Combined MEK/MDM2 inhibition demonstrates antitumor efficacy in TP53 wild-type thyroid and colorectal cancers with MAPK alterations

Most tumors with activating MAPK (mitogen-activated protein kinase) pathway alterations respond poorly to MEK inhibitors alone. Here, we evaluated combination therapy with MEK inhibitor selumetinib and MDM2 inhibitor KRT-232 in TP53 wild-type and MAPK altered colon and thyroid cancer models. In vitro, we showed synergy between selumetinib and KRT-232 on cell proliferation and colony formation assays. Immunoblotting confirmed p53 upregulation and MEK pathway inhibition. The combination was tested in vivo in seven patient-derived xenograft (PDX) models (five colorectal carcinoma and two papillary thyroid carcinoma models) with different KRAS, BRAF, and NRAS mutations. Combination therapy significantly prolonged event-free survival compared with monotherapy in six of seven models tested. Reverse-phase protein arrays and immunohistochemistry, respectively, demonstrated upregulation of the p53 pathway and in two models cleaved caspase 3 with combination therapy. In summary, combined inhibition of MEK and MDM2 upregulated p53 expression, inhibited MAPK signaling and demonstrated greater antitumor efficacy than single drug therapy in both in vitro and in vivo settings. These findings support further clinical testing of the MEK/MDM2 inhibitor combination in tumors of epithelial origin with MAPK pathway alterations.

To determine the combined growth inhibitory effect of selumetinib (MEK inhibitor) and KRT-232 (MDM2 inhibitor), we performed sulforhodamine B (SRB) assay in KRAS-mutated CRC cell lines HCT116 (KRAS G13D), LoVo (KRAS G13D), and B1011CL (KRAS A148T) and BRAF-mutated PTC cell line MDA-PDX.008CL (PDX.008CL, BRAF K601N). B1011CL and PDX.008CL cell lines were derived from B1011 and PDX.008 PDX models, respectively. SRB binds to proteins, and the amount of bound dye is used to determine the cell density and measure cell proliferation. Single-agent activity of selumetinib and KRT-232 demonstrated half maximal inhibitory concentration (IC 50 ) values < 1 µM in most cell lines, which is below clinically achievable plasma concentrations in patients 34,35 , leading to significant growth inhibition (Fig. 1B). A wide range of IC 50 values for MDM2 inhibitors in TP53 WT cell lines was previously reported 36 , similar to our observation with KRT-232 in B1011CL cells. All cell lines were then treated with both drugs for 72 h and with different dose ranges. www.nature.com/scientificreports/ combined with KRT-232 demonstrated synergistic activity across all four cell lines tested and revealed a lower IC 50 compared to single-agent treatment alone in all cell lines tested (Fig. 1C). Next, we assessed the colony forming capacity of these cell lines in presence of either single agent or the combination of selumetinib and KRT-232 by colony formation assay (CFA). We found that the combination of selumetinib and KRT-232 significantly affected anchorage-dependent growth and reduced colony formation capability in HCT116, LoVo, PDX.008CL, and B1011CL cells (Fig. 1D). Considering that long-term treatment might have a more profound effect on cells, we used CFA data and calculated IC 50 of KRT-232 and selumetinib (Fig. 1E). All cell lines were sensitive to both drugs. Then we looked at the combination effect. The combination was synergistic at all dose ratios in PDX.008CL, at some dose ratios in B1011CL and HCT116, and was additive in LoVo (Fig. 1F).
Selumetinib combined with KRT-232 affects in vitro cell cycle progression and demonstrates enhanced programmed cell death. To determine the effects of the combination of KRT-232 and selumetinib on the cell cycle, we used HCT116, B1011CL, and PDX.008CL cells. We observed a significant reduction of cells in S phase in PDX.008CL and B1011CL with combination treatment compared to vehicle (p = 0.001) ( Fig. 2A). With combination treatment, the percentage of B1011CL cells in S phase were increased compared to selumetinib (p = 0.013) and decreased compared to KRT-232 (p = 0.010) ( Fig. 2A). We also found an increase in B1011CL sub-G1 population when cells were treated with the combination regimen compared to vehicle (p = 0.042) and KRT-232 (p = 0.048), respectively. PDX.008CL demonstrated a higher proportion of cells arrested in G2/M in the combination arm compared to vehicle (p = 0.002) and selumetinib (p = 0.001). B1011CL had a higher proportion of cells arrested in G2/M in the combination arm compared to selumetinib (p = 0.039).
KRT-232 was previously observed to induce apoptosis in sarcoma cell lines 20 . After demonstrating synergistic activity between selumetinib and KRT-232, we questioned whether apoptosis was enhanced with this combination. HCT116 cells were treated for 72 h with vehicle, selumetinib, KRT-232 or a combination of both agents. Cells were collected, stained with annexin V and propidium iodide, and analyzed using flow cytometry. A significantly higher rate of apoptosis was present in cells treated with selumetinib combined with KRT-232 compared to vehicle (1.4% vs 10.1%; p = 0.04) (Fig. 2B,C). However, we observed no statistically significant difference between both single agent and the combination regimen.
Influence of selumetinib combined with KRT-232 on p53 pathway. We treated HCT116 cells with vehicle, KRT-232, selumetinib and their combination for 24 h. We lysed the cells and immunoblotted to observe the treatment effect on MDM2/p53 and Ras/Raf/MEK/ERK pathways (Fig. 2D). These signaling pathways interact with each other and both regulate apoptosis (Fig. 2E). MDM2/p53 pathway is involved in several cellular processes, such as cell cycle arrest at multiple points and senescence. Single agent KRT-232 treatment increased expression of MDM2 and p53. Likewise expression of p53 downstream proteins were regulated, such as an increase in p21 expression that regulates cell cycle, and an increase in Bax and PUMA expression that regulate apoptosis. We neither observed an increase in p-ERK expression nor MAPK activation as reported in other studies 31,32 . Selumetinib inhibited ERK1/2 phosphorylation, and DUSP4 and DUSP6 expression. There was an increase in expression of anti-apoptotic protein Mcl-1 and decrease in Mcl-1 downstream target, pro-apoptotic protein Bax, This was different from previous reports and where under similar conditions selumetinib had no effect on Mcl-1 expression 37,38 . Combination therapy had an additive effect and did not further enhance p53 activity. At this time point, we did not observe a significant increase in apoptosis either. Single agent selumetinib treatment resulted in a change in mobility of C-Raf and a significant increase in MEK1/2 phosphorylation. Inhibition of ERK1/2 phosphorylation released upstream proteins from negative feedback regulation 39 . Selumetinib combined with KRT-232 enhances in vivo antitumor efficacy in CRC PDX models compared with either single agent alone. To determine the in vivo antitumor efficacy of selumetinib combined with KRT-232, we evaluated five different CRC PDX models (C1035, B1011, C1185, C1114, and MDA-PDX.004 (PDX.004)) with different MAPK molecular characteristics (three KRAS, one BRAF, and one NRAS mutations). Figure 3 shows individual tumor volume (mm 3 ) and relative tumor volume (%) change over time. Kaplan-Meier curves demonstrated event-free survival (EFS), assessing time to tumor doubling (Fig. 3). Four of five models (B1011, C1035, C1114, and PDX.004) showed EFS benefit of treatment with the combination compared to selumetinib alone (Table 1). In four out of five models (B1011, C1035, C1114, and PDX.004), tumor growth inhibition with combination treatment was greater than both single drugs ( Table 2). C1185 model was different from the others and there was neither EFS benefit nor greater growth inhibition with combination treatment. Although the combination treatment inhibited tumor growth more than the control and other treatment groups in C1185 model, the variability in tumor sizes and particularly an outlier (the largest tumor was also treated by the combination) may have impacted the statistical analysis.

Selumetinib combined with KRT-232 induces regression in PTC PDX models.
To further investigate the in vivo effects of selumetinib plus KRT-232 across different histologies, we decided to utilize PTC PDX models. We tested two different PTC PDX models (MDA-PDX.008 (PDX.008) and PDX.020) both with presence of BRAF mutations (PDX.008 BRAF K601N and PDX.020 BRAF V600E). Significant antitumor efficacy was observed in both models. We observed complete regression of tumors in PDX.008 following 54 days of combination therapy. Dose interruption was introduced after 54 days of treatment, and mice were evaluated for regrowth of tumors. Treatments were reinitiated once tumors demonstrated 100% increase in tumor volume from baseline; this was observed 14 days after dose interruption. Despite tumor regrowth, tumors continued to respond to MDM2 combined with MEK inhibition upon rechallenge. Figure 4 shows individual tumor volume (mm 3   www.nature.com/scientificreports/ els showed EFS benefit with combination treatment compared with selumetinib or KRT-232 alone (Table 1). In both models, tumor growth inhibition with combination treatment was greater than both single drugs ( Table 2). Neither significant weight loss nor other signs of toxicity were observed during the in vivo experiments. Compared with the control, combination treatment increased cleaved caspase 3 expression in BRAF mutant PDX.004 and PDX.020 models ( Supplementary Fig. S1). Tumor samples were additionally evaluated for p53 and p21 expression by analyzing H-scores obtained from IHC. Similar to what we observed in vitro, we found both p53 and p21 to be upregulated, mainly in the KRT-232 single-agent (KRT-232 vs control, p21 p = 0.001 and p53 p = 0.002) and combination (combination vs control, p21 p = 0.004 and p53 p = 0.01) treatment groups (Fig. 5A). We also looked at Ki-67-positive cells as a measure of proliferation and found no difference among groups. Compared with the control, combination treatment decreased Ki-67 expression in KRAS mutant B1011 and C1035 models, however, even in these two models the percent of Ki-67-positive cells remained high (Supplementary Fig. S1).

Combination of MDM2 and MEK inhibitors induces apoptosis and decreases
To assess the impact of therapy on the functional proteomic profile, we performed RPPA on PDXs treated with selumetinib alone, KRT-232 alone, and the selumetinib/KRT-232 combination and untreated controls in four different models: CRC models B1011, C1035, and PDX.004 and PTC model PDX.020. Differentially expressed proteins (DEPs) are shown in Supplementary Table S1. Comparing selumetinib-treated PDXs with controls, we identified 10 DEPs, including downstream targets of MEK1/2: p-ERK1/2 Thr202/Tyr204 and dualspecificity phosphatase 6 (DUSP6) 40,41 . We corrected the random effect of models and generated a heatmap for the DEPs, in which the samples were clustered by treatment (Fig. 5B). When we compared KRT-232-treated PDXs with controls in all four models together, the only DEP was p21 (q = 0.015). The increase in p21 on RPPA was consistent with our finding of p21 upregulation in IHC data (Fig. 5A). When individual models were considered, difference in p21 expression was significant only in PDX.004 (q = 0.0013). Relative p21 expression in all models and treatment groups is shown in Fig. 5C. In PDX.004, p21 was significantly increased with KRT-232 (p < 0.001) and combination treatment (p < 0.001) but not with selumetinib treatment (p = 0.369). The difference between KRT-232 and combination treatment groups was not significant (p = 0.089). When we compared the PDXs treated with the combination of KRT-232 and selumetinib to controls, we identified 91 DEPs (Fig. 5D). Ingenuity Pathway Analysis identified p53 as the top upstream regulator (p = 1.96e−26).

Discussion
Although MAPK pathway alterations are among the most common genomic alterations, genomically informed therapeutic options are largely lacking. In this study, after inhibition of the MAPK pathway by selumetinib (MEK1/2 inhibitor) and stabilization of p53 pathway by KRT-232 (MDM2 inhibitor), we observed significantly increased growth inhibitory activity both in vitro and in vivo in CRC and PTC with MAPK alterations and TP53 WT. This confirmed our hypothesis that dual MEK and MDM2 inhibition will yield significant tumor growth inhibition in genomically matched colorectal and thyroid cancers. In PDX models of colorectal and thyroid cancers, we observed significant inhibitory effect of the combination of selumetinib and KRT-232 on tumor growth across a spectrum of MAPK genomic backgrounds, including BRAF, KRAS, and NRAS mutations, compared to single-agent treatment with either drug. BRAF or MEK inhibition alone in KRAS-and BRAF-mutant CRCs have shown limited activity 42 , likely because of MAPK pathway reactivation through adaptive responses 43 . However, Kopetz et al. showed that the combination of MEK with BRAF and EGFR inhibitors improved overall survival compared with doublet or single-agent treatment, highlighting the importance of developing combination therapies 44 . Our BRAF-mutated models demonstrated the highest level of sensitivity to combined therapy of selumetinib and KRT-232 and greatest tumor reduction, especially in our PDX.008 model (which harbors an activating BRAF K601N mutation), despite introduction of a hiatus and re-challenging with treatment. BRAF mutations in codon 601 are located in the kinase domain and have been shown to induce cell proliferation 45 . Recently it has been shown that KRT-232 inhibits tumor growth in BRAF WT melanoma PDXs, and KRT-232 in combination with BRAF or MEK inhibitors showed a synergistic effect in BRAF V600-mutant tumors 46 . A similar synergistic effect was observed with MEK inhibitor trametinib and three MDM2 inhibitors (nutlin-3, RG7388, and HDM201) in BRAF V600E-mutant and TP53 WT melanoma 47 . Thus, further study is needed to determine whether MEK/MDM2 inhibition can be an option for tumors that have progressed on BRAF inhibitors and whether a BRAF/MEK/ MDM2 inhibitor triplet would be an option to deepen and prolong responses in BRAF V600-mutant tumors.
MDM family members, MDM2 and MDM4, partner in their action, and both negatively regulate p53. In our study, we limited our experiments to models with TP53 WT. We thus cannot comment on expected results with this combination with TP53 mutant models. Stabilization of TP53 WT resolves DNA damage or induces apoptosis to protect cells from becoming tumorigenic. However, we cannot exclude the possibility that stabilization of mutant TP53 may have the opposite effect, promoting growth, metastatic properties, and genetic instability 48 . TP53 WT status is essential for KRT-232 to robustly exert its antitumor effects by blocking the interaction of p53 and MDM2. In nearly 8% of CRCs with TP53 WT, tumorigenesis is driven by MDM2 amplification and/ or overexpression 49 . Being mutually exclusive to TP53 mutations, MDM2 gene amplification has been shown to be associated with chromosome 12 gain 49 , resulting in both RNA and protein overexpression 50,51 , followed by downregulation of p53 and induction of chemoresistance 14 . Amplification is a major cause of MDM2 overexpression; however, other mechanisms have been shown to induce MDM2 overexpression, including SNP309 germline mutation 52 and MDM2 promoter binding by estrogen in hormone-positive breast cancers 53 . However, these tumors did overexpress MDM2 to the level of MDM2-amplified tumors 19 . Our findings, including immunoblotting of the p53 pathway, demonstrated a strong p53 reactivation in both single-agent KRT-232 and combination treatment, which induced apoptosis. Cell cycle analysis demonstrated variability in response, with decreases in S-phase and increases in G2/M cell populations in both KRT-232 and combination treatment groups in two of three cell lines, and there was no statistical significance between them. These findings, combined with upregulation of p53 and p21, suggest the effect on cell cycle is mainly driven by p21 through reactivation of p53 by KRT-232 and/or combination therapy. The increase in G2/M cell populations observed is most likely driven by KRT-232 in both single-agent and combination treatment arms. A similar finding was observed with MDM2 inhibitors and was found to be due to double-stranded DNA breaks' inducing a DNA damage response and blocking the entry of cells into mitosis 54 . Longer treatment and combining the compounds based on their potencies could reveal significant results in cell cycle analysis. In addition to the combination, both single agents induced apoptosis in vitro. KRT-232 increases p53 and p21 expression, which is followed by cell cycle arrest in G1 until apoptosis is induced 55 . MEK inhibitors, including selumetinib, inhibit ERK, and ERK inhibition induces the p53 upregulated modulator of apoptosis (PUMA) 56 , resulting in p53-independent apoptosis. In vivo data supported this observation. The expression of p53 and p21 were increased in the KRT-232 and combination treatment groups, yet cleaved caspase 3 was increased in all treatment groups, including the selumetinib group.
The RPPA result also produced interesting observations. The DEPs in the combination treatment vs control groups were involved in different processes, such as induction of apoptosis (Bim, Bax, IRF-1/3 upregulation, PARP1 downregulation), cell cycle arrest (p21 and p-Chk1 upregulation), cellular stress (XBP1 upregulation), fatty acid synthesis (FAS downregulation), T-cell signaling (Zap70 and Vav1 upregulation), and resistance to BRAF inhibitors (EPHA2 downregulation). Several mechanisms could be involved in drug response, which warrant further study to understand their significance. There may also be differences in the downstream effects of MEK and MDM2 inhibition including differential effects on proliferation and apoptosis. Further work is needed to determine whether downstream effects differ by genotype, histology and co-alterations and to gain deeper mechanistic insights into combination therapy efficacy observed.
Major mechanisms of resistance to MDM2 inhibitors include TP53 mutations or activation of IGFBP1-ERK signaling 31,57 . MEK inhibition resistance is mediated through activation of the PI3K pathway or FGFR signaling 58,59 . Addition of a MEK inhibitor to a MDM2 inhibitor overcame MAPK-mediated resistance to MDM2 inhibition 31 . Acquired resistance to combined MDM2 and MEK inhibition has been previously observed in colon and non-small cell lung cancer cells 60 . In our study, we did not do analysis of acquired resistance.
Limitations of our study include absence of genomic analysis of treated PDXs to determine if acquired TP53 mutations were detected; however, the tumors remained sensitive to re-challenge with combination therapy Scientific Reports | (2022) 12:1248 | https://doi.org/10.1038/s41598-022-05193-z www.nature.com/scientificreports/ after re-growth in PDX.008, suggesting that the residual cancer cells were still TP53 WT. As these combinations advance to the clinic, further studies will be needed to identify acquired resistance mechanisms. Our preclinical findings demonstrate synergistic and enhanced antitumor efficacy of the combination of MDM2 and MEK inhibition in MAPK-altered, TP53 WT tumors of colorectal and thyroid origin. These results support further investigation of the MDM2/MEK combination in tumors of epithelial origin in the clinic.

Methods
Cell lines and cultures. HCT116 and LoVo CRC cell lines were acquired from American Tissue Culture Collection (Manassas, VA, USA). MDA-PDX.008CL PTC and B1011CL CRC cell lines were generated from MDA-PDX.008 PTC and B1011 CRC PDXs, respectively. Samples were collected with informed consent under an Institutional Review Board-approved protocol for PDX generation, as described below. PDXs were collected in DMEM/F12 supplemented with 10% fetal bovine serum, chopped into 1-2 mm pieces, and incubated overnight on a shaker at 37 °C and 200 rpm. Cells were filtered through a 100 µm nylon mesh and plated in T25 ultra-low attachment flasks. Three months later, cells were further transferred into cell culture plates. Fidelity confirmation of PDX-derived cell lines was completed using short tandem repeat analysis. HCT116 cells were cultured in RPMI-1640 supplemented with 10% fetal bovine serum, and LoVo cells were cultured in F12K medium (Kaighn's Modification of Ham's F-12 Medium) with 10% fetal bovine serum at 37 °C and humidified in 5% CO 2 .
Reagents. KRT-232 (Kartos Therapeutics, Redwood City, CA, USA) and selumetinib (AstraZeneca, Cambridge, UK) were obtained through National Cancer Institute Cancer Therapy Evaluation Program. KRT-232 was prepared in 0.5% methylcellulose, 1% Tween 80 and water, while selumetinib was prepared in 0.5% methylcellulose and water for in vivo studies. Dimethyl sulfoxide was obtained from Sigma-Aldrich (St. Louis, MO, USA). All drugs used for in vitro studies were prepared in dimethyl sulfoxide.
Cell growth assays. SRB assay was used to determine the anti-proliferative activity and IC 50 61 . Cells were seeded in 96-well plates at a density of 5000 cells per well based on cell line growth characteristics. The IC 50 values were determined based on a dose-response curve generated using CalcuSyn Software, v2.11 (Biosoft, Cambridge, UK). The drugs were combined with a fixed ratio in all cell lines, whereas the selumetinib-to-KRT-232 ratios in each model were: PDX.008CL, 5:1; HCT116, 6:1; B1011Cl, 0.8:1; and LoVo, 0.5:1. Chou-Talalay methodology was used in order to obtain Combination Index (CI) values and, where CI < 1 indicates synergistic activity, CI = 1 indicates additive activity and CI values > 1 indicates antagonistic activity. Experiments were performed at least three times.
Colony formation assay. Live cells were counted using trypan blue and were seeded at a density of 1 × 10 3 for HCT116, LoVo, and PDX.008CL and 5 × 10 3 for B1011CL in each well on 60-mm plates in duplicate for each treatment group. Cells were treated the following day. Culture medium and drugs were changed twice a week. Cells were cultured for 10 days to three weeks. The colonies were fixed in 10% formalin and stained with 0.05% crystal violet in 25% methanol. Total colony area were measured using ImageJ v1.52a software (NIH, Bethesda, MD, USA). Individual IC 50 values were calculated using CalcuSyn. The Bliss independence model was used to evaluate the drug interactions across all dose combinations tested 62 3 . The mice were euthanized at the end point of the study (at 28 days or when tumor diameter reached 1.5 cm). Tumor volume was obtained by using: tumor volume (mm 3 ) = (width × width × length)/2. Tumor volumes and weights were measured twice a week.
In order to determine in vivo antitumor efficacy of selumetinib in combination with KRT-232, mice were randomized into four groups. C1035 xenografts were treated with vehicle, KRT-232 15 mg/kg, selumetinib 15 mg/kg, and combination of both (n = 5). The other six xenografts models were treated with vehicle, KRT-232 15 mg/kg, selumetinib 25 mg/kg, and combination of both (n = 5). Both selumetinib and KRT-232 were dosed daily and administered through oral route.
Sample evaluation. Tumors were harvested 4 h after last treatment. Half of the tumor was placed in formalin and the other half was snap frozen. Hematoxylin and eosin stains from formalin-fixed and paraffin-embedded PDX samples were reviewed by a board-certified pathologist for the presence of tumor cells. All but one sample had sufficient (> 100 tumor cells) tumor for immunohistochemistry (IHC) evaluation. The percentage of necrosis within the tumor was estimated on the hematoxylin and eosin stain.
Reverse-phase protein array. Reverse-phase protein array (RPPA) was performed at MDACC Functional Proteomics core facility as previously described 63 . DEPs and their association with canonical pathways were analyzed using Ingenuity Pathway Analysis software (Qiagen, Hilden, Germany).
Immunohistochemistry. From each sample, 4-µm sections were stained with antibodies against Ki-67 (clone MIB-1, code M7240, Dako/Agilent, Glostrup, Denmark), cleaved PARP (Asp214; clone: D64E10, #5625), cleaved caspase 3 (Asp175; #9661, Cell Signaling Technology), p21 (clone: F5, #sc-6246, Santa Cruz, Dallas, USA), and p53 (clone: DO-7, Leica Biosystems, Wetzlar, Germany). For Ki-67, cleaved PARP, and cleaved caspase 3, the percentage of positive tumor cells was estimated regardless of staining intensity. For p21 and p53, the percentage of positive cells as well as the intensity of staining was evaluated, resulting in an H-score (range: 0-300). Necrotic areas were excluded from evaluation. Bioinformatics analysis. The RPPA data were normalized for protein loading and transformed to log2 values for quality control and comparison analysis. The comparisons were done by fitting the linear mixed effect models using treatment/control as the fixed effect and model as the random effect. To control the false discovery rate, we calculated the q-value for each protein in each comparison and used the q-values to identify the DEPs.

Statistics.
For in vitro studies, two-group comparisons were done using the Student's t-test and p < 0.05 was considered statistically significant. The normalized log-transformed RPPA expression values were used for the analysis. Boxplot, unsupervised hierarchical clustering and Principal Component Analysis were used for the quality assessment. The comparisons were done by fitting linear mixed effect model (LMEMs) using "Group" as the fixed effect and "Model" as the random effect. A beta-uniform mixture (BUM) model was used to fit the p-values from LMEMs and identified the differentially expressed proteins (DEPs) with specified FDR.
For in vivo studies, the % tumor volume change per time point was calculated as a relative level of tumor growth change from baseline: δ t = V t −V 0 V 0 × 100 , where V t is the tumor volume a time t and V 0 is the tumor volume at baseline. An event in each animal was defined as a doubling of tumor volume from initial tumor volume. EFS was defined as the time interval from initiation of study to the first event or to the end of the study period for tumors that did not doubling in volume. The time to event was determined using linear interpolation based on the following formula: for animals for which there was no tumor volume measurement at time t but which had flanking volume measurements at time t 0 and t 1 such that t 0 < t < t 1 , we used linear interpolation to compute the measurement. That is, we computed � V t = � V 0 + β(t − t 0 ) , where β = (� V t1 − � V t0 )/(t 1 − t 0 ) . Using the Kaplan-Meier survival estimates, an EFS T/C value was defined by the ratio of the median time to tumor doubling of the treatment group and the median time to tumor doubling of the control group. If the treatment group did not reach a doubling of tumor volume from initial tumor volume, then EFS T/C was defined as greater than the ratio on the last day of the study for the treatment group divided by the median time to tumor doubling for the control group. Tumor growth curves were analyzed using a linear mixed model. A mixed effects model was fit, with treatment, time and their interaction as fixed-effects parameters, and the time trend for each subject as random effects 64 . The log-rank test was used to compare survival distribution of two treatment groups. All statistical analyses were performed using R software version 3.5.2. A p-value < 0.05 was regarded as significant.