Prophylactic tracheostomy to facilitate repeat biopsy and initiation of treatment in a patient with laryngeal tuberculosis causing critical supraglottic stenosis

Laryngeal tuberculosis (TB) accounts for less than 1% of all TB cases globally but is an important diagnostic and public health concern due to its heightened transmission potential.¹ The laryngeal location facilitates efficient aerosolisation of Mycobacterium tuberculosis during normal speech and coughing, resulting in household secondary attack rates comparable to or exceeding those of smear-positive pulmonary TB (2.5–8.8%).^2–4 This high degree of infectivity, combined with non-specific clinical features that can mimic malignancy, makes early diagnosis essential. In New Zealand, TB notification rates remain low (approximately six per 100,000 population), with the disease burden concentrated among migrants (88% of cases in 2023 occurred in overseas-born individuals).⁵

We present a patient with laryngeal TB causing critical supraglottic stenosis in whom prophylactic tracheostomy and adjunctive high-dose corticosteroids were used to protect the airway during repeat laryngeal biopsy and treatment initiation. This case highlights challenges in the diagnosis of laryngeal TB, the important intersection between transmission risk and diagnostic delay, and a proactive approach to airway management in a high-risk clinical scenario.

Case report

A woman from Southeast Asia in her mid-20s presented to the otolaryngology clinic with a 14-month history of progressive dysphonia and exertional dyspnoea with noisy breathing that resolved at rest. She was an ex-smoker with a past medical history of gastroesophageal reflux disease and asthma. On examination she had normal intraoral appearances and there was no cervical adenopathy. Flexible nasendoscopy demonstrated significant supraglottic stenosis with oedematous mucosa (Figure 1A). The initial differential diagnoses considered included autoimmune vasculitis, sarcoidosis or respiratory papillomatosis. Chest X-ray (Figure 2A) showed bilateral mid-upper zone linear opacities reported as compatible with sarcoidosis. Antineutrophil cytoplasmic antibodies, rheumatoid factor and antinuclear antibodies were negative and a serum angiotensin-converting enzyme level was normal (32.5U/L). A QuantiFERON-TB Gold demonstrated inadequate mitogen control response so was reported as indeterminate. In keeping with standard practice, human immunodeficiency virus (HIV) serology was performed following the diagnosis of TB and returned negative.

Subsequent microlaryngoscopy revealed severe supraglottic stenosis with an obliterated epiglottis and thickened aryepiglottic folds compromising the airway. The glottis, subglottis and trachea were normal. A biopsy of the glottic tissue, preserved in formalin, revealed caseating granulomatous inflammation with focal suppuration. Based on the histological findings, the formalin-fixed tissue was deparaffinised and GeneXpert MTB/RIF Ultra^® performed, which was negative.

View Figure 1–2, Table 1.

Further targeted history revealed that the patient cohabited with a sister who had been treated for abdominal TB 1 year prior. Subsequent computed tomography imaging demonstrated bilateral upper-lobe opacities and diffuse supraglottic thickening consistent with laryngeal and pulmonary TB (Figure 2B, 2C and 2D).

Due to the high index of suspicion for TB and the poor sensitivity of the GeneXpert on deparaffinised samples, a repeat biopsy was planned prior to presumptive TB treatment initiation. Given the severe fixed supraglottic stenosis, airway protection was deemed essential for safe repeat biopsy. A semi-elective tracheostomy was performed to secure the airway. Intraoperative findings confirmed severe stenosis with progressive intubation-induced oedema (Figure 1B). Repeat GeneXpert MTB/RIF Ultra^® on a fresh tissue biopsy specimen was reported as trace positive for M. tuberculosis complex but there was insufficient genetic material to determine rifampicin susceptibility. GeneXpert MTB/XDR polymerase chain reaction run on the same sample did not detect any target.

Prophylactic dexamethasone was initiated prior to the repeat biopsy with subsequent transition to a weaning course of prednisone to mitigate the risks of paradoxical airway inflammation (see Table 1 for dosing). Anti-TB treatment was commenced on day 1 post–repeat biopsy, consisting of rifampicin 600mg once daily (OD), moxifloxacin 400mg OD, pyrazinamide 1250mg OD and ethambutol 900mg OD. Moxifloxacin was used in place of isoniazid because the patient’s sister was known to have had an isoniazid-resistant strain and transmission within the household was suspected.

The patient was decannulated on day 9 and was discharged on day 15 once the treating teams were comfortable that her airway was secure and there was no further risk of transmission. Subsequent culture of the repeat laryngeal tissue biopsy specimen grew M. tuberculosis after approximately 7 weeks and whole-genome sequencing (WGS) confirmed a fully drug-susceptible strain with no resistance-associated mutations to first- or second-line drugs when interpreted using the World Health Organization mutation catalogue.^6,7 The patient was transitioned to rifampicin 600mg OD, isoniazid 300mg OD and pyridoxine 25mg OD after 8 weeks of therapy, and a total of 6 months of anti-TB treatment is planned. At subsequent review, with 3 weeks of the planned 6-month treatment course remaining, the patient had no dyspnoea or stridor and had resumed running. Phonation had improved substantially and she was able to sing, although her voice remained mildly reedy. Near-complete epiglottic loss and anterior commissure scarring of the vocal cords remained the primary drivers of residual voice changes.

Discussion

Laryngeal TB presents most commonly with dysphonia or hoarseness (71–97%), followed by odynophagia (25–40%), dysphagia, cough and weight loss.⁸ These non-specific symptoms frequently mimic malignancy, sarcoidosis or vasculitis.⁸ Misdiagnosis as laryngeal cancer occurs in up to 21% of cases when lesions appear exophytic or ulcerative.⁸ Published case series report median time between symptom onset and diagnosis of 3 months,⁹ with diagnostic delays exceeding 4 months reported in 40% of extrapulmonary TB cases in low-incidence countries.¹⁰ In our patient, the time from symptom onset to diagnosis was 15 months, including an approximately 4-month diagnostic delay following initial clinical presentation.

Histopathology confirmed caseating granulomatous inflammation, which is characteristic, but not pathognomonic, of TB (also occurring in some fungal infections).¹¹ When interpreted in the context of previous migration from a TB-endemic region, apical pulmonary changes and potential household TB contact, the probability of TB was felt to be exceptionally high and was subsequently confirmed by both molecular and culture-based methods. The initial negative GeneXpert MTB/RIF^® assay reflects well-documented GeneXpert MTB/RIF^® sensitivity limitations in paucibacillary specimens, with false-negative rates of 21.8–41.1% reported.¹² Even when positive, “trace” results occur in 12–17% of cases.¹³ Formalin fixation further impairs sensitivity.¹⁴ Multiple specimens are often necessary to obtain a diagnosis of TB and, in some cases, empiric treatment in the absence of confirmatory microbiological results is warranted. Maintaining diagnostic awareness is critical. An earlier exposure history and targeted testing for TB may have shortened the diagnostic delay and reduced the patient’s infectious period. Mortality is rare in immunocompetent individuals. Voice quality often improves following treatment for laryngeal TB, though mild dysphonia may persist with delayed diagnosis.⁸

Public health considerations

Contact tracing for infectious TB is conducted by local public health services within the National Public Health Service (NPHS) in accordance with national guidance.¹⁵

In this scenario, the person with laryngeal TB was a member of a large household, all but one of whom had migrated within the last 3 years from a country with a high incidence of TB. This increased household members’ likelihood of prior exposure to TB and therefore positive interferon-gamma release assay (IGRA) tests. Four of the six household members who were tested had positive IGRA results on their first cycle of testing. Once active TB was excluded via chest X-ray and symptom screening, these household members were referred for treatment for latent TB. Two of the newer household members returned initial negative tests and remained negative at their second test, 8 weeks post-exposure. This included one member of the household who shared a room with the case.

Further contacts were identified during the case’s first hospital admission, before TB was suspected and airborne precautions were instituted. All staff involved in the first laryngoscopy returned negative IGRA results when tested at 8 weeks post-exposure.

There was a history of TB in the patient’s household prior to moving to New Zealand. Additionally, a different member of the patient’s household was diagnosed and treated for peritoneal TB in New Zealand, 1 year prior to the diagnosis of laryngeal TB in our patient. Review of the WGS reports and the antibiotic resistance profiles indicates that the two recent cases of active TB in the household were caused by unrelated strains and that independent acquisition events for these two patients probably occurred. Although the high proportion of household contacts (50%) found to have positive IGRA tests would be consistent with the high degree of transmissibility of laryngeal TB,² alternative sources for the high proportion of family members with positive IGRA are plausible. It remains plausible that the other household members acquired latent TB infections from sources outside of the household, as is common in countries with high TB incidence.

Since none of the initial close contacts of our patient with laryngeal TB seroconverted, contacts with lesser degrees of exposure (i.e., beyond the household and exposed hospital staff) were not pursued.

Regarding the peritoneal TB case, it is important to consider source investigation for cases of non-infectious extrapulmonary TB, as this may also identify people who may have been exposed to the same source. It is standard practice for the NPHS to ask screening questions for all household close contacts for symptoms of TB (cough, haemoptysis, fever, night sweats, unexplained weight loss).¹⁵ Dysphonia is not one of the core symptoms routinely screened for and thus the yield of these questions for detecting laryngeal TB is likely to be quite low. Adding further specific symptom screening questions or a “catch-all” symptom question to the Guidelines for Tuberculosis Control in New Zealand 2019 (Ministry of Health – Manatū Hauora) could be considered, although the sensitivity and specificity of the symptoms for TB would need to be considered.

Despite New Zealand’s low overall TB incidence, TB must remain actively considered in relevant differential diagnoses, particularly in patients with epidemiologic risk factors or compatible symptoms, and high-quality clinical and public health services must be maintained to support timely diagnosis, contact tracing and treatment.

Airway management

To the best of our knowledge this is the first published report of the use of pre-emptive tracheostomy for airway stabilisation prior to biopsy and treatment initiation in laryngeal TB. Multiple cases of emergency tracheostomy for acute airway obstruction in laryngeal TB have been described, indicating that prophylactic airway management is warranted in some cases.¹⁶ In our patient, the airway narrowing was felt to be sufficiently severe that any decompensation caused by trauma or bleeding from the biopsy or paradoxical treatment-related inflammation would have been very poorly tolerated. The secure airway facilitated early discharge of the patient from the intensive care unit to the ward.

Paradoxical reactions (worsening of existing loci of disease or emergence of new lesions after initial clinical improvement on treatment) occur in 2–30% of HIV-negative TB patients, with substantially higher rates in extrapulmonary TB (25%) compared with pulmonary TB (2%).^17,18 Specific data on incidence of paradoxical reactions in laryngeal TB are lacking. Paradoxical reactions reflect delayed-type immune response to mycobacterial antigens released with effective therapy, and typically present early with two-thirds of reactions occurring within 1–4 months of initiating anti-TB therapy, (median onset 3–6 weeks).^17–19 The PredART randomised controlled trial demonstrated that prophylactic prednisone (40mg/day for 2 weeks, then 20mg/day for 2 weeks, started within 48 hours of antiretroviral therapy initiation) reduces paradoxical TB-immune reconstitution inflammatory syndrome incidence in high-risk HIV-infected patients (cluster of differentiation 4 [CD4] count <100 cells/μL, starting antiretroviral therapy [ART] within 30 days of TB treatment initiation): cumulative incidence 46.7% with placebo versus 32.5% with prednisone—a 30% relative risk reduction.²⁰ Although these data are HIV-specific, three factors justified extrapolating the prophylactic steroid strategy to our non-HIV infected patient: 1) extrapulmonary TB carries a comparatively high risk of paradoxical reaction (25%), 2) paradoxical reactions tend to occur early during treatment and thus disease-related airway narrowing was unlikely to have improved substantially by the time of onset, and 3) the consequences of any degree of paradoxical reaction in such a critical location were unacceptably high.

Steroid dose selection was based on published guidance for the use of steroids in tuberculous meningitis.²¹ Although paradoxical reactions can occur at any point during treatment, and even after treatment has been completed, the initial period during which disease-related airway narrowing was the most severe was felt to be the highest risk period. While this case cannot prove tracheostomy prevented obstruction, it illustrates a multidisciplinary strategy for a high-risk patient with laryngeal TB requiring biopsy and in this instance was associated with a very favourable clinical outcome. We suggest that semi-elective tracheostomy with adjunctive corticosteroids should be offered when critical supraglottic narrowing makes treatment-induced oedema a life-threatening risk. In such scenarios, proactive stabilisation offers a safer, more-controlled alternative to reactive emergency intervention.

Conclusion

This case illustrates two pivotal insights for New Zealand’s clinical and public health practice. First, laryngeal TB exhibits high transmissibility, with household secondary attack rates of 2.5–8.8%, matching or surpassing smear-positive pulmonary TB; a 4-month delay from initial clinical presentation to diagnosis here highlights the imperative for heightened awareness, especially among migrants from TB-endemic areas with persistent dysphonia. Second, in critical supraglottic stenosis, semi‑elective tracheostomy with adjunctive high‑dose prophylactic corticosteroids provides a proactive, safer strategy versus emergency measures, enabling secure biopsy, prompt therapy and expedited intensive care discharge. To our knowledge, this represents the first reported use of pre-emptive tracheostomy in this context. Despite New Zealand’s low overall TB incidence, TB should remain an active consideration in relevant differential diagnoses, and high-quality clinical and public health services must be protected to enable timely diagnosis, contact investigation and treatment.