To the Editor,

Milrinone is the first-line prophylaxis and treatment of low cardiac output syndrome (LCOS) after pediatric cardiac surgery.1,2 Identifying therapeutic levels (100–300 ng·mL-1) is challenging as pharmacokinetic models predict levels poorly,3 supratherapeutic levels are indistinguishable from LCOS, and subtherapeutic levels are ineffective. We conducted a pilot randomized controlled trial to assess the feasibility of real-time milrinone monitoring to improve the precision of therapy.

Patients between 36 post-conceptual weeks and 18 yr, undergoing cardiopulmonary bypass (CPB), with arterial and central venous access, anticipated to receive milrinone for ≥ 24 hr and who provided informed consent were randomized to either scheduled milrinone levels in a 1:1 ratio in three age-defined strata: (< 2, 2–10, and > 10 yr). Milrinone levels were measured using our in-house assay4 until the earlier of milrinone discontinuation or seven days in patients > 8 kg (with six-hourly levels) or five days in patients < 8 kg (with eight hourly levels). Patients < 2.0 kg or receiving extracorporeal membrane oxygenation (ECMO) were ineligible.

The primary outcome was LCOS diagnosed by clinical and biochemical values (Table) at 48 hr after enrollment.5 Feasibility was assessed using enrollment, dropouts, measured levels, dosage adjustments, and study completion (registered at ClinicalTrials.gov: NCT01841177; registered on 23 April 2013). We used a generalized estimating equation technique that accounted for clustering within patient and clinical variables to evaluate the relationship between milrinone levels and LCOS.

Of 61 patients approached, 54 (92%) consented to participate, seven (13%) withdrew (two before, five after randomization), and five were withdrawn (three ECMO, two protocol violations). The 47 who completed the study (23 intervention, 24 usual care) had a mean age of six months. Group demographics, surgical procedures, and intensive care unit (ICU) duration were similar (data not shown). The mean (standard deviation [SD]) milrinone bolus dose was 88 (22) µg·kg-1 for usual care patients and 90 (18) µg·kg-1 for intervention patients (P = 0.84). Infusions were started immediately after the bolus dose. The mean (SD) infusion rate was 0.63 (0.13) µg·kg-1·min-1 for the usual care group and 0.58 (0.16) µg·kg-1·min-1 for intervention groups (P = 0.24). During milrinone treatment, epinephrine was started in two patients and increased in six patients, one patient had norepinephrine, and one patient had vasopressin increased. Vasopressin was commenced as a new drug in the 12–18 hr interval, it was not increased. The same was the case for the norepinephrine.

Among 117 samples from the 23 patients in the intervention group, the first was taken a mean (SD) 118 (44) min after the bolus was given. Levels were available 3–17 hr after sampling. In the first 12 hr of ICU admission, there were 15 supratherapeutic (> 300 ng·mL-1) levels and six subtherapeutic levels (< 100 ng·mL-1). Three dosage adjustments were made and in one patient with a high level (493 ng·mL-1), milrinone was ceased. In another, a level of 79 ng·mL-1 led to a dose increase with subsequent level of 133 ng·mL-1. The third patient had the milrinone decreased and levels taken but there was no laboratory report available. Throughout milrinone treatment, eight (35%) patients maintained therapeutic levels, one remained subtherapeutic, and three remained supratherapeutic.

At 48 hr, LCOS was present in 13% of patients. The incidence of LCOS was 0% (0/15) in patients with normal milrinone levels, 17% (2/12) in patients with levels < 100 ng·mL-1, and 20% (4/20) in patients with levels > 300 ng·mL-1. Low cardiac output syndrome was present in one patient (4%) receiving levels and in five patients (21%) in the control group. After adjustment for loading dose, infusion duration, age, CPB, single ventricle physiology, milrinone plasma levels, creatinine levels, and low urine output, milrinone monitoring was associated with a significant reduction in the incidence of LCOS (coefficient, -2.06; 95% CI, -3.90 to -0.22; adjusted P = 0.03). Seven patients had cardiac rhythm disturbances requiring treatment (three junctional dysrhythmias, two bradycardias, one complete heart block, and one junctional ectopic tachycardia).

In conclusion, this randomized pilot trial of therapeutic drug monitoring of milrinone after pediatric cardiac surgery showed feasibility of testing, a lag time to obtain results, and modest clinical responses to drug levels by clinicians. This analysis provides further evidence of the relationship between milrinone levels and clinically important outcomes and provides preliminary data suggesting a potential clinical benefit of therapeutic drug monitoring in the small sample tested. Larger randomized studies are needed.

Table Demographic and surgical characteristics
Full size table