Escribir un comentario

Summary: 

This preprint provides a secondary analysis of Chng et al. which sampled the Tan Tock Seng Hospital in Singapore to investigate colonization patterns and the resistance of the hospital with samples from an office environment acting as a control group. This preprint uses the data from Chng et al. to investigate and compare the microbiome structure and composition of the hospital, metro stations, and the office environment. The authors analyzed Chng et al.’s data by calculating Shannon Diversity Indices for each environment and then used co-occurrence networks to perform network analysis. After characterizing the microbiomes, comparing diversity, prevalence of pathogens, and investigating the structure of the microbiomes, the authors performed metabolite modeling to predict the types of metabolites being produced and resource utilization of the microbes in the network using a SMETANA test. These findings allowed them to dive deeper into the associations found through the co-occurrence networks and create hypotheses surrounding the interactions between specific species found in the environment. Using the calculated SMETANA Scores they created a Pathogen Support Index (PSI) which provides a quantitative metric of how well microbial interactions (based on metabolite exchanges) support the persistence/presence of pathogens in that environment. 

In the study, the authors found that the metro station co-occurrence network had the highest modularity, with bacteria separating into five distinct modules compared to three and two for the hospital and office networks respectively. They found that the type of metabolic exchanges are extremely environment dependent. Office communities had more closely related and functionally redundant taxa which had high resource overlap and low interaction potential, indicating competitive environments. The hospital and metro communities had higher interaction potential and low resource overlap indicating that the communities were complementary and not as competitive. In the end, communities associated with hospitals had higher Pathogen Support Indexes. 

Recommendation:

This preprint provides insight into the underlying ecological mechanisms (metabolite exchanges) that drive pathogen persistence in urban environments which is a relatively understudied field of public health and has strong implications on infection control. Their Pathogen Support Index is a novel method that uses genomic data to predict metabolic interactions in a community to quantify the support and potential persistence of particular organisms (pathogens) in an environment. However, there are major limitations and confounders (listed below) that remain unaddressed in this preprint and require revision before publication. 

Major Concerns:

• Chng et al. collected samples at three different time points (with one week and 1.5 years between each sampling point) but this preprint did not clarity which time points were used to create the microbial networks and calculate the Pathogen Support Indices (PSI) which is a major concern as clustering samples from different time points means that the observed microbes may not have actually co-occurred or interacted. 

  • To address this, the authors should include a sentence or two in the methods section to clarify which time point samples were included in the network analyses and SMETANA tests to clarify how the temporality of samples were taken into consideration.

  • If samples were clustered and not separated by the time point in which they were sampled, that should be included in the limitations section of the discussion and the implications of this clustering on the network analysis and PSI should be discussed. 

• Unspecified significance and implications of the novel Pathogen Support Index proposed in the preprint. While reading it was unclear the novelty of the PSI that was created in this preprint. It would be useful to know how it compares to other indexes or methods of quantifying support of specific species/pathogens or what it provides. 

  • To address this, the authors should include a section in the discussion explaining the significance of their PSI, how it differs from other methods of analyzing the support of a specific species in a community network, what benefits this index provides that others do not, and include a qualitative comparison of this method to others that are used in the field of ecology.

• Unaddressed confounder of disinfection rates between environments. In the discussion, the manuscript states that the higher PSI is consistent with other findings that show hospital ecological niches promote pathogen survival and persistence despite regular disinfection. However, this differential disinfection rate can also be a co-selector for antimicrobial resistance in pathogens which provides an alternative explanation for the greater Pathogen Support Index (Sheikh et al. 2025). This alternative perspective would indicate that pathogen persistence is more likely to be higher in hospitals because of disinfection practices, not despite of.

  • The authors should consider addressing these alternative perspectives and further explain how co-selection practices for pathogens based on their environment, might be affecting their calculated PSI. 

Minor Concerns:

• Unaddressed Limitation: in the second paragraph of the results section, the preprint states, “to avoid sampling biases due to unequal to avoid sampling biases due to unequal sample sizes, 100 iterations of rarefied subsampling were performed, with 30 samples per biome randomly selected for each iteration”. However, there were only 30 samples in the office environment meaning that only one sample was able to be randomly selected from the data pool preventing appropriate control for sampling biases. In the limitations section of the discussion, unequal sample sizes are addressed, however, as the office environment is so small the methods done do not appropriately address the issue of sampling biases for the office environment. 

  • To address this, in the limitations section of the discussion the authors should include that one of the methods they used to avoid sampling biases may not have fully addressed the sampling biases, specifically for the office environment. 

Considerations for Future Analysis:

• There was lacking clarity regarding which samples were analyzed from Chng et al. and whether or not stratification based on sites was considered for the metabolite network and PSI results. Chng et al. sampled from three types of wards (isolation rooms, MDRO wards, and standard wards) and at a standardized set of sites (aerator, sink trap, bed rail, etc.). Taking into consideration the ward type could be an important control for the creation of Pathogen Support Index that could provide additional information on its accuracy and validity.

  • To address this, the authors could perform stratified PSI’s based on the type of ward and include a sentence or two explaining if there were any differences and the implications of any differences or similarities on the validity of their proposed PSI.

References:

Sheikh M, Gholipour S, Ghodsi S, Nikaeen M. Co-selection of antibiotic and disinfectant resistance in environmental bacteria: Health implications and mitigation strategies. Environ Res. (2025) 267:120708. doi: 10.1016/j.envres.2024.120708, Epub 2024 Dec 26. PMID: 39732420

Competing interests

The authors declare that they have no competing interests.

Use of Artificial Intelligence (AI)

The authors declare that they did not use generative AI to come up with new ideas for their review.