Increase in Childhood Nontuberculous Mycobacterial Infections After Bacille Calmette-Guérin Coverage Drop: A Nationwide, Population-Based Retrospective Study, Finland, 1995–2016

Abstract

Background

Epidemiological data on childhood nontuberculous mycobacterial (NTM) disease is scarce and the protective effect of bacille Calmette-Guérin (BCG) vaccination remains debated. In 2006, the BCG policy in Finland changed from universal to selective. We aimed to study the effect of the BCG coverage decrease on the incidence of childhood NTM infections in Finland.

Methods

We conducted a nationwide, population-based, retrospective study of NTM notifications recorded to the National Infectious Diseases Register between 1995 and 2016 and identified native-born children aged 0–4 years infected with NTM. Poisson log-linear model was used to estimate the change in the incidence rate of cohorts born during universal or selective BCG policy between 1995 and 2015.

Results

We identified 97 native-born children aged <5 years infected with NTM (median age, 27 months; female-to-male ratio, 2:1). The most common species was Mycobacterium avium (n = 69 [71%]). The estimated incidence rates of NTM in universal-BCG and selective-BCG cohorts were 0.2 and 3.9 per 100000 person-years, respectively. The incidence rate ratio of selective-BCG cohorts compared to universal-BCG cohorts was 19.03 (95% confidence interval, 8.82–41.07; P < .001).

Conclusions

After infant BCG coverage in Finland decreased, childhood NTM infections increased drastically. As there is no other apparent cause for the increase, this indicates that BCG offers protection against childhood NTM disease. This observation adds to the understanding of childhood NTM epidemiology and might explain why the disease is emerging in some countries.

nontuberculous mycobacteria, NTM, lymphadenitis, BCG, vaccination coverage

Nontuberculous mycobacteria (NTM) represent a large fraction of the genus Mycobacterium [1]. Other mycobacteria, Mycobacterium tuberculosis (MTB) and Mycobacterium leprae, cause tuberculosis (TB) and leprosy, respectively. Although NTM are mostly nonpathogenic, some can cause disease in humans. NTM are ubiquitous in the environment, which is the likely source of infection [1]. In children, infection predominantly presents as a cervicofacial lymphadenitis between 1 and 5 years of age [2, 3]. Although those with impaired immunity are especially susceptible, most affected children are immunocompetent [2, 4].

Mycobacterium species share many antigens, and cellular immunity is essential for the immune response to both MTB and NTM [1]. The bacille Calmette-Guérin (BCG) vaccine contains attenuated strains of Mycobacterium bovis and is used for TB prevention. In animal models, BCG has also shown a protective effect against certain NTM infections [5–7]. Furthermore, BCG has additional nonspecific or heterologous effects independent of vaccine-specific immunity that are beneficial [8, 9]. After cessation of universal BCG vaccinations in Sweden in 1975, childhood NTM infections increased [10]. While universal BCG vaccination in Finland continued, childhood NTM infections remained rare [11].

National surveillance studies and published data on childhood NTM epidemiology are scarce, and in some developed countries NTM infections seem to have emerged even without change in BCG coverage [1, 12, 13]. As recent observations after BCG coverage change have not been published, the protective effect of BCG against NTM remains debated [1]. Therefore, to understand the role of BCG and the burden of childhood NTM disease, better epidemiological data are required.

On 1 September 2006, the BCG vaccination policy in Finland changed from universal to a risk group–based approach, and consequently BCG coverage of infants dropped from >98% to an estimated 6% [14]. Such a recent and drastic change in BCG coverage has not been observed anywhere else. Furthermore, NTM isolates in Finland have been mandatorily notifiable since 1995 to the National Infectious Diseases Register (NIDR) [15].

Our aim was to study the effect of the BCG coverage change in Finland on the incidence of childhood NTM infections.

METHODS

We conducted a nationwide, population-based, retrospective study of NTM notifications recorded to the NIDR between 1 January 1995 and 31 December 2016 (accessed 5 April 2017). The NIDR collects information on the informant (source and date of notification), patient (personal identity code, gender, nationality, date and country of birth), specimen (source, type, and date of specimen), and culture results. From the registry data, we identified native-born children aged <5 years infected with NTM. We defined a case of NTM infection as having ≥1 isolate from any site and considered cases to be incident at the time of specimen collection.

As a part of national surveillance and control, clinical microbiology laboratories are obligated to notify new NTM isolates directly to the NIDR. In clinical laboratories that perform primary detection of mycobacteria, the NTM isolates obtained from patient specimens are isolated according to standard laboratory procedures: acid-fast smear with the auramin and Ziehl-Neelsen methods, and culture on solid (Lövenstein-Jensen) and liquid media (BACTEC 480 and, since 2000, BACTEC MGIT 960, Becton Dickinson, Franklin Lakes, New Jersey). Since 2010, direct detection of mycobacteria has also been performed with the GeneXpert MTB/RIF assay (Cepheid, Sunnyvale, California) and with the GenoType Mycobacteria Direct assay (Hain Lifescience, Nehren, Germany). Isolates are further submitted to the Mycobacterial Reference Laboratory at the National Institute for Health and Welfare where the identification of mycobacterial species is performed with the GenoType Mycobacterium CM/AS assays (Hain Lifescience, Nehren, Germany) and with sequencing of the 16S ribosomal RNA gene.

Population data was obtained from the Statistics Finland population database (accessed 29 March 2017) [16]. Based on the registry and population data, we estimated the annual incidence of NTM infection in native-born children aged 0–4 years. The BCG status of individual patients was not available from the NIDR or National Vaccination Register and we did not contact the patients or families directly. To assess the effect of the BCG coverage change, we estimated the incidence rates of NTM infection per 100000 person-years (PY) among birth cohorts born in Finland between 1995 and 2015. The children born in 2006, however, were further divided into 2 birth cohorts: those born during universal BCG policy from January to August and those born during risk group–based BCG policy from September to December. All birth cohorts born before September 2006 were denominated as “universal-BCG” and those born thereafter as “selective-BCG” cohorts. We assumed that the birth rate was equal year-round, so that, for example, all children born during 2015 and registered until 31 December 2016 were observed for 1.5 years. Birth cohorts that turned 5 before 31 December 2016 were observed for the maximum of 5 years. The estimated incidence rate per 100000 PY was calculated by dividing the number of NTM cases per cohort with the number of live-born children per cohort multiplied by the number of years of observation per cohort, and then multiplying the whole number with 100000.

Poisson log-linear model was used to estimate the incidence rate ratio (IRR) and relative risk reduction (RRR) between cohorts. Goodness-of-fit criteria was examined so that the ratio of the deviance to degree of freedom (value/df) was <1.50. A P value <.05 was considered statistically significant. Data were collected and analyzed with Microsoft Excel version 15 (Microsoft, Redmond, Washington) and IBM SPSS Statistics version 24 (IBM, Armonk, New York) software.

Data in this study were analyzed within the epidemiological research purposes authorized by the Finnish Communicable Diseases Act 1227/2016, 42 §. Therefore, ethical approval was deemed unnecessary.

RESULTS

A total of 100 children aged <5 years were identified from the NIDR. After excluding foreign-born children (n = 3), the rest were included in the data analysis (n = 97). The female-to-male ratio was 2:1 (65 females and 32 males). The age at the time of specimen collection ranged from 4 to 59 months with a median age of 27 months (interquartile range, 21–34). Due to the small number of cases in the universal-BCG cohorts, we did not compare demographic characteristics between the groups. The most common isolated NTM species was M. avium (n = 69 [71%]) and culture specimen was a needle aspirate or tissue biopsy (n = 78 [80%]). The sources of the culture specimens and isolated species are presented in Table 1.

Table 1.

Source of the Culture Specimens and Isolated Species of 97 Native-Born Children Aged 0–4 Years With Nontuberculous Mycobacterial Infection Registered Between 1995 and 2016 to the National Infectious Diseases Register in Finland

	No. (%)
Culture specimena
Needle aspiration or tissue biopsy from
Lymph node	7 (7)
Cervicofacial mass or anomaly	6 (6)
Unspecified abscess	4 (4)
Unspecified source	61 (63)
Purulent or other discharge material	11 (11)
Skin	2 (2)
Nasopharynx	1 (1)
Unknown	5 (5)
Isolated mycobacterial species
M. avium	69 (71)
M. lentiflavum	7 (7)
M. malmoense	6 (6)
M. intracellulare	3 (3)
M. bohemicum	2 (2)
M. interjectum	2 (2)
M. fortuitum	1 (1)
M. abscessus	1 (1)
M. gordonae	1 (1)
M. scrofulaceum	1 (1)
M. simiae	1 (1)
M. kansasii	1 (1)
NTM, unspecified	2 (2)

Abbreviation: NTM, nontuberculous mycobacteria.

^aBased on data reported to the registry.

Between 1995 and 2016, the native-born under-5 population ranged from 280049 to 322369. The annual number of NTM infections and estimated annual incidence of NTM infection in native-born children aged 0–4 years ranged from 0 to 20 per year and 0 to 6.7 per 100000 children, respectively (Figure 1).

Figure 1.

Annual number of infections and estimated annual incidence per 100000 persons of nontuberculous mycobacterial infection in native-born children aged <5 years registered between 1995 and 2016 to the National Infectious Diseases Register, Finland.

Altogether, the universal-BCG cohorts contained 678221 children observed for 3391105 PY and selective-BCG cohorts contained 549000 children observed for 2290820.5 PY. The percentage of native-born children with foreign-born mothers in the birth cohorts increased from 3.0% to 11.5%. The sex ratio (males/females) of the birth cohorts remained relatively constant between 1.04 and 1.06. Key characteristics and the estimated incidence rates for individual birth cohorts are detailed in Table 2.

Table 2.

Characteristics of Finnish Birth Cohorts From 1995 to 2015 and Estimated Incidence Rates per 100000 Person-Years of Nontuberculous Mycobacterial Infection in Children Aged <5 Years

BCG Policy	Birth Cohort	Cohort Population, No.	No. (%) With Foreign-Born Mother	Total PYa	No. of Cases	Incidence Rateb
Universal	1995	63067	1881 (3.0)	315335	0	0.0
	1996	60723	1989 (3.3)	303615	0	0.0
	1997	59329	2133 (3.6)	296645	0	0.0
	1998	57108	2267 (4.0)	285540	2	0.7
	1999	57574	2382 (4.1)	287870	0	0.0
	2000	56742	2381 (4.2)	283710	0	0.0
	2001	56189	2633 (4.7)	280945	0	0.0
	2002	55555	2696 (4.9)	277775	1	0.4
	2003	56630	2825 (5.0)	283150	1	0.4
	2004	57758	2959 (5.1)	288790	0	0.0
	2005	57745	3220 (5.6)	288725	0	0.0
	I/2006c	39801	2378 (6.0)	199005	3	1.5
Selective	II/2006d	19039	1138 (6.0)	95195	5	5.3
	2007	58729	3690 (6.3)	293645	16	5.4
	2008	59530	3923 (6.6)	297650	14	4.7
	2009	60430	4290 (7.1)	302150	11	3.6
	2010	60980	4760 (7.8)	304900	16	5.2
	2011	59961	4969 (8.3)	299805	9	3.0
	2012	59493	5415 (9.1)	267719	7	2.6
	2013	58134	5625 (9.7)	203469	8	3.9
	2014	57232	6219 (10.9)	143080	3	2.1
	2015	55472	6363 (11.5)	83208	1	1.2

Abbreviation: BCG, bacille Calmette-Guérin; PY, person-years.

^aTotal PY in cohort observed up until 31 December 2016 or 5 years of age.

^bNumber of cases per 100000 PY.

^cBorn from January to August 2006.

^dBorn from September to December 2006.

The incidence rate of NTM infection in native-born children aged 0–4 years was estimated at 0.2 per 100000 PY for the combined universal-BCG cohort and 3.9 per 100000 PY for the combined selective-BCG cohort. The IRR of the selective-BCG cohorts compared to the universal-BCG cohorts was 19.03 (95% confidence interval [CI], 8.82–41.07; P < .001; value/df 1.4). The RRR from the 2006–2015 and 2006–2013 selective-BCG cohorts was 0.10 (95% CI, .02–.18; P < .05; value/df 0.45) and 0.08 (95% CI, –.02 to .17; P = .11; value/df 0.47), respectively.

DISCUSSION

After the BCG policy in Finland changed from universal to selective, childhood NTM infections increased drastically. As the BCG policy change resulted in decreased coverage of infants, the increase indicates that non-BCG-vaccinated children are more susceptible to NTM infections.

National surveillance studies and published data on childhood NTM epidemiology are scarce [1, 17]. To our knowledge, our study represents the only one examining the effect of BCG coverage change on the incidence of NTM infection in >20 years. The only 2 previously published studies are from Sweden and the Czech Republic, where BCG policy changed in 1975 and 1986, respectively [10, 18]. The incidence of NTM presented in our study and other published studies compare well when BCG policy is taken into consideration (Table 3) [2, 10, 11, 19, 20]. In our study, the difference observed between selective-BCG and universal-BCG cohorts was clear (IRR, 19.03 [95% CI, 8.82–41.07]). Compared to the ratio between non-BCG-vaccinated and BCG-vaccinated presented by Romanus et al [10] from Sweden (5.9 [95% CI, 1.6–48.5]), it is well within the CI. In the Czech Republic, protective effect of BCG against childhood NTM infections was also previously observed [18]. In some studies childhood NTM infections have increased without significant change in BCG coverage, but none of these show a sharp increase [12, 13].

Table 3.

Reported Incidence of Nontuberculous Mycobacterial Infections in Children Aged <5 Years in Published Studies and the Bacille Calmette-Guérin Vaccination Policy of the Cohort Population or in the Respective Countries During the Study Period

Study	Country	Study Period	BCG Policya	Age Group, y	Measure of Disease Frequency	Frequency/100000
Our study	Finland	1995–2011b	Universal	0–4	Person-time incidence rate of infectionc	0.2d
Katila et al (1987) [11]	Finland	1977–1986	Universal	1–4	Annual incidence rate of cervical adenitis	0.6
Romanus et al (1995) [10]	Sweden	1969–1974	Universal	0–4	Annual incidence of extrapulmonary infection	0.06
Our study	Finland	2006–2016e	Selective	0–4	Person-time incidence rate of infectionc	3.9d
Romanus et al (1995) [10]	Sweden	1975–1980	Selective	0–4	Annual incidence of extrapulmonary infection	2.5
Romanus et al (1995) [10]	Sweden	1981–1985	Selective	0–4	Annual incidence of extrapulmonary infection	5.7
Romanus et al (1995) [10]	Sweden	1986–1990	Selective	0–4	Annual incidence of extrapulmonary infection	4.5
Hermansen et al (2017) [20]	Denmark	1991–2015	Selective	0–4	Annual incidence rate of infection	5.4
Haverkamp et al (2004) [2]	Netherlands	2001–2003	Selective	0–4	Annual incidence of extrapulmonary infection	2.3

Abbreviation: BCG, bacille Calmette-Guérin.

^aBCG vaccination policy as reported in the BCG World Atlas [19].

^bBirth cohort from January 1995 to August 2006.

^cThe annual incidence rate presented is for native-born children.

^dPer 100 000 person-years.

^eBirth cohort from September 2007 to December 2015.

In our study, there was a slight decreasing trend in the selective-BCG cohorts born between 2006 and 2015 (RRR, 0.10 [95% CI, .02–.18]). However, the 2014 and 2015 birth cohorts were observed only up to 2.5 and 1.5 years of age, respectively, and children who develop disease at a later age are missing from these birth cohorts. After excluding these birth cohorts from the model, there was no decreasing trend between the selective-BCG cohorts (RRR, 0.08 [95% CI, –.02 to .17]).

In our study and the previous study from Finland, there was a clear female predominance (female-to-male ratio of 2:1 and 3:1, respectively) [11]. Slight female predominance in adults with pulmonary NTM disease and children with any NTM infection has also been observed elsewhere [10, 12, 18, 21–24]. It is unclear why the female predominance in Finnish children seems to be more distinct.

In Finland, BCG vaccinations generally take place in maternity hospitals at infancy, most likely before first NTM exposure. After the policy change, the BCG coverage of birth cohorts is estimated to have decreased rapidly from >98% to 6%, allowing comparison between BCG-vaccinated and non-BCG-vaccinated children living in otherwise relatively similar settings [14]. Other explanations that might be considered to cause the NTM to emerge are changes in detection, notifications, investigations, exposure, pathogenicity, and/or host susceptibility.

During the study period, detection of mycobacteria from primary specimens has remained largely the same. Liquid culture methods have been used in Finland since the 1990s—first the radiometric BACTEC 480 system and, since 2000, the BACTEC MGIT 960 system. Direct detection of mycobacteria became available in Finland in 2010, yet childhood NTM infections had clearly emerged earlier and all registered cases were culture positive. Notifying a positive NTM culture to the NIDR has been mandatory since 1995 and thus allows for a reliable examination of nationwide notifications during the complete retrospective period.

The number of all requested mycobacterial cultures during the retrospective period is unknown. Heightened clinical awareness could increase investigations and diagnosed cases. Due to the previous experience from Sweden, the National Public Health Institute of Finland (predecessor of the National Institute for Health and Welfare) anticipated a rise in NTM infections when the BCG policy changed [25]. However, the decision to change the BCG policy was made precipitately [26], and the medical community was not largely informed to expect an increase of NTM infections. Furthermore, the current estimated incidence corresponds well with the incidence reported from other countries [2, 10, 20]. Therefore, in our opinion, heightened awareness explaining the radical increase is unlikely.

Suggested sources for NTM reservoirs are broad [27]. NTM species have widely been found in Finnish environment since the 1990s and early 2000s [28, 29], and we are not aware of any significant changes that may have increased occurrence of mycobacteria in the environment since. Corresponding to previous studies most cases were caused by M. avium. Interestingly, in our data, Mycobacterium lentiflavum was the second most common isolate (n = 7 [7%]), which first appeared in 2008. A study from Madrid noted that, around the same time, M. lentiflavum isolates increased [30]. In other studies, M. lentiflavum isolates have been scarce [4]. Swimming pools and potable water have been suggested as a possible source for M. lentiflavum infections [30, 31]. However, M. lentiflavum was common in Finnish drinking water systems even before the BCG policy changed [29]. Although the exact place of transmission for our cases is not known, the M. lentiflavum isolates came from 5 different healthcare districts without any specific region standing out. It is uncertain whether, regardless of BCG coverage, M. lentiflavum is emerging or the concurrence is coincidental. We are not aware of any data suggesting an increase in the pathogenicity of M. avium that represented a majority of the increase. In fact, simultaneous studies from other European countries show that M. avium isolates in children have not increased, indicating that pathogenicity has not changed [20, 32].

Many factors affecting host susceptibility are recognized in adults with NTM infection [24]. However, previous studies have found that most children with NTM infection are otherwise healthy and immunocompetent [2]. In Finland, primary immunodeficiency diseases and Mendelian susceptibility to mycobacterial diseases are rare. Although immunosuppressive medications have become more available, in children aged <5 years, they remain rare. Human immunodeficiency virus (HIV) prevalence in children is very low in Finland, and only 1 native-born pediatric case has been reported to the NIDR since 1995. Therefore, significant changes between cohorts that would affect host susceptibility before the age of 5, other than BCG, have not occurred.

Our retrospective study holds some limitations. Information on medication, HIV, and vaccination status was not available. The BCG coverage of infants, however, has decreased considerably and coincides with the clear increase in childhood NTM infections excellently. Because the date when symptoms started was not available, we could not examine the seasonal distribution of infections. We accepted all cases based on an isolation from any site and acknowledge that an isolation of NTM may also indicate colonization. However, most of the culture samples were from needle aspiration, tissue biopsy, or purulent or other discharge (n = 89 [92%]) that would likely represent relevant specimen and disease, especially in children <5 years of age. Furthermore, a study in Denmark found that positive NTM cultures among children represented definite disease in 95% of the cases [20]. Because NTM cultures have low yield (~50%), all NTM cases are not captured in the registry and actual disease burden might be higher. Also, a novel blood-based interferon-γ release assay method for diagnosing childhood NTM lymphadenitis has recently been introduced in Finland, and cultures may be considered unnecessary [33]. This may have decreased the number of cases in the selective-BCG cohorts only.

The burden of NTM disease in high-TB-incidence countries, where environmental exposure to NTM, BCG coverage, and HIV prevalence are likely higher, is largely unknown [1]. Evidence suggests that prior NTM exposure affects the efficacy of BCG vaccine against TB and tuberculin skin test as a diagnostic test for TB [2, 34, 35]. Furthermore, prior NTM exposure may provide some protection against TB [35]. BCG has also shown nonspecific or heterologous effects influencing the immune response to other vaccines and reduced all-cause mortality in resource-limited countries [8, 9]. Thus, understanding the interactions between the immune system, BCG, NTM, and TB will advance the development of new vaccines and diagnostic tests in the future. As the non-BCG-vaccinated cohorts mature, it is possible to examine whether NTM infections in adolescents and adults in Finland will increase. As the efficacy of BCG against tuberculosis is highest in young children, who are also the main risk group for NTM disease, a remarkable increase in older age groups is unlikely.

In Finland, during universal vaccinations with Statens Serum Institut vaccine, the incidence of lymph node abscesses caused by BCG from 2002 to 2004 was 150 per 100000 vaccinated [36], far higher than the estimated incidence of childhood NTM infections currently. Furthermore, serious complications, such as osteitis, arthritis, and generalized BCG, were observed in 14 per 100000 vaccinated [36]. Though other BCG strains may be less reactogenic, BCG remains a live vaccine, which may cause serious adverse effects. Considering these and the self-limiting nature of NTM lymphadenitis, it is our opinion that utilizing BCG merely for NTM prevention is not advisable.

Our study presents a large retrospective population-based cohort with complete capture of culture-confirmed childhood NTM infections nationwide over a 21-year period, including a major change in BCG coverage. We observed a drastic increase in childhood NTM infections after infant BCG coverage decreased, which supports the hypothesis that BCG offers protection against childhood NTM disease. This observation adds to the understanding of NTM epidemiology and might explain why the disease is emerging in some developed countries.

Notes

Funding. This work was supported by a grant from the Väinö and Laina Kivi Foundation to AK.

Potential conflicts of interest. All authors: No reported conflicts of interest. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.