Amy Carpenter Aquino
October 2022—William E. Winter, MD, D(ABCC), is blunt about whether to report a corrected sodium: He would worry if his name were on such a report.
“I think you have to be careful about formulas,” he said in his “hot topic” talk at the AACC meeting in July. “Whenever you’re doing an estimate, there are errors.”
He “kind of flinches,” he said, “when discussing calculations that are not well documented.”
In the case of sodium and diabetic ketoacidosis (DKA), the concept behind the formula is to reveal what the sodium would be if the patient were not hyperglycemic. “I don’t have a problem with the clinician doing that,” he says of the formula to calculate the serum or plasma sodium in DKA patients to determine the sodium deficit. “But I would be very worried if my name was on the report for a corrected sodium. I don’t believe that laboratories should be in the business of providing such tentative extrapolations.”
How to report and interpret sodium measurement isn’t always well understood, Dr. Winter, professor of pathology and pediatrics, University of Florida College of Medicine, tells CAP TODAY. “And if you don’t interpret the sodium correctly and don’t understand its ramifications, it could lead to misdiagnosis or mismanagement.” It’s also one of the most commonly measured analytes, “so we’d better do it right,” he said.
The sodium reference interval for adults 18 and over is 135 to 145 mEq/L. Abnormalities of sodium—hyponatremia (<135 mEq/L) or hypernatremia (>145 mEq/L)—are common, especially in hospitalizations. “They’re more common than abnormalities of potassium,” said Dr. Winter, who is also medical director of clinical laboratory support services (phlebotomy) and point-of-care testing.
Sodium measurements are also commonly performed in outpatients, he said. In one study of community subjects from the population-based Rotterdam Study, 15 percent of patients over age 75 had sodium abnormalities. For those 55 to 64, the prevalence was 9.8 percent, and for ages 65 to 74, nine percent (Liamis G, et al. Am J Med. 2013;126[3]:256–263).
Sodium abnormalities are associated with increased morbidity and mortality in cardiac ICU patients. In a retrospective study of 9,676 adult patients admitted to the Mayo Clinic cardiac ICU between 2007 and 2015, the authors found that hyponatremia was associated with an elevated risk of in-hospital mortality, but hypernatremia was not, whereas both were associated with a higher risk of post-discharge mortality (Breen T, et al. J Am Heart Assoc. 2020;9[2]:e014140).
Dr. Winter presented the session’s attendees with a challenge: A patient with hypoalbuminemia, massive proteinuria, massive hyperlipidemia, and pitting edema has a sodium concentration of 126 mEq/L. The physician asks if the patient is truly hyponatremic. “How should you respond?”
The answer depends on how the sodium concentration was measured, he said, noting that the case is one of nephrotic syndrome. In nephrotic syndrome, massive proteinuria causes hypoalbuminemia, which can cause a hyperlipidemia and low oncotic pressure (from the hypoalbuminemia) in which water moves from the vasculature into the interstitium, causing the pitting edema. “But the patient’s hyponatremic,” he said.
Dr. Winter
Sodium concentration in the blood is measured most commonly by direct or indirect ion-selective electrode methods. For the direct method, there’s no dilution of the sample, and the typical analyzer is a whole blood analyzer. “What we do most commonly is measure sodium on the automated analyzer”—an indirect method in which the sample of serum or plasma is diluted before analysis.
When the test is performed on whole blood, there is no physical separation of plasma from the cells. “It measures the true sodium activity in the aqueous fraction of plasma,” Dr. Winter said, “and this will exclude lipids and proteins.” The direct sodium measurements are corrected using a fixed (ion-specific) multiplier to report a concentration that agrees with that of indirect methods “and because flame emission spectrophotometry, an indirect method, is a reference method,” he said. The “uncorrected” direct method for sodium measurement is about five percent higher than the indirect method measurement.
“Does the method matter?” he asks. Usually not, but “under special circumstances there are major differences between the direct and indirect methods that could have clinical implications.”
With the indirect ISE method, serum or plasma is separated from the blood cells by centrifugation, he said. “If it’s serum, you have allowed that sample to clot.” This measures the sodium activity in an aqueous solution of diluted plasma or serum (mmol/kg H2O). Normally, the solution is about 93 percent water plus the dissolved electrolytes, and about seven percent solids, which are the proteins. “If you take that sample and dilute it, the results are corrected to report a concentration to agree with flame photometry.”
In this case of the hyponatremic patient with nephrotic syndrome, would it matter if the sodium was measured by an indirect ISE method? “If you have a normal proportion of solids to water in plasma—whether the solids are lipids or proteins—you’re going to get a clinically appropriate value,” Dr. Winter said.
However, if there is massive hyperlipidemia or massive proteinuria or both, the indirect method will measure a lower sodium, he explained, because hyperlipidemia, a characteristic of nephrotic syndrome, reduces the proportion of plasma that is water, thereby reducing the measured sodium concentration. “This is the dilutional error,” also referred to as electrolyte exclusion effect or volume displacement effect, he said. The consequence is a sodium measurement error: “The sodium measured by the indirect ISE is falsely low,” Dr. Winter said (Koch CD, et al. Clin Chim Acta. 2021;520:63–66).
“This is pseudohyponatremia because of the hyperlipidemia,” he said.
Other analytes can be affected, he said, but it’s usually not a clinically significant difference compared with the effect on the sodium measurement because the sodium reference interval is so narrow. “Your reference interval for potassium could be 3.5 to 5.2 [mEq/L], so a little bit of an error there is not going to have the same magnitude of the effect as an error would for sodium.”
If the range of sodium is 10 mEq/L, for example, and it’s divided by the mean of the reference interval (140 mEq/L), the variation between high and low is only 7.1 percent. For potassium (RI: 3.5–5.2 mEq/L), the percent range of reference interval is 39 percent [percent range = (1.7 mEq/L ÷ 4.35 mEq/L) × 100]. So it’s easier to “hit” the target (“normal range”), per se, for potassium than for sodium when measured by an indirect method, he said.
If pseudohyponatremia is suspected, the only way to find the true sodium measurement is to measure sodium directly, “so you’d ask that a whole blood specimen be sent. In our institution, we’d send it for a blood gas.”
In this case, the patient’s diagnosis was nephrosis, “and there the triglycerides can go into the thousands and cholesterol can easily go into the hundreds,” he said. If interference studies are done with Intralipid (fat emulsion), “that may not reflect what’s happening in vivo because Intralipid is not equivalent to elevations in endogenous samples” (Koch CD, et al. Clin Chim Acta. 2021;520:63–66).
Pseudohyponatremia can also occur in the clinical scenario of massive hyperproteinemia—in patients with myeloma or Waldenström macroglobulinemia, he said. “It is unusual to see a patient with myeloma or Waldenström’s with a protein of 10 or 12 mg/dL. But it’s the same concept where if you have solid material in indirect measurement of sodium, you’re going to have a pseudohyponatremia that is truly wrong.”
Katrangi, et al., studied the prevalence of clinically significant differences in sodium measurements due to abnormal protein concentrations using an ISE method (Katrangi W, et al. J Appl Lab Med. 2019;4[3]:427–432). “These folks found an average difference where the sodium by the indirect method was about 6 mmol/L lower than the direct measurement when the total protein was 8 g/dL or greater,” Dr. Winter said.
The authors wrote: “Clinical laboratories need to be aware of the impact electrolyte exclusion by total protein has on their indirect ion-selective electrode method for measuring serum or plasma sodium. This work reported that 70% of samples with total protein concentration above the reference interval had a considerably decreased sodium concentration compared with results obtained using a direct ISE method. Increased total protein concentration was observed in 1.2% of patient results reported in an academic medical center, thereby making them susceptible to pseudohyponatremia.” They added, “Sodium and total protein concentrations were ordered together in only <11% of cases, thereby making a laboratory middleware solution ineffective at detecting pseudohyponatremia.”
Direct ISE sodium measurements aren’t always the solution, Dr. Winter said. Weld, et al., studied whether direct ISE plasma sodium measurements in the laboratory (Vitros Fusion) and at the point of care in the ICU (ABL 700) show sufficient agreement to be clinically interchangeable (Weld BA, et al. J Clin Monit Comput. 017;31[5]:1103–1109). They found the mean difference was 1.74 mmol/L (range, -14 to +10 mmol/L) when the mean total protein was 5.7 g/dL, Dr. Winter said. Disagreement at ≥|4|mmol/L, ≥|3|mmol/L, and ≥|2|mmol/L was present in 4.1 percent, 13.4 percent, and 36.2 percent of pairs respectively, the authors reported. They wrote, “Disagreement was sufficient to jeopardize safe interchangeable interpretation in situations with a low tolerance for imprecision, such as hyponatremia correction.”
“Not all direct ISEs are created equal,” Dr. Winter said.
As for the influence of hemolysis on analytical measurements of sodium, he cited two studies and summed up this way: “You can argue that a little bit of hemolysis doesn’t have an effect, but a lot of hemolysis does have an effect” (Koseoglu M, et al. Biochem Med [Zagreb]. 2011;21[1]:79–85; Lippi G, et al. Clin Chem Lab Med. 2006;44[3]:311–316).
Dr. Winter presented another patient scenario: A nonlipemic sample for a stat basic metabolic panel is sent to the core laboratory from the adult emergency department. An endocrinologist interprets the sodium result of 128 mEq/L—with a glucose of 640 mg/dL (RI: 70–99 mg/dL, fasting)—as pseudohyponatremia. Clinically this patient is dehydrated, tachycardic, and tachypneic and has a history of polyuria and polydipsia. ABG with electrolytes are as follows: pH = 7.1 (RI: 7.35–7.45), pO2 = 100 mmHg (95–105), pCO2 = 22 mmHg (35–45), HCO3 = 10 mEq/L (22–30), Na+ = 127 mEq/L (135–145), K+ = 6.0 mEq/L (3.5–5.2), total protein = 7.8 g/dL (6.0–8.0).
Is this pseudohyponatremia? “I would say no,” Dr. Winter said, because there’s no analytical reason not to believe the sodium is low. “This is a hypertonic hyponatremia, with the plasma osmolality elevated because of the hypoglycemia.” (Hypertonic hyponatremia can also be related to administration of mannitol, glycine, or hyperosmolar radiocontrast media, he noted [Filippatos TD, et al. Eur J Intern Med. 2016;29:22–25].)
The case is one of diabetic ketoacidosis, Dr. Winter said, noting the direct and indirect sodium measurements were nearly equivalent. The hyponatremia seen in patients with DKA is real, not an analytical hyponatremia, and it’s one that develops in part because “the sodium loss is great,” he said. “In fact, loss of all electrolytes” is seen “from the polyuria that occurs.”
In addition, “the hyperglycemia will cause a shift in water from the intracellular space to the extracellular space”—perhaps beneficial because of the need to maintain extracellular volume to maintain circulating blood volume, he said. When circulating blood volume decreases, “tissues are not perfused, you have hypoxia, and the patient is going to be in metabolic acidosis from lactic acidosis.”
When the DKA is treated with saline and insulin, he said, the blood volume will reexpand, and the insulin will activate the sodium-potassium pump and insulin-sensitive tissues: liver, muscle, and adipose. As the cells become activated, and as the extracellular osmolality falls, the water moves from the extracellular to the intracellular space. “So the sodium will rise as part of the treatment.”
“We’re very careful when we treat patients with DKA,” Dr. Winter said, because treating too aggressively can lower extracellular osmolality too quickly and cause the brain to swell.
Important to remember, he said, is that sodium is a concentration and does not reflect total body sodium. “It may help you assess the patient, but you need to know the patient’s volume status. And in the laboratory, we may be able to assist by measuring osmolality.”
In the case of DKA, he advises laboratories not to report the corrected sodium. “I don’t have a problem with the clinician doing that, but I don’t think we have the data to support correcting sodiums.”
If a patient with congestive heart failure has a sodium of 127 mEq/L measured by a direct ISE and true hyponatremia, Dr. Winter asked, should the physician administer normal saline (155 mEq/L NaCl)? No, he said, “because if the patient has congestive heart failure and you further overload the patient, you can push the patient into pulmonary edema and kill the patient.” About 30 percent of patients with CHF will have hyponatremia, he said, and giving fluid in the “belief that the patient is not volume sufficient could have a bad outcome.”
In a patient with congestive heart failure, two major systems—the renin-angiotensin-aldosterone system and antidiuretic hormone—attempt to compensate for decreased cardiac output. In essence, a syndrome of inappropriate antidiuretic hormone secretion in patients with CHF causes the hyponatremia, he explained.
Joe El-Khoury, PhD, D(ABCC), director of the clinical chemistry laboratory, Yale New Haven Health, and associate professor of laboratory medicine, Yale University School of Medicine, says pseudohyponatremia has gone unchecked for too long because most laboratories and instrument vendors have used in their sodium interference studies an allowable error that is too lenient—10 percent (14 mmol/L), which he says is inappropriate for plasma sodium. Rather, it should be ± 4 mmol/L, he says.
In the lipemia-related study Dr. Winter cited (Koch CD, et al. Clin Chim Acta. 2021;520:63–66), Dr. El-Khoury and coauthors concluded, for users of the Roche Cobas 8000, that labs should lower their tolerance for lipemia from the currently recommended L-index cutoff of 2,000 and reflex to direct ISE when the L-index exceeds 700. They advise manufacturers and laboratories with other indirect ISE methods to evaluate the effect of lipid interference on their method using hyperlipidemic human samples, not Intralipid.
In a video presentation accessible on his YouTube channel (“Ending Pseudohyponatremia”: https://bit.ly/3B0gHDb), and supported by a grant from the Yale Department of Laboratory Medicine, Dr. El-Khoury offers five additional recommendations for clinical laboratories:
- Lower the threshold for total protein interference for sodium to at most 9 g/dL (90 g/L). In addition, set a low threshold of 3 g/dL (30 g/L) because low proteins can cause pseudohypernatremia. “That’s killing two ‘pseudos’ with one stone,” he says.
- Lower the threshold for hemolysis to about 200 on Roche. If using another vendor, reevaluate the effect of hemolysis using a total allowable error of ± 4 mmol/L, not 10 percent.
- If resources are available, validate a direct ISE method for measuring sodium in plasma and place it near the automation line. “That way, you can grab that sample when it flags, based on the new thresholds mentioned, then run it directly.” If resources are not available for this, add a comment to the report that the result is affected by proteins or lipids present in the sample and recommend that a point-of-care test by a direct method be performed.
One option, he says, is to use the “traffic light” approach to manage the workload: green—no interference present, release result; yellow—low level of interference present (total protein between 9 and 11 g/dL), release result with comment that the result is affected to alert the providers to be cautious about their interpretation; red—significant interference present, cancel and redirect, if possible, to a direct method or just state that interference is present. “This approach isn’t for everyone. It depends on what the laboratory director and clinicians are comfortable with,” Dr. El-Khoury says.
He notes that others have used calculations to correct for plasma water concentration and says he has not evaluated that and thus can’t make a recommendation.
- Do not run hemolyzed samples on a direct ISE method. Direct methods cannot account for in vitro sample dilution as occurs with hemolysis. “Just cancel and comment on those.”
- For the plasma sodium orders that do not have a total protein ordered with them, he doesn’t yet advocate running proteins on all of them, “because we don’t know the cost-to-benefit ratio,” he says. “It’s likely to be high cost and little benefit. But at the very least add a comment that sodium results are affected when total protein is greater than 9 or less than 3 g/dL.”
Amy Carpenter Aquino is CAP TODAY senior editor.