The presence of interferences such as hemolysis, icterus, and lipemia were originally detected by laboratory personnel via subjective visual evaluation of plasma after centrifugation. In contrast, most labs today employ automated tools to detect the presence of these potential interferences using calculations from absorbance measurements, collectively called serum indices. Newer instruments also incorporate cameras to detect these interferences prior to sample analysis, saving time and reagent.
In our core laboratory, all samples processed by our automation system have serum indices measured spectrophotometrically, which include an H-index (hemolysis), I-index (icterus), and L-index (lipemia/turbidity). These serum indices measurements are compared to cutoffs established by the manufacturer and/or validated in our laboratory before deciding if the result is valid. Here, we discuss special considerations for detecting and reporting results on icteric samples that exceed the acceptable I-index cutoff.
Mechanism of Interference: Conjugated vs. Unconjugated Bilirubin
Icterus, also known as jaundice, is used to describe the yellowish-greenish color observed in the sclera of the eyes or in plasma/serum samples of patients with very high concentrations of bilirubin. Such extreme elevations in bilirubin are most commonly seen in acute and chronic liver disease, biliary cirrhosis, or alcoholism. The form of bilirubin found in patient samples also depends on the patient’s disease. For example, conditions affecting the liver’s ability to conjugate bilirubin can lead to an increase in unconjugated bilirubin, while conditions such as cholestasis, which reduces bile flow, can lead to an accumulation of conjugated bilirubin.
Both forms of bilirubin interfere with analytical assays via spectral interference (absorbing strongly in the region of interest) or chemical interference (binding to dyes or acting as a reducing substance with the reagents or products formed in the test system). However, the effect and magnitude of the interference can differ between conjugated and unconjugated bilirubin, which is why it is important that in vitro diagnostic (IVD) manufacturers evaluate the effects of both separately.
Of course, this means that the manufacturers may list two tolerance cutoffs for I-index—one based on conjugated bilirubin and the other on unconjugated bilirubin spiking experiments. In this case, laboratories are advised to implement the lower I-index as their cutoff for interference because unless total and direct bilirubin are measured, it is impossible to know which form is predominantly found in the sample by I-index alone.
Automating the Solution
When our laboratory receives a sample with an I-index above the acceptable threshold, we consider three courses of action:
1. Dilute: For analytes with a validated or approved dilution matrix, we dilute the sample to lower the I-index within an acceptable range, up to the maximum dilution specified for each assay. Because the lower limit of quantitation (LLOQ) of the assay is affected by dilution, this approach is not recommended for analytes where the LLOQ itself is clinically important, such as troponin.
For example, our enzymatic creatinine assay has a manufacturer-specified I-index of 15*, LLOQ of 0.06 mg/dL, and a maximum dilution of 4-fold. For creatinine, performing a 4-fold dilution means we can resolve interferences in samples with an initial I-index up to 60, but our LLOQ is also raised 4-fold to 0.24 mg/dL on a 4-fold dilution. This approach may be applied to resolve other suspected interferences as well, but dilution should not be used to resolve interference from hemolysis. Even though dilution lowers the H-index, the mechanism of interference is typically intracellular release (not spectral or chemical), which is not resolved by dilution.
2. Result with comment: For tests where the IVD manufacturer finds no interference up to the maximum bilirubin concentrations tested (for our IVD manufacturer: I-index = 60), we release the result with the comment: “Sample is highly icteric which may affect result. Recommend interpretation within clinical context.”
In my decade of working in a clinical lab, I have never seen a sample exceed an I-index of 65. So, the likelihood that a result is affected at an I-index of 65 but not 60—and that it is clinically significant—is extremely low. Therefore, the comment advises caution during interpretation because we believe the results are more likely to be accurate than not.
3. Cancel: For analytes that have an I-index below the maximum tested (for our IVD manufacturer: I-index <60) and have no acceptable/validated dilution matrix (such as free T4 and total T3), or where dilution does not bring the I-index within the acceptable range (creatinine if I-index > 60), we automatically cancel the result because we know the interference will likely be significant. As mentioned earlier, results are also cancelled when raising the LLOQ due to dilution does not make sense clinically.
Thanks to a few of our savvy middleware superusers, our lab was able to create an automated process by developing rules to trigger the appropriate technique. So, how do we handle icteric samples? Today, we no longer do. After careful consideration of each scenario during initial validation, we now let the instrument do the work. That’s the beauty of automation.
*I-indices vary by manufacturer, with some using a 1–4 scale while others correlate the numbers to concentrations of bilirubin in mg/dL. The numbers provided here are based on Roche cobas 8000 limits.
Acknowledgments
The author would like to thank Jasmine Messina and Erick Klepacki for their mean rule-building skills and critical review of this manuscript. The author would also like to acknowledge Drs. Allison Chambliss and Adam McShane for their critical review of the manuscript.
Joe M. El-Khoury, PhD, DABCC, FADLM, is associate professor of laboratory medicine at Yale School of Medicine and co-director of the Clinical Chemistry Laboratory and Clinical Chemistry Fellowship Program at Yale-New Haven Health. +Email: [email protected]