During 2010–2016, the 13 participating centers enrolled all patients <18 years of age who were evaluated for NTM lymphadenitis into the study registry. Most patients were recruited prospectively, but 1 center in Graz, Austria, recruited patients retrospectively from a local registry containing comprehensive data on NTM cases dating back to 2001. In addition to NTM cases, we enrolled a control cohort of patients without chronic illnesses (i.e., with infectious, malignant, or immunologic diseases) who were treated in the same institutions. We matched NTM cases with controls for age, sex, and center.

For comparison with the general population of Germany, we used the nearest neighbor method to match the age and sex of the control cohort with 17,641 participants in the German Health Interview and Examination Survey for Children and Adolescents (KiGGS), 2003–2006 (12,13). We used 7-fold oversampling to achieve the most accurate matches, resulting in an age- and sex-matched healthy control cohort of 966 children. We compared data on body mass, breast-feeding history, allergies, number of siblings in the same household, smoking during pregnancy, and parents’ education level between the KiGGS-derived and the NTM cohorts.

We considered NTM likely for patients with the following symptoms: cervical lymphadenitis for >3 weeks; a lymph node size of >2 cm; exclusion of other causes, such as bartonellosis, toxoplasmosis, infectious mononucleosis, and lymphoma; and a positive tuberculin skin test (cutoff of 5 mm). We considered NTM to be the definitive diagnosis in patients with typical histology, such as presence of granuloma with or without necrosis or positive Ziehl-Neelsen staining or positive culture or PCR from lymph node samples. We documented the course of disease, including all diagnostic and therapeutic measures, antimicrobial drug therapy, associated clinical problems, prior medical history, vaccination status, and underlying conditions. Parents provided sociodemographic data, including ethnicity, place of birth, duration of breast-feeding, number of patient siblings, daycare attendance, cigarette smoke exposure, and parents’ education and employment, as well as factors potentially increasing NTM exposure, such as animal contact, exposure to water and soil sources, and travel. We tracked patients until they recovered completely. When symptoms remained at the last documented visit, we conducted telephone interviews to assess the remainder of the disease course.

We obtained parental informed consent for all children included in this study. Each participating center’s ethics committee granted ethics approval. The institutional review board of the University Medical Center, University of Freiburg, Freiburg, Germany, was the lead approval agency under IRB no. 232/10. All data were anonymized before analysis.

We performed data analysis by using R 3.5.3 (14). We used univariate analysis to screen for potential associations of individual disease history, socioeconomic factors, potential exposure to mycobacteria, and the clinical course of NTM and filtered for a correlation coefficient >0.2 and p<0.05 by using Pearson or Spearman correlations, depending on the variables. We subsequently examined the results for biological plausibility. We compared data from various categories by using the Fisher exact test and compared continuous and ordinal variables by using the Wilcoxon rank-sum test or Welch t-test. We analyzed seasonality of NTM by fitting a generalized linear regression model assuming a sinusoid Poisson distribution over the year (cosinor). Using no seasonal pattern as the null hypothesis, we assessed significance of seasonal variance by using the cosinor test (15). We considered p<0.05 statistically significant. For some analyses, we combined centers with <20 patients into a single group to attain meaningful comparisons and to ensure appropriate case anonymization.

In this study, we found that girls 18–36 months of age were at highest risk for NTM lymphadenitis, which aligns with findings of studies conducted in other geographic locations (3,4,7,18–21). The underlying cause for the predominance of NTM in girls remains unclear. We speculate that exposure to soil-residing NTM could be associated with the increased risk for NTM infection. However, the amount of time girls spent outside did not correspond to disease risk. Moreover, the predominance of NTM in female patients in varying geographic and climatic regions suggests that differences in immunity or host–pathogen interaction are more likely to be responsible for sex-related differences in NTM lymphadenitis incidence, rather than differences in exposure to NTM.

We also found a seasonal incidence pattern that peaked during colder months. Other studies reported an increase in new cases during winter (7,20), but peak incidence in relation to temperature and length of daylight hours varied among studies in different geographic regions. To expand on other studies, we recorded the duration of symptoms at each patient’s initial visit to participating centers. Using this information, we were able to correct for the time between onset of symptoms and diagnosis because differences in referral and availability of appointments might influence the seasonal pattern. We noticed that time from onset of symptoms to the first physical examination at a clinic varied greatly between centers. Year-round differences in NTM exposure also could affect disease incidence. For example, seasonal NTM occurrence in drinking water has been reported (22,23).

Furthermore, the low incidence of NTM infections in warmer months could be linked to seasonal variations in vitamin D levels. Variations in vitamin D metabolism previously have been linked to NTM susceptibility (24,25). However, in our study, patients and healthy controls did not differ in sunlight exposure, and vitamin D supplementation in the first year of life occurred more often in the NTM-infected cohort than in controls.

Given the slow replication of NTM, incubation time for NTM lymphadenitis is unknown. Thus, a critical timeframe for infection is difficult to establish from the seasonal pattern seen in our cohort. In addition, unknown host factors, such as concurring infections, might affect seasonal incidence.

M. avium causes the majority of NTM lymphadenitis cases in countries as far removed from each other as Germany, Sweden, and Australia (3,4,7,18). By contrast, NTM species isolated in pulmonary infections have much more pronounced regional differences (26). Other than M. avium, we noted variations in NTM species between centers, but the small number of cases precluded a more detailed analysis. Nevertheless, the relatively high number of rare species underlines the need for improved PCR and culture techniques for reliable diagnosis of uncommon NTM species.

Therapeutic procedures, disease outcomes, and complications varied substantially among centers and over the course of the study. Overall, patients frequently had complicated clinical courses that lasted >12 months or major complications, even after complete extirpation, the treatment most associated with a favorable outcome. More than 75% of patients in the 3 most experienced study centers had complicated clinical courses, which is a striking contrast to a report by Lindeboom et al. (16) in which early complete lymph node extirpation cured >90% of lymphadenitis cases. However, lymphadenitis frequently is caused by factors other than NTM. Our data might reflect more on clinical practices in which early extirpation is more the exception than the rule, a hypothesis further supported by the fact that clinical outcomes did not correlate with center size in our study. Consequently, surgical experience might not explain the unfavorable outcomes in our study. However, because only specialized centers participated in the study and inclusion criteria were strict, complicated surgical cases probably were overrepresented in our cohort. This finding is supported by evidence that some patients underwent primary surgery elsewhere and were referred to our participating centers after they received a definitive NTM diagnosis.

The rate of facial nerve palsy in our study was within the range reported elsewhere (7), but we saw variations among centers. Because our study relied on local clinical data, we cannot rule out differences in sensitivity thresholds. However, differences in the surgical approach reflected by the varying rates of complete extirpation most likely explain this observation. Primary complete lymph node extirpation was associated with a lower rate of facial nerve palsy than other surgical procedures. However, this observation could have a high bias because complete extirpation is performed more frequently when the affected lymph node is farther from the facial nerve. Antimicrobial therapy for NTM lymphadenitis decreased over the course of the study, which might reflect the increasing number of studies questioning the therapeutic benefit of antimicrobial drugs to treat NTM lymphadenitis.

Our study has several limitations. First, the limited number of disease controls precluded a more detailed analysis of possible individual risk factors. Second, our cohort probably does not reflect the full variety of NTM lymphadenitis phenotypes. Our strict inclusion criteria and registry-like design likely overrepresented severe cases, which are seen more frequently in secondary and tertiary health centers, and underrepresented less severe cases. Third, because we relied on data acquired by local centers and did not collect detailed information on PCR methods, we cannot rule out differences in laboratory methods or interpretation of clinical findings.

In conclusion, the NTM lymphadenitis risk profile for female patients <5 years of age, the chronic (albeit usually benign) course of disease, the worldwide predominance of M. avium, and the seasonal variability we noted in our study suggest a complex contribution of host, pathogen, environmental, and potential microbiome factors. Individual factors are insufficient to grasp the risk for unfavorable clinical outcomes. Our proposed risk score comprises multiple items and could be useful in estimating NTM lymphadenitis risk and stratifying patients to therapeutic modalities, if its validity is confirmed in a prospective study. Until then, early and complete extirpation of a suspicious lymph node remains a mainstay of NTM diagnosis and therapy.