Digoxin for reduction of circulating tumor cell cluster size in metastatic breast cancer: a proof-of-concept trial

Abstract

The presence of circulating tumor cell (CTC) clusters is associated with disease progression and reduced survival in a variety of cancer types. In breast cancer, preclinical studies showed that inhibitors of the Na⁺/K⁺ ATPase suppress CTC clusters and block metastasis. Here we conducted a prospective, open-label, proof-of-concept study in women with metastatic breast cancer, where the primary objective was to determine whether treatment with the Na⁺/K⁺ ATPase inhibitor digoxin could reduce mean CTC cluster size. An analysis of nine patients treated daily with a maintenance digoxin dose (0.7-1.4 ng ml^-1 serum level) revealed a mean cluster size reduction of -2.2 cells per cluster upon treatment (P = 0.003), meeting the primary endpoint of the study. Mechanistically, transcriptome profiling of CTCs highlighted downregulation of cell-cell adhesion and cell-cycle-related genes upon treatment with digoxin, in line with its cluster-dissolution activity. No treatment-related adverse events occurred. Thus, our data provide a first-in-human proof of principle that digoxin treatment leads to a partial CTC cluster dissolution, encouraging larger follow-up studies with refined Na⁺/K⁺ ATPase inhibitors and that include clinical outcome endpoints. ClinicalTrials.gov identifier: NCT03928210 .

Fig. 1. Study design and digoxin treatment response assessment.

a, Study flow chart. b, Representative images of a single CTC and homotypic and heterotypic CTC clusters (scale bar, 10 µm), stained with EpCAM, HER2 and EGFR (green) and CD45 (magenta). c,d, LME random coefficients showing a negative association between treatment and the average size of all CTC clusters (c) and among homotypic (regression coefficient −0.20, 95% CI −0.76 to 0.35) or heterotypic (regression coefficient −0.31, 95% CI −1.21 to 0.59) clusters, separately (d). LME coefficients are also shown for control nonrandomized patients, not receiving digoxin therapy (regression coefficient 0.48, 95% CI −0.10 to 1.07 for homotypic clusters; regression coefficient −0.06, 95% CI −0.70 to 0.58 for heterotypic clusters). The cross bar in d represents the LME fixed-effect coefficient. e, The average cluster size at baseline and posttreatment (day 3 or day 7) paired by patient (n = 9). The boxes represent the lower quantile, median and upper quantile. The vertical lines show the range of values, and the gray lines connect paired values. P values were calculated using the one-sided paired t-test. f, The fold change of the average CTC cluster size post- over pretreatment in treated patients. In control patients, the fold change of the average CTC cluster size at day 3 or day 7 (according to smaller cluster size) over baseline is shown. Each point represents an individual patient, and the cross bar represents the median. P values were calculated using the one-sided Wilcoxon rank-sum test. g, Negative association between digoxin levels and normalized size of CTC clusters at day 7 (linear regression P = 0.14, β = −4.65). The points represent individual patients, the line represents the linear regression model and the shaded area represents the 95% CI of the fitted line. CTCAE, Common Terminology Criteria for Adverse Events; BM, bone marrow; ECG, electrocardiogram; SAE, serious adverse event. Panel a was created with BioRender.com. Source data

Extended Data Fig. 1. Predicted average size of circulating tumor cell (CTC) clusters for each control and treated patient using linear mixed effect.

The lines represent the regression line for the linear model for each patient and the points represent the predicted values at each timepoint according to the model. Error bars indicate the 95% confidence interval.

Extended Data Fig. 2. CTC cluster size in control and digoxin-treated patients.

(a) Average cluster size at baseline and at follow up (smallest cluster size, day 3 or day 7) paired by patient (control cohort, n = 9). Boxes represent lower quantile, median and upper quantile. Vertical lines indicate the range of values, grey lines connect paired values. (b) Average CTC cluster size across timepoints normalized by the average CTC cluster size at baseline (0) in control and treated patients. Values in Y-axis are represented in log2 scale. The colored lines indicate individual patients.

Extended Data Fig. 3. CTC proportions in control and digoxin-treated patients.

Linear mixed effects (LME) random coefficients showing the association between assessment time points and the proportion of all types of circulating tumor cells (CTC) clusters (secondary endpoint) per control and treated patients (a) or split into heterotypic and homotypic clusters (b). The cross bar in b represents the LME fixed effect coefficient.

Extended Data Fig. 4. Metastatic potential as a function of different CTC cluster sizes.

(a) Schematic illustration of the experimental design. (b) Representative images of bioluminescent signals in mice within specified experimental groups. (c) Plot showing normalized bioluminescent signal across experimental groups (minimum n = 4 per group, where points represent individual mice) three weeks after tumor cell injection. P = 0.0089 by two-sided Mann-Whitney test. The cross bar represents the mean. Panel (a) was created with BioRender (https://biorender.com).

Extended Data Fig. 5. Longitudinal RNA sequencing of CTCs during digoxin treatment.

(a) The diagram shows the clinical timeline of patient nr. 5, including treatment history (boxes), radiological assessment (dashed lines) and serum tumor marker (CA 15-3) (grey). (b) Workflow for molecular analysis, including microfluidic separation and capture, robotic micromanipulation, pooling of circulating tumor cells (CTCs) into individual tubes and RNA sequencing. (c) Heatmap displaying differentially expressed genes (DESeq Wald test adjusted P value < 0.05) between CTC pools obtained prior to digoxin intake (n = 5 pre-digoxin pools) and post-digoxin intake (n = 2 post-digoxin pools). Values are displayed as gene-scaled (z-score) log2 counts per million mapped reads after normalization. (d) Enriched pathways among downregulated (top) and upregulated (bottom) genes of CTCs post-digoxin intake (over-representation analysis adjusted P value < 0.05). Shown is a reduced list of ontologies after simplification using semantic similarity. TP, treatment pause; PD, progressive disease; SD, stable disease; TAM, tamoxifen. Panels (a) and (b) were created with BioRender (https://biorender.com).

Extended Data Fig. 6. Pathway analysis in CTCs upon treatment with digoxin.

Top 30 enriched pathways (over-representation analysis adjusted P value < 0.05) among downregulated (top) and upregulated (bottom) genes in circulating tumor cells (CTCs) deriving from blood samples post-digoxin intake.