Localized high-risk prostate cancer harbors an androgen receptor activity-low subpopulation susceptible to HER2 inhibition

Abstract

BACKGROUNDLocalized high-risk prostate cancer (PCa) often recurs despite neoadjuvant androgen deprivation therapy (ADT). We sought to identify baseline molecular programs that predict pathologic response and reveal targetable vulnerabilities.METHODSWe profiled 147 biopsy foci from 48 MRI-visible lesions in 37 patients before 6 months of ADT plus enzalutamide and radical prostatectomy. Residual cancer burden (RCB) at prostatectomy was the primary outcome. Analyses incorporated PTEN loss, TMPRSS2:ERG status, and HER2/androgen receptor (AR) immunohistochemistry on baseline and posttreatment tissues. Findings were evaluated in an external transcriptional cohort (n = 121) and by multiplex immunostaining in an independent cohort (n = 61). Functional assays tested enzalutamide-responsive enhancers near ERBB2 and sensitivity to HER2 inhibition.RESULTSA baseline, HER2-associated transcriptional program correlated with higher RCB and inversely with AR activity, independent of PTEN and ERG. Exceptional responders had lower HER2 protein levels in pretreatment biopsy specimens. The inverse AR-HER2 relationship recurred across data sets and multiplex immunostaining, which revealed coexisting AR-high/HER2-low and HER2-high/AR-low subpopulations. Enzalutamide inhibited AR-mediated repression of ERBB2. HER2-high/AR-low cells present before therapy resisted ADT yet were sensitive to HER2 inhibitors; combining HER2 inhibitors with enzalutamide increased tumor cell killing. These findings were reproduced in the external cohort and orthogonal assays.CONCLUSIONBaseline HER2 activity marks intrinsic resistance to neoadjuvant ADT in localized high-risk PCa and identifies a preexisting, targetable AR-low subpopulation. HER2-directed therapy, alone or with AR blockade, warrants clinical evaluation.TRIAL REGISTRATIONClinicalTrials.gov registration: NCT02430480.FUNDINGProstate Cancer Foundation; Department of Defense Prostate Cancer Research Program; National Institutes of Health.

Keywords: Genetics; Molecular pathology; Oncogenes; Oncology; Prostate cancer.

Figure 1. Integrated molecular landscape of prostate tumors prior to neoadjuvant-intense ADT.

(A) Schematic of workflow in which LCM and RNA-Seq of tumor foci from image-guided baseline biopsy specimens (left) were used to assess gene expression differences that track with posttreatment pathologic tumor volumes (right). (B) Distribution of RCB (1 row per patient) plotted on a logarithmic x-axis with a pseudocount (cm³ + 1). ER, exceptional responders (residual tumor volumes <0.05 cm³). INR patients had residual tumor volumes ≥0.05 cm³. (C) Principal component (PC) analysis of 147 baseline tumor foci transcriptomes. Each dot is colored by patient, with squares representing foci from INR patients and circles representing foci from ER patients. (D) Heatmap and oncoprint depicting molecular and histologic features of baseline tumors. Each column represents 1 laser-capture microdissected tumor focus subjected to whole-transcriptome sequencing. Identical values are given to IHC profiling performed on a single tissue that was subdivided for sequencing. Black bars at the bottom indicate multiple samples from the same patient. Samples are ranked from left to right by patient-level RCB volumes. Heatmap of IHC depicts histology intensity scores reported by Wilkinson et al. (22). ADPC, adenocarcinoma; NEPC, neuroendocrine; SCL, stem cell-like; WNT, Wnt-dependent.

Figure 2. Pathologic response to neoadjuvant ADT plus enzalutamide is associated with a transcriptional signature of elevated HER2 activity at baseline.

(A) Linear mixed-effects model depicting variance in gene expression across samples within each patient (by color) versus RCB (x-axis), showing gene expression patterns for positively correlating genes. (B) Volcano plot depicting DEGs determined using a linear mixed-effects model with RCB as a fixed effect and each patient as a random effect. Horizontal boundary depicts the P_adj = 0.05 cutoff. DEGs are quantified per cubic centimeter of posttreatment tumor volume: genes to the right are more expressed at baseline in tumors with higher posttreatment volumes, and genes to the left are less expressed. (C–E) All statistically significant DEGs (P_adj < 0.05) from the linear mixed-effects model were processed with the Upstream Regulator module of IPA. The 10 most activated and inactivated pathways (with P_adj < 0.05) are shown for DEG analyses in which (C) RCB was the only fixed effect, (D) RCB and ERG status were fixed effects, and (E) RCB and PTEN status were fixed effects. The bias-corrected z score is shown on the bottom x-axis and the P_adj value is shown on the top x-axis (–log₁₀ transformed).

Figure 3. HER2 protein is expressed at baseline in tumor foci that resist therapy and is retained posttreatment.

(A) Representative micrographs of anti-HER2 and anti-pHER2 IHC in baseline biopsies and residual tumor foci, showing examples from 3 patients with matched samples. Bar: 50 μm. (B) Scatter plot showing the association of per-patient HER2 baseline H-scores (x-axis) with posttreatment H-scores (y-axis). Statistical significance was determined using Spearman’s rank correlation. Line and gray shaded area show the linear regression line and 95% CI for the regression (0.046–0.65). (C) Density plots of HER2 baseline semi-quantitative IHC, stratified by pathologic response in the final surgical specimens. Statistical significance was determined by χ² test.

Figure 4. AR activity and HER2 activity maintain an inverse relationship in human prostate tumors and PCa cell lines.

(A–C) Statistically significant DEGs (P_adj < 0.05) that were correlated with the “Hallmark_Androgen_Response” mSigDB gene set processed by single-sample GSVA (ssGSVA) were analyzed with the Upstream Regulator module of IPA for (A) our original neoadjuvant (neoadj) ADT plus enzalutamide (enza) cohort, (B) 123 tumors from the PCBN, and the (C) PCa cohort of The Cancer Genome Atlas (TCGA). PRAD, protatic adenocarcinoma. A linear mixed-effects model was used with the neoadjuvant cohort in (A) modeling repeated measures from patients as random effects. The 10 most activated and inactivated pathways (with P_adj < 0.05) are shown. (D and E) Publicly available data from the Broad Institute Dependency Map (DepMap) is shown, in which AR-positive cell lines are depicted in blue and AR-negative cell lines are depicted in red. Gene expression was summarized using ssGSVA for AR and HER2 activity signatures from mSigDB (D), and cell death (sensitivity) was plotted to compare matched enzalutamide sensitivity or ERBB2 RNAi survival scores (E). DEMETER2 is an RNAi screen analytical framework within DepMap.

Figure 5. Enzalutamide treatment of prostate tumor cells selects for a preexisting subpopulation defined by low AR activity and high HER2 activity.

(A) LNCaP cells were treated with enzalutamide (enz) for 0–3 days and subjected to flow cytometry with antibodies against HER2. HER2 fluorescent intensity is depicted as a histogram from a representative experiment. (B and C) Total HER2 (as MFI) (B) or the top 20th percentile of each individual experiment’s (C) control sample fluorescence was used as a cutoff for measuring subpopulations at the 24- and 72-hour time points. Lines indicate data medians; error bars are 95% CIs (n = 5). Statistical significance was determined by Friedman test with Dunn’s post hoc test. (D) AR ChIP-Seq, H3K27ac ChIP-Seq, and H3K27ac HiChIP profiles of LNCaP cells at the KLK3 (left, shown as a control) and ERBB2 (right) loci in enzalutamide or untreated (DMSO) conditions. (E–I) Single-cell gene expression data from LNCaP cells treated with antiandrogen (E and F) were downloaded and normalized together. (F–H) UMAP projections of each treatment condition individually (F), clustered by differential expression (G), and overlaid, colored by treatment condition (H). (I) Following trajectory and pseudobulk differential expression analysis, statistically significant DEGs (P_adj < 0.05) in the “resistant” clusters were analyzed with the Upstream Regulator module of IPA. The 10 most activated and inactivated pathways (with P_adj < 0.05) are shown.

Figure 6. HER2 inhibition selects for PCa cells with greater AR activity.

(A) Schematic depiction of in vitro screening of 22Rv1, LAPC-4, and LNCaP cells with 8 different receptor RTKis. (B) IC₅₀ values are shown for individual lots of each RTKi used, per cell line. Lines and error bars depict the mean and SE data for at least 3 independent measurements. (C) Each cell line was treated with the indicated RTKi, abiraterone (ABI), or enzalutamide (ENZ) at the IC₅₀ derived for that lot of drug. CSS was used in place of FBS (see Methods) in cell culture media. Treatments were performed over 5 days, and samples were acquired on days 0, 1, 3, and 5. RNA extracted from each sample (performed in duplicate) was subjected to whole-transcriptome sequencing, with DEGs (correlating with time) processed using the upstream regulator module of IPA. (D and E) Western blots depicting protein levels of LNCaP cells (D) and 22Rv1 cells (E) treated with AFA (at its empirically determined IC₅₀) for 0–5 days. Blots shown are representative of at least 3 independent experiments. Actin is shown as a loading control.

Figure 7. Human prostate tumors variably harbor distinct subpopulations of tumor cells with elevated levels of HER2 activity.

(A) Schematic depiction of fully quantitative machine-guided analysis of multiplex staining. (B) Representative micrographs of AR activity–high/HER2 activity–low (HER2-low) and HER2 activity–high/AR activity–low tumor cell populations in 2 representative cases. Scale bar: 1 mm; inset bar: 100 μm. (C) Scatter plot representation of individual cellular HER2/PSA ratio scores for tumor cells in 62 slides, plotted on a log₁₀ y-axis. Red bar represents median data. The number of fully segmented cells is shown below the scatter plot. The 2 cases indicated in red are the same cases depicted in B. Case 492-004-4O is enumerated to illustrate the distribution of cells. (D) Statistically significant DEGs (FDR < 0.1) that were correlated with the median HER2/PSA ratio for each case were analyzed with the Upstream Regulator module of IPA. The 10 most activated and inactivated biological pathways are shown.

Figure 8. PCa cells expressing higher levels of HER2 are more sensitive to HER2 inhibition.

(A) Schematic depiction of flow cytometry assays to measure tumor cell sensitivity to HER2 inhibition. (B) Bar graph shows the proportion of nonapoptotic cells measured in Supplemental Figure 7. (C) Representative micrographs of phenotypically dominant subpopulations from P187 and P190, showing HER2-high (P187; left) and PSA/AR-high subpopulations (P190; right), respectively. Uniform contrast enhancement was performed across both slides to enable direct visual comparison. Bar: 1 mm; inset bar: 100 μm. (D) Scatter plot representation of individual cellular HER2/PSA ratio scores of FFPE tumor sections from the radical prostatectomy specimens of tumors used to generate each organoid model, plotted on a log₁₀ y-axis. Red bar represents median data. The number of fully segmented cells are shown below the scatter plot. (E) LNCaP cells were treated with either enzalutamide or AFA ± enzalutamide (at their respective IC₂₀) for 5 days. Cell viability was measured using CTG. Data shown are the average of 2 experiments. (F and G) LNCaP (F) and VCaP (G) cells were treated with enzalutamide (Enza), AFA ± enzalutamide, or NER ± enzalutamide for 5 days. Cell viability was measured using CTG. Lines present median data; error bars represent 95% CIs (n = 8). Statistical significance was measured using a repeated-measures ANOVA test with Bonferroni adjustment for multiple comparisons.