to nylon membrane filters locally confined tumors (group A) and 25% (3112) and 9% according to standard blotting procedures. DNA probes were (1/11) of metastatic forms (group C), whereas AI on 5q, 8p, labeled with :34PdCTP by using a random primer labeling 16q and 17q was found with a similar frequency in the 2 system. The membrane filters were hybridized overnight at groups. AI on 1Oq and 18q was more frequent in localized 65C with the denatured labeled probe, washed, and autora- advanced stage tumors (group B) and metastatic forms diographed at -8OC for an appropriate period. (group C) than in locally confined tumors (group A). No A1 on Detection of microsatellite markers by PCR. Polymerase 1Oq was found in group A. Frequencies of A1 in each group chain reaction was performed in a total volume of 50 nl. are summarized in table 2. containing 50 ng. of genomic DNA, 20 mM. each primer, 1.5 mM. MgCl,, 0.1 mM. each deoxynucleotide triphosphate and DISCUSSION 1 u of Taq DNA polymerase. DNA amplification was perA s PCR amplification cannot be considered quantitative, it formed conventionally,R except that samples were subjected to 35 cycles of amplification consisting of 40 seconds of dena- is usually difficult or impossible t o distinguish between allele turation and 30 seconds of annealing. The final extension gains and losses by means of microsatellite repeat polymorstep a t 72C was lengthened to 10 minutes. The magnesium phism analysis using PCR. However, comparative genomic concentration, annealing temperature and number of ampli- hybridization (CGH), a valuable tool to distinguish between fication cycles were optimized for each primer set. Products allele gains or losses, has shown that the AI on 5q, 8p, lOq, were diluted 1:3 in denaturing loading buffer and then heatdenatured; 1.5 nl. of each sample was loaded on 6% acrylamide gels containing 7.5 M. urea. DNA was then transferred TABLE1. Allelic imbalance on the 8 chromosome arms in a series to nylon membrane filters. CA repeat probes were labeled o f 4 2 human prostate tumors with 32P dCTP by using terminal deoxynucleotidyl transAllelic ferase. The membrane filters were hybridized overnight at Markers Polymorphism lossesiinformative 42C with labeled probe, washed, and autoradiographed at type Position cases (all tumors) -8OC for an appropriate period. 3129 CA repeat 5q21 D5S669 Detection of allelic imbalance. Leukocyte and tumor DNA D5S659 3132 5q21 CA repeat from each patient were analyzed in adjacent tracks. AI de- Total 5q21 4139 (10%1 11134 CA repeat 7q31.1 termination can only be done on “informative” patients. Nor- D7S486 9125 CA repeat 7q31.1 mal DNA samples which were polymorphic a t a given locus D7S522 5121 (a1 RFLP 7q31.1 were considered informative, whereas homozygotes were MET 15133 CA repeat 7q31.2 D7S480 “uninformative”. The signal intensity of fragments was de- D7S650 10125 CA repeat 7q31.2 13130 CA repeat 7q31.2 termined by densitometry and/or scored blindly and indepen- D7S490 19/41(46%) Total 7q31 dently by three observers. Although PCR amplification can15127 CA repeat 8p22 not be considered quantitative, we optimized the PCR D8S133 19134 CA repeat 8p22 D8S261 conditions so that equal amounts of template produced equal Total 8p22 22/38158% I 7132 CA repeat 10q25 amounts of amplified product. Allelic imbalance was consid- D10S187 6/30 10q25 CA repeat ered to be present when the relative intensity of the two D10S190 9/39123%I Total 10q25 alleles in tumor DNA differed from the relative intensity in D13S289 5/29 CA repeat 13q12-ql3 lymphocyte DNA by a factor of a t least 1.5, which is a n Dl35267 9128 CA repeat 13q12-qI3 10139 (264I adequate cutoff value to determine A1 of microsatellite loci in Total 13q CA repeat 11/20 16q23 tumor samples containing greater than 60% cancer cells.1fi D16S518 13/22 CA repeat D16S402 16q24.2 Evaluations based on the dividing point of 1.5 agreed well Total 16q 17/32(53%) with evaluations made by visual inspection. Each analysis D17S800 CA repeat 10130 17q21.1 CA repeat 13/36 17q21.1 was performed at least twice to ensure reproducible detection D17S855 13/40134% 1 of A1 (another independent PCR amplification, gel separa- Total 17q21 7/23 la) RFLP 18q21.1 tion, and quantification). In an attempt to distinguish allelic DCC DlKS39 10131 CA repeat 18q21.2-q21.3 gain from LOH, comparative multiplex PCR was performed17 D18S42 CA repeat 8/17 18q21.2-q21 . . 3 16/41139%) and microsatellite markers (D4S244 or D21S222)IXwithout Total 1Rq21 known allelic imbalance’!’ were selected as internal controls. I~II RFI,P. ~rstrictionfriigment length polymorphism.
CHROMOSORIE ALLELIC IMBALANCE IN PROSTATE CANCER
2081
Gao et al.+j recently reported 47% of LOH on chromosome arm 17q in prostate cancer. Although we found a lower fre-
A, examples of allelic imbalance of microsatellite loci on different chromosome arms in representative prostate cancers. L and T indicate matched DNA samples isolated from peripheral leukocytes and tumor tissue, respectively. Allelic imbalance, in informative cases, was considered to be present when relative intensity of 2 alleles in tumor DNA (T)differed from relative intensity in lymphocyte DNA (L) by factor of at least 1.5. Faint signals in tumor DNA might be due either to contaminating normal tissue or to tumor heterogeneity. Arrows, point a t reduced or deleted alleles a t various loci. All patterns were re roducible for all patients studied. B, example of comparative m d i p l e x PCR in case 32. Internal control marker (D21S222) indicates that almost equal amounts of DNAs were analyzed in PCR and loaded in tumor (T)and lymphocyte (L) lanes. Case 32 shows loss of upper band at D13S289.
13q, 16q, 17q, and 18q in prostate cancer may be caused by a loss rather than a gain of DNA sequences.20.21 Moreover, Southern blot analysis has revealed allelic losses in prostate cancer on chromosome arms 7q, 8p, lOq, 13q, 16q, and 18q.2,3,5,22 In this study, using the RFLP markers c-met and p15-65, the informative patients in whom an imbalance was observed between the two alleles in the tumor DNA track, showed us that the nature of the imbalance at the 7q and 18q loci was classic LOH. When we compared our results with those obtained in the same region by CGH analysis or molecular techniques, they supported the notion that all the allelic imbalances observed in this study could be interpreted a s LOH. Our findings thus indicate that LOH on chromosome arms 7q, 8p, lOq, 13q, 16q, 17q, and 18q is a frequent event (table 11, in keeping with previous studies. In contrast, 5q LOH is infrequent. The most frequent site of LOH was the 8p22 band, 58% of the tumors exhibiting apparent allelic deletions. This is consistent with previous work in which Bova et al.4 identified a common region of deletion at 8p22 in prostate cancer. Likewise, the loss of 8p in various cancers, and particularly the 8p22 region,23 suggests that the putative TSG a t this site may be relevant to other human tumors. The other chromosomal site most frequently altered in the prostate cancers we studied was 16q, with 53% of LOH. This is in agreement with previous studies based on RFLP analysis of fewer cases.2.3 In a previous study we described LOH of the MET gene on chromosome band 7q31;5 Zenklusen et a1.: using microsatellite markers, recently reported a high frequency of LOH in the same region. In this study a loss of genetic information in the 7q31 region was found in 19 of 41 informative cases (46%). These results agree with Zenklusen’s findings. With regard to the long arm of chromosome 18,39%(16/41) of the tumors exhibited apparent allelic loss. We have previously described a common region of deletion on chromosome arm 18q5 and a likely target gene for LOH on 18q is the DCC (Deleted in Colorectal Carcinoma) gene; indeed, deletion of the DCC gene and an abnormal DCC gene product have both been identified in prostate cancer.24
quency of LOH in the same 17q region (34%.versus 47%’%.), possibly owing to the smaller number of cases in Gao’s study, the highest rate of deletion in the two studies was observed a t the D17S855 locus, a site recently shown to be intragenic to the BRCAl gene.15 Taken together with the data of Gao et a1.,6 our results support the idea that BRCAl is the probable target gene of the observed LOH on 17q, although additional studies are required to determine its role in both sporadic prostate tumors and familial prostate carcinoma. Frequent LOH on chromosome arm 1Oq in prostate tumors has been reported by other teams, who used RFLP to analyze fewer sample^.^.^ We found allelic loss on 1Oq in 23% (9 of 39) of informative patients, which is consistent with the aforementioned data. Positional cloning recently assigned the BRCA2 gene, which is responsible for inherited breast cancer, to the 13q12q13 region.25Arason et a1.26 proposed that genes predisposing to breast cancer also increase the risk of prostate cancer. As the BRCA2 gene might be involved in sporadic breast tum0rs,2~we postulated that the BRCA2 gene may also play an important role in sporadic prostate tumors. According to Wooster et al.25 the most likely location for BRCA2 is telomeric to D13S289 and proximal to D13S267. In this study, in which we used these two microsatellite markers, allelic loss was found in 10 of 39 informative patients (26%), suggesting the possible involvement of BRCA2 in prostate cancer. However, the highest rate of LOH was observed at the D13S267 locus (9/28; 32%). The TSG RB1 locus, which is more telomeric to D13S267, may also account for the LOH observed in this region. Even though conflicting results concerning the RB1 gene involvement have been reported,22.28 additional studies are required to determine which gene-BRCA2 or RB1-is the target TSG of the observed LOH, together with its role in prostate tumorigenesis. Chromosome arm 5q was notable for its lack of apparent involvement in prostate cancer; the background level of AI (10%) observed on the 5q21 region in this study may correspond to random events during the abnormal mitoses of cancer cells. Brewster et al.29 reported 20% of LOH within the APC gene (which is located in the same region) in prostate tumors. The discrepancy between the two studies might be due to a sampling bias (that is, the small number of informative cases in Brewster’s study), but the possibility of APC gene inactivation by point mutations or intragenic deletion cannot be ruled out, and neither can the involvement of the a catenin gene mapping to this region.30 Allelic loss on the different chromosome arms was analysed according to the tumor staging system of Bostwick et al.” The frequency of LOH at 7q31 and 13q12-13 in metastatic cancers (group C) was not higher than in locally confined tumors (group A), suggesting that inactivation of a putative TSG in these regions might occur a t an early stage of prostate development. The difference in the LOH rate in the 7q31 and 13q12-13 regions between group A and group C was not significant and could be due to the small number of cases studied; alternatively, a cell subclone with a 7q31 or 13q12-13 deletion in metastatic tumors (group C) could be masked by a dominant cell subclone bearing other genetic alterations specific to invasiveness. However, further investigations are required to confirm whether 7q31 and 13q12-13 deletions are less frequent in metastatic than in localized prostate tumors, because this would suggest that 7q31 and 13q12-13 deletions contribute to local growth of cancer cells, and are not involved in the process by which cells acquire metastatic potential. Our observations on 7q partially disagree with a recent report31 which show that 7q31.1 LOH is associated with tumor aggressiveness and progression. However when our results are considered alongside those of
CHROMOSOME ALLELIC IMBALANCE I N PROSTATE CANCER
2082
TABLE2. Allelic losses on 8 chromosome arms in prostate cancer a n d corresponding pathological features Number of allelic lossedinformative cases ( % )
5q
7q
8~
1%
13q
1%
17q
__
18q
(a I GROUP
A B
C
1/10i 10) 2/17(12) 1/11 (9)
8/13(62) 8/16150) 3/12( 2 5 )
7/11164) 8/17 (47) 7/10(70)
0/13 ( 0 ) 4/14(29) 5/12(42)
4/11( 3 6 ) 5/17(29) 1/11 (9)
4/9 (44) 8/14(571 519 (56)
5/12142) 2/13(15) 4/17(24) 8/17(47) 4/11(36)____ 6/11(55)
(a1 groups according to Bostwick et al.”: group A: patients with disease limited t o the prostate; group B: patients with local extracapsular extension; group C: patients with regional lymph node involvement or distant metastases.
Zenklusen e t al.7 on clinically localized prostate tumors, although there was no tendency towards specific coalterthey support the idea t h a t this genetic alteration on chro- ations. Interestingly 3 tumors in group A showed deletion in mosome band 7q31 might be a n early event in prostate only 1region, and in each case the allele loss was a t 7q31. We observed a positive link between the number of AI and the tumorigenesis. Similar rates of LOH were observed in group C and group Gleason score (4 of 14 tumors with grade 5 6 had only one A on chromosome arms 8p, 16q and 17q (table 2). Interest- region of LOH, whereas more than one region of LOH was ingly, the loss of 8 p was detected in a considerable number observed in 10 of 11 grade 2 8 tumors). Finally, our findings suggest that inactivation of a putative of tumors in each group, and particularly in locally contined tumors (7/11; 64%)suggesting t h a t this genetic al- TSG on 7q and 13q is an early event in prostate tumorigenteration might be a relatively early event, even though it esis. Moreover, LOH observed on 8p, 16q and 17q may occur at a n earlier stage than LOH on 1Oq and 18q, which appears has been linked to invasiveness.32.33 The E-cadherin gene located at 16q22.1,34 the product of to be related to invasiveness in prostate cancer. which plays an important role in cell-cell adhesion processes, We are indebted to Dr, B. vogelstein for providing probe might be the target of the observed LOH on 16%as suggested p15-65. The pmetH probe was obtained from the American by Isaacs et al.35 Indeed, reduced E cadherin expression is a Tissue Type Culture ~ ~ l l ~ ~ ~ i ~ ~ , powerful predictor of poor outcome, in terms of both disease progression and overall survival in prostate cancer.36 However, the incidence of LOH observed on 16q in our study was REFERENCES 1, Vogelstein, B., Fearon, E , R,, Hamilton, S. R., Kern, s. E,, high, irrespective of differences in clinical and pathological stage. The high frequency of distal 16q deletion in clinically Preisinger, A. C., Leppert, M., Nakamura, Y . , White, R., localized prostatic carcinomas (group A and group B) is in Smith, A. M. and Bos, J. L.: Genetic alterations during agreement with a recent FISH analysis.37 More studies of colorectal-tumor development. N. Engl. J. Med., 319: 525, prostate tumors are thus required to refine this 16q region. 1988. Analysis of results in the light of findings by Cher et a1.37 2. Carter, B. S., Ewing, C. M., Ward, W. S.,Treiger, B. F., Aalders, suggested that, in addition to the putative involvement of the T. W., Schalken, J. A., Epstein, J. I. and Isaacs, W. B.: Allelic E-cadherin gene at 16q22.1, inactivation of a n additional loss of chromosomes 16q and 1Oq in human prostate cancer. Proc. Natl. Acad. Sci. U.S.A., 87: 8751, 1990. TSG, at 16q24’2-qter7 might be a early event in 3, Bergerheim, U,, Kumini, K., Collins, P, V, and E b a n , p,: Deprostate tumorigenesis. Likewise, the loss of 16q, particuletion mapping of chromosomes 8, 10, and 16 in human prostatic carcinoma. Genes Chrom. Cancer, 3:215, 1991. lady the 16q22.1 and 16q24.3 regions, in breast and hepatocellular carcinomas,3s-40 suggests that Putative TSGs at 4. Bova, G. S., Carter, B. S., Bussemakers, M. J., Emi, M., these sites may be relevant to other human tumors. Fujiwara, Y., Kiprianou, N., Jacobs, S. C., Robinson, J. C., Our results on 17q agree with those of Gao et a1.,6 and Epstein, J. I., Walsh, P. C. and Isaacs, W. B.: Homozygous further indicate that the inactivation of the BRCAl gene deletion and frequent allelic loss of chromosome 8p22 loci in might be a relatively early event in human prostate cancer. human prostate cancer. Cancer Res., 53: 3869, 1993. The only significant relationship was observed between 5. Latil, A., Baron, J-c., Cussenat, o., Fournier, G., soussi, T., Boccon-Gibod, L., Leduc, A,, Rouesse, J. and Lidereau, R.: 1Oq LOH and metastatic cancers (group C) (p = 0.03). The Genetic alterations in localized prostate cancer: identification MXIl gene located at 10q24-q25, the product of which negaof a common region of deletion on chromosome 18q. Genes tively regulates MYC oncoprotein activity, might be a likely Chrom. Cancer, 11: 119, 1994. target gene for LOH on indeed, a recent study indicates 6. Gao, X., Zacharek, A,, Salkowski, A,, Grignon, D. J . , Sakr, W., that MXIl is a frequent target for mutation in prostatic Porter, A. T. and Honn, K. V.: Loss of heterozygosity of BRCAl carcinoma with deletion in this region and suggest a role for and other loci on chromosome 17q in human prostate cancer. this gene in desease pathogenesis or, more likely, in its Cancer Res., 55: 1002, 1995. evolution.4*The close link between the loss of 18q and inva7. Zenklusen, J. C., Thompson, J. C., Troncoso, P., Kagan, J. and sive phenotypes (group C) also suggests that this genetic Conti, C. J.: Loss of heterozygosity in human primary prostate carcinomas: a possible tumor suppressor gene a t 7q31.1. Canalteration, as well as 1Oq LOH, might be a relatively late cer Res., 54: 6370, 1994. event in prostate tumorigenesis. DCC, which encodes a mob 8 , Weber, J , L, and May, p, E,: Abundant class of human DNA ecule with a high degree of homology with neural cell adhepolymorphisms which can be typed using the polymerase sion molecule, might be the target of the observed LOH on chain reaction. Am. J. Hum. Genet., 44: 388, 1989. 18q, as suggested by Gao’s study.“ 9. Mellinger, G. T., Gleason, D. and Bailar, J.: The histology and The LOHs found in this study on chromosome arms 7q, 8p, prognosis of prostatic cancer. J. Urol., 97:331, 1967 111. lOq, 13q, 16q, 17q and 18q do not appear to be random events 10. Schroder, F. H., Hermanek, P., Denis, L., Fair, W. R., because of their high frequencies. These data are consistent Gospodarowicz, M. K. and Pavone-Macaluso, M.: The TNM with previous reports and indicate that a relatively large classification of prostate cancer. Prostate, Suppl 4: 129, 1992. number of chromosome loci play a part in the etiolom and 11. Bostwick, D. G., Myers, R. P. and Oesterling, J. E.: Staging in prostate cancer. Semin. Surg. Oncol., 1 0 60, 1994. Of this type’ Moreover* most Of these ge- 12. Tomfohrde, J., Wood, S.,Schertzer, M., Wagner, M. J., Wells, netic alterations are already present in locally confined tuD. E., Parrish, J . , Sadler, L. A,, Blanton, S. H., Daiger, S. P., mors, except for 1Oq LOH. In group A (13 locally confined Wang, Z . , Wilkie, P.J . and Weber, J. L.: Human chromosome tumors) and group C (12 metastatic cancers), 5 tumors 8 linkage map based on short tandem repeat polymorphisms: showed 3 or more different chromosomal regions of deletion, effect of genotyping errors. Genomics, 14: 144, 1992.
CHROMOSOME ALLELIC IMBAIA N C E IN PROSTATE CANCER
2083
13. Weissenbach, J., Gyapay, G., Dib, C., Vignal, A., Morissette, J., 26. Arason, A,, Barkardottir, R. B. and Egilsson, V.: Linkage analMillasseau, P., Vaysseix, G. and Latbrop, M. A.: Secondysis of chromosome 17q markers and breast-ovarian cancer in generation linkage map of the human genome. Nature, 359: Iceland families, and possible relationship to prostatic cancer. 794, 1992. Am. J . Hum. Genet., 52: 711, 1993. 14. Gyapay, G., Morissette, J., Vignal, A., Dib, C., Fizames, C., 27. Kerangueven, F., Allione, F., Noguchi, T., Sobol, H., Jacquemier, Millasseau, P., Marc, S., Bernardi, G., Lathrop, M. and J. and Birnbaum, D.: Patterns of loss of heterozygosity at loci Weissenbach, J.: The 1993-94 Genethon human genetic linkfrom chromosome arm 13q suggest a possible involvement of age map. Nature Genetics, 7: 246, 1994. BRCA2 in sporadic breast tumors. Genes Chrom. Cancer, 1 3 15. Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P. A,, 291, 1995. Harshman, K., Tavtigian, S., Liu, Q., Cochran, C., Bennett, 28. Sarkar, F. H., Sakr, W. A., Li, Y. W., Macoska, J., Ball, D. E. and L. M., Ding, W., Bell, R., Rosenthal, J., Hussey, C., Tran, T., Crissman, J. D.: Analysis of retinoblastoma (RB)gene deletion McClure, M., Frye, C., Hattier, T., Phelps, R., Haugen-Stano, in human prostatic carcinomas. Prostate, 21: 145, 1992. A,, Katcher, H., Yakumo, K., Gholami, Z., Shaffer, D., Stone, 29. Brewster, S. F., Browne, S. and Brown, K.: Somatic allelic loss at S., Bayer, S., Wray, C., Bogden, R., Dayananth, P., Ward, J., the DCC, APC, nm23-H1 and p53 tumor suppressor gene loci Tonin, P., Narod, S., Briston, P. K., Noms, F. H., Helvering, in human prostatic carcinoma. J. Urol., 151: 1073, 1994. L., Morrison, P., Rosteck, P., Lai, M., Barrett, J . C., Lewis, C., 30. McPherson, J . D., Morton, R. A., Ewing, E. M., Wasmuth, J . J., Neuhausen, S., Cannon-Albright, L., Goldgar, D., Wiseman, Overhauser, J., Nagafuchi, A,, Tsukita, S. and Isaacs, W. B.: R., Kamb, A. and Skolnick, M. H.: A strong candidate for the Assignment of the human a catenin gene ( C T ” A 1 ) to chrobreast and ovarian cancer susceptibility gene BRCA1. Science, mosome 5q21-q22. Genomics, 19 188, 1994. 266 66, 1994. 31. Takahashi, S., Shan, A. L., Ritland, S. R., Delacey, K A., 16. McGrogan, D., Levy, A,, Bostwick, D., Wagner, M., Wells, D. and Bostwick, D. G., Lieber, M. M., Thibodeau, S. N. and Jenkins, Bookstein, R.: Loss of chromosome arm 8p loci in prostate R. B.: Frequent loss of heterozygosity a t 7q31.1 in primary cancer: mapping by quantitative allelic imbalance. Genes prostate cancer is associated with tumor aggressiveness and Chrom. Cancer, 1 0 151, 1994. progression. Cancer Res., 5 5 4114, 1995. 17. Cairns, P., Tokino, K., Eby, Y. and Sidransky, D.: Homozygous 32. Suzuki, H., Emi, N., Komiya, A,, Fujiwara, Y., Yatani, R., deletions of 9p21 in primary human bladder tumors detected Nakamura, Y. and Shimazaki, J.: Localization of a tumor by comparative multiplex polymerase chain reaction. Cancer suppressor gene associated with progression of human prosRes., 54: 1422, 1994. tate cancer within a 1.2 Mb region of 8p22-p21.3. Genes 18. Hudson, T. J., Engelstein, M., Lee, M. K., Ho, E. C., Rubenfield, Chrom. Cancer, 13: 168, 1995. M. J., Adams, C. P., Housman, D. E. and Dracopoli, N. C.: 33. Trapman, J., Sleddens, H. F. B. M., van der Weiden, M. M., Isolation and chromosomal assignment of 100 highly informaDinjens, W. N. M.,Konig, J . J . and Schroder, F. H.: Loss of tive human simDle seauence reueat Genom- .uolvmoruhisms. heterozygosity of chromosome 8 microsatellite loci implicates a ics, 13: 622, 1992. candidate tumor suppressor gene between the loci D8S87 and 19. Watanabe, M.. Imai, H.. Shiraishi, T., Shimazaki, J., Kotate, T. D8S133 in human prostate cancer. Cancer Res., 54: 6061, and Yatani,’ R.: Microsatellite instability in human prostate 1994. cancer. Br. J . Cancer, 72: 562, 1995. 34. Natt, E., Magenis, R. E., Zimmer, J., Mansouri, A. and Scherer, 20. Cher, M. L., MacGrogan, D., Bookstein, R., Brown, J . A., G.: Regional assignment of the human loci for uvomorulin and Jenkins, R. B. and Jensen, R. H.: Comparative genomic hychymotrypsinogen B with the help of two overlapping delebridization, allelic imbalance, and fluorescence in situ hybridtions on the long arm of chromosome 16. Cytogenet. Cell Genization on chromosome 8 in prostate cancer. Genes Chrom. et., 5 0 145, 1989. Cancer, 11: 153, 1994. 35. Isaacs, W. B., Bova, G. S., Morton, R. A,, Bussemakers, M. J., 21. Visakorpi, T., Kallioniemi, A. H., Syvanen, A-C., Hyytinen, E. R., Brooks, J . D. and Ewing, C. M.: Molecular genetics and chroKarhu, R., Tammella, T., Isola, J . J . and Kallioniemi, 0-P.: mosomal alterations in prostate cancer. Cancer, s u p p l 7 2004, Genetic changes in primary and recurrent prostate cancer by 1995. comparative genomic hybridization. Cancer Res., 5 5 342, 36. Umbas, R., Isaacs, W. B., Bringier, P. P., Schaafsma, H. E., 1995. Karthaus, H. F. M., Oosterhof, G. 0. N., Debruyne, F. M. J . 22. Phillips, S. M. A., Barton, C. M., Lee, S. J., Morton, D. G., and Schalken, J. A,: Decreased E-cadherin expression is assoWallace, D. M. A., Lemoine, N. R. and Neoptolemos, J. P.: Loss ciated with poor prognosis in patients with prostate cancer. of the retinoblastoma susceptibility gene (RB1) is a frequent I Cancer Res., 54: 3929, 1994. and early event in prostatic tumorigenesis. Br. J . Cancer, 7 0 37. Cher, M. L., Ito, T., Weidner, N., Carroll, P. R. and Jensen, R. H.: 1252, 1994. Mapping of regions of physical deletion on chromosome 16q in 1 23. Fujiwara, Y., Ohata, H., Emi, M., Okui, K, Koyama, K, prostate cancer cells by fluorescence in situ hybridization Tsuchiya, E., Nakajima, T., Monden, M., Mori, T., Kurimasa, (FISH). J . Urol., 153: 249, 1995. I A. and Nakamura, Y.: A 3-Mb physical map of the chromosome 38. Tsuda, H., Zhang, W., Shimosato, Y., Yokota, J.. Terada, M., region 8p21.3-p22, including a 600-Kb region commonly deI Sugimura, T., Miyamura, T. and Hirohashi, S.: Allele loss on leted in human hepatocellular carcinoma, colorectal cancer, chromosome 16 associated with progression of human hepatoand non-small cell lung cancer. Genes Chrom. Cancer, 1 0 7, cellular carcinoma. Proc. Natl. Acad. Sci. U S A . , 87: 6791, 1 1994. 1990. 1 24. Gao, X., Honn, K. V., Grignon, D., Sakr, W. and Chen, Y. Q.: Frequent loss of expression and loss of heterozygosity of the 39. CletonJansen, A-M., Moerland, E. W., Kuipers-Dijkshoorn, N. J . K., Callen, D. F., Sutherland, G. R., Hansen, B., Devilee, 1 putative tumor suppressor gene DCC in prostatic carcinomas. P. and Cornelisse, C. J.: At least two different regions are Cancer Res., 53: 2723, 1993. involved in allelic imbalance on chromosome arm 16q in breast 25. Wooster, R., Neuhausen, S. L., Mangion, J., Quirk, Y., Ford, D., cancer. Gene Chrom. Cancer, 9 101, 1994. Collins, N., Nguyen, K., Seal, S., Tran, T., Averill, D., Fields, p., Marshall, G., Narod, S., Lenoir, G. M., Lynch, H., 40. Tsuda, H., Callen, D. F., Fukutomi, T., Nakamura, Y. and Hirohashi, S.: Allele loss on chromosome 16q24.2qter occurs Feunteun, J., Devilee, P., Cornelisse, C. J., Menko, F. H., D a b , I frequently in breast cancers irrespectively of differences in p. A., Ormiston, W., McManus, R., Pye, C., Lewis, C. M., phenotype and extent of spread. Cancer Res., 54: 513, 1994. Cannon-Albright, L. A., Peto, J., Ponder, B. A. J., Skolnick, I M. H., Easton, D. F., Goldgar, D. E. and Stratton, M. R.: 41. Eagle, L. R., Yin, X., Brothman, A. R., Williams, B. J., Atkin, N. B. and Prochownik, E. V.: Mutation of the MXIl gene in Localization of a breast cancer susceptibility gene, BRCA2, to prostate cancer. Nature Genetics, 9 249, 1995. chromosome 13q12-ql3. Science, 265: 2088, 1994. ~
,
,
,
I
i i b
THE JOURNAL OF U m o L o w Copyright 0 1996 by kwERI(.AN UROLOCICAL ASSOCIATION,
Vol. 156.2079-2083,December 1996 Printed in U.S.A.
INC.
DIFFERENTIAL CHROMOSOME ALLELIC IMBALANCE IN THE PROGRESSION OF HUMAN PROSTATE CANCER ALAIN LATIL, GEORGES FOURNIER, O L M E R CUSSENOT
AND
ROSETTE LIDEREAU*
From the Laboratoire d’Oncogenetique, Centre Rene Huguenin, St-Cloud, the Departement d’llrologie, H6pital & la Cavale Blanche, Brest, and the Departement d’Urologie, CHU Saint-Louis, Paris, France
ABSTRACT
Purpose: I t is widely accepted that a n accumulation of genetic alterations plays a n important role in the genesis of human cancers. We wished to obtain a comprehensive view of the role of genetic changes in prostate cancer. Materials and Methods: We screened 42 primary prostate tumors for allelic imbalance (AI) on 8 autosomal chromosome arms of interest (5q, 7q, 8p, lOq, 13q, 16q, 17q, 18q) by using 2 DNA probes for restriction fragment length polymorphism (RFLP) and 19 microsatellite markers (CA repeats). Results: The most frequent allelic imbalances were observed on 8p (58%)and 16q (53%).AI exceeding 20% was also observed a t sites on chromosome arms 7q (46%), 1Oq (23%), 13q (26%), 17q (34%)and 18q (39%),whereas AI was infrequent on 5q (10%). Conclusions: The data indicate that a relatively large number of chromosome loci play a part in the etiology and progression of this tumor type. Moreover, our findings suggest that inactivation of a putative tumor suppressor gene on 7q and 13q is a n early event in prostate tumorigenesis. In contrast, the close link between a n invasive phenotype and AI on 1Oq and 18q suggests that these genetic alterations occur late in prostate tumorigenesis. KEY WOROS:prostate cancer; allelic influence; loss of heterozygosity;tumor suppressor gene
cally localized prostate carcinomas underwent radical prostatectomy with lymph node excision. Thirteen patients had confined prostate tumors and twelve patients had local extracapsular extension. A radical prostatectomy specimen was first sliced thickly and samples of tissue were cut out of suspect areas. One part of the specimen was immediately placed in liquid nitrogen for high-molecular-weight DNA extraction, while adjacent sections of the initial specimen removed for molecular analysis were stained with hematoxylin and eosin to determine the proportion of tumor cells in each sample. A sample was considered suitable for DNA analysis if the proportion of tumor cells was 60% or more. Transurethral resection. The 12 patients who underwent this procedure had regional lymph node involvement (6 cases) or bone metastases (6 cases). Between 6 and 12 chips were obtained during resection. Specimens were selected for DNA extraction as described above. Needle biopsy. Samples were obtained in this way from 5 patients who had clinically localized prostate carcinomas with local extracapsular extention. The biopsy needle was introduced into suspect areas by the transrectal route using an automatic gun and a needle guide with digital control. Three 14-gauge biopsies were taken from the same site, placed in liquid nitrogen and checked histologically. Samples were selected for DNA extraction as described above. MATERIALS AND METHODS The histological diagnosis, Gleason scoreg and pathologic Patients and samples. Forty-two samples of prostate tu- tumor stage according to the tumor-nodes-metastasis classimors were obtained from patients undergoing surgery a t St fication of prostate cancerlo were determined in each case h i s Hospital (Paris)and La Cavale Blanche Hospital (Bmst). during a routine workup after surgery. The Gleason score for the 42 primary tumors ranged from The samples were obtained in 3 ways, as follows: Prostatectomy specimens. Twenty-five patients with clini- 5 to 9. By combining the Gleason score with the pathologic stage, we confirmed the well-established positive link be9 p t e d for publication April 30, 1996. Requests for reprints: Laboratoire d‘oncogenetique,Centre Rene tween grade and stage. The tumors were subdivided into Humenin, 35 rue Dailly, F-92211, St-Cloud, France. three groups corresponding to the TNM staging system as This work was supported by the Ligue Nationale de Lutte Contre described by Bostwick et al.” Group A patients had disease le Cancer; the Cornit& Regionaux des Hauts de Seine, du Val d’oise et des Yvelines; and the Association pour la Recherche sur les Tu- limited to the prostate (n = 13; 31%); group B patients had local extracapsular extension (n = 17; 40%); and group C meurs Prostatiques.
Tumorigenesis results from an accumulation of genetic abnormalities such as functional inactivation of tumor suppressor genes (TSGs) and activation of oncogenes.’ In prostate cancer the most frequent type of tumor-associated mutation is somatic loss of heterozygosity (LOH) at specific regions on chromosome arms 7q, 8p, lOq, 16q, 17q and 18q.2-7 These are believed to unmask recessive mutations inactivating tumor suppressor genes which otherwise regulate normal cellular growth and suppress abnormal cell proliferation. With the aim of elucidating the role of genetic alterations in the development of prostate cancer, we focused on these regions which have previously been shown to be deleted in prostate cancer, and also on 5q and 13q, which are regions of particular interest harboring known TSGs. We compared normal and tumor DNA from 42 patients, by using two DNA probes for restriction fragment length polymorphism (RFLP) and 19 microsatellite markers (CA repeats). We used RFLP and microsatellite markersS because they make it possible to monitor any numerical imbalance that may arise between the two parental chromosome homologues during tumorigenesis. Moreover, microsatellite repeat polymorphism is highly informative and can be identified rapidly by means of polymerase chain reaction (PCR) amplification of minimal tumor DNA.
2079
2080
CHROMOSOME ALLELIC IMBALANCE IN PROSTATE CANCER
Statistical analysis. Differences in the distribution bepatients had regional lymph node involvement or distant ,ween the different populations were tested by the chi-square metastases (n = 12; 29%). Microscopic lymph node metasta,est with Yates’ correction for adjustement of the continuity ses were seen in 6 patients in group C, and distant metastases )f the chi-square distribution. were found in the other 6 cases. We analyzed allelic imbalance (AI)on 8 autosomal chroRESULTS mosome arms (5q, 7q, 8p, lOq, 13q, 16q, 17q, 18q)by using 19 variable numbers of tandem repeat probes (CA repeats) and Eight autosomal chromosome arms were assessed a t 2 or two DNA probes for RFLP (pMetH on 7q and p15-65 on 18q) more loci for allelic imbalance (AI)in 42 cases of prostate to screen the 42 prostate tumor samples. At least two mark- Icancer. Informativity was a t least 90% (38/42),except for 16q ers were used for each region of interest in each chromosome (76%;32/42). Allelic imbalance on a t least 1of these chromoarm, to obtain a maximum of informative results. Table 1 some arms was found in all tumor DNAs. The frequency of A1 gives details of the loci investigated and their corresponding in informative cases varied considerably. Allelic imbalance locations.’z-1“ Peripheral blood leukocytes were used as a was frequent on chromosome arms 8p (58%),16q (53%) and source of normal DNA for each patient. 7q ( 4 6 8 ) and relatively frequent on arms 18q (39%), 17q Southern blot analysis. Frozen tissue samples were ground (34%), 13q (2657-1, and 1Oq (238). In contrast, the frequency of in liquid nitrogen to a fine powder using a mortar and pestle. AI was only 10% on 5q. The frequency of A1 on all the loci High-molecular-weight DNA was prepared by proteinase K tested is shown in table 1. Representative results are shown digestion and phenoVchloroform extraction, from both blood in the figure. and tissue samples. For each sample, 10 pg. of genomic DNA Allelic imbalance on the chromosome arms studied was was digested with Taql (pmetH) or Pstl (p15-65) and frac- analyzed according to the tumor staging system of Bostwick tionated by electrophoresis on 0.8% agarose gel. Leukocyte et al.11 Allelic imbalance for loci on chromosome arm 7q and and tumor DNA from each patient were analyzed in adjacent 13q was observed in respectively 62% (8/13) and 36% (4/11) of tracks. DNA was then transferred to nylon membrane filters locally confined tumors (group A) and 25% (3112) and 9% according to standard blotting procedures. DNA probes were (1/11) of metastatic forms (group C), whereas AI on 5q, 8p, labeled with :34PdCTP by using a random primer labeling 16q and 17q was found with a similar frequency in the 2 system. The membrane filters were hybridized overnight at groups. AI on 1Oq and 18q was more frequent in localized 65C with the denatured labeled probe, washed, and autora- advanced stage tumors (group B) and metastatic forms diographed at -8OC for an appropriate period. (group C) than in locally confined tumors (group A). No A1 on Detection of microsatellite markers by PCR. Polymerase 1Oq was found in group A. Frequencies of A1 in each group chain reaction was performed in a total volume of 50 nl. are summarized in table 2. containing 50 ng. of genomic DNA, 20 mM. each primer, 1.5 mM. MgCl,, 0.1 mM. each deoxynucleotide triphosphate and DISCUSSION 1 u of Taq DNA polymerase. DNA amplification was perA s PCR amplification cannot be considered quantitative, it formed conventionally,R except that samples were subjected to 35 cycles of amplification consisting of 40 seconds of dena- is usually difficult or impossible t o distinguish between allele turation and 30 seconds of annealing. The final extension gains and losses by means of microsatellite repeat polymorstep a t 72C was lengthened to 10 minutes. The magnesium phism analysis using PCR. However, comparative genomic concentration, annealing temperature and number of ampli- hybridization (CGH), a valuable tool to distinguish between fication cycles were optimized for each primer set. Products allele gains or losses, has shown that the AI on 5q, 8p, lOq, were diluted 1:3 in denaturing loading buffer and then heatdenatured; 1.5 nl. of each sample was loaded on 6% acrylamide gels containing 7.5 M. urea. DNA was then transferred TABLE1. Allelic imbalance on the 8 chromosome arms in a series to nylon membrane filters. CA repeat probes were labeled o f 4 2 human prostate tumors with 32P dCTP by using terminal deoxynucleotidyl transAllelic ferase. The membrane filters were hybridized overnight at Markers Polymorphism lossesiinformative 42C with labeled probe, washed, and autoradiographed at type Position cases (all tumors) -8OC for an appropriate period. 3129 CA repeat 5q21 D5S669 Detection of allelic imbalance. Leukocyte and tumor DNA D5S659 3132 5q21 CA repeat from each patient were analyzed in adjacent tracks. AI de- Total 5q21 4139 (10%1 11134 CA repeat 7q31.1 termination can only be done on “informative” patients. Nor- D7S486 9125 CA repeat 7q31.1 mal DNA samples which were polymorphic a t a given locus D7S522 5121 (a1 RFLP 7q31.1 were considered informative, whereas homozygotes were MET 15133 CA repeat 7q31.2 D7S480 “uninformative”. The signal intensity of fragments was de- D7S650 10125 CA repeat 7q31.2 13130 CA repeat 7q31.2 termined by densitometry and/or scored blindly and indepen- D7S490 19/41(46%) Total 7q31 dently by three observers. Although PCR amplification can15127 CA repeat 8p22 not be considered quantitative, we optimized the PCR D8S133 19134 CA repeat 8p22 D8S261 conditions so that equal amounts of template produced equal Total 8p22 22/38158% I 7132 CA repeat 10q25 amounts of amplified product. Allelic imbalance was consid- D10S187 6/30 10q25 CA repeat ered to be present when the relative intensity of the two D10S190 9/39123%I Total 10q25 alleles in tumor DNA differed from the relative intensity in D13S289 5/29 CA repeat 13q12-ql3 lymphocyte DNA by a factor of a t least 1.5, which is a n Dl35267 9128 CA repeat 13q12-qI3 10139 (264I adequate cutoff value to determine A1 of microsatellite loci in Total 13q CA repeat 11/20 16q23 tumor samples containing greater than 60% cancer cells.1fi D16S518 13/22 CA repeat D16S402 16q24.2 Evaluations based on the dividing point of 1.5 agreed well Total 16q 17/32(53%) with evaluations made by visual inspection. Each analysis D17S800 CA repeat 10130 17q21.1 CA repeat 13/36 17q21.1 was performed at least twice to ensure reproducible detection D17S855 13/40134% 1 of A1 (another independent PCR amplification, gel separa- Total 17q21 7/23 la) RFLP 18q21.1 tion, and quantification). In an attempt to distinguish allelic DCC DlKS39 10131 CA repeat 18q21.2-q21.3 gain from LOH, comparative multiplex PCR was performed17 D18S42 CA repeat 8/17 18q21.2-q21 . . 3 16/41139%) and microsatellite markers (D4S244 or D21S222)IXwithout Total 1Rq21 known allelic imbalance’!’ were selected as internal controls. I~II RFI,P. ~rstrictionfriigment length polymorphism.
CHROMOSORIE ALLELIC IMBALANCE IN PROSTATE CANCER
2081
Gao et al.+j recently reported 47% of LOH on chromosome arm 17q in prostate cancer. Although we found a lower fre-
A, examples of allelic imbalance of microsatellite loci on different chromosome arms in representative prostate cancers. L and T indicate matched DNA samples isolated from peripheral leukocytes and tumor tissue, respectively. Allelic imbalance, in informative cases, was considered to be present when relative intensity of 2 alleles in tumor DNA (T)differed from relative intensity in lymphocyte DNA (L) by factor of at least 1.5. Faint signals in tumor DNA might be due either to contaminating normal tissue or to tumor heterogeneity. Arrows, point a t reduced or deleted alleles a t various loci. All patterns were re roducible for all patients studied. B, example of comparative m d i p l e x PCR in case 32. Internal control marker (D21S222) indicates that almost equal amounts of DNAs were analyzed in PCR and loaded in tumor (T)and lymphocyte (L) lanes. Case 32 shows loss of upper band at D13S289.
13q, 16q, 17q, and 18q in prostate cancer may be caused by a loss rather than a gain of DNA sequences.20.21 Moreover, Southern blot analysis has revealed allelic losses in prostate cancer on chromosome arms 7q, 8p, lOq, 13q, 16q, and 18q.2,3,5,22 In this study, using the RFLP markers c-met and p15-65, the informative patients in whom an imbalance was observed between the two alleles in the tumor DNA track, showed us that the nature of the imbalance at the 7q and 18q loci was classic LOH. When we compared our results with those obtained in the same region by CGH analysis or molecular techniques, they supported the notion that all the allelic imbalances observed in this study could be interpreted a s LOH. Our findings thus indicate that LOH on chromosome arms 7q, 8p, lOq, 13q, 16q, 17q, and 18q is a frequent event (table 11, in keeping with previous studies. In contrast, 5q LOH is infrequent. The most frequent site of LOH was the 8p22 band, 58% of the tumors exhibiting apparent allelic deletions. This is consistent with previous work in which Bova et al.4 identified a common region of deletion at 8p22 in prostate cancer. Likewise, the loss of 8p in various cancers, and particularly the 8p22 region,23 suggests that the putative TSG a t this site may be relevant to other human tumors. The other chromosomal site most frequently altered in the prostate cancers we studied was 16q, with 53% of LOH. This is in agreement with previous studies based on RFLP analysis of fewer cases.2.3 In a previous study we described LOH of the MET gene on chromosome band 7q31;5 Zenklusen et a1.: using microsatellite markers, recently reported a high frequency of LOH in the same region. In this study a loss of genetic information in the 7q31 region was found in 19 of 41 informative cases (46%). These results agree with Zenklusen’s findings. With regard to the long arm of chromosome 18,39%(16/41) of the tumors exhibited apparent allelic loss. We have previously described a common region of deletion on chromosome arm 18q5 and a likely target gene for LOH on 18q is the DCC (Deleted in Colorectal Carcinoma) gene; indeed, deletion of the DCC gene and an abnormal DCC gene product have both been identified in prostate cancer.24
quency of LOH in the same 17q region (34%.versus 47%’%.), possibly owing to the smaller number of cases in Gao’s study, the highest rate of deletion in the two studies was observed a t the D17S855 locus, a site recently shown to be intragenic to the BRCAl gene.15 Taken together with the data of Gao et a1.,6 our results support the idea that BRCAl is the probable target gene of the observed LOH on 17q, although additional studies are required to determine its role in both sporadic prostate tumors and familial prostate carcinoma. Frequent LOH on chromosome arm 1Oq in prostate tumors has been reported by other teams, who used RFLP to analyze fewer sample^.^.^ We found allelic loss on 1Oq in 23% (9 of 39) of informative patients, which is consistent with the aforementioned data. Positional cloning recently assigned the BRCA2 gene, which is responsible for inherited breast cancer, to the 13q12q13 region.25Arason et a1.26 proposed that genes predisposing to breast cancer also increase the risk of prostate cancer. As the BRCA2 gene might be involved in sporadic breast tum0rs,2~we postulated that the BRCA2 gene may also play an important role in sporadic prostate tumors. According to Wooster et al.25 the most likely location for BRCA2 is telomeric to D13S289 and proximal to D13S267. In this study, in which we used these two microsatellite markers, allelic loss was found in 10 of 39 informative patients (26%), suggesting the possible involvement of BRCA2 in prostate cancer. However, the highest rate of LOH was observed at the D13S267 locus (9/28; 32%). The TSG RB1 locus, which is more telomeric to D13S267, may also account for the LOH observed in this region. Even though conflicting results concerning the RB1 gene involvement have been reported,22.28 additional studies are required to determine which gene-BRCA2 or RB1-is the target TSG of the observed LOH, together with its role in prostate tumorigenesis. Chromosome arm 5q was notable for its lack of apparent involvement in prostate cancer; the background level of AI (10%) observed on the 5q21 region in this study may correspond to random events during the abnormal mitoses of cancer cells. Brewster et al.29 reported 20% of LOH within the APC gene (which is located in the same region) in prostate tumors. The discrepancy between the two studies might be due to a sampling bias (that is, the small number of informative cases in Brewster’s study), but the possibility of APC gene inactivation by point mutations or intragenic deletion cannot be ruled out, and neither can the involvement of the a catenin gene mapping to this region.30 Allelic loss on the different chromosome arms was analysed according to the tumor staging system of Bostwick et al.” The frequency of LOH at 7q31 and 13q12-13 in metastatic cancers (group C) was not higher than in locally confined tumors (group A), suggesting that inactivation of a putative TSG in these regions might occur a t an early stage of prostate development. The difference in the LOH rate in the 7q31 and 13q12-13 regions between group A and group C was not significant and could be due to the small number of cases studied; alternatively, a cell subclone with a 7q31 or 13q12-13 deletion in metastatic tumors (group C) could be masked by a dominant cell subclone bearing other genetic alterations specific to invasiveness. However, further investigations are required to confirm whether 7q31 and 13q12-13 deletions are less frequent in metastatic than in localized prostate tumors, because this would suggest that 7q31 and 13q12-13 deletions contribute to local growth of cancer cells, and are not involved in the process by which cells acquire metastatic potential. Our observations on 7q partially disagree with a recent report31 which show that 7q31.1 LOH is associated with tumor aggressiveness and progression. However when our results are considered alongside those of
CHROMOSOME ALLELIC IMBALANCE I N PROSTATE CANCER
2082
TABLE2. Allelic losses on 8 chromosome arms in prostate cancer a n d corresponding pathological features Number of allelic lossedinformative cases ( % )
5q
7q
8~
1%
13q
1%
17q
__
18q
(a I GROUP
A B
C
1/10i 10) 2/17(12) 1/11 (9)
8/13(62) 8/16150) 3/12( 2 5 )
7/11164) 8/17 (47) 7/10(70)
0/13 ( 0 ) 4/14(29) 5/12(42)
4/11( 3 6 ) 5/17(29) 1/11 (9)
4/9 (44) 8/14(571 519 (56)
5/12142) 2/13(15) 4/17(24) 8/17(47) 4/11(36)____ 6/11(55)
(a1 groups according to Bostwick et al.”: group A: patients with disease limited t o the prostate; group B: patients with local extracapsular extension; group C: patients with regional lymph node involvement or distant metastases.
Zenklusen e t al.7 on clinically localized prostate tumors, although there was no tendency towards specific coalterthey support the idea t h a t this genetic alteration on chro- ations. Interestingly 3 tumors in group A showed deletion in mosome band 7q31 might be a n early event in prostate only 1region, and in each case the allele loss was a t 7q31. We observed a positive link between the number of AI and the tumorigenesis. Similar rates of LOH were observed in group C and group Gleason score (4 of 14 tumors with grade 5 6 had only one A on chromosome arms 8p, 16q and 17q (table 2). Interest- region of LOH, whereas more than one region of LOH was ingly, the loss of 8 p was detected in a considerable number observed in 10 of 11 grade 2 8 tumors). Finally, our findings suggest that inactivation of a putative of tumors in each group, and particularly in locally contined tumors (7/11; 64%)suggesting t h a t this genetic al- TSG on 7q and 13q is an early event in prostate tumorigenteration might be a relatively early event, even though it esis. Moreover, LOH observed on 8p, 16q and 17q may occur at a n earlier stage than LOH on 1Oq and 18q, which appears has been linked to invasiveness.32.33 The E-cadherin gene located at 16q22.1,34 the product of to be related to invasiveness in prostate cancer. which plays an important role in cell-cell adhesion processes, We are indebted to Dr, B. vogelstein for providing probe might be the target of the observed LOH on 16%as suggested p15-65. The pmetH probe was obtained from the American by Isaacs et al.35 Indeed, reduced E cadherin expression is a Tissue Type Culture ~ ~ l l ~ ~ ~ i ~ ~ , powerful predictor of poor outcome, in terms of both disease progression and overall survival in prostate cancer.36 However, the incidence of LOH observed on 16q in our study was REFERENCES 1, Vogelstein, B., Fearon, E , R,, Hamilton, S. R., Kern, s. E,, high, irrespective of differences in clinical and pathological stage. The high frequency of distal 16q deletion in clinically Preisinger, A. C., Leppert, M., Nakamura, Y . , White, R., localized prostatic carcinomas (group A and group B) is in Smith, A. M. and Bos, J. L.: Genetic alterations during agreement with a recent FISH analysis.37 More studies of colorectal-tumor development. N. Engl. J. Med., 319: 525, prostate tumors are thus required to refine this 16q region. 1988. Analysis of results in the light of findings by Cher et a1.37 2. Carter, B. S., Ewing, C. M., Ward, W. S.,Treiger, B. F., Aalders, suggested that, in addition to the putative involvement of the T. W., Schalken, J. A., Epstein, J. I. and Isaacs, W. B.: Allelic E-cadherin gene at 16q22.1, inactivation of a n additional loss of chromosomes 16q and 1Oq in human prostate cancer. Proc. Natl. Acad. Sci. U.S.A., 87: 8751, 1990. TSG, at 16q24’2-qter7 might be a early event in 3, Bergerheim, U,, Kumini, K., Collins, P, V, and E b a n , p,: Deprostate tumorigenesis. Likewise, the loss of 16q, particuletion mapping of chromosomes 8, 10, and 16 in human prostatic carcinoma. Genes Chrom. Cancer, 3:215, 1991. lady the 16q22.1 and 16q24.3 regions, in breast and hepatocellular carcinomas,3s-40 suggests that Putative TSGs at 4. Bova, G. S., Carter, B. S., Bussemakers, M. J., Emi, M., these sites may be relevant to other human tumors. Fujiwara, Y., Kiprianou, N., Jacobs, S. C., Robinson, J. C., Our results on 17q agree with those of Gao et a1.,6 and Epstein, J. I., Walsh, P. C. and Isaacs, W. B.: Homozygous further indicate that the inactivation of the BRCAl gene deletion and frequent allelic loss of chromosome 8p22 loci in might be a relatively early event in human prostate cancer. human prostate cancer. Cancer Res., 53: 3869, 1993. The only significant relationship was observed between 5. Latil, A., Baron, J-c., Cussenat, o., Fournier, G., soussi, T., Boccon-Gibod, L., Leduc, A,, Rouesse, J. and Lidereau, R.: 1Oq LOH and metastatic cancers (group C) (p = 0.03). The Genetic alterations in localized prostate cancer: identification MXIl gene located at 10q24-q25, the product of which negaof a common region of deletion on chromosome 18q. Genes tively regulates MYC oncoprotein activity, might be a likely Chrom. Cancer, 11: 119, 1994. target gene for LOH on indeed, a recent study indicates 6. Gao, X., Zacharek, A,, Salkowski, A,, Grignon, D. J . , Sakr, W., that MXIl is a frequent target for mutation in prostatic Porter, A. T. and Honn, K. V.: Loss of heterozygosity of BRCAl carcinoma with deletion in this region and suggest a role for and other loci on chromosome 17q in human prostate cancer. this gene in desease pathogenesis or, more likely, in its Cancer Res., 55: 1002, 1995. evolution.4*The close link between the loss of 18q and inva7. Zenklusen, J. C., Thompson, J. C., Troncoso, P., Kagan, J. and sive phenotypes (group C) also suggests that this genetic Conti, C. J.: Loss of heterozygosity in human primary prostate carcinomas: a possible tumor suppressor gene a t 7q31.1. Canalteration, as well as 1Oq LOH, might be a relatively late cer Res., 54: 6370, 1994. event in prostate tumorigenesis. DCC, which encodes a mob 8 , Weber, J , L, and May, p, E,: Abundant class of human DNA ecule with a high degree of homology with neural cell adhepolymorphisms which can be typed using the polymerase sion molecule, might be the target of the observed LOH on chain reaction. Am. J. Hum. Genet., 44: 388, 1989. 18q, as suggested by Gao’s study.“ 9. Mellinger, G. T., Gleason, D. and Bailar, J.: The histology and The LOHs found in this study on chromosome arms 7q, 8p, prognosis of prostatic cancer. J. Urol., 97:331, 1967 111. lOq, 13q, 16q, 17q and 18q do not appear to be random events 10. Schroder, F. H., Hermanek, P., Denis, L., Fair, W. R., because of their high frequencies. These data are consistent Gospodarowicz, M. K. and Pavone-Macaluso, M.: The TNM with previous reports and indicate that a relatively large classification of prostate cancer. Prostate, Suppl 4: 129, 1992. number of chromosome loci play a part in the etiolom and 11. Bostwick, D. G., Myers, R. P. and Oesterling, J. E.: Staging in prostate cancer. Semin. Surg. Oncol., 1 0 60, 1994. Of this type’ Moreover* most Of these ge- 12. Tomfohrde, J., Wood, S.,Schertzer, M., Wagner, M. J., Wells, netic alterations are already present in locally confined tuD. E., Parrish, J . , Sadler, L. A,, Blanton, S. H., Daiger, S. P., mors, except for 1Oq LOH. In group A (13 locally confined Wang, Z . , Wilkie, P.J . and Weber, J. L.: Human chromosome tumors) and group C (12 metastatic cancers), 5 tumors 8 linkage map based on short tandem repeat polymorphisms: showed 3 or more different chromosomal regions of deletion, effect of genotyping errors. Genomics, 14: 144, 1992.
CHROMOSOME ALLELIC IMBAIA N C E IN PROSTATE CANCER
2083
13. Weissenbach, J., Gyapay, G., Dib, C., Vignal, A., Morissette, J., 26. Arason, A,, Barkardottir, R. B. and Egilsson, V.: Linkage analMillasseau, P., Vaysseix, G. and Latbrop, M. A.: Secondysis of chromosome 17q markers and breast-ovarian cancer in generation linkage map of the human genome. Nature, 359: Iceland families, and possible relationship to prostatic cancer. 794, 1992. Am. J . Hum. Genet., 52: 711, 1993. 14. Gyapay, G., Morissette, J., Vignal, A., Dib, C., Fizames, C., 27. Kerangueven, F., Allione, F., Noguchi, T., Sobol, H., Jacquemier, Millasseau, P., Marc, S., Bernardi, G., Lathrop, M. and J. and Birnbaum, D.: Patterns of loss of heterozygosity at loci Weissenbach, J.: The 1993-94 Genethon human genetic linkfrom chromosome arm 13q suggest a possible involvement of age map. Nature Genetics, 7: 246, 1994. BRCA2 in sporadic breast tumors. Genes Chrom. Cancer, 1 3 15. Miki, Y., Swensen, J., Shattuck-Eidens, D., Futreal, P. A,, 291, 1995. Harshman, K., Tavtigian, S., Liu, Q., Cochran, C., Bennett, 28. Sarkar, F. H., Sakr, W. A., Li, Y. W., Macoska, J., Ball, D. E. and L. M., Ding, W., Bell, R., Rosenthal, J., Hussey, C., Tran, T., Crissman, J. D.: Analysis of retinoblastoma (RB)gene deletion McClure, M., Frye, C., Hattier, T., Phelps, R., Haugen-Stano, in human prostatic carcinomas. Prostate, 21: 145, 1992. A,, Katcher, H., Yakumo, K., Gholami, Z., Shaffer, D., Stone, 29. Brewster, S. F., Browne, S. and Brown, K.: Somatic allelic loss at S., Bayer, S., Wray, C., Bogden, R., Dayananth, P., Ward, J., the DCC, APC, nm23-H1 and p53 tumor suppressor gene loci Tonin, P., Narod, S., Briston, P. K., Noms, F. H., Helvering, in human prostatic carcinoma. J. Urol., 151: 1073, 1994. L., Morrison, P., Rosteck, P., Lai, M., Barrett, J . C., Lewis, C., 30. McPherson, J . D., Morton, R. A., Ewing, E. M., Wasmuth, J . J., Neuhausen, S., Cannon-Albright, L., Goldgar, D., Wiseman, Overhauser, J., Nagafuchi, A,, Tsukita, S. and Isaacs, W. B.: R., Kamb, A. and Skolnick, M. H.: A strong candidate for the Assignment of the human a catenin gene ( C T ” A 1 ) to chrobreast and ovarian cancer susceptibility gene BRCA1. Science, mosome 5q21-q22. Genomics, 19 188, 1994. 266 66, 1994. 31. Takahashi, S., Shan, A. L., Ritland, S. R., Delacey, K A., 16. McGrogan, D., Levy, A,, Bostwick, D., Wagner, M., Wells, D. and Bostwick, D. G., Lieber, M. M., Thibodeau, S. N. and Jenkins, Bookstein, R.: Loss of chromosome arm 8p loci in prostate R. B.: Frequent loss of heterozygosity a t 7q31.1 in primary cancer: mapping by quantitative allelic imbalance. Genes prostate cancer is associated with tumor aggressiveness and Chrom. Cancer, 1 0 151, 1994. progression. Cancer Res., 5 5 4114, 1995. 17. Cairns, P., Tokino, K., Eby, Y. and Sidransky, D.: Homozygous 32. Suzuki, H., Emi, N., Komiya, A,, Fujiwara, Y., Yatani, R., deletions of 9p21 in primary human bladder tumors detected Nakamura, Y. and Shimazaki, J.: Localization of a tumor by comparative multiplex polymerase chain reaction. Cancer suppressor gene associated with progression of human prosRes., 54: 1422, 1994. tate cancer within a 1.2 Mb region of 8p22-p21.3. Genes 18. Hudson, T. J., Engelstein, M., Lee, M. K., Ho, E. C., Rubenfield, Chrom. Cancer, 13: 168, 1995. M. J., Adams, C. P., Housman, D. E. and Dracopoli, N. C.: 33. Trapman, J., Sleddens, H. F. B. M., van der Weiden, M. M., Isolation and chromosomal assignment of 100 highly informaDinjens, W. N. M.,Konig, J . J . and Schroder, F. H.: Loss of tive human simDle seauence reueat Genom- .uolvmoruhisms. heterozygosity of chromosome 8 microsatellite loci implicates a ics, 13: 622, 1992. candidate tumor suppressor gene between the loci D8S87 and 19. Watanabe, M.. Imai, H.. Shiraishi, T., Shimazaki, J., Kotate, T. D8S133 in human prostate cancer. Cancer Res., 54: 6061, and Yatani,’ R.: Microsatellite instability in human prostate 1994. cancer. Br. J . Cancer, 72: 562, 1995. 34. Natt, E., Magenis, R. E., Zimmer, J., Mansouri, A. and Scherer, 20. Cher, M. L., MacGrogan, D., Bookstein, R., Brown, J . A., G.: Regional assignment of the human loci for uvomorulin and Jenkins, R. B. and Jensen, R. H.: Comparative genomic hychymotrypsinogen B with the help of two overlapping delebridization, allelic imbalance, and fluorescence in situ hybridtions on the long arm of chromosome 16. Cytogenet. Cell Genization on chromosome 8 in prostate cancer. Genes Chrom. et., 5 0 145, 1989. Cancer, 11: 153, 1994. 35. Isaacs, W. B., Bova, G. S., Morton, R. A,, Bussemakers, M. J., 21. Visakorpi, T., Kallioniemi, A. H., Syvanen, A-C., Hyytinen, E. R., Brooks, J . D. and Ewing, C. M.: Molecular genetics and chroKarhu, R., Tammella, T., Isola, J . J . and Kallioniemi, 0-P.: mosomal alterations in prostate cancer. Cancer, s u p p l 7 2004, Genetic changes in primary and recurrent prostate cancer by 1995. comparative genomic hybridization. Cancer Res., 5 5 342, 36. Umbas, R., Isaacs, W. B., Bringier, P. P., Schaafsma, H. E., 1995. Karthaus, H. F. M., Oosterhof, G. 0. N., Debruyne, F. M. J . 22. Phillips, S. M. A., Barton, C. M., Lee, S. J., Morton, D. G., and Schalken, J. A,: Decreased E-cadherin expression is assoWallace, D. M. A., Lemoine, N. R. and Neoptolemos, J. P.: Loss ciated with poor prognosis in patients with prostate cancer. of the retinoblastoma susceptibility gene (RB1) is a frequent I Cancer Res., 54: 3929, 1994. and early event in prostatic tumorigenesis. Br. J . Cancer, 7 0 37. Cher, M. L., Ito, T., Weidner, N., Carroll, P. R. and Jensen, R. H.: 1252, 1994. Mapping of regions of physical deletion on chromosome 16q in 1 23. Fujiwara, Y., Ohata, H., Emi, M., Okui, K, Koyama, K, prostate cancer cells by fluorescence in situ hybridization Tsuchiya, E., Nakajima, T., Monden, M., Mori, T., Kurimasa, (FISH). J . Urol., 153: 249, 1995. I A. and Nakamura, Y.: A 3-Mb physical map of the chromosome 38. Tsuda, H., Zhang, W., Shimosato, Y., Yokota, J.. Terada, M., region 8p21.3-p22, including a 600-Kb region commonly deI Sugimura, T., Miyamura, T. and Hirohashi, S.: Allele loss on leted in human hepatocellular carcinoma, colorectal cancer, chromosome 16 associated with progression of human hepatoand non-small cell lung cancer. Genes Chrom. Cancer, 1 0 7, cellular carcinoma. Proc. Natl. Acad. Sci. U S A . , 87: 6791, 1 1994. 1990. 1 24. Gao, X., Honn, K. V., Grignon, D., Sakr, W. and Chen, Y. Q.: Frequent loss of expression and loss of heterozygosity of the 39. CletonJansen, A-M., Moerland, E. W., Kuipers-Dijkshoorn, N. J . K., Callen, D. F., Sutherland, G. R., Hansen, B., Devilee, 1 putative tumor suppressor gene DCC in prostatic carcinomas. P. and Cornelisse, C. J.: At least two different regions are Cancer Res., 53: 2723, 1993. involved in allelic imbalance on chromosome arm 16q in breast 25. Wooster, R., Neuhausen, S. L., Mangion, J., Quirk, Y., Ford, D., cancer. Gene Chrom. Cancer, 9 101, 1994. Collins, N., Nguyen, K., Seal, S., Tran, T., Averill, D., Fields, p., Marshall, G., Narod, S., Lenoir, G. M., Lynch, H., 40. Tsuda, H., Callen, D. F., Fukutomi, T., Nakamura, Y. and Hirohashi, S.: Allele loss on chromosome 16q24.2qter occurs Feunteun, J., Devilee, P., Cornelisse, C. J., Menko, F. H., D a b , I frequently in breast cancers irrespectively of differences in p. A., Ormiston, W., McManus, R., Pye, C., Lewis, C. M., phenotype and extent of spread. Cancer Res., 54: 513, 1994. Cannon-Albright, L. A., Peto, J., Ponder, B. A. J., Skolnick, I M. H., Easton, D. F., Goldgar, D. E. and Stratton, M. R.: 41. Eagle, L. R., Yin, X., Brothman, A. R., Williams, B. J., Atkin, N. B. and Prochownik, E. V.: Mutation of the MXIl gene in Localization of a breast cancer susceptibility gene, BRCA2, to prostate cancer. Nature Genetics, 9 249, 1995. chromosome 13q12-ql3. Science, 265: 2088, 1994. ~
,
,
,
I
i i b