Genomic Profiling of Epidermal Growth Factor Receptor Mutation–Positive Non–Small Cell Lung Cancer after Progression on First-line Osimertinib: Phase II ORCHARD Study

Abstract

Purpose:

Osimertinib is the standard of care for first-line treatment for epidermal growth factor receptor–mutated (EGFRm) non–small cell lung cancer (NSCLC). Understanding the tumor molecular profile of patients following progression on osimertinib could help inform optimal second-line treatment.

Patients and Methods:

ORCHARD (NCT03944772), a phase II biomarker-directed study, enrolled patients with EGFRm NSCLC who progressed on first-line osimertinib to receive treatment based on their tumor molecular profile after progression. The study comprised three groups into which patients were allocated based on the molecular profile of their tumor, determined via next-generation sequencing (NGS) of a tumor biopsy. We report results from a prespecified, exploratory analysis of baseline tumor tissue and plasma samples to evaluate mechanisms of resistance to first-line osimertinib identified by tissue and plasma NGS. Agreement between tissue and plasma NGS data was also assessed.

Results:

This study provided a comprehensive dataset exploring tissue (n = 400) and plasma (n = 191) genomics, enabling characterization of the histogenomic landscape after first-line osimertinib treatment. TP53 and MDM2/4 alterations were mutually exclusive and occurred in 86% of tumors. When combining tissue and plasma genomics, resistance alterations were detected in 87% of samples, with multiple resistance alterations in 46%. Alterations in the PI3K pathway, SOX2, and MYC were frequently detected in histologically transformed tumors. Additionally, differential patterns of co-occurring EGFR mutations in tumors with L858R versus exon 19 deletion were observed.

Conclusions:

This comprehensive analysis highlights potential heterogeneous resistance to first-line osimertinib treatment, providing a rationale for combining treatments with broad activity to improve patient outcomes.

See related commentary by Gupta et al., p. 3718

Translational Relevance.

Based on the significant progression-free survival and overall survival benefit observed in the phase III FLAURA trial, osimertinib is a standard of care for the first-line treatment of patients with epidermal growth factor receptor–mutated (EGFRm) advanced NSCLC. Despite the observed efficacy of osimertinib, most patients eventually experience disease progression. A comprehensive understanding of the tumor molecular profile of patients following disease progression on first-line osimertinib treatment is essential for developing effective subsequent therapies. Through systematic histogenomic characterization, we identified resistance alterations, which were often heterogeneous, in the vast majority of patients. This informs drug development strategy for targeted versus broadly acting regimens in the second-line EGFRm NSCLC setting.

Introduction

Osimertinib is a third-generation epidermal growth factor receptor tyrosine kinase inhibitor (EGFR-TKI) that potently and selectively inhibits both EGFR-TKI–sensitizing and EGFR T790M resistance mutations, with demonstrated efficacy in EGFR-mutated (EGFRm) non–small cell lung cancer (NSCLC; refs. 1–11). Based on the significant progression-free survival and overall survival benefit observed in the phase III FLAURA trial (4, 5), osimertinib monotherapy is a standard of care for the first-line treatment of patients with EGFRm advanced NSCLC (12, 13).

Despite the observed efficacy of osimertinib, most patients eventually experience disease progression (14), so understanding the tumor molecular profile of patients following progression could help inform optimal second-line treatment. Previous reports from phase III trials analyzing resistance mechanisms to first- and second-line osimertinib therapy found MET amplification and secondary EGFR mutations (such as EGFR C797S) to be the most frequent genomic alterations (15, 16). These analyses were based on next-generation sequencing (NGS) of circulating tumor DNA (ctDNA) from paired plasma samples taken prior to osimertinib treatment and after progression or treatment discontinuation. However, ctDNA analysis has key limitations including the inability to detect histologic transformations and a lower sensitivity to detect copy-number alterations than tissue analysis (17–20). Analysis of paired tumor tissue samples obtained prior to first-line osimertinib and after progression identified histologic transformation in 15% of patients, primarily to squamous cell histology (21). Furthermore, these initial studies of osimertinib resistance mechanisms only analyze either plasma or tissue samples, limiting comparisons of these different analyses.

ORCHARD (NCT03944772), a phase II biomarker-directed platform study in patients with EGFRm advanced NSCLC with disease progression on first-line osimertinib monotherapy, was designed to assess resistance mechanisms to first-line osimertinib and the efficacy of subsequent postprogression treatment options, based on identified resistance mechanisms. Here, results from an exploratory analysis of baseline tumor tissue and plasma samples from ORCHARD, with an evaluation of the agreement between plasma and tissue detection methods, are presented.

Patients and Methods

Study design and patients

ORCHARD is a global, phase II, open-label, multicenter, biomarker-directed platform study designed to assess the efficacy and safety of multiple treatments, including osimertinib in combination with novel therapeutics, in patients with EGFRm advanced NSCLC whose disease had progressed on first-line osimertinib monotherapy. Details of the study design have been published previously (22). The study comprised three groups into which patients were allocated based on NGS analysis (FoundationOne CDx, Foundation Medicine Inc.) from a postprogression tumor biopsy (Supplementary Fig. S1).

Here, we present a prespecified, exploratory analysis of baseline tumor tissue and plasma samples to evaluate mechanisms of resistance to first-line osimertinib as identified by tissue and plasma NGS. Agreement between tissue and plasma NGS data was also assessed.

The trial was conducted in accordance with the Declaration of Helsinki, Good Clinical Practice guidelines (as defined by the International Conference for Harmonization), Council for International Organizations of Medical Sciences International Ethical Guidelines, applicable regulatory requirements, and the policy of the trial sponsor, AstraZeneca, on bioethics and human biological samples. Human investigations were performed after approval by the institutional review boards at each participating center (in accordance with local laws and guidance) and, in the United States of America, in accordance with an assurance filed with and approved by the FDA. All patients provided written informed consent prior to any study procedures or analyses. All data reported in these analyses are de-identified patient data.

Plasma ctDNA and tissue analyses

Patients provided tumor tissue for genomic profiling at screening and plasma samples at cycle 1, day 1. Tissue samples were analyzed using the FoundationOne CDx panel (Foundation Medicine Inc.), and plasma samples were analyzed using the GuardantOMNI 500 gene panel (Guardant Health). Local NGS analysis of tissue was permitted, but retrospective central NGS confirmation was required. Here, we present results from tissue and plasma samples which were centrally analyzed.

Genomic alterations were determined based on standard pipeline calls reported from the respective assays (23, 24). MET amplification was defined as ≥4 copies of MET over tumor ploidy per standard cutoffs from the manufacturer. Additional to the standard pipeline analyses, advanced genomic analyses were conducted on samples handled by Foundation Medicine Inc.

Statistical methods

This analysis was exploratory in nature, so data were summarized using descriptive statistics. Graphical methods and oncoprints were used to analyze the data. The Fisher exact test was used to evaluate co-occurrence or mutual exclusivity of alterations in patient subgroups. The median tumor mutational burden (TMB; as reported on FoundationOne CDx) between histologically transformed tumors and adenocarcinoma tumors was compared using a Mann–Whitney test. The co-occurrence of resistance alterations with EGFR mutations detected via tissue NGS was assessed using the Fisher's exact test.

Results

Patients

Overall, 706 patients were screened for ORCHARD; 483 patients provided tissue samples during screening, and 247 patients who received study drug provided plasma samples for NGS. Evaluable tissue and plasma results from central NGS testing were available from 400 and 191 patients, respectively (data cutoff October 22, 2024; Fig. 1A). The most common reasons for failed sequencing included insufficient tissue sample or insufficient DNA yields. Overall, 157 patients had matched tissue and plasma tumor biomarker status available. Baseline characteristics for the 434 patients with either tumor tissue or plasma biomarker results are shown in Table 1. The median age of patients was 62 years, 63% were female, 52% were White and 43% were Asian, and 40% had never smoked. The overall representativeness of the study population is described in Supplementary Table S1.

Figure 1.

A, Sample selection. Only results from central NGS testing are included in the analysis reported. B, Tissue-based resistance mechanisms to first-line osimertinib identified via NGS in all available samples. The cohort included patients screened for inclusion in the ORCHARD trial, with EGFRm advanced NSCLC, whose tumors had progressed on first-line osimertinib, with tumor tissue biomarker status available (n = 400). C, Presence/absence of TP53 and MDM mutations in tissue NGS. The Fisher's exact test was used to assess if TP53 and MDM mutations were mutually exclusive. The resulting OR of 0.033 (P < 0.001) can be interpreted as the MDM2/4 mutation group being less likely to have a TP53 mutation than patients without an MDM2/4 mutation (i.e., wild-type). D, Tissue-based resistance mechanisms to first-line osimertinib identified via NGS in tumors with transformed histology. E, TMB in tumors with transformed histology and adenocarcinoma tissue samples. Dots show individual samples, whereas lines show median and SEM. The comparison between median TMB between histologically transformed tumors and adenocarcinoma tumors was assessed using a Mann–Whitney test. CLIA, Clinical Laboratory Improvement Amendments; EGFRm, epidermal growth factor receptor–mutated; LCNEC, large cell neuroendocrine carcinoma of the lung; MSI, microsatellite instability; mut, mutated; NGS, next-generation sequencing; NSCLC, non–small cell lung cancer; OR, odds ratio; SEM, standard error of the mean; TMB, tumor mutational burden; WT, wild-type.

Table 1.

Baseline demographic and clinical characteristics.

Characteristic	ORCHARD baseline genomics cohort (N = 434)
Median age, years (range)	62 (26–88)
Age group, n (%)	​
≥18 to <65 years	247 (57)
≥65 years	187 (43)
Sex, n (%)	​
Male	159 (37)
Female	275 (63)
Race, n (%)	​
White	225 (52)
Asian	186 (43)
Black or African American	6 (1)
Native Hawaiian or other Pacific Islander	2 (<1)
Native American or Alaskan	2 (<1)
Other	10 (2)
Missing	3 (1)
Country, n (%)	​
United States of America	157 (36)
Republic of Korea	71 (16)
Japan	61 (14)
Spain	54 (12)
Netherlands	35 (8)
Italy	24 (6)
Norway	22 (5)
Sweden	9 (2)
Denmark	1 (<1)
Smoking status, n (%)	​
Never	172 (40)
Former	85 (20)
Current	5 (1)
Missing	172 (40)

Table shows baseline demographics and clinical characteristics for patients with a tumor tissue and/or plasma biomarker result (N = 434).

Genomic landscape of tissue-based resistance mechanisms to first-line osimertinib identified via NGS

Tissue-based resistance mechanisms to first-line osimertinib, identified via NGS and histologic evaluation, are shown in Fig. 1B and Fig. 2A (by frequency). The majority (60%) of tissue samples had known and potentially targetable genetic alterations, and 25% had multiple resistance mechanisms identified. MET alterations, including amplifications, fusions, mutations, and splice mutations, were the most common resistance mechanism, detected in 19% (76/400) of tissue samples, with amplifications being the most common MET alteration (18%; 71/400). One MET rearrangement (intragenic duplication within exon 14–exon 21), three mutations within the MET kinase domain (L1195V, H1094Y, and D1228H), and one splice site alteration were also detected. Secondary EGFR resistance mutations (C797X, L718X, and G724S) and EGFR amplifications occurred in 11% (45/400) and 35% (138/400) of tissue samples, respectively. EGFR T790M mutations were detected in 2% (7/400) of tissue samples, and one tissue sample had co-occurrence of C797S and T790M.

Figure 2.

A, Representation of tissue-based, plasma-based, and complementary tissue–based and plasma-based resistance mechanisms to first-line osimertinib identified via NGS by frequency. The cohort included patients screened on ORCHARD with EGFRm advanced NSCLC, whose tumors had progressed on first-line osimertinib, with tumor tissue biomarker status available (n = 400), with plasma biomarker status available (n = 191), and with both tumor tissue and plasma biomarker status available (n = 157). Note, rare co-occurring mechanisms across major groups occurred (see Figs. 1B, 2B and C) but were approximated here to provide a visual representation. Also, due to rounding, some percentages may not equal to 100%. B, Plasma-based resistance mechanisms to first-line osimertinib identified via NGS. The cohort included patients screened on ORCHARD with EGFRm advanced NSCLC, whose tumors had progressed on first-line osimertinib, with plasma biomarker status available (n = 191). C, Tissue- and plasma-based resistance mechanisms to first-line osimertinib identified via NGS. The cohort included patients enrolled on ORCHARD with EGFRm advanced NSCLC, whose tumors had progressed on first-line osimertinib, with both tumor tissue and plasma biomarker status available (n = 157). The percentages of tissue/plasma/total samples which were positive for the alteration noted were calculated as the total detected in tissue or plasma or both out of total evaluable samples. The proportion of positive signals detected by both tissue and plasma out of the number of positive signals in either tissue or plasma (agreement) was calculated as the total detected in both tissue and plasma out of total detected in tissue, plasma, or both. D, MET amplification in tissue and plasma by EGFR VAF status. CLIA, Clinical Laboratory Improvement Amendments; CNV, copy-number variant; ctDNA, circulating tumor DNA; EGFR, epidermal growth factor receptor; EGFRm, epidermal growth factor receptor–mutated; LCNEC, large cell neuroendocrine carcinoma of the lung; MSI, microsatellite instability; NGS, next-generation sequencing; NSCLC, non–small cell lung cancer; TMB, tumor mutational burden; VAF, variant allele frequency.

NKX2-1 amplifications and BRAF alterations were detected in 21% (82/400) and 8% (30/400) of tissue samples, respectively (BRAF fusions and V600E mutations were detected in 19 and 11 samples, respectively). Other potentially targetable alterations, including ERBB2 alterations, RAS mutations, and FGFR1/2/3 fusions, were rare, each occurring in 4% (17/400), 2% (8/400), and 2% (6/400) of tissue samples, respectively. RBM10 alterations, ERBB2 amplifications, and ERBB2 mutations were detected in 6% (25/400), 3% (13/400), and 2% (6/400) of tissue samples, respectively. The presence of MDM2/4 or TP53 alteration was detected in 86% (343/400) of tissue samples (Fig. 1B). MDM2/4 amplifications and TP53 alterations were mutually exclusive for most patients (P < 0.001; Fig. 1C).

Nineteen (5%) tissue samples had transformed histology: 13 transformed small cell lung cancers, 2 large cell carcinomas, 1 large cell neuroendocrine carcinoma, 2 adenosquamous carcinomas, and 1 squamous cell carcinoma. Baseline characteristics for the 19 patients with transformed histology are shown in Supplementary Table S2. There was a high prevalence of PI3K pathway alterations (including PIK3CA and PTEN alterations) in the transformed samples; 9 of 19 (47%) samples contained a functional mutation (Fig. 1D). SOX2 and MYC amplifications each occurred in 2 (11%) of the transformed samples and were mutually exclusive with PI3K pathway alterations. Furthermore, RB1 loss-of-function and TP53 mutations were enriched in the transformed samples [co-occurring in 68% (13/19) of transformed samples versus 10% (40/381) of nontransformed samples; P < 0.0001]. TMB was generally low across all samples; however, the median TMB was significantly higher in histologically transformed tumors versus adenocarcinoma tumors (2.52 vs. 3.78; P = 0.045; Fig. 1E).

Genomic landscape of plasma-based resistance mechanisms to first-line osimertinib identified via NGS

Plasma-based resistance mechanisms to first-line osimertinib, identified via NGS, are shown in Fig. 2A (by frequency) and Fig. 2B. In total, 23 of 191 (12%) plasma samples had no EGFR-sensitizing mutation identified. Similar to the tissue samples, MET alterations and secondary EGFR resistance mutations were the most common resistance mechanisms, detected in 17% (32/191) and 21% (41/191) of plasma samples, respectively. Most MET alterations were amplifications (16%; 30/191). NKX2-1 amplifications and RB1 deletions were each identified in 6% (12/191) of plasma samples. In total, 27% of samples had multiple resistance mechanisms identified.

Of the 19 tissue samples with transformed histology, 10 had paired plasma samples available. Co-occurrence of resistance alterations was more frequent in plasma (50%; 5/10) than tissue (16%; 3/19) samples (Fig. 1D; Supplementary Fig. S2). Of the 10 plasma samples, 4 (40%) had PI3K pathway alterations or MYC amplifications detected. As with tissue, RB1 loss-of-function and TP53 mutations were enriched in plasma samples with transformed histology and detected in 40% and 100% of samples, respectively.

Agreement between tissue and plasma NGS

The biomarker status, identified via NGS, of the 157 paired tissue and plasma samples is shown in Fig. 2A (by frequency) and Fig. 2C. Four of the paired samples (3%) had no EGFR mutation identified via tissue or plasma NGS, suggesting that these patients were erroneously enrolled into ORCHARD or had an independent new lesion after first-line osimertinib. Of these four samples, one had a KRAS G12D mutation and one had an ALK–EML4 fusion. Using tissue NGS alone or plasma NGS alone, 25% and 27% of samples, respectively, had multiple resistance mechanisms identified, which increased to 46% when combining tissue and plasma NGS. Moreover, only 14% (22/157) of samples had no known resistance alterations when using both tissue and plasma NGS (Fig. 2A).

There was strong agreement between tissue and plasma NGS for EGFR point/insertion–deletion mutations, except for patients who did not have an EGFR mutation detected via plasma NGS. Of 153 samples with EGFR-sensitizing or atypical EGFR mutations detected, 87% (133/153) were detected in both tissue and plasma samples, 12% (19/153) in tissue only, and 1% (1/153) in plasma only. Amplifications were generally more frequently detected in tissue samples; of 76 samples with EGFR amplifications detected using either modality, 30% (23/76) were detected in both tissue and plasma samples, 53% (40/76) in tissue only, and 17% (13/76) in plasma only. Similarly, of 41 samples with MET amplifications, 27% (11/41) were detected in both tissue and plasma samples, 46% (19/41) in tissue only, and 27% (11/41) in plasma only. NGS of plasma samples with low EGFR variant allele frequency (VAF; <5%) did not detect MET amplifications that were detected in tissue [1 vs. 14 samples, respectively; overall percent agreement (OPA), 85%; positive predictive value (PPV), 0%; and negative predictive value (NPV), 85%; Fig. 2D and Supplementary Table S3]. Regardless of EGFR VAF, tissue NGS was able to detect more MET amplifications than plasma NGS (30 vs. 22 samples, respectively; OPA, 81%; PPV, 50%; and NPV, 86%; Supplementary Table S4); however, plasma NGS in samples with high EGFR VAF (≥5%) detected more MET amplifications than tissue NGS (21 vs. 16 samples, respectively; OPA, 75%; PPV, 52%; and NPV, 87%; Supplementary Table S5). Agreement was lower for detection of BRAF alterations: 3 of 22 (14%) were detected in both tissue and plasma samples, 8 of 22 (36%) in tissue only, and 11 of 22 (50%) in plasma only. RAS mutations were most commonly detected in plasma only (5/10 samples). Conversely, although ALK fusions and ERBB2 mutations were rare, agreement between tissue and plasma detection was 60% and 40%, respectively.

Atypical and secondary resistance EGFR mutations

The frequency of atypical and secondary resistance EGFR mutations detected in tissue varied from 8% for C797S to <1% for E711Q and G724S (Fig. 3A). In total, 94 of 400 (24%) tissue samples had at least one atypical or secondary resistance EGFR mutation. The most commonly detected known secondary resistance EGFR mutations were C797S and L718X (Fig. 3B). L718X co-occurred exclusively with L858R (13/13, 100%), whereas C797X predominantly co-occurred with exon 19 deletion (Ex19del; 23/31, 74%; Fig. 3C).

Figure 3.

A, Tissue-based EGFR mutations identified via NGS. B, Tissue-based EGFR landscape of samples with at least one atypical or secondary resistance EGFR mutation to first-line osimertinib identified via NGS. C, Co-occurrence of EGFR atypical mutations with Ex19del and L858R in tissue samples. del, deletion; EGFR, epidermal growth factor receptor; Ex, exon; NGS, next-generation sequencing.

Atypical EGFR-sensitizing mutations, such as L861Q, tended to occur alone, and others, including E709X and S768I, co-occurred with other atypical or EGFR-sensitizing mutations (Fig. 3B). Atypical EGFR-sensitizing mutations co-occurring (not including T790M) with common EGFR mutations were predominantly with L858R (20/21, 95%) rather than Ex19del (1/21, 5%; Fig. 3C). No other alterations had strong co-occurrence patterns with atypical EGFR mutations or EGFR-sensitizing mutations (Supplementary Table S6). Of the 31 tissue samples with atypical EGFR mutations and no other EGFR-sensitizing mutations, 52% (16/31) had no known resistance mutations identified.

At least one atypical or secondary resistance EGFR mutation was detected in a similar proportion of all plasma samples (30%; 58/191; Supplementary Fig. S3) versus all tissue samples (24%; 94/400; Fig. 3B). However, there was a higher prevalence of secondary EGFR resistance mutations in plasma samples (21%) versus tissue samples (11%; Fig. 1B and Fig. 2B). Multiclonal EGFR and multiclonal PIK3CA/PTEN resistance mutations were identified in 4% (8/191) and <1% (1/191) of plasma samples, respectively. Seven samples had multiple C797S clones, one of which had a co-occurring L718Q mutation. Similar to tissue samples, EGFR T790M mutations were identified in 2% (4/191) of plasma samples. Of the 11 plasma samples with atypical EGFR mutations and no other EGFR-sensitizing mutations, 64% (7/11) had no known resistance mutations identified.

Discussion

This exploratory analysis of post-osimertinib tissue and plasma samples from ORCHARD provides the largest dataset to date evaluating resistance mechanisms to first-line osimertinib in tissue and plasma. MET alterations and secondary EGFR mutations were the most frequent mechanism of resistance detected in both tissue (19% and 11%, respectively) and plasma samples (17% and 21%, respectively). This is consistent with previous reports in which MET amplifications were detected in up to 18% of post-osimertinib samples (16, 25). The detection of secondary EGFR resistance mutations in ORCHARD was higher than that observed from NGS analysis of plasma samples in FLAURA (6%) but consistent with data from ELIOS (14% of post-osimertinib tissue samples; refs. 16, 25). MET amplification status based solely on ctDNA NGS (26) should be interpreted with caution because of its relatively low sensitivity and association with ctDNA content described here and previously (27). EGFR amplification was also frequently observed in our analysis (in 34% of tissue samples); interestingly, the ELIOS study showed that rates of EGFR amplification were similar in pre- and post-osimertinib tissue (25), suggesting that this alteration does not play a key role in osimertinib resistance.

NKX2-1 amplifications were detected in 21% of tissue samples in our analysis and have been shown to mediate tumorigenesis and confer osimertinib resistance in EGFRm NSCLC (28, 29). RBM10 alterations, identified in 6% of tissue samples, have been shown to limit response to EGFR inhibitors and are associated with poor prognosis (30, 31). Co-occurrence of RB1 and TP53 alterations in EGFRm lung cancer has been associated with a higher risk of histologic transformation (32). Notably, NKX2-1 amplifications and RB1 deletions were also each detected in 6% of plasma samples, respectively. These alterations were not identified in plasma samples from ELIOS likely due to the larger gene panel of the GuardantOMNI assay versus the Guardant360 assay used in ELIOS. EGFR T790M mutations were rarely detected in ORCHARD (2% each of tissue and plasma). T790M-mediated acquired resistance to first-line osimertinib was not observed in FLAURA (16, 25), and a recent report suggested that these T790M mutations could be a result of developmental mosaicism (33) rather than acquired resistance. Interestingly, MDM2/4 and TP53 alterations were largely mutually exclusive, consistent with the role of MDM2/4 in wild-type p53 degradation (34). The majority (86%) of tumors had an MDM2/4 or TP53 alteration detected, suggesting that the p53 pathway is suppressed in most NSCLC tumors following progression on first-line osimertinib.

NGS data were available from 157 paired tissue and plasma samples, allowing a comprehensive evaluation of the two different testing methods. Tissue samples allow for increased sensitivity because of their high tumor content and are required for assessing histologic transformation. Conversely, plasma has lower tumor content (and thus lower sensitivity) but potentially provides information on resistance mechanisms from multiple lesions (tumor heterogeneity). Consistent with this, there was high agreement between tissue and plasma NGS data for the detection of EGFR-sensitizing mutations, as would be expected for clonal driver mutations. However, amplifications were more commonly detected in tissue, reflecting the challenge for plasma-based copy-number detection in the context of low tumor content. Tissue NGS was also able to detect MET amplifications that were undetectable in plasma samples at EGFR VAF of ≤5%, suggesting that tissue NGS may provide greater sensitivity in detecting copy-number alterations in patients with low ctDNA fractions. However, plasma NGS detected more MET amplifications than tissue in plasma samples at EGFR VAF >5%, suggesting that plasma analyses may elucidate details of subclonal MET amplifications. Despite this, IHC and FISH tissue-based testing remains the gold standard methodologies for identifying MET overexpression and amplification, respectively, and are being explored in ongoing clinical trials (35–38). Agreement between tissue and plasma NGS data was lower for BRAF alterations and RAS mutations, with these alterations detected more frequently in plasma. Additionally, although alterations detected in tissue (such as MET amplification, BRAF alterations, and ERBB2 amplifications) were generally mutually exclusive, this was not the case in plasma samples. These findings further demonstrate the benefit of plasma NGS at identifying subclonal resistance and tumor heterogeneity. Overall, our data suggest that tissue and plasma NGS provides complementary information for detecting resistance mechanisms to osimertinib. Indeed, when both assays were used, only around 14% of patients had no known resistance alterations detected versus 29% with tissue and 41% with plasma alone. Additionally, utilizing both tissue and plasma samples, NGS detected multiple resistance mechanisms in 46% of samples versus 25% with tissue and 27% with plasma alone. These results align with guidelines which are moving toward recommending tissue- and/or plasma-based molecular testing (13, 39).

Of note, detection of known resistance alterations was low for patients with atypical EGFR mutations (in the absence of a common EGFR mutation), across both tissue (57%; 16/28) and plasma (64%; 7/11) analyses. Interestingly, atypical mutations and L718X secondary resistance mutations were generally found to co-occur predominantly with L858R rather than Ex19del. However, specific secondary resistance mutations, including C797X, tended to co-occur with Ex19del. Similar patterns of secondary EGFR mutations were reported in a real-world US study assessing resistance after osimertinib treatment (40) in patients with EGFRm NSCLC. Apart from EGFR amplification and RBM10 alterations, which occurred more frequently with L858R versus Ex19del, all other resistance alterations occurred at a similar frequency in tissue samples with L858R and Ex19del.

Overall, 19 (5%) samples showed transformed histology, primarily to small cell carcinoma, which is lower than that reported in previous analyses (10%–15%; refs. 21, 41), possibly due to prescreening of samples at some sites, with transformed cases not being submitted for study testing. In general, TMB was low across all samples, consistent with previous observations in patients with EGFRm NSCLC (42). However, TMB levels were higher across samples with transformed histology versus those with adenocarcinoma histology. Furthermore, there was a high rate of PI3K pathway alterations in histologically transformed tumors, as well as enrichment of RB1 loss-of-function and TP53 mutations. Two thirds of transformed samples had a PI3K, SOX2, or MYC alteration detected, supporting the potential role of these alterations in histologic transformation. This mutation profile is consistent with previous reports of histologically transformed tumors (32, 43, 44).

Several limitations should be considered when interpreting these data. As the analysis was restricted to samples after osimertinib treatment, determination of preexisting versus acquired alterations was limited. Moreover, the patient population within ORCHARD may be biased based on referral to a clinical trial, with potential overrepresentation of patients with tumors harboring rare resistance mechanisms and underrepresentation of patients with tumors harboring alterations with no targeted treatment within ORCHARD or patients with tumors with transformation to squamous or small cell lung cancer. Additionally, only 24 (13%) plasma samples had no sensitizing EGFR mutation and atypical EGFR mutation detected, slightly lower than that reported in FLAURA (19%; ref. 16), meaning that this analysis may have included more patients with more aggressive disease or larger tumors, potentially due to the tissue biopsy inclusion criteria. The ORCHARD study design evolved, with modules added as new treatments, and markers were identified; as such, our analysis may not be fully representative of a general post-osimertinib patient population. Nevertheless, this analysis provides a comprehensive overview of resistance mechanisms identified using NGS with tissue and plasma in a large dataset.

Many of the resistance mechanisms identified in this analysis are targetable; a retrospective analysis of patients who had progressed on first-line osimertinib showed that tailoring subsequent therapy based on post-progression biopsy results was associated with improved postprogression survival (45), highlighting the importance of identifying mechanisms of resistance and adapting treatment accordingly. Currently, treatment options for patients who have progressed on first-line osimertinib include platinum-based doublet chemotherapy and amivantamab plus carboplatin and pemetrexed (13). However, additional, targeted treatment options are needed for patients who progress on first-line osimertinib, as this population continues to be challenging to treat successfully (46). Promising efficacy with osimertinib plus MET inhibitors after first-line osimertinib or other EGFR-TKIs in the phase II SAVANNAH (47–49) and INSIGHT-2 (50) studies and phase III SACHI study (51) has been observed in patients with tumors harboring Met alterations, and the confirmatory phase III SAFFRON study investigating osimertinib plus savolitinib is ongoing (52). For those without targetable resistance mechanisms, favorable efficacy was observed with osimertinib plus datopotamab deruxtecan in a biomarker-unmatched module of ORCHARD (53); the phase III TROPION-Lung 15 (54) study will further investigate this combination. The ongoing modules within ORCHARD will provide further information on the efficacy and safety of specific biomarker-directed therapy after progression on osimertinib. Future analyses should also consider polyclonal resistance and its impact on treatment outcomes.

In conclusion, this study provides the most comprehensive histogenomic characterization after first-line osimertinib treatment. When combining histology analysis with tissue/plasma genomics, resistance alterations/transformations were detected in 87% of patients, with 46% of patients showing multiple resistance alterations. We have shown that the TP53–MDM2/4 pathway is frequently altered, as well as frequent genetic alterations in the PI3K pathway, SOX2, and MYC in histologically transformed tumors. Atypical EGFR and EGFR L718X mutations were most commonly associated with L858R tumors, whereas EGFR C797X mutations were most commonly associated with Ex19del tumors. This comprehensive characterization of the histogenomic landscape suggests potential heterogeneous resistance and supports the evaluation of osimertinib and broadly acting combination partners, such as datopotamab deruxtecan.

Supplementary Material

Supplementary Data

Supplementary Material Supplementary Table S1. Representativeness of study participants. Supplementary Table S2. Baseline demographic and clinical characteristics of patients with transformed histology. Supplementary Table S3. MET amplification detected by tissue and plasma NGS in patients with <5% EGFR VAF. Supplementary Table S4. MET amplification detected by tissue and plasma NGS in all patients. Supplementary Table S5. MET amplification detected by tissue and plasma NGS in patients with ≥5% EGFR VAF. Supplementary Table S6. Co-occurrence of resistance alterations with EGFR mutations detected via tissue NGS. Supplementary Figure S1. ORCHARD study design. Supplementary Figure S2. Plasma-based resistance mechanisms to first-line osimertinib identified via NGS in samples with transformed histology. Supplementary Figure S3. Plasma-based EGFR landscape of samples with at least one atypical or secondary resistance EGFR mutation to first-line osimertinib.

Data Availability

Data underlying the findings described in this article may be obtained in accordance with AstraZeneca’s data sharing policy described at https://astrazenecaclinicaltrials.com/our-transparency-commitments. Data for studies directly listed on Vivli can be requested through Vivli at www.vivli.org. Data for studies not listed on Vivli could be requested through Vivli at https://vivli.org/members/enquiries-about-studies-not-listed-on-the-vivli-platform/. AstraZeneca Vivli member page is also available outlining further details: https://vivli.org/ourmember/astrazeneca/

Authors’ Disclosures

H.A. Yu reports other support from AstraZeneca, Janssen Pharmaceuticals, Daiichi Sankyo, Taiho Pharmaceutical, SystImmune, Amgen, AbbVie, and Cullinan Therapeutics during the conduct of the study. K.H. Tang is employed by and holds shares in AstraZeneca. A. Markovets reports other support from AstraZeneca's stocks outside the submitted work. R. Hartmaier is employed by and holds shares in AstraZeneca. P.E. Smith reports receiving consulting fees from AstraZeneca. B.C. Cho reports other support from Champions Oncology, Crown Bioscience, Imagen, and PearlRiver Bio GmbH; grants from CJ Bioscience, Cyrus Therapeutics, ImmuneOncia Therapeutics, Janssen Pharmaceuticals, J INTS BIO, MSD, Yuhan Corporation, Dong-A ST, and LigaChem Biosciences; personal fees from Amgen, ArriVent BioPharma, AstraZeneca, Bristol Myers Squibb, Boehringer Ingelheim, CJ Bioscience, Cyrus Therapeutics, Gilead Sciences, GSK, Regeneron Pharmaceuticals, Janssen Pharmaceuticals, MSD, Yuhan Corporation, American Society of Clinical Oncology, AstraZeneca, Guardant Health, Roche, European Society for Medical Oncology, International Association for the Study of Lung Cancer, Korean Cancer Association, Korean Society of Medical Oncology, Korean Society of Thyroid-Head and Neck Surgery, Korean Cancer Study Group, Novartis, MSD, The Chinese Thoracic Oncology Society, Pfizer, Zai Lab, J INTS BIO, and KANAPH Therapeutics Inc.; and personal fees and other support from Yonsei University Health System and DAAN Biotherapeutics outside the submitted work. A.J. De Langen reports other support from AstraZeneca, Bristol Myers Squibb, Boehringer Ingelheim, Pfizer, Eli Lilly and Company, MSD, and Nuvalent and grants from AstraZeneca, Bristol Myers Squibb, Boehringer Ingelheim, Daiichi Sankyo, and MSD outside the submitted work. S.B. Goldberg reports grants and personal fees from AstraZeneca; grants from Boehringer Ingelheim, Mirati Therapeutics, and Adela; and personal fees from Eli Lilly and Company, Daiichi Sankyo, Johnson & Johnson, Summit Therapeutics, Merck, Regeneron Pharmaceuticals, Bayer, Synthekine, and Tubulis outside the submitted work. J.W. Goldman reports grants and personal fees from AstraZeneca during the conduct of the study, as well as grants and personal fees from AbbVie; grants, personal fees, and other support from Summit Therapeutics; and grants from SystImmune outside the submitted work. X. Le reports personal fees from AstraZeneca during the conduct of the study, as well as personal fees from Eli Lilly and Company, EMD Serono (Merck KGaA), AstraZeneca, Spectrum Pharmaceuticals, Novartis, Regeneron Pharmaceuticals, Boehringer Ingelheim, Hengrui Pharmaceuticals, Bayer, Teligene, Taiho Pharmaceutical, Daiichi Sankyo, Janssen Pharmaceuticals, Blueprint Medicines, Sensei Biotherapeutics, SystImmune, ArriVent BioPharma, Abion, BlossomHill Therapeutics, and AbbVie outside the submitted work. E. Iwama reports personal fees from AstraZeneca and Daiichi Sankyo outside the submitted work. J. Cosaert is employed by AstraZeneca. J.W. Riess reports grants and personal fees from AstraZeneca, Boehringer Ingelheim, Pfizer, Nuvalent, ArriVent BioPharma, and Merck; personal fees from Daiichi Sankyo, Natera, OncoHost, Genentech, Amgen, Verastem Oncology, Foundation Medicine, GSK, Replimune, Taiho Pharmaceutical, Nuvation Bio, and Regeneron Pharmaceuticals; and grants from Revolution Medicines, Summit Therapeutics, and IO Biotech outside the submitted work. Z. Piotrowska reports grants from AstraZeneca during the conduct of the study as well as personal fees from Boehringer Ingelheim, Bayer, Gilead Sciences, Tubulis, Taiho Pharmaceutical, Summit Pharmaceuticals, Revolution Medicines, Genmab, Eli Lilly and Company, Black Diamond Therapeutics, Natera, Merck, Sanofi, and Regeneron Pharmaceuticals; grants, personal fees, and other support from Janssen Pharmaceuticals; grants and personal fees from Takeda, AbbVie, Blossom Hill Therapeutics, Daiichi Sankyo, Blueprint Medicines, and Cullinan Therapeutics; personal fees and other support from AstraZeneca; other support from Genentech; and grants from Novartis, Spectrum Pharmaceuticals, and Tesaro/GSK outside the submitted work. In addition, Z. Piotrowska reports serving as international coordinating principal investigator for AstraZeneca and Cullinan Therapeutics. No disclosures were reported by the other authors.

Authors’ Contributions

H.A. Yu: Writing–original draft, writing–review and editing. K.H. Tang: Formal analysis, visualization, writing–original draft, writing–review and editing. A. Markovets: Data curation, formal analysis, visualization, methodology, writing–original draft, writing–review and editing. R. Hartmaier: Conceptualization, supervision, methodology, writing–original draft, writing–review and editing. P.E. Smith: Formal analysis, methodology, writing–original draft, writing–review and editing. B.C. Cho: Conceptualization, resources, data curation, formal analysis, supervision, investigation, visualization, writing–original draft, writing–review and editing. A.J. De Langen: Writing–original draft, writing–review and editing. S.B. Goldberg: Conceptualization, investigation, writing–original draft, writing–review and editing. J.W. Goldman: Conceptualization, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. X. Le: Writing–original draft, writing–review and editing. E. Iwama: Resources, writing–original draft, writing–review and editing. J. Cosaert: Conceptualization, formal analysis, supervision, investigation, writing–original draft, writing–review and editing. J.W. Riess: Data curation, writing–original draft, writing–review and editing. Z. Piotrowska: Conceptualization, writing–original draft, writing–review and editing.