The study was conducted in a 1, 400-bed general teaching hospital in Shanghai, China, from July 2016 to November 2021. Patients (≥18 years of age) with diagnosed acute myeloid leukaemia (AML), acute lymphoblastic leukaemia (ALL), chronic lymphocyte leukemia (CLL), chronic myeloid leukemia (CML), Hodgkin’s lymphoma (HL), and non-Hodgkin lymphoma (NHL), multiple myeloma (MM), myelodysplastic syndromes (MDS), hemophagocytic syndrome (HPS), poems syndrome, were included in the analysis. All hematological patients suspected with Clostridioides difficile infection (CDI) (passage of three or more unformed stools in 24 h or loose stools for more than 3 days) were tested for C. difficile. Glutamate dehydrogenase (GDH) assays and toxins A/B (TcdA/B) assays were performed on patient stool specimens using the C. DIFF QUIK CHEK COMPLETE immunochromatographic test kit (Alere, USA). Stools that tested positive for GDH were subsequently inoculated for 48 hours at 37 °C in an anaerobic environment with C. difficile CDIF selective medium (BioMerieux, France). MALDI-TOF MS was used to identify suspect isolates. Community-acquired Clostridioides difficile infection (CA-CDI) was defined as a CDI episode diagnosed within 72 hours of hospital admission in patients with no documented hospitalization in the preceding six months. Hospital-acquired Clostridioides difficile infection (HA-CDI) was defined as CDI diagnosed ≥4 days after hospital admission or CDI diagnosed within the first 72 hours of admission in patients with a documented hospitalization within the preceding six months. The collection of bacterial isolates from patient specimens and the use of related clinical information were reviewed and approved by ethics committee of Ruijin Hospital with the approval number: KY2023-083. No consent was needed for this study. Duplicate specimens from the same patients were excluded.

MLST was performed according to previous publication (Griffiths et al., 2010). The genes Adk, atpA, dxr, glyA, recA, sodA and tpi were amplified by polymerase chain reaction (PCR) and sequenced. Results submitted to the MLST database (http://pubmlst.org/C.difficile) to obtain the allele profile and sequence type (ST). The evolutionary relatedness of different strains was analyzed through the online website PHYLOViZ (https://online.phyloviz.net/index) based on goeBURST algorithm and the VivaGraphJS library. After harvesting, genomic DNA was obtained and tcdA, tcdB, cdtA and cdtB were examined by a multiplex PCR, as previously described (Pituch et al., 2005).

The minimum inhibitory concentration (MIC) of nine antibiotic, including meropenem, vancomycin, linezolid, metronidazole, moxifloxacin, teicoplanin, rifaximin, daptomycin, clindamycin, was determined by the agar dilution method, according to Clinical and Laboratory Standards Institute (CLSI) guidelines [M11-A8]. ATCC 70057 was used as control. The breakpoints for meropenem, metronidazole, moxifloxacin, clindamycin were based on CLSI recommendations for anaerobes[M100-S35]. The vancomycin resistance breakpoint (MIC >2 mg/L) was defined according to the European Committee on Antimicrobial Susceptibility Testing (EUCAST). The breakpoints for linezolid and rifaximin were based on published literature (Marin et al., 2015; Schwanbeck et al., 2025).

The Gentra Purgene Yeast/Bact. Kit was used to harvest genomic DNA from C. difficile overnight cultures(Qiagen, Germany). The genome sequence of C. difficile was sequenced by Shanghai Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China) according to Illumina instructions The SOAPdenove v2.04 tool (with GapCloser v1.12) was used to assemble the readings (http://soap.genomics.org.cn/). The ST81 cluster genomes were constructed and analyzed using CLC Genomics Workbench, with C. difficile M68 (FN668375.1) as a reference.

Statistical analyses were performed using software, SPSS Version 27.0 (SPSS, USA). Either the two-tailed unpaired Student’s t-test or the Mann–Whitney U-test was used for comparisons of continuous variables with or without a normal distribution. Chi-square were used to compare categorical variables if differences existed among groups. A P value of less than 0.05 was considered statistically significant.

Between July 2016 and November 2021, 227 diarrheal patients with hematological diseases were sampled in a teaching hospital. As shown in Table 1, the investigation included 53 (23.3%) patients with acute leukaemia, 93 (41.0%) patients with multiple myeloma (MM), 65 (28.6%) patients with non-lymphoma Hodgkin’s (NHL), and 16 (7.0%) patients with other hematological diseases. Among the participants, CDI was diagnosed in 49(21.6%) patients, with a mean age of 56.9 years and 53.1% (n=26) females. The prevalence for HA-CDI was 89.8% (44/49) and 10.2% (5/49) for CA-CDI (Not shown). Recurrence is a huge threat to CDI patients. The prevalence of recurrence was 8.1% (4/49) and ST81 was the causative strain in three of the four recurrence cases. CDI was more common in patients with acute leukemia, but the difference was not statistically significant (32.1% vs 16.1%, 20.0% and 25.0%, p = 0.152). Interestingly, 89.8% of CDI patients presented with at least one co-infectious disease. Furthermore, our analysis revealed that hematologic patients with a prior CDI within one year exhibited a significantly higher infection rate compared to those without prior infection (10.2% vs. 0%, p < 0.001). In addition, haemoglobin levels were lower in CDI patients than in the non-CDI group (74.8 ± 26.2 vs 84.0 ± 23.3, p = 0.020). There was no difference between the CDI+ and CDI- groups in other laboratory parameters such as white blood cell count, neutropenia, platelet count, albumin level, and serum creatinine level.

Of 49 toxin-positive CDI cases, 42 underwent microbial analysis, while seven culture-negative cases were excluded. Since toxin detection confirmed CDI diagnosis, culture failure in these seven cases may reflect prior antibiotic use or non-viable bacteria. As shown in Figure 1A, from 2018 to 2021, the detection rate of C. difficile increased year after year. All isolates were defined into 15 STs, with ST81 (38.1%, 16/42) being the most common clone identified, followed by ST3 (16.7%; 7/42), and ST2 (9.5%; 4/42) as illustrated in Table 2 and Figure 1B. The widely epidemic ST54, has a very low detection rate (n=2) as well as ST37(n=1), being considered as responsible for outbreak in China, Europe, and North America (Luo et al., 2018; Li et al., 2019). The phylogenetic relationships of the 42 isolates were analyzed using an online website based on the ST patterns, as shown in Figure 2. Three major clonal complexes (CCs) were identified, with clade 1 being the most common in this study. ST81, the most common clone, was classified as clade 4 (Table 2).

Four toxin genes (tcdA, tcdB, cdtA, and cdtB) were analyzed in our study. The majority of isolates (52.3%, 22/42) were positive for both tcdA and tcdB genes, while 45.3% (19/42) were negative for tcdA but positive for and tcdB genes. All ST81, the most common strains, carried only the toxin B gene. A ST1 isolate carrying four toxin genes (tcdA, tcdB, cdtA, and cdtB) was identified, representing the only strain in our study capable of producing binary toxin.

Interestingly, combining Figures 1A, B, ST81 was revealed to be a contributor of the increase in CDI in 2019 (8/11 isolates), whereas 2021 isolates exhibited greater genetic diversity. Thus, we perform an epidemiological investigation to show spatial and temporal clustering of ST81 cases. Patient 2019-P1 (Bed 25) and 2019-P7 (Bed 03)was housed in two separate rooms from other cases. Furthermore, 2019-P5 and 2019-P8 (Beds 10 and 12, same ward) had a two-month interval between admissions (Figures 3A, B). In contrast, patients 2019-P2, 2019-P3, and 2019-P4 (Beds 17, 13, and 13, respectively) shared the same ward. Notably, 2019-P3 developed CDI following May 2019 hospitalization; one month later, 2019-P4 was admitted to the same bed (Bed 13) and contracted an ST81 strain. Phylogenomic analysis demonstrated that these isolates shared high genetic relatedness (Figure 3B), indicating that they were extremely similar, suggest probable transmission. Although 2019-P6 (Bed 19) was admitted >6 months later after 2019-P2 (Bed 17), since their proximity in adjacent rooms and phylogenetic clustering of the isolates suggest potential environmental persistence or undetected transmission. Collectively, those may indicate that ST81 was spreading in patients with hematologic disease.

All C.difficile isolates were evaluated for antimicrobial resistance to nine antibiotics (meropenem, vancomycin, linezolid, metronidazole, moxifloxacin, teicoplanin, rifaximin, daptomycin, and clindamycin). Table 3 shows the antibiotic resistance patterns of all C. difficile strains. 64.3% of the strains were resistant to clindamycin, and high degree of resistance persisted from 2016 to 2021 (MIC≥32 mg/L), with a MIC50 = 32 mg/L and a MIC90 = 128 mg/L (Figure 4A). Resistance to moxifloxacin was 31%, with a decreasing trend (Figure 4B). Meropenem and rifaximin had resistance rates of 9.5% and 7.1%, respectively. No isolate showed positive for resistance to vancomycin, linezolid, metronidazole, teicoplanin, or daptomycin. The dominant strains ST81 were more resistant than non-ST81 strains (p = 0.05). More than half of them had at least two drug resistances, with 31.3% having two and 18.8% having three. ST81 had a much higher rate of resistance to moxifloxacin than the other STs (50% VS 19.2% p = 0.036). No statistically significant differences in resistance rates were observed between ST81 and other STs for meropenem (6.3% VS 2.7%), rifaximin (12.5% vs 0), or clindamycin (75% VS 53.8%) (all p>0.05).