Toward digital histopathological assessment in surgery for central nervous system tumors using stimulated Raman histology

Lisa I. WadiuraDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Lisa I. Wadiura in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Barbara KieselDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Barbara Kiesel in
jns
Google Scholar
PubMed
Close
 MD
,
Thomas Roetzer-PejrimovskyDivision of Neuropathology and Neurochemistry, Department of Neurology, Medical University of Vienna, Austria;

Search for other papers by Thomas Roetzer-Pejrimovsky in
jns
Google Scholar
PubMed
Close
 MD
,
Mario MischkulnigDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Mario Mischkulnig in
jns
Google Scholar
PubMed
Close
 MD
,
Clemens C. VogelDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Clemens C. Vogel in
jns
Google Scholar
PubMed
Close
,
Johannes A. HainfellnerDivision of Neuropathology and Neurochemistry, Department of Neurology, Medical University of Vienna, Austria;

Search for other papers by Johannes A. Hainfellner in
jns
Google Scholar
PubMed
Close
 MD
,
Christian MatulaDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Christian Matula in
jns
Google Scholar
PubMed
Close
 MD
,
Christian W. FreudigerInvenio Imaging, Inc., Menlo Park, California; and

Search for other papers by Christian W. Freudiger in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Daniel A. OrringerDepartment of Neurosurgery, New York University, New York, New York

Search for other papers by Daniel A. Orringer in
jns
Google Scholar
PubMed
Close
 MD
,
Adelheid WöhrerDivision of Neuropathology and Neurochemistry, Department of Neurology, Medical University of Vienna, Austria;

Search for other papers by Adelheid Wöhrer in
jns
Google Scholar
PubMed
Close
 MD, PhD
,
Karl RoesslerDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Karl Roessler in
jns
Google Scholar
PubMed
Close
 MD
, and
Georg WidhalmDepartment of Neurosurgery, Medical University of Vienna, Austria;

Search for other papers by Georg Widhalm in
jns
Google Scholar
PubMed
Close
 MD, PhD
Free access

OBJECTIVE

Intraoperative neuropathological assessment with conventional frozen sections supports the neurosurgeon in optimizing the surgical strategy. However, preparation and review of frozen sections can take as long as 45 minutes. Stimulated Raman histology (SRH) was introduced as a novel technique to provide rapid high-resolution digital images of unprocessed tissue samples directly in the operating room that are comparable to conventional histopathological images. Additionally, SRH images are simultaneously and easily accessible for neuropathological judgment. Recently, the first study showed promising results regarding the accuracy and feasibility of SRH compared with conventional histopathology. Thus, the aim of this study was to compare SRH with conventional H&E images and frozen sections in a large cohort of patients with different suspected central nervous system (CNS) tumors.

METHODS

The authors included patients who underwent resection or stereotactic biopsy of suspected CNS neoplasm, including brain and spinal tumors. Intraoperatively, tissue samples were safely collected and SRH analysis was performed directly in the operating room. To enable optimal comparison of SRH with H&E images and frozen sections, the authors created a digital databank that included images obtained with all 3 imaging modalities. Subsequently, 2 neuropathologists investigated the diagnostic accuracy, tumor cellularity, and presence of diagnostic histopathological characteristics (score 0 [not present] through 3 [excellent]) determined with SRH images and compared these data to those of H&E images and frozen sections, if available.

RESULTS

In total, 94 patients with various suspected CNS tumors were included, and the application of SRH directly in the operating room was feasible in all cases. The diagnostic accuracy based on SRH images was 99% when compared with the final histopathological diagnosis based on H&E images. Additionally, the same histopathological diagnosis was established in all SRH images (100%) when compared with that of the corresponding frozen sections. Moreover, the authors found a statistically significant correlation in tumor cellularity between SRH images and corresponding H&E images (p < 0.0005 and R = 0.867, Pearson correlation coefficient). Finally, excellent (score 3) or good (2) accordance between diagnostic histopathological characteristics and H&E images was present in 95% of cases.

CONCLUSIONS

The results of this retrospective analysis demonstrate the near-perfect diagnostic accuracy and capability of visualizing relevant histopathological characteristics with SRH compared with conventional H&E staining and frozen sections. Therefore, digital SRH histopathology seems especially useful for rapid intraoperative investigation to confirm the presence of diagnostic tumor tissue and the precise tumor entity, as well as to rapidly analyze multiple tissue biopsies from the suspected tumor margin. A real-time analysis comparing SRH images and conventional histological images at the time of surgery should be performed as the next step in future studies.

ABBREVIATIONS

CNS = central nervous system; HGG = high-grade glioma; LGG = low-grade glioma; SRH = stimulated Raman histology; WHO = World Health Organization.

OBJECTIVE

Intraoperative neuropathological assessment with conventional frozen sections supports the neurosurgeon in optimizing the surgical strategy. However, preparation and review of frozen sections can take as long as 45 minutes. Stimulated Raman histology (SRH) was introduced as a novel technique to provide rapid high-resolution digital images of unprocessed tissue samples directly in the operating room that are comparable to conventional histopathological images. Additionally, SRH images are simultaneously and easily accessible for neuropathological judgment. Recently, the first study showed promising results regarding the accuracy and feasibility of SRH compared with conventional histopathology. Thus, the aim of this study was to compare SRH with conventional H&E images and frozen sections in a large cohort of patients with different suspected central nervous system (CNS) tumors.

METHODS

The authors included patients who underwent resection or stereotactic biopsy of suspected CNS neoplasm, including brain and spinal tumors. Intraoperatively, tissue samples were safely collected and SRH analysis was performed directly in the operating room. To enable optimal comparison of SRH with H&E images and frozen sections, the authors created a digital databank that included images obtained with all 3 imaging modalities. Subsequently, 2 neuropathologists investigated the diagnostic accuracy, tumor cellularity, and presence of diagnostic histopathological characteristics (score 0 [not present] through 3 [excellent]) determined with SRH images and compared these data to those of H&E images and frozen sections, if available.

RESULTS

In total, 94 patients with various suspected CNS tumors were included, and the application of SRH directly in the operating room was feasible in all cases. The diagnostic accuracy based on SRH images was 99% when compared with the final histopathological diagnosis based on H&E images. Additionally, the same histopathological diagnosis was established in all SRH images (100%) when compared with that of the corresponding frozen sections. Moreover, the authors found a statistically significant correlation in tumor cellularity between SRH images and corresponding H&E images (p < 0.0005 and R = 0.867, Pearson correlation coefficient). Finally, excellent (score 3) or good (2) accordance between diagnostic histopathological characteristics and H&E images was present in 95% of cases.

CONCLUSIONS

The results of this retrospective analysis demonstrate the near-perfect diagnostic accuracy and capability of visualizing relevant histopathological characteristics with SRH compared with conventional H&E staining and frozen sections. Therefore, digital SRH histopathology seems especially useful for rapid intraoperative investigation to confirm the presence of diagnostic tumor tissue and the precise tumor entity, as well as to rapidly analyze multiple tissue biopsies from the suspected tumor margin. A real-time analysis comparing SRH images and conventional histological images at the time of surgery should be performed as the next step in future studies.

Intraoperative histopathological assessment of suspected central nervous system (CNS) tumors with analysis of frozen sections, as well as smear preparations, constitutes a crucial technique to confirm diagnostic tumor tissue during resections and biopsies.1,2 In this sense, use of conventional frozen sections support the neurosurgeon in optimizing the surgical strategy. However, preparation and review of conventional intraoperative frozen sections can take as long as 45 minutes and is thus time consuming. This limits intraoperative histopathological assessment of multiple tissue samples, including biopsies from the suspected tumor margin. Moreover, in-house neuropathological preparation, as well as evaluation, is not permanently available. Therefore, new techniques are warranted to optimize rapid intraoperative neuropathological assessment of suspected CNS tumors.

Recently, stimulated Raman histology (SRH) was introduced as a novel medical technique for providing high-resolution digital images of unprocessed tissue samples directly in the operating room.3 Recent studies showed promising first results regarding the feasibility and diagnostic accuracy of SRH, which provides histopathological features comparable to those of conventional frozen sections with H&E stainings.36 The scanning process takes approximately 3 minutes to provide digital images, and the SRH images are rapidly produced and digitally accessible online for histopathological assessment by neuropathologists.3 Another major advantage is that the unprocessed scanned tumor samples can still be applied for conventional neuropathological workup after SRH investigation.7

In this study, we aimed to compare the diagnostic accuracy of SRH imaging with that of conventional histology for suspected CNS tumors. To this end, we compared the diagnostic accuracy, tumor cell density, and presence of diagnostic histopathological characteristics determined with SRH images with those determined with conventional H&E images in a large cohort of patients with different suspected brain and spinal tumors. Additionally, we also compared the diagnostic accuracy of SRH images with that of conventional frozen sections to investigate the potential value of this novel technique for modernizing intraoperative histopathological assessment. To enable optimal comparisons of SRH with H&E images and frozen sections, we created a digital databank that included images obtained with all three modalities.

Methods

In this prospective study, adult patients (≥ 18 years) who were scheduled for neurosurgical resection or stereotactic biopsy of suspected, newly diagnosed, and recurrent CNS tumors at the Department of Neurosurgery, Medical University of Vienna, Austria, between November 2020 and March 2021 were included. Each patient underwent diagnostic magnetic resonance imaging prior to surgery, as well as contrast-enhanced T1-weighted imaging for integration into the neuronavigation system during surgery. The study protocol was approved by the local ethics committee of Medical University of Vienna.

Neurosurgical Resections/Biopsies and Intraoperative Tissue Sampling for SRH Analysis

All resections and stereotactic biopsies were conducted with neuronavigational guidance, as described previously.810 Tissue samples, especially from the suspected tumor core, were collected during neurosurgical resection or biopsy for further investigation with SRH whenever safely possible. The tissue samples (approximately 5 mm3) were positioned on a custom slide and scanned with the SRH imaging system (NIO Laser Imaging System, Invenio Imaging Inc.), as described previously.3,4 After approximately 3 minutes, digital histopathological images were available for further histopathological assessment without any staining procedure. After SRH analysis, all scanned tissue samples were transferred to the neuropathology department for histopathological workup. According to the performing neurosurgeon’s preference, additional intraoperative histology with frozen sections was conducted. Figure 1 provides a demonstration of the intraoperative workflow for the tissue samples analyzed with the SRH technique.

FIG. 1.
FIG. 1.

Intraoperative workflow for tissue samples analyzed with the SRH technique. The SRH imaging system (NIO Laser Imaging System, Invenio Imaging Inc.) is placed directly in the operating room (A). Tissue samples are derived during surgery (B) or stereotactic biopsy (C). In the next step, the collected tissue samples are positioned on a custom slide (D) and scanned with the SRH imaging system to provide digital histopathological images (E) for further neuropathological analysis.

Comparison of SRH Imaging With H&E Staining and Frozen Sections

Tissue diagnosis was established for each suspected tumor according to the valid classification system defined by the World Health Organization (WHO) that was available at the time of diagnosis.11 To allow optimal comparison of SRH images with conventional histology, one representative suspected tumor sample from each case was selected by two board-certified neuropathologists (A.W. and T.R.P.) for inclusion in this study. All H&E images and frozen sections were digitized and stored in a digital databank with the SRH images.

In the first step, the two neuropathologists assessed the image quality of each SRH case. According to their judgment, good image quality was present if appropriate evaluation of the histopathological characteristics was possible; in contrast, poor image quality was present if adequate evaluation of histopathological characteristics was not possible. Subsequently, the two neuropathologists reviewed all SRH images, H&E images, and frozen sections of each suspected tumor to make a neuropathological diagnosis. For each suspected tumor, information on patient age, tumor localization, newly diagnosed versus recurrent tumor, and suspected preoperative diagnosis were provided to the neuropathologists. In the next step, we correlated the histopathological diagnosis based on the SRH images with the corresponding neuropathological diagnosis based on the H&E images, as well as the diagnosis according to conventional frozen section analysis.

Comparison of Degree of Tumor Cellularity and Presence of Diagnostic Histopathological Characteristics Between SRH and H&E Images

Moreover, to compare SRH and H&E imaging, the 2 neuropathologists classified the degree of tumor cellularity according to a cellularity score ranging from 1 (low cellularity) to 2 (moderate cellularity) to 3 (high cellularity). Finally, the neuropathologists compared the presence of diagnostic histopathological characteristics in the SRH images with those in the H&E images and classified accordance with the following score: 0 (not present), 1 (low), 2 (good), and 3 (excellent).

Statistical Analysis

Statistical analyses were performed using SPSS version 27.0. General descriptive patient data included patient age, sex, newly diagnosed versus recurrent tumor, tumor entity, presence of histopathological characteristics, tumor localization, type of surgery, and comparison of degree of tumor cellularity between H&E and SRH images. The inferential statistical analysis aimed to investigate the correlation between cellularity in H&E images and SRH images and was performed using the Pearson correlation coefficient. Because only a single inferential analysis was conducted, no correction for multiple testing was necessary and the commonly applied cutoff of p < 0.05 for statistical significance was used.

Results

Patient Characteristics

The median (range) age of the final study group was 57.5 (21–85) years, and the cohort had a female/male ratio of 1.5:1. In this prospective study, 94 patients with different suspected CNS tumors were investigated, including 90 (96%) cranial and 4 (4%) spinal pathologies. Eighty-nine tumor resections and 5 stereotactic biopsies were performed. The most common histopathological diagnosis was meningioma in 30 cases (32%), high-grade glioma (HGG) in 28 (30%), metastasis in 19 (20%), neurinoma in 6 (6%), and low-grade glioma (LGG) in 2 (2%). Further patient characteristics are provided in Table 1.

TABLE 1.

Patient characteristics

CharacteristicValue
Patients94 (100)
Male/female sex ratio1.5:1
Age, yrs57.5 (21–85)
Recurrent tumor
 Yes20 (21)
 No74 (79)
Tumor localization
 Cranial90 (96)
  Supratentorial80
  Infratentorial10
 Spinal4 (4)
Surgery
 Resection89 (95)
 Stereotactic biopsy5 (5)
Tumor entity
 Meningioma30 (32)
  WHO grade I22
  WHO grade II8
 HGG28 (30)
  Glioblastoma WHO IV21
  Gliosarcoma WHO IV2
  Anaplastic astrocytoma WHO II4
  Anaplastic oligodendroglioma WHO II1
 LGG2 (2)
  Diffuse astrocytoma WHO II2
 Metastasis19 (20)
 Neurinoma6 (6)
 Lymphoma2 (2)
 Cavernoma2 (2)
 Epidermoid cyst1 (1)
 Focal cortical dysplasia1 (1)
 Esthesioneuroblastoma1 (1)
 Epstein-Barr virus–associated alteration1 (1)
 Inflammatory myofibroblastic tumor1 (1)
Frozen section analysis
 Yes19 (20)
 No75 (80)

Values are shown as number (%) or median (range) unless indicated otherwise.

Intraoperative Tissue Sampling, SRH Analysis, and Corresponding Image Quality

The intraoperative application of the SRH system directly in the operating room was feasible in all cases. According to the neuropathologists’ judgment, good image quality sufficient for histopathological assessment was present in all cases (100%). A histopathological diagnosis was made by the neuropathologists in 92 of 94 SRH images (98%). In contrast, a histopathological diagnosis could not be obtained in 2 SRH images. It is notable that both cases were finally diagnosed as cavernomas according to conventional histopathological analysis.

Comparison of SRH Analysis With Final Histopathological H&E Diagnosis

The final neuropathological diagnosis based on H&E images was confirmed by the two neuropathologists in 91 of the 92 cases with an established histopathological diagnosis according to SRH. Thus, we found a diagnostic accuracy of 99% for SRH. In contrast, in 1 of 92 (1%) cases, the final neuropathological H&E diagnosis was different from the SRH diagnosis. In this single case, the neuropathologists diagnosed HGG on the SRH images, whereas the final histopathological H&E diagnosis was melanoma metastasis.

Furthermore, we conducted a subgroup analysis of each tumor entity by analyzing the rate of correct neuropathological diagnosis based on SRH imaging in comparison with the final histopathological H&E diagnosis. According to our data, 30 of 30 (100%) meningiomas, 28 of 28 (100%) HGGs, 18 of 19 (95%) metastases, 6 of 6 (100%) neurinomas, 2 of 2 (100%) LGGs, 2 of 2 (100%) lymphomas, and 5 of 5 (100%) other tumor suspected lesions were diagnosed correctly by the neuropathologists on the basis of the corresponding SRH images. Details are provided in Table 2. See Fig. 2 for representative examples of common histopathologies included in this investigation.

TABLE 2.

Diagnostic accuracy

CharacteristicMeningiomasMetastasesHGGsLGGsNeurinomasOthersOverall
SRH vs final H&E diagnosis
 No. of tissue samples30192826792
 Same diagnosis30 (100)18 (95)28 (100)2 (100)6 (100)7 (100)91 (99)
 Different diagnosis1 (5)1 (5)
SRH vs fresh frozen section
 No. of tissue samples33910319
 Same diagnosis3 (100)3 (100)9 (100)1 (100)3 (100)19 (100)
 Different diagnosis
 Accordance w/ H&E3 (100)3 (100)9 (100)1 (100)3 (100)19 (100)

Values are shown as number or number (%).

FIG. 2.
FIG. 2.

Representative examples of common histopathologies included in this study. For each case, an illustrative region of interest of the SRH image (B, E, H, K, N, and Q), as well as the corresponding H&E image (C, F, I, L, O, R), is shown. WHO grade 1 spinal psammomatous meningioma (A–C), WHO grade 4 isocitrate dehydrogenase (IDH) wild-type parietal glioblastoma (D–F), IDH-mutated, WHO grade 2, left parietal diffuse astrocytoma (G–I), left frontal metastasis of a lung carcinoma (J–L), acoustic schwannoma located at the right cerebellopontine angle (M–O), and right-sided frontal cavernoma (P–R) are shown.

Comparison of SRH With Conventional Frozen Section Histology

Conventional frozen section was additionally performed during surgery in 19 of 92 (21%) cases. Frozen section histology was conducted for different suspected CNS tumors, with a final histopathological diagnosis of HGG in 9 cases, meningioma in 3 cases, and metastasis in 3 cases, as well as 1 case each of LGG, esthesioneuroblastoma, lymphoma, and Epstein-Barr virus–associated inflammation. According to the neuropathologists’ judgment, the same histopathological diagnosis was made in all SRH images (100%) compared with the corresponding frozen sections. Compared with the final H&E histopathology, the histopathological diagnosis based on SRH/frozen section was correct in all cases (100%).

Comparison of Degree of Tumor Cellularity Between SRH Images and Conventional H&E Staining

In the next step, we determined the tumor cellularity score, ranging from 1 (low cellularity) to 3 (high cellularity), for each SRH image and corresponding H&E image (Fig. 3). Overall, the median tumor cellularity score was 2 for SRH images and 2 for H&E images. According to our data, the median tumor cellularity score based on SRH images was 2 for meningiomas, HGGs, metastases, neurinomas, and LGGs and 3 for other suspected tumors. Based on the corresponding samples with H&E staining, the tumor cellularity score was 2 for meningiomas, HGGs, metastases, neurinomas, and LGGs and 3 for other suspected tumors. Altogether, the tumor cellularity scores of the SRH images and corresponding H&E images were significantly correlated (p < 0.0005, R = 0.867, Pearson correlation coefficient). A scatterplot of the results comparing tumor cellularity based on SRH images and H&E images is provided in Fig. 4.

FIG. 3.
FIG. 3.

The tumor cellularity score used in this study. An example for each score (1 [low cellularity] to 3 [high cellularity]) is demonstrated with an SRH image and corresponding H&E image.

FIG. 4.
FIG. 4.

Scatterplot comparing tumor cellularity between SRH and H&E images. In this study, we found a statistically significant correlation in the tumor cellularity scores between the SRH images and corresponding H&E images.

Presence of Diagnostic Histopathological Characteristics in SRH Images Versus Corresponding H&E Images

Finally, we investigated the presence of diagnostic histopathological characteristics (based on a cellularity score of 0 [not present] to 3 [excellent]) in each SRH image and its accordance with the corresponding H&E images. According to our data, a diagnostic histopathological characteristics score of 3 (excellent) was present in 62 of 92 (67%) cases, score of 2 (good) in 26 of 92 (28%) cases, score of 1 (low) in 3 of 92 cases (3%), and score of 0 (not present) in 1 of 92 cases (1%). More details are provided in Table 3.

TABLE 3.

Histopathological characteristics

ScoreMeningiomasMetastasesHGGsLGGsNeurinomasOthersOverall
0 (not present)1 (3)1 (5)00002 (2)
1 (low)1 (3)01 (4)0002 (2)
2 (good)3 (10)10 (53)8 (29)1 (50)1 (17)3 (43)26 (28)
3 (excellent)25 (84)8 (42)19 (67)1 (50)5 (83)4 (57)62 (67)
Total30 (100)19 (100)28 (100)2 (100)6 (100)7 (100)92 (100)

Values are shown as number (%).

Discussion

SRH was recently introduced as a novel laser-based technique to the neurosurgical field, providing rapid access to digital neuropathological images during surgery for CNS tumors.12 In 2017, Orringer et al. demonstrated the feasibility of SRH directly in the operating room for the first time and found high accuracy (> 90%) for predicting histopathological diagnosis in neurosurgical patients.3 At present, more extensive case series are urgently needed to further investigate the diagnostic accuracy of SRH during routine surgery in large cohorts of patients with different CNS tumors compared with conventional histopathological imaging methods. To this end, we designed the present study to investigate the diagnostic accuracy of SRH in comparison with those of H&E images and frozen sections in a large cohort of patients undergoing resection or stereotactic biopsy of different suspected CNS tumors. To allow optimal comparison of SRH with H&E images and frozen sections, we created a digital databank that included images obtained with all three modalities. To our knowledge, this is the first analysis to compare SRH with conventional histopathological images included in a specific digital databank that also included the corresponding H&E and frozen section histological images. Notably, the comparison of SRH with conventional H&E images was performed retrospectively and not during the neurosurgical procedure. The work reported here, in combination with other prior studies along these lines, is an essential component of the growing literature with the aim to justify the prospective use of SRH for a head-to-head diagnostic comparison in clinical practice.6,7,13

Image Quality and Feasibility of SRH

In our cohort of patients with different suspected CNS tumors derived from resection or stereotactic biopsy, good image quality formed the basis for suitable histopathological assessment in all 94 cases (100%). Similarly, Straehle et al. described good neuropathological interpretability of SRH images for approximately 96% of cases included in a cohort of different CNS tumors.4 According to neuropathological judgment, a histopathological diagnosis could be made for 92 of 94 (98%) cases on the basis of the SRH images in our study; in contrast, a histopathological diagnosis could not be obtained in 2 of our SRH images. Interestingly, both SRH images were acquired during resection of a lesion with a final histopathological diagnosis of cavernoma based on H&E images. Typically, cavernomas show distinctive vessel structures in H&E images based on the clear-cut staining technique; however, these characteristics may be absent on SRH images due to a potential squeezing effect. According to our data, the H&E images were superior to SRH images for diagnosing cavernomas. However, cavernomas do not usually constitute a relevant CNS tumor entity with a need for intraoperative neuropathological consulting.

Diagnostic Accuracy of SRH Compared With H&E Images

In the next step, we investigated the diagnostic accuracy of SRH compared with the final histopathological diagnosis based on H&E images. According to our data, SRH had a high diagnostic accuracy (99%) in our study cohort. These results are in line with those of two recent studies that reported diagnostic accuracy between 87.8%4 and 92.2%3 for SRH compared with conventional H&E images. Similarly, diagnostic accuracy between 92% and 96% for SRH compared with conventional H&E images was reported in a recent study of a group of 33 pediatric brain tumor patients.5 With regard to tumor entity, the diagnostic accuracy of SRH compared with H&E images for identifying the correct histopathological diagnosis was 100% for gliomas (LGG and HGG), lymphomas, meningiomas, and neurinomas. In contrast, the diagnostic accuracy of SRH was slightly lower (95%) for brain metastasis, with 1 case of a melanoma metastasis on H&E images that was misjudged by the neuropathologists as HGG on SRH images. In the literature, the study by Straehle et al. reported a diagnostic accuracy of 100% for SRH used to diagnose LGG and, interestingly, cerebral or spinal metastases.4 However, the authors found a lower diagnostic accuracy of 88.9% for SRH when used to diagnose HGG and meningiomas (90.9%).4 According to our data and the findings reported in the literature, the SRH technique is capable of visualizing the histopathological characteristics of suspected CNS tumors during surgery with an image quality comparable to that of conventional H&E images, which is generally accessible days after surgery during routine neuropathological workup.

Diagnostic Accuracy of SRH Compared With Frozen Section Histology

At present, intraoperative neuropathological assessment with frozen section is a crucial element during surgery for distinct CNS tumors, aiming to characterize the tumor tissue to optimize the surgical strategy during resection and to confirm the acquisition of diagnostic tumor tissue during biopsy. For example, in cases where preoperative imaging cannot reliably differentiate a lymphoma from an HGG, the neurosurgical procedure can usually be terminated after intraoperative histopathological confirmation of a lymphoma. In contrast, the tumor should generally be safely removed in the case of an intraoperative diagnosis of HGG. Therefore, we compared the diagnostic accuracy of SRH with that of intraoperative frozen sections, when available. According to our data, SRH had a diagnostic accuracy of 100% when compared with that of the corresponding frozen sections in a group of 19 patients with different suspected CNS tumors. The histopathological diagnosis established with SRH/frozen section was correct in all cases (100%) when compared with final H&E histopathological results. Comparably, recent studies found diagnostic accuracy between 92.2%3 and 91.5%6 for neuropathologists predicting the final diagnosis based on SRH images when compared with intraoperative frozen sections. In our view, SRH is a powerful technique that could be used to optimize and modernize intraoperative neuropathological assessment of suspected CNS tumors. In this sense, this innovative SRH tissue analysis technique allows visualization of histopathological details during surgery of suspected CNS tumors within 3 minutes. Because conventional intraoperative neuropathological assessment generally requires approximately 30–45 minutes, the SRH technique accelerates intraoperative estimation of histopathological tumor characteristics to optimize the surgical strategy and decision-making.

Comparison of Specific Histopathological Features in SRH and H&E Images

To further investigate the value of the SRH technique, we compared tumor cellularity and presence of diagnostic histopathological characteristics in each SRH image to the corresponding H&E images. According to our data, we found a statistically significant correlation in tumor cellularity between the SRH images and corresponding H&E images. In the study by Pekmezci et al., a semiquantitative consensus scoring system was applied to compare tissue cellularity between SRH images from the glioma margin with and without tumor infiltration.7 By using this approach, the authors found that the margin samples with tumor infiltration showed higher cellularity than samples without tumor infiltration on SRH images.7 Moreover, Hollon et al. compared the nuclear density of normal tissue with that of pediatric brain tumor tissue on SRH images by utilizing an automated image analysis application, and they found a statistically significant correlation between normal tissue, LGG, and HGG.5 Furthermore, we observed an excellent or good diagnostic histopathological characteristics score in the vast majority of cases (95%) in our study by comparing SRH with the corresponding H&E images. To our knowledge, this is the first study in the literature that investigated such a diagnostic histopathological characteristics scoring system in order to determine the value of SRH in a cohort of patients with suspected CNS tumors, including those who underwent resection and stereotactic biopsy. Altogether, these data demonstrate that the SRH technique is capable of visualizing even delicate histopathological details in suspected brain and spinal tumors.

Clinical Relevance and Future Directions

First, the time of surgery would be dramatically reduced for stereotactic or open biopsy cases if diagnostic tumor tissue could be confirmed with the SRH technique (approximately 3 minutes vs 30–45 minutes). This is of particular relevance in older patients, who constitute a vulnerable population with a higher rate of surgical and anesthesiological complications.14 Furthermore, precise estimation of a histopathological tumor entity could be markedly accelerated by using the SRH technique, and thus the surgical strategy for tumor resection could be determined at an early stage (for example, rapid intraoperative differentiation of HGG and metastasis). Additionally, SRH allows for rapid investigation of multiple tissue biopsies from the suspected tumor margin after assumed complete tumor resection within a short period of time. In this sense, Pekmezci et al. reported that the SRH technique is a powerful tool for identifying residual tumor tissue during glioma surgery.7 Moreover, the SRH technique allows digital review of histopathological images by neuropathologists and could be used to obtain a rapid reference opinion from other specialized institutions, or online access could be provided to obtain emergent expert opinions from experts working from home outside conventional working hours via an on-call service.

To improve the SRH technique, an intraoperative real-time SRH assessment should be established by developing a handheld probe for optimized CNS tumor visualization during surgery. Furthermore, the SRH technique should be combined with 5-aminolevulinic fluorescence and the conventional fluorescence technique to potentially visualize fluorescent tumor cells in an otherwise nonfluorescent LGG. Finally, recent studies have demonstrated that the combination of SRH with artificial intelligence algorithms could optimize histopathological assessment directly in the operating room.13,15 In this sense, the combination of these artificial intelligence algorithms with SRH images and intraoperative determination of the molecular profile of each individual CNS tumor would be of major interest to creating a precision medicine approach of intraoperative histopathological diagnosis in the future.

Limitations

First, a new classification for CNS tumors was released by the WHO during the study period. Because the histopathological diagnosis of each patient was established according to the valid WHO tumor classification at the time of diagnosis, not all CNS tumors were classified on the basis of the 2021 WHO criteria.11,16 Second, due to the small number of distinct tumor subgroups (e.g., LGG, lymphoma), detailed subgroup analyses were not feasible. Therefore, future large, multicenter studies are warranted to investigate the diagnostic accuracy of SRH and to enable a comprehensive subgroup analysis of different CNS tumors. Moreover, analysis of SRH compared with frozen section histology was available for only a minority of the included CNS tumors (n = 19) because frozen section analysis was not routinely performed in our study but according to neurosurgeon preference. This represents a major limitation of the current study. Therefore, we plan to systematically correlate SRH images with frozen sections at the time of surgery in a large patient cohort in order to augment the intraoperative diagnosis provided by SRH in a future study. Finally, the comparison of SRH with conventional H&E images was performed retrospectively and not during neurosurgical procedure. Therefore, our present data did not allow evaluation of intraoperative accuracy at the time of surgery because real-time diagnosis based on SRH was not attempted in this study, and thus we could not analyze differences between SRH images and frozen sections at the time of surgery.

Conclusions

We demonstrated near-perfect diagnostic accuracy of SRH compared with conventional H&E staining and frozen section analysis in a large group of patients with suspected brain and spinal tumors. According to our data, SRH is a powerful technique capable of visualizing even delicate histopathological details in suspected CNS tumors, such as tumor cellularity and diagnostic histopathological characteristics. With the SRH approach, the time of surgery can be dramatically reduced, especially for stereotactic biopsy. Moreover, rapid histopathological diagnosis can be provided for multiple biopsies of the assumed tumor margin during open resection. Because we did not perform a real-time analysis of SRH images and conventional histological findings obtained at the time of surgery, this should be investigated as the next step in future studies.

Disclosures

Dr. Freudiger is an employee, executive, and shareholder of Invenio Imaging, Inc.; and holds patents with SRS Microscopy. Dr. Orringer is an advisor and shareholder of Invenio Imaging, Inc.

Author Contributions

Conception and design: Widhalm, Wadiura, Kiesel, Roetzer-Pejrimovsky, Hainfellner, Wöhrer. Acquisition of data: Widhalm, Wadiura, Kiesel, Roetzer-Pejrimovsky, Vogel, Matula, Freudiger, Wöhrer, Rössler. Analysis and interpretation of data: Wadiura, Kiesel, Roetzer-Pejrimovsky, Mischkulnig, Vogel, Hainfellner, Matula, Orringer, Wöhrer, Rössler. Drafting the article: Widhalm, Wadiura, Mischkulnig, Orringer. Critically revising the article: all authors. Reviewed submitted version of manuscript: all authors. Approved the final version of the manuscript on behalf of all authors: Widhalm. Statistical analysis: Wadiura, Mischkulnig. Administrative/technical/material support: Widhalm, Kiesel, Hainfellner, Matula, Freudiger, Wöhrer, Rössler. Study supervision: Widhalm, Hainfellner.

References

  • 1

    Somerset HL, Kleinschmidt-DeMasters BK. Approach to the intraoperative consultation for neurosurgical specimens. Adv Anat Pathol. 2011;18(6):446449.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Roessler K, Dietrich W, Kitz K. High diagnostic accuracy of cytologic smears of central nervous system tumors. A 15-year experience based on 4,172 patients. Acta Cytol. 2002;46(4):667674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Orringer DA, Pandian B, Niknafs YS, et al. Rapid intraoperative histology of unprocessed surgical specimens via fibre-laser-based stimulated Raman scattering microscopy. Nat Biomed Eng. 2017;1(2):0027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4

    Straehle J, Erny D, Neidert N, et al. Neuropathological interpretation of stimulated Raman histology images of brain and spine tumors: part B. Neurosurg Rev. 2022;45(2):17211729.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Hollon TC, Lewis S, Pandian B, et al. Rapid intraoperative diagnosis of pediatric brain tumors using stimulated Raman histology. Cancer Res. 2018;78(1):278289.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Eichberg DG, Shah AH, Di L, et al. Stimulated Raman histology for rapid and accurate intraoperative diagnosis of CNS tumors: prospective blinded study. J Neurosurg. 2019;134(1):137143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Pekmezci M, Morshed RA, Chunduru P, et al. Detection of glioma infiltration at the tumor margin using quantitative stimulated Raman scattering histology. Sci Rep. 2021;11(1):12162.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Millesi M, Kiesel B, Wöhrer A, et al. Is intraoperative pathology needed if 5-aminolevulinic-acid-induced tissue fluorescence is found in stereotactic brain tumor biopsy? Neurosurgery. 2020;86(3):366373.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Kiesel B, Mischkulnig M, Woehrer A, et al. Systematic histopathological analysis of different 5-aminolevulinic acid-induced fluorescence levels in newly diagnosed glioblastomas. J Neurosurg. 2018;129(2):341353.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10

    Widhalm G, Minchev G, Woehrer A, et al. Strong 5-aminolevulinic acid-induced fluorescence is a novel intraoperative marker for representative tissue samples in stereotactic brain tumor biopsies. Neurosurg Rev. 2012;35(3):381391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol. 2016;131(6):803820.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    Freudiger CW, Pfannl R, Orringer DA, et al. Multicolored stain-free histopathology with coherent Raman imaging. Lab Invest. 2012;92(10):14921502.

  • 13

    Hollon TC, Parikh A, Pandian B, et al. A machine learning approach to predict early outcomes after pituitary adenoma surgery. Neurosurg Focus. 2018;45(5):E8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14

    Kiesel B, Wadiura LI, Mischkulnig M, et al. Efficacy, outcome, and safety of elderly patients with glioblastoma in the 5-ALA era: single center experience of more than 10 years. Cancers (Basel). 2021;13(23):6119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15

    Hollon TC, Pandian B, Adapa AR, et al. Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat Med. 2020;26(1):5258.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16

    Louis DN, Perry A, Wesseling P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):12311251.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Collapse
  • Expand
  • View in gallery
    FIG. 1.

    Intraoperative workflow for tissue samples analyzed with the SRH technique. The SRH imaging system (NIO Laser Imaging System, Invenio Imaging Inc.) is placed directly in the operating room (A). Tissue samples are derived during surgery (B) or stereotactic biopsy (C). In the next step, the collected tissue samples are positioned on a custom slide (D) and scanned with the SRH imaging system to provide digital histopathological images (E) for further neuropathological analysis.

  • View in gallery
    FIG. 2.

    Representative examples of common histopathologies included in this study. For each case, an illustrative region of interest of the SRH image (B, E, H, K, N, and Q), as well as the corresponding H&E image (C, F, I, L, O, R), is shown. WHO grade 1 spinal psammomatous meningioma (A–C), WHO grade 4 isocitrate dehydrogenase (IDH) wild-type parietal glioblastoma (D–F), IDH-mutated, WHO grade 2, left parietal diffuse astrocytoma (G–I), left frontal metastasis of a lung carcinoma (J–L), acoustic schwannoma located at the right cerebellopontine angle (M–O), and right-sided frontal cavernoma (P–R) are shown.

  • View in gallery
    FIG. 3.

    The tumor cellularity score used in this study. An example for each score (1 [low cellularity] to 3 [high cellularity]) is demonstrated with an SRH image and corresponding H&E image.

  • View in gallery
    FIG. 4.

    Scatterplot comparing tumor cellularity between SRH and H&E images. In this study, we found a statistically significant correlation in the tumor cellularity scores between the SRH images and corresponding H&E images.

  • 1

    Somerset HL, Kleinschmidt-DeMasters BK. Approach to the intraoperative consultation for neurosurgical specimens. Adv Anat Pathol. 2011;18(6):446449.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 2

    Roessler K, Dietrich W, Kitz K. High diagnostic accuracy of cytologic smears of central nervous system tumors. A 15-year experience based on 4,172 patients. Acta Cytol. 2002;46(4):667674.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 3

    Orringer DA, Pandian B, Niknafs YS, et al. Rapid intraoperative histology of unprocessed surgical specimens via fibre-laser-based stimulated Raman scattering microscopy. Nat Biomed Eng. 2017;1(2):0027.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 4

    Straehle J, Erny D, Neidert N, et al. Neuropathological interpretation of stimulated Raman histology images of brain and spine tumors: part B. Neurosurg Rev. 2022;45(2):17211729.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 5

    Hollon TC, Lewis S, Pandian B, et al. Rapid intraoperative diagnosis of pediatric brain tumors using stimulated Raman histology. Cancer Res. 2018;78(1):278289.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 6

    Eichberg DG, Shah AH, Di L, et al. Stimulated Raman histology for rapid and accurate intraoperative diagnosis of CNS tumors: prospective blinded study. J Neurosurg. 2019;134(1):137143.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 7

    Pekmezci M, Morshed RA, Chunduru P, et al. Detection of glioma infiltration at the tumor margin using quantitative stimulated Raman scattering histology. Sci Rep. 2021;11(1):12162.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 8

    Millesi M, Kiesel B, Wöhrer A, et al. Is intraoperative pathology needed if 5-aminolevulinic-acid-induced tissue fluorescence is found in stereotactic brain tumor biopsy? Neurosurgery. 2020;86(3):366373.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 9

    Kiesel B, Mischkulnig M, Woehrer A, et al. Systematic histopathological analysis of different 5-aminolevulinic acid-induced fluorescence levels in newly diagnosed glioblastomas. J Neurosurg. 2018;129(2):341353.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 10

    Widhalm G, Minchev G, Woehrer A, et al. Strong 5-aminolevulinic acid-induced fluorescence is a novel intraoperative marker for representative tissue samples in stereotactic brain tumor biopsies. Neurosurg Rev. 2012;35(3):381391.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 11

    Louis DN, Perry A, Reifenberger G, et al. The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathol. 2016;131(6):803820.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 12

    Freudiger CW, Pfannl R, Orringer DA, et al. Multicolored stain-free histopathology with coherent Raman imaging. Lab Invest. 2012;92(10):14921502.

  • 13

    Hollon TC, Parikh A, Pandian B, et al. A machine learning approach to predict early outcomes after pituitary adenoma surgery. Neurosurg Focus. 2018;45(5):E8.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 14

    Kiesel B, Wadiura LI, Mischkulnig M, et al. Efficacy, outcome, and safety of elderly patients with glioblastoma in the 5-ALA era: single center experience of more than 10 years. Cancers (Basel). 2021;13(23):6119.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 15

    Hollon TC, Pandian B, Adapa AR, et al. Near real-time intraoperative brain tumor diagnosis using stimulated Raman histology and deep neural networks. Nat Med. 2020;26(1):5258.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • 16

    Louis DN, Perry A, Wesseling P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):12311251.

    • Crossref
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 381 381 81
PDF Downloads 378 378 63
EPUB Downloads 0 0 0