Diagnostic accuracy and clinical utility of a simplified low cost method of counting CD4 cells with flow cytometry in Malawi: diagnostic accuracy study.
ABSTRACT To assess the diagnostic accuracy and clinical utility of a simplified low cost method for measuring absolute and percentage CD4 counts with flow cytometry.
A CD4 counting method (Blantyre count) using a CD4 and CD45 antibody combination with reduced blood and reagent volumes. Diagnostic accuracy was assessed by measuring agreement of the index test with two other assays (TruCount and FACSCount). Clinical utility was investigated by comparing CD4 counts with the new assay with WHO clinical staging in patients with HIV.
Research laboratories and antiretroviral therapy clinic at a medical school and large government hospital in southern Malawi.
Assay comparisons were performed on consecutive blood samples sent for CD4 counting from 129 patients with HIV. Comparison of CD4 count with staging was conducted on 253 consecutive new patients attending the antiretroviral therapy clinic.
Limits of agreement with 95% confidence intervals between index test and reference standards.
The limits of agreement for Blantyre count and TruCount were excellent (cell count -48.9 to 27.0 x10(9)/l for absolute counts in the CD4 range <400x10(9)/l and -2.42% to 2.37% for CD4 percentage). The assay was affordable with reagent costs per test of $0.44 ( pound0.22, euro0.33) for both absolute count and CD4 percentage, and $0.11 for CD4 percentage alone. Of 193 patients with clinical stage I or II disease, who were ineligible for antiretroviral therapy by clinical staging criteria, 73 (38%) had CD4 counts <200x10(9)/l. By contrast, 12 (20%) of 60 patients with stage III or IV disease had CD4 counts >350x10(9)/l.
This simplified method of counting CD4 cells with flow cytometry has good agreement with established commercial assays, is affordable for routine clinical use in Africa, and could improve clinical decision making in patients with HIV.
- The Lancet 04/2004; 363(9414):1074; author reply 1074-5. · 39.06 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: A CD4 monoclonal antibody which reacts with CD4+ lymphocytes but does not react significantly with monocytes was generated and used to develop a simple method for CD4 count. In a comparison with standard flow cytometry, high correlation was obtained. The developed method can be an alternative to flow cytometry for resource-limited countries.Clinical and Vaccine Immunology 06/2006; 13(5):598-601. · 2.60 Impact Factor
- [show abstract] [hide abstract]
ABSTRACT: To study the association between the clinical axis of the World Health Organization (WHO) staging system of HIV infection and disease and laboratory markers in HIV-infected Ethiopians. Cross-sectional study. Clinical manifestations and stage of HIV-positive individuals participating in a cohort study of HIV infection progression, and of HIV-positive patients hospitalized with suspicion of AIDS, were compared to CD4+ T-cell count and viral load. Of the 86 HIV-positive participants of the cohort study, 53 (62%), 16 (19%), 16 (19%), and one (1.2%) were in stage 1, 2, 3 and 4, respectively. Minor weight loss (n = 15) and pulmonary tuberculosis (n = 9) were the most commonly diagnosed conditions among the 38 (44%) symptomatic HIV-positive individuals. Although 23 (27%) HIV-positive participants had CD4+ T-cell counts less than 200 x 10(6)/l, only one was in clinical stage 4. Among 79 hospitalized HIV-positive patients, 15 (19%) and 64 (81%) were in stage 3 and 4, respectively. The majority (83.5%) had CD4+ T-cell counts < 200 x 10(6)/l. Individuals at stage 3 had lower CD4+ T-cell counts and higher viral loads when seen in hospital as compared to cohort participants (P = 0.06 and 0.008, respectively). When grouping the two study populations, the median CD4+ T-cell count decreased (337, 262, 225, 126, and 78 x 10(6)/l, P< 0.01), and the median viral load increased (4.08, 3.89, 4.47, 5.65, and 5.65 log10 copies/ml, P < 0.01), with increasing clinical stage of HIV infection (1, 2, 3 cohort, 3 hospital, and 4, respectively). Median CD4+ T-cell counts were remarkably low in HIV-negative participants (749 x 10(6)/l), and in HIV-positive participants at stage 1 and 2 (337 and 262 x 10(6)/l, respectively). There was a good correlation between WHO clinical stages and biological markers. CD4+ T-cell counts were low in Ethiopians, particularly during early stages of HIV-1 infection, and preliminary reference values at different stages of HIV-1 infection were determined. In HIV-infected Ethiopians, lymphocyte counts less than 1,000 x 10(6)/l in non-hospitalized individuals, and less than 2,000 x 10(6)/l in hospitalized patients, had high positive predictive value, but low sensitivity, in identifying subjects with low CD4+ T-cell counts (< 200 x 10(6)/l) who would benefit from chemoprophylaxis of opportunistic infections. The on-going longitudinal study will be useful to confirm the prognostic value of the WHO staging system.AIDS 02/1999; 13(3):381-9. · 6.41 Impact Factor
diagnostic accuracy study
Calman A MacLennan, Wellcome Trust research fellow and clinical lecturer in immunology,1 2Michael K P
Liu, postdoctoral immunologist,2Sarah A White, biostatistician,1Joep J G van Oosterhout, senior clinical
lecturer in medicine,3Felanji Simukonda, laboratory scientist,1Joseph Bwanali, laboratory technician,1
Michael J Moore, laboratory manager,1Eduard E Zijlstra, professor of medicine,3Mark T Drayson, senior
clinical lecturer in immunology,2Malcolm E Molyneux, director1
Objectives To assessthe diagnostic accuracy and clinical
utility of a simplified low cost method for measuring
Design A CD4 counting method (Blantyre count) using a
CD4 and CD45 antibody combination with reduced blood
and reagent volumes. Diagnostic accuracy was assessed
by measuring agreement of the index test with two other
assays (TruCount and FACSCount). Clinical utility was
with WHO clinical staging in patients with HIV.
Setting Research laboratories and antiretroviral therapy
clinic at a medical school and large government hospital
in southern Malawi.
Participants Assay comparisons were performed on
consecutive blood samples sent for CD4 counting from
129 patients with HIV. Comparison of CD4 count with
staging was conducted on 253 consecutive new patients
attending the antiretroviral therapy clinic.
Main outcome measures Limits of agreement with 95%
confidence intervals between index test and reference
Results The limits of agreement for Blantyre count and
TruCount were excellent (cell count −48.9 to 27.0 ×109/l
for absolute counts in the CD4 range <400×109/l and
−2.42% to 2.37% for CD4 percentage). The assay was
affordable with reagent costs per test of $0.44 (£0.22,
€0.33) for both absolute count and CD4 percentage, and
$0.11 for CD4 percentage alone. Of 193 patients with
clinical stage I or II disease, who were ineligible for
had CD4 counts <200×109/l. By contrast, 12 (20%) of 60
patients with stage III or IV disease had CD4 counts
Africa, and could improve clinical decision making in
patients with HIV.
In Malawi,asubSaharanAfricancountry witha popu-
lation of 12 million, an estimated million people are
infected with HIV.1In 2004 the Ministry of Health
embarked on an ambitious antiretroviral therapy pro-
gramme. By the end of March 2007, 95674 patients
had started free antiretroviral therapy in public sector
clinics,2largely on the basis of a clinical diagnosis of
WHO stage III or stage IV HIV/AIDS.3Clinical
events, however, do not fully predict immunological
status.4When clinical criteria alone are used, some
patients with stage I and stage II disease and severe
immune suppression will not receive the treatment
they need, while others with stage III and IV disease
may still have high CD4 T cell counts and the start of
antiretroviral therapy might be delayed.5
CD4 counting could therefore improve appropriate
by the Clinton Foundation and others to reduce the
price of the necessary reagents for developing nations
for Africa.8CD4 counting with flow cytometry is per-
in Malawi.9WHO guidelines state that where CD4
counting is available, adults and children over 5 years
their CD4 counts drop below 200×109/l, regardless of
clinical staging.3In children under 5 years, CD4 per-
centage of total lymphocyte count (CD4 percentage)
varies less than absolute counts with age10and so the
percentage value is recommended to help decide on
initiation of antiretroviral therapy.11
There are two main approaches for making CD4
counting more widely available in Africa: firstly, to
reduce the cost of and simplify flow cytometric CD4
counting, and, secondly, to develop alternative count-
ing methods. Flow cytometry, however, is the ideal
method and has high accuracy.612High throughput is
Effective external quality assurance schemes are
This is version 2 of the paper.
In version 1 the authors were
listed in the wrong order.
Clinical Research Programme,
College of Medicine, University of
Malawi, PO Box 30096,
Blantyre 3, Malawi
2Medical Research Council Centre
for Immune Regulation, Division of
Immunity and Infection, University
of Birmingham, Birmingham
3Department of Medicine, College
of Medicine, University of Malawi,
Private Bag 360,
Blantyre 3, Malawi
BMJ | ONLINE FIRST | bmj.compage 1 of 8
available in Africa with NEQAS (United Kingdom
national external quality assessment scheme)13and
WHO CD4 REQAS/QASI (regional external quality
tion for immunological measures relevant to HIV/
AIDS programme).14Finally, flow cytometers can
measure CD4 percentage as well as absolute counts.
The main disadvantages are that flow cytometers are
expensive and complex, reagent costs are high, and
skilled laboratory staff are required.
Alternative counting methods include enzyme
linked immunosorbent assays (ELISA),15dried whole
blood spots,16lymphocyte rosetting,17and magnetic
beads.18Such methods do not require complex equip-
ment or the same level of staff training. The major dis-
advantage of such methods is poor ability to
discriminate between CD4 T cells and monocytes,
which also express CD4,19low throughput, and poor
ability to determine CD4 percentage. Reagent costs
are similar to those of flow cytometric methods.8
Over recent years several technological develop-
ments have shown that flow cytometric CD4 counting
could be more straightforward. “Primary CD4 gating”
uses just one antibody against CD4 and side scattered
light to discriminate between lymphocytes and
monocytes.2021Gating of lymphocytes by using
rate and reproducible than using light scatter charac-
teristics alone.2223Recently, CD45 has been used to
gate all leucocytes on dual platforms, using a flow cyt-
ometer for CD4percentageand a haematological ana-
platforms (flow cytometer alone) but remains primar-
ily focused on total leucocytes rather than on lympho-
We investigated whether these technologies could
be miniaturised to reduce costs and applied them to
performed with reduced reagent costs and could accu-
rately determine both absolute and percentage CD4
with increased simplicity compared with existing flow
cytometric methods. We compared our method with
assays for diagnostic accuracy and assessed the poten-
tial impact on clinical decision making.
The study was conducted at the Malawi-Liverpool-
zabeth Central Hospital in Blantyre, the largest city in
the southern region of Malawi. The estimated preva-
lence of HIV infection among adults in Blantyre dis-
trict is 22%.25All participants gave informed consent
for CD4 counting.
We used a FACSCalibur flow cytometer, a FACS-
Count instrument (both Becton Dickinson, CA,
USA), and HmX haematological analyser (Beckman
Coulter, CA, USA). We analysed flow cytometric
data with CellQuest and MultiSet software.
Reference standards (TruCount and FACSCount assays)
counts using Multitest CD3/8/45/4 kits with TruCount
tubes and FACSCount reagent kits (Becton Dickinson)
according tothe manufacturer’s instructions. TruCount
Count as thereference standard because it isa commer-
cial CD4 counting method that was developed to be
used on the same instrument as the index test (Becton
DickinsonFACSCaliburflow cytometer) and generates
both absolute and percentage CD4. We used FACS-
Count as a second reference standard because it is one
of the most widely used CD4 counting technologies in
Africa. Both TruCount and FACSCount generate CD4
counts on a single platform, although FACSCount
010 1001000 10 000
010 100 100010 000
Fig 1 | Gating strategies for determining absolute and
percentage CD4 counts from flow cytometric data acquired
with Blantyre count. R1=CD4 positive T lymphocytes,
R2=CytoCount fluorescent microbeads, R3=total
lymphocytes, M=monocytes, N=neutrophils, L=CD4 negative
lymphocytes. Plots are for data acquired from same blood
against CD4-PE (phycoerythrin). Absolute CD4 count=(R1/
R2)×([beads] (beads/µl)/2); bottom: contour plot of side
scattered light against CD45-FITC (fluorescein
isothiocyanate). CD4 percentage=(R1 (from top)/R3)×100
page 2 of 8
BMJ | ONLINE FIRST | bmj.com
lyser to generate CD4 percentage. Both assays are used
by clinical laboratories throughout the world and have
been validated by consistently high performance in
external quality assurance schemes such as UK
NEQAS13over a period of years.
Index test (Blantyre count assay)
We used venous blood from healthy adults anti-
coagulated with EDTA to develop our assay. We
used antihuman CD4 antibody conjugated with phy-
coerythrin (CD4-PE) and antihuman CD45 antibody
conjugated with fluorescein isothiocyanate (CD45-
FITC), FACS lysing solution (all Becton Dickinson),
and CytoCount fluorescent microbeads (Dako, Den-
Gilson, France) were used for all pipetting steps. The
same pipette was used for reverse pipetting of blood
and counting beads. CD4 T cell counts and total
lymphocyte populations were determined by using
staining for CD4 and CD45 with no attempt to gate
total leucocytes or total T cells. We mixed 20 µl
whole blood with 0.5 µl CD4-PE and 0.5 µl CD45-
15 minutes in the dark at room temperature. Red
ting for pipetting blood and counting beads as precise
volumes are critical and this method is more accurate
than conventional pipetting.12Pipette calibration and
pipetting accuracy were assessed by dispensing 10 µl
and 20 µl aliquots of water on to a scientific balance.
60 seconds, were acquired by using a live gate with
acquisition threshold set on the FL1 (FITC) channel.
Analysis was performed with a CD4-PE against side
scatter dot plot with manual gating of the CD4 T cell
for manual gating of the total lymphocyte population
(R3) (fig 1 bottom).
formula: (CD4 T cell events (R1)/bead events
(R2))×([bead solution] (beads/µl)/2).
We calculated CD4 percentage with the formula:
We assessed repeatability of our assay by perform-
ingfive repeats of the assay on fourblood samplesand
examined stability of results with time by leaving a
blood sample in the laboratory at room temperature
and performing five repeats of the assay daily on the
sample over five days.
Modified Blantyre count assays
We modified our assay to reduce costs further when
only an absolute or percentage CD4 is required. The
absolute CD4 count alone variant used CD4-PE anti-
body plus beads but without CD45-FITC antibody
(Blantyre count (absolute)). The variant giving the
CD4 percentage alone used CD4 and CD45 anti-
bodies without counting beads (Blantyre count
CD4 counting comparison studies
In the main CD4 counting comparison study we
included consecutive blood samples from patients
with HIV sent to our laboratory for full blood count
and CD4 count determination from 27 January to 17
February and from 18 April to 9 May 2006 (n=134).
ple using Blantyre count, Blantyre count (absolute),
TruCount, and FACSCount assays.
We carried out a subsequent smaller study on con-
secutive blood samplesfrom patientswith HIV sent to
the laboratory in June 2006 to compare CD4 percen-
tages generated by Blantyre count and Blantyre count
(percentage) assays (n=28). Samples were not used if
were taken unless they were received after 4 pm in
which case they were processed the next morning.
Data collection was planned before the index tests
and reference standards were performed. All blood
samples from all participants meeting the inclusion
criteria underwent the index and reference standard
tests. No adverse events occurred from performing
Table1 | ReagentcostsperassaywithflowcytometryforcountingCD4cells.Costscalculatedwithpricesavailabletousin$forall
Red cell lysing
Absolute and percentage0.0560.0420.0140.3290.44
Percentage 0.056 0.042 0.014
TruCount (Multitest) (absolute and percentage)
CD4-PE=phycoerythrin conjugated antihuman CD4 antibody; CD4-FITC=fluorescein isothiocyanate conjugated antihuman CD45 antibody.
*Allowing for 12 µl/assay with reverse pipetting.
†Not provided with kit; 450 µl of 1× lysing solution required/assay.
BMJ | ONLINE FIRST | bmj.compage 3 of 8
Two authors (FS and JB) performed and read the
FACSCount assay and full blood count. Two other
authors (MKPL and CAM), both of whom had pre-
vious experience of flow cytometry, performed and
read TruCount and Blantyre count assays together
within six hours of the FACSCount assay. We have
subsequently trained local laboratory technicians
over two to three days to perform the Blantyre count
method. Manual gating of events acquired with Blan-
lated this from the FACSCount absolute CD4 count
and total lymphocyte count from the haematological
analyser using the formula (CD4 (cells×109/l)/total
lymphocyte count (cells×109/l))×100. This procedure
is prone to errorbecause CD4 percentageis generated
on a dual platform setting, which is inherently more
variable than single platform operations. For absolute
CD4 counts, we assessed agreement only for samples
the relevant range for clinical decision making. For
comparisons of CD4 percentage we used all samples.
Clinical utility study
We tested a further 253 EDTA anticoagulated venous
blood samples from new patients attending the adult
antiretroviral therapy clinic from May to July and
from September to October 2006 for CD4 count
using Blantyre count. CD4 counts and clinical staging
were compared for each patient. Clinical staff in the
antiretroviral therapy clinic performed staging blind
to CD4 count results.
External quality assurance
To determine the accuracy of Blantyre count, we
enrolled the assay for external quality assurance with
the NEQAS immune monitoring scheme.13CD4
results were determined on six NEQAS stabilised
blood samples from the UK between January and
We examined agreement between each pair of meth-
ment (=bias plus or minus 1.96×SD) with 95%
confidence intervals as described by Bland and
Altman.28Repeatability was assessed with coefficients
of variation obtained from five repeats of assays.
Refinement of Blantyre count
count assay, coefficients of variation were 4.2%, 4.2%,
6.1%, 8.1%, and 10.7%, respectively, showing a pro-
gressive increase in coefficients of variation below 10
µl of blood. There was a decrease in mean absolute
CD4 count as blood volume was reduced (712, 681,
645, 539, and 448×109/l, respectively). CD4 percen-
tages showed better overall repeatability (coefficients
of variation 2.8% with 20 µl blood) and the coefficient
Mean CD4 x109/l
Difference in CD4 x109/l
Blantyre count and TruCount
Mean CD4 percentage
Difference in CD4 percentage
Mean CD4 percentage
Difference in CD4 percentage
Mean CD4 x109/l
Difference in CD4 x109/l
FACSCount and TruCount
Blantyre count and TruCountFACSCount and TruCount
Fig 2 | Comparison of CD4 counts determined with three flow cytometric methods using 96 blood samples from patients with
HIV with CD4 counts <400×109/l for absolute CD4 cell counts (×109/l) and 129 blood samples for CD4 cell counts as a
percentage of total lymphocyte count (CD4 percentage). FACSCount CD4 percentage was obtained from FACSCount absolute
CD4 counts and total lymphocyte counts from a haematological analyser. Black lines depict bias and upper and lower limits of
agreement. Grey broken lines denote 95% confidence intervals for these values
page 4 of 8
BMJ | ONLINE FIRST | bmj.com
of variation did not noticeably increase until blood
volume was reduced to 2 µl (4.8%). We used 20 µl
blood for our assay as twice the lowest volume at
which optimal assay repeatability was maintained.
We used a similar process to determine optimal
counting bead volume. With 10 µl counting beads the
coefficient of variation was 4.2%, and similarly with 5
µl beads(4.3%), butincreased to 8.9% and5.7% with 2
and 1 µl beads, respectively. Mean CD4 counts were
not affected by bead volume. We used 10 µl counting
tities of antibody per assay. Discrimination of CD4 T
cell and total lymphocytes as discrete populations was
still possible down to 0.25 µl of CD4-PE and 0.25 µl
CD45-FITC. We chose 0.5 µl of each antibody for use
in our assay. Using these quantities, the costs of
reagents per assay were $0.44 (£0.22, €0.33) for both
absolute and percentage counts, $0.40 for an absolute
CD4 count, and $0.11 for CD4 percentage alone
Characteristics of patients
patients with HIV. One sample gave an absolute CD4
count >2000×109/l with FACSCount and was
excluded from subsequent analyses. There were no
indeterminate or missing results. The median age of
patients was 33 years (range 2-75 years); 79 were
females; and 38 (29%) were taking antiretroviral ther-
apy at the time of testing. Of the 253 new patients for
whom CD4 count was compared with clinical staging
in the antiretroviral therapyclinic, themedian agewas
33 (range 14-71), and 173 were female. None was tak-
ing antiretroviral therapy. We used a subset of 28 of
these patients for the separate comparison of Blantyre
count with the Blantyre count (percentage) variant.
The median age in this subset was 35 years (range
24-64 years), and 18 were female.
Absolute CD4 counts in agreement studies
The median CD4 count was 193×109/l (range 0 to
1884×109/l) with TruCount. The mean bias when we
with a CD4 count of <400×109/l was −11.0×109/l for
Blantyre count. Similarly, low biases were found for
other assay comparisons: 1.2×109/l for FACSCount
compared with TruCount, and 5.8×109/l for Blantyre
count (absolute) compared with Blantyre count
(table 2). Limits of agreement were −48.9 to
27.0×109/l for Blantyre count and TruCount and
were within the range −55 to 40×109/l for all other
assay comparisons (table 2, fig 2).
CD4 percentage in agreement studies
The median CD4 percentage was 13.0% (range 0.0-
44.0%) using TruCount. Agreement between CD4
percentage generated by Blantyre count and Tru-
Count was excellent over the full range of values,
with a bias of −0.03% and limits of agreement −2.42%
Count CD4 percentage with values generated using
FACSCount showed poorer agreement, with a bias of
0.92% and limits of agreement −5.83% to 7.66% for
FACSCount and TruCount and a bias of −0.94% and
limits of agreement −7.56% to 5.67% for Blantyre
count and FACSCount. Blantyre count and the Blan-
tyre count (percentage) variant could be used inter-
changeably for CD4 percentage with excellent limits
of agreements (table 2, fig 2).
Repeatability of Blantyre count
We calculated coefficients of variation on five repeats
of our assay on four blood samples with mean CD4
values 718, 712, 260, and 191×109/l and mean CD4
percentage 40.3%, 42.9%, 15.0%, and 13.8%. Mean
(SD) coefficients of variation were 5.2% (2.7%) for
absolute CD4 count and 2.5% (0.8%) for CD4 percen-
Accuracy of Blantyre count
In tests on six stabilised blood samples (CD4 count
117-1269×109/l and percentage 7.28%-60.73%) from
NEQAS with our assay, five of six absolute CD4
counts and five of six CD4 percentages were within 1
Table2 | Estimatedbiasandlimitsofagreement,with95%confidenceintervalsforpairsofflowcytometricmethodsusedto
Assay comparisonBias (95% CI)
Limits of agreement
Lower limit (95% CI)Upper limit (95% CI)
Blantyre count and TruCount
−11.0 (−14.9 to −7.1)
1.2 (−2.6 to 4.9)
−12.1 (−16.6 to −7.7)
−5.8 (−9.3 to −2.3)
−48.9 (−55.7 to −42.1)
−35.4 (−41.9 to −28.8)
−54.8 (−62.4 to −47.1)
−39.3 (−45.3 to −33.3)
27.0 (20.2 to 33.8)
FACSCount and TruCount
37.7 (31.2 to 44.3)
Blantyre count and FACSCount
30.5 (22.8 to 38.1)
27.7 (21.7 to 33.7)
Blantyre count and TruCount
−0.03 (−0.24 to 0.19)
−2.42 (−2.78 to −2.05)
−5.83 (−6.87 to −4.79)
−7.56 (−8.57 to −6.54)
−1.35 (−1.82 to −0.88)
2.37 (2.00 to 2.73)
FACSCount* and TruCount0.92 (0.32 to 1.52)
−0.94 (−1.53 to −0.35)
0.01 (−0.26 to 0.28)
7.66 (6.62 to 8.70)
Blantyre count and FACSCount*
5.67 (4.66 to 6.69)
Blantyre count and Blantyre count
1.37 (0.90 to 1.84)
*CD4 percentages with FACSCount calculated from absolute CD4 count by FACSCount and total lymphocyte count from haematological analyser.
BMJ | ONLINE FIRST | bmj.com page 5 of 8
1 and 2 SD of this value for each test. Blantyre count
values were on average 95% of the absolute NEQAS
CD4 count and 97% of the CD4 percentage.
Stability of aged samples
study, with a blood sample with day 1 CD4 count of
487×109/l and CD4 percentage of 36.1%. Daily coeffi-
cients of variation for absolute counts remained below
6% and for CD4 percentage below 2.5%. The mean
absolute CD4 count stayed within 10% and the CD4
percentage within 5% of the day 1 values.
Clinical staging and CD4 counts for new patients attending
antiretroviral therapy clinic
Of the new patients attending the antiretroviral ther-
apy clinic, 76% (193/253) were clinical stage I (n=77)
or stage II (n=116), while 24% (60/253) had stage III
(n=51) or stage IV (n=9) HIV/AIDS. The range of
fall in median CD4 counts from stage I (286×109/l) to
stage IV groups (110×109/l). Twenty five (32%)
patients with stage I disease and 48 (42%) with stage
II disease had a CD4 count <200×109/l. Eleven
(22%) patients with stage III and one (11%) patient
with stage IV HIV/AIDS had a CD4 count
>350×109/l (table 3).
Within Malawi, we have developed an affordable accu-
rate method of counting CD4 cells with flow cytometry
by refining and miniaturising existing technology.
Increasing affordability by reducing reagent costs is a
critical step in making this available in countries with
limited resources. Currently the reagent cost of a com-
98% if only CD4 percentage is required but would
decrease if the costs of competing tests are reduced.
Cost reduction was not achieved at the expense of
accuracy. Over the CD4 count range of 0-400×109/l,
our assay showed minimal bias and excellent agree-
ment compared with established CD4 counting meth-
ods (TruCount and FACSCount). Determination of
CD4 percentage by Blantyre count and TruCount
methods showed excellent agreement over the full
range of CD4 percentages. The good performance of
Blantyre count in the NEQAS immunophenotyping
scheme further shows the accuracy of this method.
in our assay have simplified CD4 counting with flow
cytometry. Use of a primary CD4 gating strategy
avoids extra sub-gating steps involving the CD3 or
CD45 cell populations as performed by other estab-
lished methods, including TruCount/Multitest and
straightforward to train technicians to gate the CD4 T
cells (R1), counting beads (R2), and total lymphocyte
populations (R3) using CD4/side scatter and CD45/
side scatter dot plots (fig 1).
Strengths and weakness of study
Our study validates the use of a simplified, affordable,
and accurate method of CD4 counting with flow cyto-
metry. Unlike many previous studies of affordable
flow cytometry,20-2224we carried out this work in a
country where affordability is of chief importance.
We looked at both absolute and percentage CD4,
which have previously been neglected. The limits of
studies of flow cytometry and narrower than studies
using other methods,15-18which are inherently less
accurate. By miniaturising the present assay, we man-
aged to reduce reagent costs further compared with
Even with the simplifications introduced, however,
CD4 counting with flow cytometry requires a level of
technical skill not always present in resource poor
settings,9a reliable power supply, and a cold chain for
reagent supplies. A flow cytometer represents a large
capital outlay, which is not always feasible, although
donor funding is sometimes available to help provide
Users of FACSCount and other methods that pro-
vide only the absolute CD4 count have needed to use
total lymphocyte counts (usually from haematological
analysers) to obtain the CD4 percentage. The poor
bined FACSCount plus haematological analysis when
compared with CD4 percentage produced by either
Blantyre count or TruCount (table 2, fig 2) shows the
inaccuracy of this laborious “reversed dual platform”
approach. It is well recognised that flow cytometers
and haematological analysers can produce signifi-
cantly different total lymphocyte counts for the same
Table3 | ComparisonofclinicalstagingwithabsoluteCD4countwithBlantyrecountfornewpatientsseenatantiretroviralclinic
Clinical stage No (%) of patients Median (range) CD4×109/l
No (%) with CD4×109/l
I77 (30) 286 (20-1020) 25 (32)25 (32)27 (35)
II116 (46)249 (4-1261) 48 (42)41 (35) 27 (23)
III 51 (20)149 (11-826)29 (57)11 (22)11 (22)
IV9 (4)110 (16-552)7 (78)1 (11)1 (11)
All253239 (4-1261)110 (43) 78 (31)65 (26)
page 6 of 8
BMJ | ONLINE FIRST | bmj.com
TruCount assays indicates that use of CD3 antibody by
be operated on less complex instruments than the
FACSCalibur, deploying only one laser and three
photomultiplier tubes to detect side scattered light and
compared with theFACSCaliburand would besimpler
and less expensive to maintain. Even on five day old
blood, our gating strategy enabled both CD4 T cells
cytes, thereby maintaining good repeatability with little
variability from day one counts.
Blantyrecount could makethe greatest impacton the
care of children under 5 with HIV. Appropriate deter-
mination of CD4 percentage has often been neglected
by investigators seeking to produce affordable CD4
counting.68Determination of CD4 percentage alone
by the Blantyre count (percentage) variant is not only
much cheaper than determining absolute CD4 counts
but also technically easier, as accurate volumetric pipet-
ting and counting beads are not required. CD4 percen-
tages were also more stable than absolute counts over
five days in the same sample.
in the antiretroviral therapy clinic confirms that use of
WHO clinical staging criteria alone for deciding who
should start antiretroviral therapy is suboptimal. About
a third of patients with clinical stage I or II disease who
would not be eligible for antiretroviral therapy on clin-
ical grounds were severely immunosuppressed with a
CD4 count of <200×109/l. Conversely, two fifths of
patients with stage III and IV disease who were eligible
to start antiretroviral therapy had CD4 counts
>200×109/l and a fifth had counts >350×109/l. Clinical
staging alone is missing many patients who urgently
need to start antiretroviral therapy, while some stage
III and IV patients are started on antiretroviral therapy
when treatment could possibly be postponed.
What would it cost?
Consideration of the economic feasibility of using the
of the flow cytometer (about $100000), the annual
service contract (about $10000), and the salary of a
laboratory technician (typical monthly salary $500)
as well as reagent prices. The contribution of these
an instrument spread over 10 years, CD4 counting
with flow cytometry would not be viable if only 200
samples were run on an instrument each month, as
“non-reagent costs” per sample would be $10.83. If,
however, 200 samples are run on a flow cytometer
ment (over 250 days this would be 50000 samples a
year), non-reagent costs are $0.52 per sample, giving
a total assay cost of $0.96. Therefore, use of the Blan-
tyre count method would be most cost effective with a
limited number of flow cytometers operating at high
sample throughput in regional centres and a coordi-
nated system for transporting samples to these centres
from peripheral clinics.
decision making in the treatment of patients with HIV
and service the whole of a country the size of Malawi
using a limited number of instruments in regional cen-
blood stabilising agents such as Transfix, permitting
delays in sending samples to regional centres.29It
remains to be seen whether such a service could be
successfully implemented in such a resource poor
We thank Emily Lifa for blood sampling in the antiretroviral therapy clinic at
Queen Elizabeth Central Hospital and all staff at the antiretroviral therapy clinic
for their assistance with this study.
optimised the Blantyre count method. CAMacL, MKPL, FS, and JB did flow
cytometric assays. CAMacL, SAW, and MKPL analysed the data. All authors
contributed to the writing of the manuscript and approved the final version.
Funding: CAMacL holds a training fellowship in clinical tropical medicine from
the Wellcome Trust (No 067902/Z/02/Z). MKPL received a travel award from
the British Society for Immunology. MEM holds a programme grant from the
Wellcome Trust (No 074124/Z/04/Z).
Competing interests: None declared.
Ethical approval: College of Medicine research and ethics committee.
Peer review and provenance: Non-commissioned, externally peer reviewed.
1 National AIDS Commission. National estimates of HIV/AIDS in
Malawi. Lilongwe, Malawi: National AIDS Commission, 2003. www.
Ministry of Health. HIV Unit, ART quarterly report for March 2007.
Lilongwe, Malawi: Ministry of Health, 2007.
WHO. Interim WHO clinical staging of HIV/AIDS and HIV/AIDS case
definitions for surveillance. Africanregion. Geneva: World Health
Organization, 2005. www.who.int/hiv/pub/guidelines/
Kassa E, Rinke de Wit TF, Hailu E, Girma M, Messele T, MariamHG,
HIV infection and disease in Ethiopia: association between clinical
stages and laboratory markers.AIDS 1999;13:381-9.
measurements,premature antiretroviral treatment and costs in
Thyolo District, Malawi.Trans R Soc Trop Med Hyg 2006;100:24-31.
WHO. CD4+ T-cell enumeration technologies. Technical information.
Geneva: World Health Organization, 2005. www.who.int/
WHAT IS ALREADY KNOWN ON THIS TOPIC
CD4 counting is the main laboratory investigation for monitoring peoplewith HIV but is often
deemed too expensive and too complex to perform in resource poor settings
CD4 counting with flow cytometry can be made more affordable by the use of simple
technical modifications, but CD4 percentages required in children under 5 years and
miniaturisation of blood and reagent volumes have received little attention
WHAT THIS STUDY ADDS
Technical modifications of flow cytometry with miniaturisation can simplify and reduce the
cost of absolute and percentage CD4 counts while maintaining diagnostic accuracy
This CD4 counting method could improve clinical decision making in patients with HIV
disease in settings with limited resources
BMJ | ONLINE FIRST | bmj.compage 7 of 8
7Schoof M. HIV-test makers agree to discounts for poorer nations.
Wall Street Journal14 Jan 2004. www.aegis.com/news/wsj/2004/
UNICEF, UNAIDS, MSF, WHO. Sources and prices of selected
medicines and diagnostics for people living with HIV/AIDS. June
2005 report and December 2006 data. http://www.unicef.org/
HarriesAD,Schouten EJ, Libamba E. Scalingup antiretroviral
treatment in resource-poor settings. Lancet 2006;367:1870-2.
10 Comans-Bitter WM, de Groot R, van den Beemd R, Neijens HJ,
Hop WC, Groeneveld K,et al. Immunophenotyping of blood
lymphocytesin childhood. Reference values for lymphocyte
subpopulations. J Paediatr 1997;130:388-93.
resource-limited settings: towards universal access. Geneva: World
Health Organization, 2006. www.who.int/hiv/pub/guidelines/
12 Brando B, Barnett D, Janossy G, Mandy F, Autran B, Rothe G, et al.
Cytofluorometric methods for assessing absolute numbers of cell
subsets in blood. European WorkingGroup on Clinical Cell Analysis.
13 Whitby L,Granger V, Storie I, Goodfellow K,Sawle A,Reilly JT, et al.
Quality control of CD4+ T-lymphocyte enumeration: resultsfrom the
last 9 years of the United Kingdom national external quality
14 Mandy F, Bergeron M, Houle G, Bradley J, Fahey J. Impact of the
immunological measures relevant to HIV/AIDS: QASI. Cytometry
Validity of an ELISA test for CD4+ T lymphocyte count and validity of
total lymphocyte count in the assessment of immunodeficiency
status in HIV infection. Ann Soc Belg Med Trop 1994;74:61-8.
16 Mwaba P, Cassol S, Pilon R, Chintu C, JanesM, Nunn A, et al. Use of
1-infected patients. Lancet 2003;362:1459-60.
rosetting method for absolute CD4+ lymphocyte countingin
18 Diagbouga S, Chazallon C, Kazatchkine MD, Van de Perre P,
Inwoley A, M’Boup S, et al. Successful implementation of a low-cost
method for enumerating CD4+ T lymphocytesin resource-limited
settings: the ANRS 12-26 study. AIDS 2003;17:2201-8.
19 Janossy G. Dried blood spot technology for CD4+ T-cell counting.
21 Janossy G,Jani I, Göhde W. Affordable CD4+ T-cell counts on ‘single-
platform’ flow cytometersI. Primary CD4 gating. Br J Haematol
CD45 for gating lymphocytes for CD3+CD4+%. Cytometry
23 Schnizlein-BickCT, Mandy FF, O’Gorman MRG, Paxton H, Nicholson
JKA, Hultin LE, et al. Use of CD45 gating in three and four-color flow
cytometric immunophenotyping: guideline fromthe National
Institute of Allergy and Infectious Diseases,Divisionof AIDS.
24 Glencross D,Scott LE, Jani IV, Barnett D, Janossy G. CD45-assisted
PanLeucogating for accurate, cost-effective dual-platform CD4+ T-
cell enumeration. Cytometry 2002;50:69-77.
25 National Statistical Office (NSO) [Malawi], and ORC Macro.Malawi
demographic and health survey 2004. HIV/AIDS prevalence.
Calverton, Maryland: NSO and ORC Macro, 2005.
26 Schnizlein-BickCT, Spritzler J, Wilkening CL, Nicholson JKA,
O’Gorman MRG. Evaluation of TruCount Absolute-Count tubes for
determining CD4 and CD4 numbers in human immunodeficiency
virus-positive adults. Clin DiagnLab Immunol 2000;7:336-43.
27 Lopez A, Caragol I,Candeias J, Villamor N, Echaniz P, Ortuño F, et al.
Enumeration of CD4+ T-cells inthe peripheral blood of HIV-infected
patients: an interlaboratory study of the FACSCount system.
28 Bland JM, Altman DG. Statistical methodsfor assessing agreement
between two methods of clinical measurement. Lancet 1986;i:307-
29 Jani IV, 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
Accepted: 15 June 2007
page 8 of 8
BMJ | ONLINE FIRST | bmj.com