Correction of Dialysis-Induced Metabolic Alkalosis

Our patient had primary metabolic alkalosis evident by her high pH of 7.56 and high total CO₂ of 36 mEq/L. The serum total CO₂ reflects the metabolic status more accurately than the blood gas bicarbonate, which is prone to calculation error by the Henderson-Hasselbalch equation. Her respiratory compensation for the metabolic alkalosis was appropriate, with an increase of arterial PaCO₂ to 47 mm Hg (expected PaCO₂ = 40 + 0.7[HCO₃-24] = 48.4). She had normal baseline PaCO₂ in the weeks prior to admission. Due to lack of residual urine output, < 50 mL/d, her metabolic alkalosis could not be attributed to the generation of bicarbonate by the kidneys, or ameliorated by dumping of excess bicarbonate, which explained why her metabolic alkalosis was severe. On the other hand, respiratory distress may have hindered the appropriate CO₂ retention.

In patients with ESRD on HD who have no residual urine output, causes of metabolic alkalosis are limited to loss of net acid or gain of alkali through the gastrointestinal tract; our patient had none of these. Similarly, all renal causes of metabolic alkalosis are not applicable to our patient, including mineralocorticoid excess and contraction alkalosis. In patients with preserved kidney function, loop diuretics can induce alkalosis through enhanced tubular absorption of HCO₃. While acetazolamide can mitigate this scenario by blocking carbonic anhydrase in the luminal border of the collecting ducts resulting in excretion of bicarbonate in the urine, our patient had negligible urine output despite being on furosemide 20 mg twice daily, making this an unlikely cause.

Severe metabolic alkalosis in dialysis patients has been reported with cocaine use, pica ingestion, and citrate load as in plasma exchange, massive transfusions, and regional anticoagulation.^2,8-11 Although calcium carbonate intake can contribute to alkalosis, her small daily dose of 1,200 mg contains approximately 12 mEq of carbonate, which is not a significant contributor to the alkalosis.

With all other causes excluded, the metabolic alkalosis in our patient is presumed to result from the bicarbonate-rich dialysate. Since the majority of patients with ESRD are acidotic before dialysis, the dialysate bicarbonate is set at a higher than normal physiologic level to bring the pH close to or even higher than normal after dialysis. The patient had been dialyzed with NaturaLyte as an outpatient, which was set at the dialysis unit default mode of 36 mEq/L. This form of alkalosis has been reported to peak immediately after treatment but in most patients returns to the predialysis acidotic state due to endogenous acid production.^1,4,12 Normally, muscles play a significant role in buffering excess bicarbonate in patients with nonfunctioning kidneys; hence, malnutrition with muscle wasting tends to propagate and maintain alkalosis, as in our patient.

Managing alkalosis in patients on dialysis can be challenging and is often directed at identifying potential causes like overzealous bicarbonate dialysate and addressing comorbidities, especially malnutrition.^6,7 Bicarbonate delivery can be set on dialysis machines as low as 20 mEq/L. However, the reliability of correcting serum bicarbonate by adjusting bicarbonate-based dialysis products is in question as these products deliver additional buffering capacity through mixing and metabolism of acetate, acetic acid, or citric acid (Table 2).

We infused a high volume of sodium chloride during dialysis to create hyperchloremic metabolic acidosis while removing the volume by UF, thereby eliminating more bicarbonate by convection. Normal saline has a pH of 5.5 and a chloride of 154 mmol/L. We have compensated for an inherent lack of flexibility in HD as it is currently practiced: dialysates are virtually all deliberately alkaline because most of the patients coming to HD have varying magnitudes of metabolic acidosis and acidemia. The dialysate concentrate that dilutes to a bicarbonate level of 30 mEq/L would have only a modest effect against this magnitude of metabolic alkalosis that this patient had at dialysis. We have compensated for this structural inadequacy of current HD by repairing the patient’s severe hypochloremic metabolic alkalosis by infusing a hyperchloremic sodium chloride solution and dialyzing off the excess sodium bicarbonate. This is the logical inverse of what usually happens in the severely acidotic patients seen prior to dialysis: dialyzing off an excess of normal saline and repairing the metabolic acidosis by transfer-in of sodium bicarbonate from the dialysate.

Fresenius Medical Care, which provides most HD machines and fluids in the United States, created charts to show the approximate degree that each contributes as additional buffer. That was in response to a class action lawsuit for metabolic alkalosis due to overdelivery of bicarbonate that resulted in alleged cardiac arrests in patients with HD.¹³ Their report cast doubt on the ability of a lower bicarbonate bath to correct metabolic alkalosis in a predictable fashion.¹ We accordingly showed that normal saline delivery is a reliable option to promptly lower serum bicarbonate level. However, this is a temporary measure and long-term bicarbonate delivery during dialysis needs to be addressed.