Gut diversity and the resistome as biomarkers of febrile neutropenia outcome in paediatric oncology patients undergoing hematopoietic stem cell transplantation

The gut microbiota of paediatric oncology patients undergoing a conditioning regimen before hematopoietic stem cell transplantation is recently considered to play role in febrile neutropenia. Disruption of commensal microbiota and evolution of opportune pathogens community carrying a plethora of antibiotic-resistance genes play crucial role. However, the impact, predictive role and association of patient´s gut resistome in the course of the therapy is still to be elucidated. We analysed gut microbiota composition and resistome of 18 paediatric oncology patients undergoing hematopoietic stem cell transplantation, including 12 patients developing febrile neutropenia, hospitalized at The Bone Marrow Transplantation Unit of the National Institute of Children´s disease in Slovak Republic and healthy individuals (n = 14). Gut microbiome of stool samples obtained in 3 time points, before hematopoietic stem cell transplantation (n = 16), one week after hematopoietic stem cell transplantation (n = 16) and four weeks after hematopoietic stem cell transplantation (n = 14) was investigated using shotgun metagenome sequencing and bioinformatical analysis. We identified significant decrease in alpha-diversity and nine antibiotic-resistance genes msr(C), dfrG, erm(T), VanHAX, erm(B), aac(6)-aph(2), aph(3)-III, ant(6)-Ia and aac(6)-Ii, one week after hematopoietic stem cell transplantation associated with febrile neutropenia. Multidrug-resistant opportune pathogens of ESKAPE, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae and Escherichia coli found in the gut carried the significant subset of patient’s resistome. Over 50% of patients treated with trimethoprim/sulfamethoxazole, piperacillin/tazobactam and amikacin carried antibiotic-resistance genes to applied treatment. The alpha diversity and the resistome of gut microbiota one week after hematopoietic stem cell transplantation is relevant predictor of febrile neutropenia outcome after hematopoietic stem cell transplantation. Furthermore, the interindividual diversity of multi-drug resistant opportunistic pathogens with variable portfolios of antibiotic-resistance genes indicates necessity of preventive, personalized approach.


Febrile neutropenia treatment with antimicrobials
The treatment regimens involved a combination of broad-spectrum antibiotics.The choice of antibiotics could reflect the underlying conditions, severity of infection or antibiotic susceptibility patterns.Standard antibiotic therapy in febrile neutropenia included Piperacillin-Tazobactam + Amikacin.Gram-positive bacteria coverage was added if central venous line (CVL) infection was suspected, skin infection was present, or fever last more than three days.For this purpose, Piperacillin, Tazobactam, and Amikacin were used.Meropenem was used if Piperacillin or Tazobactam-resistant bacteria were detected or febrile neutropenia last or worsened on the standard antibiotic combination for a couple of days.Linezolid was preferred over teicoplanin if soft tissue or lung infection with Gram-positive bacteria was present or possible.Trimethoprim-sulfamethoxazole before HSCT was used two or three days per week as prophylaxis of Pneumocystis jirovecii infection.It was paused during aplasia and re-introduced after stable engraftment on new hemopoiesis.The combination of applied antibiotic treatment varied across patients reflecting the cultivation-based microbiological findings (Supplementary Table 2.).

Sample collection, DNA isolation and high-throughput sequencing
Stool samples were collected and stored at − 80 °C until further processing.The extraction of DNA was provided by Zymobiomics DNA/RNA isolation kit (ZymoResearch, Irvine, CA) according to the manufacturer's protocol.Extracted DNA was quantified with Qubit dsDNA High Range Assay (Thermo Fisher Scientific, Waltham, MA, USA) using Qubit 4.0 Fluorometer (Invitrogen, Carlsbad, CA, USA).DNA was eluted and stored at − 20 °C.DNA libraries were prepared using Nextera kit (Illumina, San Diego, CA, USA) according to the manufacturer's instructions.Fragmenation was followed with indexing PCR and the amplified libraries were purified using Agencourt AMPure beads (Beckman Coulter, Indiana, USA).The quality and the quantity of final libraries were established by Qubit 4.0 Fluorometer and chip electrophoresis using Agilent Bioanalyser 2100 system (Agilent Technologies, Santa Clara, CA, USA).Samples were pooled in equimolar ratio and sequenced on Illumina NextSeq (Illumina, San Diego, CA, USA) platform (2 × 150 bp).All steps were performed according to the manufacturer's tutorials.

Identification and taxonomic classification of resistant bacteria
The individual contigs with detected resistance genes were analysed through BLAST 43 (optimized for highly similar sequences (megablast)) using a nucleotide collection database, and the sequence with the highest identity (min.99%) and coverage score was considered the closest relative.The position of the resistance gene within the bacterial genome (chromosome/plasmid) was recorded.

Bacteria-cancer drug interaction
ResFinder-identified resistant bacteria were searched in the databases http:// pharm acomi crobi omics.com/ and http:// www.aiddl ab.com/ MASI/.All information on bacteria-drug interactions came exclusively from databases.Only bacteria associated with anticancer therapy taken by studied patients were selected.Patients' data were screened for taken antibiotics.The effectiveness of the therapy taken, and post-transplant FN was associated with the resistant bacteria of the gut.

Statistical analysis
The data were analysed by SPSS (IBM Corp. Released 2012.IBM SPSS Statistics for Windows, Version 21.0.Armonk, NY: IBM Corp.) as well as the LefSe tool 44 .Shapiro-Wilk test was used for the normality testing.Based on this result, Kruskal-Wallis or Mann-Whitney test, respectively, were used as non-parametric tests, while the t-test or ANOVA were used as parametric tests.The tests were provided with established α-level.Correlation analysis was based on the Spearman rank correlation coefficient.Interaction networks were visualized with Gephi (Gephi version 0.10.1, www.gephi.org).The Gephi network was made using 'Force Atlas' layout and ranking used ' Average Path Length' algorithm with 'Betweeness Centrality' parameter.'Modularity class' algorithm was used for the community detection and the values were used to colorize communities.GraphPad (GraphPad Prism version 8.0.0 for Windows, GraphPad Software, San Diego, California USA, www.graph pad.com) was used for data visualization.Moreover, the Krona pie chart 45 was used for microbial composition visualization.The clust-Vis tool 46 was used for Principal Component Analysis and heatmap visualization using average clustering for antimicrobial genes based on correlation distance and for patient´s samples Ward clustering based on Manhattan clustering distance.Adobe Illustrator 2020 enabled us picture creation and editing.

Ethics approval
The study adhered to the ethical principles outlined in the Declaration of Helsinki for experiments involving human beings and its later amendments, and all steps of this study were approved by the Ethics Committee of the National Institute of Child Diseases (NICD-25/4/18).The legal representatives of all participants provided and signed written informed consent.

Change in alpha-diversity of gut microbiota one week after HSCT is associated with febrile neutropenia
Shannon and Simpson indices were used for alpha-diversity determination.The diversity of the healthy gut microbiome (median ± interquartile range) defined by Shannon index (4.27± 3.83) and Simpson index (0.96 ± 0.93) represented the benchmark for the evaluation of the deviated diversity observed in gut of patients.Significantly lower alpha diversity defined by both indices characterized both groups of patients, either with or without FN developed after HSCT, during the monitored period compared to healthy individuals (p ≤ 0.01).Although there was no significant difference between the two cohorts of patients within the same period of analysis, overall species richness (Simpson index; d-7/d + 7/d + 28) of the gut microbiome was less favourable for patients with FN (0.77 ± 0.56/0.47± 0.43/0.48± 0.29) compared to the other group (0.88 ± 0.78/0.80± 0.73/0.63± 0.41) pinpointing an important role of the alpha diversity in association with posttransplant complication.Furthermore, one week after HSCT, patients developing FN exhibited a significant decrease in alpha diversity app.by 60% compared to the timepoint before HSCT (Simpson; p = 0.016), whereas no significant change was observed in patients without FN (Fig. 1).The analysis of Shannon index provided similar results (p = 0.002) proving remarkable gut microbiota depletion after HSCT preceding FN (2.15 ± 1.47/1.03± 0.95/1.05± 0.72) compared to the group without fever development (2.47 ± 2.35/2.22 ± 1.97/1.57± 1.22).We haven´t detected any specific antimicrobial or antimicrobials combination that would be associated with selective depletion of gut microbiota after HSCT in patients developing fever.Although, consequently individually set therapy with antimicrobials was applied to FN treatment.Significantly different beta diversity discriminated healthy individuals from patients either developing or not FN before HSCT (p ≤ 0.001) as well as within one week after HSCT (p ≤ 0.001), while no difference was observed between the two cohorts of patients before HSCT (p ≤ 0.28) or after HSCT (p ≤ 0.25).The intragroup diversity did not change after treatment in any of involved groups of patients (with FN p = 0.4; without FN p = 0.5).
The bacterial profile of the patient's gut microbiome either before or after HSCT revealed a significant decrease (p ≤ 0.05) in the relative abundance of Bacteroidota, Actinomycetota, and Thermodesulfobacterota and an increase in Bacillota.Furthermore, a disbalance of Fusobacteriota and a decrease of Campylobacteriota were detected in a subgroup of patients mainly observed before and one week after HSCT (Supplementary Table 4).

Impact of HSCT on gut microbiota composition one week after HSCT
The analysis of the family level (Supplementary Table 5) provided more specific insight into the changes in the patient's gut microbiota.
One month after treatment, E. faecium remained a dominant species in gut microbiota, with an abundance of 30%.Additionally, there was a pronounced increase in the abundance of Enterococcus faecalis by 2.2%, and a similar increase of 7.1% was observed for S. haemolyticus.All the mentioned bacterial species, together with Granulicatella adiacens (2.6%), Rothia mucilaginosa (1.2%), and Bacteroides fragilis (5.8%) were detected with an abundance higher than 1% (Fig. 3).

Gut microbiome of healthy children
The gut microbiota of the healthy group was characterized by the most abundant phylum Bacteroidota (45%) and phylum Bacillota (40%).Actinomycota was detected with an abundance of 10%, Pseudomonadota was identified with an abundance of 3%, and Verrucomicrobiota with 1%.
While patients showed a significant increase in families associated with potential opportunistic pathogens, the gut microbiome of healthy controls was primarily dominated by Bacteroidaceae and Lachnospiraceae, accounting for 20% of the total abundance, followed by Oscillospiraceae (15%), Prevotellaceae (11%) and Bifidobacteriaceae (7%).

Dysbalanced SCFA-producing bacteria enable emerging of opportune pathogens
The significant changes (p ≤ 0.05; > threefold change) of the gut microbiota composition were identified on the family level of the bacterial taxonomy.The comparison of the individual time points showed a decrease in Bifidobacteriaceae, Eubacteriaceae, and Erysipelotrichaceae one week and one month after HSCT compared to samples before HSCT.Moreover, for samples taken one month after HSCT, depletion of Lachnospiraceae, Oscillospiraceae, Coprococcaceae, and Clostridiaceae was typical.
The comparison of individual time points revealed notable changes in the relative abundance of various bacterial families compared to healthy subjects.The decrease was observed in Bacteroidaceae, Prevotellaceae, Oscillospiraceae, Bifidobacteriaceae, Eubacteriaceae, Odoribacteriaceae, Sutterellaceae, and Rikenellaceae across all time points of sampling.Additionally, samples one week and one month after HSCT exhibited a depletion in Erysipelotrichaceae and Lachnospiraceae.Akkermansiaceae depletion was specifically observed before and one month after HSCT.Furthermore, the gut microbiota of patients one week after HSCT displayed a decrease in Clostridiaceae.On the other hand, there was a consistent increase in Enterococcaceae across all intervals.Notably, samples before HSCT also showed an increase in Streptococcaceae, while samples one week after HSCT exhibited an increase in Staphylococcaceae (Fig. 1).
Akkermansia muciniphila was detected in all samples of healthy individuals (0.003%-7.3%) as well as in samples of patients except for one, one week after HSCT.Before HSCT no significant difference (p = 0.16) in relative abundance between patients with FN (0.0004%-0.02%)and neutropenia without fever (0.0003-12.3%)was detected, while one week after HSCT trend to higher relative abundance (p = 0.08) in the cohort of patients without FN (without FN 0.0003-0.06%;with FN 1.10 −12 − 0.007%) was identified.We have also investigated the change in relative abundance in each individual in the course of the therapy, but also no significant differences in the ratio before HSCT/one week after HSCT between the two cohorts of patients was found.Changes in the relative abundance of Akkermansia did not indicate the onset of fever during neutropenia in paediatric patients in conditioning regimen undergoing HSCT including in this study.www.nature.com/scientificreports/

Universal antimicrobial prophylaxis treatment spanning HSCT stumble upon gut microbiome resistome
Microorganisms carrying resistance genes can directly impact the efficacy of oncology treatments.They can activate prodrugs into their active forms, but there is also a risk of increased toxicity.Additionally, the gut microbiota can indirectly influence drug metabolism and toxicity by competing with bacterial-derived metabolites for xenobiotic metabolism pathways or modulating the host's metabolic systems 48 .The resistant bacteria could play a crucial role in these mechanisms through their possibility to survive antimicrobial prophylaxis.Before HSCT, the resistome of the patients comprised a total of 57 different resistance genes (Supplementary Table 6).The most frequently found genes belonged to aac(6')-Ii detected in 8 patients, followed by msr(C) in seven patients, and tet(W) together with sul2 in six patients.We have identified genes aph(3″)-Ib, tet(O), erm(B), cfxA5, dfrG in five patients, while aac(6′)-aph(2″), ant( 6)-Ia, aph( 6)-Id, blaTEM-116 in 4 patients.
One week after HSCT, the prevalence of aac(6′)-Ii and msr(C) remained (eight patients), while dfrG and blaTEM-116 emerged as the second most abundant ones (seven patients).The number of patients with ant( 6)la carried resistance increased by 50% (six patients), while erm(T), erm(B) together with aph(3')-III were found in four patients.However, a significant decrease in tet(W) prevalence in the gut microbiome of patients was observed since this gene was detected exclusively in one gut microbiome sample one week and one month after HSCT.
Furthermore, gene aac(6')-Ii coding for chromosomal-encoded aminoglycoside acetyltransferase and aminoglycoside nucleotidyltransferase encoded by ant( 6)-la gene were detected in seven samples, even one month after HSCT.We have identified gene msr(C) conferring resistance to macrolides in six patients, followed by dihydrofolate reductase encoded by the dfrG gene in five samples.Compared to the one week after HSCT, concerning erm(B) and aph(3')-III genes of resistance, we have identified no changes (four patients).
Based on the additional statistical tests, we have found cfxA5 and tet(W) more often in the gut microbiota of patients before HSCT.At the same time, mef(A) was identified only one week after HSCT in three of 16 patients (p < 0.05).Furthermore, tet(W) significantly differed between samples, detected more often before HSCT (38% of patients before HSCT, 6% one week after HSCT).We have not detected any significant changes in gut resistome one week and one month after HSCT.These results indicate a higher abundance of specific antimicrobial resistance genes before HSCT, suggesting a potential association of the conditioning regimen with the occurrence of resistant bacteria (Fig. 4).

Antimicrobial genes in genomes of opportune pathogens and commensal bacteria
The contigs containing identified resistance genes were analysed using the BLAST tool to assign taxonomic classification and determine their origin within bacterial strains.
Our study identified fifty-two bacterial genera and five plasmid vectors carrying antibiotic-resistance genes.In healthy children, mostly commensal bacteria carried antimicrobial resistance.These bacteria involved members of phylum Bacteroidota (Parabacteroides spp., Bacteroides spp., Phocaeicola spp., Chryseobacterium spp.) but also Bacillota (Clostridium spp., Dysosmobacter spp.) or Actinomycetota, represented by C. acnes as the only identified species.The gut microbiota of the patients was rich in resistance genes observed mainly in opportunistic pathogens such as E. faecium, E. faecalis, E. coli, K. pneumoniae, E. hormaechei, S. enterica, S. aureus, S. haemolyticus, C. difficile or Campylobacter coli.
A closer analysis of patients with FN showed that in the period before HSCT, these patients had resistance genes more frequently carried by Enterococcus (E.faecium, E. faecalis, E. hirae), Staphylococcus (S. aureus, Staphylococcus pseudintermedius), Bacteroides (B.fragilis, Bacteroides thetaiotaomicron) K. pneumoniae, Streptococcus (S. gallolyticus, S. suis) S. enterica, and C. difficile compared to patients without FN Patients without FN had resistance genes carried by S. agalactiae, P. distasonis, Enterobacteriaceae bacterium, A. muciniphila, C. coli, and C. perfringens.One week after HSCT, E. faecium and E. faecalis were identified with higher prevalence in patients compared to samples before HSCT, which suggested an increase in the abundance of these species following HSCT.Additionally, during the same time, these species were more prevalent in patients with FN In patients with FN, resistance genes were observed in Streptococci species (S. gallolyticus, S. pneumoniae, S. agalactiae), E. coli, and S. enterica.In contrast, patients without FN had genes carried by S. aureus.After one month, bacteria of the genera Enterococcus, and Bacteroides, together with species K. pneumoniae, S. gallolyticus, and E. bacterium, carried resistance genes in patients with FN S. haemolyticus, S. epidermidis, S. pneumoniae, S. suis, S. infantis, A. muciniphila, E. coli, and C. coli carried resistance genes one month after HSCT in the group of patients without FN (Fig. 5).

Genotypic prediction of phenotypic resistance to antimicrobials of gut bacteria
A total of 84 genes were identified within 20 detected groups of antimicrobial resistances.From 14 healthy controls, only eight individuals carried genes for antimicrobial resistance.8/14 (57%) of all healthy individuals were resistant to tetracyclines; in 3/14 (21%), resistance to β-lactams was recorded, and in 1/14 (7%), resistance to unknown β-lactams and folate pathway antagonists were detected.
Detected genes were mainly present in the E. faecium strains.Moreover, erm(B) was also detected in Staphylococcus pseudintermedius and E. faecalis strain CVM N60443F, while aac(6)-aph (2) in S. hominis strain FDAAR-GOS_661 and erm(T) in S. gallolyticus strain FDAARGOS_666.Correlation analysis revealed that the resistome of most patients with FN clustered together, suggesting a potential association between the gut microbiome resistome and FN as one of the HSCT outcomes (Fig. 7).Our finding implies that these genes may serve as markers indicating the onset of FN.
For the rest of patients with FN, another distinct cluster of microbiota resistome characterized by a more diverse profile of resistance genes including blaCTX-M-1, aadA1, tet(B), dfrA1, aph(3)-Ia, catA1, fosA, OqxB, sul1, OqxA, blaSHV-1, tet(D), sul2, blaTEM-1B, aph(3)-Ib, aph(6)-Id, tet(M), tet(A), aac(3)-IV, tet(S), blaACT-15, dfrA14, aac(3)-IIa, qnrB1, blaOXA-1, aac( 6)-Ib-cr, blaCTX-M-15 and blaOXY-6-2 was identified.Figure 6.Graphical visualization of the spread of antimicrobial resistance genes annotated by the predicted antimicrobial group phenotype in paediatric oncology patients before (d-7), one week after (d + 7) and four weeks after (d + 28) HSCT with focus on febrile neutropenia outcome after HSCT (FN -patients with febrile neutropenia, xFN -patients without febrile neutropenia).Rows are centred; unit variance scaling was applied to rows.Both rows and columns were clustered using correlation distance and Ward linkage.www.nature.com/scientificreports/On the other hand, it was evident that the gut microbiota resistomes of patients who did not develop FN also clustered and exhibited significantly lower diversity or absence of resistance genes than those with FN.This observation suggests that patients with FN may possess a broader range of resistance genes, potentially correlating with the development of FN.Different clusters of the resistance genes may serve as biomarkers with potential in personalized approaches.
The patients without FN had more genes encoding resistance to antibiotics carried by multidrug-resistant opportunistic pathogens, mainly after HSCT (one week or one month after HSCT).In comparison, before HSCT Before HSCT, earlier occurrence of resistance genes suggests a link between the earlier invasion of potentially pathogenic resistant bacteria and the FN outcome.Furthermore, the gut microbiota of patients without posttransplant complications encompassed multidrug-resistant bacteria mainly after HSCT, indicating an association between the time of the emergence of multidrug-resistant gut bacteria and FN.These findings highlight the importance of understanding the role of a subset of resistance genes, or the whole resistome, in the context of a patient's complication development and could provide insight for a more personalized approach to patient treatment.

Common inflammation parameters and gut microbiome
Specific bacterial families, including Enterococcaceae, Streptococcaceae, and Staphylococcaceae, are commonly found in human gut.However, they can transform into pathobionts capable of translocating across the gut barrier into the blood resulting in bloodstream infections.This transition from commensals to opportunistic pathogens is a significant concern, particularly for immunocompromised individuals, who are at higher risk of infection.Moreover, these bacterial families have been identified as reservoirs of antibiotic resistance genes, further complicating their clinical outcome.The members of these bacterial families were often associated with higher mortality after HSCT and FN as well as GvHD development [51][52][53] .
We aimed to investigate the relationship between available inflammation parameters (Supplementary Table 9) included in routine analysis and resistant bacterial taxa before and after HSCT.We looked for potential links between gut bacteria and fever as a stimulus, but our analysis did not reveal any differences in CRP levels between patients with or without FN.Moreover, CRP did not show any correlation to any bacteria.However, we did find a positive correlation between Streptococcaceae and procalcitonin (PCT) levels before HSCT, suggesting a potential link between these bacteria and systemic inflammation.Interestingly, one week after HSCT, a distinct pattern emerged where Streptococcaceae were negatively correlated with lymphocyte and neutrophil counts.These findings shed light on the complex interplay between gut bacteria and inflammation and could help to stimulate future research on treatment and prevention of FN.It appears that there is a potential relationship between the loss of lymphocytes and increased abundance of Streptococcus 54 , which could be a predictor of infection development.In addition, there is evidence that Enterococcaceae may be involved in the development of FN during the post-transplant period.Based on Spearman rank correlation, one month after HSCT, there was a notable link between Streptococcaceae and Staphylococcaceae and the neutrophils to lymphocytes ratio (NLR), suggesting their potential role in driving an imbalanced immune response.These findings highlight the complex interactions between specific gut microbiota members and the host, particularly in immunocompromised patients (Fig. 8).However, other bacterial families that carry resistance genes were not significantly correlated with clinical parameters.

Personal gut resistome of patients can be associated with cancer treatment outcome
Patients undergoing cancer treatment that included antibiotics before and after HSCT exhibited resistance to specific antibiotic drugs.Biseptol (trimethoprim / sulfamethoxazole), Amikacin, and Tazocin (piperacillin / tazobactam) dominated among the antibiotics prescribed before HSCT, while after HSCT, Amikacin, Tazocin, and Targocid (teicoplanin) mainly were prescribed.Of all patients treated with antibiotics (18) mentioned earlier, resistant bacteria were found in the gut microbiome of 13 of them.More importantly, most of them, ten patients, developed FN after HSCT.Closer investigation of the gut microbiome of treated patients revealed resistance to Biseptol (trimethoprim/sulfamethoxazole; eight out of 10 patients treated; 80%) and Tazocin (Piperacillin/ Tazobactam; 4 out of 7 patients; 57%) before HSCT, while resistance to Amikacin (Amikacin; three out of five patients; 60%) and Tazocin (four out of nine patients; 44%) after HSCT.Even though each patient sustained a different portfolio of bacteria in the gut microbiome, each carrying different resistance genes, the resistome of gut bacteria represented the standard feature.Bacteria-carrying genes for resistance to trimethoprim/ sulfamethoxazole provided by drfG, drfA1, dfrA14, and sul2, were identified exclusively before HSCT.The dfrG gene carried by Enterococcus spp. was detected in the gut microbiome of two patients, the dfrA1 gene, carried by S. enterica, in one individual, and S. enterica together with K. pneumoniae encoded resistance to trimethoprim / sulfamethoxazole by dfrA14 gene in two individuals.Both microbes, identified in the gut microbiome of four other patients, also possessed sul2-encoded resistance.B. fragilis was the only commensal bacteria predicted to provide resistance to trimethoprim/sulfamethoxazole in one patient.
Potential antibiotic resistance was encoded by six bacterial genera-Enterococcus, Staphylococcus, Klebsiella, Escherichia, Salmonella, and Bacteroides.Most of the resistance genes were carried by K. pneumoniae and E. coli (Table 1).
It has been found that certain bacterial species can have an impact on cancer therapy.The Pharmacomicrobiomics and Microbiota-Active Substance Interaction Database (MASI) were searched to identify bacterial species carrying antimicrobial resistance genes.Within the gut microbiome of all patients, only resistant bacteria B. fragilis and R. intestinalis were found to have a positive interaction with the anticancer drug methotrexate (MTX), decreasing its toxicity.Prevotella, on the other hand, was associated with increased drug activity.www.nature.com/scientificreports/It is interesting to notice, that only three of the thirteen who received MTX showed the presence of these particular bacteria in their gut microbiome, while two of them developed FN: one patient with two strains of B. fragilis (B.fragilis DCMOUH0018B, B. fragilis S14) and one patient with B. fragilis FDAARGOS_763.Conversely, the in the gut microbiome of the third patient who did not exhibit FN B. fragilis BFG-79 and B. fragilis DCMOUH0017B, as well as R. intestinalis L1-82 were identified.Although we did not identify any pattern, these preliminary results show importance of considering also commensal bacteria carrying genes of antimicrobial resistance as important players also in drug metabolism.However, their potential needs to be further elucidated.

Discussion
The process of HSCT profoundly impacts the microbial composition and diversity in the gut.The gut microbiome of paediatric oncology patients changes due to the conditioning regimen, that can include chemotherapy or total body irradiation.Usually, the presence of commensal bacteria prevents colonization of the gut by pathogens.However, the use of antibiotics can alter not only the gut microbiota leading to an imbalance known as dysbiosis but also causes changes in the gastrointestinal barrier 55 .It has been observed that a decrease in gut microbial alpha-diversity is linked to higher mortality rates in patients undergoing allo-HSCT 56,57 or in correlation to GvHD associated mortality 58 .The investigation by Masetti et al. (2022) found a consistent decline in diversity among patients with extended fever duration during the post-HSCT timeframe 17 .Similarly, Rattanathammethee et al. (2020) also noted a significant decrease in alpha-diversity during the fever neutropenic phase 59 .These findings indicate that altered gut microbial diversity may have potential implications on patient outcomes.
Standard treatment protocols containing broad-spectrum antibiotics lead to an increase in the number of resistant bacteria 60 .Moreover, the patients are more susceptible to fungal or bacterial infections, including Klebsiella spp., Enterococcus spp., Staphylococcus spp., and Enterobacter spp. as the infectious agent 61 .The presence of opportunistic pathogen bacteria is a matter of significant concern.This study identified bacteria in a considerable proportion of the gut of HSCT patients, highlighting the ones with genomes enriched in resistance genes with a possible impact on the therapy.The emergence of resistant strains could be attributed to several factors, but one of the most likely culprits is prolonged antimicrobial therapy.This is often part of the prophylaxis for patients undergoing HSCT and the suppression of the immune system.E. faecium was consistently the most common bacterial species found in the gut microbiome of patients, regardless of the time intervals studied.Based on the available data, it appears that the increased abundance of the genus Enterococcus is a significant factor in the gut microbiota of patients undergoing HSCT [62][63][64][65] .Enterococci belong to opportunistic pathogens of the human gut www.nature.com/scientificreports/microbiota and are known for their multi-resistance, which can pose a significant problem in healthcare.It is important to note that in patients undergoing HSCT, Enterococcus has been found to have a positive with both mortality and the development of GvHD.The study of Smith et al. (2022) highlighted Enterococcus as an indicator of clinical outcomes 66 .E. faecium was not only the most prevalent bacteria in the cohort of patients but was identified as the primary carrier of resistance genes in the majority of patients; however, it was not observed in only three patients.The depletion in beneficial SCFA-producing bacteria, such as Bacteroidaceae, Oscillospiraceae, or Eubacteriaceae, was detected through all the samples.It has been found that a reduction in SCFA in patients can have a significant impact on their health.This can lead to changes in glucose homeostasis, gut integriy, and immunomodulation 67 , which in turn increases the risk of bloodstream infection 68 .It is important to note that SCFAs have been extensively researched regarding their implications for GvHD, where they have been shown to protect against the development of chronic GvHD.Conversely, the depletion of anti-inflammatory members of the Clostridia class has been associated with an increased risk of developing these complications 69 .Interestingly, Akkermansiaceae were depleted in the gut microbiota of patients before and one month after HSCT.On the contrary, one week after HSCT, the patient's gut microbiota showed a higher relative abundance of Akkermansiaceae compared to healthy subjects.The most known member, A. muciniphila, has been shown to play a crucial role in maintaining the integrity of the gut barrier 47 while also breaking down mucin 70 .A. muciniphila has been found to be more prevalent in patients with inflammatory diseases like inflammatory bowel diseases 71 , which could indicate an increased effort to repair the intestinal barrier.Furthermore, Akkermansia expansion in the gut of patients with acute leukaemia predicted a higher risk for neutropenic fever, probably through the regulation of microbiota-host metabolic interaction by modulating the mucosal interface 72 .The patients were often found with an injured gut barrier that can lead to bacteria translocation and system infection, which can ultimately result in sepsis.Monitoring the microbial composition and resistance occurrence in HSCT patients is crucial due to their suppressed immune system, which can result in neutropenia and a significant reduction in the number of neutrophils responsible for fighting bacterial infections.Even with broad-spectrum antibiotics, resistant microorganisms can still grow, as evidenced by our results and other studies.Opportunistic pathogens like E. faecium have been confirmed to persist even after antibiotic prophylaxis, leading to their dominance and, in some cases, monodominance.It is interesting to note that in our study, in the gut microbiota of patients undergoing HSCT, both with and without FN, E. faecium dominated the gut bacterial resistome.www.nature.com/scientificreports/As expected, the patient's resistance profile differed from that of healthy individuals carried by commensal bacteria.In the healthy subjects, the tetQ gene conferred the highest prevalence of resistance to tetracyclines, followed by resistance to unknown β-lactams, β-lactams, and sulfamethoxazole.
We have discovered the significant role, particularly of the Bacteroides strains.These strains possess a remarkable resistance to β-lactams and tetracyclines, thanks to the crucial involvement of genes such as cfxA, cfiA, cepA, and tetQ 73 .Bacteria can acquire resistance genes through mutations that spread vertically during cell division or through the horizontal gene transfer process, where resistance genes are carried on mobile genetic elements.These horizontally acquired genes can further spread vertically, contributing to the spread of resistance genes in microbial ecosystems like the gut microbiota 74 .It is worth noting that up to 80% of Bacteroides spp.isolates contain a CTnDOT conjugative transposon that carries the tetQ gene and ermF, a macrolide-resistance gene 75 .
Additionally, we have discovered different clustering patterns of resistomes of patients with FN prevalently carried by E. faecium, mainly resistant to aminoglycosides.It is interesting to note that Enterococci have an intrinsic low resistance to aminoglycosides because of their relative impermeability to aminoglycoside antibiotics that are associated with species-specific chromosomal aac(6′)-Ii gene encoding a 6'-N-aminoglycoside acetyltransferase 76 .ESKAPE is a group of pathogens responsible for most hospital infections.These pathogens, including E. faecium, S. aureus, K. pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter, are becoming increasingly resistant to antibiotics, such as aminoglycosides 77 .Among the gram-negative bacteria, aac(6′)-Ib is one of the most common [78][79][80] .Because of its localization on transposable elements, plasmids, or phages, based on the ability of these genetic elements to spread, theoretically, it can be acquired by any gram-negative bacteria.Our results show that gram-negative Enterobacteriaceae, specifically K. pneumoniae and E. coli, were prevalent resistant bacteria within the patient group and carried genes encoding resistance primarily to β-lactams and aminoglycosides.These bacterial species are frequently isolated from patients with FN 76,81 .However, high-level resistance to aminoglycosides mediated by aminoglycoside-modifying enzymes has been increasing in recent years.These enzymes are encoded by several aminoglycoside resistance genes, including aac(6′)-aph(2′′), aph(3′)-III, ant(6′)-Ia, and aph(2′′)-Ib 82 .Historically, aac(6′)-aph(2′′) was thought to be restricted only to gram-positive enterococci and staphylococci; nowadays, it was detected in Campylobacter, Streptococcus, Clostridium perfringens, or Lactococcus garvieae.Together with aph(3′)-III commonly found in S. aureus and E. faecalis, ant(6′)-Ia detected either on plasmid or chromosome of S. aureus, S. epidermidis, E. faecium and E. faecalis, as well as Streptococcus mitis, S. suis., and aph(2′′)-Ib of E. faecium and E. coli they encode all three types of aminoglycoside modifying enzymes 83 .Specifically, in the gut microbiome of patients undergoing HSCT, we have identified aac(6')-aph(2'') in the genome of E. faecium encoding bifunctional aminoglycoside-inactivating enzymes that possess activity from both enzyme components, thereby providing resistance to the combination of antibiotics; it may include acetylation, phosphorylation or nucleotidylation activity.In other E. faecium, aph(3')-III encoding O-phosphotransferase responsible for resistance to kanamycin and neomycin was identified.Within the Enterobacteriaceae family, a specific type of aac(6')-Ib, aac(6')-Ib-cr variant can induce double resistance to aminoglycosides and fluoroquinolones.Except for E. coli 84 and Enterobacter 85 , also in blood isolates, K. pneumoniae has been identified as a carrier of this type of resistance 86 .These findings correlate with our results, including an increased abundance of Enterococcus spp. that has been observed in the gut microbiome of patients post-HSCT and suggests the possibility of gut barrier translocation and increased risk of bloodstream infection 87 .
Aminoglycosides are also combined with β-lactams or vancomycin to treat some gram-positive pathogens, mainly staphylococci but also enterococci 88 .The gene of resistance to vancomycin, VanHAX, typical for patients with FN, is usually carried also by Enterococcus spp. 89.It was associated with increased mortality in patients with FN due to the prolonged duration of bacteraemia 90 .Tight correlation of resistance genes, including msr(C), dfrG, emr(T), and erm(B), can suggest their colocalization within one genome or cooccurrence of bacterial species within the gut.msr(C) gene is a chromosomal-encoded ABC-F subfamily protein carried by E. faecium that confers resistance to erythromycin and other macrolide and streptogramin B antibiotics.In a study by Mc Gann (2023), a multidrug-resistant strain of E. faecium carrying resistance genes dfrG, erm(B), and msr(C) has been identified 91 .Together with E. faecium also, Lactococcus garviae has been detected as a multidrug-resistant strain carrying an antimicrobial resistance gene at the plasmid 92 .
For immunocompromised patients, often, hospital-acquired infections are challenging to cure because opportunistic pathogens have accumulated multidrug resistance mechanisms.In our study, the gut resistome of patients with FN has been found to be mainly carried by multidrug-resistant pathobionts such as E. faecium, Enterococcus faecalis, S. aureus, and Klebsiella spp.These pathobionts carried a significant subset of resistance genes and have been suggested to be involved in the patient's treatment and metabolism of trimethoprim / sulfamethoxazole, amikacin, piperacilline/tazobactam, and teicoplanin potentially leading to alterations in the efficiency of the treatment.Resistance to piperacillin / tazobactam was mainly conferred by the genera K. pneumoniae and E. coli, consistent with other studies findings 93,94 .Bacteraemia caused by gram-positive bacteria was commonly associated with FN, with varying prevalence of gram-negative pathobionts.Prophylactic antibiotics, such as fluoroquinolones and ciprofloxacin, were linked to increased resistance among cultivation-analysed bacteria.
Except for antibiotic treatment, it is crucial also to consider the impact of MTX treatment on the gut barrier and gut microbiome status of patients, as it can contribute to an increased risk of bloodstream infections through bacterial translocation.MTX, a chemotherapy drug also used for the treatment of different types of oncology diagnoses, including acute lymphoblastic leukaemia and non-Hodkgin's lymphoma, has been found to exhibit toxic effects on multiple organs, including the gastrointestinal tract, bone marrow, heart, kidneys, and liver 95,96 .Additionally, the intestinal toxicity of MTX can lead to mucositis, resulting in symptoms such as nausea, bloating, abdominal pain, and diarrhea 97,98 .The drug triggers inflammatory events that can damage intestinal epithelium and submucosal tissue cells 99 .However, certain bacterial species, such as B. fragilis and R. intestinalis, can enhance the effects of the drug or reduce its toxicity.In particular, B. fragilis has been found to have a protective function against MTX-induced inflammatory reactions 100 We have found no clear indices regarding the association with FN since two patients with detected Bacteroides spp.and developed FN, while one with distinct strains o B. fragilis did not.However, more profound insight into strain level could indicate crucial differences in Bacteroides spp.population composition.During the surgical intervention, damaged intestinal barrier integrity, bacterial contamination, and the weak immune system of patients, B. fragilis can be found in 30-60% of cases of purulent-septic infections.There are two types of B. fragilis, either harboring silent resistance gene cfiA and with high probability resistance to carbapenems or not.Ank et al. (2015) isolated multidrug-resistant B.fragilis 101 from blood and B.fragilis strains detected in the gut microbiome of patients with FN.Strains O17, and O18 harbor th antibiotic resistance genes cfiA, tetQ, and nim 102 indicating possible activation of carbapenem resistance through insertion element and leading to possible treatment failure 103 .
Additionally, R. intestinalis is an obligately anaerobic bacterium capable of using acetate to produce butyrate due to the butyryl-CoA: acetate CoA transferase 104 .Butyrate has anti-inflammatory and metabolic modulatory effects 105,106 and serves as a source of energy for enterocytes, enhancing the integrity of the intestinal barrier.Cooccurrence of these bacterial genera can improve the treatment outcome.Vitamin supplementation, specifically vitamin C and B2, has also been shown to alleviate the clinical symptoms of MTX-induced mucositis and increase the growth of certain bacterial strains associated with intestinal inflammation.By examining the effects of these vitamins on the in vitro growth of several bacterial strains intrinsically associated with intestinal inflammation, it was found that vitamin supplementation under oxidative conditions increased the growth of Blautia coccoides and R. intestinalis 107 .Overall, it is essential to consider the potential impact of the microbiome on cancer treatment outcomes and to explore ways to mitigate the harmful effects of chemotherapy drugs.
Nevertheless, there are some limitations of this study that rely in lower number of involved patients, even though representing all the patients with febrile neutropenia hospitalized during the period of three years.Furthermore, not all the patients were able to provide sample before and after HSCT.Although, the parenteral nutrition and diet supplements were standardized, the use of antimicrobials was altered and adjusted to current health status of the patient.Additionally, the gender balance within analysed groups was slightly shifted towards male patients (11 vs. 7).While eight males developed febrile neutropenia, only four females were diagnosed with fever.Nevertheless, our study is the first one that investigates the gut microbiome composition, diversity and resistome and identifies multidrug-resistant bacteria of paediatric oncology patients undergoing haematopoietic stem cell transplantation before and after HSCT treatment.
Our research indicates that one must consider the diverse combinations of resistant or multidrug-resistant bacteria in a patient's gut microbiota.However, it is crucial to recognize the significance of clustering patients based on their resistome, which should be noticed.It is important to note that the gut resistome has the potential to connect the gut microbiome and the FN, therefore, it is a critical factor that must be considered.Continuous monitoring and adaptation of antimicrobial therapy based on local resistance patterns would be critical for effective management.Judicious use of antibiotics would also be essential in mitigating the emergence and spread of antibiotic-resistant bacteria in febrile neutropenic patients [93][94][95][96] .

Conclusion
Hematopoietic stem cell transplantation is lifesaving but still associated with adverse outcomes.We revealed a patient-specific resistome pattern rather than a typical bacterial profile associated with conditioning regimen treatment in patients undergoing HSCT.However, significant decrease of alpha-diversity was identified in the gut microbiome of patients with FN one week after HSCT.Furthermore, there appears to be a correlation between the treatment consequences and a higher prevalence of specific resistance genes tetQ and cfrA in the gut microbiome one week after HSCT.This finding is noteworthy, as it may affect future treatments and patient care.It is also important to note that a group of nine resistance genes, including msr(C), dfrG, erm(T), VanHAX, erm(B), aac(6)-aph (2), aph(3)-III, ant(6)-Ia, and aac(6)-Ii, demonstrate a significant correlation and connection to FN.The gut resistome of patients with FN has been found to be mainly carried by multidrug-resistant pathobionts such as E. faecium, E. faecalis, S. aureus, and Klebsiella spp. that have been suggested to be involved in the patient's treatment through metabolism of trimethoprim/sulfamethoxazole, amikacin, piperacilline/tazobactam, and teicoplanin.These findings need to be supported with future studies or information on the state of the intestinal barrier linking the gut with inflammation.It is worth mentioning that some gut bacteria can also be involved in cancer drug metabolism.Our results suggest screening for resistance genes in immunocompromised pediatric patients through quantitative analysis using qPCR or targeted screening for ESKAPE bacteria via PCR methods to improve management of multidrug-resistant infections.A personalized approach is necessary for assessing gut microbiome and resistome of patients undergoing HSCT in order to enhance development of effective treatment strategies and mitigate adverse outcomes.

Figure 1 .
Figure 1.Alpha diversity of gut microbiome of patients without febrile neutropenia (FN-) or developing febrile neutropenia (FN+) before (d-7) and after (d + 7) HSCT represented by Shannon index (A, B) and Simpson index (C, D).The p values were computed using Mann-Whitney test for parametric data and Wilcoxonrank test for nonparametric data with significance p ≤ 0.05 applied.Significant decrease in alpha diversity was detected in gut microbiome of patients developing febrile neutropenia after HSCT.

Figure 2 .
Figure 2. Gut microbiome composition at phylum level (I.) calculated as relative abundance visualized as bargraphs.All involved participants including healthy individuals and patients at all time-points of sampling were included.II.-relative abundance of selected bacterial families of gut microbiota of patients before (d-7), one week after (d + 7) and four weeks after (d + 28) compared to healthy individuals (CTRL).Beneficial commensal bacteria (A-Akkermansiaceae, B-Clostridiaceae, C-Lachnospiraceae) and opportunistic pathogens (D-Enterococcaceae, E-Staphylococcaceae, F-Streptococcaceae) visualized as bar charts.Significant decrease (p ≤ 0.05) in relative abundance of Bacteroidota, Akkermansiaceae, Clostridiacea, Lachnospiracea and increase of opportunistic pathogens was observed in gut microbiome of paediatric oncology patients.

Figure 3 .
Figure 3.The gut bacteriome of the patients undergoing HSCT and healthy individuals visualized using a Krona pie chart.Each chart represents gut composition calculated as average for each bacterial genus detected within the group.CTRL-control group; d-7-before HSCT; d + 7-one week after HSCT; d + 28-one month after HSCT.

Figure 4 .
Figure 4. Beta diversity of analysed samples represented by the full resistome profile of gut microbiota of paediatric oncology patients before (d-7), one week after (d + 7) and four weeks after (d + 28) HSCT.Resistome was represented by a set of antimicrobial resistance genes identified within the analysed sample.SVD with imputation was used to calculate the principal components.The X and Y axis show principal component 1 and principal component 2 that explain 19.1% and 12.4% of the total variance.Prediction ellipses represent 0.95% probability for new observation to fit the group (n = 47).

Figure 5 .
Figure 5. Graphical visualization of the absolute number of antimicrobial resistance genes clustered according to their predicted phenotype carried by bacterial genera detected in gut microbiome of healthy individuals and patients. https://doi.org/10.1038/s41598-024-56242-8

Figure 8 .
Figure 8.The network analysis visualizing the association between opportunistic pathogens Enterococcaceae, Streptococcaceae, Staphylococcaceae, antimicrobial genes of resistance, inflammation biomarkers CRP and procalcitonin and febrile neutropenia.

Table 1 .
Identified resistance to antibiotics used by patients before and one week after HSCT.Resistance genes carried by bacteria to listed antibiotics used by patients.