Prevalence of Latent Tuberculous Infection in Patients With Nontuberculous Mycobacterial Lung Disease and Colonization: A Prospective Study in an Intermediate Tuberculosis Burden Country.

BACKGROUND: Controlling latent tuberculosis infection (LTBI) is important in eliminating tuberculosis (TB); however, the prevalence of LTBI has rarely been studied in patients with nontuberculous mycobacterial (NTM) lung disease (LD) and colonization (LC).
METHODS: We prospectively recruited subjects with NTM isolated from sputum mycobacterial cultures from December 2011 to June 2019. NTM-LD and NTM-LC were defined according to the American Thoracic Society guidelines. Patients with negative cultures were recruited as controls. Patients with a history of active TB or positive TB cultures were excluded. LTBI was confirmed using a QuantiFERON-TB Gold In-tube test. The prevalence and factors associated with LTBI were analyzed.
RESULTS: A total of 406 participants were enrolled, including 171 in the NTM-LD group, 153 in the NTM-LC group, and 82 in the control group. The prevalence of LTBI was higher in the NTM-LD and NTM-LC groups than in the controls (21.6%, 20.9%, and 6.1%; P = .006). Multivariable analysis showed that old age (adjusted odds ratio [aOR], 1.021, per year increment; P = .042), NTM-LD (aOR, 4.030; P = .005), NTM-LC (aOR, 3.610; P = .011, compared with the controls), and pulmonary cavitary lesions (aOR, 3.393; P = .034) were independently associated with LTBI.
CONCLUSIONS: The prevalence of LTBI was higher in the patients with NTM-LD and NTM-LC than in the controls. Old age, pulmonary cavitation, and NTM isolated from sputum were associated with a higher risk of LTBI.

Tuberculosis (TB) remains an important infectious disease worldwide. Although the incidence of TB has decreased in the past decade [1], there were still around 10 million new cases of active TB and nearly 1.2 million deaths in 2019, and TB is still a major public health problem in some countries [2]. The World Health Organization estimated that 24.8% of the global population has latent Mycobacterium tuberculosis infection [3]. In addition to optimized treatment for active TB disease, interventions for latent tuberculosis infection (LTBI) are also important to control TB [4]. LTBI screening is recommended for individuals with active TB exposure and close contact, and also for high-risk populations including patients on dialysis, those with immunodeficiency, and those using biological agents [3].

Notably, patients with chronic respiratory diseases are not included in the targeted population for LTBI screening, even though there is a significant association between chronic respiratory diseases and the development of active TB [5]. Chronic respiratory diseases such as chronic obstructive pulmonary disease and bronchiectasis are also associated with nontuberculous mycobacterial lung disease (NTM-LD) [6].

The incidence of NTM-LD has increased in recent decades [6, 7]. A >10-fold increased risk of pulmonary TB has been reported in patients with previous NTM disease [7]. Even though NTM-LD may be associated with LTBI, few studies have investigated this issue. Furuuchi et al used the QuantiFERON-TB test to detect LTBI in Japan, a country with a low prevalence of TB, and reported a positive rate of 8% in patients with NTM-LD and 31% in those with nontuberculous mycobacterial lung colonization (NTM-LC) [8]. However, the number of patients in their study was small (N = 112), and there was no control group of patients without NTM isolated from sputum cultures.

To the best of our knowledge, no other study has focused on the association between LTBI and NTM-LD or NTM-LC. This issue is particularly important due to the increasing number of patients with NTM-LD, and to help prevent TB in these patients. Therefore, we conducted this prospective study to investigate the prevalence and characteristics of LTBI in patients with NTM-LD and NTM-LC compared to a control group without NTM in Taiwan, a country with an intermediate TB burden.

METHODS

Participant Enrollment

We prospectively recruited patients who were ≥20 years old with positive NTM sputum cultures from December 2011 to June 2019 at National Taiwan University Hospital (NTUH). The Research Ethics Committee of NTUH approved this study (institutional review board number: 201108022RC, 20142032RINC, and 201705087RINA). The exclusion criteria were patients (1) with human immunodeficiency virus infection; (2) receiving chemotherapy; (3) with a concomitant bacterial pulmonary infection; or (4) with prior active TB disease proven by a positive culture for M tuberculosis or compatible pathology or radiological findings. In addition, cultures with Mycobacterium kansasii, Mycobacterium szulgai, and Mycobacterium marinum were excluded, because these NTM species are associated with false-positive QuantiFERON-TB Gold In-tube (QFT-GIT; Qiagen, Germany) test results. All participants signed informed consent before enrollment into this study.

Participant Group Classification

NTM-LD was diagnosed according to the diagnostic guidelines recommended by the American Thoracic Society (ATS) [6]. In brief, NTM-LD was diagnosed if all of the following criteria were met: (1) 2 or more sputum culture-positive specimens for the same NTM species; (2) chest images (radiography or computed tomography) demonstrating lesions compatible with NTM-LD (ie, fibrocavitary lesions or nodular bronchiectasis); (3) presence of respiratory symptoms; and (4) no obvious alternative diagnosis at that time. Patients who had positive sputum cultures for NTM but did not fulfil the ATS diagnostic criteria for NTM-LD were classified into the NTM-LC group. We recruited a control group from our clinics, all of whom had negative chest radiograph findings or negative sputum mycobacterial cultures. TB and NTM-LD or NTM-LC were excluded during clinical workup. Mycobacterial cultures and the identification of NTM species were performed as in our previous report [9].

Investigation of Pulmonary Radiograph

Pulmonary radiographic images of the participants were examined by pulmonologists and classified into different patterns, including fibrocavitary, nodular bronchiectasis, and cavitation. Although different patterns could be present in 1 pulmonary radiograph, the predominant pattern was used for analysis. Radiographic scores were calculated systematically [9]. The lung field was divided into upper, middle, and lower parts in both lung fields, for a total of 6 areas. The scores were calculated as follows: a normal appearance in an area, 0 points; if the lung infiltration involved less than one-third of the area, 1 point; if one-thirds to two-thirds of the area was involved, 2 points; and if the area was almost totally involved, 3 points. The scores were summed, and the total score ranged from 0 to 18.

Definition of LTBI With the QFT-GIT Test and Exclusion of Active TB

Every participant underwent a QFT-GIT test after enrollment to determine LTBI status. The QFT-GIT tests (Qiagen) were performed according to the manufacturer’s instructions with a 3-tube system including an M tuberculosis antigen tube, positive control tube, and negative control tube. We incubated the participants’ blood in the 3 tubes for 16–24 hours, and then isolated the reaction supernatants and examined the concentration of interferon-gamma (IFN-γ) in the tubes.

If the IFN-γ concentration in the antigen tube minus that in the negative control tube (QFT-GIT response) was ≥0.35 IU/mL, the result was defined as being positive; if the QFT-GIT response was <0.35 IU/mL, the result was defined as being negative. If the QFT-GIT response was <0.35 IU/mL and the concentration of IFN-γ in the positive control tube was <0.5 IU/mL or the concentration of IFN-γ in the negative tube was ≥8.0 IU/mL, the QFT-GIT result was defined as being indeterminate. We defined the participants with a positive QFT-GIT test result in whom active TB disease had been excluded as having LTBI. Participants with negative or indeterminate results were classified into the non-LTBI group.

Statistical Analysis

The prevalence rates of LTBI in the control, NTM-LD, and NTM-LC groups were compared using the χ2 test. Data were analyzed using the Student t test and χ2 test for continuous and categorized variables. The risk factors for LTBI were analyzed using logistic regression analysis. In the multivariable logistic regression analysis of the risk factors for LTBI, all factors with a P value < 0.15 in the χ2 test were included in forward stepwise regression. All statistical analyses were performed with IBM SPSS version 24.0 software. A P value < .05 was considered to be statistically significant.

RESULTS

Participant Demographics

A total of 406 participants were enrolled, of whom 171 were classified into the NTM-LD group, 153 into the NTM-LC group, and 82 into the control group (Figure 1). Control group participants were younger than the NTM-LD and NTM-LC groups (mean age ± standard deviation [SD]: 57.7 ± 16.7 vs 61.8 ± 13.6 vs 65.1 ± 12.9 years, respectively; P < .001). There were fewer males (NTM-LD, 35.7%; control, 53.7%; and NTM-LC, 49.7%; P = .007) and the body mass index was lower in the NTM-LD group compared to the control and NTM-LC groups (mean ± SD: 21.0 ± 3.5, 23.5 ± 3.7, and 22.6 ± 3.8 kg/m2, respectively; P < .001) (Table 1).

Figure 1.

Flowchart of patient enrollment. ∗None of the patients had Mycobacterium szulgai or Mycobacterium marinum isolated in sputum cultures. Abbreviations: ATS, American Thoracic Society; NTM, nontuberculous mycobacteria; NTM-LC, nontuberculous mycobacterial lung colonization; NTM-LD, nontuberculous mycobacterial lung disease.

Table 1.

Demographics According to Different Nontuberculous Mycobacterial Infection Status

Characteristic	Control (n = 82)	NTM-LD (n = 171)	NTM-LC (n = 153)	P Valuea
Age, y, mean ± SD	57.7 ± 16.7	61.8 ± 13.6b	65.1 ± 12.9c,d	.001
Male sex	44 (53.7%)	61 (35.7%)b	76 (49.7%)d	.007
Current smoker	5 (6.1%)	2 (1.2%)	5 (3.3%)	.140
Ex-smoker	11 (13.4%)	21 (12.3%)	27 (17.6%)
BMI	23.5 ± 3.7	21.0 ± 3.5b	22.6 ± 3.8d	<.001
Symptoms
Cough	41 (50%)	121 (70.8%)b	96 (62.7%)	.006
Dyspnea	19 (23.2%)	41 (24.0%)	38 (24.8%)	.958
Hemoptysis	2 (2.4%)	41 (24.0%)b	33 (21.6%)c	<.001
Chest pain	6 (7.3%)	11 (6.4%)	6 (3.9%)	.478
Constitutional symptoms	5 (6.1%)	34 (19.9%)b	19 (12.4%)	.010
Underlying diseases
ESRD	1 (1.2%)	1 (0.6%)	3 (2.0%)	.533
Malignancy	5 (6.1%)	24 (14.0%)	24 (15.7%)c	.102
Cirrhosis of liver	0 (0%)	2 (1.2%)	1 (0.7%)	.589
Diabetes mellitus	9 (11.0%)	10 (5.8%)	21 (13.7%)d	.055
GERD	5 (6.1%)	14 (8.2%)	12 (7.8%)	.836
Sinusitis	3 (3.7%)	6 (3.5%)	3 (2.0%)	.653
COPD	7 (8.5%)	14 (8.2%)	17 (11.1%)	.639
Asthma	12 (14.6%)	16 (9.4%)	19 (12.4%)	.432
Autoimmune disease	3 (3.7%)	14 (8.2%)	14 (9.2%)	.300
Transplant status	0 (0%)	2 (1.2%)	3 (2.0%)	.428
Mycobacteriology
AFS positive	0	73 (42.7%)b	2 (1.4%)d	<.001
NTM culture
No growth	82 (100%)	0 (0%)b	0 (0%)c	<.001
MAC	0	122 (71.3%)b	101 (66.0%)c
M abscessus	0	43 (25.1%)b	36 (23.5%)c
RGM other than M abscessus		3 (1.8%)b	3 (2.0%)c
SGM other than MAC		0 (0%)b	2 (1.3%)c
Mixed		3 (1.8%)b,e	11 (7.2%)c,f
QFT-GIT result
Positive	5 (6.1%)	37 (21.6%)b	32 (20.9%)c	.009
Negative	72 (87.8%)	115 (67.3%)b	109 (71.2%)c
Indeterminate	5 (6.1%)	19 (11.1%)b	12 (7.8%)c
Radiographic pattern
FC	0 (0%)	21 (12.3%)b	5 (3.3%)d	<.001
NB	28 (34.1%)	131 (76.6%)b	108 (70.6%)c	<.001
Cavitation	0 (0%)	12 (7.0%)b	2 (1.3%)d	.003
Radiographic score	1.9 ± 2.1	4.6 ± 2.9b	3.7 ± 2.8c,d	<.001

Data are presented as No. (%) unless otherwise indicated. Nominal variables were analyzed using the χ2 test, including sex, smoking status, symptoms, underlying diseases, mycobacteriology, QFT-GIT result, and radiographic patterns. Numerical variables were analyzed using the Student t test, including age and BMI.

Abbreviations: AFS, acid-fast stain; BMI, body mass index; COPD, chronic obstructive pulmonary disease; ESRD, end-stage renal disease; FC, fibrocavitary; GERD, gastroesophageal reflux disease; MAC, Mycobacterium avium complex; NB, nodular bronchiectasis; NTM, nontuberculous mycobacteria; NTM-LC, nontuberculous mycobacterial lung colonization; NTM-LD, nontuberculous mycobacterial lung disease; QFT-GIT, QuantiFERON-TB Gold In-tube; RGM, rapid-growing mycobacteria; SD, standard deviation; SGM, slow-growing mycobacteria.

Comparison among the 3 groups using 1-way analysis of variance.

Significant difference between the control and NTM-LD groups using the Student t test or χ2 test as appropriate.

Significant difference between the control and NTM-LC groups using the Student t test or χ2 test as appropriate.

Significant difference between the NTM-LD and NTM-LC groups using the Student t test or χ2 test as appropriate.

Mixed NTM species in culture in the NTM-LD group: MAC and M abscessus in 2 participants; MAC and Mycobacterium fortuitum in 1 participant.

Mixed NTM species in culture in the NTM-LC group: MAC and M abscessus in 4 participants; MAC and M fortuitum in 4 participants; MAC, M abscessus, and M fortuitum in 1 participant; MAC and unidentified NTM in 1 participant; unidentified species in 1 participant.

Clinical, Microbiology, and Radiographic Patterns

With regards to the clinical symptoms, there were significantly higher rates of cough, hemoptysis, and constitutional symptoms in the NTM-LD and NTM-LC groups than in the controls; however, there was no significant difference between the NTM-LD and NTM-LC groups. There were no significant differences among the 3 groups in underlying diseases or comorbidities. Acid-fast stain positivity was significantly higher in the NTM-LD group than in the NTM-LC group. In the NTM-LC group, 98.7% of the acid-fast stain results were negative, and only 0.7% reported grade 1+, and 0.7% reported grade 2+. With regards to the mycobacterial cultures, Mycobacterium avium complex (NTM-LD group, 71.3%; NTM-LC group, 66.0%) and Mycobacterium abscessus complex (NTM-LD group, 25.1%; NTM-LC group, 23.5%) were the leading pathogens, accounting for 96.4% of the cultures in the NTM-LD group and 89.5% of the cultures in the NTM-LC group. In the lung patterns on chest radiographs, fibrocavitary disease, presence of cavitation, and increased radiographic scores were significantly higher in the NTM-LD group than in the NTM-LC group. Nodular bronchiectasis was the most common pattern, and there was no significant difference between the NTM-LD and NTM-LC groups.

LTBI Status in the NTM-LD and NTM-LC Groups and the Associated Factors

The prevalence of LTBI was significantly higher in the NTM-LD and NTM-LC groups than in the controls (21.6%, 20.9%, and 6.1%, respectively; P = .006). Age, status of NTM lung infection, and cavitation on pulmonary radiograph were associated with LTBI (Supplementary Table 1). For the QFT-GIT test, 36 cases (8.87% of all participants) had indeterminate results, including 5 cases (6.1%) in the control group, 12 cases (7.8%) in the NTM-LC group, and 19 cases (11.1%) in the NTM-LD group.

Univariable logistic regression analysis for LTBI-associated factors showed that age (odds ratio [OR], 1.023 [95% CI, 1.004–1.043]; P = .017), NTM lung infection status (control group as the reference; NTM-LD group: OR, 4.252 [95% CI, 1.604–11.274], P = .004; and NTM-LC group: OR, 4.073 [95% CI, 1.521–10.905], P = .005), and cavitation pattern (OR, 3.574 [95% CI, 1.201–10.632], P = .022) were significantly associated with LTBI (Table 2). In forward stepwise multivariable logistic regression, age (adjusted odds ratio [aOR], 1.021 [95% CI, 1.001–1.041]; P = .042), NTM lung infection status (control group as the reference; NTM-LD group: aOR, 4.030 [95% CI, 1.514–10.727]; P = .005; NTM-LC group: aOR, 3.610 [95% CI, 1.338–9.737]; P = .011), and cavitation pattern (aOR, 3.393 [95% CI, 1.097–10.499]; P = .034) were independently associated with LTBI.

Table 2.

Univariable and Multivariable Logistic Regression for Factors Associated With Latent Tuberculosis Infection in All Participants

Variables	Univariable Regression		Multivariablea	P Value
	Crude OR (95% CI)	P Value	Adjusted OR (95% CI)
Age (years), per 1 year	1.023 (1.004–1.043)	.017	1.021 (1.001–1.041)	.042
Sex (male)	1.395 (.842–2.312)	.196	…
Autoimmune disease	0.290 (.068–1.244)	.096	…
NTM infection status
Controls	Ref	.012	Ref	.019
NTM-LD	4.252 (1.604–11.274)	.004	4.030 (1.514–10.727)	.005
NTM-LC	4.073 (1.521–10.905)	.005	3.610 (1.338–9.737)	.011
Cavitation	3.574 (1.201–10.632)	.022	3.393 (1.097–10.499)	.034

aDemographic factors with P < .15 between those with and without latent tuberculosis infection (Supplementary Table 1) were included for analysis. Forward stepwise selection was used in multivariable logistic regression.

Abbreviations: CI, confidence interval; NTM, nontuberculous mycobacteria; NTM-LC, nontuberculous mycobacterial lung colonization; NTM-LD, nontuberculous mycobacterial lung disease; OR, odds ratio.

The positive percentage of QFT-GIT tests in different subgroups are shown in Figure 2. Both the NTM-LD and NTM-LC groups had significantly higher positive QFT-GIT test results than the control group. Regarding age and the presence of radiographic cavitation, patients aged >65 years and those with cavitation had a trend of higher QFT-GIT positivity compared with the corresponding counterpart groups. Regarding QFT-GIT values, the NTM-LD and NTM-LC groups had significantly higher QFT-GIT values than the controls (Supplementary Figure 1).

Figure 2.

The association between the risk factors for latent tuberculosis infection and positive percentage of QuantiFERON-TB Gold In-tube tests. ∗P < .05; ∗∗P < .01. Comparisons were performed using χ2 test. Abbreviations: NTM-LC, nontuberculous mycobacterial lung colonization; NTM-LD, nontuberculous mycobacterial lung disease; QFT-GIT, QuantiFERON Gold In-tube.

In the NTM-LD group, there was no difference in LTBI among those with different NTM subspecies (P = .564, χ2 test) (Supplementary Table 2). In contrast, in the NTM-LC group, the incidence of LTBI was higher in those with rapidly growing mycobacteria other than M abscessus, whereas the incidence of LTBI was lower in those with M abscessus and other slowly growing mycobacteria (P = .015, χ2 test) (Supplementary Table 3).

Follow-up of the TB Status of the Cohort

None of the LTBI cases in this study received treatment for LTBI. We followed their status for ≥2 years, during which 6 patients were diagnosed with new active TB, including 3 (2%) in the NTM-LC group, 2 (1.2%) in the NTM-LD group, and 1 (1.2%) in the control group (P = .821, χ2 test). Of these 6 patients, 3 (1.0%) had a negative QFT-GIT status, 2 (2.7%) had a positive status, and 1 (2.8%) had an indeterminate status (P = .445, χ2 test).

DISCUSSION

In this prospective study, the prevalence of LTBI was significantly higher in the patients with NTM-LD and NTM-LC than in the controls. There was no significant difference in the prevalence of LTBI between the NTM-LD and NTM-LC groups. In addition, old age, NTM-LD or NTM-LC status, and the presence of a cavitary pattern were associated with LTBI. The risk of LTBI in the NTM-LD group was not related to any specific species of NTM; however, since most of the enrolled NTM cases were caused by M avium complex and M abscessus, we can only conclude that the prevalence of LTBI was similar between NTM infections caused by M avium complex and M abscessus.

Previous studies of LTBI in high-risk populations in Taiwan have reported a prevalence rate of 25% in patients with end-stage renal disease [10], 20% in kidney transplant recipients [11], 28.2% in patients with diabetes mellitus [12], and 16.8% in residents and workers in long-term care facilities with a mean age of 70 years [13]. In our study, the prevalence of LTBI was around 20% in patients with NTM lung infection, compatible with other LTBI high-risk groups. This may suggest that patients with NTM-LD and NTM-LC are associated with a higher risk of M tuberculosis infection.

Although patients with prior TB disease were excluded in this study, we still found a higher prevalence of LTBI in the NTM-LD and NTM-LC groups. Due to the cross-sectional design of this study, we could not confirm whether M tuberculosis and NTM were concurrent infections or if there was a causal relationship. Nevertheless, this study is the largest to investigate LTBI in patients with NTM-LD and NTM-LC at present, and it provides important evidence for LTBI interventions.

NTM-LC is defined as patients with airway colonization by NTM but no evidence of active infection who do not usually require NTM treatment [6]. In this study, the NTM-LC group had fewer symptoms, lower NTM bacilli load, less pulmonary radiographic extent, and lower proportion of cavitation compared to the NTM-LD group. Although the patients with NTM-LC had a higher incidence of hemoptysis, a more nodular bronchiectasis pattern on imaging, and a higher radiographic score of disease extent than the control group, there were no significant differences in NTM bacilli burden and fibrocavitary pattern. The NTM-LC group were also older and had a higher percentage of malignancy than the control group. Therefore, the higher prevalence of LTBI in the NTM-LC group may be due to preexisting pulmonary structural changes, especially given the nodular bronchiectasis pattern, and underlying host factors (old age and more malignancy; Table 1).

Previous LTBI studies have reported that old age is an important risk factor for LTBI [10, 14]. The risk of LTBI in the elderly may increase with exposure and contact with TB patients as supported by previous studies of healthcare workers and household contacts [14, 15]. In addition, the adverse effect of aging on immune cells such as macrophages, neutrophils, and dendritic cells may cause T cells to dysfunction, consequently making the host more susceptible to M tuberculosis infection [16]. The trend of a decrease in the prevalence of TB in recent decades has also reduced the exposure of younger people to patients with TB [2]. These findings are consistent with our study and may explain the relationship between age and LTBI.

In the present study, lung cavitation was associated with LTBI in multivariable analysis. Although the reason for this association is not clear, it is possible that cavitation is a sign of previous active TB disease that resolved without treatment. The differential diagnosis for pulmonary cavitation includes TB disease (post–primary TB), NTM infection, bacterial pneumonia with abscess and aspergillosis, and noninfectious causes such as cancer, autoimmune diseases, and vascular lesions [17]. The presence of pulmonary cavitary lesions may imply a high risk of M tuberculosis infection in patients with cancer or autoimmune disease [18, 19].

The QFT-GIT test examines IFN-γ release in response to TB-specific peptides according to the white blood cells in the patients’ blood [20]. Immunocompromised status and prior active TB as well as some NTM species (M kansasii, M szulgai, and M marinum) can lead to false-negative and false-positive results, respectively, in the QFT-GIT test. We thus excluded patients with previous TB and the specific NTM species related to lung infection and colonization. Thirty-six cases (8.87% of all participants) had indeterminate QFT-GIT test results in this study, and the incidence increased with the severity of NTM infection status (from the controls [6.1%] to the NTM-LC group [7.8%] to the NTM-LD group [11.1%]). Previous studies have reported associations between NTM-LD and an immunocompromised status [21], poor nutrition, and other poor clinical conditions [22, 23], which are all risk factors for an indeterminate QFT-GIT test result [20, 24].

There were some limitations to this study. First, the patients’ baseline characteristics including age and sex were not exactly matched among the 3 groups. Although we adjusted for these factors in the multivariable regression analysis, selection bias is possible. Second, the pulmonary radiographic patterns were mostly assessed using plain film images as not all patients received chest computed tomography; therefore, the radiographic lesions and scores may be underestimated. Third, patients with prior TB and active TB disease were excluded by clinical history and examinations; however, some patients may still have had spontaneously resolved TB disease without a definite diagnosis or treatment. Fourth, the number of cases is too small to analyze the incidence of active TB disease. Finally, we focused on patients with NTM-LD and NTM-LC and compared them with a control group; however, the prevalence of LTBI in the general population is unknown and requires further investigations.

In conclusions, LTBI was highly prevalent in both the NTM-LD and NTM-LC patients in this study. Old age, pulmonary cavitation, NTM-LD, and NTM-LC were associated with LTBI. Further studies are needed to clarify the association between LTBI and NTM lung infection. Whether screening for LTBI is needed in populations with NTM lung infection to control TB requires further validation.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Click here for additional data file.

Click here for additional data file.