Use of Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry To Resolve Complex Clinical Cases of Patients with Recurrent Bacteremias

Abstract

Matrix-assisted laser desorption–ionization time of flight mass spectrometry (MALDI-TOF MS) is a rapid and accurate method of identifying microorganisms. Throughout Europe, it is already in routine use but has not yet been widely implemented in the United States, pending FDA approval. Here, we describe two medically complex patients at a large tertiary-care academic medical center with recurring bacteremias caused by distinct but related species. Bacterial identifications were initially obtained using the Vitek-2 system with the GPI card for Enterococcus and the API system for staphylococci. Initial results misled clinicians as to the source and proper management of these patients. Retrospective investigation with MALDI-TOF MS clarified the diagnosis by identifying a single microorganism as the pathogen in each case. To our knowledge, this is one of the first reports in the United States demonstrating the use of MALDI-TOF MS to facilitate the clinical diagnosis in patients with recurrent bacteremias of unclear source.

CASE REPORTS

Case 1.

Case 1 was a 35-year-old woman from the Dominican Republic with a history of ulcerative colitis diagnosed in 2003, a left humerus fracture in 2007 after open reduction internal fixation (ORIF) with a plate and screw device, a mitral valve prolapse with severe mitral regurgitation due to rheumatic heart disease at age 13, New York Heart Association (NYHA) class II heart failure, and a history of native valve infective endocarditis in 2009. She was admitted initially in September 2010, with multiple sets of blood cultures positive for Enterococcus faecalis on days 1 and 2 and Enterococcus gallinarum on day 4. She was treated with 4 weeks of intravenous vancomycin after clearance of peripheral blood cultures.

She was readmitted in November 2010 with Enterococcus gallinarum bacteremia on days 1, 3, and 7, for which she completed a 6-week course of intravenous ampicillin and streptomycin. She then presented in May 2011 with 1 week of fever up to 103°F at home, shaking chills, and malaise. Initial examination revealed a blood pressure of 102/66 with a pulse of 84 beats per minute, 18 respirations per minute, and a temperature of 101.6°F. Physical examination was notable for a grade 4/6 nearly holosystolic murmur over the apex, radiating to the axilla and back. Abdominal exam was normal. There were no clinical signs of infection in her left shoulder orthopedic hardware. Laboratory data revealed a peripheral leukocyte count of 7,900/μl, with 78% neutrophils. Her hemoglobin and hematocrit were 11.5 g/dl and 34.7%, respectively. Her electrolytes and renal function tests were all within normal limits. Multiple sets of blood cultures from an outpatient visit prior to admission and those obtained 4 days later on admission were positive for Enterococcus gallinarum susceptible to ampicillin (MIC = 1) and vancomycin (MIC = 2), with high-level synergy to streptomycin but not to gentamicin.

On day 2 of this admission, a third blood culture was reported as positive for Enterococcus durans but, on retesting, was identified as E. gallinarum. An infectious disease (ID) specialist was consulted, and the patient was treated with intravenous ampicillin and streptomycin for 6 weeks. Follow-up blood cultures were negative. At this juncture in the patient's hospital course, the source of her recurrent infection was unclear. A work-up by computed tomography (CT) of her thorax, abdomen, and pelvis and a gallium scan did not reveal an obvious nidus, such as an abscess or mycotic aneurysm. A transesophageal echocardiogram (TEE) showed severe mitral regurgitation but no intracardiac masses, thrombi, or vegetations. Auto-inoculation was ruled out by patient history.

The clinical suspicion of the treating ID specialist was that the recurrent enterococcal infection was caused by a common strain despite varied results obtained using the Vitek-2 system. To investigate this further, molecular testing was performed using matrix-assisted laser desorption–ionization time of flight mass spectrometry (MALDI-TOF MS) (Bruker MALDI Biotyper system) on 5 isolates, pulsed-field gel electrophoresis (PGFE) on 4 isolates (New York City Department of Health Laboratory), and 16S ribosomal DNA sequencing on 3 isolates as the gold standard (MIDI Labs, Inc.). PFGE showed that the 4 strains tested were closely related. MALDI-TOF MS demonstrated that all strains had the best match with Enterococcus faecalis (MALDI scores for 4 of 5 specimens were above 2.0, and 1 was 1.849). 16S ribosomal DNA sequencing revealed a species match with E. faecalis (Table 1). Thorough clinical investigation, 38 different blood cultures, and multiple imaging tests had failed to reveal a consistent microorganism or definitive source of infection. Due to the complexity of the patient's medical history, multiple potential sources of infection were considered. A TEE was obtained, given her history of rheumatic heart disease, severe mitral regurgitation, and a prior episode of infective endocarditis. Negative results effectively ruled out a new episode of endocarditis as the source of persistent bacteremia. A gallium scan did not reveal an infected shoulder prosthesis. A CT scan of the chest, abdomen, and pelvis did not reveal an occult source of infection. Nosocomial acquisition of infection was thought to be less likely since the patient did not have a central venous catheter, indwelling urinary catheter, or any positive urine cultures for Enterococcus. Finally, E. faecalis identification obtained by MALDI-TOF MS led to proper source identification and clinical management of the patient. As a commensal of the human gastrointestinal tract, seeding of the bloodstream with E. faecalis during active flair-ups of the patient's inflammatory bowel disease was ultimately implicated as the source of recurrent infection. This was not clinically apparent at first, since the patient had a normal abdominal exam at the time of presentation. Aggressive treatment of her underlying illness led to cessation of her episodic bacteremia, as documented by surveillance blood cultures obtained several months after discharge.

Table 1.

Summary of results for case 1

Date of culture	Vitek identification	Organism best match by MALDI-TOF MS	16S ribosomal DNA sequencing result
September 2010	E. faecalis	E. faecalis	E. faecalis
November 2010	E. gallinarum	E. faecalis	E. faecalis
April 2011	E. gallinarum	E. faecalis	Not done
May 2011	E. durans	E. faecalis	E. faecalis
May 2011	E. gallinarum	E. faecalis	Not done

Case 2.

Case 2 was a 24-year-old male with a history of familial sudden death syndrome, nonobstructive hypertrophic cardiomyopathy, and inducible ventricular tachycardia who had an automated implantable cardioverter-defibrillator (AICD) placed in 2002, which was complicated by AICD pocket infection in June 2011 that did not improve with outpatient antibiotics. Subsequent device extraction was performed in August 2011, and wound culture from the pocket grew coagulase-negative Staphylococcus on subculture. Unfortunately, a portion of the original right ventricular lead was retained despite multiple attempts to remove it from deep within the myocardium. The patient then presented to Montefiore Medical Center in January 2012 with headache, fever, and chills. His temperature was 102.8°F, blood pressure was 117/60 mm Hg, pulse was 75 beats per minute, respiratory rate was 18 breaths per minute, and pulse oximetry was 100%. Physical exam was unremarkable. His leukocyte count was 13,000/μl with 70% neutrophils. Renal function tests, electrolytes, and liver enzymes were all normal. A chest radiograph was negative for infiltrates, and a CT of the head showed no evidence of sinus disease. Multiple sets of blood cultures showed mixed infection with Staphylococcus capitis and Kocuria varians, both of which are susceptible to oxacillin, erythromycin, gentamicin, and moxifloxacin but resistant to penicillin. A transesophageal echocardiogram revealed a mobile echodensity on the abandoned right ventricular lead. An ID consultation was obtained, and the patient was treated with at least 6 weeks of cefazolin and rifampin. Multiple sets of outpatient surveillance blood cultures were negative. However, his fever to 101.7°F recurred approximately 3 weeks after discontinuation of antibiotics. Blood cultures repeated in April 2012 obtained in the outpatient clinic were again positive for Staphylococcus auricularis, as well as Kocuria varians, both of which are sensitive to oxacillin, erythromycin, gentamicin, and moxifloxacin but resistant to penicillin. The patient was readmitted for intravenous antibiotics, and subsequent blood cultures on day 1 and day 4 were also positive for S. auricularis. After clearance of peripheral blood cultures on day 5, the patient underwent thorascopic lead extraction on day 8, and subcultures from the lead grew S. capitis. Perioperative TEE did not reveal any vegetations. Postoperatively, the patient completed 6 weeks of a continuous intravenous oxacillin infusion and had multiple negative surveillance blood cultures. Due to heterogeneous results generated by the Vitek-2 system, multiple clinical isolates were retrospectively sent for additional molecular testing in June of 2012. MALDI-TOF MS analysis (Bruker MALDI Biotyper system) was performed on five isolates from blood cultures drawn on separate dates. All five were identified as Staphylococcus capitis (MALDI scores for all 5 isolates were >2.00). 16S ribosomal DNA analysis of 3 isolates (MIDI Labs, Inc.) showed an exact match with S. capitis (Table 2). As in the previous case, results obtained by MALDI-TOF MS confirmed the clinical suspicion that there was a solitary pathogen causing recurrent infection originating from a retained portion of the intracardiac device. An AICD was successfully reimplanted on 23 July 2012, and the patient has not had further episodes of fever or bacteremia.

Table 2.

Summary of results for case 2

Date of culture (no. of blood cultures)	Vitek identification	Organism best match by MALDI-TOF MS	16S ribosomal DNA sequencing result
January 2012 (1)	Coagulase-negative Staphylococcus	Staphylococcus capitis	Staphylococcus capitis
January 2012 (4)	Staphylococcus capitis and Kocuria varians, Staphylococcus capitis, Staphylococcus capitis, Staphylococcus capitis,	Staphylococcus capitis	Not done
January 2012 (1)	Staphylococcus capitis	Staphylococcus capitis	Staphylococcus capitis
April 2012 (2)	Staphylococcus auricularis and Kocuria varians (for both cultures)	Staphylococcus capitis	Not done
April 2012 (1)	Staphylococcus auricularis	Staphylococcus capitis	Staphylococcus capitis

Accurate and rapid identification of microorganisms ensures appropriate patient management. Here, we describe two complex clinical cases in which conventional analysis yielded perplexing results and prolonged the uncertainty of the diagnosis and management of patients with recurrent bacteremias. This report underscores the advantage of applying MALDI-TOF MS in the real-time management of patients. It also sheds light on the major limitations of routine phenotypic methods, despite results of prior studies that have demonstrated the accuracy of Enterococcus faecalis identification using the Vitek system GPI card (1). Perhaps the accuracy of the GPI card was affected by host factors, such as multiple prior antibiotic regimens. The clinical suspicion of the ID consultant coupled with further analysis by MALDI-TOF MS finally established the diagnosis of the patients described above.

Compared to existing culture techniques, MALDI-TOF MS is considered revolutionary due to its accuracy, speed, and cost-effectiveness (2–7). A single system can be used for a range of microorganisms, including bacteria, yeasts, and fungi, obtained from different body sites. Accurate pathogen identification is possible within hours of its isolation, expediting appropriate antimicrobial use and preventing excessive use of empirical antibiotics. Prior retrospective analyses have demonstrated the success of MALDI-TOF in comparison to existing techniques when used on reference specimens that were not obtained directly from the patient (8, 9). The MALDI-TOF database used for identification of the organisms from our patients is the same as the current Bruker systems (2013 models) and compares 8 reference strains of E. faecalis and 6 of S. capitis to members of their respective genera. In 2009, Seng et al. demonstrated its effectiveness in clinical isolates from multiple body sites studied prospectively over 16 weeks. Using the Bruker system, 95.4% of 1,160 bacterial isolates were correctly identified, 84.1% at the species level and 11.3% at the genus level. No identification occurred in 2.8% of isolates, and errors in identification occurred in 1.7% of isolates; both circumstances were attributed to human error (improper database entry). Accurate identification in this study was significantly correlated with having 10 or more reference spectra in the database (6). Similarly, van Veen et al. prospectively analyzed 980 clinical isolates of yeasts and bacteria and demonstrated that >95% of bacteria could be identified to a genus and species level. Eighty-five percent of yeasts across 12 different species were correctly identified in this study (7). Species-level identification is also achievable for a variety of Staphylococcus species by MALDI-TOF MS when it is used in conjunction with a broad reference database and retains only conserved peaks with a m/z ratio above 0.1 (3, 9). Spanu et al. identified 100% of S. aureus and 99.1% of coagulase-negative staphylococcus isolates from 450 clinical specimens of bloodstream infections using MALDI-TOF MS in a comparison with rpoB gene sequencing as a reference standard, suggesting that expensive DNA sequencing for accurate identification of staphylococcal species is unnecessary (10). As demonstrated in our study and several others, MALDI-TOF MS can provide accurate identification of staphylococci when conventional phenotypic identification fails. In 2010, Bizzini et al. prospectively analyzed 1,371 isolates of bacteria and yeasts by MALDI-TOF MS and conventional techniques. Ninety-three percent of these were identified to the species level and 5.3% to the genus level. Only 1.5% could not be identified by MALDI-TOF MS. Discordant results occurred with 63 isolates (4.9%), of which 56 were attributed to errors related to taxonomic differences in the spectral database. The study also revealed several misidentifications due to the inability of MALDI-TOF MS to distinguish bacterial species with only subtle differences in their ribosomal proteins. Two species of Shigella were misidentified as Escherichia coli, and Streptococcus pneumoniae was misidentified as Streptococcus mitis (11).

As prior studies have demonstrated, the database of reference spectra utilized by MALDI-TOF MS is not all-inclusive and has limitations for accurate identification of several clinically relevant organisms. Species that do not differ significantly in their ribosomal protein sequences remain difficult to distinguish, and therefore conventional biochemical techniques are still necessary for identification (11, 12). However, the reference database is not static and can be expanded regularly to fill in current gaps in identification. The ability of MALDI-TOF MS to detect bacterial resistance mechanisms in methicillin-resistant S. aureus (MRSA) and Enterobacteriaceae was initially thought to be limited (12, 13). More recently, Sparbier and colleagues demonstrated that beta-lactamase activity can be easily detected based on the +18-Da mass shift produced by hydrolysis of the beta-lactam ring and, in many cases, the −44-Da mass shift of the hydrolyzed product. Sensitivity and resistance patterns were detected within hours of incubation of the bacteria with a given antibiotic and were identical to those obtained by routine methods, such as the Etest (14).

The clinical implications of such technology for the rapid diagnosis and management of hospital-acquired infections are vast (14, 15). This was very recently demonstrated by Perez and colleagues at the Houston Methodist Hospital, where utilization of MALDI-TOF MS in concert with antimicrobial stewardship led to prompt identification and susceptibility testing in a group of patients with Gram-negative bloodstream infections. Results demonstrated significant reductions in time to optimal therapy, hospital lengths of stay, and total costs, estimated at $30 million annually (16).

In conclusion, MALDI-TOF MS, compared with currently employed methods, has widespread clinical applications owing to its convenience, accuracy, and cost-effectiveness when multiple clinical specimens are processed per day (9, 15, 16). Use of MALDI-TOF MS for identification of microbial pathogens will lead to improvements in patient outcomes, lengths of hospitalization, and meaningful use of antibiotics.