RPC-9 gefitinib-resistant lung adenocarcinoma cells harboring EGFR exon 19 deletion mutation and T790M were established in our laboratory (17). H1975 pulmonary adenocarcinoma cells harboring L858R and T790M were purchased from the American Type Culture Collection.

Cells were cultured in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum in a 37°C incubator with a humidified 5% CO₂ atmosphere, where the oxygen levels were maintained at either 21% (normoxia) or 1% (hypoxia). Growth inhibition was determined using a modified 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assay, as described previously (17). Each assay was performed in triplicate.

Cells were seeded in 6-well plates at a density of 5×10⁴ cells/well, grown under normoxic or hypoxic conditions for 48 h, and then grown with or without various concentrations of osimertinib. After three days, the cells were fixed with 10% formalin for 10 min, stained with crystal violet solution (Sigma-Aldrich) for 10 min, and then washed with H₂O. After the plates had been dried overnight, stained cells were quantified using ImageJ software (version 1.52a, National Institute of Health).

Cells and frozen tissues were lysed using radioimmunoprecipitation assay buffer [1% Triton X-100, 0.1% SDS, 50 mmol·L-1 Tris/HCl (pH 7.4), 150 mmol·L-1 NaCl, 1 mmol·L-1 ethylenediaminetetraacetic acid, 1 mmol·L-1 ethylene glycol tetraacetic acid, 10 mmol·L-1 β-glycerol phosphate, 10 mmol·L-1 NaF, 1 mmol·L-1 sodium orthovanadate-containing protease inhibitor tablets (Roche Applied Sciences)]. Proteins were separated by electrophoresis on polyacrylamide gels, transferred onto nitrocellulose membranes, and probed with specific antibodies that were detected using Enhanced Chemiluminescence Plus (GE Healthcare Biosciences). Bands were detected using an ImageQuant LAS-4000 imager (GE Healthcare Biosciences).

Gefitinib, cetuximab, and bevacizumab were purchased from EVERLTH. Osimertinib was purchased from Selleck Chemicals. Antibodies against phospho-EGFR (#3777), EGFR (#2232), phospho-ERK (#9101), ERK (#9102), phospho-AKT (#9271), AKT (#9272), hypoxia inducible factor-1α (HIF-1α; #36169), GAPDH (#2118), and CD31 (#77699) were purchased from Cell Signaling Technology. Anti-TGF-α antibodies (#ab9585) were purchased from Abcam.

A Human Phospho-RTK Array Kit (R&D Systems) was used according to the manufacturer's instructions. Bands and dots were detected using an ImageQuant LAS-4000 imager (GE Healthcare Biosciences). Mean pixel density was measured using ImageJ (version 1.52a, National Institute of Health).

RNA was extracted from cells using an RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. Total cDNA was synthesized and amplified using a PrimeScript RT Reagent Kit (Perfect Real Time; TaKaRa). RNA expression was analyzed using real-time quantitative reverse transcription-PCR (qRT-PCR) with SYBR Premix Ex Taq II (Tli RNase H Plus; TaKaRa), according to the manufacturer's protocol. PCR amplification was performed using a LightCycler Real-Time PCR System (Roche Applied Science), and gene dosage was calculated using a standard curve analysis. PCR was carried out using primers (forward, 5′-AGATTCCCACACTCAGTTCTGCTTC-3′; reverse, 5′-ACAGCGTGCACCAACGTACC-3′).

Female 5–7-week-old athymic mice were purchased from Charles River Laboratories. All mice were provided with sterile food and water and were housed in a barrier facility under a 12 h light/dark cycle. Cells (5×10⁶) were injected bilaterally into the back of each mouse. After 7–21 days, the mice were randomly divided into groups and then treated either with a mono-, double, or triple therapy (3–4 mice per group) consisting of a vehicle, osimertinib (per os 5 mg/kg, five times a week), cetuximab [intraperitoneal (i.p.) 1 mg/mouse, twice a week], and bevacizumab (i.p. 5 mg/kg, twice a week). Each drug was administered for 28 days, with a 28 day follow-up period. Tumor volume (width² × length/2) was measured twice a week. Euthanasia was then induced via administration of 5% isoflurane via an anesthesia machine. In addition to the assessment of inhibitory effect, the mice were treated with each drug similar to above mentioned regimen for 3 days, following which the tumor samples were collected for immunohistochemical analysis after euthanasia. Experimental protocols were approved by the Animal Care and Use Committee of Okayama University, Okayama, Japan (OKU-2017084, OKU-2020152).

Tissue samples were fixed using formalin, embedded in paraffin, and cut to a thickness of 5 µm before being placed on glass slides and deparaffinized in Hemo-De (FALMA) and graded alcohol. For antigen retrieval, sections were incubated in 10 mmol/l sodium citrate buffer, pH 6.0, for 10 min in a 95°C water bath, after which the sections were incubated with 0.3% hydrogen peroxide in methanol to block endogenous peroxidase activity. The slides were rinsed with Tris-buffered saline containing 0.1% Tween-20, and the sections were blocked with goat serum for 60 min. The sections were incubated with anti-CD31, anti-HIF-1α, or anti-TGF-α antibodies overnight at 4°C (dilution factors described in the data sheets) and amplified using biotinylated anti-rabbit antibodies and avidin-biotinylated horseradish peroxidase (HRP) conjugate for 30 min (SignalStain Boost IHC Detection Reagent (HRP, rabbit) #8114, Cell Signaling Technology, Inc.), reacted with 3,3-diaminobenzidine, and counterstained with hematoxylin.

Statistical analyses were performed using STATA software version 15.1 (StataCorp.). Differences between two groups were compared using two-tailed paired Student's t-tests. Differences between three or more groups were compared using one-way ANOVA followed by Bonferroni's test. P-values of <0.05 were considered statistically significant. The Pearson correlation coefficient was calculated using STATA software version 15.1 (StataCorp.).

First, we assessed the inhibitory effect of osimertinib monotherapy on tumor growth in a mouse xenograft model derived from RPC-9 cells harboring EGFR exon 19 deletion and T790M mutation. Two types of xenograft lung cancer models were prepared using the RPC-9 cells to achieve different starting tumor volumes (conventional: 200 mm³ or large: 500 mm³). Both models were administered with osimertinib (5 mg/kg, 5 times/week by gavage) for 28 days and then observed for another 28 days. Consistent with the in vitro data (Fig. S1A), osimertinib monotherapy significantly inhibited tumor growth compared to the vehicle in both the conventional (maximum tumor diameter at day 56; vehicle: 15.4 mm ± 0.67 vs. osimertinib: 6.9 mm ± 0.74, mean ± SE) and large models (maximum tumor diameter at day 56; vehicle: 15.0 mm ± 0.97 vs. osimertinib: 8.7 mm ± 2.2, mean ± SE) (Fig. 1A and B); however, maximum tumor regression was significantly lower in the large model than in the conventional model (−73.8% vs. −89.3% on day 24, P=0.015, t-test; Fig. 1C and D). These results suggest that although osimertinib monotherapy is effective, the magnitude of its antitumor effect in the xenograft model is affected by tumor volume.

To explore the cause of distinct tumor inhibition between the conventional and large models, we performed pathological examinations on tumors from both models. Since we suspected that hypoxia may have been induced by the increase in tumor volume, we measured HIF-1α expression. As expected, we observed higher HIF-1α expression in tumors from the large model than in the conventional model (Fig. 2A). HIF-1α regulates the transcription of various growth factors (18); therefore, we investigated the phosphorylation status of receptor tyrosine kinase (RTK) using a Phospho-RTK array (Fig. S1B). The phosphorylation of most RTKs, including EGFR, was lower in tumors from the conventional model treated with osimertinib for 7 days (Fig. 2B), whereas the inhibition of RTK phosphorylation (other than EGFR) was generally limited in the large model (Fig. 2C). In addition, the inhibitory effect of osimertinib on EGFR phosphorylation was lower in tumors from the large model than the conventional model. Due to the observed increase in HIF-1α expression and the modest inhibition of EGFR phosphorylation, we also measured TGF-α expression, finding that TGF-α expression was higher in tumors from the large model than the conventional model (Fig. 2A). As expected, a significant correlation was observed between HIF-1α and TGF-α expression levels in these tumors (Fig. S1C).

Consequently, we assessed the effect of HIF-1α expression on sensitivity to osimertinib in vitro. As reported previously (19), HIF-1α and TGF-α expression were induced in RPC-9 cells cultured under hypoxic conditions (Fig. 2D and E). In addition, sensitivity to osimertinib was significantly lower in RPC-9 cells pre-incubated under hypoxic conditions than in cells pre-incubated under normoxic conditions (Figs. 2F and S1D). Together, these results suggest that the HIF-1α/TGF-α axis may account for the differing osimertinib sensitivity observed in the conventional and large models.

Next, we tested the effect of combination therapy with osimertinib and bevacizumab (OsiBev) in both the conventional and large models derived from RPC-9 cells, since EGFR-TKI plus bevacizumab is one of the most clinically relevant combination therapies (12). In the conventional model, combination therapy with OsiBev inhibited tumor growth to the same extent as osimertinib monotherapy during the treatment period (Fig. 3A) and the maximum tumor regression did not differ significantly between therapies (osimertinib: −93.3% vs. OsiBev: −96.6% at day 24, P=0.16, t-test). However, the combination therapy significantly delayed tumor re-growth compared to osimertinib monotherapy. The regression rate of the tumor treated with OsiBev was significantly higher than that of the tumor treated with osimertinib monotherapy at day 56 (osimertinib: 57.3% vs. OsiBev: −64.0%, P=0.01, t-test) (maximum tumor diameter at day 56; osimertinib: 10.1 mm ± 1.9 vs. OsiBev: 3.5 mm ± 1.5, mean ± SE).

In the large model, no significant difference in the maximum tumor regression rate was observed between the osimertinib monotherapy and OsiBev (osimertinib: −84.4% vs. OsiBev: −83.6% on day 24, P=0.86, t-test; Fig. 3B). Moreover, combination therapy with OsiBev did not delay tumor re-growth compared to osimertinib monotherapy (tumor regression rate at day 56, osimertinib: −7.9% vs. OsiBev: −0.9%, P=0.87, t-test) (maximum tumor diameter at day 56; osimertinib: 9.6 mm ± 0.77 vs. OsiBev: 10.3 mm ± 1.8, mean ± SE) unlike in the conventional model (Fig. 3B).

To evaluate the effects of bevacizumab, we measured the expression of the vascular endothelial marker, CD31. As expected, there were significantly fewer CD31-positive cells in tumors treated with the combination therapy than osimertinib monotherapy in both the conventional and large models (Fig. 3C). Although HIF-1α expression was significantly lower and TGF-α expression did not change in the conventional model treated with the combination therapy (Fig. 3D and E), neither HIF-1α nor TGF-α expression decreased significantly in tumors from the large model treated with the combination therapy (Fig. 3D and E). Therefore, OsiBev only achieved a limited decline in HIF-1α expression and relatively high TGF-α expression was maintained in tumors from the large model compared to the conventional model.

Having observed relatively high TGF-α expression in tumors from the large model, we decided to assess the effect of TGF-α on the inhibitory function of osimertinib in vitro. As expected, adding TGF-α (100 ng/ml) to the culture medium significantly reduced the inhibitory effect of osimertinib on RPC-9 or H1975 cell viability (Fig. 4A and D). Moreover, western blotting revealed that TGF-α activated the phosphorylation of the EGFR downstream signaling protein ERK and increased HIF-1α expression in RPC-9 and H1975 cells (Fig. 4B, C, E and F). ERK phosphorylation and HIF-1α protein expression were decreased in cells treated with osimertinib and not TGF-α, but were partially restored in RPC-9 or H1975 cells treated with osimertinib and TGF-α.

We also assessed the effect of bevacizumab in vitro, finding that bevacizumab alone had little inhibitory effect on cell viability and its combination with osimertinib did not restore sensitivity to osimertinib in RPC-9 or H1975 cells incubated with TGF-α in vitro (Fig. 4A and D). Cetuximab, which inhibits EGFR activation by blocking ligand binding, had little inhibitory effect on cell viability and partially inhibited ERK phosphorylation and HIF-1α protein expression in cells treated without TGF-α. Interestingly, the combination of osimertinib plus cetuximab (OsiCet) exerted a similar inhibitory effect on cell vaibility to osimertinib monotherapy in RPC-9 or H1975 cells without TGF-α stimulation, but exhibited a superior inhibitory effect after TGF-α stimulation (Fig. 4A and D). Consistent with this, cetuximab alone had little effect but the combination of osimertinib and cetuximab showed a tendency of superior inhibitory effect on cell viability in RPC-9 cells pre-incubated under hypoxic conditions (Fig. S2). Furthermore, ERK phosphorylation and HIF-1α were lower in RPC-9 or H1975 cells treated with OsiCet and TGF-α compared to osimertinib alone and TGF-α (Fig. 4B, C, E and F). Taken together, these findings suggest that TGF-α plays an important role in mediating the effect of osimertinib, and cetuximab could restore cellular sensitivity to osimertinib.

Finally, we confirmed the effect of cetuximab in the large model. Although cetuximab monotherapy exhibited a modest effect similar to bevacizumab (Fig. S3A), OsiCet tended to exert a superior inhibitory effect on tumor growth; however, this effect was not significantly higher than for osimertinib monotherapy (Fig. S3B).

Therefore, we examined the effect of triple therapy with osimertinib, bevacizumab, and cetuximab in vivo. Although triple therapy did not inhibit tumor growth more than OsiBev in the conventional model (maximum tumor diameter at day 56; osimertinib: 8.4 mm ± 1.7 vs. OsiBev: 3.4 mm ± 1.7 vs. triplet: 3.7 mm ± 1.3, mean ± SE) (Fig. 5A), it exerted a greater tumor inhibitory effect than osimertinib monotherapy, OsiBev, or OsiCet in the large model (maximum tumor diameter at day 56; osimertinib: 11.2 mm ± 1.9 vs. OsiBev: 11.4 mm ± 0.49 vs. triplet: 4.7 mm ± 1.2, mean ± SE) (Figs. 5B and S3B). Importantly, no decrease in body weight was observed in any of the mice (Fig. 5A and B). Pathological assessment of the tumors revealed that the triple therapy tended to reduce CD31, HIF-1α, and TGF-α expression compared to the double therapies (Fig. 5C-E).

Together, the findings of this study suggest that osimertinib or its combination with bevacizumab exerted limited efficacy in lung cancer with EGFR T790M mutation and HIF-1α/TGF-α expression; however, triple therapy with osimertinib, bevacizumab, and cetuximab induces an even greater inhibitory effect (Fig. S4).