Introduction

Mycobacterium ulcerans, a slow growing environmental pathogen [1], causes a necrotizing cutaneous infection with various names but now known internationally as Buruli Ulcer (BU). The mode of transmission remains controversial [2],[3] however the key identifiable risk factor for BU is residence in or contact with a Buruli endemic region. Worldwide, most cases of BU occur in patients who live permanently in endemic areas, and because the mode of transmission is unknown it is generally impossible to estimate the incubation period (IP).

The first definitive description of M. ulcerans infection (termed Bairnsdale ulcer in that report) was in Australia in 1948 [1],[4]. Cases have been reported from at least 32 countries, in regions including Africa, Australia, Southeast Asia, China, Central and South America and the Western Pacific [5]. In Victoria, cases have been reported in Gippsland, Phillip Island [6] and the Mornington [7] and Bellarine Peninsulas [8],[9] [10]. In Queensland there is a significant established focus in Far North Queensland [11],[12]. Sporadic cases have been reported in the Northern Territory (NT) [13] and Capricorn Coast of southern Queensland [14]. Single cases have occurred in Western Australia (WA) [15] and New-South Wales (NSW) [16]. Transmission is not reported in South Australia or Tasmania.

Few published studies exist dedicated to the incubation of BU. The Uganda Buruli Group estimated that the normal IP was 4–13 weeks [17]. The shortest IP described is 2–3 weeks, in a newborn from Papua New Guinea [18]. Recently within Australia, five patients living outside endemic areas with travel to known endemic regions had reported incubation periods ranging from 2 to 5 months [19]. In Victoria where BU is legally notifiable, diagnosis by Polymerase Chain Reaction (PCR) is centralized and rapid and endemic areas are systematically mapped by members of the WHO Collaborating Centre (WHOCC) in Melbourne. Hence there is a unique opportunity to systematically estimate IP.

Methods

BU has been a legally notifiable infection by treating physicians and laboratories in Victoria since January 2004 and data has been collected systematically and recorded for several years before this. For this investigation we reviewed notifications in the 10-year period 2002 and 2012. The case definition for BU was the presence of a lesion clinically suggestive of M. ulcerans infection together with at least one of;

• Positive polymerase chain reaction (PCR) for IS2404 [20],[21],[22]
• Culture and specific identification of M. ulcerans by a Mycobacterium Reference Laboratory.

An endemic region was defined as the constant presence of the organism (e.g. M. ulcerans) within a given geographic area or population group [23]. Following literature review and consultation with the Health Department records and those of the WHOCC, endemic regions within Australia were defined and mapped (Table 1, Figure 1 & 2). Significant exposure was defined as greater than one continuous hour in an endemic area and the maximum possible IP considered was two years prior to diagnosis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Geographic representation of Bellarine Peninsula, considered endemic for BU as of 2012.

Bellarine Peninsula – east of line from Geelong to Torquay. Mornington and Westernport – southwest of line from Hampton to Tooradin (including Phillip Island).

https://doi.org/10.1371/journal.pntd.0002463.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Geographic representation of East Gippsland, considered endemic for BU as of 2012.

East Gippsland: East of Sale and south of the great divide.

https://doi.org/10.1371/journal.pntd.0002463.g002

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Regions within Eastern Australia defined as endemic for M.ulcerans.

https://doi.org/10.1371/journal.pntd.0002463.t001

Telephone interview with a structured questionnaire was undertaken with cases or next-of-kin in conjunction with review of medical records and Victorian Department of Health notification record. Data were collected for all geographical exposures to designated BU endemic areas and a map was provided to interviewees to assist them. The duration of exposure was obtained for a single visit and date of onset of symptoms recorded. If a patient was unable to recall a single date of disease exposure, a period of exposure was obtained (maximum one month) and a minimum and maximum IP value for that patient calculated (IP range). If a patient had multiple exposures to known endemic regions within Australia (Northern Territory, Queensland, Western Australia) or overseas and only one occurring within Southern Australia (Victoria and NSW), where possible, variable-number tandem repeat (VNTR) typing was performed on the isolate to identify the region of origin [19],[24]

For the 23 patients with a single visit exposure, an IP range was calculated in days and months. The midpoint of the IP range was utilized to calculate a point estimate of the IP for each case. Subsequently, a mean IP for the entire study cohort was calculated. Baseline patient characteristics and data collected were patient residential address, age, sex, location of lesion, duration of endemic exposure, date of first medical attention/form of medical attention, date of when BU first suspected, specimen provided for diagnosis, PCR confirmation, treatment and outcome(s).

Results

Of the 408 notifications of BU in the 10-year period 2002–2012, 111 (27%) patients had residential addresses outside the assigned endemic regions, 23 (6%) with single visit exposures and it was from these patients that the IP was estimated. The median age at time of diagnosis of this group was 30 years (Range: 6–73), 65% were male. The residential addresses of all patients were outside established endemic regions (Figure 1 & 2). The endemic areas visited were the Bellarine Peninsula (14; 61%), Phillip Island (3; 13%) and Mornington Peninsula (3; 13%), with single visit exposures to each of Gippsland (VIC), Darwin (NT) and Port Douglas/Mossman (QLD). The duration of exposure was predominately greater than 7 days (60%), but single day exposure was noted in 4 patients (17%) (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Characteristics of the 23 cases of BU included in the study cohort.

https://doi.org/10.1371/journal.pntd.0002463.t002

The lesions occurred primarily on the extremities, with the lower limbs the most prevalent site (14; 60%). Fifty-one percent of patients recalled mosquito bites within endemic regions, whilst only one patient reported having an open wound. The first medical professional sought for diagnosis and management was a general practitioner from a non-endemic area in 20 of the 23 cases (87%), with the diagnosis of BU suspected in only 17% of cases. The diagnosis of BU was suspected in 3 of 23 (13%) patients on first presentation overall, and 3 of 20 (15%) patients presenting initially to a non-endemic region general practitioner. The median time from symptom onset to PCR confirmed diagnosis was 71 days (range: 23–204 days). The median time from initial medical attention to diagnosis was 35 days (range: 15–103 days) for all patients.

Treatment was multimodal (medical management & surgery) in 16 (70%), surgery alone in 3 (13%) and medical alone in 4 (17%). A regime containing rifampicin and one of either clarithromycin or moxifloxacin was used in 17 (94%) patients whose medical management included antibiotic therapy. No episodes of relapse or new infection were reported after treatment completion. Follow-up ranged from 3 months to 9 years (median 24 months).

The midpoint of the reported IP range was utilized to calculate a point estimate of the IP for each case (Figure 3a). The IP point estimates for the 23 patients were normally distributed. The mean IP point estimate for the cohort was 135 days (IQR: 109–160 CI 95%: 113.9–156), corresponding to 4.5 months. The shortest incubation period noted was 32 days (1 month) and longest 264 days (9 months). Univariate analysis (Mann-Whitney U) did not identify any significant association between incubation period and sex (p = 0.489), endemic exposure (Bellarine Peninsula vs. other, p = 0.659), location of lesion (arm vs. leg, p = 0.391), age (≤30 vs. ≥31 years, p = 0.644) or the duration of exposure (<7 vs. >7 days, p = 0.207). (Table 3)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. The IP for BU infection in the study population.

The IP range (X-axis) for each patient (Y-axis) is demonstrated, arranged from shortest to longest IP. IP range midpoint values are marked as the middle line of each bar.

https://doi.org/10.1371/journal.pntd.0002463.g003

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. The IP according to variables investigated.

https://doi.org/10.1371/journal.pntd.0002463.t003

Discussion

Limited published data exists on the incubation period of Buruli ulcer. The Uganda Buruli Group in 1971 proposed an IP of 4–10 weeks in refugees who continued to develop BU following departure from an endemic refugee camp [17], with a rapid cessation of new diagnoses beyond 10 weeks. This may have been an underestimate, as the duration of exposure at the refugee camp is not entirely defined. A study of thirteen cases of BU from Port Moresby identified the shortest known incubation period (2–3 weeks) in a 6-week old baby born in an endemic region [18].

Historical data suggest incubation period estimates of 4–14 weeks [1],[13] [Hayman J, personal communication]. Reports in travelers to endemic regions have also aided research, such as a minimum IP of six weeks in a Nigerian physician working in New York City [25] and 5 months in a traveller from PNG [26]. Molecular typing has been used to identify region-specific M. ulcerans strains, from this the IP for two patients was estimated to be 3 and 7 months [17] and in five travelers from non-endemic areas to QLD and NT a range of 2 to 5 months was estimated [19].

This is the first public health investigation to systematically investigate IP in patients for whom exposure can be reasonably defined. Generally, it is not possible to do this as most patients with BU live permanently in endemic areas, hence time of transmission cannot be determined. Due to the previous systematic mapping documentation of endemic areas in Victoria and interviews with patients obtained through enhanced surveillance after notification we have estimated the mean IP for BU of 135 days, corresponding to 4.5 months or 19.2 weeks. The point estimates appeared normally distributed which suggests, but does not prove, that latency is not a major feature of M. ulcerans infection as we would expect a more right shifted distribution if this were the case, and that inoculum size is relatively similar between patients. The shortest IP noted was 1 month, the longest 9 months. We found no statistical association between individual IPs and other recorded covariables, but acknowledge this may be explained by our small final cohort size.

A secondary finding was of that BU is frequently not considered in the differential diagnosis when patients with BU first present to their doctor outside a BU endemic area. Buruli ulcer is an uncommon disease and diagnosis is often delayed when doctors are not familiar with the conditions. Diagnostic delay was not formally assessed with a matched cohort presenting in endemic areas in this investigation, however prolonged time to diagnosis in endemic compared with non-endemic areas has been reported previously [25].

This enhanced surveillance report has several limitations. The small final cohort number of patients in the group with single exposure to endemic areas limited the ability to determine meaningful associations with variables noted in Table 3. Furthermore, patient recall remains an issue particularly for earlier notifications. Nonetheless, the ability to cross-reference the telephone interview with the primary notification and enhanced surveillance forms reduced recall bias. In any investigation of BU, will always be an issue as there is a substantial period between exposure, symptom onset and diagnosis. Furthermore, as the mode of M.ulcerans transmission and exact inoculum size remains unknown, the impact of these variables is not completely understood. Nonetheless, the incubation period appears normally distributed around the point which suggests but does not prove that the inoculum size may have been similar in these 23 patients. We also acknowledge that the IP we have obtained was established in patients from Victoria only, and may or may not be generalizable to tropical Australia or the major endemic foci in sub-Saharan Africa. However our results are broadly similar to previous estimates from Africa and elsewhere.

Acknowledgments

We thank Dr. Brett A. Sutton (Senior Public Health Officer, Communicable Disease Prevention and Control Unit, Victorian Department of Health), Ms. Lucinda Franklin (Epidemiologist, Victorian Department of Health, Australia) and Dr. Rosemary Lester (Chief Health Officer, Victorian Department of Health.) for their contribution.