Background
Tuberculosis (TB) is the most common opportunistic infection in HIV-positive patients. HIV-TB co-infection was estimated to affect 1.2 million people worldwide in 2015, and to cause 400,000 deaths [
1]. HIV-positives are 20–40 times more likely to develop active TB infection compared to HIV-negatives [
2]. The most important risk factors for developing active tuberculosis among HIV-positives are low CD4 count, living in high TB-incidence regions and absence of antiretroviral therapy [
3‐
5].
High mortality has been reported in HIV-TB patients in Africa, particularly in patients with advanced disease [
6]. In Africa, Asia and Eastern Europe the leading cause of death has been TB. Data about the outcome of HIV-TB co-infections in high-income countries has been limited. A multi-centre study showed that among patients from Western Europe and Argentina, 50% of the deaths during the first three months after the TB diagnosis were TB-related [
7]. For patients who died at a later stage the cause of death was predominantly non-TB-related or unknown.
A Cochrane review of 12 randomized treatment trials of latent TB infection in HIV-positives showed a 32% reduction of the incidence of active TB [
8]. The reduction was 62% in those with positive tuberculin skin test (TST). Overall, there was no evidence that TB preventive therapy versus placebo reduced all-cause mortality. The need for treatment of latent tuberculosis infection (LTBI) in high-income countries has remained controversial. The WHO [
9] and US [
10] guidelines recommend testing of all HIV-positives for LTBI and treatment of all patients with positive result in interferon gamma release assays (IGRA) or tuberculin skin test. EACS [
11] and BHIVA guidelines [
12] recommend IGRA-testing and medication only after risk assessment based on CD4 count, country of origin and length of previous antiretroviral therapy. Despite recommendations for treatment of LTBI, the implementation of this strategy has been limited in high-income countries [
4].
The aim of this study was to describe in detail the incidence of TB among HIV-positives and the outcomes of HIV-TB co-infections in a high-income country with low tuberculosis prevalence.
Methods
Study participants
This is a retrospective register study based on the cohort of HIV-patients registered for follow up at the Infectious Disease Clinic of the Helsinki University Hospital. This is the only HIV clinic in the Helsinki region in Southern Finland with a population of 1.5 million people, where 63% of all HIV patients in Finland have been diagnosed. The treatment of HIV and tuberculosis is integrated and free of charge for patients living in Finland.
All 1939 patients in the Helsinki HIV Cohort between 1st January 1998 and 31st December 2015 were included in this study. Classified as HIV-TB co-infections were 53 patients diagnosed with an active TB infection in Finland between 1995 and 2015. We excluded four patients, whose tuberculosis diagnosis were made abroad before migrating to Finland.
Data sources
The InfCare HIV database (RealQ platform, Health Solutions, Sweden) of the Helsinki HIV Cohort has provided the basic data about the patients and details about the TB diagnosis and outcome have been collected from the electronic patient records (Uranus, CGI, Canada) and laboratory database of the Helsinki University Hospital. The categorical variables obtained from the databases were: sex, mode of HIV transmission, country of birth, outcome of TB treatment (WHO classification), MDR TB diagnosed, MAC co-infection diagnosed, AIDS defining diagnoses. The continuous variables obtained were: date of birth, date of death, date of TB diagnosis, date of HIV diagnosis, dates of TB treatment, date of initiation of HIV treatment, CD4 at time of TB diagnosis, CD4 nadir. All new HIV and TB infections in Finland are reported to the National Infectious Disease Register, and from there we have confirmed that all co-infections in the Helsinki region are included in this study. The causes of death were obtained from Statistics Finland.
Genotyping of Mycobacterium tuberculosis isolates
Genotyping was performed with the spoligotyping technique as previously described [
13]. The spoligotypes were assigned to international SIT types available in the SITVITWEB database [
14] [
15]. For spoligotypes for which a SIT code was not available, a local F (Finnish) code was given.
Statistics
For the statistical calculations, the SPSS Statistics software version 20 (IBM Corporation, NY, USA) and Stata 15.1 (StataCorp LLC, TX, USA) were used. Incidence rates and incident rate ratios and their 95% confidence intervals were calculated using Mantel-Haenszel methods. Score test was used to estimate trends of rates of change in incidence between the different 5-year periods. In the survival analyses of HIV-TB cases Log-rank test was used to calculate p-values for categorical variables and Wald test (Cox model) for continuous variables. Hazard ratios and their 95% confidence intervals were calculated using Cox model. Kaplan-Meier plots were used to show estimations of the probability of survival in different subgroups of patients with HIV-TB co-infection.
Discussion
During the last 20 years, the HIV incidence in Finland has been slightly increasing, mainly because of immigration from Africa and Asia. The tuberculosis incidence has been gradually decreasing, as the number of elderly people with exposure in childhood has declined.
The decline in the incidence of tuberculosis among HIV-infected shown in this study was expected. A comprehensive care center for marginalized HIV-infected IDUs was started in December 2000 [
17]. The outbreak of HIV among IDUs in the Helsinki region was contained in a couple of years after the start of the care center [
18]. A symptom-based interview of the clients of the care center was carried out in 2002 after the fourth case of tuberculosis among IDUs was identified. The interview yielded one additional case of tuberculosis.
We do not have a definite explanation for the decline in tuberculosis incidence among people from Sub-Saharan Africa. However, the increase in antiretroviral therapy coverage may have played a central role. The coverage of antiretroviral therapy among people from Sub-Saharan Africa in our clinical database has been rising continuously, in 1998 63%, in 2005 65%, in 2010 74%, and in 2015 93% (unpublished data).
The genotypes of the Mycobacterium tuberculosis strains showed that imported TB cases have not caused chains of infections among HIV-patients in Finland. Most TB infections in migrants have been activations of latent tuberculosis infections received in the country of birth. Only two small clusters (SIT53 and SIT42) among Finnish IDUs in the beginning of the 2000s could be recognized. This suggests that TB infections in HIV-positives in Finland are mostly diagnosed and treated at an early stage and do not cause a significant public health concern.
The rate of successful treatment of TB was 66% in our Finnish cohort. This can be compared to WHO’s global numbers of treatment success rate of 88% for HIV-negative and 74% for HIV-positive TB-patients [
1]. In Europe, the success rate for all TB treatments has increased from 67 to 75% between 1995 and 2012. For new HIV-positive TB-cases registered in Europe in 2013 the TB treatment success rate was 47% in a cohort of 9504 patients. In both the European region and the Region of the Americas, 19% of HIV-positive TB patients died during treatment, compared with just over 5% of HIV-negative TB patients.
A study from Uganda with 302 HIV-TB co-infected enrolled 2007–2009 reported an all cause mortality of 20% during an average follow up time of about one year [
6]. Among our 11 HIV-TB coinfected patients from Sub-Saharan Africa only one died, despite longer follow up times. This could be explained by early initiation of ART, which was shown to be the strongest factor protecting from death in the study from Uganda.
There was a rather high number of loss to follow-up seen in the HIV-TB patients in this cohort. Probable reasons were that a substantial part of the subjects were migrants living only temporary in Helsinki. Also IDUs might have a more unregular living situation compared to other residents.
The average duration of TB treatment was nearly 1 year. This cannot be explained by drug resistance, as only two cases of MDR TB were diagnosed. Instead, this seems to reflect the fact that most of the TB cases were treated before year 2006, when the first national Finnish guideline for TB treatment was published, which recommended shorter treatments than earlier. Until that there had been a tradition of long treatment durations in Finland.
The small number of deaths in each subgroup has to be considered when interpreting factors affecting mortality. The all-cause mortality of HIV-TB co-infections in our cohort during an average follow up time of 9 years is 30%. However, none of these 16 deaths were directly related to TB. Among the 16 fatalities, 14 were Finnish males. The all-cause mortality of Finnish males with HIV-TB co-infection was 59%. The reason for this high mortality rate is most likely an accumulation of other risk factors of early death, such as heavy alcohol consumption, smoking, injecting drug use and psychiatric illness. To reduce the mortality rate in this group, active follow-up and multidisciplinary social and psychiatric support could be useful. On the other hand, among HIV-positive females and migrants, TB-infection does not seem to affect their long-term survival.
The benefits of isoniazid preventive therapy (IPT) for latent tuberculosis has remained controversial as most of the randomized trials showing efficacy of this strategy where conducted in Africa before the beginning of highly active antiretroviral therapies [
19‐
21]. The recent TEMPRANO study from Ivory Coast showed that 6 months of IPT has a durable protective effect in reducing mortality in HIV-infected, even in people with high CD4 cell counts and who have started ART [
22,
23]. HIV cohort studies from Switzerland and the UK have shown a significant reduction in incidence of active TB after treatment of latent TB [
4,
5]. However, the number needed to treat to avoid one case of active TB was 15 and among those with ART as high as 35 [
4]. During the time period of our study, IPT was not routinely used to prevent TB in our center.