Molecular Testing of Lung Cancers - A Summary of What, Why and How

Feb. 20, 2020

Worldwide, lung cancers represent the single most common cancer type in men and third most common in women with age standardized incidence rates in 2018 as high as 0.078 percent (Hungarian males).1,2 It’s a heterogenous catchall term, with the primary stratification from both histological and behavioral considerations being between small cell lung cancer (SLLC; primarily smoking associated) and non-small-cell lung cancer (NSCLC). NSCLCs are the more common group at roughly 80 percent of presentations, of which most can be immediately further stratified into adenocarcinomas, large cell carcinomas, and squamous cell carcinomas.

Among these three NSCLC subclasses, a relatively small pool of genetic mutations including point mutations, indels, rearrangements and copy number variations are associated or mechanistically contributing (even foundational) to the cancer at high frequency. Where these mutations impact key specific cellular pathways to induce cancer (so called “driver mutations”), there can be opportunities for narrowly targeted, specific drug therapies. Compared to generic broad-spectrum antineoplastic agents, which tend to target all rapidly dividing cell populations with inherent side effects, these targeted therapies can be highly effective and target selective – but only when they are correctly matched to a cancer. Use in the wrong context is ineffectual, economically unwise (expensive therapy without benefit), and worst of all, might displace timely application of a more effective therapy. For these reasons, molecular testing of lung cancers is standard of care and key to improving medical outcomes through appropriate treatment selection.

Such testing begins with collection of test sample, an action which can influence what molecular test modalities are applicable. For primary tumor biopsy samples with histologically identifiable high preponderance (>50 percent) of cancer cells, direct whole genome sequencing by next-generation sequencing (NGS) is the most informative (and the most costly, and time consuming) approach – but with no inherent expectation bias, it can provide information on both known mutations and novel ones of potential relevance. For samples with smaller fraction of cancer cells – down to about 1 percent in practical terms – an alternate approach is direct real-time PCR with allele specific primers for particular known mutations; this is rapid, sensitive, and inexpensive to perform but can only look for what it’s designed to find.

In between these extremes falls an “NGS panel” approach where PCR amplification of key pathway genes where relevant mutations are known, followed by sequencing; while this only examines preselected gene regions, it can find novel mutations. In the case of novel mutations – either here or in the whole genome context – relevance must of course be demonstrated, although in the case of some types of mutations such as gene fusions, significant indels, or nonsense mutations, a high probability of causal relevance to the cancer may be immediately apparent.

Practical sample types

Almost any sample type as available – fresh, frozen, or formalin-fixed paraffin-embedded (FFPE) (best if it’s prepared with downstream molecular testing in mind, meaning not overfixed) – obtained by needle aspirate, bronchial wash or (insert your method of choice) – can be suitable for at least one form of the testing overviewed here, although note the caveat above about minimal tumor cell content for direct sequencing approaches.

Sampling and analysis of metastatic sites can be useful as well, although this assumes that oncogenic mutations developed early in progression and are thus common to all tumor sites; not necessarily true but observational data suggests in this context it’s often a valid assumption. One class of sample which current publications do not suggest are great here is circulating tumor cells (CTCs). While these show promise in some oncology contexts and may be relevant in lung cancer where positive results (tumor cells and associated data are detected), high false negative results have been noted and more direct tissue biopsy is preferable.3

Gene Examples

Before we proceed let’s provide some examples of genes, mutation targets and associated targeted therapies considered in lung cancer cases (see Table 1). It should be stressed this is an incomplete list, highlighting only some common mutations and therapy implications for illustrative purposes; factors out of scope for this brief article come into play in deciding best therapies in actual cases.

Application of data

While incomplete, we can see from Table 1 how molecular knowledge of a lung cancer can be applied to making informed therapy decisions. If a sample is detected as having known activating mutations in EGFR but not having the T790M mutation, then something like Gefitinib might be a good choice (again, also influenced on factors beyond our scope). If, however, the T790M mutation was simultaneously present, we’d already know Gefitinib would not work and Osimertinib would be our best choice. If instead our tumor was detected as having a rearrangement in ROS1 but unmutated in ALK, EGFR or KRAS, then neither Gefitinib nor Osimertinib would be likely of much benefit and we’d want to use something from the Crizotinib, Ceritinib, Lorlatinib, Entrectinib list.

All three of these cases could occur in something correctly identified as an NSCLC adenocarcinoma by classical pathology examination. Add in the possibility BRAF, MET, or HER-2 mutations could also be involved, each with their own best drug strategies, and the impracticality of trying to select best matched therapy by some trial and error process with its inherent cost, time and attendant disease progression before the correct match is made provide the compelling argument for molecular testing in this context.

Possible future evolution?

The utility of molecular testing in this context has been driven by the relatively small number of specific known mutations, and the pairing of these to tailored therapeutic agents. In turn, small numbers of defined targets lend themselves well to rapid, relatively low cost and complexity test modalities like allele specific PCR. As the number of gene targets and the number of known mutations per target with significance for therapeutic decision-making both increase these single target tests will become less practical. A simultaneous ongoing decrease in cost, size, and complexity of use for NGS methods is occurring.

The conclusion would seem that NGS panels combining the ability to be amplification based (that is, requiring low content both numerically and in percentage) of cancer cells in sample, plus the ability to identify both known and novel alterations in key genes of relevance, will be the method of choice for balancing cost versus completeness of data in this application. For cases where this approach does not prove fruitful, a costlier direct sequencing approach utilizing a sample enriched in tumor cells would be a second line.

Information on novel driver mutations uncovered either way may still direct therapy choices based on knowledge of signalling pathways. Consider for example Table 1, and the hypothetical detection of a suspected causal mutation in BRAF causing not just pathway activation but leaving the mutant BRAF unresponsive to agents directly binding the altered protein. Knowing BRAF signals on through MEK would suggest that Trametinib to block or reduce MEK signalling would be a likely successful intervention.

Finally, these methods – regardless of exact sample and technical approach used – will all improve over time as more and more less common significant mutations are detected, cataloged and their significance with regard to therapy choices and outcomes are known. Less Variants of Unknown Significance – VUS – mean faster, better interpretation of patient data.


  1. Shim HS, Choi YL, Kim L, et al. Molecular Testing of Lung Cancers. J Pathol Transl Med. 2017;51(3):242–254. doi:10.4132/jptm.2017.04.10
  3. Oxnard GR, Thress KS, Alden RS, et al. Association between plasma genotyping and outcomes of treatment with osimertinib (AZD9291) in advanced non-small-cell lung cancer. J Clin Oncol. 2016;34:3375–82