Stimulated Raman histology for rapid and accurate intraoperative diagnosis of CNS tumors: prospective blinded study

Daniel G. Eichberg Departments of Neurological Surgery and

Search for other papers by Daniel G. Eichberg in
Current site
Google Scholar
PubMed
Close
 MD
,
Ashish H. Shah Departments of Neurological Surgery and

Search for other papers by Ashish H. Shah in
Current site
Google Scholar
PubMed
Close
 MD
,
Long Di Departments of Neurological Surgery and

Search for other papers by Long Di in
Current site
Google Scholar
PubMed
Close
 BS
,
Alexa M. Semonche Departments of Neurological Surgery and

Search for other papers by Alexa M. Semonche in
Current site
Google Scholar
PubMed
Close
 BA
,
George Jimsheleishvili Departments of Neurological Surgery and

Search for other papers by George Jimsheleishvili in
Current site
Google Scholar
PubMed
Close
 MD
,
Evan M. Luther Departments of Neurological Surgery and

Search for other papers by Evan M. Luther in
Current site
Google Scholar
PubMed
Close
 MD
,
Christopher A. Sarkiss Departments of Neurological Surgery and

Search for other papers by Christopher A. Sarkiss in
Current site
Google Scholar
PubMed
Close
 MD
,
Allan D. Levi Departments of Neurological Surgery and

Search for other papers by Allan D. Levi in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Sakir H. Gultekin Pathology, University of Miami Miller School of Medicine; and

Search for other papers by Sakir H. Gultekin in
Current site
Google Scholar
PubMed
Close
 MD
,
Ricardo J. Komotar Departments of Neurological Surgery and
Sylvester Comprehensive Cancer Center, Miami, Florida

Search for other papers by Ricardo J. Komotar in
Current site
Google Scholar
PubMed
Close
 MD
, and
Michael E. Ivan Departments of Neurological Surgery and
Sylvester Comprehensive Cancer Center, Miami, Florida

Search for other papers by Michael E. Ivan in
Current site
Google Scholar
PubMed
Close
 MD, MBS
Full access

OBJECTIVE

In some centers where brain tumor surgery is performed, the opportunity for expert intraoperative neuropathology consultation is lacking. Consequently, surgeons may not have access to the highest quality diagnostic histological data to inform surgical decision-making. Stimulated Raman histology (SRH) is a novel technology that allows for rapid acquisition of diagnostic histological images at the bedside.

METHODS

The authors performed a prospective blinded cohort study of 82 consecutive patients undergoing resection of CNS tumors to compare diagnostic time and accuracy of SRH simulation to the gold standard, i.e., frozen and permanent section diagnosis. Diagnostic accuracy was determined by concordance of SRH-simulated intraoperative pathology consultation with a blinded board-certified neuropathologist, with official frozen section and permanent section results.

RESULTS

Overall, the mean time to diagnosis was 30.5 ± 13.2 minutes faster (p < 0.0001) for SRH simulation than for frozen section, with similar diagnostic correlation: 91.5% (κ = 0.834, p < 0.0001) between SRH simulation and permanent section, and 91.5% between frozen and permanent section (κ = 0.894, p < 0.0001).

CONCLUSIONS

SRH-simulated intraoperative pathology consultation was significantly faster and equally accurate as frozen section.

ABBREVIATIONS

CI = confidence interval; HGG = high-grade glioma; LGG = low-grade glioma; NPV = negative predictive value; PPV = positive predictive value; SRH = stimulated Raman histology; TTD = time to diagnosis.

OBJECTIVE

In some centers where brain tumor surgery is performed, the opportunity for expert intraoperative neuropathology consultation is lacking. Consequently, surgeons may not have access to the highest quality diagnostic histological data to inform surgical decision-making. Stimulated Raman histology (SRH) is a novel technology that allows for rapid acquisition of diagnostic histological images at the bedside.

METHODS

The authors performed a prospective blinded cohort study of 82 consecutive patients undergoing resection of CNS tumors to compare diagnostic time and accuracy of SRH simulation to the gold standard, i.e., frozen and permanent section diagnosis. Diagnostic accuracy was determined by concordance of SRH-simulated intraoperative pathology consultation with a blinded board-certified neuropathologist, with official frozen section and permanent section results.

RESULTS

Overall, the mean time to diagnosis was 30.5 ± 13.2 minutes faster (p < 0.0001) for SRH simulation than for frozen section, with similar diagnostic correlation: 91.5% (κ = 0.834, p < 0.0001) between SRH simulation and permanent section, and 91.5% between frozen and permanent section (κ = 0.894, p < 0.0001).

CONCLUSIONS

SRH-simulated intraoperative pathology consultation was significantly faster and equally accurate as frozen section.

In Brief

Stimulated Raman histology is a new technology that delivers accurate intraoperative pathological diagnosis and saves an average of 30.5 minutes compared to traditional frozen sectioning; it can be successfully implemented in neurosurgical operating rooms to shorten operative times.

Intraoperative neuropathological consultation efficiency is limited by current frozen section procedures. Frozen intraoperative sectioning, staining, and histopathological consultation still requires up to 20 minutes after the pathologist receives the specimen.22 Operative duration is a well-known contributor of surgical risk,7 with lengthier operating times correlating with increased risk of infection, deep vein thrombosis, and other perioperative complications.16,17 Childers and Maggard-Gibbons showed that 1 minute of operating room time can cost $36–$37, meaning that delays in intraoperative pathology consultation may have fiscal consequences for the hospital.3 Accurate histopathological assessment may be further complicated by freezing artifacts, poor staining quality, and poor section quality.14 Thus, a faster adjunct that circumvents these issues and updates the current frozen section workflow is desirable.

Stimulated Raman histology (SRH) is an emerging technology that may revolutionize the way intraoperative pathology consultations are conducted. SRH takes advantage of the unique vibrational characteristics of cellular macromolecules such as DNA, proteins, and lipids for the rapid generation of high-resolution microscopic images.10–12,19,25,26 Recently, clinical SRH microscopy has been shown to be capable of generating high-definition images of neurosurgical specimens comparable to conventional histological procedures.23 While the basic science and preliminary clinical data are promising, no clinical study has yet been published evaluating the use of SRH in a large patient sample with multiple CNS tumor types. In this prospective study, we hypothesized that the use of SRH in intraoperative neuropathology consultation would result in shorter time to diagnosis without compromising diagnostic accuracy.

Methods

The primary objectives of this study were twofold: 1) to measure the time to diagnosis via SRH and conventional histology, and 2) to measure the diagnostic accuracy and concordance of SRH versus conventional frozen section histology.

Patient Selection

This study was performed in accordance with STARD guidelines.2 After IRB approval, patients were prospectively consented for study-specific participation. Between November 7, 2018, and February 7, 2019, every consecutive patient who underwent resection of a brain or spine lesion in which the NIO Imaging System (Invenio Imaging) was used at our institution by the senior neurosurgeons (M.E.I., R.J.K., and A.D.L.) was included in this study and prospectively reviewed. Inclusion criteria were as follows: 1) patients who gave written and informed consent; 2) male or female patients aged 18 years or older undergoing open resection for a brain, spinal cord, or peripheral nerve tumor or lesion; and 3) patients in whom there was remaining leftover tumor specimen in excess of what was needed for conventional histopathological diagnosis. The exclusion criterion was patients undergoing stereotactic needle biopsy. Patient clinical information including demographic data, presenting symptoms, intraoperative findings, and postsurgical clinical status were obtained from the electronic medical record.

Intraoperative Workflow

All tissues were collected from patients who provided informed consent after IRB study approval. Intraoperative specimens were collected from patients undergoing clinically indicated open resection of brain, spinal cord, and peripheral nerve tumors or lesions and sent for frozen section and permanent section analysis (n = 82). Any remaining leftover specimen immediately underwent a 1-step squash preparation, and was analyzed by the NIO Imaging System (n = 82). Specimens for frozen section, permanent section, and SRH analysis were taken from the same area of the brain or spine tumor. Flow of participants throughout the study may be reviewed in Fig. 1.

FIG. 1.
FIG. 1.

Flow of participants throughout the study who underwent biopsy and SRH neuropathology consultation.

To image tissue with the clinical SRH microscope, a small (approximately 3-mm-thick) portion of fresh tissue was placed on a standard uncoated glass slide in the center of a small piece of two-sided tape and flattened to a thickness of 120 μm in a manner similar to a standard squash preparation. Normal saline (50 μl) was applied to the tissue and a coverslip was applied to the tissue and adhered to the slide, creating a chamber for imaging. This slide was then placed on a motorized stage and focused using standard transmission light microscopy. Using custom scripts in μ-Manager software and ImageJ software, two-channel (2845 cm−1 and 2930 cm−1) images were obtained in a mosaic fashion.

After analysis by the NIO Imaging System, files corresponding to each specimen were coded and each patient was assigned a unique study number. Files were then de-identified and stripped of all identifying information and uploaded to an online repository for storage. Standard histological techniques were used for frozen and permanent sections.

Histopathological Evaluation

A computer-based survey was developed consisting of SRH images from 82 patients and was presented to a blinded board-certified neuropathologist for review (S.H.G.). Identifying information was removed from the images. Only the patient’s age and tumor location were provided to the neuropathologist. No clinical information was provided in the evaluation of SRH, frozen section, or permanent section images. However, for evaluation of frozen and permanent sections, the pathologist had access to the patient’s chart, so pertinent clinical information may have been investigated. Reponses were recorded automatically by the survey software and scored for accuracy on two parameters: 1) for all cases, whether the image presented was lesional or nonlesional tissue; and 2) 1 of 8 categories was selected including high-grade glioma (HGG), low-grade glioma (LGG), metastasis, meningioma, pituitary adenoma, necrosis, normal brain, and other. Lesional versus nonlesional tissue diagnosis was treated as a prespecified dichotomous variable with either a positive or negative outcome for clinical frozen section, SRH, and permanent section diagnosis. The intraoperative clinical diagnosis and final pathological diagnoses were determined by a standard clinical protocol used by our Department of Pathology. Final WHO classification diagnoses using permanent sections served as the maximum possible score for each case and were considered the “true diagnosis.” Permanent sections provided the best quality for examination and diagnostic accuracy and were thus treated as the reference standard.6 In each case, the frozen section diagnosis, SRH diagnosis, and final diagnosis was used for statistical analysis. Survey responses for frozen section and SRH interpretation were compared as a measure of diagnostic concordance. Survey responses for frozen section and SRH images were compared to the “true” diagnosis from evaluation of permanent sections as a measure of diagnostic accuracy.

Time to Diagnosis

Time to diagnosis (TTD) was defined as elapsed time from when the specimen was removed to the moment the attending neurosurgeon received a final diagnosis from our neuropathologist. For frozen sections this would include all steps in a standard clinical protocol including tissue cryopreservation, sectioning, histology, and light microscopy and pathology review. For SRH, TTD was measured during a simulated pathology consultation. Intraoperatively, the time from specimen removal to SRH image acquisition was recorded prospectively. This time was added to the time taken for simulated pathology consultation—during which the blinded neuropathologist was told the patient’s age and lesion location and was able to evaluate the SRH images—to the time he rendered a diagnosis. A two-sample t-test was used to detect any significant difference in mean TTD between SRH and frozen section analysis.

Diagnostic Accuracy

To test the capability of SRH to accurately inform and contribute to intraoperative decision-making through neuropathological diagnosis, 82 specimens were imaged from 82 patients using both NIO imaging and conventional frozen sectioning and histochemical techniques. Sample portions from the same specimens were used for both frozen section histology and SRH imaging. Images generated from both SRH and frozen sectioning were reviewed by our blinded neuropathologist who responded with an intraoperative diagnosis for each case. Diagnostic accuracy was assessed through comparison of simulated diagnosis, with final diagnosis generated from evaluation of permanent sections. The same strategy was used to assess diagnostic performance in patient subgroups based on tumor pathology as a predefined approach to investigate causes of variability in diagnostic accuracy.

Statistical Analysis

SPSS software (version 24, IBM Corp.) was used for all statistical analyses. G*Power was used for all post hoc power analyses.8 Study recruitment was guided by an expected sensitivity of 95% for distinguishing lesional or nonlesional tissue with the index test (SRH) from previous studies at other prominent tertiary care centers with similar patient populations and volumes.13,23 Thus, according to Flahault et al., a minimum sample size of at least 50 cases was needed to achieve a minimal acceptable lower confidence limit of 0.8 with 95% expected sensitivity for a binary diagnostic test (lesional vs nonlesional).9

Cohen’s kappa score (κ) was used to assess the degree of agreement for each diagnosis.4 A kappa score was calculated to determine the extent of concordance between SRH and frozen section, SRH and permanent section, and frozen section and permanent section. Kappa was calculated based on agreement per category of diagnosis; the kappa value incorporates the agreement for glioblastoma, pituitary tumor, meningioma, etc., all separately and then calculates a concordance score based on the number correct in each category.

While kappa values are a statistically robust approach to comparing interest concordance, they may be difficult to interpret as a single summary statistic. Thus, we chose to also express agreement between diagnostic methods as the percentage correlation between two methodologies, in which a response was defined as “correlated” if both methods resulted in the same diagnosis. Finally, comparison of the TTD for SRH and frozen section diagnosis was also compared using two-sample t-tests.

Diagnostic accuracy was compared between SRH and frozen diagnosis through two-sample z-tests. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were compared using 95% confidence intervals (CIs). Diagnostic accuracy was compared for all samples as well as when subgrouped by pathology. Any samples in which the frozen or SRH diagnosis was indeterminate or unknown were categorized as “other.”

Results

Patient Demographics

A total of 82 patients were included in this study, 41 male (50%) and 41 female (50%). All patients underwent resection with intraoperative frozen section. On final histopathological diagnosis, 24 patients had meningioma (29.3%), 18 HGG (22.0%), 18 pituitary adenoma (22.0%), 8 metastasis (9.8%), 4 normal brain (4.9%), and 1 patient each had LGG (1.2%) and necrosis (1.2%). Eight patients were found to have “other” pathologies (schwannoma, cavernoma, etc.). A summary of patient demographic data may be reviewed in Supplementary Table 1.

Utility of SRH in Intraoperative Lesional Versus Nonlesional Tissue Diagnosis

Of 82 patient specimens, 80 (97.6%) were definitively identified as lesional tissue on permanent section while 2 samples were identified as nonlesional tissue (2.4%). With SRH, 78 samples were identified as lesional, with all 78 samples being correctly identified when compared to the gold standard of permanent section (100%). Four samples were identified as nonlesional, of which 2 were correct (50%) and 2 were incorrect (50%). With the frozen section procedure, 76 samples were identified as lesional, of which 76 were correctly identified (100%). Six samples were diagnosed as nonlesional tissue, of which 2 were correctly diagnosed (33.3%) and 4 were incorrectly diagnosed (66.7%). Additionally, the diagnostic capacity of SRH and frozen section analysis in identifying lesional versus nonlesional samples was assessed (Supplementary Table 2). Supplementary Tables 3 and 4, respectively, show SRH and frozen lesional diagnosis in relation to identification of lesional tissue on permanent sections. The diagnostic utility of SRH was excellent with a sensitivity of 97.5% (95% CI 90.4%–99.6%), specificity of 100% (95% CI 19.8%–100%), PPV of 100% (95% CI 94.1%–100%), and NPV of 50% (95% CI 9.2%–90.8%). The diagnostic capability of frozen sectioning was similar with a sensitivity of 95% (95% CI 87%–98.4%), specificity of 100% (95% CI 19.8%–100%), PPV of 100% (95% CI 94%–100%), and NPV of 33.3% (95% CI 6%–75.9%; Supplementary Tables 13).

Concordance and Correlation Between SRH, Frozen Section, and Permanent Section Diagnosis

SRH-based diagnosis and permanent section diagnosis had near-perfect concordance (κ = 0.834). Frozen and permanent section diagnoses showed a similar degree of agreement (κ = 0.894), as did frozen sections and SRH images (κ = 0.759). All κ values were statistically significant at p < 0.0001 (Fig. 2).

FIG. 2.
FIG. 2.

Diagnostic agreement between SRH, frozen sections (F), and permanent sections (P). Diagnostic performance using SRH versus permanent section and frozen versus permanent section revealed high concordance (κ = 0.834 and 0.894, respectively). Diagnostic performance using SRH versus frozen section also showed high concordance (κ = 0.759). The 95% CIs indicated no significant differences in diagnostic agreement between SRH, frozen section, and permanent section diagnostic modalities. All κ values were statistically significant at p < 0.0001.

The total proportion of samples in which the SRH and permanent section diagnoses correlated was 91.5%. The proportion of correlated samples for frozen versus permanent section diagnosis was also 91.5%. When comparing SRH and frozen section diagnosis, the percentage correlation was 87.8% (Supplementary Table 5).

Diagnostic Accuracy of SRH

SRH correctly predicted the final diagnosis in 91.5% of cases, not significantly different than the accuracy of frozen sections, which was also 91.5% (p = 1). Accuracy by lesion type is detailed in Supplementary Tables 5 and 6.

Comparison of SRH and Frozen Section Diagnostic Workflow

The mean TTD for SRH (± standard deviation) was 10.14 ± 6.93 minutes. This was significantly shorter than the TTD for frozen section analysis (40.65 ± 11.4 minutes, p < 0.0001; Fig. 3). The mean difference in TTD between SRH and frozen section was 30.49 ± 13.24 minutes (p < 0.0001). A post hoc power analysis indicated a large effect size of 4.01 using Cohen’s criteria5 with an alpha of 0.05 and a power of 1.000.

FIG. 3.
FIG. 3.

A: Histogram of time to diagnosis made using SRH versus frozen diagnosis. B: Summary of comparison of SRH and frozen histopathology TTD. TTD (in minutes) for SRH-generated histological images was significantly shorter than time to diagnosis for frozen sections. The mean difference in time taken to reach a diagnosis between SRH imaging and frozen sectioning was 30.49 ± 13.24 minutes. Error bars indicate 95% CIs of the mean time to diagnosis for SRH and frozen sections (*p < 0.0001). Figure is available in color online only.

Discussion

SRH has been investigated in both preclinical and clinical settings. SRH successfully distinguished between neoplastic and healthy tissue in in vivo human glioblastoma xenograft mouse models when distinction was not possible using standard H & E staining with light microscopy.15 Additionally, in a sample of 30 neurosurgical specimens, Orringer et al. showed near-perfect concordance between evaluation of SRH and conventionally imaged specimens when presented to a panel of 3 board-certified neuropathologists.23 High concordance was noted both in diagnostic evaluation and in distinguishing lesional versus nonlesional tissues. A second study by Hollon et al. again showed high concordance of SRH and conventional histology in the diagnosis of 25 pediatric neurosurgical specimens (20 brain tumors, 5 normal).13

We report the largest clinical series of simulated pathology consultation utilizing the SRH NIO Imaging System for intraoperative histology in the evaluation of a variety of different CNS tumor types. Additionally, ours is the first study to evaluate the utility in diagnosing a variety of tumor types.

Frozen sectioning offers several inherent advantages for intraoperative neuropathological consultation. First, frozen sections and intraoperative touch preps may reveal the following features much more clearly: detailed nuclear morphology (chromatin abnormalities, nuclear membrane irregularities, etc.); cytoplasmic detail and tinctorial qualities such as granularity, basophilia, and eosinophilia; cell border definition, which helps in identifying epithelial morphology; and identification of fibrillary cell processes and fibrillary background in glial tumors. These details help the pathologist to be more confident about identifying inflammatory pseudotumors, reactive tissues adjacent to abscesses, cerebral infarcts, lymphomas, and demyelinating lesions, all of which can mimic neoplastic lesions. Conversely, certain neoplastic processes may challenge the pathologist by an overwhelming presence of lymphocytic inflammation that may mask the tumor cells, such as CNS germinoma.

SRH analysis using the NIO Imaging System has undeniable benefits that justify its potential as a future neurosurgical tool that may supplement frozen sections as a method for rapid diagnostic consultation. While frozen sections have higher resolution, they require tissue freezing, sectioning, and labeling, which may create processing artifacts that may hinder analysis.13,21,24 SRH, which does not require these processing steps, not only saves time but also precludes any distortion of the tissue section during the process, which may contribute to diagnostic error.13,21,24 Intraoperative neuropathological diagnoses were able to be made in a quarter of the amount of time using SRH compared to frozen sections, without sacrificing diagnostic accuracy (Fig. 3).

In addition, it is clear that SRH may provide additional information about lesions versus nonlesional tissue that would be difficult to assess through conventional histological preparation. As many hypercellular brain tumors have significantly lower lipid content and higher protein content compared to the normal parenchyma, SRH images can often clearly delineate the tumor borders of well-marginated tumors such as metastases and meningiomas, as well as tissue cellularity of gliomas (Fig. 4).1,18 Microvascular proliferation in malignant brain tissue, a key hallmark of HGGs not present in the normal brain,20 is also readily apparent on SRH imaging. Compared to standard H & E staining of frozen sections, invading microvasculature appears dark and has high contrast on SRH visualization, allowing for easier identification. Additionally, tumor cells interspersed with background neurites are much more evident on SRH than H & E staining (Fig. 4A and B), making it readily apparent when a tumor is intraaxial.

FIG. 4.
FIG. 4.

Selected tumor types demonstrating hallmark histological findings on both H & E frozen section and SRH. A and B: Glioblastoma on frozen section and SRH, respectively. Cellular proliferation, high cell density, necrosis, and vascular proliferation are evident. Panel B demonstrates hypercellular tumor with background neurites (arrow), providing convincing evidence that the tumor is intraaxial and infiltrating. Such background neurites are not readily visible on frozen section (A). C and D: Meningioma on frozen section and SRH, respectively. Psammoma bodies are seen (arrows). E and F: Pituitary adenoma on frozen section and SRH, respectively. Characteristic homogeneous cytoarchitecture is noted. Figure is available in color online only.

One critical pathologic distinction to make is whether the brain lesion is lymphoma or nonlymphoma. Because lymphoma is treated with chemotherapy and radiation therapy rather than aggressive resection, early intraoperative pathologic diagnosis of lymphoma may prompt the surgeon to end the resection and spare the patient a lengthier surgery. As the presence of vasculature rules out lymphoma, and as vasculature is easily visualized on SRH, SRH may be an ideal modality to distinguish between lymphoma and other resectable lesions.

The use of SRH imaging does not minimize the need for board-certified neuropathologists in the operating room, but rather provides another intraoperative tool in addition to the frozen sections that neuropathologists can utilize.

The SRH diagnosis was wrong in comparison to the permanent section diagnosis in 7 cases. In 5 of these 7 cases, the SRH diagnosis was necrosis and the permanent section diagnosis was HGG (it was meningioma in 1 case). These errors were likely due to sampling error; some features of HGG were detected such as necrosis, but not enough to confidently diagnose HGG on SRH. Sampling error is also an issue in frozen sections, but because SRH offers a faster time to diagnosis, it is less cumbersome to analyze more tissue with SRH in order to confirm HGG when it is suspected.

In 2 cases the SRH diagnosis was meningioma but the permanent section diagnosis was HGG. While background neurites (that belong to the infiltrated brain parenchyma) and blood vessels are easier to visualize on SRH than frozen sections (and thus may greatly facilitate the diagnosis of HGG), details of nuclear morphology, background fibrillarity, or presence of fibrillary processes are much harder to detect on SRH. Thus, the absence of such histological features makes the diagnosis of HGG over other hypercellular lesions such as meningioma or metastasis more difficult. With additional experience using SRH and with the clinical information of intraaxial versus extraaxial lesions, we believe this diagnosis will become easier. However, evaluation with a large study is needed for complete assessment. Future research will leverage this digitalized information to search for additional tumor characteristics that may be too time- or labor-intensive for human investigators to identify that are important for diagnosis and prognosis.

This study had several limitations. First, our comparison of diagnostic accuracy using SRH and frozen sections should be interpreted with caution due to the large variability indicated by 95% CIs. This may be due to heterogeneity in our sample population and tumor pathology. Accuracy may be overestimated due to patient selection bias and minimal nonlesional tissue available for assessment. Additionally, specimen sampling error may affect measured accuracy.

The aim of our study was to measure the most essential interval in an intraoperative histological consultation, i.e., the time between when the specimen is taken from the surgical field and the time the pathologist calls in with a diagnosis. The most ideal study design to simulate this interval would be to perform both SRH and frozen sectioning in parallel and utilize two reading pathologists; however, logistic, staffing, and technological barriers prevented execution of this ideal study design. However, we believe our study design was rigorous enough to have clinical relevance, and thus demonstrate that TTD with SRH would be substantially shorter than conventional frozen sections.

This was a pilot study to evaluate intraoperative workflow feasibility, histopathological accuracy, and efficiency of SRH. Stereotactic needle biopsies were not included in our pilot study design because of the small anticipated tissue volume available for diagnosis; typically, minimal tissue is removed, all of which is required for conventional histopathological diagnosis. There is typically no excess tissue that could be utilized for analysis with SRH. Therefore, stereotactic needle biopsies were excluded by our IRB protocol. While the histopathological accuracy and efficiency findings of our study would likely provide similar benefit for stereotactic needle biopsy cases, further studies evaluating intraoperative time savings with SRH for stereotactic needle biopsy cases clearly need to be performed. Significant time savings during open craniotomy would be expected, for example, if SRH was able to be used to diagnose a lesion that is highly sensitive to chemotherapy or radiation therapy (such as CNS lymphoma) early in the tumor debulking. This would likely prompt the surgeon to stop the resection and shorten the length of surgery.

In this pilot study, all tumor types that underwent open resection with enough excess specimen after tissue for frozen and permanent section was sent were included. Only 1 LGG was included in the study; it is possible that during the study period most of the patients with LGG underwent stereotactic needle biopsy, which was part of our exclusion criteria due to concerns that there would not be excess specimen for SRH after tissue was sent for regular histopathological diagnosis. Of note, SRH accurately diagnosed the LGG.

The largest cohort in the study was patients with meningioma. Although some surgeons do not routinely send frozen sections on a well-circumscribed and obvious meningioma, the diagnostic accuracy, diagnostic efficiency, and feasibility of intraoperative workflow data are still pertinent and generalizable for neurosurgical operating rooms. Furthermore, in WHO grade II and III meningiomas with extensive local invasion into normal anatomical structures, frozen sectioning may provide useful information to the surgeon regarding resection boundaries of tumor infiltration. Thus, there is clinical utility in understanding the diagnostic capabilities of SRH for meningiomas.

In an effort to streamline the diagnostic comparison between SRH and frozen section analysis, we instructed the blinded board-certified neuropathologist to select from 8 prespecified diagnostic categories. As such, certain distinctions, such as atypical versus benign meningioma, were not handled by our analysis. The limit to which SRH is able to distinguish between such pathological entities more specific in nature than our 8 prespecified categories is thus beyond the scope of our study.

SRH clearly has applications in spinal tumors and peripheral nerve tumors, and we have used it as such during this study. Applications outside of neurosurgery, such as in determining tumor resection margins in head and neck surgery and other areas of surgical oncology, will likely become avid areas of future investigation.

We have shown that SRH neuropathology dramatically decreases the time needed for a neuropathological diagnosis to be conveyed to the attending neurosurgeon for intraoperative decision-making to 10.14 minutes for SRH simulation, from 40.65 minutes for frozen section (p < 0.0001).

Conclusions

SRH is a promising histological adjunct for intraoperative pathological consultation. SRH-simulated intraoperative pathology consultation was equally accurate and significantly faster compared to frozen section analysis. Our findings suggest that SRH can be successfully implemented in the workflow of a neurosurgical operating room, which can result in increased efficiency of intraoperative biopsy.

Acknowledgments

We would like to thank Linda Alberga for assistance with manuscript preparation and Roberto Suazo for assistance with figure preparation.

Disclosures

The NIO Imaging System was provided by Invenio Imaging. Dr. Levi reports receiving grant support from the NIH/NINDS and the Department of Defense, and teaching honoraria from the AANS and Medtronic. Dr. Ivan reports being a consultant to and receiving research funding from Medtronic and the NX Development Corporation.

Author Contributions

Conception and design: all authors. Acquisition of data: all authors. Analysis and interpretation of data: all authors. Drafting the article: all authors. 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: Eichberg. Statistical analysis: Eichberg, Shah. Study supervision: Ivan.

Supplemental Information

Online-Only Content

Supplemental material is available with the online version of the article.

References

  • 1

    Bentley JN, Ji M, Xie XS, Orringer DA: Real-time image guidance for brain tumor surgery through stimulated Raman scattering microscopy. Expert Rev Anticancer Ther 14:359361, 2014

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig L, et al.: STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ 351:h5527, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Childers CP, Maggard-Gibbons M: Understanding costs of care in the operating room. JAMA Surg 153:e176233, 2018

  • 4

    Cohen J: A coefficient of agreement for nominal scales. Educ Psychol Meas 20:3746, 1960

  • 5

    Cohen J: Statistical Power Analysis for the Behavioral Sciences. Boca Raton, FL: Taylor & Francis, 2013

  • 6

    Connolly JL, Schnitt SJ, Wang HH, Janina LA, Dvorak A, Dvorak HF: Role of the surgical pathologist in the diagnosis and management of the cancer patient, in Kufe DW, Pollock RE, Weichselbaum RR, et al. (eds): Holland-Frei Cancer Medicine, ed 6. Hamilton, ON: BC Decker, 2003

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Cruse PJ, Foord R: A five-year prospective study of 23,649 surgical wounds. Arch Surg 107:206210, 1973

  • 8

    Faul F, Erdfelder E, Lang AG, Buchner A: G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39:175191, 2007

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Flahault A, Cadilhac M, Thomas G: Sample size calculation should be performed for design accuracy in diagnostic test studies. J Clin Epidemiol 58:859862, 2005

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Freudiger CW, Pfannl R, Orringer DA, Saar BG, Ji M, Zeng Q, et al.: Multicolored stain-free histopathology with coherent Raman imaging. Lab Invest 92:14921502, 2012 (Erratum in Lab Invest 92:1661, 2012)

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Fu D, Lu FK, Zhang X, Freudiger C, Pernik DR, Holtom G, et al.: Quantitative chemical imaging with multiplex stimulated Raman scattering microscopy. J Am Chem Soc 134:36233626, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Fu Y, Huff TB, Wang HW, Wang H, Cheng JX: Ex vivo and in vivo imaging of myelin fibers in mouse brain by coherent anti-Stokes Raman scattering microscopy. Opt Express 16:19396–19409, 2008

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Hollon TC, Lewis S, Pandian B, Niknafs YS, Garrard MR, Garton H, et al.: Rapid intraoperative diagnosis of pediatric brain tumors using stimulated Raman histology. Cancer Res 78:278289, 2018

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Jaafar H: Intra-operative frozen section consultation: concepts, applications and limitations. Malays J Med Sci 13:412, 2006

  • 15

    Ji M, Orringer DA, Freudiger CW, Ramkissoon S, Liu X, Lau D, et al.: Rapid, label-free detection of brain tumors with stimulated Raman scattering microscopy. Sci Transl Med 5:201ra119, 2013

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Johans SJ, Garst JR, Burkett DJ, Grahnke K, Martin B, Ibrahim TF, et al.: Identification of preoperative and intraoperative risk factors for complications in the elderly undergoing elective craniotomy. World Neurosurg 107:216225, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Kim JY, Khavanin N, Rambachan A, McCarthy RJ, Mlodinow AS, De Oliveria GS Jr, et al.: Surgical duration and risk of venous thromboembolism. JAMA Surg 150:110117, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Kong K, Kendall C, Stone N, Notingher I: Raman spectroscopy for medical diagnostics—from in-vitro biofluid assays to in-vivo cancer detection. Adv Drug Deliv Rev 89:121134, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Kukura P, McCamant DW, Mathies RA: Femtosecond stimulated Raman spectroscopy. Annu Rev Phys Chem 58:461488, 2007

  • 20

    Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Jouvet A, et al.: The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114:97109, 2007

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Mahe E, Ara S, Bishara M, Kurian A, Tauqir S, Ursani N, et al.: Intraoperative pathology consultation: error, cause and impact. Can J Surg 56:E13E18, 2013

  • 22

    Novis DA, Zarbo RJ: Interinstitutional comparison of frozen section turnaround time. A College of American Pathologists Q-Probes study of 32868 frozen sections in 700 hospitals. Arch Pathol Lab Med 121:559567, 1997

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Orringer DA, Pandian B, Niknafs YS, Hollon TC, Boyle J, Lewis S, et al.: Rapid intraoperative histology of unprocessed surgical specimens via fibre-laser-based stimulated Raman scattering microscopy. Nat Biomed Eng 1:0027, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Taxy JB: Frozen section and the surgical pathologist: a point of view. Arch Pathol Lab Med 133:11351138, 2009

  • 25

    Yang Y, Li F, Gao L, Wang Z, Thrall MJ, Shen SS, et al.: Differential diagnosis of breast cancer using quantitative, label-free and molecular vibrational imaging. Biomed Opt Express 2:21602174, 2011

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Zhang X, Roeffaers MBJ, Basu S, Daniele JR, Fu D, Freudiger CW, et al.: Label-free live-cell imaging of nucleic acids using stimulated Raman scattering microscopy. ChemPhysChem 13:10541059, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Supplementary Materials

  • Collapse
  • Expand

Figure from Ulutas et al. (pp 72–83). Copyright Murat Ulutas (illustrations on left). Published with permission.

  • FIG. 1.

    Flow of participants throughout the study who underwent biopsy and SRH neuropathology consultation.

  • FIG. 2.

    Diagnostic agreement between SRH, frozen sections (F), and permanent sections (P). Diagnostic performance using SRH versus permanent section and frozen versus permanent section revealed high concordance (κ = 0.834 and 0.894, respectively). Diagnostic performance using SRH versus frozen section also showed high concordance (κ = 0.759). The 95% CIs indicated no significant differences in diagnostic agreement between SRH, frozen section, and permanent section diagnostic modalities. All κ values were statistically significant at p < 0.0001.

  • FIG. 3.

    A: Histogram of time to diagnosis made using SRH versus frozen diagnosis. B: Summary of comparison of SRH and frozen histopathology TTD. TTD (in minutes) for SRH-generated histological images was significantly shorter than time to diagnosis for frozen sections. The mean difference in time taken to reach a diagnosis between SRH imaging and frozen sectioning was 30.49 ± 13.24 minutes. Error bars indicate 95% CIs of the mean time to diagnosis for SRH and frozen sections (*p < 0.0001). Figure is available in color online only.

  • FIG. 4.

    Selected tumor types demonstrating hallmark histological findings on both H & E frozen section and SRH. A and B: Glioblastoma on frozen section and SRH, respectively. Cellular proliferation, high cell density, necrosis, and vascular proliferation are evident. Panel B demonstrates hypercellular tumor with background neurites (arrow), providing convincing evidence that the tumor is intraaxial and infiltrating. Such background neurites are not readily visible on frozen section (A). C and D: Meningioma on frozen section and SRH, respectively. Psammoma bodies are seen (arrows). E and F: Pituitary adenoma on frozen section and SRH, respectively. Characteristic homogeneous cytoarchitecture is noted. Figure is available in color online only.

  • 1

    Bentley JN, Ji M, Xie XS, Orringer DA: Real-time image guidance for brain tumor surgery through stimulated Raman scattering microscopy. Expert Rev Anticancer Ther 14:359361, 2014

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 2

    Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig L, et al.: STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies. BMJ 351:h5527, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 3

    Childers CP, Maggard-Gibbons M: Understanding costs of care in the operating room. JAMA Surg 153:e176233, 2018

  • 4

    Cohen J: A coefficient of agreement for nominal scales. Educ Psychol Meas 20:3746, 1960

  • 5

    Cohen J: Statistical Power Analysis for the Behavioral Sciences. Boca Raton, FL: Taylor & Francis, 2013

  • 6

    Connolly JL, Schnitt SJ, Wang HH, Janina LA, Dvorak A, Dvorak HF: Role of the surgical pathologist in the diagnosis and management of the cancer patient, in Kufe DW, Pollock RE, Weichselbaum RR, et al. (eds): Holland-Frei Cancer Medicine, ed 6. Hamilton, ON: BC Decker, 2003

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Cruse PJ, Foord R: A five-year prospective study of 23,649 surgical wounds. Arch Surg 107:206210, 1973

  • 8

    Faul F, Erdfelder E, Lang AG, Buchner A: G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 39:175191, 2007

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Flahault A, Cadilhac M, Thomas G: Sample size calculation should be performed for design accuracy in diagnostic test studies. J Clin Epidemiol 58:859862, 2005

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 10

    Freudiger CW, Pfannl R, Orringer DA, Saar BG, Ji M, Zeng Q, et al.: Multicolored stain-free histopathology with coherent Raman imaging. Lab Invest 92:14921502, 2012 (Erratum in Lab Invest 92:1661, 2012)

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 11

    Fu D, Lu FK, Zhang X, Freudiger C, Pernik DR, Holtom G, et al.: Quantitative chemical imaging with multiplex stimulated Raman scattering microscopy. J Am Chem Soc 134:36233626, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 12

    Fu Y, Huff TB, Wang HW, Wang H, Cheng JX: Ex vivo and in vivo imaging of myelin fibers in mouse brain by coherent anti-Stokes Raman scattering microscopy. Opt Express 16:19396–19409, 2008

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 13

    Hollon TC, Lewis S, Pandian B, Niknafs YS, Garrard MR, Garton H, et al.: Rapid intraoperative diagnosis of pediatric brain tumors using stimulated Raman histology. Cancer Res 78:278289, 2018

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 14

    Jaafar H: Intra-operative frozen section consultation: concepts, applications and limitations. Malays J Med Sci 13:412, 2006

  • 15

    Ji M, Orringer DA, Freudiger CW, Ramkissoon S, Liu X, Lau D, et al.: Rapid, label-free detection of brain tumors with stimulated Raman scattering microscopy. Sci Transl Med 5:201ra119, 2013

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 16

    Johans SJ, Garst JR, Burkett DJ, Grahnke K, Martin B, Ibrahim TF, et al.: Identification of preoperative and intraoperative risk factors for complications in the elderly undergoing elective craniotomy. World Neurosurg 107:216225, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 17

    Kim JY, Khavanin N, Rambachan A, McCarthy RJ, Mlodinow AS, De Oliveria GS Jr, et al.: Surgical duration and risk of venous thromboembolism. JAMA Surg 150:110117, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 18

    Kong K, Kendall C, Stone N, Notingher I: Raman spectroscopy for medical diagnostics—from in-vitro biofluid assays to in-vivo cancer detection. Adv Drug Deliv Rev 89:121134, 2015

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 19

    Kukura P, McCamant DW, Mathies RA: Femtosecond stimulated Raman spectroscopy. Annu Rev Phys Chem 58:461488, 2007

  • 20

    Louis DN, Ohgaki H, Wiestler OD, Cavenee WK, Burger PC, Jouvet A, et al.: The 2007 WHO classification of tumours of the central nervous system. Acta Neuropathol 114:97109, 2007

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 21

    Mahe E, Ara S, Bishara M, Kurian A, Tauqir S, Ursani N, et al.: Intraoperative pathology consultation: error, cause and impact. Can J Surg 56:E13E18, 2013

  • 22

    Novis DA, Zarbo RJ: Interinstitutional comparison of frozen section turnaround time. A College of American Pathologists Q-Probes study of 32868 frozen sections in 700 hospitals. Arch Pathol Lab Med 121:559567, 1997

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 23

    Orringer DA, Pandian B, Niknafs YS, Hollon TC, Boyle J, Lewis S, et al.: Rapid intraoperative histology of unprocessed surgical specimens via fibre-laser-based stimulated Raman scattering microscopy. Nat Biomed Eng 1:0027, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 24

    Taxy JB: Frozen section and the surgical pathologist: a point of view. Arch Pathol Lab Med 133:11351138, 2009

  • 25

    Yang Y, Li F, Gao L, Wang Z, Thrall MJ, Shen SS, et al.: Differential diagnosis of breast cancer using quantitative, label-free and molecular vibrational imaging. Biomed Opt Express 2:21602174, 2011

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 26

    Zhang X, Roeffaers MBJ, Basu S, Daniele JR, Fu D, Freudiger CW, et al.: Label-free live-cell imaging of nucleic acids using stimulated Raman scattering microscopy. ChemPhysChem 13:10541059, 2012

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 3081 253 0
Full Text Views 1299 501 157
PDF Downloads 815 157 23
EPUB Downloads 0 0 0