Table 1.
Neoadjuvant chemotherapy dataset used for analyses.
Fig 1.
Analysis of lymphocyte infiltration and immune signaling gene expression in IM subtype TNBC.
(A) Representative H&E images TNBC tumors that were scored for mild (0–10%) moderate (20–40%) or intense (>50%) levels of infiltrating lymphocytes relative to total nuclei. (B) Boxplot shows IM subtype gene expression correlation for each TCGA TNBC tumor binned into mild, moderate or intense levels by pathological evaluation of H&E slides. (C) Beeswarm plot shows the IM subtype correlation for 587 TNBC tumors according to TNBC subtypes (red). Tumors that were initially subtyped as IM, but have strong secondary correlations to other subtypes, are shown in black. (D) Heatmap shows expression of immuno-regulatory genes across 587 TNBC tumors ranked by increasing correlations to the IM TNBC centroid.
Fig 2.
LCM followed by gene expression analysis of tumor epithelium and adjacent stroma identifies normal stromal cell gene expression in the MSL TNBC subtype.
(A) Representative images of H&E (upper left), tissue before (upper right) and after (lower left) LCM and cells isolated (lower right) for gene expression analysis in adjacent tumor stroma (asterisk). (B) Scatter-plot shows differentially expressed genes (FC> 2, FDR< 0.01) between LCM-isolated tumor epithelium and stromal cells from 10 TNBC tumors. (C) Bar-plot shows the correlation of each tumor epithelial and stromal pairs to the MSL subtype.
Table 2.
Gene ontology analysis of differentially expressed genes between tumor epithelial and stromal cell.
Table 3.
TNBC subtype correlations from matched LCM tumor epithelial.
Fig 3.
Molecular subtype distribution and survival analysis of TNBC samples stratified by PAM50, TNBCtype or refined TNBCtype-4.
Piecharts show the distribution of 767 TNBC samples by (A) PAM50 (B) TNBCtype or (C) refined TNBCtype-4. Kaplan-Meier curves show overall survival for TNBC patients stratified by (D) PAM50 (E) TNBCtype or (F) refined TNBCtype-4 or relapse-free survival stratified by (G) PAM50 (H) TNBCtype or (I) refined TNBCtype-4. P-values shown were determined by logrank test. * indicates significant (p<0.05) pairwise survival differences between a subtype and all other subtypes combined not adjusted for multiple comparisons.
Table 4.
Clinical parameters of TNBC subtypes.
Fig 4.
Chemotherapy response and distant relapse-free survival of TNBC treated with neoadjuvant anthracycline and taxane relative to PAM50 or refined TNBCtype-4 subtyping.
Barplots show pCR rates achieved for patients stratified by (A) TNBC, (B) PAM50 or (C) TNBCtype-4. Dotted horizontal line indicates pCR for the individual cohort. Statistical significance determined by Fisher’s exact test. Kaplan-Meier plots display distant relapse-free survival from GSE25066 for (D) TNBC patients; (E) TNBC patients stratified by pCR or RD; (F) TNBC patients stratified by PAM50; and (G) TNBC patients stratified by refined TNBCtype-4.
Fig 5.
Combined retrospective analyses of 306 TNBC tumors treated with neoadjuvant chemotherapy stratified by molecular subtype.
Barplots show pCR for (A) all breast cancer stratified by non-TNBC and TNBC or (B) TNBC patients stratified by PAM50 or (C) TNBCtype-4. Dotted horizontal lines indicate pCR for the stratified individual cohort. Table shows distribution of pCR, residual disease (RD) and odds ratio (OR) for a pCR in TNBC patients stratified by (D) PAM50 or (E) TNBCtype-4. Forrest plots display OR for pCR in subtypes relative to all TNBC.