Fig. 5: Metabolic and immune profiling after FMT according to clinical efficacy.

a, Longitudinal plasma analysis of quinolinic acid (top) and kynurenine (bottom) metabolites measured by UHPLC−MS; for each metabolite, two representations are shown: smooth trajectories of the model-predicted mean intensity for R (blue) and NR (red) over time. Shaded bands represent 95% confidence intervals around the fitted mean; P values indicate statistical significance of each smooth term. Right: pointwise differences between the predicted mean of the two groups (R − NR) with 95% confidence intervals (gray shaded area). Pink horizontal segments indicate time periods where confidence intervals exclude zero, representing significant between-group differences. Data were fitted using generalized additive mixed models with Tweedie distribution and random subject intercepts (n = 100 samples from n = 35 patients; Methods). b, MetaCyc representation of the tryptophan pathway comparing baseline to after FMT in NSCLC R (n = 62 samples from n = 15 patients) and melanoma R (n = 71 samples from n = 15 patients); P values were computed by the Wald test (two-sided) from DESeq2 analysis. c–d, Spearmanʼs r correlation (two-sided) of flow cytometry analysis from NSCLC and melanoma patient PBMCs after FMT according to low versus high median SGB loss from baseline for frequency of CD69⁺CD8⁺ T cells (c) and CD127^lowCD25^highCD4⁺ regulatory T cells (d). Spearmanʼs r correlation (two-sided) of flow cytometry analysis from NSCLC and melanoma patient PBMCs after FMT according to low versus high plasma quinolinic acid for PD-1⁺CD8⁺ T cells (e), PD-1⁺ effector memory CD8⁺ T cells (f) and effector memory regulatory T cells (g) from n = 50 samples. Abs log₂FC, absolute log₂ fold change; EM, effector memory; Hi, high; Lo, low; PWY−6629, superpathway of L-tryptophan biosynthesis; T_reg, regulatory T; TRPSYN−PWY, L-tryptophan biosynthesis.