- Email: [email protected]
The syndrome of alcoholic ketoacidosis
The Syndrome of Alcoholic Ketoacidosis KEITH D. WRENN, M.D., COREY M. SLOVIS, M.D., ~fh-&. Georgia, and&Chester, New York, GREGG E. MINION, M.D., ROMAN RUTKOWSKI, Ph.D., Atlanta, Georgia
PURPOSE:To further
elucidate the clinical ketoacidosis (AKA). PATIENTSANDMETHODS: Acaseseriesof74patie& with AKA defiied as a wide anion gap metabolic acidosis unexplained by any other disorder or toxin, including any patient with a history of chronic alcohol abuse. The setting was the Medical Emergency Department at Grady Memorial Hospital in Atlanta, Georgia, a university-affi’jted inner-city hospital. RESULTS: AKA is a common disorder in the emergency department, more common than previously thought. The acid-base abnormalities are more diverse than just a wide-gap metabolic acidosis and often include a concomitant metabolic alkalosis, hyperchloremic acidosis, or respiratory alkalosis. Lactic acidosis is also common. Semiquantitative serum acetoacetate levels were positive in 96% of patients. Elevated blood alcohol levels were present in two thirds of patients in whom alcohol levels were determined, and levels consistent with intoxication were seen in 40% of these patients. Electrolyte disorders including hyponatremia, hypokalemia, hypophosphatemia, hyperglycemia, hypocalcemia, and hypomagnesemia were common on presentation. The most common symptoms were nausea, vomiting, and abdominal pain. The most common physical fmdings were tachycardia, tachypnea, and abdominal tenderness. Altered mental status, fever, hypothermia, or other abnormal fmdings were uncommon and reflected other underlying procemes. CONCLUSIONS: AKA is a common disorder in chronic malnourished alcoholic persons. The acid-base abnormalities reflect not only the ketoacidosis, but also associated extracellular fluid volume depletion, alcohol withdrawal, pain, sepsis, or severe liver disease. Although the pathospectrum
of alcoholic
From the Departments of Medicine (KDW. CMS. GEM) and Pathology (Clinical Laboratories) (RR), Emory University School of Medicine, Atlanta, Georgia, and the Division of Emergency Medicine, University of Rochester Medical Center (KDW, CMS). Rochester, New York. Requests for reprints should be addressed to Keith D. Wrenn. M.D., Strong Memorial Hospital, Box 655, 601 Elmwood Avenue, Rochester, New York 14642. Manuscript submitted September 24, 1990, and accepted in revised form April 22, 1991.
physiology is complex, the syndrome reversible and has a low mortality.
is rapidly
he syndrome of a wide-gap metabolic acidosis, malnutrition, and binge drinking superimposed on chronic alcohol abuse is most commonly referred to as alcoholic ketoacidosis (AKA). Numerous small reports and case studies have refined our understanding of this syndrome since its first description in 1940 [l-15]. At our institution, indigent chronic alcoholics make up a sizeable portion of the patient population and the diagnosis of AKA is often made clinically. We were interested in prospectively evaluating this population to better characterize the syndrome, specifically with respect to the acid-base problems and associated laboratory abnormalities.
T
PATIENTS AND METHODS Every patient presenting to the Medical Emergency Department of Grady Memorial Hospital in Atlanta, Georgia, with a history of alcohol abuse and a wide anion gap of greater than 15 mmol/L (normal, 12 f 2 mmol/L) was eligible for inclusion in the study. Exclusion criteria included any other disorder or toxin that could account for the widegap acidosis, such as the ingestion of aspirin, methanol, ethylene glycol, iron, or isoniaxid. Patients were excluded if a convulsion had occurred within 1 hour prior to blood sampling. Any other known cause of profound lactic acidosis present on admission (i.e., sepsis, ischemic bowel, carbon monoxide inhalation, and so forth) was also an exclusionary criterion. This study was approved by the Emory University Human Studies Committee. Patients had multiple blood samples drawn and a data intake questionnaire completed if they were eligible for the study. Blood sampling included a complete blood count, routine serum electrolyte measurements and biochemical profiles, blood alcohol level (BAL), arterial blood gases, serum acetoacetate (ACAC), /3-hydroxybutyrate (BOHB), and lactate. Not every patient had all of these tests performed because, in some instances, the diagnosis only became apparent after extracellular fluid volume repletion had already begun. Serum electrolyte measurements and biochemical profiles were perAugust
1991
The American
Journal
of Medicine
Volume
91
119
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
TABLE I Presenting
Symptoms
Symptom
and Signs in AKA Number
Nausea Vomiting Abdominal pain Shortness of breath Tremulousness Hematemesis Dizziness ye?; pain Diarrhea Syncope Seizure Melena
Percent
Physical Finding Tachycardia (pulse > loo/minute) Tachypnea (RR > 20/minute) Abdominal tenderness Heme-positive stool Hepatomegaly Altered mental status Hypotension (SBP < 100 mm Hg) Abdominal distention Hypothermia Fever Decreased bowel sounds Rebound tenderness
Number
;z 32
1; 9 8 4
3 2 1 1
Percent :: 43 18
Range lOO-150/minute 22-36/minute
it 12 5
4 3
70-93
mm Hg
33.1”34.2”C 38.1’-38~5°C
RR = respiratory rate; SBP = systolic bloodpressure.
by far the most commonly observed complaints. Common physical findings included tachycardia, tachypnea, and abdominal tenderness. Despite the frequency of abdominal complaints, signs of significant abdominal problems such as distention, decreased bowel sounds, or rebound tenderness were rarely reported. When present, these signs were always due to concomitant processes such as pancreatitis, severe hepatitis, sepsis, or pneumonia. Altered mental status, defined as an abnormality in orientation or level of consciousness, was seen in nine patients (15%). In all but one of these patients, a process other than AKA was present that explained the altered mental status: hypoglycemia in four of the patients, severe alcohol intoxication in two patients, and stroke in the remaining three patients. Fever, defined as a temperature of greater than 37.8OC, was seen in only two patients: one had alcohol withdrawal and a seizure, and the other an intracerebral hemorrhage and a seizure. Hypothermia, defined as a core temperature less than 35”C, was seen in three patients. Two of these patients were hypoglycemic and one had severe acidosis (pH 6.95).
formed on a Beckman Astra 8 or an American Monitor Parallel Analyzer. Arterial blood gas measurements were performed on a Corning 178 Analyzer (Corning). Serum ACAC values were semiquantitative determinations done using the nitroprusside reaction (Acetest reagent tablets, Ames). BOHB and lactate levels were measured with a Cobas-Bio Centrifugal Analyzer on stored refrigerated samples of blood utilizing a procedure based on the methods of Williamson et al [16] or Henry [17]. In both cases, 1 mL of whole blood was mixed with 1 mL of perchloric acid (1.1 mmol/L) to obtain protein-free filtrates prior to storage. Statistical analysis was performed with both SPSSX [18] and BMDP-85 [19] programs. Results are listed as p values representing a Pearson correlation coefficient for continuous variables or Yates’ corrected chi-square for dichotomous variables. In all cases where percentages are used, the denominator represents the number of patients in whom the test was performed. Arterial blood gases, when measured, were obtained within 2 hours of the serum chemistry values.
RESULTS
Acid-Base Disturbances The mean serum bicarbonate level was 13 mmol/L (SD = 7). Serum bicarbonate was depressed in 58 patients (79%), normal in 15 patients (20%), and elevated in one patient (1%). The mean pH was 7.30 (SD = 0.17). Five patients had a pH less than 7.0 and 17 patients less than 7.20. Ten patients (14%) were actually alkalemic (pH greater than 7.45) and 23 (31%) had a normal serum pH. The mean serum albumin concentration for the whole group was 45.8 g/L (SD = 7.7) and 92% of patients had a serum albumin level greater than 40 g/L. Of the 40 patients who had arterial blood gas levels measured, nine (23%) had a simple wide-gap metabolic acidosis with compensatory respiratory
A total of 74 patients met the study inclusion criteria during the period from August 11, 1986, to May 11, 1987. There were 39 men (53%) and 35 women (47%) ranging in age from 21 to 70 years. The average self-reported daily alcohol intake was approximately 200 g (between 1 and 2 pints of 80proof liquor per day). A total of 23% of the patients (17 patients) had experienced more than one episode: three patients had two episodes of AKA recorded during the study period and an additional 14 patients had a prior history of AKA. Signs and Symptoms Presenting symptoms and signs are listed in Table I. Nausea, vomiting, and abdominal pain were 120
August 1991 The American Journal of Medicine
Volume
91
ALCOHOLIC
KETOACIDOSIS
/ WRENN
ET AL
Figure 1. Various associated metabolic abnormalities in patients with AKA. In all cases except urine amylase, the percent equals number of patients with abnormality divided by number of patients in whom test results were available. Denominator in parentheses.
alkalosis. Another 10 patients (25%) had a primary wide-gap metabolic acidosis coexisting with a primary respiratory alkalosis, defined by a measured arterial carbon dioxide tension less than that predicted by the formula of Albert et al [20] (expected arterial carbon dioxide tension = 1.5 [serum bicarbonate] + 8 f 2). Eleven patients (28%) had a widegap metabolic acidosis combined with a primary metabolic alkalosis, defined by a decrease in serum bicarbonate that was significantly less than the increase in anion gap. Six patients (15%) had a widegap metabolic acidosis combined with a normal-gap hyperchloremic acidosis, defined by a reduction in serum bicarbonate that was significantly greater than the increase in anion gap in the absence of respiratory alkalosis. Finally, four patients (10%) had a triple acid-base disorder with a primary widegap acidosis, primary metabolic alkalosis, and primary respiratory alkalosis. The mean BOHB level was 5,760 pmol/L (SD = 3,550). BOHB levels were significantly lower in patients with higher BALs (r = 0.51, p = 0.007). However, 13 of 38 patients with an elevated BOHB also had a detectable BAL. Ten of these 13 patients had a significantly elevated BAL (greater than 22 mmol/L). Fifteen of the 38 patients with an elevated BOHB did not have a BAL measured. Of the 36 patients in whom BOHB was not measured, 15 had a .detectable BAL (eight greater than 22 mmol/L), and BALs were not determined in 15 patients. Semiquantitative serum ACAC levels were negative in only two of 47 patients (4%) and were also significantly lower in patients with higher BALs (r = 0.50, p = 0.004). Urine ketones by dipstick were positive in 24 of 27 patients (90%). Negative urine ketones
were seen most commonly in patients with higher BALs. Patients with negative nitroprusside reactions in their serum or urine had a mean blood pH of 7.47 and a mean BOHB of 1,540 gmol/L compared with a mean pH of 7.30 and a mean BOHB of 5,760 pmol/L for the group as a whole. Serum lactate levels were elevated in 23 of 38 patients (61%). Five patients (13%) had moderate elevations of serum lactate, greater than 6 mmol/L. Of these five patients, one with a severely elevated lactate level of 8.8 mmol/L eventually died of the sepsis syndrome. The other four with more moderate lactate elevations were eventually found to have severe concomitant processes such as pancreatitis, rhabdomyolysis, or severe hepatitis. Patients with more severe liver disease seemed predisposed to higher lactate levels (r = 0.50, p = 0.001). Laboratory Abnormalities Pertinent laboratory abnormalities are represented in Figure 1. Hypoglycemia was defined as a blood glucose level less than 3.3 mmol/L. Of the nine hypoglycemic patients, seven were female and two were male. Hyperglycemia, defined by a blood glucose level greater than 8.3 mmol/L (150 mg), was often present. Eight patients (10%) had blood glucose values greater than 13.8 mmol/L (250 mg/dL) but were thought to have AKA rather than diabetic ketoacidosis (DKA) because there was no history of prior diabetes mellitus and no subsequent clinical evidence of glucose intolerance after the initial treatment in the hospital. In every initially hyperphosphatemic patient for whom there was a follow-up value available 8 to 24
August
1991
The American
Journal
of Medicine
Volume
91
121
ALCOHOLIC KETOACIDOSIS / WRENN ET AL TABLE II Other Significant DiagnosesAsscciatedwith AKA Diagnosis Pancreatitis Myopathylrhabdomyolysis Gastritis/upper GI bleeding Seizures Hepatitis/cirrhosis Alcohol withdrawal Hypoglycemia Infection Septic syndrome Urosepsis Pneumonia UTI Bronchitis Alcohol intoxication CVA Neuropathy Marijuana/cocaine intoxication Vaginal bleeding New-onset atrial fibrillation Epistaxis
Number
Percent
27
36
:; 11
:i
“8
:2” 11
:
a a
[7,8,141.
: : : 2
l
:
i
i 1
i 1
= gastrointestinal; IJTI = urinary tract infection; CVA = cerebrovascular accident.
hours later, there was a significant reduction to hypophosphatemic levels. Profound hypophosphatemia (less than 0.33 mmol/L) was seen in 16 patients (22%) at some time in the course of their evaluation. Serum aspartate aminotransferase levels (AST) were often abnormal but tended to be moderately elevated (mean AST 116 U/L). Total bilirubin was also frequently elevated but again the elevation was generally mild (mean total bilirubin 25 pmol/L). Significant elevations of BAL (greater than 22 mmol/L) were seen in 17 patients (40%) and severe elevations (greater than 54 mmol/L) in nine patients (21%). Associated Medical Illnesses and Hospital Admission Other significant diagnoses noted during the hospital visit are listed in Table II. Pancreatitis, myopathy, upper gastrointestinal bleeding, a seizure disorder, hepatitis, and alcohol withdrawal were the most common concomitant disorders. There was one death related to the sepsis syndrome (case fatality rate of approximately 1%). Forty patients (54%) were admitted to the hospital. The degree of acidosis correlated most strongly with admission (r = 0.43, p
COMMENTS This series of 74 patients with AKA both validates prior beliefs and further expands the spectrum of this disease. All our patients were chronic alcoholics consuming prodigious amounts of alcohol, almost 2 pints of &)-proof liquor per day on average. The disorder was common in our popula-
122
tion, occurring about nine times per month over the study period. This frequency is most likely an underestimation at our institution. It is possible that milder cases may have been missed where no blood work was performed. Additionally, abdominal pain as the major presenting complaint (as opposed to nausea and vomiting in association with abdominal pain) is often triaged to our surgical emergency department. Prior studies have shown an incidence rate of only between 0.1 and 1.5 cases per month
August 1991 The American Journal of Medicine
Epidemiology There were approximately equal numbers of men and women among our patients. We do not have statistics that accurately reflect the prevalence of alcoholism in the female population served by Grady Memorial Hospital, but in the United States the prevalence of alcoholism is significantly less among women [21,22]. Previous series have shown a predominance of women with AKA [1,2,7], and early studies on ketosis showed that women developed fasting ketonemia and ketonuria more rapidly and to a greater degree than men [23]. Small doses of estrogen appear to augment fatty acid production in adipose tissue [24]. Pregnancy is also known to contribute to the development of ketosis, perhaps through a placental hormone or increased fetal use of maternal carbohydrate stores 1251. Although this study does not definitely answer the question of a sexual predilection for this disorder, it appears likely that female alcoholics are at increased risk. There also appears to be a tendency for AKA to recur in certain patients. Nearly one quarter of our population had either a prior history of AKA or a repeat episode during the study period. Whether this represents a genetic disorder of ketone metabolism cannot be answered by this study. Clinical Presentation Nausea, vomiting, and abdominal pain were by far the most common presenting symptoms. Vital sign changes (tachycardia and tachypnea) and abdominal tenderness were the most frequent physical findings. Despite the frequency of abdominal complaints, objective findings other than tenderness on abdominal examination were much less frequent. Abdominal distention, decreased bowel sounds, ascites, or rebound tenderness occurred rarely and only in association with other severe intra-abdominal or extra-abdominal processes such as pancreatitis or sepsis. Altered mental status was not seen in uncomplicated cases of AKA. In our patients, altered mental status was always explained by a concomitant pro-
Volume 91
ALCOHOLIC
ALCOHOLIC
1.
3.
/ WRENN
2.
ETOH + acetaldehyde
stores
t NADHINAD
Extracellular, intravascular -DT volume depletion
ET AL
KETOACIDOSIS
Starvation
Jen
KETOACIDOSIS
f catecholamines cortisol f growth hormone
-...-.-..,. I -..
13
fiat& acid prod&ion
T carnitine ---+ acyltransferase
-+
+ acetate
-
:’
S-oxidation
citric acid cycle
)
triglycerides
i beta-hydroxybutyrate
Figure 2. pathophysiology
Simplified of AKA.
scheme
of
-
acetoacetate -d
cess such as intoxication, hypoglycemia, or a cerebrovascular accident. Fever was uncommon and its presence was also an indication of another concomitant disorder. In the two patients with fever, a seizure due to alcohol withdrawal or an intracerebral hemorrhage appeared to be responsible for the fever. None of our patients who were eventually documented to have a significant infection presented with a fever. Hypothermia was often a marker of a concomitant process such as hypoglycemia or severe acidosis. Pathophysiology AKA has a complex pathophysiology. A simplified scheme is presented in Figure 2. The main predisposing events appear to be a relative deficiency of insulin coupled with an excess of glucagon caused by (1) starvation with glycogen depletion, (2) an elevated NADHLNAD ratio secondary to alcohol’s metabolism by alcohol dehydrogenase [ 141, and (3) extracellular fluid volume depletion. The core event is the relative deficiency of insulin due to glycogen depletion, decreased gluconeogenesis, and the a-adrenergic effects of epinephrine in inhibiting its secretion [14]. The elevated NADH/ NAD ratio affects gluconeogenesis, and both starvation and the presence of liver disease affect the ability to store glycogen [8]. Extracellular fluid volume depletion eventually stimulates the release of counterregulatory hormones such as growth hormone, cortisol, epinephrine, and, secondarily, glucagon [4,7,26]. Insulin deficiency promotes lipolysis and the liberation of free fatty acids [4]. Delivery of free
fatty acids to the liver is not, however, sufficient to initiate maximal ketogenesis [2,27,28]. There is a close link between the rate of ketosis and the liver carnitine content [27,28]. Fatty carnitine acyltransferase I and II constitute the carrier system by which free fatty acids are moved from the cytosol into the mitochondria where B-oxidation to the ketoacids BOHB and ACAC can take place [27,29]. Otherwise, the fatty acids would be metabolized to triglycerides via the Kreb’s cycle [29,30]. The activity of fatty carnitine acyltransferase I and II is dependent upon the concentration of malonyl coenzyme A (CoA), which inhibits their activity [29]. The concentration of malonyl CoA is decreased by a combination of decreased insulin and increased glucagon [27,28]. Glycogen depletion is required in order for the increased glucagon/insulin ratio to result in ketoacid production [29,30]. Whether pancreatic injury plays a role in the increased glucagon and decreased insulin levels is speculative, but in early acute pancreatic inflammation, decreased insulin levels combined with increased glucagon, cortisol, and catecholamine levels have been reported [31]. The elevated NADHLNAD ratio in alcoholics favors the formation of BOHB and lactate rather than ACAC and pyruvate. Diminished pyruvate dehydrogenase activity, either as a result of decreased NAD availability or magnesium or thiamine deficiency, also inhibits the utilization of pyruvate in the Kreb’s cycle, yielding more substrate for lactate production [32,33]. Alcohol itself, when chronically ingested with a calorically adequate fat-containing diet, may induce ketogenesis, and large doses of
August
1991 The American
Journal
of Medicine
Volume
91
123
ALCOHOLIC
KETOACIDOSIS
/ WRENN
ET AL
alcohol increase free fatty acid production liver [34].
in the
tients in our series had a negative nitroprusside reaction in their serum and only three of 29 patients had a negative dipstick reaction for ketones in their urine. The urine dipstick reaction is therefore approximately 90% sensitive in screening for AKA whereas the semiquantitative serum nitroprusside reaction is 96% sensitive. In the five patients who had negative serum or urine reactions for ACAC, the syndrome was very mild compared with that in the group as a whole. Serum ACAC was, like BOHB, inversely related to BAL. In the only prior series in which lactate levels were systematically measured, marked elevations of lactate occurred only in patients with shock [8]. Among our patients, 61% had an elevated lactate level. Ten patients had more than modestly elevated lactate levels (greater than 2.8 mmol/L) and all but one of these patients had a detectable BAL, implying continued drinking to within hours of admission. The mean BAL for these 10 patients was 35 mmol/L compared with a mean BAL of 8 mmol/L for the patients with lower lactate levels. Presumably, continued metabolism of ethanol further increases the NADH/NAD ratio, thereby causing increased production and decreased utilization of lactate. Moderate to severe elevations of lactate were seen in five patients, all of whom had serious concomitant disorders (sepsis, pancreatitis, rhabdomyolysis, and hypoglycemia).
Acid-Base Abnormalities Halperin et al [14] emphasized the four types of metabolic acidosis seen in AKA: (1) ketoacidosis; (2) lactic acidosis; (3) acetic acidosis; and (4) “indirect” loss of bicarbonate in the urine. Also noted were the complex interactions of the various types of metabolic acidosis with metabolic alkalosis from extracellular volume contraction. They theorized that volume contraction also contributed to inhibition of insulin release through increases in a-adrenergic activity [14]. Fulop and Hoberman [8] also noted the complexity of the acid-base disorder; half of their patients were not acidemic and almost one third were actually alkalemic. Our series reinforces the fact that AKA is not simply a wide anion gap acidosis due to accumulated ketoacids. Only 55% of the patients were acidemic and 78% had a mixed acid-base disorder. In about one quarter of patients, a primary metabolic alkalosis coexisted with the wide-gap acidosis, and in another quarter of patients, a primary respiratory alkalosis was present. Fifteen percent of patients manifested both a wide-gap acidosis and a hyperchloremic acidosis, while a triple acid-base disorder was seen in 10%. Metabolic alkalosis most likely represents the combined effects of volume contraction and vomiting. Respiratory alkalosis probably reflects underlying alcohol withdrawal, pain, or associated disorders such as sepsis or severe liver disease. Hyperchloremic acidosis probably results from “indirect” renal loss of bicarbonate associated with ketonuria [35,36]. As expected, BOHB levels were elevated in all patients in whom this measurement was performed. In a prior series, only one of 15 patients with a significantly elevated BOHB level also had a detectable BAL [15]. In our series, the BOHB levels correlated inversely with BAL. Possibly having ethanol’s “empty” calories available decreases fatty acid mobilization. In contrast to prior experience, however, many alcoholics in our series continued to ingest enough ethanol, despite vomiting, to maintain a detectable BAL and still develop significant ketoacidosis. Alcohol intoxication by itself does not cause clinically significant elevations of BOHB [37]. In prior series, BOHB levels were elevated out of proportion to ACAC levels, presumably because of an elevated NADH/NAD ratio. Because BOHB is not detected by the nitroprusside reaction normally used with urine samples to detect ketoacids, it has been stressed that a negative dipstick does not rule out AKA [8]. While this is true, only two of 47 pa-
124
August
1991 The American
Journal
of Medicine
Volume
Laboratory Findings Hyponatremia and hypokalemia were both common (Figure 1). Hyponatremia appeared to be related to vomiting and extracellular volume depletion. Hypokalemia correlated with hypomagnesemia (r = 0.39, p = 0.002). Hyperkalemia was seen in 15% of patients and correlated with lower pH values (r = 0.46, p
91
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
gestion that men are predisposed to hypoglycemia in AKA [ll], in our series hypoglycemia was more than three times as common in women. Ethanol was present in every hypoglycemic patient who had a concomitant BAL measured (mean concentration 26 mmol/L), supporting the prediction that blood glucose values should be lower in patients with more recent alcohol intake [14,39]. Prior studies in alcoholics have noted that hypoglycemia appears to be independent of pancreatic injury, liver injury, or changes in circulating insulin and glucagon [9]. Hyperglycemia was seen in 35% of patients. Mild to moderate glucose intolerance has been noted in patients in other series [2,8,15]. When blood glucose values are higher than 13.8 mmol/L, the differentiation of AKA from DKA becomes difficult. We included eight such patients in our series based on the clinical picture and their subsequent course. In chronic alcoholics without demonstrable liver disease, an impaired insulin response to glucose loading has been described [40]. Impaired glucose uptake by tissues with normal insulin responses has also been seen in alcoholic women [41]. Calcium, Phosphorus, and Magnesium Abnormalities Miller et al [9] stressed the importance of both the inhibition of NADH oxidation by the mitochondrial accumulation of acetaldehyde and the loss of mitochondrial phosphorus in contributing to ketosis [42]. In their group of patients, admission hyperphosphatemia was seen in 82%, with a mean admission phosphorus level of 2.2 mmol/L (6.8 mg/dL). In our series, only 25% of patients were hyperphosphatemic on admission but, like the patients in Miller’s series, in every case the phosphorus level fell rapidly with treatment, often to critically low levels. The greater severity of AKA in Miller’s patients (mean pH 7.14) compared with ours (mean pH 7.30) accounts for the differences in admission phosphorus concentrations. Hypophosphatemia was seen on admission in 25% of our patients. We believe our series is more representative of the spectrum of AKA as it presents to the emergency department. Other Serum Abnormalities Since the original description of AKA by Dillon et al [l], the liver has been thought to play a central role in the development of this syndrome. Diseased livers have a decreased capacity to store glycogen and promote gluconeogenesis [12,38,39]. In our patients, evidence of significant liver disease was present in the vast majority of patients. AST and bilirubin levels were elevated in 89% and 72% of patients, respectively.
Serum amylase was elevated in almost half the patients in whom it was measured. A past history of pancreatitis was present in another seven patients with normal serum amylase levels. In all, 64% of patients had evidence of pancreatic injury by history or on laboratory testing, including two patients with elevated urine amylase and normal serum amylase. Given that the serum amylase level is not infrequently within the normal range in the setting of acute pancreatitis in alcoholics [43,44], it is intriguing to speculate on the importance of pancreatic injury from alcohol in inducing the syndrome of AKA. As in AKA, the main symptoms of pancreatitis are nausea, vomiting, and abdominal pain. Anemia was common, but mild, and no correlation with available indices of gastrointestinal bleeding was noted. Leukocytosis was common but seemed more a marker of volume depletion and demargination from physical stress than of infection. Leukopenia, however, was often a marker of severe underlying disease, including infection. Thrombocytopenia was also seen relatively commonly. Concomitant disorders, such as pancreatitis, myopathy, upper gastrointestinal bleeding, seizures, hepatitis, and alcohol withdrawal, were common and undoubtedly contributed to the morbidity of the syndrome. Many patients had more than one associated problem. The case-fatality rate was very low (on the order of I%), and, as in prior series, was related more to associated disorders than to the syndrome of AKA itself, despite pH values below 7.00. The decision to admit the patient to an inhospital area seemed most related to the severity of acidosis. Slightly more than half the patients were admitted; the rest were able to be treated and discharged from the emergency department within 12 hours. Treatment On the basis of prior studies and our experience, we recommend that the treatment of AKA be directed toward reversing the three major pathophysiologic causes of the syndrome, extracellular fluid volume depletion, glycogen depletion, and an elevated NADH/NAD ratio. Extracellular volume repletion with saline is indicated to inhibit the release of counter-regulatory hormones. Whether moderate volume replacement (500 mL/hour for 4 hours followed by 250 ml/hour for 4 hours) as opposed to vigorous volume replacement (1,000 ml/hour for 4 hours followed by 500 ml/hour for 4 hours) would be more beneficial in patients without extreme extracellular volume deficits has not been studied in AKA, but in DKA a regimen of moderate extracellular volume repletion resulted in a more rapid in-
August 1991 The American Journal of Medicine
Volume 91
125
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
crease in serum bicarbonate with a significant reduction in the cost of therapy [45]. Glucose administration may also help interrupt ketogenesis by stimulation of insulin production and secretion. Dextrose-containing solutions should be administered to stimulate the oxidation of NADH and to replete glycogen stores in patients with hypoglycemia or euglycemia (blood glucose level less than 13.8 mmol/L). Miller and colleagues [9] showed that saline alone did not correct the acidosis of AKA as rapidly as dextrose and saline. Initially, dextrose-containing solutions should be avoided in patients with severe hyperglycemia (blood glucose greater than 13.8 mmol/L) because of the risk of worsening hyperglycemia in a state of relative insulin deficiency. Insulin should be used judiciously because of the risk of hypoglycemia in patients with depleted glycogen stores. Insulin in very low doses is indicated only in patients with blood glucose values greater than 13.8 mmol/L. Potassium repletion is indicated in hypokalemic patients and in normokalemic patients with acidemia. These patients should initially receive at least 10 mmol of potassium chloride per hour in the intravenous fluids. Higher rates of infusion may be required for more severe degrees of hyperkalemia as the insulin deficit and acidosis correct and potassium shifts intracellularly [46]. The subsequent rate and total dose of potassium are determined by successive potassium measurements. In the absence of renal insufficiency, magnesium repletion is indicated in all patients, not only to help restore calcium and potassium homeostasis, but also to forestall alcohol withdrawal [47,48]. Magnesium can be safely given at a rate of up to 1 g/hour [49]. Calcium repletion is rarely needed; magnesium repletion will usually allow spontaneous correction of hypocalcemia, if present, to occur [49]. The routine use of phosphorus is not encouraged but phosphorus levels should be followed closely. Routine phosphorus repletion has not been shown to improve morbidity or mortality in DKA [50,51]. When hypophosphatemia is present, intravenous phosphate repletion at a rate of 0.08 to 0.16 mmol/kg over 6 hours is recommended [52]. Adverse effects of more rapid phosphate repletion include hypocalcemia and t&any [53-551. Bicarbonate therapy is rarely if ever needed. In patients with DKA and pH values as low as 6.90, no change in morbidity or mortality has resulted from bicarbonate therapy [56]. In AKA, regeneration of bicarbonate from the metabolism of lactate, ketoacids, and acetate occurs with conservative treatment [14]. Patients’ pH values respond rapidly (within 12 hours) to treatment with volume repletion and glu-
126
August 1991 The American Journal of Medicine
case alone [9]. In our patients, no one required bicarbonate despite pH values as low as 6.95. Thiamine repletion is routinely indicated both to increase pyruvate dehydrogenase activity and to provide prophylaxis against the development of Wernicke’s encephalopathy [57]. It is safe to administer lOO-mg boluses of thiamine intravenously [58]. Because alcoholics are predisposed to multiple diseases that may be associated with AKA, a careful history and physical examination must be performed on even the most straightforward of cases. Perforated peptic ulcer, small bowel obstruction, and other abdominal catastrophes may initially be misdiagnosed as AKA. Similarly, AKA may be an accompanying illness in conjunction with pancreatitis, acute upper gastrointestinal hemorrhage, rhabdomyolysis, alcohol withdrawal, or sepsis. Conclusions AKA is a common disorder in chronically malnourished alcoholics. There is a tendency for the disorder to recur in some patients. Nausea, vomiting, and abdominal pain constitute the predominant symptoms. The physical findings are usually related to vital sign abnormalities and abdominal tenderness. Findings such as altered mental status, abdominal distention, decreased bowel sounds, rebound abdominal tenderness, fever, or hypothermia are not seen in uncomplicated AKA and imply another underlying process. The acid-base disorder of AKA is complex, and double or triple acid-base disorders are common. Primary metabolic alkalosis from vomiting and extracellular volume depletion and respiratory alkalosis from alcohol withdrawal, sepsis, or pain are particularly common. The severity of acidosis is variable and alkalemia is seen occasionally. Not only are BOHB and ACAC levels elevated, but a modest element of lactic acidosis is routinely present. The urine and serum nitroprusside reactions for ketones are almost always positive unless the syndrome is very mild. Both hyperglycemia and hypoglycemia are common. Hyponatremia, hypokalemia, hypomagnesemia, and hypocalcemia may also be expected. Both hyperphosphatemia and hypophosphatemia may be seen on presentation, but a precipitous reduction with treatment should be anticipated. Evidence of pancreatitis and myopathy are often found. A modest macrocytic anemia and leukocytosis are frequently seen. Leukopenia and thrombocytopenia, while uncommon, are markers of severe underlying disease. BALs in the intoxicated range are not unusual. Other associated disorders common to alcoholics frequently accompany AKA
Volume 91
ALCOHOLIC
and should be searched for and treated aggressively. The treatment of AKA centers on the administration of saline and dextrose to reverse extracellular volume depletion and glycogen depletion. Supportive care with electrolyte and vitamin repletion should be instituted. The role of phosphorus in treatment is unresolved but it probably should be reserved for documented hypophosphatemia. Administration of bicarbonate and insulin should generally be avoided. Further focused studies are needed to elucidate more fully the pathophysiology of AKA. Studies employing more sensitive markers of pancreatic injury than amylase would also be useful. In addition, studies addressing the role of routine phosphorus and magnesium therapy are needed.
REFERENCES 1. Dillon ES, Dyer WW, Smelo LS. Ketone acidosis in nondiabetic adults. Med Clin North Am 1940; 24: 1813-22. 2. Jenkins DW, Eckel RE. Craig JW. Alcoholic ketoacidosis. JAMA 1971; 217: 177-83. 3. Weisberger Cl., Hamilton W. Alcoholic ketoacidosis [letter]. JAMA 1971: 217: 1865-6. 4. Levy LJ, Duga J. Girgis M, Gordon EE. Ketoacidosis associated with alcoholism in nondiabetic subjects. Ann Intern Med 1973; 78: 213-9. 5. Edwards WM, Hoyt R. Alcoholic ketoacidosis. South Med J 1973; 66: 281-2. 6. Peterson G. Alcoholism and ketoacidosis [letter]. Ann Intern Med 1973; 78: 983. 7. Cooperman MT, Davidoff F, Spark R, Pallotta J. Clinical studies of alcoholic ketoacidosis. Diabetes 1974; 23: 433-9. 8. Fulop M, Hoberman HD. Alcoholic ketosis. Diabetes 1975; 24: 785-90. 9. Miller PD. Heinig RE, Waterhouse C. Treatment of alcoholic acidosis-the role of dextrose and phosphorus. Arch Intern Med 1978; 138: 67-72. 10. Podratz KC. Alcoholic ketoacidosis in pregnancy. Obstet Gynecol 1978; 52: 54s-7s. 11. Platia EV. Hsu TH. Hypoglycemic coma with ketoacidosis in nondiabetic alcoholics. West J Med 1979: 131: 270-6. 12. Lumpkin JR, Baker FJ II, Franaszek JB. Alcoholic ketoacidosis in a pregnant woman. JACEP 1979; 8: 21-3. 13. Devenyi P. Alcoholic hypoglycemia and alcoholic ketoacidosis: sequential events of the same process? Can Med Assoc J 1982; 127: 513. 14. Halperin ML, Hammeke M, Josse RG. Jungas RL. Metabolic acidosis in the alcoholic: a pathophysiologic approach. Metabolism 1983; 32: 308-15. 15. Fulop M. Ben-Ezra J. Bock J. Alcoholic ketosis. Alcoholism (NY) 1986; 10: 610-5. 16. Williamson DH. Mellanby J, Krebs HA. Enzymatic determination of D(-) Bhydroxybutyric acid and acetoacetic acid in blood. Biochem J 1962; 82: 90-6. 17. Henry RJ. Clinical chemistry-principles and technics. New York: Harper and Row. 1968: 814-5. 18. Nie NH, Hull CH, Jenkins JG, et al. SPSS: statistical package for the social sciences. 2nd ed. New York: McGraw-Hill, 1986. 19. Dixon WJ, Brown MB. BMDP-85, biomedical computer programs, P-series. Berkeley: University of California Press, 1985. 20.Albert MS, Dell RB. Winters RW. Quantitative displacement of acid-base equilibrium in metabolic acidosis. Ann Intern Med 1967; 66: 312-22. 21. Blume SB. Women and alcohol-a review. JAMA 1986; 256: 1467-70. 22. Moore RA. The prevalence of alcoholism in a community general hospital. Am J Psychiatry 1971; 128: 638-9. 23. Deuel HJ Jr, Gulick M. Studies on ketosis I. The sexual variation ketosis. J Biol Chem 1932; 96: 25-34.
in starvation
KETOACIDOSIS
/ WRENN
ET AL
24. McKerns KW. Clynes R. Sex differences in rat adipose tissue metabolism. Metabolism 1961; 10: 165-70. 25. Hartop M. Human chorionic mammosomatotropin and its clinical significance. Clin Endocrinol 1972: 1: 209-18. 26. Unger RH, Orci L. Glucagon and the A cell. Physiology and pathophysiology. N Engl J Med 1981; 304: 1518-24, 1575-80. 27. McGarry JD. Foster DW. Hormonal control of ketogenesis. Biochemical considerations. Arch Intern Med 1977; 137: 495-501. 28. Schade DS. Eaton RP. Glucagon regulation of plasma ketone body concentration in human diabetes. J Clin Invest 1975: 56: 1340-4. 29. Cahill GF Jr. Ketosis. Kidney Int 1981: 20: 416-25. 30. Blachley J. Knochel JP. Alcohol-induced disturbances in electrolyte and acid base homeostasis. in: Kokko JP, Tannen RL. editors. Fluid and electrolyte abnormalities of disease. Philadelphia: WB Saunders, 1986: 515-9. 31. Drew SI. Joffe B. Vinik A, et a/. The first 24 hours of acute pancreatitis -changes in biochemical and endocrine homeostasis in patients with pancreatitis compared with those in control subjects undergoing stress for reasons other than pancreatitis. Am J Med 1978; 64: 795-803. 32. Kreisberg RA. Lactate homeostasis and lactic acidosis. Ann Intern Med 1980; 92: 227-37. 33. Flink EB. Role of magnesium depletion in Wernicke-Korsakoff syndrome. N Engl J Med 1977; 294: 1367-70. 34. Lefevre A, Adler H, Lieber CS. Effect of ethanol on ketone metabolism. J Clin Invest 1970; 49: 1775-82. 35. Halperin ML, Bear RA, Hannaford MC, Goldstein MB. Selected aspectsof the pathophysiology of metabolic acidosis in diabetes mellitus. Diabetes 1981; 30: 781-7. 36. Hilton JG. Vandenbroucke AC, Josse RG. Buckley GC, Halperin ML The unne anion gap: the critical clue to resolve a diagnostic dilemma in a patient with ketoacidosis. Diabetes Care 1984; 7: 486-90. 37. Fulop M. Bock J. Ben-Ezra J, Antony M, Danzig J. Gage JS. Plasma lactate and 3-hydroxybutyrate levels in patients with acute ethanol Intoxication. Am J Med 1986; 80: 191-4. 38. Freinkel N. Arky RA. Singer DL, et al. Alcohol hypoglycemia IV: current concepts of its pathogenesis. Diabetes 1965; 14: 350-61. 39. Williams HE, Mortimer GE. Studies on the mechanism of ethanol-induced hypoglycemia. J Clin Invest 1963; 42: 497-506. 40. Sereny G, Endrenyi L. Mechanism and significance of carbohydrate Intolerance In chronic alcoholism. Metabolism 1978; 27: 104-6. 41. Dornhorst A, Ouyang A. Effect of alcohol onglucose tolerance. Lancet 1971; 2: 957-9. 42. Gregg CT, Lehninger AL Dependence of respiration on phosphate and phosphate acceptor in submitochondrial systems Ill. Sonic fragments. Biochim Biophys Acta 1963; 78: 27-44. 43. Spechler SJ. Dalton JW. Robbins AH, et al. Prevalence of normal serum amylase levels in patients with acute alcoholic pancreatitis. Dig Dis Sci 1983; 28: 865-9. 44. Clavier PA, Robert J. Meyer P, et a/. Acute pancreatitis and normoamylasemia-not an uncommon combination. Ann Surg 1989; 210: 614-20. 45. Adrogue HJ. Barrero J, Eknoyan G. Salutary effects of modest fluid replacement in the treatment of adults with diabetic ketoacidosis-use in patients without extreme volume deficit. JAMA 1989; 262: 2108-13. 46. Kruse JA, Carlson RW. Rapid correction of hypokalemia using concentrated intravenous potassium chloride infusions. Arch Intern Med 1990; 150: 613-7. 47. Whang R, Flink EB, Dyckner T. Webster PO, Aikawa JK. Ryan MP. Magnesium depletion as a cause of refractory potassium repletion. Arch Intern Med 1985: 145: 1686-9. 48. Wolfe SM. Victor M. The relationship of hypomagnesemla and alkalosis to alcohol withdrawal symptoms. Ann NY Acad Sci 1969; 162: 973-84. 49. Rude RK. Singer FR. Magnesium deficiency and excess. Annu Rev Med 1981; 32: 245-59. 50. Wilson HK, Kener SP. Lea S. Boyd AE Ill, Eknoyan G. Phosphate therapy in diabetic ketoacidosis. Arch Intern Med 1982; 142: 517-24. 51. Fisher JN, Kitabchi AE. A randomized study of phosphate therapy in the treatment of diabetic ketoacidosis. J Clin Endocrinol Metab 1983; 57: 177-80. 52. Lentz RD, Brown DM, Kjellstrand M. Treatment of severe hypophosphatemia. Ann Intern Med 1978; 89: 941-4. 53. Nanji AA. Symptomatic hypocalcemia due to hyperphosphatemia with alcoholic ketoacidosis [letter]. South Med J 1984: 77: 540.
August
1991
The American
Journal
of Medicine
Volume
91
127
ALCOHOLIC KETOACIDOSIS / WRENN ET AL 54. Winter RJ, Harris CJ. Phillips LS, et a/. Diabetic ketoacidosis-induction of hypocalcemia and hypomagnesemia by phosphate therapy. Am J Med 1979; 67: 897-900. 55. Zipf WB, Bacon GE, Spencer ML, et a/. Hypocalcemia. hypomagnesemia. and transient hypoparathyroidism during therapy with potassium phosphate in diabetic ketoacidosis. Diabetes Care 1979; 2: 265-8.
128
August 1991 The American Journal of Medicine
56. Morris LR, Murphy MB, Kitabachi AE. Bicarbonate therapy in severe diabetic ketoacidosis. Ann Intern Med 1986; 105: 8X-40. 57. Marx JA. The varied faces of Wernicke’s encephalopathy. J Emerg Med 1985; 3: 411-3. 58. Wrenn KD. Slovis CM, Murphy F. A toxicity study of parenteral thiamine hydrochloride. Ann Emerg Med 1989; 18: 867-70.
Volume 91
PURPOSE:To further
elucidate the clinical ketoacidosis (AKA). PATIENTSANDMETHODS: Acaseseriesof74patie& with AKA defiied as a wide anion gap metabolic acidosis unexplained by any other disorder or toxin, including any patient with a history of chronic alcohol abuse. The setting was the Medical Emergency Department at Grady Memorial Hospital in Atlanta, Georgia, a university-affi’jted inner-city hospital. RESULTS: AKA is a common disorder in the emergency department, more common than previously thought. The acid-base abnormalities are more diverse than just a wide-gap metabolic acidosis and often include a concomitant metabolic alkalosis, hyperchloremic acidosis, or respiratory alkalosis. Lactic acidosis is also common. Semiquantitative serum acetoacetate levels were positive in 96% of patients. Elevated blood alcohol levels were present in two thirds of patients in whom alcohol levels were determined, and levels consistent with intoxication were seen in 40% of these patients. Electrolyte disorders including hyponatremia, hypokalemia, hypophosphatemia, hyperglycemia, hypocalcemia, and hypomagnesemia were common on presentation. The most common symptoms were nausea, vomiting, and abdominal pain. The most common physical fmdings were tachycardia, tachypnea, and abdominal tenderness. Altered mental status, fever, hypothermia, or other abnormal fmdings were uncommon and reflected other underlying procemes. CONCLUSIONS: AKA is a common disorder in chronic malnourished alcoholic persons. The acid-base abnormalities reflect not only the ketoacidosis, but also associated extracellular fluid volume depletion, alcohol withdrawal, pain, sepsis, or severe liver disease. Although the pathospectrum
of alcoholic
From the Departments of Medicine (KDW. CMS. GEM) and Pathology (Clinical Laboratories) (RR), Emory University School of Medicine, Atlanta, Georgia, and the Division of Emergency Medicine, University of Rochester Medical Center (KDW, CMS). Rochester, New York. Requests for reprints should be addressed to Keith D. Wrenn. M.D., Strong Memorial Hospital, Box 655, 601 Elmwood Avenue, Rochester, New York 14642. Manuscript submitted September 24, 1990, and accepted in revised form April 22, 1991.
physiology is complex, the syndrome reversible and has a low mortality.
is rapidly
he syndrome of a wide-gap metabolic acidosis, malnutrition, and binge drinking superimposed on chronic alcohol abuse is most commonly referred to as alcoholic ketoacidosis (AKA). Numerous small reports and case studies have refined our understanding of this syndrome since its first description in 1940 [l-15]. At our institution, indigent chronic alcoholics make up a sizeable portion of the patient population and the diagnosis of AKA is often made clinically. We were interested in prospectively evaluating this population to better characterize the syndrome, specifically with respect to the acid-base problems and associated laboratory abnormalities.
T
PATIENTS AND METHODS Every patient presenting to the Medical Emergency Department of Grady Memorial Hospital in Atlanta, Georgia, with a history of alcohol abuse and a wide anion gap of greater than 15 mmol/L (normal, 12 f 2 mmol/L) was eligible for inclusion in the study. Exclusion criteria included any other disorder or toxin that could account for the widegap acidosis, such as the ingestion of aspirin, methanol, ethylene glycol, iron, or isoniaxid. Patients were excluded if a convulsion had occurred within 1 hour prior to blood sampling. Any other known cause of profound lactic acidosis present on admission (i.e., sepsis, ischemic bowel, carbon monoxide inhalation, and so forth) was also an exclusionary criterion. This study was approved by the Emory University Human Studies Committee. Patients had multiple blood samples drawn and a data intake questionnaire completed if they were eligible for the study. Blood sampling included a complete blood count, routine serum electrolyte measurements and biochemical profiles, blood alcohol level (BAL), arterial blood gases, serum acetoacetate (ACAC), /3-hydroxybutyrate (BOHB), and lactate. Not every patient had all of these tests performed because, in some instances, the diagnosis only became apparent after extracellular fluid volume repletion had already begun. Serum electrolyte measurements and biochemical profiles were perAugust
1991
The American
Journal
of Medicine
Volume
91
119
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
TABLE I Presenting
Symptoms
Symptom
and Signs in AKA Number
Nausea Vomiting Abdominal pain Shortness of breath Tremulousness Hematemesis Dizziness ye?; pain Diarrhea Syncope Seizure Melena
Percent
Physical Finding Tachycardia (pulse > loo/minute) Tachypnea (RR > 20/minute) Abdominal tenderness Heme-positive stool Hepatomegaly Altered mental status Hypotension (SBP < 100 mm Hg) Abdominal distention Hypothermia Fever Decreased bowel sounds Rebound tenderness
Number
;z 32
1; 9 8 4
3 2 1 1
Percent :: 43 18
Range lOO-150/minute 22-36/minute
it 12 5
4 3
70-93
mm Hg
33.1”34.2”C 38.1’-38~5°C
RR = respiratory rate; SBP = systolic bloodpressure.
by far the most commonly observed complaints. Common physical findings included tachycardia, tachypnea, and abdominal tenderness. Despite the frequency of abdominal complaints, signs of significant abdominal problems such as distention, decreased bowel sounds, or rebound tenderness were rarely reported. When present, these signs were always due to concomitant processes such as pancreatitis, severe hepatitis, sepsis, or pneumonia. Altered mental status, defined as an abnormality in orientation or level of consciousness, was seen in nine patients (15%). In all but one of these patients, a process other than AKA was present that explained the altered mental status: hypoglycemia in four of the patients, severe alcohol intoxication in two patients, and stroke in the remaining three patients. Fever, defined as a temperature of greater than 37.8OC, was seen in only two patients: one had alcohol withdrawal and a seizure, and the other an intracerebral hemorrhage and a seizure. Hypothermia, defined as a core temperature less than 35”C, was seen in three patients. Two of these patients were hypoglycemic and one had severe acidosis (pH 6.95).
formed on a Beckman Astra 8 or an American Monitor Parallel Analyzer. Arterial blood gas measurements were performed on a Corning 178 Analyzer (Corning). Serum ACAC values were semiquantitative determinations done using the nitroprusside reaction (Acetest reagent tablets, Ames). BOHB and lactate levels were measured with a Cobas-Bio Centrifugal Analyzer on stored refrigerated samples of blood utilizing a procedure based on the methods of Williamson et al [16] or Henry [17]. In both cases, 1 mL of whole blood was mixed with 1 mL of perchloric acid (1.1 mmol/L) to obtain protein-free filtrates prior to storage. Statistical analysis was performed with both SPSSX [18] and BMDP-85 [19] programs. Results are listed as p values representing a Pearson correlation coefficient for continuous variables or Yates’ corrected chi-square for dichotomous variables. In all cases where percentages are used, the denominator represents the number of patients in whom the test was performed. Arterial blood gases, when measured, were obtained within 2 hours of the serum chemistry values.
RESULTS
Acid-Base Disturbances The mean serum bicarbonate level was 13 mmol/L (SD = 7). Serum bicarbonate was depressed in 58 patients (79%), normal in 15 patients (20%), and elevated in one patient (1%). The mean pH was 7.30 (SD = 0.17). Five patients had a pH less than 7.0 and 17 patients less than 7.20. Ten patients (14%) were actually alkalemic (pH greater than 7.45) and 23 (31%) had a normal serum pH. The mean serum albumin concentration for the whole group was 45.8 g/L (SD = 7.7) and 92% of patients had a serum albumin level greater than 40 g/L. Of the 40 patients who had arterial blood gas levels measured, nine (23%) had a simple wide-gap metabolic acidosis with compensatory respiratory
A total of 74 patients met the study inclusion criteria during the period from August 11, 1986, to May 11, 1987. There were 39 men (53%) and 35 women (47%) ranging in age from 21 to 70 years. The average self-reported daily alcohol intake was approximately 200 g (between 1 and 2 pints of 80proof liquor per day). A total of 23% of the patients (17 patients) had experienced more than one episode: three patients had two episodes of AKA recorded during the study period and an additional 14 patients had a prior history of AKA. Signs and Symptoms Presenting symptoms and signs are listed in Table I. Nausea, vomiting, and abdominal pain were 120
August 1991 The American Journal of Medicine
Volume
91
ALCOHOLIC
KETOACIDOSIS
/ WRENN
ET AL
Figure 1. Various associated metabolic abnormalities in patients with AKA. In all cases except urine amylase, the percent equals number of patients with abnormality divided by number of patients in whom test results were available. Denominator in parentheses.
alkalosis. Another 10 patients (25%) had a primary wide-gap metabolic acidosis coexisting with a primary respiratory alkalosis, defined by a measured arterial carbon dioxide tension less than that predicted by the formula of Albert et al [20] (expected arterial carbon dioxide tension = 1.5 [serum bicarbonate] + 8 f 2). Eleven patients (28%) had a widegap metabolic acidosis combined with a primary metabolic alkalosis, defined by a decrease in serum bicarbonate that was significantly less than the increase in anion gap. Six patients (15%) had a widegap metabolic acidosis combined with a normal-gap hyperchloremic acidosis, defined by a reduction in serum bicarbonate that was significantly greater than the increase in anion gap in the absence of respiratory alkalosis. Finally, four patients (10%) had a triple acid-base disorder with a primary widegap acidosis, primary metabolic alkalosis, and primary respiratory alkalosis. The mean BOHB level was 5,760 pmol/L (SD = 3,550). BOHB levels were significantly lower in patients with higher BALs (r = 0.51, p = 0.007). However, 13 of 38 patients with an elevated BOHB also had a detectable BAL. Ten of these 13 patients had a significantly elevated BAL (greater than 22 mmol/L). Fifteen of the 38 patients with an elevated BOHB did not have a BAL measured. Of the 36 patients in whom BOHB was not measured, 15 had a .detectable BAL (eight greater than 22 mmol/L), and BALs were not determined in 15 patients. Semiquantitative serum ACAC levels were negative in only two of 47 patients (4%) and were also significantly lower in patients with higher BALs (r = 0.50, p = 0.004). Urine ketones by dipstick were positive in 24 of 27 patients (90%). Negative urine ketones
were seen most commonly in patients with higher BALs. Patients with negative nitroprusside reactions in their serum or urine had a mean blood pH of 7.47 and a mean BOHB of 1,540 gmol/L compared with a mean pH of 7.30 and a mean BOHB of 5,760 pmol/L for the group as a whole. Serum lactate levels were elevated in 23 of 38 patients (61%). Five patients (13%) had moderate elevations of serum lactate, greater than 6 mmol/L. Of these five patients, one with a severely elevated lactate level of 8.8 mmol/L eventually died of the sepsis syndrome. The other four with more moderate lactate elevations were eventually found to have severe concomitant processes such as pancreatitis, rhabdomyolysis, or severe hepatitis. Patients with more severe liver disease seemed predisposed to higher lactate levels (r = 0.50, p = 0.001). Laboratory Abnormalities Pertinent laboratory abnormalities are represented in Figure 1. Hypoglycemia was defined as a blood glucose level less than 3.3 mmol/L. Of the nine hypoglycemic patients, seven were female and two were male. Hyperglycemia, defined by a blood glucose level greater than 8.3 mmol/L (150 mg), was often present. Eight patients (10%) had blood glucose values greater than 13.8 mmol/L (250 mg/dL) but were thought to have AKA rather than diabetic ketoacidosis (DKA) because there was no history of prior diabetes mellitus and no subsequent clinical evidence of glucose intolerance after the initial treatment in the hospital. In every initially hyperphosphatemic patient for whom there was a follow-up value available 8 to 24
August
1991
The American
Journal
of Medicine
Volume
91
121
ALCOHOLIC KETOACIDOSIS / WRENN ET AL TABLE II Other Significant DiagnosesAsscciatedwith AKA Diagnosis Pancreatitis Myopathylrhabdomyolysis Gastritis/upper GI bleeding Seizures Hepatitis/cirrhosis Alcohol withdrawal Hypoglycemia Infection Septic syndrome Urosepsis Pneumonia UTI Bronchitis Alcohol intoxication CVA Neuropathy Marijuana/cocaine intoxication Vaginal bleeding New-onset atrial fibrillation Epistaxis
Number
Percent
27
36
:; 11
:i
“8
:2” 11
:
a a
[7,8,141.
: : : 2
l
:
i
i 1
i 1
= gastrointestinal; IJTI = urinary tract infection; CVA = cerebrovascular accident.
hours later, there was a significant reduction to hypophosphatemic levels. Profound hypophosphatemia (less than 0.33 mmol/L) was seen in 16 patients (22%) at some time in the course of their evaluation. Serum aspartate aminotransferase levels (AST) were often abnormal but tended to be moderately elevated (mean AST 116 U/L). Total bilirubin was also frequently elevated but again the elevation was generally mild (mean total bilirubin 25 pmol/L). Significant elevations of BAL (greater than 22 mmol/L) were seen in 17 patients (40%) and severe elevations (greater than 54 mmol/L) in nine patients (21%). Associated Medical Illnesses and Hospital Admission Other significant diagnoses noted during the hospital visit are listed in Table II. Pancreatitis, myopathy, upper gastrointestinal bleeding, a seizure disorder, hepatitis, and alcohol withdrawal were the most common concomitant disorders. There was one death related to the sepsis syndrome (case fatality rate of approximately 1%). Forty patients (54%) were admitted to the hospital. The degree of acidosis correlated most strongly with admission (r = 0.43, p
COMMENTS This series of 74 patients with AKA both validates prior beliefs and further expands the spectrum of this disease. All our patients were chronic alcoholics consuming prodigious amounts of alcohol, almost 2 pints of &)-proof liquor per day on average. The disorder was common in our popula-
122
tion, occurring about nine times per month over the study period. This frequency is most likely an underestimation at our institution. It is possible that milder cases may have been missed where no blood work was performed. Additionally, abdominal pain as the major presenting complaint (as opposed to nausea and vomiting in association with abdominal pain) is often triaged to our surgical emergency department. Prior studies have shown an incidence rate of only between 0.1 and 1.5 cases per month
August 1991 The American Journal of Medicine
Epidemiology There were approximately equal numbers of men and women among our patients. We do not have statistics that accurately reflect the prevalence of alcoholism in the female population served by Grady Memorial Hospital, but in the United States the prevalence of alcoholism is significantly less among women [21,22]. Previous series have shown a predominance of women with AKA [1,2,7], and early studies on ketosis showed that women developed fasting ketonemia and ketonuria more rapidly and to a greater degree than men [23]. Small doses of estrogen appear to augment fatty acid production in adipose tissue [24]. Pregnancy is also known to contribute to the development of ketosis, perhaps through a placental hormone or increased fetal use of maternal carbohydrate stores 1251. Although this study does not definitely answer the question of a sexual predilection for this disorder, it appears likely that female alcoholics are at increased risk. There also appears to be a tendency for AKA to recur in certain patients. Nearly one quarter of our population had either a prior history of AKA or a repeat episode during the study period. Whether this represents a genetic disorder of ketone metabolism cannot be answered by this study. Clinical Presentation Nausea, vomiting, and abdominal pain were by far the most common presenting symptoms. Vital sign changes (tachycardia and tachypnea) and abdominal tenderness were the most frequent physical findings. Despite the frequency of abdominal complaints, objective findings other than tenderness on abdominal examination were much less frequent. Abdominal distention, decreased bowel sounds, ascites, or rebound tenderness occurred rarely and only in association with other severe intra-abdominal or extra-abdominal processes such as pancreatitis or sepsis. Altered mental status was not seen in uncomplicated cases of AKA. In our patients, altered mental status was always explained by a concomitant pro-
Volume 91
ALCOHOLIC
ALCOHOLIC
1.
3.
/ WRENN
2.
ETOH + acetaldehyde
stores
t NADHINAD
Extracellular, intravascular -DT volume depletion
ET AL
KETOACIDOSIS
Starvation
Jen
KETOACIDOSIS
f catecholamines cortisol f growth hormone
-...-.-..,. I -..
13
fiat& acid prod&ion
T carnitine ---+ acyltransferase
-+
+ acetate
-
:’
S-oxidation
citric acid cycle
)
triglycerides
i beta-hydroxybutyrate
Figure 2. pathophysiology
Simplified of AKA.
scheme
of
-
acetoacetate -d
cess such as intoxication, hypoglycemia, or a cerebrovascular accident. Fever was uncommon and its presence was also an indication of another concomitant disorder. In the two patients with fever, a seizure due to alcohol withdrawal or an intracerebral hemorrhage appeared to be responsible for the fever. None of our patients who were eventually documented to have a significant infection presented with a fever. Hypothermia was often a marker of a concomitant process such as hypoglycemia or severe acidosis. Pathophysiology AKA has a complex pathophysiology. A simplified scheme is presented in Figure 2. The main predisposing events appear to be a relative deficiency of insulin coupled with an excess of glucagon caused by (1) starvation with glycogen depletion, (2) an elevated NADHLNAD ratio secondary to alcohol’s metabolism by alcohol dehydrogenase [ 141, and (3) extracellular fluid volume depletion. The core event is the relative deficiency of insulin due to glycogen depletion, decreased gluconeogenesis, and the a-adrenergic effects of epinephrine in inhibiting its secretion [14]. The elevated NADH/ NAD ratio affects gluconeogenesis, and both starvation and the presence of liver disease affect the ability to store glycogen [8]. Extracellular fluid volume depletion eventually stimulates the release of counterregulatory hormones such as growth hormone, cortisol, epinephrine, and, secondarily, glucagon [4,7,26]. Insulin deficiency promotes lipolysis and the liberation of free fatty acids [4]. Delivery of free
fatty acids to the liver is not, however, sufficient to initiate maximal ketogenesis [2,27,28]. There is a close link between the rate of ketosis and the liver carnitine content [27,28]. Fatty carnitine acyltransferase I and II constitute the carrier system by which free fatty acids are moved from the cytosol into the mitochondria where B-oxidation to the ketoacids BOHB and ACAC can take place [27,29]. Otherwise, the fatty acids would be metabolized to triglycerides via the Kreb’s cycle [29,30]. The activity of fatty carnitine acyltransferase I and II is dependent upon the concentration of malonyl coenzyme A (CoA), which inhibits their activity [29]. The concentration of malonyl CoA is decreased by a combination of decreased insulin and increased glucagon [27,28]. Glycogen depletion is required in order for the increased glucagon/insulin ratio to result in ketoacid production [29,30]. Whether pancreatic injury plays a role in the increased glucagon and decreased insulin levels is speculative, but in early acute pancreatic inflammation, decreased insulin levels combined with increased glucagon, cortisol, and catecholamine levels have been reported [31]. The elevated NADHLNAD ratio in alcoholics favors the formation of BOHB and lactate rather than ACAC and pyruvate. Diminished pyruvate dehydrogenase activity, either as a result of decreased NAD availability or magnesium or thiamine deficiency, also inhibits the utilization of pyruvate in the Kreb’s cycle, yielding more substrate for lactate production [32,33]. Alcohol itself, when chronically ingested with a calorically adequate fat-containing diet, may induce ketogenesis, and large doses of
August
1991 The American
Journal
of Medicine
Volume
91
123
ALCOHOLIC
KETOACIDOSIS
/ WRENN
ET AL
alcohol increase free fatty acid production liver [34].
in the
tients in our series had a negative nitroprusside reaction in their serum and only three of 29 patients had a negative dipstick reaction for ketones in their urine. The urine dipstick reaction is therefore approximately 90% sensitive in screening for AKA whereas the semiquantitative serum nitroprusside reaction is 96% sensitive. In the five patients who had negative serum or urine reactions for ACAC, the syndrome was very mild compared with that in the group as a whole. Serum ACAC was, like BOHB, inversely related to BAL. In the only prior series in which lactate levels were systematically measured, marked elevations of lactate occurred only in patients with shock [8]. Among our patients, 61% had an elevated lactate level. Ten patients had more than modestly elevated lactate levels (greater than 2.8 mmol/L) and all but one of these patients had a detectable BAL, implying continued drinking to within hours of admission. The mean BAL for these 10 patients was 35 mmol/L compared with a mean BAL of 8 mmol/L for the patients with lower lactate levels. Presumably, continued metabolism of ethanol further increases the NADH/NAD ratio, thereby causing increased production and decreased utilization of lactate. Moderate to severe elevations of lactate were seen in five patients, all of whom had serious concomitant disorders (sepsis, pancreatitis, rhabdomyolysis, and hypoglycemia).
Acid-Base Abnormalities Halperin et al [14] emphasized the four types of metabolic acidosis seen in AKA: (1) ketoacidosis; (2) lactic acidosis; (3) acetic acidosis; and (4) “indirect” loss of bicarbonate in the urine. Also noted were the complex interactions of the various types of metabolic acidosis with metabolic alkalosis from extracellular volume contraction. They theorized that volume contraction also contributed to inhibition of insulin release through increases in a-adrenergic activity [14]. Fulop and Hoberman [8] also noted the complexity of the acid-base disorder; half of their patients were not acidemic and almost one third were actually alkalemic. Our series reinforces the fact that AKA is not simply a wide anion gap acidosis due to accumulated ketoacids. Only 55% of the patients were acidemic and 78% had a mixed acid-base disorder. In about one quarter of patients, a primary metabolic alkalosis coexisted with the wide-gap acidosis, and in another quarter of patients, a primary respiratory alkalosis was present. Fifteen percent of patients manifested both a wide-gap acidosis and a hyperchloremic acidosis, while a triple acid-base disorder was seen in 10%. Metabolic alkalosis most likely represents the combined effects of volume contraction and vomiting. Respiratory alkalosis probably reflects underlying alcohol withdrawal, pain, or associated disorders such as sepsis or severe liver disease. Hyperchloremic acidosis probably results from “indirect” renal loss of bicarbonate associated with ketonuria [35,36]. As expected, BOHB levels were elevated in all patients in whom this measurement was performed. In a prior series, only one of 15 patients with a significantly elevated BOHB level also had a detectable BAL [15]. In our series, the BOHB levels correlated inversely with BAL. Possibly having ethanol’s “empty” calories available decreases fatty acid mobilization. In contrast to prior experience, however, many alcoholics in our series continued to ingest enough ethanol, despite vomiting, to maintain a detectable BAL and still develop significant ketoacidosis. Alcohol intoxication by itself does not cause clinically significant elevations of BOHB [37]. In prior series, BOHB levels were elevated out of proportion to ACAC levels, presumably because of an elevated NADH/NAD ratio. Because BOHB is not detected by the nitroprusside reaction normally used with urine samples to detect ketoacids, it has been stressed that a negative dipstick does not rule out AKA [8]. While this is true, only two of 47 pa-
124
August
1991 The American
Journal
of Medicine
Volume
Laboratory Findings Hyponatremia and hypokalemia were both common (Figure 1). Hyponatremia appeared to be related to vomiting and extracellular volume depletion. Hypokalemia correlated with hypomagnesemia (r = 0.39, p = 0.002). Hyperkalemia was seen in 15% of patients and correlated with lower pH values (r = 0.46, p
91
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
gestion that men are predisposed to hypoglycemia in AKA [ll], in our series hypoglycemia was more than three times as common in women. Ethanol was present in every hypoglycemic patient who had a concomitant BAL measured (mean concentration 26 mmol/L), supporting the prediction that blood glucose values should be lower in patients with more recent alcohol intake [14,39]. Prior studies in alcoholics have noted that hypoglycemia appears to be independent of pancreatic injury, liver injury, or changes in circulating insulin and glucagon [9]. Hyperglycemia was seen in 35% of patients. Mild to moderate glucose intolerance has been noted in patients in other series [2,8,15]. When blood glucose values are higher than 13.8 mmol/L, the differentiation of AKA from DKA becomes difficult. We included eight such patients in our series based on the clinical picture and their subsequent course. In chronic alcoholics without demonstrable liver disease, an impaired insulin response to glucose loading has been described [40]. Impaired glucose uptake by tissues with normal insulin responses has also been seen in alcoholic women [41]. Calcium, Phosphorus, and Magnesium Abnormalities Miller et al [9] stressed the importance of both the inhibition of NADH oxidation by the mitochondrial accumulation of acetaldehyde and the loss of mitochondrial phosphorus in contributing to ketosis [42]. In their group of patients, admission hyperphosphatemia was seen in 82%, with a mean admission phosphorus level of 2.2 mmol/L (6.8 mg/dL). In our series, only 25% of patients were hyperphosphatemic on admission but, like the patients in Miller’s series, in every case the phosphorus level fell rapidly with treatment, often to critically low levels. The greater severity of AKA in Miller’s patients (mean pH 7.14) compared with ours (mean pH 7.30) accounts for the differences in admission phosphorus concentrations. Hypophosphatemia was seen on admission in 25% of our patients. We believe our series is more representative of the spectrum of AKA as it presents to the emergency department. Other Serum Abnormalities Since the original description of AKA by Dillon et al [l], the liver has been thought to play a central role in the development of this syndrome. Diseased livers have a decreased capacity to store glycogen and promote gluconeogenesis [12,38,39]. In our patients, evidence of significant liver disease was present in the vast majority of patients. AST and bilirubin levels were elevated in 89% and 72% of patients, respectively.
Serum amylase was elevated in almost half the patients in whom it was measured. A past history of pancreatitis was present in another seven patients with normal serum amylase levels. In all, 64% of patients had evidence of pancreatic injury by history or on laboratory testing, including two patients with elevated urine amylase and normal serum amylase. Given that the serum amylase level is not infrequently within the normal range in the setting of acute pancreatitis in alcoholics [43,44], it is intriguing to speculate on the importance of pancreatic injury from alcohol in inducing the syndrome of AKA. As in AKA, the main symptoms of pancreatitis are nausea, vomiting, and abdominal pain. Anemia was common, but mild, and no correlation with available indices of gastrointestinal bleeding was noted. Leukocytosis was common but seemed more a marker of volume depletion and demargination from physical stress than of infection. Leukopenia, however, was often a marker of severe underlying disease, including infection. Thrombocytopenia was also seen relatively commonly. Concomitant disorders, such as pancreatitis, myopathy, upper gastrointestinal bleeding, seizures, hepatitis, and alcohol withdrawal, were common and undoubtedly contributed to the morbidity of the syndrome. Many patients had more than one associated problem. The case-fatality rate was very low (on the order of I%), and, as in prior series, was related more to associated disorders than to the syndrome of AKA itself, despite pH values below 7.00. The decision to admit the patient to an inhospital area seemed most related to the severity of acidosis. Slightly more than half the patients were admitted; the rest were able to be treated and discharged from the emergency department within 12 hours. Treatment On the basis of prior studies and our experience, we recommend that the treatment of AKA be directed toward reversing the three major pathophysiologic causes of the syndrome, extracellular fluid volume depletion, glycogen depletion, and an elevated NADH/NAD ratio. Extracellular volume repletion with saline is indicated to inhibit the release of counter-regulatory hormones. Whether moderate volume replacement (500 mL/hour for 4 hours followed by 250 ml/hour for 4 hours) as opposed to vigorous volume replacement (1,000 ml/hour for 4 hours followed by 500 ml/hour for 4 hours) would be more beneficial in patients without extreme extracellular volume deficits has not been studied in AKA, but in DKA a regimen of moderate extracellular volume repletion resulted in a more rapid in-
August 1991 The American Journal of Medicine
Volume 91
125
ALCOHOLIC KETOACIDOSIS / WRENN ET AL
crease in serum bicarbonate with a significant reduction in the cost of therapy [45]. Glucose administration may also help interrupt ketogenesis by stimulation of insulin production and secretion. Dextrose-containing solutions should be administered to stimulate the oxidation of NADH and to replete glycogen stores in patients with hypoglycemia or euglycemia (blood glucose level less than 13.8 mmol/L). Miller and colleagues [9] showed that saline alone did not correct the acidosis of AKA as rapidly as dextrose and saline. Initially, dextrose-containing solutions should be avoided in patients with severe hyperglycemia (blood glucose greater than 13.8 mmol/L) because of the risk of worsening hyperglycemia in a state of relative insulin deficiency. Insulin should be used judiciously because of the risk of hypoglycemia in patients with depleted glycogen stores. Insulin in very low doses is indicated only in patients with blood glucose values greater than 13.8 mmol/L. Potassium repletion is indicated in hypokalemic patients and in normokalemic patients with acidemia. These patients should initially receive at least 10 mmol of potassium chloride per hour in the intravenous fluids. Higher rates of infusion may be required for more severe degrees of hyperkalemia as the insulin deficit and acidosis correct and potassium shifts intracellularly [46]. The subsequent rate and total dose of potassium are determined by successive potassium measurements. In the absence of renal insufficiency, magnesium repletion is indicated in all patients, not only to help restore calcium and potassium homeostasis, but also to forestall alcohol withdrawal [47,48]. Magnesium can be safely given at a rate of up to 1 g/hour [49]. Calcium repletion is rarely needed; magnesium repletion will usually allow spontaneous correction of hypocalcemia, if present, to occur [49]. The routine use of phosphorus is not encouraged but phosphorus levels should be followed closely. Routine phosphorus repletion has not been shown to improve morbidity or mortality in DKA [50,51]. When hypophosphatemia is present, intravenous phosphate repletion at a rate of 0.08 to 0.16 mmol/kg over 6 hours is recommended [52]. Adverse effects of more rapid phosphate repletion include hypocalcemia and t&any [53-551. Bicarbonate therapy is rarely if ever needed. In patients with DKA and pH values as low as 6.90, no change in morbidity or mortality has resulted from bicarbonate therapy [56]. In AKA, regeneration of bicarbonate from the metabolism of lactate, ketoacids, and acetate occurs with conservative treatment [14]. Patients’ pH values respond rapidly (within 12 hours) to treatment with volume repletion and glu-
126
August 1991 The American Journal of Medicine
case alone [9]. In our patients, no one required bicarbonate despite pH values as low as 6.95. Thiamine repletion is routinely indicated both to increase pyruvate dehydrogenase activity and to provide prophylaxis against the development of Wernicke’s encephalopathy [57]. It is safe to administer lOO-mg boluses of thiamine intravenously [58]. Because alcoholics are predisposed to multiple diseases that may be associated with AKA, a careful history and physical examination must be performed on even the most straightforward of cases. Perforated peptic ulcer, small bowel obstruction, and other abdominal catastrophes may initially be misdiagnosed as AKA. Similarly, AKA may be an accompanying illness in conjunction with pancreatitis, acute upper gastrointestinal hemorrhage, rhabdomyolysis, alcohol withdrawal, or sepsis. Conclusions AKA is a common disorder in chronically malnourished alcoholics. There is a tendency for the disorder to recur in some patients. Nausea, vomiting, and abdominal pain constitute the predominant symptoms. The physical findings are usually related to vital sign abnormalities and abdominal tenderness. Findings such as altered mental status, abdominal distention, decreased bowel sounds, rebound abdominal tenderness, fever, or hypothermia are not seen in uncomplicated AKA and imply another underlying process. The acid-base disorder of AKA is complex, and double or triple acid-base disorders are common. Primary metabolic alkalosis from vomiting and extracellular volume depletion and respiratory alkalosis from alcohol withdrawal, sepsis, or pain are particularly common. The severity of acidosis is variable and alkalemia is seen occasionally. Not only are BOHB and ACAC levels elevated, but a modest element of lactic acidosis is routinely present. The urine and serum nitroprusside reactions for ketones are almost always positive unless the syndrome is very mild. Both hyperglycemia and hypoglycemia are common. Hyponatremia, hypokalemia, hypomagnesemia, and hypocalcemia may also be expected. Both hyperphosphatemia and hypophosphatemia may be seen on presentation, but a precipitous reduction with treatment should be anticipated. Evidence of pancreatitis and myopathy are often found. A modest macrocytic anemia and leukocytosis are frequently seen. Leukopenia and thrombocytopenia, while uncommon, are markers of severe underlying disease. BALs in the intoxicated range are not unusual. Other associated disorders common to alcoholics frequently accompany AKA
Volume 91
ALCOHOLIC
and should be searched for and treated aggressively. The treatment of AKA centers on the administration of saline and dextrose to reverse extracellular volume depletion and glycogen depletion. Supportive care with electrolyte and vitamin repletion should be instituted. The role of phosphorus in treatment is unresolved but it probably should be reserved for documented hypophosphatemia. Administration of bicarbonate and insulin should generally be avoided. Further focused studies are needed to elucidate more fully the pathophysiology of AKA. Studies employing more sensitive markers of pancreatic injury than amylase would also be useful. In addition, studies addressing the role of routine phosphorus and magnesium therapy are needed.
REFERENCES 1. Dillon ES, Dyer WW, Smelo LS. Ketone acidosis in nondiabetic adults. Med Clin North Am 1940; 24: 1813-22. 2. Jenkins DW, Eckel RE. Craig JW. Alcoholic ketoacidosis. JAMA 1971; 217: 177-83. 3. Weisberger Cl., Hamilton W. Alcoholic ketoacidosis [letter]. JAMA 1971: 217: 1865-6. 4. Levy LJ, Duga J. Girgis M, Gordon EE. Ketoacidosis associated with alcoholism in nondiabetic subjects. Ann Intern Med 1973; 78: 213-9. 5. Edwards WM, Hoyt R. Alcoholic ketoacidosis. South Med J 1973; 66: 281-2. 6. Peterson G. Alcoholism and ketoacidosis [letter]. Ann Intern Med 1973; 78: 983. 7. Cooperman MT, Davidoff F, Spark R, Pallotta J. Clinical studies of alcoholic ketoacidosis. Diabetes 1974; 23: 433-9. 8. Fulop M, Hoberman HD. Alcoholic ketosis. Diabetes 1975; 24: 785-90. 9. Miller PD. Heinig RE, Waterhouse C. Treatment of alcoholic acidosis-the role of dextrose and phosphorus. Arch Intern Med 1978; 138: 67-72. 10. Podratz KC. Alcoholic ketoacidosis in pregnancy. Obstet Gynecol 1978; 52: 54s-7s. 11. Platia EV. Hsu TH. Hypoglycemic coma with ketoacidosis in nondiabetic alcoholics. West J Med 1979: 131: 270-6. 12. Lumpkin JR, Baker FJ II, Franaszek JB. Alcoholic ketoacidosis in a pregnant woman. JACEP 1979; 8: 21-3. 13. Devenyi P. Alcoholic hypoglycemia and alcoholic ketoacidosis: sequential events of the same process? Can Med Assoc J 1982; 127: 513. 14. Halperin ML, Hammeke M, Josse RG. Jungas RL. Metabolic acidosis in the alcoholic: a pathophysiologic approach. Metabolism 1983; 32: 308-15. 15. Fulop M. Ben-Ezra J. Bock J. Alcoholic ketosis. Alcoholism (NY) 1986; 10: 610-5. 16. Williamson DH. Mellanby J, Krebs HA. Enzymatic determination of D(-) Bhydroxybutyric acid and acetoacetic acid in blood. Biochem J 1962; 82: 90-6. 17. Henry RJ. Clinical chemistry-principles and technics. New York: Harper and Row. 1968: 814-5. 18. Nie NH, Hull CH, Jenkins JG, et al. SPSS: statistical package for the social sciences. 2nd ed. New York: McGraw-Hill, 1986. 19. Dixon WJ, Brown MB. BMDP-85, biomedical computer programs, P-series. Berkeley: University of California Press, 1985. 20.Albert MS, Dell RB. Winters RW. Quantitative displacement of acid-base equilibrium in metabolic acidosis. Ann Intern Med 1967; 66: 312-22. 21. Blume SB. Women and alcohol-a review. JAMA 1986; 256: 1467-70. 22. Moore RA. The prevalence of alcoholism in a community general hospital. Am J Psychiatry 1971; 128: 638-9. 23. Deuel HJ Jr, Gulick M. Studies on ketosis I. The sexual variation ketosis. J Biol Chem 1932; 96: 25-34.
in starvation
KETOACIDOSIS
/ WRENN
ET AL
24. McKerns KW. Clynes R. Sex differences in rat adipose tissue metabolism. Metabolism 1961; 10: 165-70. 25. Hartop M. Human chorionic mammosomatotropin and its clinical significance. Clin Endocrinol 1972: 1: 209-18. 26. Unger RH, Orci L. Glucagon and the A cell. Physiology and pathophysiology. N Engl J Med 1981; 304: 1518-24, 1575-80. 27. McGarry JD. Foster DW. Hormonal control of ketogenesis. Biochemical considerations. Arch Intern Med 1977; 137: 495-501. 28. Schade DS. Eaton RP. Glucagon regulation of plasma ketone body concentration in human diabetes. J Clin Invest 1975: 56: 1340-4. 29. Cahill GF Jr. Ketosis. Kidney Int 1981: 20: 416-25. 30. Blachley J. Knochel JP. Alcohol-induced disturbances in electrolyte and acid base homeostasis. in: Kokko JP, Tannen RL. editors. Fluid and electrolyte abnormalities of disease. Philadelphia: WB Saunders, 1986: 515-9. 31. Drew SI. Joffe B. Vinik A, et a/. The first 24 hours of acute pancreatitis -changes in biochemical and endocrine homeostasis in patients with pancreatitis compared with those in control subjects undergoing stress for reasons other than pancreatitis. Am J Med 1978; 64: 795-803. 32. Kreisberg RA. Lactate homeostasis and lactic acidosis. Ann Intern Med 1980; 92: 227-37. 33. Flink EB. Role of magnesium depletion in Wernicke-Korsakoff syndrome. N Engl J Med 1977; 294: 1367-70. 34. Lefevre A, Adler H, Lieber CS. Effect of ethanol on ketone metabolism. J Clin Invest 1970; 49: 1775-82. 35. Halperin ML, Bear RA, Hannaford MC, Goldstein MB. Selected aspectsof the pathophysiology of metabolic acidosis in diabetes mellitus. Diabetes 1981; 30: 781-7. 36. Hilton JG. Vandenbroucke AC, Josse RG. Buckley GC, Halperin ML The unne anion gap: the critical clue to resolve a diagnostic dilemma in a patient with ketoacidosis. Diabetes Care 1984; 7: 486-90. 37. Fulop M. Bock J. Ben-Ezra J, Antony M, Danzig J. Gage JS. Plasma lactate and 3-hydroxybutyrate levels in patients with acute ethanol Intoxication. Am J Med 1986; 80: 191-4. 38. Freinkel N. Arky RA. Singer DL, et al. Alcohol hypoglycemia IV: current concepts of its pathogenesis. Diabetes 1965; 14: 350-61. 39. Williams HE, Mortimer GE. Studies on the mechanism of ethanol-induced hypoglycemia. J Clin Invest 1963; 42: 497-506. 40. Sereny G, Endrenyi L. Mechanism and significance of carbohydrate Intolerance In chronic alcoholism. Metabolism 1978; 27: 104-6. 41. Dornhorst A, Ouyang A. Effect of alcohol onglucose tolerance. Lancet 1971; 2: 957-9. 42. Gregg CT, Lehninger AL Dependence of respiration on phosphate and phosphate acceptor in submitochondrial systems Ill. Sonic fragments. Biochim Biophys Acta 1963; 78: 27-44. 43. Spechler SJ. Dalton JW. Robbins AH, et al. Prevalence of normal serum amylase levels in patients with acute alcoholic pancreatitis. Dig Dis Sci 1983; 28: 865-9. 44. Clavier PA, Robert J. Meyer P, et a/. Acute pancreatitis and normoamylasemia-not an uncommon combination. Ann Surg 1989; 210: 614-20. 45. Adrogue HJ. Barrero J, Eknoyan G. Salutary effects of modest fluid replacement in the treatment of adults with diabetic ketoacidosis-use in patients without extreme volume deficit. JAMA 1989; 262: 2108-13. 46. Kruse JA, Carlson RW. Rapid correction of hypokalemia using concentrated intravenous potassium chloride infusions. Arch Intern Med 1990; 150: 613-7. 47. Whang R, Flink EB, Dyckner T. Webster PO, Aikawa JK. Ryan MP. Magnesium depletion as a cause of refractory potassium repletion. Arch Intern Med 1985: 145: 1686-9. 48. Wolfe SM. Victor M. The relationship of hypomagnesemla and alkalosis to alcohol withdrawal symptoms. Ann NY Acad Sci 1969; 162: 973-84. 49. Rude RK. Singer FR. Magnesium deficiency and excess. Annu Rev Med 1981; 32: 245-59. 50. Wilson HK, Kener SP. Lea S. Boyd AE Ill, Eknoyan G. Phosphate therapy in diabetic ketoacidosis. Arch Intern Med 1982; 142: 517-24. 51. Fisher JN, Kitabchi AE. A randomized study of phosphate therapy in the treatment of diabetic ketoacidosis. J Clin Endocrinol Metab 1983; 57: 177-80. 52. Lentz RD, Brown DM, Kjellstrand M. Treatment of severe hypophosphatemia. Ann Intern Med 1978; 89: 941-4. 53. Nanji AA. Symptomatic hypocalcemia due to hyperphosphatemia with alcoholic ketoacidosis [letter]. South Med J 1984: 77: 540.
August
1991
The American
Journal
of Medicine
Volume
91
127
ALCOHOLIC KETOACIDOSIS / WRENN ET AL 54. Winter RJ, Harris CJ. Phillips LS, et a/. Diabetic ketoacidosis-induction of hypocalcemia and hypomagnesemia by phosphate therapy. Am J Med 1979; 67: 897-900. 55. Zipf WB, Bacon GE, Spencer ML, et a/. Hypocalcemia. hypomagnesemia. and transient hypoparathyroidism during therapy with potassium phosphate in diabetic ketoacidosis. Diabetes Care 1979; 2: 265-8.
128
August 1991 The American Journal of Medicine
56. Morris LR, Murphy MB, Kitabachi AE. Bicarbonate therapy in severe diabetic ketoacidosis. Ann Intern Med 1986; 105: 8X-40. 57. Marx JA. The varied faces of Wernicke’s encephalopathy. J Emerg Med 1985; 3: 411-3. 58. Wrenn KD. Slovis CM, Murphy F. A toxicity study of parenteral thiamine hydrochloride. Ann Emerg Med 1989; 18: 867-70.
Volume 91