Intraoperative confocal laser endomicroscopy: prospective in vivo feasibility study of a clinical-grade system for brain tumors

Irakliy Abramov The Loyal and Edith Davis Neurosurgical Research Laboratory, Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;
Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Irakliy Abramov in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Marian T. Park The Loyal and Edith Davis Neurosurgical Research Laboratory, Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Marian T. Park in
Current site
Google Scholar
PubMed
Close
 BS
,
Evgenii Belykh Department of Neurosurgery, Rutgers New Jersey Medical School, Newark, New Jersey

Search for other papers by Evgenii Belykh in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Alexander B. Dru The Loyal and Edith Davis Neurosurgical Research Laboratory, Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Alexander B. Dru in
Current site
Google Scholar
PubMed
Close
 MD
,
Yuan Xu The Loyal and Edith Davis Neurosurgical Research Laboratory, Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;
Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Yuan Xu in
Current site
Google Scholar
PubMed
Close
 MD
,
Timothy C. Gooldy Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Timothy C. Gooldy in
Current site
Google Scholar
PubMed
Close
 MD
,
Lea Scherschinski Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Lea Scherschinski in
Current site
Google Scholar
PubMed
Close
 MD
,
S. Harrison Farber Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by S. Harrison Farber in
Current site
Google Scholar
PubMed
Close
 MD
,
Andrew S. Little Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Andrew S. Little in
Current site
Google Scholar
PubMed
Close
 MD
,
Randall W. Porter Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Randall W. Porter in
Current site
Google Scholar
PubMed
Close
 MD
,
Kris A. Smith Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Kris A. Smith in
Current site
Google Scholar
PubMed
Close
 MD
,
Michael T. Lawton Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Michael T. Lawton in
Current site
Google Scholar
PubMed
Close
 MD
,
Jennifer M. Eschbacher Department of Neuropathology, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix, Arizona; and

Search for other papers by Jennifer M. Eschbacher in
Current site
Google Scholar
PubMed
Close
 MD
, and
Mark C. Preul The Loyal and Edith Davis Neurosurgical Research Laboratory, Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;
Department of Neurosurgery, Barrow Neurological Institute, St. Joseph’s Hospital and Medical Center, Phoenix;

Search for other papers by Mark C. Preul in
Current site
Google Scholar
PubMed
Close
 MD
Free access

OBJECTIVE

The authors evaluated the feasibility of using the first clinical-grade confocal laser endomicroscopy (CLE) system using fluorescein sodium for intraoperative in vivo imaging of brain tumors.

METHODS

A CLE system cleared by the FDA was used in 30 prospectively enrolled patients with 31 brain tumors (13 gliomas, 5 meningiomas, 6 other primary tumors, 3 metastases, and 4 reactive brain tissue). A neuropathologist classified CLE images as interpretable or noninterpretable. Images were compared with corresponding frozen and permanent histology sections, with image correlation to biopsy location using neuronavigation. The specificities and sensitivities of CLE images and frozen sections were calculated using permanent histological sections as the standard for comparison. A recently developed surgical telepathology software platform was used in 11 cases to provide real-time intraoperative consultation with a neuropathologist.

RESULTS

Overall, 10,713 CLE images from 335 regions of interest were acquired. The mean duration of the use of the CLE system was 7 minutes (range 3–18 minutes). Interpretable CLE images were obtained in all cases. The first interpretable image was acquired within a mean of 6 (SD 10) images and within the first 5 (SD 13) seconds of imaging; 4896 images (46%) were interpretable. Interpretable image acquisition was positively correlated with study progression, number of cases per surgeon, cumulative length of CLE time, and CLE time per case (p ≤ 0.01). The diagnostic accuracy, sensitivity, and specificity of CLE compared with frozen sections were 94%, 94%, and 100%, respectively, and the diagnostic accuracy, sensitivity, and specificity of CLE compared with permanent histological sections were 92%, 90%, and 94%, respectively. No difference was observed between lesion types for the time to first interpretable image (p = 0.35). Deeply located lesions were associated with a higher percentage of interpretable images than superficial lesions (p = 0.02). The study met the primary end points, confirming the safety and feasibility and acquisition of noninvasive digital biopsies in all cases. The study met the secondary end points for the duration of CLE use necessary to obtain interpretable images. A neuropathologist could interpret the CLE images in 29 (97%) of 30 cases.

CONCLUSIONS

The clinical-grade CLE system allows in vivo, intraoperative, high-resolution cellular visualization of tissue microstructure and identification of lesional tissue patterns in real time, without the need for tissue preparation.

ABBREVIATIONS

CLE = confocal laser endomicroscopy; FNa = fluorescein sodium; NPV = negative predictive value; PPV = positive predictive value; RBC = red blood cell; ROI = region of interest; TSP = surgical telepathology software platform.

OBJECTIVE

The authors evaluated the feasibility of using the first clinical-grade confocal laser endomicroscopy (CLE) system using fluorescein sodium for intraoperative in vivo imaging of brain tumors.

METHODS

A CLE system cleared by the FDA was used in 30 prospectively enrolled patients with 31 brain tumors (13 gliomas, 5 meningiomas, 6 other primary tumors, 3 metastases, and 4 reactive brain tissue). A neuropathologist classified CLE images as interpretable or noninterpretable. Images were compared with corresponding frozen and permanent histology sections, with image correlation to biopsy location using neuronavigation. The specificities and sensitivities of CLE images and frozen sections were calculated using permanent histological sections as the standard for comparison. A recently developed surgical telepathology software platform was used in 11 cases to provide real-time intraoperative consultation with a neuropathologist.

RESULTS

Overall, 10,713 CLE images from 335 regions of interest were acquired. The mean duration of the use of the CLE system was 7 minutes (range 3–18 minutes). Interpretable CLE images were obtained in all cases. The first interpretable image was acquired within a mean of 6 (SD 10) images and within the first 5 (SD 13) seconds of imaging; 4896 images (46%) were interpretable. Interpretable image acquisition was positively correlated with study progression, number of cases per surgeon, cumulative length of CLE time, and CLE time per case (p ≤ 0.01). The diagnostic accuracy, sensitivity, and specificity of CLE compared with frozen sections were 94%, 94%, and 100%, respectively, and the diagnostic accuracy, sensitivity, and specificity of CLE compared with permanent histological sections were 92%, 90%, and 94%, respectively. No difference was observed between lesion types for the time to first interpretable image (p = 0.35). Deeply located lesions were associated with a higher percentage of interpretable images than superficial lesions (p = 0.02). The study met the primary end points, confirming the safety and feasibility and acquisition of noninvasive digital biopsies in all cases. The study met the secondary end points for the duration of CLE use necessary to obtain interpretable images. A neuropathologist could interpret the CLE images in 29 (97%) of 30 cases.

CONCLUSIONS

The clinical-grade CLE system allows in vivo, intraoperative, high-resolution cellular visualization of tissue microstructure and identification of lesional tissue patterns in real time, without the need for tissue preparation.

In Brief

The authors evaluated the feasibility of using the first clinical-grade confocal laser endomicroscopy (CLE) system using fluorescein sodium for intraoperative imaging of brain tumors. CLE was safe and feasible for acquisition of noninvasive digital biopsies. The diagnostic accuracy, sensitivity, and specificity of CLE compared with those of permanent histopathological sections were 92%, 90%, and 94%, respectively. CLE shows high-accuracy real-time feedback for optically interrogated tissue, enabling improved intraoperative decision-making, thereby delivering significant advantages over traditional histopathological sections.

Confocal laser endomicroscopy (CLE), a handheld imaging technology used with fluorescent contrast agents, allows real-time intraoperative visualization of cellular-level tissue histoarchitecture. CLE facilitates rapid histopathological interrogation of abnormal tissue without the need to resect and process tissue.13 Such optical biopsies produce a series of digital histological images in seconds, which can supplement conventional intraoperative frozen-section biopsies and improve the positive yield of tissue biopsy.3

Our previous intraoperative in vivo study used an experimental Optiscan 5.1 CLE system (Optiscan, Carl Zeiss Meditec AG) with fluorescein sodium (FNa).3 This first-generation CLE system was validated in animals4 and in ex vivo and in vivo human brain tumor surgery;3 it was upgraded with enhanced imaging system features and sterility measures for routine neurosurgery.5,6 The clinical-grade CLE system (Convivo, Carl Zeiss Meditec) was evaluated in animal models5,7,8 and intraoperative ex vivo human brain tissue studies911 and was cleared by the US FDA for intracranial neurosurgical procedures.12 Although the clinical-grade system showed diagnostic and morphological concordance between CLE images and histopathology analysis, imaging was performed ex vivo.911 This is the first clinical study to evaluate the feasibility and usability of the first FDA-cleared clinical-grade CLE system for in vivo intraoperative use in human brain tumor surgery.

Methods

Study Design

A prospective feasibility study of the Convivo CLE system was conducted by Barrow Neurological Institute at St. Joseph’s Hospital and Medical Center, Phoenix, Arizona, with approval by the local institutional review board. This study was not designed to endorse the CLE system. Patients undergoing brain tumor surgery were eligible to enroll and provided informed voluntary consent. Exclusion criteria were renal failure, pregnancy or breastfeeding, age < 18 years, previous adverse reaction or hypersensitivity to FNa, and inability to provide informed consent.

The study protocol included standard microsurgical techniques for tumor removal, including image-guided navigation and operating microscope use. A dose of 5 mg/kg FNa was administered at the surgeon’s request within 5 minutes before CLE imaging. No intraoperative clinical decision was made using CLE imaging. The number of study patients was capped at 30, which was considered adequate for the study goals. The first primary end point was safety, defined as the number of adverse effects (tissue injury or postoperative infection) during CLE use that does not exceed the historical level for similar patients in whom CLE was not used. The second primary end point was feasibility, the ability of neurosurgeons to obtain noninvasive CLE digital biopsies in 90% of the patients. The secondary end points were 1) the ability of the neuropathologist to interpret the CLE biopsies as abnormal or nonabnormal in > 80% of cases, 2) the ability of the neurosurgeon to obtain interpretable images within 10 minutes of starting CLE, and 3) if the neuropathologist’s CLE interpretation matched the histopathological interpretation in ≥ 80% of tissue biopsies.

Image Acquisition

The CLE tower was positioned in the operating room with the monitor facing the neurosurgeon, and the CLE probe was inserted into the sterile sheath. The neurosurgeon moved the probe tip slowly across the lesion at different locations, interrogating the tissue and acquiring images from the regions of interest (ROIs). Three ROIs (lesion surface, core, and margin) were assessed during image acquisition for each case.

Optimal CLE image-acquisition parameters are detailed in a previous publication.5 The CLE system uses a 488-nm wavelength excitation laser light and limits the maximum laser power to 1 mW. Laser power could be adjusted from 0% to 100% during imaging; 50% power was most commonly used. The gain was maintained at a value of 2400 (range 1800–3000). The CLE system includes long-pass, band-pass, and reflection fluorescence filters. The 515- to 577-nm band-pass filter provided the best quality images, along with automatic image brightness adjustment. Of the 4 imaging speed and quality options, low-speed/high-resolution scanning (1920 × 1080 pixels, 1.3 sec/frame) was most frequently used for image recording. The CLE detector signal was synchronized with the laser light to reconstruct images parallel to the tissue surface with a 475 × 267–µm field of view. Optimal imaging depth was determined after CLE initiation. Imaging depth adjustment and recording start and stop were controllable using a foot pedal or the monitor’s touchscreen. The probe was coregistered with an image-guided navigation system, which allowed documentation of the optical and tissue biopsy site locations.

Study Personnel and Use Assessment

Four neurosurgical teams performed the surgeries, with each team including 1 attending neurosurgeon who underwent CLE training and was the predominant CLE user. Residents inexperienced with CLE occasionally used it with supervision. Neurosurgeons viewed the CLE images on the monitor as they were obtained, while the system recorded the images. A recently developed surgical telepathology software platform (TSP)13,14 was available for the last 11 cases, which provided real-time intraoperative neuropathology consultation. The user experience and usability of the CLE system were assessed using a 5-point scoring system (1–5: lowest to highest).

Tissue Biopsy Acquisition

Tissue samples were obtained from each patient, with the operating neurosurgeon determining the number of tissue biopsies. Conventional frozen-section techniques and standard H&E staining and assessments were employed. A frozen-section biopsy was not deemed necessary in every case.

CLE Image Analysis and Interpretation

For each case, 2 neurosurgeons not involved in the surgery reviewed and classified the CLE images as interpretable (images with identifiable histological features) or noninterpretable (images without identifiable histological features or with motion artifacts). The numbers of interpretable and noninterpretable images were quantified and analyzed according to imaging location (including depth), histopathological type, and duration of CLE. Interpretable images were reviewed by 1 neuropathologist and 2 neurosurgeons experienced in CLE imaging. Interpretable images were then descriptively and quantitatively compared with corresponding frozen and H&E-stained sections. "Interpretable" was defined as recognition of similar histological features in the CLE image and corresponding H&E slide. Diagnostic accuracy, specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV) of CLE images were calculated using frozen and histologically stained sections as standards. CLE biopsies with lesions were regarded as positive CLE test results. CLE biopsies without lesions included true-negative, reactive brain tissue, or noninterpretable CLE biopsies and were regarded as negative CLE results.

Statistical Analysis

Statistical analyses were performed using GraphPad Prism 9 (GraphPad Software, Inc.). Continuous variables are presented as mean (SD), and categorical variables are presented as count and percentage. The numbers and percentages of interpretable images of various lesion types and imaging locations were compared using the Mann-Whitney U-test. Specificities and sensitivities were calculated using the standard formulas. The Spearman correlation coefficient was used to assess the association between the length of time of CLE usage and the percentage or number of interpretable images; p < 0.05 was considered significant.

Results

Descriptive Analysis and CLE Safety

Thirty patients (13 men and 17 women) were consecutively enrolled (May 2020–June 2021). The mean patient age at presentation was 54.6 years (range 23–84 years). Preoperative MRI identified 26 contrast-enhancing and 5 nonenhancing lesions (1 patient had 2 lesions). Thirteen primary and recurrent gliomas, 4 treatment-related reactive brain tissue from primary brain tumors after radiotherapy or chemotherapy, 5 meningiomas, 3 brain metastases, and 6 other tumors were identified with permanent histology (Table 1). No patient experienced adverse effects from FNa administration or intraoperative CLE use, and none had postoperative wound infection.

TABLE 1.

Distribution of tumor histology among 31 tumors from 30 patients included in the study

Tumor CategoryGliomasMeningiomasOther
Primary brain tumors (n = 24)
 WHO grade IPilocytic astrocytoma (1), subependymoma (1)Meningothelial meningioma (3)Pineocytoma (1), hemangioblastoma (1), acoustic schwannoma (1), choroid plexus papilloma (1), perineurioma (1), mature teratoma (1)
 WHO grade IIMeningothelial meningioma (1), fibrous meningioma (1)
 WHO grade IIIAnaplastic astrocytoma (3), anaplastic oligodendroglioma (1)
 WHO grade IVRecurrent glioblastoma (5), primary glioblastoma (2)
Brain metastases (n = 3)Metastases of breast adenocarcinoma (1), renal cell carcinoma (1), lung adenocarcinoma (1)
Reactive brain tissue (n = 4)After primary tumor resection, radiotherapy, & chemotherapy (4)

CLE Image Acquisition

The assessment of CLE feasibility showed that 26 (65%) of 40 ratings were ≥ 3 (Fig. 1). The neurosurgeons obtained interpretable CLE images from all 30 patients: 335 ROIs were assessed; 234 (70%) were located deeply and 101 (30%) superficially. Overall, 10,713 CLE images from 335 ROIs were acquired (mean 357 [156] images per patient), and 4896 images (46%) were interpretable (mean 163 [123] per patient) (Fig. 2A).

FIG. 1.
FIG. 1.

Heat map of neurosurgeons’ (NS) ratings of CLE feasibility. The CLE system was rated on a 5-point scale, where 1 was the worst score and 5 was the best. CLE was rated by the 2 attending neurosurgeons, who were regarded as authoritative and experienced enough with CLE to assess its feasibility. Conditional formatting was applied on a scale from worst score (red) to best score (green). Asterisks indicate the neurosurgeon’s repeated evaluation of CLE technology after the conclusion of the study. Figure is available in color online only.

FIG. 2.
FIG. 2.

Graphs showing interpretation and correlation between the number or percentage of interpretable images or other factors. A: Number of interpretable and noninterpretable images obtained for each case. B–D: Scatterplots. Solid lines represent linear regression, and dashed lines indicate 95% CIs. Correlation between the number of interpretable images and progression of the study from the first case to the last case (B). Correlation between percentage of interpretable images per case and the cumulative length of time of CLE usage in the operating room throughout the study (C). Correlation between the percentage of interpretable images and time spent per case (D). E: Scatterplot showing correlation between percentage of interpretable images and the consecutive number of cases performed by each neurosurgery team (NT) during the study. F–H: Bar graphs. Error bars indicate SD. Comparison of mean percentage of interpretable images acquired from superficially and deeply located lesions (F). Mean number of CLE images by type of diagnosis (G). Comparison of mean percentage of interpretable images acquired from different types of diagnosis (H). *p < 0.05; ***p < 0.001. Figure is available in color online only.

The first interpretable image was obtained within a mean of 6 (10) images. The mean time to interpretable image acquisition was 5 (13) seconds after CLE initiation. The number of interpretable images per patient improved as the study progressed (r = 0.55, p = 0.002) (Fig. 2B). Similarly, we found positive correlations between the percentage of interpretable images and cumulative duration of CLE use (r = 0.68, p < 0.001) (Fig. 2C) and between the percentage of interpretable images and duration of CLE use per case (r = 0.48, p < 0.001) (Fig. 2D).

Neurosurgical team 1 performed 3 cases totaling 18 minutes of CLE use; team 2 performed 5 cases totaling 29 minutes of CLE use; team 3 performed 7 cases totaling 35 minutes of CLE use, and team 4 performed 15 cases totaling 116 minutes of CLE use. The mean duration of CLE use per case was 7 minutes (range 3–18 minutes). The number of interpretable images and the sequential number of cases performed were positively correlated only for team 4 (r = 0.82, p < 0.001) (Fig. 2E).

Deeply located ROIs were associated with a higher percentage of interpretable images compared with superficial ROIs (53% [SD 32%] vs 40% [27%], p = 0.02; Fig. 2F). The mean number of images taken did not differ significantly by lesion type (p = 0.46) (Fig. 2G). Similarly, the time to the first interpretable image did not differ by lesion type (p = 0.35). However, meningiomas had a lower mean percentage of interpretable CLE images (26% [36%]) compared with gliomas (57% [38%], p < 0.001), reactive brain tissue (51% [42%], p = 0.01), metastases (60% [32%], p < 0.001), and other lesions (49% [40%], p = 0.01) (Fig. 2H).

For the last 11 cases with neuropathology consulting, neurosurgeons rated the TSP highly for communication, guidance, efficiency, and coordination with the neuropathologist (Fig. 1).

CLE Image Interpretation and Comparison With Conventional Histological Sections

Fifty-five conventional tissue biopsies were acquired from 30 patients (mean 2 biopsies per patient; range 1–3) and compared with the corresponding 60 CLE optical biopsies (Fig. 3). The mean number of images acquired per tissue biopsy was 114 (SD 63). Of 55 tissue biopsies acquired, 18 were processed as frozen sections and 55 as permanent sections.

FIG. 3.
FIG. 3.

Histological features identified on CLE image (left) and corresponding H&E-stained histological specimen (right) from the same patient and lesion for various pathology types. CLE images were high quality (even from inexperienced neurosurgeons) and, in many cases, showed structural features not seen on conventional histology slides. A: Hypercellularity with atypical cells of different sizes indicative of infiltrative high-grade glioma. Bright spots likely correspond to cell surface vesicles on astrocytes, as seen with electron microscopy.24 B: Nest of atypical cells with collagen background indicative of meningioma tissue. C: Atypical tumor cells with abundant cytoplasm and nucleus on the periphery (white outlines) indicative of anaplastic astrocytoma. Dashed lines indicate a blood vessel. Transiting erythrocytes are seen in laminar form within the vessel. Erythrocytes have a laminar appearance in the CLE image because their flow is faster than the CLE scanning rate. A larger cell, interpreted as a leukocyte, is seen within the vessel (arrowhead). Arrows indicate erythrocyte infiltration. D: Numerous atypical cells (arrowheads) indicative of pineocytoma. E: Atypical cells (arrowheads) invading the blood vessel (dashed lines) indicative of glioblastoma. F: Cellular tumor and scattered atypical larger cells in a background of erythrocytes (smallest cells) correspond to metastasis of breast adenocarcinoma. Arrowheads indicate atypical mitotic cells. G: Hypocellular regions with neurodegeneration represented by corpora amylacea (white arrowheads) and axons (black arrowheads) indicative of reactive brain tissue. H: Hypocellular regions of reactive brain tissue with abundant macrophages (dashed outline). White arrowheads indicate axons. Bar = 100 µm (A–G); 200 µm (H). Figure is available in color online only.

The neuropathologist matched histology images with the CLE images in 29 (97%) of 30 cases. However, the histology and CLE images of perineurioma could not be well matched. Diagnostic accuracies for CLE versus frozen and permanent H&E sections were 94% and 92%, respectively (Table 2). CLE sensitivity and specificity versus frozen sections were 94% and 100%, respectively. CLE versus permanent histological sections had a 90% sensitivity and 94% specificity. Subgroup analyses of the PPV of CLE for all samples versus frozen sections, permanent sections, and gliomas confirmed the high PPV of CLE optical biopsies: 100%, 97%, and 100%, respectively. Similarly, a high NPV (90%) was confirmed for reactive brain tissue.

TABLE 2.

Predictive attributes of CLE compared with frozen sections and conventional histology sections in samples from 30 patients treated for brain tumors

ValueCLE vs Frozen Sections, All SamplesCLE vs Permanent Histology Sections
All SamplesGliomaReactive Brain Tissue
Diagnostic accuracy, %94929382
Sensitivity, %94 (72–100)90 (78–96)91 (72–98)0 (0–95)
Specificity, %100 (18–100)94 (74–100)100 (65–100)90 (60–100)
PPV, %100 (80–100)97 (87–100)100 (84–100)0 (0–95)
NPV, %67 (12–98)81 (60–92)78 (45–96)90 (60–100)

Values in parentheses are 95% CIs. Conventional histology sections were stained with H&E.

Illustrative Clinical Cases

Five cases of CLE in vivo application are described (Figs. 4 and 5 and Videos 1 and 2).

VIDEO 1. CLE optical biopsy acquisition process from tectal pilocytic astrocytoma. The CLE probe is moved slowly over the surface of the tumor, revealing atypical cells with Rosenthal fibers in the background. Used with permission from Barrow Neurological Institute, Phoenix, Arizona. Click here to view.

VIDEO 2. CLE optical biopsy acquisition process from sphenoid wing meningioma. A clear demarcation line is visible between the nest of tumor cells and normal dura without tumor. Used with permission from Barrow Neurological Institute, Phoenix, Arizona. Click here to view.

FIG. 4.
FIG. 4.

Images of lesions in 3 patients. A: Intraoperative neuronavigational axial MR image of a large nonenhancing lesion in the left frontal lobe, showing the location of the CLE probe for optical biopsy (left). Intraoperative photograph showing the probe tip located on the tumor margin (right). B: CLE optical biopsy showing hypercellular regions and scattered atypical cells. C: H&E-stained biopsy specimen taken from the tumor margin, showing hypercellularity with numerous atypical cells, matching the CLE image and consistent with WHO grade III anaplastic astrocytoma. D: Intraoperative neuronavigational axial MR image of a contrast-enhancing lesion in the left insular lobe, showing the location of the CLE probe (left). Intraoperative photograph showing the probe tip placed on the core of the tumor (right). E: CLE optical biopsy showing atypical hypercellular regions interpreted as high-grade glioma with axons in the background (arrowheads). F: H&E-stained tissue biopsy specimen taken from the tumor core, showing hypercellularity with numerous atypical cells matching the CLE image and consistent with WHO grade IV glioblastoma. G: Intraoperative neuronavigational MR image of a contrast-enhancing lesion in the right frontoparietal region, showing the location of the CLE optical biopsy (left). Intraoperative photograph showing the CLE probe placed on the surface of the lesion (right). H: CLE optical biopsy showing hypocellular regions with incidental atypical cells (arrowheads). I: H&E-stained biopsy specimen taken from the lesion surface, showing a hypocellular region without histoarchitectural atypia, matching the CLE image and consistent with treatment-related reactive brain tissue. Bar = 100 µm. Figure is available in color online only.

FIG. 5.
FIG. 5.

Images of lesions in 3 patients. A: Intraoperative neuronavigational MR images of a tectal lesion, showing the location of the CLE optical biopsy. B: Intraoperative photograph showing the CLE probe placed on the surface of the tumor. C: CLE optical biopsy showing hypercellular regions of atypical cells (between the dashed lines) and Rosenthal fibers in the background (arrowheads). D: H&E-stained biopsy specimen from the same location, showing concordance in hypercellularity with atypical cells (dashed outlines), matching the CLE optical biopsy and consistent with WHO grade I pilocytic astrocytoma. E: Various intraoperative neuronavigational CT images of a sphenoid wing meningioma. F: Intraoperative photograph showing the CLE probe placed on the side of the meningioma attachment to the dura. G: CLE image showing a clear demarcation between normal dura without tumor (asterisk) and a nest of tumor cells (dashed outlines). H: H&E-stained tissue biopsy specimen taken from the tumor surface, showing a nest of tumor cells (arrowheads) with collagen fibers in the background (dark pink). Bars = 100 µm. Figure is available in color online only.

During surgery, CLE was used to explore the tumor core and margins. The visual communication and analysis between the neurosurgeon and neuropathologist occurred in real time via the CLE TSP.

Discussion

Study Interpretation

This prospective study of 30 patients evaluated the feasibility and safety of in vivo use of the first FDA-cleared FNa-based CLE system for intraoperative visualization of brain tumor microstructure. The study was designed as a real-world evaluation of CLE by neurosurgeons who were not well experienced with it. The study achieved its two primary end points: CLE use was safe, and no patient experienced adverse effects related to FNa. Neurosurgeons obtained CLE images in all cases. The study also met all 3 secondary end points. The first interpretable images were acquired early enough after the start of CLE to be surgically informative. The neuropathologist identified histological features in CLE images in 29 (97%) of 30 cases, with the interpretation of CLE optical biopsies correlating to conventional histology in 92% of corresponding H&E-stained sections. The difficulty associated with the perineurioma case was related to the mismatch of optical and tissue biopsy sites rather than the interpretation of images for this tumor type.

CLE Image Acquisition and Interpretation

CLE provides immediate intraoperative cellular visualization of brain tissue without the need to extract and process tissue samples, making it a promising addition or even alternative to frozen-section assessment of the lesion, histoarchitecture, and extent of tumor resection.3,4,7,10,11,15,16 Timely, high-quality images were acquired for all patients, although more than half were uninterpretable, consistent with previous reports3,9,16 (Fig. 6). This circumstance is due to the CLE image scan rate and the relationship to image artifacts caused by motion. The first interpretable image was acquired after only a mean of 6 images within 5 seconds after CLE imaging initiation, which was less than that reported in a previous in vivo study conducted with an earlier experimental CLE model.3 The mean duration of CLE use was 7 minutes per case, and CLE did not extend the surgical operating time.

FIG. 6.
FIG. 6.

CLE images of artifacts from RBCs (A) and motion (B). Motion artifacts were greater for superficial ROIs than for deep ROIs, likely because the surgeon moves the probe to a greater degree on the surface of unconstrained tissue. It is difficult to move the probe within deep lesions, especially if the probe tip cannot be seen.

Acquisition of interpretable images correlated strongly with CLE experience, indicating a productive learning curve (Fig. 2). A major aspect of the feasibility analysis was to have neurosurgeons with various levels of CLE experience be successful using the system. Surgeons must develop expertise to properly position the probe and hold it steady. The CLE system can scan to a depth of 30 µm into the tissue, thus creating volumetric imaging.17 The surgeon determines the CLE imaging plane by adjusting the scanning depth controlled at the CLE tower by a CLE-experienced technician using a foot-pedal control.

The CLE system connects neurosurgeons directly to the tissue histoarchitecture revealed on the digital fluorescence images. Learning to recognize artifacts, such as blood, tissue motion, and moving, or focusing the CLE probe is important for CLE image acquisition of interpretable images. Because the probe must be in contact with the tissue, tactile feedback while viewing the digital images requires practice; thus, adequate training is required before intraoperative use.

Of the 4 neurosurgical teams, only team 4, which performed 15 cases, showed a significant increase in the number of interpretable images per case. Based on the characteristics of team 4’s cumulative CLE experience, training time before intraoperative CLE use should be ≥ 116 minutes. However, the neurosurgeon in the first study case acquired excellent images in a deep tectal glioma with navigational guidance for probe placement.

Superficially located lesions were associated with fewer interpretable images than deeply located lesions. CLE imaging of these lesions may be susceptible to motion artifacts due to probe instability in the operator’s hand without probe support. The probe has a working length of 150 mm. Superficial locations are often the first to be imaged. Thus, imaging quality may not be as high as later in the case. Nonetheless, better images could be acquired from deep lesions because probe movement is more constrained at greater depths.

In all patients with glioma, CLE imaging readily identified hypercellular regions of the tumor with numerous scattered atypical cells of different sizes and shapes that matched corresponding histology. The ability to quickly assess the suspected marginal area optically in real time was an advantage of CLE. Using CLE at the tumor’s margin, the neurosurgeon found hypercellular regions well demarcated from the surrounding hypocellular regions in a case of pilocytic astrocytoma (Fig. 5C). CLE can also be useful in meningiomas. CLE imaging of a sphenoid wing meningioma revealed a clear demarcation between a nest of invading tumor cells and normal dura without tumor cells (Video 2). Because of the feasibility nature of this study, this assessment is not listed as an end point. However, an analysis of specific margin images from these tumors is underway. More homogeneous tumor studies may better assess the ability of CLE to evaluate tumor margins. Although fewer interpretable images were obtained from meningiomas using CLE than from other lesions, this difference was likely related to the tumor location and whether the surgeon was experienced with CLE.

Diagnostic Accuracy

The CLE imaging and histology correlation process was more rigorous in this study than in our previous research.2,3,9 Interpretation of optical biopsies was based on the ratio of interpretable versus noninterpretable images, which resulted in a general diagnostic impression of the many CLE images for a particular optical biopsy location. The high number of uninterpretable images resulted from artifacts caused by probe movement while the images were acquired (1.3 images/sec) and processed into the video-like presentation.

In our study and others that used a system not FDA-cleared for human use, CLE showed higher sensitivity and specificity for tumor detection than frozen sections.3,9 The time to informative image acquisition and imaging duration were much less than the time required for frozen section processing and evaluation. Categorizing CLE images as interpretable and noninterpretable was used as an objective marker to assess CLE performance in different scenarios and tumor types.

CLE imaging sensitivity for gliomas in our study was similar to that observed in a previous in vivo study (91%) and higher than that in an ex vivo study (66%).3,9 The requirement for tissue harvest, nonviable tissue, and slight time delay with ex vivo imaging likely account for the differences in vivo and ex vivo CLE imaging quality. CLE had a high specificity for detecting nonlesional abnormal brain tissue and was particularly effective for discriminating reactive brain changes. The NPV is directly related to the prevalence of lesional tissue samples among different subgroups. NPV decreases with increasing lesion prevalence. The lesion prevalence and NPV were 89% and 67% among all frozen section biopsies, respectively. These values were 70% and 81% among all permanent histology sections, 77% and 78% for gliomas, and 9% and 90% for reactive brain tissue. These scores are related to the 30-patient sample and could change with larger patient cohorts. Typical features of reactive brain tissue shown on CLE images were hypocellular regions with characteristic indications of macrophage presence. The ability of CLE to reveal reactive brain tissue is promising. It provides a solid basis for reliably differentiating recurrent and primary brain tumors from radiation treatment–induced changes when examining recurrent tumors.

CLE imaging potentially provides a distinct advantage, revealing in vivo in situ histoarchitecture that cannot necessarily be correlated with conventional histology. However, for now, to achieve familiarity and legitimacy for image interpretation and usefulness, CLE imaging must be correlated to gold-standard histological assessment. A tutorial set of matched H&E-stained and CLE images significantly facilitates the learning curve for pathologists inexperienced with CLE interpretation.18 Assessment of a tutorial set before performing a blinded test resulted in diagnostic accuracy of 92.9% in CLE image interpretation.18 As CLE technology has become more common in surgical pathology, professional pathology associations have responded with appropriate training.19

CLE for Nonenhancing Gliomas

CLE can be particularly useful in detecting the extent of low- and high-grade nonenhancing gliomas on MRI. Diffuse low-grade gliomas that do not enhance on MRI studies with contrast may be malignant despite showing no enhancement.2023 Such tumors usually have a less permeable blood-brain barrier, resulting in little to no extravasation of FNa, yielding dark images on CLE (except for blood vessels filled with fluorescence) or indeterminate images given local tissue microtrauma. A case of low-grade glioma imaged with CLE revealed that high-dose FNa can delineate low-grade gliomas.24 In our study, 2 patients had diffuse nonenhancing gliomas on MRI that were thought to be low grade. However, CLE imaging revealed hypercellular tumors with numerous atypical cells (Fig. 5). Both tumors were later interpreted as high-grade gliomas, with more permeable blood-brain barriers, permitting the extravasation of FNa into the parenchyma.

Timing of FNa Administration

FNa (5 mg/kg) administered 5 minutes before CLE use appeared effective for obtaining interpretable images.3 Evaluation of CLE use and timing of FNa administration indicated that a shorter time after FNa administration resulted in better CLE image quality.6,9 FNa undergoes rapid metabolism in the liver to fluorescein monoglucuronide, which also has fluorescent properties. Fluorescein monoglucuronide (half-life of approximately 264 minutes) appears to allow CLE images collected in the later stages of surgery to remain informative. In our study, the longest interval between FNa administration and imaging was 193 minutes. CLE images obtained at longer time intervals had low contrast and brightness but still showed hypercellularity and cell atypia that led to the determination of lesional tissue. Decisions by the surgeons to not redose FNa were made because the quality of acquired CLE images was satisfactory. Redosing FNa in situations of delayed CLE use after FNa administration is feasible to achieve improved imaging, as shown in ex vivo CLE.25 The optimal protocol for FNa administration for concurrent FNa-guided tumor resection and CLE tissue interrogation warrants further investigation.

CLE Use With TSP

A recently developed web-based CLE TSP system add-on was introduced into our CLE procedure for the last 11 cases.14 TSP allows for real-time, secure, remote, shared web application communication between the neurosurgeon and the neuropathologist. As the neurosurgeon probes the tissue with CLE, images are displayed in real time to web-linked personnel for immediate intraoperative interpretation of images. Such communication is especially beneficial when the neurosurgeon attempts to identify pathologically marginal tissue to decide whether to proceed with resection.

Strengths and Limitations of CLE Use

Because this study was an intraoperative feasibility and safety study, the number of patients available was limited, and the study included heterogeneous lesions. One of the study’s purposes was to use CLE imaging for various intracranial pathologies, which may affect the analyses of specificity and sensitivity for individual tumor types. Future studies should include evaluators with different levels of CLE experience to assess CLE images in a blinded fashion and determine their validity and diagnostic accuracy.

In a previous study, a neurosurgeon CLE expert participated in most cases of CLE assessment.3 In contrast, our study involved only 1 neurosurgeon who had previously used CLE. The uneven distribution of CLE cases performed by neurosurgeons introduces user bias, whereas more experienced neurosurgeons would likely have more ease and success obtaining interpretable images. In addition, the primary surgeon occasionally allowed a resident neurosurgeon to operate the CLE, introducing further user bias.

FNa, which emits 488-nm wavelength light, is currently the only FDA-approved fluorophore for CLE use.3 Other intraoperative fluorophores (indocyanine green and 5-aminolevulinic acid) are not used with clinical-grade CLE26,27 because they cannot be detected within the CLE laser operating range. FNa should be administered close to the time of CLE imaging because of its distribution properties.25,27 Surgeons should be familiar with the techniques of fluorescence-guided surgery.

Motion and red blood cell (RBC) artifacts can obscure tissue images during CLE. An inexperienced user could falsely identify a background of RBCs as atypical hypercellularity, resulting in an inaccurate CLE image interpretation. However, RBC collections can be identified with experience. Irrigation of the probe tip and underlying tissue reduces RBC imaging interference.

The price of high CLE resolution that allows visualization of cellular-level tissue histoarchitecture could be its vulnerability to motion artifacts. Slight CLE probe movement may cause artifact and imaging difficulty. Motion artifacts can be caused by brain tissue pulsation or freehand use of the probe. Obtaining better images requires practice with the probe. Success with the CLE system depends on user dedication. The length of use and number of cases resulted in higher-quality images. The duration of CLE use per case in some centers was only about 1–2 minutes per case, while surgeons in our study averaged 7 minutes per case (3–18 minutes per case).

CLE was designed for real-time visualization of histoarchitecture and tissue cellularity instead of tissue diagnosis. CLE brings intraoperative microscopic imaging and interrogation of surgical tissues to the neurosurgeon. The goal of CLE is to detect actionable images on the basis of discriminating abnormal from nonabnormal tissue. This discernment is useful for optimizing resection at primary invasive tumor margins or inspecting eloquent tissue for tumor invasion. Future development of CLE should be directed toward improving image resolution, increasing the field of view, increasing the image acquisition rate, and improving laser stability during image acquisition.

Conclusions

We assessed the feasibility of the first FDA-cleared clinical-grade CLE for evaluating tumor tissue during in vivo brain surgery. Four neurosurgical teams used intraoperative CLE imaging and confirmed its safety and feasibility in all 30 patients. Informative and actionable CLE images were obtained, especially as users gained experience with the CLE system. The study met the secondary end points for the duration of CLE use necessary to obtain interpretable images allowing diagnostic accuracy. CLE optical biopsies correlated with permanent histological sections in 92% of biopsies. The CLE system is a promising technology that produces a real-time intraoperative optical biopsy with high-resolution visualization of tissue microstructure, allowing identification of lesional tissue.

Acknowledgments

We thank the staff of Neuroscience Publications at Barrow Neurological Institute for assistance with manuscript and video preparation.

Disclosures

Dr. Smith: direct stock ownership in Gammatile, patent holder with Osteomed, and clinical or research support for the study described from Medtronic. Dr. Lawton: consultant for Carl Zeiss Meditec AG.

Carl Zeiss Meditec provided the CLE system used and funds to offset administrative study costs but had no involvement in data analysis, manuscript preparation, or the interpretation of the conclusions in this report.

Author Contributions

Conception and design: Preul, Abramov, Belykh. Acquisition of data: Preul, Abramov, Dru, Gooldy, Farber, Little, Porter, Smith, Lawton. Analysis and interpretation of data: Preul, Abramov, Dru, Xu, Gooldy, Scherschinski, Farber. Drafting the article: Preul, Abramov, Park. Critically revising the article: Preul, Abramov, Belykh, Little, Porter, Smith, Eschbacher. Reviewed submitted version of manuscript: Preul. Approved the final version of the manuscript on behalf of all authors: Preul. Statistical analysis: Abramov. Study supervision: Preul, Abramov.

Supplemental Information

Previous Presentations

This paper was presented in part at the 89th Annual Scientific Meeting of the American Association of Neurological Surgeons, presented virtually, August 21–25, 2021.

References

  • 1

    Sankar T, Delaney PM, Ryan RW, et al. Miniaturized handheld confocal microscopy for neurosurgery: results in an experimental glioblastoma model. Neurosurgery. 2010;66(2):410418.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Sanai N, Eschbacher J, Hattendorf G, et al. Intraoperative confocal microscopy for brain tumors: a feasibility analysis in humans. Neurosurgery. 2011;68(2 Suppl Operative):282290.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Martirosyan NL, Eschbacher JM, Kalani MY, et al. Prospective evaluation of the utility of intraoperative confocal laser endomicroscopy in patients with brain neoplasms using fluorescein sodium: experience with 74 cases. Neurosurg Focus. 2016;40(3):E11.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Martirosyan NL, Georges J, Eschbacher JM, et al. Potential application of a handheld confocal endomicroscope imaging system using a variety of fluorophores in experimental gliomas and normal brain. Neurosurg Focus. 2014;36(2):E16.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Belykh E, Miller EJ, Carotenuto A, et al. Progress in confocal laser endomicroscopy for neurosurgery and technical nuances for brain tumor imaging with fluorescein. Front Oncol. 2019;9:554.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Höhne J, Schebesch KM, Zoubaa S, Proescholdt M, Riemenschneider MJ, Schmidt NO. Intraoperative imaging of brain tumors with fluorescein: confocal laser endomicroscopy in neurosurgery. Clinical and user experience. Neurosurg Focus. 2021;50(1):E19.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Belykh E, Miller EJ, Patel AA, et al. Diagnostic accuracy of a confocal laser endomicroscope for in vivo differentiation between normal injured and tumor tissue during fluorescein-guided glioma resection: laboratory investigation. World Neurosurg. 2018;115:e337e348.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Belykh E, Zhao X, Ngo B, et al. Visualization of brain microvasculature and blood flow in vivo: feasibility study using confocal laser endomicroscopy. Microcirculation. 2021;28(3):e12678.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Belykh E, Zhao X, Ngo B, et al. Intraoperative confocal laser endomicroscopy ex vivo examination of tissue microstructure during fluorescence-guided brain tumor surgery. Front Oncol. 2020;10:599250.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Belykh E, Ngo B, Farhadi DS, et al. Confocal laser endomicroscopy assessment of pituitary tumor microstructure: a feasibility study. J Clin Med. 2020;9(10):E3146.

  • 11

    Acerbi F, Pollo B, De Laurentis C, et al. Ex vivo fluorescein-assisted confocal laser endomicroscopy (CONVIVO® system) in patients with glioblastoma: results from a prospective study. Front Oncol. 2020;10:606574.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    US Food & Drug Administration. Approval letter to Carl Zeiss Meditec AG. 2018.Accessed May 20, 2022. https://www.accessdata.fda.gov/cdrh_docs/pdf18/K181116.pdf

  • 13

    Park MT, Abramov I, Gooldy TC, et al. Introduction of in vivo confocal laser endomicroscopy and real-time telepathology for remote intraoperative neurosurgery-pathology consultation. Oper Neurosurg (Hagerstown). In press.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Abramov I, Park MT, Gooldy TC, et al. Real-time intraoperative surgical telepathology using confocal laser endomicroscopy. Neurosurg Focus. 2022;52(6):E9.

  • 15

    Breuskin D, Divincenzo J, Kim YJ, Urbschat S, Oertel J. Confocal laser endomicroscopy in neurosurgery: a new technique with much potential. Minim Invasive Surg. 2013;2013:851819.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Izadyyazdanabadi M, Belykh E, Mooney MA, et al. Prospects for theranostics in neurosurgical imaging: empowering confocal laser endomicroscopy diagnostics via deep learning. Front Oncol. 2018;8:240.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Belykh E, Patel AA, Miller EJ, et al. Probe-based three-dimensional confocal laser endomicroscopy of brain tumors: technical note. Cancer Manag Res. 2018;10:31093123.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Eschbacher J, Martirosyan NL, Nakaji P, et al. In vivo intraoperative confocal microscopy for real-time histopathological imaging of brain tumors. J Neurosurg. 2012;116(4):854860.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    College of American Pathologists. In Vivo Microscopy Topic Center.Accessed May 20, 2022. https://www.cap.org/member-resources/councils-committees/in-vivo-microscopy-committee/in-vivo-microscopy-topic-center

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Ohgaki H, Dessen P, Jourde B, et al. Genetic pathways to glioblastoma: a population-based study. Cancer Res. 2004;64(19):68926899.

  • 21

    Ohgaki H, Kleihues P. Genetic pathways to primary and secondary glioblastoma. Am J Pathol. 2007;170(5):14451453.

  • 22

    Scott JN, Brasher PMA, Sevick RJ, Rewcastle NB, Forsyth PA. How often are nonenhancing supratentorial gliomas malignant? A population study. Neurology. 2002;59(6):947949.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Dropcho EJ, Soong SJ. The prognostic impact of prior low grade histology in patients with anaplastic gliomas: a case-control study. Neurology. 1996;47(3):684690.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Belykh E, Onaka NR, Zhao X, et al. High-dose fluorescein reveals unusual confocal endomicroscope imaging of low-grade glioma. Front Neurol. 2021;12 668656.

  • 25

    Abramov I, Dru AB, Belykh E, Park MT, Bardonova L, Preul MC. Redosing of fluorescein sodium improves image interpretation during intraoperative ex vivo confocal laser endomicroscopy of brain tumors. Front Oncol. 2021;11(3960):668661.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Guyotat J, Pallud J, Armoiry X, Pavlov V, Metellus P. 5-Aminolevulinic acid-protoporphyrin IX fluorescence-guided surgery of high-grade gliomas: a systematic review. Adv Tech Stand Neurosurg. 2016;(43):6190.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Martirosyan NL, Cavalcanti DD, Eschbacher JM, et al. Use of in vivo near-infrared laser confocal endomicroscopy with indocyanine green to detect the boundary of infiltrative tumor. J Neurosurg. 2011;115(6):11311138.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand

Figure from Alzahrani et al. (pp 858–867). Illustrations (left) © Rajiv Midha, published with permission.

  • FIG. 1.

    Heat map of neurosurgeons’ (NS) ratings of CLE feasibility. The CLE system was rated on a 5-point scale, where 1 was the worst score and 5 was the best. CLE was rated by the 2 attending neurosurgeons, who were regarded as authoritative and experienced enough with CLE to assess its feasibility. Conditional formatting was applied on a scale from worst score (red) to best score (green). Asterisks indicate the neurosurgeon’s repeated evaluation of CLE technology after the conclusion of the study. Figure is available in color online only.

  • FIG. 2.

    Graphs showing interpretation and correlation between the number or percentage of interpretable images or other factors. A: Number of interpretable and noninterpretable images obtained for each case. B–D: Scatterplots. Solid lines represent linear regression, and dashed lines indicate 95% CIs. Correlation between the number of interpretable images and progression of the study from the first case to the last case (B). Correlation between percentage of interpretable images per case and the cumulative length of time of CLE usage in the operating room throughout the study (C). Correlation between the percentage of interpretable images and time spent per case (D). E: Scatterplot showing correlation between percentage of interpretable images and the consecutive number of cases performed by each neurosurgery team (NT) during the study. F–H: Bar graphs. Error bars indicate SD. Comparison of mean percentage of interpretable images acquired from superficially and deeply located lesions (F). Mean number of CLE images by type of diagnosis (G). Comparison of mean percentage of interpretable images acquired from different types of diagnosis (H). *p < 0.05; ***p < 0.001. Figure is available in color online only.

  • FIG. 3.

    Histological features identified on CLE image (left) and corresponding H&E-stained histological specimen (right) from the same patient and lesion for various pathology types. CLE images were high quality (even from inexperienced neurosurgeons) and, in many cases, showed structural features not seen on conventional histology slides. A: Hypercellularity with atypical cells of different sizes indicative of infiltrative high-grade glioma. Bright spots likely correspond to cell surface vesicles on astrocytes, as seen with electron microscopy.24 B: Nest of atypical cells with collagen background indicative of meningioma tissue. C: Atypical tumor cells with abundant cytoplasm and nucleus on the periphery (white outlines) indicative of anaplastic astrocytoma. Dashed lines indicate a blood vessel. Transiting erythrocytes are seen in laminar form within the vessel. Erythrocytes have a laminar appearance in the CLE image because their flow is faster than the CLE scanning rate. A larger cell, interpreted as a leukocyte, is seen within the vessel (arrowhead). Arrows indicate erythrocyte infiltration. D: Numerous atypical cells (arrowheads) indicative of pineocytoma. E: Atypical cells (arrowheads) invading the blood vessel (dashed lines) indicative of glioblastoma. F: Cellular tumor and scattered atypical larger cells in a background of erythrocytes (smallest cells) correspond to metastasis of breast adenocarcinoma. Arrowheads indicate atypical mitotic cells. G: Hypocellular regions with neurodegeneration represented by corpora amylacea (white arrowheads) and axons (black arrowheads) indicative of reactive brain tissue. H: Hypocellular regions of reactive brain tissue with abundant macrophages (dashed outline). White arrowheads indicate axons. Bar = 100 µm (A–G); 200 µm (H). Figure is available in color online only.

  • FIG. 4.

    Images of lesions in 3 patients. A: Intraoperative neuronavigational axial MR image of a large nonenhancing lesion in the left frontal lobe, showing the location of the CLE probe for optical biopsy (left). Intraoperative photograph showing the probe tip located on the tumor margin (right). B: CLE optical biopsy showing hypercellular regions and scattered atypical cells. C: H&E-stained biopsy specimen taken from the tumor margin, showing hypercellularity with numerous atypical cells, matching the CLE image and consistent with WHO grade III anaplastic astrocytoma. D: Intraoperative neuronavigational axial MR image of a contrast-enhancing lesion in the left insular lobe, showing the location of the CLE probe (left). Intraoperative photograph showing the probe tip placed on the core of the tumor (right). E: CLE optical biopsy showing atypical hypercellular regions interpreted as high-grade glioma with axons in the background (arrowheads). F: H&E-stained tissue biopsy specimen taken from the tumor core, showing hypercellularity with numerous atypical cells matching the CLE image and consistent with WHO grade IV glioblastoma. G: Intraoperative neuronavigational MR image of a contrast-enhancing lesion in the right frontoparietal region, showing the location of the CLE optical biopsy (left). Intraoperative photograph showing the CLE probe placed on the surface of the lesion (right). H: CLE optical biopsy showing hypocellular regions with incidental atypical cells (arrowheads). I: H&E-stained biopsy specimen taken from the lesion surface, showing a hypocellular region without histoarchitectural atypia, matching the CLE image and consistent with treatment-related reactive brain tissue. Bar = 100 µm. Figure is available in color online only.

  • FIG. 5.

    Images of lesions in 3 patients. A: Intraoperative neuronavigational MR images of a tectal lesion, showing the location of the CLE optical biopsy. B: Intraoperative photograph showing the CLE probe placed on the surface of the tumor. C: CLE optical biopsy showing hypercellular regions of atypical cells (between the dashed lines) and Rosenthal fibers in the background (arrowheads). D: H&E-stained biopsy specimen from the same location, showing concordance in hypercellularity with atypical cells (dashed outlines), matching the CLE optical biopsy and consistent with WHO grade I pilocytic astrocytoma. E: Various intraoperative neuronavigational CT images of a sphenoid wing meningioma. F: Intraoperative photograph showing the CLE probe placed on the side of the meningioma attachment to the dura. G: CLE image showing a clear demarcation between normal dura without tumor (asterisk) and a nest of tumor cells (dashed outlines). H: H&E-stained tissue biopsy specimen taken from the tumor surface, showing a nest of tumor cells (arrowheads) with collagen fibers in the background (dark pink). Bars = 100 µm. Figure is available in color online only.

  • FIG. 6.

    CLE images of artifacts from RBCs (A) and motion (B). Motion artifacts were greater for superficial ROIs than for deep ROIs, likely because the surgeon moves the probe to a greater degree on the surface of unconstrained tissue. It is difficult to move the probe within deep lesions, especially if the probe tip cannot be seen.

  • 1

    Sankar T, Delaney PM, Ryan RW, et al. Miniaturized handheld confocal microscopy for neurosurgery: results in an experimental glioblastoma model. Neurosurgery. 2010;66(2):410418.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Sanai N, Eschbacher J, Hattendorf G, et al. Intraoperative confocal microscopy for brain tumors: a feasibility analysis in humans. Neurosurgery. 2011;68(2 Suppl Operative):282290.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Martirosyan NL, Eschbacher JM, Kalani MY, et al. Prospective evaluation of the utility of intraoperative confocal laser endomicroscopy in patients with brain neoplasms using fluorescein sodium: experience with 74 cases. Neurosurg Focus. 2016;40(3):E11.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Martirosyan NL, Georges J, Eschbacher JM, et al. Potential application of a handheld confocal endomicroscope imaging system using a variety of fluorophores in experimental gliomas and normal brain. Neurosurg Focus. 2014;36(2):E16.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Belykh E, Miller EJ, Carotenuto A, et al. Progress in confocal laser endomicroscopy for neurosurgery and technical nuances for brain tumor imaging with fluorescein. Front Oncol. 2019;9:554.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 6

    Höhne J, Schebesch KM, Zoubaa S, Proescholdt M, Riemenschneider MJ, Schmidt NO. Intraoperative imaging of brain tumors with fluorescein: confocal laser endomicroscopy in neurosurgery. Clinical and user experience. Neurosurg Focus. 2021;50(1):E19.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Belykh E, Miller EJ, Patel AA, et al. Diagnostic accuracy of a confocal laser endomicroscope for in vivo differentiation between normal injured and tumor tissue during fluorescein-guided glioma resection: laboratory investigation. World Neurosurg. 2018;115:e337e348.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 8

    Belykh E, Zhao X, Ngo B, et al. Visualization of brain microvasculature and blood flow in vivo: feasibility study using confocal laser endomicroscopy. Microcirculation. 2021;28(3):e12678.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Belykh E, Zhao X, Ngo B, et al. Intraoperative confocal laser endomicroscopy ex vivo examination of tissue microstructure during fluorescence-guided brain tumor surgery. Front Oncol. 2020;10:599250.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Belykh E, Ngo B, Farhadi DS, et al. Confocal laser endomicroscopy assessment of pituitary tumor microstructure: a feasibility study. J Clin Med. 2020;9(10):E3146.

  • 11

    Acerbi F, Pollo B, De Laurentis C, et al. Ex vivo fluorescein-assisted confocal laser endomicroscopy (CONVIVO® system) in patients with glioblastoma: results from a prospective study. Front Oncol. 2020;10:606574.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    US Food & Drug Administration. Approval letter to Carl Zeiss Meditec AG. 2018.Accessed May 20, 2022. https://www.accessdata.fda.gov/cdrh_docs/pdf18/K181116.pdf

  • 13

    Park MT, Abramov I, Gooldy TC, et al. Introduction of in vivo confocal laser endomicroscopy and real-time telepathology for remote intraoperative neurosurgery-pathology consultation. Oper Neurosurg (Hagerstown). In press.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Abramov I, Park MT, Gooldy TC, et al. Real-time intraoperative surgical telepathology using confocal laser endomicroscopy. Neurosurg Focus. 2022;52(6):E9.

  • 15

    Breuskin D, Divincenzo J, Kim YJ, Urbschat S, Oertel J. Confocal laser endomicroscopy in neurosurgery: a new technique with much potential. Minim Invasive Surg. 2013;2013:851819.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Izadyyazdanabadi M, Belykh E, Mooney MA, et al. Prospects for theranostics in neurosurgical imaging: empowering confocal laser endomicroscopy diagnostics via deep learning. Front Oncol. 2018;8:240.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Belykh E, Patel AA, Miller EJ, et al. Probe-based three-dimensional confocal laser endomicroscopy of brain tumors: technical note. Cancer Manag Res. 2018;10:31093123.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Eschbacher J, Martirosyan NL, Nakaji P, et al. In vivo intraoperative confocal microscopy for real-time histopathological imaging of brain tumors. J Neurosurg. 2012;116(4):854860.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    College of American Pathologists. In Vivo Microscopy Topic Center.Accessed May 20, 2022. https://www.cap.org/member-resources/councils-committees/in-vivo-microscopy-committee/in-vivo-microscopy-topic-center

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Ohgaki H, Dessen P, Jourde B, et al. Genetic pathways to glioblastoma: a population-based study. Cancer Res. 2004;64(19):68926899.

  • 21

    Ohgaki H, Kleihues P. Genetic pathways to primary and secondary glioblastoma. Am J Pathol. 2007;170(5):14451453.

  • 22

    Scott JN, Brasher PMA, Sevick RJ, Rewcastle NB, Forsyth PA. How often are nonenhancing supratentorial gliomas malignant? A population study. Neurology. 2002;59(6):947949.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Dropcho EJ, Soong SJ. The prognostic impact of prior low grade histology in patients with anaplastic gliomas: a case-control study. Neurology. 1996;47(3):684690.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Belykh E, Onaka NR, Zhao X, et al. High-dose fluorescein reveals unusual confocal endomicroscope imaging of low-grade glioma. Front Neurol. 2021;12 668656.

  • 25

    Abramov I, Dru AB, Belykh E, Park MT, Bardonova L, Preul MC. Redosing of fluorescein sodium improves image interpretation during intraoperative ex vivo confocal laser endomicroscopy of brain tumors. Front Oncol. 2021;11(3960):668661.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Guyotat J, Pallud J, Armoiry X, Pavlov V, Metellus P. 5-Aminolevulinic acid-protoporphyrin IX fluorescence-guided surgery of high-grade gliomas: a systematic review. Adv Tech Stand Neurosurg. 2016;(43):6190.

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 27

    Martirosyan NL, Cavalcanti DD, Eschbacher JM, et al. Use of in vivo near-infrared laser confocal endomicroscopy with indocyanine green to detect the boundary of infiltrative tumor. J Neurosurg. 2011;115(6):11311138.

    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 1944 218 0
Full Text Views 843 466 48
PDF Downloads 1000 535 71
EPUB Downloads 0 0 0