A Large-Scale Prospective Concordance Study of Plasma- and Tissue-Based Next-Generation Targeted Sequencing for Advanced Non–Small Cell Lung Cancer (LC-SCRUM-Liquid)

This article is featured in Highlights of This Issue, p. 1375

A variety of oncogenic drivers have been identified in non–small cell lung cancer (NSCLC), and molecular targeted therapy has greatly improved the clinical outcomes of patients with oncogenic drivers (1). Plasma cell-free DNA (cfDNA) sequencing has been developed as a less invasive method than conventional tissue genotyping for detecting various genomic alterations. Some previous retrospective studies have examined the concordance between plasma cfDNA sequencing and tissue genotyping. Previous small studies (n = 72–287) reported positive percent agreement (PPA) of plasma cfDNA sequencing compared with tissue genotyping as 58.8% to 95.8% for EGFR mutations, 75.0% for KRAS G12X, 40.0% to 100.0% for ALK fusions, and 33.3% to 100.0% for BRAF V600E (2–6). However, the concordance between plasma cfDNA sequencing and tissue genotyping has not been evaluated in detail because these results are based on smaller cohorts, and in particular, the number of patients with rare fractions of oncogenic drivers was extremely low. Therefore, to evaluate the detectability of oncogenic alterations in plasma cfDNA sequencing precisely, prospective comparative analyses with the corresponding tumor tissue genotyping in a large-scale sample size study are needed. We evaluated the concordance between plasma cfDNA sequencing and tissue assays for the detection of oncogenic alterations in patients with advanced NSCLC using a large-scale prospective study.

A large-scale lung cancer genomic screening project, LC-SCRUM-Asia, was started in February 2013, and tissue genotyping was performed to identify lung cancer patients with oncogenic drivers (UMIN number: 000010234 and 000036871; ref. 7). As of October 2021, more than 14,000 patients were already enrolled in this study.

This liquid biopsy study, LC-SCRUM-Liquid, has been conducted as an additional study in LC-SCRUM-Asia since December 2017. Blood samples were prospectively collected from patients with advanced or recurrent NSCLC within 4 weeks of tissue biopsy. Plasma cfDNA was extracted from blood samples and analyzed using next-generation sequencing (NGS). The concordance of oncogenic drivers in plasma cfDNA sequencing was evaluated, compared with tissue genotyping, which was performed independently and blindly by plasma cfDNA sequencing. The clinical outcomes of patients who received genotype-matched therapy, were also prospectively investigated.

Patients who met the following eligibility criteria were enrolled: (i) above the age of 20; (ii) with histologically/cytologically confirmed NSCLC; (iii) clinical stage III or, IV, or recurrence; (iv) diseases were unsuitable for operation or thoracic radiotherapy, but suitable for chemotherapy; (v) chemonaive or one or two prior systemic treatments for lung cancer, (vi) already enrolled in LC-SCRUM-Asia, and (vii) with blood samples taken within 4 weeks after tissue sample biopsy.

LC-SCRUM-Asia and LC-SCRUM-Liquid were approved by the Institutional Review Board of the National Cancer Center (approval number 2012–257 and 2017–222, respectively) and by each institution participating in these studies. Written informed consent was obtained from all the patients. Our studies were conducted in accordance with the guidelines for medical and health research involving human subjects specified in the Declaration of Helsinki.

Blood samples, collected using a blood collection tube, Streck Cell-Free DNA BCT (Streck Corporate, NE), were submitted to Guardant Health, a Clinical Laboratory Improvement Amendments (CLIA)-certified, and College of American Pathologists—accredited laboratory, and was subjected to plasma cfDNA sequencing, Guardant 360 panel (Guardant Health, CA), targeting 73 (until April in 2019) or 74 (afterward) cancer-related genes.

Tissue samples were mainly collected from previously untreated patients. Tissue genotyping was performed within LC-SCRUM-Asia. Tumor tissue analysis was mainly performed using fresh frozen biopsy samples. Tissue samples were submitted to a CLIA-certified clinical laboratory (SRL Incorporation, Tokyo, Japan). DNA and RNA extracted from the tissue samples were subjected to a tissue-based NGS assay, Oncomine Comprehensive Assay version 1 or 3 (Thermo Fisher Scientific, MA), targeting 143 (version 1) or 161 (version 3) cancer-related genes. In this assay, gene mutations were analyzed by DNA assay, and fusions and MET exon 14 skipping were analyzed by RNA assay.

Clinical data of patients were collected using an electronic data capture system of LC-SCRUM-Asia. The patients’ baseline characteristics were collected when the patients were enrolled in LC-SCRUM-Asia, and follow-up clinical data, including the start dates of systemic anticancer drug therapy, therapeutic regimens, tumor responses, dates of disease progression, and prognosis, were periodically collected.

Mutations in EGFR, KRAS, BRAF, HER2, NRAS, HRAS, AKT1, and MAP2K1, fusions in ALK, RET, ROS1, and FGFR3, and MET exon 14 skipping, were defined as targetable gene alterations. Among these targetable gene alterations, the concordance for alterations of eight oncogenic drivers [mutations of EGFR (insertion, deletion, and missense mutation in exons 18–21); KRAS (G12X, G13X, and Q61X); BRAF (V600E); and HER2 (insertions in exon 20): fusions of ALK, RET, and ROS1; and MET exon 14 skipping] in plasma cfDNA sequencing was assessed by estimating PPA, negative percent agreement (NPA), positive predictive value (PPV), negative predictive value (NPV), and overall percent agreement (OPA) of plasma cfDNA sequencing compared with the results of the tissue assays. These concordance analyses were performed in variants of the eight oncogenic drivers, which were covered by both the two assays.

Turnaround time (TAT) was defined as the duration from sample submission to reporting the sequencing results, and the results of plasma cfDNA sequencing and tissue assay were compared using the Wilcoxon sum rank test.

The Kaplan–Meier method was used to estimate the progression-free survival (PFS) of patients who received genotype-matched therapy. EZR software (Saitama Medical Center, Jichi Medical University, Japan) was used for the statistical analyses.

The funder of LC-SCRUM-Liquid and LC-SCRUM-Asia had no role in the study design, data collection, data analysis, data interpretation, or writing of the report.

The data generated in this study are available upon reasonable request from the corresponding author. Requests are reviewed by the research group.

From December 2017 to January 2021, 1,112 patients with advanced or recurrent NSCLC were enrolled in LC-SCRUM-Liquid. Of these, 1,065 paired blood and tissue samples were available for this study analyses. Three patients who were ineligible for inclusion were excluded. Thus, 1,062 patients (95%) were analyzed in this study (Supplementary Fig. S1).

The patient characteristics are shown in Table 1. The median age was 69 years (range: 25–91). The majority were male (61%), smokers (69%), and had stage IV disease (80%). Almost all the patients were previously untreated (93%). The histology of tumors comprised 77% adenocarcinoma, 14% squamous cell carcinoma (SCC) and other NSCLCs. Number of metastatic sites was 0 in 14%, 1 in 33%, 2 in 22%, 3 or more in 15%. There were brain metastasis in 17%, pulmonary metastasis in 31%, pleural dissemination or pleural effusion in 24%, liver metastasis in 6%, adrenal metastasis in 7%, and bone metastasis in 24%. Tissue samples for tissue assays were mainly obtained as fresh frozen (90%) and from primary lung tumor (60%), metastatic sites (29%), or pleural effusion (11%).

The success rates of genomic analysis by plasma cfDNA sequencing and tissue assay were 91% (964/1,062) and 97% (1,025/1,062), respectively. TAT in plasma cfDNA sequencing was significantly shorter than that in the tissue assay [10 days (range: 6–27) vs. 22 days (range: 12–57), P < 0·01).

In plasma cfDNA sequencing, targetable gene alterations were detected in 473 patients (44.5%). Of these, the number of eight oncogenic alterations were 255 EGFR mutations (24.0%), 129 KRAS mutations (12.1%), 10 HER2 exon 20 insertions (0.9%), 7 BRAF V600E mutation (0.7%), 26 ALK fusions (2.4%), 9 RET fusions (0.8%), 3 ROS1 fusions (0.3%), and 16 MET exon 14 skipping (1.5%; Fig. 1A). In contrast, eight oncogenic alterations were detected by tissue assay in 549 patients (51.6%). There were 281 EGFR mutations (26.4%), 145 KRAS mutations (13.6%), 11 HER2 exon 20 insertions (1.0%), 7 BRAF V600E mutation (0.7%), 45 ALK fusions (4.2%), 14 RET fusions (1.3%), 16 ROS1 fusions (1.5%), and 18 MET exon 14 skipping (1.7%) in tissue assay (Fig. 1B).

Among 147 patients with SCC, targetable gene alterations were detected in 19 patients (12.9%) by plasma cfDNA sequencing, and in 16 patients (10.8%) by tissue assay (Supplementary Fig. S2). One of the eight oncogenic alterations was detected by plasma cfDNA sequencing or tissue assay in 18 patients with SCC; 8 EGFR mutations, 6 KRAS mutations, 1 ALK fusion, 3 MET exon 14 skipping (Supplementary Table S1).

As shown in Fig. 2A, the overall PPA of plasma cfDNA sequencing was 72% (389/537). Other performance indexes of plasma cfDNA sequencing were as follows: NPA, 87% (459/525); PPV, 85% (389/455), NPV, 75% (459/607); and OPA, 79% (848/1,062; Supplementary Table S2).

For the DNA assay, PPA of plasma cfDNA sequencing was 78% (345/444; Fig. 2A): EGFR, 78% (221/281); KRAS, 75% (110/145); BRAF, 85% (6/7); HER2, 72% (8/11; Fig. 2B). Other performance indexes of plasma cfDNA sequencing for DNA assay were as follows, NPA, 90% (562/618); PPV, 86% (345/401), NPV, 85% (562/661); OPA, 85% (907/1062; Supplementary Table S2).

For the RNA assay, PPA of plasma cfDNA sequencing was 47% (44/93; Fig. 2A): MET exon 14 skipping, 66% (12/18); ALK, 46% (21/45); ROS1, 18% (3/16); RET, 57% (8/14; Fig. 2B). Other performance indexes of plasma cfDNA sequencing were as follows: NPA, 98% (959/969); PPV, 81% (44/54); NPV, 95% (959/1,008); and OPA, 94% (1,003/1,062; Supplementary Table S2).

The breakdown of discordant results between plasma cfDNA sequencing and tissue assays is shown in Fig. 3. Among the 1,062 patients, 389 showed concordant results between each assay. Among patients with oncogenic alterations detected by plasma cfDNA sequencing only, the results of tissue assay were unavailable due to unsuitable tissue samples in 70% (46/66) and no detection of oncogenic alterations in only 30% (20/66); among patients with oncogenic alterations detected by tissue assay only, the results of plasma cfDNA sequencing showed no detection of oncogenic alterations in 61% (90/148).

To investigate whether if there were any subpopulations in which plasma cfDNA sequencing was more sensitive, we evaluated PPA of plasma cfDNA sequencing according to patient characteristics. PPA of plasma cfDNA sequencing was similar regardless of smoking status (P = 0.84), stage (P = 0.47) or histology (P = 1.00), and higher in patients with 3 or more metastatic sites than in those with 2 or less metastatic sites (0, 69%; 1, 63%, 2, 71%; 3 or more, 87%; P < 0.01; Supplementary Fig. S3).

We also evaluated metastatic site and PPA of plasma cfDNA sequencing to identify subpopulations in which plasma cfDNA sequencing was more preferable. PPA was higher in patients who had brain metastasis (Brain +, 80%; Brain −, 68%; P = 0.01), liver metastasis (Liver +, 88%; Liver −, 69%; P = 0.01), adrenal metastasis (Adrenal +, 90%; Adrenal −; 69%; P = 0.01), and bone metastasis (Bone +, 85%; Bone −, 63%; P < 0.01), and was not different between patients with and without lung metastasis (P = 0.59), or pleural dissemination and effusion (P = 0.05; Supplementary Fig. S4).

There were 54 patients whose distant metastasis was present only in brain. In the 54 patients, PPA of plasma cfDNA sequencing was not different between mutation detection and fusion/exon skipping detection [60% (12/20) vs. 62% (5/8); P = 1.00; Supplementary Table S3].

To clarify whether oncogenic alterations detected by plasma cfDNA sequencing are correctly diagnosed and accurately reflect the efficacy of genotype-matched therapy, we analyzed the clinical outcomes of patients treated with genotype-matched therapy based on plasma cfDNA sequencing and tissue assays. Clinical outcome data of 115 patients treated with genotype-matched therapy were available. Among these patients, the oncogenic alterations were detected only by tissue assay in 31 patients (T group), by both tissue assay and plasma cfDNA sequencing in 71 patients (TP group), and only by plasma cfDNA sequencing in 13 patients (P group). The median PFS of T, TP, P groups were 23.0 months [95% confidence interval (CI), 12.4–not reached (NR)]; 12.4 months (95% CI, 9.1–16.3); and 12.7 months (95% CI, 5.0–13.5), respectively (Fig. 4A). Therefore, the median PFS for each group was > 12 months. The median PFS of the T and P groups was not inferior to that of the TP group. In 13 patients in the P group, in which tissue samples were unsuitable for genomic analysis due to insufficient quantity or quality of the DNA, RNA or both, the response rate of genotype-matched therapy was 85% (11/13; Supplementary Table S4).

As for patients with EGFR mutations, there were 19, 63, 11 patients in the T, TP, P groups, respectively. In the treatment with EGFR—tyrosine kinase inhibitors (TKI), the median PFS of the T, TP, P groups was 23.0 months (95% CI, 4.7–NR); 10.4 months (95% CI, 7.8–15.0); and 12.7 months (95% CI, 5.0–13.5), respectively (Fig. 4B). The median PFS of the T and P groups was not inferior to that of the TP group.

To our knowledge, this is the largest prospective concordance study for plasma cfDNA sequencing, in which tissue- and plasma-based NGS assays were simultaneously performed in patients with advanced NSCLC. The within 4-week interval for the tissue and plasma sample collections for all patients made the accurate evaluation of the concordance possible. Moreover, this study included 74 patients with rare fractions of oncogenic drivers, such as BRAF V600E (n = 8), HER2 exon 20 insertions (n = 13), MET exon 14 skipping (n = 22), and fusions of ROS1 (n = 16) and RET (n = 15). For concordance analysis, previous studies included only a few patients with rare fractions of oncogenic drivers, such as BRAF V600E mutation, ROS1 fusions, and RET fusions (2–6). This large-scale study enabled us to evaluate the clinical performance of plasma cfDNA sequencing, especially for detecting rare oncogenic drivers, which had not been previously proven precisely.

Previous reports have shown that the PPA of plasma cfDNA sequencing compared with tissue assay was 58.8% to 95.8% for EGFR mutations, and 40% to 100% for ALK fusions (2–6). However, these reports were not sufficient to evaluate the PPA of plasma cfDNA sequencing accurately because the studies were mostly conducted retrospectively, and they excluded tissue or plasma samples that were unavailable due to insufficient DNA or RNA. In this study, the PPA of plasma cfDNA sequencing was 72% to 85% for mutations in EGFR, KRAS, HER2, or BRAF, and 18% to 57% for fusions in ALK, RET, or ROS1 compared with those of tissue assays. We reveal that the detection of oncogenic alterations by plasma cfDNA sequencing was not as sensitive as previously reported but was inferior to that by tissue assay. In particular, the PPA of plasma cfDNA sequencing for gene fusions against tissue RNA assay was extremely low (less than 60%) compared with that for mutations against tissue DNA assay in our study. In a prospective report, the PPA of plasma cfDNA sequencing compared with tissue assay was 81.8% to 90% for EGFR mutations, and 62.5% for ALK fusions (3). PPA of plasma cfDNA sequencing in gene fusions was reported to be lower than that in gene mutations, because gene fusions include various variants and the capture of fusion DNA fragments is technically difficult due to the low capturing efficiency and shortness of cfDNA fragments, as indicated in a previous report (8). ROS1 fusion is known to have many partner genes compared with ALK and RET fusions; therefore, the poor detectability of ROS1 fusion in plasma cfDNA sequencing (PPA, 18%) might also be caused by the existence of various variant types. In addition, bioinformatic technologies could also influence the detectability of gene fusions. A previous study demonstrated that PPA of plasma cfDNA sequencing for ALK fusions was improved by updating bioinformatic systems for fusion detection (3, 9). Plasma cell-free RNA (cfRNA) analysis also showed a higher sensitivity for detecting fusion genes than plasma cfDNA sequencing (cfRNA, 78%; cfDNA, 33%; ref. 10). Thus, detection sensitivity for fusions in plasma assay could be improved by further advances in technology, including DNA capturing methods, bioinformatics and plasma cfRNA analysis.

There were some discordant results between plasma cfDNA sequencing and tissue assays. The main discordant reasons, in which oncogenic alterations were positive by plasma cfDNA sequencing and negative by tissue assay, were due to the unavailability of tissue samples because of the insufficient quality or quantity of DNA or RNA. When the quality and quantity of tissue samples are acceptable for genomic analysis and the results of tissue assays are negative, plasma cfDNA sequencing does not provide additional information because oncogenic alterations are rarely detected by plasma cfDNA sequencing. Therefore, plasma cfDNA sequencing could be useful for detecting oncogenic alterations only when tissue assay is unavailable.

The utility of biomarker-matched precision medicine based on plasma cfDNA sequencing has not been well investigated. In particular, the efficacy of genotype-matched therapy in patients whose oncogenic drivers are detected only by plasma cfDNA sequencing is not fully understood, although one previous study reported the responses to plasma genotype-matched therapy (11). Our study also demonstrated that, in 13 patients with oncogenic alterations identified only by plasma cfDNA sequencing, the corresponding genotype-matched therapy showed robust clinical activities. Moreover, the median PFS of patients with oncogenic alterations detected only by plasma cfDNA sequencing was over 12 months. These data were comparable with the median PFS of patients treated with tissue genotype-matched therapy (12–15). However, the median PFS of patients with oncogenic alterations detected only by plasma cfDNA sequencing tended to be shorter than that of patients with oncogenic alterations detected only by tissue assay. This is because patients with oncogenic alterations detected by plasma cfDNA sequencing often have more advanced cancers and a higher tumor burden (11, 16). Indeed, higher positivity by cfDNA sequencing was demonstrated in patients with 3 or more metastatic sites, and in patients with brain, liver, adrenal, or bone metastasis in the current study. Our results suggest that oncogenic alterations detected by plasma cfDNA sequencing are genuine for selecting the corresponding genotype-matched therapy. Therefore, treatments selected using plasma cfDNA sequencing could be suitable for patients with advanced NSCLC, especially when tissue assays are unavailable. To further validate the clinical utility of plasma cfDNA sequencing, we are presently conducting prospective umbrella trials of genotype-matched therapy stratified on the basis of this liquid biopsy study (JapicCTI number: JapicCTI-205154 and JapicCTI-205155).

This study has some limitations. First, although our study was large-scaled, patients with oncogenic alterations in HER2, BRAF, MET, RET, or ROS1 were only 74 in total. Accurate evaluation of concordance in rare fractions of oncogenic alterations was limited even in this large-scale analysis, and it requires larger-scale concordance studies with over 10,000 patients. Second, the efficacy of genotype-matched therapy in each patient was evaluated by investigators in clinical practice.

In conclusion, plasma cfDNA sequencing in patients with advanced NSCLC had a relatively high detectability for gene mutations but a lower detectability for gene fusions and MET exon 14 skipping. Our data indicated that plasma cfDNA sequencing could not fully replace tissue assays for oncogenic alterations detection. However, when positive results are obtained, plasma cfDNA sequencing has a diagnostic value equivalent to that of the tissue assay in predicting the efficacy of genotype-matched therapy for plasma oncogenic driver–positive patients. Therefore, plasma cfDNA sequencing can be a promising alternative to tissue genotyping when the tissue is unavailable because of insufficient DNA/RNA. Further, new technologies for plasma cfDNA sequencing could improve its clinical utility for NSCLC.

A. Sugimoto reports personal fees from Chugai Pharmaceutical Co. Ltd. outside the submitted work. S. Matsumoto reports grants and personal fees from Merck; personal fees from Eli Lilly, AstraZeneca, Chugai Pharmaceutical, ThermoFisher Scientific, Riken Genesis, Guardant Health, Novartis Pharma, and Amgen; and grants from Janssen Pharmaceutical outside the submitted work. H. Udagawa reports grants from Takeda and Boehringer Ingelheim GmbH outside the submitted work. S. Umemura reports personal fees from Chugai Pharmaceutical outside the submitted work. K. Nishino reports grants and personal fees from AstraZeneca, Chugai Pharmaceutical, Eli Lilly Japan, Pfizer, and Merck and personal fees from Roche Diagnostics, Nippon Boehringer Ingelheim, and Novartis outside the submitted work. I. Nakachi reports personal fees from AstraZeneca, Chugai Pharmaceutical, and Novartis Pharma outside the submitted work. S. Kuyama reports personal fees from Chugai Pharmaceutical Co. Ltd., Bristol-Myers Squibb, Boehringer Ingelheim, AstraZeneca, Pfizer, Eli Lilly, MSD, Taiho Pharmaceutical Co. Ltd., Sanofi, Kyowa Kirin Co. Ltd., Hisamitsu Pharmaceutical Co. Inc., Daiichi Sankyo, Nippon Kayaku Co. Ltd., and Novartis outside the submitted work. H. Daga reports grants and personal fees from AstraZeneca and Chugai Pharmaceutical and personal fees from Eli Lilly Japan and Ono Pharmaceutical outside the submitted work. S. Hara reports personal fees from AstraZeneca, Ono, and Chugai outside the submitted work. T. Kato reports grants from AbbVie, Blueprint, Amgen, Haihe, and Regeneron; grants and personal fees from AstraZeneca, Boehringer Ingelheim, Chugai, Eli Lilly, Merck, MSD, Novartis, Pfizer, Taiho, and Takeda; and personal fees from Bristol-Myers Squibb, Daiichi Sankyo, Ono, and Roche outside the submitted work. J. Sakakibara-Konishi reports grants from Lilly and Boehringer outside the submitted work. T. Nakagawa reports personal fees from AstraZeneca K.K., Chugai Pharmaceutical Co. Ltd., Eli Lilly Japan K.K., Ono Pharmaceutical Co., Ltd., Pfizer Japan Inc., Taiho Pharmaceutical Co. Ltd., Boehringer Ingelheim Japan, 107 Inc., and MSD K.K. outside the submitted work. T. Kawaguchi reports grants and personal fees from Chugai and Taiho and personal fees from Ono, AstraZeneca, Boehringer, Takeda, Pfizer, Kyorin, Sanofi, Novartis, MSD, Lilly, and Kyowa Hakko Kirin outside the submitted work. T. Sakai reports personal fees from AstraZeneca K.K, Chugai Pharmaceutical, MSD, Merck Serono, Thermo Fisher Scientific, and Novartis Pharma and grants from Amgen outside the submitted work. Y. Shibata reports personal fees from Pfizer Inc., Bristol-Myers Squibb K.K., AstraZeneca K.K., Ono Pharmaceutical Co. Ltd., Takeda Pharmaceutical Co. Ltd., Eli Lilly Japan K.K., and Chugai Pharmaceutical Co. Ltd. outside the submitted work. H. Izumi reports grants and personal fees from Takeda Pharmaceutical and Ono Pharmaceutical; grants from Amgen Inc., Bristol-Myers Squibb, AbbVie, and Japan Society for the Promotion of Science; and personal fees from MSD, Merck, and Chugai Pharmaceutical outside the submitted work. K. Nosaki reports personal fees from Pfizer, Taiho Pharmaceutical, Lilly, and Ono; grants and personal fees from MSD, AstraZeneca, Chugai Pharmaceutical, Janssen Pharmaceuticals, Daiichi Sankyo/UCB Japan, and Takeda; and grants from AbbVie and Amgen outside the submitted work. Y. Zenke reports grants and personal fees from AstraZeneca, MSD, and Amgen; personal fees from Bristol-Myers Squibb, Lilly, Chugai, Ono Pharmaceutical, Boehringer Ingelheim, Pfizer, Taiho Pharmaceutical, Takeda Pharmaceutical, Novartis, Nippon Kayaku, and Kyowa Kirin; and grants from Merck and Daiichi Sankyo outside the submitted work. K. Yoh reports grants and personal fees from AstraZeneca, Boehringer Ingelheim, Chugai, Daiichi Sankyo, Lilly, Taiho, and Takeda; personal fees from Bristol-Myers Squibb, Janssen, Kyowa Kirin, and Novartis; and grants from AbbVie, MSD, and Pfizer outside the submitted work. K. Goto reports grants from Merck, Takeda Pharmaceutical, Amgen, Astellas, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Chugai, Daiichi Sankyo, Eisai, Janssen, Kyowa Kirin, Lilly, Medical & Biological Laboratories, MSD, Novartis, Ono, Pfizer, Sumitomo Dainippon, Bayer Yakuhin Ltd., Haihe Biopharma Co. Ltd., Ignyta Inc., Kissei Pharmaceutical Co. Ltd., Life Technologies Japan Ltd., Loxo Oncology Inc., Merus N.V., NEC Corporation., Pfizer Japan Inc., Spectrum Pharmaceuticals Inc., Sysmex Corporation, Turning Point Therapeutics Inc., and Taiho and nonfinancial support from Guardant Health Inc. during the conduct of the study as well as grants and personal fees from Amgen, Amgen K.K., Amgen Astellas BioPharma K.K., AstraZeneca K.K., Boehringer Ingelheim Japan Inc., Bristol-Myers Squibb K.K., Blueprint Medicines Corporation, Chugai Pharmaceutical Co. Ltd., Daiichi Sankyo Co. Ltd., Eisai Co. Ltd., Eli Lilly Japan K.K., Janssen Pharmaceutical K.K., Merck Biopharma Co. Ltd., MSD K.K., Novartis Pharma K.K., Ono Pharmaceutical Co. Ltd., Taiho Pharmaceutical Co. Ltd., and Takeda Pharmaceutical Co. Ltd. and personal fees from Amoy Diagnostics Co. Ltd., Bayer U.S., Guardant Health Inc., Thermo Fisher Scientific K.K., Medpace Japan K.K., and Otsuka Pharmaceutical Co. Ltd. outside the submitted work. No disclosures were reported by the other authors.

A. Sugimoto: Resources, data curation, formal analysis, investigation, visualization, writing–original draft. S. Matsumoto: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing. H. Udagawa: Resources, investigation, writing–review and editing. R. Itotani: Conceptualization, resources, investigation, methodology, writing–original draft. Y. Usui: Conceptualization, resources, investigation, methodology, writing–review and editing. S. Umemura: Conceptualization, resources, investigation, methodology, writing–review and editing. K. Nishino: Resources, investigation, writing–review and editing. I. Nakachi: Resources, investigation, writing–review and editing. S. Kuyama: Resources, investigation, writing–review and editing. H. Daga: Resources, investigation, writing–review and editing. S. Hara: Resources, investigation, writing–review and editing. S. Miyamoto: Resources, investigation, writing–review and editing. T. Kato: Resources, investigation, writing–review and editing. J. Sakakibara-Konishi: Resources, investigation, writing–review and editing. E. Tabata: Resources, investigation, writing–review and editing. T. Nakagawa: Resources, investigation, writing–review and editing. T. Kawaguchi: Writing–review and editing. T. Sakai: Resources, investigation, writing–review and editing. Y. Shibata: Resources, investigation, writing–review and editing. H. Izumi: Resources, investigation, writing–review and editing. K. Nosaki: Resources, funding acquisition, investigation, writing–review and editing. Y. Zenke: Resources, investigation, writing–review and editing. K. Yoh: Resources, funding acquisition, investigation, writing–review and editing. K. Goto: Conceptualization, resources, supervision, funding acquisition, investigation, methodology, project administration, writing–review and editing.

We are grateful to participating patients and their families. We also thank Ms. Yuri Murata, Ms. Akiko Iizuka, and PREMIA Inc. for administrative assistance in managing clinical samples, molecular screening, and clinico-genomic database in LC-SCRUM-Liquid and LC-SCRUM-Asia.

LC-SCRUM-Liquid was funded by Guardant Health Inc., Merck Biopharma Co. Ltd., and Takeda Pharmaceutical Company Limited. This work was supported by the National Cancer Center Research and Development Fund 28-A-6 (K. Goto), and 31-A-5 (A. Ohtsu), AMED Grant Number JP21ck0106289 (K. Goto), JP21ck0106568 (K. Goto), JP17Ack0106148 (K. Goto), JP21ck0106294 (K. Yoh), JP21ck0106483 (K. Nosaki), JP20ck0106411 (S. Matsumoto), JP20ck0106449 (I. Okamoto), JP20ck0106450 (S. Niho), JP20ak0101050 (K. Tsuchihara), JP18Ik0201056 (A. Ohtsu), JP18kk0205004 (H. Nakagama), and JP17Ack0106147 (S. Yano). Tissue NGS analysis in LC-SCRUM-Asia was supported by Amgen, Astellas, AstraZeneca, Boehringer Ingelheim, Bristol-Myers Squibb, Chugai, Daiichi Sankyo, Eisai, Janssen, Kyowa Kirin, Lilly, Merck, Medical & Biological Laboratories, MSD, Novartis, Ono, Pfizer, Sumitomo Dainippon, Taiho, and Takeda.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.