During the study period, 721 patients were admitted for drug intoxication (Figure 1). One hundred and ten patients had hemodynamic failure responding to conventional treatment and 17 patients had refractory shock or cardiac arrest. In our practice, patients with refractory cardiac arrest, defined as an absence of return to spontaneous circulation after continuous cardio-pulmonary resuscitation over at least 45 minutes or refractory shock, defined as shock not responding to optimal conventional treatment, were candidates for ECLS support 23. When the decision to implant ECLS was made by a senior intensivist, a senior cardiac surgeon and a perfusionist were immediately informed and ECLS performed.

According to French legislation at the time of the study and given the observational retrospective nature of the study, no ethical committee was requested and thus no informed consent was obtained from the patients. At the time of ECLS implantation the focus was specifically on hemodynamic (including electrocardiographic and echocardiographic), neurologic, respiratory, renal, liver, and hematologic data. Measured physiological variables were used to calculate the Simplified Acute Physiology Score (SAPS II) 24 and the Sequential Organ Failure Assessment (SOFA) score 25. The toxicological screening was recorded and drugs classified as with or without cardiotoxic effect or membrane stabilizing activity (MSA). In addition, the clinical course of each patient during hospitalization was recorded. Vascular, neurologic, hemorrhagic, renal, and perfusion system complications were documented. The neurologic outcome at hospital discharge was assessed according to the cerebral performance class (CPC) categories 26: CPC 1 = good cerebral performance, CPC 2 = moderate cerebral disability, CPC 3 = severe cerebral disability, CPC 4 = coma or vegetative state, and CPC 5 = brain death or death.

Device description, cannulation technique, management, and weaning from ECLS were previously reported in detail 22. Briefly, the hardware for cardiopulmonary circulation consisted of a Biomedicus portable system (Medtronic, Inc, Minneapolis, MN, USA) incorporating a centrifugal pump console and a water pump system. The closed ECLS circuit consisted of pre-connected polyvinyl chloride tubing (Medtronic, Inc, Minneapolis, MN, USA) including a constrained vortex pump chamber, a hollow-fiber membrane oxygenator with an integral heat exchanger (Maxima PRF, Medtronic, Inc, Minneapolis, MN, USA), and a flow probe. All components were heparin-coated (Carmeda Bioactive Surface-coating). The cannulae were Biomedicus (17 F to 25 F), according to the size of patients.

Once the decision to implant ECLS support was made, the circuit was quickly primed with normal saline. Heparin was administered to the patient at 50 UI/kg immediately before cannulation of the vessels. The activated clotting time (ACT) was kept between 150 and 200 seconds at full-flow assistance. Peripheral femorofemoral cannulation was surgically set up using a modified Seldinger technique. Because cannulation-related limb ischemia was a major problem when we began this technique, additional distal limb perfusion was inserted to avoid severe leg ischemia. The distal tip of the arterial cannula was positioned in the common iliac artery or distal abdominal aorta, whereas the tip of the venous cannula was set in the right atrium under echocardiographic guidance and confirmed by chest radiograph.

Pump flow was initially set at 2.5 L/m² and vasopressor (norepinephrine and epinephrine) was used to maintain a mean blood pressure of at least 60 mmHg. An inotropic support was used even if systemic perfusion was adequately performed by ECLS to maintain a pulsatile flow through the native heart. The aim was to decompress the left heart and minimize stasis, therefore reducing the risk of intracardiac thrombosis. If necessary, to accomplish mechanical decompression of the left heart, an atrial balloon septostomy was performed. A femoral vein approach was used in which a transeptal puncture, followed by blade septostomy, was performed under combined radioscopy and echocardiographic guidance. Sequential balloon inflations were carried out to achieve left heart decompression, which was confirmed by echocardiography. The ECLS was monitored by trained ICU personnel. A perfusionist was also available for occasional monitoring visits and emergencies. Echocardiography was used serially to assess progressive myocardial recovery and exclude intracardiac thrombosis or other abnormalities. All patients were mechanically ventilated with 5 to 6 mL/kg tidal volume and 8 to 10 cmH₂O positive end-expiratory pressure, and continuous venovenous hemofiltration was used to treat acute renal failure and regulate the intravascular volume and overall fluid balance, if necessary.

The decision to discontinue ECLS support was based on evidence of multiorgan failure, overwhelming sepsis, or severe neurological injury. Patients were weaned off ECLS if the left ventricular ejection fraction, assessed by echocardiography during a reduction of pump flow to 500 to 1000 mL/min, was stable (> 50%) without deterioration in hemodynamic status. During this period, anticoagulation was adapted to adequate values of ACT (250 to 300 seconds). If the patient's cardiovascular status remained stable, ECLS was withdrawn by cardiac surgeons.

Quantitative and qualitative data are expressed as mean (± standard deviation), or median (range) and percentage, respectively.

Seventeen patients (11 females, 6 males; mean age 39 ± 18 years) were treated with ECLS following drug intoxication. All patients, except two, had ingested cardiotoxic drugs, including 11 cases of drugs with MSA. The majority (12/17) of poisonings resulted from mixed poisonings involving a combination of cardiotoxic drugs, various psychotic drugs, and alcohol (Table 1).

The baseline characteristics of patients at the time of ECLS implantation are reported in Table 2. All patients, except two, were comatose. The patients' median SAPS II score was 69 (26 to 82) and median SOFA score was 13.5 (3 to 18). The ECLS setup was performed for 13 patients in the operating room, 3 in the intensive care unit, and 1 in the emergency department. Seven patients received ECLS during external cardiac massage for refractory cardiac arrest, which occurred in six cases at hospital admissions in the emergency room. Ten other patients also received ECLS, including two after the restoration of spontaneous circulation following a brief period of asystole, during refractory shock. The mean duration of external cardiac massage was 101 ± 55 minutes.

Before the initiation of ECLS support, a severe decrease in cardiac contractility was documented by echocardiography in 14 cases. All patients were mechanically ventilated and received vasopressor. Four patients needed temporary external transthoracic electrostimulation. Six patients required continuous venovenous hemofiltration or conventional dialysis for acute renal failure before or immediately after ECLS implantation. Before connection to ECLS, the median arterial pH was 7.37 (7.34 to 7.41), partial presure of arterial oxygen/fraction of inspired oxygen ratio was 239 (180 to 261), serum bicarbonate concentration was 19.0 mmol/L (17.7 to 20.6), plasma lactate concentration was 5.9 mmol/L (3.7 to 9.7), and serum creatinine concentration was 160 μmol/L (114 to 204).

ECLS feasibility, assessed with respect to time from admission to ECLS initiation, and the percentage of successful procedures (i.e. flow rate > 2.5 L/m² and mean blood pressure > 60 mmHg) is shown in Table 3. Time from hospital admission to initiation of ECLS was 6.4 ± 7.0 hours, and the time to ECLS implant was 58 ± 11 minutes. The mean ECLS flow rate was 3.45 ± 0.45 L/min. The average ECLS duration was 4.5 ± 2.4 days. In one patient (no. 17), an atrial balloon septostomy was performed to accomplish mechanical decompression of the left heart.

Significant cannulation-related injuries of femoral vessels were reported in 10 patients: six patients with limb ischemia requiring urgent revascularization in three cases, one femoral thrombus, one cava inferior thrombus, and two cases of severe bleeding at the site of cannulation requiring a surgical revision.

Fifteen patients were weaned off ECLS support and two patients withdrawn from support because of refractory multiorgan failure and cerebral death (Table 3). Thirteen patients survived and were discharged to hospital without significant cardiovascular or neurological sequelae (CPC 1 n = 9 and CPC 2 n = 4). Two patients died of septic shock and cerebral death during the hospital stay.

Clinical experience with emergency ECLS during poisoning leading to prolonged cardiac arrest or shock that does not respond to conventional treatment is limited 345678910111213141516171819. Poly-intoxication including cardiotoxic drugs with MSA, which is known to be associated with a high mortality rate 27, was involved in a majority of cases, as previously reported 28. All patients were in prolonged cardiac arrest or refractory shock according to the definitions proposed by Baud and colleagues 28.

We confirmed the feasibility of emergency ECLS in accordance with previous reports focusing on prolonged cardiac arrest regardless of the cause 1922. The physiologic objective was to provide temporal circulatory support to the vital organs and unload the failing heart as the injured myocardium attempts to recover. In previous cohort studies 1922, survival rates were clearly higher in the toxic cardiac arrest group, as compared with other causes of cardiac arrest (3 of 12 vs 0 of 5 19 and 4 of 6 vs 4 of 34 22, respectively). The high survival rate (76%) reported in our cohort was in accordance with the general survival rate from 58% (15 of 26) in case reports of poisoned patients who benefited from ECLS 34567891011121314161719. The 5 of 7 (71%) survival rate we reported among patients with cardiac arrest was in contrast with the dramatically low survival rate of 7% and 4.5%, respectively, reported in overdoses involving cardiac arrest 2930. To our knowledge, no studies have reported the survival rate of patients with drug-induced cardiovascular shock apparently refractory to conventional treatment, limiting comparisons with our cohort. However, experimental studies with control groups demonstrated that ECLS improved survival in animal models of severe cardiotoxic drug-induced shock 3132. These results suggest that ECLS could be considered as a good emergency resuscitative tool in this setting.

Severe cannulation-related limb ischemia was the major problem when we first started the technique. Therefore, to accomplish a perfusion of the distal limb, surgeries performed an additional arterial shunt with a small 8 F catheter between the side port of arterial cannula and a point located some centimeters distally in the superficial femoral artery shunt. After this supplementary shunt was added, only distal embolic ischemia were reported. In contrast to Mégarbane and colleagues 19, the rate of cannulation-related complications, limb ischemia, and major bleeding was relatively high despite the modified Seldinger technique and additional distal limb perfusion. These differences could be explained by a higher ECLS duration in our study. However, our results were in accordance with reports of significant morbidity associated with emergency ECLS 33. In addition, no death was induced by cannulation-related complications, and all survivors were discharged without significant cardiovascular or neurological sequelae.

Firstly, because ECLS indication for drug overdose is rare, the sample size is small. Secondly, the uncontrolled retrospective observational design does not permit clarification of the role of ECLS therapy in drug-induced cardiac failure in a therapeutic algorithm. However, because usefulness of ECLS in this setting remains debatable 20, we believe this study adds important information about ECLS as a rescue therapy in patients with drug-induced cardiac arrest and refractory shock.