Methods of treatment based on molecular response to treatment

Abstract

A treatment method based on the biomolecular response of breast cancer to targeted therapy is provided. Histological evaluation of the expression levels of various biomolecules or infiltrating immune cells after initiation of human epidermal growth factor receptor 2 (HER2)-targeted therapy can be used to determine whether the breast cancer will achieve a pathological complete response. Based on the likelihood of a pathological complete response, the breast cancer can be treated accordingly. A method for diagnostically determining a pathological complete response in breast cancer includes: obtaining or obtaining an in-treatment cancer biopsy from an individual with breast cancer, the in-treatment cancer biopsy being a cancer biopsy obtained after initiation of targeted therapy; measuring or measuring the expression of one or more sets of biomolecules in at least one region of interest in the in-treatment cancer biopsy; and determining or determining whether the targeted therapy will provide a pathological complete response in the individual using a classifier and the biomolecule expression measurements.

Description

[Technical Field]

[0001] CROSS-REFERENCE TO RELATED APPLICATIONS This application claims priority to U.S. Provisional Application No. 62 / 927,557, entitled "Methods of Treatments Based Upon Molecular Response to Neoadjuvant Treatment," to Curtis et al., filed October 29, 2019, which is incorporated herein by reference in its entirety.

[0002] STATEMENT REGARDING FEDERALLY SPONSORED RESEARCH OR DEVELOPMENT This invention was made with government support under contract CA182514 awarded by the National Institutes of Health. The government has certain rights in this invention.

[0003] Technical Field The present disclosure is generally directed to methods, including diagnosis and treatment, based on the molecular characteristics of an individual's breast cancer and its molecular response to treatment. [Background technology]

[0004] background Human epidermal growth factor receptor 2-positive (HER2+) breast cancer is breast cancer that tests positive for a protein called human epidermal growth factor receptor 2 (HER2), which promotes cancer cell growth. HER2+ breast cancer accounts for 15–30% of invasive breast cancers and is associated with an aggressive phenotype. Several targeted therapies are available for HER2+ breast cancer, including trastuzumab (Herceptin), lapatinib (Tykerb), neratinib (Nerlynx), pertuzumab (Perjeta), and trastuzumab emtansine (T-DM1 or Kadcyla). Targeted therapies are often used as neoadjuvant treatments to reduce tumor size before surgery. Summary of the Invention [Means for solving the problem]

[0005] Abstract Various embodiments relate to the diagnosis and treatment of breast cancer based on molecular response to targeted treatment. In various embodiments, the molecular response of the cancer to targeted treatment is determined by measuring the expression of specific tumor-associated or immune-related biomolecules. In various embodiments, a linear model utilizes biomolecular expression to determine the likelihood of achieving a complete pathological response to the targeted treatment. In various embodiments, a specific treatment regimen is implemented based on the likelihood of achieving a complete pathological response. [Brief explanation of the drawings]

[0006] The description and claims will be more fully understood with reference to the following figures and data graphs, which are presented as exemplary embodiments of the invention and should not be construed as a complete recitation of the scope of the invention.

[0007] [Figure 1] FIG. 1 is a flow diagram of a method for treating breast cancer based on classification indicating pathological complete response (pCR), according to one embodiment of the present invention.

[0008] [Figure 2] 2 is a schematic diagram of the discovery and validation cohort analyzed with GeoMx™ Digital Spatial Profiling (DSP) technology utilized in accordance with various embodiments. Patients with invasive HER2+ breast cancer enrolled in the TRIO-US B07 clinical trial were treated with one cycle of their assigned HER2-targeted therapy followed by six cycles of their assigned HER2-targeted treatment plus chemotherapy (docetaxel + carboplatin). Tissue was obtained at three time points (pre-treatment, intra-treatment, and post-treatment / surgery).

[0009] [Figure 3]Figure 3 provides a summary of the clinical characteristics of the TRIO-US B07 DSP discovery cohort, including treatment arm, pathological complete response (pCR), estrogen receptor (ER) status, and inferred PAM50 status based on pre-treatment bulk expression data, as utilized according to various embodiments. Two-way contingency tables compare the distribution of ER status, pCR status, and treatment arm.

[0010] [Figure 4] Figure 4 provides a chart showing pathologically estimated cellularity pre- and during treatment for the discovery cohort, as utilized in accordance with various embodiments. Samples with green shading indicate those used in subsequent analysis. In the pathological complete response (pCR) column, 0 = non-pCR and 1 = pCR. For the estrogen receptor (ER) status column, 0 = ER-negative and 1 = ER-positive. Figure 4 also provides an example of an in situ area from case 30 sampled during treatment, as utilized in accordance with various embodiments. Cellularity was estimated to be 0 based on pathological review of clear tissue sections, but tumor areas were identified upon imaging of the tissue sections used in this analysis.

[0011] [Figure 5] Figure 5 is a schematic diagram summarizing the NanoString Digital Spatial Profiler workflow utilized in accordance with various embodiments. Slides are stained with a mixture of protein antibodies. The antibodies have indexing oligos bound to them, which are used for subsequent readout. ROIs (regions of interest) are selected and illuminated using UV (ultraviolet) light. The UV light cleaves the indexing oligos within the ROI for collection and per-probe quantification.

[0012] [Figure 6]Figure 6 provides a schematic diagram and images illustrating the regions of interest analyzed and utilized according to various embodiments. Multiple regions of interest (ROIs) per tissue sample were selected based on pan-cytokeratin (panCK-E) enrichment and subjected to spatial proteomic profiling of 40 tumor and immune markers. Protein counts were measured separately within the phenotypic region corresponding to the PanCK-E mask, which includes tumor cells and co-localized immune cells, and for the inverted mask corresponding to the panCK-negative region.

[0013] [Figure 7] Figure 7 provides sample images showing multiple regions of interest utilized in accordance with various embodiments. Location of spatially separated ROIs within tissue specimens for representative pCR cases (69) and empirical non-pCR cases (58). An average of four ROIs were profiled per tissue (range: 1-7).

[0014] [Figure 8] Figure 8 provides a correlation plot, generated in accordance with various embodiments, comparing percent Ki67 positivity (assessed using IHC) with normalized DSP Ki67 expression (averaged across all ROIs in separate tissue sections from the same case and time point). A total of 42 biopsies (24 pre-treatment and 18 intra-treatment) with paired Ki67 IHC and DSP data were utilized in this analysis. The Pearson correlation coefficient and corresponding p-value are also shown. Figure 8 also provides a boxplot, generated in accordance with various embodiments, comparing normalized DSP HER2 expression (averaged across all ROIs from the same case and time point) between cases that exhibited strong (3+) IHC HER2 staining (using separate tissue sections from the same case and time point) or weaker (0-2) IHC HER2 staining. A total of 44 biopsies (23 pre-treatment and 21 intra-treatment) with paired HER2 IHC and DSP data were utilized in this analysis. Significance was assessed using a Wilcoxon test.

[0015] [Figure 9] 9 provides pairwise correlations of pre-treatment protein marker expression across all ROIs in the discovery cohort, as utilized in accordance with various embodiments. Black boxes indicate probes within the same hierarchical cluster.

[0016] [Figure 10] 10 provides a chart showing inter- and intratumor variation in HER2 and CD45 protein expression in untreated HER2-positive breast tumors from a discovery cohort, corresponding to ROIs utilized according to various embodiments. Clinical characteristics, including pCR status, estrogen receptor (ER) status, and PAM50 subtype (based on gene expression profiling), are shown.

[0017] [Figure 11A] 11A and 11B provide violin plots showing CD45 and CD56 values ​​from Digital Spatial Profiling (DSP) protein data during treatment (FIG. 11A) and before treatment (FIG. 11B) in pCR versus non-pCR cases, as utilized in accordance with various embodiments. Each point represents the mean probe value for all panCK-enriched ROIs during treatment in that case. p-values ​​were derived using a linear mixed-effects model across multi-region data with blocking by patient. For each violin plot, the white box represents the interquartile range, and the black line extending from the white box represents 1.5 times the interquartile range. Analysis based on the discovery cohort. [Figure 11B]11A and 11B provide violin plots showing CD45 and CD56 values ​​from Digital Spatial Profiling (DSP) protein data during treatment (FIG. 11A) and before treatment (FIG. 11B) in pCR versus non-pCR cases, as utilized in accordance with various embodiments. Each point represents the mean probe value for all panCK-enriched ROIs during treatment in that case. p-values ​​were derived using a linear mixed-effects model across multi-region data with blocking by patient. For each violin plot, the white box represents the interquartile range, and the black line extending from the white box represents 1.5 times the interquartile range. Analysis based on the discovery cohort.

[0018] [Figure 12] 12 provides a volcano plot demonstrating treatment-related changes based on a comparison of pre-treatment versus on-treatment protein marker expression levels in the pancytokeratin-enriched (PanCK-E) region, as utilized in accordance with various embodiments. Significance, -log10 (FDR-adjusted p-value), is indicated along the y-axis.

[0019] [Figure 13] Figure 13 provides a volcano plot showing treatment-related changes based on a comparison of bulk RNA expression levels before versus during treatment, as utilized in accordance with various embodiments. RNA transcripts with corresponding Digital Spatial Profiling (DSP) protein markers were used in this analysis. Significance, -loglO (FDR-adjusted p-value), is indicated along the y-axis. Analysis based on the discovery cohort.

[0020] [Figure 14] FIG. 14 provides a table of protein antibody and gene name pairs used in the comparative analysis between DSP and bulk expression data utilized according to various embodiments.

[0021] [Figure 15]Figure 15 provides a volcano plot demonstrating treatment-related changes based on a comparison of pre- versus on-treatment protein marker expression levels in the pancytokeratin-enriched (PanCK-E) region in trastuzumab-treated cases (arms 1 and 3, n = 23). Significance-loglO (FDR-adjusted p-value) is shown along the y-axis, as utilized according to various embodiments. Analysis based on the discovery cohort.

[0022] [Figure 16A] 16A and 16B provide volcano plots showing treatment-related changes in pCR versus non-pCR cases, utilized according to various embodiments. [Figure 16B] 16A and 16B provide volcano plots showing treatment-related changes in pCR versus non-pCR cases, utilized according to various embodiments.

[0023] [Figure 17A] 17A and 17B provide pairwise correlations of protein markers in pCR versus non-pCR cases, as utilized according to various embodiments. Black boxes define hierarchical clusters. [Figure 17B] 17A and 17B provide pairwise correlations of protein markers in pCR versus non-pCR cases, as utilized according to various embodiments. Black boxes define hierarchical clusters.

[0024] [Figure 18] FIG. 18 provides a waterfall plot showing treatment-related changes (pre- to on-treatment) in ER+ and ER- cases based on protein expression, utilized according to various embodiments.

[0025] [Figure 19]Figure 19 provides a waterfall plot showing treatment-related changes (pre- to on-treatment) based on pancytokeratin-enriched (PanCK-E) regions from DSP protein expression data, as utilized in accordance with various embodiments. Input data was stratified by both estrogen receptor (ER) status and pathological complete response (pCR) outcome. Analysis based on discovery cohort.

[0026] [Figure 20] Figure 20 provides a waterfall plot showing treatment-related changes (pre- to on-treatment) in DSP protein expression in HER2-enriched and non-HER2-enriched cases (n=7 normal-like, n=2 luminal B, n=2 basal, n=1 luminal A), as utilized in accordance with various embodiments. Analysis was performed in the discovery cohort.

[0027] [Figure 21] 21 is a waterfall plot showing treatment-associated changes (pre- to on-treatment) based on pancytokeratin-enriched (PanCK-E) regions from DSP protein expression data, as utilized according to various embodiments. Samples were stratified by both PAM50 status (HER2-enriched or other) and pathological complete response (pCR) outcome.

[0028] [Figure 22] Figure 22 provides a waterfall plot generated using the pancytokeratin-enriched (PanCK-E) region from DSP protein expression data, showing treatment-related changes (pre- to on-treatment) when only one region is used to profile each sample (averaged over 100 replicates of a single region per time point) rather than the two to seven regions from each sample used in other analyses, as utilized in accordance with various embodiments. The top plot is for all patients, while the bottom plot is stratified by pathological complete response (pCR) status. Analysis based on discovery cohort.

[0029] [Figure 23]Figure 23 provides a volcano plot showing treatment-related changes from pre-treatment to surgery in tumors that did not undergo a pathological complete response (pCR) using DSP protein expression levels in pancytokeratin-enriched (PanCK-E) regions, as utilized in accordance with various embodiments. Significance, -loglO (FDR-adjusted p-value), is indicated along the y-axis. Analysis based on discovery cohort.

[0030] [Figure 24] Figure 24 provides representative in situ images of ROIs from two cases, as well as quantification of HER2 and CD45 protein levels (log2 normalized) in panCK-enriched regions, utilized according to various embodiments.

[0031] [Figure 25] FIG. 25 provides a chart showing pre- and on-treatment comparison of DSP HER2 protein levels for all regions profiled per case per time point, utilized in accordance with various embodiments.

[0032] [Figure 26] Figure 26 provides a table showing a comparison of mean squared error of DSP HER2 protein expression pre-treatment versus on-treatment within and between patients, as utilized in accordance with various embodiments. P values ​​are based on a two-tailed paired Wilcoxon signed-rank test. Analysis is based on the discovery cohort.

[0033] [Figure 27] Figure 27 provides a chart showing pre-treatment versus on-treatment heterogeneity for each DSP tumor and immune marker utilized according to various embodiments. P-values ​​are based on two-tailed paired Wilcoxon signed-rank test. Analysis is based on the discovery cohort.

[0034] [Figure 28]Figure 28 provides a chart showing pre-, mid-, and post-treatment heterogeneity of each DSP protein marker in non-pCR cases (patients with tumor cells present at the time of surgery), as utilized in accordance with various embodiments. Analysis based on discovery cohort.

[0035] [Figure 29] FIG. 29 is a chart showing during-treatment heterogeneity in DSP protein markers for pCR and non-pCR cases, utilized according to various embodiments.

[0036] [Figure 30] Figure 30 provides a chart showing pre-treatment heterogeneity of DSP protein marker expression in pCR and non-pCR cases, as utilized in accordance with various embodiments. Heterogeneity was calculated as within-patient mean squared error based on analysis of variance. P values ​​are based on a two-sided Wilcoxon matched-pairs signed-rank test. Analysis based on discovery cohort.

[0037] [Figure 31] Figure 31 provides a schematic diagram of digital spatial profiling (DSP) performed on multiple regions of interest (ROIs) per tissue sample, as utilized in accordance with various embodiments. Protein counts were measured separately within phenotypic regions corresponding to a panCK-enriched (tumor-enriched) mask containing tumor cells and co-localized immune cells, and for an inverted mask corresponding to a panCK-negative (tumor microenvironment, TME) region.

[0038] [Figure 32A] Figures 32A, 32B, and 32C provide waterfall plots of DSP protein data revealing differences in immune marker expression between immune-dense panCK-enriched regions and surrounding panCK-negative regions profiled before, during, and after treatment, as utilized according to various embodiments. Heterogeneity was calculated as within-patient mean squared error based on analysis of variance. P values ​​are based on a two-tailed pairwise Wilcoxon signed-rank test. Analysis is based on the discovery cohort. [Figure 32B] Figures 32A, 32B, and 32C provide waterfall plots of DSP protein data revealing differences in immune marker expression between immune-dense panCK-enriched regions and surrounding panCK-negative regions profiled before, during, and after treatment, as utilized according to various embodiments. Heterogeneity was calculated as within-patient mean squared error based on analysis of variance. P values ​​are based on a two-tailed pairwise Wilcoxon signed-rank test. Analysis is based on the discovery cohort. [Figure 32C] Figures 32A, 32B, and 32C provide waterfall plots of DSP protein data revealing differences in immune marker expression between immune-dense panCK-enriched regions and surrounding panCK-negative regions profiled before, during, and after treatment, as utilized according to various embodiments. Heterogeneity was calculated as within-patient mean squared error based on analysis of variance. P values ​​are based on a two-tailed pairwise Wilcoxon signed-rank test. Analysis is based on the discovery cohort.

[0039] [Figure 33A] Figures 33A and 33B provide waterfall plots generated using DSP protein data comparing immune marker expression between panCK-enriched regions and surrounding panCK-negative regions before and during treatment in pCR (n=14) and non-pCR (n=14) cases, as utilized in accordance with various embodiments. The correlation between fold-change values ​​of immune markers in pretreatment, pCR, and non-pCR cases was 0.98, indicating a similar immune distribution across panCK-enriched regions and the surrounding microenvironment regardless of pCR outcome, and this correlation remained high (0.95) during treatment. Analysis based on discovery cohort. [Figure 33B]Figures 33A and 33B provide waterfall plots generated using DSP protein data comparing immune marker expression between panCK-enriched regions and surrounding panCK-negative regions before and during treatment in pCR (n=14) and non-pCR (n=14) cases, as utilized in accordance with various embodiments. The correlation between fold-change values ​​of immune markers in pretreatment, pCR, and non-pCR cases was 0.98, indicating a similar immune distribution across panCK-enriched regions and the surrounding microenvironment regardless of pCR outcome, and this correlation remained high (0.95) during treatment. Analysis based on discovery cohort.

[0040] [Figure 34A] Figures 34A and 34B provide waterfall plots comparing immune marker expression between panCK-enriched regions and surrounding panCK-negative regions pre- and during treatment in ER-positive (n=14) and ER-negative (n=14) cases, generated using DSP protein data and utilized according to various embodiments. Analysis based on discovery cohort. [Figure 34B] Figures 34A and 34B provide waterfall plots comparing immune marker expression between panCK-enriched regions and surrounding panCK-negative regions pre- and during treatment in ER-positive (n=14) and ER-negative (n=14) cases, generated using DSP protein data and utilized according to various embodiments. Analysis based on discovery cohort.

[0041] [Figure 35]Figure 35 provides multiplex immunohistochemistry (mIHC) images showing the distribution of HER2, CD45, and CD8 signals before and during treatment of a representative tissue stamp utilized in accordance with various embodiments. The panCK mIHC channel (not shown) was used to generate panCK and tissue masks (outlined in yellow). IHC marker expression levels of HER2, CD45, and CD8 were quantified across the entire tissue section (across all digitized subimages) and within the panCK-enriched tumor region (across all digitized subimages).

[0042] [Figure 36] FIG. 36 provides an illustration of a panCK-enriched binary mask and marginal complexity-based quantification of the tumor-microenvironment boundary utilized in accordance with various embodiments.

[0043] [Figure 37] 37 provides a violin plot showing a comparison of pre-treatment marginal complexity values ​​between pCR and non-pCR cases, as utilized in accordance with various embodiments. P-values ​​calculated using a linear model and blocked by patient. Analysis is based on the discovery cohort.

[0044] [Figure 38] Figure 38 provides a violin plot showing a comparison of peripheral complexity values ​​before versus during treatment, as utilized in accordance with various embodiments. Peripheral complexity was quantified using a PanCK-enriched ROI. P-values ​​calculated using a linear model and blocked by patient. Analysis is based on the discovery cohort.

[0045] [Figure 39]Figure 39 provides a plot showing the Spearman correlation between DSP protein expression values ​​and peripheral complexity per region of interest (ROI) in pre-treatment and on-treatment tissue specimens from a discovery cohort, as utilized in accordance with various embodiments. Significantly correlated probes: p-values ​​< 0.05 are marked with an asterisk. Correlation plot of Ki-67, the marker with the highest correlation with peripheral complexity, where each dot represents an individual ROI.

[0046] [Figure 40] Figure 40 provides the area under the receiver operating characteristic (AUROC) performance of various models compared using nested cross-validation with Holm-Bonferroni correction for multiple hypotheses in a discovery (training) cohort generated according to various embodiments. Receiver operating characteristic (ROC) curves were generated using cases with DSP panCK enrichment data from both pre-treatment and on-treatment time points (n=23). ROC curves and statistical comparisons of L2-regularized classifiers trained using DSP protein marker means (averaged across ROIs) for pre-treatment, on-treatment, and a combined pre- and on-treatment ("on-treatment + pre-treatment") analysis.

[0047] [Figure 41] Figure 41 provides the area under the receiver operating characteristic (AUROC) performance of various models compared using nested cross-validation with Holm-Bonferroni correction for multiple hypotheses in a discovery (training) cohort generated according to various embodiments. Receiver operating characteristic (ROC) curves were generated using cases with DSP panCK enrichment data from both pre- and on-treatment time points (n=23). ROC curves and statistical comparisons of on-treatment versus pre-treatment trained DSP protein L2 regularized classifiers using mean values ​​of all markers, tumor markers and immune markers. This analysis used the cross-area mean marker values ​​from both pre- and on-treatment time points.

[0048] [Figure 42]Figure 42 provides the area under the receiver operating characteristic (AUROC) performance (using nested cross-validation with Holm-Bonferroni correction for multiple hypotheses) comparing DSP protein on-treatment plus pre-treatment L2 regularized classifiers trained using marker mean vs. standard error (SEM) for tumor and immune markers generated according to various embodiments. Model comparison was performed in the discovery cohort.

[0049] [Figure 43] Figure 43 provides the area under the receiver operating characteristic (AUROC) performance of various models compared using nested cross-validation with Holm-Bonferroni correction for multiple hypotheses in a discovery (training) cohort generated according to various embodiments. Receiver operating characteristic (ROC) curves were generated using cases with DSP panCK enrichment data from both pre- and on-treatment time points (n=23). ROC curves and statistical comparison of on-treatment plus pre-treatment DSP protein L2 regularized classifiers to a model trained using ER and PAM50 status. These two models were compared to a model incorporating on-treatment plus pre-treatment DSP protein data, ER, and PAM50 status.

[0050] [Figure 44]Figure 44 provides receiver operating characteristic (ROC) curves and AUROC (area under the receiver operating characteristic) quantification for the on-treatment plus pre-treatment DSP protein L2 regularized classifier using all 40 markers compared to other models generated according to various embodiments. Statistical comparison with a model trained using ROC and ER, PAM50 status, and strong (3+) HER2 IHC (immunohistochemistry) staining status, pre-treatment, n=19 patients, for whom all data was available. These two models are also compared to a model incorporating on-treatment plus pre-treatment DSP protein data, ER, PAM50 status, and HER2 IHC staining status. Statistical comparison with a model trained using ROC and ER, PAM50 status, and HER2 FISH (fluorescence in situ hybridization) ratio, pre-treatment, n=21 patients, for whom all data was available. These two models are also compared to a model incorporating on-treatment plus pre-treatment DSP protein data, ER, PAM50 status, and HER2 FISH ratio. Statistical comparisons were made with models trained using ROC and treatment-intermediate tumor-infiltrating lymphocytes (TILs) for n = 16 patients for whom all data was available. These two models were also compared with a model incorporating on-treatment plus pre-treatment DSP protein data and on-treatment TILs.

[0051] [Figure 45] Figure 45 provides an ROC and statistical comparison of on-treatment plus pre-treatment L2 regularized classifiers trained using bulk RNA expression using mean values ​​of DSP protein markers versus RNA transcripts corresponding to the DSP protein markers, generated according to various embodiments. ROC curves were generated using cases with DSP panCK enriched data and bulk expression data from both pre-treatment and on-treatment time points (n=21).

[0052] [Figure 46]Figure 46 provides a plot showing the Spearman correlation between DSP protein probes (averaged across all ROIs per case) and the corresponding bulk RNA transcripts before treatment for these markers utilized in accordance with embodiments. Significantly correlated probes (p-value < 0.05) are indicated with an asterisk. Two exemplary correlation plots are shown, with each dot representing a single case. Analysis based on the discovery cohort.

[0053] [Figure 47] Figure 47 provides a table summarizing clinical characteristics for the TRIO-US B07 Clinical Trial Digital Spatial Profiling (DSP) validation cohort used in model testing utilized according to various embodiments. Treatment arm, pathological complete response (pCR), estrogen receptor (ER) status, and inferred PAM50 status based on pre-treatment bulk expression data are included. A two-way contingency table compares the distribution of ER status, pCR status, and treatment arm.

[0054] [Figure 48] Figure 48 provides a volcano plot demonstrating treatment-related changes based on a comparison of pre-treatment versus on-treatment protein marker expression levels in the pancytokeratin-enriched (PanCK-E) region in a validation cohort, as utilized in accordance with various embodiments. Significance, -loglO (FDR-adjusted p-value), is indicated along the y-axis.

[0055] [Figure 49] Figure 49 provides a volcano plot showing treatment-related changes in the PanCK-E region versus non-pCR cases in the validation cohort, as utilized in accordance with various embodiments. Significance, -loglO (FDR-adjusted p-value), is shown along the y-axis.

[0056] [Figure 50]Figure 50 provides receiver operating characteristic (ROC) curves for the on-treatment plus pre-treatment DSP protein L2 regularized classifier in the discovery (training) cohort (n=23, assessed by cross-validation) and validation (testing) cohort (n=28, assessed by training test) using a 40-plex DSP protein marker panel generated according to various embodiments.

[0057] [Figure 51] Figure 51 provides a plot showing the coefficients for each of 40 markers in an L2 regularized in-plus-pre-treatment DSP protein model generated in accordance with various embodiments, trained in a discovery cohort, and tested in a validation cohort.

[0058] [Figure 52] Figure 52 provides a receiver operating characteristic (ROC) curve for the on-treatment plus pre-treatment DSP protein L2 regularized classifier in the discovery cohort using all cases with panCK enrichment data from both time points (n=23) and the subset of cases treated with trastuzumab or trastuzumab plus lapatinib (n=19), generated according to various embodiments. Model performance was assessed by cross-validation using the 40 DSP protein markers profiled in both cohorts.

[0059] [Figure 53] Figure 53 provides a correlation plot comparing marker coefficients for pre-treatment DSP proteins trained using all cases in the discovery cohort and only trastuzumab-treated cases (arms 1 and 3), generated according to various embodiments.

[0060] [Figure 54] Figure 54 provides a plot showing the coefficients of each marker in an L2 regularized on-plus-pre-treatment DSP protein model generated in accordance with various embodiments and trained using only cases treated with trastuzumab (arms 1 and 3).

[0061] [Figure 55] Figure 55 provides ROC curves for on-treatment plus pre-treatment L2 regularized classifiers in the discovery (training) cohort (n=23, assessed by cross-validation) and validation (testing) cohort (n=28, assessed by training test) using HER2 and CD45 from the DSP protein marker panel, generated according to various embodiments.

[0062] [Figure 56] FIG. 56 provides a chart showing the coefficients of each marker in an L2 regularized in-plus-pre-treatment DSP protein model trained using only CD45 and HER2, generated in accordance with various embodiments.

[0063] [Figure 57] Figure 57 provides ROC curves for on-treatment plus pre-treatment L2 regularized classifiers in the discovery (training) cohort (n=23, assessed by cross-validation) and validation (testing) cohort (n=28, assessed by training test) using CD45 from the DSP protein marker panel, generated according to various embodiments.

[0064] [Figure 58] FIG. 58 provides a chart showing the coefficients of each marker in an L2 regularized in-plus-before-treatment DSP protein model trained using only CD45, generated in accordance with various embodiments.

[0065] [Figure 59] Figure 59 provides ROC curves for on-treatment L1 regularized classifiers in the discovery (training) cohort (n=23, assessed by cross-validation) and validation (testing) cohort (n=28, assessed by training test) using CD45 from the DSP protein marker panel, generated according to various embodiments.

[0066] [Figure 60]Figure 60 provides a table of markers utilized in accordance with various embodiments, with signal-to-noise ratios (SNR) < 3 in the discovery cohort indicated by a caret (^), and markers with SNR < 3 in the validation cohort indicated by an asterisk (*). DETAILED DESCRIPTION OF THE INVENTION

[0067] Detailed Description Referring now to the figures and data, methods are provided for predicting pathological complete response (pCR) and treating HER2+ breast cancer based on a predicted pCR of the cancer. As understood in the art, pathological complete response is defined as the disappearance of all invasive cancer in breast tissue after completion of neoadjuvant chemotherapy. Many embodiments relate to evaluating one or more tumor biopsies of a patient diagnosed with breast cancer. In some embodiments, the individual is diagnosed with HER2+ breast cancer. In some embodiments, molecular evaluation of the tumor biopsy is performed before any treatment (i.e., pre-treatment). In some embodiments, molecular evaluation of the tumor biopsy is performed after initiation of targeted therapy (also referred to herein as a mid-treatment time point), which may occur during neoadjuvant treatment. In some embodiments, molecular evaluation of the tumor biopsy is performed shortly after initiation of targeted therapy (e.g., about 48 hours, 72 hours, 96 hours, 120 hours, 144 hours, or 168 hours after initiation). In some embodiments, molecular evaluation of the tumor biopsy is performed shortly after completion of the first cycle of targeted therapy (e.g., about 48 hours, 72 hours, 96 hours, 120 hours, 144 hours, or 168 hours after completion of the first cycle). In some embodiments, molecular evaluation of the tumor biopsy is performed both before any treatment and after initiation of targeted therapy. In some embodiments, biomolecule expression after initiation of targeted therapy is used to predict pCR. In some of these embodiments, changes in biomolecule expression that occur before any treatment and after one cycle of targeted therapy are used to predict pCR. In some embodiments, histological evaluation of immune infiltrate cells after initiation of targeted therapy is used to predict pCR.

[0068] According to several embodiments, treatment is determined by the likelihood of a response to neoadjuvant therapy to achieve pCR, which can be used to escalate or de-escalate treatment. In some embodiments, neoadjuvant therapy is used to reduce tumor size before subsequent treatment (e.g., surgery). In some embodiments, if neoadjuvant therapy is predicted to achieve pCR, de-escalated treatment is utilized, such as (for example) administering a targeted treatment directed at HER2 without systemic chemotherapy (i.e., non-targeted chemotherapy). Targeted treatments include (but are not limited to) trastuzumab, lapatinib, pertuzumab, T-DM1, and any combination thereof. In some embodiments, targeted chemotherapeutic agents are used (e.g., trastuzumab emtansine (T-DM1)). In many embodiments, if neoadjuvant therapy is not predicted to result in pCR, an escalated treatment regimen can be administered, such as (for example) targeted treatment with chemotherapy and / or dual targeted therapy, including in the neoadjuvant and / or adjuvant setting. Chemotherapeutic agents include (but are not limited to) taxanes, including paclitaxel (taxol), anthracyclines, including doxorubicin (adriamycin), cyclophosphamide, and any combination thereof.

[0069] Based on recent discoveries, the association between the expression of specific tumor and immune biomolecules and pCR after the initiation of targeted therapy is now recognized, indicating the course of treatment and monitoring. Thus, embodiments relate to classifying breast cancers based on the likelihood of achieving pCR with targeted treatment in order to determine a treatment regimen that is well suited to the breast cancer.

[0070] Treatment of breast cancer determined by molecular response Some embodiments relate to classifying breast cancer by the likelihood of pCR after targeted treatment (particularly neoadjuvant targeted treatment). In some embodiments, breast cancer classification is based on biomolecule expression in tumor biopsies determined after initiation of targeted treatment. According to some embodiments, a specific biomolecule expression pattern indicates whether a breast cancer has a high likelihood of achieving pCR. In some embodiments, breast cancer classification is based on histological evaluation of immune infiltrating cells after initiation of targeted therapy. In some embodiments, evaluation of biomolecule expression and / or immune infiltrating cells is determined before and after initiation of targeted treatment so that changes in expression and / or changes in immune cell infiltration can be determined. Based on the classification of pCR likelihood, some embodiments determine a course of treatment for breast cancer.

[0071] FIG. 1 provides a method for classifying an individual's breast cancer based on biomolecule expression and / or immune cell infiltration after initiation of targeted therapy, which indicates the likelihood of pCR and therefore the cancer is treated accordingly. In some embodiments, the breast cancer is HER2+. Process 100 begins with measuring 101 the expression of several biomolecules and / or assessing immune cell infiltration of the breast cancer after initiation of targeted treatment. In some embodiments, a breast cancer biopsy is utilized for biomolecule expression and / or immune cell infiltration analysis. In some embodiments, the biomolecule expression and / or immune cell infiltration analysis is performed in a specific region of interest in the biopsy. In some embodiments, the biomolecule expression and / or immune cell infiltration analysis is performed in an area where tumor cells interact with infiltrating immune cells. In some embodiments, the biomolecule expression and / or immune cell infiltration analysis is performed in an area with pan-cytokeratin-positive (panCK+) tumor cells, which indicates infiltrating immune cells directly interacting with tumor cells. In some embodiments, biomolecule expression and / or immune cell infiltration analysis is performed in areas with pan-leukocyte marker CD45-positive (CD45+) immune cells.

[0072] In some embodiments, biomolecule expression and / or immune cell infiltration are determined after the initiation of a targeted treatment. It is advantageous to determine biomolecule expression and / or immune cell infiltration during initial treatment so that an appropriate course of treatment can be determined and administered. In various embodiments, biomolecule expression and / or immune cell infiltration are determined after the initiation of treatment but before the completion of one cycle, after one cycle of treatment and before the second cycle of treatment, after at least one cycle of treatment and before the third cycle of treatment, after at least one cycle of treatment and before the fourth cycle of treatment, or any combination thereof. In some embodiments, biomolecule expression and / or immune cell infiltration are determined before any targeted treatment. According to several embodiments, when biomolecule expression and / or immune cell infiltration are determined at multiple time points, the dynamics of biomolecule expression can be determined. For example, in some embodiments, the change in biomolecule expression and / or immune cell infiltration from pre-treatment to after the first cycle of treatment is measured. In some embodiments, a linear mixed-effects model is utilized to quantify the dynamics of biomolecule expression from pre-treatment to after the first cycle of treatment. As mentioned above, targeted treatments include (but are not limited to) trastuzumab, lapatinib, pertuzumab, T-DM1, and any combination thereof.

[0073] As described herein, it is now understood that several biomolecules provide indicators of whether a breast cancer is likely to achieve pCR. Generally, biomolecules related to HER2 signaling and immune activation can be detected and measured. Based on recent findings, it has been found that measurement of the following HER2 signaling pathway biomolecules (RNA or protein): HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, and p-STAT3, provides an indicator of whether a breast cancer will achieve pCR (after a full course of neoadjuvant treatment). Similarly, measurement of biomolecules expressed in epithelial tumor tissues has been found to provide an indicator of whether a breast cancer will achieve pCR (after a full course of neoadjuvant treatment): PanCK, Ki67, and beta-catenin. Generally, a decrease in HER2 signaling pathway biomolecules indicates pCR. Similarly, measurement of the following immune response and activation biomolecules (RNA or protein): CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, and VISTA have been found to provide an indication of whether a breast cancer achieves pCR (after a full course of neoadjuvant therapy). Generally, an increase in immune response and activation biomolecules indicates pCR. Furthermore, measurement of the following cell survival biomolecules (RNA or protein): beta-2 microglobulin and Bcl-2 have been found to provide an indication of whether a breast cancer achieves pCR (after a full course of neoadjuvant therapy). It should be understood that other biomolecular measurements can be performed that provide an indication of whether a breast cancer achieves pCR.

[0074] It has been found that many biomolecules, including HER2, Ki67, pS6, CD45, CD56, STING, VISTA and CD66B, contribute significantly to the prediction of pCR status. Thus, in some embodiments, at least one or more of HER2, Ki67, pS6, CD45, CD56, STING, VISTA and CD66B biomolecule expression measurements are determined to predict pCR status.

[0075] It is further understood that immune cell infiltration into tumor tissue also provides an indication of whether a breast cancer is likely to achieve pCR. Generally, lymphocytes and other immune cells can be assessed by histology or immunostaining techniques. In some embodiments, cancer biopsies can be stained with hematoxylin and eosin (H&E) to count infiltrating immune cells. In some embodiments, H&E-stained cancer biopsies can be evaluated to quantify the infiltration of stromal tumor-infiltrating lymphocytes (sTILs) or intratumoral lymphocytes (iTu-Lys). In some embodiments, cancer biopsies can be evaluated by immunostaining with anti-CD45 antibodies and / or anti-CD56 antibodies to determine the number of infiltrating lymphocytes. Immunostaining can be performed in a number of ways, including, but not limited to, chromogenic immunohistochemistry (IHC), immunofluorescence, or elemental isotope staining (e.g., antibody-labeled elemental isotopes).

[0076] In many embodiments, biomolecule expression measurements and / or immune cell infiltration assessments are performed in at least one region of the tumor biopsy. In some embodiments, biomolecule expression measurements and / or immune cell infiltration assessments are performed in at least two regions of the tumor biopsy, and the measurements are combined in an appropriate manner (e.g., sum, mean, median, standard error, standard deviation, weighting). The region of interest within the tumor biopsy for biomolecule expression measurements can be determined by any appropriate method. In some embodiments, the region of interest is determined by tumor cell identification, infiltrating immune cell identification, or a combination thereof. In some embodiments, the region of interest is determined by panCK+ expression. In some embodiments, the region of interest is determined by CD45+ expression.

[0077] As shown, process 100 also utilizes the biomolecule expression measurements and / or infiltrating immune cell data as inputs to a classifier model to classify breast cancers as likely or unlikely to have a pCR after targeted treatment (103). Any suitable classifier capable of utilizing the biomolecule expression measurements and / or infiltrating immune cell data to provide a classification of pCR can be utilized. In some embodiments, the classifier is a regression model. Regression models include (but are not limited to) linear, logistic, polynomial, ridge, stepwise, LASSO, elastic net, L1 regularization, L2 regularization, and any combination thereof. In various embodiments, the classifier is one of a generalized linear model (GLM), ordinary least squares, random forest, decision tree, or neural network. The model can be trained using a collection of individuals who have biomolecule and / or infiltrating immune cell data measured at one or more time points to determine pCR after a course of treatment (particularly neoadjuvant treatment). Thus, in various embodiments, a collection of individuals with breast cancer (e.g., HER2+) who have been assessed for biomolecular and / or infiltrating immune cell data measured from tumor biopsies at baseline and / or after initiation of targeted treatment can be utilized to train a model to predict pCR. In some embodiments, the classifier model is trained to determine whether an individual should receive de-escalated treatment. In some embodiments, the classifier model is trained to determine whether an individual should receive escalated treatment.

[0078] In some embodiments, a collection of individuals with breast cancer who have been assessed for biomolecular and / or infiltrating immune cell data measured from tumor biopsies at baseline and after initiation of targeted treatment can be used to train a model that predicts pCR using dynamic measurements. As detailed in the accompanying manuscript, both static biomolecular expression measurements and / or infiltrating immune cell data after initiation of targeted treatment and dynamic biomolecular expression measurements from baseline to after initiation of targeted treatment each provide significant prediction of pCR and can be utilized as features in the regression model. Additional features can be utilized in the regression model, including, but not limited to, treatment type, ER status, PAM50 status, tumor size, tumor grade, cancer stage, patient age, and patient ethnicity.

[0079] In some embodiments, a classifier model can be trained to classify pCR based on a set of one or more biomolecule expression measurements and / or infiltrating immune cell data, including (but not limited to) expression levels and / or infiltration data for a single region, mean expression across multiple regions, total expression across multiple regions, median expression across multiple regions, standard error expression across multiple regions, and standard deviation expression across multiple regions. In various embodiments, the classifier model utilizes HER2 signaling pathway biomolecules, epithelial tumor biomolecules, immune response and activation biomolecules, cell survival biomolecules, infiltrating immune cell data, or a combination thereof. Thus, the classifier model can utilize a set of measurements of one or more of the following biomolecules: HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, p-STAT3, PanCK, Ki67, beta-catenin, CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, VISTA, beta-2 microglobulin, and Bcl-2. Similarly, the model can utilize infiltrating immune cell data determined by H&E staining or immunostaining.

[0080] In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of HER2. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of Ki67. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of pS6. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of CD45. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of CD56. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of STING. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of VISTA. In some embodiments, the classifier model utilizes a set of one or more biomolecule expression measurements that include expression of CD66B.

[0081] In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and Ki67. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and pS6. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and CD45. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and CD56. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and STING. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and VISTA. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of HER2 and CD66B.

[0082] In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and HER2. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and Ki67. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and pS6. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and CD56. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and STING. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and VISTA. In some embodiments, the classifier model utilizes a set of two or more biomolecule expression measurements comprising expression of CD45 and CD66B.

[0083] In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and Ki67. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and pS6. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and CD56. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and STING. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and VISTA. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements including expression of HER2, CD45, and CD66B.

[0084] In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and HER2. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and Ki67. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and pS6. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and STING. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and VISTA. In some embodiments, the classifier model utilizes a set of three or more biomolecule expression measurements comprising expression of CD45, CD56, and CD66B.

[0085] In some embodiments, the classifier model utilizes quantification of infiltrating immune cells. In some embodiments, the classifier model utilizes quantification of sTILs. In some embodiments, the classifier model utilizes iTu-Ly infiltration grade scores. In some embodiments, the classifier model utilizes quantification of CD45+ cells. In some embodiments, the classifier model utilizes quantification of CD56+ cells.

[0086] In some embodiments, the sensitivity, specificity, and area under the curve (AUC) metrics of the classifier can be modified to achieve desired performance. In some cases, higher specificity may be desired to ensure robust classification of individuals to ensure each individual is treated appropriately. In some cases, higher sensitivity is desired so that the detection limit is lower and the number of missed true positive results is reduced. Thus, in various embodiments, the specificity is set to about 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%, or thereabouts. Also, in various embodiments, the sensitivity is set to about 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%, or thereabouts.

[0087] Based on the cancer classification, HER2+ breast cancer is treated accordingly. 105 In some embodiments, when pCR is demonstrated, a de-escalated treatment regimen is administered, such as (for example) a HER2-targeted treatment regimen without systemic chemotherapy (i.e., non-targeted chemotherapy). Targeted treatments include (but are not limited to) trastuzumab, lapatinib, pertuzumab, T-DM1, and any combination thereof. In some embodiments, a targeted chemotherapeutic agent is used (e.g., T-DM1). In many embodiments, if pCR is not demonstrated, an escalated treatment regimen is administered, such as (for example) targeted treatment with a chemotherapy regimen or a dual targeted treatment regimen (i.e., two targeted therapeutic agents). Chemotherapeutic agents include (but are not limited to) taxanes, including paclitaxel (taxol), anthracyclines, including doxorubicin (adriamycin), cyclophosphamide, and any combination thereof.

[0088] Although specific examples of processes for molecularly classifying and treating breast cancer have been described above, those skilled in the art will understand that, according to some embodiments of the present invention, various steps of the process can be performed in different orders and certain steps may be optional. Thus, it will be apparent that various steps of the process can be used as appropriate for the requirements of a particular application. Furthermore, any of a variety of processes for molecularly classifying and treating suitable for the requirements of a given application can be utilized in accordance with various embodiments.

[0089] Methods for measuring biomolecule expression As will be appreciated by those skilled in the art, biomolecule expression can be detected and measured by several methods according to various embodiments. In some embodiments, breast cancer tumors are biopsied or surgically removed from patients, fixed, and prepared for biomolecule expression detection and measurement. Any suitable fixation method can be used, including, but not limited to, formaldehyde, formalin-fixed paraffin-embedded (FFPE), methanol, ethanol, OCT embedding, and flash freezing.

[0090] It has been found that detecting and measuring biomolecules in tumor regions of interest can provide better pCR prediction than bulk RNA profiling of tumors. Thus, in some embodiments, regions of interest or specific cell types are identified and used in biomolecule detection and measurement techniques. In some embodiments, tissues are treated and / or stained with antibodies to allow identification of the region of interest using microscopy, which allows detection and measurement of biomolecules directly on the region of interest. In some embodiments, the region of interest is identified by panCK+ tumor cells. In some embodiments, the region of interest is identified by CD45+ immune cells. In some embodiments, multiplexed spatial tissue analysis is performed to determine biomolecule expression. In some embodiments, live or fixed tissues are treated with antibodies and / or stained to allow identification and isolation of cell types by flow cytometry, which uses isolated cells to extract biomolecules for detection and measurement.

[0091] In many embodiments, multiplexed spatial tissue analysis is used to detect protein and / or RNA expression in a region of interest in fixed tissue. In some embodiments, protein and RNA expression are simultaneously assessed in a region of interest in fixed tissue. There are numerous methodologies and kits for performing multiplexed spatial tissue analysis, including, but not limited to, NanoString's GeoMx™ Digital Spatial Profiler (DSP) (Seattle, WA), Akoya Biosciences' CODEX (Menlo Park, CA), Akoya Biosciences' Vectra Polaris, Harvard Program in Therapeutic Science's Cyclic Immunofluorescence (CyCIF) (Boston, MA), IonPath's Multiplexed Ion Bea Imaging (MIBI) (Menlo Park, CA), Akoya Biosciences Opal kit, Roche-Ventana's DISCOVERY system (Oro Valley, AZ), and Genotipix-HistoRx's Automated Quantitative Analysis (AQUA) (New Haven, CT). Generally, these systems are capable of detecting multiple biomolecules within a region of interest. Further reviews of protocols for analyzing tissues can be found in E.R. Parra J. Cancer Treat. Diagn. 2018, 2, 43-53, and E.R. Parra, A. F. Fancisco-Cruz, and I.W. Wistuba Cancers (Basel). 2019, 11, E247, the disclosures of each of which are incorporated herein by reference.

[0092] One example of multiplexed spatial tissue analysis is NanoString's GeoMx™ Digital Spatial Profiler (DSP), which can detect RNA or peptide expression in selected regions using a panel of oligos for RNA expression and / or antibodies for peptide expression. Details of this machine and how it can be used are described in the exemplary embodiments. Generally, an identified region of interest (e.g., a panCK+ region) is selected, and a panel of antibodies and / or probes is incubated with the region of interest to bind and identify the biomolecule of interest. After incubation, excess unbound reagents are then washed away. Each antibody and probe in the panel has an attached oligo tail that serves as a barcode. The oligo tail barcodes are releasable by UV irradiation. After biomolecule attachment and washing, UV light releases the barcodes, which are then detected and measured using the NanoString nCounter, thereby determining the relative concentration of the biomolecule of interest (normalized to a control).

[0093] In embodiments where biomolecule extraction is required, several methods for extracting biomolecules from biological sources are well known. Generally, biomolecules are extracted from cells or tissues and then prepared for further analysis. Alternatively, biomolecules can be observed within cells, which are typically fixed and prepared for further analysis. The decision to extract biomolecules or fix tissue for direct examination depends on the assay being performed. Generally, in situ hybridization and histology samples are performed on fixed tissue, while nucleic acid amplification techniques (e.g., sequencing) and protein quantification techniques (e.g., ELISA) utilize extracted biomolecules.

[0094] In some embodiments, the cells utilized to examine the biomolecules are neoplastic cells and / or infiltrating immune cells of breast cancer, which can be directly extracted or analyzed from a biopsy. In some embodiments, solid tumor biopsies, such as (for example) primary, nodal, and / or distant tumors, are utilized. In some embodiments, the region of interest is determined by detecting tumor cells (e.g., pancytokeratin-positive (panCK+) tumor cells), infiltrating immune cells (e.g., CD45-positive (CD45+)), or a combination thereof. It should be understood that any suitable means or biomarkers for identifying the region of interest or isolating specific cell types can be utilized according to various embodiments.

[0095] Several assays are known for determining biomolecular expression in biological samples, including but not limited to hybridization techniques, nucleic acid amplification techniques, sequencing, antibody detection, and mass spectrometry. Several hybridization techniques can be used, including but not limited to in situ hybridization, microarrays (e.g., Affymetrix, Santa Clara, CA), and NanoString nCounter (Seattle, WA). Similarly, several nucleic acid amplification techniques can be used, including but not limited to PCR and RT-PCR. Additionally, several sequencing techniques can be used, including but not limited to genome sequencing of tumor tissue, exome sequencing, targeted gene sequencing, Sanger sequencing, and RNA-seq. Several antibody techniques can be used, including but not limited to in situ histology / immunohistochemistry, immunofluorescence staining and periodic immunofluorescence staining, ELISA, and Western blot.

[0096] As is understood in the art, to obtain a positive detection, it may be necessary to detect only a portion of a genomic locus, gene, or peptide. In many hybridization techniques, detection probes are typically 10 to 50 bases long, although the exact length depends on the assay conditions and the assay developer's preferences. In many amplification techniques, amplicons are often 50 to 1,000 bases long, again depending on the assay conditions and the assay developer's preferences. In many sequencing techniques, genomic loci and transcripts are identified with sequence reads of 10 to several hundred bases long, again depending on the assay conditions and the assay developer's preferences. In many antibody techniques, monoclonal or polyclonal antibodies can be used. In some embodiments, hybridization, targeted sequencing, and antibody detection techniques target the sequences of several genes of interest, such as those that provide an indication of pCR in breast cancer.

[0097] It should be understood that slight variations in gene sequences and / or assay tools (e.g., hybridization probes, amplification primers) may exist, but are expected to provide similar results in detection assays. These minor variations include (but are not limited to) insertions, deletions, single nucleotide polymorphisms, and other variations depending on the assay design. In some embodiments, the detection assay can detect genomic loci and transcripts with high but not complete homology (e.g., 70%, 80%, 90%, 95%, or 99% homology). In some embodiments, the detection assay can detect genomic loci and transcripts with an altered, deleted, or inserted 1 base pair, an altered, deleted, or inserted 2 base pairs, an altered, deleted, or inserted 3 base pairs, an altered, deleted, or inserted 4 base pairs, an altered, deleted, or inserted 5 base pairs, or an altered, deleted, or inserted more than 5 base pairs. As is understood in the art, the longer the nucleic acid polymer used for hybridization, the less homology is needed for hybridization to occur.

[0098] It should also be understood that some gene transcripts have several isoforms that are expressed. As is understood in the art, it will be understood that many alternative isoforms will provide a similar indication of molecular classification and, therefore, metastatic potential. Thus, in some embodiments, alternative isoforms of a gene transcript are encompassed.

[0099] Assessment of infiltrating immune cells Infiltrating immune cells can be detected and evaluated by several methods according to various embodiments, as will be understood by those skilled in the art. In some embodiments, breast cancer tumors are biopsied or surgically removed from patients, fixed, and prepared for the detection and evaluation of immune cell infiltration. Any suitable fixation method can be used, including, but not limited to, formaldehyde, formalin-fixed paraffin-embedded (FFPE), methanol, ethanol, OCT embedding, and flash freezing.

[0100] It has been found that detecting and evaluating infiltrating immune cells in tumor regions of interest can provide robust prediction of pCR. Thus, in some embodiments, regions of interest or specific cell types are identified and used in infiltrating immune cell detection and evaluation techniques. In some embodiments, tissues are treated with antibodies and / or stained to identify regions of interest by microscopy, allowing detection and evaluation of infiltrating immune cells to be performed directly on the regions of interest. In some embodiments, the regions of interest are identified by panCK+ tumor cells. In some embodiments, the regions of interest are identified by CD45+ immune cells.

[0101] In many embodiments, histological analysis is performed by histological staining and / or immunostaining. In some embodiments, cancer biopsies can be stained with hematoxylin and eosin (H&E) to count infiltrating immune cells. In some embodiments, H&E-stained cancer biopsies are evaluated to quantify the infiltration of stromal tumor-infiltrating lymphocytes (sTILs) or intratumoral lymphocytes (iTu-Lys). Typically, sTILs are quantified as a score of 0 to 100%, determined by the percentage of sTILs among all cells in the region of interest. iTu-Lys are typically scored by a semi-quantitative infiltration grade (0 to 3). In some embodiments, cancer biopsies can be evaluated by immunostaining with anti-CD45 antibodies and / or anti-CD56 antibodies to determine the number of infiltrating lymphocytes. Immunostaining can be performed in a number of ways, including, but not limited to, chromogenic immunohistochemistry (IHC), immunofluorescence, or elemental isotope staining (e.g., antibody-labeled elemental isotopes). Infiltrating lymphocytes can be quantified in several ways, typically as a percentage. In some embodiments, infiltrating lymphocytes are quantified as a percentage of total cells in the region of interest. In some embodiments, infiltrating lymphocytes are quantified as a percentage of total lymphocytes (e.g., the number of lymphocytes in the tumor tissue divided by the total number of lymphocytes in the tumor and surrounding tissue). In some embodiments, infiltrating lymphocytes are quantified by area (e.g., mm 2 ) is quantified as counts per 1000 cells / well. In various embodiments, histological analysis is performed by a pathologist and / or an automated image analyzer. For details of histological analysis of infiltrating immune cells, see R. Salgado et al., Ann Oncol. 2015 26(2):259-71 and C. Denkert et al., Mod Pathol. 2016 Oct;29(10):1155-64, the disclosures of which are incorporated herein by reference.

[0102] kit In some embodiments, the kit is used to determine whether a breast cancer is likely to achieve pCR after targeted treatment. The kit can be used to detect the expression of biomarkers in the region of interest of a biopsy described herein. For example, the kit can be used to detect one or more of the genetic biomarkers described herein, which can be used to determine the likelihood of pCR. The kit can include one or more agents for determining biomolecular expression, one or more agents for assessing immune cell infiltration, a container for collecting a biological sample (e.g., a biopsy) obtained from a subject, appropriate means for fixing and preparing the biological sample (e.g., reagents and materials for FFPE), and reagents for identifying the region of interest, as well as printed instructions for reacting the agent with the biological sample to detect the expression of biomarker genes derived from the sample. The agents may be packaged in separate containers. The kit can further include one or more control reference samples and reagents for performing biochemical assays, enzyme assays, immunoassays, hybridization assays, or sequencing assays.

[0103] In some embodiments, the kit is used to detect and measure biomolecules of interest. Nucleic acid detection kits, according to various embodiments, include a set of hybridizable complementary sequences and / or amplification primers specific to a set of genomic loci and / or expressed transcripts. In some examples, the kit will include additional reagents sufficient to facilitate detection and / or quantification of the set of genomic loci and / or expressed transcripts. In some examples, the kit will be capable of detecting and / or quantifying the expression of at least 5, 10, 15, 20, 25, 30, 40, or 50 biomolecules. In some examples, the kit will be capable of detecting and / or quantifying the expression of thousands or more biomolecules using sequencing techniques.

[0104] In some embodiments, the set of hybridizable complementary sequences is immobilized on an array, such as those designed by Affymetrix or Illumina. In many embodiments, the set of hybridizable complementary sequences is linked to a "barcode" to facilitate detection of the hybridized species, and hybridization is provided so that it can be performed in solution, such as in those designed by NanoString. In some embodiments, a set of primers (and optionally probes) to facilitate amplification and detection of the amplified species is provided so that PCR can be performed in solution, such as in those designed by Applied Biosystems of ThermoScientific (Foster City, CA).

[0105] The kit can include one or more containers for the compositions contained in the kit.The compositions can be in liquid form or can be lyophilized.Suitable containers for the compositions include, for example, bottles, vials, syringes, and test tubes.The containers can be made of various materials, including glass or plastic.The kit can also include a package insert containing written instructions for the method of determining biomolecule expression in tumor biopsies. Indications and Treatments for HER2+ Breast Cancer

[0106] Various embodiments are directed to breast cancer diagnosis and treatment based on an indicator of whether a cancer is likely to achieve pCR after targeted treatment, particularly short-term targeted treatment. As described herein, prognostic procedures can utilize biopsy regions of interest to detect and determine biomolecular expression and / or immune cell activation, particularly biomolecules related to HER2+ signaling and immune response and activation. Using biomolecular expression and / or immune cell activation and a trained classifier, breast cancers are classified as likely to achieve pCR, but unlikely to achieve pCR with targeted treatment alone. Based on the likelihood of achieving pCR, individuals can be administered appropriate treatments.

[0107] Diagnostic Indicators and Treatment Some embodiments are directed to obtaining a diagnostic indicator of how to treat breast cancer after initiation of targeted treatment. In some embodiments, a cancer biopsy is extracted after initiation of targeted treatment from an individual with breast cancer, and the biopsy is further analyzed.

[0108] In some embodiments, diagnostic instructions can be administered to a breast cancer patient as follows: a) At least one cycle of targeted treatment b) Extract a biopsy c) determining the static and / or dynamic expression of a set of one or more biomarkers d) Determine whether targeted treatment alone can provide pCR and determine the appropriate treatment strategy

[0109] In some embodiments, once an indication is obtained that the breast cancer can achieve pCR with the targeted treatment, de-escalated treatment is administered. In some embodiments, if pCR is indicated for the breast cancer, targeted treatment is administered without systemic chemotherapy (i.e., non-targeted chemotherapy), particularly in the neoadjuvant setting. In some embodiments, the breast cancer is HER2+ and the targeted treatment targets HER2. In some embodiments, a targeted chemotherapeutic agent is used (e.g., T-DM1). Targeted HER2 treatments include (but are not limited to) trastuzumab, lapatinib, pertuzumab, T-DM1, and any combination thereof. In many embodiments, if pCR is not indicated for the breast cancer, escalated treatment is administered in the neoadjuvant and / or adjuvant setting. In some embodiments, if pCR is not indicated, targeted treatment with chemotherapy is administered. In some embodiments, if pCR is not indicated, dual targeted treatment with chemotherapy is administered. Chemotherapeutic agents include (but are not limited to) taxanes, including paclitaxel (taxol), anthracyclines, including doxorubicin (adriamycin), cyclophosphamide, and any combination thereof.

[0110] In some embodiments, a diagnosis is determined based on a threshold. In some embodiments, the threshold is determined by the sensitivity, specificity, and / or area under the curve (AUC) metrics of the classifier. In some cases, a threshold with higher specificity may be desired to ensure robust classification of individuals to ensure that each individual is treated appropriately. For example, when classifying an individual as likely to achieve pCR, high specificity may be desirable. If an individual is likely to achieve pCR but is instead incorrectly classified as unable to achieve pCR from neoadjuvant treatment, the treatment regimen may require more severe chemotherapy and / or be prolonged, and therefore the individual would have been better off receiving targeted treatment with chemotherapy first. In some cases, higher sensitivity is desired so that the detection limit is lower and the number of missed true positive results is reduced. Thus, in various embodiments, specificity is set at about 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%, or a value therebetween. Also, in various embodiments, the sensitivity is set at about 60%, 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%, or any value therebetween.

[0111] Specific treatment regimens are also contemplated. In some embodiments, when pCR is demonstrated in HER2+ breast cancer, the following therapeutic combinations are administered in the treatment regimen: Trastuzumab and lapatinib Trastuzumab and pertuzumab T-DM1 alone T-DM1 and pertuzumab

[0112] In some embodiments, if pCR is not demonstrated for HER2+ breast cancer, the following therapeutic combinations are administered in the treatment regimen: Trastuzumab, pertuzumab, and chemotherapy drugs T-DM1 and pertuzumab, followed by weekly paclitaxel, doxorubicin, and cyclophosphamide Trastuzumab, pertuzumab, and taxanes T-DM1, pertuzumab, and anthracyclines (e.g., doxorubicin)

[0113] It is understood that the particular therapeutic combinations should not be considered limiting, and that several combinations of targeted therapeutics and / or chemotherapeutic agents may be administered.

[0114] As will be appreciated by those skilled in the art, dosing regimens and therapeutic regimens may be administered appropriately for the breast cancer being treated. For example, the following dosages may be utilized in treatment cycles according to various embodiments: Pertuzumab: 840 mg IV infusion over 60 minutes, followed by 420 mg IV infusion over 30 to 60 minutes+ Trastuzumab: Initially, 8 mg / kg IV infusion over 90 minutes, then 6 mg / kg IV infusion over 30 to 90 minutes+ Paclitaxel: 175 mg / m over 3 hours 2 IV infusion of Doxorubicin: 60 mg / m 2 IV infusion of

[0115] In some embodiments, the drug is administered in a therapeutically effective amount as part of a course of treatment.In this context, "treating" means improving at least one symptom of the disorder being treated or providing beneficial physiological effects.For example, one such symptom improvement may be reducing tumor size and / or achieving pCR.

[0116] A therapeutically effective amount can be an amount sufficient to prevent, reduce, ameliorate, or eliminate symptoms of breast cancer. In some embodiments, a therapeutically effective amount is an amount sufficient to reduce the growth and / or metastasis of breast cancer. In some embodiments, a therapeutically effective amount is an amount sufficient to achieve pCR. [Example]

[0117] Illustrative Embodiments Embodiments of the present invention will be better understood with the aid of several examples provided therein. Numerous exemplary results of the process of identifying a combined molecular signature for colorectal cancer are described. Validation results are also provided.

[0118] Example 1: Spatial proteomic characterization of HER2-positive breast tumors predicts response to neoadjuvant treatment Human epidermal growth factor receptor 2 (HER2)-positive breast cancer accounts for 15–30% of invasive breast cancers and is associated with an aggressive phenotype. The addition of HER2-targeted agents to neoadjuvant chemotherapy has dramatically improved pathological complete response (pCR) rates in early-stage HER2-positive breast cancer, yet 40–50% of patients have residual disease after treatment. Conversely, HER2 inhibition with two targeted agents and without chemotherapy can result in pCR, suggesting that it may be possible to eliminate chemotherapy in a subset of patients. Identifying biomarkers that provide an indication of response to HER2-targeted therapy may help delineate neoadjuvant treatment regimens.

[0119] Bulk gene expression profiling of pretreatment samples has identified tumor characteristics (HER2-enriched intrinsic subtypes, HER2 expression levels, and ESR1 expression levels) and microenvironmental characteristics (increased immune infiltration) associated with response to HER2-targeted therapy in the neoadjuvant setting. Because tumor cells are simultaneously profiled with both colocalized and distant stromal and immune cells, bulk expression profiling is an imperfect tool for analyzing tumor and microenvironmental changes across treatment. In particular, it is difficult to assign observed changes to specific geographic or phenotypic cell populations within the complex tumor ecosystem in which malignant tumor cells interact with fibroblasts, endothelial cells, and immune cells. Furthermore, immune cells can be further divided into those that infiltrate the tumor core and those that are eliminated. Currently, how the tumor and immune microenvironment change during treatment is poorly understood, necessitating multiplexed in situ profiling of longitudinal tissue samples.

[0120] GeoMx™ Digital Spatial Profiling (DSP, NanoString) technology was used to assay archival tissue from an initial discovery set of 28 HER2-positive breast cancer patients enrolled in the neoadjuvant TRIO-US B07 clinical trial (S. Hurvitz et al., medRxiv 2020.09.16.20194324 (2020), the disclosure of which is incorporated herein by reference) whose tumors were sampled pre-treatment, 14-21 days after HER2-targeted therapy consisting of lapatinib, trastuzumab, or both (on-treatment), and at the time of surgery after completion of combination chemotherapy with HER2-targeted therapy (post-treatment). Results were subsequently validated in an independent validation set of 29 patients from the B07 clinical trial. Importantly, the neoadjuvant setting allows for early assessment of treatment response, and pCR is a strong surrogate for long-term survival in HER2-positive disease (J. Huober et al., Eur J Cancer 118, 169-177 (2019); K R. Broglio et al., JAMA Oncol 2, 751-760 (2016); and P. Cortazar et al., Lancet 384, 164-172 (2014); the disclosures of which are incorporated herein by reference). DSP allows for geographic and phenotypic selection of tissue regions for multiplexed proteomic characterization of cancer signaling pathways and the tumor-colocalized immune microenvironment (M Toki et al., Cancer Research 77, 3810 (2017); and CRMerritt et al., Nat Biotechnol 38, 586-599 (2020); the disclosures of which are incorporated herein by reference). In particular, spatial heterogeneity was characterized in alterations of cancer signaling pathways and microenvironment composition in untreated breast tumors and matched on-treatment biopsies and post-treatment surgical samples by profiling 40 tumor and immune proteins across multiple pancytokeratin (panCK)-enriched regions per sample. Protein expression during treatment changed dramatically in tumors that went on to achieve pCR and classifiers based on these data, which robustly predicted treatment response in the validation cohort.This novel spatial proteomic biomarker outperformed established predictors such as PAM50 subtypes and classifiers based on transcriptomic data in this cohort, suggesting a new avenue for personalizing treatment in early-stage HER2-positive breast cancer.

[0121] result Spatial proteomic analysis of untreated HER2-positive breast tumors Participants in the TRIO-US B07 clinical trial (NCT00769470) in early-stage HER2-positive breast cancer received one cycle of neoadjuvant HER2-targeted therapy containing either trastuzumab, lapatinib, or both agents, followed by six cycles of their assigned HER2-targeted therapy plus docetaxel and carboplatin given every three weeks (S. Hurvitz et al., (2020), supra). Pre- and intra-treatment core biopsies were obtained 14–21 days after HER2-targeted therapy, and surgical resection specimens were obtained post-treatment (Figure 2). Initially, the discovery cohort included 28 patients for whom FFPE samples were available from all three time points (pre-treatment, intra-treatment, and surgery). The cohort was balanced for both pCR and ER status (Figures 3 and 4) and used for all exploratory analyses. For evaluation of model performance, a validation cohort of 29 patients from the TRIO-US B07 cohort with matched pre- and on-treatment FFPE samples was utilized.

[0122] DSP enables multiplexed proteomic profiling of formalin-fixed, paraffin-embedded (FFPE) tissue sections (Figure 5), allowing regions of interest (ROIs) to be selected based on both geographic and phenotypic characteristics. Using a panCK enrichment strategy, cancer cells and colocalized immune cells were profiled across an average of four regions per tissue specimen (Figure 6). CD45, panCK, and dsDNA were selected immunofluorescent markers for visualization, spatially separated regions (Figure 7), and UV-irradiation-dominated masks for protein quantification were generated based on panCK immunofluorescence. A total of 40 tumor and immune proteins were profiled using DSP, and proteins evaluated using both DSP and orthogonal techniques showed strong concordance (Figures 6 and 8). Paired pre- and intra-treatment bulk gene expression data from the same patients were utilized to infer PAM50 subtypes, enabling comparison with spatially resolved DSP data.

[0123] In untreated tumors, correlations between immune markers were significant, suggesting the coordinated action of multiple immune cell subpopulations (Figure 9). HER2 pathway members and other downstream cancer signaling markers were also highly correlated, but correlations between tumor and immune markers were minimal for most marker pairs. Inter- and intratumor variability at the proteomic level was evident before treatment, including HER2 and the pan-leukocyte marker CD45 (Figure 10). When all ROIs per patient were averaged to derive a composite score per marker, baseline HER2 levels were found to be similar in tumors that achieved pCR versus those that did not, as were CD45 levels (mean pCR cases: 14.50, mean non-pCR cases: 15.00) (mean pCR: 9.90, mean non-pCR: 9.64). Using a linear mixed-effects model with blocking by patient (Methods), individual DSP protein markers, including HER2 and CD45, were further found to not be significantly different between pretreatment pCR versus non-pCR cases (unadjusted p>0.10 for all markers).

[0124] Decreased cancer signaling and increased immune infiltration after short-term HER2-targeted therapy We used DSP to examine treatment-related changes in both breast tumor and immune markers during short-term HER2-targeted therapy by profiling on-treatment biopsies (after a single cycle of HER2-targeted therapy alone) in the discovery cohort. The protein markers most associated with pCR during treatment were CD45 (unadjusted p=0.0024) and the natural killer (NK) cell marker CD56 (unadjusted p=0.0055) (Figures 11A and 11B). The fold changes in protein levels during treatment compared to pretreatment were quantified using a linear mixed-effects model with block by patient, and the significance (false-discovery-adjusted p-values) of all markers compared to their fold changes in a volcano plot were visualized. These analyses revealed dramatic decreases in HER2 and Ki67, along with other downstream pathway members, including pAKT, AKT, pERK, S6, and pS6, with phosphorylated proteins decreasing relatively more (Figure 12). Immune markers, including CD45 and CD8, markers of cytotoxic T cells, showed the greatest increases in expression with treatment. Notably, increased expression of CD8+ T cells was similarly observed in the TRIO-US B07 transcriptome data by cell-type deconvolution; however, given the lack of a control arm that underwent repeat biopsies without intervening treatment, it is unclear whether the observed immune changes are related to HER2-targeted therapy or repeat biopsies. More generally, bulk transcriptome data at treatment versus pretreatment mirrored the changes seen at the protein level, although the fold changes were attenuated (Figure 13). For example, using genes corresponding to DSP protein markers (Figure 14), we found that expression of HER2, AKT, Ki67, and breast cancer-associated keratin genes (KRT7, KRT18, KRT19) significantly decreased with treatment, while immune markers increased. Despite the use of different analytes, measurements, and tissue sections, DSP protein and bulk RNA datasets consistently demonstrated a decrease in HER2 signaling and breast cancer-associated markers accompanied by an increase in immune cell infiltration during neoadjuvant treatment.Given that lapatinib was associated with lower pCR rates in the TRIO-US B07 trial, we further evaluated changes during treatment in trastuzumab-treated cases (arms 1 and 3, n=23), and observed a similar pattern to the full cohort (Figure 15).

[0125] Next, we examined how treatment-related changes differed based on tumor sensitivity to HER2-targeted therapy and stratified tumors based on the achievement of pCR after neoadjuvant treatment (Figures 16A and 16B). In pCR cases, numerous immune markers increased with treatment, whereas in non-pCR cases, no significant treatment-related immune changes were observed, with modest decreases in Ki67 and HER2 signaling. These patterns could also be visualized by pairwise comparison of protein marker correlations, which revealed a stronger negative correlation between immune marker clusters and cancer cell marker clusters in tumors that achieved pCR (mean fold change across all markers in pCR cases: -0.231, non-pCR cases: -0.075, two-tailed Wilcoxon rank sum test p<2.2e-16) (Figures 17A and 17B).

[0126] Because both ER status and HER2-enriched subtype are associated with response to neoadjuvant therapy, we analyzed protein marker expression changes by these covariates. ER-negative tumors showed more significant changes during treatment (compared to pretreatment) compared with ER-positive tumors (mean absolute fold change ER-negative cases: 0.59, mean ER-positive cases: 0.36, two-sided Wilcoxon rank-sum test p=0.0045, Figure 18). However, when tumors were stratified by outcome, pCR cases showed more significant changes than non-pCR cases, regardless of ER status (Figure 19), and ER status did not predict pCR in this cohort (p=0.47). Similarly, tumors classified as HER2-enriched before treatment showed significant changes in tumor and immune markers in biopsies during treatment compared with other subtypes (Figures 20 and 21). For example, CD8+ T cells significantly increased with treatment in HER2-enriched cases but slightly decreased in other cases. Similar to the entire TRIO-US B07 transcriptome cohort, HER2-enriched subtype did not predict pCR (p=0.87).

[0127] To evaluate the utility of multi-region sampling, we measured changes on treatment versus before treatment using a single randomly selected region per tissue sample averaged over 100 simulations (Figure 22). Consistent with findings based on all tumor regions, CD45 and CD8 showed the greatest increases during treatment, while HER2 and pS6 showed the greatest decreases in single-region analysis. Although the magnitude of marker fold change with treatment was greater in pCR cases than in non-pCR cases (mean absolute fold change across all markers in pCR cases: 0.87 vs. non-pCR cases: 0.33, two-tailed Wilcoxon rank sum test p=1.02e-07), individual markers did not significantly increase with treatment in single-region analysis, reflecting increased variance.

[0128] To elucidate the biology associated with the combination of HER2-targeted therapy and chemotherapy, we also examined treatment-related changes in patients with residual tumor cells present at the time of surgery (non-pCR cases). While non-pCR cases showed limited changes during treatment, by the time of surgery, there was a substantial decrease in HER2 and the downstream AKT signaling pathway, as well as a concomitant increase in immune markers in panCK-enriched areas (Figure 23). Notably, HER2 decreased more significantly than its downstream pathway members, which may reflect compensatory pathway activation contributing to resistance. Although several immune markers significantly increased in non-pCR cases at the time of surgery (n=8), the fold change was reduced compared to pCR cases sampled at the time of treatment (mean fold change in non-pCR cases after treatment: 0.30; mean fold change in pCR cases during treatment: 0.85; two-tailed Wilcoxon rank-sum test p=0.0021, Figures 16A and 16B). Among immune markers that increased at the time of surgery in non-pCR cases, CD56 was the most significant, potentially related to the role of NK cells in identifying and killing chemotherapy-stressed tumor cells. NK cells were also found to increase at the time of surgery in the TRIO-US B07 bulk expression data.

[0129] Increased heterogeneity of tumor and immune markers during HER2-targeted therapy Given that tumor heterogeneity is a defining feature of HER2-positive breast cancer, we examined the variation in HER2 protein expression within different regions of breast tumor biopsies across patients throughout neoadjuvant treatment. As shown for two illustrative cases (Figure 24), HER2 protein levels across different geographic regions within each tissue sample demonstrated relatively consistent HER2 protein levels before treatment in the majority of cases (Figure 25). Much greater heterogeneity in HER2 protein expression was observed during treatment, both between regions and between patients (Figure 26). Such regional heterogeneity may reflect vasculature, tissue architecture, immune infiltration, or pharmacokinetic differences due to the biopsy itself, highlighting the importance of profiling multiple regions per sample throughout treatment.

[0130] We also investigated regional heterogeneity across both tumor and immune protein markers during treatment. For each marker and each time point, regional heterogeneity across the cohort was calculated as the within-patient mean squared error based on ANOVA (Methods). Across all markers, DSP protein heterogeneity significantly increased during treatment compared to pretreatment, similar to that noted for HER2 (Figure 27). These changes were extensive, with greater heterogeneity for all tumor and immune markers during treatment compared to pretreatment. Probes with the greatest heterogeneity included both tumor markers (HER2, pS6) and immune markers (CD3, CD8). Among tumors that did not achieve pCR, we assessed heterogeneity throughout the course of neoadjuvant treatment. Heterogeneity between tumor markers was not significantly different during treatment compared to pretreatment (two-tailed Wilcoxon rank sum test, p = 0.52) but increased at the time of surgery (posttreatment), whereas immune marker heterogeneity increased during treatment and then decreased at surgery (Figure 28). Tumors that achieved pCR showed higher protein heterogeneity among tumor markers (including HER2) during treatment, but not among immune markers (Figure 29). Higher immune marker heterogeneity during treatment in non-pCR cases may reflect a less consistent immune response, with some regions experiencing greater immune influx than others. Higher pre-treatment HER2 heterogeneity was not observed in non-pCR cases compared with pCR cases. Also, comparable regional heterogeneity among tumor markers was observed in pCR and non-pCR cases (Figure 30).

[0131] We further analyzed the DSP data to examine the immune cell composition in the panCK-enriched region (used in other analyses) compared with the surrounding panCK-negative region, which was designed to capture the adjacent microenvironment (Figure 31). Before treatment, both T cell (CD3, CD4, CD8) and macrophage (CD68) markers were more prevalent in the surrounding microenvironment, whereas CD56-positive NK cells and immunosuppressive markers (e.g., VCTN1, PD-L1, IDO) were more prevalent in the panCK-enriched region (Figure 32A-C). These findings are consistent with T cell elimination, and IDO and PD-L1 are thought to impair intratumoral proliferation of effector T cells. A similar immune profile was observed during HER2-targeted therapy alone. However, after treatment, in non-pCR cases with residual tumor, most immune markers, including CD8 and CD68, were more prevalent in the panCK-enriched region compared with the adjacent microenvironment. Both before and during treatment, immune cell localization was similar in patients who achieved and did not achieve pCR (Figures 33A and 33B), and in ER-positive versus ER-negative cases (Figures 34A and 34B).

[0132] As a preliminary proof of principle, noting that other multiplexed imaging techniques can similarly be used to profile panCK-enriched tumors, we also used multiplexed immunohistochemistry (mIHC) with panCK enrichment to profile tissue samples from patients who achieved and did not achieve pCR. Using a panCK antibody, we defined masked regions and quantified several markers that significantly changed with DSP-based treatment, namely HER2, CD45, and CD8, across tissue sections and within the panCK-enriched region (Figure 35). As expected, changes in protein expression signal were attenuated when whole tissue sections were compared with panCK-enriched regions. These data further support the notion that panCK enrichment can be useful for defining tumor and colocalized immune changes in breast and other tumors.

[0133] The geospatial distribution of tumor cells and immune cells is associated with recurrence and survival in multiple tumor types. Here, the relationship between treatment and the tumor-microenvironment boundary was investigated using perimeter complexity, which is proportional to the perimeter of a region divided by its area (Methods, Figure 36). No significant difference in perimeter complexity was observed between pCR and non-pCR cases before treatment (p=0.299, Figure 37). However, perimeter complexity significantly decreased during treatment compared with before treatment (p=1.32e-6, Figure 38). These data suggest that treatment may affect the geographic distribution of tumor cells as well as tumor cell content. Indeed, the proliferation marker Ki67 was highly correlated with perimeter complexity (Figure 39). Thus, in highly proliferative tumors, the perimeter of the tumor-microenvironment boundary is relatively large, potentially increasing crosstalk with the surrounding microenvironment.

[0134] DSP of paired and pre- and intra-treatment biopsies reveals features associated with pCR Given the dramatic difference in treatment-related changes in pCR cases compared with non-pCR cases (Figures 16A and 16B), we next sought to evaluate whether the status of the DSP protein marker before treatment or early in the course of treatment could be used to predict pCR. Using L2-regularized logistic regression, we classified tumors by pCR status based on the mean DSP protein expression levels across multiple ROI-profiled pretreatment, on-treatment, or both pretreatment and on-treatment (denoted "on-+pre-treatment") and evaluated model performance by nested cross-validation within the discovery cohort (Methods). Tumors with data for both time points were utilized in this analysis (n = 23 cases, Figure 4). The model based on on-treatment protein expression outperformed the model based on pretreatment protein expression (mean AUROC = 0.728 vs. 0.614) and performed comparably to a model incorporating both on-treatment and pretreatment protein expression levels (mean AUROC = 0.733) (Figure 4C). Classifiers trained using both immune and tumor markers outperformed models using tumor markers alone, highlighting the utility of simultaneous tumor and immune profiling to predict treatment response (Figure 41).

[0135] For the on-treatment plus pre-treatment classifiers of DSP proteins, we investigated the importance of multi-region sampling and heterogeneity by extending the model to incorporate both the mean marker expression across all regions and the standard error of the mean (SEM) for each marker across regions (Methods). This analysis was limited to patients with at least three regions profiled at both time points (n=16, Methods). We found that utilizing mean immune values ​​and SEM for tumor markers outperformed models based on mean values ​​for both tumor and immune markers (Figure 42), suggesting that classifiers that capture heterogeneity between tumor markers may improve pCR prediction.

[0136] We compared the performance of the DSP protein on-treatment plus pre-treatment classifier with features previously associated with outcome (ER status and PAM50 subtype), and again evaluated the model by cross-validation in the discovery cohort. Notably, models based on ER status and HER2-enriched PAM50 status performed poorly in this cohort (mean AUROC = 0.589), and adding these two features or additional pathological features to the DSP protein on-treatment plus pre-treatment dataset did not improve the AUROC (Figures 43 and 44). Given the availability of bulk transcriptome data for these cases, we built a model using paired on-treatment and pre-treatment bulk RNA expression data for 37 markers overlapping with the DSP protein panel. This model also performed significantly worse than the one based on DSP protein data (Figure 45, p < 0.0001 by cross-validation). This is not surprising, as only 16 of the 37 overlapping DSP and bulk RNA expression markers were positively correlated pre-treatment (Figure 46). If protein expression is a more proximal readout of cellular phenotype, various factors, including panCK enrichment, RNA transience / degradation, and post-translational regulation, may contribute to the lack of a strong correlation between protein and RNA expression levels.

[0137] DSP predicts pCR in an independent validation cohort In light of these promising findings, we further sought to evaluate the performance of the DSP protein on-treatment plus pre-treatment classifier in an independent cohort of patients (n=29) from the TRIO B07 clinical trial (Figure 47). As with the discovery cohort, an average of four panCK-positive regions were profiled from each tumor tissue pre- and post-treatment, utilizing the same panel of 40 protein antibodies. Marker changes during treatment compared to pre-treatment were similar to those observed in the discovery cohort (Figure 48): T cell markers (CD3, CD4, and CD8) increased, while tumor markers HER2 and Ki67 showed the most significant decreases. As with the discovery cohort, in the validation cohort, the differences in protein expression during and before treatment were more dramatic in tumors that ultimately achieved pCR (Figure 49). The AUROC performance of the L2-regularized logistic regression model trained on the discovery cohort was evaluated in the validation (test) cohort. The performance of the on-treatment plus pre-treatment DSP protein model in predicting pCR was relatively high in the discovery (assessed by cross-validation, average AUROC = 0.733) and validation (assessed by training and testing, AUROC = 0.725) cohorts (Figure 50). In the pre-treatment classifier trained in the discovery cohort and tested in the validation cohort, the marker with the largest L2 regularization coefficient was on-treatment CD45 protein levels. In general, features with large coefficients included on-treatment markers representing tumor-infiltrating lymphocyte and macrophage populations (CD45, CD44, CD66B) (Figure 51). On-treatment HER2 protein expression had a negative coefficient in the model, consistent with poor outcomes associated with high on-treatment HER2 levels.

[0138] Given the widespread use of trastuzumab in the current neoadjuvant treatment paradigm, model performance was further evaluated for the best-performing on-treatment + pre-treatment DSP protein model in trastuzumab-containing cases (arms 1 and 3, n=19). Similar model performance and marker coefficients were observed in the full discovery cohort and in the subset of cases in the validation cohort treated with trastuzumab (Figures 52-54). Validation of these findings in independent cohorts demonstrates the potential of multiplex spatial proteomic profiling to predict which patients will respond early to HER2-targeted therapy so that subsequent treatment can be tailored accordingly.

[0139] Two markers of biological significance, CD45 and HER2, were selected to evaluate model performance using a reduced marker set. Again, an L2-regularized logistic regression model was trained on the discovery cohort and evaluated on the validation (test) cohort. The performance of the on-treatment plus pre-treatment DSP HER2, CD45 model in predicting pCR was high in both the discovery (assessed by cross-validation, average AUROC = 0.809) and validation (assessed by training test, AUROC = 0.754) cohorts (Figure 55). As with the entire marker panel, on-treatment CD45 had the largest coefficient (Figure 56).

[0140] Finally, a model based solely on CD45 was constructed and its performance evaluated. L2 regularization was trained on the discovery cohort and evaluated on the validation (test) cohort. The performance of the on-treatment plus pre-treatment CD45 model in predicting pCR was high in both the discovery (assessed by cross-validation, average AUROC = 0.866) and validation (assessed by training test, AUROC = 0.749) cohorts (Figure 57). Again, on-treatment CD45 had the largest coefficient (Figure 58).

[0141] A CD45-based L1 regularized model was also trained on the discovery cohort and evaluated on the validation (test) cohort. The performance of the on-treatment CD45 model in predicting pCR was high in both the discovery (assessed by cross-validation, average AUROC=0.920) and validation (assessed by training test, AUROC=0.749) cohorts (Figure 59).

[0142] Consideration Bulk genomic and transcriptomic profiling has been central to recent cancer biomarker discovery efforts. However, admixture between heterogeneous cell populations complicates the analysis of such data, a problem exacerbated when studying longitudinal samples, where changes in cell population composition and localization may reflect the biology of disease progression or mechanisms of treatment response. Indeed, efforts to establish validated biomarkers of response to HER2-targeted therapy based on bulk genomic and transcriptomic profiling have met with limited success to date in other study cohorts and in TRIO-US B07. It was reasoned that in situ proteomic profiling of the treatment-induced tumor immune microenvironment would circumvent the limitations of dissociation techniques and improve our ability to uncover features associated with response to neoadjuvant HER2-targeted therapy. Here, we used DSP technology to collectively profile 40 tumor and immune markers on single 5-µm sections of archival tissue from breast tumors sampled before, during, and after neoadjuvant HER2-targeted therapy in the TRIO-US B07 clinical trial. To enhance signal while accounting for intratumoral heterogeneity, a pan-CK masking strategy was used to enrich tumor cells and co-localized immune cells across multiple regions per sample.

[0143] DSP of longitudinal breast biopsies from this study cohort revealed treatment-related changes, including a marked decrease in HER2 and downstream AKT signaling during treatment, accompanied by increases in CD45 and CD8 expression, consistent with infiltrating leukocytes and cytotoxic T cells, respectively. By the time of surgery, after a full course of neoadjuvant treatment, the tumor-immune composition had changed considerably, with increased CD56 expression in non-pCR cases, potentially reflecting NK cell-mediated killing of chemotherapy-stressed tumor cells. Changes in both tumor and immune markers during treatment were more dramatic in tumors that went on to achieve pCR in an independent validation cohort; crucially, both on-treatment and pre-treatment protein expression robustly predicted response (AUROC = 0.725). While on-treatment protein expression levels also predicted pCR, future studies may suggest that profiling of on-treatment tissue alone may be sufficient to predict subsequent pCR. In this cohort, neither pre-treatment protein expression, established predictive features, nor bulk pre- and on-treatment gene expression data were predictive, highlighting the superiority of this novel multiplexed spatial proteomic biomarker and its potential utility for patient stratification. These findings therefore address a significant unmet clinical need, given the considerable focus on identifying subsets of patients who should undergo treatment escalation, for example, by combining HER2-targeted agents, or who should safely de-escalate their treatment, for example, by shortening or omitting chemotherapy and its associated toxicities. Numerous biomarkers are being explored to help guide personalized targeting of escalation versus de-escalation approaches in early-stage HER2+ breast cancer, including imaging, circulating tumor DNA, and pretreatment immune scores or intrinsic subtypes; however, there are currently no validated biomarkers that can guide patient stratification. The increasing number of HER2-targeted therapy options, including novel, highly effective but potentially toxic agents, combined with the large heterogeneity in response, makes HER2+ breast cancer an ideal setting for the development of optimal personalized treatments over the next decade.

[0144] More generally, the results demonstrate the feasibility and capability of multiplexed in situ proteomic analysis of archival tissue samples to provide a proximal readout of tumor and immune cell signaling induced by therapy. Many signaling proteins / phosphoproteins, including those profiled here, are considered bottlenecks in protein networks and integrate mutational and transcriptional changes, making this a particularly powerful approach for studying treatment-associated changes. Importantly, DSP antibody panels can be customized to include additional / alternative markers of interest, such as ER or other tumor-specific markers and signaling pathways. This study also highlights study design considerations, including tumor cell panCK enrichment (or other markers to enrich specific cell populations) and the value of multiregional profiling to capture local tumor heterogeneity and treatment-associated changes, a concept that should be broadly applicable to other epithelial tumor types. This, along with the quantitative and multiplexed nature of DSP, represents a notable difference compared to classical IHC. Notably, DSP measurements in this study were based on regional analysis of defined cell populations consisting of approximately 300–600 cells. Although single-cell segmentation was not required for the development of the novel biomarkers described herein (and indeed may complicate the clinical implementation of such an approach), such data can further enable the identification of cellular states and cell-cell interactions, which, facilitated by advances in segmentation and throughput of spatial profiling techniques, is likely to be an area of ​​future research.

[0145] method Cohort Selection The TRIO-US B07 clinical trial was a randomized, multicenter study involving 130 women with stage I–III unilateral HER2-positive breast cancer (S. Hurvitz et al., 2020, supra). The University of California, Los Angeles (UCLA) IRB approved the TRIO-US B07 clinical trial (ID: 08-10-035). The Stanford IRB approved the use of TRIO-US B07 clinical trial specimens for correlation studies in the Curtis Lab (eProtocol #32180). Informed consent was obtained from all participants, including consent from patients to share their samples with other researchers. Enrolled patients were randomly assigned to one of three treatment arms, which determined the type of targeted therapy they would receive: trastuzumab, lapatinib, or a combination of trastuzumab and lapatinib. Breast tumor biopsies were obtained pre-treatment and after 14–21 days of assigned HER2-targeted therapy (without chemotherapy), followed by six cycles of assigned HER2-targeted treatment plus docetaxel and carboplatin administered every three weeks, followed by surgery. For each time point, core biopsies or surgical tissue sections were obtained and stored as either fresh-frozen or FFPE material. In total, 28 cases with available FFPE samples from all three time points (pre-treatment, intra-treatment, and at the time of surgery) were selected for inclusion in the discovery cohort based on sample availability and quality, balanced by pCR status and ER status (Figures 2 and 3). To evaluate the performance of the classifier, an additional 29 cases with available FFPE samples pre- and intra-treatment were selected for the validation cohort (Figures 47–49). Notably, the validation cohort was used only to evaluate model performance; all other analyses were based on the discovery cohort. Tumor cellularity was assessed by a board-certified breast pathologist (GRB) using tumor sections stained with hematoxylin and eosin. Samples with a cellularity of 0 were omitted from further analysis. For other tissue sections estimated to have a cellularity of 0, tumor cells were identified in the FFPE sections used to perform DSP (different from the H&E sections used for pathological review) and these were included in the analysis (Figure 4).FFPE blocks were sectioned at 5 μm thickness and stored at 4°C for less than 3 weeks before DSP experiments.

[0146] DSP Data Generation and Analysis Digital Spatial Profiling (DSP, NanoString FOR RESEARCH USE ONLY. Not for use in diagnostic procedures) was performed as previously described (CR Merritt et al., (2020), supra). Briefly, tissue slides were stained with a multiplex panel of protein antibodies containing photocleavable indexing oligos to enable subsequent readout (Figure 5). Regions of interest (ROIs) were selected on the DSP prototype instrument and illuminated using UV light. The indexing oligos released from each ROI were collected and deposited into designated wells on a microtiter plate, enabling well indexing of each ROI during nCounter readout (direct protein hybridization). Custom masks were generated using an ImageJ pipeline as previously described (RNAmaria et al., Nat Med 24, 1649-1654 (2018), the disclosure of which is incorporated herein by reference). For each tissue sample, counts for each marker were obtained from the average of four (range 1-7) panCK-enriched (panCK-E) ROIs. Raw protein counts for each marker in each ROI were generated using nCounter (VA Malkov et al., BMC Res Note 2, 80 (2009), the disclosure of which is incorporated herein by reference). Approximate counts were normalized to ERCC (based on the geometric mean of three positive control markers). Histone H3 was used as a housekeeping marker, and ROIs with extreme histone H3 (more than three standard deviations from the mean) were selected (less than 1% of the ROI). Background noise was calculated using the geometric mean of two IgG antibodies, and markers with a signal-to-noise ratio <3x were noted (Figure 60). Immune markers were normalized based on ROI area to measure the total density of immune content within the region. Tumor markers were normalized using a housekeeping antibody (histone H3) to capture the status of cancer signaling pathways on a cell-by-cell basis. As a further quality control, area and housekeeping normalization factors were compared for each ROI, and ROIs were filtered with different normalization factors across the two methods (which represented 6% of all ROIs).All normalized numbers were transformed into log2 space for downstream analysis. The analyses performed in this study are comparative in nature (e.g., pre-treatment vs. on-treatment, pCR vs. non-pCR) and robust to variations in normalization methods.

[0147] Bulk mRNA expression analysis RNA was extracted using the RNeasy Mini Kit (Qiagen) and quantified using a Nanodrop One spectrophotometer (ThermoFisher Scientific). RNA samples were labeled with cyanine 5-CTP or cyanine 3-CTP (Perkin Elmer) using the Quick AMP Labeling Kit (Agilent Technologies). Gene expression microarray experiments were performed by comparing each baseline sample with a sample taken 14 to 21 days after HER2-targeted therapy (on-treatment). Each on-treatment sample was compared with a pre-treatment sample from the same patient. Limma (MERitchie et al., Nucleic Acids Research 43, e47 (2015); and MERitchie et al., Bioinformatics 23, 2700-2707 (2007); the disclosures of which are incorporated herein by reference) was used for background correction ("normexp"), intra-array normalization ("loess"), inter-array normalization, and averaging of replicate probes. For downstream analyses, including batch correction and comparison with the DSP cohort, normalized counts were transformed into log2 space. To remove potential batch effects related to the date of microarray run, Combat (W.E. Johnson et al., Biostatics 8, 118-127 (2007), the disclosure of which is incorporated herein by reference) was used. Pre- and on-treatment PAM50 status was inferred using AIMS (Absolute Intrinsic Molecular Subtyping), a N-of-1 algorithm robust to variations in dataset composition (E.R. Paquet et al., Breast Cancer Res 19, 32 (2017), the disclosure of which is incorporated herein by reference). Given the expected predominance of HER2-enriched cases in this cohort, this approach was utilized.

[0148] Correlation analysis Spearman rank correlations between DSP protein data and bulk RNA data were calculated for pretreatment samples using the average of all DSP ROIs per patient (both panCK-enriched ROIs and surrounding microenvironment-enriched ROIs). Plots showing correlations between protein markers (Figures 9, 17A, and 17B) were overlaid with hierarchical clustering in the form of black squares. Differences in the distribution of correlation values ​​in pCR vs. non-pCR cases were assessed using a two-tailed Wilcoxon two-sample t-test.

[0149] comparative analysis For comparative analysis of DSP protein data in which multiple regions were sampled per patient (e.g., pre-treatment vs. on-treatment, pCR vs. non-pCR, panCK-enriched vs. panCK-negative), we utilized a linear mixed-effects model with blocking by patient (D. Bates et al., J Stat Softw 67, 1-48 (2015), the disclosure of which is incorporated herein by reference). This model allows for comparison of marker levels in a patient-tailored manner while controlling for differences in the number of ROIs profiled per patient. The coefficient of a fixed effect is the variation attributable to that variable (x-axis of the volcano plot), and the p-value used to calculate the false discovery rate (y-axis of the volcano plot) is based on the t-value (a measure of the magnitude of the difference relative to the variability of the sample data). The false discovery rate (FDR) was calculated using the Benjamini & Hochberg procedure (Y. Benjamini and Y. Hochberg, JR Stat Soc B 57, 289-300 (1995), the disclosure of which is incorporated herein by reference), and an FDR-adjusted p-value of 0.05 was set as the significance threshold.

[0150] Area Subsampling The impact of utilizing a single randomly selected region per tissue sample, rather than multiple regions, when assessing protein expression changes during treatment versus before treatment was analyzed. For these analyses (Figure 21), 100 replicates were performed, a single region was selected from each tissue, and the average fold change and corresponding p-value was calculated across these 100 experiments. The number of random samples was chosen empirically by increasing the number of replicates beyond that required to make the resulting output robust, as the number of replicates used further increased (p-value convergence).

[0151] L2 Regularized Logistic Regression with Molecular Data Model and Features: L2-logistic regression using liblinear as the solver was used to classify pCR vs. non-pCR cases. Pre- and on-treatment marker values ​​were averaged across all ROIs to derive a composite value for each marker at that time point. Five patients were excluded from the model because data were only available at a single time point (Figure 4). Mean DSP marker expression features were used in models comparing patient time points, tumor vs. immune markers, DSP protein features vs. established predictive features (ER status and PAM50 classification), and DSP protein vs. bulk RNA features (using RNA gene transcripts corresponding to DSP protein markers). To assess heterogeneity, the standard error of the mean (SEM) was calculated for marker values ​​across all ROIs in tissues with at least three ROIs to derive a composite value for each marker at that time point. These SEM features were used in combination with mean expression features in models to assess the predictive value of heterogeneity.

[0152] Model comparison and performance evaluation by internal cross-validation: Model performance was evaluated and compared using nested cross-validation using the Python package sklearn (F. Pedregosa et al., J Mach Learn Res 12, 2825-2830 (2011), the disclosure of which is incorporated herein by reference). Data were divided into N folds using stratified sampling ("stratified cross-validation"). The number of folds was selected based on the number of cases in the non-pCR group (classes with fewer cases) so that the test data contained two cases from each class. Each model was trained on N-1 folds and scored using the average AUROC at the remaining folds. This process was repeated, maintaining a different fold each time. L2 penalty weights were selected using stratified cross-validation within the N-1 training dataset, and the weight associated with the highest average accuracy within this internal cross-validation was selected for scoring. This nested cross-validation process was repeated 100 times using randomly generated folds. Model scores were then compared using unpaired two-tailed t-tests with Holm-Bonferroni correction for multiple hypotheses. ROC curves are generated by averaging over ROC curves from 100 iterations of N-fold cross-validation, with each iteration containing a different random fold split.

[0153] Assessment of model performance in an independent cohort: As described above, pre- and on-treatment marker values ​​were averaged across all ROIs to derive a composite value for each marker at that time point. Model selection was performed using cross-validation as described above. The best-performing model was selected and trained using the entire discovery cohort. Finally, model performance based on AUROC was assessed in an independent validation (test) cohort.

[0154] Heterogeneity metrics Marker heterogeneity was calculated as the mean squared error from an analysis of variance performed in a linear model with marker value as the dependent variable and patient identity as the independent variable (datasets were dependent on the specific time point or clinical outcome of interest).

[0155] Marginal Complexity Using ImageJ (AB Watson, Mathematica 14 (2012), the disclosure of which is incorporated herein by reference), marginal complexity was calculated for each ROI for the panCK-enriched binary mask. A linear mixed-effects model with blocking by patient was used to compare marginal complexity of all panCK-enriched regions before and during treatment, as well as between cases that achieved pCR and those that did not.

[0156] Multiplex IHC analysis Unstained paraffin-embedded sections were analyzed by multiplex IHC analysis using the following markers: PanCK (AE1 / AE2), CD8, CD45 LCA, and HER2 (29D8 CST). Stained samples were scanned and digitized as a series of square subimages ("stamps") and visualized using HALO. PanCK masking and tissue region masking were performed for each stamped tissue region using Fiji (ImageJ). Briefly, the PanCK channel was used to create masks for the panCK-positive region and the entire tissue region (using the following ImageJ tools: contrast enhancement, threshold, dilation, fill holes, and create selection), and CD8, CD45, and HER2 were quantified within each mask region (using the ImageJ Measure tool). A weighted average (with a corresponding weight for each mask region) was used to calculate CD8, CD45, and HER2 levels across all scanned subimages containing tissue (either the tissue mask or the panCK mask region).

[0157] Doctrine of Equivalents While the above description includes many specific embodiments of the present invention, these should not be construed as limitations on the scope of the invention, but rather as examples of one embodiment thereof. Accordingly, the scope of the invention should be determined not by the embodiments illustrated, but by the appended claims and their equivalents. In certain embodiments, for example, the following are provided: (Item 1) 1. A method for diagnostically determining a pathological complete response in breast cancer, comprising: obtaining or having obtained an in-treatment cancer biopsy of an individual with breast cancer, wherein the in-treatment cancer biopsy is a cancer biopsy obtained after initiation of targeted therapy; measuring or having measured the expression of one or more sets of biomolecules in at least one region of interest of the cancer biopsy during the treatment; determining, or having determined, whether a targeted therapy provides a pathologic complete response in said individual utilizing a classifier and said biomolecule expression measurements; A method comprising: (Item 2) 10. The method of claim 1, further comprising administering a de-escalated treatment regimen to the individual if the targeted therapy is determined to provide a pathologic complete response. (Item 3) 3. The method of claim 2, wherein the de-escalated treatment regimen comprises a targeted therapeutic agent without systemic chemotherapy. (Item 4) 10. The method of claim 1, further comprising administering an escalated treatment regimen to the individual if the targeted therapy is determined not to provide a pathologic complete response. (Item 5) 5. The method of claim 4, wherein the escalated treatment regimen comprises a targeted therapeutic agent in combination with a chemotherapeutic agent. (Item 6) 5. The method of item 4, wherein the escalated treatment regimen comprises dual targeted therapy with two targeted therapeutic agents. (Item 7) Item 10. The method of item 1, wherein the breast cancer is HER2+. (Item 8) 2. The method of claim 1, wherein the set of one or more biological molecules comprises at least one immune response and activation biological molecule. (Item 9) 9. The method of item 8, wherein the at least one immune response and activation biomolecule is CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, or VISTA. (Item 10) 2. The method of claim 1, wherein the set of one or more biomolecules comprises at least one cell survival biomolecule. (Item 11) 11. The method of claim 10, wherein the at least one cell survival biomolecule is beta-2 microglobulin or Bcl-2. (Item 12) 2. The method of claim 1, wherein the breast cancer is HER2+ and the set of one or more biological molecules comprises at least one HER2 signaling pathway biological molecule. (Item 13) 13. The method of item 12, wherein the at least one HER2 signaling pathway biomolecule is HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, or p-STAT3. (Item 14) 2. The method of claim 1, wherein the set of one or more biomolecules comprises epithelial tumor tissue biomolecules. (Item 15) 13. The method of claim 12, wherein the at least one epithelial tumor tissue biomolecule is PanCK, Ki67, or beta-catenin. (Item 16) 2. The method of item 1, wherein the set of one or more biomolecules comprises at least two of the following biomolecules: HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, p-STAT3, PanCK, Ki67, beta-catenin, CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, VISTA, beta-2 microglobulin, or Bcl-2. (Item 17) 2. The method of claim 1, wherein the set of one or more biological molecules comprises CD45. (Item 18) 2. The method of claim 1, wherein the cancer is HER2+ and the set of one or more biological molecules includes HER2. (Item 19) 2. The method of claim 1, wherein the cancer biopsy is formalin-fixed, paraffin-embedded, OCT-embedded, or snap-frozen. (Item 20) Item 10. The method of item 1, wherein the at least one region of interest is determined by pancytokeratin-positive (panCK+) tumor cells or CD45-positive (CD45+) immune cells. (Item 21) obtaining or having obtained a pre-treatment cancer biopsy of an individual with breast cancer, wherein the pre-treatment cancer biopsy is a cancer biopsy obtained prior to the at least one cycle of targeted therapy; measuring or having measured expression of said set of one or more biomolecules in at least one region of interest of said pre-treatment cancer biopsy; determining dynamic biomolecular expression of the set of one or more biomolecules, wherein the dynamic biomolecular expression is utilized in the classifier to determine whether a targeted therapy will or will not provide a pathological complete response; Item 1, the method of claim 1 further comprising: (Item 22) 17. The method of claim 16, wherein a linear mixed effects model is used to quantify the dynamic biomolecular expression of the set of one or more biomolecules. (Item 23) 10. The method of claim 1, wherein the targeted therapy is administered as part of a neoadjuvant treatment regimen. (Item 24) 2. The method of item 1, wherein the cancer is HER2+ and at least one cycle of the targeted therapy comprises administration of trastuzumab, lapatinib, pertuzumab, or trastuzumab emtansine. (Item 25) 2. The method according to item 1, wherein biomolecule expression is determined by multiplex spatial proteomics. (Item 26) Item 10. The method of item 1, wherein the classifier is a regression model. (Item 27) 27. The method of claim 26, wherein the regression model is linear, logistic, polynomial, ridge, stepwise, LASSO, elastic net, L1 regularization, L2 regularization, or any combination thereof. (Item 28) 2. The method of claim 1, wherein the classifier incorporates a generalized linear model (GLM), ordinary least squares, random forest, decision tree, or neural network. (Item 29) 2. The method of item 1, wherein the classifier incorporates treatment type, ER status, PAM50 status, tumor size, tumor grade, cancer stage, patient age, or patient ethnicity. (Item 30) 2. The method of claim 1, wherein the cancer is HER2+ and the targeted therapy is one of trastuzumab, lapatinib, pertuzumab, or trastuzumab emtansine. (Item 31) 2. The method of claim 1, wherein the chemotherapeutic agent is one of paclitaxel, doxorubicin, or cyclophosphamide. (Item 32) 2. The method of item 1, wherein a threshold based on the sensitivity of the classifier is utilized to determine whether the targeted therapy provides a pathologic complete response. (Item 33) 33. The method of claim 32, wherein the threshold is based on a specificity of about 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%. (Item 34) assessing or having assessed immune cell infiltration within at least one region of interest of said cancer biopsy during said treatment; 2. The method of claim 1, wherein determining whether a targeted therapy provides a pathologic complete response in the individual utilizing the classifier further utilizes the immune cell infiltration assessment as a feature in the classifier. (Item 35) 35. The method of item 34, wherein the immune cell infiltration is assessed by hematoxylin and eosin staining or immunostaining. (Item 36) 35. The method of item 34, wherein said assessment of immune cell infiltration is assessment of stromal tumor-infiltrating lymphocyte infiltration. (Item 37) 35. The method of item 34, wherein said assessment of immune cell infiltration is assessment of intratumoral lymphocytes. (Item 38) 35. The method of item 34, wherein said assessment of immune cell infiltration is assessment of CD45-positive cell infiltration. (Item 39) 35. The method of item 34, wherein said assessment of immune cell infiltration is assessment of CD56-positive cell infiltration.

Claims

1. 1. A method for obtaining expression of a set of one or more biomolecules in at least one region of interest of an on-treatment cancer biopsy for use in the diagnostic determination of pathological complete response of HER2+ breast cancer, said method comprising: identifying one or more regions of interest within a HER2+ breast cancer in-treatment cancer biopsy, each region of interest being identified based on tumor cells expressing panCK or immune cells expressing CD45, wherein the in-treatment cancer biopsy is a cancer biopsy obtained from the individual after initiation of targeted therapy; measuring biomolecular expression of a set of one or more biomolecules within each region of interest to obtain a set of in-treatment biomolecular expression measurements, wherein the set of one or more biomolecules comprises immune response and activation molecules selected from CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, Granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, or VISTA; wherein the diagnostic determination comprises inputting the set of on-treatment biomolecule expression measurements into a trained machine learning classifier as an index for determining, by the trained machine learning classifier, whether a targeted therapy is sufficient to provide a pathological complete response in the individual.

2. 10. The method of claim 1, wherein the diagnostic determination that the targeted therapy is sufficient to provide a pathologic complete response indicates that a de-escalated treatment regimen should be administered to the individual.

3. 3. The method of claim 2, wherein the de-escalated treatment regimen comprises a targeted therapy without systemic chemotherapy.

4. 10. The method of claim 1, wherein the diagnostic determination that the targeted therapy is not sufficient to provide a pathologic complete response indicates that an escalated treatment regimen should be administered to the individual.

5. 5. The method of claim 4, wherein the escalated treatment regimen comprises a targeted therapeutic agent in combination with a chemotherapeutic agent.

6. 5. The method of claim 4, wherein the escalated treatment regimen comprises dual targeted therapy of two targeted therapeutic agents.

7. 2. The method of claim 1, wherein the set of one or more biomolecules comprises at least one cell survival biomolecule, and wherein the at least one cell survival biomolecule is beta-2 microglobulin or Bcl-2.

8. 10. The method of claim 1, wherein the set of one or more biomolecules comprises at least one HER2 signaling pathway biomolecule.

9. 9. The method of claim 8, wherein the at least one HER2 signaling pathway biomolecule is HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, or p-STAT3.

10. 2. The method of claim 1, wherein the set of one or more biomolecules comprises epithelial tumor tissue biomolecules, and wherein the at least one epithelial tumor tissue biomolecule is PanCK, Ki67, or beta-catenin.

11. 2. The method of claim 1, wherein the set of one or more biomolecules comprises at least two of the following biomolecules: HER2, AKT / p-AKT, S6 / p-S6, PTEN, p-ERK, p-STAT3, PanCK, Ki67, beta-catenin, CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, VISTA, beta-2 microglobulin, or Bcl-2.

12. The method of claim 1, wherein the immune response and activation biomolecule is CD45.

13. The method of claim 1 , wherein the set of one or more biomolecules comprises HER2.

14. 10. The method of claim 1, wherein the cancer biopsy is formalin-fixed, paraffin-embedded, OCT-embedded, or snap-frozen.

15. identifying one or more regions of interest within a pre-treatment cancer biopsy of the HER2+ breast cancer, each region of interest being based on tumor cells expressing panCK or immune cells expressing CD45, the pre-treatment cancer biopsy being a cancer biopsy obtained from the individual prior to initiation of the targeted therapy; measuring biomolecular expression of the set of one or more biomolecules within each region of interest to obtain a set of pre-treatment biomolecular expression measurements, wherein the set of one or more biomolecules comprises immune response and activation molecules selected from CD45, CD3, CD4, CD8, CD27, CD44, CD45RO, OX40L, ICOS, Granzyme B, CD19, CD11c, CD163, CD68, CD56, CD66B, CD14, STING, PD1 / PDL1, B7-H3, B7-H4, IDO-1, Lag3, or VISTA; 10. The method of claim 1, further comprising: The diagnostic determination may include: determining the dynamic biomolecular expression of the set of one or more biomolecules; inputting the dynamic biomolecular expression into a trained machine learning classifier as an index for determining, by the trained machine learning classifier, whether a targeted therapy provides a pathological complete response in the individual; The method further comprises:

16. 16. The method of claim 15, wherein a linear mixed effects model is utilized to quantify the dynamic biomolecular expression of the set of one or more biomolecules.

17. 10. The method of claim 1, wherein the in-treatment cancer biopsy was obtained after initiation of targeted therapy and without initiation of systemic therapy.

18. 10. The method of claim 1, wherein the targeted therapy comprises administration of trastuzumab, lapatinib, pertuzumab, or trastuzumab emtansine.

19. The method of claim 1 , wherein biomolecule expression is determined by multiplex spatial proteomics.

20. The method of claim 1 , wherein the machine-learned classifier is a regression model.

21. 21. The method of claim 20, wherein the regression model is linear, logistic, polynomial, ridge, stepwise, LASSO, elastic net, L1 regularization, L2 regularization, or any combination thereof.

22. 10. The method of claim 1, wherein the machine learning classifier incorporates a generalized linear model (GLM), ordinary least squares, random forest, decision tree, or neural network.

23. The diagnostic determination is characterized by: Treatment type, ER status, PAM50 status, tumor size, tumor grade, cancer stage, patient age or patient ethnicity 10. The method of claim 1, further comprising inputting one or more of:

24. 4. The method of claim 3, wherein the targeted therapeutic agent is one of trastuzumab, lapatinib, pertuzumab, or trastuzumab emtansine.

25. 6. The method of claim 5, wherein the chemotherapeutic agent is one of paclitaxel, doxorubicin, or cyclophosphamide.

26. 10. The method of claim 1, wherein a threshold based on the sensitivity of the machine learning classifier is utilized to determine whether the targeted therapy provides a pathologic complete response.

27. 27. The method of claim 26, wherein the threshold is based on a specificity of about 65%, 70%, 75%, 80%, 85%, 90%, 95%, 98%, 100%.

Images