Lifespring Cancer
Treatment Center

Clinical Utility of Cell-Free DNA for the Detection of ALK Fusions and Genomic Mechanisms of ALK Inhibitor Resistance in Non–Small Cell Lung Cancer

By Caroline E. McCoach, Collin M. Blakely, Kimberly C. Banks, Benjamin Levy, Ben M. Chue, Victoria M. Raymond, Anh T. Le, Christine E. Lee, Joseph Diaz, Saiama N. Waqar, William T. Purcell, Dara L. Aisne



Patients with advanced non–small cell lung cancer (NSCLC) whose tumors harbor anaplastic lymphoma kinase (ALK) gene fusions benefit from treatment with ALK inhibitors (ALKi). Analysis of cell-free circulating tumor DNA (cfDNA) may provide a noninvasive way to identify ALK fusions and actionable resistance mechanisms without an invasive biopsy.

Patients and Methods

The Guardant360 (G360; Guardant Health) deidentified database of NSCLC cases was queried to identify 88 consecutive patients with 96 plasma-detected ALK fusions. G360 is a clinical cfDNA next-generation sequencing (NGS) test that detects point mutations, select copy number gains, fusions, insertions, and deletions in plasma.


Identified fusion partners included EML4 (85.4%), STRN (6%), and KCNQ, KLC1, KIF5B, PPM1B, and TGF (totaling 8.3%). Forty-two ALK-positive patients had no history of targeted therapy (cohort 1), with tissue ALK molecular testing attempted in 21 (5 negative, 5 positive, and 11 tissue insufficient). Follow-up of 3 of the 5 tissue-negative patients showed responses to ALKi. Thirty-one patients were tested at known or presumed ALKi progression (cohort 2); 16 samples (53%) contained 1 to 3 ALK resistance mutations. In 13 patients, clinical status was unknown (cohort 3), and no resistance mutations or bypass pathways were identified. In 6 patients with known EGFR-activating mutations, an ALK fusion was identified on progression (cohort 4; 4 STRN, 1 EML4; one both STRN and EML4); five harbored EGFR T790M.


In this cohort of cfDNA-detected ALK fusions, we demonstrate that comprehensive cfDNA NGS provides a noninvasive means of detecting targetable alterations and characterizing resistance mechanisms on progression. Clin Cancer Res; 24(12); 2758–70. ©2018 AACR.

Translational Relevance

The successful treatment of patients with ALK-positive non–small cell lung cancer and identification of resistance mechanisms to targeted therapy are predicated on identifying genetic alterations in tumor cells. However, tumor tissue is not always available. Our data demonstrate that comprehensive cfDNA NGS testing can often noninvasively detect targetable alterations in newly diagnosed patients as well as resistance mutations and possible bypass pathways in patients with progression on targeted therapy. Additionally, we demonstrate the utility of cfDNA to provide a comprehensive view of the diversity and complexity of resistance mechanisms in a heterogeneous tumor cell population.


The identification and targeting of oncogenic drivers such as anaplastic lymphoma kinase (ALK), epidermal growth factor receptor (EGFR), and v-ros1 (ROS1) have had a dramatic impact on the treatment of advanced non–small cell lung cancer (NSCLC; refs. 14). Patients whose tumors harbor ALK gene fusions demonstrate significant clinical benefit from treatment with ALK inhibitors (ALKi); however, their cancer ultimately progresses (58). Repeat tumor biopsy on progression has been helpful in determining the optimal subsequent line of treatment in patients receiving oncogene-targeted therapy and is now recommended in the National Comprehensive Cancer Network (NCCN) Clinical Practice Guidelines for patients with EGFR-mutant NSCLC (9). Multiple next-generation ALKi are approved by the FDA but have differential sensitivity profiles with respect to ALK kinase domain mutations, suggesting that genomic reprofiling after failure of first-line ALKi may have a role for patients with ALK-mutant NSCLC as well (10, 11).

Sampling progressing tumor lesion(s) to identify resistance mechanisms is the standard strategy to evaluate mechanisms of drug resistance; however, this is often complicated by tumors that are inaccessible to biopsy or by tissue that is of insufficient quantity or quality to perform molecular testing. As many as 25% of tumor biopsies are inadequate or insufficient for molecular analysis (12). Additionally, sampling of a single lesion may not provide an accurate representation of the tumor genomic landscape due to tumor heterogeneity (1315). Finally, rebiopsy presents a small risk of serious complications such as pneumothorax, bleeding, or infection (16, 17).

Next-generation sequencing (NGS) of cell-free circulating tumor DNA (cfDNA) can provide a noninvasive method of obtaining and evaluating tumor DNA in patients with cancer. cfDNA-based diagnostics can identify oncogenes at initial diagnosis when tissue samples are insufficient or unobtainable, which can guide effective first-line therapy and can identify actionable resistance mechanisms at disease progression. Prior work has demonstrated the feasibility of using "hotspot" and comprehensive cfDNA profiling for identification of activating mutations and therapeutic resistance mechanisms in EGFR-mutated NSCLC (1824). However, its utility in evaluating ALK gene fusions has not been assessed on a large scale. Small studies, such as a recent report that identified two cfDNA ALK fusions in a cohort of 102 patients show that the detection of these alterations is feasible, albeit more technically challenging than somatic mutation detection, while a longitudinal evaluation of 22 patients was able to detect ALK fusions in 86% at progression (18, 23, 2527). Additionally, although numerous tumor-tissue–based studies have demonstrated mechanisms of resistance to ALKi, the use of cfDNA for the evaluation of drug resistance mechanisms in ALK-positive NSCLC has not been published.

In this study, we performed a survey of a laboratory cohort of ALK-positive patients whose cfDNA was assessed using the Guardant360 (G360; Guardant Health) assay to determine the clinical utility of plasma-based comprehensive genomic profiling for the detection of ALK fusions at diagnosis and for the evaluation of resistance mechanisms following disease progression on ALKi.

Patients and Methods

Sample identification

The G360 deidentified database of submitted cases with a reported diagnosis of NSCLC was queried to identify consecutive patients whose cfDNA contained ALK fusions or ALK kinase domain mutations reported out between February 2015 and November 2016. Information provided with sample submission was completed by the ordering provider and included age, sex, any accompanying tissue data, prior therapy data, and clinical status, when available. This information was abstracted to classify the samples into one of four cohorts. This research is approved by Quorum Institutional Review Board (IRB) for the generation of deidentified data sets for research purposes. For select patients, additional detail of treatments, outcomes, and tissue biopsy results were obtained from the treating physician as per local IRB guidelines.

cfDNA isolation and sequencing

cfDNA for the G360 panel was isolated as previously described at Guardant Health (Redwood City, CA; refs. 20, 21). The G360 panel is a CLIA-certified, College of American Pathologists (CAP)-accredited, New York State Department of Health (NYSDOH)-approved test that detects point mutations in up to 70 genes as well as copy number amplifications (CNA) in 18 genes, fusions in 6 genes, and small insertions or deletions (indels) in 3 genes (Table 1).

Table 1. Guardant360 70-gene panel
Point mutations (SNVs) with complete or critical exon coverage in 70 genes*
  • NOTE: G360 is a CLIA-laboratory cfDNA test that detects point mutations in 70 genes and select amplifications (18 genes), fusions (6 genes), and small indels (3 genes).
  • *Boldface indicates complete exon coverage.
Amplifications (CNAs; 18 genes) Fusions (6 genes) Indels (3 genes)

Following isolation of cfDNA, 5 to 30 ng of DNA was subjected to oligonucleotide barcoding for preparation of a digital sequencing library. This library was amplified and then enriched for the target genes using biotinylated custom baits. Each of the 70 cancer-related genes was then paired-end sequenced on an Illumina HiSeq 2500. This sequencing covered 146,000 base pairs, and each base was sequenced at average coverage depth of 10,000×. After sequencing, algorithmic reconstruction of the digitized sequencing signals was used to reconstruct the cfDNA fragments. Analytic and clinical validation has been previously reported (20, 28). The molecular barcoding and alignment allow sensitive detection of cfDNA fusion events, detected by merging overlapping paired-end reads and forming a sequenced cfDNA molecule representation; this is followed by alignment and mapping to the original sequence (28). Specific reporting thresholds were determined by retrospective and training set analyses.

The Illumina sequencing reads were mapped to the hg19/GRCh37 human reference sequence, and genomic alterations in cfDNA were identified from the sequencing data by Guardant Health's proprietary bioinformatics algorithms. The algorithms quantify the absolute number of unique DNA fragments at a given nucleotide position, thus enabling cfDNA to be measured as a quantitative percentage of the total cfDNA (which is primarily germline cfDNA with a small amount of tumor cfDNA). The mutant allele frequency (MAF) for a given somatic mutation was calculated as the fraction of cfDNA molecules harboring that mutation divided by the total number of unique cfDNA molecules mapping to the position of the mutation and was reported as %cfDNA. The reportable range for single-nucleotide variants (SNV), indels, fusions, and CNAs in cfDNA by the G360 assay is ≥0.04%, ≥0.02%, ≥0.04%, and ≥2.12 copies, respectively (28). Plasma copy number of 2.4 is the 50th percentile in the Guardant Health database and reported as 2+, and >4.0 copies is the 90th percentile and reported as 3+.

Somatic mutation testing in tumor tissue

Tissue evaluation for ALK fusions and EGFR mutations was performed at the discretion of the patient's physician during standard of care disease management. Information regarding the specific methods of tissue testing was abstracted from the records submitted with the G360 clinical order and was not available for all patients.

RNA-based tissue NGS

Anchored Multiplex PCR-based enrichment and library preparation, examining RNA from selected regions of targeted genes in patient C4-3, was carried out using the FusionPlex Solid Tumor sequencing panel (ArcherDx, Inc.) at the University of Colorado Molecular Correlates Laboratory (CMOCO). Bioinformatics analysis was carried out using version-controlled Archer Analysis (

ALK IHC evaluation

IHC evaluation for ALK protein expression in patient C4-3 was performed at CMOCO using ALK D5F3 antibody (Ventana Medical Systems), according to the manufacturer's instructions.


During the study period, samples were received from 8,744 unique patients with a diagnosis of NSCLC with 7,852 patients having cfDNA detected (89.8%). A total of 91 consecutive patients were identified who met the inclusion criteria (1.2%). Eighty-eight patients and 96 cfDNA-detected ALK fusions were identified across 60 institutions, both within the United States and internationally. An additional 3 patients (3 samples) were identified with an ALK resistance mutation but no reported ALK fusion in the cfDNA. Based on data provided upon sample submission, patients were separated into four cohorts (Fig. 1). Cohort 1 contained 42 patients with new discovery of an ALK fusion (new diagnosis or prior diagnosis with new ALK fusion finding). Per clinical data provided at G360 order, patients in this cohort were newly diagnosed (N = 23), or had been treated with a nontargeted agent following ALK tissue results that were negative/quantity not sufficient (N = 7), or had not been exposed to ALKi therapy, either by clinician report (N = 4), or by absence of known ALKi resistance mechanisms of G360 (N = 8). Cohort 2 consisted of 31 patients (and 34 samples) with known or presumed ALK-positive NSCLC disease progression. Patients in this cohort either had clinical information provided and known prior ALKi therapy (N = 18) or had a co-occurring resistance alteration, known to develop after treatment with ALKi therapy (N = 13). Cohort 3 included 13 patients whose samples were submitted without additional clinical information and therefore were unable to be distributed into other cohorts, and cohort 4 consisted of 6 patients with a prior EGFR mutation–positive lung cancer, treated with anti-EGFR–targeted therapy, who were found to have an ALK fusion by G360. Patient clinical characteristics are shown in Table 2. Each cohort and the entire patient group had an equal distribution of male/female patients. In the entire patient group, the mean age at cfDNA collection was 54 years (range, 27–84). In the 96 cfDNA ALK fusions, the majority were to EML4 (85.4%), then STRN (6.3%), followed by KCNQ, KLC1, KIF5B, PPM1B, and TGF (totaling 8.3% across the five). ALK fusions were not identified by cfDNA in 3 patients with an ALK resistance mutation identified in cfDNA (3 of 91; 3.3%). All 3 patients (3 samples) were in cohort 2 (previously treated with an ALKi). In two additional cohort 2 patients with longitudinal samples (C2-1 and C2-2), the ALK fusion was identified in one of their serial samples, but not the other.

Figure 1. CONSORT diagram.
Figure 1: Consort Diagram
Table 2. Patient demographics
  Cohort 1 (new Dx or new ALK Dx) Cohort 2 (ALK TKI progression) Cohort 3 (unknown clinical status) Cohort 4 (EGFR TKI progression) All*
  • *One patient in cohorts 1 and 2 counted once in "All".
  • **One patient in cohort 3 had EML4 and STRN–ALK fusions.
Patients (N) 42 31 13 6 91
Female 22 (52%) 15 (48%) 6 (46%) 3 (50%) 46 (51%)
Male 20 (48%) 16 (52%) 7 (54%) 3 (50%) 45 (49%)
Age, years
Average (range) 54.6 (27–84) 50.7 (27–73) 58.6 (46–82) 61.2 (43–71) 54 (27–84)
cfDNA-detected fusion (by sample)
EML4–ALK 40 24 13 5** 82
STRN–ALK 6** 6
KLC1–ALK 3 1 4
Not detected 5 5
Total 42 34 14 11 101

Newly diagnosed ALK fusion–positive NSCLC or new detection of ALK fusion not previously known (cohort 1)

The genomic landscape of cohort 1 is illustrated in Fig. 2. The ALK mutation status in tumor tissue was known for 10 of the 42 patients with 5 patients ALK positive in tissue by fluorescence in situ hybridization (FISH) and 5 patients ALK negative in tissue (4 by FISH and 1 by NGS). In 11 patients (26%) the tissue sample was insufficient to test for ALK status; thus, the ALK fusion was only identified by cfDNA. In the remaining 21 patients, ALK status in tissue was not provided.

Figure 2. Cohort 1 (newly identified ALK fusion) genomic landscape. Individual patient results and cfDNA identified alterations for cohort 1, newly identified ALK fusions. Tumor tissue ALK status was known for 21 (50%) of cohort 1 cases: 5 negative (NEG), 5 positive cases (POS), 11 insufficient tissue to perform analysis or unable to obtain tissue for analysis (QNS). The remainder of the samples had tissue status that was unknown (Unk). Alterations identified are in the following columns and denoted by color: blue shades, fusion; green, RAS/RAF/EGFR/MET variant; red, amplification; light gray, tumor suppressor/other pathway gene mutations. Asterisk (*) indicates instances in which only the reciprocal ALK–EML4 fusion was detected in cfDNA. If known, the prior systemic therapy is listed, as are the days between diagnosis and blood draw for Guardant360 (median 21 days; range, 1–1,056 days).
Figure 2: Cohort 1 Genomic Landscape

Within cohort 1, 31 patients were newly diagnosed ALK-positive NSCLC (23 treatment-naïve, 8 treatment status unknown) and 11 patients had prior treatment for NSCLC but the ALK fusion was not previously identified (4 were tissue insufficient, 3 were tissue negative, 4 tissue status not reported). Among this subgroup of patients of previously unidentified ALK fusion (N = 11), the ALK fusion was identified in cfDNA at a median of 13.5 months (range, 5–34 months) post-initial diagnosis; 9 patients received chemotherapy prior to identification of the ALK fusion, 1 patient received immunotherapy and 1 received chemotherapy and immunotherapy. Clinical follow-up was available for all 3 patients with prior negative tissue testing and prior therapy (C1-2, C1-3, and C1-4, Fig. 2). Patient C1-2 had pretreatment tissue NGS of a lung lesion obtained by CT-guided core biopsy which revealed a TP53 variant, but no other actionable mutations. Tissue was insufficient for ALK or ROS1 testing by FISH. The patient was initiated on chemotherapy and while on treatment had a repeat biopsy of the liver for additional molecular testing and FISH results were negative for ALK and ROS1 fusions. After the patient progressed, blood was procured for G360 testing. Results were positive for an EML4–ALK fusion at an MAF of 0.9%. Based on these results, the patient was started on crizotinib. Pre-crizotinib CT scan of a representative lesion and repeat imaging performed at 10 weeks demonstrated a dramatic response (Supplementary Fig. S1A and S1B). The patient remained on crizotinib for 7 months before progression.

Patient C1-3 had pretreatment tissue testing by rtPCR and FISH, which was negative for EGFR mutations and ALK fusions. Over a 3-year period, the patient was treated with chemotherapy and immunotherapy. At progression, G360 identified an EML4–ALK fusion at 0.3%. After progression on second-line chemotherapy, treatment was switched to crizotinib with a response to therapy that was still ongoing at the most recent imaging (Supplementary Fig. S1C and S1D).

Patient C1-4 had local laboratory pretreatment FISH testing of a bone metastasis that was deemed of insufficient quantity. Follow-up targeted NGS testing of a lymph node was negative for any ALK fusions or other oncogenic mutations. The patient was treated with stereotactic radiation to the brain followed by palliative chemotherapy until progression. Surgery was performed for spinal cord decompression and was followed by palliative radiation to the spine and additional sites of bony metastases, followed by immunotherapy with pembrolizumab until progression. Tissue NGS was again attempted locally and was positive for an ALK fusion. G360 ordered at the same time was also positive for the EML4–ALK fusion at 0.1%. The patient was started on crizotinib until progression at 6 months, at which time he was transitioned to alectinib and continues to have stable disease after 8 months of treatment.

Other genomic alterations, including SNVs and CNAs, were also identified in cohort 1 samples. TP53 alterations were identified in 18 of 42 patients (43%) in cohort 1. This is consistent with prior reports of the frequency of TP53 alterations in lung cancer and similar to that seen in ALK fusion–positive NSCLC (10, 29, 30). One patient had co-occurring KRAS mutations (G13D and V14I) which were observed in trans with each other. Three patients had CNAs in one or multiple genes. Patient C1-17 demonstrated an ERBB2 CNA, patient C1-27 demonstrated a BRAF and PIK3CA CNA, and patient C1-16 demonstrated CNA in BRAF, CCND1, CDK6, EGFR, KIT, MET, and PDGFRA (Fig. 2). As PDGFRA/KIT are located on chromosome 4 and BRAF/EGFR/MET/CDK6 are on chromosome 7, this may reflect aneuploidy in the tumor cell as opposed to focal gene CNA.

Known or presumed ALK fusion–positive patients whose cfDNA had been drawn at progression (cohort 2)

Cohort 2 contained 31 patients with a known or presumed ALK fusion who had received an ALKi (Fig. 3). Overall line of treatment and complete treatment history is unknown. Resistance mutations in the ALK kinase domain were detected in 16 patients (52%). The most common resistance mutations identified were G1202R (8 patients), F1174C/V/L (6 patients), and I1171T/N (5 patients).

Figure 3. Cohort 2 genomic landscape. Individual patient results are shown in rows, and cfDNA identified alterations are shown in each column. Patients C2-1, C2-2, and C2-3 had multiple progression samples collected, denoted as 1 and 2. The most recent treatment is noted in the following column, if known. The mutational allele frequency (MAF) is shown in each box for fusions, resistance mutations, and somatic mutations in alternative oncogenes. The maximum somatic alteration allele frequency for each sample is shown in the far right column. "T" denotes fusion previously detected in tissue but not detected in cfDNA at progression. Asterisk (*) indicates instances in which the reciprocal ALK–EML4 fusion was also detected in cfDNA. Color legend: blue/purple shades, fusion; orange, on-target resistance mutation; green, RAS/RAF/EGFR/MET mutation; red, amplification. Sample C2-17 is the progression sample from patient C1-26 in cohort 1.
Figure 3: Cohort 2 Genomic Landscape

In the 8 patients with a G1202R mutation, the most recent treatment was alectinib for one; ceritinib, then alectinib in a second patient; chemotherapy (unspecified) for a third; and not provided for the remainder. The MAF (0.27% and 0.14% for patients C2-1_2 and C2-3_2, respectively) is comparatively low for both patients for the G1202R mutations compared with other co-occurring mutations, consistent with more recent development (Fig. 3).

For the ALK mutation F1174C/V/L, the most recent ALK tyrosine kinase inhibitor (TKI) was known for 4 of the 6 and included ceritinib then alectinib in 2 patients, crizotinib in 1 patient, and lorlatinib in 1 patient. In the 5 patients in which I1171T/N was found, the most recent treatment was known in 2 patients, 1 patient received chemotherapy and in the second patient, two separate I1171 mutations were found at different treatment time points: (i) after treatment with crizotinib, I1171T was identified (MAF 4.71%) and (ii) after treatment with ceritinib then alectinib, I1171N was identified (MAF 0.29%). The I1171T was no longer identified at the second analysis.

Using prior treatment as a comparator, in the 9 patients who received crizotinib, 2 (22%) developed resistance mutations, 1 with a single mutation (F1174V, C2-3_1) and 1 patient with a dual mutation (G1269A and I1171T, C2-1_1). Seven patients received alectinib as their most recent treatment, and 4 (54%) demonstrated resistance mutations; 2 patients with single mutations (G1202R, C2-3_2; L1196Q, C2-7), and 2 patients with 3 mutations each (I1171N, F1174L, and G1202R in C2-1_2; F1174L, C1156Y, and D1203N in C2-2_2).

Three patients had two separate postprogression cfDNA evaluations after progression on different treatments (C2-1, C2-2, and C2-3). Interestingly, the second assessments demonstrated an entirely different complement of resistance mutations for all 3 patients (Fig. 3). The MAFs for these samples is shown demonstrating the relative frequencies of each kinase domain mutation.

In 6 patients, concurrent resistance mutations were identified; 4 patient samples demonstrated 3 mutations (C2-1_2, C2-2_2, C2-20, C2-24) and 3 patient samples demonstrated 2 concurrent mutations (C2-2_1, C2-28, C2-29). Two of the 3 patients described in the prior paragraph who had serial testing developed a different spectrum of resistance mutations in the later sample. In 5 samples, an ALK kinase domain mutation was identified in cfDNA but the ALK fusion was not detected in cfDNA despite prior tissue testing showing an ALK fusion (samples denoted by a "T" in Fig. 3).

In addition to mutations in the ALK kinase domain, multiple additional cancer-related genes demonstrated mutations or CNAs. Seven patients had a mutation in a potential alternative oncogenic driver in addition to detected ALK fusion. Specifically, 4 patients had mutations in the RAS pathway, including 2 with KRAS G12C/V (C2-8, C2-24), 1 patient with HRAS Q61L (C2-12) and 1 patient with KRAS G13C (C2-31). Three patients had individual mutations in BRAF V600E (C2-23), EGFR E330K (C2-15), or a MET splice-site mutation (C2-14). Two patients with ALK kinase domain mutations also demonstrated an activating mutation in an alternate oncogene [BRAF V600E (C2-23) and KRAS G12C/V (C2-24)] and 5 were found to have CNAs (C2-1_1/2, C2-2_2, C2-3_2, C2-25, C2-28).

Across the cohort, 8 patients demonstrated CNAs. These were primarily single gene amplifications with C2-2_2 demonstrating amplification of CCND2 and FGFR2 and patient C2-31 demonstrating amplification of EGFR, MYC, and FGFR1. Notably, ALK was shown to regulate the MYC signaling axis and together with these results suggest that MYC amplification may be able to partially bypass ALK signaling (31). Patient C2-8 demonstrated amplification in 7 genes. Similar to patient C1-16, amplified genes were clustered on the same chromosome (BRAF/EGFR/MET/CDK6 are located on chromosome 7 while CCND2/KRAS/CDK4 are located on chromosome 12); therefore, this likely represented aneuploidy as opposed to independent focal gene amplification events.

As noted above, 3 patients underwent more than one cfDNA evaluation during their disease trajectory. The shifting resistance mutation profile of 1 of these patients, C2-3, is illustrated in Fig. 4. At diagnosis, the patient's cfDNA and tissue demonstrated an EML4–ALK fusion and was noted to also have a mutation in ARID1A. The patient began treatment with crizotinib, but switched to ceritinib due to side effects. After progression on ceritinib the patient's cfDNA was again evaluated. In addition to the original fusion gene, the ALK F1174C resistance mutation was also detected. The patient was then started on alectinib to which the patient clinically responded. At progression, cfDNA was reassessed; the original fusion gene was again identified; however, the F1174C mutation was no longer identified in cfDNA, but G1202R was present.

Figure 4. NSCLC case with multiple cfDNA time points across disease trajectory. Patient C2-3: At initial diagnosis, EML4–ALK fusion was detected in cfDNA (and tissue). Crizotinib was initiated with 32% reduction in target lesion in 3 months. Treatment was switched to ceritinib due to side effects. G360 was drawn again when progression occurred on ceritinib, and the original fusion was detected along with ALK F1174V, a mutation conferring resistance to crizotinib and ceritinib. Alectinib was then initiated, and after initial response, progression was noted and G360 was drawn again: F1174V was no longer present in circulation, but G1202R was identified, a mutation conferring resistance to all FDA-approved ALKi but predicted to be sensitive to lorlatinib and brigatinib. Lorlatinib currently is available only by clinical trial.
Figure 4: NSCLC Case Disease Trajectory

Recently, Lin and colleagues and Ou and colleagues separately presented data demonstrating a difference in the development of ALK kinase domain resistance mutations depending on the specific EML4–ALK fusion variant (32, 33). We evaluated the data based on the two most common variants, EML4 exon 13 to ALK exon 20 (variant 1, n = 9) an EML4 exon 6 to ALK exon 20 (variant 3, n = 7). In cohort 2, ALK kinase domain mutations were observed in 4/9 (44.4%) of samples with variant 1 and 6/8 (75%) of samples with variant 3 (P = 0.2145). G1202 was observed in 0/9 (0%) variant 1 and 4/8 (50%) variant 3 samples. (P = 0.0186). Cohort 2 breakpoint data for each sample is listed in Supplementary Table S1.

Patients with unknown clinical status (cohort 3)

For 14 patients, we did not have sufficient clinical data to classify them into one of the other cohorts. Their molecular characteristics are illustrated in Fig. 5. Thirteen (93%) patients had fusions to EML4 and 1 patient had a KLC1–ALK fusion. Eight patients demonstrated SNVs in other gene. No CNAs or resistance mutations were identified.

Figure 5. Genomic landscape of cohort 3. Cohort 3, genomic landscape for patients with unknown clinical status. Individual patient results are shown in rows, and cfDNA-identified alterations are shown in each column. The second column lists the tissue status; POS, FISH positive; NEG, FISH negative; UNK, unknown. Color legend: blue shades, fusion; green, RAS/RAF/EGFR/MET mutation; red, amplification; light gray, tumor suppressor/other pathway gene mutations. Asterisk (*) indicates instances in which only the reciprocal ALK–EML4 fusion was detected in cfDNA.
Figure 5: Cohort 3 Genomic Landscape

Newly acquired ALK fusions as resistance mechanisms (cohort 4)

Cohort 4 contained 6 patients who were known to have an EGFR-activating mutation by tissue testing (exon 19 deletion in 4 patients and L858R in 2 patients; Fig. 6). These patients demonstrated progression on EGFR TKI after a median of 2.5 years (range, 0.5–7.4 years). Notably, cfDNA at progression on an EGFR TKI demonstrated ALK fusions: 1 patient had both an EML4–ALK and STRN–ALK fusion, 1 patient had an EML4–ALK fusion, and 4 patients had STRN–ALK fusions. The initial diagnostic biopsy tissue testing results were available for 3 patients and were positive for the same EGFR-activating mutation identified on cfDNA, EGFR T790M negative, and ALK negative by FISH in all three (C4-3, C4-4, and C4-5). Five patients also demonstrated an EGFR T790M resistance mutation (four by cfDNA and one by tissue). The most recent treatment at the time of the cfDNA draw included osimertinib for 2 patients, erlotinib for 1 patient, chemotherapy for 1 patient, nivolumab (previous afatinib and osimertinib) for 1, and unknown for the remaining patient. As shown in Fig. 6, cfDNA evaluation also identified 2 to 5 gene amplification events in 5 patients and SNVs in AKT1 (1), CDKN2A (1), PIK3CA (1), and TP53 (6).

Figure 6. Cohort 4 genomic landscape. A, Individual patient results are shown in rows, and cfDNA identified alterations are shown in each column. The EGFR mutation subtype is shown in parentheses. The most recent treatment is noted in the far-right column. The mutant allele frequency (MAF) is shown in each box for fusions, resistance mutations, and somatic mutations in alternative oncogenes. Color legend: blue shades, fusion; orange, on-target resistance mutation; green, RAS/RAF/EGFR/MET mutation; red, amplification; light gray, tumor suppressor/other pathway gene mutations Eleven ALK fusions (6 STRN–ALK, 5 EML4–ALK) were identified in 6 patients, drawn when progression occurred on an EGFR TKI (of 1450 NSCLC cases with EGFR driver mutations in the Guardant database). EGFR T790M was also present in 5 of 6 patients (4 by cfDNA in 7 tests, 1 by tissue [T]) along with multiple amplification events. In 3 patients (*), tissue testing results from initial diagnosis were available and were EGFR positive, ALK fusion negative. In patient C4-3, progression tissue biopsy, the presence of the STRN–ALK fusion was confirmed by an RNA-based NGS assay. B, Relative to the highest MAF variant in circulation, the EGFR driver mutations appear to be clonal, whereas both T790M and the fusions appear to be subclonal. This information, combined with available treatment-naïve tissue testing results, suggests that the ALK fusion events are emergent alterations.
Figure 6: Cohort 4 Genomic Landscape

MAF of variants in circulation was also evaluated for patients in cohort 4. Figure 6B displays the clonality relative to the highest MAF in the sample. In this analysis, the EGFR-activating mutation had the highest relative MAF, the EGFR resistance mutation, T790M, had a lower relative MAF, consistent with a subclonal population and the ALK fusion protein had even lower clonality, consistent with the presence of a small subclonal population. The available "ALK-negative" tissue results from initial diagnosis combined with the low subclonality of the cfDNA-detected ALK fusion at progression suggest that the fusion event is either a resistance mechanism emerging under treatment selection or represents a small subclone not detected at initial diagnosis that was selected for under EGFR-targeted therapy.

Patient C4-3 in this cohort is a 43-year-old Caucasian man who was initially diagnosed with advanced NSCLC after presenting with back pain. NGS revealed an EGFR exon 19 deletion as well as a TP53 mutation (L114*). Testing for alterations in ALK, ROS1, RET, MET amplification as well as the 26-gene TruSight NGS tumor sequencing panel was negative. FISH testing on this diagnostic biopsy sample was also negative for an ALK rearrangement. The patient was started on erlotinib and bevacizumab. After 6 months, imaging demonstrated osseous progression. Given the challenge of molecular analysis of bone biopsies, G360 was performed and demonstrated the original EGFR exon 19 deletion and TP53 mutation as well as a T790M mutation, an EGFR amplification, a CDKN2A mutation (H83Y), MYC mutation (D173A), amplification (3+), STRN–ALK fusion, and MET amplification (3+). Notably, a biopsy performed after cfDNA analysis confirmed the presence of an STRN–ALK fusion using an RNA-based NGS assay (Supplementary Fig. S2). Further, the EGFR exon 19 deletion and the T790M mutation were demonstrated by DNA-based tumor NGS, the ALK expression was confirmed by IHC (Supplementary Fig. S3), and MET amplification was confirmed by FISH (MET/CEP7 ratio 5.04). Given the MET amplification and the ALK fusion, the patient's therapy was changed to osimertinib and crizotinib with radiation to painful sites of disease in the thoracic and lumbar spine. The treatment was tolerated well and resulted in radiologically stable disease for eight months, demonstrating the clinical benefit of plasma-based NGS testing to identify resistance mechanisms and determine next lines of treatment.


In this retrospective study, we examined the efficacy of using cfDNA to (i) identify ALK fusions in plasma and (ii) identify resistance mechanisms through utilization of a targeted 70-gene cfDNA NGS test.

In cohort 1 patients (either newly diagnosed lung cancer or in whom an ALK fusion gene was not previously identified), we were able to demonstrate an ALK fusion in 16 patients who had previously been reported as tissue negative or tissue insufficient, in addition to confirming the molecular diagnosis in five patients and providing an ALK fusion diagnosis in 25 patients. Consistent with our finding, a recent tissue-based study reported a FISH false-negative rate of 35% in a cohort of 47 patients who were found to be ALK positive by NGS, suggesting that NGS-based assessment for ALK fusions may be warranted in patients with higher probability of ALK fusion and whose FISH analysis is negative (34). The importance of this is illustrated by 3 patients in our cohort who, despite a negative ALK FISH were ALK positive by cfDNA and went on to respond to ALKi treatment. Clinical data were not available for the remaining two tissue-negative but cfDNA-positive ALK fusion patients. Notably, among the 8 patients who received prior lines of non-ALKi therapy due to tissue insufficiency/false-negative results, two received immunotherapy which is known to have an inferior response to treatment in ALK positive lung cancer (35). For the 11 cases whose tissue was "quantity not sufficient" for biomarker testing, the cfDNA analysis salvaged the molecular testing by providing an oncogene result without requiring an additional biopsy. Thus, across this cohort, the use of cfDNA to complement tissue testing provided effective treatment options in these patients. Additionally, the utility of cfDNA in fusion detection is not limited to ALK fusions. A recent study of tumors with RET fusions used cfDNA to identify several patients in their cohort (36). As the current cohort of patients was selected based on a positive cfDNA ALK result, we do not have an accurate estimate for the false-negative rate. Therefore, this testing should be viewed as a rule-in versus a rule-out test.

We also explored the genomic landscape of known or presumed ALK-positive patients whose cfDNA was interrogated at the time of disease progression. In this cohort, a possible mechanism of resistance was identified in 24 of the 31 patients (77%). Recently reported cohorts of ALK-positive patients have identified similar percentages of resistance mechanisms (including kinase domain mutations, alternative oncogenic mutations, and copy number gains) in patients who have progressed on at least one type of ALKi therapy (10, 37, 38).

In evaluation of resistance mutations, 16 of 31 patients (51.6%) were identified with at least one mutation in the kinase domain. This result is similar to other studies reporting between 44% and 56% of patients with kinase domain mutations (10, 27, 37, 38). Notably, 7 of the 16 (43.8%) patients with resistance mutations in ALK demonstrated more than one mutation. This contrasts with a recently published series of ALK-positive patients who underwent tissue-based NGS after progression on second-generation ALKi in which 6 of 48 (12%) patient specimens contained compound mutations (10). The increased percentage in our cohort may reflect the nature of cfDNA, which contains tumor DNA shed from multiple tumor sites throughout the body, whereas tissue biopsy of a single site may not fully represent tumoral heterogeneity both within an individual lesion and across multiple metastatic sites (1315). Therefore, it is important to consider that cfDNA may present a more accurate picture of tumor heterogeneity and the challenges of overcoming resistance to targeted therapy given multiple complementary mechanisms of resistance.

In three patients from cohort 2 we collected two cfDNA time points during the patients' disease trajectory. In each case, the second cfDNA assessment demonstrated a complete shift in the resistance mutation spectrum, with ALK fusions and/or initial ALK resistance alterations becoming undetectable and new ALK resistance alterations appearing. Additionally, in 2 patients, the number of resistance mutations increased (C2-1 and C2-2). The shifts in the mutation spectrum likely reflect the selective pressure of different ALK targeted agents. This again illustrates the benefit of using cfDNA resistance profiles to give a picture of the complexity of resistance to targeted therapy in a heterogeneous tumor cell population.

The interpretation of MAFs in cfDNA is evolving. The MAFs for cohort 2 are shown in Fig. 3. In general, the initial driver alterations, often a TP53-inactivating mutation and ALK fusion, are at the highest MAF pretreatment, suggesting an early, truncal event. At progression on crizotinib, the ALK resistance mutation is often at lower MAF, reflecting its more recent development as a branched event. In instances in which there are multiple ALK resistance mutations identified in a single sample at progression on a given ALKi, it is interesting to consider whether the resistance mutation with the highest allele frequency indicates the mutation that is driving resistance or reflects more nuanced factors that influence tumor DNA shedding, such as location, tumor size, or blood supply. Regardless, high prevalence in blood does not necessarily equate to being the dominant driver of resistance as additional somatic mutations both within the ALK kinase domain and in other genes may shift resistance and sensitivity profiles. Additional studies need to be done to further clarify the utility of using MAF in longitudinal cfDNA interpretation.

Finally, in samples from 5 patients from cohort 2 (14.7% of cohort 2 samples), the ALK fusion was not detected, but a resistance mutation was identified in the ALK kinase domain indicating the presence of the ALK fusion despite the absence of detection by cfDNA. In 2 of these patients, the ALK fusion was detected in the cfDNA at a different time point. The sensitivity of fusion detection in cfDNA is known to be lower than that for SNVs or indels. cfDNA is highly fragmented, making it more prone to interference leaving insufficient mappable sequence to identify the fusion event [e.g., complex fusion events involving multiple partners or generation of random sequences (due to double strand break rescue gap-fill) which do not map to the human genome] and fusion molecules can be lost due to fusion hybrid capture inefficiencies. Due to these technical explanations, and known biological reasons for low detection rate (e.g., low tumor DNA shedding on treatment), as mentioned above, cfDNA should be utilized as a rule-in versus a rule-out test. In these 5 patients, the identification of the ALK resistance mutation is pathognomonic for the presence of the ALK fusion, even if the latter is present below the reportable range for the cfDNA assay.

In cohort 4, interrogation of cfDNA identified ALK fusions in 6 patients known to have EGFR-mutant NSCLC who had progressed on prior therapy. The detection of ALK fusions as a mechanism of resistance to EGFR TKI therapy has been previously reported (39). Conversely, EGFR-activating mutations have been identified as a mechanism of resistance in patients initially identified as ALK fusion positive, both in this series (patient C2-15; Fig. 3) and elsewhere (40). The incidence of emerging ALK fusions in patients treated with EGFR TKIs is unknown, but is likely infrequent. In a recently presented cohort of over 5,000 patients with advanced treatment-naïve and progressing NSCLC, tested by G360, 26.4% (N = 1,361) had detectable EGFR driver alterations. cfDNA T790M was detected in 654 patients (48%). In the current study, we reviewed the results for over 8,000 treatment-naïve and progressing patients with advanced NSCLC and identified only 6 patients with an EGFR-activating mutation and ALK fusion at progression on prior treatment (41). The MAFs demonstrate the clonality of the primary oncogenic driver (EGFR), and the subclonal populations of the T790M resistance mutation, and the ALK fusion, each with decreasing MAFs.

As illustrated in Table 2, the majority of fusion partners identified across cohort 4 patients were EML4; however, in cohort 4 we identified 5 STRN fusion events in 6 patients. The STRN–ALK fusion has been previously reported in thyroid cancer, with an increased frequency in poorly differentiated (9%) and anaplastic thyroid cancer (4%) compared with papillary thyroid cancer (1.6%; refs. 42, 43). STRN–ALK fusions have been reported in patients with NSCLC, with a recent case report describing a patient with this rare fusion event and resistance to alectinib (44). This is the first report of STRN–ALK fusions in a cohort of patients treated with EGFR TKIs. The true incidence of these STRN fusions in NSCLC as oncogenic drivers or potential therapeutic resistance alterations remains unknown, though likely very rare. The STRN gene is located on chromosome 2 and encodes a calmodulin binding protein thought to be involved in Ca2+ depending scaffolding (45). It was initially localized in neurons and its coiled-coil domain has been previously reported to lead to MAPK signaling via dimerization (42). Its high prevalence in this molecularly defined cohort may indicate a possible preferential fusion event in patients who develop ALK fusions in response to the selective pressure of an EGFR TKI. Notably, in patient C4-3′s tumor sample, this fusion was confirmed by two orthogonal methods, RNA NGS and ALK IHC. The patient example from this cohort achieved prolonged disease stabilization with dual TKI therapy. Efficacy of combination treatment with ALKi and EGFR inhibitors has been demonstrated previously in other clinical situations but has been limited by toxicity (4648). It is notable that several groups have now reported the finding of oncogene fusions involving ALK, RET, NTRK1, and FGFR3 fusions in the setting of EGFR TKI resistance and suggests that broad testing for these targetable alterations at resistance (in addition to EGFR T790M) may allow for attempts to overcome resistance using combinations of targeted agents (39, 4952).

This study has several limitations. First, this is a retrospective analysis reliant on clinical information provided on sample submission. Therefore, complete treatment history and clinical follow-up is not available (and cannot be verified) for all patients. This includes patient demographic information, type and length of prior therapies, local tissue testing modality, and prior molecular testing results both at diagnosis and progression rebiopsy. Further, there are limitations to the cfDNA platform, including the identification of multiple subclonal populations, which may not be clinically relevant to resistance. Additionally, given G360 is a clinical cfDNA assay, only ALK fusion events that occur with partners with known biologic significance are reported. Finally, in this study we identified 6 patients in cohort 2 whose ALK fusion were not identified by cfDNA, instead they were identified by the presence of the ALK resistance mutation. This reflects the complexity of fusion proteins and the fact that ALK has numerous fusion variants that may hinder identification by small fragment cfDNA analyses (53). Additionally, we are unable to estimate the true false negative rate of cfDNA in detecting ALK fusions given the database search parameters.

In conclusion, in the largest cohort of cfDNA ALK fusions reported to date, our data demonstrate that comprehensive cfDNA NGS testing is an additional tool that provides a noninvasive means of detecting targetable alterations in newly diagnosed patients, as well as resistance mutations and possible bypass pathways in patients progressing on targeted therapy. In this cohort, we were able to demonstrate the evolving and dynamic resistance profile in the longitudinally assessed patients with ALK fusions. We also describe STRN–ALK fusions as a potential emerging target upon progression in patients with EGFR-driven NSCLC. cfDNA provides a comprehensive view of the diversity and complexity of resistance mechanisms in a heterogeneous tumor cell population, highlighting the need to consider novel and combinatorial therapies in patients to help attenuate resistance.

Disclosure of Potential Conflicts of Interest

C.E. McCoach is a consultant/advisory board member for Guardant Health and Takeda. C.M. Blakely is a consultant/advisory board member for Jazz Pharmaceuticals, and reports receiving commercial research grants from Ignyta, MedImmune, Mirati, and Novartis. B. Levy is a consultant/advisory board member for AstraZeneca, Celgene, Eli Lilly, Genentech, Merck, and Takeda. A.T. Le has ownership interests (including patents) at Molecular Abbott. D.L. Aisner is a consultant/advisory board member for AbbVie and Bristol-Myers Squibb, and reports receiving commercial research grants from Genentech. R.B. Lanman has ownership interests (including patents) at Guardant Health. A.T. Shaw is a consultant/advisory board member for Ariad/Takeda, Genentech, Novartis, Pfizer, and TP Therapeutics. R.C. Doebele has ownership interests (including patents) at Rain Therapeutics, is a consultant/advisory board member for Ariad, AstraZeneca, Clovis Oncology, Ignyta, OncoMed, Pfizer, Spectrum Pharmaceuticals, Takeda, and TrovaGene, and reports receiving commercial research grants from Ignyta, Loxo Oncology and Mirati Therapeutics. No potential conflicts of interest were disclosed by the other authors.

Authors' Contributions

Conception and design
C.E. McCoach, K.C. Banks, B. Levy, B.M. Chue, V.M. Raymond, R.B. Lanman, R.C. Doebele
Development of methodology
C.E. McCoach, K.C. Banks, B.M. Chue, R.B. Lanman, R.C. Doebele
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.)
C.E. McCoach, C.M. Blakely, K.C. Banks, B. Levy, B.M. Chue, V.M. Raymond, S.N. Waqar, W.T. Purcell, D.L. Aisner, R.B. Lanman, A.T. Shaw, R.C. Doebele
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis)
C.E. McCoach, K.C. Banks, B. Levy, B.M. Chue, V.M. Raymond, A.T. Le, K.D. Davies, R.B. Lanman, A.T. Shaw, R.C. Doebele
Writing, review, and/or revision of the manuscript
C.E. McCoach, C.M. Blakely, K.C. Banks, B. Levy, B.M. Chue, V.M. Raymond, S.N. Waqar, D.L. Aisner, K.D. Davies, R.B. Lanman, A.T. Shaw, R.C. Doebele
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases)
C.E. McCoach, K.C. Banks, V.M. Raymond, C.E. Lee, J. Diaz, R.C. Doebele
Study supervision
C.E. McCoach, R.C. Doebele


This work was supported by a Career Enhancement Award from the University of Colorado Lung Cancer SPORE (funded by the NCI of the NIH grant P50CA058187). We would also like to thank Stephen Fairclough, PhD, for his assistance with the ALK breakpoint analysis.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.


Note: Supplementary data for this article are available at Clinical Cancer Research Online (

Received September 15, 2017.

Revision received January 6, 2018.

Accepted March 20, 2018.

Published first April 2, 2018.

©2018 American Association for Cancer Research.


  1. Mok TS, Wu YL, Thongprasert S, Yang CH, Chu DT, Saijo N, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361:947-57.
  2. Mitsudomi T, Morita S, Yatabe Y, Negoro S, Okamoto I, Tsurutani J, et al. Gefitinib versus cisplatin plus docetaxel in patients with non-small-cell lung cancer harbouring mutations of the epidermal growth factor receptor (WJTOG3405): an open label, randomised phase 3 trial. Lancet Oncol 2010;11:121-8.
  3. Rosell R, Carcereny E, Gervais R, Vergnenegre A, Massuti B, Felip E, et al. Erlotinib versus standard chemotherapy as first-line treatment for European patients with advanced EGFR mutation-positive non-small-cell lung cancer (EURTAC): a multicentre, open-label, randomised phase 3 trial. Lancet Oncol 2012;13:239-46.
  4. Shaw AT, Ou SH, Bang YJ, Camidge DR, Solomon BJ, Salgia R, et al. Crizotinib in ROS1-rearranged non-small-cell lung cancer. N Engl J Med 2014;371:1963-71.
  5. Shaw AT, Yeap BY, Solomon BJ, Riely GJ, Gainor J, Engelman JA, et al. Effect of crizotinib on overall survival in patients with advanced non-small-cell lung cancer harbouring ALK gene rearrangement: a retrospective analysis. Lancet Oncol 2011;12:1004-12.
  6. Camidge DR, Bang YJ, Kwak EL, Iafrate AJ, Varella-Garcia M, Fox SB, et al. Activity and safety of crizotinib in patients with ALK-positive non-small-cell lung cancer: updated results from a phase 1 study. Lancet Oncol 2012;13:1011-9.
  7. Shaw AT, Kim DW, Nakagawa K, Seto T, Crinó L, Ahn MJ, et al. Crizotinib versus chemotherapy in advanced ALK-positive lung cancer. N Engl J Med 2013;368:2385-94.
  8. Solomon BJ, Mok T, Kim DW, Wu YL, Nakagawa K, Mekhail T, et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. N Engl J Med 2014;371:2167-77.
  9. Ettinger DS, Wood DE, Aisner DL, Akerley W, Bauman J, Chirieac LR, et al. Non-small cell lung cancer, version 5.2017, NCCN clinical practice guidelines in oncology. J Natl Compr Canc Netw 2017;15:504-35.
  10. Gainor JF, Dardaei L, Yoda S, Friboulet L, Leshchiner I, Katayama R, et al. Molecular mechanisms of resistance to first- and second-generation ALK inhibitors in ALK-rearranged lung cancer. Cancer Discov 2016;6:1118-33.
  11. Shaw AT, Friboulet L, Leshchiner I, Gainor JF, Bergqvist S, Brooun A, et al. Resensitization to crizotinib by the lorlatinib ALK resistance mutation L1198F. N Engl J Med 2016;374:54-61.
  12. Meric-Bernstam F, Brusco L, Shaw K, Horombe C, Kopetz S, Davies MA, et al. Feasibility of large-scale genomic testing to facilitate enrollment onto genomically matched clinical trials. J Clin Oncol 2015;33:2753-62.
  13. Piotrowska Z, Drapkin B, Engelman JA, Nagy RJ, Lanman RB, Sequist LV. Plasma T790M result alters treatment options in a previously T790 wild-type EGFR-mutant lung cancer. J Thorac Oncol 2016;11:e95-7.
  14. Goyal L, Saha SK, Liu LY, Siravegna G, Leshchiner I, Ahronian LG, et al. Polyclonal secondary FGFR2 mutations drive acquired resistance to FGFR inhibition in patients with FGFR2 fusion-positive cholangiocarcinoma. Cancer Discov 2017;7:252-63.
  15. Jamal-Hanjani M, Wilson GA, McGranahan N, Birkbak NJ, Watkins TBK, Veeriah S, et al. Tracking the evolution of non-small-cell lung cancer. N Engl J Med 2017;376:2109-21.
  16. Lokhandwala T, Bittoni MA, Dann RA, D'Souza AO, Johnson M, Nagy RJ, et al. Costs of diagnostic assessment for lung cancer: a medicare claims analysis. Clin Lung Cancer 2017;18:e27-e34.
  17. Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, et al.National Lung Screening Trial Research T, Aberle DR, Adams AM, Berg CD, Black WC, Clapp JD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med 2011;365:395-409.
  18. Thompson JC, Yee SS, Troxel AB, Savitch SL, Fan R, Balli D, et al. Detection of therapeutically targetable driver and resistance mutations in lung cancer patients by next-generation sequencing of cell-free circulating tumor DNA. Clin Cancer Res 2016;22:5772-82.
  19. Schwaederle M, Husain H, Fanta PT, Piccioni DE, Kesari S, Schwab RB, et al. Detection rate of actionable mutations in diverse cancers using a biopsy-free (blood) circulating tumor cell DNA assay. Oncotarget 2016;7:9707-17.
  20. Lanman

Originally published on March 29, 2018, in American Association for Cancer Research - Clinical Cancer Research.