Molecular Evidence for the Aerial Route of Infection of Mycobacterium leprae and the Role of Asymptomatic Carriers in the Persistence of Leprosy

Background

Leprosy persists as a public health problem. The chain of transmission and mechanism of infection are not completely understood. In the current study, we investigated the route of infection and of disease onset, from airway exposure, colonization, and bloodstream dissemination.

Methods

Mycobacterium leprae DNA was detected through quantitative polymerase chain reaction in nasal vestibule, nasal turbinate mucosa, and peripheral blood samples, along with anti–phenolic glycolipid I serology and skin tests from the same individual, from 113 leprosy patients and 104 household contacts of patients (HHCs). Bivariate statistics and multiple correspondence analysis were employed.

Results

The rates of DNA positivity among patients were 66.4% (75 of 113) for nasal swab samples, 71.7% (81 of 113) for nasal turbinate biopsy samples, 19.5% (22 of 113) for blood samples, with seropositivity of 62.8% (71 of 113 samples) and with increasing incidences toward the multibacillary pole of the clinical spectrum. Positivity among HHCs were as follows: 49% (51 of 104) for nasal swab samples, 53.8% (56 of 104) for nasal biopsy samples, 6.7% (7 of 104) for blood samples, and 18.3% (19 of 104 samples) for anti–phenolic glycolipid I serology. During the follow-up of 5–7 years, out of 104 HHCs, 7 developed leprosy (6.7%). Risk for the disease outcome was estimated by comparing results in HHCs who develop leprosy with those not affected. Neither nasal passage nor mucosa positivity was determinant of later disease onset; however, blood presence increased the risk for disease development (relative risk/positive likelihood ratio, 5.54; 95% confidence interval, 1.30–23.62), as did seropositivity (positive likelihood ratio, 3.69 [1.67–8.16]; relative risk, 5.97 [1.45–24.5]).

Conclusions

Our findings strongly suggest that the aerosol route of infection and transmission is predominant and that HHCs contribute to the infection risk to themselves and probably to others.

Leprosy, Epidemiology, Mycobacterium leprae, Quantitative Real-Time PCR, phenolic glycolipid I

(See the Editorial Commentary by Frade and Foss on pages 1421–2.)

Leprosy is one of the oldest infectious diseases to affect humans and continues to be a public health problem, particularly in Brazil [1]. Persons with untreated leprosy are generally considered the main source of transmission. However, this contrast with >30 years of an effective and globally available treatment and the occurrence of new cases among persons with no previous contact with those with untreated leprosy, indicating that there must be other undisclosed sources of infection [2–6].

Several studies have demonstrated widespread dissemination of bacilli in endemic regions and showed that contact with a person with leprosy was associated with increased risk for disease onset [2, 5–7]. On the other hand, most exposed individuals do not manifest the disease, which supports the hypothesis of asymptomatic infection and its potential link to the chain of transmission of Mycobacterium leprae [2, 3, 5, 7]. Leprosy bacilli cannot be cultivated in vitro and have an extremely long generation time (about 13 days). Experimental investigations are further limited because bacilli are not detected by available tests in some leprosy cases [8] and the armadillo is the only natural host and the best animal model [9].

To investigate the aerosol route of leprosy infection and transmission, we proposed in the current study to identify the presence of M. leprae DNA in nasal swab samples from nasal vestibules, nasal turbinate biopsy samples, and peripheral blood samples from the same individual, collected from persons with untreated leprosy and their household contacts (HHCs). The participants’ humoral and cell-mediated immune responses to M. leprae were also investigated.

MATERIAL AND METHODS

Subjects

During the period from 2003 to 2015, samples were collected from 113 patients with untreated leprosy and 104 HHCs. We included in the study only individuals who had a complete set of clinical and laboratory parameters and were followed up for a minimum of 5–7 years. All leprosy cases were diagnosed by leprosy experts and categorized according Ridley-Jopling clinical classification (as tuberculoid [TT], borderline-tuberculoid [BT], midborderline [BB], borderline-lepromatous [BL], or lepromatous leprosy [LL]). For treatment purposes, patients with untreated leprosy were categorized into paucibacillary (PB) or multibacillary (MB) operational classifications. The BT subclassification into PB or MB is described elsewhere [10].

HHCs were defined as persons who live or have lived in the same dwelling as a patient with leprosy in the last 5 years before the leprosy diagnosis. All HHCs were initially examined by leprosy experts for symptoms of the disease. Results for HHCs corresponded to the time point of diagnosis for the corresponding index cases. The HHCs enrolled in the study were not matched to the patient group, and each had a separate index case patient. For data analysis, HHCs were stratified according to the clinical (TT, BT, BB, BL, LL) and operational (PB, MB) classifications of their corresponding index case patients. For follow-up analysis, HHCs were further stratified into categories of affected (HHCs in whom leprosy developed) and healthy (HHCs without any sign or symptom of disease). Among the HHCs affected by leprosy, we excluded those who presented signs of leprosy at first examination (coprevalence). Subjects with incomplete data, inconclusive results in any of the tests, or comorbid conditions or treatments that could interfere with results and subjects who decided not to participate were excluded from this study.

The research protocol was approved by the institutional review board at the Federal University of Uberlandia (#099/2003) and was conducted in accordance with the guidelines of the Declaration of Helsinki. All participants agreed to take part in this study with written informed consent. Patients and HHCs who developed leprosy received free healthcare and adequate treatment for their illness.

Methods

The presence of M. leprae DNA was evaluated in samples collected from the nose vestibule; the nose mucosa, and the peripheral blood. The collection of samples and DNA extraction were described in the Supplementary Methods. We also used enzyme-linked immunosorbent assay (ELISA) to identify the presence of serum antibodies against M. leprae–native phenolic glycolipid I (PGL-I). The specific delayed hypersensitive response to M. leprae antigens was also evaluated with the intradermal Mitsuda test, also known as the lepromin skin test.

To detect M. leprae DNA, we used a real-time quantitative polymerase chain reaction (qPCR) primer/probe assay, described elsewhere [11], to target the M. leprae species-specific genomic element of dispersed repeats (RLEP) (ABI 7300 real-time PCR system; Applied Biosystems). We also used the same qPCR assay to evaluate skin biopsy and slit-skin smear samples from the patients with untreated leprosy. We determined the sensitivity and lower limit of detection for 100 bacilli by analyzing 2-fold serial dilutions of mouse footpad–derived M. leprae with a starting quantity of 1.0 × 10⁷ bacilli per qPCR reaction, from which DNA was extracted and amplified according to the same qPCR protocol. A 10-fold standardized dilution of M. leprae DNA was used in each assay to establish the standard curve. We used the qPCR instrument software to quantify the samples, using interpolation from the standard curve to yield the number of bacilli per reaction.

To assess the immunological status of the participants, we performed indirect ELISA in serum samples to detect circulating immunoglobulin M antibodies against the M. leprae–derived PGL-I antigen (obtained from BEI Resources), as described elsewhere [12]. To moderate intra-plate and inter-plate variations between assays, we converted absorbance results measured at 492 nm into an ELISA index (EI), in which the optical density (OD) of the sample was divided by the OD of the cutoff (EI = OD_sample/OD_cutoff). The positivity threshold was 1.1.

To evaluate the cell-mediated hypersensitive response, we conducted the Mitsuda test, which involves intradermal injection of a heat-killed M. leprae suspension (obtained from Institute Lauro de Souza Lima, Brazil). After 21–28 days, we measured the transverse diameter in millimeters of induration (not erythema) at the injection site, with positive results defined as measurements of ≥7 mm.

Statistical Analysis

Because the data set did not present a Gaussian distribution, we applied nonparametric statistical methods, determining the pairwise bivariate association, linear dependence, correlation coefficient (r), agreement coefficient (K), and congruence. To eludicate the complex relationship between all results at the same time, we used the exploratory multivariate technique of multiple correspondence analysis (MCA). Briefly, this technique allows analysis of the relations between variables and between different categories/levels of each variable, offering at the same time, in comparison with other methods, statistical results that can be seen both analytically and visually [13, 14]. When the data matrix (table of all possible pairwise tabulations of the categorical data) is represented in a graphic display, the distance between points (variables, individuals, and results) represents the association between those variables among the observed data (ie, points closer together in space are more likely to have similar distributions).

For the analysis of follow-up data, the test performances were evaluated and the probabilities for later leprosy onset were estimated through the positive likelihood ratio (LR+), negative likelihood ratio, relative risk (RR), and associated 95% CI. Statistical analyses were performed using the following software: GraphPad Prism (version 5; GraphPad Software), XLStat (version 2015.6.01.24996; Addinsoft), and BioEstat (version 5.3; Mamiraua Institute for Sustainable Development). The threshold for statistical significance was set at an α level of 5%.

RESULTS

The qPCR positivity rates for the detection of M. leprae DNA among patients was 66.4% (75 of 113) for nasal swab samples, 71.7% (81 of 113) for nasal turbinate biopsy samples, and 19.5% (22 of 113) for peripheral blood samples. The mean (standard deviation) quantities of M. leprae detected in the different patient samples, given as number of bacilli per reaction, were 1.8 × 10⁷ (7.4 × 10⁷) for nasal swab, 4.1 × 10⁷ (9.0 × 10⁷) for nasal turbinate biopsy, and 9.3 × 10³ (1.6 × 10⁴) for peripheral blood samples (Kruskal–Wallis test, P < .001). The evaluation of the humoral response measured by anti–PGL-I ELISA demonstrated a total seropositivity rate of 62.8% (71 of 113 samples), with 3.3 as mean EI value.

The observed frequencies for all test results increased toward the bacteriologically positive lepromatous side of the leprosy spectrum (Table 1), except for the Mitsuda test results, in which PB patients had 97% of positive reactions ≥7 mm and only 11% of MB patients had positive reactions (z test and Mann–Whitney, P < .001). The comparisons between the PB and MB patients' results showed statistically significant differences for all parameters evaluated (z test and Mann–Whitney, P ≤ .05). The differences between results for the clinical manifestations were also significant (Monte Carlo and Kruskal–Wallis tests, P < .001) (Table 2).

Table 1.

Laboratory Tests Frequency According to Operational Classifications of Patients With Leprosy

Laboratory Test	PB Patients (n = 32)	P Value	MB Patients (n = 81)
Anti–PGL-I ELISA
Positivity, No. (%)a	3 (9.4)	<.001b	68 (83.9)
EI, mean (SD)	0.7 (0.6)	<.001c	4.4 (3.9)
Mitsuda test
Positivity, No. (%)d	31 (96.9)	<.001b	9 (11.1)
Induration, mean (SD), mm	7.6 (2.7)	<.001c	1.2 (2.3)
qPCR results by sample site
Nasal swab
Positivity, No. (%)	14 (43.8)	<.001b	61 (75.3)
Quantity, mean (SD), bacilli/reaction	2.3 × 104 (2.9 × 104)	<.001c	2.2 × 107 (8.2 × 107)
Nasal biopsy
Positivity, No. (%)	15 (46.9)	<.001b	66 (81.5)
Quantity, mean (SD), bacilli/reaction	4.6 × 104 (7.5 × 104)	<.001c	5.0 × 107 (9.8 × 107)
Blood
Positivity, No. (%)	1 (3.1)	.003b	21 (25.9)
Quantity, mean (SD), bacilli/reaction	2.8 × 103 (0)	…	1.6 × 104 (2.1 × 104)
Skin biopsy
Positivity, No. (%)	3 (9.4)	<.001b	60 (74.1)
Quantity, mean (SD), bacilli/reaction	3.6 × 103 (1.1 × 103)	.03c	7.3 × 107 (1.5 × 108)
Skin smear
Positivity, No. (%)	2 (6.2)	<.001b	63 (77.8)
Quantity, mean (SD), bacilli/reaction	2.0 × 103 (2.4 × 102)	.0504c	1.3 × 108 (2.4 × 108)

Abbreviations: EI, ELISA index; ELISA, enzyme-linked immunosorbent assay; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; SD, standard deviation.

^a For anti–PGL-I serology, an EI ≥1.1 was considered a positive result.

^b Comparison between proportions in PB and MB patients (z test).

^c Comparison between means in PB and MB patients (Mann–Whitney test).

^d For the Mitsuda test, a node ≥7 mm was considered a positive result. Patients receiving corticoid or other interfering treatment were excluded from positivity calculations.

Table 2.

Laboratory Test Frequency According to Clinical Classifications of Patients With Leprosy

Laboratory Test	Index Case Patient Classification						P Valuea
	TT (n = 15)	BT (PB) (n = 17)	BT (MB) (n = 28)	BB (n = 16)	BL (n = 19)	LL (n = 18)
Anti–PGL-I ELISA
Positivity, No. (%)b	1 (6.7)	2 (11.8)	14 (50)	16 (100)	19 (100)	18 (100)	<.001
EI, mean (SD)	0.6 (0.5)	0.8 (0.6)	1.97 (2.3)	4.2 (2.9)	4.4 (3.2)	8.2 (4.5)
Mitsuda test
Positivity, No. (%)c	15 (100)	16 (94.1)	9 (47.4)	0 (0)	0 (0)	0 (0)	<.001
Induration, mean (SD), mm	8.5 (2.5)	6.6 (2.8)	3.1 (3.2)	0.5 (1.4)	0 (0)	0 (0)
qPCR results by sample site
Nasal swab
Positivity, No (%)	6 (40)	8 (47.1)	11 (39.3)	13 (81.3)	19 (100)	18 (100)	<.001
Quantity, mean, bacilli/reaction	1.8 × 104	3.3 × 104	2.4 × 104	1.2 × 105	2.6 × 107	4.9 × 107
Nasal biopsy
Positivity, No. (%)	8 (53.3)	7 (41.2)	16 (57.1)	13 (81.3)	19 (100)	18 (100)	<.001
Quantity, mean, bacilli/reaction	8.0 × 103	9.0 × 104	2.5 × 105	1.1 × 107	8.0 × 107	9.2 × 107
Blood
Positivity, No. (%)	1 (6.7)	0 (0)	2 (7.1)	0 (0)	8 (42.1)	11 (61.1)	<.001
Quantity, mean, bacilli/reaction	2.8 × 103	0	2.7 × 103	0	9.5 × 103	2.3 × 104
Skin biopsy
Positivity, No. (%)	2 (13.3)	1 (5.9)	11 (39.3)	14 (87.5)	17 (89.5)	18 (100)	<.001
Quantity, mean, bacilli/reaction	3.8 × 103	3.3 × 103	1.7 × 104	1.7 × 107	4.1 × 107	1.9 × 108
Skin smear
Positivity, No. (%)	1 (6.7)	1 (5.9)	10 (35.7)	16 (100)	19 (100)	18 (100)	<.001
Quantity mean, bacilli/reaction	1.9 × 103	2.2 × 103	8.3 × 103	1.1 × 107	1.5 × 108	2.9 × 108

Abbreviations: BB, midborderline; BL, borderline-lepromatous; BT, borderline-tuberculoid; EI, ELISA index; ELISA, enzyme-linked immunosorbent assay; LL, lepromatous leprosy; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; SD, standard deviation; TT, tuberculoid.

^aP values for comparisons of both proportions (Monte Carlo test) and means (Kruskal–Wallis test) by clinical patient group.

^b For anti–PGL-I serology, an EI ≥1.1 was considered a positive result.

^c For the Mitsuda test, a node ≥7 mm was considered a positive result. Patients receiving corticoid or other interfering treatment were excluded from positivity calculations.

The bivariate exploration of patients' data revealed an intricate contingency relationship between all pairs of variables (Table 3). The Fisher exact test (2 tailed) demonstrated that for each pairwise comparison of dichotomous data (positive or negative), there were substantial associations between every 2 variables examined (all ≤ .003 for comparison of qPCR of blood samples and anti–PGL-I ELISA results). The linear statistical dependence between pairs of observations, expressed by the Spearman r coefficient (Table 3), also showed significant correlations between every pair of continuous data (qPCR mean quantities of bacilli and anti–PGL-I EIs), except for the comparison of qPCR results for blood and nasal swab samples (all P ≤ .02 for qPCR of nasal swab samples vs anti–PGL-I ELISA results). To assess the agreement between tests, the kappa test κ coefficients were obtained (Table 3). The odds ratio (OR) output from logistic regressions also indicated a high probability of congruence between paired tests (Table 3).

Table 3.

Statistical Coefficients Obtained From Pairwise Analysis of Tests Results in Patients With Leprosy^a

Coefficients by Continuous Variable	Anti–PGL-I ELISA	qPCR by Sample Site
		Nasal Swab	Nasal Biopsy	Blood	Skin Biopsy
qPCR of nasal swab sample
Spearman r	0.23	…	…	…	…
Kappa κ	0.42	…	…	…	…
OR (95% CI)	6.6 (2.8–15.5)	…	…	…	…
qPCR of nasal biopsy sample
Spearman r	0.64	0.32	…	…	…
Kappa κ	0.40	0.63	…	…	…
OR (95% CI)	6.7 (2.8–16.3)	24.9 (8.5–73.3)	…	…	…
qPCR of blood sample
Spearman r	0.37	0.10	0.45	…	…
Kappa κ	0.19	0.19	0.17	…	…
OR (95% CI)	7.8 (2.0–31.1)	14.4 (2.6–79.3)	Positive infinity	…	…
qPCR of skin biopsy sample
Spearman r	0.69	0.27	0.67	0.45	…
Kappa κ	0.60	0.30	0.33	0.29	…
OR (95% CI)	18.7 (7.1–49.1)	3.8 (1.7–8.7)	5.1 (2.1–12.3)	24.5 (4.5–135)	…
qPCR of skin smear sample
Spearman r	0.76	0.26	0.74	0.46	0.85
Kappa κ	0.63	0.48	0.38	0.24	0.75
OR (95% CI)	22.3 (8.3–59.7)	9.5 (3.9–23.2)	7.1 (2.9–16.3)	10.2 (2.6–40.4)	49.9 (16.7–141)

Abbreviations: CI, confidence interval; ELISA, enzyme-linked immunosorbent assay; OR, odds ratio; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction.

^a The Spearman correlation coefficient (r) was used to determine the statistical linear dependence between pairs of continuous observations. The kappa coefficient (κ) was used to assess the agreement between tests. To measure the probability of congruence between paired tests, the OR and 95% CI were estimated through logistic regression analysis.

To further explore the complex interaction among the tests results and the classification of the patients, we used the multivariate statistical method of MCA; results are presented graphically in Figure 1. The MCA intuitive visual interpretation denoted the patient profile for all tests at the same time. The main clusters observed were points representing negative results for qPCR of skin biopsy and skin smear samples and anti–PGL-I ELISA, close to the PB and TT patient classifications; points representing positive results for qPCR of skin biopsy, skin smear, and blood samples and anti–PGL-I ELISA, close to the MB pole and its clinical manifestations BB, BL, and LL; and grouping of positivity or negativity for qPCR results in both nasal swab and nasal turbinate biopsy samples (Figure 1).

Figure 1.

Graphic display of the multivariate exploration of data from patients with leprosy through multiple correspondence analysis. Abbreviations: BB, midborderline; BL, borderline-lepromatous; BT, borderline-tuberculoid; ELISA, enzyme-linked immunosorbent assay; LL, lepromatous leprosy; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; TT, tuberculoid.

For the HHCs, the overall qPCR positivity rates for the different samples were 49% (51 of 104) for nasal swab, 53.8% (56 of 104) for nasal turbinate biopsy, and 6.7% (7 of 104) for blood samples. The mean (standard deviation) quantities of M. leprae, given as number of bacilli per reaction, were 1.8 × 10⁴ (2.2 × 10⁴) for nasal swab, 1.3 × 10⁵ (1.9 × 10⁵) for nasal turbinate biopsy, and 1.4 × 10⁴ (9.4 × 10³) for peripheral blood samples (Kruskal–Wallis, P = .002).

With anti–PGL-I serology, the overall positivity rate for the HHCs was 18.3% (19 of 104), with 0.7 the mean EI. Serum antibodies against PGL-I were positive for 21% (17 of 81) of the HHCs with MB index case patients (Table 3) but only 4.5% (2 of 23) of those with a PB index case patient; this difference was not significant, however (z test, P = .20). The mean EIs for the HHCs of PB patients differed significantly from those for the HHCs of MB patients (Mann–Whitney, P = .004) (Table 4).

Table 4.

Laboratory Test Frequency According to Operational Classifications of Index Case Patients for Household Contacts (n = 104)

Laboratory Test	PB Index Case Patients (n = 23)	P Value	MB Index Case Patients (n = 81)
Anti–PGL-I ELISA
Positivity, No. (%)a	2 (8.7)	.20b	17 (21)
EI, mean (SD)	0.5 (0.5)	.004c	0.8 (0.7)
Mitsuda test
Positivity, No. (%)d	18 (78.3)	.18b	50 (61.7)
Induration, mean (SD), mm	7.7 (3.1)	.80c	8.2 (4.0)
qPCR results by sample site
Nasal swab
Positivity, No. (%)	9 (39.1)	.39b	42 (51.8)
Quantity, mean (SD), bacilli/reaction	9.3 × 103 (1.3 × 104)	.19c	2.0 × 104 (3.0 × 104)
Nasal biopsy
Positivity, No. (%)	13 (56.5)	.96b	43 (53.0)
Quantity, mean (SD), bacilli/reaction	6.4 × 104 (1.4 × 105)	.42c	1.5 × 105 (4.6 × 105)
Blood
Positivity, No. (%)	2 (8.7)	>.99b	5 (6.2)
Quantity, mean (SD), bacilli/reaction	1.2 × 104 (1.3 × 104)	.46c	1.4 × 104 (1.7 × 104)

Abbreviations: EI, ELISA, index; ELISA, enzyme-linked immunosorbent assay; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; SD, standard deviation.

^a For anti–PGL-I serology, an EI ≥1.1 was considered a positive result.

^b Comparison between proportions in PB and MB patients (z test).

^c Comparison between means in PB and MB patients (Mann–Whitney test).

^d For the Mitsuda test, a node ≥7 mm was considered a positive result. Patients receiving corticoid or other interfering treatment were excluded from positivity calculations.

Interestingly, logistic regression analysis of the data for HHCs yielded significant congruency of results when comparing qPCR results for nasal swab and turbinate biopsy samples (OR, 2.3; 95% CI, 1.0–5.1). The other noteworthy output from the statistical analysis of HHC results was for the comparison between qPCR results for nasal turbinate samples and anti–PGL-I ELISA serology (Fisher exact test, P = .046; OR, 4.2; 95% CI, 1.2–14.8).

In the multivariate MCA exploration of the HHC data, we observed cluster associations between positive results for both qPCR of blood samples and anti–PGL-I serology; positive results for qPCR of both nasal swab and nasal turbinate biopsy samples; negative results for both qPCR of blood samples and anti–PGL-I serology; and negative results for qPCR of both nasal swab and nasal turbinate biopsy samples (Figure 2).

Figure 2.

Graphic display of the multivariate exploration of data from household contacts of patients with leprosy through multiple correspondence analysis. Clinical and operational classifications are those of the index case patient corresponding to the household contact. Abbreviations: BB, midborderline; BL, borderline-lepromatous; BT, borderline-tuberculoid; ELISA, enzyme-linked immunosorbent assay; LL, lepromatous leprosy; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; TT, tuberculoid.

Comparison between the mean qPCR values obtained for patients and for HHCs revealed that for both sites in the nose (surface and mucosa), the means differed significantly degree from each other (Mann–Whitney, P < .001). The anti–PGL-I mean EI values observed for patients and HHCs also differed significantly (Mann–Whitney, P < .001), but the mean values did not (P = .82).

After the initial evaluation, the group of HHCs was monitored. Of the 104 HHCs, 7 (6.7%) developed leprosy during a follow-up of 5–7 years. The probability of disease outcome was estimated, comparing the results in HHCs who were affected by disease with the results in those without clinical manifestations during follow-up (Table 6). An evaluation that takes into account the risk assessment for both likelihood (LR+) and RR ratios showed a clear probability of developing disease when results were positive for both qPCR in blood samples (LR+ and RR, 5.54; 95% CI, 1.30–23.62) and anti–PGL-I titers (LR+, 3.69 [95% CI, 1.67–8.16]; RR, 5.97 [1.45–24.5]) (Table 6). (A table with the combined qPCR and ELISA results was added as Supplementary Table.)

Table 5.

Laboratory Test Frequency According to Clinical Classifications of Index Case Patients for Household Contacts (n = 104)

Laboratory Test	Index Case Patient Classification						P Valuea
	TT (n = 3)	BT (PB) (n = 20)	BT (MB) (n = 20)	BB (n = 16)	BL (n = 19)	LL (n = 26)
Anti–PGL-I ELISA
Positivity, No. (%)b	0 (0)	2 (10)	3 (15)	3 (18.7)	2 (10.5)	9 (34.6)	.22
EI, mean (SD)	0.4 (0.2)	0.5 (0.6)	0.7 (0.5)	0.75 (0.8)	0.6 (0.4)	1.0 (0.8)	.046
Mitsuda test
Positivity, No. (%)c	3 (100)	15 (75)	10 (50)	12 (75)	14 (73.7)	14 (53.8)	.20
Induration, mean (SD), mm	11 (3.6)	7.3 (2.8)	7.5 (3.5)	7.9 (3.2)	9.6 (5.0)	7.9 (4.0)	.31
qPCR results by sample site
Nasal swab
Positivity, No. (%)	1 (33.3)	8 (40)	8 (40)	7 (43.7)	9 (47.4)	18 (69.2)	.30
Quantity, mean, bacilli/reaction	1.5 × 104	2.0 × 103	3.1 × 103	1.5 × 103	5.8 × 103	2.5 × 104	.02
Nasal biopsy
Positivity, No. (%)	2 (66.7)	11 (55)	11 (55)	6 (37.5)	12 (63.2)	14 (53.8)	.77
Quantity, mean, bacilli/reaction	1.8 × 105	1.5 × 104	1.2 × 104	1.8 × 105	8.2 × 104	6.1 × 104	.50
Blood
Positivity, No. (%)	0 (0)	2 (10)	2 (10)	2 (12.5)	0 (0)	1 (3.8)	.64
Quantity, mean, bacilli/reaction	0	1.2 × 103	3.8 × 102	3.9 × 103	0	2.5 × 103	.61

Abbreviations: BB, midborderline; BL, borderline-lepromatous; BT, borderline-tuberculoid; EI, ELISA, index; ELISA, enzyme-linked immunosorbent assay; LL, lepromatous leprosy; MB, multibacillary; PB, paucibacillary; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; SD, standard deviation; TT, tuberculoid.

^a Comparisons of proportions (Monte Carlo test) and means (Kruskal–Wallis test).

^b For anti–PGL-I serology, an EI ≥1.1 was considered a positive result.

^c For the Mitsuda test, a node ≥7 mm was considered a positive result. Patients receiving corticoid or other interfering treatment were excluded from positivity calculations.

Table 6.

Results for Quantitative Polymerase Chain Reaction, anti–PGL-I ELISA, and Mitsuda Tests Based on Positivity at the Time of Diagnosis of the Index Case, Observed in Healthy Household Contacts (HHCs) (n = 97) and HHCs Affected by Leprosy (n = 7)

Test and Follow-up Outcome	Results, No.		Diagnostic Parameter	LR+ (95% CI)	RR (95% CI)
	Positive	Negative
qPCR of nasal swab sample
Affected	6	1	Sensitivity: 85.7%	1.85 (1.28–2.68)	6.24 (0.78–50.03)
Healthy	45	52	Specificity: 53.6%
qPCR of nasal biopsy sample
Affected	4	3	Sensitivity: 57.1%	1.07 (0.55–2.08)	1.14 (0.27–4.86)
Healthy	52	45	Specificity: 46.4%
qPCR of blood sample
Affected	2	5	Sensitivity: 28.6%	5.54 (1.30–23.62)	5.54 (1.30–23.62)
Healthy	5	92	Specificity: 94.9%
Anti–PGL-I ELISAa
Affected	4	3	Sensitivity: 57.1%	3.69 (1.67–8.16)	5.97 (1.45–24.5)
Healthy	15	82	Specificity: 84.5%
Mitsuda testb
Affected	2	5	Sensitivity: 28.6%	2.24c (1.29–3.88)	0.21 (0.04–1.04)
Healthy	66	31	Specificity: 31.96%

Abbreviations: CI, confidence interval; ELISA, enzyme-linked immunosorbent assay; LR+, positive likelihood ratio; PGL-I, phenolic glycolipid I; qPCR, quantitative polymerase chain reaction; RR, relative risk.

^a For anti–PGL-I serology, an ELISA index ≥1.1 was considered a positive result.

^b For the Mitsuda test, a node ≥7 mm was considered a positive result.

^c Negative likelihood ratio.

DISCUSSION

Previous reports have addressed M. leprae PCR [15], M. leprae DNA in upper airways [2, 16–18], and M. leprae DNA in blood of contacts [3] and blood donors [5]. In the current study, we evaluated all these sites in the same individuals, with a much more sensitive technique of real-time qPCR.

Even though there is no reason to discard the possibility of opportunistic routes by which M. leprae could infect the host, it is apparent that the evolutionary success of the oldest pathogen associated with human disease was not fortuitous, and there must be a preferred route of infection that conferred the necessary fitness to survive as an exclusively human obligate intracellular pathogen. M. leprae has been detected in soil and water and probably contaminates other surfaces within households in endemic settings, but it is not clear how bacilli from environmental sources could infect a person.

Our thorough analysis of the data set and careful characterization of the study participants generated robust findings and provided molecular evidence in support of our theory that the upper respiratory tract, particularly the nose, is the main portal for the entry and exit of M. leprae; exposure leads to infection of nasal mucosa, which elicits immune responses (cell mediated and humoral); if colonization is successful, M. leprae is transported through the bloodstream and disseminated to favorable sites of growth (ie, myelin-producing Schwann cells).

This hypothesis was evidenced by comparisons among nasal swab, nasal biopsy, and peripheral blood samples from patients and HHCs. The bacilli mean for nasal swab and biopsy samples from patients was 2–3 logs higher than that for the HHCs, whereas in peripheral blood samples the patient mean was just slightly higher than that for HHCs (P > .05). It is possible that infected circulating macrophages in the blood are rapidly deposited or cleared from the blood as they migrate to skin and/or nerves. Based on this hypothesis, the exposure to aerosolized M. leprae and its retention in the nasal passage is the first step in the route of infection, which was confirmed by the detection of M. leprae DNA in nasal swab samples from 6 of 7 HHCs in whom clinical symptoms of leprosy were later diagnosed during follow-up.

The respiration captures any particulate matter suspended in the air into the nose passage, and the organism has several mechanisms of defense against microbial invasion. The most plausible transmission mechanism of M. leprae is that it evades airways epithelial clearance [19], which allows mucosal colonization, with subsequent airborne transmission, from person to person, of exhaled and inhaled respiratory droplets bearing M. leprae. The presence of M. leprae DNA in half the nasal swab samples from HHCs reflected the high prevalence of bacilli in the noses of healthy persons. Although this finding was not markedly associated with later infection or its onset, it was indirect evidence of the role of asymptomatic carriers as a neglected dispersal mode for M. leprae.

The second step in the route of infection was characterized by the likewise high prevalence of M. leprae in nasal biopsy samples from asymptomatic individuals, with higher rates of positivity than nasal swab samples, indicating more permanent infection within the mucosa. After establishing infection in macrophages in the nasal cavity, these cells can migrate to other sites via the circulatory or lymphatic system, and the bacilli will rapidly induce innate-immunity and chemokine-associated genes in the very early stages of infection, resulting in neuropathogenesis [20]. The observed associations and increased likelihood of positive results in both nasal swab and turbinate biopsy samples from patients and HHCs indicated that subclinical infection seems to be a far more common outcome in the host-pathogen interaction than generally accepted.

After initial infection, the adaptive immune response directed at sites of infection will determine whether there is localized induction of either a protective T-helper 1 inflammatory cytokine response with spontaneous healing or an ineffective T-helper 2 cytokine response, allowing growth and further spread of the bacilli, resulting in disease development within the broad clinical spectrum of leprosy [21]. Nevertheless, we once again demonstrated the convincing existence of M. leprae asymptomatic carriage and subclinical infection [2, 18, 19], which is of paramount importance in the insidious persistence of transmission.

By the same token, after its invasion of the mucosa, M. leprae probably drains into the vast lymph capillary network, which originates from the nasal turbinate. The drainage to lymph nodes may well induce the production of antibodies against M. leprae. Therefore, it seems that anti–PGL-I production is present during the early stages of M. leprae–human host interaction. Our present results corroborate previous findings demonstrating that anti–PGL-I serology can help determine those with subclinical infection and those who greater risk of disease [2, 3, 7, 22].

The human nose is well vascularized, and we hypothesize that after infection and colonization, M. leprae–infected cells are transported through bloodstream circulation, leading to secondary and more favorable sites for M. leprae establishment. When M. leprae was detected in blood samples, we confirmed that it can successfully enter the circulatory system, substantiating the systemic infection. However, the low prevalence observed indicated that the presence of M. leprae in blood is a transitory event, mostly linked to BL and LL in patients, regarded as systemic manifestations. Nonetheless, all the patients with untreated leprosy and HHCs who developed the disease and had positive qPCR findings in blood samples also had positive nasal biopsy samples, strengthening the argument for our proposed route of infection. Accordingly, our results verified that the presence of M. leprae in peripheral blood determined high risk for the development of leprosy, as previously reported but with detection of a different M. leprae–specific gene fragment [3]. In addition, the only access to a normal nontraumatized nerve endoneurial compartment would be hematogenous, given that the highly protective barrier of the perineurium isolates the Schwann cells.

The overall frequency of positive results among HHCs revealed that M. leprae is widespread among them; that HHCs have a high bacillary burden regardless of the disease manifestation in their index case patient; and that they comprise a recognizable group of individuals who contribute to the infection risk to themselves, and probably to others. The exposure of the airways to aerosolized M. leprae, invasion of host cells, and colonization of the nasal mucosa tissue, followed by bloodstream dissemination of bacilli, is the main route of infection, evidenced by qPCR findings. Our results presented herein highlight the great amount of subclinical infection and asymptomatic carriage, favoring the persistence of leprosy disease.

Notes

Acknowledgements. We are grateful for the commitment of the staff of the National Reference Center for Sanitary Dermatology and Leprosy (CREDESH) in providing outstanding patient care for the individuals involved in this research, as well for the technical support of the staff of the Laboratory of Molecular Pathology and Biotechnology at CREDESH, particularly Thiago Pescador. We appreciated the reviewers' valuable contributions and the careful evaluation of our work.

Author contributions. S. A., L. R. G., and I. M. B. G. conceived and designed the study; S. A. and L. O. F. carried out the laboratory analysis; S. A., L. R. G., and I. M. B. G. interpreted data and prepared the manuscript; and all authors read and approved the final manuscript.

Financial support. This study was supported by the Foundation for Research Support of the State of Minas Gerais, the Brazilian National Council for Scientific and Technological Development, Brazilian Coordination for Improvement of Higher Education Personnel, and the National Fund for Health/Brazilian Ministry of Health.

Potential conflicts of interest. All authors: No potential 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.