Duration of targeted therapy in advanced non-small cell lung cancer patients identified by circulating tumor DNA analysis

Abstract

Background:

Outcomes of therapy targeting molecular driver alterations detected in advanced non-small cell lung (NSCLC) using circulating tumor DNA (ctDNA) have not been widely reported in patients who are targeted therapy naïve.

Methods:

We performed a multicenter retrospective review of patients with unresectable stage IIIB-IV NSCLC who received matched therapy after a targetable driver alteration was identified using a commercial ctDNA assay through usual clinical care. Eligible patients must not have received targeted therapy prior to ctDNA testing (prior chemotherapy or immunotherapy was permitted). Kaplan-Meier analysis was used to estimate median duration of targeted therapy. Patients still on targeted therapy were censored at last follow-up.

Results:

76 patients met inclusion criteria. Median age of diagnosis of NSCLC was 64.5 years (range 31–87y), 67% were female, 74% were never smokers, and 97% had adenocarcinoma histology. Twenty-one patients (28%) received systemic treatment prior to targeted therapy, including chemotherapy (n=17), immunotherapy (5), and/or a biologic (4). Thirty-three patients (43%) remain on targeted therapy at the time of data analysis. Median time on targeted therapy was similar to what has been reported for tissue-detected oncogenic driver mutations in the targeted therapy naïve setting.

Conclusions:

Patients with ctDNA-detected drivers had durable time on targeted therapy. These treatment outcomes data compliment previous studies that have shown enhanced targetable biomarker discovery rates and high tissue concordance of ctDNA testing when incorporated at initial diagnosis of NSCLC. Identification of NSCLC driver mutations using well-validated ctDNA assays can be used for clinical decision-making and targeted therapy assignment.

MicroAbstract

Identifying targetable genomic driver alterations is critical for optimal treatment selection in advanced non-small cell lung cancer (NSCLC). Previous studies have shown that ctDNA testing increases the identification of informative biomarkers at initial diagnosis. In this study, we show that patients with drivers detected by a validated ctDNA next generation sequencing assay had durable times on targeted therapy, comparable to what has been reported for tissue-detected drivers in the targeted therapy naïve setting.

Introduction

The identification of patients with advanced non-small cell lung cancer (NSCLC) with therapeutically targetable genomic alterations is critical for optimal therapy selection. Guidelines issued by the National Comprehensive Cancer Network (NCCN) for the treatment of NSCLC and jointly by the College of American Pathologists (CAP), International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) recommend testing seven genes in NSCLC with an adenocarcinoma component: EGFR, ALK, ROS1, BRAF, ERBB2 (HER2), MET, and RET^1,2. The number of relevant genomic and other biomarkers such as PD-L1 continues to expand, and the NCCN added a recommendation for testing for NTRK fusions due to therapeutic options for tumors with these alterations^3,4.

Targeted therapy is generally the preferred first-line treatment for patients with targetable mutations in EGFR, ALK, and ROS1 who have improved outcomes with lower toxicity when given genomically-matched therapies relative to chemotherapy and tumors with these mutations are often resistant to immune checkpoint inhibition^5–17. Combination dabrafenib and trametinib is FDA-approved for the treatment of BRAF V600E advanced NSCLC and has shown efficacy in the first-line setting¹⁸. The FDA also recently approved larotrectinib and entrectinib for certain NTRK fusion positive metastatic solid tumors, including NSCLC^3,4. Furthermore, novel agents targeting exon 20 insertions in EGFR, HER2 mutations, MET exon 14 skipping or amplification, RET fusions, or KRAS G12C mutations have shown promise in NSCLC and are noted as emerging targets^19–27.

However, rates of complete tissue genotyping are low. In recent multicenter studies of real world genotyping rates in the United States, EGFR and ALK were tested in 65–83% of patients, ROS1 in 25–58%, BRAF V600E in 19–35%, and all NCCN-recommended biomarkers in 8–19%^28–30. Lack of genotyping may be due to lack of clinician order or insufficient tissue for evaluation of all guideline-recommended genes, especially given the growing number of biomarkers that must be assessed, often from limited quantities of tissue. Furthermore, the need for subsequent biopsy and/or long turnaround times for receipt of complete tissue results can delay treatment decision-making, and in some situations patients are initiated on chemotherapy and/or immunotherapy regimens while genomic test results are still pending^28,30,31. A critical emerging concern is the safety of administering targeted therapy in combination with or following immunotherapy, based on toxicity observed in these settings with targeted drugs such as osimertinib and crizotinib^32–35.

Highly sensitive and specific plasma circulating tumor DNA (ctDNA) profiling assays are clinically available and provide a means to obtain tumor genetic sequencing without an invasive tissue biopsy. In a recent multicenter prospective trial of 282 newly diagnosed patients with advanced NSCLC undergoing simultaneous physician-directed tissue genotyping and ctDNA analysis using a commercially available comprehensive assay (Guardant360®), concordance of FDA-approved targets was >98% with 100% positive predictive value for ctDNA-detected mutations²⁹. The addition of ctDNA testing to tissue analysis increased the identification of informative biomarkers by 48% with a significantly shorter turnaround time. It is expected that patients with mutations identified by ctDNA will have similar outcomes to targeted therapies as patients treated based on tissue analysis; however, given the relatively recent adoption of ctDNA analysis into clinical practice, only limited data are available using small patient populations^11,36–41. In this study, we describe the clinical outcomes of targeted therapy naïve patients with NSCLC who received matched treatment after being found to have a sensitizing alteration in EGFR, ALK, ROS1, BRAF, ERBB2, MET, or RET using a commercially available comprehensive ctDNA test.

Methods

This was a multicenter retrospective review of patients with NSCLC who received matched targeted therapy following identification of a driver alteration on a validated commercial ctDNA assay (Guardant360®, Redwood City, CA). Guardant360 is a CLIA-certified, College of American Pathologists-accredited, New York State Department of Health-approved comprehensive next-generation sequencing (NGS) test. Analytic and clinical validation have been previously reported^42,43. During the study course, this assay evaluated 54–73 genes, depending on the time the test was run, and includes assessment of single-nucleotide variants (SNV) and insertion-deletions (indel), fusions, and copy number gain in select genes.

Eligible patients were identified from one of three US-based academic medical centers and included consecutive patients with stage IIIB-IV NSCLC who 1) had a targetable mutation in EGFR, ALK, ROS1, BRAF, ERBB2, MET or RET identified by Guardant360, 2) received targeted therapy following the ctDNA testing, 3) had at least one clinical follow-up, and 4) had not received targeted therapy prior to ctDNA testing. These seven genes were selected as they were listed by professional guidelines as being an established or emerging targetable biomarker in NSCLC at the time the study was initiated. NTRK fusions became an NCCN-recommended target after the initiation of this study and were thus not included. The Guardant360 assay sequences the coding regions of all seven of these genes, and fusions known to be biologically relevant are reported. ctDNA testing was completed as part of routine clinical care between April 2014 and February 2019. Targeted therapy was administered as part of standard care using FDA-approved therapies or via a clinical trial using investigational matched targeted agents. Patients who had received targeted therapy prior to ctDNA testing were excluded in order to focus on outcomes from first targeted therapy, however prior chemotherapy or immunotherapy was permitted to improve inclusion of patients with rare drivers that may not be routinely tested at the time of diagnosis of advanced disease.

Clinical information (histology, demographics, performance status at start of therapy, records of treatment history, tissue results when available, and overall survival from time of diagnosis) was extracted from medical records from each respective institution and maintained in a de-identified database. We chose time on targeted therapy as the primary endpoint for this study given that RECIST-evaluated response data were not readily available. Additionally, we wanted to capture potential clinical benefit of targeted therapy beyond the initial progression event (e.g. patients with oligoprogressive disease who could be managed using local interventions while remaining on targeted therapy), as well as situations where patients may have been switched from one targeted therapy to another for reasons other than efficacy, such as tolerability, cost, or availability of newer more efficacious drugs.

Kaplan-Meier analysis was used to estimate median duration of targeted therapy for first targeted therapy and all sequential targeted therapies, where applicable (e.g. osimertinib following erlotinib). Patients were censored at time of targeted therapy discontinuation (evaluated as time on first targeted therapy as well as including all sequential targeted therapies). Patients still on targeted therapy at the data cutoff were censored at last follow-up. The log-log transformation was applied to the survivor function to obtain the 95% confidence interval of the median time elapsed. The relationship between the EGFR driver variant allele fraction (VAF) and time on targeted therapy was evaluated using Pearson correlation analysis as well as by the Kaplan-Meier method (data stratified by VAF ≥ 1% vs. <1% and analyzed using the log-rank univariate test).

This research was approved by each institution’s Institutional Review Board (IRB) and the Quorum IRB for the generation of deidentified data sets for research purposes.

Results

A total of 76 patients met inclusion criteria. Demographic characteristics were typical of an oncogene positive cohort, including a predominance of females (67%), adenocarcinoma histology (97%), and never smokers (74%, Table 1). Of note, 26% and 2.6% of targetable biomarker positive patients had previous history of smoking or squamous histology, respectively, therefore restricting genomic testing to non-smokers and adenocarcinoma histology only could miss patients with targetable biomarkers.

Table 1.

Cohort Characteristics

		n (%)
Median age at diagnosis (range)	64.5 years (31–87y)	76
Sex	Male	25 (33%)
	Female	51 (67%)
Smoking history	Ever	20 (26%)
	Never	56 (74%)
Stage	IIIB	3 (4%)
	IV	73 (96%)
ECOG Performance Status at the start of therapy	0	24 (32%)
	1	36 (47%)
	2	12 (16%)
	Unknown	4 (5%)
Histology	Adenocarcinoma	74 (97%)
	Squamous cell carcinoma	2 (3%)
Systemic therapy received prior to targeted therapy	None	55 (73%)
	Chemotherapy	17 (22%)
	Immune checkpoint inhibitor	5 (7%)
	Biologic	4 (5%)
Mutation	Any EGFR sensitizing mutation	48 (63%)
	EGFR L858R	20
	EGFR exon 19 deletion	24
	Other*	4
	BRAF V600E	10 (13%)
	ALK fusion	9 (12%)
	MET exon 14 skipping	4 (5%)
	ROS1 fusion	2 (3%)
	ERBB2 exon 20 insertion	2 (3%)
	RET fusion	1 (1%)
Status at last follow-up	Alive on targeted therapy	33 (43%)
	Alive, no longer on targeted therapy	13 (17%)
	Deceased	30 (40%)

^*Other EGFR sensitizing mutations included G719A, G719A+R776C, and L861Q (n=2)

The most common mutations were EGFR exon 19 deletion (n=24) or L858R (n=20) followed by BRAF V600E (n=10), however there was at least one patient with each of the seven NCCN-recommended NSCLC drivers. For most patients (55/76; 73%), targeted therapy was administered as the first-line treatment, however 21 patients received treatment prior to targeted therapy, including chemotherapy (n=17), an immune checkpoint inhibitor (n=5), and/or a biologic (n=4).

At the time of the data cutoff, 26 patients (34%) were still alive and receiving their first targeted therapy, while an additional seven (9%) were still alive and receiving a subsequent targeted therapy. Thirty (40%) were deceased and the remaining 13 (17%) were alive but had discontinued targeted therapy (Table 1). For the 33 patients still on targeted therapy, at least 6 months of follow-up data were available for all but three patients. The median follow-up from initiation of targeted therapy was 15.7 months (range 3 weeks-6.5 years).

The most common first targeted therapies were osimertinib (n=23/48) or erlotinib (n=21/48) for EGFR+, combination dabrafenib plus trametinib for BRAF V600E+ (n=10/10), alectinib (n=6/10) for ALK+, crizotinib (n=3/4) for MET+, ado-trastuzumab emtansine (n=2/2) for ERBB2+, and investigational TKIs for ROS1 or RET+. Median time on first targeted therapy could be calculated for osimertinib, erlotinib, and combination of dabrafenib and trametinib and was similar to what has been reported for tissue-detected oncogenic driver mutations (Figure 1)^18,44–46. Sixteen EGFR+ and three ALK+ patients received more than one sequential TKI (most commonly initial erlotinib followed by osimertinib, n=12). The median duration of anti-EGFR therapy including any sequential EGFR TKI was 19.5 months (95%CI;11.0, 23.3; Figure 2A). For the remaining oncogenic drivers, the overall number of patients was low, precluding estimation of median time on therapy. However individual times on matched targeted therapies were generally durable (Figures 2B–D).

Figure 1. Median time on targeted therapy for oncogene-positive advanced NSCLC identified by Guardant360.

Kaplan-Meier survival analysis showing time on first targeted therapy for erlotinib (blue), osimertinib (grey), and dabrafenib plus trametinib (purple). Patients were censored at time of targeted therapy discontinuation or at last follow-up if still on targeted therapy

Figure 2. Individual times on targeted therapy for oncogene-positive advanced NSCLC identified by Guardant360.

Swimmers’ plots of patients treated with one or more matched targeted therapies for activating driver mutations in A. EGFR, B. ALK, C. BRAF, or D. other (MET exon 14 skipping, ROS1 fusion, ERBB2 activating mutation, or RET fusion). Patients still being treated at the time of data cutoff are indicated with green arrows. Rare EGFR driver mutations are denoted with an asterisk and include L861Q (patients 5 and 12), G719A (patient 9), and G719A+R776C (patient 11).

Of the 76 patients, 58 (76%) had tissue testing for the ctDNA-detected driver. The driver mutation detected by ctDNA was also detected in tissue in 55/58 (95%) patients (53 from tissue sampled at the time of initial diagnosis of NSCLC, and two who had the driver confirmed on a subsequent tissue biopsy). The remaining 3/58 patients with tissue testing had a negative tissue result for the driver identified in ctDNA. These included one EGFR L858R and one EGFR exon 19 deletion tested in tissue using PCR, and one EML4-ALK fusion tested in tissue using FISH. Of the 18 patients without tissue testing for the ctDNA detected driver, nine patients never had tissue testing ordered for the driver identified by ctDNA, seven had insufficient quantity or quality of tissue DNA to complete oncogene analysis, and two had tissue blocks that could not be obtained from an outside institution for testing. Two of the three patients who had ctDNA-detected drivers that were tested for but not detected in tissue had good clinical outcomes on targeted therapy. The patient with an EGFR exon 19 deletion remains on first-line osimertinib after 24 months (patient 41 in Figure 2A), and the patient with an EML4-ALK fusion was treated with alectinib for approximately 1 year (patient 3 in Figure 2B). These data indicate that the tissue assay for these two patients resulted a false negative rather than the ctDNA assay resulting a false positive, as these outcomes would be unexpected in patients without a true underlying driver mutation. The third patient (patient 1 in Figure 2A) with an EGFR L858R mutation was in his 80s at diagnosis and received osimertinib for approximately 2 weeks, however opted to discontinue therapy.

We next explored whether the mutant allele fraction of the EGFR driver mutation, type of EGFR mutation (rare vs. classic), and/or any co-occurring mutations detected in ctDNA were associated with duration on EGFR TKIs. The median mutant allele fraction of the EGFR driver mutation was 2.5% (range 0.04–75%) and was not correlated with time on targeted therapy, consistent with findings from a previous study (Supplemental Figure 1)⁴⁰. Four patients had non-classic EGFR sensitizing mutations, including L861Q (patients 5 and 12 in Figure 2A), G719A (patient 9), and G719A+R776C (patient 11). Time on EGFR TKI was shorter than the median time on first TKI overall (5 and 21 weeks for the two who received first-line osimertinib and 20 and 23 weeks for the two who received first line erlotinib; Figure 2A).

Most EGFR-positive patients (77%) had additional functionally significant mutations in 21 other cancer-related genes detected in ctDNA, most commonly in TP53 (65%) or GNAS (13%). Co-existing amplification of EGFR was detected in 10% of patients. There was no obvious association between co-occurring mutations and time on EGFR TKI (Figure 3, supplemental Figure 2). Interestingly, two EGFR-positive patients had co-occurring KRAS G12 mutations. One patient had squamous cell lung cancer with an early stage synchronous colon adenocarcinoma treated surgically. In addition to the EGFR driver and KRAS alteration, this patient’s ctDNA showed two APC truncating mutations and two TP53 alterations. Notably, the mutant allele fractions of the APC and TP53 alterations were in line with the KRAS alteration, and all were more than a half-fold lower than the mutant allele fraction of the EGFR driver. As somatic APC mutations are common in colorectal cancer, the ctDNA profile may have reflected the genomic profile of both primary cancers as has been described previously⁴⁷. This patient was started on erlotinib but was discontinued after 18 weeks following convincing evidence of progression. This patient had a second Guardant360 drawn at the time of progression which showed an increase in the VAF of the EGFR L858R mutation, consistent with progressive disease, as well as mutations in TP53 and RB1. However, a tissue biopsy at progression was not available to evaluate for small cell transformation. The other patient with co-occurring EGFR and KRAS mutations was treated with osimertinib after front line carboplatin/pemetrexed combination therapy and remains on treatment with clinical benefit after 17 months with no evidence of a second primary cancer.

Figure 3. Co-occurring mutations detected in ctDNA and impact on duration of EGFR targeted therapy.

Each column represents a patient with an activating EGFR mutation (patient numbers correspond to those in Figure 1B), colored squares indicate co-occurring mutations (only mutations predicted to be functionally relevant are included, uncertain variants and synonymous variants were excluded). Patients are grouped by those who have completed targeted therapy (panel A) and those with targeted therapy ongoing (panel B). Times indicate all sequential EGFR TKIs. Data organized by time on first TKI only can be found in Supplemental Figure 2.

Discussion:

Multiple previous studies have shown the utility of ctDNA analysis in detection of clinically relevant mutations in advanced NSCLC and have demonstrated high concordance, sensitivity, and positive predictive value as compared to tissue^29,38,43. There is increasing recognition that incorporation of ctDNA testing as an initial diagnostic option can improve the proportion of patients who have complete genotyping prior to first-line therapy, especially in time-constrained or tissue limited clinical scenarios^29,38. A “blood first” paradigm, where ctDNA is used first to identify genomic biomarkers, reserving tissue for PD-L1 staining and reflexing to tissue for genomic testing if blood is uninformative, has been proposed for patients with newly diagnosed NSCLC, however one critique has been the paucity of treatment outcomes data demonstrating equivalency to tissue-detected mutations.

In this first of its kind study, we demonstrate that patients treated based on ctDNA-detected drivers had durable times on targeted therapy, with outcomes comparable to those guided by tissue testing. In patients with ctDNA-identified EGFR sensitizing mutations treated with an EGFR TKI as their first targeted therapy, the median time on erlotinib or osimertinib was 7.6 and 17.4 months, respectively. Median time on erlotinib was comparable to the median duration of erlotinib treatment in the EURTAC trial (8.2 months) and similar to that reported in two recent studies using claims data to estimate real-world duration of first-line anti-EGFR therapy in large cohorts of US patients with advanced NSCLC (Lim et al – median of 9.9 months, Hess et al - mean of ~7.7 months)^9,44,46. The median duration of total treatment exposure to osimertinib in the phase III FLAURA trial was 20.7 months, similar to our observed median duration of therapy⁷. Of note, data from FLAURA, which showed superior efficacy of osimertinib compared to other EGFR TKIs in the first-line setting, was published during the time at which some of the patients reported here were initiated on targeted therapy and patients may have been switched from another EGFR TKI for reasons other than lack of efficacy or the development of an EGFR T790M mutation. For that reason, overall time on any sequential EGFR TKI may be a better comparator in this real-world study. The median time on any sequential EGFR TKI was 19.5 months in this ctDNA detected cohort. This is comparable to the FLAURA crossover data which demonstrated a median time to EGFR TKI discontinuation of 23 months for osimertinib first or 16 months for other EGFR TKIs first⁴⁸.

Of note, there were four patients in this study who had a remarkably poor outcome on first-line osimertinib with less than 13 weeks of overall survival and less than 5 weeks on targeted therapy (patients 1, 3, 4, and 5). We did not observe evidence of an effect of the driver variant allele fraction or co-occurring mutations on duration of EGFR TKIs, however one of these patients had a rare EGFR mutation (L861Q), which may be associated with shorter time to progression on EGFR TKIs compared to patients with exon 19 deletions or L858R^7,49. A recent meta-analysis evaluated the prognostic significance of concurrent TP53 mutations on outcomes in over 1,000 patients with advanced NSCLC receiving EGFR TKIs across 11 studies⁵⁰. TP53 mutations were associated with inferior overall and progression-free survival. We observed concurrent TP53 mutations in 65% of our EGFR-positive patients, including those with durable times on targeted therapy. While we observed no obvious association with time on targeted therapy (Figure 3; Supplemental Figure 2), our small sample size of 31 EGFR/TP53 double mutant and 17 EGFR+/TP53− patients limits our ability to analyze this association in greater depth. Across multiple studies, roughly 5–10% of patients have disease that is refractory to EGFR TKIs (higher for patients with uncommon EGFR mutations) and additional research is needed to understand the underlying mechanisms of these poor responses^9,45,49,51.

While the cohort is smaller, outcomes of combination BRAF-directed therapy observed in this study (12.9 months) were favorable compared to times on first-line dabrafenib (9.0 months) and trametinib (9.5 months) reported in 36 patients with BRAF V600E mutated NSCLC in a Phase 2 non-randomized trial by Planchard et al¹⁸. We also observed generally durable times on ALK, ROS1, MET, ERBB2, and RET-directed therapies though the number of patients was too small to estimate median times on therapy. These mutations are individually rare in NSCLC (less than 2% each), are often not routinely assessed in clinical practice, and so are difficult to study in the real-world. However, our data show they can be detected in ctDNA and targeted with good clinical outcome.

This is a retrospective real-world study of clinical experience with ctDNA testing, tissue testing, and clinical decision making and has inherent limitations While we attempted to focus on patients treated with targeted therapy in the first line setting, we allowed prior non-targeted therapies in order to improve inclusion of patients with rare mutations that may not be routinely tested. Our data highlight this issue in that 27% of the cohort received non-targeted therapy first line, including five patients who received an immune checkpoint inhibitor. Inclusion of these patients may have biased our cohort to include those with higher performance status who tolerated chemotherapy and/or immune checkpoint inhibition and were still eligible for further treatment. However, these patients may have experienced toxicity or symptoms from cancer following front-line therapy and had poorer functional status, which may have resulted in shorter times on targeted therapy than what would have been observed if we had only included patients treated in the first-line setting. We also do not have an internal control group and rely on published literature to compare outcomes of this ctDNA-detected cohort to those treated based on tissue-detected drivers. Comparator groups are often derived from highly controlled clinical trials which may not reflect the same patient population as those treated in a real-world clinical practice and claims data are retrospective and may contain errors. We also used time on targeted therapy as our primary outcome measure, and therefore this study will include patients who were treated beyond progression whereas many trials utilize progression free survival as the primary outcome. We were unable to evaluate overall survival in this study given insufficient data. A separate study is needed to evaluate emerging biomarkers (for example NTRK fusions) that were not yet included in the NCCN guidelines at the time of data collection for this study.

Finally, the study design and eligibility criteria precluded estimation of the sensitivity of this ctDNA assay to detect targetable driver mutations in the first line setting. However, recent studies have addressed this question. The sensitivity of the Guardant360 assay was 80% for tissue-detected guideline-recommended biomarkers in a recent prospective trial comparing temporally matched tissue and plasma analysis in patients with newly diagnosed metastatic NSCLC²⁹. Furthermore, the addition of the assay to tissue analysis was found to increase the biomarker detection rate by 48% after accounting for patients with tissue genomic profiling that was negative, incompletely assessed, or insufficient for analysis

Despite these limitations, this study offers a description of real-world outcomes in patients with ctDNA-detected targetable genomic biomarkers, and we demonstrate that a majority of patients achieve durable times on targeted therapy, comparable to what has reported in the literature. These findings demonstrate that a well-validated, comprehensive ctDNA assay can be used for clinical decision making in advanced NSCLC.

Supplementary Material

Supp.Materials

Supplemental Figure 1. Duration of EGFR TKI as a function of the EGFR driver variant allele fraction A. Kaplan-Meier survival analysis with patients stratified by EGFR VAF ≥ 1% vs. <1% and all sequential EGFR TKIs included. B. Correlation between EGFR driver VAF and duration of targeted therapy (all sequential EGFR TKIs included). Patients with therapy ongoing are shown in green. No associations with time on therapy and EGFR driver VAF were found when the Kaplan-Meier and Pearson correlation analyses were repeated using only time on first EGFR TKI (e.g. either osimertinib or erlotinib); data not shown.

Supplemental Figure 2. Co-occurring mutations detected in ctDNA and impact on duration of EGFR targeted therapy Patients are grouped by first EGFR TKI received: erlotinib (panel A) or osimertinib (panel B). Times in green indicate patients whose targeted therapy is ongoing.

Highlights.

• Duration of targeted therapy is similar for driver mutations detected by ctDNA and tissue testing

• Neither driver allele fraction nor co-occurring mutations impacted targeted therapy duration

• ctDNA testing aids in the identification of patients eligible for targeted therapy

Clinical Practice Points.

Identifying targetable genomic driver alterations is critical for optimal treatment selection in advanced non-small cell lung cancer (NSCLC). Previous studies have shown that incorporating testing for circulating tumor DNA (ctDNA) at initial diagnosis of NSCLC can improve biomarker discovery rate and speed time to complete genotyping, however therapeutic outcomes of ctDNA-detected drivers at the time of first targeted therapy are not widely reported. In this study, we show that patients with drivers detected by a validated 54–73-gene commercial ctDNA next generation sequencing assay had durable times on targeted therapy, comparable to what has been reported for tissue-detected drivers in the targeted therapy naïve setting. We also found no obvious influence of other ctDNA findings, such as the driver variant allele fraction or presence of co-occurring mutations, on targeted therapy outcomes. These patient outcomes data complement previous studies evaluating the utility of ctDNA testing at initial diagnosis of advanced NSCLC and demonstrate that a well-validated comprehensive ctDNA assay can be used for clinical decision making in advanced NSCLC.

Abbreviations:

TKI

tyrosine kinase inhibitor;

O

osimertinib;

A

afatinib;

E

erlotinib;

G

gefitinib,

RTK

receptor tyrosine kinase;

AMP

gene copy number amplification