Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
[CANCER RESEARCH55, 5217-5221, November 15, 1995] Advances in Brief p53 Point Mutation and Survival in Colorectal Cancer Patients' Hak-Su Goh, Jocelyn Yao, and Duncan R Smith2 Colorectal Cancer Research Laboratory, Department of Colorectal Surgery, Singapore General Hospital, Outram Road, Singapore 0316, Republic of Singapore Abstract We have examined the relationship between point mutation of the p53 tumor suppressor gene and survival in colorectal cancer patients We found that patients with tumors harboring mutated p53 genes showed a significantly poorer prognosis than did those patients with genes without point mutations, and, moreover, patient response to postoperative thera pies depended significantly on mutation status in both adjuvant and palliative treatment cohorts. However, not all point mutations were the same functionally; point mutations within the conserved domains of the p53 tumor suppressor gene were inherently more aggressive than tumors with point mutations outside ofthese domains, and mutations ofcodon 175 were particularly aggressive. These results suggest that knowledge of a patient's p53 status, both with respect to the presence of point mutations and to the specific nature of the lesion, may be required to accurately predict both the course of the disease and the response of the disease to postoperative therapeutic interventions, especially those on the induction of apoptosis in the neoplastic cell. therapies based Introduction The prediction of the biological behavior of a tumor is of primary importance in the management of colorectal cancer patients. Many staging systems have been introduced in an attempt to provide accu rate prognostication, but, in general, these provide accurate informa tion about cohorts of patients rather than individuals. Staging systems are often used as a basis for deciding patient management, especially with regard to additional therapies such as radiotherapy, chemother apy, or combinations thereof (1). The mainstay of treatment for colorectal carcinoma is surgical resection (1). The intent of surgical intervention may be either cura tive or palliative, depending on the spread of the disease and the tumor clearance during surgery. Selected patients from both curative model system that tumors with functional to treatment became unresponsive whether and by grants from the Shaw Foundation, of the acquisi the Lee requests for reprints should or metachronous, in the colon were excluded. consisted of single of the colorectum, Patients at the time of surgery with dysplastic were also excluded. Dukes' original staging modification of (13). DNA Sequence Analysis. Templateswere analyzedby directsequencing of cDNA-PCR the dsCycle products Sequencing corresponding system to exons 4—9of the p53 gene by using (GIBCO-BRL, Gaithersburg, MD). Sequenc ing was carried out by using a pane] of p53 cDNA-specific primers (sequences available on request). Point mutations were confirmed analysis by DNA sequence on both DNA strands. Data Analysis and Statistics. Patient follow-up was determined to be the time between surgery and the last departmental contact (scheduled follow-up, or telephone package contact) (interquartile (SPSS, range: or patient death. Median 14, 34 months). patient Survival follow up data was ana plots (14) and the log rank test by using the SPSS Inc., Chicago, IL). Models were tested by using Cox regression analysis, with entry into the model dependent on P < 0.05. Data Foundation, and the Singapore Totalisator Board. 2 To whom elsewhere lyzed by Kaplan-Meier 18 U.S.C. Section 1734 solely to indicate this fact. supported The patient cohort is essentially as described pre Tumors were staged Dukes' A—Daccording to Tumbull's computer charges. This article must therefore be hereby marked advertisement in accordance with was synchronous adenomas mail response, Received 8/9/95; accepted 10/2/95. The costs of publication of this article were defrayed in part by the payment of page work and Tumors. was 24 months tion of p53 point mutations (5). Given that the p.5.3tumor suppressor gene is one of the most commonly mutated genes in a wide variety of tumors (6—8)these results are potentially of great clinical importance. I This and Methods Patients p.53 genes respond because although the basis of this is not readily apparent (1 1). Nearly 98% of point mutations of the p53 gene detected to date fall in an approximately 600-nucleotide-long stretch of the p53 message, corresponding roughly to exons 4—9of the p53 gene (6, 7). The p53 protein shows a high degree of evolutionary conservation, but, in particular, five domains (domains I—V)show almost complete con servation from trout to man (12). Four of these domains (domains II—V)occur within the 200-amino acid stretch encoded by the region in which almost all point mutations occur (6, 7). We have shown previously that point mutation of the p53 gene is associated with disease progression, and in particular, with lymphatic dissemination (9); in this study we wished to examine the influence of PS3 point mutation in determining both the course of the disease and how such lesions influence the response to postoperative therapies. viously (9). All tumors selected for DNA sequencing adenocarcinomas, and patients with multiple carcinomas to radiation and chemotherapeutic agents, whereas those tumors with out functional p53 genes do not. Moreover, tumors that were initially responsive (8). Not all mutantproteinshavethe sameoncogenicpotential, Materials palliative cohorts may be judged to be suitable for additional postop erative treatment in an attempt to either prevent the growth of sec ondary cancers or to reduce the tumor burden of the patient. Such postoperative therapies may consist of either radiotherapy, chemo therapy, or a combination of both. Currently, such postoperative therapies are believed to act through the induction of apoptosis, or programmed cell death, in the neoplastic cells (2), and several studies suggest that induction of apoptosis is dependent on the presence of functional copies of the p.53 tumor suppressor gene (2—4).More recently, Lowe et a!. (5) have shown in a mouse Point mutation of the p53 gene occurs in approximately 50% of colorectal carcinomas (6—8) and has recently been shown to be associated strongly with lymphatic dissemination (9) and possibly with a poorer patient prognosis (10). The vast majority of point mutations detected in p.5.3genes represent missense mutations, which result in amino acid substitutions in the p53 protein (6—8). Such mutant proteins have been shown to have an array of altered biolog ical and biochemical properties, including an extended half-life and the ability to cooperate with activated ras genes to transform cell lines be addressed. were entered by using the Forward:Wald method of entry. Results The p53 point mutation status was determined for 192 colorectal adenocarcinomas by single-stranded conformational polymorphisms, with the cohort consisting essentially of tumors described previously (9). Point mutations were detected in 57% (109 of 192) of cases. Kaplan-Meier survival plots (14) were generated for patients with or without point-mutated p53 genes (Fig. 1). Log rank analysis of this 5217 Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research. @ @, —@ p53 AND PROGNOSIS A. Suiv1v@ @tIa@ D. Swvival @pS3pok*ai@1os @iI[Tz::z@:I L::i-iI ., @io@ trill p@isu*s with ——@-----@--- S.' S.. I .$ 3 I .3 .3 @ tr@p— II 3• —@-——---—-‘ I r@-@Ms@) E. Survival fimctlo B. Suavivalftanctios@ krco..srved aid @ @ i_ I I @I @ir@whcte @‘s aiwive @I* ., q_ T@;. TTIM 3 .3 .3 I SI a a • a a ?0@ F. Survivslñmctions f@p@ie,@s where tr@M is pslllMivsin heeiit C. Survival fbnctioai ofcodon 175 aid allothermsgalons I I I I F.— F.— Fig. 1. Mutation status and survival. A, Kaplan-Meier survival analysis for 192 patients with respect to p53 mutation status. Deaths due to causes unrelated to colorectal cancer were treated as censored events. The two curves are significantly different (P = 0.0054 by log rank analysis). , no p53 point mutation detected. —, point mutation detected. B, Kaplan-Meier plots for 53 patients with characterized mutations of the p53 gene. Patients are analyzed after dividing into two groups based on the presence (—)or absence ( of point mutations in the conserved domains. The two groups are significantly different (P = 0.0201 by log rank analysis). C, Kaplan-Meier survival plot for 53 patients with characterized point mutations of the p53 gene. Patients are divided into two groups, those with mutations of codon 175 (—) and those with other mutations ( ). The two groups are significantly different (P = 0.0007 by log rank analysis). D, Kaplan-Meier plots for 54 colorectal cancer patients who underwent postsurgical therapies and includes both palliative and adjuvant therapy patients. Patients are divided into two groups based on the presence (—)or absence ( ) of a point mutation in the p53 gene. The two groups are significantly different (P = 0.0006 by log rank analysis). E, Kaplan-Meier plots for colorectal cancer patients who underwent curative resection and therapy. Patients are divided into two groups based on the presence (—) or absence ( ) of a point mutation in the p53 gene. The two curves are significantly different (P = 0.038 by log rank analysis). F, Kaplan-Meier plots for colorectal cancer patients who underwent palliative treatment. Patients are divided into two groups based on the presence (—) or absence ( ) of a point mutation in the p53 gene. The two curves are significantly different (P = 0.0048 by log rank analysis). @ plot showed a significant association between point mutation and a in the nature of point mutations, we examined 50% of the mutation positive tumors from the above cohort to determine the exact nature of poorer patient prognosis (P = 0.0054). Cox regression analysis was the point mutation. Hence, 53 templates were subjected to a full DNA used to determine the contribution of several factors toward patient sequence analysis. Results are shown in Table 1. The point mutation survival. Factors available for entry into the model included Dukes' spectrum determined is representative of the local population (15), Stage, differentiation, and mutation status, as well as patient age and which is consistent with, but not identical to, that published on other sex. Entry into the model was via the Forward:Wald method and population cohorts (6, 7), and as such probably reflects regional conditional on P < 0.05. The final model contained the terms for influences in mutation events (7, 15). Fig. 2 shows an analysis of point Dukes' Stage (P < 0.0001) and mutation status (P = 0.0310). Overall mutation distribution with respect to lymph node involvement. We significance of the model was = 65.001 (P < 0.0001). note that point mutations occurring within the highly conserved do To determine whether there were qualitative differences occurring 5218 Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research. p53 AND PROGNOSIS Table 1 p53 point mutations characterized in 53 colorectal carcinomas. Characterization of 53 carcinomas carrying mutated p53 genes and associated clinicopathological parameters. CodonLab no.Base changeAmino acid change'@Stage―LN'Dm DF'Follow-up@CD@111333CTGtoCCGLeutoProA———45N12781TCCtoCCCSertoProC+——52Y132184AAG to AGOLys ArgC+—+{15}YI 35784TGC 1}Y141584TGC PheC+—+35Y151185CCC ThrB———10N157242GTC PheB———22N157672GTC PheB———32N162684ATC to to to to to StopA---31NI to AATC ATSdDm to to TyrD+—+{1 to to to to TACCys ‘Il'CCys ACCPro ‘Il'CVal T1'CVal 66138AGC to AGGC StopB———53N168194CACtoCGCHistoArgB———32NI 73825GTG MetC+——{1}Y1751CGC to ATOVal to HisD+++{20}Y17540CGC to CACMg to HisC+—+{14}Y17568CGCtoCACArgtoHisC+—+{36}Y175120CGCtoCACArgtoHisD+——{5}Y175149CGC to CACMg to HisD+++{12}Y175246CGC HisD+——{6}Y1 75731CGC HisA———29Y175857CGC to CACMg to to to CACArg to CACArg to HisD+++{5}Y175782CGCtoCTCArgtoLeuC+——25Y1755CGCtoCTCArgtoLeuD++—{5}Y17591COC to to CACArg HisC+—+{24}Y179838CAT to CACMg to LeuC+——25Y179864CAT to CITHis to LeuC+——24Y18112CGC to CTTHis to CysB———59Y197147GTG to TGCArg to GlyC+——46N237190ATG to 000Val to IleC+——51Y237356ATGtoAAGMettoLysA———43Y238630TGT to ATFMet to to TATCys TyrA———34Y242236TGC to to ‘fl'CCys PheD+++{10}Y245297GGCtoAGCGlytoSerC+——46Y245792GGCtoAGCGlytoSerD+—+19Y248161CGG to TrpC+—+{32}Y248199COG GlnA---50Y248257CGG GlnC+—+{36}Y248654COG TrpD++-1Y248852COG GinD+++{5}Y249332AGO TipD+++{7}Y257823Cr0 GluA---20Y266840GGA to TGGMg to CAGMg to CAGMg to TOGArg to CAGMg to to TGGArg to to to to to to to CAGLeu to GAAGly to ATGVal to TOTMg to TATCys to TOOMg to TOGArg to TOGArg GluC+——24N272681GTG MetD+++29Y273817COT CysD+—+{23}Y275321TOT TyrC+—+{12}Y28275COG TrpC+—+{36}Y282109COG TrpC+——{36}Y282216COG TrpC+——47Y282660COG TrpD-++4Y282765COO LeuD+++22Y282835COO TrpB---8Y28285 1COG TrpC+—+24Y282903COG to to to to to to to to to to to TOGArg to Cr0Arg to TOGMg to TOGArg TrpD+++{2}Y285724CAOtoAAOAsptoLysA———29Y to TOGArg a Amino acid b Stage, Dukes' C LN, presence d Dm ATh, change, amino stage (+) presence of tumor or absence (+) acid change in three-letter code. according to Tumbull's modification (—) of or absence detectable lymphatic (—) of detectable e Dm DF, presence (+) or absence (—)of detectable to to involvement distant organ (13). at time metastasis of surgery. at the time of surgery. distant organ metastasis at time of surgery or during subsequent follow-up. 1Follow-up, patient follow-up in months since surgery. Numbers in brackets indicate patient is deceased. a CD, point mutation is in (Y) or outside (N) conserved domains. mains of the p53 gene show a significant association with lymph node involvement (Table 2; P = 0.001 by Fishers' exact test). Mutations occurring outside of these regions show a much lower rate of lymph node involvement [Fig. 2; 22% (2 of 9) compared to 80% (35 of 44)]. There is no association between the presence of point mutations in the conserved domains and the presence of distant metastasis at the time of surgery (Table 2). However, when all cases are analyzed during the follow-up period, there is a strong association between point mutation in the conserved domains and patients who have or develop distant organ metastasis (Table 2). Indeed, as yet, none of the patients with mutations outside of the highly conserved regions has developed distant metastasis, although the sample population of mu tants outside of the conserved regions is relatively small (nine pa tients). In contrast 52% of patients with point mutations in the con served regions have been diagnosed with metastatic disease dissemination. Similarly, of the 22 deaths in the mutation-positive cohort, all have occurred in patients with mutations in the conserved regions of the p53 gene. Patients with mutations in the conserved regions of the p53 gene showed a significantly poorer prognosis than did patients with mutations outside of these regions (Fig. 1; P = 0.02, as assessed by log rank analysis). In particular, mutations of codon 175 were associated with aggressive cancers. We note that 82% (9 of 11) of patients with mutations in codon 175 had died of causes related to cancer, in contrast to 3 1% of patients with mutations at other bases (see Fig. 1; P = 0.0007). Lesions of the p53 gene have been implicated recently in the response of tumors to radiation and chemotherapy (5). Therefore, we selected a new cohort of 54 patients who had undergone postsurgical 5219 Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research. p53 AND PROGNOSIS @ anim a@ Discussion + @ ++ @1@ - ICrYQOSYGFR LcwLH@iA@Ic@ svrmsp4u Dou@nfl m@n n@ I 10 120 - wPcrmvkAM 130 140 ISO The 160 + + + + + + + + -+ + + - + AIYKQSQHMT LRVEYLDOftN 1FRNSVVvPY Do@@inill 170 180 190 200 210 220 + However, + + + ________________________ ++ + ++ EFS'EVGSDCITI*'NYMØJS S@GGM?@P @0 230 260 : sGNIi.GRNs[@ 270 EVRVCAL@(_@iii@k KKG@1*ELP 290 300 Fig. 2. Lymphatic dissemination and p53 point mutation. Partial sequence of the p53 protein (amino acids 101—300)showing the distribution of 53 mutants detected by DNA sequence analysis (Table 1). +, presence of lymphatic dissemination; —,absence of lymphatic dissemination. Conserved domains are boxed. @ @ @ properties of mutant p53 has been equivalent observed (1 1), and that not the data all point generated mutations by DNA are sequenc codon 175 mutants being particularly aggressive. within + +- 210 biochemical mutations occurring within these regions. Moreover, point mutations + +++ and ing support this. We find that point mutations occurring within the evolutionarily conserved domains show a significantly poorer prog nosis than do those mutations occurring outside these domains. We further note that lymphatic dissemination, and the occurrence of subsequent distant organ metastasis are strongly associated with point ILTIrn..Ebs ; + biological it functionally Doi@n IV 230 specific proteins have been the subject of intense investigation. The observa tion that nearly 80% of mutations of the p53 gene are point mutations leading to amino acid substitutions (missense mutations) has lead to the speculation that such mutants may confer an additional growth advantage over and above that resulting from the loss of suppressor function (6). This speculation is apparently supported by experiments that show a “gain of function―for p53 mutants (16). These observa tions would suggest that tumors with p53 point mutations would show a growth advantage over tumors with no p53 mutation. In patient terms, a growth advantage of a tumor would correlate to a poorer patient prognosis, and we show here that patients with p53 point mutations show a significantly poorer prognosis than do patients who do not show such mutations. radiotherapy or chemotherapy (or both). Tumors from these patients were analyzed for point mutations of the p53 gene by single-stranded conformational polymorphisms as described elsewhere (9, 15). Point mutations were detected in 54% (29 of 54) of cases. Analysis of patient survival with respect to point mutation was then undertaken by Kaplan-Meier plots (Fig. 1). The two survival curves are significantly different (P = 0.0006 by log rank analysis). Patients with point mutations of the p53 gene showed a significantly higher patient mortality rate than did those patients whose tumors did not show evidence of point mutations occurring within the p53 gene. However, this cohort of patients consists of patients who underwent either palliative or curative resections. The data was therefore re analyzed with respect to the surgical intent. When only curative resections were considered, a smaller but still significant difference was found between patients with and without mutations (Fig. 1). Again, patients with point mutations of the p.53 gene showed a significantly poorer prognosis than did patients without detectable point mutations [number of patients, 37; mutation rate, 51% (19 of 37); P = 0.038 by log rank analysis]. A more substantial difference was seen in the palliative treatment cases (Fig. 1). The response of patients to palliative therapy was markedly dependent on the p53 status of the primary colorectal cancer. All of the patients undergoing palliative postoperative treatment who had point mutations of the p53 gene have died subsequently, in contrast to only one patient who died from the cohort of patients not showing detectable point mutation [number of patients, 17; mutation rate, 59% (10 of 17); P = 0.0048, log rank analysis]. these regions clearly show differential aggressiveness, with In this report, we have shown three levels of increasing risk for colorectal cancer patients: (a) patients with point mutations of the p53 gene show a poorer prognosis than do patients without point muta tions; (b) patients with point mutations in the conserved domains show a poorer prognosis than do those with mutations outside these domains; and (c) patients with mutations of codon 175 show the poorest prognosis of all. Detecting point mutations of the p53 gene is technically demanding and determining the specific nature of the lesion by DNA sequencing is even more so. However, work from other areas has shown that routine screening of colorectal carcinomas for p53 point mutations may become a prime prerequisite in deter mining patient management. Recently, Lowe et a!. (5) have shown that the status of the p53 gene plays an integral role in determining a tumor's responsiveness to adjuvant therapies. Tumors that have no PS3 or have mutated p53 evince treatment resistance to both chemo therapy and radiotherapy, probably by disruption of the cells' apop totic pathways (5). Moreover, initially sensitive tumors may become resistant by the selection of cells containing mutated p53 (5). Indeed, in confirmation of this, in the two experimental cohorts reported here, the response of patients with point-mutated p53 genes to therapies that are believed to act via the induction of apoptosis is significantly Table 2 Conserved domain mutations and disease dissemination. The distribution of p53 mutations was analyzed by Fishers' exact test in respect to disease dissemination both at the time of surgery and during patient follow-up. LN-― LN+― Dm ATS@c 9 nCD― P (Fishers') a absence b LN+, presence DM ATS+― Dm DF@e 35 32 12 7 2 P = 0.0018 9 0 of detectable of detectable involvement lymphatic at time involvement 23 9 0 P = 0.0033 P = NS' lymphatic Dm DF+― 21 of at time surgery. of surgery. CDm ATh—, absenceof detectabledistantorganmetastasisat time of surgery. d Dii. ATS+, e Dm DF—, presence absence of of detectable detectable distant distant organ organ metastasis metastasis at at time time of of surgery. surgery or during subsequent follow-up. 1Dm DF+, presence of detectable distant organ metastasis at time of surgery or during subsequent follow-up. gCD,mutation present inoneoftheconserved domains. ,, nCD, mutation S NS, not present outside of the conserved domain. significant. 5220 Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research. p53 AND PROGNOSIS poorer (asjudged by patient survival) than it is for patients who do not show evidence of point-mutated p53 genes in their primary colorectal adenocarcinomas. Hence, in colorectal cancer patients, point mutation of the p53 gene plays a pivotal role, in both determining, to some extent, the biological behavior of colorectal carcinomas, as well as modifying the effectiveness of postoperative therapies. As such, correcting defects of the p53 tumor suppressor gene would seem an ideal target for novel gene-based therapies for the treatment of meta static colon carcinoma. Acknowledgments We are grateful to Chui-Sien Chan for technical assistance and Stephanie Fook Chong for statistical advice. References 1. Kronborg, 0. Staging and surgery for colorectal cancer. Eur. J. Cancer, 29A: 575— 583, 1993. 2. Kerr, J. F. R., Winterford, C. M., and Harmon, B. V. Apoptosis: its significance in cancer and cancer therapy. Cancer (Phila.), 73: 2013—2026,1994. 3. Lawe, S. W., Ruley, H. E., Jacks, T., and Housman, D. E. p53-dependent apoptosis modulates the cytotoxicity of anticancer agents. Cell, 74: 957—967,1993. 4. Shaw, P., Bovey, R., Tardy, S., Sahli, R., Sordat, B., and Costa, J. Induction of apoptosis by wild type p53 in a human colon tumor-derived cell line. Proc. Natl. Acad. Sci. USA, 89:4495-4499,1992. 5. Lawe, S., Bodis, S., McClatchey, A., Remington, L., Ruley, H. E., Fisher, D. E., Houseman, D. E., and Jacks, T. p53 status and the efficacy of cancer therapy in vivo. Science (Washington DC), 266: 807-810, 1994. 6. Hollstein, M., Sidransky, D., Vogelstein, B., and Harris, C. C. p53 mutations in human cancers. Science (Washington DC), 253: 49—53,1991. 7. Greenblatt, M. S., Bennett, W. P., Hollstein, M., and Harris, C. C. Mutations in the p53 tumor suppressor gene: clues to cancer etiology and molecular pathogenesis. Cancer Res., 54: 4855-4878, 1994. 8. Levine, A. J., Perry, M. E., Chang, A., Silver, A., Dittmer, D., Wu, M., and Welsh, D. The role of the p53 tumor suppresor gene in tumorigenesis. The 1993 Walter Hubert Lecture. Br. J. Cancer, 69: 409—416, 1994. 9. Goh, H-S., Chan, C-S., Khine, K., and Smith, D. R. p53 and the behaviour of colorectal cancer. Lancet, 344: 233—234,1994. 10. Hamelin, R., Laurent-Puig, P., Olschwang, S., Jego, N., Maclain, B., Remivikos, Y., Girodet, J., Salmon, R. J., and Thomas, G. Association of p53 mutations with short survival in colorectal cancer. Gastroenterology, 106: 42—48,1994. 11. Vogelstein, B., and Kinzler, K. W. p53 function and dysfunction. Cell, 70: 523—526, 1992. 12. Soussi, T., Caron de Fromentel, C., and May, P. Structural aspects of the p53 protein in relation to gene evolution. Oncogene, 5: 945—952,1990. 13. Turnbull, R. B., Watson, F. R., and Spratt, J. Cancer of the colon: the influence of the no-touch isolation technic on survival rates. Ann. Surg., 166: 420—427, 1967. 14. Kaplan, E. L., and Meier, P. Nonparametric estimation from incomplete observations. J. Am. Stat. Assoc., 53: 475—481,1958. 15. Smith, D. R., Chan, C-S., and Goh, H-S. A detailed analysis of p53 point mutations in colorectal cancers in Singapore. Ann. Acad. Med. Singapore, 23: 803—809, 1994. 16. Dittmer, D., Pati, S., Zambetti, G., Chu, S., Teresky, A. K., Moore, M., Finlay, C., and Levine, A. J. Gain of function mutations in p53. Nat. Genet., 4: 42—46,1993. 5221 Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research. p53 Point Mutation and Survival in Colorectal Cancer Patients Hak-Su Goh, Jocelyn Yao and Duncan R. Smith Cancer Res 1995;55:5217-5221. Updated version E-mail alerts Reprints and Subscriptions Permissions Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/55/22/5217 Sign up to receive free email-alerts related to this article or journal. To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at [email protected]. To request permission to re-use all or part of this article, contact the AACR Publications Department at [email protected]. Downloaded from cancerres.aacrjournals.org on June 15, 2017. © 1995 American Association for Cancer Research.