Enrichment of HIV-1 Subtype AD Recombinants in a Ugandan Cohort of Severely Septic Patients

Background Several population-wide HIV-1 subtype distribution studies in Uganda have evaluated relatively healthy clinic patients. Given the differences in HIV-1 disease progression based on subtype, we examined HIV-1 subtype distribution and disease outcomes among hospitalized patients with severe sepsis. Methods Patients with severe sepsis were enrolled at two hospitals in Uganda. Data collected included demographics, Karnofsky scores, highly active antiretroviral therapy (HAART) use, HIV-1 serostatus, CD4+ T cell concentration, whole blood lactate concentration, and blood cultures. HIV-1 subtypes were determined by sequencing parts of the gag and env genes, followed by phylogenetic analysis. Results Of the 267 patients evaluated, 228 (85.4%) were HIV infected. The predominant HIV-1 subtypes were A (46%), D (17%), and AD recombinants (30%). HIV-1 subtypes B, C, and other recombinants were uncommon. Patients infected with HIV-1 subtypes A, D and AD viruses were similar in demographics, CD4+ T cell concentration, HAART use, Karnofsky scores, whole blood lactate concentration, and positive blood cultures. There was no difference in 30-day mortality from severe sepsis between the 3 groups (p = 0.99). Conclusion A high proportion of HIV-1 subtypes A and AD recombinants was observed in this cohort of severely septic patients. The proportion of AD recombinants was higher in this cohort than in previous cohorts of Ugandan HIV-1 patients. No difference in baseline demographics, clinical factors or 30-day mortality was seen across HIV-subtypes.

The rate of disease progression among HIV-1 infected patients varies depending on subtype. Subtype D is associated with a faster rate of CD4 + T-cell decline and more rapid disease progression to AIDS and death compared to subtype A in both adult and pediatric HIV-infected populations in Uganda [13][14][15], Kenya [16] and Tanzania [17]. However, HIV-1 subtype does not always correlate with severity of HIV-associated illness. For instance, HIV-1 subtype association with neurocognitive performance has yielded conflicting results. In a Ugandan adult population, subtype D infection was associated with a higher prevalence of dementia compared to subtype A infection [18], while subtype A-infected Ugandan children demonstrated poorer neurocognitive performance compared to subtype D-infected children [19].
Sepsis from severe blood stream infections has been recognized as a significant cause of morbidity and mortality among patients infected with HIV-1. There is a 76-84% prevalence of HIV among patients who present with febrile illness or sepsis in sub-Saharan Africa [20][21][22][23][24][25], where the HIV pandemic has claimed most of its victims [26]. Given the association between certain HIV-1 subtypes and rate of CD4 decline and mortality, we sought to determine HIV-1 subtype distribution and mortality in a cohort of predominantly HIV-infected patients with severe sepsis in Uganda.

Study Design and PATIENT Recruitment
This was a prospective cohort study done in Uganda at Mulago Hospital, the national referral hospital in Kampala, and Masaka Regional Referral Hospital, 80 miles southwest of Kampala, Uganda. Between May 2008 and May 2009, we enrolled 426 severely septic patients admitted to the medical wards of these 2 hospitals in an intervention study of fluid resuscitation [27]. Adult patients (age $18 years) triaged to the hospitals' Medical Casualty Units were enrolled based on the following inclusion criteria: (1) suspected infection as determined by the admitting medical officer; (2) two of the following: a) axillary temperature .37.5uC or ,35.5uC, b) heart rate .90 beats/min, c) respiratory rate .20 breaths/min; (3) systolic blood pressure (SBP) #100 mmHg; and (4) either whole blood lactate concentration .2.5 mmol/L or Karnofsky Performance Scale (KPS) score #40. Patients with acute cerebrovascular events, gastrointestinal hemorrhage, suspected surgical or obstetric emergencies or shock without suspected infection were excluded.

Ethics Statement
Ethical approval was obtained from the research and/or ethics committees of the University of Virginia, Makerere University, Mulago Hospital, Infectious Diseases Institute, and Uganda National Council of Science and Technology. Written informed consent was obtained from each patient or a surrogate if the patient was too obtunded to provide consent.

Data Collection and Clinical Assessments
At the time of enrollment, baseline information including demographic data (age, gender), clinical history (highly active antiretroviral therapy [HAART] use) and clinical data (body temperature, respiratory rate, blood pressure and KPS scores) were collected. Laboratory evaluation included: HIV serostatus, CD4+ T cell concentration (if HIV-infected), whole blood lactate concentration, and blood cultures. Thirty day mortality and length of hospital stay were also recorded.

Phylogenetic Analysis and Subtype Determination
Using ClustalW, the sequences were aligned with those from the HIV sequence database (Los Alamos National Laboratory). The phylogenetic relationship was then analyzed using the neighbor-joining method in the PHYLIP software package. The references. If both gag and env sequences of a sample were the same subtype, that sample was assigned a single subtype. If, however, the gag and env sequences of the same sample resulted in different subtypes, this was considered a recombinant. In addition to the above phylogenic analysis, we also utilized the Recombinant Identification Program (RIP) on the HIV sequence database webpage (www.hiv.lanl.gov/content/ sequence/HIV/HIVTools.html/) to confirm the subtypes.

Statistical Analysis
Analysis was performed on patients infected with the 3 most common HIV subtypes -subtype A, D and AD. Differences in baseline demographics between the subtypes were compared using a t-test for continuous variables and Chi-square test for categorical variables. Kaplan-Meier estimates of mortality were compared with a log-rank test and length of hospital stay among patients who survived to discharge was compared using a Mann-Whitney test. Cox proportional hazards analysis was used to examine the crude and adjusted relationship between HIV subtype and 30-day mortality. Multivariable analysis identified whether an independent association existed between 30-day mortality and the following independent variables: gender, age, CD4 cell count, HAART use, lactate concentration (, or $4 mmol/L), and aerobic and mycobacterial positive blood cultures.

HIV-1 Prevalence and Subtype Distribution
Of the 426 patients enrolled in the fluid resuscitation cohort, blood samples were available from 267 patients that were included in this study. Two hundred and twenty eight (85.4%) of the 267 individuals in our study were HIV-1 infected (Fig. 1). HIV-1 subtype determination was successful in 209 (91.7%) of the 228 HIV-1 infected patients. Nineteen samples failed both env and gag amplification while 34 samples had either the gag or the env successfully amplified but not both.

Clinical Features of HIV-1 Infected Patients with Subtypes A, D and AD Recombinants
Since the majority (.90%) of the HIV positive patients were infected with subtypes A, D or AD recombinant viruses, analyses were focused on these three subytpes. Baseline characteristics including age, gender, median CD4 + T cell count, HAART use,

Disease Outcome
Disease outcomes recorded in this study included 30-day mortality and length of hospital stay. There was no statistically significant difference in 30-day mortality due to sepsis between patients infected with HIV-1 subtypes A, D and AD recombinants (p [log rank] = 0.99) (Fig. 3). At day 5, it appeared that survival in AD recombinant-infected patients was lower (77%) compared to subtype A-and D-infected patients (85.2% and 86.1%, re-  Results of the Cox proportional analysis to examine the relationship between HIV-1 subtypes, other covariates and 30day mortality are represented in table 2. Covariates included in the multivariate analysis included age, gender, median CD4 count, receipt of HAART, lactate concentration, and positive blood cultures for aerobic and Mycobacterium tuberculosis. None of these factors changed the relationship between HIV-1 subtypes and 30day mortality (Table 2). However, the multivariate analysis revealed two independent predictors of 30-day mortalityelevated lactate concentration (HR, 2.02; 95% CI, 1.15-3.52; p = 0.014) and blood cultures positive for Mycobacterium tuberculosis (HR, 2.19; 95% CI, 1.2-4.02; p = 0.011).

Discussion
Although the HIV-1 subtype distribution in sub-Saharan Africa has been widely studied, to our knowledge, this is the first study characterizing HIV-1 subtype distribution and its effects in a severely septic cohort. Our findings show a higher proportion of HIV-1 AD recombinant viruses in our cohort of severe sepsis than previously reported non-critically ill cohorts. HIV prevalence in patients with severe sepsis in this cohort was 85% which is similar to earlier studies performed by our group on patients with sepsis in Uganda [21], but much higher than the national HIV prevalence in Uganda (6.5%) [26], and the HIV prevalence among patients presenting to emergency departments in Kampala hospitals (50%) [28]. These data may suggest that infection with HIV-1 increases the possibility for patients to develop severe sepsis.
Data on HIV-1 subtype distribution in clinic patients in the Kampala region of Uganda around the time of our study showed a slight predominance of subtype A over subtype D [12]. Among treatment naïve patients, 49% were subtype A, 28.7% subtype D and 8.1% were AD recombinants. Among the treatment experienced patients, the subtypes A, D and AD recombinant distribution was 41%, 37.9% and 7.4% respectively. While results from our cohort of severely ill patients showed a comparable proportion of patients infected with subtype A virus (46%), a lower proportion of subtype D viruses (17%) and a higher proportion of AD recombinant viruses (30%) existed in our cohort. Without subtype data from a comparison control group of patients without sepsis, our study is limited in being able to definitely conclude whether the enrichment of subtype AD recombinant viruses in our sepsis cohort is significantly greater than the general population of HIV-1 infected patients in Uganda. Notably, the high proportion of subtype AD recombinant viruses among critically ill HIV-1infected patients has also been reported in a Kenyan cohort in which 6 out of 14 (42.8%) patients with suspected immune reconstitution inflammatory syndrome were infected with HIV-1 subtype AD recombinant viruses [29]. Ultimately, more studies are required to determine if the enrichment of recombinant viruses in critically ill patients is a common phenomenon.
The AD recombinant viruses in these patients could have risen from an initial infection with an AD recombinant virus or from recombination of 2 different viruses that infected the same patient. If the latter is true, infection with multiple HIV-1 viruses might impose a heavier burden on a patient's immune system, making it more difficult to control other infections leading to sepsis. Recent data from Uganda supporting this theory show that the rate of late stage WHO events was 3 times greater among patients with multiple HIV-1 infections compared to those infected with a single HIV-1 subtype [30].
Previous studies showed that subtype D progressed to AIDS and death faster than subtype A [14,16,17]. In addition, patients infected with HIV-1 AD recombinant viruses were noted to progress even faster than subtype D infected patients [14]. Despite the enrichment of AD recombinants in this severe sepsis cohort, we found no significant difference in 30-day mortality between HIV-1 subtypes. In the previous studies, patients were followed prospectively from the time of HIV-1 seroconversion. Our study is limited because patients admitted with severe sepsis only represent a cross-section of the patient's HIV-1 infection course; thus, the time of HIV seroconversion is unknown. Time to first development of sepsis or frequency of septic episodes may be related to length of HIV seropositivity, rate of CD4 decline, HIV viral load rise and HAART use. Since HIV-1 subtype affects the rate of CD4-cell decline and HIV viral load changes [15], it would therefore affect how early and how often an individual may develop sepsis. Once an individual develops sepsis however, the factors that are most likely to influence mortality due to sepsis are level of immunosuppression and goal directed therapy for sepsis. In this cohort, all the subjects received the same treatment for sepsis [27]. Survival from an acute illness like sepsis seems unlikely to be influenced by HIV-1 subtype. In addition, level of immune suppression represented by CD4 cell count did not significantly affect the relationship between HIV-1 subtype and mortality in this cohort likely due to the overall low level of immune suppression among all sepsis cohort patients (median CD4 count ,100 lymphocytes/mm 3 ). More studies are required to determine the relationship between the HIV subtype and acute illness.

Conclusions
This is the first report on HIV-1 subtype distribution in severely septic patients in sub Saharan Africa. We found a high proportion of subtype A and AD recombinant viruses in a cohort of severely septic HIV-1 infected patients from Uganda. The proportion of AD recombinant viruses was higher than previously reported in non-critically ill cohorts. Although it has been shown to affect long-term disease progression, HIV-1 subtype did not affect the outcomes of patients with severe sepsis in our study.