SummaryBackgroundNeonates born to women infected with HIV are at increased risk for invasive group B streptococcus (GBS) disease. We aimed to compare safety and immunogenicity of trivalent glycoconjugate GBS vaccine in pregnant women with and without HIV in Malawi and South Africa.MethodsIn our non-randomised phase 2, open-label, multicentre study, we recruited pregnant women attending two antenatal clinics, one in Blantyre, Malawi, and one in Soweto, Johannesburg, South Africa. Participants were divided into three groups on the basis of their HIV infection status (no infection, infection and high CD4 cell count [>350 cells per μL], and infection and low CD4 cell count [>50 to ≤350 cells per μL]) and received a 5 μg dose of glycoconjugate GBS vaccine (serotypes Ia, Ib, and III, with CRM197 [Novartis Vaccines, Siena, Italy]) intramuscularly at 24–35 weeks' gestation. GBS serotype-specific antibody concentrations were measured before vaccination (day 1), day 15, day 31, and at delivery, and in infants at birth and day 42 of life. The primary outcomes were safety in mothers and infants and the amount of placental transfer of GBS serotype-specific antibodies from mothers to their infants. All immunogenicity and safety analyses were done on the full analysis set, including participants who, or whose mother, correctly received the vaccine and who provided at least one valid assessable serum sample. This study is registered with ClinicalTrials.gov, number NCT01412801.Findings270 women and 266 infants were enrolled between Sept 26, 2011, and Dec 4, 2012 (90 women and 87 infants without HIV, 89 and 88 with HIV and high CD4 cell counts, and 91 and 91 with HIV and low CD4 cell counts, respectively). Seven women were lost to follow-up, six withdrew consent, one died, and two relocated. Eight infants died or were stillborn and two were lost to follow-up. Across serotypes, fold change in antibody concentrations were higher for the HIV-uninfected group than the HIV-infected groups. Transfer ratios were similar across all three groups (0·49–0·72; transfer ratio is infant geometric mean antibody concentration in blood collected within 72 h of birth divided by maternal geometric mean antibody concentration in blood collected at delivery); however, at birth, maternally derived serotype-specific antibody concentrations were lower for infants born to women infected with HIV (0·52–1·62 μg/mL) than for those born to women not infected with HIV (2·67–3·91 μg/mL). 151 (57%) of 265 women reported at least one solicited adverse reaction: 39 (45%) of 87 women with HIV and low CD4 cell counts, 52 (59%) of 88 women with HIV and high CD4 cell counts, and 60 (67%) of 90 women in the HIV-uninfected group. 49 (18%) of 269 women had at least one adverse event deemed possibly related to the vaccine (six [7%] in the HIV and low CD4 cell count group, 12 [13%] in the HIV and high CD4 cell count group, and 21 [23%] in the HIV-uninfected group), as did three (1%) of 266 neonates (zero, two [1%], and one [1%]); none of these events was regarded as serious.InterpretationThe vaccine was less immunogenic in women infected with HIV than it was in those not infected, irrespective of CD4 cell count, resulting in lower levels of serotype-specific maternal antibody transferred to infants, which could reduce vaccine protection against invasive GBS disease. A validated assay and correlate of protection is needed to understand the potential protective value of this vaccine.FundingNovartis Vaccines and Diagnostics division (now part of the GlaxoSmithKline group of companies), Wellcome Trust UK, Medical Research Council: Respiratory and Meningeal Pathogens Research Unit.