CD4 T cell count as predictor of Pneumocystis carinii pneumonia in children born to mothers infected with HIVBMJ 1994; 308 doi: http://dx.doi.org/10.1136/bmj.308.6926.437 (Published 12 February 1994) Cite this as: BMJ 1994;308:437
- D Dunn,
- M L Newell,
- T Ades,
- c Peckham,
- Correspondence to: Dr M-L Newell
- Accepted 28 October 1993
Objective : To assess the value of CD4 T cell count in predicting Pneumocystis carinii pneumonia in infants born to mothers infected with HIV, with reference to the guidelines from the Centers for Disease Control on prophylaxis against pneumocystis.
Design : Prospective birth cohort study.
Setting : Hospitals in 10 European cities participating in the European collaborative study.
Subjects : 924 children born to mothers known to be infected with HIV at or before delivery. Main outcome measures - The incidence of P carinii pneumonia. CD4 T cell counts in children before diagnosis of the pneumonia. The proportions of children infected and uninfected with HIV who fulfilled the criteria for primary prophylaxis.
Results : Fourteen children were diagnosed with P carinii pneumonia. The cumulative incidence by the age of 6 years was 2% (95% confidence interval 0.9 to 3.0%). Of the 11 children with a CD4 T cell count predating diagnosis, only three fulfilled the criteria from the Centers for Disease Control for prophylaxis. Prophylaxis was indicated by 1 year of age for 62% of infected children who had not developed P carinii pneumonia and for at least 10% of uninfected children.
Conclusions : Monitoring CD4 T cell count seems to be of limited value in deciding when to start prophylaxis against P carinii pneumonia in children born to mothers infected with HIV. The alternative approach of giving prophylaxis to all children born to infected mothers would be difficult to justify given the low incidence of the pneumonia.
The Centers for Disease Control recommend prophylaxis against Pneumocystis carinii pneumonia when CD4 T cell count falls below a critical level
In this study of children born to mothers infected with HIV the cumulative incidence of Pneumocystis pneumonia was 16% by 6 years of age in children who had HIV infection
If the guidelines were to be followed prophylaxis would be indicated for at least 10% of uninfected children by 1 year of age
Monitoring CD4 T cell count was of limited value in deciding when to start prophylaxis in children born to mothers infected with HIV
A better strategy would be to start prophylaxis in infants under 6 months of age as soon as HIV infection is diagnosed.
Registry based studies have shown that Pneumocystis carinii pneumonia is the most common indicator disease of AIDS in children infected with HIV.1,2 It does not necessarily follow, however, that children born to mothers infected with HIV are at high risk of P carinii pneumonia. This risk depends on the probability of transmission of HIV from mother to child and the incidence of P carinii pneumonia in infected children, variables which can be reliably estimated only from large, prospective studies.
It is recommended that the CD4 T cell counts of children born to infected mothers are closely monitored and that prophylaxis with trimethoprim-sulphamethoxazole (co-trimoxazole) is started when the count falls below a critical value.3 Although there is strong evidence that the number of CD4 T cells is a reliable indicator of risk for P carinii pneumonia in adults infected with HIV,4 the evidence is weaker in children since this pneumonia most commonly occurs in the early months of life and immunological data are seldom available before its diagnosis.5 An alternative approach would be to give prophylaxis to all children born to infected mothers irrespective of whether they are infected or of their CD4 T cell count.6 The impact of any policy on prophylaxis depends on knowledge that a child is at risk of infection, and widening the provision of antenatal testing has therefore been recommended.7
In the European collaborative study children born to mothers known to be infected with HIV at or before delivery are followed prospectively from birth, with regular determination of CD4 T cell counts.8 These data have been examined to estimate the incidence of P carinii pneumonia in children infected with HIV and in the cohort as a whole and to assess the role of monitoring CD4 T cell counts in primary prophylaxis against P carinii pneumonia with particular reference to the guidelines from the Centers for Disease Control.3 Specifically, we sought to determine how many children who developed P carinii pneumonia fulfilled the criteria for prophylaxis before its diagnosis; the age at which children infected with HIV became eligible for prophylaxis in relation to the age specific incidence of P carinii pneumonia; and the proportion of uninfected children who would have been treated. Subjects and methods
According to the original protocol of the European collaborative study children were examined at birth, every three months up to 2 years of age, and every six months thereafter.8 The follow up schedule recommended in the guidelines from the Centers for Disease Control (see appendix) is similar except that the initial assessment is at 1 month of age and monitoring becomes more intensive when a CD4 T cell count is just above the threshold for prophylaxis. In 1991 the protocol was modified to include more frequent examinations in the first six months of life, but most children described in this report received the less intensive follow up.
Lymphocyte phenotyping and assessment of antibody specific for HIV are performed on blood samples taken at each visit after birth. Clinical and immunological data from additional visits are also recorded. The median age at the initial determination of CD4 T cells was 3.1 months (the 25th and 75th centiles being 2.2 and 4.8 months), with a median interval of 3.0 months (2.5 and 4.1 months) between determinations during the first two years of life. Definition of infection with HIV9 and lymphocyte phenotyping procedures10 have been described elsewhere. Results on lymphocyte subsets in uninfected children were consistent with those reported from studies of normal children, and the extent of systematic variation between laboratories was small compared with variation between children and over repeated measurements within children.10
Some of the findings could be influenced by treatment in infected children. Eight infected children received prophylaxis against P carinii pneumonia (in all cases with trimethoprim-sulphamethoxazole). In the analysis these children were not considered to be at risk of P carinii pneumonia after prophylaxis had been started. This could lead to an underestimation in incidence as they were possibly at higher risk. Any such bias would be marginal, however, and only three of the eight children received prophylaxis during the period of highest risk in the first six months of life. Thirty seven infected children received zidovudine (before developing AIDS), which could have affected CD4 T cell counts. From studies on adults, however, it seems that any effect would probably be small and transient, and only eight children received zidovudine before 6 months of age.
The main aim of the analysis was to examine the predictive value of CD4 T cell count for P carinii pneumonia, but we also report other clinical signs and symptoms that were observed before the diagnosis of the pneumonia by using the Centers for Disease Control's classification of infection with HIV in children.11
The proportion of children, by age, with a CD4 T cell count (either absolute number or as a percentage of total lymphocytes) below the threshold for prophylaxis was derived by the Kaplan-Meier method12 with follow up curtailed at the last CD4 T cell determination. For uninfected children a CD4 T cell count below the threshold after disappearance of transplacentally acquired antibody to HIV was ignored. For infected children CD4 T cell counts are relevant only if measured before the onset of P carinii pneumonia. The analysis, therefore, was restricted to children who had not yet developed P carinii pneumonia by using the conditional probability relationship: probability (CD4 <threshold (?) no P carinii pneumonia)= 1-[1-probability (CD4<threshold or P carinii pneumonia)]/probability (no P carinii pneumonia).
The analyses excluded 16 infected and 73 uninfected children who had not had a CD4 T cell determination in the first year of life.
Incidence of p carinii pneumonia
Between December 1984 and January 1993, 924 children were recruited from 10 centres. Of these, 111 children were infected with HIV, 571 were uninfected, and 242 had an indeterminate state of infection. P carinii pneumonia was diagnosed in 14 children (histologically confirmed in 11). The incidence of P carinii pneumonia was highest between 3 and 6 months of age (table 1). All children in whom the pneumonia had been diagnosed in the first year of life were aged between 2.9 and 6.2 months (table II). The cumulative incidence by 6 years of age was 16.0% (95% confidence interval 7.3% to 23.9%) in infected children and 2.0% (0.9% to 3.0%) in all children born to infected mothers. None of the children with P carinii pneumonia received antipneumocystis or antiretroviral treatment before diagnosis.
CD4 T Lymphocyte count as predictor for p carinii
The following analysis was restricted to 11 of the 14 children for whom at east one CD4 T cell count was available before diagnosis of P carinii pneumonia. Figure 1 shows the CD4 T cell counts around the time of diagnosis for the children who developed P carinii pneumonia in the first year of life. Despite close monitoring, in which all children had lymphocyte phenotyping performed within two and a half months of diagnosis of P carinii pneumonia, only two children had counts which crossed the threshold for prophylaxis of 1500x106 CD4 T cells/l before the pneumonia was diagnosed. Another two children had CD4 counts between 1500x106 cells/l and 2000 x 106 cells/l when last assessed 1.4 and 2.0 months before diagnosis of the pneumonia and, had the guidelines been adopted, their CD4 T cell counts would have been reassessed one month later, which could have triggered prophylaxis
Of the two children who developed P carinii pneumonia after 1 year of age, one had a CD4 T cell count consistently below the threshold for prophylaxis (fig 2). The second child's CD4 T cell count fell sharply during the two months before diagnosis of the pneumonia, but the criteria for prophylaxis were not fulfilled. Distinct decreases in CD4 T cell count before diagnosis of pneumonia were also evident in many of the younger children (fig 1), although these are less obvious because of the narrow time scale.
When the threshold CD4 T cell count was expressed as a percentage of total lymphocytes rather than an absolute number, no additional children were identified but one was missed. In summary, of the 11 children with CD4 T cell determinations predating diagnosis of P carinii pneumonia, only three clearly fulfilled the criteria for primary prophylaxis.
Clinical signs and symptoms as predictors for P carinii pneumonia
Clinical signs and symptoms before the onset of P carinii pneumonia could provide an alternative or complementary basis for prophylaxis (table II). According to the classification of the Centers for Disease Control seven children had no symptoms (P1) before onset of P carinii pneumonia and two others (cases 6 and 7) met the criteria for symptomatic infection at the time of diagnosis of the pneumonia. Thus, only five children had clinical signs and symptoms before the onset of P carinii pneumonia, two of whom were not eligible for prophylaxis on the basis of their CD4 T cell count.
Age when children cross threshold for prophylaxis
Figure 3 shows, by state of infection, the proportion of children at each age who had not developed P carinii pneumonia but whose CD4 T cell count fell below the threshold for prophylaxis. Prophylaxis was indicated for 6% of infected children by 3 months of age, rising to 33% by 6 months and 62% by 12 months. A low CD4 T cell count was observed in 10% of uninfected children before the disappearance of maternal antibody to HIV. These analyses are sensitive to the frequency and ages at which lymphocyte determinations are performed. Under the slightly more intensive schedule recommended in the guidelines from the Centers for Disease Control even higher proportions of children would be anticipated to fulfil the criteria for prophylaxis. This was confirmed by simulation analyses which increased the estimate of the proportion of uninfected children who were eligible for prophylaxis to 18%.
Our findings, based on 111 children infected with HIV and 813 uninfected children all born to mothers infected with HIV suggest that monitoring CD4 T cell count is of limited value in deciding when to initiate prophylaxis against P carinii pneumonia. In only three of the 11 children who developed P carinii pneumonia and who had a CD4 T cell determination before diagnosis was the value below the threshold for prophylaxis. Moreover, 62% of infected children who had not developed P carinii pneumonia had a CD4 T cell count below the threshold for prophylaxis during the first year of life, as did 10% of uninfected children before the disappearance of maternal antibody. These findings confirm previous results.13,14
Guidelines from the centers for disease control
The sensitivity of the guidelines from the Centers for Disease Control may have been overestimated because of the retrospective nature of the data on which they are based. The threshold of 1500x106 CD4 cells/l for children aged 1-11 months was based on the observation that 63 of 71 children diagnosed with P carinii pneumonia in the first year of life had a CD4 T cell count below this value.*RF 5,6,15-18* In most of these cases, however, lymphocyte phenotyping was performed at or after diagnosis, exaggerating the predictive value of CD4 T cell counts. In addition, since the infection with HIV of many of the children came to light only as a result of the episode of P carinii pneumonia delay between initial symptoms and diagnosis of the pneumonia may have been longer than for the closely monitored children in the European collaborative study. Moreover, a higher proportion of the children had other symptoms related to HIV infection before or concurrent with diagnosis of the pneumonia (possibly reflecting selective inclusion of symptomatic children in these series) than in the European collaborative study,5,15 and immunological variables are more likely to be impaired in children with symptomatic than with asymptomatic HIV infection.19
In view of the apparent low sensitivity and specificity of the guidelines consideration could be given to their possible modification.3 One possibility would be to use clinical signs and symptoms instead of or as well as immunological markers. Prophylaxis is advisable for children with serious manifestations related to HIV infection, but we have shown that the absence of signs and symptoms does not preclude the possibility of P carinii pneumonia. Evidence of a rapid decline in CD4 T cell count around the time of onset of the pneumonia in some children points to a possible advantage in increasing the frequency of lymphocyte phenotyping. Indeed, a rapid decline in CD4 T cell count has been suggested as sufficient grounds for therapeutic intervention in adults,20 but large fluctuations of no immunological importance are common, particularly in children, because of large variability within people.10,21 Although higher thresholds for prophylaxis would result in more children receiving treatment before they developed P carinii pneumonia, this would further decrease the specificity of the guidelines.
An alternative approach would be to give prophylaxis to all children infected with HIV,6,13 which would enable prophylaxis to be initiated before the period of highest risk for P carinii pneumonia. Routine early diagnosis of infection with HIV, however, is not yet practicable.22 Adequate protection for infected children would therefore necessitate giving treatment to all children born to infected mothers, including children who are uninfected, at least during the period of high risk for P carinii pneumonia. This would be difficult to justify since the cumulative incidence of this pneumonia by 6 years of age was only 2%.
Several factors need to be balanced against the benefit of delaying the progression of disease in children infected with HIV. These include the theoretical possibility of side effects resulting from early and prolonged treatment23; the cost of the drug and its delivery20; and the need to monitor children for CD4 T cell counts and adverse haematological reactions.3 It is also uncertain for how long prophylaxis should be given and whether it is as effective in preventing P carinii pneumonia in infants as it is in adults.24
Because of the low incidence of P carinii pneumonia no single study can provide a definitive analysis of the predictive value of CD4 T cell counts. It is important that data from other studies be examined to confirm our findings.
Collaborators in the European collaborative study are Dr C Giaquinto, Dr A De Rossi, Professor L Chieco-Bianchi, Professor F Zacchello (Universita degli Studi di Padova); Dr I Grosch-Worner, Dr M Vocks-Hauck, Dr M Langhof (Universitatsklinikum Rudolf Virchow, Berlin); Dr J Mok, Dr G Bird (City Hospital, Edinburgh); Dr F Omenaca Teres, Dr I Bates, Dr M C Garcia-Rodriguez (Hospital Infantil La Paz, Madrid); Dr C Canosa, Dr D M Galbis (Hospital La Fe, Valencia); Dr H Scherpbier, Dr G Mulder-Kampinga, Dr K Boer (Academisch Medisch Centrum, Amsterdam); Dr A B Bohlin, Dr E Belfrage (Huddinge and Danderyd Hospitals, Stockholm); Dr A Ferrazin, Dr C Gotta (Hospital San Martino, Genoa); Dr J Levy, Dr A Alimenti (Hospital St Pierre, Brussels); Dr A Mur, Dr H Yazbeck, Professor J Llorens (Hospital del Mar, Laboratorio Referencia de Cataluna, Barcelona).
We acknowledge the help we have had from Ms J Callis. We thank Dr R D'Elia, Dr A M Laverda, Dr E Ruga, Dr S Cozzani, Dr A Giacomelli, Dr A Pagliaro, Dr A Condini, Dr C Cattelan, Dr A Mazza, Professor B Grella, Dr A R Del Mistro, Dr A Amadori (Padua); Dr A Schafer, Dr B Zorr, Dr M Mielke, Professor U Wahn, Professor B Stuck, Dr J Woweries, Dr H Abdul-Khaliqu (Berlin); Dr R Hague, Dr F D Johnstone, Dr P L Yap, Dr S Burns, Dr J Inglis, Dr J Peutherer, Mrs S Rebus, Mrs F Mitchell, Mrs C Lockhart (Edinburgh); Dr G Fontan-Casariego, Dr R Martinez-Capico, Dr M D Jose, Dr M L Gonzalez, Dr M L Prieto (Madrid); Dr F Asensi, Dr M Otero, Dr D Perez Tamarit, Dr D Granda, Dr G Molina (Valencia); Dr C A B Boucher, Dr M Roos, the participants of the Dutch collaborative study of HIV infected women and their children (Amsterdam); Dr F Melica, Dr C Cirillo, Dr G Barigione, M Bellomo, F Caocci (Genoa); Dr S Sprecher, Prof N Clumeck (Brussels); and Dr E Gimenez, Dr C Sierra (Barcelona).
This study is a concerted action of the European Commission. The Medical Research Council provides support to the coordinating centre. Collaborating centres were supported by grants from the Ministero della Sanita-Istituto Superiore di Sanita, progetto AIDS 1989-1992 (Padua, Genoa); the Medical Research Council, the AIDS Virus Education Research Trust, the Scottish Office Home and Health Department (Edinburgh); Praeventiefonds number 28-1704 (Amsterdam); Bundesminister fur Gesundheit (Berlin); and Fonds de la Recherche Scientifique Medicale (Brussels).