Pediatric multi-drug-resistant tuberculosis in Germany – diagnostic and therapeutic challenges of an “orphan disease”

Delay in diagnosing multidrug-resistant tuberculosis (MDR-pTB) in children prolongs time to effective treatment. Data on risk factors for pediatric MDR from low-incidence countries are scarce. Retrospective nationwide case–control study to analyze MDR-pTB cases in Germany between 2010 and 2020 in comparison to a drug-susceptible (DS)-pTB group. We included 52 MDR cases (24 tuberculosis (TB), 28 TB infection (TBI); mean age 7.3 years) and 56 DS cases (31 TB, 26 TBI; mean age 7.9 years). Groups were similar for sex, household size, and migration background. Compared to the DS group, more children with MDR were born in the Commonwealth of Independent States (CIS) (22% MDR-pTB vs. 13% DS-pTB, n.s.) and had more MDR index cases (94% MDR-pTB, 5% DS-pTB, p < 0.001). The interval between first healthcare contact and initiation of effective therapy was significantly longer in MDR-pTB (47 days) than in DS-pTB (11 days, p < 0.001), correlating with disease progression. Treatment for MDR-pTB was successful in 74%, but 22% experienced long-term adverse effects (e.g., hepatopathy, hearing loss). Conclusions: Close contact to MDR cases or birth in MDR-TB-high-incidence countries are risk factors for MDR-pTB. Early identification of potential MDR index cases by contact investigation, and susceptibility testing in children from high-burden MDR-TB countries are essential for timely diagnosis and treatment, reducing the severity of disease and treatment side effects. Trial Registration: Deutsches Register Klinischer Studien (https://www.drks.de/drks_web/navigate.do?navigationId=trial.HTML&TRIAL_ID=DRKS00023817), DRKS00023817, 2020–09-08. What is Known: •Management of children with MDR-TB remains challenging due to difficulties in diagnosing MDR-TB (lack of information on MDR index case, lack of microbiological confirmation in paucibacillary disease). •Choice of treatment regimen and monitoring of side effects. What is New: •Children with an MDR-TB index or born in a MDR-TB-high-incidence country are at higher risk of developing MDR-TB in a low incidence country. •The time lag to initiate treatment in MDR-TB is longer than in DS-TB and MDR-TB treatment involves a higher risk of adverse effects in longer treatment regimens especially with injectables.

Germany is a low-incidence country with 4,127 TB cases in 2020 (5 per 100,000), of which 163 were under the age of 15 years.Pediatric TB incidence in 2020 (1.4 per 100,000) was lower than in previous years, highest under the age of five (2.1 per 100,000) [6].
The WHO defines multidrug-resistant TB (MDR-TB) as resistant to rifampin and isoniazid [1].In 2019, 2% of all children with pediatric TB (pTB) in Germany had MDR-pTB [7].Adults with drug-susceptible TB (DS-TB) can develop into MDR-TB by secondary drug resistance, whereas children with infection by resistant strains almost exclusively acquire these from adult index cases [5,8].Due to frequently missing reference to index patients and the absence of sensitive diagnostic tools, time-to-treatment for MDR-pTB is often extended [9,10].Repeated microbiologic sampling is essential for diagnosing children, particularly in those with paucibacillary [8,11] or extrapulmonary disease [12].Drug susceptibility testing (DST) is commonly phenotypic and depends on microbiological culture.Genotypic detection of mutations is not routinely used [9,13].The challenging diagnostic process and lack of contact investigations delays administering appropriate treatment [10,14].
Children with MDR-pTB are treated with anti-TB drugs adapted to their index resistance pattern [1,4,8,11,15].Pharmacokinetic and pharmacodynamic data in children are limited, and recommended regimens changed twice during the study period.Recent WHO guidelines recommend 9-12 months of treatment for children with MDR-TB, regimens may be shorter (6-9 months) in non-severe or longer (9-12 months) in severe cases [1,3,16].Drug adverse events (DAE) are common, possibly due to the limited evidence on dosing and treatment duration in children [3,4].According to own experience and South African literature [10], delayed contact investigation and lack of knowledge of MDR-TB contributes to prolonged time-to-treatment.
We hypothesized that the time-to-treatment in children with MDR-pTB was significantly longer than for DS-pTB in Germany due to often missing microbiological results, complex diagnostics, and frequently unknown index.We aimed to determine causes for this putative delay, focusing on possible risk factors for MDR and analyzing epidemiological and clinical characteristics of such cases.Primary endpoint was the time-to-treatment.Secondary endpoints were severity of disease, DAE, and therapy outcome.

Study design
A nationwide retrospective multicenter case-control study, with non-matched control sample from the same centers.

Ethics approval
The Ethics Committee of Ruhr-University Bochum approved this study (No. 19-6545-BR, approved: 29 May 2019) and waived the need to obtain informed consent from the included patients due to the retrospective nature of this study.The study was registered with the Deutsches Register Klinischer Studien (ID: DRKS00023817; registered: 8 September 2020).

Data collection
Cases were identified through the German Society for Pediatric Pneumology at hospitals in Bochum, Berlin, Hannover, Regensburg, Kaiserslautern, Dinslaken, Soest, Essen, Karlsruhe, Wangen, Erlangen, and Bremen.Data were retrieved from the patients' medical files and by interviewing the attending physicians.Information was recorded using the Research Electronic Data Capture (REDCap) software (powered by Vanderbilt University, Nashville, Tennessee).This database was established by the Pediatric Tuberculosis Network European Trials Group (ptbnet) [17].We used the database of the Robert-Koch-Institute (RKI), where TB is notified in Germany, to verify our case numbers.

Study population
MDR was defined as resistance to isoniazid and rifampin in the child's or the index`DST [1].Inclusion criteria were a bacteriologically (culture, PCR) or clinically (symptoms, medical imaging, immunological testing) [17] confirmed tuberculosis in children below 18 years, living in Germany and diagnosed between 2010 and 2020 with informed consent.Children with drug-susceptible TB (DS-pTB) were included in equal proportions from the same centers as cases and acted as controls.Positive immunological testing without clinical, microbiological, or radiological pathology suggestive of tuberculosis was classified as tuberculosis infection (TBI) [1][2][3][4][5].Children with TBI were assumed as DS or MDR according to their index-patient.
Time-to-treatment was defined as the time between first contact with healthcare professionals for the current episode and the date of diagnosis and effective treatment initiation [10].Reasons for healthcare contact were grouped as contact tracing, migrant screening, or presence of symptoms.Symptoms were prioritized when there were multiple reasons.Country of birth was classified based on WHO disease burden definitions [15].Refugees, immigrant children and international adoptions counted as migration background.The index case was the person assumed to have infected the child [4].
Clinical history, including previous anti-TB medication and Bacillus Calmette-Guérin (BCG) vaccination, comorbidities, inherited or acquired immunodeficiency, human immunodeficiency virus (HIV) status, and TB signs and symptoms, were collected, and analyzed.Nutritional status was assessed based by the body mass index (BMI) Z-score using Ped(Z) calculator [19] based on the Central European standard values, with Z-scores under -1.96 indicating malnutrition.

Follow-up
Treatment adherence and completion were monitored, and DAEs were documented throughout treatment and up to two-year follow-up.WHO-defined treatment outcomes for TB included cured (treatment completed, clinical symptoms resolved, radiological improvement, sputum culture conversion), probably cured (missing culture conversion), treatment failed (culture remains positive or missing clinical or radiological improvement), died, and lost to followup.Children with TBI remained asymptomatic, developed TB disease, or were lost to follow-up.Recurrence was possible after treatment completion [4].

Statistical analysis
Continuous variables are expressed as mean ± standard deviation and were compared using the two-sample t-test.Discrete variables are presented as percentages and were compared using the χ 2 or Fisher's exact test.Bivariate analysis assessed associations between time-to-treatment and drug-susceptibility.Risk factors for MDR were determined by univariate analyses (odds ratio), while the best factor combination was assessed with the fast-and-frugal decision trees (FFTrees) [20].Logistic regression was not applied because high interactions between factors were assumed.We chose a dfan algorithm, a high-sensitivity aim, and a goal of 0.6-sensitivity-weighted accuracy.Statistical analysis was performed using R (R Core Team) [21] and Microsoft Excel (Version 16.58, 2022).Twosided p < 0.05 was considered statistically significant.
Distribution of sex, household size, and migration background was similar between groups.More children with MDR-pTBI (52%) and MDR-pTB (22%) were born in in the Commonwealth of Independent States (CIS) than children with DS-pTB (13%) or DS-pTBI (0%; p < 0.001).Unlike children with DS-pTB (5%), most index cases of children with MDR had drug-resistant mycobacteria (94% MDR-pTB, 100% MDR-pTBI).The annual number of cases with MDR-pTB in this study increased from one in 2010 to five in 2020.In comparison to RKI, we identified more cases because we included children clinically diagnosed as MDR-TB.

Diagnostic management
At the time of diagnosis, the country of birth was known for 75% of those with MDR-pTB, 97% DS-pTB, 96% MDR-pTBI, and 92% DS-pTBI.An index case was identified for two-thirds of the pTB patients, all those with MDR-pTBI, and 60% of those with DS-pTBI, predominantly among household members.The index` DST was available for 94% of those with MDR-pTB, 67% DS-pTB, 96% MDR-pTBI, and 53% DS-pTBI.In TBI-cases without index´ DST drug-susceptibility was presumed based on response to first line treatment.
MDR-pTB and DS-pTB differed in the rates of children with disseminated (21% vs. 0%), and extrapulmonary (25% vs. 13%) disease (p = 0.006).Severity of disease correlated with time-to-treatment (r = 0.28, p = 0.040) and treatment duration (r = 0.474, p = 0.001).Time-to-treatment was strongly correlated with treatment duration (r = 0.839, p = 0.001).Most patients with MDR-pTB (71%) were confirmed bacteriologically, over 50% by culture (vs.61% and 37%, respectively, of those with DS-pTB) (Table 2).More samples were needed to confirm MDR-pTB than DS-pTB (mean 2.8 vs. 1.7 samples; p = 0.004).Common sample Table 1 Clinical data summary of all children included in the study.The participants were grouped into those with tuberculosis disease (TB disease) or tuberculosis infection (TBI) and sub-grouped into multidrug-resistant (MDR) and drug-susceptible (DS) tuberculosis BMI body mass index, BCG Bacillus Calmette-Guérin, DST drug susceptibility testing, SD standard deviation, DR drug resistant, CIS Commonwealth of Independent States p < 0.05 is considered as statistically significant and presented in bold a Including human immunodeficiency virus (HIV) b Refugee, immigrant child, international adoption c Only countries where study participants were born d Commonwealth of Independent States (CIS): Azerbaijan, Moldova, Russia e Europe: Italy, the Netherlands, Czech Republic, Hungary, Poland, Albania, Bosnia, Romania, Serbia, Bulgaria, Ukraine, Estonia, Latvia f Africa: Angola, Gabon, Guinea, Somalia, Tunisia, Libya g Australasia: Iraq, Syria, Afghanistan, Australia (Norfolk Island)

Risk factors for MDR mycobacterial infection
Risk factors for MDR were an index case with MDR mycobacteria, birth in an MDR mycobacterial high-burden country, and disseminated disease.The optimal decision tree had 100% sensitivity and 95% specificity (53 of 56 with MDR) based on the index` DST, country of birth, site of TB, and immune competence.We could classify 72% of our study population into the MDR or control group according to their index` drug susceptibility and country of birth.In the remaining, site of TB and immune competence fitted best as decision factors (Fig. 1).

Time-to-treatment
It took over four times longer to treat MDR-pTB than DS-pTB (mean, median [range] in days: 46.96, 26  vs. 10.68,5.5 [0-81]), irrespective of the patient's age (Fig. 2).MDR-pTBI time-to-treatment was over ten times longer than DS-pTBI (40.11, 25 [0-188] vs. 3.88, 3 [0-18]).All patients with TBI were diagnosed based on their index` DST (n = 27; mean time-to-diagnosis, 40.1 days).Two were immediately diagnosed as MDR, the others index cases were Table 2 Diagnostic, treatment regimen, and outcome parameters of all study participants.The patients were grouped into those with tuberculosis disease (TB disease) or TB infection (TBI).and sub-grouped into multidrug-resistant (MDR) and drug-susceptible (DS) TB p < 0,05 is considered as statistically significant and presented in bold a time-to-treatment = time from the first healthcare contact for the current episode to the day on which the diagnosis was made and therapeutic regimen was decided b Cured, probably cured, or remained asymptomatic once treatment completed  presumed as DS until DST-results were available and partly treated with isoniazid.Time-to-treatment of cases with MDR-pTB and DS-pTB varied over the observational period (Table 2; Figs. 3 and  4).Diagnosis took longest in children with symptomatic TB and unidentified index case.Contact tracing, migrant screening, and an index with MDR mycobacteria contribute to shorter time-to-treatment (Fig. 4).Immunologic evidence of TB became available early, but targeted therapy initiation took long (mean ± standard deviation (SD) in days: 45 ± 95 MDR-pTB vs. 12 ± 54 DS-pTB) (Table 2).
All cases with DS-pTBI received isoniazid and rifampin for three months.Among children with MDR-pTBI, five were monitored without medication, while the others received DST-adapted regimens with an average duration of nine months (Fig. 5).
One child with MDR-pTBI developed TB disease, and three with DS-pTBI were lost to follow-up.The other children with TBI remained asymptomatic (Table 2).

Discussion
Despite recent improvements in the diagnostic and therapeutic management of MDR-pTB, our study in a low-incidence country confirms diagnostic latency, which is associated with a higher risk of advanced disease, long treatment duration and adverse effects.The main risk factors for MDR-TB were an index with MDR-TB and birth in MDR-TB highincidence countries like the CIS.

Time-to-treatment
In MDR-pTB cases, time to effective treatment initiation was four times longer than in DS-pTB.Immunologic evidence of TB usually became available early, but initiation of targeted therapy was often postponed, showing a wide interindividual range.Diagnosis took the longest in children with symptomatic TB disease and an unidentified index case.Low TB incidence among children in Germany and the even lower incidence of MDR-TB might contribute to the delayed diagnosis.When an adult is diagnosed with TB, detailed contact investigation is essential to detect exposed children early.Children often have paucibacillary disease and microbiological confirmation and DST is difficult.Three cases with MDR (1 pTB, 2 pTBI) and a known MDR index were diagnosed as MDR immediately.When the index case was unknown, diagnosis of drug resistance often took long and required repeated microbiological sampling due to the paucibacillary disease and infrequent use of genotypic confirmation.Children with MDR-pTBI were verified by the index DST, which was not always immediately available and only a surrogate of the child's DST.
In 2014, a South African study found that the time from sputum collection to diagnosis using GeneXpert in adults was significantly shorter than by culture [22].In our pediatric study population, GeneXpert or other nucleicacid amplification techniques were not applied regularly.Diagnosis was established by culture in 53% of the cases with MDR-pTB (37% DS-pTB).The remaining cases were diagnosed clinically and through index investigation.In an American study population, 25% were bacteriologically Fig. 1 a Risk factors for multidrug resistance in children with tuberculosis (ordinate).Data are presented in a logarithmic scale, with odds ratios and confidence intervals; BCG, Bacillus Calmette-Guérin.b Fast-and-frugal decision tree based on risk factors with the highest odds ratio to identify pediatric multidrug-resistant tuberculosis cases [n = 108; 52 with multidrug-resistant tuberculosis (MDR-TB) and 56 with drug-susceptible tuberculosis (DS-TB)].Three false-negative patients with MDR-TB were detected.Sensitivity was 100%, and specificity was 95% ◂ confirmed [23], while 70% of the participants in a British study had positive culture or polymerase chain reaction (PCR) [24].More extensive use of molecular tests (Gen-eXpert, Line Probe assays, NGS) might have facilitated the diagnosis of MDR-pTB resulting in earlier initiation of treatment [25,26].
Schaaf et al. observed an extended time-to-treatment among patients with MDR-pTB in South Africa in 2003 (median, 283 days), but shorter in children with a known source case (median, 246 days) or a known MDR-TB index case (median, 2 days) [10].Time-to-treatment varied from two days to 46 months in a meta-analysis from 2012 [27].To shorten this interval, timely contact investigations and better cooperation between public health authorities and pediatricians are needed to identify index patients or transmission dynamics [28].

Risk factors for MDR
Risk factors for MDR were exposure to an index with MDR-TB or birth in a CIS country.A flowchart for cases with pTB at high risk for MDR revealed that primary contact tracing contributed greatly to identifying those with MDR.
An index case was identified in 67% of our patients with MDR-pTB, 94% of whom had MDR-TB, and all were household members.All contacts in the American study [23] and 21/39 in the South African study had MDR-TB [10].High prevalence of TB infection among MDR-TB household members was observed in high-burden countries [29].Estimated 25,000-32,000 children develop MDR-TB disease each year but only 3-4% of them receive treatment.Effective contact investigations around adult MDR-TB might contribute to identify twelve times more MDR-pTB cases [30].These cases with MDR-pTB could be treated according to their index` DST since equivalent resistance profiles among household members were assumed [31,32], shortening the time to adequate treatment [10].Due to the assumption that in a low-incidence setting every child with TBI and an index with MDR had MDR-pTBI, 100% of the index cases of MDR-pTBI were identified.Other studies identified birth in a foreign country and previous TB treatment as risk factors for MDR-TB in lowincidence countries [33].In Germany, many children with TB had a migration background, involving language and cultural barriers.Asylum seekers in Germany have limited access to healthcare services [28].The groups differed significantly with respect to countries of birth: most patients with MDR-pTB were born in the CIS, while most patients with DS-pTB were German-born.Glasauer et al. described high rates of drug resistance among German-born and foreign-born TB-patients, particularly high in patients born in Russia (33%) [33].
One third of children in our study were BCG-vaccinated, significantly more children with DS mycobacteria than MDR, depending on country-specific WHO recommendations.BCG-vaccination is neither effective in preventing DS-or MDR-TB.In high-burden countries children should receive a single BCG-dose, significantly reducing their risk of severe TB.In Germany BCG-vaccination was abandoned in 1998 due to low-incidence.So, children born in highburden countries who immigrate to Germany are not vaccinated [1,34].

Treatment
Patients with MDR-pTB were mainly treated with linezolid (71%), moxifloxacin (67%), and cycloserine (67%), according to WHO [1].Based on previous recommendations, 54% received injectables which cause a high rate of adverse effects.In our study, hearing impairment occurred four times in patients receiving amikacin, two children did not recover.New drugs (bedaquiline and delamanid) were approved for TB treatment in Europe during this study [35].Linezolid effectiveness in patients with drug-resistant TB was documented in several studies [36].The efficacy of a reduced dose has been evaluated since linezolid is associated with frequent DAEs [37].
There is low evidence for preventive treatment or observation of MDR-TB contacts [38].Based on observational studies [39], WHO suggests levofloxacin in combination with ethambutol or ethionamide for 6 months.The combination of levofloxacin and pyrazinamide is not recommended due to high risk of adverse events [1].Clinical trials for levofloxacin in other combination [40] or monotherapy ((V-QUIN MDR-TB, Vietnam [41]), TB-CHAMP (children < 5 years, South Africa) and PHOENIx (delamanid, international)) are ongoing [42].In a South African study a 6-months-regimen with ofloxacin, ethambutol and isoniazid was well tolerated and only 4% of pediatric MDR contacts developed TB [43].

Adverse events and outcome
Five of 17 children treated with linezolid in our study had DAEs, none of seven treated with bedaquiline, and one of four treated with delamanid.Recommendations and treatment have changed over the observation period.We found no difference in treatment outcomes between MDR-pTB and DS-pTB, but patients with MDR-pTB required a longer treatment, and transient or permanent DAEs occurred in almost half of them.The DAEs in our study population (hearing or vision impairments, hypothyroidism, hepatitis, neutropenia, nausea) are commonly described in MDR-pTB [27].Two children in our study experienced permanent hearing loss, probably caused by aminoglycoside treatment [28].Treatment was successful in 74% of MDR-TB cases in our study, comparable to 78% in a meta-analysis [44].
No deaths were reported in our study, whereas 4/39 children with MDR-pTB died in the South African study [10].Two children with poor outcomes in the South African study and one in our study were infected with HIV.The pTB mortality rate in India is 5.4%, mainly due to drug resistance, delayed presentation to healthcare, or disseminated disease [45].Outcomes in developing countries are worse than in developed countries [46].

Further findings
Most cases with pTB in Europe were not diagnosed in specialized hospitals [15,28].Patients with MDR-TB in the US are cared for in public health settings [47].Although hospitals treating children with MDR-pTB in Germany differed in their specialization level, no differences in outcomes were detected [28].All but three centers treated adult index patients and children separately.We found more cases than reported by the RKI because they only considered culture-confirmed TB disease cases in children under the age of 15 as pTB [6].

Limitations
Although we aimed to include all children diagnosed with MDR-pTB in Germany from 2010 to 2020, the sample size was small corresponding to a low-incidence setting.
Selection bias could not be excluded, particularly since diagnosis of TBI relied on an index case and was assumed as MDR-TBI according to the index´ DST.The retrospective study design was another limitation.Although all original findings and documentation were studied, some data remained incomplete.Moreover, we could not include cases with DS-pTB from all participating centers, but similar numbers of controls were included from hospitals with many MDR cases over the same period.

Conclusion
Children with close contact to an MDR index case or from MDR-TB-high-incidence countries are at higher risk for developing MDR-TB.Early identification of potential MDR index cases, repeated collection of microbiological samples, and DST in children from high-burden MDR-TB countries are essential for timely diagnosis and treatment, reducing the severity of disease and treatment side effects.Bacteriological confirmation, including PCR-based DST, should always be attempted in children whose index had MDR-TB or who were born in a highincidence country.Initial treatment should follow the index` DST and resistance profile of the country of birth.The effect of contact tracing requires evaluation in further prospective studies.commented on previous versions of the manuscript.All authors read and approved the final manuscript.
Funding Open Access funding enabled and organized by Projekt DEAL.

Fig. 2 Fig. 3
Fig. 2 Density proportions (ordinate) of the time-to-treatment (in days, logarithmic scale, abscissa) of multidrug-resistant tuberculosis (MDR TB) and drug-susceptible tuberculosis (DS TB).Density probabilities were estimated using normal distribution

Fig. 4 Fig. 5 Fig. 6
Fig. 4 Factors (ordinate) affecting the time to tuberculosis (TB) treatment (abscissa, days) in children with multidrugresistant (MDR) TB (bars).Mean time to MDR-TB treatment (vertical line) was 43.26 days.Information is based on the available n as indicated.* Only available for TB disease