Introduction

Ethylene glycol (EG) poisoning is associated with a high likelihood of acute kidney injury (AKI) [1,2,3] and mortality [4, 5]. In 2020, the US poison control centers reported 6036 calls relating to EG, 586 of which had at least moderate clinical effects and 30 of which resulted in death [6]. Hemodialysis was first reported in the management of an EG-poisoned patient in 1959 [7] and became a critical component of the management of EG poisoning [8]. However, with the advent and wider availability of fomepizole, the indications for extracorporeal treatments (ECTRs), such as intermittent hemodialysis and continuous kidney replacement therapy (CKRT), have evolved and the role of ECTRs is  currently being challenged [9,10,11].

The EXtracorporeal TReatments In Poisoning (EXTRIP) workgroup is composed of international experts representing diverse specialties and professional societies (Additional file 1: Table S1). Its mission is to provide recommendations on the use of ECTRs in poisoning (http://www.extrip-workgroup.org). The rationale, background, objectives, methodology, and its initial recommendations are previously published [12,13,14]. The objective of this article is to present EXTRIP’s systematic review of the literature and recommendations for the use of ECTR in patients poisoned with EG. Although diethylene glycol and other alcohols share common characteristics with EG, this review is restricted to EG poisoning.

Physicochemical characteristics and toxicokinetics

The toxicokinetics of EG are summarized in Table 1. EG, like other alcohols, is a small water-soluble molecule that is quickly and completely absorbed in the gastrointestinal tract. It has negligible protein binding and distributes in total body water. One third of absorbed EG is eliminated unchanged in urine while two-thirds are oxidized in the liver by the enzyme alcohol dehydrogenase (ADH) to glycolaldehyde, which is then rapidly converted to glycolate by aldehyde dehydrogenase [15]. Glycolate is converted to glyoxylate by glycolate oxidase which is the rate-limiting step. Glyoxylate is later metabolized by various pathways to oxalate and non-toxic products such as glycine and α-hydroxy-β-ketoadipate. EG elimination follows first-order pharmacokinetics but a biphasic elimination profile is described [16]. Total body clearance of EG is approximately 100 mL/min, a fourth of which is attributable to kidney clearance and is directly proportional to kidney function [15]. Consequently, the half-life (T1/2) of EG is prolonged in patients with a decreased glomerular filtration rate (GFR). Ethanol and fomepizole both compete with EG for ADH, so their administration prevents the metabolism of EG, prolonging the apparent elimination T1/2 of EG. Because fomepizole has a stronger affinity and inhibition of ADH compared to ethanol [17, 18], EG elimination T1/2 is longer during fomepizole than during ethanol therapy (Table 1).

Table 1 Chemical characteristics and toxicokinetics of ethylene glycol
Full size table

Review of ethylene glycol toxicity

EG is the main component of commercial antifreeze and is present in many industrial products including coolants and de-icers. EG itself has minimal toxicity, but its metabolites are responsible for most of the clinical effects; glycolate contributes to the acidemia, while deposition of calcium oxalate crystals in tissues causes AKI and neurological complications [19,20,21].

The initial clinical manifestations of EG poisoning mimic those of ethanol ingestion, namely inebriation and ataxia. As EG is metabolized, metabolic acidemia appears after a latent period of approximately 3–6 h after ingestion. Thereafter, progressive neurotoxicity (coma, cerebral edema, cranial nerve palsies, and seizures), cardiotoxicity (tachycardia with hypertension or hypotension), respiratory distress, and AKI occur. The incidence of AKI varies between 30 and 70% [22,23,24,25,26,27,28,29,30,31,32,33,34]. Multiorgan failure and death can occur at this stage. Cranial nerve palsies, radiculopathy, and other neuropathies may appear several days after ingestion, despite treatment [35,36,37]. Rarely, brainstem and basal ganglia injuries are reported [38, 39].

A threshold dose for toxicity is poorly defined in humans. Aircraft de-icing workers systemically exposed to an estimated 27 mg/kg from aerosolized EG (≈ 2 mL of pure EG) did not demonstrate any adverse effects [40]. Self-experiments with EG revealed no harm with pure EG ingestions of 10–30 mL [41,42,43]. In one cohort of 86 unintentional ingestions of < 100 mL EG, all patients survived [44] and only one patient developed mild AKI, although most were treated with ethanol and/or hemodialysis within 3 h of ingestion. The often-quoted lethal dose in an untreated 70 kg adult is 100 mL [45,46,47,48], although there are several cases of toxicity and even death below this dose (Additional file 1: Table S7) [37, 49,50,51,52,53,54,55,56]. EG dose is prognostic of outcomes only if there is a delay to treatment [46, 57, 58]. Similarly, a concentration threshold for toxicity is unknown; some sources quote a peak EG concentration > 3.2 mmol/L (20 mg/dL) as a risk for toxicity [59], over which treatment should be initiated [11]. Some authors propose a treatment threshold of 10 mmol/L (62 mg/dL) in asymptomatic patients if base deficit is below 10 mmol/L, based on molar-molar conversion of EG to glycolate in the absence of supporting clinical data [60]; this is not a routine recommendation by most poison centers. Toxicity did not occur in 7 untreated patients who had EG concentrations < 4.8 mmol/L (30 mg/dL) [44, 61, 62]. Reports of toxicity when the EG concentration is < 3.2 mmol/L (20 mg/dL) are exclusively in patients who have already metabolized EG when testing is performed and does not represent peak EG concentrations. Because EG itself causes little toxicity, the EG concentration is poorly predictive of mortality [23, 24, 27, 28, 32, 63,64,65,66,67,68,69].

Other than dose and concentration, EG toxicity is modulated by co-ingestion with ethanol because this decreases EG metabolism [25, 29, 69, 70]. Patients in whom medical care is delayed, for example late presentations, develop more toxicity due to a higher concentration of toxic metabolites [23, 63]. Data suggest that a delay of 6–12 h between EG ingestion and treatment initiation is associated with an increased risk of immediate and long-term complications [32, 55, 58], although this was not confirmed in other studies [52, 63, 65].

Complications from EG poisoning are predicted by the concentration of plasma glycolate and associated acid–base disorders [23, 24, 29, 63, 66, 68, 71,72,73]. The development of AKI is closely correlated with other outcomes including death [29, 31, 32, 52], as AKI is a marker of metabolite-mediated organ injury, and it delays kidney excretion of EG. Death very seldom occurs if AKI is not present [23, 27, 29]. Other clinical markers of mortality are coma [23, 27, 31, 32, 63, 65, 67, 74], respiratory failure [23, 27, 31, 32], hypotension [23, 31], and seizures [31, 63, 65].

With better supportive care, increasing awareness of treatment priorities, widespread use of antidotes, and greater availability of ECTR, mortality from intentional EG poisoning has steadily decreased: The mortality exceeded 80% prior to 1960 [7, 75, 76] and decreased to 30–40% in the 1970s and 1980s [22, 23, 53, 56, 77,78,79]. This trend has continued to improve during the 1990s [24, 27, 31, 52, 80] with mortality declining to < 10% today [32, 33, 67, 81]. High mortality rates are still reported when antidotes and/or ECTR are not readily available [4, 5, 51, 63, 82, 83].

Persisting sequelae are unusual in survivors. AKI lasts approximately 7–10 days [27, 65, 84, 85] and kidney function returns to baseline in most patients [27, 86] However, there are cases of residual chronic kidney disease (CKD) [80, 87,88,89] including patients who remain dialysis-dependent [26, 36, 84, 90,91,92] one year later. Similarly, neuropathy regresses over time although there are cases of chronicity [93,94,95,96,97,98,99]. The incidence of these sequelae is unclear but appears to be less than 1% [68].

Standard care for patients with an EG exposure includes assessment for and treatment of abnormalities of the airway, breathing and circulation, correction of acid–base disorders, and ADH-inhibitor therapy (Additional file 1: Table S2). Ethanol and fomepizole decrease EG metabolism through competitive inhibition of ADH, and both prevent toxicity and death in EG-poisoned animals [18, 100,101,102,103]. Ethanol has been used as an antidote in humans since the 1960s [104]. Fomepizole was approved in the USA in the late 1990s [28] and has largely replaced ethanol as the antidote of choice for EG poisoning in many countries [81]. Thiamine and pyridoxine are used to facilitate the conversion of glyoxylate to non-toxic metabolites rather than oxalate, but their clinical utility has never been determined.

Methods

The workgroup developed recommendations on the use of ECTR in EG exposures, following the EXTRIP methodology previously published [13], with modifications, updates, and clarifications. The methods and glossary are presented in full in the Supplementary Material, including the PRISMA checklist (Additional file 1: Table S3), dialyzability criteria (Additional file 1: Table S4), quality assessment of individual toxicokinetic studies (Additional file 1: Table S5), and evidence (Additional file 1: Table S6). These data were assessed according to GRADE methodology (Additional file 1: Figure S1) which also informed the voting process for recommendations (Additional file 1: Figure S2). Cases in which ECTR was performed solely for anuria or related complications were excluded.

Complementary searches were also performed to answer specific questions regarding 1) dialyzability of EG metabolites, 2) dialyzability of ethanol/fomepizole, 3) dialyzability of pyridoxine and thiamine.

Results

Results of the primary literature search (first performed on March 1, 2019, and last updated February 1, 2022) are presented in Fig. 1.

Fig. 1
figure 1

Article selection

Full size image

A total of 1296 articles were identified after removal of duplicates. In the final analysis, 226 articles were included for qualitative analysis, including two animal experiments [105, 106], five cost-evaluation studies [107,108,109,110,111], 31 cohort studies with pooled data [5, 22,23,24, 27, 28, 31,32,33,34, 56, 58, 67, 69, 73, 84,85,86,87, 112,113,114,115,116,117,118,119,120,121,122,123], one toxicokinetic modeling study [124], 181 case reports and case series with individual patient-level data [7, 25, 26, 29, 30, 37,38,39, 44, 51, 53, 64, 66, 70, 80, 89, 104, 125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218,219,220,221,222,223,224,225,226,227,228,229,230,231,232,233,234,235,236,237,238,239,240,241,242,243,244,245,246,247,248,249,250,251,252,253,254,255,256,257,258,259,260,261,262,263,264,265,266,267,268,269,270,271,272,273,274,275,276,277,278,279,280,281,282,283,284,285,286,287,288], three comparative studies [52, 55, 63], and three systematic reviews [289,290,291]. No randomized controlled trials were identified. One fourth of all selected articles were in languages other than English.

Other articles were obtained, although they were not part of the initial systematic literature search, namely relating to dialyzability of EG metabolites [292,293,294,295,296,297,298,299,300,301,302,303,304,305,306], dialyzability of ethanol [307,308,309,310,311,312,313,314,315,316,317], dialyzability of fomepizole [318,319,320,321,322,323], and dialyzability of pyridoxine and thiamine [324,325,326,327,328,329,330,331,332,333,334,335,336,337,338,339,340,341]. Data from publications reporting the same subjects were merged but the citations were only counted once in the systematic review, e.g., [15, 28, 211] and [19, 147, 151, 342].

Summary of the evidence: dialyzability

Dialyzability of EG

EG and its metabolites have characteristics of dialyzable poisons [343], namely small size, high water solubility, absence of protein binding, and a low volume of distribution (Table 1). EG clearance with high-efficiency intermittent hemodialysis is similar to that of urea [133, 172], approximating plasma flow and can surpass 200 mL/min (Table 2) [15, 148, 209]. EG clearance from hemodialysis increases total clearance by at least 800%, compared to endogenous clearance (assuming adequate ADH blockade). As for other small molecules, increasing blood and effluent flow and using a dialyzer with a higher surface area will increase EG clearance [163]. Mass removal of EG can exceed 100 g during a 6-h hemodialysis [148, 202, 287, 342].

Table 2 Toxicokinetics of ethylene glycol and metabolites during various extracorporeal treatments
Full size table

Continuous and hybrid techniques (e.g., CKRT, sustained low-efficiency hemodialysis, extended daily dialysis) provide inferior EG clearances and mass removal compared to standard intermittent hemodialysis, due to their lower blood and/or effluent rates (Table 2) [196, 287]. Nevertheless, these techniques still achieve a substantial increase in total EG clearance. As EG has negligible protein binding, no advantage would be expected from therapeutic plasma exchange, liver support devices, and hemoperfusion; in one case, EG clearance during hemoperfusion only reached 50 mL/min and quickly decreased because of extensive cartridge saturation [143]. Low-efficiency techniques like peritoneal dialysis have modest effect on removal of both EG or metabolites [128, 132, 139, 303], but will provide a clearance that exceeds endogenous clearance in the presence of AKI and adequate ADH blockade. No toxicokinetic data exist for exchange transfusion. After completion of ECTR, a rebound increase in EG concentration was observed in 21% of the cohort and the median magnitude of this rebound was 30% of the immediate post-ECTR concentration; in one case [144], the rebound was substantial (200%).

As mentioned, once ADH is blocked by ethanol or fomepizole, endogenous EG clearance is at best 30 mL/min, which is modest relative to extracorporeal EG clearances with modern-day efficient ECTRs (Table 2) [148, 192, 209, 211]. For this reason, dialyzability was not graded based on kidney function. Although most of the toxicokinetic data are dated with technology considered substandard today, EG was considered “dialyzable” with intermittent hemodialysis (level of evidence = B, Table 3), “moderately dialyzable” with CKRT (level of evidence = D), “slightly dialyzable” with peritoneal dialysis (level of evidence = C), and “slightly dialyzable” with hemoperfusion (level of evidence = D). No data exist on intermittent hemodiafiltration, but it is expected to perform as well as hemodialysis, based on achievable urea clearance. In the rare scenario in which no antidote is available, high-efficiency hemodialysis would increase endogenous clearance at least 100% [278], i.e., “moderate” dialyzability.

Table 3 Final toxicokinetic grading of ethylene glycol and glycolate during extracorporeal treatments
Full size table

Dialyzability of ethylene glycol metabolites

EG metabolites have physicochemical characteristics indicative of being dialyzable, as is confirmed by data; however, caution is required when grading dialyzability according to half-life comparison, during and off ECTR, as these may be influenced by ongoing variable production of metabolites, especially when ADH inhibition is inadequate.

Glycolate: High glycolate clearance (> 150 mL/min) and high mass removal of glycolate (up to 50 g) are reported during hemodialysis [19, 164, 211]. There are reports of glycolate concentrations increasing modestly during ECTR suggesting that production surpassed elimination and that ADH blockade was inadequate [165]. Glycolate can also reaccumulate after ECTR, especially if ADH blockade is not continued [181, 185, 344]. Based on 4 patients in whom dialyzability could be assessed, glycolate was rated as “dialyzable” with hemodialysis (level of evidence = C) (Table 3).

Oxalate: Oxalate was detected in blood in only 4 of the 10 patients in whom it was measured, in concentrations at least 20 times lower than those observed for glycolate [150, 241, 250, 252]. In one report, three sessions of hemodialysis removed on average of 380 mg of oxalate, although dialyzability could not be estimated [131]. From studies in dialysis-dependent CKD patients with primary or secondary oxalosis, oxalate clearance surpasses 150 mL/min with hemodialysis and hemodiafiltration [301, 303,304,305], which is at least 300% more than the kidney elimination capacity [292, 297]. Oxalate clearance in peritoneal dialysis is consistently less than 8 mL/min [139, 293, 295, 296, 303].

Glyoxylate: Glyoxylate extracorporeal clearance by hemoperfusion–hemodialysis was 71 mL/min in one patient [176].

Dialyzability of ethanol/fomepizole

Both ethanol [316] and fomepizole [318, 322, 345] are extensively removed by ECTR. If the dose of either of these antidotes is not increased during the ECTR session, a risk of inadequate inhibition of EG metabolism during ECTR exists. The elimination T1/2 of ethanol and fomepizole during hemodialysis and hemodiafiltration ranges between 1.5 and 3.0 h [308, 310, 312, 313, 315, 319, 321, 345, 346], and ECTR clearance surpasses 100 mL/min [198, 202, 272, 308, 310, 319, 345]. Considerably lower amounts of these antidotes are removed during CKRT [272, 320, 347, 348], and particularly during peritoneal dialysis [307, 311].

Dialyzability of pyridoxine and thiamine

Pyridoxine: In vivo clearance of pyridoxal-5’-phosphate with cellulose filter membranes averaged 170 mL/min in 6 subjects [328], while it was < 1 mL/min with peritoneal dialysis [329].

Thiamine: Thiamine concentration decreased between 5 and 40% during hemodialysis, but extracorporeal clearance was not calculated [325, 332, 336, 340].

Summary of the evidence: pre-clinical data

Three animal studies with group comparisons were identified [105, 106, 349]. In one experiment, an LD400 dose of EG was given to 23 dogs; 13 were treated with intravenous NaHCO3 and 10 were treated with a single session of hemodialysis for 20–24 h. All died in the NaHCO3 group while two died in the hemodialysis group (p < 0.0001), suggesting a beneficial effect of hemodialysis [105]. In one experiment of six EG-poisoned dogs, five received 3 h hemoperfusion within 5 h of poisoning and one was a control. All died and no improvement was seen from hemoperfusion [106].

Summary of the evidence: clinical data

Comparative data

We identified one retrospective study in which supportive care with ECTR (n = 28) was directly compared to supportive care alone (n = 28, Table 4) [63]. The mortality was the same in both groups (8/28 = 28.6%), although the ECTR group was sicker at baseline (higher EG dose, lower pH or HCO3− concentration, higher anion gap, greater percent with coma and respiratory failure, longer delay to admission, longer delay to antidotal therapy). An important limitation is that only about half of the entire cohort received antidotal therapy (ethanol).

Table 4 Included studies comparing the effect of extracorporeal treatments vs no extracorporeal treatments
Full size table

Cohorts in which the effect of ECTR could be analyzed were extracted and analyzed (Table 4) [5, 24, 25, 29, 32, 44, 107, 118]. Unfortunately, none of these studies were designed to appropriately determine the effect of ECTR, and all contained critical methodological flaws that preclude meaningful conclusions. These limitations include small sample size, retrospective design (except for poison control data which have other limitations), unclear patient selection, variable definitions of EG poisoning (reported amount ingested vs laboratory testing), unreported baseline characteristics and exposure details (especially co-ingestion with ethanol, volume EG ingested, time from EG ingestion to admission, antidote used), imprecise indications for ECTR, and confounding-by-indication, i.e., ECTR is preferentially used in those with more severe clinical features. Some cohorts included patients that could have been plausibly managed without ECTR [44]. One study suggested that hemodialysis increased the odds of severe outcomes 17-fold [118], after adjusting for age, gender, year, addition of bittering agent, administration of antidote and admission to critical care. This likely quantifies the extent of confounding-by-indication and selection bias in real-life presentations. There were studies that compared types of ECTR [5, 24, 52, 55] (Additional file 1: Tables S8 and S9), number of treatments [65], or the impact of time to ECTR initiation on clinical outcomes [32, 56] (Additional file 1: Tables S8 and S9). No meaningful conclusions can be inferred from these studies, as the same major limitations apply. Therefore, clinical data were not considered suitable for inclusion in a meta-analysis comparing ECTR to no ECTR.

Clinical cases

A total of 446 cases had sufficient patient-level data, the summary of which are shown in Table 5. Most patients received ethanol rather than fomepizole for ADH blockade, reflecting older literature or country of origin. Intermittent hemodialysis was by far the most used ECTR. Fifty-four percent of patients received more than one ECTR session (usually for management of uremia), comparable to other cohorts [32]. Acidemia was corrected quickly in most cases receiving high-efficiency hemodialysis, usually within four hours. Peritoneal dialysis was switched to hemodialysis because of clinical failure in one patient [160]. Mortality from the entire cohort was 18.7% and the median time to death was 96 h after ingestion. Some patients with massive ingestions (> 1 L) or very high EG concentrations (> 200 mmol/L or > 1240 mg/dL) survived [123, 214, 238, 274, 281, 282], as did some with extreme acid–base abnormalities (e.g., pH < 6.60 or HCO3−  < 2 mmol/L) [25, 26, 29, 104, 180, 190, 191, 215, 223, 226, 236, 244, 248, 254, 257, 264, 276]. As suggested in one review [71], poor outcomes were infrequent when the glycolate concentration is < 12 mmol/L or the anion gap (with potassium, calculated as Na+ + K+– Cl- – HCO3−) is < 28 mmol/L (Additional file 1: Table S10); three such patients who received ECTR died, in two cases there were limited details reported [51], and one died without receiving an ADH inhibitor [126]. Mortality in patients who had an anion gap over 28 mmol/L was much higher (20.4%).

Table 5 Clinical summary of included cases with ethylene glycol poisoning
Full size table

Several complications occurred during ECTR although in most cases, these can be attributed to EG poisoning rather than the procedure itself. Complications assessed as likely related to ECTR included hypotension [7, 127, 130, 153, 166, 181, 258, 265, 275, 276], bleeding related to heparin [141], catheter-related thrombosis [291], catheter-related bacteremia [166, 183], cardiac arrest [167], and death [153, 155]. ECTR can potentially increase intracranial pressure [350,351,352] and may have aggravated EG-related cerebral edema leading to seizures [141, 177, 178]. Some authors mentioned concerns related to rapid fluid and solute shifts during ECTR, although it is unclear if these resulted in injury [178, 219]. In one cohort of 72 patients receiving hemodialysis for toxic alcohol poisoning (34 for EG), 20 patients experienced a hemodialysis-related adverse reaction including three cases of hypotension and one case of arterial tear during catheter insertion leading to internal bleeding, shock, and cardiac arrest (the patient eventually recovered) [353].

Persistent sequelae were seldomly reported in case reports and observational cohorts. The median duration of kidney replacement therapy for patients who developed AKI was 9 days [IQR 3,14], while the median duration of serum creatinine concentration elevation was 21 days [IQR 7,40], similar to results in published cohorts [23, 25, 27, 65, 84, 85]. Some degree of CKD was present in 16.8% of patients and 2.9% remained dialysis-dependent; however, in most cases, these were noted on hospital discharge and the duration of follow-up, when reported, was often short, so it is expected that these numbers are overestimated. On extended follow-up, < 1% of patients remained dialysis-dependent at 6 weeks and < 5% had CKD at 6 months [30, 52, 65, 68, 86], although there are rare reports of patients who remained dialysis-dependent after 1 year [26, 36, 84, 90,91,92]. Among other sequelae, there were rare cases of anoxic brain injury, basal ganglia injury, and irreversible cranial nerve palsies [37,38,39, 80, 94, 128, 151, 201, 215, 221, 254, 354].

Cost analysis of ECTR

In patients presenting prior to the development of acidemia or AKI, ECTR may shorten length of stay and associated morbidity due to nosocomial complications and reduce overall healthcare costs. Studies (mostly performed in the USA) report a cost advantage when hemodialysis is added to fomepizole, especially at higher EG concentrations (Table 6) [107,108,109,110,111, 227, 267, 355]. However, these data are dependent on many factors, including health care delivery model, patient location (medical ward vs high-dependency unit), type of ECTR, initial concentration of EG, need for transfer to another institution, and cost of fomepizole, all of which vary between countries and institutions, and are therefore not generalizable.

Table 6 Summary of studies evaluating costs of ECTR vs no ECTR in ethylene glycol poisoned patients
Full size table

Clinical data informing the evidence table

The evidence table is shown in Table 7. It was not possible to reliably populate the evidence table for patients with advanced or late EG poisoning (e.g., glycolate concentration > 12 mmol/L or anion gap with potassium > 28 mmol/L) because very few controls (i.e., not treated with ECTR) exist. The exceptions are patients i) too sick to undergo hemodialysis; ii) who died before hemodialysis was initiated; iii) in whom EG poisoning was not diagnosed; or iv) when ECTR was unavailable [85, 356, 357]. For obvious reasons, these patients are not adequate controls. Patient cohorts published before hemodialysis became widely available report extremely high mortality, although this could also be explained by inadequate standard care and absent antidotal therapy. A prospective randomized trial of ECTR in late EG poisoning would not be considered ethical because of a lack of clinical equipoise. (It is assumed that ECTR would produce a large survival effect.)

Table 7 Evidence profile table: ECTR + standard care compared to standard care in patients with “early” ethylene glycol poisoning#
Full size table

The evidence table was therefore populated with the subset of patients known to have a good prognosis without ECTR, i.e., those with a glycolate concentration ≤ 12 mmol/L or an anion gap with potassium ≤ 28 mmol/L (“Early EG poisoning,” Table 7) [29, 66, 71, 358]. As expected, the addition of ECTR to fomepizole did not seem to improve the prognosis of these patients, namely mortality, length of stay, or requirement of ECTR for AKI. ECTR appears to reduce costs, although this must be weighed against possible risks of complications of ECTR. When ethanol is used in low-risk patients, ECTR reduces the duration of time during which a patient is exposed to risks related to ethanol (decreased consciousness, dysphoria) and may limit the risk of ethanol failure. In rare cases in which neither fomepizole nor ethanol can be used (a situation described by some panel members), ECTR would theoretically markedly reduce the risk of mortality and AKI.

Clinical recommendations

The EXTRIP recommendations for ECTR in EG poisoning are presented in Table 8. Current indirect evidence suggests that ECTR is lifesaving when significant acidemia and/or Stage 2 or 3 AKI are present as it corrects metabolic disequilibria and removes EG, oxalate, and glycolate. EG-related mortality increases substantially once the plasma glycolate concentration exceeds 12 mmol/L [71]. Uncertainty remains on the expected magnitude of effect of ECTR compared to no ECTR on mortality and other patient-important outcomes at different plasma glycolate concentrations. When neither AKI nor acidemia is present, the advantages of ECTR when added to an ADH inhibitor are mainly to limit costs, reduce length of hospitalization, and limit risks of ethanol therapy when used, rather than reducing the occurrence of major adverse outcomes from EG. These recommendations may be used to prioritize and triage ECTR in cases of group poisoning [24, 359]. Because of the lack of reliable comparative studies and the aforementioned uncertainties, the quality of evidence was very low for all recommendations.

Table 8 Final EXTRIP recommendations on the use of ECTR in ethylene glycol poisoning
Full size table

Indications

The following indications should be considered independent of each other. For example, if a criterion recommending ECTR is met, then evaluation of the other criteria is not necessary.

EG Dose
  1. a.

    In patients presenting with EG poisoning, we recommend against ECTR based solely on the reported EG dose ingested (strong recommendation, very low-quality evidence)

Rationale: A reported EG dose ingested is never by itself an indication for ECTR as it may be imprecise and requires complementary confirmation from other diagnostic cues such as the presence of EG in blood, an elevated osmol gap and/or an elevated anion gap, or other non-specific tests (calcium oxalate crystals in urine). However, a history of ingestion may prompt early contact and consideration of transfer to a center where ECTR can be performed should this be required. Similarly, neither oxalate crystals in urine, urine immunofluorescence nor hypocalcemia nor a history of EG exposure alone are indications for ECTR (but may help to diagnose EG poisoning).

Plasma EG concentration
  1. a.

    Fomepizole is used:

    1. i.

      In patients presenting with EG poisoning, we suggest ECTR if EG concentration is > 50 mmol/L (> 310 mg/dL) (weak recommendation, very low-quality evidence)

  2. b.

    Ethanol is used

    1. i.

      In patients presenting with EG poisoning, we recommend ECTR if EG concentration is > 50 mmol/L (> 310 mg/dL) (strong recommendation, very low-quality evidence)

    2. ii.

      In patients presenting with EG poisoning, we suggest ECTR if EG concentration is 20 to 50 mmol/L (124 to 310 mg/dL) (weak recommendation, very low-quality evidence)

  3. c.

    No antidote is available

    1. i.

      In patients presenting with EG poisoning, we recommend ECTR if EG concentration is > 10 mmol/L (> 62 mg/dL) (strong recommendation, very low-quality evidence)

Rationale: The EG concentration is poorly prognostic of clinical outcomes because EG itself causes little toxicity. In fact, patients receiving fomepizole alone have excellent outcomes regardless of the EG concentration, assuming there is no kidney impairment and minimal academia [358]. The benefit of ECTR in this context is to presumably reduce length of stay and total hospital costs, especially at high EG concentration, rather than reducing the incidence of outcomes such as mortality or AKI and explains why this is "suggested" rather than "recommended" [107,108,109,110,111, 227, 267]. However, the workgroup acknowledges that these decisions need to be individualized (which GRADE emphasizes for weak/conditional recommendations) as cost considerations are dependent on the setting and institution; for example, costs for ECTR and fomepizole may exceed fomepizole alone if a patient needs to be transferred to another center for ECTR.

The same EG concentration cutoff was chosen when ethanol was used as an antidote, although this was a recommendation. The rationale being that ADH blockade with ethanol is more unpredictable and there are cases of treatment failure even with little to no acidosis or kidney impairment present on admission [358]. Prolonged ethanol therapy also carries risks and requires admission to a high-dependency unit which may be shortened by using ECTR. With EG concentration > 50 mmol/L, assuming an endogenous EG T1/2 = 14 h during ethanol therapy, a patient would need treatment for > 48 h before reaching a safe EG concentration, during which the risks of side effects from ethanol (central nervous system depression, dysphoria) and/or therapeutic failure become considerable. If no ADH inhibitor can be used, an EG concentration > 10 mmol/L (> 62 mg/dL) should prompt ECTR, as adverse outcomes are generally reported in untreated patients over this concentration; however, until there is a better understanding of threshold ethylene glycol concentrations, the workgroup recognizes that a more cautious cutoff may be preferable (see research gap below). From a known EG concentration and using local cost of fomepizole, ethanol, hospitalization, and ECTR, and using the EG T1/2 during specific circumstances (ECTR, AKI, antidote used, Table 1 and Table 2), clinicians can estimate the time to reach a safe concentration and decide if ECTR would be cost-effective in the specific scenario (Additional file 1: Figure S3). The decision to transfer a patient to another institution to receive ECTR should be individualized.

At extremely high EG concentrations with an associated high plasma osmolality, there is a potential risk of inducing osmotic disequilibrium with ECTR. However, these complications are considered unlikely because of the acute onset of the hyperosmolality and were only reported in one out of the 27 cases with an EG concentration > 100 mmol/L(> 620 mg/dL) [268]. The workgroup’s recommendations remain applicable for these patients.

The osmol gap (calculated as measured osmolality − calculated osmolarity, in SI units and adjusted for ethanol) when there is evidence of EG exposure
  1. a.

    Fomepizole is used

    1. i.

      In patients presenting with EG poisoning, we suggest ECTR if the osmol gap is > 50 (weak recommendation, very low-quality evidence)

  2. b.

    Ethanol is used

    1. i.

      In patients presenting with EG poisoning, we recommend ECTR if the osmol gap is > 50 (strong recommendation, very low-quality evidence)

    2. ii.

      In patients presenting with EG poisoning, we suggest ECTR if the osmol gap is 20 to 50 (weak recommendation, very low-quality evidence)

  3. c.

    No antidote is available

    1. i.

      In patients presenting with EG poisoning, we recommend ECTR if the osmol gap is > 10 (strong recommendation, very low-quality evidence)

Rationale: EG assays are seldom available locally in an appropriate time frame to influence clinical decisions [360, 361], and so the osmol gap is often used as a surrogate to predict the EG concentration [113, 119, 172, 342, 362,363,364]. Unfortunately, many clinical conditions and ingestion of other alcohols increase the osmol gap [365]. Conversely, the osmol gap may be “normal” or even below 0 if EG is already metabolized or if too little EG is ingested [24, 366,367,368,369]. For these reasons, the osmol gap is a poor screening test for EG ingestion, especially at low osmol gap values [370,371,372,373]. However, at high EG concentration, the osmol gap correlates linearly with the EG concentration (Additional file 1: Table S11), despite considerable inter- and intra-patient variability [371]. The panel proposed that the same cutoffs for osmol gap and EG concentrations be used for initiation of ECTR, especially if there is a confirmed history of EG ingestion. Since the osmol gap may overestimate the EG concentration, the panel acknowledges that using these cutoffs may lead to unnecessary ECTRs. If no antidote is available, an osmol gap > 10, in the context of EG exposure, is a reasonable criterion for hemodialysis, with the above caveats [371]. The workgroup did not provide an osmol gap cutoff when there is no/very low suspicion of EG poisoning. The workgroup also acknowledges that there are many formulas to calculate to osmol gap [368, 369, 374, 375].

Plasma glycolate concentration
  1. a.

    In patients presenting with EG poisoning, we recommend ECTR if the glycolate concentration is > 12 mmol/L (strong recommendation, very low-quality evidence)

  2. b.

    In patients presenting with EG poisoning, we suggest ECTR if the glycolate concentration is > 8 mmol/L (weak recommendation, very low-quality evidence)

Rationale: Glycolate is the EG metabolite in highest concentration in blood [19] and correlates with AKI and death [71]. There was only one death reported when the glycolate concentration was < 12 mmol/L [376]; however, the mortality rate rises substantially once the glycolate concentration exceeds 12 mmol/L [71].

Anion gap when there is evidence of EG exposure
  1. a.

    In patients presenting with EG poisoning, we recommend ECTR if the anion gap (calculated as Na+ + K+– Cl−– HCO3.−) is > 27 mmol/L (strong recommendation, very low-quality evidence)

  2. b.

    In patients presenting with EG poisoning, we suggest ECTR if the anion gap is 23–27 mmol/L (weak recommendation, very low-quality evidence)

Rationale: Glycolate assays are not available in most institutions. The anion gap is by far the best surrogate marker for glycolate and correlates linearly with glycolate and is associated with clinical outcomes [29, 67, 71, 72, 74]. Based on previous reviews [71], the workgroup agreed that an anion gap of 24–28 mmol/L and > 28 mmol/L, respectively, would best correlate with the suggested and recommended indications for ECTR based on plasma glycolate concentrations. To support this, there were only 3 patients out of 84 with an anion gap ≤ 28 mmol/L or glycolate concentration ≤ 12 mmol/L on admission who died; in all these cases, the clinical and metabolic data were incomplete (Additional file 1: Table S10). Because the relationship between glycolate concentration and prognosis require confirmation, the workgroup chose slightly more conservative cutoffs, accepting that this may lead to unnecessary ECTRs.

The anion gap may overestimate (e.g., concomitant AKI or ketoacidosis) or underestimate (e.g., hypoalbuminemia or co-ingestions of lithium or barium) [377,378,379] the glycolate concentration. In such circumstances, other acid–base parameters such as pH, HCO3−, and base excess can also be consulted with due consideration of the extent to which they are also influenced by other factors such as inadequate respiratory compensation or exogenous bicarbonate. It is important to note that the anion gap is only useful to predict glycolate concentrations if there is a high pre-test probability of EG exposure. Its value to predict the need for ECTR is poor if it is used indiscriminately; when there is little or no evidence of EG exposure, an elevated anion gap is not by itself an indication for ECTR as this may be caused by various factors.

An elevated glycolate concentration can falsely elevate the plasma lactate concentration on some analyzers [239, 380,381,382,383,384], which has prompted some to suggest using the “lactate gap” as a surrogate of glycolate. However, this requires knowledge of the specific analyzer’s cross-reactivity so it cannot be simply formalized into a recommendation.

Some authors suggest that hemodialysis may be obviated even in cases of severe academia [385] or that fomepizole may lower glycolate concentrations faster than hemodialysis [124], although the EXTRIP panel strongly cautions against these viewpoints.

Clinical indications
  1. a.

    Coma

    1. i.

      In patients presenting with coma due to EG poisoning, we recommend ECTR (strong recommendation, very low-quality evidence)

  2. b.

    Seizures

    1. i.

      In patients presenting with EG poisoning and seizures, we recommend ECTR (strong recommendation, very low-quality evidence)

  3. c.

    Kidney Impairment

    1. i.

      In patients presenting with EG poisoning and CKD (eGFR < 45 mL/min/1.73 m.2), we suggest ECTR (weak recommendation, very low-quality evidence)

    2. ii.

      In patients presenting with EG poisoning and AKI (KDIGO stage 2 or 3), we recommend ECTR (strong recommendation, very low-quality evidence)

Rationale: AKI is correlated with mortality [27, 29] and kidney impairment reduces endogenous elimination of EG, which is pertinent once antidotal therapy is started, as elimination of EG and metabolites are dependent on functional kidneys. The presence of kidney impairment lowers the EG concentration thresholds for initiating ECTR. Coma and seizures are associated with a poor prognosis and become a clinical justification for ECTR [23, 27, 31, 32, 63, 65, 67, 74]. Mild inebriation may be due to EG early after ingestion and is not an indication for ECTR.

Other clinical manifestations are not recommendations for ECTR; respiratory failure and pulmonary edema would occur after already stated indications for ECTR. Cranial nerve defects may occur long after exposure despite repeated ECTR.

Special populations

The workgroup proposed that these recommendations remain applicable for other populations. A lower EG threshold concentration for ECTR may be applicable in children being treated with ethanol, to minimize adverse effects of ethanol. In pregnancy, ethanol is potentially teratogenic and fomepizole is classed Category C [386]. Although fomepizole was used without complication in the second [387] and third trimesters [388], there may be a preference for lower ECTR threshold to reduce the exposure to these antidotes in the pregnant patient, although ECTR also carries risks in this population. These considerations were discussed with a patient representative who endorsed the recommendations.

Modality

  1. a.

    In patients presenting with EG poisoning requiring ECTR, when all ECTR modalities are available, we recommend using intermittent hemodialysis rather than any other type of ECTR (strong recommendation, very low-quality evidence)

  2. b.

    In patients presenting with EG poisoning requiring ECTR, we recommend using continuous kidney replacement therapy (CKRT) over other types of ECTR if intermittent hemodialysis is not available (strong recommendation, very low-quality evidence)

Rationale: Hemodialysis remains the ECTR that is most widely available so can be initiated quicker than other ECTRs and is also less costly [389]. High-efficiency hemodialysis is also the most efficient ECTR to remove both EG and metabolites, although intermittent hemodiafiltration would be expected to be equally effective. Intermittent hemodialysis was also shown, in methanol poisoning, to correct acidemia quicker than CKRT [390], although clinical outcomes were comparable with both techniques [391]. CKRT is preferred if it can be initiated faster than hemodialysis (e.g., unavailability of hemodialysis, nursing limitations), leading to attainment of a safe EG concentration, as illustrated in Additional file 1: Figure S3. CKRT is preferable if a patient has marked brain edema, as it increases intracranial pressure to a lesser degree than intermittent hemodialysis [392]. Clinicians performing ECTR should optimize operator settings to maximize EG clearance (e.g., higher blood flow, higher effluent production, filters with higher surface area).

In resource-restricted regions where hemodialysis or CKRT cannot be performed, rapid-exchange peritoneal dialysis using non-lactate-based solutions will at least double EG clearance if kidney impairment is present. However, peritoneal dialysis should not replace hemodialysis if the latter is available, as there are many cases of clinical worsening during peritoneal dialysis [150, 158, 393]. Although there are authors suggesting adding peritoneal dialysis to hemodialysis, the EXTRIP panel could not imagine a clinical or kinetic rationale for this, as peritoneal dialysis would only add 5% to the clearance obtained with hemodialysis [52] but at much higher cost and complication rate.

Cessation

  1. a.

    We recommend stopping ECTR when the AG (calculated as Na+ + K+– Cl−– HCO3.−) is < 18 mmol/L (strong recommendation, very low-quality evidence)

  2. b.

    We suggest stopping ECTR when the EG is < 4 mmol/L (25 mg/dL) (weak recommendation, very low-quality evidence)

  3. c.

    We suggest stopping ECTR when acid–base abnormalities are corrected (weak recommendation, very low-quality evidence)

Rationale: Once ECTR is initiated, it is recommended that all acid–base parameters have normalized, in particular a confirmed and sustained normalization of the glycolate concentration and/or anion gap before stopping ECTR. Some clinical markers such as seizures may reverse quickly once ECTR is initiated and others, such as cranial nerve palsies and AKI in particular, may take weeks to resolve and should therefore not guide cessation. Once acid–base homeostasis is restored, then an acceptable and conservative endpoint for cessation is an EG concentration < 4 mmol/L. If ECTR must be terminated when the EG concentration is higher (e.g., multiple poisoned patients), continuation of ADH blockade is reasonable to prevent further EG metabolism. As noted above, some authors have suggested that an EG concentration < 10 mmol/L is an acceptable endpoint in asymptomatic patients [60], but supporting clinical data are not currently available and there are few downsides to continuing ECTR until lower cutoffs are reached. The clinical significance of rebound in EG concentration after ECTR is uncertain but can be addressed with a repeat ECTR session if clinically indicated. Although the EG concentration is linearly correlated with the osmol gap when elevated, the workgroup did not propose a specific osmol gap cutoff for cessation as there are too many imprecisions at low osmol gap values. The use of validated equations based on expected decay of EG in plasma can be used to predict dialysis time [67, 115, 233]. These formulas assume high-efficiency hemodialysis, with good and constant blood flows, and no sampling errors [394]. These formulas may be useful for planning purposes, especially in cases of multiple poisoned patients where ECTR triage is required [67], although the workgroup reiterates that all acid–base abnormalities should be reversed before stopping ECTR.

Miscellaneous

There are various reported approaches to ADH blockade once ECTR is started. Some centers switch from fomepizole to ethanol and others use ethanol in the dialysate bath. Both fomepizole and ethanol are readily dialyzable (see above) and require increased dosage during ECTR (Additional file 1: Table S12). Some centers withhold ADH blockade during ECTR, claiming that ECTR will remove metabolites quickly enough to avoid harm [395]. This appeared to be safe in 5 patients in a retrospective study [123], but more data are required and this approach cannot be currently recommended.

Several EG-poisoned patients have concomitant alcohol use disorder, and are at risk for alcohol withdrawal, especially if ECTR is performed [282]; usual measures should be in place to mitigate this risk. Cerebral hemorrhage is not a common manifestation of EG poisoning (as opposed to methanol) but occasionally occurs [396, 397], and so the decision to anticoagulate the ECTR circuit should be individualized.

EXTRIP strongly advocates for widespread and rapid (2–4 h) hospital availability of glycolate and EG measurements, as recommended by the National Academy of Clinical Biochemistry [398]. These would permit precise diagnosis and targeted treatment, avoiding costly unnecessary treatments [399, 400].

Research gap

Cost and clinical outcome studies should be performed to evaluate the use of fomepizole versus ethanol versus no ADH blockade during ECTR. Further, in patients with no acidosis, it would be useful to determine if the addition of ECTR to ADH blockade significantly reduces cost and/or length of stay. These questions can be ideally answered by prospective randomized trials, but adequately powered studies require multicenter collaborations which can be complicated. Well-conducted single-center studies with complete reporting of relevant clinical outcomes including biomarkers of EG toxicity (for example, changes in acidosis and kidney function), preferably with statistical matching of baseline variables, will also be informative. Of course, cost-effectiveness studies will vary between institutions and health systems, so the development of algorithms that allow for the incorporation of all relevant components and their cost to facilitate local decision-making is anticipated to be informative. Studies should identify what represents “safe” EG and glycolate concentrations to better define initiation and cessation criteria for ECTR and/or ADH blockade. Cases of osmotic demyelination syndrome in those that have a very high concentration of EG (e.g., > 100 mmol/L or > 620 mg/dL) and of “dialysis disequilibrium” during ECTR should be reported to assess the incidence of these phenomena [268, 401]. Cases of EG poisoning treated with exchange transfusion (when no other modality is available) should be reported with proper toxicokinetic data.

Conclusion

The EXTRIP workgroup reviewed the available literature pertaining to the role of ECTR in ethylene glycol poisoning to propose treatment recommendations based on blood tests with and without the use of ADH blockers. The presence of kidney impairment decreased the thresholds for ECTR use. The workgroup recommends using intermittent hemodialysis over other ECTRs, and to evaluate the cost-benefit between fomepizole use vs ECTR for less severe poisoning. The dosage of antidotes needs to be adjusted during ECTR. The workgroup strongly advocates for the widespread availability of resources for the rapid measurement of glycolate and EG concentrations.