Maternal plasma viral load, zidovudine and mother-to- child transmission of HIV-1 in Africa: DITRAME ANRS
049a trial
ValeÂriane Leroy, CreÂpin Montchoa, Olivier Manigartb, Philippe Van de Perreb, FrancËois Dabis, Philippe Msellatic,
Nicolas Medab, Bruno Youa, Arlette Simononb and Christine Rouziouxd for the DITRAME Study Group (ANRS 049 clinical trial)
Objective: To study the relationship between maternal plasma RNA levels and mother-to-child transmission (MTCT) of HIV-1 in African breastfed children.
Design: Nested case±control study within a randomized trial assessing the ef®cacy of a short maternal zidovudine (ZDV) regimen to reduce MTCT.
Methods: Eligible women received either 300 mg of ZDV twice a day until labour, 600 mg at the beginning of labour and 300 mg twice a day for 7 days post-partum or a placebo. The diagnosis of paediatric HIV-1 infection was based on PCR tests at days 1±8, 45, 90 and 180 then on serology performed at 3 monthl intervals. Plasma HIV-1 RNA was measured at inclusion and on day 8 after delivery for all women who did transmit HIV to their children (cases) using a Chiron branched DNA assay (sensitivity 50 copies/ml) and compared with women who did not transmit (two per case) matched for phase trial, treatment allocation and site.
Results: At inclusion, mean log10 viral load was 4.6 among 55 transmitting mothers and 3.7 among 117 non transmitters (P0.0001). Among transmitters, the mean difference in log10viral load between day 8 post-partum and inclusion wasÿ0.13 in the ZDV group (n23) versus 0.27 in the placebo group (n32;P0.01); among non transmitters it wasÿ0.35 for the ZDV group (n47) versus 0.27 in the placebo group (n70; P,10ÿ4). In multivariate logistic regression analysis, odds ratios for MTCT were 8.7 (95% con®dence interval, 3.7±20.6) for 1 log10 increase of maternal RNA at inclusion and 4.2 (95% con®dence interval, 1.7±10.3) for 1 log10 increase difference from inclusion to day 8 post-partum.
Conclusion: High maternal viral load at inclusion strongly predicts MTCT of HIV in Africa. A short ZDV treatment regimen decreases signi®cantly maternal viral load from its pretreatment level. &2001 Lippincott Williams & Wilkins
AIDS 2001,15:517±522
Keywords: HIV, mother to child transmission, zidovudine, breastfeeding
From the INSERM U330, ISPED, Universite Victor Segalen, Bordeaux, France, theaCeDReS, Programme PAC-CI, Abidjan, CoÃte d'Ivoire,bCentre MURAZ, Bobo-Dioulasso, Burkina Faso,cIRD, Abidjan, CoÃte d'Ivoire, anddLaboratoire de virologie, CHU Necker, Paris, France.See Appendix.
Requests for reprints to: V. Leroy, Unite INSERM 330, Universite Victor Segalen - Bordeaux 2, 146, rue LeÂo Saignat, 33076 Bordeaux Cedex, France.
Note: This study was reported in part at theXIII International AIDS Conference. Durban, July 2000 [abstract WePpB1302].
Received: 7 July 2000; revised: 5 January 2001; accepted: 10 January 2001.
ISSN 0269-9370&2001 Lippincott Williams & Wilkins 517
Introduction
Two clinical trials in Africa have reported the ef®cacy of an oral maternal short regimen of zidovudine (ZDV) in reducing the risk of mother-to-child transmission (MTCT) of HIV-1 at 6 and 15 months of age in breastfed populations [1±3]. However, the effect of short ZDV regimens on the reduction of the maternal viraemia was not included in these reports. In addition, it remains unclear whether the reduction of MTCT obtained by using ZDV is due to a direct effect of the drug on maternal viraemia or/and to other yet uni- denti®ed mechanism [4].
This report examins the relationship between maternal plasma HIV-1 RNA levels and the risk of MTCT of HIV in the context of a clinical trial of ZDV in Africa.
Methods
The DITRAME ANRS 049a trial was conducted in two large cities of West Africa, Abidjan, CoÃte d'Ivoire and Bobo Dioulasso, Burkina Faso. The methodology of this randomized double-blind placebo-controlled trial has been published previously [1]. Brie¯y, eligible HIV-infected pregnant women were randomized at 36±38 weeks gestation to receive ZDV (250 mg twice a day during phase 2 and 300 mg during phase 3) or a matching placebo, until the beginning of labour, then a single oral dose of 500/600 mg until delivery, and a 7 day post-partum treatment of 500/600 mg per day.
No treatment was given to the neonate.
The diagnosis of paediatric HIV-1 infection was based on PCR tests at day 1-8, 45, 90 and 180 then on serology performed at 3 month intervals until at least 18 months of age or 2 months after complete cessation of breastfeeding if the child was still breastfed at 18 months of age. The blood sample collected at day 180, or an earlier one when this was not available, was analysed by PCR. PCR was then applied to all the preceding available samples if the ®rst one tested was positive. Non-commercial DNA PCR was used in Abidjan. In Bobo-Dioulasso samples were analysed ®rst by non-commercial DNA PCR and quantitative plas- ma RNA PCR (Amplicor HIV Monitor, version 1.5;
Roche Diagnostics Systems Inc, Branchburg, New Jersey, USA) that gave concordant results, and then by RNA PCR only. The diagnosis of paediatric HIV-1 infection was considered on the basis of one positive PCR [5]. Serum samples collected between 9 and 15 months of age were screened for HIV-1 and HIV-2 antibodies by a commercial enzyme-linked immuno- sorbent assay (ELISA; Genelavia Mixt, Diagnostics Pasteur, Paris, France or Murex ICE1-O-2, Murex Biotech Ltd, Dartford, UK). Con®rmation on the same
sample was obtained with a synthetic peptide ELISA (Peptilav 1-2, Diagnostics Pasteur). A positive antibody test at 15 months or beyond was also a diagnostic criteria of paediatric infection. Children who had no sample available for PCR and could not be followed beyond 6 months of age were considered to be of indeterminate HIV status.
For the present study, the design was a nested case±
control study within the DITRAME prospective co- hort of HIV-1-infected pregnant women and their liveborn children. Women who did transmit HIV to their children and had maternal plasma HIV-1 RNA levels measured at baseline were considered as cases and compared with women who did not transmit HIV and had samples available (of a suf®cient amount, >1 ml).
Two controls per case were selected that were compar- able for trial phase, study site and treatment allocation.
Branched DNA assay with a limit of detection of 50 HIV RNA copies/ml (Bayer-Chiron, Quantiplex 340M, version 3.0; E. Walpole, Massachusetts, USA) was used; this technology was chosen because of its particular ability to quantify correctly non-B HIV subtypes.
Plasma HIV-1 RNA viral load results were log10- transformed to obtain a normal distribution and de- scribed using mean and standard error. Absolute differ- ence between log10 viral load at day 8 post-delivery and inclusion, and its relative decrease from inclusion were estimated when possible. Comparisons of mater- nal RNA levels between transmitting and non-trans- mitting mothers and between treatment groups were made using Student's t test. Median CD4 cell count and maternal age were described and compared using Wilcoxon non-parametric test. Univariate logistic re- gression analysis estimated the odds ratio of HIV transmission for maternal viral load at entry, at day 8 post-partum, for the difference in viral load between day 8 and inclusion, other known determinants of MTCT (low birth weight, ,2500 g; prematurity, ,37 weeks gestation; prolonged rupture of mem- branes, >4 h) and adjustment variables. The latter (phase, site and treatment) and the variables associated in univariate analysis to MTCT of HIV with P,0.25 were included in a multivariate logistic regression stepwise descendant model.
Results
From September 1995 to February 1998, 421 women were enrolled and delivered 401 live infants (200 ZDV, 201 placebo). Ten children with indeterminate HIV status and seven mothers with indeterminate CD4 cell count (including one transmitter) were not eligible for this analysis.
Maternal RNA levels were available and measurable for 55 cases (23 ZDV, 32 placebo) of the 94 mothers of children diagnosed as infected up to 15 months of age. The plasma of the 39 mothers of other infected children was insuf®cient to measure maternal RNA.
These two groups were comparable in terms of treat- ment allocation (P0.74), maternal age at birth (P0.82), proportion of low birth weight (P0.54), prematurity (P0.39) and prolonged rupture of mem- branes (P0.99). Median maternal CD4 cell count was 3293106/l (range, 41±13553106/l) for the cases included in this analysis versus 4813106/l (range, 14±18173106/l) among those not included (P0.046). Infected children whose mothers were included in the present analysis were more likely to come from the Abidjan site than those not included, 69% versus 38%, respectively (P0.003) and were more often enrolled during the phase 3 of the trial, 89% versus 28%, respectively (P0.001).
Among the 290 mothers of children uninfected at 15 months of age, 117 controls (47 ZDV, 70 placebo) were selected for comparison with the 55 infected cases (Table 1). None of these women had reached the World Health Organization de®nition of AIDS.
Baseline mean log10HIV copies/ml was comparable between the two treatment groups, 4.09 log10 (SE, 0.92) among the 70 mothers in the ZDV group and 3.91 log10 (SE, 0.90) in the placebo group (P0.12).
At inclusion, mean log10HIV copies/ml was higher by 0.91 log10 among transmitting than among non-trans- mitting mothers: 4.61 log10 HIV copies/ml and 3.69 log10 HIV copies/ml, respectively (P0.0001;
Table 1). Among transmitting mothers, the mean log10
HIV copies/ml at inclusion was higher by 0.37 log10
in the ZDV group than in the placebo group (signi®- cant at P0.01), whereas there was no statistical difference between the two treatment groups among non-transmitting mothers (Table 2). The minimum viral load among transmitting mothers was 1590 HIV copies/ml (3.2 log10) at inclusion and 91 HIV copies/
ml (1.96 log10) at day 8 post-partum, both in the placebo group. At day 8 post-partum, mean log10 HIV copies/ml was 1 log10 higher among transmitting mothers than among non-transmitting mothers (P0.0001; Table 1). Among transmitting mothers, the mean change of log10 HIV copies/ml from inclusion to the end of the peri-partum treatment was ÿ0.13 in the ZDV group versus 0.27 in theplacebo group (P0.01); among non-transmitting mothers, it was ÿ0.35 in the ZDV group versus 0.27 in the placebo group (P,10ÿ4; Table 2). Compared with the baseline viral load, the mean change between inclusion and day 8 post-partum was smaller in the ZDV group, both in transmitting and non-transmitting mothers whereas it tended to increase in placebo
groups (Table 2). Table1.MaternalplasmaHIV-1RNAlevels,treatmentgroup,andotherdeterminantsofmother-to-childtransmissionofHIV-1.Univariatelogisticregression.Nestedcase±controlstudywithinthe DITRAMEANRS049atrial.AbidjanandBobo-Dioulasso,1995±1998. TransmittingmothersNon-transmittingmothers95%Con®dence (n55)(n117)OddsratiointervalP aAbidjansite(%)69.175.20.730.4±1.50.39 aPhase2trial(%)10.920.52.100.8±5.50.12 aZidovudinnematernaltreatment(%)41.840.11.070.5±2.10.83 bMedianmaternalage[years(range)]25(18±42)24(18±41)1.170.6±1.70.60 6caMedianCD4cellcountatentry[310/l(range)]329(136±1355)534(931±1456)1.351.2±1.50.0001 aLowbirth-weight(,2500g)(%)25.9(14/54)11.2(13/116)2.771.2±6.40.017 Prematurity(gestationalage,37weeks)(%)11.1(5/45)9.0(9/100)1.260.4±4.00.69 Prolongedruptureofmembranes(.4h)(%)31.7(16/51)22.8(23/101)1.550.7±3.30.26 daMeanmaternalplasmaHIVRNAatentry4.61(0.08)3.69(0.08)5.072.8±9.00.0001 [logcopies/ml(SE)]10 daMeanmaternalplasmalogHIVRNAday8post-partum4.73(0.10)3.73(0.08)5.993.1±11.50.000110 [logcopies/ml(SE)]10 dMeandifferenceinHIVRNAbetweenday8post-partumandentry0.10(0.08)0.02(0.07)1.200.7±2.00.49 [logcopies/ml(SE)]10 abc6dVariablesincludedinthemultivariatestepwisedescendantlogisticregressionmodel.Foranincreaseof10years.Foradecreaseof100310/lCD4cells.Foronelogincrease.10
In univariate logistic regression analysis, maternal CD4 cell count at inclusion, low birth weight, mean viral load at inclusion and at day 8 post-partum were signi®cantly associated with the risk of MTCT while study site, phase trial, maternal treatment, maternal age, prematurity and prolonged rupture of membranes were not (Table 1). In the ®nal multivariate model adjusted also for study site, phase trial and maternal treatment, the only variables signi®cantly associated with the risk of MTCT were: maternal CD4 cell count at inclusion [adjusted odds ratio (aOR), 1.22 for a CD4 cell count decrease of 1003106/l; 95% con®dence interval (CI), 1.03±1.41); maternal plasma RNA level at inclusion (aOR, 8.72 for a 1 log10 increase; CI, 3.69±20.57);
absolute difference between day 8 post-partum and inclusion plasma RNA levels (aOR, 4.23 for a 1 log10
increase; CI, 1.74±10.29) and low birth weight (aOR, 9.17; CI, 2.54±33.13).
Discussion
This case±control study nested within a randomized clinical trial allowed estimation of the contribution of maternal plasma viral load as a risk factor for MTCT adjusted for other known determinants of MTCT and controlled for maternal ZDV treatment in an African population. As there were less than 2% undetectable HIV RNA values in this sample, this could not bias the comparisons between the two groups of transmitting and non-transmitting mothers. By selecting mothers who had a more severe immunode®ciency compared with those not included in this study, this potential bias could have overestimated the fraction of MTCT attributable to this risk factor. As our ®nal logistic regression model was adjusted on matching variables and maternal CD4 cell count, we concluded that independently of maternal CD4 cell count, baseline maternal plasma viral load and its relative decrease at day 8 post-partum are signi®cant risk factors for MTCT.
In this mainly breastfed population, maternal plasma viral load at the end of the pregnancy and its evolution after delivery were strong independent predictors of MTCT of HIV. This increased risk of MTCT asso- ciated with a high maternal viral load con®rms the
®ndings of previous studies in non-breastfed popula- tions with different HIV strains in the USA [4,6±10], Europe [4,11,12], and Thailand [13]. This is also consistent with the results of the recent report of an observational cohort study performed in the Gambia [14]. However, as transmission was demonstrated from women with very low HIV load at day 8 post partum, it is not possible to de®ne a viral load threshold that could predict con®dently an absence of transmission, in
Table2.MaternalCD4cellcount,plasmalog10HIV-1RNAlevelsatinclusion,day8post-partumandevolutionaccordingtomaternaltreatmentandriskofmother-to-childtransmissionofHIV-1. Nestedcase±controlstudywithintheDITRAMEANRS049atrial.AbidjanandBobo-Dioulasso,1995±1998. TransmittingmothersNon-transmittingmothers ZDV (n23)Placebo (n32)PZDV (n47)Placebo (n70)P BaselinemedianCD4cellcount[3106/l(IQR)]307(168±412)358(263±623)0.19487(327±757)563(407±756)0.16 BaselineHIVRNA(log10copies/ml) Mean(SE)4.82(0.08)4.45(0.12)0.013.73(0.13)3.66(0.11)0.70 Minimum±maximum3.80±5.483.20±5.691.70±5.271.69±5.33 Undetectablevalue(n)a0003 Day8post-partumHIVRNA(log10copies/ml)(n20)(n28)(n42)(n63) Mean(SE)4.70(0.12)4.74(0.15)0.843.29(0.13)4.03(0.09),10ÿ4 Minimum±maximum3.75±5.691.96±5.691.69±4.551.69±5.69 Undetectablevalue(n)a0031 Absolutedifferenceday8±baselineinHIVRNA(log10copies/ml)(n20)(n28)(n42)(n63) Mean(SE)ÿ0.13(0.12)0.27(0.11)0.01ÿ0.35(0.10)0.27(0.07),10ÿ4 Minimum±maximumÿ1.58±0.85ÿ1.60±1.52ÿ1.50±1.47ÿ1.06±2.09 Relativedifferenceday8±baseline/baselineinHIVRNA(log10copies/ml)(n20)(n28)(n42)(n63) Mean(SE)ÿ0.02(0.11)0.07(0.15)0.008ÿ0.08(0.19)0.10(0.22)10ÿ4 Minimum±maximumÿ0.29±0.22ÿ0.45±0.42ÿ0.38±0.52ÿ0.21±1.21 a,50copiesofHIV-1RNA.IQR,Interquartilerange;ZDV,ziduvudine.
contrast with a conclusion made in a single report so far [6]
Several studies have identi®ed maternal viral load as a determinant of MTCT but few adjusted in the same analysis for the other known risk factors of MTCT such as delivery or host determinants [10,12,13].
Cesarean section was too rare in our population to be taken into account but we studied other delivery and neonatal risk factors and could not retain them in the
®nal model; the exception was low birth weight which was found to be an independent risk factor of MTCT adjusted for maternal viral load and CD4 cell count.
Others have also reported that prematurity remains strongly associated with MTCT even after including the effect of viral load [12,13] but we did not observe this association. It is unclear whether low birth weight and prematurity are foetal risk factors or if they re¯ect of the effect of HIV infection itself.
In the placebo group, the evolution of maternal plasma viral load between inclusion and day 8 post partum showed a mean increase of 0.27 log10 HIV copies/ml, consistent between transmitting and non-transmitting mothers. Our maternal short ZDV regimen decreased signi®cantly the maternal viral load at day 8 post- partum from its baseline level when compared with the placebo group. This decrease was in¯uenced by the initial viral load as the effect seemed to be greater among non-transmitting mothers than transmitting mothers with higher initial viral load. This important
®nding may suggest that ZDV effect on the reduction of MTCT could be greater in women with lower initial viral load and thus less advanced HIV disease at inclusion in the intervention arm. However, our design does not allow for a full assessment of the contribution of ZDV to the reduction of MTCT through the change in viral load; this would have required a systematic prospective assessment of maternal viral load at inclusion and at day 8 post-partum.
Our ®ndings support the role of high viral load in overall MTCT of HIV-1 in Africa but also suggest the role of ZDV in decreasing the maternal plasma viral load and consequently reducing the risk of MTCT even after a few weeks of treatment. Maternal plasma viral load is an important factor in MTCT but it is not the only one. In an African context, transmission attributable to breastfeeding is substantial, with 44% of all HIV infections attributable to breast milk [15]. This postnatal transmission could have profound negative implications on the overall effect of peri-partum anti- retroviral interventions given to mothers to prevent MTCT [16,17]. Thus, further analysis should pay attention to the reduction of viral load during preg- nancy and after delivery in relation to ZDV use; the timing of MTCT in this context of peri-partum and breastfeeding transmission should be considered also.
Acknowledgements
The authors acknowledge the support obtained from the ANRS and the French Ministry of Foreign Affairs, the National AIDS authorities in CoÃte d'Ivoire and Burkina Faso and women and children enrolled in the trial. The authors thank F. Rouet and K. Castetbon for their help in the preparation of the manuscript and to P. Combe for the laboratory organisation in Abidjan.
References
1. Dabis F, Msellati P, Meda N,et al.Six months ef®cacy, tolerance and acceptability of a short regimen of oral zidovudine in reducing vertical transmission of HIV in breast-fed children. A double blind placebo controlled multicentre trial, ANRS049a, CoÃte d'Ivoire and Burkina Faso.Lancet1999,353:786±792.
2. DITRAME ANRS 049 Study Group. 15-month ef®cacy of oral zidovudine to decrease vertical transmission of HIV-1 in breastfed African children.Lancet1999,354:2050±2051.
3. Wiktor S, Ekpini E, Karon J,et al.Short-course oral zidovudine for prevention of mother-to-child transmission of HIV-1 in Abidjan, CoÃte d'Ivoire: a randomized trial. Lancet 1999, 353:781±785.
4. Sperling R, Shapiro D, Coombs R, et al. Maternal viral load, zidovudine treatment and the risk of transmission of Human Immunode®ciency Virus type 1 from mother to infant.N Engl J Med1996,335:1621±1629.
5. Dabis F, Msellati P, Newell M,et al.Methodology of intervention trials to reduce mother-to-child transmission of HIV-1 with special reference to developing countries. AIDS1995,9(suppl A):S67±S74.
6. Fang G, Burger H, Grimson R, et al. Maternal plasma human immunode®ciency virus type 1 RNA level: a determinant and projected threshold for mother-to-child transmission.Proc Natl Acad Sci USA1995,92:12100±12104.
7. Thea D, Steketee R, Pliner V,et al.The effect of maternal viral load on the risk of perinatal transmission of HIV-1.AIDS1997, 11:437±444.
8. Dickover RE, Garratty EM, Herman SA, et al.Identi®cation of levels of maternal HIV-1 RNA associated with risk of perinatal transmission: effect of maternal zidovudine treatment on viral load.JAMA1996,275:599±605.
9. Garcia PM, Kalish LA, Pitt J, et al.Maternal levels of plasma human immunode®ciency virus type 1 RNA and the risk of perinatal transmission.New Engl J Med1999,341:394±402.
10. Mofenson L, Lambert J, Stiehm E,et al.Risk factors for perinatal transmission of human immunode®ciency virus type 1 in women treated with zidovudine.New Engl J Med1999,341:385±393.
11. Mayaux M, Dussaix E, Isopet J,et al.Maternal virus load during pregnancy and mother-to-child transmission of human immuno- de®ciency Virus type 1: the French perinatal cohort study.
J Infect Dis1997,175:172±175.
12. European Collaborative Study.Maternal viral load and vertical transmission of HIV-1: an important factor but not the only one.
AIDS1997,13:1377±1385.
13. Shaffer N, Roogpisuthipong A, Siriwasin W,et al.Maternal viral load and perinatal HIV-1 subtype E transmission, Thailand.
J Infect Dis1999,179:590±599.
14. O'Donovan D, Ariyoshi K, Milligan P, et al. Maternal plasma viral RNA levels determine marked differences in mother-to- child transmission rates of HIV-1 and HIV-2 in the Gambia.
AIDS2000,14:441±448.
15. Nduati R, John G, Mbori-Ngacha D,et al.Effect of breastfeeding and formula feeding on transmission of HIV-1. A randomized clinical trial.JAMA2000,283:1167±1174.
16. Dabis F, Leroy V, Castetbon K, Spira R, Newell M, Salamon R.
Preventing mother-to-child transmission of HIV-1 in Africa in the year 2000.AIDS2000,14:1017±1026.
17. De Cock K, Fowler M, Mercie E,et al.Prevention of mother-to- child HIV transmission in resource-poor countries. Translating research into policy and practice.JAMA2000,283:1175±1182.
Appendix
The DITRAME Study Group
Coordination: INSERM U 330, Universite Victor Segalen Bordeaux 2, Bordeaux, France (F. Dabis).
Principal investigators: Centre Hospitalier Universitaire de Yopougon, Abidjan, CoÃte d'Ivoire (C. Welffens- Ekra), Maternite Cochin Port-Royal, Paris, France (L.
Mandelbrot). Abidjan Center (CoÃte d'Ivoire):
CeDReS, Centre Hospitalier Universitaire de Treich- ville (D. Bonard, P. Combe, N. Elenga, R. Likikouet, C. Montcho, V. Noba, F. Sylla-Koko, I. Viho, B.
You); Centre Hospitalier Universitaire de Yopougon (R. Camara, M. Dosso, M. TimiteÂ); DITRAME Project (G. Gourvellec, R. Ramon); ORSTOM Petit Bassam (P. Msellati, local coordinator) and the Health Centers of Anonkoua-Koute, Ouassakara, Yopougon-
Attie and Yopougon. Bobo-Dioulasso Center (Burkina Faso): Centre Muraz (M. Cartoux, A.M. Cassel- Beraud, L. Gautier-Charpentier, O. Ky-Zerbo, O.
Manigart, N. Meda (local coordinator), A OuangreÂ, O Sanou, A. Simonon, I. SombieÂ, S. Tiendrebeogo, S.
Yaro); Centre Hospitalier National Souroà Sanou (A.
BazieÂ, B. Dao, B. Nacro, F. Tall) and the Health Centers of Accart-Ville, Farakan and Social Security.
Data management: INSERM U 330, Bordeaux (L.
Dequae-Merchadou). Methodology: INSERM U 330, Bordeaux (V. Leroy, R. Salamon); Centre MURAZ (P. Van de Perre); Laboratoire de Virologie, HoÃpital Necker-Enfants Malades, Paris, France (C. Rouzioux).
Data and Safety Monitoring Board: J-F. Delfraissy (President), D. Costagliola and C. Chouquet (statisti- cians), B. Bazin, P. Lepage, B. Masquelier and K.
Toure Coulibaly.
