Supplemental Figure 2. Comparison Between AQUA Scores Within the Total Compartment And

Supplemental Figure 1. Schematic for multiplexed immunofluorescencestaining for tumor infiltrating lymphocytes. Antibodies against three markers for TIL subpopulations (CD3, CD8, and CD20) are multiplexed along with cytokeratin and DAPI. HRP-conjugated secondary antibodies are matched by primary antibodies by isotype. This results in the use of five separate fluorescent channels – DAPI, FITC, Cy3, Cy5, and Cy7. Benzoic hydrazide is used to quench HRP activation before incubation with additional secondary antibodies.

Supplemental Figure 2. Comparison between AQUA scores within the total compartment and AQUA scores calculated within the stromal compartment and tumor mask. The stromal compartment was created by subtracting the tumor mask, consisting of cytokeratin positive epithelial cells, from the total compartment, derived from the DAPI signal. (A) Correlation between AQUA scores for CD3 within the total compartment stromal compartment was excellent, whereas (B) correlation between AQUA scores for CD3 in the total compartment and tumor mask was inferior. Likewise, AQUA score for CD8 in the total compartment correlated with (C) CD8 AQUA score in the stromal compartment than with (D) CD8 in the tumor mask. Finally, (E) correlation between CD20 AQUA score in the total compartment and stromal compartment was excellent, but (F) correlation between CD20 AQUA score in the total compartment and tumor mask was poor.

Supplemental Figure 3. Reproducibility of quantitative immunofluoresence assay on serial sections of an invasive breast tissue microarray. AQUA scores for (A) stromal CD3 expression, (B) stromal CD8 expression, (C) stromal CD20 expression have excellent correlation on two serial sections of index microarrays stained independently alongside the neoadjuvant cohort.

Supplemental Figure 4. Correlation between TIL AQUA scores and clinicopathologic characteristics. Box plots demonstrating the correlation between stromal AQUA Scores for (A) CD3, (B) CD8, and (C) CD20 and estrogen receptor, progesterone receptor, and HER2 status. Correlations between all four scores and triple negative subtype were also determined. Significant associations (p < .05) are denoted in red.

Supplemental Figure 5. Determination of cutpointsbetween low and high expression of TIL biomarkers. Joinpoint software was used to select three possible cutpoints based on trends in the sequential arrangement of AQUA scores from low to high. The central joinpoint was ultimately chosen as the cutpoint between high and low expression for CD3, CD8, and CD20.

Supplemental Figure 6. Sensitivity and specificity for AQUA analysis of CD3, CD8, and CD20 and histopathologic assessment of TILs on H&E slides. Sensitivity and specificity were assessed by generation of ROC curves for AQUA scores within the stromal compartment for (A) CD3, (B) CD8, and (C) CD20 and for (D) TILs evaluated on H&E slides.

Supplemental Figure 7. Correlation between AQUA Score for CD3, CD8, and CD20 and response stratified by molecular subtype. Correlation between stromal AQUA scores for (A) CD3, (B) CD8, and (C) CD20, and pCR stratified by ER status, HER2 status, and among patients with the triple negative subtype. Significant p-values (p < .05) are denoted in red.

Supplementary Figure 1

Supplementary Figure 2

Supplementary Figure 3

Supplementary Figure 4

Supplementary Figure 5

Supplementary Figure 6

Supplementary Figure 7