Molecular Testing in Lung Cancer

Protein Expression

Immunohistochemistry (IHC) is most commonly used for protein expression assessment in the clinical context. IHC is a process that is easily performed by investigators because of the short time needed to complete testing and low cost, and due to its applicability to formalin-fixed paraffin-embedded (FFPE) rather than fresh frozen tissue. In addition, IHC may help investigators assess protein expression at the cellular level, thus allowing them to evaluate cellular localization (e.g., membranous, nuclear, or cytoplasmic), topography (e.g., tumor or stromal cells), and heterogeneity of expression and is also applicable to very small specimens, including cytologic samples. However, many preanalytic and analytic factors may influence IHC reactions, resulting in potentially variable staining that may affect the interpretation of the results. Therefore optimizing and standardizing the protocols and conditions are required for each marker tested. Interpreting the results is also observer dependent and may vary between observers, thus requiring standardization of protocols and conditions. Lastly, the scores for defining positive or negative IHC results for their prognostic or predictive value of specific biomarkers need to be well defined and validated in multiple independent cohorts/institutions and clinical trial samples. However, despite the mentioned limitations, IHC is considered to be an easy and inexpensive clinically applicable assay, which already is available in most pathology departments.

Gene Mutations

The technologies available for mutation analyses are associated with different sensitivity. Analytic sensitivity is defined as the lowest percentage of tumor cells or tumor cell DNA concentration in which a mutation is detectable with confidence within replicate assays. The standard method for detecting mutations has been direct sequencing by the Sanger method. This method allows the detection of a minimum of 25% of mutated allele frequency from tissue containing 50% cancer cells cellularity, if the mutation is heterozygous and in the absence of gene amplification. However, mutated driver oncogenes, such as EGFR and KRAS , are commonly amplified implying that a lower number of tumor cellularity in the sample may yield 25% mutated alleles. Bidirectional sequencing and confirmation by repeat sequencing on independently amplified polymerase chain reaction (PCR) products should be performed especially on FFPE tissue (see later). The impact of the lower sensitivity of the Sanger sequencing method resulting in a substantial false-negative response rate (approximately 30%) in the detection of EGFR mutations has been documented.

To overcome the generally lower sensitivity of the Sanger sequencing method, various technologies are available that allow mutation detection at significantly higher sensitivity with a tumor cellularity of as low as 1% to 5% or mutated allele. These more sensitive technologies involve a mutated allele-enriching strategy, including the peptide nucleic acid/locked nucleic acid amplification, the coamplification at lower denaturation temperature-PCR, or the enzymatic digestion of wild-type sequences. The US Food and Drug Administration (FDA) has approved two assays for the detection of EGFR mutation analyses in advanced NSCLC: the Scorpion-amplification refractory mutation system (ARMS) and cobas technologies (Roche Molecular Diagnostics, Pleasanton, CA, USA). Several other methods may be used to detect EGFR mutations ( Table 18.1 ). The Scorpion-ARMS technology is available as a commercial kit that allows investigators to test for 29 EGFR mutations and has a sensitivity of at least 5%. The cobas EGFR mutation test is a reverse transcription-PCR-based (RT-PCR) test for the qualitative detection of exon 19 deletion and exon 21 L858R mutations of EGFR in DNA extracted from FFPE tissue and was used by the investigators in the European Randomized Trial of Tarceva versus Chemotherapy (EURTAC) and LUX-Lung 3 trials.

Several other available assays are based on different technologies. Their sensitivities vary, and only certain techniques have the ability to detect new mutations and/or insertions and deletions. In contrast to the Sanger method that lacks sensitivity for samples with fewer tumor cells, the more sensitive methods might give rise to false-positive results and a lower specificity. Therefore it is crucial that the appropriate positive and negative controls always be included in the assay. Of note, the Sanger method detects any mutation including previously unidentified mutations in the sequenced exons, but other assays are designed for specific mutation testing, as in the case of the digital droplet PCR that has a very high sensitivity. Another option could be a two-step procedure, starting with a highly sensitive detection of the presence of a mutation and the subsequent characterization of the mutation. When finding a mutation that has not or has rarely been reported, the results should not be considered as errors until the replicate test confirms or denies it. However, testing all mutations can require more time and might not be suitable for the clinical situation when treatment must be initiated without delay. In such cases, another approach can be used that tests the most common mutations first and then completes the screening of less frequent mutation.

The recent and rapid development of NGS accomplishes massive parallel gene mutation analyses and discovery and requires a small amount of tissue, preferably fresh frozen. This technology uses miniaturized and parallelized platforms for sequencing of millions of short nucleotides (50–400 bases). The different platforms all have in common a technical paradigm of massively parallel sequencing via clonally amplified, spatially separated DNA templates or single DNA molecules in a flow cell. Currently, NGS is used for research purposes rather than to test specific biomarkers. However, with the expansion of the use of molecular biomarkers and the rapidly growing targeted therapies available, using NGS to have a full molecular profiling of the tumor or at least multiplex mutation testing of a panel of biomarkers of interest might become preferable in the future in order to spare time and tissue. In addition, in a recent study, even for individual gene mutation analyses, NGS has been shown to have better sensitivity, as it detected all relevant EGFR mutations for prediction of response to EGFR TKIs in 24 tumors, compared with the Sanger method and pyro-sequencing, which resulted in four and two false-negative results, respectively.

Nevertheless, the clinician’s chosen approach needs to take into account the tumor content available, the possibility of detecting all mutations, the timing required for testing, and the urgency to start the patient’s treatment. Usually a laboratory investigator will decide to choose a specific method based on equipment availability and cost and will conduct an assay optimization and standardization exercise and test for assay sensitivity and specificity.

Changes in Gene Structure and Copy Number

Fluorescence in situ hybridization (FISH) is the standard method for assessing changes in gene structure and copy number. Similar to IHC, FISH can be performed on FFPE tissue but requires standardized protocols ( Fig. 18.1 ). Interpreting and reading FISH specimens is observer dependent and requires a dark room and a specific microscope, and the reader needs specific training and expertise in order to achieve reproducible results. Of note is that the tissue structure is not well visualized, and this factor necessitates the preselection of areas to assess in order to discriminate tumor cells from nonmalignant cells. Furthermore, the fluorescence probes are unstable and fade over a short period, which can limit the possibility of revisiting the specimens. Therefore imaging the specimens in a short time frame is important. By using probes against multiple targets labeled with different fluorescent dyes, it is possible to assess multiple markers on the same sections. In addition to detecting the gene copy number, FISH is used to assess structural changes including fusions between genes. The example of ALK fusion is detailed later in this chapter.

Several alternative techniques to FISH have been developed, including chromogenic in situ hybridization and silver in situ hybridization. These techniques are used primarily in research and give results comparable with FISH; however, they are not commonly used in routine clinical testing in lung cancer. Silver in situ hybridization is now FDA approved for human epidermal growth factor receptor-2 (HER2) determination in breast cancer and is widely used for that indication. Multicolor assays are in development. Finally, gene copy number can be assessed by array-comparative genomic hybridization technique. However, comparative genomic hybridization is used mainly to probe for a large number of markers in exploratory discovery research, rather than to assess specific markers for clinical applications.

Another assay that is used for assessing gene copy number is PCR, which is a very sensitive method that requires specific primers and probes as are used in gene rearrangement in the ALK gene. More recently, computational algorithms have also been developed to derive gene copy number estimate using high coverage NGS data.

Preanalytic Factors

Based on expert consensus opinion mutation testing should be performed on FFPE, frozen, or alcohol-fixed tissue specimens. The main advantage of using FFPE tissue is that it is the most commonly used method to process tissue for routine histology. FFPE also allows for a better evaluation of tumor cell content, which is also possible with fresh tissue, but is less convenient and requires cutting and staining of frozen sections adjacent to the section used for DNA extraction. The results of mutation testing with alcohol-fixed tissue specimens are also excellent. This fixation method is often used on cytologic specimens, which are then suitable for mutation testing. DNA isolated from fresh or frozen tissues may yield fragments of 1000 base pairs (bp) and longer. Fixation of tissue in formalin induces crosslinks between DNA, RNA and proteins, and DNA fragmentation that results in DNA fragments of 300 bp or less. Formalin fixation also creates random nucleotide base exchange, resulting in false-positive results. This type of problem mostly occurs with low DNA yield and/or with ultrasensitive assays. Tissue treated with acidic or heavy-metal fixatives, including lead, cobalt, chromium, silver, mercury, and sometimes even uranium and decalcifying solutions, may reduce the success rate of mutation testing and should be avoided when alternate FFPE specimens are available. In molecular biology, heavy-metal fixatives inhibit the DNA polymerases used in PCR testing. Acidic solutions, including the decalcifying solutions that are used to process samples obtained from bone metastasis, can induce a high rate of DNA fragmentation. For these types of specimens that are obtained specifically for molecular testing, nonacidic methods of decalcification, such as nonacidic chelating decalcifying solutions, should be used in the sample-processing step.

IHC and FISH should be performed on FFPE tissue, ideally on cut sections that have been stored for fewer than 6 weeks, to avoid the oxidation process that occurs over time. Whatever the method used, standardizing the fixation procedure and storage conditions is required. The fixation should be performed within hours after the sample has been obtained. Fixation duration should be controlled and not exceed 12 hours for small biopsy specimens and 18 hours to 24 hours for resected specimens.

Data have shown that molecular testing (i.e., mutation testing or FISH) can be performed on liquid biopsies (circulating DNA extracted from plasma or circulating tumor cells). These assays are still experimental and need to be reproduced and standardized before any clinical application.

Sample Processing and Analysis

Tumor tissue is heterogeneously composed of a mixture of tumor cells and host cells. Host cells include inflammatory cells and vascular endothelial and stromal fibroblasts and their abundance is highly variable but may have a substantial impact on the sensitivity of mutation testing ( Fig. 18.2 ). The proportion of tumor cells in the specimen, as compared with normal tissue and inflammatory cells, may affect the result of the mutation analyses, mainly with less sensitive methods for mutation detection, as a low copy number of the DNA template generates artifacts in the results. To avoid false-negative results, specimens with a minimum proportion of tumor cells ideally should be selected for mutation analyses.

Routinely, DNA is extracted from five to ten scratched unstained sections, depending on the size of tissue sample. However, in some cases, a very low amount of DNA is obtained. To overcome the low amount of DNA extracted from small tissue specimens, different techniques can be used. Whole-genome amplification has been developed and is used in research; however, this technique has not been implemented in clinical testing yet, as it may introduce bias. Performing the assay in duplicate, and ideally in triplicate, may ensure the accurate interpretation of the results, but these methods may not be practical in a clinical laboratory because of the lack of tissue and the time and labor needed to duplicate (or triplicate) testing. Different methods of tissue enrichment can be used for tissue with heterogeneity in areas with tumor cells, including gross macrodissection, coring areas with tumor cells out of an FFPE block, microdissection from the glass, flow sorting, or laser capture microdissection (LCM). Macrodissection is used for clinical testing, but LCM is not routinely used because it is labor intensive and because the effects of a laser on mutation testing are unknown and must be evaluated. In addition, even if LCM produces a very pure sample of tumor cells, it also provides a very low DNA yield.

The advantage of IHC and FISH is that the evaluation of the protein expression level or genomic aberration can be analyzed more specifically on individual tumor cells. In IHC and FISH, cells are analyzed individually. Therefore the tumor cell’s content is less crucial than for mutation testing. However, focusing the analysis on tumor-rich areas is important in FISH. Thus a corresponding hematoxylin and eosin (H&E)-stained section should be used to select areas for analysis. When using IHC, a larger sample size provides a better evaluation of the tumor heterogeneity and percentage of cells expressing the biomarker. However, obtaining larger samples is not easily controlled, as the sample size depends on the type of specimen that can be obtained from the tumor.