Methods for Intratumoral Microdialysis Probe Targeting and Validation in Murine Brain Tumor Models

Abstract

Background:

Microdialysis is a well validated sampling technique that can be used for pharmacokinetic studies of oncological drugs targeting the central nervous system. This technique has also been applied to evaluate tumor metabolism and identify pharmacodynamic biomarkers of drug activity. Despite the potential utility of microdialysis for therapeutic discovery, variability in tumor size and location hamper routine use of microdialysis as a preclinical tool. Quantitative validation of microdialysis membrane location relative to radiographically evident tumor regions could facilitate rigorous preclinical studies. However, a widely accessible standardized workflow for preclinical catheter placement and validation is needed.

New method:

We provide methods for a workflow to yield tailored placement of microdialysis probes within a murine intracranial tumor and illustrate in an IDH1-mutant patient-derived xenograft (PDX) model. This detailed workflow uses a freely available on-line tool built within 3D-slicer freeware to target microdialysis probe placement within the tumor core and validate probe placement fully within the tumor.

Results:

We illustrate use of this workflow to validate microdialysis probe location relative to implanted IDH1-mutant PDXs, using the microdialysis probes to quantify levels of extracellular onco-metabolite D-2 hydroxyglutarate.

Comparison with Existing Methods: Previous methods have used 3D slicer to reliably measure tumor volumes. Prior microdialysis studies have targeted expected tumor locations without validation.

Conclusions:

The new method offers a streamlined and freely available workflow in 3D slicer to optimize and validate microdialysis probe placement within a murine brain tumor.

Introduction

Gliomas and other solid tumors are characterized in part by altered metabolism which promotes tumorigenesis and malignant progression [1]. Improved understanding tumor metabolism may reveal therapeutic vulnerabilities and candidate pharmacodynamic biomarkers. Given dynamic tumor heterogeneity, longitudinal access to the intratumoral microenvironment may improve mechanistic understanding of tumor evolution and therapeutic resistance.

Microdialysis is an important technique for continuous sampling of extracellular analytes within tissues of interest. The method has been used to analyze metabolites [2], neurotransmitters [3], and cytokines [4] in several models of brain injury [5]. It is also used in pharmacokinetic studies to measure drug concentrations in brain and tumor [6, 7] and has been utilized to identify tumor- and therapy-associated metabolic alterations in GBM [8]. A rigorous preclinical workflow for intracranial microdialysis could facilitate important translational efforts by defining candidate metabolic biomarkers of tumor subtype, vulnerability and therapeutic response.

Here, we have adapted 3D slicer (http://www.slicer.org) as a tool to guide microdialysis probe implantation into established intra-cranial tumors and to validate placement relative to the center and margin of variably located tumors. 3D slicer is a freely available research software platform for visualization, manipulation, co-registration and analysis of biomedical images. Within 3D slicer, we have built 3D models for commonly utilized commercially available Eicom microdialysis probes that can be aligned with the CT-visible portion of the guide cannula. Once aligned to the CT, the model demonstrates the position of the microdialysis membrane otherwise unseen on imaging. This methodology was utilized to facilitate microdialysis probe placement in an orthotopic mouse model of isocitrate dehydrogenase 1 (IDH-1) mutant patient-derived xenografts (PDX). From a subset of the microdialysate collected, we found (D) 2-hydroxyglutarate (D2HG) was detectable in all samples obtained from microdialysis probes placed within and immediately adjacent to the radiographically evident tumor.

Methods and Materials:

Basic Protocol # 1: Intracranial tumor implantation

Short Term Explant Cell Cultures

Short-term explant cultures (GBM164; GBM196) were derived from flank PDXs established in female athymic nude mice as previously described utilizing female Hsd:athymic Nude-Foxn1nu, ages 8–9 week old (Envigo, Indianapolis, IN, USA) [9, 10]. Briefly, mechanically disaggregated tumors were plated on reduced matrigel coated flasks. PDX164 were maintained in neural stem cell media: Stem Pro NSC SFM: (Invitrogen, Carlsbad, CA, USA) and PDX196 maintained in DMEM (Corning) supplemented with 10% fetal bovine serum (FBS) (Atlanta Biologicals, Minneapolis, MN, USA). All xenograft lines were routinely tested for mycoplasma at six months intervals.

Intracranial tumor implantation

All animal experiments were reviewed, approved by, and conducted in accordance with, the Institutional Animal Care and Use Committee (IACUC) of the Mayo Clinic. The procedure for establishing intracranial tumors has been described previously [11, 12]. Briefly, cells from short-term explant cultures were harvested by trypsinization and suspended in phosphate buffered saline (PBS). Mice were anesthetized by delivering a Xylazine/Ketamine anesthetizing cocktail (20mg/ml of xylazine and 100mg/ml of ketamine) given intraperitoneally (dose=0.01 mL/g/mouse). Anesthetized mice were immobilized in a Kopf stereotaxic frame with the toothbar positioned to ensure a level skull between bregma and lambda. Coordinates for implantation were 1mm anterior, 2mm lateral, and 2mm deep to bregma. A high-speed drill was used to prepare the burr hole. Three microliters of cell suspension (1×10⁵ cells / μL) were implanted using a 10 μl Hamilton syringe with a 26G needle at 1 μl/minute; the needle was withdrawn after 1 minute.

The burr hole was then filled sealed with bone wax to minimize the potential for extra-cerebral cranial extension of the tumor tissue growth. The scalp wound was sutured with 4–0 vicryl using an rb-1 needle (Ethicon, Cincinnati, OH, USA). Mice were placed in clean cages on a heating pad and allowed to recover. The animals received a standard mouse diet and water ad libitum, and 3 days of post-operative analgesic care that consisted of children’s liquid ibuprofen mixed with drinking water (1mg/ml). Animals were assessed daily after surgery for body condition and monitored for neurological symptoms for three days post-surgery.

Magnetic Resonance Imaging

Approximately 43 days after tumor implantation, Magnetic Resonance Imaging (MRI) imaging was performed using a Bruker Avance 300 MHz, (7 Tesla), vertical bore Nuclear Magnetic Resonance (NMR) spectrometer (Bruker Biospin, Billerica, MA, USA). During data acquisition, animal core temperatures were maintained at 37°C by a flow of warm air. Isoflurane anesthesia (1.5–2.5%) in oxygen was delivered via nose cone.

A 20 mm diameter volume coil was used as the radiofrequency transmitter and receiver. Temperature of the coil was maintained by a heating block built into the gradient system. Respiration was monitored by a respiration gate throughout the entire scan. A single three-dimensional (3D) T2-weighted fast spin echo sequence (FSE) was acquired with echo time (TE)/repetition time (TR) = 45.20/2000ms, bandwidth (BW) = 110kHz. The total imaging time per mouse was ~10 minutes.

Basic Protocol # 2: Microdialysis probe targeting

Each MRI image was converted from Brukers format to NIFTI using bruker2nifti software (http://joss.theoj.org/papers/2ee6a3a3b1a4d8df1633f601bf2b0ffe) for easier management in 3D Slicer [13]. An MRI of a control mouse brain was used as a template to co-register with a rendered mouse skull (http://digimorph.org/specimens/Mus_musculus/). Both are placed within the 3D slicer “Stage Space,” wherein bregma is defined as the origin. The template is positioned in stage space so that lambda has the same z- and x-coordinates as bregma. The following procedures are followed individually for each mouse. Please see supplemental online methods for stepwise details. Acquired brain MRIs of tumor-bearing mice are registered to the template in the Stage Space [14] with general registration or manual fine alignment using the template as a fixed image. Translation and Rigid Body Rotation transforms are then applied with 0.1 step sizes over 10,000 steps evaluated using Mattes metric. The center of tumor mass is manually determined for each mouse and an entry site identified relative to Bregma such that the vertical trajectory passes through the center of the tumor core. This entry site was compared to the originally placed burr-hole to determine whether or not the original burrhole would provide catheter positioning fully within the tumor volume. If not, the catheter model was adjusted anterior or posterior, left or right within the stage of the mouse to identify the new optimal burrhole coordinates. In early pilot studies we found that oblique placement of the guide cannula through the original burrhole toward the center of the tumor resulted in decreased guide cannula stability, so thereafter utilized a new burrhole when tumor location required. A representative MRI of a tumor-bearing mouse is shown co-registered to the provided template in Figure 1.

Figure 1: 3D slicer modules.

Location of bregma and lambda on a co-registered MRI scan. Anatomical landmarks, such as the skull plate fusion sites bregma and lambda, appear in the appropriate stage space locations following co-registration of the MRI images to the Template MRI space.

Basic Protocol # 3: Surgical Implantation of guide cannula and microdialysis probe

Surgical Implantation

Approximately 65 days (PDX 196) and 45 days (for PDX 164) after tumor implantation, mice underwent placement of guide cannulas; CXG-02 (Amuza, San Diego, CA, USA). Mice were anesthetized and placed in the stereotaxic frame in the same manner as above. Incision sites from the first surgery were opened, and the original burr hole identified. If the original burrhole was found to be adequate in step #2 above, bone wax was removed.. Otherwise, a new burr hole was prepared at the newly defined coordinates directly overlying the tumor. A guide cannula was implanted at the injection site and secured with two anchor screws and dental glue. A “dummy probe” CXD(T)-2 (Amuza, San Diego, CA, USA) was placed into the guide cannula after guide cannula implantation for all animals in Table 1. For animals in Table 2, probe CXD(T)-2 built by the Division of Engineering (DOE) at the Mayo Clinic was used. Post-operative care was provided as described above.

Table 1:

Validation of final catheter placement. Tumor center to probe center (mm) to estimated percentage of probe within the tumor: Tumor volume (mm³), distance (mm) of tumor center to probe center, and estimated % of probe within the tumor for each mouse bearing IDH1-mutant PDX (196/164).

Mouse ID	PDX	Tumor Volume (mm3)	Distance : tumor center to probe center (mm)	Estimated % of probe within tumor
1	196	17.56	3.20	50
2	196	8.86	2.03	0
3	196	10.95	1.10	60
4	196	2.82	1.05	40
5	164	18.12	0.49	95
6	164	9.54	0.23	100
7	164	8.36	0.59	75
8	164	32.78	0.70	100
9	164	22.35	0.60	100

Table 2:

Levels of D2HG in microdialysate: From a subset of mice with GBM 164 IDH-1 mutant PDX, D2HG levels were measured using targeted LC/MS mass spectrometry.

Mouse ID	PDX	Tumor Volume (mm3)	Distance : tumor center to probe center (mm)	Estimated % of probe within tumor	D 2 hydroxyglutarate (D2HG) μM
10	164	8.49	0.70	100	1.07
11	164	24.19	0.99	100	2.35
12	164	11.10	0.90	90	3.37
13	164	8.41	0.58	65	2.231
14	164	25.12	0.534	100	3.80

Brain Microdialysis

For microdialysis, the dummy probe was removed from the guide cannula and replaced with a brain microdialysis probe CX-I series; model CX-I-2–02 (Amuza, San Diego, CA, USA) that contains two ports connected to a membrane; one referred to as the inlet and other as the outlet. A Hamilton gastight syringe 2.5 ml (Hamilton Reno, NV, USA) was placed in an ESP101 series pump (Amuza, San Diego, CA, USA). Catheter from the syringe was used to pump Ringer’s solution into a two-channel TCS2–21 cannula swivel (Amuza, San Diego, CA, USA). The inlets and the outlet of the microdialysis probe inlet was connected to the swivel. A catheter from the cannula swivel was connected to a microdialysis fraction collector FC90 (Amuza, San Diego, CA) to collect microdialysate. The cannula swivel allows for sampling from a freely moving mouse. Prior to placement, the microdialysis probe was perfused at a flow rate of 3 μL/min for at least 20 min with sterile filtered and degassed Ringer’s solution (147 mM NaCl, 2.7 mM KCl, 1.2 mM CaCl₂, and 0.85 mM MgCl_2,, pH=7.5). The dummy probe was then removed and the microdialysis probe inserted into the guide cannula, after which the mouse was permitted to move freely during microdialysis, performed at 1 μL/min. Given dead space within the tubing, microdialysate collected during the first 30 minutes was discarded. Thereafter, microdialysate was collected for 2 hrs with samples maintained at 4–6°C for the duration of microdialysis. Samples were transferred to sterile 200 μL polypropylene tubes after collection and stored at −80°C until analyzed.

Basic Protocol # 4: Tumor target validation

CT scans

After the dialysate had been collected, cone-beam CT (CBCT) scans were obtained using a calibrated X-RAD SmART irradiator (Precision X-ray, North Branford, CT, USA). Isoflurane-anesthetized mice were cranially fixed on the stereotactic stage using a bite block as described in [15] and imaged via a high resolution CBCT protocol (40 kVp, 8.0mA, 2.0 mm Al filter, 256 images, 100 μm3 voxel size).

Co-registration of the CT and MRI scans, Tumor Segmentation, and Microdialysis Probe Placement Validation

CT scans were converted into NIfTI (Neuroimaging Informatics Technology Initiative) (PMID: 31231202) format and loaded into the MRI-containing Stage space (please see supplemental methods) for each animal. CT scans were manually co-registered with the previous animal-specific MRI using 3D Slicer’s “Transforms” Module such that the skull volume visible in the CT scan, accurately encompassed the brain volume visualized in the MRI scan. In this manner the original models were used only to define Bregma as assisting with alignment. Final colocalization was performed using the experimentally acquired CT and MRI averting any concern regarding anatomical differences between the model and the experimental animal. Tumor volume and dimensions were then defined within the pre-operative MRI using the “Segment Editor” Module within 3D Slicer to highlight and select tumor-containing voxels. As such, a volumetric reconstruction of the tumor was obtained within the 3D view panel of 3D Slicer (Figure 2). Microdialysis probe placement validation was performed by loading a 3D model catheter probe model and aligning in the “Transforms” Module to the microdialysis probes as visualized in the post-surgical CT scan.

Figure 2: A representative image in 3D slicer.

Example of tumor segmentation (3D Slicer - Segment Editor Module; MRI reconstruction) and probe placement (3D Slicer – Models Module; CT-based alignment) represented in axial slice (top-left), sagittal slice (bottom-left), coronal slice (bottom-right), and 3-dimensional (top-right) views.

Determination of Tumor and Probe Spatial Placement Coordinates

Tumor volume was determined via the “Segment Statistics” Module within 3D Slicer, using the aforementioned tumor “Segmentation” reconstruction and the study animal MRI as the scalar volume. Tumor center was determined by measuring the three orthogonal tumor dimensions, including their bregma coordinate boundaries, and placing a fiducial at the mid-point for each orthogonal dimension. “Tumor center-to-probe center” was determined algebraically from the 3D coordinates of the “tumor center” and “membrane center” fiducials. The approximate percentage of the microdialysis probe surface area located within the tumor mass was estimated by visualizing the evaluating probe location within the virtual tumor reconstruction using orthogonal image planes and the 3D reconstruction. Since the catheter was sometimes seen to pass obliquely along the edge of the tumor, visual inspection of this reconstructed 3D relationship was used to estimate the percentage of the total probe surface area within the tumor to the nearest 5%.

Quantification of D and L 2-HG using LCMS

Microdialysate samples were processed by the Mayo metabolomics core for analysis of L and D forms of 2D-hydroxyglutarate (2-HG). Concentration of L and D isomers of 2-HG were separated and quantified by liquid chromatograph mass spectrometry (LC/MS) as previously described [16, 17] with a few modifications. Briefly 20ul of internal standard solution containing U-¹³C labeled 2-hydroxyglutarate were added to 60ul of microdialysate. Proteins were precipitated by adding 500ul of chilled 80% methanol solution to the sample mixture. The mixture was allowed to incubate on ice for 40 minutes prior to removal of the supernatant. After drying the supernatant in a speed vacuum under medium heat, the samples were derivatized with Diacetyl-L-Tartaric Anhydride (DATAN, 25 mg/ml in 4:1 dichloromethane: acetic acid). Samples were dried and resuspended in 100uL of water before being analyzed on a Nexera X2 UPLC module coupled with an AB Sciex QTrap 6500 mass spectrometer. Metabolites were separated on an Acquity HSS-T3, 1.8 μm, 2.1 × 50 mm column (Waters Corp, Milford, MA, USA), held at 40 °C, using 99% water, 1% acetonitrile, and 5 mM ammonium formate in water, adjusted to pH 3.3 with formic acid as mobile phase A, and 99% acetonitrile, 1% water, and 0.1% formic acid as mobile phase B. The flow rate was 0.3 mL/min and 2-HG D- and L-isomers were separated with an isocratic elution (99%A, 1%B) for 8 minutes. The mass spectrometer was operated in electrospray ionization - mode, monitoring mass transitions of m/z 152 -›134 for DATAN labeled 2-HG and 147-›129 for 2-HG. Concentrations of both isomers were measured against 10-point calibration curves that underwent the same derivatization.

Results

MRI and CT scans of mice implanted with either PDX 196 or 164 PDX were used to analyze probe sampling location using 3D slicer. For each mouse, tumor volume, distance between tumor center and probe center, and an estimated percentage of probe within tumor were analyzed. The results for each mouse are summarized in Table 1.

We initially performed pilot studies with PDX 196 (n=4). Tumor size around 65 days was found to be quite variable ranging from 2.82–17.56mm³. Direct contact between the catheter membrane and the tumor was demonstrated for 3 of 4 animals. The catheter for mouse #2 fell just outside of the edge of the tumor, with the catheter center 2.03mm from the tumor center. GBM 196 was found to have indiscrete margins on T2-weighted MRI leading to some challenges in accurately defining tumor borders. As such we proceeded with GBM 164 which we found to be more discretely defined on T2-weighted MRI. Segmented tumor size on MRI ranged from 8.36–32.78mm³. All catheters (10/10) for PDX164 fell <1mm from the tumor center.

Microdialysate from the final 5 animals was utilized for targeted analysis of D2HG and L2HG. Levels of L2HG were below the limit of detection for all samples tested, in a similar manner to prior reports [18]. However, elevated D2HG (1.07–3.80μM) was recovered from 5/5 animals. The assay limit of detection was 20nM. Non-tumor bearing mice had undetectable levels of D2HG (not shown). The results for each mouse are summarized in Table 2.

Discussion

There is an unmet clinical need to identify and quantify biomarkers of GBM burden, progression and therapeutic responsiveness. Sampling directly within the tumor extracellular space via microdialysis may enhance our knowledge of these tumors and facilitate novel therapies.

Metabolomics is an increasingly important tool for understanding cancer mechanisms [19–21]. Most low grade gliomas harbor a mutation in IDH1 that produces 2HG [22]. Accumulation of 2HG has been shown to modify the epigenetic landscape of the gliomas, inducing DNA hypermethylation [23]. Metabolomic biomarkers have been proposed to help facilitate early diagnosis, predict disease progression, and quantify response to therapy [24]. Metabolic gradients may reflect regional tumor burden and heterogeneity, increasing need for precise probe placement and validation.

Here, we developed a workflow leveraging image-guided co-registration between MRI and CT using 3D-slicer to optimize and verify the accuracy of probe placement. Although image coregistration tools can be found in expensive proprietary clinical software packages, the methods outlined empower preclinical basic scientists to target catheter placement relative to individualized tumor anatomy with sub-millimeter accuracy and validate placement accuracy in each animal. Small animal MRI and CT facilities are increasingly available and medical research facilities. However, standard software lacks application features for import and co-registration of images from disparate sources. Critically, the now standardized NIFTI open file format was developed to convey not only imaging information but also spatial orientation that had been lacking from prior file formats. With use of the models provided herein and 3D slicer freeware, microdialysis with validated quantitative accuracy can be routinely accessible to those without specialized expertise. Even if a lab has no specific expertise related to methods of MRI or CT acquisition, relying on core facilities or collaborators, the NIFTI-exported files can be leveraged to perform the desired catheter targeting and registration. Through tumor targeting validation approach from the samples tested, the probe was typically present within the tumor (Table 1). With the exception of one sample, where the probe was confirmed to have missed the tumor altogether, all other tumors had intratumoral probe placements. We detected D2HG to provide proof-of-principle application for reliable detection of analytes sampled through the outlined approach (Table 2). None of the microdialysate had detectable L-2HG levels (data not shown), similar to previous in vivo orthotopic brain tumor model [18].

By combining mathematical modeling and quantifying in vitro release of D2HG from glioma cells, Linninger et al estimated that approximately 3 mM D2HG may be present in the interstitial fluid up to 2cm away from the center of an IDH-mutant glioma [25]. In this study, all mice were maintained for a fixed time point after tumor implantation. Given this suggested diffusion distance, and the relatively small murine xenografts sampled at a fixed time point, it is not surprising that D2HG was detected in the catheter positioned just outside the tumor volume. However, we found in pilot studies (not shown) that catheters placed contralateral to the tumor failed to consistently recover measurable D2HG. However, the method outlined can be the foundation to model microfabricated microdialysis probes with small cross-sections [26] to refine spatial resolution within the tumor and levels of D2HG.

We here performed targeted quantification of an established IDH-mutant glioma-associated biomarker. However, tumor-associated microdialysate may also have utility for biomarker discovery (ongoing work). While PDX 164 and 196 were obtained from the Brain Tumor PDX National Resource and is well characterized genomically and phenotypically [27], the approach described here may facilitate metabolic profiling of novel tumor lines and correlation with genetic phenotypes.

Potential Limitations

Current work is based on imaging and validation of MRI-guided probe placement within MRI visible tumor boundaries. As infiltrative lesions, the boundaries may be inherently indistinct. As such, defining parameters for acceptable intratumoral catheter placement may need to be tailored to the radiographic features of individual tumor lines. Placement of microdialysis catheters with the aid of pre-op MRI, post-op CT and image coregistration can enable optimal confidence in samples obtained and could be particularly valuable for individualized correlations on an animal-by-animal basis. However, the associated time and expense may not be practical for high throughput applications. Microdialysis probe implantation leads to a local tissue disruption-associated metabolic response that could add variability between animals and could partially overlap with certain tumor-associated biomarkers themselves. As such, appropriate controls are necessary to define which biomarkers are specific to the tumor of interest.

Alternate Applications

The methods described can be adapted to target any intracranial target; some of which may be variably displaced by the tumor requiring pre-operative MRI localization. For example, ventricular anatomy is typically markedly altered by the tumor but may be targeted to sample CSF biomarkers. The current methods utilize a metallic guide cannula and metal anchor screws. However, plastic screws and MRI-compatible microdialysis systems are available and could empower longitudinal real-time updates via serial MRIs to evaluate catheter location relative to therapy-altered tumor anatomy. While the models are provided for the commonly utilized Amuza microdialysis probes CX-I series, minor modifications to the models can be made to account for catheters of different length and diameter. Moreover, the workflow is fully compatible for placement of multiple catheters, such as bilateral microdialysis, wherein the contralateral hemisphere serves as a control.

Time Considerations

Intracranial implantations are typically performed using four injection jigs with two laboratory-trained staff, enabling up to sixteen mice to be implanted per hour. Once tumors are established, 3D T2-weighted MRI requires a 10-minute scan and associated set-up time. Given multiple manual steps even once accustomed to the 3D-slicer software workflow, tumor segmentation and probe targeting requires approximately 15 minutes per mouse. Surgical microdialysis probe implantation is typically performed with a single surgeon using one stereotactic jig apparatus requiring at least 15 minutes per mouse. CBCT scanning takes about 4 minutes per mouse. Image co-registration for validation of probe placement requires another 15 minutes per mouse.

Highlights.

• Instructions sufficient for first-time users targeting murine brain tumors

• Described methods leverage freely available software, 3D-slicer

• All needed templates and files provided for intracranial microdialysis

• Flexible adaptable workflow for catheter targeting and validation

Abbreviations:

GBM

glioblastomas

IDH-1

isocitrate dehydrogenase I

PDX

patient derived xenografts

2HG

2 hydroxyglutarate

MRI

Magnetic Resonance Imaging