Mechanisms and Timing of Perinatal HIV Transmission

Mechanisms of Transmission

HIV may be transmitted from mother to infant in three ways:

Diagnosis of Perinatally Acquired HIV Infection

Bryson and colleagues[38] have proposed a definition of the timing of HIV infection. Early transmission (or in utero) is defined as that which may be detected within the first 48 hours of life. Late transmission (or intrapartum) is defined as negative virologic evaluations during the first week, with evidence of HIV detection between 7 and 90 days of age in the absence of breastfeeding. Employing viral culture, approximately 15% of HIV-infected infants have detectable levels of virus in the blood within 48 hours of birth,[39,40] indicating that these infections may have occurred during in utero development. Another study using HIV culture as the diagnostic method demonstrated a sensitivity of 24% in the first week of life which increased to 85% by 1 month of age.[41] Using polymerase chain reaction (PCR) technology, HIV-1 DNA was detected in 38% of 271 infected infants studied within 48 hours of birth.[42] These data would indicate that as much as 62% to 85% of perinatal transmission may occur in the intrapartum or neonatal periods. Approximately 93% of HIV-infected infants will demonstrate presence of viral markers by 14 days of life with PCR testing, and essentially 100% within 30 to 90 days of birth.[39-43] This excludes infants who are breastfed by HIV-infected mothers, for in this setting HIV transmission may occur throughout the duration of breastfeeding.

Factors Which Influence the Risk of Perinatal HIV Transmission

Maternal HIV viral load. Clinically or immunologically advanced HIV disease in the mother has consistently been associated with higher likelihood of transmission to the infant.[16,37] Thus, the European Collaborative Study found that a CD4+ cell count less than 700 cells/mm3 was associated with increased risk of maternal-fetal transmission,[16] while Mayaux and colleagues noted a 43% transmission rate among women whose CD4+ cell counts were less than 200 cells/mm3, versus 15% when the CD4+ count was over 600 cells/mm3.[44] Of note, the subsequent course of HIV disease in the infected children has also been correlated with the severity of maternal HIV disease during pregnancy. Blanche and colleagues demonstrated a statistically increased risk of death during the first 18 months of life in those infants whose mothers had more advanced clinical disease, presence of p24 antigenemia, and/or lower CD4+ cell count at the time of delivery.[45] In addition, studies have demonstrated that an increased maternal virus load is associated with an increased risk of in utero transmission.[46,47] In utero transmission, in turn, appears to correlate with an increased rate of disease progression in the infant.[48-50]

Multiple studies have confirmed the importance of maternal viral load in predicting the risk of HIV transmission to the infant.[51-53] Dickover and colleagues studied 92 HIV-seropositive pregnant women and their 97 infants, of whom 20 (21%) were subsequently proved to be perinatally infected.[54] A maternal HIV-1 RNA level over 50,000 copies/mL at delivery was associated with increased transmission: 75% of the 20 transmitters had viral load greater than 50,000 copies/mL compared with only 4 (5.3%) of the nontransmitters. In contrast, none of the 63 women with viral load less than 20,000 copies/mL transmitted HIV to their infants. Forty-two of the mothers received zidovudine during pregnancy, resulting in significant decreases in HIV-1 RNA levels in plasma; none of these women transmitted HIV to the newborn. However, 4 zidovudine-treated women with persistently high HIV-1 RNA levels did transmit HIV-1. These data would suggest that maternal HIV-1 viral load is an important determinant of perinatal transmission, and also suggests the possibility of a threshold below which perinatal transmission would probably not occur. However, transmission may occur even with undetectable levels of HIV RNA.[55]

The importance of HIV-1 viral load in predicting perinatal transmission was also addressed by Sperling and colleagues, as part of the Pediatric AIDS Clinical Trials Group (PACTG) Protocol 076 Study Group, which evaluated the efficacy of zidovudine in the prevention of perinatal transmission.[55] In this placebo-controlled trial of 402 mother-infant pairs, the rate of transmission of HIV-1 was 7.6% with zidovudine, and 22.6% in those who received placebo (P<.001). In the placebo group, a high viral load at study entry (14 weeks of gestation) or delivery was associated with increased likelihood of HIV transmission, with a 40% transmission rate in those women with the highest levels of HIV-1 (>15,700 copies/mL on the RT-PCR assay, and >7,530 copies/mL on the branched DNA assay). Of interest, while zidovudine was associated with a reduction in perinatal transmission and a median 0.24 log10 decrease in plasma viral RNA, a reduction in viral RNA from baseline to delivery was not significantly associated with the risk of transmission. Further, zidovudine was effective in reducing transmission at all HIV-1 RNA levels, even the lowest. These data suggest that high HIV-1 viral load is associated with perinatal transmission in untreated women. Further, the ability of zidovudine to reduce such transmission is only partially explained by its ability to lower viral load in the mother. It is possible that ZDV may reduce perinatal transmission by providing prophylaxis directly to the infant, since this antiretroviral readily crosses the placenta.

Despina and colleagues recently assessed the predictive value of maternal HIV-RNA load in perinatal HIV transmission by performing a meta-analysis of 9 cohorts involving 1115 maternal-child pairs.[56] The overall rate of transmission in untreated women was 21.3%. As the HIV-1 viral load increased, so did the rate of transmission, rising from 5% in women with <1000 copies/mL to 15% in those with 1000 to 9999 copies/mL and 37% in those with HIV-RNA levels over 10,000 copies/mL. In pregnant women receiving antiretroviral therapy, the perinatal transmission rate was 5% for those with <1000 copies/mL; 7% for those with viral loads between 1000 and 9999 copies/mL; and 18% for those with over 10,000 copies/mL. The risk of transmission, based solely on HIV-1 RNA levels in plasma, thus appears attenuated in patients receiving antiretroviral therapy. However, while maternal HIV-RNA levels are quite helpful in predicting the average risk of transmission in a group of women, such information is of more limited use in predicting outcome in individual patients, since an absolute threshold of safety has not been proven.[55]

Table I: Predictive Value of Maternal Viral Load for HIV Transmission[56]

Viral load (copies/mL)

Rate of Perinatal HIV Transmission

Untreated Women

Treated Women










Garcia and colleagues measured HIV-1 RNA in 552 pregnant, HIV-1-infected women enrolled in the Women and Infants Transmission Study (WITS).[57] Increasing geometric mean levels of plasma HIV-1 RNA were associated with an increasing risk of transmission (Table II).

Table II: Maternal Viral Load and HIV Transmission in the WITS Study [57]

HIV-1 RNA (copies/mL)

Number transmitting/total

Percent transmission
















On multivariate analysis, higher maternal plasma HIV-1 RNA level, lack of receipt of zidovudine therapy according to ACTG protocol 076, low birth weight of the infant, and time from rupture of membranes of delivery >4 hours were all statistically associated with risk of perinatal transmission. This study would thus suggest that while HIV-1 RNA level is highly predictive of transmission, other factors are also important.

Further analysis of Pediatric ACTG study 185 was also recently published by Mofenson and colleagues.[58] In this study of 480 pregnant women with advanced HIV-1 disease, all received zidovudine during and after pregnancy, as did the infants. Half of the women were also randomized to receive HIV-1 hyperimmune globulin monthly during pregnancy and once to the neonates at birth. In this study, baseline HIV RNA at the time of first testing during pregnancy was statistically correlated with eventual perinatal transmission.

At the time of delivery, several factors were associated with increased risk of transmission, including higher maternal HIV-1 RNA levels, higher levels of maternal HIV p24 antibody, and presence of chorioamnionitis. No perinatal transmission occurred among the 84 women who had undetectable HIV RNA (<500 copies/mL) in plasma at baseline, or the 107 women who had undetectable levels at delivery.

These studies would thus indicate that HIV-1 RNA is a powerful predictor of HIV-1 transmission. However, obstetrical factors are also important, such as prolonged intervals between amnionic membrane rupture and delivery, and presence of chorioamnionitis.

Other maternal factors. Aside from more advanced HIV disease, additional maternal factors have been associated with an increased chance of perinatal transmission.

Maternal cigarette smoking has been associated with a statistically significant, 3-fold increased risk of perinatal HIV-1 transmission, in women with low CD4+ cell counts.[59] Further, mothers who transmitted HIV to their infants were more likely to have clinical chorioamnionitis than those who did not transmit, as demonstrated in a study by Nair and colleagues.[60]

While use of illicit injection drugs during pregnancy has been associated with an increased risk of perinatal HIV infection, a past history of injection drug use poses no increased risk.[59,61] Furthermore, geographic origin, educational history, parity, or marital status are not associated with change in risk for transmission.[16,45,60]

Older maternal age, with increasing risk for each 5-year increment over 25 years, has been associated with an increased risk of perinatal transmission,[44] although the majority of studies have not confirmed this relationship.[16,45,60]

Obstetrical factors: premature rupture of the membranes. Premature rupture of the amniotic membranes has been associated with higher risk of perinatal HIV transmission, probably related to increased duration of fetal exposure to infected cervicovaginal secretions. The Women and Infants Transmission Study (WITS) enrolled 525 women who delivered live singleton infants, in whom the serostatus was known. In those mothers whose membranes ruptured more than 4 hours prior to delivery, the rate of perinatal transmission was 25%, versus 14% among those whose membranes ruptured 4 hours or less prior to the time of delivery.[61] This increased risk was independent of the mode of delivery. On multivariate analysis, ruptured membranes for over 4 hours nearly doubled the risk of transmission. Premature rupture was also shown to be a significant factor in predicting perinatal transmission by Burns and colleagues,[59,62] independent of maternal CD4+ cell count. The impact of membrane rupture over 4 hours is also evident among women who have been treated with antiretroviral agents during pregnancy.[61,63]

Obstetrical factors: mode of delivery. Initial studies indicated similar rates of HIV transmission in children born after vaginal or cesarean deliveries.[9,25,64] However, recent data would indicate that elective cesarean section (C-section) may be very effective in reducing perinatal transmission of HIV-1. These discrepancies may be secondary to the fact that elective C-section negates the possibility of premature rupture of the membranes. Differences in results could also have been caused by the possible inclusion of women who underwent nonelective C-section; and by the potential bias that some obstetricians may be more or less likely to perform C-sections on HIV-infected patients with more advanced clinical disease.

The European Collaborative Study, based on 721 children born to 701 mothers, demonstrated that elective C-section was associated with a decreased risk of perinatal transmission (relative risk = 0.56); however, these results were not statistically significant.[16] A subsequent report from this group again noted a decreased risk of perinatal HIV transmission after C-section.[22] In another study, Duliege and colleagues evaluated data from prospectively identified twins whose mothers were HIV-infected.[65] Among first-born twins, 52% of the transmission risk was found to be related to vaginal delivery.

More recently, Read and colleagues performed a meta-analysis of North American and European studies, each consisting of at least 100 mother-child pairs, to determine the role of elective C-section in decreasing perinatal transmission.[66] The primary analysis included data from 8533 mother-child pairs. After adjusting for use of antiretroviral therapy, maternal stage of HIV infection, and infant birth weight, the risk of perinatal transmission was decreased by approximately 50% in those women who underwent elective C-section, when compared with other modes of delivery (odds ratio, 0.43; 95% confidence limit, 0.33 to 0.56). Similar results were found when the control population was limited only to those women with rupture of the membranes immediately prior to delivery. When all C-sections were compared with all vaginal deliveries in this meta-analysis, use of C-section remained predictive of decreased risk of transmission, although not to the same degree as that seen with elective C-section alone.

Among women who did not receive antiretroviral therapy, a transmission rate of 10.4% was documented after elective C-section, versus 19% after vaginal delivery.[66] Of importance, use of antiretroviral therapy during pregnancy as well as elective C-section resulted in a decrease of perinatal transmission of approximately 87%. Among the 196 women who took antiretroviral agents and had an elective C-section, the vertical transmission rate was 2%. Among 1255 mothers who took antiretroviral agents but underwent other modes of delivery, the perinatal transmission rate was 7.3%. It is important to note, however, that no maternal virologic data were available from the 15 studies that were included in this meta-analysis. Further, even though there was clearly a significant association between elective C-sections and decrease in perinatal HIV transmission, no conclusions can be drawn about what additional benefit, if any, an elective C-section would have in preventing perinatal transmission from mothers with undetectable plasma levels of virus.

Table III: Perinatal HIV Transmission Rates Among Women Receiving Antiretroviral Therapy and/or Elective Cesarean Section[66]


Antiretroviral Therapy

No Antiretroviral Therapy

Elective C-section



No Elective C-section



Of interest, a second study performed in a French cohort of patients, evaluating the association between elective C-section and perinatal HIV-1 transmission rates, only noted a clear benefit of this procedure when it was performed in conjunction with antiretroviral treatment of the mother.[67] Mandelbrot and colleagues reported a perinatal transmission rate of 17.5% for vaginal deliveries, 15.6% for nonelective C-sections, and 17.5% for elective C-sections in a group of 1877 women who received no antiretroviral therapy during pregnancy. However, for 872 women who received ZDV treatment during gestation there was a significant association between mode of delivery and perinatal transmission. Transmission rates were 6.6% for vaginal deliveries, 11.4% for nonelective C-sections, and 0.8% for elective C-section (P=0.002) with an odds ratio for elective C-section of 0.2 (95% CI, 0.0-0.9). Similar to the Read study,[66] maternal virologic data were not available for analysis in the French study.

A recent prospective randomized clinical trial has confirmed the advantage of elective C-section.[68] The European Mode of Delivery Collaboration study randomized 370 infants born to HIV-infected mothers to delivery via either elective C-section (170 infants) or vaginal delivery (200 infants). At 18 months, the rate of HIV transmission to the infants was 1.8% among those born by C-section, and 10.5% among those assigned to vaginal delivery. No serious side effects were noted among women who underwent elective C-section, and the rate of postpartum complications was very small.

Recent data on the rate of infectious complications after surgical deliveries in HIV-infected women are conflicting. WITS reported a morbidity rate associated with elective cesarean delivery of 19%.[69] Morbidity due to infections, including endometritis, wound infections, and urinary tract infections, occurred in 11% of elective C-sections and in 21% of nonelective C-sections. For vaginal deliveries these rates were 8% for women receiving instrumental assistance and 4% for women with noninstrumented deliveries. A second perinatal study, ACTG 185, also reported an increased rate of infectious complications in HIV-infected women undergoing surgical deliveries: 26% for elective C-sections and 40% for C-sections performed after rupture of amniotic membranes.[70] Vaginal deliveries had a lower rate of infectious complications: 13% for spontaneous deliveries and 19% for assisted vaginal deliveries. Of interest, in this study absolute CD4+ cell count did not correlate with an increased rate of infectious complications. These results clearly demonstrate that risks and benefits of surgical procedures must be taken into account when considering elective C-sections for HIV-infected women.

Nonetheless, the current data do demonstrate a significant advantage to elective C-section delivery in HIV-infected mothers. Such a mode of delivery would theoretically be associated with decreased likelihood of microtransfusions of infected blood to the fetus during labor. Further, C-section would result in avoidance of direct fetal contact with infected maternal sections in the birth canal. Unfortunately, however, elective C-sections do not prevent against in utero infection. Furthermore, the current studies were conducted prior to the advent of highly-active antiretroviral therapy (HAART). Thus, it is unknown whether HIV transmission rates after maternal use of potent combination antiretroviral therapy would be further reduced by elective surgical deliveries.

In terms of vaginally delivered infants, rates of transmission are increased in deliveries in which episiotomy, scalp electrodes, forceps, or vacuum extractors were used, but only in those centers where these procedures are not routinely performed.[16]

Factors related to the fetus. A low birth weight of less than 2500g[51,61,71] and/or gestational age <34 weeks[16] or <38 weeks[51,71] have each been associated with an increased risk of perinatal HIV transmission. This association between prematurity and transmission may simply be a consequence of HIV infection in utero, resulting in abnormalities in fetal development and premature birth. Additionally, women with more advanced HIV disease, who are more likely to transmit to their infants, are also more likely to deliver premature infants. It is also possible that infants born prematurely may have less developed immune systems, and thus be more susceptible to infection during labor and delivery. One way to address this issue is to determine the timing of infection. Recent studies of this type in HIV-exposed infants have suggested that the rate of intrapartum transmission might be higher among premature babies, implying that acquisition of infection occurred because they were born prematurely and not vice versa.[72,73]

Aside from prematurity or low birth weight, the birth order of twins has also been associated with differing risks of infection. Both Goedert and colleagues[25] and Duliege and associates[65] have shown that the majority of twin pairs are concordant in terms of HIV-1 infection. However, a significantly increased risk of HIV-1 infection is also apparent among first-born ("A") as opposed to second-born ("B") twins, independent of mode of delivery. Thus, as shown by Duliege, HIV-1 infection occurred in 35% of vaginally-delivered "A" twins versus 15% of vaginally-delivered second-born twins. In twin pairs born after C-section delivery, the risk of HIV infection was 16% among "A" twins, and 8% among "B" twins.[65] Although the precise mechanisms for this increased risk in first-born twins are unknown, it is certainly consistent with the fact that perinatal infection often occurs at the time of delivery.

Table IV: Perinatal HIV Transmission Rates in Twins[65]

Mode of delivery

Perinatal transmission rates

First-born "A" twins

Second-born "B" twins

Vaginal delivery



C-section delivery



Role of breastfeeding. HIV-1 has been isolated from cell-free breast milk,[74] and HIV DNA has been demonstrated in the majority of milk specimens from HIV-infected mothers. Thus, Ruff and colleagues detected HIV DNA in 70% of milk specimens collected from 47 HIV-seropositive women from 0 to 4 days postpartum, and in about 50% of specimens collected from 6 to 12 months postpartum.[75] Evidence of HIV transmission via breastfeeding has come from reports in which children were exposed to HIV-1 only through breastfeeding,[76] and from studies in which breastfed infants had an increased risk of perinatal HIV transmission, when compared with children who were bottlefed.[9,16,19,77]

While the potential for HIV transmission through maternal milk is clearly evident from these studies, the actual risk for such transmission has been uncertain. Studying a group of mothers with primary HIV infection, Van de Perre and colleagues found that at least 4 of 11 (36%) infants became infected with HIV through breastfeeding.[34] Plasma viral load is known to be high during primary HIV infection,[21] making it difficult to assess the applicability of these data to other stages of HIV infection. In fact, a meta-analysis of five different studies of postnatal transmission reported an excess transmission risk by breastfeeding of 14% for women with established HIV infection, and an excess risk of 29% among women who developed primary infection during the postpartum period.[78]

De Martino and coworkers studied a group of 168 breast-fed and 793 bottlefed children born to seropositive mothers in Italy.[79] Breastfeeding was shown to increase the risk of HIV-1 transmission, with an estimated adjusted odds ratio for 1 day of breastfeeding versus bottlefeeding of 1.19 (95% CI, 1.10-1.28). The odds ratio of transmission increased with the duration of breastfeeding, a finding also reported by Nagelkerke and colleagues.[80]

Recent work from Malawi in Africa has demonstrated a clear relationship between duration of breastfeeding and the likelihood of HIV transmission to the infant. During months 2 to 6 postpartum, the incidence of transmission through breastfeeding was 0.7% per month. During later months, the risk of transmission decreased slightly but was still present, at 0.6% per month from 6 to 11 months, and 0.3% per month from 12 to 18 months. Thus, the cumulative risk of HIV transmission was 3.5% at 5 months, 7.0% at 11 months, 8.9% at 17 months, and 10.3% at 23 months of breastfeeding.[81]

Key Points

  • Perinatal HIV transmission may occur in utero, during delivery, or through breastfeeding.
  • Higher maternal viral load and premature rupture of the amniotic membranes are independently associated with an increased risk of HIV transmission to the infant.
  • In studies conducted prior to the widespread use of HAART during pregnancy, elective C-section has been associated with a reduced risk of perinatal transmission.
  • There is a linear relationship between duration of breastfeeding and the likelihood of postnatal HIV transmission.


Case Study (continued)
The patient is begun on antiretroviral therapy, including zidovudine and lamivudine given as Combivir (1 tablet twice daily) and indinavir (800 mg every 8 hours). A repeat CD4+ cell count 3 months later reveals an increase to 560 cells/mm3, while the HIV-1 RNA has dropped to below the limit of quantitation (<200 copies/mL). The patient is maintained on this regimen.

In January 1997, the patient is seen for nausea and vomiting of approximately 3 weeks' duration. Physical examination and laboratory analysis subsequently confirm that the patient is pregnant. After discussing the pregnancy with her family and boyfriend, the patient decides to keep the child. The patient is subsequently educated and counseled regarding her options for use of antiretroviral therapy during the pregnancy and delivery. Since very little information is available regarding use of antiretroviral agents during the first trimester of pregnancy, and since the patient's viral load has remained nondetectable, she decides to stop all antiretroviral drugs until week 14. She is advised to stop cigarette smoking, at least during the remainder of the pregnancy.