Multi-strain Clostridioides difficile infection: Increased detection and clinical implications

Abstract

Clostridioides difficile infection (CDI) is an antibiotic-associated diarrheal disease. Mixed infection has been previously reported. However, its clinical implications are not well defined. In this study, we evaluated the presence of multi-strain (MS) isolates in fecal specimens from patients diagnosed with CDI at a tertiary hospital in Virginia. C. difficile colonies (three per sample) were obtained from fecal specimen cultures (N=174) from patients diagnosed with CDI from April 2021 to February 2022. The toxin gene profile of each isolate was determined by PCR using specific primers for triose phosphate isomerase (tpi), toxin A (tcdA), toxin B (tcdB), binary toxin A (cdtA), and binary toxin B (cdtB). A subset of fecal specimens (n=62) with Ct values 25 for tcdB underwent total bacterial isolation in TCCFA and were processed for whole-genome shotgun sequencing. Based on the toxin gene profile of each isolate, MS infection was identified in 32.8% (57/174) of patients. Among all the isolates (N=522), tcdA⁺tcdB⁺cdtA⁻cdtB⁻ (32.38%), tcdA⁺tcdB⁺cdtA⁺cdtB⁺ (32.18%), and tcdA⁺tcdB⁺cdtA⁺cdtB⁻ (28.5%) were the most common toxin gene profiles identified. An analysis of clinical data of 160 patients showed that age, gender, recurrence, severity of disease, mortality, and being infected by a strain expressing the binary toxin gene did not distinguish between being infected by MS or SS (single strain). Among the MS cases (N=52), the presence of non-toxigenic strains decreased the risk of severe CDI (p=0.0098). Whole-genome shotgun sequencing analysis confirmed mixed infection in our cohort and identified that the C. difficile strains S-0352 (clade 1, ST2) and W0023a (clade 1, ST42) are the most common sequence types in MS and SS cases, respectively. Overall, our findings indicated a high prevalence of MS infection among C. difficile-infected patients, and a non-toxigenic C. difficile strain association with less severe disease in MS cases. These results suggest that specific strains may differentially predominate in MS or SS infection and that the presence of non-toxigenic C. difficile strains in MS infection influences disease severity. The impact of MS infection on the outcomes of anti-C. difficile treatments warrant further investigation.

Introduction

Clostridioides difficile (C. difficile) is a gram-positive, anaerobic, spore-forming bacterium and the causative agent of C. difficile infection (CDI). C. difficile is the leading cause of hospital-associated diarrhea worldwide.^1,2 Moreover, cases of community-acquired CDI have been increasing.³ In 2022, the Centers for Disease Control and Prevention (CDC) reported a higher incidence of community-acquired CDI cases (62.1 cases per 100,000 individuals) compared with healthcare-associated cases (54.0 cases per 100,000 individuals). In the United States of America (USA), CDI was responsible for a total of 216, 083 deaths between 1999 and 2023.⁴ In Europe, the incidence of CDI ranges from 1.49 to 4.1 per 10,000 patients/day,^5,6 while in Asia, it is estimated that 5.3 episodes of CDI occur per 10,000 patients/ day.⁷ Patients experiencing a first episode of CDI can have a severe or non-severe disease.⁸ Moreover, 13.7–25% of patients who experience an initial episode of CDI develop recurrent disease,^9–12 which is characterized by a new diarrhea episode with detection of the bacteria, within 2 to 8 weeks of completing treatment.¹³ Among those with recurrent disease, up to 45–65% will suffer from further multiple recurrences.⁹ The direct medical costs attributable to recurrent CDI range from $67,837 to $82,268 per patient/year.¹⁴

Different types of C. difficile strains can infect patients, each belonging to distinct clades with unique characteristics, virulence profiles, antimicrobial resistance, and toxin expression. The classification of C. difficile strains is often conducted using genomic sequencing typing, which divides isolates into five clades, clades 1 to 5, and PCR-based ribotyping.¹⁵ Geographically, ribotype distribution varies.^15–17 In the USA, the most prevalent ribotypes of C. difficile are 027, 014/020, 106, 001, and 056, according to a multicenter observational surveillance study conducted in hospitals across the USA between 2011 and 2017.¹⁶ In Europe, a multicenter study revealed a diversity of C. difficile ribotypes, with predominance of 078, 126, 014, 181, and 027.¹⁵ In Asia, RT 017, followed by RTs 002, 018, and 014, are the most common circulating strains, according to a recent multicenter study of CDI conducted across countries in the Asia-Pacific region.¹⁷ These ribotypes may coexist in an individual host. Emerging evidence suggests that mixed infections with multiple strains of C. difficile may be more common than previously recognized and contribute to relapse.¹⁸ Some studies reported that around 16–78% of CDI second episodes are not caused by the primary strain.^18–21 A longitudinal study performed with hospitalized patients in the USA found that mixed infections, which were characterized by the detection of more than one ribotype in fecal specimens, were higher in patients who later developed recurrence (11/14, 78.6%) in comparison to nonrecurrent patients (7/19, 36.8%).¹⁸

Multi-strain (MS) infections, also known as mixed infection, in patients with CDI are not well established. The clinical relevance and a better methodology to detect this multi-strain infection are still needed. Molecular strain characterization and antibiotic susceptibility studies have traditionally focused on a single isolated colony from clinical samples, with few investigations addressing the prevalence of mixed infections in patients with CDI disease and their impact on clinical outcomes. Here, we aimed to identify the occurrence of MS infections in fecal samples collected from adult patients using toxin gene profiling and to verify whether biological variables, clinical disease variables (severity and recurrence), and the presence of the binary toxin gene could influence their occurrence. In a subset of samples, we confirmed the presence of MS infections by whole-genome shotgun sequencing.

Material and Methods

Human fecal samples and clinical data collection

All fecal samples and experimental procedures involving human specimens were conducted after the University of Virginia (UVA) Institutional Review Board (IRB) approval (IRB# 20813). For this study, a total of 174 fecal specimens from CDI-positive patients, obtained from the Clinical Microbiology Laboratory at the UVA Medical Center between April 2021 and February 2022 and stored at −20 °C, were analyzed. Samples were classified as positive based on the detection of C. difficile toxins (TcdA and TcdB) by ELISA and tcdB gene by PCR, which were performed by the UVA Clinical Microbiology Laboratory. Upon receipt, each fecal sample was divided into five 1 mL aliquots and stored at −20 °C for subsequent analyses. An initial screening to confirm the positivity of each sample was performed to detect the tcdB and cdtB genes using quantitative PCR (qPCR) on DNA extracted directly from 500 μL of stool. Samples with cycle threshold (Ct) values ≤32 were classified as positive. Melt curve analysis confirmed amplification specificity, with peak values matching those of the positive control (UK1, ribotype 027) for tcdB (75.00 °C ± 2) and cdtB (76.00 °C). Based on these criteria, all 174 samples (100%) tested positive for the tcdB gene, while 12 samples (6.9%) were positive for the cdtB gene.

From each fecal sample, three C. difficile colonies were randomly selected (one single colony from each quadrant) and subjected to two successive subcultures on taurocholate cycloserine-cefoxitin fructose agar (TCCFA) plates under anaerobic conditions (Fig. 1A). A random selection of colonies involves examining the entire culture plate to identify C. difficile colonies and collecting a single colony from three distinct quadrants. In TCCFA plates, C. difficile colonies appear flat, with an irregular shape, and gray or white color.²² This procedure yielded three clinical isolates per sample, resulting in a total of 522 isolates. C. difficile strains can differ at the genetic, toxigenic, phenotypic, and epidemiological levels. Among these, differences in toxin production represent a clinically relevant determinant, directly influencing virulence and disease severity.^23,24 Bacterial DNA was extracted from each isolate, and qPCR was performed to detect five target genes: tpi (triose phosphate isomerase), tcdA, tcdB (toxins A and B), and cdtA/cdtB (genes codifying the binary toxin). For toxin gene profiles determination, isolates with a cycle threshold (Ct) value ≤30 and a melt peak within ± 2 °C of the positive control were considered positive. Samples were classified as MS if more than one C. difficile toxin gene profile was identified among the three isolates. Samples were designated CDT⁺ (binary toxin genes positive) if at least one isolate carried both cdtA and cdtB.

Fig. 1.

Evaluation of multi-strain C. difficile in human stool samples through sequential isolation on TCCFA plates and qPCR analysis. (A) Isolation workflow for C. difficile. Fecal specimen aliquots were individually treated with alcohol, and pellets were plated on TCCFA plates (Isolation 0). From each plate, three colonies were randomly selected and subjected to two successive subcultures on TCCFA plates (Isolation 1 and Isolation 2). Following the final isolation, a single colony was cultured in BHIS broth. Aliquots from this culture were used for: (1) DNA extraction, (2) glycerol stock preparation, and (3) chopped meat broth (MB) stock preparation. (B) PCR screening for C. difficile toxin genes in isolates from stool specimens from 174 CDI-positive patients. Screening targeted tcdA and tcdB (toxins A and B) and binary toxin genes cdtA and cdtB (CDT⁺). Samples were considered positive for tcdA, tcdB, cdtA, or cdtB if at least one of the three isolates from the same specimen had a Ct value ≤30 and a melt curve peak within ± 2 °C of the positive control (C. difficile UK1, ribotype 027): tpi = 75.50 °C, tcdA = 76.50 °C, tcdB = 75.00 °C, cdtA = 75.50 °C, cdtB = 76.00 °C. MS: multi-strain; SS: single-strain.

Biological (age and gender) and clinical variables (severity, recurrence, and mortality in the last 6 months) from each patient were collected using the electronic medical record. Recurrence and severity were determined as previously described.^10,13 Recurrent CDI case included patients diagnosed with a new episode of diarrhea (three or more loose or watery bowel movements within 24 h) with a positive toxin gene or antigen stool assay following an initial improvement of symptoms after treatment of a first episode of CDI (typically occurring 2–8 weeks later).²⁵ Severe cases included patients with a white blood cell (WBC) count of > 15, 000 cells/mm³ or serum creatinine of 1.5 mg/dL.

Isolation of C. difficile from fecal samples

For C. difficile isolation, 500 μL aliquots of each fecal specimen were mixed with 500 μL of ethanol (alcohol shock) to reduce vegetative bacterial cells and enrich spores. The mixtures were incubated at room temperature for 1 h, centrifuged at 4,500 g for 10 min at 4 °C (Eppendorf 5810 R), and the supernatant was discarded. Pellets were resuspended in 500 μL of 1× PBS and stored at −20 °C for subsequent isolation procedures. Isolation of C. difficile was performed on TCCFA (taurocholate-cefoxitin-cycloserine-fructose agar) plates incubated at 37 °C under anaerobic conditions (85% N₂, 5% H₂, 10% CO₂; COY Vinyl Double, Coy Laboratory). Then, 100 μL of each homogenate was plated onto fresh TCCFA plates (Plate 0). After 48 h of incubation, three C. difficile colonies from each sample were randomly selected (one single colony for each quadrant) and streaked onto fresh TCCFA plates (Plate 1). This subculturing step was repeated once more (Plate 2). Following the final plating, a single isolated colony was inoculated into 5 mL of BHIS (Brain Heart Infusion) supplemented with l-cysteine (1 g/L) and vitamin K (1 mg/L), and incubated anaerobically at 37 °C for 24 h. Isolates were stored in 50% glycerol at −80 °C and in chopped meat broth (Anaerobe systems, AS-811) at room temperature. Additionally, a 1 mL aliquot of each culture was collected for DNA extraction and PCR analysis of the genes of interest.

DNA extraction from fecal samples

DNA extraction from fecal samples was performed using the QIAamp DNA Stool Mini Kit (QIAGEN) and the automated QIAcube system following the manufacturer’s instructions (QIAGEN).

DNA extraction from C. difficile isolates

C. difficile genomic DNA from each isolate (1 mL aliquot of C. difficile clinical isolates grown in BHIS) was obtained as previously described.¹⁰

PCR detection of C. difficile toxin genes

Detection of tpi, tcdA, tcdB, cdtA (C. difficile binary toxin A), and cdtB (C. difficile binary toxin B), in DNA extracted from C. difficile clinical isolates or fecal samples was performed by PCR. Primer sequences are listed in Table S1. Each reaction was prepared in a final volume of 20 μL containing 2 μL of primer mix (5 μM each primer), 4 μL of nuclease-free water (Invitrogen), 10 μL of iTaq Universal SYBR® Green Supermix (Bio-Rad), and 4 μL of DNA template. PCR amplification was carried out on a CFX96 Real-Time System or CFX Connect Real-Time System (Bio-Rad) using the following conditions: initial denaturation at 95 °C for 5 min; 39 cycles of 95 °C for 5 s and 55 °C for 30 s; followed by a melt curve analysis from 55 °C for 5 s to 95 °C with 0.5 °C increments for 5 s. Samples were considered positive if the Ct value was ≤30 and the melt curve peak matched that of the positive control (UK1, ribotype 027) within ± 2 °C: tpi = 75.50 °C, tcdA = 76.50 °C, tcdB = 75.00 °C, cdtA = 75.50 °C, and cdtB = 76.00 °C.

Whole-genome shotgun sequencing of clinical isolates and analysis

Of the 174 fecal specimens used in our initial study, 62 fecal specimens were selected for bacterial whole-genome shotgun sequencing. Low Ct values (≤25) for the tcdB gene, which represent the most relevant CDI cases in terms of disease outcome,^26,27 in fecal specimens was used as a criterion to select the samples. Strain UK1 (ribotype 027) served as the positive control, while a CDI-negative fecal sample was used as the negative control. To perform the sequencing of all the C. difficile strains present in each fecal specimen, 100 μL fecal samples submitted to alcohol shock (500 μL aliquots of each fecal specimen were mixed with 500 μL of ethanol, incubated at room temperature for 1 h, centrifuged at 4,500 g for 10 min at 4 °C, and the supernatant was discarded) were resuspended in 500 μL of PBS, plated on TCCFA and incubated anaerobically at 37 °C for 4 days. All resulting colonies were harvested by adding 1 mL of PBS, transferred to 1.5 mL microtubes, and centrifuged at 10,000 rpm for 5 min at 4 °C. The supernatant was discarded, and pellets were stored at −20 °C. Genomic DNA was extracted using the QIAamp PowerFecal® Pro DNA Kit (QIAGEN) according to the manufacturer’s instructions. Extracted genomic DNA was quantified and stored at −20 °C. Aliquots of 100 μL containing 10 ng/μL of DNA were prepared in TE buffer (Invitrogen, 12090–015) and submitted for sequencing at the Genome Analysis and Technology Core. Prior to library preparation, the purity and quality of DNA samples were checked using the Agilent TapeStation platform (D5000 HS). Illumina sequencing libraries were prepared using P3 XLEAP-SBS reagent kit (200 cycles) and sequenced on an Illumina NextSeq 500 machine with 150-base paired-end reads.

Methodology for strain level identification of C. difficile

We had an average 20 million reads per sample (range 10–40 million). The quality of raw sequence reads was examined using FastQC, and quality control (QC) reports were summarized using MultiQC. After reviewing the FastQC report, adapter contamination was identified, and the same was trimmed out using Cutadapt.²⁸

For bacteria species characterization, we performed multi-locus sequence typing (MLST) analysis on metagenomic data using the program metaMLST (https://github.com/SegataLab/metamlst.git) by mapping reads against publicly available MLST housekeeping gene sequences of C. difficile (adk, atpA, dxr, glyA, recA, sodA, and tpi; 8,42 sequences), along with their ST information, which was retrieved from PubMLST (https://pubmlst.org). We created a metaMLST database by concatenating all seven-housekeeping gene FASTA files, adding ST information to the database, and generating a Bowtie2 index following the metaMLST protocol. Mapping was performed using Bowtie2 in local alignment mode.²⁹ MetaMLST was then run on BAM files, followed by metaMLST-merge to assign sequence types (STs) based on housekeeping gene alleles identified in the samples.³⁰

For strain-level identification of C. difficile, we used StrainGE, which identifies strains based on k-mer comparisons between reference genomes and sample data.³¹ A total of 177 C. difficile whole-genome sequences were retrieved from NCBI using the ncbi-genome-download tool, and a StrainGE database was constructed using the built-in database construction pipeline. The function “straingst kmerize” was used to generate k-mers for the reference genomes, followed by “straingst kmersim” to compute all-versus-all genome similarity using the Jaccard index. Genomes sharing ≥99% of k-mers with a Jaccard similarity ≥0.99 were grouped using “straingst cluster”, resulting in 94 representative genomes selected for building a pangenome database for the given genus and species using “straingst created”. The clustering step combines highly similar genomes into clusters and selects an optimal representative genome for each cluster, thereby minimizing redundancy in the resulting pangenome database.

Similarly, k-mers were generated for samples using “straingst kmerize”, and sample k-mers were compared against k-mers in the pangenome database to estimate strain-level relative abundance using “straingst run” in each sample. Sample-wise strain abundance profiles were then combined with sequence type (ST) information obtained from metaMLST analysis to annotate each sample with strain-level and ST assignments.

Statistical analysis

GraphPad Prism software 10.1.1 was used to perform all the statistical analyzes. The chi-square or Fisher’s exact test was used to assess associations between variables. A p-value of < 0.05 was considered to indicate significance.

Results

Characterization of the C. difficile isolates

Toxin gene profile analysis of the 174 specimens (three isolates from each sample) revealed a high prevalence of binary toxin positive (CDT⁺) samples: 71 (40.8%) were positive for both cdtA and cdtB. 57 (32.8%) of the specimens were MS, i.e., containing more than one infective strain (Fig. 1B). Notably, CDT⁺ prevalence was higher among MS samples (47.4%, 27 of 57) compared to single-strain (SS) samples (37.6%, 44 of 117) (Fig. 1B).

Nine distinct toxin gene profiles were identified among 522 C. difficile isolates from 174 fecal specimens (Fig. S1). The most prevalent profiles were tcdA⁺ tcdB⁺ cdtA⁻ cdtB⁻ (32.38%, 169 isolates), tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁺ (32.18%, 168 isolates) and tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁻ (28.50%, 149 isolates). Less frequent profiles included, tcdA⁻ tcdB⁺ cdtA⁻ cdtB⁻ (3.26%, 17 isolates), tcdA⁻ tcdB⁻ cdtA⁻ cdtB⁻ (1.53%, 8 isolates), tcdA⁺ tcdB⁻ cdtA⁻ cdtB⁻ (1.34%, 7 isolates), tcdA⁻ tcdB⁺ cdtA⁺ cdtB⁻ (0.38%, 2 isolates), tcdA⁻ tcdB⁺ cdtA⁺ cdtB⁺ (0.19%, 1 isolate), and tcdA⁺ tcdB⁺ cdtA⁻ cdtB⁺ (0.19%, 1 isolate) (Fig. S1). Analysis of the three clinical isolates from each specimen revealed toxin gene profile diversity within samples, with MS detected in 57 specimens (32.8%). As shown in Fig. 2A, the most common toxin gene profile in MS samples were tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁻ (41 specimens), tcdA⁺ tcdB⁺ cdtA⁻ cdtB⁻ (30 specimens), and tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁺ (27 specimens). Notably, seven specimens from MS samples (P04, P17, P37, P97, P115, P157, P182) displayed the tcdA⁻ tcdB⁻ cdtA⁻ cdtB⁻ profile but were positive for the tpi gene, indicating that they were non-toxigenic C. difficile strains (Fig. 2A). The remaining 117 SS samples (Fig. 2B-C) exhibited five distinct toxin gene profiles. The most frequent were tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁺ (44 specimens), tcdA⁺ tcdB⁺ cdtA⁻ cdtB⁻ (41 specimens), and tcdA⁺ tcdB⁺ cdtA⁺ cdtB⁻ (28 specimens). The two less common profiles tcdA⁻ tcdB⁺ cdtA⁻ cdtB⁻ and tcdA⁺ tcdB⁻ cdtA⁻ cdtB⁻ were each found in two specimens (Fig. 2B-C).

Fig. 2.

Toxin gene profiles distribution of C. difficile clinical isolates. A total of 522 C. difficile isolates from 174 CDI-positive patients were analyzed in this study to identify the toxin genes carried by isolates from single-strain (SS) and multi-strain (MS) infections. Each isolate was subjected to qPCR using specific primers to the genes of interest (tpi, tcdA, tcdB, cdtA, and cdtB). Distribution of toxin gene profiles across the analyzed specimens. (A) Isolates from MS infections. (B-C) Isolates from SS infections. (A-C) The colors represent the number of strains identified in each fecal specimen (Black: 3 isolates; Dark green: 2 isolates; Light green: 1 isolate; White: no isolates).

Clinical correlations in C. difficile-infected patients with multi-strain infections

Next, we investigated associations between MS infections and biological (age and gender) or clinical variables (recurrence, disease severity, gender, and mortality) in patients with CDI. Among 174 fecal specimens collected, complete clinical data were available for only 160 patients. Among these patients, 52 (32.5%) exhibited MS, while 108 (67.5%) had SS infections. Interestingly, patients under 65 years of age showed a higher frequency of mixed infections (66.4%, 34/52, Table 1), although this difference was not significant.

Table 1.

Clinical data analysis among multi-strain (MS) and single-strain (SS) samples.

CDI positive patient characteristics from UVA Hospital (N= 160)
Summary characteristics	Multi-strain (MS)a n=52	%	Single-strain (SS)b n=108	(%)	P=values
Age
≥ 65	18/52	34.6	46/108	42.6
< 65	34/52	66.4	62/108	57.4	P=0.3347
Recurrence
Recurrent	11/52	21.2	20/108	18.5
Nonrecurrent	41/52	78.8	88/108	81.5	P=0.6928
Severity
Severe	26/52	50.0	45/108	41.7
Non-severe	26/52	50.0	63/108	58.3	P=0.3204
Gender
Male	23/52	44.2	48/108	44.4
Female	29/52	55.8	60/108	55.6	P=0.9797
Death
Yes	7/52	13.5	16/108	14.9
No	45/52	86.5	92/108	85.1	P=0.8192
Binary+ (cdtA+/cdtB+)
Positive	23/52	44.2	40/108	37.0
Negative	29/52	55.8	68/108	63.0	P=0.3830
Non-toxigenic strains
(tcdA-/tcdB-/ cdtA-/cdtB-)
Yes	7/52	13.5	0/108	0.0	***
No	45/52	86.5	108/108	100.0	P=0.0003

^(a)Specimens MS showed the presence of more than one toxin gene profile among 3 isolates.

^(b)Specimens SS showed only one toxin gene profile among the isolates. Chi-square test was used for all the analyses (exception: analysis on non-toxigenic strains, for which Fisher’s exact test was applied).

Non-toxigenic C. difficile strains were identified in 7 of 52 (13.5%) patients infected with MS strains (Table 1). Among the 7 unique cases of MS infections with at least one non-toxigenic C. difficile strain, the majority were patients who were more than 65 years old (5/7, 71.4%), female (4/7, 57%), first CDI cases (5/7, 71.4%), non-severe (7/7, 100%) and post-infection survivors (5/7, 71.4%). Comparing the presence of non-toxigenic C. difficile strains versus their absence in MS infections, we found that being infected with at least one non-toxigenic strain was more associated with less severe disease (P=0.0098) but did not distinguish recurrent cases or mortality outcome (Table 2).

Table 2.

Clinical impact of the presence of non-toxigenic strains in C. difficile-infected patients with MS infections.

CDI positive patients with Multi-strain (MS) infections (N= 52)
Summary characteristics	Presence of non-toxigenic strains n=7	%	Absence of Non-toxigenic strains n=45	(%)	P=values
Recurrence
Recurrent	2/7	28.6	9/45	20.0
Nonrecurrent	5/7	71.4	36/45	80.0	P=0.6303
Severity
Severe	0/7	0	26/45	57.8	**
Non-severe	7/7	100.0	19/45	42.2	P=0.0098
Death
Yes	2/7	28.6	5/45	11.1
No	5/7	71.4	40/45	88.9	P=0.2346

Fisher’s exact test was applied.

No statistically significant differences were found between MS and SS groups for recurrence, disease severity, mortality, gender, or binary toxin presence (Table 1). However, some of these characteristics (recurrence, severe disease, female gender and presence of binary toxin genes) tended to be more frequent in MS cases: recurrent CDI was observed in 21.2% (11/52) of MS patients compared to 18.5% (20/108) of SS patients, severe CDI was identified in 50% (26/52) of MS versus 41.7% (45/108) of SS patients, females represented 55.8% (29/52) of MS cases versus 55.6% (60/108) in SS, and binary toxin positivity was detected in 44.2% (23/52) of MS versus 37.0% (40/108) of SS patients (Table 1).

DNA sequencing by whole-genome shotgun sequencing in enriched culture from 62 human fecal samples

Given that the analysis of only three independent clinical isolates cultured from fecal samples of CDI-positive patients may be insufficient in detecting MS in all fecal samples, we performed whole-genome shotgun sequencing in an enriched isolation culture system to validate our findings (Fig. 3A). Based on this technique, we evaluated 62 specimens exhibiting a Ct value ≤ 25 for tcdB, which represent the most clinically relevant cases.^26,27 Sequencing analysis showed that 75.8% of the specimens (47/62) contained a single infecting strain (SS), whereas 24.2% (15/62) harbored MS. Similar to earlier data, a high prevalence of binary toxin genes (cdtA⁺/cdtB⁺) was observed in MS infection: 53.3% of MS (8/15) and 27.3% of SS samples (12/44) tested positive.

Fig. 3.

Bacterial DNA extraction and whole-genome shotgun sequencing for the detection of multiple C. difficile strains. The presence of SS (single-strain) or MS (multi-strain) in 62 human fecal samples with CDI diagnostic and low Ct value for tcdB gene (≤25), which represent the most clinically relevant samples, was assessed by whole-genome shotgun sequencing bacterial DNA extracted from fecal aliquots cultured anaerobically for 4 days on selective medium. (A) Workflow for bacterial DNA extraction from cultured fecal samples. (B) Identification of multi-strain (MS) and single-strain (SS) infections in fecal specimens based on whole-genome shotgun sequencing analysis (using StrainGE analysis for classification). *The percentage of SS infections containing a binary-positive strain was calculated using a total of 44 specimens, for which the identification of binary-positive or binary-negative status was well established.

Our whole-genome sequencing analysis revealed 27 distinct strains across the 62 fecal specimens evaluated, suggesting a high level of genetic heterogeneity. Among MS samples, the most frequent strain detected was Clos_difficile_S0352 (RT014, ST2 and clade 1) ³² and Clos_difficile_Cd10 (ribotype and clade not defined) found in 33.3% (5/15) and 20% (3/15) of MS specimens, respectively (Fig. 4A). In SS samples, Clos_difficile_W0023a (ST42, clade 1) ³³ and Clos_difficile_S0253 (RT002, ST8, clade 1)³² were the predominant strains detected, corresponding to 21.3% (10/47) and 10.6% (5/47), respectively (Fig. 4B).

Fig. 4.

Frequency of C. difficile strains identified in human fecal samples. Whole-genome shotgun sequencing was used to identify C. difficile strains in the fecal samples (N=62) from patients with MS infections (n= 15) and SS infections (n=47). (A) Strain frequency in samples containing more than one C. difficile strain (MS infections). (B) Strain frequency in samples containing a single C. difficile strain (SS infection).

Among these 62 specimens, a complete clinical dataset was available for 58 patients. Among these 58 patients, 22.4% (13/58) had MS infection and 77.5% (45/58) had SS infection. In this cohort, being infected with a C. difficile strain expressing binary toxin genes trended to be higher in MS (53.8%, 7/13) infection than SS (27.9%, 12/43) infection (p= 0.0835, Table 3). Of note, all MS infections carrying a C. difficile strain positive for binary toxin gene were followed by a binary negative C. difficile strain. Despite no differences between mixed infection and SS infection, mortality was not reported on those patients with MS infection. Whereas 13.3% of patients with SS infections died post-CDI (Table 3).

Table 3.

Clinical data analysis among sequenced multi-strain (MS) and single-strain (SS) samples.

Patients with complete clinical dataset (N= 58)
Summary characteristics	Multi-strain (MS)a n=13	%	Singlestrain (ss)b n=45	(%)	P=values
Age c
≥ 65	6/13	46.2	19/45	42.2
< 65	7/13	53.8	26/45	57.8	P=0.8009
Recurrence d
Recurrent	3/13	23.1	11/45	24.4
Nonrecurrent	10/13	76.9	34/45	75.6	P=0.9999
Severity d
Severe	4/13	30.8	15/45	33.3
Non-severe	9/13	69.2	30/45	66.7	P=0.9999
Gender d
Male	4/13	30.8	23/45	51.1
Female	9/13	69.2	22/45	48.9	P=0.2248
Death d
Yes	0/13	0.0	6/45	13.3
No	13/13	100.0	39/45	86.7	P=0.3205
Binary+ (cdtA+/cdtB+) c
Positive	7/13	53.8	12/43*	27.9
Negative	6/13	46.2	31/43*	72.1	P=0.0835

^(a)Specimens MS showed the presence of more than one C. difficile strain by using StrainGE.

^(b)Specimens SS showed only one C. difficile strain by using StrainGE.

^(c)Chi-square test or

^(d)Fisher’s exact test was applied.

^*Binary toxin gene classification is available only for 43 strains present in this study.

Overall, our whole-genome shotgun sequencing data validated the presence of MS infection and identified which strains are potentially causing disease in our small cohort.

Discussion

In our cohort, we found that 32.8% of patients with CDI exhibit MS infection, as determined by toxin gene profiles using qPCR from single C. difficile isolates (three isolates per specimen). In fecal specimens with low Ct values for the tcdB gene, which indicate a higher infection burden, we identified 24.2% of MS infections using StrainGE classification in our whole-genome shotgun sequencing dataset of all recovered C. difficile colonies from each fecal specimen. Although MS CDI did not impact recurrence, disease severity and mortality, in the mixed infections cohort, colonization by a non-toxigenic C. difficile strain was more commonly associated with non-severe cases. In addition, whole-genome shotgun sequencing allowed us to identify distinct sequence types among patients infected with SS versus MS, with strains W0023a (ST42, clade 1) and S-0352 (ST2, clade 1) being the most prevalent in each group, respectively.

While we found a 32.8% prevalence of mixed infection in CDI patients, a higher prevalence of MS infection (59%) has been previously reported in C. difficile-infected patients receiving care in a hospital setting in Michigan/USA.¹⁸ Seekatz et al. ¹⁸ isolated 40 colonies from each fecal specimen and used PCR-ribotyping to determine mixed infections. Despite using a large number of isolates (ranging from 6 to 93) from each stool sample and PCR-ribotyping and/or MLST typing to identify C. difficile strains and determine mixed infections in the cohort, other studies have reported a low prevalence of 8% to 17%.^20,34 Although these previous studies have used the same PCR-ribotyping method to confirm the presence of MS, variability in how colonies were selected may have contributed to the reported prevalence. Although most studies reporting the possibility of MS infection in CDI were published more recently, the presence of MS was first described in 1983 by Borriello and Honour,³⁵ who reported the coexistence of cytotoxigenic and non-cytotoxigenic C. difficile strains in the same stool samples from six patients with antibiotic-associated diarrhea and one with non-antibiotic-related diarrhea.³⁵

In our study, recurrence was not significantly higher in mixed infections (21.2%) than in SS infections (18.5%). The study performed in Michigan/USA found a considerably high rate of mixed infections (71%) among patients who developed recurrence compared to nonrecurrent cases (46.2%).¹⁸ A smaller cohort study including 23 patients identified MS infection in 4 patients, of whom half relapsed.³⁶ Overall, mixed infection appeared to be associated with recurrence, which is a major concern in patients recovering from CDI. The underlying mechanisms of recurrences in MS infections may likely be multifactorial, including potential presence of strains with propensity to hyper-sporulation,³⁷ variable susceptibility to antibiotics used for treatment,³⁸ inadequate antibody response to certain strains,³⁹ delays in host microbial or metabolic recovery and others that are yet to be proven in infection models and human studies. Interestingly, we found that the presence of a non-toxigenic strain appears to be protective against recurrence, consistent with the known beneficial effect of specific non-toxigenic strain in CDI.⁴⁰ Given that this association is currently limited to C. difficile-infected patients in the USA, the worldwide significance of mixed infection remains unclear.

Here, we did not find any association between mixed infection and biological variables (age and gender), disease severity, or mortality. Similarly, a previous study performed in a hospital in China did not find an association between MS infections and gender or variables that can impact disease severity (blood leukocyte count, serum creatinine, and albumin levels).³⁴ However, Zhang et al.³⁴ found that age up to 60 years and the presence of ulcerative colitis were risk factors for mixed CDI. In our study, age, which is a major risk factor for CDI severity, was not pointed out as a driver for MS infections. It is possible that certain underlying host conditions or the intestinal milieu are more prone to colonization by multiple C. difficile strains. However, more longitudinal prospective studies are needed to better define risk factors and avoid misinterpretations due to differences in populations and timelines.

Interestingly, we detected non-toxigenic strains in mixed infections but not in SS infections. In addition, the presence of a non-toxigenic strain in patients with MS infections was associated with better outcomes, including non-severe disease and fewer recurrent cases. In accordance with our findings, a previous study conducted in China showed that hospitalized patients colonized with non-toxigenic C. difficile strains were more likely to present a non-severe disease and decreased the risk of CDI development within 90 days.⁴¹ It is possible that the presence of non-toxigenic C. difficile strains in MS infections is linked to less severe disease through a strain-competition mechanism resulting to prevention of colonization and/or expression of virulence by a toxigenic strain. Indeed, colonization with a non-toxigenic strain, NTCD-M3, prevented recurrent CDI in a phase 2 randomized controlled trial.⁴²

Our whole-genome shotgun sequencing dataset confirmed the presence of MS infections in our cohort. Further, we identified that S-0352, a ST2 type strain belonging to ribotype 014 and clade 1 ³², was the most prevalent strain in mixed infection cases. The ST2/RT014 strain is one of the more common strains found in many Asian countries, Australia,¹⁷ and the United States ⁴³, occurring in both community-acquired cases and hospital settings. RT014 has been associated with interspecies transmission and is characterized by a large open pan-genome, numerous prophages, and clinically important antimicrobial resistance elements.⁴⁴ In SS infections, we identified W0023a, a ST42 type strain belonging to ribotype 106 and clade 1,³³ as the most frequent strain. Clinically, infections by RT014 are more likely to be associated with less severe disease than infections by RT106 strains.⁴⁵ It is yet unclear if SS infections tend to be of more virulent strains that may have outcompeted others.

Additionally, within this small group of CDI cases with higher pathogen loads in fecal samples, we observed higher rates of binary-positive strains in MS infections than in SS infections. CDI cases associated with strains producing binary toxins may result in poorer outcomes due to the heightened virulence of these strains.^46,47 Nonetheless, in this cohort, all patients with MS infections survived after CDI. Since every MS infection with binary-positive strains was followed by a binary-negative strain, it is possible that competition between strains reduces the virulence of binary-positive strains, a hypothesis that warrants further research.

Our study is limited by the use of toxin gene profiles to distinguish strains that share identical or highly similar characteristics, potentially leading to missed cases of MS infections. However, we have shown that even with selection of only three distinct isolates in the culture plate (something that can be performed by clinical laboratories), MS infection has already been detected and confirmed by sequencing all colonies in a subset of specimens. Toxin gene profiling of all colonies that grow on a culture plate will be labor-intensive in the current state of clinical laboratories. Future potential alternative methods are high-throughput culture-based and direct DNA sequencing techniques applied to clinical specimens, methods that remain costly and inaccessible for most clinical laboratories as these would require specialized equipment or sending out to research or commercial laboratories with these capabilities. Another important limitation of our study is the lack of granular data on comorbidities, diet, antibiotic exposure, medications, prior healthcare exposure, other medical/surgical interventions, functional abilities, and numerous other clinical factors, as well as intestinal microbiota or metabolic profiles that may influence infection outcomes. A large, longitudinal, prospective study potentially aided by machine learning would be needed to identify all variables of interest in MS infection and SS infection. Future investigations are also needed to determine the impact of MS infection across multiple recurrences, the transmission of genes related to antimicrobial resistance, and the response to therapies such as antimicrobial agents and live biotherapeutic product transplantation.

In conclusion, our findings indicate that MS infection is highly prevalent among patients with CDI. Toxin gene profiles of three different isolates from each fecal specimen, as well as whole-genome shotgun sequencing, are powerful tools to detect mixed infection. There may be specific strains with propensity to be in MS versus SS infections. MS infections containing a non-toxigenic C. difficile strain appear to be more likely to result in better clinical outcomes.

Appendix A. Supporting information

Supplementary data associated with this article can be found in the online version at doi:10.1016/j.jinf.2026.106800.

Data availability

All data and details are provided in the manuscript and supplementary materials. For additional questions regarding data, analytical methods and study materials, please reach out to the corresponding author.