3.1 The combination of pemetrexed and aumolertinib exhibits a sequence-dependent synergistic effect in EGFR-mutant NSCLC cell lines.
We initially determined the IC50 values of pemetrexed or aumolertinib on different NSCLC cell lines including EGFR-wide type A549, EGFR-del19 HCC827 and EGFR L858R/T790M H1975 (Fig.S1). And then three different combination strategies were designed (Fig.1A). The integration index of each combination strategy was measured with various concentrations of pemetrexed and aumolertinib at the ratios of their natural IC50 values, ranging from 0.25 times IC50 concentration to 4 times IC50 concentration. As shown in Fig.1B, P-A was a superior strategy in H1975 and HCC827, with significantly stronger inhibitory effects on survival rates than single drug treatment and also other combination strategies. Furthermore, the CI value showed that only the anti-proliferative effect of the strategy P-A resulted in synergy (CI < 0.75) at all concentration points in both H1975 and HCC827 (Fig.1C). In comparison, strategy P+A just generated additive effect (0.75<CI<1.45) at all concentration points in H1975 as well as high concentration in HCC827, and even antagonistic effect (CI>1.45) at low concentration points in HCC827. The most inferior strategy A-P, generated antagonistic effect at all concentration points in H1975 and high concentration in HCC827 with only additive effect at low concentration points in HCC827. However, different from EGFR-mutant cell lines, all combination strategies exhibited antagonistic effect at all concentration points in EGFR wide type cell line A549. Next, this sequence-dependent synergistic effect of proliferative inhibition upon EGFR-mutant NSCLC cell lines were further validated by colony formation analysis (Fig.1D). In addition, apoptosis plays a crucial role in the response of cancer to EGFR TKIs7 and also pemetrexed27. Our results showed that pro-apoptosis effect markedly enhanced post P-A sequence treatment, revealed by significantly increased expression of cleaved Caspase 3 and cleaved PARP1 (Fig.1E-1F).
Similar to the antiproliferative effect, P-A also exhibited superior inhibition on cell migration and invasion. The migration rate of H1975 and HCC827 after P-A treatment was only 17.24% and 12.39%, significantly lower than the single drug treatment groups. In comparison, P+A and A-P treatment failed to enhance the inhibition effect of aumolertinib on cell migration (Fig.2A). Transwell migration and invasion assay further validated the synergistic effect of pemetrexed and aumolertinib combination at P-A sequence (Fig.2B). Moreover, we examined several factors involved in EMT since the EMT is a crucial step for EGFR-induced tumor metastasis. As shown in Fig.2C, β-catenin, vimentin and snail were significantly reduced after drug treatments, among which, P-A exhibited the strongest inhibition.
3.2 The combination of pemetrexed and aumolertinib exhibits a sequence-dependent synergistic effect in EGFR-mutant NSCLC bearing mice.
Based on the difference of integration effect among various combination strategies in vitro, we hence inferred that P-A would exhibit superior anti-tumor effect over other combination strategies. This hypothesis was confirmed by a small-scale in vivo assay (Fig.3A). As shown in Fig.3B, the tumor volume of P-A group was significantly smaller than that of P+A treatment group and A-P treatment group. At the end of drug treatment, tumors were collected and photographed (Fig.3C). Among all drug treatment groups, the tumor burden of P-A was lightest, suggesting the superior anti-tumor effect of P-A.
Next, we applied a systemic experiment to determine the synergy of P-A sequence on both H1975 and HCC827 tumor bearing mice (Fig.3D). As shown in Fig.3E, the progress of H1975 tumor was slower with drug treatment, among which, P-A exhibited significant anti-tumor effect at the earliest and at the most extent. In comparison of monotherapy treatment and combination drug treatment, we found that the tumor volume of P-A group was significantly smaller than that of pemetrexed treatment group since day 14 and aumolertinib treatment group since day 18, indicating a synergistic effect (Fig.S2A). At the end of drug treatment, tumors were collected (Fig.3F) and tumor weights were assayed. Among all drug treatment groups, the tumor burden of P-A was lightest (Fig.3G). Furthermore, in agreement with in vitro assay, P-A exhibited stronger inhibition on tumor metastasis, proved by the obviously reduced expression of β-catenin, vimentin and snail (Fig.3H) in tumor and lower metastasis in liver (Fig.3I).
The synergistic effect of P-A sequence was further substantiated by the markedly increased antitumor effect when the treatment strategy was applied to another EGFR-mutant NSCLC bearing mice. In comparison with H1975, HCC827 was more sensitive to drug treatment, in accordance with the smaller IC50 assayed in vitro. As shown in Fig.3J and Fig.S2B, P-A exhibited a significant anti-tumor effect since day 4 post drug treatment, and both pemetrexed and aumolertinib treatment group showed anti-tumor effect after 6-days drug treatment. Despite the different drug sensitivities between HCC827 and H1975, P-A sequence treatment also exhibited superior anti-tumor effect on HCC827 tumor bearing mice, similar to the phenomenon observed in H1975 tumor bearing mice. The tumor volume of P-A group was significantly smaller than pemetrexed and aumolertinib monotherapy group since day 10. Furthermore, the tumor burden of P-A sequence group was lightest after 5-cycles drug treatment, with only 23.3 % volume and 23.4 % weight of the control, significantly smaller than both pemetrexed and aumolertinib monotherapy groups (Fig.3K and 3L).
During the whole period of drug treatment, the mice in all groups showed no obvious changes of food and water intake, and the body weight kept stable, indicating the safety of drug administration (Fig.S2C).
3.3 P-A sequence improves the suppression of EGFR signaling pathway.
The binding of EGF and dimerization of EGFR, a transmembrane glycoprotein, activates EGFR and the downstream PI3K-AKT and ERK signaling pathway, which can regulate cell survival, proliferation, anti-apoptosis, and metastasis (Fig.4A). Suppression of EGFR signaling pathway was the crucial mechanism for the anti-proliferation and anti-metastasis effect of EGFR TKIs. Herein, we firstly detected the activation of EGFR and its downstream signaling pathway upon H1975 and HCC827 to gain the insight of the molecular mechanism behind the synergistic effects during P-A sequence treatment. As shown in Fig.4B and 4C, we noted a significant inhibition effect on phosphorylating EGFR and concomitantly the reduced expression of p-ERK and p-AKT after aumolertinib or P-A sequence treatment, in contrast pemetrexed showed no effect on EGFR pathway. Interestingly, despite no direct effect of pemetrexed on EGFR signaling pathway, the levels of p-AKT and p-ERK expression in P-A sequence treatment group were much lower than that of aumolertinib monotherapy, which were validated in two EGFR-mutant cell lines. We also tested whether the enhanced suppression of EGFR signaling pathway post P-A sequence treatment can be found in tumor mass. Consistent with the results found in the cell lines, the expression of p-ERK and p-AKT in tumor mass after P-A sequence treatment was significantly lower than that in aumolertinib treatment group (Fig.4D). However, when sequence alternating, the enhanced suppression of EGFR signaling pathway disappeared after A-P sequence treatment (Fig.S3). Besides, in line with resembled anti-tumor effect, aumolertinib and osimertinib suppressed EGFR signaling pathway to similar magnitude.
3.4 P-A sequence improves the accumulation of pemetrexed in tumor via vascular normalization.
Since the latest report about an altered secretion of VEGF after EGFR activating mutation occurring in NSCLC cells28, we sought to investigate whether aumolertinib influences the secretion of VEGF from H1975 and HCC827. Fig.5A shows a significant reduction of VEGF secreted into the cell medium after aumolertinib treatment. Similarly, the amount of VEGF was also significantly reduced in the tumor mass after aumolertinib administration (Fig.5B). The distinct change of VEGF secretion following the suppression of EGFR signaling pathway in NSCLC cells suggested an indirect influence of EGFR TKIs on tumor vessels. Expectedly, the migration of endothelial cell HUVEC was significantly promoted after the co-culture with H1975, while the migration was inhibited when HUVECs were priorly exposed to aumolertinib for 24 hours. In addition, this inhibition could not be observed when HUVEC directly exposure to aumolertinib (Fig.5C). Based on the important role of VEGF on pathological angiogenesis in tumor mass and also the normalization of the abnormal structure and function of tumor vasculature during the treatment with antiangiogenic agents targeting VEGF/VEGFR229, we hence inferred that vascular normalization would occur in tumor mass after EGFR TKIs treatment. This hypothesis was confirmed in Fig.5D-5G with the significant decreases of VEGF and other pro-angiogenic factors (HIF-1α, TGFβ, ANG) as well as the obviously increases of anti-angiogenesis factors (sFLT, Angiostatin) after EGFR TKIs treatment including aumolertinib or osimertinib monotherapy, and P-A sequence combination therapy. Accompanied by the rebalance of pro-angiogenic and anti-angiogenesis factors, the normalization of the abnormal tumor vasculature structure was observed. Double staining of vascular endothelial cells marker CD31 (Red) and smooth muscle cells marker α-SMA (Green) was the indicator of vascular maturity. Fig.5G shows that originally thin, short and clutter tumor vessels in the control group were markedly prolonged and simultaneously exhibited higher co-location of CD31+ and α-SMA+ post EGFR TKIs treatment. Among three groups involving EGFR TKIs, P-A groups exerted superior effect on vascular normalization, in agreement with the strongest suppression of EGFR signaling pathway.
Next, we examined whether the tumoral vasculature normalization makes it more efficient for drug delivery. As shown in Fig.6A, P-A sequence treatment did not change the plasma exposure of both pemetrexed and aumolertinib, indicated by the identical drug concentration curve against time and similar kinetic parameters with monotherapy groups (Table.S2). However, the intratumoral concentration of pemetrexed after P-A sequence administration was remarkably higher than that in pemetrexed treatment group with 4.66-fold increase in H1975 tumors and 8.61-fold increase in HCC827 tumors (Fig.6B). In contrast, the accumulation of pemetrexed in other major organs showed no significant difference between P-A sequence and pemetrexed treatment group (Fig.S4). Besides, P-A exhibited negligible influences on the accumulation of aumolertinib in the tumor mass.
3.5 Superior anti-tumor effect of P-A sequence in patients harboring EGFR mutant
From April, 2020 to January, 2022, 50 patients had received aumolertinib monotherapy as first-line therapy and 15 patients received combination therapy (pemetrexed administered one week prior to aumolertinib). As shown in Table.1, no significance was observed in the age distribution, sex distribution and EGFR mutant type distribution in two groups. After two-cycle treatment, 14 patients (93.3%) in combination therapy group exhibited partial response revealed by more than 30% reduction in the longest diameter of target lesion. In comparison, only 64% patients exhibited partial response in aumolertinib monotherapy. Notably, no patients exhibited tumor progression after combination therapy group while 4 patients (8%) underwent target lesion increases by at least 20% or the appearance of new lesion after aumolertinib monotherapy. Overall, the objective response rate (ORR) and disease control rate (DCR) in combination therapy group are 93.3% and 100% respectively, obviously higher than 64% and 92% in aumolertinib monotherapy group.
For five representative cases who were received combination therapy (pemetrexed administered one week prior to aumolertinib) as neoadjuvant therapy, the significant tumor regressions were observed in patients harboring EGFR mutant after 2-3 cycle treatment (Table.2), independent of the different concomitant mutations. Pathological response was assessed by local pathologists, who measured the percentage of residual viable tumor in primary tumors resected from each patient during surgery. Surprisingly, four patients reached major pathological response changing from clinical stage III/IV to postoperative pathological stage I and one patient reached complete pathological response changing from clinical stage IIIB to postoperative pathological T0N0M0. Thereinto, two patients (Patient No.1 and No.2) occurred tumor metastasis and PET/CT was performed for further evaluation. As shown in Fig.7, both the primary tumor and the metastatic lesion regressed significantly. Meanwhile, the priorly high accumulation of 18F-FDG disappeared or significantly decreased.