Clinical Characteristics of Patients with a Response to Gefitinib
Patients with advanced, chemotherapy-refractory non–small-cell lung cancer have been treated with gefitinib as a single agent since 2000 at Massachusetts General Hospital. A total of 275 patients were treated, both before its approval on May 2003 by the Food and Drug Administration (FDA), as part of a compassionate-use expanded-access program, and subsequently, with the use of a commercial supply. During this period, 25 patients were identified by physicians as having clinically significant responses to the drug. A clinically significant response was defined as a partial response according to the response evaluation criteria in solid tumors18 for patients with measurable disease; for patients whose tumor burden could not be quantified with the use of these criteria, the response was assessed by two physicians.
A computed tomographic scan of the chest in Patient 6 shows a large mass in the right lung before treatment with gefitinib was begun (Panel A) and marked improvement six weeks after gefitinib was initiated (Panel B).
Table 1 shows the clinical characteristics of nine patients for whom tumor specimens obtained at the time of diagnosis were available. Tissue was not available from the other patients with a response to gefitinib, most commonly because diagnostic specimens were limited to needle aspirates. As a group, the nine patients derived a substantial benefit from gefitinib therapy. The median duration of survival from the start of drug treatment exceeded 18 months, and the median duration of therapy was greater than 16 months. Consistent with previous reports, we found that most patients with a response to gefitinib were women, had never smoked, and had bronchoalveolar tumors.11,12 Patient 6 was representative of the cohort. This patient, a 32-year-old man with no history of smoking, presented with multiple brain lesions and bronchoalveolar carcinoma in the right lung. He was treated with whole-brain radiotherapy, followed by a series of chemotherapy regimens (carboplatin and gemcitabine, docetaxel, and vinorelbine) to which his tumor did not respond. With a declining functional status and progressive lung-tumor burden, he started therapy with 250 mg of gefitinib per day. His dyspnea promptly improved, and computed tomography of the lung six weeks after the initiation of treatment revealed a dramatic improvement (Figure 1).
EGFR Mutations in Patients with a Response to Gefitinib
We hypothesized that patients with non–small-cell lung cancer who had striking responses to gefitinib had somatic mutations in the EGFR gene that would indicate the essential role of the EGFR signaling pathway in the tumor. To search for such mutations, we first looked for rearrangements within the extracellular domain of EGFR that are characteristic of gliomas15; none were detected. We therefore sequenced the entire coding region of the gene using PCR amplification of individual exons.
Panels A, B, and C show the nucleotide sequence of the EGFR gene in tumor specimens with heterozygous in-frame deletions within the tyrosine kinase domain (double peaks). Tracings in both sense and antisense directions are shown to demonstrate the two breakpoints of the deletion; the wild-type nucleotide sequence is shown in capital letters, and the mutant sequence is in lowercase letters. The 5′ breakpoint of the delL747–T751insS mutation is preceded by a T-to-C substitution that does not alter the encoded amino acid. Panels D and E show heterozygous missense mutations (arrows) resulting in amino acid substitutions within the tyrosine kinase domain. The double peaks represent two nucleotides at the site of heterozygous mutations. For comparison, the corresponding wild-type sequence is also shown. Panel F shows dimerized EGFR molecules bound by the EGF ligand. The extracellular domain (containing two receptor ligand [L] domains and a furin-like domain), the transmembrane region, and the cytoplasmic domain (contai
ning the catalytic kinase domain) are highlighted. The position of tyrosine1068 (Y1068), a site of autophosphorylation used as a marker of receptor activation, is indicated, along with downstream effectors activated by EGFR autophosphorylation – STAT3, MAP kinase (MAPK), and AKT. The locations of tumor-associated mutations, all within the tyrosine kinase domain, are shown in red.
Heterozygous mutations were observed in eight of nine patients, all of which were clustered within the tyrosine kinase domain of EGFR (Table 2 and Figure 2). Four tumors had in-frame deletions, removing amino acids 746 through 750 (delE746–A750) in Patient 1, 747 through 751 (delL747–T751insS) in Patient 2, and 747 through 753 (delL747–P753insS) in Patients 3 and 4. The second and third deletions were associated with the insertion of a serine residue, resulting from the generation of a novel codon at the deletion breakpoint. Remarkably, all these deletions overlapped, sharing the deletion of four amino acids (leucine, arginine, glutamic acid, and alanine at codons 747 through 750) within exon 19.
Another three tumors had amino acid substitutions within exon 21: leucine to arginine at codon 858 (L858R) in Patients 5 and 6 and leucine to glutamine at codon 861 (L861Q) in Patient 7. The L861Q mutation is of particular interest, since the same amino acid change in the mouse egfr gene is responsible for the Dark Skin (dsk5) trait, associated with altered EGFR signaling.19 A fourth missense mutation in the tyrosine kinase domain resulted in the substitution of cysteine for glycine at codon 719 within exon 18 (G719C) in Patient 8.
Matched normal tissue was available for Patients 1, 4, 5, and 6 and showed only the wild-type sequence, indicating that the mutations had arisen somatically during tumor formation. By comparison, no mutations were observed in seven patients with non–small-cell lung cancer who had had no response to gefitinib (P<0.001 by a two-sided Fisher’s exact test).
Prevalence of Specific EGFR Mutations in Non–Small-Cell Lung Cancer and Other Types of Cancer
Unlike gliomas, in which rearrangements affecting the EGFR extracellular domain have been extensively studied,15 the frequency of EGFR mutations in non–small-cell lung cancer has not been defined. We therefore sequenced the entire coding region of the gene in tumors from 25 patients with primary non–small-cell lung cancer who were not involved in the gefitinib study, including 15 with bronchoalveolar lung cancer, which has been associated with responsiveness to gefitinib in previous clinical trials.11,12 Heterozygous mutations were detected in two patients with bronchoalveolar cancers. Both had in-frame deletions in the kinase domain that were identical to those found in the patients with a response to gefitinib – namely, delL747–P753insS and delE746–A750 (Table 2). Given the apparent clustering of EGFR mutations, we sequenced exons 19 and 21 in a total of 95 primary tumors and 108 cancer-derived cell lines, representing diverse tumor types (see the Supplementary Appendix). No mutations were detected, suggesting that only a subgroup of cancers, in which EGFR signaling may play a critical role in tumorigenesis, harbor EGFR mutations.
Increase in EGF-Induced Activation and Gefitinib-Induced Inhibition of Mutant EGFR Proteins
Panel A shows the time course of ligand-induced activation of the delL747–P753insS and L858R EGFR mutants, as compared with wild-type EGFR, after the addition of EGF to serum-starved cells. The autophosphorylation of EGFR is used as a marker of receptor activation, with the use of Western blotting with an antibody that specifically recognizes the phosphorylated tyrosine1068 (Y1068) residue of EGFR (left side), and compared with the total concentrations of EGFR expressed in Cos-7 cells as control (right side). Autophosphorylation of EGFR is measured at intervals after the addition of EGF (10 ng per milliliter). Panel B also shows the EGF-induced phosphorylation of wild-type and mutant EGFR. Autoradiographs from three independent experiments were quantified with the use of National Institutes of Health image software; the intensity of EGFR phosphorylation has been adjusted for the total protein expression and is shown as the mean (±SD) percent activation of the receptor. Panel C shows the dose-dependent inhibition of the activation of EGFR by gefitinib. Autophosphorylation of EGFR tyrosine1068 is demonstrated by Western blot analysis of Cos-7 cells expressing wild-type or mutant receptors and stimulated with 100 ng of EGF per milliliter for 30 minutes. Cells were untreated (U) or pretreated for three hours with increasing concentrations of gefitinib (left side). Total amounts of EGFR expressed are shown on the right side (control). Panel D also shows the mean (±SD) inhibition of EGFR by gefitinib. Concentrations of phosphorylated EGFR were adjusted for total protein expression.
To study the functional properties encoded by these mutations, we expressed the receptor with the L747–P753insS deletion and the receptor with the L858R missense mutation in cultured cells. Transient transfection of wild-type and mutant constructs into Cos-7 cells demonstrated equivalent expression levels, indicating that the mutations do not affect the stability of the protein. EGFR activation was quantified by measuring phosphorylation of the tyrosine1068 residue, commonly used as a marker of the autophosphorylation of EGFR.20 In the absence of serum and associated growth factors, neither wild-type nor mutant EGFR demonstrated autophosphorylation (Figure 3A
and Figure 3B). However, the addition of EGF doubled or tripled the activation of both mutant EGFRs, as compared with the activation of the wild-type receptor. Moreover, whereas the activation of normal EGFR was down-regulated after 15 minutes, consistent with the internalization of the receptor, the two mutant receptors demonstrated continued activation for up to three hours (Figure 3A). Similar results were obtained with the use of antibodies to measure the total phosphorylation of EGFR after the addition of EGF (data not shown).
Since seven of the eight EGFR tyrosine kinase mutations reside near the ATP cleft, which is targeted by gefitinib, we assessed whether the mutant proteins have altered sensitivity to the inhibitor. EGF-induced autophosphorylation of EGFR was measured in cells pretreated with various concentrations of gefitinib. Remarkably, both mutant receptors were more sensitive than the wild-type receptor to inhibition by gefitinib. Wild-type EGFR was inhibited by 50 percent at a gefitinib concentration of 0.1 μM and was completely inhibited by a concentration of 2.0 μM, whereas the respective values for the two mutant proteins were 0.015 μM and 0.2 μM (Figure 3C and Figure 3D). This difference in drug sensitivity may be clinically relevant, since pharmacokinetic studies indicate that daily oral administration of 400 to 600 mg of gefitinib results in a mean steady-state trough plasma concentration of 1.1 to 1.4 μM, whereas the currently recommended daily dose of 250 mg leads to a mean trough concentration of 0.4 μM.21