New insights into the molecular profile of lung adenocarcinoma and implications for therapy

Abstract

Lung cancer is a molecularly heterogeneous disease. The advent of next-generation sequencing techniques has significantly advanced our understanding of the complex molecular underpinnings of lung cancer. Furthermore, the development of targeted therapies has significantly altered the landscape of lung cancer therapy over the past decade. There is hence an increasing interest in developing a classification system that guides clinical management and also incorporates relevant genomic information. Here, we highlight the molecular features of lung adenocarcinoma as highlighted by several independent groups, and more recently The Cancer Genome Atlas and discuss their potential clinical significance.

Keywords:

• adenocarcinoma
• genome sequencing
• lung cancer
• molecular classification
• NSCLC

Lung cancer is the leading cause of cancer-related mortality in the world [1]. Non-small cell lung cancer (NSCLC) is the most common histological subtype of lung cancer [2]. Broadly, the histological subtypes of NSCLC include adenocarcinoma, squamous cell carcinoma and large cell carcinoma. The genomic landscape of a typical lung cancer caused by tobacco smoking is quite complex [3,4].

Summary of findings from genomic studies on lung adenocarcinoma

Comprehensive genomic studies involving lung adenocarcinoma show several striking features [4–8]. The mutational burden in tobacco-related lung adenocarcinoma is significantly higher than many common cancers (median of ∼10.5 mutations/Mb of the genome) [5]. In contrast, lung adenocarcinoma from never smokers is associated with a very low mutational burden (median of 0.6 mutations/Mb of the genome). Smoking history also correlates with the prevalence of nucleotide transversions within tumors [4]. Transversion-high (smokers) and transversion-low (never-smokers) tumors are associated with different gene mutations, suggesting that distinct mechanisms drive these tumors. Furthermore, significant intra- and inter-tumoral heterogeneity has been observed in lung adenocarcinomas through next-generation sequencing. The evolutionary history of lung cancer clones constructed through multi-region sequencing suggests that driver alterations in lung cancer occur early in cancer development [7,8]. Although RAS/RAF/AKT pathways are the most commonly involved (76% cases) in the molecular pathogenesis of lung adenocarcinoma, alterations in the TP53 and cell cycle regulator pathways are observed in approximately two-third of cases [4].

Molecular classification of lung adenocarcinoma

Based on tumor expression profiling, Hayes et al. [9] demonstrated that lung adenocarcinomas cluster into three groups – bronchioid, magnoid and squamoid. The Cancer Genome Atlas (TCGA) investigators have validated these findings and identified correlation between molecular subtypes of adenocarcinoma with clinical features [4]. For instance, bronchioid tumors harbor EGFR mutations and tyrosine kinase (TK) fusions more frequently than other subtypes, are seen more frequently in never-smokers and associated with better prognosis. When applied to tumor samples that were a part of the JBR.10 trial, which demonstrated benefit with adjuvant cisplatin and vinorelbine in early-stage NSCLC, this classification system demonstrated that only magnoid subtype tumors showed a significant disease-free survival benefit with adjuvant therapy [10]. It is speculated that magnoid tumors, which show a high degree of chromosomal instability compared with other subtypes, harbor DNA repair pathway defects that render them more sensitive to platinum therapy. TCGA investigators have proposed to change the nomenclature from bronchioid, magnoid and squamoid subtypes to terminal respiratory unit, proximal proliferative and proximal inflammatory subtypes respectively [4]. Integrative clustering using copy-number, mutation, DNA methylation and mRNA expression data suggested the presence of six clusters that were associated with distinct molecular profiles.

Implications for therapy

Targeted therapies directed against lung cancers harboring mutations in EGFR TK and rearrangements involving anaplastic lymphoma kinase (ALK) have significantly improved outcomes in a subset of patients [11–14]. These alterations are particularly prevalent among never- or light-smokers [11,15,16]. Unfortunately, EGFR mutations are present in the tumor cells in only about 16% of patients with advanced stage NSCLC [17]. Similarly, about 5% of patients with lung adenocarcinoma have ALK gene rearrangements in the tumor cells. A median progression-free survival of 7.7 and 11 months has been associated with the use of crizotinib in previously treated and chemotherapy naïve patients [18–20]. Ceritinib has recently been approved for use in ALK rearranged NSCLC [21]. A number of novel ALK inhibitors are under development. Similarly, fusions involving RET and ROS1 oncogenes have also been reported in 1–2% of patients with NSCLC, and partial responses with crizotinib and cabozantinib have been reported in these patients, respectively [22]. Shaw et al. [23] recently reported an objective response rate of 72% with a median PFS of 19.2 months with crizotinib in ROS1 rearrangement-positive NSCLCs.

An issue of considerable clinical importance with the use of targeted therapies is the eventual emergence of resistance to these drugs. Mechanisms of resistance include the development of alterations that affect drug binding through steric hindrance or loss of affinity, such as the EGFR gatekeeper T790M or ALK L1196M mutations [24]. Few patients with resistance to ALK inhibition demonstrate copy-number gains in rearranged ALK. Activation of bypass signaling pathways and phenotypic transformation of tumors, including epithelial-mesenchymal transformation or differentiation to a small cell histology, represent other mechanisms of acquiring resistance to targeted therapies. TK inhibitors such as AZD9291 and CO1686 have recently demonstrated significant activity in NSCLCs harboring the EGFR T790M mutation [25,26]. Similarly, encouraging response rates with several novel agents have been reported in crizotinib-resistant ALK rearranged tumors [24,27]. These are currently being actively investigated in clinical trials [28–31].

The introduction of next-generation sequencing technologies has enabled an unbiased characterization of cancer genomes on a massive scale. This has led to the discovery of several potentially targetable genetic alterations in lung cancer over the past few years, giving rise to newfound optimism and promise in the field of cancer therapeutics. Such efforts have led to the identification of new driver alterations in ‘oncogene negative’ lung adenocarcinomas – a subset of tumors that lack known somatic driver alterations capable of activating the RTK/RAS/RAF pathway [4]. Among these, RIT1 mutations were reported in 2% of adenocarcinomas by TCGA. Berger et al. [32] demonstrated a potential role for combined MEK and PI3K inhibition in these tumors. Similarly, Vaishnavi et al. [33] reported oncogenic NTRK1 gene fusions in 3% of oncogene-negative tumors. Cells expressing the fusion protein were sensitive to CEP-701 (lestaurtinib) and crizotinib. Comprehensive genomic analysis by TCGA also suggests that distinct mechanisms underlie oncogene activation. For instance, MET activation through exon 14 skipping as a consequence of splice site alterations was reported in 10 sequenced adenocarcinoma samples [4]. Overall, these findings emphasize the utility of next-generation sequencing technologies in unraveling the complex molecular mechanisms that drive oncogenesis, and identifying targetable alterations. Such characterization of molecular alterations in tumors through ‘multiplexed assays’ can potentially aid in the administration of genotype-matched therapies to patients harboring specific mutations – an approach that was recently demonstrated by the Lung Cancer Mutation Consortium [34]. Tumors from 1007 patients were tested for at least 1 and 733 patients for 10 genetic alterations as a part of this initiative. These results were used to select a targeted therapy or trial in 275 (28%) of 1007 patients. The median survival was 3.5 years among 260 patients with an oncogenic driver who received genotype-directed therapy, compared with 2.4 years in patients with oncogenic drivers who did not receive such therapy. Next-generation sequencing also carries the potential to determine occult biomarkers that underlie drug sensitivity. Whole genome sequencing of the tumor of a patient with metastatic bladder carcinoma who achieved a durable remission with everolimus aided in the identification of a loss-of-function TSC1 mutation that correlated with drug sensitivity [35]. Similarly, Lovly et al. [36] recently identified a therapeutic synergism between ALK and IGF 1 receptor (IGF-1R) inhibitors through whole genome sequencing of an ALK fusion-positive tumor that had an exceptional response to IGF-1R-specific antibody.

Future directions

It is critical to identify rare variants that have not so far been identified. The ongoing ALCHEMIST study will screen 8000 patients with lung adenocarcinoma for targetable alterations in resected early stage NSCLC. There are plans to study comprehensively the genomic alterations in this trial to identify rare variants. In addition, the contribution made by clonal evolution in the process of metastases and treatment resistance should be further studied using multi-region and multi-site sampling. It is likely that clinical trials in the future will incorporate genomic studies, to understand the molecular mechanisms underlying exceptional responses to novel therapies.

Financial & competing interests disclosure

R Govindan has served as a consultant for Boehringer Ingelheim, GlaxoSmithKline, Pfizer, Merck, Covidien, Bristol Myers-Squibb, Genentech (Roche), Mallinckrodt Pharmaceuticals. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.