Unexpected survival after deliberate phosphine gas poisoning: An Australian experience of extracorporeal membrane oxygenation rescue in this setting

Abstract

Phosphine poisoning is responsible for hundreds of thousands of deaths per year in countries where access to this pesticide is unrestricted. Metal phosphides release phosphine gas on contact with moisture, and ingestion of these tablets most often results in death despite intensive support. A 36-year-old woman presented to a regional hospital after ingesting multiple aluminium phosphide pesticide tablets and rapidly developed severe cardiogenic shock. In this case, serendipitous access to an untested Extracorporeal Membrane Oxygenation (ECMO) service of a regional hospital effected a successful rescue and prevented the predicted death. We discuss the toxicology, management and the evidence for and against using ECMO in this acute poisoning.

Introduction

Acute poisoning presenting to hospital in the Western world has a relatively low mortality, well under 3%,^1–3 because of the interventions that can be provided: mechanical ventilatory support with intubation supports sedation-induced respiratory depression, renal failure and metabolic derangement can be bridged with renal support therapies, and even transplantation has been used in drug-induced fulminant liver failure. However, in the rare instances when the circulatory system continues to fail despite heroic support with infused fluids and medications, mechanical circulatory support is infrequently considered or provided. Cardiac pump failure is often the commonest reason for failed circulatory support and death in toxicology.

We describe a case in which severe shock due to cardiac failure was caused by aluminium phosphide (AlP) poisoning, and where a likely fatal outcome was averted by extracorporeal support. AlP poisoning is a rare presentation of poisoning in Australia, with the last published case in Australia occurring in 1999 and dying despite all interventions. ⁴ Written consent was obtained from the patient for publication of this report.

Case

A 36-year-old woman with no past medical history ingested multiple AlP tablets with suicidal intent. A relative, who handles AlP in his profession, found a glass with residual fragments of at least eight tablets in water. Other remaining tablets were sealed in plastic bags before the arrival of emergency services.

On arrival in the emergency department the patient was triaged directly to a negative pressure room due to the risk of cross contamination to the staff. She was awake with a blood pressure of 80/50 mmHg, and was tachypnoeic and mottled. Oxygen saturations were 93% on Fi0² 0.6 with a venous blood gas demonstrating pH 7.36, PCO² 22 mmHg, HCO³ 16 mmol/l and lactate of 8 mmol/l. She rapidly became less responsive and more cyanotic, and was intubated emergently with staff in full personal protective equipment. Further volume resuscitation, noradrenaline and adrenaline infusion, N-acetylcysteine and 50 g of activated charcoal via nasogastric tube were administered. A focused echocardiogram demonstrated severely impaired left and right heart systolic contractility, with a calculated left ventricular ejection fraction (EF) of 20%.

After transfer to the intensive care unit (ICU) there was rapid escalation of vasoactive requirements, to 35 μg/min of noradrenaline and 35 μg/min of adrenaline, to maintain a mean arterial pressure (MAP) of 60 mmHg. The lactate deteriorated to 13.8 mmol/l and pH to 7.1. A repeat echocardiogram by the cardiology service revealed further impairment of left ventricular systolic function with a calculated EF of less than 10%.

At this point support of the heart with veno-arterial extracorporeal membrane oxygenation (VA-ECMO) was proposed, and support for this decision was sought from two experienced centres, but there was no precedent or experience with phosphine toxicity. The toxicology service strongly considered the mortality to be close to 100%, so a decision was made to commence VA-ECMO support.

Percutaneous femoro-femoral cannulation configuration (multi-stage 21Fr venous access cannula to left common femoral vein, a 17Fr arterial return cannula to right common femoral artery and a 10Fr backflow (distal limb perfusion) cannula to right superficial femoral artery) was used to establish VA-ECMO support. Transthoracic rather than transoesophageal echocardiographic guidance was used for line placements due to the risks of exposing the operator and staff to gastric phosphine gas, and concern about oesophageal injury after ingestion.

The ECMO system used was Cardiohelp system (MAQUET Getinge, Rastatt, Germany) with HLS Set Advanced 7.0. The patient was therapeutically heparinised just before insertion of vascular cannulae, and for the duration of ECMO support. ECMO blood flow was commenced at 4 l/min, along with additional vasopressor support to achieve a MAP of 60 mmHg. Continuous renal replacement therapy (CRRT) via the ECMO circuit was continued to mitigate the severe metabolic acidosis. The patient remained on CRRT for 39 h to clear the metabolic acidaemia.

Serial echocardiograms showed improving cardiac function and the ECMO support was required for four days, during which time there was an uneventful semi-urgent circuit change due to oxygenator failure with progressively rising transmembrane pressure despite adequate heparinisation. The ECMO support period was also complicated by a transiently ischaemic leg on the arterial return cannula side, occurring despite a well-functioning backflow cannula, and this did not require vascular surgery. There was some evidence of disseminated intravascular coagulopathy, which would not be unexpected in a severe shock state requiring VA-ECMO.

Planned decannulation of the right femoral artery, superficial femoral artery and left common femoral vein with on-table thrombectomy and closure of the arterial puncture sites by vascular surgeons was straightforward. The weaning study echocardiogram demonstrated moderately reduced systolic function with left ventricular EF 42%. The patient was discharged from the ICU after a seven-day admission in ICU to a voluntary mental health facility. At discharge she appeared to have a full neurological and cardiovascular recovery.

Discussion

Phosphine (PH₃) is a gas released from solid AlP by exposure to moisture, and in concentrations as low as 5.0–10.0 ppm will kill both vertebrate and invertebrate pests. It is a colourless and odourless gas, but due to the manufacturing process is usually described as smelling of garlic or rotten fish. Although most usage is in the farming and food transport and storage industries, alternative sources of phosphine exposure may occur in the semiconductor industry, during manufacture of flame retardants, and potentially from terrorist threats. AlP poisoning is a rare presentation of poisoning in Australia, with the last published case in Australia occurring in 1999, and dying despite all interventions. ⁴

Phosphine has been shown to inhibit various cytochrome complexes of the mitochondrial respiratory chain, and this may be exacerbated by the generation of reactive oxygen species, as well as by the inhibition of detoxifying enzyme systems like glutathione reductase.^5,6 This disruption of mitochondrial function and resulting metabolic crises in more active tissues leads to cardiovascular collapse, cardiogenic shock and the pronounced metabolic acidosis. Patients may also present with agitation, lethargy, pulmonary oedema and hepatic dysfunction. There are no known antidotes, and none of the historical therapies proposed, such as infusions of magnesium, bicarbonate, and N-acetylcysteine, or lavage with potassium permanganate, coconut oil, sodium bicarbonate and activated charcoal have demonstrated efficacy.

The mortality from phosphine gas in developed health systems depends greatly on the type of exposure. Public health strategies in the developed world mean that most exposures are transient, accidental, and usually only to the gas in low concentrations. ⁷ Mortality rates after deliberate ingestion of the tablets remain very high, at greater than 60%, even in advanced health systems. ⁸ After deliberate and accidental ingestion occurring in less developed countries with less stringent controls and delayed presentation, mortality rates are described as exceeding 80%.^9–11 AlP and paraquat account for a huge proportion of self-harm deaths in these countries, almost 80% in some areas of India, and the worldwide prevalence of pesticide-related deaths is calculated to be 233,997 to 325,907 per year. AlP ingestion is so common in developing countries that several authors have identified features associated with likely death (Table 1), and management algorithms have been proposed to avoid wasting scarce resources on those likely to die.^11–17

Table 1.

Features associated with likely death after phosphine gas poisoning.

Prediction tools	Our patient
PGI* score(12) : (without ECMO use) one point each for:• pH < 7.25• GCS < 13• SBP < 87This test has a sensitivity of 55% and specificity of 98.2%and a 96.4% positive predictive value that the patient with the score of 3 will die.	Score was 3/3.
Farzenah et al. 16 nomogram uses the following inputs for prediction of death:• SBP, mmHg• Urine output, ml/day• GCS• Bicarbonate level, mmol/l	0.99 chance of death
Shadnia et al. 17 observed that a pH less than 7 measured during admission was associated with 100% mortality	pH 7.07
Mathai and Bhanu 11 observed a correlation between severity of illness scores and mortality:• APACHE II > 8 = 73% mortality• APACHE II > 25 = 100% mortality• SAPS II > 30 = 69.2%	APACHE II score was 17SAPS II score was 50 points

*PGI score: ‘P’ stands for pH, ‘G’ for Glasgow Coma Scale (GCS) score, and ‘I’ for impaired or low SBP (systolic blood pressure) (‘PGI’ also represents the short form of the name of the authors’ institute of PGIMER, Chandigarh, India).ECMO: extracorporeal membrane oxygenation; APACHE: Acute Physiology and Chronic Health Evaluation; SAPS: Simplified Acute Physiology Score.

Our patient had most of these high-risk features and features of both tissue metabolic paralysis and cardiac decompensation. Using the prognostic markers derived from experience in the developing world, the mortality without ECMO was predicted to be between 70% and 100%. This high risk of death, the rapid progression of cardiac pump failure occurring early in the course of the poisoning, the knowledge that the patient had ingested a large number of tablets and, most importantly, our awareness that in the case reports with survival there was well-described recovery of cardiac function over several days, encouraged our decision to commence VA-ECMO.

The use of ECMO in acute toxicity is rapidly increasing,^18–20 and is tentatively supported by a recent meta-analysis. ²¹ Its role seems to be in both cardiogenic shock due to cardiac-active drugs, and in severe hypoxaemia due to primary lung injury such as aspiration and hydrocarbon exposure. The greatest experience is in cardiotoxic drug toxicities, particularly calcium channel blockers, beta-blocker overdose, and tricyclics, with excellent survival rates of 80%–100% even in less experienced centres,^21–24 and despite the occurrence of cardiac arrest and ischaemic complications. ²⁵ The experience with veno-venous (VV) ECMO in secondary lung injury due to toxins is less consistently good, but survival has been described even in traditionally fatal ingestions of paraquat and other direct and indirect lung toxins.^26–28

The past decade has seen a parallel increase in the use of ECMO support in phosphine toxicity, and although individual cases have been reported from Israel, USA and Germany, most case series have been published from developing countries.^8,29–33 The largest series was an Indian retrospective observational study which identified 83 patients during this decade with severe AlP toxicity, and reported a mortality of 87% in those who did not receive ECMO, compared with only 33% mortality in those who did receive ECMO, despite the latter having worse baseline left ventricular EF and pH <7. The ECMO support period in all these series was typically less than seven days, and most people had normal recovery.

In most parts of the world, where ECMO is concentrated in high-volume metropolitan centres, a highly unstable ECMO candidate or a patient in cardiac arrest in a regional hospital would require either a rapid-response mobile ECMO service, or precipitously risky inter-hospital transport. In our setting of Australia, the inevitable delays would likely have been fatal. The hub-and-spoke model of lower volume regional centres that can initiate ECMO and transfer once achieving cardiovascular stability is an alternative application of the technology.

The feasibility of this model was supported by our experience as a regional hospital establishing a nascent ECMO service as part of statewide COVID-19 pandemic preparedness. Staff who had previously undertaken training in ECMO were identified and put through intensive ECMO training and credentialling, and the service was certified by the statewide ICU network within ten weeks. An in-house credentialling process was developed, a full-time ECMO nursing educator was appointed and ECMO readiness was certified by the statewide ICU network in weeks. Ostensibly the expectation was for only critically ill COVID-19 patients to be considered for ECMO support, but in view of the diagnostic uncertainty occurring in the early days of the COVID-19 pandemic, and considering the ethics of withholding an available and cost-equivalent resource, any suitable candidate was to be considered as long as they fit recognised criteria, and were acceptable to the metropolitan ECMO service.

Establishing any new therapy is likely to face arguments of lack of proof of benefit, expense and safety. Although we experienced some complications with this case, the intensive preparation and simulation, in addition to the technique of peripheral cannulation, meant we were able to recognise and manage them with local resources. Although ECMO has yet to have a large body of evidence of benefit, there is consistent support from clinicians who recognise its utility in transient severe cardiac dysfunction and cardiac arrest, and there is now emerging evidence from trials on the efficacy and cost-effectiveness of VA-ECMO.^34–38 In the setting of catastrophic but recoverable cardiac dysfunction caused by mechanisms such as poisonings and arrhythmias, the risk–benefit ratio of providing ECMO in low-volume centres is dramatically altered.

Conclusion

We described a rare but severe toxicity involving phosphine that had a high risk of death or of developing secondary organ failures, which would have required significant hospital resources but responded well to VA-ECMO support. The application of ECMO for acute cardiovascular collapse and toxicities is rapidly being established, and should be considered early, before established organ failures, and may change the paradigm of cardiovascular support.