Incidence of Severe Neutropenia in HIV-Infected People Starting Antiretroviral Therapy in West Africa

Background In sub-Saharan Africa, antiretroviral therapy (ART) including drugs with potential toxicity such as Zidovudine (ZDV) are routinely prescribed. This study aimed at estimating the incidence of severe neutropenia and associated factors after ART initiation in five West African countries. Methods A retrospective cohort analysis was conducted within the international epidemiologic database to evaluate AIDS (IeDEA) collaboration in West Africa. All HIV-infected adults, initiating ART between 2002 and 2014, with a baseline and at least one follow-up absolute neutrophil count (ANC) measurement were eligible. Incidence of severe neutropenia (ANC <750 cells/mm3) was estimated with 95% confidence interval (CI) according to age, gender, HIV clinic, hemoglobin, CD4 count, clinical stage, and ART duration. A Cox proportional hazard model was used to identify factors associated with severe neutropenia, expressed with their adjusted hazard ratios (aHR). Results Between 2002 and 2014, 9,426 HIV-infected adults were enrolled. The crude incidence rate of a first severe neutropenia was 9.1 per 100 person-years (95% CI: 8.6–9.8). Factors associated with severe neutropenia were exposure to ZDV <6 months (aHR = 2.2; 95% CI: 1.8–2.6), ≥6–12 months (aHR = 2.1; 95% CI: 1.6–2.8) and ≥12 months (aHR = 1.6; 95% CI: 1.2–2.2) [Ref. no ZDV exposure], CD4 count <350 cells/mm3 (aHR = 1.3; 95% CI: 1.1–1.5) and advanced clinical stage at ART initiation (aHR = 1.2; 95% CI: 1.0–1.4). Conclusion The incidence of severe neutropenia after ART initiation in West Africa is high and associated with ZDV exposure and advanced HIV disease. In this context, efforts are needed to scale-up access to less toxic first-line ART drugs and to promote early ART initiation.

Introduction Sub-Saharan Africa is by far the most affected region of the world by the HIV pandemic, with 25.8 million of people living with HIV (69.9% of the total) [1]. However, during the last decade, access to antiretroviral treatment (ART) has substantially improved in this part of the world and in 2014, 10.7 million of HIV-infected people (41.5% of the world total) were receiving ART in sub-Saharan Africa, an increase of 20% from 2010 [2]. These figures (absolute number and proportion) are likely to increase thanks to the recent World Health Organization (WHO) guidelines which recommends to initiate ART in all adults living with HIV regardless of CD4 cell count [3].
Facing an unprecedented number of people initiating ART in sub-Saharan Africa, the monitoring of treatment-related toxicity must be adapted to inform recommendations for the use of ART. Indeed, in low and middle-income countries long-term safety data of ART is still limited due to the lack of pharmacovigilance systems. Nevertheless, adverse events related to ART are common and represent the first cause of treatment modification in HIV-infected people on ART in sub-Saharan Africa [4,5]. The WHO classifies ART related toxicities into four grades of severity. In case of severe toxicity (grade [3][4] it is usually recommended to remove the incriminated drug. Mild or moderate neutropenia are usually asymptomatic but when severe, can lead to life-threatening infectious conditions and thus fit the grade 3-4 toxicity classification. Several studies have shown an increased risk of bacterial and fungal infections in HIV-infected people with neutropenia, most of them being conducted in the pre-ART era in industrialized countries [6][7][8][9][10]. In addition, a study has also revealed an increased risk of mortality associated with neutropenia in Spanish patients living with HIV [11]. Etiological factors leading to neutropenia in patients with HIV infection are multiple. Indeed, neutropenia can occur as a direct consequence of HIV infection and its related immunodeficiency, usually in the late stages of the HIV disease [12]. Another cause of neutropenia in people living with HIV is the use of myelotoxic agents. Despite the WHO 2013 guidelines recommending to treat everyone with tenofovir (TDF) in first-line [13], zidovudine (ZDV), one of the first-line antiretroviral drugs recommended since 2000, is still massively used.
There have been several studies reporting the impact of ZDV on developing neutropenia in a context of no or limited access to ART [14][15][16][17]. Other drugs such as cotrimoxazole (CTX), which is part of the minimal package of care for the prevention of opportunistic infection in people living with HIV, has also been shown to be a common cause of neutropenia [17][18][19].
In West Africa, where ZDV is still the usual drug in first-line ART regimens, there is a need to document the occurrence of hematologic toxicity in real-life conditions. The main objective of this study was to estimate the incidence of severe neutropenia and associated factors in people living with HIV initiating ART in West Africa.

Study population
The present analysis was conducted within the International Epidemiologic Databases to Evaluate AIDS (IeDEA) in West Africa consortium (www.iedea.org/) that aims at collecting a large set of data necessary for clinical and epidemiological research in the context of ART use [20]. This collaboration currently involves 17 HIV clinics from nine countries: Benin, Burkina Faso, Côte d'Ivoire, Ghana, Guinea-Bissau, Mali, Nigeria, Senegal and Togo. The IeDEA West Africa database maintains individual records related to socio-demographic, clinical and biological data of patients at ART initiation and during follow-up visits in participating health facilities. In each health center, patients are expected to have full blood count and selected biological tests at ART initiation and every six months thereafter. Data generated by these HIV clinics are reported, once a year, to the regional center based in Abidjan, Côte d'Ivoire using a standardized data format allowing data merging.
Adult patients (age !16 years) living with HIV and initiating a first-line ART regimen between 2002 and 2014 that contained three or more drugs, had an absolute neutrophil count (ANC) >750 cells/mm 3 within six months prior to ART initiation and at least one follow-up ANC in the first six months on ART were included. HIV clinics not reporting ANC within their database were excluded from this project, leading a total of eight HIV clinics participating in this analysis: the Centre National Hospitalier Universitaire de Cotonou (CNHU) in Benin, the Centre Hospitalier Universitaire Souro Sanou of Bobo Dioulasso (BOBO) and the Service

Data collection and outcome definition
For the present analysis, characteristics extracted from the IeDEA West Africa database were gender, date of birth, height, weight, ART regimen, clinical staging (either WHO or CDC definition), CD4 cell count and full blood count. Baseline was defined as the date of ART initiation.
The primary outcome of this study was severe neutropenia. Neutropenia was classified according to the NIH Division of AIDS scale: grade 1 (ANC <1,300 cells/mm 3 ), grade 2 (ANC <1,000 cells/mm 3 ), grade 3 (ANC <750 cells/mm 3 ), and grade 4 (ANC <500 cells/mm 3 ) [21]. Severe neutropenia was defined as an ANC 750 cells/mm 3 (grade 3 or 4). The main exposure was ART and more specifically the use of ZDV. Exposure to ZDV was considered as a timeupdated variable in order to take into account ART modifications over time. The cumulated exposure to ZDV was then classified into four categories: not exposed (reference group), exposed 6 months, exposed] 6-12 months] and exposed >12 months.

Ethical considerations
The IeDEA West Africa Collaboration obtained authorization from the Ethics committee "Comité de Protection des Personnes Sud-Ouest et Outre-mer III" (N˚06-09) in Bordeaux, France, and from the national ethics committees of each participating countries: Benin (IRB 00006860), Burkina Faso (N˚2011-8-46), Cote d'Ivoire (IRB 00009111), Senegal (IRB 00002659) and Mali (IRB 00001465). A waiver for individual informed consent was given by the Institutional Review Boards as the data was collected from routine patients' charts. The study procedure did not involve any personal contact with the patients.

Statistical analysis
The baseline characteristics were summarized using descriptive statistics. Missing values at baseline were imputed by the most updated data within a range of three months and data still missing were treated as a specific category. Longitudinal data from patients with a baseline ANC measure and at least one ANC measure after six month were computed to estimate the incidence rate of severe neutropenia and evaluate the factors associated with the occurrence of neutropenia. The incidence rate was defined as the number of patients with a first episode of severe neutropenia per 100 person-years (PY) of follow-up. Poisson regression method was used to estimate the incidence rate of neutropenia with its 95% confidence interval (CI). Incidence rates were adjusted for HIV clinic, gender, age, calendar year for ART initiation, baseline neutrophil count, baseline hemoglobin level, baseline CD4 count, first-line ART regimen and baseline clinical stage.
The role of ART and other characteristics in the development of first-time neutropenia was analyzed by univariate and multivariate Cox proportional hazards regression models. All the explanatory variables were considered at baseline except ZDV exposure and CD4 count that were computed as time-updated variables. In order to take into account time-updated variables we did a counting process format style of input which consists of a sequence of time intervals over which the values of any time-varying explanatory variable are constant for each participating individual. The final multivariate Cox regression model was stratified on HIV clinic and only significant variables (p <0.05) and possible confounding factors were kept in the model. Each patient contributed to the time at risk from the time they entered into the study (i.e. the date of ART initiation) either until the closing date (December 31 st , 2014), date of death or loss-to-follow-up, date of transfer out, date of last visit with ANC available in an interval of six months or date of first severe neutropenia. We restricted the analysis for a twoyear period after ART initiation. In addition, we described the ART modifications that occurred in the month following the development of severe neutropenia.
All statistical analysis was performed with the SAS 9.4 software (SAS Institute Inc., Cary, NC, USA).

Results
Of the 25,833 patients with at least one ANC measure available enrolled in the participating IeDEA West Africa cohorts between 2002 and 2014, 15,456 (60%) had an ANC measure available at baseline (ART initiation). Of these, 9,880 (65%) had at least another ANC during the first six months but 454 of them were excluded because of severe neutropenia at ART initiation. Overall, 9,426 patients were included in this analysis (Fig 1).
Patients with at least one ANC measure available but who were excluded (n = 16,407) initiated ART in earlier calendar year than the 9,426 patients included in the analysis (median year: 2008 versus 2009; p <10 −4 ). Conversely, there were no significant differences concerning age (p = 0.11), gender (p = 0.20), baseline CD4 count (p = 0.18) and ART regimen distribution (p = 0.49) (S1 Table).
As depicted in Fig 2, the proportion of patients initiating a D4T-containing regimen in our study population fell from 50% between 2002 and 2008 to 12% during the period 2009-2011 and to only 1% for the period 2012-2014. This decreased was compensated by an increased prescription of ZDV and tenofovir (TDF)-containing regimens. Indeed, before 2009, ZDV was prescribed just as much as D4T and its proportion increased to >60% for the period 2012-2014 and remained the most frequently prescribed combination since then. The proportion of patients initiating a TDF-containing regimen increased from <8% before 2009 to 27% for the period 2009-2011 and 32% for the period 2012-2014.

Factors associated with severe neutropenia
The results of Cox regression models are shown in Table 3. The risk to develop a first episode of severe neutropenia was significantly higher in patient exposed to ZDV compared to those initiating another first-line ART regimen, irrespective of their time spent Thereafter, we looked at ART modification after a first episode of severe neutropenia (Table 4). Of the 888 patients with severe neutropenia, 128 (14.4%) had a change of ART regimen in the following month corresponding to a total of 134 drug modifications. The first drug  stopped was ZDV (49%) followed by Nevirapine (24%). Among the 65 patients who stopped ZDV after severe neutropenia, 43% switched to D4T and 46% to TDF.

Discussion
We found that the risk of developing severe neutropenia was associated with ZDV-containing ART regimen, initiating ART before 2009, having a low baseline neutrophil count, an advanced clinical stage and a more profound immune deficiency. In addition we found that the risk of severe neutropenia persisted over six months of treatment with ZDV-containing ART regimens in a context where the majority of ART initiation were based on ZDV including in the recent years, despite the change in WHO recommendations to replace it by TDF [13].
Our estimated incidence of severe neutropenia was higher than among patients living with HIV before access to ART in comparable settings. Indeed, the incidence rate of neutropenia reported by Toure et al. in a cohort study in Cote d'Ivoire was 4.3 per 100 PY after initiation of CTX prophylaxis [18]. Similarly, Anglaret et al. found an overall incidence rate of 1.9 per 100 PY during a CTX versus placebo clinical trial in Cote d'Ivoire and the incidence in the CTX arm was 2.8 per 100 PY [19]. Although theses variations might be related to methodological differences, they may also suggest that the incidence of severe neutropenia was lower in sub-Saharan Africa before access to ART. Conversely, our estimate is much lower than the rate of 56.3 per 100 PY found by Moh et al. in their cohort study in Cote d'Ivoire during the six months after ZDV-based ART initiation [22]. This variation can be explained by the fact that the majority of their study population (80%) was exposed to both ZDV and CTX prophylaxis at ART initiation, a combination known to increase the risk of severe neutropenia. As the impact of CTX on ANC is supposed to be acute, information on CTX use over time was needed to capture the impact of this molecule on neutropenia. Information on CTX exposure was incomplete in our study and we could not incorporate this variable in our analysis. Thus we could not fully compare our findings with the previous reports and their wide range of estimates of severe neutropenia.
We found an increased risk of severe neutropenia in patients exposed to ZDV and this risk was persistent after one year of treatment. These results are consistent with the known myelosuppressive effect of ZDV [14][15][16][17]. Hoffman et al. in South Africa reported an increased incidence of severe neutropenia after ZDV-based ART initiation in their cohort [17]. However, to the best of our knowledge, it is the first time that a study considers ZDV-exposure as a timeupdated variable and quantifies the duration of exposure.
We also found that advanced immunodeficiency and advanced clinical stage were independent risk factors for severe neutropenia. These results are consistent with previous findings and emphasize the need of an early ART initiation as recently recommended by the WHO guidelines [3,14,18].
Finally, we found significant differences in the occurrence of severe neutropenia according to clinics. Several explanations may be formulated. The clinics participating to the IeDEA West Africa collaboration have their own organization and patient management procedures. Thus, the frequency of follow-up visits and biological measurements may vary substantially from one clinic to another, inducing a potential information bias. Second, unmeasured Despite the WHO recommendations suggesting to adapt the ART regimen when adverse effects of grade 3 or 4 occur, the rate of patients with severe neutropenia who changed ART regimen following neutropenia was low, leading to a few hypotheses. First, we were not able to document exposure to other drugs co-administered with ART such as CTX prophylaxis. CTX has a known acute hematologic toxicity and might be discontinued prior to any other drugs. If neutropenia resolves, CTX needs to be permanently evicted from patient's prescriptions and no additional ART modifications are needed. Secondly, the limited availability of alternative ART regimens might have prevented any ART modification in the absence of clinical consequences associated with severe neutropenia. Thirdly, the reported ANC might have not been properly graded, thus limiting the applications of guidelines recommendations. Indeed, a previous report from our collaboration in West Africa showed that only a limited number of ART prescribers declared grading adverse drug reactions [4].
We acknowledge several limitations to our study. First, we selected patients with at least one ANC at baseline and one ANC during the first six months of follow-up. It might have introduced a bias of selection if patients included and excluded were different. However, we found no significant differences between these two groups concerning the CD4 count and exposure to ZDV at ART initiation. Second, the occurrence of neutropenia is expected in patients initiating a ZDV-based ART regimen leading to a potential bias of information if these patients were better monitored. However, we did not find any significant differences concerning the number of follow-up visits and the duration of follow-up between patients exposed to ZDV and patients not exposed. Another important limitation was the lack of information concerning the exposure to CTX. Without this information we were not able to describe the combined impact of exposure to ZDV and CTX. In addition, we could not take into account the CTX prophylaxis interruptions in our description of treatment modifications after severe neutropenia.

Conclusion
The incidence of severe neutropenia was high among people living with HIV on ART in West Africa. Initiating ART with a ZDV-based regimen was an important risk factor associated with the occurrence of a severe neutropenia and this risk persisted over time. In addition, an advanced HIV disease and a low CD4 count at ART initiation were also associated with a subsequent risk to develop a severe neutropenia.
The recent WHO guidelines recommend initiating a TDF-based ART regimen in HIVinfected persons regardless of their CD4 count. In a context where ZDV is still massively prescribed and most patients initiate ART at an advanced stage of the disease with low CD4 count, the implementation of the WHO recommendation will be crucial in order to reduce the risk of severe neutropenia.
Severe neutropenia is a known risk factor for bacterial infections although we were not able to document this clinical outcome in our population followed for a relatively short period of time. Thereafter, a better documentation of morbidity is needed in order to assess the potential impact of drug therapies on clinical outcomes among HIV-infected persons in sub-Saharan Africa.