Hemodynamic derangements are common in critically ill neonatal foals, and monitoring cardiovascular function in those animals is an important part of patient management.1–6 Many of those hemodynamic derangements adversely affect tissue perfusion and delivery of oxygen to tissues.6,7 Cardiac output is a major determinant of oxygen delivery to and perfusion of tissues.4,6,7 Although physical examination and monitoring of basic hemodynamic variables are useful for general assessment of cardiovascular function, direct evaluation of CO provides more accurate information and can improve patient management.6,8
In horses, direct monitoring of CO is limited because accurate and noninvasive methods for such monitoring are currently lacking.2,6,7,9 Methods to monitor CO in adult horses and foals include 2DE and the Fick, UDCO, and indicator (ie, pulmonary thermodilution and lithium dilution) methods.2,6,9 The Fick and indicator methods are not suitable for critically ill patients because they are invasive and require systemic administration of an indicator solution, which can have deleterious effects.2,6,10
The UDCO is a minimally invasive technique for hemodynamic monitoring that involves administration of a small bolus of a physiologic and noncumulative signal solution (saline [0.9% NaCl] solution).5,6,11–13 This technique provides information about cardiac function, including CO and blood volumes.6,11 Measurement of CO by use of the UDCO has been validated against techniques that are considered reference standards in humans and animals such as neonatal foals5 and juvenile horses.11 In human medicine, UDCO is used in intensive care units to measure the CO in pediatric8,14 and adult15 patients. Use of UDCO in critically ill foals is promising because it requires access to only peripheral veins and arteries, which is readily achieved in those patients.
Two-dimensional transthoracic echocardiography is well described in horses and other species and can provide information about both cardiovascular function and morphology.1–3,9 However, the accuracy of 2DE is limited by image orientation, geometric assumptions, and boundary tracing errors.16 Consequently, estimation of CO by 2DE is only moderately correlated with measurement of CO by reference standard techniques.6,16,17 Newer, more sensitive, and minimally invasive imaging modalities include 2-D TEE and gated cMRI.17–19 Use of TEE has been described in various species, including adult horses, for assessment of hemodynamic variables, cardiac morphology, CO, and volume status,17,18,20–24 but to our knowledge, its use has not been described in neonatal foals. With TEE, images can be acquired in unique planes without interference caused by ribs, lungs, and hair; therefore, TEE may be more accurate than standard transthoracic imaging.
Cardiac MRI uses nonionizing radiation, is considered the reference standard for evaluation of cardiac morphology and measurement of cardiac function,25–27 and is well described in humans, dogs, and cats.19,28–33 Compared with echocardiography, advantages of cMRI include the ability to characterize myocardial tissue and obtain cross-sectional images of the heart in any plane, thereby avoiding geometric assumptions and improving accuracy and reproducibility of measurements owing to unrestricted fields of view. It allows cardiologists to view the heart in all planes and can be used to create 3-D reconstructed images of the heart.25,34,35 Veterinary patients need to be anesthetized for cMRI, which limits its routine use. However, cMRI examination of healthy foals can provide valuable information that can be used as a reference against which similar data obtained by other modalities can be compared.
Neither TEE nor cMRI has been described in neonatal foals. Thus, measurements of CO derived from 2DE, TEE, cMRI, and UDCO have not been compared. The objectives of the study reported here were to describe TEE and cMRI in neonatal foals and to evaluate the accuracy of CO measured by 2DE, TEE, and UDCO, compared with that measured by cMRI. We hypothesized that the CO measured by TEE and UDCO would more closely mirror that measured by cMRI than the CO measured by 2DE.
Materials and Methods
Animals
All study procedures were reviewed and approved by the University of Illinois Urbana-Champaign Animal Care and Use Committee (protocol No. 16135). Six university-owned, healthy neonatal Standardbred foals that ranged in age from 5 to 9 days old and body weight from 66 to 77 kg were evaluated in the study. Each foal was maintained with its dam at the University of Illinois Equine Breeding Farm until the day before initiation of experimental procedures, at which time the foal and its dam were transported to the University of Illinois Veterinary Teaching Hospital. Each foal was housed with its dam in an appropriately sized stall in the large animal hospital, and all animals were monitored in accordance with standard hospital protocols.
All foals were determined to have adequate colostral immunity on the basis of results of an IgG enzyme immunoassaya that was performed on blood obtained from each foal at 24 hours of age. All foals were determined to be healthy on the basis of results of a physical examination, CBC,b and serum biochemical analysis,c which were performed on blood samples obtained prior to transport to the veterinary teaching hospital, and results of a venous blood gas analysisd that was performed immediately before anesthesia induction. Each foal underwent a 2DE examination while restrained in a standing position prior to anesthesia induction to verify that it did not have any gross structural cardiac abnormalities. Any foal with evidence of structural or functional cardiac disease or any other medical condition was excluded from the study.
Anesthesia and instrumentation
For each foal, a 14-gauge over-the-wire cathetere was aseptically placed in the left jugular vein for administration of all anesthetic medications and IV fluids and measurement of CO by UDCO. Immediately before imaging, each foal received butorphanol tartratef (0.05 mg/kg, IV) and midazolamg (0.1 mg/kg, IV). Once sedated, the foal was positioned in sternal recumbency on a raised gurney, and anesthesia was induced with propofolh (4 to 6 mg/kg, IV, to effect). After orotracheal intubation with an appropriately sized cuffed endotracheal tube (internal diameter, 12 to 14 mm), anesthesia was maintained with isofluranei in oxygen (mean flow rate, 2 to 3 L/min) delivered by a standard anesthesia machinej via a rebreathing circle system. Each foal was allowed to breath spontaneously throughout the UDCO and echocardiographic procedures but was mechanically ventilated (8 to 10 breaths/min; tidal volume, 10 to 15 mL/kg) during cMRI scanning to control abdominal and thoracic wall movement during image acquisition.
Once each foal was anesthetized but before UDCO measurement and cMRI scanning, a 20-gauge, 4.2-cm catheterk was aseptically placed in the right metatarsal artery. While a foal was anesthetized, it was instrumented with a multiparameter patient monitorl that provided a continuous base-apex ECG tracing and was used to monitor the following physiologic variables at 5-minute intervals: respiratory rate, end-tidal partial pressure of CO2, oxygen saturation as measured by pulse oximetry, end-tidal isoflurane concentration, and direct MAP and systolic and diastolic arterial pressures.
Each foal received a balanced electrolyte solutionm (10 mL/kg/h, IV) to maintain normovolemia while it was anesthetized. Dobutaminen (2 to 12 μg/kg/min, IV, to effect) was administered as necessary to maintain the MAP > 60 mm Hg.
Experimental design
Each foal underwent UDCO measurement and complete 2DE, TEE, and cMRI examinations. The UDCO measurement was performed simultaneously with 2DE. For each foal, a random number generatoro was used to determine whether UDCO measurement with simultaneous 2DE examination or cMRI was performed first. All echocardiographic examinations were performed by a board-certified veterinary cardiologist (RCF). All echocardiographic and cMRI loops and images were stored digitally for off-line analysis at a later time. Images were analyzed in a randomized order by investigators who were unaware of (blinded to) subject information and study number. Echocardiographic images were independently evaluated by 3 investigators (RCF, SK, and JPS). The cMRI loops and images were also independently evaluated by 3 investigators (RCF, JPS, and KML).
2DE
Echocardiographic examinations were performed by use of an ultrasonographic scannerp equipped with a phased-array 5-MHz transducer, 6T TEE transducer, and simultaneous ECG monitoring. Each foal was positioned in sternal recumbency on a raised gurney for 2DE, which included 2-D, M-mode, spectral, and color-flow Doppler imaging. Digital images were obtained in accordance with standard techniques and imaging planes as described36 and sent to a workstation equipped with a software packageq for off-line analysis.
To assess the sizes of individual chambers, images were optimized to include the apex of both ventricles in the R4CH and biplane images, with care taken to ensure that the maximal length of the LV was captured in the images. Measurements were repeated for 5 discrete cardiac cycles obtained during the same anesthetic session, and mean values were calculated and used for statistical analysis.
M-mode images of the LV were acquired in real time from 2DE-guided images, and the LV internal dimension was measured at end diastole and end systole as described.36 The LVEDV and LVESV were calculated by use of the Teichholz method.37 Cardiac output was determined by use of information acquired from the instantaneous ECG (R-R interval and heart rate) during image acquisition.
The monoplane SMOD was used to measure the LV on R4CH and biplane images as described.10,38 Briefly, the LV endocardial border was manually traced at end diastole and end systole. The papillary muscles and trabeculae were not traced around and were included in the LV volume calculations (Figure 1). The LVEDV, LVESV, and LVEF were automatically calculated as the summation of 15 elliptical disks by use of a clinical software program.q Cardiac output was determined by use of the heart rate obtained during imaging acquisition.
Representative R4CH 2DE images of the heart of a healthy 1-week-old Standardbred foal obtained at end diastole (A) and end systole (B) that depict the 15 elliptical disks that were created and summed by a clinical software program to calculate the LVEDV (green lines) and LVESV (blue lines), respectively. The LV endocardial border was manually traced without exclusion of the papillary muscles and trabeculae; thus, the papillary muscles and trabeculae were included in the volume calculations. The simultaneous ECG tracing at the time of image acquisition appears at the bottom of each image.
Citation: American Journal of Veterinary Research 81, 12; 10.2460/ajvr.81.12.930
2-D TEE
A TEE transducerr (diameter, 10.5 mm; length, 110 cm) was advanced through a circular mouth guard and into the esophagus until images of the LV were obtained from the middle position as described.39 Briefly, the middle position was achieved by unflexing the probe tip while imaging in the transverse plane until a longitudinal 4-chamber view of the heart was obtained. Images were optimized for maximum LV chamber size, endocardial detail, and visualization of the apex, and LV measurements were obtained.
Monoplane SMOD was used to measure LV volume. The LV endocardial border was manually traced at end diastole and end systole. The LVEDV and LVESV were determined as described for 2DE, and CO was determined by use of the heart rate obtained during image acquisition.
cMRI
All cMRI sequences were acquired by use of a 3-T MRI scanners with an 18-channel, phased-array body coil and with system-associated software.s Scout images were used to identify the long- and short-axis views of the LV as well as the 2-, 3-, and 4-chamber views of the heart. Thereafter, cine images of 3 long-axis (2-, 3-, and 4-chamber) views were acquired by use of a b-SSFP sequence in combination with parallel imaging and retrospective gating during an inspiratory breath hold. Frequency scout scans were obtained before each cine image, and an optimal frequency was subjectively assessed and determined by the operator (RCF). Frequencies (range, −200 to 200 Hz) were optimally adjusted prior to each cine loop on the basis of the frequency scout.34 Left ventricular systolic function and morphology were imaged by use of b-SSFP sequencing, with coverage of the heart from the aortic root to the LV apex by means of gapless short-axis slices with retrospective gating and multiple end-expiratory breath holds. Gated b-SSFP cine mode was used to acquire dynamic loops of the heart and stacks of short-axis images from the aortic root to LV apex with the following parameters: slice thickness, 5 mm; field of view, 500 mm; repetition time, 53.5 milliseconds; echo time, 1.9 milliseconds; and flip angle, 44°. Images were stored digitally for off-line analysis. Volumes were calculated by use of system software.t All measurements were calculated for 3 cardiac cycles, and the means were calculated and used for statistical analyses.
Determination of LV volume was assessed from the first basal slice in which 50% of the circumference of the LV cavity was surrounded by myocardial tissue to the last apical slice of the LV cavity. For each slice, the endocardial border was manually traced beginning with slices obtained at end diastole (point of the cardiac cycle at which the LV cavity was at its largest) to those obtained at end systole (point of the cardiac cycle at which the LV cavity was at its smallest). The papillary muscles and trabeculae were traced around and therefore excluded from calculations of LV volumes (Figure 2). The LVEDV, LVESV, and EF were calculated with the SMOD equation by the system software,t and CO was determined by use of the heart rate obtained during image acquisition.
Representative short-axis b-SSFP cMRI sequence of the heart of a healthy 1-week-old Standardbred foal that depicts the dimensions used to calculate the LVEDV and LVESV. Determination of LV volume was assessed from the first basal slice in which 50% of the circumference of the LV cavity was surrounded by myocardial tissue to the last apical slice of the LV cavity. For each slice, the endocardial border was manually traced (red line) beginning at end diastole (point of the cardiac cycle at which the LV cavity was at its largest; A) to end systole (point of the cardiac cycle at which the LV cavity was at its smallest; I). Notice that the papillary muscles and trabeculae were traced around and therefore excluded from LV volume calculations. The LVEDV and LVESV were calculated with the SMOD equation by the cMRI system software.
Citation: American Journal of Veterinary Research 81, 12; 10.2460/ajvr.81.12.930
UDCO
A minimally invasive hemodynamic measurement systemu was used to calculate the CO by UDCO as described in humans,14 foals,5 juvenile horses,12 and dogs.13 Briefly, customized tubingv was used to establish an extracorporeal circuit between the catheter in the right metatarsal artery (arterial catheter) and the catheter in the left jugular vein (venous catheter). Two ultrasound velocity sensorsw were placed along the circuit to detect changes in blood velocity; 1 was positioned upstream from the venous catheter (venous sensor), and 1 was positioned downstream from the arterial catheter (arterial sensor). A 30-mL bolus of warm (37°C) sterile saline solution was injected into the venous catheter. The bolus of saline solution caused a transient dilution of blood, which produced changes in blood velocity that were detected by the arterial sensor. A specialized computer with automated softwareu was used to integrate information delivered from the velocity sensors to generate dilution curves and quantify CO. For each foal, the CO was determined from the mean of 3 measurements obtained at 3-minute intervals. All UDCO measurements were performed by the same investigator (SCC-P).
Statistical analysis
For all measurement modalities (biplane, R4CH, and M-mode 2DE, TEE, UDCO, and cMRI), CI was calculated as the CO divided by body surface area. Data distributions for continuous variables were assessed for normality by use of the Shapiro-Wilk and Anderson-Darling tests. Descriptive statistics were generated,x and the mean ± SD, median (range), and CV were reported for all variables of interest (LVEDV, LVESV, LVEF, LVSV, CO, CI, heart rate, and respiratory rate). The effect of measurement modality on each variable of interest was assessed with a 1-way ANOVA,y which included a random effect to account for repeated measurements within each foal. Data underwent a logarithmic transformation to normalize the distribution of nonparametric variables prior to ANOVA. When necessary, post hoc pairwise comparisons were performed with the Bonferroni adjustment to control for type I error inflation.
Additionally, for each variable of interest, the Spearman correlation coefficienty (rs) and OVLy were calculated between the measurement obtained by cMRI (referent) and the respective corresponding measurements obtained by the 2DE, TEE, and UDCO modalities. The Spearman correlation coefficient was calculated instead of the Pearson correlation coefficient because of the small study population (n = 6) and fact that the data were not normally distributed for some variable-modality combinations. The OVL is a measure of the area intersected by 2 or more probability density functions. It is not constricted by assumptions about the data distributions for the variables being compared and can range in value from 0 (no overlap of data distributions) to 1 (complete overlap of data distributions). The Bland-Altman methodz was used to calculate the mean bias and associated interquartile ranges and 95% confidence limits between measurements obtained by cMRI and the respective corresponding measurements obtained by the other modalities. Additionally, the Bland-Altmanz method was used to assess the mean bias and associated 95% confidence limits between UDCO and 2DE with respect to CO and CI. Values of P < 0.05 were considered significant for all analyses.
The ICC was calculated to estimate the overall concordance among the measurements recorded by the 3 independent observers. The ICC could range in value from 0 to 1, with 0 indicating no agreement and 1 indicating perfect agreement among the measurements obtained by the 3 observers. Intraclass correlation coefficients < 0.5, between 0.5 and 0.75, between 0.75 and 0.9, and > 0.9 were considered an indication of poor, moderate, good, and excellent concordance of the measurements among observers, respectively.40
Results
Foals
Complete 2DE examinations, UDCO assessments, and cMRI scans were obtained for all foals. However, TEE examinations were completed in only 5 of the 6 foals. A TEE examination was attempted but not completed for 1 foal because of the inability to advance the TEE transducer to the level of the esophagus where images could be obtained. Two foals had hemodynamically irrelevant patent ductus arteriosus identified by TEE, which was not detected by physical examination or 2DE. For all 6 foals, the mean ± SD duration of anesthesia was 159 ± 11 minutes. The time required to complete cMRI scanning ranged between 50 and 60 minutes. It took 1 to 3 minutes to acquire each 2DE view and 15 to 25 minutes to acquire all TEE images.
The mean ± SD respiratory rate was 11 ± 4 breaths/min during echocardiographic imaging and UDCO measurements. The mean ± SD heart rate was 97 ± 23 beats/min for the entire experimental period. However, the mean ± SD heart rate during cMRI sequencing (79 ± 10 beats/min) was significantly (P = 0.002) lower than that during acquisition of R4CH images (98 ± 14 beats/min). The mean heart rate did not differ significantly between any other imaging or measurement modalities. For all 6 foals, the mean MAP remained > 60 mm Hg for the duration of the experimental period, although the mean ± SD MAP during cMRI scanning (66.13 ± 3.6 mm Hg) was significantly (P = 0.02) lower than that during echocardiographic imaging and UDCO measurements (75 ± 4.6 mm Hg).
LV measurements
Descriptive data for LVEDV, LVESV, LVEF, LVSV, CO, and CI as determined by all modalities were summarized (Table 1). Compared with cMRI, all 2DE and TEE modalities underestimated LVEDV and LVESV and overestimated LVEF, CO, and CI (Table 2). Compared with cMRI, LVSV was underestimated by the biplane, R4CH, and TEE modalities and overestimated by the M-mode and UDCO modalities. The magnitude of bias was small for CO and CI measurements derived from R4CH and TEE images, although the 95% confidence limits were wider for CO and CI measurements derived from TEE images.
Descriptive statistics for select LV variables derived from various 2DE, TEE, and cMRI images and UDCO for 6 healthy 1-week-old Standardbred foals.
| Variable | Measurement modality | Mean ± SD | Median (range) | CV (%) |
|---|---|---|---|---|
| LVEDV (mL) | Biplane* | 159.11 ± 33.07 | 155 (116–208) | 20.78 |
| R4CH | 163.22 ± 40.50 | 154 (110–233) | 24.81 | |
| M-mode | 183.78 ± 37.58 | 185 (115–246) | 20.45 | |
| TEE† | 157.53 ± 20.4 | 154 (122–189) | 12.95 | |
| cMRI | 214.85 ± 25.61 | 217.3 (173–249.3) | 11.92 | |
| LVESV (mL) | Biplane | 65.61 ± 16.97 | 61.5 (45–107) | 25.86 |
| R4CH* | 74.56 ± 20.30 | 65.5 (51–109) | 27.23 | |
| M-mode* | 64.83 ± 22.03 | 57 (39–105) | 33.98 | |
| TEE*† | 67.47 ± 10.73 | 70 (40–79) | 15.90 | |
| cMRI | 117.02 ± 15.79 | 123.4 (96.8–133.9) | 13.49 | |
| LVEF (%) | Biplane | 58.67 ± 6.37 | 58.29 (45.74–69.31) | 10.86 |
| R4CH | 54.02 ± 7.05 | 53.54 (43.48–67.20) | 13.05 | |
| M-mode | 64.55 ± 9.87 | 67.90 (40.68–78.19) | 15.29 | |
| TEE† | 56.44 ± 9.66 | 55.43 (35.25–74.52) | 17.12 | |
| cMRI | 45.46 ± 4.60 | 44.38 (40.78–54.06) | 10.12 | |
| LVSV (mL) | Biplane | 93.50 ± 22.97 | 93 (59–140) | 24.57 |
| R4CH | 89.44 ± 25.22 | 83.50 (56–139) | 28.20 | |
| M-mode | 119.61 ± 30.27 | 123.5 (70–166) | 25.31 | |
| TEE† | 89.47 ± 23.98 | 91 (43–127) | 26.80 | |
| UDCO | 133.63 ± 13.95 | 134.5 (113–159) | 10.44 | |
| cMRI | 97.82 ± 16.21 | 98.45 (76.50–115.30) | 16.57 | |
| CO (L/min) | Biplane | 9.34 ± 2.17 | 9.14 (5.02–12.88) | 23.23 |
| R4CH* | 8.63 ± 2.22 | 8.25 (5.95–14.04) | 25.72 | |
| M-mode | 12.35 ± 4.22 | 13.51 (6.58–20.05) | 34.17 | |
| TEE† | 8.45 ± 3.99 | 7.57 (3.80–14.26) | 47.19 | |
| UDCO | 10.85 ± 1.51 | 10.6 (8.1–13.1) | 13.91 | |
| cMRI | 7.74 ± 1.67 | 7.19 (5.97–10.72) | 21.58 | |
| CI (L/min/m2) | Biplane | 5.93 ± 1.83 | 5.86 (2.97–9.54) | 30.92 |
| R4CH* | 4.96 ± 1.78 | 4.80 (2.29–8.68) | 35.95 | |
| M-mode | 7.48 ± 2.71 | 8.07 (3.23–11.93) | 36.21 | |
| TEE† | 5.06 ± 2.12 | 4.52 (2.29–8.88) | 41.83 | |
| UDCO | 6.24 ± 1.18 | 6.38 (4.08–7.94) | 18.89 | |
| cMRI | 4.53 ± 0.98 | 4.36 (3.11–5.99) | 21.60 | |
| Heart rate (beats/min) | Biplane* | 102 ± 22 | 91 (83–132) | 21.84 |
| R4CH | 98 ± 14 | 100 (81–118) | 13.85 | |
| M-mode* | 102 ± 23 | 97 (79–137) | 22.21 | |
| TEE† | 91 ± 26 | 90 (63–128) | 28.49 | |
| UDCO | 80 ± 9 | 81 (66–91) | 11.63 | |
| cMRI | 79 ± 10 | 80 (63–93) | 12.64 |
Data were not normally distributed for this variable–measurement modality combination.
A TEE examination was attempted but not completed for 1 foal because of the inability to advance the TEE transducer to the level of the esophagus where images could be obtained; therefore, values represent data for only 5 of the 6 foals.
Results of Bland-Altman analyses and OVL and Spearman correlation coefficients for comparisons of select measures of LV volume and function derived from various 2DE and TEE images and UDCO with those derived from cMRI (referent) for the foals of Table 1.
| Bias | Spearman correlation | |||||
|---|---|---|---|---|---|---|
| Variable | Measurement modality | Mean ± SD | 95% confidence limits | OVL | rS | P |
| LVEDV (mL) | Biplane | −55.74 ± 25.31 | −68.33 to −43.15 | 0.45 | 0.66 | 0.16 |
| R4CH | −51.63 ± 29.59 | −66.34 to −36.91 | 0.52 | 0.77 | 0.07 | |
| M-mode | −31.07 ± 20.43 | −41.23 to −20.91 | 0.67 | 0.60 | 0.21 | |
| TEE | −65.63 ± 29.63 | −82.04 to −49.21 | 0.30 | 0.30 | 0.62 | |
| LVESV (mL) | Biplane | −51.41 ± 15.07 | −58.90 to −43.91 | 0.20 | 0.66 | 0.16 |
| R4CH | −42.46 ± 13.71 | −49.28 to −35.65 | 0.36 | 0.89 | 0.02 | |
| M-mode | −52.18 ± 17.04 | −60.66 to −43.71 | 0.28 | 0.83 | 0.04 | |
| TEE | −53.59 ± 15.75 | −62.32 to −44.87 | 0.08 | 0.80 | 0.10 | |
| LVEF (%) | Biplane | 13.20 ± 7.02 | 9.71 to 16.69 | 0.31 | 0.37 | 0.47 |
| R4CH | 8.56 ± 9.55 | 3.81 to 13.31 | 0.53 | 0.09 | 0.87 | |
| M-mode | 19.09 ± 9.95 | 14.14 to 24.03 | 0.26 | 0.03 | 0.96 | |
| TEE | 10.71 ± 10.91 | 4.67 to 16.76 | 0.62 | 0.50 | 0.39 | |
| LVSV (mL) | Biplane | −4.32 ± 22.87 | −15.69 to 7.06 | 0.80 | 0.32 | 0.54 |
| R4CH | −8.37 ± 27.09 | −21.85 to 5.10 | 0.73 | 0.14 | 0.78 | |
| M-mode | 21.79 ± 26.81 | 8.46 to 35.13 | 0.60 | 0.52 | 0.29 | |
| TEE | −12.61 ± 28.74 | −28.53 to 3.30 | 0.81 | 0.05 | 0.93 | |
| UDCO | 36.89 ± 21.91 | 13.90 to 59.88 | 0.39 | 0.20 | 0.70 | |
| CO (L/min) | Biplane | 1.61 ± 1.88 | 0.67 to 2.54 | 0.66 | 0.49 | 0.33 |
| R4CH | 0.89 ± 2.69 | −0.44 to 2.23 | 0.84 | 0.09 | 0.87 | |
| M-mode | 4.62 ± 3.67 | 2.79 to 6.44 | 0.45 | 0.43 | 0.40 | |
| TEE | 0.36 ± 3.97 | −1.84 to 2.56 | 0.62 | 0.10 | 0.87 | |
| UDCO | 3.12 ± 2.92 | 0.06 to 6.18 | 0.43 | 0.31 | 0.56 | |
| CI (L/min/m2) | Biplane | 0.99 ± 1.13 | −0.19 to 2.17 | 0.54 | 0.49 | 0.36 |
| R4CH | 0.55 ± 1.60 | −1.13 to 2.23 | 0.82 | 0.24 | 0.91 | |
| M-mode | 2.75 ± 2.27 | 0.36 to 5.14 | 0.24 | 0.43 | 0.42 | |
| TEE | 0.26 ± 2.46 | −2.79 to 3.31 | 0.67 | 0.10 | 0.95 | |
| UDCO | 1.86 ± 1.72 | 0.06 to 3.66 | 0.64 | 0.18 | 0.54 | |
The OVL is a measure of the area intersected by 2 or more probability density functions and can range in value from 0 (no overlap of data distributions) to 1 (complete overlap of data distributions).
The R4CH-derived LVESV was strongly and positively correlated with the cMRI-derived LVESV (rs = 0.89; P = 0.02), despite the fact that the mean bias between the 2 measures was large (Table 2).
When comparing UDCO with 2DE, UDCO overestimated CO, compared with R4CH (mean bias, 2.22 L/min; 95% confidence limits, −3.05 to 7.49 L/min) and biplane (mean bias, 1.51 L/min; 95% confidence limits, −3.31 to 6.33 L/min) and overestimated CI, compared with R4CH (mean bias, 1.31 L/min/m2; 95% confidence limits, −1.90 to 4.51 L/min/m2) and biplane (mean bias, 0.87 L/min/m2; 95% confidence limits, −1.89 to 3.63 L/min/m2).
ICC
The ICCs indicated there was good to excellent concordance among the 3 observers for all LV and CO measurements derived from 2DE and cMRI images (Table 3). However, concordance among the 3 observers was poor to moderate for all TEE-derived measurements.
Intraclass correlation coefficients for select measures of LV volume and function derived from various 2DE, TEE, and cMRI images by 3 independent observers for the foals of Table 1.
| Variable | Measurement modality | ICC |
|---|---|---|
| LVEDV | Biplane | 0.90 |
| R4CH | 0.89 | |
| M-mode | 0.91 | |
| TEE | 0.36 | |
| cMRI | 0.75 | |
| LVESV | Biplane | 0.72 |
| R4CH | 0.77 | |
| M-mode | 0.92 | |
| TEE | 0.14 | |
| cMRI | 0.66 | |
| LVEF | Biplane | 0.67 |
| R4CH | 0.63 | |
| M-mode | 0.86 | |
| TEE | 0.27 | |
| cMRI | 0.83 | |
| LVSV | Biplane | 0.88 |
| R4CH | 0.81 | |
| M-mode | 0.87 | |
| TEE | 0.40 | |
| cMRI | 0.90 | |
| CO | Biplane | 0.84 |
| R4CH | 0.77 | |
| M-mode | 0.93 | |
| TEE | 0.69 | |
| cMRI | 0.93 |
Discussion
The present study was the first to compare measurements of LV volume and function derived from 2DE, TEE, and UDCO modalities with those derived from cMRI, the currently accepted reference standard for such measurements, in healthy neonatal foals. Quantification of cardiac size and function in foals is most commonly performed by use of 2DE alone or in combination with M-mode echocardiography and is dependent on operator technique, acoustic windows, and geometric assumptions regarding heart shape. In the present study, multiple statistical methods were used to draw conclusions regarding which imaging or measurement modalities yielded measurements that most closely agreed with cMRI-derived measurements.
In the present study, we calculated descriptive statistics (mean ± SD, median [range]) for LV measurements derived from the various modalities and compared those with measurements derived from cMRI by various statistical methods. It is important to note that the fractile of a mean varies with the shape of the data distribution and sample size. Thus, a statistically significant difference does not imply anything about the clinical relevance of the difference between a measurement derived from one particular modality and that derived from cMRI. Consequently, a measurement derived from an alternate modality must have both a correlation with and OVL distribution similar to the corresponding cMRI-derived measurement to retain qualitative and quantitative relevance. Even when those requirements are met, a significant difference between means may or may not be clinically relevant.41
For the foals of the present study, all LV measurements derived from 2DE and TEE modalities were significantly underestimated, compared with LV measurements derived from cMRI. The R4CH obtained by 2DE might be considered superior for assessing LVEDV and LVESV relative to the other echocardiographic imaging planes evaluated in the present study because those measurements were strongly and positively correlated with and had fairly high OVLs, compared with the LVEDV and LVESV derived from cMRI. Nevertheless, LV dimensions derived from 2DE should not be used interchangeably with those derived from cMRI. Significant underestimation of LV dimensions by 2DE modalities relative to cMRI was most likely attributable to the geometric assumptions required for calculation from 2-D images and the inherent restrictions in viewing planes during 2DE imaging.16 Contrary to reports17,33,34 involving dogs, the use of TEE did not improve the accuracy of LV dimension estimates relative to those estimated from cMRI for the foals of the present study.
Results of the present study suggested that TEE was possible, albeit with substantial challenges, in neonatal foals. Images obtained by TEE were often of poor diagnostic quality, and LV measurements derived from those images had high CV and low repeatability. When TEE images of adequate diagnostic quality were successfully obtained, the imaging planes, particularly those of the heart base, were superior to those obtained by 2DE, which resulted in the identification of patent ductus arteriosus that was not detected during physical examination or 2DE imaging for 2 of the 6 foals.
In the present study, biplane, R4CH, and TEE images all yielded measurements of CO and CI with acceptable bias relative to cMRI-derived estimates. Surprisingly, of all modalities assessed, the TEE-derived CO (mean bias, 0.36 L/min; 95% confidence limits, −1.84 to 2.56 L/min) and CI (mean bias, 0.26 L/min/m2; 95% confidence limits, −2.79 to 3.31 L/min/m2) had the lowest mean bias, with moderate OVLs (CO, 0.62; CI, 0.67) relative to cMRI-derived estimates. The R4CH-derived estimates of CO (mean bias, 0.89 L/min; 95% confidence limits, −0.44 to 2.23 L/min) and CI (mean bias, 0.55 L/min/m2; 95% confidence limits, −1.13 to 2.23 L/min/m2) had the next lowest mean bias, but greater OVLs (CO, 0.84; CI, 0.82) relative to cMRI-derived estimates, compared with the OVLs for the TEE-derived estimates. For the foals and experimental conditions of the present study, UDCO resulted in CO (mean bias, 3.12 L/min; 95% confidence limits, 0.06 to 6.18 L/min) and CI (mean bias, 1.86 L/min/m2; 95% confidence limits, 0.06 to 3.66 L/min/m2) estimates with unacceptable bias relative to cMRI-derived estimates.
Although cMRI is considered the reference standard for evaluation of cardiac size and function, it is not currently readily available or widely used in veterinary medicine. Therefore, cMRI may not be accessible or practical for diagnostic use in critically ill veterinary patients. For the foals of the present study, UDCO-derived estimates of LV function were compared with those derived from routine 2DE images by means of Bland-Altman analyses. For neonatal foals, measurement of CO and CI from biplane and R4CH images is reported to be an accurate and noninvasive method for estimating those variables.10 The UDCO overestimated both CO and CI when compared with CO and CI measurements derived from R4CH and biplane images. The mean bias was smallest between the UDCO and biplane methods for both CO (mean bias, 1.51 L/min; 95% confidence limits, −3.31 to 6.33) and CI (0.87 L/min/m2; 95% confidence limits, −1.89 to 3.63 L/min/m2). Results of other studies indicate that estimation of CO by UDCO5 or select 2DE images10 is acceptable for anesthetized neonatal foals when CO estimated from a lithium dilution method is used as the referent. Results of the present study suggested that, although estimation of CO and CI by UDCO and from R4CH and biplane images may provide useful information for neonatal foals, it may not be appropriate to use those measurements interchangeably for monitoring of clinical patients.
In the present study, the ICCs calculated for 3 independent observers (an indication of the repeatability of the measurements) were classified as good to excellent for quantitative measurements of LV volume and function derived from biplane, R4CH, M-mode, and cMRI images. The ICCs calculated for LVESV were generally lower than those calculated for the other variables assessed for all measurement modalities except M-mode echocardiography, likely owing to the subjectiveness associated with determining end systole. When measurements are obtained from M-mode images, systole is defined by the apogee of the interventricular septum; therefore, the ICCs for LVEDV (0.91) and LVESV (0.92) were similar, and both measures were considered to have excellent ICCs. The ICCs determined for measurements of LV volume and function derived from TEE images were generally poor (< 0.50).
The present study had several limitations. We chose to use cMRI-derived measurements as the reference standard, and our conclusions might have differed had measurements derived from a different modality (eg, thermodilution or lithium dilution) been used as the reference standard. Additionally, all foals assessed in the present study were clinically normal and anesthetized for cardiac imaging. The estimated measures of LV volume and function may differ for foals that are not anesthetized, particularly those with cardiac disease or hemodynamic disorders. The acquisition of cardiac images by use of the various modalities could not be conducted simultaneously. Therefore, although the foals were anesthetized and attempts were made to maintain the foals at a consistent hemodynamic steady state for all imaging, the duration of anesthesia at the time of image acquisition could have affected comparison of measurements among the modalities. The order in which images were obtained by the various modalities was randomized to minimize the effect of anesthesia duration on comparison of measurements among modalities. Moreover, UDCO was performed simultaneously with 2DE image acquisition. We do not believe that the CV within each measurement modality was affected by anesthesia duration because all images were acquired over a short period of time, during which there was little change in the hemodynamic variables of individual foals. However, the foals of the present study were anesthetized and immobile during cardiac image acquisition, which might have minimized the CVs reported for the measurements in this study, and a similar precision and accuracy should not be expected in unanesthetized foals.
The foals of the present study were mechanically ventilated and remained completely immobile throughout cMRI scanning, and all cMRI images were obtained at end-expiration. Conversely, the foals were allowed to breathe spontaneously during the acquisition of all echocardiographic images and UDCO measurements; therefore, echocardiographic images were obtained at various times throughout the respiratory cycle. Although we avoided capturing echocardiographic images at peak inspiration, the acquisition of images at various times during the respiratory cycle can affect cardiac size and function and may have contributed to the discrepancies observed among measurements derived from UDCO and echocardiographic and cMRI images. Those discrepancies may have also been at least partially attributable to small differences in hemodynamic variables within individual foals. The position of each foal had to be modified slightly to optimize visualization of cardiac structures during 2DE and TEE image acquisition, and changes in subject positioning may have contributed to the discrepancies in MAP and heart rate observed between echocardiographic and cMRI image acquisition. Changes in the heart rate between the echocardiographic and cMRI imaging might have affected the magnitude of the bias calculated for variables dependent on heart rate, such as CO and CI.
On the basis of the results of the present study, we concluded that 2DE, TEE, UDCO, and cMRI can be used to assess LV volume and function in neonatal foals; however, TEE may not be possible in all foals. For assessment of cardiac function on the basis of heart size, measurements derived from cMRI should be used as the reference standard when accuracy is of paramount importance. Although measurements of LV volume and function derived from all assessed modalities appeared to be clinically acceptable, they should not be used interchangeably when monitoring patient progress.
Acknowledgments
Supported by USDA Hatch funds awarded during fiscal year 2017.
Presented in part at the World Congress of Veterinary Anesthesiology, Venice, Italy, September 2018, and the American College of Veterinary Internal Medicine Forum, Phoenix, 2019.
ABBREVIATIONS
| 2DE | 2-D transthoracic echocardiography |
| Biplane | Left apical 4-chamber and 2-chamber view |
| b-SSFP | Balanced steady-state free precession |
| CI | Cardiac index |
| cMRI | Cardiac MRI |
| CO | Cardiac output |
| CV | Coefficient of variation |
| ICC | Intraclass correlation coefficient |
| LV | Left ventricle |
| LVEDV | Left ventricular end-diastolic volume |
| LVEF | Left ventricular ejection fraction |
| LVESV | Left ventricular end-systolic volume |
| LVSV | Left ventricular stroke volume |
| MAP | Mean arterial pressure |
| M-mode | Right parasternal short-axis M-mode view |
| OVL | Overlapping coefficient |
| R4CH | Right parasternal long-axis 4-chamber view |
| SMOD | Simpson method of discs |
| TEE | Transesophageal echocardiography |
| UDCO | Ultrasound velocity dilution cardiac output method |
Footnotes
Idexx Snap foal IgG test kit, Idexx Laboratories Inc, Westbrook, Me.
Cell-dyn 3700 hematology analyzer, GMI Inc, Ramsey, Minn.
Beckman Coulter AU680, Beckman Coulter Inc, Indianapolis, Ind.
Critical Care Express, Nova Biomedical, Waltham, Mass.
MILA International Inc, Florence, Ky.
Butorphic (10 mg/mL), Lloyd Laboratories Inc, Shenandoah, Iowa.
Hospira Inc, Lake Forest, Ill.
PropoFlo (10 mg/mL), Abbott Laboratories, North Chicago, Ill.
IsoFlo, Abbott Laboratories, North Chicago, Ill.
Millennium small animal machine, Eagle Eye Anesthesia Inc, Jacksonville, Fla.
Becton, Dickinson, and Co, Franklin Lakes, NJ.
Datascope Passport, Datascope Corp, Paramus, NJ.
Lactated Ringer solution, Hospira Inc, Lake Forest, Ill.
Bedford Laboratories, Bedford, Ohio.
Random integer generator, Randomness and Integrity Services Ltd, Dublin, Ireland. Available at: www.random.org. Accessed Apr 2, 2017.
Vivid E9, GE Medical System, Waukesha, Wis.
EchoPac version BT13, GE Medical System, Waukesha, Wis.
6Tc transesophageal transducer, GE Medical System, Waukesha, Wis.
MAGNETOM Skyra 3T, Siemens Healthcare, Erlangen, Germany.
Argus version syngo MR D11D, Siemens Healthcare, Erlangen, Germany.
COstatus system, Transonic Systems Inc, Ithaca, NY.
COStatus AV Loop HCS3011, Transonic Systems Inc, Ithaca, NY.
COStatus flow/dilution sensors, HC2T, Transonic System Inc, Ithaca, NY.
Systat Inc, San Jose, Calif.
SAS, version 9.4, SAS Institute Inc, Cary, NC.
Prism 8, GraphPad Software, San Diego, Calif.
References
1. Marr CM. The equine neonatal cardiovascular system in health and disease. Vet Clin North Am Equine Pract 2015;31:545–565.
2. Corley KT, Donaldson LL, Durando MM, et al. Cardiac output technologies with special reference to the horse. J Vet Intern Med 2003;17:262–272.
3. Corley KTT, Donaldson LL, Furr MO. Comparison of lithium dilution and thermodilution cardiac output measurements in anaesthetised neonatal foals. Equine Vet J 2002;34:598–601.
4. Valverde A, Giguère S, Sanchez LC, et al. Effects of dobutamine, norepinephrine, and vasopressin on cardiovascular function in anesthetized neonatal foals with induced hypotension. Am J Vet Res 2006;67:1730–1737.
5. Shih A, Giguère S, Sanchez LC, et al. Determination of cardiac output in neonatal foals by ultrasound velocity dilution and its comparison to the lithium dilution method. J Vet Emerg Crit Care (San Antonio) 2009;19:438–443.
6. Shih A. Cardiac output monitoring in horses. Vet Clin North Am Equine Pract 2013;29:155–167.
7. Mellema M. Cardiac output monitoring. In: Silverstein D, Hopper K, eds. Small animal critical care medicine. St Louis: Saunders Elsevier, 2009;894–898.
8. Floh AA, La Rotta G, Wermelt JZ, et al. Validation of a new method based on ultrasound velocity dilution to measure cardiac output in paediatric patients. Intensive Care Med 2013;39:926–933.
9. McConachie E, Barton MH, Rapoport G, et al. Doppler and volumetric echocardiographic methods for cardiac output measurement in standing adult horses. J Vet Intern Med 2013;27:324–330.
10. Giguère S, Bucki E, Adin DB, et al. Cardiac output measurement by partial carbon dioxide rebreathing, 2-dimensional echocardiography, and lithium dilution method in anesthetized neonatal foals. J Vet Intern Med 2005;19:737–743.
11. Vigani A, Shih A, Queiroz P, et al. Quantitative response of volumetric variables measured by a new ultrasound dilution method in a juvenile model of hemorrhagic shock and resuscitation. Resuscitation 2012;83:1031–1037.
12. Shih AC, Queiroz P, Vigani A, et al. Comparison of cardiac output determined by an ultrasound velocity dilution cardiac output method and by the lithium dilution cardiac output method in juvenile horses with experimentally induced hypovolemia. Am J Vet Res 2014;75:565–571.
13. Shih A, Giguère S, Vigani A, et al. Determination of cardiac output by ultrasound velocity dilution in normovolemia and hypovolemia in dogs. Vet Anaesth Analg 2011;38:279–285.
14. Lindberg L, Johansson S, Perez-de-Sa V. Validation of an ultrasound dilution technology for cardiac output measurement and shunt detection in infants and children. Pediatr Crit Care Med 2014;15:139–147.
15. Tsutsui M, Matsuoka N, Ikeda T, et al. Comparison of a new cardiac output ultrasound dilution method with thermodilution technique in adult patients under general anesthesia. J Cardiothorac Vasc Anesth 2009;23:835–840.
16. Chukwu EO, Barasch E, Mihalatos DG, et al. Relative importance of errors in left ventricular quantitation by two-dimensional echocardiography: insights from three-dimensional echocardiography and cardiac magnetic resonance imaging. J Am Soc Echocardiogr 2008;21:990–997.
17. Domenech O, Oliveira P. Transesophageal echocardiography in the dog. Vet J 2013;198:329–338.
18. Vegas A, Meineri M. Three-dimensional transesophageal echocardiography is a major advance for intraoperative clinical management of patients undergoing cardiac surgery: a core review. Anesth Analg 2010;110:1548–1573.
19. Drees R, Johnson RA, Stepien RL, et al. Quantitative planar and volumetric measurements using 64 MDCT and 3T MRI versus standard 2D and M-mode echocardiography: does anesthetic protocol matter? (Erratum published in Vet Radiol Ultrasound 2016;57:450). Vet Radiol Ultrasound 2015;56:638–657.
20. Sobrino A, Basmadjian AJ, Ducharme A, et al. Multiplanar transesophageal echocardiography for the evaluation and percutaneous management of ostium secundum atrial septal defects in the adult. Arch Cardiol Mex 2012;82:37–47.
21. Gouveia V, Marcelino P, Reuter DA. The role of transesophageal echocardiography in the intraoperative period. Curr Cardiol Rev 2011;7:184–196.
22. Porciello F, Caivano D, Giorgi ME, et al. Transesophageal echocardiography as the sole guidance for occlusion of patent ductus arteriosus using a canine ductal occlude in dogs. J Vet Intern Med 2014;28:1504–1512.
23. Glaus T, Sommerfield N. Using technology to find the secret places of the heart. Vet J 2014;200:216–217.
24. Linton RA, Young LE, Marlin DJ, et al. Cardiac output measured by lithium dilution, thermodilution, and transesophageal Doppler echocardiography in anesthetized horses. Am J Vet Res 2000;61:731–737.
25. Constantine G, Shan K, Flamm SD, et al. Role of MRI in clinical cardiology. Lancet 2004;363:2162–2171.
26. Asferg C, Usinger L, Kristensen TS, et al. Accuracy of multislice computed tomography for measurement of left ventricular ejection fraction compared with cardiac magnetic resonance imaging and two-dimensional transthoracic echocardiography: a systematic review and meta-analysis. Eur J Radiol 2012;81:e757–e762.
27. Albertí JF, de Diego JJG, Delgado RV, et al. State of the art: new developments in cardiac imaging [in Spanish]. Rev Esp Cardiol 2012;65(suppl 1):24–34.
28. Sieslack AK, Dziallas P, Nolte I, et al. Comparative assessment of left ventricular function variables determined via cardiac computed tomography and cardiac magnetic resonance imaging in dogs. Am J Vet Res 2013;74:990–998.
29. Meyer J, Wefstaedt P, Dziallas P, et al. Assessment of left ventricular volumes by use of one-, two-, and three-dimensional echocardiography versus magnetic resonance imaging in healthy dogs. Am J Vet Res 2013;74:1223–1230.
30. MacDonald KA, Kittleson MD, Garcia-Nolen T, et al. Tissue Doppler imaging and gradient echo cardiac magnetic resonance imaging in normal cats and cats with hypertrophic cardiomyopathy. J Vet Intern Med 2006;20:627–634.
31. Contreras S, Vázquez JM, Miguel AD, et al. Magnetic resonance angiography of the normal canine heart and associated blood vessels. Vet J 2008;178:130–132.
32. Mai W, Weisse C, Sleeper MM. Cardiac magnetic resonance imaging in normal dogs and two dogs with heart base tumor. Vet Radiol Ultrasound 2010;51:428–435.
33. Fries RC, Gordon SG, Saunders AB, et al. Quantitative assessment of two- and three-dimensional transthoracic and two-dimensional transesophageal echocardiography, computed tomography, and magnetic resonance imaging in normal canine hearts. J Vet Cardiol 2019;21:79–92.
34. Coon PD, Pollard H, Furlong K, et al. Quantification of left ventricular size and function using contrast-enhanced real-time 3D imaging with power modulation: comparison with cardiac MRI. Ultrasound Med Biol 2012;38:1853–1858.
35. Sugeng L, Mor-Avi V, Weinert L, et al. Quantitative assessment of left ventricular size and function: side-by-side comparison of real-time three-dimensional echocardiography and computed tomography with magnetic resonance reference. Circulation 2006;114:654–661.
36. Thomas WP, Gaber CE, Jacobs GJ, et al. Recommendations for standards in transthoracic two-dimensional echocardiography in the dog and cat. Echocardiography Committee of the Specialty of Cardiology, American College of Veterinary Internal Medicine. J Vet Intern Med 1993;7:247–252.
37. Teichholz LE, Kreulen T, Herman MV, et al. Problems in echocardiographic volume determinations: echocardiographic-angiographic correlations in the presence or absence of asynergy. Am J Cardiol 1976;37:7–11.
38. Wess G, Mäurer J, Simak J, et al. Use of Simpson's method of disc to detect early echocardiographic changes in Doberman Pinschers with dilated cardiomyopathy. J Vet Intern Med 2010;24:1069–1076.
39. Kienle RD, Thomas WP, Rishniw M. Biplane transesophageal echocardiography in the normal cat. Vet Radiol Ultrasound 1997;38:288–298.
40. Koo TK, Li MY. A guideline of selecting and reporting intraclass correlation coefficients for reliability research (Erratum published in J Chiropr Med 2017;16:346). J Chiropr Med 2016;15:155–163.
41. Martens EP, Pestman WR, de Boer A, et al. The use of the overlapping coefficient in propensity score analysis. Pharmacoepidemiol Drug Saf 2007;16.
