The international community recognizes the pressing medical, moral, social and economic imperatives to expand access to antiretroviral therapy to the many people living with HIV/AIDS in the developing world. As CD4 cell counts are the best indicator of the risk of occurrence of opportunistic infections, the time of initiation of antiretroviral treatment, and of immunologic response to therapy, it is urgent that CD4 cell monitoring becomes more widely available in resource-limited settings .
Flow cytometry (FC) is the accepted standard method for enumerating CD4+ T lymphocytes [2,3]. However, the high cost of the equipment and of reagents, and problems associated with maintenance, prevent FC from being widely used in developing countries. Simple and less expensive methods for the enumeration of CD4+ T lymphocytes, not requiring complex laboratory equipment, have been developed [4–9]. The introduction of such techniques should allow for the follow-up of large numbers of patients in sites that are not equipped for FC, particularly in rural zones distant from major cities.
In a previous study , we found that the Dynabeads (Dynal Biotech, Oslo, Norway) technique, an alternative method based on CD4 T-cell isolation using anti-CD4 monoclonal antibody-coated magnetic beads, compared better than other alternative non-cytofluorometric methods with standard FC. In addition, the Dynabeads method appears to be simple and less expensive than other alternative methods, emphasizing its potential usefulness in countries with limited resources.
In the present study we evaluated the feasibility and the relevance of the implementation of the Dynabeads technique for enumerating CD4+ T cells in HIV-infected patients in five countries of West Africa.
Patients and methods
The study was conducted in West Africa between December 2000 and December 2001 with the main objective of comparing the CD4 T-lymphocyte enumeration provided in the field by the two techniques. Study participants included 35 HIV-infected individuals from Abidjan, Côte d'Ivoire (89 pairs of samples, see below), 28 individuals from Bamako, Mali (68 pairs of samples), 86 individuals from Bobo-Dioulasso, Burkina Faso (181 pairs of samples), 55 individuals from Dakar, Senegal (54 pairs of samples), 43 individuals from Lome, Togo (134 pairs of samples) and 54 individuals from Ouagadougou, Burkina Faso (131 pairs of samples). All participants recruited gave signed or thumbprint informed consent for their participation in the study. Prophylactic drugs including cotrimoxazole were offered by the study sponsoring agency for patients with CD4 cell counts below 500 × 106 cells/l. Patients were invited to attend each run of the dual CD4 T-lymphocyte enumeration organized every 3 months.
Sites and training
Six laboratory sites (Abidjan, Bamako, Bobo-Dioulasso, Dakar, Lome, Ouagadougou) and 43 laboratory technicians with little or no experience in immunological techniques from five countries, participated in the study which had been approved in each country by the national health authorities and ethic committees. The Centre Muraz in Bobo-Dioulasso was the reference site for the study. All participating laboratory technicians were trained locally for 2 days by the principal investigator and one technician from the reference site. Standardized training included the transfer of the Dynabeads technique and information on the processing of blood samples for FC.
The Dynabeads technique for CD4 T-cell enumeration
The Dynabeads technique uses magnetic beads coated with anti-CD4 monoclonal antibodies (mAbs) to capture and isolate CD4+ T lymphocytes from whole blood. One hundred and twenty-five microlitres of freshly-obtained EDTA-anti-coagulated blood were added to 350 μl of phosphate buffer saline (PBS). Twenty-five microlitres of suspended magnetic beads coated with anti-CD14 mAb were then added and the mixture incubated for 10 min at room temperature on a Dynal mechanical rotator in order to deplete blood from monocytes. Magnetic separation of monocytes was performed using a magnetic particle concentrator, as recommended by the manufacturer. One aliquot of 200 μl was taken from the supernatant of monocyte-depleted blood and dispensed into 200 μl of PBS. Twenty-five microlitres of beads coated with anti-CD4 mAb were then added prior to incubation at room temperature for 10 min on the rotator. The beads were separated using the magnetic particle concentrator and washed twice with PBS. After addition of 50 μl of lysing solution, cells were stained with 50 μl of a solution of acridine orange and layered on a 10-band Malassez slide where they were integrally enumerated using an epifluorescent microscope. Results were expressed as number of positive cells per microlitre of whole blood. The current reagent cost of a single assay by this technique is US$ 3.
Single platform flow cytometry technique
CD4 cell counts were performed using three-color direct immunofluorescence according to standard procedures using TruCount [Becton Dickinson Immunocytometry Systems (BDIS), San Jose, California, USA]. Fifty microlitres of freshly-obtained whole blood were stained with a combination of anti-CD45-perCP, anti-CD3-FITC and anti-CD4-PE mAbs for 15 min at room temperature in the presence of a fixed number of fluorochrome-labeled, polystyrene reference beads. Red blood cells were lysed using a Facs Lysing Solution (BDIS) for 15 min at room temperature. Stained cells were then fixed by addition of a Cell Fix solution (BDIS) and sent by air within 12 h to the reference site for FC analyses except in the case of the Dakar site that had performed the analyses locally.
Samples were analyzed using a FACScan flow cytometer (BDIS) and the Multiset and Atractor software (BDIS) for calculating absolute values of CD4 cells. The current cost of a single assay by this technique is approximately US$ 25.
The study consisted in a dual CD4 T-lymphocyte enumeration by the Dynabeads technique and the FC technique of all the blood samples collected across the sites during five experimental runs. Blood samples were collected from HIV-infected patients and separated into two aliquots. Each aliquot was used for a single CD4 enumeration by either the Dynabeads assay or FC. Instrument quality control procedures were performed before each experimental run and included optics alignment and flow cytometer standardization with fluorochrome-labeled microbeads. Quality controls using stabilized blood (CD Chex-Plus; Becton Dickinson) were performed for CD4 enumeration by FC. Stabilized samples were not useable with the Dynabeads assay for direct calibration of this technique. Dual CD4 cell enumeration by both techniques were performed independently and in a blinded manner.
Assay reproducibility was assessed in a separate experiment using 130 duplicate samples for the Dynabeads technique and 65 for FC.
Statistical analyses were performed using the SAS-PC software (version 8.2, SAS Institute, Cary, North Carolina, USA). Analyses were performed on samples for which results were obtained by both FC and the Dynabeads methods. The variable analysed was the absolute CD4 cell count. P-values were the results of two-sided tests and considered as significant if below 5%. All confidence intervals were two-sided with a significance level of 5%. Difference between the two techniques was computed on each sample as the subtraction of the FC result to the Dynabeads result. Result from FC was used as a reference for CD4 levels in the analyses. Continuous variables were summarized by means, standard deviations, ranges and quartiles. Categorical data were summarized by counts and percentages. The correlation coefficient between the two techniques was calculated and a model of regression then applied to the data. The difference between the two techniques was determined overall and by levels of CD4 cell counts. The absolute value of the difference, that is, the amplitude of the difference between the results of the two techniques in a given sample, was used to characterize the within sample distance of the two techniques and summarized in the same way as the difference.
We also determined the number and proportion of pairs of results not consistently classified by both techniques at various thresholds of CD4 cells relevant for the clinical management of HIV-infected patients, with a specific emphasis on the 200 × 106 cells/l threshold. A confidence interval of this proportion was built and a test of kappa performed. Considering a difference of up to 100 × 106 cells/l may occur in about 10% of individual patients with CD4 cell counts of around 200 × 106 cells/l tested in two close separate occasions , we also restricted the analyses of discrepant pairs to those with a difference greater than 100 × 106 cells/l. For pairs of discrepant results at the threshold of 200 × 106 cells/l, with an absolute difference greater than 100, provided that the patient had longitudinal data, we implemented an algorithm to detect which technique was failing [This algorithm calculates overall the results for a patient, how many results are less than 200 and how many are greater or equal than 200. A technique was determined to be failing when it gave the same result as the minority of cases.]. The numbers and proportions [confidence interval (CI)] of discrepant pairs of results with a difference exceeding 100 were also calculated for the thresholds of 350 and 500 × 106 cells/l.
Six hundred and fifty-seven pairs of values of CD4 cell counts were generated by both FC and the Dynabeads technique from 684 blood samples. Results from FC were not available for 23 samples, due to shipping problems. Results from Dynabeads technique were missing for four samples. Samples were obtained from 301 HIV-infected patients seen in one (n = 112), two (n = 61), three (n = 75), four (n = 40) or five (n = 13) occasions in 12 outpatient clinics.
Among the 657 samples, CD4 cell counts, as measured by the FC reference method, were below 200 × 106 cells/l in 266 cases (40.5%), between 200 and 349 × 106 cells/l in 128 cases (19.5%), between 350 and 499 × 106 cells/l in 62 cases (9.4%), between 500 and 999 × 106 cells/l in 132 cases (20.1 %) and above 1000 × 106 cells/l in 69 cases (10.5%). Overall, the median number of circulating CD4 cells was 260 × 106 cells/l, with an interquartile range (IQR) of 109–613; the mean number was 416 × 106 cells/l (standard error: 429 × 106 cells/l). The mean decrease in CD4 cell count between visit 1 and 2 was 38 cells in 90 patients, 56 between visit 1 to 3 in 95 patients, 57 between visit 1 and 4 in 83 patients and 100 cells between visit 1 and 5 in 13 patients.
Figure 1 depicts the correlation plot for the two methods for all the data analysed. The correlation coefficient was 0.89 (P < 10−4). The overall median difference between the Dynabeads technique and FC was −16 (95% CI, −22 to −8) × 106 cells/l. Median differences between the two techniques was insignificant (+7.5 cells) for CD4 cell counts below 200 × 106 cells/l and increased with CD4 levels. Thus, the systematic median difference was −23 cells for CD4 cell counts between 200 and 350 × 106 cells/l, −43.5 cells for CD4 cell counts between 350 and 500 × 106 cells/l, −96.5 cells for CD4 cell counts between 500 and 1000 × 106 cells/l and −269 cells for CD4 cell counts above 1000 × 106 cells/l. The analysis of the residuals from a regression model indicated that the fit to a linear model was not adequate since the dispersion of the residuals increased with CD4 levels.
An additional experiment was performed on 15 samples to assess whether the concentration of anti-CD4 mAb-coated beads was a limiting factor in enumerating the upper values of CD4 T cells. Samples exhibiting a median value of CD4 cell counts of 839 × 106 cells/l, were analysed by the Dynabeads technique using concentrations of mAb-coated beads suspension increasing from 25 to 40 μl by steps of 5 μl. CD4 cell counts were summarized by the level of bead concentration. We did not find any significant difference in the results of CD4 cell enumeration while increasing the amount of anti-CD4 mAb-coated beads (data not shown).
We further assessed the median (interquartile) amplitude of the difference between results of the two techniques by levels of CD4 cells. The overall median amplitude was 56 (22–137). The median (interquartile) amplitude of the difference was 25 (11–51) for CD4 cell counts below 200 × 106 cells/l, 54 (24–95) for CD4 cell counts between 200 and 350 × 106 cells/l, 90 (37–165) for CD4 cell counts between 350 and 500 × 106 cells/l, 126 (70–214) for CD4 cell counts between 500 and 1000 × 106 cells/l and 320 (160–493) for CD4 cell counts above 1000 × 106 cells/l.
In 583 of 657 cases (88.7%), patients were consistently classified at the threshold of 200 × 106 cells/l by both methods Among the 74 discrepant pairs of values, 43 exhibited a difference of less than 100 cells (of which 22 exhibiting a difference of less than 50) and only 31 (4.7%; 95% CI, 3.1, 6.3) exhibited a difference of more than 100 × 106 cells/l. The algorithm (see above) applied to the 29 discrepant results from individuals with longitudinal follow-up, suggested that the failing technique was Dynabeads in 19 cases, FC in eight cases and undetermined in two cases. Restricting the analysis to the patients with CD4 cell counts below 500 × 106 cells/l, to better target the population requiring CD4 cell monitoring, the proportion of discrepant pairs of values was 69 of 456 (15.1%; 95% CI, 11.8, 18.4), of which 26 (5.7%; 95% CI, 3.6, 7.8) exhibited a difference greater than 100 × 106 cells/l. Thus patients were consistently classified in 88.7% of the cases (strict definition) or in 95.3% of the cases allowing for a difference up to 100 × 106 cells/l (Table 1).
Discrepant results with a difference exceeding 100 × 106 cells/l in the study samples, were observed for the thresholds of 350 and 500 × 106 cells/l, in 8.8% (95% CI, 6.7, 11.0) and 7.6% (95% CI, 5.6, 9.6), respectively.
The comparison of the results of the consecutive dual CD4 cell enumeration suggested that maximal quality was reached from the first run and was sustained thereafter. Thus, as illustrated in Figure 2, correlation coefficients between the results of the two techniques ranged from 0.86 to 0.92, achieved at the third and first dual CD4 cell enumeration respectively. Moreover, discrepant results with a difference exceeding 100 × 106 cells/l at the threshold of 200 × 106 cells/l were observed in 4.3 % of cases at the first run, 7.4% at the second run, 4.7% at the third run and 1.9% at the fourth run. Among the sites, correlation coefficients ranged from 0.72 to 0.94.
The reproducibility of each technique was assessed in a specific set of experiments in which 130 and 65 duplicate samples were assayed using the Dynabeads technique and FC, respectively. The coefficient of variation for the Dynabeads technique was 8.4%, ranging from 5.3 to 14.6% in the different sites. The coefficient of variation for FC performed in the study reference site was 8.3%.
The impact of the delay in sample handling was assessed in 28 samples that were assayed by the Dynabeads technique at hours 0, 4, 8, 12, and 24 after obtaining the sample. Whereas median values varied from 318 × 106 cells/l at time 0 to 296 × 106 cells/l at hour 24, mean values decreased from 370 × 106 cells/l at time 0 to 309 × 106 cells/l at hour 24, indicating that a noticeable decrease in CD4 cell counts occurred in a proportion of samples (Table 2). However, the mean and median variations within 8 h were insignificant.
The aim of the present study was to assess the feasibility, accuracy, reproducibility and the relevance of implementation of a technique alternative to FC for enumerating circulating CD4+ T cells in HIV-infected patients in resource-limited settings. We have selected the Dynabeads technique, which, in preliminary studies, appeared effective in terms of performance, cost and technical simplicity [5,10]. We performed a large-scale study in six sites in five countries of West Africa involving 43 laboratory technicians. Six hundred and fifty-seven pairs of values of CD4 cell counts were generated by both the ‘state-of-the-art’ single platform FC and the Dynabeads technique in up to five runs of dual CD4 cell enumeration within 1 year.
Most patients included in the study were at an advanced stage of HIV disease since more than 40% of the 657 values analysed were below 200 × 106 cells/l. Results from the Dynabeads technique and FC were highly correlated (r = 0.89) with an insignificant systematic difference between the two methods for CD4 cell counts below 500 × 106 cells/l. For CD4 levels above 500 × 106 cells/l, results obtained with Dynabeads were lower than those obtained by FC with a median difference reaching 269 cells for CD4 cell counts above 1000 × 106 cells/l. The latter difference was not related to a sub-optimal concentration of anti-CD4 Mab-coated beads in the assay, as it has been found with other techniques . It could be related, at least in part, to the fact that the results obtained by the TruCount method deviate more to the hight from those obtained by conventional FC for upper CD4 values .
The ability of the Dynabeads technique to classify patients in a consistent fashion with FC, at thresholds of CD4 cell counts critical for clinical care, was close to 95% ignoring differences of less than 100 × 106 cells/l. The lack of stabilized samples useable with both techniques for CD4 count calibration and external control was overcome in this study by the availability of longitudinal data in 61% of the patients which allowed, in most cases, to determine which of the two techniques could be failing. Results obtained using a predetermined algorithm were consistent with clinical judgement and suggested that Dynabeads contributed slightly more than FC to these infrequent discrepancies.
In experimental conditions which included the counting of the whole Malassez slide, the reproducibility of the Dynabeads technique was high, similar to that expected from FC [13,14]. The overall coefficient of variation was 8.4%, ranging from 5.3 to 14.6% in the different sites suggesting that the performance of the assay was mastered in a homogeneous way in all the sites. These results were achieved after a single 2-day training of local laboratory technicians. In the study reference site in Bobo-Dioulasso, the coefficient of variation for FC was 8.3%, consistent with previous reports [12–14]. However, in our limited investigation of repeatability, the coefficient of variation of FC appeared to be high (median: 24.2%) on samples that were shipped to the reference site for FC analysis after having been processed locally for staining and paraformaldehyde fixation. Inaccuracy in sampling the blood volume, stocking conditions during shipping  could partly explain the increase in variability that was observed. It is thus possible that the correlation between the two methods is even higher than that reported here. The Dynabeads technique, implemented in this program in six West African laboratories following a 2-day local training of 43 technicians, gave optimal results in terms of correlation with FC, from the very first run, emphasizing the ease of use of the method. A batch of six samples was processed by a single technician in about 1 h. The cost of an epifluorescent microscope, which has low maintenance requirements, is currently approximately half of the cheapest equipment for FC and the current reagent cost per assay is US$ 3, which is about 12% of the current cost of a CD4 cell count determination by the currently used ‘state-of-the-art’ FC in West Africa. In laboratories already equipped for FC, the newly described ‘primary CD4 gating’ technology should lower the cost per CD4 cell count determination, especially if used with generic mAbs, . The Dynabeads technique, which is not likely to be used by laboratories already equipped for FC to further decrease their per-test costs, will remain a valuable option for measuring CD4 cell counts in low-resource laboratories that cannot afford FC technology.
Several issues may further be addressed to facilitate the routine implementation of the Dynabeads technique. The reading of the whole Malassez slide, as done in this study, was effective in terms of assay reproducibility but is time-consuming. Additional studies need to be performed however before recommending to read only part of the Malassez slide. Studies are currently in progress to assess whether light microscopy may be used instead of epifluorescent microscopy. The results were optimal when the technique was performed on blood samples drawn up to 8 h, a delay which should be manageable in most situations for a technique that would be locally implemented. Finally, this robust and easy to implement technique still requires rigorous quality controls be set up prior to large-scale implementation. These quality controls could be managed by users’ networks and rely on the shipping of whole blood samples stabilized with standardized fixatives, such as the recently introduced TransFix, from the local sites to national or international reference laboratories .
Since the availability of CD4 cell count determination is key to clinical management of HIV-infected patients and to antiretroviral therapy monitoring, an expanded access to low-cost, reliable techniques now appears to be urgent need. Our results demonstrate that the implementation of such an alternative technique to FC in the current context in Africa is now feasible.
Sponsorship: This work was supported by the Agence Nationale de Recherches sur le SIDA (ANRS) and the VIHPAL program of the French Ministry of Research.
1. World Health Organization. Scaling up Antiretroviral Therapy in Resource-limited Settings: Guidelines for a Public Health Approach. Geneva: World Health Organization, 2002.
2. Landay A, Ohlsson-Wilhelm B, Giorgi JV. Application of flow cytometry to the study of HIV infection. AIDS 1990, 4: 479–497.
3. Centers for Disease Control and Prevention. 1997 revised guidelines for performing CD4+ T-cell determinations in persons infected with human immunodeficiency virus (HIV). MMWR Morb Mortal Wkly Rep 1997, 46:1–29.
4. Landay A, Ho JL, Hom D, Russel T, Zwerner R, Minuty JG, et al. A rapid manual method for CD4+ T-cell quantitation for use in developing countries. AIDS 1993, 7:1565–1568.
5. Lyamuya EF, Kagoma C, Mbena EC, Urassa WK, Pallangyo K, Mhalu FS, et al. Evaluation of the FACScount, TRAx CD4 and Dynabeads methods for CD4 lymphocyte determination. J Immunol Methods 1996, 195:103–112.
6. Gernow A, Lisse IM, Bottiger B, Christensen L, Brattegaard K. Determination of CD4+ and CD8+ lymphocytes with the cytosphere assay: a comparative study with flow cytometry and the immunoalkaline phosphatase method. Clin Immunol Immunopathol 1995, 76:135–141.
7. Johnson D, Hirschkorn D, Busch MP. Evaluation of four alternative methodologies for determination of absolute CD4+ lymphocyte counts. The National Heart, Lung, and Blood Institute Retrovirus Epidemiology Donor Study. J Acquir Immune Defic Syndr Hum Retrovirol 1995, 10:522–530.
8. Ferreira OC, Jr., Suleiman G, Brites C, Novoa P, Piovesana M, Suleiman J, et al. A rapid and reliable assay to enumerate CD4+ T lymphocytes in whole blood. J Acquir Immune Defic Syndr 1994, 7:1224–1227.
9. Carriere D, Vendrell JP, Fontaine C, Jansen A, Reynes J, Holzmann C, et al. Whole blood capcellia CD4/CD8 immunoassay for enumeration of CD4+ and CD8+ peripheral T lymphocytes. Clin Chem 1999, 45:92–97.
10. Didier JM, Kazatchkine MD, Demouchy C, Moat C, Diagbouga S, Sepulveda C, et al. Comparative assessment of five alternative methods for CD4+ T-lymphocyte enumeration for implementation in developing countries. J Acquir Immune Defic Syndr 2001, 26:193–195.
11. Lathey JL, Hughes MD, Fiscus SA, Pi T, Jackson BJ, Rasheed S, et al. Variability and prognostic values of virologic and CD4 cell measures in human immunodeficiency virus type 1-infected patients with 200-500 CD4 cells/mm(3) (ACTG 175). AIDS Clinical Trials Group Protocol 175 Team. J Infect Dis 1998, 177:617–624.
12. Nicholson JK, Stein D, Mui T, Mack R, Hubbard M, Denny T. Evaluation of a method for counting absolute numbers of cells with a flow cytometer. Clin Diagn Lab Immunol 1997, 4:309–313.
13. Aboulker JP, Autran B, Beldjord K, Touraine F, Debre P. Consistency of routine measurements of CD4+, CD8+ peripheral blood lymphocytes. J Immunol Methods 1992, 154:155–161.
14. Schnizlein-Bick CT, Spritzler J, Wilkening CL, Nicholson JK, O'Gorman MR. Evaluation of TruCount absolute-count tubes for determining CD4 and CD8 cell numbers in human immunodeficiency virus-positive adults. Site Investigators and The NIAID DAIDS New Technologies Evaluation Group. Clin Diagn Lab Immunol 2000, 7:336–343.
15. Janossy G, Jani I, Gohde W. Affordable CD4(+) T-cell counts on 'single-platform’ flow cytometers I. Primary CD4 gating. Br J Haematol 2000, 111:1198–1208.
16. Jani V, Janossy G, Iqbal A, Mhalu FS, Lyamuya EF, Biberfeld G, et al. Affordable CD4+ T cell counts by flow cytometry. II. The use of fixed whole blood in resource-poor settings. J Immunol Methods 2001, 257:145–154.
© 2003 Lippincott Williams & Wilkins, Inc.