Functional Heterogeneity in MET Pathway Activation in PDX Models of Osimertinib-resistant EGFR-driven Lung Cancer

Abstract MET pathway activation is one of the most common mechanisms of resistance to osimertinib in EGFR-mutant non–small cell lung cancer (NSCLC). We previously demonstrated spatial and temporal heterogeneity in MET pathway activation upon osimertinib resistance in EGFR-mutant NSCLC; however, the functional relevance of these findings is unclear. Here, we generated 19 patient-derived xenografts (PDX) from 9 patients with multi-region and temporal sampling of osimertinib-resistant tumor tissue from patients with EGFR-mutant NSCLC. MET pathway activation was a putative mechanism of osimertinib resistance in 66% (n = 6/9) patients from whom PDXs were generated. Significant spatial and temporal heterogeneity in MET pathway activation was evident. Osimertinib-resistant PDXs with MET amplification by FISH (defined as MET/CEP7 ratio ≥2.0 or mean MET ≥ 6.0 copies/cell) and high-level phospho-MET, but not c-MET expression, had better responses to osimertinib and savolitinib combination than to osimertinib alone. MET polysomy tumors by FISH from both PDXs and patients had evidence of subclonal phospho-MET expression. Select MET polysomy PDX tumors with phospho-MET expression responded better to osimertinib and savolitinib combination than MET polysomy PDX tumors without phospho-MET expression. Our results suggest osimertinib and savolitinib combination is most effective for osimertinib-resistant EGFR-mutant tumors with MET pathway activation as evidenced by phospho-MET. As subclonal MET amplification may be evident in MET polysomy tumor progression, MET polysomy warrants close clinical follow-up with phospho-MET IHC in parallel with FISH diagnostic. Significance: Using a novel cohort of in vivo PDX models of MET pathway activation with acquired resistance to osimertinib in EGFR-mutant lung cancer, we demonstrate that phospho-MET may be a clinically relevant assay to guide treatment selection with osimertinib and savolitinib combination. In addition, our work shows that patients with MET polysomy tumors may have subclonal MET amplification and therefore require close follow up for the use of osimertinib and savolitinib combination.


Introduction
EGFR tyrosine kinase inhibitors (TKI) are the standard-of-care therapy for advanced EGFR-mutant non-small cell lung cancer (NSCLC).Despite excellent after EGFR TKI resistance.Indeed, the TATTON study (19) was designed to test whether the combination of osimertinib and savolitinib (a potent, selective MET TKI) could overcome MET-mediated osimertinib resistance in EGFRmutant NSCLC.This study reported an objective partial response rate of 33% (n = 23/69) among patients with EGFR-mutant NSCLC previously treated with third-generation TKI such as osimertinib (19).While the efficacy results from this trial are encouraging, a better understanding of MET-mediated osimertinib resistance is required to further optimize patient selection for osimertinib and savolitinib combination.For example, the type of assay used to detect METmediated osimertinib resistance may be important for patient selection.In the TATTON study, however, no major difference in efficacy was evident based on the type MET assay used, that is, MET amplification by FISH (MET/CEP ratio ≥2) or next-generation sequencing (≥20% tumor cells, coverage of ≥200 × sequencing depth and ≥5 copies of MET over tumor ploidy), MET polysomy by FISH (copy number ≥5 if MET/CEP ratio is <2) or MET protein expression by tissue IHC (MET +3 expression in ≥50% of tumor cells; ref. 19).
These results suggest additional MET assays may be beneficial to aid in patient selection of osimertinib and savolitinib combination.Heterogeneity of MET pathway activation in osimertinib-resistant EGFR-mutant NSCLC, as we previously have shown (10), may also be an important factor in assessing response to osimertinib and savolitinib combination.Unfortunately, due to the limitations of single-tissue biopsies, the TATTON study was not able to assess for potential heterogeneity in MET-mediated resistance.
One of the challenges of studying MET-directed therapies for osimertinibresistant EGFR-mutant NSCLC is lack of relevant preclinical models, as the current evidence for MET-directed therapies is largely based on in vitro model systems (13).In this study, we developed a novel cohort of EGFR-mutant NSCLC patient-derived xenograft (PDX) models with heterogenous MET pathway activation, and evaluated osimertinib and savolitinib combination treatment in vivo to address the following questions: (i) What are the optimal MET assays to determine efficacy of osimertinib and savolitinib combination?and (ii) Does MET pathway heterogeneity impact the efficacy of osimertinib and savolitinib combination?
On the basis of our results, we propose that phospho-MET detection by IHC is the MET-directed assay that best predicts in vivo response to osimertinib and savolitinib combination.Moreover, we found MET polysomy by FISH (MET ≥ 4.0 and <6.0 copies/cell; four or more MET signals observed in at least 40% of cells; or five or more MET signals observed in at least 10% of cells if MET/CEP ratio is <2) to be a potential early precursor of MET amplification by FISH (defined as MET/CEP ratio ≥2.0 or mean MET ≥ 6.0 copies/cell) and thus, MET polysomy may represent an important subset of patients for osimertinib rechallenge and osimertinib and savolitinib combination.Altogether, our results demonstrate functional heterogeneity in MET pathway activation upon osimertinib resistance in EGFR-mutant NSCLC, which may help guide the use of osimertinib and savolitinib combination in the clinic.

Tissue Acquisition and PDX Generation
PDXs were generated from patients enrolled in a single-arm, single-institution, open-label phase II study of osimertinib treatment and local ablative therapy (LAT) upon progression on osimertinib in EGFR-mutant metastatic lung adenocarcinoma (NCT02759835).Tumor samples were obtained for PDX generation at the time of osimertinib resistance and at any point after as clinically indicated.One tissue core was split in half and implanted into nod scid gamma mice (RRID:IMSR_JAX:005557, the Jackson Laboratory) mice at 8 weeks of age for PDX development.

In Vivo Drug Studies
Cohorts for efficacy studies were implanted with PDX tumors at passage 4 (LAT001_9B), (LAT001_6B), and (LAT006_0118), and at passage 5 for (LAT006_2B-1216).Tumor size was monitored by caliper measurement, and treatment in efficacy studies started when tumor volume reached between 200 and 500 mm 3 .Recruitment of paired mice in equal numbers to different treatment groups was staggered as necessary with 5 to 10 mice per group treated in any given study.Pharmacodynamic studies were performed using passage 6 donor from PDX-LAT006_2B-1216 (representing MET polysomy by FISH), and PDX-LAT006_0118 (representing MET amplification by FISH).Osimertinib (AZD9291) was provided by AstraZeneca (SN#1123772831, AZ#13552748-030) and savolitinib was purchased from Chemietek (AZD9291, CT-A9291, free base).Osimertinib was dissolved in 1% polysorbate (tween, Sigma) 80 in water; savolitinib was dissolved in 0.5% carboxymethylcellulose sodium at pH 2.1.Both drug formulations were administered orally.Once enrolled in efficacy studies, mice were treated with osimertinib daily at 5 mg/kg (10 mL/kg of a 0.5 mg/mL solution) or 10 mg/kg (10 mL/kg of a 1.0 mg/mL solution), and savolitinib daily at 10 mg/kg (10 mL/kg of a 1.0 mg/mL solution) or 25 mg/kg (10 mL/kg of a 2.5 mg/mL solution).Treatment continued until all vehicletreated mice had reached 3,000 mm 3 tumor size endpoint.If euthanasia was necessary during the treatment period, it was performed 4 hours after the last dose.For all mice, tumor tissues were collected for further analysis.One tumor fragment was preserved in neutral buffered formalin (NBF) for histopathology, and two flash-frozen tumor fragments were saved for molecular analyses.Treatment in the pharmacodynamic studies commenced when tumors reached 600 mm 3 .Mice were then treated for 5 days, and tissue harvested 4 hours after the last dose.Treatment with osimertinib at 10 mg/kg, savolitinib at 10 mg/kg, and the drug combination was compared with vehicle-treated mice.

IHC and Quantitative IHC
Paraffin sections (5 μm) were prepared before staining and subjected to either EDTA antigen retrieval with the following primary antibodies: p-Met (1:300, Cell Signaling Technology #3077, RRID:AB_2143884), Cytokeratin 7 (CK-7, Abcam #181598, RRID:AB_2783822), thyroid transcription factor (TTF-1, 1:100, Abcam #227652, no RRID), or citrate-based antigen retrieval c-MET (1:500, Abcam #51067, RRID:AB_880695).Primary antibody incubation was performed for 1 hour at room temperature on a Leica BondMax autostainer (Leica) utilizing the Polymer Refine kit (Leica).All slides were counterstained with hematoxylin, dehydrated, and permanently mounted.For quantitative IHC, slides were scanned and analyzed using the Aperio scanning/analysis system.For automated quantification of c-MET and phospho-MET IHC, slides were scanned and analyzed using Aperio ImageScope (ImageScope, RRID:SCR_014311, Leica).For automated quantification, the entire slide area was selected for analysis to avoid any sampling bias.This resulted in approximately 10,000 to 200,000 cells quantified per tumor.The H ("histology") score is a semiquantitative weighted method that considers the relative staining intensity of positive cells.It allows for better comparison across samples, especially for membrane-specific stains that vary in intensity.The H-score reflects the percentage of positive cell number representing all levels of intensity.

Real-time PCR Assays for Copy-number Variation Analyses
Real-time PCR was conducted using an ABI Viia7 Real-Time PCR System (Applied Biosystems).Multiplex reactions (10 μL) containing FAM-MGB MET copy-number assay (Applied Biosystems: Hs01602615_cn) and VIC-TAMRAlabeled telomerase reverse transcriptase (TERT) copy-number reference assay (Applied Biosystems) were performed in quadruplicate using 384-well plates with 10 ng template and 1x Universal Master Mix (Applied Biosystems without Amp Erase UNG).Determination of copy number was performed using the comparative Ct method (delta, delta Ct) with normalization to TERT as an internal reference for copy number (Applied Biosystems ViiA 7Real-Time PCR System Getting Started Guides).Patient germline genomic control DNA samples carrying two copies each of MET and TERT were used as calibrator samples.Samples were analyzed in quadruplicate, and values expressed as the mean ± SE.Copy-number analyses are based on the premise that all controls and unknowns carry two copies of TERT; thus, the ratio of MET to TERT can be used to assign MET copy number.

FISH
FISH for MET was performed by Chromosome Pathology Unit, Lab. of Pathology, NCI, NIH.MET was considered amplified when MET/CEP ≥ 2.0 or mean MET ≥ 6.0 copies/cell.MET polysomy was defined as mean MET ≥ 4.0 and <6.0 copies/cell; four or more MET signals observed in at least 40% of cells; or five or more MET signals observed in at least 10% of cells.MET was considered negative when above criteria are not satisfied and indeterminate when technical issues prevented the test from being reported as positive, negative, or equivocal.For each FISH assay, a minimum of 100 cells were examined.

Quantification and Statistical Analysis
All figures and graphs were generated using the "ggplot2" package available through the R statistical program.Correlations and t tests were conducted though the R base packages.All tests were two tailed and P values less than 0.05 were considered significant.Statistical analyses were performed with Prism 8 Software, version 8.4.3, 2020.Results are shown as mean ± SD.Mantel-Cox, Gehan-Breslow-Wilcoxon, unpaired t test, and multiple t tests were used for multiple comparison between control group and treatment groups.For efficacy studies, treatment groups had 10 to 11 mice, except slow growing PDX (LAT006_2B), which had 5 to 6 mice recruited per group.For the pharmacodynamic study and phospho-MET analysis, there were 5 mice per treatment group.Representative studies with treatment of interest are reported.

Data Availability
The experimental data and PDXs generated in this study are available upon request from the corresponding author.

Generation of Osimertinib-resistant EGFR-mutant NSCLC PDX Models with Spatial and Temporal Heterogeneity in MET Pathway Activation
We generated PDX models using multi-region spatial and temporal sampling of tumors from patients with EGFR-mutant NSCLC enrolled in a prospective clinical trial assessing LAT after osimertinib resistance (NCT02759835; Fig. 1A).Overall, we generated 19 PDXs from a total of 9 patients (Fig. 1B; Supplementary Table ).As we reported previously (10), MET pathway activation was a putative mechanism of osimertinib resistance in 66% (n = 6/9) of patients from whom PDXs were generated (Fig. 1B).Two or more PDXs were generated from patients LAT001, LAT006, and LAT015, in which MET pathway activation was the likely mechanism of resistance to osimertinib (Fig. 1B).Within each of these 3 patients, there was heterogeneity in MET genomic alterations, as PDXs gener-  inhibitor (savolitinib) with a third-generation EGFR TKI (osimertinib).PDXs with spatial heterogeneity in MET pathway activation (LAT001_6B and LAT001_9B), PDXs with temporal heterogeneity in MET pathway activation (LAT006_2B and LAT006_0118) and an additional validation PDX (LAT015_6B) were treated with vehicle, osimertinib, savolitinib, and osimertinib plus savolitinib combination followed by assessment of efficacy and identification of predictive markers.
spatially separated into nine sections for analysis, of which three were implanted into immunodeficient (NSG) mice, and all three successfully developed into PDXs (Fig. 1B).Among the LAT001 PDXs, 9B was MET amplified (mean MET copies: 7.8, MET/CEP ratio: 2.2) and 6B had MET polysomy (mean MET copies: 4.0, MET/CEP ratio: 1.1).Patient LAT006 was treated with osimertinib as first-line therapy and upon osimertinib resistance had a left lower lung lobectomy for local ablation.The tumor-bearing lobe was spatially separated into seven sections of which three were injected into NSG mice, and all successfully developed into PDXs (LAT006 1B, 2B, 4B; Fig. 1B).In addition, LAT006 PDX 0317 was developed from a supraclavicular lymph node biopsy taken after the patient's second progression on osimertinib, that is, progression after LAT and osimertinib re-challenge, and LAT006 PDX 0118 was developed from a biopsy taken after combination treatment with chemotherapy and immunotherapy (Fig. 1B).Among the LAT006 first progression PDXs, 2B showed MET polysomy (mean MET copies: 4.8, MET/CEP ratio: 0.9).LAT006 PDX 0317 showed MET polysomy (mean MET copies: 4.6, MET/CEP ratio: 1.2).The LAT006 PDX 0118 was MET amplified (mean MET copies: 8.5, MET/CEP ratio: 2.3).Patient LAT015, who was treated with osimertinib as second-line therapy for development of EGFR T790M mutation had a partial hepatectomy (LAT) from which five PDXs were successfully developed (Fig. 1B).LAT015 PDX 6B exhibited MET polysomy (mean MET copies: 4.7, MET/CEP ratio: 1.0).We confirmed the origin of the PDXs used for efficacy studies by comparing Sanger sequencing of driver EGFR mutation(s) in PDX tumor tissue with patient tumor tissue (Supplementary Table ).Additional details of these PDXs and clinical information of the patients from which they were derived are available in Supplementary Table and have also been described previously (10).

Differential Efficacy of Osimertinib and Savolitinib Combination in MET Amplified and MET Polysomy Osimertinib-resistant EGFR-mutant PDXs
To assess the functional relevance of MET genomic heterogeneity, we selected PDXs from individual patients with MET amplification (LAT001_9B and LAT006_0118) and MET polysomy tumors (LAT001_6B, LAT006_2B, and LAT015_6B; Fig. 1C).Efficacy studies were performed on MET amplified and MET polysomy PDXs with treatment arms consisting of vehicle, osimertinib, savolitinib, and osimertinib and savolitinib combination (Figs.1D, 2A and B).Treatment continued until all vehicle-treated mice had reached tumor size endpoint.We found that the MET polysomy PDX tumors retained sensitivity to osimertinib as a single agent (Fig. 2A), while the MET amplified PDX tumors demonstrated only partial sensitivity (Fig. 2B).Both MET polysomy and amplification tumors showed no significant difference in response to savolitinib monotherapy compared with vehicle treatment (Fig. 2A and B).In MET amplified PDX models, osimertinib and savolitinib combination inhibited tumor growth and significantly lengthened time to tumor size endpoint compared with osimertinib or savolitinib alone (Fig. 2B).In contrast, there was no significant difference in extent of tumor growth inhibition in the osimertinibtreated tumors compared with the osimertinib and savolitinib combination arms within the MET polysomy PDXs (Fig. 2A).However, LAT006_2B MET polysomy origin demonstrated response heterogeneity in individual mice to both single drug and combination treatment (Fig. 2A).Several tumors in this cohort of mice exhibited increased time to progression and long-term durable response after withdrawal of both savolitinib monotherapy and combination treatment but not osimertinib monotherapy (Fig. 2A), suggesting a sole dependence on MET pathway activity in these select tumors, which has been described previously (20).Overall, these in vivo experiments suggest MET amplified tumors and select MET polysomy osimertinib-resistant EGFR-mutant NSCLC tumors can benefit from combination therapy with osimertinib and savolitinib.

Predictors of Response to Osimertinib and Savolitinib Combination in MET Amplified and MET Polysomy Osimertinib-resistant EGFR-mutant PDXs
We next sought to identify clinically relevant predictors of response to osimertinib and savolitinib combination across our cohort of osimertinib-resistant PDXs.In addition to MET FISH, we performed MET copy-number assays on PDX tumors from each efficacy study and evaluated phospho-MET and c-MET protein expression by IHC.MET/CEP FISH ratio and higher MET copy number were significantly associated with response to osimertinib and savolitinib combination (Fig. 2B).Expression of phospho-MET by IHC, but not c-MET was associated with response to osimertinib and savolitinib combination (Fig. 2C).PDX tumors with response to osimertinib and savolitinib combination exhibited strong phospho-MET expression throughout the tumor (quantified by H-score) whereas PDX without response had overall low phospho-MET expression (Fig. 2C and D).PDXs with and without response to osimertinib and savolitinib combination displayed high c-MET expression (Fig. 2D).Thus, MET FISH amplification and phospho-MET expression are predictors of response to osimertinib and savolitinib combination in these osimertinib-resistant PDX tumors, whereas MET expression does not appear to be a predictor.

Osimertinib and Savolitinib Combination Suppresses AKT Signaling in MET Amplified Osimertinib-resistant EGFR-mutant NSCLC PDXs
We next sought to assess whether the differences in response to osimertinib and savolitinib combination treatment in MET amplified and MET polysomy tumors could be explained by the drugs' effect on EGFR, MET, and downstream AKT and MAPK signaling pathways.We conducted pharmacodynamic studies in which mice bearing MET amplified and MET polysomy PDX tumors from patient LAT006 were treated for a period of 5 days with vehicle, savolitinib, osimertinib, or osimertinib and savolitinib combination.Tumor tissues were harvested after the last drug dose and analyzed by immunoblotting.Both phospho-EGFR and phospho-MET were expressed in the MET amplified PDX (LAT006_0118; Fig. 3A).Osimertinib treatment alone decreased phospho-EGFR expression, but knockdown of both phospho-EGFR and phospho-MET was observed only in the osimertinib and savolitinib combination treatment arms (Fig. 3A and C).Combination therapy also resulted in increased inhibition of phospho-AKT and phospho-S6 compared with single agent osimertinib (Fig. 3A and C).In contrast, there were no differences in MEK1/2 and ERK phosphorylation in tumors treated with osimertinib and savolitinib combination compared with single agent osimertinib (Fig. 3A and C).
The MET polysomy PDX, LAT006-2B, also expressed phospho-EGFR, although at lower levels compared with the MET amplified PDX (Fig. 3B and  D).However, one PDX treated with savolitinib had high phospho-EGFR expression potentially due to EGFR amplification, which we previously reported in this patient (ref.10; Fig. 3B and D).Phospho-MET was only evident in the osimertinib-treated MET polysomy PDX, suggesting that inhibition of phospho-EGFR may result in feedback activation of MET signaling in these tumors (Fig. 3B and D).In contrast to the MET amplified PDX, osimertinib treatment in the MET polysomy PDX led to reduction in phospho-AKT with no increased inhibition with combination treatment (Fig. 3B and D).Thus, these results suggest that the effectiveness of osimertinib and savolitinib combination in osimertinib-resistant MET amplified PDXs may be through inhibition of downstream AKT signaling.Furthermore, baseline phospho-MET expression as an indicator of MET pathway activation predicts the sensitivity to osimertinib and savolitinib combination in osimertinib-resistant PDXs.

Heterogeneity in MET Polysomy and MET Amplification in Osimertinib-resistant EGFR-mutant NSCLC
Despite the overall low phospho-MET H-scores in MET polysomy tumors (Fig. 2C), closer examination of phospho-MET IHC revealed focal areas of phospho-MET staining varying from low to high intensity (Fig. 4A).
In addition, we found evidence of low-frequency MET amplification by FISH (range: 2%-8% MET amplified cells) across the three MET polysomy osimertinib-resistant PDX models (LAT001_6B, LAT006_2B, and LAT015_6B) demonstrating early, subclonal MET amplification (Fig. 4A) occurring at a higher frequency than previously reported among untreated patients with EGFR-mutant NSCLC (21).To further understand how MET amplification may evolve within a given patient, we performed phospho-MET IHC and MET FISH of LAT006 patient tumors at first and second progression on osimertinib (MET polysomy) and at a later timepoint after chemoimmunotherapy (MET amplification; Fig. 4B).There was a low percentage of MET amplified cells (6% of cells) at first progression on osimertinib (Fig. 4B, left) concordant with the LAT006_2B MET polysomy PDX (MET amplification in 2% of cells; Fig. 4A, middle) generated from the same patient tumor.Strikingly, the patient tumor at second progression on osimertinib, although classified as MET polysomy (mean MET copies 4.8 and MET/CEP ratio of 1.5), showed approximately 26% MET amplified cells and focal areas with positive membranous phospho-MET staining (Fig. 4B, middle).After chemoimmunotherapy, the patient exhibited a MET amplified tumor (mean MET copies 11.2 and MET/CEP ratio of 3.6) with MET amplification in approximately 93% of cells and concordant strong, diffuse membranous phospho-MET IHC staining (Fig. 4B, right).
These results provide evidence from serial, temporal sampling that defined as MET polysomy tumors may concurrently have subclonal MET amplification and phospho-MET expression.

Phospho-MET IHC to Assess MET Pathway Activation in Pre-and Post-osimertinib Resistant EGFR-mutant NSCLC Patient Tumors
To further assess the added value of phospho-MET detection as a complement to MET FISH assays, we performed phospho-MET IHC along with c-MET IHC and MET FISH in two patient tumors (LAT028 and LAT021) pre-and post-osimertinib resistance.Pre-osimertinib tumor biopsies from patient LAT028 showed MET amplification by copy number (total of 7.7 MET copies; Fig. 5A) and patient LAT021 showed MET polysomy (based on four or more MET signals observed in at least 40% of cells with a MET/CEP ratio <2; Fig. 5B).Both patient tumors displayed evidence of subclonal MET amplification with 42% and 5% MET/CEP > 2.0 cells, respectively in LAT028 (Fig. 5A) and LAT021 (Fig. 5B).Intriguingly, while the post-osimertinib MET/CEP ratios across multiple sampled tumors from patient LAT028 were similar to the pre-osimertinib MET/CEP ratio (none reached the >2.0 MET/CEP ratio threshold), all of the post-osimertinib LAT028 tumors exhibited strong, diffuse phospho-MET staining compared with negative phospho-MET IHC staining in the pre-osimertinib tumor biopsy suggesting additional factors may contribute to MET phosphorylation in this patient (Fig. 5A).Thus, despite evidence of pre-osimertinib MET pathway activation based on MET FISH, patient LAT028 did not have evidence of active (phosphorylated) MET pre-treatment that would suggest benefit from osimertinib and savolitinib combination.In patient LAT021, the pre-osimertinib biopsy was negative for phospho-MET by IHC and the MET/CEP ratio was 1.1, indicating no MET amplification.The post-osimertinib biopsy was positive for phospho-MET by IHC and MET FISH indicating MET amplification, which thereby demonstrates how these two tests can also be concordant (Fig. 5A).
Overall, based on our in vivo and patient data, we suggest a clinical pathway incorporating phospho-MET IHC staining in addition to standard MET FISH assays to guide treatment for patients with osimertinib-resistant EGFR-mutant lung cancer (Fig. 5C).

Discussion
MET pathway activation is the most common mechanism of resistance to osimertinib in EGFR-mutant NSCLC (4,7,9,10,22).To our knowledge, this is the first study to systematically generate and functionally assess the role of osimertinib and savolitinib combination in osimertinib-resistant EGFR-mutant NSCLC PDX models with heterogenous MET pathway activation.Our results in PDX models are partially in agreement with results from the TATTON study (19), as tumors with a MET/CEP ratio ≥2 by FISH are likely to respond to osimertinib and savolitinib combination.However, in this cohort of osimertinib-resistant PDXs, total MET protein expression by IHC did not correlate with response in our PDX models, suggesting that this criterion alone may not identify patients likely to benefit from osimertinib and savolitinib combination.Total MET protein expression by IHC has previously been shown to poorly correlate with MET FISH amplification (23).In fact, recent data suggest that patients with EGFR-mutant lung cancer selected for high tumor MET

AACRJournals.org
Cancer Res Commun; 4(2) February 2024    Our extensive interrogation of the spatial and temporal heterogeneity of MET pathway activation upon osimertinib resistance in EGFR-mutant lung cancer may also explain other recent findings from the TATTON study (19).In this study, patients with MET polysomy (defined as copy number ≥5 if MET/CEP copy-number ratio <2) upon osimertinib resistance had similar overall response rate (28%) compared with patients with MET amplification (31%).We propose that MET polysomy patients who benefited from the osimertinib and savolitinib combination may still have had sensitivity to osimertinib or may have had MET amplification in other tumor areas not uncovered by a single-site biopsy.Alternatively, some MET polysomy tumors may have high MET copy number and phospho-MET staining, but without a MET/CEP copy-number ratio ≥2 such as we demonstrated in patient LAT028.
There are several limitations to our study.While our PDXs showed similar heterogeneity in MET pathway activation as the patient tissues from which they were derived, the PDX were generated from independent tumor areas.We also cannot rule out that MET pathway activation may have fluctuated upon generation of our PDXs, as has been previously demonstrated in patients on and off erlotinib (25), as the PDXs were off osimertinib during their establish-ment.However, all PDX tumors were rigorously evaluated at each passage for retention of the characteristics of the original biopsy.
In conclusion, we functionally assessed MET pathway biomarkers using in vivo efficacy studies from a novel cohort of osimertinib-resistant EGFR-mutant NSCLC PDX models.We demonstrated that the heterogeneity and differential drug sensitivity found in patients with lung adenocarcinoma can be recapitulated in PDX models, providing a valuable tool for evaluation of therapeutics.
Future studies could include long-term treatment of polysomy tumor PDXs with osimertinib to evaluate subclonal changes in MET activation, as well as exploration of additional combined targeted therapies for drug resistance.
Our current study supports the use of osimertinib and savolitinib combination in patients with osimertinib-resistant EGFR-mutant NSCLC tumors and MET pathway activation.Evaluation of MET pathway activation by MET phosphorylation and more careful consideration of MET polysomy tumors in osimertinib-resistant EGFR-mutant NSCLC may help inform ongoing prospective studies of osimertinib plus savolitinib such as SA-VANNAH (NCT03778229), ORCHARD (NCT03944772), and SAFFRON (NCT05261399).
ated from different regions within the tumors showed either MET amplification by FISH (MET/CEP ≥ 2.0 or mean MET ≥ 6.0 copies/cell) or MET polysomy by FISH (MET ≥ 4.0 and <6.0 copies/cell; four or more MET signals observed in at least 40% of cells; or five or more MET signals observed in at least 10% of cells if MET/CEP ratio is <2).Patient LAT001 was treated with osimertinib as first-line therapy, and upon osimertinib resistance had a partial hepatectomy for local ablation.The osimertinib-resistant tumor from patient LAT001 was

FIGURE 1
FIGURE 1 Generation of spatial and temporally heterogenous osimertinib-resistant EGFR-mutant NSCLC PDX models and treatment study design.A, Schematic diagram of the prospective clinical trial of LAT for osimertinib-treated EGFR-mutant lung cancer (RT: radiotherapy).PDXs were generated from osimertinib-resistant tumor tissue either at first or second progression on osimertinib or after standard of care (S.O.C) therapy.B, Multi-region and temporal tumor samples from surgical resections or biopsies used for PDX generation are shown for each individual patient.Putative mechanism of resistance to osimertinib as evidenced by exome and or transcriptome sequencing (Continued on the following page.)

FIGURE 2
FIGURE 2 Efficacy and determinants of response to osimertinib and savolitinib combination among osimertinib-resistant EGFR-mutant NSCLC PDX models with spatially and temporally heterogenous MET pathway activation.Tumor growth inhibition studies in MET polysomy (A) and MET amplified (B) PDXs.Dashed vertical lines delineate when treatment was stopped.Asterisks signify statistical significance between osimertinib and savolitinib combination and osimertinib alone treatment arms.P values were calculated by t test.P values < 0.05 were considered significant.C, Association between response to osimertinib and savolitinib combination and IHC features (phospho-MET, c-MET), copy number and FISH parameters (number of MET copies and MET/CEP7 ratio).Response is defined as >25 days until reaching tumor size endpoint in osimertinib and savolitinib combination compared with osimertinib treatment alone.Individual circles represent a unique tumor for each represented PDX model.D, Representative c-MET and phospho-MET IHC images of PDX tumors with and without response to osimertinib and savolitinib combination.Scale bars, 50 μm.

FIGURE 3
FIGURE 3 Osimertinib and savolitinib combination suppresses AKT signaling in MET amplified osimertinib-resistant EGFR-mutant NSCLC PDXs.EGFR and MET pathway protein expression in MET amplified (A) and MET polysomy (B) LAT006 PDXs treated with either vehicle, osimertinib, savolitinib, and osimertinib plus savolitinib.Bar graphs show the relative quantification of phosphorylated proteins normalized to total protein expression in MET amplified (C) and MET polysomy (D) LAT006 PDXs.

FIGURE 4
FIGURE 4 Heterogeneity in MET polysomy and MET amplification in osimertinib-resistant EGFR-mutant NSCLC.A, Representative phospho-MET IHC (several are reused from Fig. 2D), MET FISH images and MET FISH scoring from PDXs with MET polysomy.B, Representative phospho-MET IHC, MET FISH images and MET FISH scoring from longitudinally collected tumor samples from patient LAT006 at first progression on osimertinib, second progression after osimertinib rechallenge, and upon further progression on chemoimmunotherapy.Scale bars, 50 μm.

FIGURE 5
FIGURE 5 Phospho-MET expression is an indicator of MET activity post-osimertinib treatment; and proposed clinical flow diagram for treating EGFR-mutant NSCLC with evidence of MET pathway activation after osimertinib resistance.Representative phospho-MET IHC, MET FISH images and MET FISH scoring from pre-and post-osimertinib resistant tumors from patient LAT028 (multiple spatially heterogenous post-osimertinib resistant tumors shown; A) and from patient LAT021 (B).Scale bars, 50 μm.C, Clinical flow diagram for osimertinib-resistant EGFR-mutant NSCLC with evidence of MET pathway activation.
Tumors were passaged as fragments of approximately 5 mm implanted subcutaneously into recipient mice.PDX model generation was conducted following procedures under an approved Animal Study Protocol and according to Frederick National Laboratory Animal Care and Use Committee guidelines.NCI-Frederick is accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International and follows the Public Health Service Policy for the Care and Use of Laboratory Animals.Animal care was provided in accordance with the procedures outlined in the "Guide for Care and Use of Laboratory Animals" (National Research Council; 1996; National Academy Press; Washington, D.C.).
(19)ein expression have infrequent co-occurring MET amplification(24), whereas the majority of patients with MET FISH amplified tumors are also MET IHC positive(19), indicating that correlations among the various methodologies for detecting MET activity are not well defined.In the osimertinib-resistant EGFR-mutant NSCLC PDX models presented here, phospho-MET IHC and MET/CEP ratio ≥2 by FISH predicted response to osimertinib and savolitinib combination.Thus, our data suggest that incorporation of a phospho-MET IHC assay, particularly among patients with MET polysomy, or otherwise equivocal MET FISH report, may offer additional information regarding the use of MET kinase inhibitors.While phospho-MET IHC, to our knowledge, has not been systematically assessed in large patient cohorts, our evaluation of phospho-MET in patient samples showed the feasibility of using such a test in the clinic.Various IHC biomarkers are FDA approved as companion diagnostics and many FDA-approved biomarkers encompass and rely on multiple platforms including RT-PCR, FISH, or NGS with corresponding therapeutic indications.Further retrospective and prospective assessment of phospho-MET IHC among MET amplified and MET polysomy tumors (based on FISH) is warranted.Our in vivo data have implications for the management of patients with osimertinib resistance with evidence of MET activation, in particular MET polysomy (defined in our study as MET ≥ 4.0 and <6.0 copies/cell if MET/CEP ratio is <2).As our data demonstrate that MET polysomy PDXs can be sensitive to osimertinib as a single agent, we believe patients with MET polysomy tumors should be analyzed in separate cohorts from patients with MET amplified tumors in clinical trials of osimertinib and MET kinase inhibitors.Practically, if MET polysomy is present in a biopsy without concurrent strong, diffuse phospho-MET, rechallenge with osimertinib alone may offer clinical benefit.Importantly, as our serial biopsy sampling revealed subclonal MET amplification in MET polysomy tumors, repeat biopsies of patients with MET polysomy to test for MET amplification and/or phospho-MET IHC is warranted after rechallenge with osimertinib or additional therapies.Repeat tumor biopsies may also be particularly important given the observation in the TATTON study that relying on ctDNA mediated assessment of MET amplification may result in a high level of false negatives(19).