Improving identification of symptomatic cancer at primary care clinics: A predictive modeling analysis in Botswana.

In resource-limited settings, augmenting primary care provider (PCP)-based referrals with data-derived algorithms could direct scarce resources towards those patients most likely to have a cancer diagnosis and benefit from early treatment. Using data from Botswana, we compared accuracy of predictions of probable cancer using different approaches for identifying symptomatic cancer at primary clinics. We followed cancer suspects until they entered specialized care for cancer treatment (following pathologically confirmed diagnosis), exited from the study following noncancer diagnosis, or died. Routine symptom and demographic data included baseline cancer probability assessed by the primary care provider (low, intermediate, high), age, sex, performance status, baseline cancer probability by study physician, predominant symptom (lump, bleeding, pain or other) and HIV status. Logistic regression with 10-fold cross-validation was used to evaluate classification by different sets of predictors: (1) PCPs, (2) Algorithm-only, (3) External specialist physician review and (4) Primary clinician augmented by algorithm. Classification accuracy was assessed using c-statistics, sensitivity and specificity. Six hundred and twenty-three adult cancer suspects with complete data were retained, of whom 166 (27%) were diagnosed with cancer. Models using PCP augmented by algorithm (c-statistic: 77.2%, 95% CI: 73.4%, 81.0%) and external study physician assessment (77.6%, 95% CI: 73.6%, 81.7%) performed better than algorithm-only (74.9%, 95% CI: 71.0%, 78.9%) and PCP initial assessment (62.8%, 95% CI: 57.9%, 67.7%) in correctly classifying suspected cancer patients. Sensitivity and specificity statistics from models combining PCP classifications and routine data were comparable to physicians, suggesting that incorporating data-driven algorithms into referral systems could improve efficiency.

HIV-1 drug resistance mutations among individuals with low-level viraemia while taking combination ART in Botswana.

OBJECTIVES: To assess whether a single instance of low-level viraemia (LLV) is associated with the presence of drug resistance mutations (DRMs) and predicts subsequent virological failure (VF) in adults receiving ART in 30 communities participating in the Botswana Combination Prevention Project. METHODS: A total of 6078 HIV-1 C pol sequences were generated and analysed using the Stanford HIV drug resistance database. LLV was defined as plasma VL = 51-999 copies/mL and VF was defined as plasma VL ≥ 1000 copies/mL. RESULTS: Among 6078 people with HIV (PWH), 4443 (73%) were on ART for at least 6 months. Of the 332 persons on ART with VL > 50 copies/mL, 175 (4%) had VL ≥ 1000 copies/mL and 157 (4%) had LLV at baseline. The prevalence of any DRM was 57 (36%) and 78 (45%) in persons with LLV and VL ≥ 1000 copies/mL, respectively. Major DRMs were found in 31 (20%) with LLV and 53 (30%) with VL ≥ 1000 copies/mL (P = 0.04). Among the 135 PWH with at least one DRM, 17% had NRTI-, 35% NNRTI-, 6% PI- and 3% INSTI-associated mutations. Among the 3596 participants who were followed up, 1709 (48%) were on ART for ≥6 months at entry and had at least one subsequent VL measurement (median 29 months), 43 (3%) of whom had LLV. The OR of experiencing VF in persons with LLV at entry was 36-fold higher than in the virally suppressed group. CONCLUSIONS: A single LLV measurement while on ART strongly predicted the risk of future VF, suggesting the use of VL > 50 copies/mL as an indication for more intensive adherence support with more frequent VL monitoring.

Mediation Analysis of a Theory-Based Culture and Age-Appropriate HIV/STI Prevention.

Few studies have investigated mediator effects of HIV prevention interventions on adolescents in sub-Saharan Africa. Herein, we report on a secondary analysis of an intervention that increased intentions to use condoms, abstain from sexual intercourse, and seek safe male circumcision among adolescents in Botswana. In a study conducted in Botswana, 806 grade 9 students from 21 public Junior Secondary Schools were randomly assigned to either the OWN THE FUTURE: Pulling Together We Will" ( PTWW) intervention group or a health promotion control group. Both conditions consisted of 12 1-h modules, with two modules delivered during each of the six sessions on six consecutive school days. The students in both groups completed confidential computer-based surveys at several time points: pre-, immediately post-, 3, 6, and 12 months post-intervention. Mediation was assessed using the product-of-coefficients approach in a generalized estimating equations (GEE) framework. The analyses showed that condom use beliefs were significant mediators of the intervention effect on the intention to consistently use condoms over time. Also, negative socio-cultural beliefs, prevention beliefs, and HIV/STI knowledge were significant mediators of the intervention's effects on the intention to abstain from sex. Additionally, normative beliefs, prevention beliefs, parental negotiation, and circumcision knowledge were significant mediators of the intervention's effect on intention to seek safe male circumcision. The mediation analysis delineated a theoretical model and isolated activities that positively impact condom use, abstinence from sex, and circumcision intentions of Batswana middle school adolescents.

Efficient estimation of human immunodeficiency virus incidence rate using a pooled cross-sectional cohort study design.

Development of methods to accurately estimate human immunodeficiency virus (HIV) incidence rate remains a challenge. Ideally, one would follow a random sample of HIV-negative individuals under a longitudinal study design and identify incident cases as they arise. Such designs can be prohibitively resource intensive and therefore alternative designs may be preferable. We propose such a simple, less resource-intensive study design and develop a weighted log likelihood approach which simultaneously accounts for selection bias and outcome misclassification error. The design is based on a cross-sectional survey which queries individuals' time since last HIV-negative test, validates their test results with formal documentation whenever possible, and tests all persons who do not have documentation of being HIV-positive. To gain efficiency, we update the weighted log likelihood function with potentially misclassified self-reports from individuals who could not produce documentation of a prior HIV-negative test and investigate large sample properties of validated sub-sample only versus pooled sample estimators through extensive Monte Carlo simulations. We illustrate our method by estimating incidence rate for individuals who tested HIV-negative within 1.5 and 5 years prior to Botswana Combination Prevention Project enrolment. This article establishes that accurate estimates of HIV incidence rate can be obtained from individuals' history of testing in a cross-sectional cohort study design by appropriately accounting for selection bias and misclassification error. Moreover, this approach is notably less resource-intensive compared to longitudinal and laboratory-based methods.

High prevalence of albuminuria among adult males living with HIV in Botswana.

Chronic HIV disease is associated with a fivefold increase in albuminuria outside of sub-Saharan Africa. However, very little is known about albuminuria risk among people living with HIV (PLWH) in sub-Saharan Africa. Therefore, we conducted a cross-sectional observational HIV clinic-based study of albuminuria among 1533 adults aged 21 years or older between January 2020 and January 2021 in Gaborone, Botswana. Clinical albuminuria was defined using a sex-based albumin‒creatinine ratio (ACR) of 25-355 mg/g for females and 17-250 mg/g for males. The study population mean age was 48.5 (SD 10.3) years, and 764/1533 (49.7%) were female. The overall prevalence of albuminuria was 20.7% (95% CI 18.7%, 22.8%). A higher proportion of males were more likely to be categorized as having albuminuria than females, 25% (95% CI 22.0, 28.2) versus 16.4% (95% CI 13.8,19.2), P value < 0.001. In the final multivariate models, predictors of albuminuria differed by sex group. Larger longitudinal studies are required to evaluate the impact of albuminuria among PLWH with particular emphasis on the effect of sex on the risk of albuminuria.

Reversal of CSF HIV-1 Escape during Treatment of HIV-Associated Cryptococcal Meningitis in Botswana.

Cerebrospinal fluid (CSF) viral escape has been poorly described among people with HIV-associated cryptococcal meningitis. We determined the prevalence of CSF viral escape and HIV-1 viral load (VL) trajectories in individuals treated for HIV-associated cryptococcal meningitis. A retrospective longitudinal study was performed using paired CSF and plasma collected prior to and during the antifungal treatment of 83 participants recruited at the Botswana site of the phase-3 AMBITION-cm trial (2018−2021). HIV-1 RNA levels were quantified then CSF viral escape (CSF HIV-1 RNA ≥ 0.5 log10 higher than plasma) and HIV-1 VL trajectories were assessed. CSF viral escape occurred in 20/62 (32.3%; 95% confidence interval [CI]: 21.9−44.6%), 13/52 (25.0%; 95% CI: 15.2−38.2%) and 1/33 (3.0%; 95% CI: 0.16−15.3%) participants at days 1, 7 and 14 respectively. CSF viral escape was significantly lower on day 14 compared to days 1 and 7, p = 0.003 and p = 0.02, respectively. HIV-1 VL decreased significantly from day 1 to day 14 post antifungal therapy in the CSF but not in the plasma (ß = −0.47; 95% CI: −0.69 to −0.25; p < 0.001). CSF viral escape is high among individuals presenting with HIV-associated cryptococcal meningitis; however, antifungal therapy may reverse this, highlighting the importance of rapid initiation of antifungal therapy in these patients.

Low-frequency HIV-1 drug resistance mutations in antiretroviral naïve individuals in Botswana.

BACKGROUND: Individuals living with human immunodeficiency virus (HIV) who experience virological failure (VF) after combination antiretroviral therapy (cART) initiation may have had low-frequency drug resistance mutations (DRMs) at cART initiation. There are no data on low-frequency DRMs among cART-naïve HIV-positive individuals in Botswana. METHODS: We evaluated the prevalence of low-frequency DRMs among cART-naïve individuals previously sequenced using Sanger sequencing. The generated pol amplicons were sequenced by next-generation sequencing. RESULTS: We observed low-frequency DRMs (detected at <20% in 33/103 (32%) of the successfully sequenced individuals, of whom four also had mutations detected at >20%. K65R was the most common low-frequency DRM detected in 8 individuals. Eighty-two of the 103 individuals had follow-up viral load data while on cART. Twenty-seven of the 82 individuals harbored low-frequency DRMs. Only 12 of 82 individuals experienced VF. The following low-frequency DRMs were observed in four individuals experiencing VF: K65R, K103N, V108I, and Y188C. No statistically significant difference was observed in the prevalence of low-frequency DRMs between individuals experiencing VF (4/12) and those not experiencing VF (23/70) (P = .97). However, individuals with non-nucleoside reverse transcriptase inhibitors-associated low-frequency DRMs were 2.68 times more likely to experience VF (odds ratio, 2.68; 95% confidential interval, 0.4-13.9) compared with those without (P = .22). CONCLUSION: Next-generation sequencing was able to detect low-frequency DRMs in this cohort in Botswana, but these DRMs did not contribute significantly to VF.

The Global Cohort of Doctoral Students: Building Shared Global Health Research Capacity in High-Income and Low- and Middle-Income Countries.

Doctoral students in high- and low-income countries pursuing careers in global health face gaps in their training that could be readily filled through structured peer-learning activities with students based at partnering institutions in complimentary settings. We share lessons learned from the Global Cohort of Doctoral Students, a community of doctoral students based at the Harvard T. H. Chan School of Public Health, Haramaya University, University of Gondar, University of Botswana, and University of Rwanda College of Medicine and Health Sciences. Students in the Global Cohort program engage in collaborative research, forums for constructive feedback, and professional development activities. We describe the motivation for the program, core activities, and early successes.

Use of a mutation-specific genotyping method to assess for HIV-1 drug resistance in antiretroviral-naïve HIV-1 Subtype C-infected patients in Botswana.

Background: HIV-1 drug resistance poses a major threat to the success of antiretroviral therapy. The high costs of available HIV drug resistance assays prohibit their routine usage in resource-limited settings. Pan-degenerate amplification and adaptation (PANDAA), a focused genotyping approach based on quantitative PCR (qPCR), promises a fast and cost-effective way to detect HIV drug resistance mutations (HIVDRMs).  Given the high cost of current genotyping methods, we sought to use PANDAA for screening key HIVDRMs in antiretroviral-naïve individuals at codons 103, 106 and 184 of the HIV-1 reverse transcriptase gene. Mutations selected at these positions have been shown to be the most common driver mutations in treatment failure.  Methods: A total of 103 samples from antiretroviral-naïve individuals previously genotyped by Sanger population sequencing were used to assess and verify the performance of PANDAA. PANDAA samples were run on the ABI 7500 Sequence Detection System to genotype the K103N, V106M and M184V HIVDRMs. In addition, the cost per sample and reaction times were compared. Results: Sanger population sequencing and PANDAA detected K103N mutation in three (2.9%) out of 103 participants.  There was no evidence of baseline V106M and M184V mutations observed in our study. To genotype the six HIVDRMs it costs approximately 40 USD using PANDAA, while the reagents cost per test for Sanger population sequencing is approximately 100 USD per sample. PANDAA was performed quicker compared to Sanger sequencing, 2 hours for PANDAA versus 15 hours for Sanger sequencing. Conclusion: The performance of PANDAA and Sanger population sequencing demonstrated complete concordance. PANDAA could improve patient management by providing quick and relatively cheap access to drug-resistance information.

Differences in human immunodeficiency virus-1C viral load and drug resistance mutation between plasma and cerebrospinal fluid in patients with human immunodeficiency virus-associated cryptococcal meningitis in Botswana.

To determine effects of cryptococcal meningitis (CM) on human immunodeficiency virus (HIV)-1C cerebrospinal fluid (CSF) viral escape, CSF/plasma viral discordance, and drug resistance mutation (DRM) discordance between CSF and plasma compartments, we compared CSF and plasma viral load (VL) and DRMs in individuals with HIV-associated CM in Botswana.This cross-sectional study utilized 45 paired CSF/plasma samples from participants in a CM treatment trial (2014-2016). HIV-1 VL was determined and HIV-1 protease and reverse transcriptase genotyping performed. DRMs were determined using the Stanford HIV database. CSF viral escape was defined as HIV-1 ribonucleic acid ≥0.5 log10 higher in CSF than plasma and VL discordance as CSF VL > plasma VL.HIV-1 VL was successfully measured in 39/45 pairs, with insufficient sample volume in 6; 34/39 (87.2%) participants had detectable HIV-1 in plasma and CSF, median 5.1 (interquartile range: 4.7-5.7) and 4.6 (interquartile range:3.7-4.9) log10 copies/mL, respectively (P≤.001). CSF viral escape was present in 1/34 (2.9%) and VL discordance in 6/34 (17.6%). Discordance was not associated with CD4 count, antiretroviral status, fungal burden, CSF lymphocyte percentage nor mental status. Twenty-six of 45 (57.8%) CSF/plasma pairs were successfully sequenced. HIV-1 DRM discordance was found in 3/26 (11.5%); 1 had I84IT and another had M46MI in CSF only. The third had K101E in plasma and V106 M in CSF.Our findings suggest that HIV-1 escape and DRM discordance may occur at lower rates in participants with advanced HIV-disease and CM compared to those with HIV associated neurocognitive impairment.

Explaining disparities in oncology health systems delays and stage at diagnosis between men and women in Botswana: A cohort study.

PURPOSE: Men in Botswana present with more advanced cancer than women, leading to poorer outcomes. We sought to explain sex-specific differences in time to and stage at treatment initiation. METHODS: Cancer patients who initiated oncology treatment between October 2010 and June 2017 were recruited at four oncology centers in Botswana. Primary outcomes were time from first visit with cancer symptom to treatment initiation, and advanced cancer (stage III/IV). Sociodemographic and clinical covariates were obtained retrospectively through interviews and medical record review. We used accelerated failure time and logistic models to estimate standardized sex differences in treatment initiation time and risk differences for presentation with advanced stage. Results were stratified by cancer type (breast, cervix, non-Hodgkin's lymphoma, anogenital, head and neck, esophageal, other). RESULTS: 1886 participants (70% female) were included. After covariate adjustment, men experienced longer excess time from first presentation to treatment initiation (8.4 months) than women (7.0 months) for all cancers combined (1.4 months, 95% CI: 0.30, 2.5). In analysis stratified by cancer type, we only found evidence of a sex disparity (Men: 8.2; Women: 6.8 months) among patients with other, non-common cancers (1.4 months, 95% CI: 0.01, 2.8). Men experienced an increased risk of advanced stage (Men: 67%; Women: 60%; aRD: 6.7%, 95% CI: -1.7%, 15.1%) for all cancers combined, but this disparity was only statistically significant among patients with anogenital cancers (Men: 72%; Women: 50%; aRD: 22.0%, 95% CI: 0.5%, 43.5%). CONCLUSIONS: Accounting for the types of cancers experienced by men and women strongly attenuated disparities in time to treatment initiation and stage. Higher incidence of rarer cancers among men could explain these disparities.

High incidence of tuberculosis in the first year of antiretroviral therapy in the Botswana National antiretroviral therapy programme between 2011 and 2015.

OBJECTIVE: Tuberculosis (TB) remains one of the leading causes of mortality and morbidity among people living with HIV. We sought to estimate the incidence of TB in a national database of HIV-infected patients receiving antiretroviral therapy (ART) in Botswana. DESIGN: A retrospective analysis of HIV-infected adult patients (≥18years) who initiated ART between 2011 and 2015 in the Botswana ART program. METHODS: Multivariable analysis using Cox regression included sex, age, viral load and CD4 T-cell counts. RESULTS: Of 45 729 patients, with a median follow-up of 1.7 years Q1 : Q3, 0.5, 3.1), 1791 patients developed TB over a median of 1.5 years (Q1 : Q3, 0.3, 3.1) of follow-up (incidence rate 1.9 per 100 person-years; 95% CI 1.8-2.0). At baseline, the median CD4 T-cell count was 272 cells/µl (Q1, Q3 146, 403). The risk of TB was greatest within the first year of ART (incidence rate 2.9 per 100 person-years; 95% CI 2.7-3.1) and in patients with CD4 T-cell counts below 50 cells/µl (incidence rate 8.3/100 person-years; 95% CI 7.1-9.7). Patients with viral loads above 10 000 copies/ml at 3 months post-ART initiation had two times higher risk of TB, hazard ratio 2.5 (95% CI 1.8-2.3). CONCLUSION: We report a high incidence of TB within the first year of ART and in patients with advanced immunodeficiency. Improved screening strategies and virologic monitoring during this early period on ART, coupled with TB preventive treatment, will reduce the burden of TB.

Immunological non-response and low hemoglobin levels are predictors of incident tuberculosis among HIV-infected individuals on Truvada-based therapy in Botswana.

BACKGROUND: There is a high burden of tuberculosis (TB) in HIV antiretroviral programmes in Africa. However, few studies have looked at predictors of incident TB while on Truvada-based combination antiretroviral therapy (cART) regimens. METHODS: We estimated TB incidence among individuals enrolled into an observational cohort evaluating the efficacy and tolerability of Truvada-based cART in Gaborone, Botswana between 2008 and 2011. We used Cox proportional hazards regressions to determine predictors of incident TB. RESULTS: Of 300 participants enrolled, 45 (15%) had a diagnosis of TB at baseline. During 428 person-years (py) of follow-up, the incidence rate of TB was 3.04/100py (95% CI, 1.69-5.06), with 60% of the cases occurring within 3 months of ART initiation. Incident cases had low baseline CD4+ T cell counts (153cells/mm3 [Q1, Q3: 82, 242]; p = 0.69) and hemoglobin levels (9.2g/dl [Q1, Q3: 8.5,10.1]; p<0.01). In univariate analysis, low BMI (HR = 0.73; 95% CI 0.58-0.91; p = 0.01) and hemoglobin levels <8 g/dl (HR = 10.84; 95%CI: 2.99-40.06; p<0.01) were risk factors for TB. Time to incident TB diagnosis was significantly reduced in patients with poor immunological recovery (p = 0.04). There was no association between baseline viral load and risk of TB (HR = 1.75; 95%CI: 0.70-4.37). CONCLUSION: Low hemoglobin levels prior to initiation of ART are significant predictors of incident tuberculosis. Therefore, there is potential utility of iron biomarkers to identify patients at risk of TB prior to initiation on ART. Furthermore, additional strategies are required for patients with poor immunological recovery to reduce excess risk of TB while on ART.

Correction: Immunological non-response and low hemoglobin levels are predictors of incident tuberculosis among HIV-infected individuals on Truvada-based therapy in Botswana.

[This corrects the article DOI: 10.1371/journal.pone.0192030.].