Folate Receptor Overexpression in Human and Canine Meningiomas—Immunohistochemistry and Case Report of Intraoperative Molecular Imaging

Abstract

BACKGROUND

Meningiomas are well-encapsulated benign brain tumors and surgical resection is often curative. Nevertheless, this is not always possible due to the difficulty of identifying residual disease intraoperatively. We hypothesized that meningiomas overexpress folate receptor alpha (FRα), allowing intraoperative molecular imaging by targeting FRα with a near-infrared (NIR) dye.

OBJECTIVE

To determine FRα expression in both human and canine meningioma cohorts to prepare for future clinical studies. Present a case study of a meningioma resection with intraoperative NIR fluorescence imaging.

METHODS

Tissue samples of 27 human meningioma specimens and 7 canine meningioma specimens were immunohistochemically stained for FRα along with normal dura, skeletal muscle, and kidney tissue. We then enrolled a patient with a pituitary adenoma and tuberculum sella meningioma in a clinical trial in which the patient received an infusion of folate-linked, NIR fluorescent dye prior to surgery.

RESULTS

In the cohort of human meningiomas, 9 WHO grade I, 12 grade II, and 6 grade III tumors were identified. Eighty-nine percent of WHO grade I, 67% of grade II, and 50% of grade III tumors overexpressed FRα. In the 7 canine meningioma samples, 100% stained positively for FRα. Both human and canine normal dura from autopsy samples demonstrated no evidence of FRα overexpression. In the case study, the meningioma demonstrated a high NIR signal-to-background-ratio of 4.0 and demonstrated strong FRα immunohistochemistry staining.

CONCLUSION

This study directly demonstrates FRα overexpression in both human and canine meningiomas. We also demonstrate superb intraoperative imaging of a meningioma using a FRα-targeting dye.

ABBREVIATIONS

5-ALA

5-aminolevulinic acid

FRα

folate-receptor alpha

H-score

histo score

IRB

Institutional Review Board

MRI

magnetic resonance imaging

NIR

near-infrared

SBR

signal-to-background-ratio

Meningiomas are the most common benign brain tumor in adults, accounting for 33.8% of all primary central nervous system tumors diagnosed each year in the United States.¹ Most of these tumors are encapsulated and benign (WHO grade I) but may cause serious problems due to mass effect on critical structures. Atypical (WHO grade II) and malignant (WHO grade III) meningiomas account for approximately 5% to 20% of all meningiomas.^1–3 Surgical resection is the primary mode of treatment for symptomatic meningiomas given their benign nature. However, despite surgery and adjuvant radiation therapy, median time to radiographic progression is 12, 7, and 2 yr for WHO grade I, II, and III meningiomas respectively.^1,4

Still considered the gold standard for meningioma resection classification, the classic 1957 paper by Simpson,⁵The Recurrence of Intracranial Meningiomas After Surgical Treatment, details meningioma recurrence rates as well as a grading system for surgical resection. In short, the more extensive a resection, the less chance for recurrence. This surgical principle continues to guide the neurosurgeon's efforts intraoperatively. As such, various surgeons have attempted to use fluorescent dyes intraoperatively to aid in distinguishing neoplasm from normal brain parenchyma. The 5-aminolevulinic acid-induced protoporphyrin IX (5-ALA), a prodrug in the porphyrin family that facilitates tumor identification, has been of particular interest and was recently FDA-approved for glioma resection. Valdes et al⁶ observed fluorescence in 80% of the 15 meningioma patients.⁶ Both Morofuji et al⁷ and Puppa et al⁸ used 5-ALA to identify bone invasion during meningioma surgery. Despite these benefits, 5-ALA fluoresces in the visible light range, which has limited tissue penetration and is hindered by autofluorescence from surrounding tissue. On the other hand, near-infrared (NIR) fluorescence does not suffer from these disadvantages.

Folate is a necessary vitamin for nucleic acid biosynthesis, and thus the proliferative state of the cell determines the rate of folate internalization.⁹ Folate receptors are glycosylphosphatidylinositol linked plasma protein that mediate delivery of folate into cells.¹⁰ Folate-receptor alpha (FRα) is overexpressed in ovarian, renal, lung, pituitary, and breast cancers.^11,12 Importantly, FRα can internalize small molecules or protein conjugated to folate, making it an attractive receptor for chemotherapeutic drugs or fluorescent imaging agents.¹³ Indeed, van Dam et al¹⁴ first reported use of a folate-fluorescein dye as a means of identifying ovarian cancer intraoperatively, providing proof-of-principle and demonstrating potential benefit.

Meningiomas have been known for decades to express estrogen and progesterone receptors, but folate-receptor expression has not been well studied.¹⁵ Based on prior work by Ross et al¹⁶ on FRα and FRβ expression in tumors suggesting that meningiomas may be an excellent target for intraoperative molecular imaging, we hypothesized that meningiomas overexpress FRα. Here, we report our findings in ex vivo human and canine meningioma specimens as well as a case report of a surgical resection using NIR fluorescence in a human patient.

METHODS

Human Clinical Pathology Samples

A panel of previously resected meningioma specimens was chosen based on histology and study convenience. All patients were enrolled in Institutional Review Board (IRB) approved studies that allow for tissue banking of human specimens with subsequent future deidentified analysis. All surgeries were performed between 2011 and 2016. Demographic and surgical information was queried from the patient's electronic medical records. Cases were selected at random but in such a way as to increase the number of WHO grade II and III cases in our panel. The tumor specimens from each patient were chemically fixed and embedded in paraffin blocks at 5 μm thickness. These permanent sections were stained with hematoxylin and eosin staining. Immunohistochemistry against FRα was performed using murine monoclonal antibodies against the folate receptor (∼3.3 μg/mL, NCL-L-FRalpha, Leica, Wetzlar, Germany). Normal human dura from a previous resection was also stained. Human kidney and skeletal muscle (quadriceps muscle) served as positive and negative controls, respectively.

Canine Meningioma Pathology Samples

A panel of previously resected canine meningioma specimens were identified through the Pathology Service at the School of Veterinary Medicine at our institution. Specimens were formalin-fixed and embedded in paraffin blocks at 5 μm thickness. Sections were stained with hematoxylin and eosin. Immunohistochemistry was performed using a rabbit anti FOLR1/FRα antibody (∼1.3 μg/mL, LSBio #LS-B5727, Seattle, WA) diluted 1:750 in Antibody Diluent (003118, Life Technologies, Fredrick, Maryland). Normal canine dura was stained. In addition, normal canine kidney and normal skeletal muscle was stained as positive control and negative control, respectively.

Quantification of FRα Expression

In order to quantify the amount of FRα receptor expression, an H-score (or “histo” score) was calculated for each specimen. The H-score ranges from 0 to 300, with 300 implying strong staining in all cells of the specimen. An H-score is designed to give more relative weight to higher-intensity membrane staining in a given specimen. A single neuropathologist at our institution, blinded to all fluorescence data, scored all specimens.

H-Score Calculation for FRα

H-Score = [1 × (% cells 1+) + 2 × (% cells 2+) + 3 × (% cells 3+)].^16,17 Our cutoff of the H-score to define overexpression is currently arbitrary. The best way to determine the appropriate cutoff would be to determine threshold at which the surgeon could distinguish elevated signal as a function of increasing H-score. Unfortunately, this clinical study is not yet available.

Data Analysis

The data from 27 patients were analyzed using STATA 10^TM (StataCorp LLC, Lakeway Drive College Station, Texas). Linear Regression analysis was used to predict the FRα H-score. Independent variables tested included gender, age, body mass index (BMI), tumor diameter, fluid-attenuated inversion recovery sequence on magnetic resonance imaging (MRI), tumor location, history of prior radiation, Ki-67 index, and WHO grade. Image analysis was performed using VisionSense software (VSPlayer, v1.8.05.01, VisionSense, Philadelphia, Pennsylvania) and ImageJ (https://imagej.nih.gov/ij/; National Institute of Health (NIH), Bethesda, Maryland).

Human Case Example

We enrolled a patient with a somatotroph pituitary adenoma in an IRB-approved phase I, open-label pituitary imaging clinical trial using OTL38. OTL38 (OnTarget Laboratories, West Lafayette, Indiana) is a folate-analog linked to a NIR dye IR783.¹⁸ This prospective clinical study was not designed for patients with meningiomas but rather, pituitary adenomas. Preliminary results from this pituitary trial have been published.¹⁹ Because 1 patient also had a tuberculum sella meningioma, the extended endonasal approach provided access to both tumors in the same setting, thus allowing for intraoperative visualization of both the pituitary adenoma and WHO grade I meningioma.

Approximately 3 hr prior to the surgery, the patient received an intravenous 0.025 mg/kg infusion of the study drug over 45 min. A standard, endonasal, endoscopic trans-sphenoidal opening was performed by an otorhinolaryngologist. Upon exposure of the sella, the neurosurgeon employed a NIR-capable camera attached to a 4-mm endoscope to image the tumor before dural opening. The VisionSense Iridium^TM (VisionSense) camera's laser excitation light source is tuned to 785 nm, and a dual-light path and sensor design is employed so as to visualize both white light and NIR that can be superimposed in real time. All images with the VisionSense were recorded for posthoc analysis. To quantify the degree of fluorescence, the VisionSense software and ImageJ were used (NIH). Three to five points were placed on the fluorescing samples and surrounding normal tissue to generate a signal-to-background-ratio (SBR). As stated in the informed consent form, the extent of surgery was not modified based on the experimental NIR findings. Surgery proceeded with visible light only, and only upon completion of surgery was the NIR camera reintroduced in order to study margins.

RESULTS

Human Meningioma Specimens

A total of 27 meningioma specimens from patients who had previously undergone meningioma resection (18 females, 9 males; mean age 62 ± 13.3) were studied. Pathology revealed 9 WHO grade I, 12 WHO grade II, and 6 WHO grade III tumors (Table 1). The locations of the tumors were as follows: 10 convexity meningiomas, 10 parasagittal, 4 sphenoid wing, 2 olfactory groove, and 1 tuberculum sella. Four patients had undergone prior craniotomies for tumor resection. Three of those four patients also received radiation following the initial resection. The maximum diameter of the tumors ranged from 15.2 to 85.4 mm (mean: 24.1 mm). The range and mean for the Ki-67 index for the WHO grade I, II, and III meningiomas were 1 to 5 (mean: 2.7), 1 to 18 (mean: 8.5), and 10 to 40 (mean: 26), respectively.

TABLE 1.

Description of Patients—Demographics & Tumor Parameters

	ID	Age	Gender	BMI	Prior craniotomy (yr)	Prior radiation (yr)	Location	FLAIR (none, mild, significant)	Maximum tumor diameter (mm)
WHO Grade 1
	OTL-11	74	F	31.6			Tuberculum sella	Mild	30
	ICG-21	74	M	28			Convexity	Significant	85.4
	ICG-33	64	F	24.7			Sphenoid wing	None	23.9
	ICG-01	32	F	24.4			Olfactory groove	None	15.2
	ICG-513	46	F	34			Convexity	None	33
	ICG-44	52	M	31.9			Olfactory groove	None	42.7
	ICG-29	73	F	19			Parasagittal	Mild	39.8
	ICG-40	64	F	24.5			Convexity		45.2
	ICG-59	54	F	38.4			Sphenoid wing	Mild	24
WHO Grade 2
	ICG-03	54	F	23	6.8	4.2	Parasagittal	Significant	22.3
	101	66	F	22.8			Convexity	Mild	34.2
	102	58	F	24.2			Parasagittal	Significant	37.2
	103	34	M	27.4			Parasagittal	Significant	44.3
	ICG-42	66	F	30			Convexity	Significant	51.5
	104	73	M	26.3			Convexity	Significant	50.2
	105	68	M	37			Sphenoid wing	Significant	33.9
	ICG-53	68	F	36.7			Parasagittal	Significant	55.5
	106	48	M	24.4			Parasagittal	Significant	30.2
	ICG-55	60	F	35.4	10.5		Sphenoid wing	Significant	53.8
	107	74	M	23.8			Parasagittal	Significant	59.6
	108	54	F	33.6			Parasagittal	Significant	35.6
WHO Grade 3
	109	64	F	25.5	7.1	6.8	Convexity	Significant	30.5
	110	50	F	27.4			Parasagittal	Mild	28.9
	111	56	F	30.5	1, 1.4	1.2	Convexity	Significant	23.1
	112	89	M	27.1			Convexity	Mild	32.0
	113	80	F	33.1			Convexity	Mild	35.9
	114	74	M	36.5			Parasagittal	Significant	51.3

Folate-Receptor Immunohistochemistry Staining

Human kidney (positive control) had an h-score of 180 and human skeletal muscle (negative control) had an h-score of zero. Normal dura demonstrated no expression of FRα, with an H-score of zero (Figure 1). Consequently, we defined overexpression of FRα in meningiomas as any H-score greater than 10. Eight of the nine (89%) WHO grade I, 8 of the 12 (67%) WHO grade II, and 3 of the 6 (50%) WHO grade III tumors overexpressed FRα (Figure 2). The average H-scores for the WHO grade I, II, and III meningiomas stained were 89.4 ± 67.2, 53.3 ± 62, and 28.3 ± 46.7, respectively (Table 2). The 2 best predictors of FRα overexpression was WHO grade and BMI; increasing WHO grade resulted in a lower likelihood of FRα overexpression (P = .0575), while increasing BMI resulted in lower likelihood of FRα overexpression (P = .0567).

FIGURE 1.

FRα staining by WHO grade and dura. The first vertical column displays the hematoxylin and eosin stained specimens, and the second vertical column displays the corresponding FRα immunohistochemical stained specimens. A and B, WHO grade I meningioma with an FRα H-score of 165. C and D, WHO grade II meningioma with FRα H-score of 75. E and F, WHO grade III meningioma with FRα H-score of zero. G and H, Sample of human dura with FRα H-score of zero.

FIGURE 2.

Distribution of human meningioma FRα H-scores. A box plot of FRα H-score on the y-axis for meningiomas in each of the WHO grade categories (x-axis). Overall, higher grade meningiomas expressed lower levels of FRα and had more specimens that expressed no FRαm.

TABLE 2.

FRα H-scores in Human Meningiomas

ID	1+	2+	3+	FRα staining H-score	KI-67 proliferation index
Meningioma WHO grade I
OTL-11	30	60	10	180	2
ICG-21	10	25	35	165	2
ICG-33	20	30	20	140	1
ICG-01	40	40	0	120	4
ICG-513	30	30	5	105	5
ICG-44	20	15	0	50	4
ICG-29	25	0	0	25	3
ICG-40	20	0	0	20	1
ICG-59	0	0	0	0	3
Meningioma WHO grade II
ICG-03	20	35	30	180	5
101	10	30	20	130	18
102	40	30	10	130	2
103	25	25	0	75	15
ICG-42	50	0	0	50	10
104	30	0	0	30	15
105	20	2	0	24	5
ICG-53	15	0	0	15	1
106	0	0	1	3	5
ICG-55	2	0	0	2	2
107	0	0	0	0	12
108	0	0	0	0	12
Meningioma WHO grade III
109	25	25	15	120	40
110	30	0	0	30	10
111	20	0	0	20	20
112	0	0	0	0	40
113	0	0	0	0	25
114	0	0	0	0	21
Human dura	0	0	0	0	N/a

Seven canine meningioma specimens were immunohistochemically stained for FRα. All 7 (100%) of the canine meningiomas stained positively for FRα with H-scores ranging from 30 to 160 (mean: 96.4 ± 41.5; Table 3). The canine dura and skeletal muscle did not stain for FRα (H-score = 0), whereas the kidney specimen stained positive for FRα with an H-score of 190.

TABLE 3.

Canine FRα H-Scores

ID	1+	2+	3+	FRα staining H-score
B1220330	20	40	0	100
B1314771	10	20	20	110
B1319032	30	0	0	30
B1410969	15	50	0	115
B1412657	20	10	40	160
B1416273	80	10	0	100
B1606517	20	20	0	60
Canine dura	0	0	0	0
Canine kidney	40	30	30	190
Canine skeletal muscle	0	0	0	0

Case Report

A 74-yr-old woman presented with bitemporal hemianopsia, significant weight gain, diabetes mellitus, and enlarging interphalangeal joints. The patient's serum IGF-1 was 480 (reference range 34-245). MRI revealed a heterogeneous, hypoenhancing 1.0 × 1.0 × 0.8 cm mass in the right side of the pituitary gland, as well as a separate 2.2 × 3.0 × 2.0 cm extra-axial suprasellar mass along the tuberculum sellae, presumed to be a meningioma (Figure 3).

FIGURE 3.

Case example—intraoperative molecular imaging. Each vertical column displays a type of imaging performed by the camera system. The first column displays the visible light image only, the middle column displays the NIR image fused over a visible light image and the third column displays the near infrared only image. A-C, Preoperative MRI images showing a heterogeneous hypoenhancing 1.0 × 1.0 × 0.8 cm pituitary adenoma in the right side of the pituitary gland, as well as a separate 2.2 × 3.0 × 2.0 cm extra-axial suprasellar meningioma along the tuberculum sellae. D-F, Intraoperative imaging seen with the dura intact and not yet opened. The NIR signal can be seen localizing just under the dura at the presumed location of the tumor (E, F). G-I, Intraoperative images after the dura is opened; the meningioma is significantly brighter than the pituitary adenoma (H, I, top right corner). J-L, Intraoperative detection of fluorescent margin from the meningioma that was biopsied and later determined to tumor by pathology.

Intraoperatively, after removal of the sellar bone and the planum sphenoidale, the NIR camera was set at an appropriate distance that allowed both the meningioma and the adenoma to be captured in the same view. The adenoma fluoresced with an SBR of 2.4 and the meningioma fluoresced with an SBR of 4.0 with the dura intact for both (Figure 3). After the senior neurosurgeon resected the obvious portions of the adenoma, the NIR endoscope was used to visualize the area, revealing a small area with an SBR of 2.7. A specimen was resected from the area and pathology determined it to contain both adenoma and meningioma. Once the pituitary adenoma had been resected, the meningioma was gradually debulked. Once the senior neurosurgeon was satisfied with the resection, the NIR camera was used to establish whether there was any remaining fluorescence at the margins. An isolated area of SBR 3.1 appeared to be emanating from the inferior frontal lobe (Figure 3). This specimen was not believed to be neoplastic by the surgeon's impression but was biopsied; pathology revealed it to be positive for meningioma.

Immediately after resection, the excised specimens were imaged ex vivo, using an exoscope for NIR fluorescence. The adenoma sample fluoresced with an absolute signal of 95 units at a gain of 100% while the meningioma sample fluoresced with an absolute signal of 102 units at a gain of only 11%, for a gain-corrected signal of 927 units (ImageJ, NIH; Figure 4).

FIGURE 4.

Case example—FRα staining and ex vivo imaging. A, FRα immunohistochemical staining of the pituitary adenoma specimen demonstrating low H-score of 20. B, Ex vivo NIR imaging fused over visible light imaging of the pituitary adenoma specimen. The camera gain percent was 100% during this portion of imaging. C, FRα immunohistochemical staining of the meningioma specimen demonstrating high H-score of 180. D, Displays the back table, ex vivo NIR imaging fused over visible light imaging of the meningioma specimen. Importantly, the camera gain percent was only 11% during this portion of imaging.

Hematoxylin and eosin staining confirmed that the pituitary adenoma was consistent with gonadotroph adenoma. Immunohistochemistry for FRα on the pituitary samples revealed minimal expression of FRα, with an H-score of only 20. Conversely, the meningioma samples stained very intensely for folate receptor alpha, with an H-score of 180 (Figure 4). Based on current WHO criteria, this meningioma sample was consistent with WHO grade I benign meningioma.

DISCUSSION

FRα has the intrinsic ability to internalize folate conjugated to small particles and proteins. Thus, several strategies including anti-FRα-antibodies, drugs conjugated to folate, oncolytic viruses, nanoparticles, T-cells, and imaging compounds have been targeted at cancers overexpressing this receptor.¹³ Our group is interested in exploring the capabilities of targeting FRα for NIR imaging in certain intracranial tumors. We have previously demonstrated that pituitary adenomas often overexpress FRα and can be imaged with NIR imaging. In this study, we examined whether meningiomas also overexpress FRα. The level of FRα expression was first measured in human and canine meningioma specimens. Then, based on our findings, we conducted a proof-of-principle study in a patient with a meningioma and performed intraoperative, real-time visualization with our FRα-targeting NIR contrast agent.

Human Meningiomas Demonstrate Overexpression of Folate-Receptor

We analyzed the FRα expression in 27 meningioma specimens at our institution. Overall, 78% of the tumors stained for FRα. We found that 89% of the WHO grade I and 67% of WHO grade II meningiomas overexpressed FRα. We found great heterogeneity in the expression of the receptor across all tumors, with H-scores ranging from 0 to 180 (Figure 2). Using linear-regression analysis, the 2 best predictors of FRα-overexpression were WHO grade and BMI. The positive association between BMI and meningioma has been established in prior studies; the suggested mechanism is that adipose tissue may convert androgens to estrogen via aromatase, driving meningioma growth.¹⁹ We were encouraged by the staining seen in the grade I and II tumors. However, grade III meningiomas failed to show the same pattern. This lack of FRα expression in the WHO grade III malignant meningiomas is unfortunate as this is where we envisioned NIR imaging to be particularly useful in aiding with resection and providing longer progression-free survival. It is not clear why the higher grade meningiomas show reduced FRα expression, but we hypothesize that aggressive tumors demonstrate less differentiation than their more benign variants, leading to reduced FRα expression. With almost all grade I and II meningiomas overexpressing FRα, we believe these tumors would be excellent targets for FRα-targeted NIR imaging. In addition, the dura and normal brain parenchyma demonstrate undetectable FRα expression, thus making background signal less likely.^20,21

Ross et al¹⁶ quantitated folate receptors on the surface of meningiomas (along with other normal and pathological tissues) by looking at mRNA levels.¹⁶ They then obtained crude plasma membrane preparations from these tissues, subjected them to low pH conditions to remove endogenously bound folates and assayed them for specific binding to [³H]-folate. They found the 6 meningiomas studied to all express folate receptors in varying amounts. The level of FRα mRNA levels expressed in arbitrary units as compared to the negative control ranged from 1.9-77.7. They then looked at the ratio of FRβ mRNA levels to FRα mRNA levels and found the ratio varied from 0.2 to 14.1. The expression of [³H]-folate binding proteins in the meningiomas was 21.0 pmol/mg protein. Thus, they demonstrated a correlation between the level of folate receptor mRNA quantified and the active synthesis of both isoforms into functional folate receptors. Of note, the WHO grade for the 6 meningiomas studied was not given. In agreement with our study, Ross et al¹⁶ found heterogeneous folate receptor expression across the meningiomas studied.

Feasibility of a Canine Model for Meningioma NIR Imaging

We next demonstrated that 7 out of 7 canine meningiomas overexpressed FRα. In comparison to the human meningiomas tested, the canine tumors H-scores averaged slightly higher than the human WHO grade I tumors (96.4 vs 89.4). We also stained a section of normal canine dura for FRα that was negative just as with the human dura. Thus, we were able to demonstrate that canine meningiomas behave similarly to human meningiomas, at least with respect to FRα expression in the tumor and surrounding tissue. This is valuable, as canine models to study NIR fluorescence imaging in meningiomas is now feasible for future studies.

Intraoperative Visualization of a Meningioma using NIR Fluorescence

Finally, we enrolled a patient with both a pituitary adenoma and a meningioma into our ongoing clinical study investigating the use of FRα-targeted NIR imaging in pituitary adenomas. Prior literature has suggested that pituitary adenomas, particularly nonfunctional adenomas, overexpress FRα.²² This patient happened to also have a tuberculum sellae meningioma, which was scheduled for resection along with the pituitary adenoma, allowing us to observe the utility of FRα-targeted NIR imaging in meningiomas.

The somatotroph adenoma in this patient showed minimal expression of FRα, with an H-score of 20. Accordingly, the adenoma demonstrated weak NIR signal (95 absolute units of fluorescence vs the 927 units for the meningioma). The meningioma, which showed a much stronger signal both in vivo and ex vivo, turned out to have a very high expression of FRα with an H-score of 180. After standard resection, we detected an area of fluorescence in the resection cavity. This specimen was believed to be non-neoplastic tissue by the surgeon's impression but was biopsied and confirmed by pathology to be meningioma. On postoperative MRI, a small area of residual neoplasm was detected, consistent with the area of fluorescence we had seen. Thus, we demonstrated that FRα-targeted NIR imaging has a higher sensitivity for neoplastic tissue than the surgeon's impression alone, which is in line with our previous findings in pituitary adenomas. If this finding is replicated in future studies with a larger number of subjects, NIR imaging could potentially enhance intraoperative detection of residual neoplasm, leading to greater gross-total-resection rates and improved patient outcomes. Furthermore, NIR imaging can easily be incorporated into existing operating suites and workflow, unlike intraoperative MRI or other potential methods of increasing tumor sensitivity, by the simple addition of NIR excitation/sensor modules to existing microscopes in many cases or by using a dedicated exoscope/endoscope setup. Thus, NIR imaging has great potential and should be studied further in intracranial tumors.

Limitations

Our study is limited by several factors. This study was designed to examine FRα expression in meningiomas with the larger objective of investigating the use of FRα-targeted NIR imaging to aid in resection. The study involved a relatively small number of patients and the number of WHO grade III samples was limited by the total number of these specimens at our institution.

In addition, this study did not involve immunohistochemistry for FRβ, which may also be expressed in meningiomas. It is possible that OTL38 could cross-react with FRβ, which should be investigated in future studies to avoid potential confusion.

Finally, with the endoscopic approach, there is a learning curve to proper usage of NIR fluorescence. With exoscopes, imaging is simple. However, we have observed that the VisionSense endoscope tends to over-estimate fluorescence near the center of its view compared to the periphery (manuscript currently undergoing review). This can cause false-positive NIR signal reading, as we have described previously in pituitary adenomas.²³ Thus, a certain distance must be maintained between the endoscope and the tumor in order to accurately measure its NIR fluorescence. The proper distance can be estimated by maintaining the medial-opticocarotid-recess spanning less than one-third of the screen.

CONCLUSION

Complete resection of meningiomas remains the gold standard for surgical treatment of meningiomas. Intraoperative molecular imaging techniques to improve surgery for meningiomas is not yet currently available, but based on this preliminary study, we suggest that intraoperative imaging of meningiomas is feasible using folate-receptor-targeting NIR fluorophores.

Because normal dura appears not to express FRα, while 89% of WHO grade I meningiomas and 67% of grade II do, we believe this technique can provide strong tumor/nontumor contrast in real-time. This technique may provide value in neurosurgical patients at high risk of meningioma recurrence.

Disclosures

This work was partially supported by the National Institutes of Health R01 CA193556 (SS), Institute for Translational Medicine and Therapeutics of the Perelman School of Medicine at the University of Pennsylvania (JYKL), and National Center for Advancing Translational Sciences of the National Institutes of Health under Award Number UL1TR000003 (JKYL). Dr Lee owns stock options in VisionSense^TM. The other authors have no personal, financial, or institutional interest in any of the drugs, materials, or devices described in this article.

COMMENT

Meningiomas are tumors of the central nervous system presumably derived from arachnoid cap cells of the leptomeninges and with possible epithelial and mesenchymal origins. The wide histological and genetic diversity of these tumors bears testimony to that. Abnormality of the NF-2 gene on chromosome 22 is the most frequent change associated with convexity meningiomas but inactivation of a host of other tumor suppressor genes and activation of proto-oncogenes contribute to meningioma initiation and progression. Surgical removal of a meningioma and its dural base and any involved bone remains the optimal strategy to treat these tumors. Single fraction or fractionated radiation therapy also has a role in the management of surgically inaccessible or progressive higher-grade meningiomas. Most convexity meningiomas are easily accessible; cranial base or other deeper location tumors are more challenging but skilled surgeons can inflict a significant removal in almost half these cases as well. Visualization of the tumor itself is not hard; it has a fleshy yellowish-brown appearance in most cases that is distinct from surrounding brain and frequently an arachnoid plane around the tumor lends itself to a better resection. The involved dura and bone, which is frequently hyperostotic, is a different challenge. While visual inspection of the dura and intraoperative neuronavigation is helpful in guiding the extent of dural resection, a fluorescent dye that can stain abnormal dura would be ideal.

For neurosurgeons, intraoperative fluorescent imaging provides objective, real-time visualization of diffuse or encapsulated intracranial tumors during surgery. Judiciously used, it allows for more complete resection of the tumor - and possibly better outcome. Passive fluorescent probes such as fluorescein sodium or indocyanine green (ICG) or metabolic probes such as 5-aminolevulinic acid (5-ALA) have emerged as the predominant agents used for brain tumor resections although numerous other fluorophores await development and approval.¹ Most of these probes have the ability to accumulate in tumor tissue - large molecular weight or non-lipophilic probes rely on defects in the blood brain barrier to facilitate their accrual. The optical modalities are important as well - photons with a longer wavelength in the near-infrared spectrum have greater tissue penetration and sensitivity to tumor tissue.¹ For intracranial neoplasms, 3 predominant fluorescent contrast agents have emerged; fluorescein sodium, ICG, and 5-ALA. We have had cautious optimism with the use of fluorescein sodium in the resection of malignant gliomas, meningiomas, and metastatic brain tumors - as previously reported by other groups.² The agent is easily administered by intravenous injection following induction of anesthesia and visualization under the yellow-560 module facilitates tumor visualization as a bright-green fluoroscence. Specificity and quantification of the degree of fluorescence remains a challenge and can be affected by various factors such as absorption, scatter, anisotropy, and autofluoroscence.¹

This manuscript describes a targeted probe that was being evaluated for the resection of pituitary adenomas and that was incidentally found to have a use for meningioma resection. The folate-receptor alpha (FRα) probe (OTL-38), described in this manuscript, is a targeted probe with the potential to isolate neoplastic meningothelial cells. OTL-38 has been used with ovarian and lung cancer resections and is a targeted agent with excellent penetration and low local tissue autofluorescence.^3,4 The authors expanded their study to a tissue bank of human and canine meningiomas and report robust FRα expression in benign WHO grade 1 cranial meningiomas, while FRα expression was lacking in normal dura. They suggest this might serve as a novel target for intraoperative visualization of the tumor and its dural base facilitating more complete resections of intracranial, and potentially spinal, meningiomas.

As such, there does not appear to be a clear linkage between FRα expression and the precise embryological lineage of neoplastic cell lines. It is seen in benign and malignant neoplasms of ectodermal or mesodermal origin - thus, it may simply be a reflection of dysregulated cellular metabolism. FRα is overexpressed in pituitary tumors, and certain glioma cell lines.⁵ These authors have previously demonstrated the ability to image pituitary adenoma FRα expression with near-infrared imaging.⁶ That has great potential benefit particularly in small ACTH-secreting microadenomas in Cushing's disease or to facilitate more complete resections of larger macroadenomas. For meningiomas, the World Health Organization (WHO) grade 1 tumors appear to have the best expression of FRα. One would expect increased FRα expression with increasing grades of meningiomas but WHO grade II and III meningiomas in this series had progressively less expression. Also, no clear correlation was noted between FRα expression and meningioma subtypes although meningothelial and transitional meningiomas were predominant in this series. Whether expression of FRα has the same prognostic inference as the presence of progresterone receptors in meningiomas remains to be seen. It does however, have the potential to facilitate better resections of meningiomas and numerous studies have validated the impact of complete surgical resection on reducing the progression or recurrence-free survival and overall survival of patients with intracranial meningiomas. If an agent can be easily administered and has the ability to improve visualization of the tumor target, facilitate resection and impact outcome without a significant risk of side-effects, then clearly it is worthy of consideration - OTL-38 appears to meet these criteria. It will likely mature over time with gathered experience and neurosurgeons will determine for themselves whether its impact on the quality of tumor resection is significant.

Daniel M. Heiferman

Kevin N. Swong

Vikram C. Prabhu

Douglas E. Anderson

Maywood, Illinois