Recent Patents on Medical Imaging, 2011, 1, 000-000

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor by Multimodal Imaging: Quantitative Measurement by Fusion of MRI, PET Imaging with MALDI and Histology
Rakesh Sharma*,1,2 and Jose K. Katz1
1 2

Departments of Radiology and Medicine, Columbia University, New York, NY 10032, USA Amity Institute of Nanotechnology, Amity University UP, Noida, UP, 201303, India

Received: 02 June 2010; Revised: 09 June 2010; Accepted: 27 June 2011

Abstract: Integration of imaging data with immunohistology is a new art. Increased PET and MRI image intensities on rat breast tumor MRI-PET images were reviewed for possible correlation with tumor histology and MALDI imaging tumor characteristics in the light of recent inventions and patents. Increased signal intensities of intracellular (IC) sodium MRI and flouro-2-deoxy-glucose utilization by PET from apoptosis protein rich MALDI visible regions of tumors were positively correlated to chemosensitivity of Taxotere. MCF-7 cancer cell line induced rat tumor MRI-PET images and histology digital images were compared for correlation in pre- and post taxotere treated tumors. For MALDI imaging, iterated protein ion mass spectrometry peak analysis was done using data from laser raster over tumor slices in sequence and 3D tumor volume was simulated for specific peak(s) distribution. A criterion was developed to evaluate malignancy by histology and MRI-PET imaging. For correlation, regression analysis was done using MRI-PET imaging, histology and MALDI imaging data from MCF-7 tumor after 24 hours post-taxotere treatment. Apoptosis indices were calculated by histostaining using pentachrome, feulgen and ss-DNA antibody assay. Review showed sodium MRI and PET signal intensity distribution comparable and measurable in tumor tissue regions. In tumors, taxotere induced an increase in IC-Na MRI signal with decreased tumor size and micro-PET showed FDG uptake increase with decreased tumor size than that of control tumors after 24 hours. Histology features indicated tumor risk (high 'IC/EC ratio', high mitotic index and apoptotic index), decreased tumor viability (reduced mitotic figures, reduced diploidy or aneuploidy and proliferation index) after Taxotere treatment. These features in co-registered IC-Na, PET hypermetabolic and monoclonal antibody (ss-DNA) sensitive regions showed 6% difference. MALDI imaging showed tumor specific protein ion species and their distribution showed empirical correlation (limited visual match) with MRI-PET signal intensities but comparable match with histology features. Recent patents strongly suggest the possibility of sodium MRI and PET multimodal imaging integrated with MALDI-imaging as an non-invasive chemosensitivity assay to monitor the anticancer effect.

Keywords: MRI and PET integration with MALDI, apoptosis index, prostate tumor, validation, MALDI imaging, texotere chemosensitivity. INTRODUCTION Magnetic resonance imaging (MRI) combined with positron emission tomography (PET) microimaging multimodal technique was invented and recently emerged as multimodal molecular imaging tool in experimental tumor pharmaco-dynamic characterization. Matrix Assisted Laser Desorption/Ionization (MALDI) based imaging was invented and proposed as diagnostic technique and it is emerging now as imaging technique to visualize the cancer specific protein(s) for time-dependent monitoring of anticancer chemosensitivity by sodium MRI as research tool in cancer therapy [1]. However, MRI/PET visible tumor image characteristics and association with MALDI-imaging of tumor specific protein still remain a puzzle to correlate them with tumor physiology and histology characteristics due to difficulty of interpretating physical complexity of MRI-PET,
*Address correspondence to this author at the Center of Biotechnology, Innovations And Solutions Inc., 3945 west Pensacola Street, #98, Tallahassee, Florida 32304, USA; Tel: 18507027661; Fax: 18503395361; Email: rs2010@columbia.edu, jk32@columbia.edu

signal physico-chemical complexity of MALDI signal and cytomorphic complexity of histopathology structural details of tumor [2]. In last 5 years, extensive efforts are made to develop time-of- flight MALDI (TOF-MALDI) as real-time fast imaging technique of accurate analysis of proteins and peptides with detail information of minute protein species by mass spectroscopy (up to nanomoles based on m/z ratio) by combining it with other variant mass spectroscopy SELDI, MALDI-LC, MALDI-TLC methods and modifying sample positioning, matrix composition and laser desorption/ absorption as represented in recent patents [3-13]. These techniques were of limited use to analyze protein composition as m/z peak intensities with or no information of protein distribution. In attempt to develop tissue MALDI imaging as ion distribution maps of selected m/z peaks of high intensities, we report Monte Carlos simulation technique to convert the MALDI peaks as points and display them as digitized simulated m/z protein ion peak maps at matched tissue locations on histology tissue sections coregistered with MRI-PET signal intensity distribution maps of tissue sodium-glucose uptake or oxygen contents. Currently, efforts are focused on integration or fusion of MRI-PET data from in vivo images and MALDI-histology 2011 Bentham Science Publishers Ltd.

2210-6847/11 $100.00+.00

2 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

immunostaining data from ex vivo tissue slides to reconstruct the three-dimensional tissue volume with details of biophysicochemical, structural and molecular makeup of tissue. The present review focuses on physical principles and quantitative approach to explore the possibility of integration and fusion of in vivo and ex vivo tissue data. Intracellular sodium MRI signal intensity increase in glioma, necrosis, and apoptosis was considered due to leaching out intracellular sodium after the membrane damage [1, 14]. Measurement of intracellular sodium by triple quantum, inversion recovery MRI imaging methods emerged as technique risk-free frominvasive infusion of Schiff reagents [15]. The 18F-Fluoro-deoxy glucose positron emission tomography (18FDG-PET) dynamic scanning indicated hyper-glycolytic regions in tumor while MRI generated static images [16]. In following sections, we review our study of MCF-7 injected rat breast tumor illustrating the static deformed PET registration with slice-by-slice MRI sections to demonstrate the point-wise match between MRI-PET signal intensities and protein MALDI peaks or images. A quantitative MRIPET-MALDI criterion was developed to validate and correlate the MRI/PET microimaging for tumor intracellular sodium signal intensities and PET active hyper-glycolytic regions, with MALDI protein peaks and histology tumor features. The assumptions of tumor cells were: 1. Loss of membrane sodium pump/symporter is associated with glucose pump and loss of oxygen (low oxidative phosphorylation makes high glycolysis); 2. In tumorigenesis, low oxidative phosphorylation high glycolysis apoptosis necrosis cell proliferation cell death occurs in a sequence; 3. The events of tumorigenesis or drug antitumor action are detectable by in vivo oxidative phosphorylation and intracellular sodium (by MRI), in vivo high glycolysis and oxygen (by PET), apoptosis proteins (by MALDI protein peaks), ex vivo cytomorphometric changes of apoptosis, proliferation, necrosis, cysts (by histology); 3. Multimodal hybrid molecular imaging provides a finger print of tumorigenic kinetics and antitumor pharmacokinetics or therapeutic monitoring (Quantitative Theradiagnostics). Relationship between sodium MRI and PET signal: The sodium pump Na+/K+ ATPase exchanges intracellular sodium across cell membrane simultaneously with glucose transport. MRI/PET signal intensities represent the less known status of sodium and glucose transport [17]. The sodium MRI T1 signal is sensitive to intracellular sodium release while PET is sensitive to glucose uptake. The echo delay time (TE) applied during multiple flip-angle gradientecho multi-slice imaging is a function of T1 or T2 signal intensity and concentration of sodium while biotransformation of 18FDG standard uptake value (SUV) in PET is function of glycolysis. Relationship between breast tumor protein (proteomics by MALDI) and MRI-PET imaging: Protons act as coin with two faces (one sensitive to MALDI and other sensitive to MRI-PET). MALDI imaging is protein specific to metabolic disorder of rat tumor cells in various breast malignant tumor stages with following presumptions as recently described [18]:

m/z of specific proteins act as molecular MALDI signatures of m proton mass and z ionic charge on proteins (bound with intracellular sodium) in tumors Sodium Magnetic Resonance Relaxation relationship with proteins (electrochemical protein-ionic forms or protons) serves as molecular signatures of sodium pump and sodium symporter protein in tumor cells (IC/zC ratio for unit mass proton) Intracellular sodium ions (bound with proteins) are visible at specific NMR frequency and single charged sodium bound protein ions or protons from protein generate distinct m/z peaks by MALDI Fusion of intracellular sodium (bound with proteins) MRI map with MALDI protein ion map displays empirical protein information Comparison of MRI signal vs MALDI signal gives matched characteristic with high: accuracy, repeatability, sensitivity At tumor specific physiological conditions of pH and temperature, specific protein molecule net charge and molecular mass ratio values predict specific protein(s) by MALDI. Standard peptide (proteins)samples of CHACA and HABA show specific m/z peak patterns. The breast tumor has specific hsp27/60, 14-3-3-annexin A2, phophoglycerate kinase-1, calreticulin, protein disulfide isomerase and phosphoproteins Pea 15 and Fabp5 [2,18]. By Monte-Carlos simulations, the specific m/z peak(s) may be displayed as 3D reconstruct (image volume) or peak(s) spatial distribution in different regions of tumors [2, 19].

MALDI imaging principle: Protein detection probability from MALDI images displays several distinct protein m/z peak intensities after baseline denoising, peak intensity thresholding correction via apodization and regression or calculated normalized peak intensity with accuracy [20]. Reference acids (hydrocinnamic acid and sinapinic M + Analyte-H +) show high gas acid) or (MH+ +Analyte phase or proton affinity (basicity) and they are desorbed and accelerated to a fix kinetic energy in a time-of-flight analyzer (a vaccum field free tube). In tube, protein ions are separated as a function of their m/z depending on velocity and time of flight as following: m/z = 2Vt2/L2(MALDI) ICMR-Na(MRI) SUV(PET) Eq 1 where m is the ion mass, z is number of charges, V is source potential, t is time of flight and L is length of TOF analyzer. For unit mass, (m=1)/z is proportional with intracellular sodium (IC) MR signal intensity and PET intensity (SUV or oxygen content from glycolysis). Single proton m/z, MRI, and PET intensities are known in standard calibration [20]. Due to possibility of multi-ionic charged proteins, laser pulse receives all charged protein ions with no delay and sends them to detector same time to detect all proteins in one step to generate high resolution m/z spectra or 2-dimensional ion density m/z maps of protein molecules from tumor tissue

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

sample. Due to no exact internal standard, each peptide concentration is measured indirectly from peak amplitudes at different positions as representative of peptide size. The relative peptide concentration is calculated from MALDI peak as: I=1/ [I predicted intensity pI(s) ] Eq 2 Suppose s is peptide sequence and I is intensity of MS peak. MASCOT software for peptide mass finger printing was used to search protein analysis in terms of: 1. protein mass in range in Daltons (Da); 2. isoelectric pI of specific proteins [2]; 3. protein hit score and matched other common peak pattern of spectra.; 4. Narrowing down number of common sharp spectral peaks per protein and find matched spectral peak pattern common among proteins; 5. preprocessing for deconvolution, baseline correction and noise peak removal; 6. identification of m/z peak intensities; 7. narrow down search of protein specific m/z peaks [2]. MALDI in protein ion 3D display: After imaging, coregistered tumor sections at same histology match are placed on MALDI plate and laser beam rasterizes over tumor sections in MS spectrometer at different locations to generate an array of MS peak digitized points appearing as distribution maps of proteins as outlined in Fig. (1). MATERIALS AND METHODS Tumor propagation and Microimaging: Techniques were reviewed from patents (see Table 1) for MCF-7 induced tumor propagation inlocally raised rats and treated with Taxotere (Aventis Pharmaceuticals, Bridgewater, NJ) similar to earlier report [20]. The sodium MRI (1 mm slice) and 18FDG-PET images of MCF-7 induced rats (n = 6) and control rats (n = 3) were scanned with Animal Care and Use Committee guidelines in force at Columbia University as described in previous report [1, 21]. In brief, rats were placed inside 26 mm birdcage coil and imaged in 4.23 T clinical imager to generate T1, T2 parametric images. Sodium IR MRI T1 maps were obtained using multiple flip-angle gradientecho multi-slice images (128 128 10, FOV = 25 25 1.8 mm). For T1 map, TR = 200 msec, TE = 4.5 msec, = {15, 30, 45, 60, 75, 90}) were chosen. For T2 map, TR = 4 sec, TE = {12, 24, 36, 48} msec) were chosen. The T1 maps were generated by non-linear fitting the gradient-echo multi-slice images (T1 weighted images) to the following function: T1w( ) = C0(1-e-TR/T1) sin( )/(1-cos( ) e-TR/T1) + C1, Eq 3 where, T1w( ) is the per-voxel measured signal at each flip angle, TR is the repetition time for the MR experiment, T1 is the fitted parameter, and C0 and C1 are constants used to account for baseline signal and measurement offset, respectively. T1 values were measured at each voxel on voxel-by-voxel T1 map. The T2 maps were obtained by non-linear fitting the gradientecho multi-slice images T1 weighted images to the following function: T2w(TE) = C0*e-TE/T2 + C1 Eq 4

where, T2w(TE) is the per-voxel measured signal at each echo-time, TE is the echo-time for the MR experiment, T2 is the fitted parameter and C0 and C1 are used to account for baseline signal and measurement offset, respectively. After magnetic resonance imaging, each animal was immediately killed, perfused with saline and frozen in an ice block to minimize any postmortem protein degradation. The frozen tumor tissue was sectioned to 20 micrometers thick on a cryo-microtome (Leica) in the cranial-caudal direction. Tumor sections from each segment in the blockface volume were prepared for MALDI-IMS acquisition. MALDI IMS images were taken and collected every 150 micrometers in an anterior to posterior fashion. The tumor tissue ion images were collected at lateral resolution of 150-300 micrometers, with each pixel represented by accumulation of 300 laser shots. MALDI spectrum of mass-charge ratio data was processed for spectral smoothing and base-line correction in MATLAB and ProTSData (Biodesix) software (see Fig. 1). MicroPET facility in Milstein Hospital building was used to demonstrate hyper- or hypometabolic tumor regions in Taxotere treated Sprague-Dowley rats after injection of 100 Curie 18FDG by intravenous route. Technique Acquisition Development for MALDI-IMS Data

Mass spectrometer was tuned and controlled in its operations for TOF-MALDI MS spectroscopy mode and it was also used as the data source to acquire, process, store, and print. Most of the analog electrical signals reach the computer after analog-to-digital converter is used. In reverse order, digital signal can be converted to analog signal. However, in MALDI, transputer is used as digital device to convert its electrical signals in the form of pulses or proportional m/z peak intensities. A mass spectrum has m/z values (peaks) each showing peak height proportional to number of protein ions with unit charge. The m/z peak shapes from selected tumor tissue locations on MALDI slide (protein molecules) generate a set of electrical signals at preset voltage (checking is important in different Scanning modes). By manipulation of mass spectra data, accurate mass measurement was done (relevant peak sorting by thresholding at a certain peak height) to gather important m/z peaks and compare with calibrated reference peaks of reference CHACA and HABA calibrated compounds. By match and try a set of peptides was selected to determine the protein make-up (proteomics finger print) using MOSCOT and Swiss library search for molecule identification as shown in Fig. (1) at the bottom. Tumor tissue samples were prepared for MALDI data using techniques described in a previous study [1]. Collected tissue sections were transferred using rice paper to gold-coated MALDI target plates (Applied Biosystems Inc.) and spray-coated with a 25 mg/mL sinapinic acid matrix solution prepared in 60% acetonitrile, 0.5% trifluoroacetic acid. Approximately 10 mL of matrix solution were needed to produce a homogeneous matrix crystal layer. Matrix coated samples were then analyzed on a linear MALDI-time-of-flight mass spectrometer (Autoflex II, Bruker Daltonics Inc.) equipped with a Smartbeam laser operating at 100 Hz. Each peak was analog signal varied

4 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

Fig. (1). Outline of MALDI-MRI-PET-immunostaining integration of digital tumor images is shown to generate a MALDI map of proteins to highlight the m/z peak selection and peak digitization to reconstruct proteomics map and 3D tumor volume. After MRI-PET imaging, tumor is excised and processed for histology and MALDI to generate a rasterized information of tumor cytometry (shown as 1-5 and a to g regions) with corresponding (spectra 1,2.. n) of protein distribution in breast tumor tissue. Note the spatial information of protein ions in 3D cube on right obtained by thresholding and baseline correction. The 3D tumor digital information is fused with MRI-PET images. See Fig. (4) for better visualization.

from base line. First signal was digitized and simulated by Monte-Carlos approximated 3D reconstruction volume. An analog to digital converter (ADC) converted a continuous signal to series of digital pulses in which the voltage represented snapshots of the biochemical nature of molecules as analog signal taken at regular time intervals (discrete digital readings in Fig. 1). The voltage from ion detector varies as ions were introduced and focused. As each m/z value was focused, a peak ion current generated the voltage change proportional to each m/z amplitude or peak intensity on spectra. 3D Spatially Resolved MALDI-IMS Post-Processing Image post-processing was performed to recreate spatially resolved 3D MALDI IMS data. The first step was to align the MALDI mass spectra to the targeting image used by the mass spectrometer. The MS spectrometer reports the

origin of MS spectra within the tumor field-of-view (FOV) in terms of internal motor coordinates. These tumor locations are used as training fiducials related to a targeting image(s) selected at the time of data acquisition. Pixel locations of the training fiducials in the targeting image were registered to their corresponding motor coordinates using an affine alignment. Tumor locations in internal motor coordinates within the mass spectrometer were transformed into images pixel space (spatial locations) by alignment via contour based registration of shape features in each image of the targeting images as reconstructed block face image (outline of the rat tumor section). The contours were manually highlighted by iterative closest point (ICP) algorithm to align the corresponding contours in targeting images. This transformation placed the targeting image coordinates into block face image coordinates. The slice location in zdirection or axial direction was annotated for each slice number and the thickness of each slice to append it with each

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

2D block face coordinate to provide a 3D physical space location for each MALDI IMS spectra or a continuous transformation from the 2D motor coordinates to 3D animal specific coordinates. Breast Tumor 3D Reconstruction Volume Spatially resolved three dimensional volume reconstruction MALDI IMS images were reconstructed by post-processing steps. The ice-block was manually set across a stationary blade to section the rat tumor tissue at the interval of 1 mm distance and 5 micron thickness. Retrospective inter-slice registration was used to align consecutive blockface images. For each set of blockface images, the second slice was aligned to the first slice. The registered MRI image slice was used as the target blockface to align the next consecutive histology slice in the tumor volume. This process was iterated upon until all of the MRI images had been registered, and concatenating these together yielded an accurately reconstructed blockface volume. Tumor Registration Three dimensional MALDI-IMS and magnetic resonance images were correlated by geometrical deformation and rigid co-registration (tumor high signal intensities on images as solid object) of blockface volume (PET frames) to the magnetic resonance volume (3D slices) of tumor. The heterogeneous tissue nature of breast tumor prevents centroid registration large scale deformable changes in anatomy and deconvolution method cannot realign them. A six-degree of freedom rigid body (position and orientation) normalized mutual based registration was used to associate and align implicitly the blockface PET data (moving frames in x angular direction) with magnetic resonance volumes (ascending image slices in z direction) shown in Fig. (3). Registration algorithms: IDL (Research Systems, Inc., Boulder, CO) on Pentium IV platform was used to use Automated Image Registration (AIR) algorithm and mutual information (MI) algorithm for following subsampling at coordinates 884, 442, 221, or 111 (XYZ subsampling) after prealignment by manual reorientation of tumor volumes selected in the capture range as reported elsewhere [22]. Rigid geometric transformation was used using six SUV and MRI imaging characteristics without smoothing by AIR and MI algorithms for convergence optimization method [23]. However, singular value decomposition (SVD) algorithm was used to estimate the registration errors to give perfect registration transformation matrix [24, 25]. This transformation was applied to a set of points spaced in and around the mouse tumor. The SVD-transformed set of points was then retransformed with the inverse of the transformation matrix arising from the multimodality registration to visualize tumor pixels. The mean Euclidean distance between these final points and the maximum distance between these points measured the registration accuracy and the functional performance of both algorithms [26]. Histology features and tumor evaluation criteria: A stereotactic MRI-PET match criteria was applied at different 16 locations of each tumor on histology slide to indicate different tumor features [27].

Registration and similarity match: Tumor was assumed as rigid body. Based on rigid body registration principle, first tumor anatomic features were segmented from a MRI data set. Intensity-based, pattern-matching algorithm was used for inter-slice registration based upon Normalized Mutual Information [28]. NMI was optimized for x-y translation and in-plane rotation to register sequential images in the reconstruction process. The specific implementation of the algorithm used in this paper was provided by the Insight Toolkit (http: //www.itk.org). It is important to note that registration via NMI requires no fiduciary systems [29]. Validation method: For validation of PET, MRI image intensities, biotransformation was optimized and registration methods were compared to the ground truth histology sections and registration accuracy was evaluated by root mean square (RMSE) as described earlier [21]. Imaging and histology data of tumor area with different cytomorphic features were compared by image processing using Optimas 6.5 and statistical correlation was calculated using PRISM 3.03 softwares. RESULTS Reviewed data showed that subcutaneous tumors measured 0.1 to 0.5 cm in diameter as shown in Fig. (2). The MRI and PET image acquisition systems generate images in different format so making intrinsic MRI, PET image registration difficult with blockface of tumor on microtome for histology. The blockface volume reconstruction was completely automatic. Normal rat images did not show hyperintense regions. The tumor regions on MRI and PET images were prealigned by manual reorientation using tumor shape as pseudo marker as shown in Figs. (2, 3). The breast tumor areas in five rats were selected to image the whole tumor mass in each. The volume trimming removed the extra image points not common to both PET and MRI data sets from tumor. The rigid geometric transformation by AIR and MI algorithms generated convergence optimizationof common data points of tumor visible on both MRI and PET images as shown in Figs. (1, 2). The singular value decomposition (SVD) algorithm estimated the registration errors up to 10.5% and generated registration transformation matrix by applying transformation to a set of points spaced in and around the breast tumor. The transformation matrix visualized the tumor pixels up to 1.5 mm. The mean Euclidean distance between these final points in fused coregistered images and the maximum distance between these points measured up to +10% accuracy as functional performance of SVD algorithm. Taxotere effect on 18FDG PET hyperintensities (SUV values): Hyperintensities on MCF-7 induced rat tumor in vivo dynamic PET images are shown in Fig. (2). The locations of tumor were distinct and measurable with resolution of 0.5 mm, as shown in Fig. (2). After segmentation, control breast tumor showed distinct tumor boundaries. Different colors indicated the metabolic activity distribution at different tumor locations. In 24 hours post Texotere treated rats the tumor size was reduced in comparison with untreated tumor bearing control rats at same location as shown in Fig. (3) as extracted tumor areas. The distribution of metabolic activity at different tumor

6 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

Fig. (2). A rat breast tumor after 21 weeks of propagation is shown in first row on top with excised histology slide (in middle) for registration with mitotic figures, MCF-7 cell MALDI, tumor cytomorphometry, tunnel maps, DNA cycle M/S bars (first row on right A-D panels) and MRI-PET images at different slice levels (second row on left). Notice the histology of tumor provides the ultimate cytomorphometric details (different tumor i-m stages with tissue features) while same tissue locations show specific TOF-MALDI peaks and Monte-Carlos simulated 3D tumor reconstructed volume as display of 3D protein map in registration with MRI-PET piled up slice volume. The detailed protein distribution (on bottom panel at left) with MS MALDI peaks (on bottom panel in the middle) provide peptide informatics or tumorigenic protein 3D makeup (shown with tumor locations 1-5 with 4 at back side) in the tumor volume.

location a d was significant to correlate these malignant regions and MRI signal intensities as shown in Table 1. The distribution of both sodium MRI signal intensities and PET hyperintensities SUV values in different tumor locations was significant to correlate MRI-PET visible regions with premalignancy characteristics by histology shown in Table 2. Taxotere effect on extracellular and intracellular sodium image intensities: High MRI signal intensities are shown on MCF-7 induced rat tumor in vivo extracellular (EC) sodium and intracellular (IR) sodium images in Fig. (3). The sodium MRI signal intensities of tumors were comparable with sodium phantom images at optimized inversion time (TI = 25 ms) and echo time (TE = 10 ms). The IR-MRI images showed better contrast between breast tumor and normal rat tissue compared to the EC-MRI images on T1 weighted contrast scan settings as shown in Fig. (2).

The sodium MRI images of in vivo MCF-7 tumors are shown in Figs. (2, 3), at baseline, 24 hours and 48 hours after intravenous injection of 1 mg/ml in 0.25 ml of Taxotere in the femoral vein. Both EC-MRI and IR-MRI image showed contrast resolution similarity. However, contrast enhancement is shown as the IR-MRI line profiles following Taxotere chemotherapy in Fig. (3). The visible intracellular sodium[Na]i increase was distinct by IR-MRI imaging but less visible by EC-MRI imaging after Taxotere chemotherapy. Using a standard high temperature sensitive superconducting copper coil enhanced the signal-to-noise ratio of IR-MRI images by a factor of 10.5 times in less imaging time by a factor of 10 times. Tumor histology evaluation criteria and MRI-PET imaging comparison: The histology distinguished the tumor necrosis, apoptosis and different malignant tumor features

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

under high power field microscopy as shown in Fig. (2) in pre-treatment tumors. Different tumor features were measured of cyst size up to 40-200 microns, membrane blabbing, beaded nuclei 2-5 per cell and extracellular space up to 100-300 microns observed under high power field on slide and each division showed distinct MRI/PET signal intensities as shown in Table 1. The cyst showed low MRI signal intensities and extracellular space showed high MRI intensities. The PET showed high signal intensities in both cyst and extracellular space features. However, necrosis, cell proliferation regions showed isointense MRI images as shown in Fig. (2) coregistred with panels (i-m for different stages) on histology. The tumor evaluation quantitative molecular imaging criterion suggested a simple, reproducible tumor grading scheme with minimum intraoperator bias to evaluate the tumor features and changes after anticancer taxotere drug effect on tumor chemosensitivity. The delineated areas on both histology digital images and MRI-PET fused images showed a comparable tumor morphometric measurement with r2 = 0.997, P value = 0.002 as shown in Figs. (2, 3). However, histology method showed 20% less areas due to shrinkage of tumor tissues as shown in Figs. (2, 3). Registration error and accuracy measurement: Rigid registration method showed two different points annotated as bright tumor regions and dark tumor regions as PET images. The brighter point set included all brighter nodes of the tumor from Moriginal. The dark point set included all dark-gray voxels of tumor from tumor from Moriginal. The RMSE accuracy was measured in millimeters for rigid registration method [1]. The tumor regions from Moriginalwere registered with static (Ptr and Pem) PET transmission and emission volumes. Monitoring of therapeutic response by tumor histology and signal intensity changes in IC Na MRIand PET images: The effect of Taxotere after 12 hours was not visible by MRI and PET techniques. Taxotere enhanced the localized tumor signal intensities on intracellular sodium on MRI by 25% and glucose uptake on PET by 10% after 24 hours. However, later the effect was reduced after 48 hours as shown in Fig. (3). The post treated 48 hours end-point histology suggested the tumor cytomorphology features of enhanced apoptosis, cyst size and mitotic index comparable with enhanced IR sodium and PET signal intensities at these tumor locations. Apoptosis was evident by Annexin V immunostaining and cell cycle CAS 200 histograms showed M and S phases of neoplasia in selected histology matched tumor regions shown in Fig. (2). However, these less defined tumor tissue features were not feasible for histology and imaging co-registration. MALDI imaging with MRI-PET and histology: The excised tumor specimens and their matched locations in MRI-PET showed a coregistered data set of MALDI peak profile and histology cytomorphic features on small tissue sections are shown with i-m panels and MALDI peak spectra in Fig. (2) in high power microscopy fields. Same data is shown for registered MRI-PET signal intensities with MALDI array and registered histology details of each cytomorphic feature shown in Fig. (3) and Table 2. For achieving tumorigenic protein details from MALDI images, individual peaks A and B were characterized to sort out from

electrophoresis out of several spots of proteins as shown in Fig. (2). Between the individual m/z values (between the peaks), no voltage change was observed because no ions arrive at the detector. The time between individual peaks was very much longer than the time taken by one peak width. The base line correction eliminated the noisy part of base line to record desired signal. Such processing was useful in identification of specific pattern of ion peak output voltage values for specific molecule in the sample. A total of 3 peaks (m/z with 6,630, 8,139 and 8,942 Da) were screened out by support vector machine to construct the classification model with high discriminatory power in the training set as shown in Figs. (1, 2). The sensitivity and specificity of the model were 96.45 and 94.87%, respectively, in the blind-testing set. The candidate biomarker with m/z of 6,630 Da was found to be down-regulated in breast cancer tumors, and was identified as apolipoprotein C-I (unpublished data). Another two candidate biomarkers (8,139, 8,942 Da) were found upregulated in breast cancer and identified as C-terminaltruncated form of C3a and complement component C3a, respectively (unpublished data). In addition, the level of apolipoprotein C-I progressively decreased with the clinical stages I, II, III and IV, and the expression of C-terminaltruncated form of C3a and complement component C3a gradually increased in tumor locations different from nontumor locations (unpublished data). Other set of three MALDI peaks (m/z with 11250 Da (A), 13750 Da (B), and small peaks (C) at 13700 Da and 15200 Da positions) showed match with MRI-PET fusion signal intensities as shown in Figs. (3, 4). MALDI Image Processing: Post-processed peaks and digitized ion images (m/z peak specific) for tumor volume reconstructions represented a set of peptide(s) and protein(s) indicated by respective spectral m/zpeaks as shown in Fig. (2). Monte-Carlos digitization performed well to generate point spread or point on MALDI-IMS image. Several cycles ofiterations (N = 10) further enhanced the sensitivity of protein or peptide specific peak selection and digitized peak point as simulated MALDI image as shown in Fig. (3). MALDI IMS imaging: Integrating three-dimensional volume reconstructs of spatially resolved MALDI IMS ion images of whole breast tumor with fused high resolution MRI-PET images showed correlation between proteomic profiles with in vivo distribution of sodium and glucose uptake shown in Figs. (1-3). Each laser spot on MALDI plate corresponds to a pixel in a two-dimensional array of protein molecule or peptide make-up profile (proteomics content) of the selected point (on tumor slice on MALDI plate) predicting m/z peaks at every pixel. Three dimensional tissue volume reconstruct (display of simulated m/z data in 3 dimensions) by MALDI IMS construct provided the information of the MALDI IMS data or simulated 3D spatial finger print of proteomics relationship (protein pattern) with reconstructed 3D tumorigenic events in tumor 3D tissue volume. The MALDI-IMS information enhanced the protein-specific mass spectra as 3D anatomical distribution of protein annotation or distribution map for proteins (m/z) as visual ion volumes. Spatially resolved MALDI IMS data and volume rendered ion volumes fused with in vivo MRIPET data suggested the possibility of breast tumor proteins

8 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

Fig. (3). (top panel) MRI-PET-MALDI data integration method is sketched. (bottom panel) A rat breast tumor (PET image on left) after taxotere treatment is shown. First row (in middle at top) with MRI-PET image fusion at different MRI slice levels (second column in right). Notice the high color coded signal intensities of tumor provides the taxotere effect while same tissue locations show specific TOF-MALDI peaks as finger print of taxotere effect (second raw in middle) and Monte-Carlos simulated 3D tumor reconstructed volume as display of 3D protein map in registration with MRI-PET piled up slice volume of shrunk tumor size (at bottom panel in middle). The detailed protein distribution (on bottom panel at right) with MS-MALDI peaks provide peptide informatics or tumorigenic protein 3D makeup (peaks A and B shown in tumor locations A-E) in the tumor volume.

as associated with structural and functional events in the tumor cells (sodium symporter-glucose uptake status by MRI-PET) during tumorigenesis as drug evaluation testing method (see Figs. 2, 3). The tumor areas with presence of tumor-specific proteins (apoptosis related high protein concentration or high m/z ion volume on MALDI image) showed correlation with T1 contrast IC sodium variations in magnetic resonance and blockface volumes in the overlay renderings. In 24 hours post-texotere treated animals, IC sodium variation was

maximum without change in MALDI image proteomics profile. In selected voxels inside breast tumor and treated breast tumors showed protein peak (m/z) ion intensity values + 3 sd above mean value. Unpaired f-testand t-test showed unequal variances of tumors (n = 16) vs normal tissue (n =3). Magnetic resonance images of rat breast tumor on a 4.23 Tesla clinical imager produced structural quantitative sodium parametric MRI images while PET generated physiological distribution of glucose uptake. Parametric comparison on MRI-PET images showed pixel intensities in both RGB colors and gray scale as biomarkers.

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Fig. (4). (on top row) A tumor histology section with high power microscopy area (see arrows for apoptosis) is shown with corresponding MALDI optimized peaks A and B. The peaks A and B were digitized by Monte Carlo simulation (A and B shown with arrows) and integrated with MRI-PET images to generate a 3 dimensional tissue reconstruct. The tumor proteomics-image volume was used to compare chemosensitivity.

The results indicated the correlation of ex vivo postmortem 3D proteomics and 2D histology correlation with in vivo 3D MRI-PET structural-physiological imaging. The ex vivo proteomics features were aligned and fused with corresponding ex vivo histology regions visible in the in vivo MRI-PET data at the resolution of 1.5 mm. The texotere

treated animal tumor cells showed distinct cytomorphic features on histology, distinct MALDI peaks, distinct MRIPET signal intensities, distinct 3D MRI-PET-MALDI simulated construct display, different from normal tumor respective tumor features.

10 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

Table 1.

Quantitative evaluation of MCF-7 rat breast tumor shows comparison of MRI-PET imaging with cytomorphic indices and MALDI peaks to interpret the power of signal intensities before and after 24 hours taxotere treatment in mouse tumor to represent chemosensitivity
Sodium MRI Intensity SUV (kBq/ml)Peaks MALDI (in HPF) A,B A,B A A A,B,C Histology (A.I.) 4.450.2 4.370.3 65-75 60-70 4525 635021 35 25 48 35 104 104 B or C B or C A or B A or B C C 6120 7311 4412 3012 12525 8222 KI-67 (in HPF) 160 120 P.I. (Bars) 260 125 3.70.2 3.20.4 M-DNA M-DNA S-DNA M or S-DNA M/S-DNA

Pre-Drug treated@ and Postdrug Treated# Tumor Features

Tumor area (mm2; n=16)@ Tumor area (mm ; n=3) IC/EC space
@ 2 #

4.400.3 4.350.2 60-70 84-95 gray bright dark dark bright bright gray gray
(#)

84

IC/EC space# Necrosis* (squares)@ Necrosis* (squares) Viable** (squares)@ Viable** (squares)# Apoptosis*** (nuclei)@ Apoptosis*** (nuclei)# Cyst**** (size in m)@ Cyst**** (size in m)#
(@)

Pre-drug treated tumor and post-treated tumor by sodium MRI image intensity and histology. By usingeyepiece-micrometer square counter, necrosis*(<25% cells in micrometer square), viable cells** (<60% cells inmicrometer square) and apoptosis*** (20-40 apoptotic nuclei in HPF) and cyst space****(< 100 m) per HPF werepremalignant histology characteristics. IC/EC space (% space in HPF), necrosis, viable cells are shown as number ofmicrometer squares with <25% necrosis area in HPF by histology. Apoptotic index (A.I.) and proliferation index (P.I.) are shown as average number of apoptotic nuclei per HPF and number of mitotic figures per HPF. S-DNAhistogram area was measured by CAS 200 system in arbitrary units. Single strand-DNA mAb area was measured indigital images by Optimas 6.5 and ss-DNA mAb density was measured in arbitrary units of photomultiplier scanner. The major three peaks were visible at 11250(A), 13750(B), and small peaks (C) at 13700 and 15200 positions (in m/z) on spectra as characteristics of rat breast tumor proteins shown in Figs. (3, 4).

Statistical analysis of MALDI data showed spectra (in the range of 1-20 kDa) baseline subtracted and averaged peaks by centroid function gave peak resolution (FWHM) 1000-1500 (S/R = 8 for m/z 950 peak signals) on linear TOF mode setup as shown in Figs. (3, 4). Three MALDI peaks showed frequency graph of Gaussian distribution showed averaged distribution 0.83(95% confidence intervals 0.780.88) and peak intensities differed CV 20% with score between 0.7-1.0 by linear regression. It means lowest and highest relative intensities differ each other for < 30%. Cartesian plot analysis for influence of m/z over similarity score of each peak (within 0.7-1.0 range) showed reproducible data in tumors from rats as shown in Fig. (5). DISCUSSION The integrated MRI-PET-MALDI data fusion for evaluation of rat breast tumor is explored to identify apoptotic protein(s). Explanted MCF-7 cell lines are routine in experimental rat breast tumor propagation and testing anticancer drug chemosensitivity [27]. In vivo Taxotere treatment experimental doses used in rats were comparable to human clinical doses in previous study. Early tumor features and malignant lesions in rats induced with MCF-7 cell line showed biological, morphological similarities with many characteristics that closely mimic human breast cancer [1]. The proteomics information of MCF-7 cells is well documented by MALDI MS spectroscopy and PEG electrophoresis to identify potential protein biomarkers to predict response to chemotherapy in breast cancer [30-32]. Present study also indicated the possibility of protein

identification in tumor by MS peaks and presence of different proteins by PEG electrophoresis. The selection of breast tumor areas with high signal intensity pseudo markers on MRI and PET images provided fine tumor alignment. However, better spatial resolution was still a challenge. The convergence optimization of both MRI and PET common data points was reasonable to fuse them and approach was comparable with previous reports [21,23]. The SVD functional performance was comparable to calculate registration error with acceptable accuracy [33,34]. The distinct sodium MRI and glucose uptake PET signatures in tumor solid sites and semi-solid or fluid filled cyst regions in breast tumors were comparable with previous reports of different pre-malignant or malignant stages in tumor [1,14,20]. The tumor chemosensitivity to Taxotere was associated with solid tumor shrinkage or tumor cyst development showing up with low IR-MRI signal and enhanced EC-MRI signal on T1 weighted images. This observation corroborated with other reports indicating more loss of bound sodium in breast tumor after 48 hours than at 24 hours [1, 20]. The MRI/PET similarity measure and validation by root mean square error (RMSE) was feasible for PET and MRI tumor size [35]. However, the PET method used an assumption that deformation to the intermediate tumor graydark point was proportional to the full deformation to the dark tumor regions in images. Poor MRI/PET similarity measurement accuracy artifacts were minimized by simulating MRI tumor size from segmented tumor size or

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

11

Table 2.

A tumor histology evaluation criteria of comparing histology features with MRI/PET image signal intensities is shown. The histology cytomorphic features of cystic fluid, apoptotic cells per high power field, extracellular volume, necrosis, viable cells are shown in mice prostate tumors. The degree of malignancy is shown with different extents as + for mild, ++ for moderate, +++ for intense, ++++ for severe cell damage. Several malignancy features were characteristic.
MALDI m/z Peaks* Sodium MRI Signal Intensity IR Na MRI ++ + PET SUV Histology Tumor Features Tumor Characteristics Tumor Stage Pre-Malignant Stage: Intraductal proliferation, ductal hyperplacia, apoptosis Malignant stage: Carcinoma (papillary; invasive comedo and cribriform) sarcoma neoplasia

B.C

+ +

Viable cells Active necrosis

A,B,C

++++

+++

apoptosis

dark/bright/gray

apoptosis cyst, extracellular space

*The major three peaks were visible at 11250(A), 13750(B), and small peaks at 13700 and 15200(C)positions (in m/z) on spectra as characteristics of rat breast tumor proteins in tissue as shown in Fig. (3).

Fig. (5). The statistical analysis is for reproducibility of m/z data (frequency vs score) on left panel and linear regression analysis on right panel.

computing NMI between static PET transmission volume (Ptr) and small MRI volumes chosen along the line of maximum 0.6 0.9 voxels away from zero translation in order to avoid segmentation errors as reported elsewhere [28]. The rigid registration of whole tumor borders in all directions also improves measurement accuracy [29]. Moreover, the deformations were restricted to anatomically tumor areas by locating the sphere centers on the triangle nodes. The quantitative criterion served as numerical approach to measure tumor areas or size showing different tumor cytomorphic features, MALDI peaks and MRI/PET signal intensities at matched locations up to 100 micron resolution in pre-treatment or post-Taxotere treatment. Main features were: cyst hypointensity on MRI; necrosis with cell proliferation hyperintensity on PET, semisolid mass hypointensity on MRI and PET both modalities; Time of Flight-MALDI (TOF-MALDI) or Electrospray Ionization

(ESI) Mass Spectrometry of phosphopeptides from trypsin protein digests to have a large number of peaks; distinct cytomorphic features of apoptosis, necrosis, proliferation, cyst mass, aneucleosis by excised tumor histology. Earlier studies showed MRI-PET with histology as powerful tools for characterization and identification of phosphorylation sites [35]. Status of MALDI imaging as adjunct still remains disputed because of several artifacts including low intrinsic abundance, inefficient ionization, and/or signal suppression of most common peptides may limit or even prevent detection, unless the apoptosis sensitive phosphopeptide(s) content is significantly enriched by electrophoresis prior to MALDI analysis. Similar to the present study showing peaks (m/z with 11250(A), 13750(B) Da major peaks and m/z with 13700 and 15200(C) Da small peaks shown in Fig. 3), three major peaks (m/z with 6,630, 8,139 and 8,942 Da) were earlier sorted out by support vector machine to construct the

12 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

Sharma and Katz

confirmation and classification model with high discriminatory power in the training breast tumor data set [35]. Investigators reported primary invasive cancer proteins (C6, C11, C14, C16, and C26) different from five normal ones (N4, N15, N32, N33, and N36). Mass spectra on IMAC-Cu chip arrays using 1 g of total protein showed protein expression profiles confirmed the breast cancer by two clusters: C6, C14, N32, N33, and N36 and the other C11, C16, C26, N4, and N15. A supervised cluster analysis by ProPeak (3Z Informatics, Charleston, SC) and biomarkers both separated the cancer data from noncancer data [35]. The sensitivity and specificity of the model were reported 96.45% and 94.87%, respectively, in the blind-testing set. The candidate biomarker with m/z of 6,630 Da was found as down-regulated in breast cancer patients, and was identified as apolipoprotein C-I. Other two candidate biomarkers (8,139, 8,942 Da) were found up-regulated in breast cancer and identified as C-terminal-truncated forms of C3a and complement component C3a, respectively. In addition, the level of apolipoprotein C-I progressively was decreased with the clinical stages I, II, III and IV, and the expression of Cterminal-truncated form of C3a and complement component C3a gradually increased in higher clinical stages [18, 35]. 3D MALDI protein sorting and display by Monte Carlos distribution as digital image of specific protein peak (appeared as digital map similar with MRI-PET digital images) was similar to previous study on 3D volume reconstruction of proteins and peptides of breast cancer [19, 36]. Sinha et al. developed an iteration method to display phosphopeptide Pea15 and Fabp5 proteins of glioma and confirmed them by microscale affinity capture technique and calibrated with standard phosphopeptides (phosphoserine, phosphothreonine, and phosphotyrosine) [2]. Using same approach in MCF-7 induced breast cancer model of explants tumor we performed proteomics analysis to identify and characterize tumor-associated protein variants associated with apoptosis by two-dimensional electrophoresis (shown in Fig. 1) and MALDI mass spectrometry. We characterized the influence of N-methyl-N-nitrosourea (genotoxic nitroso compound) as tumor-inducing agent on the protein pattern of breast malignant tumor in rat. We found several tumor apoptosis-associated variants AKR1C1 or -C3, AKR1B1 representing the proteins of the aldo-keto reductase superfamily. We believed that apoptosis-associated protein induction and/or protein inhibition were related to the carcinogen MCF-7 bio-oxidation used for tumor induction. The most prominent tumor-associated protein, rat aldose reductase-like protein-1 (rARLP-1) (69% sequence identity to lens aldose reductase) and three additional types of rARLP-1 were detected in nitroso compound-induced rat tumors, while rat aldo-keto reductase protein-c (Rak-c), a novel tumor-associated variant (65% sequence identity with 3 -hydroxysteroid dehydrogenase) was identified in Nmethyl-N-nitrosourea-induced breast tumors (unpublished data). Reduced 3 -hydroxysteroid dehydrogenase and 4 -3ketosteroid-5 -reductase enzymes both were tumor-specific detoxification independent of MCF-7 induction. We believe that MCF-7 carcinogen leaves a specific fingerprint(s) at the proteomics level to manifest breast tumors. In contrast, members of the aldo-keto reductase superfamily were not

reported as associated with MCF-7 induced changes in proteomics peaks in breast tumor [37]. At this point much remains to explore and investigate the apoptosis proteins in breast tumor associated with induction or due to MCF-7 induction. 3D volume construction and 3D MALDI imaging coregistration with MRI/PET digital images was somewhat trivial because of MALDI sensitivity to protein mass; MRI sensitivity to protein bound sodium and PET sensitivity to radiolabel [18F] bound glucose. Moreover, apoptosis associated proteins and intracellular sodium bound proteins may or may not be dependent on glucose uptake (sodium symporter) or sodium pump and oxygen state in cell [38, 39]. Earlier reports suggested the active role of intracellular sodium and elevated glycolysis and reduced apoptosis in tumorigenesis and reversing or arresting or slowing down by anticancer drug chemosensitivity [40]. However, coregistration of 3D co-ordinates on MRI/PET and histology digital images was decisive and comparable with other report [41, 42]. The present study extended one more imaging MALDI modality to get composite information of tumor protein molecular details. The relationship of PET (18-FDG-glucose signal intensities) and MRI (sodium signal intensities)with MALDI (protein distribution at different tumor locations) and matched cytomorphological details (histology) may serve the purpose of cytomorphometrics and glycolytic tumor characterization with location of proteins (proteomics maps) to monitor realtime tumor chemosensitivity as a tool with possibility of in vivo diagnostic and therapeutics interpretation. The real time monitoring of docetaxal (Taxotere) drug chemosenstivity effect during 0-48 hours was demonstrated in present study in terms of shrunken tumor mass by sodium MRI and decrease in hyperglycolytic tumor tissue with possible MALDI-IMS visible premalignancy malignancy specific tumor protein(s). However, the chemosensitivity effect was reduced at 48 hours end-point in comparison with chemosensitivity effect at 24 hours end-point. The tumor ex vivo cytomorphometry features added quantitative evaluation of drug chemosensitivity and supported to our previous observations [1, 20, 43]. However, there are several limitations of the use MRI/PET and predicting the tumor features. First, the tumor shrinkage during tissue processing of 4 micron tissue histology section limits the measurement of tumor histology area compared with 1000 micron MRI images. Second, 18FDG PET images are dynamic and their projection images limit the correlation of glycolysis rich signals with sodium MRI signals. Third, different tumor features observed under high power fields may not always true representative of tumor staging as assumed in criteria. Fourth, identification of major carcinogenic responsible tumor proteins is a challenge because MALDI peaks are showing m/z peaks of proteins or peptides from a very small tumor region (difficult to take away specimen from big mass of tumor) while PEG electrophoresis protein/peptide map shows presence of tumor proteins (with different pI) in large number without any confirmation of responsible tumorgenic or apoptotic or premalignant protein(s). However, the intracellular sodium and glycolysis relationship stands valid in progressing tumor

Taxotere Chemosensitivity Evaluation in Rat Breast Tumor

Recent Patents on Medical Imaging, 2011, Volume 1, No. 2

13

cells as indicated earlier [43]. However, it remains to determine if tumor proteomics signatures have any impact on in vivo MRI-PET imaging signal contrast. Other important issue was how spatial and quantitative information from proteomics may extend the protein predictability and implications of in vivo MRI-PET image contrast distribution due to presence of odd numbered sodium nuclei and radiolabeled glucose. Proteomics automated image processing data [44] from in vivo MALDI-IMS studies may have potentials to test drug action or to predict functional regulatory protein information responsible of tumor apoptosis and angiogenesis (proteomics profiling), signaling mechanism and molecular mechanism of programmed tissue degradation (protein expression) and cancer protein mapping [45-50]. The spatial distribution of tumor cell protein molecules as false color maps can open window to the visible biochemical changes with insight of biophysical basis of MRI image contrast and physiological basis of PET contrast [51]. In future, technical advancements may accomplish the purpose of quantitative noninvasive MALDI imaging combined with multinuclear in vivo protonintracellular sodium and glycolysis imaging indicators of tumorigenesis (apoptosis, necrosis, proliferation, premalignancy or malignancy) to test anticancer drug chemosensitivity [52]. It remains to see the new inventions how advanced techniques solve the problem of integrating in vivo imaging data with ex vivo molecular imaging data to construct threedimension tumor volume of molecular details or IMAGINGTHERAPROTEOMICS to test anticancer drug effects. CURRENT AND FUTURE DEVELOPMENTS Recently several inventions and patents have suggested the possibility of multimodal imaging by integrating digital data from morphometric imaging with molecular imaging such as MALDI, immunostaining. The present review showed the distribution of 18-FDG-PET and sodium MRI signal intensities in tumor as measurable and diagnostic by imaging methods. Still there is no consensus on best configuration for PET/MRI system. There are three main approaches of PET/MRI integration architecture: sequential, insert and integrated types [53]. Major challenges are: 1. Potential cross talk effects in front-end electronics due to fluctuations in light yield of scientillators in PET detectors caused by rapidly changing MR gradients and RF signals; 2. Magnetic inhomogeneities; 3. Compensation of Eddy currents and better shimming; 4. Better PET attenuationscatter-random coincidence correction algorithms; 5. Detector technology with matching scintillation crystals combined with less sensitive light sensors. In future new technology of magnetic field insensitive avalanche photodiodes, design shielded PET electronics will be available to avoid electromagnetic interference. In future, quantitative MRIPET-MALDI-histoimmunostaining criterion can or will distinguish apoptosis-rich and benign or malignant tumor features for theradiagnostics. Sodium MRI and PET image intensities is a new information showing positive correlation with histology and apoptosis premalignancy proteomics indices as rapid drug monitoring time-dependent assay. In this direction, recently inventors modified and suggested design of transparent MALDI slides [54], antibody-peptide conjugate mediated MALDI imaging by fast fragmentation

method [55] and new thresholding techniques of MALDI peak selection. 3D digital mapping of MALDI is in infancy. CONCLUSION The physical basis of MALDI imaging and MRI-PET data integration is explored and patents are reviewed with a focus on the progress of quantitative MRI-PET and MALDI protein detection applications to test anticancer drug. Review of patents showed the approach of integrated MRI/PET imaging and immunostaining, histology and MALDI data may show correlation as sensitive, tumor specific, accurate reproducible and precise to define apoptosis in theradiagnostics of breast tumors in experimental rats. ACKNOWLEDGEMENTS This manuscript in part was presented at peer-reviewed AFLAC award at AACR meet 2002, ISMRM workshop 2001 and ISMRM annual meet 2002. MALDI-IMS data was presented by Doris Terry at ASMS 2007. The authors wish to acknowledge the experimental data and expertise provided by Drs. Ed X. Wu, Paul Cannon, van Heertum, Kenny Hess at Radiology department and Dr. Matthias Schbolcs and Dr. Mansukhani at Pathology department and helping in these imaging and continuing tumor histology experiments. Authors wish to acknowledge the MALDI-IMS and peak analysis done by Dr. Doris Terry at Florida State University, Tallahassee, Florida. Grant source: Aventis Pharmaceuticals Company, Bridge-water, NJ. Figures were improved by Mr. Magesh Sadasivam at Amity Institute of Nanotechnology, Amity University UP, NOIDA, India. CONFLICT OF INTEREST Authors do not have any financial or commercial conflict. REFERENCES
[1] [2] [3] [4] [5] [6] [7] [8] [9] [10] [11] [12] Kline RP, Wu EX, Petrylak DP, et al. Rapid in vivo monitoring of chemotherapeutic response using weighted sodium magnetic resonance imaging. Clin Cancer Res 2000; 6: 2146-56. Sinha TK, Khatib-Shahidi S, Yankeelov TE, et al. Integrating spatially resolved three-dimensional MALDI IMS with in vivo magnetic resonance imaging. Nat Methods 2008; 5: 57-9. Fournier I, Thomy V, Salzet M, Wisztorski M, Verplanck N. Masks useful for MALDI imaging of tissue sections, processes of manufacture and uses thereof. US20090197295, 2009. Truche JL, Overney GT, Fisher WD, Tella RP. MALDI sample plate imaging workstation. US7495231, 2009. Caprioli RM. Method and apparatus for imaging biological samples with MALDI MS. US5808300, 1998. Diamond SL, Gosalia D. Method and devices for running reactions on a target plate for maldi mass spectrometry. US0298515, 2007. Ogawa K, Takeuchi S, Harada T, Ueno Y, Inoue F, Setou M. Imaging mass spectrometer. US20070114388, 2007. Bamberger C, Bamberger A. Imaging mass spectrometry principle and its application in a device. US0261243, 2009. Bogyo MS, Blum G, von Degenfeld G. Imaging of protease activity in live cells using activity based probes. US20070036725, 2007. Bui HA. Reduction of scan time in imaging mass spectrometry. US20070141719, 2007. Shiea, J. Mass spectrometric imaging method under ambient conditions using electospray-assisted laser desorption ionization mass spectrometry. US7196525, 2008. Sparkman OD, Colby SM. Sample imaging. US7196525, 2007.

14 Recent Patents on Medical Imaging, 2011, Volume 1, No. 2 [13] [14] [15] [16] Holle A. Imaging mass spectrometry for small molecules in twodimensional samples. US0296488, 2008. Schepkin VD, Ross BD, Chenevert TL, et al. Sodium magnetic resonance imaging of chemotherapeutic response in a rat glioma. Magn Reson Med 2005; 53: 85-92. Ouwerkerk R, Bleich KB, Gillen JS, Pomper MG, Bottomley PA. Tissue sodium concentration in human brain tumors as measured with 23Na MR imaging. Radiology 2003; 227: 529-37. Somer EJ, Marsden PK, Benatar NA, Goodey J, O'Doherty MJ, Smith MA. PET-MR image fusion in soft tissue sarcoma: Accuracy, reliability and practicality of interactive point-based and automated mutual information techniques. Eur J Nucl Med Mol Imaging 2003; 30: 54-62. Unlu MZ, Krol A, Magri A, et al. Computerized method for nonrigid MR-to-PET breast-image registration. Comput Biol Med 2010; 40: 37-53. Fan Y, Wang J, Yang Y, et al. Detection and identification of potential biomarkers of breast cancer. J Cancer Res Clin Oncol 2010; 136: 1243-54. Knochenmuss R, Zhigilei LV. Molecular dynamics simulations of MALDI: Laser fluence and pulse width dependence of plume characteristics and consequences for matrix and analyte ionization. J Mass Spectrom 2010; 45: 333-46. Sharma R, Kline R. Flow cytometry, MRI, PET and NMR spectroscopy methods of non-invasive drug monitoring in prostate tumor: Technical Note. Proc of 16th IEEE Comput Based Med Syst 2003; 16: 263- 8. Maddalo G, Petrucci F, Iezzi M, et al. Analytical assessment of MALDI-TOF imaging mass spectrometry on thin histological samples. An insight in proteome investigation. Clin Chim Acta 2005; 357: 210-8. Vaquero JJ, Desco M, Pascau J, et al. PET, CT and MR image registration of the rat brain and skull. IEEE Trans Nucl Sci 2001; 48: 1440-5. Press WH, Teukolsky SA, Vetterling WT, Flannery BP. Numerical recipes in C. 2nd ed. Cambridge: Cambridge University Press 1992. Bttcher P, Maierl J, Hecht S, Matis U, Liebich HG. Automatic image registration of three-dimensional images of the head of cats and dogs by use of maximization of mutual information. Am J Vet Res 2004; 65: 1680-7. Arun KS, Huang TS, Blostein SD. Least squares fitting of two 3D point sets. IEEE Trans Pattern Anal Mach Intell 1987; 9: 699-700. West J, Fitzpatrick JM, Wang MY, et al. Comparison and evaluation of retrospective intermodality brain image registration techniques. J Comput Assist Tomogr 1997; 21: 554-66. Sharma R, Kline RP, Wu EX, Katz JK. Rapid in vivo taxotere quantitative chemosensitivity response by 4.23 tesla sodium MRI and histo-immunostaining features in N-Methyl-N-Nitrosourea induced breast tumors in rats. Cancer Cell Int 2005; 5: 26. Maes F, Collignon A, Vandermeulen D, Marchal G, Suetens P. Multimodality image registration by maximization of mutual information. IEEE Trans Med Imaging 1997; 16: 187-98. Fei B, Wang H, Muzic RF Jr, et al. Deformable and rigid registration of MRI and microPET images for photodynamic therapy of cancer in mice. Med Phys 2006; 33: 753-60. Chen ST, Pan TL, Tsai YC, Huang CM. Proteomics reveals protein profiles changes in doxorubicin-treated MCF-7 human breast cancer cells. Cancer Lett 2002; 181: 95-107. Chuthapisith S, Layfield R, Kerr ID, Hughes C, Eremin O. Proteomic profiling of MCF-7 breast cancer cells with chemoresistance to different types of anticancer drugs. Int J Oncol 2007; 30: 1545-51. Perera CN, Spalding HS, Mohammed SI, Camarillo IG. Identification of proteins secreted from leptin stimulated MCF-7 breast cancer cells: A dual proteomic approach. Exp Biol Med 2008; 233: 708-20. Woods RP, Mazziotta JC, Cherry SR. MRI-PET registration with automated algorithm. J Comput Assist Tomogr 1993; 17: 536-46. [34] [35] [36] [37]

Sharma and Katz Skerl D, Likar B, Pernus F. A protocol for evaluation of similarity measures for rigid registration. IEEE Trans Med Imaging 2006; 25: 779-91. Li J, Zhao J, Yu X, et al. Identification of biomarkers for breast cancer in nipple aspiration and ductal lavage fluid. Clin Cancer Res 2005; 11: 8312-20. Sanders ME, Dias EC, Xu BJ, et al. Differentiating proteomic biomarkers in breast cancer by laser capture microdissection and MALDI MS. J Proteome Res 2008; 7: 1500-7. Zeindl-Eberhart E, Klugbauer S, Dimitrijevic N, Jungblut PR, Lamer S, Rabes HM. Proteome analysis of rat hepatomas: Carcinogen-dependent tumor-associated protein variants. Electrophoresis 2001; 22: 3009-18. Andersson M, Groseclose MR, Deutch AY, Caprioli RM. Imaging mass spectrometry of proteins and peptides: 3D volume reconstruction. Nat Methods 2008; 5: 101-8. Hayakawa N, Uemura K, Ishiwata K, et al. A PET-MRI registration technique for PET studies of the rat brain. Nucl Med Biol 2000; 27: 121-5. Sharma R, Katz JK. Taxotere chemosensitivity evaluation in mice prostate tumor: Validation and diagnostic accuracy of quantitative measurement of tumor characteristics by MRI, PET and histology of mice tumor. Technol Cancer Res Treat 2008; 7: 175-85. Lyons SK. Advances in imaging mouse tumour models in vivo. J Pathol 2005; 205: 194-205. Humm JL, Ballon D, Hu YC, et al. A stereotactic method for the three-dimensional registration of multi-modality biologic images in animals: NMR, PET, histology, and autoradiography. Med Phys 2003; 30: 2303-14. Sharma R, Kline RP, Katz JK. Apoptosis characterization by flouro-deoxy glucose and intracellular sodium distribution using MicroPET/MRI in PC3 induced prostate cancer in mice tumor. Nanotech 2005; 1: 71-3. Lee RE, Welch EB, Cobb JG, Sinha T, Gore JC, Yankeelov TE. Implementation of a semi-automated post-processing system for parametric MRI mapping of human breast cancer. J Digit Imaging 2009; 22: 424-36. McDonnell LA, Heeren RM. Imaging mass spectrometry. Mass Spectrom Rev 2007; 26: 606-43. Caldwell RL, Caprioli RM. Tissue profiling by mass spectrometry: A review of methodology and applications. Mol Cell Proteomics 2005; 4: 394-401. Reyzer ML, Caprioli RM. MALDI-MS-based imaging of small molecules and proteins in tissues. Curr Opin Chem Biol 2007; 11: 29-35. Stoeckli M, Staab D, Schweitzer A. Compound and metabolite distribution measured by MALDI mass spectrometric imaging in whole-body tissue sections. Int J Mass Spectrom 2007; 260: 195202. Chaurand P, Norris JL, Cornett DS, Mobley JA, Caprioli RM. New developments in profiling and imaging of proteins from tissue sections by MALDI mass spectrometry. J Proteome Res 2006; 5: 2889-900. Francese S, Dani FR, Traldi P, Mastrobuoni G, Pieraccini G, Moneti G. MALDI mass spectroscopy imaging from its origins up to today the state of the art. Comb Chem High Throughput Screen 2009; 12: 156-74. Cornett DS, Reyzer ML, Chaurand P, Caprioli RM. MALDI imaging mass spectrometry: Molecular snapshots of biochemical systems. Nat Methods 2007; 4: 828-33. Franck J, Arafah K, Elayed M, et al. MALDI imaging mass spectrometry: State of the art technology in clinical proteomics. Mol Cell Proteomics 2009; 8: 2023-33. Delso G, Martinez-Mller A, Bundschuh RA, et al. Evaluation of the attenuation properties of MR equipment for its use in a wholebody PET/MR scanner. Phys Med Biol 2010; 55: 4361-74. Sun K, Young HM. MALDI imaging slide. KR980214B1, 2010. Remi L, Isabelle F, Michel S, Michel D, Jean-Claude TE, Gottfried P, Ivo R, Marc L. MALDI tissue imaging using conjugates cleavable fast fragmentation. EP2322920A1, 2011.

[17] [18] [19]

[38] [39] [40]

[20]

[41] [42]

[21]

[43]

[22] [23] [24]

[44]

[45] [46] [47] [48]

[25] [26] [27]

[28] [29] [30] [31]

[49]

[50]

[51] [52] [53] [54] [55]

[32]

[33]

Differentiating Proteomic Biomarkers in Breast Cancer by Laser Capture Microdissection and MALDI MS
Melinda E. Sanders,*,,O Eduardo C. Dias, Baogang J. Xu,,+ James A. Mobley,,+ Dean Billheimer,| Heinrich Roder, Julia Grigorieva, Mitchell Dowsett,# Carlos L. Arteaga,,,O and Richard M. Caprioli*,,O,+
Departments of Pathology, Medicine, Biochemistry, Statistics, Cancer Biology, Breast Cancer Research Program, and the Mass Spectrometry Research Center, Vanderbilt University, Nashville, Tennessee 37232, Biodesix, Inc., Steamboat Springs, Colorado 80477, and Royal Marsden Hospital and Institute of Cancer Research, London, U.K.
Received December 03, 2007

We assessed proteomic patterns in breast cancer using MALDI MS and laser capture microdissected cells. Protein and peptide expression in invasive mammary carcinoma versus normal mammary epithelium and estrogen-receptor positive versus estrogen-receptor negative tumors were compared. Biomarker candidates were identied by statistical analysis and classiers were developed and validated in blinded test sets. Several of the m/z features used in the classiers were identied by LC-MS/MS and two were conrmed by immunohistochemistry.
Keywords: MALDI MS laser capture microdissection breast cancer

Introduction
Breast cancer is the leading cause of cancer the USA for women and ranks second in cancer deaths, with an estimated 182 460 new cases of invasive cancer, 67 770 cases of noninvasive breast cancer, and 40 480 deaths in 2008. Rather than one disease, it is a heterogeneous group of neoplasms, some of which are locally aggressive and may metastasize early, while other forms proliferate slowly and may be cured by surgical excision alone. Among the subset of special type carcinomas1 an excellent prognosis may be indicated by histology alone (e.g., Pure tubular carcinoma); however, these represent less than 15% of all breast cancers. The majority of no special type (aka. ductal carcinomas) have by denition no distinctive features. Clinical decision making including the need for systemic adjuvant therapy25 is currently based on a combination of estrogen (ER) and progesterone (PR) receptor status and expression levels, presence or absence of Her2-neu gene amplication, tumor size, grade, proliferative rate, and stage.6
* To whom correspondence should be addressed. Melinda E. Sanders, M.D., Vanderbilt University Medical Center, 23rd and Pierce Ave., 4918-B TVC Bldg., Nashville, TN 37232. Phones, 615-322-1410(ofce), 615-3439060(secretary); fax, 615-343-9563; e-mail, melinda.sanders@vanderbilt.edu. Richard M. Caprioli, Ph.D., Mass Spectrometry Research Center, 465 21st Avenue South, 9160 MRB-III, Vanderbilt University School of Medicine, Nashville, TN 37232, Phone, 615-322-4336, fax, 615-343-8372, e-mail, r.caprioli@vanderbilt.edu. Department of Pathology, Vanderbilt University. O Breast Cancer Research Program, Vanderbilt University. Department ofMedicine, Vanderbilt University. Department of Biochemistry, Vanderbilt University. + Mass Spectrometry Research Center, Vanderbilt University. | Department of Statistics, Vanderbilt University. Biodesix, Inc. # Royal Marsden Hospital and Institute of Cancer Research. Department of Cancer Biology, Vanderbilt University.

Retrospective patient analyses including gene expression proling suggest that differences in intrinsic biology of individual tumors have important implications for therapy and prognosis and that these differences are often not discernible on a histological basis. Subsequent predictors of prognosis in breast cancer based on cDNA expression710 have been developed, some of which are in use in clinical trials.1113 However, one would expect there ultimately to be limits to their predictive power because mRNA expression is poorly correlated with the functional protein component. Accordingly, proteomics which studies the active mediators of cellular processes is a required complement to gene expression analyses. Proteomic expression among breast cancer subtypes is largely unexplored and should prove to be an important complement to microarray studies and an excellent mechanism for further understanding these different phenotypes. Matrix-assisted laser desorption/ionization mass spectrometry (MALDI MS) can prole proteins at high sensitivity up to 50 kDa in tissues.14 This technology can directly measure many peptides and proteins in tumor tissue sections and can also be used for high resolution imaging of individual biomolecules present in tissue sections.1517 Coupled with laser capture microdissection (LCM), MALDI MS is an ideal approach for generation of separate protein proles of the invasive tumor and normal epithelial components of breast tumors and tissues. In addition, epithelial elements usually compose only 515% of normal breast tissue making LCM mandatory in most cases to ensure a dominantly epithelial sample for evaluation. We aimed to use MALDI MS to assess protein expression proles in approximately 2000 cells from frozen sections of surgically resected breast tumors and reduction mammoplasty tissue, and to assess the resulting data using ProTS Marker software (Biodesix, Inc., Steamboat Springs, CO). The goal of this project
10.1021/pr7008109 CCC: $40.75 2008 American Chemical Society

1500 Journal of Proteome Research 2008, 7, 15001507

Published on Web 04/04/2008

Biomarkers in Breast Cancer

Table 1. Clinical and Pathologic Characteristics of Samples Across Centers
a

research articles
centers 1 2 3 4 5 Totals

IMC Average age yrs (no.)b Grade Low Intermediate High Histologic type No Special type Special typec Stageb I II III Hormone Receptor Status ER+/PR+ ER+/PRER-/PR+ ER-/PRNME Average age (yrs)

24 66 (n ) 22)

9 57 (n ) 4)

13 65 (n ) 13)

3 49 (n ) 3)

73 53 (n ) 73)

122

7 15 2

1 7 1

0 2 11

0 0 3

8 27 38

16 51 55

23 1

6 3

13 0

3 0

70 3

115 7

5 6 5

0 0 2

1 4 3

0 0 2

11 19 15

17 31 27

10 11 0 3 0 N/A

6 1 0 2 26 36

5 3 0 5 91 33

0 0 0 3 31 33

20 14 0 34 19 31

41 29 0 47 167

a Centers: (1) Royal Marsden Hospital, U.K.; (2) CHTN Eastern Division, University of Pennsylvania, Philadelphia, PA; (3) CHTN Midwestern Division, Ohio State University; (4) CHTN Southern Division, University of Alabama, Birmingham, AL; (5) CHTN Western Division, Vanderbilt University. b Information was not available for all women. No reduction mammoplasty specimens were obtained from center 1. c Special type cancers: center 1, lobular-2; center 2, mucinous carcinoma-1, lobular carcinoma-2; center 5, tubular-1, mucinous carcinoma-2.

was to provide a distinctive protein prole of each tumor and assess the ability of our analysis algorithms to classify the tumors into ER-positive and ER-negative subgroups based on differences in these patterns. Recent results have shown that MALDI MS-based diagnostics can be highly reproducible across different laboratories, and may overcome some of the ambiguities arising from other techniques.18

Experimental Procedures
Tissue Collection and Evaluation. A total of 122 invasive mammary carcinomas (IMC) and normal mammary epithelium (NME) from 167 reduction mammoplasty specimens were analyzed in this study. These samples were derived from 289 women. These tissue samples were collected and distributed to our laboratory in a deidentied fashion by the four divisions of the Cooperative Human Tissue Network and the Royal Marsden Institute, U.K. (M.D.). Table 1. details the distribution of clinical and pathologic characteristics across centers. None of these women had received preoperative hormonal, chemo-, or radiation therapy. These tissues were obtained at the time of the womans primary surgery, snap-frozen in liquid nitrogen within 30 min after removal from the patient, and stored at 80 C until analyzed. The presence of tumor or NME was conrmed by a board-certied pathologist who specializes in breast disease (M.E.S.) who examined a frozen section of each tissue block and subsequently performed LCM on appropriate areas. Tissue Sample Preparation. Sections for microdissection were prepared according to our previously developed protocol.19 In brief, using a cryostat, 7 m frozen tissue sections were

mounted on uncharged glass slides without the use of embedding media and placed immediately in 70% ethanol for 1 min. Subsequent dehydration was achieved using graded alcohols and xylene treatments as follows: 95% ethanol for 30 s (2 times), 100% ethanol for 30 s (2 times), and xylene for 5 min (2 times). Slides were then dried in a laminar ow hood for 10 min prior to microdissection. Laser Capture Microdissection. LCM was performed using the PixCell IIe LCM system (Arcturus, Mountain View, CA). Depending on the size of the lesion, 500-1000 shots using the 7.5 or 15 m infrared laser beam were utilized to obtain an average of 2000 cells. All samples were microdissected in duplicate. Preparation of Microdissected Cells for MALDI MS. MALDI MS was performed directly on the LCM acquired cells. After LCM, the thermoplastic lm was removed from the LCM cap using forceps and placed onto the MALDI plate using conductive double-sided tape. A nely pulled glass capillary was employed to deposit as little as 10 nL of matrix solution as required to cover the captured cells under microscopic visualization. The matrix solution consisted of sinapinic acid at 20 mg/mL in 6/4/0.01 (v/v/v) acetonitrile/water/TFA. MALDI MS Analysis. MALDI MS analysis was performed using a Voyager DE-STR MALDI time-of-ight mass spectrometer (Applied Biosystems, Framingham, MA) with a 337 nm nitrogen laser. Acquisition was achieved in the linear positive ion mode under optimized delayed extraction conditions as described previously.14,15,2024 Approximately 750 laser shots were averaged to create a single spectrum from the captured cells. In most cases, we generated three spectra per sample
Journal of Proteome Research Vol. 7, No. 4, 2008 1501

research articles
which were then combined to create one average spectrum which was used in the statistical analysis. In a subset of cases where the number of tumor cells was very small, we were able to generate only one spectrum. In this analysis, signals in the mass-to-charge (m/z) range from 2000 to 30 000 Da were considered. Immunohistochemistry. Estrogen receptor (ER) and progesterone receptor (PR) expression were evaluated using the Zymed PREDILUTED ER antibody 6F11 (S., San Francisco, CA) and the DAKO PgR636 PR antibody (Carpinteria, CA) each at a 1:100 dilution and incubated for 1 h at room temperature. Immunohistochemistry was performed on frozen sections from the same tissue block on which LCM was performed. The results were scored as the percentage of tumor cells with nuclear staining. Tumors were considered to be ER-positive or PR-positive when >10% nuclear staining was observed at any intensity. Data Processing and Statistical Analysis. 1. Spectral Preprocessing. Preprocessing of spectra is necessary to render spectra comparable and to ensure the reproducibility of the statistical analysis procedure. Preprocessing was performed using proprietary analysis tools developed by Biodesix, Inc., CO. A detailed description can be found at http://www.biodesix.com in the technology section. Raw spectra were sent to Biodesix (Steamboat Springs, CO) for analysis. Mass spectra generated over 2 years time, although run by the same personnel and on the same instrument, may exhibit variation. To enable analysis of these spectra, we applied a suite of preprocessing procedures2529 and developed some additional procedures. In brief, the background and noise were estimated and then subtracted from each spectrum based on local noise estimators using a local (in m/z) robust asymmetric estimator2529 and normalized to total ion current (TIC). Peaks were detected using a signal-to-noise ratio (S/N) cutoff of 4.0, which was found to be a good compromise between overdetection and sensitivity. Spectra were aligned sequentially in three steps. The set of common peaks appear in a plurality of spectra whose m/z values differ by < (5 Da was used for the alignment process using a polynomial up to quadratic terms. The sets of points for the alignment were selected based on the criteria of communality (present in at least 2/3 of all spectra), S/N (S/N > 5 in the rst alignment, S/N > 4 in the second and third alignment), and roughly even distribution along the m/z range. The following set of common peaks was used in the rst phase of the alignment process: m/z 3373.5, 4939.7, 5359.1, 5654.4, 6548.4, 7670.7, 8570.2, 10091.7, 10839.3, 11309.6, 11349.3, 11650.0, 12346.6, 13781.1, 14005.6, 15346.4, 15860.2, 17885.1, 17926.1. Spectra were saved after the rst alignment, reloaded, aligned again using a second set of peaks, and saved (m/z 3449.0, 4939.8, 5360.7, 5653.9, 6175.2, 6546.9, 6650.0, 6890.2, 7004.7, 8091.9, 9154.7, 10093.9, 10841.5, 11306.4, 113450.0, 11653.78, 12347.9, 13780.2, 14009.9, 15342.6, 17893.3, 20945.4). Replicate spectra after the second alignment were used to create an average spectrum for each individual LCM sample. The third alignment was performed on the averaged spectra using a third set of alignment peaks (m/z 4047.0, 4937.7, 5358.3, 5652.8, 5939.5, 6175.8, 6277.6, 6547.7, 6665.1, 6890.6, 7005.8, 8413.1, 8568.4, 9155.0, 9971.6, 10094.7, 10142.6, 10843.6, 11309.1, 11653.4, 12349.0, 12652.4, 13780.1, 14010.0, 15341.1, 15867.9, 17897.5, 20758.4, 26600.1). The preprocessing procedure was optimized using the training set and held xed for the classication of all test sets.
1502 Journal of Proteome Research Vol. 7, No. 4, 2008

Sanders et al.
2. Feature Denition and Selection. Each MALDI spectrum is characterized by a set of features, which are dened as integrated, background-subtracted, and normalized spectral intensities integrated over a chosen m/z range containing a peak. The m/z range for each feature was calculated from the alignment error and the local peak width of each spectrum. The features were predened from a set of peaks that were common (within a predened tolerance of 0.5 Da) to at least 3 spectra of each clinical group. A combination of a selected subset of features and of the algorithm (and its parameters), which assigns a clinical label to a spectrum, constitutes a classier. Candidate features for the classication algorithms were identied as differentially expressed m/z values from spectra from IMC versus NME and ER-positive versus ERnegative tumors. The initial selection of signicant features for the classier was based on a calculation of univariate p-values from MannWhitney U-tests (Wilcoxon rank sum test). Then a variant of a oating search method was applied.30 The oating search looks iteratively at combinations of the signicant features to see how they perform as a classier. As an optimization criterion, we used the leave-one-out cross validation (LOOCV) error on the training set. Finally, a visual inspection of features was carried out using the graphs of the group averaged spectra in the ProTS Marker software. In some cases, the feature widths were manually adjusted during the training process to take asymmetric peak shapes into account. 3. Classication Algorithm. The classication algorithm used was a straightforward implementation of a k-nearestneighbor (KNN) algorithm.31 KNN requires as parameters a set of representative and labeled instances (i.e., list of selected feature values from a training sample set). Samples from each clinical group of interest (IMC vs NME and ER-positive vs ERnegative tumors) were randomly split into training and test sets. Aligned average spectra were used for each sample instance, both in the training and in the test sets. To classify a new spectrum, the KNN algorithm rst calculates the Euclidean distance of the feature values of the new spectrum to those of the training set spectra. This calculation yields a list of distances from the test spectrum to each representative spectrum. For the k-nearest neighbors (those with the k smallest distances), the labels are compared. Finally, the assigned label is a simple majority vote over the k-nearest neighbor labels. The number k of the nearest neighbors was the only classier parameter used. 4. Cross-Validation of a Classier and Validation. Crossvalidation is an important tool in the assessment of classier performance during the training phase. A xed number (one for LOOCV or a prescribed number N for LNOCV) of instances was removed from the training set, a classier was generated using the remaining instances, and the performance of this classier was evaluated by applying it to the left-out instances and comparing classier and true labels. Ideally, this analysis is performed for all possible permutations of left-out and kept instances in the training set. The average of the classication performance over these permutations is an estimate of the expected performance of the classication. Classier parameters and the set of selected features were optimized using these cross-validation procedures. After training and optimization, all parameters were frozen. No changes in the classication algorithm were allowed during the validation performed on the independently and randomly selected test spectra. 5. Assignment of Training and Test Groups. Average spectra of IMC samples (122 instances) and NME samples (167

Biomarkers in Breast Cancer

instances) were split into test and training sets. A random numbers generator was used to assign each sample a number, then all samples were sorted by their numbers; the rst 62 IMC instances and 84 NME instances were assigned to the training set and used for generation of a classier to distinguish IMC from NME. The remaining IMC and NME spectra were assigned to the test set. The same approach was used to randomly split the group of 117 IMC spectra with known ER status into ER+ and ER- training and test sets. The training set consisted of 36 ER+ and 25 ER- tumors, while the test set consisted of the remaining 32 ER+ and 24 ER- tumors. Protein Identication. Once candidate molecular weight markers were selected by the class prediction model, we utilized a combination of MS techniques previously validated by the VUMSRC15,32 to isolate and identify several of the protein species of interest. The remainder of each specimen used for LCM was kept frozen to enable subsequent protein extraction and identication. Tissues with the highest relative intensity of the features of interest were selected for use in the protein identication. A total of 12 tissue fragments, 5 IMC and 7 normal, were subsequently selected from this list because the MALDI spectra generated from these samples contained the largest total number of the signicant differentially expressed peaks. Protein extracts were prepared using 3 to 4 mm3 portions of mammary tissue in 1:20 (w/v) Tissue Protein Extraction Reagent (T-PER) plus Halt Protease Inhibitor Cocktail (10 L/ mL) (Pierce, Rockford, IL). Protein extracts were subjected to a cleanup step using a disposable hand-packed C18 preparative column from Waters (Milford, MA). Samples were then fractionated with a Vydac (Hesperia, CA) 208TP5315 reversed-phase C8 polymeric column at 40 C using a Waters Alliance HPLC system (Milford, MA) using a ow rate of 0.5 mL/min. Thirty second fractions were collected into a 96 well plate using a Gilson Fraction Collector (Middleton, WI) and further analyzed by MALDI MS and Flex Analysis software (Bruker-Daltonics, Billerica, MA) to determine the fractions containing the peaks of interest. The remaining volume of the fractions containing peaks of interest was separated by one-dimensional gel electrophoresis. Bands were excised corresponding to the m/z peaks of interest. Bands were in-gel-digested with trypsin, and subjected to LC-MS/MS analysis on a Thermo LTQ linear ion trap instrument equipped with a Thermo nanoelectrospray source, Surveyor LC system, and autosampler (Thermo Fisher, San Jose, CA). Tandem MS spectra were search against the UniRef human database using SEQUEST (Thermo Electron, San Jose, CA) and data ltered based upon the following ltering criteria: cross correlation (Xcorr) value of >1.9 for singly charged ions, > 2.2 for doubly charged ions, and >3.75 for triply charged ions. A RSp (ranking of primary score) value of <4 and a dCN value of g0.1 were also required for positive peptide identications.33 A more detailed description of the protein identication methods is given in the Supporting Information. Validation of Differentially Expressed Features. Two of the differentially expressed features were conrmed using two commercially available tissue microarrays composed of parafn-embedded tissue. The AcuMax A202IV array (ISU Abxis Co.) contained 45 breast cancers and 4 adjacent normal tissues in duplicate cores. The US Biomax array consisted of 24 breast tumors with self-matching adjacent tissue and normal tissue (Ijamsville, MD), and was accompanied by ER, PR, and Her2/ neu immunohistochemistry results and staging information. Each array was stained with the Sigma calcyclin/Anti-S100A6 antibody (St. Louis, MO) at a 1:125 dilution for 1 h at room

research articles
temperature and the DAKO calgranulin A/clone MAC 387 antibody (Carpinteria, CA) overnight at 4 C. Antigen retrieval for both antibodies was performed with proteinase K. The results were scored based on the intensity of cytoplasmic staining on a scale of 0 to 3+ and percentage of cells staining positively. The Goodman Kruskal was used as a measure of association between the ordered categorical calcyclin and calgranulin A staining and the clinicopathological variables. For all tests, differences were considered signicant for P-values less than 0.05. In this exploratory study, we seek to identify potentially interesting relationships (with p < 0.05), rather than control for experiment-wise error rate by using a reduced signicance level for individual tests. Representative photomicrographs showing staining with each antibody were taken with an Olympus DP-70 digital camera attached to an Olympus BX40 microscope with a 22 ocular and a 20 lens.

Results and Discussion

Reproducibility and Inuence of Clinical Covariates. The intersample reproducibility of spectra is illustrated in Supporting Information. Spectral replicas (after ProTS Marker preprocessing see above) were highly reproducible with a slightly higher variance of spectra in cancer samples than in normal samples. We have also compared groups of samples originating from different research centers using methods described above to evaluate possible differences attributable to specimen handling at the different centers. In subsequent analyses, we ensured that none of the features that had statistical signicance in interinstitutional sample comparisons were used in the classication of clinical groups. Details of this analysis are reported in the Supporting Information (Table 1). We performed similar comparisons of the NME from women 25-35 years and greater than 45 years to examine the possibility that age differences between samples may bias the classier. In subsequent analyses, we also ensured that none of the features that had statistical signicance in the age comparisons were used in the classication of clinical groups. Details of this analysis are reported in the Supporting Information (Table 2). Cancer versus Normal. To detect proteomic patterns in breast tumors, we assessed the protein expression proles of 122 IMC and 167 examples of NME from reduction mammoplasty specimens utilizing laser capture microdissection and MALDI MS. Spectra were obtained from an average of 2000 cells dissected from frozen breast tissue by a breast pathologist (M.E.S.) using a serial hematoxylin and eosin stained section as a guide. Using wrapper methodology and cross-validation as a criterion, we created a classier optimized for the correct classication of the training set. From 88 features considered, a set of 14 features was selected to minimize the LOOCV error in the analysis of the training set comparing IMC and NME (Table 2). The details of this feature selection and the LOOCV error for various k-values are presented in the Supporting Information (Tables 35). Using the class prediction model based on the selected signals and a k ) 7, we were able to distinguish IMC versus NME with 97% accuracy in the training cohort and 94% accuracy in the testing cohort with a sensitivity and specicity of 89% and 98%, respectively (Table 3). Discriminating features subsequently identied by the protein identication studies are shown in Figure 1A. The complete set of discriminating features is shown in the Supporting Information (Table 2). While tumor can be distinguished from normal mammary epithelium based on histology alone, we believe that demonstrating a high accuracy
Journal of Proteome Research Vol. 7, No. 4, 2008 1503

research articles
Table 2. Features Used in Classiers
Classier Greatest Relative expression
a

Sanders et al.
An advantage of our approach was the use of LCM-acquired cells. The genomic studies classifying these tumor types used whole tumor tissues; thus, the presence of stroma, inammatory cells, and small vessels also contribute to the tumor expression proles. Protein Identication. Although the spectral proles may be useful for classication and prognosis, clues to the underlying biology of neoplastic transformation and progression can be obtained from identication and functional investigation of these peptides and proteins. For protein identication, we prepared a tissue extract containing portions of 12 tissues (5 IMC and 7 normal) remaining after LCM and MALDI analysis and then fractionated the proteins by HPLC. The tissues were selected based on the fact that they had the highest relative intensity of the features of interest. Monitoring of the fractions by MALDI MS permitted identication of the fractions containing the peaks of interest. Fractions of interest were then run on a trycine gel, and in-gel trypsin digests were performed on bands or molecular weight regions of interest. The resulting peptide extracts were subjected to LC-MS/MS analysis. By this methodology, we were able to identify 126 proteins by 2 or more unique peptides; however, only 3 were among the statistically signicant discriminator peaks, ubiquitin m/z 8568, calcyclin (S100A6) 10094 m/z, and calgranulin A (S100A8) 10842 m/z (Table 2). Two of these classiers, calcyclin and calgranulin A, were conrmed by immunohistochemistry using two different commercially available TMAs, an AccuMax array containing duplicate cores of 45 tumors which were accompanied by ER, PR, and Her2/neu staining results and a US Biomax TMA which contained 24 tumors and matched normal tissues. Figure 2 shows the spectrum of staining intensity observed with both antibodies. By our MALDI MS analysis, calcyclin expression was greater in normal compared to tumor tissues in the IMC versus NME comparison (Table 2). Correspondingly, on the US Biomax TMA, the number of normal tissues and the intensity of expression of calcyclin was signicantly greater than in tumor tissues (Table 4; p ) 0.02). Interestingly, in a subset of cases, the myoepithelial component also stained positively for calcyclin; however, this relationship did not reach statistical signicance. By our MALDI MS analysis, calgranulin A showed increased expression in IMC relative to NME and served as a classier in the IMC versus NME comparison (Table 2), but we found no statistically signicant difference in the staining intensities on the Biomax array (Table 4). By our MALDI MS analysis, calgranulin A served as one of the features constituting the classier for ER-negative tumors (Table 2). Calgranulin A staining is negatively associated with ER status for tumors on the AccuMax array ( ) -0.53, p ) 0.007, see Table 4). Although not serving as a discriminator by our MALDI analysis, calcyclin expression is also negatively correlated with ER status () -0.67, p < 0.0001), which might be improved with performance of a true multivariate analysis as in the KNN classier based on protein proles. The S100 Ca(2+)-binding proteins, a subfamily of EF-hand Ca(2+)-binding proteins, recently became of major interest because of their differential expression in neoplastic tissues, their involvement in metastatic processes, and the clustered organization of at least 10 S100 genes on human chromosome 1q21, a region frequently rearranged in several tumors. Calcyclin has implied roles in the regulation of cell growth and division, exhibits deregulated expression in association with cell transformation, and is found in high abundance in certain

IMC vs NME

IMC

NME

m/z

4205, 4938, 5421, 5827, 7176, 8435, 8568, 10842, 11654

6891, 7651, 10094, 13782, 22602

Classier Greatest Relative expression

ER+ vs ER-

ER+ tumors

ER- tumors

m/z

7177

6548, 9155, 10842

a The m/z values in boldface were subsequently identied as ubiquitin (m/z 8568), calcyclin (m/z 10094), and calgranulin A (m/z 10842).

for a classier distinguishing these groups is a necessary proof of concept. The next task is correlation of the proteomic proles with known biomarkers such as estrogen receptor status. ER Status. Sufcient residual tissue was available for 117 tumors to evaluate ER status by immunohistochemistry (Table 1). Of these, 61 tumors were ER+ and 56 tumors were ER-. On the basis of a training set of 36 ER+ tumors and 25 ERtumors, a set of 4 features from among 94 features considered was selected to minimize the LOOCV error (Table 2). The details of this feature selection and the classication LOOCV error for various k-values are shown in the Supporting Information (Tables 68). Using the class prediction model based on the selected signals, a k-value of 5, and the remaining 32 ER+ and 24 ER- tumors, we were able to distinguish ER+ and ERtumors with 85% concordance with immunohistochemistry in the training cohort and 66.1% accuracy in the testing cohort with a sensitivity and specicity of 53% and 87.5%, respectively (Table 3). One discriminating feature subsequently identied by the protein identication studies is shown in Figure 1B. To our knowledge, this is the rst study to examine differences in proteomic expression among ER+ and ER- tumors utilizing MALDI MS of LCM acquired tumor cells. The study of protein expression within these tumor subsets should provide insights complementing those indicated by gene array studies because mRNA expression cannot always indicate which proteins are actually expressed and how their activity might be modulated after translation.34,35 Therefore, analysis of the proteome directly from tumor tissue may provide a better molecular snapshot of the pathological status of the cancer than gene expression patterns. As MALDI MS is an easy to use, reproducible, high-throughput technology, it may in the long run provide a cheaper and faster alternative to genetic and immunohistochemical approaches. Classication accuracy for the ER+ and ER- tumors was less accurate than anticipated. There are likely several factors contributing to this phenomenon. First, these subsets are themselves heterogeneous groups as is well-documented by gene array studies.36 Second, we were limited by the small number of samples available. Having 32 spectra in ER+ and 24 spectra in ER- groups of the training cohort was apparently not enough to create a robust classier. Thus, the results for ER-status classication should be considered as preliminary. Still, identication of Calgranulin A (see below) as a signicant discriminator indicates that even with a limited amount of samples we were able to obtain valuable information using our approach. We expect the classication to be noticeably improved and the protein identities of other signicant features identied once we have increased the sample size.
1504 Journal of Proteome Research Vol. 7, No. 4, 2008

Biomarkers in Breast Cancer

research articles

Figure 1. Representative features selected for classiers. (A) Three representative features from the IMC vs NME classier. The features m/z 8568, m/z 10094, and m/z 10842 in spectra generated from IMC are shown in blue and from NME in red. (B) A representative feature for the ER-positive vs ER-negative classier. The feature 10842 m/z from ER-positive tumors is shown in blue and the feature for ER-negative tumors is shown in red. The spectra in the left column represent median spectra from the two groups. Bold lines represent the median spectrum for each group, and the thin lines represent 25th and 75th percentile. The spectra in the right column show the specic peak as it appears in each individual spectrum. Table 3. Error Rates for the Classication of the Independent Test Setsa
Classier test set IMC (n ) 61) NME (n ) 83) total (n ) 144)

IMC vs NME

Correct Error

54 7

81 2
Classier

135 9

test set

ER+ (n ) 32)

ER(n ) 24)

total (n ) 56)

ER+ vs ER-

Correct Error

17 15

21 3

37 19

a Sensitivity, specicity, and accuracy for the IMC vs NME comparison are 88.5%, 97.6%, and 93.7%, respectively. Sensitivity, specicity, and accuracy for the ER+ vs ER- comparison are 53.0%, 87.5%, and 66.1%, respectively.

sion.39 It is interesting that in our study the ER- and Her2tumors showed the highest expression of S100A8 and these triple negative tumors are the subtype which contains tumors from patients carrying BRCA-1 mutations.36,40 Mutant forms of BRCA-1 may be incapable of S100A8 repression. Finally, Ohuchida et al. have shown that inhibition of S100A6 decreased proliferation and invasiveness of pancreatic cancer cell lines41 and Vimalachandran et al. demonstrated the high expression of S100A6 (Calcyclin) is signicantly associated with poor survival in pancreatic cancer patients.42 While we do not have specic follow-up information on the patients in this study, it is striking that the tumor types with the highest expression of calcyclin in our study, triple negative and Her2-overexpressing, are known to have a poor prognosis.

Conclusions
The work reported here represents the rst stage in the analysis of proteomic expression in human breast tumors utilizing MALDI MS and LCM-acquired cells from frozen tissue sections. Following spectral alignment and processing, biomarker candidates were identied by statistical analysis. We developed classiers for distinction of breast tumor versus normal mammary epithelium and ER+ versus ER- tumors using test sets and then successfully used these classiers in blinded test sets. Two of the m/z features, 10094 and 10842, were subsequently identied by LC-MS/MS as calcyclin (S100A6) and calgranulin A (S100A8) and conrmed by immuJournal of Proteome Research Vol. 7, No. 4, 2008 1505

breast cancer cell lines. In an immunohistochemical survey of S100 protein expression in 28 tissue types and 21 tumor types, Cross et al. found expression of S100A6 and S100A8 in 12% and 29% of breast cancers.37 Our ndings are consistent with those of Carlsson et al. who found down-regulation of S100A6 regardless of pathological stage and up-regulation of S100A8 in breast cancer using serial analysis of gene expression (SAGE).38 Kennedy et al. have shown that BRCA1 is capable of repressing several of the members of the S100A family including S100A8 and that functional BRCA1 is required for this repres-

research articles

Sanders et al.
Table 4. US Immunohistochemistry Staining Results for Tissue Microarrays
US Biomax TMA Breast Cancer Cases vs Normal Breast interpretable cores (%) invasive tumor (%) luminal epithelium

antigen

myoepithelium

Calcyclin+ CalcyclinTotal P-value Calgranulin+ CalgranulinTotal P-value

37 (77)

11 8 19 7 12 19

36 (75)

14 4 18 0.02a 5 12 17 0.51a

5 13 18 0.20a 0 0 0 0.06a

AccuMax TMA Breast Cancer Cases and Tumor Subsets interpretable cores (%) positive staining

antigen

ER+

ER-

Her2+

Her2-

Calcyclin+ CalcyclinTotal P-value Calgranulin+ CalgranulinTotal P-value

41 (91)

32 8 40 25 16 41

41 (91)

22 10 8 0 30 10 <0.0001b 16 9 14 2 30 11 0.007b

11 21 4 4 15 25 <0.14b 10 15 5 11 15 26 0.51b

a Values are calculated with respect to invasive tumor. b Hypothesis tests are computed based on the IHC scoring from 0 to 3+, but are collapsed to positive and negative staining for this table.

Figure 2. Calcyclin and calgranulin A immunohistochemistry. (A) Calcyclin (upper panel). The photomicrographs show representative examples of calyclin staining seen in tumor on the AccuMax tissue microarray. The staining intensity ranged from 3+ (1), 2+ (2), 1+ (3), to 0 (4). All photos were shot with a 22 ocular and 20 lens with an Olympus DP-70 digital camera. (B) Calgranulin A (lower panel). The photomicrographs show representative examples of calgranulin A staining seen in tumor on the AccuMax tissue microarray. The staining intensity ranged from 3+ (1), 2+ (2), 1+ (3), to 0 (4). All photos are shot with a 22 ocular and 20 lens with an Olympus DP-70 digital camera.

Supporting Information Available: (1) Detailed protein identication methods, (2) reproducibility analyses, (3) interinstitutional sample variability, (4) variability of samples with age, (5) complete list of features dened for the IMC vs NME comparison, (6) subset of features used for the IMC vs NME classier, (7) IMC vs NME classication LOOCV error depending on selection of k-value for the KNN algorithm, (8) complete list of features dened for the ER+ vs ER- comparison, (9) subset of features used for the ER+ vs ER- classier, (10) ER+ vs ER- classication LOOCV error depending on selection of k-value for the KNN algorithm, (11) detailed protein identication data. This material is available free of charge via the Internet at http://pubs.acs.org. References

nohistochemistry. Additional studies to characterize the function and biological role of these proteins in breast cancer will be undertaken. Proteomic expression in breast cancer was evaluated by MALDI MS of laser captured microdissected cells. Protein and peptide expression in cancer versus normal mammary epithelium and estrogen receptor-positive versus estrogen receptornegative tumors were compared. Biomarker candidates were identied by statistical analysis and classiers developed using a training set and validated in independent test sets. Several m/z features used in the classiers were identied by LC-MS/ MS and two were conrmed by immunohistochemistry.

Acknowledgment. Dr. Sanders is the recipient of a Komen Foundation Award. Dr. Dias is the recipient of an AVON-AACR Scholarship Award. This project was supported in part by The Vanderbilt Breast Cancer Specialized Program in Research Excellence (3P50 CA098131), NIH RO1 CA80195 (C.L.A), and Cancer Center Support Grant P30 CA68485.
1506 Journal of Proteome Research Vol. 7, No. 4, 2008

(1) Page, D. L. Special types of invasive breast cancer, with clinical implications. Am. J. Surg. Pathol. 2003, 27 (6), 8325. (2) Early Breast Cancer Tralists Collaborative Group (EBCTCG). Effects of chemotherapy and hormonal therapy for early breast cancer on recurrence and 15-year survival: an overview of the randomized trials. Lancet 2005, 365 (9472), 1687-717. (3) International breast Cancer Study (IBCSG). Endocrine responsiveness and tailoring adjuvant therapy for postmenopausal lymph node-negative breast cancer: a randomized trial. J. Natl. Cancer Inst. 2002, 94 (14), 105465. (4) Romond, E. H.; Perez, E. A.; Bryant, J.; et al. Trastuzumab plus adjuvant chemotherapy for operable HER2-positive breast cancer. N. Engl. J. Med. 2005, 353 (16), 167384. (5) Piccart-Gebhart, M. J.; Procter, M.; Leyland-Jones, B.; et al. Trastuzumab after adjuvant chemotherapy in HER2-positive breast cancer. N. Engl. J. Med. 2005, 353 (16), 165972. (6) Carlson, R. W.; Brown, E.; Burstein, H. J.; et al. NCCN task force report: adjuvant therapy for breast cancer. J. Natl. Compr. Cancer Network 2006, 4, S126. (7) Perou, C. M.; Sorlie, T.; Eisen, M. B.; et al. Molecular portraits of human breast tumors. Nature 2000, 406 (6797), 74752. (8) van de Vijver, M. J.; He, Y. D.; vant Veer, L. J.; et al. A geneexpression signature as a predictor of survival in breast cancer. N. Engl. J. Med. 2002, 347 (25), 19992009.

Biomarkers in Breast Cancer

(9) Wang, Y.; Klijn, J. G.; Zhang, Y.; et al. Gene-expression proles to predict distant metastasis of lymph-node-negative primary breast cancer. Lancet 2005, 365 (9460), 6719. (10) Paik, S.; Shak, S.; Tang, G.; et al. A multigene assay to predict recurrence of tamoxifen-treated, node-negative breast cancer. N. Engl. J. Med. 2004, 351 (27), 281726. (11) Bogaerts, J.; Cardoso, F.; Buyse, M.; et al. Gene signature evaluation as a prognostic tool: challenges in the design of the MINDACT trial. Nat. Clin. Pract. Oncol. 2006, 3 (10), 54051. (12) Mauriac, L.; Debled, M.; MacGrogan, G. When will more useful predictive factors be ready for use. Breast 2005, 14 (6), 61723. (13) Sparano, J. A. TAILORx: trial assigning individualized options for treatment (Rx). Clin. Breast Cancer 2006, 7 (4), 34750. (14) Caprioli, R. M.; Farmer, T. B.; Gile, J. Molecular imaging of biological samples: localization of peptides and proteins using MALDI-TOF MS. Anal. Chem. 1997, 69 (23), 475160. (15) Chaurand, P.; DaGue, B. B.; Pearsall, R. S.; Threadgill, D. W.; Caprioli, R. M. Proling proteins from azoxymethane-induced colon tumors at the molecular level by matrix-assisted laser desorption/ionization mass spectrometry. Proteomics 2001, 1 (10), 13206. (16) Chaurand, P.; Schwartz, S. A.; Caprioli, R. M. Imaging mass spectrometry: a new tool to investigate the spatial organization of peptides and proteins in mammalian tissue sections. Curr. Opin. Chem. Biol. 2002, 6 (5), 67681. (17) Stoeckli, M.; Chaurand, P.; Hallahan, D. E.; Caprioli, R. M. Imaging mass spectrometry: a new technology for the analysis of protein expression in mammalian tissues. Nat. Med. 2001, 7 (4), 4936. (18) Taguchi, F.; Solomon, B.; Gregorc, V.; et al. Mass spectrometry to classify non-small-cell lung cancer patients for clinical outcome after treatment with epidermal growth factor receptor tyrosine kinase inhibitors: a multicohort cross-institutional study. J. Natl. Cancer Inst. 2007, 99 (11), 83846. (19) Xu, B. J.; Caprioli, R. M.; Sanders, M. E.; Jensen, R. A. Direct analysis of laser capture microdissected cells by MALDI mass spectrometry. J. Am. Soc. Mass Spectrom. 2002, 13 (11), 12927. (20) Cazares, L. H.; Adam, B. L.; Ward, M. D.; et al. Normal, benign, preneoplastic, and malignant prostate cells have distinct protein expression proles resolved by surface enhanced laser desorption/ ionization mass spectrometry. Clin. Cancer Res. 2002, 8 (8), 2541 52. (21) Chalmers, M. J.; Gaskell, S. J. Advances in mass spectrometry for proteome analysis. Curr. Opin. Biotechnol. 2000, 11 (4), 38490. (22) Chaurand, P.; Stoeckli, M.; Caprioli, R. M. Direct proling of proteins in biological tissue sections by MALDI mass spectrometry. Anal. Chem. 1999, 71 (23), 526370. (23) Stoeckli, M.; Farmer, T. B.; Caprioli, R. M. Automated mass spectrometry imaging with a matrix-assisted laser desorption ionization time-of-ight instrument. J. Am. Soc. Mass Spectrom. 1999, 10 (1), 6771. (24) Todd, P. J.; Schaaff, T. G.; Chaurand, P.; Caprioli, R. M. Organic ion imaging of biological tissue with secondary ion mass spectrometry and matrix-assisted laser desorption/ionization. J. Mass Spectrom. 2001, 36 (4), 35569. (25) Amann, J. M.; Chaurand, P.; Gonzalez, A.; et al. Selective proling of proteins in lung cancer cells from ne-needle aspirates by matrix-assisted laser desorption ionization time-of-ight mass spectrometry. Clin. Cancer Res. 2006, 12 (17), 514250. (26) Chaurand, P.; Norris, J. L.,; Cornett, D. S.; Mobley, J. A.; Caprioli, R. M. New developments in proling and imaging of proteins from tissue sections by MALDI mass spectrometry. J. Proteome Res. 2006, 5 (11), 2889900. (27) Cornett, D. S.; Mobley, J. A.; Dias, E. C.; et al. A novel histologydirected strategy for MALDI-MS tissue proling that improves throughput and cellular specicity in human breast cancer. Mol. Cell. Proteomics 2006, 5 (10), 197583.

research articles
(28) Meistermann, H.; Norris, J. L.; Aerni, H. R.; et al. Biomarker discovery by imaging mass spectrometry: transthyretin is a biomarker for gentamicin-induced nephrotoxicity in rat. Mol. Cell. Proteomics 2006, 5 (10), 187686. (29) Roder, H.; Gringorieva, J.; Tsypin, M. The use of mass spectra for cancer biomarker detection. Available at: http://www.biodesix.com/Documents/MarkerWhitePaper.pdf, Biodesix, 2005. (30) Theodoritis, S.; Koutroumbas, K. Pattern Recognition, 3rd ed.; Academic Press: Amsterdam, Boston; 2006. (31) Webb, A. Statistical Pattern Recognition; Wiley: Chichester, U.K. ; 2002. (32) Yanagisawa, K.; Shyr, Y.; Xu, B. J.; et al. Proteomic patterns of tumor subsets in non-small-cell lung cancer. Lancet 2003, 362 (9382), 4339. (33) Zimmerman, L. J.; Wernke, G. R.; Caprioli, R. M.; Liebler, D. C. Identication of protein fragments as pattern features in MALDIMS analyses of serum. J. Proteome Res. 2005, 4 (5), 167280. (34) Anderson, L.; Seilhamer, J. A comparison of selected mRNA and protein abundances in human liver. Electrophoresis 1997, 18 (3 4), 5337. (35) Wilkins, M. R.; Sanchez, J. C.; Williams, K. L.; Hochstrasser, D. F. Current challenges and future applications for protein maps and post-translational vector maps in proteome projects. Electrophoresis 1996, 17 (5), 8308. (36) Sorlie, T.; Perou, C. M.; Tibshirani, R.; et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc. Natl. Acad. Sci. U.S.A. 2001, 98 (19), 1086974. (37) Cross, S. S.; Hamdy, F. C.; Deloulme, J. C.; Rehman, I. Expression of S100 proteins in normal human tissues and common cancers using tissue microarrays: S100A6, S100A8, S100A9 and S100A11 are all overexpressed in common cancers. Histopathology 2005, 46 (3), 25669. (38) Carlsson, H.; Petersson, S.; Enerback, C. Cluster analysis of S100 gene expression and genes correlating to psoriasin (S100A7) expression at different stages of breast cancer development. Int. J. Oncol. 2005, 27 (6), 147381. (39) Kennedy, R. D.; Gorski, J. J.; Quinn, J. E.; et al. BRCA1 and c-Myc associate to transcriptionally repress psoriasin, a DNA damageinducible gene. Cancer Res. 2005, 65 (22), 1026572. (40) Sorlie, T.; Tibshirani, R.; Parker, J.; et al. Repeated observation of breast tumor subtypes in independent gene expression data sets. Proc. Natl. Acad. Sci. U.S.A. 2003, 100 (14), 841823. (41) Ohuchida, K.; Mizumoto, K.; Ishikawa, N.; et al. The role of S100A6 in pancreatic cancer development and its clinical implication as a diagnostic marker and therapeutic target. Clin. Cancer Res. 2005, 11 (21), 778593. (42) Vimalachandran, D.; Greenhalf, W.; Thompson, C.; et al. High nuclear S100A6 (Calcyclin) is signicantly associated with poor survival in pancreatic cancer patients. Cancer Res. 2005, 65 (8), 321825. (43) Adam, B. L.; Qu, Y.; Davis, J. W.; et al. Serum protein ngerprinting coupled with a pattern-matching algorithm distinguishes prostate cancer from benign prostate hyperplasia and healthy men. Cancer Res. 2002, 62 (13), 360914. (44) Paweletz, C. P.; Trock, B.; Pennanen, M.; et al. Proteomic patterns of nipple aspirate uids obtained by SELDI-TOF: potential for new biomarkers to aid in the diagnosis of breast cancer. Dis. Markers 2001, 17 (4), 3017. (45) Petricoin, E. F.; Ardekani, A. M.; Hitt, B. A.; et al. Use of proteomic patterns in serum to identify ovarian cancer. Lancet 2002, 359 (9306), 5727. (46) Petricoin, E. F., III; Ornstein, D. K.; Paweletz, C. P.; et al. Serum proteomic patterns for detection of prostate cancer. J. Natl. Cancer Inst. 2002, 94 (20), 15768.

PR7008109

Journal of Proteome Research Vol. 7, No. 4, 2008 1507