• TB is responsible for an enormous burden of human disease yet is still understood poorly at a fundamental level.
• High-quality clinical research in developing countries is possible; field-based clinical and basic research can impact diagnosis, care, management, and vaccine development.
• Sequencing of TB genomes linked to patient histories can reveal patterns of drug resistance, identify molecular markers and pathways for new and improved drugs and diagnostics, and provide key datasets for molecular epidemiology and surveillance efforts.
• The integration of epidemiology with systems biology is creating a “systems epidemiology” that can provide a comprehensive view of infection, disease progression, and transmission.
• Meta-analyses of data gathered from many sites can answer questions that cannot be answered with data from a single site.
a Identified by individual speakers.
Five speakers at the workshop discussed new forms of TB research that could have a dramatic effect on TB prevention and treatment. New technologies and innovative analytic methods offer opportunities not available with traditional approaches.
Around the year 2000, Anne E. Goldfeld, Harvard Medical School and GHC, started to see wasting and dying caused by AIDS in Cambodia, which previously had been largely spared from that epidemic. Many children were taking care of their dying parents, and others were themselves infected. “We were overwhelmed about how to handle this,” said Goldfeld. From private sources, she and her colleagues were able to obtain the funds needed to keep 100 people alive for 1 year. They also began conducting research through the Comprehensive International Program for Research on AIDS (CIPRA), partly because they knew this work would help get AIDS drugs into the country. And they convinced the photographer James Nachtwey to come to Cambodia to document the country’s joint epidemic of AIDS and TB, helping to shed light internationally on this problem, which was under-appreciated at the time. Goldfeld noted that between a quarter and a half of the 30 million people with HIV infection who have died have actually died of TB, which is curable even in a setting with a high prevalence of AIDS.
Goldfeld and her colleagues have been performing hypothesis-driven research in Cambodia to understand the molecular basis of TB’s natural history and to develop operational models for treating the disease, particularly in patients coinfected with HIV. TB and HIV infection each result in a host immune response, which in turn boosts the natural progression of the other pathogen. TB infection usually occurs first, followed by HIV infection, progression to AIDS, and extraordinarily high mortality, particularly when the CD4 level dips below 200. (Goldfeld pointed out that reinfection with TB also is very common in immunosuppressed HIV positive individuals.) When Goldfeld’s team began their research, the international recommendation from WHO was to start antiretroviral therapy for HIV at 2 months following diagnosis. Some data from retrospective studies indicated that an earlier start to therapy produced better outcomes, but drug–drug interactions were a concern for coinfected patients—for example, rifampicin speeds up the cytochrome P450 system of the liver, which increases side effects and the burden of taking medicines.
Goldfeld and her colleagues proceeded to explore the hypothesis that early initiation of antiretroviral therapy would increase survival in coinfected patients despite a more complex initial clinical management. With the support of the Agence Nationale de Recherche sur le SIDA, the NIH Division of AIDS, and the Institut Pasteur in Cambodia, the research team, together with the Cambodian Health Committee, developed five clinical sites in
1 This section is based on the presentation by Anne E. Goldfeld, Professor of Medicine, Harvard Medical School; and Co-Founder, GHC.
Cambodia, which at that point had no real research infrastructure. The first patient was recruited to the trial in 2006, and recruitment ended in 2009, by which time AIDS drugs were available in Cambodia through the Global Fund. A total of 661 patients were randomized, about half to the early and half to the late arm. Most were culture positive. Their CD4 count averaged 25—“unbelievably low,” according to Goldfeld—and their BMIs were very low as well. Fewer than 2 percent of patients were lost to follow-up, and fewer than 1 percent of almost 9,000 protocol visits were missed. The project was called CAMELIA, for Cambodian Early versus Late Introduction of Antiretrovirals, and the results were published in 2011 in the New England Journal of Medicine (Blanc et al., 2011).
The findings from this research were “stunning,” said Goldfeld. The early arm resulted in a 34 percent reduction in mortality. By using drugs that were already available, death rates were markedly reduced. “Obviously we need many new drugs, but one can do better with what one has in hand,” Goldfeld asserted. She added that having MDR TB resulted in an eightfold increased risk of death, partly because of delays in diagnosis.
The research also gave the team an opportunity to learn about the impact of TB on the reconstitution of the immune system during antiretroviral therapy. (Goldfeld noted that securing funding to pursue this basic scientific question proved to be extremely difficult, and the research was possible in part because of a private donation from the Annenberg Foundation.) Goldfeld described some of the initial results of this research, which are elucidating some basic principles involved in the immune response to HIV and TB and which were still being analyzed at the time of her presentation.
The research conducted by Goldfeld’s team revealed that, on a worldwide basis, early antiretroviral therapy could save as many as 150,000 of the 450,000 lives lost to combined HIV and TB infection. The results of the research have already affected national policy in Cambodia, and a WHO follow-up recommendation is expected.
Nobody believed that a high-quality international clinical trial was possible in a setting such as Cambodia, said Goldfeld, but this assumption was proved wrong. She emphasized that the combination of basic science and a clinical network is a powerful way to learn more about TB and HIV infection. It also had numerous other benefits, including vastly improved research infrastructure and capacity. Goldfeld concluded by noting that field-driven clinical and basic research can impact diagnosis, care, management, and vaccine development.
Maria Y. Giovanni, Director, Office of Genomics and Advanced Technologies; Assistant Director, Microbial Genomics and Advanced Technology; Division of Microbiology and Infectious Diseases, National Institute of Allergy and Infectious Diseases (NIAID), NIH, explained that NIAID funds basic research along with the development of new drugs, diagnostics, and vaccines. In the area of TB, it funds basic research and natural history studies in addition to drugs, diagnostics, and vaccines. NIAID has been taking advantage of new sequencing technologies by investing in genomics to provide resources and tools, including drug discovery and drug development tools, for researchers. For example, three Genomic Sequencing Centers for Infectious Diseases create databases, tools, 3-D structures, protein clones, and predictive models that are freely distributed to the scientific community. NIAID also has developed a research agenda for MDR and XDR TB (NIAID Working Group, 2007).
In 2012, NIAID and international collaborators launched the Large-Scale TB Genome Sequencing Project. Using high-throughput, next-generation sequencing technology along with bioinformatics analysis and tools, the project has the goal of sequencing more than 1,000 TB strains from Korea, Russia, South Africa, Uganda, and other countries. These TB genomes will be linked to patient histories to advance understanding of genetic patterns of drug resistance, identify molecular markers and pathways for new and improved drugs and diagnostics, and provide key datasets for molecular epidemiology and surveillance efforts. NIAID also is a founding member of the TB International Genome Consortium, which is investigating drug resistance in different TB strains.
NIAID has developed a database called the Pathosystems Resource Integration Center (PATRIC) that has a strong focus on TB. To build synergy, PATRIC has deidentified genomic as well as many other kinds of data, including data contributed by members of the TB community. NIAID has been emphasizing the collection of clinical data and metadata to enhance the richness and utility of the data it provides. Almost every string sequenced in PATRIC has associated clinical data and metadata that are available for use.
The Institute has made a major commitment to systems biology, which makes it possible to combine different kinds of datasets and construct predictive models. Through this work, many gaps in understanding of the regulatory networks for TB have been filled in the past few years, with
2 This section is based on the presentation by Maria Y. Giovanni, Director, Office of Genomics and Advanced Technologies; Assistant Director, Microbial Genomics and Advanced Technology; Division of Microbiology and Infectious Diseases, National Institute of Allergy and Infectious Diseases (NIAID), NIH.
the number of demonstrated protein-gene interactions increasing almost 17-fold since 2008.
To reduce the risk and expense of drug development, NIAID offers services, provided by contractors, to academic and industrial investigators worldwide. The objective is to facilitate research at all stages—from basic research through clinical applications—by developing critical information needed to move a product along the product development pathway. In addition, the Institute provides organisms, reagents, assays, animal models, in vitro and in vivo screening for antimicrobial activity, and candidate therapeutics.
Finally, Giovanni briefly mentioned NIAID’s support for diagnostics development. The goal is to develop integrated diagnostic platforms that are rapid, inexpensive, easy to use, sensitive, and specific for detecting multiple targets to support diagnosis of infectious diseases, determination of antimicrobial resistance, and monitoring of treatment and prevention.
Comas and Gagneux (2009) have suggested that control of DR TB will remain a formidable challenge for many years to come unless two new technological developments are rapidly applied, noted Gail Cassell, Harvard Medical School and Infectious Disease Research Institute. The first is high-throughput genomic technologies, which have already contributed greatly to the understanding of TB epidemiology, comparative genomics, evolution, and host-pathogen interactions. Cassell listed several examples of advances these technologies have produced. They have
• demonstrated the importance of ongoing transmission and the existence of drug resistance even in places where little TB treatment has taken place;
• demonstrated that the mycobacterial genome has more genomic plasticity than expected, as well as shown the evolution and geographic distribution of different strains;
• provided the first rigorous evidence that previous exposure to TB does not protect against reinfection, a finding that has major implications for vaccine development;
• allowed differentiation of patients who relapsed because of treatment failure from those reinfected with a different strain;
• enabled pragmatic public health efforts, such as detection of outbreaks and ongoing TB transmission; and
3 This section is based on the presentation by Gail Cassell, Visiting Professor, Harvard Medical School; and Vice President of TB Drug Discovery, Infectious Disease Research Institute.
• pointed to possible associations between genomic content and disease severity.
The second critical technological development cited by Comas and Gagneux entails the integration of epidemiology with systems biology to create a new “systems epidemiology.” This integrated field would systematically link genomic data with other forms of “omics” data and with clinical data to provide a systematic view of infection, disease progression, and transmission.
TBResist is a global consortium for whole-genome sequencing of DR TB strains that seeks to both take advantage of and advance these two technological developments. It has several major goals:
• to better address DR TB and such comorbidities as HIV infection and diabetes;
• to better understand latency, including the incidence and effects of mixed infection and the characteristics of replicating versus non-replicating organisms;
• to better predict the future disease trajectory of disease; and
• to develop new tools, including drugs, diagnostics, and vaccines.
Achieving these goals will be possible only by aggregating data from thousands of patients and thousands of strains, as is being done in other disease areas, said Cassell. The consortium will bring together many diverse investigators who are willing to share their data and build large databases of genomic data from well-characterized patient populations.
The founding members of the consortium are the Laboratory Information for Public Health Excellence, NIAID, the Broad Institute, and IBM, which is supplying its powerful HIV data-mining tool EuResist, along with bioinformaticists who specialize in mining data from medical databases. Many other public and private groups also have joined the consortium, including several from BRICS countries. A strong interdisciplinary team is building a mathematically and statistically robust analytical framework to make sense of the data deluge. TBResist sequencing centers are located at the Broad Institute and in China, Russia, and Taiwan.
The consortium plans to publicly release all of the sequence data, including RNA data, as rapidly as possible. For the clinical data and metadata, each participant must have a data sharing and release plan, although the release can be delayed up to 9 months or until publication. Patrick Tao Li, Scientific Representative, BGI (formerly Beijing Genomics Institute), described his organization’s collaborative efforts to develop methods for TB diagnostics and DST (Box 11-1).
BGI (formerly Beijing Genomics Institute), which has been extensively involved in DNA sequencing for plants, animals, and human diseases, is making a major investment in TB, said BGI’s Scientific Representative, Patrick Tao Li. BGI uses multiple generations of sequencing to ensure validated sequences, and then uses these data to find single nucleotide polymorphisms relevant to resistance. BGI also is analyzing the transcriptome from TB samples and DNA methylation, which affects gene expression.
In collaboration with other research groups from around the world, BGI is using these data to develop methods for diagnosing TB and testing for drug resistance. A large portion of these data is stored in the BGI cloud so researchers with permission can access BGI’s supercomputer through their laptops to analyze data and make contributions to the project.
a This box is based on the presentation by Patrick Tao Li, Scientific Representative, BGI (formerly Beijing Genomics Institute).
Meta-analyses provide a way to pool information gathered from many sites to answer questions that cannot be answered with data from a single site, said Kathryn DeRiemer, Associate Professor, Department of Public Health Sciences, School of Medicine, University of California, Davis. She described a meta-analysis with which she was involved that looked at treatment regimens and patient outcomes for more than 9,000 patients (Ahuja et al., 2012). The study pooled data from 71 co-authors and 32 countries. Earlier reviews had identified 93 studies conducted prior to 2009 that included at least one MDR TB case. Of these studies, 26 were excluded because they represented the same or overlapping cohorts. Other studies were excluded because the original investigators no longer had access to the data or did not have data on DST, or for other reasons. Patients were excluded who had XDR TB or exhibited only extrapulmonary TB, or for whom information on treatment outcomes was lacking.
4 This section is based on the presentation by Kathryn DeRiemer, Associate Professor, Department of Public Health Sciences, School of Medicine, University of California, Davis.
The objectives of the meta-analysis were to assess the impacts of specific drugs, the number of drugs used, and the duration of treatment on the clinical outcomes of patients with pulmonary MDR TB. Once approvals had been obtained from each of the local ethics boards at the participating institutions, deidentified data were sent to a central data center at the Montreal Chest Institute of McGill University. The meta-analysis investigators consulted extensively with the authors of the original studies to ensure that the data were accurate and reflected the defined outcomes.
The meta-analysis estimated the odds of treatment success, defined as cure or treatment completion versus treatment failure or relapse; combinations of treatment failure, relapse, and default; and combinations of treatment failure, relapse, death, and default. Random-effects multivariate logistic regression models were used to estimate odds ratios and confidence intervals.
Of the 9,153 individuals with MDR TB in the study, 66 percent were sputum smear positive, and 52 percent had cavitary lesions. A number of drugs were used to treat these MDR TB patients. Some of these drugs were those that were available at the time the patient required treatment. No set regimen was followed in each of the sites, and limited information is available on the specific regimen followed by each patient over long periods of time.
The overall treatment success rate was 54 percent. Fifteen percent of patients in the study died, which translates to more than 1,300 deaths. A substantial number—23 percent—defaulted, transferred out, or had unknown outcomes, a concern with respect to the transmission of MDR TB.
Use of later-generation ethionamide, ofloxacin, and quinolones or a related drug was associated with treatment success compared with treatment failure, relapse, or death. In the intensive phase of treatment, MDR TB patients who were treated with four or more effective drugs (based on DST) had better treatment outcomes. In the continuation phase, use of at least three effective drugs was a predictor of treatment success.
For the initial treatment phase, most patients underwent 7 to almost 9 months of therapy. The total duration of treatment was about 18 to 22 months for these MDR TB patients.
This study represents the first attempt to pool data on MDR TB patients from many different countries and different sites and studies, which created some limitations:
• Different field studies used different guidelines and case management strategies, depending on the country and clinical practice.
• The availability of SLDs varied among studies.
• Specimens were not available for further analysis, such as additional DST and genome sequencing.
Nevertheless, this research points to the potential of a future prospective study looking across different sites, with standardized data collection and multiple specimens per patient at different points during therapy.
From a clinical and program perspective, said DeRiemer, treatment consists of at least four steps: (1) screening/testing, (2) diagnosis, (3) treatment, and (4) outcome. In the past, the goal has been to develop guidelines for good clinical practice that provide guidance on these steps, with the assumption that these guidelines would be appropriate for all cases everywhere. But MDR TB treatment is now taking much longer than it has in the past, which DeRiemer ascribed to the continuous nature of M.tb. infection. The goal of the pathogen is to multiply and transmit itself to other people. As a result, the bacterial burden in an individual may be increasing over days, weeks, months, years, or a lifetime. Therapy produces a selective pressure, which leads to mycobacterial diversification through mutations and genomic deletions. The goal of treatment therefore must be to reduce or eliminate the bacterium by the time treatment is completed. Accomplishing this goal requires asking questions about the pathogen burden, heterogeneity, and diversification.
DeRiemer concluded by urging collaborative prospective human studies with multiple time points for specimen and data collection. She also recommended incorporating new tools in these studies that can illuminate how mycobacterial changes such as mutations affect treatment outcomes.
This page intentionally left blank.