Mass spectrometry imaging as a tool for surgical …dunnlab.bwh.harvard.edu/wp-content/uploads/2015/04/...Mass spectrometry imaging as a tool for surgical decision-making David Calligaris,a†

Special feature: perspective

Received: 23 July 2013 Revised: 3 October 2013 Accepted: 10 October 2013 Published online in Wiley Online Library

(wileyonlinelibrary.com) DOI 10.1002/jms.3295

1178

Mass spectrometry imaging as a tool forsurgical decision-makingDavid Calligaris,a† Isaiah Norton,a† Daniel R. Feldman,b Jennifer L. Ide,a

Ian F. Dunn,a Livia S. Eberlin,c‡ R. Graham Cooks,c Ferenc A. Jolesz,d

Alexandra J. Golby,a,d Sandro Santagatab and Nathalie Y. Agara,d,e*

Despite significant advances in image-guided therapy, surgeons are still too often left with uncertainty when deciding toremove tissue. This binary decision between removing and leaving tissue during surgery implies that the surgeon should
be able to distinguish tumor from healthy tissue. In neurosurgery, current image-guidance approaches such as magneticresonance imaging (MRI) combined with neuronavigation offer a map as to where the tumor should be, but the only definitivemethod to characterize the tissue at stake is histopathology. Although extremely valuable information is derived from thisgold standard approach, it is limited to very few samples during surgery and is not practically used for the delineation oftumor margins. The development and implementation of faster, comprehensive, and complementary approaches for tissuecharacterization are required to support surgical decision-making – an incremental and iterative process with tumor removedin multiple and often minute biopsies. The development of atmospheric pressure ionization sources makes it possible toanalyze tissue specimens with little to no sample preparation. Here, we highlight the value of desorption electrosprayionization as one of many available approaches for the analysis of surgical tissue. Twelve surgical samples resected from apatient during surgery were analyzed and diagnosed as glioblastoma tumor or necrotic tissue by standard histopathology,and mass spectrometry results were further correlated to histopathology for critical validation of the approach. The use ofa robust statistical approach reiterated results from the qualitative detection of potential biomarkers of these tissue types.The correlation of the mass spectrometry and histopathology results to MRI brings significant insight into tumor presentationthat could not only serve to guide tumor resection, but that is also worthy of more detailed studies on our understanding oftumor presentation on MRI. Copyright © 2013 John Wiley & Sons, Ltd.
Additional supporting information may be found in the online version of this article at the publisher’s web site.

Keywords: DESI-MSI; brain tumors; real-time diagnosis; surgery; image-guided therapy

* Correspondence to: Nathalie Y.R. Agar, Departments of Neurosurgery andRadiology, Brigham and Women’s Hospital, and Department of CancerBiology, Dana-Farber Cancer Institute, Harvard Medical School, Boston, MA02115. E-mail: [email protected]

† David Calligaris and Isaiah Norton contributed equally to this work.

a Department of Neurosurgery, Brigham and Women’s Hospital, HarvardMedical School, Boston, MA, 02115, USA

b Department of Pathology, Brigham and Women’s Hospital, Harvard MedicalSchool, Boston, MA, 02115, USA

c Department of Chemistry and Center for Analytical Instrumentation Develop-ment, Purdue University, West Lafayette, IN, 47907, USA

d Department of Radiology, Brigham and Women’s Hospital, Harvard MedicalSchool, Boston, MA, 02115, USA

e Department of Cancer Biology, Dana-Farber Cancer Institute, Harvard MedicalSchool, Boston, MA, 02115, USA

‡ Current address: Department of Chemistry, Stanford University, Stanford, CA,94305-5080, USA

Abbreviations: MALDI; DESI; MS; MSI; H&E; MRI; SVM; pLSA; PCA; GBM

Introduction

Surgery is typically the first step for the treatment of braintumors. To minimize the removal of functional healthy tissue,brain mapping techniques are often used prior to and duringsurgery. During the procedure, surgeons can use intraoperativeultrasound[1] and magnetic resonance imaging (MRI)[2–5] incenters where the technology is available, but these tools stillprovide limited temporal resolution (MRI) and discriminativecapability (ultrasound). In addition, neither ultrasound nor MRIdirectly samples the tumor to determine the molecular charac-teristics of the tissue, thereby providing only an indirect assess-ment of the tumor.Over several decades, various methods have been proposed to

provide tissue discrimination including infrared or Raman spec-troscopy,[6–8] flow-cytometry,[9–11] in vivo labeling techniquescoupled with spectroscopy,[12,13] and scintillation counting[14]

for the characterization of tissues in the operating room. Becauseof issues of complexity, limited sensitivity for properly discrimi-nating tissues or limited compatibility with the surgical environ-ment, none of these techniques has yet gained widespread use.A wealth of reports have been published over the past decade on

the ability ofmass spectrometry (MS) to discern and characterize bi-ological tissues with increasing sensitivity and specificity.[15–17] It,therefore, becomes very natural to return mass spectrometers back

J. Mass Spectrom. 2013, 48, 1178–1187

into the operating room where they were routinely used in the1980s to sample airway gases from anesthetized patients.[18] Now,

Copyright © 2013 John Wiley & Sons, Ltd.

Intraoperative mass spectrometry

117

however, they would permit the precise molecular characterizationof tissue and serve, as an analytical tool in image-guided therapy.Different MS platforms will likely find themselves interfacing withsurgical decision-making at various points in the clinical workflow.MS has already proven to be useful for the characterization of intactbiological tissues.[19–21] For over a decade, matrix-assisted laserdesorption/ionization (MALDI) mass spectrometers have success-fully been used for the profiling of peptides and proteins fromtissues and cells in the research setting[19] and has recently beenincreasingly employed for the analysis of small molecules such aslipids, drugs, and their metabolites.[22–30] MALDI mass spectrometryimaging (MALDI-MSI) analyses of tissue have become an extremelypromising tool to support decision-making in histopathologyevaluation of tissue.[20] With its ability to capture essentially acomplete mass range of biomolecules that include acceptedbiomarkers such as proteins, MALDI-MSI should assist in diagnosisproviding enhanced discriminating power over visual inspectionof tissue.[19] A higher level and certainty of diagnosis providedduring frozen section analysis would certainly benefit surgicaldecision-making in better understanding the disease faced by thesurgeon. Typically, one or two samples are sent for frozen sectionanalysis during a surgical case, and MALDI-MSI could find a wayto fit within comparable timelines to standard analysis. For thedelineation of tumor margins, though, multiple minute specimenswould need to be analyzed, and the analysis should result in real-time feedback. Currently, the sample preparation steps requiredfor MALDI-MSI would not be compatible with such a workflow.

With the development of ambient ionization methods such asdesorption electrospray ionization (DESI), it is possible to performMS analysis with essentially no sample preparation, hence, makingsuch methods compatible with the time restrictions required forintraoperative tumor diagnosis and margin delineation.[31,32] InDESI, a pneumatically assisted electrospray produces chargeddroplets that are directed to collide with the surface of a sample.[33]

As the charged droplets collide with the sample surface, theycreate a thin-liquid film into which analytes are extracted; theimpact of subsequent primary droplets releases secondarymicrodroplets in a process termed droplet pick-up.[33] Followingthis pick-up mechanism, the standard electrospray solvent evapo-ration processes occur, followed by the production of dry ions ofanalyte either by the field desorption or charge residue process.

Desorption electrospray ionization is one of multiple atmo-spheric pressure ionization sources. Aimed at ease of implemen-tation and execution, these enabling technologies produceinstantaneous results from solids, aerosols, vapors, and liquids situ-ated externally to the MS, in their native environment.[34] Examplesinclude methods in which the energetic beam is metastable gas-phase atoms and reagent ions (i.e. DART,[35] DAPCI,[36–38] FAPA,[39]

LTP[40,41]), energetic droplets (i.e. DESI,[42,43] EASI,[44] JeDI[45]), andcombinations of laser radiation and electrospray ionization (i.e.ELDI,[46] MALDESI,[47,48] LAESI[49,50]). Ambient methods have manyapplications including imaging biological tissue[51] and thin layerchromatography plates,[52] as well as the direct analysis ofpharmaceutical tablets[53] and inks on banknotes[54] and manyother surfaces. DESI is readily implemented on existing commercialinstruments that have a direct interface with the atmosphere andon small, field portable MS systems.[55,56] Because sampling occursoutside the vacuum system of the instrument, a broad range ofsamples and sample forms can be presented to the massspectrometer.

Another critical feature of DESI is that it allows MSI of sectionsof tissue. MSI enables to record spatially-defined biochemical

J. Mass Spectrom. 2013, 48, 1178–1187 Copyright © 2013 John

information in two-dimension and three-dimension (3D). DESI-MSI analysis is commonly performed by rastering the samplesurface with respect to the stationary continuous flux of spray-charged droplets through an array of predefined coordinates whilecollecting a mass spectrum at each position containing mass-to-charge (m/z) and relative abundance information. The resultingdata are concatenated into an array and selected m/z values areplotted to assess spatial distribution of intensity at specificm/z values. DESI coupled with MSI is particularly valuable in thefield of tissue diagnosis for comparison with standard clinical diag-nosis performed on hematoxylin and eosin (H&E) stained histolog-ical tissue sections.[31,57–59] In contrast to extractive techniques,such as liquid chromatography MS, tissue sections that have beenimaged with DESI-MSI are relatively well preserved and can still bestained after the MS sampling, therefore, allowing MSI data to becorrelated to the exact area of tissue that was analyzed.[31,32]

Desorption electrospray ionization has successfully beenemployed for the study of small molecules[60] including theinvestigation of lipid distributions in a variety of healthy anddiseased animal and human tissues,[51,58,61–66] exemplifying theutility of the method for determining diagnostically-relevantinformation by MS with minimal sample preparation. In compar-ison to existing MS and optical imaging modalities, the ambientionization methods show only modest spatial resolution. Despitethis limitation, these methods have considerable benefits: theyfacilitate measurements outside the vacuum of the instrument,require no contrast agents or chemical-tags, and do not requirefurther sample treatment. Although very high spatial resolutionis desirable for research and development, for example thenanometer range resolution achieved by technologies such assecondary ion MS, the modest spatial resolution and fast analysistime provided by ambient MS technologies is ideal for applica-tions in the clinical setting, especially, during surgery. The minia-turization of mass spectrometers could also eventually facilitateclinical implementation.[67]

General workflow

Surgery remains the most important and usually the first treat-ment modality for devastating brain tumors such as gliomas aswell as other primary and metastatic tumors. Although maximalsurgical excision with the goal of gross total tumor resection isdesirable, in practice, delineation of resection margins is verydifficult because tumors can closely resemble normal tissue andfrequently infiltrate into surrounding normal brain structures. Inaddition, tumors often abut or directly involve critical brainregions – too large a resection margin may increase the risk forpostoperative neurologic deficits. Preoperative localization byMRI of brain tumors is used to plan the surgical interventionand to minimize postoperative deficits. But the shift in theposition of brain structures that occurs following a craniotomycan lead to spatial inaccuracies.[68]

Molecular information obtained rapidly during a surgicalprocedure could provide surgeons with a powerful tool forperforming real-time, image-guided surgery. A variety of mappingtechniques (i.e. Raman imaging,[69] Fourier transform infraredspectroscopy imaging,[70] diffusion tensor imaging,[71] positronemission tomographic/single-photon emission computed tomog-raphy,[72] electrocortical stimulation,[73] and functional MRI[74–77])have been developed to provide surgeons with such understand-ing of the relationship of the tumor to surrounding key corticalareas for neurosurgery. Intraoperative MRI developed at Brigham

Wiley & Sons, Ltd. wileyonlinelibrary.com/journal/jms

9

D. Calligaris et al.

1180

and Women’s Hospital (BWH) has provided unprecedented intrao-perative visualization.[2–5]

Histopathological evaluation of frozen sections from tumorbiopsies is currently the only method available to provide sur-geons with information about tumor type and grade. Althoughcustomarily used, evaluating tumors with frozen sections has anumber of significant limitations that are disruptive to the surgi-cal workflow, in particular, the analysis of each sample requires20min or more, and typically no more than a few samples arepractical to analyze during any one surgical procedure. Moreover,visual review of stained tissue sections does not provide any di-rect molecular information about a tumor. The use of DESI-MScould help with some of these problems, by allowing continuoussampling of multiple areas within the surgical field, by providingspecific information about tumor type, grade, and possiblyprognosis rapidly (within seconds), and by offering very specificmolecular information about a sample including levels of bio-markers or therapeutic compounds. The imaging capabilities ofDESI can be used to develop a well-annotated reference systemcorrelating specific molecular signatures to standard histopathol-ogy information. For real-time applications, though, the rapidprofiling capabilities of DESI can be used.[78] The data acquiredin a seconds-scale profiling fashion can then be mathematicallycompared with a well-defined and well-validated reference sys-tem, providing the surgeon with critical information of the tissueat stake in a real-time.We describe results highlighting the use of MS as a powerful

tool in characterizing tissue for surgical-decision making. Morespecifically, we used DESI-MS to distinguish necrotic tumor tissuefrom viable glioblastoma (GBM) tumor. We first established corre-lation between histopathological staining and DESI-MS to distin-guish viable from nonviable tumor tissue and built a classificationmodel representative of the histological evaluation. We thenused a robust statistical method to validate the detection ofpotential biomarkers. Direct correlation of MS and histopathologyresults offers a level of validation that cannot be bypassed forachieving the goals of introducing this promising analytical toolin the surgical decision-making workflow and of gaining wide-spread acceptance by medical teams. In our approach to imple-ment MS into the operating suite, we push this validationfurther by correlating MS and histopathology results to preoper-ative and intraoperative MRI. In doing so, we not only ensure thevalidity of the information acquired from our MS experiment andits data analysis, but we also enable clinicopathologic correlationsas presented in the succeeding text. The case presented here ad-dresses the discrimination between necrosis and viable tumor,which challenges pre-existing knowledge of the characteristicsof such tissue on MRI. Our work demonstrates that MS could playa significant role in the near-time and real-time diagnosis oftumors, assist in tumor delineation, and complement MRI.

Experimental section

Sample collection

Research subject (surgical case 9) was recruited from surgicalcandidates at the neurosurgery clinic of the BWH and gavewritten informed consent to the Partners Healthcare InstitutionalReview Board protocols. Samples were obtained in cooperationwith the BWH Neurooncology Program Biorepository collectionand analyzed under Institutional Review Board-approvedresearch protocol.

wileyonlinelibrary.com/journal/jms Copyright © 2013 Joh

Image-guided neurosurgery

All surgeries were performed with auxiliary image guidance ofthe BrainLab Cranial 2.1 neuronavigation system (BrainLab).Preoperative MRI sequences included full T2 (1 × 1 × 2mm,100 × 100 slice matrix) and post-contrast T1 (1 × 1× 1mm,256 × 256 slice matrix, 176 slices), processed in the BrainLabiPlanNet 3.0 software. Standard clinical protocols were observedto obtain primary diagnosis from stained frozen sections.

Stereotactic sample acquisition

After clinical frozen-section diagnosis was confirmed, additionalsamples were acquired during the course of clinical resectionand stored at �80 °C. Each sample site was localized by theneurosurgeon using the neuronavigation system pointer, andthe locations were transferred for offline visualization using theOpenIGTLink protocol (client: open-source 3D Slicer software onwww.Slicer.org; server: BrainLab Cranial 2.1 with OpenIGTLinklicense option).[79]

Desorption electrospray ionization mass spectrometryimaging

Tissue sections were prepared on a Microm HM 550 (ThermoScientific, USA) with the microtome chamber chilled at �21 °Cand the specimen holder at �20 °C. 10μm thickness coronalsections were prepared and thaw mounted onto a glass slides.Following thaw mounting of tissue sections, slides were allowedto dry for 10min in a desiccator. DESI-MSI was performed usingan amaZon speed(TM) ion trap mass spectrometer (BrukerDaltonics) equipped with a commercial DESI ion source fromProsolia, Inc. DESI-MSI was performed in a line-by-line fashionwith a lateral spatial resolution of 200μm. MS instrumentalparameters used were 200 °C heated capillary temperature, 5 kVspray voltage, and 4 lmin-1 dry gaz. MS data were acquired fromm/z=50 to 1100 in UltraScan mode (32 500m/z s�1) with a targetmass of m/z = 600 and trap drive level of 100%. Seventeenmicroscans were averaged for each pixel in the images for ascan time of 1 s. The spray solvent was 1:1 acetonitrile:dimethylformamide and the solvent flow rate was 0.7 μl min-1.

Hematoxylin and eosin staining

The following protocol for H&E staining was performed onsections previously analyzed by DESI-MSI: (1) fix in MeOH(2min), (2) rinse in water (10 dips), (3) stain in Harris modifiedhematoxylin solution (1.5min), (4) rinse in water (10 dips), (5) bluein 0.1% ammonia (a quick dip), (6) rinse in water (10 dips), (7)counterstain in Eosin Y (8 s), (8) rinse and dehydrate in 100% EtOH(10 dips), (9) rinse and dehydrate again in 100% EtOH (10 dips),(10) dip in xylene (6 dips), and (11) dip in xylene again (6 dips).Sections were dried at room temperature in hood and coveredwith histological mounting medium (Permount®, Fisher Chemicals,Fair Lawn, NJ) and a glass cover slide.

Statistical analysis

Classification models for glioma subtype, grade, and tumor cellconcentration of gliomas had been previously developed usingsupport vector machine analysis in Bruker ClinProTools 3.0.[80]

New support vector machine classification models were calculatedto classify spectra for each surgical sample (‘GBM’ multiforme vs.

n Wiley & Sons, Ltd. J. Mass Spectrom. 2013, 48, 1178–1187

http://www.Slicer.org;

Table 1. Classification results for samples from surgical case 9

Tissue type (%)

Name Histopathologydiagnostic

GBM Necrosis

D32 GBM/necrosis 12 88

D33 necrosis 2 98

D34 necrosis 1 99

D35 necrosis 0 100


D37 necrosis 4 96

D38 necrosis 0 100

D39 GBM 99 1

D40 GBM 100 0

D41 GBM 96 4



Results indicate the percent of pixels within each image that wereassigned to a given class. Surgical samples used as reference tobuild the support vector machine classifier are in boldface (D38and D40). GBM, glioblastoma.


necrosis). Principal component analysis (PCA) and probabilisticlatent semantic analysis (pLSA) were also carried out usingClinProTools 3.0 software (Bruker Daltonics). PCA is a mathematicaltechnique designed to extract, display, and rank the variancewithin a data set.[81] With PCA, important information that ispresent in the data is retained, whereas the dimensionality of thedata set is reduced. For DESI-MSI, each mass spectrum presents aseries of m/z values with specific intensities. With PCA, wefactorized the set of spectra such that the constituent principalcomponent vectors are ranked in the order of variance. In MSI,the first three principal components (PCs) generally differentiatethe most the samples. PCA also provides loading values(comprised between �1 and 1), originating from the calculationof the PCs, that make it easy to select the contributing peaks ofeach PC for further analysis. pLSA has been introduced in the MSliterature as a technique to divulge latent tissue-type specificmolecular signatures.[82] For each tissue, a distinct distributioncan be considered and mass spectra acquired from this tissue areanalyzed as a specific combination of m/z values. In contrast toPCA, pLSA allows to directly visualize the discriminating peaks fora specific tissue type within a mass spectrum.

Desorption electrospray ionization mass spectrometry imagingdata were converted for import to ClinProTools 3.0 using in-house software. Extracted DESI mass spectra were internallyrecalibrated on common spectra alignment peaks withinClinProTools 3.0. An average mass spectrum created from allsingle spectra was used for peak selection using the ClinProTools3.0 internal method (based on vector quantization). For statisticalanalyses, mass spectra were selected from the tissue from repre-sentative areas (GBM vs. necrosis). Extracted DESI-MS spectraacquired from D43 surgical sample were imported intoClinProTools 3.0 software. Normalization, baseline subtraction,peak peaking, and spectra recalibration were automaticallyperformed using the software. The initial peak integrationwindows were manually verified against the average spec-trum to ensure that no over-calculation or under-calculationwere present.

Visualization of magnetic resonance imaging and massspectrometry data

Magnetic resonance imaging data obtained were plotted in 3DSlicer (www.Slicer.org) (version 4.1). The results of MS datasubjected to the described classification system were overlaidas stereotactic points rendered in color scales representing thedifferent tissue types.

118

Results and discussion

Mass spectrometric evaluation of a glioblastoma resection

Twelve surgical samples (D32 to D43) were taken from a braintumor. After a full pathologic evaluation, a final report was is-sued that diagnosed the tumor as a GBM. This report was issued9 days following the operation. Stereotactic information wasregistered for ten of the biopsies (D32 to D41). Frozen sectionsfrom these surgical samples were analyzed by DESI-MSI andsubsequently stained with H&E. Review of the H&E stained sec-tions by light microscopy revealed some of these surgical sam-ples were entirely composed of viable tumor, whereas otherswere entirely composed of nonviable tumor tissue (i.e. necroticGBM tissue) (Table 1). Because GBM tumors are composed of


rapidly proliferating cells, these tumors will frequently displayregions of necrosis, either focally or in large regions (termedgeographic necrosis).

Hematoxylin and eosin stained tissue sections of surgicalsample D40 showed typical histological features of GBM witha high concentration of viable tumor cells [inset of Fig. 1(a)],whereas sample D38 was entirely composed of necrotic tissue[inset of Fig. 1(b)]. In negative-ion mode, mass spectraacquired from D40 and D38 frozen tissue sections demon-strated distinct profiles (Fig. 1) with certain ions exclusivelyobserved in viable GBM [e.g. m/z= 279.0 and m/z= 391.3 fromD40, Fig. 1(a)] and others in the necrosis region [m/z= 544.5,m/z= 626.6, and m/z = 650.6 for D38, Figure 1(b)]. We alsonoted some ions were present with a higher relative abun-dance in one of the two surgical samples [e.g. m/z= 437.3and m/z=491.3 for D40, Fig. 1(a) and m/z 572.7 for D38, Fig. 1(b)].Corresponding ion images indicate that these ions are presentthroughout the tissue sections of D40 [m/z = 279.0, m/z = 391.3,m/z= 437.3, and m/z = 491.4 ions, Figure S1A) and D38(m/z = 544.5, m/z = 626.6, m/z = 650.6, and m/z = 572.7 ions,Figure S1B).

We have previously shown that tissue specimens can bediscriminated based upon the presence of specific lipidpatterns.[31,57–59] To validate the ability to distinguish viablefrom necrotic GBM by DESI-MS molecular profiling, we nextturned to surgical specimens from this GBM resection thatcontain within the same tissue section both viable and necrotictumor tissue. As shown in Figs. 2 and S2, H&E staining revealeddistinct boundaries between viable GBM and necrotic tumor (N)in both surgical samples D43 [Fig. 2(a)] and D42 (Figure S2A).The DESI-MS data revealed that both of the lipid patterns thatwe had observed in sample D40 and D38 (Fig. 1) were nowpresent in the same sample [Figs. 2(b), 2C, S2B, and S2C;m/z values in red in each Figure] and were located in theappropriate histologic regions – the ion images in the insetsof Figs. 2 and S2 highlight both the areas of viable GBM [ionat m/z= 279.0 Figs. 2(b) and S2B] and the necrotic GBM [ions


1

http://www.Slicer.org

Figure 1. Desorption electrospray ionization mass spectrometry imaging lipid profiles of surgical samples D40 and D38.

Figure 2. Histological evaluation and desorption electrospray ionization mass spectrometry imaging analyses of surgical sample D43.


1182

at m/z = 572.7 and m/z = 544.5 Figs. 2(b) and S2B, respectively].We observed similar results for other ions that we had previ-ously identified as discriminating viable and necrotic tumor


(m/z= 391.3, m/z = 437.3, m/z = 491.3 for GBM, and m/z= 626.6,m/z = 650.6 for necrosis; ion images of Fig. 4 for D43 and S4for D42).



Toward the validation of desorption electrosprayionization mass spectrometry imaging for real-timemolecular diagnostic

We are developing DESI-MSI as a platform for intraoperativediagnostics. In prior studies, we were able to discriminate tumorsof the central nervous system. This was possible not only fortumors that are highly distinct from one another (e.g. gliomafrom meningioma) but also for tumors that are histologically sim-ilar (e.g. discriminating low-grade gliomas such as oligoden-droglioma from low grade astrocytoma).[31,57–59]

Here, we further demonstrate that we can build a classifica-tion method as a proof of concept based on a small trainingset for discriminating viable from nonviable tumor tissue. Thiswas readily achieved by building a classification model basedon machine learning and then determining the rate of cross-val-idation and recognition capability between GBM and necrotictissues in other samples. The cross-validation and recognitioncapability of the classifier was 98% and 100% in the trainingdataset. The results for the test dataset are reported in Table 1.For D43 and D42 surgical samples, each mass spectra

Figure 3. Spectral classification and principal component analysis from daimaging analysis of surgical sample D43.

Figure 4. Probabilistic latent semantic analysis from desorption electrospray io


contributing to classify tissues as GBM or necrosis were mappedon binary images in Figs. 3(a) and S3A.

Principal component analysis [Figs. 3(b) and S3B] and pLSA(Figs. 4 and S4) are two statistical tools that were used in ad-dition to the machine learning approaches to further identifydiscriminating peaks between tissue types. According to thefirst two principal components, PCA results show that massspectra acquired in each region belong to the same tissuetype delimited in Fig. 3(a) [left panel of Fig. 3(b)]. Moreover,the loading model of Fig. 3(b) (right panel) and the statisticaldata of Table 2 clearly indicate that m/z values presented inFig. 2(b) and 2(c) are specific of each tissue type accordingto the Wilcoxon/Kruskal–Wallis test. Finally, pLSA data confirmthe relevance of these m/z values to discriminate the twotissue types [Figs. 4(a) and S4A]. Regarding the statisticalstudy of DESI-MS data of surgical case 9, we can assume thatpotential markers of GBM and necrosis could have beendefined, and further studies should be undertaken to specifi-cally identify the nature of these biomolecules and assignedtargeted peaks as previously described.[31,57–59]

ta acquired from desorption electrospray ionization mass spectrometry

nization mass spectrometry imaging analysis data from surgical sample D43.


1183

Table 2. p-values obtained for the eight peaks from t-tests

Index Mass PTTA PWKW PAD Ave1 Ave2 StdDev1 StdDev2

GBM 162 279.0 <0.000001 <0.000001 <0.000001 1.64 7.19 1.53 3.87

208 391.3 <0.000001 <0.000001 <0.000001 2.2 4.15 1.65 2.22

216 437.3 0.191 0.0141 < 0.000001 3.48 3.82 2.11 1.61

226 491.3 <0.000001 <0.000001 <0.000001 4.33 6.28 2.36 2.63

Necrosis 228 544.5 <0.000001 0 <0.000001 9.55 2.62 4 1.06

232 572.7 <0.000001 0 <0.000001 105.94 23.82 53.11 9.7

251 626.6 <0.000001 0 <0.000001 17.68 6.29 7.36 1.98

259 650.6 <0.000001 0 <0.000001 24.4 7.35 9.95 2.45

The p-values of the Wilcoxon/Kruskal–Wallis (PWKW) test and the Anderson–Darling Test (PAD) indicate a significant difference between theglioblastoma and the necrosis data sets for each m/z value of Figure 2B and 2C (≤0.05 and >0.05, respectively). All the average intensityvalues for the m/z values 279.0, 391.3, 437.3, and 491.3 are also increased in the glioblastoma average mass spectrum (Ave2 values) and theothers (m/z values 544.5, 572.7, 626.6, and 650.6), in the necrosis average mass spectrum (Ave1 values).

Index, sequence of peak; Mass, m/z; PTTA, p-value of t-test (two classes).

GBM, glioblastoma.


1184

Desorption electrospray ionization mass spectrometry imag-ing and magnetic resonance imaging: is the Whole Greaterthan the Sum of its Parts?

Samples from surgical case 9 were classified as GBM or necrotictissue based on mass spectral information, and the results werevalidated by histopathology evaluation of each specimen.Although lipid profiling provides highly specific data to discrimi-nate tissues and define boundaries between tumor and healthybrain tissue, DESI-MSI is still an invasive technique requiringdirect contact with the tissue of interest. Conversely, MRI is a non-invasive technique that may supply mm-scale localization of thetumor, but with limited information on the tumor’s chemistry.As shown in Fig. 5, 3D magnetic resonance (MR) structural scanscan delineate the tumor volume [Fig. 5(a)] and axial gadolinium-enhanced T1-weighted MR images demonstrate the spreading ofthis bilateral GBM across the hemisphere boundary [Fig. 5(b)].The majority of the images in Fig. 5(b) show a hypodense centralcore, commonly associated with necrosis. This core is circled by athick-irregular ring with a shaggy inner margin typical of GBM.GBM has prominent neovascularity with abnormal blood–brainbarrier, and breakdown of this barrier is thought to cause leakageof the contrast agent (i.e. gadolinium) into tissues and to be

Figure 5. Label-free 3D molecular imaging of tumor presentation with deso


responsible for a ring-enhanced signal on enhanced T1-weightedMR images.[83] The highest neovascularity and therefore viabletumor concentration is typically associated with the enhancingtumor ring.

By using stereotactic data about the location of the biopsiesfrom surgical case 9, we mapped information derived from ourclassifiers (GBM or necrotic tissue) onto the MR images (Fig. 5).The 3D MR rendering of the segmented tumor in Fig. 5(a) showsthe relative distribution of surgical samples as they relate to tumorpresentation, whereas individual axial MR images more specificallycorrelate tissue characteristics with the uptake of contrast [Fig. 5(b)]. As shown in Fig. 5(b), DESI-MS data mapping indicates thatthe tumor presents necrotic components both in the central andperipheral portions of the tumor. Some studies have reported thatnecrosis is present in 85% of cases diagnosed as GBM,[84–86] but itis mainly associated with the central region of the tumor. Previousstudies have also reported the propensity of radiation-inducednecrosis that is the result of inflammatory cascades activated byradiation injury and exacerbated by the chronic hypoxia fromendothelial remodeling.[87] In GBM, this radiation-induced necrosisis generally observed in the periphery of the tumor; however, thepatient (case 9) had not received prior radiotherapy.

rption electrospray ionization mass spectrometry.



Conclusion

Surgery is the primary treatment for most brain tumors. Surgicaldecision-making could be improved with tools that rapidly pro-vide molecular information about multiple biopsies or continu-ous sampling at the time of surgery. Ambient MS techniquesthat can provide near-real time molecular information fromtissue samples hold great potential in this area, but they haveto be carefully validated using well-annotated histopathologyevaluation of the tissue. With DESI-MS, we have previously beenable to classify tumors, define tumor subtypes, and identifytumor grade. Here, we show that in surgical resection speci-mens we can readily identify necrotic tumor tissue, an indicatorof a high-grade malignancy, and we can distinguish necrotictumor tissue from viable tumor regions. As we apply DESI-MSto a broad range of human malignancies, we will be able todefine the molecular correlates of a range of histologic features,many of which have become diagnostic hallmarks of cancer(such as necrosis in the diagnosis of GBM). Many of theseinsights will rely on the use of powerful machine learning andstatistical tools to assist in turning the vast data sets acquiredby MS into usable tumor classifiers that are ultimately usefulfor real-time applications. As more and more is carried out,DESI-MS could have a significant role for a broad range of diag-nostic applications including defining the boundaries betweentumor and normal tissue, diagnosing image-guided needlebiopsies, and determining prognostic and predictive informa-tion for guiding patient care. One significant disadvantage ofMS over optical approaches in characterizing tissue is that mol-ecules need to leave the tissue for MS analysis, therefore,disrupting it. Because surgery innately exposes and disrupts tis-sue, MS-based approaches for real-time tissue characterizationdo not pose more risk to the patient. Some of the significant ad-vantages of MS toward surgical decision-making applicationsinclude the following: (1) the ability to analyze any molecule,at least in principle, (2) acquire complex signatures that canincrease specificity over a single biomarker paradigm, (3) nomolecular labeling is required, and (4) rapidity of execution, es-pecially when interfaced with ambient ionization methods. Oursiting of a mass spectrometer into the AMIGO at BWH provideswith invaluable opportunities to validate MS findings for a vari-ety of surgical diseases tackled by the growing field of MSI andto continue technology development with the hope of improv-ing patient care.

118

Acknowledgements

The authors are grateful to their patients and families whoconsent to participate in research. The authors would also liketo acknowledge support from the Advanced MultimodalityImage Guided Operating (AMIGO) suite team.

The work received support from Daniel E. Ponton fund for theNeurosciences, the DFCI Pediatric Low-Grade Astrocytomaprogram, and the NIH Director’s New Innovator Award (grant1DP2OD007383-01 to N.Y.R. Agar); the Klarman Family Founda-tion (A.J. Golby). S.S. is supported by NIH grant K08NS064168.LSE, and RGC also thank the U.S. National Institute of Health(Grant 1R21EB009459). This project was supported by theNational Center for Research Resources and the National Instituteof Biomedical Imaging and Bioengineering of the NationalInstitutes of Health through Grant Numbers P41EB015898 andP41RR019703.


References

[1] H. Tang, H. Sun, L. Xie, Q. Tang, Y. Gong, Y. Mao, Q. Xie, M. Zheng,D. Wang, H. Zhu, J. Zhu, X. Feng, Z. Yao, X. Chen, L. Zhou.Intraoperative ultrasound assistance in resection of intracranialmeningiomas. Chin. J. Cancer. Res. 2013, 25, 339–345.

[2] J. Kettenbach, D. F. Kacher, S. K. Koskinen, S. G. Silverman, A. Nabavi,D. Gering, C. M. Tempany, R. B. Schwartz, R. Kikinis, P. M. Black, F. A.Jolesz. Interventional and intraoperative magnetic resonance imag-ing. Annu. Rev. Biomed. Eng. 2000, 2, 661–690.

[3] P. M. Black, E. Alexander, 3rd, C. Martin, T. Moriarty, A. Nabavi, T. Z.Wong, R. B. Schwartz, F. Jolesz. Craniotomy for tumor treatment inan intraoperative magnetic resonance imaging unit. Neurosurgery1999, 45, 423–431; discussion 431–423.

[4] P. M. Black, T. Moriarty, E. Alexander, 3rd, P. Stieg, E. J. Woodard, P. L.Gleason, C. H. Martin, R. Kikinis, R. B. Schwartz, F. A. Jolesz. Develop-ment and implementation of intraoperative magnetic resonance im-aging and its neurosurgical applications. Neurosurgery 1997, 41,831–842; discussion 842–835.

[5] F. A. Jolesz, A. Nabavi, R. Kikinis. Integration of interventional MRI withcomputer-assisted surgery. J. Magn. Reson. Imaging 2001, 13, 69–77.

[6] A. Becker, C. Hessenius, K. Licha, B. Ebert, U. Sukowski, W. Semmler,B. Wiedenmann, C. Grotzinger. Receptor-targeted optical imagingof tumors with near-infrared fluorescent ligands. Nat. Biotechnol.2001, 19, 327–331.

[7] A. S. Haka, K. E. Shafer-Peltier, M. Fitzmaurice, J. Crowe, R. R. Dasari,M. S. Feld. Diagnosing breast cancer by using Raman spectroscopy.Proc. Natl. Acad. Sci. U. S. A. 2005, 102, 12371–12376.

[8] N. Kosaka, M. Ogawa, P. L. Choyke, H. Kobayashi. Clinical implicationsof near-infrared fluorescence imaging in cancer. Future Oncol. 2009,5, 1501–1511.

[9] M. Ito, Y. Minamiya, H. Kawai, S. Saito, H. Saito, K. Imai, J. OgawaIntraoperative detection of lymph node micrometastasis with flowcytometry in non-small cell lung cancer. J. Thorac. Cardiovasc. Surg.2005, 130, 753–758.

[10] A. H. Mesiwala, L. D. Scampavia, P. S. Rabinovitch, J. Ruzicka, R. C.Rostomily. On-line flow cytometry for real-time surgical guidance.Neurosurgery 2004, 55, 551–560; discussion 560–551.

[11] T. Shioyama, Y. Muragaki, T. Maruyama, T. Komori, H. Iseki.Intraoperative flow cytometry analysis of glioma tissue for rapiddetermination of tumor presence and its histopathological grade.J. Neurosurg. 2013, 118, 1232–1238.

[12] H. Kishimoto, M. Zhao, K. Hayashi, Y. Urata, N. Tanaka, T. Fujiwara,S. Penman, R. M. Hoffman. In vivo internal tumor illumination bytelomerase-dependent adenoviral GFP for precise surgical naviga-tion. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 14514–14517.

[13] B. J. Tromberg, N. Shah, R. Lanning, A. Cerussi, J. Espinoza, T. Pham,L. Svaasand, J. Butler. Non-invasive in vivo characterization of breast tu-mors using photon migration spectroscopy. Neoplasia 2000, 2, 26–40.

[14] P. Garcia-Talavera, J. R. Garcia-Talavera, C. Gonzalez, E. Martin,M. Martin, A. Gomez. Efficacy of in-vivo counting in parathyroidradioguided surgery and usefulness of its association with scintigra-phy and intraoperative PTHi. Nucl. Med. Commun. 2011, 32, 847–852.

[15] C. Schone, H. Hofler, A. Walch. MALDI imaging mass spectrometry incancer research: combining proteomic profiling and histologicalevaluation. Clin. Biochem. 2013, 46(6), 539–545.

[16] P. M. Angel, R. M. Caprioli. Matrix-assisted laser desorption ionizationimaging mass spectrometry: in situ molecular mapping. Biochemistry2013, 52, (22), 3818–3828.

[17] J. Pol, M. Strohalm, V. Havlicek, M. Volny. Molecular mass spectrom-etry imaging in biomedical and life science research. Histochem. cellbiol. 2010, 134, (5), 423–443.

[18] G. M. Ozanne, W. G. Young, W. J. Mazzei, J. W. Severinghaus.Multipatient anesthetic mass spectrometry: rapid analysis of datastored in long catheters. Anesthesiology 1981, 55, 62–70.

[19] M. Stoeckli, P. Chaurand, D. E. Hallahan, R. M. Caprioli. Imaging massspectrometry a new technology for the analysis of protein expres-sion in mammalian tissues. Nat. Med. 2001, 7, 493–496.

[20] A. Thomas, N. H. Patterson, M. M. Marcinkiewicz, A. Lazaris,P. Metrakos, P. Chaurand. Histology-driven data mining of lipid sig-natures from multiple imaging mass spectrometry analyses: appli-cation to human colorectal cancer liver metastasis biopsies. Anal.Chem. 2013, 85, 2860–2866.

[21] P. Chaurand, J. L. Norris, D. S. Cornett, J. A. Mobley, R. M. Caprioli.New developments in profiling and imaging of proteins from tissue


5


1186

sections by MALDI mass spectrometry. J. Proteome Res. 2006, 5,2889–2900.

[22] J. Bunch, M. R. Clench, D. S. Richards. Determination of pharmaceu-tical compounds in skin by imaging matrix-assisted laser desorp-tion/ionisation mass spectrometry. Rapid Commun. Mass Spectrom.2004, 18, 3051–3060.

[23] D. M. Drexler, T. J. Garrett, J. L. Cantone, R. W. Diters, J. G. Mitroka,M. C. Prieto Conaway, S. P. Adams, R. A. Yost, M. Sanders. Utility ofimaging mass spectrometry (IMS) by matrix-assisted laserdesorption ionization (MALDI) on an ion trap mass spectrometer inthe analysis of drugs and metabolites in biological tissues.J. Pharmacol. Toxicol. Methods 2007, 55, 279–288.

[24] Y. Hsieh, J. Chen, W. A. Korfmacher. Mapping pharmaceuticals intissues using MALDI imaging mass spectrometry. J. Pharmacol.Toxicol. Methods 2007, 55, 193–200.

[25] Y. Hsieh, R. Casale, E. Fukuda, J. Chen, I. Knemeyer, J.Wingate, R.Morrison,W. Korfmacher. Matrix-assisted laser desorption/ionization imagingmassspectrometry for direct measurement of clozapine in rat brain tissue.Rapid Commun. Mass Spectrom. 2006, 20, 965–972.

[26] S. Khatib-Shahidi, M. Andersson, J. L. Herman, T. A. Gillespie, R. M.Caprioli. Direct molecular analysis of whole-body animal tissuesections by imaging MALDI mass spectrometry. Anal. Chem.2006, 78, 6448–6456.

[27] M. L. Reyzer, Y. Hsieh, K. Ng, W. A. Korfmacher, R. M. Caprioli. Directanalysis of drug candidates in tissue by matrix-assisted laser desorption/ionization mass spectrometry. J. Mass Spectrom. 2003, 38, 1081–1092.

[28] H. Y. Wang, S. N. Jackson, J. McEuen, A. S. Woods. Localization andanalyses of small drug molecules in rat brain tissue sections. Anal.Chem. 2005, 77, 6682–6686.

[29] F. J. Troendle, C. D. Reddick, R. A. Yost. Detection of Pharmaceuticalcompounds in tissue by matrix-assisted laser desorption/ionization(MALDI) and laser desorption/chemical ionization (LD/CI) MS/MSwith a quadrupole Ion trap. J. Am. Soc. Mass Spectrom. 1999, 10,1315–1321.

[30] D. S. Cornett, S. L. Frappier, R. M. Caprioli. MALDI-FTICR imaging massspectrometry of drugs and metabolites in tissue. Anal. Chem. 2008,80, 5648–5653.

[31] L. S. Eberlin, I. Norton, D. Orringer, I. F. Dunn, X. Liu, J. L. Ide, A. K.Jarmusch, K. L. Ligon, F. A. Jolesz, A. J. Golby, S. Santagata, N. Y. Agar,R. G. Cooks. Ambient mass spectrometry for the intraoperative mo-lecular diagnosis of human brain tumors. Proc. Natl. Acad. Sci.U. S. A. 2013, 110, 1611–1616.

[32] R. G. Cooks, Z. Ouyang, Z. Takats, J. M. Wiseman. Detection technol-ogies. Ambient mass spectrometry. Science 2006, 311, 1566–1570.

[33] A. Venter, P. E. Sojka, R. G. Cooks. Droplet dynamics and ionizationmechanisms in desorption electrospray ionization mass spectrome-try. Anal. Chem. 2006, 78, 8549–8555.

[34] C. Wu, A. L. Dill, L. S. Eberlin, R. G. Cooks, D. R. Ifa. Mass spectrometryimaging under ambient conditions. Mass Spectrom. Rev. 2013, 32,218–243.

[35] R. B. Cody, J. A. Laramee, H. D. Durst. Versatile new ion source for theanalysis of materials in open air under ambient conditions. Anal.Chem. 2005, 77, 2297–2302.

[36] H. Chen, J. Zheng, X. Zhang, M. Luo, Z. Wang, X. Qiao. Surface de-sorption atmospheric pressure chemical ionization mass spectrome-try for direct ambient sample analysis without toxic chemicalcontamination. J. Mass Spectrom. 2007, 42, 1045–1056.

[37] J. P. Williams, V. J. Patel, R. Holland, J. H. Scrivens. The use of recentlydescribed ionisation techniques for the rapid analysis of somecommon drugs and samples of biological origin. Rapid Commun.Mass Spectrom. 2006, 20, 1447–1456.

[38] Y. Song, R. G. Cooks. Atmospheric pressure ion/molecule reactionsfor the selective detection of nitroaromatic explosives using acetoni-trile and air as reagents. Rapid Commun. Mass Spectrom. 2006, 20,3130–3138.

[39] J. T. Shelley, G. C. Chan, G. M. Hieftje. Understanding the flowingatmospheric-pressure afterglow (FAPA) ambient ionizationsource through optical means. J. Am. Soc. Mass Spectrom. 2012,23, 407–417.

[40] Y. Liu, Z. Lin, S. Zhang, C. Yang, X. Zhang. Rapid screening of activeingredients in drugs by mass spectrometry with low-temperatureplasma probe. Anal. Bioanal. Chem. 2009, 395, 591–599.

[41] J. D. Harper, N. A. Charipar, C. C. Mulligan, X. Zhang, R. G. Cooks,Z. Ouyang. Low-temperature plasma probe for ambient desorptionionization. Anal. Chem. 2008, 80, 9097–9104.


[42] J. M. Wiseman, D. R. Ifa, A. Venter, R. G. Cooks. Ambient molecularimaging by desorption electrospray ionization mass spectrometry.Nat. Protoc. 2008, 3, 517–524.

[43] Z. Takats, J. M. Wiseman, B. Gologan, R. G. Cooks. Mass spectrometrysampling under ambient conditions with desorption electrosprayionization. Science 2004, 306, 471–473.

[44] P. M. Lalli, G. B. Sanvido, J. S. Garcia, R. Haddad, R. G. Cosso, D. R.Maia, J. J. Zacca, A. O. Maldaner, M. N. Eberlin. Fingerprinting andaging of ink by easy ambient sonic-spray ionization massspectrometry. Analyst 2010, 135, 745–750.

[45] Z. Takats. Method and device for desorption ionization by liquid jet.PCT Int. Appl. 2007, 29 pp, WO 200713837.

[46] J. Shiea, M. Z. Huang, H. J. Hsu, C. Y. Lee, C. H. Yuan, I. Beech,J. Sunner. Electrospray-assisted laser desorption/ionization massspectrometry for direct ambient analysis of solids. RapidCommun. Mass Spectrom. 2005, 19, 3701–3704.

[47] J. S. Sampson, K. K. Murray, D. C. Muddiman. Intact and top-downcharacterization of biomolecules and direct analysis using infraredmatrix-assisted laser desorption electrospray ionization coupled toFT-ICR mass spectrometry. J. Am. Soc. Mass Spectrom. 2009, 20,667–673.

[48] J. S. Sampson, A. M. Hawkridge, D. C. Muddiman. Direct characteriza-tion of intact polypeptides by matrix-assisted laser desorptionelectrospray ionization quadrupole Fourier transform ion cyclotronresonance mass spectrometry. Rapid Commun. Mass Spectrom.2007, 21, 1150–1154.

[49] P. Nemes, A. A. Barton, Y. Li, A. Vertes. Ambient molecular imagingand depth profiling of live tissue by infrared laser ablationelectrospray ionization mass spectrometry. Anal. Chem. 2008, 80,4575–4582.

[50] P. Nemes, A. Vertes. Laser ablation electrospray ionization foratmospheric pressure, in vivo, and imaging mass spectrometry. Anal.Chem. 2007, 79, 8098–8106.

[51] J. M. Wiseman, D. R. Ifa, Q. Song, R. G. Cooks. Tissue imaging at atmo-spheric pressure using desorption electrospray ionization (DESI)mass spectrometry. Angew. Chem. Int. Ed. Engl. 2006, 45, 7188–7192.

[52] G. Paglia, D. R. Ifa, C. Wu, G. Corso, R. G. Cooks. Desorptionelectrospray ionization mass spectrometry analysis of lipids aftertwo-dimensional high-performance thin-layer chromatography par-tial separation. Anal. Chem. 2010, 82, 1744–1750.

[53] H. Chen, N. N. Talaty, Z. Takats, R. G. Cooks. Desorption electrosprayionization mass spectrometry for high-throughput analysis ofpharmaceutical samples in the ambient environment. Anal. Chem.2005, 77, 6915–6927.

[54] L. S. Eberlin, R. Haddad, R. C. Sarabia Neto, R. G. Cosso, D. R. Maia,A. O. Maldaner, J. J. Zacca, G. B. Sanvido, W. Romao, B. G. Vaz, D. R.Ifa, A. Dill, R. G. Cooks, M. N. Eberlin. Instantaneous chemical profilesof banknotes by ambient mass spectrometry. Analyst 2010, 135,2533–2539.

[55] N. L. Sanders, S. Kothari, G. Huang, G. Salazar, R. G. Cooks. Detectionof explosives as negative ions directly from surfaces using a minia-ture mass spectrometer. Anal. Chem. 2010, 82, 5313–5316.

[56] G. Huang, W. Xu, M. A. Visbal-Onufrak, Z. Ouyang, R. G. Cooks. Directanalysis of melamine in complex matrices using a handheld massspectrometer. Analyst 2010, 135, 705–711.

[57] L. S. Eberlin, I. Norton, A. L. Dill, A. J. Golby, K. L. Ligon, S. Santagata,R. G. Cooks, N. Y. Agar. Classifying human brain tumors by lipidimaging with mass spectrometry. Cancer Res. 2012, 72, 645–654.

[58] L. S. Eberlin, A. L. Dill, A. J. Golby, K. L. Ligon, J. M. Wiseman, R. G.Cooks, N. Y. Agar. Discrimination of human astrocytoma sub-types by lipid analysis using desorption electrospray ionizationimaging mass spectrometry. Angew. Chem. Int. Ed. Engl. 2010,49, 5953–5956.

[59] N. Y. Agar, A. J. Golby, K. L. Ligon, I. Norton, V. Mohan, J. M. Wiseman,A. Tannenbaum, F. A. Jolesz. Development of stereotactic mass spec-trometry for brain tumor surgery. Neurosurgery 2011, 68, 280–289;discussion 290.

[60] J. M. Wiseman, D. R. Ifa, Y. Zhu, C. B. Kissinger, N. E. Manicke, P. T. Kissinger,R. G. Cooks. Desorption electrospray ionization mass spectrometry:Imaging drugs and metabolites in tissues. Proc. Natl. Acad. Sci. U. S. A.2008, 105, 18120–18125.

[61] J. M. Wiseman, S. M. Puolitaival, Z. Takats, R. G. Cooks, R. M. Caprioli.Mass spectrometric profiling of intact biological tissue by usingdesorption electrospray ionization. Angew. Chem. Int. Ed. Engl.2005, 44, 7094–7097.



[62] A. L. Dill, L. S. Eberlin, D. R. Ifa, R. G. Cooks. Perspectives in imag-ing using mass spectrometry. Chem. Commun. (Camb.) 2010, 47,2741–2746.

[63] A. L. Dill, L. S. Eberlin, C. Zheng, A. B. Costa, D. R. Ifa, L. Cheng, T. A.Masterson, M. O. Koch, O. Vitek, R. G. Cooks. Multivariate statisticaldifferentiation of renal cell carcinomas based on lipidomic analysisby ambient ionization imaging mass spectrometry. Anal. Bioanal.Chem. 2010, 398, 2969–2978.

[64] A. L. Dill, D. R. Ifa, N. E. Manicke, A. B. Costa, J. A. Ramos-Vara, D. W.Knapp, R. G. Cooks. Lipid profiles of canine invasive transitional cellcarcinoma of the urinary bladder and adjacent normal tissue by de-sorption electrospray ionization imaging mass spectrometry. Anal.Chem. 2009, 81, 8758–8764.

[65] L. S. Eberlin, D. R. Ifa, C.Wu, R. G. Cooks. Three-dimensional vizualizationof mouse brain by lipid analysis using ambient ionization mass spec-trometry. Angew. Chem. Int. Ed. Engl. 2010, 49, 873–876.

[66] L. S. Eberlin, A. L. Dill, A. B. Costa, D. R. Ifa, L. Cheng, T. Masterson,M. Koch, T. L. Ratliff, R. G. Cooks. Cholesterol sulfate imaging inhuman prostate cancer tissue by desorption electrospray ionizationmass spectrometry. Anal. Chem. 2010, 82, 3430–3434.

[67] M. Yang, T. Y. Kim, H. C. Hwang, S. K. Yi, D. H. Kim. Development of apalm portable mass spectrometer. J. Am. Soc. Mass Spectrom. 2008,19, 1442–1448.

[68] M. H. Reinges, H. H. Nguyen, T. Krings, B. O. Hutter, V. Rohde, J. M.Gilsbach. Course of brain shift during microsurgical resection ofsupratentorial cerebral lesions: limits of conventional neuronavigation.Acta. Neurochir. (Wien) 2004, 146, 369–377; discussion 377.

[69] M. F. Kircher, A. de la Zerda, J. V. Jokerst, C. L. Zavaleta, P. J. Kempen,E. Mittra, K. Pitter, R. Huang, C. Campos, F. Habte, R. Sinclair, C. W.Brennan, I. K. Mellinghoff, E. C. Holland, S. S. Gambhir. A brain tumormolecular imaging strategy using a new triple-modality MRI-photoacoustic-Raman nanoparticle. Nat. Med. 2012, 18, 829–834.

[70] N. Bergner, B. F. Romeike, R. Reichart, R. Kalff, C. Krafft, J. Popp.Tumor margin identification and prediction of the primary tumorfrom brain metastases using FTIR imaging and support vectormachines. Analyst 2013, 138, 3983–3990.

[71] J. W. Henson, P. Gaviani, R. G. Gonzalez. MRI in treatment of adultgliomas. Lancet Oncol. 2005, 6, 167–175.

[72] C. A. Buchpiguel, J. B. Alavi, A. Alavi, L. C. Kenyon. PET versus SPECTin distinguishing radiation necrosis from tumor recurrence in thebrain. J. Nucl. Med. 1995, 36, 159–164.

[73] F. E. Roux, K. Boulanouar, J. P. Ranjeva, M. Tremoulet, P. Henry,C. Manelfe, J. Sabatier, I. Berry. Usefulness of motor functional MRIcorrelated to cortical mapping in Rolandic low-grade astrocytomas.Acta. Neurochir. (Wien) 1999, 141, 71–79.

[74] W. A. Hall, H. Liu, C. L. Truwit. Functional magnetic resonanceimaging-guided resection of low-grade gliomas. Surg. Neurol.2005, 64, 20–27; discussion 27.

[75] T. Krings, R. Topper, K. Willmes, M. H. Reinges, J. M. Gilsbach,A. Thron. Activation in primary and secondary motor areas inpatients with CNS neoplasms and weakness. Neurology 2002, 58,381–390.


[76] I. D. Wilkinson, C. A. Romanowski, D. A. Jellinek, J. Morris, P. D.Griffiths. Motor functional MRI for pre-operative and intraoperativeneurosurgical guidance. Br. J. Radiol. 2003, 76, 98–103.

[77] W. C. Liu, S. C. Feldman, M. Schulder, A. J. Kalnin, A. I. Holodny,A. Zimmerman, R. Sinensky, S. Rao. The effect of tumour type anddistance on activation in the motor cortex. Neuroradiology 2005, 47,813–819.

[78] P. Chaurand, S. A. Schwartz, M. L. Reyzer, R. M. Caprioli. Imaging massspectrometry: principles and potentials. Toxicol. Pathol. 2005, 33, (1),92–101.

[79] J. Tokuda, G. S. Fischer, X. Papademetris, Z. Yaniv, L. Ibanez, P. Cheng,H. Liu, J. Blevins, J. Arata, A. J. Golby, T. Kapur, S. Pieper, E. C.Burdette, G. Fichtinger, C. M. Tempany, N. Hata. OpenIGTLink: anopen network protocol for image-guided therapy environment. Int.J. Med. Robot. 2009, 5, 423–434.

[80] T. Alexandrov, J. Decker, B. Mertens, A. M. Deelder, R. A. Tollenaar,P. Maass, H. Thiele. Biomarker discovery in MALDI-TOF serum proteinprofiles using discrete wavelet transformation. Bioinformatics 2009,25, 643–649.

[81] H. Han. Nonnegative principal component analysis for mass spectralserum profiles and biomarker discovery. BMC Bioinformatics 2010,11(Suppl 1), S1.

[82] M. Hanselmann, M. Kirchner, B. Y. Renard, E. R. Amstalden, K. Glunde,R. M. Heeren, F. A. Hamprecht. Concise representation of mass spec-trometry images by probabilistic latent semantic analysis. Anal.Chem. 2008, 80, 9649–9658.

[83] J. G. Smirniotopoulos, F. M. Murphy, E. J. Rushing, J. H. Rees, J. W.Schroeder. Patterns of contrast enhancement in the brain andmeninges. Radiographics 2007, 27, 525–551.

[84] S. F. Smith, J. M. Simpson, J. A. Brewer, L. H. Sekhon, M. T. Biggs, R. J.Cook, N. S. Little. The presence of necrosis and/ormicrovascular prolifer-ation does not influence survival of patients with anaplastic oligoden-droglial tumours: review of 98 patients. J. Neurooncol 2006, 80, 75–82.

[85] C. R. Miller, C. P. Dunham, B. W. Scheithauer, A. Perry. Significance ofnecrosis in grading of oligodendroglial neoplasms a clinicopatho-logic and genetic study of newly diagnosed high-grade gliomas.J. Clin. Oncol. 2006, 24, 5419–5426.

[86] T. Homma, T. Fukushima, S. Vaccarella, Y. Yonekawa, P. L. Di Patre,S. Franceschi, H. Ohgaki. Correlation among pathology, genotype,and patient outcomes in glioblastoma. J. Neuropathol. Exp. Neurol.2006, 65, 846–854.

[87] D. S. Lee, M. Yu, H. S. Jang, Y. S. Kim, B. O. Choi, Y. N. Kang, Y. S. Lee,D. C. Kim, Y. K. Hong, S. S. Jeun, S. C. Yoon. Radiation-induced braininjury: retrospective analysis of twelve pathologically proven cases.Radiat. Oncol. J. 2011, 29, 147–155.

Supporting information

Additional supporting information may be found in the onlineversion of this article at the publisher’s web site.


1187

Documents

Mass spectrometry imaging as a tool for surgical …dunnlab.bwh.harvard.edu/wp-content/uploads/2015/04/...Mass spectrometry imaging as a tool for surgical decision-making David Calligaris,a†