Bronchoscopic tissue yield for advanced molecular testing: are we getting enough?

The treatment of advanced lung cancer has become increasingly personalized over the past decade as a result of the improved understanding of tumor molecular biology and anti-tumor immunity. An adequate tumor sample is central to targetable mutation analysis, and immunologic profiling. The majority of lung cancer patients currently present at an advanced disease stage, so that diagnosis and staging are largely based on small biopsy and cytology specimens. Flexible bronchoscopy techniques play a prominent role in the acquisition of these diagnostic specimens. This narrative review summarizes the available evidence with regards to the role of various conventional and advanced flexible bronchoscopy techniques in acquiring sufficient tissue for mutation analysis and programmed death-ligand 1 (PD-L1) testing. 2020 Journal of Thoracic Disease. All rights reserved.

Introduction

The treatment of lung cancer has evolved beyond the simple histologic distinction of small cell and non-small cell lung cancer (NSCLC). Novel treatments for NSCLC target specific gene alterations. Trials have demonstrated the significant benefit of such targeted therapies over standard cytotoxic chemotherapy. Patients with epidermal growth factor receptor (EGFR) mutations, typically on exons 19 and 21, exhibit dramatic responses to specific tyrosine kinase inhibitors (TKI) such as gefitinib, erlotinib, afatinib and osimertinib (10–15% of NSCLC) (1,2). Rearrangement of the anaplastic lymphoma kinase (ALK) gene (5% of NSCLC) predicts response to treatment with ALK inhibitors such as crizotinib, ceritinib, alectinib, and brigatinib (3). ROS-1 rearrangement (1–2% of NSCLC) also predicts response to crizotinib (4). KRAS mutations, detected predominately in smokers with adenocarcinoma, are associated with brain metastasis and a worse prognosis (5). KRAS inhibitors are currently undergoing early phase clinical trials.

Over the past decade, the understanding of anti-tumor immunity has remarkably improved. Interaction of the programmed death-ligand 1 (PD-L1) with its receptor (PD-1), has been recognized as an essential immune escape mechanism for tumor cells, and is now targeted in lung cancer treatment. The level of PD-L1 expression on tumor cells, as evaluated by immunohistochemistry (IHC), currently guides immunotherapy treatment decisions. Several immune checkpoint inhibitors are used in the treatment of advanced NSCLC (6). Patients with advanced NSCLC and PD-L1 expression on at least 50% of tumor cells (a.k.a. tumor proportion score or TPS 50%) benefit from first-line treatment with the anti-PD-1 agent pembrolizumab, with improved survival and reduced adverse effects compared to standard chemotherapy (7). Pembrolizumab and platinum based chemotherapy is indicated as first-line treatment in patients with PD-L1 1–49% (8,9). Pembrolizumab can also be administered as second-line treatment in the presence of PD-L1 expression on at least 1% of tumor cells (10).

Advanced molecular testing, to identify targetable mutations and quantify PD-L1 expression, thus guides the personalized treatment of lung cancer. The majority of patients with advanced lung cancer are diagnosed and staged with minimally invasive techniques, and the resultant small biopsy and cytological samples must be judiciously processed for molecular testing. Small biopsy and cytopathology specimens also play a central role in the context of disease progression after treatment (11). The selection of a biopsy site is based on review of available CT and/or PET imaging, keeping in mind that invasive sampling should target the site associated with the highest suspected disease stage, whenever feasible (12,13). Bronchoscopic procedures must successfully sample target lesions (node, nodule or mass) and provide sufficient tissue for both histological characterization, and molecular analysis. This narrative review summarizes the available evidence with regards to the role and limitations of various conventional and advanced flexible bronchoscopy techniques in acquiring sufficient tissue for mutation analysis and PD-L1 testing.

Molecular testing of small biopsy and cytology specimens

The increasing role of small biopsy and cytology specimens was recognized in the 2015 World Health Organization Classification of Lung Tumors (14). The most recent guidelines from the College of American Pathologists (CAP), the International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) recommend molecular testing at the time of initial diagnosis in patients with advanced lung adenocarcinoma regardless of clinical characteristics (15). The NCCN has expanded the indication for EGFR and ALK testing to non-squamous NSCLC and NSCLC not otherwise specified (NOS), as well as patients with metastatic squamous cell lung carcinoma who are non-smokers (6). A basic panel, including EGFR mutation, ALK rearrangement, mutations in V-raf murine sarcoma viral oncogene homolog B (BRAF), and ROS-1 mutation, is recommended for routine testing in lung adenocarcinomas. An expanded panel with RET, ERBB2 (HER2), KRAS, and MET can be tested in the context of clinical trials (15).

The accuracy of molecular profiling depends on the quality of tissue samples, both in terms of the quantity and quality of tumor cells acquired. Adequate technique for both specimen collection and processing are essential, and collaboration between proceduralists and pathologists is required (16). The required number of tumor cells for molecular analysis has not been clearly established, although studies have suggested a range of 100–400 tumor cells (17,18). Sun et al. reported that cytological specimens are comparable to histological samples, as long as they provide DNA concentrations >25 ng·mL-1, >30 tumor cells, or >30% tumor cellularity (19). Cytology cell blocks are preferred to simple smears for the purpose of molecular analysis (20). Small biopsy and cytological samples generally provide lower amounts of DNA, so that sensitive methods are required to detect targetable mutations.

The CAP/IASLC/AMP guideline recommends using sensitive analytical assays that can detect mutations with tumor cell proportions as low as 20% of the test sample (15). The NCCN guideline lists a range of methods for tissue genotyping including next-generation sequencing (NGS), real-time polymerase chain reaction (PCR) for specific targetable mutations, Sanger sequencing, other multiplex approaches (e.g., SNaPshot, MassARRAY), fluorescence in situ hybridization (FISH) for detecting gene rearrangements, and immunohistochemistry (IHC) for certain analyses or (mutation) screening. Sanger sequencing, the first generation of DNA sequencing, examines one or a few genes at a time, and requires more than 25–30% of tumor cells (6). In contrast, NGS allows the detection of numerous mutations at the same time, making it particularly advantageous when limited tissue samples are available. The interested reader is referred to the website of the American Association for Clinical Chemistry for a user-friendly description of these techniques (https://labtestsonline.org/genetic-testing-techniques).

There is limited data on the molecular testing of samples acquired with conventional diagnostic bronchoscopy. Many studies have combined bronchoscopy samples with other small biopsy or cytology specimens, making specific evaluation more difficult. The percentage of malignant cells from bronchoscopic biopsies [endobronchial biopsy or radial-endobronchial ultrasound (R-EBUS) guided transbronchial lung biopsy (TBB)], endobronchial ultrasound with transbronchial needle aspiration (EBUS-TBNA), and surgical biopsies was examined by Dooms et al: 57% of diagnostic endoscopic biopsies, 44% of EBUS-TBNA samples, and 90% of surgical biopsies provided >25% tumor cells, which is considered adequate for Sanger sequencing. Surgical biopsies provided more DNA than bronchoscopic biopsies and EBUS-TBNA, with medians of 2,500 vs. 1,610 and 1,440 ng, respectively (21). Hagmeyer and colleagues examined the role of brush specimens acquired from central lung cancers for molecular testing using NGS. The combined brush smear and brush tip washing provided diagnostic sensitivity of 69%, and NGS was feasible in 18 of 29 samples (62%) (22). Perhaps not surprisingly, NGS has been shown to be more sensitive than Sanger sequencing for detection of EGFR mutations in both BAL and pleural fluid samples (23).

EBUS-TBNA samples and molecular testing

Labarca and colleagues recently reviewed the adequacy of EBUS-TBNA samples for molecular analysis in patients with NSCLC. A total of 33 studies, encompassing 2,698 participants, were included in their meta-analysis. The pooled probability of obtaining an adequate sample for EGFR and ALK testing was 94.5% and 94.9%, respectively, and the prevalence of EGFR mutation and ALK rearrangement were 15.8% and 2.8%, respectively (24). A small study from Chile reported 10 of 12 (83.3%) EBUS-TBNA specimens were adequate for ROS1 testing (25). A retrospective review by Cicek et al. showed 90 of 98 (91.8%) EBUS-TBNA samples were adequate for the ROS1 FISH test (26). Xie et al. showed 100% concordance for EGFR, ALK, and ROS1 results between NGS and conventional analytical assays in EBUS-TBNA samples, however NGS provided information on 12 additional mutations (27).

The recent CHEST Guideline on technical aspects of EBUS-TBNA recommends that additional samples, beyond the minimum of 3 separate needle passes acquired to establish the diagnosis of lung cancer, be obtained for molecular analysis (28). Labarca et al. suggested that the minimum number of needle passes per lymph node in order to obtain adequate tissue for molecular analysis was 3 with rapid on-site evaluation (ROSE) or 4 without ROSE (24). Trisolini et al. conducted a randomized trial to evaluate whether ROSE increased the yield of EBUS-TBNA for molecular analysis. The result showed 90.8% of EBUS-TBNA specimens from the ROSE arm and 80% from non-ROSE arm were suitable for genotypic analysis (P=ns). Patients in the ROSE arm were more likely to have the bronchoscopy terminated after a single biopsy site (29). The available evidence does not suggest a difference in diagnostic yield and/or feasibility of molecular analysis according to needle size used (21 or 22G), and data is still limited on the newer 19G needles (30,31). Larger size needles, however, have been associated with more blood in specimens (30).

Guided bronchoscopy samples and molecular testing

Molecular testing guides the treatment of patients with advanced lung cancer, in whom isolated peripheral pulmonary nodules are rarely targeted for sampling. Accordingly, there is limited data on molecular analysis of specimens acquired with guided bronchoscopy techniques. Moon et al. demonstrated the feasibility of molecular testing using R-EBUS-guided TBB peripheral lung lesions: 63 of 64 specimens (98%) were found to be adequate for EGFR testing, 60 of 60 (100%) adequate for ALK IHC, and 16 of 17 (94%) adequate for PD-L1 IHC (32). Kim et al. compared EGFR mutations and ALK rearrangements of paired R-EBUS-guided TBB and subsequent surgical resection specimens. The result demonstrated 97% agreement in EGFR mutations and 100% in ALK translocation (33). Guisier et al. retrospectively reviewed the adequacy of R-EBUS-guided TBB and brushings for molecular testing. Molecular analysis was feasible in 88/113 non-squamous NSCLC samples, with 73 of 86 diagnostic biopsy specimens (85%) being suitable for testing. Factors associated with successful molecular testing included upper or middle lobe location and at least three biopsies taken (34). However, when diagnosis was based on brushings (smears) alone, only 15/27 smears were suitable for molecular analysis. In contrast, two studies from Taiwan and Spain demonstrated a 95% success rate of molecular analysis from R-EBUS-guided “waste” brushing samples; the brushes were immediately snipped off and immersed into fixative solution (35,36). The NAVIGATE study examined the diagnostic yield of ENB-guided sampling for peripheral pulmonary lesions, and reported 86.2% of samples (75/87 specimens) were adequate for molecular/genetic testing (37). If tissue sampling is successful, then advanced molecular testing is highly feasible in samples from peripheral pulmonary lesions acquired using guided bronchoscopy techniques.

PD-L1 testing of small biopsy and cytology specimens

The current NCCN guidelines recommend quantitative evaluation of tumor cell PD-L1 expression at initial diagnosis of advanced NSCLC (6). PD-L1 expression is quantified by the percentage of membrane stained viable tumor cells relative to all viable tumor cells in the sample, known as tumor proportion score (TPS). A minimum of 50–100 viable tumor cells is required, depending on the assay used. Several antibodies and platforms for PD-L1 IHC testing have been developed and marketed, and lack of standardization remains a significant limitation. Clinical trials have used different PD-L1 assays and varying cut-off levels (38). Most clinical trials supporting the use of immune checkpoint inhibitors performed PD-L1 testing on formalin-fixed paraffin-embedded (FFPE) core biopsy or surgical resection specimens, however this practice does not reflect the clinical care of most patients with advanced lung cancer (39).

Concerns have been raised with regards to the use of cytology specimens to evaluate with immune landscape in NSCLC (38). Small biopsies and cytology samples could increase the risk of false-negative PD-L1 results, based on the known temporal and spatial heterogeneity of its tumoral expression (40,41). Bigras et al. evaluated the impact of small biopsy size on PD-L1 results in vitro. More biopsies <2 mm2 were associated with false-negative results (PD-L1 TPS <1%), while larger specimens more frequently demonstrated PD-L1 TPS 1–49%. However, the number of samples which detected PD-L1 TPS ≥50% was not different between small and large biopsies (42).

The concordance of PD-L1 expression as assessed by bronchoscopic small biopsies and surgical resection specimens has been examined. Kitazono et al. reviewed the PD-L1 expression of 79 patients in whom both small biopsy specimens and surgical resection samples were available. The bronchoscopic samples included 59 TBB, 12 TBNA, and 8 CT-guided needle biopsies. The concordance of PD-L1 expression was 92.4% concordance between small biopsies and surgical resections, using a hybrid IHC score assessment (43). The study of Ilie and colleagues provides contrasting results: 160 patients with both surgical specimens and small biopsies were considered, in whom 110 (69%) had TBB specimens. PD-L1 expression was underestimated by small biopsy specimens, with 48% overall discordance rate. However, overall concordance was evaluated based on PD-L1 expression in both tumor cells and tumor-infiltrating immune cells using the PD-L1 SP142 antibody clone (44).

Several recent studies have examined the feasibility of PD-L1 testing using small biopsy and cytological samples (43-46), in particular the use of EBUS-TBNA samples for PD-L1 IHC analysis (). Sakakibara et al. demonstrated that a significantly larger number of tumor cells were obtained with EBUS-TBNA compared with TBB, with median numbers of 1,149 vs. 435 (P<0.001); crush rate was significantly lower with EBUS-TBNA vs. TBB (47). Sakata et al. retrospectively compared the results of PD-L1 testing in EBUS-TBNA samples vs. surgical resection specimens in 61 patients and reported that sensitivity, specificity, PPV, and NPV for PD-L1 ≥1% were 72%, 100%, 100%, and 80%, respectively. The concordance rate between EBUS-TBNA vs. surgical specimens was 87% for PD-L1 ≥1% and 82% for PD-L1 ≥50%. However, the sensitivity of EBUS-TBNA samples declined from 72% to only 47% at a PD-L1 expression cut-off of ≥50% (vs. ≥1%), raising concerns of false-negative PD-L1 results in EBUS-TBNA specimens (48). In contrast, we have reported a recent series of PD-L1 testing in 120 consecutive EBUS-TBNA samples. Comparison of PD-L1 results in the subgroup of 18 patients with both EBUS-TBNA and histological samples revealed moderate concordance, with no false-negative results (50).

Table 1

PD-L1 testing on EBUS-TBNA specimens: feasibility and concordance with histology samples

Author, year	PD-L1 assay	TPS cut-offs	Number of EBUS-TBNA samples	Feasibility of PD-L1 testing	Number and type of paired samples	Agreement analysis	Comments
Sakakibara 2017 (47)	EPR1161	N/A	EBUS-TBNA 97	99%	Resected primary tumor 6, resected LN 5, TBB 16	Pearson correlation for lung resection r=0.75, for LN metastasis r=0.93, for TBB r=0.75	1 of 97 EBUS-TBNA sample had <100 tumor cells
Sakata 2018 (48)	22C3	≥1%, ≥50%	EBUS-TBNA 61	84%	Surgical resection 61	Concordance 87% for cut-off ≥1% and 82% for cut-off ≥50%	–
Wang 2019 (49)	22C3	≥50%	EBUS-TBNA	86.8%	Paired surgical biopsy 34	Concordance 91.3%	–
Smith 2019 (50)	22C3	≥50%	EBUS-TBNA	92%	Surgical resection 11, TBB 1, core biopsies 4, pleural biopsy 1, autopsy 1	Concordance 78%	–
Stoy 2018 (51)	28-8	<1%, 1–49%, ≥50%	EBUS-TBNA 16, endobronchial FNA 4, peripheral TBNA nodule 2	88% for 16 EBUS-TBNA samples	FOB biopsy 2 (same location), 1 (different location)	Concordance in 2 of 3 cases (for which cytology and histology were from same location)	–
Biswas 2018 (52)	22C3	<1%, 1–49%, ≥50%	EBUS-TBNA 50	86%	N/A	N/A	–
Fernandez-Bussy 2018 (53)	E1L3N	<1%, 1–50%, >50%	EBUS-TBNA 23	100%	N/A	N/A	–

Table adapted with permission from Smith et al. (50). PD-L1, programmed death-ligand 1; CT, computed tomography; EBUS-TBNA, endobronchial ultrasound with transbronchial needle aspiration; FNA, fine needle aspiration; FOB, flexible bronchoscopy; LN, lymph node; N/A, not available; TBB, transbronchial biopsy.

PD-L1 IHC testing has been approved in formalin-fixed paraffin embedded (FFPE) tissue samples (54). Concerns have been raised with regards to the use of other agents such as methanol-based fixatives for subsequent PD-L1 assessment. Recent studies have reported on the feasibility of PD-L1 testing in Cytolyte-fixed specimens (50,51), with two recent studies reporting concordance rates of 78% and 91%, respectively, for PD-L1 expression in EBUS-TBNA specimens initially processed with methanol-based fixatives vs. FFPE histological specimens (49,50). Similar to molecular testing, limited data is currently available on the role of ROSE (55) and needle gauge in ensuring adequate EBUS-TBNA samples are acquired for assessment of PD-L1 expression.

The available evidence thus suggests that PD-L1 testing of EBUS-TBNA samples is feasible, and points towards at least moderate concordance with PD-L1 expression assessed in histological samples. Future work will need to determine the role of PD-L1 expression as assessed in cytological samples for predicting response to immunotherapy.

Summary

The treatment of advanced lung cancer has become increasingly personalized over the past decade, as a result of improved understanding of tumor molecular biology and anti-tumor immunity. Advanced diagnostic bronchoscopy techniques play a central role in the evaluation of patients with suspected lung cancer. An adequate tumor sample is central to targetable mutation analysis and immunologic profiling. To the question “Are we getting enough?” the answer is “it depends”—mainly on the procedure being performed, but also partly on the analysis being requested. Communication between bronchoscopists and pathologists is key: clinicians should know which technique is being used locally for molecular profiling, minimum number of tumor cells required, as well as preferred samples types and preparation. Minimally invasive needle techniques such as EBUS-TBNA have become the first-line procedures in patients with adenopathy, and/or centrally located tumors. The available data suggests that EBUS-TBNA samples are largely adequate for molecular testing, and seem to provide a reliable assessment of PD-L1 expression. Advanced diagnostic bronchoscopy techniques for sampling of peripheral pulmonary lesions still suffer from a limited and variable diagnostic yield, which technological advances actively seek to remedy. The role of conventional diagnostic bronchoscopy techniques is similarly limited to centrally located lesion that can be sampled under direct vision. Technological advances in both bronchoscopy procedures and analytical processes are expected to further consolidate the role of small biopsy and cytological specimens for molecular testing in patients with advanced lung cancer.

The article’s supplementary files as