- Email: [email protected]
Biological variability of lymphocyte subsets of human adults' blood
Clinica Chimica Acta 424 (2013) 159–163
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Biological variability of lymphocyte subsets of human adults' blood Francesca Tosato ⁎, Daniela Bernardi, Maria Colomba Sanzari, Giorgia Pantano, Mario Plebani Department of Laboratory Medicine, University Hospital of Padova, Via Giustiniani 2, 35128 Padova (PD), Italy
a r t i c l e
i n f o
Article history: Received 20 March 2013 Received in revised form 4 June 2013 Accepted 4 June 2013 Available online 11 June 2013 Keywords: Biological variability Lymphocyte subsets Quality specifications
a b s t r a c t Background: Alterations in lymphocyte subpopulations are present in several immune diseases, and clinicians and researchers recognise the importance of investigating the distribution and changes in lymphocyte subsets over relatively long periods of time in order to evaluate the effectiveness of treatment and follow the course of disease. Yet further insight is required on the biological variability (BV) of lymphocyte subsets, which is crucial to the correct interpretation of longitudinal changes and provides essential information for setting desirable quality specifications and defining the usefulness of reference values. Methods: Four-colour-flow cytometry was used to investigate the BV of lymphocyte populations (LP) in the peripheral blood of 20 healthy adults recruited from our laboratory staff and followed for three months. The total lymphocyte count was measured, and the relative frequencies determined for T-cells (CD3 +), T-helper cells (CD3 + CD4 +), cytolytic T-cells (CD3 + CD8 +), B-cells (CD3 − CD19 +), NK-cells (CD3 − CD16 +/56 +), non-MHC restricted cytolytic T-cells (CD3 + CD56 +) and activated T-cells (CD3 + HLA-DR +). Results and conclusions: Data on the components of BV were applied to set quality specifications for allowable precision, bias and total error. Analytical performances were established, and they were more than desirable for all the markers considered in our study. By comparing within-subject and between-subjects BV, we were able to define the uselessness of reference ranges in the evaluation of changes in CD serial results. Data on within-subject BV and analytical precision were thus used to determine the reference change values, in order to identify the significance of changes in serial results. The findings made in the present study provide further evidence of the relevance of BV in the evaluation of immunological markers of LP. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Lymphocytes have both regulatory and effector roles and the distribution of the different populations in peripheral blood can be altered in many primary immune diseases or diseases with secondary involvement of the immune system. To follow the course of disease and evaluate the effectiveness of treatment, the distribution of, and changes in, the lymphocyte populations (LP) are often monitored for long periods, both for clinical and research purposes. In order to accurately estimate longitudinal changes it is important to take into consideration biological variability (BV) of the measurement (i.e. random fluctuation around a homeostatic setting point, the so-called within-subject or intra-individual BV). Pure estimates of the average within-subject BV and the between-subjects BV are obtained by carefully controlling pre-analytical variability and by designing experiments that allow the quantification of analytical variability [1].
⁎ Corresponding author. Tel.: +39 0498211313; fax: +39 0498211915. E-mail address: [email protected] (F. Tosato). 0009-8981/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cca.2013.06.001
Pre-analytical variability should be minimised as far as possible. Therefore candidates for study must be: willing to provide a number of samples over a long period of time, healthy, and not on any drugs that might affect the analytes under investigation; nor should they have unusual lifestyles or habits, or have an alcohol intake in excess of the units specified [1]. BV can be used to: set quality specifications; consider the utility of conventional population-based reference values; determine the change occurring in an individual's serial results before it becomes significant (reference change values (RCV)) [1]. A number of early studies available in literature evaluating LP BV reported variations occurring during a day [2–6] and/or during a year [7,8,3,9–11]. More recent studies demonstrate that the circadian variability of lymphocyte subsets is very relevant, and correlates with the neuro-endocrine system and changes with aging [12–14]. However, in spite of its well-defined clinical role, no reliable data on the BV occurring in lymphocyte subsets have yet appeared in literature.The present study therefore reports findings obtained by making a definitive assessment of the analytical and biological components of variability of lymphocyte subsets using an accurately designed experimental and statistical protocol.
160
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
2. Materials and methods 2.1. Subjects A valid estimate of the components of variability is obtained from relatively small numbers of specimens collected from a small group of subjects over a relatively short period of time [1]. Therefore, over a three-month time-period (February 2012–May 2012), four blood specimens were drawn from each of 20 healthy volunteers (11 women, 9 men; age range, 25–58 years) on the same day, at regular time intervals (start time, after 1 week, 1 month and 3 months). Apparently healthy subjects were studied to ensure that any LP fluctuation in peripheral whole blood would truly reflect biology and not modifications due to pathological processes. In accordance with the Helsinki II Declaration, the design and execution of the experiment were explained thoroughly to the subjects, and fully informed consent was obtained in writing. 2.2. Experimental design The group of subjects selected for studies evaluating BV should be considered “reference individuals”; in order to minimise pre-analytical variability, exclusion or inclusion criteria are established before the accrual of subjects [1]. Since its aim was to obtain data on BV, our experimental protocol minimised the numerous pre-analytical factors that can influence the results of the tests, such as lifestyle, time of sampling, phlebotomist, specimen handling and storage [1]. The subjects meeting inclusion criteria had no current illness or infection, were non-smokers, had not exercised, and had neither taken medication nor consumed substantial quantities of alcohol; none of the women were pregnant. During the study period, subjects were asked to continue with their dietary habits and usual activities. To minimise circadian variability in subjects, venous blood was obtained within the same time-period of the day (between 09.00 and 10.00). The same phlebotomist collected blood samples with minimal stasis into vacuum collection tubes with EDTA as the anticoagulant (Becton-Dickinson, Milan, Italy). The total lymphocyte count of all the subjects was within the reference range (1100–4800/μL). Samples were stored at room temperature until staining. 2.3. Methods The total lymphocyte count was measured with an ADVIA 2120 haematology analyser (Siemens). The following 4-colour combinations of monoclonal antibodies (Beckman-Coulter) were used: CD45-FITC/CD4-PE/CD8-ECD/CD3-PeCy5, CD45-FITC/CD56-PE/CD19-ECD/CD3-PeCy5, and CD16-FITC/HLA-DR-PE/ CD3-ECD/CD45-PeCy5. Sample preparation was performed according to the NCCLS (H42-A2) guidelines [15]. In particular, 100 μL of EDTA blood was added at analiquot of monoclonal antibody as suggested by the manufacturer, and incubated for 15 min at room temperature. After incubation, erythrocytes were lysed with TQ-prep (Beckman-Coulter). Samples were then acquired on the flow cytometer. Staining was always undertaken on the day of sampling, within 3 h. Four-colour-flow cytometric analysis was performed on a Navios (Beckman-Coulter). Acquisition was run until 25,000 events were detected. Data analyses were made with CXP software. Using the appropriate “gating” approach, the relative frequencies were determined for T-cells (CD3 +), T-helper cells (CD3 + CD4+), cytolytic T-cells (CD3 + CD8+), B-cells (CD3 − CD19 +), NK-cells (CD3 − CD16 +/56 +), non-MHC restricted cytolytic T-cells (CD3 + CD56 +) and activated T-cells (CD3 + HLA-DR+). Since we work with a dual platform, the absolute frequencies of the cells subsets
were calculated on the basis of the relative frequencies on total lymphocytes. Internal quality control (IQC), the so-called Immunotrol, performed daily, consisted of a liquid preparation of human stabilised red and white blood cells that was processed as a routine sample with the reagents in use. Before being included in routine use, the new batch of Immunotrol was analysed simultaneously with the old batch for at least three days, to determine whether any outside range results should have been attributed to unsuitability of the new batch or to other causes. The control values were stored in the instruments in a specific file of the program called Quality Control (QC). If the IQC was within the acceptable range, sample acquisition and analysis were performed. The instrument was calibrated, according to the operational manual, once a month and also whenever there were problems with the IQC or instrumental breakdowns occurred. If the calibration was not accepted, it was repeated using a new calibrator and, if it was still not acceptable, external technical assistance was sought. The calibrator was processed as indicated on the package insert, and the calibration was stored in the instruments. All the calibrators were conserved in cold room at radio controlled temperature. The UK Neqas External Quality Assessment “Leucocyte Immunophenotyping” external QC was used for assessing CD3, CD4, CD8 and CD19 values. 3. Statistical analysis All data were investigated for outliers. Cochran's Q test was performed for outlier identification among observations and withinsubject variations, whereas Reed's criterion was used to identify outliers among the mean values of subjects [1]. After outlier exclusion, the Kolmogorov-Smirnov test was applied separately to the set of results from each individual to check data distribution. The analytical, within-subject, and between-subjects components of variability were calculated by nested ANOVA from replicate analyses [1]. Within-subject BV was estimated from the within-subject total variability minus one-half of the analytical variability, and betweensubjects BV from the total variability of the data minus the analytical and intra-individual components. Precision was calculated by means of the IQC program of the laboratory. In order to assess the analytical uncertainty of laboratory results and provide a more correct clinical interpretation and utilisation of results, the analytical total error was calculated for each analyte as follows: bias + 1.65 CVa, where bias was the mean of bias (result − target value/target value*100) obtained at significant clinical concentrations, in an annual cycle, from reports of the External Quality Assessment Scheme; CVa represents the analytical variation coefficient (CVa%) obtained by determining internal control materials at clinically significant concentrations [1]. The index of individuality (II), a ratio calculated as (SDintra2/ SDbetween2)1/2, allowed the objective assessment of the value of population-based reference values. When the II is greater than 1.4, the conventional population-based reference values are relevant. When the ratio is low, particularly less than 0.6, conventional population-based reference values are of little relevance in the correct evaluation of results because individuals may have marked changes in marker levels that are significantly different from their own usual values and all results may still lie within the usual reference interval [16]. The index of heterogeneity (IH) was calculated as the ratio between the observed CV of the set of individuals (including analytical) to the theoretical CV, which is (2/k − 1)1/2, k being the number of specimens collected per subject. If the index differed from 1.0 by more than 2
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
Standard Deviation (SD) of the difference between this ratio and its expected value of unity, the heterogeneity was considered significant. The SD of the difference between this ratio and its expected value of unity (under the hypothesis of no heterogeneity of true withinsubject variations) is 1/(2 k)1/2. On assessing four specimens per subject, the IH values allowing us to define the homogenous withinsubject data would be IH b1.82. Data on the components of BV were used in assessing the significance of changes in serial results from an individual. The total variation required for a significant change to have occurred (RCV) was calculated by the addition of the intra-subject and analytical variability and then the SD was multiplied by 2.77 so as to attain 95% certainty that the two results were different. The value obtained was then transformed into a percentage of the overall mean. In the formula we used the analytical CV from the IQC of our laboratory at the appropriate clinical decision-making level [1].
161
Table 2 Optimal, desirable, and minimum analytical goals for imprecision, bias, and total error calculated in relation to biological variability. Quality specifications
CD3
CD4
CD8
CD19
CD56
CD16
Imprecision CV desirable CV minimum CV optimum Bias Bias desirable Bias minimum Bias optimum Total error TE desirable TE minimum TE optimum
0.007 0.020 0.030 0.010 0.005 0.027 0.040 0.013 0.016 0.060 0.090 0.030
0.014 0.040 0.060 0.020 0.026 0.047 0.070 0.023 0.049 0.113 0.169 0.056
0.028 0.025 0.038 0.013 0.057 0.066 0.099 0.033 0.010 0.107 0.161 0.054
0.040 0.055 0.083 0.028 0.010 0.078 0.116 0.039 0.057 0.168 0.252 0.084
0.077 0.125 0.188 0.063 0.013 0.138 0.206 0.069 0.114 0.344 0.516 0.172
0.077 0.115 0.173 0.058 0.013 0.126 0.190 0.063 0.114 0.316 0.474 0.158
4. Results
in order to identify the significance of changes in serial results. The IH and the RCV are shown in Table 3.
Within-subject BV and between-subjects BV expressed as CV using the overall means are shown in Table 1.
5. Discussion
4.1. Quality specifications Optimal, desirable, and minimum analytical goals for imprecision, bias, and total error calculated in relation to BV are shown in Table 2. 4.1.1. Imprecision All CD studied achieved the analytical goals, as shown in Table 2. The amount of variability added to the test result when CVA is different fractions of CVI was calculated (Fig. 1). Most of the tests studied were in the optimum performance range, the increase in result variability being 1% for CD3 and CD4, 4% for CD16 and CD56 and 6% for CD19 while the ratios of analytical imprecision to within-subject BV (SDA/SD2I ) were 0.17, 0.30 and 0.36, respectively. CD8 was within the desirable performance range, the increase in result variability being 14%; the ratio of analytical imprecision to within-subject BV was 0.56. 4.1.2. Bias and total error All CD studied achieved the analytical goal, being within the optimum performance range (Table 2). 4.2. Usefulness of reference values The values calculated for the II are shown in Table 3. All the indices were lower than 0.6, thus being of limited utility for conventional population-based reference intervals in the evaluation of changes in CD serial results. 4.3. RCV As the IH was not significant, all values being b 1.82 (1 + 2SD), within-subject variability was used to calculate the RCV. Therefore, the RCV, i.e. the minimal significant difference (p b 0.05) between two consecutive LP measurements in the same subject, were estimated Table 1 Within-subject and between-subjects biological variability expressed as CV using the overall means.
CVintra CVbetween
CD3
CD4
CD8
CD19
CD56
CD16
CD56 + 3+
3DR
0.04 0.10
0.08 0.17
0.05 0.26
0.11 0.29
0.25 0.49
0.23 0.45
0.38 0.62
0.23 0.56
Data on the components of BV were determined in order to set CD quality specifications, evaluate the utility of conventional population-based reference values, and to assess the significance of changes in serial results from an individual. Laboratory test results are used for many purposes. We use test results clinically for four different purposes: diagnosis, case finding, screening and monitoring. Quality specifications for precision and bias should ensure that all these objectives are achieved. If we develop separate quality specifications for precision and bias, then we can calculate specifications for total allowable error. In the present study, based on the biological data obtained, we established appropriate analytical performances, which were more than desirable for all the markers considered [1]. All the immunological markers fulfilled the homogeneity condition. Therefore the mean for within-subject variability observed for these indicators constitutes the basis for calculating the RCV appropriate for all patients and useful to other laboratories working with the same analytical imprecision. Moreover, since indices of individuality are calculated at less than 1, it appears difficult to identify an abnormal value in a subject if conventional population-based reference intervals are used; the RCV appears to be a better tool for identifying a true biological change in sequential measurements [17–19]. In longitudinal studies of LP, the variability between different sampling values depends on methodological and biological factors. Although it is of utmost clinical importance to evaluate differences within and between sampling values, little attention has been paid to this aspect in the literature and nor have efforts been made to take either methodological or BV into consideration. The literature contains large databases for the components of BV of many biological markers, but no studies reporting reliable data on the BV of lymphocyte subsets have yet appeared in the literature, the only exception being studies on circadian variability, an important pre-analytical factor that was minimised in our experimental protocol. In 1990 Malone et al. observed large fluctuations in repeated CD4+ cell counts in HIV + patients, and concluded that they could be explained in part by the CD4 + cell count diurnal cycle and in part by a high variability in total lymphocyte counts [20]. In a more recent study several methods were used over a long period of time in order to record monthly variations in CD4+ and CD8+ cells in HIV + patients [10]; in patients with a disease affecting the CD4/CD8 ratio a different pattern from that in healthy controls is to be expected. We analysed samples from healthy subjects who had total lymphocyte counts within the reference range, in order to obviate the effect of disease and total lymphocyte fluctuations on test results.
162
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
45
% increase in results variability
40 35 30 MINIMUM
25 20
CD8
15 DESIRABLE 10 CD19 CD16 e 56
5 OPTIMUM CD3 e 4
0 0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
analytical imprecision/within-subject biological variation Fig. 1. Test result variability (percentage) increases as a function of the ratio of analytical imprecision to within-subject biological variability. Most of the tests range around the optimum performance, the increase in result variability being 1% for CD3 and CD4, 4% for CD16 and CD56 and 6% for CD19 while the ratios between analytical imprecision and within-subject biological variability are 0.17, 0.30 and 0.36, respectively. CD8 ranges within the desirable performance, the increase in result variability and the ratio of analytical imprecision to within-subject biological variability being 14% and 0.56, respectively.
In studies evaluating LP BV, daily [2–6] and/or yearly [7–11] variations have been reported, but these findings are now several years old and, since they appeared in literature, major improvements have been made to immune phenotyping methods. Most of the available publications on seasonal variations of LP were the fruit of pioneering studies conducted in the wake of the introduction of monoclonal antibodies but before the widespread use of flow cytometry involving the use of microscopy, ficoll-separated mononuclear cells and antibody combinations [7,8,3,9]. Although monthly and seasonal fluctuations were noted in these studies, it is not surprising that patterns found were inconsistent. However, recent studies demonstrated that the circadian variability of lymphocyte subsets is highly relevant. A number of recent papers point to the important circadian rhythmicity of lymphocyte subpopulations, which seems to be related with neuro-endocrine system [12–14]. In particular, regarding LP, a clear circadian rhythm was validated for CD8 and CD16 with acrophase in the morning and at noon, and for CD3, CD4, CD4/CD8 ratio, HLA-DR and CD20 with acrophase at night [12].Therefore, in the present study we obtained samples at the same time of day in order to minimise the effect of circadian variability in subjects. Moreover, since age-related changes in 24-h rhythms have been found in older humans, in one study on circadian variability in lymphocyte subpopulations were also evaluated in subjects aged 67 to 79 years and compared with those in subjects 36 to 55 years. The authors concluded that aging is associated with enhanced T cell compartment responsiveness and alterations in circadian rhythmicity [13,14]. In order to obviate any age-related interference we recruited subjects in the age range of 25 to 58 years for the present study. Esquifino et al. observed in growing rats that the availability of nutrients may also affect the mechanisms modulating the circadian Table 3 Individuality and heterogeneity indices and reference change values.
Individuality index Heterogeneity index RCV
CD3
CD4
CD8
CD19
CD56
CD16
0.13 0.04 53%
0.20 0.09 67%
0.03 0.06 27%
0.14 0.13 26%
0.28 0.30 64%
0.21 0.29 65%
variability of immune response. On determining over a 24-h time span, the lymphocyte subset populations in the submaxillary lymph nodes of rats that had been submitted to a calorie restriction diet, the authors found that T and CD4+ cell number and CD4/CD8 ratio augmented, whereas B cell number decreased [21]. Therefore we asked the subjects in our study to continue with their dietary habits and avoid calorie restriction. Bearing in mind that other factors affecting LP are smoking habit [22], exercise training [23], menstrual cycle [24] and pregnancy [25], we excluded smokers, samples were drawn at least 12 h after exercise, no samples were collected from women at the same time in the menstrual cycle and pregnant women were excluded. In a longitudinal study [26] on the biological and methodological variability of lymphocyte subsets in the blood of human adults, the authors concluded that there were no major fluctuations of LP over a year, and that the variability noted in the longitudinal study included both biological and methodological variability, this being the most important contributory factor for variability in repeated measurements over time. Our study, which reports on a final assessment of the analytical and biological components of variability in lymphocyte subsets, used an accurately designed experimental and statistical protocol that enabled us to delete accurately the numerous pre-analytical confounding factors reported by other authors. 6. Conclusions The results obtained in our study provide further evidence of the relevance of BV in the evaluation of immunological markers of LP. As quality specifications based on BV are widely accepted criteria for ensuring valid clinical decision-making, they should be considered prerequisites for setting IQC and quality assurance goals. Quality specifications represent “the level of performance required to facilitate clinical decision-making” [1] and should be used for internal quality management procedures as well as for facilitating clinical reasoning, decision-making and patient management. Thus, quality specifications are prerequisites for implementing a valid quality management system.
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
The data obtained in the present study demonstrate that analytical performance goals based on BV were achieved for all CDs studied. Moreover, as the II for evaluated CDs does not indicate that the traditional comparison between patient results and reference values is useful, we have identified appropriate RCV for a valid interpretation of data, namely data obtained in patient monitoring through serial results.
References [1] Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409–37. [2] Ritchie AW, Oswald I, Mickelm HS, et al. Circadian variation of lymphocyte subpopulations: a study with monoclonal antibodies. Br Med J (Clin Res Ed) 1983;286:1773–5. [3] Lévi FA, Canon C, Tuoitou Y, Reinberg A, Mathé G. Seasonal modulation of the circadian time structure of circulation T and natural killer lymphocyte subsets from healthy subjects. J Clin Invest 1988;81:407–13. [4] Pasqualetti P, Colantonio D, Casale R, Colangeli S, Natali G. Circadian rhythm of human lymphocyte subpopulations. Quad Sclavo Diagn 1988;24:89–95. [5] Fukuda R, Ichikawa Y, Takaya M, Ogawa Y, Masumoto A. Circadian variations and prednisolone-induced alterations of circulating lymphocyte subsets in man. Intern Med 1994;33:733–8. [6] Mazzoccoli G, Bianco G, Correra M, et al. Circadian variation of lymphocyte subsets in healthy subjects. Recent Prog Med 1998;89:569–72. [7] Bratescu A, Teodorescu M. Circannual variations in the B cell/T cell ratio in normal human peripheral blood. J Allergy Clin Immunol 1981;68:273–80. [8] Abo T, Miller CA, Cloud GA, Blach CM. Annual stability in the levels of lymphocyte subpopulations identified by monoclonal antibodies in blood of healthy individuals. J Clin Immunol 1985;5:13–20. [9] Afoke AO, Eeg-Olofsson O, Hed J, Kjellman NI, Lindblom B, Ludvigsson J. Seasonal variation and sex differences of circulating macrophages, immunoglobulins and lymphocytes in healthy school children. Scand J Immunol 1993;37:209–15. [10] Paglieroni TG, Holland PV. Circannual variation in lymphocyte subsets, revisited. Transfusion 1994;34:512–6. [11] Termorshuizen F, Geskus RG, Roos MT, Coutinho RA, Van Loveren H. Seasonal influences on immunological parameters in HIV-infected homosexual men: searching for the immunomodulating effects of sunlight. Int J Hyg Environ Health 2002;205:379–84.
163
[12] Mazzocolli G, De Cata A, Greco A, Carughi S, Giuliani F, Tarquini R. Circadian rhythmicity of lymphocyte subpopulations and relationship with neuro-endocrine system. J Biol Regul Homeost Agents 2010;24:341–50. [13] Mazzocolli G, Vendemiale G, La Viola M, et al. Circadian variation of cortisol, melatonin and lymphocyte subpopulations in geriatric age. Int J Immunopathol Pharmacol 2010;23:289–96. [14] Mazzocolli G, De Cata A, Greco A, et al. Aging related changes of circadian rhythmicity of cytotoxic lymphocyte subpopulations. J Circadian Rhythms 2010;8:6. [15] NCCLS H42–A2 guidelines. Enumeration of immunologically defined cell populations by flow cytometry; approved guideline – second edition. Vol. 27 No. 16. 2007. [16] Harris KH. Effects of intra- and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974;20:1535–42. [17] Ross SM, Fraser CG. Biological variation of cardiac markers: analytical and clinical considerations. Ann Clin Biochem 1998;35:80–4. [18] Fraser CG, Lippi G, Plebani M. Reference change values may need some improvement but are invaluable tools in laboratory medicine. Clin Chem Lab Med 2012;50:963–4. [19] Plebani M, Lippi G. Biological variation and reference change values: an essential piece of the puzzle of laboratory testing. Clin Chem Lab Med 2012;50:189–90. [20] Malone JL, Simms TE, Gray GC, Wagner KF, Burge JR, Burke DS. Sources of variability in repeated T-helper lymphocyte counts from human immunodeficiency virus type 1-infected patients: total lymphocyte count fluctuations and diurnal cycle are important. J Acquir Immune Defic Syndr 1990;3: 144–51. [21] Esquifino AI, Chacon F, Cano P, Marcos A, Cutrera RA, Cardinali DP. Twentyfour-hour rhythms of mitogenic responses, lymphocyte subset populations and amino acid content in submaxillary lymph nodes of growing male rats subjected to calorie restriction. J Neuroimmunol 2004;156:66–73. [22] Schaberg T, Theilacker C, Nitschke OT, Lode H. Lymphocyte subsets in peripheral blood and smoking habits. Lung 1997;175:387–94. [23] Nehlsen-Cannarella SL, Nieman DC, Balk-Lamberton AJ, et al. The effects of moderate exercise training on immune response. Med Sci Sports Exerc 1991;23: 64–70. [24] Northern AL, Rutter SM, Peterson CM. Cyclic changes in the concentrations of peripheral blood immune cells during the normal menstrual cycle. Proc Soc Exp Biol Med 1994;207:81–8. [25] Matthiesen L, Berg G, Ernerudh J, Håkansson L. Lymphocyte subsets and mitogen stimulation of blood lymphocytes in normal pregnancy. Am J Reprod Immunol 1996;35:70–9. [26] Backteman K, Ernerudh J. Biological and methodological variation of lymphocyte subsets in blood of human adults. J Immunol Methods 2007;322:20–7.
Contents lists available at ScienceDirect
Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim
Biological variability of lymphocyte subsets of human adults' blood Francesca Tosato ⁎, Daniela Bernardi, Maria Colomba Sanzari, Giorgia Pantano, Mario Plebani Department of Laboratory Medicine, University Hospital of Padova, Via Giustiniani 2, 35128 Padova (PD), Italy
a r t i c l e
i n f o
Article history: Received 20 March 2013 Received in revised form 4 June 2013 Accepted 4 June 2013 Available online 11 June 2013 Keywords: Biological variability Lymphocyte subsets Quality specifications
a b s t r a c t Background: Alterations in lymphocyte subpopulations are present in several immune diseases, and clinicians and researchers recognise the importance of investigating the distribution and changes in lymphocyte subsets over relatively long periods of time in order to evaluate the effectiveness of treatment and follow the course of disease. Yet further insight is required on the biological variability (BV) of lymphocyte subsets, which is crucial to the correct interpretation of longitudinal changes and provides essential information for setting desirable quality specifications and defining the usefulness of reference values. Methods: Four-colour-flow cytometry was used to investigate the BV of lymphocyte populations (LP) in the peripheral blood of 20 healthy adults recruited from our laboratory staff and followed for three months. The total lymphocyte count was measured, and the relative frequencies determined for T-cells (CD3 +), T-helper cells (CD3 + CD4 +), cytolytic T-cells (CD3 + CD8 +), B-cells (CD3 − CD19 +), NK-cells (CD3 − CD16 +/56 +), non-MHC restricted cytolytic T-cells (CD3 + CD56 +) and activated T-cells (CD3 + HLA-DR +). Results and conclusions: Data on the components of BV were applied to set quality specifications for allowable precision, bias and total error. Analytical performances were established, and they were more than desirable for all the markers considered in our study. By comparing within-subject and between-subjects BV, we were able to define the uselessness of reference ranges in the evaluation of changes in CD serial results. Data on within-subject BV and analytical precision were thus used to determine the reference change values, in order to identify the significance of changes in serial results. The findings made in the present study provide further evidence of the relevance of BV in the evaluation of immunological markers of LP. © 2013 Elsevier B.V. All rights reserved.
1. Introduction Lymphocytes have both regulatory and effector roles and the distribution of the different populations in peripheral blood can be altered in many primary immune diseases or diseases with secondary involvement of the immune system. To follow the course of disease and evaluate the effectiveness of treatment, the distribution of, and changes in, the lymphocyte populations (LP) are often monitored for long periods, both for clinical and research purposes. In order to accurately estimate longitudinal changes it is important to take into consideration biological variability (BV) of the measurement (i.e. random fluctuation around a homeostatic setting point, the so-called within-subject or intra-individual BV). Pure estimates of the average within-subject BV and the between-subjects BV are obtained by carefully controlling pre-analytical variability and by designing experiments that allow the quantification of analytical variability [1].
⁎ Corresponding author. Tel.: +39 0498211313; fax: +39 0498211915. E-mail address: [email protected] (F. Tosato). 0009-8981/$ – see front matter © 2013 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cca.2013.06.001
Pre-analytical variability should be minimised as far as possible. Therefore candidates for study must be: willing to provide a number of samples over a long period of time, healthy, and not on any drugs that might affect the analytes under investigation; nor should they have unusual lifestyles or habits, or have an alcohol intake in excess of the units specified [1]. BV can be used to: set quality specifications; consider the utility of conventional population-based reference values; determine the change occurring in an individual's serial results before it becomes significant (reference change values (RCV)) [1]. A number of early studies available in literature evaluating LP BV reported variations occurring during a day [2–6] and/or during a year [7,8,3,9–11]. More recent studies demonstrate that the circadian variability of lymphocyte subsets is very relevant, and correlates with the neuro-endocrine system and changes with aging [12–14]. However, in spite of its well-defined clinical role, no reliable data on the BV occurring in lymphocyte subsets have yet appeared in literature.The present study therefore reports findings obtained by making a definitive assessment of the analytical and biological components of variability of lymphocyte subsets using an accurately designed experimental and statistical protocol.
160
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
2. Materials and methods 2.1. Subjects A valid estimate of the components of variability is obtained from relatively small numbers of specimens collected from a small group of subjects over a relatively short period of time [1]. Therefore, over a three-month time-period (February 2012–May 2012), four blood specimens were drawn from each of 20 healthy volunteers (11 women, 9 men; age range, 25–58 years) on the same day, at regular time intervals (start time, after 1 week, 1 month and 3 months). Apparently healthy subjects were studied to ensure that any LP fluctuation in peripheral whole blood would truly reflect biology and not modifications due to pathological processes. In accordance with the Helsinki II Declaration, the design and execution of the experiment were explained thoroughly to the subjects, and fully informed consent was obtained in writing. 2.2. Experimental design The group of subjects selected for studies evaluating BV should be considered “reference individuals”; in order to minimise pre-analytical variability, exclusion or inclusion criteria are established before the accrual of subjects [1]. Since its aim was to obtain data on BV, our experimental protocol minimised the numerous pre-analytical factors that can influence the results of the tests, such as lifestyle, time of sampling, phlebotomist, specimen handling and storage [1]. The subjects meeting inclusion criteria had no current illness or infection, were non-smokers, had not exercised, and had neither taken medication nor consumed substantial quantities of alcohol; none of the women were pregnant. During the study period, subjects were asked to continue with their dietary habits and usual activities. To minimise circadian variability in subjects, venous blood was obtained within the same time-period of the day (between 09.00 and 10.00). The same phlebotomist collected blood samples with minimal stasis into vacuum collection tubes with EDTA as the anticoagulant (Becton-Dickinson, Milan, Italy). The total lymphocyte count of all the subjects was within the reference range (1100–4800/μL). Samples were stored at room temperature until staining. 2.3. Methods The total lymphocyte count was measured with an ADVIA 2120 haematology analyser (Siemens). The following 4-colour combinations of monoclonal antibodies (Beckman-Coulter) were used: CD45-FITC/CD4-PE/CD8-ECD/CD3-PeCy5, CD45-FITC/CD56-PE/CD19-ECD/CD3-PeCy5, and CD16-FITC/HLA-DR-PE/ CD3-ECD/CD45-PeCy5. Sample preparation was performed according to the NCCLS (H42-A2) guidelines [15]. In particular, 100 μL of EDTA blood was added at analiquot of monoclonal antibody as suggested by the manufacturer, and incubated for 15 min at room temperature. After incubation, erythrocytes were lysed with TQ-prep (Beckman-Coulter). Samples were then acquired on the flow cytometer. Staining was always undertaken on the day of sampling, within 3 h. Four-colour-flow cytometric analysis was performed on a Navios (Beckman-Coulter). Acquisition was run until 25,000 events were detected. Data analyses were made with CXP software. Using the appropriate “gating” approach, the relative frequencies were determined for T-cells (CD3 +), T-helper cells (CD3 + CD4+), cytolytic T-cells (CD3 + CD8+), B-cells (CD3 − CD19 +), NK-cells (CD3 − CD16 +/56 +), non-MHC restricted cytolytic T-cells (CD3 + CD56 +) and activated T-cells (CD3 + HLA-DR+). Since we work with a dual platform, the absolute frequencies of the cells subsets
were calculated on the basis of the relative frequencies on total lymphocytes. Internal quality control (IQC), the so-called Immunotrol, performed daily, consisted of a liquid preparation of human stabilised red and white blood cells that was processed as a routine sample with the reagents in use. Before being included in routine use, the new batch of Immunotrol was analysed simultaneously with the old batch for at least three days, to determine whether any outside range results should have been attributed to unsuitability of the new batch or to other causes. The control values were stored in the instruments in a specific file of the program called Quality Control (QC). If the IQC was within the acceptable range, sample acquisition and analysis were performed. The instrument was calibrated, according to the operational manual, once a month and also whenever there were problems with the IQC or instrumental breakdowns occurred. If the calibration was not accepted, it was repeated using a new calibrator and, if it was still not acceptable, external technical assistance was sought. The calibrator was processed as indicated on the package insert, and the calibration was stored in the instruments. All the calibrators were conserved in cold room at radio controlled temperature. The UK Neqas External Quality Assessment “Leucocyte Immunophenotyping” external QC was used for assessing CD3, CD4, CD8 and CD19 values. 3. Statistical analysis All data were investigated for outliers. Cochran's Q test was performed for outlier identification among observations and withinsubject variations, whereas Reed's criterion was used to identify outliers among the mean values of subjects [1]. After outlier exclusion, the Kolmogorov-Smirnov test was applied separately to the set of results from each individual to check data distribution. The analytical, within-subject, and between-subjects components of variability were calculated by nested ANOVA from replicate analyses [1]. Within-subject BV was estimated from the within-subject total variability minus one-half of the analytical variability, and betweensubjects BV from the total variability of the data minus the analytical and intra-individual components. Precision was calculated by means of the IQC program of the laboratory. In order to assess the analytical uncertainty of laboratory results and provide a more correct clinical interpretation and utilisation of results, the analytical total error was calculated for each analyte as follows: bias + 1.65 CVa, where bias was the mean of bias (result − target value/target value*100) obtained at significant clinical concentrations, in an annual cycle, from reports of the External Quality Assessment Scheme; CVa represents the analytical variation coefficient (CVa%) obtained by determining internal control materials at clinically significant concentrations [1]. The index of individuality (II), a ratio calculated as (SDintra2/ SDbetween2)1/2, allowed the objective assessment of the value of population-based reference values. When the II is greater than 1.4, the conventional population-based reference values are relevant. When the ratio is low, particularly less than 0.6, conventional population-based reference values are of little relevance in the correct evaluation of results because individuals may have marked changes in marker levels that are significantly different from their own usual values and all results may still lie within the usual reference interval [16]. The index of heterogeneity (IH) was calculated as the ratio between the observed CV of the set of individuals (including analytical) to the theoretical CV, which is (2/k − 1)1/2, k being the number of specimens collected per subject. If the index differed from 1.0 by more than 2
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
Standard Deviation (SD) of the difference between this ratio and its expected value of unity, the heterogeneity was considered significant. The SD of the difference between this ratio and its expected value of unity (under the hypothesis of no heterogeneity of true withinsubject variations) is 1/(2 k)1/2. On assessing four specimens per subject, the IH values allowing us to define the homogenous withinsubject data would be IH b1.82. Data on the components of BV were used in assessing the significance of changes in serial results from an individual. The total variation required for a significant change to have occurred (RCV) was calculated by the addition of the intra-subject and analytical variability and then the SD was multiplied by 2.77 so as to attain 95% certainty that the two results were different. The value obtained was then transformed into a percentage of the overall mean. In the formula we used the analytical CV from the IQC of our laboratory at the appropriate clinical decision-making level [1].
161
Table 2 Optimal, desirable, and minimum analytical goals for imprecision, bias, and total error calculated in relation to biological variability. Quality specifications
CD3
CD4
CD8
CD19
CD56
CD16
Imprecision CV desirable CV minimum CV optimum Bias Bias desirable Bias minimum Bias optimum Total error TE desirable TE minimum TE optimum
0.007 0.020 0.030 0.010 0.005 0.027 0.040 0.013 0.016 0.060 0.090 0.030
0.014 0.040 0.060 0.020 0.026 0.047 0.070 0.023 0.049 0.113 0.169 0.056
0.028 0.025 0.038 0.013 0.057 0.066 0.099 0.033 0.010 0.107 0.161 0.054
0.040 0.055 0.083 0.028 0.010 0.078 0.116 0.039 0.057 0.168 0.252 0.084
0.077 0.125 0.188 0.063 0.013 0.138 0.206 0.069 0.114 0.344 0.516 0.172
0.077 0.115 0.173 0.058 0.013 0.126 0.190 0.063 0.114 0.316 0.474 0.158
4. Results
in order to identify the significance of changes in serial results. The IH and the RCV are shown in Table 3.
Within-subject BV and between-subjects BV expressed as CV using the overall means are shown in Table 1.
5. Discussion
4.1. Quality specifications Optimal, desirable, and minimum analytical goals for imprecision, bias, and total error calculated in relation to BV are shown in Table 2. 4.1.1. Imprecision All CD studied achieved the analytical goals, as shown in Table 2. The amount of variability added to the test result when CVA is different fractions of CVI was calculated (Fig. 1). Most of the tests studied were in the optimum performance range, the increase in result variability being 1% for CD3 and CD4, 4% for CD16 and CD56 and 6% for CD19 while the ratios of analytical imprecision to within-subject BV (SDA/SD2I ) were 0.17, 0.30 and 0.36, respectively. CD8 was within the desirable performance range, the increase in result variability being 14%; the ratio of analytical imprecision to within-subject BV was 0.56. 4.1.2. Bias and total error All CD studied achieved the analytical goal, being within the optimum performance range (Table 2). 4.2. Usefulness of reference values The values calculated for the II are shown in Table 3. All the indices were lower than 0.6, thus being of limited utility for conventional population-based reference intervals in the evaluation of changes in CD serial results. 4.3. RCV As the IH was not significant, all values being b 1.82 (1 + 2SD), within-subject variability was used to calculate the RCV. Therefore, the RCV, i.e. the minimal significant difference (p b 0.05) between two consecutive LP measurements in the same subject, were estimated Table 1 Within-subject and between-subjects biological variability expressed as CV using the overall means.
CVintra CVbetween
CD3
CD4
CD8
CD19
CD56
CD16
CD56 + 3+
3DR
0.04 0.10
0.08 0.17
0.05 0.26
0.11 0.29
0.25 0.49
0.23 0.45
0.38 0.62
0.23 0.56
Data on the components of BV were determined in order to set CD quality specifications, evaluate the utility of conventional population-based reference values, and to assess the significance of changes in serial results from an individual. Laboratory test results are used for many purposes. We use test results clinically for four different purposes: diagnosis, case finding, screening and monitoring. Quality specifications for precision and bias should ensure that all these objectives are achieved. If we develop separate quality specifications for precision and bias, then we can calculate specifications for total allowable error. In the present study, based on the biological data obtained, we established appropriate analytical performances, which were more than desirable for all the markers considered [1]. All the immunological markers fulfilled the homogeneity condition. Therefore the mean for within-subject variability observed for these indicators constitutes the basis for calculating the RCV appropriate for all patients and useful to other laboratories working with the same analytical imprecision. Moreover, since indices of individuality are calculated at less than 1, it appears difficult to identify an abnormal value in a subject if conventional population-based reference intervals are used; the RCV appears to be a better tool for identifying a true biological change in sequential measurements [17–19]. In longitudinal studies of LP, the variability between different sampling values depends on methodological and biological factors. Although it is of utmost clinical importance to evaluate differences within and between sampling values, little attention has been paid to this aspect in the literature and nor have efforts been made to take either methodological or BV into consideration. The literature contains large databases for the components of BV of many biological markers, but no studies reporting reliable data on the BV of lymphocyte subsets have yet appeared in the literature, the only exception being studies on circadian variability, an important pre-analytical factor that was minimised in our experimental protocol. In 1990 Malone et al. observed large fluctuations in repeated CD4+ cell counts in HIV + patients, and concluded that they could be explained in part by the CD4 + cell count diurnal cycle and in part by a high variability in total lymphocyte counts [20]. In a more recent study several methods were used over a long period of time in order to record monthly variations in CD4+ and CD8+ cells in HIV + patients [10]; in patients with a disease affecting the CD4/CD8 ratio a different pattern from that in healthy controls is to be expected. We analysed samples from healthy subjects who had total lymphocyte counts within the reference range, in order to obviate the effect of disease and total lymphocyte fluctuations on test results.
162
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
45
% increase in results variability
40 35 30 MINIMUM
25 20
CD8
15 DESIRABLE 10 CD19 CD16 e 56
5 OPTIMUM CD3 e 4
0 0,0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1,0
analytical imprecision/within-subject biological variation Fig. 1. Test result variability (percentage) increases as a function of the ratio of analytical imprecision to within-subject biological variability. Most of the tests range around the optimum performance, the increase in result variability being 1% for CD3 and CD4, 4% for CD16 and CD56 and 6% for CD19 while the ratios between analytical imprecision and within-subject biological variability are 0.17, 0.30 and 0.36, respectively. CD8 ranges within the desirable performance, the increase in result variability and the ratio of analytical imprecision to within-subject biological variability being 14% and 0.56, respectively.
In studies evaluating LP BV, daily [2–6] and/or yearly [7–11] variations have been reported, but these findings are now several years old and, since they appeared in literature, major improvements have been made to immune phenotyping methods. Most of the available publications on seasonal variations of LP were the fruit of pioneering studies conducted in the wake of the introduction of monoclonal antibodies but before the widespread use of flow cytometry involving the use of microscopy, ficoll-separated mononuclear cells and antibody combinations [7,8,3,9]. Although monthly and seasonal fluctuations were noted in these studies, it is not surprising that patterns found were inconsistent. However, recent studies demonstrated that the circadian variability of lymphocyte subsets is highly relevant. A number of recent papers point to the important circadian rhythmicity of lymphocyte subpopulations, which seems to be related with neuro-endocrine system [12–14]. In particular, regarding LP, a clear circadian rhythm was validated for CD8 and CD16 with acrophase in the morning and at noon, and for CD3, CD4, CD4/CD8 ratio, HLA-DR and CD20 with acrophase at night [12].Therefore, in the present study we obtained samples at the same time of day in order to minimise the effect of circadian variability in subjects. Moreover, since age-related changes in 24-h rhythms have been found in older humans, in one study on circadian variability in lymphocyte subpopulations were also evaluated in subjects aged 67 to 79 years and compared with those in subjects 36 to 55 years. The authors concluded that aging is associated with enhanced T cell compartment responsiveness and alterations in circadian rhythmicity [13,14]. In order to obviate any age-related interference we recruited subjects in the age range of 25 to 58 years for the present study. Esquifino et al. observed in growing rats that the availability of nutrients may also affect the mechanisms modulating the circadian Table 3 Individuality and heterogeneity indices and reference change values.
Individuality index Heterogeneity index RCV
CD3
CD4
CD8
CD19
CD56
CD16
0.13 0.04 53%
0.20 0.09 67%
0.03 0.06 27%
0.14 0.13 26%
0.28 0.30 64%
0.21 0.29 65%
variability of immune response. On determining over a 24-h time span, the lymphocyte subset populations in the submaxillary lymph nodes of rats that had been submitted to a calorie restriction diet, the authors found that T and CD4+ cell number and CD4/CD8 ratio augmented, whereas B cell number decreased [21]. Therefore we asked the subjects in our study to continue with their dietary habits and avoid calorie restriction. Bearing in mind that other factors affecting LP are smoking habit [22], exercise training [23], menstrual cycle [24] and pregnancy [25], we excluded smokers, samples were drawn at least 12 h after exercise, no samples were collected from women at the same time in the menstrual cycle and pregnant women were excluded. In a longitudinal study [26] on the biological and methodological variability of lymphocyte subsets in the blood of human adults, the authors concluded that there were no major fluctuations of LP over a year, and that the variability noted in the longitudinal study included both biological and methodological variability, this being the most important contributory factor for variability in repeated measurements over time. Our study, which reports on a final assessment of the analytical and biological components of variability in lymphocyte subsets, used an accurately designed experimental and statistical protocol that enabled us to delete accurately the numerous pre-analytical confounding factors reported by other authors. 6. Conclusions The results obtained in our study provide further evidence of the relevance of BV in the evaluation of immunological markers of LP. As quality specifications based on BV are widely accepted criteria for ensuring valid clinical decision-making, they should be considered prerequisites for setting IQC and quality assurance goals. Quality specifications represent “the level of performance required to facilitate clinical decision-making” [1] and should be used for internal quality management procedures as well as for facilitating clinical reasoning, decision-making and patient management. Thus, quality specifications are prerequisites for implementing a valid quality management system.
F. Tosato et al. / Clinica Chimica Acta 424 (2013) 159–163
The data obtained in the present study demonstrate that analytical performance goals based on BV were achieved for all CDs studied. Moreover, as the II for evaluated CDs does not indicate that the traditional comparison between patient results and reference values is useful, we have identified appropriate RCV for a valid interpretation of data, namely data obtained in patient monitoring through serial results.
References [1] Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409–37. [2] Ritchie AW, Oswald I, Mickelm HS, et al. Circadian variation of lymphocyte subpopulations: a study with monoclonal antibodies. Br Med J (Clin Res Ed) 1983;286:1773–5. [3] Lévi FA, Canon C, Tuoitou Y, Reinberg A, Mathé G. Seasonal modulation of the circadian time structure of circulation T and natural killer lymphocyte subsets from healthy subjects. J Clin Invest 1988;81:407–13. [4] Pasqualetti P, Colantonio D, Casale R, Colangeli S, Natali G. Circadian rhythm of human lymphocyte subpopulations. Quad Sclavo Diagn 1988;24:89–95. [5] Fukuda R, Ichikawa Y, Takaya M, Ogawa Y, Masumoto A. Circadian variations and prednisolone-induced alterations of circulating lymphocyte subsets in man. Intern Med 1994;33:733–8. [6] Mazzoccoli G, Bianco G, Correra M, et al. Circadian variation of lymphocyte subsets in healthy subjects. Recent Prog Med 1998;89:569–72. [7] Bratescu A, Teodorescu M. Circannual variations in the B cell/T cell ratio in normal human peripheral blood. J Allergy Clin Immunol 1981;68:273–80. [8] Abo T, Miller CA, Cloud GA, Blach CM. Annual stability in the levels of lymphocyte subpopulations identified by monoclonal antibodies in blood of healthy individuals. J Clin Immunol 1985;5:13–20. [9] Afoke AO, Eeg-Olofsson O, Hed J, Kjellman NI, Lindblom B, Ludvigsson J. Seasonal variation and sex differences of circulating macrophages, immunoglobulins and lymphocytes in healthy school children. Scand J Immunol 1993;37:209–15. [10] Paglieroni TG, Holland PV. Circannual variation in lymphocyte subsets, revisited. Transfusion 1994;34:512–6. [11] Termorshuizen F, Geskus RG, Roos MT, Coutinho RA, Van Loveren H. Seasonal influences on immunological parameters in HIV-infected homosexual men: searching for the immunomodulating effects of sunlight. Int J Hyg Environ Health 2002;205:379–84.
163
[12] Mazzocolli G, De Cata A, Greco A, Carughi S, Giuliani F, Tarquini R. Circadian rhythmicity of lymphocyte subpopulations and relationship with neuro-endocrine system. J Biol Regul Homeost Agents 2010;24:341–50. [13] Mazzocolli G, Vendemiale G, La Viola M, et al. Circadian variation of cortisol, melatonin and lymphocyte subpopulations in geriatric age. Int J Immunopathol Pharmacol 2010;23:289–96. [14] Mazzocolli G, De Cata A, Greco A, et al. Aging related changes of circadian rhythmicity of cytotoxic lymphocyte subpopulations. J Circadian Rhythms 2010;8:6. [15] NCCLS H42–A2 guidelines. Enumeration of immunologically defined cell populations by flow cytometry; approved guideline – second edition. Vol. 27 No. 16. 2007. [16] Harris KH. Effects of intra- and interindividual variation on the appropriate use of normal ranges. Clin Chem 1974;20:1535–42. [17] Ross SM, Fraser CG. Biological variation of cardiac markers: analytical and clinical considerations. Ann Clin Biochem 1998;35:80–4. [18] Fraser CG, Lippi G, Plebani M. Reference change values may need some improvement but are invaluable tools in laboratory medicine. Clin Chem Lab Med 2012;50:963–4. [19] Plebani M, Lippi G. Biological variation and reference change values: an essential piece of the puzzle of laboratory testing. Clin Chem Lab Med 2012;50:189–90. [20] Malone JL, Simms TE, Gray GC, Wagner KF, Burge JR, Burke DS. Sources of variability in repeated T-helper lymphocyte counts from human immunodeficiency virus type 1-infected patients: total lymphocyte count fluctuations and diurnal cycle are important. J Acquir Immune Defic Syndr 1990;3: 144–51. [21] Esquifino AI, Chacon F, Cano P, Marcos A, Cutrera RA, Cardinali DP. Twentyfour-hour rhythms of mitogenic responses, lymphocyte subset populations and amino acid content in submaxillary lymph nodes of growing male rats subjected to calorie restriction. J Neuroimmunol 2004;156:66–73. [22] Schaberg T, Theilacker C, Nitschke OT, Lode H. Lymphocyte subsets in peripheral blood and smoking habits. Lung 1997;175:387–94. [23] Nehlsen-Cannarella SL, Nieman DC, Balk-Lamberton AJ, et al. The effects of moderate exercise training on immune response. Med Sci Sports Exerc 1991;23: 64–70. [24] Northern AL, Rutter SM, Peterson CM. Cyclic changes in the concentrations of peripheral blood immune cells during the normal menstrual cycle. Proc Soc Exp Biol Med 1994;207:81–8. [25] Matthiesen L, Berg G, Ernerudh J, Håkansson L. Lymphocyte subsets and mitogen stimulation of blood lymphocytes in normal pregnancy. Am J Reprod Immunol 1996;35:70–9. [26] Backteman K, Ernerudh J. Biological and methodological variation of lymphocyte subsets in blood of human adults. J Immunol Methods 2007;322:20–7.