Reference interval transference of common clinical biomarkers

Abstract

Clinical examination has become an important method of disease diagnosis, curative effect evaluation, prognosis judgment and health monitoring, and the biological reference interval is the reference standard to interpret test results and analyses of test information. In clinical tests, the reference interval is often affected by race, sex, age, geographical location and growth and development, so it is very important to establish a suitable reference interval for each laboratory. It is a huge and arduous task for each laboratory to establish its own reference interval. It is unrealistic for different measurement systems to establish reference intervals. According to the C28-A3c guideline from the Clinical and Laboratory Standards Institute (CLSI), clinical laboratories can appropriately transfer the reference intervals provided by other laboratories. This paper reviews whether the biomarkers in multiregional laboratories can transfer reference intervals between different measurement systems to expand the application of reference interval databases and ensure the accuracy and consistency of the test results.

Keywords:

• Reference interval
• transference
• biomarker
• bias
• measurement system

Introduction

The reference interval is known under many terms, including ‘reference range’, ‘expected value’, ‘normal value’ and similar expressions. The C28-A3c guideline from the Clinical and Laboratory Standards Institute (CLSI) [1] indicates the reference interval as the value between the lower and the upper reference limits. An accurate reference interval can help clinicians distinguish between healthy people and sick people. Usually, the measured value from the patient or the observed laboratory test results are compared with the reference interval, and the detected value above or below the reference interval is usually considered to be abnormal. Therefore, it is particularly important for clinical laboratories and manufacturers to provide accurate and reliable reference intervals.

The concentration of a given biomarker may be affected by factors such as race, sex, age, geographical location, living environment and eating habits. Establishing a reliable reference interval requires large quantities of samples and multicenter studies, and many studies have established common reference intervals [2,3]; however, most studies are based on reference intervals established by a single measurement system. Although the metrological traceability of current analyses has improved, the reference interval established by one measurement system cannot be directly used for other systems, which is a common problem faced by many laboratories. A solution is to establish separate reference intervals using multiple measurement systems, but this option is expensive and time-consuming and requires large samples of reference individuals. In addition, with the increasing number and usage of new biomarkers and measurement systems, it is unrealistic that each clinical laboratory establishes its own reference interval for all measurands.

Given the complexity, challenges and high cost of establishing reference intervals, the C28-A3c guideline proposes that it is feasible to transfer the reference interval from one laboratory to another through a lower cost and more convenient process and proposes a transfer method from a laboratory (donor) to other laboratories (receivers) based on CLSI document EP9-A3 [1,4]. The laboratory transference of reference intervals should be based on the two most important variables: the comparability of the measurement system and the comparability of the test subject population. If the measurement system is comparable and the test subject population is consistent, the reference interval can be transferred directly between the two systems. When the test subject population is consistent and a new measurement system is introduced, fresh patient samples should be used for method comparison and bias estimation according to the EP9 guideline. If the bias is within the specified range, it can be transferred. After transference, the reference interval should be verified according to the method recommended in the C28-A3c guideline. Once the comparability is found to be clinically unacceptable, the reference interval cannot be transferred, and the laboratory must establish a new reference interval.

At present, many laboratories in many countries and regions, such as Canada, Northern Europe and China, have successfully transferred the reference intervals of some biomarkers. This paper reviews the comparability of determination results of many common biomarkers under different conditions, aiming to explore the feasibility of reference interval transference, expand the applicability of reference interval databases, and improve the accuracy and practicability of established reference intervals.

The method of reference interval transference

In Canada, the Canadian Society of Clinical Chemists Taskforce assessed the feasibility of establishing common reference intervals using the CALIPER (Canadian Laboratory Initiative on Pediatric Reference Intervals) databases as the basis [5]. Most of the CALIPER reference intervals initially established on the Abbott Architect measurement system were later transferred to other measurement systems, including Beckman Coulter, Ortho, Roche and Siemens [6]. In China, many laboratories have carried out method comparison and bias estimation on the determination results of routine biomarkers through a variety of measurement systems in the same laboratory. According to the transference standard in the C28-A3c guideline, if the measurement system and reference population are comparable with others respectively, the reference interval of the corresponding project can be transferred [7].

Conditions of transference

The method recommended by the C28-A3c guideline from CLSI is often used for reference interval transference, which has been applied in many studies, such as Higgins et al. [8]. The biomarker concentrations were measured in the comparative system (X) and the test system (Y), and then, the linear regression model, scatter plot and bias plot were drawn by statistical methods. Concentrations/activities obtained with the comparative system (X) were plotted against the corresponding concentrations/activities obtained with the test system (Y). In most instances, visual examination of the data did not reveal any obvious outliers. In rare cases, gross outliers (identified by visual inspection) were removed. The results below the lower end of the reportable range were excluded. The sample selection for each biomarker should cover a wide range of age, sex, and other conditions, excluding specimens with interference factors for the test items. The correlation coefficient r² was calculated, and the best fitting line was determined. If r² ≥ 0.95, the least square method was used for simple linear regression; if 0.70 < r² < 0.95, Deming regression was used as the statistical method; lastly, if r² < 0.70, the correlation between measurement systems was inadequate, and the corresponding reference interval could not be transferred. The transfer requirements of the CALIPER project are more stringent than those of CLSI. To assess the appropriateness of a linear model with normally distributed data points, quantile–quantile (Q–Q), studentized residual, and Bland–Altman plots were generated for each assay [8]. The Q-Q plot was used to evaluate the normal distribution of residuals. The standardized residual plot evaluated the random distribution of the residuals. The Bland-Altman plot was used for consistency analysis. If the analysis of the three plots meets the statistical criteria, the best fit line equation is used to transfer the reference interval of the biomarker. Otherwise, the reference interval cannot be transferred and a new reference interval needs to be established. It should be noted that even if there is a strong correlation between the two measurement systems, the reference interval of the biomarker cannot be transferred if the residual is biased or non-normal in one of the plots.

If the laboratory wants to use two different methods to determine a biomarker and transfer the reference interval in the same measurement system, the methodological comparison and bias evaluation between the two methods should be carried out according to the standards proposed by CLSI EP9-A3 to judge whether the test results are comparable or not [4]. Similarly, if r² ≥ 0.95, it shows that the slope and intercept calculated by the regression equation are more reliable, the range of sample data is reasonable, and the correlation between the two methods is good. The prediction deviation and its 95% confidence interval are then calculated according to the medical decision level concentration Xc. The acceptable bias range is calculated according to the half allowable total error (TEa) by the clinical laboratory improvement amendment (CLIA’88) [9]. If the acceptable bias is greater than the higher limit of the confidence interval of the predicted bias, the results measured by the two methods are considered to be comparable, and the reference interval measured by the evaluated method is equivalent to the reference interval measured by the reference method.

Conditions of verification

The C28-A3c guideline recommends that clinical laboratories can quote and transfer the reference intervals established by other laboratories and verify the applicability to determine whether they are suitable for this laboratory [1]. Twenty qualified reference individuals should be selected for verification, the sample selection requirements are consistent with the above, and outliers are detected. If the data outside the reference interval do not exceed 2 cases, then the verification will be passed; if more than 3 cases exceed the limit, another 20 reference individuals can be selected for verification; if the data falling outside the reference interval do not exceed 2 cases, then verification will be passed; if more than 3 cases exceed the reference interval, a new reference interval should be established. For some important biomarkers, the laboratory can increase the sample size (n = 60) to verify the reference interval. It is worth noting that whether large or small samples are used for verification, the factors applied by the user laboratories (receivers) must be consistent with those of the reference laboratory (donor).

While according to the C28-A3c usually only 20 reference samples are enough to verify the transferred reference interval, the CALIPER project needs more verification reference samples. A CALIPER white paper on pediatric reference intervals drafted by a number of Canadian experts pointed out [6] that approximately 100 healthy CALIPER reference samples should be used to verify the transferred reference interval, and the selected samples should cover the age range and sex as much as possible. The total percentage of samples falling within the 95% confidence interval of the upper and lower limits of the reference interval is then calculated. If ≥ 90% of the reference samples fall within the 95% confidence interval of the transferred reference interval, the reference interval is considered to have passed the verification.

The result of reference interval transference

Most of the reference intervals in the CALIPER database have been successfully transferred to multiple clinical measurement systems. Taking Abbott Architect as the comparative system and Beckman Coulter, Ortho, Roche and Siemens as the test systems, the transference of the biomarker reference interval was shown in Table 1. When the test system was Roche Cobas 6000, 13/16 (81%) of the biomarkers met the transfer conditions, five of which were not verified: Cr, GGT, ASO, hsCRP, and Apo-B, respectively. In the Roche Modular-P system, 31/36 (86%) of biomarkers met the transference conditions, and 12 of them failed to pass the validation, namely, Cr, phosphate, UA, AMY, ALP, ASO, CRP, IgA, IgG, IgM, TP, and HDL-C [8]. Among the 34 biomarkers transferred from Abbott Architect to Ortho VITROS 5600, 32 were successfully transferred, 29 were verified, and TB, Mg and LDH failed to be verified [10]. In the process of transference to Siemens Vista, 26 biomarkers were transferred successfully, of which CRP was not clearly verified [11]. In the two measurement systems of Beckman Coulter, Beckman Coulter AU accepted most of the CALIPER reference intervals established by Abbott Architect, and 30 of the 32 biomarkers detected were successfully transferred [13]; 31 of 36 biomarkers detected by Beckman Coulter DxC met the statistical conditions of transfer, of which TB, Mg and ASO did not meet the verification criteria [12].

Table 1. Biomarkers that can be transferred between measurement systems using Abbot Architect as comparative system.

	Biomarkers	Roche Cobas 6000 [8]	Roche Modular-P [8]	Ortho VITROS 5600 [10]	Beckman Coulter [11]	Beckman Coulter DxC [12]	Beckman Coulter AU [13]	Siemens Vista [11]
Chemistry	Bilirubin, direct		Y	Y	Y		Y	Y
	Bilirubin, total	Y	Y		Y	Y	Y	Y
	Calcium		N	N	Y	N	N	Y
	CO2			N	N	N	N	N
	Creatinine (enzymatic)	Y	Y	Y		Y	Y	Y
	Creatinine (Jaffe)				Y
	Magnesium	N	N	Y	N	Y	Y	N
	Iron		Y	Y		Y	Y	Y
	Phosphate		Y	Y	N	Y	Y	Y
	Phosphorus					N
	Bicabonate (HCO3^–)	N	N
	Urea		Y	Y		Y	Y
	Uric acid		Y	Y	Y	Y		Y
Enzymes	Alkaline phosphatase (ALP)		Y	Y	Y	Y	Y	Y
	ALT (with pyridoxal phosphate)		Y	Y	Y	Y	Y	Y
	ALT (without pyridoxal phosphate)		Y
	Amylase（AMY）	Y	Y	Y	Y	Y	Y	Y
	AST AY (with pyridoxal phosphate)		Y	Y	Y	Y	Y	Y
	AST (without pyridoxal phosphate)		Y
	Gamma-glutamyltransferase (GGT)	Y	Y	Y	N	Y	Y	N
	Lactate dehydrogenase (LDH)		Y	Y	Y	Y	Y	Y
	Lipase			Y				Y
	Cholinesterase			Y
Lipids/lipoproteins	Apolipoprotein B (Apo B)	Y	Y	Y		Y	Y	Y
	Apolipoprotein A1 (Apo A1)			Y
	Cholesterol (CHO)		Y	Y		Y	Y	Y
	HDL-cholesterol (HDL-C)	Y	Y	Y	Y	Y	Y
	Triglycerides (TG)	Y	Y	Y	Y	Y	Y	Y
Proteins	Albumin G		Y	Y
	Albumin P	Y	Y	Y	Y	Y	Y	Y
	ComplemeN C3 (C3)	Y	Y	Y		Y	Y	Y
	ComplemeN C4 (C4)	Y	Y			Y	Y	Y
	C-Reactive protein (CRP)				Y	N		Y
	High–Sensitivity C–Reactive Protein (hsCRP)	Y	Y	Y			Y
	Haptoglobin		Y			N	Y	Y
	Immunoglobulin A (IgA)		Y	Y	Y	Y	Y	Y
	Immunoglobulin G (IgG)		Y	Y	Y	Y	Y	Y
	Immunoglobulin M (IgM)		Y	Y	Y	Y	Y	Y
	Prealbumin		N	Y	Y	Y	Y
	Transferrin	Y	Y	Y		Y	Y	Y
	ANistreptolysin-O (ASO)	Y	Y	Y		Y	Y
	Total protein (TP)		Y	Y	Y	Y	Y	Y
	Alpha-1 ANitrypsin (A1AT)			Y
	Rheumatoid FaYor (RF)	N	N

Y: Can be transferred based on statistical criteria; N: Not transferable based on statistical criteria.

In addition to CALIPER, several laboratory studies [14–16] have shown that the reference intervals of biomarkers could be transferred between different measurement systems (Table 2). The Nordic Reference Interval Project (NORIP) has established the reference interval of LDH in the Roche modular system. The reference limits for the female and male groups (18 to <70 years) were 105–205 U/L. The serum LDH concentration was analyzed on the Dimension Vista 1500 system with an IFCC method with bias of +2.1% and +2.7% against NFKK Reference Serum X and ERM-AD453/IFCC, respectively, showing verification of transference of the NORIP reference interval [19]. As researchers suspected that the NORIP upper reference interval limit for LDH (205 U/L) would not fit with the many clinical observations, they retrospectively verified the reference interval for LDH. The indirect reference interval was calculated on the assumption that the reference interval limits are constant in the interval from 18 to <70 years. The indirect finding of an upper limit of 240 U/L (90% CI: 234–243 U/L), and the relatively high number of test results >205 U/L, suggest that the NORIP upper limit should be adjusted [17,18].

Table 2. Transference of biomarkers between measurement systems.

Comparative system	Roche Modular P-800				Roche Cobas-8000		Vitros-750	Vitros-250	Beckman DXC-800	Hitachi-7600-020
Test system	Olympus-AU5400 [17]	Vitros-950 [17]	Beckman DXC-800 [18]	Vista Dimension [19]	Vitros-5600 [14][7]	Hitachi-7170 [7]	Hitachi-7600 [20]	Hitachi-7080 [7]	Hitachi-7600-020 [21]	Vitros-350 [15]	Vitros-5600 [16]
Chemistry
Bilirubin, total						Y	N		Y	Y	Y
Bilirubin, direct						Y	N		Y
Urea	Y	Y			Y	Y
Uric acid	Y	Y			Y	Y	Y		Y	Y
Calcium	Y	Y			Y	N
CO2					N	N
Glu	Y	Y			Y	Y	Y		Y	Y
Creatinine	N	Y			Y	Y	Y		Y	Y	Y
Mg					N	N
Cl					N	N
Na^+					N	Y
BUN							Y		N	Y
Enzyme
ALT			Y		Y	Y	Y	Y	Y	Y
AST					Y	Y	Y	Y	Y	Y
ALP					Y	Y	Y
GGT					Y	Y	Y	Y		Y
CK			Y		Y	Y
LDH				Y	Y	Y
Proteins
TP	Y	Y	Y		Y	Y	Y		Y	Y
ALB	N	N			N	Y	Y		Y	Y
Lipids/lipoproteins
TC	Y	Y	Y							Y	Y
TG	Y	Y				Y	Y		Y	Y	Y
CHO						Y	Y		Y
HDL-C									Y	Y	Y
LDL-C									Y	Y	Y

Y: Can be transferred based on statistical criteria; N: Not transferable based on statistical criteria; ALT: alanine transaminase; BUN: blood urea nitrogen; AST: aspartate transaminase; ALP: alkaline phosphatase; GGT: gamma-glutamyl transpeptadase; CK: creatine kinase; LDH; Lactic dehydrogenase; TP: total protein; ALB: albumin; TC: total cholesterol; TG: triglyceride; CHO: cholesterol; HDL-C: high-density lipoprotein cholesterol; LDL-C: low-density lipoprotein cholesterol; Mg: magnesium; Glu: glucose.

Similarly, to determine whether the reference interval of the biomarker is universal, several laboratories in China use different measurement systems or detection methods to transfer reference intervals. With Roche Modular P-800 as the comparative system, the reference interval of most biomarkers could be transferred to the test system. However, some biomarkers did not meet the appropriate transfer conditions. For example, the reference interval of ALB could not be transferred to three measurement system including Olumpus AU5400, Vitros 950, Beckman DXC 800, and the reference interval of Cr could not be transferred to the Olympus AU 5400 [17,18]; the reference interval of CK could not be transferred to the Beckman DXC800 [18].

For Vitros 5600 and Roche Cobas 8000, except ALB, Na+, Mg and Cl, the other 13 biomarkers were closely related (R² ≥ 0.95) [7]. Vitros 750 and Vitros 250 were used as comparative systems, and the test systems were Hitachi 7600 and Hitachi 7080, respectively. The reference intervals of TB and DB could not be transferred [20].

In the comparative system Beckman DXC 800 and the test system Hitachi 7600-020, the predicted biases of 14 biomarkers at the medical decision level were ≤ 1/2 CLIA′88 TEa, except that the low concentration predicted biases of urea nitrogen (BUN) were ≥ 1/2 CLIA′88 TEa, indicating that the determination results were comparable between the two systems [21].

Discussion

Several initiatives in Canada, China, Europe, and other places have performed the transference of reference intervals. Different biomarkers were measured, compared, analyzed and evaluated by different multicenter measurement systems. The reference intervals of most biomarkers could be successfully transferred. However some biomarkers could not be successfully transferred. The transfer and validation conditions follow the C28-A3c guideline issued by CLSI. If the measurement system is comparable and the study reference population is consistent, the reference interval can be transferred. The reference intervals of some biomarkers could not be transferred, mainly because of weak correlation between the two measurement systems, the predicted bias between the systems exceeding the clinically acceptable range, or an inconsistent study reference population.

The correlation between the two measurement systems is poor

Most biomarkers have a strong correlation between the Abbott Architect and Roche measurement systems. The reference interval of the Roche system according to a specific age and sex is similar to the reference interval of the CALIPER project originally established using Abbott Architect [22]. The main reason that the reference intervals of biomarkers cannot be transferred is that the correlation between the two measurement systems is poor: that is, r² < 0.7. Both bicarbonate and magnesium did not transfer from Abbott to either Roche assay due to poor correlation (r² ≤ 0.70) [8,11]. The correlation values of Mg between Abbott Architect and Beckman Coulter, Siemens Vista and Ortho Vitros were less than 0.7, and therefore the reference interval of Mg could not be effectively transferred between different measurement systems [8,11]. In Abbott Architect and Roche Modular-P systems, the r² values of total calcium and prealbumin were 0.53 and 0.27, respectively. The correlation was poor and the reference interval could not be transferred [8,11]. The correlation of Ca was poor and could not be transferred among Ortho Vitros 5600, Beckman Coulter DXC and Beckman Coulter AU measurement systems [10,12,13]. The CO₂ reference interval could not be transferred because the linear correlation coefficient r² was less than 0.7 in the five test systems, including Ortho VITROS 5600, Beckman Coulter, Beckman Coulter DxC, Beckman Coulter AU and Siemens Vista [11–13]. Phosphate was measured by the Beckman Coulter measurement system with r² = 0.68, and so its reference interval could not be transferred [8,11]. There are numerous reasons for the poor correlation between the two measurement systems, including e.g.: poor stability or traceability of biomarkers; influence of pre-analysis factors; different methods; different reagent dyes or calibration used between measurement systems.

Due to the influence of pre-analytical factors, the correlation between the reference intervals of biomarkers in different analytical systems is weak. Poor stability represents pre-analytical factor. CO₂ is volatile and could easily escape from samples during storage, transportation and aerobic treatment. Because of the instability of CO₂, it is impossible to transfer the reference interval in most measurement systems. Pre-analytical factors may be partly responsible for the invalidated LDH reference upper limit measured on the Vista dimension and that measured by the NORIP project on Roche Modular [19]. The factors affecting the results of LDH are centrifugation time, temperature, type of sample, etc. [23]. There may also be rhythmic 24-h variation of LDH [24]. In brief, determination of LDH levels usually requires that time to centrifugation (transport time) must be kept at a minimum [25–27]; LDH levels in frozen samples are decreased [23,28,29], and plasma contains higher LDH values compared with serum [30–32]. When the NORIP limits were measured on the Roche Modular, the time to centrifugation was kept short under standardized conditions. The frozen serum was used in the sample [2], while the fresh serum was used in the Vista Dimension, which led to the lower reference upper limit of LDH measured by NORIP [33].

The difference in methodology may also be the reason for the difference in the final measured values, resulting in a weak correlation between measurement systems. For ASO assays, Abbott Architect uses the Rantz–Randall (hemolytic) method, a semiquantitative measurement of the inhibition of Streptolysin-O-induced hemolysis of erythrocytes by ASO [8]. In contrast, Roche platforms use an immunoturbidimetric assay . The results obtained using Roche assays were approximately 30% and 60% higher for the 1 to ＜6 and 6 to ＜19 year age partitions, respectively, compared to values obtained with the Abbott assay [8]. Another CALIPER transference study between Abbott and the Beckman Coulter DxC 800 also found a notable difference between the ASO assays [12]. The Mg assay, however, likely failed to transfer due to methodology differences. The Roche method is a dye binding assay, whereas the Abbott method is an enzymatic assay [34]. Serum albumin is usually measured with a dye‐binding assay, such as the bromocresol green (BCG) and bromocresol purple (BCP) methods. Researchers aimed to examine differences among albumin measurements quantified using the BCG method via the ADVIA 2400 instrument and albumin measurement quantified using the BCP method via the Dimension RxL instrument from 165 serum samples [35]. Final results show that albumin results from the BCP and BCG methods may result in unacceptable differences and clinical confusion, especially at lower albumin concentrations. Due to the influence of acute phase globulin, the concentration of albumin in the BCG method is increased, especially when the concentration is low. BCP is more specific than BCG. A superior method would be capillary zone electrophoresis. However, this assay is not without problems. Unlike BCG, the BCP method underestimates the albumin concentrations in the serum of patients with renal insufficiency. Similarly, it has been described that the BCP method underestimates albumin concentrations in patients undergoing hemodialysis. The interfering substance has not been clearly identified [35]. It is worth noting that different reagents/dyes used in the measurement process also indirectly cause methodological differences. Both Abbott and Roche measure the concentration of total calcium using colorimetric assays. However, the two methods use different dyes. Abbott Architect uses arsenazo III dye binding [36], while the Roche Modular-P method uses 5-nitro-5′-methyl-BAPTA [8].

The use of different calibrators in the measurement system leads to differences in traceability chain and subsequent poor correlation. For prealbumin, for example, both Abbott and Roche assays use an immunoturbidimetric assay for measurement, although the calibration reference materials are different. Abbott uses a WHO International Standard Preparation of Rheumatoid Arthritis Serum (NIBSC code 64/2) [36], while the Modular-P method is studentized against Institute for Reference Materials and Measurements (IRMM) certified reference material (ERM-DA470k/IFCC) [8].

Haptoglobin uses an immunoturbidimetric assay and may be subject to antibody lot-to-lot variability. Antibody-based assays are difficult to transfer because subtle differences in antibody binding affinities can result in different responses across biomarker concentrations [37].

The Abbott Architect measurement system utilizes an arsenazo III colorimetric method to determine Ca [36], while the Beckman Coulter DxC800 extrapolates the total calcium concentration by an ion-specific electrode (ISE) [38]. Calcium measurements via various ion-selective electrodes are susceptible to fluctuations in pH brought about by extended storage or exposure to the atmosphere. Poor traceability of biomarkers may lead to weak correlation between reference intervals of biomarkers among different measurement systems.

The reference population is not comparable

According to the C28-A3c guideline, if the clinical laboratory wants to transfer the reference interval established by another laboratory or testing manufacturer, in addition to ensuring comparability of the test results, the reference population and other pre-analysis factors in reference interval studies should also be comparable. Due to different factors such as race, sex, age, environmental conditions, and lifestyle, even if the concentration of biomarkers exhibits a good correlation between different measurement systems, the reference intervals cannot be transferred between the measurement systems. One study [39] showed that the concentrations of biomarkers may vary by ethnicity. Researchers discussed significant differences in serum prostate-specific antigen (PSA) concentration between various ethnicities, with relatively higher concentrations in African American subjects and lower concentrations in Asian subjects. In a reference interval study in Malaysia [40], it was found that the reference intervals of some biomarkers in healthy people also varied with sex and country. For example, the serum ferritin values of healthy women in Malaysia are lower than those of the normal asymptomatic female British population.

The bias between measurement systems is not clinically acceptable

Except for cases with no significant difference between the results of the two measurement systems and when there is good correlation (r² ≥ 0.95), the relative bias (SE%) of the medical decision level of each biomarker should be less than 1/2 CLIA’88 or other country 's TEa, which was considered as a clinically acceptable error. In the evaluation of the bias between the comparative system and test system, the SE% values of TB, DB, Cr, BUN, ALT, GGT, and ALB at the medical decision point are greater than 1/2 CLIA'88 TEa (see Table 2). The difference in detection methodology between the two measurement systems is one of the reasons for the significant difference between the two measurement systems. In the comparative system, Roche Modular P-800, the Jaffe rate method was used, while the enzyme method was used to detect Cr in Olympus AU5400. Its SE% = 32.46% at the medical decision point is greater than 1/2 CLIA’88 TEa, which is 7.5%. The test result exhibits an obvious positive bias [17]. Although both Roche Modular P-800 and Vitros 950 systems used the bromocresol green (BCG) method to detect albumin, wet chemical analysis was used in the former, while dry chemical analysis was used in the latter, and the results were still inconsistent. Finally, the SE% of the low and middle values of albumin at the medical decision point was greater than 1/2 CLIA’88 TEa, indicating that the biomarker was quite different between the two measurement systems [17]. The routine field methods are non-specific and that endogenous substances affect the accuracy of these tests, such as Cr.

Similarly, because of the different methods, the results of low concentration values of BUN in the two measurement systems are different. The SE% relative bias at the medical decision point is greater than 1/2 CLIA’88 TEa. The electrode method is used in the comparative system, Beckman DXC 800, while the method used by Hitachi 7600-020 as the test system is the enzyme coupling rate method [21].

Conclusion

In summary, the consistency of determination results of different measurement systems is the key to realizing the standardization of laboratories. Although the degree of standardization of the laboratory is constantly improving, the reference intervals of different measurement systems vary greatly. Cross-system detection readily causes inconsistent reference intervals of biomarkers, so it is necessary to find a simple and effective method to make the different measurement systems use common reference intervals. At present, China, Canada, Nordic nations and other countries have made progress in transferring the reference intervals, but some reference intervals have not been transferred successfully. Only when the reference population and the measurement system are comparable and also work is performed to reduce the variability of pre-analyical factors, may the corresponding reference interval be transferred. The comparability of the measurement system is the basis of the transference of the reference interval. When r² ≥ 0.95 and the number of tests is sufficient, the transferred reference interval of the linear regression equation can be established. Considering the applicability of the transference reference intervals, a verification method should be used to evaluate the transfer result. Although there are many influencing factors, transference is an important way to expand the reference interval databases, and the method is simple and low cost: thus, it is necessary to choose appropriate methods and reasonable methods for further research.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

The present research was financially supported by First Hospital Translational Funding for Scientific & Technological Achievements, Jilin Science and Technology Development Program, Jilin Science and Technology Development Program.