Evaluation of Matrix-Assisted Laser Desorption Ionization–Time of Flight Mass Spectrometry for Identification of Vibrio cholerae

Abstract

We evaluated the use of matrix-assisted laser desorption ionization–time of flight mass spectrometry (MS) for the identification of Vibrio cholerae. MS identified all 42 isolates of V. cholerae O1 and O139 and 7 of 9 non-O1/O139 isolates. MS correctly discriminated between all Aeromonas and V. cholerae isolates. Overall, MS performed as well as or better than biochemical methods.

TEXT

Vibrio cholerae is unique among waterborne bacterial pathogens in its ability to cause global pandemics of disease (1). V. cholerae serogroup O1 is the predominant cause of cholera globally, although previous epidemics have also been caused by serogroup O139 (1). Some non-O1/O139 strains may possess cholera toxin and cause sporadic cases of severe dehydrating diarrheal illness, while others do not produce cholera toxin and are associated with sporadic cases of gastroenteritis and occasionally sepsis (1).

Given the importance of V. cholerae as a cause of severe diarrheal illness globally, its potential for rapid epidemic spread, and its listing as a biothreat level B agent (http://www.bt.cdc.gov/agent/agentlist-category.asp, accessed 22 October 2014), it is important for clinical microbiology laboratories to be able to correctly identify it and differentiate it from other bacterial species. In particular, differentiating V. cholerae from aeromonads, which are also stool pathogens, can be challenging. Vibrio, but not Aeromonas, species generally grow in saline and on thiosulfate citrate bile salt sucrose agar (TCBS) and are susceptible to the vibriostatic compound O/129. However, certain isolates of V. cholerae are known to be resistant to O/129 (2), some Aeromonas isolates are sensitive to vibriostatic compounds, strains of A. caviae occasionally grow on TCBS (3), and some Aeromonas isolates grow in saline (4). Thus, more thorough biochemical testing is typically required for definitive identification. In places such as the United States, where V. cholerae is rarely isolated, a clinical laboratory may rely on the results of a standardized biochemical identification system. Unfortunately, these commonly used identification systems have previously been shown to perform poorly with Vibrio species (5–7). In one study, the accuracy of identification of V. cholerae was 50% with the API 20E system and 67% with Vitek GNI+ (6). Others have similarly reported that these systems perform only slightly better with Aeromonas species, with 77% accuracy for API 20E and 83% accuracy for the Vitek 2 ID-GN card (4). Specifically, one report noted that an Aeromonas isolate was misidentified as V. cholerae (5). Given that matrix-assisted laser desorption ionization–time of flight (MALDI-TOF) mass spectrometry (MS) is increasingly being used in the clinical microbiology laboratory for identification of bacteria, we compared the accuracy of this method with that of two biochemical identification systems, Vitek 2 GN and API 20E, in the identification of V. cholerae.

Fifty-one clinical isolates presumed to be V. cholerae and 26 presumed to be Aeromonas species were obtained from laboratories in Bangladesh (44 isolates of Vibrio species obtained between 2000 and 2013 and 10 isolates of Aeromonas sp. from the International Centre for Diarrhoeal Disease Research), Haiti (7 isolates of Vibrio sp. from Le Hôpital St. Marc), and the United States (8 isolates of Aeromonas sp. from Massachusetts General Hospital, Boston, and 8 isolates of Aeromonas sp. from Barnes-Jewish Hospital, St. Louis, MO). Each contributing laboratory identified the isolates according to their own internal protocol. Multiple isolates obtained from the same patient were not included in this study. Serotyping of V. cholerae isolates was performed by slide agglutination at the laboratory that originally obtained the isolate with polyclonal O1 and O139 antisera, followed by serotype-specific (Ogawa and Inaba) antisera for all serogroup O1 isolates. Isolates that did not group in O1 or O139 were labeled as non-O1/O139. Among the V. cholerae isolates, 27 were serogroup O1 (including 17 Ogawa and 10 Inaba isolates), 15 were serogroup O139, and 9 were non-O1/O139 serogroup isolates. All isolates were shipped to the Massachusetts General Hospital and stored at −70°C prior to testing. All subsequent analysis, described below, was performed at the Massachusetts General Hospital.

The identification of all of the V. cholerae isolates was verified by sequencing of the rpoB gene. Among the 26 presumed Aeromonas isolates, 23 were verified by rpoB sequencing as Aeromonas species, including Aeromonas hydrophila (n = 6), A. caviae (n = 3), A. veronii (n = 8), A. trota (n = 4), A. taiwanensis (n = 1), and one that could not be reliably identified to the species level. The remaining three isolates were identified as Vibrio fluvialis (n = 1) and Shewanella species (n = 2). In all cases, crude lysates were amplified and sequenced with primers 5′-GTAGAAATCTACCGCATGATG-3′ and 3′-ACCGCCTGACGTTGCATGTT-5′. The resulting sequences were aligned and edited by using ChromasPro (Technelysium). The edited sequences were used to search GenBank and Bio Informatic Bacteria Identification (BIBI) to provide the definitive identification.

After an initial culture on Trypticase soy agar with 5% sheep blood (BAP; Becton Dickinson and Company, Sparks, MD), a single colony of each isolate was subcultured onto BAP, MacConkey II agar (MAC; Becton Dickinson and Company), and TCBS agar (Becton Dickinson and Company) to assess colony morphology and perform all subsequent testing. Colonies isolated from the MAC plate were tested on the API 20E (bioMérieux, Durham, NC) and Vitek 2 GN (bioMérieux) biochemical testing platforms, as well as the Vitek MS v2 MALDI-TOF MS platform with Knowledgebase 2.0 (bioMérieux). All three testing platforms have been validated and cleared for clinical use by the U.S. FDA. Testing was performed in accordance with the manufacturer's package insert to simulate clinical testing. Colonies isolated from BAP were used to perform a spot oxidase test (Becton Dickinson and Company) and Gram staining. A colony isolated from TCBS was also tested with Vitek MS.

Among the V. cholerae isolates, all appeared as curved Gram-negative rods upon Gram staining and grew as convex yellow colonies on TCBS agar. Colony morphologies on BAP and MAC were consistent with non-lactose-fermenting Gram-negative rods for the majority of the isolates, although eight were beta-hemolytic on BAP and four appeared as lactose fermenting on MAC. Biochemical testing indicated that 30 of the 51 Vibrio isolates were O/129 resistant, including all of the O1 isolates from both Haiti and Bangladesh, while the O139 isolates were sensitive. These results are consistent with what is known about O/129-resistant isolates from Bangladesh and Haiti during the time period in which these isolates were obtained. All of the verified Aeromonas isolates appeared as Gram-negative rods upon Gram staining and as non-lactose fermenting on MAC and failed to grow on TCBS.

As shown in Table 1, Vitek MS gave a single correct result for 49 (96%) of the V. cholerae isolates tested from MAC. Notably, it was 100% accurate for the O1 and O139 serogroups of V. cholerae (the causative agents of epidemic cholera) but only 78% accurate for the non-O1/O139 strains (which cause sporadic disease). The same isolates from TCBS medium tested by Vitek MS were not as likely to be identified (P < 0.0001, Fisher exact test). The only exceptions were isolates of serogroup O1 Inaba, which were all correctly identified regardless of the agar they were grown on. As shown in Table 2, none of the V. cholerae isolates were misidentified by Vitek MS; rather, no identification or a low-discrimination result of V. cholerae and V. mimicus was reported. API 20E correctly identified 48 of the 51 V. cholerae isolates (Table 1), including all O1 and O139 strains. Of the nine non-O1/O139 strains, seven were correctly identified while two produced a low-discrimination result, including V. cholerae and V. alginolyticus. Vitek 2 GN correctly identified 45 of the 51 V. cholerae isolates (Table 1). Unlike the other methods, misidentification occurred for the O1, O139, and non-O1/O139 serogroups. Among these were low-discrimination results that included A. sobria and V. cholerae, one incorrect identification as A. sobria, and two isolates that could not be identified.

TABLE 1.

Numbers of isolates identified correctly by each method

Reference IDa (total no. of isolates)	No. (%) of isolates identified correctly by:
	API 20E	Vitek 2 GN	Vitek MS MAC	Vitek MS TCBS
V. cholerae (51)	48 (94)	45 (88)	49 (96)	26 (51)
O1 Ogawa (17)	17 (100)	15 (88)	17 (100)	9 (53)
O1 Inaba (10)	10 (100)	9 (90)	10 (100)	10 (100)
O139 (15)	15 (100)	14 (93)	15 (100)	5 (33)
Non-O1/O139 (9)	7 (78)	7 (78)	7 (78)	2 (22)
Aeromonas sp.b (21)	20 (95)	21 (100)	21 (100)	NA
A. hydrophila/caviae (9)	0 (0)	9 (100)	9 (100)	NA
A. veronii/sobria (12)	0 (0)	11 (92)	6 (50)	NA

^aReference identification was done by sequencing. For V. cholerae, serogrouping was performed at the clinical laboratory that originally obtained the isolate.

^bIdentification as an Aeromonas species was considered correct if the test method gave a single correct result that matched the reference identification or a low-discrimination result that included only the Aeromonas species complex.

TABLE 2.

Misidentifications by method

Reference ID	Test method	Result(s) (no. of isolates)
V. cholerae O1 Ogawa	Vitek 2 GN	A. sobria/V. cholerae (2)
	Vitek MS TCBS	No ID (6), V. cholerae/V. mimicus (2)
V. cholerae O1 Inaba	Vitek 2 GN	A. sobria/V. cholerae (1)
V. cholerae O139	Vitek 2 GN	A. sobria (1)
	Vitek MS TCBS	No ID (10)
V. cholerae non-O1/O139	API 20E	V. alginolyticus/V. cholerae (2)
	Vitek 2 GN	Unidentified (2)
	Vitek MS MAC	No ID (1), V. cholerae/V. mimicus (1)
	Vitek MS TCBS	No ID (5), V. cholerae/V. mimicus (2)
Aeromonas species	API 20E	Unacceptable profile, Aeromonas sp./V. cholerae (1)

Vitek MS and Vitek 2 GN correctly identified all of the Aeromonas isolates to the genus level. API 20E identified 20 of the 23 Aeromonas isolates correctly to the genus level. The remaining isolate could not be definitively identified to the genus level, giving a split between Aeromonas species and V. cholerae. All three identification methods failed to discriminate accurately among the Aeromonas species and often gave low-discrimination results. For the majority of the isolates, the Vitek MS and Vitek 2 GN results included only members within the same complex (Table 1). API 20E was unable to differentiate between the two complexes. The one isolate of V. fluvialis and the two Shewanella isolates were correctly identified by Vitek MS and Vitek 2 GN but could not be identified by API 20E (data not shown).

In aggregate, Vitek MS was the most accurate of the methods, correctly discriminating between V. cholerae and Aeromonas species with an accuracy of 97%, compared to 92% for Vitek 2 GN, and 94% for API 20E. Although the difference in accuracy among the methods was not statistically significant (P > 0.5, Friedman's test), the lack of errors in the identification of the O1 and O139 serogroups of V. cholerae and correct identification of all Aeromonas isolates to the genus level do suggest that the Vitek MS may be the superior method for identification, although the current cost of this technology will limit its use in resource-limited settings, where V. cholerae infection commonly occurs.

The one striking limitation of Vitek MS was the relative inability to identify V. cholerae cultured on TCBS. The isolates that were not readily identified from TCBS plates tended to be sticky and more challenging to spot onto the target slide, which may explain their lack of identification. In general, mucoid, dry, and sticky colonies can be challenging to spot onto the target slide and occasionally require multiple attempts to get a definitive identification. An alternative explanation for the lack of identification from TCBS may be that differences in protein expression between isolates grown on TCBS and those grown on MAC result in changes in the mass spectra such that the isolate can no longer be identified with the current database and algorithms. Additional studies need to be performed to rule this out.

This study provides additional data supporting MALDI-TOF MS as an accurate method for the identification of bacteria in the clinical microbiology laboratory. In addition to being the largest study to date of the identification of V. cholerae (4–9), we found that the API 20E and Vitek 2 GN methods performed better than expected on the basis of the available literature. This may be due to reformulation of the reagents or other changes in these identification systems since similar studies were last performed. We were also able to show that the accuracy of these identification methods does appear to vary between the O1/O139 and non-O1/O139 serogroups, a point that had not been previously reported. In conclusion, the accuracy of MALDI-TOF MS in the identification of V. cholerae O1/O139 and some other Vibrio species suggests that it is a versatile and robust tool for the rapid identification of V. cholerae and its discrimination from aeromonads.