Exploratory Analysis of Plasma Angiotensin-Converting Enzyme 2 and Angiotensin Peptides as Candidate Discriminatory Signals in Breast Cancer: A Pilot Case–Control Study

Figures

• 

  Figure 1. CONSORT style flowchart illustrating sample enrollment, processing, and analysis. The diagram shows the progression of the study from initial sample of n = 38 through technical analysis and final model inclusion. A total of 21 breast cancer patients and 17 healthy controls were enrolled. 5 samples in total including 1 control and 4 patients were excluded from the RAS peptide analysis due to hemolysis and technical failure. For the ACE2 ELISA experiment, 1 control and 2 patient samples were excluded due to their coefficient of variance between the replicated were higher than 20% threshold. For the cytokine analysis, 3 samples were omitted for IL-10 analysis, whereas the rest of the cytokine panel (IL-1α, IL-1β, IL-6, IL-8, IFN-γ, and TNF-α) had 50% of samples below LOQ. Therefore, many samples were excluded, making the finding for these 6 specific analytes unreliable.

• 

  Figure 2. The boxplots overlayed with scatterplots of plasma concentrations of Ang 1-7 (a), Ang II (b), and the Ang 1-7/Ang II ratio (c) in the control group (n = 16) and breast cancer patients (n = 17). The boxplot of plasma ACE2 concentration (d) in the control group (n = 17) and breast cancer patients (n = 19). Statistical comparisons were performed using the Mann–Whitney U test; statistical significance was set at p < 0.05.

• 

  Figure 3. Bar graphs with scatter plots of plasma concentrations of IL-10 (a), and IL-8 (b) in the control group and in breast cancer patients. Data are presented as mean ± SEM. Statistical comparison between the groups (Controls, n = 15 and Patients, n = 19) was performed using Student’s t-test for IL-10, and statistical significance was set at p < 0.05. No group comparison was carried out for IL-8 (n = 12, 7) due to at least one group having more than 50% samples below limit of quantification.

• 

  Figure 4. The correlation graphs of plasma concentrations (ng/mL) of Ang II against BMI (kg/m2) (a), Ang 1-7 (ng/mL) against platelets (counts per µL) (b), Ang II (ng/mL) against Hemoglobin (g/dL) (c), and Ang 1-7/Ang II ratio against hemoglobin (g/dL) (d). A statistical test was performed using Spearman’s correlation coefficient (r), with p < 0.05 considered statistically significant.

• 

  Figure 5. Receiver operating characteristic (ROC) curves showing the performance of the ACE2 univariate model (a) and the multivariate model using ACE2, Ang 1-7, and Ang II together (b) at an arbitrary threshold of 0.5. The ROC curves with calculated Youden’s index (J) for the ACE2 univariate model (c) and the multivariate model using ACE2, Ang 1-7 and Ang II together (d). The dashed diagonal line represents the line of no discrimination (random classifier; AUC = 0.5). ACE2, Angiotensin Converting Enzyme; Ang, Angiotensin; AUC, Area under curve; Youden’s index (J) represented as cutoff.

• 

  Figure 6. Forest plot of multivariable logistic regression showing the odds ratios (OR) and 95% CI for angiotensin converting enzyme (ACE2), Angiotensin 1-7 (Ang 1-7) and Ang II with all biomarkers treated as continuous covariates adjusted for each other. The OR is presented on a logarithmic scale. Black dots and lines represent statistically significant associations (p < 0.05), and gray-scale circles and lines represent non-significant associations. The vertical dashed lines represent OR of 1 corresponding to no effect.

• 

  Figure 7. The Decision Curve shows that the RAS Multi-Analyte Model (bold grey line) maintains a superior net benefit over the ‘Screen All’ strategy (dashed line) across most threshold probabilities, particularly in the 20% to 50% risk range.