Efficacy of Gefitinib, an Inhibitor of the Epidermal Growth Factor Receptor Tyrosine Kinase, in Symptomatic Patients With Non–Small Cell Lung Cancer:  A Randomized Trial FREE

More persons in the United States die from non–small cell lung cancer (NSCLC) than from breast, colorectal, and prostate cancer combined.¹ Each year, more than 60 000 persons develop stages IIIB and IV NSCLC; nearly all go on to die from metastatic spread. In addition, most individuals experience symptoms caused directly by lung cancer. These symptoms are often the first manifestations of the illness and increase in frequency and severity as the disease progresses. Cough, shortness of breath, weight loss, loss of appetite, and chest tightness impair the quality of lives already cut short by NSCLC. For patients with advanced lung cancer, physical well-being and changes in quality of life correlate with survival.² Especially among patients with metastatic NSCLC, who are rarely cured, lessening lung cancer symptoms without adding burdensome adverse effects is an important goal of care. No anticancer therapy has been proven to be beneficial to persons seeking additional care after receiving 2 or more chemotherapy regimens.

The epidermal growth factor receptor (EGFR) mediates cancer cell growth, proliferation, invasion, and metastasis, and inhibits apoptosis.³ When ligands bind to the receptor, the molecule is phosphorylated by constitutive tyrosine kinases, causing activation of downstream pathways.⁴ Preclinically, drugs targeting these tyrosine kinases block EGFR activation and the intracellular events that follow.⁵ Compounds disrupting EGFR tyrosine kinases inhibit the growth of human tumors that express EGFR and cause overexpressing tumors to regress.^5- 7 These same agents reduce levels of vascular endothelial growth factor and enhance apoptosis.^8- 9

Many lines of evidence suggest that EGFR has relevance to patients with NSCLC and thus may serve as a potential therapeutic target. Expression of EGFR has been detected by immunohistochemistry testing in from 62%^8- 11 to 93% of resected primary tumors,¹² and EGFR mRNA has been found in 100%.¹¹ The overexpression of EGFR has been variably correlated with clinical outcomes.^11,13- 15

The oral drug gefitinib (ZD1839, Iressa, AstraZeneca Pharmaceuticals, Wilmington, Del) blocks EGFR tyrosine kinases and prevents epidermal growth factor–induced proliferation in cell culture. It inhibits growth and causes regressions in human tumor xenografts with EGFR overexpression.⁵ When given to patients with cancer, gefitinib inhibits EGFR activation in skin.¹⁶ Phase 1 trials identified diarrhea as dose-limiting at daily oral gefitinib doses of 700 to 1000 mg.^17- 20 A continuous gefitinib dosing schedule was developed because it was determined to be the best schedule to counter the continuous oncogenic signaling through this receptor, as seen in animal models⁵ and presumed to occur in persons with cancer. An acne-like rash was also noted. Unlike conventional chemotherapy, gefitinib did not cause myelosuppression, neuropathy, or significant alopecia. In these same phase 1 studies, rapid symptom improvement and radiographic regressions were documented in patients with NSCLC who had previously received chemotherapy.²¹

In light of preclinical activity in EGFR-expressing tumors, evidence of EGFR expression in NSCLC, and antitumor effects in patients with NSCLC persisting after chemotherapy in the phase 1 trials, we initiated this phase 2 trial of gefitinib. We hypothesized that blocking EGFR tyrosine kinases with gefitinib would lead both to symptomatic benefits and objective regressions in patients with NSCLC. We further tested whether there were important differences in outcomes or adverse effects comparing 250-mg and 500-mg doses of gefitinib using a randomized, double-blind, phase 2 design.

From November 2000 to April 2001, 221 patients were enrolled at 30 sites in the United States for the second Iressa Dose Evaluation in Advanced Lung Cancer (IDEAL2) trial. Patients were included if they had pathological confirmation of NSCLC; stage IIIB or IV disease extent²²; treatment with 2 or more regimens containing cisplatin or carboplatin and docetaxel, given either concurrently or as separate regimens; disease progression or unacceptable toxicity with the last chemotherapy regimen; symptomatic NSCLC as determined by a score of 24 or lower out of 28 using the lung cancer subscale of the Functional Assessment of Cancer Therapy-Lung (FACT-L) quality-of-life instrument^23- 24; measurable or evaluable indicator lesions²⁵; World Health Organization performance status of 0-2; and if they had provided written informed consent. Patients were excluded if they had received chemotherapy or irradiation within 14 days; unresolved toxicity greater than grade 2 from prior chemotherapy; neutrophil count less than 1.5 ×10⁹ cells/L, platelet count less than 75 ×10⁹ cells/L, bilirubin level more than 1.25 times the upper limit of normal, and alanine aminotransferase or aspartate aminotransferase levels more than 2.5 times the upper limit of normal; and creatinine clearance less than 30 mL/min (0.50 mL/s).

Patients were randomized to receive either two 250-mg tablets of gefitinib (500-mg total dose) or one 250-mg gefitinib tablet and 1 matching placebo tablet (250-mg total dose) daily. These dosages were chosen, based on phase 1 study results, to maximize the potential for therapeutic activity with an ample safety margin. Responses had been observed at doses as low as 150 mg in the phase 1 studies.²¹ Gefitinib and placebo (both supplied by AstraZeneca Pharmaceuticals) were dispensed on day 1 of each 28-day treatment cycle. One blinded dose reduction from 250 to 100 mg or from 500 to 250 mg was permitted.

Symptom assessments were measured using the FACT-L instrument.^23,26 This instrument was completed pretreatment and then every 28 days. Weekly, patients recorded the presence and severity of 7 symptoms using the lung cancer subscale: shortness of breath, weight loss, clarity of thinking, cough, appetite, chest tightness, and difficulty breathing. Severity was assessed using a 0-4 scale (0-1, most symptomatic; 2-3, less symptomatic; 4, asymptomatic); thus, on the 0 to 28 summed lung cancer subscale score, a score of zero denotes the worst symptoms and 28, none of the 7 symptoms. A 2-point change in the summed score has been proven to correlate with both survival and performance status.^23- 24 Symptom improvement required confirmation of the 2-point increase with no interim worsening over the subsequent 4 weeks. Time to symptom improvement was measured from randomization. The duration of improvement was measured from the first visit at which a 2-point change in the summed lung cancer subscale score was observed until the visit when worsening occurred. Patients were removed from the study based on radiographic criteria or toxicity.

Partial radiographic responses required a greater than 50% decrease in lesion size.²⁵ Imaging studies were repeated 4 and 8 weeks after randomization, then every 8 weeks. As prespecified, the investigators' assessment of response was used in the analysis. Eighty-eight percent of participants had at least 1 bidimensionally measurable indicator lesion. The National Cancer Institute Common Toxicity Criteria (NCI CTC version 2.0) were used to describe and quantitate adverse effects.

Tumor specimens for immunohistochemical analysis for EGFR were requested from all patients. Of the 97 specimens received, 70 were suitable for analysis. Results of these immunohistochemical tests were presented²⁷ and will be reported separately. Baseline characteristics of the 70 patients with adequate samples were similar to the entire group enrolled.

Patients were stratified by World Health Organization performance status (0-1 vs 2) and number of prior chemotherapy regimens (2 vs 3 vs 4 or more). Coprimary end points were rates of symptom improvement and radiographic tumor regression. We included all patients who received any gefitinib. Sample size was chosen to independently evaluate each coprimary end point for each dose. For both rates, 100 patients per group yielded a power of 0.90 for a 1-sided .0125 significance-level test that the rate of symptom or radiographic improvement is 5% or less when the true rate is 15%.²⁸

Secondary end points included overall survival by dose, frequency, and severity of adverse events, and overall quality of life using the FACT-L instrument. Quality of life analyses have been presented²⁹ and will be reported separately. Kaplan-Meier plots were calculated by dose using an unadjusted log-rank test. Analyses were planned to correlate the symptom improvement and radiographic response rates with each other and with survival. We used logistic regression and χ² tests to explore the coprimary outcomes in relation to disease and demographic factors. The data cutoff date for this analysis was August 1, 2001. Radiographic responses were updated to December 17, 2001, when the trial closed. Survival data were updated May 7, 2002. Statistical analyses were carried out using SAS version 8.1 (SAS Institute Inc, Cary, NC) and StatXact version 4 (Cytel Software Corp, Cambridge, Mass); P<.05 was used to determine statistical significance.

From November 7, 2000, to April 6, 2001, 221 patients were enrolled at 30 sites in the United States (listed at the end of this article). The flow of patients through the trial is illustrated in Figure 1. Five individuals who never received gefitinib were excluded after randomization. Baseline patient characteristics are summarized in Table 1. No significant differences were observed between the treatment groups receiving 250 mg and 500 mg of gifitinib. With their last chemotherapy regimen before starting gefitinib, 170 patients (79%) had disease progression and 38 (18%) had intolerable toxicity, mainly peripheral neuropathy. Overall, 58% of patients had received 3 or more prior chemotherapy regimens. Table 2 shows the severity of 7 lung cancer symptoms at baseline. Pulmonary symptoms were most common. The median number of days receiving gefitinib was 56 and 53 for the groups receiving 250 mg and 500 mg, respectively. The median number of days in the study was 56 for both groups. Patients completed 84% of lung cancer subscale assessments while in the study.

Table 3 lists outcome measures. The symptom improvement rate was 43% (95% confidence interval [CI], 33%-53%) for patients who received 250 mg of gefitinib and 35% (95% CI, 26%-45%) for those who received 500 mg (P = .26). Fifty-five percent of symptom improvements in patients receiving 250 mg and 58% of symptom improvements in patients receiving 500 mg were apparent after 1 week of treatment. For those patients with symptom improvement, the median durations of benefit were not reached (range, 1+ to 7+ months for patients receiving 250 mg and 1+ to 8+ months for those receiving 500 mg). For all participants, the best median symptom score improved by 25%, from a baseline of 16 (out of 28) to 22 after treatment with gefitinib (P<.001). For all but 1 study week, the mean change in the summed lung cancer subscale score for all patients was 2 or more, the predefined level for significant symptom improvement (Figure 2). The greatest mean improvement in summed lung cancer subscale scores (4.8) occurred in the patients with partial radiographic responses. Mean changes were 2.6 for individuals with stable disease and 1.0 for those with progression. The change in mean summed lung cancer subscale scores for patients with disease progression did not improve by the prespecified 2-point cutoff defining improvement.

The response rate (all partial) was 12% (95% CI, 6%-20%) for the group receiving 250 mg of gefitinib and 9% (95% CI, 4%-16%) for the group receiving 500 mg (P = .51). The P value for the test that the true rate is greater than 5% was .005 for the group receiving 250 mg and .06 for the group receiving 500 mg. The median duration of radiographic response was 7 (range, 3 to 9+) months for patients receiving 250 mg and 6 (range, 3 to 8+) months for patients receiving 500 mg. Symptoms improved in 96% of patients with partial responses, 73% of those with no partial response but no progression (stable disease), and 17% of those with disease progression (P<.001). Rates of symptom improvement were improved comparing the patients with partial response and those with stable disease (P = .02). Figure 3 displays correlations between symptom improvement and radiographic response.

Symptom improvement and radiographic responses were observed in all patient subgroups. Symptom improvement was more common with adenocarcinoma than with other histologic types (43% vs 30%, P = .06). Response rates were 13% for adenocarcinoma vs 4% for other types (P = .046). The incidence of adenocarcinoma was 79% in women and 58% in men. A multivariable comparison (which included sex, histologic subtype, performance status, age, number of prior regimens, and months from initial diagnosis) demonstrated only female sex to be predictive of response both for symptom improvement (50% vs 31% for women vs men, respectively; P = .006) and radiographic regression (19% vs 3%, P = .001). Eighteen of 22 partial responses (82%) occurred in women.

Partial response rates did not differ significantly whether patients had received 2 (8%), 3 (10%), or 4 or more (15%) prior chemotherapy regimens (P = .38), and rates of symptom improvement were similar based on the number of prior regimens (P = .38). Radiographic response rates did not differ among patients with performance status of 0, 1, or 2 (12%, 9%, and 14%, respectively; P = .53). Rates of symptom improvement also were unaffected by performance status (P = .53).

The projected median survival was 7 months for patients receiving 250 mg of gefitinib and 6 months for those receiving 500 mg (P = .40). The estimated 1-year survival was 27% for patients receiving 250 mg and 24% for those receiving 500 mg (P = .54). Figure 4 displays the overall survival by dose. Using a landmark analysis to examine the survival of only patients who lived at least 2 months (the time needed to assess radiographic response), median survival differences were observed among partial responders (13 months), patients with no partial response but no progression (ie, stable disease) (9 months), and those with progression (5 months) (P<.001). Also using the landmark method, patients with symptom improvement had a median survival of 13 months vs 5 months for those without symptom benefit (P<.001).

Table 4 presents deaths, discontinuations, withdrawals, and gefitinib-related diarrhea and skin toxicities. Skin toxicity, described variably as rash, acne, dry skin, or pruritus, was observed in 62% of patients receiving 250 mg of gefitinib vs 75% of those receiving 500 mg (P = .04). The rash appeared on the face, neck, and trunk, and commonly faded or improved despite continuing therapy. It occurred during the first treatment cycle in 82% of patients. Although all 22 patients with partial responses had some skin toxicity, 65% of those who did not have a partial response also experienced this adverse effect. Skin toxicity was documented in 86% (72/84) of patients with symptom improvement and in 58% (76/132) of those whose symptoms did not improve (observed difference, 28%; 95% CI, 17%-39%).

Diarrhea was noted in 57% of patients receiving 250 mg of gefitinib and in 75% of those receiving 500 mg (P = .006). No routine prophylactic antidiarrheal medication was given. Only 1 patient receiving 250 mg had diarrhea greater than grade 2 (up to 6 daily bowel movements) compared with 6 patients in the group receiving 500 mg. Two individuals receiving 500 mg withdrew because of diarrhea. In 76% of patients, diarrhea was observed during the first treatment cycle. Symptoms were generally controllable with loperamide taken after each bowel movement. Approximately one third took an antidiarrheal medication. Diarrhea was documented in 82% (69/84) of patients with symptom improvement and in 56% (74/132) whose symptoms did not improve (observed difference, 26%; 95% CI, 14%-38%).

Nineteen percent of patients reported grade 1 or 2 eye toxicities such as redness or itchiness. All were self-limited and in no case led to study withdrawal. Treatment-related vomiting or nausea (grade 1 or 2 only) was observed in 15% and 10% of patients, respectively. There were no cases of investigator-identified interstitial lung disease following gefitinib administration (observed rate, 0%; 95% CI, 0%-1.7%). Pulmonary events (collected as pneumonia, aspiration pneumonia, lung disorder, respiratory distress syndrome) were noted in 13 patients (6 of grade 3 or 4) receiving 250 mg of gefitinib and in 14 patients (8 of grade 3 or 4) receiving 500 mg. None of the pulmonary events were considered drug-related by the investigators. One patient had grade 3 thrombocytopenia and 3 had reversible grade 3 elevations of alananine aminotransferase and aspartate aminotransferase levels that were deemed drug-related. No grade 3 or 4 neutropenia, anemia, or neuropathy occurred.

Only 1 possible treatment-related death was recorded. This patient experienced cavitation of his primary tumor, developed massive hemoptysis, and died on day 11. Only 1 patient receiving 250 mg of gefitinib and 5 patients receiving 500 mg experienced a drug-related adverse event leading to study withdrawal. Dose reductions for toxicity occurred in 1 patient receiving 250 mg and 10 receiving 500 mg. Grade 3 or 4 drug-related adverse events were observed in 7 patients receiving 250 mg and 20 patients receiving 500 mg. Thirty-day all-cause mortality was 3.8% for patients receiving 250 mg and 8.8% for those receiving 500 mg. Sixty-day all-cause mortality was 8.8% for patients receiving 250 mg and 18% for those receiving 500 mg.

This trial demonstrated that oral gefitinib given once daily caused rapid symptom improvement and tumor regressions in patients with NSCLC. Until now, only chemotherapy, surgery, and radiotherapy have demonstrated the ability to cause lung tumors to regress. Once these modalities have been exhausted, only supportive care measures remain. Gefitinib, which was designed to achieve its anticancer effects through a different mechanism, can help fill this therapeutic void.

There is no comparable prospective series treating a cohort of symptomatic patients who had received both cisplatin or carboplatin and docetaxel. In a retrospective review of 43 individuals treated with various chemotherapies, the response rate for a third regimen was 2%.³⁰ One study randomized individuals who had received 1 or more chemotherapy regimen(s) to either docetaxel or supportive care alone. The response rate with docetaxel was 6%. Those who did not receive chemotherapy had a median survival of 5 months.³¹ In another "second-line" trial conducted in the United States, radiographic tumor regressions were induced in 7% of patients receiving docetaxel vs in 1% of those receiving vinorelbine or ifosfamide.³² The 10% response rate with gefitinib, achieved without myelosuppression or neurotoxicity, and with virtually no hair loss, is provocative in comparison. The results of this trial are consistent with the gefitinib phase 1 experience in patients with NSCLC.²¹ The recent international phase 2 trial (IDEAL1) also compared 250-mg and 500-mg gefitinib doses, but in patients pretreated with 1 or 2 prior chemotherapy regimens who were not required to have symptoms at trial entry. For the 250-mg dose, they reported an 18% radiographic response rate.³³ Similar efficacy has been observed with the EGFR tyrosine kinase inhibitor erlotinib (OSI-774, Tarceva, OSI Pharmaceuticals, Melville, NY).³⁴ In our study, approximately 15% of patients with the "best response" of progression had some symptomatic improvement by study criteria. This likely reflects either a placebo effect or the resolution of adverse effects of the prior chemotherapy regimens.

Adverse effects of gefitinib were generally mild, manageable, noncumulative, and reversible with the cessation of the drug and sometimes even with continued use. Among individuals receiving 250 mg, some degree of skin toxicity occurred in 62% and of diarrhea in 57%. For the 250-mg dose, toxicity caused just 1 patient to stop taking gefitinib and 1 to reduce the dosage. Interstitial lung disease has been associated with use of gefitinib in Japan and reported to occur in 1% to 2% of patients.³⁵ This is a recognized but uncommon adverse effect of cytotoxic drugs³⁶ and also has been described following treatment with the tyrosine kinase inhibitor imatinib (STI571, Gleevec, Novartis Pharmaceuticals, Basel, Switzerland).³⁷ With no case of interstitial lung disease reported in this trial, the 95% confidence limit for the true incidence of this complication after gefitinib administration ranges from 0% to 1.7%, lower than that observed with many chemotherapeutic agents.³⁶

There were no significant differences in the incidence of symptomatic or radiographic improvement between the 250-mg and 500-mg doses of gefitinib. The incidence and severity of both rash and diarrhea, however, were higher among patients receiving 500 mg. Consistent with the proposed mechanism of action of gefitinib, once plasma levels adequate to block tyrosine kinases such as EGFR have been achieved, additional dose escalations are unlikely to improve response and will increase toxicity. That is likely what we observed in this study. We recommend the 250-mg dose. The IDEAL1 trial reached the same conclusion.³³

Can the amount of EGFR protein in tumors predict response to gefitinib? The answer appears to be no. In a combined analysis of tumor EGFR expression levels determined by immunohistochemistry in 157 analyzable specimens submitted from patients enrolled in the IDEAL1 trial and in this study, there were no consistent associations between levels of EGFR expression and radiographic or symptomatic improvements.²⁷ In clinical trials of the EGFR inhibitors cetuximab and erlotinib, response also did not correlate with the degree of EGFR expression measured by immunohistochemistry.^34,38

Of all clinical characteristics, only female sex and adenocarcinoma histological type showed a correlation with response, both in our study and in the international study.³³ Patients in Japan also had higher rates of radiographic response than those enrolled from other countries.³³ Investigators from Memorial Sloan-Kettering Cancer Center also have reported that in a multivariable analysis of 140 patients that included 6 people treated as part if this trial, individuals who have never smoked cigarettes and those with any bronchioloalveolar histologic features in their tumor specimens are more likely to respond to gefitinib.³⁹ Although female sex has been associated with longer survival in patients with advanced NSCLC, it has not been found to be a predictor of radiographic response with chemotherapy.⁴⁰ It is likely that the consistently higher sensitivity to gefitinib in women, Japanese patients, persons with adenocarcinoma histological type (especially bronchioloalveolar), and those who have never smoked cigarettes has a biological basis. Understanding these observations may help reveal the mechanisms that underlie response to gefitinib. At this time, we do not recommend using any pretreatment characteristics to routinely select patients to receive gefitinib. Although we observed higher rates of response in women, radiographic regressions were nonetheless documented in men and symptoms were improved in 31%. For now, the only way to predict which patients will benefit from gefitinib is to administer the drug and observe its effects, which are quickly apparent. In this trial, symptom improvement was documented in 75% of responding patients within 3 weeks and in 85% by 4 weeks.

Why do these results not appear as striking as those from other "targeted" approaches? The pathobiology of growth-factor signaling in lung cancer may be fundamentally different from that in other tumors, in which blockade of growth-factor receptors has proved to be of benefit. In breast cancer, persons who respond to trastuzumab (Herceptin, Genentech Inc, San Francisco, Calif) uniformly have high expression of HER2 protein caused by HER2 gene amplification. In lung cancer, even high-level EGFR protein expression is not commonly linked to gene amplification.¹⁰ Growth-factor–mediated proliferation in NSCLC also is different from that observed in gastrointestinal stromal tumors, in which specific mutations of either the c-KIT or PDGFRA genes cause ligand-independent activation of receptor proteins that drive tumor growth.^41- 44 No similar controlling mutations of EGFR have been reported in NSCLC. The targeted therapy imatinib inhibits several tyrosine kinases, including those expressed by c-KIT and PDGFRA.⁴⁵ Imatinib causes dramatic regressions in gastrointestinal stromal tumors regardless of whether the controlling mutations are in the PDGFRA or c-KIT genes, because it effectively blocks both tyrosine kinases and the resulting oncogenic signals emanating from either receptor. Similarly, gefitinib may inhibit other tyrosine kinases in addition to EGFR. If this occurred in patients with lung cancer, oncogenic signaling in tumors in the fraction of patients who derive substantial benefit from gefitinib could be driven at least in part by activation of other tyrosine kinases in addition to or instead of EGFR.

Preclinical studies have demonstrated the uncoupling of the effects of EGFR tyrosine kinase inhibition and tumor growth.^46- 47 This observation may be relevant in NSCLC. Since multiple genetic lesions are necessary to initiate lung cancers, many cell-signaling pathways may be aberrant and provide multiple mechanisms to maintain tumors and permit "signaling redundancy." For example, Kristen-ras mutations,⁴PTEN promoter hypermethylation,⁴⁸PI3-Ká amplification,⁴⁹ and p53 mutations⁵⁰ are all downstream of EGFR and occur in lung tumors. Any or all of these aberrations may be present in patients whose tumors fail to respond to gefitinib. We may be able to help individuals with gefitinib-resistant tumors by determining which downstream factor works in concert with EGFR and then designing a combination therapy inhibiting both EGFR and the downstream aberration as well. Characterization of the presence or absence of these additional lesions in a given patient's tumor also could lead to more discriminate use of gefitinib.

The strength of downstream oncogenic signaling can be determined by which member of the HER family dimerizes with EGFR.⁴ Identifying the presence of other HER family members in tumors and their degree of dimerization with EGFR may help identify persons with lung cancer more likely to be sensitive to gefitinib. Investigations also have revealed that amino acid substitutions in the ATP binding pocket in regions sequentially distant but conformally important to the imatinib binding site have been identified in both clinically and laboratory induced mutations of the Abelson tyrosine kinase in chronic myelogenous leukemia.⁵¹ These substitutions confer resistance to imatinib. This mechanism may have relevance to understanding gefitinib resistance and provides yet another direction for research.

The results of this trial proved our study hypothesis. Blocking EGFR tyrosine kinase with gefitinib leads to symptom improvement and radiographic regressions in patients with NSCLC. These findings support the use of gefitinib for the treatment of patients with NSCLC who have received cisplatin or carboplatin and docetaxel, and other agents. The magnitude and duration of benefits, coupled with the safety of gefitinib, justifies its use in patients previously treated with chemotherapy. These results further demonstrate that a therapy that disrupts biological pathways specific for cancer cells can improve the outcome of patients with advanced NSCLC.