CLIN. CHEM. 25/7, 1329-1330 (1979)
Diabetic Ketoalkalosis
John Koett,1 James Howell,1 Steven Steinberg,2 and Paul Wolf3
The usual metabolic derangement inuncontrolled
diabetes
mellitus is metabolic acidosis, with an increase in the anion
gap because of increased ketoacids and lactate. However,
diabeticketoalkalosismay occasionallybe encountered,
the prominent clinical
featureof which isvomiting,with
depletion of potassium, chloride,and hydrogen ions.
Self-medication with absorbable alkali
may also contribute
to the alkalosis. It would be dangerous to treathyperglycemic patients with alkali if their condition is ketoalkalosis
instead of ketoacidosis.
The metabolic distortion in uncontrolled diabetes mellitus
is usually metabolic acidosis, with an increase in the anion gap
due to increased ketoacids and lactic acid; however, diabetic
ketoalkalosis may occasionally be encountered (1). The
prominent clinical feature of diabetic ketoalkalosis is vomiting, which causes depletion of potassium, chloride, and hydrogen ions. Self-medication with alkali may also contribute
to the alkalosis. It would be dangerous to treat hyperglycemic
patients with alkali if their condition is ketoalkalosis instead
of ketoacidosis. Thus, arterial blood-gases must be routinely
determined for all patients with diabetic ketosis, to ascertain
what replacement therapy is appropriate. Ketoalkalosis in
uncontrolled diabetes is rare. We report such a case.
A 54-year-old woman, an adult-onset diabetic, was admitted
to the hospital with uncontrolled diabetes. She had been doing
well taking 16 USP units of Neutral Protamine Hagedorn and
8 USP units of regular insulin for her diabetes. Four weeks
before admission she developed a persistent cough. Three days
before admission she noted polyuria, polydipsia, blurring of
vision, anxiety, and rapid breathing. She had essential hypertension, which was being treated with 40mg of furosemide
twice daily and 0.1 mg of clonidine#{149}HC1
(Catapres) twice
daily.
Physical examination
woman who
blood pressure was
152/76 mmHg, temperature 35.8 #{176}C,
pulse 108 beats/mm, and
respiratory rate 32/mm. The head and neck were normal,
except for a dry mouth. The chest was normal on auscultation
and percussion. The heart was tachycardic, with no murmurs.
The abdominal examination revealed no masses, epigastric
tenderness, nor organomegaly. Results of neurological examination were entirely normal.
Laboratory
results on admission:
Hematocrit 47%, glucose
9580
mg/L, calcium
117 mgfL,
K 2.5 mmol/L,
HCO3- 38
mmol/L, uric acid 120 mgfL, serum urea nitrogen 650 mg/L,
Na 123 mmolfL, lactate 1 mmol/L, Cl 56 mmol/L, anion gap
35 mmol/L, pH 7.66, pco2 23 mmHg, po2 65 mmHg. Urine:
was anxious
showed a well-developed
and hyperventilating.
Her
Departments of Laboratory Medicine and 2 Medicine, Naval
Regional Medical Center, San Diego, CA 92134.
2 Department of Pathology, University of California, La Jolla, CA
92093.
The opinions or assertions herein are the private views of the authors and are not to be construed as official or as reflecting the views
of the Department of Defense or the Department of the Navy.
Received March 12, 1979; accepted May 7. 1979.
relative density 1.015,4+ glucose, no protein,
pH 5.0,sediment
negative, no ketones, Na 40 mmol/L, K 46 mmol/L, and Cl 26
mmol/L.
The test for ketones in serum collected 2 h after
admission was positive ma twofold dilution of the serum with
isotonic saline; the semiquantitative test for urine ketones
showed 1000 mg/L, that for acetoacetate 100 mg/L.
Clinical course in hospital: The patient was admitted to
the Medical Intensive Care Unit and treated with regular
crystalline zinc insulin and rehydration with isotonic saline
and potassium supplementation. Assays on samples of arterial
blood collected 1 h later showed a pH of 7.84, p02 of 87 mmHg,
and Pco2 of 16 mmHg. The patient then was begun on a rebreathing mask and an hour later her arterial blood gaseswere
pH 7.80, pco2 19 mmHg, and po2 188 mmHg (Figure 1). Because the patient was extremely agitated and in both metabolic and respiratory alkalosis, she was given a tranquilizer
(Valium), with subsequent improvement in the alkalosis, the
arterial Pco2 increasing to 30 mmHg. Her uncontrolled diabetes stabilized with continuing infusion of regular crystaffine
zinc insulin. She was transferred to a regular medical ward,
placed on a daily regimen of 16 USP units of Neutral Protamine Hagedorn and 8 USP units of regular crystalline zinc
insulin. Blood glucose in a sample collected while the patient
was fasting was 1410 mg/L. Furosemide and Catapres were
not resumed.
Discussion
Diabetic ketoalkalosis has been postulated to be secondary
to development of a fluid-volume-contraction
alkalosis, initiated by vomiting in a diabetic patient. In the few reported
cases, this has been a relatively common antecedent feature
of this rather unusual metabolic pattern (2, 3). The patient
we describe was an insulin-dependent adult-onset diabetic
with associated hypertension, which was being treated with
Catapres and furosemide. In contrast to other patients, this
patient’s vomiting was inconsequential in amount and her
volume contraction was most likely secondary to her therapy
with diuretics and the osmotic diuresis induced by hyperglycemia. Her decreased sodium, potassium, and chloride were
probably secondary tothe therapy with a diuretic and, in part,
secondary to the mild vomiting. The hypokalemia was probably sustained by the contraction alkalosis with hypochloremia. In the presence of a low serum chloride concentration,
sodium is not reabsorbed in the proximal tubules and is presented to the distal tubule, where it can exchange for potassium ions as well as hydrogen ions. This is reflected in our
patient by the relatively high urinary sodium (40 mmol/L)
with a concomitant K+ value of 46 mmol/L. Moreover, with
volume depletion, secondary hyperaldosteronism is induced,
with subsequent exchange of sodium ions for potassium ions
as well as hydrogen ions in the distal tubules. The end result
of this interplay of factors is the maintenance of an alkalotic
state. The patient’s sodium of 123 mmol/L can readily be
explained on the basis of sodium loss from diuretic therapy
as well as the hyponatremia caused by the osmotic effect of
the above-normal glucose.
An interesting phenomenon in our patient was theincreased
CLINICAL CHEMISTRY,
Vol. 25, No. 7, 1979
1329
7.62
750
traction alkalosis, with simultaneous
impairment
of electron
transport.
Under these circumstances
NADH, which is gen-
7.41
erated early in the Embden-Meyerhof
40
T
.
E
20 ‘
S 2C
10
10
pathway through the
oxidation of 3-phosphoglyceraldehyde
to 1,3-diphosphoglycerate, must be regenerated.
This regeneration
of NADH
usually occurs by the reduction of pyruvate to lactate. If this
indeed occurs in these individuals
with severe contraction
alkalosis, then one would expect that lactate concentrations
would be increased. However, lactate values measured in individuals with diabetic ketoalkalosis have been normal, as in
our patient. The origin of the large unaccountable
anion gap
therefore
remains
an enigma.
However, certain possibilities should be considered in future
.1
91011
2345678
HOURS
URINE:
pH
k.lOflS
Na/K/Ce
SERUM:
5
7
fl9
g,5’
AFTER
121314t5161718192021222324
AOMISSON
75
nig
40/48/26
k.ton.
Fig. 1. Blood-gas
.1:2
n.g
data obtained during the treatment
of the dia-
betic ketoalkalosis
Electrolyte values are in mrnol/L
anion gap with marked alkalosis instead of an acidosis. Usually
a large anion gap is associated with an increase in the weak
base of an organic acid, as is seen in lactic acidosis. In contrast,
our patient, who had a large anion gap, had prominent alkalosis and a normal lactate concentration.
One should consider
the presence of several organic anions that might be produced
in diabetic ketoalkalosis,
which theoretically
could explain
this aberration.
Initially,
the urinalysis was negative for ketones. Serum ketones were not measured on admission but
presumably also were negative, because no ketones were detected in the urine. Serum ketones, measured 2 h after admission, were positive (ketones 1000 mg/L, acetoacetate 100
mg/L). On admission,
the patient was somewhat volume
contracted,
with decreased peripheral
oxygenation
leading
to presumably
increased (not measured) beta-hydroxybutyrate from acetoacetate. This would result from a shift of the
redox state of the cell with increased reductive potential secondary to the loss of the electron acceptor (oxygen) as peripheral perfusion is compromised.
With re-expansion of her
vascular volume, the shift in her redox state would favor the
conversion of beta-hydroxybutyrate
to acetoacetate
as intracellular concentrations
of NADH decrease with improving
electron transport as oxygen is made available at the cellular
level. This is reflected by the positive test for serum acetone
2 h later. By the conventional
methods of measuring serum
ketones, only acetoacetate
and ketones are measured, not
beta-hydroxybutyrate.
Beta-hydroxybutyrate
possibly
is
substantially
increased in these patients as a consequence of
decreased peripheral
perfusion induced by the volume contraction. However, it is unlikely that it has any buffering effect
because its pKa is approximately
4.4. We could find little reported information
on this.
Undoubtedly,
peripheral oxygenation
is decreased in con-
1330
CLINICAL CHEMISTRY,
Vol. 25, No. 7, 1979
investigations,
the first being that hydrogen ions generated
by the production of lactate are maintained at an absolute
minimum by the sustained alkalosis resulting from the volume
contraction.
This possibility is reflected in our patient by her
low urinary pH. Secondly, the relatively normal lactate values
encountered
in diabetic ketoalkalosis
may be secondary to a
high rate of utilization of lactate in the Con cycle, enhanced
by the above-normal
gluconeogenic hormone concentrations.
There is decreased production of pyruvate through glycolysis,
because glucose entry into the cellular compartment
in certain
tissues would be restricted in the absence of sufficient insulin.
Thirdly, the high reductive state of the body cells secondary
to loss of oxygen (terminal
electron sink) would favor the
conversion of acetoacetate to beta-hydroxybutyrate
as pyruvate becomes limiting as a reducible substrate. This would
lower the measurable serum acetone, and may be responsible
for the large anion gap seen in these patients.
The pco2 was 16 mmHg (hypocapnea), which is consistent
with respiratory
alkalosis. However, the measured serum bicarbonate was 36 mmol/L,
which is inconsistent
with the
measured pco2 (16 mmHg) and is more consistent with a
metabolic alkalosis. By the Henderson-Hasselbalch
equation,
the calculated serum bicarbonate would be 26 mmolfL if this
patient had a pure respiratory
alkalosis with a pco2 of 16
mmHg. Thus, she evidently had a combined metabolic and
respiratory alkalosis. The respiratory alkalosis component was
related to anxiety and to the presence of a reactive catecholamine excess secondary to the volume-contraction
state that
caused hyperventilation.
Thus, we emphasize that the diabetic patient who presents
with signs and symptoms of diabetic ketoacidosis should be
quickly evaluated for arterial pH before therapy is begun,
because alkali would be hazardous in diabetic
ketoalkalosis.
References
I. Bleicher, S., Keto- is not always acidosis:
relevant. Diabetes Outlook 3,3-4 (1967).
“Heartburn”
can he
K. C.,and Walsh, C. H., Diabetic ketoalkalosis: A readily
misdiagnosed entity. Br. Med. J. 2, 19 (1976).
:1. Roggin, G. M., Moses, D., Kautcher, M., et al., Ketosis and metaholic alkalosis in a patient with diabetes. J. Am. Med. Assoc., 211,296
(1970).
2. Lim.