If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Pathology and Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkMarie-Josee and Henry R. Kravis Center for Molecular Oncology, Memorial Sloan Kettering Cancer Center, New York, New York
Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Pathology and Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkMarie-Josee and Henry R. Kravis Center for Molecular Oncology, Memorial Sloan Kettering Cancer Center, New York, New York
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Human Oncology and Pathogenesis Program, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Pathology and Laboratory Medicine, Memorial Sloan Kettering Cancer Center, New York, New York
Corresponding author. Address for correspondence: Helena A. Yu, MD, Thoracic Oncology Service, Department of Medicine, Memorial Sloan Kettering Cancer Center, 545 E 73rd Street, New York, NY 10021.
Department of Medicine, Memorial Sloan Kettering Cancer Center, New York, New YorkDepartment of Medicine, Weill Cornell Medical College, New York, New York
Preferred first-line treatment for patients with metastatic EGFR-mutant lung cancer is osimertinib, yet it is not known whether patient outcomes may be improved by identifying and intervening on molecular markers associated with therapeutic resistance.
Methods
All patients with metastatic EGFR-mutant lung cancer treated with first-line osimertinib at the Memorial Sloan Kettering Cancer Center (n = 327) were identified. Available pretreatment and postprogression tumor samples underwent targeted gene panel sequencing and mutational signature analysis using SigMA algorithm. Progression-free survival (PFS) and overall survival were estimated using the Kaplan-Meier method.
Results
Using multivariate analysis, baseline atypical EGFR (median PFS = 5.8 mo, p < 0.001) and concurrent TP53/RB1 alterations (median PFS = 10.5 mo, p = 0.015) were associated with shorter PFS on first-line osimertinib. Of 95 patients with postprogression biopsies, acquired resistance mechanisms were identified in 52% (off-target, n = 24; histologic transformation, n = 14; on-target, n = 12), with MET amplification (n = 9), small cell lung transformation (n = 7), and acquired EGFR amplification (n = 7), the most frequently identified mechanisms. Although there was no difference in postprogression survival on the basis of identified resistance (p = 0.07), patients with subsequent second-line therapy tailored to postprogression biopsy results had improved postprogression survival (hazard ratio = 0.09, p = 0.006). The paired postprogression tumors had higher tumor mutational burden (p = 0.008) and further dominant APOBEC mutational signatures (p = 0.07) compared with the pretreatment samples.
Conclusions
Patients with EGFR-mutant lung cancer treated with first-line osimertinib have improved survival with treatment adaptation on the basis of identified mechanisms of resistance at time of progression using tissue-based genomic analysis. Further survival gains may be achieved using risk-based treatment adaptation of pretreatment genomic alterations.
NSCLC with activating EGFR alterations is the single largest lung cancer genotype treated with molecularly targeted therapies and often considered a paradigm for oncogene-addicted cancers. The FLAURA trial established osimertinib, a highly central nervous system–penetrant, third-generation EGFR tyrosine kinase inhibitor (TKI) as the preferred first-line treatment for metastatic EGFR-mutant lung cancer, with improvement in overall survival (OS)
Mediators of resistance to osimertinib differ between first- versus later-line treatment setting, with on-target acquired EGFR alterations more common when osimertinib is used after other EGFR TKIs.
Landscape of acquired resistance to osimertinib in EGFR-mutant NSCLC and clinical validation of combined EGFR and RET inhibition with osimertinib and BLU-667 for acquired RET fusion.
Tumor analyses reveal squamous transformation and off-target alterations as early resistance mechanisms to first-line osimertinib in EGFR-mutant lung cancer.
Mechanisms of resistance to first-line osimertinib reported in small, primarily plasma-based series are similar to later-line osimertinib, including emergence of EGFR C797S,
Acquired resistance of EGFR-mutant lung cancer to a T790M-specific EGFR inhibitor: emergence of a third mutation (C797S) in the EGFR tyrosine kinase domain.
Tumor analyses reveal squamous transformation and off-target alterations as early resistance mechanisms to first-line osimertinib in EGFR-mutant lung cancer.
albeit at differing frequencies. Importantly, although tailoring treatment to an identified mechanism of acquired resistance in small series or individual cases has induced disease responses,
Response to dual crizotinib and osimertinib treatment in a lung cancer patient with MET amplification detected by liquid biopsy who acquired secondary resistance to EGFR tyrosine kinase inhibition.
Tolerable and effective combination of full-dose crizotinib and osimertinib targeting MET amplification sequentially emerging after T790M positivity in EGFR-mutant non-small cell lung cancer.
the impact of this approach on a larger scale has not been systematically studied. Furthermore, studies focusing on clinicogenomic correlates affecting outcomes on first-line osimertinib are limited, with previous studies including patients treated with later-line osimertinib,
Landscape of acquired resistance to osimertinib in EGFR-mutant NSCLC and clinical validation of combined EGFR and RET inhibition with osimertinib and BLU-667 for acquired RET fusion.
Tumor analyses reveal squamous transformation and off-target alterations as early resistance mechanisms to first-line osimertinib in EGFR-mutant lung cancer.
In this work, we identified patients with EGFR-mutant NSCLC treated with first-line osimertinib at the Memorial Sloan Kettering (MSK) Cancer Center. We focused on patients with pre- and post-osimertinib progression tumor tissue who underwent next-generation sequencing (NGS) by the Food and Drug Administration–cleared MSK-Integrated Mutation Profiling of Actionable Targets assay (MSK-IMPACT) platform.
Memorial Sloan Kettering-integrated mutation profiling of actionable cancer targets (MSK-IMPACT): a hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology.
Using the largest reported data set of patients treated with first-line osimertinib with both clinical and genomic annotation, we sought to characterize pretreatment clinicogenomic correlates that are associated with time to progression, mechanisms of acquired resistance, and the impact of identifying and treating mechanisms of resistance on post-osimertinib progression survival.
Materials and Methods
Patients
The study was conducted in accordance with the Declaration of Helsinki and good clinical practice guidelines following approval by MSK institutional review board/privacy board (MSK IRB 21-416). Patients with NSCLC and EGFR alterations who received treatment with first-line osimertinib at MSK were retrospectively identified using an automated query
and Idylla tumor or plasma testing. Inclusion criteria comprised disease confirmation of metastatic lung cancer, identification of an EGFR-sensitizing alteration, and documented first-line osimertinib treatment for metastatic disease. Receipt of concurrent neoplastic agents with first-line osimertinib with or without receipt of prior neoplastic treatments in the adjuvant setting for locally advanced disease were allowed. Patient and tumor characteristics, including baseline EGFR alteration and concurrent genomics, treatment history, imaging, and pathology reports were extracted from the electronic health record.
Genomic Analyses
Samples from the included patients were queried for somatic tumor genomic testing obtained for routine clinical purposes by the Food and Drug Administration–cleared MSK-IMPACT
Memorial Sloan Kettering-integrated mutation profiling of actionable cancer targets (MSK-IMPACT): a hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology.
and/or other institutional or commercial gene panels (Supplementary Fig. 1). Patients with a post-osimertinib progression tumor sample were retrospectively queried for testing with next-generation RNA sequencing (RNAseq) by Archer FusionPlex Custom Solid Panel, a custom RNAseq panel designed to detect gene fusions in 123 cancer-related genes involved in chromosomal rearrangements.
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.
Somatic genomic alterations, including mutations, copy number alterations, and structural variants, were filtered for “oncogenic,” “likely oncogenic,” and “predicted oncogenic” variants on the basis of OncoKB annotation.
Mutational signature analyses, fraction genome altered (FGA), whole genome doubling (WGD), and tumor mutational burden (TMB) were only derived from tumor samples with greater than or equal to 20% purity. TMB was calculated as the total number of nonsynonymous mutations divided by the length of the genomic target region captured with the exome assay, as previously described.
The FGA was computed as the fraction of log2 copy number variation (gain or loss) greater than 0.2 divided by the size of the genome whose copy number was profiled, as previously described.
Patients were considered to have “tailored treatment” if their second-line treatment was directly on the basis of postprogression biopsy results, including treatment for SCLC transformation or targeted therapy directly against an acquired genomic alteration with or without continued osimertinib. Patients who either did not receive further therapy or their second-line treatment was not on the basis of biopsy results were in the “not tailored” group.
Mutational Signature Analysis
For tumor samples that underwent targeted sequencing by MSK-IMPACT, had at least five somatic synonymous and nonsynonymous single-nucleotide variants, and had tumor purity greater than or equal to 20%, mutational signatures were computed using Signature Multivariate Analysis (SigMA),
SigMA was run with customized settings (i.e., data type: msk, cancer type: lung, add_sig3: true, check msi: true). A dominant signature for each sample was determined on the basis of the proposed category assigned by SigMA, as previously reported.
Comparisons of categorical and continuous variables were performed by Fisher’s exact and Mann-Whitney U tests, respectively. Comparisons of frequencies of genes altered by somatic genetic alterations were performed using Fisher’s exact test and logistic regression. Multiple testing correction using the Benjamini-Hochberg method was applied to control for the false discovery rate whenever appropriate. Kaplan-Meier estimates of real-world PFS (rwPFS), time to treatment discontinuation (TTD), and OS were computed. The log-rank test was used to assess for differential effects in EGFR alteration, concurrent genomics, brain metastases, and type of osimertinib treatment received. rwPFS was defined from the start date of osimertinib therapy to the date of progression or death, with dates of progression determined on the basis of the AACR GENIE BPC PRISSMM framework
as the time point at which both imaging reports and medical oncology notes concurred with disease progression. TTD was defined as date of osimertinib initiation to last administered dose, and OS was defined as the date of osimertinib initiation to date of death or last follow-up. The cutoff date for this analysis was March 1, 2022. Hazard estimates, starting from the time of progression, were used to evaluate the postprogression effect of tailored treatment and the mechanism of resistance on OS. The post-baseline effects were tested using a time-dependent covariate Cox model.
Results
Baseline Clinicogenomic Correlates Associated With Time to Progression on Osimertinib
Since 2012, tumor samples from 8221 patients with NSCLC underwent somatic genomic testing at MSK, including 2925 patients harboring an EGFR driver alteration identified by MSK-IMPACT. A total of 327 patients were treated with first-line osimertinib for metastatic disease (Table 1 and Supplementary Fig. 1). Most patients were treated with osimertinib alone (n = 252, 77%). Others received concurrent chemotherapy (including antivascular epithelial growth factor (VEGF) therapy, n = 57), a second EGFR TKI (n = 16), or immune checkpoint inhibitor (ICI) (n = 2, Supplementary Table 1). In addition, 29 patients harbored atypical EGFR alterations, defined as activating EGFR alterations that were neither L858R nor exon 19 deletions (exon 19 del) (Table 2).
Table 1Patient and Tumor Characteristics
Characteristics
Full Cohort N = 327, n (%)
Paired Biopsies Subcohort N = 74, n (%)
Post-Tx Biopsy Subcohort N = 21, n (%)
Age, median (IQR)
64 (57–71)
60 (52–69)
61 (53–66)
Sex
Female
220 (67.2)
50 (67.6)
15 (71.4)
Male
107 (32.7)
24 (32.4)
6 (28.6)
Race
White
197 (60.2)
46 (62.2)
10 (47.6)
Asian
98 (30.0)
21 (28.4)
8 (38.1)
Black
17 (5.2)
3 (4.1)
3 (14.3)
Other
8 (2.4)
2 (2.7)
0 (0)
Unknown
6 (1.8)
2 (2.7)
0 (0)
Smoking history
Never smoker
251 (76.8)
62 (83.8)
17 (81.0)
<15 py history
50 (15.3)
8 (10.8)
3 (14.3)
>15 py history or current
26 (8.0)
4 (5.4)
1 (4.7)
Baseline brain metastases
Present
135 (41.3)
25 (33.8)
7 (33.3)
Absent
183 (56.0)
49 (66.2)
14 (66.7)
Unknown
9 (2.8)
0 (0)
0 (0)
Histology
Adenocarcinoma
309 (94.5)
67 (90.5)
21 (100.0)
Nonadenocarcinoma
18 (5.5)
7 (9.5)
0 (0)
EGFR alteration
Exon 19 deletion
185 (56.6)
46 (62.2)
10 (47.6)
L858R
113 (34.6)
22 (29.7)
6 (28.6)
Atypical
29 (8.9)
6 (8.1)
5 (23.8)
Concurrent pre-tx alterations
TP53
141 (43.1)
40 (54.1)
1 (4.8)
TP53/RB1
50 (15.3)
15 (20.3)
2 (9.5)
Neither
92 (28.1)
19 (25.7)
2 (9.5)
Unknown
44 (13.5)
0 (0)
16 (76.2)
Note: Patient and tumor characteristics are illustrated. Patients with unknown brain metastases status did not have pre-tx brain imaging performed.
With a median follow-up of 23.9 (interquartile range: 13.6–33.9) months, 250 patients experienced disease progression on osimertinib and 77 patients continued first-line osimertinib at the time of data censoring. Median rwPFS was 13.7 months (95% confidence interval [CI]: 12.6–16.8 mo; Fig. 1A). Among all 327 patients, 152 patients had died at data cutoff with a median OS of 35.2 months (95% CI: 32.4–38.5 mo; Fig. 1B). Median TTD was 16.4 months (95% CI: 13.8–18.4 mo; Supplementary Fig. 2A), reflecting the common practice of continuing EGFR TKI beyond initial disease progression.
Continuing EGFR-TKI beyond radiological progression in patients with advanced or recurrent, EGFR mutation-positive non-small-cell lung cancer: an observational study.
Figure 1Efficacy of osimertinib and impact of baseline factor rwPFS on osimertinib. (A) rwPFS and (B) OS for all patients (N = 327) treated with first-line osimertinib, with median times and 95% CIs noted. (C) rwPFS stratified by baseline EGFR alterations (n = 29 atypical, n = 185 exon 19 deletion, and n = 113 L858R). (D) rwPFS stratified by baseline concurrent genomic alterations for patients with available data (n = 92 neither TP53/RB1 alterations, n = 141 TP53 comutations, n = 50 TP53/RB1 comutations). CI, confidence interval; del, deletion; med, median; OS, overall survival; rwPFS, real-world progression-free survival.
We next evaluated pretreatment biomarkers that may influence time to disease progression on osimertinib. rwPFS was shorter for patients with atypical EGFR alterations (median rwPFS = 5.8 mo, 95% CI: 4.9–11.9 mo) compared with patients with exon 19 del (15.2 mo, 95% CI: 12.7–18.8 mo, p = 0.001) or L858R (16.4 mo, 95% CI: 13.1–18.8 mo, p = 0.004; Fig. 1C). No difference in rwPFS between patients with L858R and exon 19 del was observed (p = 0.8). Of 318 with baseline brain imaging, patients with baseline brain metastases had shorter rwPFS compared with patients without brain metastases (median PFS = 11.9 mo; 95% CI: 10.5–15.2 mo versus 16.5 mo, 95% CI: 13.1–18.8 mo, p = 0.04; Supplementary Fig. 2B). There was no difference in rwPFS on the basis of whether osimertinib was used alone or in combination, although patients who received combination therapy had numerically longer rwPFS (median PFS = 13.7 mo osimertinib plus chemotherapy or immune checkpoint inhibitor versus 12.5 mo for osimertinib alone; Supplementary Fig. 2C).
To investigate the impact of baseline somatic alterations on osimertinib efficacy, survival analysis was conducted in patients with available pretreatment NGS (n = 283). The presence of baseline TP53/RB1 and TP53 mutations was associated with shorter rwPFS compared with patients without TP53/RB1 (rwPFS =10.5 mo, 95% CI: 9.0–15.2 mo), at 13.5 mo (95% CI: 11.6–17.9 mo) and 18.8 mo (95% CI: 15.7–24.5 mo), respectively (p = 0.002; Fig. 1D). Concurrent TP53/RB1 alterations were enriched in tumors with EGFR exon19 del (n = 31, 62%) compared with L858R (n = 18, 36%) and atypical (n = 1, 3%). Notably, although concurrent TP3/RB1 alterations increase the risk of histologic transformation,
only 15 of 50 patients (30%) with concurrent TP53/RB1 alterations in our series experienced histologic transformation, indicating that transformation alone does not fully explain shorter rwPFS for these patients. At univariate analysis, atypical EGFR alterations, concurrent TP53/RB1 alterations, and brain metastases were individually associated with rwPFS. Baseline atypical EGFR alterations and concurrent TP53/RB1 alterations remained independently associated with shorter rwPFS in a multivariate model (p < 0.001 and 0.015, respectively; Supplementary Table 2).
Mechanisms of Acquired Resistance to First-Line Osimertinib
Focusing on patients with a postprogression tumor biopsy sample who underwent NGS (n = 95), genomic mechanisms of acquired resistance to first-line osimertinib were identified in 50 of 95 patients (53%) (Fig. 2A). Off-target alterations were found in 24 patients (26%) and represented the most frequently identified mechanism of acquired resistance. Gene amplifications (MET, n = 9; HER2, n = 3; MYC, MDM2/CDK4, and CCND1, n = 1 each), single-gene mutations (PIK3CA, n = 3; KRAS G12A, n = 1), and acquired fusions (RET fusion, n = 2; BRAF fusion, n = 1) were detected. In addition, two patients acquired oncogenic loss-of-function alterations of RB1. On-target EGFR alterations comprised 13% of acquired alterations (EGFR amplification, n = 7; C797S, n = 3; G724S, n = 2). Acquired alterations in post-osimertinib samples that occurred in one or more patients in the cohort are found in Figure 2B. Oncogenic alterations in pre- and post-progression samples with greater than 5% frequency are found in Supplementary Figure 3.
Figure 2Mechanisms of acquired resistance to first-line osimertinib. (A) Identified mechanisms of resistance for 95 patients with postprogression tumor biopsy who underwent NGS with MSK-IMPACT. (B) Oncoprint of oncogenic, acquired alterations in postprogression samples with frequency. (C) Swimmers plot detailing clinical course of patients with histologic transformation. Amp, amplification; FGA, fraction of genome altered; HRD, homologous recombination deficiency; LCNEC, large cell neuroendocrine cancer; NGS, next-generation sequencing; WGD, whole genome doubling.
Lineage plasticity is of particular interest in EGFR-mutant lung cancer. A total of 14 patients (15%) initially had adenocarcinoma and underwent histologic transformation (SCLC, n = 6; squamous cell carcinoma [SQCC], n = 4; large cell neuroendocrine [LCNEC], n = 2; combined LCNEC/SCLC, n = 1; combined LCNEC/SQCC, n = 1) at the time of disease progression. Eight of these patients had baseline TP53/RB1 alterations (SCLC, n = 7; LCNEC/SCLC, n = 1), three had TP53 alterations only (SQCC, n = 2; LCNEC, n = 1), and three had neither (SQCC, n = 2; LCNEC/SQCC, n = 1). Clinical courses are found in Figure 2C. Median TTD of osimertinib in this subgroup was 15.2 mo. Median survival after transformation was 24.4 mo. Six patients were treated with local therapy; three of these cases continued osimertinib after the local therapy. Furthermore, 10 patients received platinum-etoposide after transformation, with time on treatment ranging from 2 to 33 months.
We assessed whether identified resistance mechanisms differed when patients received first-line osimertinib monotherapy versus combination therapy. The proportion of patients with transformation initially treated with osimertinib combination therapy was higher than those treated with osimertinib monotherapy (27% versus 7%, p = 0.02; Supplementary Fig. 4A and B). Ten patients receiving combination treatment and eight patients receiving osimertinib monotherapy had baseline concurrent EGFR/TP53/RB1 alterations. Conversely, gene amplifications and fusions were identified as mechanisms of resistance almost exclusively in patients treated with osimertinib monotherapy (p < 0.001). We further considered whether patients with baseline atypical EGFR alterations may have unique mechanisms of resistance to first-line osimertinib. In total, 15 patients with atypical EGFR alterations had postprogression biopsies (n = 8, paired; n = 7, post-progression only). Off-target (n = 6, total; MET amp [3], acquired RB1 loss, PIK3CA E545K, MDM2/CDK4 amp), on-target (n = 2, EGFR amp), and transformation (n = 1, squamous) were identified mechanisms of resistance. Although sample size was too small for formal comparison, the distribution of identified resistance mechanisms resembled that of patients with classical EGFR alterations.
Site of biopsy may affect identification of resistance mechanisms, particularly if biopsy is done on nonprogressive disease areas for safety or accessibility. Of the 95 patients with postprogression biopsies, 89 were performed at a progressive disease site (Supplementary Table 3). Of the six patients with biopsies at nonprogressive disease sites, five did not have an identified resistance mechanism. We further considered that certain biopsy sites may limit identification of resistance mechanism owing to tissue quality or quantity or that resistance mechanisms may differ by site of metastases, but no patterns emerged between category of resistance and biopsy site (Supplementary Fig. 4C).
Role of Identifying and Treating Resistance Mechanisms
There was no statistical difference in post-osimertinib progression survival between patients with on-target, off-target, histologic transformation, or unknown resistance mechanisms (transformation 27.1 mo, on-target 23.3 mo, unknown 13.3 mo, off-target 12.6 mo, p = 0.07; Fig. 3). In contrast, when using a time-dependent Cox model, there was a difference in post-osimertinib survival between patients who received no subsequent therapy (n = 10), subsequent tailored second-line therapy on the basis of biopsy results (n = 26), and subsequent second-line therapy not tailored to biopsy results (n = 59, p < 0.001; Supplementary Fig. 5A). The difference in survival was most notable between the tailored therapy cohort and those who received nontailored therapy (24-mo hazard ratio = 0.09). For the above-mentioned analyses, small cell-directed therapies were included in tailored second-line therapies; the survival benefit with tailored therapy was still present when histologic transformation cases were considered separately (24-mo hazard ratio = 0.02; Supplementary Fig. 5B and Supplementary Table 4).
Figure 3Mechanisms of resistance and postprogression survival. Post-progression survival for patients stratified by mechanism of identified resistance. CI, confidence interval; NR, not reached.
Resistance mechanisms were not identified for nearly half of patients with somatic tumor NGS testing, prompting exploration of whether additional methods for identifying drivers of resistance are warranted. Given that structural rearrangements mediate resistance in patients with EGFR-mutant NSCLC,
Landscape of acquired resistance to osimertinib in EGFR-mutant NSCLC and clinical validation of combined EGFR and RET inhibition with osimertinib and BLU-667 for acquired RET fusion.
Tumor analyses reveal squamous transformation and off-target alterations as early resistance mechanisms to first-line osimertinib in EGFR-mutant lung cancer.
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.
In the postprogression biopsy cohort, 40 of 95 patients (42%) underwent Archer FusionPlex, including 25 of 45 patients (56%) without an identified resistance mechanism (Fig. 4A). Three of the 25 patients with unknown resistance mechanisms had acquired fusions of unknown significance on Archer not identified by MSK-IMPACT (TGF-GPR128, MARS-GLI1, and GOLGA3-STAT6 fusions). Among patients with identified mechanisms of resistance on MSK-IMPACT (n = 15), Archer testing confirmed the one fusion, a TRIM24-BRAF fusion, also detected on MSK-IMPACT. Resistance mechanisms can be heterogeneous with multiple present in the same patient and may change over time. In Figure 4B, we illustrate a case with initial ERBB2 amplification pre-osimertinib, with acquired MET amp post-osimertinib. A subsequent sample after progression on osimertinib with capmatinib was positive for a CD74-ROS1 fusion identified on Archer. This case reveals clonal evolution in response to the selective pressures of targeted therapies that may be missed with somatic NGS DNA testing alone.
Figure 4Methods for investigating unknown resistance. (A) Results of RNAseq testing among 95 patients with postprogression tumor biopsy sample. (B) Single patient with serial biopsies analyzed using Archer, ROS1 IHC staining, and MSK-IMPACT. Figure illustrating clonality of each variant, total copy number of ERBB2/MET, and the presence of ROS1 fusion. Clonality is presented by RAF. (C) Sankey plot revealing evolution of mutational signatures derived from MSK-IMPACT before and after osimertinib treatment. (D) Patterns of mutational signatures and TMB in patients with known and unknown mechanisms of resistance. IHC, immunohistochemistry; PD, progressive disease; RAF, relative allele frequency; RNAseq, RNA sequencing; RT, radiation; T1, tumor biopsy 1; T2, tumor biopsy 2; T3, tumor biopsy 3; TMB, tumor mutational burden.
Alternatively, resistance to first-line osimertinib may be mediated by tumor evolution beyond acquisition of somatic single-gene alterations, including large-scale genomic events. Work from our group has validated identification of mutational signatures from targeted NGS data.
The most frequent dominant mutational signatures identified in paired postprogression tumor samples were APOBEC (28%) and clock/aging mutational signatures (36%) (Fig. 4C). There was a nonstatistically significant enrichment in APOBEC-dominant signature in the post-treatment compared with pretreatment samples (28% versus 10%, p = 0.07), suggesting that larger-scale mutational processes may contribute to the accumulation of somatic alterations driving osimertinib resistance. In addition, among 39 paired samples with tumor purity of greater than 20%, TMB was significantly higher in postprogression tumors compared with paired pretreatment tumors (median 5.3 versus 4.4 mutations per megabase [mut/Mb], p = 0.008; Supplementary Fig. 6A). Interestingly, four postprogression tumor samples had greater than or equal to 10 mut/Mb (Fig. 4D), including three cases with histologic transformation, suggesting an association between genomic instability and lineage plasticity. Neither FGA (p = 0.12; Supplementary Fig. 6B) nor WGD (Supplementary Fig. 6C) was significantly different between pre- and post-progression tumors, although the proportion of WGD in postprogression samples was higher (WGD in 71% versus 61%, p = 0.3; Supplementary Fig. 6C).
Discussion
The approval of first-line osimertinib was a significant paradigm shift in EGFR-mutant lung cancers. Rather than treating patients sequentially with multiple EGFR TKIs, the FLAURA trial revealed that upfront treatment with osimertinib prolongs OS, presumably owing to better on-target inhibition and avoiding development of heterogeneous-resistant subclones selected for by treatment with sequential EGFR TKIs.
Despite this success, further understanding of the evolutionary pressures driving resistance to first-line osimertinib is needed, as the approach of first-line osimertinib monotherapy for all patients is likely suboptimal. Furthermore, there are ongoing studies evaluating the efficacy of osimertinib combinations as first-line treatment. Little is known whether resistance mechanisms may differ with combination treatment. Using postprogression tumor samples that underwent NGS with MSK-IMPACT, including the largest retrospective data set to date of patients treated with first-line osimertinib, this work highlights the importance of identification of biomarkers mediating response and resistance and the limitations of current methods of identifying resistance.
The study used a real-world cohort of patients with EGFR-mutant lung cancers receiving first-line osimertinib with a median rwPFS shorter than what was reported in prospective trials of first-line osimertinib.
Reasons for this include no official Response Evaluation Criteria in Solid Tumors (RECIST) evaluation of disease progression, inclusion of patients with atypical EGFR alterations (not included in FLAURA), and capturing a broader, real-world population likely less fit and more diverse than patients on clinical trials.
For example, brain metastases were twice as high in this real-world cohort than the FLAURA study (41% versus 21%, respectively). Our cohort reinforced that baseline genomic alterations, namely concurrent TP53/RB1 alterations and atypical EGFR mutations, are associated with shorter rwPFS. Although limited gene panel sequencing or immunohistochemistry may rapidly identify EGFR alterations, comprehensive sequencing with NGS may risk stratify patients with poor prognostic genomic features at baseline. Patients with these pretreatment alterations may benefit from escalation of care such as adding chemotherapy to osimertinib and/or may prompt clinicians to pursue a change in treatment earlier at first signs of disease progression.
Although the landscape of acquired resistance to osimertinib has been reported in small series,
this is the largest data set of patients receiving first-line osimertinib who underwent a tumor tissue biopsy at progression and reveals that resistance to osimertinib is characterized primarily by accumulation of off-target events, histologic transformation, and unknown mechanisms of resistance. Interestingly, the landscape of mechanisms of resistance may differ for osimertinib monotherapy compared with combination treatment. Incidence of histologic transformation was higher with combination therapy whereas gene amplifications and gene fusions were more often found with osimertinib monotherapy. With several osimertinib combination studies ongoing including FLAURA2,
(NCT04035486), it will be paramount to understand how combination therapy alters the selective pressures that drives resistance. Most of the tumor biopsy samples (93%) obtained in this series were from progressive disease sites, reducing the likelihood of missing subclonal events perpetuating resistance. Furthermore, mechanisms of resistance identified did not seem to be associated with site of metastatic disease where biopsy was done.
Importantly, we report that categories of resistance are not prognostic; patients without an identified mechanism of resistance do not have shorter postprogression survival compared with patients with an identified mechanism of resistance. Nevertheless, intervening on an acquired mechanism improves postprogression survival, including recognition of and treatment for small cell transformation. Although small cell transformation is associated with a poor prognosis,
several patients in this study did remarkably well with a median post-transformation survival of 24 months, in part owing to local therapy interventions and systemic therapy directed at transformation. Patients presumably benefit from early identification of histologic transformation when the disease is still histologically heterogeneous and local therapy can address subclones that have transformed and osimertinib can continue to control the remaining adenocarcinoma. This is a strong rationale to encourage tumor tissue biopsies to identify cases with histologic transformation earlier.
The improvement in postprogression survival gained from intervening on identified mechanisms of resistance in the second-line setting underscores the importance of expanding methods for identifying and treating mediators of resistance. Postprogression tumor biopsies have distinct advantages than plasma testing in EGFR-mutant lung cancer for several reasons. Not only is tissue necessary to diagnose histologic transformation, but tissue may capture mechanisms not detected in plasma alone and can be used for additional exploratory analyses, such as RNAseq, immunohistochemistry, and/or epigenetic analyses. In our cohort, four patients had increased TMB post-osimertinib (TMB ≥ 10 mut/Mb), potentially selecting for patients who may benefit from immunotherapy,
Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study.
In lung cancer, there are no therapeutic interventions that target mutational signatures, but homologous repair deficiency signals sensitivity to PARP inhibitors in ovarian cancer,
establishing a paradigm where mutational signatures dictate therapeutic strategies. Further exploration is needed to see whether mutational signatures will result in additional treatment options in lung cancers, but changes in mutational signatures reveal larger-scale drivers of evolution under the selective pressures of targeted therapies. Postprogression samples were enriched for APOBEC or clock/aging mutational signatures, which are both associated with acquisition of subclonal events.
Although clock/aging-like mutational signatures naturally occur in either normal and tumor cells over time and are affected by the rate of cell division,
Concurrent inhibition of APOBEC enzymatic activity is therefore a possible avenue for addressing drug resistance to osimertinib treatment and worthy of prospective evaluation, although at present, no methods of direct inhibition of APOBEC enzymes are clinically available. Synthetic lethality-based therapeutic approaches targeting APOBEC mutagenesis are an area of active research in preclinical tumor models, including lung cancer,
In our view, indeed, EGFR-mutant lung cancers could represent an ideal subset where the inhibition of APOBEC enzyme activity should be explored to prevent the acquisition of subclonal mutations and intratumor heterogeneity and enhance the antitumor efficacy of available TKIs.
This work has several important limitations. First, rwPFS, although consistent with how clinicians define disease progression in real-world settings, is more subjective than RECIST–based PFS. Concurrent treatment with osimertinib and other cancer therapies was also allowed in this study, which may limit the generalizability of some results; however, it also provides valuable information on how mechanisms of acquired resistance differ when more intensive therapy (i.e., addition of chemotherapy) is used upfront. The sample size of patients with postprogression biopsies, although significantly larger than previously reported studies, may be underpowered to detect differences in postprogression survival among the different resistance subtypes. Finally, although patient and tumor characteristics of patients in the tumor biopsy cohort were similar to the larger clinical cohort treated with first-line osimertinib, there may be intrinsic differences in patients who underwent postprogression tumor biopsies compared with those who did not.
In conclusion, with only 50% of resistance to osimertinib driven by single-gene alterations, the mechanisms by which tumors escape EGFR-directed therapies are multifaceted and not fully characterizable by somatic NGS testing alone. Exploration should begin at the genome-wide level and narrow to the epigenome, gene, RNA transcript, and protein level as each analysis may reveal relevant mediators of resistance to validate and therapeutically harness. More practically, expanding clinical uptake of comprehensive molecular testing performed at disease progression on osimertinib may improve patient outcomes by identifying resistance mechanisms to target therapeutically.
CRediT Authorship Contribution Statement
Noura J. Choudhury: Conceptualization, Data curation, Formal analysis, Investigation, Writing—original draft, Writing—review and editing.
Antonio Marra: Conceptualization, Data curation, Formal analysis, Methodology, Investigation, Writing—original draft, Writing—review and editing.
Jane S. Y. Sui: Conceptualization, Data curation, Formal analysis, Investigation, Writing—original draft, Writing—review and editing.
Jessica Flynn: Formal analysis, Methodology, Writing—review and editing.
Soo-Ryum Yang: Formal analysis, Writing—review and editing.
Christina J. Falcon: Data curation, Project administration, Writing—review and editing.
Pier Selenica: Formal analysis, Writing—review and editing.
Adam J. Schoenfeld: Investigation, Writing—review and editing.
Natasha Rekhtman, Daniel Gomez, Mike F. Berger, Marc Ladanyi, Maria Arcila, Charles M. Rudin, Gregory J. Riely, Mark G. Kris: Resources, Writing—review and editing.
Glenn Heller: Formal analysis, Supervision, Methodology, Writing—review and editing.
Jorge S. Reis-Filho: Supervision, Methodology, Writing—review and editing.
Helena A. Yu: Conceptualization, Funding acquisition, Investigation, Methodology, Supervision, Writing—original draft, Writing—review and editing.
Acknowledgments
Dr. Yu is supported in part by the Haussler Fund, EGFR Resisters/LUNGevity Lung Cancer Research Award, Starr Foundation, and R01CA264078. Memorial Sloan Kettering authors are supported by the National Cancer Institute Cancer Center Core Grant number P30-CA008748.
Landscape of acquired resistance to osimertinib in EGFR-mutant NSCLC and clinical validation of combined EGFR and RET inhibition with osimertinib and BLU-667 for acquired RET fusion.
Tumor analyses reveal squamous transformation and off-target alterations as early resistance mechanisms to first-line osimertinib in EGFR-mutant lung cancer.
Acquired resistance of EGFR-mutant lung cancer to a T790M-specific EGFR inhibitor: emergence of a third mutation (C797S) in the EGFR tyrosine kinase domain.
Response to dual crizotinib and osimertinib treatment in a lung cancer patient with MET amplification detected by liquid biopsy who acquired secondary resistance to EGFR tyrosine kinase inhibition.
Tolerable and effective combination of full-dose crizotinib and osimertinib targeting MET amplification sequentially emerging after T790M positivity in EGFR-mutant non-small cell lung cancer.
Memorial Sloan Kettering-integrated mutation profiling of actionable cancer targets (MSK-IMPACT): a hybridization capture-based next-generation sequencing clinical assay for solid tumor molecular oncology.
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.
Continuing EGFR-TKI beyond radiological progression in patients with advanced or recurrent, EGFR mutation-positive non-small-cell lung cancer: an observational study.
Association of tumour mutational burden with outcomes in patients with advanced solid tumours treated with pembrolizumab: prospective biomarker analysis of the multicohort, open-label, phase 2 KEYNOTE-158 study.
Drs. Choudhury, and Marra contributed equally to this work.
Disclosure: Dr. Choudhury reports receiving institutional research funding from Merck, AbbVie, and Monte Rosa Therapeutics and royalties from Wolters Kluwer (Pocket Oncology). Dr. Marra reports receiving royalties from ililli. Dr. Schoenfeld reports having consulting/advising role to J&J, KSQ Therapeutics, Merck, Enara Bio, Perceptive Advisors, and Heat Biologics; and receiving research funding from GlaxoSmithKline (inst), PACT Pharma (inst), Iovance Biotherapeutics (inst), Achilles Therapeutics (inst), Merck (inst), Bristol Myers Squibb (inst), and Harpoon Therapeutics (inst). Dr. Gomez reports receiving grants from Varian, AstraZeneca, Merck, and Bristol Myers Squibb; receiving personal fees from Varian, AstraZeneca, Merck, US Oncology, Bristol Myers Squibb, Reflexion, WebMD, Vindico, and Medscape; and having served on the advisory board for AstraZeneca. Dr. Ladanyi reports having consulting or advisory roles for Takeda Oncology and Janssen Pharmaceuticals. Dr. Berger reports receiving personal fees from Roche; receiving grants from Illumina and Grail, outside of the submitted work; and having a patent Systems and Methods for Detecting Cancer Via cfDNA Screening pending. Dr. Arcila reports receiving personal fees for ad hoc board participation from AstraZeneca, Bristol Myers Squibb, and Jansen Global Services and speaker fees from Biocartis, Invivoscribe, RMEI Medical Education, Physician Education Resources, PeerView Institute for Medical Education, and I3 Health Education on Point outside of the submitted work. Dr. Rudin reports receiving personal fees from AbbVie, Amgen, Ascentage, AstraZeneca, Bicycle, Celgene, Daiichi Sankyo, Genentech/Roche, Ipsen, Jansen, Jazz, Lilly/Loxo, Pfizer, PharmaMar, Syros, Vavotek, Bridge Medicines, and Harpoon Therapeutics, outside of the submitted work Dr. Riely reports receiving grants from the National Institutes of Health/National Cancer Institute; having been an uncompensated consultant to Daiichi, Pfizer, and Mirati; receiving institutional research support from Mirati, Takeda, Merck, Roche, Pfizer, and Novartis; and having pending patents US20170273982A1 and WO2017164887A8. Dr. Kris reports receiving personal fees from Novartis, Genentech, Sanofi-Genzyme, AstraZeneca, Pfizer, Janssen, and Daiichi Sankyo; receiving honoraria for participation in educational programs from WebMD, OncLive, Physicians Education Resources, Prime Oncology, Intellisphere, Creative Educational Concepts, PeerView, i3 Health, Paradigm Medical Communications, AXIS, Carvive Systems, AstraZeneca, and Research to Practice; receiving travel support from AstraZeneca, Pfizer, Regeneron, Daiichi Sankyo, and Genentech; and receiving editorial support from F. Hoffmann-La Roche. Memorial Sloan Kettering has received research funding from The National Cancer Institute (USA), The Lung Cancer Research Foundation, and Genentech Roche for research conducted by Dr. Kris. Dr. Reis-Filho reports receiving personal/consultancy fees from Goldman Sachs, REPARE Therapeutics, Paige.AI, and Eli Lilly; having membership of the scientific advisory boards of VolitionRx, REPARE Therapeutics, and Paige.AI; having membership of the Board of Directors of Grupo Oncoclinicas; having ad hoc membership of the scientific advisory boards of Roche Tissue Diagnostics, Ventana Medical Systems, Novartis, Genentech, and InVicro, outside the scope of this study; and owning Paige.AI and REPARE Therapeutics stocks. Dr. Yu reports having consultant or advisory role for AstraZeneca, Janssen, Blueprint Med, Black Diamond, C4 Therapeutics, and Daiichi; and receiving research funding from AstraZeneca (inst), Pfizer (inst), Daiichi (inst), Cullinan (inst), Lilly (inst), Novartis (inst), and ERASCA (inst). The remaining authors declare no conflict of interest.
01932-3&id=gr1.jpg){kind=link}
01932-3&id=gr2.jpg){kind=link}
01932-3&id=gr3.jpg){kind=link}
01932-3&id=gr4.jpg){kind=link}