ResearchResearch summary
Fecal Candida signals before sepsis in very preterm infants
A small matched study found fecal Candida and fungal signals before culture-proven sepsis in very preterm infants; it did not show where infection began.
In this pathway
- Current topic
- Fecal Candida signals before sepsis in very preterm infantsCurrent

Bottom line
In this small matched study, fecal fungal and Candida signals differed between 8 very preterm infants who developed culture-proven Candida late-onset sepsis and 16 matched controls. Samples were collected during the five days before diagnosis and on the day of diagnostic workup, but the findings do not show that infection traveled from the gut to the bloodstream.
Why this source matters
The study looked at stool profiles collected before a separate, culture-defined invasive infection. That gives a narrow view of how fecal signals and later Candida late-onset sepsis were associated in a very preterm NICU cohort. It does not show where the bloodstream infection began or establish a route of transmission.
Study and population
This observational matched case-control analysis was nested in a prospective cohort with daily stool and clinical data. Ten neonatal intensive care units in the Netherlands and Belgium enrolled infants from October 2014 through February 2023. Eligibility was 24–30 weeks' gestation through March 2021 and 24–28 weeks thereafter. The analysis included 8 infants with Candida late-onset sepsis and 16 matched controls. The cases had fungal infection confirmed by blood and/or cerebrospinal-fluid culture.
What was tested
Researchers compared fecal fungal ITS1 and bacterial 16S profiles, PCR amplicon yields, and the ratio of ITS to 16S yields. Samples covered the day of diagnostic workup (t=0) and the preceding five days (t=−1 through t=−5). This was an observational comparison, not a test of treatment or prevention. Stool profiling was separate from the blood/CSF culture used to define sepsis.
Results
Before read-quality exclusions, the 8 cases contributed 22 stool samples and the 16 pooled controls contributed 56. The reviewed full text does not give the final per-assay sample counts, so these sample totals are not the denominator for every reported analysis.
Across the t=−5 through t=0 interval, mean fungal amplicon yield was 16.23 ± 9.43 ng/µL in cases versus 5.45 ± 7.17 ng/µL in controls (P<.001). Mean bacterial amplicon yield was 8.68 ± 5.53 ng/µL versus 14.39 ± 3.32 ng/µL (P=.02). These PCR yields are measurement proxies, not direct counts of viable organisms.
The relative fecal Candida-abundance measure was 0.998 ± 0.003 in cases versus 0.673 ± 0.368 in controls across the interval (P=.036). The ITS/16S amplicon-yield ratio was about 15-fold higher in cases (P<.001), but the difference was not significant when analysis was restricted to the day before diagnosis. These group comparisons do not establish that fecal Candida caused, predicted, or seeded the bloodstream infection.
Applicability
The result applies to the eight very preterm infants with culture-proven Candida late-onset sepsis and their matched NICU controls in this study. It does not establish risk or diagnosis for an individual infant, and it does not apply to term infants, older children, or adults with nonspecific symptoms.
Limitations
Only 8 cases and 16 controls were analyzed, and case stool was available for a selected subset of the cohort. There were 22 case and 56 control samples before read-quality exclusions, but final assay denominators are not available in the reviewed full text. The fecal measures were relative sequencing profiles and PCR-yield proxies, not direct viable-organism counts. The study did not compare stool and blood/CSF strains by whole-genome analysis, so it cannot establish gut-to-blood transmission; skin colonization and a skin-to-blood route remain plausible. Cases and controls also differed in antibiotic exposure and feeding/delivery characteristics, leaving room for confounding and reverse causation.
Contribution to the evidence
Contextual. The study adds a small, selected neonatal observation that fecal fungal and Candida signals were associated with later culture-proven Candida late-onset sepsis. It supports neither a causal gut-source conclusion nor an adult broad-Candida claim. It does not alter the linked evidence map's claims or their practical guidance.
Practical change
No practical change. The study does not establish the source or route of infection and does not test neonatal care, treatment, or prevention. It cannot be used to make recommendations for infants or to guide care for adults with broad Candida-related symptom claims.
Funding and conflicts
The work received an unrestricted grant from MDL Fonds and a grant from Stichting Reggeborgh. Disclosures included funding through INTERCEPT IHI; Health Holland TKI and the Helmsley Foundation Charity Fund; Health Holland TKI projects MICROPIAL and BIOMARKERS; EU Horizon METHYLOMIC; and support from Bristol Myers Squibb, Friesland Campina, Trained Immunity xD, AcelaBio, Alimentiv, and Nutricia. Other authors reported no potential conflicts; all authors submitted ICMJE forms.
Source
Read the original publication (published online October 10, 2025; Journal of Infectious Diseases 2026;233(2):e570–e581).
Study type, population, and sample scale
- Study type
- Observational multicenter matched case-control analysis embedded in a prospective cohort
- Population
- Very preterm infants in ten NICUs in the Netherlands and Belgium, with matched controls; the analyzed cases had culture-proven Candida late-onset sepsis in blood and/or cerebrospinal fluid
- Sample scale
- 8 Candida-LOS cases and 16 matched controls; 22 case and 56 pooled-control stool samples were analyzed before read-quality exclusions. Final per-assay sample counts are not available in the reviewed full text.
Results
- Across the t=−5 through t=0 interval, mean fungal amplicon yield was 16.23 ± 9.43 ng/µL in the 8-case group versus 5.45 ± 7.17 ng/µL in the 16-control group (P<.001). The groups contributed 22 and 56 stool samples, respectively, before read-quality exclusions; final per-assay counts are not available in the reviewed full text.
- Across the same interval, mean bacterial amplicon yield was 8.68 ± 5.53 ng/µL in the 8-case group versus 14.39 ± 3.32 ng/µL in the 16-control group (P=.02); sample counts before read-quality exclusions were 22 and 56, respectively.
- The relative fecal Candida-abundance measure was 0.998 ± 0.003 in the 8-case group versus 0.673 ± 0.368 in the 16-control group across the interval (P=.036); the pre-exclusion sample counts were 22 and 56, and final per-assay counts are not available in the reviewed full text.
- The ITS/16S amplicon-yield ratio was about 15-fold higher in cases than controls across the interval (P<.001), but the difference was not significant when limited to the day before diagnosis. The groups included 8 and 16 infants and contributed 22 and 56 samples before read-quality exclusions.
Limitations
- The analysis included 8 Candida-LOS cases and 16 matched controls; 22 case and 56 control samples were available before read-quality exclusions.
- Only a selected fraction of the cohort contributed case stool, and final per-assay denominators are not available in the reviewed full text; estimates from this small sample may be unstable.
- Fungal profiles and amplicon yields were measured from stool.
- Relative sequencing and PCR-yield measures are not direct counts of viable fungi and do not show that organisms entered the bloodstream from the gut.
- Stool and blood/CSF isolates were not compared by whole-genome strain analysis.
- The study cannot establish a gut-to-blood route; coexisting skin colonization and a skin-to-blood route remain plausible.
- The case and control groups differed in antimicrobial exposure and feeding/delivery characteristics.
- Matching does not remove all confounding, and reverse causation is possible.
Applies to
- Associations between fecal fungal/Candida signals and culture-proven Candida late-onset sepsis in the eight very preterm NICU cases and their matched controls
Does not apply to
- Term infants, older children, or adults; the cause or route of infection in an individual; stool testing as a diagnostic or predictive test; neonatal treatment, prevention, or care recommendations
Contribution to the current conclusion
Contextual
The study reports an association between fecal fungal/Candida signals and later culture-proven Candida late-onset sepsis in a very small, selected group of very preterm infants. It did not match stool and blood/CSF strains and does not establish a gut-to-blood route, support a neonatal care recommendation, or change adult broad-Candida claims.
Related evidence mapPractical change or no change
No practical change. This is a small observational association in very preterm infants; it does not prove a gut source, support neonatal treatment or prevention advice, or change the claims or practical action on the linked evidence map.
Funding and conflicts
- Funding
- Unrestricted grant from MDL Fonds and a grant from Stichting Reggeborgh.
- Conflicts
- I.A.M. van Thiel was funded through INTERCEPT IHI; I.A.M. Kreulen through Health Holland TKI and the Helmsley Foundation Charity Fund; W.J. de Jonge through Health Holland TKI projects MICROPIAL and BIOMARKERS and EU Horizon METHYLOMIC, and received funding from Bristol Myers Squibb, Friesland Campina, Trained Immunity xD, AcelaBio, and Alimentiv; T. de Meij and H. Niemarkt received an unrestricted grant from Nutricia. Other authors reported no potential conflicts, and all submitted ICMJE forms.
- Paper date
- 2025-10-10
Sources
View all 1 source
- The Gut as a Source of Infection for Fungal Pathogens: Increased Fecal Candida albicans Precedes Onset of Candida Late-Onset Sepsis in Very Preterm Infants
- Source type
- Observational study
- Published
- 2025-10-10