Lipid emulsion in the poisoned patient – a few answers, but more questions remain

In this issue, Levine et al. [1] report their analysis of cases entered in the lipid emulsion sub-registry of the Toxicology Investigators Consortium (ToxIC). The authors have attempted to determine whether the octanol-water partition coefficient (log P) explains the beneficial effect (if any) of lipid emulsion in poisoned patients. Specifically, they focus on survival – the outcome of greatest interest. In this analysis, the octanol-water partition coefficient comes up short as a predictor of outcome, a finding identified by Fettiplace and Weinberg [2].

Even if the lipid solubility were to explain part of the mechanism of action for lipid emulsion, the octanol-water partition coefficient (log P) has two limitations. First, it includes only the partition of non-ionized drug. Second, the measurement may occur over a range of pH values that include pH values far from the range of pH values that we may see even in our sickest patients. The preferred parameter is the octanol-water partition coefficient at a pH of 7 (log D) [3]. This parameter also includes both ionized and non-ionized forms of the drug in question. The two parameters may or may not be similar. For example, diphenhydramine has a log P of 3.65 but a log D of 1.92 at pH 7 [4].

The lipid emulsion sub-registry has highly detailed information about the included patients. This is both an advantage and a disadvantage. The sub-registry nested within the Core Registry includes numerous data fields to provide granularity in the observations. However, the additional detail in the sub-registry may be an impediment to clinicians entering some cases. We do not know whether this reflects a reporting bias of entering cases with successful outcomes or not. Within the Core Registry, 182 patients received lipid emulsion, but 44 patients – nearly one-quarter – did not have data added to the lipid emulsion sub-registry. We do not know their outcomes, but 81% of the patients in the sub-registry survived. Likewise, we do not know the identity of the drugs among these 44 lipid-treated patients. It is possible that reporting bias increased the likelihood of reporting apparent successes and decreased the likelihood of reporting apparent failures.

The high survival rate, regardless of the lipid solubility of the primary drug, may result from another aspect of the Core Registry. All patients in the Registry received treatment involving bedside consultations by medical toxicologists who were active participants in the Toxicology Investigators Consortium [5]. Simply receiving this expertise may confer a survival advantage compared to patients in non-participating hospitals with similarly severe poisonings.

The authors found 29 adverse events among 21 patients [1], but the degree of harm to patients is more difficult to ascertain. Eleven patients (8%) had interferences with some laboratory tests, although this is an expected finding [6–8]. Six patients (5%) reportedly developed pancreatitis after lipid emulsion, but it is unclear whether the authors classified these based only on elevated lipase activity or used the definition of elevated lipase activity plus abdominal pain, nausea, or vomiting as used in a previous case series [6]. Lipase is one of the colorimetric tests subject to lipemic interference [7].

Twelve patients (9%) in the present study had an increase in oxygen requirement [1], but we know little more about these patients. Were these patients who received 2 L/min of oxygen by nasal cannula or patients who required endotracheal intubation? Were these patients who had laboratory interference with blood gas measurements without clinical deterioration? Was the primary intoxicant or the lipid emulsion more culpable? The 2016 systematic review by the Lipid Emulsion Workgroup left this unclear [8]. The studies cited by Hayes et al. [8] with pulmonary abnormalities included 20 studies comprising 280 patients and eight healthy volunteers receiving various lipid emulsions for total parenteral nutrition (eight studies comprised 117 premature infants). Hayes et al. [8] found six studies (three case reports, two published case series, one case series in an abstract) comprising 30 patients receiving lipid emulsion as antidotal rescue. Of these 30 patients, ten had acute respiratory distress syndrome (ARDS) or ventilation/perfusion (V/Q) mismatch. Of these one received a bolus of between 500 and 1,000 mL for verapamil toxicity [9], while another received lipid emulsion continuously for 17 days for amitriptyline poisoning [10].

As with a previous review of fatal outcomes among poisoned patients receiving lipid emulsion [11], all patients received lipid emulsion with no comparison to similar patients without lipid emulsion.

We cannot draw clear conclusions about the role of lipid emulsion in the outcomes in either group of patients. In both study groups, nearly all patients received other treatments. Two-thirds of patients in the present study received vasopressors [1]. In the fatality study, authors classified 93% of the cases as “cocktail”, “last resort”, or “cardiac arrest” [11]. Giving lipid emulsion when cardiac arrest is imminent or has occurred is almost certainly futile. From these two studies, we cannot unequivocally attribute success or failure to lipid emulsion.

The two studies also had diverse intoxicants, and both shared the same five leading drugs or drug classes in a slightly different order [1,11]. The wide array of drugs complicated the analyses in both studies as they comprised drugs for which lipid emulsion might be useful, among other drugs for which it may offer no benefit or might even cause harm. Without uniform laboratory confirmation, the drug identifications frequently relied upon patient history. The final conclusion from Smolinske et al. [11] still stands: "The evaluation of the role and efficacy of intravenous lipid emulsion therapy in non-local anesthetic poisoning needs robust controlled clinical trials".

Until such trials occur – if ever they do – studies should seek to compare similar poisonings. For example, a recent systematic review of case reports focused on diphenhydramine poisonings [12], which are common and should be easier to study. Ostensibly, the review by Clemons et al. [12] found no significant change in QRS duration or corrected QT interval but did see a significant increase in mean arterial pressure among patients who received lipid emulsion than among those who did not. This mirrors a finding in the present review [1], and both together suggest another mechanism of action besides the “lipid sink”.

A path not yet taken by Levine et al. [1] is a retrospective case-control study. Among the 53,286 patients in the Core Registry who did not receive lipid emulsion, there surely must be other cases of poisoning by the leading drugs or drug classes (beta-adrenergic antagonists, calcium channel blockers, tricyclic antidepressants, bupropion, citalopram, and diphenhydramine). It should be possible to match multiple control patients to each index patient who received lipid emulsion. Each intoxicant or class would be suitable for a separate study.

As we await prospective, controlled trials that may never come, a case-control study would bring us closer to the answers we seek. Until then, it is difficult to move much beyond the 2016 recommendations of the Lipid Emulsion Workgroup [13].

Disclosure statement

No potential conflict of interest was reported by the author.

Additional information

Funding

The author reported there is no funding associated with the work featured in this article.