Abstract
BACKGROUND AND AIMS: Multimorbidity and polypharmacy are common in long-living patients and have been linked to alterations in gut microbiota. However, the prognostic relevance of these microbiota changes remains unclear in this population. We hypothesized that clinical parameters and gut microbial profiles are associated with survival outcomes. This exploratory study aimed to investigate the relationships between intestinal microbiota, clinical variables, and 18-month mortality in long-living male patients undergoing multi-drug therapy. METHODS: 33 long-living male adults (mean age 94 ± 4 years) were enrolled. Upon admission, clinical parameters and stool samples were collected. The composition of intestinal microbiome was examined using 16S rRNA gene sequencing technology. Subsequently, the patients were categorized into death and no-death groups based on their survival status after 18 months. Risk factors associated with 18-month mortality in patients were evaluated. RESULTS: The 18-month mortality rate was 48.5%. In this exploratory analysis, higher CIRS-G scores and creatinine levels were independently associated with mortality (HR: 1.323 and 1.007, respectively). CIRS-G had high prognostic value (AUC = 0.967). IL-6 levels and renal/hepatic/pancreatic comorbidities were also elevated in the death group. Notably, Erysipelatoclostridium was enriched in survivors and Sutterella in non-survivors, with correlations to clinical parameters. CONCLUSIONS: CIRS-G scores and serum creatinine levels demonstrated potential predictability for prognostic value in forecasting mortality among long-living male patients. Moreover, a potential association between CIRS-G score and intestinal microbiome was observed. These results underscore the intricate impact of comorbidities and polypharmacy on intestinal microbiome composition and clinical consequences. Clinically, the CIRS-G score could serve as a valuable tool for identifying high-risk elderly patients who may benefit from closer monitoring and individualized interventions. However, the small sample size, monocentric design, and potential confounders limit generalizability. These findings support further validation in larger cohorts.