Analytical validation of hemolysis detection on GEM Premier 7000

To the Editor,

Hemolysis, traditionally defined as complete breakdown of red blood cells (RBCs) resulting in release of intracellular components into the surrounding fluid, is generally classified as either “in vivo” or “in vitro” [1]. In vivo hemolysis, referred also to as hemolytic anemia, occurs when erythrocytes are injured within the body vessels due to a variety of pathological processes including autoimmune disorders (e.g., autoimmune hemolytic anemia), hemoglobinopathies (e.g., thalassemia, sickle cell disease), mechanical injury (e.g., artificial heart valves, microangiopathic hemolytic anemias), severe systemic infections (e.g., malaria, sepsis), toxins exposure (e.g., snake venoms, certain chemicals), inherited enzyme deficiencies (e.g., glucose-6-phosphate dehydrogenase and pyruvate kinase deficiencies), and rarer conditions such as paroxysmal nocturnal hemoglobinuria [1]. In vitro hemolysis, also known as spurious hemolysis, occurs instead outside the body, and is typically caused by errors during specimen collection, handling or processing. This artifact can significantly compromise the accuracy of some laboratory test results, especially potassium and other intracellular molecules (e.g., hemoglobin, lactate dehydrogenase) [1].

Prompt detection of hemolyzed samples has become a critical component of laboratory quality assurance, leading to development and widespread implementation of the so-called hemolysis index (HI) across most clinical chemistry, immunochemistry and, more recently, coagulation analyzers [2]. Hemolysis is typically quantified by these analyzers in serum or plasma samples following centrifugation. Nevertheless, detecting hemolysis in whole blood samples, which are typically used for hematological testing and blood gas analysis, has remained a longstanding challenge [3]. Regarding blood gas analysis, epidemiological evidence indicates that hemolysis may occur in approximately 1–4 % of samples (i.e., syringes) [4], 5]. The introduction of a new generation of blood gas analyzers, specifically featuring dedicated modules for assessing the HI in whole blood, has represented a significant advancement in this area [6]. Accordingly, this study aims to validate the measurement of HI on the novel GEM Premier 7000 blood gas analyzer (Werfen, Bedford, USA).

The GEM Premier 7000 is a next-generation blood gas analyzer capable of measuring a vast array of blood gases and metabolic parameters, using sample volumes between 65 and 150 µL. In addition to standard measurements, analyzers equipped with iQM3 system and the hemolysis module detect sample hemolysis within approximately 45 s. Although a detailed description of this function has been reported elsewhere [6], it may be helpful to briefly summarize the hemolysis detection methodology. The hemolysis module incorporates an acoustofluidic flow cell designed to separate plasma from blood cells, coupled with an optical detector and LED light source. The separated plasma is illuminated, and absorbance is measured at two wavelengths (570 and 610 nm). Based on the Beer–Lambert law, HI is quantified in arbitrary units (AU) that the analyzers automatically converts into six ranges of mg/dL of cell-free hemoglobin (i.e., 0–50, 51–115, 116–200, 201–300, 301–400, and ≥401 mg/dL). When the HI exceeds the predefined interference threshold of ≥116 mg/dL, potassium results are flagged, alerting the user to potential unreliability and prompting further decision-making. Other than the default qualitative categorization in mg/dL, Werfen supported the extraction of quantitative HI values for this evaluation, which were reported in grams per liter (g/L) in accordance with SI (Système International) unit standards. In our validation of HI on GEM Premier 7000, we assessed intra-assay imprecision, linearity, and we also compared HI values in heparinized whole blood against those obtained in paired heparinized plasma samples analyzed on a clinical chemistry platform (Roche Cobas c702; Roche Diagnostics, Basel, Switzerland), which uses a hemolysis detection assay highly correlated with the reference cyanmethemoglobin technique [7].

Intra-assay imprecision was evaluated by measuring 10 consecutive replicates of three centrifuged plasma samples collected into 6.0-mL lithium-heparin blood tubes (Vacutest Kima, Padova, Italy) and containing increasing concentrations of cell-free hemoglobin (low, medium, and high). Heparinized whole blood was found unsuitable for this study due to time-dependent hemolysis, which began as early as 15 min after the first measurement on GEM Premier 7000 and increased progressively up to 35–40 min when the last replicate was assayed, as mirrored by the gradual increase in both HI and potassium (Supplementary Figure 1). Linearity was assessed by preparing serial dilutions (from 1:10 to 10:1) of a high-HI whole blood sample with a low-HI whole blood sample, both collected in 6.0-mL lithium-heparin blood tubes. Sample comparison was performed using specimens collected from 26 laboratory staff members into 6.0-mL lithium-heparin tubes. Individual samples were divided into six identical aliquots, and mechanical hemolysis was induced from the second aliquot onward through sequential passages (one to five times) through a fine needle (i.e., 25 gauge) attached to an insulin syringe, according to a standard protocol validated in previous studies [8]. The HI was first measured on GEM Premier 7000, the samples were immediately centrifuged at 1,500×g for 10 min at room temperature, and lithium-heparin plasma was then transferred into plastic cups for HI measurement on a Cobas c702 analyzer, which uses bichromatic absorbance readings at 570/600 nm and reports HI in AU, where 1 AU corresponds to 0.01 g/L.

Statistical analysis was conducted with Analyse-it (Analyse-it Software Ltd, Leeds, UK), and included the calculation of intra-assay imprecision (expressed as coefficient of variation, CV%), assessment of linearity by Pearson’s correlation, and evaluation of sample comparability by Spearman’s correlation, Deming fit, Kappa statistic, Bland–Altman plots and receiver operating characteristic (ROC) curve analysis. The study was conducted under the ISO 15189:2022 accreditation in agreement with the Declaration of Helsinki and under the conditions of relevant local legislation. The study was cleared by the local Ethical Committee (Verona and Rovigo provinces; protocol number: 971CESC, date of approval: 25 July, 2016).

Table 1 shows the results of the imprecision studies, with values ranging from 4.1 % in heparinized plasma samples with low HI values to 0.5 and 0.2 % in samples with medium and high HI values, respectively. The assay demonstrated excellent linearity across the HI range between 0.06 and 7.27 g/L, with Spearman’s correlation coefficient of 0.999 (95 % confidence interval [CI]: 0.998–1.000; p<0.001) (Supplementary Figure 2). The comparison of HI values measured in heparinized whole blood on GEM Premier 7000 and heparinized plasma on Cobas c702 is shown in Figure 1 (n=143; 13 samples were excluded as values exceeded the measuring range of GEM Premier 7000). The Spearman’s correlation coefficient between continuous HI values on both analyzers was 0.990 (95 % CI: 0.986–0.993; p<0.001), slightly decreasing to 0.954 (95 % CI: 0.936–0.966; p<0.001) when comparing continuous values on Cobas c702 with categorized (six classes) HI values on GEM Premier 7000 (Supplementary Figure 3). Deming regression analysis for continuous HI values yielded the following equation: [GEM Premier 7000] = 1.12 × [Cobas c702] – 0.04. The Bland-Altman plot analysis showed no significant systematic bias between the two assays (mean difference: −3.2 %; 95 % CI: −10.2 to 3.8 %; p=0.364). A high level of concordance was observed between the two analyzers when HI values were classified into six predefined ranges of cell-free hemoglobin, yielding Cohen’s kappa coefficient of 0.87 (95 % CI: 0.80–0.93; p<0.001).

Figure 1:

Comparison of hemolysis index (HI) values measured on paired samples using heparinized whole blood on GEM Premier 7000 (continuous values) and heparinized plasma on Cobas c702 (Deming fit). HI, hemolysis index.

ROC curve analyses at 0.50 g/L and 1.16 g/L HI thresholds in heparinized plasma on Cobas c702 are shown in Supplementary Figure 4. The area under the curve (AUC) was 1.00 (95 % CI: 1.00–1.00; p<0.001) at both thresholds. At the 0.50 g/L cutoff for both analyzers, GEM Premier 7000 demonstrated a sensitivity of 0.99 (95 % CI: 0.94–1.00) and a specificity of 0.96 (95 % CI: 0.86–1.00), while both sensitivity and specificity were 1.00 (95 % CI: 0.96–1.00 and 0.94–1.00, respectively) at the 1.16 g/L threshold.

The results of our study, encompassing a comprehensive analytical validation of HI measurement on the novel GEM Premier 7000 blood gas analyzer, demonstrated very low intra-assay imprecision and a wide, linear measurement range, extending up to 7.27 g/L. We observed excellent comparability with a reference clinical chemistry platform (Cobas c702), with Spearman’s correlation of 0.990 for continuous HI values, statistically negligible bias, and almost optimal accuracy at the clinically relevant HI thresholds of 0.50 and 1.16 g/L.

Our findings confirm and extend earlier results reported by Balasubramanian et al. [6], as both studies consistently demonstrated that GEM Premier 7000 effectively detects hemolysis in whole blood samples without requiring plasma separation – a significant innovation for point-of-care (POC) testing [3]. Notably, the threshold of cell-free hemoglobin reported to influence potassium measurements with GEM Premier 7000 (≥1.16 g/L) was validated in both investigations. Excellent agreement rates were also observed between GEM Premier 7000 and conventional clinical chemistry analyzers, with high concordance regardless of the technique used for generating hemolysis (i.e., hemolysis-spiked samples in the study by Balasubramanian et al., vs. mechanical hemolysis in our current investigation) [6].

In conclusion, our validation confirms that GEM Premier 7000 represents a significant technological advancement for detecting hemolysis in whole blood with blood gas analyzers [3], 9], combining analytical robustness with clinical applicability. Its implementation holds considerable potential to enhance patient safety by enabling timely identification of hemolyzed whole blood samples, either in vitro or in vivo, improving the reliability of critical laboratory results – especially potassium – in emergency and intensive care settings [10].