May 1998
This section provides details on the methods to be used for measurement of lipids in the WHO MONICA Project and their standardization.
List of abbreviations |
||
|---|---|---|
| Apo-B | Apoprotein B | |
| CDC | Centers for Disease Control, Atlanta, Georgia 30333, USA | |
| EDTA | Ethylene diamine tetraacetic acid | |
| EQA | External quality assessment | |
| G | Gravity | |
| HDL | High density lipoproteins | |
| HDL-C | High density lipoprotein cholesterol | |
| IKEM | Institute for Clinical and Experimental Medicine, Prague, Czechoslovakia | |
| IQC | Internal quality control | |
| LDL | Low density lipoproteins | |
| LRC | Lipid Research Clinics | |
| MDC | MONICA Data Centre | |
| MMC | MONICA Management Centre | |
| MQC | MONICA Quality Control Centre | |
| OD | Optical Density | |
| QA | Quality Assurance | |
| R.p.m. | Revolutions per minute | |
| RV | Reference value | |
| SCN | Serum thiocyanate | |
| St | Total standard deviation | |
| TC | Total cholesterol | |
| TG | Triglycerides (Triacylglycerols) | |
| VLDL | Very low density lipoproteins | |
| WHO-HQ | WHO Headquarters | |
| WHO-KRIS | WHO-Kaunas-Rotterdam Intervention Study | |
| WHO-LW | WHO Lipid Workshop (Prague, 26-28 January 1982) | |
| WHO-RLRC | World Health Organization Regional Lipid Reference Centre (WHO Collaborating Centre for Blood Lipid Research In Atherosclerosis and Ischaemic Heart Disease at the Institute for Clinical and Experimental Medicine (IKEM), Budejovicka 800, 14000 Prague-4, Czech Republic) | |
1.1 Total cholesterol (TC) is an obligatory blood (serum, plasma) measurement in the MONICA surveys. Wherever possible, samples should also be analysed for HDL-C. Laboratory directors who attended WHO-LW agreed upon HDL-C measurements. Serum thiocyanate (SCN) (for SCN see Part III, Section 3) was originally an obligatory parameter, but became optional after the initial survey.
1.2 Other optional but complementary measurements might include, e.g.: triacyglycerols (triglycerides), lipoprotein typing and composition (electrophoresis, ultracentrifugation, ultrafiltration and nephelometry, chromatography of lipoproteins, etc.), apoprotein concentrations [12] (immuno-electrophoretic, radial-diffusion and/or nephelometric methods), glucose, uric acid, urea, electrolytes, creatinine, thyroxine, haemoglobin, carboxyhaemoglobin, and in certain situations, trace elements.
2.1 Centres should collect MONICA blood samples in each survey over the same seasons to prevent seasonal variations in the levels of serum/plasma parameters. MCC collecting MONICA blood samples over the course of the entire year should randomize blood taking to ensure equal representation of age and sex subgroups over time.
Lipid laboratories may want to measure lipoprotein fractions and triglycerides and so requires a 16-hour fast. For epidemiological purposes such as in MONICA, where the primary lipid of interest is serum total cholesterol and secondarily high density lipoprotein cholesterol (HDL-C), blood can be taken at any time of day with the subject non- fasting; this facilitates the scheduling appointments at the survey clinic. For comparability with other MONICA centres it is desirable that blood pressure measurements be spread throughout the day and made in non-fasting subjects: for this reason any MCC that wishes to measure fasting lipid specimens should consider doing this separately from the routine survey clinic where blood pressure is measured.
For the purposes of longitudinal comparisons, the methods of measurement of lipids should be consistent between one MONICA survey and another. Any proposed changes such as of time of day and fasting/non-fasting status should be avoided or tested in pilot studies to ensure that they do not introduce bias.
2.1.1 Venipunctures should be carried out with the subject or patient in a sitting position. Prolonged venous occlusion can cause changes in the apparent concentrations of blood constituents [3]. Use of a tourniquet should therefore be avoided. If a good flow cannot be obtained in some subjects and the tourniquet has to be used, it must be released before the withdrawal of blood.
Standardization of the position (sitting position is recommended) is necessary since plasma volume changes occur when a standing subject assumes a recumbent position [20,3].
2.2 Serum should be used in preference to plasma. It is realized that the group of laboratories standardized with CDC may have to continue using these methods for MONICA. It follows that these laboratories may continue using EDTA plasma prepared according to the LRC instructions [20].
2.2.1 Either 10 ml vacuum tubes or syringes and glass tubes may be used to collect blood. Glass tubes should be equipped with stoppers made from material which does not react with blood constituents. When vacuum tubes are used, the type with stoppers un-lubricated with glycerol should be selected (glycerol causes interference with TG assays in enzymatic methods).
2.2.2 The use of vacuum tubes containing EDTA is recommended if plasma is used (the group of laboratories following LRC methods).
2.3 For serum preparation, blood samples are allowed to clot at not more than 20°C usually for up to one hour before centrifugation. There is evidence (personal communication from the Helsinki Centre) that this period can be prolonged by up to three hours.
Blood specimens should be centrifuged at a temperature of not more than 20°C (warning: Prolonged running of a non-refrigerated centrifuge may result in considerable warming of the centrifuge compartment and centrifuged samples) at a minimum of 1500 G for at least 10 minutes to separate serum from the clot. If a refrigerated centrifuge is unavailable, it may be necessary to cool blood samples before centrifugation (for instance, in a refrigerator or on melting ice). With a refrigerated centrifuge, centrifugation should be preferably done at 4°C. Whole blood samples must not be frozen during processing (this will cause haemolysis).
2.4 For plasma preparation the tube(s) filled with blood must be stoppered immediately (if vacuum tubes are not used) and inverted gently about 10 times to ensure prompt and thorough mixing of blood sample(s) with EDTA. Mixing should not be vigorous. According to LRC recommendations the blood samples are then cooled on melting ice. Within 3 hours (and preferably within one hour) the tubes should be centrifuged at 4°C in a refrigerated centrifuge at 1500 G for 30 minutes. If a refrigerated centrifuge is not available within 3 hours of collection, the samples may be centrifuged at room temperature within 1 hour of collection, and the plasma stored at 4°C.
2.5 After centrifugation, the serum/plasma should be promptly separated from clot or cells and transferred to a clean tube. The white cell layer (buffy coat) is not transferred with the plasma.
2.6 Haemolytic serum/plasma samples should be discarded and fresh samples should be taken from the subjects and analysed.
3.1 It is recommended that the TC and HDL-C levels should be estimated on the day of sample collection.
3.2 Samples can be stored for up to four days at +4°C [1,3,5]. If analyses of TC cannot be performed within 4 days the serum or plasma samples should be immediately stored at -20°C or lower in tightly stoppered glass vials [1].
3.3 Precipitation of Apo-B containing (LDL + VLDL) lipoproteins (see 4.5) should preferably be done on fresh (non-frozen) serum aliquots on the day of blood collection. Should it prove impossible to perform precipitation on fresh samples, the serum or plasma for HDL-C determination should be immediately frozen at -20°C or lower [1,3]. Precipitation should then be performed within 14 days. The sample should be thawed only once and well mixed. It should be recognized, however, that freezing may introduce a systematic error of up to 4% in HDL-C measurements [3].
3.4 Storage of serum/plasma samples for other purposes than immediate analysis of TC, HDL-C and SCN: if samples are to be stored for voluntary analysis, laboratories must select adequate storage conditions at which the measured parameters' concentrations would be stable.
3.5 Despite the problems associated with storage, MONICA centres are encouraged to archive replicate samples of serum as an insurance. If the laboratory's external quality control proves to be seriously deficient, the replicate samples give the opportunity for providing valid data, if the results from the primary specimens are disqualified.
4.1 It is recommended that an enzymatic cholesterol method with practically 100% cholesterol ester hydrolysis be used. Advantages include non-corrosive reagents, easy and fast operational steps, specificity, the possibility of using automated methods as well as manual methods with inexpensive instruments, the standardization of methodology in MONICA, etc. However, it is realized that a group of laboratories standardized with CDC using LRC methods based on extraction into isopropanol, and several laboratories using so called "direct methods" (see also 4.3) may continue using their methods for MONICA if the purpose of their own research should require such continuity.
4.2 The use of a number of commercial brands of enzymatic reagents for cholesterol analysis might diminish comparability of results between laboratories due, for example, to non-homogeneity of production lots or due to loss of kit enzyme activities on storage. Information collected at the WHO-LW showed that most of the participating laboratories have worked, or are going to work with the same commercial brand which is widely used in Europe. This is fortunate for the Project. The suitability of selected enzymatic methods will be checked by means of the EQA.
4.3 It is possible that a few laboratories taking part in MONICA may be unable to comply, for local reasons, with the recommendation to use an enzymatic method. These laboratories probably will continue to use the so-called "direct", "one- step", "Liebermann-Burchard" method(s) in manual or automated version(s), etc. The WHO-RLRC will circulate a questionnaire concerning analytical and associated procedures to be used in the different centres, with the aim of obtaining full information on the use of alternatives to LRC or enzymatic methods. It is the policy of MONICA to accept, in exceptional cases, participants not able to use the recommended methods. In such cases particular attention to methods for ensuring comparability will be necessary if performance within allowable EQA limits is unattainable with use of primary pure standards. The laboratories concerned may have to calibrate with serum calibrators provided by the WHO-RLRC (MQC).
4.4 Every participating laboratory should provide a detailed description of their entire procedure of sample collection and analysis to the MMC (WHO - Cardiovascular Diseases Unit) and to Dr D. Grafnetter (WHO-RLRC) at least six months before analysis of MONICA samples begins in the MCC concerned. An example of a recommended sequence of quality control standardization and participant samples for total cholesterol and HDL-cholesterol determination are shown in Figure 1.
4.5 Isolation of the HDL-fraction:
4.5.1 It was recommended before the start of MONICA that laboratories should use a "Phosphotungstate-Mg2+" (PT) precipitation method [2, 10, 11, 15, 31, 32, 33] for isolation of HDL from LDL + VLDL. In this method final concentrations after mixing serum (or plasma) with precipitation reagent(s) are: 3.6 g/litre for phosphotungstic acid and 0.045 M for Mg2+ [21, 24]. According to these papers, precipitation of LDL + VLDL is complete if the pH of the final mixture (serum/plasma + PT reagents) remains below pH 7.6. This method (see 4.5.2 - 4.5.7.3) was in general use at the time this Manual was initiated. By 1986, many centres had already switched to a new PT method described under 4.5.9.
4.5.1.1 Some laboratories, e.g. those with the LRC methodology, may need to continue with other precipitation techniques (Mg2+ -dextran, Mn2+ -heparin). Acceptance of these methods is based on the same policy as mentioned under 4.3.
4.5.2 Preparation of PT method [31, 32] reagents for precipitation at pH below 7.6 (for alternative concentrations and use of PT reagents see subsections 4.5.7-4.5.9):
4.5.2.1 Sodium phosphotungstate, 40.0 g/l, pH 7.4. Dissolve four grams of phosphotungstic acid (reagent grade) in about 60 ml distilled water. Add gradually while mixing and using a pH meter as much of 1 mol/l NaOH so as to reach pH 7.4 (usually somewhat less than 16 ml). Then make the volume up to 100 ml.
4.5.2.2 Magnesium chloride, 2.0 mol/l: Dissolve 40.6 grams of MgCl2·6H2O (analytical grade) in about 80 ml of distilled water and make the volume up to 100 ml.
4.5.2.3 Reagents should be kept at 4°C in between use and can be kept for up to 6 months, provided they show no bacterial growth.
4.5.3 Use of PT reagents is described under 4.5.2.1 and 4.5.2.2:
4.5.3.1 Add 100 µl of the phosphotungstate solution (4.5.2.1) to 1.0 ml of sample and mix (preferably by vortexing). Add 25 µl of magnesium chloride solution (4.5.2.2) and again mix well. Let each precipitated sample stand for ten minutes at room temperature and then centrifuge. For centrifugation conditions see subsection 4.5.4.
4.5.3.2 Volumes of sample and reagents could be reduced (e.g. halved) for the precipitation step.
4.5.3.3 It is also permissible to use the pre-mixed reagent. This is prepared and used as follows: phosphotungstate and magnesium chloride solutions are mixed on the day of use in the ratio of 100:25 (e.g. 4 ml + 1 ml). Then 125 µl of the pre-mixed reagent are added to 1.0 ml of sample and the solution is mixed well.
4.5.4 Centrifugation after precipitation: Since centrifugation at +4°C on one hand and +20°C on the other can lead to certain differences in HDL-C levels in some reconstituted lyophilized quality control materials (observations made in the WHO-RLRC) but not in fresh sera, it is recommended (for all but the LRC methods laboratories) to centrifuge "HDL samples" after PT precipitation of Apo-B containing lipoproteins at room temperature (not below +15°C or above +25°C) at 2000 G for 30 minutes.
4.5.5 After centrifugation the supernatant should be immediately transferred to a clean and dry tube.
4.5.5.1 The HDL-C concentration obtained after the colorimetric measurement in the supernatants has to be multiplied by 1.125 (due to sample dilution by the precipitant).
4.5.6 With study samples, only clear (non-turbid) supernatants should be used for subsequent cholesterol analysis. With the PT method (if TG are elevated) the supernatant may occasionally show some turbidity. In this case the precipitation step should be repeated with the serum/plasma sample diluted 1:1 with 0.9% sodium chloride solution in distilled water (saline, physiological solution). This usually results in a clear supernatant (but the precision of the cholesterol assay is decreased; remember that the result must be multiplied by two if the 1:1 diluted sample is used - see also subsection 4.6.4). Should the dilution of serum/plasma sample not give clear supernatants after precipitation and centrifugation, do not proceed. Any such irregularity and/or difficulty in obtaining clear supernatants should be recorded on the results form.
4.5.6.1 In some lyophilized quality control materials slight turbidity of supernatants may be unavoidable. The WHO-RLRC will distribute instructions for use of pools with individual shipments of EQA sets.
4.5.7 Alternative concentrations and use of PT method reagents:
Special (adjustable or constriction) pipettes are necessary to deliver the unusual volumes (e.g. 25 µl or 125 µl) of the above "classical" PT method reagents (see 4.5.1-4.5.6.1). The following modified reagent concentrations [33] and their use (see sections 4.5.7.1, 4.5.7.2 and 4.5.7.3) yield the same result but enable the use of normal precision pipettes:
4.5.7.1 Sodium phosphotungstate, 48.0 g/l, pH 7.4: dissolve 4.8 grams of phosphotungstic acid (reagent grade) in about 50 ml distilled water. Add gradually while mixing, and using a pH meter, as much of 1 mol/1 NaOH as necessary to reach pH 7.4. Then make the volume up to 100 ml.
4.5.7.2 Magnesium chloride, 3.0 mol/l: dissolve 60.9 grams of MgCl2·6H2O (analytical grade) in about 80 ml of distilled water and make the volume up to 100 ml.
4.5.7.3 Use of the modified PT method reagents:
Precipitation reagent is prepared by mixing 5 parts of sodium phosphotungstate (4.5.7.1) with one part of magnesium chloride (4.5.7.2) solution (e.g. 5 ml + 1 ml) on the day of use. Then 100 µl of the premixed reagent are added to 1.0 ml of serum/plasma sample and the whole is mixed well. Further steps (centrifugation, etc.) are the same as those starting at section 4.5.4. However, the multiplication (dilution) factor will be 1.1 in this case (not 1.125 as in 4.5.5.1).
4.5.8 Some laboratories have used a commercial PT precipitation kit and followed its working instructions, and they may wish to continue doing so. This is possible if results obtained with the use of such a kit are found compatible with the EQA criteria.
4.5.9 A modified precipitation method [35, 36] for determination of HDL- cholesterol using PT was developed and has been commercially available for several years (kit "HDL-cholesterol. Precipitant", Cat. no. 543004; Boehringer Mannheim GmbH). This technique is described in detail in the operation sheet inserts with each kit. The method has been shown to compare well with the reference ultracentrifugation method. It uses a precipitant solution with decreased phosphotungstic acid and magnesium chloride concentrations; only 200 µl of sample are required for one determination on a sample using the semimicro version of the method. Samples become more diluted than in the conventional PT method(s); thus, turbid supernatants occur only very exceptionally. In short, the procedure is as follows:
500 µl (or 200 µl in the semi-microversion) of sample are mixed with 1000 µl (500 µl) of the reagent. In the semimicro version the precipitant solution must be diluted with water (4 + 1 volumes) to obtain the precipitation reagent. After mixing, the mixture is left to stand 10 minutes at room temperature and then centrifuged for 10 minutes at 4000 r.p.m. or 2 minutes at 12 000 r.p.m. Cholesterol is determined in the supernatant using the enzymatic method (e.g. Monotest Cholesterol or Peridochrom, Boehringer Mannheim GmbH). 200 µl of sample (or blank, or standard) supernatant are mixed with 2000 µl of the Monotest or Peridochrom reagent in the conventional manual method (one can also use proportionally less of sample and reagent in the automated methods, or where allowed by the cuvette volume of the photometric instrument). After mixing and 10 minute incubation at 20°C - 25°C (or 5 minutes at 37°C) the developed colour is measured at Hg 546nm or at 500nm, using a 1 cm cuvette. Concentration is calculated on the basis of "water soluble" standard(s) (e.g. Preciset Cholesterol; Boehringer Mannheim GmbH), or using the manufacturer's factor if the kit procedure is followed exactly as described in the insert, of a factor derived from use of a certified serum calibrator.
The Precipitant solution used in this method contains phosphotungstic acid (0.55 mmol/l) and magnesium chloride (25 mmol/l) in distilled water. For use in the semimicro version of the method (200 µl sample + 500 µl reagent), the Precipitant should be diluted as mentioned above, i.e. 4 + 1 (4 volumes Precipitant + 1 volume distilled water). This method yields good results not only on fresh serum or plasma material, but also on many non-outdated lyophilized control materials. Its use makes control of HDL-cholesterol methods by means of lyophilized control sera easier [35].
4.6 Determination of cholesterol in the HDL containing supernatant:
4.6.1 The determination should preferably be made shortly after precipitation of LDL+VLDL and the centrifugation step (on the same day). If it is necessary to store samples, analysis is recommended within 4 days if stored at +4°C. Prolonged storage requires at least -20°C. Storage is recommended in small volume glass tubes (vials) with leak-proof stoppers to prevent volume and concentration changes (evaporation, freezing out). After storage and before analyses the samples should be brought to room temperature and gently mixed.
4.6.2 Cholesterol should be determined in the HDL-containing supernatants by the same method as used in the laboratory concerned with the TC estimation. Should the methods sensitivity be low (optical density readings in unreliably low range) it might be necessary to use a photometric cuvette with greater optical length and/or to increase the supernatant sample volume. Reliability of the method must be checked carefully in such cases.
4.6.3 When serum/plasma sample is diluted 1:1 before use (see section 4.5.6) again the results must be multiplied by two to obtain the final serum/plasma HDL-C concentration.
4.6.4 Duplicate or single TC and HDL-C measurements: If possible, all MONICA serum/plasma samples should be analysed in duplicate. For the analysis of cholesterol by extraction chemical methods, two separate extracts will be required for duplication (do not perform two analyses on one extract). For HDL-C analysis duplicate LDL+VLDL precipitations will be required. Mean results are reported if analyses are performed in duplicate. (The MDC must be notified as to whether single analysis or mean values are reported, the "power" of the means is greater in statistical evaluations.)
4.6.5 For IQC and EQA samples all analyses are performed in duplicate (but mean values are not reported to the MQC, both values are recorded and reported).
4.6.6 The IQC rules shown below (see section 7.7) give criteria for each single measurement result on control samples, from which the maximal allowable difference between side-by-side duplicates is in fact also defined. Daily use of IQC information is of prime importance for maintaining good precision, since EQA information is only retrospective and primarily devoted to following interlaboratory comparability.
It is recommended in addition that laboratories analyzing not only control, but also MONICA Project samples in duplicate, should reanalyse the samples with duplicates differing more than 0.40 mmol/l for TC (15.0 mg/dl), or 0.20 mmol/l (8.0 mg/dl) for HDL-C. Should duplicate results frequently differ that much, the method should be revised and every effort be made to improve precision.
Recommendation: Primary standards and/or secondary serum/plasma calibrators are used in duplicate at least for calibration.
5.1.0 Each participating laboratory is responsible for its own analytical primary standards and/or secondary serum calibrators.
5.1.1 It is assumed that all participating laboratories will use the purest and most appropriate substances and method reagents. The cholesterol substance used for preparation of standards should be of more than 99% purity.
5.1.2 Secondary (serum) calibrators should preferably be prepared and/or labelled with correct TC and/or HDL-C concentrations in the WHO-RLRC and/or CDC. It is understood, however, that staffing and budgetary restraints may be a limiting factor in such an undertaking.
5.1.3 The WHO-RLRC distributed during the pre-standardization period a set of three cholesterol standards for use in testing linearity over the working range. There was a shipment of a commercial cholesterol standards pack (6 standards with concentrations of 50-400 mg/dl/1.3-10.4 mmol/l) to each laboratory. Such sets should enable control of linearity and at the same time calibration of several enzymatic and/or extraction cholesterol methods. Some cholesterol methods (the so-called "direct" Liebermann-Burchard chemical methods) cannot be calibrated by these "water soluble" standards (falsely high results may be obtained without special arrangements), but the linearity response of these methods can be judged on the basis of these standards.
5.1.4 Each standard (calibrator) should be run at least in duplicate.
5.1.5 Each laboratory should perform tests on linearity over the usual working range (0.5 - 10.0 mmol/l of TC), even if the WHO-RLRC or other body is not able to supply them free with appropriate control material for this purpose.
5.1.6 It is essential that linearity should be checked repeatedly during the prestandardization period. During the study linearity should be checked regularly, and particularly if greater than allowable imprecision and/or inaccuracy (allowable limits on accuracy and precision - see 8.3.8) is detected. Ideally, linearity should be checked with at least three standards in each run in which MONICA samples are analysed. It is understood, however, that this will not always be possible in all laboratories and that it may not be necessary to use standards to judge linearity where adequate data can be obtained from IQC and EQA samples (see 7.3.2). Laboratories are free to check on linearity by any reasonable method guaranteeing acceptable results within the limits of accuracy and precision.
It is understood that linearity enabling the use of a mean calculation factor (which requires calibration line passing through the X,Y intercept after subtraction of the reagent blank) may not be attainable in some participating laboratories (e.g. because of limitations in the photometric equipment available for the assay). Non-linearity need not necessarily mean wrong results. Linearity must be checked, and every effort should be made to attain it. If unattainable the laboratory should derive results mathematically from regression equations or graphically from standard curves covering the working range of the method. (Enzymatic cholesterol methods can usually provide linearity up to 11.5 mmol/l. When a higher serum/plasma cholesterol concentration is expected, analysis should be repeated after appropriate dilution of the sample according to the instructions of the manufacturer of the enzymatic kit).
5.1.7 More detailed information on checking on linearity, calibration, and QA can be found in the literature [4, 14, 19, 20].
Recommendation: All results should ideally be reported to 2 decimals in the International SI Units [23].
However, the older units (e.g. mg/dl) are still used in some laboratories because continuity of methods is desirable until certain related projects are finished. In the MONICA Project results for TC and HDL-C will therefore be accepted both in mmol/l and mg/dl since units can easily be converted by computer.
6.1 Results obtained in mmol/l should be calculated and reported to two decimals.
6.2 Results obtained in mg/dl should be calculated to one decimal, but TC concentrations should then be reported, rounded off to whole numbers and HDL-C concentrations as calculated, i.e. to one decimal.
6.3 The conversion factor for cholesterol from mg/dl to mmol/l is: 0.025864. The conversion factor for triglyceride from mg/dl to mmol/l is: 0.0114.
Example: (250mg/dl) x (0.025864) = 6.466 = 6.47 mmol/l for TC.
(250mg/dl) x (0.0114) = 2.85 mmol/l for TG.
Recommendation: At least two serum control pools are used for IQC. The preparation, checking, and use of these pools are described under 7.3 and 7.7.
7.1 Performance of methods within IQC limits is a prerequisite for successful participation in the EQA programme and for reliable analysis of MONICA samples.
7.2 The following recommendations on the preparation of IQC pools and how to proceed with IQC represent the views of the WHO-LW on the desirable methods of quality control. It is understood however that some laboratories will not be in a position to comply with all of these recommendations, and others will already have instituted adequate but different quality control procedures, which they are unable to change. In these cases each participating laboratory should inform the WHO-RLRC centre of their quality control and standardization methods. In other words, systems other than the recommended internal quality control system are permissible as long as there is adequate indication that analytical results, as judged by IQC and EQA, are within the IQC and EQA control limits described below. It follows that use of self-made and/or commercial frozen or lyophilized internal quality control materials is allowed. If lyophilized material is used for HDL-C, it should be material specifically designed for this purpose.
7.3 Establishment of pools: Each laboratory should establish at least two control serum pools for internal quality control. Each pool should last through the whole of one survey phase of MONICA and in any case should be sufficient to last for one year of normal operations. The pools under 7.3.1 and 7.3.2 are the only two pools that are strictly required for MONICA.
7.3.1 TC + HDL-C pool: This pool should be prepared from non-turbid human serum containing "normal" TC and TG concentrations. The pool should be distributed in suitable aliquots into tightly closed glass vials and ideally kept frozen at -60°C or below1. Should it prove impossible to use a self-made or commercial frozen serum pool, a lyophilized pool could be used. In the latter cases, there should be a guarantee that the pool was prepared only from human sera, without use of additives or enrichments which could affect TC and HDL-C determinations, and it should be suitable for the HDL-C determination. The TC + HDL-C pool would be used both for TC and HDL-C method control, including the precipitation step in the latter case (the quality control specimen should be treated in exactly the same way as the test samples). Lyophilized pools do not require -60°C for storage. Even +4°C to +6°C may be satisfactory for a defined expiry period.
Note1: CDC laboratories found -20°C unsatisfactory for long term storage (personal communication). However, a number of MONICA laboratories may only have storage at -20°C. In such cases the TC + HDL-C pool should be periodically checked for HDL-C concentration stability in runs containing EQA samples.
7.3.2 Low total cholesterol pool (LTC pool): This pool should contain 1.30-1.60 mmol/l cholesterol and it should be used for control in the low TC range. It could be prepared from a human serum pool diluted to the appropriate cholesterol concentration with 0.15m NaCL. Alternatively, animal (bovine, horse) serum pool, eventually slightly diluted to the desired CH concentration level, could be used.
7.3.2.1 The appropriately diluted TC + HDL-C pool (under 7.3.1) could be used as a LTC pool with the advantage that both pools could also be used for linearity checking (plot of absorbance against concentrations or dilution factors should result in a line passing through the X, Y intercept). This checking will work only with enzymatic and/or extraction cholesterol methods. In the direct chemical cholesterol methods the concomitant dilution of interference (by bilirubin, tryptophane, proteins, etc.), as well as changed viscosity in automated versions, may result in non-proportionality of absorbances to dilution.
7.4 Some laboratories may elect to establish a further (or several more) pool(s) for TC and/or HDL-C measurements, e.g. one with the cholesterol concentration in the upper part of the working range. The HDL-C pool(s) should have a low TG concentration (under 1.5 mmol/l).
7.5 Useful information concerning the preparation of self-made pools can be found in the literature [13, 14, 16, 17, 18, 22, 26, 27, 28, 29].
7.6 Use of IQC material to ensure analytical precision for TC and HDL-C analysis:
7.6.1 Once opened, an ampoule (vial) of quality control materials should not be used for more than one day's operations. Accurate reconstitution of lyophilized materials must be ensured (to prevent introduction of errors unrelated to the method itself). If a lyophilized pool is used for HDL-C control (precipitation included) each laboratory concerned should standardize the time period between reconstitution and analysis.
7.6.2 Each analytical run should begin with calibration standard(s). This could be the primary standard(s) and/or secondary serum calibrator(s) (see example in Appendix I).
7.6.3 IQC samples should follow. All quality control samples should be analysed in side-by-side duplicates. The results obtained from quality control samples following the standard(s) are used to indicate whether the run is in control and whether the analysis of study samples can be begun (for criteria see 7.7).
7.6.4 When a new pool is introduced both the next and the old pool should be in use together (overlap period) for at least the number of runs (analyses) required to establish the starting mean (see 7.6.6), to ensure adequate calibration of the new pool and method control.
7.6.5 Control of drift in a run:
In each analytical run (or part of analytical run) containing MONICA samples, quality control material should be placed in every tenth position among the samples (to monitor within-run stability). This can be other than the LTC+HDL-C or LTC pool material and need not necessarily be of human origin nor accurately calibrated because it serves only for monitoring within-run reproducibility (see Figure 1).
7.6.6 Before a pool is introduced, its starting concentration mean
must be established. This is done as follows: at least 40 values are collected on at least
20 pool vials randomly distributed in independent analytical runs over at least 20 days (do
not use 40 times in one run). Finally, the whole set of values should be examined
for potential outliers [19]. If any value differs from the overall
mean (
)
by a number greater than 3.3 St (i.e. if any Xi > + 3.3 St),
the pair of results in which it occurs is discarded and recalculated. Only one such
outlier is permissible. Should there be more outliers, the method should be revised and
complemented with further runs and checks on outliers as above.
(St is calculated only for the purpose of checking for outliers; it is not needed for IQC in MONICA, common control limits are stated below.)
Total (overall) standard deviation (St) is calculated on the basis of individual repeated measurements as follows:
where Xi = individual measurements results
= the
mean of all measurements
n = the number of measurements
= the sum of
the measurements results.
Relative standard deviation (coefficient of variation-C.V.) is calculated:
For an example of mean and St calculation see Appendix 1.
7.7 IQC criteria for acceptability of the run: Once the acceptable
pool mean () is obtained, the precision control limits are defined as follows (for an
example of the quality control chart for visual follow-up of method performance see Figure
2):
Warning limits for TC (levels above 2.60 mmol/l or 100 mg/dl)
±3% from the mean (±0.03).
Example: at a 5.17 mmol/l (200 mg/dl) level:
Maximum allowable limits for TC measurements (levels above 2.6 mmol/l
or 100 mg/dl):
±4.5% from the mean (±0.045)
Example: at a 5.17 mmol/l (200 mg/dl) level:
Warning limits for HDL-C: ±6% from the mean (±0.06)
Maximum allowable limits for HDL-C: ±9% from the mean (±0.09).
7.7.1 Ideally, in runs containing MONICA samples, no single quality control pool value (not the mean of a duplicate) should surpass the warning limits. Should this occur, and the result be still within maximum allowable limits, the run result can be used but it is a strong signal that method(s) should be checked (see standards, calibrator(s), reagents, OD reading instrument(s), pipettes, dispensers, quality control pool expiry dates (see 7.3), eventual microbial growth, possible calculation errors, etc.). The best quality volumetric glass, dispensers, and measurement instruments are recommended.
7.7.2 In no case should single control value(s) fall beyond the maximum allowable limits. If this occurs, the run must be stopped, results cannot be used, the cause of trouble must be eliminated, and the method must be brought under control before MONICA samples are started.
Recommendation: the WHO-RLRC is responsible for sending the EQA samples.
8.1 EQA is complementary to the IQC and its main purpose will be to check on accuracy (assessment of bias), although it will also supply laboratories with evaluations concerning overall, between-day (between-run) and within-run variability. (Regular use of IQC is a prerequisite for successful participation in the EQA programme.)
8.2 EQA will be carried out from the MQC at the WHO-RLRC (Chief: Dr R. Poledne (former chief: Dr D. Grafnetter), Institute for Clinical and Experimental Medicine (IKEM), Videnska 800, PO Box 10, 140 00 Prague 4, Czech Republic).
8.3 EQA assistance to participating laboratories:
8.3.1 1982 was considered as a pre-checking (pre-calibration, pre- standardization) period. Each laboratory assigned to analyse MONICA samples (all addresses should be made available to the WHO-RLRC by WHO/HQ, see also 4.4) were expected before the start of the proper EQA to analyse with "blind" samples (concentration unknown to the laboratories) in one or two self-evaluation sets (concentrations known) provided by the WHO-RLRC. Self-evaluation sets contain vials with control sera selected from lyophilized pools. They are provided, if needed, with instructions for use and with the list of reference TC, HDL-C, TG (and possibly also Apoprotein AI and B) concentrations. Self-evaluation sets enable a check on methods and/or their calibration.
After the "open" period, a "blind" EQA system was started to evaluate laboratory performances. Its principles and methods were explained in detail to the laboratory directors who attended the WHO-LW. Others who are interested or who join the Project later may obtain information from the literature [6, 7, 8, 9]. It is only necessary to mention here that the system has been based on repeated analysis of control sera shipped at regular intervals as lyophilizate to laboratories ("sets" composed of 14-21 samples). This system of "blind" EQA sets will be used throughout MONICA.
8.3.2 Sets are sent periodically to participating laboratories with clear instructions for use, as well as with relevant information, reporting forms, and eventually also questionnaires. Samples contained in the shipments may originate from commercial pools, as well as from pools prepared in the WHO-RLRC or elsewhere. It is the responsibility of the WHO-RLRC to check, in cooperation with CDC or other bodies, the suitability of pools.
8.3.3 Control samples should be used in the laboratories according to the accompanying instructions (reconstitution, sequence of samples, etc.). Each sample should be analysed in side-by-side duplicates on the day of reconstitution. If more measurements are performed, only the first two should be reported to the WHO-RLRC. The rest may serve for internal purposes, e.g. analysis of drift.
8.3.4 The decision on how many samples (or sets) should be distributed to individual laboratories should be made on an individual basis, depending on the questionnaires completed and results reported by the laboratories. In most cases 4-5 EQA sets per year should be distributed to each laboratory during the respective survey period. Each set should contain about 20 samples, sufficient for about 2 months.
8.3.5 Some laboratories may wish to analyse MONICA survey samples (each survey covers approximately 1600 samples) in batches separated by intervals of days or weeks, while others may spread their analyses over year(s). Frequency of EQA sets should be adjusted to their needs. Only sets falling into the survey periods of sample analyses (not collection and/or storage) will be used for evaluation of MONICA results.
Laboratories will be requested to complete analysis of individual sets and report results within 2 months, at the latest . Failure to do so may mean that a further set will not be received, since only the receipt of results will be a signal to the WHO-RLRC to indicate that a further set should be sent. Failure of a laboratory to undertake sufficient EQA control may result in elimination of its TC and HDL-C results from MONICA evaluations.
8.3.6 Based upon the EQA results received, the WHO-RLRC will as soon as possible supply the participating laboratory (and WHO-HQ if needed) with information on the analysis of the EQA set, i.e. on the calculated overall, between-run and within-run standard deviation, and on bias, from which acceptability of results for the different pools and methods can be judged. Performance acceptability criteria are shown in Table 1 and discussed under 8.3.8.
8.3.7. Reference values (RV) will be obtained in the WHO-RLRC as before by the modified Abell Kendall method [18], in cooperation with CDC. CDC will be requested to analyse the EQA pools and make results available to the WHO-RLRC, to ensure comparability between WHO-CDC and WHO-RLRC evaluations. As a secondary check in obtaining the RVs a CHOD-PAP enzymatic cholesterol method with practically 100% cholesterol-ester hydrolysis may be used by WHO-RLRC. For LDL+VLDL precipitation in the HDL-C reference value estimation the PT reagents and procedures described in sections 4.5 and 4.6 will be used.
8.3.8 Performance acceptability criteria in EQA (limits on maximum allowable inaccuracy and imprecision) (see Table 1):
Total cholesterol: for any of the control pools, calculated bias (based on the EQA set mean and on RV) should not be greater than +5%. At the same time st should be smaller than that shown for individual reference values in Table 1. Table 2 shows allowable limits for CV% according to increasing levels of cholesterol concentration.
HDL-cholesterol: for any control pool, calculated bias (based on the EQA set mean and RV) should not be greater than +7.5%. At the same time St should be smaller than that shown for individual reference values in Table 2 (i.e. CV<6.5%).
| Table
1. Performance acceptability criteria in EQA (limits on maximum allowable
inaccuracy and imprecision). Individual standard deviations can be obtained for any RV concentration by interpolation. |
|||||
| Reference value(RV) | Maximum allowable overall standard deviation (St) |
Maximum allowable deviation of set pool mean from RV (bias)* |
|||
| mmol/l | mg/dl | mmol/l | mg/dl | ||
| TOTAL | 2.5864 | 100 | 0.1681 | 6.5 | ±5% |
| CHOLESTEROL | 5.1728 | 200 | 0.1940 | 7.5 | ±5% |
| 7.7592 | 300 | 0.2198 | 8.5 | ±5% | |
| 10.3456 | 400 | 0.2457 | 9.5 | ±5% | |
| HDL- | 0.5173 | 20 | 0.0336 | 1.3 | ±7.5% |
| CHOLESTEROL | 1.0346 | 40 | 0.0672 | 2.6 | ±7.5% |
| 1.5518 | 60 | 0.1009 | 3.9 | ±7.5% | |
| 2.0691 | 80 | 0.1345 | 5.2 | ±7.5% | |
*Calculated by comparing the average of all determinations for a given pool in a set to the reference value for that pool, i.e. with 0.05×RV (total cholesterol) or 0.075×RV (HDL-cholesterol)
| Table 2. MONICA - Total Cholesterol Limits on maximal allowable imprecision | ||||||
| Reference value mmol/l |
Overall standard deviation mmol/l |
Coefficient of variation % |
Reference value mmol/l |
Overall standard deviation mmol/l |
Coefficient of variation % |
|
| 2.6 | 0.1682 | 6.5 | 6.4 | 0.2062 | 3.2 | |
| 2.7 | 0.1692 | 6.3 | 6.5 | 0.2072 | 3.2 | |
| 2.8 | 0.1702 | 6.1 | 6.6 | 0.2082 | 3.2 | |
| 2.9 | 0.1712 | 5.9 | 6.7 | 0.2092 | 3.1 | |
| 3.0 | 0.1722 | 5.7 | 6.8 | 0.2102 | 3.1 | |
| 3.1 | 0.1732 | 5.6 | 6.9 | 0.2112 | 3.1 | |
| 3.2 | 0.1742 | 5.4 | 7.0 | 0.2122 | 3.0 | |
| 3.3 | 0.1752 | 5.3 | 7.1 | 0.2132 | 3.0 | |
| 3.4 | 0.1762 | 5.2 | 7.2 | 0.2142 | 3.0 | |
| 3.5 | 0.1772 | 5.1 | 7.3 | 0.2152 | 2.9 | |
| 3.6 | 0.1782 | 5.0 | 7.4 | 0.2162 | 2.9 | |
| 3.7 | 0.1792 | 4.8 | 7.5 | 0.2172 | 2.9 | |
| 3.8 | 0.1802 | 4.7 | 7.6 | 0.2182 | 2.9 | |
| 3.9 | 0.1812 | 4.6 | 7.7 | 0.2192 | 2.8 | |
| 4.0 | 0.1822 | 4.6 | 7.8 | 0.2202 | 2.8 | |
| 4.1 | 0.1832 | 4.5 | 7.9 | 0.2212 | 2.8 | |
| 4.2 | 0.1842 | 4.4 | 8.0 | 0.2222 | 2.8 | |
| 4.3 | 0.1852 | 4.3 | 8.1 | 0.2232 | 2.8 | |
| 4.4 | 0.1862 | 4.2 | 8.2 | 0.2242 | 2.7 | |
| 4.5 | 0.1872 | 4.2 | 8.3 | 0.2252 | 2.7 | |
| 4.6 | 0.1882 | 4.1 | 8.4 | 0.2262 | 2.7 | |
| 4.7 | 0.1892 | 4.0 | 8.5 | 0.2272 | 2.7 | |
| 4.8 | 0.1902 | 4.0 | 8.6 | 0.2282 | 2.7 | |
| 4.9 | 0.1912 | 3.9 | 8.7 | 0.2292 | 2.6 | |
| 5.0 | 0.1922 | 3.8 | 8.8 | 0.2302 | 2.6 | |
| 5.1 | 0.1932 | 3.8 | 8.9 | 0.2312 | 2.6 | |
| 5.2 | 0.1942 | 3.7 | 9.0 | 0.2322 | 2.6 | |
| 5.3 | 0.1952 | 3.7 | 9.1 | 0.2332 | 2.6 | |
| 5.4 | 0.1962 | 3.6 | 9.2 | 0.2342 | 2.5 | |
| 5.5 | 0.1972 | 3.6 | 9.3 | 0.2352 | 2.5 | |
| 5.6 | 0.1982 | 3.5 | 9.4 | 0.2362 | 2.5 | |
| 5.7 | 0.1992 | 3.5 | 9.5 | 0.2372 | 2.5 | |
| 5.8 | 0.2002 | 3.5 | 9.6 | 0.2382 | 2.5 | |
| 5.9 | 0.2012 | 3.4 | 9.7 | 0.2392 | 2.5 | |
| 6.0 | 0.2022 | 3.4 | 9.8 | 0.2402 | 2.5 | |
| 6.1 | 0.2032 | 3.3 | 9.9 | 0.2412 | 2.4 | |
| 6.2 | 0.2042 | 3.3 | 10.0 | 0.2422 | 2.4 | |
| 6.3 | 0.2052 | 3.3 | ||||
8.4 Laboratories should not begin analysis of MONICA samples before they are able to show that their analyses of at least two successive blind EQA sets are within the limits of acceptability (see 8.3.8) for every control pool. In order to allow time for all this, participating laboratories should begin standardization at least six months before MONICA population survey work begins, and this should be repeated before each further MONICA population survey. In order to detect trends in serum cholesterol level, the laboratory must be well standardized on all occasions.
8.4.1 If EQA results suggest trouble with a method a decision will have to be made on the admissibility of data from the laboratory. It is strongly recommended that assistance be provided to help solve such problems with the laboratory concerned, using QA information, including IQC data. Laboratories should keep good records on their IQC data and be able to present them upon request.
8.4.2 A situation might arise, however, when a laboratory demonstrates greater than permissible but fairly stable analytical bias with acceptable precision (small st). In such a case the WHO MONICA appointed adviser, who receives evaluation copies from WHO-RLRC through WHO-HQ, should consider the case and advise the laboratory and WHO-RLRC immediately on the course to be followed. (Results might perhaps be corrected mathematically provided that the bias was practically the same at different concentration levels in successive tests, and that st did not surpass allowable limits). Once st is above the acceptable limit no allowance for bias consideration can be made, and the laboratory will be requested to revise and correct the method before proceeding with the MONICA study samples.
8.5 To help overcome problems of retrospectivity in EQA, the WHO-RLRC will include in sets, if possible, a pool with nominated RVs to be made known to participating laboratories for continual self-evaluation.
9.1 MONICA results reported to WHO-HQ and/or an outside adviser elsewhere will be analysed under the auspices of WHO. WHO staff will design forms for data reporting suitable for computer analysis.
9.2 Laboratories should at all times indicate whether they report single measurements results or means.
9.3 WHO-RLRC will provide laboratories with reporting forms prepared for EQA sets.
9.3.1 Laboratories will be identified in the EQA system by code numbers which will not be made publicly available without written consent from the respective laboratories. However, the code will be open for WHO-HQ and for MONICA appointed advisers (who are also bound to keep it confidential) and WHO MONICA meetings.
Initial action should be to make every effort to bring the method(s) into the required performance limits, with the use of information obtained and with special samples for method calibration. WHO-RLRC or CDC should be advised about the problem.
If all efforts fail, WHO may assist in nominating and supporting a visit of a consultant to the laboratory concerned.
Laboratories are advised to store replicate samples under good conditions. Should quality control results be poor and the fault cannot be found, these replicates can then be processed by a well-standardized laboratory elsewhere. If the local laboratory's problems are solved, it can process the replicates itself.