REVIEW Mutated circulating tumor DNA as a liquid biopsy in lung cancer detection and treatment Martyna Filipska1,2 and Rafael Rosell1,2 1 Germans Trias i Pujol Research Institute and Hospital, Badalona, Spain 2 Autonomous University of Barcelona, Cerdanyola del Valles, Spain Keywords Over the past decade, substantial developments have been made in the cancer detection; cfDNA; ctDNA; liquid detection of circulating tumor DNA (ctDNA)—cell-free DNA (cfDNA) biopsy; LUAD; NSCLC; TMB fragments released into the circulation from tumor cells and displaying the genetic alterations of those cells. As such, ctDNA detected in liquid biop- Correspondence M. Filipska, IGTP, Ed. Muntanya, Ctra de sies serves as a powerful tool for cancer patient stratification, therapy guid- Can Ruti, Cami de les Escoles s/n, 08916 ance, detection of resistance, and relapse monitoring. In this Review, we Badalona, Barcelona, Spain describe lung cancer diagnosis and monitoring strategies using ctDNA E-mail: mfilipska@igtp.cat detection technologies and compile recent evidence regarding lung cancer- related mutation detection in liquid biopsy. We focus not only on epider- (Received 29 December 2020, revised 30 mal growth factor receptor (EGFR) alterations, but also on significant April 2021, accepted 7 May 2021, available co-mutations that shed more light on novel ctDNA-based liquid biopsy online 26 May 2021) applications. Finally, we discuss future perspectives of early-cancer detec- doi:10.1002/1878-0261.12983 tion and clonal hematopoiesis filtering strategies, with possible inclusion of microbiome-driven liquid biopsy. personalization in the field of oncology, it is necessary 1. Introduction to have an easily accessible and less invasive way to Personalized medicine—specifically precision oncology determine and follow the molecular makeup of a tumor —nowadays provides molecular characterization of a from the moment of its detection and over the treatment patient’s tumor via tissue biopsy and can help guide of the disease. One such approach is through a liquid treatment decisions. However, to fully implement biopsy, where the genetic characterization of the tumor Abbreviations AGR2, anterior gradient 2; AKT1, AKT serine/threonine kinase 1; ALK, ALK receptor tyrosine kinase; ASXL1, ASXL transcriptional regulator 1; BRAF, B-Raf proto-oncogene; CAPP-Seq, cancer profiling by deep sequencing; CDK4, cyclin-dependent kinase 4; cfDNA, cell-free DNA; cffDNA, cell-free fetal DNA; CH, clonal hematopoiesis; CNV, copy number variation; CTC, circulating tumor cell; ctDNA, circulating tumor DNA; cTMB, corrected tumor mutation burden; CTNNB1, catenin beta 1; DNMT3A, DNA methyltransferase 3 alpha; ds, double-stranded; EGFR, epidermal growth factor receptor; ERBB2, erb-b2 receptor tyrosine kinase 2; FGF, fibroblastic growth factor; FOXA1, forkhead box A1; JAK, Janus kinase; KEAP1, kelch-like ECH-associated protein 1; KRAS, KRAS proto-oncogene GTPase; LCMC, Lung Cancer Mutation Consortium; LSCC, lung squamous cell carcinoma; LUAD, lung adenocarcinoma; mbDNA, microbial DNA; MDM2, MDM2 proto-oncogene; MEK1, serine/threonine protein kinase MEK1; MET, MET proto-oncogene; mtDNA, mitochondrial DNA; NGS, next-generation sequencing; NKX2-1, NK2 homeobox 1; NRAS, NRAS proto-oncogene GTPase; NSCLC, non-small-cell lung cancer; OS, overall survival; PFS, progression- free survival; PIK3CA, phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; PTEN, phosphatase and tensin homolog; RB1, RB transcriptional corepressor 1; RET, ret proto-oncogene; ROS1, ROS proto-oncogene 1; RTK, receptor tyrosine kinase; SEER, Surveillance, Epidemiology and End Results; SMARCA4, SWI/SNF-related, matrix-associated, actin-dependent regulator of chromatin, subfamily a, member 4; ss, single-stranded; ST7, suppression of tumorigenicity 7; STK11, serine/threonine kinase 11; TCGA, The Cancer Genome Atlas; TET2, tet methylcytosine dioxygenase 2; TGFA, transforming growth factor alpha; TKI, tyrosine kinase inhibitor; TMB, tumor mutation burden; TP53, tumor protein p53; TTF-1, transcription termination factor 1. Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1667 Federation of European Biochemical Societies. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell can be assessed through a biofluid sample. The term ‘liq- concentration [20]. Later, Stroun et al. [21] demon- uid biopsy’ refers to tumor-derived analytes sampled strated that some of these DNA fragments were of from various biological fluids, usually blood, but also tumor origin, due to their genomic instability. In the other clinical specimens, such as urine, saliva, ascites, early 1990s, two independent studies noted the pres- and cerebrospinal fluid [1]. ence of specific KRAS proto-oncogene GTPase The constant development of technologies to detect (KRAS) and NRAS proto-oncogene GTPase (NRAS) cell-free DNA (cfDNA) with high sensitivity has facili- mutations in cfDNA from pancreatic adenocarcinoma tated the employment of liquid biopsies in diverse clin- [22] and patients with acute myelogenous leukemia ical applications, including in oncology. Analysis of [22,23]. This fraction of cfDNA was later termed as circulating tumor DNA (ctDNA) obtained from ‘circulating tumor DNA’ (ctDNA). plasma at multiple time points throughout the course ctDNA refers to cfDNA fragments that are released of the disease allows for patient stratification for treat- into the bloodstream from primary tumor or meta- ment (known as ‘companion diagnostics’), screening, static cells and display tumor-specific point mutations, monitoring response to the selected treatment, and chromosomal rearrangements, copy-number variation detection of minimal residual disease after surgery. (CNV), and DNA methylation [24–26]. Importantly, In this article, we define the terms of cfDNA and ctDNA is more fragmented than cfDNA, resulting in ctDNA, describe their properties, and outline the his- a much higher <100 bp fraction in the plasma [27]. torical breakthroughs in ctDNA detection. Most The 10-bp periodicity observed for fragments smaller importantly, we then aim at summarizing the most than 167 bp [3,28] corresponds to a turn of the DNA recent state-of-the-art developments in ctDNA utiliza- helix wrapped around the histone. This might protect tion in diagnosis, including early-stage detection, treat- one part of the DNA from the nucleases present in the ment selection, and follow-up, in non-small-cell lung blood. This specific fragmentation pattern suggests cancer (NSCLC) patients. that apoptosis may be a major source of cfDNA and that histones may be the key protein complex associ- ated with DNA in the blood. The release of longer 2. cfDNA and ctDNA: history, ctDNA fragments from tumor cells has been associ- definitions, and properties ated with necrotic cell death and occurs via active pro- cfDNA refers to degraded DNA fragments of usually cesses in living cells [11,29,30]. ctDNA fragments that 167 bp length that are released into the blood [2,3]. are released into the circulation mirror the tumor sta- The specific length may result from the action of a tus, its evolution, and the genomic alterations present caspase-dependent endonuclease that cleaves DNA in primary and/or metastatic tumors [11]. after a core histone and its linker. Recent studies have In a pan-cancer study involving 640 patients, Bette- shown a distinct nuclear fragmentation pattern, with gowda et al. [31] demonstrated that ctDNA analysis variable fragment lengths of cfDNA, from different might allow monitoring of the therapeutic response, tissue of origin [4–7]. Several different types of cfDNA tracking resistance, and, in some cases, early detection have been described in the circulation, including both of localized malignancies. They have also shown a cor- double-stranded (ds) and single-stranded (ss) DNA relation between tumor burden and stage of the dis- particles. In humans, cfDNA originates from all cells, ease. Significant differences in ctDNA levels were seen yet the vast majority is known to be of hematopoietic between cancer types, and the median cfDNA concen- provenance [8–11]. Within the ‘classical’ cfDNA, we tration was shown to be 100-fold higher in patients can distinguish more specific subclasses based on its with stage IV versus stage I disease [31]. site of origin or mechanism of release, for example, ctDNA has a notably short half-life in the blood- mitochondrial DNA (mtDNA) [12], cell-free fetal stream [32], and this characteristic is an important fea- DNA (cffDNA) [13], extrachromosomal circular DNA ture in analyzing dynamics of the mutations and [14], as well as microbial DNA (mbDNA) [15–17]. tumor burden after surgery or systemic treatment The first record of cfDNA detection in blood serum throughout the disease. Thus, real-time tumor dynam- and plasma was in 1948, by Mandel and colleagues ics might be monitored through ctDNA analysis for [18]. Later, in 1977, higher levels of cfDNA were early prediction and assessment of drug response, as detected in patients with pancreatic cancer compared well as early intervention independent of detection by with healthy controls [19], which led to the hypothesis imaging examinations or clinical symptoms [26]. A that tumors release DNA fragments to the circulation. recent study developed a mathematical model to pre- In 1983, Shapiro and colleagues confirmed correlations dict the shedding rate of early-stage NSCLC [33]. between benign vs. malignant tumors and cfDNA From this study, it has been estimated that there 1668 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring would be an average of only 1.7 genome copies of Gr€olz et al. [53] explain the importance of preserving ctDNA in 15 mL of blood for lung tumors with a vol- whole-blood specimens after blood drawn for use as ume of 1 cm3. liquid biopsies, and summarize preservation solutions In general, detection of ctDNA requires the presence that are currently available. Through entities like of typical mutations that can be readily detected by CANCER-ID or the International Liquid Biopsy Stan- simple sequencing techniques, proving the presence of dardization Alliance (ILSA) [54], the importance of tumor. With the advent of genomic information from working toward the global use of liquid biopsy in the most recent cancer genome sequencing studies, it oncology practice is being well recognized. has become clear that practically all cancers of every type harbor somatic alterations. Cancer somatic muta- 3. A glimpse at lung cancer tions occur at minor frequencies in normal cell popula- tions and therefore provide impeccably specific Lung cancer remains the most common cause of can- biomarkers from a biological perspective [34]. Some cer death worldwide, with an estimated 1.8 million historic studies provide temporal analyses of the total deaths each year [55]. About 85% of patients are his- tumor burden, as well as identifying specific mutations tologically grouped as NSCLC, of which lung adeno- that appear during therapy [29,35–40]. carcinoma (LUAD) and lung squamous cell carcinoma Identification of somatic mutations within white (LSCC) are the most common subtypes [56,57]. Until blood cells might be a recurring source of discordance 2004, all NSCLC subtypes—LUAD, LSCC, and large between tumor and total cfDNA genotyping. This phe- cell carcinoma—were treated in the same manner with nomenon is called clonal hematopoiesis (CH) and is chemotherapy (cisplatin or carboplatin in combination an aging-related phenomenon whereby nonmalignant with either docetaxel, paclitaxel, gemcitabine, or hematopoietic stem and progenitor cells acquire vinorelbine). The FDA approval of gefitinib—the first somatic alterations that can confer a selective advan- EGFR inhibitor—was a game changer in NSCLC tage [41]. CH mutations are similar to the mutations treatment, leading to stratification of patients with detected in plasma and may involve both canonical activating EGFR mutations to targeted therapy [58]. CH genes, such as DNA methyltransferase 3 alpha Lung cancer treatment is stage-specific. Early-stage (DNMT3A); tet methylcytosine dioxygenase 2 (TET2); disease can be cured by surgical resection, while locally ASXL transcriptional regulator 1 (ASXL1); and Janus advanced disease demands multimodal treatment, kinase (JAK), and driver mutations related to tumors, including chemotherapy, radiotherapy, and surgery for such as KRAS, phosphatidylinositol-4,5-bisphosphate chosen cases. Intrinsically, due to the usually late 3-kinase catalytic subunit alpha (PIK3CA), and epider- prognosis, NSCLC is a metastatic disease (henceforth, mal growth factor receptor (EGFR) mutations [42–45]. stage IV or metastatic is applied to NSCLC or Because hematopoietic cells are the primary source of LUAD) with gloomy prognosis. Its response rate is cfDNA [46] and contribute somatic variants to the about 30%, progression-free survival (PFS) 4– cfDNA pool [43,44], several approaches to distinguish 6 months, and median overall survival (OS) about mutations derived from CH from their tumor-derived 12 months; however, toxicity usually applies to all counterparts have been proposed [42,47,44]. Studies treatment strategies. In the past, some physicians with- that indicated that CH and tumor cfDNA have frag- held chemotherapy on the basis of patients’ age (with ments of a different size distribution might also help the cutoff arbitrarily set at 70 years) [57]. to distinguish between the two [48–50]. Chan et al. [51] elaborate elegantly on further clinical implications of 3.1. Early detection the CH phenomenon. In addition to considering CH before implementing Although the understanding of lung cancer pathobiol- ctDNA analysis in the clinic, other strict guidelines ogy has significantly improved over the last few dec- and standard operating procedures need to be formu- ades, poor disease prognosis is partially attributed to lated. Factors like preanalytical standardization need late stages at diagnosis, given that there are very few to be well optimized. For instance, the CANCER-ID early symptoms [59]. When diagnosed at an early consortium was funded between public and private stage, patients with NSCLC have a 5-year survival rate sector units with the aim of establishing standard of about 71%. For patients diagnosed with stage IV protocols for and clinical validation of ctDNA- and disease, it is less than 2% [60]. Early diagnosis may circulating tumor cell (CTC)-based biomarkers. thus improve patient outcome [61], especially if Lampignano et al. [52] compare the preanalytical and ctDNA-guided adjuvant therapy administration analytical workflows of cfDNA-based techniques, and reaches the bedside, as proposed in the TRACERx Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1669 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell consortium study [62,63]. Therefore, there is still a 3.2. Personalized treatment for NSCLC considerable need to develop noninvasive integrative biomarkers for early detection of lung cancer. Targetable oncogenic drivers account for almost 25% Despite the fact that several alterations have been of LUADs, of which EGFR mutations are the most noted in histologically normal bronchial epithelium frequent [69]. The use of targeted therapies has specimens of smokers [64], early diagnosis in NSCLC reduced lung cancer mortality [70]. Based on Surveil- remains low. Molecular changes are detected in hyper- lance, Epidemiology and End Results (SEER) cancer plasia and dysplasia, stages that precede carcinoma registries, among men, incidence-based mortality from in situ and invasive carcinoma [57]. Recent studies show NSCLC decreased by 6.3% annually from 2013 that aneuploidy and driver mutations also precede can- through 2016. Similar patterns were found among cer diagnosis by several years [65]. Driver fusion onco- women with NSCLC [70]. genes in LUAD could arise in the early decades of life, The Lung Cancer Mutation Consortium (LCMC) with a long latency period before diagnosis [66]. performed the first large-scale study in LUADs for In 2018, Cohen et al. [67] developed CancerSEEK— detection of oncogenic driver mutations in the United a multianalyte test that can detect eight human cancer States. Multiplex genotyping for mutation detection types through determination of the levels of circulating was performed in several sites, using any of three proteins and mutations in ctDNA. For lung cancer, methods: matrix-assisted laser desorption/ionization the probability of detection reached 75%. It seems that time-of-flight mass spectrometry (Sequenom, Arizona ctDNA combined with protein biomarkers might serve Research Laboratories), multiplex single-nucleotide as an opportunity to detect cancers before they metas- extension sequencing (SNaPshot, Applied Biosystems), tasize, when it is not yet evident radiologically. At this or Sanger sequencing with peptide nucleic acid probes stage, patients can be cured in up to 50% of cases with [71]. The LCMC prioritized genotyping EGFR, KRAS, systemic therapies. Nonetheless, further studies should erb-b2 receptor tyrosine kinase 2 (ERBB2), AKT ser- be performed for the possible implementation of Can- ine/threonine kinase 1 (AKT1), B-Raf proto-oncogene cerSEEK in the clinical practice. (BRAF), serine/threonine protein kinase MEK1 As the majority of cfDNA originates from (MEK1), NRAS, PIK3CA, ALK receptor tyrosine hematopoietic cells, these CH mutations are detected kinase (ALK), and MET proto-oncogene (MET). in plasma and, without appropriate controls, can be KRAS mutations were the most frequent, found in 182 incorrectly attributed to tumor. High-sensitivity of 733 specimens (25%), followed by sensitizing EGFR cfDNA analyses have identified CH mutations in 60– mutations (exon 19 deletions, L858R, L861Q, and 90% of individuals without cancer and shown them to G719X) in 122 of 733 specimens (17%). This kind of be age-related [43,44]. To avoid calling these false- EGFR mutation results in sensitivity to tyrosine kinase positive CH mutations, white blood cell controls, frag- inhibitors (TKIs). ALK rearrangements occurred in 57 ment length discrimination, CH-associated variant fil- of 733 specimens (8%). The median survival of tering, and deep-error controlled sequencing are patients with each of the five most common oncogenic required (Fig. 1). drivers ranged from 2.0 years (mutations in two genes) To address this issue, in their latest discovery, Cha- to 4.3 years (ALK-rearranged tumors). The 260 bon et al. [48] implemented certain improvements to patients with an oncogenic driver and treatment with a cancer profiling by deep sequencing (CAPP-Seq) [68] targeted drug had a median survival of 3.5 years; the of ctDNA analysis at lung cancer screening. Although 318 patients with a driver and no targeted therapy had very low levels of ctDNA were detected in early-stage 2.4 years median survival; and the 360 patients with lung cancer patients, its presence was confirmed as no driver identified had 2.1 years (P < 0.001) [71]. strongly prognostic. The mutational profile of total Other nationwide studies were performed following cfDNA from lung cancer patients and risk-matched the LCMC study [71], including a genomic screening controls revealed nonrecurrent CH, and the mutations network (LC-SCRUM-Japan) and the French nation- were detected on longer cfDNA fragments. A wide IFCT-InCa project, as well as a project in machine-learning method called ‘lung cancer likelihood Cologne, Germany (reviewed in ref. [69]). All these ini- in plasma’ (Lung-CLiP) allowed for robust discrimina- tiatives paved the way for the current genomic classifi- tion of early-stage lung cancer patients from risk- cation of LUAD and set the basis for the genomic matched controls [48]. These findings give hope for classification of LSCC [69,72]. employment of ctDNA-based screening methods into Recently, the TARGET study used a ctDNA assay the clinical practice, hence reducing cancer-related involving a panel of 641 cancer-associated genes. For mortality by increasing the early-detection rate. the first 100 patients, ctDNA data showed good 1670 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring Fig. 1. Strategies to filter clonal hematopoiesis (CH) from tumor-derived mutated circulating free DNA (cfDNA). Plasma-derived cfDNA, including circulating tumor DNA (ctDNA), is subjected to fragment size analysis to better discriminate between tumor- and non-tumor- derived cfDNA. Then, along with white blood cell (WBC)-derived genomic DNA (gDNA), cfDNA is sequenced and analyzed via a rigorous bioinformatic pipeline. CH may be excluded by filtering nonsynonymous mutations except for the positive selection analysis, mutational signature analysis, and genes canonically associated with CH [48]. concordance with tissue (78% concordance) and iden- (10%, n = 20) [74]. FOXA1 is a transcription factor tified potentially actionable mutations in 41% of that is frequently mutated in prostate, breast, bladder, patients. Of these patients, 11 of 41 (27%) went on to and salivary gland tumors [75]. The median PFS was a matched therapy [73]. 11 months; however, if the TP53 mutation was also present, then the median PFS was reduced (6 months) [74]. PFS was 5 months when pretreatment MET 3.2.1. EGFR-mutant LUAD genomic assessment amplification (2%, n = 4) was noted. The presence of Targeted next-generation sequencing in tumor tissue TP53 alterations was also associated with shorter sur- from metastatic NSCLC is gradually becoming more vival [74]. The TP53 concurrent mutation also pre- used for the identification of targetable driver muta- dicted shorter PFS to EGFR TKI in EGFR-mutated tions and gene fusions. However, the first question NSCLC [76]. More recent studies [77] reconfirm the that comes up when a driver actionable mutation is Helena Yu co-mutation plot of genomic alterations [74] detected is: What is the meaning of the other, often and help to gain insights on co-acquired alterations co-occurring, mutations and gene alterations reported? through the evolution of TKI therapy in NSCLC EGFR-mutant NSCLC was analyzed using the MSK- patients [77]. PIK3CA mutations showed a domain- IMPACT assay, a clinical test that detects mutations, dependent effect on PFS. Mutations in the kinase copy-number alterations, and select fusions involving domain (Y1021H and H1047R), helical domain 341 (version 1), 410 (version 2), or 468 (version 3) (E542K), and C2 domain (N345K) were associated with cancer-associated genes [74]. The median number of co- poorer PFS, while mutations in the p85-binding domain mutations was 5 (range 0–19) in samples prior to (R88Q, R108H, and K111E) were associated with an EGFR TKI therapy. The most frequent co-occurring improved survival [77]. Intriguingly, multiple driver mutations were tumor protein p53 (TP53) (60% mutations occur in the same gene, especially in PIK3CA n = 119), PIK3CA (12%, n = 23), catenin beta 1 (10% of samples) and EGFR (10%) [78]. In fact, The (CTNNB1) (9%, n = 18), and RB transcriptional core- Cancer Genome Atlas (TCGA) research network pressor 1 (RB1) (10%, n = 19). The most frequent con- showed that the principal driver alterations in NSCLC current amplifications were EGFR (22%, n = 45), NK2 (i.e., LUAD)—either EGFR mutations or KRAS muta- homeobox 1/transcription termination factor 1 (NKX2- tions—include the co-occurrence of several others, com- 1/TTF-1) (15%, n = 29), MDM2 proto-oncogene monly with TP53 mutations, as well as serine/threonine (MDM2) (12%, n = 23), cyclin-dependent kinase 4 kinase 11 (STK11) and kelch-like ECH-associated pro- (CDK4) (10%, n = 21), and forkhead box A1 (FOXA1) tein 1 (KEAP1) mutations and alterations [79]. Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1671 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell Evaluation) [83]. In tissue-positive patients, the bio- 3.2.2. Detection of EGFR mutations in ctDNA marker was identified alone (12/60) or concordant with In our group [80], EGFR mutations in ctDNA were ctDNA (48/60), with an 80% ctDNA sensitivity. assessed in a large cohort of 1026 NSCLC patients, Importantly, the ctDNA median turnaround time was and sensitizing EGFR mutations were found in 113 faster than tissue (9 vs 15 days; P < 0.0001) [83]. We patients (11%). More than 50% of samples had conducted a similar prospective study in Spain [84] in <10 pg mutated genomes per µL, with allelic fractions 186 NSCLC patients, confirming that the Guardant360 below 0.25%. Patients treated first line with TKI had NGS ctDNA assay was not inferior to standard-of- an objective response rate of 72% and a median PFS care tissue testing in detecting recommended biomark- of 11 months. Of the 105 patients screened after pro- ers, further confirming that ctDNA-based first-line gression to EGFR TKIs, sensitizing mutations were therapy produces outcomes similar to tissue-based test- found in 56%, and the acquired T790M mutation was ing [84]. found in 35%. Detection of EGFR mutations in As confirmed in multiple studies, detection of EGFR plasma ctDNA was used as a selection criterion for mutations in plasma serves as a satisfactory surrogate first-line gefitinib in patients with LUAD (BENEFIT for tissue biopsy. Significant weight should be put on study) [81]. The objective response rate among the co-occurring genetic alterations, since their detec- 188 patients was 72%, while median PFS was tion serves as an important prognostic marker. 9.5 months. Of 167 patients with available blood samples, 147 (88%) had clearance of EGFR mutations 3.2.3. Significance of EGFR L858R mutation in ctDNA in ctDNA at week 8, and median PFS was longer for these patients than for the 20 patients whose EGFR We carried out a large-scale ctDNA screening of mutations persisted at week 8 (11 months vs patients with NSCLC for EGFR mutations in Spain. 2.1 months, P < 0.0001). From baseline next- EGFR mutations were found in 350 of 2015 patients generation sequencing (NGS) data in 179 patients [us- (16.6%). For 217 patients who received erlotinib, the ing an ultra-deep (20 0009) 168-gene panel named adjusted hazard ratios for the duration of PFS were LungPlasma (Burning Rock Biotech, Guangzhou, 2.94 for men (P < 0.001); 1.92 for the presence of the China)], three subgroups were seen: those with EGFR L858R mutation, as compared with a deletion in exon mutations alone (n = 58), those with mutations in 19 (P = 0.02); and 1.68 for the presence of the L858R EGFR and tumor-suppressor genes [TP53 (65%), RB1 mutation in paired serum DNA, as compared with the (8%) or phosphatase and tensin homolog (PTEN), absence of the mutation (P = 0.02) [85]. After examin- (3%) (n = 97)], and those with mutations in EGFR ing the ctDNA of 97 patients included in the EUR- and oncogenes [MET, ERBB2, KRAS, BRAF, ret TAC phase 3 trial [86] using a peptide nucleic acid- proto-oncogene (RET) or ROS proto-oncogene 1 mediated 5’ nuclease real-time polymerase chain reac- (ROS1) (n = 24)]. Corresponding median PFS in these tion (TaqMan) assay, it was determined that median subgroups was 13.2 months, 9.3 months, and OS was shorter in patients with the L858R mutation 4.7 months, respectively [81]. in ctDNA than in those with the exon 19 deletion An earlier seminal study deciphered the evolution (13.7 vs 30 months, P < 0.01). Moreover, univariate and clinical impact of co-occurring genetic alterations analysis of patients with EGFR mutations in ctDNA in 1122 EGFR-mutated NSCLC patients by means of identified the L858R mutation in tumor tissue or in the Guardant 360 ctDNA assay, which covered single- ctDNA as a marker of shorter OS (hazard ratio, 2.70, nucleotide variants, small insertions/deletions (indels), P < 0.001) and PFS (hazard ratio, 2.04, P = 0.008). gene rearrangements/fusions, and copy-number gains For patients with the L858R mutation detected pri- across 68 clinically relevant cancer genes [82]. PIK3CA marily in tissue, median OS was 13.7 months for and CTNNB1 mutations were noted in tumor tissue, patients with the L858R mutation in ctDNA and as described above [74]. Interestingly, concurrent geno- 27.7 months for those with no mutation detected mic alterations detectable in EGFR-mutant NSCLC (P = 0.03). The conclusion was that the L858R muta- patients encoding the EGFR C797S mutation were also tion in ctDNA might be a novel surrogate prognostic found [82]. Using new versions of the Guardant360 marker [87]. ctDNA assay, covering 73 cancer-related genes, two The FLAURA phase 3 trial of osimertinib vs gefi- studies have reported great concordance between the tinib in first line also showed a shorter PFS and OS in ctDNA assay and tissue-based clinical genotyping. In EGFR-mutant NSCLC harboring the L858R mutation the United States, 282 patients were evaluated in the in tumor tissue [88,89]. Anterior gradient 2 (AGR2), a NILE study (Non-Invasive versus Invasive Lung disulfide isomerase that promotes lung tumorigenesis, 1672 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring is significantly expressed (using TCGA datasets) in the combination of first- and third-generation TKIs EGFR L858R airway tumors induced by transforming can restore EGFR inhibition. Conversely, if the two growth factor alpha (TGFA; an EGFR ligand), but mutations are in cis (on the same allele), the cells are not in normal lung. Experimental data points out that refractory to any of the EGFR TKIs, as well as the AGR2 may contribute to the growth of EGFR L858R combination of first- and third-generation inhibitors airway lung tumors induced by TGFA. In addition, [93]. It was foreseen that clinical assessment of the cis TGFA induces the expression of AGR2 in human versus trans configuration can be examined by NGS, EGFR-mutant LUAD. As explained below, TGFA- since T790M and C797S mutations are in close enough mediated fibrosis associated with EGFR-mutant lung proximity to coexist on a significant number of indi- tumors in vivo may induce growth factors (e.g., FGF) vidual sequencing reads [93,94]. Interestingly, it was that confer resistance to EGFR TKIs, gefitinib, erloti- shown that, at a concentration of 1 µmolL 1, afatinib nib, afatinib, and osimertinib. It has been posited that can inhibit mutant EGFR with C797S in the absence TGFA is a therapeutic target for recurrent EGFR- of T790M [93]. mutant lung cancer (e.g., osimertinib-resistant LUAD) Osimertinib is the preferred first-line therapy for [90]. Although the biological background of L858R EGFR-mutant NSCLC [88,89]; however, resistance mutations has not been elucidated, it is remarkable unavoidably develops in patients. Resistance is medi- that multiple driver mutations in the same EGFR gene ated by acquired secondary mutations in EGFR. In can occur for L858R, in contrast with the EGFR exon addition to C797S, others also occur, such as L718Q 19 deletion [91]. A number of rare (minor) mutations [95]. Analysis of ctDNA data from patients disclosed are found in extracellular and transmembrane that L718Q mutations usually appear in the context of domains, together with the L858R major driver muta- an L858R driver mutation [96]. This adds evidence to tion. These findings lead us to speculate that multiple the observation that additional mutations occur in the mutations in the same oncogene cooperate to potenti- presence of EGFR L858R, rather than in EGFR exon ate its tumor-promoting activity [90]. 19 deletion [78]. On the same lines as Niederst’s observations [93], treatment in mice revealed that both erlotinib and afa- 3.2.4. Landscape of acquired EGFR mutations, other tinib caused regression of osimertinib-resistant C797S- mutations, and gene fusions in EGFR-mutant NSCLC containing tumors, whereas only afatinib was effective EGFR-mutant NSCLC patients treated with TKI often in L718Q mutant tumors. Combination of first-line develop acquired EGFR mutations that, until now, osimertinib plus erlotinib could prevent the emergence were only examined in re-biopsies. As an example, a of secondary mutations in EGFR [96]. patient that harbored EGFR exon 19 deletion was A novel EGFR G724S mutation, causing resistance treated with afatinib [92]. After progression, the sec- to osimertinib, occurs with exon 19 deletion, but not ond biopsy still revealed the presence of EGFR exon L858R. In addition, the exon 19 deletion/G724S 19 deletion and the acquired EGFR exon 20 T790M. retains sensitivity to afatinib, but not to erlotinib [97]. Treatment was then switched to osimertinib (a EGFR C724S was identified in 4% of postosimertinib mutant-specific third-generation EGFR TKI). When patients treated with first-line osimertinib, whereas the third biopsy was performed, EGFR exon 19 dele- C797X was identified in 29% of postosimertinib tion was pervasively identified, with disappearance of patients treated with later-line osimertinib [98]. EGFR exon 20 T790M and emergent EGFR exon 20 Targeted NGS for 416 cancer-related genes was car- C797S (acquired with osimertinib). The patient was ried out in 93 osimertinib-resistant NSCLC patient sam- treated with gefitinib (EGFR inhibitor, first genera- ples, mainly in ctDNA, and matched pretreatment tion) which induced tumor response once more [92]. samples of 12 patients. A co-mutation plot of Previously, it was suggested that combining first- postosimertinib-treated patients revealed two subgroups and third-generation TKIs in first-line therapy could of patients: those with major EGFR tertiary mutations be crucial, because neither a T790M, nor a C797S at the positions of L718/G719, L792, and G796/C797 mutation, alone, would be able to drive resistance to (identified with a frequency of 9.7%, 10.8%, and the combination [93]. For T790M-positive, erlotinib- 24.7%, respectively). In most cases, mutations were also resistant NSCLCs that develop a C797S mutation fol- noted in EGFR T790M and TP53 [99]. Almost all lowing therapy with a third-generation TKI, the con- patients without EGFR resistance mutations showed figuration of the T790M and C797S mutations TP53, MET, KRAS, or PIK3CA mutations [99]. influences how the cells can respond to therapy. If the Median OS after osimertinib progression (osimer- two mutations are in trans (on separate alleles), then tinib given as second-line in EGFR T790M mutant Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1673 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell NSCLC patents) was 5.4 months in 40 patients from generation ALK inhibitors, including alectinib, briga- the AURA study, whose plasma was available after tinib and ensartinib, and lorlatinib, a third-generation disease progression [100]. Twelve (30%) of these had ALK inhibitor [103]. The combination of ALK and the T790M mutation (four of whom also had C797S). MET inhibitors is emerging as a plausible efficient Patients without detectable EGFR-activating mutations combination, since MET amplification has been in plasma before treatment had the best overall and detected in 15% of tumor biopsies from patients postprogression survival (22.4 months and relapsing on next-generation ALK inhibitors, including 10.8 months, respectively). Loss of T790M but pres- 12% and 22% of biopsies from patients progressing ence of EGFR-activating mutations in plasma was on second-generation inhibitors or lorlatinib, respec- associated with the shortest PFS (median 2.6 months). tively [104]. Also, two tumor specimens harbored sup- Importantly, in 22 postprogression tumor samples, one pression of tumorigenicity 7 (ST7)-MET squamous cell and two small-cell transformations were rearrangement [105]. A recent phase 2 study, VISION, seen. In addition, the number of patients was small evaluates tepotinib (a MET inhibitor) for the treat- and T790M was found in 50% of samples, C797S in ment of NSCLC with MET exon 14 skipping muta- 17%, MET amplification in 50%, BRAF mutations in tions. PFS with tepotinib was 11 months (tissue 8%, and KRAS mutations in 8%. biopsy, n = 60), 8.5 months (liquid biopsy, n = 66) and Once we have the data from a ctDNA analysis, 8.5 months (combined biopsy, n = 60) [106]. In addi- what should we do with it? What do we do when fac- tion, for MET exon 14 skipping mutations, patients ing the multiplicity of genomic aberrations described? treated with crizotinib (MET inhibitor; PROFILE Is it a basic rule in medicine to bear in mind their fre- 1001), PFS was significantly shorter in patients with quency? Secondly, are the acquired driver co- MET mutations detected in ctDNA versus nondetected alterations druggable? A recent study has shed light on in ctDNA, with a median of 3.6 months vs 7.8 months these issues [98]. MSK-Fusion Solid, a custom RNA- (hazard ratio = 2.27, P = 0.06) [107]. seq panel, was used to detect fusions in cases where no It is clear that targeted NGS in tissue and plasma resistance mechanism was identified by NGS and suffi- provides complementary theranostic information in the cient tissue was available. Among 62 patients, histolog- current era of single targeted therapy. However, the ical squamous transformation was identified in 15% of genomic evaluation could become complex with first-line osimertinib cases and 14% of later-line cases. the cornucopia of genomic alterations that are being Nineteen percent of patients treated with first-line developed, and tracking plasma is the most suitable osimertinib had off-target genetic resistance (2 MET advance that the practitioner has, although it poses amplifications, 1 KRAS mutation, 1 RET fusion, and multiple hurdles on how to treat the patient. One 1 BRAF fusion), whereas 4% had an acquired EGFR appealing case described an EGFR L858R patient that mutation (EGFR G724S). Patients with squamous also displayed EGFR G719S in four different regions transformation acquired PIK3CA mutation, chromo- of the resected lung tumor, in different allelic propor- some 3q amplification, and fibroblastic growth factor tions, as well as SWI/SNF-related, matrix-associated, (FGF) amplification. The compound mutation EGFR actin-dependent regulator of chromatin, subfamily a, S768 + V769L and the mutation MET H1094Y were member 4 (SMARCA4) [108]. At the time of relapse, also identified. Longitudinal analysis of two patients, she received afatinib and a liver metastasis biopsy was who received later-line osimertinib with emergence of performed where L858R was found in an allelic frac- ALK fusion, revealed that one patient acquired an tion of 14%, together with other gene mutations. After EGFR C797S mutation and lost the ALK fusion after 2 years of afatinib, a resection of an ovarian metasta- treatment with osimertinib and alectinib. The other sis revealed mutation of the EGFR gene in six different patient acquired the ALK G1202R mutation after regions: L858R, G719S (allelic fraction range, 40– treatment with osimertinib and alectinib [98]. Recent 50%), and EGFR C797S (allelic fraction range 9–14%) studies confirm the validity of ctDNA analysis for in five of the six areas of the ovary. SMARCA4 muta- response assessment to osimertinib [101] and others tion was found in all six ovarian areas analyzed, with regarding the prognostic impact of TP53 mutations, an allelic fraction of more than 40%. After the ovary suggesting that EGFR-mutant and TP53 wild-type metastasis resection, the patient continued afatinib for patients may benefit from the combination of EGFR more than 2 years without recurrence [108]. Almoner- TKI with bevacizumab [102]. tinib is a third-generation EGFR TKI with high selec- Great progress has also been made in the use of tivity for EGFR-sensitizing and T790M resistance ALK inhibitors for the treatment of patients with mutations that show great inhibitory activity against ALK-positive NSCLC, from crizotinib to second- T790M, T790M/L858R, and T790M/Del19 [109]. 1674 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring The secondary objective of a recently opened multicen- antiviral immunity [120]. The current management of ter, open-phase II clinical study (NCT04841811), in metastatic NSCLC has evolved from chemotherapy to patients with unresectable stage III NSCLC is to assess chemotherapy plus immune checkpoint inhibitors in a the safety of different treatment decisions guided by large number of NSCLC patients with no identifiable ctDNA monitoring [110]. driver oncogene mutations or fusions [56,121]. Targeted NGS ctDNA assays are warranted to man- Despite the progress in immunotherapy, not all age the complexity of genomic alterations; however, NSCLC patients respond. Whole-exome sequencing of intrinsic and acquired resistance also involves the identi- 104 patients treated with immune checkpoint inhibitors fication of pathways that influence sensitivity, such as in identified that corrected TMB (cTMB) adjusted for KRAS G12C-mutant NSCLC, where the combination tumor purity predicted the benefit of immunotherapy, of the KRAS G12C inhibitor, PI3K inhibitor, and as well as smoke-related mutational signature and SHP2 inhibitor caused tumor regression in mouse mod- human leukocyte antigen status. However, mutations els with acquired resistance to AMG510 [111]. Co- in receptor tyrosine kinase (RTK) genes were indica- treatment in EGFR-mutant NSCLC is highly recom- tive of no response [122]. mended to improve PFS and OS. We and other groups have shown that targeting STAT3, YAP1, and SHP2 4. Conclusions and perspectives could have great synergism in vitro and in vivo in EGFR-mutant cell line models [112–114] and that, by In the hope of overcoming several obstacles and chal- preventing NF-jB signaling activation, the formation of lenges resulting from classical tissue biopsy-based diag- acquired EGFR mutations could be prohibited [115]. nosis, liquid biopsy emerged as a robust tool for ctDNA monitoring and disease detection and monitor- ing. Importantly, liquid biopsies trump tissue biopsies 3.2.5. ctDNA as a biomarker for immunotherapy when there is insufficient material for testing or its Most recently, the introduction of immunotherapy quality is unsatisfactory. In addition, a patient’s poor with promising responses in subgroups of cancer performance status and tumor accessibility are often a patients preceded the search for biomarkers to better substantial concern. stratify patients. This raised the question of whether Tumors are highly heterogeneous, and sampling in studying a tumor microenvironment from the tissue its entirety is challenging: How well does a small tissue biopsy was crucial for predictive biomarker discovery biopsy sample represent the whole tumor? In patients or whether a biomarker could be reasonably sought in with multiple metastases, to gain a holistic view of the a peripheral blood sample. It has been demonstrated disease biopsy, samples should be collected from all of that the latter may be achieved via analysis of tumor the (known) metastatic sites (Fig. 2). Since the blood mutation burden (TMB) in ctDNA, adding another reaches most tumor sites in patients with advanced liquid biopsy biomarker to a constantly expanding list cancers, it is perhaps reasonable to speculate that of blood tests for the management of cancer. An inter- blood-based liquid biopsies might better reflect tumor esting approach to combine two liquid biopsy bio- heterogeneity. Moreover, tumors evolve over time and sources would be to complement TMB detected on can modify their molecular fingerprint, making clinical ctDNA with PD-L1 assessment on CTCs [116]. The decisions based on historical biopsy data insignificant. recently opened BESPOKE clinical trial The limitation of acquiring tissue biopsy samples lon- (NCT04761783) is to examine the impact of SIGNA- gitudinally to determine disease response or monitor TERATM (a personalized and tumor-informed 16-plex relapse is also reduced by the liquid biopsy approach, NGS assay to detect ctDNA) on clinical decision- since multiple samples can be collected noninvasively making regarding immunotherapy for treatment of over time (Fig. 2). solid tumors [117]. Although early-detection strategies based on ctDNA Interestingly, fusion-driven LUADs often have are promising, numerous hindrances must be SETD2 mutations, which are not seen in LUADs with addressed before they can be robustly applied in the EGFR, KRAS, BRAF, or MET mutations [66]. This clinic. False-positive results can be precarious for any finding is of interest, since ALK or ROS1 fusion- screening assay. Experience thus far suggests that driven LUADs show poor benefit with immune check- benign tumors and non-neoplastic conditions do not point inhibitors (anti-PD-1 or anti-PD-L1 monoclonal generally give rise to ctDNA [29], so the ‘overdiagno- antibodies) [118,119]. SETD2 methyltransferase medi- sis’ of benign tumors is not likely to pose a major ates STAT1 methylation on lysine 525, being an essen- problem. Moreover, strategies to filter out CH signal tial signaling event for interferon-alpha-dependent are being developed (Fig. 1) [34–36]. Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1675 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell Fig. 2. ctDNA-based liquid biopsy in clinical use. Lung cancer was diagnosed at early stage, EGFR exon 19 deletion (ex19del) was detected, and the tumor was subjected to resection. Subsequently, two liquid biopsies detected no mutated circulating tumor DNA (ctDNA), indicating no progression of the disease. At the recurrence, EGFR ex19del and L858R mutation (L858Rmut) was detected in ctDNA and erlotinib was administered to the patient. L858R mutant cells metastasized to the liver and developed resistance to erlotinib (yellow line). Tumor derived from the EGFR ex19del clone responded well to the therapy (gray line). Solid arrows indicate liquid biopsy sample collection time points; dashed arrows indicate treatment procedures. ctDNA can be robustly detected in plasma when a companion diagnostics approaches where the mutation significant number of copies of mutant ctDNA are is druggable. shed into blood. However, when the amounts of Despite its potential, ctDNA analysis is not suitable ctDNA are too low, analysis of individual mutant loci to diagnose all cancers, since some tumor types (e.g., might result in a negative result because of sampling gliomas and sarcomas) are poor ctDNA shedders. background noise even when using an assay with per- Moreover, ctDNA-based assays are applicable in fect analytical sensitivity [123]. It is much easier to tumors with higher TMB, whereas, for example, detect a single mutation in the follow-up of an glioblastoma multiforme or pancreatic adenocarci- advanced disease and more demanding in the early noma are barely detectable in blood, especially at ear- stage where higher depth for detecting a broad panel lier stages, due to their low shedding and TMB. Thus, of mutations requires more ctDNA input. Where the Poore et al. [17] have approached this limitation: They amount ctDNA isolated is not sufficient for an NGS addressed the possibility of using mbDNA to discrimi- analysis, alternative platforms, such as NanoString nate between cancer and healthy patients. In short, nCounter, can be of service [124]. their findings suggest that mbDNA may serve as a bio- A revolutionary approach for cancer management marker for cancer diagnosis and detection even for was proposed by Wan and colleagues [125], in which low-TMB cancers. This new path in liquid biopsy can- tumor-guided personalized ctDNA screening is per- cer diagnostics has been opened by highlighting formed in longitudinally collected plasma samples. another clinical application of microbiome-based Such patient-specific mutation lists provide an oppor- assays [17,126]. tunity for highly sensitive monitoring from a range of In conclusion, ctDNA-based liquid biopsies can be a sequencing data types using methods for signal aggre- powerful tool for cancer diagnosis, monitoring, prog- gation, weighting, and error suppression [125]. All this nosis, and individualized treatment and can completely makes ctDNA still the most useful technique for change the current paradigms of cancer management, 1676 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring especially in a multianalyte setting. However, consider- 8 Elzanowska J, Semira C & Costa-Silva B (2020) DNA able research and development are still needed to in extracellular vesicles: biological and clinical aspects. improve the isolation, enrichment, and downstream Mol Oncol 15, 1701–1714. analysis of all circulating biomarkers. In our view, in 9 Lam WKJ, Gai W, Sun K, Wong RSM, Chan RWY, current clinical practice, ctDNA measurements need to Jiang P, Chan NPH, Hui WWI, Chan AWH, Szeto be combined with standard-of-care approaches and CC et al. (2017) DNA of erythroid origin is present in ideally combined with other blood-based biosources. human plasma and informs the types of anemia. Clin Chem 63, 1614–1623. https://doi.org/10.1373/clinchem. 2017.272401. Acknowledgements 10 Moss J, Magenheim J, Neiman D, Zemmour H, Loyfer N, Korach A, Samet Y, Maoz M, Druid H, The authors thank Stephanie Davies for language Arner P et al. (2018) Comprehensive human cell-type editing. This project has received funding from the methylation atlas reveals origins of circulating cell-free European Union’s Horizon 2020 research and innova- DNA in health and disease. Nat Commun 9, 5068. tion program under the Marie Skłodowska-Curie 11 Thierry AR, El Messaoudi S, Gahan PB, Anker P & grant agreement no 765492. Stroun M (2016) Origins, structures, and functions of circulating DNA in oncology. Cancer Metastasis Rev Conflict of interest 35, 347–376. 12 Fernandes J, Michel V, Camorlinga-Ponce M, Gomez The authors declare no conflict of interest. A, Maldonado C, De Reuse H, Torres J & Touati E (2014) Circulating mitochondrial DNA level, a noninvasive biomarker for the early detection of References gastric cancer. Cancer Epidemiol Biomarkers Prev 23, 1 Pantel K & Alix-Panabieres C (2010) Circulating 2430–2438. tumour cells in cancer patients: challenges and 13 Alberry M, Maddocks D, Jones M, Abdel Hadi M, perspectives. Trends Mol Med 16, 398–406. Abdel-Fattah S, Avent N & Soothill PW (2007) Free 2 Diehl F, Li M, Dressman D, He Y, Shen D, Szabo S, fetal DNA in maternal plasma in anembryonic Diaz LA, Goodman SN, David KA, Juhl H et al. pregnancies: confirmation that the origin is the (2005) Detection and quantification of mutations in the trophoblast. Prenat Diagn 27, 415–418. plasma of patients with colorectal tumors. Proc Natl 14 Kumar P, Dillon LW, Shibata Y, Jazaeri AA, Jones Acad Sci USA 102, 16368–16373. DR & Dutta A (2017) Normal and cancerous tissues 3 Lo YM, Chan KC, Sun H, Chen EZ, Jiang P, Lun release extrachromosomal circular DNA (eccDNA) FM, Zheng YW, Leung TY, Lau TK, Cantor CR into the circulation. Mol Cancer Res 15, 1197–1205. et al. (2010) Maternal plasma DNA sequencing reveals 15 Li SK, Leung RK, Guo HX, Wei JF, Wang JH, the genome-wide genetic and mutational profile of the Kwong KT, Lee SS, Zhang C & Tsui SK (2012) fetus. Sci Transl Med 2, 61ra91. Detection and identification of plasma bacterial and 4 Cristiano S, Leal A, Phallen J, Fiksel J, Adleff V, viral elements in HIV/AIDS patients in comparison to Bruhm DC, Jensen S, Medina JE, Hruban C, White healthy adults. Clin Microbiol Infect 18, 1126–1133. JR et al. (2019) Genome-wide cell-free DNA 16 Moustafa A, Xie C, Kirkness E, Biggs W, Wong E, fragmentation in patients with cancer. Nature 570, Turpaz Y, Bloom K, Delwart E, Nelson KE, Venter 385–389. JC et al. (2017) The blood DNA virome in 8,000 5 Snyder MW, Kircher M, Hill AJ, Daza RM & humans. PLoS Pathog 13, e1006292. Shendure J (2016) Cell-free DNA comprises an in vivo 17 Poore GD, Kopylova E, Zhu Q, Carpenter C, nucleosome footprint that informs its tissues-of-origin. Fraraccio S, Wandro S, Kosciolek T, Janssen S, Cell 164, 57–68. Metcalf J, Song SJ et al. (2020) Microbiome analyses 6 Sun K, Jiang P, Cheng SH, Cheng THT, Wong J, of blood and tissues suggest cancer diagnostic Wong VWS, Ng SSM, Ma BBY, Leung TY, Chan SL approach. Nature 579, 567–574. et al. (2019) Orientation-aware plasma cell-free DNA 18 Mandel P & Metais P (1948) Nuclear acids in human fragmentation analysis in open chromatin regions blood plasma. C R Seances Soc Biol Fil 142, 241–243. informs tissue of origin. Genome Res 29, 418–427. 19 Leon SA, Shapiro B, Sklaroff DM & Yaros MJ (1977) 7 Ulz P, Thallinger GG, Auer M, Graf R, Kashofer K, Free DNA in the serum of cancer patients and the Jahn SW, Abete L, Pristauz G, Petru E, Geigl JB et al. effect of therapy. Cancer Res 37, 646–650. (2016) Inferring expressed genes by whole-genome 20 Shapiro B, Chakrabarty M, Cohn EM & Leon SA sequencing of plasma DNA. Nat Genet 48, 1273–1278. (1983) Determination of circulating DNA levels in Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1677 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell patients with benign or malignant gastrointestinal (2020) A mathematical model of ctDNA shedding disease. Cancer 51, 2116–2120. predicts tumor detection size. Sci Adv 6, https://doi. 21 Stroun M, Anker P, Maurice P, Lyautey J, Lederrey C org/10.1126/sciadv.abc4308 & Beljanski M (1989) Neoplastic characteristics of the 34 Vogelstein B, Papadopoulos N, Velculescu VE, Zhou DNA found in the plasma of cancer patients. S, Diaz LA & Kinzler KW (2013) Cancer genome Oncology 46, 318–322. landscapes. Science 339, 1546–1558. 22 Sorenson GD, Pribish DM, Valone FH, Memoli VA, 35 Dawson SJ, Tsui DW, Murtaza M, Biggs H, Rueda Bzik DJ & Yao SL (1994) Soluble normal and OM, Chin SF, Dunning MJ, Gale D, Forshew T, mutated DNA sequences from single-copy genes in Mahler-Araujo B et al. (2013) Analysis of circulating human blood. Cancer Epidemiol Biomarkers Prev 3, tumor DNA to monitor metastatic breast cancer. N 67–71. Engl J Med 368, 1199–1209. 23 Vasioukhin V, Anker P, Maurice P, Lyautey J, 36 Diaz LA Jr, Sausen M, Fisher GA & Velculescu VE Lederrey C & Stroun M (1994) Point mutations of the (2013) Insights into therapeutic resistance from whole- N-ras gene in the blood plasma DNA of patients with genome analyses of circulating tumor DNA. myelodysplastic syndrome or acute myelogenous Oncotarget 4, 1856–1857. leukaemia. Br J Haematol 86, 774–779. 37 Diaz LA Jr, Williams RT, Wu J, Kinde I, Hecht JR, 24 Corcoran RB & Chabner BA (2018) Application of Berlin J, Allen B, Bozic I, Reiter JG, Nowak MA cell-free DNA analysis to cancer treatment. N Engl J et al. (2012) The molecular evolution of acquired Med 379, 1754–1765. resistance to targeted EGFR blockade in colorectal 25 Heitzer E, Ulz P & Geigl JB (2015) Circulating tumor cancers. Nature 486, 537–540. DNA as a liquid biopsy for cancer. Clin Chem 61, 38 Holdhoff M, Schmidt K, Donehower R & Diaz LA Jr 112–123. (2009) Analysis of circulating tumor DNA to confirm 26 Oliveira KCS, Ramos IB, Silva JMC, Barra WF, somatic KRAS mutations. In J Natl Cancer Inst 101, Riggins GJ, Palande V, Pinho CT, Frenkel- 1284–1285. Morgenstern M, Santos SEB, Assumpcao PP et al. 39 Leary RJ, Sausen M, Kinde I, Papadopoulos N, (2020) Current perspectives on circulating tumor Carpten JD, Craig D, O’Shaughnessy J, Kinzler KW, DNA, precision medicine, and personalized clinical Parmigiani G, Vogelstein B et al. (2012) Detection of management of cancer. Mol Cancer Res 18, 517–528. chromosomal alterations in the circulation of cancer 27 Mouliere F, Robert B, Arnau Peyrotte E, Del Rio M, patients with whole-genome sequencing. Sci Transl Ychou M, Molina F, Gongora C & Thierry AR (2011) Med 4, 162ra154. High fragmentation characterizes tumour-derived 40 Murtaza M, Dawson SJ, Tsui DW, Gale D, Forshew circulating DNA. PLoS One 6, e23418. T, Piskorz AM, Parkinson C, Chin SF, Kingsbury Z, 28 Jiang P, Chan CW, Chan KC, Cheng SH, Wong J, Wong AS et al. (2013) Non-invasive analysis of Wong VW, Wong GL, Chan SL, Mok TS, Chan HL acquired resistance to cancer therapy by sequencing of et al. (2015) Lengthening and shortening of plasma plasma DNA. Nature 497, 108–112. DNA in hepatocellular carcinoma patients. Proc Natl 41 Steensma DP, Bejar R, Jaiswal S, Lindsley RC, Acad Sci USA 112, E1317–E1325. Sekeres MA, Hasserjian RP & Ebert BL (2015) Clonal 29 Diehl F, Schmidt K, Choti MA, Romans K, hematopoiesis of indeterminate potential and its Goodman S, Li M, Thornton K, Agrawal N, Sokoll L, distinction from myelodysplastic syndromes. Blood Szabo SA et al. (2008) Circulating mutant DNA to 126, 9–16. assess tumor dynamics. Nat Med 14, 985–990. 42 Hu Y, Ulrich BC, Supplee J, Kuang Y, Lizotte PH, 30 Stroun M, Lyautey J, Lederrey C, Olson-Sand A & Feeney NB, Guibert NM, Awad MM, Wong KK, Anker P (2001) About the possible origin and J€anne PA et al. (2018) False-positive plasma mechanism of circulating DNA apoptosis and active genotyping due to clonal hematopoiesis. Clin Cancer DNA release. Clin Chim Acta 313, 139–142. Res 24, 4437–4443. 31 Bettegowda C, Sausen M, Leary RJ, Kinde I, Wang 43 Liu J, Chen X, Wang J, Zhou S, Wang CL, Ye MZ, Y, Agrawal N, Bartlett BR, Wang H, Luber B, Alani Wang XY, Song Y, Wang YQ, Zhang LT et al. RM et al. (2014) Detection of circulating tumor DNA (2019) Biological background of the genomic in early- and late-stage human malignancies. Sci Transl variations of cf-DNA in healthy individuals. Ann Med 6, 224ra224. Oncol 30, 464–470. 32 Kustanovich A, Schwartz R, Peretz T & Grinshpun A 44 Razavi P, Li BT, Brown DN, Jung B, Hubbell E, Shen (2019) Life and death of circulating cell-free DNA. R, Abida W, Juluru K, De Bruijn I, Hou C et al. Cancer Biol Ther 20, 1057–1067. (2019) High-intensity sequencing reveals the sources of 33 Avanzini S, Kurtz DM, Chabon JJ, Moding EJ, Hori plasma circulating cell-free DNA variants. Nat Med SS, Gambhir SS, Alizadeh AA, Diehn M & Reiter JG 25, 1928–1937. 1678 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring 45 Xia L, Li Z, Zhou B, Tian G, Zeng L, Dai H, Li X, incidence and mortality worldwide for 36 cancers in Liu C, Lu S, Xu F et al. (2017) Statistical analysis of 185 countries. CA Cancer J Clin 71, 209–249. mutant allele frequency level of circulating cell-free 56 Herbst RS, Morgensztern D & Boshoff C (2018) The DNA and blood cells in healthy individuals. Sci Rep biology and management of non-small cell lung cancer. 7, 7526. Nature 553, 446–454. 46 Lui YY, Chik KW, Chiu RW, Ho CY, Lam CW & 57 Spira A & Ettinger DS (2004) Multidisciplinary Lo YM (2002) Predominant hematopoietic origin of management of lung cancer. N Engl J Med 350, 379–392. cell-free DNA in plasma and serum after sex- 58 Lynch TJ, Bell DW, Sordella R, Gurubhagavatula S, mismatched bone marrow transplantation. Clin Chem Okimoto RA, Brannigan BW, Harris PL, Haserlat 48, 421–427. SM, Supko JG, Haluska FG et al. (2004) Activating 47 Li BT, Janku F, Jung B, Hou C, Madwani K, Alden mutations in the epidermal growth factor receptor R, Razavi P, Reis-Filho JS, Shen R, Isbell JM et al. underlying responsiveness of non-small-cell lung cancer (2019) Ultra-deep next-generation sequencing of to gefitinib. N Engl J Med 350, 2129–2139. plasma cell-free DNA in patients with advanced lung 59 Gridelli C, Rossi A, Carbone DP, Guarize J, cancers: results from the Actionable Genome Karachaliou N, Mok T, Petrella F, Spaggiari L & Consortium. Ann Oncol 30, 597–603. Rosell R (2015) Non-small-cell lung cancer. Nat Rev 48 Chabon JJ, Hamilton EG, Kurtz DM, Esfahani MS, Dis Primers 1, 15009. Moding EJ, Stehr H, Schroers-Martin J, Nabet BY, 60 Goldstraw P, Chansky K, Crowley J, Rami-Porta R, Chen B, Chaudhuri AA et al. (2020) Integrating Asamura H, Eberhardt WE, Nicholson AG, Groome genomic features for non-invasive early lung cancer P, Mitchell A & Bolejack V (2016) The IASLC lung detection. Nature 580, 245–251. cancer staging project: proposals for revision of the 49 Leal A, van Grieken NCT, Palsgrove DN, Phallen J, TNM stage groupings in the forthcoming (eighth) Medina JE, Hruban C, Broeckaert MAM, Anagnostou edition of the TNM classification for lung cancer. V, Adleff V, Bruhm DC et al. (2020) White blood cell J Thorac Oncol 11, 39–51. and cell-free DNA analyses for detection of residual 61 Cassim S, Chepulis L, Keenan R, Kidd J, Firth M & disease in gastric cancer. Nat Commun 11, 525. Lawrenson R (2019) Patient and carer perceived 50 Marass F, Stephens D, Ptashkin R, Zehir A, Berger barriers to early presentation and diagnosis of lung MF, Solit DB, Diaz LA & Tsui DWY (2020) cancer: a systematic review. BMC Cancer 19, 25. Fragment size analysis may distinguish clonal 62 Abbosh C, Birkbak NJ, Wilson GA, Jamal-Hanjani hematopoiesis from tumor-derived mutations in cell- M, Constantin T, Salari R, Le Quesne J, Moore DA, free DNA. Clin Chem 66, 616–618. Veeriah S, Rosenthal R et al. (2017) Phylogenetic 51 Chan HT, Chin YM, Nakamura Y & Low SK (2020) ctDNA analysis depicts early-stage lung cancer Clonal hematopoiesis in liquid biopsy: from biological evolution. Nature 545, 446–451. noise to valuable clinical implications. Cancers (Basel) 63 Li BT, Stephens D, Chaft JE, Rudin CM, Jones DR, 12, 2277. https://doi.org/10.3390/cancers12082277 Rusch VW, Rimner A & Isbell JM (2017) Liquid 52 Lampignano R, Neumann MHD, Weber S, Kloten V, biopsy for ctDNA to revolutionize the care of patients Herdean A, Voss T, Groelz D, Babayan A, Tibbesma with early stage lung cancers. Ann Transl Med 5, 479. M, Schlumpberger M et al. (2020) Multicenter 64 Wistuba II, Mao L & Gazdar AF (2002) Smoking evaluation of circulating cell-free DNA extraction and molecular damage in bronchial epithelium. Oncogene downstream analyses for the development of 21, 7298–7306. standardized (Pre)analytical work flows. Clin Chem 66, 65 Gerstung M, Jolly C, Leshchiner I, Dentro SC, 149–160. Gonzalez S, Rosebrock D, Mitchell TJ, Rubanova Y, 53 Gr€olz D, Hauch S, Schlumpberger M, Guenther K, Anur P, Yu K et al. (2020) The evolutionary history of Voss T, Sprenger-Haussels M & Oelm€ uller U (2018) 2,658 cancers. Nature 578, 122–128. Liquid biopsy preservation solutions for standardized 66 Lee JJ, Park S, Park H, Kim S, Lee J, Youk J, Yi K, pre-analytical workflows-venous whole blood and An Y, Park IK, Kang CH et al. (2019) Tracing plasma. Curr Pathobiol Rep 6, 275–286. oncogene rearrangements in the mutational history of 54 Connors D, Allen J, Alvarez JD, Boyle J, Cristofanilli lung adenocarcinoma. Cell 177, 1842–1857.e1821. M, Hiller C, Keating S, Kelloff G, Leiman L, 67 Cohen JD, Li L, Wang Y, Thoburn C, Afsari B, McCormack R et al. (2020) International liquid biopsy Danilova L, Douville C, Javed AA, Wong F, Mattox standardization alliance white paper. Crit Rev Oncol A et al. (2018) Detection and localization of surgically Hematol 156, 103112. resectable cancers with a multi-analyte blood test. 55 Sung H, Ferlay J, Siegel RL, Laversanne M, Science 359, 926–930. Soerjomataram I, Jemal A & Bray F (2021) Global 68 Newman AM, Lovejoy AF, Klass DM, Kurtz DM, cancer statistics 2020: GLOBOCAN estimates of Chabon JJ, Scherer F, Stehr H, Liu CL, Bratman SV, Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1679 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell Say C et al. (2016) Integrated digital error suppression 80 Mayo-de-Las-Casas C, Jordana-Ariza N, Garz on- for improved detection of circulating tumor DNA. Nat Iba~nez M, Balada-Bel A, Bertran-Alamillo J, Viteri- Biotechnol 34, 547–555. Ramırez S, Reguart N, Mu~ noz-Quintana MA, Lianes- 69 Rosell R & Karachaliou N (2016) Large-scale Barragan P, Camps C et al. (2017) Large scale, screening for somatic mutations in lung cancer. Lancet prospective screening of EGFR mutations in the blood 387, 1354–1356. of advanced NSCLC patients to guide treatment 70 Howlader N, Forjaz G, Mooradian MJ, Meza R, decisions. Ann Oncol 28, 2248–2255. Kong CY, Cronin KA, Mariotto AB, Lowy DR & 81 Wang Z, Cheng Y, An T, Gao H, Wang K, Zhou Q, Feuer EJ (2020) The effect of advances in lung-cancer Hu Y, Song Y, Ding C, Peng F et al. (2018) Detection treatment on population mortality. N Engl J Med 383, of EGFR mutations in plasma circulating tumour 640–649. DNA as a selection criterion for first-line gefitinib 71 Kris MG, Johnson BE, Berry LD, Kwiatkowski DJ, treatment in patients with advanced lung Iafrate AJ, Wistuba II, Varella-Garcia M, Franklin adenocarcinoma (BENEFIT): a phase 2, single-arm, WA, Aronson SL, Su PF et al. (2014) Using multicentre clinical trial. Lancet Respir Med 6, 681– multiplexed assays of oncogenic drivers in lung cancers 690. to select targeted drugs. JAMA 311, 1998–2006. 82 Blakely CM, Watkins TBK, Wu W, Gini B, Chabon 72 Rosell R, Karachaliou N & Arrieta O (2020) Novel JJ, McCoach CE, McGranahan N, Wilson GA, molecular targets for the treatment of lung cancer. Birkbak NJ, Olivas VR et al. (2017) Evolution and Curr Opin Oncol 32, 37–43. clinical impact of co-occurring genetic alterations in 73 Rothwell DG, Ayub M, Cook N, Thistlethwaite F, advanced-stage EGFR-mutant lung cancers. Nat Genet Carter L, Dean E, Smith N, Villa S, Dransfield J, 49, 1693–1704. Clipson A et al. (2019) Utility of ctDNA to support 83 Leighl NB, Page RD, Raymond VM, Daniel DB, patient selection for early phase clinical trials: the Divers SG, Reckamp KL, Villalona-Calero MA, Dix TARGET study. Nat Med 25, 738–743. D, Odegaard JI, Lanman RB et al. (2019) Clinical 74 Yu HA, Suzawa K, Jordan E, Zehir A, Ni A, Kim R, utility of comprehensive cell-free DNA analysis to Kris MG, Hellmann MD, Li BT, Somwar R et al. identify genomic biomarkers in patients with newly (2018) Concurrent alterations in EGFR-mutant lung diagnosed metastatic non-small cell lung cancer. Clin cancers associated with resistance to EGFR kinase Cancer Res 25, 4691–4700. inhibitors and characterization of MTOR as a 84 Palmero R, Taus A, Viteri S, Majem M, Carcereny E, mediator of resistance. Clin Cancer Res 24, 3108–3118. Garde-Noguera J, Felip E, Nadal E, Malfettone A, 75 Parolia A, Cieslik M, Chu SC, Xiao L, Ouchi T, Sampayo M et al. (2021) Biomarker discovery and Zhang Y, Wang X, Vats P, Cao X, Pitchiaya S et al. outcomes for comprehensive cell-free circulating tumor (2019) Distinct structural classes of activating FOXA1 DNA versus standard-of-care tissue testing in alterations in advanced prostate cancer. Nature 571, advanced non–small-cell lung cancer. JCO Precis 413–418. Oncol 5, 93–102. https://doi.org/10.1200/PO.20.00241 76 Kim ES, Roy UB, Ersek JL, King J, Smith RA, 85 Rosell R, Moran T, Queralt C, Porta R, Cardenal F, Martin N, Martins R, Moore A, Silvestri GA & Jett J Camps C, Majem M, Lopez-Vivanco G, Isla D, (2019) Updates regarding biomarker testing for non- Provencio M et al. (2009) Screening for epidermal small cell lung cancer: considerations from the national growth factor receptor mutations in lung cancer. N lung cancer roundtable. J Thorac Oncol 14, 338–342. Engl J Med 361, 958–967. 77 Jin Y, Bao H, Le X, Fan X, Tang M, Shi X, Zhao J, 86 Rosell R, Carcereny E, Gervais R, Vergnenegre A, Yan J, Xu Y, Quek K et al. (2020) Distinct co- Massuti B, Felip E, Palmero R, Garcia-Gomez R, acquired alterations and genomic evolution during Pallares C, Sanchez JM et al. (2012) Erlotinib versus TKI treatment in non-small-cell lung cancer patients standard chemotherapy as first-line treatment for with or without acquired T790M mutation. Oncogene European patients with advanced EGFR mutation- 39, 1846–1859. positive non-small-cell lung cancer (EURTAC): a 78 Saito Y, Koya J, Araki M, Kogure Y, Shingaki S, multicentre, open-label, randomised phase 3 trial. Tabata M, McClure MB, Yoshifuji K, Matsumoto S, Lancet Oncol 13, 239–246. Isaka Y et al. (2020) Landscape and function of 87 Karachaliou N, Mayo-de las Casas C, Queralt C, de multiple mutations within individual oncogenes. Aguirre I, Melloni B, Cardenal F, Garcia-Gomez R, Nature 582, 95–99. Massuti B, Sanchez JM, Porta R et al. (2015) 79 Network CGAR (2014) Comprehensive molecular Association of EGFR L858R mutation in circulating profiling of lung adenocarcinoma. Nature 511, free DNA with survival in the EURTAC Trial. JAMA 543–550. Oncol 1, 149–157. 1680 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies. M. Filipska and R. Rosell ctDNA in lung cancer detection and monitoring 88 Ramalingam SS, Vansteenkiste J, Planchard D, Cho 98 Schoenfeld AJ, Chan JM, Kubota D, Sato H, Rizvi H, BC, Gray JE, Ohe Y, Zhou C, Reungwetwattana T, Daneshbod Y, Chang JC, Paik PK, Offin M, Arcila Cheng Y, Chewaskulyong B et al. (2020) Overall ME et al. (2020) Tumor analyses reveal squamous survival with osimertinib in untreated. N Engl J Med transformation and off-target alterations as early 382, 41–50. resistance mechanisms to first-line osimertinib in 89 Soria JC, Ohe Y, Vansteenkiste J, Reungwetwattana EGFR-mutant lung cancer. Clin Cancer Res 26, 2654– T, Chewaskulyong B, Lee KH, Dechaphunkul A, 2663. Imamura F, Nogami N, Kurata T et al. (2018) 99 Yang Z, Yang N, Ou Q, Xiang Y, Jiang T, Wu X, Osimertinib in untreated EGFR-mutated advanced Bao H, Tong X, Wang X, Shao YW et al. (2018) non-small-cell lung cancer. N Engl J Med 378, 113– Investigating novel resistance mechanisms to third- 125. generation EGFR tyrosine kinase inhibitor osimertinib 90 Tomoshige K, Guo M, Tsuchiya T, Fukazawa T, in non-small cell lung cancer patients. Clin Cancer Res Fink-Baldauf IM, Stuart WD, Naomoto Y, Nagayasu 24, 3097–3107. T & Maeda Y (2018) An EGFR ligand promotes 100 Lin CC, Shih JY, Yu CJ, Ho CC, Liao WY, Lee JH, EGFR-mutant but not KRAS-mutant lung cancer Tsai TH, Su KY, Hsieh MS, Chang YL et al. (2018) in vivo. Oncogene 37, 3894–3908. Outcomes in patients with non-small-cell lung cancer 91 Kim EY, Cho EN, Park HS, Hong JY, Lim S, Youn and acquired Thr790Met mutation treated with JP, Hwang SY & Chang YS (2016) Compound EGFR osimertinib: a genomic study. Lancet Respir Med 6, mutation is frequently detected with co-mutations of 107–116. actionable genes and associated with poor clinical 101 Beagan JJ, Bach S, van Boerdonk RA, van Dijk E, outcome in lung adenocarcinoma. Cancer Biol Ther 17, Thunnissen E, van den Broek D, Weiss J, Kazemier 237–245. G, Pegtel DM, Bahce I et al. (2020) Circulating tumor 92 Chic N, Mayo-de-Las-Casas C & Reguart N (2017) DNA analysis of EGFR-mutant non-small cell lung Successful treatment with gefitinib in advanced non- cancer patients receiving osimertinib following small cell lung cancer after acquired resistance to previous tyrosine kinase inhibitor treatment. Lung osimertinib. J Thorac Oncol 12, e78–e80. Cancer 145, 173–180. 93 Niederst MJ, Hu H, Mulvey HE, Lockerman EL, 102 Cheng Y, Ma L, Liu Y, Zhu J, Xin Y, Liu X, Wang Garcia AR, Piotrowska Z, Sequist LV & Engelman JA Y, Zhang T, Yang C, Wang S et al. (2020) (2015) The allelic context of the C797S mutation Comprehensive characterization and clinical impact of acquired upon treatment with third-generation EGFR concomitant genomic alterations in EGFR-mutant inhibitors impacts sensitivity to subsequent treatment NSCLCs treated with EGFR kinase inhibitors. Lung strategies. Clin Cancer Res 21, 3924–3933. Cancer 145, 63–70. 94 Thress KS, Paweletz CP, Felip E, Cho BC, Stetson D, 103 Shaw AT, Bauer TM, de Marinis F, Felip E, Goto Y, Dougherty B, Lai Z, Markovets A, Vivancos A, Liu G, Mazieres J, Kim DW, Mok T, Polli A et al. Kuang Y et al. (2015) Acquired EGFR C797S (2020) First-line lorlatinib or crizotinib in advanced. N mutation mediates resistance to AZD9291 in non-small Engl J Med 383, 2018–2029. cell lung cancer harboring EGFR T790M. Nat Med 104 Dagogo-Jack I, Rooney M, Lin JJ, Nagy RJ, Yeap 21, 560–562. BY, Hubbeling H, Chin E, Ackil J, Farago AF, Hata 95 Bersanelli M, Minari R, Bordi P, Gnetti L, Bozzetti C, AN et al. (2019) Treatment with next-generation ALK Squadrilli A, Lagrasta CA, Bottarelli L, Osipova G, inhibitors fuels plasma. Clin Cancer Res 25, 6662– Capelletto E et al. (2016) L718Q mutation as new 6670. mechanism of acquired resistance to AZD9291 in 105 Recondo G, Bahcall M, Spurr LF, Che J, Ricciuti B, EGFR-mutated NSCLC. J Thorac Oncol 11, e121– Leonardi GC, Lo YC, Li YY, Lamberti G, Nguyen T e123. et al. (2020) Molecular mechanisms of acquired 96 Starrett JH, Guernet AA, Cuomo ME, Poels KE, resistance to MET tyrosine kinase inhibitors in VanAlderwereltVan Rosenburgh IK, Nagelberg A, patients with MET exon 14-mutant NSCLC. Clin Farnsworth D, Price KS, Khan H et al. (2020) Drug Cancer Res 26, 2615–2625. sensitivity and allele specificity of first-line osimertinib 106 Paik PK, Felip E, Veillon R, Sakai H, Cortot AB, resistance. Cancer Res 80, 2017–2030. Garassino MC, Mazieres J, Viteri S, Senellart H, Van 97 Brown BP, Zhang YK, Westover D, Yan Y, Qiao H, Meerbeeck J et al. (2020) Tepotinib in non-small-cell Huang V, Du Z, Smith JA, Ross JS, Miller VA et al. lung cancer with MET exon 14 skipping mutations. N (2019) On-target resistance to the mutant-selective Engl J Med 383, 931–943. EGFR inhibitor osimertinib can develop in an allele- 107 Drilon A, Clark JW, Weiss J, Ou SI, Camidge DR, specific manner dependent on the original EGFR- Solomon BJ, Otterson GA, Villaruz LC, Riely GJ, activating mutation. Clin Cancer Res 25, 3341–3351. Heist RS et al. (2020) Antitumor activity of crizotinib Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of 1681 Federation of European Biochemical Societies. ctDNA in lung cancer detection and monitoring M. Filipska and R. Rosell in lung cancers harboring a MET exon 14 alteration. 117 BESPOKE study of ctDNA guided immunotherapy. Nat Med 26, 47–51. https://ClinicalTrials.gov/show/NCT04761783 108 Kunimasa K, Hirotsu Y, Miyashita Y, Goto T, 118 Gainor JF, Shaw AT, Sequist LV, Fu X, Azzoli CG, Amemiya K, Mochizuki H, Samamoto I, Ohki T, Piotrowska Z, Huynh TG, Zhao L, Fulton L, Schultz Oyama T, Honma K et al. (2020) Multiregional KR et al. (2016) EGFR mutations and ALK sequence revealed SMARCA4 R1192C mutant clones rearrangements are associated with low response rates acquired EGFR C797S mutation in the metastatic site to PD-1 pathway blockade in non-small cell lung of an EGFR-mutated NSCLC patient. Lung Cancer cancer: a retrospective analysis. Clin Cancer Res 22, 148, 28–32. 4585–4593. 109 Yang JC, Camidge DR, Yang CT, Zhou J, Guo R, 119 Skoulidis F, Byers LA, Diao L, Papadimitrakopoulou Chiu CH, Chang GC, Shiah HS, Chen Y, Wang CC VA, Tong P, Izzo J, Behrens C, Kadara H, Parra ER, et al. (2020) Safety, efficacy, and pharmacokinetics of Canales JR et al. (2015) Co-occurring genomic almonertinib (HS-10296) in pretreated patients With alterations define major subsets of KRAS-mutant lung EGFR-mutated advanced NSCLC: a multicenter, adenocarcinoma with distinct biology, immune open-label, phase 1 trial. J Thorac Oncol 15, 1907– profiles, and therapeutic vulnerabilities. Cancer Discov 1918. 5, 860–877. 110 ctDNA guiding treatment after almonertinib induction 120 Chen K, Liu J, Liu S, Xia M, Zhang X, Han D, Jiang therapy for EGFRm+ NSCLC in the MDT diagnostic Y, Wang C & Cao X (2017) Methyltransferase model. https://ClinicalTrials.gov/show/NCT04841811 SETD2-mediated methylation of STAT1 is critical for 111 Adachi Y, Ito K, Hayashi Y, Kimura R, Tan TZ, interferon antiviral activity. Cell 170, 492–506.e414. Yamaguchi R & Ebi H (2020) Epithelial-to- 121 Hanna NH, Temin S & Masters G (2020) Therapy for mesenchymal transition is a cause of both intrinsic and stage IV non-small-cell lung cancer without driver acquired resistance to KRAS G12C inhibitor in alterations: ASCO and OH (CCO) joint guideline KRAS G12C-mutant non-small cell lung cancer. Clin update summary. JCO Oncol Pract 16, e844–e848. Cancer Res 26, 5962–5973. 122 Anagnostou V, Niknafs N, Marrone K, Bruhm DC, 112 Chaib I, Karachaliou N, Pilotto S, Codony Servat J, White JR, Naidoo J, Hummelink K, Monkhorst K, Cai X, Li X, Drozdowskyj A, Servat CC, Yang J, Hu Lalezari F, Lanis M et al. (2020) Multimodal genomic C et al. (2017) Co-activation of STAT3 and YES- features predict outcome of immune checkpoint associated protein 1 (YAP1) pathway in EGFR- blockade in non-small-cell lung cancer. Nat Cancer 1, mutant NSCLC. J Natl Cancer Inst 109, djx014. 99–111. https://doi.org/10.1093/jnci/djx014 123 Wan JCM, Massie C, Garcia-Corbacho J, Mouliere F, 113 Ito M, Codony-Servat C, Karachaliou N & Rosell R Brenton JD, Caldas C, Pacey S, Baird R & Rosenfeld (2019) Targeting PKCι-PAK1 in EGFR-mutation N (2017) Liquid biopsies come of age: towards positive non-small cell lung cancer. Transl Lung implementation of circulating tumour DNA. Nat Rev Cancer Res 8, 667–673. Cancer 17, 223–238. 114 Karachaliou N, Chaib I, Cardona AF, Berenguer J, 124 Gimenez-Capitan A, Bracht J, Garcıa JJ, Jordana- Bracht JWP, Yang J, Cai X, Wang Z, Hu C, Ariza N, Garcıa B, Garz on M, Mayo-de-Las-Casas C, Drozdowskyj A et al. (2018) Common co-activation of Viteri-Ramirez S, Martinez-Bueno A, Aguilar A et al. AXL and CDCP1 in EGFR-mutation-positive non- (2021) Multiplex detection of clinically relevant smallcell lung cancer associated with poor prognosis. mutations in liquid biopsies of cancer patients using a EBioMedicine 29, 112–127. hybridization-based platform. Clin Chem 67, 554–563. 115 El Kadi N, Wang L, Davis A, Korkaya H, Cooke A, 125 Wan JCM, Heider K, Gale D, Murphy S, Fisher E, Vadnala V, Brown NA, Betz BL, Cascalho M, Mouliere F, Ruiz-Valdepenas A, Santonja A, Morris Kalemkerian GP et al. (2018) The EGFR T790M J, Chandrananda D et al. (2020) ctDNA monitoring mutation is acquired through AICDA-mediated using patient-specific sequencing and integration of deamination of 5-methylcytosine following TKI variant reads. Sci Transl Med 12, eaaz8084. https:// treatment in lung cancer. Cancer Res 78, 6728–6735. doi.org/10.1126/scitranslmed.aaz8084 116 Heidrich I, Ackar L, Mossahebi Mohammadi P & 126 Newsome RC & Jobin C (2020) Microbiome-derived Pantel K (2021) Liquid biopsies: potential and liquid biopsy: new hope for cancer screening? Clin challenges. Int J Cancer 148, 528–545. Chem 67, 463–465. 1682 Molecular Oncology 15 (2021) 1667–1682 ª 2021 The Authors. Molecular Oncology published by John Wiley & Sons Ltd on behalf of Federation of European Biochemical Societies.