Anion and osmolal gaps in the diagnosis of methanol poisoning: Clinical study in 28 patients

Abstract

To evaluate anion and osmolal gaps as diagnostic tools in methanol poisoning.
Clinical observational study.
In a recent methanol outbreak, the initial triage and treatment decisions in 28 patients were based mainly upon the values of the osmolal and anion gaps on admission. Methanol and formate levels were later compared to these gaps by linear regression analysis.
The correlation between the osmolal gaps and serum methanol concentrations on admission was linear (y = 1.03x+12.71, R2 = 0.94). The anion gaps correlated well with the serum formate concentrations (y = 1.12x+13.82, R2 = 0.86). Both gaps were elevated in 24 of the 28 subjects upon admission. Three patients had an osmolal gap within the reference area (because of low serum methanol), but elevated anion gap because of formate accumulation. One patient with probable concomitant ethanol ingestion had a high osmolal gap and a normal anion gap.
Osmolal and anion gaps are useful in the diagnosis and triage of methanol-exposed subjects. Confounders are low serum methanol and concomitant ethanol ingestion.

Full text

Intensive Care Med (2004) 30:1842–1846
DOI 10.1007/s00134-004-2373-7
BRIEF REPORT
Knut Erik Hovda
Odd Helge Hunderi
Nina Rudberg
Sten Froyshov
Dag Jacobsen
Anion and osmolal gaps in the diagnosis
of methanol poisoning:
clinical study in 28 patients
Received: 14 October 2003
Accepted: 3 June 2004
Published online: 8 July 2004
 Springer-Verlag 2004
K. E. Hovda (
)
) · S. Froyshov ·
D. Jacobsen
Department of Acute Medicine,
Ullevaal University Hospital,
0407 Oslo, Norway
e-mail: knuterik.hovda@ulleval.no
Tel.: +47-22-119100
Fax: +47-22-119181
O. H. Hunderi
Department of Medicine,
Ostfold Central Hospital,
1603 Fredrikstad, Norway
N. Rudberg
Department of Clinical Chemistry,
Ullevaal University Hospital,
Oslo, Norway
Abstract Objective: To evaluate an-
ion and osmolal gaps as diagnostic
tools in methanol poisoning. Design
and setting: Clinical observational
study. Patients and methods: In a
recent methanol outbreak, the initial
triage and treatment decisions in 28
patients were based mainly upon the
values of the osmolal and anion gaps
on admission. Methanol and formate
levels were later compared to these
gaps by linear regression analysis.
Results: The correlation between the
osmolal gaps and serum methanol
concentrations on admission was
linear (y = 1.03x+12.71, R
2
= 0.94).
The anion gaps correlated well with
the serum formate concentrations
(y = 1.12x+13.82, R
2
= 0.86). Both
gaps were elevated in 24 of the 28
subjects upon admission. Three pa-
tients had an osmolal gap within the
reference area (because of low serum
methanol), but elevated anion gap
because of formate accumulation.
One patient with probable concomi-
tant ethanol ingestion had a high os-
molal gap and a normal anion gap.
Conclusion: Osmolal and anion gaps
are useful in the diagnosis and triage
of methanol-exposed subjects. Con-
founders are low serum methanol and
concomitant ethanol ingestion.
Keywords Methanol · Formate ·
Diagnosis · Acidosis · Osmolality
Introduction
Methanol poisoning is characterized by increasing meta-
bolic acidosis because of accumulation of the toxic me-
tabolite formic acid, visual disturbances, and respiratory
and cardiovascular failure [1, 2]. Despite effective treat-
ment with alkali, antidotes (ethanol or fomepizole) and
dialysis, patients suffer from complications such as visual
impairment and a Parkinson-like syndrome because of
delayed diagnosis [1, 2]. An important reason for this is
that few hospitals analyze methanol on a 24-h basis.
Formic acid from methanol metabolism results in met-
abolic acidosis with high anion gap (AG) [3]. Ingestion of
alcohols, such as methanol, significantly increases S-os-
molality and the osmolal gap (OG) because of their high
molar concentrations (Fig. 1). The use of these gaps as a
diagnostic tool has not been evaluated in a larger meth-
anol outbreak. Their usefulness has also been questioned
[4]. We evaluated the use of the OG and the AG as a
diagnostic tool in 28 patients from the latest methanol
outbreak in Norway.
Patients and methods
Patients
During the outbreak, 46 patients were admitted to hospital and
confirmed as poisoned by methanol, of whom six died. All con-
sumed illegal spirits consisting of 20% methanol and 80% ethanol.
Most patients were symptomatic upon admission, with dyspnea and
visual disturbances being most common. Treatment was hypertonic
(0.5 mmol/ml) sodium bicarbonate (27 of 28), antidote (ethanol or
fomepizole), and hemodialysis (23 of 28). Three patients died and
five were discharged with permanent visual and/or cerebral se-
quelae.

1843
Methods
Methanol in serum was measured by gas chromatography using a
headspace injector (Fisons GC 8000; Carlos Erba Instruments,
Rodano, Italy) (sensitivity 1.3 mmol/l and day-to-day coefficient of
variation 5%). Calibrators and controls were made by dilution of
100% methanol (Merck, Damstadt, Germany). Formate was mea-
sured enzymatically on a Cobas Mira analyzer (Roche Diagnostics,
Basle, Switzerland) using formate dehydrogenase (Roche) and
nicotinamid adenine dinucleotid (NAD) (Sigma, St. Louis, USA)
(sensitivity 0.1 mmol/l, reference range 0.4 mmol/l, day-to-day
coefficient of variation 5%). Statistical analyses were performed by
the use of linear regression method.
Anion and osmolal gaps
The anion gap (AG) in serum was determined from the equation:
AG ¼ Na
þ
þ K
þ
ðÞCl

þ HCO

3

The reference range is 13€8 mmol/L (mean € 2SD) [5].
The osmolal gap (OG) in serum is the difference between the
measured osmolality (MO) and calculated osmolality, determined
from the following equation [1]:
OG ¼ MO 
1:86 Na þ glucose þ urea
0:93
Osmolality was measured by freezing point depression meth-
od on a Fiske one-ten osmometer (Bergman Diagnostika, Oslo,
Norway). The reference range for the OG is 5€14 mOsm/kgH
2
O
(mean€2SD) [5].
The osmolal contribution from ethanol (four patients) was
subtracted from the measured osmolality. Ethylene glycol was not
detected. As such, these alcohols did not influence the OG pre-
sented here. The osmolal contribution of a 100 mg/dl concentration
of methanol is 34 mOsm/kgH
2
O, ethanol 24 mOsm/kgH
2
O, iso-
propanol 18 mOsm/kgH
2
O, and ethylene glycol 17 mOsm/kgH
2
O.
Results
Many patients (18 of 28) were admitted late with pro-
nounc ed metabolic acidosis (base deficit 20 mmol/l).
The median pH on admi ssion was 7.12 (range 6.57 –
7.50), mean pCO
2
was 3.2 kPa (range 1.3–7.9), mean
HCO
3
-
7.3 mmol/l (range 1.0–28.0), and mean base
deficit 21 mmol/l (range 0–30). Mean S-m ethanol
was 54.3 mmol/l (range 8.4–146.9), mean S-ethanol
7.6 mmol/l (range 2.2–10.9, n=4), mean OG 68 mOsm/
kgH
2
O (rang e 14–159 ), and mean AG 35 mmol/l (range
16–50).
The regression line between S-methanol concentra-
tions and OG on admission (y = 1.03x+12.71, R
2
= 0.94)
showed a very good correlation (Fig. 2). The sensitivity
was, however, poor for the detection of S-methanol con-
centrations below 20 mmol/l (65 mg/dl). The mean ref-
erence value for the OG in this population was 13 mOsm/
kgH
2
O (x = 0).
Hypertonic sodium bicarbonate infusion did not af-
fect the usefulness of the OG as a diagnostic tool; re-
gression line was then: y = 1.02x+4.33, R
2
= 0.96 [n = 47,
Fig. 1 Three stages of methanol
poisoning (theoretical levels).
Early: little formate accumulat-
ed from methanol metabolism
(AG normal); Intermediate:
both methanol and formate
present and both gaps are ele-
vated; Late: little or no metha-
nol left, all metabolized to for-
mate (high AG, normal OG)

1844
time from admission: 1.5–40 h, mean S-methanol
34.5 mmol/l (range 2.8–138.1 mmol/l), mean OG 40
(range 1–140 mOsm/kgH
2
O)].
Figure 3 presents the correlation between several con-
comitantly measured AGs and S-formate levels in eight
patients. The correlation was good (y = 1.12x + 13.82, R
2
= 0.86), indicating a mean reference value of an AG of
14 mmol/l in this population. The factor 1.12 indicates that
the increase in the AG was slightly higher than the in-
crease in the respective S-formate. This was most probably
accounted for by an increasing accumulation of lactate in
the most acidotic patients. In six patients (Table 1), S-
lactate was measured (2.4–14.1 mmol/l); reference range
0.3–1.5 mmol/l. As seen from Table 1, there was a trend
towards more lactate in the most acidotic patients.
Discussion
Formic acid is responsible for metabolic acidosis in the
early stage of methanol poisoning [3]. Later, formate in-
hibition of cytochrome oxidase may explain the lactate
production reported in some late admitted cases [1] and in
our patients with the most severe metabolic acidosis
(Table 1). In contrast to a small series of patients where
formate accumulation alone accounted for the increased
AG [3], the present series of patients were more seriously
poisoned as judged from clinical features and their more
pronounced metabolic acidosis. The present patients were
admitted late because of their own symptoms, while in the
other series the patients were admitted early because of
recommendations through other media, most before
clinical symptoms had developed [3, 6]. This fact, com-
bined with more alcohol abuse among the present pa-
tients, may explain the difference in anion composition,
with more lactate in our patients [7]. Only one of our
patients had a normal AG, probably because of early
admission and/or concomitant ethanol ingestion.
Methanol metabolism is slow (about 2.5 mmol·l·h or
8 mg·dl·h) [8] and symptomatic acidosis therefore does
not develop before 15–20 h after ingestion, and even later
if the antidote ethanol is co-ingested [2]. Latency periods
as long as 96 h have been reported in such patients [6]. In
the present outbreak, a mixture of methanol in ethanol
was ingested, thus increasing the latency period even lon-
ger than this. Diagnosis based on the history was therefore
difficult to obtain.
Most drugs do not raise the OG significantly because of
their low molar concentrations [2]. However, if ethanol is
also ingested in methanol-poisoned patients, the osmolal
contribution from ethanol must be subtracted (see meth-
ods). As methanol metabolism proceeds, the OG decreases
and the AG increases (formic acid accumulates; Fig. 1).
Fig. 2 S-methanol versus OG upon admission in 28 patients poi-
soned with methanol. Osmolality from ethanol subtracted (see
methods)
Fig. 3 S-formate versus AG in methanol poisoned patients with
elevated S-formate levels

1845
An elevated OG was found in 25 of our patients. The very
good correlation between the S-methanol concentration
and the OGs (Fig. 2) was still present after start of bi-
carbonate treatment. The diagnostic value of the OG was
therefore not limited to the samples taken upon admission.
The three patients without an elevated OG all had low S-
methanol concentrations ( 15.6 mmol/l or 50 mg/dl) and
no S-ethanol. They were all acidotic with high AGs, re-
flecting the accumulation of a significant amount of for-
mate from methanol metabolism (Fig. 1).
Several conditions in medicine increase the AG or the
OG [2], but except for methanol or ethylene glycol poi-
soning, few conditions increase both at the same time.
The few exceptions previously reported include diabetic
coma [9], acidosis in alcoholics [10], chronic renal failure
[11] and patients with shock following major trauma [12].
However, these reports were published before the new
reference range of OG in acutely admitted patients was
proposed (9 to 19 mOsm/kgH
2
O) [5]. Patients with
chronic renal failure and alcoholic ketoacidosis will, ac-
cording to this, have OGs within the reference range
[10, 11]. In diabetic ketoacidosis, S-acetone may rise
to 13 mmol/l [9], an osmolal contribution of 13/0.93 =
14 mOsm/kgH
2
O, which will hardly raise the OG above
30 mOsm/kgH
2
O. In these patients a high S-glucose will
also point to the diagnosis.
In patients with metabolic acidosis of unknown origin,
we propose a new decision level for the OG of 25 mOsm/
kgH
2
O before considering therapeutic interventions with
antidotes, provided that the osmolal contribution from
ethanol is subtracted. This approach reduces sensitivity,
but increases specificity. The high cost of fomepizole also
requires a high degree of specificity concerning the di-
agnosis of methanol poisoning. The proposed algorithm
can be used in order to simplify the understanding of
diagnosis and triage of methanol poisoning (Fig. 4).
Patients with suspected methanol intoxication and
OG<25 (below the decision level), can be separated into
two categories (Fig. 4): patients with a small intake of
methanol, or late admitted patients who have already
metabolized most of the methanol to formic acid. The first
category are not severely poisoned and may not be can-
didates for fomepizole treatment (if necessary, ethanol can
be used). In the second group, the diagnosis of methanol
poisoning may be supported by history, symptoms (visual
disturbances), findings (hyperventilation and pseudopa-
pilitis) [2], and always an elevated AG. In our material, all
seven patients with an OG<25 (Table 2) belonged to this
latter category, whereas six had visual disturbances [2].
In conclusion, the use of the AG and OG in patients
presenting with metabolic acidosis of unknown origin
helps in diagnosing methanol (or ethylene glycol) poi-
soning at an early stage where effective treatment may
still reduce morbidity and mortality. The proposed deci-
sion level for the OG (25 mOsm/kgH
2
O) improves di-
agnostic specificity.
Table 1 Acid/base-status on admission in six methanol-poisoned patients sorted after increasing AG and increasing formate concentrations (first admissi on to local hospital in
parenthesis in case 1) (CP chest pain, VD visual disturbances, GI GI-symptoms, D dyspnoea, BP back pain, VS visual sequela, CS cerebral sequela)
Case Methanol
(mmol/l)
Ethanol
(mmol/l)
OG (mOsm/
kgH
2
O)
AG
(mmol/l)
Formate
(mmol/l)
Lactate
(mmol/l)
pH pCO
2
(kPa)
Base deficit
(mmol/l)
HCO
3
-
(mmol/l)
Clinical
features
Sequelae
1 73.4 28.3
a
104 14 2.3 2.4 7.37 5.1 3 21.5 None None
(102.8) (0) (101) (24) (6.9) (-) (7.33) (3.9) (9) (15) None
2 140.6 8.7 138 23 3.3 3.4 7.50 4.8 5 28.0 None None
3 8.4 8.7
a
24 28 11.7 3.6 7.26 2.5 17 8.1 CP, VD, GI, D VS
4 77.5 0 113 39 15.7 12.7 6.60 6.1 28 4.5 CP, D CS
5 15.6 0 16 40 20.4 7.1 6.92 1.9 30 2.8 CP, VD, D None
6 32.5 0 61 40 21.0 14.1 6.87 2.9 29 3.8 VD, D, BP VS, CS
a
Measured after treatment with ethanol i.v.

1846
References
1. Barceloux DG, Bond GR, Krenzelok
EP, Cooper H, Vale JA (2002) Ameri-
can Academy of Clinical Toxicology
practice guidelines on the treatment of
methanol poisoning. J Toxicol Clin
Toxicol 40:415–446
2. Jacobsen D, McMartin KE (1997) An-
tidotes for methanol and ethylene glycol
poisoning. J Toxicol Clin Toxicol
35:127–143
3. Sejersted OM, Jacobsen D, Ovrebo S,
Jansen H (1983) Formate concentra-
tions in plasma from patients poisoned
with methanol. Acta Med Scand
213:105–110
4. Hoffman RS, Smilkstein MJ, Howland
MA, Goldfrank LR (1993) Osmol gaps
revisited: normal values and limitations.
J Toxicol Clin Toxicol 31:81–93
5. Aabakken L, Johansen KS, Rydningen
EB, Bredesen JE, Ovrebo S, Jacobsen D
(1994) Osmolal and anion gaps in pa-
tients admitted to an emergency medi-
cal department. Hum Exp Toxicol
13:131–134
6. Jacobsen D, Jansen H, Wiik-Larsen E,
Bredesen JE, Halvorsen S (1982)
Studies on methanol poisoning. Acta
Med Scand 212:5–10
7. Hojer J (1996) Severe metabolic aci-
dosis in the alcoholic: differential di-
agnosis and management. Hum Exp
Toxicol 15:482–488
8. Jacobsen D, Webb R, Collins TD,
McMartin KE (1988) Methanol and
formate kinetics in late diagnosed
methanol intoxication. Med Toxicol
Adverse Drug Exp 3:418–423
9. Sulway MJ, Malins JM (1970) Acetone
in diabetic ketoacidosis. Lancet 2:736–
740
10. Cooperman MT, Davidoff F, Spark R,
Pallotta J (1974) Clinical studies of al-
coholic ketoacidosis. Diabetes 23:433–
439
11. Sklar AH, Linas SL (1983) The osmolal
gap in renal failure. Ann Intern Med
98:481–482
12. Boyd DR, Folk FA, Condon RE, Nyhus
LM, Baker RJ (1970) Predictive value
of serum osmolality in shock following
major trauma. Surg Forum 21:32–33
Fig. 4 Algorithm for diagnosis
and triage in suspected metha-
nol poisoning based on the dif-
ferent combinations of gaps
(osmolal contribution from eth-
anol subtracted). The principles
of the algorithm may also be
also valid in ethylene glycol
poisoning. For details in treat-
ment, see [1, 2]. (VD visual
disturbances, HD hemodialysis)
Table 2 Seven patients with OG below the decision level upon admission (CP chest pain, VD visual disturbances, GI GI-symptoms, D
dyspnoea, F fatigue, VS visual sequela)
Case Methanol
(mmol/l)
Ethanol
(mmol/l)
OG (mOsm/
kgH
2
O)
AG
(mmol/l)
pH pCO
2
(kPa)
Base deficit
(mmol/l)
HCO
3
-
(mmol/l)
Clinical
features
Sequelae
3 8.4 8.7
a
24 28 7.26 2.5 17 8 CP, VD, GI,
D
VS
7 9.4 0 15 38 7.18 1.7 21 5 VD, GI none
8 12.5 0 20 32 7.22 2.8 21 6 VD, D, F none
9 15.6 0 14 39 7.25 2.2 20 7 D none
10 15.6 0 21 34 7.12 1.6 25 4 VD, GI, F none
5 15.6 0 16 40 6.92 1.9 30 3 CP, VD, D none
11 21.9 2.2 24 30 7.12 2.4 22 6 VD, GI none
a
Measured after treatment with ethanol i.v.