Can the imaging manifestations of melioidosis prognosticate the clinical outcome? A 6-year retrospective study

Melioidosis is an important cause of community-acquired sepsis in the tropical and subtropical countries being well documented in Southeast Asia and northern Australia [1, 2]. Although sporadic case reports from India were noted by Cheng et al. [1] in 2005, most of the Indian subcontinent is recognised endemic for melioidosis since 2008 [3]. The disease is caused by a saprophytic gram-negative bacillus, Burkholderia pseudomallei (B. pseudomallei) [1, 4‐6], freely living in soil and water. Human infection occur following percutaneous inoculation, inhalation of aerosolised bacteria or accidental ingestion; rarely person-to-person transmission, nosocomial transmission and laboratory-acquired infection can occur [1, 7]. Along with the expansion in human travel and movement, and climate change, there is a risk of imported infection worldwide and extension in the boundaries of endemic melioidosis [3, 8]. Clinical manifestations vary from localised cutaneous infections at sites of inoculation, chronic multisystem involvement mimicking tuberculosis to acute fulminant sepsis and death [9, 10]. Fatal cases have identified risk factors; the known risk factors are diabetes, excessive alcohol consumption, chronic renal and lung disease, thalassemia and other haematological conditions, immunocompromised state and occupational exposure to the pathogen [8, 11]. Mortality rates are 10 to 14% in Australia, 40 to 50% in Thailand and 30% in Malaysia [1, 5, 9, 12]. The reported mortality rate from the published literature in endemic countries varied from up to 58% in 2003 [13] and 9 to 25% in recent years [14, 15].

The standard of diagnosis is by the culture of a clinical specimen: however, challenges in laboratory diagnosis are lack of resources and expertise, misidentifying or underreporting of the bacilli with methods used routinely in clinical laboratories [16, 17]. Imaging is important in the evaluation of multisystem involvement of melioidosis; organ involvement manifests most commonly as abscesses [18]. Certain characteristics of abscesses caused by melioidosis or combinations of organ involvement may contribute to diagnosis in an appropriate clinical scenario [19, 20]. Though plain radiograph or ultrasonogram is often the first line of imaging, it is surpassed by cross-sectional studies in its usefulness to describe the extent and assess the complications of infection. There are studies that have analysed clinical and lab parameters indicating mortality [21, 22]; however, there are very few research in the usefulness of imaging in predicting the clinical outcome. In this study, we aim to describe the imaging manifestations of melioidosis and to analyse the organ involvement which could predict unfavourable outcome, and evaluate the association between the organ imaging manifestation and known risk factors/predisposing conditions.

Early and accurate diagnosis of melioidosis is warranted due to high fatality of melioidosis which could be as high as 80% amongst those with ineffective antibiotics and up to 50 to 90% amongst septic melioidosis [16, 25]. Hence, imaging is a complementary diagnostic tool as well as a predictor of disease outcome and mortality to guide patient management. This study, being one of the largest cohorts with culture-proven melioidosis, described the various imaging manifestations of melioidosis and prognosticate the clinical outcome. There is limited published literature which has evaluated the association of imaging features of meloioidosis with clinical outcome.

Multiple abscesses are the most frequent imaging manifestation of melioidosis [18]. The presence of respiratory system involvement showed significant association with a more severe form of infection with unfavourable outcome and predicted higher risk of mortality independently, compared to other organ or system involvement. Involvement of the lungs was most prevalent in our study population, as in previous studies [26‐28]. Contradictorily, pleural effusion, not a previously described major feature, was seen in more than half of our cases. Perhaps this could be because smaller effusions are being more discernible on cross-sectional studies as compared to plain radiography. While a significant number of pleural effusions were part of the constellation of lung findings, a small number were reactive to splenic or liver manifestations of the disease.

Patients with diabetes mellitus were statistically more prone for infection of the bones and soft tissue. This explains the most common involvement of the bones and soft tissue in our cohort who had a 75% prevalence of diabetes as a risk factor. The proportion of the bones and joint involvement in our study population and the commoner affectation of the lower limbs to upper limbs agree with previous observations by other researchers [29, 30]. Additionally, infection of the bones and soft tissues was found to have statistically significant association with no mortality in our study population. Imaging pattern of appendicular or axial bone involvement was non-specific from other causes of infection. Soft tissue involvement was seen as abscesses with surface breakdown resulting in a site for discharging pus.

There was also significant association between diabetic mellitus and splenic abscess. The spleen was the most common organ to manifest abscess foci in disseminated disease; no splenic involvement was seen in isolation. Other studies have also made this observation with note that splenic lesions occurring along with an infective focus in the body would favour the diagnosis of melioidosis [19, 31, 32] especially in a diabetic patient. It is possible that disseminated disease is common amongst patients with diabetes mellitus where the infection spreads readily to the spleen.

Solid organ abscesses with internal septations giving the abscess a honeycomb appearance have been described in previous studies as well [18, 20, 33, 34]; while these observations were made in the liver, similar characteristics were seen in renal and prostate abscesses as well. Pancreatic involvement is seen as focal pancreatitis or contiguous involvement in disseminated disease with adjacent thrombophlebitis as has been described in earlier studies [35]. The reason for the absence of genitourinary tract involvement in disseminated melioidosis, in those patients with past history or concurrent tuberculosis, could not be ascertained in this study. Neurological manifestations were seen in 5% of our study population, the proportion is similar to previous studies in Northern Australia [36]. When regional lymph nodes were involved, it predicted overall favourable outcome which was statistically significant.

Tuberculosis is a close mimicker of melioidosis on imaging. Other pyogenic infections also share some of the imaging characteristics as melioidosis. The lung nodules in melioidosis are commonly larger and irregular. Lung nodules in active tuberculosis are commonly smaller, miliary nodules being typical. In our cohort, we did not encounter isolated lung cavities, though there were cavities within the consolidation. Melioid cavities are thin walled, and air-fluid levels are unusual findings, as compared to tuberculosis [37]. Isolated pleural effusion or pleural calcification could be seen with tuberculosis as well as melioidosis but uncommon in other pyogenic infection. Melioid abscesses in the liver could be larger or smaller. Honeycomb sign is found to be specific for melioidosis [38]. Abscesses in the liver and spleen are not commonly encountered in tuberculosis, whereas they are common with other pyogenic infections. Larger pyogenic abscesses can also demonstrate honeycomb sign, though not specific. Superficial soft tissue abscess along with deep solid visceral organ abscess is common with melioidosis and pyogenic infection, but uncommon with tuberculosis. Superficial soft tissue abscess along with underlying bone or joint infection is common with melioidosis and tuberculosis but uncommon with other pyogenic infections.

The radiologist has the potential to be the first to alert the clinician and microbiologist for enhanced diagnosis. Being a retrospective study, the timeline of clinical presentation and imaging investigation may not be uniform in our cohort of patients. Hence, any difference in the imaging manifestation of various organ involvement with respect to the duration of illness cannot be evaluated in our study. The pattern of resolution of imaging findings could not be determined in most of the cases due to the absence of follow-up imaging.

In this large cohort of culture-proven melioidosis, we have described the spectrum of imaging manifestation of melioidosis and evaluated the association between clinical outcome and organ involvement as evaluated on imaging. Respiratory system involvement in melioidosis showed significant association with unfavourable outcome and predicted higher risk of mortality independently. Honeycomb sign which is specific for melioid liver abscesses is found in other solid organ melioid abscesses. Diabetes mellitus being a common risk factor for melioidosis is more prone for infection of the spleen and musculoskeletal system. Thus, awareness of imaging manifestations of melioidosis can complement microbiological diagnostic tests for accurate early diagnosis and timely medical management of melioidosis. Hence, imaging is an essential tool to assess the extent of disease, prognosticate and influence overall patient management and clinical outcome.