The influence of analytical bias on diagnostic misclassifications

Quality specifications for analytical imprecision and bias based on the state of the art; 'biology' and 'analysis of clinical situations' have been proposed by several scientists. Most interesting is the assessment of 'diagnostic misclassifications' based on direct evaluation of the consequences of analytical bias on the percentage of false positives and false negatives from a clinical decision situation, or based on the percentage of healthy individuals outside each reference limit when common reference intervals are used. With use of graphical or computer simulations assuming increasing (positive or negative) analytical bias, the expected percentage of misclassifications can be estimated- and, for the error for which the outcome (the fraction of misclassifications) is considered unacceptable, the maximum allowable analytical bias can be defined. An overview is given of previous proposals for specification of allowable analytical bias, and new examples are presented: (i) for S-transferrin. an analytical bias of +10% will increase the percentage of healthy individuals with measured concentration values above the upper reference limit from 2.5 to 10% (ii) the percentage of healthy men with concentration values for S-cholesterol above 6.2 mmol/l (240 mg/dl) will vary between 25 and 85% for analytical bias from - 1.0 to +1.0 mmol/l (+/- 16%): (iii) for glycated haemoglobin, two examples are given which illustrate the effect of analytical bias on the risk of retinopathy and so-called 'microalbuminuria' for measured values identical to the target 7.5% and 10.1% glycated haemoglobin, respectively. It is concluded that analytical bias may have significant impact on diagnostic performance, better standardization is needed, and quality specifications for allowable analytical bias should be based on medical usefulness criteria or, if such data are not available, on biological criteria.

Consensus of a group of professional societies and diagnostic companies on guidelines for interim reference ranges for 14 proteins in serum based on the standardization against the IFCC/BCR/CAP Reference Material (CRM 470). International Federation of Clinical Chemistry. Community Bureau of Reference of the Commission of the European Communities. College of American Pathologists

The release in 1993 of a new reference material for serum proteins, CRM 470/RPPHS 5 has given rise to a great improvement in the between-laboratory variability of serum protein measurements worldwide. Conversion to the new reference material results in significant changes in reference values for some proteins. The establishment of new reference ranges will take a considerable time, and in the interim several professional societies and diagnostic companies have agreed to use consensus reference ranges based on studies already undertaken.

Robust nonradioactive oligonucleotide ligation assay to detect a common point mutation in the CYP2D6 gene causing abnormal drug metabolism

A new nonradioactive oligonucleotide ligation assay for the detection of a common point mutation in the CYP2D6 gene is presented. The assay takes advantage of simultaneous time-resolved fluorescence measurements of lanthanide-labeled probes and the specificity of T4-DNA ligase in combination with the polymerase chain reaction. This strategy makes it possible to perform the assay using both the wild-type-specific and mutant-specific probes simultaneously, securing an internal control in each reaction. We show that the allele-specific ligation part of the assay can be performed with great accuracy over a wide range of temperatures, salt concentrations, and T4-DNA ligase concentrations. This eliminates the risk of false-positive or false-negative reactions due to variations in these factors. Because the assay is simple to perform, is very reliable, and can be partly automated, we conclude that it is well-suited for analysis in a routine laboratory.

Robust nonradioactive oligonucleotide ligation assay to detect a common point mutation in the CYP2D6 gene causing abnormal drug metabolism

A new nonradioactive oligonucleotide ligation assay for the detection of a common point mutation in the CYP2D6 gene is presented. The assay takes advantage of simultaneous time-resolved fluorescence measurements of lanthanide-labeled probes and the specificity of T4-DNA ligase in combination with the polymerase chain reaction. This strategy makes it possible to perform the assay using both the wild-type-specific and mutant-specific probes simultaneously, securing an internal control in each reaction. We show that the allele-specific ligation part of the assay can be performed with great accuracy over a wide range of temperatures, salt concentrations, and T4-DNA ligase concentrations. This eliminates the risk of false-positive or false-negative reactions due to variations in these factors. Because the assay is simple to perform, is very reliable, and can be partly automated, we conclude that it is well-suited for analysis in a routine laboratory.

Serum lactate dehydrogenase isoenzyme 1. An early indicator of relapse in patients with testicular germ cell tumors

The present study aimed to evaluate serum lactate dehydrogenase isoenzyme 1 (S-LD-1) as an indicator of relapse for patients with testicular germ cell tumors. Twenty-seven patients were investigated with repeated determinations of S-LD-1 from diagnosis to relapse; 9 had seminoma and 18 nonseminomatous tumors. Eleven of 21 had raised S-LD-1 at relapse (4 with seminoma). Seven of the 27 patients had a raised S-LD-1 for median 2 months (1.4-4.5 months) before relapse was detected. Thus, S-LD-1 is of use, complementary to clinical examinations, determinations of serum alpha fetoprotein and human chorionic gonadotropin, and CT scans, in monitoring patients with testicular germ cell tumors for relapse.

An enzyme-linked immunosorbent assay for plasma-lactoferrin. Concentrations in 362 healthy, adult blood donors

Very different concentrations of plasma-lactoferrin in healthy adults have been reported in the literature. We compared three commercially available lactoferrins and lactoferrin purified in our laboratory as calibrators in an ELISA. No statistical differences among these preparations of lactoferrin were detected. The concentration of purified lactoferrin was measured by dry weight, and efforts were made in order to minimize loss of purified lactoferrin by adhesion to tubes etc. and thus, secure accuracy of the method. Dilutions were made in PBS 0.01 mol l-1 with NaCl 0.436 mol l-1, (NH4)2SO4 0.5 mol l-1, BSA 5 gl-1 and normal rabbit IgG 10 mg l-1, which was shown to give parallel dilution curves of primary calibrator, secondary calibrator and plasma samples. No significant difference in the content of lactoferrin in neutrophils (median; range) among men (1.78; 0.83-4.48 micrograms 10(-6) neutrophils; n = 20) and women (2.12; 1.16-9.30 micrograms 10(-6) neutrophils; n = 14) was found. Lactoferrin was analysed in EDTA-plasma obtained from 135 female and 227 male blood donors. Median concentrations were 84.7 and 97.8 micrograms l-1 respectively, while 2.5% and 97.5% reference limits (with 90% confidence intervals) were estimated to 42.9 (38.7-47.4) micrograms l-1 and 166.9 (151.0-186.3) micrograms l-1 for women and 52.3 (49.1-55.6) micrograms l-1 and 189.9 (175.9-206.4) micrograms l-1 for men, respectively.

Appropriate sera for calibration and control of specific protein assays

An ultracentrifugation technique is described which makes it possible to prepare protein calibrators and control sera which (i) are stable (more than 8 years documented for nine proteins), (ii) are clear (remains clear for more than 8 years at -80 degrees C), (iii) contain genuine proteins (documented electrophoretically and immunologically), and (iv) are reproducible to prepare (collection and preparation).

Discrepancy in HbA1c measurements performed at different local laboratories and at a selected central reference laboratory

As participants in a general practice intervention study, 66 patients had their HbA1c measured both at a local and at a selected central reference laboratory. A discrepancy in the results was observed, as 97% of the results measured locally were lower than the centrally determined results. Bias (as calculated from mean value of measured HbA1c) between local laboratories and the central laboratory was measured to -1.47% HbA1c. A bias of this magnitude gave "problems" both to the general practitioners, patients and laboratories. To reduce the "problems" a bias of 0.5% HbA1c is estimated to be acceptable. But, to avoid these "problems" totally, a bias of 0.25% HbA1c is estimated to be the highest allowed bias. For HbA1c, a control system for both control of method standardisation and for specificity is described.

Analytical bias by contamination from hemolysis in determination of serum lactate dehydrogenase isoenzyme 1 in patients with testis germ cell tumors

We investigated the impact of correction for contamination from hemolysis on serum lactate dehydrogenase isoenzyme 1 (S-LD-1) determinations. A study of hemolysates from 7 control patients showed a mean correction factor for the contamination of 0.1 U/L S-LD-1 for each 1 mg/L serum hemoglobin (S-Hb). S-LD-1 in a series of blood samples from 44 patients (EJC 1992;28:410-5) would decrease median 24 U/L (range 8-70 U/L) if the measurements were corrected with this factor. So we advice to correct S-LD-1 determinations for the contamination with a common correction based on the S-Hb concentrations in the samples.

A Nordic reference serum suitable for use as trueness control in the clinical routine laboratory

The described reference serum is characterized by: liquid human serum at "normal" level stored in frozen state at -80 degrees C; minimum damage of proteins; aseptic preparation; cryoprecipitate and excess fibrin removed; serum cleared by ultracentrifugation; pH at 7.2-7.6; available in sealed glass ampoules with inert gas (one ml serum in each); specified components among most frequently analyzed analytes; homogeneity assured and stability monitored; produced under strict rules for good manufacturing practices (GMP). The assigned values are traceable to reference measurement procedures and reference materials of highest achievable metrological level; according to the present proposal the maximum allowable uncertainty of the assigned value is based on biological variation (shared common reference intervals); the uncertainty should ideally not exceed 1/5 of the maximum allowable bias of results obtained on patients samples (even 1/2 would theoretically be acceptable and, for a practical guide approximately < 1% may suffice). The present document provides some guidance of how the reference serum could be established in practice. The document also indicates the use of the material and further extension of the concept. The present work is done as a NORDKEM project.

External analytical quality assurance for proteins

In the Nordic Protein Project an external control scheme (external quality assessment) was combined with the two other indispensable aspects of analytical quality, i.e. standardization (with a common high quality calibrator) and specification of needed analytical quality for sharing common reference intervals for nine serum proteins in the Nordic countries. The quality specifications are reliable for the purpose and given in clinical chemical terms--ready for application to the control systems. Further, a control design for disclosing external and internal errors, separately, is designed with respect to calibration and robustness towards analytical interference from turbid patient samples.

A programme for assigning target values for external quality assessment schemes in countries with no authorized reference laboratories. Annex. Experiences with deviating results on Ektachem 700 XR

The use of consensus values in external quality assessment schemes (EQAS) involves several problems and should preferably be replaced with target values obtained by methods of high metrological level. However, such values are difficult to obtain. In the present study we transferred values from the NIST (former NBS) certified reference serum SRM 909 to lyophilized and frozen test sera for various inorganic components using flame absorption or flame emission spectrometry. Enzyme values were assigned by laboratories of members of the former Scandinavian Enzyme Committee. The assignment was based on 2-4 determinations each day through 3 days of experiment. A total of 10 laboratories participated in the work. The results were utilized in a Danish EQAS. One practical concern is the fairly long time (9 months) which was needed for production, collection and compiling all data. To get an impression of how much dry chemistry analysers, e.g, could influence consensus values a Kodak Ektachem 700 XR was studied using lyophilized and frozen sera. The results are reported in the annex. On NIST SRM 909 the values found for sodium(I) were 6% too high even though the findings on frozen human sera were accurate. For aspartate aminotransferase a result three times the target values was found on a human lyophilized serum, while the values on the frozen sera only were slightly too high.

Serum lactate dehydrogenase isoenzyme 1 activity in patients with testicular germ cell tumors correlates with the total number of copies of the short arm of chromosome 12 in the tumor

The aim of our study was to assess the relationship between the serum lactate dehydrogenase isoenzyme 1 (S-LDH-1) activity in patients with testicular germ cell tumors and the number of copies of the short arm of chromosome 12 (12p) present in tumor. Twenty-seven adult patients with measurable tumor lesions were studied. Twenty-five had three or more copies of chromosome 12 per cell in the tumors. Nineteen had one or more copies of a specific chromosomal abnormality, an isochromosome of the short arm of chromosome 12, i(12p). Fourteen had increased S-LDH-1 levels. S-LDH-1 activity correlated significantly with the product of total tumor volume and the total number of copies of the short arm of chromosome 12 present per cell (total tumor 12p). We conclude that the total number of copies of the short arm of chromosome 12 in the tumors is most probably a factor contributing to the LDH-1 activity released from the tumors.

Serum lactate dehydrogenase isoenzyme 1 and tumour volume are indicators of response to treatment and predictors of prognosis in metastatic testicular germ cell tumours

44 patients with metastatic testicular germ cell tumours treated with cisplatin-based chemotherapy were evaluated for prognostic implications of clinical characteristics. 22 obtained complete remission by the initial chemotherapy, and 30 are disease-free. S-LDH-1 had an overall predictive value regarding the response of 80%, S-LDH of 64%, S-AFP of 62%, and S-hCG of 62%. In multivariate analysis regarding response, only tumour volume classified according to the Royal Marsden system (P = 0.0036) and S-LDH-1 (P = 0.0069) yielded information. Regarding survival, S-LDH-1 (P = 0.0141) and an estimate of total tumour mass (P = 0.0171) had most impact with additional information from S-hCG only (P = 0.0536). We conclude that S-LDH-1 may be used as a tumour marker in addition to S-hCG and S-AFP in patients with metastatic testicular germ cell tumour.

Screening for haemochromatosis: prevalence among Danish blood donors

Hereditary haemochromatosis is an autosomal recessive disease that is genetically expressed by excessive accumulation of iron in the tissues, resulting in cirrhosis, diabetes mellitus, cardiomyopathy and hypogonadism. As the disease may be diagnosed before the appearance of symptoms, and prevented by repeated phlebotomies, there are strong implications for adoption of a screening procedure. Determinations of transferrin saturation (TS) and serum ferritin concentration (SF) were used to screen 4302 blood donors, who were selected for follow-up studies if they fulfilled any of the following three criteria: (i) TS greater than or equal to 0.7; (ii) TS greater than or equal to 0.5 together with SF greater than or equal to 150 micrograms l-1; (iii) SF greater than or equal to 300 micrograms l-1. A total of 58 subjects who fulfilled at least one of these criteria were reinvestigated, after which 18 individuals still fulfilled at least one criterion. Fifteen subjects having SF greater than or equal to 300 micrograms l-1 were offered liver biopsy and thirteen of these accepted. In one individual, no stainable iron was detected, and two subjects did not fulfil the previously established diagnostic criteria for the diagnosis of hereditary haemochromatosis. Ten subjects who had a high TS and liver iron grade 2-4 according to Bassett were classified accordingly as homozygotes. On the basis of these results, the prevalence of haemochromatosis in Denmark was estimated to be 0.0037-0.0046.

Screening for haemochromatosis: influence of analytical imprecision, diagnostic limit and prevalence on test validity

Serum ferritin (S-ferritin) and the saturation of transferrin iron-binding capacity (TIBC-sat) were evaluated as screening procedures for idiopathic haemochromatosis in a non-diseased population. Special attention was paid to the influence of the analytical quality, but the effects of prevalence and discrimination limits were also considered. Changes in the analytical quality for S-ferritin and TIBC-sat used as single tests highly influenced the number of false test-positives. Increasing the coefficient of variation from 0.0 to 0.2 resulted in nearly a doubling of the number of false test-positives to be further investigated. Using even a high, yet achievable analytic quality with a low coefficient of variation of 0.056 and 0.059 for S-ferritin and TIBC-sat respectively, screening procedures had unacceptably high fractions of false test-positives and false test-negatives associated with any discrimination limit. If the prevalence of haemochromatosis is 0.003, the predictive value of a positive test result did not exceed 0.05, accepting a fraction of false test-negatives of 0.025. This was found to be too low for a screening test to be used in the general population. The combined use of S-ferritin and TIBC-sat resulted in higher performance with a sensitivity of 0.90, a specificity of 0.99 and a predictive value of a positive test result of 0.29 if a fraction of misclassification of 0.01 is accepted for each.

Analytical goal setting prior to selection of a method for glycated haemoglobin

Without a consensus on either a reference method or a single glycated haemoglobin standard, the individual laboratory has to establish and secure its own assay method. Prior to selection of a method for glycated haemoglobin we used an assessment model and defined our clinical goals. Based on clinical goals it was possible to set up goals of analytical quality and evaluate available literature concerning the performance characteristics of different assay methods. We wanted a method that measured HbAlc, without measuring the labile intermediate pre-HbAlc, and that provided separate detection of haemoglobin variants. Low imprecision was required to secure the measured, non-diabetic reference interval, and to minimize random analytical error. We found the isoelectric focusing method the only one able to meet our goals, and a method evaluation was carried out. Clinical goals of analytical quality should always be defined before method selection and implementation. Using this model, we were able to establish a precise isoelectric focusing method that fulfilled our clinical goals by measuring HbAlc with a total coefficient of variation of 2.1% and a non-diabetic reference interval from 5.2% to 6.8% HbAlc.