A Phase II Pharmacodynamic Study of Erlotinib in Patients with Advanced Non–Small Cell Lung Cancer Previously Treated with Platinum-Based Chemotherapy

Erlotinib, an epidermal growth factor receptor (EGFR) tyrosine-kinase inhibitor (TKI) significantly prolongs survival in unselected patients with advanced non–small cell lung cancer (NSCLC) who have previously received chemotherapy (1). A relevant question is whether we can more accurately predict which patients will obtain clinical benefit from erlotinib. Several studies indicate that EGFR gene mutations, and increased EGFR gene copy number, are linked with response to EGFR TKIs, with conflicting results for survival (2–10).

Pharmacodynamic studies can provide valuable information on the effectiveness of targeted agents, and identify potential markers of clinical benefit. Pharmacodynamic studies of both erlotinib and gefitinib have been done using skin biopsies (11, 12). However, this approach has limitations (13): drug distribution may differ between skin and tumor tissue, and the response of skin and tumor cells to EGFR inhibition may be qualitatively different. Furthermore, tumors may harbor gene mutations and/or changes leading to constitutive activation of intracellular signaling pathways. Ideally, therefore, the pharmacodynamic effects of EGFR inhibition should be studied in tumor tissue.

We conducted a pharmacodynamic study of erlotinib, involving the prospective collection of tumor tissue samples from NSCLC patients. Our primary objectives were to determine the genetic profile of patients who benefit from erlotinib treatment, and to analyze the effect of erlotinib on EGFR and its downstream signaling pathway. We also prospectively assessed the feasibility of performing biopsies for research purposes in patients with NSCLC.

Study design. This phase II trial was conducted in two institutions: Vall d'Hebron University Hospital and the Grosshansdorf Hospital. Tumor biopsies were taken from all patients at baseline (≤2 wk before treatment start). Patients received erlotinib (150 mg/d p.o.) until disease progression, unacceptable toxicity, or consent withdrawal. A second biopsy was to be done after 6 wk' treatment. All biopsies were initially done using an 18-gauge needle, guided by computed tomographic scan in the presence of a pathologist. After the inclusion of 36 patients, 1 patient died during biopsy. The trial was temporarily stopped and the protocol amended to increase the safety of biopsies: patients with previous thoracic radiotherapy were excluded; spirometry was required, forced expiratory volume in 1 s having to exceed 50% of the age-adjusted volume; and the 6-wk biopsy was suspended. Also, bronchoscopy was included as an alternative biopsy method, after the experience of one participating institution (Grosshansdorf Hospital).

Patient eligibility. Eligibility criteria were histologically/cytologically confirmed advanced NSCLC, with progression after ≥1 platinum-based chemotherapy regimen; tumor tissue accessible at biopsy; age of >18 y; Eastern Cooperative Oncology Group performance status of 0 to 1; life expectancy of ≥12 wk; ≥14 d (radiotherapy) or 28 d (chemotherapy) since last treatment; adequate bone marrow, hepatic, and renal functions; and written informed consent. The trial complied with the Declaration of Helsinki and principles of Good Clinical Practice, and was approved by appropriate ethics committees. A sample size of 80 patients was calculated to give 8 to 10 patients with response to erlotinib.

Assessments. Medical histories, physical examinations, and biochemical and hematologic assessments were done at screening and were repeated every 3 wk. Computed tomographic scans were obtained within 28 d before enrollment. Radiologic investigations were done every 6 wk.

Tumor response and survival. Response was assessed using Response Evaluation Criteria in Solid Tumors. Clinical benefit was defined as a complete or partial response, or stable disease lasting ≥12 wk. Early progressive disease was defined as progression before 12 wk.

Progression-free survival was measured from enrollment to disease progression (or death, whichever occurred first). Patients with no documented progression were counted as censored observations. The duration of overall survival was measured from enrollment to the date of death, or until the last study visit for survivors.

Tolerability. Adverse events were assessed using the National Cancer Institute Common Toxicity Criteria version 2.0.

Pharmacodynamic assessments. Formalin-fixed, paraffin-embedded tumor tissue samples were processed as described previously (14). All samples were blinded as to whether they were pretreatment or on treatment, and assessed by pathologists, who also verified the presence of adequate amounts of tumor tissue.

Table 1 lists the probes used in all analyses. Immunohistochemical analyses of EGFR signaling pathway components included EGFR, phosphorylated EGFR (p-EGFR), phosphorylated p42/44 mitogen-activated protein kinase (p-MAPK), phosphorylated AKT (p-AKT), and cyclin-dependent kinase inhibitor p27^kip1. A marker of tumor proliferation (Ki-67) was also evaluated. For assessment of immunohistochemical staining, the percentage of target cells stained was evaluated in 10 optical microscope fields per tissue section (×400 magnification) and the average percentage staining was calculated. Staining intensity was scored as 1 (mild), 2 (moderate), and 3 (intense). H-scores were calculated as the percentage of positive staining (0-100) × the correspondent staining intensity (0-3). Specimens with H-scores of ≤150 and >150 were classified as immunohistochemical negative and immunohistochemical positive, respectively.

Levels of apoptosis were determined using a terminal deoxynucleotidyl-transferase–mediated dUTP nick end labeling assay. Four-micrometer sections of paraffin-embedded tissue were stained with 12-dUTP-fluorescein (Roche Diagnostics GmbH) and counterstained with 0.1 μg/mL 4-diamidino-2-phenylindone. Prepared sections were examined using a fluorescence microscope (Nikon Eclipse E400) at ×400 magnification. The apoptotic index was calculated as the mean percentage of fluorescent cells assessed in 10 randomly selected microscope fields.

EGFR gene copy number was evaluated in 4-μm sections of paraffin-embedded tissue using a dual-target, dual-color fluorescence in situ hybridization (FISH) assay, as described previously (8). The number of EGFR gene copies was determined using a fluorescence microscope to assess at least 100 nonoverlapping nuclei with intact morphology. The EGFR FISH status of tumors was classified as described previously (7).

Mutations in exons 18 to 21 of the EGFR gene and exons 2 to 3 of the K-ras gene were evaluated by PCR followed by sequence analysis. The relevant exons were amplified from tumor lysate (a minimum of 1,000 tumor cells) using a nested PCR protocol. For each fragment, two independent PCR products were sequenced on both strands, using dye terminator chemistry on an ABI 3730 sequencer. Mutations were identified using Polyphred mutation analysis, and confirmed if they were present in at least two independently obtained PCR products.

Statistical analyses. Progression-free survival, overall survival, and 95% confidence intervals (CI) were calculated by the Kaplan-Meier method (15). χ² or Fisher's exact tests were used to evaluate the relationship between the levels of different markers and clinical outcome. The Wilcoxon signed-rank test was used to compare the results from paired samples obtained before treatment and after 6 wk' treatment. Two-sided P values of <0.05 were considered significant. All analyses were done using SPSS version 10.0 (SPSS, Inc.) and SAS version 8.2 (SAS Institute, Inc.).

Between September 2002 and June 2005, 83 patients were enrolled (Table 2). Table 3 summarizes the biopsies done and the samples suitable for analysis. Not all analyses were possible for every patient, due to the limited amount of tissue obtained. The priority order of molecular analyses was immunohistochemical analysis, FISH, and mutation analysis.

One patient, a 45-year-old woman with bilateral lung metastases who had progressed after three lines of chemotherapy, died during her baseline computed tomographic–guided fine-needle biopsy. At inclusion, the patient had moderate dyspnea due to lung metastases. During biopsy, the patient coughed and expectorated a moderate amount of blood. She immediately became unconscious and respiratory arrest followed. She was intubated and, despite intensive medical support, died 24 hours later. Intrapulmonary hemorrhage in a patient with limited pulmonary reserve was considered the most likely cause of death. No autopsy was done. Due to this event, study enrollment was temporarily interrupted until the protocol was amended as described in Materials and Methods. No other biopsy complications occurred.

Seven of 73 evaluable patients (10%) achieved a partial response with erlotinib, 28 (38%) had stable disease, and 38 (52%) had progressive disease. Eighteen patients (25%) experienced clinical benefit. The median progression-free survival was 44 days (95% CI, 42-77 days) and median overall survival was 120 days (95% CI, 103-193 days). The 1-year survival rate was 16.6%.

Erlotinib was well-tolerated, the only clinically relevant toxicities being rash and diarrhea. Rash occurred in 61% of patients (grade 1, 28%; grade 2, 26%; grade 3, 7%) and diarrhea in 23% (grade 1, 14%; grade 2, 7%; grade 3, 2%). Other toxicities, which were generally mild or moderate, included nausea (two patients), fatigue (five), eye toxicity (three), and liver enzyme elevation (four). No lung toxicity was observed.

Relationship between baseline expression of EGFR markers and clinical outcome. Baseline levels of EGFR markers were determined in tumor biopsies from 53 patients. Of these, 13 patients had clinical benefit (3 partial responses and 10 stable disease >12 weeks), 36 had progressive disease, and 4 were nonevaluable for response. There were no significant differences in the baseline expression of EGFR, p-EGFR, p-MAPK, p-AKT, p27, and Ki-67 between patients with clinical benefit and those with early progressive disease (data not shown).

Response and survival according to EGFR gene copy number.EGFR amplification or high polysomy (FISH positive) was detected in 15 of 57 patients (26%). The characteristics of FISH-positive and FISH-negative patients were similar with regard to gender, smoking status, histology, and performance status (data not shown). Five patients with FISH-positive tumors had partial response (33%), whereas no partial responses were recorded in the FISH-negative group. Ten of the FISH-positive patients (67%) had clinical benefit, compared with 5 of the FISH-negative patients (12%; Pearson χ² test, P < 0.001).

The median progression-free survival in FISH-positive patients was 137 days (95% CI, 79-227), compared with 43 days (95% CI, 41-73) for the FISH-negative group [Fig. 1A; P_log-rank = 0.002; hazard ratio (HR), 0.37]. Median overall survival was also longer in the FISH-positive group (226 days; 95% CI, 132-318) than the FISH-negative group (106 days; 95% CI, 79-193), but the difference was not statistically significant (Fig. 1B; P_log-rank = 0.267; HR, 0.70). The EGFR FISH–positive tumors had higher levels of EGFR (P = 0.028) and p-EGFR (P = 0.0005) than those which were EGFR FISH negative.

Influence of EGFR and K-ras mutations on response and survival.EGFR mutations were identified in 5 of 39 patients (13%); all patients with mutations had adenocarcinoma. Of the five patients with EGFR mutations, FISH analysis was done in three; two were EGFR FISH negative and one was FISH positive. For patients with mutations, there were 2 partial responses (L747-S752 deletion; L747-T751 deletion), 1 stable disease (L861Q mutation) and 1 progressive disease (R776H mutation), compared with 1 partial response, 9 stable disease, and 21 progressive disease among the 34 patients with wild-type EGFR. The median progression-free survival among patients with EGFR mutations was 205 days (95% CI, 21-307), compared with 43 days (95% CI, 36-50) for those with the wild-type gene (P_log-rank = 0.0878; HR, 0.354). Overall survival was 205 days (95% CI, 21-325) in patients with EGFR mutations, compared with 113 days (95% CI, 86-169) in those with wild-type EGFR (P_log-rank = 0.8468; HR, 1.109).

K-ras mutations were identified in 7 of 39 patients (18%); all patients with mutations were current or former smokers (no patients had both EGFR and K-ras mutations). No patient with a K-ras mutation had an objective tumor response during erlotinib treatment. The median progression-free survival was 43 days in patients with K-ras mutations (95% CI, 27-118) and in those with wild-type K-ras (95% CI, 34-77; P_log-rank = 0.8955; HR, 1.058). The median overall survival in patients with K-ras mutations was 134 days (95% CI, 59-220), versus 111 days (95% CI, 86-205) in those with the wild-type gene (P_log-rank = 0.6454; HR, 1.238).

Effect of treatment with erlotinib on biomarker expression. Of the 14 patients with suitable paired biopsy samples, 5 experienced clinical benefit. Table 4 summarizes the molecular markers of these 14 patients and clinical benefit observed. Overall, after 6 weeks' treatment with erlotinib, there were significant reductions in p-EGFR (P = 0.002), p-MAPK (P = 0.001), and Ki-67 (P = 0.025). There were no significant changes in EGFR (P = 0.332), p-AKT (P = 0.455), p27 (P = 0.272), or in the apoptotic index (P = 0.410; Fig. 2). However, apoptosis was significantly increased among patients who had clinical benefit, compared with those with early progressive disease (P = 0.029; Fig. 3).

In this study, treatment with erlotinib (150 mg/d) achieved a tumor response in 10% of patients and stable disease in a further 38%. These findings resemble those of the BR.21 placebo–controlled phase III study of erlotinib in patients with chemotherapy-refractory advanced NSCLC (1).

Identifying the NSCLC patients likely to obtain clinical benefit from EGFR TKI treatment is relevant for the development of these agents. Studying drug exposure and sensitivity markers in tumors is challenging as it is difficult to obtain sequential tumor samples for research. The feasibility of obtaining serial biopsies was previously shown in breast cancer patients (14). In the present study, in NSCLC patients, the success rate of obtaining samples suitable for analysis was higher using computed tomographic–guided fine-needle biopsy (68.8%) than it was with bronchoscopy (54.2%). However, one patient died during a guided fine-needle biopsy procedure; potential complications with this procedure are pneumothorax, hemorrhage, and systemic embolism, and the risks of these are probably higher when using an 18-gauge needle rather than one of smaller diameter (16). Bronchoscopy is generally considered a safer method to obtain samples.

A minority of patients with treatment-refractory NSCLC achieve a radiologic response to treatment with EGFR TKIs. However, more achieve prolonged stable disease, which contributes to the survival benefit observed with erlotinib (1). In the present trial, we used tumor tissue obtained immediately before starting erlotinib therapy to examine the relationship between baseline levels of various biomarkers and clinical benefit with erlotinib. We found no differences in the baseline levels of EGFR or p-EGFR (measured using immunohistochemical analysis) between patients with or without clinical benefit. To date, no clear association has been identified between levels of EGFR expression and response or survival with EGFR-targeted agents (17–20). A retrospective analysis of the BR.21 trial suggested that erlotinib prolonged survival in EGFR-positive patients, but in EGFR-negative patients, there was no apparent survival advantage (9). In the phase III Iressa Survival in Lung Cancer study, gefitinib failed to significantly improve survival compared with placebo in relapsed NSCLC, and although survival seemed to be improved in patients with EGFR-expressing tumors, compared with EGFR-negative tumors, a significant overall benefit was not observed in either group (21, 22).

In our study, no apparent relationship was found between baseline levels of p-MAPK, p-AKT, or p27 and clinical benefit. Some studies have suggested that high levels of p-AKT are associated with better outcomes in patients treated with gefitinib (23, 24), although in the Iressa Survival in Lung Cancer study, p-AKT was not significantly associated with survival outcome on gefitinib (22).

Somatic mutations of the EGFR gene have been associated with sensitivity to EGFR TKIs (2–5, 25–27). In this study, EGFR mutations were identified in five patients, all of whom had adenocarcinoma; three experienced clinical benefit with erlotinib and one progressed during treatment. Although not statistically significant, a substantial difference in progression-free survival was noted for patients with EGFR mutations compared with those who had wild-type EGFR.

In the present study, as previously reported (28, 29), no patients with K-ras mutations had concomitant EGFR mutations, and no responses to erlotinib were observed in these patients. Eberhard et al. (30) reported that during treatment with erlotinib plus chemotherapy, patients with K-ras mutations had reduced time to progression and reduced overall survival. However, in the present study, erlotinib-treated patients with K-ras mutations had similar survival times to those with wild-type K-ras.

High EGFR gene copy number, assessed using FISH, has also been proposed to be a potential predictive marker for clinical benefit with EGFR TKIs (7, 8). A retrospective subanalysis of the BR.21 trial suggested a difference in response rate in favor of patients with FISH-positive tumors (20% versus 2.4%, P = 0.03; ref. 9). However, a multivariate analysis showed that EGFR FISH status was not a significant predictive factor for survival. In the Iressa Survival in Lung Cancer study, gefitinib was marginally beneficial in patients with high EGFR gene copy number (HR, 0.61; 95% CI, 0.36-1.04; P = 0.067) but not in patients without high EGFR gene copy number (22). In the current study, we observed differences in objective response rate and progression-free survival in favor of patients with FISH-positive tumors. These results indicate that further study of the relationship between EGFR gene copy number and clinical outcome with erlotinib is warranted.

Erlotinib has been shown to induce the cell-cycle inhibitor p27 and to suppress the cell cycle promoter cyclin D1, thereby blocking cell-cycle progression at the G₁ phase (31). In skin biopsy samples, erlotinib attenuated EGFR activation and increased expression of p27 (11). In the current study, although erlotinib had no significant effect on EGFR levels, there were significant reductions in the levels of p-EGFR and p-MAPK, as well as in the proliferation marker, Ki-67. However, as in breast cancer, the inhibition of the EGFR pathway in NSCLC is not sufficient for an antitumor effect in all patients.

A biologically relevant observation from this study is the increased apoptosis observed in those patients who obtained clinical benefit. Increased apoptosis may therefore be a marker for benefit from erlotinib therapy and, in preoperative studies, could identify at surgery those patients who have benefited from therapy, thereby assisting in the choice of postoperative adjuvant therapy. This view is supported by preclinical studies. In a transgenic mouse model of EGFR-mutant lung adenocarcinoma, tumor regression associated with erlotinib treatment was accompanied by decreased mitoses and increased apoptosis (assessed by measurement of activated caspase-3; ref. 32). Gong et al. (33), in a panel of human lung cancer cell lines that harbor EGFR mutations, showed that EGFR kinase inhibition in drug sensitive cells provokes apoptosis via the intrinsic pathway of caspase activation. Although specific mechanisms underlying erlotinib-induced cell death are yet to be clearly defined, p-AKT induction seems to be required for erlotinib-induced apoptosis, as do other signaling components downstream of EGFR. In our study, there was a slight decrease in pAKT levels in those patients who had clinical benefit, albeit without statistical differences. The second tumor sample was obtained after 6 weeks of erlotinib treatment and one point to clarify is the effect of when changes in pAKT levels actually take place. Our findings also support further study in NSCLC of erlotinib combined with other apoptosis-enhancing agents, such as mammalian target of rapamycin or phosphoinositide 3-kinase inhibitors.

In summary, the findings of this study concur with those of previous research and suggest that EGFR gene copy number assessed by FISH should be evaluated further as a predictive marker for clinical benefit with erlotinib in patients with NSCLC. Erlotinib suppressed EGFR-mediated signaling, and apoptosis increased in patients with clinical benefit.

A. Heller is employed by Roche Diagnostics GmbH; B. Klughammer, H. Maacke, D. Foernzler, and J. Mocks are employed by F. Hoftmann-La Roche Ltd. U. Gatzemeirer has a commercial research grant with Roche and receives other commercial research support from Lilly. The following authors have received honoraria from the indicated companies: M. Reck, Hoftmann-La Roche, Lilly, Boehringer, Bayer Healthcare; U. Gatzemeier, Roche, Lilly, Astra Zeneca; J. Baselga, Roche.

For Ulrich Brennscheidt, MD, who regrettably died before finalization of this manuscript, for his invaluable contribution to the design and conduct of the study. We thank M. Simpson, PhD, of Gardiner-Caldwell Communications, who provided medical writing support funded by F. Hoffmann-La Roche Ltd.