MolDX: Inivata™, InVisionFirst®, Liquid Biopsy for Patients with Lung Cancer

Title XVIII of the Social Security Act (SSA), §1862(a)(1)(A) allows coverage and payment for only those services that are considered to be reasonable and necessary for the diagnosis or treatment of illness or injury or to improve the functioning of a malformed body member.

42 CFR §410.32 Diagnostic x-ray tests, diagnostic laboratory tests, and other diagnostic tests: Conditions

CMS Internet-Only Manual, Pub. 100-03, Medicare National Coverage Determinations Manual, Chapter 1, Part 2, §90.2 Next-Generation Sequencing for Patients with Advanced Cancer

CMS Internet-Only Manual, Pub. 100-02, Medicare Benefit Policy Manual, Chapter 15, §80 Requirements for Diagnostic X-Ray, Diagnostic Laboratory, and Other Diagnostic Tests, §80.1.1 Certification Changes

This test is a “liquid biopsy". It is intended to assist physicians caring for patients who suffer from a common form of lung cancer and who have advanced disease.

This policy provides limited coverage for InvisionFirst^® - Lung (Inivata^™, Research Triangle Park, NC) (hereafter InVision^®) a plasma-based, somatic comprehensive genomic profiling (CGP) test for patients with advanced (Stage IIIB/IV) non-small cell lung cancer (NSCLC):

• At diagnosis
• When results for EGFR single nucleotide variants (SNVs) and insertions and deletions (indels); rearrangements in ALK and ROS1; and SNVs for BRAF are not available AND when tissue-based CGP is infeasible [i.e., quantity not sufficient (QNS) for tissue-based CGP or invasive biopsy is medically contraindicated],

or

• At progression
• For patients progressing on or after chemotherapy or immunotherapy who have not been tested for EGFR SNVs and indels; rearrangements in ALK and ROS1; and SNVs for BRAFs, and for whom tissue-based CGP is infeasible; or
• For patients progressing on EGFR tyrosine kinase inhibitors (TKIs).

If no genetic alteration is detected by InVision^® or if circulating tumor DNA (ctDNA) is insufficient/not detected, tissue-based genotyping should be considered.

or

• At progression
  • For patients progressing on or after chemotherapy or immunotherapy who have not been tested for EGFR SNVs and indels; rearrangements in ALK and ROS1; and SNVs for BRAF, and for whom tissue-based CGP is infeasible;
  • For patients progressing on EGFR tyrosine kinase inhibitors (TKIs).

If no genetic alteration is detected by InVision or if circulating tumor DNA (ctDNA) is insufficient/not detected, tissue-based genotyping should be considered.

Background

It is estimated that more than 222,500 new cases of lung cancer will be diagnosed in the United States (US) this year.¹ This represents roughly 13% of all new cancer diagnoses and 26% of cancer deaths.¹ At least 87% of lung cancer is NSCLC.² The estimated 5-year survival rate for all NSCLC cancer patients is 17%, and only 4% for patients with advanced (stage IIIB/IV) disease.³

The pathophysiological development of lung cancer is complicated, with several known genomic alterations found individually or in combination in many patients. These alterations may be due to toxic exposure or underlying genetic factors, and not all alterations have the same impact on disease development or prognosis. Some alterations appear to be integral to the transformation and ongoing growth of the tumor (driver alterations).

Among the best-studied genomic alterations are EGFR SNVs and indels and EML4-ALK rearrangements/fusions. EGFR-mutated NSCLC comprises up to 15% of all NSCLC patients in the US, with higher prevalence in certain ethnic groups (e.g., 40% in Asian Americans and 26% in Latin Americans).⁴ These mutations convey a more favorable prognosis and predict response to treatment with oral EGFR inhibitors such as erlotinib, gefitinib, osimertinib or afatinib. Rearrangements of ALK and EML4, or with other less common fusion partners, occur in approximately 4% of all NSCLC patients and predict response to treatment with oral ALK-targeted inhibitors such as crizotinib, ceritinib, or alectinib.⁵ Recently, dabrafenib in combination with trametinib has been approved for BRAF V600E positive metastatic NSCLC.⁶

Genomic alterations in NSCLC vary by smoking history, ethnicity and age. Sequencing of tumor specimens in never-smokers demonstrates a higher mutation prevalence of EGFR than in smokers. Some non-smoking ethnic groups, such as Asian women, have a much higher mutation prevalence than their Caucasian counterparts.⁷ Prevalence of ALK rearrangements is also higher in non-smokers.⁸ In contrast, smokers have a higher prevalence of targetable alterations in the MET and BRAF genes.^9-10

Tumor Tissue Genotyping

Failure of oncologists to order genotyping, inadequate quantity or quality of tissue specimen, and the necessity for repeat invasive biopsies with their associated complications and costs are just a few issues that confound tumor-tissue genotyping. Traditionally, tumor genotyping has been conducted by direct interrogation of tumor tissue obtained through invasive tissue sampling procedures. However, this diagnostic approach is limited by the availability of sufficient tumor tissue and the ability of patients to undergo invasive procedures.

In a recent study of more than 100 community-based oncologists, nearly one-third of NSCLC patients were not tested for EGFR or ALK mutations, and more than 75% were not tested for ROS1 fusions. Fewer than 10% of NSCLC patients were tested for all guideline-recommended alterations.¹¹ These results are similar to a study in a single academic center where 58% of non-squamous NSCLC were tested for EGFR and 40% for ALK fusions, despite repeat invasive biopsies to obtain sufficient tissue for genomic testing in 13% of patients.^12-13 Tissue availability was similarly limited in several recent series, some of which reported that more than 50% of NSCLC patients had insufficient or unobtainable material for tissue-based CGP.^14-16

Even when adequate tissue for next-generation sequencing (NGS) testing is available for testing, many specimens do not yield a complete result for a variety of reasons. Pre-analytical variables in tissue preservation are known to impact the quality and success of the NGS testing. Some of these variables include tissue fixation and processing variables, the volume of tissue (needle biopsy or resection specimen) available for testing, and the fraction of tumor cells within the specimen. Evaluating somatic mutations in formalin-fixed, paraffin-embedded (FFPE) tissue below 5% allele frequency is challenging due to these pre-analytical variables.^17-19

Recently, the Oncomine^TM Dx Target Test (Thermo Fisher Scientific Inc., Waltham, MA) and FoundationOne^TM CDx (Foundation Medicine, Cambridge, MA) tissue-based NGS panel assays received Federal Drug Administration (FDA) approval and Medicare coverage as alternatives to multiple, individual FDA-approved companion diagnostics (CDx).^20,21 The Oncomine™ panel is considered the least tissue intensive of the CGP tissue profiling panels. In the Oncomine™ clinical validation studies submitted for FDA approval, only 60% of samples had sufficient tissue for successful testing. The Oncomine™ validation set included surgical resections specimens that generally have a higher tumor load than diagnostic biopsies and may represent an over-estimate of successful testing specimens.²⁰

Currently, a variety of techniques are used to test for genomic alterations in plasma specimens to determine if a patient is a candidate for targeted therapy, including the FDA-approved Cobas^® EGFR Mutation Test (tissue or plasma samples) for erlotinib and osimertinib. This assay interrogates specific regions in EGFR to determine whether the genomic alteration of interest is present.²² For various reasons, these CDx and other existing lab developed test (LDT) techniques may miss deleterious EGFR mutations, ALK rearrangements, and other genomic alterations that can be targeted with FDA-approved drugs, though efficacy data for the patient’s specific indication may be limited. For example, alterations may occur outside the sequenced region or involve complex alterations (e.g. indels, copy number alterations, or rearrangements) that are not detectable by certain tests.²²

Within the InVision® clinical validation studies, only 33% of the prospectively recruited NSCLC patients had sufficient tissue for complete CGP. The remaining 67% either had no tissue for genomic analysis (31%) or had only enough tissue for some, but not all markers required (36%). This data underscores the marked limitation of available tissue specimens for tissue CGP testing and emphasizes the importance of plasma-based CGP testing.

Even when successful, tissue acquisition procedures pose a significant morbidity and mortality risk to Medicare patients. In a recent report, 19% of all lung tissue acquisition procedures resulted in a serious adverse event.²³ The National Lung Cancer Screening Trial reported 1-2% mortality rates in their cohorts.²⁴

Given that the majority of lung cancer diagnoses are based on needle biopsy, and that only 30%-60% of tissue specimens provide full informative results by CGP, plasma-based CGP (ctDNA testing) identifies genetic alternations for use of targeted therapies without delay in therapy,²⁵ and without the risks and costs of repeat invasive biopsy.^23,26  InVision® detects genomic targets linked to targeted drug therapies used at diagnosis and/or progression with response rates similar to those patients identified using tissue-based CGP and tissue-based CDx.

InVision® Test Description and Performance

InVision® is a plasma-based circulating tumor DNA (ctDNA) NGS assay for detection of genomic alterations consisting of 36 commonly mutated genes. It utilizes technology first developed by the Cancer Research UK (CRUK)-funded Cambridge Institute at the University of Cambridge.^27-29 The group was first to publish industry standard ctDNA methods, including hybrid capture and the highly sensitive tagged amplicon, deep sequencing or TAm-Seq^TM technology. The InVision® assay utilizes an enhanced version of the TAm-Seq™ method developed by Inivata™ to detect clinically relevant cancer mutations of low allele fractions in cell free DNA (cfDNA) including substantial improvements and optimizations to maximize sensitivity and specificity of the assay.²⁸

Approximately 76% of patients with NSCLC are known to have a genomic alteration in tumor tissue for 1 of 8 genes (EGFR, ALK, ROS1, BRAF, MET, ERBB2, KRAS, STK11).³⁰ These alterations constitute actionable driver alterations (EGFR, ALK, ROS1, BRAF, MET, ERBB2 - rule-ins) associated with FDA-approved therapies or are recognized as mutually exclusive for actionable changes (STK 11 and KRAS-rule-outs). These alterations have not been described as significant mutations contributing to clonal hematopoiesis of indeterminate potential.³¹

Analytical Validation

The analytical validation of the InVision® assay was conducted according to the deliverables outlined in the Molecular Diagnostic Services Program (MolDx®) document M00135 v2.0. Using contrived samples and tested with multiple users, multiple reagent lots and across multiple days, the sensitivity, specificity, reproducibility and level of detection (LoD) is summarized for all 4 variant types in the test system: SNV, Structural Variants (SV), indel, and CNV. Using patient samples, orthogonal comparison to digital PCR/fluorescent in-situ hybridization (dPCR/FISH) techniques was generated for SNVs, indels and SVs. Interference of somatic mutation detection was investigated with both spiked EDTA or Streck BCT plasma with fragmented cell line DNA and detection was shown to be comparable.

In the analysis of the contrived sample sets (Table 1), the positive percent agreement (PPA) is 96.6% for SVs, 100% for SNVs, 97.4% for indels and 100% for CNVs. The positive predictive value (PPV) is 100% for SVs, 99.8% for SNVs, 100% for indels and 98.3% for CNVs. Specificity was shown to be acceptable for all variants. One CNV false positive was seen. No false positives were seen with indels or SVs. In normal donor analysis, 1 SNV is considered to be a false positive. Specific variants described in M00135 were further analyzed for specificity and were shown to be of high specificity (>99% negative percent agreement (NPA). Reproducibility for all variants as analyzed at the LoD region was shown to be acceptable within reagent lots, within operators, and overall.

Table 1. Analytical Performance for the InVisionFirst® assay

Sensitivity for both the detection of contrived samples and patient samples is shown in Table 2. The goals for meeting the lower 95% CI for the M00135 guidance was met at both the LoD region and at the region >3x or 2x LoD. Specific variants described in M00135 were further analyzed for sensitivity and were shown to be of high sensitivity (>97% PPA) in all variants (Table 3). Orthogonal testing,³² Table 4, showed very good agreement with dPCR with the region of greatest disagreement between the 2 technologies occurring at the LoD90-3xLoD90 region. Above the 3xLoD90 region the agreement between dPCR and NGS was 100%. 8/9 ALK or ROS1 fusions were detected where the tissue was shown to be FISH positive for the fusions.³²

Table 2- Performance Characteristics - All Variants Tested Contrived Samples

Table 3- Performance Characteristics – Specific Variants Tested Contrived Samples

Table 4- Performance Characteristics - Orthogonal Testing¹ (dPCR SNV/Indel; FISH SV), All Variants Tested¹

¹ CNA not tested with orthogonal method.

² Neither non-NGS method or NGS method returns quantitative values.

Clinical Validation

The InVision® test was investigated prospectively in advanced untreated patients with non-squamous NSCLC blood samples.³³ Clinical validation data consists of combined analysis of 3 studies. Two prospective multicenter studies (NCT02906852 and NCT03116633) demonstrated the concordance of the InVision® assay with tissue-based CGP in 254 patients with untreated advanced (stage IIIB/IV) non-squamous NSCLC. A third study consisted of a small group of banked matched tissue and plasma samples (n=10) from an equivalent patient population that were procured from a commercial bio-repository and used to supplement the prospective collections. Across the 264 patients, only 165 patients (62.5%) had tissue available for testing for any point mutations/indels. For 159 patients (60.2%), tissue was tested for ROS1 and/or ALK fusions. One hundred nineteen patients (45% of patients) underwent CGP.

The performance of InVision® is highlighted by the following factors:

• High Sensitivity: for each of the key 8 genes used for therapeutic treatment decision with an overall plasma sensitivity of 73.9% (73.9% of tissue results are identified in plasma);
• High Proportion of Informative Results: based on the utilization of 8 specific genes (present in tumors independently of each other occurring in 53% of NSCLC) each with individual high gene sensitivity;
• High Specificity: for each of the 8 panel genes that supports correct therapeutic determination (CTD) in the 53% of patients with informative results.

Based on literature evidence of the prevalence of the genetic alterations in the 8 genes of most interest (76%) and an assumed 70% clinical sensitivity for ctDNA testing, the authors predicted an informative result in 50% of patients with untreated advanced NSCLC (0.76x0.7). Clinical CTD performance based on InVision® profiling results yielded actionable genes in 18.2% of patients and rule-out findings in 35.6%, an informative result in 53.8% of patients (95%; CI 41%-56.2%) (Table 5). Of the plasma positive gene results, when either a molecular change in the 6 actionable driver genes (rule-in), or a non-targetable gene (rule-out) was detected, the CTD was 100%. This clinical validity is consistent across the entire intended use population (n=264), both those with and without tissue for profiling.

Table 5: InVision® - ctDNA Clinical Validity

Performance characteristics for clinically actionable alterations in 8 genes that effect clinical patient management were: PPV-97.8%, NPV-97.1%, sensitivity-73.9% and specificity-99.8% (Table 6).

Table 6: Concordance of Combined and Individual Actionable Driver Genomic Alterations

*“Key 8 genes” refers to the combination of all directly actionable mutations (ALK/ROS1 fusions, BRAF V600E, EGFR exons 18-21, ERBB2 insertions, MET exon 14 splice) and KRAS and STK11 variants.

Clinical Utility

Clinical utility has been demonstrated with prospective outcome collection from within the clinical validation study and within additional studies at the Institute Gustave Roussy (Paris, France) and Centre Leon Berard (Lyon, France), and in 3 groups of patients; namely,

1. Patients not exposed to any prior therapy and receiving targeted therapy directed by the assay
2. Patients with no prior targeted therapy but other therapy and targeted directed by the assay, and
3. Patients with prior anti-EGFR targeted therapy and now progressing with the specific osimertinib sensitive mutation T790M detected by the assay

As detailed recently by the FDA, time on treatment was used as the endpoint for clinical impact of targeted therapy.³⁴ When targeted therapies are used in patients without a specific target, average time on treatment is well under 2 months.³⁵ With current directed targeted treatment in NSCLC, it is unusual to come off therapy before 3 months.²⁹ Regardless of which group mentioned above was assessed, disease control at 3 months was approximately 80% or more, which is equivalent to the best outcomes of any target agent reported. This is strong evidence that therapeutic determination based on InVision® results is equivalent to outcomes reported in clinical trials, and most importantly, unlikely to be causing patient harm (Table 7).

Implied clinical utility of Inivatas™ prospective clinical validation study (NCT02906852) and 3 unpublished internal studies identified actionable alterations in patients who received FDA-approved drugs and tracked clinical outcomes (Table 7).

Table 7: Actionable Genomic Alterations Detected by InVision®:

Patients Treated with Appropriate Targeted Therapy and Remaining on Therapy at 3 Months

Professional Society Clinical Practice Guidelines

National Comprehensive Cancer Network (NCCN) clinical practice guidelines (v4.2018) for non-small cell adenocarcinoma recommend a broad molecular profile panel. NCCN recommends molecular testing in never-smokers regardless of histology or mixed histology, and in small biopsies with the goal of identifying rare driver mutations for which effective drugs may be available. Tissue profiling is recommended to include EGFR and ERBB2 point mutations and indels; BRAF mutations; ALK, ROS1, and RET rearrangements; and MET amplification and deletion/skipping of exon. The guidelines indicate that if tissue biopsy is not feasible, plasma biopsy should be considered. If plasma biopsy is negative, then repeat tissue biopsy is recommended, if feasible.³⁶

Level of Evidence:

Quality – Moderate
Strength – Limited
Weight – Limited

The InVision® assay provides a minimally invasive methodology to detect actionable mutations with an informative test rate of 50-70% for providing valuable guidance for patient genomic profile stratification.  Clinical utility has been demonstrated equivalent to tissue-based profiling outcomes.  Patients are limited to 1 test assay per primary diagnosis.

N/A

N/A

1. Siegal RL, Miller KD, Jemal A. Cancer statistics, 2017. CA Cancer J Clin. 2017;67(1):7-30.
2. Govindan R, Page N, Morgensztern D, et al. Changing epidemiology of small-cell lung cancer in the United States over the last 30 years: analysis of the surveillance, epidemiologic, and end results database. J Clin Oncol. 2006;24(28):4539-4544.
3. Owonikoko TK, Ragin C, Chen Z, et al. Real-world effectiveness of systemic agents approved for advanced non-small cell lung cancer: a SEER-Medicare analysis. Oncologist. 2013;18(5):600-610.
4. Arrieta O, Cardona AF, Martín C, et al. Updated frequency of EGFR and KRAS mutations in nonsmall-cell lung cancer in latin america: the latin-american consortium for the investigation of lung cancer (CLICaP). J Thorac Oncol. 2015;10(5):838–843.
5. Chia PL, Mitchell P, Dobrovic A, John T. Prevalence and natural history of ALK positive non-small-cell lung cancer and the clinical impact of targeted therapy with ALK inhibitors. Clin Epidemiol. 2014;6:423-432.
6. Planchard D, Besse B, Groen H, et al. Dabrafenib plus trametinib in patients with previously treated BRAF^V600E-mutant metastatic non-small cell lung cancer: an open-label, multicentre phase 2 trial. Lancet Oncol. 2016;17(7):984–993.
7. Li S, Choi YL, Gong Z, et al. Comprehensive characterization of oncogenic drivers in asian lung adenocarcinoma. J Thorac Oncol. 2016;11(12): 2129–2140.
8. Dong Y, Ren W, Qi J, et al. EGFR, ALK, RET, KRAS and BRAF alterations in never-smokers with non-small cell lung cancer. Oncol Lett. 2016;11(4): 2371–2378.
9. Caparica R, de Castro G, Gil-Bazo I, et al. BRAF mutations in non-small cell lung cancer: has finally Janus opened the door? Crit Rev Oncol/Hematol. 2016;101:32–39.
10. Tong JH, Yeung SF, Chan A, et al. MET amplification and exon 14 splice site mutation define unique molecular subgroups of non-small cell lung carcinoma with poor prognosis. Clin Cancer Res. 2016;22(12):3048–3056.
11. Gutierrez ME, Choi K, Lanman RB, et al. Genomic profiling of advanced non-small cell lung cancer in community settings: gaps and opportunities. Clin Lung Cancer. 2017;18(6):651-659.
12. Thompson JC, Yee SS, Troxel AB, 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(23):5772-5782.
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(8):e95–e97.
14. Hagemann IS, Devarakonda S, Lockwood CM, et al. Clinical next-generation sequencing in patients with non-small cell lung cancer. Cancer. 2015;121:631-639.
15. Villaflor V, Won B, Nagy R, et al. Biopsy-free circulating tumor DNA assay identifies actionable mutations in lung cancer. Oncotarget. 2016;7(41):66880-66891.
16. Thompson JC, Yee SS, Troxel AB, 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(23):5772–5782.
17. Dedhia P, Tarale S, Dhongde G, Khadapkar R, Das B. Evaluation of DNA extraction methods and real time PCR optimization on formalin-fixed paraffin-embedded tissues. Asian Pacific J Cancer Prev. 2007;8:55–59.
18. Srinivasan M, Sedmak D, Jewell S. Effect of fixatives and tissue processing on the content and integrity of nucleic acids. Am J Pathol. 2002;161(6):1961–1971.
19. Morris S, Subramanian J, Gel E, et al. Performance of next-generation sequencing on small tumor specimens and/or low tumor content samples using a commercially available platform. PLoS ONE. 2018;13(4):e0196556.
20. Oncomine^TM Dx Target Test. Summary of Safety and Effectiveness Data (SSED). PMA P170019. June 22, 2017. Accessed August 24, 2021.
21. FoundationOne CDx^TM. Summary of Safety and Effectiveness Data (SSED). PMA P170019. November 30, 2017. Accessed August 24, 2021.
22. cobas^® EGFR Mutation Test v2. Summary of Safety and Effectiveness Data (SSED). PMA P150044. September 28, 2016. Accessed August 24, 2021.
23. Heerink WJ, de Bock GH, de Jonge GJ, Groen HJM, Vliegenhart R, Oudkerk M. Complication rates of CT-guided transthoracic lung biopsy: meta-analysis. Eur Radiol. 2017;27:138–148.
24. Chudgar NP, Bucciarelli PR, Jeffries EM, et al. Results of the national lung cancer screening trial: where are we now? Thorac Surg Clin. 2015;25(2):145-153.
25. Lim C, Tsao MS, Le LW, et al. Biomarker testing and time to treatment decision in patients with advanced nonsmall-cell Lung cancer†. Ann Oncol. 2015;26(7):1415–1421.
26. Lokhandwala T, Bittoni MA, Dann RA, et al. Costs of diagnostic assessment for lung cancer: a medicare claims analysis. Clin Lung Cancer. 2017;18(1):pe27–e34.
27. Forshew T, Murtaza M, Parkinson C, et al. Noninvasive identification and monitoring of cancer mutations by targeted deep sequencing of plasma DNA. Sci Transl Med. 2012;4(136):136ra68.
28. Plagnol V, Woodhouse S, Howarth K, et al. Analytical validation of a next generation sequencing liquid biopsy assay for high sensitivity broad molecular profiling. PLoS ONE. 2018;13(3):e0193802.
29. Wan JCM, Massie C, Garcia-Corbacho J, et al. Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat Rev Cancer. 2017;17(4):223-238.
30. The Cancer Genome Atlas Research Network. Comprehensive molecular profiling of lung adenocarcinoma. Nature. 2014;511:543-550.
31. Coombs CC, Gillis NK, Tan X, et al. Identification of clonal hematopoiesis mutations in solid tumor patients undergoing unpaired next-generation sequencing assays. Clin Cancer Res. 2018;24(23):5918-5924.
32. Guibert N, Hu Y, Feeney N, et al. Amplicon-based next-generation sequencing of plasma cell-free DNA for detection of driver and resistance mutations in advanced non-small cell lung cancer. Ann Oncol. 2018;29:1049-1055.
33. Pritchett MA, Camidge DR, Patel M, et al. Prospective clinical validation of the InVisionfirst-lung circulating tumor dna assay for molecular profiling of patients with advanced nonsquamous non-small cell lung cancer. JCO Precis Oncol. 2019.
34. Zhou J, Rajeshwari S, Theoret MR, Mishra-Kalyani PS. Time to treatment failure (TTF) as a potential clinical endpoint in real-world evidence (RWE) studies of melanoma. J Clin Oncol. 2018;36(15):abstr 9578.
35. Mok TS, Yi-Long W, Thongprasert S, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med. 2009;361(10):947–957.
36. National Comprehensive Cancer Network. Non-small cell lung cancer. Accessed August 24, 2021.