Differential Efficacy of Bevacizumab and Erlotinib in Preclinical Models of Renal Medullary Carcinoma and Fumarate Hydratase-Deficient Renal Cell Carcinoma

Abstract

Renal medullary carcinoma (RMC) and fumarate hydratase (FH)-deficient renal cell carcinoma (RCC) are rare and highly aggressive cancers. Although the combination of vascular endothelial growth factor (VEGF) inhibition by bevacizumab and epidermal growth factor receptor (EGFR) inhibition by erlotinib is clinically used for both diseases, the differential effect of each component has not been investigated. Transcriptomic profiling revealed that RMC and FH-deficient tumor tissues demonstrate increased EGFR but not VEGF expression compared with adjacent normal kidney. Subsequent in vitro studies revealed that RMC and FH-deficient cell lines are sensitive to erlotinib treatment, whereas clear cell RCC cell lines are resistant. We developed patient-derived xenograft (PDX) models of tumors exposed to first-line therapies to represent treatment-experienced RMC and FH-deficient RCC models. These models were then used to determine tumor growth response to angiogenesis inhibition by bevacizumab alone or in combination with erlotinib. The FH-deficient RCC PDX model responded to either bevacizumab or erlotinib alone or in combination while the RMC PDX responded only to erlotinib consistent with clinical and preclinical data suggesting that RMC is refractory to angiogenesis inhibition. Statistically higher expression of EGFR was observed in the RMC PDX model compared to the FH-deficient model; while higher phosphorylated tyrosine-416 SRC expression was observed in FH-deficient PDX model compared to the RMC model. Our preclinical data suggest that EGFR signaling differentially modulates tumor growth in RMC and FH-deficient RCC and that angiogenesis inhibition is a valid target in FH-deficient RCC but not RMC.

Figure 1.. Pathways analyses of patient tissue bulk RNA-sequencing.

A, Oncogenic signatures for EGFR pathway in RMC versus adjacent normal kidney; B, EGFR pathway in FH-deficient RCC versus adjacent normal kidney; C, VEGF-VEGFR pathway in RMC versus adjacent normal kidney; D, and VEGF-VEGFR pathway in FH-deficient RCC versus adjacent normal kidney.

Figure 2.. In vitro effect of erlotinib.

A, Percent of viable cells by MTT after 5 days of treatment with 5 μM of erlotinib. Results for ccRCC cell lines are shown in black, RMC cell lines are in red, and FH-deficient RCC cell lines are in gray. B, The IC₅₀ of RMC2C1, UOK353, UOK360, and UOK268 determined using CellTiter-Glo^® with 25 μM, 15 μM, 10 μM, 8 μM, 5 μM, 2.5 μM, 1 μM, 0.5 μM, 0.25 μM, 0.125 μM, and 0.05 μM of erlotinib. C, Clonogenic assays with 2 μM of erlotinib or 0.01% DMSO (vehicle). Statistical significance was determined using Mann-Whitney U test (****, p < 0.0001; ***, p < 0.001). D, Representative images are shown of immunoblot of 786-O and RMC cell lines treated with 0.01% DMSO (1) or 2 μM erlotinib (2) for 2 hours prior to harvesting; and immunoblot of 786-O, UOK268, and UOK262 cells treated with 0.01% DMSO (1) or 2 μM erlotinib (2) for 30 minutes. E, Quantitation of normalized p-EGFR/EGFR band with and without erlotinib for the six cell lines are shown in the scatter plots. F, Quantitation of pAKT/AKT and pSRC/SRC of all samples. Statistical significance was determined using one-way ANOVA with p-values adjusted for multiple comparisons (****, p < 0.0001; ***, p < 0.001; **, p < 0.005, *, p < 0.05, ns, not statistically significant). All results were obtained from at least three independent biological replicates.

Figure 3.

Histological evaluation of PDX models. Images of H&E, INI-1, and S-2-succino-cysteine (2-SC) staining of PDX tissue from RMC32X and FH-RCC1X models. The box in the left corner of each microphotograph represents the scale bar (300 μm).

Figure 4.

In vivo therapeutic experiments. Days of treatment are on the x-axis while mean tumor size in mm³ is on y-axis. Error bars represent standard error of the mean. Statistical significance was determined using 2-way ANOVA with multiple comparisons. A, Tumor growth curves for RMC32X PDX model treated with vehicle (V), bevacizumab (B), erlotinib (E), and erlotinib plus bevacizumab (E+B). For RMC32X on day 24 of treatment, V vs. E and V vs. E + B the p-values were found to be < 0.0001. B, Tumor growth curve for FH-RCC1X PDX model. For FH-RCC1X on day 30 of treatment, the p-values for V vs. B and V vs. E + B were found to be p = 0.0015 and p = 0.013 respectively.

Figure 5

Protein quantitation between models and treatment cohorts. A, Microphotograph images of RMC32X and FH-RCC1X tumor tissues stained with hematoxylin and eosin (H&E) and EGFR. Scale bar in lower left corner of all images is 300 μm. Representative images for vehicle (V), erlotinib (E), bevacizumab (B), or erlotinib + bevacizumab (B+E) treatment shown. Percent EGFR positive cells shown in graph and statistical significance was determined by unpaired 2-tailed t-test (**, P < 0.005). B, Immunoblots determining the differences in phosphorylated Y1069 EGFR, total EGFR expression, phosphorylated Y416 SRC, total expression of SRC, phosphorylated S473 AKT, and total expression of AKT in RMC32X and FH-RCC1X models. (V) Vehicle cohorts are labeled in black, (B) bevacizumab cohorts are labeled in blue, (E) erlotinib cohorts are labeled in red, and (E +B) erlotinib plus bevacizumab treatment cohorts are labeled in green. Each black box is separate gel with its corresponding loading control shown. C, Densitometry measurements for total EGFR, SRC, and AKT over the loading control for each model and the ratio of pEGFR(Y1069)/EGFR, pSRC(Y416)/SRC, and pAKT(Ser473)/AKT are shown. For visualization purposes, phosphorylated protein over total protein ratios were multiplied by 100. One-way ANOVA with p-values adjusted for multiple comparisons (****, p < 0.0001). Each well is a separate tumor. D, Illustration of the differences we observe in phosphorylated and total protein between RMC32X and FH-RCC1X models. This illustration was created using the Smart Servier Medical Art library (https://smart.servier.com/), which is licensed under a Creative Commons Attribution 3.0 Unported License.