Systems Pharmacology Modeling Identifies a Novel Treatment Strategy for Bortezomib-Induced Neuropathic Pain

Abstract

Chemotherapy-induced peripheral neurotoxicity is a common dose-limiting side effect of several cancer chemotherapeutic agents, and no effective therapies exist. Here we constructed a systems pharmacology model of intracellular signaling in peripheral neurons to identify novel drug targets for preventing peripheral neuropathy associated with proteasome inhibitors. Model predictions suggested the combinatorial inhibition of TNFα, NMDA receptors, and reactive oxygen species should prevent proteasome inhibitor-induced neuronal apoptosis. Dexanabinol, an inhibitor of all three targets, partially restored bortezomib-induced reduction of proximal action potential amplitude and distal nerve conduction velocity in vitro and prevented bortezomib-induced mechanical allodynia and thermal hyperalgesia in rats, including a partial recovery of intraepidermal nerve fiber density. Dexanabinol failed to restore bortezomib-induced decreases in electrophysiological endpoints in rats, and it did not compromise bortezomib anti-cancer effects in U266 multiple myeloma cells and a murine xenograft model. Owing to its favorable safety profile in humans and preclinical efficacy, dexanabinol might represent a treatment option for bortezomib-induced neuropathic pain.

Keywords: bortezomib; dexanabinol; multiple myeloma; peripheral neuropathy; pharmacodynamics; systems pharmacology.

FIGURE 1

Network simulations to identify combinatorial treatment strategies for BIPN. (A) Network model simulations of select intracellular components in the presence of proteasome inhibition. The Boolean network model was converted to normalized HillCube differential equations using Odefy, and default parameter values were used (tau = 1; k = 0.5; n = 3). Simulations were performed for 25-time steps (arbitrary units). Results from attractor analysis for the relative activation frequency of (B) network nodes and (C) neuronal apoptosis across identified attractors for eight network perturbations: proteasome inhibition (bortezomib) in combination with a TNFα inhibitor, NMDA receptor antagonist, and/or ROS inhibitor. Monotherapies, two-target combinations, and the three-target combinations are highlighted in red, blue, and purple.

FIGURE 2

Assessment of in vitro cytotoxicity of bortezomib and dexanabinol. Cell viability was measured using WST-1. (A) Cytotoxicity of SH-SY5Y cells exposed to various concentrations of bortezomib in the absence (red) or presence of 1 μM (green) and 10 μM (blue) dexanabinol. (B) Cytotoxicity of U266 multiple myeloma cells exposed to a range of bortezomib and dexanabinol concentrations for (left) 24, (middle) 48, and (right) 72 h. Antagonistic, additive, and synergistic pharmacological relationships are colored in red, green, and blue. (C) Concentration-effect relationship of dexanabinol on LPS-induced TNFα production in primary rat macrophages.

FIGURE 3

Assessment of dexanabinol neuroprotective effects using nerve-on-a-chip. The effect of dexanabinol on bortezomib-induced decreases in (A) proximal action potential amplitude (APA), (B) distal action potential amplitude, (C) proximal nerve conduction velocity, and (D) distal nerve conduction velocity. Action potential amplitudes and nerve conduction velocity were normalized to treatment naïve control (black). Dexanabinol (10 and 25 μM), bortezomib (100 nM), and the combination of both drugs are displayed as blue, red, and purple bars. GraphPad prism v7.04 was used to perform a one-way ANOVA with Tukey’s correction for multiple comparisons across all groups. Only significant comparisons to control are displayed. p values are reported as: ns (not significant), * (<0.05), ** (<0.01), and *** (<0.001).

FIGURE 4

Nerve conduction studies of bortezomib and dexanabinol treatment in rats. Neurophysiological values were determined at baseline, 4 weeks of treatment, and 8 weeks of treatment. Top panels show neurophysiological changes in (A) proximal caudal SAP amplitude and (B) distal caudal SAP amplitude. Bottom panels show neurophysiological changes in (C) proximal caudal sensory NCV and (D) distal caudal sensory NCV. Bortezomib (Bort) was administered IV (0.2 mg/kg) three times a week, and dexanabinol (Dex) was administered IP (10 mg/kg) three times a week approximately 30 min prior to bortezomib. GraphPad prism v7.04 was used to perform a Kruskal-Wallis test with Dunn’s correction for multiple comparisons across all groups. Only significant comparisons to vehicle (Vehicle) are displayed. p values are reported as: * (<0.05), ** (<0.01), *** (<0.001).

FIGURE 5

Efficacy of dexanabinol for preventing bortezomib-induced mechanical allodynia and thermal hyperalgesia in rats. Mechanical threshold (grams) was determined using dynamic test at baseline, 4 weeks of treatment, and 8 weeks of treatment (A). The withdrawal latency to an infrared heat stimulus (seconds) was determined using a dynamic plantar analgesiometer at baseline, 4 weeks of treatment, and 8 weeks of treatment (B). Bortezomib (Bort) was administered IV (0.2 mg/kg) three times a week, and dexanabinol (Dex) was administered IP (10 mg/kg) three times a week approximately 30 min prior to bortezomib. GraphPad prism v7.04 was used to perform a Kruskal-Wallis test with Dunn’s correction for multiple comparisons across all groups. p values are reported as: Bort vs vehicle (Vehicle), ** (<0.01), *** (<0.001) and Bort vs Dex + Bort, ○ (<0.05), ○ ○ (<0.01).

FIGURE 6

Efficacy of dexanabinol on bortezomib-induced decreases in intraepidermal nerve fiber (IENF) density. IENF density was measured at the end of the 8 weeks of drug treatment by an immunochemistry analysis using PGP 9.5. GraphPad prism v7.04 was used to perform a Kruskal-Wallis test with Dunn’s correction for multiple comparisons across all groups. Only significant comparisons to vehicle are displayed. p values are reported as: * (<0.05), *** (<0.001).

FIGURE 7

Pharmacokinetics and pharmacodynamics of dexanabinol and bortezomib in a MM1.S multiple myeloma mouse model. Model predicted mouse plasma concentrations of (A) 1 mg/kg of bortezomib administered on Days 1, 5, 8, 12, 15, and 19 and (B) 10 mg/kg of dexanabinol administered on Days 1, 5, 8, 12, and 15. (C) Pharmacodynamics of bortezomib and dexanabinol on MM1.S tumor volume in SCID mice. Tumors were grown to approximately 100 mm³ prior to drug administration. Mice (n = 20) were divided into four treatment groups: treatment naïve (black), dexanabinol (blue), bortezomib (red), and their combination (purple). Bortezomib and dexanabinol were administered IP twice weekly (1 and 10 mg/kg). Tumor volume measurements and pharmacodynamic model fitted profiles are displayed as solid markers and dashed lines. Asterisks indicate statistical significance from control (corrected p value < 0.5), which were obtained via multiple t-tests with Holm-Sidak Bonferroni correction (GraphPad Prism v7.04).