Acute Hematomas: Effects of Deoxygenatlon, flematocrit, and Fibrin-Clot Formation and Retraction on T2 Shortening’ 1 From the Gates and Crellin Laboratories of Chemistry, California Institute of Technology, Pasa- dena, CA 91 126 (R.A.C., J.D.R.), and the MR Imaging Laboratory, Huntington Medical Research Institutes, Pasadena (A.T.W., W.G.B.). From the 1988 RSNA annual meeting. Received May 1, 1989; revision requested June 6; revision received November 6; accepted December 20. Supported by the California Institute of Technology Summer Undergraduate Research Fellowship Program and E. I. Du Pont. Address reprint requests to J.D.R. 2 Current address: Department of Radiological Sciences, UCLA Medical Center, Los Angeles. 3 Current address: Department of Radiology, Long Beach Memorial Medical Center, Long Beach, Calif. (ci RSNA, 1990 Rachael A. Clark, BS #{149} Alyssa 1. Watanabe, MD2 #{149} William G. Bradley, Jr, MD, PhD3 #{149} John D. Roberts, PhD 201 Acute hematomas can appear hy- pointense on T2-weighted magnetic resonance (MR) images at field strengths as low as 0.35 T. Using Ra- man spectroscopy to measure blood oxygenation and taking T2 measure- ments at 2.1 and 9.4 T, the authors examined the relaxation mecha- nisms acting during deoxygenation, increases in hematocrit, and fibrin- clot formation and retraction. Indi- vidual contributions to overall T2 from deoxyhemoglobin and the in- teractions of water with protein hy- dration layers in hemoglobin, plas- ma proteins, and fibnin were mea- sured. Overall T2 values estimated by summing individual relaxation rates were in reasonable agreement with the T2 values of clotted blood. Results suggest that deoxygenation may be most important in T2 short- ening, followed by increased hemat- ocnit. T2 shortening from fibrin po- lymerization was minimal at the field strengths used. Effects of deox- ygenation and increasing hemato- cnit are more sensitive to field strength than fibrin T2 shortening. Effects of fibnin may be more sig- nificant at middle and low field strengths. Index terms: Blood, coagulation #{149} Blood, MR studies #{149} Brain, hemorrhage, 10.367, 10.43. Brain, MR studies, 10.i2i4 #{149} Hemoglobin. Magnetic resonance (MR), experimental #{149} Mag- netic resonance (MR), spectroscopy Radiology 1990; 175:201-206 T HE purpose of this study was to examine and quantitate the me- laxation mechanisms underlying physiologic changes that arc believed to cause enhanced T2 relaxation in acute hematomas. BACKGROUND INFORMATION Proton Populations in Blood Water molecules in blood are di- yided into intracellular and extracel- lular populations by red cell mem- branes. Diffusion across red cell membranes appears to be reasonably rapid, and Andrasko (1) has calculat- ed that the average lifetime of a wa- ter molecule inside an erythrocyte is 17 msec at 24#{176}C. If there is fast cx- change on the nuclear magnetic reso- nance (NMR) time scale between these two populations, a single trans- verse relaxation time is expected for the water protons. A further subdivision of proton populations is necessary when blood is recognized as a protein solution. The primary protein component is hemoglobin, but there are others present in lesser concentrations, such as fibrinogen and lipoproteins. All of these proteins interact with water protons and affect the relaxation be- hayior of the water protons. Protons in protein solutions can be divided on the basis of their mobility into three populations (2): (a) protons of water molecules that are not bound to protein (free water), (b) pro- tons of water molecules in the pro- tein hydration layers, and (c) cova- lently bound protons of protein mol- ecules. The covalently bound protons of protein molecules have very short T2 relaxation times, and the spin- echo technique employed to measure T2 in this study effectively disre- gards signals from this group. Pro- tons in the second group, those bound in the hydration layer, genen- ally have the same translational and rotational mobility as the protein molecules themselves (3). Fast cx- change between water in the protein hydration layers and free water will produce a single resultant relaxation time, which is a function of the pro- portion of water in the protein hy- dration layers and of the rotational correlation time characterizing the molecules of each of the kinds of pro- teins (2). The conditions of fast cx- change predict that T2 will be linear- ly related to protein concentration, at least up to concentrations at which the proteins strongly interact with one another (3). We assume through- out that the T2 relaxation of the wa- ten in blood will be observed as a sin- gle exponential signal decay. Factors That Can Cause T2 Relaxation in Acute Hematomas The physiologic changes currently implicated in T2 shortening include deoxygenation of hemoglobin (4,5), increased hematocrit and hemogbo- bin concentration resulting from clot formation and retraction (6), and fi- brim polymerization and clot retrac- tion (7). The hemoglobin in blood cells is expected to enhance transverse relax- ation by two separate mechanisms. One is associated with paramagnetic deoxyhemogbobin formed in the de- oxygenation of blood that will create Abbreviations: CPMG Carr-Purcell-Mei- boom-Gill, EDTA ethylenediaminetetraacetic acid, UV = ultraviolet.
6
Embed
Acute Hematomas: Effects ofDeoxygenatlon, flematocrit, and ...mriquestions.com/uploads/3/4/5/7/34572113/clark... · Other proteins inblood (eg,lipo-proteins) alsocontribute toT2relax-ation.
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Acute Hematomas: Effects of Deoxygenatlon,flematocrit, and Fibrin-Clot Formationand Retraction on T2 Shortening’
1 From the Gates and Crellin Laboratories of Chemistry, California Institute of Technology, Pasa-
dena, CA 91 126 (R.A.C., J.D.R.), and the MR Imaging Laboratory, Huntington Medical ResearchInstitutes, Pasadena (A.T.W., W.G.B.). From the 1988 RSNA annual meeting. Received May 1, 1989;revision requested June 6; revision received November 6; accepted December 20. Supported by theCalifornia Institute of Technology Summer Undergraduate Research Fellowship Program and E. I.Du Pont. Address reprint requests to J.D.R.
2 Current address: Department of Radiological Sciences, UCLA Medical Center, Los Angeles.3 Current address: Department of Radiology, Long Beach Memorial Medical Center, Long Beach,
Calif.(ci RSNA, 1990
Rachael A. Clark, BS #{149}Alyssa 1. Watanabe, MD2 #{149}William G. Bradley, Jr, MD, PhD3 #{149}John D. Roberts, PhD
201
Acute hematomas can appear hy-pointense on T2-weighted magneticresonance (MR) images at fieldstrengths as low as 0.35 T. Using Ra-man spectroscopy to measure bloodoxygenation and taking T2 measure-ments at 2.1 and 9.4 T, the authorsexamined the relaxation mecha-nisms acting during deoxygenation,increases in hematocrit, and fibrin-clot formation and retraction. Indi-vidual contributions to overall T2from deoxyhemoglobin and the in-teractions of water with protein hy-dration layers in hemoglobin, plas-ma proteins, and fibnin were mea-sured. Overall T2 values estimatedby summing individual relaxationrates were in reasonable agreementwith the T2 values of clotted blood.Results suggest that deoxygenationmay be most important in T2 short-ening, followed by increased hemat-
ocnit. T2 shortening from fibrin po-lymerization was minimal at thefield strengths used. Effects of deox-ygenation and increasing hemato-cnit are more sensitive to fieldstrength than fibrin T2 shortening.Effects of fibnin may be more sig-nificant at middle and low fieldstrengths.
Index terms: Blood, coagulation #{149}Blood, MRstudies #{149}Brain, hemorrhage, 10.367, 10.43.
laxation mechanisms underlyingphysiologic changes that arc believedto cause enhanced T2 relaxation inacute hematomas.
BACKGROUND
INFORMATION
Proton Populations in Blood
Water molecules in blood are di-yided into intracellular and extracel-lular populations by red cell mem-branes. Diffusion across red cellmembranes appears to be reasonably
rapid, and Andrasko (1) has calculat-ed that the average lifetime of a wa-
ter molecule inside an erythrocyte is
17 msec at 24#{176}C.If there is fast cx-change on the nuclear magnetic reso-
nance (NMR) time scale betweenthese two populations, a single trans-verse relaxation time is expected for
the water protons.
A further subdivision of protonpopulations is necessary when bloodis recognized as a protein solution.The primary protein component ishemoglobin, but there are otherspresent in lesser concentrations, suchas fibrinogen and lipoproteins. All ofthese proteins interact with waterprotons and affect the relaxation be-hayior of the water protons.
Protons in protein solutions can bedivided on the basis of their mobilityinto three populations (2): (a) protonsof water molecules that are notbound to protein (free water), (b) pro-tons of water molecules in the pro-tein hydration layers, and (c) cova-
lently bound protons of protein mol-ecules. The covalently bound protons
of protein molecules have very shortT2 relaxation times, and the spin-echo technique employed to measureT2 in this study effectively disre-gards signals from this group. Pro-tons in the second group, thosebound in the hydration layer, genen-ally have the same translational androtational mobility as the proteinmolecules themselves (3). Fast cx-change between water in the proteinhydration layers and free water willproduce a single resultant relaxation
time, which is a function of the pro-portion of water in the protein hy-dration layers and of the rotational
correlation time characterizing themolecules of each of the kinds of pro-teins (2). The conditions of fast cx-change predict that T2 will be linear-
ly related to protein concentration, atleast up to concentrations at which
the proteins strongly interact withone another (3). We assume through-
out that the T2 relaxation of the wa-ten in blood will be observed as a sin-gle exponential signal decay.
Factors That Can Cause T2Relaxation in Acute Hematomas
The physiologic changes currently
implicated in T2 shortening includedeoxygenation of hemoglobin (4,5),increased hematocrit and hemogbo-bin concentration resulting from clotformation and retraction (6), and fi-brim polymerization and clot retrac-tion (7).
The hemoglobin in blood cells isexpected to enhance transverse relax-ation by two separate mechanisms.One is associated with paramagneticdeoxyhemogbobin formed in the de-oxygenation of blood that will create
local magnetic gradients that act todephase the proton moments in dif-fusing water molecules. The other in-volves water protons, which also willhave shortened T2 values during theperiod that they arc sequestered inthe hydration layers of the protein.
Deoxyhemogbobin within mcd cellsis expected to create an average dif-fenence in magnetic susceptibility be-tween the interior and the exterior ofthe cells, although it has been report-ed that the effect is not more than0.15 ppm (8). There will also be gna-dients within the cell, but their ef-fects may be averaged by rapid diffu-sion of the water molecules. The pre-cessing magnetic moments of waterprotons in different parts of these bo-cal field gradients will not stay inphase, and as diffusion occurs, unlessit is fast, there will be further oppon-tunities for dephasing, thus shorten-ing T2 (4). This mode of proton relax-ation is dependent on the square ofthe magnetic field strength, andThulborn et a! (4) expressed doubtthat it would be significant at fieldstrengths of 1.5 T on bower. However,
Gomoni et a! (5) observed significantenhancement of T2 at 1.5 T using along 2TCPMG, the Carr-Purcell-Mci-boom-Gill intenecho interval. Trans-verse relaxation was reported to bedependent on both the intenecho in-terval and on the square of the fieldstrength at 1.5 T. It was suggestedthat this mode of transverse relax-ation depends on the paramagnetichemoglobin being sequestered in medblood cells because, in these experi-ments, the T2 values of blood bysateswere found to be independent offield strength and intcrecho interval.
Hemoglobin can also contribute totransverse relaxation by hydration-layer exchanges, as mentioned above.Hayman et al (6) found that lysingred cells does not, in fact, eliminateT2 shortening. These workers also es-tablished a linear relationship be-tween T21 and hematocnit for bothoxygenated and deoxygenated blood.The part of the protein-hydration-layer relaxation mechanism, which isassociated with the mcd cells, is cx-pected to depend on hemoglobinconcentration (and hematocnit) be-cause it is dependent on the propor-tion of water in the hemoglobin hy-dration layers.
Other proteins in blood (eg, lipo-proteins) also contribute to T2 relax-ation. One of particular interest is fi-
bninogen, which, when converted tofibmin, polymerizes and forms a clot.
Previous studies have shown that T2decreases in fibnin solutions during
the polymerization phase of clotting
(2,3). Daszkiewicz et al (2) suggested
that T2 shortening is the result ofgreatly increased correlation time ofthe water protons in the hydrationlayer of the fibnin molecules. The
molecular correlation time reflectsthe notational mobility of the fibninmolecules and would become largerwith increasing length of the fibminpolymers and the increasing rigidity
of the fibnin network (2).When several relaxation mecha-
nisms contribute to a single relax-
ation time, the overall Ti on T2 is me-bated to the individual relaxationmechanisms by the following cqua-
tion (9):
i/T2 = �i/T2, (1)
where the sum is over i individual me-
baxation mechanisms. Given that themechanisms mentioned above are in-dependent, we can predict the relax-ation time of the water in a sample ofclotted blood of known oxygenationand hematocnit by summing the mdi-vidual relaxation rates:
i/T20b5 i/T2Hbpar + l/T2Hbprot
+ 1 /T2 fib prot � � /T2p!as prot
(2)
where 1 /T2Hb par is the contributioncharacteristic of paramagnetic T2
shortening from diffusion throughmagnetic gradients in deoxyhemo-gbobmn, 1 /T2Hb prot is the contribution
associated with hemoglobin (oxy-,deoxy-, and methemogbobin) proteinconcentrations, 1 /T2fIb prot is the con-
tnibution from polymenized fibnin
molecules, and 1 /T2plas prot is the con-tnibution characteristic of intemac-tions with other plasma proteins.
In this study, we sought to isolateand measure each of these individual
relaxation rates. We studied the be-
havior of T2 with deoxygenation,
which was expected to change
1 /T2Hb par but have minimal effects
on the other mates. We examined thedependence of T2 on hematocnit inboth oxygenated and dcoxygenatcd
blood. Altering blood hematocmitchanges the protein concentrationscorresponding to hemoglobin andplasma proteins and also alters the
distribution of the field and localfield gradients, thus changing1 /T2Hb pare 1 /T2Hb prot� and 1 /T2pias prot
but leaving 1 /T2f�b prot unaffected.
MATERIALS AND METHODS
The T2 values of samples of varying ox-ygenation, samples of varying hemato-crit, and clotted samples were measured
to determine the dependence of the trans-
verse relaxation time on the degree ofblood oxygenation, the hematocrit, and
presence of clotting.
Sample Preparation
Blood was obtained from human volun-teers and stored for a few hours at 4#{176}Cuntil used.
Samples of varying hematocrit.-Hemato-crit was measured by centrifuging sam-
ples and visually measuring the propor-tion of red blood cells to total sample vol-ume. The hematocrit was increased bycentrifuging samples and removing the
supemnatant plasma. Samples with lower
hematocrit were prepared by mixing frac-tions of whole blood and plasma.
Samples of varying oxygenation-Bloodwas deoxygenated completely by addingsodium dithionite, 5 mg/mL, and stored
under argon. Oxygenation was increasedby incubating untreated blood in an at-
mosphere of molecular oxygen. Glass-ware was coated with paraffin to mini-mize red cell lysis. Samples were stored at4#{176}Cbetween measurements of T2 and ox-ygenation.
Samples with and without clots-Plasmasamples with and without buffy coat
were clotted to determine the relaxationrate associated with fibrin polymeriza-tion. Samples without buffy coat were
clotted because such samples undergonormal clot formation but contain far
fewer platelets (therefore there is littleclot retraction), allowing us to estimate
the relative contributions of fibrin poly-
merization and clot retraction to overallfibrin T2 shortening.
Heparinized blood was clotted in NMRtubes for clotting experiments with use of
0.2 mg protamine per milliliter of whole
blood and 0.04 mL reconstituted bovine
thrombin per milliliter of whole blood.
Only samples with well-formed clots
were used.Plasma with buffy coat was obtained by
removing plasma from above settled redcells. The buffy coat was removed from
plasma by centrifugation to produce plas-ma clots free of all cellular elements.
Clots of whole deoxygenated blood
were made from samples treated with so-dium dithionite. Clots of oxygenated
blood were made from whole blood sam-
ples incubated for 1.5 hours under oxy-gen at 25#{176}C.The 12 values of samples ofclotted oxygenated and deoxygenatedblood were compared with those calculat-
ed with use of the individual rates men-tioned in the Background section.
Determination of HemoglobinOxygenation
Blood oxygenation was determined by
means of a new quantitative application
of Raman spectroscopy. Raman spectros-copy, a light-scattering technique com-
monly used to probe the electronic struc-
ture of complex molecules, examines the
coherent laser light scattered by a sample.
25000
20000
C 5000
C 0000
(3)
- A +B+ 1’
mole fraction deoxyhemoglobin
= A X oxyhemoglobmn fraction, (6)
Volume 175 #{149}Number 1 Radiology #{149}203
Figure 1. Laser Raman spectrum of awhole blood sample. Oxygenation of sam-ples was measured by comparing the peak
intensities of band I, a hemoglobin oxida-tion-state marker band, with band V. a spin-state marker band. Sample shown contains77% oxyhemoglobin, 33% deoxyhemoglobin,and less than 1% methemoglobin. The laserwavelength was 441.6 nm. Hb deoxyhe-moglobin, Hb02 oxyhemoglobin, MHbmethemoglobin.
Some of the incident photons scattered bythe sample lose or gain energy. Examin-ing these scattered photons of altered en-ergy can provide detailed information asto the electronic and vibrational struc-tures of the sample molecule. Laser Ra-man spectroscopy differs from ultraviolet(UV) spectroscopy in that the Raman ef-fect involves light scattering, instead oflight absorption. In UV spectroscopy,photons with frequencies correspondingto energy differences in the molecule areabsorbed, and the molecule is raised to anexcited state. The excited molecule canlater lose its excitation energy through ra-diation or other mechanisms. In the Ra-man effect, the incident photons are scat-tered, not absorbed, and the frequenciesof the incident photons have no relation-ship to energy differences of the mole-cule.
Oxy-, deoxy-, and methemogbobin eachhave distinctive Raman marker bands(Fig i) with laser Raman irradiation at441 .6 nm. Band I, an oxidation-state mark-em band, is thought to reflect the electronoccupancy of the porphyrin ir* orbitals.High occupancy of these orbitals weakensthe porphyrin bonds and decreases theirvibrational frequency (i0-i2). Band I ap-pears at 1,358 cm’ for deoxyhemogbobinand at 1,377 cm� for oxy- and methemo-globin. Band V. a spin-state marker band,reflects changes in expansion of the por-phyrin core (13,14). Band V of oxyhemo-
globin occurs at 1,640 cm’.Spectra of pure oxy-, deoxy-, and met-
hemoglobin were used to measure inten-sities of bands I and V at 100% oxygen-ation, deoxygenation, and oxidation. In-
tensities of bands I and V in purehemoglobin samples, measured relativeto a 0.4 mol/L Na2SO4 standard, areshown in the Table. Oxygenation of testsamples was measured by comparing theintensities of bands I and V to theirknown maximum intensities from the
spectra of pure hemoglobin standards.That is, maximum peak heights for both
bands were measured on samples of iOO%oxy-, deoxy-, and methemogbobin, andthe percent maximum intensity of thesebands in the test sample was directly re-lated to the percent oxygenation.
Comparison of band I intensity for de-oxyhemoglobin (1,358 cm’) relative tothe intensity of oxyhemoglobin band V(1,640 cm’) yields A, the mole fraction of
deoxygenated blood divided by the molefraction of oxygenated blood:
A = -�--- X ‘L358
35.7 ‘1,640
= mole fraction deoxyhemoglobin
mole fraction oxyhemoglobin
where ‘1,358 intensity at 1,358 cm’, ‘1,640
= intensity at 1,640 cm1, and 35.7’ is astandardization factor that relates the in-tensity of deoxyhemoglobin band I at
100% deoxygenation to that of oxyhemo-globin band V at 100% oxygenation.
Likewise, the relative intensities ofbands I (oxy- and methemoglobin, 1,377cm’) and band V (oxyhemoglobin, 1,640cm’) minus the maximum intensity ofband I at 100% oxygenation (2.13) is equalto the mole fraction of methemoglobindivided by the mole fraction of oxyhemo-globin:
B= 0.28.�-�.�-2.13‘1,640
= mole fraction methemoglobin�
mole fraction oxyhemoglobin
We then have
mole fraction oxyhemoglobin
and
mole fraction methemoglobin
= B X oxyhemoglobin fraction. (7)
The relative intensities of Raman bandswere determined by measuring the peakheights of plotted spectra (Fig 1). The in-tensity of band V of oxyhemoglobin is
very small, especially in samples consist-ing mainly of deoxyhemoglobin (Fig 1).Care must be taken to minimize noise onthe Raman spectrum so that this peak canbe accurately measured. Noise can be re-duced by taking several scans and by us-ing dilute samples (samples of low he-matocrit). Ethylenediaminetetraaceticacid (EDTA) is the preferred anticoagu-lant in these experiments, because hepa-rinized samples fluoresced at 441.6 nm.However, it was necessary to use heparinas an anticoagulant when the blood was
clotted with protamine. For these cases,
the degree of oxygenation was measuredby taking two identically prepared sam-
ples; one with EDTA was used to measure
the initial oxygenation level while theheparinized sample was clotted.
Previous studies have used UV absorp-tion to determine oxygenation of blood
samples (4,5). We chose not to use UVspectroscopy because it would requiretransferring oxygen-sensitive samplesfrom NMR tubes to cuvettes. Raman spec-troscopy allowed us to make both T2 andoxygenation measurements on the samesealed capillary tube. In all experiments,T2 was measured before Raman spectros-copy because exposure of blood to laserlight causes photolysis of hemoglobin.Changes in blood oxygenation were mm-imized by allowing less than 1 hour be-tween measurements.
Measurement of TransverseRelaxation Time
Measurements of the proton T2 relax-ation time were made on 89.55-MHz and399.65-MHz Fourier-transform NMRspectrometers (Jeol Ltd. Tokyo) operatingat 2.1 and 9.4 1, respectively. A CPMG se-quence was used (15): 9O� - (TCPMG - i8O�- TCpMG),�. We restricted 2TCPMG to shortvalues (less than 6 msec for the 89.55-MHz measurements and 2 msec for the399.65-MHz measurements) to avoid pos-sible distortion of the results by the inho-mogeneities of the magnetic fields of ourspectrometers. Intensities of NMR waterresonance peaks were recorded at evenvalues of 2TCPMG, and T2 was calculatedby a least-squares fit to a single exponen-
(4) tial. Only measurements that had stan-dard deviations of less than 5% from theleast-squares fit were used. Duplicate T2measurements were taken of all samples;plotted values are the average of the twomeasurements. Standard deviations ofduplicate readings were within 3% of theT2 for the plasma-clot samples and within5% for all other samples. Clotting samples
were often observed to have two peakswith distinctive relaxation times. Theslowly relaxing component had relax-ation times characteristic of plasma andwere judged to arise from plasma sur-rounding the retracted clot. The shorterrelaxation times were believed to be from
water within the clot; these shorter relax-ation times are reported here as T2 of theclot. All measurements were made at37#{176}C.
RESULTS
Dependence of T2 onOxygenation at 90 MHz
Raman measurements showed thatthere was i% on less methemogbobmnin all samples of oxy- and deoxyhe-mogbobin. Figure 2 shows a linear me-bationship between T2’ and the
square of the mole fraction of oxyhe-mogbobin, with a correlation coeffi-
Figures 2-4. (2) The linear relationship between T2’ and the square of the mole fraction of oxyhemoglobin at 90 MHz (2.1 1). 2TCPMG = 6msec. Hematocrit was 37%. Other workers (4) have reported a linear relationship between 12_i and the square of the mole fraction of deoxy-hemoglobin. (3) Linear correlation of 12 with hematocnit at 90 MHz. 2TCPMG 4 msec. The minimum 12 is at a hematocrit of 100% for bothdeoxy- and oxyhemoglobin. (4) Linear correlation of T2 with hematocrit at 400 MHz (9.4 T). 2TCPMG = 2 msec.
Hous after Clotwi9
5. 6.
Figures 5, 6. (5) Changes in 12 of clotting plasma with and without buffy coat at 90 MHz.2TCPMG 4 msec. Zero-hour reading was taken immediately after adding protamine andthrombin. (6) Changes in T2 of clotting plasma at 400 MHz. 2TCPMG = 2 msec.
Dependence of T2 on Hematocritin Oxygenated and DeoxygenatedBlood
Raman measurements confirmedthat 99% on more of dithionite-treat-ed blood hemoglobin was in the de-oxygenated form. Raman measure-ments of oxygenated samples showedthat they contained 96% or more oxy-hemoglobin. Figures 3 and 4 showlinear relationships between hemato-crit and T2� obtained at 89.55 and399.65 MHz for both oxy- and deoxy-
hemoglobin. Correlation constantsfor all four linear fits were .99. Thelinear plots of Figures 3 and 4 are inaccord with the results of Hayman etal (6).
Dependence of T2 on Clotting
Plasma clots.-The T2 of plasma
clots with buffy coat decreased from438 to 365 msec at 89.55 MHz and
from i25 to i0i mscc at 399.65 MHzwithin 38 hours after adding pnot-amine and thrombin (Figs 5, 6). Atboth field strengths, the T2 of plasmaclots without buffy coat decreased by
roughly half the above amount. Aplasma sample with 2.5-3.0 times thenormal concentration of buffy coatunderwent a 54% decrease in T2 over24 hours.
Oxyhemoglobin clots.-The T2 of
oxyhemogbobin clots decreased from
i97 to i45 msec within 37 hours afterclotting at 89.55 MHz (Fig 7). Similar-
by, T2 at 399.65 MHz decreased from
57 to 46 mscc within 37 hours (Fig 8).No red cell lysis was observed inthese experiments, and the intense
med color of the clots did not change,
indicating no significant conversionto deoxy- on methemogbobin.
Deoxyhemoglobin clots.-At 89.55MHz, T2 decreased from 73 to 3imsec by 38 hours after clotting (Fig9), while at 399.65 MHz, the decreasewas from i2.9 to 6.2 mscc (Fig 10). Nomed cell lysis on colon change was ob-served during these experiments.
DISCUSSION
Equations (1) and (2) express theoverall T2 relaxation time of clottedblood in terms of rates of individual me-laxation mechanisms. If we have prop-emly quantitated the rates of the major
relaxation mechanisms responsible forT2 shortening, summation of the mdi-vidual mates should produce overall T2values for clotted blood samples thatare in reasonable agreement with theexperimentally measured T2 values.
Measurement of 1 /T2pias prot is
straightforward. To obtain a samplewith a hematocmit of 0%, red blood cells
were removed by centnifugation, leav-
ing plasma proteins in solution. The T2
of this 0%-hematocrit sample is deter-mined by interactions of water withplasma proteins; thus, 1 /T2plas prot �5
equal to 2.28 sec’ at 90 MHz. This rateincludes contributions from lipopro-
teins and fibrmnogen, the unpolymer-ized precursor of fibrin. Because fibnin-
ogen is included in the first term,1 /T2fIb prot �5 0 in uncbotted samples.
Clotted samples of plasma with andwithout buffy coat were prepared tomeasure i /T211b prot- Samples with buffycoat (containing normal concentrations
of plasma proteins, platelets, and fi-bninogen) showed a 16%-i7% decreasein transverse relaxation time to 0.37seconds by 38 hours after clotting. Ifwe assume that the degree of fibrin po-lymerization after 38 hours is the sameas would be observed in a whole bloodclot over this period, then the small me-la.xation contribution from fibrin poly-
merization (38 hours after clotting) isgiven by the equation i /T2f(b proc
(1/0.37) - i/T2ptasprot 0.42 sec’ at 90MHz. 1 /T2plas prot �5 subtracted fromthis rate because plasma proteins notparticipating in clotting also influence
the relaxation rate in these samples. Werecognize that the value of 1 /T2pias prot
contains a contribution from the unpo-
lymerized fibrmnogen molecules. Clear-ly, when these molecules are converted
to fibrin and form a clot, they no long-
019
018
T2. sec 017
016
015
0058
0056
0054
T2. sec0052
0050
0.048
0046
8.
) 10 20 30
Hours after Clown9
40
008
10 2�:� �
Hours after Clotwt�
007
006
005
004
003
0012
T2. sec0-010
0008
10 20 30
Hours after Clow�
msec.
) 1�0 20 30
Hours after ClouLn�
10.
Figures 9, 10. (9) Changes in T2 of clotting deoxygenated whole blood at 90 MHz. 2TCPMG
= 4 msec. (10) Changes in 12 of clotting deoxygenated whole blood at 400 MHz. 2TCPMG 2
Volume 175 #{149}Number 1 Radiology #{149}205
er contribute to 1 /T2pias prot- Thus,i /T2plas prot will decrease when clottingoccurs. We have neglected this de-crease in calculating the relaxation ratefrom fibnin polymerization, expecting
that because fibninogen is highly mo-
bile with a relatively short correlationtime, it should not greatly affect T2. Onthe other hand, networks of fibrinpolymers should greatly affect T2 be-cause of their long correlation times.For these reasons, the T2 change fromfibnin polymerization is probably large
enough to render the small error in itscalculation relatively insignificant.
The two rate constants for relaxationvia hemoglobin paramagnetic and pro-tein effects, 1 /T2Hb par and 1 /T2Hb prot�
remain to be determined. Whole bloodis a mixture of plasma and red blood
cells. The hematocnit, the percentage ofthe total volume occupied by red bloodcells, expresses the relative abundanceof each fraction. Theme are two majorchanges in the transverse relaxationrate as the hcmatocmit increases. First,the contribution of plasma proteins tothe total relaxation time decreases asthe solution becomes more concentrat-ed in red blood cells. The change incontribution of the plasma proteins isexpected to lead to an approximatelylinear decrease of 1 /T2plas prot with in-creasing hcmatocmit. Second, the in-creasing concentrations of red bloodcells in samples of increased hematocnitshorten T2 by both paramagnetic ef-fects (with deoxyhemogbobin) and pro-tein effects (with oxy- and deoxyhemo-globmn). Because diamagnetic effects
are usually small, it seems reasonable to
assume that the diamagnetic effects ofdcoxy- and oxyhcmogbobin are at beastcomparable. The relationships of 1 /T2to hematocnit can then be formulated asfollows. For deoxygenated, uncbottedblood,
i/T2 = (l/T2piasprot)[(i Hct)/iOO}
+ (i/T2Hbpar + i/T2Hbprot)
x (Hct/iOO), (8)
and for oxygenated, uncbotted blood,
i/T2 = (i/T2piasprot)[(i Hct)/100]
+ (i /T2Hb prot)U�ct/ iOO), (9)
where i /T2plas prot is the relaxation rateof the 0%-hematocnit (Hct) sample fromFigures 3 and 4 (2.28 seC1 at 90 MHz)and (1 /T2Hb par + i /T2Hb prot) is the me-laxation rate of the deoxygenated 100%-hematocrit sample (29.6 scc� at 90MHz). The constant i /T2Hb prot containssmall contributions to relaxation fromdifferences in diamagnetic susceptibil-ities. Oxyhemogbobin is not paramag-netic, so i /T2Hb par is equal to 0 forthese samples.
Estimated versus Observed T2 forDeoxygenated Clots
Clots of 99% deoxygenated wholeblood retracted to half their originalvolume by 37 hours after clotting. Thestarting hematocrit was approximately45%; clotting and retraction produced aclot with a hematocnit of approximately90% (45 + 50).
The estimated overall T2 (T2est) was
I Figures 7, 8. (7) Changes in T2 of clottingoxygenated whole blood at 90 MHz. 2TcPMG
= 4 msec. Zero-hour reading was taken im-mediately after adding protamine andthrombin. (8) Changes in T2 of clotting oxy-genated whole blood at 400 MHz. 2i’CPMG =
2 msec.
obtained as follows: i/TZ9� =
i /T2unclotted + i /T2f�b prot = 27.3 sec1 at90 MHz (1 /T2unclotted comes from Eq[8]). This gives T2�51 as 37 msec, in good
agreement with 35 msec observed forT2 at 38 hours after clotting. A similarcalculation with use of the data taken at400 MHz yields an estimated T2 of 5.7msec, and the observed T2 was 6.2 msec(Fig iO).
Estimated versus Observed T2 forOxygenated Clots
40 Clots of 96% oxygenated whole bloodretracted roughly the same amount asdid those of deoxygenated blood, andthe approximate hematocnit was 90% af-ten 37 hours.
The estimated T2 was as follows:1 /T2�5� = 1 /T2unclotted + 1 /T211b prot
7.67 sec� at 90 MHz (i/T2unc1otte,j
comes from Eq [9]). The T2est �5 130msec, in reasonable agreement with thei45 msec observed for T2 of oxygenat-ed clots at 38 hours after clotting (Fig7). At 400 MHz, the estimated T2 was3i msec, and the observed value is inpoor agreement at 46 msec.
Figures i i and 12 show estimates ofthe proportional contribution to theoverall T2 rate from each individual me-laxation rate at 90 and 400 MHz. Con-tributions were calculated as percent-ages of the total mate.
As shown in Figures i i and i2, withthe high hematocrits characteristic of ne-tracted blood clots (greater than 90%),hemoglobin protein is mainly responsi-ble for relaxation enhancement in bothoxygenated and deoxygenated blood at
both field strengths we used. Also, de-oxygenation of blood is more importantto T2 shortening than the protein ef-fects of both oxy- and deoxyhemogbobinat these field strengths; the contributionfrom the i /T2Hb par relaxation rate, cal-cuiated by subtracting the oxyhemoglo-bin relaxation rate from the deoxyhe-moglobin rate, was 72% of the overall
nate at 90 MHz and 82% at 400 MHz.The contributions to relaxation in oxy-genated clots were the same (to within2%) at both field strengths, as were thefibrin and plasma protein rates in deox-
ygenated clots. The only major differ-ence between the two field strengthswas the increased paramagnetic contni-bution at 400 MHz; the pamamagnetic T2shortening of deoxyhemogbobmn de-pends on the square of the magneticfield strength, and as expected, its con-tribution increases with field strength.
Figures 11, 12 Pie charts show estimates of the proportional contributions of the individual relaxation rates associated with the paramag-netism of deoxyhemoglobin and the concentrations of hemoglobin, fibnin, and plasma proteins to the overall 12 relaxation rate at 90 (11)and 400 (12) MHz for oxygenated and deoxygenated clotted blood. The sizes of the pies are not to scale but are drawn differently to empha-
size that the total relaxation rates are not the same for oxygenated and deoxygenated clots.
CONCLUSIONS
In clinical practice, acute intracene-bra! hemorrhage may appear hypoin-tense on T2-weighted images at fieldstrengths as low as 0.35 T (16). Of thephysiologic changes that we evaluatedthat could act to shorten the T2 of acutehematomas-deoxygcnation of blood,increase in hematocrit, and fibrin-cbotformation and netraction-deoxygen-ation appears to be the most important.Obviously, the T2 change arising fromdeoxygenation will depend on the de-gree to which the blood becomes deox-ygenated and on the field strength.
The T2 values of blood are alsostrongly affected by hematocnit, andthe increased hematocnit resulting fromclot retraction in hematomas can be cx-pected to shorten T2 significantly.
Fibrin polymerization and clot me-
traction had small effects on T2 at thefield strengths used in this study. Ap-proximately half of the fibmin-inducedT2 shortening appeared to result fromclotting and half from clot retraction.Although the T2 shortening from clot-ting was quite small in our expemi-mcnts, it appears to be sensitive toplatelet concentration. Not every he-matoma clots and retracts, and thosethat do may differ in platelet concen-tration, the degree of clotting, and theamount of clot retraction. Consequent-by, some degree of uniqueness must beexpected in the overall T2 decrease fordifferent hemorrhages.
Because the effects on T2 from dcox-ygenation and increasing hematocnitarc more sensitive to field strengththan are those arising from fibrin-cbotformation and retraction, the latter maybe relatively more important causes of
T2 shortening in acute hematomas atmiddle and bow field strengths.
In this study T2 was measured withshort interecho times (4-6 msec). Clini-
cal imagers commonly use interecho(ic, echo delay) times of around 20-40
mscc. With bong interecho times, the T2
shortening resulting from diffusionthrough localized magnetic gradientsin deoxygenated blood should increase.Gomori ct a! (5) reported that the ap-parent values of T2 from intracellular
deoxyhemogbobin decreased by a factor
of three when the intemecho time wasincreased from 4 to 64 msec. However,it is not completely clear how much of
the reported decrease was caused bydiffusion in the local gradients associ-ated with hemoglobin and how muchcould be ascribed to inhomogeneitiesin the static magnetic field. Irrespectiveof this, it is clear that in applying theresults of the present study to clinical
imaging, differences in interecho times
must be taken into account. U
Acknowledgments: Special thanks to RobertCopeland for invaluable help with Ramanspectroscopy and to S. I. Chan for providing
samples of pure oxy-, deoxy-, and methemoglo-bin. The authors also express appreciation toDominic V. McGrath for instruction in andsupport of NMR instrumentation and to David
A. Stauffer for helpful suggestions.
References1. Andrasko J. Water diffusion permeability of
human erythrocytes studied by a pulsed gra-dient NMR technique. Biochim Biophys Acta1976; 428:304-311.
2. Daszkiewicz OK, Jedrychowski A, Kirch-mayer S. Studies on blood clotting by thespin echo technique. Acta Med Pol 1965;1:95-102.
3. Jedrychowski A. Nuclear magnetic reso-nance studies of fibrin polymerization.Thromb Diath Haemorrh 1968; 20:513-527.
4. Thulborn KR, Waterton JC, Matthews PM,Radda GK. Oxygenation dependence of thetransverse relaxation time of water protons inwhole blood at high field. Biochim BiophysActa 1982; 714:265-270.
5. Gomori JM, Grossman RI, Yu-Ip C. Asakura I.NMR relaxation times of blood: dependenceon field strength, oxidation state, and cell in-tegrity. J Comput Assist Tomogr 1987; 11:684-690.
6. Hayman LA, Ford JJ. Taber KH. Saleem A,Round ME, Bryan RN. 12 effect of hemoglo-bin concentration: assessment with in vitroMR spectroscopy. Radiology 1988; 168:489-491.
7. Hayman LA. Taber KH. Ford JJ, Saleem A,Gurgun M, Mohamed 5, Bryan RN. Effect ofclot formation and retraction on spin-echoMR images of blood: an in vitro study. AJNR1989; 10:1155-1158.
8. Brunetti A, Alger J. Brooks R, Di Chiro G.On the origin of paramagnetic inhomogenei-ty in blood (abstr). In: Book of abstracts: Soci-ety of Magnetic Resonance in Medicine 1988.Vol 1. Berkeley, Calif: Society of MagneticResonance in Medicine, 1988; 233.
9. Levy GC. Carbon-13 nuclear spin relaxation.In: Topics in carbon-13 NMR spectroscopy.Vol 1. New York: Wiley. 1974; 96.
10. Kitagawa I, lizuka I, Saito M, Kyogoku Y.Resonance Raman scattering from hemopro-teins: the nature of the bond between thesixth ligand and the heme iron in ferrouslow-spin derivatives of hemoglobin. ChemLett 1975; 849-852.
11. Kitagawa I, Ozaki Y, Kyogoku Y. ResonanceRaman studies on the ligand-iron interactionsin hemoproteins and metalloporphyrins. AdvBiophys 1978; 11:153-196.
12. Spiro 1G. Stong JD, Stein P. Protein controlof porphyrin conformation: comparison ofresonance Raman spectra of heme proteinswith mesoporphyrin IX analogues. J AmChem Soc 1976; 98:5482-5489.
13. Lanir A, Yu NT, Felton RH. Conformationaltransitions and vibronic couplings in acid fer-ricytochrome c: a resonance Raman study.Biochem 1979; 18:1656-1660.
14. Spiro TG, Stong JD, Stein P. Porphyrin coreexpansion and doming in heme proteins:new evidence from resonance Raman spectraof six-coordinate high-spin iron(III) hemes.Am Chem Soc 1979; 101:2648-2655.
15. Meiboom S. Gill 0. Modified spin-echomethod for measuring nuclear relaxationtimes. Rev Sci Instr 1958; 29:688-691.
16. Bradley WG. MRI of hemorrhage and iron inthe brain. In: Stark DO, Bradley WG, eds.Magnetic resonance imaging. St Louis: Mos-by. 1988; 359-374.