Saturday 2 February

1991

No 8736

ORIGINAL ARTICLES

Children born to
natural

women

history

with HIV-1 infection: and risk of transmission

EUROPEAN COLLABORATIVE STUDY*

600 children born to HIV-infected mothers by June 15, 1990, in ten European centres were followed to study the natural history of HIV infection and the vertical transmission rate. They were seen at birth, every 3 months up to 18 months of age, and every 6 months thereafter. At last follow-up, 64 children were judged to be HIV infected and 343 had lost antibody and were presumed uninfected. The initial clinical feature in infected children was usually a combination of persistent lymphadenopathy, splenomegaly, and hepatomegaly, though 30% of children presented with AIDS, or with oral candidosis followed rapidly by AIDS. An estimated 83% of infected children show laboratory or clinical features of HIV infection by 6 months of age. By 12 months, 26% have Al DS and 17% die of H IV-related disease. Subsequently, the disease progresses more slowly and most children remain stable or even improve during the second year. The vertical transmission rate, based on results in 372 children born at least 18 months before the analysis, was 12·9% (95% Cl 9·5-16·3%). Virus has been repeatedly isolated in an additional small proportion of children (2·5%, 95% Cl 0·7-6·3%) who lost
maternal

specialist centres.l-6 Prospective studies of vertically infected children should provide a more accurate understanding/-9 though increasing use of treatment could bias results.
Information about earlier or less severe features of the disease and its progression is needed for the design of clinical trails, and to assist in management and early diagnosis of infected children. The European Collaborative Study (ECS) is a prospective study of children born to HIVinfected mothers .10-12 We here report results relating to the natural history of HIV infection and the vertical transmission rate.

Children and methods

By June 15, 1990, 600 children bom to HIV-1 seropositive mothers had been recruited and followed from birth in ten

European
were

(Padua, Berlin, Edinburgh, Madrid, Valencia, Amsterdam, Stockholm, Genoa, Brussels, Barcelona). Children
centres

included only if their mothers were known to be infected at or before delivery; all such children were included.1O-12

*Prepared by
Peckham,
MD

Dr A. E. Ades, PhD, Dr M L.

Newell, MB, Prof C. S.

(coordinating centre), Department of Paediatric Epidemiology, Institute of Child Health, 30 Guilford Street,
London WC1 N 1 EH, U K. Collaborators Dr C. Giaquinto, Prof F. Zacchello, Dr A. De Rossi, Prof L. Chieco-Bianchi (Universita degli Studi di Padova, Italy); Dr I. Grosch-Wörner, Dr S. Koch (Freie Universitat, Universitätsklinikum Rudolf Virchow, West Berlin), Dr J Y Q. Mok, Dr R Hague (City Hospital, Edinburgh, UK), Dr F. Omeñaca Teres, Dr M. C. Garcia Rodriguez, Dr R Martinez-Zapico, Dr M. I. de Jose, Dr G. Fontan Casariego (Hospital Infantil La Paz, Madrid, Spain), Dr C. A Canosa, Dr M. C Otero, Dr D. Perez Tamarit, Dr D Granda, Dr M Gobernado, Dr A. Gonzalez Molina (Departamento de Pediatria "La Fe", Valencia, Spain); Dr H. J Scherpbier, (Academisch Medisch Centrum, Amsterdam, the Netherlands), Dr A B. Bohlin, Dr M. Forsgren (Department of Paediatrics, Huddinge Hospital and Central Microbiological Laboratory, Stockholm, Sweden), Dr A Ferrazin, Dr A De Maria, Dr C Gotta, Prof A Terragna (I Clinica Mallattie Infettive Università, Istituto G. Gaslini, Genova, Italy), Dr J. Levy (Hospital St Pierre, Brussels, Belgium); Prof J. Llorens-Terol, Dr A. Mur, Dr H Yazbeck, Dr F. Prats (Hospital del Mar,

antibody and have remained clinically and immunologically normal. Without a definitive

virological diagnosis, the monitoring of immunoglobulins, CD4/CD8 ratio, and clinical signs could
identify HIV infection in 48% of infected children by 6 months, with a specificity of more than 99%.
Introduction

Knowledge of the natural history of vertically acquired human immunodeficiency virus-1(HIV-1) is based mainly on studies of children presenting with symptoms to

Barcelona, Spain).

Correspondence to Prof C. S. Peckham.

258

recognise 48% of infected infants by 6 months. None of the 325 children who lost antibody and who were followed for
over

12 months met this criterion in the first 12 months. The

specificity is therefore estimated to be 100% and with this sample size is unlikely to be less than 99-1% (lower 90% confidence limit).
Vertical transmission rate
Of 419 children born 18 months or more before June 15, 1990, 372 were of known infection status and of these 48 were infected (table ]). This represents a transmission rate of 12-9% (95% CI 9-5-16-3%), which agrees closely with the estimate of the surviving fraction of 13-4% in fig 2. There were no statistically significant differences in transmission rates between centres, nor were there changes over time. Of the 353 children who lost antibody, virus isolation had been attempted in 163; in 10 (6-1%), positive results were obtained at least once. With the exception of 1 child in whom a static encephalopathy uncharacteristic of HIV developed following a Reye-like syndrome/2,15 these 10 children did not seem to differ clinically or immunologically from the other 343 children who lost antibody. 1 child had hyperIgG, another a low T4/T8 ratio, and a third lymphadenopathy and hepatomegaly. However, these signs all resolved within a few months, and taken together they represent an incidence of the same order as would be expected in children who lost antibody. In 4 children the initial positive culture was confirmed on subsequent samples by repeat virus culture and/or polymerase chain reaction (PCR). But in the remaining 6, a total of eighteen attempted cultures were negative, and laboratory error or contamination cannot be completely ruled out. The data do not, therefore, support an estimate greater than 4/163 (2-5%,95% CI 0-7-6-3%).

infected, this implies a positive predictive value of only 80%, and may not therefore be an indication for therapies with side-effects. A combination of any two of hyperIgG,A,M, low T4/T8 ratio, or HIV-ISS was highly specific
are

and could be used as a criterion for treatment or inclusion in a therapeutic trial. However, these findings should not be applied to children whose mothers are not known to be positive, because immunological signs could have very low predictive value in the general population. Moreover, our findings would not be applicable in countries where hyper-IgG, lymphadenopathy, and hepatosplenomegaly, due to chronic infections are common. 27 The relative frequency of AIDS indicator diseases in the ECS is similar to earlier reports.7.2028 The possible association between failure-to-thrive and encephalopathy would point to a central nervous system mechanism for failure-to-thrive that is different from the opportunistic diarrhoeal infections that probably underlie failure-to-thrive in African studies.

Our findings have implications for clinical management and for the design of clinical trials. Firstly, treatment will have to be started very early-within the first 2 or 3 months-to prevent the early onset of AIDS in about 30% of infected children.7 Except in laboratories with high expertise, such an early diagnosis would be hard to achieve at present. Also, during the second and third years of life, the progression of HIV disease seems to be slow, with more children improving than deteriorating. Long-term treatment of symptom-free children or those with mild symptoms may be necessary to show any benefit; this not only creates difficulties in the design of trials for treatments with side-effects, but also raises doubts about earlier, open therapeutic studies that attributed to treatment improvements that might have occurred spontaneously.z9.3o
Our 13% estimated transmission rate is lower than rates published in all but one study,31 which was a subset of the ECS. It is possible that a higher frequency of risk factors such as breastfeeding or mothers clinical condition could account for these differences. However, estimates in many earlier studies may have been biased upwards. For example, in one study, virus cultured from cord blood was taken as proof of infection.22 In a study citing 39% transmission, 47% of the cohort had been lost to follow-up;24 and those remaining may have been more likely to have been ill. A 29% rate was cited in a New York Study.32 However, that analysis included antibody-negative children only if they had been followed for 15 months, whereas infected children were included with shorter follow-up. In studies in which some children are recruited retrospectively, the requirement that the mother should be seropositive may be insufficient to prevent a bias towards selective recruitment of sick children. A further requirement is that collaborating centres report all children born to seropositive mothers. This type of bias may have occurred in the Italian Multicentre Study" (33% / transmission), which is register based; in the French Study, which has 51 collaborating centres (27% transmission); and also in earlier reports from the ECS.lO,l1 The requirement for inclusion in the transmission rate that children should have an antibody test after 18 months"rather than have a birth date 18 months earlier also biases estimates upwards because parents of children who lose antibody earlier are less likely to bring the children back for continuing follow-up. Even if all the children who were lost to follow-up were infected, the transmission rate would still only be 23%. Deaths in children of indeterminate infection status, of which there were 7 in the ECS, also make the calculations of

Discussion

description of the natural history of HIV-1 infection life, we believe that the ECS is substantially unbiased by treatment. While confirming the wide spectrum of disease found by other investigators,-, 9,16-23 the ECS provides additional information on the timing of initial symptoms, and the frequencies of clinical and immunological features. Birthweight, gestation, head circumference at birth, or birthweight adjusted for gestation did not discriminate between infected and uninfected children; this also confirms earlier results.7,1l The lower birthweight of children born to women with AIDS or with symptomless HIV infection than of those born to seronegative women reported in two African-based studies2224 is more likely to be
As
a

up to diagnosis of AIDS in the first 3 years of

due to differences in the mothers socioeconomic or health circumstances than to interuterine infection, since these findings have not been seen in New York 25 or Scotland.26 Our findings show that without a definitive diagnosis, immunological and clinical signs can be regarded as predictors of HIV infection in children under 18 months old. Although several signs and symptoms are associated with HIV infection, the ECS shows that many are not specific enough to be useful markers either of infection or of the development of AIDS. Otitis media, rhinitis, noncryptosporidium diarrhoea and unexplained fever were all poor indicators of infection compared with hyperIgG,A,M, which identified 77% of infected children at 6 months with 97% specificity. Inapopulation of whom 15%