oxygen-enhanced and dynamic contrast- enhanced ... · convergence and technologies oxygen-enhanced...

Convergence and Technologies

Oxygen-Enhanced and Dynamic Contrast-Enhanced Optoacoustic Tomography ProvideSurrogate Biomarkers of Tumor VascularFunction, Hypoxia, and NecrosisMichal R. Tomaszewski1,2, Marcel Gehrung1,2,3, James Joseph1,2, Isabel Quiros-Gonzalez1,2,Jonathan A. Disselhorst3, and Sarah E. Bohndiek1,2

Abstract

Measuring the functionalstatus of tumor vasculature,including blood flow fluc-tuations and changes in oxy-genation, is important incancer staging and therapymonitoring. Current clini-cally approved imagingmodalities suffer long proce-dure times and limitedspatiotemporal resolution.Optoacoustic tomography(OT) is an emerging clinicalimaging modality that mayovercome these challenges.By acquiringdata atmultiplewavelengths,OT can interro-gate hemoglobin concentra-tion and oxygenation direct-ly and resolve contributionsfrom injected contrast agents. In this study, we testedwhether two dynamicOT techniques, oxygen-enhanced (OE) and dynamiccontrast-enhanced (DCE)-OT, could provide surrogate biomarkers of tumor vascular function, hypoxia, and necrosis. We foundthat vascular maturity led to changes in vascular function that affected tumor perfusion, modulating the DCE-OT signal.Perfusion in turn regulated oxygen availability, driving the OE-OT signal. In particular, we demonstrate for the first time a strongper-tumor and spatial correlation between imaging biomarkers derived from these in vivo techniques and tumor hypoxiaquantified ex vivo. Our findings indicate that OT may offer a significant advantage for localized imaging of tumor response tovascular-targeted therapies when compared with existing clinical DCE methods.

Significance: Imaging biomarkers derived from optoacoustic tomography can be used as surrogate measures of tumorperfusion and hypoxia, potentially yielding rapid, multiparametric, and noninvasive cancer staging and therapeutic responsemonitoring in the clinic.

Graphical Abstract: http://cancerres.aacrjournals.org/content/canres/78/20/5980/F1.large.jpg. Cancer Res; 78(20); 5980–91.�2018 AACR.

IntroductionAngiogenesis, the growth of new blood vessels from surround-

ing host vasculature, can be a rate-limiting process in tumor

development and progression. The resulting tumor vasculatureis often chaotic and tortuous, leading to high intratumoral het-erogeneity in vascular density and function (1). A high density of

1Department of Physics, University of Cambridge, Cambridge, United Kingdom.2Cancer Research UKCambridge Institute, University of Cambridge, Cambridge,United Kingdom. 3Werner Siemens Imaging Center, Preclinical Imaging andRadiopharmacy, University of Tuebingen, Tuebingen, Germany.

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

Corresponding Author: Sarah E. Bohndiek, University of Cambridge, CavendishLaboratory, JJ Thomson Avenue, Cambridge CB3 0HE, United Kingdom. Phone:44-1223-337267; Fax: 44-1223-337000; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-18-1033

�2018 American Association for Cancer Research.

CancerResearch

Cancer Res; 78(20) October 15, 20185980

on March 1, 2021. © 2018 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst August 16, 2018; DOI: 10.1158/0008-5472.CAN-18-1033

http://crossmark.crossref.org/dialog/?doi=10.1158/0008-5472.CAN-18-1033&domain=pdf&date_stamp=2018-10-1

http://cancerres.aacrjournals.org/content/canres/78/20/5980/F1.large.jpg

http://cancerres.aacrjournals.org/

tumor vasculature does not necessarily translate into efficientoxygen and nutrient transport (2). Diffusion-limited hypoxiaemerges early in tumor development, as rapidly proliferatingcancer cells experience a gradient of hypoxia with increasingdistance from the nearest perfused blood vessel (3). Perfusion-limited (or "cycling") hypoxia occurs in cells close toblood vesselsthat experience rapid spatiotemporal fluctuations in local oxygendelivery due to highly variable blood flow (4). Hypoxia in solidtumors has been associatedwith both chemo- and radioresistance(4), as well as poor prognosis (5, 6). Furthermore, antiangiogenicand vascular disrupting therapies are under active clinical devel-opment, with highly variable rates of success (7, 8). A rapid test toprobe the functional status of the tumor vasculature, includingblood flow fluctuations and changes in oxygenation, could there-fore improve cancer patient management, for example, in distin-guishing benign frommalignant tumors, in monitoring responseto chemo- and radiotherapy, and in aiding development of novelvascular-targeted therapies (1).

Noninvasive imaging of tumor vascular function in the clinicusually requires administration of an exogenous untargeted con-trast agent followed by longitudinal imaging of wash-in andwash-out kinetics, referred to as dynamic contrast-enhanced(DCE) imaging (9). DCE MRI has been broadly applied tointerrogate tumor perfusion by tracking the dynamics of aninjected gadolinium-based small-molecule contrast agent (10).Unfortunately, an increasing number of reports suggesting long-term toxicity of gadolinium chelates (11)may limit future use andtypical voxel sizes of approximately 2 � 2 � 4 mm3 (12) limitinterrogation of spatial heterogeneity (5). To avoid the use ofcontrast agents, label-free MRI techniques that are sensitive toperfusion such as arterial spin labeling (ASL) may be used;however, ASL can suffer from low signal-to-noise ratio, typicallyrequiring voxel sizes of over approximately 3 � 3 � 4 mm3 inpatients (13). A more established MRI technique is blood oxygenlevel-dependent (BOLD) MRI, sensitive to deoxyhemoglobincontent, which can reach submillimeter in-plane spatial resolu-tion and temporal resolution of < 10 seconds clinically (14). TheBOLD signal appears to reflect tumor perfusion and hypoxia,based on correlations with immunohistochemistry (15), and canbe applied, for example, to indicate prognosis in chemo- andradiotherapy (14, 16). Despite this promise, attempts to directlycorrelate the BOLD and DCE-MRI signals have shown no signif-icant relationship (17), and it has been suggested that someambiguity remains in the biological interpretation of the BOLDsignal (14, 18).

Considering other clinical imaging modalities, PET contrastagents are available clinically for visualization of vascularfunction (e.g., H2

15O) and hypoxia (e.g., 18F-MISO). Althoughthese approaches benefit from the exquisite sensitivity of PET,difficulties arise from the fundamental spatial resolution limits(19) and the requirement to administer a radiopharmaceutical,which is a particular challenge for short-half-life agents such as15O (t1/2 � 2 minutes; ref. 20). A more cost-effective optionmay be possible with diffuse optical spectroscopic imaging,which measures concentrations of oxy- and deoxyhemoglobinas surrogate markers of hypoxia and is in clinical trials (21),though this all-optical imaging approach suffers from very poorspatial resolution (�1 cm). A cost-effective and high-resolutionsolution could be available using DCE ultrasound with gas-filled microbubbles as an exogenous contrast agent (22), yetsafety concerns related to injection of microbubbles have been

raised in patients (23). Thus, there remains a need for cost-effective, noninvasive imaging of tumor vascular function withhigh spatiotemporal resolution, ideally available without con-trast agent administration.

Optoacoustic tomography (OT) is an emerging imagingmodality (24) that is currently in clinical trials (25). OT revealsthe distribution of tissue optical absorption in real time (26).Because the optical absorption spectra of oxy- and deoxyhemo-globin are distinct, acquiring OT data at multiple wavelengthsmakes it possible to derive imaging biomarkers that relate tototal hemoglobin concentration (THb) and oxygenation (SO2).These imaging biomarkers provide complementary hemody-namic information to thosemeasured clinicallywithDCE-basedtechniques and also the label-free MRI-based techniques intro-duced above. OT has been shown to monitor the evolution oftumor vasculature during disease development (27, 28) and todetect response to vascular-targeted therapies (29, 30). OT hasalso been combined with DCE ultrasound (31, 32), showingrelationships between hemoglobin parameters and perfusionmetrics. For these reasons, OT has already been deployed inclinical studies in breast, ovarian, and prostate cancers amongothers, achieving localized imaging at depths of up to 7 cm withspatial resolution of 500 mm or better and wavelength tuningrates of up to 100 Hz (25). Importantly, numerous clinical trialsare underway world-wide, which are beginning to show greatpromise for the technology (despite the aforementioned depthlimitations) for detecting tumor vascularization and differenti-ating benign and malignant lesions, particularly in the breast(33–35).

In addition to the "static" measurements of hemoglobin con-centration and oxygenation available with existing OT, newtechniques have recently emerged that directly report on vascularmaturity and function. Inspired by clinically approved oxygen-enhanced (OE) MRI methods, OE-OT (36) measures the changein hemoglobin oxygenation following a change in respiratory gasfrom air to 100% oxygen. Contrary to the static measurement ofoxygenation, these "dynamic" OE-OT biomarkers have beenshown to correlate with histopathologic analysis of tumor vas-cular function and substantially outperform the static biomarkersin terms of robustness and repeatability (36). DCE-OT is alsoavailable, using the clinically approved fluorescent agent indo-cyanine green (ICG) as an untargeted blood pool agent (37).Taking multiwavelength OT data over time makes it possible toseparate ICG signals from oxy- and deoxyhemoglobin, giving thepotential to extract spatially resolved relationships betweentumor oxygenation and tumor perfusion in a clinical setting usingnontoxic, noninvasive OT imaging.

The purpose of the present study was to evaluate the potentialof OT to be used for rapid, multiparametric, noninvasive assess-ment of tumor vascular function, hypoxia, and necrosis. Here, weperform coregistered OE-OT and DCE-OT in two tumor models,showing for the first time a quantitative spatial per-pixel corre-lation betweenOTmetrics derived in vivo and the histopathologicassessment of vascular maturity and tissue hypoxia ex vivo. Fur-thermore, we resolved the key determinants ofOE-OT response intermsof oxygendelivery via the blood supply andoxygendemandin the tissue. Our findings suggest that OE-OT– and DCE-OT–derived imaging biomarkers can be used as surrogate measures oftumor perfusion and hypoxia. We also note that OE-OT mayprovide a label-free alternative to DCE approaches for evaluatingtumor perfusion that can be readily implemented into the

Optoacoustic Tomography Measures Tumor Vascular Function

www.aacrjournals.org Cancer Res; 78(20) October 15, 2018 5981




imaging protocol of the emerging clinical optoacoustic technol-ogy thanks to its negligible toxicity risk.

Materials and MethodsAnimal experiments

All animal procedures were conducted in accordance withproject (70-8214) and personal licenses (IDCC385D3) issuedunder the United Kingdom Animals (Scientific Procedures) Act,1986, andwere approved locally under compliance form numberCFSB0671. Subcutaneous tumors were established inmale BALB/c nude mice (Charles River). Note that 1.5 � 106 PC3 prostateadenocarcinoma cells (donated by Dr. Jason Carroll's lab at theCRUK Cambridge Institute in July 2015) suspended in a mixtureof 50 mL PBS and 50 mL matrigel (354248; Corning) wereinoculated subcutaneously in both lower flanks of 9mice (result-ing in 18 tumors). Also note that 1� 106 K8484mouse pancreaticadenocarcinoma cells suspended in 100 mL PBS were inoculatedsubcutaneously in both lower flanks of a further 4mice (resultingin 7 tumors). The K8484 cells (38) were derived from a pancreaticadenocarcinoma of a transgenic mouse model (39, 40) and werekindly donated inOctober 2017 byProfessorDuncan Jodrell's labat the CRUK Cambridge Institute, providing validation of thefindings in amodel of distinctmorphology to theprostate tumors.Authentication of PC3 cells using Genemapper ID v3.2.1 (Genet-ica) bySTRGenotyping (1/2015) showed94%match.Noauthen-tication was performed in K8484 cells. Both cell types weremycoplasma tested by RNA-capture ELISA prior to use (PC3,March 13, 2017; K8484, September 05, 2017). Cells were usedat 4 passages from thawing from frozen stocks. Tumor growthwasmonitored by calipers, and imaging was performed when tumorsreached approximately 8 mm in any linear dimension.

To investigate the effect of vascular disruption on the optoa-coustic measurements and evaluate its relationship with vascularmaturity, 1.5 � 106 PC3 prostate adenocarcinoma cells wereinoculated as described above in a further 8 mice (resulting in16 tumors). In 4 animals, both tumors could not be visualized in asingle imaging slice, resulting in 4 tumors being excluded, leaving12 tumors for analysis. Combretastatin 4A phosphate (CA4P,C7744; Sigma-Aldrich), a potent Vascular Disrupting Agent withwell-established efficacy in preclinical models, was used (30, 41,42). When tumors reached approximately 8 mm linear dimen-sion, mice were randomly allocated into two groups: treated(CA4P, 8 mL/kg of 12.5 mg/mL solution dosed intraperitoneallyto achieve a dose of 100mg/kg, n¼ 7 tumors, 4mice) and vehicle(PBS 8 mL/kg intraperitoneally, n ¼ 5 tumors, 4 mice).

OTA commercial multispectral OT (MSOT) system (inVision

256-TF; iThera Medical GmbH) was used for this study (43).Briefly, a tunable optical parametric oscillator pumped by anNd:YAG laser provides excitation pulses with a duration of 9 nsfor wavelengths ranging from 660 to 1,300 nm at a repetitionrate of 10 Hz, wavelength tuning speed of 10 ms, and a peakpulse energy of 90 mJ at 720 nm. Ten arms of a fiber bundleprovide uniform illumination of a ring-shaped light strip ofapproximately 8 mm width. For ultrasound detection, 256toroidally focused ultrasound transducers with a center fre-quency of 5 MHz (60% bandwidth), organized in a concavearray of 270 degree angular coverage and a radius of curvatureof 4 cm, are used.

Mice were prepared for OT according to our standard operatingprocedure (44). Briefly, mice were anaesthetized using <3% iso-flurane, placed on a heat pad, and a catheter (home-made with30G needle) was placed in the tail vein and fixed in place usingtissue glue (TS1050071F; TissueSeal). The mouse was subse-quently moved into a custom animal holder (iThera Medical)wrapped in a thin polyethylene membrane, with ultrasound gel(Aquasonic Clear; Parker Labs) used to couple the skin to themembrane. The holder was placedwithin the imaging chamber ofthe MSOT system filled with degassed heavy water (617385;Sigma-Aldrich) maintained at 36�C, with the end of the catheterline available outside of the imaging chamber for contrast agentinjection. Heavywater was used due to other studies performed inparallel on the MSOT system and is not essential to the studydescribed here because the optical absorption of water and heavywater is similar in the spectral range interrogated.

Mice were allowed to stabilize their physiology for 15 minuteswithin the system prior to initialization of the scan, and theirrespiratory rate was then maintained in the range 70 to 80 bpmwith approximately 1.8% isoflurane concentration for the entirescan. The respiration rate wasmonitored by observing the breath-ing motion of the animal using a video feed from an opticalcamera positioned within the imaging chamber and counting thebreaths over a minute using a stopwatch. We first performed OEOT (OE-OT; ref. 36), in which the breathing gas was switchedmanually from medical air (21% oxygen) to pure oxygen (100%oxygen), using separate flowmeters (according to the schedule inSupplementary Fig. S1). A single slice was chosen for imagingshowing the largest cross-sectional area of the tumors on bothflanks where possible. Images were acquired in the single sliceusing 10 wavelengths (700, 730, 750, 760, 770, 800, 820, 840,850, and 880 nm) and an average of 6 pulses per wavelength; anentire single slice multiwavelength data acquisition was 5.5seconds in duration. In the CA4P- and vehicle-treatedmice, whereimaging was performed twice, the imaging slice in the secondsession was chosen to be as close as possible to the first one byvisual alignment to the reconstructed images of the first scan.

Following OE-OT, the breathing gas was switched back tomedical air and after 10 minutes allowed for equilibration, theDCE-OT was initiated in the same imaging slice. Images wereacquired using 5 wavelengths (700, 730, 760, 800, and 850 nm)and an average of 10 pulses. After 1minute of continuous imagingto establish the baseline signal, a bolus of ICG (40 nmol/20 gmouse in PBS; ref. 45) was injected intravenously through thecatheter, followed by a pulse of PBS to flush the line. OT wascontinued for a further 15 minutes to sample the enhancementcurve.

All mice underwent the full OT procedure at least once. Micereceiving CA4P or vehicle were imaged at 48 hours before treat-ment to ensure clearing of the injected contrast agent and thenagain at 4 hours after treatment.

Histopathologic tumor stainingFollowing the last OT procedure, mice were immediately

sacrificed by cervical dislocation while still under anesthesia. Thetumors were then excised, taking care for the orientation to bepreserved, and cut in half along the imaging plane. The top- andleft-hand sides of the tumors were marked with green and redtissue marking dyes (RCD-0727-3, RCD-0727-5, Cell Path) tolater indicate the orientation of histopathologic sections relativeto the in vivo imagingprocedure.Onehalfwas thenfixed inneutral

Tomaszewski et al.

Cancer Res; 78(20) October 15, 2018 Cancer Research5982




buffered 10% formalin for 24 hours prior to paraffin embedding.Fixed blocks were sectioned at 3-mm thickness at 4 separate levelswithin the tumor spaced by 500mmapart.Hematoxylin and eosin(H&E) staining and immunohistochemistry were performed.Adjacent sections from each of the 4 levels were stained withCD31 (anti-mouse; BD Biosciences, 553370) to indicate vesseldensity; alpha smooth muscle actin (ASMA; anti-mouse; Abcam,ab5694) to indicate smooth muscle coverage; and CAIX (anti-human, BioScience Slovakia, AB1001) to indicate hypoxicregions. We also performed pimonidazole staining for hypoxiain a representative tumor to qualitatively assess the reliability ofCAIX staining for hypoxia visualization in the PC3 model. Notethat 60 mg/kg pimonidazole hydrochloride (Hypoxyprobe) inPBS was injected intraperitoneally 60 minutes before sacrifice.IHC staining (Mab-1 antibody, 4.3.11.3, Hypoxyprobe) wasperformed on the sections. Spatial colocalization between CAIXand pimonidazole staining was observed (Supplementary Fig.S2). All immunostainings were performed with DAB as substrate.All sections were digitized at 20x with an Ariol System (AperioTechnologies Ltd.).

OT image analysisAll OT analysis was performed in MATLAB 2017b (Math-

works) using custom software. OT images were reconstructedusing an acoustic backprojection algorithm (iThera Medical)with an electrical impulse response correction, to account forthe frequency-dependent sensitivity profile of the transducers(46), and a speed-of-sound adjustment, to focus the images.Images were reconstructed with a pixel size of 75 mm � 75 mm,which is approximately equal to half of the in-plane resolutionof the InVision 256-TF, to facilitate region drawing. It should benoted that the out-of-plane resolution of this system is approx-imately 0.9 mm (47). Regions of interest (ROI) were drawnmanually around the tumor area (excluding the skin) and ahealthy, well-vascularized tissue region around the spine, in the800 nm (isosbestic) image taken from the first frame of the OE-OT scan. The reconstructed images were downsampled to 200mm � 200 mm pixel size for further analysis, to improveresponse classification, as described below.

For OE-OT analysis, a pseudoinverse matrix inversion (pinvfunction inMATLAB2017b)was used for spectral unmixing of therelative weights of oxy-[HbO2] and deoxyhemoglobin [Hb] inde-pendently in each pixel. Because OT is not able to accuratelymeasure the absolute SO2 without the precise knowledge ofoptical energy distribution, we denote the approximate oxygen-ation metric derived in this study as the apparent SO2

MSOT ratherthan absolute SO2. SO2

MSOT was computed as the ratio of HbO2

to total hemoglobin THb ¼ [HbO2 þHb]. Average SO2MSOT was

calculated in each pixel for air and oxygen breathing periods anddenoted SO2

MSOT(Air) and SO2MSOT(O2), respectively. The

amplitude of response to the oxygen gas DSO2MSOT ¼ SO2

M-

SOT(O2) – SO

2MSOT(Air) was calculated for each pixel (illustrated

in Fig. 1A). The variability of the signal was also assessed bycalculating the standard deviation SDOE of the SO2

MSOT valuesbetween the individual scans acquired during air breathing. Eachpixel was classified as responding to the oxygen challenge ifDSO2

MSOT exceeded 2 � SDOE (see Supplementary Fig. S3). Asmall fraction of pixels showed artifactual negative Hb or HbO2

levels due to low signal andwere classified as nonresponding. TheOE responding fraction (RF)was subsequently calculated for eachtumor and scan as the ratio of the number of tumor pixels

classified as responding to the total number of pixels in thetumor ROI.

DCE-OT analysis was performed similarly. The same ROIs asfor the corresponding OE-OT scans were used, because the imag-ing was performed in the same slice and the movement of theanaesthetized animal between the scans was negligible. Afterdown-sampling the reconstructed image, linear spectral unmixingas abovewas performed forHbO2,Hb, and ICG. The amplitude ofICG enhancement, DICG, was quantified as the differencebetween the average baseline ICG signal and themaximum signalrecorded in the first 3 minutes after injection (illustrated inFig. 1A) to capture the perfusion rather than accumulation effectof the dye. Variability of the ICG signal was also measured as theSD of the individual images acquired before contrast agentinjection (SDDCE). Each pixel was then classified as enhancingwhen DICG exceeded 2 � SDDCE (see Supplementary Fig. S3),with artifactual negative pixels classified as nonenhancing. DCERF was computed accordingly for each tumor.

Correlations between OE and DCE signals were calculated foreach tumor on a per-pixel basis. The results presented are from allmice (n ¼ 12þ18 tumors). The small fraction of pixels showingartifactual SO2

MSOT or ICG signal values was excluded from thecorrelation analysis. Spearman rank correlation coefficient wascalculated (MATLAB) and quoted, due to the apparent nonlinearmonotonic relationship between the metrics.

Histopathologic image analysis and data coregistrationFor each tumor, 4 sectionswere analyzed (see "Histopathologic

Tumor Staining"). Necrosis was identified from H&E sectionsusing a Convolution Neural Network (CNN) approach. Theschematic of the CNN layer architecture is presented in Supple-mentary Fig. S4. The viable and necrotic patches of H&E sectionsfor training the model were identified manually. A threshold of0.5 was applied to the necrosis score maps to discriminate thenecrotic from viable regions as the probabilistic output had arange from 0 to 1, meaning that values below or above 0.5 have ahigher likelihood of being viable or necrotic tissue, respectively.The necrotic fraction was quantified as the ratio of the totalnecrotic area to total tumor area across a whole section. Modelperformance was assessed by qualitative comparison with H&Esections, an example of which is demonstrated in SupplementaryFig. S5A and S5B, and quantitative comparison to results ofmanual segmentation, performed in Imagescope (Aperio Tech-nologies Ltd.). A strong, significant correlation was observedbetween the model and manual quantification (r ¼ 0.75; P <0.0001, see Supplementary Fig. S6).

Hemorrhagic areas were identified in H&E sections based ontheir color.Quantificationof hemorrhagic fractionwas performedautomatically using Halo (Indica Labs) image analysis software.Analysis of CD31 and ASMA coverage was also performed usingHalo software, quantifying the CD31-positive area (to measurethe amountof vasculature), aswell as theCD31-positive areas thatwere alsopositive forASMA inadjacent sections to identifymaturevasculature with smoothmuscle coverage (2). The fraction of areapositive for both CD31 and ASMA to the CD31-positive area wasquoted as a metric. CAIX analysis was performed using customcode written inMATLAB. The areas of CAIX-positive staining wereidentified based on color deconvolution (48) of the antibody, cellnuclei, and background. The correct colors for the 3 classes werecomputed by manually outlining example areas in two sections.CD31/ASMA analysis was performed in the CA4P/vehicle-treated






cohort (n ¼ 12), to increase the vascular maturity range probed.The CAIX and necrosis analysis was performed in the other cohortonly (n ¼ 18), as the different experimental protocol (2 imagingsessions for CA4P cohort, 1 for untreated cohort) made thehistology datasets not eligible to be combined.

To evaluate the relationship between OT metrics of vascularfunction and histopathologic assessment of tissue hypoxia, pointset registration was performed with points determined by theapplied tissue marking dyes (49).

The DSO2MSOT and DICG OT images were then compared

spatially with CAIX sections. Kernel density estimation, assuminga bimodal intensity distribution with low and high values, wasapplied on pooled CAIX stain intensity values across all tumors toobtain a discriminative threshold for binarization of stain inten-sity in individual sections. The binarized CAIX and necrosis maps

were overlaid, and the necrotic areas were excluded from theanalysis. Mean DSO2

MSOT and DICG values in CAIX-positive and-negative viable regions were then calculated. The differencesbetween the DSO2

MSOT and DICG values in the CAIX-positiveand -negative regions were then extracted for each tumor, with avalue significantly different from 0 taken as a measure of differ-ential response. This informed on spatial colocalization and co-occurrence of high/low DSO2

MSOT with negative/positive CAIXregions. It should be noted that this approach to image coregis-tration is subject tohuman error in tissuehandling and sectioning,which leads to a high rate of exclusion for the analysis. Out of the18 tumors analyzed in the cohort, 8 had to be excluded due tofailure of the registration, arising from distorted or torn tissuesections, resulting inmisplaced tissuemarking dyes, which in turnobstructed accurate registration of the image pairs.

Figure 1.

TumorOE-OT andDCE-OT responses are strongly correlated. A strong spatial relationshipwas observed between the responsemaps for OE-OT (A) andDCE-OT (B)in both PC3 (top) and K8484 (bottom) tumors. OE and DCE kinetic curves (C and D) were used to quantify metrics, as denoted, that were then compared incorrelation analyses on a per-pixel basis in each tumor. E, Exemplar per-pixel correlations for each tumor type. F, When comparing correlations extractedfrom the entire tumor cohort (each data point represents one tumor), significantly stronger correlations were observed in the dynamic OE-OT metric DSO2

MSOT

than either static metrics of SO2MSOT(Air) and SO2

MSOT(O2). Red dashed horizontal line, no correlation (correlation coefficient of 0). Data in A to E areexemplars taken from one representative tumor for each type. Data in F are taken from the entire tumor cohort (n ¼ 30 PC3; n ¼ 7 K8484). �� , P < 0.001 bypaired two-tailed t test. Boxes between 25th and 75th percentiles; line at median. a.u., arbitrary unit.

Tomaszewski et al.





Statistical analysisAll errors are quoted as the SEM unless otherwise stated. All

statistical analyses were performed in OriginPro 9 (OriginLab).Paired two-tailed t test compared different metrics in each tumorand changes in parameters due to CA4P treatment; unpaired two-tailed t test assuming equal variances compared tumors in dif-ferent cohorts. One-tailed t test was used to assess whether thedifferences inOT parameters between low and high CAIX stainingregions are significantly above 0 for the coregistered histopathol-ogy andOT image analysis. Only the last scan immediately beforesacrifice was used for correlations with histology, and the Pearsonrank test was performed to assess the significance. P < 0.05 wasconsidered statistically significant.

ResultsOE and DCE-OT responses are strongly correlated

Using our intrinsically coregistered OE-OT and DCE-OT data,we first sought to examine the spatial correlations between thetumor OE and DCE responses. The spatial distribution (Fig. 1Aand B) was examined, and amplitudes of these responses,DSO2

MSOT (Fig. 1C) and DICG (Fig. 1D), respectively, werecompared on a per-pixel basis. Highly significant correlations(P value < 10�6 in all cases) were observed between these twometrics (Fig. 1E). The correlation deviates from linearity for theextreme values, suggesting Spearman rank correlation coefficientas a more informative estimate of the relationship (Fig. 1E).

Although very strong correlations between OE and DCE responsewere observed in both tumor cohorts (Spearman r¼ 0.64� 0.02,n ¼ 30 PC3 tumors; Spearman r ¼ 0.65 � 0.07, n ¼ 7 K8484tumors), no correlation was observed to the static metrics ofSO2

MSOT(Air) or SO2MSOT(O2) measured at baseline, indicating

that these metrics are not sensitive to tumor perfusion [Spearmanr ¼ –0.16 � 0.05 and r ¼ –0.11 � 0.05 for SO2

MSOT (Air) andSO2

MSOT(O2), respectively for PC3, n¼ 30 PC3 tumors, r¼ –0.04� 0.09 and r ¼ 0.25 � 0.09 respectively for K8484]. The correla-tions for all tumors analyzed are summarized in Fig. 1F.

ICG retaining pixels in DCE-OT show weak or no OE-OTresponse

Examining the DCE-OT data in greater depth, it was clear thattwo classes of pixels showing distinct ICG kinetics were present.The first such group, referred to as "clearing," consistentlyshowed an obvious enhancement peak followed by exponentialclearance of the contrast down to a plateau (Fig. 2A and B, blue).The second such group, referred to as "retaining," showed anenhancement after injection, but displayed no clearance; thelevel of signal either remained high and stable over the durationof the experiment, or even increased gradually (Fig. 2A and B,red). As might be expected based on established differences invascular maturity (36), clearing regions tended to be moreprevalent in the rim than the core of the tumor, with the fractionof the rim occupied by clearing pixels being significantlyhigher than the corresponding fraction of the core (0.51 �

Figure 2.

Two distinct classes of DCE kinetics also possess different OE kinetics. Spatially distinct regions were segmented showing ICG clearance or retention (A) followinginjection, according to the DCE-OT response kinetics (B). The retaining regions show little or no OE-OT response (C), reflecting the poorer vascular functionin these areas. Response maps of OE-OT (D) and DCE-OT (E) are also shown, with the clearing, retaining, and nonenhancing regions denoted in light blue,red, and black ROIs, respectively. These further indicate that the strongest OE response occurs in clearing regions, as suggested in C. Data shown are from arepresentative PC3 tumor. a.u., arbitrary unit.






0.04 vs. 0.39 � 0.05, P ¼ 0.002, n ¼ 30 PC3 tumors; 0.44 � 0.13vs. 0.16 � 0.06, P ¼ 0.01, n ¼ 7 K8484 tumors).

Interestingly, the OE-OT responses of these two distinct classesof DCE response also showed significant differences (Fig. 2C).Retaining regions demonstrate weaker OE response and have a

significantly lower OE RF than clearing regions (0.55 � 0.05 vs.0.25� 0.02, P < 10�5, n¼ 30 PC3 tumors). The retaining regionsalso show a weaker correlation between the DSO2

MSOT and DICG(Fig. 2D and E) than the clearing regions (0.57 � 0.03 vs. 0.44 �0.04, P ¼ 0.007).

Figure 3.

OE-OT and DCE-OT RFs show a significant positive correlation with vascular maturity. A, Overlaid CD31- and ASMA-stained sections were used to evaluatethe fraction of blood vessels positive for ASMA (red on CD31-stained section). The OE RF (B) and DCE RF (C) both show a significant correlation to tumorvascular maturity (B), with DCE showing a stronger relationship. The analysis includes tumors treated with combretastatin A4 phosphate (red points, n ¼ 7 PC3tumors), which show clearly lower ASMA coverage than vehicle-treated tumors (black points, n ¼ 5 PC3 tumors). � , P < 0.05 and �� , P < 0.01 show thestrength of the correlation assessed using a Pearson rank test. Line of best fit with 95% confidence intervals is also shown in the graphs.

Figure 4.

OE-OT is also strongly related to tumor hypoxia and necrosis. A, Representative CAIX-stained sections were used to quantify the extent of tumor hypoxia,to which OE-OT RF shows a strong inverse correlation, whereas the DCE-OT RF shows a weaker relationship. B, H&E-stained sections were used to quantifythe extent of tumor necrosis (green line outlines necrotic area), to which OE-OT again showed a strong inverse correlation, whereas DCE-OT response was notsignificant. Analysis shown from 18 PC3 tumors. n.s., not significant; � , P < 0.05; �� , P < 0.01. Line of best fit with 95% confidence intervals is shown inthe graphs where significant relationships are identified.

Tomaszewski et al.





Tumor DCE-OT signal is driven predominantly by vascularmaturity, whereasOE-OT is also strongly related tohypoxia andnecrosis

The relationships observed between OE-OT and DCE-OT sug-gested that similar vascular characteristics may underpin theirresponses. We next broadly explored the correlations between thein vivo OT responses and the ex vivo histopathologic analysisrelating to vascular maturity (ASMA coverage of CD31-positiveblood vessels), hypoxia (CAIX positivity), and tumor viability(necrosis assessedwithH&E) on aper-tumor basis in PC3 tumors,where our study was sufficiently well powered to identify signif-icant correlations.

Vascular maturity (Fig. 3A) showed a significant positive cor-relation with bothOE RF (OE RF, r¼ 0.58, P¼ 0.048, n¼ 12 PC3tumors; Fig. 3B) and DCE RF (r ¼ 0.78, P ¼ 0.002, n ¼ 12 PC3tumors; Fig. 3C). As could be expected given the direct influenceof vascular maturity on vessel function and subsequently onperfusion, a higher andmore significant correlation was observedfor DCE RF than for OE RF.

A significant negative correlation was found between hypoxia(based on CAIX-positive area fraction, Supplementary Fig. S7A)andOERF (Fig. 4A; Supplementary Fig. S7B; r¼ –0.68, P¼ 0.002,n ¼ 18 PC3 tumors). The negative correlation between hypoxiaand DCE RF was weaker (Fig. 4A; r ¼ –0.49, P ¼ 0.04, n ¼ 18tumors). A negative correlation was also observed betweenOE RFand the tumor necrotic fraction (Fig. 4B; Supplementary Fig. S8Aand S8B; r ¼ –0.56, P ¼ 0.016, n ¼ 18 tumors); however, nosignificant relationship was seen for DCE RF (Fig. 4B; Supple-mentary Fig. S8B; r ¼ –0.42, P ¼ 0.08, n ¼ 18 tumors).

Given the strong relationship between hypoxia and OE-OTresponse, we investigated further by examining the spatial colo-calization of high DSO2

MSOT and low CAIX signals. Taking eachCAIX-stained section (Fig. 5A) andH&E-stained section (Fig. 5B),we performed an image coregistration and down-sampled thespatial resolution of the CAIX image (binarized into high and low

staining regions) to match that of the in vivo OT image (Fig. 5C).The resulting coregistered CAIX data were then compared withDICG (Fig. 5D) and DSO2

MSOT (Fig. 5E) on a per-pixel basis.Where successful coregistration was possible (n¼ 10 PC3 tumors,seeMaterials andMethods), the analysis revealed that areas of lowCAIX staining (considered to reflect low tissue hypoxia) wereassociated with a notably higher mean DSO2

MSOT level than theareas of high CAIX staining (Fig. 5E). The difference betweenaverage DSO2

MSOT taken in low CAIX compared with that in highCAIX regions was significantly higher than 0 when comparedacross all tumors (difference¼ 0.014� 0.005 vs. 0, P¼ 0.007, n¼10PC3 tumors). The averageDICGwas also significantly higher innormoxic than in hypoxic tumor regions in line with the per-tumor findings (difference¼ 0.42� 0.17 vs. 0, P¼ 0.018, n¼ 10PC3 tumors).

OE-OT and DCE-OT are highly sensitive to treatment with avascular disrupting agent

Having established relationships betweenOE-OT andDCE-OTimaging biomarkers with vascular function, hypoxia, and necro-sis, we then sought to evaluate their utility in detecting response toa vascular disrupting agent. Imaging studies were performed bothbefore and 4 hours after administration of the vascular disruptingagent combretastatin A4 phosphate. The 4-hour time point waschosen so as to observe the induced vascular disruption prior tothe development of substantial tumor necrosis (30).

The dramatic effect of the treatment on tumor vasculature wasconfirmed histologically, through induction of hemorrhage (Sup-plementary Fig. S9A and S9B) and change in vascular maturity(Supplementary Fig. S9C). As desired, this was not followed byinduction of significant necrosis in the treated tumors (necroticfraction ¼ 0.28 � 0.09, n ¼ 6 vs. 0.18 � 0.06, n ¼ 5 treated vs.vehicle, P ¼ 0.38). The induced vascular disruption was qualita-tively observed in maps of DSO2

MSOT and DICG (SupplementaryFig. S10A) as well as in the quantification of the kinetic responses

Figure 5.

Spatial coregistration allows comparison of OE-OT and DCE-OT response in hypoxic tumor tissue. CAIX-stained sections (A) were binarized into low and high stainareas. This information was overlaid with necrosis map obtained from H&E sections (B), then coregistered and downsampled (C) for comparison with theoptoacoustic images. DCE-OTDICG (D) andOE-OTDSO2

MSOT (E) could thenbe comparedwith the degree ofCAIX staining in viable areas. Theboxplots show that forthe analyzed tumors, the difference between DICG (D) and DSO2

MSOT (E) between areas of low and high CAIX hypoxia staining is significantly higher than 0(indicatedwith red dashed line). Images inA toC are from a representative PC3 tumor. Analysis inD and E is presented from 10 PC3 tumors. � , P <0.05; �� , P <0.01 byone-tailed t test (deviation from 0). Box between 25th and 75th percentiles; line at median.






in both cases (Supplementary Fig. S10B and S10C). As expected,treated tumors were dominated by ICG retaining areas (as definedin Fig. 2). Clear changes in the spatial distribution of respondingpixels could be observed in both OE-OT and DCE-OT images indrug-treated tumors (Fig. 6A), but not in the vehicle-treated ones(Supplementary Fig. S11). These changes were reflected in theresponding fractions in vehicle- and CA4P-treated tumors (Fig.6B).OERF showed a significant decrease between 48hours beforeand 4 hours after treatment (0.47 � 0.05 vs. 0.16 � 0.03, P ¼0.0005, n ¼ 7 PC3 tumors). No significant change was seen invehicle-treated control animals (0.38 � 0.06 vs. 0.30� 0.05, P ¼0.44, n ¼ 5 PC3 tumors). Similarly, DCE RF showed a significantdecrease (0.63 � 0.05 vs. 0.32 � 0.02, P ¼ 0.0001, n ¼ 7 PC3tumors), whereas control did not (0.74�0.06 vs. 0.68�0.05,P¼0.34, n ¼ 5 PC3 tumors).

DiscussionThe balance of oxygen supply and demand in solid tumors can

be a key determinant of prognosis and response to therapy. The

aim of this work was to evaluate the potential of imaging bio-markers accessible using OT to be used in rapid, multiparametric,and noninvasive assessment of tumor vascular function andmonitoring response to therapy.

Wefirst examined the relationshipbetween the twoOT imagingbiomarkers under study: DSO2

MSOT, accessible without the intro-duction of a contrast agent using OE-OT; and DICG, requiringadministration of the clinically approved and nontoxic contrastagent ICG and imaged through a DCE-OT technique. Thesedynamic biomarkers were strongly spatially correlated in bothtumor models examined, suggesting that perfusion is a strongdeterminant of response in both techniques. The kinetics of theDCE-OT response also showed strong differences between "clear-ing" and "retaining" regions, the latter of which have beenpreviously described as associated with the enhanced permeabil-ity and retention effect in areas of immature and leaky vasculature(50, 51). These regions also showed distinct OE-OT responses,with greater DSO2

MSOT seen in clearing regions.We then established how these OT imaging biomarkers were

connected with ex vivomeasurements of vascular function, as well

Figure 6.

OE and DCE RFs show similarly highsensitivity in detecting changes invascular function. BothOERF andDCERF show a drastic drop due to vascularshutdown causedby the treatment, as seen inrepresentative enhancement mapsfrom a PC3 tumor (A) and in boxplots representing the cohort (n ¼ 7treated; n ¼ 5 vehiclePC3 tumors). n.s., not significant;�� , P < 0.001 by paired two-tailedt test. Box between 25th and 75thpercentile; line at median.

Figure 7.

Summary of the relationships governing tumor physiology that have been established with OT imaging biomarkers. The physiologic relationships underpinthe correlations that we observed between the in vivo and ex vivo measurements of physiologic processes. As indicated with the color bar, the moredisconnected the physiologic parameters are from each other, the weaker the observed correlations. � , P < 0.05; �� , P < 0.01.

Tomaszewski et al.





as tumor hypoxia and necrosis; these relationships are summa-rized in Fig. 7. Vascular maturity leads to changes in vascularfunction that affect tumor perfusion, modulating the DCE-OTsignal. Perfusion in turn regulates oxygen availability, driving theOE-OT signal. Insufficient oxygen supply leads to tissue hypoxiaand eventually necrosis (Fig. 7, bottom row; ref. 4). These relation-ships, and hence our understanding of the OE and DCE-OTsignals, were directly confirmed by the correlations observedbetween our in vivoOT and ex vivo histopathologic measurements(Fig. 7, middle row). The strength of the correlation reflected howclosely the individual measurement is linked with the underlyingphysiologic process (Fig. 7, top row), overall revealing a complex,yet consistent network of relationships in the tumor vascularmicroenvironment. These findings indicate that the response forDCE-OT is driven most strongly by perfusion and vascular func-tion, whichwould be expected given that ICG shows strong serumbinding in vivo. The response for OE-OT appears to also begoverned strongly by perfusion and vascular function but isfurther modulated by the tumor oxygen demand. The strong andsignificant relationships observed between theOERF andhypoxiaarea on a per-tumor basis were also confirmed on a spatial per-pixel basis.

Treatment with a potent vascular disrupting agent, combretas-tatin A4 phosphate, was used to induce vascular shutdown,causing a dramatic perfusion drop, resulting in a significantdecrease in DCE RF, as expected. Interestingly, the OE RF showedequally high sensitivity to the vascular shutdown, indicating thatit could be used as an alternative to the contrast agent–basedDCEmethods for detecting response to vascular-targeted therapies.Both of our surrogate biomarkers were able to sensitively detectresponse to the vascular-targeted therapy.

In line with our previous findings (36), static OT biomarkerssuch as SO2

MSOT(O2) and SO2MSOT(Air) showed little relation-

ship to perfusion or hypoxia. Similar relationshipswere examinedpreviously comparing tumor oxygenation assessed using staticOTwith DCE ultrasound (31, 32) or pimonidazole staining (52).Although some spatial relationships were noted, particularly inrelation to the necrotic tumor core, these studies were limitedrespectively by a lack of histologic validation, poor sensitivity ofthe optoacoustic imaging approach applied, and small numbersof biological replicates.

There remain some limitations to the presented work thatmustbe addressed in future studies. From a biological perspective,vascularization of subcutaneous models differs from that oforthotopic xenografts and spontaneous tumors (53) and maynot be representative of the vascular function found clinically insolid tumors. These findings should therefore be validated inorthotopic and transgenic tumor models prior to application ofdynamicOTmetrics in studies of cancer biology or in the clinic. Inour subcutaneous PC3 model, good colocalization was observedbetween CAIX and pimonidazole staining, which we took as anindication that CAIX staining indeed reflected hypoxia in thismodel. Although CAIX staining is well-documented to be regu-lated by the activation of hypoxia-inducible factor (54) and hasbeen widely used for ex vivo hypoxia identification, nonspecificeffects can be observed in some models; therefore, if ourfindings are to be further validated in other tumor models, itwould be prudent to use multiple methods to assess hypoxiaex vivo.

Some further limitations exist in the efficient clinical transla-tion of OT and associated imaging biomarkers. Penetration

depths of up to 3 to 7 cm (25) have been reported in patients,enabling access to superficial cancer sites, such as those in thebreast or head and neck. With the ongoing development ofendoscopic probes, imaging organs such as the prostate (55) isalso expected tobepossible, yet access to somedeep-seatedorganswill remain limited even with these technological advances. Thelocalized nature of OTmeans that it would bemost appropriatelyplaced in the patient management pathway after diagnosis oridentification of a suspicious lesion using another imaging tech-nique. Light attenuation at depth in tissue poses an additionalchallenge for signal quantification. Methods available to performlight fluence correction of OT data have received only limitedvalidation in vivo (47). Futurework is required to directly relateOTdata to absorbed optical energy density and enable absolutequantification if desired. However, qualitative features derivedfrom clinical optoacoustic images have also shown significantprognostic value (56).

In summary, we have shown that noninvasive and nontoxicOE-OT and DCE-OT techniques can be used to interrogatetumor vascular function, hypoxia, and necrosis. The compre-hensive histopathologic validation of the OT imaging biomar-kers presented here indicates that despite the aforementionedtechnical challenges that face the technology, OT is capable ofproviding a unique and rapid insight into the tumor vascularmicroenvironment. Although DCE-OT requires administrationof a contrast agent, OE-OT provides a completely noninvasive,label-free measurement; our findings indicate that the oxygenchallenge approach could be used as a safe alternative forexogenous contrast injection as it has been used clinically withno associated risk (57). OT is already being tested, with prom-ising results in numerous clinical trials in patients with cancer(33–35), despite some technical limitations of the technology.In the future, the low cost, portability, and simplicity of OTmayoffer significant advantage for localized imaging of tumorresponse to vascular-targeted therapies when compared withexisting clinical DCE methods, particularly in the neoadjuvantsetting.

Disclosure of Potential Conflicts of InterestM.R. Tomaszewski reports receiving other commercial research support from

iThera Medical. S.E. Bohndiek reports receiving other commercial researchsupport from iThera Medical and PreXion Inc. No potential conflicts of interestwere disclosed by the other authors.

Authors' ContributionsConception and design: M.R. Tomaszewski, S.E. BohndiekDevelopment of methodology: M.R. Tomaszewski, J. Joseph, I. Quiros-Gonzalez, S.E. BohndiekAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): M.R. TomaszewskiAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis):M.R. Tomaszewski, M. Gehrung, I. Quiros-Gonzalez,S.E. BohndiekWriting, review, and/or revision of the manuscript: M.R. Tomaszewski,J. Joseph, I. Quiros-Gonzalez, J.A. Disselhorst, S.E. BohndiekStudy supervision: J.A. Disselhorst, S.E. Bohndiek

AcknowledgmentsWe would like to thank the CRUK CI Core Facilities for their support of

this work, in particular, the Biological Resource Unit, Histopathology, andBiorepository. We would also like to thank Emma Brown for helpful commentson the draft article.

Data associated with this article can be found online at https://doi.org/10.17863/CAM.23164. This work was supported by Cancer Research UK





https://doi.org/10.17863/CAM.23164

https://doi.org/10.17863/CAM.23164


(C47594/A16267 andC14303/A17197) and the EPSRC-CRUKCancer ImagingCentre in Cambridge and Manchester (C197/A16465 and C8742/A18097).

The costs of publication of this article were defrayed in part by thepayment of page charges. This article must therefore be hereby marked

advertisement in accordance with 18 U.S.C. Section 1734 solely to indicatethis fact.

Received April 5, 2018; revised May 22, 2018; accepted August 13, 2018;published first August 16, 2018.

References1. Gillies RJ, Schornack PA, Secomb TW, Raghunand N. Causes and effects of

heterogeneous perfusion in tumors. Neoplasia 1999;1:197–207.2. Eberhard A, Kahlert S, Goede V, Hemmerlein B, Plate KH, Augustin HG.

Heterogeneity of angiogenesis and blood vessel maturation in humantumors: implications for antiangiogenic tumor therapies. Cancer Res2000;60:1388–93.

3. Nagy JA,ChangS-H,DvorakAM,DvorakHF.Whyare tumour bloodvesselsabnormal and why is it important to know? Br J Cancer 2009;100:865–9.

4. Michiels C, Tellier C, Feron O. Cycling hypoxia: a key feature of thetumor microenvironment. Biochim Biophys Acta - Rev Cancer 2016;1866:76–86.

5. O'Connor JPB, Rose CJ, Waterton JC, Carano RAD, Parker GJM, Jackson A.Imaging intratumor heterogeneity: role in therapy response, resistance, andclinical outcome. Clin Cancer Res 2015;21:249–57.

6. Lundgren K, Holm C, Landberg G. Hypoxia and breast cancer: prognosticand therapeutic implications. Cell Mol Life Sci 2007;64:3233–47.

7. Gacche RN, Meshram RJ. Angiogenic factors as potential drug target:Efficacy and limitations of anti-angiogenic therapy. Biochim BiophysActa- Rev Cancer 2014;1846:161–79.

8. Thorpe PE. Vascular targeting agents as cancer therapeutics. Clin Cancer Res2004;10:415–27.

9. O'Connor JPB, Tofts PS, Miles KA, Parkes LM, Thompson G, Jackson A.Dynamic contrast-enhanced imaging techniques: CT and MRI. Br J Radiol2011;84 Spec No:S112–20.

10. Paldino MJ, Barboriak DP, Folkman J, Knopp MV, Giesel FL, Marcos H,et al. Fundamentals of quantitative dynamic contrast-enhanced MRimaging. Magn Reson Imaging Clin N Am 2009;17:277–89.

11. Gulani V, Calamante F, Shellock FG, Kanal E, Reeder SB, InternationalSociety forMagnetic Resonance inMedicine.Gadoliniumdeposition in thebrain: summary of evidence and recommendations. Lancet Neurol2017;16:564–70.

12. Little RA, Barjat H, Hare JI, Jenner M,Watson Y, Cheung S, et al. Evaluationof dynamic contrast-enhanced MRI biomarkers for stratified cancer med-icine: how do permeability and perfusion vary between human tumours?Magn Reson Imaging 2018;46:98–105.

13. Grade M, Hernandez Tamames JA, Pizzini FB, Achten E, Golay X, Smits M.A neuroradiologist's guide to arterial spin labeling MRI in clinical practice.Neuroradiology 2015;57:1181–202.

14. Jiang L, Weatherall PT, McColl RW, Tripathy D, Mason RP. Blood oxygen-ation level-dependent (BOLD) contrast magnetic resonance imaging(MRI) for prediction of breast cancer chemotherapy response: a pilotstudy. J Magn Reson Imaging 2013;37:1083–92.

15. Hammond EM, Asselin M-C, Forster D, O'Connor JPB, Senra JM, WilliamsKJ. The meaning, measurement and modification of hypoxia in thelaboratory and the clinic. Clin Oncol 2014;26:277–88.

16. Rodrigues LM, Howe FA, Griffiths JR, Robinson SP. Tumor R2� is aprognostic indicator of acute radiotherapeutic response in rodent tumors.J Magn Reson Imaging 2004;19:482–8.

17. Jiang L, Zhao D, Constantinescu A, Mason RP. Comparison of BOLDcontrast and Gd-DTPA dynamic contrast-enhanced imaging in rat prostatetumor. Magn Reson Med 2004;51:953–60.

18. Howe FA, Robinson SP, McIntyre DJ, Stubbs M, Griffiths JR. Issues in flowand oxygenation dependent contrast (FLOOD) imaging of tumours. NMRBiomed 2001;14:497–506.

19. MosesWW. Fundamental limits of spatial resolution in PET. Nucl InstrumMethods Phys Res A 2011;648:S236–40.

20. Lopci E, Grassi I, Chiti A, Nanni C, Cicoria G, Toschi L, et al. PET radio-pharmaceuticals for imaging of tumor hypoxia: a review of the evidence.Am J Nucl Med Mol Imaging 2014;4:365–84.

21. Tromberg BJ, Zhang Z, Leproux A, O'Sullivan TD, Cerussi AE, CarpenterPM, et al. Predicting responses to neoadjuvant chemotherapy in breast

cancer: ACRIN 6691 trial of diffuse optical spectroscopic imaging. CancerRes 2016;76:5933–44.

22. Lassau N, Chami L, Benatsou B, Peronneau P, Roche A. Dynamic contrast-enhanced ultrasonography (DCE-US) with quantification of tumor per-fusion: a new diagnostic tool to evaluate the early effects of antiangiogenictreatment. Eur Radiol 2007;17:F89–98.

23. ter Haar G. Safety and bio-effects of ultrasound contrast agents. Med BiolEng Comput 2009;47:893–900.

24. Taruttis a., van Dam GM, Ntziachristos V. Mesoscopic and macroscopicoptoacoustic imaging of cancer. Cancer Res 2015;75:1548–60.

25. Zackrisson S, van de Ven SMWY, Gambhir SS. Light in and sound out:emerging translational strategies for photoacoustic imaging. Cancer Res2014;74:979–1004.

26. Dima A, Burton NC, Ntziachristos V. Multispectral optoacoustic tomog-raphy at 64, 128, and 256 channels. J Biomed Opt 2014;19:36021.

27. Wilson KE, Bachawal SV, Tian L, Willmann JK. Multiparametric spectro-scopic photoacoustic imaging of breast cancer development in a transgenicmouse model. Theranostics 2014;4:1062–71.

28. Quiros-Gonzalez I, TomaszewskiMR, Aitken SJ, Ansel-Bollepalli L,McDuffusL-A, Gill M, et al. Optoacoustics delineates murine breast cancer modelsdisplaying angiogenesis and vascular mimicry. Br J Cancer 20182018;1.

29. Rich LJ, Seshadri M. Photoacoustic imaging of vascular hemodynam-ics: validation with blood oxygenation level-dependent MR imaging.Radiology 2015;275:110–8.

30. Dey S, Kumari S, Kalainayakan SP, Campbell J, Ghosh P, ZhouH, et al. Thevascular disrupting agent combretastatin A-4 phosphate causes prolongedelevation of proteins involved in heme flux and function in resistant tumorcells. Oncotarget 2018;9:4090–101.

31. Bar-Zion A, Yin M, Adam D, Foster FS. Functional flow patterns andstatic blood pooling in tumors revealed by combined contrast-enhanced ultrasound and photoacoustic imaging. Cancer Res 2016;76:4320–31.

32. Shah A, Bush N, Box G, Eccles S, Bamber J. Value of combining dynamiccontrast enhanced ultrasound and optoacoustic tomography for hypoxiaimaging. Photoacoustics 2017;8:15–27.

33. Becker A, Masthoff M, Claussen J, Ford SJ, Roll W, Burg M, et al. Multi-spectral optoacoustic tomography of the human breast: characterisationof healthy tissue and malignant lesions using a hybrid ultrasound-optoacoustic approach. Eur Radiol 2018;28:602–9.

34. Diot G, Metz S, Noske A, Liapis E, Schroeder B, Ovsepian SV, et al. Multi-Spectral Optoacoustic Tomography (MSOT) of human breast cancer. ClinCancer Res 2017;23:6912–22.

35. Heijblom M, Piras D, van den Engh FM, van der Schaaf M, Klaase JM,SteenbergenW, et al. The state of the art in breast imaging using the twentephotoacoustic mammoscope: results from 31 measurements on malig-nancies. Eur Radiol 2016;26:3874–87.

36. Tomaszewski MR, Gonzalez IQ, O'Connor JP, Abeyakoon O, Parker GJ,Williams KJ, et al. Oxygen enhanced optoacoustic tomography (OE-OT)reveals vascular dynamics in murine models of prostate cancer. Theranos-tics 2017;7:2900–13.

37. Weber J, Beard PC, Bohndiek SE. Contrast agents for molecular photo-acoustic imaging. Nat Methods 2016;13:639–50.

38. Courtin A, Richards FM, Bapiro TE, Bramhall JL, Neesse A, Cook N, et al.Anti-tumour efficacy of capecitabine in a genetically engineered mousemodel of pancreatic cancer. PLoS One 2013;8:e67330.

39. Olive KP, Jacobetz MA, Davidson CJ, Gopinathan A, McIntyre D,Honess D, et al. Inhibition of Hedgehog signaling enhances deliveryof chemotherapy in a mouse model of pancreatic cancer. Science2009;324:1457–61.

40. Schreiber FS,Deramaudt TB, Brunner TB, BorettiMI, GoochKJ, StoffersDA,et al. Successful growth and characterization of mouse pancreatic ductal

Tomaszewski et al.





cells: functional properties of theKi-RASG12Voncogene.Gastroenterology2004;127:250–60.

41. Liu L, Su X, Mason RP. Dynamic contrast enhanced fluorescentmolecular imaging of vascular disruption induced by combretas-tatin-A4P in tumor xenografts. J Biomed Nanotechnol 2014;10:1545–51.

42. Salmon BA, Siemann DW. Characterizing the tumor response to treatmentwith combretastatin A4 phosphate. Int J Radiat Oncol Biol Phys 2007;68:211–7.

43. Morscher S, Driessen WHP, Claussen J, Burton NC. Semi-quantitativemultispectral optoacoustic tomography (MSOT) for volumetric PK imag-ing of gastric emptying. Photoacoustics 2014;2:103–10.

44. Joseph J, Tomaszewski MR, Quiros-Gonzalez I, Weber J, Brunker J, Bohn-diek SE. Evaluationof precision in optoacoustic tomography for preclinicalimaging in living subjects. J Nucl Med 2017;58:807–14.

45. Taruttis A, Morscher S, Burton NC, Razansky D, Ntziachristos V. Fastmultispectral optoacoustic tomography (MSOT) for dynamic imaging ofpharmacokinetics and biodistribution in multiple organs. PLoS One2012;7:e30491.

46. Caballero MAA, Rosenthal A, Buehler A, Razansky D, Ntziachristos V.Optoacoustic determination of spatiotemporal responses of ultra-sound sensors. IEEE Trans Ultrason Ferroelectr Freq Control 2013;60:1234–44.

47. Brochu FM, Brunker J, Joseph J, Tomaszewski MR, Morscher S, BohndiekSE. Towards quantitative evaluation of tissue absorption coefficients usinglight fluence correction in optoacoustic tomography. IEEE Trans MedImaging 2017;36:322–31.

48. Ruifrok AC, Johnston DA. Quantification of histochemical staining bycolor deconvolution. Anal Quant Cytol Histol 2001;23:291–9.

49. Hill DLG, Hawkes DJ, Crossman JE, Gleeson MJ, Cox TCS, Bracey EECML,et al. Registration of MR and CT images for skull base surgery using point-like anatomical features. Br J Radiol 1991;64:1030–5.

50. Onda N, Kimura M, Yoshida T, Shibutani M. Preferential tumor cellularuptake and retention of indocyanine green for in vivo tumor imaging. Int JCancer 2016;139:673–82.

51. Herzog E, Taruttis A, Beziere N, Lutich AA, Razansky D, Ntziachristos V.Optical imaging of cancer heterogeneity with multispectral optoacoustictomography. Radiology 2012;263:461–8.

52. Gerling M, Zhao Y, Nania S, Norberg KJ, Verbeke CS, Englert B, et al. Real-time assessment of tissue hypoxia in vivo with combined photoacousticsand high-frequency ultrasound. Theranostics 2014;4:604–13.

53. Eklund L, Bry M, Alitalo K. Mouse models for studying angiogenesis andlymphangiogenesis in cancer. Mol Oncol 2013;7:259–82.

54. Wykoff CC, Beasley NJ, Watson PH, Turner KJ, Pastorek J, Sibtain A, et al.Hypoxia-inducible expression of tumor-associated carbonic anhydrases.Cancer Res 2000;60:7075–83.

55. Lediju BellMA,GuoX, SongDY, Boctor EM. Transurethral light delivery forprostate photoacoustic imaging. J Biomed Opt 2015;20:036002.

56. Neuschler EI, Butler R, Young CA, Barke LD, Bertrand ML, B€ohm-V�elez M,et al. A pivotal study of optoacoustic imaging to diagnose benign andmalignant breast masses: a new evaluation tool for radiologists. Radiology2018;287:398–412.

57. Moreton FC, Dani KA, Goutcher C, O'Hare K, Muir KW. Respiratorychallenge MRI: practical aspects. NeuroImage Clin 2016;11:667–77.






2018;78:5980-5991. Published OnlineFirst August 16, 2018.Cancer Res Michal R. Tomaszewski, Marcel Gehrung, James Joseph, et al. Function, Hypoxia, and NecrosisTomography Provide Surrogate Biomarkers of Tumor Vascular Oxygen-Enhanced and Dynamic Contrast-Enhanced Optoacoustic

Updated version

10.1158/0008-5472.CAN-18-1033doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2018/08/16/0008-5472.CAN-18-1033.DC1

Access the most recent supplemental material at:

Overview

Visual

http://cancerres.aacrjournals.org/content/78/20/5980/F1.large.jpgA diagrammatic summary of the major findings and biological implications:

Cited articles

http://cancerres.aacrjournals.org/content/78/20/5980.full#ref-list-1

This article cites 56 articles, 12 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/78/20/5980.full#related-urls

This article has been cited by 10 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/78/20/5980To 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-18-1033

http://cancerres.aacrjournals.org/content/suppl/2018/08/16/0008-5472.CAN-18-1033.DC1

http://cancerres.aacrjournals.org/content/78/20/5980/F1.large.jpg

http://cancerres.aacrjournals.org/content/78/20/5980.full#ref-list-1

http://cancerres.aacrjournals.org/content/78/20/5980.full#related-urls

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

mailto:[email protected]

http://cancerres.aacrjournals.org/content/78/20/5980


oxygen-enhanced and dynamic contrast- enhanced ... · convergence and technologies oxygen-enhanced...

Documents