Key Points
Question
Does concurrent RNA next-generation sequencing (RNA-NGS) and DNA-NGS detect more actionable structural variants (aSVs) in patients with advanced lung cancer than DNA-NGS alone?
Findings
In this cohort study of 5570 patients with advanced non–small cell lung cancer who underwent concurrent RNA-NGS and DNA-NGS, 8.8% had at least 1 aSV detected by 1 or both assays. Concurrent RNA-NGS and DNA-NGS testing identified 15.3% more patients harboring an aSV compared with DNA-NGS alone.
Meaning
These results suggest that concurrent RNA-NGS and DNA-NGS testing should be used clinically to maximize structural variant detection relative to DNA-NGS alone.
This cohort study assesses clinical evidence from patients with advanced lung adenocarcinoma and compares the detection of National Comprehensive Cancer Network–recommended actionable structural variants (aSVs) via concurrent DNA next-generation sequencing (NGS) and RNA-NGS vs DNA-NGS alone.
Abstract
Importance
The National Comprehensive Cancer Network (NCCN) guidelines for non–small cell lung cancer suggest that RNA next-generation sequencing (NGS) may improve the detection of fusions and splicing variants compared with DNA-NGS alone. However, there is limited adoption of RNA-NGS in routine oncology clinical care today.
Objective
To analyze clinical evidence from a diverse cohort of patients with advanced lung adenocarcinoma and compare the detection of NCCN-recommended actionable structural variants (aSVs; fusions and splicing variants) via concurrent DNA and RNA-NGS vs DNA-NGS alone.
Design, Setting, and Participants
This multisite, retrospective cohort study examined patients sequenced between February 2021 and October 2023 within the deidentified, Tempus multimodal database, consisting of linked molecular and clinical data. Participants included patients with advanced lung adenocarcinoma and sufficient tissue sample quantities for both RNA-NGS and DNA-NGS testing.
Exposures
Received results from RNA-NGS and DNA-NGS solid-tissue profiling assays.
Main Outcomes and Measures
Detection rates of NCCN guideline–based structural variants (ALK, ROS1, RET and NTRK1/2/3 fusions, as well as MET exon 14 skipping splicing alterations) found uniquely by RNA-NGS.
Results
In the evaluable cohort of 5570 patients, median (IQR) age was 67.8 (61.3-75.4) years, and 2989 patients (53.7%) were female. The prevalence of actionable structural variants detected by either RNA-NGS or DNA-NGS was 8.8% (n = 491), with 86.7% (n = 426) of these detected by DNA-NGS. Concurrent RNA-NGS and DNA-NGS identified 15.3% more patients harboring aSVs compared with DNA-NGS alone (491 vs 426 patients, respectively), including 14.3% more patients harboring actionable fusions (376 vs 329 patients) and 18.6% more patients harboring MET exon 14 skipping alterations (115 vs 97 patients). There was no significant association between the assay used for aSV detection and aSV-targeted therapeutic adoption or clinical outcome. Emerging structural variants (eSVs) were found to have a combined prevalence to be 0.7%, with only 47.5% of eSVs detected by DNA-NGS.
Conclusions and Relevance
In this cohort study, the detection of structural variants via concurrent RNA-NGS and DNA-NGS was higher across multiple NCCN-guideline recommended biomarkers compared with DNA-NGS alone, suggesting that RNA-NGS should be routinely implemented in the care of patients with advanced NSCLC.
Introduction
In advanced non–small cell lung cancer (NSCLC), targeted therapies for patients harboring select structural variants (SVs) have revolutionized the clinical landscape by improving survival compared with standard chemotherapy.1 Therefore, genomic testing is recommended for advanced stage NSCLC to identify patients who might benefit from targeted therapies,2,3,4,5 including gene fusions in ALK, RET, ROS1, and NTRK1/2/3, as well as MET exon 14 skipping alterations.6,7,8,9,10,11,12,13,14
Next-generation sequencing (NGS) of DNA is widely used to detect structural variants, including gene fusions and MET exon 14 skipping, in the clinical setting. However, targeted DNA panels must balance assay sensitivity and read coverage. For fusion detection, this frequently entails opting for deeper coverage in exonic regions instead of comprehensive coverage in large—and often repetitive—intronic regions, making it difficult to detect complex fusion rearrangements where breakpoints may be deeply intronic.15,16 Detection of MET exon 14 skipping events also has challenges, as it is difficult to determine the optimal region to identify DNA variants associated with splicing events, particularly for rare variants.17,18 Overall, detecting complex rearrangements via DNA-NGS is technically challenging,19,20,21 resulting in the potential underdetection of clinically actionable variants.22,23
RNA-NGS overcomes several limitations of DNA-NGS for SV detection by enabling the direct observation of gene fusions and splicing junctions, with the potential to improve the detection of actionable SV (aSV) targets.24,25,26 For this reason, clinical guideline bodies have recommended RNA-NGS for SV detection. In particular, the National Comprehensive Cancer Network (NCCN) guidelines describe RNA as the preferred NGS method for detecting fusions and MET exon 14 skipping alterations,2 the European Society for Medical Oncology suggests RNA-NGS is the criterion standard for detecting NTRK fusions,27 and a recent American Society for Clinical Oncology guideline considered RNA-based methods generally superior for fusion detection across multiple targets and recommended RNA-based tests in cases where no oncogenic drivers are detected in DNA or other standard-of-care options.4,28 Finally, as clinical guidelines expand based on emerging clinical evidence for targeted treatment of rare SVs, RNA-NGS can be used to optimize SV detection without assay optimization, as is needed for DNA-NGS assays.
While the clinical benefits of using RNA-NGS to detect aSVs in patients with advanced NSCLC has been previously reported,17,24,25,26 RNA-NGS is not widely used in clinical practice.29 In this study, we compared the detection of NCCN-recommended aSVs via concurrent RNA-NGS and DNA-NGS compared with DNA-NGS alone. We also evaluated the use of whole transcriptome sequencing for the detection of emerging SVs (eSVs) under rapidly evolving clinical guidelines.
Methods
Cohort Selection
The full cohort consisted of patients from the deidentified Tempus multimodal database who received tissue-based DNA next generation sequencing (NGS) and whole-transcriptome RNA-NGS results between February 2021 and October 2023. The Tempus multimodal database contains longitudinal deidentified data from geographically diverse oncology practices, including integrated delivery networks, academic institutions, and community practices linked with deidentified claims data from the Komodo Healthcare Map. Race and ethnicity, which we used to assess diversity of the patient cohort, may be self-declared by patients or reported by clinicians. See eMethods in Supplement 1 for more details.
All patients in this study were diagnosed with stage IIIB-IV metastatic lung adenocarcinoma prior to tissue sample collection. For patients with multiple tests that met the cohort criteria, tests with a fusion detected were used for analysis. If no fusion was detected on any test, the most recent test was used for analysis. This study was conducted on deidentified health information subject to an institutional review board exempt determination (Advarra Pro00072742) and did not involve human participants research. Patient consent was not necessarily obtained, in accordance with this exempt status designation, due to deidentification protocols. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies.
Definitions of aSVs and eSVs
The aSVs included in the primary analyses are those with recommended targeted therapies by NCCN NSCLC Guidelines2: ALK, RET, ROS1, and NTRK1/2/3 fusions and MET exon 14 skipping variants. NCCN-recommended targeted therapies for these variants are listed in eTable 1 in Supplement 1.
The eSVs were determined as follows: (1) EGFR and BRAF fusions were assessed because preliminary data suggests therapies targeting other variants in these genes may also target fusions.30,31 (2) NRG1 and FGFR2/FGFR3 fusions were assessed because these fusions activate ERBB2- and FGFR-mediated pathways respectively, and may be targeted by ERBB2- and FGFR-targeted therapies approved for other cancers.32,33
Tissue-Based Sequencing
Tissue-based DNA-sequencing was performed via the Tempus xT assay (version 4), which assesses variants in 648 genes with 500 × coverage. All sequenced samples had a minimum of 20% tumor purity by pathology review.
All samples had successful tissue-based whole transcriptome RNA sequencing performed via the Tempus xR assay, with all assessed samples passing quality control metrics for both RNA-NGS and DNA-NGS. See eMethods in Supplement 1 for more details.
Fusion Classification by Assay
RNA-NGS and DNA-NGS pipelines are run separately and manually reviewed as part of the Tempus clinical workflow. Fusions were considered detected by DNA if they had an intact kinase domain, the fusion read support was greater than 35, or if the fusion had any read support from DNA-NGS and read support from RNA-NGS for the same partner gene. Fusions were considered detected by RNA if they had an intact kinase domain and the fusion had any read support from RNA-NGS. For fusions with multiple partners, we consider the dominant partner as the partner with the highest read support. See eMethods in Supplement 1 for more details.
Classification of MET Exon 14 Skipping Alterations by Assay
Sample DNA was labeled as positive for MET exon 14 skipping alterations if a detected variant exhibited prespecified alterations. Consistent with the American College of Medical Genetics standards, DNA variants were classified and reported as pathogenic after manual review.34
RNA-based detection of MET exon 14 skipping alterations was performed as part of the Tempus altered splicing pipeline. Positive detection of a MET exon 14 skipping alteration required at least 10 reads of a corresponding isoform. See eMethods in Supplement 1 for more details.
Cohort Selection for Therapy Adoption and Time-to-Next-Treatment Analyses
Patients from the full cohort were included in the therapy adoption cohort if they had 1 or more medication events that occurred more than 90 days post sequencing. Patients who received a targeted therapy prior to NGS testing with the Tempus assays were excluded from analysis. Patients were classified as adherent if they received a targeted therapy consistent with NCCN guidelines following sequencing (eTable 1 in Supplement 1).
Patients from the full cohort were included if they had 1 or more medication initiation events that occurred more than 90 days post–tissue sample collection. The index date was defined as the first documented date of treatment with variant targeted therapy after tissue collection. The end point for the time-to-next-treatment (TTNT) analysis was the start date of a new treatment regimen that did not include the variant targeted therapy or death. This definition excluded regimen updates where new medications were merely added to variant-targeted therapy, whether due to Food and Drug Administration (FDA) approval or insurance coverage. Event-free patients were censored at the earliest of medication end date or last known follow-up. See eMethods in Supplement 1 for more details.
Statistical Analysis
The χ2 test was used to assess differences in categorical variables according to SV status (if a category had less than 5 patients, Fisher exact test was used). The Kruskal-Wallis test was used on continuous variables (eg, age). Fisher exact test was used to compare the association between variant detection modality and patient adherence to targeted therapy. The association between variant detection modality and TTNT was visualized using Kaplan-Meier curves and tested using a Cox proportional hazards model, with a null hypothesis that there was no difference in TTNT based on variant detection modality. Statistical significance was assessed using a Wald test. All statistical tests were performed using P < .05 as the threshold for statistical significance, and all tests were 2-sided (eMethods in Supplement 1).
Results
SV Prevalence
In the full cohort of 5570 deidentified patients with stage IIIb-IV lung adenocarcinoma who underwent successful concurrent RNA-NGS and DNA-NGS testing (eMethods in Supplement 1), median (IQR) age was 67.8 (61.3-75.4) years; 2989 patients (53.7%) were female; and 3925 patients (70.5%) were known current or former smokers (Table).
Table. Clinical and Demographic Characteristics of the Cohort, Stratified According to Structural Variant Status.
| Clinical characteristic | Overall (N = 5570) | aSV-negative (n = 5079) | aSV-positive (n = 491) | P value |
|---|---|---|---|---|
| Tissue site, No. (%) | ||||
| Primary tumor | 3863 (69.4) | 3540 (69.7) | 323 (65.8) | .16 |
| Metastasis | 1640 (29.4) | 1480 (29.1) | 160 (32.6) | |
| Unknown | 67 (1.2) | 59 (1.2) | 8 (1.6) | |
| Cancer stage, No. (%) | ||||
| Stage IIIB | 324 (5.8) | 297 (5.8) | 27 (5.5) | .59 |
| Stage IIIC | 91 (1.6) | 81 (1.6) | 10 (2.0) | |
| Stage IV | 5089 (91.4) | 4638 (91.3) | 451 (91.9) | |
| Other | 42 (0.8) | 41 (0.8) | 1 (0.2) | |
| Unknown | 24 (0.4) | 22 (0.4) | 2 (0.4) | |
| Smoking status, No. (%) | ||||
| Current or former smoker | 3925 (70.5) | 3753 (73.9) | 172 (35.0) | <.001 |
| Nonsmoker | 962 (17.3) | 729 (14.4) | 233 (47.5) | |
| Unknown | 683 (12.3) | 597 (11.8) | 86 (17.5) | |
| Sex, No. (%) | ||||
| Female | 2989 (53.7) | 2714 (53.4) | 275 (56.0) | .30 |
| Male | 2581 (46.3) | 2365 (46.6) | 216 (44.0) | |
| Age at tissue collection, median (IQR), y | 67.8 (61.3-75.4) | 68.0 (61.7-75.5) | 65.1 (54.4-74.3) | <.001 |
| Race, No. (%) | ||||
| American Indian or Alaska Native | 7 (0.1) | 7 (0.1) | 0 | .03 |
| Asian | 201 (3.6) | 172 (3.4) | 29 (5.9) | |
| Black or African American | 462 (8.3) | 430 (8.5) | 32 (6.5) | |
| Native Hawaiian or Other Pacific Islander | 3 (0.1) | 3 (0.1) | 0 | |
| White | 2836 (50.9) | 2606 (51.3) | 230 (46.8) | |
| Unknown | 1847 (33.2) | 1671 (32.9) | 176 (35.8) | |
| Other racea | 214 (3.8) | 190 (3.7) | 24 (4.9) | |
| Ethnicity, No. (%) | ||||
| Hispanic or Latino | 171 (3.1) | 151 (3.0) | 20 (4.1) | .25 |
| Not Hispanic or Latino | 2434 (43.7) | 2232 (43.9) | 202 (41.1) | |
| Unknown | 2965 (53.2) | 2696 (53.1) | 269 (54.8) |
Abbreviation: aSV, actionable structural variant.
Note that the other race category includes any individual with an abstracted race that does not explicitly match one of the listed categories.
Overall, aSVs with matched NCCN-recommended targeted therapies were identified in 8.8% of patients (491 of 5570). Prevalences for each aSV were 4.5% for ALK, 1.1% for RET and ROS1, 0.1% for NTRK1/2/3, and 2.1% for MET exon 14 skipping (Figure 1A). Patients harboring an aSV were more likely to be nonsmokers (aSV-positive nonsmokers: 233 [47.5%] vs aSV-positive smokers: 172 [35.0%] vs aSV-positive unknown smoking status: 86 [17.5%]; P < .001) and younger (median [IQR] age of those who were aSV-positive: 65.1 [54.4.-74.3] years vs aSV-negative: 68.0 [61.7-75.5] years; P < .001) (Table), whereas patients with MET exon 14 skipping variants were older (median [IQR] age of those with MET exon 14 skipping variants: 75.3 [66.8-83.6] years vs those without: 67.7 [61.3-75.2] years; P < .001) (eTable 2 in Supplement 1).
Figure 1. Assessment of Actionable Structural Variant (aSV) Detection via Combined RNA Next-Generation Sequencing (NGS) and DNA-NGS.
A, The prevalence of 5 NCCN guideline–recommended aSV biomarkers, with additional categories for all fusions and all aSVs. Whiskers indicated 95% CIs. B, For each aSV and combined categories, stacked bars indicate the percentage of variants detected via RNA-NGS, DNA-NGS, or both assays. C, The percentage increase in detection from concurrent RNA-NGS and DNA-NGS compared with only DNA-NGS.
For patients with an aSV, RNA-NGS uniquely identified aSVs in 13.2% of patients (65 of 491). The percentage of individual aSVs identified solely by RNA-NGS and not DNA-NGS varied slightly by aSV, with the largest percentage identified in ROS1 (25.4%) (Figure 1B). Overall, the addition of RNA-NGS testing identified 15.3% more patients harboring aSVs relative to DNA-NGS alone (491 vs 426 patients), including 14.3% more patients harboring actionable fusions (376 vs 329 patients) and 18.6% more patients harboring MET exon 14 skipping alterations (115 vs 97 patients) (Figure 1C). Count data underlying these findings is provided in eTable 3 in Supplement 1.
Fusion Partners and Fusion Isoforms Detected by RNA-NGS and DNA-NGS
The most common fusion partners detected for each fusion were ALK-EML4, RET-KIF5B, and ROS1-CD74 (eFigure 1 in Supplement 1). These canonical partners were more likely detected by RNA-NGS than by DNA-NGS, with the latter detecting less common fusion partners (eFigure 1 in Supplement 1). For example, 93.2% of patients with ALK fusions identified by RNA-NGS (233 of 250) had EML4 as the dominant partner, whereas 82.1% of patients with ALK fusions identified by DNA-NGS (184 of 224) had EML4 as the dominant partner. In contrast, 9.8% of patients with ALK fusions identified by DNA-NGS (22 of 224) had fusion partners that were only detected in 1 patient, whereas in patients with ALK fusions identified by RNA-NGS, this number was only 3.2% (8 of 250). Similar trends were observed for other fusions (eFigure 1 in Supplement 1).
We next quantified the concordance of fusion partners by assay in the subset of patients that had a fusion detected by both RNA-NGS and DNA-NGS (n = 327). In this subset, fusions were classified as detected by DNA-NGS either if they had at least 35 supporting reads (78.6% [257 of 327]) or had fewer than 35 supporting reads but had the same fusion partner that was detected by RNA-NGS (21.4% [70 of 327]). Because DNA-NGS and RNA-NGS may detect multiple fusion partners of a driver fusion at varying levels of read support, we assessed concordance of fusion partners in 2 ways. First, we assessed whether the fusion detected by DNA-NGS with the highest read support had the same partners as that detected by RNA-NGS. The majority of patients had the same fusion partner detected by both assays, with the highest concordance observed in NTRK (100% [4 of 4]) and ROS1 (95.5% [42 of 44]) and the lowest observed in ALK (87.4% [195 of 223]) and RET (80.4% [45 of 56]) (Figure 2A). Next, we assessed concordance of fusion partners allowing for any fusion detected by DNA-NGS, regardless of how much read support there was for that fusion. With this approach, concordance among ALK fusions increased from 87.4% to 93.7%, and increased for RET from 80.4% to 91.1% (Figure 2B).
Figure 2. Fusion Partner Concordance When Fusions Are Detected by Both RNA Next-Generation Sequencing (NGS) and DNA-NGS.
A, For each fusion gene, the percentage of patients for which the highest evidence gene fusion partner detected by DNA-NGS (quantified by read support) was the same (matched) as the highest evidence gene fusion partner detected by RNA-NGS. B, For each fusion gene, the percentage of patients for which there was any evidence (quantified by at least 1 read) in DNA-NGS for the highest evidence gene fusion partner detected by RNA-NGS (eg, the fusion with the highest read support).
For the most common fusion partners for each fusion gene, we additionally assessed the distribution of individual isoforms according to detection modality. For the ALK-EML4 fusion, we observed differences in the distribution of fusion isoforms detected only by RNA-NGS vs those detected by both NGS assays, with a higher frequency of EML4-Exon6+ALK-Exon20 detected by RNA-NGS only (eFigure 2 in Supplement 1). However, there were few distinct isoforms detected only by RNA-NGS as hypothesized. Similar trends were seen for RET-KIF5B as well as ROS1-CD74 and ROS1-EZR isoforms (eFigure 2 in Supplement 1).
Targeted Therapy Adoption for SVs Detected via RNA-NGS and DNA-NGS
Of the 491 patients with an aSV detected, 104 were naive to targeted therapy at the time of Tempus NGS sequencing and had longitudinal medication information (see eMethods in Supplement 1). Of these 104 patients, 84 (80.8%) received the matched NCCN-recommended targeted therapy for the identified aSV following sequencing; adoption rates varied by SV (Figure 3A). Patients who did not receive matched therapy (n = 20; 3 with ALK fusions, 3 with RET fusions, 12 with MET exon 14 skipping alterations, and 2 with ROS1 fusions) received combination chemotherapy and immunotherapy. Clinical and demographic characteristics were similar across patients who did and did not receive the targeted therapy (eTable 4 in Supplement 1).
Figure 3. Therapy Adoption and Time-To-Next-Treatment Stratified by Detection Modality for a Subset of Patients With Enriched Clinical Data.
A, Targeted therapy adoption rates are shown according to whether the actionable structural variant was detected by either DNA next-generation sequencing (NGS) or RNA-NGS only for each of the indicated variants. (Note that 1 NTRK patient detected in DNA +/− RNA is not shown but is included in the All category.) Whiskers indicated 95% CIs. B, Lines depict time-to-next-treatment according to whether actionable structural variants were detected by both RNA-NGS and DNA-NGS or RNA-NGS only (hazard ratio = 1.3 [95% CI, 0.6-3.2]; P = .50 [Wald test]). Dashed lines indicate median time-to-next-treatment for the 2 categories: (1) RNA only and (2) RNA and DNA. Results further stratified for select variants can be found in eFigure 3 in Supplement 1.
We evaluated whether there was any association between assays used to detect aSVs and targeted therapy adoption. Although rates of therapy adoption were higher for patients with variants detected by DNA-NGS (83.1% [74 of 89], irrespective of RNA-NGS detection) than for patients with variants detected only by RNA-NGS (66.6% [10 of 15]), these rates were not significantly different in aggregate or by specific aSV (Figure 3A).
TTNT for SVs Detected via RNA-NGS and DNA-NGS
We next assessed whether there were differences in TTNT for patients with an aSV detected by both RNA-NGS and DNA-NGS vs RNA-NGS only that were treated with targeted therapy. We identified 124 patients who received an NCCN guideline–recommended targeted therapy after tissue collection and for whom we could assess TTNT (eMethods in Supplement 1). Of these patients, 14 had variants detected only via RNA-NGS, while the remaining 110 had variants detected by both RNA-NGS and DNA-NGS. Clinical and demographic characteristics were similar between the 2 groups (eTable 5 in Supplement 1). No significant difference in TTNT was observed between the groups (hazard ratio, 1.3 [95% CI, 0.5-3.3]; Figure 3B); median TTNT was 19.7 months in patients with variants detected by both RNA-NGS and DNA-NGS compared with 14.9 months in patients with variants detected by RNA-NGS alone. The TTNT findings—stratified by variant—showed similar results as the aggregate analysis (eFigure 3 in Supplement 1).
Evaluation of eSV Biomarkers in NSCLC
We next examined the eSVs of BRAF, NRG1, EGFR, and FGFR2/3. The demographics of the patients with eSVs detected were similar with that of patients without eSVs (eTable 6 in Supplement 1). As for patients with aSVs, patients with eSVs were less likely to be smokers (P < .001). The overall prevalence of these fusions in our cohort was 0.7% (40 of 5570): 0.3% (18 of 5570) BRAF, 0.2% (10 of 5570) NRG1, 0.1% (5 of 5570) EGFR, and 0.1% (7 of 5570) FGFR2/3 (Figure 4A). With the exception of FGFR3 (for which all fusion partners were TACC3), there were no dominant fusion partners identified for any of the driver fusions (eTable 7 in Supplement 1). Fifty-three percent (n = 21) of the eSVs were detected solely by RNA-NGS—including 66.7% (12 of 18) of BRAF, 30.0% (3 of 10) of NRG1, 40.0% (2 of 5) of EGFR, and 57.1% (4 of 7) of FGFR2/3 (Figure 4B), corresponding to a more than doubling (110.5% increase) in the eSV detection rate when incorporating RNA-NGS testing compared with DNA-NGS alone (Figure 4C). Count data underlying these figures is provided in eTable 8 in Supplement 1.
Figure 4. Assessment of Emerging Fusion Variants via Combined RNA Next-Generation Sequencing (NGS) and DNA-NGS.
A, The prevalence of 4 emerging fusion biomarkers. Whiskers indicate 95% CIs. B, For each fusion variant (and combined), stacked bars indicate the percentage of variants detected by RNA-NGS, DNA-NGS, or both assays. C, The percentage increase in detection rate from concurrent RNA-NGS and DNA-NGS compared with only DNA-NGS.
Of the 40 patients with eSVs, 32.5% (n = 13) also harbored an NCCN NSCLC guideline-recommended, actionable covariant (eMethods and eTable 9 in Supplement 1). In comparison, the covariant prevalence for patients with an NCCN-defined aSV was significantly lower at only 2.2% (11 of 491) (P < .001). The covariant prevalence for patients with eSVs varied by fusion type: 50.0% (9 of 18) of BRAF, 10.0% (1 of 10) of NRG1, 0% of EGFR, and 43% (3 of 7) of FGFR2/3 (eTable 9 in Supplement 1). Of the 9 patients with a BRAF fusion and actionable covariant, all covariants were EGFR exon 19 deletions and 2 of these patients had additional EGFR p.T790M variants. Among 8 patients with cooccurring BRAF fusions and actionable EGFR variants for whom treatment information was available, 7 were treated with EGFR-targeted therapy prior to sequencing, suggesting the clonal emergence of BRAF fusions as a possible resistance mechanism to anti-EGFR therapy.
Discussion
In a comprehensive analysis of actionable structural variants in patients with advanced adenocarcinoma NSCLC in a diverse, clinical setting, we found that concurrent RNA-NGS and DNA-NGS identified 15.3% more patients harboring actionable SVs over DNA-NGS alone. The benefits of concurrent RNA-NGS and DNA-NGS were consistently found across individual SVs, with ROS1 exhibiting the highest increase at 34.1% among NSCLC guideline-recommended biomarkers compared with DNA-NGS alone. ALK fusions represented the largest number of detected aSVs and had 12.1% more detected due to the inclusion of RNA-NGS compared with DNA-NGS alone. We found that 78.0% of patients treated in the first-line setting received an NCCN-recommended targeted therapy following receipt of testing results. An analysis of TTNT, a surrogate end point for multisite progression-free survival, showed that disease progression following aSV-targeted therapy did not significantly differ according to the assay used for aSV detection, confirming the clinical benefit of RNA-NGS testing alone as previously demonstrated in clinical trials using DNA-NGS.35
While DNA-NGS assays need to be optimized to detect complex SVs and must consider trade-offs between breadth and depth of coverage for large intronic regions, RNA-NGS is a more direct representation of chimeric proteins. Consistent with other studies, the large increase in SV detection by RNA-NGS in ROS1 is likely due to the large intronic region with repetitive elements,24,36 a finding that is likely to generalize to other genes with complex intronic regions. Furthermore, although most patients with fusions detected by both RNA-NGS and DNA-NGS had the same fusion partner, canonical fusion partners were more likely to be detected by RNA-NGS, suggesting that there may have been complex rearrangements in DNA that make the exact fusion partner difficult to detect prior to splicing.
Clinical guidelines recommend broad panel testing to detect actionable genomic alterations.2 Specifically, guidelines suggest RNA-NGS following DNA-NGS if no actionable results are detected from DNA.2 However, as our study shows, 13.2% of patients who had an aSV would have experienced a meaningful delay of several weeks or more in the detection of a matched targeted therapy if RNA-NGS and DNA-NGS had been performed sequentially rather than concurrently. Furthermore, the clinical benefits of targeted therapy adoption via NGS results (whether DNA or RNA) have not yet been fully realized, as our targeted therapy adoption analysis demonstrated that about 1 in 5 patients did not receive NCCN-recommended first-line targeted therapies, highlighting the ongoing need for education about implementing NGS results in clinical practice. For aSVs detected by RNA-NGS, this barrier may be even higher: although not statistically significant difference, we observed 79.5% of patients receiving targeted therapy due to DNA-NGS aSV detection compared with 70.6% due to RNA-NGS detection alone.
The benefits of RNA-NGS testing in NSCLC are particularly important because the FDA-approved landscape of SV-targeted therapies in patients with advanced NSCLC continues to expand. We therefore examined the prevalence of emerging eSVs not included in current guidelines, but for which promising clinical evidence is accumulating. Although the overall prevalence of BRAF, EGFR, and NRG1 fusions in our dataset was low (<1%), concurrent RNA-NGS and DNA-NGS detected 100% more eSVs relative to DNA-NGS alone, expanding the number of potential patients that may benefit from targeted therapies.
The co-occurrence of rare BRAF fusions and actionable EGFR variants in patients pretreated with standard of care anti-EGFR therapy highlights a potential resistance mechanism for the emergence of BRAF fusions, consistent with other published studies.37 Consequently, these findings support a possible evaluation of upfront dual variation targeted treatment to mitigate this resistance mechanism. Further prospective clinical studies are needed to validate these findings, and incorporation of RNA-NGS as a companion diagnostic test in future registrational studies should be considered.
Limitations
This study has limitations. As SV detection accuracy may vary among commercial platforms, a limitation of our study is the use of a single commercial NGS platform. However, as aSV prevalences observed in this study by DNA-NGS were comparable with published studies using alternative assays,17,29,31,36,38 the primary finding of increased SV detection by RNA-NGS is likely to generalize. Another limitation is incomplete longitudinal data for therapy adoption and TTNT analyses due to the retrospective nature of the study, resulting in small cohort sizes, which may have limited the ability to detect statistical differences between RNA-NGS and DNA-NGS. However, there is no published evidence to suggest that differences in DNA or RNA fusion detection may contribute to differences in outcomes.38 Additionally, we note that in this study we do not account for differences in assay failure rates between RNA-NGS and DNA-NGS. Differences in failure rates may be due, in part, to nucleic acid extraction protocols, which often prioritize DNA extraction over RNA, and the decreased stability of RNA compared with DNA.39 A prior study found that out of 275 lung adenocarcinomas that were fusion-negative based on DNA-NGS, 7.6% (n = 21) of cases were not sequenced due to low RNA quality.24 Though differences in failure rate are important to consider clinically, we focused on a cohort that received successful RNA-NGS and DNA-NGS in order to directly compare SV detection rates.
Conclusions
To our knowledge, this is the largest retrospective study that expands clinical evidence supporting NCCN guideline recommendations for the use of concurrent RNA-NGS and DNA-NGS in NSCLC. Our findings suggest that RNA-NGS has a robust clinical sensitivity, enhances SV detection when used in conjunction with DNA-NGS for both existing and emerging actionable structural variants, and should be routinely offered with DNA-NGS in the clinic to maximize SV detection in patients with advanced adenocarcinoma NSCLC.
eMethods. Supplementary Methods
eFigure 1. Fusion Partner Identity Amongst Fusions Detected by DNA-NGS and RNA-NGS
eFigure 2. Fusion Isoform Proportions Detected by RNA-NGS for Most Common Fusion Partners, Stratified by Whether the Patient Had the Fusion Detected Only by RNA-NGS or by Both RNA-NGS and DNA-NGS
eFigure 3. Real-World Time to Next Treatment (rwTTNT) Stratified According to Structural Variant Detection Modality for Individual Fusions
eTable 1. List of Targeted Therapies for the Primary Structural Variants Assessed
eTable 2. Clinical and Demographic Characteristics of the Cohort Stratified by Fusion and MET Exon 14 Skipping Status
eTable 3. Count Data Underlying Findings Displayed in Figure 1
eTable 4. Clinical and demographic characteristics of the sub-cohort selected for therapy adoption analysis.
eTable 5. Clinical and Demographic Characteristics of the Sub-Cohort Selected for rwTTNT Analysis
eTable 6. Clinical and Demographic Characteristics of the Cohort Stratified by Emerging Fusion Status
eTable 7. List of Fusion Partner Frequency for the Individual Emerging Fusions Stratified According to Detection Modality
eTable 8. Count Data Underlying Findings Displayed in Figure 4
eTable 9. Prevalence of NCCN Guideline Recommended Co-Variants for Actionable and Emerging Structural Variants
Data Sharing Statement
References
- 1.Thai AA, Solomon BJ, Sequist LV, Gainor JF, Heist RS. Lung cancer. Lancet. 2021;398(10299):535-554. doi: 10.1016/S0140-6736(21)00312-3 [DOI] [PubMed] [Google Scholar]
- 2.National Comprehensive Cancer Network . Non-small cell lung cancer (version 2.2023). Accessed March 22, 2023. https://www.nccn.org/professionals/physician_gls/pdf/nscl.pdf
- 3.ESMO . ESMO Clinical Practice Guidelines: Lung and Chest Tumours. Accessed July 27, 2023. https://www.esmo.org/guidelines/guidelines-by-topic/esmo-clinical-practice-guidelines-lung-and-chest-tumours
- 4.Owen DH, Singh N, Ismaila N, et al. Therapy for stage IV non-small-cell lung cancer with driver alterations: ASCO living guideline, version 2023.2. J Clin Oncol. Published online July 11, 2023. doi: 10.1200/JCO.23.01055 [DOI] [PubMed]
- 5.Lindeman NI, Cagle PT, Aisner DL, et al. Updated molecular testing guideline for the selection of lung cancer patients for treatment with targeted tyrosine kinase inhibitors: guideline from the College of American Pathologists, the International Association for the Study of Lung Cancer, and the Association for Molecular Pathology. J Thorac Oncol. 2018;13(3):323-358. doi: 10.1016/j.jtho.2017.12.001 [DOI] [PubMed] [Google Scholar]
- 6.Camidge DR, Kim HR, Ahn MJ, et al. Brigatinib versus crizotinib in ALK-positive non-small-cell lung cancer. N Engl J Med. 2018;379(21):2027-2039. doi: 10.1056/NEJMoa1810171 [DOI] [PubMed] [Google Scholar]
- 7.Shaw AT, Bauer TM, de Marinis F, et al. ; CROWN Trial Investigators . First-line lorlatinib or crizotinib in advanced ALK-positive lung cancer. N Engl J Med. 2020;383(21):2018-2029. doi: 10.1056/NEJMoa2027187 [DOI] [PubMed] [Google Scholar]
- 8.Wolf J, Seto T, Han JY, et al. ; GEOMETRY mono-1 Investigators . Capmatinib in MET exon 14-mutated or MET-amplified non-small-cell lung cancer. N Engl J Med. 2020;383(10):944-957. doi: 10.1056/NEJMoa2002787 [DOI] [PubMed] [Google Scholar]
- 9.Paik PK, Felip E, Veillon R, et al. Tepotinib in non-small-cell lung cancer with MET exon 14 skipping mutations. N Engl J Med. 2020;383(10):931-943. doi: 10.1056/NEJMoa2004407 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Drilon A, Laetsch TW, Kummar S, et al. Efficacy of larotrectinib in TRK fusion-positive cancers in adults and children. N Engl J Med. 2018;378(8):731-739. doi: 10.1056/NEJMoa1714448 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Doebele RC, Drilon A, Paz-Ares L, et al. ; trial investigators . Entrectinib in patients with advanced or metastatic NTRK fusion-positive solid tumours: integrated analysis of three phase 1-2 trials. Lancet Oncol. 2020;21(2):271-282. doi: 10.1016/S1470-2045(19)30691-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Subbiah V, Wolf J, Konda B, et al. Tumour-agnostic efficacy and safety of selpercatinib in patients with RET fusion-positive solid tumours other than lung or thyroid tumours (LIBRETTO-001): a phase 1/2, open-label, basket trial. Lancet Oncol. 2022;23(10):1261-1273. doi: 10.1016/S1470-2045(22)00541-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Gainor JF, Curigliano G, Kim DW, et al. Pralsetinib for RET fusion-positive non-small-cell lung cancer (ARROW): a multi-cohort, open-label, phase 1/2 study. Lancet Oncol. 2021;22(7):959-969. doi: 10.1016/S1470-2045(21)00247-3 [DOI] [PubMed] [Google Scholar]
- 14.Shaw AT, Riely GJ, Bang YJ, et al. Crizotinib in ROS1-rearranged advanced non-small-cell lung cancer (NSCLC): updated results, including overall survival, from PROFILE 1001. Ann Oncol. 2019;30(7):1121-1126. doi: 10.1093/annonc/mdz131 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Jennings LJ, Arcila ME, Corless C, et al. Guidelines for validation of next-generation sequencing-based oncology panels: a joint consensus recommendation of the Association for Molecular Pathology and College of American Pathologists. J Mol Diagn. 2017;19(3):341-365. doi: 10.1016/j.jmoldx.2017.01.011 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Wong D, Yip S, Sorensen PH. Methods for identifying patients with tropomyosin receptor kinase (TRK) fusion cancer. Pathol Oncol Res. 2020;26(3):1385-1399. doi: 10.1007/s12253-019-00685-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Davies KD, Lomboy A, Lawrence CA, et al. DNA-based versus RNA-based detection of MET Exon 14 skipping events in lung cancer. J Thorac Oncol. 2019;14(4):737-741. doi: 10.1016/j.jtho.2018.12.020 [DOI] [PubMed] [Google Scholar]
- 18.Socinski MA, Pennell NA, Davies KD. MET Exon 14 skipping mutations in non-small-cell lung cancer: an overview of biology, clinical outcomes, and testing considerations. JCO Precis Oncol. 2021;5:PO.20.00516. doi: 10.1200/PO.20.00516 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Koboldt DC. Best practices for variant calling in clinical sequencing. Genome Med. 2020;12(1):91. doi: 10.1186/s13073-020-00791-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.van Belzen IAEM, Schönhuth A, Kemmeren P, Hehir-Kwa JY. Structural variant detection in cancer genomes: computational challenges and perspectives for precision oncology. NPJ Precis Oncol. 2021;5(1):15. doi: 10.1038/s41698-021-00155-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Liu Z, Roberts R, Mercer TR, Xu J, Sedlazeck FJ, Tong W. Towards accurate and reliable resolution of structural variants for clinical diagnosis. Genome Biol. 2022;23(1):68. doi: 10.1186/s13059-022-02636-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Zhang NN, Liu YT, Ma L, et al. The molecular detection and clinical significance of ALK rearrangement in selected advanced non-small cell lung cancer: ALK expression provides insights into ALK targeted therapy. PLoS One. 2014;9(1):e84501. doi: 10.1371/journal.pone.0084501 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Li W, Wan R, Guo L, et al. Reliability analysis of exonic-breakpoint fusions identified by DNA sequencing for predicting the efficacy of targeted therapy in non-small cell lung cancer. BMC Med. 2022;20(1):160. doi: 10.1186/s12916-022-02362-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Benayed R, Offin M, Mullaney K, et al. High yield of RNA sequencing for targetable kinase fusions in lung adenocarcinomas with no mitogenic driver alteration detected by DNA sequencing and low tumor mutation burden. Clin Cancer Res. 2019;25(15):4712-4722. doi: 10.1158/1078-0432.CCR-19-0225 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Cohen D, Hondelink LM, Solleveld-Westerink N, et al. Optimizing mutation and fusion detection in NSCLC by sequential DNA and RNA sequencing. J Thorac Oncol. 2020;15(6):1000-1014. doi: 10.1016/j.jtho.2020.01.019 [DOI] [PubMed] [Google Scholar]
- 26.Yang SR, Aypar U, Rosen EY, et al. A performance comparison of commonly used assays to detect RET fusions. Clin Cancer Res. 2021;27(5):1316-1328. doi: 10.1158/1078-0432.CCR-20-3208 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Oncology PRO . Algorithm for NTRK gene fusion testing. Accessed July 27, 2023. https://oncologypro.esmo.org/oncology-in-practice/anti-cancer-agents-and-biological-therapy/targeting-ntrk-gene-fusions/algorithm-for-ntrk-gene-fusion-testing
- 28.Chakravarty D, Johnson A, Sklar J, et al. Somatic genomic testing in patients with metastatic or advanced cancer: ASCO provisional clinical opinion. J Clin Oncol. 2022;40(11):1231-1258. doi: 10.1200/JCO.21.02767 [DOI] [PubMed] [Google Scholar]
- 29.Lin HM, Wu Y, Yin Y, et al. Real-world ALK testing trends in patients with advanced non-small-cell lung cancer in the United States. Clin Lung Cancer. 2023;24(1):e39-e49. doi: 10.1016/j.cllc.2022.09.010 [DOI] [PubMed] [Google Scholar]
- 30.Nokin MJ, Ambrogio C, Nadal E, Santamaria D. Targeting infrequent driver alterations in non-small cell lung cancer. Trends Cancer. 2021;7(5):410-429. doi: 10.1016/j.trecan.2020.11.005 [DOI] [PubMed] [Google Scholar]
- 31.Konduri K, Gallant JN, Chae YK, et al. EGFR fusions as novel therapeutic targets in lung cancer. Cancer Discov. 2016;6(6):601-611. doi: 10.1158/2159-8290.CD-16-0075 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Patil T, Carrizosa DR, Burkard ME, et al. Abstract CT229: CRESTONE: a phase 2 study of seribantumab in adult patients with neuregulin-1 (NRG1) fusion positive locally advanced or metastatic solid tumors. Cancer Res. 2023;83(8)(suppl):CT229. doi: 10.1158/1538-7445.AM2023-CT229 [DOI] [Google Scholar]
- 33.Ellis H, Goyal L. Are FGFR fusions and mutations the next tumor-agnostic targets in oncology? JCO Precis Oncol. 2024;8:e2400113. doi: 10.1200/PO.24.00113 [DOI] [PubMed] [Google Scholar]
- 34.Li MM, Datto M, Duncavage EJ, et al. Standards and guidelines for the interpretation and reporting of sequence variants in cancer: a joint consensus recommendation of the Association for Molecular Pathology, American Society of Clinical Oncology, and College of American Pathologists. J Mol Diagn. 2017;19(1):4-23. doi: 10.1016/j.jmoldx.2016.10.002 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.FDA . PMA P170019: FDA summary of safety and effectiveness data (FoundationOne CDxTM). Accessed February 16, 2024. https://www.accessdata.fda.gov/cdrh_docs/pdf17/P170019B.pdf
- 36.Davies KD, Le AT, Sheren J, et al. Comparison of molecular testing modalities for detection of ROS1 rearrangements in a cohort of positive patient samples. J Thorac Oncol. 2018;13(10):1474-1482. doi: 10.1016/j.jtho.2018.05.041 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Rosell R, González-Cao M, Codony-Servat J, Molina-Vila MA, de Las Casas CM, Ito M. Acquired BRAF gene fusions in osimertinib resistant EGFR-mutant non-small cell lung cancer. Transl Cancer Res. 2023;12(3):456-460. doi: 10.21037/tcr-22-2888 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Xia P, Zhang L, Li P, et al. Molecular characteristics and clinical outcomes of complex ALK rearrangements identified by next-generation sequencing in non-small cell lung cancers. J Transl Med. 2021;19(1):308. doi: 10.1186/s12967-021-02982-4 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Bruno R, Fontanini G. Next generation sequencing for gene fusion analysis in lung cancer: a literature review. Diagnostics (Basel). 2020;10(8):521. doi: 10.3390/diagnostics10080521 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eMethods. Supplementary Methods
eFigure 1. Fusion Partner Identity Amongst Fusions Detected by DNA-NGS and RNA-NGS
eFigure 2. Fusion Isoform Proportions Detected by RNA-NGS for Most Common Fusion Partners, Stratified by Whether the Patient Had the Fusion Detected Only by RNA-NGS or by Both RNA-NGS and DNA-NGS
eFigure 3. Real-World Time to Next Treatment (rwTTNT) Stratified According to Structural Variant Detection Modality for Individual Fusions
eTable 1. List of Targeted Therapies for the Primary Structural Variants Assessed
eTable 2. Clinical and Demographic Characteristics of the Cohort Stratified by Fusion and MET Exon 14 Skipping Status
eTable 3. Count Data Underlying Findings Displayed in Figure 1
eTable 4. Clinical and demographic characteristics of the sub-cohort selected for therapy adoption analysis.
eTable 5. Clinical and Demographic Characteristics of the Sub-Cohort Selected for rwTTNT Analysis
eTable 6. Clinical and Demographic Characteristics of the Cohort Stratified by Emerging Fusion Status
eTable 7. List of Fusion Partner Frequency for the Individual Emerging Fusions Stratified According to Detection Modality
eTable 8. Count Data Underlying Findings Displayed in Figure 4
eTable 9. Prevalence of NCCN Guideline Recommended Co-Variants for Actionable and Emerging Structural Variants
Data Sharing Statement
