and GERALD D. BUCKBERG JOSEPH UTLEY ...

Total and Regional Myocardial Blood Flow Measurements with 25µ, 15µ, 9µ, and Filtered 1-10µ Diameter Microspheres and Antipyrine in Dogs and Sheep JOSEPH UTLEY, EDWIN L. CARLSON, JULIEN I. E. HOFFMAN, HUGO M. MARTINEZ and GERALD D. BUCKBERG Circ Res. 1974;34:391-405 doi: 10.1161/01.RES.34.3.391 Circulation Research is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 1974 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7330. Online ISSN: 1524-4571

The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circres.ahajournals.org/content/34/3/391

Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation Research can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation Research is online at: http://circres.ahajournals.org//subscriptions/

Downloaded from http://circres.ahajournals.org/ by guest on May 30, 2012

Total and Regional Myocardial Blood Flow Measurements with 25/x, 15/x, 9/JL , and Filtered 1-10/x Diameter Microspheres and Antipyrine in Dogs and Sheep By Joseph Utley, Edwin L. Carlson, Julien I. E. Hoffman, Hugo M. Martinez, and Gerald D. Buckberg ABSTRACT The accuracy of regional myocardial blood flow measurements made with microspheres of different sizes is uncertain. Therefore, we simultaneously injected radioactive microspheres of different sizes into the left atria of dogs and sheep; the microsphere diameters were 25/A, 15/A, 9/X, 1-10/X, and filtered 1-10/x (most >7/x). Antipyrine was sometimes simultaneously infused for 15-60 seconds. Myocardial blood flow was altered by hemorrhage, tachycardia, supravalvar aortic constriction, or infusion of methoxamine or adenosine triphosphate; left coronary artery branches were occluded on two occasions. Sometimes the percent of untrapped microspheres was estimated. All sizes of microspheres measured similar total myocardial blood flows when the percent of untrapped microspheres was known. All indicators were distributed identically to the right and left ventricular free walls and septum; ischemic areas had a 1% excess of antipyrine. With any pair of microspheres, the larger had a subendocardial excess except for 25/x microspheres, which were in excess compared with 15/u. microspheres in the higher flow layer whether it was subendocardial or subepicardial. The greatest difference for any pair of microspheres was 10.53% of flow in a layer. Antipyrine did not define which size of microspheres measured true regional blood flow, since many previously unemphasized limitations were discovered. Nevertheless, we believe that microspheres 9/x in diameter are probably the best for measuring regional myocardial blood flow. KEY WORDS microsphere trapping transport function

partition coefficient acute myocardial infarction subendocardial ischemia coronary occlusion coronary blood flow

• Measurements of blood flow to small myocardial regions cannot be made with flowmeters or the direct Fick technique, since the vascular inflow to and outflow from these regions cannot be isolated. It is therefore necessary to use techniques involving an indicator which is distributed to various myocardial regions in proportion to the regional blood flows; the amounts of indicator entering, retained in. and leaving the regions must be deter-

From the Cardiovascular Research Institute, the Department of Pediatrics, and the Department of Biochemistry and Biophysics, University of California, San Francisco, California 94143, and the Department of Surgery, University of California, Los Angeles, California 90014. This study was supported by U. S. Public Health Service Grant HL-06285 from the National Heart and Lung Institute, by grants in aid from the Beaumont Foundation, and by U. S. Public Health Service Grant CM-13242 (Dr. Buckberg), NHLI Graduate Training Grant HL05251 (Drs. Utley and Carlson), and NHLI Cardiothoracic Surgery Training Grant HL05911 (Dr. Utley). Received August 13, 1973. Accepted for publication December 19, 1973. Circulation Research, Vol. XXXIV, March 1974

mined. At present, both particulate and diffusible indicators are used to measure regional blood flow, but proof of their accuracy is often not available. Rudolph and Heymann (1) first demonstrated that radioactive microspheres could simultaneously measure flow to all the organs, and Buckberg et al. (2) and Archie et al. (3) described some of the factors needed to make this method accurate. Total myocardial blood flow measured by injecting radioactive microspheres is usually within 10% of that measured by collecting the coronary venous drainage (2, 4). However, no one has proved that microspheres can measure flows to small regions within any organ, including the heart. Domenech et al. (4) found that the ratio of the number of microspheres per gram in the subendocardial muscle to that in the subepicardial muscle was 2.7 for microspheres 51-61/x, in diameter, 1.4 for those 20-23^ in diameter, and 1.2 for those 14/x in diameter. They believed that the larger microspheres overestimated subendocardial blood flow and underestimated subepicardial blood flow. 391


392

UTLEY, CARLSON, HOFFMAN, MARTINEZ, BUCKBERG

Both in physical models (5) and in animals (6), large particles in a flowing stream migrate axially more rapidly than do small particles. It is thus possible that axial migration of the larger microspheres causes skimming of microsphere-poor blood into the proximal subepicardial arteries and delivery of microsphere-rich blood to the terminal subendocardial arteries. Recently, Yipintsoi et al. (7) injected diffusible indicators and microspheres 15/i., 35/i, and 50/x in diameter to study flow to small regions of the myocardium. They showed that the diffusible indicators and the 15/A diameter microspheres measured flows to any given small region with only slight differences, but they could not tell which, if any, gave true flow measurements. They also found in one experiment in which microspheres 15/x and 35fx in diameter were injected simultaneously that the larger microspheres were overrepresented in regions of greater flows no matter where they were anatomically. The explanation given was based on arguments provided by Fung (8). T h e r e is a t h i r d possible reason why microspheres might fail to measure flows to small regions correctly, namely, regional variations in nontrapping of microspheres. The fraction of microspheres that fail to be trapped can be measured for whole organs (3) but not for regions within organs. There is thus a possibility for error with the smaller microspheres that, for other reasons, might be more useful for measuring regional myocardial flows. Diffusible indicators avoid these problems but add others. Rubidium has a clearance that varies with changes in flow rate (9). The inert indicators can be used to measure regional flows provided the time course of arterial concentration of the indicator, the partition coefficient, the tissue concentration, and the rate of equilibration of indicator in blood and tissue are known (10, 11). The inert indicator method has been used to measure regional blood flows within the brain (11, 12) and the heart (7), but validation of the results by an independent method has not yet been done. Furthermore, many assumptions used with these indicator methods have not been evaluated in previous studies. We therefore designed experiments to determine the ability of microspheres of different diameters to measure total and regional myocardial blood flows. In addition, we compared these results with those obtained using antipyrine infusions and made a theoretical analysis of some of the problems inherent in the antipyrine method.

Methods Microspheres. — Microspheres with nominal mean diameters of 25/x, 15/*, and 9/x were obtained from the Minnesota Mining and Manufacturing Company; they were labeled with 125I, 141Ce, 85Sr, and 46Sc. The diameters were measured with an ocular micrometer, and the means ± SD of different batches were 25.5 ± 3.5/x and 28.1 ± 2.7/;., 15.3 ± 2.7/t and 18.5 ± 2.1/Lt, and 9.28 ± 0.98/t and 9.06 ± 0.84/*. In some experiments, we obtained microspheres ranging in diameter from 1-10/t (about 40-80% were 7/* or less in diameter) and filtered them through 8/x Nuclepore filters (General Electric); varying amounts (usually 25-31%) of the retained microspheres were under 7/t in diameter. Microspheres subjected to this process will be referred to as filtered 1-10/* microspheres. All the batches of microspheres were suspended in saline to which Tween-80 had been added to a final concentration of 0.5% to reduce aggregation. Before injection, the microspheres were shaken and ultrasonicated to break up clumps, and the absence of aggregation was confirmed by microscopy before each injection. With each batch of microspheres the supernatant liquid was checked periodically for eluted radioactivity; none was found. With each injection, 500,000-3,000,000 microspheres were given. Animal Preparations.—Sheep and mongrel dogs were anesthetized with Diabutal (20-30 mg/kg, iv) and ventilated with room air by an endotracheal tube and a Harvard respirator. Catheters were placed in the left atrium for injection of the microspheres through a thoracotomy (two left atrial catheters were placed if antipyrine was also infused), in the femoral vein, and in either a brachial or a femoral artery for withdrawal of an arterial reference sample (4). In some animals, reference samples were also collected from the coronary sinus from a catheter placed via the left hemiazygos vein in sheep or through the wall of the coronary sinus in dogs; transit times to the right atrium and the coronary sinus of indocyanine green injected into the femoral vein showed that the coronary sinus sample contained no right atrial blood (13). Both arterial and coronary sinus reference samples were drawn at steady rates of 13-15 ml/min for 1 or 1.5 minutes by Holter pumps (Extracorporeal Medical Specialities). The sample collections began just before the microspheres were injected into the left atrium through specially designed vials with 5-10 ml of saline warmed to body temperature; the injection lasted about 20 seconds and did not cause arrhythmias or changes in blood pressure. Batches of microspheres of different diameters were injected simultaneously from separate vials connected in parallel. To change total and regional myocardial blood flows, we used a variety of maneuvers: blood pressure was kept at a mean of 35 mm Hg for 2 hours, adenosine triphosphate was infused into the left coronary artery at a rate of 0.76 mg/min to cause vasodilation, methoxamine was infused intravenously to obtain peak systemic systolic arCirculation Research. Vol XXXIV, March 1974


393

REGIONAL MYOCARDIAL FLOW BY INDICATORS

terial blood pressures of 200-250 mm Hg, the atrium was paced at rates of 180-200 beats/min, the aorta was constricted just above the aortic valve to produce peak systolic left ventricular pressures of 200-230 mm Hg and cause subendocardial ischemia (14), and the left anterior descending and circumflex coronary arteries beyond their first major branches were ligated to produce regional ischemia. With all these interventions, coronary blood flows were measured when heart rate and blood pressure were stable. Another 50 dogs were prepared similarly and placed on right heart bypass (4). A drain from the right ventricle to the venous reservoir could be diverted to a measuring cylinder for timed collection of coronary venous blood during microsphere injections. On 14 occasions the collected total coronary venous drainage was examined to determine the amount of radioactivity not trapped in the heart. At the end of the experiments, the animal was killed with Diabutal, the heart was excised, the epicardial fat and large vessels were removed, and pieces of myocardium were placed in plastic vials for gamma counting. The heart was dissected to yield subendocardial, midmyocardial, and subepicardial layers of the left ventricular free wall, the left and right sides of the ventricular septum, and the right ventricular free wall, the latter sometimes being cut into subendocardial and subepicardial layers. In five hearts the free wall of the left ventricle was first cut into three subdivisions. The lower third of the ventricle and apex formed one subdivision, and the remaining wide basal zone was cut into anterior and posterior halves. Each subdivision was then divided into three layers, as above. Tissues and blood samples were analyzed by determining counts in the region of interest of the photoelectric peak of each nuclide with a multichannel pulse-height analyzer (Nuclear Chicago) and vial changer. The total counts in the region of interest of each nuclide, the weight of each specimen, the flow rate of blood in the arterial reference samples, the background counts, and the estimated number of counts per minute per microsphere (determined by us on receipt of the microspheres and corrected for subsequent decay) were entered on IBM punch cards and then processed on an IBM 360 computer. The percent of total counts for each nuclide in each region of interest was corrected for the height of tissue in the counting vial, and corrections for the efficiency of counting tissues and blood were included in the computer program. Calculations.—Total myocardial blood flow (Qh[m\/ min] ) and regional myocardial blood flow (Q([ml/min] ) were calculated from the measured counts per minute for each nuclide in the heart or region (lh or /„ respectively), the counts per minute in the arterial reference sample (Zor), and the flow rate of the arterial reference sample (@ar[ml/min] ). QH= Qar- hi Lr, (D

If coronary sinus sampling at a rate of Qcr{m\/min) contained 7er counts per minute then total myocardial flow could be corrected for incomplete trapping of microspheres by

Qh = h/(Iar/Qar-hrIQcr).

(3)

This correction is based on the fraction of microspheres that are not trapped in an organ and so appear in its venous drainage (3). Antipyrine Flows. — Antipyrine was infused into the left atrium through a separate catheter, and the infusion began 0-40 seconds after the microspheres were injected; reference samples for the microspheres were collected as described above. The arterial concentration of antipyrine was obtained from a multihole catheter inserted through the carotid artery and placed just above the aortic valves. Aortic blood was withdrawn with a constant-flow Sarns pump at a rate of 80 ml/min and collected at 2-second intervals in a fraction collector (Harvard Instruments). When the antipyrine infusion was ended 20-60 seconds after it had begun, the heart was rapidly removed (within 15 seconds) and divided as quickly as possible into pieces as described above. Flow for each region was determined from microspheres by gamma counting as previously described. Antipyrine concentrations were determined in two ways. 125I-Antipyrine was measured by gamma counting and a modification of the computer program. With u C-antipyrine, the pieces of myocardium were dissolved in 2N methanolic potassium hydroxide and samples were counted for beta emission in a liquid scintillation counter before and after standard amounts of 14C-toluene were added. Only the supernatant liquid was counted to avoid errors from beta emissions of the microspheres. Antipyrine in the blood samples was determined after they had been bleached with hydrogen peroxide and 25% acetic acid in a 75°C oven for 1 hour. For calculations, we used the technique described by Reivich et al. (12) but collected samples at 2- rather than 6-second intervals; for the smearing correction, we used a first-order equation instead of the second-order equation used by Reivich et al. (12). A separate computer program was used to determine the flow to each region of the myocardium (12). The partition coefficient, the time course of the arterial concentration, the catheter transport function, and the concentration of antipyrine in each piece of tissue were entered into the computer, which then printed out the values for flow per gram necessary to give the observed concentrations. Autoradiographs. — Autoradiographs were made from two additional dog hearts after thoracotomy. In one dog, a bolus of dissolved 133Xe was injected into the left coronary artery, and the heart was rapidly removed 15 seconds later and plunged into acetone and Dry Ice. In the other dog, 14C-antipyrine was infused at a constant rate into the left coronary artery and the heart was rapidly removed and frozen 20 seconds later. After

Circulation Research, Vol. XXXIV, March 1974



394

600r

TABLE 1

Percent of Untrapped Microspheres t

Diameter

% Not trapped

N

Q UJ

Mean

Range

27.6

10.2-49.0 2.9-17.5 0.0-2.4 0.0 - 3.8 0.0-2.2

t z> 4 0 0 -

6.3 1.1 1.1 0.4

5

01

9 15 25

3 8 5 13 7

CAL

(1

1-10 unfiltered l-IO filtered

8 200-

>.•••

N = number of observations.

freezing was complete, the hearts were sliced with a band saw and the surfaces were smoothed; the left ventricular cross sections were placed in contact with a photographic emulsion (type A Industrial x-ray film) in a container of Dry Ice. After appropriate exposures, the films were developed.

Results Comparisons of Calculated and Measured Total Coronary Blood Flow.— Figure 1 gives the results of 250 comparisons made during right heart bypass and includes 102 measurements reported before (2, 4). The equation relating calculated flow (Y) to measured coronary venous drainage (X) was Y = 1.00446X; the equation was fitted by a line passing through the origin with SD proportional to X. The slope was not significantly different from one and the percent SD was 12.97%; 66% and 90% of the calculated flows were within 10% and 20%, respectively, of the measured flows. There were no differences for microspheres with diameters ranging from 9/A to 61/x. Total Coronary Flow and Trapping of Microspheres.—The percents of microspheres not trapped by the myocardium draining into the coronary sinus were estimated by the method of Archie et al. (3) (Table 1). When, in separate experiments, untrapped microspheres were counted in the total coronary venous drainage, the percent untrapped ranged from 0.8 to 1.1% in 3 studies with 15/x diameter microspheres and from 0.6 to 3.2% in 11 studies with 9/JL diameter microspheres. In these studies, the coronary arteries were maximally dilated by acetylcholine infusion. When coronary blood flows were measured by simultaneous injection of 25fju and 15^ or 15/n and 9^. diameter microspheres, the regression lines relating them passed through the origin and the slopes did not differ significantly from the line of identity. How-

O a. o o 0

200

400

600

CORONARY BLOOD FLOW MEASURED :(ml • min"') FIGURE 1

Measured coronary sinus drainage in 250 right heart bypasses plotted against coronary blood flow calculated from microsphere injections. The solid center line is the line of identity; the dashed and dotted lines are 10% and 20% deviations from the line of identity, respectively.

ever, in comparisons of the 15/u. and either the filtered or the unfiltered 1-10/Lt diameter microspheres, the smaller microspheres underestimated flow; this discrepancy disappeared when allowance was made for the fraction of untrapped microspheres. Flow to Ventricular Free Walls and Septum.— To eliminate variability due to the distribution of microspheres in the reference samples, we added the counts in each ventricular free wall and those in the septum and calculated the percent of this total in each of the three ventricular regions. The percent distribution of microspheres of a given diameter to these three regions was then compared with the percent distribution obtained from simultaneous injections of microspheres of another diameter. The mean differences were evaluated by paired f-tests. Differences between microsphere distributions and proportional flows derived from antipyrine were similarly compared (Table 2). The mean differences were small, and with one exception they did not significantly differ from zero. To determine if the percent distribution of microspheres to these large regions depended on the magnitudes of flows to the regions, we calculated the regression lines for each region relating the percent differences in flow measured by Circulation Research, VoL XXXIV. March 1974



395 TABLE 2

Differences in Percent Distribution of Indicators to the Ventricular Free Walls and Septum Comparisons Region

LV

25 vs. 15 25 vs. 1-10 filtered

S

-0.25 (0.28)

1.07 (0.75) -0.90" (0.46)

RV

-0.25 (0.45)

-0.15 (0.39)

13

12

N

0.50 (0.72)

15 vs. 9

15 vs. 1-10 25 vs. 1-10 15 vs. 1-10 25 vs. AP filtered unfiltered unfiltered

15 vs. AP

1-10 filtered vs. AP

-0.29 (0.99) -0.04 (0.36)

0.28 (0.28) -0.45° (0.23)

0.06 (0.26) 0.71 (0.32)

-0.37 (1.0) 0.57 (0.56)

-0.33 (0.98) -1.88 (1.80)

-0.50 (2.07) -0.38 (1.55)

-0.14 (0.57) -1.26 (1.41)

0.31 (0.57)

0.20 (0.25)

-0.77* (0.06)

-0.17 (0.41)

2.25 (1.05)

0.92 (1.23)

1.43 (1.03)

8

13

3

3

6

8

7

Values are mean differences in percents to each region, the values for the smaller microspheres being subtracted from those for the larger microspheres and those for antipyrine being subtracted from those for microspheres; LV and RV = free walls of left and right ventricles, respectively, S = septum, N = number of comparisons, and AP = antipyrine. Numbers at the heads of columns are microsphere diameters in microns. Standard deviation of mean difference is given in parentheses. °0.10 >P>0.05. 1 0.0>P.

microspheres of two different diameters to the actual average percent flow to that region. In each comparison, the slope of the regression line was close to and not significantly different from zero. Flow to Large Subdivisions of the Left Ventricle.—The left anterior descending and circumflex coronary arteries of two sheep were occluded prior to injection of microspheres of different sizes (25/A, 15/x, and filtered 1-10/x diameter in one; 15/x and 9/x diameter in the other). In the sheep in which coronary sinus blood was sampled, the fractions of 15/x and 9/x diameter microspheres passing through the myocardium were 2.88% and 2.45%, respectively; these values are among the largest for microspheres of these diameters, but they still are not large in absolute terms. The fraction of left ventricular flow going to the ischemic zones was the same for microspheres of all different sizes. One of these sheep had markedly reduced flows to six pieces of subendocardial and five pieces of subepicardial muscle (this wall was cut with only two layers). In these pieces of subendocardial muscle, proportional antipyrine flow averaged 0.54% above the p e r c e n t flow measured with microspheres (PP. +0.05 >P>0.01. *0.10>/>>0.05.

in the midmyocardial layer, and the microspheres had a slight excess in subepicardial muscle. The subendocardial antipyrine excess was significant in one comparison and nearly so in another. No relationship of these differences to subendocardialsubepicardial flow ratios was found. Although 9fi and filtered 1-10/A diameter microspheres were not compared directly, the differences of each from the 15/x diameter microspheres were compared by an unpaired ttest; no significant difference was found. In eight experiments in which we measured the fraction of filtered l-l0fi diameter microspheres emerging in the coronary venous blood, the relative excess in the subendocardial layer of 15/x diameter microspheres was not related to the fraction of small microspheres that were not trapped in the myocardium. This finding was also true for three comparisons between 15/x and unfiltered 1-lOfi diameter microspheres. Microsphere Distribution and Relative Flow Rates. — To examine further the possibility that the differences in the distributions of microspheres of different diameters might depend on relative regional flows rather than on regional anatomy, we first calculated the ratio of flow per gram in the total subendocardial and subepicardial layers of

each left ventricular free wall. This flow ratio ranged from 0.36 to 3.52 and was calculated from the smallest microsphere size used for that injection. Then we compared the percent distributions of proportional flows of simultaneously injected pairs of microspheres of different diameters to the subendocardial, midmyocardial, and subepicardial layers for those with subendocardial-subepicardial flow ratios that were low (under 0.8), normal (0.8-1.2), and high (over 1.2). There was a significant interaction between relative regional flows and the distribution of 25/A compared with 15/n diameter microspheres (Table 4). Compared with 15/n diameter microspheres, the percent of 25/A diameter microspheres in the subendocardial layer was less for the hearts with low flow ratios and more for the hearts with high flow ratios. Some tendency for interaction was also shown by comparing 25/u. and filtered l-10/oi diameter microspheres, since the excess of larger microspheres in the subendocardial layer was greater at high flow ratios than it was at low flow ratios. Comparing 15/A diameter microspheres with 9/LA or filtered l-l0fi diameter microspheres, the overestimate of subendocardial flow by larger microspheres seemed to be independent of the relative flow ratios. Circulation Research, Vol. XXXIV, March 1974


397

REGIONAL MYOCARDIAL FLOW BY INDICATORS TABLE 4

Distribution of Microspheres Related to Microsphere Diameters and Relative Flows Subendocardial-subepicardial flow ratios

Diameter

to

Layers

Interaction

3.20 1.10 2.97

25 vs. 15

1.88

10.21°

3.96 -2.46 0.97

8.36 2.07 6.19

25 vs. 1-10 filtered

14.67°

3.36*

5.58 1.20 -2.51

1.26 -2.15 2.61

3.88 1.01 3.48

15 vs. 1-10 filtered

12.84°

7.00 -0.90 -2.44

10.53 -1.86 -7.12

2.90 3.93 6.24

15 vs. 9

15.61°

< 0.8

0.8-1.2

Subendocardial Midmyocardial Subepicardial

-3.54 0.78 1.04

2.64 -0.19 -1.53

Subendocardial Midmyocardial Subepicardial

3.04 1.84 -1.74

Subendocardial Midmyocardial Subepicardial Subendocardial Midymyocardial Subepicardial

P. t0.10>P>0.05.

TABLE 5

Variability within Layers.—In two dogs the layers of the left ventricular free wall were cut up into many small pieces—41 in one dog and 21 in the other. The respective subendocardial-subepicardial flow ratios were 3.52 and 0.50. For each piece of tissue, the flow as a percent of total flow to the free wall of the left ventricle was divided by the weight of that piece as a percent of the total weight of the free wall of the left ventricle and multiplied by 100; a value under 100% indicates that flow to that piece was below the average for the free wall of the left ventricle. For each layer of each heart, the distribution of these percents was examined (Table 5). There was marked variability of flow within a layer, and in seven of the nine layers examined the observed SD was considerably greater than the square root of the mean percent for that layer. This variability was also shown in two other hearts studied by autoradiography. Adjacent dark and light areas with no specific anatomical distribution were seen with 133Xe washout as well as with 14 C-antipyrine washin. Flow in Layers of Septum and Right Ventricle.— We examined flow to the left and right sides of the septum, which have been shown previously to act like the subendocardial and subepicardial layers, respectively, of the free wall of the left ventricle (15). The proportional flow to the left side of the

Variability of Flow within Layers of the Left Ventricular Free Wall Percent flow x 100/percent

weight Region

Dog

1

2

Range

SO

93 3.2 25 15 21 6 15 21 9 18

Subendocardial b Midmyocardial Subepicardial

78 60

91-418 9.5-20.4 54-131 34-80

Subendocardia! Midmyocardial Subepicardial Subendocardial Midmyocardial Subepicardial

63 93 121 56 86 127

31-91 86-104 93-139 29-88 69-94 90-143

a

2

Mean

215

14.3

In dog 1 the subendocardial distributions were markedly skewed so that the actual values (a) are accompanied by a square root transformation of the data (b). Dog 2 had two pairs of microspheres injected without any interventions between them.

septum averaged 0.60% greater for 15/z. than for filtered l-10fi diameter microspheres (P< 0.05) independent of the subendocardial-subepicardial flow ratios. A similar but not significant difference

Circulation Research. VoL XXXIV. March 1974



398

occurred for 25/A compared with filtered 1-10/u. diameter microspheres; however, when 25/LA and 15^ diameter microspheres were compared, a slightly but not significantly lower proportion of the larger microspheres went to the left side of the septum. There was a suggestion that more of the larger microspheres went to the left side of the septum when the flow ratio was high than when it was low, but the differences were not significant. When the subendocardial and subepicardial layers of the right ventricle were examined in three dogs, no differences in the distribution of microspheres of different sizes to these layers was noted. Antipyrine and Microsphere Distributions in Left Ventricular Subregions.—In each of eight animals, we compared the percents of flow determined by simultaneously injected indicators (either antipyrine and small microspheres, or two different sizes of microspheres) going to each piece of left ventricular free wall. The number of pieces per animal ranged from 10 to 38. For each comparison we determined the line of best fit by a linear leastsquares method, the correlation coefficient, and the coefficient of variation (SD from regression as a percent of the mean value of Y). For the antipyrine and microsphere comparisons (one with 9/it and the rest with filtered 1-10/A diameter microspheres), the correlation coefficients ranged from 0.7209 to 0.9904 and the coefficients of variation from 11.69% to 52.64%. In each animal, the comparison of one of these smaller microspheres with 15/u. diameter microspheres gave higher correlation coefficients which ranged from 0.8700 to 0.9964 and lower coefficients of variation which ranged from 6.54% to 20.07%, respectively. Furthermore, in comparing different sizes of microspheres, each regression line had a slope not significantly different from one and an intercept not significantly different from zero. On the other hand, in four of the eight antipyrine comparisons, the slope was significantly below one and the intercept significantly above zero. Discussion Total Coronary Flow.—We confirmed previous findings (2, 4) that total coronary blood flow can be measured with an error of usually under 20% and in about two-thirds of studies under 10%; no bias in the measurements was shown. Nontrapping of microspheres above 7/x in diameter by the heart is usually below 3% whether it is measured directly or indirectly (3); the percent of trapping is not significantly altered by stressing the coronary circula-

tion. Thus, coronary flow can be measured equally well by any size of microsphere provided that the fraction of microspheres not trapped by the myocardium can be estimated. For convenience, microspheres above 7fi in diameter that are almost all trapped should be used. Flow to Large Regions.—Since there were no differences in the relative distribution of microspheres of any of the diameters injected to each ventricular free wall and the septum or to subdivisions of the left ventricle, including ischemic zones, the distribution of well-mixed microspheres into the right, septal, left anterior descending, and left circumflex coronary arteries was independent of microsphere size. These findings were expected, since there is not time for axial streaming at the orifices of these large arteries. Table 2 also suggests that nontrapping of the smallest microspheres after injecting 1-lOfj. diameter microspheres was uniform for each ventricular wall and the septum. Antipyrine flow distribution matched that of the microspheres and supports the contention that all of these indicators can accurately measure flow to these large regions. When there were ischemic zones, however, slight differences between antipyrine and microsphere flow distributions were noted. In both sheep, proportional antipyrine flow exceeded proportional microsphere flow in subendocardial muscle, probably because of passage of antipyrine into subendocardial muscle from the ventricular cavity (16, 17). In one sheep, too, there were minute amounts of antipyrine in outer muscle layers when no microspheres were present. This finding could indicate the inaccuracy expected when microspheres are used to measure extremely low flows or else the passage of antipyrine through small channels that do not allow passage of microspheres. The error was negligible in these acute studies. We emphasize, however, that these studies do not give evidence that microspheres can measure flow to an ischemic area after collaterals have developed. Flow within Layers of the Left Ventricular Free Wall.—The larger of any pair of microspheres tended to be present in greater percent in the subendocardial layer and in lower percent in the subepicardial layer. These differences were not significant for the three experiments with the unfiltered l-10pi diameter microspheres, probably because of the small number of observations since the differences were the largest seen. In the experiments with the smallest microspheres (filtered Circulation Research, Vol. XXXIV. March 1974



and unfiltered 1-10/x diameter), the relative overrepresentation of the larger paired microsphere (25/A or 15/x) in the subendocardial layer was unrelated to the fraction of small microspheres that were not trapped. This finding suggests that the degree of nontrapping was similar in the different layers of the free wall of the left ventricle, since, if a large fraction was untrapped, selective passage through one layer would have shown up as marked maldistribution. Differential trapping is thus not a cause of the different distributions of batches of microspheres of different diameters. One comparison, that between 25/J. and 15/u. diameter microspheres, showed small, nonsignificant differences in distribution of the two sizes. When examined further in relation to subendocardial-subepicardial flow ratios, the proportional distribution of 25(JL diameter microspheres exceeded that of 15/x diameter microspheres in the regions of highest flows, whether these were subendocardial or subepicardial. However, when flows were similar in the two layers, the larger microspheres appeared in slightly greater proportion in the subendocardial layer (Table 4). To some extent these results confirm what Yipintsoi et al. (7) have found comparing 35/u, and 15/A diameter microspheres. This relationship of anatomical distribution to relative flow rate was less clear-cut when comparing 25/x and filtered 1-lOju. diameter microspheres, since the larger microspheres were always overrepresented in the subendocardial layer; however, the degree of overrepresentation varied with the flow ratio as shown by an almost significant interaction term. For microspheres 15/x or less in diameter, there was always proportionately more of the larger of the pair of microspheres in the subendocardium, independent of the flow ratio. These findings suggest that forces which act on microspheres 35/u, and 25/x in diameter might be less important for microspheres 15/n. in diameter or less. It is also possible that the distribution of microspheres of different diameters in the free wall of the left ventricle is a function not only of the geometry of the vasculature and the relative flow rates to different layers but also of the actual velocities of flow in each layer. Since subendocardial muscle probably receives all or almost all of its flow in diastole (14), then in diastole the subendocardial flow rate usually exceeds that in subepicardial muscle. Our experiments did not allow us to examine this other possible source of variability of distribution. Our antipyrine studies showed a slight excess of

399

proportional antipyrine flows in subendocardial muscle irrespective of the relative subendocardial and subepicardial flows. The magnitude of the excess is consistent with passage of antipyrine from the ventricular cavity (16,17) and does not by itself indicate inaccuracy of the microsphere method. The data in Table 4 indicate that, when smaller microspheres (9/x or filtered 1-10/u. diameters) are compared with either 25/u or 15/A diameter microspheres, the larger microspheres could give left ventricular subendocardial flows up to 11% greater than those calculated from the smaller ones and subepicardial flows up to 7% less than those calculated from the smaller microspheres. These data give subendocardial-subepicardial flow ratios per gram that are 5-10% greater for the larger microspheres than they are for the smaller microspheres. We conclude that microspheres over 25fi. in diameter may not measure flow to small myocardial regions very accurately, since their distribution in different layers of the wall may be influenced by relative flows in these layers. Whether 9JU. or 15/x diameter microspheres give the true flow values cannot be determined without an independent method, but it is reasonable to suggest that the smaller the microsphere the more it is likely to be distributed like red blood cells. If this assumption is true, the 9/i. diameter microspheres might well be the best tracers presently available for regional myocardial blood flow, since they are almost all trapped in the myocardium and the degree of nontrapping does not differ in different regions of the heart. Flow to Layers of the Right Ventricle or Septum.—We did not have enough experiments in which the right ventricle was divided into layers to decide the relative merits of microspheres of different diameters in measuring regional flow in this ventricle. However, the septum behaved like the free wall of the left ventricle in that larger microspheres tended to overestimate flow to the left ventricular side of the septum. Since the septal artery in dogs runs near the right ventricular margin of the septum, the small branches coming off it supply the septum in a manner anatomically similar to the way in which the branches of the left anterior descending and circumflex coronary arteries supply the free wall of the left ventricle. The difference in distribution of microspheres of different diameters in the septum may thus depend on the same forces as those which act in the free wall of the left ventricle. Variability within Layers.—In any one layer of

Circulation Research, Vol. XXXIV, March 1974


400


the left ventricular free wall, there was marked variability of flow to adjacent areas (Table 5). With mere random variability we would have expected the SD to approximate the square root of the mean. The fact that it was usually much greater needs explanation. In these studies there was a minimum of 800 microspheres in each piece of the heart, and usually the number was over 1,500 per piece; therefore, there should have been enough microspheres to reduce the random variability of the method to under 5% (2). Even if there had been small clusters of microspheres going to any one small region, the total number should have been sufficient to damp out variations due to this cause. Thus, either there is a cause of nonrandom distribution within each layer that makes adjacent pieces of tissue get microspheres out of proportion to local blood flow or there are considerable variations in blood flow within adjacent small regions. Not only is the former unlikely, but this study and the study by Yipintsoi et al. (7) found similar variability of flows with microsphere and diffusible indicators. Different small regions of the left ventricle may thus not only have different flows at any one moment but also may be doing different amounts of work; similar observations have been made in skeletal muscle (18). The implications of this finding for the measurement of How by diffusible indicators will be discussed below, but other implications should be pointed out. If some small regions receive little flow, then measurements of tissue extravascular space by indicators like sucrose or inulin that leave the vascular bed might be influenced by what proportion of the bed is fully open at any one time. This phenomenon might explain the oscillations seen in inulin concentrations in coronary sinus blood by Klassen (19). A second implication is that, if some of the muscle is not working as hard as an adjacent portion of muscle, then this nonworking or slightly working muscle will contribute to part of the series elasticity of the heart. Finally, this finding suggests that recruitment of working portions of the ventricle can occur and that the ability of the heart to perform more work from the same end-diastolic fiber length and tension can be achieved either by an increased contractility of each muscle fiber or else by recruitment of more muscle fibers without any increase in contractility. ANALYSIS OF ANTIPYRINE TECHNIQUES

There was a close relationship between flows measured by antipyrine and microspheres in large

regions of the heart, in layers of the left ventricle, and in ischemic and nonischemic regions of the left ventricle. Although there was very close agreement between the fractional distributions of antipyrine-measured and microsphere-measured flows in pieces of the left ventricular free wall, the agreement between proportional antipyrinemeasured and microsphere-measured flows was always less good than that between flows measured with pairs of microspheres of different sizes. This finding demonstrates the precision of microsphere measurements, not their accuracy, and does not tell us which method measures regional blood flows correctly. Since diffusible indicators are theoretically capable of accurately measuring the flow to small regions (10, 11), we had hoped that these comparisons would allow us to decide which size of microscope —if any—measured regional myocardial blood flows correctly. However, we found that several factors could interfere with the accuracy of the antipyrine method. Since these factors have not been discussed adequately in the literature, we have attempted to analyze their importance. Choice of Indicator.—Antipyrine is a lipophilic, freely diffusible indicator (20) that has been regarded as flow limited in skeletal muscle (21) and heart (22). It has been labeled with radioactive iodine and 14C, with respective molecular weights of 318 and 188. Yipintsoi and Bassingthwaighte (23) compared these two forms of antipyrine and concluded that they gave similar flow measurements over a wide range of coronary flows. There are, however, potential difficulties with iodoantipyrine. It slowly dissociates into free iodine and antipyrine; the rate increases with temperature, light, and time (24, 25). With substantial amounts of free iodine, there might be errors relating radioactivity to indicator concentration (21); therefore, preliminary testing for and separation of any free iodine is essential. Furthermore, in some species iodoantipyrine breaks down rapidly in the tissues (26, 27) so that infusion of pure iodoantipyrine is no guarantee of its value as a diffusible indicator. For these reasons we did most of our experiments with l4 C-antipyrine. Partition Coefficient—By definition, the partition coefficient \i/j is the concentration in phase i divided by the concentration in phase j at equilibrium (28). Concentrations are quantities per unit volume so that for any indicator its equilibrium distribution per unit volume of tissue and blood must be determined. The concentration of an indicator in blood is easily measured, but that in Circulation Research, Vol. XXXIV. March 1974



401

tissues is obtained with more difficulty. Although tissue volume can be calculated readily from tissue weight and specific gravity, the calculated total tissue volume and the volume of distribution of the indicator are not necessarily synonymous. Antipyrine was originally used to measure total body water (29, 30), and there is added evidence that the volumes of distribution of 14C-antipyrine and tritiated water are identical in the lung (31). However, the volumes of distribution of 14C-antipyrine and tritiated water may not always coincide, and mean coronary transit time at a given flow is less for water than it is for 14C-antipyrine or iodoantipyrine (23). Furthermore, Effros and Chinard (31) have noted that, when blood is incubated with 14C-antipyrine and tritiated water, the red cells contain about 10% more antipyrine than expected from the tritiated water concentrations in cells and plasma. Presumably, antipyrine might be concentrated in cell lipid or protein. In fact, antipyrine has been found to be concentrated by mitochondrial pellets, even after brief exposure at 0°C (32). This finding is particularly important, because mitochondria form about one third of heart muscle. On the other hand, O'Brien and Brierley (33) have shown that isolated beef heart mitochondria may have two compartments, one of which is impermeable to most solutes. The volumes of distribution of antipyrine and tritiated water in tissue and blood may thus not be the same after several minutes as they are after several seconds. Even if we assume that antipyrine is not barrier limited, there is still the relationship between hematocrit and partition coefficient to be considered. By definition, the partition coefficients of water for red cells (c) and plasma (p) and for muscle {t) and plasma are

KP = (Qc/vc)/(Qp/vp), *,P = (Q,/v,)/(g p /v p ),

(4) (5)

where Q is the quantity of water in V, the volume under consideration. Then the partition coefficient of water for muscle and whole blood (fo) is *