Cabozantinib and nivolumab with or without live bacterial supplementation in metastatic renal cell carcinoma: a randomized phase 1 trial

Abstract

Supplementation with CBM588, a bifidogenic live bacterial product, has been associated with improved clinical outcomes in persons with metastatic renal cell carcinoma (mRCC) receiving nivolumab and ipilimumab. However, its effect on those receiving tyrosine kinase inhibitor-based combinations is unknown. In this open-label, randomized, investigator-initiated, phase 1 study, 30 participants with locally advanced or mRCC with histological confirmation of clear cell, papillary or sarcomatoid component were randomized in a 2:1 fashion to receive cabozantinib (an inhibitor of vascular endothelial growth factor receptor, MET and AXL) and nivolumab (anti-programmed cell death protein 1) with or without CBM588 as first-line treatment. Metagenomic sequencing was performed on stool samples to characterize their gut microbiome at baseline and 13 weeks into treatment. The primary endpoint was a change in the relative abundance of Bifidobacterium spp.; secondary endpoints included objective response rate (ORR), progression-free survival (PFS) and toxicity profile. The primary endpoint of the study was not met and the addition of CBM588 to cabozantinib and nivolumab did not result in a difference in the relative abundance of Bifidobacterium spp. or alpha diversity (as measured by the Shannon index). However, ORR was significantly higher in participants treated with CBM588 compared to those in the control arm (14 of 19, 74% versus 2 of 10, 20%; P = 0.01). PFS at 6 months was 84% (16 of 19) and 60% (6 of 10) in the experimental and control arms, respectively. No significant difference in toxicity profile was seen between the study arms. Our results provide a preliminary signal of improved clinical activity with CBM588 in treatment-naive participants with mRCC receiving cabozantinib and nivolumab. Further investigation is needed to confirm these findings and better characterize the underlying mechanism driving this effect.ClinicalTrials.gov identifier: NCT05122546.

Fig. 1. CONSORT (Consolidated Standards of Reporting Trials) diagram.

CONSORT diagram showing the flow of participant enrollment and treatment.

Fig. 2. Microbiome assessment in participants with mRCC treated with cabozantinib + nivolumab with or without CBM588 revealed no significant changes in Bifidobacterium spp. with treatment.

Analyses were performed using n = 58 stool samples from n = 29 participants (n = 10 participants in the cabozantinib + nivolumab arm and n = 19 participants in the cabozantinib + nivolumab with CBM588 arm). a, Change in relative abundance of Bifidobacterium spp. from baseline to week 13 in participants by treatment arm. A two-sided Wilcoxon signed rank test was used to perform comparisons between two time points within the same treatment arm and a two-sided Mann–Whitney U test was used for comparisons between the two arms. b, Difference in the relative abundance of several bacterial species in baseline samples from participants receiving CBM588 (n = 19) compared to those in the control arm (n = 10). The ANCOM-BC was used to perform comparisons in the CBM588 arm relative to the control arm at baseline. Data are presented by effect size depicting features with LFC > 1 and P < 0.05 (per two-sided z-test using the Wilcoxon test statistics). Error bars represent the effect size error (SE). c, Difference in the relative abundance of several bacterial species in week 13 samples from participants receiving CBM588 (n = 19) compared to those in the control arm (n = 10). ANCOM-BC was used to perform comparisons in the CBM588 arm relative to the control arm at week 13. Data are presented by effect size depicting features with LFC > 1 and P < 0.05 (per two-sided z-test using the Wilcoxon test statistics). Error bars represent the SE. d, Microbial richness between baseline and week 13 in participants with the cabozantinib + nivolumab treatment. The Shannon entropy diversity metric was used to compare two time points. The median and interquartile range are depicted, with whiskers extending to the minimum and maximum values. e, Microbial richness between baseline and week 13 in participants with the cabozantinib + nivolumab with CBM588 treatment. The Shannon entropy diversity metric was used to compare two time points. The median and interquartile range are depicted, with whiskers extending to the minimum and maximum values. f, Relative similarities of microbiome composition as a principal coordinate analysis (PCoA) of Bray–Curtis distances between control samples. g, Relative similarities of microbiome composition as a PCoA of Bray–Curtis distances between samples of participants receiving CBM588.

Fig. 3. Differentially abundant microbial metabolic pathways in participants with mRCC treated with cabozantinib + nivolumab with or without CBM588.

a, Differentially abundant microbial metabolic pathways between baseline and week 13 in participants with the cabozantinib + nivolumab treatment (n = 10). b, Differentially abundant microbial metabolic pathways between baseline and week 13 in participants with the cabozantinib + nivolumab with CBM588 treatment (n = 19). ANCOM-BC was used to perform comparisons between two time points within the same treatment arm (differential features with LFC > 1 and P < 0.05 are indicated). The P value was calculated through a two-sided z-test using the Wilcoxon test statistics. Error bars represent the SE.

Fig. 4. Clinical outcomes in participants with mRCC treated with cabozantinib + nivolumab with or without CBM588.

a, ORR. b, Best change in target lesions. c, PFS. The data are from n = 29 participants (19 participants in the cabozantinib + nivolumab with CBM588 arm and 10 participants in the cabozantinib + nivolumab arm). The Kaplan–Meier log-rank test was used to compare survival between the two arms.

Extended Data Fig. 1. Differentially abundant bacterial species in stool microbiome.

Heatmap of differentially abundant bacterial species in stool microbiome across patients in (a) control arm (n = 10) and (b) experimental arm (n = 19) of the study at baseline and week 13 (top 100 most abundant bacterial species are presented).

Extended Data Fig. 2. Differentially abundant microbial feature assessment in patients with or without objective response.

(a) Difference in the relative abundances of several bacterial species in baseline samples of patients without objective response (n = 13) compared to patients with response (n = 16). The ANCOM-BC (Analysis of Composition of Microbiomes with Bias Correction) was used to determine the differentiating features in patients without response relative to those with response at baseline (differential features with log fold change (LFC) greater than 1 and p < 0.05 are indicated). The P value was calculated through two-sided z-test using the W test statistics. Error bars represent effect size error (SE). (b) Difference in the relative abundances of several bacterial species in week 13 samples of patients without objective response (n = 13) compared to patients with response (n = 16). The ANCOM-BC was used to determine the differentiating features in patients without response relative to those with response at week 13 (differential features with LFC greater than 1 and p < 0.05 are indicated). The P value was calculated through two-sided z-test using the W test statistics. Error bars represent effect size error (SE).

Extended Data Fig. 3. Changes in levels of circulating cytokines from baseline to week 13 by treatment arm – Part 1.

A total of 53 samples from 30 patients were available for the final analysis. Two-sided Wilcoxon matched-pairs test was used to compare the levels of cytokines at the two prespecified timepoints. Two-sided Mann–Whitney U test was used for comparisons between the two arms.

Extended Data Fig. 4. Changes in levels of circulating cytokines from baseline to week 13 by treatment arm – Part 2.

A total of 53 samples from 30 patients were available for the final analysis. Two-sided Wilcoxon matched-pairs test was used to compare the levels of cytokines at the two prespecified timepoints. Two-sided Mann–Whitney U test was used for comparisons between the two arms.

Extended Data Fig. 5. Assessment of circulating cytokine levels in patients with or without response – Part 1.

Changes in circulating cytokine levels from baseline to week 13 in patients with or without response were assessed using the two-sided Wilcoxon matched-pairs test. Comparison of circulating cytokine levels at baseline and week 13 between patients with or without response was performed using the two-sided Mann-Whitney U test.

Extended Data Fig. 6. Assessment of circulating cytokine levels in patients with or without response – Part 2.

Changes in circulating cytokine levels from baseline to week 13 in patients with or without response were assessed using the two-sided Wilcoxon matched-pairs test. Comparison of circulating cytokine levels at baseline and week 13 between patients with or without response was performed using the two-sided Mann-Whitney U test.

Extended Data Fig. 7. Changes in circulating immune cell populations between baseline and week 13 by treatment arm.

Changes in the populations of (a) CD8+ T cells and (b) CD4+ regulatory T cells from baseline to week 13 in cabozantinib-nivolumab arm or cabozantinib-nivolumab with CBM588 arm. Two-sided Wilcoxon matched-pairs test was used to compare the immune cell populations between the two prespecified timepoints.