Lung cancer is the most frequent invasive malignancy and the common cause of cancer death worldwide with an estimated 1.60 million new cases and 1.37 million deaths in 2008.¹ Non small-cell lung cancer (NSCLC), the broad category that accounts for approximately 87% of all patients with lung cancer, usually presents at an advanced stage, where the treatment is essentially palliative. The survival improvement for unselected patients with metastatic NSCLC has been modest, with a large surveillance, epidemiology, and end results (SEER) study, from the periods 1990–1993 to 2002–2005, showing increased survivals at 1 and 2 years of 13.2% to 19.4% and 4.5% to 7.8%, respectively.² More recently, however, there has been a significant improvement in the understanding of the biology of lung cancer, with the discovery of new targets and development of several drugs with novel mechanisms of action. This chapter reviews the current knowledge of cancer genomics, the use of molecular markers, and results from clinical trials that are changing the therapeutic landscape of NSCLC.

Genomics is defined as the study of the entire set of genetic information of a person, encoded in the structure of deoxyribonucleic acid (DNA). Cancer genomics is the study of DNA-associated abnormalities associated with the development of cancer. The DNA in normal cells is constantly damaged by environmental and normal cellular processes. Although the majority of the damage is repaired, a small fraction is converted into fixed mutations. Mutations may be broadly subdivided into germline and somatic. Whereas germline mutations are present in the fertilized egg, inherited from the parents and therefore present in all somatic cells, the somatic mutations are acquired after conception. Although somatic mutations are distributed throughout the genome, a subset of them occurs in key genes. These “driver mutations” are implicated in oncogenesis by allowing the malignant clone to expand more than the normal cells. In contrast, the “passenger mutations” are carried along during clonal expansion, do not contribute to cancer development, and are not associated with growth advantage.^3,4

The majority of malignancies is sporadic and occurs as a consequence of the accumulation of genomic alterations that lead to dysregulation of protein-encoding genes. As normal cells evolve to a neoplastic state, they acquire several essential complementary capabilities including sustained proliferative signaling, resistance to apoptosis, evasion of growth suppressors, and induction of angiogenesis, invasion, and metastasis.⁵ Cancer cells, however, often are physiologically dependent or “addicted to” to the continued activity of specific oncogenes, and this dependency has been explored for drug discovery.⁶ The pivotal study of chronic myeloid leukemia that showed excellent response rates (RRs) and good tolerability for imatinib in patients who progressed after interferon therapy, validated the concept of targeting driver mutations and essentially started the era of targeted therapy in cancer treatment.⁷

Recent advances in DNA sequencing have permitted significant advances in cancer genomics. The major breakthrough came with the development of next-generation sequencing, also known as second-generation sequencing, which increased the efficiency for detecting the main types of somatic cancer genome alterations including nucleotide substitutions, small insertions and deletions (indels), copy number gains or losses, chromosomal rearrangements, and microbe infections.⁸ Platforms for massive parallel DNA sequence reads, such as the 454 Genome Sequencer FLX, Illumina Genome Analyzer and Applied Biosystems SOLiD, combined with powerful computational methods derived from significant progress in bioinformatics, made possible the sequencing of exome, transcriptosome, and epigenome.⁹

Molecular markers, also known as biomarkers, are defined as a characteristic that can be objectively measured and evaluated as an indicator of normal biologic processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.¹⁰ Prognostic markers provide information about the natural history of the disease by describing outcomes independently of therapeutic intervention. Their main applications are to evaluate which patients should be treated. This information is particularly useful in the adjuvant setting.^11,12 Predictive markers in contrast, provide information on the probability of benefit or toxicity from a specific therapy, and therefore guide the choice of therapy. Since cancer therapy typically benefits only a small fraction of patients, the identification of those most likely to respond to each specific treatment is one of the top priorities in new drug development.

NSCLC is broadly subdivided into three main histologies, namely, adenocarcinomas, squamous cell carcinomas, and large cell carcinomas. Traditionally, patients with advanced stage NSCLC, good performance status, and no specific contraindications have been treated as a single entity with platinum-based doublets. With this empirical approach, modern third-generation regimens have shown an RR of 19%, median time to progression (TTP) 3.6 months, median overall survival (OS) 7.9 months, and 1-year OS of 33%.¹³ These regimens are associated with significant toxicity and reached a plateau in efficacy. Therefore, treatment has been chosen based on personal preferences and toxicity profile.

The advances in DNA sequencing made a significant impact on the understanding and treatment of NSCLC through the identification of multiple driver mutations. These driver mutations occur mostly in the genes encoding signaling proteins, particularly the tyrosine receptor kinases (TKIs).¹⁴ It soon became clear that a substantial percentage of lung cancer patients harbor mutant signaling proteins, encoded by “driver mutations,” which could be used as targets for drug therapy. Most of the initial mutations were described predominantly or exclusively in patients with adenocarcinoma histology, including those in the epidermal growth factor receptor (EGFR), KRAS, BRAF, PIK3CA, and other genes, as well as the anaplastic kinase lymphoma (ALK) fusion gene (Fig. 169-1). The most common abnormalities in patients with squamous cell carcinomas are amplifications of fibroblast growth factor receptor 1 (FGFR1) and SOX2 and mutations of DDR2 and PIK3CA, with the amplifications each occurring in 20% of cases in an almost mutually exclusive pattern, and the mutations occurring in less than 5% of patients (Fig. 169-2).^15–17

The EGFR is a transmembrane receptor tyrosine kinase that belongs to a family of four related proteins, also including ERBB2 (HER2), ERBB3 (HER3), and ERBB4 (HER4). Each receptor consists of three domains including extracellular ligand-binding, transmembrane domain, and intracellular. Upon binding to one of the ligands, the inactive EGFR monomers undergo conformal changes that lead to homodimerization or heterodimerization with the other members of the family, most commonly HER2, which lacks a specific ligand, followed by autophosphorylation of tyrosine residues within the cytoplasmic tail, which work as docking sites for several proteins leading to activation of downstream intracellular signaling pathways.¹⁸ EGFR may be targeted by two broad mechanisms, monoclonal antibodies, and small-molecule tyrosine kinase inhibitors (TKIs). The promising results from randomized phase II studies comparing chemotherapy alone or in combination with the anti-EGFR monoclonal antibody cetuximab in previously untreated patients¹⁹ led to the two large randomized phase III trials evaluating the role of cetuximab in this setting. The first study, the FLEX trial,²⁰ randomized 1125 chemotherapy-naïve patients with EGFR-expressing tumors to cisplatin plus vinorelbine alone or in combination with cetuximab. The cetuximab arm was associated with increased RRs (36% vs. 29%, p = 0.01), identical progression-free survival (PFS) of 4.8 months, and a modest improvement in median OS (11.3 vs. 10.1 months, p = 0.04). In the BMS099 trial,²¹ 676 patients were randomized to carboplatin, a taxane (either paclitaxel or docetaxel) with or without cetuximab. Compared to chemotherapy alone, the cetuximab arm improved the RRs (25.7% vs. 17.2%, p = 0.007) but did not lead to a significant improvement in PFS (4.4 vs. 4.2 months, p = 0.23) or median OS (9.6 vs. 8.3 months, p = 0.16). In the correlative study evaluating KRAS mutations, EGFR mutations, EGFR protein expression, and EGFR copy number, there were no predictive markers for response to cetuximab.²² In contrast to monoclonal antibodies, the discovery of predictive molecular markers for response to EGFR TKIs in NSCLC resulted in a marked improvement in the outcomes from the initial studies in unselected populations to the most recent trials restricted to patients harboring the specific driver mutation.

The initial experience with EGFR TKIs showed good tolerance to gefitinib, with the most common side effects including rash, diarrhea, nausea, and anorexia.^23–26 Among the combined 100 heavily pretreated patients with NSCLC enrolled into the phase I studies, 10 had partial response and an additional 24 achieved stable disease, increasing interest for gefitinib in this setting and leading to the development of phase II studies. The Iressa Drug Evaluation in Advanced Lung Cancer (IDEAL) I and II studies were conducted simultaneously worldwide and in the United States, respectively.^27,28 Both studies enrolled previously treated patients and had two arms consisting of gefitinib at doses of 250 or 500 mg daily. RRs were observed in approximately 10% of Caucasian patients and 18% of Japanese patients with no significant differences according to the treatment doses. The treatment was overall well tolerated, with predicted toxicities similar to the phase I trials. Attempts to increase the RR by combining either gefitinib^29,30 or erlotinib³¹ to chemotherapy did not improve outcomes compared to chemotherapy alone. RRs to erlotinib or gefitinib compared to placebo in two randomized clinical trials enrolling previously treated patients with NSCLC were 8.9% and 8%, respectively, reflecting a modest efficacy for EGFR TKIs in unselected populations.^32,33 Nevertheless, subset analyses from multiple studies identified subgroups of patients most likely to respond to EGFR TKIs based on clinical characteristics including good performance status, women, never smokers, adenocarcinoma histology, and Asians.^27,32,33 The first breakthrough toward personalized medicine in patients with NSCLC occurred in 2004, when two independent groups described the association between activating EGFR tyrosine kinase mutations and response to EGFR TKIs.^34,35 The majority of EGFR-activating mutations are a point deletion in exon 21 with the substitution of arginine for leucine (L858R) and a small in-frame deletion of four amino acids around the LREA motif centered at codons 746 to 750 in exon 19. These mutations were present more commonly in patients with the same previously described clinical characteristics that predicted response to EGFR TKIs, explained the earlier observations, and provided the first genotype-driven therapy in NSCLC. In a pooled analysis of 268 EGFR-mutant patients treated with TKIs, 210 patients (78%) responded, compared to 68 out of 659 (10%) patients with wild-type EGFR.³⁶ Several prospective phase II studies evaluated the use of EGFR TKIs in patients with activating mutations, with RRs ranging from 55% to 84%, PFS from 8.9 to 12.9 months, and 1-year survival from 73% to 83%.^37–42 Since the RRs and PFS for TKIs in patients with activating EGFR mutations were clearly superior to historical studies using chemotherapy, the next step was to perform a direct comparison in the first-line setting.