MALDI-TOF MS: An alternative approach for ribotyping Clostridioides difficile isolates in Brazil

Abstract

Clostridioides difficile is an important organism causing healthcare-associated infections. It has been documented that specific strains caused multiple outbreaks globally, and patients infected with those strains are more likely to develop severe C. difficile infection (CDI). With the appearance of a variant strain, BI/NAP1 ribotype 027, responsible for several outbreaks and high mortality rates worldwide, the epidemiology of the CDI changed drastically in the United States, Europe, and some Latin American countries. Although the epidemic strain 027 was not yet detected in Brazil, there are ribotypes exclusively found in the country, such as, 131, 132, 133, 135, 142 and 143, which are responsible for outbreaks in Brazilian hospitals and nursing homes. Although PCR-ribotyping is the most used method in epidemiology studies of C. difficile, it is not available in Brazil. This study aimed to develop and validate an in-house database for detecting C. difficile ribotypes, usually involved in CDI in Brazilian hospitals, by using MALDI-TOF MS. A database with 19 different ribotypes, 13 with worldwide circulation and 6 Brazilian-restricted, was created based on 27 spectra readings of each ribotype. After BioNumerics analysis, neighbor-joining trees revealed that spectra were distributed in clusters according to ribotypes, showing that MALDI-TOF MS could discriminate all 19 ribotypes. Moreover, each ribotype showed a different profile with 42 biomarkers detected in total. Based on their intensity and occurrence, 13 biomarkers were chosen to compose ribotype-specific profiles, and in silico analysis showed that most of these biomarkers were uncharacterized proteins or well-conserved peptides, such as ribosomal proteins. A double-blind assessment using the 13 biomarkers correctly assigned the ribotype in 73% of the spectra analyzed, with 94%–100% of correct hits for 027 and for Brazilian ribotypes. Although further analyses are required, our results show that MALDI-TOF MS might be a reliable, fast and feasible alternative for epidemiological surveillance of C. difficile in Brazil.

1. Introduction

Symptoms of Clostridioides difficile infection (CDI) range from mild diarrhea to severe colonic inflammation, known as pseudo-membranous colitis (PMC) [1]. Severe CDI can require prolonged antibiotic therapy, such as metronidazole, fidaxomicin and vancomycin, in hospitalized patients [2], generating substantial mortality, morbidity and economic burden [3]. In the United States, for instance, more than 500,000 patients are infected annually and approximately 14,000 die [4].

Most strains involved in CDI are toxigenic, which is associated with the ability to produce two major toxins, TcdA, an enterotoxin, and TcdB, a potent cytotoxin [5]. In the early 2000s, the epidemiology of CDI drastically changed with the emergence of BI/NAP1/ribotype 027, an epidemic and hypervirulent strain. This strain caused numerous outbreaks and 29.000 deaths in North America (2000–2010), Canada and several countries in Europe (3700 deaths/year) [6,7]. More recently, other toxigenic C. difficile ribotypes, such as RT001, RT014 and RT078 [8], have also emerged as main causes of CDI. In Latin America, CDI cases associated with the epidemic strain RT027 were reported in Costa Rica [9], Panama [10], Chile [11] and Colombia [12]. A ribotype 027 strain susceptible to fluoroquinolones was detected in Argentina [13].

In Brazil, however, the epidemic ribotype 027 strain has never been reported and CDI epidemiology is still underexplored. This is partly because detection of anaerobic bacteria is not a routine procedure in clinical laboratories, mainly because of the lack of specialized technologies and facilities [14]. In addition, few research groups are working with C. difficile and most of the studies are held in hospitals in metropolitan areas [15,16]. Nevertheless, a number of C. difficile ribotypes involved in CDI cases have been detected in Brazilian hospitals, including RT014, RT043, RT046, RT106, RT132, RT133, RT134, RT135, RT142 and RT143 [14,17,18]. Some of those ribotypes were considered new, because they were not at the time in the PCR library created by Stubbs et al. [19] and among those, ribotype 133 was the most prevalent. Even more, the ribotypes 133, 135, 142 and 143 are exclusively found in Brazil [20].

The determination of C. difficile ribotypes is important to enable comparison of strains recovered in different places and assessment of the global epidemiology of CDI. Polymerase Chain Reaction (PCR) ribotyping remains the gold standard technique for this purpose. However, it is a time-consuming method (at least 48 h) and is performed only in laboratories with a vast strain collection for fingerprinting comparison, which most countries do not have, including Brazil. Consequently, the development of a simpler and faster method for gathering epidemiological data on C. difficile strains is urgently needed.

Matrix-Assisted Laser Desorption Ionization – Time-of-Flight Mass Spectrometry (MALDI-TOF MS) has become a widely used asset in clinical laboratories [21]. It discriminates microbial species and subspecies through mass spectra patterns derived from ribosomal and other conserved and abundant bacterial proteins (biomarkers) [22]. The process is fast (less than 12 h), sensitive and cost-effective. Besides genus and species identification, MALDI-TOF MS has been used for bacterial typing and prediction of antimicrobial sensitivity in clinically relevant bacterial species, including Streptococcus pneumoniae [23], Streptococcus pyogenes [24] and methicillin-resistant Staphylococcus aureus (MRSA) [25]. For C. difficile, however, MALDI-TOF MS has been used mainly for species identification [26], and its ability to discriminate among different C. difficile strains has been only recently assessed [27,28]. Our work aimed to develop a method to identify and type C. difficile ribotypes, usually involved in C. difficile infection (CDI) in Brazilian hospitals, by evaluating a MALDI-TOF MS typing scheme for predicting C. difficile ribotypes, including those prevalent in Brazil and worldwide.

2. 2-. Material and methods

2.1. Clostridioides difficile strains and clinical isolates

A total of 18 C. difficile human isolates recovered from 2001 to 2017 from fecal specimens of patients with CDI, including children and adults, representing 18 different ribotypes were included in this study. Six of them are ribotypes exclusively found in Brazil (131, 133, 135, 142, 143 and 233) (Table 1). All 18 isolates belong to the culture collection of the Laboratory Biologia de Anaeróbios at IMPG, UFRJ, Brazil. The epidemic strain R20291 (BI/NAP1/ribotype 027; CDC ANA #2004016) was also included in the study and was kindly given by Dr. Angela Thompson from CDC, Atlanta, USA (Table 1), totalizing 19 isolates. All strains were previously PCR-ribotyped [19] by Dr. Jon Brazier from the Anaerobe Reference Unit, Public Health Wales Microbiology, University Hospital of Wales, Cardiff, Wales, United Kingdom.

Table 1.

Clostridioides difficile strains used in this study depicting strain name, ribotype, antimicrobial susceptibility testing and toxin profile.

Strainsa	Ribotypes	ASTb			Toxin profilec
		Met	Van	Clin	A	B	CDT
CBA0206	001	S	S	S	+	+	−
CBA1125	010	S	S	R	−	−	−
630	012	S	S	S	+	+	−
CBA0822	014	S	S	R	+	+	−
CBA0251	015	S	S	S	+	+	−
HU24	020	S	S	S	+	+	−
R20291	027	S	S	S	+	+	+
CBA0168	031	S	S	I	−	−	−
CBA1129	038	S	S	R	−	−	−
CBA0025	043	S	S	N/A	+	+	−
CBA1127	046	S	S	S	+	+	−
CBA0166	060	S	S	I	−	−	−
CTI	106	S	S	R	+	+	−
CBA0165	131	S	S	S	−	−	−
HU09	133	S	S	S	−	+	−
CBA0178	135	S	S	S	+	+	−
CBA0202	142	S	S	I	+	+	−
CBA0204	143	S	S	S	+	+	−
CBA1130	233	S	S	S	+	+	−

^aStrains from the culture collection of the Laboratório de Biologia de Anaeróbios, Instituto de Microbiologia Paulo de Góes, Universidade Federal do Rio de Janeiro, (UFRJ), Rio de Janeiro, RJ, Brazil, and donated by Angela Thompson, Centers for Disease Control and Prevention, Atlanta, Georgia, USA (CDC ANA #2004016).

^bAntimicrobial susceptibility testing for vancomycin [44] (Van), clindamycin (Clin) and metronidazole (Met). S: susceptible; I: intermediate; R: resistant; N/A: not available.

^cPCR-multiplex for toxin profile [45]. A: toxin A (tcdA), B: toxin B (tcdB); CDT: binary toxin (cdtB).

Isolates were cultivated on supplemented (5 μg/mL hemin and 10 μg/mL menadione) blood agar (5% defibrinated sheep blood) plates overnight at 37 °C under anaerobic conditions. Colonies were collected for species confirmation by MALDI-TOF MS. For this purpose, a single colony of each strain was spotted in duplicate onto MALDI polished steel target plate, covered with 1 μL of a saturated solution of CHCA matrix (α-cyano-4-hydroxycinnamic acid; Bruker Daltonics, Bremen, Germany) in 50% acetonitrile and 2.5% tri-fluoroacetic acid; and allowed to dry at room temperature. Escherichia coli DH5α was used for mass calibration instrument and parameter optimization. Runs, taking up to an hour, were carried out on a Microflex LT MALDI-TOF MS system using Flex Control and Biotyper software (Bruker Daltonics) on default mode.

2.2. MALDI-TOF MS data acquisition

For MALDI-TOF MS typing, the following extraction protocol was performed, in accordance with the equipment manufacturer’s recommendations. All strains were cultivated on supplemented (5 μg/mL hemin and 10 μg/mL menadione) blood (5% defibrinated sheep blood) agar plates overnight at 37 °C under anaerobic conditions. For bacterial protein extraction, five bacterial colonies were homogenized in 300 μL of high-performance liquid chromatography (HPLC) water. Then, 900 μL of ethanol was added and vortexed. The suspension was then centrifuged at 16,000×g for 2 min. The pellet was allowed to dry at room temperature for approximately 20 min. Afterwards, 20 μL of 70% formic acid and 20 μL of acetonitrile were added to the pellet, homogenized and centrifuged. The supernatant was used for MALDI-TOF MS analysis. We transferred 1 μL of this extract to a MALDI polished steel target plate and allowed it to dry at room temperature. The sample was then coated with 1 μL of CHCA matrix. Spectra were obtained on a Microflex LT MALDI-TOF MS system using Flex Control software. We used a laser frequency of 60 Hz, ion source voltages of 2.0 and 1.8 kV, and lens voltage of 6 kV, which generated spectra in the range of 2000–20,000 m/z. All 19 C. difficile isolates included in this study were run in up to 27 technical replicates.

2.3. In silico spectral analyses

All spectra were imported to BioNumerics (v7.6, Applied Maths), where they were preprocessed and normalized using default parameters (baseline subtraction by rolling disc and peak detection using a signal-to-noise ratio of 10). After preprocessing, spectra were used to generate a neighbor joining tree based on Pearson correlation.

We used a tolerance of ±0.002 m/z for peak matching. All peaks (or biomarkers) detected in at least 30% of all replicates were exported to a spreadsheet, which we visually analyzed. Selection of ribotype-specific biomarkers was based on the intensity (higher than 2000 a.u.) and occurrence (present in at least 20 technical replicates) of peaks. For protein in silico prediction, we searched the UniProt database (www.uniprot.org) for peptides and proteins having molecular masses equal or most similar to the m/z values of biomarkers selected. After peak matching, spectra of all technical replicates of each isolate were combined into a single spectrum, which was then considered to be ribotype-specific [23]. All 19 summarized spectra were then used to generate an unweighted pair group method with arithmetic mean (UPGMA) dendrogram.

To evaluate and validate the MALDI ribotyping proposal in an unbiased way, a double-blind test was performed. One laboratorian ran and collected all spectra on MALDI-TOF MS. Another laboratorian analyzed the data on BioNumerics by assigning ribotypes to each spectrum using the ribotype-specific profiles determined here, without being previously aware of strain ribotypes.

3. 3-. Results

All 19 isolates were confirmed to be C. difficile by MALDI-TOF MS with scores ≥2.0. For BioNumerics analysis, only replicates that generated scores ≥2.0 were considered; thus, fifty spectra needed to be removed and analyses were performed with a total of 463 spectra, which represented at least 24 technical replicates of each one of all 19 ribotypes included. In the neighbor joining tree, spectra were distributed in clusters according to ribotype (Fig. 1A and B). Each ribotype was represented by one well-defined cluster when all technical replicates of a single run were considered (Fig. 1A) and also when all replicates of all runs were summarized in one single spectrum (Fig. 1B). Although ribotypes 131 and 135 were located in nearby clusters, it was still possible to differentiate the Brazilian ribotypes, which remained in separate clusters. The UPGMA dendrogram built from summarized spectra confirmed that all 19 ribotypes have distinguishable spectra profiles, with 77%–92% of similarity among them (Fig. 2). Those results suggest that MALDI-TOF MS could discriminate all 19 ribotypes evaluated, including the Brazilian C. difficile ribotypes (131, 133, 135, 142, 143 and 233).

Fig. 1.

Neighbor joining tree based on Pearson correlation, built with the MALDI-TOF MS spectra of Clostridioides difficile clinical isolates showing main clusters and distribution. (A) Each cluster represents 27 spots analysis of a single day of extraction and each colorful dot represent the spectrum of a single C. difficile ribotype. (B) Each dot represents the combination of 27 spectra of each ribotype, extracted through 5 consecutive days. BioNumerics software was used to build the trees.

Fig. 2.

Dendrogram generated from cluster analysis of MALDI-TOF MS mass spectra of C. difficile ribotypes. BioNumerics software (v7.6, Applied Maths) was used for this construction, after all spectra were collected and compiled.

A total of 42 biomarkers (supplementary material 1) were detected among the 19 C. difficile isolates. Of those, 13 were chosen to compose ribotype-specific profiles based on peak intensity and occurrence in each ribotype (Table 2 and Fig. 3). In silico prediction suggested that most of these biomarkers are uncharacterized proteins or well-conserved peptides, such as ribosomal proteins (Table 3).

Table 2.

Selected biomarkers for differentiation of Clostridioides difficile ribotypes. Biomarkers are related to the ribotypes and the respective m/z found in the mass spectra.

Ribotypes	Biomarkers (m/z)
	2574	3539	3962	3975	4448	4982	5059	5437	5826	5960	7264	7527	7553
001			X			X					X		X
010		X				X		X		X	X
012		X				X		X	X
014						X
015	X	X				X		X		X			X
020					X	X		X
027		X				X	X			X
031		X				X		X		X			X
038													X
043						X				X	X		X
046			X	X		X					X
060		X	X			X					X
106					X	X			X
131						X	X						X
133			X			X					X
135						X							X
142							X	X
143												X
233			X			X	X				X		X

Fig. 3.

Mass spectra obtained after gathering 27 mass spectral profile readings of six Clostridioides difficile ribotypes (boxes). Arrows indicate differences detected among ribotypes. Biotyper software was used to generate the spectra.

Table 3.

Relation between biomarkers and proteins of Clostridioides difficile.

Biomarkers (m/z)	Protein	Microorganism	Gene	Mass (Da)	Length	Entry name
2574.80	Truncated TcdC	Clostridioides difficile	tcdC	2568	21	A7UJ27
3539.37	Uncharacterized protein	Phage phiCDHM14 Clostridium	phiCDHM14_gp34	3537	30	A0A090DBM2
3962.45	Uncharacterized protein	Phage phiMMP04 Clostridium	phiMMP04_gp42	3967	34	J9QE91
3975.77	Uncharacterized protein	Clostridioides difficile E10 strain	BN166_640025	3977	34	U3VIL1
4448.02	Uncharacterized protein	Clostridioides difficile 630 strain	CD630_32561	4449	39	F3Y655
4982.10	Putative phage protein	Clostridioides difficile T5 strain	BN163_1150018	4983	40	U3UC47
5059.48	Uncharacterized protein	Clostridioides difficile E10 strain	BN166_1780024	5062	43	U3VA00
5437.60	Uncharacterized protein	Clostridioides difficile E10 strain	BN166_2860004	5436	45	U3VF68
5826.04	Uncharacterized protein	Clostridioides difficile T5 strain	BN163_60040	5829	45	U3UKS4
5960.86	50S ribosomal protein L33	Clostridioides difficile 630 strain	rpmG CD630_00581 CD0058A	5959	49	Q18CE3
7264.86	50S ribosomal protein L35	Clostridioides difficile	rpmI IM33_02980	7262	64	A0A0N1HX66
7527.32	Transcriptional regulator	Phage phiCD211 from Clostridium genus	PHICD211_20164	7526	64	A0A0A8WF92
7553.91	Negative protein regulator	Clostridioides difficile	yxlE NCKUH21_00337	7553	66	A0A286PIL5

In the double-blind assessment using the 13 biomarker profiles for determining ribotypes, the percentage of correct assignments varied according to ribotype, ranging from 0% (in two ribotypes) to 100% (in seven ribotypes). In general, 73% (339) of the 463 spectra evaluated were correctly ribotyped (Table 4). Moreover, most (13) of the 19 ribotypes evaluated, including the epidemic BI/NAP1/ribotype 027, and all the Brazilian ribotypes had from 94% to 100% correct hits (Table 4).

Table 4.

Validation of Clostridioides difficile typing by MALDI-TOF MS. Dark gray indicates no correlation, medium gray indicates good correlation and light gray indicates optimal correlation.

Ribotype (n)	Correct % (n)	Wrong % (n)	Uncertain % (n)
001 (22)	23% (5)	0	77% (17)
010 (20)	45% (9)	0	55% (11)
012 (24)	21% (5)	4% (1)	75% (18)
014 (24)	17% (4)	0	83% (20)
015 (25)	96% (24)	4% (1)	0
020 (25)	96% (24)	0	4% (1)
027 (25)	100% (25)	0	0
031 (26)	96% (25)	4% (1)	0
038 (25)	0	100% (25)	0
043 (25)	94% (24)	4% (1)	0
046 (25)	94% (24)	4% (1)	0
060 (25)	100% (25)	0	0
106 (24)	96% (23)	4% (1)	0
131 (24)	100% (24)	0	0
133 (25)	100% (25)	0	0
135 (25)	100% (25)	0	0
142 (25)	100% (25)	0	0
143 (23)	100% (23)	0	0
233 (25)	0	36% (9)	64% (16)
All (463)	73% (339)	9% (40)	18% (84)

4. 4-. Discussion

Bacterial typing is critical to tackle the spread of bacterial pathogens, but most of current methods may have drawbacks that hinder their wide application, especially in clinical laboratories of low- and middle-income countries [29]. With the integration of MALDI-TOF MS into the microbiology laboratory workflow, microbial species identification became faster and cost-effective [21,30]. However, robust epidemiological data rely on gathering information on specific bacterial strains, usually based on molecular methods [31,32]. For C. difficile, determining the ribotype involved in the infection is extremely important. However, most techniques used to type C. difficile, such as PCR-ribotyping, are time-consuming, costly, labor-intensive and not widely available [29,33]. In Latin America countries, including Brazil, the PCR-ribotyping is still not available, making it difficult to evaluate C. difficile epidemiology [16].

Recently, global independent research groups have successfully used MALDI-TOF MS to discriminate subgroups within a single microbial species [34,35]. Special pathogens involved in outbreaks, such as MRSA, Haemophilus influenzae, Streptococcus pneumoniae [23–25] and even pathogenic Acanthamoeba spp. [36], were studied. For C. difficile, few research groups have tried to correlate mass spectra of certain strains with their respective ribotypes. Reil et al. evaluated 29 different ribotypes, but only managed to discriminate four of them; two ribotypes were very similar and were grouped as a single ribotype [37]. These researchers acquired spectra from whole bacterial cells (without any extraction protocol) and used SARAMIS to analyze spectra. Rizzardi et al. [38] successfully used mass spectra to differentiate 20 ribotypes, using the Bruker Biotyper system and spectra from purified extracts containing high molecular mass proteins recovered from C. difficile surface layers. More recently, a group of researchers from China used 204 C. difficile isolates to evaluate MALDI-TOF MS as a tool to identify ribotype 017 strains belonging to sequence type 037 (ST37). ST37 is the most prevalent genotype circulating in that country. The authors were able to identify two peaks (3242 m/z and 3286 m/z), with specificity of 98.7%, that were present only in ST37 strains [39].

Emele et al. [27] used 109 C. difficile strains (5 MLST clades) to create a proteotyping scheme based on MS-detectable biomarkers coupled with translated National Center for Biotechnology Information (NCBI) genome sequences to create isoforms (biomarkers genes). The authors identified nine biomarkers (ribosomal proteins), but their proteotyping scheme could only differentiate ribotypes 027 and 176 from non-027 and non-176 strains. Although the method could be used for a specific epidemiological question, it is limited and could not distinguish the C. difficile ribotypes 027 from 176. Such discrepancies among different studies might be due to the variety of methods used for obtaining and analyzing MALDI-TOF MS spectra. That highlights the importance of standardizing these steps so that results from different studies can be compared. Cheng et al. conducted a MALDI-TOF MS spectra peak statistics study. The authors used 160 C. difficile isolates recovered from three large teaching hospitals, including C. difficile MLST clade 4 strains, which present high level resistance to drugs, and are the most prevalent in China. In this study, all 41 clade 4 C. difficile isolates circulating in Beijing clustered together and identified five biomarkers that provide simple and accurate method for identifying this specific clade [40]. The authors concluded that MALDI-TOF MS is a very simple method for identifying and monitoring C. difficile clones in China. Corver et al. also studied peptide markers to identify C. difficile clades through MALDI-TOF MS. In their study, 44 strains, belonging to eight different MLST, were analyzed through ultrahigh-resolution Fourier transform ion cyclotron resonance (FTICR) MS. The amino acid sequence of peptide markers MLST-specific were determined by MALDI-TOF MS/MS. The authors concluded that the combination of MALDI and FTICR MS could identify peptide markers and distinguish MLST-1 and MLST-11 C. difficile [41].

Ortega et al. [42], used a methodology established by Rizzardi and Akerlund in 2015 [42], using MALDI-TOF profiling of C. difficile. The study analyzed 1000 isolates of toxin-positive C. difficile from different hospitals in Sweden and compared them with isolates from United Kingdom The authors could detect three outbreaks in different hospitals caused by the epidemic RT027 strain. Their method proved a very useful and cost-effective to detect outbreaks of C. difficile in early stages and are implemented as a routine diagnostic tool in Sweden hospitals.

In our study, we aimed to identify and type at the same time C. difficile ribotypes, typically involved in CDI cases in Brazilian hospitals. For that, 19 C. difficile strains representing different ribotypes were evaluated, using parameters that were chosen for practicality and performance, including: i) Supplemented blood agar media because C. difficile growth rate is satisfactory on this medium and is usually available in microbiology laboratories worldwide; ii) The extraction protocol recommended by Bruker on C. difficile colonies before generating spectra. This method is fast, simple and might reduce background when compared to whole cell analysis; iii) Twenty-seven technical replicates of each isolate in our analyses to help increase peak reproducibility and correctly align all peaks with the similar m/z.

Using such parameters, it was possible in this study to successfully differentiate the Brazilian ribotypes from those circulating worldwide. This can be observed in the neighbor joining trees built with all 27 replicates of all isolates, in the UPGMA dendrogram generated from summarized spectra and on the list of the 13-biomarker profiles that we have identified. Because we had only one strain representing a particular ribotype, two neighbor-joining trees were constructed to disclose consistency in the identification method. The results remained consistent either when the extraction and run were performed on a single day or on different days.

In the double-blind analysis carried out to validate the 13-biomarker profiles that were determined, accuracy was at least 94% for 13 of the 19 ribotypes included in the study, and 73% of all spectra evaluated were correctly typed. Ribotypes that had 100% of correct hits included 133 and 135, which are exclusive to Brazil, and 027. One of the advantages of determining ribotype-specific profiles of biomarkers is that other research groups can use them to analyze their C. difficile isolates, without the need for dedicated software such as BioNumerics. A number of methodologies can be used to type C. difficile isolates. For example, in a study conducted by Berger et al., six C. difficile epidemiologic important ribotypes (001, 002, 010, 014, 027 and 078) were sent to 21 different laboratories in Europe. The laboratories could use any methodology (MLST, PCR-ribotyping, slpAST, DNA based microarray, PFGE, MALDI-TOF MS and WGS) to identify and type the strains. According to the authors, any of the methodologies carried out by all the laboratories participants, could identify all six ribotypes. For instance, three participants used PCRs and MALDI-TOF MS and they identified only the classical ribotypes, 027 and 078. MALDI-TOF could not discriminate between ribotypes 001 and 078. The authors argued that those discrepancies among the methodologies used in the study might be related to the epidemiology of C. difficile, which is permanently in change, and the considerable differences among C. difficile strain composition. They also explain that sub-typing of C. difficile in Europe and other regions of the world is dominated by in-house solutions that serves as a basis for new typing schemes [43] and help with the epidemiology data.

Although we demonstrated great efficacy in identifying strains, this study presented a few limitations. We are aware that our work would benefit from a larger sample size, especially when gathering data on the clustering analysis, which would improve accuracy. Also, since our database had one specimen for each ribotype, the validation had to be in silico. This type of validation was necessary to overcome the use of technical replicates, so we could obtain a more robust analysis. Since this is a preliminary study, further studies are required to validate the proposal and improve its reproducibility, especially for dealing with other strains and ribotypes, making it more robust for effective future application.

Mass spectrometry has a wide variety of functions from fast bacterial identification in food-borne disease outbreak, water quality control, antibiotics susceptibility/resistance tests, fast infectious disease diagnosis, to biomarker discovery that may assist and identify precisely closely related organisms [29]. Some of those methods have become routine in microbiological and research laboratories as an attempt to reduce turnaround time, cost, and overall labor [3]. Given the widespread use of the MALDI-TOF MS for routine bacterial identification in Brazil in research institutes and clinical laboratories, it becomes a cost-effective method of screening for the C. difficile strains in the country and for specific ribotypes involved in CDI outbreaks in Brazilian hospitals by using cluster spectral analysis as shown in this study. This approach could help to enhance the epidemiological data about this pathogen in the country and laboratories that are not able to perform PCR-ribotyping, or any other typing methdology, to achieve and gather robust epidemiological data on CDI. Ultimately, the approach could support and advance efforts to better understand this clinically relevant pathogen and associated diseases.