Effect of Tilarginine Acetate in Patients With Acute Myocardial Infarction and Cardiogenic Shock: Title and subTitle BreakThe TRIUMPH Randomized Controlled Trial FREE

Cardiogenic shock is the leading cause of death among hospitalized patients with acute myocardial infarction (MI), with mortality rates in excess of 50%.^{1 - 8} Early revascularization improves survival; however, early mortality rates remain high, particularly among patients with continued shock after revascularization.⁹ The understanding of the pathophysiology of cardiogenic shock complicating MI has recently evolved toward an appreciation of the role of systemic inflammation, including cytokine release and expression of inducible nitric oxide synthase (NOS).^{10 - 12} Excessive NOS results in high levels of nitric oxide that, in turn, lead to inappropriate systemic vasodilatation, progressive systemic and coronary hypoperfusion, and myocardial depression.^{10 - 12}

Inhibition of NOS is a theoretically appealing approach to treatment of this high-risk population. Early single-center studies with isoform-nonselective NOS inhibitors were promising and suggested a substantial beneficial effect on survival.^{12 - 14} The phase 2, dose-ranging trial SHOCK-2 (Should We Inhibit Nitric Oxide Synthase in Cardiogenic Shock 2) investigated the safety and tolerability of L-N^G-monomethylarginine (L-NMMA) (tilarginine acetate injection; ArgiNOx Pharmaceuticals, Redwood Shores, Calif) in this population.¹⁵ In SHOCK-2, tilarginine was given as a bolus (0.15 to 1.5 mg/kg) followed by 5-hour infusion (0.15 to 1.5 mg/kg per hour) and resulted in modest early changes in hemodynamic parameters.¹⁵ There was no effect on survival; however, SHOCK-2 was not powered to assess the effect of tilarginine on mortality.

The Tilarginine Acetate Injection in a Randomized International Study in Unstable MI Patients With Cardiogenic Shock (TRIUMPH) trial was designed to test the effect of NOS inhibition with tilarginine on mortality due to persistent cardiogenic shock complicating MI despite an open infarct artery.

TRIUMPH was a prospective, international, multicenter, randomized, double-blind, placebo-controlled trial that tested the hypothesis that tilarginine compared with placebo would reduce by 25% 30-day all-cause mortality in patients with MI complicated by cardiogenic shock despite successful revascularization of the infarct artery.

TRIUMPH was conducted at 130 centers in 8 countries in North America and Europe. Participants, or their legally authorized representatives, provided written informed consent. Institutional review board or ethics committee approval was obtained at all sites. Inclusion required all of the following: (1) MI, confirmed by ischemic symptoms for at least 30 minutes with elevated cardiac markers and/or ST-segment elevation or left bundle-branch block; (2) patency (<70% stenosis) of the infarct artery, either occurring spontaneously and confirmed at angiography or after percutaneous revascularization; (3) refractory cardiogenic shock of less than 24 hours’ duration, confirmed by peripheral signs of tissue hypoperfusion and systolic blood pressure less than 100 mm Hg despite vasopressor therapy (dopamine ≥7 μg/kg per minute or norepinephrine or epinephrine ≥0.15 μg/kg per minute) continuing longer than 1 hour after infarct artery patency; (4) clinical or hemodynamic evidence of elevated left ventricular filling pressures; and (5) left ventricular ejection fraction of less than 40%. Hemodynamics and requirement for vasopressor treatment were reconfirmed after randomization just prior to study drug administration; patients with resolving shock were excluded.

Major exclusion criteria included suspected or documented infection, other causes of shock (tachyarrhythmia or bradyarrhythmia, hypovolemia, hemorrhage, or anaphylaxis), shock due to acute mitral regurgitation or rupture of the ventricular septum or free wall, severe valvular heart disease, predominant right ventricular failure or severe right ventricular dysfunction of any cause, serum creatinine level greater than 3.0 mg/dL (>264 μmol/L) or end-stage renal disease requiring dialysis, adult respiratory distress syndrome, anoxic brain injury precluding survival, irreversible multisystem failure, recent thoracic or abdominal surgery, primary pulmonary hypertension, and the need for emergency coronary artery bypass graft surgery within 24 hours.

Participants had to be aged at least 18 years and both men and women were eligible. Self-reported race was collected to perform a subgroup analysis based on race for regulatory reporting.

Eligible patients were randomized 1:1 to receive either tilarginine, 1.0 mg/kg intravenous bolus followed by 1.0 mg/kg per hour of intravenous infusion for 5 hours, or matching placebo. This dose was chosen based on prior studies that suggested an association with improved survival.^{13 - 14} The protocol strongly recommended avoiding a decrease in vasopressor doses during study drug infusion. Eligible patients were randomized, at least 1 hour after infarct artery patency, via an interactive voice-response system. Randomization was blocked with block sizes of 2 or 4 and stratified by site and by patient age (<75 or ≥75 years).

Other than administration of study drug and the recommendation regarding vasopressor dosing, patients were managed at the discretion of the treating physician based on current standards of care as recommended by the American College of Cardiology, American Heart Association, European Society of Cardiology, and Canadian Cardiovascular Society guidelines.^{16 - 20}

The primary outcome was all-cause mortality at 30 days overall and stratified by age (<75 or ≥75 years) among patients who received any study medication. Additional secondary outcome measures included the number of participants with resolution of shock, duration of shock, duration of mechanical ventilation, duration of intra-aortic balloon pump support, New York Heart Association (NYHA) functional class at 30 days, and 6-month mortality. Resolution of shock was defined as discontinuation of vasopressors, other than low-dose dopamine, and cessation of intra-aortic balloon pump support for at least 24 hours.

The planned sample size of 658 treated patients had 90% power to detect a 25% relative reduction in mortality from a projected placebo mortality rate of 50% at an α level of .05.^{13 - 15} Interim efficacy and futility analyses were planned at 50% and 75% enrollment. A recommendation to stop the trial for efficacy was to be based on an O’Brien-Fleming-type boundary and Lan-DeMets α spending function. Futility analyses were performed to determine the conditional probability of reaching a statistically significant result at the final analysis. A conditional power of 20% or less at either the first or second interim analysis was to trigger a recommendation to stop the trial. The α for the final primary analysis was .044.

The primary prespecified analysis was based on a modified intent-to-treat cohort composed of all randomized patients who received any study medication. An intent-to-treat analysis was performed on all randomized patients as a sensitivity analysis. All other analyses were conducted on the intent-to-treat population.

Dichotomous outcomes are expressed as numbers and frequencies. Continuous variables are expressed as medians (interquartile range [IQR]). All discrete analyses were performed using the Cochran-Mantel-Haenszel or χ² test. All continuous analyses were performed using the general linear model (GLM) with the exception of duration of shock. Assumptions for using continuous variables in the GLM were evaluated; only duration of shock was in violation. Duration of shock was not normally distributed, and no transformation was found to make it such. It was therefore analyzed using the log-rank statistic, with time measured from randomization to the resolution of shock. For patients who died prior to resolution of shock, assigned duration of shock was the longest reported duration of shock plus 1 hour.

Mortality at 30 days was analyzed using logistic regression. Mortality at 6 months was analyzed using the log-rank statistic, with calculation of a hazard ratio and 95% confidence interval (CI) by Cox proportional hazards modeling. Patients with missing data were not included in relevant analyses. Logistic regression modeling was used to assess the relationship between change in systolic blood pressure and mortality. Changes in systolic blood pressure were reported as a linear spline with different slopes at less than or equal to 0 and greater than 0. Other than the primary analysis of 30-day mortality, all other analyses should be considered hypothesis-generating. We considered P<.01 as strong evidence of an association. All analyses were performed using SAS software version 8.0 (SAS Institute Inc, Cary, NC).

A prespecified futility analysis was performed by the data and safety monitoring board when data were available on approximately 50% of the planned sample. Based on the recommendation of the board, recruitment into the trial was discontinued following an assessment of less than 10% conditional power to meet the specified primary objective of the trial. When the trial was stopped, data collection and cleaning ceased, resulting in some missing data. All participating centers were closed by the trial sponsor. Academic study leadership subsequently attempted to ascertain 6-month vital status on all randomized patients. Despite the withdrawal of financial support by the sponsor and the initial closure of the protocol at many sites, we were able to obtain nearly complete 6-month follow-up (through February 2007).

Of 1611 patients with MI complicated by cardiogenic shock entered into the TRIUMPH screening database between January 2005 and August 2006, 398 met the study inclusion criteria and were enrolled (Figure 1). On average, 3.65 patients (0.25 patients per site per month) were enrolled at each of 130 centers in 8 countries.

Baseline characteristics of the population are shown in Table 1. All baseline characteristics were well balanced between the treatment groups. More than one quarter of the population was older than 75 years; the majority were male and of white race. More than half of the patients had hypertension and one third had diabetes; 84 (21%) had a history of heart failure and almost a third of those had advanced heart failure symptoms in the 6 weeks prior to enrollment. A quarter of the patients had baseline creatinine levels of 1.7 mg/dL (150 μmol/L) or higher.

The median supported blood pressure just prior to study drug administration was 88/52 mm Hg. Most patients were supported with a single vasopressor at the time of study drug administration (Table 2). The majority of patients presented with anterior, ST-segment elevation MI with left anterior descending infarct artery location. Percutaneous coronary intervention was performed in nearly all patients to achieve the requirement for less than 70% infarct artery stenosis before study entry.

A total of 15 (3.8%) randomized patients (5 [2.4%] tilarginine and 10 [5.3%] placebo, P = .14) did not receive study drug because their hemodynamics prior to study drug administration were no longer consistent with entry criteria. Study drug administration was interrupted in 5 (2.5%) patients assigned to tilarginine and 6 (3.3%) patients assigned to placebo (P = .62), and the mean duration of study drug use was similar (4.7 vs 4.4 hours; P = .06).

In-hospital procedures are shown in Table 3. Almost all patients were treated with intra-aortic balloon counterpulsation and mechanical ventilation, most before randomization, with no difference in use or duration between groups. Forty-seven (12%) patients received left ventricular assist device support after a median of 69 (IQR, 15-117) hours; however, only a minority of these ultimately underwent cardiac transplantation. Coronary artery bypass graft surgery was performed during hospitalization in 37 (9.3%) patients.

Concomitant medication use both between randomization and 24 hours after randomization and between 24 hours after randomization and 7 days after randomization is shown in Table 4. There was no statistically significant or clinically meaningful difference in the use of any nonstudy medication between groups. Most patients were treated with multiple antiplatelet and anticoagulant agents including aspirin, thienopyridines, glycoprotein IIb/IIIa inhibitors, and heparin. Most patients had heparin administered through 7 days. All patients received vasopressor therapy at randomization; in most cases, vasopressors were continued through at least 7 days. There was substantial use of early antiarrhythmic therapy, which increased further by day 7. A small number of patients received β-blockers and/or angiotensin-converting enzyme inhibitors or angiotensin II receptor blockers early in the course despite having hypotension that required vasopressor support. By day 7, roughly half of patients were treated with β-blockers despite the fact that a substantial percentage had continued administration of sympathomimetic amines. A total of 229 (69%) patients received insulin therapy during the study period.

Hemodynamic outcomes are shown in Table 5. Patients assigned to tilarginine had a greater increase in systolic blood pressure at 2 hours compared with patients assigned to placebo (12.0 mm Hg vs 7.0 mm Hg; P = .001). This effect tended to be more evident in patients aged 75 years or older than those younger than 75 years (interaction P = .02). Timing of resolution of shock was similar among groups; the duration of shock was shorter in patients assigned to tilarginine, but this was not statistically significant (P = .16) (Figure 2). Similar findings were observed when duration of shock was assessed only in survivors. There was no effect of tilarginine on renal function.

Major clinical outcomes at 30 days are shown in Table 6. There was no difference in the primary outcome of all-cause, 30-day mortality between patients who received tilarginine and those who received placebo (97/201 [48%] vs 76/180 [42%]; risk ratio, 1.14; 95% CI, 0.92-1.41; P = .24). These results were unchanged when the 15 patients who did not receive study drug were included in the analysis. Also, when we adjusted for baseline differences in the use of vasopressors (Table 2), the results were unchanged (odds ratio, 1.27; 95% CI, 0.84-1.94; P = .26)

Tilarginine had no effect on 30-day mortality in any prespecified subgroup including those based on age, sex, diabetes status, infarct artery location, left ventricular ejection fraction, or history of heart failure. When renal function was assessed as a continuous function, patients with higher creatinine levels tended to have worse outcomes with tilarginine (interaction P = .07) (Figure 3). The proximate cause of death was cardiac in 135 patients (78%) in both groups, with 67 (50%) of these due to pump failure. Tilarginine had no effect on the rate of 30-day myocardial reinfarction or 30-day New York Heart Association (NYHA) heart failure class. At 30 days, 64 (17%) surviving patients remained hospitalized. At 6 months, 58% of patients assigned to tilarginine and 59% of patients assigned to placebo had died (hazard ratio,  1.04; 95% CI, 0.79-1.36; P = .80) (Figure 4). These results were unchanged when the 15 patients who did not receive study drug were excluded from the analysis, and there was no interaction between treatment and age.

A post-hoc analysis revealed that changes in systolic blood pressure between baseline and 2 hours were predictive of 30-day mortality; different relationships were observed for decreases and increases in blood pressure. In patients who manifested a decrease in systolic blood pressure between baseline and 2 hours, larger decreases in systolic blood pressure were associated with higher 30-day mortality (hazard ratio,  1.07 [95% CI, 1.04-1.11], χ² = 17.3; P<.001) with no interaction by treatment (interaction P = .62). Increases in systolic blood pressure, however, were not associated with lower 30-day mortality (hazard ratio,  1.00 [95% CI, 0.98-1.01]; χ² = 0.66; P = .42).

There were a total of 274 serious adverse events reported (130 in the placebo group and 144 in the tilarginine group) (Table 7). No significant differences in safety were found.

In this international, multicenter, randomized trial, isoform-nonselective NOS inhibition with tilarginine (L-NMMA) did not reduce 30-day or 6-month mortality rates in patients with MI complicated by cardiogenic shock persisting after the infarct artery was patent, either in the overall group or when stratified by age. The observation of higher mortality rates among patients with baseline renal dysfunction treated with tilarginine may be due to chance but requires further investigation. Overall, tilarginine was well tolerated but had no effect on the resolution of cardiogenic shock, on reinfarction, or on renal function.

There was, however, a significant increase in blood pressure with tilarginine. This hemodynamic effect of NOS inhibition in patients with hypotension refractory to sympathomimetic amines supports the hypothesis that excess nitric oxide may play a role in the genesis and/or persistence of cardiogenic shock. However, the modest increase in systemic arterial pressure did not translate into improvement in outcome, even though higher blood pressure is well known to be associated with survival in individuals with cardiogenic shock.^{21 - 22} Perhaps isoform-nonselective NOS inhibition, with endothelial as well as inducible NOS inhibition, is not the optimal treatment to reverse the effects of excess inducible NOS–generated nitric oxide. The discordant findings for blood pressure response and survival serve as a reminder that there may be no adequate surrogate outcome or marker for mortality in cardiogenic shock complicating MI. This poses enormous challenges for the development of new therapies, particularly dose finding, in the treatment of this critical illness because each new therapy must be tested in randomized clinical trials with mortality as the end point. An adaptive, event-driven trial design might be best in this setting.²³ Such a design would require a commitment on the part of sponsors to provide funding for enrollment of an uncertain number of patients, which is likely to exceed the number needed based on traditional designs. This study did establish the feasibility of conducting an adequately powered, multicenter, international randomized trial of a pharmacologic agent in persistent cardiogenic shock with a mortality end point.

Patients enrolled in TRIUMPH were selected for particularly high risk, based on the persistence of shock despite vasopressor/inotropic support for at least 1 hour after successful percutaneous revascularization. This patient subset was similar to that studied in the phase 1, single-center trials of NOS inhibition in cardiogenic shock, and the dose and duration of NOS inhibition were the same as in these studies.^{13 - 14} However, as is often the case, these early experiences with NOS inhibition in cardiogenic shock were conducted at a single institution without rigorous randomized trial methods. It is unlikely that the use of L-nitroarginine-methylester (L-NAME) in one of these studies accounted for the different results, because similar beneficial effects were seen in the other study in which L-NMMA was used.^{13 - 14} There remains the possibility that a higher dose, longer duration, or alternative dosing regimen of tilarginine could have a more pronounced effect on hemodynamics and/or survival; however, this seems unlikely in the absence of any signal of benefit in the present study. Furthermore, the SHOCK-2 phase 2 study of a 10-fold dose range of L-NMMA for cardiogenic shock found no benefit with the highest dose of 1.5-mg/kg bolus and infusion.¹⁵ Prior studies of prolonged NOS inhibition in patients with septic shock showed hemodynamic toxicity and excess mortality at higher doses with increases in pulmonary vascular resistance that would be expected to be prohibitive in the treatment of cardiogenic shock.²⁴ In addition, animal models of cardiogenic shock have shown that higher-dose treatment with NOS inhibition is associated with an increase in afterload with no change in contractility.²⁵

It has been postulated that inhibition of the systemic inflammatory response, which may be triggered by a large MI via different pathways, would reduce the occurrence of shock. A phase 2 trial of the complement inhibitor pexelizumab suggested that baseline cytokine levels were predictive of cardiogenic shock onset and that complement inhibition might reduce the incidence of shock and death.^{26 - 27} This action was believed to involve, at least in part, a reduction in production of nitric oxide by iNOS. However, a large randomized trial of pexelizumab in patients with MI undergoing reperfusion showed no effect on mortality.²⁸ Unfortunately, the promise shown in small initial studies of new therapies is too often not borne out in appropriately powered multicenter randomized trials.^{29 - 30}

Recent reports have suggested that outcomes have improved over time with advances in percutaneous intervention and supportive care, and the mortality rates for persistent cardiogenic shock in our study are consistent with this trend.^{6 ,31 - 33} The death rate was 39% for those patients younger than 75 years. Importantly, the majority of survivors were in NYHA functional class I or II at 30 days. Furthermore, the mortality rate between 1 and 6 months was low, even in this very high-risk cohort. This is consistent with findings from the SHOCK (Should We Emergently Revascularize Occluded Coronaries for Cardiogenic Shock) trial and reinforces how essential and rewarding it is to treat these most critically ill patients early in their hospital course.⁹ Aggressive treatment in this cohort included percutaneous revascularization, treatment with multiple vasopressors at moderate to high doses, inotropic therapy, intra-aortic balloon counterpulsation, mechanical ventilation, anticoagulant and antiplatelet therapy, as well as insulin in many cases. The use of very early vasodilator therapy (angiotensin-converting enzyme inhibition, angiotensin II receptor blockade, nitrates) and β-blockade in a small percentage of patients is surprising. These treatments are the mainstay of secondary prevention following MI in these highest-risk patients but should be reserved for the period after shock resolution.

Tilarginine, at the dose and duration studied, had no effect on mortality in patients with MI complicated by refractory cardiogenic shock. The increase in blood pressure in response to NOS inhibition suggests that excess nitric oxide may play a role in the pathophysiology of cardiogenic shock. Additional innovations in therapy and improved health care delivery systems are needed to establish earlier reperfusion to prevent shock following MI and to reduce the high short-term mortality rate of patients who develop cardiogenic shock.