Diagnostic and economic value of biomarker testing for targetable mutations in non-small-cell lung cancer: a literature review

Lung cancer is the most common cancer worldwide and is the leading cause of cancer death for men and the second leading cause for women [1]. Non-small-cell lung cancer (NSCLC) affects approximately 85% of all patients with lung cancer [2]. These patients typically have a poor prognosis, with 5-year survival rates of 24% for all NSCLC patients and of only 6% among those with metastatic disease [3].

Targeted therapy is an important treatment option for patients with NSCLC who have driver mutation-positive tumors. Current treatment guidelines recommend tyrosine kinase inhibitor therapy for patients with actionable driver mutations, including EGFR, BRAF and RAS mutations, ALK, ROS1, RET or NTRK fusions and MET tyrosine kinase abnormalities (i.e., high-level MET amplification and MET exon 14 skipping mutation) [4,5]. Genetic testing is thus required, either to guide appropriate selection of available therapies or to assess patient suitability for a clinical trial of a new therapy [6]. However, with each biomarker testing procedure performed using a biopsy from an individual patient, the amount of tissue available for further biomarker testing is reduced. As the number of potential genetic targets increases, prioritization of limited tissue is essential [7].

Strategies to ensure maximum testing yield from available tissue and to limit invasive procedures for the patient include multigene sequencing with next-generation sequencing (NGS) and liquid biopsy (LBx) techniques, respectively. NGS is tissue-sparing compared with conventional sequencing because it allows identification of a panel of genes using a single sample, but it has not replaced conventional sequencing despite progressive cost reduction [8]. LBx techniques can be used to test for circulating tumor DNA (ctDNA), circulating tumor cells, exosomes, platelets and microRNAs [9]. The role of these biomarker techniques in NSCLC, including their diagnostic and economic value, has not been clearly defined.

The overall objectives of this literature review were to assess the diagnostic evidence and economic impact of first, NGS versus single-gene testing and second, LBx compared with standard tissue biopsy (TBx) in adults with unresectable, advanced or metastatic NSCLC.

Methods

A literature review was conducted to identify publications related to NGS and LBx in adult patients (aged ≥18 years) with advanced, recurrent and/or metastatic NSCLC. The review was guided by the population, intervention, comparison, outcome, study type (PICOS) framework [10].

Data sources

Embase^® and MEDLINE^® were searched for records from 2015 to 2020 using targeted keyword searches (Supplementary Table 1). To supplement the database searches, conference abstracts (2017–2020) from American Society of Clinical Oncology (ASCO), European Society for Medical Oncology (ESMO), International Society for Pharmacoeconomics and Outcomes Research (ISPOR), American Association for Cancer Research (AACR), World Conferences on Lung Cancer (WCLC), International Association for the Study of Lung Cancer (IASLC) North American Congress on Lung Cancer (NACLC), IASLC Targeted Therapies and International Lung Cancer Congress were searched manually. Bibliographies of review articles were also searched manually, and the Tufts Cost–Effectiveness Analysis (CEA) Registry was searched for economic evidence [11,12].

Study selection

Eligible studies were controlled clinical trials (including both randomized and non-randomized), single-arm studies, observational studies (excluding case reports/series), systematic reviews, surveys and economic evaluations published in English. Studies had to include ≥100 adult patients of any race or sex with unresectable, advanced and/or metastatic NSCLC, regardless of histology, for any biomarker or mutation. Any study meeting the above criteria and reporting biomarker types, molecular testing methods and application, and their associated challenges and advantages in NSCLC, were considered.

Review procedure

All articles were downloaded into a systematic review database. First screening (titles and abstracts) and second screening (full-text papers) were conducted by a single reviewer, followed by a quality check from a second independent reviewer (Supplementary Figure 1). Data were extracted by a single reviewer and verified by an independent reviewer. Where more than one publication was identified describing a single trial, the data were compiled into a single entry in the data extraction table to avoid double counting of patients and studies.

Results

Literature review

A total of 2602 records were identified, from which 375 full-text studies were screened (Figure 1). Of these 375 publications, the review included 43 relevant publications describing 42 studies. Fourteen studies examined NGS versus standard molecular testing techniques in NSCLC. Twenty-eight studies reported evidence on LBx versus TBx (29 publications describing 28 studies).

NGS versus standard molecular testing

Six studies, two of them USA-based, reported diagnostic outcomes [13–18], while nine (five USA-based) reported economic evidence [17,19–27]. Of these, one study conducted in Singapore by Tan et al. reported both diagnostic and economic outcomes [17].

Diagnostic evidence

All studies reporting diagnostic outcomes were observational; two were retrospective and four were prospective. Among studies reporting demographic data (n = 4 studies with a range of 174–533 patients), the median age of patients ranged from 67 to 70 years, and 38–62% of patients were male [13,15–17]. Two studies from Europe [15,16] and one from Asia [17] reported concordance rates ranging from 70% to 99% for NGS versus standard molecular testing across clinically actionable mutations, including EGFR and ALK fusion (Table 1). Sensitivity for NGS versus standard molecular testing ranged from 86% to 100% for clinically actionable mutations and was reported as 55.6% in one study based on acquired resistant mutations [13–15,17]. Of the two USA-based studies included, one (Dagogo-Jack et al.) reported sensitivity and specificity of 98% and 100%, respectively, for a rapid EGFR PCR assay using NGS as the reference assay [13]. Other studies included did not report specificity data. The second USA-based study (Yu et al.) reported higher rates of test initiation and completion using less tissue compared with single-gene testing for four or more biomarkers [18]. Based on the two USA-based studies, median turnaround times were longer using NGS than with single-gene testing (14–17 vs 7–11 days), but this was not the case if multiple sequential single-gene tests were required (e.g., three single-gene tests ordered in sequence would require ∼21–∼33 days total turnaround time) [13,18].

Economic impact

Among the ten economic studies, six assessed cost–effectiveness [17,19,23–26], one reported costs [27], two assessed budget impacts [20,21] and one reported a cost–consequence analysis (Table 2) [22].

Among the five USA-based studies that reported economic evidence, two found tumor tissue NGS versus sequential exclusionary testing or hotspot panel testing (excluding treatment costs) to be cost saving [20,22], with one (Dalal et al.) reporting the shortest wait time, earlier initiation of effective targeted therapy and lower costs in patients receiving upfront NGS [22]. Two studies found tumor tissue NGS versus single-gene testing (including treatment costs) to be associated with increased budget, although that was balanced by evidence of cost–effectiveness for NGS testing [19,21]. In the study conducted by Yu et al. the budget increase over single-gene testing was minimal, at US$0.0072 per member per month, but NGS was expected to identify more patients with activating mutations, enabling better selection for targeted therapy [21]. Another study, by Steuten et al., found ctDNA NGS versus standard of care molecular testing (no additional genomic testing) to be cost-effective, with 8% more patients identified with targetable mutations and expected survival increasing by 0.06 years versus single-gene testing [19].

Studies conducted in other regions (e.g., Europe, Asia) were aligned with USA-based studies regarding economic and diagnostic outcomes. In Spain, Simarro et al. reported that NGS implementation was feasible and could be done at reasonable cost [27]. In Singapore, Tan et al. found that routine upfront NGS was cost-effective compared with sequential sequencing [17].

LBx versus TBx

Of the 25 studies included that compared LBx with TBx (both NGS and single-gene assays) [28–52], four were from the USA [32–35], five were from Europe [36–40], 12 were from Asia [41–52] and four were global [28–31]. The majority (88%) of diagnostic studies were observational. Among studies reporting demographic data (n = 16 studies with a range of 102–1026 patients), the median age of patients ranged from 57 to 70 years (n = 11), and 32–70% of patients were male (n = 16) [32–38,40,45–51].

Three studies comparing economic outcomes between LBx and TBx were a cost study by Arnaud et al. from the USA [53], a cost–consequence analysis from Italy by Gancitano et al. [54] and a Canadian cost–effectiveness and budget impact analysis by Ontario Health [55].

Diagnostic evidence

Measures of diagnostic accuracy, including sensitivity, concordance, positive predictive values (PPVs) and negative predictive values (NPVs), for LBx versus TBx in detecting clinically actionable genes were examined across studies (Table 3).

The range of specificity values reported across mutations and across studies was 42.5–100%, with 11 of 17 studies reporting a specificity ≥90% for all tests [28,30–33,35–37,40–43,47–49,51,52]. Specificity values <90% for EGFR_T790M were observed in two of two studies reporting specificity specifically for this variant [28,31]. Sensitivity values varied widely across mutations and across studies (range: 0–100%), and 14 of 18 studies reported sensitivity of ≥60% for all tests [28,30–33,35–37,40–43,47–52].

Concordance rates using LBx to detect targetable mutations were reported in 18 studies [28–30,34–43,45–49]. Concordance rates for all mutations tested were ≥70% in 16 studies and >90% in seven [28–30,35–37,39–43,45–49]. In ten studies reporting PPV for LBx versus TBx, the range of PPVs for clinically actionable mutations was 77.8–100% [32,34–37,41,45,47,48,51].

In two USA-based studies, 100% PPV was reported for all clinically actionable mutations except EGFR_T790M (79%) [32,35]. In seven studies that reported NPV of LBx versus TBx, the range of NPVs for clinically actionable mutations was 52–98% [34,35,41,45,47,48,51]. Faster turnaround times were reported for LBx versus TBx (range across three studies: 2–10 vs 5–25 days) [29,32,35].

Economic impact

Three economic studies found that incorporating LBx into the treatment pathway was associated with lower testing cost per patient (Table 4) [53–55].

A USA-based study conducted from a Medicare reimbursement perspective compared the clinical costs and complications of LBx with TBx (computed tomography [CT]-guided biopsy and navigational bronchoscopy) in a hypothetical biomarker case that tested for EGFR, KRAS, ALK and BRAF mutations [53]. The authors found that LBx was significantly less expensive, having a lower total cost of biopsy and biomarker testing (US$836) compared with CT-guided biopsy (US$4130) and navigational bronchoscopy (US$8284). Overall, LBx was associated with cost reductions of ≥US$3200 per patient. LBx also reduced time from screening to treatment and was associated with fewer complications for patients compared with CT-guided biopsy or navigational bronchoscopy, which was associated with pneumothorax, lung collapse, hemorrhage and respiratory distress in some patients.

A cost–consequence analysis was performed in a hypothetical cohort with EGFR mutations in Italy [54]. This model compared three different diagnostic pathways: TBx (for first- and second-line treatment), combined (TBx for first line and LBx for second line if the outcome was unknown) and potential (TBx or LBx for tissue-ineligible patients as first line and LBx as second line). The potential pathway provided the greatest number of correctly identified cases. The average cost per correctly identified case in the potential pathway (€685) was lower than for the combined pathway (€732) or the TBx pathway (€1004). Overall, the addition of LBx was associated with a cost reduction of approximately €300 per patient in this setting. These authors concluded that a correct diagnostic pathway is essential to optimize cancer therapies. This analysis also highlights the value of upfront LBx in the diagnostic pathway for both first- and second-line treatment.

A recent Canadian Health Technology Assessment reported the cost–effectiveness of TBx alone, LBx alone or LBx as a triage test in a cohort with EGFR_T790M mutations [55]. LBx alone or as a triage test was less costly and more effective (i.e., resulted in fewer tissue biopsies and more correct decisions) than TBx in terms of test-related costs and effects only. With regard to lifetime costs, LBx as a triage test was the most effective and produced the most life-years and quality-adjusted life-years, but it was the costliest of the three options. However, this was largely driven by higher long-term treatment cost as more patients were correctly identified to receive targeted therapy [55].

Discussion

This targeted literature search conducted from January 2015 to March 2020 identified studies reporting diagnostic and economic aspects of NGS versus single-gene testing and LBx versus TBx. The studies were global, with most conducted in North America (including the USA and Canada), followed by Europe and Asia. Overall, the studies suggest that NGS is highly concordant with conventional molecular testing in patients with NSCLC. Two studies reporting <90% concordance (55.6% [14] and 82% [37]) between NGS and single-gene tests used specialized single-gene tests for EGFR testing, and the discordances (missed by NGS) were mainly in the low mutant allelic fractions. While less sensitive than conventional methods, NGS resulted in broader genomic coverage, which may reveal diverse mechanisms of resistance among patients with advanced NSCLC.

NGS can also measure tumor mutational burden (TMB), an emerging biomarker to select patients for immunotherapy, and will likely need to be used in conjunction with PD-L1 immunohistochemistry [5]. The sequencing of targeted therapies and immunotherapies as recommended in treatment guidelines will continue to evolve as the treatment landscape changes, but clinicians may make use of the wider range of genetic information available with NGS to facilitate the selection of the right therapy for an individual patient. The potential role of additional genetic screening – via NGS or single-gene testing – in a patient whose disease develops resistance to initial therapy needs to be clarified in future studies.

In terms of cost and cost–effectiveness, NGS leads to a greater proportion of patients assigned to targeted therapy and increased life-years gained while being cost neutral or cost saving. NGS was generally found to be cost-effective at typical thresholds. With current treatment guidelines recommending targeted therapies for eight specific genetic biomarkers plus additional recommendations based on PD-L1 and TMB status [5], the additive costs of multiple single-gene tests should be considered. This review provides indirect evidence on the question of how the costs of NGS compare with those of multiple single-gene tests, and future studies on the costs of biomarker testing in NSCLC may provide clearer evidence.

In all studies, concordance between LBx and TBx for all mutations tested was generally high (≥70%), with six studies reporting >90% concordance. LBx exhibits high specificity to detect targetable mutations in patients with NSCLC, but it may have lower sensitivity than TBx. Overall, the LBx studies reported shorter turnaround times from blood sample collection to report delivery compared with TBx. The faster turnaround time and high PPVs of LBx enable faster treatment decisions in patients with NSCLC who have targetable mutations. LBx may also provide additional genetic material for subsequent testing at the time when a patient develops resistance to initial therapy, limiting invasive procedures and potentially improving patient experience and outcomes. Upfront cost savings may be achieved using LBx as an initial screening method in complement to TBx, although identification of a greater number of cases may lead to increased treatment costs.

Several limitations to this review should be noted. This review includes only English-language papers published as journal articles in the last 5 years. While unpublished or non-English-language studies may contain valid results that may conflict with the conclusions of this review, the broad search strategy used here and the large number of citations screened make this unlikely. The review excludes diagnostic studies with small sample size (≤100 patients); however, these are likely to be exploratory studies that could introduce bias. Additional parameters that inform the quality of biomarker testing (e.g., test failure rate) were not included in the data extraction. Only three economic evaluations were found comparing LBx and TBx, suggesting that data may be limited in this area.

Future perspective

The findings of our review may have implications regarding recommendations for the timing of LBx and NGS in future NSCLC treatment guidelines. Among studies included in this review, the results of NGS and conventional single-gene testing were highly concordant. Comparisons of TBx and LBx indicated that these techniques also generally have high concordance. NGS and LBx separately showed benefits in terms of correctly identifying more patients for targeted therapy, enabling faster turnaround and quicker treatment decisions. In terms of cost and cost–effectiveness, these methods were associated with reductions in short-term treatment-related costs. In the long term, increased use of NGS could result in a minimal increase of budget largely driven by more patients receiving targeted treatment.