JACC Vol. 19. No. I
March I5, 1992 :759-s4
759
METHODS
Intravascular Ultrasound Imaging of Saphenous Vein Grafts In Vitro ;
Comparison With Histologic and Quantitative Angiographic Findings
JOHN E . WILLARD, MD," t DYMPHNA NETTO, MD ." SABA E . DEMIAN, MD,'
DON R. HAAGEN . RCFL,v M. ELIZABETH BRICKNER . M U .rt ERIC J . EICHHORN . MD, FACC,*t
PAUL A . GRAYBURN. M D . FACC*t
Dallas, Texas
The ubiquity of coronary artery disease and the resultant widespread use of saphenous veins for coronary artery bypass surgery
has stimulated considerable interest in the morphologic and
pathophysiologic alterations these vessels undergo after implantation, This study was undertaken to determine the ability of
intravascular ultrasound to identify and chs :acterize abreormali .
lies in saphenous vein grafts. Ten saphenous vein grafts excised at
autopsy and nine saphenous vein segments harvested during
coronary artery bypass surgery were examined with intravascular
ultrasound imaging, quantitative overeat-3 a,lgiogrzphic techniques and histologic analysis .
lntravasenlar ultrasound lumen measurements were strongly
correlated with quantitative coronary arterfographic measurements (r 0.91, SEE 0.5 mm), Wall thickness was significantly
greater in the vein grafts after bag-term implantation than in the
freshly harvested veins (average thickness 1.4 m 0.5 vs. 0.7 =
0 .2 mm, p < 0,007); lhs0ndne ;veeelxted histeiogicatly with sum
wall fibrosis . There was good correlation between ntlrasmmd
imaging and histologic analysis, with the ability to distinguish
among normal intim, iastimbi hyperpfaxia, vein wait fibrosis and
atheromatous plaque.
Thus, this preliminary study, deaaastrates the ability of mrra
vascular titraseand to provide real-time crvvs-seetiomal images of
saphenous veins land morphologic characterization of their was .
This modality may have important clinical applications, including
the ability to detect serial changes in vein graft iatimal hyperplasta
and atheroscleross .
(j Am Call Canna! 1992 ;19:759-64)
The ubiquity of severe coronary ariery disease has resulted
in the widespread use of reversed autogenous saphenous
veins for coronary artery bypass surgery, Consequently,
there is considerable interest in the morphologic and
pathophysiologic alterations these vessels undergo after
implantation . Histologic analysis of grafts excised at the
time of reoperation or autopsy has revealed most of what
is known about saphenous vein graft disease (1-3) . There
have been several attempts to correlate cineangiographic
findings with histologic analysis at variable intervals after
implantation (4-6). Although these studies provide important assessment of saphenous vein grafts, cineargiography
provides little more than a two-dimensional image of the
lumen contour without a significant correlation with the
volume of atheroma present or the pathologic changes in
vessel wall morphology demonstrated by necropsy his-
tologic studies (7-10) . These studies are further flawed by
the extended time intervals between cine:aigiography and
histologic analysis that frequently occur in human auiusy
studies . thus allowing for considerah!e pathologic changes to
From the •Cardiac Catbetedzation Laboratory and Department of Interml Medicine, Divisan tfCardialogy . Dallas Veterans Altars Medical Center
and ttiniversily of Texas South-- Medical Center, Dallas, Texas. This
study Was presented in pmt al the 63rd Scientific Sessions of the American
Heart Association, Dallas . Texas, November 1590.
Manuscript received Jane 25.1991 : revised manuscript received August
26 . 1991. accepted September 12. 1991 .
Address for reprints: pant A . Graybum. MD. Division of Cardiology
011A, . Veterans Affairs Medical Cemer .4500Soulh Lamaaler Road. Dallas.
Texas 75219,
01992 by the Anreacan Colks* or Cardwlogy
occur between data points . An in vivo modality capable of
accurately defining the morphologic features of saphenous
vein grafts that evolve after implantation viould have goal
clinical potential for identifying and chaexeterizing serial
pathologic changes such as intimal hyperplasia and atherosclerosis,
Intravascular ultrasound is an imaging modality that
could potentially provide such information- Recent work
has shown that it can generate detailed images from
within the lumen of arteries (11) . distinguish muscular from
elastic arteries and differentiate normal vegsrt wnlt from
atherosclerosis (12-18) . Unfortunately, few data exist regarding the utility of in : avascular utrresoend imaging
in sapheuous vein grafts . Therefore, the purpose of this in
vitro study was twofold : I) to compare the ability of intravascular ultrasound imaging to accurately measure
lumen diameter with that of quantitative angiography, and 2)
to compare the ability of intravascular ultrasound imaging to
distinguish among normal intima, intimal hyperplasia . vein
wall fibrosis, and atheromatous plaque with that of histologic
study .
0725-io97.5295 .00
760
IACC Vat . 19. No. 4
March 15, 1992 :759-64
WILLARD F.T At
INTRAVASCULAR ULTRASOUND IMAGING OF GRAFTS
l-P
0
tine
r-0 .9t
y =o.7s. .t .5
5EE=0 .0 mm
Figure 1 . A saphenous vein s°gment w attached with a silk
ligature at each end to an SF hemostatic sheath . The side arm of one
sheath was connected ton fluid-filled pressure transducer calibrated
to mercury . The other side arm was attached to a fluid reservoir
containing a 50VVr, mixture of saline solution and iodinated contrast
medium . The reservoir could he raised or lowered to maintain a
constant irtraluminal distending pressure, (:incangiooraphy (tine)
Figure 2. Relation between measurement,
was performed with the intravascular ultrmound (IVUS) catheter
diameter by intravascular ultrasound ly axis) and quantitative cor-
positioned at the site of measurements .
onary arteriography (QCA) (x axis) is depicted as compared by
linear regression analysis . Other abbreviations as in Figure 1 .
Methods
. We examined 10 saphenous vein
Experimental protocol
grafts excised at autopsy and 9 freshly harvested saphenous
vein segments obtained during coronary artery bypass surgery . The vessels were subjected to intravascular ultrasound
imaging, quantitative coronary angiography and histologic
study. A saphenous vein segment was attached with a silk
ligature at each end to an 8F hemostatic sheath (Fig. U . The
side arm of one sheath was connected to a fluid-filled
pressure transducer calibrated to mercury. The other side
arm was attached to a fluid reservoir containing a 50%
mixture of saline solution and iodinated contrast medium .
The reservoir could be raised or lowered to maintain a
constant intraluminal distending pressure . We empirically
used a constant pressure of 30 mm Hg to facilitate comparisons between the imaging modalities employed. The vessel
was submerged in a water bath to allow for both intravascular ultrasound imaging and cineongiography .
Intravascular ultrasound. A 4 .81' intravascular ultrasound catheter (Boston Scientific) was introduced into the
sheath and positioned in the mid-portion of the vein . The
catheter was operated at a carrier frequency of 20 MHz by a
Diasonics Intravascular Ultrasound imaging console . This
system employ s a mechanical transducer rotating at 900 rpm
to image the entire vessel circumference at a 10° Irok .
forward angle . Measurements of lumen dia tteter and wall
thickness were made an-line by using the software package
built into the instrument . Images were recorded on 0 .5 in.(1 .27-cm) VHS videotape .
Quantitative aupiography . For each saphenous vein segment
. single-plane 35-mm cineangiographic images were
acquired. The cineangiograms were made with the intravascular ultrasound catheter still in position to allow for diameter measurement at the same site used by the ullrasouttd
transducer. Quantitative coronary angiography was performed with use of a Tagarno projector interfaced with the
Coronary Angiographic Analysis System . A calibration fac
.
for for image magnification was determined by caliper mea-
4
5
6
7
OCA Diameter Imm)
of maximal
(Max) lumen
suretnent of the external diameter of the hemostatic sheath .
Measurements of lumen diameter, with correction for pincushion distortion, were then made on-line as previously
described (19,20) .
Histology . After the final cineangiogram each vein segment was distended with 10% buffered formalin to the same
distending pressure used for ultrasound and angiographic
imaging. The vein segments were then submerged in formalie to allow for overnight fixation. Transverse sections were
prepared from veins at specific sites where ultrasound images had been recorded, and at I cm intervals . The histologic
sections were stained with hematoxylin-eosin, Verhoeff-van
Gieson and Masson trichrome stains . The sections were then
examined microscopically by a pathologist who had no prior
knowledge of the ultrasound appearance of the vein segments .
Statistical analysis . Maximal vessel lumen diameters obtoined by intravascular ultrasound imaging and quantitative
coronary angiographic measurements were compared with
each other by linear regression analysis to determine the
standard error of the estimate (SEE) . The wall thickness, as
determined by intravascular ultrasound imaging, of chronically implanted vein grafts was compared with that of freshly
harvested veins by an unpaired Student I test . A p value <
0-05 was accepted as statistically ,ignificaat .
Results
Comparison of intravascular ultrasound and angiugrnphlc
measurements . Figure 2 illustrates the relation between
measurements of maximal lumen diameter by intravascular
ultrasound ly axis) and quantitative coronary arteriography
(x axis). Linear regression analysis revealed a strong correlation (r 0.91, SEE 0.5 mm) . Although these two imaging
techniques were concordant in measuring lumen diameter,
intravascular ultrasound was able to identify encroachment
of the lemon by diffuse pathologic processes such as intimal
hyperplasia . Conversely . angiography- which shows only
JACC Val . 19. No.4
March 15. 1997 :759-64
the column of contrast medium within the vessel lumen . was
generally unable to identify diffuse narrowing of the lumen .
Freshly harvested vein graft morphology . Low power
microscopy of a freshly harvested saphenous vein, performed with use of a Verhoeff-van Gieson elastic stain (Fig .
3), typically revealed a normal thin-walled vein with a widely
patent lumen without evidence of intimal hyperplasia . The
dark black fibers represent elastin, which probably accounts
for the bright echogenic appearance on intravascular ultrasound imaging. A typical ultrasound image of a freshly
harvested vein reveals a thin-walled vessel with an absence
of echo reflectance within the vessel lumen .
Chronically implanted vein graft morphology . Chronically implanted saphenous vein grafts may demonstrate a
variety of pathologic changes . including intimal proliferation
WILLARD ET AL.
INTRAVASCELAR ULTRASOUND IMAGING OF GRAFTS
76 1
Figure 3. Left panel, Histologic cross-sectional view of a freshly
harvested saphetlous vein stained with Verhoeffvan Gieson stain
Imagnificalion X6 .25, reduced by 287c). A normal thin-walled vein
with a widely patent lumen is demonstrated. The dark black fibers
represent elaslin. which accounts for the bright echogeni- epy arance on intresascular ultrasound imaging (right panel). Note the
absence of echo reflectance within the vessel lumen .
Figure 4L Len pod, ilistotogic cross-sectional view ot a tong-tern
implanted saphenous vein graft stained with Vedloei< van C`eson slain
fmagnificalion x615, reduced by 359c) . Concentric thickening and
sclerosis of the vein wall tract anon) and narrowing of the vessel
lumen by inrirnal hypcrplasia (lbn arrow) are evident . Right panel,
Intravascular ultraseund imaging revealed a bright highly relkCtive
region that correlates with the llickened vein wall (rmdr ron) and a
faint echolucent band within the vessel lumen that coeetaes with The
intimat hyperplasia observed on histologic study (thin rmw)-
762
WILL ARD ET AL .
INTRAVASCULAR ULTRASOUND IMAGING OF GRAFTS
Figures . Len panel, Histologic cross-sectional view of a t0-year old
saphenous vein graft stained with Verhoef-van Gieson slain (magnificalion x6,25, reduced by 28)5) . lmimal hyperplasia and atherosclerosis involving the intima and media (small arrows) with microcystic degeneration (thick arrow) are evident . Right panel,
Ultrasound image from the same area of the graft reveals a highly
reflective thickened vein wall (open arrow) with an echogenic
atheroma from the 4 o'clock to the 9 o'clock position (small arrows)
that coreiates with the plaque on Verhoeff-van Gieson staining . On
intravascular ultrasound imaging atherosclerosis appears more heterogeneous than does intimal hyperplasia . Left panel, Magnification
x6.25. reduced by 20W.
and atherosclerosis . Figure 4 shows uniform narrowing of
the vessel lumen by intimal hyperplasia . The vein wall itself
is thickened and sclerotic in contrast to the freshly harvested
vein seen previously. Intravascular ultrasound imaging revealed a bright, highly reflective region that cur relates with
the thickened vein wall ; intimal hyperplasia appears as a
faint echolucent band within the vessel lumen.
Fix... 5 is u representtniv, ennnlple ofthe appearance of
athernsclerosis. The ultrasound image of a 10-year old
saphenous vein graft reveals a highly reflective . thickened
vein wall with an echogenic atheroma from the 4 o'clock
to the 9 o'clock position that correlates with the plaque on
Verlloeff-van Gieson =laming (Fig . 3) . Inlimal hyperplasia is
also present and appears as a faint echolucent band within
the vessel lumen .
Wall thickness was significantly greater in the vein grafts
after long-term implantation than in the freshly harvested
veins, Specifically • wall thickness averaged 1 .4 1 0 .5 nim in
the old vein grafts versus 0 .7 *_ 0.2 turn in the new saphenous
veins . As demonstrated histologically . this difference in vein
wall thickness was due primarily to vein wall sclerosis .
Despite the remarkable refinement of this technology and
JACC Vo1 . 19. No. a
March 15. 199175'4-64
miniaturization of the ultrasound catheters, their access is
confined to lumens of moderate-sized diameter . Figure 6
illustrates a severely diseased portion of a graft that could
not be imaged because of the small lumen diameter .
Discussion
Intravascular ultrasound imaging . This preliminary study
demonstrates the ability of intravascular ultrasound to provide real-time truss-sectional images of saphenous vein
grafts and to distinguish among the pathologic abnormalities
of the vein graft wall that evolve after implantation . This new
modality provides detailed information regarding the alterations in vessel wall morphology, thus allowing the detection
of diffuse disease encroaching on the vessel lumen that
would be missed by contrast angiography . Although this
preliminary study involved a small number of vein grafts
(especially those containing intimal lipid deposition), the
correlation between histologic analysis and ultrasound images suggest that this methodology could be used to establish
the sequential changes of autogenous vein grafts after implantation . This capability would eliminate the potential for
histologic processing artifacts and the temporal disparity
between cineangiographic and histologic analysis common
LO autopsy studies .
Three are sere,, nmrphologir aIterations that ran nrcur
in m+tngenous vein grafts idler implantation into the arterial
circulation : I) endothelial damage, 2) medial hypertrophy,
3) medial necrosis, 4) graft wall fibrosis (media and adventitia) . 5) intimal hyperplasia• 6) atherosclerosis, and 7) aneurysmal dilation (1) . This study suggests that intravascular
ultrasound can distinguish between the normal appearance
of freshly harvested veins and I) vein graft wall fibrosis
JACC Vol. 19. No. J
March 15 . 1992.759-64
IN rRAV,sSC LuL .R
W I LLARD ET AL.
t :LTBASOutuD rmAGtNG OF GRAFTS
263
Figure 6. Histologie cros •s ectional viof the graft depicted in Figure 5 . Note the
extensive compromise or the lumen by
severe vein graft atherosclerosis and disrupted intimal plaque . This view illustrates
one of the limitations of intraluminal ultrasound in which severe lumen -wing
presents justapositional maneuvering of
the iruravascular ultrasound catheter .
(media and adventilia), 2) intimal hyperplasia, and 3) atherosclerosis .
Wall thickness. A marked increase in vessel wall thickness was detected among the chronically implanted saphenous vein grafts in comparison with wall thickness in the
freshly harvested saphenous vein segments . Histologic analysis correborated this Boding and identified graft wall fibrosis as the primary pathologic lesion responsible for the
increased wall thickness, with a very thin layer of intimal
hyperplasia present and identifiable on intravascular
imaging. It was also possible with microscopy to identify
etastin within the vessel wall by the use of an elastic stain
(Verhneff-van Gieson) and to correlate its presence with its
bright echogenic appearance within the media on ultrasound
imaging,
Lumen diameter . The lumen diameters obtained from the
intravascular ultrasound images in this study correlated
closely with our angiugraphic measurements (Fig . 21. This
finding is in agreement with the previously reported (211
comparison of intravascular ultrasound and quantitative
coronary angiography in muscular arteries, However . it
appears that this Close correlation may not be applicable
after angioplasty . Angioplasty often results in tissue dissection planes that can fill with contrast medium, resulting in a
hazy appearance on angiography that may lead to significant
overestimation of the residual lumen (22-25) . We have not
evaluated the accuracy of intravascular ultrasound measurements after saphenous vein graft angioplasty .
Clinical implications . The ability to perform serial in vivo
characterization of vein graft morphologic appearance would
allow serial assessment of pharmacologic interventions directed at reducing the rapidity of vein graft failure and
restenosis after vein graft angioplasty . The use of intravascular ultrasound imaging during interventional procedures
may yield a more accurate assessment of one lumen dimensions. thus allowing for the optimal depth of tissue eartucl ion
with atixerectomy or ratoablation . 11 may also permit better
assessment of the need for additional dilations during augioplasty or the presence of a flap that is partially occlusive of
the lumen and that could be excised with adjunctive atherectea y. Furthermore
. the use of intravascular ultrasound
could be extended to the operative suite to evaluate patency
of the distal anastomotic site, thus ensuring that it is not flow
restrictive . Moreover. inlravascularultrasoundimagiuguray
be valuable
selectut th_ -pt;m&I dicta] art,autoniy site
and thus avoiding placement of the anastomotic site in the
midst of or just proximal to a significant corontry artery
stenosis.
Limitations of the study. This study was performed in
vitro under idealized conditions and is therefore subject to
several potential limitations, In viva . small vein grafts or
those with a significantly reduced lumen diameter, due to
atherosclerosis. intimal hyperplasia or thrombus . might preclude proper positioning of the intravascular sitaisound
catheter . Because the vein grafts obtained after long-term
implantarlan were excised dnringauropsy, their press vasion
and processing may have resulted in inadvertent dislodgmeat of loosely adherent thrombus . Moreover, we were
unable to evaluate thromhoscd vein grafts because the
residual lumen was generally obliterated and did not allow
room for the imaging catheter . Thus, we did not acquire any
data an the ultrasound appearance ofthrombus in saphenous
vein grafts. Laboratory processing may also have resulted in
artifacts that influenced the histologic appearance of the vein
grafts. Finally, this preliminary study included a small number of vein grafts for interrogation_ therefore the significance
of these findings is subject to the limitations of the small
sample size .
76 4
WILLARD CT AL .
INTRA V ASCULAR ULTRASOUND IMAGING OF GRAFTS
Conclusions . In this in vitro study we compared !he
information obtained by iniravascular ultrasound imaging,
cineangiography and histologic study of saphenous vein
grafts. fnlravascular ultrasound imaging is capable of accurate measurement of lumen dimensions of saphenous vein
grafts as compared with results obtained Gvith quantitative
cineangiogaphy . Although angiography is limited to outlining a vessel's lumen, intravusculnr alnrasound imaging also
provides morphologic chart-Aericat :co of the v .eel vial litre
aprons with histologic data. The clinical relevance of determining vessel wall morphologic features of saphenous vein
grails by intravascular ultrasound imaging avails the results
of ongoing in vivo studies,
References
I . nary
Spray TL • Rooms WC . Chengc ; in saphannvs "inn usud en
bypass Oman. Am Hcan J 197784 :500-16.
2. Smith SH, Gee, JC . Morphulugy of a mbenous vein-coronary artery
bypass grafts. Arch Panhel Lab Med 1987 :107 : i3--8.
3. Lie JT. Law- GM. Morris GC. Anmomrorsry bypass saphenous vain
graft stheroslerosic anatomic study of99 vein grafts from normal and
hypedipopraeiaamic patients up to 75 months postopemlively. Am I
Cardial 1977;40:906-14 .
4. Lawrie GM, Lie 1T, Morris GC Jr. Beazley HL . Vein gmn potency and
imimal proliferation after aortocoronary bypass early and loeig .arrm
argirOralhdogic oomlations . Am 1 Cordial 1976 .38 :856-62.
5 . Bj6rk VC, teen T, Lamina C . Anglor,raphic changes in mtemal mammon'
artery and s, pheraas vein grafls-two weeks . one yearam i five years after
ovary Imp ess surgery. Stand f Thorac Cardiovasv Surg 1981 ;IS:L'+-70 .
6 . Camp- L, Enjalbert M . Lmperance J . Vaislic C . Grondin CM .
Bourassa MG, Atherosclerosis and lme closure at aar000nmlary saphenous vein grafts: sequential angiographic studin,at E creeks . 1 year.510
7 years, and 10 to I^_ years after surgery . Circuhtion 1983t68lsuppl
Iti'~11-I-?.
7 . Vlndaver /., Frech R, Van Tassel RA . Edwards IF . Carrrdalmn at inc
anmmmtem coronary arteriagram and the postmurlnm specimen . Clrcuralion 1n•1, .47 :r6Z-9.
8 . Arnett ET . raiser 161, Redwood DR . el al . Coronary artery ma-AC in
ca-my heart dictate : Campannan of eineanglagrapldr and necropsy
Poi Ann late. Med 1979 :91750-6.
9, EnstermanlH,AchorRWP .KincaidOW.BrassALJr.Athcrasclcmtic
disease of the ccaunary anencs. a vatlwlupk-r,i iulogla-aetelative sicdy .
Circulation 191,2:76 :1788-955
in. Wmrer BE PaStOr WC. Arnounl of narrowing by mheroselrtat:r plaque
JACC Vet, 19 . No . 4
March 15, 1992:759-64
M 44 non.bypass .d and 52 bypassed -jot epicardial coronary arteries in
12 neempsy patio ms who died within m e month oiaonotOa000ry bypass
grafting, ,4m 1 Card-1 1980 ;46:9.56-62.
11 . Ysn:k PG. Johnson EL. Linker Dr, Inlrasascular ultrasound: development and clinical porentiol . Am 1 Cnmiee Imaging 1988 :2:185-93.
12 . Gasaenhoven El . Essed CE, Lancer C T . et al . Arteual wall &1
lies determined by iruwnaueular uhrasound hnaging : an in vitro study .
J .4m Call Cards 1999:14:947-521), Oussenheven El, Easrd CE. Frielman P, er al . Inbavascular ultrasonic
imaging: histologic and echographic coroeimion EurJ Vans Surg 1989 :3:
570-1i .
14 . Giessenhoven EJ . Essed CE, Frietman F, heravascular echographic
assesamcel of vessel wall chamctern ica1 a eorrelanirn with histolney . ]At
J Camine Imaging 1919:4:103-16,
15 . Tobis SM . ValteryJ, lahon D.ct al .lalrovascularullresound imatingof
human cocaar; arteries in viva. analysis of tissue charavlen .alioas with
. Circulation 199n13 :9t7comparison to in vitm hisrological specimens
2G .
16 . Landini L. Samelli R, Pwano E, Salvation M . Evaluation of frequency
dcpcndcnce of backscancr eaM6eient in normal and mherosclerulic conic
walls. Ultrasound Mall Bial 1936 .12 :897-41,1 .
17 . Barzilai B . Saffiz 3E . Miller JG . Sobel BE
. Quantitative ultrasonic
chxmeleriro;rmv ordhe tram re of al lwroseleroli, plaques in human aorta .
Cite Rae 1987160 :459-63 .
18. Linker DT. Torts PG . Grpnnimgsaether A . Johansen E. Angelsen BAK.
Analysis or backrcattemd ultrasound from normal and diseased arterial
wall . Inn I Cord Imaging 1989:4:177-85 .
19. RaberJHC.KooiimonCJ .SlagerCI •etal .Coronary artery dimensions
from eangielra s :methadalogyandvalidelionofacoMputer-assisted
analysis procedure. IEEE Traps Med Imaging 1964;MI.3 :131-41 .
20. Reiber JHC, Semi PW. Kooijman CJ, et al. Assessment of short.,
medium-, and Iang.term variations in arteral dimensions from computer.
assisted quantilation of coronary cincangiagrams Circulation 1985;71 :
280
.8 .
21 . Nissan SE. Cmines CL. Gurley 3C . el al . Appkcarion of a new phasedarray ultrasound imaging catheter in the assessment of vascular dimeavivo eompmi on to cinsangiogmphy .Cireulation 1990 :81:660-6 .
22. Tobis 1M. hamkry JA, Gessen 1, et al, Inbavascular ultrasound crosssectioned anoint imaging before and after balloon angioplasly, in vitro .
Cimulallan 191910;873-82.
23. Mnllery JA . Table JM, Grifhlh 1, el al . Assessment of normal and
athnoaelemlic arterial wag thickness with an irnnavarcutar ultrasound
imaging call Am Heart J 1995,119'.1192-4W.
24. Pnrkire BN . Banorelli AL . Gessen 3M, e1 al. Coronary artery imaging
with irtnvascular highh frequency uhresoued. Circulation 1990 :81 :157585 .
25, Yak PG. Frail PI. Linker DT. .Asses BA!. int .vescular Alan-ad guidanw fur calhelnr-based wroaary inten'enlions, 1 Am Call
Cardurl193p173Wi-459 .