Reference Interval Harmonization: Harnessing the Power of Big Data Analytics to Derive Common Reference Intervals across Populations and Testing Platforms

BACKGROUND

Harmonization in laboratory medicine is essential for consistent and accurate clinical decision-making. There is significant and unwarranted variation in reference intervals (RIs) used by laboratories for assays with established analytical traceability. The Canadian Society of Clinical Chemists (CSCC) Working Group on Reference Interval Harmonization (hRI-WG) aims to establish harmonized RIs (hRIs) for laboratory tests and support implementation.

METHODS

Harnessing the power of big data, laboratory results were collected across populations and testing platforms to derive common adult RIs for 16 biochemical markers. A novel comprehensive approach was established, including: (a) analysis of big data from community laboratories across Canada; (b) statistical evaluation of age, sex, and analytical differences; (c) derivation of hRIs using the refineR method; and (d) verification of proposed hRIs across 9 laboratories with different instrumentation using serum and plasma samples collected from healthy Canadian adults.

RESULTS

Harmonized RIs were calculated for all assays using the refineR method, except free thyroxine. Derived hRIs met proposed verification criterion across 9 laboratories and 5 manufacturers for alkaline phosphatase, albumin (bromocresol green), chloride, lactate dehydrogenase, magnesium, phosphate, potassium (serum), and total protein (serum). Further investigation is needed for some analytes due to failure to meet verification criteria in one or more laboratories (albumin [bromocresol purple], calcium, total carbon dioxide, total bilirubin, and sodium) or concern regarding excessively wide hRIs (alanine aminotransferase, creatinine, and thyroid stimulating hormone).

CONCLUSIONS

We report a novel data-driven approach for RI harmonization. Findings support feasibility of RI harmonization for several analytes; however, some presented challenges, highlighting limitations that need to be considered in harmonization and big data analytics.

Performance of StatSensor Point-of-Care Device for Measuring Creatinine in Patients With Chronic Kidney Disease and Postkidney Transplantation

Background:

The StatSensor is a point-of-care device which measures creatinine in capillary whole blood. Previous studies reported an underestimation of the creatinine measurements at high creatinine concentrations and were performed in the prestandardization era for creatinine.

Objective:

This accuracy-based study evaluates the use of this device in kidney-transplanted patients and those with chronic kidney disease (CKD).

Design:

Cross-sectional diagnostic accuracy study.

Setting:

Nephrology outpatient clinic in an urban tertiary center.

Participants:

Adults with CKD or a functioning kidney transplant.

Measurements:

Duplicate StatSensor creatinine measurements were performed on capillary whole blood samples collected by direct fingerstick and SAFE-T-FILL collection device. Results were compared with simultaneous venous blood sampling for serum and plasma creatinine measured by an enzymatic method on the Roche Integra 400 mainframe analyzer with traceability to the ID-GC-MS (isotope dilution gas chromatography mass spectrometry) reference method.

Methods:

Deming regression, Pearson correlation coefficient, and Bland-Altman analysis were used to assess accuracy and comparability between capillary whole blood measured by StatSensor and plasma creatinine measured by routine analyzer with traceability to the reference method. Estimated glomerular filtration (eGFR) rates were calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation and concordance with Kidney Disease Improving Global Outcomes (KDIGO) CKD stage classification was evaluated.

Results:

There were 60 participants (mean age = 61.9 ± 15.0 years, 55% men, 33% transplant, mean plasma creatinine = 137 ± 59 µmol/L). Bland-Altman analysis indicated a positive mean bias of 12.7 µmol/L between StatSensor fingerstick creatinine measurement and plasma creatinine. Comparison of eGFR (CKD-EPI) calculated from the StatSensor fingerstick creatinine versus plasma creatinine showed misclassification across all KDIGO CKD stages. Postanalytical correction of the bias did not improve misclassifications. The use of mean of duplicate StatSensor creatinine results did not improve performance compared with the use of singlet results.

Limitations:

Single center, limited participant numbers.

Conclusions:

The results of our study suggest that the limiting characteristics of the StatSensor device are not only bias, but also imprecision. The level of imprecision observed may influence clinical decision-making and limit the usefulness of StatSensor as a CKD screening tool. If choosing to utilize it for either screening for or monitoring CKD, it is essential that clinicians understand the limitations of point-of-care devices and apply this knowledge to test interpretation.

A Summary of Worldwide National Activities in Chronic Kidney Disease (CKD) Testing

Chronic kidney disease (CKD) is a major public health issue worldwide and is associated with adverse health outcomes, especially in low- and middle-income countries. In a cash limited healthcare system, guidelines that improve the efficiency of health care free up resources needed for other healthcare services. This short review presents some examples from national acitivities in CKD testing, including countries throughout the globe: Mexico in North America, Uruguay in South America, Italy in Europe, Nigeria in Africa and India in Asia. Considering the fact that treatment of CKD is cost-effective and improves outcomes, this observation argue in favor of including CKD in national guidelines and noncommunicable chronic disease (NCD) programs. This diverse example of national activities fullfil the very first step in achieving this goal.

State of the art in trueness and interlaboratory harmonization for 10 analytes in general clinical chemistry

CONTEXT

Harmonization and standardization of results among different clinical laboratories is necessary for clinical practice guidelines to be established.

OBJECTIVE

To evaluate the state of the art in measuring 10 routine chemistry analytes.

DESIGN

A specimen prepared as off-the-clot pooled sera and 4 conventionally prepared specimens were sent to participants in the College of American Pathologists Chemistry Survey. Analyte concentrations were assigned by reference measurement procedures.

PARTICIPANTS

Approximately 6000 clinical laboratories.

RESULTS

For glucose, iron, potassium, and uric acid, more than 87.5% of peer groups meet the desirable bias goals based on biologic variability criteria. The remaining 6 analytes had less than 52% of peer groups that met the desirable bias criteria.

CONCLUSIONS

Routine measurement procedures for some analytes had acceptable traceability to reference systems. Conventionally prepared proficiency testing specimens were not adequately commutable with a fresh frozen specimen to be used to evaluate trueness of methods compared with a reference measurement procedure.

Regional implementation of creatinine measurement standardization

Because patients may receive care at multiple locations within a geographic area, serum creatinine measurements must be standardized across laboratories to enable comparisons of reported estimated glomerular filtration rate (eGFR). The results of a successful creatinine standardization program designed to minimize the contribution of laboratory error to the reporting of eGFR are reported; 107 laboratories, which tested creatinine on 124 analyzers from six different manufacturers, voluntarily participated. Each laboratory received a correction factor to apply to its creatinine measurements to standardize them to the isotope dilution mass spectrometry reference method. The adjusted values were then used to calculate eGFR using the Modification of Diet in Renal Disease (MDRD) equation. The standardization program reduced the average total error in the measurement of creatinine from 23.9 to 8.7% and the average analytical bias from 16.5 to 2.7%. Implementing this program on a larger scale could reduce the rate of incorrect classification of stage 3 chronic kidney disease by 84%.

Hypertriglyceridemic hyperapoB in type 2 diabetes

OBJECTIVES

Much less attention has been paid to LDL in type 2 diabetes than to VLDL or HDL. In particular, there are few data on apoB levels in these patients. Moreover, most reports have focused on mean lipoprotein levels and consequently there is little information on the frequencies of the various dyslipidemic phenotypes.

RESEARCH DESIGN AND METHODS

Plasma and lipoprotein lipids, apoB and apoA1 were measured by standardized methods. LDL particle size was determined by PAGE. The total cohort was divided into phenotypes by two different methods. The first was based on triglycerides (> or = or <1.5 mmol/l) and LDL cholesterol (> or = or <4 mmol/l), whereas the second was based on triglycerides (> or = or <1.5 mmol/l) and apoB (> or = or <120 mg/dl).

RESULTS

For the overall cohort, plasma triglycerides were elevated (2.13 +/- 1.6 mmol/l), total and LDL cholesterol were normal (5.34 +/- 1.1 and 3.28 +/- 0.88 mmol/l, respectively), and peak LDL size was reduced (252.9 +/- 5.8 A). HDL cholesterol was between the 25th and 50th percentiles of the general population (1.12 +/- 0.36 mmol/l). The average level of apoB was 114 +/- 29 mg/dl, a value that is between the 50th and 75th percentiles of the general population and is higher than that for LDL cholesterol, which was between the 25th and the 50th percentiles of the population. The results of the phenotyping analysis were as follows. Using the conventional approach, only 23% has abnormal LDL, i.e., an elevated LDL cholesterol level. Using the new approach, almost 40% has an elevated apoB and therefore an elevated LDL particle number. Only 12.8% has combined hyperlipidemia based on the conventional approach, whereas almost one-third had the equivalent, hypertriglyceridemic hyperapoB-based on the new algorithm. The severity of the dyslipoproteinemia in this group was noteworthy. Although the average LDL cholesterol was 3.91 mmol/l, a value just below the 75th percentile of the general population, the average apoB was 145 mg/dl, a value that approximates the 95th percentile of the population.

CONCLUSIONS

The dyslipidemic profile of patients with type 2 diabetes is not uniform. A substantial group have normal lipids and normal LDL particle number and size whereas others have markedly abnormal profiles. Diagnosis based on triglycerides and apoB rather than triglycerides and LDL cholesterol revealed that more than one in five had hypertriglyceridemic hyperapoB, which is characterized by hypertriglyceridemia, marked elevation of LDL particle number, small dense LDL, and low HDL, a constellation of abnormalities that is associated with markedly accelerated atherogenesis and therefore justifies intensive medical therapy.

Effect of L-carnitine treatment on very low density lipoprotein kinetics in the hyperlipidemic rabbit

This study examined the hypolipidemic effect of 4 weeks of L-carnitine treatment (170 mg/kg b.w./day) in New Zealand White rabbits fed a high fat diet (5% corn oil/0.5% cholesterol). Specifically, [3H] glycerol and [125I] very low density lipoprotein (VLDL) turnover studies were conducted to examine the effect of treatment on VLDL kinetics. The masses of plasma VLDL-triglycerides (VLDL-TG) and VLDL-apoprotein B (VLDL-apoB) were significantly increased by the high-fat diet. Four weeks of treatment with L-carnitine significantly reduced these masses. Kinetic analysis indicated that fat feeding reduced the fractional catabolic rates (FCRs) of VLDL-TG and VLDL-apoB relative to chow-fed controls. The transport of these VLDL components was not altered by the diet. L-carnitine treatment had no effect on the FCRs of VLDL-TG and VLDL-apoB or on the transport of VLDL-apoB. Yet, treatment significantly lowered the transport of VLDL-TG. These data indicate that the lipid-lowering effect of L-carnitine in this animal model was due, in part, to a decrease in the transport and not due to an alteration in the fractional catabolic rate of VLDL-TG.

Production of acyl-carnitines from the metabolism of [U-14C]3-methyl-2-oxopentanoate by rat liver and skeletal muscle mitochondria

A sensitive method of continuous on-line radio-high performance liquid chromatography (HPLC) was used to detect the specific radio-labelled acyl-carnitine esters derived from the oxidation of [U-14C]3-methyl-2-oxopentanoate by rat liver and muscle mitochondrial fractions. The recoveries of carnitine, acetyl-carnitine, propionyl-carnitine, 2-methylbutyryl-carnitine, and hexanoyl-carnitine were 98.7% (+/- 5.4; SEM, n = 3), 91.4% (+/- 7.6), 89.4% (+/- 5.2), 84.6% (+/- 6.8), and 87.9% (+/- 7.8), respectively, from quenched mitochondrial incubations. This method demonstrated that rat liver and muscle mitochondria generate acetyl-carnitine, propionyl-carnitine and 2-methylbutyryl-carnitine when incubated with [U-14C]3-methyl-2-oxopentanoate in the presence of carnitine. The production of acetyl-carnitine was almost similar in the 2 tissues. Muscle mitochondria produced higher amounts of propionyl-carnitine and 2-methylbutyryl-carnitine than liver mitochondria. These observations suggest a limited utilization of propionyl-CoA by muscle mitochondria which, through a mechanism of feed-back inhibition, may have contributed to the accumulation of 2-methylbutyryl-CoA. This study provides further evidence for the importance of carnitine in the modulation of the mitochondrial [acyl-CoA/[CoA] pool.

Lack of effect of oral L-carnitine treatment on lipid metabolism and cardiac function in chronically diabetic rats

L-Carnitine is necessary for the transfer of long-chain fatty acids into the mitochondrial matrix where energy production occurs. In the absence of L-carnitine, the accumulation of free fatty acids and related intermediates could produce myocardial subcellular alterations and cardiac dysfunction. Diabetic hearts have a deficiency in the total carnitine pool and develop cardiac dysfunction. This suggested that carnitine therapy may ameliorate alteration in cardiac contractile performance seen during diabetes. In this study, heart function was studied in streptozotocin diabetic rats given L-carnitine orally. Oral L-carnitine treatment (50-250 mg.kg-1.day-1) of 1- and 3-week diabetic rats increased plasma free and total carnitine and decreased plasma acyl carnitine levels. In both groups, myocardial total carnitine levels were increased. However, L-carnitine (200 mg.kg-1.day-1) treatment of diabetic rats for 6 weeks had no effect on plasma carnitine levels. Similarly, plasma lipids remained elevated whereas cardiac function was still depressed. These studies suggest that in the chronically diabetic rat, the route of administration of L-carnitine is an important factor in determining an effect.

Low concentration galactose determination in plasma adapted to the Cobas-Bio

Galactose elimination at blood concentrations lower than 2.22 mmol/L has been advocated as a measure of functional liver blood flow. We have adapted an assay employing galactose dehydrogenase (EC 1.1.1.48) to the Cobas-Bio to measure low galactose concentrations in plasma. The collection of blood in sodium fluoride/potassium oxalate anticoagulant tubes eliminated the necessity for the plasma deproteinization step required in similar, manual methods. The between run CV's for plasma samples spiked with galactose to concentrations of 0.13-0.5 mmol/L were 3.6% and 3.2%, respectively. Our automated assay was more precise and had a greater range of linearity than a manual galactose oxidase (EC 1.1.3.9) method set up in our laboratory (0.04-1.10 mmol/L as compared to 0.06-0.56 mmol/L). The total assay time was 20 min.

Therapeutic monitoring of cyclosporine: impact of a change in standards on 125I-monoclonal RIA performance in comparison with liquid chromatography

This study examines the measurement of cyclosporine (CsA) by 125I-monoclonal RIA, and describes the impact of the recent change in the standard curve provided. CsA concentrations in serum and whole-blood control samples measured by 125I-RIA were initially 8-18% higher than those by HPLC. During the first two months of 1989, a significant and sustained deviation in the 125I-RIA produced results that exceeded the HPLC results by 21-28% (P less than 0.001). Introduction of the new standard curve in March 1989 returned the concentration of the whole-blood controls to the previous range (11-12% above HPLC, P less than 0.001). Measurement of clinical samples from heart, liver, and bone-marrow graft recipients by 125I-RIA by both old and new kit standards produced a close linear correlation (y = 0.89 x - 19.02; r = 0.99; n = 75, range = 40-850 micrograms/L), with use of the new standards yielding results 82 (SD 8)% of those with the preceding assay. However, even with the new standard curve, CsA concentrations by 125I-RIA in the clinical samples exceeded those by HPLC by a factor of 1.37 (SD 0.18) to 1.52 (SD 0.19). Segregation for transplant type did not affect the RIA/HPLC ratio. The results suggest cross-reactivity of the 125I-RIA with material present in vivo.

Therapeutic monitoring of cyclosporine: impact of a change in standards on 125I-monoclonal RIA performance in comparison with liquid chromatography

This study examines the measurement of cyclosporine (CsA) by 125I-monoclonal RIA, and describes the impact of the recent change in the standard curve provided. CsA concentrations in serum and whole-blood control samples measured by 125I-RIA were initially 8-18% higher than those by HPLC. During the first two months of 1989, a significant and sustained deviation in the 125I-RIA produced results that exceeded the HPLC results by 21-28% (P less than 0.001). Introduction of the new standard curve in March 1989 returned the concentration of the whole-blood controls to the previous range (11-12% above HPLC, P less than 0.001). Measurement of clinical samples from heart, liver, and bone-marrow graft recipients by 125I-RIA by both old and new kit standards produced a close linear correlation (y = 0.89 x - 19.02; r = 0.99; n = 75, range = 40-850 micrograms/L), with use of the new standards yielding results 82 (SD 8)% of those with the preceding assay. However, even with the new standard curve, CsA concentrations by 125I-RIA in the clinical samples exceeded those by HPLC by a factor of 1.37 (SD 0.18) to 1.52 (SD 0.19). Segregation for transplant type did not affect the RIA/HPLC ratio. The results suggest cross-reactivity of the 125I-RIA with material present in vivo.

Human amnion as a bioprosthesis for bile duct reconstruction in the pig

Despite technical advances in management, the complication of late stricture formation and biliary sepsis still occur in bile duct reconstruction. In an attempt to avoid bilioenteric anastomosis, which bypasses the biliary sphincter mechanism, various biologic and artificial materials have been employed clinically and experimentally to replace the damaged bile duct. No satisfactory biliary replacement material has yet been found. In the experimental model of bile duct stricture that has been presented, human amnion bile duct injuries mimicking those seen in clinical practice were repaired using human amnion as a free graft. Noncircumferential duct loss appeared to be satisfactorily repaired using amnion, and the amnion repair was found to be as good as or superior to plastic repair; however, circumferential duct loss was not adequately repaired with the amnion graft.

Oral L-carnitine administration after jejunoileal by-pass surgery

Jejunoileal bypass surgery was performed on 14 morbidly obese women. Seven were treated with 3/g/day oral L-carnitine for ten days before surgery and for another ten days at six weeks following surgery. The other seven were not treated with carnitine. Carnitine administration was found to be without effect on plasma lipids, free fatty acids and ketones and on all other tested indicators. However, plasma levels of carnitine and the rate of urinary elimination of carnitine both after treatment and after an acute oral load were increased before and after surgery, indicating that carnitine was well absorbed by the shortened gut. Rate of elimination of short chain carnitine esters via the urine was also elevated after treatment. Surgery resulted in a marked decrease of plasma lipids but in no change in apolipoprotein A and B concentrations.

Perinatal changes in plasma carnitine levels in 4 species of mammal

Plasma levels of carnitine, acetylcarnitine, and beta-hydroxybutyrate rise perinatally in rats, guinea-pigs and sheep but not in rabbits. In the fetus, carnitine levels are high in rabbits and guinea-pigs but not in rats and sheep.

Plasma carnitine levels during intravenous feeding of the neonate

The premature infant has a limited capacity for fatty acid oxidation. This study shows that solutions commonly used for intravenous feedings in the newborn infant contain no carnitine. Infants maintained on this solution have significantly lower total, free, and acylcarnitine levels as compared to when they are fed orally with expressed human milk or a proprietary formula, which is known to contain carnitine. The exogenous supply of carnitine to the premature infant may have a significant influence on the ability to stimulate optimal fatty acid oxidation.

Regulation of phosphoenolpyruvate carboxykinase and fatty acid synthetase in brown fat of suckling rats

Infant rats were injected with prednisolone (0.5-5 mg/100 g wt). This caused phosphoenolpyruvate carboxykinase (PEPCK) activity to rise in liver and to decrease in brown fat. Fatty acid synthetase (FAS) activity remained unchanged in liver but increased in brown fat. A single injection of prednisolone caused hepatic PEPCK activity to remain elevated for at least 7 days. Brown fat FAS also remained high for that period. However, brown fat PEPK activity returned to normal on the third day after the injection. A single injection of prednisolone or cortisone to 5-day-old rats caused a transient elevation of the blood level of insulin and a prolonged decrease in that of growth hormone. No effect on the level of glucagon was noted. Injections of insulin had effects similar to those of prednisolone, i.e. a rise in hepatic and a fall in brown fat PEPCK. Using antibodies prepared to hepatic PEPCK it was shown that the observed changes were due to changes in the rate of synthesis of the enzyme. Using actinomycin D indirect evidence was obtained that changes in FAS activity of brown fat were also due to changes in the synthetic rate.