Article Text

Download PDFPDF
Comparison of two next-generation sequencing-based approaches for liquid biopsy analysis in patients with non-small cell lung cancer: a multicentre study
  1. Silvia Bessi1,
  2. Francesco Pepe2,
  3. Gianluca Russo2,
  4. Pasquale Pisapia2,
  5. Marco Ottaviantonio1,
  6. Francesca Biancalani2,
  7. Antonino Iaccarino2,
  8. Maria Russo2,
  9. Mauro Biancalani3,
  10. Giancarlo Troncone2,
  11. Umberto Malapelle2
  1. 1Departmental Structure of Oncological Molecular Pathology, Azienda USL Toscana Centro, Prato, Italy
  2. 2Public Health, University of Naples Federico II, Naples, Italy
  3. 3Morphological Diagnostic and Biomolecular Characterization Area, Complex Unit of Pathological Anatomy, Azienda USL Toscana Centro, Prato, Italy
  1. Correspondence to Professor Giancarlo Troncone, Public Health, University of Naples Federico II, Naples, Italy; giancarlo.troncone{at}


In the era of personalised medicine, testing for an increasing number of predictive biomarkers is becoming a priority. However, tissue biopsies from these patients are oftentimes insufficient for conventional approaches, a common issue that deprives them of the clinical benefits of biomarker-directed treatments. To tackle this problem, many clinical laboratories are resorting to circulating tumour DNA (ctDNA), which is becoming increasingly appreciated as a valuable source for biomarker testing. In this context, next-generation sequencing (NGS) has become essential. Indeed, different NGS systems are able to detect several clinically relevant low-frequency hot-spot mutations simultaneously in a single run. However, their reproducibility in the analysis of ctDNA has not yet been investigated. The purpose of this study was to evaluate the reproducibility of using Illumina MiSeq and Thermo Fisher Ion S5 Plus platforms to assess pathogenic alterations in non-small cell lung cancer (NSCLC) liquid biopsy specimens. Using the in vitro diagnostic (IVD) NGS panel Myriapod NGS Cancer panel DNA (Diatech Pharmacogenetics) on MiSeq platform (Illumina), we reanalysed ctDNA extracted from a retrospective series of n=40 patients with advanced NSCLC previously tested with a custom NGS panel (SiRe) on Thermo Fisher Ion S5 Plus system. Overall, 13 out of 40 (32.5%) ctDNA samples displayed pathogenic alterations in at least two genes, namely, EGFR and KRAS. A concordance rate of 100% was identified between the two methodologies in terms of sample mutational status and total number of detected variables. All NGS platforms featured a high degree of concordance.

  • Pathology, Molecular
  • Lung Neoplasms
  • LUNG

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


In the era of personalised medicine, scores of predictive biomarkers for patients with lung cancer clinical stratification have been assessed in recent years.1–4 In this regard, international associations for cancer research have established a panel of must-test genes to identify the eligibility of patients with non-small cell lung cancer (NSCLC) for biomarker-targeted treatments.5 Unfortunately, for these patients, tissue samples are often too scant for accurate and meaningful molecular assessment. In this scenario, cytological or small biopsy specimens continue to represent a low abundant source of valuable nucleic acids to identify diagnostic and predictive biomarkers in routine clinical practice.6 However, single gene assays, routinely adopted in the vast majority of clinical laboratories, are inadequate to carry out large-scale analyses of the ever-increasing number of clinically relevant alterations in patients with NSCLC.7 A valid alternative to these assays is next-generation sequencing (NGS) platforms. First introduced at the beginning of the 21st century, they have gained great momentum in clinical practice especially in the last decade or so, offering clinicians the opportunity to detect several clinically relevant hot-spot mutations at very low frequency simultaneously in a single analysis. Despite the implementation of NGS platforms, a considerable percentage (15%–25%) of scant diagnostic samples (15%–25%) do not harbour sufficient neoplastic cells for the molecular analysis.8 In these cases, adopting integrating diagnostic sources of nucleic acids can be a reliable approach for molecular testing.9 In this context, liquid biopsy has recently been the focus of numerous molecular studies. In general, this test is by far a much less invasive option than histological or cytological tests to analyse the EGFR molecular status, as it requires only a simple blood draw.10 11 Notably, molecular testing for the analysis of circulating tumour DNA (ctDNA) has been approved for the detection of sensitising EGFR mutations for first or second Tyrosine kinase inhibitor (TKI) generation when tissue samples are deemed inadequate.12 13 In addition, some researchers have demonstrated the feasibility of adopting liquid biopsy even to identify acquired resistance mutation (EGFR exon 20 p.T790M) after first-line or second-line TKI treatments.14 15 Because of the ever-increasing need for biomarker testing in clinical practice, a plethora of NGS platforms have been made commercially available.16 These systems are generally based on different chemical processes involving library preparation and sequencing.17 However, the technical performance of each platform is controversial owing to several technical parameters.18

In this report, we evaluated the technical feasibility of applying two NGS-based liquid biopsy platforms to detect clinically relevant hot-spot mutations in 16 NSCLC driver genes. To this aim, we reanalysed ctDNA extracted from a retrospective series of n=40 patients with advanced NSCLC previously tested with a custom NGS panel (SiRe) on Ion S5 plus system (Thermo Fisher Scientific, Waltham, Massachusetts, USA) to establish the concordance rate between the platforms.

Study design

We selected a retrospective series of n=40 liquid biopsy specimens from patients with advanced NSCLC previously tested with a custom NGS assay (SiRe panel) on Ion S5 system (Thermo Fisher Scientific) at the Department of Public Health of the University of Naples Federico II. This panel can simultaneously cover clinically relevant hot-spot mutations in KRAS, EGFR, NRAS, BRAF, KIT, PDGFRα and PIK3CA.17 These data were classified as ‘reference results’ (RRs). Then, we analysed ctDNA extracted at the Departmental Structure of Oncological Molecular Pathology, S. Stefano Hospital-Prato, Oncological Department Azienda USL Toscana Centro, by using the in vitro diagnostic (IVD) NGS panel ‘Myriapod NGS Cancer panel DNA’ (Diatech Pharmacogenetics, Jesi, Italy) on the MiSeq platform (Illumina, San Diego, California, USA). The analysis enable us to evaluate the technical feasibility of detecting clinically relevant hot-spot mutations in 16 genes (ALK, BRAF, EGFR, ERBB2, FGFR3, HRAS, IDH1, IDH2, KIT, KRAS, MET, NRAS, PDGFRα, PIK3CA, RET and ROS1) from liquid biopsy samples obtained in routine clinical practice. These molecular data were named ‘index results’ (IRs). Concordance rate in detected mutations, analytical sensitivity and analytical specificity was evaluated. Written informed consent was obtained from all patients and documented in accordance with ‘The Italian Data Protection Authority’ ( 2485392). All information regarding human material was managed using anonymous numerical codes, and all samples were handled in compliance with the Declaration of Helsinki (

Sample preparation and ctDNA extraction

Liquid biopsy samples were processed as follows: 10 mL of peripheral blood was drawn from patients with NSCLC and collected in Vacutainer tubes (BD, Plymouth, UK). The tubes were immediately centrifuged at 2300 rpm for 10 min for plasma isolation containing ctDNA. A second centrifugation step at 2300 rpm for 10 min was then performed before storage of aliquoted plasma samples at −80°C. Cell-free DNA was automatically extracted and purified from 1.2 mL of plasma specimens. If the plasma volume ranged from 1.0 and 1.2 mL, phospate-buffered saline was complemented to reach the standardised volume for ctDNA extraction. ctDNA was extracted with the QIAsymphony platform (Qiagen, Hilden, Germany). QIAsymphony DSPVirus/Pathogen Midi Kit was adopted for ctDNA extraction as previously demonstrated.17 All preanalytical procedures for the management of liquid biopsy samples were standardised with a dedicated workflow at the Department of Public Health of the University of Naples Federico II.

NGS SiRe analysis

The liquid biopsy specimens were previously analysed with the SiRe panel. In general, this assay enables molecular analysis of hot spot clinically relevant alterations in DNA-based biomarkers that are key players in patients with NSCLC clinical stratification. Briefly, 15 µL of n=8 extracted ctDNA samples was automatically processed using Ion AmpliSeq DL8 (Thermo Fisher Scientific) on Ion Chef platform (Thermo Fisher Scientific). Then, two sets of amplified libraries were diluted at 60 pM and combined for template preparation on Ion Chef platform (Thermo Fisher Scientific) by using Ion Torrent Ion 510 and Ion 520 and Ion 530 Kit (Thermo Fisher Scientific), according to the manufacturer’s instructions. Finally, DNA fragments were charged on Ion 520 Chip (Thermo Fisher Scientific) and sequenced on Ion S5 System (Thermo Fisher Scientific), as previously described.17

Data analysis was carried out with default base-caller parameters: variant calling and coverage analysis were carried out on plug-in on Torrent Suite V.5.0.2 with SiRe(R) customised bed files plug-in (V. In particular, variant calling was automatically carried out with variant caller plug-in (V. in combination with optimised parameters of the SiRe panel workflow. Additionally, raw data extracted from primary analysis (BAM files) were inspected with a secondary analysis software (Golden Helix Genome Browser V.2.0.7, Bozeman, Montana, USA). In detail, only variants with 5× allele coverage and a quality score of ×20 were filtered with an amplicon coverage of at least 1000× alleles. Finally, the mutational frequency of these selected alterations were recorded.

Myriapod NGS cancer panel DNA analysis

Overall, 20 µL of previously extracted and characterised ctDNA was reanalysed with the Myriapod NGS Cancer panel DNA (Diatech Pharmacogenetics) on the MiSeq platform (Illumina) at the Departmental Structure of Oncological Molecular Pathology, S. Stefano Hospital-Prato, Oncological Department Azienda USL Toscana Centro. This panel allows investigation of a total of 107 DNA regions involving the hot-spot regions of 16 distinctive cancer related genes (ALK, BRAF, EGFR, ERBB2, FGFR3, HRAS, IDH1, IDH2, KIT, KRAS, MET, NRAS, PDGFRa, PIK3CA, RET and ROS1).

Briefly, 5 µL of each extracted ctDNA sample was quantified and their fragmentation profile was evaluated by real-time PCR analysis. Then, 25 µL of ctDNA was processed in three different batches (n=10, n=10 and n=20). Samples were amplified in two steps following the manufacturer’s instructions: PCR1 was performed to amplify hot-spot regions and PCR2 was performed to provide the indexed fragments. Moreover, ctDNA was quantified on the EasyPGX qPCR platform with the EasyPGX Analysis Software V.4.0.10 (Diatech Pharmacogenetics). Finally, libraries were diluted at 12 pM and pooled together for template generation. The sequencing phase was performed on the MiSeq platform (Illumina) according to the manufacturer’s instructions.

Data analysis for coverage and variant calling inspection was carried out by Myriapod NGS Data analysis Software V.5.0.4 (Diatech Pharmacogenetics). In detail, samples with minimal coverage of 2000× and a variant alteration of≥1% were selected. The IGV software (Broad Institute and the Regents, University of California) allowed us to integrate variant calling with molecular alterations ranging from 0.05% to 1% VAF. Finally, mutational frequency was recorded.

Statistical analysis

The sample set was divided into two groups by using the RR as the gold standard, depending on the absence or presence of at least one significant variant in at least one of the genes shared by both panels. Concordance rate between the two assays was assessed. Subsequently, molecular alterations detected by both assays and their VAF values were analysed. Finally, the concordance rate was evaluated. In addition, sensitivity and specificity were reported.

Cohen’s kappa coefficient (κ) was used to measure the degree of agreement between RR and IR. Pearson correlation coefficient (r) was instead used to measure the linear correlation between the percentage of allelic frequencies found in the two variants.


Remarkably, our molecular analysis was successfully carried out in all instances. In particular, 13 out of 40 (32.5%) liquid biopsy specimens showed pathogenic alterations in at least two genes: EGFR and KRAS. More precisely, our analysis revealed n=11 alterations in EGFR (n=8 exon 19 deletions/insertions, n=2 exon 20 p.T790M mutation and n=1 exon 20 insertion) and n=4 alterations in KRAS (n=2 exon 2 p.G12C, n=1 exon 3 p.Q22K and n=1 exon 4 p.A146T) (table 1). Moreover n=6 exon 18 p.V824V PDGFRα polymorphic variants were also detected. The concordance rate between the methodologies was 100%. Indeed, both showed the same mutational status and number of detected variants (tables 2 and 3, respectively). Interestingly, Myriapod NGS Cancer panel DNA (Diatech Pharmacogenetics) generated no false positives. On the other hand, the SiRe panel KIT exon detected one 11 p.541L mutation only in n=11 patients, most likely because this alteration is not covered by the Myriapod NGS Cancer panel DNA reference range.

Table 1

Molecular results from SiRe panel (RR) and Myriapod NGS cancer panel (IR)

Table 2

Concordance between RR and IR referring to MUT+ and MUT− samples

Table 3

Concordance between RR and IR referring to total identified variants

Both technologies efficiently detected the gene variants. Indeed, the SiRe panel identified a total of 21 significant gene alterations in the 17 positive samples (table 3). Likewise, the Myriapod NGS Cancer panel DNA correctly recognised all 21 mutations (100%). Overall, both technologies showed perfect sensitivity and specificity (100%; 81 (95% CI 6 to 1) and 85 (95% CI 7 to 1), respectively) (table 3).

Agreement between the two methodologies was classified as perfect (Cohen’s κ=1). Moreover, we collected 21/21 (100%) VAF values, and the concordance for each variant was very high (Pearson’s r=0.92, figure 1).

Figure 1

Schematic representation of comparable VAF correlation between RR and IR molecular data. IR, index result; RR, reference result; VAF, variant allele frequency.


Our data highlight the technical feasibility of applying two NGS assays with different sequencing procedures to liquid biopsy samples obtained from patients with advanced-stage NSCLC. In particular, our data clearly indicate that hybrid capture-based and amplicon-based sequencing platforms are totally interchangeable . In fact, they showed a concordance rate of 100% in a retrospective series of n=40 plasma specimens. Regarding the molecular alterations within the referenced genes, both EGFR sensitive or EGFR exon 20 p.T790M resistance point mutations were correctly identified by both molecular assays. These results reflect the efficiency of these NGS-based liquid biopsy assays to select patients with advanced lung cancer for EGFR TKI treatments,15 thereby having important clinical implications for patients with NSCLC.15 Beyond being able to detect EGFR mutations, these assays were also able to detect n=2 KRAS exon 2 p.G12C. Currently, KRAS gene alterations, and in particular exon 2 p.G12C, have acquired a relevant clinical role in the treatment decision making of patients with advanced stage NSCLC, particularly after the recent Food and Drug Administration approval of two TKI inhibitors, namely, adagrasib and sotorasib.19 20

The clinical application of NGS platforms to liquid biopsy samples is instrumental in informing treatment decisions in patients with NSCLC.17 In fact, owing to the low amount of ctDNA, the adoption of ultradeep technologies able to identify mutations at a very low VAF is crucial. However, a major obstacle to implementing NGS-based liquid biopsy in routine clinical practice is that each of the currently available platforms uses different panels. For this reason, harmonisation studies are necessary to evaluate the interchangeability of different diagnostic platforms, particularly for these complex samples. Indeed, consistency and interchangeability of platforms would ensure that patients with advanced NSCLC receive biomarker-directed treatments in the shortest time possible . In a previous experience, we evaluated the reproducibility of our custom NGS panel (SiRe) in five Italian clinical laboratories involved in molecular predictive pathology analysis. Results showed a high concordance rate between each centre. Similarly, in this study, we highlighted the feasibility of adopting Myriapod NGS Cancer panel DNA analysis (Diatech Pharmacogenetics) on MiSeq platform (Illumina) to detect pathogenic alterations in NSCLC liquid biopsy specimens by comparing our present results with those obtained in a retrospective series analysed with the SiRe NGS panel.8 The main advantages and disadvantages of these two platforms are summarised in table 4. Considering gene fusions, amplicon-based approaches are limited at DNA level due to the presence of large intronic regions involved in gene rearrangements, whereas at RNA level, it should be evaluated as expression imbalances between 5′ and 3′ regions of the target genes, enabling the identification of a gene rearrangement even when the fusion partner is unknown. Taking into account hybrid-capture approaches that adopt a gene-specific enrichment by a hybridisation step, probes are designed to match intronic, exonic and intergenic regions when DNA is adopted, whereas when RNA is used, probes are designed to target only exonic regions. In this latter approach, known and novel fusion variants can be detected by the panel. Overall, the major limitations of our study are represented by the small sample cohort size and the absence of ALK and ROS1 fused cases. In addition, it should be borne in mind that a crucial parameter for molecular analysis is represented by the nucleic acid assessment. Despite a high quality, ctDNA concentration into the torrent blood is variable and depends on biological mechanisms of tumour shedding and clinical status of patients. NGS platforms, which require a minimum amount of starting material to successfully perform molecular analysis, are influenced by this issue for the molecular assessment of predictive biomarkers in plasma samples. For these reasons, further studies are required to better evaluate the concordance of different ultradeep NGS approaches on liquid biopsy samples.

Table 4

Pros and cons of amplicon-based and hybrid capture-based platforms

In conclusion, our results demonstrate the high performance rate of the ultradeep NGS approach on liquid biopsy samples in routine molecular predictive practice, as evidenced by the perfect concordance rate (100%) between the two NGS systems.

Ethics statements

Patient consent for publication

Ethics approval

This study involves human participants. Written informed consent was obtained from all patients and documented in accordance with the general authorisation to process personal data for scientific research purposes from 'The Italian Data Protection Authority' (, accessed on 3 February 2022). All information regarding human material was managed using anonymous numerical codes, and all samples were handled in compliance with the Helsinki Declaration (, accessed on 3 February 2022). According to the aforementioned national guidelines, the double-blinded study did not require an ethical committee approval since it did not affect the clinical management of the involved patients’ samples. The participants gave informed consent to participate in the study before taking part.


We thank Dr Paola Merolla for editing the manuscript.



  • SB and FP are joint first authors.

  • Handling editor Runjan Chetty.

  • Twitter @PasqualePisapia, @UmbertoMalapel1

  • SB and FP contributed equally.

  • Contributors Conceptualisation: UM; writing—original draft preparation: GR, FP, PP and UM; supervision and project administration: SB, GT and UM; funding acquisition: GT; methodology, software, validation, formal analysis, investigation, resources, data curation, writing—review and editing and visualisation: all authors.

  • Funding Monitoraggio ambientale, studio ed approfondimento della salute della popolazione residente in aree a rischio—In attuazione della D.G.R. Campanian.180/2019. POR Campania FESR 2014–2020 Progetto 'Sviluppo di Approcci Terapeutici Innovativi per patologie Neoplastiche resistenti ai trattamenti—SATIN'.

  • Competing interests PP received personal fees as a speaker bureau of Novartis, unrelated to the current work. GT reported personal fees (as member of the speaker bureau or advisor) from Roche, MSD, Pfizer, Boehringer Ingelheim, Eli Lilly, BMS, GSK, Menarini, AstraZeneca, Amgen and Bayer, unrelated to the current work. UM received personal fees (as consultant and/or speaker bureau) from Boehringer Ingelheim, Roche, MSD, Amgen, Thermo Fisher Scientifics, Eli Lilly, Diaceutics, GSK, Merck and AstraZeneca, Janssen, Diatech, Novartis and Hedera, unrelated to the current work.

  • Provenance and peer review Not commissioned; externally peer reviewed.