Occurrence of extrapulmonary tuberculosis is associated with geographical origin: spatial characteristics of the Frankfurt TB cohort 2013–2018

Introduction Tuberculosis (TB) is caused by M. tuberculosis complex (MTB) and pulmonary tuberculosis (PTB) is its classical manifestation. However, in some regions of the world, extrapulmonary TB (EPTB) seems to be more frequent. Methods We performed a retrospective cohort study of all TB patients treated at University Hospital Frankfurt, Germany, for the time period 2013–2018. Patient charts were reviewed and demographic, clinical, and microbiological data recorded. Patients were subdivided according to their geographic origins. Results Of the 378 included patients, 309 were born outside Germany (81.7%). Three WHO regions were significantly associated with the occurrence of isolated EPTB: the South-East Asian Region (OR 3.37, CI 1.74–6.66, p < 0.001), the African Region (2.20, CI 1.25–3.90, p = 0.006), and the Eastern Mediterranean Region (OR 3.18, CI 1.78–5.76, p < 0.001). On a country level, seven countries of origin could be demonstrated to be significantly associated with the occurrence of isolated EPTB: India (OR 5.58, CI 2.30–14.20, p < 0.001), Nepal (OR 12.75, CI 1.73–259.28, p = 0.027), Afghanistan (OR 3.64, CI 1.14–11.98, p = 0.029), Pakistan (OR 3.64, CI 1.14–11.98, p = 0.029), Eritrea (OR 3.32, CI 1.52–7.47, p = 0.003), Somalia (OR 7.08, CI 2.77–19.43, p < 0.001), and Turkey (OR 9.56, CI 2.52–47.19, p = 0.002). Conclusion Geographical origin is a predictor for the occurrence of extrapulmonary TB. This might be linked to a delay in diagnosis in these patients, as well as specific responsible impairments of the host’s immune system, possible virulence factors of MTB, and relevant comorbidities. Supplementary Information The online version contains supplementary material available at 10.1007/s15010-022-01921-9.


Introduction
Tuberculosis (TB) is a multi-systemic disease caused by the M. tuberculosis complex (MTB, comprising among others M. tuberculosis, M. bovis, and M. africanum) [1]. It is presumed that approximately 1.7 billion individuals are at least latently infected with MTB [2]. However, only a fraction of those patients develop clinically symptomatic disease within their lifetimes [3]. The Global Tuberculosis Report of 2021 estimates that there were approximately 10 million new TB cases for the year 2020. Of those, 3-4% of patients suffer from multidrug-resistant TB or rifampicin-resistant TB (MDR/RR-TB). Eight high incidence countries (India, Pakistan, Nigeria, South Africa, China, Bangladesh, Indonesia, and the Philippines) account for two-thirds of all notified cases worldwide [4]. In high-income countries, such as Germany, TB has become mainly an imported infection among foreign-born patients [5].
TB can be described as a clinical spectrum ranging from latent tuberculous infection (LTBI) to active disease [6]. Pulmonary TB (PTB) is the most common and classical manifestation of TB. On the other hand, there is a large variety of extrapulmonary TB (EPTB) such as tuberculosis of the pleura, lymph nodes, bone, the central nervous system or the genitourinary tract [7]. There seem to be differences in host susceptibility to developing these different clinical manifestations. Recently, an association between the occurrence of extrapulmonary TB with different geographical origins has been described: Sotgiu et al. found an African origin or being from the Indian Subcontinent to be positive predictors for extrapulmonary TB, while Hayward et al. showed that migrants from South-East Asia and sub-Saharan Africa tend to suffer more frequently from extrapulmonary TB (EPTB) than those from other geographic regions [8,9]. In addition, impairments of cellular immunity (such as HIV or immunosuppressive treatment), as well as specific defects in the interferon-gamma axis (e.g., Mendelian susceptibility to mycobacterial disease) are known risk factors for mycobacterial infections and lead to disseminated infections that might require prolonged antimycobacterial treatment [10].
At Frankfurt University Hospital, we take care of a large number of TB patients in the Metropolitan Region. This group involves, among others, patients with a migration background, patients with HIV infection, as well as immunocompromised patients, for example after solid organ or stem cell transplantation. In this study, we aimed at elucidating whether geographic origin is a predictor for different clinical forms of TB in our cohort.

Database query, inclusion criteria, and exclusion criteria
We performed a laboratory database query for all patients with positive cultures or PCR tests for MTB, as well as a patient database query for all patients coded with ICD-codes A15-A19 for the time period 2013-2018. Patients, for whom no clinical data were available, were excluded, as well as patients that had established the initial diagnosis of TB prior to the observation period. Therefore, we included all patients with a clinical, microbiological, or histological diagnosis of TB from 2013 to 2018.
Using the local hospital patient information system (ORBIS, Agfa Health Care, Bonn, Germany), we performed a chart review to retrieve relevant patient information: age, gender, geographical origin (by country and by WHO regions), education, microbiological results (mycobacterial species, culture, PCR results, susceptibility testing and site of infection), antimycobacterial therapies, clinical manifestations, comorbidities, and the occurrence of lethal events. Observation time was recorded as time from TB diagnosis to the last clinical contact. Data collection forms were adapted from an ongoing TBnet study (https:// www. tbnet. eu/ migra nt-proje ct).
Resistance classifications were applied following the WHO definitions (before 2018) and German guideline definitions [11]: fully drug susceptible (DS-TB), monoresistance (resistance to one first-line TB drug), multidrug-resistant TB (MDR, resistance to isoniazid and rifampicin), extensively drug-resistant TB (XDR, resistance to isoniazid, rifampicin, to at least one fluoroquinolone and one of the injectables), and polyresistance (resistance to more than one first-line TB drug, but not meeting MDR or XDR definitions).
PTB was defined as affection of the lung parenchyma with positive radiological signs or positive microbiological specimens from a respiratory sample. EPTB included all other sites of infections including pleura, lymph nodes, abdominal manifestation, bone affection, urogenitary TB, and affection of the central nervous system, the spine, or others. This study was approved by our local ethics committee under file number 2021-270.

Statistical analysis
All data were analyzed in R v. 4.1.2 "Bird Hippie" [12]. Continuous data are depicted as mean with range for normally distributed data and as median with interquartile range (IQR) for non-normally distributed data. Categorical variables are shown as numbers and percentage. We used the Wilcoxon signed-rank test to detect differences in continuous data and the Fisher exact for differences in categorical variables between groups. Univariate logistic regression was conducted in R using a linear model. First, all WHO regions were tested against the European Region as a reference. Second, all countries of origin were tested against a German origin as a reference. Multivariate analysis was performed including the geographic origins of patients (in form of the WHO region), HIV status, administration of immunosuppressive therapies, and age. Odds ratios (OR), as well as confidence intervals (CI), were recorded. For all statistical tests, a confidence level of alpha = 0.05 was used. All graphs were drawn using the ggplot2 package within the tidyverse [13,14].
Multivariate analysis showed the same geographic regions to be associated with the occurrence of isolated extrapulmonary TB (Table 5). On the other hand, an HIV infection or an immunosuppressive therapy was a negative predictor (OR 0.25, CI 0.10-0.55, p = 0.001, and OR 0.20, CI 0.06-0.54, p = 0.004, respectively).

Discussion
In this study, we demonstrate that different geographical origins are associated with different clinical manifestations in TB.
In Germany, a majority of TB cases are observed in patients born outside the country, while the incidence in the autochthonous population is constantly decreasing. This is reflected in our study population, in which 309 patients (81.7%) were born abroad. Overall, we show a very low case fatality rate of 3.4%. In comparison to other mycobacterial infections, such as infections with non-tuberculous mycobacteria (NTM), this is an indicator for the excellent treatment options of TB in a high resource setting.
Origin from three WHO regions (SEAR, AFR and EMR) and especially seven countries (India, Pakistan, Nepal Afghanistan, Eritrea, Somalia, Turkey) was shown to be significantly associated with the occurrence of isolated extrapulmonary TB, while a majority of patients in our cohort originated from the Horn of Africa (Eritrea, Somalia, and Ethiopia) or the Indian Subcontinent (India, Nepal, Bangladesh, Pakistan, and Sri Lanka). However, only WHO regions with ten or fewer subjects did not qualify as being significantly associated with isolated EPTB. Nevertheless, our results are in line with health claims data on a European scale: Sotgiu et al. have shown that provenance from Africa or the Indian Subcontinent was significantly associated with the occurrence of extrapulmonary TB, while Hayward observed an association with origin from South-East Asia and sub-Saharan Africa [8,9].
These differences might be linked to a delay in diagnosis as primary affection of the lung might not be treated in the respective home countries of patients born abroad. In addition, constraints that arise during a strenuous travel to Germany and socioeconomic factors might contribute to this effect. However, most patients were living for more than 3 years in Germany and most were living in their own flats or housings. Interestingly, patients from the seven countries associated with the occurrence of isolated extrapulmonary TB were significantly younger and suffered less frequently from relevant comorbidities. The fact that patients from Germany received immunosuppressive therapies more frequently and were suffering more frequently from HIV might be another explanation four our findings: these factors were shown to be negative predictors for the occurrence of isolated extrapulmonary TB and were positively associated with pulmonary TB (alone or with extrapulmonary foci). This might be linked to easier dissemination of the disease in immunocompromised hosts. Therefore, the geographical differences observed in our study might be partially explained by this effect. Amirkhani et al. have demonstrated that HIV infection was negatively associated with EPTB in Ethiopia, as well [15]. However, Khalife et al. for example describe a predominance of extrapulmonary TB in HIV-positive patients in Ukraine [16]. Besides obvious impairments of the immune system, such as HIV and medical immunosuppressive therapy, there might be susceptibility factors within a specific host attributable to geographic origin. Hypothetically, differences between different populations in the interferon gamma pathway and therefore the immune response to mycobacterial infection might be responsible for the variable clinical manifestations. In addition, different mycobacterial lineages have been shown to be associated with geographical origin [17]. This could be another factor explaining for different clinical manifestations. However, our analysis showed that other factors such as HIV infection and immunosuppressive therapies are unevenly distributed among the study population. This study has several limitations: first, it is a monocentric study; second, we could not provide typing or whole genome sequencing data of bacterial isolates to be correlated with geographic origin; third, although for a single center case numbers are considerable, bigger study populations would be needed to underscore the shown effects.
In conclusion, we show that isolated extrapulmonary TB is more frequent in patients from India, Nepal, Pakistan, Afghanistan, Eritrea, Somalia, and Turkey at our tertiary care center. This observation gives a hint that geographical PTB pulmonary TB, EPTB extrapulmonary TB, OR odds ratio, CI confidence interval, NA not applicable, Inf infinite *For the occurrence of isolated EPTB. Significant associations are shown in bold. ORs and p values for countries with less than three patients are not shown   Specific impairments of the host's immune system, possible virulence factors of the bacterium, as well as a delay in diagnosis and relevant comorbidities contributing to this effect will have to be investigated in future studies.