Noninvasive Investigation of Blood Oxygenation Dynamics of Tumors by Near-Infrared Spectroscopy

mbtn

Noninvasive investigation of blood oxygenationdynamics of tumors by near-infrared spectroscopy

Hanli Liu, Yulin Song, Katherine L. Worden, Xin Jiang, Anca Constantinescu, andRalph P. Mason

The measurement of dynamic changes in the blood oxygenation of tumor vasculature could be valuablefor tumor prognosis and optimizing tumor treatment plans. In this study we employed near-infraredspectroscopy ~NIRS! to measure changes in the total hemoglobin concentration together with the degreeof hemoglobin oxygenation in the vascular bed of breast and prostate tumors implanted in rats. Mea-surements were made while inhaled gas was alternated between 33% oxygen and carbogen ~95% O2, 5%CO2!. Significant dynamic changes in tumor oxygenation were observed to accompany respiratorychallenge, and these changes could be modeled with two exponential components, yielding two timeconstants. Following the Fick principle, we derived a simplified model to relate the time constants totumor blood-perfusion rates. This study demonstrates that the NIRS technology can provide an effi-cient, real-time, noninvasive means of monitoring the vascular oxygenation dynamics of tumors andfacilitate investigations of tumor vascular perfusion. This may have prognostic value and promisesinsight into tumor vascular development. © 2000 Optical Society of America

OCIS codes: 170.1470, 170.3660, 170.4580, 170.5280, 290.1990, 290.7050.

Com

a

1. Introduction

The presence and the significance of tumor hypoxiahave been recognized since the 1950’s. There is in-creasing evidence that tumor oxygenation is clini-cally important in predicting tumor response toradiation, tumor response to chemotherapy, overallprognosis, or all three. Hypoxic cells in vitro and inanimal tumors in vivo are documented to be 3 timesmore resistant to radiation-induced killing comparedwith aerobic cells.1 Recent studies show that hy-poxia may have a profound impact on malignant pro-gression and on responsiveness to therapy.2,3

Numerous studies on tumor oxygen tension ~pO2!easurements have been conducted in recent years

y use of a variety of methods, such as microelec-rodes,2 phosphors,4 electron paramagnetic reso-ance,5 or magnetic resonance imaging6 ~MRI!.

H. Liu [email protected]!, Y. Song, L. Worden, and X. Jiang arewith the Joint Graduate Program in Biomedical Engineering, Uni-versity of Texas at Arlington, Arlington, Texas 76109. A. Con-stantinescu and R. P. Mason are with the Department ofRadiology, University of Texas Southwestern Medical Center, Dal-las, Texas 75390.

Received 16 March 2000; revised manuscript received 25 June2000.

0003-6935y00y285231-13$15.00y0© 2000 Optical Society of America

omparing needle-based, oxygen-sensitive electrodesr electron paramagnetic resonance and MRI foreasuring pO2 shows that the latter two offer the

advantage of facilitating multiple repeated measure-ments to map pO2 noninvasively. However, mag-nets are large, and the methods are not readilyportable. A versatile method for monitoring intra-tumor oxygenation rapidly and noninvasively istherefore very desirable for tumor prognosis and tu-mor treatment planning.

In the near-infrared ~NIR! region ~700–900 nm!the major chromophores in tissue are oxygenated he-moglobin and deoxygenated hemoglobin, which differin their light absorption. Measurements of the ab-sorption of light travelling through the tissue understudy allow us to evaluate or quantify blood oxygen-ation, such as the concentrations of oxygenated he-moglobin ~HbO2!, and deoxygenated hemoglobin ~Hb!nd the hemoglobin saturation SO2. In the past de-

cade, three forms of NIR spectroscopy ~NIRS! thatuse pulsed-laser light in the time domain, amplitude-modulated laser light in the frequency domain, andcw light in a dc form were developed for blood oxy-genation quantification in tissue.7 Significant in-vestigations in both laboratory and clinical settingsby use of NIRS were conducted for noninvasive, quan-titative measurements and imaging of cerebraloxygenation8–12 and blood oxygenation of exercised

1 October 2000 y Vol. 39, No. 28 y APPLIED OPTICS 5231

13–17

piopa

hmH

oS

reO

c6

dtpie

5

muscle in vivo. Although NIR techniques wereused extensively in conjunction with cryospectropho-tometry to investigate tumor blood-vessel oxygen-ation in biopsies,18 only a few reports19–22 were

ublished on using the NIR techniques for monitor-ng tumor oxygenation in vivo. In principle, the the-retical model, i.e., the diffusion approximation to thehoton transport theory, works well for only largend homogeneous media.23,24 Accurate quantifica-

tion of tumor oxygenation by use of the NIR approachis limited because of the considerable heterogeneityand the finite sizes of tumors.

It is understood and documented25 that the NIRtechnique used for blood oxygenation monitoring issensitive to vascular absorption in the measured or-gan. The NIR method is not limited to measure-ments of blood oxygenation in arteries ~c.f., pulseoximetry! or in veins but interrogates blood in theentire vascular compartment, including capillaries,arterioles, and venules, i.e., the vascular bed. A va-riety of terms like cerebral oxygenation, tissue hemo-globin oxygenation, and mean hemoglobinoxygenation are used in the literature7,24,26 to indi-cate this concept. Although tissue hemoglobin oxy-genation is not rigorous because hemoglobinmolecules are located in only blood, the term is usedspecifically to differentiate between the hemoglobinsaturation in the tissue vascular bed, as measured bythe NIR method, and the arterial hemoglobin satu-ration SaO2, as measured by a pulse oximeter.

The goal of this paper is to demonstrate the NIRtechnique as a real-time, noninvasive means of mon-itoring hemoglobin oxygenation dynamics, i.e.,changes in the concentrations of total hemoglobin~Hbt! and oxygenated hemoglobin ~HbO2!, in the vas-cular bed of breast and prostate rat tumors in re-sponse to respiratory challenge. Compared withprevious NIR studies of tumors in vivo, our approach

as the following features: ~1! The transmissionode, as opposed to the reflectance mode used byull et al.,22 interrogates deeper regions ~central

parts! of the tumor. ~2! Only two wavelengths, aspposed to the spectrum of 300–1100 nm used byteen et al.,21 are employed and provide a fast and

low-cost instrument. ~3! A source–detector separa-tion of 1–2 cm interrogates a large tumor noninva-sively, as opposed to the needlelike probe used bySteinberg et al.20 More innovatively, on the basis ofthe experimental observation of tumor hemoglobinoxygenation dynamics, we developed a tumor he-moperfusion model that provides important insightinto tumor blood perfusion.

This paper is organized as follows: In Section 2,we describe our animal model, the NIR instrument,and the algorithm for calculations of tumor bloodoxygenation. In Section 3, we show experimentalresults measured from both breast and prostate tu-mors under respiratory interventions and calculatetime constants for the hemoglobin oxygenation dy-namics of the tumors. In Section 4, we develop atumor hemoperfusion model to interpret the experi-mental data obtained in the tumor-intervention stud-

232 APPLIED OPTICS y Vol. 39, No. 28 y 1 October 2000

ies and to relate the time constants to tumor bloodperfusion. Finally, in Section 5, we discuss the re-sults, the future extensions, and the potential uses ofthe NIR technique as a novel diagnostic–prognostictool for tumor therapy and cancer research.

2. Materials and Methods

A. Animal Model and Measurement Geometry

NF13762 breast tumor was implanted in adult fe-male Fisher rats, and Dunning prostate adenocarci-noma R3327-AT1 was implanted in adult maleCopenhagen rats. The tumors were grown inpedicles27 on the forebacks of the rats until the tu-mors were approximately 1–2 cm in diameter. Ratswere anesthetized with 200-ml ketamine hydrochlo-ide ~100 mgyml! and maintained under general gas-ous anesthesia with 33% inhaled O2 ~0.3 dm3ymin2, 0.6 dm3ymin N2O, and 0.5% methoxyflurane!

through a mask placed over the mouth and nose.Tumors were shaved to improve the optical contactfor transmitting light. Body temperature was main-tained with a warm-water blanket. In some cases, afiber-optic pulse oximeter ~Nonin, Inc., Model 8600V!that was manufacturer calibrated was placed on thehind foot to monitor arterial oxygenation SaO2, and afiber-optic probe was inserted rectally to measuretemperature. The tumor volume V ~in centimetersubed! was estimated as V 5 ~4py3! @~L 1 W 1 H!y#3, where L, W, and H are the three respective or-

thogonal dimensions.Most measurements were performed with 33% oxy-

gen as inhaled gas to achieve a stable baseline for aperiod of 5 to 15 min. The inhaled gas was thenswitched to carbogen ~95% oxygen, 5% carbon diox-ide! for at least 20 min and then switched back to 33%O2 for approximately 15 min. The complete cyclelasted 1 hour. Sometimes repeated carbogen inter-ventions were performed sequentially to evaluate thereproducibility of the time profiles of the tumors. Incertain cases alternative gases were used, as definedin the results and figures, and some rats were sacri-ficed by KCl-induced cardiac arrest.

Figure 1 shows the measurement geometry: Hor-izontally, the delivering and the detecting fiber bun-dles were face to face in the transmittance mode, andboth were in contact with the tumor surface withouthard compression. The separation of the two bundlesurfaces was between 1.0 and 2.5 cm, depending onthe tumor size. Vertically, the two bundle tips ~with

iameters of 0.5 cm! were placed around the middle ofhe tumor. Thus the current setup of the probesrovides an optimal geometry for the NIR light tonterrogate deep tumor tissue with minimal interfer-nce from the foreback of the rat.

B. Near-Infrared Instrument and Data Analysis

As shown in Fig. 1, we used a homodyne frequency-domain photon-migration system28,29 that was capa-ble of determining the amplitude and the phasechanges of amplitude-modulated light passingthrough tumors. In this setup a rf source modulates

du

tf

p

the light from two laser diodes ~wavelengths of 758and 782 nm! at 140 MHz. The laser light passesthrough a combined fiber-optic bundle, is transmittedthrough the tumor tissue, and is collected by a secondfiber bundle. The light is then detected by a photo-multiplier tube and demodulated with a commer-cially available in-phase and quadrature ~IQ!demodulator chip into its I and Q components. Afterthese components are put through a low-pass filterthey can be used to calculate the amplitude and thephase changes caused by the tumor. These steps areexpressed mathematically by

I~t! 5 2A sin~vt 1 u!sin~vt!

5 A cos~u! 2 A cos~vt 1 u!O¡low

passIdc

5 A cos~u!, (1)

Q~t! 5 2A sin~vt 1 u!cos~vt!

5 A sin~u! 1 A sin~vt 1 u!O¡low

passQdc

5 A sin~u!, (2)

u 5 tan21~QdcyIdc!, (3)

A 5 ~Idc2 1 Qdc

2!1y2, (4)

where A and u are the amplitude and the phase of theetected light, respectively, and v is the angular mod-lation frequency ~52p 3 140 MHz!.The two laser lights were time shared, and the con-

rolling process and the data acquisition both inter-aced through a 12-bit analog-to-digital board ~Real

Time Devices, Inc., Model AD2100! with a maximumsampling rate of 4 Hz.28 However, slower samplingrates were used in measurements to compensate forexperimental noise. Simple time averaging among afew adjacent data points was performed during dataanalysis to further decrease the noise. However, datasmoothing was not applied for ~1! calculating the ex-perimental uncertainty ~error bars! or ~2! fitting thetime constants to prevent the fast-changing compo-nent from being oversmoothed and overlooked. Thepulse-oximeter data were not averaged because theywere recorded manually and appear discrete compared

Fig. 1. Experimental setup of a one-channel, NIR, frequency-dphotomultiplier tube for detecting light; IQ, in-phase and quadratPass, low-pass filter. The 5-mm-diameter fiber bundles deliver an

with the NIR data. The experimental uncertaintiesfor arterial saturation and changes in hemoglobin con-centrations were calculated by use of the baseline datataken over 5–10 min without respiratory perturbationto the rat. Nonlinear curve fitting based on the Mar-quardt algorithm30,31 was performed by use of Kalei-daGraph.32 The software also provided the errors ~oruncertainties! for each fitted parameter, the optimizedx2 values, and the fitting correlation coefficient R, to-gether with the goodness of the fit R2.33 The signifi-cance of changes was assessed on the basis of Fisherprotected-least-significant-difference analysis of vari-ance by use of Statview software.

C. Calculation for Changes in the HemoglobinConcentration

It is well known that the NIRS of tissue can be usedto determine the total hemoglobin concentration Hbtand the hemoglobin oxygen saturation SO2 of an or-gan in vivo.7 When two NIR wavelengths are used~758 and 782 nm, in this case! it is assumed thattissue background absorbance is negligible and thatthe major chromophores in organs are oxygenatedand deoxygenated hemoglobin molecules. In princi-ple, because the IQ system can give both phase andamplitude values, we should be able to obtain abso-lute calculations of HbO2, Hb, and SO2.7,28 How-ever, given the tumor’s small size and large spatialheterogeneity, it is very difficult to obtain such ab-solute quantification accurately with conventionalalgorithms34 that are based on the diffusion approx-imation. Instead, on the basis of the modified Beer–Lambert law, we can use the amplitude of the lighttransmitted through the tumor to calculate concen-tration changes in HbO2, Hb, and Hbt ~expressed asDHbO2, DHb, DHbt, respectively! of the tumor thatare caused by respiratory intervention. Thesechanges can be derived25 and expressed as ~see Ap-

endix A for derivations and justifications!

DHb 5 Hb~transient! 2 Hb~baseline!

5

eHbO2

l1 logSAb

AtDl2

2 eHbO2

l2 logSAb

AtDl1

L~eHbl2 eHbO2

l1 2 eHbl1 eHbO2

l2 !, (5)

in IQ instrument for tumor oxygenation measurement. PMT,emodulator for retrieving amplitude and phase information; Lowtect the laser light through the tumor in transmittance geometry.

omaure dd de


a

itlpsoce

D

tpr

t

da

2n

1

5

DHbO2 5 HbO2~transient! 2 HbO2~baseline!

5

eHbl2 logSAb

AtDl1

2 eHbl1 logSAb

AtDl2

L~eHbl2 eHbO2

l1 2eHbl1 eHbO2

l2 !, (6)

where eHbl and eHbO2

l are extinction coefficients35 ofdeoxygenated and oxygenated hemoglobin, respec-tively, at wavelength l; the variable Ab is a constantmplitude of baseline; At is the transient amplitude

under measurement; and L is the optical path lengthbetween the source and the detector.

Using the approach suggested by Cope andDelpy,10 we can express L as L 5 DPF 3 d, where ds the direct source–detector separation in centime-ers and DPF is the ratio between the optical pathength and the physical separation and is tissue de-endent. The DPF for tumors has not been welltudied; for simplicity, we assume the DPF to be 1 inur calculations. The justification for this simplifi-ation is given in Section 5. After substituting thextinction coefficients35 at 758 and 782 nm in Eqs. ~5!

and ~6! with values of eHb758 5 0.359, eHbO2

758 5 0.1496, eHb782

5 0.265 and eHbO2

782 5 0.178, respectively, in units ofinverse millimoles times inverse centimeters, we ar-rive at

DHb [email protected] log~AbyAt!

758 2 6.17 log~AbyAt!782#

L,

(7)

HbO2 5

@210.92 log~AbyAt!758 1 14.80 log~AbyAt!

782#

L,

(8)

DHbt 5 D~HbO2 1 Hb!

[email protected] log~AbyAt!

758 1 8.63 log~AbyAt!782#

L,

(9)

where the units are in millimoles. Equations ~7!and ~8! permit the calculation of changes in Hb andHbO2 that are due to respiratory challenge, respec-tively, whereas Eq. ~9! quantifies a relative increasein the total hemoglobin concentration that is causedby the intervention. The last quantity also reflects achange in blood volume because it is proportional tothe total Hb concentration.

3. Results

A. Instrument Drift Tests

The stability of the NIR instrument was tested interms of baseline drift after a warm-up period of30 min by use of a tissue phantom25,36 with stableoptical properties. Figure 2 shows an example ofa phantom measurement that displays the variationof relative changes in apparent HbO2 and Hbt con-


centrations, as calculated from Eqs. ~8! and ~9!. Inhis example, the standard deviations over the entireeriod of 100 min were less than 0.007 and 0.004 mM,espectively, for DHbO2 and DHbt. Furthermore, we

calculated uncertainties for both of these quantitieson the basis of the propagation of errors, and theresults are consistent with those shown in Fig. 2.

B. Breast Tumors

Figure 3~a! shows the results taken from a breastumor ~4.5 cm3! with a source–detector separation of

1.8 cm. The data were smoothed, and the measure-ment uncertainties are shown at only discrete loca-tions. The figure shows the relative changes in totalhemoglobin concentration DHbt and oxygenated he-moglobin concentration DHbO2. The arterial Hbsaturation was also obtained to show a relativelyrapid change in arterial signals when the inhaled gaswas switched from 33% O2 to carbogen. Respiratorychallenge caused a sharp rise in DHbO2 ~p , 0.01after 1 min, p , 0.0001 by 1.5 min! that was followedby a further slow, gradual, but significant, increaseover the next 25 min ~p , 0.001!. DHbt also changedsignificantly ~p , 0.001! within the first minute, butthe total change was only approximately 10% of thatof DHbO2. Given the exponential appearance of therising part of DHbO2, we used single-exponential anddouble-exponential expressions to fit the data in therising portion to better understand and quantify thedynamic features of DHbO2. The unsmoothed dataand the fitted curves are shown in Figure 3~b!. The

ouble exponential appears to give a much better fit,s is confirmed by the respective R values ~0.98 ver-

sus 0.81!. Time constants of 0.18 6 0.02 min and7.8 6 3.9 min were obtained for fast and slow dy-amic changes, respectively, in the tumor HbO2 con-

centration.Figure 4~a! was obtained from a second breast tu-

mor ~5.9 cm3! with a source–detector separation of.6 cm. Here DHbO2 increased rapidly after the ini-

Fig. 2. Results of a drift test of the NIR instrument by use of atissue phantom. The thicker solid curve represents relativechanges in the oxygenated hemoglobin concentration, i.e., DHbO2,and the thinner solid curve represents relative changes in the totalhemoglobin concentration, i.e., DHbt. DHbO2 and DHbt were cal-culated by use of Eqs. ~8! and ~9!, respectively.

ec

2t

srD

5

o

fDcTr

1w0

f

tial gas switch but did not exhibit the continued slowrise afterward. DHbt was found to increase with car-bogen inhalation, although the magnitude wassmaller than that of DHbO2 during the period of theintervention. Again, changes in DHbO2 were mod-led by a single-exponential term that yielded a timeonstant of 2.00 6 0.04 min ~R 5 0.97! and by a

double-exponential formula with two time constantsof 0.8 6 0.2 min and 3.0 6 0.3 min ~R 5 0.98!. Inthis case both expressions fit the data well, as shownin Fig. 4~b!.

To demonstrate the reproducibility of the dynamicchanges in response to respiratory challenge, we sub-jected one animal to repeat carbogen inhalation.Figure 5~a! shows measurements taken from a breasttumor ~6.7 cm3! with a source–detector separation of

cm. In this case air with 1.2% isoflurane ~anes-hetic! was used as the baseline instead of 33% O2.

This figure shows a very consistent pattern in tworepeated time responses with a fast and a slow in-crease in DHbO2. Here DHbt shows a similar dy-

Fig. 3. ~a! Results obtained with the NIR instrument from a4.5-cm3 rat breast tumor while the breathing gas was switchedrom 33% O2 to carbogen. The thicker solid curve representsHbO2, the thinner solid curve represents DHbt, and the dashedurve with the filled circles represents arterial saturation. ~b!he unsmoothed data and the fitted curves: The solid curvesepresent the best fits to the DHbO2 data at the rising portion.

The best fit to the two exponential terms is 0.143$1 2 exp@2~t 22.5!y0.18#% 1 0.36$1 2 exp@2~t 2 12.5!y27.8#%, with R 5 0.98,hereas the best fit with one exponential term is expressed as.322$1 2 exp@2~t 2 12.5!y5.1#%, with R 5 0.81.

namic pattern, i.e., a rapid rise followed by a slowcontinuation. Figures 5~b! and 5~c! show the un-moothed data together with the fitted curves for theising portions of the two repeated increases inHbO2. Again, the double-exponential expression

with two time constants produced much better fitsthan did the single-exponential term in both pro-cesses with two averaged time constants of t1 ~mean!

0.26 6 0.11 min and t2 ~mean! 5 8.2 6 1.8 min.Individual, respective time constants and coefficientsare summarized in Table 1. Furthermore, single-exponential and double-exponential expressions werefitted to obtain time constants for the decay processesafter the inhaled gas was switched repeatedly back tothe baseline conditions. Similarly, the double-exponential expression fits the data better with twomean time constants of t1

decay ~mean! 5 0.17 6 0.07min and t2

decay ~mean! 5 12.2 6 0.7 min for the twodecay processes.

To further validate our experimental observations,we subjected some rats to cardiac arrest ~with KCl! tobserve the changes in HbO2 and Hbt on death. Fig-

ure 6 shows an example of cardiac arrest on a rat with

Fig. 4. ~a! Results obtained with the NIR instrument from a5.9-cm3 rat breast tumor while the breathing gas was switchedrom 33% O2 to carbogen. ~b! The solid curves show the best fits

to the HbO2 data at the rising portion. In this case a doubleexponential with values of 0.09$1 2 exp@2~t 2 9.8!y0.8#% 10.16$1 2 exp@2~t 2 9.8!y3.0#%, with R 5 0.98, and a single expo-nential with a value of 0.250$1 2 exp@2~t 2 9.8!y2.00#%, with R 50.97, provided similarly good fits.


fisro

Tab

le1.

Sum

mar

yo

fth

eV

ascu

lar

Oxy

gen

Dyn

amic

s

Tum

orD

oubl

e-E

xpon

enti

alF

itti

ngD

HbO

25

A1@1

2ex

p~2

tyt 2

!#1

A2@1

2ex

p~2

tyt 2

!#S

ingl

e-E

xpon

enti

alF

itti

ngD

HbO

2

5A

1@1

2ex

p~2

tyt!

#

Typ

eV

olum

e~c

m3!

t 1~m

in!

t 2~m

in!

A1

~mM

!A

2~m

M!

R2

t 1yt

2

g1

g2

5A

1

A2

f 1 f 25

A1y

A2

t 1yt

2t

~min

!A

~mM

!R

2

Bre

ast

~Fig

.3!

4.5

1.8

60.

0227

.86

3.9

0.14

36

0.00

30.

366

0.03

0.96

0.00

66

0.00

10.

406

0.03

61.3

56

0.01

5.1

60.

30.

322

60.

005

0.66

Bre

ast

~Fig

.4!

5.9

0.8

60.

23.

06

0.3

0.09

60.

020.

166

0.02

0.96

0.27

60.

070.

566

0.14

2.11

60.

182.

006

0.04

0.25

06

0.00

10.

94B

reas

t~F

ig.5

!6.

70.

186

0.01

6.93

60.

090.

232

60.

002

0.36

86

0.00

20.

970.

026

60.

001

0.63

60.

0124

.27

60.

012.

806

0.03

0.55

06

0.00

10.

806.

70.

332

60.

003

9.47

60.

120.

321

60.

001

0.23

36

0.00

10.

990.

035

60.

001

1.38

60.

0139

.30

60.

011.

386

0.02

0.48

56

0.00

10.

66P

rost

ate

~Fig

.7!

8.2

0.26

56

0.00

76.

026

0.15

0.09

06

0.00

10.

064

60.

001

0.93

0.04

46

0.00

11.

416

0.03

31.9

56

0.01

1.13

60.

020.

140

60.

001

0.67

Pro

stat

e~F

ig.8

!10

.80.

36

4.5

15.6

63.

10.

004

60.

010.

386

0.03

0.94

0.02

60.

200.

016

0.03

0.6

61.

214

.86

1.6

0.37

60.

020.

94

5

Fig. 5. ~a! Relative changes in the HbO2 detected with the NIRinstrument from a rat breast tumor ~6.7 cm3! while the breathinggas was alternated between air ~21% O2! and carbogen. The best

ts to the HbO2 data by use of both the double-exponential and theingle-exponential expressions for ~b! the first and ~c! the secondespiratory challenges are shown. The fitted equations that werebtained from ~b! are 0.232$1 2 exp@2~t 2 19.5!y0.18#% 1 0.368$1 2

exp@2~t 2 19.5!y6.93#%, with R 5 0.98, and 0.550$1 2 exp@2~t 219.5!y2.80#%, with R 5 0.89, respectively. The fitted equationsthat were obtained from ~c! were 0.321$1 2 exp@2~t 2 67!y0.332#%1 0.233$1 2 exp@2~t 2 67!y9.47#%, with R 5 0.99, and 0.485$1 2exp@2~t 2 67!y1.38#%, with R 5 0.81, respectively.


~D

D

ie

de

wmlt

tdK

b

t

a breast tumor ~5.3 cm3!. Both DHbt and DHbO2dropped significantly, immediately after KCL wasadmitted intravenously. Within 1 min DHbtreached a plateau, whereas DHbO2 decreased rapidlywithin the first 30 s and then was followed by a slowprolongation.

C. Prostate Tumors

Figure 7~a! was obtained from a large prostate tumor8.2 cm3!. In common with the breast tumors,HbO2 showed a rapid initial increase that was fol-

lowed by a slower continuation. DHbt increased rap-idly and then reached a plateau. Figure 7~b! showsthat the double-exponential equation fits the un-smoothed data better ~R 5 0.96! than does the single-exponential term ~R 5 0.82!. Here the fast and theslow time constants are 0.265 6 0.007 min and 6.02 60.15 min, respectively.

Figure 8~a! was obtained from another large pros-tate tumor ~10.8 cm3 with a source–detector separa-tion of 2.5 cm!. Here DHbO2 displayed a gradualincrease throughout the entire period of carbogen in-halation, whereas the increase in DHbt was consid-erably delayed. Variations in arterial hemoglobinsaturation SaO2 are also shown and were very rapidin comparison with DHbO2, in common with Fig. 3.

HbO2 dropped rapidly when the inhaled gas wasswitched back from carbogen to 33% O2. Both thesingle-exponential and the double-exponential ex-pressions were used to obtain time constants for therising portion of DHbO2 that was due to carbogenntervention. In this case both expressions gavequally good fits, as shown in Fig. 8~b! and Table 1.

For the decay process, we obtained t1decay 5 0.6 6 0.2

min and t2decay 5 6.6 6 1.7 min with R 5 0.94 for the

ouble-exponential fitting, whereas the single-xponential fitting resulted in t 5 2.8 6 0.4 min with

R 5 0.88. For comparison the rat was also chal-lenged with 100% O2.

In summary, we observed dynamic changes inHbO2 that were due to carbogen intervention for bothbreast and prostate tumors. In most cases thesechanges were modeled better by a double-exponential

Fig. 6. Influence of KCl-induced cardiac arrest on the values ofHbO2 and Hbt of a breast tumor ~5.3 cm3!, while the rat wasreathing air.

expression with a fast and a slow time constant thanthey were by a single-exponential fitting. Dynamicchanges in arterial saturation preceded those inHbO2. The detailed parameters regarding tumorsize, fitted time constants, corresponding magni-tudes, and R2 are listed in Table 1.

4. Model for the Blood Oxygenation Dynamics ofTumors

As was shown in Section 3, the temporal changes inHbO2 caused by respiratory challenge can be fittedwith an exponential equation that has either one ortwo time constants ~fast and slow!. In this section,

e further derive and simplify a hemoperfusionodel to interpret these time constants and to corre-

ate the experimental findings with the physiology ofhe tumors.

To develop the model, we follow an approach usedo measure regional cerebral blood flow ~rCBF! withiffusible radiotracers, as originally developed byety37 in the 1950’s. The basic model was modified

in a variety of ways to adapt it to positron emissiontomography studies.38,39 By analogy, we can evalu-ate tumor hemodynamics such as tumor blood flow~perfusion! by using the respiratory-intervention gasas a tracer.

Fig. 7. ~a! Influence of respiratory challenges ~switching from airo carbogen! on the values of HbO2 and Hbt of a large rat prostate

tumor ~8.2 cm3!. ~b! The best-fitted equations are 0.090$1 2exp@2~t 2 12!y0.265#% 1 0.064$1 2 exp@2~t 2 12!y6.02#%, with R 50.96, and 0.140$1 2 exp@2~t 2 12!y1.13#%, with R 5 0.82, for thedouble-exponential and the single-exponential expressions, respec-tively.


t

at

Tb

ohD

t

pe

tp

a

5

In general, the Fick principle can be stated as fol-lows38: The rate of change of tracer concentration ina regional area of an organ equals the rate at whichthe tracer is transported to the organ in the arterialcirculation minus the rate at which it is carried awayinto the venous drainage, i.e.,

dCt

dt5 f ~Ca 2 Cv!, (10a)

where f is the blood flow ~or perfusion rate!, Ct is theracer concentration in tissue, and Ca and Cv are the

time-varying tracer concentrations in the arterial in-put and the venous drainage, respectively. Ca canbe measured from a peripheral artery, but Cv is rel-atively difficult to obtain regionally. Therefore abrain–blood partition coefficient, l 5 CtyCv, was de-veloped by Kety, and Eq. ~10a! becomes37

dCt

dt5 fSCa 2

Ct

l D . (10b)

Fig. 8. ~a! Variations in SaO2, DHbO2, and Hbt of which the latterwo were detected with the NIR instrument, from a large ratrostate tumor ~10.8 cm3! during respiratory challenge. ~b! The

solid curves represent the best fits to the DHbO2 data at the risingportion during carbogen inhalation. The fitted equations are0.004$1 2 exp@2~t 2 21.5!y0.3#% 1 0.38$1 2 exp@2~t 2 21.5!y15.6#%nd 0.37$1 2 exp@2~t 2 21.5!y14.8#% for the double-exponential and

the single-exponential expressions, respectively, with R 5 0.97 forboth.


In Eq. ~10b!, f and l are constants, whereas Ct is atime-dependent variable that is written as Ct~t!. Inprinciple, the arterial-tracer concentration Ca is atime-varying quantity. If a certain concentration ofthe arterial tracer is administrated continuouslystarting at time 0, Ca can be expressed mathemati-cally as a constant value of Ca~0! after time 0. ThenEq. ~10b! can be solved as

Ct~t! 5 lCa~0!@1 2 exp~2ftyl!#. (11)

Equation ~11! indicates that, at time t after the onsetof tracer administration, the local tissue ~traditional-ly brain! Ct~t! concentration depends on the bloodflow f, the arterial time–activity curve Ca~0!, and thepartition coefficient l.

In response to respiratory intervention, a suddensmall change is introduced into the arterial O2 satu-ration SaO2, and the resulting increase in arterialHbO2 concentration ~DHbO2

artery! can be consideredas an intravascular tracer.40 Following Kety’smethod and assuming that changes in dissolved O2are negligible,40 we have

ddt

~DHbO2vasculature! 5

fSDHbO2artery 2

DHbO2vasculature

g D .

(12)

where f still represents blood flow ~or perfusion rate!nd g is defined as a vasculature coefficient of theumor. The coefficient g is the ratio of the HbO2

concentration change in the vascular bed to that inveins and equals ~DHbO2

vasculature!y~DHbO2vein!.

his definition implies that a change in the venouslood oxygenation DHbO2

vein is proportional to achange in the Hb oxygenation in the vascular bed,DHbO2

vasculature.In Eq. ~12!, f and g are constants, whereas

DHbO2vasculature is a time-dependent variable. By

analogy to Eq. ~11!, DHbO2vasculature can be solved

rigorously given a constant input H0 for DHbO2artery

after time 0. Our data ~Figs. 3 and 8! demonstratethat changes in the arterial HbO2 ~SaO2! are muchfaster than in the vascular bed. Then solving Eq.~12! leads to

DHbO2vasculature~t! 5 gH0@1 2 exp~2ftyg!#. (13)

Equation ~13! indicates that, at time t after the onsetf respiratory intervention, the change in oxygenatedemoglobin concentration in the tumor vasculatureHbO2

vasculature ~t! depends on the blood perfusionrate f, the arterial oxygenation input H0, and thevasculature coefficient of the tumor g.

As indicated by Eq. ~8!, our NIR instrument is ableo measure an increase in the vascular HbO2 concen-

tration DHbO2vasculature. Equation ~13! gives an ex-

onential of the same form as that used to fit ourxperimental data, indicating that the measured

irCe

psecusemot

tvtD

csa

sfi

time constant is associated with the blood perfusionrate f and the vasculature coefficient g of the tumor inthe measured area. If the measured volume in-volves two distinct regions, we then involve with twodifferent blood-perfusion rates f1 and f2, two differentvasculature coefficients g1 and g2, or all four. Heret is reasonable to assume that the measured signalesults from both regions, as illustrated in Fig. 9.onsequently, Eq. ~13! can be modified with a double-xponential expression and two time constants as

DHbO2vasculature~t! 5 g1 H0@1 2 exp~2f1 tyg1!#

1 g2 H0@1 2 exp~2f2 tyg2!#

5 A1@1 2 exp~2f1 tyg1!#

1 A2@1 2 exp~2f1 tyg2!#, (14)

where f1 and g1 are the blood-perfusion rate and thevasculature coefficient, respectively, in region 1, f2and g2 are the same for region 2, A1 5 g1H0, and A2 5g2H0. The two time constants are equal to t1 5 g1yf1and t1 5 g2yf2. Then, if A1, A2, and the two timeconstants are determined from our measurements,we arrive at the ratios for the two vasculature coef-ficients and the two blood-perfusion rates:

g1

g25

A1

A2,

f1

f25

A1yA2

t1yt2. (15)

With these two ratios, we can obtain insight intothe tumor vasculature and blood perfusion. For ex-ample, a ratio of g1yg2 near 1 from a measurementimplies that the vascular structure of the measuredtumor volume is rather uniform. Then the coexist-ence of two time constants reveals two mechanisms ofregional blood perfusion in the tumor. A large time

Fig. 9. Schematic diagram showing ~1! a tumor model with twovascular perfusion regions, ~2! the source and the detector fibersand their geometry with respect to the tumor model, and ~3! thelight patterns that propagate in the tumor tissue. A represents aportion of the detected signal, which interrogates the well-perfusedregion, and B represents another portion of the detected signal,which passes mainly through the poorly perfused region. Wehave assumed that the total detected signal is the sum of A and B.

constant implies slow perfusion through a poorly per-fused area, whereas a small time constant indicatesfast perfusion through a well-perfused area. In themeantime, the ratio of the perfusion rates in thesetwo areas can also be obtained quantitatively. Fur-thermore, a ratio of g1yg2 . 1 ~i.e., A1yA2 . 1! meansthat the measured signal results more from region 1than from region 2 within the measured tumor vol-ume. Therefore, by studying tumor blood oxygen-ation dynamics and obtaining time constantstogether with their amplitudes, we can gain impor-tant information on regional blood perfusion and vas-cular structures of the tumor within the measuredvolume.

Our experimental data ~Table 1! reveal that all themeasurements can be fitted with the double-exponential model equivalently to or better than thesingle-exponential fitting. Ratios of t1yt2, g1yg2,and f1yf2 are also shown in Table 1 for respectivecases.

5. Discussion

Using NIRS, we have measured relative changes inHbt and HbO2 in breast and prostate rat tumors inresponse to respiratory intervention. We have ob-served that respiratory challenge caused the HbO2concentration to rise promptly and significantly inboth breast and prostate tumors but that the totalconcentration of hemoglobin sometimes increasedand sometimes remained unchanged. The dynamicchanges of tumor oxygenation can be modeled by ei-ther one exponential term with a slow time constantor two exponential terms with fast and slow timeconstants. This relation suggests that there may betwo vascular mechanisms in the tumor that are de-tected by the NIRS measurement. As indicated byEqs. ~13! and ~14!, these time constants are inversely

roportional to the blood-perfusion rates of the mea-ured volumes of the tumors. Based on the double-xponential model, determination of the two timeonstants and their corresponding amplitudes allowss to determine the relations between the two perfu-ion rates and between the vascular structures, asxpressed in Eq. ~15!. Further investigation withore measured quantities may lead to quantification

f each parameter individually by use of the NIRechnique.

To develop a model for interpreting the NIR dataaken during carbogen inhalation, we have defined aasculature coefficient g. It is a proportionality fac-or between DHbO2

vein and DHbO2vasculature, i.e.,

HbO2vein 5 DHbO2

vasculatureyg. We expect that gdepends on ~1! the oxygen consumption and ~2! theapillary density of the tumor. If the oxygen con-umption, the capillary density, or both of the tumorre large, changes in the venous HbO2 concentration

will be small; if the oxygen consumption, the capillarydensity, or both of the tumor are small, changes in thevenous HbO2 concentration will be large. Furthertudies are necessary to learn more about this coef-cient and to confirm our speculation.Our current NIR system allows us to quantify the


H

t2tumetm

19

s

sD

5

ratio of gyf by using the single-exponential model orto quantify the ratios of g1yg2 and f1yf2 by using thedouble-exponential model. We can obtain impor-tant information on the blood perfusion of the tumor:a large time constant usually represents slow bloodperfusion, whereas a small time constant indicatesfast blood perfusion. The coexistence of two timeconstants implies a combination of well-perfused andpoorly perfused mechanisms of blood perfusion. In-deed, some tumor lines have only 20% to 85% of ves-sels perfused.41 Furthermore, tumor structures andoxygen distribution6,42 are highly heterogeneous.Therefore it is likely that our measurement detects awell-perfused region, or a poorly perfused region, or amixture of both in the tumor, depending on the posi-tion or the location of the source and the detector ofthe NIR instrument.

Hull et al.22 reported carbogen-induced changes inrat mammary tumor oxygenation by using spatiallyresolved NIRS. To compare our results to theirs, weapplied our curve-fitting procedure to their publishedhemoglobin saturation curve and obtained a timeconstant of 0.27 min ~or 16 s! for the rising edge,which is consistent with our fast component. Theirdata do not show a slow component, suggesting thattheir measurement was dominated by active tumorvasculature. This difference may be explained asfollows:

~1! The tumor volume mentioned in the paper byull et al.22 was approximately 1.5 cm3, much

smaller than the volumes of the tumors that we mea-sured ~in Figs. 3 to 8, the largest tumor was 10.8 cm3,whereas the smallest tumor was 4.5 cm3!.

~2! Their measurement was in reflectance geome-ry with multiple detectors located at distances 1 to0 mm away from the source, and in the calculationhe tumor was assumed to be homogenous in order tose diffusion theory. Thus their measurement wasore sensitive to the superficial area of the tumor,

mphasizing the tumor periphery, which is often bet-er vascularized than the central part of the tu-or.42,43

The fast component observed in our tumors is con-sistent with the rapid changes detected in SaO2 in theleg as measured by use of the pulse oximeter, provid-ing further evidence that this relates to the well-vascularized, highly perfused region of the tumor.

The data shown in Fig. 8 in this paper are repre-sentative of the measurement of a poorly perfusedregion in which the measured tumor was large andthe portion for the fast oxygenation response wassmall. The data given in Figs. 3–5 and 7 resultedfrom a mixture of well-perfused and poorly perfusedareas in the tumors and exhibited a mixture of fastand slow oxygenation responses to hyperoxygen con-ditions. It is reasonable to expect that larger tumorshave more poorly perfused regions than do smallertumors. The time constant of the slow componentobserved here approaches that observed previouslyfor changes in tissue pO2 in an AT1 prostate tumor


measured by use of F nuclear magnetic resonancespectroscopy to interrogate interstitial oxygenation.44

In general, well-oxygenated tumor regions had alarge and rapid response to respiratory challenge,whereas poorly oxygenated regions were much moresluggish.45 One would indeed expect changes invascular oxygenation to precede changes in the tis-sue, and combined investigations by NIRS and nu-clear magnetic resonance spectroscopy in the futurewill provide further insight into the delivery of oxy-gen to tumors.

Dynamic changes in vascular oxygenation havebeen assessed previously by several other techniques.Following the infusion of Green 2W dye intrave-nously into EMT-6 tumor-bearing mice, Vinogradovet al.4 were able to image changes in the surfacevascular pO2. On switching from air to carbogeninhalation, they observed a very rapid increase in pO2with a rate similar to the fast component, which wehave seen here. Although the phosphorescencemethod provides vascular pO2, NIR methods gener-ally provide HbO2 or SO2 because there is some un-certainty in the local affinity of hemoglobin for tumoroxygen: the pO2–SO2 dissociation curve is subject topH, temperature, and other allosteric effectors, suchas 2,3-diphosphoglycerate in the immediate milieu.A promising new approach is the blood-oxygen-level–dependent ~BOLD! contrast 1H MRI, which is sensi-tive to vascular perturbations. Robinson et al.46

explored the response to respiratory challenge in var-ious tumors and showed reversible regional changeson switching from air to carbogen inhalation. Incommon with our NIR data, their changes were oftenbiphasic with a large change occurring within thefirst 2 min and followed by slower increases.46 How-ever, interpreting the BOLD MRI results is compli-cated by variations in vascular volume and flow, andthere is no direct measure of HbO2 in tumors.

The time constants are not source–detector sepa-ration sensitive. Equations ~8! and ~9! have demon-trated that DHbO2 and DHbt are proportional to 1yd,

where d is the source–detector separation. This re-lation indicates that a different d value will onlytretch or compress the entire temporal profile ofHbO2, but it does not change the transient behavior

of the time response. The same argument can applyto the DPF. In this study, we have assumed a DPFratio of 1 for simplicity. If the DPF value is largerthan 1, the values of DHbO2 and DHbt will decreaseby a factor of DPF. But this modification does notaffect the time constants t1 and t2, which constitutethe dynamic responses of DHbO2 of the tumors torespiratory intervention.

Given the evidence for intratumoral heterogeneityfrom MRI6,46 and histology,47 we believe it will beimportant to advance our NIR system to have multi-ple sources, multiple detectors, or both to study notonly dynamic but also spatial aspects of blood oxy-genation in tumor vasculature. Nonetheless, we be-lieve the preliminary results described here are aproof of principle for the technique, laying a founda-tion for more extensive tests to correlate tumor size

p

bptmfi

w

ot

b

tB

o

ci

D

c

with the rates of change of HbO2 and Hbt with respectto respiratory challenge. Although Hull et al.22 andwe have focused on respiratory challenge, we notethat previous NIR studies of tumors also examinedthe influence of chemotherapy,21 pentobarbitaloverdose,21 ischemic clamping,20 and infusion of

erfluorocarbon blood substitute.19 These studiesdemonstrate the potential versatility of the NIR ap-proach and its application for diverse future studies.

In summary, we have demonstrated that the NIRtechnology can provide an efficient, real-time, nonin-vasive means for monitoring vascular oxygenationdynamics in tumors during hyperoxygen respiratorychallenge. HbO2 concentrations measured fromoth breast and prostate tumors often exhibit arompt rise that is followed by a gradual persistencehroughout the intervention. By developing a he-operfusion model with two exponential terms andtting the model to the increased HbO2 data, we are

able to recognize two mechanisms for blood perfusionin the tumor and to quantify the ratios of the twoperfusion rates and those of the two vasculature co-efficients. Thus the technique can enhance our un-derstanding of the dynamics of tumor oxygenationand the mechanisms of tumor physiology under base-line and perturbed conditions. Moreover, it appearsthat the NIRS may have a great potential for moni-toring tumor angiogenesis because the method canprovide information on blood perfusion and oxygenconsumption of the measured tumor.

Appendix A

A. Derivation

It has been shown that, in the NIR range, the majorlight absorbers in tissue are oxygenated and deoxy-genated hemoglobin molecules.24,25 With thisknowledge, the absorption coefficients ~in inversecentimeters! at two wavelengths can be associatedwith the concentrations of HbO2 and Hb by

mal1 5 eHbl1 Hb 1 eHbO2

l1 HbO2, (A1)

mal2 5 eHbl2 Hb 1 eHbO2

l2 HbO2, (A2)

here eHbl , and eHbO2

l are the extinction coefficients ~ininverse centimeters times inverse millimoles! of de-xygenated and oxygenated hemoglobin, respec-ively, at wavelength l, and HbO2 and Hb are the

oxyhemoglobin and the deoxyhemoglobin concentra-tions. Because eHb

l and eHbO2

l are constants, changesin HbO2 and Hb in tissue vasculature result inchanges in ma

l. In turn, changes in HbO2 and Hb cane determined by measuring changes in ma

l at twowavelengths and can be expressed as

DHb 5 Hb~transient! 2 Hb~baseline!

5eHbO2

l1 Dmal2 2 eHbO2

l2 Dmal1

eHbl2 eHbO2

l1 2 eHbl1 eHbO2

l2, (A3)

DHbO2 5 HbO2~transient! 2 HbO2~baseline!

5eHb

l2 Dmal1 2 eHb

l1 Dmal2

eHbl2 eHbO2

l1 2 eHbl1 eHbO2

l2, (A4)

where DHb and DHbO2 refer to the change in thedeoxyhemoglobin and the oxyhemoglobin concentra-tions between the baseline condition and the tran-sient, or perturbed, condition, respectively, and Dma

l

represents the change in the absorption coefficient atl relative to the baseline condition. However, ourcurrent experimental setup with one source and onedetector does not provide adequate information forquantifying ma

l values for a solid rat tumor. Thus weake an approximate approach by using the modifiedeer–Lambert relation to calculate DHb and DHbO2.According to the modified Beer–Lambert law,10 an

ptical density ~OD! can be defined as OD 5 log~I0yI!5 maL, where I0 and I are the incident and the de-tected optical intensities, respectively, and L is theoptical path length traveled by light inside the tissue.When an organ or a tumor undergoes a change fromits baseline condition to a transient condition underphysiological perturbations, a change in the OD atwavelength l will occur and can be expressed as

DODl 5 ODl ~transient! 2 ODl ~baseline!

5 log~IbyIt!l 5 Dma

lLl, (A5)

where Ib and It are measured optical intensities un-der baseline and transient conditions, respectively.Thus we arrive at

Dmal 5 log~IbyIt!

lyLl. (A6)

With our current NIR instrument, we can obtain theratios of ~IbyIt!

l1 and ~IbyIt!l2 from the tumor mea-

surement. By assuming a constant path length, i.e.,Ll1 ~baseline! ' Ll2 ~baseline! ' Ll1 ~transient! ' Ll2

~transient! ' L, we next substitute Eq. ~A6! into Eqs.~A3! and ~A4! and arrive at Eqs. ~5! and ~6! for cal-ulations of tumor hemoglobin oxygenation dynam-cs. Note that Ib, It have been replaced with Ab, At in

Eqs. ~5! and ~6!.

B. Justification

The assumption of a constant path length as givenabove makes it possible to use relatively simple equa-tions, for example, Eqs. ~5! and ~6!, to quantify the

Hb and the DHbO2 of tumors under respiratory in-tervention. However, in principle, the optical pathlength L through tissue is wavelength dependent andcould be variable under physiological perturbations.Therefore it is useful to know whether the relativeerror for calculated DHb and DHbO2 caused by thisassumption is within a reasonable range.

According to the diffusion approximation, the opti-cal path length L of the NIR light traveling in tissuean be expressed approximately as24

L 5Î32

d Sm9smaD1y2

, (A7)


1

5

where d is the source–detector separation and ma andm9s are the absorption and the reduced scattering co-efficients, respectively. Because in the NIR regionthe m9s of tissue is not sensitive to either wavelengthor perturbation, we assume that a change in L resultsfrom only a change in ma, which is both wavelengthand perturbation dependent. With this assumption,Eq. ~A7! leads to

DLL

5 21Dma

2ma. (A8)

Equation ~A8! allows us to determine the relativeerrors of DLyL that are caused by ~a! the wavelengthdependence of ma and ~b! the perturbation depen-dence of ma in the tumor. For case ~a!, we calculatedthis error by using

ma758 2 ma

782

2ma758

under the baseline and the perturbed conditions; forcase ~b!, we employed

mal ~transient! 2 ma

l ~baseline!

2mal ~baseline!

at both l 5 758 nm and l 5 782 nm for the errorcalculation. The ma values used here were takenfrom Hull et al.22 Although the rat tumor used intheir study was different from ours, the absorptioncoefficients of the tumors should be in a similar orderand follow a similar dynamic trend. The calculationshows that, with 758 and 782 nm under carbogenperturbation, the maximum value of DLyL is 12%.This result implies that the assumption of a constantpath length that was used for Eqs. ~5! and ~6! givesrise to a maximal relative error of 12% in L.

On the basis of Eqs. ~5! and ~6! @or Eqs. ~7! and ~8!#,we arrive at DXyX 5 2DLyL, where X can be DHbO2,DHb, or DHbt. Thus the assumption of a constantpath length leads to a maximal relative error of 12%for the magnitude of the changes that we detectedwith regard to respiratory challenge. Although 12%is not completely negligible, the measurement andthe calculation with the assumption of a constantpath length are still worthwhile. Such an approachmakes it possible, as a first-order approximation, toquantify the DHbt and the DHbO2 of tumors underrespiratory intervention, providing deep insightinto tumor vascular phenomena and mechanismsof modulating tumor physiology for therapeuticenhancement.

This study was supported in part by The WhitakerFoundation RG-97-0083 ~H. Liu!, The American Can-cer Society RPG-97-116-010CCE ~R. P. Mason!, andThe Department of Defense Breast Cancer InitiativeBC962357 ~Y. Song!. We thank Peter P. Antich andEric W. Hahn for their continued collegial support.


References1. L. Gray, A. Conger, M. Ebert, S. Hornsey, and O. Scott, “The

concentration of oxygen dissolved in tissues at time of irradi-ation as a factor in radio-therapy,” Br. J. Radiol. 26, 638–648~1953!.

2. M. Hockel, K. Schlenger, B. Aral, M. Mitze, U. Schaffer, and P.Vaupel, “Association between tumor hypoxia and malignantprogression in advanced cancer of the uterine cervix,” CancerRes. 56, 4509–4515 ~1996!.

3. T. Y. Reynolds, S. Rockwell, and P. M. Glazer, “Genetic insta-bility induced by the tumor microenvironment,” Cancer Res.56, 5754–5757 ~1996!.

4. S. A. Vinogradov, L.-W. Lo, W. T. Jenkins, S. M. Evans, C.Koch, and D. F. Wilson, “Noninvasive imaging of the distribu-tion of oxygen in tissues in vivo using near-infrared phos-phors,” Biophys. J. 70, 1609–1617 ~1996!.

5. J. A. O’Hara, F. Goda, K. J. Liu, G. Bacic, P. J. Hoopes, andH. M. Swartz, “The pO2 in a murine tumor after irradiation:an in vivo electron paramagnetic resonance oximetry study,”Radiat. Res. 144, 222–229 ~1995!.

6. R. P. Mason, A. Constantinescu, S. Hunjan, D. Le, E. W. Hahn,P. P. Antich, C. Blum, and P. Peschke, “Regional tumor oxy-genation and measurement of dynamic changes,” Radiat. Res.152, 239–245 ~1999!.

7. D. T. Delpy and M. Cope, “Quantification in tissue near-infrared spectroscopy,” Philos. Trans. R. Soc. London B 952,649–659 ~1997!.

8. M. Fabiani, G. Gratton, and P. M. Corballis, “Noninvasivenear-infrared optical imaging of human brain function withsubsecond temporal resolution,” J. Biomed. Opt. 1, 387–398~1996!.

9. R. Wenzel, H. Obrig, J. Ruben, K. Villringer, A. Thiel, J. Ber-narding, U. Dirnagl, and A. Villringer, “Cerebral blood oxygen-ation changes induced by visual stimulation in humans,”J. Biomed. Opt. 1, 399–404 ~1996!.

0. M. Cope and D. T. Delpy, “A system for long-term measure-ment of cerebral blood and tissue oxygenation in newborninfants by near-infrared transillumination,” Med. Biol. Eng.Comput. 26, 289–294 ~1988!.

11. B. Chance, E. Anday, S. Nioka, S. Zhou, L. Hong, K. Worden,C. Li, T. Murray, Y. Ovetsky, D. Pidikiti, and R. Thomas, “Anovel method for fast imaging of brain function, noninvasively,with light,” Opt. Express 2, 411–423 ~1998!, http:yywww.epubs.osa.orgyopticsexpress.

12. A. M. Siegel, J. A. Marota, J. Mandeville, B. Rosen, and D. A.Boas, “Diffuse optical tomography of rat brain function,” inOptical Tomography and Spectroscopy of Tissue III, B. Chance,R. R. Alfano, and B. J. Tromberg, eds., Proc. SPIE 3597, 252–261 ~1999!.

13. S. Homma, T. Fukunaga, and A. Kagaya, “Influence of adiposetissue thickness on near-infrared spectroscopic signals in themeasurement of human muscle,” J. Biomed. Opt. 1, 418–424~1996!.

14. M. Ferrari, Q. Wei, L. Carraresi, R. A. De Blasi, and G. Zac-canti, “Time-resolved spectroscopy of the human forearm,” J.Photochem. Photobiol. B: Biol. 16, 141–153 ~1992!.

15. M. Ferrari, R. A. De Blasi, S. Fantini, M. A. Franceschini, B.Barbieri, V. Quaresima, and E. Gratton, “Cerebral and muscleoxygen saturation measurement by a frequency-domain near-infrared spectroscopic technique,” in Optical Tomography,Photon Migration, and Spectroscopy of Tissue and ModelMedia: Theory, Human Studies, and Instrumentation, B.Chance and R. R. Alfano, eds., Proc. SPIE 2389, 868–874~1995!.

16. H. Long, G. Lech, S. Nioka, S. Zhou, and B. Chance, “CWimaging of human muscle using near-infrared spectroscopy,”in Advances in Optical Imaging and Photon Migration, J. G.Fujimoto and M. S. Patterson, eds., Vol. 21 of OSA Trends in

Optics and Photonics Series ~Optical Society of America, 33. J. L. Hintze, NCSS, Version 6.0, User’s Guide II: Statistical

4

4

4

Washington, D.C., 1998!, pp. 256–259.17. B. Chance, S. Nioka, J. Kent, K. McCully, M. Fountain, R.

Greenfield, and G. Holtom, “Time-resolved spectroscopy ofhemoglobin and myoglobin in resting and ischemic muscle,”Anal. Biochem. 174, 698–707 ~1988!.

18. B. M. Fenton, S. F. Paoni, J. Lee, C. J. Koch, and E. M. Lord,“Quantification of tumor vasculature and hypoxia by immuno-histochemical staining and HbO2 saturation measurements,”Br. J. Cancer 79, 464–471 ~1999!.

19. H. D. Sostman, S. Rockwell, A. L. Sylva, D. Madwed, G. Cofer,H. C. Charles, R. Negro-Villar, and D. Moore, “Evaluation ofBA 1112 rhabdomyosarcoma oxygenation with microelec-trodes, optical spectrometry, radiosensitivity, and MRS,”Magn. Reson. Med. 20, 253–267 ~1991!.

20. F. Steinberg, H. J. Rohrborn, T. Otto, K. M. Scheufler, and C.Streffer, “NIR reflection measurements of hemoglobin andcytochrome aa3 in healthy tissue and tumors,” Adv. Exp.Med. Biol. 428, 69–77 ~1997!.

21. R. G. Steen, K. Kitagishi, and K. Morgan, “In vivo measure-ment of tumor blood oxygenation by near-infrared spectros-copy: immediate effects of pentobarbital overdose orcarmustine treatment,” J. Neuro-Oncol. 22, 209–220 ~1994!.

22. E. L. Hull, D. L. Conover, and T. H. Foster, “Carbogen-inducedchanges in rat mammary tumor oxygenation reported by near-infrared spectroscopy,” Br. J. Cancer 79, 1709–1716 ~1999!.

23. M. S. Patterson, B. Chance, and B. C. Wilson, “Time-resolvedreflectance and transmittance for the noninvasive measure-ment of tissue optical properties,” Appl. Opt. 28, 2331–2336~1989!.

24. E. M. Sevick, B. Chance, J. Leigh, S. Nioka, and M. Maris,“Quantitation of time- and frequency-resolved optical spectrafor the determination of tissue oxygenation,” Anal. Biochem.195, 330–351 ~1991!.

25. H. Liu, A. H. Hielscher, F. K. Tittel, S. L. Jacques, and B.Chance, “Influence of blood vessels on the measurement ofhemoglobin oxygenation as determined by time-resolvedreflectance spectroscopy,” Med. Phys. 22, 1209–1217 ~1995!.

26. S. J. Matcher, M. Cope, and D. T. Delpy, “In vivo measure-ments of the wavelength dependence of tissue-scattering coef-ficients between 760 and 900 nm measured with time-resolvedspectroscopy,” Appl. Opt. 36, 386–396 ~1997!.

27. E. W. Hahn, P. Peschke, R. P. Mason, E. E. Babcock, and P. P.Antich, “Isolated tumor growth in a surgically formed skinpedicle in the rat: a new tumor model for NMR studies,”Magn. Reson. Imaging. 11, 1007–1017 ~1993!.

28. Y. Yang, H. Liu, X. Li, and B. Chance, “Low-cost frequency-domain photon migration instrument for tissue spectroscopy,oximetry, and imaging,” Opt. Eng. 36, 1562–1569 ~1997!.

29. H. Y. Ma, Q. Xu, J. R. Ballesteros, V. Ntziachristors, Q. Zhang,and B. Chance, “Quantitative study of hypoxia stress in pigletbrain by IQ phase modulation oximetry,” in Optical Tomogra-phy and Spectroscopy of Tissue III, B. Chance, R. R. Alfano,and B. J. Tromberg, eds., Proc. SPIE 3597, 642–649 ~1999!.

30. P. R. Bevington, Data Reduction and Error Analysis for thePhysical Sciences ~McGraw-Hill, New York, 1969!.

31. W. H. Press, B. P. Flannery, S. A. Teukolsky, and W. T. Vet-terling, Numerical Recipes ~Cambridge U. Press, Cambridge,1988!.

32. KaleidaGraph, Version 3.08 ~Synergy Software, 2457 Perki-omen Avenue, Reading, Pa. 19606, 1996!.

System for Windows ~Number Cruncher Statistical Systems,Kaysville, Utah, 1996!.

34. J. B. Fishkin and E. Gratton, “Propagation of photon-densitywaves in strongly scattering media containing an absorbingsemi-infinite plane bounded by a straight edge,” J. Opt. Soc.Am. A 10, 127–140 ~1993!.

35. W. G. Zijlstra, A. Buursma, and W. P. Meeuwsen-van derRoest, “Absorption spectra of human fetal and adult oxyhemo-globin, deoxyhemoglobin, carboxyhemoglobin, and methemo-globin,” Clin. Chem. 37, 1633–1638 ~1991!.

36. H. Liu, C. L. Matson, K. Lau, and R. R. Mapakshi, “Experi-mental validation of a backpropagation algorithm for three-dimensional breast tumor localization,” IEEE J. Select. Top.Quantum Electron. 5, 1049–1057 ~1999!.

37. S. S. Kety, “The theory and applications of the exchange ofinert gas at the lungs and tissue,” Pharmacol. Rev. 3, 1–41~1951!.

38. H. Watabe, M. Itoh, V. Cunningham, A. A. Lammertsma, P.Bloomfield, M. Mejia, T. Fujiwara, A. K. P. Johes, T. Johes, andT. Nakamura, “Noninvasive quantification of rCBF usingpositron emission tomography,” J. Cerebr. Blood Flow Metab.16, 311–319 ~1996!.

39. S. S. Kety, “Cerebral circulation and its measurement by inertdiffusible racers,” Israel J. Med. Sci. 23, 3–7 ~1987!.

40. A. D. Edwards, C. Richardson, P. Van Der Zee, C. Elwell, J. S.Wyatt, M. Cope, D. T. Delpy, and E. O. R. Reynolds, “Mea-surement of hemoglobin flow and blood flow by near-infraredspectroscopy,” J. Appl. Physiol. 75, 1884–1889 ~1993!.

41. H. J. J. A. Bernsen, P. F. J. W. Rijken, T. Oostendorp, and A. J.van der Kogel, “Vascularity and perfusion of human gliomasxenografted in the athymic nude mouse,” Br. J. Cancer 71,721–726 ~1995!.

42. R. P. Mason, P. P. Antich, E. E. Babcock, A. Constantinescu, P.Peschke, and E. W. Hahn, “Noninvasive determination of tu-mor oxygen tension and local variation with growth,” Int. J.Radiat. Oncol. Biol. Phys. 29, 95–103 ~1994!.

43. B. P. J. van der Sanden, A. Heerschap, A. W. Simonetti,P. J. F. W. Rijken, H. P. W. Peters, G. Stuben, and A. J. van derKogel, “Characterization and validation on noninvasive oxy-gen tension measurements in human glioma xenografts by19F-MR relaxometry,” Int. J. Radiat. Oncol. Biol. Phys. 44,649–658 ~1999!.

44. S. Hunjan, R. P. Mason, A. Constantinescu, P. Pescheke, E. W.Hahn, and P. P. Antich, “Regional tumor oximetry: 19F NMRspectroscopy of hexafluorobenzene,” Int. J. Radiat. Oncol. Biol.Phys. 41, 161–171 ~1998!.

5. S. Hunjan, D. Zhao, A. Constantinescu, E. W. Hahn, P. P.Antich, and R. P. Mason, “Tumor oximetry: an enhanceddynamic mapping procedure using fluorine-19 echo planarmagnetic resonance imaging,” Int. J. Radiat. Oncol. Biol. Phys.~to be published!.

6. S. P. Robinson, F. A. Howe, L. M. Rodrigues, M. Stubbs, andJ. R. Griffiths, “Magnetic resonance imaging techniques formonitoring changes in tumor oxygenation and blood flow,”Semin. Radiat. Oncol. 8, 198–207 ~1998!.

7. B. M. Fenton, “Effects of carbogen plus fractionated irradiationon KHT tumor oxygenation,” Radiother. Oncol. 44, 183–190~1997!.


Noninvasive Investigation of Blood Oxygenation Dynamics of Tumors by Near-Infrared Spectroscopy

Documents

Transcript of Noninvasive Investigation of Blood Oxygenation Dynamics of Tumors by Near-Infrared Spectroscopy