I. SIGNIFICANCE/CONTEXT AND IMPORTANCE OF THE QUESTION OR SUBJECT STUDIED
Surgical therapy for brain tumors presents a number of challenges, including differentiating tumor from normal tissue and achieving maximal resection without injuring adjacent functional tissue. A growing body of evidence demonstrates that extent of resection is one of the key determinants of progression-free and overall survival in patients with low-grade glioma (LGG) and high-grade glioma (HGG).
A number of different technologies have been developed in an attempt to increase extent of resection. These include the use of neuronavigation, intraoperative ultrasound, and intraoperative magnetic resonance imaging (MRI). Intraoperative fluorescent tumor labeling is an emerging technology with potential to directly differentiate tumor from normal tissue during resection.
Currently, the only widespread application of fluorescence-guided brain tumor resection involves HGG labeling with protoporphyrin IX fluorescence induced by 5-aminolevulinic acid (5-ALA). Stummer et al 1 demonstrated in a phase IIIa clinical trial in patients with HGG that 5-ALA-induced fluorescence-guided resection significantly increases the gross total resection rate (65% vs 36%) and 6-month progression-free survival (41% vs 21%). Although used internationally, 5-ALA has not yet been US Food and Drug Administration (FDA)-approved for brain tumor surgery in the United States.
Fluorescence-guided surgery requires specialized operating microscopes that provide the proper illumination and optical filtering to visualize fluorescently labeled tissues. This works well for malignant tumors with bright tumor fluorescence, such as HGG labeled with 5-ALA, but the relatively low-power magnification of the operating microscope rarely yields visible 5-ALA fluorescence in LGG, even though the tumor demonstrates quantifiable fluorescence.2 Fluorescence is often visible with higher levels of magnification, especially with the resolution and clarity provided by laser-scanning confocal microscopy. Although the confocal microscopes traditionally used in basic research are not compatible with intraoperative use, a miniaturized fiber optic point-scanning confocal microscope has been developed that allows for handheld use (Optiscan FIVE 1, Optiscan, Notting Hill, Victoria, Australia). This handheld confocal microscope allows for cellular level resolution, with a stated lateral resolution of 0.7 μm and an optical slice thickness of approximately 7 μm.3
The intraoperative diagnosis of brain tumors has traditionally been performed by biopsy and frozen sectioning. This method is prone to sampling error. Furthermore, frozen sectioning does a relatively poor job of maintaining tissue cytoarchitecture to allow for reliable diagnosis and grading of tumors. Intraoperative confocal microscopy potentially allows for the imaging of tissue with a similar cellular resolution, without disrupting the tissue. Moreover, more areas could be sampled in the same time frame, decreasing the potential for sampling error.
II. ORIGINALITY OF THE WORK
The confocal microscope allows for a level of resolution far beyond current intraoperative techniques. This miniaturized confocal microscope has been used previously to evaluate pathology in other organ systems, including the bladder, gastrointestinal tract, and skin. Schlosser et al previously reported using the confocal microscope to investigate human glioblastoma multiforme tissue, but only after excision of the tumor tissue. The utility of the handheld confocal microscope for in vivo brain tumor imaging was first demonstrated in a mouse glioblastoma model by Sankar et al3 of the Barrow Neurological Institute (BNI). In this article they used topically applied acriflavine, a nuclear stain, and intravenously administered fluorescein to provide real-time images of the cellular architecture of glioblastoma. A clear difference between normal tissue and tumor was observed, and the margin of tumor infiltration was identified.
In the current study, Sanai et al demonstrate for the first time the feasibility of intraoperative fluorescent confocal microscopy for imaging a number of different human brain tumors in vivo at a cellular level. The handheld confocal microscope has the potential to provide intraoperative brain tumor diagnosis and identify tumor margin.
III. APPROPRIATENESS OF STUDY DESIGN OR EXPERIMENTAL APPROACH
The study qualitatively compares the ability of the intraoperative confocal microscope to visualize fluorescent labeling to identify characteristic cytoarchitectural features of a wide variety of different neurosurgery pathologies vs permanent histopathology.
Specifically, 33 patients diagnosed with neurosurgical lesions planned for standard resection with the aid of neuronavigation were evaluated. Once tumor was exposed, intravenous sodium fluorescein was administered, and images were acquired with the handheld confocal probe. The location of the probe was recorded via MRI neuronavigation, and biopsies for histopathologic examination were taken at the site of image acquisition. Images were also taken of healthy brain; however, no biopsies were obtained from healthy brain. The lesions represented included high- and low-grade astrocytoma, oligodendroglioma, mixed oligoastrocytoma, central neurocytoma, meningioma (grade I and II), metastasis, and radiation necrosis.
The authors note that the intraoperative confocal images demonstrated very good microscopic resolution, especially of tumor microvasculature. Figures are shown demonstrating characteristic pathologic features of oligodendroglioma, central neurocytoma, meningioma, and anaplastic astrocytoma obtained with intraoperative confocal microscopy. These are compared with corresponding histologic specimens.
IV. ADEQUACY OF EXPERIMENTAL TECHNIQUES—WERE THESE NECESSARY AND SUFFICIENT TO ANSWER THE QUESTIONS/ANY DRAWBACKS?
Sanai et al claim that the images were of sufficient resolution for their neuropathologist to obtain an intraoperative preliminary diagnosis. A significant limitation of the article is a lack of objective evidence to support this claim beside the few select examples included as figures. A blinded comparison of the neuropathologist's diagnosis based on intraoperative confocal images vs the diagnosis based on permanent histology would have considerably improved their claim. Moreover, despite discussing the potential of intraoperative confocal microscopy to augment or replace frozen-section pathology, frozen sections are not provided to compare with the confocal images, and there is no attempt to compare the accuracy of the 2 methods for obtaining diagnosis.
V. SOUNDNESS OF CONCLUSIONS AND INTERPRETATION—WAS THE ANALYSIS OF THE DATA APPROPRIATE (INCLUDING STATISTICS); DOES THE ANALYSIS SUPPORT THE CONCLUSIONS?
The figures provided do show a favorable qualitative correlation between the confocal microscope images and permanent hematoxylin and eosin–stained sections, suggesting that this technology could provide an intraoperative method to identify tumor type and margin. Moreover, the microscope demonstrated utility with a wide variety of pathologies and did not significantly add to operative time, with an imaging time of approximately 10 minutes per patient and a total additional time added to the case of approximately 15 to 20 minutes. However, the sensitivity and specificity of the confocal microscope to reliably identify tumor margin and tumor type cannot be evaluated because of the lack of quantifiable measurements.
VI. RELEVANCE OF DISCUSSION
The discussion covers a number of interesting potential future uses for the intraoperative confocal microscope. There is also a discussion of some of the practical limitations, including obscuration of the probe by blood. More detailed analysis of the time required for setup and use of the probe would be beneficial. Specifically, it would be helpful to have documented how many sites can be examined in the 10 minutes of imaging time mentioned, and whether it would be practical to use the probe to survey an entire resection cavity.
One aspect of feasibility not discussed in the article is the cost and availability of the Optiscan FIVE 1 system, which costs approximately $161 000 (personal correspondence), and is not FDA approved for clinical use. Proven clinical benefit will be required to justify a relatively high cost.
VII. CLARITY OF WRITING, STRENGTH, AND ORGANIZATION OF THE ARTICLE
Overall, the article has clear, understandable prose, even when discussing technical specifications. One issue with the organization of the article is the presentation in the protocol section of data regarding the time added to the surgery by the use of the intraoperative confocal microscope. The additional intraoperative time required for use of confocal microscopy is one of the important variables that will determine the real-world practicality of this imaging modality, and it should have been prospectively studied and presented in the results section.
VIII. ECONOMY OF WORDS
The article does a good job of concisely describing the new intraoperative confocal microscope technology and rationale for investigating the probe as an adjunct to tumor diagnosis and tumor margin delineation. The results section includes appropriate descriptions of the appearance of various pathologies as they compare with permanent histopathology. The enumeration of tumor location and surgical approaches could have been shortened to a more general statement about the ability to use the probe in a variety of locations and surgical approaches. The discussion is relatively lengthy and includes speculative discussions that are not always related to specific findings of the article; however, this is reasonable in a article introducing a new technology so as to present avenues for future research.
IX. RELEVANCE, ACCURACY, AND COMPLETENESS OF BIBLIOGRAPHY
Sanai et al appropriately referenced articles describing the limitations of intraoperative pathology and current surgical adjuncts. They also provided a number of references to the use of the handheld confocal microscope in identifying tumor pathology in other organ systems. One notable exception to the completeness of the bibliography was the omission of the article by Schlosser et al that described the first use of the Optiscan confocal microscope with human glioblastoma multiforme tissue.4 Schlosser et al used the microscope to visualize excised tumor treated with acriflavine. Their study provides important information about the capabilities of the Optiscan confocal microscope and should have been referenced, as Sanai et al admit in response to a letter to the editor identifying the oversight.
X. NUMBER AND QUALITY OF FIGURES, TABLES, AND ILLUSTRATIONS
The article includes a number of excellent figures demonstrating the capabilities of the Optiscan handheld confocal microscope and comparing the images obtained to corresponding permanent histopathology. The article would have been strengthened if examples of all the various pathologies had been included, including the cases of radiation necrosis. More examples of corresponding normal brain images would also be beneficial to allow the reader to gauge the ability to differentiate tumor from normal parenchyma. Furthermore, examples of tumor margin would have helped in demonstrating the potential of the microscope to aid in tumor resection. The intraoperative MRI neuronavigation images could have been limited to one representative image to conserve space and allow for more tissue images. There is obviously a limited amount of space available for additional figures, so the various pathologies not included in the print article perhaps could have been included as online supplementary material.
XI. WHAT FUTURE/NEXT STEP DOES THIS PAPER LOGICALLY LEAD TO?
The logical next step would be to perform a blinded trial to quantitatively evaluate the ability of neurosurgeons and neuropathologists to provide an accurate diagnosis using intraoperative confocal images vs frozen pathology sections in comparison with corresponding permanent pathology sections.
Another important avenue of research is determining the optimal fluorescent labeling agent. Sodium fluorescein, a relatively nonspecific agent that mostly highlights vasculature, was likely chosen because it is FDA approved. The BNI has subsequently investigated the use of 5-ALA and indocyanine green as alternative fluorescent labels.2,5 Newer agents with increased tumor specificity could enhance the ability to identify tumor margin and aid in surgical resection.
Most importantly, as a feasibility study, the article focuses on preliminary diagnosis, but is unable to answer the more significant question of whether the potential ability to identify tumor margins afforded by intraoperative confocal microscopy meaningfully improves extent of resection. Sanai et al do provide a preliminary answer to this question in another case series presented in a 2011 article published in Journal of Neurosurgery.2 The study involved LGG patients who underwent resection with the aid of 5-ALA and intraoperative confocal microscopy. As expected with LGG, there was no fluorescence visible when using the usual fluorescent operative microscope; however, all 10 of the LGG displayed fluorescence when evaluated with the confocal probe. Additionally, in some patients, the use of 5-ALA labeling and intraoperative confocal microscopy led to increased tumor resection beyond the margin based on standard operative microscopy and MRI neuronavigation. The BNI is currently conducting a phase IIIa randomized placebo-controlled trial, the BALANCE study, which will evaluate the effect of 5-ALA fluorescence and intraoperative confocal microscopy on the volume of residual disease with LGG resection.2
Intraoperative confocal microscopy-assisted fluorescence-guided surgery is an intriguing new technology that requires further investigation before it is considered for widespread clinical use. Additional research is needed to identify optimal fluorescent tumor labels, to determine the correlation of intraoperative confocal images with tumor pathology and brain cytoarchitecture, and to determine the impact of intraoperative confocal microscopy on extent of tumor resection and patient survival.
The authors have no personal financial or institutional interest in any of the drugs, materials, or devices described in this article. This work is not supported by or affiliated with any funding source.
We gratefully acknowledge the residents that participated in this journal discussion: Drs Amrendra S. Miranpuri, Andrew M. Bauer, Christopher M. Nickele, Casey J. Madura, Jane Ng, Solomon M. Ondoma, Kutluay Uluç. Dr John S. Kuo also provided valuable assistance and guidance for this Journal Club Report.