Fecal Microbiota Transplantation Commonly Failed in Children With Co-Morbidities

Abstract

Objectives: 

Fecal microbiota transplantation (FMT) is arguably the most effective treatment for recurrent Clostridioides difficile infection (rCDI). Clinical reports on pediatric FMT have not systematically evaluated microbiome restoration in patients with co-morbidities. Here, we determined whether FMT recipient age and underlying co-morbidity influenced clinical outcomes and microbiome restoration when treated from shared fecal donor sources.

Methods: 

Eighteen rCDI patients participating in a single-center, open-label prospective cohort study received fecal preparation from a self-designated (single case) or two universal donors. Twelve age-matched healthy children and four pediatric ulcerative colitis (UC) cases from an independent serial FMT trial, but with a shared fecal donor were examined as controls for microbiome restoration using 16S rRNA gene sequencing of longitudinal fecal specimens.

Results: 

FMT was significantly more effective in rCDI recipients without underlying chronic co-morbidities where fecal microbiome composition in post-transplant responders was restored to levels of healthy children. Microbiome reconstitution was not associated with symptomatic resolution in some rCDI patients who had co-morbidities. Significant elevation in Bacteroidaceae, Bifidobacteriaceae, Lachnospiraceae, Ruminococcaceae, and Erysipelotrichaceae was consistently observed in pediatric rCDI responders, while Enterobacteriaceae decreased, correlating with augmented complex carbohydrate degradation capacity.

Conclusion: 

Recipient background disease was a significant risk factor influencing FMT outcomes. Special attention should be taken when considering FMT for pediatric rCDI patients with underlying co-morbidities.

What Is Known/What Is New

What Is Known

• Fecal microbiota transplantation (FMT) is effective in treating Clostridioides difficile infection (CDI) recurrence
• FMT-derived microbiome reconstitution is common in recurrent Clostridioides difficile infection (rCDI) responders compared with non-responders.

What Is New

• Careful clinical precautions should be taken when considering rCDI patients with co-morbidities for FMT.
• Post-FMT microbiome composition in medically complex (with co-morbidities) rCDI patients may not correlate with clinical outcomes.
• Post-FMT microbiome composition in non-complicated pediatric rCDI responders more closely represented age matched healthy controls than their respective adult donors.
• Expansion of Bifidobacteriaceae after FMT was unique to pediatric rCDI responders.

Clostridioides difficile infection (CDI) is a frequently reported nosocomial pathogen ⁽¹⁾. Severe diarrhea associated with CDI is common leading to significant hospitalization, morbidity and mortality ⁽²⁾. Recent epidemiologic observations indicate normalizing or declining trends in the United States ⁽³⁾. CDI causes approximately half a million infections and over $1.5 billion in excess medical costs annually in the United States alone ^(2,4). The incidence of pediatric CDI is lower, but shows similar trends as in adults ^(5,6).

CDI disease recurrence is 20–30%, and the risk increases further with repeated infections despite new antibiotic treatments ⁽⁷⁾. Although still classified as an investigational procedure, fecal microbiota transplantation (FMT) is now regarded as the most effective clinical therapy for treating recurrent CDI (rCDI) ^(8,9), supported by a systematic review of 844 patients ⁽¹⁰⁾. Similar cure rates are reported in children ^(11,12), but associated microbiome analyses are less common ^(13–16). Special consideration for FMT must be taken with pediatric rCDI cases given the high C difficile asymptomatic carriage rate and the more frequent underlying clinical co-morbidities when compared to adults ⁽¹⁷⁾. These complications were recognized by the Infectious Diseases Society of America (IDSA) and Society for Healthcare Epidemiology of America (SHEA) ⁽¹⁸⁾, emphasizing the need to critically evaluate FMT outcomes in pediatric rCDI cases.

To assess clinical efficacy and evaluate microbiome restoration, we conducted fecal 16S rRNA amplicon sequencing in a single-center cohort of eighteen pediatric rCDI patients who underwent FMT. The majority (94%) of cases received FMT from two universal donors, which created an opportunity to examine recipient-based microbiome restoration to FMT in children over a wide age range and co-morbidities. We compared clinical and microbiome differences in pediatric recipients who shared the same donor, thereby limiting biologic variation in clinical and microbiome outcomes.

METHODS

Study Design and Outcome Measures

Eighteen rCDI patients (8 females and 10 males; average 9.3 years; range 1.5–16) received FMT from February 2013 to December 2015 under IRB-approved informed consent (#H-31066) at Baylor College of Medicine (Figure 1, https://links.lww.com/MPG/C560 Supplemental Digital Content). The investigational nature of FMT was highlighted during consenting according to U.S. Food and Drug Administration regulations. CDI diagnosis was based on enzyme immunoassay (EIA), and/or culture, toxin polymerase chain reaction (PCR) positivity, along with clinical complaints of three or more diarrheal stools per day. Any of the specific laboratory tests positive for toxigenic C difficile in the beginning of the recurrent infection were considered as diagnostic. Since most patients were referred to our tertiary center, and repeat testing during the course of rCDI is not recommended ⁽¹⁹⁾, we did not incorporate a laboratory-based diagnostic algorithm. We recently highlighted ⁽²⁰⁾ that no laboratory-based diagnostic is available to differentiate between infection and colonization with C difficile in children ⁽²¹⁾.

Five patients had inflammatory bowel disease (IBD), one received a heart transplant, and three had significant neurologic impairment as underlying conditions (Table 1). All patients reported recurrent (within 2 months) or ongoing diarrheal symptoms in spite of at least two trials of CDI-directed antibiotics. All patients received at least one course of metronidazole (10–14 days) and at least one 14-day course of vancomycin orally (except for P02 who following a course of metronidazole and attempted home-based FMT from a parent donor, using a turkey baster with verbal instructions from the family doctor; she shortly had recurrent symptoms not responding to 4 days of vancomycin before FMT). Sixteen patients received vancomycin until 1 day before FMT, except for two patients (P01 and P18) who completed vancomycin therapy 2 and 3 weeks before FMT, respectively and remained symptom free before FMT. These two rCDI patients had multiple recurrences historically and the clinical decision was made to provide FMT without allowing for a repeated recurrence. Absence of diarrhea was the main outcome measure within 8 weeks post FMT to stratify responders (R) versus non-responders (NR).

TABLE 1 - Summary of FMT for rCDI patients with or without underlying conditions

Patient ID	Age (y)	Gender	Category	Comorbidities	Diarrhea resolved	Group
P01	14	Male	Chronic GI condition	UC, GERD s/p fundo, hypogammaglobulinemia	No	Non-responder
P02	15	Female	No underlying GI disease	–	Yes	Responder
P03	15	Female	Neurologic impairment	Prematurity 26 wk, cerebral palsy, global developmental delay	No (paradoxical diarrhea)	Non-responder
P04	4	Male	No underlying GI disease	–	Yes	Responder
P05	2	Male	No underlying GI disease	GERD, transient hypogammaglobulinemia of infancy, chronic cough with recurrent resp tract infections, autism, previous exposure to sewage/mold	Yes (returned 7 mo later, cleared with antibiotics)	Responder
P06	2	Female	Heart transplant	Hypoplastic left heart syndrome s/p cardiac transplant, VUR	Yes (returned 2 mo later after antibiotic therapy)Yes (after second FMT as well)	Responder
P07	5	Male	Chronic GI condition	CD	No (symptoms resolved only after several months once switched to vedolizumab)	Non-responder
P08	2	Female	Neurologic impairment	Spastic quadriplegia, global developmental delay, infantile spasms, RAD, chronic lung disease, G-tube dependence, recurrent UTI	Yes after second FMT; recurrence >1 y later; no symptom resolution after third FMT when fidaxomicin added and constipation addressed	Responder
P09	15	Male	No underlying GI disease	–	Yes	Responder
P10	16	Female	Chronic GI condition	UC	Yes	Responder
P11	16	Male	No underlying GI disease	–	Yes	Responder
P12	5	Female	No underlying GI disease	–	Yes	Responder
P13	2	Male	Neurologic impairment	Global developmental delay, GERD s/p GT placement/Nissen, aspiration/choking episodes, upper GI dysmotility	No (intermittent diarrhea/pain even after second FMT; working diagnosis of mito-disease)	Non-responder
P14	7	Female	Chronic GI condition	UC	No	Non-responder
P15	13	Male	Chronic GI condition	UC	No (symptoms returned after 1 mo; transiently improved while on intermittent vancomycin for 5 mo + adalimumab then later vedolizumab)	Non-responder
P16	17	Female	No underlying GI disease	Wolff-Parkinson-White syndrome s/p cardiac ablation procedures	Yes (until 1 y later, had recurrence of C difficile diarrhea that responded to vancomycin)	Responder
P17	3	Male	No underlying GI disease	Recurrent otitis media s/p multiple courses of antibiotics	Yes (until 6 mo later, developed C difficile colitis despite receiving vanc ppx after penicillin injection; improved after vancomycin treatment)	Responder
P18	2	Male	No underlying GI disease	Allergic rhinitis, eczema, multiple sinus infections	Yes	Responder

CD = Crohn disease; rCDI = recurrent Clostridioides difficile infection; FMT = fecal microbiota transplantation; fundo, fundoplication; GERD, gastro-esophageal reflux disease; GI = gastrointestinal; UC, ulcerative colitis; UTI = urinary tract infection; VUR = vesico-ureteral reflux; RAD = reactive airway disease; –, data/records/outcome are not available.

Fecal samples from twelve age-matched healthy pediatric subjects were examined as controls (IRB protocols #H-35071 and #H-36399). Additionally, we examined pre- and post-treatment fecal samples from three patients with ulcerative colitis (UC) (PCR negative for C difficile toxin genes), who received four treatment series (one patient enrolled twice 1 year apart) of >20 FMTs (one colonoscopy and others via enema) over 8–12 weeks. Details of the non-CDI UC, serial FMT trial are already published ⁽²²⁾. These cases served as disease controls for FMT without CDI and recent antibiotic exposure.

Donor and Preparation

Patients received filtered, frozen-thawed fecal preparations from two, screened, standardized donors D2 (38–40 years male during donations) for six patients and D3 (38–40 years male during donations) for 10 patients. Patient P03 received FMT from both D2 and D3, and patient P06 received a self-designated donor D5 (recipient's father). All patients received the first FMT via colonoscopy, followed by enema or nasogastric FMT (in three cases) if clinically indicated (Table 1). The donor screening and fecal preparation procedures were approved by the U.S. Food and Drug Administration (IND#15743) and are published previously ⁽²²⁾.

Stool Collection and Microbiome Analysis

Fecal samples were collected from patients the day before FMT and 8–9 weeks following treatment at a follow-up visit (Figure 2, https://links.lww.com/MPG/C561 Supplemental Digital Content), or were sent on wet ice from home. The reason for this timing of stool collection was to match the clinic or phone based follow-up, which determined the treatment response by the commonly recommended timeline (ie, 8 weeks) defining recurrence versus re-infection. Follow-up fecal samples from patients with UC were obtained 2 weeks after serial FMT completion (patients in this study received 22–30 FMTs) in a weaning fashion from daily to weekly enemas; we wished to examine post FMT microbiomes 2 weeks after the last treatment before a potential clinical flare would have occurred ⁽²²⁾. All samples were processed and stored and 16S V4 sequenced as described by Kellermayer et al ⁽²²⁾.

Raw Illumina paired-end reads were merged by VSEARCH (version 2.9) ⁽²³⁾. The default DADA2 package (version 1.8) ⁽²⁴⁾ was used for sequence denoising with specific modification on sequence quality filtering using a maximum expected error of 2. Taxonomy of amplicon sequence variants (ASVs) were annotated by IdTaxa function ⁽²⁵⁾ with its pre-built SILVA reference database (release 132) using the stringent confidence threshold of 70. PICRUSt2 ⁽²⁶⁾ was used to infer metagenomic functions based on ASV-assigned sequences. Downstream analyses including alpha-diversity (Shannon index) and beta-diversity were performed using the Non-metric Multidimensional Scaling (NMDS)-ordination method for Bray-Curtis dissimilarity profiling. ASV-ASV correlation networks were generated using Cytoscape (version 3.7.2) ⁽²⁷⁾. Raw amplicon data was deposited to NCBI under accession number PRJNA735699.

Statistical Analysis

Wilcoxon rank-sum test was applied to assess group differences using the R-based pairwise.Wilcox.test. Spearman-correlations were assessed using the R-based cor.test. Fischer exact testing was used for group comparisons. Multiple-comparisons were adjusted using the Benjamini-Hochberg procedure with statistical significance considered at P < 0.05.

RESULTS

Patient Outcomes

Age did not affect treatment outcomes (patients <8 years [n = 10] vs >12 years [n = 8]; P = 1). All nine patients without underlying gastrointestinal (GI) disease (Table 1) had symptomatic resolution for more than 2 months following a single FMT. The success of FMT significantly differed between recipients without underlying chronic conditions and those with IBD, and/or immunosuppression, or significant neurodevelopmental delay (Fischer exact test, P = 0.009). This separation was biased by the IBD patients (1/5 [20%] who responded (Fischer exact test, P = 0.005 compared to patients without underlying GI disease). Patients with other complicating conditions also showed diminished efficacy, but this did not reach significance (2/4 [50%] responded, Fischer exact test, P = 0.077 compared to patients without underlying diseases).

Of the nine patients with complicating clinical conditions, only three (33.3%: P06, P08, and P10) had obvious clinical benefit from C difficile-directed antibiotics before FMT. These latter patients responded well to a single FMT, including the youngest solid organ transplant patient receiving treatment ⁽²⁸⁾; however, two of the “chronic disease” responders were treated with antibiotics for upper respiratory infections within the 2-month follow-up period, and experienced recurrence of CDI. Both received repeat vancomycin followed by intra-gastric FMT from the same donor as during their first FMT with full resolution of their symptoms for over 2 months.

One rCDI patient with UC (P01) required three FMT treatments to clear toxigenic C difficile but remained symptomatic and still required colectomy during the subsequent clinical course despite being pathogen free by PCR-based testing. Our second patient with UC (P10; maintenance mesalamine therapy only) had rapid resolution of symptoms, and remained symptom free for 9 weeks, then experienced a C difficile-positive flare of UC and was started on steroids and 6-mercaptopurine along with oral vancomycin for 2 weeks. The patient responded well to the intervention and subsequently tested negative for C difficile 2 weeks after stopping vancomycin. Another patient with UC (P14) cleared the infection by PCR but remained symptomatic and required augmentation of immunosuppressive treatments. Our fourth patient with UC (P15) did not clear C difficile after FMT, but became symptom free only after intense IBD treatment optimization.

One patient with global developmental delay (GDD, P03) was diagnosed with poorly managed constipation during her second enema FMT and continued with C difficile carriage based on toxin-PCR testing, but clinically improved with constipation directed treatments. Another GDD patient (P13) cleared C difficile after FMT but remained symptomatic. There were seven patients among the responders who had more than 5-month clinical follow-up. None of these patients developed CDI or obesity, or any new medical condition based on the available records to 3 years of maximal follow-up. They all remained below 70th percentile body mass index (BMI) during this period (Figure 3, https://links.lww.com/MPG/C562 Supplemental Digital Content).

Microbiome Restoration Differed Between Pediatric Fecal Microbiota Transplantation Responders and Non-Responders

Irrespective of adult donor fecal source, microbiota transplant significantly increased the Shannon index in pediatric CDI responders and non-CDI UC patients post-FMT, whereas alpha-diversity in CDI non-responders remained unaltered (Fig. 1A); however, the Shannon index of baseline pre-FMT samples demonstrated large differences between CDI and non-CDI UC patients suggesting that therapy with C difficile-directed antibiotics had significant impact on microbiome diversity in CDI patients. Further beta-diversity analysis demonstrated restoration of microbiome composition and function in pediatric rCDI responders, with the exception of patients P06 and P08 (Fig. 1B). Diarrhea-resolved P06 and P08 were both younger patients with underlying conditions (Table 1) and remained dysbiotic even after two FMT procedures. Our analysis also demonstrated that CDI non-responders P03 (pre-FMT) and P13 (post-FMT) had low dissimilarity distance scores compared with healthy controls even though both patients remained symptomatic (Fig. 1B). Misdiagnosis/CDI independent diarrhea associated with sensitive PCR-based C difficile toxin testing could explain these findings. Notably, metagenomic sequencing would have supported one case of clinical CDI misdiagnosis in one child (P03) who was later recognized to suffer from chronic constipation with overflow (paradoxical) diarrhea (Fig. 2A). Exclusion of patient P03 from the microbiome analyses did not alter the study findings, demonstrating significant changes in microbiome community structure in FMT responders only (Figure 4, https://links.lww.com/MPG/C563 Supplemental Digital Content).

FIGURE 1: FMT promotes the reconstitution of microbiome composition and function, and alpha-diversity in rCDI responders. (A) Shannon index of pediatric CDI and UC patients receiving identical donor material. (B) Beta-diversity analysis of microbiome composition (ASVs) and predicted function (PICRUSt2) using Bray-Curtis dissimilarity metric. Significance denotations: n.s., not significant; ^∗ P < 0.05; ^∗∗∗ P < 0.001. ASV = amplicon sequence variant; rCDI = recurrent Clostridioides difficile infection; UC = ulcerative colitis.

FIGURE 2: Successful microbiome reconstitution is highly conserved but age-dependent in pediatric rCDI recipients. (A) Hierarchical clustering of family abundance highlights the reconstitution of conserved microbiome (box with dashed line) in pediatric rCDI responders. Notably, CDI-P3 (pre=FMT) clustered with healthy controls suggesting potential C difficile colonization rather than active infection. (B) Responder microbiomes of pediatric rCDI patients were similar to pediatric healthy controls. Bifidobacteriaceae expansion is a pediatric-specific microbiome feature in post-FMT samples when compared to adult donor microbiomes. Significance denotations: ^∗ P < 0.05; ^∗∗ P < 0.01; ^∗∗∗ P < 0.001. rCDI = recurrent Clostridioides difficile infection; FMT = fecal microbiota transplantation; UC = ulcerative colitis.

Reconstitution of Microbiome Composition in Pediatric Recurrent Clostridioides difficile Infection Responders

Hierarchical clustering analysis of family abundance profiles demonstrated consistent reconstitution of Bacteroidaceae, Lachnospiraceae, Ruminococcaceae, Bifidobacteriaceae, and Erysipelotrichaceae, which were significantly elevated in rCDI responders after FMT, similar to age-matched healthy controls (Fig. 2). We also observed that Enterobacteriaceae abundance decreased significantly post-FMT in rCDI responders (Fig. 2B). Interestingly, post-FMT Bifidobacteriaceae, Lachnospiraceae, and Ruminococacceae in rCDI responders showed similar abundance with pediatric healthy controls, but were significantly higher than in adult donors (Fig. 2B). Whereas Bacteroidaceae increased among FMT responders, this family remained significantly lower in abundance compared with the adult donors and was similar to healthy age-matched controls. These findings represent a previously unappreciated feature in pediatric rCDI recipients who receive FMT from an adult donor.

Enterobacteriaceae De-Colonization is Linked With Augmented Complex Carbohydrate Utilization

Enterobacteriaceae dominance is observed in treatment-native CDI patients, and remains high after C difficile clearance before FMT ⁽²⁹⁾. Our ASV-ASV correlation network analysis identified four dominant Enterobacteriaceae ASVs that demonstrated significant negative correlations with Bacteroidaceae, Lachnospiraceae, Ruminococcaceae associated with a healthy microbiome (Fig. 3A). These dominant AVS's accounted for over 95% abundance of Enterobacteriaceae and were largely absent in post FMT specimens (Fig. 3B). Correlation analysis between ASV abundance and PICRUSt2 pathway abundance identified top significant hits for the degradation of complex carbohydrates including glycogen, starch and mannans that negatively correlated with decolonization of Enterobacteriaceae following FMT (Fig. 3B). These pathways are highly abundant in healthy microbiota and are linked with short-chain fatty acid (SCFA) production, which we have previously reported as being associated with restored microbiota function after successful FMT ⁽³⁰⁾.

FIGURE 3: Decolonization of Enterobacteriaceae is associated with carbohydrate utilization of healthy dominant commensals engrafted in rCDI responders. (A) ASV-ASV correlation network analysis highlights the strong negative correlation of four Enterobacteriaceae with healthy microbiota. (B) Four ASVs with diverse genera represented a major contributor of the Enterobacteriaceae abundance in pre-FMT specimens. (C) Enterobacteriaceae is negatively correlated (P < 0.001) with complex carbohydrate degradation of healthy microbiomes. ASV = amplicon sequence variant; rCDI = recurrent Clostridioides difficile infection; FMT = fecal microbiota transplantation.

DISCUSSION

This study represents the largest metagenomic investigation of microbiome restoration following FMT in pediatric rCDI with age-matched healthy and non-CDI UC patient controls. It also provides the longest clinical outcome information in pediatric rCDI patients following FMT. Utilization of an institutional stool bank providing shared/universal donors between recipient patients created a unique opportunity to evaluate microbiome restoration in both rCDI and UC after FMT. This approach decreased donor microbiome variation when characterizing donor-recipient interactions, compared with other FMT studies that utilized self-designated or commercial fecal donors. All FMT procedures in this study were performed within a 2-year period, during which the universal donors did not change diet and had a stable BMI. Under such circumstances, fecal microbiomes from adult individuals are remarkably stable over time ⁽³¹⁾, in agreement with our findings in serial donor specimens that clustered closely over this investigational FMT study. Similar to others ⁽¹¹⁾, we found FMT to be safe and highly effective in preventing CDI recurrence in pediatric patients without complicating clinical conditions. Moreover, similar to adult patients with CDI receiving FMT ^(32–37), post-FMT microbiome diversity of pediatric CDI responders increased to a level comparable with healthy pediatric controls. Unlike antibiotic-induced gut dysbiosis in rCDI patients, pre-FMT microbiome diversity remained high in patients with non-CDI UC in clinical remission. As such, microbiome diversity may not be a good marker for evaluating FMT outcomes in patients with non-CDI UC. We also found that microbiome shifts in antibiotic treated rCDI patients were larger after a single FMT than in patients with UC without antibiotic exposure even though they received 22–30 FMTs. This finding shows that antibiotic pretreatment and/or recurrent CDI creates a more significant dysbiosis than UC during immunomodulatory-induced remission.

Our study also demonstrated that microbiome reconstitution is generally in agreement with FMT outcomes in rCDI patients without predisposing clinical conditions. This was not evident in patients with co-morbid conditions indicating that metagenomics alone is not suitable to predict disease outcomes of FMT in complicated pediatric rCDI patients. We found that diagnostic ⁽²¹⁾ and clinical ⁽³⁸⁾ screening for rCDI before considering FMT is challenging in medically complex patients and the real-life cohort in this work represents the evolution of our program's experience. Our findings indicate that microbiome analysis incorporating healthy age matched controls may help distinguish carrier/colonized states from true infections in select cases in the future (P03 in this case). Additionally, the observation that those patients who remained symptomatic during C difficile directed antibiotic therapy were all non-responders is consistent with a prior report on pediatric FMT ⁽¹²⁾ and suggests that this criterion should be considered in the treatment paradigm of rCDI.

We also found increased abundance of Bifidobacteriaceae in pediatric rCDI patients post-FMT, which has not been observed previously even in adult rCDI ^(33–37), but has been reported after FMT in healthy volunteers ⁽³⁹⁾ and patients with hepatic encephalopathy ⁽⁴⁰⁾. Furthermore, we demonstrated that taxonomic similarity in multiple bacterial families was closer between post-FMT responder and healthy age-matched control than donor microbiomes. This finding implicates the “enslapment” nature of FMT in children ⁽⁴¹⁾. Engraftment versus “enslapment” could not be directly examined in this study because of the lack of deep sequencing and requires further investigation, especially in regard to the expansion of SCFA-producing bacteria that can suppress C difficile and Enterobacteriaceae growth via intracellular acidification ⁽⁴²⁾. Butyrate ^(43,44) is protective against CDI in preclinical models and carbohydrate catabolism is a major SCFA producing pathway in healthy microbiomes ⁽⁴⁵⁾.

“Non-responder” Patient P03 deserves further discussion since this case highlighted the clinical importance of addressing the possibility of paradoxical/overflow diarrhea in pediatric patients with complex conditions who are considered for FMT due to presumed rCDI. Current clinical diagnostic tests cannot distinguish between CDI and C difficile colonization in children ⁽²¹⁾. Consequently, it is difficult to diagnose rCDI and we recently reported that up to 25% of pediatric FMT candidates may have alternative diagnoses, of which overflow diarrhea was the most common ⁽³⁸⁾. Our finding herein that the pre-FMT microbiome of Patient P03 clustered with age-matched controls (Figs. 2B and 3A) indicates that 16S-sequencing may aid in differentiating between patients with paradoxical (constipation-associated) diarrhea and C difficile colonization versus rCDI. This conclusion will need confirmation in larger prospective studies.

Limitations include the small single-center, real-life study of pediatric rCDI patients receiving FMT. This indication, however, is thankfully rare in children making our cohort “sizable,” considering its single-center nature. Not all subjects submitted post-intervention stools around the same time-frame. Some of the subjects did not respond to anti-CDI treatment and one subject was found to have constipation raising concerns for colonization rather than true infection, making interpretation of data difficult. In spite of these limitations, our findings provide valuable information to practicing clinicians and microbiome investigators in respect to pediatric FMT outcomes, underscoring the need for cautious consideration when approaching medically-complex pediatric cases suffering from presumed rCDI.

The ultimate goal of microbiome-based therapeutic interventions is to identify disease-specific combinations with the most limited number of microbial species involved ⁽⁴⁶⁾. A recurrent theme in metagenomic observations during FMT for rCDI is the recovery of Bacteroidaceae abundances in adults and children ^(8,15). Our findings support the notion that Bacteroidaceae may represent key protective members of the healthy microbiome and could form the basis for precision-therapy against rCDI ^(47,48). Finally, our findings newly implicate Bifidobacteriaceae as being associated with treatment success in pediatric rCDI, meriting further investigation as a model of age-dependent precision medicine. Bifidobacteria have a butyrogenic effect in the human colon ⁽⁴⁹⁾ due to cross-feeding between other SCFA-producing colonic bacteria ⁽⁵⁰⁾, supporting the notion that the butyrate-pathway may be an important target for microbial therapeutics in children.