To determine and compare the clinical and immunologic outcomes for HIV-infected women initiated on antiretroviral therapy (ART), with and without previous exposure to single-dose nevirapine in the MTCT-Plus programme – Kampala, Uganda, from 2003 to 2011.
Retrospective comparison of prospectively collected programmatic data of clinical and immunologic treatment outcomes among HIV-infected Ugandan women, with and without prior exposure to sdNVP, who received NNRTI-based ART for a median follow-up of 6 years.
Of the 408 women in the programme, 289 (70.8%) were started on ART, of whom 205 (70.9%) had prior exposure to sdNVP. Clinical, immunologic and combined (clinical and or immunologic) treatment failure occurred in 29 (10.0%), 132 (45.7%) and 142 (49.1%) women, respectively. There was no significant difference in the distribution of time to immunologic failure for women by exposure to sdNVP (log-rank P = 0.98). In Cox proportional hazard modelling, exposure to sdNVP was not associated with immunologic failure [adjusted hazard ratio (HR) = 0.89, 95% confidence interval (CI): 0.61–1.30]. CD4 count >100 cells/mm3 at initiation was associated with reduced incidence of immunologic failure in adjusted analyses (HR = 0.32, 95% CI: 0.22–0.48).
HIV-infected Ugandan women initiated on an NVP-based ART regimen had similar immunologic treatment outcomes irrespective of previous NVP exposure. CD4 cell count prior to initiating HAART was a key prognostic factor for successful long-term immunologic treatment outcomes. In poor settings, regular follow-up of patients on HAART with adequate counselling to promote adherence and safe disclosure may promote low clinical failure rates.
Déterminer et comparer les résultats cliniques et immunologiques des femmes infectées par le VIH engagées dans le traitement antirétroviral, avec et sans exposition antérieure à la dose unique de névirapine (sdNVP) dans le Programme sur la Transmission Mère-Enfant Plus – Kampala, en Ouganda, de 2003 à 2011.
Comparaison rétrospective de données programmatiques recueillies de façon prospective des résultats des traitements cliniques et immunologiques de femmes infectées par le VIH en Ouganda, avec et sans exposition préalable à sdNVP, qui ont reçu un traitement antirétroviral à base d'inhibiteur non nucléotidique de la transcriptase inverse (INNTI) et un suivi médian de six ans.
Sur les 408 femmes dans le programme, 289 (70,8%) ont été mises sous ARV, dont 205 (70,9%) avec une exposition antérieure à sdNVP. L’échec clinique, immunologique et combiné (clinique et/ou immunologique) du traitement est survenu chez 29 (10,0%), 132 (45,7%) et 142 (49,1%) femmes, respectivement. Il n'y avait pas de différence significative dans la distribution de la durée à l’échec immunologique pour les femmes selon l'exposition à sdNVP (log-rank P = 0,98). Dans la modélisation des risques proportionnels de Cox, l'exposition à sdNVP n’était pas associée à un échec immunologique [rapport de risque ajusté (RR) = 0,89, intervalle de confiance 95% (IC): 0,61 à 1,30]. Un taux de CD4 > 100 cellules/mm3 à l'initiation était associé à une incidence réduite de l’échec immunologique dans les analyses ajustées [HR = 0,32, IC95%: 0,22 à 0,48].
Les femmes infectées par le VIH en Ouganda, engagées dans un schéma thérapeutique à base de NVP avaient les mêmes résultats immunologiques thérapeutiques, quelle que soit l'exposition précédente à la NVP. Le taux de CD4 avant le début du HAART était un facteur clé pour un pronostic positif à long terme des résultats immunologiques du traitement. Dans les pays pauvres, un suivi régulier des patients sous HAART avec des conseils appropriés pour promouvoir l'adhésion et la sécurité de la communication peuvent favoriser de faibles taux d’échecs cliniques.
Determinar y comparar los resultados clínicos e inmunológicos de mujeres infectadas con VIH que han iniciado terapia antirretroviral, con y sin exposición previa a la nevirapina, dentro del programa MTCT-Plus en Kampala, Uganda entre el 2003 y el 2011.
Comparación retrospectiva de datos del programa recogidos de forma sobre los resultados clínicos e inmunológicos de mujeres infectadas con VIH en Uganda, con y sin una exposición previa a sdNVP, que recibieron terapia antirretroviral basada en inhibidores no nucleosídicos de la transcriptasa (NNRTI) durante un periodo de seguimiento de 6 años de mediana.
De las 408 mujeres participando en el programa, 289 (70.8%) habían comenzado TAR, y de ellas 205 (70.9%) habían tenido una exposición previa a sdNVP. Hubo fallo del tratamiento a nivel clínico, inmunológico y combinado (clínico e inmunológico) en 29 (10.0%), 132 (45.7%) y 142 (49.1%) de las mujeres, respectivamente. No había una diferencia significativa en la distribución del tiempo al fallo inmunológico para mujeres con exposición a sdNVP (log-rank P = 0.98). En el modelo de riesgos proporcionales de Cox, la exposición a sdNVP no estaba asociada a un fallo inmunológico [cociente de riesgos instantáneos ajustado (HR) = 0.89, IC 95%: 0.61–1.30]. El conteo de CD4 > 100 células/mm3 al comienzo estaba asociado con una incidencia reducida de fallo inmunológico en el análisis ajustado [HR = 0.32, IC 95%: 0.22–0.48].
Las mujeres de Uganda infectadas con VIH que iniciaron un TAR basado en NVP tenían resultados inmunológicos similares, independientemente de su exposición previa a la NVP. El conteo de CD4 anterior al inicio de la Terapia Antirretroviral Altamente Activa (TAAA) era un factor clave de prognosis de unos resultados inmunológicos exitosos a largo plazo. En emplazamientos pobres, un seguimiento regular de los pacientes que están recibiendo TAAA, con un asesoramiento adecuado que promueva la adherencia y una divulgación segura, podrían promover unas bajas tasas de fallo clínico.
Uganda's prevention of mother-to-child transmission (PMTCT) of HIV guidelines evolved over time. More efficacious antiretroviral therapy (ART) regimens have replaced sdNVP, which was promoted from 2000 to 2005 (MOH Uganda 2009). But progress towards full coverage of PMTCT has been slow; by the end of 2009, only 53% of women and their infants accessed PMTCT services and only half of these received combination short-course ART. The low PMTCT coverage, coupled with late first antenatal visits and drug stock-outs, may result in some women receiving only sdNVP instead of more efficacious regimens recommended for PMTCT.
Exposure to sdNVP prior to ART initiation has been associated with poor treatment outcomes in some but not all studies, particularly when women initiate treatment within the first 12-18 months after sdNVP exposure (Jourdain et al. 2004; Lockman et al. 2007). Benjamin Chi et al. (2007) found no difference in mean CD4 cell change by exposure to NVP at 6 and 12 months after HAART initiation (Benjamin Chi et al. 2007). A related study in Côte d'Ivoire also showed similar increases in CD4+ T cell count after 36 months of follow-up in women irrespective of sdNVP exposure.
We analysed data for women on ART enrolled in the MTCT-Plus family care programme at Mulago National Referral Hospital – Kampala to observe the impact of sdNVP on long-term immunologic and clinical treatment outcomes within the Ugandan context where sdNVP has been widely used. Many women received sdNVP for PMTCT, and others did not, providing an opportunity to study differences in treatment outcomes associated with sdNVP exposure. The objective was to determine the immunologic failure and combined clinical and immunologic failure after initiating ART treatment.
This was a retrospective analysis using data from a longitudinally followed Uganda cohort of HIV-infected women to evaluate the clinical, immunologic and combined ART treatment outcomes of HIV-infected women within the context of scaling up of PMTCT 2003–2011.
Study setting and population
At Mulago Hospital, HIV-infected women identified through the PMTCT programme or perinatal studies were referred to the MTCT-Plus programme for HIV/AIDS care. The programme was run by Makerere University–Johns Hopkins University (MU-JHU) Research Collaboration with technical support from Columbia University, USA (Toro et al. 2012). From March 2003 to October 2007, HIV-infected women and their families were enrolled into the MTCT-Plus programme to receive long-term HIV/AIDS care. At enrolment, demographic, social, clinical, and laboratory data were collected. For this analysis, women had to have initiated ART by 31 December 2010. We evaluated immunologic and clinical outcomes of 289 women during the follow-up period March 2003 to 31 December 2011.
Exposure to single-dose nevirapine
Uganda PMTCT guidelines evolved during this study. The exposed women received a 200-mg tablet of sdNVP for PMTCT at onset of labour (i) alone when HIV infection was identified during labour, (ii) after receiving short-course zidovudine (ZDV) (300 mg bd), starting at 28–36 weeks'; gestation, or (iii) after receiving short-course ZDV plus 3TC (150 mg bd), initiated from 3 weeks of gestation and continued until 7 days post-delivery. Details on the procedure of how sdNVP was administered in the PMTCT programme have been previously described (Namukwaya et al. 2011).
Participants initiated ART in accordance with World Health Organization (WHO) and Ugandan Ministry of Health (MOH) ART eligibility criteria (MOH Uganda 2003): WHO stage 4 disease regardless of CD4+ T-lymphocyte count, or WHO stage 3 plus CD4+ T-lymphocyte count <350 cells/mm3, or CD4+ T-lymphocyte count <200 cells/mm3, or Tuberculosis and CD4 cell count between 200 and 350 per mm3. First-line HIV treatment was a backbone of ZDV (300 mg) and lamivudine (3TC) (150 mg) bd or 3TC (150 mg) and stavudine (D4T) (30 mg) twice daily, plus NVP (200 mg) od for 2 weeks escalated to bd or efavirenz (EFV) (600 mg) od. Twenty-five (8.7%) patients started on D4T switched to ZDV or tenofovir (TDF) (300 mg) od in accordance with WHO and Ugandan MOH revised guidelines for ART treatment (MOH Uganda 2003, 2008).
At monthly visits, patients were seen by a nurse for vital signs and a counsellor to discuss psychosocial support, disclosure status and adherence. At the same visit, the physician would obtain an interval history, perform a physical examination, record new diagnoses, and encourage the patients to return in case of intercurrent illness or side effects. Data were recorded in patient charts and entered into a central database by trained clerks after each visit. For women who missed visits, telephone calls or home visits were made by health visitors to establish reasons. If a death occurred outside the hospital, the health visitor would conduct a verbal autopsy. The CD4 cells were counted at enrolment and every 6 months thereafter using the BD FACS Caliber instrument (Becton, Dickinson & Company). HIV viral load and drug resistance testing were not performed in the MTCT-Plus programme. All laboratory tests were performed at MU-JHU Core Laboratory, Uganda, which is certified by the College of American Pathologists.
Adherence was assessed by patients'; self-reports of number of pills ingested during the last 6 days prior to clinic visit. Adherence was reinforced with patient-targeted, pre-approved unannounced home visits. The majority of patients adhered to their regime well (≥95% of pills) (Byakika-Tusiime et al. 2009). Patients with poor adherence (<95%) were offered special adherence counselling to supplement routine counsellors'; visits.
The primary outcomes were clinical, immunologic and combined treatment failure. Clinical failure was defined as a new OI of WHO stage 3 or 4, and/or death at least 6 months after initiating ART with assumption of good adherence. Death was ascertained from verbal autopsy or hospital records. Immunologic failure was defined as: (i) fall of CD4 cell count to baseline, (ii) 50% fall from on-treatment peak or (iii) persistent CD4 cell count 100 cells/mm3 at least 6 months after ART initiation. Combined failure was defined as having either or both clinical and immunologic failures.
Patients'; characteristics were compared between groups (sdNVP-exposed verses unexposed) using Pearson's chi-square (or Fisher's test for sparse data) for categorical variables and t-tests (Wilcoxon rank-sum test if non-normal distributions) for continuous variables. Kaplan–Meier estimates were used to describe distributions of time-to-event endpoints. If a patient experienced more than one event, time was counted to the first occurrence. All available follow-up data were included, censoring at loss to follow-up, transfer out of the programme or on 31 December 2011, whichever came first. HR estimates from Cox regression were used to quantify differences in outcomes by previous sdNVP exposure during PMTCT, in both unadjusted and multiple regression models. Variables with a P-value <0.2 or previously known to have clinical association with treatment failure were considered for multiple Cox regression analysis (previous sdNVP exposure, baseline CD4 count, WHO stage, parity and disclosure to partner). All testing was two-sided at 5% significance level. HRs were reported with 95% confidence intervals (CI). Analyses were performed in SAS, version 9.1, and Intercooled STATA, version 10.
After removing patient identifiers from collected clinical care data, the study was exempted from human subject review by the Uganda National Council of Science and Technology and University of Medicine and Dentistry of New Jersey Institutional Review Board.
From 2003, 408 women were followed up in MTCT-Plus programme, and by 31 December 2010, 289 women had initiated ART and were followed for median duration of 6.0 years [interquartile range (IQR) 3.6–7.1]. Seventy-one per cent of 205 women on ART were previously exposed to sdNVP for PMTCT and 29% (84) were not (Figure 1). 3.4% (14) of the women were lost to follow-up during the study period. Women exposed to sdNVP were younger than women who were not exposed (P = 0.003). More unexposed women were classified as WHO stage 3 and 4 than sdNVP-exposed women (P = 0.009). There were no significant differences in sdNVP-exposed and unexposed women's baseline CD4 cell counts and other baseline characteristics (Table 1). At ART initiation, women exposed to sdNVP had a median (IQR) duration of 27.9 (9.0–56.8) months since NVP exposure; 45.4% (93) of women had ≤24 months and 54.6% (112) had >24 months since last exposure. The vast majority (96.9%) of women initiated NVP-based ART regimens.
Table 1. Baseline characteristics for HIV-infected women (N = 289) enrolled in the Mulago MTCT-Plus programme April 2003 to October 2007 and initiated on antiretroviral therapy by December 2010
Parity (number of pregnancies) at enrolment: median (IQR) years
Marital status (n = 286): n (%)
Duration since last NVP exposure to initiation of treatment (months): median (IQR)
Among the 84 women unexposed to sdNVP, 7 (8.3%) experienced new OIs and three (3.6%) died. One woman who developed a new OI died later; therefore, nine (10.7%) sdNVP-exposed women experienced clinical failures. In comparison, among the 205 women exposed to sdNVP, 16 (7.8%) experienced new OIs and 7 (3.4%) died; three of those with new OIs died; 20 (9.8%) had a clinical event. The clinical failure rates did not differ by sdNVP exposure (log-rank P = 0.88). The OIs included severe bacterial infection (9, 39.1%), pulmonary tuberculosis (5, 21.7%), grade IV wasting (3, 13.0%), Kaposi's sarcoma (2, 8.7%), extrapulmonary tuberculosis (2, 8.8%), suspected disseminated candidiasis (1, 4.3%) and Pneumocystis jiroveci pneumonia (PCP) (1, 4.3%).
Immunologic (CD4 cell count) failure
In total, 132/289 women, including 90/205 (43.9%) of sdNVP-exposed and 42/84 (50.0%) unexposed women, experienced immunologic failure (log-rank P = 0.33). Five of the 10 deaths occurred subsequent to immunologic failure. Among 90 sdNVP-exposed women with immunologic failure, 62 (68.9%) resulted from CD4 count dropping to half the peak, 26 (28.9%) from CD4 count returning to baseline level, and two (2.2%) from CD4 dropping to <100 cells/mm3. Of the 42 sdNVP non-exposed women classified as failing immunologically, 29 (69.0%) failed due to CD4 cell level dropping to half the peak, eight (19.0%) from CD4 count returning to baseline level, and five (12.0%) from to CD4 count dropping to <100 cells/mm3.
There was no significant difference in distributions of time to immunologic failure for women exposed and unexposed to sdNVP. For sdNVP-exposed women, the Kaplan–Meier estimates of the percentages with immunologic failure were 15.7%, 34.0%, 48.8% and 51.0% at 1, 3, 5 and 6 years compared with 23.8%, 44.6%, 51.1% and 51.1%, respectively, of unexposed women (log-rank P = 0.88) (Figure 2). The median (IQR) CD4 cell count at 6-year follow-up was 466 (324–614) cells/mm3 for the sdNVP-exposed group compared with 517 (332–711) cells/mm3 for unexposed group (P-value = 0.22).
In Cox models, exposure to sdNVP was not associated with increased risk of immunologic failure [unadjusted HR 0.86 (95% CI: 0.59–1.24)] even after adjusting for baseline CD4 cell count, WHO stage, age and parity of woman [adjusted HR 0.91 (95% CI: 0.62–1.33)] (Table 2). Women with baseline CD4 counts <100 cell/mm3 were at higher risk of immunologic failure. WHO stage 3 and 4 and parity >3 were associated with increased hazard of immunologic failure before but not after adjustment.
Table 2. Cox regression analysis to predict immunologic or combined clinical and immunologic outcome in HIV-infected women enrolled in the Mulago MTCT-Plus programme, April 2003 to October 2007 and initiated on antiretroviral therapy by December 2010 using follow-up through December 2011 (n = 289)
Combined clinical and immunologic outcome
ref, reference category; CI, confidence interval.
We adjusted for variables with P < 0.2 in bivariate analyses or those known to have association with immunologic and clinical failure.
Previous sdNVP exposure
Non NVP exposed
Baseline CD4 (cells/mm3)
WHO stage category
1 and 2
3 and 4
Mother's age in years
Disclosed to partner
Among sdNVP-exposed women, the risk of immunologic failure was smaller in the 113 women with remote (>24 months) exposure to sdNVP than in those with more recent exposure [HR = 0.56 (95% CI: 0.36–0.87), P = 0.009]. This risk of failure was less significant (HR = 0.71 (95% CI: 0.45–1.12), P = 0.14] after adjusting for baseline CD4 cell count. Similar findings were observed when using 12 and 18 months as cut-off values for defining remote exposure.
A total of 142/289 women failed treatment based on either immunologic or clinical failure, which only included 10 additional endpoints over the immunologic failure endpoint. Among the sdNVP-exposed women, 98/205 had combined failure, whereas 44/84 unexposed (log-rank P = 0.35) did. For sdNVP-exposed women, the Kaplan–Meier estimates of the percentage with combined failure were 19.3%, 35.8%, 50.7% and 54.7%; they were 27.1%, 46.4%, 52.8% and 52.8% of unexposed women at 1, 3, 5 and 6 years, respectively (log-rank P = 0.98) (Figure 3). Cox modelling showed no significant difference in combined outcomes between groups (Table 2).
For sdNVP-exposed women, risk of combined failure was smaller in those who had remote exposure (>24 month) to sdNVP than for women with more recent exposure [HR = 0.66 (95% CI: 0.44–0.99), P = 0.048] but not after adjusting for baseline CD4 cell count [HR = 0.71 (95%CI: 0.55–1.32), P = 0.48].
About half of the women initiated on a NNRTI-containing regimen experienced treatment failure 6 years after ART initiation by immunologic or combined immunologic and clinical criteria. Failure rates and distributions were similar using the two outcomes because of the small numbers of women who failed clinically without first failing immunologically. There was no difference in the distributions of immunologic or combined failure in women by previous exposure to sdNVP for PMTCT. Our findings are similar to previous studies that found no difference in rates of failure by sdNVP exposure after 1 year of ART (Benjamin Chi et al. 2007; Ekouevi et al. 2010).
Of importance, the data demonstrate a high rate of immunologic and combined failure, which is similar to that shown in a large antiretroviral treatment programme in Nigeria (Rawizza et al. 2011), but differs from other studies that demonstrated only mild declines in CD4+ counts (Lifson et al. 2011; Mugyenyi et al. 2010). The difference noted in our study may have arisen from the variation in study populations, ART regimen/previous exposure to sdNVP and definition of immunologic treatment outcomes. For example, unlike in the DART trial (Mugyenyi et al. 2010), our participants were all female. Few (3%) of the DART participants had previous exposure to sdNVP: only 7% initiated NVP-based regimens, with the majority (85%) on ZDV, 3TC plus TDF or abacavir, whereas in our study >71% were sdNVP exposed, 96% initiated on NVP-based ART regimen, and none started on PI or triple NRTI-based ARVs. Despite all these differences, our results of clinical outcomes were similar to those in DART trial. Our study is unique in that we have substantially longer follow-up compared with most of the previous studies.
NNRTI treatment failure among HIV-infected women exposed to sdNVP depends in part on the time interval since exposure to ART initiation. Women starting NNRTI-based regimens shortly after sdNVP exposure are at higher risk of immunologic failure than those who delayed initiation (Benjamin Chi et al. 2007; Ekouevi et al. 2010). A prolonged interval between sdNVP exposure and treatment initiation (median of 26 months in our study) may be responsible for lack of a difference in failure rates between groups, as waning of resistance mutations occurs over time. Reduction in risk of immunologic failure in women with remote exposure (≥24 months) may have resulted from the strong confounding effect of the CD4 count at ART initiation.
Our findings are consistent with previous studies conducted in HIV-infected women starting HAART after exposure to sdNVP and followed for shorter durations (≤2 years; Benjamin Chi et al. 2007; Ekouevi et al. 2010; Jourdain et al. 2004). Currently, WHO adult HIV treatment guidelines recommend initiation of NNRTI-based ART in women exposed to sdNVP, with or without an NRTI tail, only if the duration since prior exposure is >12 months and then monitoring with viral load testing at 6 months (WHO 2009). In case viral load testing is not available, WHO recommends continued monitoring using clinical criteria plus CD4 cell count. Because viral load testing was not routinely conducted in Uganda during the study time frame, we cannot ascertain when failing women might have first demonstrated evidence of virologic failure. Nevertheless, our long-term data demonstrate that women who initiate NNRTI-based ART after sdNVP exposure, with remote duration since sdNVP exposure, may have similar longer-term clinical and immunologic outcomes as unexposed women. Even after adoption of WHO-PMTCT option B-plus, PMTCT coverage in Uganda falls short of nationwide coverage. Hence, these results are useful in Uganda, where large number of HIV-infected women may have received only sdNVP, with or without NRTI tail, for PMTCT and later initiate NNRTI-based ART.
There were certain limitations to this study. It was observational and non-randomised, so there may have been differences in risk factors for disease progression that could have affected the results. We tried to address this concern by adjusting for known risk factors using Cox regression models while comparing women who did or did not receive sdNVP. Also, integrity of data recorded collected mainly for patient management may be less robust than clinical trial data due to less strict quality control of recorded data, and/or ascertainment of outcomes. Further, given the sample size of women available over 6 years follow-up, comparison of treatment outcomes between sdNVP-exposed and unexposed women may have been underpowered to detect small differences; the sample size of 289 had 76% power of detecting HR of 2.0 between groups, at the 5% significance level.
Given these caveats, there were a number of strengths to the study including longer follow-up compared with previous studies, good self-reported adherence to taking ARVs and high retention. We attribute this high retention and adherence to the MTCT-Plus family model of care, which enhanced disclosure and family support in other settings. This study also provides insight into long-term treatment outcomes of HIV-infected women in poor countries similar to Uganda.
Summary and Conclusions
After a median of 6-year follow-up after initiation of NNRTI-based treatment regimens, half of the women enrolled in a family model of HIV care failed immunologically, with similar failure rates irrespective of the women's prior exposure to sdNVP. This study also highlights that CD4 cell count prior to initiating ART is a key prognostic factor for long-term HAART treatment outcomes. Use of viral load as it becomes more available will be useful in elucidating reasons for high failure rates in this and similar cohorts. With viral load and ART resistance testing for monitoring treatment outcomes still limited in resource constrained settings, ART programmes should have regular follow-up of patients on HAART and have adequate counselling to promote treatment adherence and safe disclosure to minimise failure rates.
We would like to thank the clients and the entire MTCT-Plus programme staff at MUJHU. Deep appreciation is extended to Columbia University for the financial and technical support provided to the MTCT-Plus Initiative programme since its inception, particularly Drs. Elaine Abrams and Patricia Toro. Our gratitude goes to Dr. James Koola and the IMPAACT International Resource Committee Program for their support of New Investigators, which helped support this work.