Determining Bacterial Communities in the Lungs of HIV-infected Individuals With COPD in Uganda.

Investigating the Association Between Altered Lung MicroBiome and HIV-associated Chronic Obstructive Pulmonary Disease in a Ugandan Cohort

Research question

Is there any association between altered lung bacterial communities and HIV-associated Chronic Obstructive Pulmonary Disease (COPD)?

Rationale

Sub-Saharan Africa has experienced dramatic increases in COPD related-morbidity and mortality. Longitudinal studies have shown that people living with HIV develop worsening airflow obstruction with a prevalence higher than that of the general population (i.e 3.4 to 21% compared to 0.4 to 12.2%). It is still unknown why HIV-infected individuals develop COPD at a prevalence higher than their HIV-negative counterparts. It's been hypothesized that a change in the lung bacterial communities in the setting of HIV drives inflammation leading to lung damage. There is a need to explore the dynamics of lung bacterial communities and elucidate mechanisms responsible for irreversible lung damage that may follow lung disturbances in bacterial richness and diversity. In addition, understanding the bacterial communities of the lung in normal subjects is an essential step in providing negative controls to interpret lung microbe in disease states for-example COPD. Insights from this research will inform efforts to design optimal screening and treatment strategies for COPD in the HIV-infected population in sub Saharan Africa.

Methods

A cross sectional study will be conducted in which lung bacterial communities in 63 HIV infected participants ≥ 35 years with and without COPD will be compared with 63 HIV negative participants with and without COPD. Participants will be recruited from COPD/HIV and LINK Nakaseke cohorts, which were population based studies conducted in the same study setting. Sputum samples will be collected using sputum DNA collection, preservation and isolation Kits. Extracted bacterial DNA will be sequenced and used to determine all bacterial species in the processed samples using available online metagenomics databases.

Analysis plan

A histogram will be used to display the frequencies of the identified bacterial species in the processed samples. Bacterial richness and diversity of samples in the 4 groups will be compared to determine any differences.

Background

The improvements in access to antiretroviral therapy (ART) among people living with HIV/AIDs (PLWHA) has resulted in a decrease in HIV-associated morbidity and mortality. This is particularly true in low- and middle-income countries (LMICs), which bear the largest burden of HIV. The reduction in mortality has substantially increased life expectancy, with estimates among PLWHA now approaching that of the general population (Asiki, Reniers et al. 2016). Consequently, there has been increased attention among survivors to the emerging burden of non-communicable diseases (NCD), such as chronic obstructive pulmonary disease (COPD) (Geneau, Stuckler et al. 2010). Sub-Saharan Africa, which has the highest density of PLWHA, has experienced dramatic increases in COPD related-morbidity and mortality (van Zyl Smit, Pai et al. 2010, Asiki, Reniers et al. 2016). Studies are urgently needed to further elucidate the pathogenesis of COPD, and to determine optimal screening and treatment strategies (Asiki, Reniers et al. 2016, Drummond, Kunisaki et al. 2016). Associations between lung dysbiosis and COPD exacerbation phenotypes have been demonstrated in the general population(Wang, Bafadhel et al. 2016). However, it still remains unknown why PLWHA have higher prevalence of COPD compared with the general population (Drummond, Kunisaki et al. 2016). No data currently exists on lung microbiome in the general Sub Saharan African population including Uganda. There is a need to explore the dynamics of the lung microbiome (Cui, Morris et al. 2014, Wang, Bafadhel et al. 2016) and elucidate immune-mediated responses responsible for irreversible lung damage that may follow lung dysbiosis in the setting of HIV infection (Hutchinson, Vlahos et al. 2014, Wang, Bafadhel et al. 2016). Understanding the role of lung dysbiosis in the pathogenesis of HIV-associated COPD is of utmost significance in the African setting with the highest HIV/AIDs burden(Cassol, Cassetta et al. 2010, Morris, George et al. 2011).

Study hypothesis

Altered lung microbiome in HIV infected individuals is associated with chronic obstructive pulmonary disease (COPD).

Specific aims

1. To determine the lung microbiome among HIV-infected and uninfected individuals without COPD (healthy controls).
2. To determine the lung microbiome among HIV-infected individuals with COPD.
3. To compare lung microbiome among HIV-infected individuals with COPD and HIV negative individuals with COPD.
4. To determine the association between lung microbiome and COPD in HIV-infected individuals.

Problem Statement

Whereas multiple studies on lung microbiome and its role in COPD exacerbations are currently being carried out in the western world (Cui, Morris et al. 2014, Sze, Hogg et al. 2014, Wang, Bafadhel et al. 2016) , there is limited literature on the role of HIV in COPD pathogenesis (Morris, George et al. 2011, Drummond, Kunisaki et al. 2016). It is still unknown why HIV-infected individuals develop COPD with a prevalence higher than their HIV-negative counterparts (Morris, George et al. 2011). No data currently exists on lung microbiome in the general Sub Saharan African population including Uganda. Research is urgently needed to describe the lung microbiome in individuals without COPD and to elucidate the pathogenesis of COPD in HIV (Morris, George et al. 2011, Drummond, Kunisaki et al. 2016).

Justification

Establishing an association between lung microbiome and HIV-associated COPD is of utmost significance in the African setting with the highest HIV/AIDs burden.Knowledge from such a study will to guide the development of optimal screening and treatment strategies of COPD in HIV population.

Methods

A cross sectional study will be conducted among HIV-infected individuals attending ART clinics in Nakaseke and HIV negative individuals from Lung function in Nakaseke (LiNK) study.

Estimated sample size

Applying the hypothesis testing and power calculations for taxonomic-based human microbiome data using the Human Microbiome project-R (HMP-R) statistical package (version1.4.3) which operates on the Dirichlet-multinomial model[33], For significance level (alpha set at = 5%), number of participants N = 50; number of sequence reads ≥20,000 (considered as cut off for quality control), power ≥99.99% (this power is sufficient to detect the effect size that is anticipated to be observed in the sequence data). Considering 20% non-participation rate, the sample size for each group will be 63 participants.

Study site selection

Nakaseke district ART clinics have been chosen as the study sites because of the ongoing COPD/HIV study whose objectives are to determine the prevalence and factors associated with COPD in HIV. Over 752 HIV-infected individuals have been screened for COPD following standard guidelines. In addition, the LiNK study (Lung Function in Nakaseke and Kampala (LiNK), PI: Kirenga, Checkley) was also conducted in Nakaseke. It was a population based observational study assessing COPD prevalence, risk factors and symptomatology in rural communities of Nakaseke using a stratified random sample of 1000 individuals above 35 years of age and full time residents of Nakaseke.

Sampling procedure

Using Epi tool random number generator (La Rosa, Brooks et al. 2012), the study staff will select 63 participants in the target groups from our respective cohorts. The research assistant will document their age, sex and smoking history. The COPD+HIV+ group will be used to identify participants from other groups.

Study plan

Trained research assistants and a medical officer will be based at Nakaseke ART clinics. Randomly selected participants will be contacted by phone. The study team will explain the protocol to the participants and if interested, a study visit will be scheduled by the research assistant. They will obtain informed consent from participants and a baseline spirometry will be done. The study staff will carry out sputum induction procedure following standard operating procedures. Induced sputum specimens will be collected using sputum DNA collection, preservation and isolation Kits following manufacturer's instructions. The components of the preservative will allow the collected samples to be stored for more than 2 years without any detectable DNA degradation.

Specimen handling

Specimen will be handled as per the standard operating procedures. Briefly, sputum specimens will placed in a leak proof biohazard bag with sealed lids and absorbent material. All specimens will be transported in compliance with local and national regulations governing the transport of potentially infectious materials.

Core laboratory for specimen processing

Molecular diagnostics laboratory and Integrated Biorepository laboratory located at Makerere University College of Health Sciences on the 3rd floor of the Medical Microbiology building will be used for sample processing.

Genomic DNA extraction

Bacterial genomic DNA will be extracted from 200 microlitres of sputum samples using commercially available kits.

Genomic DNA extraction controls

ZymoBIOMICS Microbial Community Standard will be used as a positive sample control when preparing DNA samples. Storage Buffer from DNA Collection Kit with no sample added will be used as negative control.

V3 and V4 hypervariable region polymerase chain reaction (PCR) amplification

The V3 and V4 hypervariable region of the 16S rRNA gene will be PCR amplified utilizing commercially available primers.

16S rRNA sequencing

DNA sequencing will be done in batches utilizing MiSeq sequencing platform following manufacturer's protocol. Each batch will consist of randomized samples from the two groups. A 1x26 MiSeq run will be performed to check cluster density and normalization of samples. Illumina metagenomics workflow by MiSeq Reporter version 2.3 will be used for demultiplexing indexed reads, generating sequence files, and classifying reads. Stringent criteria will be used to remove low quality and chimeric reads.

Statistical analysis

For specific aim 1, 2 and 3

1. Clustering 16SrRNA sequence reads into OTUs:

   After quality control and data cleaning, the remaining reads will be subjected to an open reference operational taxonomical unit (OTU) picking (97% identity cut-off) in which reads will be firstly clustered against the Greengenes reference sequences. OTUs will be rarefied to the lowest number of reads among all samples, and the rarefied OTU table will be used for assessing alpha and beta diversity . A jack-knifing Principal Coordinate Analysis (PCoA) will be performed to assess the robustness of the results.

2. Composition of the lung microbiome in the target compared with control groups:

The investigators will compute the mean percentage abundance of all the identified taxonomic clusters OTUs) or bacterial species in the target population vs. the control group. This data will be summarized and presented in histogram.

For specific aim 3

1. A multivariate model between OTUs, clinical status (COPD/HIV) and/or clinical data will be developed. This will be achieved by performing a general linear regression for each continuous variable and the binary outcome. To establish the relationship between OTUs and a group of clinical data, a canonical correspondence analysis will be performed.

   For the microbiome dataset, OTUs that will be present in at least 10% of all samples will be included in the analysis. For the clinical dataset, variables that will be missing in more than 50% of all samples will be excluded to minimize the effects of missing values. Collinearity will be addressed using pairwise Pearson's correlation test on microbiome and clinical variables. Significant model components will be selected by cross validation using the Auto-fit option with default criteria in the SIMCA-P (Wang, Bafadhel et al. 2016).

2. Measuring association between OTUs, HIV, COPD status and clinical data A co-occurrence network will be established by correlation of individual OTUs. Pairwise Pearson's correlation test will be used to assess the significance of co-occurrence relationships, and only significant correlations will be retained (adjusted P-value < 0.05). The False Discovery Rate (FDR) method will be used throughout to adjust P-values (adj. P) for multiple tests (Wang, Bafadhel et al. 2016). Unstable edges whose scores will not be within 95% confidence interval of bootstrap distribution will be removed from the network. Missing values will be omitted from correlation calculations. For the OTU network, the 100 top-ranking and 100 bottom-ranking edges will be displayed in the final network (Wang, Bafadhel et al. 2016).

Study limitations

In this study, the scope of our investigations will involve bacterial communities in the lungs which the investigators will identify by 16SrRNA sequencing. The investigators acknowledge that other microbe communities (viral, fungal and parasites) may play a role in COPD pathogenesis and exacerbation.

Ethical consideration

The study was approved by the Uganda National Council of Science and Technology (UNCST) and Mulago Hospital Research Ethics Committee (MHREC). No invasive procedures are being done to the participants and all screening will include standard point of care approaches.

Informed consent process:

Confidentiality

Participants will be given unique identification numbers to replace their identifiable data. No participant identifiers will be attached to participant data. Study staff will have access to raw data. All data will be stored in a password protected, fully encrypted database, accessible only to the study staff and investigators responsible for analysis.

Inclusion Criteria:

• Male and female individuals atleast 35 years of age
• Both HIV seropositive and seronegative.
• Spirometry confirmed COPD and no COPD

Exclusion Criteria:

• Participants with asthma
• Participants with significant respiratory disease other than COPD
• Failure to perform spirometry
• Pulse rate greater than 120 beats per minute
• Blood pressure greater than 140(systolic)/90( diastolic)
• History of headaches in the past 6 months
• History of eye, chest or abdominal surgery
• History of hernia or chest trauma
• Pregnant women
• Bed ridden patients
• Mentally incapacitated patients