Hunting down a double gap metabolic acidosis

Abstract

Double gap metabolic acidosis occurs in the setting of unmeasured active osmoles in the serum (osmolal gap) and anion gap (AG) metabolic acidosis. We describe a 67-year-old woman with acute respiratory failure on mechanical ventilator from pneumonia and anuric acute on chronic renal failure (urea nitrogen 21.4 mmol/L, creatinine 530.4 μmol/L) requiring haemodialysis (HD). On hospital day 5, she was found to have progressive metabolic acidosis (serum pH 7.16, PCO₂ 4.38 kPa, HCO₃ ⁻ 12.1 mmol/L and AG 21 mmol/L). There was no evidence of hypoxaemia, hypoperfusion or haemodynamic instability. Normal serum ketone and l-lactate but high serum osmolal gap (89.4 mmol/kg) was detected. A search for toxic alcohols revealed a high serum propylene glycol (PG 32.9 mmol/L), a stabilizing solvent for intravenous formulations of lorazepam, which was being used as sedation for mechanical ventilation. Unexpectedly, serum l- and d-lactate as metabolites of PG were not elevated. Although extended HD for eight hours completely removed serum PG and the osmolal gap, the predialysis high AG metabolic acidosis persisted, potentially related to hypercatabolism and anuric renal failure. PG should be in the differential diagnosis of the disorders with high osmolar gap and may not always be associated with l- or d-lactic acidosis.

Introduction

A high anion gap (AG) and osmolal gap (double gap) metabolic acidosis is an important clue of toxic alcohol intoxication, such as methanol and ethylene glycol. Propylene glycol (PG, 1,2-propanediol), a less-recognized alcohol, is believed to be innocuous and has been widely used as a solvent and preservative in cosmetics, foods and pharmaceuticals, including intravenous lorazepam. Although lorazepam is the drug of choice for maintenance sedation in intensive care units (ICU) due to its appreciated safety, high-dose continuous infusion (>0.1 mg/kg/h) have been implicated in the development of PG accumulation and toxicity. ^1–4

PG is synthetically prepared from propylene and glycerol as a racemic mixture of d- and l-isomers and will be metabolized to d- and l-lactic acid. Although l-lactate is the major anionic metabolite of PG, both d- and l-lactate may contribute to PG-related AG metabolic acidosis. ⁵ However, the well-established relationship between lactate and PG-associated high AG metabolic acidosis may lead to causes other than lactate being overlooked. Herein, we present a patient with anuric renal failure and double gap metabolic acidosis after five days of high-dose intravenous lorazepam. Quantitative measurement of PG metabolites and direct measurement of serum d- and l-lactate concentration clearly indicated that PG only accounted for the high serum osmolal gap without the resultant d- or l-lactic acidosis.

Case report

A 67-year-old woman presented to the emergency department with shortness of breath and fever for one day. She had poorly controlled hypertension for 10 y and a two-year history of stage 4 chronic renal insufficiency. On admission, her blood pressure was 118/76 mmHg, heart rate 102 beats/min, respiratory rate 24/min, body temperature 38.2°C and body weight 50 kg. Physical examination showed jugular venous distention, use of accessory muscles for respiration, diffuse inspiratory crackles and grade II pitting oedema. The remainder of the physical examination was unremarkable. Laboratory studies showed a white cell count of 14.0 × 10⁹/L, haemoglobin 95 g/L and platelets 146 × 10⁹/L. Serum biochemistries demonstrated acute renal failure (urea nitrogen 21.4 mmol/L, creatinine 530.4 μmol/L) and normal liver function (Table 1). Her chest X-ray showed bilateral pneumonia in the lower lobes and abdominal ultrasonography showed bilateral atrophic kidneys. The patient was intubated for acute respiratory failure and lorazepam continuous infusion uptitrated to 10 mg/h to achieve adequate sedation for synchronized ventilation. Ticarcillin/clavulanate 2.0 g every 12 h intravenously was begun for treatment of pneumonia. She also underwent haemodialysis (HD) on days 2 and 3 of hospitalization for acute renal failure.

Table 1

Serial serum laboratory investigations

Serum concentration	Reference range	Day 1	Day 5	Day 6	Day 7
Na+	136–145 mmol/L	129	133	135	134
K+	3.5–5.1 mmol/L	4.0	4.7	2.5	4.6
Cl−	98–107 mmol/L	99	98	100	99
Urea nitrogen	2.5–7.2 mmol/L	15.8	32.8	6.8	30.0
Creatinine	44.2–88.4 μmol/L	503.9	645.3	141.4	335.9
pH	7.35–7.45	7.41	7.23	7.34	7.29
HCO3 −	24–26 mmol/L	21.0	11.9	18.6	13.8
Anion gap	10–14 mmol/L	9.0	23.1	15.7	19.2
Measured osmolality	285–295 mmol/kg	–	395	292	307
Osmolal gap*	5–10 mmol/kg	–	89.4	9.0	10
Propylene glycol	0 mmol/L	–	32.9	0	0
l-Lactate	0.7–2.1 mmol/L	–	1.9	0.2	0.4
d-Lactate	<0.1 mmol/L	–	0.04	0	0

BUN, blood urea nitrogen

*Osmolal gap = measured plasma osmolality − calculated osmolality = measured osmolality − (2 × [Na⁺] + [BUN]/2.8 + [glucose]/18) ⁶

By the fifth hospital day, her pneumonia improved markedly, but high AG metabolic acidosis (pH 7.16, PO₂ 14.9 kPa, PCO₂ 4.4 kPa, HCO₃ ⁻ 12.1 mmol/L, AG 21 mmol/L) (Table 1) was noticed without hypoxaemia, tissue ischaemia or severe sepsis. A work-up for the excessive anions including l-lactate, β-hydroxybutyrate, acetate and salicylate was unrevealing. An extremely high plasma osmolal gap (89.4 mmol/kg) was found without ingestion of ethanol, methanol or ethylene glycol. She had been receiving a continuous intravenous infusion of lorazepam at 10 mg/h for five days and each vial (2 mg/mL) contains 0.8 mL (10.9 mmol/mL) of PG. Her PG concentrations which was measured by gas chromatography reached 32.9 mmol/L. In light of her normal serum l-lactate, its stereoisomer, d-lactate, in the serum was measured by enzymatic method of d-lactic dehydrogenase and was also normal. The lorazepam was replaced by midazolam, non-PG-containing sedatives. Extended HD with mean blood flow 300 mL/min and dialysate flow 500 mL/min for eight hours effectively eliminated the serum osmolal gap and PG with the extracorporeal PG clearance estimated to be 130 mg/min (1.7 mmol/min) (Figure 1). However, high AG metabolic acidosis still persisted the following day and worsened two days later (Table 1). The patient eventually died from ventilator-associated pneumonia after a two-month hospital stay.

Figure 1

Efficacy on removal of propylene glycol by long-extended haemodialysis

Discussion

This patient developed double gap metabolic acidosis following the continuous administration of high-dose PG-containing lorazepam (0.2 mg/kg/h, containing PG 54.5 mmol/h) for five days. PG overdose was considered after the exclusion of ethanol, methanol and ethylene glycol intoxication. The total accumulated dose of PG was estimated to be at least 2616 mmol (54.5 mmol × 24 h × 2 d) within the two days from the last HD to the measurement of PG because HD presumably is effective in the clearance of PG. ⁷ Her serum PG of 32.9 mmol/L represented an osmolality of approximately 33 mmol/kg, which did not fully account for the osmolal gap 89.4 mmol/kg. In addition to PG, other solvents and unmeasured intermediate metabolites such as methylglyoxal or lactaldehyde may in part contribute to her osmolal gap. ⁸ Extended HD for eight hours effectively eliminated the serum PG and osmolal gap. However, another hidden cause was responsible for the worsening AG metabolic acidosis after dialysis because her serum d-and l-lactate concentration were normal.

Intravenous lorazepam contains lorazepam at a concentration of 2 mg/mL suspended in PG (80%, molecular weight [MW] 76 Da), polyethylene glycol-400 (PEG-400, 18%, MW 400 Da) and benzyl alcohol (2%, MW 108 Da). PG contributes 23 and 57 times as many osmotically active particles as PEG-400 and benzyl alcohol and exerts toxicity at a concentration exceeding 2.36 mmol/L. ⁹ In general, a maximal PG dose of 1 g/kg/d (13.2 mmol/kg/d) is presumed to be safe in the absence of risk factors. The half-life of PG is 1.4–3.3 h in normal individuals. The elimination kinetics of PG is a saturable capacity-limited metabolism, indicating first-order elimination for low-to-moderate plasma concentration but zero-order for high concentration. ¹⁰ Approximately 12–45% of PG is excreted unchanged in urine. The rest is oxidized to l- or d-lactaldehyde and then l- or d-lactic acid via hepatic alcohol dehydrogenase and aldehyde dehydrogenase (ADH, ALDH) (Figure 2). ⁹ Children younger than four years, pregnant woman, patients with impaired renal or hepatic function and those treated with disulphiram are prone to PG toxicity because of either diminished renal PG excretion or low enzymatic activity of ADH or ALDH. ¹¹ As in this case, impaired renal function increases the risk of PG intoxication and a 50% reduction in maximum dose is recommended in the presence of risk factors.

Figure 2

Metabolism of propylene glycol (propanediol)

PG toxicity has been hypothesized to result from direct cytotoxicity of the parent compound and its metabolites. A significant correlation between cumulative lorazepam dose and end-of-therapy PG concentration has been demonstrated in paediatric ICU and in adult patients on high-dose lorazepam infusion for more than 48 h. ^11,12 Serum PG concentrations showed considerable interpatient variation with respect to the rate of infusion and accumulated dose in adults (Table 2). ^9,13–22 We analysed the prior cases in the literature and found no significant correlation between the serum PG concentration and daily dose of PG, serum lactate concentration or AG. As seen in Figure 1, the serum osmolal gap trended together with PG concentration during HD and may be a sensitive and convenient index to assess the degree of PG intoxication.

Table 2

Literature review of intravenous lorazepam-induced propylene glycol intoxication

Reference	Age (yrs)	Accumulated dose of PG (mmol)	Infusion rate of PG (mmol/day)	Serum PG (mmol/L)	Osmolal gap (mmol/kg)	Anion gap (mmol/L)	Serum lactate (mmol/L)	Liver/renal function	Note
Arbour R, et al. 9	60	3947	7.9	10.3	21	12	3	Normal
Arbour R, et al. 13	45	78724	26.2	22.3	36	11	NS	Normal
Reynolds HN, et al. 14	36	∼15789	22.4	NS	25	NS	4.7	Normal
Parker MG, et al. 15	34	∼39447	16.4	68.4	97	18	2.7	Renal failure	HD
Al-Khafaji AH, et al. 16	28	NS	28.4	17.2	38	20	1.2	Renal failure	CVVHD
Tayar J, et al. 17	34	9263	28.6	21.7	165	17.8	4.5	Renal failure	HD
Tuohy KA, et al. 18	56	2355	11.8	61.4	32	15	NS	Normal
Yaucher NE, et al. 19	48	11145	3.3	33.4	64.2	13	NS	Liver dysfunction
	67	10487	22.9	10.3	12	10	NS	Normal
	52	9829	24.6	45.4	49	13	3.0	Normal
	39	5237	27.9	3.5	33	11	1.9	Liver dysfunction
Hayman M, et al. 20	46	7263	6.9–10.3	3.9	15	NS	2.4	Renal failure	CVVHD
Wilson KC, et al. 21	34	4026	20.1	23.0	57	16	NS	Alcohol abuse
	60	6724	22.4	7.6	41	14	2.3	Alcohol abuse
	35	6447	32.2	NS	35	12	NS	Alcohol abuse
Neale BW, et al. 22	24	67987	65.5	144.7	186	15	4.7	Nil
This case	67	2616	12.6	32.9	89	23	1.9	Renal failure	HD

CVVHD, continuous veno-venous hemodialysis; HD, hemodialysis; NS,not shown; PG, propylene glycol

The metabolic acidosis with normal serum l-lactate concentration in this patient indicates that significant lactate production from PG metabolism was not the cause of AG metabolic acidosis. PG is a racemic mixture of d-, l-1,2-propanediol (d/l ratio 1:1) and can be metabolized to either d- or l-lactate. One previous case with intentional ingestion of PG-containing gel developed d- but not l-lactic acidosis. ⁵ However, the rate of conversion to l-lactate by lactate dehydrogenase is five times more rapid than d-lactate by d-2-hydroxyacid dehydrogenase. ²³ In this patient, the normal l- and d-lactate concentrations pointed to possibly other underlying causes of AG metabolic acidosis.

Other unmeasured anionic metabolites, like pyruvate, α-ketoglutarate, sulphate and phosphate, may contribute to the high AG metabolic acidosis in this patient. Because this patient was dialysed with conventional alkali dialysate, endogenous acid production from protein oxidation during the interdialytic period rather than intradialytic factors was a potentially important contributor for her AG metabolic acidosis. ²⁴ Protein hypercatabolism in the face of severe critical illness may be a significant source of increased endogenous acid production in this patient who had no ability to excrete acid renally.

In conclusion, PG is an important cause of high osmolal gap but not necessarily high AG metabolic acidosis. The serum osmolal gap is a reliable surrogate for monitoring and assessing retained PG. Whether d- or l-lactate accounts for the AG depends on the accumulated dose of PG and metabolic capacity of liver and kidneys. One should not forget other concomitantly possible causes of metabolic acidosis, especially in critically ill patients.

DECLARATIONS

Competing interests: None.

Funding: None.

Guarantor: S-HL, MD.

Contributorship: MTY helped with the clinical discussion of the case as well as with the references. CJC established the first clinical contact with the patient and provided the clinical data. TC helped revise the final version of this manuscript. S-HL is the main author of this report and conducted most of the bibliographical search and coordination of the discussion. There are no conflicts of interest and no funding.

Acknowledgements: None.