Eligible patients were men and women with STEMI referred for primary percutaneous coronary intervention (PCI) within 6 h after chest symptom onset. STEMI was defined as chest pain lasting more than 15 min with at least one of the following criteria: ST elevation ≥1 mm in at least two contiguous leads or presumed or certain appearance of a Q wave in three contiguous leads. Only patients with a TIMI coronary flow score of 0 on the first angiogram were included. Exclusion criteria were fibrinolysis, known allergy to morphine hydrochloride, uncontrolled epilepsy, brain injury or intracranial hypertension, previous STEMI or coronary artery bypass grafting (CABG), resuscitated cardiac arrest, cardiogenic shock or mechanical ventilation at admission, mechanical complication, sustained ventricular arrhythmia, high-grade atrioventricular block, decompensated respiratory insufficiency, serious hepatocellular insufficiency, and contraindication to cardiac MRI or gadolinium injection.

The study was performed in accordance with the ethical principles stated in the Declaration of Helsinki. The protocol was approved by the institutional review board of Pitie-Salpetriere hospital and all patients gave written informed consent. All data were measured off-line by a physician blinded to group allocation.

Patients in both the morphine and placebo groups routinely received intravenous aspirin and a loading dose of P2Y12 inhibitors before PCI. Intravenous heparin was administered to maintain an activated clotting time > 250 s.

All coronary angiography and stent implantation procedures were performed using standard interventional techniques. Thromboaspiration, balloon predilation, stent type selection, and use of glycoprotein IIb/IIIa inhibitors were at the operators’ discretion.

Randomization (1:1) sequences were generated by the biostatistician using random number tables and stratified by center using random blocks of 4. Pharmacists prepared sequentially numbered identical syringes of morphine (1 mg in 3 mL of saline) or placebo (3 mL of saline) according to the randomization sequence. Interventional cardiologists were blinded to allocation group. The study treatment was administered immediately after the artery was reopened (TIMI flow 2 or 3).

MRI was performed between day 3 and day 5 on a clinical 1.5 T system (Magnetom Avanto; Siemens Healthcare; Erlangen; Germany) using a 6-channel phased-array cardiac coil to assess myocardial injury. Unenhanced cine Steady State Free Precession (SSFP) sequences were acquired in the short axis plane. Eight to 10 sections encompassing the left ventricle were acquired for each patient. T2W–short tau inversion recovery (T2W-STIR) sequences were acquired in the short axis section covering the whole left ventricle using the following parameters (voxel size 2 × 1.5 × 8 mm, flip angle 90/180°, effective echo time 47 ms, bandwidth 235 Hz/pixel). Fifteen minutes after injection of 0.2 mmol/Kg of gadolinium (Dotarem; Guerbet; Aulnay sous Bois; France), late gadolinium enhanced (LGE) images were acquired using a segmented three-dimensional IR Gradient Echo T1-weighted technique. Left ventricular (LV) end-diastolic and end-systolic volume indexes and ejection fraction were measured from SSFP images using dedicated software (SyngoVia; Siemens Healthcare). The extent of myocardial edema was quantified on the T2W images using semiautomatic detection with the full width at half-maximum approach on a dedicated software (cvi42; Circle Cardiovascular Imaging Inc., Calgary, Alberta, Canada). Same approach (full width at half-maximum method) was used to calculate infarct size. LV mass was calculated by countering inner and outer boundaries of the LV wall on LGE images. Infarct size and LV mass were expressed in grams (myocardial density: 1.05 g/ml). The relative proportion of myocardial infarction was calculated by dividing infarct size by total LV mass.

Transthoracic echocardiography was performed on days 1 and 6. Left ventricular ejection fraction (LVEF) was calculated from parasternal long axis and apical four-chamber views using the biplane Simpson’s method.

The primary endpoint was infarct size/LV mass ratio measured by MRI between days 3 and 5. Secondary endpoints included the ratio of infarct size over area at risk measured by MRI, the areas under the curves (AUCs) of serum creatine kinase and troponin T levels (measured every 4 h during the first 24 h then every 6 h during the next 48 h), LVEF measured by echocardiography on days 1 and 6, infarct size/LV mass ratio by MRI after 1 year, and clinical outcome after 3 months and 1 year.

A previous study reported a mean infarct size measured by MRI of 14% ± 6% of LV mass [10]. The estimated relative decrease in infarct size/LV mass after intracoronary morphine in preclinical studies was at least 33% [6]. To have 90% power for detecting a 33% decrease with alpha set at 5%, 36 patients were needed in each group. Assuming that 30% of patients would not be evaluable (e.g., due to a contraindication to MRI, consent withdrawal, or poor MRI acquisition), 91 patients were included in this study.

Continuous variables are described as mean ± standard deviation and categorical variables as numbers (%). For the primary outcome, patients were assessed according to randomized treatment group under the intent-to-treat (ITT) principle. Continuous variables were compared using the Student or Mann-Whitney test, depending on the normality of distributions as assessed by the Shapiro-Wilk test, and categorical variables using the chi-square test or Fisher’s exact test, as appropriate. Pearson or Spearman correlation coefficients were computed to assess relationships between selected continuous parameters (i.e. troponin T and creatine kinase AUCs against infarct size/left ventricular mass ratio), as appropriate. All p values were two-sided, and p values lower than 0.05 were considered statistically significant. All analyses were performed using Stata V14.1 (StataCorp, College Station, TX, USA).

We included 91 patients with STEMI, who were randomly assigned to intracoronary injection of either morphine or placebo at the time of reperfusion. One patient withdrew consent after study inclusion, leaving 90 patients in the intention-to-treat population (Fig. 1).

There was no difference in baseline characteristics and concomitant therapies between both groups. The distribution of the culprit arteries was well balanced between the two arms. Most patients (93.3%) received glycoprotein (GP) IIb/IIIa receptor inhibitors or bilavirudin (Table 1).

Intracoronary injection of the study medication was associated with asymptomatic hypotension in two patients, one in each group. At the end of the procedure, the groups had no significant differences in rates of slow flow, distal embolism, or intra-stent residual thrombosis. The proportions of patients with myocardial blush grade II (21.1% vs. 30.0% in morphine and placebo groups, respectively) or III (47.4% vs. 40.0%) were similar in the two groups (p = 0.96) (Table 2).

Myocardial injury was assessed by MRI (Fig. 1). Infarct size/LV mass ratio was not significantly improved by intracoronary morphine (27.2 ± 15.0% vs. 30.5 ± 10.6% with placebo, p = 0.28). Infarct size remained similar in the two groups when adjusted on the area at risk (Table 3). A per-protocol analysis led to the same conclusion as did the intention-to-treat analysis.

These results are supported by cardiac enzyme release. Morphine did not significantly reduce AUC_{0–72 h} values for troponin T (70,979 [36,726-100,070] A.U. vs. 61,443 [46,330-97,460] A.U. with placebo, p = 0.71) and creatine kinase (18,815 [11,829-28,389] A.U. vs. 18,721 [13,515-25,994] A.U. with placebo, p = 0.79). Neither were peak troponin T or creatine kinase values significantly different (Fig. 2). Infarct size assessed by MRI correlated strongly with enzyme release (troponin T: r = 0.73, p < 0.0001; creatine kinase: r = 0.68, p < 0.0001).

LVEF assessed by echocardiography was similar in the morphine and placebo groups on day 1 (49.7% ± 10.3% vs. 49.3% ± 9.3%, p = 0.84) and day 6 (48.5% ± 10.2% vs. 49.0% ± 8.5%, p = 0.86). The percentage of patients with LV systolic dysfunction defined as LVEF below 45% was not significantly lower with morphine (29.5% vs 26.7% with placebo, p = 0.76). Diastolic function assessed based on the E/E’ ratio was similar in the two groups (8.2 ± 2.5 vs 7.9 ± 3.5, p = 0.73). Similarly, MRI showed no significant differences in LV structural or functional parameters on day 3–5 (Table 3).

MRI performed after one year showed similar infarct size/LV mass ratio (21.4 ± 9.8% vs. 23.0 ± 8.0% in placebo, p = 0.51) and LVEF (52.2% ± 10.3% vs. 53.3% ± 11.4%, p = 0.71) values in the two groups. The frequency of major adverse cardiac events (MACEs) after 12 months was not lower in the morphine group: MACEs occurred in 7 (15.6%) patients in the morphine group and 8 (18.6%) patients in the placebo group. The most common MACE was unplanned admission for decompensated heart failure. Stent thrombosis occurred in one patient in each group. All MACEs were ascribed by the investigators to the underlying cardiovascular disease (Table 4).

Post-hoc analyses were performed to examine the consistency of results across subgroups and to generate hypotheses to explain the study results. All tests for interaction were non-significant for age, sex, symptom-to-balloon time, and hypercholesterolemia, with p values > 0.20. Diabetes was common in the study population and may have been a confounder. The test for interaction between diabetes and treatment effect found a p value of 0.17. An analysis in the subgroup without diabetes showed a trend toward an improvement in the primary endpoint in the morphine group (22.5% [14–33] vs. 27.5% [22–41], p = 0.053). In contrast, no effect was evidenced in the subgroup of diabetic patients (35% [29–48] vs. 31% [28–36], p = 0.36).