Diabetes and tuberculosis: An important relationship

In the current issue of the Journal of Diabetes, Wang et al.1 bring up an important aspect of the interrelationship between diabetes and tuberculosis, finding that, controlling for the effect of age, among 2280 people with pulmonary tuberculosis in seven clinics, those with diabetes or prediabetes (confirmed with fasting glucose measurement) were more likely to have delayed their visit for evaluation of symptoms. Following phagocytosis, Mycobacterium tuberculosis alters macrophage function in such a fashion as to avoid lysosomal degradation, as well as reducing intraphagosomal lipolysis, increasing nutrient availability to the infecting bacterium.2 Clinically, tuberculosis (TB) follows a sequence with inefficient innate immunity superimposed on latent infection, followed by a proliferative cellular response to infection. Among predisposing factors to TB are malnutrition, smoking, and immunocompromise, particularly from HIV/AIDS; in this setting, it is a great concern that diabetes has emerged as another leading factor,3 with studies performed over the past two decades showing it to be associated with an approximate threefold increased risk of TB,4, 5 a frequency that continues to be shown in in low- and medium-income countries,6 where approximately 95% of patients with TB and 70% of patients with diabetes mellitus live.7 For a given individual, HIV/AIDS has greater effect, increasing TB risk more than 20-fold, but diabetes, which affects a far larger proportion of the population, is likely to play an increasing role in contributing to global TB incidence,8 and currently is only slightly less important than HIV in its association with TB,9 an association that has been characterized as a “looming co-epidemic.”10 According to the World Health Organization (WHO),11 India’s TB prevalence dropped from 217 to 171 per 100 000 population from 2000 to 2013, but patients with TB and diabetes already outnumber those co-infected with TB and HIV, with tuberculosis rates around 650–900/100 000 in people with diabetes, although screening is hampered because no more than half the people with diabetes in India have been identified. In 2015, there were 10.4 million new TB cases, including 1.2 million among HIV-positive people and 1.0 million among children, with 1.4 million TB deaths in 2015, roughly comparable to the estimated direct worldwide mortality from diabetes, with an additional 0.4 million deaths among HIV-positive people with TB.12 Based on this relationship, projections suggest that simply stopping the rise of diabetes would avoid 6.0 million cases and 1.1 million tuberculosis deaths in 13 countries during the next 20 years, whereas interventions reducing diabetes incidence by 35% over the next decade would avoid a further 7.8 million cases and 1.5 million TB deaths by 2035.13 In the Mexican National Tuberculosis Registry from 2000 to 2012,14 more than 190 000 patients were registered, 19% had diabetes (compared with 9.2% of the adult population). Further, during the study period, pulmonary TB incidence rates in people with diabetes increased by 83%, whereas rates in non-diabetic people decreased by 27%.14 In the Pacific Island nation of Kiribati, diabetes prevalence was 37% among cases with TB, but 19% among controls, with an adjusted odds ratio for diabetes of 2.8.15 Much of the diabetes associated with TB may be undiagnosed. Of 4000 TB patients screened with capillary glucose measurement in 56 directly observed treatment centers in Nigeria, 480 (12%) had diabetes, with only one-third previously known.16 Given the strength of the evidence that diabetes is a risk factor for TB and for poor outcome of TB treatment, recommendations have been made for bidirectional screening to decrease the proportion of undiagnosed individuals with TB or diabetes, recognizing that integrated screening and care will enable patients with both diseases to receive appropriate treatments.17 Such programs are time-consuming and difficult. In a community screening in China, TB was diagnosed in 14 of 4085 diabetic patients, a prevalence of 343 per 100 000 diabetic people, more than eightfold the rate of 43 per 100 000 people in the local general population.18 In another study in China, 2942 diabetic patients from 10 community health clinics were screened for TB; two had known TB and 278 had symptoms (cough for >2 weeks, fever or night sweats for >4 weeks, weight loss etc.), of whom 205 had chest X-rays, with one case of TB being discovered.19 The relationship between TB and diabetes is likely to be different in high-income countries. Data from the UK Clinical Practice Research Datalink including 222 731 patients with diabetes diagnosed from 1990 to 2013 and 1 218 616 controls showed that, among 969 people identified with TB, those with diabetes had just a 30% increase in risk.20 International migration is likely to become an increasingly important factor in high-income countries (the number of migrants increased from 173 to 244 million between 2000 and 2015),21 with physical inactivity and increased availability of high-caloric foods promoting the development of diabetes. In high-income countries, rates of TB among native-born people have been static or decreased, but incidence rates among immigrants, particularly those from South America, Asia, and Africa, have been 8.7- to 18.4-fold greater, exceeding half of all TB cases in many high-income countries, either because of active TB prior to arrival, reactivation of latent TB infection, or TB acquired through local transmission in the host country.21 The interrelationship between TB and diabetes is likely to be bidirectional. In San Francisco, where 85% of patients with TB came from Mexico, China, or the Philippines, analysis of a clinic population showed that 16% of people with tuberculosis had diabetes, compared with 5% and 7% of those who did not have TB or had latent TB, respectively.22 In another study in California, the overall TB rate among people with diabetes was 21 cases per 100 000, with a relative risk of 3.5, but with a multiplicative effect of foreign birth, with diabetic foreign-born people having an overall TB rate of 142/100 000, for a >30-fold relative risk, so that the number of people needed to be screened to find one case of TB decreased 10-fold among foreign-born people with diabetes.23 In the US, during the period from 1998 to 2011, hospitalizations for TB decreased but hospitalizations for TB with a secondary diagnosis of diabetes were relatively constant, with the frequency of diabetes as a secondary diagnosis increasing from 11.2% to 19.5% of TB hospitalizations.24 In diabetes, the combination of susceptibility to infection with oxidative stress and increased tissue inflammation may well lead to greater risk of acquiring and reactivating TB, as well as predisposing to worse outcome, with recent studies suggesting poor response to mycobacterial response occurs twice as frequently with coexisting diabetes.25 In a study of 2189 adults treated at a center in Brazil, comparing 135 diabetic with 273 non-diabetic people, the diabetic patients were more likely to be symptomatic and were more likely to have positive sputum smear at treatment initiation, as well as after 30 days treatment.26 In a study of 89 diabetic and 120 non-diabetic patients with TB, treatment success rates (based on sputum smears) were 89.9% and 95.8%, respectively.27 Greater likelihood of treatment failure was also found in the Mexican National Tuberculosis Registry.14 Mortality more than doubled, from 6% in non-diabetics to 14% in diabetic patients, at 6 months in an Ethiopian study of 1314 TB patients at 44 clinics.28 Patients with diabetes–TB co-infection are more likely to remain positive after initial antitubercular treatment, and to exhibit higher resistance to frontline drugs. Moreover, these people are more likely to have multidrug-resistant TB, treatment failure, and relapse, as reported in another study from Mexico.29 The relationship between TB and glycemic control among diabetic people has not been as widely studied. Of 1594 people with culture-positive pulmonary TB in a study from Taiwan, half had diabetes and, in this group, 39% had HbA1c >9%.30 The proportions of patients with any symptom, cough, hemoptysis, tiredness, and weight loss were all highest in the group with high HbA1c, who had the highest sputum smear positivity grade and the lowest likelihood of treatment response.30 Among 661 patients with pulmonary TB in Korea, 157 had diabetes, 108 with HbA1c ≥7%; the likelihood of positive culture at 2 months in the group not having diabetes, in those with HbA1c <7% and diabetes, in patients with HbA1c between 7% and <9%, and in those with HbA1c ≥9% was 7.6%, 5.7%, 11.9%, and 34.3%, respectively.31 Type 1 diabetes (T1D) often is associated with higher levels of glycemia and, as such, TB should be considered when symptoms are present. Among 151 T1D patients in India screened for TB, mean HbA1c was 9%, 20 were asymptomatic but had a history of TB (12 of whom had positive sputum culture) and five had symptoms of cough, dyspnea, fever, and/or weight loss (of whom four had positive culture).32 Active screening of symptomatic T1D patients with sputum culture may be particularly important, as Mantoux skin testing and other ancillary diagnostic approaches may give equivocal findings.32 Compared with 20 780 Taiwanese people not having T1D seen between 2002 and 2011 in the National Health Database, there was a 4.2-fold age-, gender-, and comorbidity-adjusted increase in likelihood of developing TB among 5195 Taiwanese patients with T1D over a mean 5-year period of follow-up.33 What are the implications of the diabetes–TB relationship? Particularly in developing countries, physicians should be well aware of the potential coexistence of diabetes and TB, and that TB may not respond as rapidly to standard treatment protocols and may be multidrug resistant. In addition, TB investigations should be considered for any patient with diabetes having prolonged cough, weight loss, fever, or any of a plethora of other gastrointestinal, musculoskeletal etc. symptoms. Few diabetes clinics in China (or, for that matter, in other countries) screen patients with cough and separate them from other patients, and there is little systematic use of masks for staff, or of maintenance of ventilation of waiting and examination areas.34 In regions with sufficiently high TB prevalence, such measures appear important. Psychosocial factors pertaining to TB evaluation should be considered, such as those implied by the findings of Wang et al.1 in the current issue of the Journal. As such, the economic cost of diabetes management is high; the expenditure in south Asia (which has a high prevalence of pulmonary TB) has been estimated to be US$6 billion.35 The economic cost in diabetes–TB co-infection is going to be higher, and most of it is out-of-pocket expenditure, particularly in those belonging to low socioeconomic strata, as shown in a study from Kyrgyzstan.36 The cost of treating new cases of diabetes detected in TB patients for during a 6-month Directly Observed Tuberculosis Therapy Short (DOTS) program has been estimated as being US$3–56 million in South East Asia and US$5–92 million in Africa,37 which may pose significant road blocks in achieving UN Millennium Development Goals and targets set for diabetes and TB controls in individual countries. Greater collaboration between communicable and non-communicable disease (NCD) practitioners in health systems of developing countries could be useful in addressing this, perhaps using DOTS programs in each country for diabetes patients as well, just as these programs have been used for the treatment of patients with HIV.38 Training community workers already in place in the care of patients with TB, HIV and NCD, and affordable testing of TB and diabetes would be other proposed steps. The unique characteristics of diabetes, with overnutrition leading to insulin resistance and inflammation, coupled with the unique mechanism of M. tuberculosis infection involving abnormality of macrophage function may suggest novel approaches to TB treatment. Aspects of the dysmetabolism of diabetes increasing susceptibility to TB may be important,39 and efforts to better characterize relationships between TB and glycemia, dyslipidemia, hypercoagulability, inflammation, and insulin resistance could be fruitful. The recent suggestion that metformin may be beneficial for TB control by reducing deleterious inflammation and enhancing antimycobacterial activity of immune cells is of interest.40 One wonders what interrelationships between insulin action, glucose metabolism, and TB are yet to be discovered. 在本期杂志中,Wang等1提出了糖尿病与结核病之间具有重要的相互关系,他们发现,控制年龄的影响之后,在7个诊所就诊的2280名肺结核患者中,合并糖尿病或者糖尿病前期(经空腹血糖测量确认)的患者更有可能延迟就诊去评估他们的症状。 结核分支杆菌被吞噬之后可以通过改变巨噬细胞的功能来避免被溶酶体降解,除此之外它还可以减少吞噬体内的脂肪分解,使导致感染的细菌能够得到更多的养分2。在临床上,结核病(tuberculosis,TB)往往发生在先天免疫缺陷的患者叠加了潜伏感染之后,随后就是对感染应答的细胞增殖反应。TB的易感因素包括营养不良、吸烟以及免疫功能低下,特别是由于HIV/AIDS所导致的;在这种背景下,值得密切关注的是,糖尿病成了另外一个重要的相关因素3。过去的二十年中进行的研究结果表明,糖尿病与TB的发生风险增加约3倍相关4,5,在中低收入国家中的发生率一直是如此6,生活在这些国家的TB患者大约占95%,糖尿病患者占70%7。对于一个既定患者来说,HIV/AIDS对他的影响更大,可导致TB的风险增加20倍以上,但是受到糖尿病影响的人口比例却要大得多,对于全球TB发病率来说糖尿病所造成的影响可能越来越大8,糖尿病与TB之间关系的重要性目前可能只略低于HIV与TB之间的关系9,这种关系被称为“逐渐显现的并发流行病” 10。根据世界卫生组织数据11,印度每10万人口中的TB患病率从2000年的217名降到了2013年的171名,但是同时合并TB与糖尿病的患者已经在数量上超过了同时感染TB与HIV的患者,在糖尿病患者中TB的患病率大约为650-900/10万人,但接受筛查的患者数量有限,因为在印度糖尿病患者中接受筛查的患者数量不超过一半。在2015年有1040万名新发TB患者,其中包括120万名HIV阳性患者以及100万名儿童患者,并且在2015年有140万名患者死于TB,估计与全世界直接死于糖尿病的患者数量差不多,另外还有40万名死亡的患者同时合并HIV阳性与TB12。根据这种相关性,预测未来20年期间在13个国家中只需要防止糖尿病患病率进一步增加就可以减少600万例的TB以及110万例的TB死亡,而在未来10年期间经过干预后如果能够使糖尿病的发病率下降35%,那么到了2035年就可以进一步减少780万例的TB以及150万例的TB死亡13。 在2000-2012年墨西哥全国结核病登记系统中14,登记的患者超过了19万名,其中有19%的人患糖尿病(对照的成年人群糖尿病患病率为9.2%)。此外,在研究期间,糖尿病患者的肺结核发病率增加了83%,而非糖尿病人群的TB发病率却下降了27%。在太平洋岛国基里巴斯,在TB患者中糖尿病的患病率为37%,但是对照组中的患病率却只有19%,经过校正后的糖尿病患病率比值比为2.815。许多糖尿病合并TB的病例可能未得到诊断。使用毛细血糖测量值筛查了在尼日利亚56个直接观察治疗中心的4000名TB患者,发现有480名(12%)患糖尿病,并且只有三分之一的患者既往已经诊断糖尿病16。考虑到有充分证据表明糖尿病是TB的一个危险因素且TB的治疗效果很差,所以推荐进行双向筛查以减少未诊断TB或者糖尿病的患者比例,应认识到要将综合筛查与卫生保健进行整合才能使合并这两种疾病的患者接受适当的治疗17。这样的项目既费时又费力。在一项中国社区筛查研究中,4085名糖尿病患者中有14名被诊断为TB,每10万名糖尿病患者中的TB患病率为343名,是当地普通人群的患病率(43/10万人)的8倍多18。另外一项中国研究在2942名来自10个社区卫生诊所的糖尿病患者中筛查TB;有2名为已知TB患者,278名有相关症状(咳嗽超过2周,发热或者盗汗超过4周,体重减轻等),其中205名患者进行了胸部X线摄片,发现了1例TB患者19。 在高收入国家中TB与糖尿病之间的关系可能有所不同。来自英国临床实践研究数据库(UK Clinical Practice Research Datalink)的数据表明,从1990年至2013年共有222731名患者被诊断为糖尿病,另外还有1218616名对照者,结果发现在969名确诊TB的患者中,合并糖尿病患者的TB风险只增加了30%20。目前在高收入国家中国际人口迁移很可能成为一个日益重要的因素(人口迁移数量从2000年的1亿7千3百万增加到2015年的2亿4千4百万)21,并且缺乏体力活动以及高热量食物摄入的增加进一步促进了糖尿病的发生。在高收入国家中,本土出生人口的TB发病率已经较为稳定或者正在下降,但是在移民中,特别是来自南美洲、亚洲以及非洲的移民,发病率要高出8.7到18.4倍,在许多高收入国家中占了所有TB病例的一半以上,他们有的是在移民之前就已经有活动性TB,有的是潜伏的TB感染复活,有的是到东道国之后通过本地传播获得TB。TB与糖尿病之间的相互关系很可能是双向的。在旧金山,85%的TB患者来自墨西哥、中国或者菲律宾,一项对诊所人群的分析结果显示,在结核病患者中有16%合并糖尿病,而在对照的没有TB或者潜伏TB的人群中糖尿病的患病率分别为5%与7%22。在另外一项加州研究中,每10万名糖尿病患者中总的TB发生率为21名,相对风险为3.5,但是国外出生的患者具有倍增效应,每10万名字在国外出生的糖尿病患者中总的TB发生率为142名,相对风险大于30倍,因此在国外出生的糖尿病患者中要筛查出1例TB所需的患者数量要少10倍23。在美国,从1998年至2011年期间,因TB住院的患者在不断减少,但是第二诊断为糖尿病的TB住院患者数量却相对不变,在TB住院患者中第二诊断为糖尿病的患者比例从11.2%增加到了19.5%24。 在糖尿病患者中,对感染的易感性加上氧化应激与组织炎症的增加都有可能导致患者感染TB以及TB再次活化的风险增大,除此之外还可导致预后变差,最近有研究表明,患者合并糖尿病后对抗结核治疗反应差的机率翻倍25。在巴西的一项研究中,有2189名成年患者在研究中心接受治疗,比较了135名糖尿病患者与273名非糖尿病参与者,结果发现糖尿病患者更有可能出现症状,并且在治疗起始阶段以及治疗30天后都更有可能出现痰涂片阳性26。在一项纳入了89名有糖尿病的TB患者以及120名非糖尿病的TB患者的研究中,治疗成功率(根据痰涂片结果)分别为89.9%与95.8%27。在墨西哥全国结核病登记系统中发现糖尿病患者TB治疗失败的可能性也更高14。在一项为期6个月的埃塞俄比亚研究中,44个诊所共纳入了1314名TB患者,结果发现非糖尿病患者的死亡率为6%,糖尿病患者的死亡率为14%,糖尿病患者的死亡率要高出1倍以上28。同时感染TB的糖尿病患者经过初期的抗结核治疗后更有可能保持TB阳性,并且对一线药物表现出了更高的耐药性。 此外,据另一项来自墨西哥的研究报告,这些患者更有可能出现多重耐药TB、治疗失败以及复发29。 在糖尿病患者中,TB与血糖控制之间的关系尚未被广泛研究。来自中国台湾的一项研究结果表明,在1594名肺TB细菌培养阳性的患者中,有一半的患者有糖尿病并且在这些患者中有39%的人HbA1c > 9%30。在这组高HbA1c的患者中,出现任何相关症状(例如咳嗽、咯血、疲劳以及体重减轻)的患者比例都是最高的,他们的痰涂片阳性度也是最高的,对治疗有反应的可能性也是最低的。在韩国的一项研究中纳入了661名肺TB患者,其中有157例合并糖尿病,108例的HbA1c ≥ 7%;第2个月时在未合并糖尿病组、HbA1c < 7%、HbA1c ≥ 7%且 < 9%以及HbA1c ≥ 9%的合并糖尿病患者组中TB细菌培养出现阳性的可能性分别为7.6%、5.7%、11.9%与34.3%31。 1型糖尿病(T1D)常与更高的血糖水平相关,因此当患者出现相关症状时应该考虑是否合并TB。印度的一项研究筛查了151名T1D患者是否合并TB,平均HbA1c为9%,有20名患者无症状但是有TB病史(其中有12名痰培养阳性)32,并且有5名患者具有咳嗽、呼吸困难、发热和/或体重减轻症状(其中有4名痰培养阳性)。对有症状的T1D患者要利用痰培养进行有效的筛查,这一点可能特别重要,而结核菌素皮肤试验以及其他辅助诊断方法却可能给出模棱两可的结果。使用2002至2011年间的健康数据库,与20780名没有合并T1D的中国台湾人群相比,校正年龄、性别与合并疾病后,发现在平均5年的随访期间,5195名合并T1D的患者发生TB的可能性要增加4.2倍33。 糖尿病与TB之间的关系意味着什么?特别是在发展中国家,医生应该充分认识到糖尿病与TB有可能同时共存,并且使用常规治疗方法治疗TB可能不会迅速起效,而且还可能发生多重耐药。另外,对任何出现长时间咳嗽、体重减轻、发热或者出现任何过多的其他胃肠道与肌肉骨骼等症状的糖尿病患者,都应该考虑进行TB检查。在中国(或者,在其他国家也有这个问题)很少有糖尿病诊所对咳嗽的患者进行筛查并且将他们与其他患者隔离开来,并且很少有工作人员系统性地使用口罩或保持候诊区与检查区的通风34。在TB流行率非常高的地区这些措施显得非常重要。应该考虑对TB相关的心理社会因素进行评估,就像当前这期杂志中Wang等1的研究结果所提示的那样。 因此,糖尿病管理的经济成本很高。据估计南亚(该地区的肺TB患病率很高)的相关费用就达到了60亿美元35。同时合并糖尿病与TB的经济成本将更高,并且大部分都是自费支出,特别是对属于低社会经济阶层的患者来说,来自吉尔吉斯斯坦的一项研究证实了这一点36。在一项为期6个月的直接观察TB治疗效果的短期项目(Directly Observed Tuberculosis Therapy Short,DOTS)中,据估计治疗TB患者中新诊断出的糖尿病病例的费用在东南亚需要300-5600万美元,在非洲需要500-9200万美元37,这对于要达到联合国千年发展目标以及各个国家的糖尿病与TB控制目标来说可能都是一个严重的障碍。为了有效解决这个问题,在发展中国家的卫生系统中,传染性疾病与非传染性疾病的医务从业人员可能需要进行更紧密的协作,也许在每个国家也要实施针对糖尿病患者的DOTS项目,就像那些针对HIV患者治疗的项目一样38。要培训社区工作者来护理TB、HIV与非传染性疾病患者,而另外一个提议就是要进行可以负担的TB与糖尿病检测。 糖尿病的显著特征就是营养过剩导致的胰岛素抵抗与炎症,加上结核分枝杆菌感染的独特机制使得巨噬细胞功能出现异常,这意味着需要新的TB治疗方法。糖尿病代谢紊乱可导致患者对TB的易感性增加这一点可能很重要39,另外我们可能还需要更好地了解TB与血糖、血脂代谢紊乱、高凝状态、炎症以及胰岛素抵抗之间的关系。最近有一项令人感兴趣的研究表明二甲双胍可能对控制TB有益,它可以减少有害的炎症反应以及增强免疫细胞抗结核分支杆菌活性40。人们还想知道的是目前尚未明确的胰岛素作用、葡萄糖代谢与TB之间的存在着怎样的相互关系。