Characterization of a Tsukamurella Pseudo-Outbreak by Phenotypic Tests, 16S rRNA Sequencing, Pulsed-Field Gel Electrophoresis, and Metabolic Footprinting

Abstract

We report a pseudo-outbreak of Tsukamurella due to improperly wrapped scissors used for processing of tissue specimens. A polyphasic approach, involving biochemical, genetic, and metabolomic techniques, was used in the laboratory investigation. This report highlights that early recognition of pseudo-outbreaks is important in preventing unnecessary and incorrect treatment of patients.

TEXT

Laboratory contamination of clinical specimens can lead to erroneous microbiology reports which can adversely affect patient management (1). Early recognition of such pseudo-outbreaks is important for clinical microbiology laboratories. In 2010, five isolates of Tsukamurella were isolated from the tissue specimens of four patients within 10 days in our clinical microbiology laboratory. Outbreak investigation revealed that the recovery of Tsukamurella was due to laboratory contamination during processing of the tissue samples. In this article, we report the investigation of this pseudo-outbreak and characterization of the outbreak strains using a polyphasic approach.

The index patient was a 28-year-old female veterinarian, admitted for fever of unknown origin with right axillary lymphadenopathy (Table 1, patient 1). A direct Gram smear of the right axillary lymph node biopsy specimen in our laboratory revealed Gram-positive bacilli. A lymph node biopsy specimen was culture positive for Tsukamurella. However, the histological section, processed by the pathology laboratory, showed coalescent epithelioid cell granulomas, and no organisms were revealed by Gram staining, Ziehl-Neelsen staining, periodic acid-Schiff diastase staining, or Grocott staining. Bartonella henselae serology showed a >4-fold rise in IgG level for sera collected 6 days apart. PCR amplification and sequencing for the 16S rRNA gene of B. henselae was also positive on the right axillary lymph node. Since the clinical syndrome was compatible with cat scratch disease, there was a suspicion of contamination of the lymph node biopsy specimen. Contamination at the level of specimen processing in the microbiology laboratory was suspected because it was seen in the direct Gram smear that was processed in the microbiology laboratory but not in the biopsy specimens processed in the pathology laboratory.

Table 1.

Line listing of patients with clinical specimens positive for Tsukamurella

Patient no.	Sexa/age (yr)	Date(s) of specimen collection	Specialty	Specimen type	Clinical syndrome	Histopathological finding	Other microorganism(s) recovered
1	F/28	Sep 17	Medicine	Right axillary lymph node biopsy	Cat scratch disease	Granulomatous reaction; Gram stain and Ziehl-Neelsen stain did not reveal any organisms	Bartonella henselae
2	M/63	Sep 17 and Sep 24	Orthopedics	Left plantar granulation tissue	Chronic osteomyelitis	Chronic osteomyelitisb	Diphtheroid bacilli
3	F/56	Sep 24	Surgery	Lung tissue	Left lower lobe nodule	Adenocarcinomab	No growth
4	F/24	Sep 26	Intensive care unit	Peritoneal tissue	Peritonitis	Edematous fibroblastic tissue, with scattered neutrophils and lymphocytesb	Klebsiella pneumoniae, Escherichia coli, Enterococcus gallinarum

^aF, female; M, male.

^bGram stain and Ziehl-Neelsen staining were not performed.

Within a period of 9 days, 4 additional tissue specimens from 3 patients also grew Tsukamurella. A direct Gram smear of these 4 specimens performed in the clinical microbiology laboratory was negative. Line listing was performed for the 4 cases as described previously (Table 1) (2). A review of our laboratory records revealed that in the 24-month period before the outbreak, only 3 cases of Tsukamurella were reported from our laboratory and all were isolated from eye swabs (Fig. 1). Since the overall clinical histories of the 4 patients were not typical for Tsukamurella infections, with a sudden surge in the number of Tsukamurella strains isolated in a period of 10 days, an outbreak due to laboratory contamination was suspected.

Fig 1.

Number of clinical samples culture positive for Tsukamurella species from September 2008 to December 2010.

On the 3 days in which Tsukamurella was isolated, there were a total of 254 miscellaneous specimens (specimens other than blood, cerebrospinal fluid, respiratory tract specimens, urine, or stool) submitted to our laboratory for bacterial culture. Of 17 tissue specimens, 5 (29.4%) were culture positive for Tsukamurella, while none of the 237 nontissue specimens were positive (P < 0.001). A case-control study confirmed that only tissue specimens were significantly associated with the isolation of Tsukamurella (P < 0.001) (data not shown).

Procedures for processing of tissue samples were reviewed. In our laboratory, tissue specimens were minced with sterile surgical scissors in sterile saline, which would then be transferred to a glass slide for staining and the appropriate medium for culture. This step was not necessary for body fluids or swabs. A detailed examination of the procedures found that the scissors used in cutting the specimens were not properly wrapped before autoclaving (Fig. 2A). This procedure was quickly rectified and the responsible laboratory staff retrained, and there were no more cases of laboratory contaminations (Fig. 2B). Since laboratory contamination was suspected before the laboratory reports were issued to the clinician, none of the patients received unnecessary treatment.

Fig 2.

(A) Improper wrapping of scissors leading to pseudo-outbreak. (B) Proper wrapping of scissors after rectification of the procedure.

The four Tsukamurella isolates recovered from the tissue samples of patients 2, 3, and 4 had the same biochemical profile, while that from patient 1 differed from the others by the assimilation of l-fucose and d-arabitol but not maltose (data not shown). The biochemical profiles of all the five Tsukamurella isolates did not match the profile of any of the six known human-pathogenic species (3–7). Sequencing of the 16S rRNA genes of the five Tsukamurella isolates was performed as described previously (8, 9) and showed that they were identical, with no base difference from that of Tsukamurella tyrosinosolvens DSM 44234 (GenBank accession no. AY238514).

Pulsed-field gel electrophoresis (PFGE) showed that except for the Tsukamurella isolate recovered from patient 1 (lane 1), all of the other four isolates (lanes 2 to 5), recovered from the tissue samples of patients 2, 3, and 4, had the same PFGE patterns (Fig. 3). The extracellular metabolite contents of Tsukamurella isolates were examined by ultrahigh-performance liquid chromatography–electrospray ionization–quadruple time of flight mass spectrometry (UHPLC-ESI-Q-TOF-MS) analysis. In principal component analysis (PCA), 86.27% of the total variance in the data was represented by the first three principal components (Fig. 4A). The 3D-PCA score plot revealed that the five patient isolates are closely related and can be distinguished from the other two control isolates based on the first three principal components, with the five isolates clearly separated from control isolate 6 along PC1, which represented 36.12% of the variance, and also well separated from control isolate 7 on PC2, which represented 28.75% of the variance (Fig. 4A). A hierarchical clustering dendrogram also indicated the close grouping of the five patient isolates and their separation from the other two control Tsukamurella strains (Fig. 4B). The metabolic profiles of isolates 2 to 5 were clustered randomly, whereas that of isolate 1 was clustered to one side, suggesting that metabolic profiling could distinguish isolate 1 from isolates 2 to 5, concurring with the results of PFGE.

Fig 3.

PFGE of XbaI-digested genomic DNA of the five Tsukamurella isolates of the present study and the two control isolates of T. tyrosinosolvens recovered from patients with conjunctivitis. M, lambda ladder marker; 1, isolate from patient 1; 2, first isolate from patient 2; 3, second isolate from patient 2; 4, isolate from patient 3; 5, isolate from patient 4; 6 and 7, control isolates of T. tyrosinosolvens recovered from patients with conjunctivitis.

Fig 4.

(A) Principal component analysis and (B) hierarchical clustering dendrogram of metabolic footprints of the five Tsukamurella isolates of the present study and two control isolates of T. tyrosinosolvens recovered from patients with conjunctivitis based on positive ionization mode. A total of 4,431 molecular features (defined by retention time and mass pair) were included in the analysis. The 3D-PCA score plot for component PC1 versus component PC2 versus component PC3 was presented. The percentage of variance in the data set reflected by the first three PCs for each sample is shown. Hierarchical clustering of different Tsukamurella isolates was represented by heat map and dendrogram. Clustering was performed using Euclidean distance and average linkage. Blue and red indicate a negative and positive log ratio, respectively. Three biological and two technical replicate experiments were performed for each sample. 1, isolate from patient 1; 2, first isolate from patient 2; 3, second isolate from patient 2; 4, isolate from patient 3; 5, isolate from patient 4; 6 and 7, control isolates of T. tyrosinosolvens recovered from patients with conjunctivitis.

The possibility of laboratory contamination was first suspected when Tsukamurella was isolated from the tissue sample of the index patient who had cat scratch disease. A pseudo-outbreak was further suspected when additional strains of Tsukamurella were isolated from the tissue samples of three additional patients. Among these three additional cases, patients 2 and 4 had chronic osteomyelitis and primary peritonitis, respectively, which had never been reported to be associated with Tsukamurella infection. Phenotypic identification also revealed that all five strains of Tsukamurella had biochemical profiles that were not compatible with any of the Tsukamurella species associated with human infections. Line listing and case-control study confirmed that the pseudo-outbreak was associated with tissue specimens, and a review of the laboratory procedures revealed that improper wrapping of the scissors used for mincing the tissue specimens was the source of the contamination, rectification of which led to a prompt stop of the pseudo-outbreak. This explains why a Gram stain of the histological section of the lymph node did not show any bacteria, because the histological section was not processed by the contaminated scissors, but a direct Gram smear of the lymph node biopsy specimen showed Gram-positive bacilli, as the Gram smear was performed after the tissue was processed by the contaminated scissors. In fact, this is the first report of pseudo-outbreak due to contaminated scissors, although scissors carried by nurses have been found to be frequently contaminated by microorganisms (10). In the only Tsukamurella pseudo-outbreak reported in the literature, the source of the contamination was traced to a reused bottle of saline solution, when saline from the bottle was added to the tissue specimens before grinding (11).

Environmental contamination of medical and laboratory equipment resulting in outbreaks and pseudo-outbreaks is often due to multiple species/strains of microorganisms (12, 13). This is in contrast to the situation in outbreaks and pseudo-outbreaks due to point sources, where a single strain of microbe is often involved (14). In the present pseudo-outbreak, the four Tsukamurella strains isolated from patients 2 to 4 had identical biochemical profiles, indicating that they were likely to be the same species. Further typing by PFGE and UHPLC-ESI-Q-TOF-MS also showed that these four strains were identical, indicating that they were also the same strain. In fact, this is the first Tsukamurella pseudo-outbreak that involved more than one Tsukamurella species/strain (11).

Early recognition of pseudo-outbreaks is as important as early recognition of outbreaks, as it prevents unnecessary and incorrect treatment of patients (1). This pseudo-outbreak also highlights the role of the clinical microbiologist, in that correlation with clinical information and histological information is important when facing unusual culture results. As Tsukamurella can be a genuine pathogen or a contaminant, a high index of suspicion is required to assess its role in every case of isolation.