Linearity Verification and Evaluation of Within and Between-run Precision of the Albumin Assay Kit   IntroductionAlbumin is a protein, synthesised in the liver and most abundantly found in plasma. It represents 50% of the body’s total content of protein. It is extremely soluble in water, as a result of it having a negative charge at a physiological pH. It plays a pivotal role in maintaining oncotic pressure. Albumin is involved in the transportation of steroids, calcium ions, ligands and bilirubin. Clinically, high levels of this protein are rarely encountered and are seen in cases of dehydration. Low levels possess greater diagnostic value and can be seen in hepatic disease, gastrointestinal loss, inflammation and ascites. In the Albumin Assay Kit that is utilised in this practical, the albumin binds to the bromocresol green (BCG) indicator. An albumin-BCG complex forms and the absorbance of this is measured spectrophotometrically at 578nm. The concentration of albumin in the sample is directly proportional to the intensity of the colour that is formed by the complex (Caraceni et al., 2013).The term linearity refers to there being a linear relationship between two variables. These variables that are both increasing, are directly proportional and so, the data will be represented in a straight line (Schneider et al., 2010). In the case of this practical, the linearity of the Albumin Assay Kit was evaluated and the manufacturer’s claim of this assay being linear up to 62.2g/l is to be verified. An 80g/l albumin concentration standard was given and dilutions of this standard were made to verify the assay’s linearity. The absorbance of each of the standards were plotted versus the concentration of each standard. The relationship between both variables was then evaluated to test the manufacturer’s claim. The CLSI’s document, EP15-A2 is  intended to verify if the performance of a laboratory is consistent to that claimed by the manufacturer, i.e verification of the precision of the manufacturer’s claim (Chesher, 2008).Another aspect of this practical that was evaluated is the within-run precision, also known as intra-assay precision and the between-run precision or inter-assay precision, as it is also referred to as. Within-run precision is used to determine the precision of the assay being used. This is carried out by measuring the absorbance of a pooled plasma sample a number of times, assessing the repeatability of the results and so, the precision of the results. The results are assessed on how close they are in agreement to each other after being measured successively under the same conditions. In this practical the sample was measured 10 times. The ideal amount of times a sample should be repeated at a given level, in a single day and in a single run is 20. This will give an idea of the repeatability of the assay. The mean, standard deviation and percentage coefficient of variation were all obtained in order to determine how precise the assay was. Between-run precision was also evaluated. This is a better indicator of the precision of a method than that of within run. This is so as it is a measure of random error of the method. One of these errors being different people carrying out the assay. The EP05-A2, a document produced by the CLSI, is intended to determine the precision of the methods used by the laboratory. This assessment is carried out at two levels, containing at least ten samples per run and should contain at least one sample of quality control (Chesher, 2008). The class results for the albumin concentration (g/l) for both Control 1 and Control 3 were used for this evaluation. Again, the mean, standard deviation and percentage coefficient of variation were all determined and compared to that of the manufacturers. This enabled the comparison of the class results for each control and the manufacturer’s control results in terms of precision MaterialsBeakerEppendorf tubesTube rack1000µl pipette tips10µl pipette tips100 – 1000µl pipette. Manufacturer: Labnet. Serial number: 7440605492 – 20µl pipette. Manufacturer: Labnet. Serial number: 7044030281Control 1Control 3Calibrator. 32g/l albumin concentrationSalineAlbumin standardPatient 1 plasma samplePatient 2 plasma samplePatient 3 plasma samplePatient 4 plasma sampleAlbumin BCG reagent. Lot number: 3054. Expiry: 2019-04VWR lab dancer vortex. Model: S040.Timer. Manufacturer is VWR. Serial number: F153504Pooled plasma sampleUV spectrophotometer at 600nmThe intensity of the colour that is formed by the complex is directly proportional to the concentration of albumin in the sample. MethodsTable 1: Serial dilution table, showing the concentration of albumin (g/l), the dilution that was carried out for each concentration and the volumes of albumin (µl) and saline, that were used to make up each albumin standard.Albumin standard (g/l)Dilution Volume of albumin (µl)Volume of saline (µl)808/810 x 2 = 200707/8 – 7 parts sample + 1 part saline70 x 2 = 14010 x 2 = 20606/8 – 6 parts sample  + 2 parts saline60 x 2 = 12020 x 2 = 40404/8 / ½ = equal amounts of both sample and saline50 x 2 = 10050 x 2 = 100202/8 = 2 part sample + 6 parts saline20 x 2 = 4060 x 2 = 12000/8 = 0 parts sample + 8 parts saline010 x 2 = 20Volumes were multiplied by 2 as 100µl to 200µl of each standard was required in the practicalFor all other methods refer to Audit Diagnostics – Albumin BCG Multi-Purpose (MPR) Liquid Reagent kit insert. ResultsThe following are the results obtained for the albumin colourimetric manual method. The Audit Albumin Kit was used in this practical. The linearity of this kit was assessed to verify the manufacturer’s claim that this assay is in fact linear to 62.2g/l. This kit was also used to evaluate both the within-run precision for the assay, using a pooled plasma sample and the between-run precision of the assay using the class albumin concentrations (g/l) of Control 1 and Control 3. The between-run precision that was obtained was inferior to that of the manufacturer’s, while the within-run precision obtained was quite good.Tube Absorbance (A) of first reading (a)Absorbance (A) of second reading (b)Average Absorbance (A)Concentration of albumin (g/l)Calibrator0.5490.5520.55132Control 10.5450.4870.5163080g/l albumin standard1.4061.4321.419/70g/l albumin standard1.2791.2981.289/60g/l albumin  standard0.9731.2291.101/40g/l albumin standard0.7150.7030.709/20g/l albumin standard0.1180.1260.122/0g/l albumin standard0.000/0.000/Patient 10.5830.5100.54732Patient 20.4470.6730.56033Patient 30.9830.9810.98257Control 30.9310.9160.92454Patient 40.6120.6040.60835Table 2: Absorbance (A) of the first reading (a), absorbance (A) of the second reading (b) average absorbance (A)  and albumin concentration (g/l) of the calibrator, control 1, control 3, standards ranging from 0g/l to 80g/l of albumin and the four patient samples using the Audit Albumin Kit colourimetric manual method.  The absorbance readings were taken at 600nm on a spectrophotometer.Reference range for control 1: 23-33g/l  Reference range for control 3: 39-54g/l.Both controls are within range and are highlighted yellow to signify this.  Sample calculation of how to obtain the average absorbance (A) for each sample:Patient 4 – Absorbance (A) from first reading + Absorbance (A) from second reading / 2                                                                =    Average absorbance (A)0.612A+ 0.604A / 2 = 0.608ASample calculation of how to obtain the albumin concentrations (g/l) for the controls – C1 and C3 and the patient samplesAlbumin concentration = OD of the sample / OD of the calibrator x Concentration of the calibrator.e.g. C1: 0.516A / 0.551A x 32g/l = 29.97 = 30 g/l.Figure 1: Linear graph depicting the absorbance (A) of the albumin standards versus their expected concentrations (g/l).This linear graph verifies the manufacturer’s claim that this albumin assay is indeed linear to 62.2g/l. After this concentration, the linearity ceases and a curve begins to develop      Table 3: Evaluation of the within-run/ intra-assay precision of the Audit Albumin Kit using the manual colourimetric method. Absorbance (A) readings, average absorbance (A) and albumin concentration (g/l) of the calibrator, Control 1, Control 3 and the pooled plasma sample that had its absorbance read ten times as a measurement of within-run precision. A spectrophotometer was used at 600nm. Also included in this table are the mean (g/l), standard deviation and the percentage coefficient of variation (%CV) of the pooled plasma sample.TubeAbsorbance readings (A)Average absorbance (A)Albumin concentration (g/l)Calibratora)0.868           b)0.8690.86932Control 1 a)0.852           b)0.8490.85131Control 3 a)1.264           b)1.3291.29748Pooled plasma 11.014/37Pooled plasma 21.014/37Pooled plasma 31.029/38Pooled plasma 41.060/39Pooled plasma 51.001/37Pooled plasma 61.115/41Pooled plasma 71.011/37Pooled plasma 8 1.079/40Pooled plasma 91.111/41Pooled plasma 101.111/41Mean of pooled plasma replicates (g/l)38.8Standard deviation of pooled plasma replicates1.81%CV of pooled plasma replicates4.66Reference range for control 1: 23-33g/l Reference range for control 3: 39-54g/l.Both controls are within range and are highlighted yellow to signify this.%CV= Percentage coefficient of variationMean and standard deviation of the pooled plasma replicates was carried out on a Casio fx-83GT Plus calculation in the STAT mode.Calculation on how to obtain the %CV:       Standard deviation / Mean (g/l) x100 = %CV. Table 4: Evaluation of the between-run/ inter-assay precision of the Audit Albumin Kit using the manual colourimetric. The class results (1-14 readings) for the albumin concentration (g/l) for Control 1 and Control 3 were used. The reference range (g/l) for each control is included in the table, along with the manufacturer’s and class’s mean (g/l), standard deviation and %CV of Control 1 and Control 3.  The manufacturers took 20 readings of each control. Control 1 albumin concentration (g/l)Control 3 albumin concentration (g/l)Reference range (g/l)23-3339-54Manufacturer’s mean  (g/l)45.3531.04Manufacturer’s standard deviation 0.7960.620Manufacturer’s (%CV) 1.752.00133522314933352425  3752953625477315183148933531032541131541229501333601430 50 Class mean (g/l)30.4350.71Class standard deviation2.685.07Class (%CV)8.8110.00 Two of the class results for Control 3 were outside of the reference range (37 is below and 60 is above the reference range) and are highlighted red to emphasize this.DiscussionThe controls were all within range in the first part of this practical, where the manufacturer’s claim was verified. This can be seen is table 2. As these controls were within range, the results for the assay were valid and so, the assay did not have to be repeated. Again, for the within-run precision assay, as seen in table 3, the controls were all within the reference range. However, for the between-run assay, as shown in table 4, two of the class results for Control 3 were out of range. This could have been due to inaccurate pipetting of the 10µl of the controls or contamination of the control or reagent, causing the absorbance readings to be out of range.From observing figure 1, it is evident that the Albumin Assay Kit is in fact linear up to 62.2g/l, verifying the manufacturer’s claim. After this concentration, linearity appears to dissipate and a curve begins to develop. For this reason, all concentrations higher than 62.2g/l should be diluted, in order to continue to achieve linearity. From observing the graph, it is evident that the 20g/l albumin standard is not in linearity with and does not correlate with the rest of the standards. Reasons for this may be that the exact 10µl volume of the albumin standard was not pipetted accurately to the Eppendorf tube or that the tube was not vortexed prior to incubation, resulting in an erroneous absorbance reading. Contamination could also be the reason for this low albumin concentration. The four patient albumin concentrations fell within the linear range of the albumin standards. However, no further comment can be made on these results as no patient details were given.In table 3, it is clear that the within-run precision was quite good. A %CV of less than 5 was obtained, emphasising good agreement within the replicates (Westgard.com, 2018). The precision could be further increased and a lower %CV obtained if the pooled plasma sample was repeated until the sample was read 20 times. Hence, this amount of repeatability is a good estimate of within-run precision.The between-run precision obtained by the class was far below that of the manufacturer. Both controls achieved a %CV of greater than 5, highlighting imprecision in the assay (Westgard.com, 2018).  This is depicted in table 4. However, the manufacturer’s between-run precision is superior as they tested the sample 20 times, as opposed to the class’s 14.  Precision increases upon repeatability of a sample.  As the results that were obtained were that of the class, this could be a reason for the poor precision. Had one person obtained and reported results for all 14 readings of each control, more precision may have been seen.        ConclusionIn all, the manufacturer’s claim was verified as the Albumin Assay Kit utilised in this practical was linear up to 62.2g/l. Both the within-run and between-run precision were evaluated. The between-run precision obtained by the manufacturer was shown to be far more precise than those obtained by the class, while the within-run precision was shown to be quite precise. The aims have been met and understood for this practical. ReferencesCaraceni, P., Tufoni, M., Bonavita, M.E. (2013) Clinical use of albumin. Blood Transfusion [Online] 4 pp.18-25.Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3853979/pdf/blt-11-s18.pdf  [Accessed on: 28th September 2018)Chesher, D. (2008) Evaluating assay precision. The Clinical Biochemist Reviews [Online] 29 pp.s23-s26. Available at: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2556577/pdf/cbr29spgs23.pdf [Accessed on: 28th September 2018]Schneider, A., Hommel, G. and Blettner, M. (2010) Linear Regression Analysis. Deutsches Ärzteblatt International. [Online] 107(44) pp.776–782. Available at: https://pdfs.semanticscholar.org/fe01/f0df14412c9265b0186fdfaecaf1462b56e3.pdf?_ga=2.142852981.1947834640.1538344743-1818968726.1537298205  [Accessed on: 28th September 2018].Westgard.com (2018) Mean, Standard Deviation, And Coefficient Of Variation. [Online]Available at: https://www.westgard.com/lesson34.htm [Accessed on: 28th September 2018]