Determination of hemolysis index thresholds for biochemical tests on Siemens Advia 2400 chemistry analyzer

Background In vitro hemolysis is still the most common source of pre‐analytical nonconformities. This study aimed to investigate the hemolytic effects on commonly used biochemical tests as well as to determine the hemolysis index (HI) thresholds on Siemens Advia 2400 chemistry analyzer. Methods Peripheral blood samples were collected from forty healthy volunteers. Hemolysis was achieved using syringes. Five hemolysis levels were produced including the no hemolysis group, slight hemolysis group, mild hemolysis group, moderate hemolysis group, and heavy hemolysis group. We then used the bias from baseline (no hemolysis) and HI to construct regression functions. The HI corresponding to the bias limits was considered as HI thresholds. We chose the total allowable error (TAE) as the bias limit. Results Of the twenty‐eight analytes, ten analytes had clinical significance. Creatine kinase‐MB, creatine kinase, potassium, aspartate aminotransferase, and hydroxybutyrate dehydrogenase were all positively affected; the corresponding HI threshold was 45.2, 99.96, 4.07, 10.16, and 7.94, respectively. Lactate dehydrogenase was also positively interfered, but we failed to calculate the HI threshold. Total bile acid, uric acid, and sodium were all negatively affected, and the HI threshold was 42.23, 500 and 501.8, respectively. Glucose was also negatively interfered, but it failed to achieve the HI threshold. Conclusions When the HI value was higher than its threshold, the corresponding analyte was considered inappropriate for reporting. The implementation of the assay‐specific HI thresholds could provide an accurate method to identify analytes interfered by hemolysis, which would improve clinical interpretations and further boost laboratory quality by reducing errors associated with hemolysis.

The total allowed error for HBDH was assumed to be 30% according to experience.
a ±10% bias was set as the accepted TAE for the analytes because they are not included in the CLIA'88 regulations. b To compare with the TAE defined as ERROR in CLIA'88, the hemolysis effects on Ca, K, Na, and CK-MB are expressed as bias directly. c Significant differences exist among the five groups when compared with each other (P < 0.05).

TA B L E 1 (Continued)
tubes. 2,7 Traditionally, hemolysis is detected by visual detection, but this method is time-intensive, arbitrary, and rather subjective, which consequently impact clinical decisions. 8 Moreover, it is difficult to visually detect subtle differences in color between hemolysis and icteric samples. The continuous-flow automatic system leaves little chance for visual detection, and it has been reported that intravenous catheters and vacuum blood-drawing technology might result in a higher risk of hemolysis; 9 therefore, the increasing use of these technologies makes it more challenging to quickly identify hemolysis specimens.
Hemolysis index (HI) generated by analyzers is an effective tool to counteract the hemolysis challenge, as it can standardize the process of identifying hemolytic specimens and estimate the hemolysis interferences quantitatively. [10][11][12][13][14][15][16][17] Even though it was reported that the hemolysis index (HI) was accurate and highly reproducible among different platforms and laboratories, 16   to the corresponding reagent protocols. The assay reagents were obtained from the same vendor as the analyzer system (Siemens Healthcare Diagnostics Inc, Deerfield, IL, USA). The hemoglobin (Hb) level was measured on an XE-5000 hematology analyzer (Sysmex Corporation, Kobe, Japan).

| Statistical analysis
All statistical analyses were carried out using the Statistical The 2-tailed t test was used to compare analyte concentrations between hemolysis groups and the baseline group (no hemolysis).
These results revealed a higher bias than TAE limits, and statistical differences from baseline concentrations were considered to be clinically significantly interfered by hemolysis. In order to identify HI thresholds for interference analysts, we used the "curve estimation" in SPSS including linear, logarithmic, inverse, quadratic, and cubic models to select the model with the highest R2 and the lowest P values. We then used the GraphPad Prism 7 to produce graph and formula of regression curves chosen from the curve estimation, through which we could precisely locate the x (HI) and y (bias) coordinates on the curves. The HI corresponding to the bias limits (TAE or ±10%) was considered as the HI threshold. A P value of <0.05 was considered statistically significant.  Table 1.

| RE SULTS
At first, the Hb concentrations in the five groups were measured to evaluate the relationship between HI and Hb. The HI values among the five groups were significantly different when compared to each other (P < 0.05); there was a strong association between HI and Hb concentrations ( Table 1, r = 0.982, P < 0.05, Supplement Figure S1).
We then compared analyte concentrations between hemolysis groups and the baseline group. The concentrations of AST, TBA, ALB, LDH, CK, CKMB, HBDH, GLU, UA, K, and Na in hemolysis groups were significantly different from that of the NH (Table 1, Figure 1, TAE, total allowed error recommended by CLIA'88 regulations; NA, not available. The total allowed error for hydroxybutyrate dehydrogenase was assumed to be 30% according to experience. a ±10% bias was set as the accepted TAE for the analytes because they are not included in the CLIA'88 regulations. Hemoglobin was estimated from equation HI = 120*Hb-68, supplement Figure S1. In our study, though the hemolysis interference on biochemistry analytes was dependent on the analyzer system, the interference on CK, AST, LDH, and K was consistent with former studies using the Cobas 6000 c501 analyzer 22 or Roche analyzers. 15 This confirmed that a common mechanism underlies the observed hemolysis interference. The mechanisms behind the hemolysis interferences include the additive interferences of released intracellular substances (eg, LDH, AST, K, and HBDH) and the chemical interferences when the released substances interacted with the measured analyte (eg, CK and CKMB); 23 it was reported that intracellular adenylate kinase might interfere with the CK assay. 12 In addition, our results showed that positive hemolysis interferences on CK-MB activity started to increase at lower HI values compared with CK activity, which was in accordance with Oğuzhan Özcan's study. 24 The reason may be that the errors from the interfering agents released by hemolysis were amplified by multiplying a constant; this constant parameter is commonly used to calculate the CKMB activity in the assay. We also observed that UA, GLU, Na, and TBA decreased due to hemolysis, which may result from the dilution effects caused by the leakage of intracellular components into the surrounding fluid. However, GLU was less affected in MH than SH, which was also reported in another study. 12 This phenomenon might be due to the interaction between the spectral interference of the released hemoglobin and the dilution effects owing to the leakage of intracellular components.
We found that the HI thresholds for AST ( In conclusion, this is the first study to our knowledge that investigated HI thresholds using the Advia 2400 analyzer, which extended these HI studies. [10][11][12][13][14][15][16][17][20][21][22]24 Our results provide HI thresholds for eight analytes (CKMB, CK, K, AST, HBDH, TBA, Na, and UA). These analyses would be inappropriate for reporting when their HI values are higher than the corresponding HI thresholds. The implementation of the assay-specific HI thresholds can provide an accurate means to identify the extent to which hemolysis interferes with analytes. This would lead to better clinical interpretations and may improve the laboratory test quality by reducing errors associated with hemolysis.

ACK N OWLED G M ENTS
All the authors have accepted responsibility for the entire content of this submitted manuscript and approved the submission. This study was supported by grants from Tianjin Science and Technology

Project (15ZXLCSY0040) and the Basic Research Project of Logistics
College of Armed Police Forces (WHJ2016024).

CO N FLI C T O F I NTE R E S T
No conflict of interest exists in this manuscript, and the manuscript is approved by all authors for publication. The funding organization(s) played no role in the study design; in the sample collection, analysis, and interpretation of data; in the writing of the report; or in the decision to submit the report for publication.