Between 1 January 1993 and 31 December 1998, the paediatric onco-haematology unit of the Armand-Trousseau hospital in Paris followed 122 children living with HIV. A dissertation written within that team, in the unit led by Professors Christian Courpotin and Guy Leverger, analysed retrospectively the 78 complete records of that caseload.
Those six years are not six years like any others. They span precisely the moment when HIV in children stops being a fatal disease and becomes a chronic one. Reading the work today means watching a turning point as it happens.
Who these children were
| Caseload 1993-1998 | 122 children; 18 died, 22 lost to follow-up or followed irregularly; 78 complete records analysed |
|---|---|
| Age | From 18 months to 21 years — mean 8.5 years, median 8 |
| Route of infection | 69 mother-to-child, 5 through transfusion, 4 undetermined |
| Schooling | 97.5% of children over three were in school |
Those five children infected through transfusion, among the oldest in the cohort, recall a history now almost forgotten: before blood products were secured, paediatric HIV was not only a matter of maternal transmission.
What the figures show
Over the six years, 18 children died — 15% of the cohort. All were at the most severe stage of the paediatric classification at death. Age at death traces clearly the double distribution characteristic of the disease in children: seven children aged three or under, nine aged ten or over — the fulminant early forms, and the slow ones.
But the curve bends from 1995. Comparing the end of the period with its beginning is striking:
| 1993-1994 | 1997-1998 | |
|---|---|---|
| Deaths | 6 | 1 |
| Mother-to-child transmission | 15% | 3% |
| Mean length of hospital stay | 14 days | 5.6 days |
| Nutritional support | Frequent | Discontinued |
| Schooling | Interrupted for many | All children in school |
A fivefold fall in transmission in four years, in a single unit: that is the result of trials that had just reported.
The three trials that changed everything
- ACTG 076 / ANRS 024. A Franco-American trial: zidovudine from the fourteenth week of pregnancy, through delivery, then six weeks in the newborn. Result: 8% transmission against 26% on placebo. The founding trial.
- ANRS 075. Opened in 1997, adding lamivudine to zidovudine from the thirty-second week. Result: 2.6% against 6.5% with zidovudine alone.
- PENTA 7. An international protocol treating every infected infant as soon as infection was established, before any clinical sign — the conceptual break with the 1993 recommendations, which reserved treatment for children already symptomatic.
The dissertation also notes, with the caution the period required, the alarm raised by two deaths among uninfected children exposed to prophylaxis, in whom mitochondrial injury had been suspected. The conclusion of the time — maintain prophylaxis while informing mothers — illustrates how health policy is decided when the data are incomplete.
The race through treatment lines
By 1998 no child in the cohort was on monotherapy; a quarter remained on dual therapy, the majority had moved to three drugs or more. Almost all had passed successively through monotherapy, then dual therapy; some were on their third or fourth triple regimen.
The work records a signal that the following years confirmed: the more powerful the combination, the faster it is changed. Mean time on a given regimen was about a year on dual therapy and only seven months on triple therapy. Behind that acceleration lay three new problems that would occupy the next twenty years: resistance, adherence, and sustaining efficacy over time.
Two children, two lessons
The work details two opposite trajectories. The first child responds immediately and durably on all three fronts — clinical, immunological and virological. The second is in apparent failure: the viral load stays poor. And yet the child improves — gastrostomy stopped, growth resumed, back at school, CD4 up from 20 to 400 per cubic millimetre. The conclusion, written in 1999, has not aged: you have to know how to wait, and perhaps you should not chase an undetectable viral load at any cost.
What has changed since
| Late 1990s | Today | |
|---|---|---|
| Who is treated? | Symptomatic or immunosuppressed children first; universal early treatment was still a trial protocol | Every infected child, whatever the age, stage or CD4 count, as early as possible |
| With what? | Nucleoside analogues, the first protease inhibitors and non-nucleosides; multiple syrups, several doses a day | Triple therapy around an integrase inhibitor, dolutegravir, from the first weeks of life and from three kilograms |
| In what form? | Awkward liquid formulations, doses by body surface area | Paediatric dispersible tablets for small weights — the most recently lifted bottleneck |
| Mother-to-child transmission | 26% untreated; 8% then 2.6% with the first prophylaxis | Under 1% when the mother is treated and her viral load undetectable |
| The main problem | Finding effective, tolerable drugs | Getting them to children: just over one child in two is treated worldwide, against nearly eight adults in ten |
What this cohort still says
Two findings from 1999 have not been contradicted. First: a child's adherence does not depend on the child but on the structure around them — and the work notes one child receiving no treatment at all, the mother having refused, the team judging that no treatment was better than poor adherence. Second: children lost to follow-up outnumbered children who died. Twenty-five years on, globally, that is still the weak link.
For the current state of knowledge, see HIV in children and mother-to-child transmission.