In a cohort study at Memorial Sloan Kettering Cancer Center in New York, which included adult recipients of allogeneic hematopoietic cell transplant (allo-HCT) from 2017 to 2022, participants were invited to share their dietary intake three times a week after being admitted for the procedure. Neutrophil engraftment was marked by the first of three consecutive days with a neutrophil count of at least 500 per microliter. Five patients who died prior to achieving engraftment did not factor into the median time analysis. However, there were no issues with follow-up for other patients. For the mortality analysis, one patient who passed away before day 12 was excluded. Data collection for this study wrapped up in April 2023.
Regarding nutrition, each meal tray had a printout from the hospital’s kitchen system, prompting patients to indicate their consumption levels—0, 25%, 50%, 75% or 100%—right after each meal. Dieticians or research staff collected this data three times weekly, filling in any gaps at the bedside and encouraging patient participation. The collected data was entered into the kitchen software for analysis and verified by a research dietitian to ensure objectivity, especially in cases of obvious data errors. The focus of the analysis was on the dietary intake during the hospital stay, not on patterns prior to admission.
Each food item was assigned an eight-digit code based on the U.S. Food and Nutrient Database for Dietary Studies (FNDDS). This coding system helps distinguish various foods within larger categories. During the analysis, water content was removed from food weights to better assess nutrient intake. For enteral nutrition administered as a liquid, the nutrient weight was calculated by adding the weights of protein, fat, and carbohydrate components. Thus, references to food increases—like a 100g rise in sweets—indicate the dehydrated weight of the foods, ensuring a more accurate nutrient comparison.
A food tree was created featuring 622 unique FNDDS items consumed by patients, divided into nine major categories, including grains, vegetables, meats, and fruits. Any foods with multiple classifications in the FNDDS were manually categorized based on the best fit. The assessment of dietary data utilized a method that visualizes similarities in meal compositions over time, using specialized statistical tools to find correlations between dietary intake and microbiome dynamics.
Patient stool samples were processed under specific criteria and transported to labs for microbiome profiling through 16S rRNA sequencing. Various methods refined sample integrity to ensure high-quality results. Statistical analyses were conducted to evaluate associations between dietary patterns and clinical outcomes, utilizing Bayesian models for a thorough understanding of the data.
Mice were also included in related studies to analyze dietary and antibiotic effects on microbiome health, examining consumption patterns and physiological responses to different diet and treatment regimens. Several statistical analyses evaluated these factors to identify any significant relationships between diet composition, antibiotic exposure, and microbiome diversity.
Ethical approvals were secured for both human and animal studies, adhering to established protocols to ensure participant safety and scientific integrity.
