Very low calorie diets and hypokalaemia: the importance of ammonium excretion

Transcription

1 Nephrol Dial Transplant (2005) 20: doi: /ndt/gfh582 Teaching Point (Section Editor: K. Ku hn) Supported by an educational grant from Very low calorie diets and hypokalaemia: the importance of ammonium excretion Tane Liu 1,2, Glenn T. Nagami 1,2, Marcia L. Everett 1 and Barton S. Levine 1,2 1 Nephrology Section and Departments of Medicine VAGLAHS West Los Angeles and 2 The David Geffen School of Medicine at UCLA, Los Angeles, CA, USA Keywords: acute renal failure; ammonia; fasting; hypokalaemia; ketoacidosis; metabolic alkalosis Introduction besity has become a worldwide epidemic and is associated with significant morbidity and mortality. More than two-thirds of adults in the United States are either trying to lose weight or prevent weight gain. Very low calorie diets, diets with energy levels between 200 and 800 kcal/day, are used commonly for rapid weight loss. These diets are generally safe and well-tolerated and have been reported to improve blood pressure, glycaemic control and lipid profiles. Minor side effects, including fatigue, constipation, nausea and diarrhoea, often occur, but serious adverse events, including death, may result if the very low calorie diets are not used appropriately [1]. Hypokalaemia may occur with very low calorie diets but is usually mild [1,2]. Below, we present a case of severe hypokalaemia, which developed after 2 weeks of a very low calorie diet. The pathophysiological mechanisms that lead to the development of significant potassium depletion are discussed. Case A 61-year-old man was admitted to the inpatient service of the VAGLAHS for hypokalaemia. A serum potassium of 2.8 meq/l was obtained during a Correspondence and offprint requests to: Barton S. Levine, Nephrology Section (111L), VAGLAHS West Los Angeles, Wilshire Boulevard, Los Angeles, CA 90073, USA. barton.levine@med.va.gov routine outpatient visit [Table 1 (18 November 2002) and Figure 1]. He was instructed to return to the emergency room where he received 100 meq KCl over an 8 h period. His serum potassium remained at 2.9 meq/l, prompting his admission. The patient was morbidly obese with a body mass index of 43.6 kg/m 2. He had several medical problems associated with obesity, including hypertension, diabetes mellitus, hyperlipidaemia, obstructive sleep apnoea, microalbuminuria and osteoarthritis. ther past history included bipolar disorder and benign prostatic hypertrophy. In an attempt to lose weight, he self-initiated a low calorie, high protein diet 15 days prior to admission (Medifast Inc, winop Mills, Maryland, USA; see Table 2 for contents). He was ingesting three to six servings of Medifast as his sole energy source and denied eating other food. He achieved a weight loss of 13.6 kg during the 2 week period. The patient was asymptomatic at the time of his outpatient visit. He denied a history of vomiting, diarrhoea or excessive sweating. His daily medications included fosinopril (20 mg), hydrochlorothiazide (25 mg), potassium chloride (20 meq), felodipine (10 mg), lovastatin (40 mg) and aspirin (325 mg). The patient had been on the same therapeutic regimen for >1 year and before this episode of hypokalaemia, the serum potassium had been within or near the normal range (Figure 1). n physical examination, vital signs were normal and, aside from his morbid obesity and 2þ oedema of the lower extremities, the remainder of his examination was unremarkable. His serum and urine chemistries and an arterial blood gas are shown in Table 1. f note, he was hypokalaemic, his anion gap was 19 but his bicarbonate was elevated and he had a mild alkalaemia. His calculated transtubular potassium gradient (TTKG) was 16 and his serum ketones were positive the day after admission. Random plasma renin activity was ß The Author [2005]. Published by xford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please journals.permissions@oupjournals.org

2 Very low calorie diets and hypokalaemia 643 Table 1. Serum and urine chemistries and arterial blood gas (ABG) 16 months prior to admission 12 months prior to admission admission a after admission Serum Na (meq/l) K (meq/l) Cl (meq/l) C 2 (mmol/l) BUN (mg/dl) Creatinine (mg/dl) Glucose (mg/dl) smolality (mosm/kg) 296 Anion gap (meq/l) Urine Na (meq/l) <10 K (meq/l) 83 Cl (meq/l) 22 Creatinine (mg/dl) 377 smolality (mosm/kg) 538 TTKG 16 ABG ph/pc 2 (U/mmHg) 7.45/49.2 HC 3 /P 2 (mmol/l/mmhg) 34.3/58.1 Lactate (mmol/l) 0.9 a Date of admission. Admission Table 2. Content of Medifast meq/l Serum HC3 Serum K 1 Pack 4.5 Packs Protein (g) Carbohydrate (g) Fat (g) Calories Na (mmol) K (mmol) ng/ml/h and serum aldosterone was 13.5 ng/dl. His electrocardiogram was notable for T-wave inversion in leads I, AVL and V4 V6 and Q waves in V1 and V2. ne day after admission, the patient agreed to start a regular diet and intravenous fluids were given to correct his volume depletion and hydrochlorothiazide was discontinued. The patient was given a total of 220 meq KCl for the first 24 h after admission and his serum potassium rose to 3.1 meq/l. The patient continued to require 120 meq of supplemental KCl daily for the next 3 days after which his serum potassium normalized (Figure 1). Days Fig. 1. Serial values for serum potassium and bicarbonate. Arrow indicates date of admission. Discussion The patient in this report, despite taking potassium supplements, developed severe hypokalaemia 2 weeks after initiating a very low calorie diet. The large amount of potassium required to correct the hypokalaemia suggested a large body deficit of potassium. Several factors probably contributed to potassium depletion in this patient. Hypokalaemia occurs commonly with the use of thiazide diuretics and is usually apparent during the first weeks of therapy. Thereafter, the serum potassium remains stable unless other disturbances of potassium homeostasis are superimposed. This patient had been taking hydrochlorothiazide for several years, but only manifested borderline hypokalaemia until the present admission. Mild hypokalaemia is also common after fasting or ingesting a low calorie diet. Significant renal losses of potassium occur during the first week of fasting due to the release of intracellular potassium resulting from tissue catabolism, in conjunction with obligatory losses from the urinary excretion of ketoacid anions generated during fasting; the excretion of ketoacid anions in the urine obligates the excretion of

3 644 T. Liu et al. Table 3. Equations and calculations for the UAG, UG, DPI, UUN and 24 h urine volume 1. UAG ¼ [Na þ ]u þ [K þ ]u [Cl ]u a. UAG ¼ (5 þ 83) 22 ¼ DPI a ¼ 6.25(UUN þ NUN) 3. UUN ¼ (DPI 6.25 NUN)/6.25 a. UUN ¼ (63 [ ])/6.25 ¼ UG ¼ [Uosm]m [Uosm]c 5. [Uosm]c ¼ 2([Na þ ]u þ [K þ ]u) þ Gu (mmol/l) þ Uu (mmol/l) a. [Uosm]c ¼ 2(5 þ 83) þ 0 þ 352 ¼ 528 b. UG ¼ ¼ Estimate of 24 h urine volume (UV) ¼ 100 (Cr) 24 h /(Cr) spot a. UV ¼ /377 ¼ 577 ml a Assuming there are no urinary or gastrointestinal losses of protein. [Na þ ]u, concentration of urine sodium (meq/l); [K þ ]u, concentration of urine potassium (meq/l); [Cl ]u, concentration of urine chloride (meq/l); [Uosm]m, measured urine osmolality (mosm/kg); [Uosm]c, calculated urine osmolality (mosm/kg); Gu, concentration of glucose in urine (mmol/l); Uu, concentration of urea in urine (mmol/l); UUN (g/day); NUN, non-urea nitrogen [estimated as bodyweight (g/day)]; (Cr) 24 h, 24 h urine creatinine excretion (mg/day); (Cr) spot, creatinine concentration in spot urine (mg/dl). an accompanying cation, such as sodium or potassium, to maintain electroneutrality. Consistent with the presence of ketosis in this patient was the elevated serum anion gap of 19 on admission, probably reflecting the presence of ketoacids (acetoacetate and/or b-hydroxybutyrate), and a positive test for urine ketones. In addition, the high urine anion gap (UAG; Table 3, equation 1) suggested the presence of unmeasured anions. With fasting, continued urinary excretion of potassium along with ketoacid anions could eventually lead to profound hypokalaemia. However, after the first week of fasting the renal losses of potassium are usually minimized as ketones are excreted with ammonium generated in response to the ketoacidosis, rather than with sodium or potassium [2,3]. As a result, serum potassium levels only fall slightly below normal. In the present case, severe hypokalaemia occurred after the initiation of the low caloric diet, even though the patient was ingesting 20 meq/day of supplemental KCl in addition to an average intake of 51 meq/day of potassium from the Medifast supplement (assuming an average intake of 4.5 packs of Medifast). Furthermore, the patient was on fosinopril, an agent that blunts diuretic-induced hypokalaemia through its effect on the renin angiotensin aldosterone system. Why then did this patient s serum potassium drop to 2.8 meq/l? From the data at hand it is clear that the severe hypokalaemia observed in this case resulted from excessive renal losses of potassium, as evidenced by a high TTKG [4] and the lack of any history of gastrointestinal losses. What is unusual in this case is that excessive renal potassium excretion continued after the initial week of the low caloric diet, when, as already mentioned, renal potassium losses are minimized with fasting and the TTKG is low [2,3]. Examination of the urine electrolytes reveals that potassium was being excreted at a high concentration with a non-chloride anion(s), reflecting the presence of ketoacid anions in the urine. Thus, a contributing factor to his continued renal loss of potassium was ketonuria, present 1 day after admission. But, as already mentioned, renal losses of potassium should have been minimal after 2 weeks of the low caloric diet, since at this stage the ketoacid anions should have been predominantly excreted with ammonium [2,3]. As evidenced by his urine electrolytes (Table 1), the patient was excreting minimal amounts of sodium due to volume contraction and a low salt intake, a typical finding after 2 weeks of fasting [2]. Therefore, the only cations that would have been available for excretion with the ketoacid anions were ammonium and potassium; any impairment in ammonium excretion would have obligated the continued renal excretion of potassium with the ketoacid anions and aggravated the hypokalaemia. Was urinary ammonium excretion impaired in this case? To implicate a lack of adequate urinary ammonium for the renal wasting of potassium in this instance, it would be helpful to know the approximate level of urinary ammonium excretion. Direct measurements of urinary ammonium excretion are not available readily in clinical practice. Therefore, means of estimating urinary ammonium excretion have been devised using specific urine indices, the UAG and urine osmolal gap (UG) [5]. The UAG is calculated from a spot urine sample by subtracting the concentration of urine chloride from the sum of the urine sodium and potassium concentrations obtained from the same specimen (Table 3, equation 1). A negative UAG indicates that chloride is being excreted with another cation besides sodium or potassium, namely ammonium. By contrast, a positive UAG in the presence of metabolic acidosis indicates that urinary ammonium excretion is impaired. Caution should be exercised in interpreting the UAG under certain circumstances [5]. If an anion other than chloride is present in the urine in significant amounts, it too must be accompanied by a cation. In this circumstance, ammonium may be excreted with the unmeasured anion, such as ketoacid anions, and the UAG would be positive despite the presence of substantial amounts of ammonium in the urine. Therefore, when an unmeasured anion is present in the urine in significant quantities the UAG cannot be used to accurately estimate ammonium excretion. The UG (Table 3, equation 4) is calculated by subtracting the calculated urine osmolality ([Uosm]c; Table 3, equation 5) from the measured urine osmolality ([Uosm]m). The UG/2 provides an estimate of the urine ammonium excretion, but is not subject to many of the vagaries that muddle the interpretation of the UAG, such as ketonuria. A drawback to the UG is that urine urea and glucose concentrations, necessary for the [Uosm]c, are not measured routinely. With both hypokalaemia and a high anion gap metabolic acidosis, a marked increase in urinary ammonium excretion would be expected. But the UAG was markedly positive in this case (Table 3, equation 1a),

4 Very low calorie diets and hypokalaemia 645 consistent with an inappropriately low level of ammonium excretion. However, the patient was also excreting ketones when the spot urine electrolytes were obtained, making the interpretation of the UAG problematic for the reasons alluded to earlier. In this situation the UG would have provided more information, but no urine urea or glucose measurements were obtained. In calculating UG the glucose term can be ignored if the dipstick reading for glucose is negative. An estimate of urea excretion and 24 h urine volume can be made to obtain a urea concentration. Two independent methods were used to estimate urine urea excretion. In the first, his glomerular filtration rate (GFR) was assumed to be reduced given his pre-renal state. His blood urea nitrogen was 29 mg/dl or mmol/l of urea. If his GFR was decreased to 60 ml/min due to volume contraction (assuming his GFR declined 50% given his baseline serum creatinine was 0.9 mg/dl and his serum creatinine was mg/dl when the urine was collected), then over the 24 h period he would have filtered 86.4 l/day and 894 mmol/day of urea. Given his volume-contracted state and that he was on a diuretic, it can be assumed that he had a fractional excretion (FE) of urea of 30% [6] and, therefore, his total daily urea excretion was at 268 mmol; if his FE urea were 25% or 20% his total urea excretion would be mmol/day and mmol/day, respectively. Another estimate of his urea excretion can be obtained utilizing the formula for evaluating dietary protein intake (DPI; Table 3, equation 2) and rearranging it to solve for urine urea nitrogen (UUN; Table 3, equation 3). Since he was ingesting three to six packs of Medifast per day as his sole dietary source of protein it can estimated that he was ingesting, on average, 63 g of protein per day (4.5 packs of Medifast per day). Using his body weight of 142 kg, his estimated UUN is 5.7 g/day (Table 3, equation 3a), which is equivalent to a urine urea of 203 mmol/day. Since the patient was likely catabolic, the estimate using the DPI formula would tend to underestimate urea excretion. The 24 h urine output of the patient on the day the spot urine was obtained can be estimated using a 24 h urine creatinine of 2100 mg measured 2 days later. Since the creatinine excretion should be similar on any given day in the steady state, the urine output on the day of the spot urine can be estimated by dividing the creatinine in the 24 h urine by the spot urine value (Table 3, equation 6). Using this formula his estimated urine volume on the day in question was 577 ml/24 h (Table 3, equation 6a). This value is probably an overestimate of his urine output since his serum creatinine was still decreasing during the 24 h urine collection, which would result in a higher level of creatinine excretion. The low urine output would be consistent with his pre-renal state. Using the estimates of daily urea excretion and urine volume, the concentration of urine urea would be 268 mmol/0.577 l or 464 mmol/l urea if the FE urea were 30%, 387 mmol/l if the FE urea were 25% or 310 mmol/l if the FE urea were 20%. Using the DPI estimated urea, the concentration of urea would be 203 mmol/0.577 l or 352 mmol/l urea. The UG can now be calculated using the spot urine electrolytes (Table 2) and the estimated concentrations for urea. Assuming an FE urea of 25%, his estimated UG would be 538 (180 þ 387) or 29, while using the 20% estimate it would be 48. Using the data obtained from the DPI formula, the UG would be 10 (Table 3, equation 5a,5b). The urine ammonium excretion, calculated as the UG/2, would be low regardless of which estimate is used, leaving potassium as the sole cation available to accompany the excretion of the ketoacid anions. Why might this patient s ammonium excretion have been impaired? In this case several factors were present that could impair ammonia production and/or excretion, including pre-renal azotaemia resulting in a decrease in distal sodium delivery, ketosis, disruption of the angiotensin-ii system and metabolic alkalosis. f these, the most important factor was the development of metabolic alkalosis, as evidenced by his alkalaemia and elevated bicarbonate level during the admission in question [7]. Although his bicarbonate level had been elevated chronically, this appears to have been due to a chronic respiratory acidosis, as evidenced by an arterial blood gas obtained 1 month after the episode of severe hypokalaemia (ph 7.38; PC mmhg; P mmhg; HC meq/l). The acute metabolic alkalosis may have arisen after he initiated a low sodium chloride intake along with the continued use of a diuretic resulting in volume depletion. The acute metabolic alkalosis would have suppressed his ammonium excretion, hence aggravating potassium excretion and his hypokalaemia. In turn, the hypokalaemia may have further aggravated his metabolic alkalosis. Teaching points (i) Patients on very low calorie diets are predisposed to develop renal potassium wasting mediated, in part, by the need to excrete a cation with the increased production and excretion of ketoacid anions. (ii) In most individuals potassium wasting is limited by the excretion of ammonium instead of potassium. Conditions that impair ammonium excretion, such as metabolic alkalosis, may result in a higher degree of renal potassium loss. (iii) Pre-existing or concurrent conditions, such as the use of thiazide diuretics, which may cause potassium depletion in themselves, further predispose patients to more severe degrees of hypokalaemia. (iv) Therefore, patients who initiate a very low caloric diet require careful monitoring of their serum potassium, should avoid the use of concomitant

5 646 T. Liu et al. potassium wasting diuretics and should be given appropriate potassium supplementation so that potassium intake is meq/day. Acknowledgements. We wish to acknowledge Dr Arnold Felsenfeld for his careful review of the manuscript and his editorial assistance. Conflict of interest statement. None declared. References 1. Stevens A, Robinson DP, Turpin J, Groshong T, Tobias JD. Sudden cardiac death of an adolescent during dieting. South Med J 2002; 95: Lin SH, Cheema-Dhadli S, Wowrishankar M, Marliss EB, Kamel KS, Halperin ML. Control of excretion of potassium: lessons form studies during prolonged total fasting in human subjects. Potassium excretion in prolonged fasting. Am J Physiol 1997; 273: F796 F Kamel KS, Lin SH, Cheema-Dhadli S, Marliss EB, Halperin ML. Prolonged total fasting: a feast for the integrative physiologist. Kidney Int 1998; 53: Kamel KS, Quaggin S, Seich A, Halperin ML. Disorders of potassium homeostasis: an approach based on pathophysiology. Am J Kidney Dis 1994; 24: Halperin ML, Margolis BL, Robinson LA et al. The urine osmolal gap: clue to estimate urine ammonium in hybrid types of metabolic acidosis. Clin Invest Med 1988; 11: Carvounis CP, Nisar S, Guro-Razuman S. Significance of the fractional excretion of urea in the differential diagnosis of acute renal failure. Kidney Int 2002; 62: Lemann J, Jr, Lennon EJ, Goodman D, Litzow JR, Relman AS. The net balance of acid subjects given large loads of acid or alkali. J Clin Invest 1965; 44: