Epidemiology and clinical impact
In the 1920s, before the introduction of insulin, it was estimated that infections killed 20% of all diabetes patients, commonly tuberculosis (Bell & Hockaday 1996). At that time, most diabetes was insulin-dependent, juvenile diabetes (type 1 diabetes). As mortality from type 1 diabetes declined with effective treatment, type 2 diabetes gained in significance. Towards the end of the 20th century, type 2 diabetes predominates. Today, more than 90% of diabetes is type 2 diabetes, and much is associated with the global pandemic of obesity (Centers for Disease Control & Prevention 2006).
The total number of people with diabetes (half of whom are undiagnosed) rose to an estimated 371 million in 2012 and is projected to rise further (International Diabetes Federation 2012; World Health Organisation 2012). Even in sub-Sahara Africa, where obesity has not been generally considered a major problem, the estimated number of persons with diabetes is expected to nearly double to 23.9 million by 2030 (82% undiagnosed) (Hall et al. 2011; International Diabetes Federation 2012). China with 92 million and India with 63 million have the largest numbers of people with diabetes. Four-fifths of people with diabetes live in LMICs where infectious diseases remain highly prevalent and medical care less available.
Recently, the re-emergence of the association of diabetes and tuberculosis has become widely recognised globally (Jeon & Murray 2008; Young et al. 2009; Harries et al. 2011). However, many other pulmonary pathogens also result in increased morbidity and mortality in diabetes, particularly influenza, pneumococcal and staphylococcal pneumonias. Also important are pneumonia with opportunistic pathogens such as Klebsiella pneumonia, Pseudomonas aeruginosa and many fungi (Peleg et al. 2007; Santhosh et al. 2011). Other pneumonias are also seen in association with diabetes, particularly those due to Mycoplasma pneumoniae, Chlamydia pneumoniae, Legionella spp. and Hemophilus influenzae. Interestingly, one report of all hospitalised patients with pneumonia returning to India from Makkah following the Hajj found 55% had diabetes (Asghar et al. 2011). With the current estimated burden of diabetes predicted to rise to 553 million by 2030, all-cause pneumonia deaths in patients with diabetes will have increasing global impact (International Diabetes Federation 2012).
The re-emergence of tuberculosis associated with type 2 diabetes is relatively recent, but is now well documented globally (Olmos et al. 1989; Mugusi et al. 1990; Swai et al. 1990). The adjusted odds ratio for this association in Hispanics was reported in 1997 to range from 2.95 (95% CI 2.61, 3.33) to in 2004, 6.4-fold (95% CI 5.7, 8.2) (Pablos-Mendez et al. 1997; Ponce-De-Leon et al. 2004). Independently our own studies showed similar associations in South Texas (Perez et al. 2006; Restrepo et al. 2007, 2011). Reports from many countries across the globe followed rapidly, particularly in Asia (Leung et al. 2008; Dooley & Chaisson 2009; Viswanathan et al. 2012). Even pre-diabetes increases the risk of tuberculosis (Viswanathan et al. 2012). In 2008, a working group met to review a comprehensive meta-analysis of the literature, assess the global situation and determine the research agenda (Jeon & Murray 2008; Harries et al. 2009, 2010). This group then developed the WHO framework for management of diabetes and tuberculosis (Harries et al. 2011; World Health Organisation, Diabetes Program 2011). Predictions of the impact of diabetes on tuberculosis have been extensively reviewed by Harries et al., and it is now recognised that diabetes threatens tuberculosis control globally (Young et al. 2009; Dye & Williams 2010; Ruslami et al. 2010; Goldhaber-Fiebert et al. 2011; Hall et al. 2011; Harries et al. 2011). The increase in tuberculosis due to diabetes is particularly felt in high-burden countries such as Peru and the Russian Federation, and in India, where diabetes is estimated to have increased the number of cases by 46% (Dye et al. 2011; Bygbjerg 2012).
Severity of diabetes is also important. A large study in Hong Kong among elderly persons found those with poorly controlled diabetes had higher risk of tuberculosis (HR 1.77 95% CI 1.41, 2.24) (Leung et al. 2008). In people with HbA1c >7% the hazard ratio was 3.11 (95% CI 1.63, 5.92) for active pulmonary tuberculosis. Also of considerable significance is that diabetes makes tuberculosis management much more challenging and may increase drug resistance (including XDR-TB) and tuberculosis mortality (Baker et al. 2011; Tang et al. 2011; Jimenez-Corona et al. 2013). The risk ratio in a recent meta-analysis for treatment failure and/or death in tuberculosis with diabetes is 1.69 (95% CI 1.36, 2.12) (Baker et al. 2011).
Epidemiological data on pneumococcal, staphylococcal and influenza pneumonia in diabetes are scarcer than those for tuberculosis. However, it has been widely recognised from the early 20th century that certain pathogens have higher morbidity and mortality in the patient with diabetes (Smith & Poland 2000). This has been amply demonstrated in an extensive review of hospital-based and community-acquired studies of influenza and pneumococcal pneumonia mostly from developed countries (Smith & Poland 2000). Some of the deaths in influenza were, however, attributed to secondary bacterial infections particularly in patients with end-organ complications of diabetes such as renal failure (Smith & Poland 2003).
Streptococcus pneumoniae kills more than 40 000 people in the United States and one million globally each year. However, bacteriological diagnosis, particularly in LMICs, is often unavailable and is in any event insensitive, leading to major under-reporting. Where data are available, minimum estimates of incidence range from 35 per 100 000 in adults aged 20–59 years to 69 per 100 000 in those over 60 years (Fedson & Scott 1999). A case–control study of community-acquired pneumococcal pneumonia found an adjusted odds ratio for having diabetes of 1.5 (95% CI 1.1, 2.0), particularly in younger adults without coexisting morbid conditions and in males (Thomsen et al. 2005). Although not statistically significant, the risk of death from pneumococcal pneumonia in a US-based study from 1978 to 1997 was reported to be twice as high in patients over 50 years of age with diabetes (95% CI 0.8, 4.7) (Mufson & Stanek 1999). Nasal carriage of staphylococcus aureus is reported in 30.5% persons with diabetes compared with 11.4% in controls without diabetes (Lipsky et al. 1987). Staphylococcal infections are common in diabetes patients; however, the increased risk of staphylococcal pneumonia in diabetes is not clear (Breen & Karchmer 1995; Joshi et al. 1999).
The most complete documentation of influenza pneumonia and diabetes came again from the 2009 pandemic of S-OIV. In one series of hospitalised patients with influenza (n = 160), the adjusted odds ratio (OR) for having diabetes was 4.72 (95% CI 1.81, 12.3), making this a greater risk than that from cardiac disease (adjusted OR 1.77, 95% CI 0.61, 5.16). Interestingly, these patients were over three times as likely to be 20–40 years of age (average age 28 years). Among the 31 who were admitted to the intensive care unit, ten had diabetes. Although small and biased towards hospitalised patients, this study shows the clearest evidence of the increased risk of diabetes for influenza pneumonia, independent of other risk factors (Allard et al. 2010). Other reports do not allow discrimination of risks independent of obesity, but risks are compounded when the conditions coexist (Gill et al. 2010).
Pathogenesis and immunology
The increased morbidity and mortality due to pneumonia in diabetes is rooted in similar defects in immune surveillance and responses to those in obesity. Damage to lung microvasculature characteristic of diabetes could also be involved, but there is little evidence to support this hypothesis. The defects appear once more to be closely related to the chronic inflammatory syndrome in which both the innate and adaptive immune systems are chronically up-regulated. Although responses to infectious antigens are brisk, we are beginning to understand that, similar to obesity, the effectiveness of both innate and immune systems in controlling infections and killing invading organisms is considerably impaired.
In type 2 diabetes, there are more data implicating specific immune system defects. Diabetes is characterised by a progressive impairment of glucose control and insulin secretion leading to insulin resistance and pancreatic β-cell dysfunction, and disposing to cardiovascular, renal and other chronic conditions (Seshasai et al. 2011). Hazard ratios for morbidity and mortality in diabetes are nonlinear, increasing markedly with deteriorating glucose control, and other comorbidities of diabetes, including obesity, compound the risks (Leibovici et al. 1996; Bertoni et al. 2001). As previously stated, many patients with type 2 diabetes are obese, and with the addition of diabetes, metabolic dysregulation broadly affects both the innate and adaptive immune systems (Bastard et al. 2006; Mathis & Shoelson 2011). Defects specifically affect minimum lung function, antibody response, neutrophil and macrophage function, CD4+/CD8 ratios, and natural killer cell function (Pickup 2004).
Innate and acquired immune responses are not separate entities but operate in concert with a complex response both to new invaders and recognised antigens. Alterations in innate immune responses in diabetes are documented, but laboratory data have been difficult to reconcile in part due to variations in methodology and lack of reproducibility, such that the picture is fragmented and hard to relate to clinical observations (Peleg et al. 2007). In summation, it appears that neutrophil adherence to vascular endothelium is increased, but chemotaxis and transmigration into tissue reduced (Peleg et al. 2007). Neutrophils from persons with diabetes have been found to have impairments in migration (Sawant 1993), phagocytosis (Krol et al. 2003), production of reactive oxygen species (Marhoffer et al. 1994) and apoptosis (Tennenberg et al. 1999). The downstream effects of these impairments are decreased ability to respond to sites of infection, inability to effectively destroy and clear pathogens and promotion of excessive, damaging inflammation (Droemann et al. 2000; Kobayashi et al. 2010).
Macrophages and monocytes in diabetes have also been shown to have functional impairments in phagocytosis, activation and antigen presentation (Dooley & Chaisson 2009) Alveolar macrophages from persons with diabetes have been shown to have reduced activation and antimicrobial activity in response to challenge with M. tuberculosis (Wang et al. 1999). Decreased populations of monocytes with complement receptor 3 (used for adherence and phagocytosis of pathogens) are associated with diabetes (Chang & Shaio 1995). Macrophages have compromised functionality of Fc receptors used for antigen processing and presentation, including diminished internalisation of Fc-receptor-bound material under conditions of insulin resistance, so that the ability to process and present information to other immune effector cells is inhibited (Abrass 1991). Dysregulation of adhesion molecules E-selectin, vascular adhesion molecule-1 and intracellular adhesion molecule-1 have been reported. These molecules are up-regulated during the innate immune response and aid in the recruitment of macrophages and other leucocytes to sites of infection (Andreasen et al. 2010).
There are elevated levels of pro-inflammatory cytokines in diabetes, both at baseline and after immune stimulation. In diabetes, interleukin-8, interleukin-1b, IL-6 and TNF-α have higher baseline and expression levels after experimental stimulation of cells with bacterial lipopolysaccharide. Innate and type 1 cytokine responses were significantly higher in tuberculosis patients with diabetes, nevertheless these patients fail to control the mycobacterium adequately (Restrepo et al. 2008). Conversely, in a rodent model, Th1-related cytokines and expression of inducible nitric oxide synthetase were reduced in experimentally infected animals with diabetes (Yamashiro et al. 2005). In our own studies, we have documented significantly higher levels of pro-inflammatory cytokines in resting plasma samples from our cohort of Mexican Americans with high rates of diabetes (Mirza et al. 2012). Stimulation of whole blood from diabetes patients with heat-killed pneumococci resulted in a 10-fold increase in the secretion of interferon-γ, IL-6 and interleukin-17 compared with blood from normal controls. For example, we found that neutrophils from diabetes patients producing neutrophil extracellular traps (NetS) when exposed to S. pneumoniae exhibited impaired ability to killed bacteria.
Also promising are mouse studies that have implicated a shift in the balance of subsets of T cells in diabetes, including increases in pro-inflammatory cytokine-producing T helper (Th)-17 cells and decreases in regulatory T cells (T regs) that mediate and control inflammation (Feuerer et al. 2009; Jagannathan-Bogdan et al. 2011). An emerging hypothesis suggests that the elevated levels of cytokines IL-6 and interleukin-1β (which promote Th-17 cell differentiation) in diabetes skew the balance of T cells to produce higher levels of Th-17 cells and suppresses the levels of inflammation-mediating regulatory T cells (T regs) (Jagannathan-Bogdan et al. 2011). The end result of this is the chronic, heightened pro-inflammatory state that further contributes to tissue injury and dysregulation of immune responses. This hypothesis provides a novel explanation for the role of effector T cells in immune impairment among persons with diabetes. Additional studies are necessary to further elucidate the mechanisms behind these observations as the targeting of this balance of Th-17 and T regs could provide novel approaches to future therapies.