label-free raman spectroscopy detects stromal adaptations ... · here, we have investigated lungs...

Integrated Systems and Technologies

Label-Free Raman Spectroscopy DetectsStromal Adaptations in Premetastatic LungsPrimed by Breast CancerSantosh Kumar Paidi1, Asif Rizwan2, Chao Zheng1,3, Menglin Cheng2,Kristine Glunde2,4, and Ishan Barman1,4

Abstract

Recent advances in animal modeling, imaging technology, andfunctional genomics have permitted precise molecular observa-tions of themetastatic process. However, a comprehensive under-standing of the premetastatic niche remains elusive, owing to thelimited tools that can map subtle differences in molecular med-iators in organ-specific microenvironments. Here, we report theability to detect premetastatic changes in the lung microenviron-ment, in response to primary breast tumors, using a combinationof metastatic mouse models, Raman spectroscopy, and multivar-iate analysis of consistent patterns in molecular expression. Weused tdTomato fluorescent protein expressing MDA-MB-231 andMCF-7 cells of high and low metastatic potential, respectively, togrow orthotopic xenografts in athymic nude mice and allow

spontaneous dissemination from the primary mammary fat padtumor. Label-free Raman spectroscopic mapping was used torecord the molecular content of premetastatic lungs. Thesemeasurements show reliable distinctions in vibrational features,characteristic of the collageneous stroma and its cross-linkers aswell as proteoglycans, which uniquely identify the metastaticpotential of the primary tumor by recapitulating the composi-tional changes in the lungs. Consistent with histological assess-ment and gene expression analysis, our study suggests thatremodeling of the extracellular matrix components may presentpromising markers for objective recognition of the premetastaticniche, independent of conventional clinical information.Cancer Res; 77(2); 247–56. �2016 AACR.

IntroductionWhile local breast cancers are largely responsive to current

therapeutic strategies, treatments to permanently eradicatemetas-tasis are yet to be developed. Consequently, nearly all breastcancer-related deaths today result from metastatic disease thatinvolves distant organs (1). The distribution of metastases is anon-random process with each tumor type manifesting a char-acteristic pattern of metastatic involvement in distant vital organs(2, 3). Stephen Paget's "Seed and Soil" hypothesis originally

shifted the attention from a sole focus on the behavior of primarytumor cells to the important role of the stroma at the secondarysite (4, 5). Seeking to understand the basis of metastasis organo-tropism, his seminal hypothesis postulated that a receptivemicro-environment at the secondary organ (soil) is crucial to theengraftment of circulating tumor cells (seed). This also provideda conceptual framework for later observations in experimentalmetastasis assays that cancer cells derived from a distant sitedisplay enhanced metastatic ability to that specific organ (6). Yet,it is only with recent advances in animal metastasis assays,genomic profiling, and real-time imaging techniques that themolecular components that drive organ-specific metastasis havebeen specifically probed. Translation of the preclinicalfindings onthemetastaticmicroenvironment into a clinical test, however, hasnot yet been realized.

Building on the seed and soil hypothesis, emerging evidencesuggests the formation of a premetastatic niche (7, 8), i.e., col-lective changes at the target metastasis sites prior to the arrival ofthe first tumor cells. This niche development in the preferredmetastatic sites appears to be driven by soluble growth factorssecreted by the primary tumor and recruitment of tumor-associ-ated cells (9). The priming of the secondary organs was initiallyattributed to the localization of hematopoietic bone marrowprogenitor cells expressing vascular epithelial growth factor recep-tor 1 (VEGFR-1) due to VEGFbeing secreted by the primary tumor(7). Exosomes secreted from primary tumors have also beenreported to play a significant role in mobilizing these progenitorcells to the premetastatic sites (10). The recruitment of tumor-associated cells provides an increased availability of chemokines,growth factors, matrix degrading factors, and adhesionmolecules

1Department of Mechanical Engineering, Johns Hopkins University, Baltimore,Maryland. 2Division of Cancer Imaging Research, The Russell H. Morgan Depart-ment of Radiology and Radiological Science, The Johns Hopkins UniversitySchool of Medicine, Baltimore, Maryland. 3Department of Breast Surgery, TheSecond Hospital of Shandong University, Jinan, Shandong, China. 4The SidneyKimmel Comprehensive Cancer Center, The Johns Hopkins University School ofMedicine, Baltimore, Maryland.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

S.K. Paidi, A. Rizwan, and C. Zheng contributed equally to this article.

Corresponding Authors: Ishan Barman, Johns Hopkins University, 3400 NorthCharles Street, Latrobe 103, Baltimore, MD21218. Phone: 410-516-0656; Fax: 410-516-4316; E-mail: [email protected]; and Kristine Glunde, Division of CancerImaging Research, The Russell H. Morgan Department of Radiology and Radio-logical Science, The Sidney Kimmel Comprehensive Cancer Center, The JohnsHopkins University School of Medicine, 720 Rutland Avenue, Traylor 203,Baltimore, MD 21205. Phone: 410-614-2705; Fax: 410-614-1948; E-mail:[email protected]

doi: 10.1158/0008-5472.CAN-16-1862

�2016 American Association for Cancer Research.

CancerResearch

www.aacrjournals.org 247

on June 11, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 15, 2016; DOI: 10.1158/0008-5472.CAN-16-1862

http://cancerres.aacrjournals.org/

that initiate the metastatic cascade (8, 9). This process is reportedto be accompanied by remodeling of the extracellular matrix(ECM) in the premetastatic niche, notably through the upregu-lated expression of matrix metalloproteinases (MMP; ref. 11),transformation of local fibroblasts, and focal expression of fibro-nectin. For instance, a recent series of investigations have revealedthat lysyl oxidase (LOX), an enzyme secreted by hypoxic tumorcells, modulates the ECM in premetastatic sites by cross-linkingcollagen fibrils, thereby making it more receptive to furthermyeloid cell infiltration (12, 13).

While promising, these findings also highlight the need forfurther research to reveal a holistic picture of the premetastaticstage that trigger (or inhibit) engraftment and proliferation. This,in turn, demands molecular-specific and quantitative analyticaltools that canprovide direct readouts frommultiple biomoleculeswithout necessitating individual labeling. Such a tool wouldinform if and how the compositional contributors of the stromalmicroenvironment in metastatic sites are changing in response toa spontaneously disseminating primary tumor—but prior to thearrival of tumor cells. Vibrational spectroscopy offers a promisingtool to meet these demands, owing to the wealth of intrinsicmolecular information (that obviates the need for imagingprobes), extensive multiplexing capability, and facile readout(14–17).

Spontaneous Raman spectroscopy, in particular, has emergedas an attractive technique for the diagnosis of cancers with highspecificity and free of interobserver variability (18). Based oninelastic scattering of light arising from the interactions with thetissue being analyzed, Raman spectroscopy affords subcellularsignal localization and can easily be extended to in vivoapproaches (19, 20). Recently, its ability to discern pathologiesin advance of their clinical manifestations has also been shown(21). Malins and colleagues elegantly demonstrated the earlydetection sensitivity of vibrational spectroscopy in a study, wherespectral changes in theDNAof primary tumorwere noted 57 daysprior to the appearance of histologic changes (22). We hypoth-esized that the utility of Raman spectroscopic information couldalso be extended to identifying the premetastatic niche, due to theunique structural and chemical changes associated with theevolving soil. Important clues also come from a recent report byKwak and colleagues, demonstrating the utility of infrared (IR)spectroscopic imaging in predicting cancer recurrence by exploit-ing molecular features of the tumor microenvironment (23), andour recent observation that lymph nodes in mice with metastatictumor xenografts displayed an increased collagen I density (24).Consistent with these recent literature reports, we suspected thatthe collagen architectural modifications, in part, preceded theseeding of metastatic cancer cells. Because Raman spectra reportvibrational features characteristic of collagen and its cross-linkingmoieties as well as glycoproteins, our goal in this study was toidentify Raman spectral patterns that are able to detect charac-teristic molecular changes in the premetastatic niche.

Here, we have investigated lungs from mouse models thatrecapitulate spontaneously disseminating breast cancer cells oflow and high metastatic potential and exploited the molecularbasis of Raman spectroscopy to probe the premetastatic niche(Fig. 1). Raman spectroscopic mapping measurements revealedsubtle, but consistent, changes in the vibrational features of ECMcomponents of the lungs, in particular in their collagen fibermatrix and proteoglycan content. The definition of the premeta-static adaptations in spectral terms facilitated the development of

a decision algorithm,which accurately differentiates lungs inmicewithmetastaticMDA-MB-231 tumor xenografts from that inmicewithMCF-7 xenografts and normal controls. A continuousmodelof ECM modifications, based on the metastatic potential of theprimary tumor, is proposed to explain the differential signa-tures—in the confirmed absence of any tumor cells in the lungs.This model is in agreement with observations from Masson'strichrome staining and gene expression analysis performed onmicroarray data of premetastatic lung samples frommice harbor-ing breast tumor xenografts. Taken together, this study demon-strates the potential of Raman spectroscopy as a rapid, objective,and label-free tool in the recognition of premetastatic changes.Weenvision that our findings here will also accelerate the use ofRaman spectroscopy in identifying distinct biochemical signa-tures in organ-specific niches, thereby enabling a better under-standing of organotropism.

Materials and MethodsTissue preparation and histopathology

Six-week-old female athymic nu/nu mice (NCI, MD) wereorthotopically inoculated with 2 � 106 cells of the human breastcancer cell lines MDA-MB-231 (n ¼ 3), or MCF-7 (n ¼ 3) in theirfourth right mammary fat pad, as detailed in our previous article(25). For comparison, control mice (n ¼ 3) without tumor cellimplantation were used in the study. Cell lines were obtainedfrom the ATCC and stably transfected with a construct containingcDNA of tdTomato as described in our previous report (24). Celllines tested negative formycoplasma and were authenticated usingshort tandem repeat (STR) profiling prior to inoculation in mice.Cell lines were maintained in RPMI 1640 (Sigma Aldrich) sup-plemented with 10% fetal bovine serum (Sigma Aldrich) and 1%penicillin–streptomycin (Sigma Aldrich) in a humidified incuba-tor at 37 �C/5% CO2. Prior to implantation of MCF-7 cells, micewere supplemented with 17b-Estradiol (Innovative Research ofAmerica, SE#121, 0.72 mg/pellet, 60-day release) in their neckregion (26). Primary tumor size was monitored, and mice weresacrificed within 8 to 12 weeks of cell implantation when primarytumors grew to approximately 500mm3 in volume. Control micewere also sacrificed in this timeframe. Freshly excised lungs ofmicewere cleaned in phosphate buffered saline (PBS) andfixed informalin for 24 hours. Formalin-fixed lung tissue samples wererinsed thoroughly in excess PBS to remove any residual formalinbefore acquiring Raman spectra. Following spectral acquisition,tissues were stored in 70% ethanol and sent to JHU HistologyServices for paraffin embedding and serial sectioning, after whichone of the sections was used for haematoxylin and eosin (H&E)staining. The unstained slides were utilized in our laboratory toperform Masson's trichrome staining for collagen as detailed inour previous study (24). The Institutional Animal Care and UseCommittee at the Johns Hopkins University School of Medicineapproved the protocol of this study.

Acquisition of Raman spectraFormalin-fixed lung specimens were rinsed in PBS, flattened,

and placed on a clean aluminum block. There was no interferenceof the tissue Raman spectrum from the aluminum substrate,which also ensured a consistent probe-tissue imaging distance.A custom-built portable, fiber-probe–based Raman spectroscopysystem was used for spectral acquisition (27). Briefly, an 830-nmdiode laser (500mWmaximumpower, Process Instruments) was

Paidi et al.

Cancer Res; 77(2) January 15, 2017 Cancer Research248




used to excite the sample. A lensed fiber-optic Raman bundledcontact probe (Emvision LLC) having a diameter of 2mm(and anestimated tissue sampling volume of 1 mm3) was used to deliverthe excitation beam through its central fiber and collect the back-scattered light through an annular ring of optical fibers. Thescattered light was directed to a spectrograph (Holospec f/1.8i,Kaiser Optical Systems). The spectra were then recorded using athermoelectrically cooledCCD camera (PIXIS 400BR, 20� 20mmpixels, 1,340�400array, Princeton Instruments). The laser powerat the lung tissue samples was maintained at around 15 mW inthis study and the tissue was kept moist throughout the period oflaser exposure by intermittent addition of PBS. A total collectiontime of 10 seconds (10 accumulations of 1 second each to preventCCD saturation) was used for acquisition of each spectrum.Spectroscopic mapping was pursued to overcome the limitationsof conventional fiber probe-based point spectroscopy that onlyexamines a small area of tissue and suffers from undersampling.Wide areamapping, over the entire lung specimen,was performedby scanning the optical probe using a pair of motorized transla-tion stages (travel range: 13 mm, T-LS13M, Zaber TechnologiesInc.) in each orthogonal direction. Zaber console (open-sourcesoftware) was used to control the raster scan through the PC serialports. The mapping protocol also ensured the collection ofsufficient spectra (approximately 300 spectra per mouse) for thedevelopment of robust classification models.

Data analysisThe Raman instrument was wavenumber-calibrated using 4-

acetamidophenol (Tylenol) spectra. Raman spectra recorded frommouse lungswere restricted to the fingerprint wavenumber region(500–1,850 cm�1 ) for analysis and normalized to lie between 0and1 inorder to remove the effects of potential differences in laserpower at the sample. Principal component analysis (PCA) wasused to reduce the dimensionality of the spectral dataset to a fewdimensions characteristic of themaximum variance in the dataset

(28). This transformation converts the set of spectral recordingsinto a set of values of linearly uncorrelated variables that form anorthogonal basis set. The spectral dataset of each mouse modelwas subjected to PCA using the statistical toolbox of MATLAB2015b (Mathworks) to obtain principal component (PC) scoresand loadings that highlight the spectral features characteristic ofthe class. The use of these key patterns (PCs) enhances sensitivityof the analysis by not focusing on small differences in Ramansignatures that may arise from natural variation or sampling.

To visualize the differences among the classes, radial visuali-zation maps were plotted using the Radviz tool of Orange datamining toolbox (29). Here, we utilized the scores of select PCsobtained from subjecting the entire spectral dataset to PCA.Guided by the Vizrank algorithm, the PCs were chosen to max-imize class separation. In the radial visualization plot, the scoresof a spectrum determine the position of the corresponding datapoint relative to the PC pivots. Partial least squares discriminantanalysis (PLS-DA), a supervised classification technique based onpartial least squares regression,was used to create decisionmodelsfrom the acquired Raman spectra for identifying the premetastaticniche (30). PLS-DA-derived classification models were built andtrained using a leave-m-out cross-validation approach that uti-lizes randomly chosen training data consisting of 60% of the dataof each class and test data constituted by the remaining 40% ofthe spectra. Randomized equalization of classes was implemen-ted prior to PLS-DA model development to avoid skewing themodel through disproportionate class sizes. Multiple iterationsof class equalization and splitting into testing and training sets(10 � 100) were performed to obtain average performancemetrics of the PLS-DA derived classification models.

Collagen quantification of Masson's trichrome stained tissueslides was achieved using FIJI (Image-J-based open-source soft-ware; ref. 31) and MATLAB (Mathworks). The color deconvolu-tion feature provided by FIJI was used to extract an 8-bit frame(dense collagen presence ¼ 0 and no collagen presence ¼ 256)

t = 0 t = 8–12 Weeks t > 15 Weeks

MDA-MB-231 XenograftRaman shift (cm–1)

Inte

nsi

ty (

a.u

.)

14

12

10

8

6

4

2

0

1,000 1,200 1,400 1,600 1,800

ControlMCF-7MDA-MB-231

A

B C

Figure 1.

Raman spectroscopic profiling of premetastaticlungs. A, Mouse models, orthotopicallyxenografted with human breast cancer cells ofdifferent metastatic potential (MCF-7 andMDA-MB-231), were used to study stromaladaptations in the lung prior to seeding of tumorcells. B, Representative in vivo brightfield (left)and fluorescence (right) images of mousegrowing a tdTomato-expressing breast tumorxenograft. C, Mean Raman spectra (with theshadow representing �1 standard deviation)acquired from lungs of normal mice, andpremetastatic lungs of MCF-7 and MDA-MB-231xenografted mice are shown.

Raman Spectroscopy Detects Premetastatic Adaptations

www.aacrjournals.org Cancer Res; 77(2) January 15, 2017 249




corresponding to the color, indicative of collagen content in thetrichrome stains. The color was defined by average RGB values ofpixels in a small user-selected region of interest (ROI) chosen inthe image. Using in-house MATLAB code, the intensity of thepixels was converted to obtain a measure of collagen density ineach frame. The data were averaged over the entire lung tissuesection with n >35 fields of view (FOV) per class, where each FOVwas approximately 1.75mm�1.33mm. Statistical significance ofdifferences across the classes was evaluated using the Student ttest. A conventional criterion of P value less than 0.05 was used toconsider differences as statistically significant.

Microarray datasetThe gene expression microarray dataset GSE62817 from the

Gene Expression Omnibus (GEO) of the National Center forBiotechnology Information (http://www.ncbi.nlm.nih.gov/geo/)was used in this study (32). This dataset contains gene expressiondata frompremetastatic lungs of BALB/cmice injectedwith tumorcells into their fourth mammary fat pad. In particular, 67NR(nonmetastatic) and 4T1 (metastatic) breast carcinoma cell lineswere used and lung tissue was collected when the tumors reacheda volume of 50 mm3. Control mice with no tumor cell injectionswere utilized for comparison. Briefly, RNA was extracted using aQiagen kit, andAffymetrixmicroarrays (Mouse 430-v2)were usedto analyze the expression profile of tissue samples. The heat mapwas generated using Gene-e matrix visualization and analysissoftware (http://www.broadinstitute.org). We used the moderat-ed F-test statistic for selecting relevant genes. Consistent with thenumber of different groups and number of samples per group inthe dataset, a threshold F-test statistic of 2.53 (correspondingto a ¼ 0.125 level of significance) was used.

Results and DiscussionLung was selected as the target organ in the current pilot study,

as it offers a favorable site for spontaneous disseminationof breastcancer and is the most commonly studied metastatic site inanimal models (9, 33). Primary orthotopic MDA-MB-231, and

also eventually MCF-7, breast tumor xenografts used in our studypreferentially metastasize to the lungs (34, 35). Spectroscopicmapping of the lungs, as opposed to a limited number of discretepointmeasurements, was pursued to encompass a large FOVwithhigh spectral contrast. This would also account spectroscopicallyfor the intrinsic biological variation in lung tissue that couldotherwise suppress the subtle differences expected from premeta-static adaptations. Figure 1C shows average Raman spectrarecorded from lung samples of control mice (control) as well asmice bearing MCF-7 (MCL) and MDA-MB-231 (MDL) tumorxenografts. The spectra shown here were background subtractedfor the tissue autofluorescence component. While gross visualinspection reveals limited spectral variations, we reason that asubset of pixels (representing specific molecular moieties) haspredictive power that is lost in examining the average value of thespectra across the lung specimen. In an effort to focus on eluci-dating the differentiating biochemical characteristics, we usedPCA. To preserve the subtle spectral features, we performed PCAon the normalized spectra recorded from the specimen withoutbackground subtraction. For comparison, the results obtainedfollowing fifth-order best-fit polynomial based autofluorescencebackground removal have also been provided alongside (and inSupplementary Information).

Consistent differences in Raman spectra reflect biochemicalchanges in premetastatic lungs

Figure 2 shows the first 7 PC loadings in order of spectralvariance for each of the three classes, control, MCL, andMDL. Thefirst few PCs in each class are evidently influenced by the broadtissue autofluorescence signal; the characteristic Raman featuresare more prevalent in PCs 4 through 7. The PCs derived from thespectra belonging to the lungs of control mice exhibit notableRaman features at 859 cm�1 (C–C stretch of proline in collagen),1,003 cm�1 (C–C stretching vibration of the aromatic ring in thephenylalanine side chain), 1,442 cm�1 (CH2 deformations inlipids), 1,592 cm�1 (tentatively attributed to carbon particles)and 1,653 cm�1 (amide-I feature of proteins with potentialcontributions of C¼C stretching in lipids) with a weaker peak at

10

8

6

4

2

0

10

8

6

4

2

0

10

8

6

4

2

0800 1,000 1,200 1,400 1,600 1,800 800 1,000 1,200 1,400 1,600 1,800 800 1,000 1,200 1,400 1,600 1,800

Raman shift (cm–1) Raman shift (cm–1) Raman shift (cm–1)

Inte

nsi

ty (

a.u

.)

Inte

nsi

ty (

a.u

.)

Inte

nsi

ty (

a.u

.)

856

859

859

917

1001

1061

917

1061

1003

1304

1442

1592

1653

PC7

PC6

PC5

PC4

PC3

PC2

PC1

PC7

PC6

PC5

PC4

PC3

PC2

PC1

PC7

PC6

PC5

PC4

PC3

PC2

PC1

A B C

Figure 2.

PCA of the acquired Raman spectra. A, PC loadings derived from spectra of lungs from control mice, i.e., bearing no tumor xenograft. B, PC loadings derivedfrom spectra of lungs belonging to mice bearing MCF-7 xenografts (labeled as MCL in the text). C, PC loadings derived from spectra of lungs belonging tomice with MDA-MB-231 xenografts (labeled as MDL in the text). Dotted and dot–dashed lines highlight collagen and proteoglycan features, respectively.

Paidi et al.





1,304 cm�1 (in-plane CH2 twisting modes of lipids). Thesefeatures are concordant with prior observations in the literature(36–38). Table ST1 (Supplementary Information) lists the prom-inent peaks observed in the PCs and their characteristic bandassignments.

Visual inspection of the PC loadings shows an enhancement ofthe 859 cm�1 peak, which can be attributed to collagen, for theMDL specimen in comparison with MCL and control as well as anew peak at 917 cm�1 (C–C stretch of proline ring; ref. 17) for thenon-control samples. These spectral differences suggest a positivecorrelation of collagen density in the lung specimens with thepresence of a primary tumor xenograft and, importantly, with itsmetastatic potential. Previous studies have discussed the role ofcollagen in the premetastatic niche and have shown evidence ofcollagen cross-linking and the creation of a metastatic growthpermissive fibrotic microenvironment at secondary sites, whichwas mediated by LOX secreted by hypoxic tumors (39, 40).Inhibition of LOX synthesis in human breast cancer cells hasbeen shown to reduce the accumulation of CD11bþmyeloid cellsin premetastatic organs of mice with orthotopic tumors andprevent metastasis (12). Another pertinent peak was observed atapproximately 1,061 cm�1 in the MCL and MDL PCs, which isknown to be a key spectralmarker for proteoglycans (41, 42). Thisfinding offers an intriguing insight into the nature of molecularmodifications in the premetastatic niche, particularly in light ofthe study of Gao and colleagues. This study demonstrated thatmyeloid cells in premetastatic lungs (recruited by primary tumorderived secretory factors) aberrantly expressed versican, an ECMproteoglycan (43). Versican stimulated mesenchymal-to-epithe-lial transition of metastatic tumor cells by reducing phospho-Smad2 levels, which led to elevated cell proliferation and accel-

erated metastases. In fact, lung metastasis in mouse models wasfound to be significantly impaired through knockdown of versi-can, reinforcing the importance of proteoglycan content as apremetastatic site marker. Furthermore, the gradual increase inthe prominence of proteoglycan marker in PCs with increasingmetastatic potential is in agreement with the seminal report ofKaplan and colleagues, which showed that recruitment of bonemarrow–derived cells is correlated to the aggressiveness of theprimary tumor (7). On the other hand, a significant suppressionof the peaks at approximately 1,302 cm�1 and 1,442 cm�1 wasnoted with a smaller reduction in the intensity of the 1,653 cm�1

feature. Because the former two peaks are characteristic of lipidsand the latter also has lipid contributions, one can reasonablyinfer a relative reduction in the lipid content corresponding tospectra from lungs of mice bearing primary tumor xenografts.

Given the large dimensionality of the spectral data, however, itis challenging to judge whether the differences across the classesare significant from visual inspection of the PC loadings alone. Toobserve these differences better, we used radial visualization plotsthatmap the scores ofmultiple PCs onto a two-dimensional spacefor the purpose of clustering. Figure 3 shows a representativeradial visualization plot constructed by using PCs derived from arandomized spectral selection with 300 points per class (control,MCL, and MDL). These were chosen from the total set consistingof approximately 900 spectra/class, which in turn were constitut-ed by approximately 300 spectra acquired from spatially distinctpoints in the lung lobes of each mouse. Supplementary Fig. S1shows the corresponding radial visualization map after subtrac-tion of tissue autofluorescence background. In order to obtaininformative projections of the class-labeled data, the VizRankalgorithm was used to grade the PCs by their ability to visually

PC6

PC1

PC2

PC4

PC7

ControlMCF-7MDA-MB-231

Figure 3.

Visualization of spectroscopic differences due topremetastatic adaptations. Radial visualization plotshowing clusters formed by spectra recorded from lungsamples of sacrificed mice bearing MDA-MB-231 andMCF-7 breast cancer xenografts as well as controlswithout xenografts.






discriminate between classes (44). Evidently, there are pro-nounced differences in the Raman spectra acquired from lungspecimens of control, MCL, and MDLmouse models, most likelyowing to differential priming through factors secreted by theprimary tumor. The presence of a small overlap of clusters fromcontrol and MDL mice indicates a limited development of thepremetastatic niche in some of the latter cases and requires furtheranalysis, as detailed in the ensuing paragraphs. While the PCscore-based plot offers a satisfactory tool for preliminary dataexploration, it does not provide quantitative information aboutthe potential of Raman spectroscopy in recognizing the class(metastatic potential) and in understanding how the lung(s) ofan individual mouse responds to the primary tumor xenograft.

Thus, we used partial least squares–discriminant analysis (PLS-DA)-based classification models for translating the spectroscopicmeasurements in the premetastatic lungs to identification of thetype of primary tumor xenograft. We used an equal number ofspectra belonging to each class (control,MCL, andMDL) and theirclass labels to train the classification algorithm. To ensure robust-ness, we evaluated the classifier by testing on a separate validationdataset as detailed in the Data analysis section. Average correctrates of prediction of 90.1%, 97.7%, and 78.4% (95.4%, 95.6%,and 75.1% after autofluorescence background subtraction) wereobtained for the spectra belonging to control, MCL, and MDL,respectively. The relevant confusion matrix of the reference andpredicted labels is shown in Supplementary Table ST2. The lowercorrect classification rate for MDL spectra in both the cases is inagreement with the overlap of the MDL and control clustersobserved on the radial visualization plot in Fig. 3.

In order to understand the root cause of the MDL spectramisclassifications, we repeated the former analysis by leaving onemouse out of the dataset each time (Table 1 and SupplementaryTable ST3 after autofluorescence background subtraction).Removing mouse MD #3 (arbitrary numbering of mice used fortabulating results) yields near-perfect classification accuracy indi-cating significantly lesser premetastatic adaptations in the lungs ofthis animal. Furthermore, removingmouseMD#3 also improvedthe classification rate of spectra belonging to control mice due toenhanced contrast in the training data. Notably, removal of anyother mouse from the classification protocol did not result in assignificant a change in the accuracy levels. This reinforces the factthat the improvement observed on removal ofmouseMD#3 datawas not due to overtraining of the model on smaller numbers, asotherwise similar enhancements would have been noted in all theother cases. The inadequate priming of the MD #3 lungs is also

supported by application of Chauvenet's criterion to the set ofclassification rates obtained for the MDL class (Table 1). Thelatter results in designation of MD #3 as the sole outlier in thegroup due to its significant deviation from the mean by morethan the maximum allowable number of standard deviations(tmax ¼ 1.96 for a sample size of n ¼ 10). Application ofChauvenet's criterion also facilitates determination of individ-ual sample eligibility for training the PLS-DA classifier. Thespectroscopic measurements, thus, capture the inherent vari-ability in metastasis, which is commonly regarded as an inef-ficient process that only a subset of tumor cells can successfullynavigate (45, 46) and is known to exhibit sporadic occurrenceacross a cohort of animals.

Finally, we conducted a negative control study to verify thatthe predictive power of the developed algorithms was notdriven by potential spurious correlations in the spectral dataset(47). For this validation study, we assigned random class labelsto the spectra irrespective of their true class origins and used thePLS-DA-derived classification models after similar splitting of thedata into training and test sets. This resulted in an average correctclassification rate of 33.3% with a standard deviation of 1.4%(and 33.6% with a standard deviation of 1.4% after backgroundsubtraction) for 1,000 iterations. The significantly low rate ofcorrect classification (consistent with the likelihood of randomselection of the true class label, 1/3) underscores the absence ofchance correlations in the developed model.

Histologic assessment of the premetastatic niche in mice lungsDue to their high metastatic potential and preference for

metastasis to lungs, orthotopic MDA-MB-231 xenografts are fre-quently used to replicate breast cancer metastasis and organo-tropism (33, 48). Aggressive subpopulations ofMDA-MB-231 areoften derived through multiple rounds of in vivo selection andreimplantation and have been recently reported to result inmacro-metastases to the lungs in 100% of all tested mice (35).In our study, we observed no cancer cell seeding in lungs of micebearing MDA-MB-231 tumor xenografts (time of sacrifice: 8–12weeks post orthotopic tumor inoculation). Prior optical trackingstudies by Winnard and colleagues showed that orthotopicallyimplantedMDA-MB-231 cells reached lungs only after�15weeksof implantation in SCID mice (34). They also observed theabsence of distant metastases after 8 weeks, consistent with thetime period of sacrifice in our study. MCF-7 cells, often classifiedas nonmetastatic (49), were likewise not expected to engraft inthe lungs within this 8- to 12-week time frame. However, it is

Table 1. Correct classification rates (%) of the PLS-DA–derived model using leave-one-mouse-out protocol

Correct classification rate (%)Chauvenet's criterion for MDL

(n ¼ 10; tmax ¼ 1.96)Mouse excluded Control MCL MDL t ¼ |xi � xmean|/s Result

None 90.1 97.7 78.4 0.26 RetainMD #1 81.2 97.0 75.7 0.61 RetainMD #2 83.8 96.8 76.2 0.54 RetainMD #3 100.0 98.6 99.4 2.54 EliminateMC #1 88.8 98.0 78.5 0.24 RetainMC #2 89.6 97.3 77.6 0.36 RetainMC #3 88.8 98.3 77.3 0.40 RetainControl #1 87.2 96.4 73.1 0.96 RetainControl #2 92.9 97.3 80.2 0.01 RetainControl #3 92.9 97.4 86.4 0.81 Retain

NOTE: MD and MC refer to mouse models with MDA-MB-231 and MCF-7 tumor xenografts, respectively.

Paidi et al.





noteworthy that MCF-7 cells are known to eventually metastasizeto lungs in immunodeficient mice such as NSG (35).

Here, the lung tissue sections from each mouse were H&Estained to check for the onset (or the lack thereof) of metastasis.Also, to histologically examine the differences in collagen contentacross the classes, serial sections were processed with Masson'strichrome stain. Figure 4 shows representative images of H&E andMasson's trichrome stained lung sections belonging to each class(control, MCL, and MDL). The H&E images corroborate the lackof any metastatic lesions in the lung specimens. The Masson'strichrome-stained sections were used for quantification of themean collagen density for each class (detailed in Materials andMethods). Figure 5A shows the mean bar plot that highlightsthe differences in collagen density for control mice and micebearingMCF-7 andMDA-MB-231 xenografts.We observe that the

metastatic potential of the primary tumor is positively correlatedwith the collagen density in the premetastatic niche. Yet, thedifferences in the mean collagen density values between MCLand MDL samples do not reach statistical significance (P < 0.05).Based on our spectroscopic findings, we suspected that the lungspecimens of mouse MD #3 may possibly skew the collagendensity values of the MDL set. Accordingly, we recalculated thevalues by removing the images of the lungs of this mouse, asshown in Fig. 5B. With this modification, the differences amongeach pair of classes were found to be statistically significant. Thisimprovement of contrast in collagen density corresponds wellwith our spectroscopic findings and reflects the biochemicalsensitivity of the vibrational spectroscopic data.

In light of the spectroscopic identification of stromal adapta-tions, we further sought to investigate the genetic underpinnings

A

D

G H I

E F

B C

Figure 4.

Histologic assessment ofpremetastatic lungs shows stromalchanges. Top (A–C) and middle (D–F)panels display representativemicroscopic images of H&E- andMasson's trichrome–stained slides at�5 and �10 magnifications,respectively. The H&E-stainedsections confirm the absence of tumorcell seeding in the lungs of controls.Masson's trichrome stain delineatescollagen fibers in the extracellularmatrix and is quantified through imageprocessing, as shown inG–I. Left (A, D,and G), lung sections derived fromcontrol mice; middle (B, E, and H) andright (C, F, and I), lung sections frommice bearing MCF-7 (nonmetastatic)and MDA-MB-231 (metastatic) tumorxenografts, respectively. The scalebars in the top and middle panelsrepresent 1,000 and 500 mm,respectively.

25

20

15

10

5

0

25

20

15

10

5

0

Co

llag

en d

ensi

ty (

a.u

.)

Co

llag

en d

ensi

ty (

a.u

.)

Control MCF-7 MDA-MB-231 Control MCF-7 MDA-MB-2310.034 0.0340.123 0.0067

0.0014 < 0.0001

A B

Figure 5.

Quantification of collagen fiberdensity in premetastatic lungs. A, Barplot showing mean and standarddeviation of collagen density acrossthe three classes (with all miceincluded) alongwith pairwise Studentt test P values. B, Bar plot showingmean and standard deviation ofcollagen content across the threeclasses (after exclusion of MDA-MB-231 xenograft bearing mousedisplaying atypical Raman data)along with pairwise Student t testP values.






of premetastatic priming of lungs. We performed gene expres-sion analysis on publicly available microarray data (GSE62817) to determine markers in premetastatic lungs inresponse to primary breast tumors of divergent metastaticpotential (32). Specifically, the data included gene expressionlevels corresponding to the lungs of normal mice (n¼ 5) as wellas premetastatic lungs of mice injected with nonmetastatic67NR breast carcinoma cells (n ¼ 5) and with metastatic4T1 breast carcinoma cells (n ¼ 4). Seeking to isolate genesrelevant to our study, we restricted our search to genes encodingfor key stromal constituents and significantly overexpressed inpremetastatic lungs of 4T1 tumor–bearing mice. Figure 6 showsthe heatmap representing expression levels of these genes alongwith corresponding moderated F-statistic. Premetastatic lungsof the 4T1 tumor bearing mice demonstrate a selective upre-gulation of genes related to ECM constituents, notably colla-gen, fibronectin, versican, and glypican. Importantly, each ofthese ECM components exhibits a decreasing gradient of valuesfrom 4T1 to 67NR and then to control cases. The differentialexpression of stromal genes in response to primary tumordevelopment can, thus, help explain our observations of dis-cernible biochemical alterations in premetastatic lungs of micebearing MCF-7 xenografts, even though these cells rarely metas-tasize in the mouse model used.

Taken together, our findings suggest that remodeling of theECM, such as an increase in collagen and proteoglycan content,occurs in response to primary tumor–derived factors, whichprecedes the actual seeding of tumor cells at the distantmetastaticsite. The data in this study support a continuous premetastaticniche formation model from primary tumors with low and highmetastatic potential, rather than discrete premetastatic adapta-tions that are representative of the highly metastatic model alone.This would also imply that premetastatic adaptations are a nec-essary condition for further progression but not predictive of theeventual success of metastases.

In conclusion, the current study proposes Raman spectros-copy as a label-free molecular-specific tool for detection ofpremetastatic adaptations in the stromal environment. Usingbreast cancer metastasis to the lungs as the paradigm, we havedemonstrated that Raman spectroscopy accurately detectschanges in the ECM of premetastatic lungs, which correlatewith the metastatic potential of the respective primary tumorxenograft. We identified spectral markers corresponding tocollagen and proteoglycan that offer molecular insights intothe formation of the premetastatic niche while also facilitatingobjective detection. The data presented here are unique andcomplementary to other microenvironment profiling methodssuch as genomic assays and mass spectrometry. While breastcancer metastasis to the lungs has been chosen for the currentstudy, it should be noted that this approach can be extendedto study the development of premetastatic niches at anysecondary target organ from primary breast and non-breastmalignancies.

We envision that the use of Raman spectroscopic imaging inconjunction with further biochemical assays will offer detailedmechanistic insights into premetastatic niche formation andevolution. As such, this offers a unique research tool that com-bines microenvironment and cellular profiling through nonper-turbative, multiplexed measurements of proteins, nucleic acids,lipids, and metabolites. Building on the ability to detect suchsubtle changes in tissue composition, and as discussed in recentreports (18, 23), we anticipate that Raman spectroscopic imagingcan, with further refinement, facilitate surgical margin assessmentin tissue conserving surgery and provide prediction of tumorrecurrence. Integration of Raman spectroscopy with minimallyinvasive biopsy needles can also permit real-time, in situ detectionof malignancies (19, 50).

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Premetastatic lung tissue in mouse harboring -

No tumor (control)67NR Breast 4T1 Breastcancer allograft cancer allograft

Min Max

Symbol

Col15a1

Col5a1

Col4a3

Fn1

Vcan

Gpc4

Description ID

Collagen, type XV, alpha 1

Collagen, type V, alpha 1

Collagen, type IV, alpha 3

Fibronectin 1

Versican

Glypican 4

1448755_at

1416741_at

1438779_at

1426642_at

1421694_a_at

1443620_at

Mod.F-Value

3.88

3.83

2.55

3.31

3.61

4.25

Figure 6.

Gene expression changes in premetastatic lungs as a function of metastatic potential of primary tumor. Microarray gene expression data heat map wasobtained by analyzing the publicly available dataset GSE62816 on the Gene-e data visualization and analysis platform. The sample cohort includes lungs of micebearing breast tumor xenografts of different metastatic potential. Total RNA was isolated from the premetastatic lungs and hybridized on an AffymetrixMouse Genome 430 2.0 Array. Genes that are relevant to spectral markers identified in the current study and overexpressed in response to the metastaticpotential of the primary tumor were analyzed. Moderated F-value of 2.53 was set as the criterion for inclusion.

Paidi et al.





Authors' ContributionsConception and design: S.K. Paidi, A. Rizwan, K. Glunde, I. BarmanDevelopment of methodology: A. Rizwan, I. BarmanAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): S.K. Paidi, A. Rizwan, C. Zheng, M. ChengAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): S.K. Paidi, A. Rizwan, C. Zheng, K. GlundeWriting, review, and/or revision of the manuscript: S.K. Paidi, A. Rizwan,K. Glunde, I. BarmanAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): M. ChengStudy supervision: K. Glunde, I. Barman

Grant SupportS.K. Paidi and I. Barman acknowledge the JHU Whiting School of Engineer-

ing Startup Funding. C. Zheng acknowledges the support of the NationalConstruction of High Quality University Projects of Graduates from the ChinaScholarship Council (CSC; Grant No. 201406170141). A. Rizwan, M. Cheng,and K. Glunde acknowledge the support of NIH R01 CA154725.

The costs of publication of this article were defrayed in part by the paymentof page charges. This article must therefore be hereby marked advertisementin accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Received July 11, 2016; revisedOctober 13, 2016; acceptedOctober 30, 2016;published OnlineFirst November 15, 2016.

References1. Sporn MB. The war on cancer. The Lancet 1996;347:1377–81.2. Hess KR, Varadhachary GR, Taylor SH, Wei W, Raber MN, Lenzi R, et al.

Metastatic patterns in adenocarcinoma. Cancer 2006;106:1624–33.3. Lee YNM. Breast carcinoma: Pattern of metastasis at autopsy. J Surg Oncol

1983;23:175–80.4. Paget S. The distribution of secondary growths in cancer of the breast. The

Lancet 1889;133:571–3.5. Fidler IJ, Kripke ML. Metastasis results from preexisting variant cells within

a malignant tumor. Science 1977;197:893–5.6. Fidler I. Selection of successive tumour lines for metastasis. Nature

1973;242:148–9.7. Kaplan RN, Riba RD, Zacharoulis S, Bramley AH, Vincent L, Costa C, et al.

VEGFR1-positive haematopoietic bone marrow progenitors initiate thepre-metastatic niche. Nature 2005;438:820–7.

8. Hiratsuka S, Watanabe A, Aburatani H, Maru Y. Tumour-mediated upre-gulation of chemoattractants and recruitment of myeloid cells predeter-mines lung metastasis. Nat Cell Biol 2006;8:1369–75.

9. Psaila B, Lyden D. The metastatic niche: Adapting the foreign soil. Nat RevCancer 2009;9:285–93.

10. Costa-Silva B, Aiello NM, Ocean AJ, Singh S, Zhang H, Thakur BK, et al.Pancreatic cancer exosomes initiate pre-metastatic niche formation in theliver. Nat Cell Biol 2015;17:816–26.

11. Hiratsuka S, Nakamura K, Iwai S, Murakami M, Itoh T, Kijima H, et al.MMP9 induction by vascular endothelial growth factor receptor-1 isinvolved in lung-specific metastasis. Cancer cell 2002;2:289–300.

12. Erler JT, Bennewith KL, Cox TR, Lang G, Bird D, Koong A, et al. Hypoxia-induced lysyl oxidase is a criticalmediator of bonemarrow cell recruitmentto form the premetastatic niche. Cancer Cell; 15:35–44.

13. Erler JT, Bennewith KL, NicolauM,Dornh€ofer N, KongC, LeQT, et al. Lysyloxidase is essential for hypoxia-induced metastasis. Nature 2006;440:1222–6.

14. Matousek P, Stone N. Development of deep subsurface Raman spectros-copy for medical diagnosis and disease monitoring. Chem Soc Rev2016;45:1794–1802.

15. Fernandez DC, Bhargava R, Hewitt SM, Levin IW. Infrared spectroscopicimaging for histopathologic recognition. Nat Biotechnol 2005;23:469–74.

16. Walsh MJ, Reddy RK, Bhargava R. Label-free biomedical imaging withmid-IR spectroscopy. IEEE J Sel Top Quantum Electron 2012;18:1502–13.

17. Stone N, Kendall C, Shepherd N, Crow P, Barr H. Nearinfrared ramanspectroscopy for the classification of epithelial precancers and cancers.J Raman Spectrosc 2002;33:564–73.

18. Kong K, Rowlands CJ, Varma S, PerkinsW, Leach IH, Koloydenko AA, et al.Diagnosis of tumors during tissue-conserving surgery with integratedautofluorescence and raman scattering microscopy. Proc Natl Acad SciU S A 2013;110:15189–94.

19. Jermyn M, Mok K, Mercier J, Desroches J, Pichette J, Saint-Arnaud K, et al.Intraoperative brain cancer detection with raman spectroscopy in humans.Sci Transl Med 2015;7:274ra19.

20. Matth€aus C, Krafft C, Dietzek B, Brehm BR, Lorkowski S, Popp J. Nonin-vasive imaging of intracellular lipidmetabolism inmacrophages by ramanmicroscopy in combination with stable isotopic labeling. Anal Chem2012;84:8549–56.

21. Pandey R, Paidi SK, Kang JW, Spegazzini N, Dasari RR, Valdez TA, et al.Discerning the differential molecular pathology of proliferative middle earlesions using raman spectroscopy. Scientific reports 2015;5:13305.

22. MalinsDC,AndersonKM,GilmanNK,GreenVM,Barker EA,Hellstr€omKE.Development of a cancer DNA phenotype prior to tumor formation. ProcNat Aca Sci U S A 2004;101:10721–5.

23. Kwak JT, Kajdacsy-Balla A, Macias V, Walsh M, Sinha S, Bhargava R.Improving prediction of prostate cancer recurrence using chemical imag-ing. Scientific Reps 2015;5:8758.

24. Rizwan A, Bulte C, Kalaichelvan A, ChengM, Krishnamachary B, BhujwallaZM, et al. Metastatic breast cancer cells in lymph nodes increase nodalcollagen density. Scientific Reps 2015;5:10002.

25. Rizwan A, Serganova I, Khanin R, Karabeber H, Ni X, Thakur S, et al.Relationships between LDH-A, lactate, and metastases in 4T1 breasttumors. Clin Cancer Res 2013;19:5158–69.

26. Yarden RI, Lauber AH, El-Ashry D, Chrysogelos SA. Bimodal regulation ofepidermal growth factor receptor by estrogen in breast cancer cells. Endo-crinology 1996;137:2739–47.

27. Paidi SK, Siddhanta S, Strouse R, McGivney JB, Larkin C, Barman I. Rapididentification of biotherapeutics with label-free raman spectroscopy. AnalChem 2016;88:4361–8.

28. Jolliffe I. Principal component analysis. Wiley Online Library; 2002.29. Dem�sar J, Curk T, Erjavec A, Gorup C, Hocevar T, Milutinovic M, et al.

Orange: Data mining toolbox in python. J Mac Lear Res 2013;14:2349–53.

30. Brereton RG. Applied chemometrics for scientists. John Wiley & Sons;2007.

31. Schindelin J, Arganda-Carreras I, Frise E, Kaynig V, Longair M, Pietzsch T,et al. Fiji: An open-source platform for biological-image analysis. NatMeth2012;9:676–82.

32. Kowanetz M, Wu X, Lee J, Tan M, Hagenbeek T, Qu X, et al. Granulocyte-colony stimulating factor promotes lung metastasis through mobilizationof Ly6GþLy6Cþ granulocytes. Proc Natl Acad Sci U S A 2010;107:21248–55.

33. Minn AJ, Gupta GP, Siegel PM, Bos PD, Shu W, Giri DD, et al. Genes thatmediate breast cancer metastasis to lung. Nature 2005;436:518–24.

34. Winnard PT, Kluth JB, Raman V. Noninvasive optical tracking of redfluorescent protein-expressing cancer cells in a model of metastatic breastcancer. Neoplasia 2006;8:796–806.

35. Iorns E, Drews-Elger K, Ward TM, Dean S, Clarke J, Berry D, et al. A newmouse model for the study of human breast cancer metastasis. PLoS One2012;7:e47995.

36. Huang N, Short M, Zhao J, Wang H, Lui H, Korbelik M, et al. Full rangecharacterization of the Raman spectra of organs in amurinemodel. Opt Ex2011;19:22892–909.

37. Frushour BG, Koenig JL. Raman scattering of collagen, gelatin, and elastin.Biopolymers 1975;14:379–91.

38. Czamara K, Majzner K, Pacia M, Kochan K, Kaczor A, Baranska M. Ramanspectroscopy of lipids: A review. J Raman Spectrosc 2015;46:4–20.

39. Joo YN, Jin H, Eun SY, Park SW, Chang KC, Kim HJ. P2Y2R activation bynucleotides released from the highly metastatic breast cancer cell MDA-MB-231 contributes to pre-metastatic niche formation by mediating lysyloxidase secretion, collagen crosslinking, andmonocyte recruitment. Onco-target 2014;5:9322–34.






40. Cox TR, BirdD, Baker AM, BarkerHE,HoMW, LangG, et al. LOX-mediatedcollagen crosslinking is responsible for fibrosis-enhanced metastasis.Cancer Res 2013;73:1721–32.

41. Kunstar A, Leijten J, van Leuveren S,Hilderink J,OttoC, vanBlitterswijkCA,et al. Recognizing different tissues in human fetal femur cartilage by label-free raman microspectroscopy. J Biomed Opt 2012;17:116012.

42. Gamsjaeger S, Klaushofer K, Paschalis EP. Raman analysis of proteoglycanssimultaneously in bone and cartilage. J Raman Spectrosc 2014;45:794–800.

43. GaoD, JoshiN, ChoiH, Ryu S,HahnM,Catena R, et al.Myeloid progenitorcells in the premetastatic lung promote metastases by inducing mesen-chymal to epithelial transition. Cancer Res 2012;72:1384–94.

44. Leban G, Zupan B, Vidmar G, Bratko I. Vizrank: Data visualization guidedby machine learning. Data Min Kno Dis 2006;13:119–36.

45. Weiss L. Metastatic inefficiency. Adv Cancer Res 1990;54:159–211.

46. Wong CW, Lee A, Shientag L,Yu J, Dong Y, Kao G, et al. Apoptosis: An earlyevent in metastatic inefficiency. Cancer Res 2001;61:333–8.

47. ArnoldMA, Burmeister JJ, Small GW. Phantom glucose calibrationmodelsfrom simulated noninvasive human near-infrared spectra. Anal Chem1998;70:1773–81.

48. Lu X, Kang Y. Efficient acquisition of dual metastasis organotropism tobone and lung through stable spontaneous fusion between MDA-MB-231variants. Proc Nat Aca Scis 2009;106:9385–90.

49. Rizwan A, Cheng M, Bhujwalla ZM, Krishnamachary B, Jiang L, Glunde K.Breast cancer cell adhesome and degradome interact to drive metastasis.NPJ Breast Cancer 2015;1:15017.

50. Barman I, Dingari NC, Saha A, McGee S, Galindo LH, Liu W, et al.Application of raman spectroscopy to identify microcalcifications andunderlying breast lesions at stereotactic core needle biopsy. Cancer Res2013;73:3206–15.


Paidi et al.




2017;77:247-256. Published OnlineFirst November 15, 2016.Cancer Res Santosh Kumar Paidi, Asif Rizwan, Chao Zheng, et al. Premetastatic Lungs Primed by Breast CancerLabel-Free Raman Spectroscopy Detects Stromal Adaptations in

Updated version

10.1158/0008-5472.CAN-16-1862doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2016/11/15/0008-5472.CAN-16-1862.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/77/2/247.full#ref-list-1

This article cites 47 articles, 10 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/77/2/247.full#related-urls

This article has been cited by 1 HighWire-hosted articles. Access the articles at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/77/2/247To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-16-1862

http://cancerres.aacrjournals.org/content/suppl/2016/11/15/0008-5472.CAN-16-1862.DC1

http://cancerres.aacrjournals.org/content/77/2/247.full#ref-list-1

http://cancerres.aacrjournals.org/content/77/2/247.full#related-urls

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/77/2/247


label-free raman spectroscopy detects stromal adaptations ... · here, we have investigated lungs...

Documents