Molecular profiling of lung cancer specimens and liquid biopsies using MALDI-TOF mass spectrometry

Lung cancer is the leading cause of cancer death worldwide. Non-small cell lung cancers (NSCLCs), primarily adenocarcinoma (ADC) and squamous cell carcinoma (SCC), account for approximately 80% of lung cancer cases [1].

With the introduction of the Epidermal Growth Factor Receptor/Tyrosine Kinase inhibitors (EGFR-TKIs), which target cancer cells harboring activating EGFR mutations, the detection of somatic mutations became relevant to treatment choices for lung ADC [2]. Erlotinib, gefitinib, and afatinib are used to target EGFR-activating mutations. More recently, the new drug osimertinib was introduced. This molecule can inhibit EGFR kinase activity in the presence of the EGFR T790 M mutation, which confers resistance to the other inhibitors [3‐6]. Yet another drug, crizotinib, inhibits ALK, ROS1, and MET when their kinase activities are aberrantly activated [7‐10].

Ongoing clinical trials are investigating emerging agents capable of avoiding acquired tumor resistance to the common TKIs, or of targeting other activated proteins, such as PI3K, AKT1, ERBB2, MEK1, and DDR2 [10, 11].

Mutations in KRAS (found in 25–40% of ADC) are a negative prognostic biomarker for NSCLC, since no drugs have been developed to inhibit the mutant protein. Alternative strategies, such as inhibition of MEK, have been suggested as treatment for patients with KRAS-mutated cancers [12].

The most frequent activating mutations in lung ADC, other than KRAS, involve EGFR (15%), whereas BRAF, ERBB2, and MEK1 are mutated in less than 2% of cases. PIK3CA mutations are present in approximately 1–3% of NSCLCs, and are more common in SCCs (15%). DDR2 mutations are present in 2% of SCCs. ALK and ROS1 translocations and MET amplifications are typical of ADCs, representing 5%, 4%, and 2% of cases, respectively. AKT1 mutations are found in 1% of lung cancers, more frequently in SCCs [10, 11].

A list of druggable molecular markers and pathways in lung cancer is provided in Fig. 1.

Based on the growing knowledge of inhibitors that target abnormally activated kinases and the resultant clinical inclusion of new drugs, the optimal choice of treatment of NSCLC patients relies critically on screening the tumor-related genetic alterations.

Different issues should be taken into account for the molecular characterization of NSCLC. First, NSCLC are heterogeneous and cells harboring a specific mutation may represent a minor clone in a mixture of neoplastic cells, as well as of non-neoplastic, stromal and inflammatory cells [13]. In addition, the availability and/or the quality of the specimens suitable for molecular evaluations could be scare, since formalin-fixed paraffin-embedded (FFPE) samples from small biopsies or cytology specimens could represent the only available material. [14]. Moreover, in recent years, non-invasive approaches (collectively called “liquid biopsy”) have been developed to identify the molecular profile of tumor circulating cells (TCCs) or circulating tumor DNA (ctDNA) [15‐18] and therefore very sensitive detection methods are required. For all of these reasons, it is critical to use reliable and sensitive diagnostic methods capable of simultaneously detecting a wide range of mutations also in poor quality samples.

Different molecular approaches, such as Sanger sequencing, real-time PCR, pyrosequencing, MALDI-TOF mass spectrometry (MS) and next-generation sequencing (NGS) are currently available [19‐22]. Among them, Sanger sequencing is the less sensitive (at least 20% of mutant alleles); MS is considered a robust approach for the genotyping of known mutations, able to combine the advantages of multiplexing, high sensitivity, and specificity with rapid turnaround, easy sample handling, and cost-effectiveness [19, 20]. Finally, although NGS is a very robust approach with the highest sensitivity [23], it is less affordable than the other approaches and poses several additional challenges, including validation and data handling for diagnostic purposes. Besides, to preserve the cost per test and avoid wasting resources, a consistent number of cases should be simultaneously analyzed.

Using the MS genotyping approach, we tested a cohort of 92 NSCLCs, investigating a wide spectrum of actionable mutations currently targeted by specific therapies or for which clinical trials are ongoing. Our aim was to verify the performance and sensitivity of the method using low levels of tumor DNA. Finally, we evaluated the performance of MS on plasma DNA from 13 lung cancer patients with EGFR- or KRAS-mutated tumors.

A large number of proteins activated in cancer and involved in the intracellular signaling pathways have been investigated as approved or potential targets for biological inhibitors. These discoveries led to a significant increase in molecular testing on tumor samples. Consequently, it becomes crucial to select a robust diagnostic method capable of identifying a wide spectrum of mutations in low-quality samples and in a cost-effective manner. We investigated whether MS would be reliable for this purpose, considering several challenges of tumor genotyping: the frequent paucity of available biological material, the low yield of DNA from FFPE specimens or liquid biopsy, and tumor heterogeneity, which is associated with variable frequencies of somatic mosaicism.

We found that mutations at 10% frequency could always be detected using our multiplexed genotyping panel, which covers 158 mutations in six genes. However, we confirmed that the sensitivity of MS depended on the specific assay, and genetic alterations could also be detected when the mutated allele frequency is very low, e.g., 5% or 2.5% [19, 20]. On this basis, assuming that mutations are heterozygous, a 10% mutated allele frequency corresponds to 20% of cells carrying the mutation [27]. In turn, this implies that a minimum of 20% of tumor cells in a tissue specimen should be required to detect mutations at 10%. However, considering the tumor heterogeneity, it is also possible that a target mutation could be present only in a subset of cancer cells, and thus identifiable by MS, depending on MS performance for each mutation and the DNA quality.

Another crucial issue is the minimum amount of DNA needed for analysis. Frequently, the available amount of DNA is limited by the small sizes of biopsies or cytology specimens. We observed a relationship between the frequency of a mutated allele in the sample and the amount of DNA required for detection. This is partially in contrast to the findings of Magliacane et al. (2015), who reported that mutations could be detected even using very small amounts of DNA (1 ng per reaction) [28]. In particular, we observed that, in the presence of highly represented mutations (about 20%), only a very small amount of DNA (about 5 ng) was sufficient for the detection. By contrast, the quantity of DNA is crucial when identifying mutations with a frequency lower than 10%. Mutations represented at 10% and 30% were detected in all cases, but, when the mutated allele frequency was lower than 10%, the amount of template DNA influenced the performance of the MS analysis.

Taken together, these results showed that, to define the robustness of molecular profiling using MS, it is necessary to set the sensitivity of the method for each mutation to be investigated.

The validity of the mutation detection by MS was confirmed by comparison with the results reported in the literature: KRAS is the most frequently mutated gene in ADCs (37.2%), followed by EGFR (14.4%), whereas PIK3CA is frequently mutated in SCCs (27.3%) [10].

Three NSCLC cases were not further classified because of the paucity of specimens. Our molecular results identified the KRAS G12A mutation in one of them, suggesting that it could be classified as ADCs. This case confirms the importance of molecular profiling evaluation in lung cancer to complete the histopathological diagnosis, as already indicated for brain tumors [29].

In summary, MS allows the genotyping of several samples simultaneously by screening many known mutations in a single and cost effective test. It has high sensitivity, an important feature when a minority of mutant alleles must be distinguished from abundant wild-type alleles, and also allows mutations to be detected from a small amount of low-quality DNA, such as that typically obtained from poor-quality tissue specimens. Therefore, our experience emphasizes that it may not be appropriate to decline to perform molecular diagnosis on tissue specimens with a tumor component lower than 20%, or on poor biological materials.

Regarding the ctDNA evaluation, we observed concordant mutation detection between tumor tissue and plasma in four out of thirteen cases using MS, whereas, when using real-time PCR, we detected the tumor mutation in an additional case. The low sensitivity of both methods could be due to the low level of the mutated allele in the bloodstream or to tumor heterogeneity, although the small number of analyzed cases prevents us from drawing definitive conclusions. MS seems to be quite less sensitive than real-time PCR; however, MS has the advantage to screen panels of variations simultaneously. Our preliminary data suggest that, although in some cases analysis of ctDNA alone can be insufficient, MS has the potential to analyze liquid biopsy and monitor patients during treatment. Moreover, MS allows the estimation of alleles percentages, a data that could have a clinical significance in monitoring patients.

Our paper provides important advises for the use of MS in predictive analyses.

By analyzing 158 sequence alterations of the most frequently NSCLCs mutated genes in 92 tumor specimens, we confirmed that the method is able to detect mutated alleles present at 2,5% in the tumor specimens, and that the sensitivity can vary, depending on the mutation analyzed [19].

Interestingly, we noticed that the amount of DNA could affect the analysis. Specifically, when the mutation is present in more than 10% of alleles it is detectable even using low DNA amount (5 ng), but when the mutated alleles are less than 10%, the mutation detection can be compromised when using as low as 10 ng of DNA.

Finally, a proof-of concept investigation on liquid biopsy testing suggests that MS can be a reliable approach for this purpose even though it needs to be improved.

MS is a powerful and high throughput method for detecting known mutations, and allows to genotype scarce component of tumor cells in the tissue specimen, this has an important impact on patient clinical management.