Gefitinib and chemotherapy combination studies in five novel human non small cell lung cancer xenografts. Evidence linking EGFR signaling to gefitinib antitumor response



The epidermal growth factor receptor (EGFR) signaling pathway is often activated in NSCLC, and thus represents a promising therapeutic target. We studied the antitumor activity of gefitinib (Iressa™), an orally active EGFR-tyrosine kinase inhibitor, alone and in combination with standard chemotherapy in 5 recently established human NSCLC xenografts with wild-type EGFR. Mice were treated with 2 protocols of chemotherapy based on cisplatin (CDDP) combined with either gemcitabine (GEM) or vinorelbine (VNR). Gefitinib alone significantly inhibited tumor growth (TGI) in 4 of the 5 tumor xenografts (mean TGI of 58%, range: 25–70%). CDDP+VNR alone failed to achieve any significant responses, while CDDP+GEM achieved significant responses in 2 xenografts (TGI of 93 and 47%). Addition of gefitinib to CDDP+GEM potentialized chemotherapy in the 3 CDDP+GEM-resistant xenografts, but did not potentialize the CDDP+VNR combination. The effect of gefitinib treatment on the activity of extra cellular-regulated kinase (Erk), Akt, JNK and p38 kinases was assessed in IC9LC11 and IC1LC131, two NSCLC xenografts selected for their sensitivity and resistance to gefitinib, respectively. In IC9LC11, gefitinib strongly inhibited Erk, Akt and Jnk phosphorylation, but P38 remained active. Inversely, in IC1LC131, Erk and Akt pathways remained active, while Jnk and P38 pathways were inhibited by gefitinib. The data indicate that the antitumor activity of gefitinib in NSCLC, alone or in combination with chemotherapy, is tumor-dependent and is influenced by downstream signaling events independent of EGFR status. © 2006 Wiley-Liss, Inc.

In 2000, it was estimated that over 1 million lung cancer deaths occurred worldwide, making it the leading cause of cancer deaths.1 Nonsmall cell lung cancer (NSCLC) is the largest histological subset of all lung cancers, accounting for approximately 80% of cases. Patients with NSCLC have a poor prognosis, with nearly half of the patients presenting with advanced or metastatic disease at diagnosis. Survival remains unimpressive, with a median 5-year survival for NSCLC (all stages included) of approximately 3–23%.2

Chemotherapy is the standard first-line treatment option for advanced/metastatic NSCLC and usually comprises a platinum agent (cisplatin, carboplatin) in combination with gemcitabine or vinorelbine, or taxanes (paclitaxel or docetaxel).3 Although chemotherapy is associated with a modest increase in survival compared to best supportive care,4 4 of the most commonly used chemotherapy regimens demonstrate similar results in terms of response rate and survival, indicating that the efficacy of existing first-line treatment has reached a plateau.5 Relapse of patients with NSCLC who have received first-line chemotherapy is commonly observed.

Thus, in NSCLC, there is an urgent need for rationally designed, targeted agents displaying improved efficacy and tolerability compared to existing treatments. The epidermal growth factor receptor (EGFR) has been a leading focus for the clinical development of such new agents. This tyrosine kinase receptor is involved in key tumorigenic processes, such as proliferation, invasion, survival and angiogenesis,6 and is often highly expressed in NSCLC. Several EGFR inhibitors are now in clinical development or approved in NSCLC, including small molecule inhibitors targeting the kinase domain of the receptor, such as gefitinib or erlotinib, or the monoclonal antibody cetuximab, reviewed in.7 (Iressa™) is an orally active EGFR-tyrosine kinase inhibitor (EGFR-TKI) that inhibits phosphorylation of EGFR and thereby blocks downstream signal transduction pathways. In 2 large Phase II trials, gefitinib monotherapy (250 mg/day) in NSCLC patients previously treated by chemotherapy was demonstrated to be active with response rates of 18.4% and 11.8% and was generally well tolerated.8, 9

The effectiveness of combining gefitinib with standard first-line doublet platinum-based chemotherapy has been explored. Previous results from preclinical studies showed additive or synergistic growth inhibitory activity when gefitinib was combined with cisplatin, carboplatin, paclitaxel or docetaxel.10 However, 2 Phase III trials comparing first-line chemotherapy with and without gefitinib in patients with advanced NSCLC found similar survival outcomes for both treatment arms.11, 12

In the present study, the antitumor effect of gefitinib alone or in combination with standard gemcitabine- and vinorelbine-containing platinum-based chemotherapies was investigated in 5 new human NSCLC xenograft models obtained by direct transplantation of human tumor samples into nude mice. In addition, to search for tumor markers correlating with response to gefitinib or chemotherapy, we analyzed the mitogen-activated protein kinase (MAPK) pathways downstream of EGFR in a gefitinib-sensitive and a gefitinib-resistant tumor xenograft.

Material and methods

Human NSCLC xenografts

A selection of 5 NSCLC tumor xenografts was obtained from patients at the Curie Institute and Marie-Lannelongue Hospital after informed written consent. None of these patients had received prior chemotherapy except for IC9LC11, who received Vinorelbine. IC9LC11 and IC1LC131 (primary tumors), and IC7, IC8LC10 and IC14LC16 (skin, brain and adrenal metastases, respectively) were established as xenografts by subcutaneous implantation of fresh tumor fragments (surgery waste) into the scapular area of nude Swiss mice. The tumors were maintained by subcutaneous transplantation (2–3 passages) and cryopreserved in liquid nitrogen. Histology was carried out with conventional (hematoxilin/eosin) staining. IC1LC131, IC8LC10, IC9LC11 and IC14LC16 were adenocarcinomas, and IC7 was a squamous cell carcinoma. Histological examination showed that these features were maintained in corresponding xenografts.

In vivo studies

Gefitinib (‘Iressa’; provided by AstraZeneca, UK) was suspended in 0.05% Tween 80 (Sigma Chemical, St. Quentin Fallavier, France) in sterile water at a concentration of 18 mg/ml (120 mg/kg dose) or 6 mg/ml (40 mg/kg dose), and administered orally (0.2 ml) once daily for 2 or 8 weeks. Gemcitabine (GEM; provided by Lilly, France) was diluted to a concentration of 40 mg/ml (60 mg/kg dose) and administered once weekly. Cisplatin (CDDP; obtained from Aventis, France) was resuspended at 0.15 mg/ml (1 mg/kg dose) and administered at 21-day intervals. Vinorelbine (VNR; provided by Pierre Fabre Oncologie, France) was diluted to 0.19 mg/ml (1.25 mg/kg dose) and administered at 10-day intervals. Working solutions of GEM, CDDP and VNR were prepared in 0.9% NaCl (pH 7) and administered by intraperitoneal injection in a total volume of 0.2 ml. Two different chemotherapy regimens were used alone or combined with gefitinib: CDDP+GEM and CDDP+VNR were administered during all the experiment.


The anti-EGFR (SC101), anti-Akt (H135) and anti-actin (C2/SC-8432) antibodies were purchased from Santa Cruz Biotechnology Inc. (Tebu-Bio, Le Perray en Yvelines, France). Antibodies to anti-P-EGFR (Tyr1068), anti-Erk1/2 and anti-P-Erk1/2 (Thr202/Tyr204), anti-Jnk1/SAPK and anti-P-SAPK/Jnk (Thr183/Tyr185), anti-P38 and anti-P-P38 (Thr180/Thr182) were from Cell Signaling Technology (Ozyme, Saint Quentin Yvelines, France).

Experimental therapeutics

Nude mice (Swiss nu/nu; 8–10 weeks old) bearing subcutaneous NSCLC (100–250 mm3) were randomly assigned to either control or treatment groups (n = 8–12 mice/group); in all experiments, the yield of tumor take in mice was at least 80% in all treatment arms, which allows to treat all treatment arms simultaneously. The volume of each tumor was measured usually twice weekly by caliper measurement of two perpendicular diameters. Tumor volume (V) was calculated using the following previously reported formula13: V = A × B2, where A is the width of the tumor in millimetres and B the length. The individual relative tumor volume (RTV) was calculated using the following formula: RTV = Vx/V1 where Vx is the volume in mm3 at a given time and V1 at the start of treatment. Growth curves were obtained by plotting mean RTV against time. The percentage tumor growth inhibition (TGI) was calculated as the mean RTV of treated mice/mean RTV of control mice × 100. Tumor growth delay (TGD) was calculated as the time necessary for a 5-fold increase in tumor volume. Animals were ethically sacrificed when the tumor volume exceeded 2,000 mm3. For the global analysis of responses, we estimated the number of partial responses, complete tumor regressions (PR and CR), and cures; PR was defined as TGI up to 40 % and CR as tumors that had regressed and could no longer be palpated. Mice were considered to be cured if tumor-free at the end of experiment. Individual TGI was calculated as individual RTV of treated mouse divided by mean control RTV on the day of optimal effect, and data were pooled in treatment groups regardless of the NSCLC xenografts.

Analyses of mRNA and proteins in NSCLC xenografts

Tumors were grafted into nude mice and the mice were treated with CDDP+GEM or CDDP+VNR alone or in combination with gefitinib. Subcutaneous tumors were excised from mice 3 mice per group) 24 hr after the start of treatment, and were dissected to remove any necrotic tissue before freezing in liquid nitrogen. Approximately 20 and 40 mg of frozen tumor samples were pulverized under liquid nitrogen with a mortar and pestle to extract total RNA and protein, respectively.

Western immunoblot analyses

To examine protein phosphorylation under basal conditions or after treatment, pulverized tumor-tissue was mixed with ice-cold lysis buffer (50 mM Hepes pH 7.9, 1 mM EDTA, 120 mM NaCl, 1% NP-40, 5 mM NaF, 1 mM Na3V04, 1 mM AEBSF and 1 tablet Complete™ in 50 ml lysis buffer). The protein concentration was determined using BCA Protein Assay Reagent (Pierce Chemical, Rockford, IL). Tumor lysates (20 or 30 μg) were then separated on 12% SDS-PAGE polyacrylamide gels and transferred into nitrocellulose membrane filters. The blots were incubated with blocking solution and probed with various antibodies and washed well before incubation with HRP-conjugated secondary antibody. Proteins were visualized by enhanced chemiluminescence (ECL, Amersham).

cDNA synthesis

RNA was isolated using the RNA plus® kit reagent (Qbiogene, Illkirch, France) according to the manufacturer's instructions, and purified RNA was stored in RNAse-free distilled water at −80°C. RNA quality was confirmed by agarose gel electrophoreses and ethidium bromide staining, and RNA was quantified by spectrophotometry at 260 nm. RNA (1 μg) was reverse-transcribed in a final volume of 20 μL containing 1× reverse transcriptase buffer (Amersham, France; 1.25 mM each dNTP, 6.7 mM MgCl2, 2.5 Units of RNAse inhibitor), 5 μM/L random Hexamer (Bæhringer–Mannheim, Germany) and 10 Units of Moloney Murine Leukemia Virus Reverse Transcriptase (Life Technologies, Gaithersburg, MD). The reaction mix was incubated at 42°C for 30 min, before inactivation. cDNA was stored at −80°C.

Real-time PCR amplification

EGFR, HER-2, MDR1, MRP1, MRP5 and LRP transcripts were quantified using real-time quantitative reverse transcription (RT)-PCR assays. Primers and probes were selected using Primer Express Software (Applied Biosystems, Foster City, CA) and the nucleotide sequences were checked against GenBank, EMBL and DDBJ databases to confirm gene specificity. To avoid amplification of contaminating genomic DNA, 1 of the 2 primers was placed in a separate exon. Nucleotide and probe sequences and PCR conditions for EGFR, HER-2 and MDR1 were as previously described.14 Calibration curves (1:25–800) were created for the remaining target gene cDNAs (MRP1, MRP5 and LRP). The sequences of the primers and Taqman probes used for quantification of MRP1, MRP5 and LRP were the following: MRP1, forward primer: 5′ TAC-TCA-TTG-CAG-GTC-ACC-ACG-TAC-TT 3′; reverse primer: 5′ GAA-TTC-CAC-TCG-GCC-CAC-C 3′; and probe: AAG-GAG-GCG-CCC-TGG-CAA-AT; MRP5, forward primer: 5′ GCC-CTC-ATC-ACC-ACC-ACG 3′; reverse primer: 5′ CCC-CGT-TAA-CTG-GAC-AGC-ATA-AG 3′; and probe: CAC-GGG-CAG-ATT-CCC-CAG; LRP, forward primer: 5′ GGA-GAT-CAT-TCA-GGC-CAC-CAT 3′; reverse primer: 5′ AGA-ACC-TCC-TCA-AAC-ACC-GCT-G 3′; and probe: CTG-GGA-CCG-GGA-CGG-CAA.

PCR reactions were performed using an ABI Prism 7700 sequence detection System (Applied Biosystems), and real-time detection was performed using qPCR Core Reagent Kit (Eurogentec, Belgium). Fluorescent probes were synthesized by Applied Biosystems and primers were synthesized by Invitrogen (Paisley, UK). Thermal cycling conditions comprised an initial denaturation step at 95°C for 10 min, then 40 cycles at 95°C for 15 sec and 1 min at a target-dependent annealing temperature. HPRT (Applied Biosystems) and β-2 microglobulin15 were also analyzed as reference genes. All measurements were performed in duplicate. Results were expressed as N-fold differences in target gene expression relative to reference gene (both HPRT and β-2 microglobulin) and the calibrator was defined as ‘N target’ and determined using the formula: N target = Etarget(Ct calibrator−Ct sample)/Ereference gene(Ct calibrator−Ct sample), where E is the efficiency of PCR measured using the slope of the calibration curve.14

Two-sided Student's t test for equal variances was used to assess statistical significance for tumor growth inhibition and tumor growth delay data controlling for repeated measure. The χ2 test was used to compare the different treatment groups for the % of partial and complete responses.

Analysis of K-ras mutations and EGFR amplification status

The level of EGFR amplification was determined by quantitative genomic PCR essentially as described previously.16 Briefly, tumor or normal genomic DNA was amplified with 3 independent sets of primers across exons 1, 7 and 20 of the EGFR gene. Sequences of the primers are available upon request. PCR reactions were run on an Applied Biosystems 7500 RT-PCR System using SYBR Green detection. EGFR DNA amounts were normalized against GAPDH and EGFR amplification levels in tumor samples were calculated after calibration against normal human lymphocytes, using the 2-δCt method.

The presence of K-ras codon 12 or 13 mutations was analyzed by direct sequencing of PCR products from tumor genomic DNA. A nested PCR was used as previously described17 and PCR products were directly sequenced using a Big Dye Terminator sequencing kit (BD Biosciences).


Tolerance studies

The maximal tolerated dose (MTD) was defined as the maximum dose associated with ≤10% weight loss and no toxic deaths. The MTD was 150 mg/kg/day (5 days/week for 2 weeks) for gefitinib, and 3, 2.5 and 80 mg/kg for CDDP, VNR and GEM, respectively. It is noticeable that the MTD of gefitinib is about 20-fold higher in mice compared to humans. When gefitinib was co-administered with a combination of two cytotoxic agents, the following reduced doses of each compound were used: 120, 1, 1.25 and 60 mg/kg of gefitinib, CDDP, VNR and GEM, respectively. With this schedule, gefitinib did not affect the tolerability of the GEM, VNR or platinum-based chemotherapy, even when administered for 8 successive weeks.

Sensitivity of the different NSCLC tumor xenografts to gefitinib

The dose-effect of gefitinib (120 mg/kg and 40 mg/kg) on tumor growth was evaluated after a 2-week treatment period and 8-week in case of IC14LC16 xenograft (Table I, Figs. 1a and b). Dose-dependent tumor growth inhibition was observed: at 40 mg/kg no significant tumor response was observed, except in IC14LC16 tumor treated for 8 weeks, with a TGI of 48% and a prolonged growth delay of 13 days compared to controls. After a two-week treatment at 120 mg/kg, 4 out of the 5 tumors responded significantly, with a mean TGI of 58% and a mean TGD of 25 days. In case of IC1LC131, 2 independent experiments confirmed the absence of response to gefitinib. Response to gefitinib was not associated with either EGFR or MAPK kinase constitutive expression levels, or with their relative degree of phosphorylation (Fig. 4).

Figure 1.

Time and dose effect on the growth of IC14LC16 after gefitinib treatment. IC14LC16 tumor xenografts were treated with gefitinib at 40 and 120 mg/kg/day for 8 weeks [(a) and (b)] and treated at 120 mg/kg/day for 2 weeks [(c) and (d)]. Figures (b) and (d) show the survival time of mice treated after gefitinib: the cumulative percentage of individual tumors (mice) reaching their respective TGD is plotted as a function of time. Animals were sacrificed when the tumor volume reached 2000 mm3. Error bars represent calculated standard deviation.

Table I. Dose-Dependency of Gefitinib for a 2-Week Treatment in NSCLC Xenografts
NSCLC xenograftsDrug dosage per injection
40 mg/kg120 mg/kg
TGI (%)1GD (days)2TGI (%)1GD (days)2
  •  n.s.: Not significantly different from control groups.

  • 1

    Tumor Growth Inhibition: calculated as the ratio of treated/control relative tumor volume on the day of the ethical sacrifice of the first control mouse tumor volume = 2,000 mm3.

  • 2

    Growth Delay: difference between tumor growth delays (TGD) in treated vs. control tumors.

  • 3

    Mean tumor growth inhibition from 2 independent experiments.

  • *

    Significantly different from control groups, two-sided Student's t test p < 0.05.

IC78 (n.s)2870*28
IC8LC1028 (n.s)3053*30
IC1LC1310025 (13,38)38 (6,10)3
IC9LC11  60 (55,62)325 (25,25)3
IC14LC16  49 (P=0.015)15
IC14LC16 (8 weeks)48 (p = 0.03)1360 (p = 0.001)25

To evaluate potential time-dependence, IC14LC16 xenografts were treated with gefitinib at 120 mg/kg for 2 or 8 weeks, Table I and Figures 1c and 1a, respectively. Increased growth inhibition and growth delay was obtained when the treatment was prolonged for an 8 weeks period. The TGI was increased from 49% to 60% inhibition and the TGD was also prolonged from 15 to 25 days (Table I), and shown respectively in Figures 1a, 1c and Figures 1b, 1d.

Antitumor activity of gefitinib and the CDDP + VNR combination

None of the tumors responded significantly to the CDDP+VNR combined treatment and no complete response was observed. The greatest response, although not significant, was observed in IC8LC10, with a TGI of 39% and a tumor growth delay prolongation of 17 days. The addition of gefitinib (120 mg/kg) to CDDP+VNR chemotherapy was not associated with increased antitumor activity (Fig. 2). Data analysis indicates no significant benefit of this combination in the 5 tumors studied (Table II).

Figure 2.

Effect of gefitinib and CDDP plus VNR chemotherapy alone or in combination on the growth of the NSCLC tumor xenografts panel. Gefitinib, was given daily per os, at a dose of 120 mg/kg/day for 2 successive weeks; CDDP, was given I.P. at 1 mg/kg/day, every 21 days; and VNR was given I.P. at 1.25 mg/kg/day, every 10 days. Treatment started when subcutaneous growing tumor volumes were of 100–250 mm3; the tumor growth was measured by measuring 2 perpendicular diameters, using a caliper; tumor volume and the relative tumor volume (RTV) was calculated as described in Material and Methods section; growth curves were obtained by plotting mean RTV against time. Error bars represent calculated standard deviation.

Table II. Responses of the 5 NSCLC Xenografts to Gefitinib Alone or Combined with CDDP + VNR
NSCLC xenograftsTreatmentMean ± (SD) TGD (days)1GD (days)2TGI (%)3Nb mice /groupNb PR4Nb CR5Nb of cures6
  • Significantly different from control groups: *0.001 < p < 0.05; **10−6.

  • 1

    Tumor Growth Delay: delay in days required to reach a 5-fold increase of individual tumor volume from their size at the start of treatment (100–250 mm3).

  • 2

    Growth Delay: difference between tumor growth delays (TGD) in treated vs. control tumor.

  • 3

    Tumor Growth Inhibition: calculated as the ratio of treated/control relative tumor volume on the day of the ethical sacrifice of the first control mouse tumor volume = 2000 mm3.

  • 4

    Partial response: up to 40% of individual growth inhibition, calculated at the ethical sacrifice of the first control mouse.

  • 5

    Complete regressions.

  • 6

    Tumor-free animal was defined as nonpalpable tumor at the end of the experiment.

IC7Controls30 ± 9  10   
Gefitinib58 ± 22*2870**12282
CDDP + VNR41 ± 811012100
Combined39 ± 7935*8140
IC1LC131Controls14 ± 4  10   
Gefitinib24 ± 8*1037*10500
CDDP + VNR14 ± 30229000
Combined25 ± 9*1147*10201
IC9LC11Controls12 ± 2  7   
Gefitinib37 ± 9*2555*7501
CDDP + VNR15 ± 13187100
Combined46 ± 11*3470*8422
IC8LC10Controls60 ± 8  8   
Gefitinib90 ± 11*3053*7301
CDDP + VNR77 ± 917399400
Combined74 ± 814445300
IC14LC16Controls33 ± 11  9   
Gefitinib48 ± 3*1549*9600
CDDP + VNR36 ± 13309100
Combined65 ± 26*32649600

Antitumor activity of gefitinib and the CDDP+GEM combination

Two NSCLC tumors (IC8LC10 and IC9LC11) responded significantly to the CDDP+GEM combination and three were resistant (IC7, IC1LC131 and IC14LC16) (Table III, Fig. 3). IC9LC11 was the most responsive to chemotherapy, with a TGI of 93% and a 4-fold increase in TGD. Adding gefitinib to chemotherapy did not improve this response. This was also the case for IC8LC10, in which combined treatment was less than additive. Conversely, in the 3 CDDP+GEM-resistant NSCLC tumors, the addition of gefitinib to chemotherapy was associated with an increased response, which was significant in term of TGI for IC1LC131 (p = 0.006 for combination versus gefitinib alone and p = 0.017 for combination versus chemotherapy alone) and in term of TGD for IC14LC16 (p = 0.001 for combination vs. chemotherapy) (Table III).

Figure 3.

Effect of gefitinib and CDDP plus GEM chemotherapy alone or in combination on the growth of the NSCLC xenografts panel. Gefitinib, was given daily per os, at a dose of 120 mg/kg/day for 2 successive weeks; GEM was given I.P. at 60 mg/kg/day, weekly and CDDP, was given I.P. at 1 mg/kg/day, every 21 days. Treatment started when subcutaneous growing tumor volumes were of 100–250 mm3; the tumor growth was measured by measuring 2 perpendicular diameters with a caliper; tumor volume and the relative tumor volume (RTV) was calculated as described in Material and Methods section; growth curves were obtained by plotting mean RTV against time. IC14LC16 was treated over a period of 8 weeks. Error bars represent calculated standard deviation.

Table III. Responses of the 5 NSCLC Xenografts to Gefitinib Alone or Combined with CDDP+GEM
NSCLC xenograftsTreatmentMean ± (SD) TGD (days)1GD (days)2TGI (%)3Nb mice /groupNb PR4Nb CR5Nb of cures6
  • Significantly different from control groups: 0.001 < p < 0.05; **10−6< p < 10−4; ***p < 10−7.

  • 1

    Tumor Growth Delay: delay in days required to reach a 5-fold increase of individual tumor volume from their size at the start of treatment (100–250 mm3).

  • 2

    Growth Delay: difference between tumor growth delays (TGD) in treated vs. control tumor.

  • 3

    Tumor Growth Inhibition: calculated as the ratio of treated/control relative tumor volume on the day of the ethical sacrifice of the first control mouse tumor volume = 2,000 mm3.

  • 4

    Partial response: up to 40% of individual growth inhibition, calculated at the ethical sacrifice of the first control mouse.

  • 5

    Complete regressions.

  • 6

    Tumor-free animal was defined as non palpable tumor at the end of the experiment.

  • 7

    Treatment duration of 8 weeks.

IC7Controls30 ± 9  10   
Gefitinib58 ± 22*2870**12560
CDDP + GEM30 ± 230012001
Combined68 ± 22**3885**9351
IC1LC131Controls17 ± 2  10   
Gefitinib23 ± 2*61310000
CDDP + GEM24 ± 2*71810200
Combined34 ± 8*1748*10201
IC9LC11Controls17 ± 1  8   
Gefitinib42 ± 8**2562*9521
CDDP + GEM64 ± 3***4793**10820
Combined61 ± 2***4494**10280
IC8LC10Controls34 ± 4  10   
CDDP + GEM43 ± 2947*8200
Combined38 ± 1435*12110
IC14LC167Controls23 ± 5  8   
Gefitinib48 ± 8*2560*6310
CDDP + GEM28 ± 155227100
Combined72 ± 23**4981**7501

Global analysis of responses to treatments

Individual responses of mice bearing the 5 NSCLC xenografts were analyzed as a whole, and 3 classes of responses were considered, based on growth inhibition of individual tumors: (i) partial responses, and complete responses (ii) with or (iii) without recurrences (cures) (Table IV). Sixty percent (60%) of NSCLC-bearing mice responded to gefitinib, with a cure rate of 4%. CDDP+VNR was less effective than CDDP+GEM (global responses in 15% and 36% of mice, respectively), but the difference was not statistically significant. The addition of gefitinib to CDDP+VNR did not increase the overall response rate or affect the quality of the response compared to gefitinib alone. On the other hand, the addition of gefitinib to CDDP+GEM increased the frequency of complete regressions (29% vs. 12%, p < 0.05) when compared to gefitinib alone (Table IV).

Table IV. Pooled Responses (Tumor Growth Inhibition) of NSCLC to Gefitinib, CDDP + VNR Alone or Combined with Gefitinib and to CDDP + GEM Alone or Combined with Gefitinib
TreatmentTotal number of treated micePartial responses1 (%)Complete regression2 (%)Cures3 (%)Global responses4 (%)
  • 1

    Partial response 50% of TGI.

  • 2

    Complete regression, not palpable for a transient period.

  • 3

    Cures, no recurrence after complete regressions.

  • 4

    Total number of responses.

  • *

    The χ2 test was used to compare the different treatment groups for the % of partial and complete responses. The level of statistical significance (p) was 0.05.

Gefitinib7029 (41)9 (12)3 (4)41 (60)
CDDP + VNR467 (15)007 (15)
CDDP + VNR + Gefitinib4016 (40)6 (15)3 (7.5)25 (62)
CDDP + GEM4713 (28)2 (4)2 (4)17 (36)
CDDP + GEM + Gefitinib4813 (27)14 (29*)3 (6)30 (62)

Constitutive and gefitinib-induced changes in MAPK signaling downstream of EGFR

As shown in Figure 4, the 5 NSCLC tumors displayed different levels of EGFR expression, which defined 2 groups of tumors: 3 high EGFR-expressing tumors (IC7, IC1LC131 and IC8LC10) and two low EGFR-expressing tumors (IC9LC11 and IC14LC16). The data in Figure 4 also show that the ratio level of phosphorylated/total EGFR in these tumors was much more pronounced in IC9LC11 than in the 4 other tumors. Expression of total or phosphorylated MAPK varied much less among these tumors. Conversely, the IC1LC131 tumor had significantly less phosphorylated Akt, as compared to the other tumors. IC9LC11 expressed higher levels of phosphorylated Jnk but the level of phosphorylated P38 was less. In order to characterize downstream EGFR signaling that could be related to tumor responses, we also examined the effect of gefitinib on the expression of these MAPK. Total and phosphorylated EGFR and MAPK levels were analyzed before and 24 hr after gefitinib treatment in IC1LC131 and IC9LC11, 2 xenografts respectively resistant and sensitive to gefitinib. As shown in Figure 5, western blotting analysis confirmed that EGFR was predominantly phosphorylated in IC9LC11, which denotes a high level of activity. In contrast, low levels of P-EGFR were detected in IC1LC131, which expressed the highest EGFR basal level. Gefitinib treatment strongly reduced the phosphorylation level of EGFR in both tumors. Inhibition of P-Erk and P-Akt was also observed in IC9LC11 treated with gefitinib. Conversely, P-Erk and P-Akt activity remained unchanged after gefitinib treatment in IC1LC131. Gefitinib inhibited both Jnk and P38 phosphorylation in IC1LC131, whereas only P-Jnk was strongly inhibited in IC9LC11.

Figure 4.

Expression of EGFR and related downstream signaling MAPK in NSCLC tumor xenografts. Western blot analysis: 20–30 μg sample of total tumor cell lysates from 5 NSCLC xenografts were separated by SDS-PAGE, transferred to a PVDF membrane, and incubated with a specific antihuman antibody and then with HRP-conjugated secondary antibody. Actin serves as loading control to ensure equal amounts of protein.

Figure 5.

Effects of gefitinib treatment on EGFR and MAPK signaling. Analysis of EGFR and MAPK phosphorylation in IC9LC11 and IC1LC131 before and 24 hr after treatment of mice with gefitinib 120 mg/kg/day. Western blot analysis: 20–30 μg sample of total tumor cell lysates from IC9LC11 and IC1LC131 NSCLC xenografts were separated by SDS-PAGE, as described in Material and Methods section. Actin serves as loading control to ensure equal amounts of protein.

mRNA expression of EGFR, HER-2, MDR-1, MRP-1, MRP-5 and LRP in chemosensitive and chemoresistant NSCLC

The expression level of several candidate genes was compared between the chemoresistant IC1LC131 tumor and the chemosensitive IC9LC11 tumor by quantitative RT-PCR. LRP, MRP-1 and HER-2 mRNAs were 10-, 2- and 3-fold increased respectively in IC1LC131, while the expression of MRP-5 and MDR1 was increased in IC9LC11. Notably, MDR1 transcripts were undetectable in IC1LC131 cells. EGFR mRNA expression was similar in both tumors, in contrast with the large difference observed at the protein level (Table V).

Table V. Real-Time PCR Evaluation of Candidate Gene Expression in IC1LC131 AND IC9LC11 Tumors
TumorArbitrary units
ControlGefitinibCDDP + VNRGefitinib + CDDP + VNRCDDP + GEMGefitinib + CDDP+GEM
  • Numbers represent N-fold difference in target gene expression relative to reference gene. Bold numbers are statistically different from control.

  • 1

    u.d. = undetectable.


The IC1LC131 chemoresistant tumor was characterized by a lack of variation of gene expression (MDR-1, MRP-1, MRP-5, LRP, EGFR and HER-2) after treatment with chemotherapy and/or gefitinib. In contrast, the IC9LC11 (chemosensitive) tumor demonstrated differential expression of receptor and resistance genes upon treatment. Globally, expression increased in a dose-dependent manner after treatment with gefitinib alone. MRP-1, MRP-5 and LRP gene expression were increased over 1.5-fold after gefitinib treatment, both alone and in combination with CDDP+GEM. EGFR expression increased after gefitinib treatment, whereas it had a tendency to decrease in response to chemotherapy. Chemotherapy-induced reduction of EGFR expression was reversed by the combination of chemotherapy with gefitinib. HER-2 expression was not affected by chemotherapy, but was increased 1.5-, 1.2- and 1.7-fold in response to gefitinib alone (120 mg) and gefitinib in combination with CDDP+VNR and CDDP+GEM, respectively. Although the observed changes in gene expression were relatively small, their consistent observation in the IC9LC11 tumor alone and reproducibility in the 3 mice studied (SD ≤ 10% of the mean, not shown) in each treatment group indicates that these changes were indeed real and did not result from experimental variability alone.

EGFR amplification and K-ras mutation status in NSCLC xenografts

The presence of EGFR amplification and of K-ras mutations at codons 12 and 13 was investigated by quantitative genomic DNA PCR and direct sequencing of PCR amplification products respectively. While no NSCLC tumor xenograft harbored K-ras mutations, 1 tumor, IC8LC10, had an amplified EGFR gene across all exons tested (between 14- and 18-fold, Table VI).

Table VI. EGFR Amplification and K-RAS Mutation Status in NSCLC Xenografts
TumorEGFR exon 11EGFR exon 71EGFR exon 201K-ras2
  • 1

    Fold EGFR amplification over normal human lymphocyte DNA.

  • 2

    WT = wild-type sequence of codons 12 and 13 sequence.



As a general rule, NSCLC responds poorly to chemotherapy. The level of chemoresistance is variable, but is observed with all chemotherapy agents. Combination chemotherapies have been shown to have a better therapeutic index than monotherapies, but this benefit is limited to a few patients. Several hypotheses have been advanced to explain the biological basis of drug resistance in NSCLC, including overexpression of drug transport genes,18 but the exact mechanisms remain unknown. The de novo resistance of NSCLC may be attributable, at least in part, to lung tissue characteristics contributing to an exceptional survival propensity in the face of constant contact with xenobiotics, namely tobacco-charged air.19 This survival potential of normal lung tissue, which appears to be amplified in the tumors to which they give rise, may be dependent on the expression of resistance genes, as well as on the activation of survival pathways, such as those mediated by receptor tyrosine kinases (RTK). Indeed, EGFR, 1 of the 4 members of class I RTKs, is highly expressed in numerous epithelial cancers, including NSCLC.20

The NSCLC model used here recapitulates the general features of NSCLC. These xenografts were obtained in our laboratory by directly grafting 27 human tumor samples into the subcutaneous tissue of nude mice. Twelve xenografts were established and 5 were selected in this study. Patient-derived tumor xenografts are more clinically relevant than cell line-derived xenografts, since they very closely resemble the patient's tumors with respect to histology and chemosensitivity. Of the 5 tumors investigated, 2 were chemoresponsive (IC8LC10 and IC9LC11) and 3 were not (IC1LC131, IC7 and IC14LC16). Indeed, the survival time of patients correlated with the xenografts response data: patient survival time was superior to 12 months in case of IC8LC10 and IC9LC11, it was inferior to 12 months in case of IC1LC131, IC7 and IC14LC16. Globally, the drug responses observed in the current study are in accordance with clinical practice,5 with responses ranging from absent to modest, regardless of the chemotherapy regimen used. Only one NSCLC xenograft (IC8LC10) responded to CDDP+VNR (although the response was not statistically significant), while two showed a significant response to CDDP+GEM (IC8LC10 and IC9LC11). It is noteworthy that IC8LC10, which originated from a patient with chemosensitive disease, responded to both chemotherapies. In our model, EGFR was variably but constantly expressed at the mRNA and at the protein level, giving us the opportunity to test the efficacy of new EGFR-TK inhibitors and to investigate the relationship between EGFR expression and signaling and gefitinib response.

Strategies for EGFR-targeted cancer therapies include inhibition of the intracellular tyrosine kinase domain of EGFR by small molecules such as gefitinib. Phase II clinical studies demonstrated that gefitinib as single agent induced major clinical responses in a small percentage of patients.8 In these multiinstitutional randomized Phase II trials, which included more than 200 patients with NSCLC previously treated by chemotherapy, the response rates were 10 and 19 %, after administration of gefitinib at 250 and 500 mg, respectively, with better tolerability at the lower dose. These findings were confirmed by other studies.21 Potential biological markers of response/resistance were not investigated. However, in vitro studies have previously shown that the sensitivity of lung cancer cell lines to EGFR-TKIs is unrelated to the expression level of EGFR,9 and these findings are confirmed in the present study.

We demonstrated dose-dependent, gefitinib-mediated growth inhibition of NSCLC xenografts: at 120 mg/kg, gefitinib reduced tumor growth and prolonged survival in 4 of the 5 tumors tested, whereas at 40 mg/kg no tumor responded significantly. However, prolonged treatment up to 8 weeks of the IC14LC16 tumor with low dose (i.e., 40 mg/kg) gefitinib resulted in significant growth inhibition and prolonged survival, which was equivalent to the activity of a high dose given for a shorter period. Thus, by giving a high dose or by increasing the time of treatment, gefitinib seems to demonstrate higher activity in these NSCLC tumor xenografts than in NSCLC patients with wild-type EGFR. Proportionally, the doses of gefitinib used in our study are 17 and 5.7-fold higher respectively than the highest tolerated clinical dose of 500 mg. There is a 20-fold increase in the MTD of gefitinib in mice compared to man, probably reflecting the large differences in metabolism observed between mice and men. Although we used gefitinib at lower percentages of the MTD (i.e., 80% and 27%) than what was used clinically, pharmacodynamic studies would be required to confirm whether similar intratumoral gefitinib concentrations were achieved in both settings. A recent assay was developed for measuring gefitinib plasma concentration in human and mouse.22 The described method allows a rapid and accurate analysis of the pharmacokinetics of gefitinib by using less sample, as compared to others. The higher dose of gefitinib used in our preclinical study may explain why gefitinib was generally more effective than chemotherapy (see later discussions).

Gefitinib conferred the greatest benefit to chemotherapy when used at a high dose. However, the addition of gefitinib modestly but significantly potentiated CDDP+GEM-based chemotherapy in the 3 chemoresistant tumors but had no effect on the response of the 2 chemosensitive tumors. On the other hand, gefitinib treatment had no additive effect when added to CDDP+VNR-based chemotherapy. The global analysis strategy, in which all data was regrouped without consideration of tumor differences, is a close representation of clinical trial data (each xenograft model representing a patient subgroup).23 Whereas clinical practice precludes accurate comparison of the relative effects of treatment modalities because of patient (tumor) heterogeneity, preclinical evaluation allows such comparison. In the current global analysis, gefitinib yielded a high response rate (60%; predominantly partial responses, with few cures 4%). Chemotherapy was less effective, yielding response rates of 15% and 36% with CDDP+VNR and CDDP+GEM, respectively. The addition of gefitinib to chemotherapy generally conferred an apparent benefit compared to chemotherapy alone, but this was due predominantly to the efficacy of gefitinib itself. The sole benefit of combination with gefitinib was in the extent of response, with an increased rate of complete responses with CDDP+GEM+gefitinib compared to CDDP+GEM (29% vs. 4%, p = 0.05). Therefore, our data are quite similar to the results from the large Phase III clinical studies that failed to show a clinical benefit from adding EGFR inhibitors to chemotherapy in untreated NSCLC patients. This contrasts with earlier preclinical studies using xenografted tumor cell lines, in which the combination of gefitinib with chemotherapy was impressively additive.10 Potential antagonism between gefitinib and chemotherapy has been raised to explain the negative data from the clinical combination studies, reviewed in Ref 23. However, our data are not suggestive of antagonism with our combination therapies; rather they show a higher antitumor efficacy of gefitinib (at high dose) and a lack of potentiation by gefitinib of the efficacy of chemotherapy.

In the IC9LC11 chemosensitive subgroup, 14 of 16 mice (88%) responded to treatment with gefitinib (including 50% complete responses), while only 25% of the IC1LC131 chemoresistant subgroup responded to gefitinib (no complete responses). These data further suggest that the clinical use of RTK inhibitors could be improved by targeting potential responders using predictive biological markers.24

Chemoresistance is a major problem in the therapy of NSCLC. Several mechanisms are thought to be involved in drug resistance, including those associated with drug transport. Here, we investigated the predictive value of biological determinants of antitumor response by quantitative RT-PCR, using both a chemosensitive (IC9LC11) and a chemoresistant tumor (IC1LC131). We focused on genes involved in drug transport and treatment targets (EGFR and HER-2). MDR1 and MRP belong to the ABC (ATP binding cassette) transporter proteins, which contribute to drug resistance by excluding drugs from tumor cells.25 MRP1 was cloned from a human lung carcinoma cell line in 1992.26 LRP is a major vault protein (MVP) found in the cytoplasm and nuclear membrane, and is thought to drive drugs (predominantly taxanes and platinum derivatives) away from the nucleus.27

When analyzing the treatment targets, we observed significant expression of EGFR and HER-2 as previously described in NSCLC.28 EGFR mRNA-expression was similar in both tumors, but HER-2 was lower in the sensitive IC9LC11. EGFR, HER-2, MRP and LRP expression increased in a dose-dependent manner after gefitinib treatment in the IC9LC11 sensitive tumor, while chemotherapy yielded no effect. However, in the IC1LC131 resistant tumor, neither gefitinib nor chemotherapy yielded a transcriptional response. This gefitinib-induced gene expression in the sensitive tumor is compatible with a regulatory role of the EGFR pathway in this tumor, although to our knowledge, no obvious link has been previously established between EGFR signaling and transcriptional regulation of drug transport genes.

Our data are in agreement with the reported lack of predictive value of EGFR expression in patient's response to gefitinib.29, 30 This led us to evaluate other biological markers of gefitinib efficacy. Recently, NSCLC of patients responding to gefitinib were found to have mutations in the EGFR gene.31, 32 In the present study, we performed sequencing analysis in exons 18–21, and found no specific mutations in this NSCLC xenograft panel (data not shown). Therefore, the highly variable response to gefitinib observed in these 5 tumors indicates that factors other than EGFR mutations may significantly influence the antitumor activity of gefitinib in NSCLC. This finding is in agreement with the clinical observations, which indicate that EGFR mutations alone cannot account for the interpatient variability of the clinical response to gefitinib.

Although there is generally no correlation between expression levels of EGFR and response to gefitinib, some recent data point to a possible link between EGFR gene amplification and gefitinib responsiveness in NSCLC, reviewed in.7 However, the presence of amplified EGFR in IC8LC10, one of the low dose gefitinib-resistant xenograft, does not confirm this assumption in the present study.

Likewise, K-ras mutations have been reported to be mutually exclusive with EGFR mutation in NSCLC and are therefore considered as a potential gefitinib resistance marker.33 Since all our tumor xenograft were found to harbor a wild-type K-ras gene at the codon 12 and 13 mutational hot-spot, it cannot explain the differences observed in gefitinib response between our tumor xenografts.

Because activated EGFR signals mainly through downstream MAPK pathways, we examined the effect of gefitinib on MAPK activity using phosphospecific antibodies in IC1LC131 and IC9LC11, 2 xenografts respectively resistant and sensitive to gefitinib. Despite comparable inhibition of EGFR phosphorylation, the extra cellular signal-regulated kinase (Erk) pathway, whose activation is associated with cell proliferation and survival, was inhibited by gefitinib in the sensitive tumor, whereas it remained unchanged in the resistant tumor. Likewise, the gefitinib-sensitive IC9LC11 tumor displayed a strong decrease of phosphorylated Akt in response to gefitinib treatment, while P-Akt was unaffected by gefitinib in the resistant IC1LC131 tumor. These results are consistent with previous studies correlating resistance to gefitinib with the presence of residual Erk or Akt phosphorylation after gefitinib exposure in NSCLC cell lines.34, 35 EGFR-independent activation of the MAPK or Akt pathways that may render tumor cells insensitive to EGFR inhibition could be due to inappropriate activation or inactivation of other regulatory components of the pathways, such as ras, which is frequently activated by mutations in NSCLC.36 Activation of the JNK and p38 pathways is involved in growth arrest and apoptosis induced by stress and various inflammatory cytokines.37 Resistance to gefitinib in IC1LC131 may also be linked to the down regulation of both Jnk and P38 pathways, which was not seen in IC9LC11, where P-P38 expression remained unchanged after gefitinib. The imbalance resulting from the loss of pro-survival (erk and Akt) and the unchanged pro-apoptotic (p38) MAP kinase activities in response to gefitinib may result in higher growth inhibition and/or cell death in the IC9LC11 NSCLC. Indeed, several studies indicated that inhibition of the ERK or PI3K/Akt signaling pathways concomitant with activation of the p38/JNK pathways could switch the equilibrium toward apoptosis in various tumor cell lines.38, 39, 40

In conclusion, pharmacological blockade of EGFR by gefitinib was well tolerated and could be feasibly implemented in combination with chemotherapy in mice. The NSCLC tumor xenografts used here were globally refractory to standard chemotherapy combination and displayed variable sensitivity to gefitinib treatment alone. Combined with chemotherapy, gefitinib yielded globally modest benefits if any, in agreement with data from Phase III clinical studies. The variable responses of NSCLC xenografts to gefitinib were not due to EGFR mutations, K-ras mutations or EGFR gene amplification. Further efforts to identify additional response determinants to gefitinib in NSCLC and other epithelial tumors are thus warranted and may lead to improved clinical use of EGFR inhibitors through rational patient selection and the design of synergistic therapeutic approaches.


We are grateful to Mrs. Claire Chevalier and Mr. Vincent Bordier for excellent technical assistance in animal experimentation. We also thank Dr. Pascale Fehlbaum for Q-RT-PCR-primer and probes sets design and validation, Mrs. Evelyne Boudou and Ms. Catherine Barbaroux for excellent technical assistance (Unité de Pharmacologie, Institut Curie) and Dr. Samir Agrawal, (New Malden, England) for helpful critical review and reviewing the English used in this article.