How to verify the analytical and clinical performance of ELISA immunoanalysis in the real laboratory practice. PCSK9 as an example.

BACKGROUND: Manufacturers and diagnostic companies often recommend on-site verification of analytical performance in the clinical laboratory. The validation process used by manufacturers is rarely described in detail, and certain information on analytical performance is missing from the product sheet, especially for immunoanalytical methods. We describe an approach to the detailed validation of an ELISA method for the measurement of proprotein convertase subtilisin/kexin type 9 (PCSK9) plasma concentrations. We compared manufacturers' claims of analytical performance with data obtained in the field laboratory using several approaches. METHODS: We used the Human Proprotein Convertase 9/PCSK9 Quantikine ELISA diagnostic kit (R&D systems, Bio-Techne Ltd., Abingdon Science Park, Abingdon, UK) and three levels of quality control solution Quantikine Immunoassay Control Group 235 (R&D systems, Bio-Techne Ltd., Abingdon Science Park, Abingdon, UK) to verify precision. We measured the concentration of PCSK9 using the DS2 ELISA Reader (Dynex Technologies GmbH, Denkendorf, Germany). We used analysis of variance (ANOVA) and the R statistical package (R core team, version 1.4.5). Statistical analysis and terminology were performed according to protocol CLSI EP15-A3, and the reference interval was checked according to CLSI/IFCC C28-A3c. RESULTS: We found a significant difference between the manufacturer's claims of analytical performance and real data measured in the routine clinical laboratory. The calculated CV (%) for repeatability (calculated by simple estimation of the mean and SD, as used by the manufacturer) was between 5.5% and 7.4%, but the manufacturer's claim was between 4.1% and 6.5%. Using ANOVA, the true repeatability was between 5.0% and 6.9%. Similarly, ANOVA revealed values of CV (%) for within-laboratory imprecision between 6.5% and 9.1%, while the manufacturer's claims were between 4.1% and 5.9%. The recovery ranged from 105.5% to 121.8%. The manufacturer's recommended reference interval was checked and we didn't find any significant difference between men and women. CONCLUSIONS: We describe a comprehensive approach to verify the analytical performance of an ELISA method using the measurement of PCSK9 plasma concentration as an example. We found differences between the results of this approach based on the CLSI EP15-A3 protocol and data provided by the manufacturer. We recommend the verification of analytical performance by more complex statistical tools in laboratory practice.

Biological variation of PIVKA-II in blood serum of healthy subjects measured by automated electrochemiluminescent assay.

Background: Prothrombin/Protein Induced by Vitamin K Absence-II (PIVKA-II) is a candidate biomarker of hepatocellular cancer, recommended both for diagnostics and monitoring. The aim was to evaluate biological variation (BV) of serum PIVKA-II. Methods: Within-subject (CVI) and between-subject (CVG) BV estimates were assessed in 14 healthy volunteers in a 6-week protocol. Serum concentrations of PIVKA-II were measured by a Roche Elecsys PIVKA-II diagnostic kit (cobas e8000). Precision (CVA) was assessed from duplicate measurements of all volunteers' samples. Two methods were used for the estimation of CVI: SD-ANOVA and CV-ANOVA method. We calculated the index of individuality (II) and reference change value. The experiment was fully compliant with EFLM database checklist. Results: The CVI of PIVKA-II in healthy persons, as calculated by two statistical methods, were 8.2% (SD-ANOVA with CVA of 3.2%) and 9.4% (CV-ANOVA) with CVA of 2.7%). The CVG was 19.5% (SD-ANOVA), and respective II and RCV were 0.42 and 24.4%. Conclusions: CVI and CVG of PIVKA-II were 8.2% and 19.5%, respectively, with CVA below 4%. The low II and RCV below 25% enable the use of this biomarker both for diagnostics and monitoring. More data are needed before the introduction of PIVKA-II into clinical practice.

Biological variation of proprotein convertase subtilisin/kexin type 9 (PCSK9) in human serum.

BACKGROUND: Proprotein convertase subtilisin/kexin type 9 (PCSK9) is involved in the regulation of LDL receptors. Inhibition of PCSK9 increase uptake of LDL-particles and pathogen-associated molecular patterns (PAMPs). The aim of our study was to evaluate biological variation of serum PCSK9. METHODS: Within-subject (CVI) and between-subject (CVG) biological variations were assessed in 14 healthy volunteers in a 6-week protocol (7 samples, equidistant time intervals). Serum concentration of PCSK9 was measured by a Quantikine ELISA assay (R&D systems, Bio-Techne Ltd., UK) on a DS2 ELISA reader (Dynex Technologies GmbH, Germany). Precision (CVA) was assessed by duplicate measurements. Two methods with different levels of robustness were used for the estimation of CVI, SD-ANOVA and CV-ANOVA method. We calculated the index of individuality and reference change values. The experiment was fully compliant with EFLM database checklist. RESULTS: The within-subject values of PCSK9 in healthy persons, as calculated by two statistical methods, were 23.2% (SD-ANOVA with CVA of 5.6%) and 26.6% (CV-ANOVA with CVA of 4.8%). The CVG was 10.9% (SD-ANOVA), index of individuality and RCV were 2.13 and 66.3%, respectively. CONCLUSIONS: The high index of individuality indicates that common reference intervals can be used to interpret serum PSCK9 values.

Presepsin: what is important for correct interpretation.

INTRODUCTION: Presepsin (soluble CD14 subtype) is a new biomarker of infection and sepsis. Correct interpretation is based on the knowledge of analytical reliability, biological variation, decision limits, and diagnostic effectivity. AIM: The aim of the study was to verify analytical precision of presepsin measurements, to assess biological variation in healthy subjects, to verify reference and decision limits, to assess diagnostic effectivity, and to compare data with commonly used septic biomarkers - procalcitonin (PCT), CRP and interleukin 6 (IL6). MATERIAL AND METHODS: Analyti-cal precision (repeatability and intermediate precision) was estimated by repeated measurements of commercial control materials. Biological variation was evaluated in a group of 20 healthy volunteers in a 7-week experi-ment. Reference ranges were extracted from the literature and compared with data from healthy subjects. RESULTS: Precisions of presepsin measurements were 2.0-4.0 % (“within-run”) and 6.1-9.5 % (“between-run”). Intraindividual biological variation of presepsin was 22.3 %, interindividual variation 20.8 %. Index of individuality was 1.07, reference change value (RCV - critical difference) was 72 %. Upper reference limit was around 180 ng/l. CONCLUSION: Ana-lytical quality of presepsin measurement is suitable for clinical use. Biological variation parameters enable the use of reference limits, upper reference limit of presepsin is around 180 ng/l. None of the tested biomarkers has universal cut-off value, multiple biomarkers are needed and repeated measurements are advisable.

Biological variation of intact fibroblast growth factor 23 measured on a fully automated chemiluminescent platform.

BACKGROUND: Fibroblast growth factor 23 (FGF23), a potent regulator of phosphate and vitamin D metabolism, is a new biomarker of kidney, bone and cardiovascular disorders. The aim of this study was to assess the biological variation of intact fibroblast growth factor 23 (iFGF23). METHODS: The within-subject (CVI) and between-subject (CVG) biological variations were assessed in 14 healthy volunteers in a six-week protocol (seven samples). Imprecision (CVA) was assessed by duplicate measurements and the EP15-A2 protocol. Intact FGF23 was measured using a fully automated chemiluminescent assay (Liaison XL, DiaSorin S.p.A., Saluggia, Italy). Two methods with different sensitivities to non-Gaussian distribution were used to estimate the CVI, SD ANOVA and CV ANOVA methods. We calculated the index of individuality (II) and reference change values. RESULTS: Depending on the statistical method used, the CVI and CVA were 14.2 and 3.7% (SD ANOVA) or 12.5 and 3.9% (CV ANOVA), respectively. The corresponding reference change values were 40.5 and 36.4%, respectively. The CVG was 13.4% (SD ANOVA was the only option), and the total imprecision (EP15-A2) was less than 7%. CONCLUSIONS: The measurement of iFGF23 demonstrated a CVA less than 4% during the experimental estimation of biological variation. The total imprecision was less than 7% in the EP15-A2 experiment. The CVI values of iFGF23 in healthy persons were 14.2 (SD ANOVA) and 12.5% (CV ANOVA), respectively. The CVG was 13.4%, and the resulting index of individuality was 1.06. The reference change value was less than 41%. The availability of this automated assay for iFGF23 with well-characterized biological variation data delivers opportunities for improved availability and application of this assay clinically.

Tacrolimus has immunosuppressive effects on heavy/light chain pairs and free light chains in patients after heart transplantation: A relationship with infection.

The aim of the study was to investigate the relationship between tacrolimus (TAC) immunosuppressive treatment and serum concentrations of immunoglobulin heavy/light chain pairs (sHLC) and free light chains (sFLC) in patients after heart transplantation (HTX) and to use these biomarkers to predict the risk of infection in these patients. A total of 88 patients with an immunosuppressive regimen involving tacrolimus who underwent HTX were analyzed over 24 months of follow-up. sFLC and sHLC levels were determined before and at three time points after HTX. TAC concentrations were determined at several time points after HTX, and mean TAC concentrations and areas under the curve (AUCs) of TAC concentration were calculated. Relevant clinical data were obtained from patients' medical records. A larger AUC of TAC was associated with decreases in the concentrations of IgG total (p < 0.05); similarly, cumulative AUC of TAC during 18 post-transplant months correlated inversely with sHLC IgG kappa (r = -0.228, p < 0.05) and IgG total (r = -0.352, p < 0.05). Concentrations of sFLC kappa, sFLC lambda, sHLC IgG kappa, and sHLC IgG total were significantly lower in infected patients (in the 9th month after HTX, all p < 0.05). Combined criteria for increased AUC (greater than the median of 12.9 mg·d/l) and decreased sFLC kappa (less than the median of 12.5 mg/l) correlated with the presence of infection (p < 0.03) in the 9th month after HTX. Ratio of concentration of TAC to sFLC kappa or lambda was significantly higher in infected patients (both p < 0.05). Intensive treatment with tacrolimus after HTX is possibly reflected by decreases in sFLC and sHLC (mainly sHLC IgG). Patients with decreased concentrations of these biomarkers are at increased risk for infection, primarily in the 9th month after HTX, when the concentrations of tacrolimus were the highest.

Enhanced liver fibrosis (ELF) score: Reference ranges, biological variation in healthy subjects, and analytical considerations.

BACKGROUND: The Enhanced Liver Fibrosis score has been recognized as a non-invasive test for liver fibrosis. However, reference intervals, biological variation and analytical performance have not been studied in detail so far. The aim was to acquire data that are essential for correct interpretation. METHODS: A total of 40 apparently healthy volunteers were evaluated for reference ranges of serum concentration of hyaluronic acid, aminoterminal propeptide of type III collagen, and tissue inhibitor of metalloproteases-1, and calculated ELF score. A subgroup of 20 subjects was evaluated repeatedly for 7 weeks. For all variables, reference intervals, within-subject and between-subject biological variations, reference change values, and the indexes of individuality were assessed. Analytical performance (intermediate precision) and interlaboratory comparison were also evaluated. RESULTS: The reference ranges were 5.1-62.7 µg/L for HA, 3.56-12.6 µg/L for PIIINP, 143.6-265.3 µg/L for TIMP-1, and 7.14-9.55 for the ELF score. The within-subject variations were 32.7, 10.6, 4.2, and 3.2% for HA, PIIINP, TIMP-1, and ELF score, respectively. Similarly, the between-subject variations were 59.0, 13.3, 12.8, and 5.2%. For the ELF score, RCV was 10.1% and II was 0.62. The intermediate precisions were <5%, <6%, and <10% for HA, PIIINP, and TIMP-1, respectively. CONCLUSION: The reference range of the ELF score overlap with the area defined as moderate fibrosis by the manufacturer. High biological variation of HA was diminished by the natural logarithm in the calculation of the ELF score. The use of the ELF score has suitable analytical and acceptable biological performance characteristics for clinical practice. However, the transfer of results evaluated in healthy persons to the populations with chronic liver diseases deserves caution.

Serial measurement of presepsin, procalcitonin, and C-reactive protein in the early postoperative period and the response to antithymocyte globulin administration after heart transplantation.

Differentiation between systemic inflammatory response syndrome and sepsis in surgical patients is of crucial significance. Procalcitonin (PCT) and C-reactive protein (CRP) are widely used biomarkers, but PCT becomes compromised after antithymocyte globulin (ATG) administration, and CRP exhibits limited specificity. Presepsin has been suggested as an alternative biomarker of sepsis. This study aimed to demonstrate the role of presepsin in patients after heart transplantation (HTx). Plasma presepsin, PCT, and CRP were measured in 107 patients serially for up to 10 days following HTx. Time responses of biomarkers were evaluated for both noninfected (n=91) and infected (n=16) patients. Areas under the concentration curve differed in the two groups of patients for presepsin (P<.001), PCT (P<.005), and CRP (P<.001). The effect of time and infection was significant for all three biomarkers (P<.05 all). In contrast to PCT, presepsin was not influenced by ATG administration. More than 25% of noninfected patients had PCT above 42 µg/L on the first day, and the peak concentration of CRP in infected patients was reached on the third post-transplant day (median 135 mg/L). Presepsin seems to be as valuable a biomarker as PCT or CRP in the evaluation of infectious complications in patients after HTx.

Analysis of Four Automated Urinalysis Systems Compared to Reference Methods.

BACKGROUND: The aim of this study was to compare four automated urinalysis systems: the Iris iQ200 Sprint (Iris Diagnostics, U.S.A.) combined with the Arkray AUTION MAX AX 4030, Iris + AUTION, Arkray AU 4050 (Arkray Global Business, Inc., Japan), Dirui FUS 2000 (Dirui Industrial Co., P.R.C.), and Menarini sediMAX (Menarini, Italy). METHODS: Urine concentrations of protein and glucose (Iris, Dirui) were compared using reference quantitative analysis on an Abbott Architect c16000. Leukocytes, erythrocytes, epithelia, and casts (Iris, Arkray, Diuri, Menarini) were compared to urine sediment under reference light microscopy, Leica DM2000 (Leica Microsystems GmbH, Germany) with calibrated FastRead plates (Biosigma S.r.l., Italy), using both native and stained preparations. RESULTS: Total protein and glucose levels were measured using the Iris + AUTION system with borderline trueness, while the Dirui analysis revealed worse performances for the protein and glucose measurements. True classifications of leukocytes and erythrocytes were above 85% and 72%, respectively. Kappa statistics revealed a nearly perfect evaluation of leukocytes for all tested systems; the erythrocyte evaluation was nearly perfect for the Iris, Dirui and Arkray analyzers and substantial for the Menarini analyzer. The epithelia identification was connected to high false negativity (above 15%) in the Iris, Arkray, and Menarini analyses. False-negative casts were above 70% for all tested systems. CONCLUSIONS: The use of automated urinalysis demonstrated some weaknesses and should be checked by experienced laboratory staff using light microscopy.

The role of timely measurement of galectin-3, NT-proBNP, cystatin C and hsTnT in predicting prognosis and heart function after heart transplantation.

BACKGROUND: Changes of biomarkers measured soon after heart transplantation (HTx) can reflect different processes: cardiomyocyte necrosis (troponins, high-sensitivity cardiac TnT and TnI), heart function (natriuretic peptides, BNP and NT-proBNP), fibrosis (galectin-3 and ST2), and global cardiorenal risk (cystatin C). We assessed the prognostic role of hsTnT, NT-proBNP, galectin-3 and cystatin C during the early post-transplant period. METHODS: A total of 121 consecutive post-HTx patients were assessed. The main outcomes were survival, left ventricular ejection fraction (LVEF) and rejection periods. Survival was assessed after intermediate (12 months) and long periods (total follow-up during study, median of survival 763 days, IR 527-1038 days). LVEF was assessed 12 months after HTx. Rejection was evaluated during follow-up. We report biomarker concentrations measured 10 days and 12 months after HTx. RESULTS: Ten days after HTx, cystatin C and hsTnT predicted death both under univariable and multivariable analysis. These two biomarkers along with galectin-3 were increased in patients with decreased LVEF measured 1 year after HTx. NT-proBNP did not show early prognostic power. None of the measured biomarkers predicted rejection, but hsTnT and NT-proBNP were increased significantly 12 months after HTx in patients with at least one rejection. CONCLUSIONS: Cystatin C and hsTnT measured 10 days after HTx can provide prognostic information on survival and galectin-3 measured at the same time may display a relationship to heart function assessed 1 year after HTx. Further study should be carried out in a large cohort of patients.

A clinical and laboratory approach used to elucidate discordant results of high-sensitivity troponin T and troponin I.

BACKGROUND: Careful interpretation of discordant results in high-sensitivity troponin measurements is necessary in cases of suspect immunoassay interferences. We describe several procedures taken in a case of a polymorbid patient with chest pain, without clear evidence of myocardial necrosis and with increased high-sensitivity cardiac troponin T (hs-cTnT). We checked the Vafaie's algorithm for the evaluation of suspect interference in troponin measurements. METHODS: We conducted a case report analysis, additional measurements, a dilution test and pretreatment of plasma with blocking agents. RESULTS: Concentration of hs-cTnT (99 th percentile of "healthy" population 14 ng/L) increased from 120.1 ng/L to 280.4 ng/L during an 8-month period and decreased to 216.3 ng/L during the following month with repeatedly negative troponin I (TnI), hs-cTnI, myoglobin and creatine kinase MB (CK-MB). Suspected false positivity of hs-cTnT was further confirmed by treatment of plasma with an antiheterophile blocking agent (hs-cTnT before treatment 280.4 ng/L, after 16.53/16.23 ng/L). This outcome was further confirmed by the manufacturer's experiments. CONCLUSIONS: The false-positive results of hs-cTnT were caused by the presence of extremely rare high molecular weight protein, presumably IgM, most likely HAMA (human anti-mouse antibody). Only the pre-treatment of plasma with a blocking agent provided a reliable indication of the interference. Cooperation among clinicians, laboratory personnel and the manufacturer is essential.

Long-term biological variability of galectin-3 after heart transplantation.

BACKGROUND: Galectin-3 is an emerging biomarker of heart failure and of myocardial fibrosis risk. Monitoring of galectin-3 is essential during treatment with galectin-3 inhibitors. The aim of our study was to assess long-term biological variability in a specific group of unhealthy subjects. METHODS: The biological variability of galectin-3 was measured in a group of 44 patients after heart transplantation (HTx). Six samples were taken from each patient during a 12-month period. Galectin-3 was measured with an Abbott Architect automated immunoassay. RESULTS: Intraindividual (CVi) and interindividual (CVg) variabilities were calculated together with the reference change value (RCV), the log-normal RCV for increase (RCV+), and the log-normal RCV for decrease (RCV-). The CVi, CVg, RCV, RCV+, and RCV- were 28.2%, 35.6%, 78.6%, 116%, and -53.7%, respectively. The index of individuality was 0.79. CONCLUSIONS: The concentrations of galectin-3 in patients followed 12 months after HTx fluctuated around the homeostatic point, with CVi of approximately 28%. RCVs of +116% (log-normal increase) and -54% (log-normal decrease) mean that the concentration of galectin-3 would need to approximately double or decrease by half to indicate a new process.