What’s new in microbiota research on the outcomes after allogeneic hematopoietic stem cell transplantation?

Allogeneic hematopoietic stem cell transplantation (allo-HSCT) is associated with alterations of intestinal microbiota and a low microbiota diversity, leading to an increased risk of graft-versus-host disease (GvHD) and a negative impact on overall survival (OS).¹ Two studies presented at the 61^st American Society of Hematology Meeting & Exposition (ASH), Orlando, FL, US, in December 2019, have extended this knowledge, providing correlating evidence that gut fungal microbiota (mycobiota) and certain antibiotics and dietary factors can influence the microbial composition and impact the outcome after allo-HSCT.^2,3

Gut mycobiota and outcomes after allo-HSCT²

Although fungi represent less than 1% of microbes in the gut, they contribute considerably to biomass and host-microbe interaction as their size is 100 times bigger than bacteria. In a study presented by Florent Malard from the Clinical hematology and cell therapy, Hospital Saint Antoine, Paris, FR, fecal samples were collected after completion of the conditioning regimen and just before stem cell graft infusion (Day 0), with fecal mycobiota profiles characterized by internal transcribed spacer 2 (ITS2) sequencing using MiSeq technology.

Patients

• 68 patients (37 males, 31 females)
• Median age of 60 years (range 22–74 years)
  • 34 (50%) had acute myeloblastic leukemia (AML)
  • Seven (10%) had myelodysplastic syndrome
  • 15 (12%) had myeloproliferative neoplasms
  • Five (7%) had acute lymphoblastic leukemia
  • Four (6%) had lymphoma
  • Three (4%) had aplastic anemia
• 49 (72%) patients had received a myeloablative (reduced toxicity) conditioning regimen and 19 (28%) had received a reduced-intensity conditioning regimen
• 5 patients had received antifungal medications at the time of allo-HSCT and, 54% of patients have had antibiotics with broad anaerobic coverage at the time of allo-HSCT

Results

• As was previously described, patients with low bacterial diversity had lower OS and GvHD-free survival
• A low fungal diversity score was observed in all patients; this had no impact on overall survival
• A low abundance of Candida albicans in the gut was associated with a significant increase in OS (p= 0.026) and GvHD-free, disease-free survival (p= 0.013) relative to a high abundance
• A multivariate Cox hazard analysis that included the most important parameters associated with patient outcomes yielded similar results, with a low number of Candida albicans associated with a significantly improved OS (hazard ratio (HR)= 0.235; 95% CI, 0.087–0.64; p= 0.004) and a significant improvement in GvHD-free, disease-free survival (HR= 0.37; 95% CI, 0.166–0.82; p= 0.014)
• Similarly, low numbers of Enterococcus and Klebsiella correlate with improved OS in a multivariate analysis. By contrast, abundance of Oscillibacter was associated with better OS

Conclusions

The investigators concluded that the fecal mycobiota diversity is significantly disrupted in patients undergoing allo-HSCT, with an abundance of Candida albicans found to be an independent predictor of decreased OS and GvHD-free, disease-free survival post-transplant.

Results from the ODYSSEE trial of fecal microbiota transfer in patients with AML can be found on the AML Hub.

Antibiotics, dietary factors, and dysbiosis in allo-HSCT³

Antonio Gomes from the Sloan Kettering Cancer Center in New York, NY, US, presented the results from a study assessing the impact of antibiotic exposure and dietary factors on gut bacterial composition in patients receiving allo-HSCT. The initial goal of the study was to identify meaningful clusters of microbiota composition among patients receiving allo-HSCT in order to test the impact of various environmental factors on cluster dynamics. To achieve this, 16S rRNA deep-sequencing was used to determine the bacterial composition of almost 8,000 fecal samples from 1,076 allo-HCT patients (mean age at hematopoietic SCT, 54 ± 12.9 years; 60.4% male; 34.7% with AML). Microbiota composition diversity and time patterns were visualized in t-distributed stochastic neighbor embedding (tSNE) projection, with ten distinct clusters identified. Each cluster mapped to specific taxonomic groups; each had a different microbiota diversity and a different risk of mortality. The mortality HR increased in lower-diversity clusters, with four clusters found to have a statistically significant association compared with the reference high-diversity cluster 1 (e.g. cluster 4 had a HR of 1.87 [1.01–3.5] compared to cluster 10 with a HR of 2.45 [1.48–4].

To evaluate how antibiotics and diet might impact on microbiota dynamics, regression-based predictive approaches were used to model cluster transition probabilities in terms of maintaining stability of a cluster – remaining in the same cluster over time (self-weight) or attracting transitions from other clusters over time (attractor-weight). The impact of the three most commonly used nonprophylactic antibacterial drugs was then determined using 2,359 samples from 391 allo-HCT patients collected between Day –14 to Day 7 relative to transplant.

Results

• High-diversity cluster 1, in which Ruminococcaceae and other Clostridiales were the most commonly observed taxa, was significantly destabilized by piperacillin-tazobactam (pip-tazo) exposure (β= –0.83, p< 0.05), while low diversity cluster 9, which was a Streptococcus-dominated cluster, was significantly stabilized by pip-tazo exposure (β= 1.24, p< 0.05)
• In contrast, exposure to cefepime demonstrated a trend towards increased stability both, for cluster 1 and 9
• Exposure to pip-tazo significantly increased the transition probability to cluster 9 (β= 1.81, p< 0.05), while exposure to cefepime (β= 2.86, p< 0.05) significantly increased the transition probability to Enterococcus-dominated cluster 10

The same concepts were then applied to evaluating the impact of diet on cluster probabilities in 46 allo-HCT patients for whom dietary information was available, with 242 fecal samples sequenced. In these samples, the investigators observed that:

• Protein consumption was associated with destabilization of cluster 1 (β= –0.10, p< 0.05)
• Fat consumption was associated with an increased stability of cluster 1 (β= 0.14, p< 0.05)
• No significant associations were found in terms of the attracting transition to other clusters

Conclusions

This study has confirmed that clusters of microbiota can provide meaningful data in terms of patient outcomes in allo-HSCT, providing a framework that predicts the effect of antibiotics and environmental exposures in microbiota dynamics that may be used in future research to potentially improve outcome in allo transplanted patients.