A Collaborative Analysis of Data from Cohorts in Thailand, South Africa, Botswana, and the United Kingdom

Internat ional Col laborat ive Study of Pediatr ic Diagnost ic Tests

JULY 19 , 2011

Time to DNA-PCR Positivity in Non-Breastfed HIV-Infected Infants

(Primarily Non-B HIV Subtype)

Background

Accurate diagnostic tests to detect HIV infection in infants are critical to ensure early treatment

HIV DNA-PCR has imperfect sensitivity in the first two weeks of life

Previous studies found that time to HIV DNA-PCR positivity increase by about 15% with ZDV prophylaxis compared with no antiretrovirals (ARV)

Impact of combination ARV prophylaxis is not well characterized; no single cohort can adequately address this question

Goals

Combine data from several cohorts of non-breastfeeding HIV-infected mother-infant pairs Initial phase: cohorts with primarily non-B HIV

subtype Estimate the time to DNA-PCR positivity in

non-breastfed HIV-infected infants Assess differences in time to DNA-PCR

positivity according to maternal/infant ARV regimen

Inclusion/Exclusion Criteria

Inclusion criteria: Infant HIV-infected and has at least one DNA-

PCR result before age 3 months Maternal HIV diagnosis before or within 1 month

after birth Excluded diagnostic tests with missing

result or missing age at time of blood drawExcluded one infant whose mother’s ARV

exposure was unknown

Statistical Methods

Used methods for interval-censored data Timing of HIV infection uncertain; in interval between last

negative and first positive DNA-PCR testEstimated the cumulative probability of DNA-PCR

positivity according to age and ARV Non-parametric methods; Turnbull algorithm, Wilcoxon

testExploratory regression modeling to adjust for

potential confounders Parametric methods; assumed Weibull distributions with

proportional hazards; likelihood ratio test

Participating Cohorts

Cohort Country Birth YearsNumber of

Infants (Tests)MASHI Botswana 2001-2003 33 (91)

Infant Diagnostic Study South Africa 2001-2003 26 (83)

NSHPC/HPA UK (African origin) 2000-2008 74 (181)

PHPT Thailand 1997- 2003 177 (678)

Thai/CDC Thailand 1992-1998 122 (370)

Total 432 (1403)

Infants Grouped by Most Complex Maternal/Infant ARV

Infant ARV Prophylaxis

Mat

erna

l AR

V du

ring

Trim

este

r of D

eliv

ery

No ARVSingle NRTI

sdNVP (+/- ZDV) >3 ARVs

No ARV 125 4 5 4Single NRTI 18 137 8 2sdNVP (+/-

ZDV) 0 21 71 1>3 ARVs 0 14 2 20

No ARV = Mother and infant did not receive ARV at any time (n=125)Single NRTI = Mother or infant received single NRTI (n=159)sdNVP (+/- ZDV) = Mother or infant received sdNVP +/- ZDV (n=105)>3 ARVS = Mother or infant received >3 ARVs (n=43)

Demographics

No ARV (n=125)

Single NRTI (n=159)

sdNVP +/- ZDV (n=105)

>3 ARVs(n=43)

Year of Birth 1992-1994 67 0 0 0 1996-1999 37 113 0 0 2000-2003 12 43 104 14 2004-2008 9 3 1 29Country Where HIV Acquired Botswana 0 11 20 2 South Africa 0 0 27 2 Zimbabwe 8 3 2 10 Nigeria 4 1 0 6 Other African Country 7 5 0 21 Thailand 105 139 56 1

Characteristics (median [25th-75th percentile])

No ARV(n=123)

Single NRTI (n=159)

sdNVP +/- ZDV (n=79)

>3 ARVs(n=43)

P value(Kruskal-Wallis)

Maternal CD4 Near Delivery

410 [290-529]

296 [190-415]

284 [185-424]

290 [149-400]

< 0.0001

Maternal Viral Load Near Delivery

67,590 [26,263-191,174]

14,125[2,963 – 39,363]

41,496[12,448 – 94,575]

6,000[419-

50,666]

< 0.0001

Gestational Age at Delivery

39[35-43]

38[33-43]

38[33-43]

37[30-45]

< 0.001

Delivery Type: VaginalCS no labor/ROMCS with labor/ROM

92%2%6%

78%9%

13%

80%5%

15%

25%40%35%

< 0.0001

Cumulative Probability of Positive DNA-PCR by Age (Non-Parametric)

P value (Wilcoxon) =0.0002

ARV Group

Number of

infantsBirth-1 day

<14 days

< 30 days

<42 days

<90 days

<370 days

No ARV 125 0.31 0.94 0.94 0.94 1.00 1.00Single NRTI 159 0.63 0.63 0.91 0.91 0.96 1.00

sdNVP +/- ZDV 105 0.71 0.82 0.82 0.96 1.00 1.00>3 ARVs 43 0.67 0.67 0.67 0.79 0.94 1.00

Cumulative Probability of Positive DNA-PCR by Age (Subset: 143 infants negative at birth or day 1)

P value (Wilcoxon) =0.0007

ARV GroupNumber

of infants<14

days <30

days <42

days<90

days<370 days

No ARV 48 0 0.87 0.87 1.00 1.00Single NRTI 70 0 0.86 0.86 0.97 1.00

sdNVP+/- ZDV 13 0 0 0.80 1.00 1.00>3 ARVs 12 0 0.35 0.35 0.88 1.00

Cumulative Probability of Positive DNA-PCR by Age (Parametric - separate Weibull models, unadjusted)

Adjustment for Confounders: Infants who Received ARV(Parametric, Proportional Hazards; HR >1 Means Earlier Positivity)

ARV Group Unadjusted Hazard Ratio

[95% CI]

Adjusted Hazard Ratio [95% CI]

Single NRTI 1.44 [1.00 – 2.09] 1.44 [0.92-2.28]sdNVP+/- ZDV 1.56 [1.06 – 2.30] 1.39 [0.84-2.30]>3 ARVs 1.0 1.0P value for ARV Group

0.06 0.11*

*P value for CD4 = 0.80, viral load =0.89, gestational age=0.44, delivery type =0.91

Summary and Conclusions

Lower DNA-PCR positivity at birth and greater increase in positivity by 14 days of age in HIV-infected, non-breastfed infants who had no ARV vs. those who had maternal or infant ARV Suggests ARV prevents a large proportion of

intrapartum transmissionNonparametric estimates of the probability of

DNA-PCR positivity by age differed significantly according to ARV group Time to DNA-PCR positivity was later with receipt of

>3 ARVs than with single NRTI or sd-NVP (+/- ZDV)

Summary and Conclusions (2)

In preliminary parametric regression modeling, the association between ARV group and time to positivity did not remain statistically significant However, adjustment did not change the hazard ratios

and the confidence intervals were mostly above 1.0 The small number of infants exposed to > 3 ARVs limited

the statistical power to detect a difference; further study is needed

Our results may have implications for scheduling final HIV PCR diagnostic testing, particularly when resources are limited.

International Collaborative Study of Pediatric Diagnostic Tests – Collaborators

R. Balasubramanian D.E. Shapiro M.G. FowlerK. DominguezP. TookeyJ. MastersJ. TosswillM. Lallemant N. Ngo-Giang Huang M. McConnellP. Mock

G. ShermanS. LockmanV. NovitskyP. PalumboS. NesheimB. BohannonK. RichM. HughesWe gratefully

acknowledge the study participants