The association between clinical characteristics and HIV-infection in very young infants
Polly Clayden, HIV i-Base
There are very few descriptions of characteristics of very young HIV-positive and HIV-exposed infants (<60 days). In settings with no access to PCR or CD4% quantification where clinical presumptive diagnosis is often used, these data could guide diagnostic algorithms for infants.
In an oral presentation, Heather Jaspen from the Faculty of Health Science, University of Stellenbosch, Cape Town, South Africa reported findings from an evaluation of the clinical and immunological parameters of HIV-positive infants and clinical characteristics of exposed, uninfected (EU), and unexposed, uninfected (UU) infants aged 4 to 10 weeks.
HIV-positive infants in this study were enrolled from the CHER study (n=540), and EU (n=125) and UU (n=125) infants were from a vaccine study in Cape Town and Soweto.
The median age of all infants was 44 days (range 28 to 78 days).
The investigators found weight below the 10th centile, oral thrush, lymphadenopathy and hepatomegaly to be significantly associated with HIV-positive status in infants in this study (all p<0.001). Nappy rash, was also more likely in infants with HIV (p<0.05).
Anaemia, neutropenia, gastroenteritis and gastric reflux (GERD) were also associated (p<0.005).
When the investigators looked at clinical characteristics and severe immunosupression (</=25%) in the HIV-positive infants only weight (OR 1.6, 95% CI 1.1-2.4) and lower respiratory tract infection (OR 0.3, 95% CI 0.1-0.7) were associated in an age adjusted analysis.
So the study largely focused on the association between clinical characteristics and HIV infection.
After age adjustment, the investigators found weight in the lowest 10th percentile (OR 3.3, 95% CI 1.6-6.6), oral thrush (OR 5.6, 95%CI 3.0 to 10.2), any lymphadenopathy (OR 8.9, 95%CI 3.8 to 29.8), generalized LAD (OR 9.2, 95% CI 2.9-29.8) and nappy rash (OR 2.4, 95% CI 1.5-4.0) were associated with HIV infection.
When the investigators performed a sensitivity analysis including the following symptoms: oral thrush, any LAD, hepatomegaly, splenomegaly, GERD and weight below the 10th percentile, presence of one symptom gave a sensitivity of 49.6% and specificity of 78.4%. Increasing the cut off to two or more symptoms decreased sensitivity but rapidly increased specificity. Two symptoms gave a specificity of 97.6% with 25% sensitivity. Three or four symptoms gave 100% specificity with 10.2% and 5.6% sensitivity for three and four symptoms respectively.
Dr Jaspen remarked that this is at least as good as the IMCI and WHO algorithms for diagnosis in older children. But, she explained, Many HIV-infected children will still be missed, therefore PCR in resource limited settings is essential. She added that these findings need to be validated in different populations before an algorithm can be developed.
A previous small study from the Cote dIvoire [Rouet et al. AIDS 2002, 16:2303 2309] described a mononucleosis-like syndrome, dermatitis, and generalised lymphadenitis as independent factors associated with acute retroviral infection in breast fed infants. Generalised lymphadenopathy had a specificity of 96% and a sensitivity of 32%.
The negative and positive predictive values depended on the MTCT rate: the higher the rate of transmission, the higher the PPV but the lower the NPV. Clinical algorithms are no substitute for affordable sensitive and specific tests, when it comes to decision making about starting ARVs.
Jaspan H, Myer L, Violari A et al. Clinical and immunological characteristics of very young infants with HIV infection: Children with HIV Early Antiretroviral Study. 15th CROI, February 2008, Boston, MA, USA. Oral abstract 76, 2008.