Predictive value of absolute CD4 cell count

for disease progression in untreated

HIV-1-infected children
HIV Paediatric Prognostic Markers Collaborative StudyM
Objectives: To describe the relationship between absolute CD4 cell count and the
short-term risk of disease progression in HIV-1-infected children.
Design: A meta-analysis of individual longitudinal data on HIV-1-infected children
enrolled in trials and cohort studies in Europe and the USA.
Methods: The risks of progression to death and AIDS (or death) within 12 months, in
terms of age and the most recent CD4 cell count, were estimated using parametric
survival models. The analysis was restricted to measurements before the start of
antiretroviral therapy except zidovudine monotherapy. The values of the absolute
CD4 cell count and percentage predicting selected levels of disease progression risk
were determined from this and previous models.
Results: A total of 566 deaths was observed over 9128 person-years of follow-up, and
992 children progressed to AIDS or death over 7309 person-years of follow-up. In
children older than 4 or 5 years, the estimated risk of disease progression increased
sharply when the CD4 cell count fell below 200300 cells/ml. As with other immunological markers, CD4 cell count was less prognostic in younger children. The CD4 cell
count values predicting a 12-month risk of death of 25% and of AIDS of 510% were
much more strongly influenced by age than equivalent CD4 cell percentage values.
Conclusion: This study suggests it may be appropriate to extend CD4 cell count criteria
for initiating antiretroviral therapy in HIV-1-infected adults to children as young as 4 or
5 years. Monitoring by CD4 cell count in younger children is problematical because age
2006 Lippincott Williams & Wilkins
is a highly influential variable.

AIDS 2006, 20:12891294

Keywords: CD4, epidemiology, natural history, paediatrics, prognosis

Introduction
In HIV-1-infected children, disease progression and
response to antiretroviral therapy (ART) have traditionally been monitored using CD4 cell percentage (i.e. the
number of CD4 cells expressed as a percentage of total
lymphocytes) rather than the absolute CD4 cell count,
the marker generally used in adults. This is primarily
because of the large natural decline in CD4 cell count in
early life [1]. In line with clinical practice, most paediatric
studies of prognostic markers for clinical HIV disease
progression have focussed on CD4 cell percentage, and
comparatively little information is available on CD4 cell
count [2]. It is important to fill this gap in the current

knowledge for two reasons. First, many of the cheaper

CD4 cell assays being developed for use in resourcelimited settings estimate the absolute CD4 cell count but
not the CD4 cell percentage [3,4]. Paediatricians in these
settings may therefore have no choice but to monitor HIV
infection with the CD4 cell count, or else to embark on a
more complex course involving the separate measurement of differential white blood cell counts and deriving
the CD4 cell percentage from the CD4 cell count and
total lymphocyte count. Second, in Europe and the
United States increasing numbers of perinatally infected
children are surviving into late childhood and adolescence [5,6]. Treatment guidelines suggest that CD4 cell
count as well as CD4 cell percentage should be considered

Correspondence and reprint requests to David Dunn, MRC Clinical Trials Unit, 222 Euston Road, London NW1 2DA, UK.
Tel: +44 0207 670 4739; fax: +44 0207 670 4815; e-mail: d.dunn@ctu.mrc.ac.uk


See Appendix for details of Steering Committee.
Received: 14 December 2005; revised: 13 February 2006; accepted: 24 February 2006.

ISSN 0269-9370 Q 2006 Lippincott Williams & Wilkins

1289

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1290

AIDS 2006, Vol 20 No 9

when deciding when to initiate ART in older children or

adolescents [7,8]. However, it has not been established
whether the CD4 cell count levels for initiating ART in
adults are also appropriate for older children, nor, if this is
confirmed, the age from which they should be used. To
inform these issues we have examined the relationship
between absolute CD4 cell count and the short-term risk
of disease progression in a large group of prospectively
followed HIV-infected children.

exp(k.CD4). The parameters a, b, and k were allowed to

depend on age: loge (a) a1 (a2.age), loge (b)
b1 (b2.age), loge (k) k1 (k2.age). The fit of the
model was significantly improved by the age transformation: 4 0.1
(age  4) if  4 years. Model validity was
assessed through plots comparing cumulative numbers
(by CD4 cell count) of predicted and observed events.
Models that also included study as a factor showed that
this variable did not confound the relationship between
disease progression risk and the joint effects of age and
CD4 cell count (not shown).

Methods
The organizers of all major European and USA cohort
studies and trials of HIV-1-infected children with data
before the introduction of combination ART were
invited to participate in the HIV Paediatric Prognostic
Markers Collaborative Study (HPPMCS). Longitudinal
data were received and pooled from a total of 17 studies
(five birth cohort studies, three general cohort studies,
nine randomized trials), covering the period 19832002.
Full details of the individual studies have been described
elsewhere [9]. Variables provided by the studies included
dates of initial AIDS diagnosis, last clinical examination,
death, and when last known to be alive; dates when
zidovudine and any other antiretroviral drugs were first
started; and CD4 cell count and percentage. Details of
immunological methods used were not available, but it is
presumed that most or all CD4 cell counts were derived
from dual-platform methodology, i.e. multiplying CD4
cell percentage obtained from flow cytometry and total
lymphocyte count obtained from a haemocytometer.
Separate analyses were conducted to estimate the 12month risk of progression to death and the 12-month risk
of progression to AIDS or death in the absence of effective
ART. To minimize presentation bias, children in nonbirth cohort studies were excluded from the analysis if
they experienced an endpoint within one month of
diagnosis of HIV infection. To take account of the
multiple CD4 cell measurements per child, each value
contributed a unit of observation to a survival analysis, i.e.
the timescale was reset to 0 at each new measurement,
with current age and CD4 cell count defined as the
baseline covariates. Follow-up was censored at the earliest
of the following: date of last clinical visit (for analyses of
AIDS) or date last known to be alive (for analyses of
death); 12 months after the last measurement of CD4 cell
count; 6 months after starting any antiretroviral drug
other than zidovudine monotherapy, which has a
marginal clinical effect [10]. The rationale for the
6-month extension was to reduce bias resulting from
the selective use of combination therapy in children with
a poor prognosis [9].
Disease progression estimates were derived from exponential survival models with a hazard rate l a b.

Results
A total of 3944 children with a median of six
[interquartile range (IQR) three to 10] CD4 cell count
measurements per child were included; the median
interval between successive measurements was 2.8
months (IQR 1.84.6). A total of 566 deaths were
observed over 9128 person-years of follow-up (47% during
zidovudine monotherapy), and 992 children progressed
to AIDS or death over 7309 person-years of follow-up
(44% during zidovudine monotherapy). Data after the age
of 10 years were infrequent, comprising approximately
5% of total follow-up. The distribution of the CD4 cell
count was highly age-specific: the proportion of
measurements less than 1000 cells/ml increased from
28% at 6 months, to 34% at one year, 48% at 2 years, 79%
at 5 years, and 97% at 10 years; the corresponding
proportions less than 500 cells/ml were 13, 16, 21, 30, and
69%.
The estimated 12-month risks of death and AIDS
according to current CD4 cell count and age shows two
main features (Fig. 1a,b). First, younger children
experience a greater risk of disease progression than
older children at the same level of CD4 cell count,
although this effect is much less pronounced after
approximately 4 years of age, particularly for mortality.
Poisson regression models restricted to this older age
group, adjusting for study and calendar year, showed that
the effect of age was statistically non-significant for
progression to death (P 0.8) and for progression to
AIDS (P 0.2), although the power of this analysis is
limited by the relatively small number of events
(147 deaths and 269 AIDS diagnoses after age 4 years,
of which only 20 and 141, respectively, were immediately
preceded by a CD4 cell count above 100 cells/ml).
Second, the increase in the risk of disease progression
with declining CD4 cell count occurs much more sharply
for older children than for younger children, the former
having a low and stable risk above a CD4 cell count of
200300 cells/ml. For example, the estimated 12-month
risk of AIDS for a 10-year-old child increases from 2.9%
at 400 cells/ml, 3.3% at 300 cells/ml, 5.3% at 200 cells/ml,

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Predictive value of CD4 cell count in children HIV Paediatric Prognostic Markers Study
(a)

the risk for a CD4 cell percentage of 15% is slightly lower,

ranging from 0.5 to 1.4%. At intermediate ages (15
years) the age gradient is much steeper for a fixed level of
CD4 cell count (500 cells/ml) than for the CD4 cell
percentage (15%); a similar pattern is seen if alternative
CD4 cell count/percentage values are used. This
difference between the two markers can be explained,
at least partly, by the much larger decline in the absolute
CD4 cell count than in the CD4 cell percentage in young
normal children [1].

50

Probability (%)

40

30

20

10

0
0

1000

2000

3000

CD4 cell count (cells/l)

(b)
80
70
60

Probability (%)

1291

50
40
30
20
10
0
0

1000

2000

Another way of showing these data, which has been used

in guidelines to justify the timing of therapy initiation, is
to compare the estimated values of CD4 cell percentage
and CD4 cell count that relate to a specified level of risk
[7]. Table 1 shows the 12-month risk of death of 2 and 5%
and of AIDS of 5 and 10%, and further illustrates the
strong influence of age on these values. Changes are
particularly striking for the CD4 cell count at younger
ages. For example, for a 5% risk of death, the CD4 cell
count threshold would need to change from 1162 cells/ml
at one year to 295 cells/ml at 3 years, a fourfold reduction.
In contrast, the reduction for the CD4 cell percentage
would be only 1.8-fold, from 24 to 13%. The change
would be similar for the two markers between the ages of
4 and 10 years (1.5-fold reduction for CD4 cell count
from 153 to 103 cells/ml; 1.6-fold reduction for CD4 cell
percentage from 11 to 7%). It should be noted that at
certain ages the specified level of risk is not attainable at
any value of CD4 cell count or percentage, for example, a
5% risk of AIDS below the age of 2 years. This is a
consequence of the intrinsically weak prognostic value of
both markers in infancy.

3000

CD4 cell count (cells/l)

Fig. 1. Estimated 12-month risk of (a) death (b) progression

to AIDS by current CD4 cell count at selected ages. 6 months
(highest risk), 1 year, 18 months, 2 years, 3 years, 4 years, 10
years (lowest risk). Curves are truncated at the 5th and 95th
centiles for that age (age 6 months extends to 4100 cells/ml,
age 1 year to 3500 cells/ml, age 18 months to 3200 cells/ml).

and 14.9% at 100 cells/ml. The corresponding values for

the 12-month risk of death are 0.2, 0.3, 1.0, and 5.3%.
The Centers for Disease Control and Prevention have
defined severe immune suppression in HIV-1-infected
children in terms of the CD4 cell percentage (< 15% at all
ages) and CD4 cell count (< 750 cells/ml for < 1 year,
< 500 cells/ml for 15 years, < 200 cells/ml for 612
years) [11]. Figure 2 shows the estimated 12-month risk of
death at these levels of CD4 cell count and CD4 cell
percentage. In infancy, the pattern is similar for both
markers, a very high risk after birth, declining rapidly to a
moderately high risk (12%) at the age of one year.
Between 6 and 12 years of age, a CD4 cell count of 200
cells/ml predicts a 12-month mortality risk of 0.72.1%;

Discussion
Treatment guidelines for HIV-1-infected adults currently
recommend that ART should be deferred in asymptomatic individuals until the CD4 cell count declines to
200350 cells/ml [12,13]. The rationale for this
recommendation is the very low short-term risk of
AIDS above this range of values in untreated individuals,
coupled with increasing concerns about the long-term
toxicity of ART [14]. A key finding from our analysis is
the demonstration of a similar CD4 cell count threshold
effect in children older than 4 or 5 years, consistent with
the results of a previous smaller study that found a
threshold of 200 cells/ml in children older than 6 years
[2]. Although different AIDS case definitions apply for
HIV-1-infected adults and children, it is nonetheless of
interest to compare its frequency of occurence between
the two groups. An annual AIDS incidence of
approximately 4% among untreated adults with a CD4
cell count of 200350 cells/mm3 was reported from a
large cohort of seroconverters [15]. This is comparable to
the level observed among older children in the present

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1292

AIDS 2006, Vol 20 No 9

Probability of death (%)

30

25

20

15

10

0
0

10

Age (years)

Fig. 2. Estimated 12-month risk of death at CD4 cell count and percentage levels that define severe immune suppression.
CD4 cell percentage: < 15%.
CD4 cell count (cells/ml): < 750 (< 1 year), < 500 (15 years), < 200 (612 years).
From Centres for Disease Prevention and Control [11].

study; for example, the estimated annual incidence for a

10-year-old child is 5.3%, and 3.0% at a CD4 cell count of
200 cells/ml and 350 cells/ml, respectively.

age 55 years than at age 25 years [15]. However, it is

noteworthy that adult guidelines do not emphasize this
substantial age gradient in adulthood.

Evidence from this study therefore suggests that it may be

appropriate to extend CD4 cell count criteria for
initiating ART in HIV-1-infected adults to children,
possibly as young as 4 or 5 years of age, i.e. considerably
earlier than currently suggested by European and USA
paediatric guidelines [7,8]. There was some indication
that the risk of disease progression at a given level of CD4
cell count may continue to decline in older children, but
this was not conclusive. In adults, the risk increases rather
than decreases with age; controlling for the CD4 cell
count, AIDS incidence is approximately twice as high at

One approach in deciding marker thresholds for ART

initiation is to choose values that predict a specified
level of risk of disease progression. As indicated by
Fig. 2 and Table 1, this would require, at minimum, a
lowering of the CD4 cell count threshold at each
birthday in the first few years of life. This is likely to be
too complex for routine clinical practice, particularly in
resource-limited settings. Monitoring with the CD4
cell percentage allows the use of wider age bands
because its association with disease progression risk is
less age dependent. The interpretation of longitudinal

Table 1. Values for CD4 cell percentage and CD4 cell count corresponding to selected levels of 12-month risk of death or AIDS.
5% Risk of death
Age (years)
0.5
1
1.5
2
2.5
3
4
5
6
7
8
9
10

2% Risk of death

10% Risk of AIDS

5% Risk of AIDS

CD4 cell %

CD4 cell count

CD4 cell %

CD4 cell count

CD4 cell %

CD4 cell count

CD4 cell %

CD4 cell count

36
24
19
16
14
13
11
10
9
8
8
7
7

1747
1162
808
572
410
295
153
144
135
126
118
110
103

31
25
21
19
16
15
13
12
11
11
10

1451
899
618
439
230
216
203
190
179
168
158

39
28
23
20
17
15
13
11
10
10
9
8

3059
1479
940
628
428
205
191
178
165
154
143
133

39
30
23
19
17
15
14
13
12

1124
333
307
283
262
242
224
208

A missing value indicates that the estimated risk always exceeds the given level regardless of the marker value. The values in this Table are not
intended for clinical application; they are considerably lower than the treatment initiation thresholds recommended in clinical guidelines.

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

Predictive value of CD4 cell count in children HIV Paediatric Prognostic Markers Study

measurements within an individual child is also more

problematical for CD4 cell count because of the large
natural decline in this parameter over the first 5 years of
life [1].
Standard technology for CD4 cell count measurements
currently remains impractical in many resource-limited
settings because of issues of cost and technical expertise.
Given that many alternative low-cost assays enumerate
the CD4 cell count but not the CD4 cell percentage
[3,4], the risk estimates presented here provide a context
for deciding when to initiate ART when only CD4 cell
count measurements are available. A research priority is
to validate CD4 cell count levels for initiating ART
using data from cohorts of untreated HIV-1-infected
children in resource-limited settings. Finally, an
unresolved question for developed countries is whether
CD4 cell count or CD4 cell percentage should be the
preferred marker for monitoring, particularly in older
children, or whether it may be advantageous to consider
both markers in conjunction. The observation that the
total lymphocyte count is highly prognostic suggests that
the CD4 cell count may be the more powerful marker
[16], and this is currently being investigated in further
analyses.

References
1. Wade AM, Ades AE. Incorporating correlations between measurements into the estimation of age-related reference ranges.
Stat Med 1998; 17:19892002.
2. Palumbo P, Raskino C, Fiscus S, Pahwa S, Fowler MG, Spector
SA, et al. Predictive value of quantitative plasma HIV RNA and
CD4 lymphocyte count in HIV-infected infants and children.
JAMA 1998; 279:756761.
3. Forum for Collaborative HIV Research. QA/QC of CD4 and
viral load assays in the resource limited setting. Available
at: www.hivforum.org/publications/QAQC.pdf. Accessed: 12
December 2005.
4. World Health Organization. CD4 T-cell enumeration technologies. Technical information. Available at: http://www.who.int/
diagnostics_laboratory/publications/en/cd4_is_draft.pdf. Accessed: 12 December 2005.
5. Lindegren ML, Byers R, Bertolli J, Dominguez K, Thomas P.
Increasing numbers of adolescents living with perinatal HIVinfection in the United States. In: XIIIth International AIDS
Conference. Durban, South Africa, 914 July 2000 [Abstract
TuPeC3351].
6. Foster CJ, Lyall EGH, Doerholt K, Duong T, Tookey PA, Sharland
M, et al., on behalf of the Collaborative HIV Paediatric Study
(CHIPS). HIV infected adolescents: an evolving UK cohort. In:
British HIV Association 10th Anniversary Conference. Cardiff,
UK, 2004 [Abstract O17].
7. Sharland M, Blanche S, Casltelli G, Ramos J, Gibb DM, on
behalf of the PENTA Steering Committee. PENTA (Paediatric
European Network for Treatment of AIDS) guidelines for the
use of antiretroviral therapy. HIV Med 2004; 5 (Suppl. 2):61
86.
8. The Working Group on Antiretroviral Therapy. Guidelines for
the use of antiretroviral agents in pediatric HIV infection. Available at: http://www.aidsinfo.nih.gov/guidelines/pediatric/PED_
032405.pdf. Accessed: 12 December 2005.
9. HIV Paediatric Prognostic Markers Collaborative Study. Shortterm risk of disease progression in HIV-1-infected children
receiving no antiretroviral therapy or zidovudine monotherapy: a meta-analysis. Lancet 2003; 362:16051611.

1293

10. Paediatric European Network for Treatment of AIDS (PENTA).

Five-year follow-up of vertically HIV-infected children in a
randomised double blind controlled trial of immediate versus
deferred zidovudine: the PENTA 1 trial. Arch Dis Child 2001;
84:230236.
11. 1994 Revised classification system for human immunodeficiency virus infection in children less than 13 years of age.
MMWR Morb Mortal Wkly Rep 1994; 43:No.RR-12.
12. The Department of Health and Human Services. Guidelines
for the use of antiretroviral agents in HIV-1-infected adults
and adolescents. Available at: http://www.aidsinfo.nih.gov/
guidelines/adult/AA_102904.pdf. Accessed: 12 December
2005.
13. The British HIV Association (BHIVA). Guidelines for the treatment of HIV-infected adults with antiretroviral therapy. Available at: http://www.bhiva.org/guidelines/2003/hiv/index.html.
Accessed: 12 December 2005.
14. Phillips AN, Cozzi Lepri A, Lampe F, Johnson M, Sabin CA.
When should antiretroviral therapy be started for HIV infection? Interpreting the evidence from observational studies.
AIDS 2003; 17:18631869.
15. CASCADE Collaboration. Short-term risk of AIDS according to
current CD4 cell count and viral load in antiretroviral drugnaive individuals and those treated in the monotherapy era.
AIDS 2004; 18:5158.
16. HIV Paediatric Prognostic Markers Collaborative Study. Use of
total lymphocyte count for informing when to start antiretroviral therapy in HIV-infected children: a meta-analysis of longitudinal data. Lancet 2005; 366:18681874.

Appendix: Steering Committee

D.T. Dunn, D.M. Gibb, T. Duong, A.G. Babiker (Medical
Research Council Clinical Trials Unit, London, UK); J.P.
Aboulker (INSERM SC10, Villejuif, France); M. Bulterys
(Division of HIV/AIDS Prevention, Centers for Disease
Control and Prevention, Atlanta, GA, USA; Perinatal AIDS
Collaborative Transmission Study); M. Cortina-Borja
(Institute of Child Health, University College London,
London, UK; European Collaborative Study); C. Gabiano
(Department of Pediatrics, University of Turin, Turin, Italy;
Italian Register for HIV Infection in Children); L. Galli
(Department of Pediatrics, University of Florence, Florence,
Italy; Italian Register for HIV Infection in Children); C.
Giaquinto (Department of Pediatrics, University of Padova,
Padova, Italy; Paediatric European Network for Treatment
of AIDS; PENTA); D.R. Harris (Westat, Rockville, MD,
USA; National Institute of Child Health and Human
Development; NICHD Intravenous Immunoglobulin
Study Group); M. Hughes (Harvard School of Public
Health, Boston, MA, USA; Pediatric AIDS Clinical Trials
Group; PACTG); R. McKinney (Duke University Medical
Center, Durham, NC, USA; PACTG); L. Mofenson
(National Institute of Child Health and Human Development (NICHD), National Institutes of Health, Rockville,
MD, USA; NICHD Intravenous Immunoglobulin Study
Group); J. Moye (NICHD; Women and Infants Transmission Study); M.L. Newell (Institute of Child Health,
University College London, London, UK; European
Collaborative Study); S. Pahwa (North ShoreLIJ Research
Institute, Manhasset, NY, USA; PACTG); P. Palumbo
(UMDNJ Medical School, Newark, NJ, USA; Perinatal
AIDS Collaborative Transmission Study); C. Rudin
(University Childrens Hospital, Basel, Switzerland; Swiss

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.

1294

AIDS 2006, Vol 20 No 9

Mother and Child HIV Cohort Study); M. Sharland
(St Georges Hospital Medical School, London, UK;
Collaborative HIV Paediatric Study [CHIPS] of UK and
Ireland); W. Shearer (Baylor College of Medicine,
Houston, TX, USA; Pediatric Pulmonary and Cardiovas-

cular Complications of HIV Infection Study); B. Thompson (Clinical Trials and Surveys Corp, Baltimore, MD,
USA; Women and Infants Transmission Study); P. Tookey
(Institute of Child Health, University College London,
London, UK; CHIPS).

Copyright Lippincott Williams & Wilkins. Unauthorized reproduction of this article is prohibited.