Secondary Logo

Journal Logo

Brief Reports

Detection of Mycobacterium Tuberculosis DNA in Buccal Swab Samples from Children in Lima, Peru

Flores, Juan A. MSc*,†; Calderón, Roger MSc‡,§; Mesman, Annelies W. PhD; Soto, Martin BSc; Coit, Julia MPH; Aliaga, Juan ScD§; Mendoza, Milagros BSc; Leon, Segundo R. MSc*; Konda, Kelika PhD; Mestanza, Francisco M. MD**; Mendoza, Carlos J. MD††; Lecca, Leonid MD; Murray, Megan B. ScD; Holmberg, Rebecca C. PhD‡‡; Pollock, Nira R. PhD§§; Franke, Molly F. SD

Author Information
The Pediatric Infectious Disease Journal: November 2020 - Volume 39 - Issue 11 - p e376-e380
doi: 10.1097/INF.0000000000002828


Conventional tuberculosis (TB) diagnostics rely on a sputum sample, which many children cannot produce. Gastric aspirates are an alternative,1 but the procedure is invasive and often unavailable. Less invasive alternatives such as the enteric string2 or noninvasive sample types such as stool3–5 and urine6 have been widely studied in children, but none surpass the sensitivity of sputum or gastric aspirate nor consistently diagnose TB in children with clinically diagnosed unconfirmed TB.

Recent studies from South Africa and Peru used molecular diagnostics to detect Mycobacterium tuberculosis from oral swab samples from adults with TB, with promising results.7–9 In South Africa, investigators found that M. tuberculosis also could be detected in buccal swabs from children with confirmed TB and from those with clinically diagnosed TB that was not bacteriologically confirmed.10 We evaluated the sensitivity and specificity of M. tuberculosis DNA detection from buccal samples in children.


Study Population

From May 2015 through February 2018, we enrolled children in Lima, Peru, in a study that aimed to identify alternative sample types for pediatric TB diagnosis. Participants were less than 15 years old, had a history of contact with an adult with pulmonary TB and had one or more of the following symptoms consistent with intrathoracic TB, which was persistent and unexplained: cough, weight loss, persistent fever or fatigue/lethargy.11

Participants received the Peru National Tuberculosis Program standard of care for TB.12 Children provided 2 respiratory specimens for culture and smear microscopy and underwent a chest radiograph, tuberculin skin test and a physical examination. Pediatric pulmonologists used this information to diagnose or rule out TB, consistent with standardized case definitions for intrathoracic TB.13 Children with unconfirmed TB lacked bacteriologic confirmation but presented with one or more TB symptoms, had a chest radiograph consistent with TB and were diagnosed with TB by a pediatric pulmonologist. If a child could not spontaneously expectorate sputum, gastric aspiration or, less commonly, sputum induction was performed. Passive retrospective follow-up of all participants was conducted at 6 months based on medical chart review. Due to the low HIV prevalence among children in Peru (≈0.02%),14 we did not prospectively test for HIV; however, HIV status was reported by caregivers.

This analysis includes all children with TB (bacteriologically confirmed or clinically diagnosed unconfirmed cases) and a subset of children in whom TB was ruled out (ie, controls) that could be most closely matched to cases by age and sample collection date.

Buccal Sample Collection

We collected a single buccal sample from each child using 1 of 2 collection devices (based on availability). The OmniSwab (Whatman, Maidstone, UK, catalog no. WB100035) consisted of a sterile swab head that was ejected into sterile lysis buffer.8 The EasiCollect device (Whatman, catalog no. WHAWB120210) consisted of a sponge-like swab that folded to contact and transfer the sample onto an Flinders Technology Associates (FTA) card. Trained study staff collected buccal samples by gently and firmly swabbing the inside of each of the child’s cheeks (right and left) for 10 seconds. OmniSwab samples in lysis buffer were vortexed, and the swab was removed before freezing at −80C. FTA cards were stored dry at −20C until processing. For cases, sample collection occurred before initiation of TB treatment. Criteria for sample exclusion included incorrect labeling or transportation conditions that did not meet protocol requirements (ie, outside of 2°C–8°C for OmniSwab or not room temperature for EasiCollect).

Laboratory Procedures

Sputum and neutralized gastric aspirates underwent smear microscopy and culture by BACTEC MGIT 960 System (BD Franklin Lakes, NJ). Positive cultures were confirmed to belong to M. tuberculosis complex by the BD MGIT TBc Identification Test (Becton and Dickinson). Polymerase chain reaction (PCR) on sputa or gastric aspirates is not performed as part of standard TB diagnostic procedures in Peru and was therefore not used for diagnosis as part of this study.15

We extracted M. tuberculosis DNA from swab samples using the QIAamp DNA Mini Kit (Qiagen, MD; catalog no. 51306).8 Initially, we used a sample input volume of 0.25 mL for OmniSwabs and 3 punches from the FTA card. To maximize sensitivity, we subsequently increased these volumes to 0.5 mL and to the entire sponge contact surface area of the FTA card, respectively. The FTA card was cut into pieces using a sterile scalpel and placed in a tube with 300 µL 0.1 N NaOH, 0.3 mM ethylenediaminetetraacetic acid, pH 13.0 solution and incubated for 5 minutes at room temperature. After which, 300 µL of 0.1 M Tris-HCl, pH 7.0 was added to neutralize, followed by an incubation at 90°C for 20 minutes to release DNA from the card. The liquid was then processed through the QIAamp Mini Spin Column. OmniSwabs in lysis buffer were processed as previously described.8 The final DNA from both collection types was eluted in 50 µL of elution buffer prewarmed to 42°C. M. tuberculosis DNA was amplified in duplicate using IS6110 insertion element primers by real-time PCR (Instrument Roche 480 Lightcycler; Roche, Basel, Switzerland). The real-time PCR assay was validated in sputum and has been used for detection of M. tuberculosis from oral swabs.8–10,16,17 Real-time PCR positivity was established using a predetermined assay cut point of <37 for the cycle threshold and a fluorescence of ≥6, the latter which was based on our prior work in stool.5 Laboratory staff were blinded to case status.

Statistical Analysis

Sensitivity and specificity, and corresponding exact confidence intervals, were calculated among cases and controls, respectively. We estimated sensitivity overall and stratified by collection type, age group, sex, tuberculin skin test result, smear result and presence of cavitary disease on chest radiograph. To examine factors associated with M. tuberculosis detection from the buccal samples, we conducted univariable binomial regression analyses. The small number of children with culture-confirmed TB precluded a comprehensive multivariable analysis; however, we adjusted analyses of collection device (OmniSwab vs. EasiCollect) by smear status because it happened that a higher percentage of children with an OmniSwab had a positive sputum smear (71% vs. 29%). Statistical analyses were conducted in Stata, version 14.2 (StataCorp, College Station, TX).

Ethical Considerations

Ethics Committees of the Peru National Institute of Health and Harvard Medical School approved the study. Caregivers provided written informed consent, and children 8 years old and older provided written informed assent. The Ethics Committee of the Universidad Peruana Cayetano Heredia del Peru approved analyses.


Study Population

The analysis included 89 children with intrathoracic/pulmonary TB (24 with culture-confirmed TB, 65 with clinically diagnosed unconfirmed TB) and 199 controls (Figure, Supplemental Digital Content 1, Among all children, 43%, 36% and 21% were 0–4, 5–9 and 10–14 years, respectively. Fourteen samples (5 from confirmed TB cases, 2 from clinically diagnosed unconfirmed cases and 7 from controls) had discrepant replicates and were considered negative. Of 24 children with confirmed TB, 29% (7/24) had a positive sputum smear (Table, Supplemental Digital Content 2, EasiCollect swabs were collected from 140 children (including 13 confirmed cases and 42 clinically diagnosed unconfirmed cases), and OmniSwabs were collected from 148 children (including 11 confirmed cases and 23 clinically diagnosed unconfirmed cases). No child was reported to be living with HIV infection.

Sensitivity and Specificity Analysis

Sensitivity among all TB cases (confirmed and clinically diagnosed unconfirmed) was 9% (8/89; 95% confidence interval [CI]: 4%–17%) but varied widely based on smear and culture results. Sensitivity was 21% among culture-confirmed cases (5/24; 95% CI: 7%–42%) with higher sensitivities observed among children with smear-positive disease (4/7 = 57%; 95% CI: 19%–90%) versus smear-negative disease (1/17 = 5.9%; 95% CI: 0%–29%, P = 0.005). Among clinically diagnosed unconfirmed TB cases, sensitivity was 4.6% (3/65; 95% CI: 1%–13%), notably lower than for confirmed TB cases (P = 0.018) (Table 1). Specificity was 99% (197/199; 95% CI: 96%–100%). Use of a smaller input volume for early samples did not appear to affect results (Table, Supplemental Digital Content 3,

TABLE 1. - Sensitivity of Mycobacterium tuberculosis DNA Detection From Buccal Samples Using Real-time PCR, Overall and by Select Factors
Confirmed TB (N = 24) OmniSwab Unconfirmed, Clinically Diagnosed TB (N = 65) OmniSwab
Overall EasiCollect Overall EasiCollect
n/N Sensitivity (95% CI) n/N Sensitivity (95% CI) n/N Sensitivity (95% CI) n/N Sensitivity (95% CI) n/N Sensitivity (95% CI) n/N Sensitivity (95% CI)
Overall 5/24 21 (7–42) 1/13 7.7 (0–36) 4/11 36 (11–69) 3/65 4.6 (1–13) 2/42 4.8 (1–16) 1/23 4.3 (0–22)
Age (yr)
 0–4 0/8 0 (0–37) 0/5 0 (0–52) 0/3 0 (0–70) 1/34 2.9 (0–15) 0/20 0 (0–17) 1/14 7.1 (0–33)
 5–9 1/7 14 (0–58) 0/4 0 (0–60) 1/3 33 (1–91) 2/20 10 (1–32) 2/13 15 (2–45) 0/7 0 (0–41)
 10–14 4/9 44 (14–79) 1/4 25 (1–81) 3/5 60.0 (15–95) 0/11 0 (0–28) 0/9 0 (0–34) 0/2 0 (0–84)
 Female 4/12 33 (10–65) 1/5 20 (1–72) 3/7 43 (10–82) 2/39 5.1 (1–17) 1/26 3.8 (0–20) 1/13 7.7 (0–36)
 Male 1/12 8.3 (0–38) 0/8 0 (0–37) 1/4 25 (1–81) 1/26 3.8 (0–20) 1/16 6.2 (0–30) 0/10 0 (0–31)
 Yes 0/1 0 (0–98) 0/1 0 (0–98) 0/0 0/2 0 (0–84) 0/2 0 (0–84) 0/0
 No 5/23 22 (7–44) 1/12 8.3 (0–38) 4/11 36 (11–69) 3/63 4.8 (1–13) 2/40 5 (1–17) 1/23 4.3 (0–22)
Ability to spontaneously expectorate sputum†
 Yes 5/13 38 (14–68) 1/4 25 (1–81) 4/9 44 (14–79) 1/27 3.7 (0–19) 1/17 5.9 (0–29) 0/10 0 (0–31)
 No 0/11 0 (0–28) 0/9 0 (0–34) 0/2 0 (0–84) 2/37 5.4 (1–18) 1/24 4.2 (0–21) 1/13 7.7 (0–36)
Tuberculin skin test (≥10 mm)†
 Positive 0/15 0 (0–22) 0/10 0 (0–31) 0/5 0 (0–52) 3/32 9.3 (2–25) 2/25 8.0 (1–26) 1/7 14 (0–58)
 Negative 1/2 50 (1–99) 0/0 1/2 50 (1–99) 0/29 0 (0–12) 0/15 0 (0–22) 0/14 0 (0–23)
Cavitary disease on chest radiograph
 Yes 3/6 50 (12–88) 1/2 50 (1–99) 2/4 50 (68–93) 0/0 0/0 0/0
 No 2/18 11 (1–35) 0/11 0 (0–28) 2/7 29 (4–71) 3/65 4.6 (1–13) 2/42 4.8 (1–16) 1/23 4.4 (0–22)
Sputum smear result
 Positive 4/7 57 (18–90) 1/2 50 (1–99) 3/5 60 (15–95)
 Negative 1/17 5.9 (0–29) 0/11 0 (0–28) 1/6 17 (0–64)
*In children <5 years old, undernutrition was defined as <−2 standard deviation (SD) in Z score for weight-for-height. In children ≥5 years old, undernutrition was defined as <−2 SD in Z score for body mass index for age, according to World Health Organization growth guidelines.
†Because of missing data, denominators do not sum to the total N.

Factors Associated With Buccal Sample Test Positivity Among Children With Culture-confirmed TB

In unadjusted analyses among children with culture-confirmed TB, a positive sputum smear (prevalence ratio = 9.7; 95% CI: 1.3–72; P = 0.03), cavitary disease on chest radiograph (prevalence ratio = 4.5; 95% CI: 0.97–21; P = 0.05) and ability to spontaneously expectorate sputum (5/13 M. tuberculosis DNA detection for children who could produce sputum vs. 0/11 for children unable to produce sputum; P = 0.04) were all positively associated with M. tuberculosis detection (Table, Supplemental Digital Content 4, We did not find an association between buccal sample collection method (OmniSwab vs. EasiCollect) and M. tuberculosis detection in either univariable or multivariable analyses adjusting for the child’s smear status (adjusted prevalence ratio = 2.4; 95% CI: 0.2–24; P = 0.45).


We detected M. tuberculosis DNA in buccal samples from children younger than 15 years of age with TB, using 2 different sample collection methods. Moreover, a positive sputum smear, the ability to spontaneously produce sputum and cavitary disease on chest radiograph were associated with M. tuberculosis detection from buccal samples among children with culture-confirmed TB.

A study from South Africa also evaluated the sensitivity of M. tuberculosis detection from pediatric buccal swabs.10 Using similar methods, that study reported a sensitivity of 30% from a single swab among children with culture-confirmed TB; this is higher, although within range of our observed sensitivity of 21% (95% CI: 7–42). Importantly, in that study M. tuberculosis was detected from a single buccal sample in 15% of children with clinically diagnosed unconfirmed TB, when compared with only 5% (n = 3) in our study. The addition of buccal sample testing increased the percent of TB cases with microbiologic confirmation from 27% (24/89) to 30% (27/89) in Peru (1 buccal sample) and from 33% (40/121) to 49% (59/121) in South Africa (2 buccal samples).10

A number of factors may contribute to the lower sensitivities in our study. First, cavitary disease on chest radiograph and a positive sputum smear were associated with M. tuberculosis detection from buccal samples. This has also been reported in pediatric TB molecular diagnostic studies of stool.4,5 The study from South Africa did not report smear status; therefore, differences in disease severity or disease presentation may have contributed to different sensitivities. Stratification by smear result or other markers of disease severity will facilitate comparisons between sample collection and detection methods and study populations. A higher false-positive rate in the South Africa study, ethanol DNA precipitation of negative samples and/or random variability are other possible explanations.

The detection of M. tuberculosis DNA in buccal samples from children in whom TB could otherwise not be confirmed represents an exciting possibility and highlights the need to optimize methods for this sample type. A number of factors, including the use of flocked swabs and sample collection from the tongue, improved sensitivity in adults and should be evaluated in children.9 We did not find differences in sensitivity by sample collection type; however, the low overall sensitivity combined with the small number of confirmed cases limited power. Collection of additional samples also improves yield: Nicol et al10 found that the addition of a second sample (in most cases collected immediately after the first) increased sensitivity from 30% to 43% among children with confirmed TB and from 15% to 24% among children with unconfirmed TB. Finally, imperfect pediatric TB diagnostic tools complicate interpretation of findings: if children without TB were misclassified as having TB unconfirmed clinically diagnosed TB, we will have underestimated sensitivity in this group. Similarly, missed TB diagnoses could result in misclassification of children with TB as controls, underestimating specificity. An extensive workup based on standard-of-care diagnostic procedures and evaluation by pediatric pulmonologists likely minimized this misclassification.

In conclusion, buccal samples are simple to collect in children and, with additional optimization, may increase the frequency with which TB disease can be confirmed in children.


The authors are grateful to the guardians and children who participated in this study as well as the staff at the Socios En Salud, Sucursal Peru in Lima city, where the study was managed. Additionally, the first author gives special thanks to the Department of Global Health and Social Medicine at Harvard University for providing him with a productive training and mentorship by Dr. Franke.


1. Oberhelman RA, Soto-Castellares G, Gilman RH, et al. Diagnostic approaches for paediatric tuberculosis by use of different specimen types, culture methods, and PCR: a prospective case-control study. Lancet Infect Dis. 2010; 10:612–620
2. Tafur KT, Coit J, Leon SR, et al. Feasibility of the string test for tuberculosis diagnosis in children between 4 and 14 years old. BMC Infect Dis. 2018; 18:57C4
3. MacLean E, Sulis G, Denkinger CM, et al. Diagnostic accuracy of stool xpert MTB/RIF for detection of pulmonary tuberculosis in children: a systematic review and meta-analysis. J Clin Microbiol. 2019; 57:e02057-18
4. Mesman AW, Rodriguez C, Ager E, et al. Diagnostic accuracy of molecular detection of Mycobacterium tuberculosis in pediatric stool samples: a systematic review and meta-analysis. Tuberculosis (Edinb). 2019; 119:101878
5. Mesman AW, Soto M, Coit J, et al. Detection of Mycobacterium tuberculosis in pediatric stool samples using TruTip technology. BMC Infect Dis. 2019; 19:563
6. Nicol MP, Allen V, Workman L, et al. Urine lipoarabinomannan testing for diagnosis of pulmonary tuberculosis in children: a prospective study. Lancet Glob Health. 2014; 2:e278–284
7. Mesman AW, Calderon R, Soto M, et al. Mycobacterium tuberculosis detection from oral swabs with Xpert MTB/RIF ULTRA: a pilot study. BMC Research Notes. 2019; 12:349
8. Wood RC, Luabeya AK, Weigel KM, et al. Detection of Mycobacterium tuberculosis DNA on the oral mucosa of tuberculosis patients. Sci Rep. 2015; 5:8668
9. Luabeya AK, Wood RC, Shenje J, et al. Noninvasive detection of tuberculosis by oral swab analysis. J Clin Microbiol. 2019; 57:e01847-18
10. Nicol MP, Wood RC, Workman L, et al. Microbiological diagnosis of pulmonary tuberculosis in children by oral swab polymerase chain reaction. Sci Rep. 2019; 9:10789
11. Cuevas LE, Browning R, Bossuyt P, et al. Evaluation of tuberculosis diagnostics in children: 2. Methodological issues for conducting and reporting research evaluations of tuberculosis diagnostics for intrathoracic tuberculosis in children. Consensus from an expert panel. J Infect Dis. 2012; 205(suppl 2):S209–215
12. MINSA. Modificatoria de la Norma Técnica de Salud para la atención integral de las personas afectadas por tuberculosis (NTS N°104 – MINSA/DGSP V.01. Ministerio de Salud del Peru. Available at: Accessed 10 March, 2020
13. Graham SM, Cuevas LE, Jean-Philippe P, et al. Clinical case definitions for classification of intrathoracic tuberculosis in children: an update. Clin Infect Dis. 2015; 61(suppl 3):S179–187
14. UNAIDS. UNAIDS Data 2019. UNAIDS. Available at: Accessed June, 2020
15. MINSA. Norma Técnica de Salud para la Atención Integral de las Personas Afectadas por Tuberculosis, aprobada por Resolución Ministerial N° 715-2013/MINSA. MINSA. Available at: Accessed June, 2020
16. Savelkoul PH, Catsburg A, Mulder S, et al. Detection of Mycobacterium tuberculosis complex with Real Time PCR: comparison of different primer-probe sets based on the IS6110 element. J Microbiol Methods. 2006; 66:177–180
17. Thakore N, Norville R, Franke M, et al. Automated TruTip nucleic acid extraction and purification from raw sputum. PLoS One. 2018; 13:e0199869

oral swab; real-time polymerase chain reaction; tuberculosis; childhood tuberculosis; diagnostic

Supplemental Digital Content

Copyright © 2020 Wolters Kluwer Health, Inc. All rights reserved.