Liquid biopsy detects genomic drivers in NSCLC without EGFR mutations by single‐plex testing: WJOG13620L

Abstract Background Actionable tumor genomic alterations, primarily EGFR mutations, occur in nearly 70% of Japanese advanced nonsquamous non‐small cell lung cancer (NSCLC) patients. Standard assessment of tumor tissue includes rapid testing for EGFR mutations, ALK fusions and ROS1 fusions. We conducted a prospective observational study (WJOG13620L) of follow‐on next‐generation sequencing of circulating tumor DNA (ctDNA) in patients without driver alterations after EGFR testing. Methods Patients with untreated advanced (Stage IIIB–IV or relapsed) nonsquamous NSCLC without EGFR mutations according to single‐plex testing of tumor tissue, were enrolled into this study. Patients with other known driver mutations or who underwent comprehensive genomic profiling were excluded. Plasma was analyzed by Guardant360, and the primary endpoint was the proportion of patients with pathogenic gene alterations in at least one of nine genes. Results Among the 72 patients enrolled, ALK and ROS1 fusions were tested in 86.1% and 65.2%, respectively. Alterations in pre‐defined genes were detected in 21 patients (29.2%; 95% confidence interval: 19.0–41.1, p < 0.001 [one‐sided null hypothesis proportion of 10%]), including RET fusion (n = 1) and mutations in KRAS (n = 11), EGFR (n = 5), ERBB2 (n = 3), and BRAF (n = 1). Median time from sample submission to results was 8 days (range, 5–17 days). Conclusion Rapid follow‐on comprehensive testing of ctDNA should be considered prior to first‐line treatment for patients with advanced nonsquamous NSCLC when no alterations are detected after single‐plex tissue testing.


| INTRODUCTION
Current National Comprehensive Cancer Network guidelines recommend that all patients with newly diagnosed advanced lung adenocarcinoma should be tested for driver alterations, as tumors harboring these driver oncogene alterations can be treated with molecularly targeted therapies. 1Targeted drugs for EGFR, ALK, ROS1, BRAF, RET, MET, KRAS and NTRK alterations have already been approved in Japan and corresponding molecular tests have been introduced to clinical practice simultaneously with approval of the respective molecularly targeted drugs.Each test requires a sufficient tissue sample, whereas specimens collected from advanced non-small cell lung cancer (NSCLC) patients are often either small or cytology specimens.Conventional single-plex gene tests have been performed to identify patients who may be responsive to molecularly targeted drugs.3][4][5][6][7][8][9] Given this background, multiplex genetic analysis is highly advantageous in clinical practice, enabling the analysis of limited amounts of biopsied tissue samples.
In Japan, participation in the national health insurance system is mandatory and covers most medical expenses.Under this system, medical doctors request the use of medical procedures, equipment and drugs that have been approved by the Japanese Ministry of Health, Labour and Welfare. 10,11Support for treatment with the approved precision medicine requires the use of an approved assay for the detection of the relevant biomarker.Although multi-gene assays are approved for assessing most actionable biomarkers prior to first-line treatment, single-plex gene analysis is still used in patients with limited tissue samples.In the absence of accurate identification, opportunities for implementing appropriate molecularly targeted therapies against advanced NSCLC may be missed.On the other hand, repeated biopsies may not be feasible for many patients because of anatomical difficulties, existing comorbidities, and/or clinical deterioration forcing rapid initiation of medical treatment.
In daily practice, cases are sometimes encountered in which only some driver genes can be tested because of insufficient or unavailable tissue samples.In this case, multiplex gene analysis of circulating tumor DNA (ctDNA) may be useful.To verify the utility of ctDNA next-generation sequencing (NGS) in such situations, we planned this prospective study to assess the clinical performance of ctDNA NGS in a multi-institutional prospective cohort of patients diagnosed with advanced-stage nonsquamous NSCLC who had been tested for at least EGFR mutations by single-plex gene analysis and were negative for any driver alterations.

| Study design and patients
This prospective observational study (WJOG13620L/ STARLIGHT) evaluated the detection rate of actionable gene alterations by ctDNA NGS in patients with nonsquamous NSCLC for whom gene alterations were not detected in tissue-based single-plex assays.This multicenter, prospective study recruited patients at 24 institutions in Japan.Patients were enrolled if they met the following eligibility criteria: (1) histologically confirmed nonsquamous NSCLC; (2) Stage IIIB-IV or recurrent disease after surgery or chemoradiotherapy; (3) age ≥ 20 years; (4) EGFR gene confirmed as wild type by single-plex assay and "unknown" or "negative" results for the other eight genes (ALK, ROS1, BRAF, KRAS, MET, RET, ERBB2, and NTRK) tested at the time of enrollment; (5) comprehensive genomic profiling had not been performed; (6) no prior treatment with systemic anti-cancer treatment (e.g., cytotoxic agent, molecularly targeted therapy, or immune checkpoint inhibitor) for advanced-stage NSCLC (excluding postoperative adjuvant chemotherapy or chemoradiotherapy); (7) disease considered treatable with anti-cancer drugs; (8) life expectancy ≥3 months; and (9) written informed consent from the patient after sufficient explanation of the study.The key exclusion criteria were as follows: (1) presence of other malignancies with metastases; (2) serious mental disease; and (3) not fit for this trial as determined by the treating physician.
This study was conducted in accordance with the Declaration of Helsinki and the Japanese ethical Guidelines for Medical and Health Research Involving Human Subjects.The study protocol was approved by the institutional review board of each participating institution and registered at the University Hospital Medical Information Network Clinical Trial Registry (protocol no.UMIN000041583).

K E Y W O R D S
circulating tumor DNA, liquid biopsy, lung cancer, molecularly targeted therapy, multiplex gene analysis 2.2 | Blood samples, ctDNA isolation, and ctDNA sequencing The NGS analysis of ctDNA was conducted using Guardant360 (Guardant Health, Redwood City, CA) at Guardant Health, a Clinical Laboratory Improvement Amendments-certified, College of American Pathologist-accredited, New York State Department of Health-approved assay, as previously described. 12This assay interrogates 74 genes and detects single-nucleotide variants (SNVs), insertions and deletions, fusions, and copy number alterations.Whole blood (20 mL) was collected in Streck tubes (Streck, La Vista, Nebraska) during routine phlebotomy before starting first-line treatment, and samples were shipped at ambient temperature overnight to California from Japan.Analyzed results were promptly reported to physicians via an electronic portal site.

| Data analysis
Patient backgrounds were evaluated for all enrolled patients.The primary endpoint of this study was the proportion of patients with an alteration found by ctDNA NGS in any one of nine clinically relevant genes: EGFR; ALK; ROS1; BRAF; KRAS; MET; RET; ERBB2; and NTRK.We set the clinically useful discovery rate (H 1 ) as 25%, roughly half the value reported in a prior study using a tissue-based follow-on multiplex assay. 13We set the threshold rate (H 0 ) as 10% because the test seems to be clinically useful if the multiplex test finds about 10% of driver gene alterations in patients for whom no driver gene alterations were found by the single-plex test.The required sample size was calculated as 65 patients using a binomial test (one-sided α = 0.025, β = 0.10).The planned total sample size was set at 72 considering the potential ineligibility of patients or inadequate quality of samples.The two-tailed 95% confidence interval (CI) was used to evaluate statistical significance.Secondary endpoints were the percentages of patients with each of the nine major alterations detected, turnaround time (TAT), rate of patients showing gene alteration detected using companion diagnostic (CDx) methods to confirm the gene alteration detected in ctDNA, and the rate of patients treated with molecularly targeted therapies based on the ctDNA result, up to 6 months after study enrollment.We defined TAT as the period from the day blood was collected to the day that the ctDNA result was obtained from the website.All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA).

| Patient characteristics
In total, 72 patients were enrolled between October 2020 and May 2021.The baseline characteristics of patients are summarized in Table 1.Median age was 72.5 years (range, 47-86 years) and 50 patients (69.4%) were male.Sixteen patients (22.2%) were never-smokers, whereas 56 (76.4%) patients had a history of smoking.The performance status of these patients was generally 0 or 1 (91.6%).Percentages of patients with adenocarcinoma and with Stage IV disease were 91.7% and 73.6%, respectively.

| TAT
Median time from the day blood was collected to the day that the result was available was 8 days (range, 5-17 days; Figure 4).

| Follow-up tissue testing
The demographics and test results for the 21 patients for whom driver gene alterations were detected in ctDNA are summarized in Table 5.Among all enrolled patients, four (5.6%; 95% CI, 1.5%-13.6%)underwent re-analysis of tissue samples.For two patients, this confirmed the presence of driver alterations that were detected in ctDNA (L858R in EGFR exon 21 by single-plex test and V600E in BRAF exon 15 by multiplex test).On the other hand, for the two patients in whom EGFR exon 18 deletion and exon 20 insertion were detected in ctDNA, the same alteration could not be confirmed by tissue-based single-plex test.In another patient without a detectable driver alteration by single-plex and ctDNA testing, ALK fusion was eventually detected by tissue-based multiplex analysis (Table 5, Patient 22).

| Targeted therapy for patients based on ctDNA results
Among all 72 patients, two patients (2.8%; 95% CI, 0.3%-9.7%)were treated by targeted therapies according to ctDNA results.One patient showed L858R in EGFR exon 21, which was undetected at first with a tissue-based single-plex test using a bone metastasis sample and later confirmed by performing an approved single-plex test performed on a primary tumor specimen that had been surgically resected 10 years prior.This patient was treated with osimertinib (Table 5, Patient 16).A patient whose tumor showed deletion in EGFR exon 18 detected by ctDNA, underwent confirmatory tissue-based multiplex analysis again, whereas gene alterations were not detected (Table 5, Patient 12).That patient was treated with afatinib and showed partial response with a treatment duration of 6 months.Tyrosine kinase inhibitor (TKI) treatment was not administered in two patients harboring KRAS G12C mutations because confirmation by companion diagnosis had not been performed in these patients.Those patients in whom BRAF mutation or RET fusion was detected were responding to previous treatments and had not been treated with TKIs.

| DISCUSSION
This is the first prospective study to demonstrate the ability of follow-on ctDNA testing to detect informative genomic alterations in patients with advanced NSCLC for whom initial tumor tissue testing was negative for at least EGFR mutations.Using this approach, we identified relevant alterations in 29.2% (95% CI, 19.0%-41.1%) of patients tested.Alterations were identified in some previously tested genes as well as in genes that were not tested, reflecting the challenges of single-plex biomarker testing in clinical practice.Potentially actionable driver gene alterations, according to standards established in international guidelines, were detected in 16.7% of samples tested (12/72; 12 patients who were excluded had tumors harboring KRAS mutations other than G12C).
For follow-on testing, examination of ctDNA was less invasive than histological biopsy.Furthermore, the median time from blood testing to result confirmation tended to be shorter (8 days; range, 5-17 days), against 11 days (range, 8-14 days) associated with one of the tissue-based multi-gene assays available in Japan. 14ecause some previous reports have shown that selecting the appropriate first-line molecularly targeted therapy leads to benefits in terms of long-term outcomes, 3,8 the results of shortening TAT may also be important in clinical settings.
In Japan, the Oncomine Dx Target test Multi CDx System (Thermo Fisher Scientific Inc., Waltham, MA, USA) was approved as a CDx test to identify alterations in five driver genes for first-line targeted treatments in NSCLC are available: EGFR, ALK, ROS1, BRAF (exon 15 V600E), and RET.This tissue-based multiplex assay was studied in the Lung Cancer Genomic Screening Project for Individualized Medicine in Japan (LC-SCRUM-Japan).Among 1688 advanced NSCLC patients without EGFR gene mutations according to prior testing, the following driver alterations were detected: 17.0% for KRAS mutations; 8.4% for EGFR mutations; 6.3% for ERBB2 mutations; 4.2% for BRAF mutations; 4% for ROS1 fusion; 3% for RET fusion; 3% for ALK fusion; 2% for MET exon 14-skipping mutation; and 1% for the KRAS mutations and BRAF mutations combined (total rate, 48.9%). 13ere, we showed with a comprehensive liquid biopsy and median TAT of 8 days that the total detection rate was 29.2% (p = 0.001; 95% CI, 19.0%-41.1%),including some patients who were tested for more than EGFR mutations.
Although direct comparisons cannot be made because of differences in patient background, study sample sizes and sources of tumor tissue, the detection rates we observed with ctDNA analysis were somewhat lower but confirmed that testing with a platform that did not rely on tissue could identify a clinically meaningful proportion of patients who could be candidates for first-line targeted therapy.
In our study, EGFR mutations were detected in the second highest number of patients (n = 5, 6.9%) using ctDNA analysis despite the fact that prior tissue testing did not find EGFR mutations.We did not collect information on which assays were used for determining EGFR-negative status.Most single-plex EGFR gene mutation assays focus on common sites of alterations, as so-called "hot spots," and do not fully interrogate the gene.Therefore, less common but potentially actionable alterations such as V774M/ H773L in exon 20, L747P in exon 19, and exon 18 deletion, as detected in our study, may not be found using currently approved companion diagnostic tests. 15Theoretically, the EGFR L858R and exon 20 insertion should have been detected by single-plex testing.The failure of currently available hotspot testing to achieve such detection may be due to tumor sampling and the low proportion of tumor cells in samples.The same reasons were considered for the two patients showing positive results for KRAS mutations, and one patient showing a positive result for ERBB2 mutation in ctDNA analysis, with negative results by single-plex analysis of each clinically relevant gene.
Leighl et al. reported that ctDNA NGS identified eight genomic biomarkers, including EGFR mutations, at a rate at least as high as that of tissue genotyping, with high tissue concordance, more rapidly and more completely than tissue-based genotyping among 282 patients with newly diagnosed metastatic NSCLC. 16In addition, Zugazagoitia et al. analyzed the clinical utility of the same assay in patients with inadequate tumor samples for tissue genotyping with 93 advanced-stage lung adenocarcinoma patients, identifying actionable alterations in 13 patients (14%). 17Similar to our approach, other researchers have found that plasma ctDNA analysis and tissue genotyping  are complementary tools for therapeutic decision-making for advanced NSCLC. 18However, because the ctDNA analysis used in this study has not been approved for first-line decision making in Japan, the results could not be used for treatment decisions.The approval of ctDNA analysis as a comprehensive genome-profiling test before first-line treatment is desired in Japan, but not yet available.
One key limitation in this study was bias in patient selection.When the sample quantity is limited, prioritizing multiplex tissue-based testing over single-plex gene analysis is preferred.Since our study excluded patients who had undergone multiplex gene analysis, some patients may have preferentially chosen to undergo single-plex gene analysis for the purposes of enrolment in the study.The BRAVE study investigated how to select first-line treatment based on testing of EGFR, ALK, ROS1, and PD-L1 in nonsquamous NSCLC patients in 11 medical centers in Japan and showed that, while 197 of 202 patients (97.5%) were tested for EGFR, only 39.1% were concurrently tested for all three genomic biomarkers. 19Another important limitation is the effectiveness of the molecular targeted drug was not confirmed for the patients whose gene alteration was detected.Aggarwal et al. reported the therapeutic effect of molecularly targeted drug in 42 NSCLC patients for whom driver alterations were detected by ctDNA. 20According to those results, 36 (85.7%) of the 42 patients with evaluable results achieved either complete response (n = 1), partial response (n = 19), or stable disease (n = 16).Paik et al. also reported similar efficacy in a study using tepotinib for lung cancer patients harboring MET exon 14-skipping mutation between tissue-positive patients and ctDNA-positive patients. 21These results support the reliable positive percentage agreement of the ctDNA analysis used in our study.Finally, in the developed world, where resources are readily available and reimbursed, the trend is to perform multiplex testing as the initial examination.However, this is not the practice in all situations.In some cases, using a single-plex examination may be more practical and cost-effective, such as when gene analysis results need to be obtained quickly or to identify common alterations that account for the majority of actionable biomarkers for which treatments are accessible.For example, in Asia, single-plex testing for EGFR mutation remains an important examination because of the high frequency of EGFR mutation.This study was a single-arm trial with a relatively small sample size and was exploratory in nature.The present results should therefore be confirmed by a large clinical study to evaluate the detection rate for driver gene alterations using cfDNA and a tissue sample.
In conclusion, this multicenter, prospective observational study demonstrated that comprehensive next-generation sequencing of ctDNA from plasma samples identified driver gene alterations in 29.2% of patients with untreated advanced nonsquamous NSCLC.These results were obtained within a reasonable timeframe.The use of multiplex testing of tissue specimens has become more common for lung cancer, but in situations where the probability of a driver alteration is not rare and the results of initial testing are negative, rapid follow-on comprehensive liquid biopsy could be useful to ensure that actionable driver alterations missed by initial testing are detected prior to initiation of first-line treatment for advanced NSCLC.

F I G U R E 2
Detection rates for the nine driver-gene alterations detected by multiplex gene analysis.Pie chart shows the number of cases.No driver alterations were detected in NTRK, ROS1, ALK, or MET.T A B L E 3 Summary of EGFR and KRAS mutations.

F I G U R E 4 T A B L E 5
Abbreviations: AD, adenocarcinoma; SA, sarcomatoid carcinoma; VAF, variant allele frequency.
Characteristics of the 72 patients in the present study.
T A B L E 1 Characteristics of patients with (n = 21) and without (n = 51) clinically informative alterations detected in ctDNA.