The potential of theragnostic ¹²⁴I-8H9 convection-enhanced delivery in diffuse intrinsic pontine glioma

Abstract

Background

Reasons for failure in prior human glioma convection-enhanced delivery (CED) clinical trials remain unclear. Concentration-dependent volume of distribution (V_d) measurement of CED-infused agents in the human brain is challenging and highlights a potential technical shortcoming. Activity of iodine isotope 124 (¹²⁴I ) in tissue can be directly measured in vivo with high resolution via PET. With the potential therapeutic utility of radioimmunotherapy, we postulate ¹²⁴I conjugated to the antiglioma monoclonal antibody 8H9 may serve as a “theragnostic” agent delivered via CED to diffuse intrinsic pontine glioma.

Methods

Fifteen rats underwent CED of 0.1–1.0 mCi of ¹³¹I-8H9 to the pons for toxicity evaluation. Six additional rats underwent CED of 10 µCi of ¹²⁴I-8H9 to the pons for dosimetry, with serial microPET performed for 1 week. Two primates underwent CED of gadolinium-albumin and 1.0 mCi of ¹²⁴I-8H9 to the pons for safety and dosimetry analysis. Serial postoperative PET, blood, and CSF radioactivity counts were performed.

Results

One rat (1.0 mCi ¹³¹I-8H9 infusion) suffered toxicity necessitating early sacrifice. PET analysis in rats yielded a pontine absorbed dose of 37 Gy/mCi. In primates, no toxicity was observed, and absorbed pontine dose was 3.8 Gy/mCi. Activity decreased 10-fold with 48 h following CED in both animal models. Mean V_d was 0.14 cc³ (volume of infusion [Vi] to V_d ratio = 14) in the rat and 6.2 cc³ (V_d/V_i = 9.5) in primate.

Conclusion

The safety and feasibility of ¹²⁴I dosimetry following CED via PET is demonstrated, establishing a preclinical framework for a trial evaluating CED of ¹²⁴I-8H9 for diffuse intrinsic pontine glioma.

Diffuse intrinsic pontine glioma (DIPG) is a uniformly lethal malignancy of childhood, with a median survival of <1 year.^1–4 Palliative radiotherapy is the standard of care. Despite numerous clinical trials, including those investigating various novel chemotherapeutic agents, high-dose myeloablative chemotherapy, and hyperfractionated irradiation, survival has not been affected.^3,5–12 A critical need persists for the development of original therapeutic approaches to this tumor.

Convection-enhanced delivery (CED) is a mode of local delivery that utilizes a cannula directly implanted into target tissue or tumor through which drug is infused under a constant pressure gradient. This achieves high regional concentrations and uniformity of therapeutic molecules.^13–16 From both preclinical animal studies and clinical applications, there is agreement that CED in the brain stem has a promising therapeutic application.^17–20 Moreover, selection of an appropriate therapeutic molecule to deliver via this route is not limited to conventional agents, as CED bypasses the blood–brain barrier (BBB).

Monoclonal antibody (mAb) 8H9 recognizes B7-H3, an outer membrane protein that exhibits complex interactions with T-cells and natural killer cells. This is expressed by the vast majority of human glial tumors and not normal neurons or glia.^18,21 Microarray and immunohistochemical analysis of DIPG samples suggests increased transcription levels in DIPG. Our laboratory has successfully characterized the distribution and toxicity of this antibody delivered by CED in the naïve brain stem as well as orthotopic gliomas in a rat.¹⁸ We have also shown preclinical antitumor efficacy of a recombinant immunotoxin based on the Pseudomonas exotoxin using 8H9 as a targeting mechanism (8H9-PE38) against human glioma.²²

The paucity of in vivo dosimetry data of infused drugs following CED has been a major limitation of major clinical trials in glioblastoma using this delivery technique. Preclinical CED studies have endeavored to determine a prediction for volume of distribution (V_d) as a function of infusion volume (V_i); however, this ratio (V_d/V_i) can vary, particularly as a variable of infusion drug or tracer concentrations.^{17–19,23,24} In addition to its therapeutic potential, CED of an antibody-radioisotope conjugate has the desirable property of establishing dosimetry using PET or other modern imaging techniques. The mAb 8H9 as a radioimmunotherapeutic agent (both iodine isotope 124 [¹²⁴I]–8H9 and ¹³¹I-8H9) has already been used via an intrathecal route in clinical studies against CNS-relapsed neuroblastoma with encouraging results.²⁵ With this established, and given the radiation responsiveness of DIPG, ¹²⁴I conjugated to the antiglioma mAb 8H9 produces a theoretical “theragnostic” agent against this tumor. By using CED of ¹²⁴I-8H9 and ¹³¹I-8H9 to the brain stem in the preclinical setting, we hypothesized that this treatment approach would be safe and would offer a method for determining the distribution and dose in live animals. Further, it is believed that these infusion parameters and results would ultimately be integrated into the design of a clinical trial in children with DIPG.

Materials and Methods

Radioisotope-Antibody Conjugation

The murine IgG₁ 8H9 was produced by hyperimmunizing BALB/c mice with human neuroblastoma, as previously described. The mAb 8H9 was radiolabeled with the following technique. The mAb was allowed to react for 5 min with ¹²⁴I or ¹³¹I (¹²⁴I produced on a Sloan-Kettering cyclotron; ¹³¹I from PerkinElmer Life and Analytical Sciences) and 1 mg/mL chloramine T (Sigma-Aldrich) in 0.3 M phosphate buffer, pH 7.2, at room temperature at an antibody to chloramine T molar ratio of 1:700. The reaction was stopped by adding 1 mg/mL sodium metabisulfite (Sigma-Aldrich) in 0.3 M phosphate buffer, pH 7.2, for 2 min. Radiolabeled mAb 8H9 was separated from free iodine using an A1GX8 resin column (BioRad) saturated with 1% human serum albumin (New York Blood Center) in phosphate buffered saline, pH 7.4. Peak radioactive fractions were pooled, and radioactivity was measured using a radioisotope calibrator (Bristol-Meyers Squibb). Iodine incorporation and specific activities were calculated. Precipitation of trichloroacetic acid (Fisher Scientific) was used to assess the percentage of protein-bound ¹²⁴I or ¹³¹I. Thin-layer chromatography (TLC) was performed by running 1 L of iodinated mAb, in a silica gel-on-glass TLC plate (Sigma-Aldrich) and scanning it with a System 200 Imaging Scanner (Bioscan). A final specific activity of 2.0 mCi to 1.0 mg of antibody was achieved.

Convection-enhanced Delivery of ¹³¹I-8H9, Safety, and Toxicity in a Rodent Model

All rodent procedures were performed in accordance with guidelines and approved protocols from the Weill Cornell Medical College and the Memorial Sloan-Kettering Cancer Center Institutional Committee of Animal Care and Use. To test for the safety and tolerability of radioiodine-labeled 8H9 CED to the brain stem, a series of 15 Sprague-Dawley rats weighing 200–300 g underwent stereotactically guided CED of ¹³¹I-8H9 to the brain stem at a rate of 0.2 µL/min. The procedure was performed as we have previously published.¹⁸ Animals received a dose of 0.1, 0.3, or 1.0 mCi in groups of 5 rats each, all in volumes of 0.01 mL of ¹³¹I-8H9.

Following infusion, animals were examined on a daily basis for 3 months for evidence of clinical or neurological toxicity. If no apparent clinical toxicity was observed, sacrifice was performed at 3 months following CED. The brain was harvested, paraffin embedded, cut, stained with hematoxylin and eosin, and examined by a blinded neuropathologist (M.A.E.).

PET Imaging and Dosimetry of ¹²⁴I-8H9 in a Rodent Model

To evaluate anatomical distribution and biologically absorbed dose of radiation, CED of 1.0 µCi ¹²⁴I-8H9 in a volume of 0.01 mL was performed as already noted to the pons of 6 rats. PET scans were performed on a microPET Focus 120 rodent scanner (Concorde Microsystems) within 5 h after infusion and then repeated on a daily basis for a 7-day period. List-mode data were acquired for 5–30 min (with longer imaging times at later times post-injection) using a γ-ray energy window of 350–750 keV and a coincidence timing window of 6 ns. The resulting list-mode data were sorted into 2D histograms by Fourier rebinning, and transverse images were reconstructed by filtered back-projection using MicroPET Manager software (Concorde Microsystems). The counts in the reconstructed images were then converted to activity concentration (in percent of the injected dose per gram of tissue [%ID/g]) using a system calibration factor (mCi/mL/cps/voxel) derived from imaging of a mouse-size water-equivalent phantom. The reconstructed images were analyzed using ASIPro VM software (Concorde Microsystems).

At each imaging time point, the mean activity concentration (in %ID/g) in the infusion site was determined by region-of-interest analysis using the microPET's ASIPro VM software. The resulting time-activity concentration data were fit to biexponential functions, which then analytically integrated (incorporating the effect of physical decay of ¹²⁴I and of ¹³¹I) to determine the cumulated activity concentrations, or the total number of decays per gram of tissues, of ¹²⁴I and of ¹³¹I, respectively, in the infusion site. The ¹²⁴I and ¹³¹I mean absorbed doses to these tissues were then obtained by multiplying the cumulated activity concentrations of ¹²⁴I and ¹³¹I by their respective equilibrium dose constants for nonpenetrating radiations (ie, β-rays).

Safety and Dosimetry in a Primate Model

To more directly assay the potential toxicity and feasibility of in vivo imaging and dosimetry in a model relevant to the human condition, CED of ¹²⁴I-8H9 was performed in the primate brain stem with co-infusion of gadolinium (Gd)-bound albumin. The covalently bound Gd-albumin complex has precedence for use as a CED tracer in the brain, where its properties of stability and retention in host tissue may be advantageous (as opposed to its potential use as an intravascular contrast agent).^26,27 All procedures in primates were performed in accordance with guidelines and approved protocols from the Cornell and Memorial Sloan-Kettering Cancer Center Institutional Committee of Animal Care and Use.

Surgery was performed in 2 male cynomolgus monkeys, age 2 years and weights 2.0–2.2 kg. On the morning of surgery, the primate was placed under general endotracheal anesthesia and fitted into an MRI-compatible stereotactic frame (David Kopf Instruments). For targeting purposes, a T1-weighted MRI was performed using 1.2-mm slice thickness. Following incision and subperiosteal dissection, a burr hole was created using a high-speed drill (Stryker Instruments). A silica step-design cannula with outer-inner tubes was used for infusion (PlasticsOne). The outer 22-gauge cannula was lowered into the infusion target using a frame-compatible stereotactic arm (David Kopf Instruments). The inner 28-gauge cannula, connected to plastic tubing, was affixed to a 2.5-mL Hamilton syringe, connected to a microprocessor-driven infusion pump. The inner cannula was placed through the outer cannula to the target depth and CED was performed. Following the cessation of infusion, both cannulae were left in place for 5 min and then slowly withdrawn. The incision was closed and the primate was removed from the frame and transported to MRI for an immediate postoperative scan to confirm accurate targeting and delivery. Following this MRI, the animal was extubated and recovered.

In the first primate, to demonstrate safety of the tracer, 30 µL Gd-albumin (25 mg/mL) was infused to the brain stem at a constant rate of 0.2 µL/min. To determine whether infusion caused any delayed or chronic toxicity, this primate also underwent MRI 6 months following CED. In the second primate, 1 mCi of ¹²⁴I-8H9 was co-infused with 30 µL Gd-albumin (25 mg/mL) in a total infusion volume of 560 µL. The infusion rate was started at 0.5 µL/min and was escalated by 0.5 µL/min every 10 min to reach a maximum rate of 7.5 µL/min. PET imaging was performed 2 h postoperatively for dosimetry evaluation in addition to an immediate postoperative MRI following infusion. A second PET scan was performed 40 h after surgery, and blood and CSF were drawn for radioactivity counts 12 days after CED of ¹²⁴I-8H9.

After surgery, observation was conducted on a daily basis with clinical and neurological examination to evaluate for toxicity. Evaluation of complete blood counts and serum chemistries was performed on a monthly basis. At the end of an observation period of 21 months for primate 1 and 9 months for primate 2, the animals were sacrificed and brains were histologically examined.

Results

Toxicity Assessment in the Rodent

Of the 15 rats that underwent infusion of this isotope, 14 showed no evidence of clinical toxicity in terms of neurological dysfunction, behavioral abnormalities, or weight loss (Table 1). One rat who received 0.1 mCi of ¹³¹I-8H9 experienced wound dehiscence 3 days following CED, necessitating debridement and wound revision. This rat ultimately made an uneventful recovery. One rat who received a dose of 1.0 mCi of ¹³¹I-8H9 suffered severe hemiparesis, necessitating sacrifice on post-CED day 6. Postmortem histology demonstrated a significant necrotic cavity at the infusion site in the right pons (Fig. 1).

Table 1.

Toxicity in rats following CED of ¹³¹I-8H9

Dose	N	Toxicity
0.1 mCi	5	One wound dehiscence
0.3 mCi	5	One wound infection
1.0 mCi	5	One severe hemiparesis

Fig. 1.

 (A) Necrotic cavity in the right pons (arrow) in rat with severe toxicity from 1.0 mCi ¹³¹I-8H9 infusion. This rat had significant hemiparesis, necessitating early sacrifice. (B) A 10× hematoxylin/eosin section from a rat that underwent infusion of 1.0 mCi ¹³¹I-8H9 without any evidence of clinical toxicity.

Dosimetry of ¹²⁴I-8H9 in the Rodent

Activity following CED was readily measured by microPET, and the 3 predominant sites of activity were the brain stem, brain insertion site, and thyroid (Table 2). PET revealed a mean maximum V_d of 0.14 cc³ (V_d/V_i = 14). Mean absorbed dose of radiation per mCi of 124I-8H9 administered was 37.8 ± 6.9 cGy (range, 16.0–55.9 cGy). By 72 h following infusion, <5% of infusion activity concentration remained in the primary infusion site in all rats (Fig. 2).

Table 2.

Dosimetry of ¹²⁴I-8H9 following CED in rats and primates

Animal	Injection Site
	Activity Concentration (%ID/g)	Vd (mL)	Total Activity (%ID)	124I Absorbed Dose (Gy/mCi)
Rat 1	18	0.41	7.2	29.4
Rat 2	17	0.23	3.9	16.1
Rat 3	26	1.2	31	23.6
Rat 4	56	1.2	69	52.6
Rat 5	52	1.1	58	48.6
Rat 6	75	1.7	126	55.9
Primate	2.4	4.1	9.8	3.80

Fig. 2.

Radiation activity in the rat brain stem target and brain (“hot brain”) as a function of time following CED.

Toxicity Assessment in the Primate

No intraoperative complication was experienced in either primate. Both primates had immediate postoperative MRI, which confirmed accurate infusion to the brain stem (based on visualization of the Gd tracer; Fig. 3A and B). For primate 1, no abnormalities in clinical behavior or neurological examination were noted over a period of 6 months of postoperative examination. An MRI performed 6 months following surgery showed only minimal hemosiderin deposition at the infusion site, without any evidence of cystic or necrotic change in the brain stem (Fig. 3C). This primate was ultimately sacrificed 21 months following infusion. Histologic examination showed discoloration on the surface of the right frontal lobe, presumably from hemosiderin deposition, the center at the intersection of the superior frontal sulcus and the precentral sulcus and thinning of cortex here (Fig. 4). The cannula insertion tract was seen passing through the right internal capsule, ending at the center of the right pons. A dural adhesion on the right frontal side of the skull at the burr hole site was also appreciated.

Fig. 3.

MRI performed immediately following CED to the brain stem of both primates confirmed accurate targeting (A and B), and MRI in the primate performed 6 months following CED of Gd-albumin showed no evidence of cystic or necrotic change (C). The animal never demonstrated evidence of clinical toxicity from this infusion.

Fig. 4.

Histologic evaluation of primates 1 and 2 depicted at 400× magnification. No evidence of microscopic toxicity was seen in primate 1 (A), and in primate 2 a focal region of neuronal loss, gliosis, and hemosiderin-laden macrophages were seen at the site of infusion in the pons (B).

In primate 2, postoperative recovery was also uneventful. The monkey showed no evidence of neurological toxicity or any other behavioral abnormality otherwise. The wound healed well, and there were no abnormalities on routine serum chemistry and hematologic evaluations. No abnormality was seen on MRI (both T1 and T2, all 3 axes) performed 7 months following surgery. The primate was ultimately sacrificed 9 months following infusion, and histology from the infusion site in the right pons showed no evidence of cystic or necrotic change but some mild microscopic neuronal loss and gliosis at the infusion site compared with the normal left side (Fig. 4).

Dosimetry in the Primate

PET imaging detected high activity counts in the brain stem, with anatomical distribution of high levels of radiation following co-registration with MRI brain of the primate encompassing the entire pons and midbrain (Fig. 5). Activity was also present on the brain surface, spinal CSF, and thyroid. Activity concentration (%ID/g) decreased almost 10-fold between 2 and 36 h following infusion, from 2.4% to 0.35%. Biological absorbed dose was 380 cGy. Volumes of distribution of radiation in the brain stem were 4.1 and 6.2 cc³ observed 2 and 40 h, respectively, following infusion, as measured by PET (maximum V_d/V_i = 9.5). In contrast, V_d estimate from Gd distribution on T1 MRI performed 1 h following infusion was 0.9 cc³. Doses to the thyroid, spinal CSF, and the brain surface detected by PET scan were 200, 70, and 30 cGy, respectively. There was no residual activity in CSF and blood drawn 12 days after surgery.

Fig. 5.

PET/CT co-registered in the primate following CED of ¹²⁴I-8H9, post-infusion at 2 and 40 h.

Discussion

The necessity for novel modes of therapy in malignant glioma and DIPG remains high. The past decade has been witness to numerous preclinical and human trials investigating the potential application of local pressure-driven drug infusion, or CED, into the brain for incurable malignant tumors.^{14,17,20,28–36} The increasingly widespread investigation of this delivery methodology is largely attributable to its inherent properties of bypassing the BBB, resulting in enhanced local therapeutic uniformity and distribution. An additional appealing feature of local delivery is the avoidance of any systemic toxicity of the agent.

CED in human clinical studies has been explored with malignant gliomas in adults. One noteworthy effort in this regard evaluated the safety and efficacy of targeted recombinant immunotoxin IL13-PE38QQR. After years of preclinical investigation demonstrating this agent's safety and potential efficacy in preclinical models, CED of this agent was explored in humans with recurrent glioblastoma in phase I and II trials with relative safety.³⁷ A phase III investigation (the PRECISE study group) compared efficacy in terms of survival against implantable carmustine wafers and failed to show any survival benefit for CED of IL13-PE38QQR.³⁸ There was no attempt in this study to assess in vivo drug distribution following delivery.

This result is the most compelling impetus for an improved methodology to assess drug dosimetry following local delivery to the brain. High-grade glioma, DIPG included, poses significant biological resistance to various well-designed targeted therapies. However, potential reasons for failure of a study like the PRECISE trial are not limited to therapeutic properties of the drug. The diffusely infiltrative nature of these tumors has been cited as the reason why CED may portend an advantage (as infusate “infiltrates” in a manner similar to tumor cells), but it also heightens the importance of achieving a large V_d that spans tumor boundaries. If distribution is inadequate, treatment will predictably fail. This was well recognized by the investigators of the PRECISE trial, who have since even attempted to evaluate positioning of catheters in relation to therapeutic outcome.^39,40 An accurate methodology for in vivo dosimetry assessment remains a challenge.

Surrogate tracers, especially Gd-bound agents with MRI, have shown a relatively good ability to demarcate distribution on radiographic studies.^19,23,24 However, the ideal methodology for dose estimation would directly assay the distribution and concentration of the therapeutic agent itself. For one, distribution and dosimetry of targeted agents may theoretically differ from therapeutically inert agents, such as Gd, especially in the presence of the tumor target (the cellular basis of targeted therapy to some extent implies this). Indeed, our results showed that the V_d/V_i of PET ¹²⁴I-8H9 activity was 4.6 times greater than the V_d/V_i of the Gd tracer on MRI measured within a 1-h interval post-infusion in the primate. In addition, effective distribution could be limited by a proximal pial, ependymal, or resection cavity surface, all of which can act as a sink for infusate diversion.⁴¹ Further, concentration and dose of the therapeutic agent play a major role in V_d and may not be accurately represented by tracer of another standard concentration.^18,23,42,43 Lastly, while surrogate tracers have been shown to approximate the distribution of therapeutic agents immediately after CED, no assessment has been done on relative clearance rates between these molecules.

Given that (i) there are neither proximal resection cavity nor cyst surfaces to the central brain stem infusion site in DIPG, (ii) all current conventional chemotherapy in DIPG meets with relatively rapid failure, and (iii) DIPG does respond to radiotherapy, we have selected this tumor as a focus for the development of CED-mediated radioimmunotherapy. The current investigation sought to test the preclinical safety of CED of the radioiodine isotope-conjugated antiglioma antibody ¹²⁴I-8H9 and to more critically establish an effective in vivo methodology to assess post-infusion dosimetry that could be translated in the setting of a clinical trial. It was found that doses up to 1.0 mCi of ¹³¹I-8H9 and ¹²⁴I-8H9 were tolerated in the brain stem in all but 1 rat and the primate and that those infusion activities produced a mean absorbed dose of 37.8 and 3.80 Gy per mCi of 124I-8H9 administered in rats and primates, respectively. The ratios of V_d/V_i were 14 and 9.5 in these naïve rats and monkeys, respectively. Measurement of absorbed dose and anatomical distribution of dose absorption was feasible in vivo using PET imaging following CED of ¹²⁴I-8H9. The thyroid was the predominant site of radiation absorption outside the CNS; we postulate that this is most likely due to a combination of CSF clearance and vascular uptake of ¹²⁴I-8H9 secondary to mechanical BBB disruption, which occurs following cannula insertion for CED. A thyroid blocking agent, such as potassium iodide, would need to be implemented in the clinical setting.

The V_d/V_i ratio observed in this series is comparable to a range of 2–14 observed in other preclinical and clinical investigations.^17,19,23,24 The disparity between absorbed radiation dose in the rat and primate, however, was not anticipated and cannot be explained with certainty. Although it is possible (and has been seen in other studies⁴⁴) that increase in the V_i and size of the target distribution may have played a role in a roughly 10-fold reduction in the ratio of absorbed dose to infusion activity, the exact reason for this phenomenon is unknown. Regardless, the relative unpredictability of drug dosimetry with various therapeutic agents following CED, especially in a tumor target, highlights the importance of implementing in vivo dose absorption measurements.

The extensive aforementioned preclinical and clinical work with CED, the proven efficacy of ^131/124I-8H9 in the treatment of CNS neuroblastoma, and the current investigation have all laid the foundation for a structured clinical trial employing CED of ¹²⁴I-8H9 in children with DIPG. This trial would have the especially desirable ability to directly evaluate drug distribution and dosimetry in the brain following treatment. Such a study would offer a novel arm of therapy for children without hope of lasting survival benefit from conventional therapy. It could also provide previously lacking in vivo dosimetry information for investigators who aim to elucidate reasons for previous failure, and improve drug delivery techniques for incurable malignant brain tumors in the future.

Funding

This work was supported by the Christian Rivera Foundation, the Beez Foundation, and the Lehman Brothers Foundation.

Conflict of interest statement. A potential conflict of interest for investigators at Memorial Sloan-Kettering Cancer Center (MSKCC) exists: 8H9 was licensed by MSKCC to United Therapeutics Inc, Silver Spring, MD.