Mycobacterium kansasii in Immunocompromised Patients and Hospital Water Sources

Background: M. kansasii is an uncommon environmental contaminant that can cause serious pulmonary or disseminated infections resembling tuberculosis, with immunocompromised patients at highest risk. Understanding the clinical features of patients infected with M. kansasii and identifying potential sources of contamination within the hospital environment are essential for prevention and diagnoses.

Objective: To assess the clinical features of nosocomial M. kansasii infections in patients at a tertiary care medical center over time. To compare M. kansasii from clinical isolates and hospital water sources which were obtained as part of the investigation.  

Methods: IRB approval was obtained for a retrospective chart review of patients with M. kansasii isolated between 1995-2005 to include demographics, underlying conditions, and patient room locations with a focus on 1997, in which there was a perceived outbreak with water as the most likely source. To increase the yield of isolating M. kansasii from water, 1 L samples were obtained from multiple water sources at the medical center; 50 mL aliquots of water did not yield any positive M. kansasii cultures. These samples were mixed with 0.04% cetylpyridinium chloride for 24 h, filtered, washed, and transferred to Middlebrook 7H11 agar. M. kansasii positive water samples were compared to patient isolates using pulsed-field gel electrophoresis (PFGE) with the ATCC strain as a control.

Results: Eleven M. kansasii positive cases were identified in 1997; in 1995, 1996, and 1999 there were 7, 5, and 7 cases per year respectively; the other years had 0-3 cases per year. In 1997, median age was 38 years (range 21-76), 7 male, 10 Caucasian, and 1 African American. Co-morbid illnesses included: CML in 5 patients, 2 AML, 1 HIV, 1 lung cancer with COPD, 1 recurrent pulmonary emboli, and 1 kidney/pancreas transplant recipient. All 7 cases with CML/AML had chemotherapy and bone marrow transplant (BMT). M. kansasii grew from: 5 BALs, 2 sputums, 1 CVC, 1 shin lesion, 1 stool, and 1 surgical wound. Median hospital length of stay was 39 days (range 4-143). At time of positive culture, 9 of 11 patients were clustered within 4 units of the cancer hospital. Seven of 11 patient isolates and 3 (2 from BMT unit patient rooms, 1 from BMT unit ice machine) of 4 water sources matched PFGE banding patterns of the M. kansasii ATCC strain. Two patient isolates and the remaining water source were subtypes of this strain. Two patient isolates were different strains and corresponded to the 2 patients in a hospital other than the cancer hospital. Sterile water and ice were initiated for patients on BMT, leukemia/lymphoma, and solid organ transplant units. 

 

Conclusions: All nosocomial M. kansasii cases in 1997 involved immunocompromised patients. As the strain/strain subtype of M. kansasii from water sources matched isolates from patients clustered in the cancer hospital, water was confirmed as a likely source of infection.