bs_bs_banner

© 2014 Wiley Periodicals, Inc. and International Center for Artificial Organs and Transplantation

Thoughts and Progress Derivation of Indices of Left Ventricular Contractility in the Setting of Continuous-Flow Left Ventricular Assist Device Support *†Sunil Gupta, *†‡Kavitha Muthiah, *†Kei Woldendorp, *Desiree Robson, *Paul Jansz, and *†‡Christopher S. Hayward *Heart Failure and Transplant Unit, St. Vincent’s Hospital; †Faculty of Medicine, The University of New South Wales; and ‡Victor Chang Cardiac Research Institute, Sydney, New South Wales, Australia Abstract: It is important to accurately monitor residual cardiac function in patients under long-term continuousflow left ventricular assist device (cfLVAD) support. Two new measures of left ventricular (LV) chamber contractility in the cfLVAD-unloaded ventricle include IQ, a regression coefficient between maximum flow acceleration and flow pulsatility at different pump speeds; and K, a logarithmic relationship between volumes moved in systole and diastole. We sought to optimize these indices. We also propose RIQ, a ratio between maximum flow acceleration and flow pulsatility at baseline pump speed, as an alternative to IQ. Eleven patients (mean age 49 ± 11 years) were studied. The K index was derived at baseline pump speed by defining systolic and diastolic onset as time points at which maximum and minimum volumes move through the pump. IQ across the full range of pump speeds was markedly different between patients. It was unreliable in three patients with underlying atrial fibrillation (coefficient of determination R2 range: 0.38–0.74) and also when calculated without pump speed manipulation (R2 range: 0.01– 0.74). The K index was within physiological ranges, but poorly correlated to both IQ (P = 0.42) and RIQ (P = 0.92). In four patients there was excellent correspondence between RIQ and IQ, while four other patients showed a poor relationship between these indices. As RIQ does not require pump speed changes, it may be a more clinically appropriate measure. Further studies are required to determine the validity of these indices. Key Words: Contractility—Left ventricular assist device—Continuous flow—Heart failure.

doi:10.1111/aor.12292 Received October 2013; revised December 2013. Address correspondence and reprint requests to Dr. Christopher S. Hayward, Heart Failure and Transplant Unit, St. Vincent’s Hospital, Victoria Street, Darlinghurst, New South Wales 2010, Australia. E-mail: [email protected]

It has been documented that long-term continuous-flow left ventricular assist device (cfLVAD) support can lead to myocardial recovery, with potential for device explantation (1–7). Thus, it is vital to regularly assess all cfLVAD patients for changes in intrinsic LV chamber function following implantation. Currently, echocardiographic markers such as ejection fraction and fractional shortening remain the mainstay measures, despite inaccuracy related to limited views, user subjectivity, mechanical ventilation, and swelling in the early postoperative period (8,9). Assessment of unloaded LV volumes necessitates cessation of pump function, which is unacceptable due to blood regurgitation through the idle pump (10). Alternatively, markers of left ventricular chamber function derived from pressure–volume measures can be used. However, they are of limited use in the clinical outpatient setting as they require intraventricular catheterization, a time-consuming and invasive process (8,9). Some also rely on generating several pressure–volume loops via inferior vena caval occlusion (11), a pertinent issue as associated reductions in preload can lead to LV suction. The quest for a suitable marker has led to the development of two new hemodynamic indices, IQ and K, specifically formulated for the purpose of monitoring myocardial function under cfLVAD support. In 2010, the IQ index was proposed by Naiyanetr and colleagues as an alternative to pressure–volume measures of myocardial function (10). It is defined as the regression coefficient between the maximum derivative of pump flow (dQ/dtmax) and peak-to-peak variation in flow rate (QP2P) following a series of pump speed changes. However, the requirement of pump speed changes is of clinical concern as it is time-consuming, and excessive variation can lead to the development of adverse symptoms. Furthermore, there are no guidelines specifying the degree of pump speed variation required for IQ calculation. Our study aims to determine the viability of an alternative index, RIQ, which we defined as a simple ratio between dQ/dtmax and QP2P at baseline pump speed only. This would eliminate the need for pump speed changes, negating clinical concerns and time required, and hence making it more clinically applicable. Artificial Organs 2014, ••(••):••–••

2

THOUGHTS AND PROGRESS

The K index is a hemodynamic marker of LV function developed by Ferreira and colleagues in 2011 (12). It is defined as the coefficient of a logarithmic relationship between the total volume of blood moved through the pump in systole and that moved in diastole. Determination of LV volumes in the different phases of the cardiac cycle requires a corresponding electrocardiogram (ECG) trace. This is cumbersome and prone to errors, particularly if there is asynchronous behavior between ECG and pump equipment. In our study, we propose an alternative method of volume calculation, one that relies on an integrated form of the flow rate output data readily available from cfLVAD controllers. PATIENTS AND METHODS Patient selection We conducted a single-center prospective study at St. Vincent’s Hospital, Sydney, between January 2010 and December 2012. Willing participants with an implanted HeartWare cfLVAD (HVAD; HeartWare, Framingham, MA, USA) who were older than 15 years were studied. Patients were excluded if they required support with a right ventricular assist device (RVAD), inotropic therapy, or extracorporeal membrane oxygenation (ECMO) at time of study. The study was approved by the Human Research Ethics Committee of St. Vincent’s and Mater Health Services, Sydney. Study protocol Patients were requested to lie supinely for the duration of the study. A high-frequency (50 Hz) sampling software program (HeartWare CDAS) installed on a handheld ultramobile computer was used to continuously record HVAD pump parameter data (speed and flow rate) for offline determination of the indices. At rest, baseline measurements were taken, consisting of mean arterial pressure (MAP) calculation (ultrasonic Doppler probe Model 811-B, Parks Medical Electronics, Inc, Beaverton, OR, USA) and echocardiography (Acuson Cypress; Acuson, Mountain View, CA, USA) to determine ventricular dimensions. Following baseline measurements, pump speed was increased at intervals of 80 rpm every 2 min until 240 rpm above the baseline. This was slowly brought back down to baseline over a 2-min period. Pump speed was then reduced at intervals of 80 rpm every 2 min until 240 rpm below the baseline, after which it was returned to baseline, thereby ending the protocol. With each change in pump speed, MAP calculation and echocardiography were repeated. Protocol Artif Organs, Vol. ••, No. ••, 2014

termination was indicated with development of a ventricular arrhythmia. With speed increases, a left ventricular end-diastolic dimension (LVEDD) reduction of 20% or flow rate increase by 30% also warranted termination. With speed decreases, criteria included an increase in LVEDD by 10% or decrease in flow rate by 20%, as previously defined (13). Data analysis Microsoft Excel and Visual BASIC for Applications programming software were used to derive the IQ, RIQ, and K indices. IQ was determined by two different methods. The first was based on regression between dQ/dtmax and QP2P for all cardiac beats at each pump speed interval (all-data), while the second was based on regression between average dQ/dtmax and QP2P values at each pump speed interval (average-data). IQ was also determined at baseline pump speed only, as a regression between all values of dQ/dtmax and QP2P obtained at that pump speed. To determine volumes moved in the HVAD pump throughout the cardiac cycle, we integrated the flow rate curve obtained from the pump with the subsequent waveform representing the relative volume of blood moved (Fig. 1). Systolic onset, clinically defined as mitral valve closure following the commencement of LV contraction, coincided with the integrated waveform being at a local minimum (point A). This suggested the onset of systole, as volume increases following this local minimum would be due to LV contraction. Furthermore, point A occurred after the QRS complex and continued past the onset of the T wave, reflective of the delay between the heart’s electrical and mechanical events. Similarly, diastolic onset, clinically defined as aortic valve closure following the commencement of LV relaxation, coincided with the integrated waveform being at a local maximum (point D). As volume decreases following this local maximum would be due to LV relaxation, it appeared that this point coincided with the onset of diastole. These definitions of systolic and diastolic onset enabled calculation of the K index without an ECG trace. Statistical analysis The best method of calculating the IQ index (alldata versus average-data) was determined using linear regression and Bland–Altman analysis. We then assessed the validity of the IQ index by comparing values obtained with varying degrees of pump speed changes. Furthermore, linear regression and Bland–Altman analysis using the most accurate IQ result, as determined by the R2 coefficient, and RIQ

M F M M M M F F M M M

Gender

Mean 1.84 (SD 0.27)

2.00 1.65 1.90 2.35 2.12 1.91 1.62 1.40 1.96 1.56 1.77

BSA DCM DCM DCM IHD DCM IHD HCM DCM IHD HCM DCM

HF etiology

Median 2700

2800 2500 2700 2600 2800 2800 2600 2600 2700 2800 2800

Pump speed (rpm)

Mean 232 (SD 186)

126 61 99 78 540 369 146 636 97 612 148

Days post-op

W, A, C, B, AA, L W, A, C, B, L, A2RB, Di W, A, C, AA, ACEi, Di, S W, A, C, AA, L, A2RB, Amio W, A, B, L, Di W, A, C, AA, L, Di, S W, C, AA, ACEi, Di, Amio, GTN W, A, C, B W, A, C, B, AA, L, Di, Amio, GTN, HCT W, A, C, B, Di W, A, AA, ACEi, Amio

Medications

* Indicates underlying atrial fibrillation rhythm. † Indicates biventricular pacing. BSA, body surface area; HF, heart failure; HCM, hypertrophic cardiomyopathy; DCM, dilated cardiomyopathy; IHD, ischemic heart disease; W, warfarin; A, aspirin; C, clopidogrel; B, beta-blockers; ACEi, angiotensin-converting enzyme inhibitors; A2RB, angiotensin II receptor blockers; Di, digoxin; Amio, amiodarone; GTN, glyceryl trinitrate; AA, aldosterone antagonist; L, loop diuretic; S, statins; HCT, hydrochlorothiazide.

Average

51 22 42 48 62 57 53 40 61 51 54

Mean 49 (SD 11)

Eleven patients (eight male, three female; mean age 49 ± 11 years) were studied (Table 1). Three patients had underlying atrial fibrillation, one of whom was being paced by a biventricular automatic implantable cardioverter–defibrillator. All

1 2 3 4 5 6 7 8 9*† 10* 11*

RESULTS

Age (years)

were performed to determine interchangeability between the two indices. Similar analyses were also performed for the K index for comparability with IQ and RIQ. All continuous variables are presented as the mean ± standard deviation. Statistical significance was set at a P value < 0.05, with statistical analysis performed in Microsoft Excel.

Patient

FIG. 1. The cardiac cycle under continuous-flow LVAD support, adapted from Carl Wiggers’ diagram of the cardiac cycle (14). It depicts the LVAD pump flow rate and the relative volume of blood moving through the LVAD.

TABLE 1. Patient demographics (n = 11)

THOUGHTS AND PROGRESS 3

Artif Organs, Vol. ••, No. ••, 2014

4

THOUGHTS AND PROGRESS TABLE 2. Comparisons between IQ at different pump speeds, K, and RIQ (n = 11) IQ index (baseline speed)

Patient 1 2 3 4 5 6 7 8 9*† 10* 11* Excluding AF patients Mean SD Including AF patients Mean SD

IQ index (−240 to +240 rpm)

K index (baseline speed)

RIQ index (baseline speed)

Value

R2

Value

R2

Value

SD

Value

SD

7.4 1.47 2.38 4.93 8.88 6.06 5.27 7.28 8.68 2.51 7.47

0.31 0.03 0.01 0.20 0.34 0.25 0.74 0.37 0.09 0.13 0.55

9.72 10.84 9.44 8.15 7.08 7.22 5.81 6.30 13.53 5.25 −10.03

0.87 0.96 0.94 0.81 0.67 0.99 0.99 0.96 0.48 0.74 0.38

1.02 0.94 1.57 1.84 1.45 1.51 1.58 0.96 1.71 1.76 1.61

0.28 0.26 0.25 0.31 0.54 0.36 0.39 0.08 0.19 0.24 0.14

6.07 7.53 8.74 7.41 6.03 6.24 7.94 8.38 8.01 6.70 7.20

0.47 1.05 0.86 0.57 1.22 0.47 1.04 0.97 0.83 1.73 0.54

5.46 2.53

0.28 0.23

8.07 1.78

0.90 0.11

1.36 0.34

0.31 0.13

7.29 1.07

0.83 0.29

5.67 2.60

0.27 0.22

6.61 3.12

0.52 0.27

1.45 0.33

0.28 —

7.30 0.94

0.89 —

* Indicates underlying atrial fibrillation rhythm. † Indicates biventricular pacing. AF, atrial fibrillation.

patients were able to tolerate the full range of pump speed variation, with no requirement for protocol termination. There was a strong positive correlation between values of IQ derived from all data and average data (R2 = 0.96, P < 0.001). However, a Bland–Altman analysis of this data showed the presence of systematic error (R2 = 0.68, P = 0.011). This suggested that the IQ based on average data yielded higher values and that the discrepancy between the two methods was wider at larger values. All-data results may have been skewed due to an uneven number of cardiac beats sampled at different pump speeds. Therefore, to standardize results, further analysis only utilized the average-data method for IQ determination. For patients with an underlying atrial fibrillation, results for the IQ index with pump speed variation (±240 rpm from baseline) were inaccurate (IQ range: −10.03 to 13.53, R2 range: 0.38 to 0.74; Table 2). Comparatively, the RIQ index appeared to yield results for the atrial fibrillation group that were similar to the results of those without an arrhythmia.

With regard to the remaining eight patients (those with normal cardiac rhythm), IQ at baseline pump speed was not accurate, with R2 values from 0.01 to 0.74 (mean 0.28 ± 0.23) (Table 2). A pump speed range of ±240 rpm from baseline provided the most accurate results for IQ determination (mean R2 = 0.90 ± 0.11, P < 0.001 to 0.024; Table 2). With lower pump speed ranges, mean R2 values were not as reliable (±160 rpm: mean R2 = 0.85 ± 0.15; ±80 rpm: mean R2 = 0.75 ± 0.33). With ±240 rpm results yielding the most accurate IQ results, these values were compared to RIQ and K. No correlation was found between RIQ and IQ (R2 < 0.01, P = 0.91), RIQ and K (R2 < 0.01, P = 0.92), or IQ and K (R2 = 0.11, P = 0.42) when considering the eight patients with normal cardiac rhythm. Looking more closely at the comparisons between IQ and RIQ, we see that of the eight patients in sinus rhythm, there were four patients with discrepancies between IQ and RIQ (Fig. 2). Only one of these had an aortic valve that opened at lower pump speeds, which characteristically results in overestimation of

FIG. 2. Regression and Bland–Altman plots comparing the IQ and RIQ indices of patients in sinus rhythm. Solid black circles represent patients with a discrepancy between IQ and RIQ results.

Artif Organs, Vol. ••, No. ••, 2014

THOUGHTS AND PROGRESS

5

FIG. 3. Impact of flow rate and LVEDD on the discrepancy between RIQ and IQ. Solid black circles represent patients with a greater discrepancy between IQ and RIQ results.

IQ (10). There was a tendency for high flow rates and large resting ventricular volumes to result in underestimation of RIQ relative to IQ (Fig. 3). Comparatively, low flow rates and small resting ventricular volumes resulted in overestimation of RIQ relative to IQ. DISCUSSION Given suggestions that use of long-term cfLVADs may result in myocardial recovery, it is imperative to be able to accurately monitor LV chamber function. Given that up to half of all explanted patients require reimplantation or cardiac transplantation due to subsequent decompensation (4,15–20), current measures based on echocardiography are clearly ineffective in assessing load-independent measures of contractility. While various measures such as IQ and K index have been developed, there has been little clinical uptake. Using the RIQ index instead of IQ may improve utilization due to its ease of calculation without requirement for pump speed alteration. The potential advantage of such an index is that it enables an objective marker of cardiac function to be determined during clinic visits. Interestingly, we found that the IQ index demonstrated significant intrapatient variability when derived from different pump speed ranges. The greater accuracy demonstrated with larger pump speed ranges may be due to a greater number of overall data points being utilized for analysis. However, as there are no current guidelines in relation to the degree of pump speed variation required, it appears that RIQ may be a more clinically relevant index. As demonstrated through this study, a further advantage of RIQ over IQ is that it can be derived for a series of pulses (30–60 s), even in irregular rhythm, by averaging dQ/dtmax and QP2P. In contrast, the IQ regression slope is unreliable in the setting of atrial fibrillation or irregular ventricular rhythm. When considering LVEDD as a variable, our regression analysis revealed a statistically significant discrepancy between RIQ and IQ. However, as the number of

data points used for this analysis was limited due to the small sample size, larger studies are required to assess the impact of LVEDD on these indices. Correlation between RIQ and K was poor, indicating that other factors may be affecting the validity of one or both indices. A recent study suggested that the sampling frequency with which the HVAD flow rate is estimated (50 Hz) results in under-estimation of dQ/dtmax and thus RIQ (21). As a result, current bandwidth may be insufficient in detecting ventricles with greater levels of functioning, thus causing skewed results, potentially masking any correlation between K and RIQ. Thus, further assessments of correlation between K and RIQ are required. We suggest intrapatient comparisons be made longitudinally or with active alteration in contractility state. We were able to calculate the K index without an ECG trace by using the timing of minima and maxima points on the relative-volume waveform to define systolic and diastolic onset, respectively. As our study was limited by the small sample size, future studies should assess larger cohorts. Furthermore, as we have proposed alternatives in the form of the RIQ index and in the method of determining the K index, both will need to undergo validation studies to determine the effect of variation in clinically relevant variables such as pump speed, heart rate, loading conditions, and inotropy. CONCLUSION As RIQ is calculated without the need for any intervention, it has the potential to be a clinically relevant index of myocardial function. The K index could be calculated without an ECG trace. Both indices require further validation before they can be applied in the clinical setting to monitor myocardial function in patients with continuous flow left ventricular assist devices. Conflict of Interest: CSH has received research funding and travel support from HeartWare, Inc. unrelated to the current project. None of the remaining authors have relevant conflicts of interest. Artif Organs, Vol. ••, No. ••, 2014

6

THOUGHTS AND PROGRESS REFERENCES

1. Kirklin JK, Naftel DC, Kormos RL, et al. The Fourth INTERMACS Annual Report: 4000 implants and counting. J Heart Lung Transplant 2012;31:117–26. 2. Lahpor J. State of the art: implantable ventricular assist devices. Curr Opin Organ Transplant 2009;14:554–9. 3. Xydas S, Rosen RS, Ng C, et al. Mechanical unloading leads to echocardiographic, electrocardiographic, neurohormonal, and histologic recovery. J Heart Lung Transplant 2006;25:7–15. 4. Maybaum S, Mancini D, Xydas S, et al. Cardiac improvement during mechanical circulatory support: a prospective multicenter study of the LVAD working group. Circulation 2007; 115:2497–505. 5. Topilsky Y, Oh JK, Atchison FW, et al. Echocardiographic findings in stable outpatients with properly functioning HeartMate II left ventricular assist devices. J Am Soc Echocardiogr 2011;24:157–69. 6. Krabatsch T, Schweiger M, Dandel M, et al. Is bridge to recovery more likely with pulsatile left ventricular assist devices than with nonpulsatile-flow systems? Ann Thorac Surg 2011; 91:1335–40. 7. Starling RC, Naka Y, Boyle AJ, et al. Results of the post-US Food and Drug Administration-approval study with a continuous flow left ventricular assist device as a bridge to heart transplantation: a prospective study using the INTERMACS (Interagency Registry for Mechanically Assisted Circulatory Support). J Am Coll Cardiol 2011;57:1890–8. 8. Chung N, Nishimura R, Holmes D Jr, Tajik A. Measurement of left ventricular dp/dt by simultaneous Doppler echocardiography and cardiac catheterization. J Am Soc Echocardiogr 1992;5:147–52. 9. Rhodes J, Udelson JE, Marx GR, et al. A new noninvasive method for the estimation of peak dP/dt. Circulation 1993; 88:2693–9. 10. Naiyanetr P, Moscato F, Vollkron M, Zimpfer D, Wieselthaler G, Schima H. Continuous assessment of cardiac function during rotary blood pump support: a contractility index derived from pump flow. J Heart Lung Transplant 2010;29:37–44. 11. Hayward CS, Kalnins WV, Rogers P, Feneley MP, Macdonald PS, Kelly RP. Left ventricular chamber function during

Artif Organs, Vol. ••, No. ••, 2014

12.

13.

14. 15. 16.

17.

18.

19.

20. 21.

inhaled nitric oxide in patients with dilated cardiomyopathy. J Cardiovasc Pharmacol 1999;34:749–54. Ferreira AL, Wang Y, Gorcsan J, Antaki JF. Assessment of cardiac function during mechanical circulatory support: the quest for a suitable clinical index. Presented at the Annual International Conference of the IEEE Medicine Biology Society, Boston, 2011. Hayward CS, Salamonsen R, Keogh AM, et al. Effect of alteration in pump speed on pump output and left ventricular filling with continuous-flow left ventricular assist device. ASAIO J 2011;57:495–500. Wiggers CJ. The Henry Jackson Memorial Lecture: dynamics of ventricular contraction under abnormal conditions. Circulation 1952;5:321–48. Birks EJ, Tansley PD, Hardy J, et al. Left ventricular assist device and drug therapy for the reversal of heart failure. NEJM 2006;355:1873–84. Dandel M, Weng Y, Siniawski H, et al. Prediction of cardiac stability after weaning from left ventricular assist devices in patients with idiopathic dilated cardiomyopathy. Circulation 2008;118(14S1):S94–105. Dandel M, Weng Y, Siniawski H, Potapov E, Lehmkuhl HB, Hetzer R. Long-term results in patients with idiopathic dilated cardiomyopathy after weaning from left ventricular assist devices. Circulation 2005;112(S9):37–45. Farrar DJ, Holman WR, McBride LR, et al. Long-term follow-up of Thoratec ventricular assist device bridge-torecovery patients successfully removed from support after recovery of ventricular function. J Heart Lung Transplant 2002;21:516–21. Mancini DM, Beniaminovitz A, Levin H, et al. Low incidence of myocardial recovery after left ventricular assist device implantation in patients with chronic heart failure. Circulation 1998;98:2383–9. Simon MA, Kormos RL, Murali S, et al. Myocardial recovery using ventricular assist devices. Circulation 2005;112(Suppl. 9):132–6. Granegger M, Moscato F, Casas F, Wieselthaler G, Schima H. Development of a pump flow estimator for rotary blood pumps to enhance monitoring of ventricular function. Artif Organs 2012;36:691–9.