Early infant diagnosis for HIV: is it taking place early enough?

Current early infant diagnosis (EID) protocols may need to be revised in the light of current WHO guidelines on the prevention and treatment of HIV-infection in low-resourced settings, according to Dr. Gayle Sherman of Wits University, South Africa, and senior specialist in Paediatric Haematology with the country's National Health Laboratory Services, speaking at the 19^th Conference on Retroviruses and Opportunistic Infections held last week in Seattle.

Dr Sherman described the dramatic advances that have been made in the field over the last several years, but noted that advances in both technology and guidelines have presented new challenges.

“Six weeks may not be the right time to be testing,” she said, referring to the current protocol for performing EID at the six week immunisation visit. Testing at this time, she went on,  delivers diagnoses a bit too late to take full advantage of lifesaving early antiretroviral therapy (ART) for infected infants, and does not account for the effects that prolonged daily nevirapine prophylaxis could have on diagnostic accuracy (see below).

“We’ve come a very long way in a very short space of time but it’s very difficult to complete a journey when we keep changing the maps: in the forms of technologies and guidelines,” Dr Sherman said.

From serial antibody screening to large-scale PCR tests in less than a decade

HIV-exposed infants carry their mother’s antibodies to the virus for months after childbirth or after their last exposure via breast milk, so standard HIV tests (ELISA tests that look for HIV-antibodies) cannot be used to make a positive HIV diagnosis in young infants. However, HIV antibody tests can be useful for detecting if an infant has been exposed to HIV or to show if an infant is HIV-negative, or if an infant who previously had antibodies seroreverts — in other words, loses those antibodies, when the HIV tests are repeated serially in infants.

If antibodies persist well beyond the last exposure, a positive diagnosis can also be made, typically around the time of 18 months. Since many HIV-infected children don’t survive to this point without treatment, as demonstrated by the CHER study, final results of which were presented on Tuesday at the conference, in 2004, WHO began recommending PCR testing to make an earlier diagnosis in children.

At the time, prevention of HIV transmission to infants was very different. With single-dose nevirapine (sdNVP) having recently been introduced, the preferred feeding option for infants was exclusive formula feeding, and there was no ART available for treatment in low-resourced settings. WHO’s EID diagnostic guidelines for LRS therefore recommended performing a single HIV PCR test at six weeks of age because it was believed that an HIV PCR test at that time would detect virtually all in-utero and intrapartum infections; and because six weeks happened to coincide with an established immunisation visit.

“For the next couple of years, we frantically spent our energies scaling up early infant diagnosis – both in the clinic and in the laboratory. We developed job-aids to explain to nurses how to take blood on babies for a test they’d never heard of; and we crammed our laboratories full of instruments to deliver these tests,” said Dr Sherman. She added that in South Africa, the laboratory services went from performing a couple of thousand PCR tests per year to doing more than 275,000 tests by 2010.

However, there were disparities between, and within, countries. PCR laboratories had varying capacities with turn-around times anywhere between a couple of days to a couple of months of testing. To this day, there are areas where children have to rely on a WHO presumptive diagnosis, rather than a PCR test result, to access ART. Even where PCR laboratories were functioning well, large numbers of exposed infants were not getting into be tested. Finally, many of “those that were getting PCR-positive tests weren’t getting on to treatment,” she said.

Refinements to EID

The next few years were spent on interventions to refine EID and prevent children falling through the gaps in the cascade of services required for effective programmes to prevent vertical transmission.

There have also been technological improvements, with a number of tests that now fulfill WHO recommended requirements for performance, and which can be done on whole blood or plasma or dried blood spots. In general, these are all improved in terms of automation (reducing the lab tech’s workload in the lab, and lab turnaround time), more specific (less false positives) and more sensitive, which means they detect HIV at lower levels and may have the potential to diagnose HIV early in life — possibly even in the presence of ARVs given to the infant to prevent HIV infection.

Point-of-care tests for EID are also highly anticipated, but like other point-of-care tests must meet ‘Assured’ criteria, which means: they need to be affordable (< $5), sensitive (95% or ideally 98%) and specific (98%), user-friendly (require minimal training), robust (no cold chain) and rapid (< hour), as equipment-free as possible (battery-operated, few moving parts), and deliverable (commercially available & approved). Many of these assays have already performed well when assessed against their laboratory counterparts, and have been assessed in the field as well but at present none of them have been launched.

“I’m not sure that we’ve exactly worked out where they would best fit in healthcare systems and it would differ from country to country,” said Dr Sherman.

Similarly, countries need to use the right rapid HIV tests at the right time and in the right way (which could reduce the need for PCR), but at present, according to Dr Sherman, they are under-utilised in under 18-month children. “I think in our programmes to get PCR testing accepted early on, we scared health workers off by using antibody tests in infants.”

But she also added that another challenge is that different rapid HIV test perform differently children under 18 months old than in adults. Dr Sherman said that studies she’s been involved in suggest that some rapid HIV tests do not even meet WHO performance criteria in infants, which means it’s probably best to test the mothers to identify HIV-exposed infants. If their mothers are not available, programmes should choose the rapid test that performs best to screen for exposure; and later, if testing for seroreversion, programmes should choose the rapid test that works best for that clinical application (which may be different).

Dilemmas raised for EID by the new WHO guidelines

Determining which of the evolving technologies to detect exposure and diagnose HIV infection in infants has been complicated by changes in the prevention and treatment of HIV in infants, however, Dr Sherman said.

Some of these complications could be a matter of life and death.

The need for an earlier early-infant diagnosis.

The profound reduction in infant mortality seen in the CHER study may have largely been due to when ART was initiated — at around seven weeks of age in the immediate treatment arms of that study. Dr Sherman pointed out that judging from the death records in South Africa, there is a peak of early infant mortality at around 8 to 12 weeks of age.

“If we test at 6 weeks of age, in general children get results by 10 weeks of age and are hardly ever initiated on ART before 12 weeks of age — a little bit late” she said.

But it may be possible to do better. Although a meta-analysis performed in 1995 found that if infants were tested at birth, only 38% of early infections (in-utero and intra-partum infections) would be detected, and if tested 28 days after birth, 96% of all infections could be detected —these data were drawn from studies using early PCR technology and came from a time when most women weren’t even receiving any ARVs to prevent HIV transmission to the infant. Now there are new more sensitive PCR assays, and prophylaxis consists of multiple ARVs for much longer periods of time.

There are few published data on how the newer EID assays perform before six weeks of age – but Dr Sherman shared some of NHLS’s unpublished data showing how the CAP/CTM and Aptima assay performed on samples from birth and later.

838 mother-infant pairs were enrolled into a programme at a Johannesburg delivery unit. Of those, an HIV status was established in 710 (667 were HIV-negative). 43 infants were infected (five were only detected late after six months of age), but there were 29 in-utero infections, defined by a positive PCR at birth, and nine intra-partum infections, as defined by a negative PCR at birth and a positive PCR by six weeks of age. Later however, when this routine programme tested infants at six weeks of age, it only managed to detect 26 infected infants. So even though testing at birth would have missed the intra-partum infections, it would have detected more infections.

“This means, with the new assays – and this is true for both of these assays that are more sensitive – we could detect 76% of ALL early infections at birth,” she said. “We also need to consider that these in-utero infected infants may well be those that are the rapid progressors, and need to initiate ART earlier. Added to that, if we had identified them at birth, we could have spared them the daily dose nevirapine which would have almost certainly made them resistant!”

The need to prevent nevirapine resistance in HIV-infected children

Indeed, this is a potential downside of the current WHO guidelines to prevent HIV-transmission to infants exposed by breastfeeding — while daily nevirapine for the duration of breastfeeding protects exposed uninfected infants, it poses a serious risk of resistance for any HIV-infected child taking it. Although studies seem to suggest that exposure to sdNVP may not totally eliminate the benefit of a subsequent nevirapine-based regimen, daily dosing of nevirapine monotherapy almost certainly would lead to significant drug resistance in children who are already infected — and leave ART programmes with little option to use more expensive Kaletra-based regimens (which also requires cold storage).

The potential reduction in EID accuracy associated with new regimens to prevent vertical transmission

Exclusive breastfeeding is now the preferred infant feeding option, so programmes also need to ensure that algorithms can identify any children that are infected post-natally and that testing continues until there is no further risk of post-natal transmission. What is less appreciated however is that prophylactic regimens, such as daily dosed nevirapine, may also affect the accuracy of tests used for EID.

“There is evidence emerging now that multi-drug infant and maternal prophylaxis does delay a diagnosis, and certainly reduces viral loads — but what about infant daily dose nevirapine? No data,” said Dr Sherman.

In the Johannesburg cohort, Dr Sherman described above, the infants and mothers had taken maternal AZT and sdNVP, and as already noted, 29 were infected at birth using the sensitive CAP CTM and Aptima assays.  However, when Sherman and her colleagues tested 18 infants with blood samples available at two weeks, four infants had become negative… and then became positive again at 4 weeks.

“This was probably number one, a nevirapine effect, and probably because we detected their viral loads at very low levels at birth because we were using very sensitive assays, but if this is what a single dose of nevirapine at birth does, it’s highly possible – or probable – that daily-dose nevirapine is going to extend the time that it takes to detect intra-partum infection. In other words, it will delay detecting that infection beyond six weeks of age.

Research priorities

More data needs to be collected to determine the best revision of a testing algorithm, but Dr Sherman believes that it may be better to perform testing at birth — possibly with point-of-care tests — which would give healthcare workers a better chance of retaining HIV-infected infants in care. For intra-partum infection and post-partum infections, screening should perhaps come a bit later.

This may sound alarming to some programme managers.

“Whenever I say this to our Department of Health, they see two PCR tests now – not one,” said Dr Sherman. “So to try and reduce the number of PCR tests that we would require… we could possibly use rapid tests to try and show seroreversion in some children and reduce the number of PCR tests that would be required. The rapid tests would have to be evaluated on infants in the field to know which rapid tests to use when looking for exposure, and which to use for seroreversion and minimize the number of infected infants that we miss.”

Finally, she concluded that the implementation of the whole PMCTC programme – including EID – needs to be improved “and to that end, the laboratory information system data is going to be very useful.”