ISSN: (Print) (Online) Journal homepage:

Transcription

1 Infectious Diseases ISSN: (Print) (Online) Journal homepage: Incidence of tuberculosis and the need of prophylactic treatment in persons living with HIV in Stockholm during the era of anti-retroviral therapy Maria Norrby, Carolina Wannheden, Anna Mia Ekström, Ingela Berggren & Lars Lindquist To cite this article: Maria Norrby, Carolina Wannheden, Anna Mia Ekström, Ingela Berggren & Lars Lindquist (2018) Incidence of tuberculosis and the need of prophylactic treatment in persons living with HIV in Stockholm during the era of anti-retroviral therapy , Infectious Diseases, 50:11-12, , DOI: / To link to this article: Society for Scandinavian Journal of Infectious Diseases. Published by Informa UK Limited, trading as Taylor & Francis Group Published online: 26 Oct Submit your article to this journal Article views: 225 View Crossmark data Full Terms & Conditions of access and use can be found at

2 INFECTIOUS DISEASES, 2018; VOL. 50, NO , ORIGINAL ARTICLE Incidence of tuberculosis and the need of prophylactic treatment in persons living with HIV in Stockholm during the era of anti-retroviral therapy Maria Norrby a,b, Carolina Wannheden c, Anna Mia Ekstr om d,e, Ingela Berggren f and Lars Lindquist a a Unit of Infectious Diseases, Department of Medicine, Huddinge, Karolinska Institutet, Stockholm, Sweden; b Department of Infectious Diseases, Danderyds Hospital, Stockholm, Sweden; c Medical Management Centre, Department of Learning, Informatics, Management and Ethics (LIME), Karolinska Institutet, Stockholm, Sweden; d Department of Public Health Sciences, Karolinska Institutet, Stockholm, Sweden; e Department of Infectious Diseases, Karolinska University Hospital, Huddinge, Sweden; f Department of Communicable Disease Control and Prevention, Stockholm County Council, Stockholm, Sweden ABSTRACT Background: The aim of this observational cohort study was to determine the incidence and risk factors of active tuberculosis (TB) in persons living with HIV in a low endemic setting over a 17-year time period when combination antiretroviral therapy (ART) has been available. We thereby aimed to understand the usefulness of TB chemoprophylaxis among HIV patients with latent TB. Methods: All 2127 adult patients diagnosed with HIV January 1996 December 2013 at the Karolinska University Hospital in Stockholm County were eligible. After exclusion of 259 patients transferred to other clinics, 1868 were followed until TB diagnosis, death or end of study period (December 2013). The median follow-up time was 7.9 years (interquartile range, ). Results: Active TB was diagnosed in 92 patients, corresponding to an incidence rate of 6.2 cases (95% CI ) per 1000 person-years with a significant decline over time. The majority (52%) of TB cases were diagnosed within 1 month from HIV diagnosis. Being a migrant from a TB-endemic region, was the only patient characteristic associated with significantly higher risk of active TB (Hazard Ration, HR: 8.54, 95% confidence interval, CI: in a multivariate regression analysis controlling for route of HIV transmission, year of HIV diagnosis and CD4-cell count and viral load at HIV diagnosis. The number needed to treat to prevent one case of TB among patients in this high-risk group was 22 (95% CI 26 47). Conclusion: The incidence of active TB in persons living with HIV in Stockholm County declined significantly after the introduction of ART but was still 80 times higher than in the general population at the end of the study period. The therapeutic gain of chemoprophylaxis in low endemic settings should be weighed against costs and side effects. KEYWORDS HIV latent TB antiretroviral treatment TB incidence low endemic region Sweden ARTICLE HISTORY Received 29 March 2018 Revised 30 May 2018 Accepted 31 May 2018 CONTACT Maria Norrby maria.norrby@sll.se Department of Infectious Diseases, Danderyds Hospital, Stockholm, Sweden ß 2018 Society for Scandinavian Journal of Infectious Diseases. Published by Informa UK Limited, trading as Taylor & Francis Group This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License ( which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in any way

3 808 M. NORRBY ET AL. Introduction Globally, tuberculosis (TB) is the leading cause of morbidity and mortality among persons living with HIV and caused 374,000 deaths in 2016 [1]. Low CD4-cell counts and viral replication are known risk factors for opportunistic infections, including TB, in persons living with HIV infection [2]. The risk for active TB infection in persons living with HIV is strongly reduced by early antiretroviral treatment (ART) [3]. Sweden is a low-burden country for both HIV and TB with an HIV incidence rate of 4.3 cases per 100,000 person-years and a TB incidence rate of 7.3 cases per 100,000 person years in 2016 (Public Health Agency of Sweden). In 2016, Sweden was the first country in the world to reach the UNAIDS targets, i.e. that 90% of people living with HIV should know their HIV status, that 90% of these should be on treatment with ART and that 90% of the these should be virally suppressed [4,5]. The World Health Organization (WHO) aims to reduce the TB incidence rate by 50% and the TB-associated deaths by 75% by 2025 (compared to 2015) [1]. The WHO and the European Center of Disease Control (ECDC) have agreed on an action framework for lowincidence countries. It is currently recommended to screen individuals with higher risk of active TB, such as people living with HIV, with interferon gamma release assay (IGRA) or tuberculin skin test and to provide treatment for latent tuberculosis infection (LTBI) [6,7]. The Swedish National Board of Health and Welfare also recommends screening and treatment for LTBI in persons living with HIV. However, the National Reference Group for ART in Sweden (RAV, an independent expert group whose members are appointed by the Swedish Society of Medicine) recommends neither screening nor treatment for LTBI in people infected with HIV. The RAV and the Swedish Medical Products Agency recommendations for when to start ART in persons living with HIV have changed over time. In 1997, the CD4-cell count threshold for starting ART was 250 cells/ ml, in 2009 it was raised to 350 cells/ml and since 2014 ART is recommended for all HIV infected persons independent of CD4-cell count. The aim of this observational cohort study was to determine the incidence of and risk factors for developing active tuberculosis (TB) and to identify risk factors in persons living with HIV in a low endemic setting in the time period 1996 until 2013, when combination ARTART was available. We used a complete patient cohort in Stockholm County including all new adults diagnosed with HIV over 18 years ( ) to analyze risk factors of late TB in the cohort and to estimate the number of individuals with HIV infections needed to treat with chemoprophylaxis at diagnosis to prevent one case of active TB. Methods Study population We performed an observational cohort study at Karolinska University Hospital in Stockholm, from January 1996 through December The study included all adult patients (aged 18 years) diagnosed with HIV at Karolinska University Hospital representing 70% of all persons living with HIV in Stockholm, based on data from SmiNet, the Swedish national reporting system and registry for communicable diseases. All persons living with HIV-TB coinfection in Stockholm were exclusively treated at this hospital during this period. The unique Swedish personal identity number was used for patient identification. Persons diagnosed with HIV during the study period were identified in InfCare HIV, the Swedish national quality registry which includes more than 99% of residents living with HIV in Sweden since 1983 [5]. The patients who were transferred to other clinics during the study period (N ¼ 259) were excluded from the study as the time at risk for TB infection could not be calculated for these patients (Figure 1). From the InfCare HIV registry we extracted information on sex, year of birth, route of HIV transmission, country of origin, date of HIV diagnosis, HIV viral load and CD4 count at HIV diagnosis, CD4 nadir (during the study period) and where applicable date of death. For patients who were diagnosed with TB during the study period additional data were extracted: date of TB diagnosis, CD4-count and HIV RNA viral load at TB diagnosis, date for ART initiation and ART regimen. Information on TB manifestation was retrieved from SmiNet. Other information regarding TB including diagnostic procedures, microbiological results and treatment regimens were extracted from medical records. The patient cohort was divided into two groups, based on the year of HIV diagnosis ( and ) due to the changed guideline recommendations for the introduction of ART in During the first part of the study period ( ), guidelines recommended ART for patients with a CD4-cell count of 250 cells/ml or less; in the later part of the study period

4 INFECTIOUS DISEASES 809 Figure 1. Flow chart of included and excluded patients in the study. ( ), guidelines recommended ART for patients with a CD4-cell count of 350 cells/ml or less. Ethical approval for the study was obtained from the Regional Ethical Review Board in Stockholm (Dnr 2006/ /, 2013/ and 2015/ ). Definitions of terms and concepts Active TB was defined as a specimen positive for acid fast bacilli in smear microscopy or Polymerase Chain Reaction (PCR) or culture for Mycobacterium tuberculosis or if radiography or histology was characteristic for active TB and lead to treatment for active TB. TB diagnosed within 1 month from HIV diagnosis was not assumed to be preventable with LTBI treatment and was defined as early TB. In contrast, TB diagnosed more than 1 month after HIV diagnosis was considered to be potentially preventable with LTBI treatment and was defined as late TB in coherence with previous studies [8,9]. Patients were considered to be at risk of preventable TB from 1 month after HIV diagnosis until TB diagnosis or death or end of follow-up. Country of origin was classified as TB endemic versus low endemic, based on the WHO estimates of TB incidence in different countries [1]. Based on these estimates, countries in Sub Saharan Africa and Asia were classified as TB endemic and all other countries as low endemic. Clinical manifestation of TB was defined as isolated pulmonary if only the lungs were affected; extrapulmonary if other organs than the lungs were affected, including the pleura and thoracic lymph nodes; or disseminated if the lungs and at least one extrapulmonary site were involved. In patients from TB-endemic countries, latent TB was considered possible in absence of regular tuberculin skin test or IGRA test results. CD4 nadir was defined as the lowest CD4 value measured within the study period. ART was defined as a combination therapy involving two nucleoside reverse-transcriptase inhibitors plus one of the following: a protease inhibitor, a non-nucleoside reverse-transcriptase inhibitor, or a fusion inhibitor. ART was classified as: possibly effective if ART had been administered for more than three consecutive months; or confirmed effective if ART had been administered for more than three consecutive months and HIV-RNA was undetectable (<50 cop/ml) according to the RAV recommendations. Statistical analysis Data were analyzed using SPSS (version 24.0, SPSS, Inc. Chicago, IL). Continuous data are presented as median and interquartile ranges (IQR). Comparisons were made by non-parametric tests (Mann Whitney). Categorical data are presented as frequencies and proportions and comparisons were made by Chi-square tests. Incidence rates were calculated as the annual number of patients diagnosed with TB divided by the annual number of person years at risk of TB in the HIV cohort. Trends in incidence rates were calculated using a Poisson regression model. Sex, age at HIV diagnosis and origin were included in the model as categorical variables and calendar year as a continuous variable. The age variable was dichotomized into two equally large groups, one including patients aged years at TB diagnosis, the other including patients aged 35 and over, respectively. Results from the model are presented as percent change in incidence rate per year. To analyze the changing proportion of patients diagnosed with HIV from TB endemic regions over time, we divided the study period into three 6-year intervals ( , , ) and used v 2 -test for trend.

5 810 M. NORRBY ET AL. Cox regression was used to analyze the association of patient characteristics at HIV diagnosis with active TB more than 1 month after HIV diagnosis. The association of active TB was tested in univariate analyses with the following variables: sex (male, female), age (<35, 35), route of HIV transmission (heterosexual, other), origin (endemic, low endemic), year of HIV diagnosis ( , ), CD4-cell count and HIV viral load at HIV diagnosis (continuous variables). Variables significantly associated with risk of TB were included in a multivariate analysis adjusted also by sex and age. All statistical tests were considered significant at p- value of <.05. The numbers needed to treat (NNT) was calculated as the reciprocal of the incidence of TB (the number of TB cases/the numbers of patients at risk for TB), under the assumption that all patients had latent TB at HIV diagnosis and with a postulated effectiveness of LTBI treatment of 100% (1/X, where X is the incidence of TB). Results Patient characteristics During the study period ( ) 2127 patients were newly diagnosed with HIV. Of these, 259 patients were transferred to other clinics and were thus excluded. The remaining 1868 patients were followed until TB diagnosis or death or until the end of observation. During the study period 92 patients (5%) in the cohort were diagnosed with TB (Figure 1). The median age at HIV diagnosis in the study population was 36 years (IQR, 29 44). The majority of the patients were men (58%). Most of the patients had acquired HIV through heterosexual transmission (65%) and were originally from TB endemic countries (52%). Patients diagnosed with TB during the study period had a significantly lower median nadir CD4-cell count and median CD4-cell count at HIV diagnosis (p <.001) and a significantly higher HIV viral load (p <.001). Of the 92 patients who were diagnosed with TB, six patients (7%) died, whereof two from TB. Characteristics of the HIV cohort are presented in Table 1. The proportion of patients from TB endemic regions diagnosed with HIV increased significantly over the study period ( : 45%, : 53%, : 58%, test for trend: p ¼.004). Incidence and prevalence of TB Overall incidence of TB. The 1868 persons living with HIV were followed over 14,740 person years which corresponds to a median follow-up time of 7.9 (IQR, ) years. Of the 92 TB cases, 48 (52%) were diagnosed within 1 month after HIV diagnosis (early TB) and 44 (48%) were diagnosed more than 1 month after HIV diagnosis (late TB). The overall incidence rate of TB during the study period was 6.2 (95% CI ) per 1000 person years. During the study period we observed a significant decrease in incidence of TB; this was equivalent to a mean annual rate decline of 10.1% [95% CI, 15.1 ( 6.8)] (Figure 2). The incidence of TB during the overall study period was 1.0 (95% CI ) per 1000 person years in patients from low TB-endemic regions and 11.3 (95% CI ) per 10 3 person years in patients from TB-endemic regions. Prevalence of early TB. We estimated the prevalence of TB at the time of HIV diagnosis under the assumption that patients diagnosed early with TB, i.e. within 1 month from their HIV diagnosis, represent prevalent rather than incident cases. The overall prevalence of early TB during the study period was 2.6% (95% CI ). The prevalence of early TB increased over the entire study period; 1.6% (95% CI ) during the first three years and 4.4% (95% CI ) during the final 3 years (Figure 3). However, the observed trend of an increasing prevalence of early TB did not reach the level of statistical significance (p ¼.064). Incidence of late TB. The overall incidence of late TB was 3.0 per 1000 person years (95% CI, ). A significant decreasing incidence rate of late TB was observed during the study period [ 11.1% (95% CI, 16.9 ( 5.0)] (Figure 2). Characteristics of TB patients Of the 92 patients with HIV-TB co-infection, TB disease was culture confirmed in 79 of the patients (86%). Three were diagnosed by PCR, two by sputum microscopy, five by radiography and three by histopathology. Resistant TB strains were detected in four cultures whereof two were multi-drug resistant. Active TB was diagnosed within one year from HIV diagnosis in 68 (73%) of the 92 co-infected patients. A tuberculin skin test or IGRA test was performed in ten of the 92 patients and was positive in six. Characteristics of the HIV-TB co-infected patients (n ¼ 92) during the study period (grouped into early and

6 INFECTIOUS DISEASES 811 Table 1. Characteristics of all adult patients diagnosed with HIV in Stockholm county (n ¼ 1868). All patients (n ¼ 1868) Non-TB patients (n ¼ 1776) TB-patients (n ¼ 92) Non-TB vs. TB p-value Age at HIV-diagnosis, median (IQR) 36 (29 44) 36 (29 44) 36 (32 43).827 Sex, n (%) Female 777 (42%) 725 (41%) 52 (55%).003 Male 1091 (58%) 1051 (59%) 40 (45%) Continent of origin, n (%) TB-low endemic 891 (48%) 882 (50%) 9 (10%) <.001 TB-endemic 977 (52%) 894 (50%) 83 (90%) Route of HIV-transmission, n (%) Heterosexual 1207 (65%) 1130 (64%) 77 (84%).002 IDU 267 (14%) 260 (15%) 7 (8%) Homo/bi-sexual 285 (15%) 283 (16%) 2 (2%) Other 35 (2%) 33 (0%) 2 (2%) Not known 74 (4%) 70 (4%) 4 (4%) Year of HIV-diagnosis, n (%) (74%) 1324 (74%) 68 (74%) (26%) 452 (26%) 24 (26%) CD4 nadir (cells/ml) median (IQR) 210 ( ) 220 ( ) 100 (40 176) <.001 CD4 at HIV-diagnosis, (cells/ml) median (IQR) 310 ( ) 320 ( ) 170 (67 337) <.001 HIV viral load at HIV-diagnosis (cop/ml) median (IQR) 44,500 ( ,000) 41,000 ( ,000) 111,500 (33, ,000) <.001 Dead, n (%) 181 (10) 175 (10) 6 (7) IQR: interquartile range; IDU: Injection drug use; BT: Blood transfusion. p-values refer to comparisons of the distribution of patient characteristics between persons in the study population with TB and without TB. Continuous variables were compared using the Mann-Whitney test and categorical data using Chi-square tests. Statistically significant (p <.05) different distributions of patient characteristics are bolded. Figure 2. Incidence rate of TB in patients living with HIV in Stockholm county , divided into 3-year periods. The upper curve represents the incidence rate of TB overall. The lower curve represent the incidence of the subset of patients diagnosed with late TB (diagnosed >1 month after HIV diagnosis). late TB cases) are presented in Table 2. Disseminated TB and lower median viral load at TB diagnosis were significantly more often seen among the 44 (48%) late TB patients compared to the 48 (52%) early TB patients. In the late TB group, 34 (77%) patients were ART naïve or had less than 3 months of preceding ART. Seven (16%) had had effective ART (more than 3 months of treatment and a viral load <50 cop/ml at the time for TB diagnosis). Risk factors of late TB in the HIV cohort In a univariate Cox regression analysis we found significant associations between heterosexually transmitted Figure 3. Prevalence of early TB cases diagnosed within 1 month after HIV diagnosis, divided into 3-year periods, in patients diagnosed with HIV in Stockholm county HIV and origin from TB-endemic continent for an increased risk for active TB more than 1 month from HIV diagnosis. After adjustment in the multivariate analysis, continent of origin was the only characteristic that had a statistically significant. Numbers needed to treat (NNT) to prevent one case of active TB Based on the assumption that only the 44 patients (out of the 1795 HIV patients at risk of late TB) diagnosed with late TB could have been prevented with LTBI treatment, and, anticipating full protection for active TB with LTBI treatment, 41 (95% CI, 31 56) patients would have to be treated with LTBI to prevent one case of active TB (during a median follow-up time of 7.4 (IQR, )

7 812 M. NORRBY ET AL. Table 2. Characteristics of TB-patients with HIV (n ¼ 92) grouped into early and late TB. All TB patients (n ¼ 92) Early TB (n ¼ 48) Late TB (n ¼ 44) Early vs. Late TB p-value CD4 at TB-diagnosis, (cells/ml) median (IQR) 160 (61 260) 168 (64 259) 154 (52 282).759 HIV viral load at TB-diagnosis (cop/ml) median (IQR) 90,100 ( ,000) 346,000 (56, ,000) 10,295 (50 253,250).045 Patients diagnosed per time period, n (%) (69%) 31 (65%) 33 (75%) (31%) 17 (35%) 11 (25%) Time on ART at TB-diagnosis, n (%) No ART 64 (70%) 37 (77%) 27 (61%) 3 months 18 (20%) 11 (23%) 7 (16%) >3 months 10 (10%) 0 10 (23%) Continent of origin, n (%) TB-low endemic 10 (11%) 4 (8%) 6 (14%).414 TB-endemic 82 (89%) 44 (92%) 38 (86%) Clinical manifestation of TB, n (%) Isolated pulmonary 38 (41%) 27 (56%) 11 (25%).010 Extrapulmonary 22 (24%) 9 (19%) 13 (29%) Disseminated 32 (35%) 12 (25%) 20 (46%) Time from HIV to TB-diagnosis, n (%) <1 months 48 (52%) 48 (100%) 1 month 1 year 20 (22%) 20 (46%) 1 2 years 8 (9%) 8 (18%) 2 6 years 15 (16%) 15 (34%) >6 years 1 (1%) 1 (2%) IQR: interquartile range. p-values refer to comparisons of the distribution of disease characteristics between patients with early respectively late TB-diagnosis. Statistically significant (p <.05) differences in proportions of patient characteristics are bolded. Table 3. Results from Cox regression analysis with active TB more than 1 month after HIV-diagnosis as dependent variable by characteristics at HIV diagnosis in persons diagnosed with HIV in Stockholm County (n ¼ 1795). HIV (n) TB (n) Univariate analysis HR (95% CI) p-value Multivariate analysis HR (95% CI) p-value Total (n) Sex Male ( ) ( ) Female Reference Age (years) < ( ) ( ) Reference Route of HIV-transmission Heterosexual ( ) ( ).920 Other Reference Continent of origin TB endemic ( ) < ( ) <.001 TB low endemic Reference Year of HIV-diagnosis ( ).726 Not included Reference CD4 at HIV-diagnosis (cells/ml) < ( ).095 Not included Reference VL at HIV-diagnosis (cop/ml) < Reference.060 Not included ( ) HR: hazard ratio; CI: confidence interval; VL: viral load. Variables with a p-value >.05 in bivariate analysis were not included in the multivariate analysis except for sex and age that were considered to be potential confounding factors. Statistically significant odds ratios (p-value <.05) are bolded. years). In patients originating from TB-endemic regions the NNT was 22 (95% CI, 15 30) (during a median follow-up time of 6.7 (IQR, ) years), also anticipating a full protection of LTBI treatment. NNT could be handled and reported in different ways. If we instead reported it as an annual number, assuming a protective effect of LTBI treatment of 60%, the NNT would be 271 per year in patients from TBendemic regions. Discussion In this complete cohort of adult people living with HIV on ART in Stockholm a clear declining incidence of TB was observed, corresponding to a mean annual decrease of approximately 10%. The decline was more profound in the first half of the study period and the earlier introduction of ART launched in 2009 seemed to have little further impact on the TB incidence rate. The majority

8 INFECTIOUS DISEASES 813 (52%) of TB cases in this population were diagnosed within 1 month from HIV diagnosis. Origin from TB endemic region was the only significant risk factor for developing TB more than 1 month after HIV diagnosis. However, even within this high risk group of migrants from a TB endemic country, living with HIV, approximately 22 would have to be treated with LTBI treatment at the time of HIV diagnosis to prevent one case of active TB. The overall TB incidence and the observed significant decline in incidence rate of TB over time in our cohort are in the same range as in most other contemporary studies from low TB incidence countries [8 14]. Studies in both endemic and low endemic regions have shown that the individual risk of TB decreased significantly two to five years after the introduction of ART [11,15,16]. We did not have individual ART data on all patients with HIV diagnosis but all patients in our cohort were followed from one center and changes in treatment routines were introduced uniformly. The continuous reduction in the incidence rate of late TB during the study period, in parallel with the practice of gradually earlier introduction of ART, lends support to a highly protective effect of ART against TB. We also found that the majority (77%) of patients with potentially preventable TB (late TB) were ART naïve or treated with ART less than 3 months at time of TB diagnosis which underscores the importance of the nowadays implemented strategy of introducing ART already at diagnosis in all patients diagnosed with HIV to prevent active TB. Acknowledging the probable protective effect of immunological reconstitution after ART it is still worth noticing that the incidence of late TB in our cohort at the end of the study period was roughly 80 times higher than that of the general population in Stockholm during the same time period according to the Public Health Agency. TB still is a potentially lethal infection in persons living with HIV. To identify patients who could possibly benefit from LTBI treatment, we analyzed the association between certain patient characteristics at the time of HIV diagnosis and the risk of developing late TB. In this adjusted analysis, originating from a TB-endemic country was the only characteristic that was significantly associated with active TB development. The risk associated with origin in a TB-endemic country among persons living with HIV is not surprising and has also been demonstrated in recent studies from other low incidence countries [8,9,10 13]. Low CD4-cell count and high HIV viral load at HIV diagnosis were also associated with late TB diagnosis in two contemporary studies from Great Britain [9] and Germany but our study did not lend any support to a predictive role of these two characteristics for risk of developing late TB when modern ART was introduced early. Thus low CD4-cell count and high HIV viral load at HIV diagnosis, important predictors in the era of early ART is no longer predictive of later TB susceptibility. The remaining increased risk of TB in HIV patients in our cohort might be mediated by other so far insufficiently understood vulnerabilities in the immune system not associated to CD4-cell count. In our study the majority of active TB patients were diagnosed at or within 1 month from HIV diagnosis. The prevalence of active TB 1 month after HIV diagnosis (2.6%) in our cohort was significantly higher than found in a British study covering the time span (0.9%) [9]. This latter observation is in congruence with a recent Swedish study which shows that an increasing proportion of persons living with HIV are diagnosed late and present with advanced disease [17]. It should be emphasized that preventing TB, even in the era of ART, must include promotion of generous HIV screening and testing routines in the community. An early diagnosis reduces the risk of opportunistic infections. The current WHO guidelines still recommend screening for and treatment of LTBI in patients diagnosed with HIV in low endemic countries [7,18]. This strategy is also supported by The Public Health Agency of Sweden in their recently updated recommendations. The policy to screen all patients diagnosed with HIV for LTBI and to treat those with positive tests is supported by the success of a Swiss study from 2007 in which a significant decrease in subsequent TB incidence was shown after the implementation of a LTBI screening program [19]. This observational nationwide cohort study, where the majority of the 6160 persons living with HIV were tested with tuberculin skin test, also demonstrated that a positive screening test was a strong risk factor for later active TB (HR 25, 95% CI, 11 57) and that none of the 193 tuberculin skin test positive patients that received preventive treatment developed TB. However, current WHO recommendations are being questioned in several European countries including Sweden and most HIV clinicians neither screen nor treat newly diagnosed HIV positive patients with LTBI. The recommendation, by the RAV not to screen for and treat LTBI in persons living with HIV is justified by the fact that Sweden has reached the UNAIDS targets, that ART is available for all individuals in Sweden with a known HIV infection and that all patients are required

9 814 M. NORRBY ET AL. by law to attend regular monitoring of their HIV infection (including CD4-cell count and viral load at least twice per year). It is assumed that viral suppression makes the great majority of people living with HIV less vulnerable to active TB and that the mandatory monitoring system sufficiently facilitates early detection and treatment of any active TB case [4,5]. The RAV statement is also grounded on the reported low capacity of tuberculin skin test and IGRA screening tests to define a population at high risk of developing TB. The suggested strong protective effect of early ART (and low predictive value of screening) gains support from a European multi-center study from 2014 which shows that none of the study s 338 patients with HIV with viral suppression at the time of LTBI screening developed TB during the following years, regardless of the LTBI screening result [2]. The fear of adverse event, when combining ART and LTBI treatment (isoniazid) could not be confirmed in a Cochrane meta-analysis from 2015 [20]. However, more recent evidence in a not yet published study from high TB-endemic areas including pregnant and postpartum women, found that treatment with isoniazid during pregnancy and postpartum was associated with increased mortality due to liver toxicity [21]. In our cohort we estimated that 22 people living with HIV in Sweden who had their origin from a TB-endemic country, had to be treated with chemoprophylaxis to prevent one case of late active TB. This figure is significantly higher than the estimated NNT in the formerly cited Swiss study, where only, 8, during a median follow up time of 4.4 years, in patients with HIV with an origin from endemic countries and a positive LTBI screening test had to be treated to prevent on active case of TB [19]. However, our Swedish cohort results align with the estimated NNT in our study in the recent systematic Cochrane review of TB endemic countries where the overall NNT for LTBI treatment in tuberculin skin test positive persons living with HIV was estimated to 20 patients [22]. Important to note is however that the estimation of NNT is very much dependent on length of follow-up. While most patients in the studies from endemic countries were followed for only a year or two, the patients in our study were followed for a median time of 7.4 years. The importance of follow up time for the comparison of NNT is highlighted by the recent British study that report a seemingly high NNT of 264 to prevent one case of TB annually. Had we though instead reported our results as annual NNT and as the British study assumed a restricted protective isoniazid effect of 60%, we would end up with roughly the same estimate that 271 high risk patients would need to be treated with chemoprophylaxis to prevent one case of active TB per year. It should be noted that 30% of all persons living with HIV in Stockholm during the study period were treated at a hospital not included in this study according to the Department of Communicable Disease Control and Prevention, Stockholm. However, only few of the patients at the other hospital (15% compared to 50% at the Karolinska University Hospital) originated from TB endemic countries and the patients who developed TB were referred to the Karolinska University Hospital and thus included in our study. Because of the higher proportion of patients from TB endemic countries included in our cohort, the risk of TB shown in our study could be overestimated compared to what could be expected in the overall population of people living with HIV in Stockholm. A limitation of our study is that we did not have information on socioeconomic factors known to be related to risk of TB infection when analyzing the association of patient characteristics with risk of late TB. A possible shortcoming of our study is that it reports the outcome of persons living with HIV in the early ART era and does thus not reflect the most modern ART strategy Test and Treat, thus likely underestimating the NNT. It is worth noticing that the vast majority (37 out of 44 (84%)) of the patients diagnosed with late TB in our study had not started ART or had started ART but had not yet reached suppressed viral loads (<50 cop/ ml) at TB diagnosis. Hypothetically, these 37 cases might have been avoided with the recommendation of starting ART in all persons diagnosed with HIV regardless of CD4 count, which was implemented in If this was true only the remaining seven late TB cases, all from endemic countries, might have been avoided with LTBI treatment, which theoretically corresponds to a fivefold increase in NNT to prevent one case of active TB. However, even carefully monitored patients will probably also in the future have a substantially larger risk of developing TB, compared to the general population. Early identification and chemo prophylactic treatment of patients at high risk of TB could probably further decrease the remaining over-risk of TB in persons living with HIV. The gap between current recommendations from the WHO as well as the Swedish Public Health Agency which stills recommends screening and treatment of all pregnant women (including persons with HIV) from TB endemic countries for latent TB and

10 INFECTIOUS DISEASES 815 current practice among HV clinicians and expert opinions based on clinical observations, is striking. The HIV expert group opinion is supported by the study large study mentioned above, calling for a revision of current LTBI guidelines for pregnant and postpartum women living with HIV due to eth low risk of TB in relation to the risk for adverse toxicity events [21]. Although tuberculin skin test and IGRA test are far from optimal, there are currently no better screening tests available to identify patients at risk of active TB. As the patients in our study cohort were not routinely screened for LTBI at HIV diagnosis we, could not evaluate the predictive value of LTBI screening. The absence of LTBI screening also limits our possibility to determine if TB infection was really present at HIV diagnosis or if it in some cases might have been contracted at a later stage (and thus not preventable with early screening and prophylaxis). However, from the Swiss experience we could expect that screening would have reduced the necessary NNT to prevent one case of late active TB to one third of our estimate [19]. Taking all these premises into consideration we could hypothesize that approximately 50 screening positive persons from TB-endemic countries diagnosed with HIV today will have to be treated with isoniazide (expecting a protective effect of 80%) to prevent one case of TB during the coming 7.4 years. Whether, in this scenario, the gain of screening and prophylactic chemotherapy outweighs the costs and risks associated with tubercolostatics is a question to be debated among experts in both TB and HIV. In Sweden health care is free and of high quality. Persons living with HIV are also required by law to attend regular monitoring of their HIV infection. These circumstances explain why none of the patients in our cohort were lost to follow-up. The InfCare HIV register has a full coverage of all persons living with HIV in Stockholm and we could thus find information regarding viral load and CD4-cell count for all patients in the cohort. Despite the relatively long follow-up time of over 14,740 person years, only 92 patients with HIV were diagnosed with active TB in Stockholm during the study period. Regardless of the complete data of all persons living with HIV the scarcity of active TB in the cohort limited our capacity to find statistically significant associations. Our study confirms that the incidence rates of active TB have decreased substantially in persons living with HIV in Stockholm following the introduction of ART. Nevertheless, the incidence of TB among persons living with HIV was much higher, even during the end of the study period, than in the general Swedish population. A generous HIV screening and testing program to identify persons with HIV before their immune system has deteriorated seem to be of major importance to further decrease the remaining over-risk of TB in persons living with HIV. The conflicting recommendations from experts in TB respectively HIV calls for collaborative common guidelines, grounded on evidence and sound judgment, on screening and treatment of LTBI in persons living with HIV in low endemic. Acknowledgements We thank Eva-Lena Fredriksson at the Clinic for Infectious Diseases, Karolinska University Hospital for data from the InfCare HIV registry. Disclosure statement No potential conflict of interest was reported by the authors. References [1] WHO Global Tuberculosis Report Available from: _13Nov2017.pdf?ua ¼1 [2] Sester M, van Leth F, Bruchfeld J, et al. Risk assessment of tuberculosis in immunocompromised patients A TBNET study. Am J Respir Crit Care Med. 2014;190: [3] Suthar AB, Lawn SD, del Amo J, et al. Antiretroviral therapy for prevention of tuberculosis in adults with HIV: a systematic review and meta-analysis. PLoS Med. 2012;9:e [4] UNAIDS An ambitious treatment target to help end the AIDS epidemic Available from: unaids.org/sites/default/files/media_asset/ _en.pdf [5] Gisslen M, Svedhem V, Lindborg L, et al. Sweden, the first country to achieve the Joint United Nations Programme on HIV/AIDS (UNAIDS)/World Health Organization (WHO) continuum of HIV care targets. HIV Med. 2017;18: [6] L onnroth K, Migliori GB, Abubakar I, et al. Towards tuberculosis elimination: an action framework for low-incidence countries. Eur Respir J. 2015;45: [7] WHO Policy on Collaborative TB/HIV Activities: Guidelines for National Programmes and Other Stakeholders Geneva Available from: /tb_hiv_policy_ /en/ [8] Maniewski U, Payen M-C, Delforge M, et al. Is Systematic Screening and Treatment for Latent Tuberculosis Infection in HIV Patients Useful in a Low Endemic Setting? Acta Clin Belg. 2017;72: [9] Manavi K, Hodson J. An observational study on the incidence of tuberculosis among a cohort of HIV infected adults in a setting with low prevalence of tuberculosis. HIV Clinical Trials 2016;17:

11 816 M. NORRBY ET AL. [10] Girardi E, Sabin CA, d'arminio MA, et al. Incidence of Tuberculosis among HIV-infected patients receiving highly active antiretroviral therapy in Europe and North America Clin Infect Dis ;41: [11] Lodi S, del Amo J, d'arminio Monforte A, et al. Risk of tuberculosis following HIV seroconversion in high-income countries. Thorax 2013;68: [12] Karo B, Haas W, Kollan C, et al. Tuberculosis among people living with HIV/AIDS in the German ClinSurv HIV Cohort: long-term incidence and risk factors. Bmc Infect Dis. 2014;14:14. [13] Grant AD, Bansi L, Ainsworth J, et al. Tuberculosis among people with HIV infection in the United Kingdom: opportunities for prevention?. AIDS. 2009;23: [14] Taarnhoj GA, Engsig FN, Ravn P, et al. Incidence, risk factors and mortality of tuberculosis in Danish HIV patients BMC Pulm Med. 2011;11: [15] Lawn SD, Badri M, Wood R. Tuberculosis among HIVinfected patients receiving HAART: long term incidence and risk factors in a South African cohort. AIDS 2005;19: [16] Dembele M, Saleri N, Carvalho AC, et al. Incidence of tuberculosis after HAART initiation in a cohort of HIV-positive patients in Burkina Faso. Int J Tuberc Lung Dis. 2010;14: [17] Br annstr om J, Svedhem Johansson V, Marrone G, et al. Deficiencies in the health care system contribute to a high rate of late HIV diagnosis in Sweden. HIV Med. 2016; 17: [18] Getahun H, Matteelli A, Abubakar I, et al. Management of latent Mycobacterium tuberculosis infection: WHO guidelines for low tuberculosis burden countries. Eur Respir J. 2015;46: [19] Elzi L, Schlegel M, Weber R, et al. Reducing tuberculosis incidence by tuberculin skin testing, preventive treatment, and antiretroviral therapy in an area of low tuberculosis transmission. Clin Infect Dis. 2007;44: [20] Ayele HT, Mourik MS, Debray TP, et al. Isoniazid prophylactic therapy for the prevention of tuberculosis in HIV infected adults: a systematic review and meta-analysis of randomized trials. PLoS One. 2015;10:e [21] Gupta M, Aaron Theron, et al. Randomized trial of safety of isoniazid preventive therapy during or after pregnancy [abstract 142LB] In: Electronic Materials of Conference on Retroviruses and Opportunistic Infections, Boston, Massachusetts, [22] Akolo C, Adetifa I, Shepperd S, et al. Treatment of latent tuberculosis infection in HIV infected persons. Cochrane Database Syst Rev. 2010;1:CD