Post-tuberculosis treatment paradoxical reactions

Abstract

Paradoxical reactions (PR) to tuberculosis (TB) treatment are common during treatment, but have also been described after treatment. A presentation with recurrent signs or symptoms of TB after cure or completion of prior treatment needs to be differentiated between microbiological relapse and a paradoxical reaction. We searched all published literature on post-treatment PR, and present a synthesis of 30 studies, focusing on the epidemiology, diagnosis and management of this phenomenon. We report an additional case vignette. The majority of studies were of lymph node TB (LN-TB), followed by central nervous system TB (CNS-TB). A total of 112 confirmed and 42 possible post-treatment PR cases were reported. The incidence ranged between 3 and 14% in LN-TB and was more frequent than relapses, and between 0 and 2% in all TB. We found four reports of pulmonary or pleural TB post-treatment PR cases. The incidence did not differ by length of treatment, but was associated with younger age at initial diagnosis, and having had a PR (later) during treatment. Post-treatment PR developed mainly within the first 6 months after the end of TB treatment but has been reported many years later (longest report 10 years). The mainstays of diagnosis and management are negative mycobacterial cultures and anti-inflammatory treatment, respectively. Due to the favourable prognosis in LN-TB recurrent symptoms, a short period of observation is warranted to assess for spontaneous regression. In CNS-TB with recurrent symptoms, immediate investigation and anti-inflammatory treatment with the possibility of TB retreatment should be undertaken.

Keywords: Lymph node; Paradoxical reaction; Paradoxical response; Paradoxical upgrading; Recurrence; Tuberculosis.

Fig. 1

Flow diagram of literature screening and inclusion

Fig. 2

Timing of post-treatment PR, stratified by category; a months since the end of TB treatment and b months after the start of TB treatment. Four recurrences in the first month after treatment are included in the second graph, but not in the first; as the primary data reported was insufficient to exclude them from the first [20]. The recurrences from the 1985 BTS publication were all included as possible PRs as the data stratified by month did not provide information on culture results [14]. As per the exclusion criteria, we excluded cases of PR that occurred at the end of treatment or in the first month after treatment. The data underlying this Figure are included in Table S3 of the appendix. An outlier (a confirmed PR) was reported at 132 months or 11 years (a) and 120 months or 10 years (b), respectively. Our case vignette was a confirmed case reported at 100 months (8.3 years) and 106 months (8.9 years), respectively