TS表达-免疫组化

The Role of Thymidylate Synthase Expression inPrognosis and Outcome of Adjuvant Chemotherapyin Patients With Rectal CancerBy Patrick G. Johnston, Edwin R. Fisher, Howard E. Rockette, Bernard Fisher, Norman Wolmark,James C. Drake, Bruce A. Chabner, and Carmen J. AllegraPurpose: We assessed the prognostic importance ofthe level of thymidylate synthase (TS) expression in pa-tients with primary rectal cancer and whether, for Dukes'B and C cancer patients, the benefit of chemotherapy wasassociated with TS expression.Patients and Methods: The level of TS expression inthe primary rectal cancers of 294 of 801 patients en-rolled on protocol R-01 of the National Surgical AdjuvantBreast and Bowel Project (NSABP) was immunohisto-chemically assessed with the monoclonal antibodyTS 106.Results: Forty-nine percent of patients whose tumorshadl0-wTS levels (n = 91) were disease free at 5 yearscompared with 27% of patients with high levels of TS (n= 203; P < .01). Moreover, 60% of patients with low TSlevels were alive after 5 years compared with 40% ofpatients with high TS levels (P < .01). The level of TSprotein was significantly associated with Dukes' stage(P < .01); patients with a more advanced Dukes' stagehad a significantly higher level of TS. The level of TS ex-pression remained prognostic for both disease-free sur-HYMIDYLATE SYNTHASE (TS; EC.2.1.1.45) isa critical therapeutic target for the fluoropyrimidinecytotoxic drugs that are the mainstay of treatment forpatients with locally advanced and metastatic gastrointesti-nal cancers.1'2 Both fluorouracil (5-FU) and fluorodeoxyur-idine (FUdR) are converted in tumor cells to 5-fluoro-deoxyuridine monophosphate, which forms a tight-bindingcovalent complex with TS in the presence of the folatecofactor 5,10-methylenetetrahydrofolate.2-4 TS providesthe sole de novo source of thymidylate for DNA syn-thesis.The clinical importance of TS in determining fluoro-pyrimidine cytotoxicity has been established in both clini-cal and preclinical studies. Increased expression of TST vival (P < .01) and survival (P < .05) independent ofDukes' stage and other pathologic characteristics evalu-ated. Thirty-eight percent and 54% of patients with highTS levels (n = 71) were disease free and alive, respec-tively, after 5 years when treated with chemotherapy,compared with 17% and 31%, respectively, of similarpatients when treated with surgery alone (n = 64) (P <.01). No difference was noted in disease-free survival(P = .46) or survival (P = .43) in patients with low TSlevels.Conclusion: The expression of TS is an important in-dependent prognosticator of disease-free survival andsurvival in patients with rectal cancer. Adjuvant fluoro-uracil (5-FU)-based chemotherapy demonstrated sig-nificant improvement in disease-free and overall survivalfor patients with high TS levels. Prospective studies mea-suring TS levels will be needed to understand further therole of TS as a prognosticator of survival and chemother-apeutic benefit.J Clin Oncol 12:2640-2647. O 1994 by American So-ciety of Clinical Oncology.protein due to underlying gene amplification has beendescribed in cell lines selected for drug resistance byexposure to 5-FU.56 'Other studies have demonstrated thatthe level of TS protein correlates inversely with sensitiv-ity and response to 5-FU in human cancer cell lines.7-9Moreover, several investigators have demonstrated invivo as well as in vitro that TS enzyme levels in neoplasticcells increase acutely when cells are exposed to cytotoxicagents such as 5-FU.'0\" Thus, the ability of a tumorto overexpress TS enzyme may play a key role in thedevelopment of tumor resistance and may represent animportant protective mechanism in response to 5-FU. Inview of this critical role of TS, considerable work hasbeen directed toward quantitating TS enzyme levels intumor tissues in an effort to understand the potential pre-dictive role of this enzyme in patients.The quantitation of TS has traditionally been performedusing enzymatic biochemical assays; however, theseassays have major limitations when applied to humantumor tissue samples.'2\"3 We have recently developedmonoclonal antibodies to human TS that have the re-quired sensitivity and specificity to detect and quantitateTS enzyme in formalin-fixed tissue sections.'4Several randomized clinical trials recently demon-strated that treatment with adjuvant chemotherapy withFrom the National Cancer Institute-Navy Medical OncologyBranch, Division of Cancer Treatment, National Cancer Institute,MD, and the National Surgical Adjuvant Breast and BowelBethesda, Project, University of Pittsburgh School of Medicine, Pittsburgh,PA.Address reprint requests to Patrick G. Johnston, MD, NationalCancer Institute-Navy Medical Oncology Branch, Building 8, Room5101, Naval Hospital, Bethesda, MD 20889.© 1994 by American Society of Clinical Oncology.0732-183X/94/11212-0019$3.00/02640of Clinical Oncology, Vol 12, No 12 (December), 1994: pp 2640-2647Journal Copyright © 1994 American Society of Clinical Oncology. All rights reserved.

TS EXPRESSION AND PROGNOSIS IN RECTAL CANCERor without radiation therapy improved disease-free sur-vival and survival of patients with rectal cancer.15,16 Arecent National Institutes of Health consensus conferencerecommended adjuvant chemotherapy using a 5-FU-con-taining regimen along with radiation in patients with stage2 and 3 (Dukes' B and C) rectal cancer.\" In this study,our aim was to determine whether levels of TS expressionin the primary rectal cancer might be prognostic for dis-ease-free survival and overall survival or useful as a deter-minant of outcome to 5-FU-based therapy. To achievethis, we studied TS levels in rectal cancer from patientswho were part of a prospective randomized clinical trialof the National Surgical Adjuvant Breast and Bowel Proj-ect (NSABP) protocol R-01.PATIENTS AND METHODSA total of 801 eligible patients with rectal cancer were enteredonto NSABP protocol R-01 from 1977 to 1986. Patients with Dukes'B and C rectal cancer were randomized to receive either surgeryalone, surgery plus radiation, or surgery plus lomustine (meth-2641ylCCNU), 5-FU, and vincristine (MOF) chemotherapy. Dukes' A(n = 181) and Dukes' D (n = 80) patients were entered for follow-up only. The TS study samples consisted of tissue sections fromparaffin-embedded blocks available from 312 patients. These repre-sented all available tissue sections from the R-01 protocol. Tissueblocks were not sent for all patients, as this was not a requirement ofthe protocol. These samples had been stored in the NSABP pathologyheadquarters at Shadyside Hospital, Pittsburgh, PA. The clinical andpathologic data available for all patients were kept separately in acomputerized data bank in the Department of Statistics at the Univer-sity of Pittsburgh. The length of clinical follow-up at the time ofthis analysis was 131 months. Of the 312 patients, only 294 speci-mens were assessable for analysis. Eighteen were not assessable dueto the lack of available clinical data (n = 4), incorrect numericallabeling (n = 8), or uninterpretable immunohistochemical staining(n = 6). The clinical results of the NSABP R-01 study have beenpreviously published.\"6Immunohistochemical LabelingParaffin-embedded tissue sections 6-pM thick were deparaffinizedin xylene, rehydrated through graded alcohols, and washed in phos-phate-buffered saline (PBS). The TS 106 monoclonal antibody wasapplied using the avidin-biotin complex (ABC) immunohistochemi-cal technique.'\"Fig 1. Immunohistochemical staining of two rectal tumors using the TS 106 monoclonal antibody. Tissues were prepared as described inPatients and Methods. The tissues were counterstained in hematoxylin, mounted, and coverslipped with permount (Fisher Scientific, Fairlawn,NJ). The tissue on the right demonstrates high-intensity TS staining, whereas the tissue on the left demonstrates low-intensity TS staining.Copyright © 1994 American Society of Clinical Oncology. All rights reserved.

2642Table 1. Comparison of Dukes' Stage in Thymidylate Synthase StudyPopulation With Overall R-01 Population of Rectal Cancer PatientsThymidylate SynthaseSample Dukes' Stage No. % Total R-01 PopulationNo. % TotalJOHNSTON ET ALRESULTSPatient DemographicsTwo hundred ninety-four patient tumor samples wereassessable for TS staining. The TS study populationtended to have more patients who were eligible for ran-domization to either surgery or surgery plus chemother-apy (Dukes' B and C patients) and fewer patients whowere included in the observational arm of the study(Dukes' A and D patients). When compared with theoverall R-01 population, the distribution among Dukes'stage and the survival and disease-free survival patternswithin each Dukes' stage were similar. The patient char-acteristics and their treatments are listed in Table 1.Pathologic CorrelationsDukes' A Dukes' BSurgery Radiation MOF Total Dukes' CSurgery Radiation MOF Total Dukes' D Total 32 37173892 642755146 24 294 11 181 2331 204 2650 8 336 80 8014210NOTE. When compared with the remainder of the R-01 population, thesurvival and disease-free survival patterns of the thymidylate synthase studygroup were similar within each Dukes' stage.The TS staining in rectal tumors was predominantly agranular cytoplasmic staining pattern that was noted inthe tubular and cribriform glands of rectal cancer. TSstaining was also present in lymphocytes adjacent to thetumors and some minor crossreactivity was observed withsmooth-muscle cells. Heterogeneity of TS staining wasapparent both within individual tumors and between dif-ferent tumors. The TS staining intensity was not signifi-cantly associated with the histologic grade (P = .20),although an association of borderline significance wasnoted with nuclear grade (P = .06). Those rectal cancerswith poor nuclear grade had high TS levels.Clinical CorrelationsTS expression was inversely correlated with both dis-ease-free survival and survival (P < .01; Table 2). Forty-nine percent of patients whose tumors had low TS levelswere disease-free at 5 years compared with 27% of pa-tients who had high levels of TS expression (P < .01;Tissue EvaluationThe slides were examined and scored independently by two ob-servers (P.G.J.) and E.R.F.) blinded to both the clinical and patho-logic data. TS expression was quantitated using a visual gradingsystem based on the intensity of staining (0 to 3+) as well as itsextent, focal (< 25% of tumor staining positive) or diffuse (> 25%of tumor staining positive). Intensity levels 0 to 1 were groupedtogether and considered low-intensity (Fig lA), whereas 2 to 3 stain-ing intensity was considered high-intensity staining (Fig 1B). Therewas close agreement (> 85%) in the TS evaluation between bothinvestigators. In those cases of disagreement, final grading was deter-mined by consensus. Duplicate samples were included in the studyas internal controls for TS immunostaining and interpretation. TheTS staining intensity was similar from one immunohistochemicalexperiment to the next and the interpretation had 100% concordance.Statistical AnalysisDisease-free survival and overall 5-year survival was obtainedusing Kaplan-Meier estimates.\" Since the number of events waslarge, the survival curves were plotted from actuarial life tables20 where data were grouped into 3-month intervals.According toNSABP criteria, when disease-free survival was used an an endpoint, an event included recurrence of disease, death from cancer,death from noncancer causes, and the occurrence of a second nonco-lorectal primary. The log-rank test was used to test the differencebetween life-table distributions.2 We also analyzed the data ad-justing the log rank for Dukes' stage and treatment group. In thesubgroup of Dukes' B and C patients randomized to either surgeryalone or surgery plus MOF, we investigated the relative benefit ofchemotherapy to surgery alone in the low- and high-TS groups.Tests for interaction of TS classification and disease-free survivalor survival were conducted within the framework of a proportional22 hazards modelwhere significance levels for quantitative interactionwere based on the likelihood ratio test.Table 2. Correlation of Thymidylate Synthose LevelWith Disease-Free Survival and SurvivalThymidylateSynthose Intensity No. of Patients Disease-FreeSurvival (%) Survival 1%)Low intensity(0, 1) High intensity(2, 3) 91 203 49 (39, 60)' 27 (30, 33) P < .01 60 (50, 70)40 (33, 47)P < .01NOTE. The thymidylate synthase staining intensity was compared withthe percentage of patients disease-free or surviving 5 years as describedin Patients and Methods. The log-rank test was applied to test the differencebetween life-table distributions.* 95% confidence interval.Copyright © 1994 American Society of Clinical Oncology. All rights reserved.

TS EXPRESSION AND PROGNOSIS IN RECTAL CANCER2643gi_w0U.ccRc'U4P550 6 12 18 24 30 36 42MONTHS SINCE SURGERY48 54 600 6 12 18 24 30 36 42 MONTHS SINCE SURGERY48 54 60Fig 2. The correlation of the TS level with disease-free survival (A) and survival (B) in patients with rectal cancer. The TS staining intensitywas compared with the percentage of patients who are disease free or surviving 5 years as described in Patients and Methods. The log-ranktest adjusted for Dukes' stage and treatment group was used to test the difference between life-table distributions.Fig 2A). Sixty percent of patients with low levels werealive after 5 years compared with 40% of patients withhigh TS levels (P < .01; Fig 2B). Thus, for patients withlow TS levels, 81% more were disease-free and 50% morewere alive after 5 years when compared with patientswith high TS levels. The level of TS protein was alsosignificantly associated with Dukes' stage (P < .01). Pa-tients with more advanced Dukes' stage had significantlyhigher levels of TS. When the data were stratified forDukes' stage, nuclear grade, and treatment group, thelevel of TS expression remained prognostically significantfor both disease-free survival (P < .01) and survival (P< .05) independent of both Dukes' stage and nucleargrade. Comparisons of disease-free survival and survivalpatterns for low and high TS groups within each Dukes'stage are shown in Fig 3. Dukes' D patients were notincluded, because 23 of 24 Dukes' D patients had highTS levels. No difference in disease-free survival (P =.98) or survival of patients was noted (P = .85) betweena focal or diffuse TS tissue-staining pattern (data notshown).Correlation of TS Expression With Outcome toAdjuvant ChemotherapyThe disease-free survival and survival of 194 patientswith Dukes' B and C rectal cancer who were treated byeither surgery alone or surgery plus chemotherapy wereanalyzed to determine if TS was a predictor of outcometo MOF (methylCCNU, 5-FU, vincristine) chemotherapy(Table 3). In patients with high TS levels, only 17% weredisease-free after 5 years when treated with surgery alonecompared with 38% disease-free when treated with MOFchemotherapy (P < .01) (Fig 4A). A similar trend insurvival was apparent; 54% of patients were alive at 5years when treated with adjuvant MOF chemotherapycompared with 31% of those treated with surgery alone(P < .01) (Fig 4B; Table 3). Thus, twofold more patientswhose tumors contained high levels of TS were diseasefree and 74% more patients were alive at five years whentreated with MOF chemotherapy. In patients with low TSlevels, no significant difference in disease-free survival(36% v 43%) or survival (50% v 57%) was noted (Table3). Among the 44 patients who received radiation therapy,no difference in disease-free survival or survival wasnoted compared with patients who received surgery alone.Thus, MOF chemotherapy demonstrated a beneficial ef-fect on disease-free survival and survival only in patientswith high levels of TS.DISCUSSIONThymidylate synthase plays a key role in DNA nucleo-tide precursor synthesis and represents an important thera-peutic target for 5-FU, the most effective single agent inthe treatment of rectal cancer.'\"In this study we have examined the clinical importanceof TS enzyme expression as a prognostic marker of recur-rence and survival and as a predictor for chemotherapeuticbenefit in patients with rectal cancer. The results showthat the level of TS expression in rectal cancer correlateswith disease-free survival and survival. In patients whosetumors contained low TS levels, 81% more were disease-free and 50% more were alive at 5 years compared withpatients with high TS levels. The level of TS protein wasassociated with Dukes' stage and nuclear grade. However,even when patients were stratified for Dukes' stage andnuclear grade, the level of TS protein remained a signifi-cant independent prognosticator of disease-free survivaland survival. Thus, patients whose tumors have high lev-Copyright © 1994 American Society of Clinical Oncology. All rights reserved.

2644DUKES A RECTAL CANCER--- 24HIH 111JOHNSTON ET AL-----oN-•owso4020*N-14oN-Is0 12 IS 4 'S 4 U M4 * 12 Is N N I ER4 0 MOMTh ONCE SURoGYDUKES B RECTAL CANCER-I10-1001LOW TS5HIGH TSso-S5I0 •1I2 16e 24 21 36 .4 4 11 6 40_S 12 MONTH SINCE SURGERY42 0 is 24 24 3 MONT'M SINCE SURGERY01 1Fig 3. The correlation of TS intensity level withdisease-free survival and survival for Dukes'stage A, B, and C. The log-rank test showed asignificant prognostic effect of TS on both dis-ease-free survival and survival independent ofDukes' stage.DUKES C RECTAL CANCER100i-ISON-a TKINW LOW ISISIIS6 $ 12 01 20-ON-l\"2ONT41 a2 s 2 42 U Us4 0 a 12 s SURVIVALa 34 ONCE SUMMERDISEASE FREE SURVIVALSURVIVALels of TS have a poorer disease-free survival and overallsurvival.Dukes' staging represents the most useful prognosticfactor in rectal cancer, yet within each Dukes' stage con-siderable variation is present in the biologic behavior of-Tumor markers such as carcinoembry-these tumors.2325 Table 3. Relationship Between Thymidylate Synthase Level and theEffect of MOF Chemotherapy on Disease-Free Survival and SurvivalThymidylate Synthose Intensity Low intensitySurgery MOF High intensitySurgery MOF No. of Patients 37 22 Disease-FreeSurvival (%) 43 (27, 59)* 36 (16, 56) P= .46 17(8,26) 38 (27, 49) P < .01 Survival (%)57 (41, 73)50 (29, 71)P= .4331 (23, 46)54 (42, 65)P < .0164 71 NOTE. The test for interaction was significant, indicating a different rela-tive effect of chemotherapy in the two thymidylate synthase groups.* 95% confidence interval.onic antigen and CA 19-9 have been assessed for theirability to predict the clinical course of this disease buthave not been found to provide helpful information in'Other studies have examined theclinical practice.2627 prognostic significance of overexpression or mutations ofoncogenes such as c-myc, H-ras, and the tumor-suppres-sor gene p53 in colorectal cancer. These studies haveproven inconclusive.28-30 Recent studies have suggestedthat tumor DNA content and cell proliferation as deter-mined by flow cytometry may provide prognostic infor-mation. Several studies have suggested that patients withdiploid tumors generally have improved clinical outcomecompared with patients with aneuploid tumors.31-34 Witziget al32 examined DNA content and cell proliferation inpatients with stage B2 and C colorectal cancer (Astlerand Coller stage) and found that these parameters wereindependent prognostic factors for both colon and rectalcancer. In other studies, aneuploid tumors tended to havea worse clinical course, but there was no significant rela-tionship between ploidy and survival, independent ofDukes' stage.33:34 A previous study performed by FisherCopyright © 1994 American Society of Clinical Oncology. All rights reserved.

TS EXPRESSION AND PROGNOSIS IN RECTAL CANCER26454-JuJ4>_2646examining the role of TS as a predictor of chemotherapeu- tic efficacy in patients treated with 5-FU-based chemo- therapy will be necessary to fully define the role of TS as a determinant of outcome in patients treated with flu- oropyrimidine-based therapies. In conclusion, the level of TS expression predicts for JOHNSTON ET ALstudies measuring TS levels are necessary to better under-stand the role of TS as a prognosticator of survival andchemotherapeutic benefit in patients with rectal and otherepithelial cancers, such as gastrointestinal and breast can-cers, treated with fluoropyrimidine-based therapy.ACKNOWLEDGMENTThe authors thank Dr Timothy Kinsella for his insightful andhelpful discussions, Judy Jones for her help with the statistical analy-ses, and Kathy Moore for her editorial assistance in the preparationof the manuscript.disease-free survival and survival independent of Dukes'stage in patients with rectal cancer. This study also sug- gests that patients with high TS levels may derive the greatest benefit from adjuvant chemotherapy. Prospective REFERENCES1. Moertel CG: Current concepts in cancer: Chemotherapy ofgastrointestinal cancer. N Engl J Med 299:1049-1052, 19782. Danenberg PV: Thymidylate synthase: A target enzyme in can-cer chemotherapy. Biochim Biophys Acta 473:73-92, 19773. Santi DV, McHenry CS, Sommer M: Mechanisms of interac-tion of thymidylate synthase with 5-fluorodeoxyuridylate. Biochem-istry 13:471-480, 19744. Santi DV, Danenberg PV: Mechanisms of interaction of thy-midylate synthase with 5-fluorodeoxyuridylate folates in pyrimidinenucleotide biosynthesis, in Blakely RL, Benkovic SJ (eds): Folatesand Pterins. New York, NY, Wiley, 1984, pp 385-3985. Berger SH, Chung-Her J, Johnson LF, et al: Thymidylate syn-thase overproduction and gene amplification in fluorodeoxyuridine-resistant human cells. Mol Pharmacol 28:461-467, 19876. Clark JL, Berger SL, Mittelman A, et al: Thymidylate synthasegene amplification in a colon tumor resistant to fluoropyrimidinechemotherapy. Cancer Treat Rep 71:261-265, 19877. Spears CP, Antranik AH, Moran RG, et al: In vivo kinetics ofthymidylate synthase inhibition in 5-fluorouracil-sensitive and-resistant murine colon adenocarcinomas. Cancer Res 42:450-456,19828. Washtein WL: Increased levels of thymidylate synthase in cellsexposed to 5-fluorouracil. Mol Pharmacol 25:171-177, 19849. Johnston PG, Drake JC, Trepel J, et al: Immunological quanti-tation of thymidylate synthase using the monoclonal antibody TS106 in 5-fluorouracil-sensitive and -resistant human cancer cell lines.Cancer Res 52:4306-4312, 199210. Swain SM, Lippman ME, Egan EF, et al: Fluorouracil andhigh-dose leucovorin in previously treated patients with metastaticbreast disease. J Clin Oncol 7:890-896, 198911. Chu E, Koeller D, Johnston PG, et al: The regulation ofthymidylate synthase in human colon cancer cells treated with 5-fluorouracil and interferon-gamma. Mol Pharmacol 43:527-533,199312. Moran RG, Spears CP, Heidelberger C: Biochemical determi-nants of tumor sensitivity to 5-fluorouracil: Ultrasensitive methodsfor the determination of 5-fluoro-2-deoxyuridylate and thymidylatesynthase. Proc Natl Acad Sci USA 76:1456-1460, 197913. Roberts D: An isotope assay for thymidylate synthetase. Bio-chemistry 5:3546-3548, 196614. Johnston PG, Liang CM, Henry S, et al: The production andcharacterization of monoclonal antibodies that localize human thy-midylate synthase in the cytoplasm of human cells and tissues. Can-cer Res 51:6668-6676, 199115. Krook J, Moertel C, Gunderson LL, et al: Effective surgicaladjuvant chemotherapy for high-risk rectal carcinoma. N Engl J Med324:709-715, 199116. Fisher B, Wolmark N, Rockette H, et al: Postoperative adju-vant chemotherapy or radiation therapy for rectal cancer: resultsfrom NSABP protocol R-01. J Natl Cancer Inst 80:21-29, 198817. National Institutes of Health Consensus Conference: Adju-vant therapy for patients with colon and rectal cancer. JAMA264:1444-1450, 199018. Hsu SM, Raine L, Fanger H: A comparative study of theperoxidase-antiperoxidase method and an avidin biotin complexmethod for studying polypeptide hormones with radioimmunoassayantibodies. Am J Clin Pathol 75:734-738, 198119. Kaplan EL, Meier P: Non-parametric estimation from incom-plete observations. J Am Stat Assoc 53:457-458, 195820. Cutler SJ, Ederer F: Maximum utilization of the life-tablemethod in analyzing survival. J Chronic Dis 8:699-712, 195821. Peto R, Peto J: Asymptotically efficient rank invariant testprocedures. J R Stat Soc A 135:185-206, 197222. Cox DR: Regression models and life tables. J R Stat Soc34:187-220, 197223. Fisher ER, Sass R, Palekar A, et al: Dukes' classificationrevisited. Cancer 64:2354-2360, 198924. Dukes CE, Bussey HJR: The spread of rectal cancer and itseffect on prognosis. Br J Cancer 12:309-320, 195825. Astler VB, Coller FA: The prognostic significance of directextension of carcinoma of the colon and rectum. Ann Surg 139:846-852, 195426. Wanebo HJ, Rao B, Pinsky C: Preoperative carcinoembryonicantigen level as a prognostic indicator in colorectal cancer. N EnglJ Med 299:446-451, 197827. Scott N, Wieand H, Moertel C, et al: Colorectal cancer Dukes'stage tumor site, preoperative CEA plasma level and patient progno-sis related to tumor DNA ploidy pattern. Arch Surg 122:1375-1379,198728. Sun X-F, Cartensen JM, Zhang H, et al: Prognostic signifi-cance of cytoplasmic p53 oncoprotein in colorectal adenocarcinoma.Lancet 340:1369-1373, 199229. Sikora K, Chan S, Evan G, et al: C-myc oncogene expressionin colorectal cancer. Cancer 59:1289-1295, 198730. Gallick GE, Kurzock R, Kloetzer WS, et al: Expression ofp21 ras in fresh primary and metastatic human colorectal tumors.Proc Natl Acad Sci USA 82:1795-1799, 198531. Bauer KD, Lincoln ST, Vera-Roman JM, et al: Prognosticimplications of proliferative activity and DNA aneuploidy in colonicadenocarcinomas. Lab Invest 57:329-335, 1987Copyright © 1994 American Society of Clinical Oncology. All rights reserved.

TS EXPRESSION AND PROGNOSIS IN RECTAL CANCER32. Witzig TE, Loprinzi CL, Gonchoroff NJ, et al: DNA ploidyand cell kinetic measurements as predictors of recurrence and sur-vival in stages B2 and C colorectal cancer. Cancer 68:879-888, 199133. Chang KJ, Enker WE, Melamed M: Influence of tumor cellDNA ploidy on the natural history of rectal cancer. Am J Surg153:184-188, 198734. Fisher ER, Siderits RH, Sass R, et al: Value of assessmentof ploidy in rectal cancers. Arch Pathol Lab Med 113:525-528, 198935. Navalgund LG, Rossano C, Muench AJ, et al: Cell cycleregulation of thymidylate synthase gene expression in culturedmouse fibroblasts. J Biol Chem 255:7386-7390, 1980264736. Pestalozzi BC, McGinn CJ, Trepel J, et al: Immunologicalquantitation of thymidylate synthase and cell cycle in 5-fluorouracil-sensitive and -resistant human cancer cell lines. Proc Am AssocCancer Res 34:415, 1993 (abstr)37. Chu E, Voeller DM, Jones KL, et al: Identification of a thy-midylate synthase ribonucleoprotein complex in human colon cancercells. Mol Cell Biol 14:207-213, 199438. Lenz H-J, Horikoshi T, Leichman CG, et al: Thymidylatesynthase gene expression predicts response of gastric tumors to 5-FU leucovorin cisplatin. Proc Am Soc Clin Oncol 12:199, 1993(abstr)Copyright © 1994 American Society of Clinical Oncology. All rights reserved.