Plasma and cerebrospinal fluid pharmacokinetics of select chemotherapeutic agents following intranasal delivery in a non-human primate model

Abstract

The blood–brain barrier (BBB) limits entry of most chemotherapeutic agents into the CNS, resulting in inadequate exposure within CNS tumor tissue. Intranasal administration is a proposed means of delivery that can bypass the BBB, potentially resulting in more effective chemotherapeutic exposure at the tumor site. The objective of this study was to evaluate the feasibility and pharmacokinetics (plasma and CSF) of intranasal delivery using select chemotherapeutic agents in a non-human primate (NHP) model. Three chemotherapeutic agents with known differences in CNS penetration were selected for intranasal administration in a NHP model to determine proof of principle of CNS delivery, assess tolerability and feasibility, and to evaluate whether certain drug characteristics were associated with increased CNS exposure. Intravenous (IV) temozolomide (TMZ), oral (PO) valproic acid, and PO perifosine were administered to adult male rhesus macaques. The animals received a single dose of each agent systemically and intranasally in separate experiments, with each animal acting as his own control. The dose of the agents administered systemically was the human equivalent of a clinically appropriate dose, while the intranasal dose was the maximum achievable dose based on the volume limitation of 1 mL. Multiple serial paired plasma and CSF samples were collected and quantified using a validated uHPLC/tandem mass spectrometry assay after each drug administration. Pharmacokinetic parameters were estimated using non-compartmental analysis. CSF penetration was calculated from the ratio of areas under the concentration–time curves for CSF and plasma ( AUC_CSF:plasma). Intranasal administration was feasible and tolerable for all agents with no significant toxicities observed. For TMZ, the degrees of CSF drug penetration after intranasal and IV administration were 36 (32–57) and 22 (20–41)%, respectively. Although maximum TMZ drug concentration in the CSF (C_max) was lower after intranasal delivery compared to IV administration due to the lower dose administered, clinically significant exposure was achieved in the CSF after intranasal administration with the lower doses. This was associated with lower systemic exposure, suggesting increased efficiency and potentially lower toxicities of TMZ after intranasal delivery. For valproic acid and perifosine, CSF penetration after intranasal delivery was similar to systemic administration. Although this study demonstrates feasibility and safety of intranasal drug administration, further agent-specific studies are necessary to optimize agent selection and dosing to achieve clinically-relevant CSF exposures.

Introduction

A major challenge to successful treatment of CNS tumors is overcoming the blood–brain barrier (BBB) in order to achieve adequate chemotherapeutic exposure at the tumor site. The BBB is essential to protect the CNS from toxins and maintain a homeostatic environment [1, 2]. The specialized tight junctions and active efflux transporters which form a neurovascular unit comprising the BBB limit entry of systemically administered chemotherapeutic agents, especially those that are large or hydrophilic, resulting in inadequate delivery of agents to the tumor site. While the BBB and blood:tumor barriers may be disrupted in some tumors, as evidenced by contrast enhancement on MRI, this disruption is incomplete and heterogeneous. Invasive cells at the leading edge of infiltrative tumors are frequently in areas where the BBB is intact. The BBB of some tumors such as diffuse intrinsic pontine glioma (DIPG) [3] and low grade glioma, are frequently intact, restricting drug delivery. Attempts to overcome this barrier have included systemic administration of higher chemotherapeutic doses, intra-arterial delivery, and BBB disruption, each of which are associated with significant toxicities [4, 5].

An alternate route of drug delivery to the CNS is intranasal administration. Intranasal delivery is non-invasive, may bypass the BBB, and has been successfully used for administration of drugs for other neurologic diseases [6–10]. Its utility for delivery of chemotherapeutic agents has yet to be clinically established. Intranasal drug delivery may overcome issues with systemic drug delivery including poor bioavailability, slow absorption, drug degradation, adverse gastrointestinal tract side effects, and first-pass hepatic metabolism [11, 12]. The exact mechanism of CNS delivery after intranasal drug administration is not completely understood, although animal studies have shown evidence of direct extracellular and intracellular transport pathways between the epithelial layer of the nasal cavity and the CNS. Drugs administered by the intranasal route can enter the CNS directly, as well as via transport along the olfactory and trigeminal nerve branches, predominantly by following the respective perivascular and neuronal channels [1, 10, 13–17].

Animal models have previously shown the potential for intranasal delivery of anti-tumor agents [18–22], such as the oligonucleotide telomerase inhibitor GRN163 [21, 22]. A limited number of clinical trials involving a small number of patients with refractory CNS tumors have been performed [23]. A recent Phase I/II study of intranasal administration of perillyl alcohol, an agent that affects the TGF-β and/or Ras signaling pathways, was well tolerated in adult patients with recurrent malignant gliomas when administered via inhalation. The authors report modest activity with radiographic evidence of tumor regression and longterm stability [24]. No clinical trials of intranasally delivered chemotherapeutic agents have been performed in children.

The primary objectives of this study were to determine if intranasal chemotherapeutic drug delivery is safe and feasible, and to establish a proof-of-principle that clinically meaningful exposure could be achieved in the CSF after intranasal delivery. The agents selected for this study, TMZ, valproic acid, and perifosine, were chosen because of their varying degrees of CSF penetration and potential anti-tumor activity. The alkylating agent TMZ is small (MW 194.15 Da) and hydrophilic with a CSF penetration of approximately 33% [25]. Valproic acid is also small (MW 144.20 Da) and hydrophilic, but with a lower CSF penetration of approximately 13% [26]. Commonly used as an antiepileptic, it has an epigenetic role as an HDAC inhibitor in CNS malignancies, although clinically relevant anti-tumor exposures are difficult to achieve because of toxicity. The Akt inhibitor, perifosine, is a large (MW 461.66 Da) and lipophilic agent, with a CSF penetration of only approximately 0.16% [27].

Materials and methods

NHPs

The National Cancer Institute Animal Care and Use Committee approved this study. Adult male rhesus macaques (Macaca mulata) weighing 9.5–14.2 kg were used. Each animal had an indwelling, subcutaneous central venous catheter in the femoral or saphenous vein for blood sampling, and a fourth or lateral ventricular reservoir for CSF sampling. The NHPs were socially housed, fed chow with water ad libitum, and cared for in accordance with the National Research Council (NRC) Guide for the Care and Use of Laboratory Animals [28].

Agents

Temodar^® (temozolomide; Merck, Whitehouse Station, NJ) was supplied as lyophilized powder in 100 mg-vials. Valproic acid was obtained as Valproate Sodium in solution for injection (APP, Fresenius Kabi, USA LLC, Lake Zurich, Il). Perifosine was supplied as lyophilized powder (IV administration: Cancer Therapy Evaluation Program and IN administration: Sigma-Aldrich, St. Louis, MO). All agents were stored at 4 °C. Prior to intranasal administration, the drugs were prepared at room temperature and dissolved either in sterile water (TMZ) or normal saline (perifosine).

Study design and procedure

Systemic administration technique and dosing

The NHP acted as its own control. Each animal received an individual agent systemically at the following doses: TMZ: 7.5 mg/kg (human equivalent dosing (HED) of 150 mg/m²) infused IV over 1 h, valproic acid: 18.5 mg/kg (HED of 370 mg/m²) PO, perifosine: 7 mg/kg (HED of 140 mg/m²) PO. Serial paired whole blood (1 mL) and CSF samples (0.3 mL) were collected prior to all experiments and at multiple set time points within the sampling period depending on the half-life of each agent. Collection times were TMZ: 0–24 h, valproic acid: 0–48 h, perifosine: 0–8 weeks (Supplementary Table 1). After collection, plasma was immediately separated by centrifugation. For TMZ, CSF and plasma samples were placed in tubes prepared with 30 and 100 μL of 1 N hydrochloric acid (10% by volume), respectively, as a preservative. Plasma and CSF were stored at −20 °C then transferred to storage at −80 °C until analysis.

Intranasal administration technique and dosing

Intranasal dosing was limited to the maximum concentration achievable within a total volume of 1 mL (0.5 mL/nare), as this volume was determined to be within safe limits to administer to the animals. Resulting mean doses and numbers of animals receiving each agent were: TMZ (n = 4), intranasal (1.8 mg/kg) and IV (7.5 mg/kg); valproic acid (n = 4), intranasal (9.2 mg/kg) and PO (11.2 mg/kg); and perifosine (n = 3), intranasal (6.8 mg/kg) and PO (7.1 mg/kg). Animals were sedated with intramuscular ketamine (10 mg/kg) and dexmedetomidine (1000 mcg/m²), intubated to prevent aspiration during intranasal administration, and placed in the Mygind position, a dorsal recumbency with the head angled at 30°–45°. This position is optimal for intranasal administration of agents to gain maximum exposure to the olfactory epithelium [29]. Intranasal administration was performed using a mucosal atomization device (MAD, LMA MAD Nasal™, Fig. 1a). Each agent (0.5 mL) was sprayed in each nare followed by 0.2 mL sterile saline flushes. The primates remained in the Mygind position for 15 min after dosing followed by further recovery in the upright position. Blood and CSF sample collection and sampling time points were similar to those used for systemic administration.

Fig. 1.

a Mucosal atomization device (MAD) and b Image of NHP nasal turbinates prior to and 24 h after administration of intranasal agents. Incidental finding of benign epithelial pigmentation (blue arrows) were observed without any identified acute changes directly correlated with intranasal administration of all chemotherapeutic agents

Safety monitoring

Vital signs were monitored during the procedure and throughout recovery. The nasal cavity of each animal was visually evaluated using a WiFi flexible boroscope endoscope for iPhone/iPAD/Android (Vividia), 3.9 mm scope prior to, and at least 24 h following intranasal delivery. (Fig. 1b). They were monitored through clinical chemistries and complete blood counts, which remained within normal limits. Care of the animals was performed in accordance with the National Research Council (NRC) Guide for the Care and Use of Laboratory Animals [28].

Sample analysis

Drug concentrations in plasma and CSF were quantitatively measured using validated ultra-high performance liquid chromatography-tandem mass spectrometry (uHPLC-MS/MS) assays with a lower limit of quantitation (LLQ) of 5, 25, and 0.5 ng/mL, for TMZ [30], valproic acid, and perifosine, respectively.

Pharmacokinetic analysis

Non-compartmental analyses of plasma and CSF pharmacokinetic parameters were calculated using Phoenix^® WinNonlin 6.4 (Certara, Cary, NC). Any plasma concentration measured below the LLQ was excluded from analyses. The maximum concentration (C_max) and time to C_max (T_max) were recorded as observed values. The area under the plasma concentration versus time curve was calculated using the linear up/log down trapezoidal rule, both to the last quantifiable time point (AUC_last) and to time infinity (AUC_INF). The elimination rate (K_el) was calculated as the slope of the best-fit line drawn through at least three terminal concentration points. Half-life (t_1/2) was calculated as ln2/K_el. The mean and sample standard deviation of each of the above parameters was calculated using Microsoft Excel 2011 for Mac, version 14.4.2.

Results

Feasibility

Total planned drug volumes were successfully delivered intranasally to each animal using the MAD. All animals tolerated the procedure well; there were no systemic effects following drug administration; no changes in the nasal cavity were detected by borescope visualization after administration of any agent with the exception of a minimal increase in brown pigmentation around the nasal turbinates (see Fig. 1). This phenomenon was seen with all agents.

Pharmacokinetics

Intranasal and IV TMZ

Logarithmic plots of mean plasma and CSF concentrations versus time by method of administration for TMZ are shown in Fig. 2. Comparisons between the mean (range) C_max, time to C_max (T_max) and half-life (t_½) are shown in Supplementary Table 2. After intranasal administration, median (range) CSF penetration of TMZ (as measured by the AUC_CSF:plasma ratio) was 36 (32–57)%. After IV TMZ administration, CSF penetration was 22 (20–41)% (Fig. 5a). TMZ CSF exposures after IN administration were 40% that in the plasma, compared to 24% after systemic administration.

Fig. 2.

Plasma (a) and CSF (b) concentration (ng/mL) × time (h) logarithmic curves of TMZ (intranasal and IV) up to approximately 8 h

Fig. 5.

AUC_CSF:Plasma comparison between systemic and intranasal (IN) routes for a temozolomide, b valproic acid, and c perifosine

Intranasal and PO valproic acid

The logarithmic plots of the mean of all plasma and CSF concentrations versus time between intranasal and PO valproic acid are shown in Fig. 3. Comparisons between the mean (range) C_max, T_max and t_½ are shown in Supplementary Table 3. For intranasal valproic acid, CSF penetration was 7 (6–9)%; for PO valproic acid, CSF penetration was also 7 (6–8)%. (Fig. 5b).

Fig. 3.

Plasma (a) and CSF (b) concentration (ng/mL) × time (h) logarithmic curves of valproic acid (intranasal and PO) up to 48 h

Intranasal and PO perifosine

The logarithmic plots of the mean of all plasma and CSF concentrations versus time between intranasal and PO perifosine are shown in Fig. 4. Comparisons between the mean (range) C_max and time to C_max (T_max) are shown in Supplementary Table 4. For intranasal perifosine, CSF penetration was 0.4 (0.2–0.9)%; for PO perifosine, CSF penetration was also 0.4 (0.2–0.9)%. (Fig. 5c).

Fig. 4.

Plasma (a) and CSF (b) concentration (ng/mL) × time (h) logarithmic curves of perifosine (intranasal (IN) and PO) up to approximately 1400 h

Discussion

This study demonstrates the safety and feasibility of intranasal administration of chemotherapeutic agents in a non-human primate model. It also demonstrates the proof-of-principle that agents delivered intranasally can reach the CSF. However, the potential pharmacokinetic benefit of intranasal delivery over systemic delivery varies by drug, and may be limited by maximum drug concentration in a deliverable volume.

As expected from the smaller dose administered intranasally, the maximum concentration of TMZ in plasma and CSF was decreased compared to IV administration. However, when comparing the degree of CSF penetration as defined by the AUC_CSF:plasma, a greater percentage of TMZ entered the CSF after intranasal administration compared to IV administration, and exposures achieved in the CSF were in the effective exposure range of TMZ. This suggests that intranasal delivery of TMZ may provide equivalent CNS exposure with less systemic exposure (and hence, less systemic toxicity) compared to IV administration. This may be due to its high CSF penetration, which would allow rapid entry into the CSF and avoidance of first pass effect. In contrast, plasma and CSF exposures for valproic acid and perifosine were similar for both methods of delivery.

This study is the first to use a non-human primate model to investigate the intranasal pharmacokinetics of clinically available chemotherapeutic agents. Several limitations were encountered in the process. Although this study was designed as a proof-of-principle approach, a major limitation was the maximum volume placed on intranasal dosing, precluding direct comparison of CSF penetration by each route of administration. While the study could be designed to administer lower doses systemically, our goal was to evaluate clinically relevant doses, and to determine if effective CSF exposures could be achieved safely after intranasal administration. In addition, our study was limited by the animal model itself. Although prior studies have demonstrated that the NHP model is predictive of pharmacokinetics in children after systemic administration, this species has a greater percentage of olfactory:respiratory epithelium in the nares compared to humans. However, this model is closer to humans than the prior studies evaluating intranasal administration in a rodent species, which has an even greater olfactory:respiratory epithelium ratio [31]. Another consideration in this model is the absence of a closed loop system. Although the Mygind position is the optimal positioning to direct drug towards the olfactory region, some drug is likely absorbed across mucosal membranes or inhaled into the pulmonary system. However, this is a challenge that would also be present in patients. Finally, in this NHP model, due to limitations with CNS tissue access, CSF drug concentrations are utilized as a surrogate for CNS tissue delivery, with the assumption that there is a direct, although not absolute, correlation of CSF and CNS tissue penetration. Studies examining CNS extracellular fluid concentrations after IN delivery will therefore better define drug penetration in the tissue.

In summary, intranasal delivery of chemotherapeutic agents appears safe and feasible. For certain agents, it may offer a pharmacokinetic advantage, potentially improving the efficiency of CNS delivery and limiting toxicities by reducing systemic exposure. The pharmacokinetics of select agents for intranasal delivery should be studied individually to determine the potential benefit over systemic delivery.