The TandemHeart PTVA system (CardiacAssist, Inc., Pittsburgh, Pennsylvania), is a promising new percutaneous ventricular assist device (pVAD) being utilized in adults.1–3 A 21Fr venous cannula is introduced into the left atrium (LA) from the femoral vein after trans-septal puncture. Blood is then returned to the femoral artery by a centrifugal pump. Use of the current system in smaller children is limited by the fact that LA cannulae both small enough in caliber for a child's femoral vein and long enough to reach the LA from the femoral vein cannot provide adequate flow.
In order to overcome this system limitation, we sought to develop an alternative route for device placement. Trans-hepatic diagnostic and interventional catheterization is well described in pediatrics, and has been shown to be safe and effective.4–6 Compared with the femoral route, trans-hepatic LA cannulation offers both a shorter distance to the LA and access to larger-caliber veins. Thus, trans-hepatic cannulation would allow the use of larger, shorter cannulae, which could potentially allow adequate flow rates.
We hypothesized that trans-hepatic placement of a pVAD was feasible. We developed prototype cannulae based on direct patient measurements and our experience with trans-hepatic catheterization.4,5 Benchtop testing of our prototype cannulae demonstrated that adequate flow rates could indeed be delivered under simulated use conditions.
Materials and Methods
Approval was obtained from the University of Michigan Institutional Review Board. Two hundred consecutive patients admitted to the Pediatric Cardio-Thoracic Intensive Care Unit at the C.S. Mott Children's Hospital were evaluated. Patients 14 years or older or with dextrocardia were excluded.
In order to estimate the length from the trans-hepatic insertion site to the lateral wall of the LA, perioperative chest x-rays were evaluated. The distance from the skin at the mid-axillary line between the 10th and 11th ribs to the left-heart border was measured. When available, catheterization reports were also reviewed for direct intra-cardiac measurements of this distance. The proposed cannula positions are shown in Figure 1.
As the tip of the venous drainage cannula must lie entirely within the LA, we then sought to determine the LA size. Contemporary echocardiograms were reviewed and the linear distance from the atrial septum to the left atrial free wall was measured from the sub-costal coronal projection. This measurement was felt to best represent the maximal length of the perforated portion of the LA cannula. Echocardiograms were not evaluated if a distinct atrial septum was not visualized (i.e., in single ventricle patients after atrial septectomy).
Finally, linear regression was performed on left atrial distance (LAD) and left atrial size (LAS) against patient age and weight. The regression results were used to determine the total length of the trans-hepatic cannula.
A major challenge for pediatric VAD application is the wide range of patient sizes and amount of support needed. We broke our patient population into five groups based on weight (2–3.9 kg, 4–6.9 kg, 7–11.9 kg, 12–19.9 kg, and 20–40 kg), with estimated flow requirements listed in Table 1. Estimated flow needs were calculated based on a typical cardiac index of 3.0 lpm/m2 and the assumption that the pump will provide 50% support. We considered patients over 40 kg to be adequately served by the adult TandemHeart® system.
Proper cannula selection is critical for system design. Because appropriate arterial cannulae are commercially available, only the venous cannulae were developed. The length of the venous cannula selected for each group of patients was based on the LAD at two standard deviations above the mean, with a minimum of 10% extra working length to account for three-dimensional curves in the path. This can be expressed by the equation:
The perforated end of the cannula was designed to fit within the LA as measured from the LAS regression curve. Maximum French sizes for venous cannulae for each group were picked based on our reported experience with trans-hepatic catheterization.4,5 Because arterial return is pump-driven, it is a secondary aspect of the design. Therefore, arterial cannula size was based on our clinical experience with femoral artery catheterization.
After venous cannula lengths for all patient groups were decided, prototype cannulae of the proposed length and diameter were fabricated. For each patient group, the proposed cannula characteristics are listed in Table 1.
System Validation Testing
A prototype pediatric TandemHeart pump7 was connected to the system simulation loop shown in Figure 2. The system was filled with a glycerin and saline mixture mixed to simulate a blood viscosity of 4.0 cP. Benchtop testing was then performed with each of the cannulae pairings listed in Table 1 with a maximum pump speed of 8500 rpm. The tip of the venous cannula was inserted into a fluid column simulating the LA. A Setra pressure transducer was connected to the bottom of the column to monitor the simulated LA pressure (LAP). Simulated mean arterial pressure (MAP) was measured at the tip of the arterial cannula. MAP was adjusted with a tubing clamp.
For each patient group, the fluid level in the reservoir was adjusted to achieve the desired LAP levels (Table 2). After the pump was started at 8500 rpm, the tubing clamp was adjusted to reach the desired MAP. Flow rates in Table 2 were recorded after the flow had stabilized. The pressure at the inlet of the pump was also recorded. This value was used to determine whether the negative pressure in the system was high enough to cause pump cavitation and subsequent hemolysis.
Prism 4 software (GraphPad Software Inc. San Diego, CA) was used for regression curve formation. Square root and logarithmic transformation of the independent variables was performed to establish the best linear fit to the data. Bland-Altman analysis was performed to compare actual to predicted cannula lengths in patients with appropriate catheterization data.8 This method allows the direct comparison of two clinical measurement methods and assesses the systematic differences between them. If one measure consistently underestimates or overestimates the other, it will report this as a bias. In contrast, the correlation coefficient may report perfect correlation despite the consistent differences between the two variables.
Of the 200 evaluated patients, 174 met our inclusion criteria. The median age was 6.5 months with a range from 1day to 155.9 months. The median weight was 6.6 kg with a range from 2.2 to 73.9 kg.
Left atrial distance was plotted against weight and age. The best linear fit was obtained with square root transformation of the independent variables. From this we generated the linear regression curves (Figure 3):
A total of 147 patients had interpretable echocardiograms. Left atrial size was plotted against weight and age. We then generated the regression curves (Figure 4):
In six patients, direct intracardiac measurements of the LAD were available from catheterization data having been taken for other reasons. These distances were compared with our calculated LAD. In these few patients, this showed our calculated LAD compared with our measured LAD with a correlation coefficient of r = 0.86. The overall tendency was to underestimate the direct measurement by an average of 9.7% with a Bland-Altman bias of –1.3 (Figure 5). This bias fell within the amount of extra working length added to each cannula.
Using the proposed cannula pairings, system simulation tests were conducted. The achieved flow rate at the maximum pump speed of 8500 rpm for each pairing is listed in Table 2 with the recorded pump inlet and outlet pressures. In all cases, the proposed design was able to deliver the estimated flow needs for each patient group.
The saturated vapor pressure of plasma at 37°C is 47 mm Hg absolute.9 Pressure below 47 mm Hg absolute will cause cavitation and therefore hemolysis. The lowest pressure in the system was –98 mm Hg gauge pressure, corresponding to 662 mm Hg absolute. Therefore, cavitation caused hemolysis is not predicted to occur.
Pediatric use of ventricular assist devices is increasing steadily.10–12 In critically ill patients, surgical implantation is fraught with complications.13–15 The only alternative currently available to pediatric patients is extracorporeal membrane oxygenation. Although this option is well established, there are many potential complications as well.16,17
We recently implanted the TandemHeart pVAD in the youngest patient to date—a 14-year-old girl—and have successfully bridged her to cardiac transplantation. Although we have been able to use the device in a young adolescent, this technology is currently unavailable to smaller children. In our current study, we propose a mechanism that would allow its use in infants and describe the results of our preliminary feasibility analysis.
The major limiting factor to the system is the size of the cannulae necessary to allow adequate inflow into the pump. The resistance to flow through the cannula is dependent on cannula diameter, length and viscosity and can be expressed by the equation:
where η = viscosity, L = length and r = radius.18
We ran our system simulation tests with potential femoral cannulae and, as is shown in Table 3, the pressure drops are prohibitive. Because the trans-hepatic route allows us to both decrease the length and increase the radius of the cannulae, the resistance to flow is significantly lower. Additionally, access to the LA is relatively simple from the trans-hepatic route due to the straight course to the foramen ovale.4 This offers a substantial advantage over the approach from the internal jugular, as access to the LA from the neck is challenging. Based on our benchtop testing, we have been able to develop cannulae that meet the requirements of the system. Our results suggest that the system could be adapted to infants and children: The proposed cannula pairings are predicted to meet or exceed the estimated flow needs without excessive hemolysis.
We also examined the size of the LA to ensure that we would have an adequate chamber in which to position our cannulae. This measurement proved to be adequate for at least a 1 cm multiple-holed cannula. Because LA size tends to vary more by disease state, there was more variability with this measurement. We would anticipate that in patients requiring left ventricular assist, the LA size would tend to be larger.
One weakness of this feasibility study is the noninvasive approximation of the distance from the skin to the LA that was used for cannula design. Our numbers appear to correlate well in the few patients with direct measurements, although further study is clearly needed. To minimize this underestimation, all cannulae were selected to be at least two standard deviations greater than our predicted mean and at least 10% extra working length was added to all cannulae. This gives the added advantage that should significant body wall edema occur there is sufficient cannula length to minimize the risk of cannula migration from the LA.
The next step in our investigation is to confirm our noninvasive measurements with direct patient measurements. Additionally, animal modeling of cannula flow properties will be necessary to confirm our benchtop results. Finally, although no significant effects on liver enzymes are seen during cardiac catheterization,4 it is possible that a longer-term indwelling catheter would adversely affect the liver. Given the liver's regenerative ability, we do not expect this to be a problem, but clearly the safety and efficacy of the system in animal models will have to be confirmed before human applications can be investigated.
Trans-hepatic left atrial cannulation for percutaneous VAD placement is feasible. The shorter distance and larger vein size from the trans-hepatic route allows cannulae to be designed that can deliver higher flow rates. Adequate flow can be delivered by the proposed pediatric pVAD system to support pediatric patients from newborns to adolescents.
The authors thank Dr. Caren Goldberg for statistical assistance.
1.Aragon J, Lee MS, Kar S, et al: Percutaneous left ventricular assist device: “TandemHeart” for high-risk coronary intervention. Cathet Cardiovasc Intervent
65: 346–352, 2005.
2.Pitsis AA, Dardas P, Mezilis N, et al: Temporary assist device for postcardiotomy cardiac failure. Ann Thorac Surg
77: 1431–1433, 2004.
3.Thiele H, Lauer B, Hambrcht R, et al: Reversal of cardiogenic shock by percutaneous left atrial-to-femoral arterial bypass assistance. Circulation
104: 2917–2922, 2001.
4.Shim D, Lloyd TR, Cho KJ, et al: Transhepatic cardiac catheterization in children. Circulation
92: 1526–1530, 1995.
5.Shim D, Lloyd TR, III RHB: Transhepatic therapeutic cardiac catheterization: a new option for the pediatric interventionalist. Cathet Cardiovasc Intervent
47: 41–45, 1999.
6.McLeod KA, Houston AB, Richens T, et al: Transhepatic approach for cardiac catheterization in children: Initial experience. Heart
82: 694–696, 1999.
7.Wang DHJ, Smith DE, Bacha EA, et al: Development of a percutaneous pediatric ventricular assist device. ASAIO J
51: 551–556, 2005.
8.Bland JM, Altman DG: Statistical methods for assessing agreement between two method of clinical measurement. The Lancet
327: 307–310, 1986.
9.Chambers SD, Bartlett RH, Ceccio SL: Hemolytic potential of hydrodynamic cavitation. J Biomech Eng
122: 321–326, 2000.
10.Reinhartz O, Stiller B, Eilers R, et al: Current clinical status of pulsatile pediatric circulatory support. ASAIO J
48: 455–459, 2002.
11.Throckmorton AL, Allaire PE, Gugesell HP, et al: Pediatric circulatory support systems. ASAIO J
48: 216–221, 2002.
12.Weiss WJ: Pulsatile pediatric ventricular assist devices. ASAIO J
51: 540–545, 2005.
13.Rao V, Oz MC, Flannery MA, et al: Revised screening scale to predict survival after insertion of a left ventricular assist device. J Thorac Cardiovasc Surg
125: 855–862, 2003.
14.Oz MC, Goldstein DJ, Pepino P, et al: Screening scale predicts patients successfully receiving long-term implantable left ventricular assist devices. Circulation
92: 169–173, 1995.
15.Morgan JA, John R, Rao V, et al: Bridging to transplant with the HeartMate left ventricular assist device: The Columbia Presbyterian 12-year experience. J Thorac Cardiovasc Surg
127: 1309–1316, 2004.
16.Levi D, Marelli D, Plunkett M, et al: Use of assist devices and ECMO to bridge pediatric patients with cardiomyopathy to transplantation. J Heart Lung Transplant
21: 760–770, 2002.
17.Pagani FD, Aaronson KD, Swaniker F, et al: The use of extracorporeal life support in adult patients with primary cardiac failure as a bridge to implantable left ventricular assist device. Ann Thorac Surg
71: S77–81, 2001.
18.Allen HD, Gutgesell HP, Clark EB, et al: Moss and Adams' Heart Disease in Infants, Children, and Adolescents Including the Fetus and Young Adult. 6th ed.
Philadelphia, Lippincott Williams & Wilkins, 2001.