Defining the research agenda to reduce the joint burden of disease from Diabetes Mellitus and Tuberculosis

Abstract

The steadily growing epidemic of diabetes mellitus (DM) poses a threat for global tuberculosis (TB) control. Previous studies have identified an important association between DM and TB. However, these studies have limitations: very few were carried out in low-income countries, with none in Africa, raising uncertainty about the strength of the DM-TB association in these settings, and many critical questions remain unanswered. An expert meeting was held in November 2009 to discuss where there was sufficient evidence to make firm recommendations about joint management of both diseases, to address research gaps and to develop a research agenda. Ten key research questions were identified, of which 4 were selected as high priority: i) whether, when and how to screen for TB in patients with DM and vice versa; ii) the impact of DM and non-DM hyperglycaemia on TB treatment outcomes and deaths, and the development of strategies to improve outcomes; iii) implementation and evaluation of the tuberculosis “DOTS” model for DM management; and iv) the development and evaluation of better point-of-care diagnostic and monitoring tests, including measurements of blood glucose and glycated haemoglobin A_1c (HbA_1c) for patients with DM. Implementation of this research agenda will benefit the control of both diseases.

Background

The global burden of disease from diabetes mellitus (DM) and tuberculosis (TB) is immense. In 2010, there will be an estimated 285 million people living with DM with approximately 4 million deaths (International Diabetes Federation 2009). In 2007, there were an estimated 14.4 million people living with TB, 9.2 million new cases and 1.7 million deaths (WHO 2009). While it is widely known that 95% of TB patients live in the low- and middle-income countries, 70% of patients with DM patients also live in these same countries, especially in South-east Asia and the Western Pacific.

Among several risk factors for TB, which include HIV/AIDS, silicosis, malnutrition, alcoholism and smoking, DM has received recent recognition. A systematic review of the literature in 2008 identified 13 age-adjusted, quantitative, observational studies in North America, UK, Russia, Mexico, Korea, Taiwan and India, finding a relative risk of TB in DM patients of 3.1 in cohort studies and odds ratios that ranged from 1.16–7.83 in case control studies (Jeon & Murray 2008). These findings were similar to those reported in a previous systematic review (Stevenson et al. 2007a), were supported by data included in an epidemiological model indicating that, in India, DM might account for nearly 15% of pulmonary tuberculosis (PTB) cases (Stevenson et al. 2007b), and have been further endorsed by a review in 2009 that provided a synopsis of the evidence of the role of DM in influencing the clinical presentation and response to treatment for TB (Dooley & Chaisson 2009).

The important association between DM and TB is not in doubt. However, the previous studies published between 1965 and 2009 have limitations. Many are health facility-based case control studies, using medical chart diagnoses of DM, and may be subject to confounding. Most studies are from industrialised countries. The evidence-base from low-income countries is weak, with none at all from Africa, raising uncertainty about the strength of the DM-TB association in these settings. There are also many critical unanswered questions: should patients with DM be screened for TB in DM clinics and vice versa, and how and when?; what is the effect of DM on the presentation of TB and subsequent TB treatment outcomes?; what is the natural history of multidrug-resistant TB (MDR-TB) in DM patients, and what steps are needed for prevention and management?; should TB preventive therapy be considered in patients with DM?; would life-style changes be helpful in preventing both diseases?

Systematic review and expert meeting

As a result of these uncertainties and questions, a systematic review of the literature was commissioned from the Harvard School of Public Health, USA, in May 2009. The review addressed key questions such as screening for TB and DM in routine clinics, TB chemoprophylaxis and the impact of DM on clinical and programmatic management of TB. The systematic review was completed by the end of August, and the findings presented and discussed at an expert meeting in November 2009 at the International Union Against Tuberculosis and Lung Disease, Paris, France. The main objectives of the meeting were to determine whether there was enough evidence to make policy recommendations about joint diagnosis and management of both diseases, address research gaps and develop a research agenda around these gaps. This paper summarises the research agenda that emerged at the meeting, which we believe if carried out will assist in reducing the joint burden of disease from DM and TB.

Research agenda for Diabetes and Tuberculosis

The key research questions are shown in the Box, divided into high, medium and low priority. We identified 4 high priority research questions: whether and how to screen for TB in patients with DM and vice versa; the impact of DM or non-DM hyperglycaemia on TB treatment outcomes and deaths, and the development and testing of strategies to improve outcomes; implementation and evaluation of the tuberculosis “DOTS” model for DM management; and development and evaluation of better point-of-care diagnostic tests for patients with DM.

Box. Key research questions and methodology for improving the prevention, management and care of diabetes mellitus and tuberculosis.

Key research questions	Priority	Study design and methodology
Screening for Disease: Screening patients with DM for active TB Screening patients with TB for DM	High	Prospective observational cohort studies of DM patients routinely attending diabetes clinics and screened for TB, and TB patients starting anti-TB treatment and screened for DM
TB treatment outcomes in patients with DM and with non-diabetes hyperglycaemia, including a more detailed assessment of death during anti-TB treatment, and the development and testing of strategies to improve outcomes for both diseases	High	Prospective observational cohort studies using standardised TB regimens and standardised treatment outcomes and focusing on defined primary outcomes Prospective observational cohort studies to determine when death occurs in relation to start of TB treatment, the aetiology and whether case fatality is reduced by better control of DM or hyperglycaemia or modification to TB drug regimens, duration of therapy and TB drug doses
Implementing and evaluating the “DOTS” model for standardised case management of DM	High	Operational research that includes quarterly cohort reporting of new cases, treatment outcomes of cumulative cases including frequency of co-morbidities such as TB, and survival analysis
Development and evaluation of better point-of-care diagnostic and monitoring tests for DM	High	Developmental work to produce a reliable low cost finger stick test for measuring blood glucose and glycated haemoglobin A1c (HbA1c) in rural areas, which then needs to be tested for efficacy and feasibility in the field
Rates of hospitalisation and additional medical costs associated with diagnosis and management of dual disease	Medium	Cross-sectional and case-control studies
Use of the community to improve diagnosis, management and care of patients with diabetes and TB	Medium	Operational research
Household contact tracing of adult patients with smear- positive Pulmonary TB	Medium	Prospective observational studies to determine the yield of screening household contacts of index Pulmonary TB patients for TB infection, active TB, HIV and DM, and assess whether DM influences the risk of TB infection
Radiographic findings in DM patients with tuberculosis	Medium	Systematic review of the literature, and prospective cross-sectional studies if further evidence is required, to determine the common radiographic patterns that are associated with DM
Modelling the effect of the DM epidemic on the TB epidemic	Medium	Mathematical modelling studies, ideally informed by higher quality studies of the association between DM and TB, particularly from low-income settings
TB preventive therapy in patients with DM	Low	Randomised controlled trial assessing efficacy and safety of isoniazid preventive therapy in reducing risk of active TB in patients with DM

DM = diabetes mellitus; TB = tuberculosis; HIV = human immunodeficiency virus

Screening

Patients with DM, particularly those with sub-optimal control, should be screened for active TB in areas of high TB prevalence. Prospective observational cohort studies should be conducted in DM clinics with a focus on adults and stratified by quality of DM control. Key questions, similar to those being asked about intensified TB case finding in people living with HIV (Havlir et al. 2008; Kranzer et al. 2010), include:- a) what type of screening algorithm will be most effective, i.e., should all DM patients be investigated by standard TB laboratory investigations or should these investigations be targeted to those with symptoms suggesting active TB or with poor control of DM as defined by symptoms or measurements of blood glucose or glycated haemoglobin A_1c (HbA_1c); b) how often should screening be conducted; and c) what are the most appropriate screening tools (sputum smear examination and chest radiography for pulmonary disease, ultrasound for extra-pulmonary abdominal lymphadenopathy) and how do these compare to the gold standard of sputum culture for Mycobacterium tuberculosis. In determining the most appropriate strategy, the impact of DM on clinical manifestations of TB should be considered. The value of formal clinic education in diabetes clinics with TB posters, leaflets or group talks about the links between DM and TB needs to be explored in terms of improving health care staff and patient awareness and for future guidelines and training materials. Urban and other settings known to have high TB and DM incidence would be the most appropriate places to conduct such studies, for example Dar es Salaam in Tanzania, Blantyre in Malawi, Chennai in India and Beijing in China all being possibilities.

For TB patients, previous studies suggest that it is more reliable to screen for DM later in the course of anti-TB treatment rather than at the start (Dooley & Chaisson 2009), because TB as a chronic infectious disease may elevate blood glucose levels due to cytokine stimulation resulting in false positive DM diagnoses if investigations are performed too early. However, delayed screening may be a missed opportunity for subsequently modifying treatment, and many TB programmes, especially in Africa, have decentralised services to peripheral facilities where it is difficult to get laboratory investigations performed (Edginton 1990; Drabo et al. 2006). Research is required to determine the optimal time and best methods for diagnosing DM in TB patients, focusing on adults stratified by type of disease (smear-positive PTB, smear-negative PTB and extra-pulmonary TB). The most appropriate ways of screening should be explored (urine, random or fasting blood glucose, and/or glycated haemoglobin A_1c (HbA_1c)).

TB treatment outcomes

Patients identified by the two screening strategies discussed earlier need to be entered into prospective studies assessing the impact of DM or non-DM hyperglycaemia on TB treatment outcomes, using standardised TB regimens and outcomes, in which other confounding factors (age, smoking status, alcohol, body mass index and HIV) are taken into account. The level of hyperglycaemia and quality of diabetes control, measured for example by HbA_1c, are important factors to consider. Primary treatment outcomes should include:- a) liver function tests; b) pharmacokinetic levels of rifampicin and oral diabetes medications; c) TB treatment outcomes; d) recurrence of TB one year after completion of TB treatment as determined by sputum culture; and e) culture and drug-sensitivity testing, at the start of treatment and at the time of failure or TB recurrence in order to assess linkages and associations with drug-resistant TB.

Death is reported to be more frequent in DM patients on anti-TB treatment (Dooley & Chaisson 2009), but research is required to address unanswered questions such as when death occurs in relation to start of anti-TB treatment, the aetiology, and whether better DM control or modified TB drug regimens, duration of anti-TB therapy and TB drug doses reduce case fatality.

The TB DOTS model for managing DM

The concept of using components of the Tuberculosis “DOTS Model” for managing DM has already been proposed (Harries et al. 2008), and diabetes clinics in urban areas in high burden countries need to pilot and evaluate this approach through operational research, and particularly to assess whether quarterly cohort reporting of incident cases, cumulative outcomes, complications and survival analysis can lead to better management and care, more rational drug forecasting and uninterrupted drug supplies (Harries et al. 2009).

Better point-of-care diagnostic and monitoring tests

In the same way that the rapid point-of-care (POC) HIV test revolutionised counselling and HIV testing replacing cumbersome and slow HIV-ELISA methodology (De Cock & Odhiambo 2006), better POC tests need to be developed for DM management. These tests could include measurements of blood glucose and HbA_1c, the latter helping to differentiate DM from stress-related hyperglycaemia, and they should be assessed both for diagnosis and monitoring of the disease. If such tests are to be used widely in resource-poor settings, they should:- be simple to use; rely on finger-prick blood sampling; be independent of instrumentation or electronics; be robust and able to withstand elevated ambient temperatures without cold-chain shipment or storage; have a long shelf-life; and be inexpensive (for example, costing less than USD$2 per test). These would have the potential to be used in peripheral clinics and remote health centres in Africa and Asia and would improve access to DM diagnosis and care.

Other research questions are listed in the Box. Among these, we consider TB preventive therapy to have low priority. Two studies conducted before the 1970s indicated that active TB in DM patients could be reduced through TB chemoprophylaxis (Pfaffenberg & Jahler 1958; Lesnichii & Karpina 1969). Both studies were flawed, and therefore the true benefit of chemoprophylaxis remains unknown. The question can only be properly addressed through a randomised controlled trial (RCT). However, given the expense of conducting RCTs, the low uptake of isoniazid preventive therapy in people living with HIV despite proven efficacy from a number of well conducted RCTs (WHO 2009), and other issues such as feasibility of implementation, this research is not given high priority because we feel it is unlikely to be carried forward to policy and practice. Better DM control might be safer and more cost-effective in preventing TB and reducing other DM complications.

Conclusion

The epidemic of non-communicable diseases, especially DM, is steadily growing. DM in particular threatens TB control efforts and the achievement of the 2015 TB targets (WHO 2006). In the same way that HIV threatens TB control and has lead to the development of an HIV-TB research agenda (Smart 2009), the threat of DM requires a research agenda focussed on providing tools for national and international agencies charged with the control of both these diseases. Many critical questions regarding the association between DM and TB remain unanswered either because of poorly conducted studies or no studies at all. These major concerns led to the assembly of an expert group who systematically reviewed existing data and developed a research agenda providing a clear basis for future action.