Incidence, clinical spectrum, diagnostic features, treatment and predictors of paradoxical reactions during antibiotic treatment of Mycobacterium ulcerans infections

Abstract

Background: Paradoxical reactions from antibiotic treatment of Mycobacterium ulcerans have recently been recognized. Data is lacking regarding their incidence, clinical and diagnostic features, treatment, outcomes and risk factors in an Australian population.

Methods: Data was collected prospectively on all confirmed cases of M. ulcerans infection managed at Barwon Health Services, Australia, from 1/1/1998-31/12/2011. Paradoxical reactions were defined on clinical and histological criteria and cases were determined by retrospectively reviewing the clinical history and histology of excised lesions. A Poisson regression model was used to examine associations with paradoxical reactions.

Results: Thirty-two of 156 (21%) patients developed paradoxical reactions a median 39 days (IQR 20-73 days) from antibiotic initiation. Forty-two paradoxical episodes occurred with 26 (81%) patients experiencing one and 6 (19%) multiple episodes. Thirty-two (76%) episodes occurred during antibiotic treatment and 10 (24%) episodes occurred a median 37 days after antibiotic treatment. The reaction site involved the original lesion (wound) in 23 (55%), was separate to but within 3 cm of the original lesion (local) in 11 (26%) and was more than 3 cm from the original lesion (distant) in 8 (19%) episodes. Mycobacterial cultures were negative in 33/33 (100%) paradoxical episodes. Post-February 2009 treatment involved more cases with no antibiotic modifications (12/15 compared with 11/27, OR 5.82, 95% CI 1.12-34.07, p = 0.02) and no further surgery (9/15 compared with 2/27, OR 18.75, 95% CI 2.62-172.73, p < 0.001). Six severe cases received prednisone with marked clinical improvement. On multivariable analysis, age ≥ 60 years (RR 2.84, 95% CI 1.12-7.17, p = 0.03), an oedematous lesion (RR 3.44, 95% CI 1.11-10.70, p=0.03) and use of amikacin in the initial antibiotic regimen (RR 6.33, 95% CI 2.09-19.18, p < 0.01) were associated with an increased incidence of paradoxical reactions.

Conclusions: Paradoxical reactions occur frequently during or after antibiotic treatment of M. ulcerans infections in an Australian population and may be increased in older adults, oedematous disease forms, and in those treated with amikacin. Recognition of paradoxical reactions led to changes in management with less surgery, fewer antibiotic modifications and use of prednisolone for severe reactions.

Figure 1

Local paradoxical reaction following complete surgical excision of original lesion (see surgical scar) manifest by the appearance of 3 new lesions within 3 cm of the surgical site 6 weeks after antibiotic commencement.

Figure 2

Illustration of clinical features representing a wound paradoxical reaction.a-b: Buruli ulcer lesion posterior calf at commencement of antibiotic treatment with induration 40x42 mm diameter (a). Lesion after 5 weeks antibiotic treatment complicated by a paradoxical reaction manifest by increased induration around lesion (60x40 mm diameter), increased serous discharge and new ulceration adjacent to initial lesion (b).

Figure 3

Cumulative incidence of paradoxical reactions. a: Cumulative incidence of first paradoxical reactions. b: Cumulative incidence of first paradoxical reactions by age. c: Cumulative incidence of first paradoxical reactions by lesion type. d: Cumulative incidence of first paradoxical reactions by use of amikacin.

Figure 4

Number and proportion of patients developing paradoxical reactions per calendar year.

Figure 5

Illustration of histological features representing a paradoxical reaction. a-d: Initial lesion before antibiotic treatment showing a sparse acute inflammatory reaction around necrotic fat and subcutaneous tissue (a) with high numbers of extracellular acid-fast bacilli (in pink) on Wade-Fite stain (b). After 8 weeks of antibiotic treatment showing a paradoxical reaction manifest by a dense inflammatory reaction including multinucleated giant cells (c) and no extracellular acid-fast bacilli seen with Wade-Fite stain (d).