Lung Cancer (2004) 45, 381—386

Phase II trial of the histone deacetylase inhibitorpivaloyloxymethyl butyrate (Pivanex, AN-9) inadvanced non-small cell lung cancer

Tony Reida, Frank Valoneb, William Liperac, David Irwind,Warren Parolye, Ron Natale f, Sunil Sreedharanb, Harold Keerb,Bert Lumg, Frank Scappaticcib, Anish Bhatnagarb,*

a VA Medical Center and Stanford University, 3801 Miranda Drive, MC111 ONC,Palo Alto, CA 94304, USAb Titan Pharmaceuticals Inc., 400 Oyster Point Boulevard, Suite 505, South San Francisco,CA 94080, USAc North Shore Hematology Oncology Associates, 235 N. Belle Mead Road, E. Setauket, NY 11733, USAd Alta Bates Cancer Center, 2001 Dwight Way, Berkeley, CA 94704, USAe Sidney Kimmel Cancer Center, 10835 Altman Row, San Diego, CA 92121, USAf The Cedars Sinai Comprehensive Cancer Center, 8700 Beverly Boulevard, RM AC-1237, Los Angeles,CA 90048, USAg Stanford University, 300 Pasteur Drive, Stanford, CA 94305, USA

Received 15 October 2003 ; received in revised form 8 March 2004; accepted 15 March 2004

KEYWORDSPivanex;Histone deacetylaseinhibitor;Non-small cell lungcancer;AN-9

Summary This multicenter phase II trial evaluated the therapeutic activity and safetyprofile of pivaloyloxymethyl butyrate (Pivanex, AN-9) as a single agent in refractorynon-small cell lung cancer (NSCLC). Pivanex (2.34 g/m2 per day) was administered asa 6-h continuous intravenous infusion, daily for 3 days, and repeated every 21 daysuntil disease progression. Forty-seven patients were treated. More than 90% of pa-tients had received both a platinum compound and a taxane and 32% had receivedthree or more prior chemotherapy regimens. The most common toxicities were tran-sient grade 1—2 fatigue (34%), nausea (17%), and dysgeusia (11%). Three patients hadpartial responses (6.4 and 95%; CI 1.4—18.7%) and 14 patients had stable disease for≥12 weeks (30%). Median survival for all patients was 6.2 months with 1-year survivalof 26%. For patients who received fewer than three prior chemotherapy regimens,median survival was 7.8 months and 1-year survival was 31%. Pivanex is well toler-ated and appears to be active as a single agent in patients with advanced NSCLCrefractory to previous chemotherapy. Based on its therapeutic activity and favorablesafety profile, further studies of Pivanex in NSCLC, particularly in combination withcurrent chemotherapeutic agents, are warranted.© 2004 Elsevier Ireland Ltd. All rights reserved.

*Corresponding author. Tel.: +1-650-244-4990;fax: +1-650-244-4956.E-mail address: abhatnagar@titanpharm.com (A. Bhatnagar).

1. Introduction

Pivanex is a small molecule inhibitor of histonedeacetylases (HDACs) that has anti-proliferative,

0169-5002/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.doi:10.1016/j.lungcan.2004.03.002

382 T. Reid et al.

differentiating, and pro-apoptotic effects [1,2].Pivanex inhibits clonogenicity of fresh tumorcells from a wide range of human solid tu-mors including non-small cell lung carcinoma(NSCLC), colo-rectal cancer, breast cancer, andmelanoma [3]. It is also cytotoxic to fresh hu-man chronic lymphocytic leukemia (CLL) cells[4] and acute lymphocytic leukemia (ALL) cells[5]. Pivanex inhibits growth of Lewis lung, EBC-1,SK-Mes-1, Calu-6, PC9, and AS49 lung cancer celllines in vitro [6—8]. In pre-clinical animal mod-els, Pivanex decreases pulmonary metastasesafter intravenous injection of 3LLD122, Lewislung carcinoma cells [6]. Pivanex also increasessurvival of mice injected with Lewis lung car-cinoma [6] or monocytic leukemia MmA cells[8].

Inhibition of histone deacetylases by Pivanexresults in hyperacetylation of histones leadingto altered gene expression and cell differentia-tion or cell death [9—11]. Pivanex decreases ex-pression of the anti-apoptotic protein Bcl-2 andincreases expression of the pro-apoptotic pro-tein BAX in fresh human CLL cells and this mayaccount for Pivanex’s apoptotic effect on CLLcells [4]. Furthermore, Pivanex increases expres-sion of p53, which may contribute to tumor cellapoptosis [7]. Inhibition of cell proliferation mayalso result from Pivanex-induced increases in ac-tive (hypophosphorylated) retinoblastoma proteinand decreases in inactive (hyperphosphorylated)retinoblastoma protein [10]. Finally, Pivanex pro-motes differentiation of leukemia and solid tu-mor cell lines, perhaps through modulation ofexpression of oncogenes such as c-myc and c-jun[7,11].

Pivanex was developed as a pro-drug of the differ-entiating agent butyric acid [1]. Pre-clinical stud-ies, however, demonstrated that Pivanex is substan-tially more potent than butyric acid for alteringgene expression and inducing tumor cell differen-tiation or death [1,8,11]. More rapid and completecell uptake may account for Pivanex’s enhanced ef-ficacy [12].

In a phase I trial, doses of Pivanex up to 3.3 g/m2

per day were well tolerated when given on days1—5 of a 21-day treatment cycle [13]. One patientwith squamous cell lung cancer had an objectiveclinical response to Pivanex. Based on the exten-sive pre-clinical data that suggested Pivanex mightbe active in NSCLC, we undertook a phase II trialof Pivanex in patients with chemotherapy-resistantNSCLC. The goals of the trial were to assessanti-cancer activity of Pivanex as a single agent andto obtain additional safety information on potentialtoxicities.

2. Patients and methods

2.1. Patient eligibility

Patients with histologically or cytologically con-firmed NSCLC that was locally recurrent ormetastatic and had progressed after standardchemotherapy were eligible for the trial. Addi-tional eligibility criteria were: (1) age ≥18; (2)ECOG performance status ≤2; (3) no chemotherapy,radiotherapy, or investigational therapy within 4weeks prior to Pivanex treatment; (4) one or moreindex pulmonary lesions that were bidimensionallymeasurable by radiographs; (5) adequate hemato-logical function (WBC >2500 �l−1, absolute neu-trophil count >1500 �l−1, platelets ≥100,000 �l−1,and hemoglobin ≥9 g/dl); (6) adequate hepaticfunction (total bilirubin level ≤1.5 times the up-per normal limit, aspartate aminotransferase andalanine aminotransferase of ≤2.0 times the upperlimit of normal); (7) normal serum creatinine; (8)serum amylase <1.5 times the upper normal limit;(9) serum triglycerides <400mg/ml; (10) New YorkHeart Association Class I or II; and (11) no coexist-ing medical conditions that would interfere withtreatment. Written informed consent was obtainedfrom patients according to federal and institu-tional guidelines. The study was approved by theappropriate ethics committees at each institution.

2.2. Treatment

Pivanex (Titan Pharmaceuticals Inc., South SanFrancisco) was infused at a dose of 2.34 g/m2 over6 h by peripheral vein. Treatment was given ondays 1—3 and repeated every 21 days until diseaseprogression. For the infusions, Pivanex was mixedwith 20% intralipid at a final concentration of 27mgPivanex/ml.

2.3. Evaluation of response and toxicity

Bidimensionally measurable tumor masses wereevaluated at baseline and at the end of every twocycles. Complete response (CR) was defined as a100% disappearance of all tumor masses (measur-able or not) and partial response (PR) was definedas 50—99% decrease in the sum of products of thelongest perpendicular diameters (SPPD) of all mea-surable tumors. Stable disease (SD) was definedas <50% decrease or <25% increase in SPPD of allmeasurable tumors. Progressive disease (PD) wasdefined as ≥25% increase in SPPD or appearance ofnew lesions. Reported responses are best responsesseenwhile on study based on theWHO criteria. Toxi-

Phase II trial of Pivanex, AN-9 in advanced non-small cell lung cancer 383

cities were graded according to the National CancerInstitute Common Toxicity Criteria (version 1.0).

2.4. Statistical analysis

Statistical analyses were performed using SASv.8e (SAS Institute Inc., Cary, North Carolina).Survival probabilities were computed using theKaplan—Meier product limit estimates and plotted(survival plots) using the STAT/GRAPH procedures(PROC PLOT). Patient variables were of binomialnature (yes/no) and included histology (squa-mous and non-squamous), and number of priorchemotherapy regimens (<3 and ≥3). Analysis ofthe influence of patient variables on survival wasperformed using the Logrank test to compare sur-vival distributions. Time to disease progression andoverall survival endpoints were calculated as thetime from the first Pivanex dose to the specifiedevent. All enrolled patients were evaluated for thestudy’s endpoints.

3. Results

3.1. Patient characteristics

Patients were enrolled at 10 separate clinical cen-ters in this multicenter trial from 14 January 1999to 16 July 2001. The median age of patients was66 (range 43—82). The patients were a relativelypoor prognostic group with only seven patients(15%) having a performance status of 0 and eightpatients (17%) having a performance status of 2.Forty-three percent of patients had squamous celltumors and the rest had a variety of non-squamouscell tumor histologies. A majority of patients (72%)had metastatic disease (stage IV) at disease pre-sentation and the remainder presented with stageIIIa/IIIb disease that was refractory or relapsedat the time of protocol registration. The patientswere heavily pre-treated with 15 patients (32%)having received ≥3 different chemotherapy regi-mens. Forty-five patients (96%) received a platinumcompound (cisplatin and/or carboplatin) and 42patients (89%) had received a taxane (paclitaxelor docetaxel). Prior chemotherapy agents and thenumber of patients who received them were car-boplatin (40), cisplatin (12), paclitaxel (38), doc-etaxel (6), gemcitabine (18), navelbine (17), andother (23).

The dose of Pivanex was not reduced for toxicityin any patient. Eight patients experienced rapid dis-ease progression (n = 5) or an adverse event (n = 3)leading to treatment discontinuation before receiv-

Table 1 Frequent (≥5%) adverse events potentiallya

related to Pivanex (n = 47)

Event n (%)

Overall Grade 3 Grade 4

Fatigue 16 (34%) 1 (2%) 0Nausea 8 (17%) 0 0Dysgeusia 5 (11%) 0 0Asthenia 5 (11%) 1 (2%) 0Cough 5 (11%) 0 0Anorexia 4 (9%) 0 0Dyspnea 4 (9%) 0 2 (4%)Injection site pain 4 (9%) 0 0Injection site burning 3 (6%) 0 0Vision blurred 3 (6%) 0 0Headache 3 (6%) 0 0Chest pain 3 (6%) 1 (2%) 1 (2%)Pyrexia 3 (6%) 0 0

a Includes events considered unlikely to be related toPivanex.

ing two full courses of Pivanex therapy. The me-dian number of Pivanex courses was two (mean: 3.6;range: 0—25).

3.2. Adverse events

Pivanex was generally well tolerated. Hematologictoxicity was minimal and no patient experiencedgrade 3 or 4 neutropenia. One patient experi-enced grade 3 thrombocytopenia (45,000mm−3)and one patient experienced grade 4 anemia(hemoglobin = 5.4 g/dl). There were no episodes ofhepatotoxicity or nephrotoxicity. Table 1 presentsthe most frequent (≥5%) clinical adverse eventsthat were scored as Pivanex-related even if theinvestigator considered the relationship to be un-likely. Almost all treatment-related adverse eventswere mild/moderate in severity (grades 1—2),occurred during the Pivanex infusion, and wererapidly reversible. Two grade 3 adverse eventswere reported as possibly related to Pivanex: fa-tigue and hypokalemia. One episode of grade 4hypersensitivity was reported as possibly relatedto Pivanex. This episode occurred at the onset ofthe patient’s second dose of Pivanex. The patientwas treated with supportive care, with completeresolution (without hospitalization), and Pivanexwas discontinued.

3.3. Tumor response and patient survival

For the overall population, three patients experi-enced partial responses (6.4 and 95%; CI 1.4—18.8%)(two confirmed, one unconfirmed) and 14 patients

384 T. Reid et al.

Table 2 Efficacy of Pivanex in advanced NSCLC

All patients (n = 47) Patients who received <3prior therapies (n = 32)

Best tumor responsePartial response 3 (6.4%) 3 (9.4%)Stable disease (≥12 weeks) 14 (30%) 12 (40%)

SurvivalProgression-free survival, median 1.5 months (3—634 days) 2.5 months (6—634 days)Overall, median 6.2 months (6—772 days) 7.8 months (6 to >772 days)1-year survival 26% 31%

(30%) experienced stable disease for ≥12 weeks(Table 2). The three responding patients receivedfewer than three prior chemotherapy regimens. Dis-ease stabilization ranged from 12 to >38 weeks withone long-term responding patient completing 25 cy-cles of Pivanex before stopping therapy while stillin remission. The protocol did not collect quality oflife data prospectively.

For all patients median progression free sur-vival was 1.5 months and median overall survivalwas 6.2 months (Fig. 1). Twenty-six percent of pa-tients were alive at 1 year. Median progression-freesurvival was significantly greater for the 32 pa-tients who were treated with fewer than threeprior chemotherapy regiments (2.5 months) com-pared to patients who received three or more priorchemotherapy regimens (1.4 months) (P = 0.007).As can be seen in Fig. 1, patients who receivedfewer than three prior chemotherapy regimens

Fig. 1 Kaplan—Meier plot of overall survival according to the number of prior chemotherapy regimens.

also had significantly improved overall survivalwhen compared to patients who received three ormore prior regimens (median survival = 7.8 and2.5 months, respectively; P = 0.003).

4. Discussion

The results of this study show that Pivanex is welltolerated in patients with advanced non-small celllung cancer and suggest that Pivanex has therapeu-tic activity in this setting. Overall, most adverseevents related to Pivanex were mild to moderatein severity and transient in duration. The mostcommon adverse events were fatigue, nausea, anddysgeusia. Unlike conventional chemotherapeuticagents, Pivanex has minimal hematological, renal,and hepatic toxicity. Pivanex’s safety profile com-pares favorably with other HDAC inhibitors. In a

Phase II trial of Pivanex, AN-9 in advanced non-small cell lung cancer 385

phase I trial, dose-limiting toxicities for depsipep-tide were fatigue, nausea and vomiting, throm-bocytopenia, and cardiac arrhythmia [14]. Majortoxicities with suberoylanilide hydroxamic acid(SAHA) include fatigue, diarrhea, anorexia, de-hydration, and myelosuppression [15]. More than25% of patients receiving MS-275 experience grade1—2 fatigue, headache, nausea and vomiting, anddose-limiting toxicities include elevated AST, pleu-ral effusion, and epigastric pain [16]. Phenylbu-tyrate has a safety profile comparable to Pivanexwith the most frequent adverse events being re-versible fatigue and somnolence [17].

This study suggests that Pivanex is active fortreatment of advanced NSCLC. Three patients(6.4%) experienced partial remissions with oneresponse lasting >38 weeks. An additional 14 pa-tients (30%) experienced disease stabilization for12 or more weeks. Pivanex induces tumor celldifferentiation and this effect could result in pro-longed survival without tumor regression. In thisheavily pre-treated group of NSCLC patients, me-dian survival was 6.2 months with 1-year survivalof 26%. For patients who received Pivanex as sec-ond or third line therapy, median survival was7.8 months and 31% of patients were alive at 1year. Although patient selection may contribute toPivanex’s apparent favorable effect on survival,the trial was a multicenter, community-based studymaking such selection bias less likely. These resultsare comparable to those observed for advancedNSCLC patients treated with docetaxel [18 (95% CI3.1—13.1%), 19 (95% CI not reported)] or gefitinib[20 (95% CI 11.5—27.3%), 21 (95% CI not reported)].For patients who had received one or two priorchemotherapy regimens, docetaxel achieves re-sponse rates of 5.5 and 6.7%. Median survival withdocetaxel treatment is approximately 6 monthswith 1-year survival of approximately 35%. Patientstreated with gefitinib experience higher responserates of 18.4—19%. However, median survival is 7.6and 8.0 months with 1-year survival of 29 and 35%.Similarly, median survival and 1-year survival ratesfor patients who received more than two priorchemotherapy regimens are similar after treatmentwith Pivanex, docetaxel, and gefitinib [18—21].However, the current study is not a randomizedcontrolled study and the primary endpoint was notsurvival. Pivanex-treatment causes fewer adverseevents than docetaxel and gefitinib, suggestingthat Pivanex has a favorable risk-benefit profile.Additional studies that formally assess quality oflife after Pivanex treatment are planned to confirmthis suggestion.

Pivanex’s favorable toxicity profile suggests thatfull doses of this agent could be combined safely

with standard cytotoxic agents. Pre-clinical stud-ies of Pivanex combined with standard cytotoxicagents show synergistic interactions for combi-nations of Pivanex with docetaxel, gemcitabine,doxorubicin, and daunorubicin [6—8,22], as well ascisplatin, carboplatin, irinotecan, and topotecan(unpublished observations). Synergy appears to beschedule-dependent as maximal synergy with do-cetaxel requires pre-treatment with Pivanex [22]whereas pre-treatment is not required for synergywith doxorubicin [6]. These observations suggestthat more than one molecular mechanism under-lies the synergistic interactions of Pivanex andcytotoxic agents.

5. Conclusion

The data from this phase II trial show that Pivanexis well tolerated in patients with advanced NSCLCand is indicative of anti-cancer activity. Moreover,the response rate and patient survival characteris-tics are similar to other agents recently approvedin this disease setting. Given Pivanex’s excellentsafety profile and pre-clinical data showing synergywith cytotoxic agents used for treatment of NSCLC,further clinical development of Pivanex will focuson combination chemotherapy regimens. A phase Itrial of the combination of Pivanex and docetaxelindicate that the combination is well tolerated [23].We recently opened a randomized phase IIb trialcomparing docetaxel monotherapy with the combi-nation of Pivanex and docetaxel as second line ther-apy in patients with NSCLC.

Acknowledgements

We would like to thank David Magnuson for criticalreview of this manuscript and Kim Lang for adminis-trative assistance. This study was supported by Ti-tan Pharmaceuticals Inc.

References

[1] Rephaeli A, Rabizadeh E, Aviram A, Shaklai M, Ruse M,Nudelman A. Derivatives of butyric acid as potentialanti-neoplastic agents. Int J Cancer 1991;49:66—72.

[2] Nudelman A, Ruse M, Aviram A, Rabizadeh E, Shaklai M,Zimrah Y, et al. Novel anticancer prodrugs of butyric acid2. J Med Chem 1992;35:687—94.

[3] Siu LL, Von Hoff DD, Rephaeli A, Izbicka E, Cerna C, GomezL, et al. Activity of pivaloyloxymethyl butyrate, a novelanticancer agent, on primary human tumor colony-formingunits. Invest New Drugs 1998;16:113—9.

[4] Rabizadeh E, Bairey O, Aviram A, Ben Dror I, Shaklai M,Zimra Y. Doxorubicin and a butyric acid derivative ef-

386 T. Reid et al.

fectively reduce levels of Bcl-2 protein in the cells ofchronic lymphocytic leukemia patient. Eur J Haematol2001;66:263—71.

[5] Batova A, Shao LE, Diccianni MB, Yu AL, Tanaka T, RephaeliA, et al. The histone deacetylase inhibitor AN-9 has se-lective toxicity to acute leukemia and drug-resistant pri-mary leukemia and cancer cell lines. Blood 2002;100:3319—24.

[6] Niitsu N, Kasukabe T, Yokoyama A, Okabe-Kado J,Yamamoto-Yamaguchi Y, Umeda M, et al. Anticancerderivative of butyric acid (pivalyloxymethyl butyrate)specifically potentiates the cytotoxicity of doxorubicin anddaunorubicin through the suppression of microsomal gly-cosidic activity. Mol Pharmacol 2000;58:27—36.

[7] Reid T, Lee Y, Eng L, Gadol N, Sreedharan S, Keer H, etal. Pivanex represses oncogene expression and sensitizestumor cells to chemotherapeutic agents. Proc Am AssocCancer Res 2002;43:944a.

[8] Kasukabe T, Rephaeli A, Honma Y. An anti-cancer derivativeof butyric acid (pivalyloxmethyl butyrate) and daunoru-bicin cooperatively prolong survival of mice inoculatedwith monocytic leukaemia cells. Br J Cancer 1997;75:850—4.

[9] Aviram A, Zimrah Y, Shaklai M, Nudelman A, RephaeliA. Comparison between the effect of butyric acid andits prodrug pivaloyloxymethylbutyrate on histones hyper-acetylation in an HL-60 leukemic cell line. Int J Cancer1994;56:906—9.

[10] Sreedharan SP, Wirtz UF, Carrizale G. Pivanex, a short-chainfatty acid histone deacetylase inhibitor, induces changes inkey regulatory protein expression in human non-small celllung carcinoma cells [abstract]. AACR-NCI-EORTC Interna-tional Conference–—Molecular Targets and Cancer Thera-peutics; 2003.

[11] Rabizadeh E, Shaklai M, Nudelman A, Rephaeli A. Rapidalteration of c-myc and c-jun expression in leukemic cellsinduced to differentiate by a butyric acid prodrug. FEBSLett 1993;328:225—9.

[12] Zimra Y, Nudelman A, Zhuk R, Rabizadeh E, Shaklai M, Avi-ram A, et al. Uptake of pivaloyloxymethyl butyrate intoleukemic cells and its intracellular esterase-catalyzed hy-drolysis. J Cancer Res Clin Oncol 2000;126:693—8.

[13] Patnaik A, Rowinsky EK, Villalona MA, Hammond LA, BrittenCD, Siu LL, et al. A phase I study of pivaloyloxymethylbutyrate, a prodrug of the differentiating agent butyricacid, in patients with advanced solid malignancies. ClinCancer Res 2002;8:2142—8.

[14] Sandor V, Bakke S, Robey RW, Kang MH, Blagosklonny MV,Bender J, et al. Phase I trial of the histone deacetylaseinhibitor, depsipeptide (FR901228, NSC 630176), in patients

with refractory neoplasms. Clin Cancer Res 2002;8:718—28.

[15] Heany M, O’Connor OA, Richon V. Clinical experience withthe histone deacetylase (HDAC) inhibitor suberoylanilidehydroxamic acid (SAHA) in heavily pretreated patientswith hematological malignancies. Proc Am Soc Clin Oncol2003;22:200A (abstract 2321).

[16] Ryan QC, Headlee D, Sparreboom A, Figg W, Zhai J, TrepelJ, et al. A phase I trial of an oral histone deacetylaseinhibitor, MS-275, in advanced solid tumor and lymphomapatients. Proc Am Soc Clin Oncol 2003;22:200A (abstract802).

[17] Gore SD, Weng LJ, Zhai S, Figg WD, Donehower RC, DoverGJ, et al. Impact of the putative differentiating agentsodium phenylbutyrate on myelodysplastic syndromes andacute myeloid leukemia. Clin Cancer Res 2001;7(8):2330—9.

[18] Fossella FV, DeVore R, Kerr RN, Crawford J, Natale RR,Dunphy F, et al. Randomized phase III trial of doc-etaxel versus vinorelbine or ifosfamide in patients withadvanced non-small-cell lung cancer previously treatedwith platinum-containing chemotherapy regimens. The TAX320 non-small cell lung cancer study group. J Clin Oncol2000;18(12):2354—62.

[19] Shepherd FA, Dancey J, Ramlau R, Mattson K,Gralla R, O’Rourke M, et al. Prospective random-ized trial of docetaxel versus best supportive carein patients with non-small-cell lung cancer previouslytreated with platinum-based chemotherapy. J Clin Oncol2000;18(10):2095—103.

[20] Fukuoka M, Yano S, Giaccone G, Tamura T, NakagawaK, Douillard JY, et al. Multi-institutional randomizedphase II trial of gefitinib for previously treated patientswith advanced non-small-cell lung cancer. J Clin Oncol2003;21(12):2237—46.

[21] Cappuzzo F, Gregorc V, Rossi E. Gefitinib in pretreatednon-small-cell lung cancer (NSCLC): analysis of efficacy andcorrelation with HER2 and epidermal growth factor recep-tor expression in locally advanced or metastatic NSCLC. JClin Oncol 2003;21(14):2658—63.

[22] Sreedharan S, Revilla M, Estlack L, Wirtz U, Valone F, Pi-azza G. Pivanex, a histone deacetylase inhibitor, is syn-ergistic with chemotherapy in inhibiting growth of humannon-small cell lung cancer lines. Proc Am Assoc Cancer Res2003;44:853.

[23] Reid T, Cosgriff TM, Yen K, Scislowski P, Valone F, BhatnagarA. Dose escalation study of Pivanex (a histone deacety-lase inhibitor) in combination with docetaxel for advancednon-small cell lung cancer. Lung Cancer 2003;41(2):S182(abstract P-358).