American College of Cardiology

STEVENSON AND KORMOS, ET AL., MECHANICAL CARDIAC SUPPORT 2000
JACC Vol. 37, No. 1, January 2001:340-70

A. Indications for Device Support

The appropriate population for a trial of mechanical circulatory support is comprised of the patients whose current quality of life and prognosis are measurably worse than expected outcomes for the device being tested. The population should be defined as broadly as possible to maximize generalization of the results. Although the specific entry criteria will vary for each device and indication, there are general categories of patients who can be considered along a scale of disease severity (Table 3).

As the severity of disease increases, there is greater certainty regarding imminent death, and less certainty is required regarding the device performance and patient outcome after device implantation. In general, however, increasing disease severity also increases the risk of adverse outcomes attributable more to the patient than to the device. At lesser grades of severity, when death is not imminent, details regarding the expected function and quality of life with mechanical circulatory support become more critical. In one study, a majority of patients anticipating continued heart failure symptoms at rest expressed willingness to trade >50% of their remaining time, or take >50% risk of death, for a chance to return to more normal function.⁴²

1. Cardiogenic Shock
a. Critical low output state from exacerbation of chronic heart failure. Most of the current experience with mechanical support as bridging to transplantation derives from the population of patients with chronic HF that decompensates to a critically low output state threatening tissue perfusion and organ viability. In the absence of reversible factors, this state usually leads to death before hospital discharge. When transplantation, and thus bridging to transplantation, is not an option or when current bridging techniques are not applicable, this population could be considered for trials of newer support systems. Early identification of such patients would be desirable for these trials, but it is confounded by difficulty in distinguishing reversibility of organ system dysfunction and by the rapidity of clinical progression. One study evaluated the ability of the Acute Physiology and Chronic Health Evaluation II (APACHE II) scoring system to determine optimum timing of VAD implantation in patients with lung rales, S3, peripheral edema, ejection fraction <30%, systolic blood pressure <80 mm Hg, progressive prerenal azotemia, altered level of consciousness, gastrointestinal ischemia or congestion or persistent but reversible pulmonary hypertension.⁴³ By the end of the follow-up period, the VAD patients had survived longer (560 vs. 256 days). Kaplan-Meier analysis of non-VAD patients at low (≤10), medium (11–20) and high (>20) baseline APACHE II scores revealed a decreasing survival with increasing APACHE II scores. Similar outcomes were seen in VAD-treated patients. Patients with low APACHE II scores had similar outcome regardless of whether or not they received VAD support. However, when VAD and non-VAD patients with medium APACHE II scores were compared, VAD-treated patients had better survival, which was confirmed in a model after controlling for baseline APACHE II scores. Although this study concluded that the severity of illness measured by APACHE II might be used to time insertion of devices for bridging to transplant, it might also be used to identify patients for urgent destination therapy. However, use of the APACHE II score to predict short-term mortality in patients with primary cardiovascular disease is limited, and it is complicated by variances in interpretation of the scoring system and errors in data capture.⁴⁴ The use of a modified APACHE II scoring system may improve the accuracy and reproducibility of these methods.⁴⁵ Extensive prospective evaluations of the APACHE II system (or a modification) are needed to further define the role of this method of risk stratification of potential candidates for mechanical support.

The frequency of CS complicating HF in transplant candidates is difficult to estimate from the 15% of recipients of “bridges” to transplantation, as the increased recognition of the benefits of mechanical support have broadened the application to patients with impending or anticipated circulatory failure. In addition, this population also includes patients bridged for more common causes of CS, such as MI and post-cardiotomy failure.

b. Cardiogenic shock after acute myocardial infarction. It is estimated that 1.1 million patients suffer an acute MI in the U.S. each year. Of these, approximately one third die prior to presentation.⁴⁶ In the large multicenter Global Utilization of Streptokinase and Tissue-Plasminogen Activator for Occluded Coronary Arteries (GUSTO) trial, CS occurred in 7.2% of patients, but it accounted for 58% of all deaths in the entire trial.⁴⁷ The estimated yearly incidence in the U.S. is 50,000 in the hospital with post-infarction CS. In the SHOCK registry, in-hospital mortality was approximately 60% in patients with post-MI CS.⁴⁸ Of the patients developing shock, it was present initially in 10.6% and developed after admission in the remaining 89.4%, usually within 48 h.⁴⁹ In one sub-study of GUSTO a prognostic algorithm predicted with high accuracy the 30-day mortality in patients with CS complicating an acute MI. Increased age was the strongest demographic variable predicting 30-day mortality, and shock at presentation had better outcome than shock presenting later. Clinical predictors focused on findings of peripheral hypoperfusion such as an altered sensorium, cold and clammy skin and oliguria. Significant hemodynamic predictors were a cardiac output <1.5 L/min or a pulmonary arterial wedge pressure >20 mm Hg. A serious limitation of this prognostic algorithm is the lack of consideration of revascularization, found in another GUSTO substudy to reduce the 30-day mortality rate and in the SHOCK trial to reduce six-month mortality.³⁶ Based on these data, a patient with CS after MI, especially if not a candidate for revascularization, could be a candidate for long-term mechanical support.

c. Post-cardiotomy shock. Post-cardiotomy shock is described in approximately 1.5% of the 400,000 patients undergoing cardiac operations each year in the U.S. As discussed above, survival to discharge is in the range of 20% to 40%.^1–6 In the minority of patients who proceeded through bridging devices to transplantation, the overall survival rate to discharge was 40% to 60%. Patients who are not candidates for transplantation could be considered for trials of permanent mechanical support, but it should be recognized that the factors rendering them ineligible for transplantation would also affect outcome on devices. The post-surgical state may also predispose to worse outcome because the results of mechanical bridging to transplant have been slightly less favorable in this population than in primary bridging experiences.

2. Heart failure dependent on intravenous inotropic support.
The population of patients requiring intravenous inotropic support is increasingly being considered as a potential candidate group for newer heart failure therapies, particularly those that carry significant risk. This population definition is less precise, however, than others based on immediately measurable parameters. Many patients hospitalized for heart failure exacerbations receive brief courses of intravenous inotropic therapy to facilitate diuresis or redesign an effective oral regimen, following which the inotropic therapy is discontinued. Persistent efforts to achieve fluid balance, the substitution or combination of different vasodilators to avoid symptomatic hypotension and severe renal dysfunction, and enrollment in heart failure management programs frequently allow patients previously on intravenous infusions to maintain a reasonable quality of life on oral regimens.50

Specific criteria for determining the ongoing need for intravenous inotropic therapy have not been established, despite numerous reports of chronic and intermittent intravenous inotropic therapy for ambulatory patients with heart failure. The classification of disease severity is ambiguous because patients on inotropic infusions may initially be reclassified to the clinical level of class III symptoms, while deterioration after discontinuation may take over 24 h to become apparent. Definitions of “failed weaning attempts” have been proposed,¹⁴ but identification of “symptomatic hypotension” and “worsening renal dysfunction” remains subjective. Despite the lack of uniform criteria for intravenous inotropic support, the prognosis for patients receiving either chronic or intermittent inotropic infusions outside the hospital is remarkably consistent. This may reflect a greater homogeneity of the population than recognized and/or a dominant adverse effect of the infusions themselves. The mortality reported in representative series generally ranges between 30% and 50% by six months.

Among patients listed for transplantation as Status II in the multicenter pre-transplant database, intravenous inotropic infusions were being administered at the time of listing in approximately 10% of the patients, of whom over 90% had died or deteriorated to Status I by the end of one year.⁵¹ In the pilot trial before REMATCH, 80% of the patients were on inotropic infusions at the time of randomization, and 53% of the patients in the medical arm were receiving inotropic infusions after hospital discharge.52 Mortality in the medical treatment arm of the pilot experience for the REMATCH trial was 30% at three months, consistent with the reported experiences on inotropic therapy. There are not yet sufficient data regarding quality of life to compare chronic intravenous inotropic therapy, which requires maintenance of an indwelling catheter and infusions, with LVADs, which require other equipment. Regardless of the difficulties of establishing true dependence on intravenous inotropic therapy, patients in this group would appear to be reasonable candidates for consideration of mechanical support devices, with which intermediate outcomes are expected to be comparable or better.

3. Outpatients with symptomatic heart failure—who is at intermediate risk?
It is relatively easy to identify critically ill patients not likely to survive until hospital discharge. For this population, patient-associated factors related to infection, renal failure, hepatic failure and malnutrition may play a greater role than device characteristics in post-operative survival. It is also relatively easy to find patients with good functional capacity and mild symptoms of heart failure who are likely to survive at least two years. This population adds relatively little patient-related risk to a new procedure but does not offer large opportunity for measurable improvement in outcome. Defining a population with intermediate risk and mortality remains a major challenge.

Most of the information regarding outcomes in heart failure derives from multicenter heart failure trials, dominated by mild-to-moderate heart failure and one-year mortality of <20%. Even the trial populations intended to include advanced class IIIb and class IV generally have actual one-year mortality of <30%, suggesting less severe disease. Various biochemical, structural and functional characteristics have been identified singly and in composite scores that predict mortality in these populations but are more uniformly abnormal among patients who would be considered for mechanical support. Among two series of patients with class III and class IV heart failure referred for transplantation—representing a total of almost 1,000 patients—the combined end point of death and urgent transplantation occurred in approximately 50% of the patients by two years.^53,54 A multivariate model including continuous variables of heart rate, LVEF, mean blood pressure, presence of intraventricular conduction delay, peak oxygen consumption and serum sodium identified 19% of the population with a one-year survival of 30% to 40% without urgent transplantation.⁵³ The other study indicated that patients referred with class IV symptoms could be divided approximately in half by serum sodium or LV dimension, with a <50% one-year survival for either a serum sodium of <134 mEq/L or an LV diastolic dimension >75 mm.⁵⁴ From a multicenter study of 967 patients listed as Status II for transplantation, class IV symptoms, higher creatinine, higher pulmonary capillary wedge pressure, diagnosis of ischemic heart disease and inotropic therapy at listing predicted worse outcomes.⁵¹ Even among patients awaiting transplantation, however, outcomes were relatively good for patients having a non-urgent status listing, with only 30% dying or deteriorating to an urgent status within the next year—most deaths occurring suddenly. The previous risk predictions will be compromised in future applications by broader use of implantable defibrillators.

Among patients out of the hospital on oral therapy, the major distinctions are made on a clinical basis. Many patients referred with class IV symptoms can regain stability—some immediately, some after a prolonged period of closely monitored adjustment of the medical regimen. From a practical standpoint, many patients exhibit a dynamic state that fluctuates over months, with exacerbations related to dietary indiscretion, seasonal viral infections and other exogenous factors. For ambulatory patients with heart failure, a large component of the decision to receive investigational therapy, either medical or surgical, is the degree to which the current clinical status is unacceptable. Patients able to regain and maintain freedom from congestion during close follow-up have a two-year survival of almost 80% despite an initial admission with class IV symptoms.⁵⁵ Patient preference for quality of life versus survival shows remarkable variation at every level of disease severity.⁴² For an individual patient with severe heart failure being evaluated for heart transplantation, certain pre-transplant risk factors may make transplantation a relatively high-risk option.⁵⁶ Although transplantation may be offered to such a patient despite this increased risk, an alternative mode of therapy may be mechanical assistance. In addition, an individual patient may decline transplantation because of social, psychological or religious reasons. Although transplantation may be indicated by medical standards in such patients, mechanical assistance may also offer improved quality of live and life span. It is not clear whether eligible patients refusing transplantation should be excluded from clinical studies of devices.

4. Uncontrollable ventricular arrhythmias.
Approximately half of the deaths from heart failure occur suddenly.⁵⁷ Unexpected cardiac death is usually due to tachyarrhythmias, but it may result from bradyarrhythmias or electromechanical dissociation in 10% of the general series, more often as the end stages of cardiac disease are reached. The implantable cardioverter-defibrillator is of limited efficacy in the therapy of rapidly recurrent or incessant arrhythmias because of limited battery life, high defibrillation thresholds in the advanced cardiomyopathic ventricle, and the downward spiral of hemodynamic instability. In addition, the quality of life can become unbearable under the shadow of frequent defibrillations without anesthesia.

Therapy with amiodarone or combination of other anti-arrhythmic agents may reduce the number of device discharges to a tolerable frequency. Recurrent tachyarrhythmias from an identifiable focus may be amenable to catheter ablation techniques. If symptomatic ventricular tachyarrhythmias are not controllable by all available means they may lead to the need for ventricular assist or the insertion of a total artificial heart.⁵⁸ Ventricular assist has been used successfully to provide hemodynamic support and allow effective pharmacologic arrhythmia suppression as a bridge to transplantation for refractory arrhythmias.^59–62 Although LV support has frequently been adequate for bridging patients with refractory tachyarrhythmias to transplantation, permanent support may be better provided by total support devices.

5. Cardiac allograft dysfunction and/or cardiac allograft vasculopathy.
The intermediate-term survival of patients with severe allograft CAD is very poor. Keogh and colleagues reported the mortality of patients with severe CAD in a study of 353 heart transplant recipients from Stanford University with a mean follow-up post transplant of 5.5 years.⁶³ In this study, the mean survival for patients dying from CAD was 15 months from the detection of any coronary disease (range 1 to 74 months). Survival was statistically worse in patients with >70% stenosis in a primary epicardial coronary artery. Survival at two years was 13%. Actuarial survival after the diagnosis of >70% stenosis in three primary epicardial vessels in this population was <50% at one year, half of these patients dying within the first six months. In a recent study from the Cardiac Transplant Research Database, CAD was defined as “severe” if the left main coronary artery or two or more primary vessels had stenoses of >70% or if there were isolated branch vessel stenoses >70% in all three coronary artery systems.⁶⁴ In 46 patients with severe CAD, actuarial freedom from death due to CAD (n = 17) or re-transplantation for CAD (n = 6) was only approximately 36% by two years after the diagnosis of coronary disease. Although the use of intracoronary stents may alter the natural history of patients who develop more proximal lesions amenable to this mode of therapy,⁶⁵ the majority of patients who develop CAD will have progressive disease with a similar rate of progression irrespective of when the disease is initially diagnosed.⁶⁶

Although re-transplantation is offered to patients with severe disease at some institutions, this practice is discouraged elsewhere in recognition of the limited supply of donor hearts and the lower survival after repeated transplantation.³⁹ Therefore, the population of patients with severe allograft CAD is one that may be considered for studies of biventricular support and the use of the total artificial heart. Current estimates suggest that there may currently be about 2,000 such patients. Table 4 outlines the potential advantages and disadvantages of this population as recipients for long-term destination therapy of mechanical circulatory support devices.

B. Evaluation for Exclusion Criteria

Patients who are acutely ill, including those without prior known cardiovascular disease suffering an acute MI with CS, will often develop some degree of non-cardiac end-organ and systemic dysfunction. Indices of organ dysfunction place the patient into a risk group in which support devices are warranted but in which they also increase the likelihood of post-operative complications. Although current experiences are not large enough for extensive multivariable analysis of risk factors for death and complications after mechanical support, experience with current implantable VADs has revealed some predictors of poor outcome during or post-device implantation.^53,67

In almost every major registry of VAD follow-up^68,69 and a single center review,⁷⁰ poor renal function or renal failure has been a significant predictor of death following LVAD implantation. Although renal insufficiency has customarily been defined by an elevated serum creatinine, oliguria in the face of adequate filling pressures may be more predictive in acute decompensation because the creatinine may not increase quickly, especially in a cachectic patient. In the combined Columbia Presbyterian Hospital and Cleveland Clinic experience, oliguria, defined as urine output <30 cc/h despite maximal medical therapy with diuretics, was the most important predictor of perioperative death, with a risk ratio of 3.9.⁶⁷ In this analysis, the second most important predictor was respiratory failure, defined as the need for intubation, with a relative risk of 3.0. The presence of a coagulopathy, defined as the inability to correct the prothrombin time to <16 s indicated significant liver dysfunction and carried a risk ratio of 2.4. Other pre-operative risk factors identified in this study included a central venous pressure ≥16 mm Hg (relative risk 3.1), the LVAD placement as reoperation (relative risk 1.8) and a leukocyte count >15,000/mm³ (relative risk 1.1).

If the placement of an LV support system alone (without an RV support system) is being considered, the condition of the RV should be assessed. In addition to the possibility of an elevated pulmonary arterial pressure secondary to HF,⁷¹ the reactive pulmonary hypertension associated with cardiopulmonary bypass and thromboxane A₂ release may predispose to significant RV failure early post-isolated LVAD placement. Also, the LVAD may suddenly markedly improve RV filling, leading to worsening RV failure. In one series, although the need for perioperative RV support was low, a low preoperative pulmonary arterial pressure (indicating decreased RV function) and a low RV stroke work index were significant risk factors for RVAD use.⁷² Others have shown that strong predictors of subsequent RV dysfunction after LVAD implantation were the pre-implant medical condition, presence of end-organ failure, pulmonary edema and coagulation abnormalities.⁷³ Factors to be considered in all patients are prior surgical history, prior radiation therapy, the general medical and nutritional condition of the patient and the patient’s social support structure.⁷⁴

C. Selection of Devices

Studies of new mechanical support devices should be targeted toward specific populations with high anticipated mortality with conventional therapy but a reasonable chance of surviving device placement and the perioperative period. Two broad categories of potential device recipients can be identified: 1) those with acute, potentially reversible conditions, and 2) those with chronic generally irreversible disease. In the first category, ideally devices should be inexpensive and easily inserted and removed. The second category of patients would benefit from devices with greater longevity, even if the device is more difficult to insert and is more expensive. Patients with intractable malignant arrhythmias and severe transplant vasculopathy will require the capability of biventricular support. Heart transplant candidates requiring mechanical bridging remain an excellent population in which to assess the feasibility of new potential long-term devices. Table 5 outlines the relative importance of various device characteristics as applied to potential recipient populations.

D. End Points for Outcomes

It is clear that appropriate end points to be incorporated into future clinical trial designs for mechanical circulatory support devices will need to vary according to the nature of the patient population to be included in each trial and the particular device being subjected to trial. For instance, simple all-cause mortality at six months might be an appropriate end point in a group of patients selected who had a >50% probability of death within six months, whereas more complex measures of “quality-adjusted survival” would be appropriate in a less sick population. All trials should be designed to incorporate measures of cost, cost-effectiveness and tracking of device malfunction and device failure. Quality of life will become an increasingly important end point to assess and should be compared with valid control groups of patients rather than relying on the patients’ own perceptions of their quality of life before and after placement of the device. It should be recognized that quality of life is a subjective and individual assessment and that the currently available tools to measure quality of life are imperfect and have not been well validated in advanced heart failure. It may be necessary to revise and validate tools for this specific patient population.

The following sections outline some generic suggestions for appropriate primary and secondary end points for patient groups of differing severity of illness. Each end point may have time-related “midpoints” to be assessed as well.

1. The end points for critical populations. Survival over the next three to six months is a major challenge for patients who are New York Heart Association (NYHA) functional class IV and compromised enough to depend on ongoing intravenous inotropic support to maintain secondary organ function and overall circulatory sufficiency. When trials of mechanical systems commence in these patient populations, it is suggested that end points of such trials include the components listed in Table 6.

2. Ambulatory heart failure on oral therapy. Patients with NYHA functional class IV symptoms who are candidates for chronic mechanical circulatory support and who are not recurrently hospitalized or dependent on intravenous inotropic agents are generally not “as sick” as patients dependent on intravenous inotropic support. These patients can experience discomfort during any physical activity and may have discomfort while at rest. The hypothesis is that a mechanical circulatory support device will provide such patients with an improved physiologic and functional quality of life and for a duration that extends well beyond the 30-day post-implant period. As discussed above, the probability of survival at a specific time is not well established.

The primary end point for clinical studies of devices intended for use in these patients would be all-cause mortality at a specified duration, such as six months, one or two years, although mortality due specifically to cardiac events should also be captured. End points of quality of life may assume more importance for these patients, for whom a sustained improvement in quality of life may be considered a significant benefit even if survival is equivalent.⁷⁶ Quality of life is a multidimensional construct measuring outcomes in the following domains: emotional state, general health perception, pain, social function and physical functioning. There is considerable debate about appropriate measurements for quality of life, but experience in assessing these aspects is rapidly being gained.^77,78

These domains can be analyzed and integrated in the context of patient preferences for health-related quality of life versus length of life. These measurements seek to capture the overall value or preference that a patient holds for a particular health outcome. Both the time trade-off instrument and the standard gamble questionnaire have been used to determine the relative value placed by an individual patient on the degree of perceived health versus remaining survival time or risk of death while pursuing better health.^42,76,79 They may have greater relevance to decision-making than abstract scores. Preference ratings can serve as the quality adjustment factors for calculating quality adjusted survival, measured in quality adjusted life years (QALY). Such measures are expressed as numeric values on a uniform scale (0 to 1). They are particularly useful for summarizing overall changes in health-related quality of life because they are expressed as a single score.

Morbidity parameters as listed in Table 6 should be secondary end points, but they will assume increasing significance and may become primary end points in trials of less sick patients for whom, if survival is equivalent and is associated with significantly less morbidity, significant benefit may be considered to have been demonstrated. The frequency of each event and the time to each event should be captured for reporting in the application for approval for marketing by the FDA. In addition, device (system) malfunctions and device (system) failures are adverse events that should be captured for purposes of facilitating design improvements. The location where each morbidity event occurs and where each device malfunction and device failure occurs should be documented to establish device (system) safety in its intended user environment (in-hospital vs. out-of-hospital).

Because the relationship between cost and benefit is a significant issue in the evaluation of these devices, all cost information associated with this therapy should be collected for comparison with costs incurred by patients who do not receive a device. This includes costs associated with hospitalizations, caregivers in and out of the home, travel and medications. Cost-effectiveness is an analytical technique that looks at the rate paid to obtain a measure of health. This is often expressed in dollars per life year saved. When quality of life is taken into consideration, this is expressed as dollar cost per QALY saved. This form of analysis provides the optimal means to allocate health care resources to maximize the health benefits achieved.

Some might argue that certain therapies that are shown to have a defined benefit would prove to be too expensive for society to bear. On the other hand, we recognize that in some cases society has been willing to expend significant resources for a limited benefit to the population as a whole. It is conceivable that, although the actual cost may be extremely expensive for mechanical circulatory support, this therapy may significantly improve quality of life and return large numbers of individuals to a productive role in society and thus ultimately be considered cost-effective. Analysis of cost-effectiveness during the current stage of device development may not adequately reflect the eventual value or beneficial impact of mechanical circulatory support therapies, but such assessment can be expected to become more favorable as experience with devices, quality of devices and scope of their use expand in the future.