Identifying brain tumors by differential mobility spectrometry analysis of diathermy smoke

Ilkka Haapala Unit of Neurosurgery, Tampere University Hospital;

Search for other papers by Ilkka Haapala in
Current site
Google Scholar
PubMed
Close
 MD
,
Markus Karjalainen Faculty of Medicine and Health Technology, Tampere University;

Search for other papers by Markus Karjalainen in
Current site
Google Scholar
PubMed
Close
 MSc
,
Anton Kontunen Faculty of Medicine and Health Technology, Tampere University;

Search for other papers by Anton Kontunen in
Current site
Google Scholar
PubMed
Close
 MSc
,
Antti Vehkaoja Faculty of Medicine and Health Technology, Tampere University;

Search for other papers by Antti Vehkaoja in
Current site
Google Scholar
PubMed
Close
 MSc, PhD, DSc(Tech)
,
Kristiina Nordfors Department of Pediatrics, Tampere University Hospital;

Search for other papers by Kristiina Nordfors in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Hannu Haapasalo Fimlab Laboratories Ltd., Tampere University Hospital;

Search for other papers by Hannu Haapasalo in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Joonas Haapasalo Unit of Neurosurgery, Tampere University Hospital;
Faculty of Medicine and Health Technology, Tampere University;

Search for other papers by Joonas Haapasalo in
Current site
Google Scholar
PubMed
Close
 MD, PhD
,
Niku Oksala Faculty of Medicine and Health Technology, Tampere University;
Centre for Vascular Surgery and Interventional Radiology, Tampere University Hospital; and

Search for other papers by Niku Oksala in
Current site
Google Scholar
PubMed
Close
 MD, PhD, DSc(Med)
, and
Antti Roine Faculty of Medicine and Health Technology, Tampere University;
Department of Surgery, Tampere University Hospital, Hatanpää Hospital, Tampere, Finland

Search for other papers by Antti Roine in
Current site
Google Scholar
PubMed
Close
 MD, PhD
Full access

OBJECTIVE

There is a need for real-time, intraoperative tissue identification technology in neurosurgery. Several solutions are under development for that purpose, but their adaptability for standard clinical use has been hindered by high cost and impracticality issues. The authors tested and preliminarily validated a method for brain tumor identification that is based on the analysis of diathermy smoke using differential mobility spectrometry (DMS).

METHODS

A DMS connected to a special smoke sampling system was used to discriminate brain tumors and control samples ex vivo in samples from 28 patients who had undergone neurosurgical operations. They included meningiomas (WHO grade I), pilocytic astrocytomas (grade I), other low-grade gliomas (grade II), glioblastomas (grade IV), CNS metastases, and hemorrhagic or traumatically damaged brain tissue as control samples. Original samples were cut into 694 smaller specimens in total.

RESULTS

An overall classification accuracy (CA) of 50% (vs 14% by chance) was achieved in 7-class classification. The CA improved significantly (up to 83%) when the samples originally preserved in Tissue-Tek conservation medium were excluded from the analysis. The CA further improved when fewer classes were used. The highest binary classification accuracy, 94%, was obtained in low-grade glioma (grade II) versus control.

CONCLUSIONS

The authors’ results show that surgical smoke from various brain tumors has distinct DMS profiles and the DMS analyzer connected to a special sampling system can differentiate between tumorous and nontumorous tissue and also between different tumor types ex vivo.

ABBREVIATIONS

CA = classification accuracy; DMS = differential mobility spectrometry; GBM = glioblastoma; LDA = linear discriminant analysis; LGG = low-grade glioma; LOOCV = leave-one-out cross-validation; OCT = optical coherence tomography; REIMS = rapid evaporate ionization mass spectrometry; 10-f-CV = 10-fold cross-validation.

OBJECTIVE

There is a need for real-time, intraoperative tissue identification technology in neurosurgery. Several solutions are under development for that purpose, but their adaptability for standard clinical use has been hindered by high cost and impracticality issues. The authors tested and preliminarily validated a method for brain tumor identification that is based on the analysis of diathermy smoke using differential mobility spectrometry (DMS).

METHODS

A DMS connected to a special smoke sampling system was used to discriminate brain tumors and control samples ex vivo in samples from 28 patients who had undergone neurosurgical operations. They included meningiomas (WHO grade I), pilocytic astrocytomas (grade I), other low-grade gliomas (grade II), glioblastomas (grade IV), CNS metastases, and hemorrhagic or traumatically damaged brain tissue as control samples. Original samples were cut into 694 smaller specimens in total.

RESULTS

An overall classification accuracy (CA) of 50% (vs 14% by chance) was achieved in 7-class classification. The CA improved significantly (up to 83%) when the samples originally preserved in Tissue-Tek conservation medium were excluded from the analysis. The CA further improved when fewer classes were used. The highest binary classification accuracy, 94%, was obtained in low-grade glioma (grade II) versus control.

CONCLUSIONS

The authors’ results show that surgical smoke from various brain tumors has distinct DMS profiles and the DMS analyzer connected to a special sampling system can differentiate between tumorous and nontumorous tissue and also between different tumor types ex vivo.

In Brief

There is a need for intraoperative tissue identification technology in neurosurgery. The authors developed a method based on differential mobility spectrometry (DMS) analysis of diathermy smoke and tested it to discriminate brain tumors and control samples ex vivo. To their knowledge, this is the first demonstration of DMS analysis in brain tumor identification. Such a system could potentially become an affordable, user-friendly, and near–real-time instrument for intraoperative brain tumor tissue identification.

In present clinical practice, frozen section analysis is the gold standard for intraoperative tumor identification, yet the histopathological examination is often limited to a very small number of separate samples and the diagnosis is achieved with a delay.15 Especially in cases of low-grade tumors, frozen section analysis is difficult, and often the pathologist can only provide a morphological description of the tumor, with a lot of interobserver variability. Also, the assessment of more accurate molecular characteristics of the tumor is achieved only after several days and extensive tissue preparation.11 Furthermore, a precise intraoperative tissue identification would, in many cases, have a direct effect on the surgeon’s resection strategy. For example, in the case of a low-grade glioma (LGG), a surgeon would attempt an aggressive gross-total resection for maximal survival gain, whereas in glioblastoma (GBM) resections it is especially important to preserve remaining functions and be careful with eloquent cerebral areas.4,14 Therefore, there is a need for real-time, intraoperative tissue identification technology.

Several technologies are under development for intraoperative brain tumor identification. Optical coherence tomography (OCT) provides a high-resolution microscopic view of the tissue by illuminating the tissue with near-infrared light and measuring the reflected light with a spectrometer. OCT has been used to provide structural information about the dissected tissue.3

Raman spectroscopy is a modality that gives spectral tissue characteristics based on molecular signatures resulting from inelastic scattering of incident light. The resulting spectrum provides a fingerprint by which different molecular species can be identified. Raman spectroscopy reached sensitivity of 93% and specificity of 91% in binary classification when used for WHO grades II–IV brain tumor detection.8

The rapid evaporative ionization mass spectrometry (REIMS) setup, developed by the Imperial College, London, United Kingdom, has already been incorporated into a commercialized device by Waters Corp. as the “intelligent knife” or iKnife. It has reached an impressive identification capability with 100% of the given diagnoses matching the postoperative histopathology.1 A mass spectrometry–based sampling system has also been coupled with the Cavitron Ultrasonic Surgical Aspirator (CUSA) with a formidable ex vivo identification performance of 100% in GBMs and healthy brain samples, which the authors speculated to be a slight overestimation due to the limited number of samples.16 There is also a laser-based adoption of iKnife REIMS technology. Laser-based sampling systems such as PIRL and SpirerMass have also been developed.5,20 Both have been validated in analysis of animal tissues, but data from human tissues are not yet available. MasSpec Pen is yet another mass spectrometry–based sampling approach that uses water flushing of the analyzed tissue as the sampling modality.21 The system has achieved high performance in ex vivo analysis on cancerous tissues and has been demonstrated in vivo in a murine model, although sensitivity and specificity were not reported. On their website, the team states that research on MasSpec Pen in CNS malignancies is underway.

Electrosurgical resection is currently a widely used technique in neurosurgery. It produces surgical smoke, which is evacuated from the resection cavity to improve visibility and to avoid inhalation of toxic smoke. Surgical smoke carries information about the excised tissue in the form of biomarkers or tissue-specific metabolites.17 Electric discharge of the diathermy blade disperses particulates and causes evaporation of molecules from the tissue. The gold standard for the analysis of such molecules is mass spectrometry, which identifies molecules based on their mass-to-charge ratio. In various REIMS studies, the phospholipid content of the tissue has been identified as a key distinguishing factor.1 Mass spectrometry operates in a high vacuum, which poses considerable costs and mechanical challenges. In particular, ion mobility spectrometry (IMS) characterizes substances in atmospheric pressure based on cross-sectional area and electrical charge of the molecules.18 Differential mobility spectrometry (DMS) is an advanced adaptation of IMS. It uses a high-frequency electric field to break molecule clusters, thus creating additional information about the cluster strength that can be used in the analysis. Compared with mass spectrometry, the DMS is technically less complex and does not require a vacuum. Therefore, DMS is potentially a more affordable solution with less required maintenance and a smaller size.

Previously, the DMS has been studied as a diagnostic tool in various cancers with promising results.7 We have previously shown that DMS is capable of discriminating animal tissues and benign and malignant human breast tissue with high accuracy from surgical smoke.10,19 To our knowledge, this is the first time when DMS is used for brain tumor identification and classification.

Methods

Tissue Samples

We retrospectively obtained samples from 28 patients who had undergone neurosurgical operations in Tampere University Hospital between 2005 and 2017. The original samples included 6 meningiomas (WHO grade I), 3 pilocytic astrocytomas (grade I), 3 other grade II LGGs, 9 GBMs (grade IV), 5 CNS metastases, and 3 control samples. Primary tumors for the metastases were lung, endocervix, breast, endometrium, and prostate carcinomas. A total of 9 GBM samples were collected from 8 patients; otherwise, we had one sample per patient. The control samples were hemorrhagic or traumatically damaged brain tissue. Seventeen samples were originally preserved in Tissue-Tek (Sakura Finetek), a gel-like medium consisting of polyethylene glycol and polyvinyl alcohol.2 It penetrates the tissue and preserves the sample’s suitability for morphological examination with a microscope. Diagnoses of the tumor types were obtained from an experienced neuropathologist. The study was approved by the ethics review board of Pirkanmaa Hospital District, Finland.

All samples were stored in a freezer at −70°C. The samples were thawed and cut into smaller specimens that were placed in a custom-made aluminum well plate. The wells were 3.9 mm deep and 3 mm in diameter. The size of the smaller specimens was determined by the size of the wells. Each well contained a fitting of 3–5 wt/wt agar and 0.9 wt/wt NaCl to simulate body conductivity and to avoid arcing from the blade to the well plate that acted as a grounding electrode for the electrosurgical blade. Samples were analyzed with the DMS system in 9 sessions with a total numer of 694 tissue specimens prepared; 331 specimens were derived from patient samples preserved in Tissue-Tek. The specimens included in total 121 meningiomas, 154 CNS metastases, 35 pilocytic astrocytomas, 257 GBMs, 32 LGGs, and 20 control samples.

Tissue Analysis System

The measurement system consisted of 4 distinct parts: a tissue sampling unit (I), sample conditioning unit (II), DMS sensor (III), and computational data analysis (IV) as illustrated in Fig. 1 (a photograph of the system appears in Supplemental Fig. 2). The sampling system is described in detail in an earlier study.10

FIG. 1.
FIG. 1.

Illustration of the measurement system consisting of a tissue sampling unit (I), sample conditioning unit (II), DMS sensor (III), and computational data analysis (IV). The tissue sampling unit further includes a diathermy power unit (A), robot stage (B), sample well plate (C), surgical smoke evacuator (D), filter and dilution system (E), and ENVI-AMC differential mobility spectrometer (F). The numbered values depict the input flow rate to each device. The LDA is a sketch of the operating principle, where 2 classes, + and o, are projected onto a 2D plane.

An automated tissue sampling unit (I) was developed to produce standardized incisions from tissue samples. This eliminated the sampling variations arising from free-hand use of the surgical blade. The sampling unit consisted of a commercial diathermy blade (Itkacut 350 MB, Innokas Medical) used with a nominal 100-W cut mode. The high power was used to avoid adhesion of the samples to the blade. The blade was attached to a modified 3D printer (RepRap Mendel Prusa i3, Kit Printer 3D) and a smoke evacuator (Surtron Evac, LED SpA), which directed the surgical smoke to the sample conditioning unit. The diathermy blade incised each specimen once. The duration of time from the capture of smoke to the readout was 1 minute. The blade used in this study was 2.5 mm in width; thus, the spatial resolution is in the same order.

The sample conditioning unit (II) separated contaminating particulate matter from the smoke and kept the sample concentration constant. In addition to biomarkers, electrosurgery produces a significant amount of particulate matter,9 which has 2 unwanted consequences. First, the particulate matter contaminates the sensor and second, it causes increased carryover. Thus, an electric particle filter was used as the first active component in the sample conditioning unit for removing the particulate matter from the sampling stream. After filtering, we diluted the sample with cleaned air, using variable dilution ratio from 1:85 to 1:440 due to the high concentration of the sample. Compressed diluting air was cleaned with active carbon and 5-Å molecular sieves. Moreover, the sample flow path of the sample conditioning unit was heated to 40°C–50°C to minimize carryover from previous sampling events.

The sample was passed from the conditioning unit to a DMS analyzer (III) (ENVI-AMC, Environics Oy). The DMS consists of a radioactive 241Am source to ionize the molecules and a sensor unit, which separates ions with low and high electric fields. A 2D spectrum is obtained by altering the radiofrequency electric field strength and direct current field. With these fields, ions partially separate and partially stay clustered. This ion separation can be used as a characteristic dispersion spectrum, or “smell fingerprints,” of the sample, as seen in Fig. 2.

FIG. 2.
FIG. 2.

Example spectral images from a single control tissue sample (left) and a GBM (right). The areas of the dispersion plot that are in the 1- to 2-V compensation voltage range are the most prone to exhibit the differences in the DMS spectra between the tissues. In this instance, the differential mobility peaks near the 2-V compensation voltage differ in shape and size. VC = compensation voltage; VRF = peak-to-peak amplitude of the radiofrequency waveform voltage.

Analysis of the measurement data (IV) was performed using MATLAB (MathWorks Ltd.)–based processing and classification algorithms. The samples were classified into 7, 5, or 2 categories with linear discriminant analysis (LDA). LDA is a simple and commonly used supervised classification method that is based on feature projection to reduce the dimensionality of the initial measurement data. The principle of LDA classification is presented in Fig. 3. The LDA classification was performed using the full raw data matrices (1620 pixels) of the DMS output spectra as features. To avoid overfitting, leave-one-out cross-validation (LOOCV) or 10-fold cross-validation (10-f-CV) was performed, depending on the sample size. LOOCV is a valid method for small sample size classifications, but when the size increases to hundreds, a more unbiased estimation is achieved with 10-f-CV. Furthermore, in cases in which the sample sizes of the binary classification were significantly skewed, the classes were balanced by only including a portion of the samples from the larger class in the classification. Exclusion of the samples from the larger class was randomized. In addition to the cross-validated and balanced LDA models, blind hierarchical clustering was also carried out to test unsupervised classification.

FIG. 3.
FIG. 3.

The principle of LDA classification. The LDA classification model aims to find a linear combination that best separates two or more classes based on given features. In this study, the features were the pixels in the DMS spectra, but for visualization purposes, the data for GBM and LGG are presented in 2 dimensions as Sammon projections of the DMS spectrum data.

Results

When classifying the full measurement data set to the 7 tissue classes, a classification accuracy (CA) of 50% was achieved. The average CA by chance would be 14%. The confusion matrix is presented in Table 1. Blind hierarchical clustering did not show relevant grouping.

TABLE 1.

Results of 7-class classification

Assigned Class
MeningiomaNo SampleMetPAGBMLGGControlTotalSensitivitySpecificityCA
True classMeningioma630222331012152%91%
No sample272001007596%99%
Met182636631115440%77%
PA211121711356%98%
GBM2707531445325756%68%
 LGG4110015203299%
 Control005311012099%
Total50%

Met = metastasis; PA = pilocytic astrocytoma.

Correctly classified samples appear in boldface type. Sensitivity is the percentage of correctly identified positive samples. Specificity is the percentage of correctly identified negative samples. In the sensitivity and specificity columns, the positive sample is determined as the sample on the given row, while all others are considered as negative samples. For example, in the first row for meningioma, there are 121 positive samples and 573 negative samples, which gives a sensitivity of 63/121 = 52% and specificity of 520/573 = 91%.

When only focusing on distinguishing CNS metastases from meningiomas, the cross-validated classifier produced a CA of 74% (sensitivity 71%, specificity 77%). For separating the control samples from tumors (the class consisted of 4 samples of each tumor type), the CA was 83% (sensitivity 78%, specificity 89%). In differentiation of LGGs (grade II) from GBMs, the CA was 94% (sensitivity 88%, specificity 100%). For the differentiation of pilocytic astrocytomas from diffusely infiltrating gliomas (grades II and IV), the CA was 70% (sensitivity 77%, specificity 63%). The differentiation of LGGs (grade II) from control samples produced a CA of 94% (sensitivity 97%, specificity 90%). Table 2 shows the results for each binary classification case.

TABLE 2.

Results of selected relevant binary classifications

ComparisonNo. of SamplesCASensitivitySpecificityPPVNPV
Tumor (+) vs control (–)*,4083%78%89%87%80%
Meningioma (+) vs met (–)27574%71%77%70%77%
LGG (grade II) (+) vs GBM (–)6494%88%100%100%89%
PA (+) vs diffusively infiltrating glioma (–; grades II & IV)7070%77%63%68%73%
LGG (+; grade II) vs control (–)5294%97%90%94%95%

NPV = negative predictive value; PPV = positive predictive value; + = positive; – = negative.

Positive predictive value is the ratio of true positives to all (predicted and true) positives. Negative predictive value is the ratio of true negatives to all (predicted and true) negatives.

Five randomly selected tumor samples of every class.

LOOCV.

10-f-CV.

Classifications were also conducted separately in samples that were not preserved in Tissue-Tek, which improved the CA significantly. The results for the classification of the remaining 5 classes are presented in Table 3. In addition, the binary classification (LDA with 10-f-CV) of meningioma from metastases without the Tissue-Tek samples produced a CA of 95% with 87% sensitivity and 98% specificity.

TABLE 3.

Classification results for samples not preserved in Tissue-Tek

Assigned Class
MeningiomaNo SampleMetPAGBMTotalSensitivitySpecificityCA
True classMeningioma3213033982%99%
No sample0740017599%99%
Met018221710280%90%
PA11101162938%99%
GBM0017010111886%89%
Total83%

Correctly classified samples appear in boldface type.

Discussion

Our results show that surgical smoke produced by various brain tumors has distinct DMS profiles. The DMS accompanied with a specialized sampling system can distinguish different brain tumors from each other ex vivo. With 7-class classification by LDA, we reached the CA of 50%, which is considerably better than the discrimination of 14% achieved by chance. We also carried out binary classifications in order to better simulate the factual intraoperative use of the DMS. In clinical reality, the physician often has some thought as to the tumor diagnosis even before surgery, based on imaging study findings, patient age, tumor location, and other factors. Therefore, the diagnostic alternatives for DMS can often be narrowed down beforehand. Binary classifications yielded significantly higher CAs. Particularly promising is the 94% CA in LGG (grade II) versus control setting. Low-grade glioma patients compose a subgroup that may have the greatest survival benefit from more accurate resection,14 and their identification by frozen section analysis is especially difficult. The CA also largely improved when the samples originally conserved in Tissue-Tek were excluded from the analysis, suggesting that Tissue-Tek was a major confounding factor. As an example, a Sammon projection image about the dispersion of GBMs and empty samples is shown in Supplemental Fig. 1.

From a clinician’s standpoint, a DMS-based tumor identification system has several advantages. It can be connected to the instrumentation already present in neurosurgical operating rooms and does not require a separate probe. The DMS provides data that can be translated by relatively simple algorithms to tissue classifications that can be interpreted by the surgical staff without substantial training. Thus, the system is user friendly, provides near–real-time information, and preserves the surgical workflow. OCT requires a visual analysis of the image and thus is ultimately a subjective method for cancer margin evaluation. The image analysis is sensitive to surgical artifacts (blood and cauterized tissue), and surgical personnel would need additional training and qualification to manage the analysis. In OCT, the analysis does not happen in vivo but instead the surgically removed tissue piece is imaged on a separate platform.3 Raman spectroscopy requires a specialized probe to be inserted into the resection cavity.8 Therefore, all such systems are likely to interfere with the surgeon’s workflow.

Compared with mass spectrometry, a clinical use of DMS is analogous. DMS is a technically less complex system than mass spectrometry and does not require a vacuum. The adaptability of the iKnife for standard clinical use has been hindered by its high cost, expertise required for its use, need for regular and intensive maintenance, and impractically large size for operating rooms. The same challenge applies to other described mass spectrometry–based methods. DMS is a potentially more affordable solution with less required maintenance and a smaller size.

A major limitation of our study is that we had only 29 patient samples from which we obtained 694 specimens for the analysis. However, in the studies of many competing solutions, the number of patient samples is similar or even smaller.3,8,16 Even though a small number of patients may introduce selection bias and affect the generalizability of the results, it was considered to be enough for a proof-of-concept study.

We used hemorrhagic or traumatically damaged brain tissue as control samples. Even though these samples were nontumorous, pathological injuries could have had an impact on metabolic profiles of these cells. Therefore, these specimens cannot be directly contrasted with healthy brain tissue.

When the entire sample is taken into consideration, our CA does not seem to match the identification capabilities of the competing solutions. The performance, however, is significantly improved when the samples processed with Tissue-Tek were excluded, with a total CA of 84% (5 classes) and CAs of greater than 90% in several binary scenarios. This suggests that tissue preservation medium is a significant confounding factor. Tissue-Tek penetrates the tissue and preserves it for morphological examination with a microscope.2 Yet because it penetrates the tissue, it is likely that the electrosurgical smoke will contain some of the medium, thereby disturbing the DMS classifier. Since 2016, Tissue-Tek has been a part of standard processing of brain tumor samples in the Department of Pathology, Tampere University Hospital, Finland. We collected our samples retrospectively, so all the samples newer than 2016 were conserved in Tissue-Tek.

Further shortcomings in classification accuracy may be explained by intratumoral heterogeneity. It is obvious that CNS metastases are histologically heterogeneous, since they arise from completely different primary tumors.12 Thus, it is expected that they can be troublesome for the classifier. Also, intratumoral histological heterogeneity is a common phenomenon in gliomas, especially in GBMs. Different areas of a glioma (grades II–IV) has been shown to vary in terms of cellular density, nuclear pleomorphism, necrosis, histological architecture, vasculature, and mitoses, which causes the WHO grade to vary within a tumor.13 Structural differences between separate tumor areas also cause variation in tissue impedance.22 We dissected the original tissue samples into smaller pieces. Therefore, different tissue pieces with the same pathological diagnosis most likely varied in terms of histological structure and bioelectrical qualities, which may have affected the results. The classifier also performed better with meningiomas, which are considered to be rather homogeneous tumors.6

In addition to histological heterogeneity, different brain tumors also vary a lot in terms of macroscopic composition. Therefore, even with the custom-made well plate, standardized incisions, and our best preparation of the samples, the depth of the incisions and the amount of smoke provided was somewhat varied. This issue was only partly managed with the supervised learning method in the data analysis (LDA) with its ability to reduce the variance and noise of the original data set. Since significantly better results were achieved by excluding samples preserved in Tissue-Tek, we examined the possibility of bias in distribution of sample types analyzed in different sessions with no significant bias observed.

In the future, the DMS will be tested to detect elevated levels of D-2-hydroxyglutarate and thereby intraoperatively determine the IDH mutation status of a tumor. Further studies are also needed to validate the DMS analyzer for tissue identification in vivo, with more accurate histopathological validation. Instead of the current diathermy blade, we will need to develop a bipolar scissors–integrated sampling system that conforms the routine instruments used by neurosurgeons. The speed of the analysis will also be improved by optimizing the DMS analysis window and the sampling system. Along with further in vivo research and technical development, we can also assess the system’s suitability for the evaluation of neurosurgical resection margins.

Conclusions

This proof-of-concept study demonstrates that surgical smoke from various brain tumors has distinct DMS profiles, and that the DMS connected to a special sampling system can differentiate between tumorous and nontumorous brain tissue and also between different tumor types ex vivo. A DMS analyzer could potentially become a simple, user-friendly, and affordable tool for intraoperative brain tumor tissue identification.

Acknowledgments

This study was supported by grants from the following foundations: Finnish Foundation for Technology Promotion (TES); Tampereen Tuberkuloosisäätiö (Tampere Tuberculosis Foundation); Emil Aaltonen Foundation; Finnish Cultural Foundation; Finnish Medical Foundation; Pediatric Research Foundation; Finnish Neurosurgical Society; Competitive State Research Financing of the Expert Responsibility area of the Tampere University Hospital and Pirkanmaa Hospital District grants 9s045, 151B03, 9T044, 9U042, 150618, 9U042, and 9V044; and Academy of Finland, grant 292477.

Disclosures

Mr. Karjalainen, Mr. Kontunen, Dr. Oksala, and Dr. Roine: direct stock ownership in Olfactomics Ltd. Dr. Roine: employee of Olfactomics Ltd.

Author Contributions

Conception and design: Karjalainen, H Haapasalo, J Haapasalo, Oksala, Roine. Acquisition of data: H Haapasalo, J Haapasalo. Analysis and interpretation of data: Haapala, Karjalainen, Kontunen, H Haapasalo, Roine. Drafting the article: Haapala, Karjalainen, Kontunen, Roine. Critically revising the article: all authors. Reviewed submitted version of manuscript: Haapala. Approved the final version of the manuscript on behalf of all authors: Haapala. Statistical analysis: Haapala, Karjalainen, Kontunen. Administrative/technical/material support: Karjalainen, Kontunen, Vehkaoja, Nordfors, H Haapasalo, J Haapasalo, Oksala, Roine. Study supervision: J Haapasalo, Oksala, Roine.

Supplemental Information

Online-Only Content

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

Previous Presentations

Portions of this work were presented in poster form at the 3rd Annual Symposium on Brain Tumors, Finnish Brain Tumor Research Association, Turku, Finland, October 26–27, 2017.

References

  • 1

    Balog J, Sasi-Szabó L, Kinross J, Lewis MR, Muirhead LJ, Veselkov K, et al.: Intraoperative tissue identification using rapid evaporative ionization mass spectrometry. Sci Transl Med 5:194ra93, 2013

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

    Barthel LK, Raymond PA: Improved method for obtaining 3-microns cryosections for immunocytochemistry. J Histochem Cytochem 38:13831388, 1990

  • 3

    Böhringer HJ, Lankenau E, Stellmacher F, Reusche E, Hüttmann G, Giese A: Imaging of human brain tumor tissue by near-infrared laser coherence tomography. Acta Neurochir (Wien) 151:507517, 2009

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Brown TJ, Brennan MC, Li M, Church EW, Brandmeir NJ, Rakszawski KL, et al.: Association of the extent of resection with survival in glioblastoma: a systematic review and meta-analysis. JAMA Oncol 2:14601469, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Fatou B, Saudemont P, Leblanc E, Vinatier D, Mesdag V, Wisztorski M, et al.: In vivo real-time mass spectrometry for guided surgery application. Sci Rep 6:25919, 2016

  • 6

    Filippi CG, Edgar MA, Uluğ AM, Prowda JC, Heier LA, Zimmerman RD: Appearance of meningiomas on diffusion-weighted images: correlating diffusion constants with histopathologic findings. AJNR Am J Neuroradiol 22:6572, 2001

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Fitzgerald JE, Bui ETH, Simon NM, Fenniri H: Artificial nose technology: status and prospects in diagnostics. Trends Biotechnol 35:3342, 2017

  • 8

    Jermyn M, Mok K, Mercier J, Desroches J, Pichette J, Saint-Arnaud K, et al.: Intraoperative detection of glioma invasion beyond MRI enhancement with Raman spectroscopy in humans. Proc SPIE 9318:93180D-193180D-6, 2015

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Karjalainen M, Kontunen A, Saari S, Rönkkö T, Lekkala J, Roine A, et al.: The characterization of surgical smoke from various tissues and its implications for occupational safety. PLoS One 13:e0195274, 2018

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

    Kontunen A, Karjalainen M, Lekkala J, Roine A, Oksala N: Tissue identification in a porcine model by differential ion mobility spectrometry analysis of surgical smoke. Ann Biomed Eng 46:10911100, 2018

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

    Longuespée R, Wefers AK, De Vita E, Miller AK, Reuss DE, Wick W, et al.: Rapid detection of 2-hydroxyglutarate in frozen sections of IDH mutant tumors by MALDI-TOF mass spectrometry. Acta Neuropathol Commun 6:21, 2018

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

    Nayak L, Lee EQ, Wen PY: Epidemiology of brain metastases. Curr Oncol Rep 14:4854, 2012

  • 13

    Paulus W, Peiffer J: Intratumoral histologic heterogeneity of gliomas. A quantitative study. Cancer 64:442447, 1989

  • 14

    Sanai N, Berger MS: Glioma extent of resection and its impact on patient outcome. Neurosurgery 62:753766, 2008

  • 15

    Santagata S, Eberlin LS, Norton I, Calligaris D, Feldman DR, Ide JL, et al.: Intraoperative mass spectrometry mapping of an onco-metabolite to guide brain tumor surgery. Proc Natl Acad Sci U S A 111:1112111126, 2014

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

    Schäfer KC, Balog J, Szaniszló T, Szalay D, Mezey G, Dénes J, et al.: Real time analysis of brain tissue by direct combination of ultrasonic surgical aspiration and sonic spray mass spectrometry. Anal Chem 83:77297735, 2011

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

    Schäfer KC, Dénes J, Albrecht K, Szaniszló T, Balog J, Skoumal R, et al.: In vivo, in situ tissue analysis using rapid evaporative ionization mass spectrometry. Angew Chem Int Ed Engl 48:82408242, 2009

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

    Schneider BB, Nazarov EG, Londry F, Vouros P, Covey TR: Differential mobility spectrometry/mass spectrometry history, theory, design optimization, simulations, and applications. Mass Spectrom Rev 35:687737, 2016

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

    Sutinen M, Kontunen A, Karjalainen M, Kiiski J, Hannus J, Tolonen T, et al.: Identification of breast tumors from diathermy smoke by differential ion mobility spectrometry. Eur J Surg Oncol 45:141146, 2019

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Tata A, Gribble A, Ventura M, Ganguly M, Bluemke E, Ginsberg HJ, et al.: Wide-field tissue polarimetry allows efficient localized mass spectrometry imaging of biological tissues. Chem Sci (Camb) 7:21622169, 2016

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

    Zhang J, Rector J, Lin JQ, Young JH, Sans M, Katta N, et al.: Nondestructive tissue analysis for ex vivo and in vivo cancer diagnosis using a handheld mass spectrometry system. Sci Transl Med 9:eaan3968, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Zhang Y, Xu S, Min W, Shen L, Zhang Y, Yue Z: Surg-25. A novel bio-impedance spectroscopy system real-time intraoperatively discriminates glioblastoma from brain tissue in mice. Neuro Oncol 19:240, 2017

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

Diagram from Kondziolka et al. (pp 1–2).

  • FIG. 1.

    Illustration of the measurement system consisting of a tissue sampling unit (I), sample conditioning unit (II), DMS sensor (III), and computational data analysis (IV). The tissue sampling unit further includes a diathermy power unit (A), robot stage (B), sample well plate (C), surgical smoke evacuator (D), filter and dilution system (E), and ENVI-AMC differential mobility spectrometer (F). The numbered values depict the input flow rate to each device. The LDA is a sketch of the operating principle, where 2 classes, + and o, are projected onto a 2D plane.

  • FIG. 2.

    Example spectral images from a single control tissue sample (left) and a GBM (right). The areas of the dispersion plot that are in the 1- to 2-V compensation voltage range are the most prone to exhibit the differences in the DMS spectra between the tissues. In this instance, the differential mobility peaks near the 2-V compensation voltage differ in shape and size. VC = compensation voltage; VRF = peak-to-peak amplitude of the radiofrequency waveform voltage.

  • FIG. 3.

    The principle of LDA classification. The LDA classification model aims to find a linear combination that best separates two or more classes based on given features. In this study, the features were the pixels in the DMS spectra, but for visualization purposes, the data for GBM and LGG are presented in 2 dimensions as Sammon projections of the DMS spectrum data.

  • 1

    Balog J, Sasi-Szabó L, Kinross J, Lewis MR, Muirhead LJ, Veselkov K, et al.: Intraoperative tissue identification using rapid evaporative ionization mass spectrometry. Sci Transl Med 5:194ra93, 2013

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

    Barthel LK, Raymond PA: Improved method for obtaining 3-microns cryosections for immunocytochemistry. J Histochem Cytochem 38:13831388, 1990

  • 3

    Böhringer HJ, Lankenau E, Stellmacher F, Reusche E, Hüttmann G, Giese A: Imaging of human brain tumor tissue by near-infrared laser coherence tomography. Acta Neurochir (Wien) 151:507517, 2009

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 4

    Brown TJ, Brennan MC, Li M, Church EW, Brandmeir NJ, Rakszawski KL, et al.: Association of the extent of resection with survival in glioblastoma: a systematic review and meta-analysis. JAMA Oncol 2:14601469, 2016

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 5

    Fatou B, Saudemont P, Leblanc E, Vinatier D, Mesdag V, Wisztorski M, et al.: In vivo real-time mass spectrometry for guided surgery application. Sci Rep 6:25919, 2016

  • 6

    Filippi CG, Edgar MA, Uluğ AM, Prowda JC, Heier LA, Zimmerman RD: Appearance of meningiomas on diffusion-weighted images: correlating diffusion constants with histopathologic findings. AJNR Am J Neuroradiol 22:6572, 2001

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 7

    Fitzgerald JE, Bui ETH, Simon NM, Fenniri H: Artificial nose technology: status and prospects in diagnostics. Trends Biotechnol 35:3342, 2017

  • 8

    Jermyn M, Mok K, Mercier J, Desroches J, Pichette J, Saint-Arnaud K, et al.: Intraoperative detection of glioma invasion beyond MRI enhancement with Raman spectroscopy in humans. Proc SPIE 9318:93180D-193180D-6, 2015

    • PubMed
    • Search Google Scholar
    • Export Citation
  • 9

    Karjalainen M, Kontunen A, Saari S, Rönkkö T, Lekkala J, Roine A, et al.: The characterization of surgical smoke from various tissues and its implications for occupational safety. PLoS One 13:e0195274, 2018

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

    Kontunen A, Karjalainen M, Lekkala J, Roine A, Oksala N: Tissue identification in a porcine model by differential ion mobility spectrometry analysis of surgical smoke. Ann Biomed Eng 46:10911100, 2018

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

    Longuespée R, Wefers AK, De Vita E, Miller AK, Reuss DE, Wick W, et al.: Rapid detection of 2-hydroxyglutarate in frozen sections of IDH mutant tumors by MALDI-TOF mass spectrometry. Acta Neuropathol Commun 6:21, 2018

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

    Nayak L, Lee EQ, Wen PY: Epidemiology of brain metastases. Curr Oncol Rep 14:4854, 2012

  • 13

    Paulus W, Peiffer J: Intratumoral histologic heterogeneity of gliomas. A quantitative study. Cancer 64:442447, 1989

  • 14

    Sanai N, Berger MS: Glioma extent of resection and its impact on patient outcome. Neurosurgery 62:753766, 2008

  • 15

    Santagata S, Eberlin LS, Norton I, Calligaris D, Feldman DR, Ide JL, et al.: Intraoperative mass spectrometry mapping of an onco-metabolite to guide brain tumor surgery. Proc Natl Acad Sci U S A 111:1112111126, 2014

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

    Schäfer KC, Balog J, Szaniszló T, Szalay D, Mezey G, Dénes J, et al.: Real time analysis of brain tissue by direct combination of ultrasonic surgical aspiration and sonic spray mass spectrometry. Anal Chem 83:77297735, 2011

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

    Schäfer KC, Dénes J, Albrecht K, Szaniszló T, Balog J, Skoumal R, et al.: In vivo, in situ tissue analysis using rapid evaporative ionization mass spectrometry. Angew Chem Int Ed Engl 48:82408242, 2009

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

    Schneider BB, Nazarov EG, Londry F, Vouros P, Covey TR: Differential mobility spectrometry/mass spectrometry history, theory, design optimization, simulations, and applications. Mass Spectrom Rev 35:687737, 2016

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

    Sutinen M, Kontunen A, Karjalainen M, Kiiski J, Hannus J, Tolonen T, et al.: Identification of breast tumors from diathermy smoke by differential ion mobility spectrometry. Eur J Surg Oncol 45:141146, 2019

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 20

    Tata A, Gribble A, Ventura M, Ganguly M, Bluemke E, Ginsberg HJ, et al.: Wide-field tissue polarimetry allows efficient localized mass spectrometry imaging of biological tissues. Chem Sci (Camb) 7:21622169, 2016

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

    Zhang J, Rector J, Lin JQ, Young JH, Sans M, Katta N, et al.: Nondestructive tissue analysis for ex vivo and in vivo cancer diagnosis using a handheld mass spectrometry system. Sci Transl Med 9:eaan3968, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation
  • 22

    Zhang Y, Xu S, Min W, Shen L, Zhang Y, Yue Z: Surg-25. A novel bio-impedance spectroscopy system real-time intraoperatively discriminates glioblastoma from brain tissue in mice. Neuro Oncol 19:240, 2017

    • Crossref
    • PubMed
    • Search Google Scholar
    • Export Citation

Metrics

All Time Past Year Past 30 Days
Abstract Views 1514 134 0
Full Text Views 297 124 43
PDF Downloads 287 98 28
EPUB Downloads 0 0 0