In the late 1960s, the radioactive microsphere (RM) technique was introduced for the determination of regional blood flows (1). Experimentally validated for measuring these flows (2), as well as cardiac index (3), the RM technique is still regarded as the reference method for estimating blood flows in cardiovascular pharmacology. However, because the use of radioactivity is becoming increasingly problematic for security, environmental, and regulatory reasons, a number of new replacement techniques have recently been developed including colored (4,5) or fluorescent microspheres (6,7).
All these techniques are based on the same principle as the radioactive microspheres. The spheres are trapped during their first passage in each organ, and the flow of each organ is directly proportional to the number of spheres recovered within that organ. However, whereas radiolabeled microspheres are directly quantified in a gamma counter, colored or fluorescent microspheres need first to be separated from tissues before dye or fluorescence assessment.
The fluorescent microsphere (FM) technique has been demonstrated to be a reliable alternative to radioactive microspheres for measuring organ perfusion in pigs and dogs (6,7) but has not yet been validated in smaller species. The goal of this study was thus to apply and validate this technique for the measurement of cardiac index and regional blood flows in rats.
We therefore investigated successively (a) the agreement between the FM technique and the reference RM technique by comparing cardiac index and regional blood flow values obtained simultaneously in the same animal by the two approaches, (b) the repeatability of the measurements performed with the FM technique, and (c) the ability of the FM technique to detect and quantify drug-induced changes in systemic and regional blood flows by using a selective coronary vasodilating drug, dipyridamole. This is the first validation study of this new technique in rats, a widely used species in experimental cardiovascular pharmacology and physiology investigations.
Preparation of animals. Male Wistar rats (body weight, 300-340 g; n = 20; Iffa Credo, L'Arbresle, France) were anesthetized with sodium pentobarbitone (50 mg/kg, i.p.). A catheter was placed in the left femoral artery. A second catheter was inserted into the left ventricle via the right carotid artery, its proper position being verified by observing the characteristic left ventricular pressure waveform on a Video Screen (Statham P10EZ transducer; Gould amplifier 13-4615-10 model, ES 2000 V12; Gould Instruments, Cleveland, OH, U.S.A.). Arterial blood pressure [systolic (SAP) and diastolic (DAP)] and heart rate (Biotach 13-4615-66 model; Gould Instruments) were continuously recorded on a polygraph (ES 2000 EW; Gould Instruments).
Animal instrumentation and subsequent experiments were performed in accordance with the regulations of the Ministère Français de l'Agriculture.
Microsphere injection and quantification. A mixture containing ∼80,000 radioactive microspheres (103Ru, 15 ± 0.1 m; specific activity, 5.17 mCi/g; NEN Company, Boston, MA, U.S.A.) and 200,000 fluorescent microspheres (blue or yellow-green fluorescent label, 15 ± 0.1 m; Triton, San Diego, CA, U.S.A.) was sonicated, vortexed for 10 min, and injected into the left ventricle of the anesthetized animals with a Hamilton injection syringe (model 1750 LT) over 10 s followed by 0.2 ml saline. The number of RM classically reported in the literature is 80,000, and 200,000 is the minimal number of FM yielding an accurately measurable fluorescence signal in all organs, especially in that with the lowest flow (i.e., skeletal muscle). Preliminary experiments showed that the simultaneous injection of 80,000 RM and 200,000 FM did not alter systemic hemodynamics in the rat.
The radioactivity of the injection syringe was counted before and after injection (Compugamma 1280; LKB, Turku, Finland) to determine by difference the total injected radioactivity.
The total injected fluorescence was estimated each week as the mean of the fluorescence measured in four FM samples randomly collected with the same Hamilton syringe as that used for the in vivo injection, and directly processed. Sixteen samples were thus collected per FM flask, and the calculated coefficient of variation was found to be 4.5% and 4.8% for blue or yellow-green FM, respectively.
A reference blood sample was withdrawn from the femoral artery at a rate of 0.85 ml/min into a preweighed heparinated syringe (Harvard 907A; Harvard, South Natick, MA, U.S.A.) starting 10 s before microsphere injection and lasting for 90 s. The syringe containing the reference blood sample was weighed and assayed for radioactivity in the gamma counter. The animals were then killed, and hearts, brains, kidneys, and samples of skeletal muscle were removed, blotted, weighed, and assayed for radioactivity.
All tissue samples (range, 1-2 g) and reference blood samples (1.2 ml) were then processed for fluorescence quantification. Tissue samples were individually digested in 8 ml of 4N KOH with 2% Tween 80 for 24 h at room temperature, whereas reference blood samples were digested in 1.2 ml of 16N KOH. The digested samples were then individually filtered through 10-μm pore polycarbonate filters (Triton) to recover the FM, and the vials and filtering burette were rinsed twice with 10 ml of phosphate buffer, pH 7, and once with 70% ethanol. The filters were carefully removed and placed in individual polypropylene test tubes, and 1 ml of cellosolve acetate was added to each tube to extract the fluorescent dye. After 1 h, a dye/solvent sample from each organ piece was pipetted into individual wells of 96-well microplates.
Fluorescence was determined with a Perkin Elmer LS 50B luminescence spectrophotometer (Perkin-Elmer, Beaconsfield, U.K.) equipped with a 96-well microplate reader. The fluorescence intensity of each dye/solvent sample was measured at the optimal excitation/emission wavelength pair of each dye (blue, 360 and 415 nm; and yellow-green, 490 and 510 nm) with excitation and emission slit widths of 2.5 nm and 5 nm, respectively.
Microsphere recovery was tested according to Van Oosterhout et al. (7): 200 μl of a suspension containing ∼2,000 blue-green microspheres were either directly extracted by the solvent (n = 58) or added as an internal reference to vials (n = 67) containing blood or tissue samples, the latter being processed as previously described. The blue-green fluorescence signals (optimal excitation/emission wavelength pair, 435 and 470 nm) obtained (a) from the microspheres directly extracted, and (b) from the microspheres added to tissue or blood samples were compared and the latter (mean fluorescences of the kidney, 394.2 ± 2.7; of the heart, 401.1 ± 5.5; of the muscle, 389.3 ± 5.5; of the brain, 381.0 ± 3.8; and of the blood, 380.3 ± 2.8) were found to be ∼8.7-13.4% smaller than the former (mean fluorescence, 439.3 ± 3.1).
Preparation of animals. The animals (body weight, 320-400 g; n = 14) were anesthetized intraperitoneally with a mixture of ketamine (70 mg/kg) and xylazine (10 mg/kg) and instrumented as previously described. All the catheters were filled with 5% polyvinyl pyrrolidone containing heparin (50 IU/ml), tunnelized, and exteriorized at the back of the neck of the animals. The animals were then allowed to recover for ≥4 h after completion of surgery.
Protocol. Cardiac output and renal and myocardial blood flows were then estimated twice successively, at a 10-min interval in the conscious animals, by using two different fluorescent dyes (blue, 3 × 105; and yellow-green, 2 × 105 per injection).
Tissue and reference blood sample processing and fluorescence assessment were performed as previously described. Mean arterial pressure (MAP) and heart rate were continuously recorded from the femoral artery before and after each microsphere injection.
Hemodynamic changes induced by dipyridamole
Instrumentation of the animals (body weight, 310-400 g; n = 45) was performed as previously described for the repeatability study, a third catheter being additionally inserted into the jugular vein.
After recovery from anesthesia, fluorescent microspheres (blue, 3 × 105; and yellow-green, 2 × 105 per injection) were injected in the conscious animals before and after a 10-min dipyridamole infusion via the jugular vein at four different doses (2, 4, 6, and 8 mg/kg·min; infusion rate, 0.4 ml/kg·min; n = 9-13 for each dose of dipyridamole; Harvard 33 double syringe pump). Cardiac output and renal and myocardial blood flows were determined twice, before and after drug infusion.
MAP and heart rate were continuously recorded from the femoral artery, except during the microsphere injections.
In an additional group of five anesthetized rats, the potential interference of dipyridamole fluorescence with the different dyes was checked. The animals were instrumented as previously described and received a 10-min infusion of dipyridamole at the highest dose used (8 mg/kg·min) without any microsphere injection. Fluorescence of blood and of tissue (heart and kidneys) samples was measured after processing at the optimal excitation/emission wavelength of dipyridamole (435 and 470 nm, respectively). Fluorescence signal obtained from all samples was similar to the background solvent fluorescence signal, indicating that no dipyridamole trace could be detected on the filters after sample processing.
MAP was calculated as (SAP + 2 DAP)/3.
Cardiac output (CO) was calculated according to the formula: Equation (1) where T is the total injected fluorescence or radioactivity, Rf is the fluorescence or radioactivity of the reference blood sample, and Qf is the reference flow calculated as: Equation (2)
Cardiac index (CI, ml/min·kg) was calculated by dividing CO by the animal body weight (kg).
Regional blood flows (Qo, ml/min·g of organ) were calculated for each organ or part of organ according to the formula: Equation (3)
Total peripheral resistance (TPR, mm Hg·min·kg/ml) was calculated by dividing MAP recorded just before each microsphere injection by CI.
Dipyridamole was obtained from Sigma (Saint-Quentin-Fallavier, France), and ketamine (Imalgene 1000) and xylazine (Rompun 2%) from Centravet (Plancoët, France).
Values of all hemodynamic parameters are given as mean ± SEM.
In the agreement study, RM and FM techniques measuring the same flows were compared according to the recommendations of Bland and Altman (8). The correlation between each set of paired values obtained by both techniques was investigated (equation of the linear relation, correlation coefficient, r and p value). This first step allowed the detection of possible bias (proportional or additional under- or overestimations), followed if necessary by a model allowing expression of the compared flows in comparable units. When the FM and RM values were expressed in the comparable units, differences within pairs (Di) were tested, and error probability distribution was estimated by calculating a mean intermethod error ((Equation 4)) and a 95% confidence interval, ±1.96 √Σ(Di)2/n.
When error depended on measured flow value (proportional error), a log transformation of the data was performed and provided an estimation of agreement (95% confidence interval), which varied with the considered flow value (8).
In the repeatability study (measurement of the same flow by two successive FM injections), we followed the same procedure to calculate mean intramethods error and 95% confidence intervals, but no prior modeling was needed, as the two measurements were performed in the same scale. In addition, coefficients of variation between the two flow values obtained for each organ in each rat (calculated as the ratio of Di/mean of the pair) were plotted against the mean number of fluospheres trapped in each sample to test the relation between these two parameters.
In the dipyridamole-induced vasodilation study, systemic and regional effects of dipyridamole in the four groups of animals were compared by using an analysis of variance for repeated measurements followed by intergroup comparisons by using a two-sided t test with Bonferroni's correction. A p value <0.05 was considered significant.
Mean basal values of MAP and heart rate calculated before microsphere injection in 20 anesthetized rats were 131 ± 3 mm Hg and 411 ± 7 beats/min, respectively.
Mean values of CI and of renal, myocardial, muscular, and cerebral blood flows obtained by the RM and FM methods are listed in Table 1.
Cardiac index. Correlation study (Fig. 1a) establishes a linear relation between the two series of measurements (r = 0.82; p < 0.001), the corresponding equation (slope, 1.04 ± 0.17; intercept, −52.5 ± 61.2) being not statistically different from the identity line. The difference between the two techniques (Di) does not appear to depend on the considered CI value (Fig. 1b), and the mean error ((Equation 4), 38.1 ± 52.7) does not significantly differ from zero.
Thus limits of agreement can be expressed as a 95% confidence interval: difference between two determinations of the same cardiac index (one pair of measurements) by FM and RM is expected to be <±125 ml/min·kg [±1.96 √Σ(Di2)/n], with a probability of 95%.
Regional blood flows. When all organ blood flow data are pooled, the correlation study (Fig. 2a) shows that the linear relation between regional blood flow values obtained by FM and RM differs from the identity line (r = 0.99; p < 0.001; slope, 0.83 ± 0.01; intercept, 0.06 ± 0.05), FM regional blood flow values being somewhat globally smaller than those obtained with the RM method. Therefore, before studying intermethods error, flows must be expressed in the same scale. Thus FM values were translated into RM comparable units (FM′)... by using the formula derived from the correlation equation: FM′ = (FM − 0.06)/0.83.
Such a transformation allows the study of intermethods errors (Di), which establishes (Fig. 2b) that Di increases with the measured blood flow value. Log transformation provides an estimation of a proportional error that varies with the considered regional blood flow value. In Table 2 are examples of regional blood flow values with their corresponding 95% confidence intervals.
The microsphere injection and the reference sample withdrawal did not per se significantly affect MAP (131 ± 3 mm Hg after vs. 133 ± 3 mm Hg before) and heart rate (372 ± 12 beats/min after vs. 362 ± 6 beats/min before). Table 3 shows mean CI and regional blood flow values measured in the same conscious animals twice successively with two different FM sets as well as their corresponding 95% confidence intervals for one pair of measurements.
Figure 3 plots the individual coefficients of variation between the two flow values obtained for each organ in each rat against the number of fluospheres trapped in each sample. This figure shows that there is no relation between these two parameters (r = 0.126; p = 0.36). Mean coefficients of variation for each organ are shown in Table 3.
Changes induced by dipyridamole in cardiac output and regional blood flows
Mean basal values of the different investigated parameters before dipyridamole infusion were not significantly different among the four groups of animals (Table 4).
Changes induced by dipyridamole at the different doses used are illustrated in Figs. 4 and 5.
Dipyridamole dose-dependently decreased MAP and TPR and increased heart rate and CI. Myocardial blood flow, up to 6 mg/kg·min, was dose-dependently increased in the left as well as in the right ventricle. In contrast, whatever the dose used, dipyridamole did not affect renal blood flow. It also appears from Figs. 4 and 5 that whereas the highest tested dose (8 mg/kg·min) of dipyridamole was still more effective at reducing MAP and TPR than the other doses, the maximal increases in myocardial blood flow were already achieved with the 6 mg/kg·min dose.
The feasibility of the fluorescence technology for measuring regional blood flows was initially evaluated in vitro by Glenny et al. (6). FM present numerous advantages over RM. The first is that spectrofluorometry provides sensitive measurements with good spectral separation, which allows the use of several different dyes in the same experiment without requiring corrections for signal spillover (6,7), but the major advantage of this technique is clearly the lack of radioactive exposure.
This study is the first attempt to validate fully the FM technique, a nonradioactive approach to measure regional blood flows in the rat (i.e., a species that is widely used in cardiovascular research). Furthermore, it is also the first study dealing with the validation of CI measurements. We therefore investigated the agreement between the FM technique and the reference RM technique, as well as the ability of the FM technique to perform repeatable measurements in the same animal. Finally, we applied this method to a pharmacologic test of selective regional vasodilation (i.e., the dipyridamole test).
Concerning CI measurements, agreement between the two methods allows us to conclude that FM can be used instead of RM. Indeed, there is only a 5% probability that the absolute difference within one pair of CI values obtained by the two techniques will be >125 ml/min·kg. When applied to pragmatic experimental designs in which means are calculated within samples, dispersion of the error is then inversely proportional to the square root of the sample size. For example, intermethods difference of a mean CI value within a sample of 25 animals is unlikely (<5%) to be >±25 ml/min·kg (confidence interval/√25. In such a sample, a mean CI of 300 ml/min·kg obtained by the FM technique (Table 1) may correspond to a mean CI value within the 275-325 ml/min·kg range with the RM technique, which can be considered as an acceptable error.
Concerning pooled regional blood flows, agreement varies with the measured flow (the higher the flow, the larger the difference between the estimates provided by the two methods). This proportional error between the two methods has been described but not discussed in previous studies (6,7). Under these conditions, agreement must be expressed in reference to the measured value. For instance, intermethods error in the determination of one myocardial blood flow of, for example, 5 ml/min·g, is likely to be less than (6.48 − 3.86)/2 = 1.31 ml/min·g (Table 2) and for a mean myocardial blood flow within a sample of 25 animals <1.31/√25 = 0.26 ml/min·g, which is also an acceptable error. In addition, by pooling the flow data obtained from four particular organs, this study shows that the FM technique is in agreement with the RM reference technique for measuring any regional blood flow, the value of which is comprised in the range studied (i.e., 0.1-7 ml/min·g. Furthermore, it allows the prediction of the corresponding error 95% confidence interval (Table 2).
Regarding cardiac index, the linear relation between the RM and FM evaluations is not statistically different from the identity line. However, in regional blood flow measurements, our results demonstrate that the FM technique globally provides somewhat lower estimates of regional blood flows than the RM technique. This underestimation was not previously found by Glenny et al. (6) in pigs and by Van Oosterhout et al. (7) in dogs, but it must be stressed that besides the species difference, the method used for measuring fluorescence in our study (microplate readings) was different from that used by the other mentioned authors (direct readings in cuvettes). Another explanation for the lower estimates obtained in our study by the FM technique could be a loss of microspheres during processing. Our in vitro experiments show that there is a loss in the measured fluorescence (−8.7 to −13.4%, depending on the organ) when a similar amount of FM (a) is added to blood or tissues subsequently processed (digestion, filtration, extraction) or (b) is directly assessed for fluorescence. However, as this loss is almost identical in blood and tissues, it cannot explain the underestimation of flows by FM versus RM. But it must be stressed that this is an in vitro experiment in which FM are added to previously excised tissues, whereas in vivo injected FM penetrate and are trapped within the tissues. And it may well be that the recovery of tissue-trapped FM is less (because of unadequate digestion, for instance) than in blood, which could explain the underestimation of organ flows by FM versus RM. Another important argument in favor of this hypothesis is that when measuring regional blood flows with the colored microsphere technique in which an extraction step is also involved, the blood flow values obtained in the rat are also smaller than those measured with the RM technique (5). In these conditions, the needlessness of tissue processing for cardiac index measurements with FM probably explains why we found a better adequation between the values obtained by the FM and RM techniques for this parameter.
The differences between the regional blood flow values obtained by both techniques remain small and of no importance when intramethod comparisons result from repeated measurements in the same animals.
Regarding the repeatability study and because of the small blood volume in rats, we limited our investigation to two successive measurements (i.e., to two reference blood samplings), a situation that allows assessing the effect of one pharmacologic intervention. As shown by our data, two successive FM injections did not induce any systemic hemodynamic change (MAP and heart rate), which is an essential requisite for accurately measuring regional blood flows. Table 3 indicates that in conscious animals, there is only a 5% probability that the absolute difference between two successive assessments (one pair) of left ventricular myocardial blood flow (normal value, 6 ml/min·g) will exceed 2.25 ml/min·g. This confidence interval is perfectly comparable to that calculated for RM from the data by Kobrin et al. (9), which has a value of ±2.41 ml/min·g (one pair) for a normal myocardial blood flow value of 5.43 ml/min·g. Furthermore, if we consider that, in pharmacologic studies involving regional flow measurements, animal group size is ≥12, it can be calculated from our data that, for a normal left ventricular myocardial blood flow value of 6 ml/min·g (Table 3), the probability that two successive measurements will differ by >0.65 ml/min·g (i.e., 10.8%) is only 5%. This is fully satisfactory, as pathology-induced or drug-induced flow variations largely exceed 10.8%. In this study, for instance, the lowest dose of dipyridamole used increased left ventricular myocardial blood flow by 140% (Fig. 5). This probability approach also allows us to calculate the size of experimental groups to minimize method-induced variations under the level of the expected effects of a pharmacologic intervention. Finally, our data show that, within the range of 760-20,000 fluospheres trapped within a sample, repeatability of FM measurements does not vary (Fig. 3), a finding in agreement with that of Prinzen and Glenny (10) for values of FM >400 per sample.
Regarding the pharmacologic application, our study demonstrates that in the conscious rat, the FM method is able to show qualitative and to evaluate quantitative flow variations after the infusion of a selective coronary vasodilator, dipyridamole. And indeed, qualitatively, dipyridamole, whichever the dose used, significantly increased CI and left and right ventricular myocardial blood flows without simultaneously affecting renal blood flow. Quantitatively, our results demonstrate that dipyridamole, up to 6 mg/kg·min, induces a dose-dependent increase in both left and right myocardial blood flows. It should again be stressed here that all the variations in flow observed in this study (with the exception of those of renal blood flow) largely exceed the limits of their corresponding error 95% confidence interval as obtained from the repeatability study. These effects are thus induced by dipyridamole per se as they cannot reasonably be accounted for by method-induced artefacts. Incidentally, the demonstration that the 4 mg/kg·min dose of dipyridamole induces significantly greater increases in both left and right myocardial blood flows than the 2 mg/kg·min dose without simultaneously producing a greater decrease in MAP (i.e., in coronary perfusion pressure), questions the adequacy of the generally used 2 mg/kg·min dose of dipyridamole (11,12) for the assessment of coronary flow reserve in rats.
In conclusion, the FM technique (a) is a reliable method for measuring cardiac index and regional blood flows in rats, (b) displays a good agreement with the reference radioactive technique for which it can thus be advantageously substituted, (c) provides repeatable measurements allowing at least two successive flow determinations in the same conscious animal, and (d) permits the quantification of drugs' systemic and regional hemodynamic effects in the conscious rat.
Acknowledgment: This work was supported by a joint grant from INSERM, Paris, France (97 AN 49), and Merck, Sharp & Dohme, West Point, PA, U.S.A.
M.G. is a recipient of a Fellowship Grant from the Ministère Français de l'Enseignement Supérieur, de la Recherche et de la Technologie. We thank Ms. M-F. Dauby for her excellent editorial assistance.