INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

DISTRIBUTION OF REGIONAL myocardial blood flow is strongly heterogeneous (1, 32). This is not merely a result of chance but has been shown to represent the adaptation to regionally different substrate needs (7). Relative dispersion (RD; SD/mean) is a conventionally used measure of spatial perfusion heterogeneity. However, this parameter is dependent on the spatial resolution of the measurement; i.e., RD increases, obeying a power law if spatial resolution of the measurement is increased (1, 32). Comparison of data obtained with different techniques at different scales is therefore inadequate. Clinical perfusion measurements have a much lower resolution than many experimental methods, and technical advances will allow an increase in spatial discrimination of perfusion. These differences in resolution complicate comparison of data from different studies and extrapolation of experimental results to human physiology.

The dilemma has been resolved by the introduction of fractal analysis of myocardial blood flow heterogeneity (1, 32). Fractal analysis relates the increasing resolution of a measurement with the concomitant increase in RD. This relationship obeys a power law. A single parameter, the fractal dimension (D), can therefore be determined to describe this relationship. D is calculated as 1 minus the slope of the linear relationship between the natural logarithms of measurement resolution and perfusion heterogeneity measured by RD.

The formal relationship is given in Eq. 1

(1)

The natural logarithm of the relationship of RD measured at some resolution (i.e., mass of myocardial samples; RD_m) and RD measured with the original sample mass [RD_(m0)] equals the natural logarithm of the relationship of the respective masses m and m₀ times 1  D. If 1  D is the slope of the relationship, D = 1  slope and can empirically be found from a linear equation obtained by linear regression analysis.

Fractal properties of the coronary vascular tree have been demonstrated (19, 31), and it is not surprising that the RD of regional myocardial perfusion (RMP) is also fractal. Therefore, heterogeneity of RMP lends itself to fractal analysis. Fractal analysis is increasingly accepted to further elucidate the physiology of myocardial blood flow in experimental animals (6, 18, 26, 27). Absolute regional myocardial blood flow can be measured in humans, and heterogeneity of RMP is recognized to be of relevance in coronary artery disease (5, 9). Visser et al. (33) have calculated the D of RMP of syndrome X patients and compared these data with data from healthy volunteers. They have shown that D is reduced in syndrome X patients.

The D of RMP heterogeneity, as calculated from an increasing RD, may vary from 1.0 to 1.5 (1). A D value of 1.0 denotes uniform distribution or RMP balanced on all scales of measurement, whereas a D value of 1.5 shows that perfusion heterogeneity increases greatly with increased resolution and that spatially neighboring myocardial regions display uncorrelated rates of perfusion. A reduction of D can also be interpreted as a homogenization of RMP.

Ischemic heart disease is a common condition in surgical patients (23, 23). Therefore, the combination of both coronary stenosis and hypotension or shock is likely to be common in accidental or surgical trauma. The model described in the present report attempts to model this situation in a standardized way. In a previous study we have investigated the effects of this deleterious combination on fractal heterogeneity of RMP for the first time (22). We have shown that hemorrhagic shock decreases the D of RMP. Surprisingly, it was found that this effect was not reversed on restoration of hemodynamic determinants of coronary perfusion by means of small-volume resuscitation or conventional fluid therapy (22).

Infusion of diaspirin-cross-linked hemoglobin (DCLHb) has been demonstrated to attenuate postischemic reperfusion injury in hamster muscular tissue (28). Mortality from hemorrhagic shock in rats was reduced to zero by resuscitation with a hemoglobin-based oxygen carrier (HBOC), whereas conventional fluid therapy did not achieve similar results (4). Fluid infusion for treatment of hypovolemia dilutes endogenous hemoglobin and therefore reduces the oxygen transport capacity of blood. Although extreme hemodilution with a HBOC preserves muscle tissue oxygenation, hemodilution to the same degree with conventional colloids is associated with tissue hypoxia (30). Furthermore, a HBOC can increase oxygen delivery to poststenotic tissues beyond the amount supplied by blood without a HBOC (16).

We therefore hypothesized that resuscitation from hemorrhagic shock with DCLHb would be more effective than small-volume resuscitation to reverse shock-induced alterations of the distribution of RMP.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This paper represents a further analysis of data from our study published by Habler et al. (13). The aforementioned study does not contain data on the heterogeneity of RMP. To avoid repetition, macrohemodynamic parameters are only briefly summarized.

Animal preparation. We used an animal preparation and technical procedures that have been described before in detail (22, 34). Animals were housed and treated in conformance with German animal-protection legislation. Twenty experiments were performed in open-chest pigs weighing 26 ± 4.2 (SD) kg. The animals were intravenously anesthetized with an opioid [48 µg · kg body wt (BW)¹ · h¹ fentanyl] and a benzodiazepine (1.2 mg midazolam · kg BW¹ · h¹), paralyzed with 4 mg/h pancuronium, and mechanically ventilated with 50% oxygen in air. The catheters listed below were inserted into femoral vessels, carotid arteries, or external jugular veins, respectively, after surgical exposure. Animals were instrumented with two micromanometers (PC 370 Millar Instruments, Houston, TX) to measure left ventricular (LV) and aortic pressure. Cardiac output and left anterior descending coronary artery (LAD) flow were assessed with ultrasonic flow probes (diameter 14 and 2 mm; Transonic, Ithaca, NY). For injection of microspheres, a catheter (5F, Arrow, Reading, PA) was inserted into the left atrium through a small incision and secured in place with a circular suture. A critical coronary stenosis was induced by micromanipulator-assisted narrowing of the LAD with a thin, Teflon-coated copper-wire sling distal to the coronary flow probe. Absence of hyperemic response after complete occlusion with a second, more distal, copper wire for 10 s was controlled for to verify the critical degree of stenosis (10).

Microsphere methodology. RMP was measured with radioactive microspheres. The reference-sample technique was used, which allows quantitative determinations (15). For each measurement, one nuclide was randomly selected (⁸⁵Sr, ¹¹⁴In, ⁹⁵Nb, ¹⁴¹Ce, ⁴⁶Sc; diameter 15 ± 0.1 µm; NEN-TRAC, DuPont, Wilmington, DE). The number of microspheres labeled with an individual nuclide in sequential measurements was optimized to reduce methodological error by using specialized software and amounted to 4.8 × 10⁶ ± 0.7 × 10⁶ microspheres/measurement (20). Microspheres were injected slowly into the left atrium while the reference sample was drawn for 3 min from the abdominal aorta through a pigtail catheter (7F, Cordis, Miami, FL) with a calibrated pump at 3.24 ml/min (model 640A, Harvard Apparatus, South Natick, MA). After completion of the protocol, the animals were killed by left atrial injection of potassium chloride. The hearts were removed and fixed in 10% formaldehyde for 4-5 days. Other organs were removed but not analyzed for perfusion heterogeneity. The isolated LV myocardium was dissected into six slices parallel to the base of the heart. Four slices were cut into 12 radial sectors; the most apical slices were dissected into only 10 sectors, which were separated into subendocardial, midmyocardial, and subepicardial samples. In total, 204 LV tissue samples were obtained from each animal, with a mean weight of 331 ± 280 mg. Tissue and reference sample radioactivity were counted for 180 s with a calibrated 1,024-channel gamma counter (model 5650; Packard Instruments, Downers Grove, IL) with a 3-in. NaI (Tl) crystal detector connected to a multichannel analyzer (series 35plus MCA, Canberrra Industries, Meriden, CT). Gamma spectra were separated, background activity was subtracted, and spillover and nuclide decay were corrected with a computer program (MIC-III) (11). RMP was calculated as

(2)

where _M is regional myocardial blood flow (ml/min), I_M is the intensity of emitted radioactivity from the myocardial tissue sample, _a is the reference-sample flow (3.24 ml/min), and I_a is the intensity of emitted radioactivity from the reference sample.

Experimental protocol. Baseline values were recorded and microspheres were injected after mean arterial pressure (MAP), heart rate, and cardiac output had been stable for 15 min. Results from this measurement are labeled as "baseline" in this report.

1

1

Data analysis. D was calculated by assessing the heterogeneity of regional perfusion (RD; i.e., SD/mean) repeatedly while reducing the resolution of measurement by averaging adjacent tissue samples. Resolution describes the minimal size of the object that can be observed with a certain method. When RMP is measured with radioactive microspheres, resolution can be changed at will by merely decreasing the size of myocardial samples during dissection of ventricles. Within limits, this is only limited by the number of microspheres within one sample. If microsphere counts are too low, increased methodological error results (2, 35). With the physical act of dissection, the resolution of RMP measurement is set; the minimal size of myocardial regions cannot be changed.

1

1

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Probability density functions (PDF) of regional myocardial perfusion (RMP) in diaspirin-cross-linked hemoglobin (DCLHb; A) and human serum albumin (HSA; B) groups. Dotted lines, subendocardial myocardium; solid lines, subepicardial myocardium; P, relative probability of samples to fall into RMP classes of 0.2 ml · min1 · g1 width.

Statistics. Statistical analyses were performed with the R statistical software (version 0.63.1). Normal distribution of data was tested with the Shapiro-Wilks test. Myocardial blood flow data were not normally distributed. All other parameters, including heterogeneity of RMP, displayed a Gaussian distribution. Analysis of variance on data or on ranked data (not normally distributed parameters) was used to test the presence of global differences. If this indicated a significant effect, paired t-tests or Wilcoxon signed-rank tests, as appropriate, were used to compare data from one measurement with data from the subsequent protocol time point. Comparisons between groups and between myocardial regions were made with the unpaired t-test or with the Wilcoxon rank-sum test, as appropriate. Normally distributed data are presented as means ± SD; otherwise, the median (25th/75th percentile) is used. Increased alpha error from repeated testing was corrected for by adjusting the alpha error threshold from P < 0.05 to P < 0.05/n, where n is the number of sequential tests performed. In the manuscript P < 0.05 is used for denoting significant differences, because corrected alpha thresholds serve to ensure that P values are actually <0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

For induction of hemorrhagic shock, 675 ± 97 ml (274 ml/kg BW; DCLHb group) or 811 ± 118 ml (303 ml/kg BW; HSA group) blood were withdrawn within 15-20 min. This led to arterial hypotension according to the protocol (MAP: DCLHb group, 45 ± 2 mmHg; HSA group, 46 ± 4 mmHg). Within 15-20 min after completion of volume therapy and before the final measurement in the protocol, 5 of 10 HSA-treated animals died from ventricular fibrillation. This was obviously due to arterial hypotension (MAP: HSA group, 69 ± 16 mmHg; DCLHb group, 115 ± 30 mmHg) and LV failure. High LV end-diastolic pressures (LVEDP) indicated reduced LV compliance (LVEDP: 18 ± 9 mmHg in the HSA group vs. 9 ± 4 mmHg in the DCLHb group). At baseline DCLHb- and HSA-treated groups had similar cardiac indexes (DCLHb 4.0 ± 0.8 l · min¹ · m²; HSA 4.6 ± 0.5 l · min¹ · m²; not significant). Induction of LAD stenosis did not change these values (DCLHb 4.0 ± 0.8 l · min¹ · m²; HSA 4.5 ± 0.4 l · min¹ · m²; not significant). After 60 min of hemorrhagic shock, there was a severe reduction of CI by 68% (1.3 ± 0.4 l · min¹ · m²) in the DCLHb group and 67% in the HSA group (1.5 ± 0.4 l · min¹ · m²). After infusion of DCLHb the CI was increased 2.5-fold to 3.2 ± 0.7 l · min¹ · m², whereas HSA treatment resulted in a rise of 3.7-fold to 5.6 ± 0.7 l · min¹ · m².

At baseline, endocardial-to-epicardial blood flow ratio (EER) was >1, indicating absence of subendocardial ischemia (Table 1). Induction of critical coronary stenosis did not induce any change of EER (Table 1). After hemorrhagic shock for 60 min, EER was significantly reduced to <1 in both groups. In the surviving five animals, after the infusion of HSA there was a nonsignificant reduction of EER. In contrast, application of DCLHb significantly elevated and normalized EER in all animals.

As predicted by the definition of critical stenosis, induction of LAD stenosis did not result in reduction of regional perfusion in the LV subendocardial or subepicardial myocardium (Table 1). During shock, RMP was well preserved in the subepicardial myocardium and only insignificantly reduced in the subendocardial myocardium. There was a tendency toward enhanced blood flow secondary to primary resuscitation, but this varied widely within both groups and was insignificant.

In the DCLHb-treated group, there was a significant increase in blood flow in the normally supplied LV wall and a tendency toward enhanced blood flow in the poststenotic area. Irrespective of transmural position (subendocardial, subepicardial), blood flow to the normally perfused myocardium was completely preserved during shock in both groups. In the poststenotic area, there was no significant reduction compared with baseline; however, there was a significantly lower perfusion than in the prestenotic areas in both groups.

Coronary perfusion pressure (CPP) was calculated as diastolic aortic pressure minus LVEDP. There were no differences between groups nor significant alterations of CPP in response to induction of critical coronary stenosis. During shock, CPP was uniformly reduced to approximately one-third of the baseline value. Although HSA infusion (in 5 surviving animals) did not consistently increase CPP, there was almost complete normalization of this parameter on infusion of DCLHb.

PDFs of subendocardial and subepicardial RMP are shown in Fig. 1. Flows >10 ml · min¹ · g¹ were not plotted to improve graphical appearance of the majority of RMP values. The slight rightward shift of subendocardial PDFs at baseline and after induction of coronary stenosis is caused by the physiological preferential perfusion of the subendocardial myocardium. During shock, this pattern was reversed; the subendocardial PDF was found slightly left of the subendocardial PDF. In both myocardial layers, a broadening of the distribution occurred; more low-flow and more high-flow regions were present. This broadening was further enhanced after fluid resuscitation. Only in the DCLHb group, the subendocardial PDF was right shifted again in comparison to the subepicardial PDF and resulted in a normalized EER (see Table 1).

RD at the smallest resolution of subendocardial as well as subepicardial perfusion is shown in Fig. 2. Average figures are given in Table 2. There were only changes of the RD in the LV subendocardial myocardium. Induction of coronary stenosis produced a reduction of subendocardial RD in the HSA group. The numerical decrease was small (28% of the increase during shock) but consistent and statistically significant. Shock significantly increased subendocardial RD in both groups. Infusion of HSA further increased subendocardial RD in four of five animals; however, this was not statistically tested because of the small number of observations. Administration of DCLHb led to a decrease in RD in 8 of 10 animals. There was a statistical error of P = 0.036 to reject the hypothesis "no difference in RD"; however, this was above the corrected threshold of P = 0.025. In the subepicardial myocardium of animal da26, very high RD of RMP was calculated. Values for subendocardial myocardium were not different from the other animals. No methodological error in RMP measurement was detected. Subepicardial PDF curves of animal da26 did not display abnormalities except a rightward shift and broadening that were absent in PDFs of subendocardial myocardium.

View larger version (35K): [in this window] [in a new window]   Fig. 2.   Relative dispersion (RD; time points in protocol) of RMP at smallest resolution in DCLHb (A) and HSA (B) groups. B, bottom: in the heart of animal da26, highly heterogeneous RMP was measured. Ordinates were scaled to yield optimal results for majority of animals. Values of animal da26 were beyond this scale. For plotting, 50% was subtracted from each value. Subepicardial fractal dimension (D) values were also highest in this animal (see Fig. 3). * Significant difference compared with previous time point.  Significant difference between HSA- and DCLHb-treated groups.

Subendocardial D as well as subepicardial perfusion are depicted in Fig. 3. Average values of D (together with corresponding error bars) of the complete LV, and subendocardial and subepicardial D are shown in Table 2. D and error bars of D of individual animals are presented in Table 3. At baseline, subendocardial D was significantly higher than subepicardial D in both groups. Induction of critical stenosis increased subendocardial D in the DCLHb group. Although hemorrhagic shock did not affect subepicardial D, there was a significant reduction of subendocardial D in the DCLHb group. The decrease during shock was much greater than the slight but consistent decrease after induction of coronary stenosis in the DCLHb animals.

View larger version (36K): [in this window] [in a new window]   Fig. 3.   D of RMP measured against time points in protocol (x-axis) for DCLHb (A) and HSA (B) groups. * Significant difference compared with previous time point of protocol, P < 0.05.  Significant difference between HSA- and DCLHb-treated groups, P < 0.05. § Significant difference between subepicardial and subendocardial D of RMP, P < 0.05.

In the HSA group, there were insignificant reductions of D in 7 of 10 animals in response to critical coronary stenosis (P = 0.12) and in 6 of 10 animals during shock (P = 0.11). These changes were smaller and less consistent than in the DCLHb animals.

In the DCLHb group, there appeared a tendency toward normalized subendocardial D, but this was not significant. Albumin infusion did not result in large changes in the surviving animals.

As stated in METHODS, the assumption underlying fractal analysis is that the relationship between RD of RMP and resolution of measurement (mass of aggregated samples) is governed by an inverse power law. After the natural logarithms of both parameters are taken, this translates into an inverse linear relationship. Individual r values of all linear fits of ln(RD) and ln(mass) are shown in Fig. 4. Average values are given in Table 2.

View larger version (35K): [in this window] [in a new window]   Fig. 4.   Coefficients of determination (r) of linear fits of natural logarithms of RD and mass of aggregated myocardial samples for DCLHb (A) and HSA (B) groups. x-Axis, time points in protocol. * Significant difference compared with previous time point of protocol, P < 0.05.

There was a small increase in r after induction of coronary stenosis and no significant change thereafter in the DCLHb group. In contrast, in the HSA group during shock, r decreased in seven animals and remained stable in three.

The error bar of the slope of the regression line of ln(RD) and ln(mass) was used to assess the uncertainty of this linear fit. This error was small and did not change in the course of the experiment in any group (see Tables 2 and 3).

Calculation of any heterogeneity parameters implies generalization; no conclusion can be drawn from these parameters as to which changes in perfusion were induced by any intervention in one individual myocardial region. This latter information is presented in Figs. 5 and 6. In Fig. 5, RMP measured at each time point in the protocol is plotted against RMP of the identical myocardial region calculated at the following measurement. Thus if no changes occurred, all data pairs were on a line of identity. In Fig. 6, the distributions of differences of RMP between two measurements are depicted as PDF curves.

View larger version (35K): [in this window] [in a new window]   Fig. 5.   Changes in perfusion induced by experimental interventions in individual myocardial pieces from DCLHb (A) and HSA (B) groups. Left and right halves of each panel denote subendocardial and subepicardial data, respectively; x-axis, RMP measured during 1 time point in protocol; y-axis, RMP measured during following time point; solid lines, lines of identity. Lack of any change between measurements would result in all data pairs lying on these lines.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   PDFs of changes in perfusion induced by experimental interventions in individual myocardial pieces from DCLHb (A) and HSA (B) groups. Dotted lines, PDFs of subendocardial flow differences; solid lines, PDFs of subepicardial flow differences; solid vertical lines, lines through 0 to ease evaluation of PDFs; P, relative probability of differences of perfusion during 2 measurements to fall into RMP classes of width 0.2 ml · min1 · g1; x-axis, difference in regional myocardial perfusion.

Induction of critical coronary stenosis led to a marginal increase of subendocardial as well as subepicardial RMP in both groups. This is represented by right-shifted PDFs (Fig. 6) and a clustering of data points above the line of identity (Fig. 5). Generally, all high-flow regions remained such during coronary stenosis and vice versa.

The differences in blood flow during shock and blood flow during stenosis alone were predominantly negative in subendocardial myocardium in both groups. However, there were many large increases in flow to individual pieces and more, albeit less extreme, decreases in individual perfusion values.

Resuscitation with DCLHb increased subendocardial perfusion more than subepicardial perfusion. There were many large decreases as well as increases in RMP; however, a large proportion of regions with small RMP was converted to high-flow regions by DCLHb.

In contrast to DCLHb, resuscitation with HSA increased subepicardial perfusion more than subendocardial perfusion. There were less extreme changes than in the DCLHb group. Many low-to-moderate flow regions were transformed to extremely low flow regions in subendocardial myocardium. This latter phenomenon was not present in subepicardial myocardium, where flow almost exclusively increased.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The main result of this study was that the fractal model fitted the empirical distributions of RMP during stenosis, shock, and resuscitation. There were, however, decreases in the goodness of fit for the subendocardial myocardium during shock in the HSA group. Hemorrhagic shock induced increases in RD of subendocardial perfusion, but the relationship of resolution of measurement and RD became less steep, which translates into a reduced D. In subepicardial myocardium, shock did not elicit this effect. Resuscitation with DCLHb induced tendencies toward normalized subendocardial RD and D but did not systematically alter subepicardial parameters. Only five animals survived until measurement 60' R . Therefore, the effects of HSA on RD and D could not reliably be evaluated. DCLHb infusion led to a normalized EER in all animals, whereas the surviving HSA-treated animals suffered from a persistently reversed EER.

We have previously shown that hemorrhagic shock reduces D of RMP (22). However, in the cited study resuscitation with normal or hypertonic saline, respectively, did not induce normalization of D. We performed this study to verify or falsify the hypothesis that more effective fluid resuscitation from hemorrhagic shock would reverse shock-induced reduction of D of regional myocardial blood flow.

In our previous study we used either 80% of shed blood volume (SBV) of isotonic saline or one-eighth of this amount of a solution of hypertonic saline and hyperoncotic dextran (small-volume resuscitation) (22). In the present study we attempted to improve resuscitation by using a stable solution of DCLHb. This solution is a hyperoncotic (oncotic pressure 42 mmHg) oxygen carrier and acts as a vasoconstrictor because of nitric oxide scavenging, increased sensitivity of adrenergic receptors, and upregulation of the endothelin system (8, 12). Nitric oxide scavenging has recently been challenged as the sole mechanism of hemoglobin-induced vasoconstriction (29).

DCLHb is a favorable solution for primary resuscitation of hemorrhagic shock, as previous studies in animals have shown. Mortality from hemorrhagic shock in rats can be reduced by DCLHb to the same degree as is achieved by autologous whole blood (4). Gulati and Sen (12) reported that resuscitation from shock with DCLHb produced preferential redistribution of blood flow to the heart and the brain. In the report from these authors, it was also shown that a dose of 50% of SBV resulted in satisfactory resuscitation, which could not be improved by increasing the dose to 100% of SBV. The latter result was the rationale for determining the dose of DCLHb in the present study, namely, the volume of plasma removed with the SBV. Hematocrit is always below 40% in juvenile pigs; this regime led to a DCLHb dose of ~70% of SBV.

Normalization of blood volume after hypovolemic shock dilutes red blood cells and thus reduces arterial oxygen content. Because hemoglobin in a cellular solution can transport oxygen (14, 16, 30), the dilutional effect of fluid therapy of shock can be offset by DCLHb. As a control solution, 8.0% HSA was used. This solution has the same oncotic pressure as DCLHb and, as a protein, has similar pharmacokinetic properties, at least over the limited posttreatment observation period in this study.

At baseline, subepicardial myocardium D values were in the range of what has been reported previously for baboons, sheep, and rabbits (1). Subendocardial D, however, was higher than reported before for whole ventricles and significantly above subepicardial values. We also calculated D for complete ventricles and found values in the range that was previously reported. Differences in D from subendocardial myocardium and subepicardial myocardium between our results and those from previous studies were absent after induction of hemorrhagic shock.

D values of the whole ventricle were lower in this study than in our previous paper (22). In contrast to the latter, in the present article subepicardial D values were in the range of LV D values, and subendocardial D values were higher than both LV and subepicardial D. Despite these, in part extreme, differences, the effects of hemorrhagic shock were similar in both studies. Hemorrhagic shock reduced subendocardial D more than subepicardial D. However, in contrast to our previous study LV D values were not altered by shock. The most prominent difference between both studies is that hypertonic and normal saline further changed LV D in the same direction as did shock, whereas DCLHb but not HSA evoked a tendency toward an increase in subendocardial D.

The differences in D cannot be explained by differences in calculations. We used identical numbers of RD-mass data pairs for linear regression, and similar algorithms (implemented in different programming languages) were applied. All radioactive microsphere injections and measurements were performed by the same person in both studies. Despite major differences in different parameters, the main result of fractal analysis in both studies, reduction of subendocardial D during shock, was identical in both reports.

There was no significant difference of RD values at maximum resolution between left ventricular subepicardial and subendocardial myocardium. However, before shock, average epicardial RD values were higher than subendocardial values, but this was not statistically significant. In our previous study, RD of RMP was lower than the values in this paper. This reflects the higher resolution of RMP measurement in the present study (10% lower sample weight). Subendocardial RD was lower than subepicardial RD in both studies. Whereas in this paper subepicardial RD was not altered during shock, in the saline study there was an increase in subepicardial RD. This increase was much smaller than subendocardial changes, paralleling the tendency observed in this study.

It is no contradiction that higher RD at maximum resolution and significantly lower subendocardial myocardium D occurred. RD at one resolution represents only one point on the linear log-log plot of heterogeneity and resolution, whereas D represents a measure of the slope of this line (see Fig. 7). The line may rotate around one point, leaving RD values of one resolution unchanged, whereas others are increased or decreased. Because of different techniques of perfusion measurement and different resolution, comparison of RD from the present study with data from previous reports of other research groups is difficult. However, taking into account the dependence of RD on sample mass, there appears to be no difference between baseline RD values in the present study compared with previous reports (1, 18, 25).

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Typical example of relationship of RD of RMP and aggregated mass of myocardial samples. Data are from animal da22 at baseline. Left y-axis, natural logarithm of RD of RMP [ln(RD); %]; right y-axis, RD of RMP (RD; %); top x-axis, aggregated mass of myocardial samples (mass; mg); bottom x-axis, natural logarithm of aggregated mass of myocardial samples [ln(mass); mg]. Data points having equal or approximately equal mass are derived from different possibilities to recombine larger samples (for further explanation, see Data analysis).

It is striking that, in either experimental group, D values derived from subepicardial myocardium did not change in response to shock, whereas subendocardial D values were significantly or tendentially reduced after induction of hemorrhagic shock. RD of the RMP of subendocardial and subendocardial myocardium, respectively, confirmed these findings. There was no alteration of subepicardial RD in shock, but subendocardial RD was significantly increased in subendocardial myocardium in both groups.

These results confirm our previous findings (22) that shock in the presence of critical coronary stenosis most strongly affects subendocardial perfusion heterogeneity. Subendocardial edema and consecutive compression of intramyocardial vessels (3), as well as preferential swelling of myocytes in the subendocardial myocardium (17), might be mechanisms causing this difference.

In conclusion, we have confirmed that hemorrhagic shock in the presence of coronary stenosis reduces subendocardial RMP D. Blood flow to subendocardial myocardium is more susceptible to changes in flow distribution, whereas subepicardial perfusion is relatively unaffected.

Despite the insignificant change in D values of subendocardial RMP (P = 0.066) and RD (P = 0.036), it cannot be ignored that there was a strong tendency toward normalization after infusion of DCLHb.

We therefore conclude that our hypothesis can cautiously be considered verified: rapid and effective fluid therapy with DCLHb did induce a tendency toward restoration of normal distribution of regional myocardial blood flow. In previous experiments (22), we have investigated the effects of resuscitation from shock with isotonic and hypertonic saline solutions on RMP distribution. None of these fluids has proven efficacious in this respect. It remains to be investigated whether resuscitation with autologous or homologous blood is able to restore heterogeneity of RMP after hemorrhagic shock.