Comprehensive Spatial Transcriptomic Profiling of Cribriform, Intraductal, Atypical Intraductal Proliferation, and Ductal Adenocarcinoma of the Prostate

Ting Zhao; Cole Nawrocki; Linjie Xiong; Linda T Nieman; Todd E Abbott; Philip J Saylor; Francis J McGovern; David T Miyamoto; David T Ting; Douglas M Dahl; Chin-Lee Wu

doi:10.1016/j.labinv.2026.106154

. Author manuscript; available in PMC: 2026 Oct 4.

Published in final edited form as: Lab Invest. 2026 Jul 17;106(9):106154. doi: 10.1016/j.labinv.2026.106154

Comprehensive Spatial Transcriptomic Profiling of Cribriform, Intraductal, Atypical Intraductal Proliferation, and Ductal Adenocarcinoma of the Prostate

Ting Zhao ¹, Cole Nawrocki ², Linjie Xiong ¹, Linda T Nieman ², Todd E Abbott ³, Philip J Saylor ⁴, Francis J McGovern ⁵, David T Miyamoto ^2,⁶, David T Ting ^2,⁴, Douglas M Dahl ⁵, Chin-Lee Wu ¹

PMCID: PMC13633661 NIHMSID: NIHMS2208807 PMID: 42468875

Abstract

Purpose:

Cribriform (Crib) acinar adenocarcinoma (AAC), intraductal carcinoma of the prostate (IDC-P), atypical intraductal proliferation (AIP) and ductal adenocarcinoma (DAC) are linked to poor outcomes in prostate cancer (PCa). We analyzed their gene expression using spatial transcriptomics.

Materials and Methods:

A tissue microarray of 18 FFPE cores from 17 prostatectomies was profiled for expression of 18,676 mRNAs with the Bruker GeoMx Digital Spatial Profiler platform. 53 areas of interest were selected based on H&E, PIN4 IHC and fluorescence markers. Gene set enrichment analysis was conducted for Reactome gene sets.

Results:

Most patients had grade group ≥3 (94%) and ≥pT3 disease (65%). At a median follow-up of 47 months (range, 18–130), 31% developed metastases. 16 tumors showed mixed morphology. 3 DACs had adjacent intraductal components with preserved basal cells (ductal IDC-P). 9/10 acinar IDC-P and 6/7 AIP presumably represented intraductal spread (invasive type); one was mostly non-invasive (95% IDC-P/AIP, putative precursor type). Crib AAC, acinar IDC-P, and AIP demonstrated substantial transcriptomic similarity, with only 4 differentially expressed genes between acinar IDC-P and crib AAC and between acinar IDC-P and AIP. Crib AAC exhibited Notch signaling and homologous recombination DNA repair enrichment versus low-grade AAC. Acinar IDC-P upregulated GADD45G/ERRFI1 and downregulated CCN3/TMEFF2, with reduced PKN1-mediated androgen receptor signaling versus crib AAC. AIP showed reduced GPCR signaling versus IDC-P and crib AAC. Acinar IDC-P upregulated TSPAN8/CA4, and downregulated TRPM8/DHCR24 versus AIP. Putative precursor-type IDC-P/AIP downregulated oncogenic and stemness programs versus invasive type. DAC was transcriptomically more similar to crib AAC than to non-crib AAC. Ductal IDC-P downregulated luminal epithelial programs, and upregulated antigen presentation/immune signaling versus DAC.

Conclusions:

Crib AAC, acinar IDC-P, and AIP demonstrated substantial transcriptomic similarity despite distinct differences. DAC showed greater transcriptomic similarity to crib AAC than to non-crib AAC.

Keywords: prostate cancer, spatial transcriptomics, IDC-P, AIP, cribriform, ductal

Introduction

Prostate cancer (PCa) is the most commonly diagnosed malignancy in men and the second leading cause of cancer-related death. Cribriform (Crib) acinar adenocarcinoma (AAC), intraductal carcinoma of the prostate (IDC-P), and ductal adenocarcinoma (DAC) are frequently associated with poor outcomes.¹ Crib AAC, IDC-P, DAC and atypical intraductal proliferation (AIP) are generally considered within the spectrum of large-gland proliferations of the prostate, in contrast to the more common small acinar-type PCa.² Since 2019, the International Society of Urological Pathology (ISUP) has recommended mandatory reporting of cribriform morphology owing to its significant prognostic and therapeutic implications.³ The most widely accepted criterion for acinar IDC-P is solid or dense cribriform growth within pre-existing, typically expanded duct-acinar structures.⁴ Most acinar IDC-P is regarded as representing retrograde intraductal spread of invasive PCa and was recently proposed by the Genitourinary Pathology Society (GUPS) and ISUP Joint Expert Consultation group to be termed “invasive type”.⁵ Rare cases of acinar IDC-P identified in radical prostatectomy (RP) specimens without associated invasive PCa or with only grade group (GG) 1 PCa have also been reported;⁶ these have been interpreted as putative precursors of PCa and proposed to be termed “precursor type”.⁵ The molecular features distinguishing the presumed invasive-type from the precursor-type remain to be characterized. While most IDC-P are of the acinar type, PCa with ductal morphology may also retain a basal cell layer (ductal IDC-P), which is less well characterized than acinar IDC-P and may represent intraductal spread of DAC.^5,7 Historical conflicting practice recommendations regarding the grading of IDC-P from two leading uropathology societies have caused confusion for both pathologists and clinicians. The recent GUPS and ISUP Joint Expert Consultation group proposed that although pure IDC-P should not be graded, the IDC-P component associated with invasive PCa of any grade and in any specimen should be included in the Gleason score unless it is clearly spatially distinct.⁵ AIP in which acinar IDC-P is favored but the diagnostic criteria are not fully met has been proposed as “AIP, suspicious for IDC-P” and should not be graded. AIP on prostate biopsy has been associated with unsampled adverse pathology⁸ and may also represent intraductal spread of high-grade PCa. The molecular relationships of AIP relative to acinar IDC-P and AAC remain unclear. Acinar IDC-P and crib AAC have often been grouped together in previous studies,^9,10 and it remains uncertain whether these two entities possess distinct molecular characteristics. Additionally, conflicting data have been reported regarding the clonal relationship between DAC and AAC.^11–13

Spatial transcriptomics (ST) is a powerful tool for investigating molecular relationships within spatially and histologically heterogeneous PCa. While conventional transcriptomic approaches valuable transcriptional information, tissue homogenization prior to sequencing results in loss of spatial context. Recent technological advances in imaging-based methods and next-generation sequencing (NGS) have enabled ST to quantify gene expression across tissue sections while preserving spatial information.¹⁴ ST has highlighted the immunosuppressive tumor microenvironment (TME) in PCa.¹⁵ It also enables high-resolution correlation of histological components with molecular features, uncovering intratumoral heterogeneity that conventional transcriptomic approaches cannot resolve. ST profiling of aggressive PCa with heterogeneous histological components may enhance understanding of the biological behavior and molecular characteristics of each component, help clarify conflicting findings and address ongoing controversies, and ultimately identify potential therapeutic targets and inform the development of personalized treatment strategies.

The objective of this study is to investigate and compare gene expression in histologically heterogeneous PCa with aggressive features using ST.

Materials and Methods

Histological Review and Specimen Acquisition

This study was approved by the Institutional Review Board, and informed consent was obtained. Seventeen patients who had PCa enriched with crib AAC, IDC-P, AIP, and DAC in RP specimens were included. Histological slides were reviewed by two genitourinary pathologists (TZ and CLW). Cribriform architecture was defined according to the 2021 ISUP consensus as a confluent sheet of contiguous malignant epithelial cells with multiple glandular lumina lacking intervening stroma or mucin.¹⁶ AIP was defined as a large-gland intraductal proliferation that exceeds the architectural complexity and atypia of high-grade prostatic intraepithelial neoplasia (HGPIN) but falls short of the diagnostic criteria for IDC-P.⁵ Formalin-fixed, paraffin-embedded (FFPE) tissue from RP specimens was used to construct a tissue microarray (TMA) comprising 18 cores, each 3 mm in diameter. One DAC specimen was represented by two cores sampled from distinct tumor areas. Hematoxylin and eosin (H&E) and PIN4 (p63, HMWCK, AMACR/P504S; Biocare, Pacheco, CA, USA) immunohistochemical (IHC) staining were performed on TMA sections. Clinicopathological data, including patient age, Gleason score, tumor stage, treatment history, and follow-up data, were reviewed.

GeoMx^® Digital Spatial Profiler (DSP)

The GeoMx DSP platform (Bruker Spatial Biology, Seattle, WA, USA) was used to generate spatially resolved transcriptomic data from transcribed mRNA. TMA sections were hybridized with the GeoMx DSP Human Whole Transcriptome Atlas, consisting of a mixture of unique oligonucleotide-labeled probes targeting 18,676 mRNAs. Slides were co-stained with fluorescent markers, including pan-cytokeratin (pan-CK, green; Bruker), CD45 (Magenta; Bruker), and alpha-smooth muscle actin (α-SMA, cyan; clone 1A4, Novus Biologicals, Centennial, CO, USA) conjugated with Alexa Fluor 647 (Abcam, Waltham, MA, USA), to visualize epithelial, immune, and myofibroblast/smooth muscle compartments, respectively, across distinct emission spectra. A total of 54 regions of interest (ROIs), representing heterogeneous histological components, were manually annotated by two genitourinary pathologists (TZ and CLW) using corresponding H&E- and PIN4 IHC-stained sections and fluorescent markers as references. In addition to Crib AAC, IDC-P, AIP, and DAC, the adjacent non-cribriform components in 16 of the 17 cases were also annotated and subsequently served as the comparator group in ST. Pan-CK fluorescence was used to delineate tumor cells. ROIs corresponding to distinct histological components were selected, and the entire regions were designated as areas of interest (AOIs) for downstream profiling. After quality control (QC; details under “GeoMx DSP Computational Methods”), 53 AOIs were retained for downstream analyses, including 7 crib AAC, 7 high-grade (HG; defined as GG ≥3) non-crib AAC, 8 low-grade (LG) non-crib AAC, 10 acinar IDC-P, 7 AIP, 9 DAC, and 3 ductal IDC-P. Duplicate AOIs representing the same histological component were sampled from two different tumor regions in two patients. The AOIs from each patient are listed in Table 1. Representative AOIs are shown in Fig. 1E and 1H. Oligonucleotide barcodes, linked via ultraviolet light-cleavable bonds, were released from AOIs and collected into a 96-well plate. Collected oligos were pooled according to the Bruker GeoMx-NGS Readout Library Preparation User Manual (MAN-10117–03). Libraries were prepared using Illumina i5 × i7 dual indexing adapters, followed by amplification and ligation. Final library concentrations were quantified by quantitative PCR (qPCR) using KAPA Library Quantification on a Roche LightCycler^® 480 system. Sequencing was performed on an Illumina NextSeq 1000/2000 platform with a 1% PhiX spike-in using a 2 × 100-cycle Mid Output v3 flow cell. Resulting FASTQ files were processed with the GeoMx NGS pipeline through Illumina’s DRAGEN service to generate probe-specific count data, which was exported in DCC format.

Table 1.

Clinicopathological features of patients and selected areas of interest.

Pt	Age, y	Gleason score on RP	Tumor stage	AOIs¹	Treatment	Follow-up (m)

1	64	4 + 3	T3a N0	DAC, non-crib AAC (4+4)	RP	NED (130)
2	76	4 + 3	T3a N0	DAC	RP	NED (109)
3	68	4 + 4	T2 N0	DAC, ductal IDC-P, non-crib AAC (3+3)	RP	NED (52)
4	55	3 + 4	T2 N0	DAC, ductal IDC-P, non-crib AAC (3+3)	RP	NED (40)
5	52	4 + 4 + 5	T2 N0	DAC, non-crib AAC (3+3)	RP, ADT, RT	BCR (28)
6	64	4 + 3	T3a N0	DAC, non-crib AAC (3+4), AIP	RP, ADT, RT	BCR (34)
7	68	4 + 4	T2 N0	DAC, non-crib AAC (4+4)	RP	NED (18)
8	73	4 + 4	T2 N0	DAC, ductal IDC-P, non-crib AAC (4+3), acinar IDC-P	RP, ADT	NED (69)
9	59	4 + 5	T3b N1	Crib AAC, non-crib AAC (4+4), acinar IDC-P	RP, ADT	AWM (74)
10	76	4 + 5	T3b N1	DAC, crib AAC, non-crib AAC (3+3), acinar IDC-P	RP, ADT, RT	AWM (44)
11	52	4 + 5	T3b N1	Crib AAC, non-crib AAC (4+3), acinar IDC-P, AIP	RP, ADT, RT	AWM (84)
12	62	4 + 4	T3a N0	Crib AAC, non-crib AAC (4+3), acinar IDC-P, AIP	RP, ADT, RT	BCR (66)
13	62	4 + 5	T3b N0	Non-crib AAC (4+3), acinar IDC-P, AIP	RP, ADT, RT	AWM (50)
14	62	4 + 4	T3b N0	Crib AAC, non-crib AAC (3+3), acinar IDC-P	RP, ADT	LOF
15	63	4 + 3	T2 Nx	Non-crib AAC (3+3), acinar IDC-P, AIP	RP	NED (21)
16	69	4 + 5	T3b N1	Crib AAC, non-crib AAC (3+3), acinar IDC-P, AIP	RP, ADT	AWM (20)
17	71	4 + 4	T3a N0	Crib AAC, non-crib AAC (4+3), acinar IDC-P, AIP	RP	NED (20)

Open in a new tab

Abbreviations: AAC, acinar adenocarcinoma; ADT, androgen deprivation therapy; AIP, atypical intraductal proliferation; AOIs, areas of interest; AWM, alive with metastasis/metastases; crib, cribriform; BCR, biochemical recurrence; DAC, ductal adenocarcinoma; IDC-P, intraductal carcinoma of the prostate; LOF, loss of follow-up; m, months; NED, no evidence of disease; pt, patient; RP, radical prostatectomy; RT, radiotherapy; y, years.

Gleason scores in the “AOIs” column reflect tumor grade in the selected areas of interest.

Representative H&E and PIN-4-stained images and AOIs. Sixteen tumors exhibited mixed morphology, including adjacent DAC and AAC (A, B), adjacent DAC, crib AAC, and acinar IDC-P (K, L), and adjacent crib AAC and acinar IDC-P (M, N). Three DACs showed adjacent intraductal components with preserved basal cells (ductal IDC-P; F-J), whereas one tumor was composed exclusively of DAC (C-E). Acinar IDC-P/AIP predominated in a largely non-invasive tumor composed of approximately 95% IDC-P/AIP (putative precursor type) and 5% AAC, predominantly Gleason pattern 3 (O, P; green circles indicate rare AAC). Representative AOIs are shown (E, H), with green fluorescence indicating cytokeratin and blue indicating nuclei. PIN4 IHC staining: AMACR (red); p63 and high-molecular-weight cytokeratin (brown). AAC, acinar adenocarcinoma; AIP, acinar intraductal proliferation; AOIs, areas of interest; DAC, ductal adenocarcinoma; IDC-P, intraductal carcinoma of the prostate; IHC, immunohistochemical.

GeoMx DSP Computational Methods

Using the GeomxTools R package (Bruker), DCC files were further processed into a counts matrix, which was subject to QC. Briefly, AOIs with low read alignment quality, low total read counts, or insufficient segment area were excluded from downstream analyses. Next, outlier probes were excluded, and the data was aggregated by gene target, producing a gene by AOI expression matrix. Finally, genes failing to meet the limit of quantification (LOQ) in ≥10% of AOIs were filtered. After QC, data for 14,994 genes remained across 53 AOIs. Filtered expression data were then “Q3 normalized” by aligning all AOIs’ 75th percentiles of expression to the same value. To achieve this alignment, a size factor was derived for each AOI. Specifically, a given AOI’s size factor was defined as the 75th percentile count value across its expression profile, divided by the geometric mean of all such 75th percentile values across all AOIs. Dividing the expression profile of each AOI by its corresponding size factor produced the Q3 normalized matrix, which was transformed to log scale and batch-corrected with limma for downstream visualization.¹⁷

The DESeq2 R package¹⁸ was used for differential expression analysis. Notably, the size factors computed for Q3 normalization were provided to DESeq2 ahead of modeling, and a patient identifier effect was added to the model formula, when applicable. For each comparison, a Wald test was conducted, and a two-sided p-value < 0.05, after FDR adjustment, was considered statistically significant. Gene set enrichment analysis (GSEA) was conducted for Reactome gene sets with the fgsea algorithm as implemented in the clusterProfiler R package.¹⁹ Importantly, GSEA was predicated on gene lists created by pre-ranking differential expression results by Wald statistic. Lastly, single-sample gene set enrichment analysis (ssGSEA) was performed with the ssGSEA2 R package to compare putative precursor-type IDC-P and AIP from one patient with presumed invasive-type from other patients. Pathway-level changes were described descriptively, and no statistical comparisons were performed for this analysis. All analysis code is available in the following GitHub repository: ccnawrocki/wu-gu-geomx-project.

Results

Clinicopathological Features

All patients underwent RP. The median patient age was 64 years (range, 52–76); 94% of tumors were HG (defined as GG ≥3), and 65% demonstrated ≥pT3 disease. At a median follow-up of 47 months (range, 18–130), all patients with available follow-up were alive; 31% developed metastases and 19% developed biochemical recurrence (BCR) without clinical metastasis (Table 1). In addition to RP, 4 patients (24%) received additional androgen deprivation therapy (ADT), and 6 patients (35%) received both ADT and radiotherapy.

Sixteen tumors exhibited mixed morphology, including adjacent DAC and AAC (Fig. 1A, B), adjacent DAC, crib AAC, and acinar IDC-P (Fig. 1K, L), and adjacent crib AAC and acinar IDC-P (Fig. 1M, N). All acinar IDC-P in this study demonstrated a dense cribriform morphology. Three DACs had adjacent intraductal components with preserved basal cells (ductal IDC-P; Fig. 1F–J), while one tumor was composed exclusively of DAC (Fig. 1C–E; Case #2 in Table 1). The majority of acinar IDC-P (9/10) and AIP (6/7) presumably represented intraductal spread in the setting of extensive invasive cancer (invasive type). In contrast, acinar IDC-P/AIP predominated in a largely non-invasive tumor composed of approximately 95% IDC-P/AIP and 5% AAC (predominantly GG1) (putative precursor type; Fig. 1O, P).

Overview of Spatial Transcriptomic Comparisons

Crib AAC, acinar IDC-P, and AIP demonstrated substantial transcriptomic similarity, with only 4 differentially expressed genes between acinar IDC-P and crib AAC and between acinar IDC-P and AIP, although distinct transcriptional differences exist. Acinar IDC-P exhibited significant transcriptomic differences compared with adjacent HG non-crib AAC and LG AAC within the same tumor. DAC was transcriptomically more similar to crib AAC than to HG non-crib AAC or LG AAC. Crib AAC, acinar IDC-P, AIP, and DAC shared a luminal epithelial transcriptional phenotype and were enriched for tumor-intrinsic genes and pathways, whereas non-crib AAC demonstrated enrichment of TME-associated programs.

Comparison of Gene Expression Between Crib AAC and Other AAC

Compared with non-crib AAC, crib AAC preferentially expressed canonical luminal epithelial genes, such as FOLH1 (PSMA) and KLK11, as well as androgen-regulated genes including TMEFF2 and SCUBE2. Additional upregulated genes included JAG1, a mediator of Notch signaling, and GMNN, a regulator of cell-cycle progression (Fig. 2A). In contrast, genes enriched in non-crib AAC included MYH11, MYL9, ACTA2, and CCN2, consistent with stromal activation and myofibroblastic TME. Subgroup analyses comparing crib AAC with HG non-crib AAC and LG AAC yielded largely similar findings (Fig. 2B, C).

Differential gene expression and pathway enrichment in crib AAC versus other AAC components. Volcano plots showing differentially expressed genes (DEGs) between crib AAC and all non-crib AAC (A), HG non-crib AAC (B), and LG AAC (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; crib, cribriform; HG, high grade; LG, low grade.

Distinct pathway enrichment between crib AAC and non-crib AAC was observed (Table 2, comparisons 1.1–1.3; Fig. 2). Crib AAC was enriched for Notch signaling (Notch1 and Notch4) relative to overall non-crib AAC and LG AAC. Compared with HG non-crib AAC, crib AAC showed enrichment of translation-related programs associated with protein synthesis and ribosome biogenesis, whereas, relative to LG AAC, crib AAC exhibited enrichment of homologous recombination-mediated DNA repair, G-protein-coupled receptor (GPCR)/amine receptor signaling, and sterol regulatory element-binding protein (SREBP)-mediated cholesterol/lipid regulation. In contrast, non-crib AAC exhibited enrichment of TME programs, consistent with a microenvironment-driven phenotype.

Table 2.

Interpretation of gene set enrichment analysis (GSEA) pathway results.

Comparisons	Significantly enriched pathways		Interpretation

1.1 Crib AAC vs overall non-crib AAC	Crib AAC: Notch1 signaling activation	Overall non-crib AAC: ECM remodeling (collagen, integrin, proteoglycans), stromal muscle contraction, platelet activation, IL-4/IL-13/interferon signaling	Crib AAC: Notch signaling Overall non-crib AAC: TME interactions
1.2 Crib AAC vs HG¹ non-crib AAC	Crib AAC: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting to membrane	HG non-crib AAC: ECM organization, integrin interactions, collagen/elastic fiber formation, stromal contraction, platelet & immune signaling	Crib AAC: Translation-related programs HG non-crib AAC: TME interactions
1.3 Crib AAC vs LG¹ AAC	Crib AAC: Notch4 signaling activation, homologous recombination-mediated DNA repair (D-loop, Holliday junction), GPCR/amine receptors, SREBP & cholesterol regulation, neuronal/synaptic signaling	LG AAC: Translation, ribosome biogenesis, SRP-dependent targeting, selenoamino acid metabolism	Crib AAC: Notch signaling; homologous recombination DNA repair; GPCR/amine receptor signaling; SREBP-mediated cholesterol/lipid regulation LG AAC: Translation-related programs
2.1 Acinar IDC-P vs all AAC	Acinar IDC-P: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino metabolism, SLIT/ROBO signaling	All AAC: ECM organization, collagen remodeling, integrin/adhesion, stromal contractility, platelet activation/vascular signaling	Acinar IDC-P: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling All AAC: TME interactions
2.2 Acinar IDC-P vs crib AAC	Acinar IDC-P: Translation, ribosome programs, selenoamino metabolism, SLIT/ROBO signaling, interferon signaling	Crib AAC: DNA repair (base excision repair, nonhomologous end-joining), nucleosome assembly, telomeric packaging, DNA methylation, histone modification	Acinar IDC-P: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling Crib AAC: Error-prone DNA repair (base excision repair, nonhomologous end-joining); epigenetic and chromatin regulatory activity
2.3 Acinar IDC-P vs HG non-crib AAC	Acinar IDC-P: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino metabolism	HG non-crib AAC: ECM remodeling, integrin/collagen programs, platelet activation/degranulation, IGF signaling	Acinar IDC-P: Translation-related programs; redox stress adaptation HG non-crib AAC: TME interactions
2.4 Acinar IDC-P vs LG AAC	Acinar IDC-P: Apoptosis-related adhesion cleavage, neuronal/synaptic signaling, keratinization, cornified envelope	LG AAC: Translation/ribosome programs, ECM organization/degradation, neuronal signaling	Acinar IDC-P: Epithelial differentiation-associated programs LG AAC: Translational activity; TME interactions
3.1 Paired² acinar IDC-P vs all AAC	Acinar IDC-P: Translation, ribosome biogenesis, SRP-dependent targeting, selenoamino metabolism, SLIT/ROBO signaling	All AAC: ECM remodeling, collagen/proteoglycans formation, integrin/adhesion, muscle contraction, platelet activation	Acinar IDC-P: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling All AAC: TME interactions
3.2 Paired acinar IDC-P vs crib AAC	Acinar IDC-P: Translation, ribosome programs, SRP-dependent targeting, selenoamino metabolism, SLIT/ROBO signaling	Crib AAC: Pre-Notch expression, PKN1-AR-KLK2/KLK3 signaling, DNA methylation, histone modifications, cholesterol biosynthesis	Acinar IDC-P: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling Crib AAC: Notch signaling; PKN1-AR-KLK2/KLK3 signaling; epigenetic and chromatin regulatory activity
3.3 Paired acinar IDC-P vs HG non-crib AAC	Acinar IDC-P: Translation, ribosome biogenesis, SRP-dependent targeting, selenoamino metabolism, SLIT/ROBO signaling	HG non-crib AAC: ECM remodeling, collagen/elastic fiber formation, integrin/adhesion, muscle contraction, platelet activation, IL-4/IL-13 signaling	Acinar IDC-P: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling HG non-crib AAC: TME interactions
3.4 Paired acinar IDC-P vs LG AAC	Acinar IDC-P: Mitochondrial metabolism (ETC, TCA cycle), translation, RNA splicing, Golgi trafficking, metabolic genes regulated by p53, DNA repair	LG AAC: ECM remodeling, collagen/elastic fiber formation, platelet activation, vascular adhesion, IL-4/IL-13 signaling	Acinar IDC-P: Bioenergetic/metabolic programs; DNA repair LG AAC: TME interactions
4.1 AIP vs all AAC	AIP: Translation, ribosome biogenesis, selenoamino acid metabolism, SLIT/ROBO signaling	All AAC: ECM organization, collagen formation/degradation, platelet activation, GPCR signaling	AIP: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling All AAC: TME interactions
4.2 AIP vs crib AAC	AIP: Translation, ribosome biogenesis, selenoamino acid metabolism, SLIT/ROBO signaling	Crib AAC: GPCR signaling	AIP: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling Crib AAC: GPCR signaling
4.3 AIP vs HG non-crib AAC	AIP: Translation, ribosome biogenesis, selenoamino acid metabolism, SLIT/ROBO signaling	HG non-crib AAC: ECM organization, collagen formation/degradation, integrin/adhesion, platelet activation	AIP: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling HG non-crib AAC: TME interactions
4.4 AIP vs LG AAC	AIP: Mitochondrial metabolism (ETC, TCA cycle), RNA splicing/processing, proteasome degradation, metabolic genes regulated by p53, SLIT/ROBO signaling	LG AAC: ECM organization, collagen formation, integrin/adhesion, platelet activation, IL-4/IL-13 signaling	AIP: Bioenergetic/metabolic programs; SLIT/ROBO signaling LG AAC: TME interactions
5 AIP vs Acinar IDC-P	AIP: Translation/ribosome programs; SRP-dependent targeting; selenoamino acid metabolism, SLIT/ROBO signaling	IDC-P: GPCR signaling; immune/inflammation signaling (TNF, ADORA2A)	AIP: Translation-related programs; redox stress adaptation; SLIT/ROBO signaling Acinar IDC-P: GPCR and inflammatory signaling
6. IDC-P and AIP³: Presumed invasive-type vs putative precursor-type	Presumed invasive-type: DNA replication/repair, RNA biogenesis control, VEGF-VEGFR interactions/dimerization; MET-PI3K/AKT, RUNX3-WNT, and p38-MAPK signaling; POU5F1 (OCT4), SOX2, NANOG programs; mitochondrial/energy metabolism, lipid metabolism, GAG/ECM anchorage	Putative precursor-type: Adaptive immunity/checkpoint activation, interferon/innate signaling, inflammatory cytokines, pyroptosis/celldeath pathways, antigen presentation/complement; ion-channel signaling; SLIT/ROBO signaling	Presumed invasive-type: Oncogenic stemness and replicative/metabolic signaling Putative precursor-type: Immune, inflammatory and cell-death signaling; SLIT/ROBO signaling
7.1 DAC vs crib AAC	DAC: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	Crib AAC: Nef-mediated down-regulation of MHC I (immune-evasion module)	DAC: Translation-related programs; redox stress adaptation Crib: Antigen presentation down-regulation
7.2 DAC vs HG non-crib AAC	DAC: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	HG non-crib AAC: ECM remodeling; integrin/non-integrin ECM interactions, platelet activation	DAC: Translation-related programs; redox stress adaptation HG non-crib AAC: TME interactions
7.3 DAC vs LG AAC	DAC: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	LG AAC: ECM organization; collagen/elastic fiber formation, integrin/non-integrin ECM interactions, platelet activation, neutrophil degranulation, IL-4/IL-13/interferon signaling	DAC: Translation-related programs; redox stress adaptation LG AAC: TME interactions
8.1 Paired DAC vs all AAC	DAC: Translation, ribosome biogenesis, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	All AAC: ECM organization/degradation; integrin/non-integrin ECM interactions; collagen/elastic fiber formation, GPCR/PDGF/RTK signaling	DAC: Translation-related programs; redox stress adaptation All AAC: TME interactions; GPCR/PDGF/RTK signaling
8.2 Paired DAC vs HG non-crib AAC	DAC: Translation, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	HG non-crib AAC: ECM remodeling; collagen/elastic fiber formation, integrin/non-integrin ECM interactions, GPCR/PDGF/RTK signaling	DAC: Translation-related programs; redox stress adaptation HG non-crib AAC: TME interactions and GPCR/PDGF/RTK signaling
8.3 Paired DAC vs LG AAC	DAC: Translation, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	LG AAC: ECM remodeling; collagen/elastic fiber formation, integrin/non-integrin ECM interactions, IL-4/IL-13 signaling, GPCR/RTK signaling	DAC: Translation-related programs; redox stress adaptation LG non-crib AAC: TME interactions and GPCR/RTK signaling
9.1 DAC vs ductal IDC-P	DAC: Translation, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	Ductal IDC-P: Immune/antigen presentation, inflammation, neutrophil degranulation; complement regulation; PD-1 pathway; TCR/BCR signaling; IL-1/4/10/13 signaling, IFN-α/β/γ signaling	DAC: Translation-related programs; redox stress adaptation Ductal IDC-P: Antigen presentation and immune pathways, including checkpoint and T-cell signaling
9.2 Paired DAC vs ductal IDC-P	DAC: Translation, rRNA processing, SRP-dependent targeting, selenoamino acid metabolism, EIF2AK4 (GCN2) amino-acid stress	Ductal IDC-P: Robust immune/T-cell programs, PD-1 signaling, antigen presentation, phosphorylation of CD3/TCRζ, ZAP-70 translocation, TLR cascades; chemokine receptor binding; IL-4/10/13/20 signaling, IFN-α/β/γ signaling	DAC: Translation-related programs; redox stress adaptation Ductal IDC-P: Antigen presentation and immune pathways, including checkpoint and T-cell signaling

Open in a new tab

Abbreviations: AAC, acinar adenocarcinoma; AR, androgen receptor; ADORA2A, adenosine A2a receptor; AIP, atypical intraductal proliferation; crib, cribriform; BCR, B cell receptor, DAC, ductal adenocarcinoma; ECM, extracellular matrix; EIF2AK4, eukaryotic translation initiation factor 2 alpha kinase 4; ETC, electron transport chain; D-loop, displacement loop; GAG, glycosaminoglycan; GCN2: general control nonderepressible 2; GPCR, G-protein-coupled receptor; HG, high-grade; IDC-P, intraductal carcinoma of the prostate; IGF, insulin-like growth factor; IFN, interferon; IL, interleukin; KLK, kallikrein-related peptidase; LG, low-grade; MAPK, mitogen-activated protein kinase; MET, mesenchymal-epithelial transition factor; MHC, major histocompatibility complex; Nef, negative factor; PI3K, phosphatidylinositol-3-kinase; PDGF, platelet-derived growth factor; PKN1, protein kinase N1; POU5F1, POU class 5 homeobox 1; ROBO, roundabout; RTK, receptor tyrosine kinase; RUNX3, runt-related transcription factor 3; rRNA, ribosomal RNA; SLIT, slit guidance ligand; SOX2, SRY-box transcription factor 2; SREBP, sterol regulatory element-binding protein; SRP, signal recognition particle; TCA, tricarboxylic acid; TCR, T cell receptor, TLR, Toll-like receptor; TME, tumor-microenvironment; TNF, tumor necrosis factor; VEGF, vascular endothelial growth factor; VEGFR, vascular endothelial growth factor receptor; vs, versus.

High grade was defined as grade group ≥3, whereas low grade was defined as grade group <3.

Paired, adjacent areas of interest within the same tumor.

Acinar IDC-P/AIP predominated in a largely non-invasive tumor. Single-sample gene set enrichment analysis (ssGSEA) was performed. Descriptive pathway changes were reported, and no statistical comparisons were conducted.

Comparison of Gene Expression Between Acinar IDC-P and AAC (crib and non-crib)

Among all comparisons with AAC, acinar IDC-P showed the fewest differentially expressed genes relative to crib AAC, with differences observed in only 4 genes. Acinar IDC-P upregulated GADD45G and ERRFI1, genes involved in cell cycle and growth regulatory pathways, whereas it downregulated CCN3, an extracellular matrix (ECM)-associated signaling protein, and TMEFF2, an androgen-regulated gene associated with luminal epithelial differentiation (Fig. 3B). Compared with AAC overall, acinar IDC-P demonstrated modest upregulation of CDK19 and AMD1, genes involved in transcriptional regulation and polyamine metabolism, respectively (Fig. 3A). Compared with both HG non-crib AAC and LG AAC, acinar IDC-P upregulated genes associated with luminal epithelial differentiation and androgen signaling, including CRISP3, SPON2, SCUBE2, and TMEFF2 (Fig. 3C, Supplementary Fig. S1A).

Differential gene expression and pathway enrichment in acinar IDC-P versus AAC. Volcano plots showing differentially expressed genes (DEGs) between acinar IDC-P and all AAC (A), crib AAC (B), and HG non-crib AAC (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; crib, cribriform; HG, high grade; IDC-P, intraductal carcinoma of the prostate; LG, low grade.

Acinar IDC-P exhibited enrichment of translation-related programs and redox stress adaptation compared with AAC overall (which showed enrichment of TME interactions), crib AAC, and HG non-crib AAC (Table 2, comparisons 2.1–2.4; Fig. 3). Acinar IDC-P also upregulated SLIT/ROBO signaling relative to AAC overall and crib AAC. Pathway enrichment analysis further revealed reduced error-prone DNA repair capacity, specifically nonhomologous end-joining and base excision repair, and decreased epigenetic and chromatin regulatory activity in acinar IDC-P relative to crib AAC (Fig. 3B). Compared with LG AAC, acinar IDC-P showed enrichment of pathways related to epithelial differentiation-associated programs (Supplementary Fig. S1A).

Paired analyses between adjacent acinar IDC-P and AAC components within the same tumor yielded findings similar to those observed in the unpaired comparisons. No significant differentially expressed genes were identified between paired acinar IDC-P and crib AAC, suggesting a similar transcriptomic profile (n = 7; Fig. 4B). Consistent with the unpaired analyses, paired acinar IDC-P was enriched for genes associated with luminal epithelial differentiation and androgen signaling, including TMPRSS2, VIPR1, SCUBE2, and RDH11, relative to both HG non-crib AAC and LG AAC (Fig. 4C, Supplementary Fig. S1B).

Differential gene expression and pathway enrichment in paired adjacent acinar IDC-P versus AAC within the same tumor. Volcano plots showing differentially expressed genes (DEGs) between paired acinar IDC-P and all AAC (A), crib AAC (B), and HG non-crib AAC (C) within the same tumor. Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; crib, cribriform; IDC-P, intraductal carcinoma of the prostate.

Paired pathway-level trends were also concordant with those observed in the unpaired analyses (Table 2, comparisons 3.1–3.4; Fig. 4). Acinar IDC-P additionally showed statistically significant, reduced Notch signaling and PKN1-mediated activation of androgen receptor (AR) target genes (KLK2/KLK3) relative to crib AAC, although no significant single-gene differences were detected (n = 7; Fig. 4B). Compared with LG AAC, acinar IDC-P demonstrated enrichment of bioenergetic (mitochondria-related) and metabolic programs and DNA repair pathways (no specificity for a particular repair mechanism) in paired analyses (Supplementary Fig. S1B).

Comparison of Gene Expression Between AIP and AAC

Among all comparisons with AAC, AIP showed the fewest differentially expressed genes relative to crib AAC, mirroring the pattern observed in acinar IDC-P, with upregulation of 12 genes, including genes involved in biosynthetic/metabolic programs (DHCR24, HPGD, TRPM8), redox and cellular stress adaptation (CREB3L1, GDF15, TGM2, DHCR24), growth factor signaling and cell survival (MDK), apoptosis-related signaling (TNFSF10), luminal epithelial differentiation (ANO7 and SLC45A3), lineage plasticity (SOX9), and ECM organization (COL9A2) (Fig. 5B). Compared with all AAC, AIP was enriched for genes associated with luminal epithelial differentiation and androgen signaling, including TRPM8, ANO7, and SLC45A3 (Fig. 5A). Similar findings were observed in comparisons with HG non-crib AAC, in which AIP was enriched for CRISP3, TRPM8, STEAP4, and KLK11 (Fig. 5C). Compared with LG AAC, AIP remained enriched for TRPM8 and additionally upregulated MDK, COL9A2, and GDF15, genes involved in growth factor signaling, ECM organization, and cellular stress responses, respectively (Supplementary Fig. S1C).

Differential gene expression and pathway enrichment in AIP versus AAC. Volcano plots showing differentially expressed genes (DEGs) between acinar AIP and all AAC (A), crib AAC (B), and HG non-crib AAC (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; AIP, acinar intraductal proliferation; HG, high grade; IDC-P, intraductal carcinoma of the prostate.

Similar to acinar IDC-P, AIP upregulated translation-driven programs, redox stress adaptation, and SLIT/ROBO signaling compared with AAC overall, crib AAC, and HG non-crib AAC (Table 2, comparisons 4.1–4.4; Fig. 5). AIP also demonstrated reduced GPCR signaling compared with crib AAC. In addition, similar to acinar IDC-P, AIP demonstrated enrichment of bioenergetic and metabolic programs when compared with LG AAC (Supplementary Fig. S1C).

Comparison of Gene Expression Between Acinar IDC-P and AIP

Only 4 genes were differentially expressed between acinar IDC-P and AIP, with upregulation of genes associated with epithelial membrane organization and plasticity (TSPAN8), microenvironmental pH regulation and metabolic adaptation (CA4), and downregulation of genes involved in calcium signaling (TRPM8) and cholesterol biosynthesis/oxidative stress response (DHCR24) in acinar IDC-P compared with AIP (Fig. 6A). Pathway enrichment analysis further showed greater upregulation of translation-related programs, redox stress adaptation, and SLIT/ROBO signaling in AIP, although both AIP and acinar IDC-P exhibited enrichment of these pathways relative to AAC. In addition, AIP demonstrated downregulation of GPCR and inflammatory signaling pathways compared with acinar IDC-P (Table 2, comparison 5).

Differential gene expression and pathway enrichment in acinar IDC-P versus AIP, and in acinar IDC-P/AIP as presumed invasive type versus putative precursor type. Volcano plots showing differentially expressed genes (DEGs) between acinar IDC-P and AIP (A), and between acinar IDC-P/AIP as presumed invasive type versus putative precursor type (B). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). Single-sample gene set enrichment analysis was performed to compare putative precursor-type IDC-P and AIP from one patient with presumed invasive-type from other patients. AAC, acinar adenocarcinoma; AIP, acinar intraductal proliferation; IDC-P, intraductal carcinoma of the prostate.

Comparison of Gene Expression Between Acinar IDC-P and AIP as Presumed Invasive Type versus Putative Precursor Type

The putative precursor-type acinar IDC-P/AIP showed upregulation of immune, inflammatory and cell death-related pathways and SLIT/ROBO signaling and downregulation of oncogenic signaling (VEGF-VEGFR, MET-PI3K-AKT, WNT, and p38-MAPK) and stemness-associated programs (POU5F1 [OCT4], SOX2, and NANOG), compared with presumed invasive-type (Table 2, comparison 6; Fig. 6B).

Comparison of Gene Expression Between DAC and AAC

DAC was transcriptionally more similar to cribriform AAC than to non-cribriform AAC (Fig. 7). Compared with crib AAC, DAC demonstrated higher expression of the oncogene MYC together with genes involved in cellular metabolism and associated with luminal epithelial differentiation and androgen signaling, including FABP5, ANPEP, AMACR, and SLC45A3, whereas crib AAC was enriched for a distinct set of luminal differentiation-associated and androgen-regulated genes, including PLA2G2A, FOLH1 (PSMA), SPON2, and STEAP4 (Fig. 7A). Compared with both HG non-crib AAC and LG AAC, DAC was enriched for genes associated with luminal epithelial differentiation and androgen signaling, including CRISP3, TFF3, DPP4, ANPEP, and TMEFF2 (Fig. 7B, C). In addition, compared with HG non-crib AAC, DAC demonstrated enrichment of the oncogene MYC and genes involved in ribosome biogenesis and protein translation (RPL8, RPL28, and RPL36A) (Fig. 7B). On pathway analysis, DAC exhibited enhanced translation-related programs and redox stress adaptation compared with crib AAC, HG non-crib AAC, and LG AAC (Table 2, comparisons 7.1–7.3; Fig. 7). DAC also demonstrated upregulation of antigen presentation pathways and reduced immune evasion compared with crib AAC.

Differential gene expression and pathway enrichment in DAC versus AAC. Volcano plots showing differentially expressed genes (DEGs) between DAC and crib AAC (A), HG non-crib AAC (B), and LG AAC (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; crib, cribriform; DAC, ductal adenocarcinoma; HG, high grade; LG, low grade.

Paired comparisons between adjacent DAC and AAC components within the same tumor revealed upregulation of MYC, RPL39 and CPNE7 in DAC, consistent with enhanced proliferative, translational, and epithelial signaling programs, along with similar pathway-level trends (Table 2, comparisons 8.1–8.3; Fig. 8A). In these paired analyses, DAC additionally showed downregulation of TME interactions and GPCR, PDGF, and RTK signaling relative to adjacent AAC. Paired comparison between DAC and crib AAC was not performed due to the presence of only a single such pair.

Differential gene expression and pathway enrichment in paired adjacent DAC versus AAC within the same tumor and in DAC versus ductal IDC-P. Volcano plots showing differentially expressed genes (DEGs) between paired adjacent DAC and AAC within the same tumor (A), DAC and ductal IDC-P (B), and paired adjacent DAC and ductal IDC-P within the same tumor (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10(adjusted p value). AAC, acinar adenocarcinoma; DAC, ductal adenocarcinoma; IDC-P, intraductal carcinoma of the prostate.

Comparison of Gene Expression Between DAC and Ductal IDC-P

Compared with ductal IDC-P, DAC demonstrated higher expression of genes associated with luminal epithelial differentiation and androgen signaling, including NPY, TMEFF2, EGR1, and SPON2 (Fig. 8B). In contrast, ductal IDC-P showed enrichment of metallothionein genes (MT1G, MT1M, and MT1A), together with genes involved in antigen presentation and immune signaling, including HLA-DRA, TAP1, and B2M. No significantly differentially expressed genes were identified in the paired comparison between adjacent DAC and ductal IDC-P within the same tumor (n = 3; Fig. 8C).

Ductal IDC-P showed downregulation of translation-related programs and redox stress adaptation, while exhibiting upregulation of antigen presentation and immune pathways, including checkpoint and T-cell signaling, compared with DAC (Table 2, comparison 9.1; Fig. 8B). Paired comparisons between adjacent DAC and ductal IDC-P demonstrated statistically significant, similar pathway-level trends, although no statistically significant differences in individual gene expression were detected (Table 2, comparison 9.2; Fig. 8C).

Discussion

We compared gene expression across histologically overlapping components of aggressive PCa using ST, including crib AAC, IDC-P, AIP, and DAC. Notably, Crib AAC, acinar IDC-P, and AIP exhibited substantial transcriptomic similarity, although distinct transcriptional differences were also identified. The shared luminal epithelial transcriptional phenotype suggests that these entities retain a common differentiation state and may represent related manifestations of aggressive PCa biology. These findings may improve our understanding of the biology of these aggressive components and help inform several ongoing controversies in prostate pathology. Although further investigation and validation in independent cohorts are required, our findings provide biological support for the recent proposal from the GUPS and ISUP Joint Expert Consultation Group that IDC-P associated with invasive PCa should be incorporated into Gleason grading. In addition, AIP (“AIP, suspicious for IDC-P”) has increasingly been recognized to be associated with adverse pathological findings.⁸ The transcriptomic similarity between acinar IDC-P and AIP support the hypothesis that these lesions might represent a shared biological process, differing in the degree of epithelial proliferation and ductal involvement, while exhibiting subtle differences in pathways related to redox stress adaptation and metabolic responses observed in our study. The observed transcriptomic resemblance among Crib AAC, acinar IDC-P and AIP may reflect several non-mutually exclusive mechanisms, including shared clonal lineage, intraductal spread, phenotypic convergence, and/or influences of the surrounding microenvironment. Although our findings indicate molecular relatedness and overlapping transcriptional programs among cribriform AAC, acinar IDC-P, and AIP, the precise biological relationships and evolutionary trajectories underlying these patterns require further investigation.

We demonstrated that DAC was transcriptomically more similar to crib AAC than to non-crib AAC. DAC can exhibit cribriform architecture and may show histological overlap with crib AAC, although DAC typically comprises tall columnar pseudostratified cells and frequently demonstrates papillary architecture. DAC and crib AAC share aggressive clinical behavior,²⁰ and both are assigned Gleason pattern 4. Although previous studies have compared DAC with AAC in general, conflicting data have been reported regarding the clonal relationship between DAC and AAC,^11–13 and studies directly comparing DAC and crib AAC are scarce. Interestingly, DAC with cribriform architecture has been shown to be associated with more adverse prognostic features than non-cribriform DAC.²¹ Therefore, the greater transcriptomic similarity between DAC and crib AAC than between DAC and non-crib AAC may provide biological support for their shared aggressive clinical behavior, although further investigation is required to elucidate the precise biological relationship between these entities.

Acinar IDC-P has recently been proposed by the GUPS and ISUP Joint Expert Consultation group to be classified as “invasive type” and “precursor type.”⁵ Although analogous terminology has not been formally discussed for AIP, it may share a similar etiology with acinar IDC-P, and extending the invasive-type and precursor-type classification to AIP might facilitate research and academic discussion. In the current study, the putative precursor-type acinar IDC-P/AIP showed upregulation of immune, inflammatory, and cell death-related pathways, and downregulation of oncogenic signaling and stemness-associated programs compared with the presumed invasive type, suggesting of the greater proliferative and aggressive nature of invasive-type IDC-P/AIP. Given the rarity of putative precursor-type acinar IDC-P/AIP, multicenter collaborations will be essential to determine whether these differences represent reproducible biological distinctions.

Ductal IDC-P is less well characterized than acinar IDC-P and may represent intraductal spread of DAC.^5,7 The diagnostic criteria for ductal IDC-P remain to be defined, and the criteria established for acinar IDC-P may not be directly applicable. In our study, ductal IDC-P was defined as an intraductal component morphologically resembling DAC and retaining a basal cell layer. Some invasive DAC cases may lack significant cellular proliferation within the lumen (Fig. 1I–J). Introducing a category of “ductal AIP,” based on a loose growth pattern within the luminal space (analogous to acinar AIP), does not appear appropriate. Furthermore, the prognostic significance of introducing a “ductal AIP” category remains unclear. Therefore, based on the current knowledge, we do not propose introducing such a category. Additionally, it remains unclear whether an analogous precursor-type ductal IDC-P exists. Ductal IDC-P demonstrated downregulation of luminal epithelial and translation-related programs and upregulation of antigen presentation and immune signaling pathways relative to DAC, suggesting that DAC may retain a stronger luminal lineage identity and proliferative state and exhibit greater immune evasion than ductal IDC-P.

We revealed that crib AAC was enriched for Notch signaling compared with non-crib AAC and acinar IDC-P. In PCa, increased NOTCH1 expression may confer a survival advantage, and levels of Notch family members have been reported to increase with higher tumor grade.²² Notch signaling has been implicated in lineage plasticity and castration resistance (CR) in PCa.²³ Increased Notch signaling has also been reported in a study combining crib AAC and acinar IDC-P with cribriform morphology, where it was associated with TME interactions and angiogenesis.²⁴ Nevertheless, the specific role of Notch signaling in crib AAC warrants further investigation.

More than 20% of metastatic PCa harbor DNA damage response (DDR) defects, predominantly homologous recombination deficiency (HRD) involving genes such as BRCA1/2 and ATM, which are associated with aggressive clinical behavior and sensitivity to poly(ADP-ribose) polymerase (PARP) inhibitors.^25,26 Crib AAC showed enrichment of homologous recombination-mediated DDR programs relative to LG AAC in our study, potentially reflecting increased genomic instability and heightened demand for high-fidelity DNA repair. In contrast, compared with acinar IDC-P, crib AAC demonstrated enrichment of nonhomologous end joining and base excision repair programs without significant differences in homologous recombination DDR pathways, likely suggesting greater reliance on alternative, often error-prone mechanisms to resolve double-strand breaks and oxidative DNA lesions. Approximately 70% of crib AAC and acinar IDC-P with cribriform morphology harbor DDR defects,²⁷ which may indicate overwhelmed or dysfunctional repair capacity despite transcriptional activation of DDR pathways. HRD has been shown to force cells to rely on error-prone repair mechanisms, promoting accumulation of genomic damage and tumor evolution.²⁸ These findings raise the possibility that crib AAC may undergo greater genomic evolution than acinar IDC-P, although further studies are needed. In addition, upregulated GPCR (including aminergic) signaling has been implicated in AR activation, androgen-independent tumor growth, and CR in PCa.^29–31 SREBP-mediated cholesterol and lipid metabolic programs are frequently elevated in PCa and have been linked to tumor stemness, invasion, and CR through mechanisms including intratumoral androgen synthesis and oncogenic pathways such as c-MYC activation.^32,33 Pharmacologic inhibition of SREBP has recently demonstrated significant antitumor efficacy in PCa cell lines.³³ Nevertheless, the upregulation of GPCR signaling and SREBP-mediated lipid programs in crib AAC relative to LG AAC, as well as the upregulation of GPCR signaling in crib AAC and acinar IDC-P relative to AIP, may represent transcriptional programs associated with more aggressive PCa biology.

The inflammatory signaling was more activated in acinar IDC-P than in AIP, potentially reflecting a metabolically stressed or resource-limited state in acinar IDC-P. This may relate to its more crowded architecture and higher cellular density, which may contribute to nutrient depletion and hypoxia and correlate with the comedonecrosis observed in some cases. Both acinar IDC-P and AIP showed enrichment of SLIT/ROBO signaling relative to AAC overall, with greater upregulation in AIP than in acinar IDC-P, and in the putative precursor-type acinar IDC-P/AIP compared with the presumed invasive-type. In PCa, SLIT expression has been reported as both upregulated and downregulated across studies.³⁴ Recent data indicate that upregulation of SLIT1, SLIT3, ROBO3, and ROBO4 is associated with advanced-stage PCa, whereas SLIT2 expression has been linked to more favorable outcomes.³⁵ However, higher SLIT2 and ROBO1 expression has also been associated with disease progression in PCa, whereas ROBO1 has been reported to inhibit tumor migration and invasion through modulation of the TGF-β/SMAD signaling pathway.³⁶ Collectively, these findings highlight the complex and context-dependent nature of SLIT/ROBO signaling in PCa that warrants further investigation.

Compared with crib AAC, acinar IDC-P demonstrated upregulation of GADD45G and ERRFI1, and downregulation of CCN3 and TMEFF2. Repression of GADD45G mediated by NF-κB has been shown to represent a critical step for cancer cells to evade programmed cell death, including in PCa.³⁷ Upregulated CCN3 expression has been correlated with epithelial-mesenchymal transition (EMT)-like phenotypes through activation of FAK/AKT/HIF-1α signaling in PCa, contributing to invasive and metastatic behavior.³⁸ Accordingly, the relative downregulation of GADD45G and upregulation of CCN3 in crib AAC compared with acinar IDC-P may reflect enhanced survival capacity and increased invasive and metastatic potential of crib AAC. Knockdown of ERRFI1, a regulator of EGFR family signaling, enhances pro-survival AKT signaling through increased phosphorylation of AKT and its downstream targets in PCa cells, suggesting a tumor-suppressive role.³⁹ The relative downregulation of ERRFI1 in crib AAC compared with acinar IDC-P may therefore reflect a more pronounced pro-survival phenotype in crib AAC, although further investigation is warranted. The role of TMEFF2, an androgen-regulated gene associated with luminal epithelial differentiation, in PCa remains complex and context-dependent, with both tumor-suppressive and oncogenic functions described;⁴⁰ full-length TMEFF2 has been shown to sensitize PCa cells to apoptosis, whereas the shed ectodomain appears to exert opposing effects.⁴¹ The biological significance of TMEFF2 downregulation in acinar IDC-P relative to crib AAC warrants further mechanistic investigation.

Acinar IDC-P demonstrated reduced PKN1-mediated activation of AR target genes, including KLK2 and KLK3 (the latter encoding prostate-specific antigen [PSA]), relative to crib AAC, in addition to the previously discussed downregulation of error-prone DDR programs and Notch signaling. PKN1 has been shown to regulate the function of androgen-responsive serum response factor (SRF), and its overexpression is associated with CR, and shorter survival in animal models of PCa.⁴² Although dysregulation of AR-interacting proteins has been reported more frequently in PCa with cribriform morphology and may contribute to pathogenesis,⁴³ PKN1-mediated AR signaling has not been described in crib AAC and warrants further investigation. Compared with LG AAC, acinar IDC-P showed enrichment of pathways related to bioenergetic and metabolic programs, DDR activity, and epithelial differentiation, likely reflecting high proliferative demand, stress-adaptive responses, increased DNA damage burden, and potential lineage plasticity.

DAC showed upregulation of MYC, RPL39, and CPNE7, with additional downregulation of TME interactions, as well as GPCR, PDGF and RTK signaling relative to adjacent AAC. MYC overexpression has been shown to antagonize the canonical AR transcriptional program and to contribute to PCa initiation, progression, and CR.⁴⁴ RPL39 is an oncogenic ribosomal protein in a subset of breast cancer, where it promotes stemness, chemoresistance, metastasis, and poor survival.⁴⁵ Ribosomal proteins other than RPL39 (e.g., RPL22L1) have also been reported to promote PCa proliferation and invasion,⁴⁶ although the role of RPL39 in PCa remains incompletely understood. Higher CPNE7 expression has been associated with worse outcomes in PCa.⁴⁷ Although the specific roles of MYC, RPL39, and CPNE7 in DAC remain unclear, their upregulation is concordant with the increased translation-related programs identified on pathway analysis and may indicate a proliferative, high-risk molecular phenotype. In addition to mediating androgen-independent tumor growth in PCa,³⁰ GPCR signaling regulates TME adhesion and ECM remodeling, thereby influencing angiogenesis and tumor invasion.⁴⁸ PDGF signaling is implicated in PCa development and EMT and also plays a key role in TME crosstalk.⁴⁹ RTKs activate the PI3K/AKT and RAS/MAPK pathways, which are central to proliferation and survival across cancers.⁵⁰ RTK signaling additionally mediates TME interactions and contributes to tumor invasion and metastasis.⁵¹ The observed downregulation of GPCR, PDGF, and ERK signaling in DAC relative to AAC is concordant with reduced TME interactions in the current study, possibly reflecting relatively greater tumor-stroma crosstalk in AAC. Compared with crib AAC, DAC showed enrichment of antigen presentation pathways, likely suggesting relatively greater immune evasion in crib AAC.

This study has several limitations. The sample size is relatively small. In addition, putative precursor-type acinar IDC-P/AIP is exceedingly rare on RP specimens, and only one case was included in our cohort. Therefore, these results should be considered exploratory and hypothesis-generating, and the observed transcriptomic differences between precursor-type and invasive-type IDC-P/AIP require validation in larger, independent cohorts. TMA-based spatial transcriptomic approach has inherent limitations, particularly with respect to intratumoral heterogeneity. Although multiple histological components and, in selected cases, duplicate AOIs from different tumor regions were included, the use of TMAs may not fully capture the spatial and molecular heterogeneity present within entire tumor. Nevertheless, this approach allowed us to increase the sample size and include a broader spectrum of histological components within the constraints of available resources. We also agree that integration of transcriptomic findings with clinical outcomes would further enhance the clinical relevance of the study. However, given the relatively small cohort size, limited number of outcome events, and the exploratory nature of this study, we were underpowered to perform robust correlations between specific transcriptomic features and clinical endpoints such as metastasis or recurrence. Future studies with larger, clinically annotated cohorts will be necessary to investigate the prognostic significance of the identified transcriptomic programs. Nonetheless, using high-resolution ST, we identified overlapping and distinct transcriptomic features among IDC-P, AIP, crib AAC, and DAC that may improve understanding of these lesions and help clarify current controversies.

In conclusion, crib AAC, acinar IDC-P, and AIP demonstrated substantial transcriptomic similarity, although distinct transcriptional differences are present. In addition, DAC is transcriptomically more similar to crib AAC than to non-crib AAC. While further investigation is warranted, these findings may enhance understanding of the biology of these aggressive components and help clarify current controversies.

Supplementary Material

Supplementary Figure S1. Differential gene expression and pathway enrichment in acinar IDC-P versus LG AAC and in AIP versus LG AAC. Volcano plots showing differentially expressed genes (DEGs) between acinar IDC-P and LG AAC (A: overall; B: paired within the same tumor), and between AIP and LG AAC (C). Red points indicate significant DEGs (adjusted p < 0.05 and |log2FC| > 1). Heatmaps display batch-corrected expression of significant DEGs with hierarchical clustering. Reactome pathway enrichment analysis results are shown on the right. Positive values indicate relative enrichment in the indicated subtype, whereas negative values indicate relative enrichment in the comparator group. Bar plots display normalized enrichment scores (NES), with color intensity representing −log10 (adjusted p value). AAC, acinar adenocarcinoma; AIP, acinar intraductal proliferation; IDC-P, intraductal carcinoma of the prostate; LG, low grade.

NIHMS2208807-supplement-1.pdf^{(1MB, pdf)}

Funding

This research was supported by NIH-NCI 1P50CA272390 (Wu) and the Urologic Oncology Multidisciplinary Translational Research Fund (Wu, Dahl, Zhao, Saylor) at Massachusetts General Hospital, Harvard Medical School.

Footnotes

Declaration of Competing Interests

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Ethics Approval and Consent to Participate

Approval of this study was granted by the Massachusetts General Hospital (MGH) Institutional Review Board. Informed consent was obtained from subjects.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Data Availability Statement

The raw data used in the preparation of this manuscript are available from the authors upon reasonable request.

References

1.Shi Y, Wang H, Golijanin B et al. Ductal, intraductal, and cribriform carcinoma of the prostate: Molecular characteristics and clinical management. Urol Oncol. 2024;42(5):144–154. 10.1016/j.urolonc.2024.01.037 [DOI] [PubMed] [Google Scholar]
2.Zhou M High-grade prostatic intraepithelial neoplasia, PIN-like carcinoma, ductal carcinoma, and intraductal carcinoma of the prostate. Mod Pathol. 2018;31(S1):S71–79. 10.1038/modpathol.2017.138 [DOI] [PubMed] [Google Scholar]
3.van Leenders G, van der Kwast TH & Iczkowski KA The 2019 International Society of Urological Pathology Consensus Conference on Prostate Cancer Grading. Eur Urol. 2021;79(6):707–709. 10.1016/j.eururo.2020.08.004 [DOI] [PubMed] [Google Scholar]
4.Guo CC & Epstein JI Intraductal carcinoma of the prostate on needle biopsy: Histologic features and clinical significance. Mod Pathol. 2006;19(12):1528–1535. 10.1038/modpathol.3800702 [DOI] [PubMed] [Google Scholar]
5.Shah RB, Varma M, Zhou M et al. The Genitourinary Pathology Society and International Society of Urological Pathology Joint Expert Consultation Recommendations on intraductal carcinoma of the prostate. Histopathology. 2026;88(1):8–23. 10.1111/his.70046 [DOI] [PMC free article] [PubMed] [Google Scholar]
6.Khani F, Wobker SE, Hicks JL et al. Intraductal carcinoma of the prostate in the absence of high-grade invasive carcinoma represents a molecularly distinct type of in situ carcinoma enriched with oncogenic driver mutations. J Pathol. 2019;249(1):79–89. 10.1002/path.5283 [DOI] [PubMed] [Google Scholar]
7.Seipel AH, Wiklund F, Wiklund NP & Egevad L Histopathological features of ductal adenocarcinoma of the prostate in 1,051 radical prostatectomy specimens. Virchows Arch. 2013;462(4):429–436. 10.1007/s00428-013-1385-5 [DOI] [PubMed] [Google Scholar]
8.Bhattarai R, McKenney JK, Alaghehbandan R et al. Atypical Intraductal Proliferation in Prostate Needle Core Biopsy: Validation as a Marker of Unsampled Adverse Pathology in a Clinicopathologic Series of 142 New Patients. Am J Surg Pathol. 2025;49(5):515–522. 10.1097/PAS.0000000000002376 [DOI] [PubMed] [Google Scholar]
9.Bottcher R, Kweldam CF, Livingstone J et al. Cribriform and intraductal prostate cancer are associated with increased genomic instability and distinct genomic alterations. BMC Cancer. 2018;18(1):8. 10.1186/s12885-017-3976-z [DOI] [PMC free article] [PubMed] [Google Scholar]
10.Chua MLK, Lo W, Pintilie M et al. A Prostate Cancer “Nimbosus”: Genomic Instability and SChLAP1 Dysregulation Underpin Aggression of Intraductal and Cribriform Subpathologies. Eur Urol. 2017;72(5):665–674. 10.1016/j.eururo.2017.04.034 [DOI] [PubMed] [Google Scholar]
11.Lotan TL, Toubaji A, Albadine R et al. TMPRSS2-ERG gene fusions are infrequent in prostatic ductal adenocarcinomas. Mod Pathol. 2009;22(3):359–365. 10.1038/modpathol.2008.236 [DOI] [PMC free article] [PubMed] [Google Scholar]
12.Gillard M, Lack J, Pontier A et al. Integrative Genomic Analysis of Coincident Cancer Foci Implicates CTNNB1 and PTEN Alterations in Ductal Prostate Cancer. Eur Urol Focus. 2019;5(3):433–442. 10.1016/j.euf.2017.12.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
13.Sanati S, Watson MA, Salavaggione AL & Humphrey PA Gene expression profiles of ductal versus acinar adenocarcinoma of the prostate. Mod Pathol. 2009;22(10):1273–1279. 10.1038/modpathol.2009.103 [DOI] [PubMed] [Google Scholar]
14.Rao A, Barkley D, Franca GS & Yanai I Exploring tissue architecture using spatial transcriptomics. Nature. 2021;596(7871):211–220. 10.1038/s41586-021-03634-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
15.Hirz T, Mei S, Sarkar H et al. Dissecting the immune suppressive human prostate tumor microenvironment via integrated single-cell and spatial transcriptomic analyses. Nat Commun. 2023;14(1):663. 10.1038/s41467-023-36325-2 [DOI] [PMC free article] [PubMed] [Google Scholar]
16.van der Kwast TH, van Leenders GJ, Berney DM et al. ISUP Consensus Definition of Cribriform Pattern Prostate Cancer. Am J Surg Pathol. 2021;45(8):1118–1126. 10.1097/PAS.0000000000001728 [DOI] [PubMed] [Google Scholar]
17.Ritchie ME, Phipson B, Wu D et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47. 10.1093/nar/gkv007 [DOI] [PMC free article] [PubMed] [Google Scholar]
18.Love MI, Huber W & Anders S Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. 10.1186/s13059-014-0550-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
19.Yu G Thirteen years of clusterProfiler. Innovation (Camb). 2024;5(6):100722. 10.1016/j.xinn.2024.100722 [DOI] [PMC free article] [PubMed] [Google Scholar]
20.Marra G, van Leenders G, Zattoni F et al. Impact of Epithelial Histological Types, Subtypes, and Growth Patterns on Oncological Outcomes for Patients with Nonmetastatic Prostate Cancer Treated with Curative Intent: A Systematic Review. Eur Urol. 2023;84(1):65–85. 10.1016/j.eururo.2023.03.014 [DOI] [PubMed] [Google Scholar]
21.Au S, Villamil CF, Alaghehbandan R & Wang G Prostatic ductal adenocarcinoma with cribriform architecture has worse prognostic features than non-cribriform-type. Ann Diagn Pathol. 2019;39:59–62. 10.1016/j.anndiagpath.2019.02.007 [DOI] [PubMed] [Google Scholar]
22.Carvalho FL, Simons BW, Eberhart CG & Berman DM Notch signaling in prostate cancer: a moving target. Prostate. 2014;74(9):933–945. 10.1002/pros.22811 [DOI] [PMC free article] [PubMed] [Google Scholar]
23.Jiang Y, Cheng S, Li L et al. Notch as a Driver of Lineage Plasticity and Therapeutic Target in Enzalutamide-Resistant Prostate Cancer. bioRxiv. 2025. 10.1101/2025.05.26.656166 [DOI] [Google Scholar]
24.Wong HY, Sheng Q, Hesterberg AB et al. Single cell analysis of cribriform prostate cancer reveals cell intrinsic and tumor microenvironmental pathways of aggressive disease. Nat Commun. 2022;13(1):6036. 10.1038/s41467-022-33780-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
25.Piombino C, Pipitone S, Tonni E et al. Homologous Recombination Repair Deficiency in Metastatic Prostate Cancer: New Therapeutic Opportunities. Int J Mol Sci. 2024;25(9). 10.3390/ijms25094624 [DOI] [PMC free article] [PubMed] [Google Scholar]
26.Na R, Zheng SL, Han M et al. Germline Mutations in ATM and BRCA1/2 Distinguish Risk for Lethal and Indolent Prostate Cancer and are Associated with Early Age at Death. Eur Urol. 2017;71(5):740–747. 10.1016/j.eururo.2016.11.033 [DOI] [PMC free article] [PubMed] [Google Scholar]
27.Shah RB, Palsgrove DN, Desai NB et al. Enrichment of “Cribriform” morphologies (intraductal and cribriform adenocarcinoma) and genomic alterations in radiorecurrent prostate cancer. Mod Pathol. 2022;35(10):1468–1474. 10.1038/s41379-022-01093-9 [DOI] [PubMed] [Google Scholar]
28.Wagener-Ryczek S, Merkelbach-Bruse S & Siemanowski J Biomarkers for Homologous Recombination Deficiency in Cancer. J Pers Med. 2021;11(7). 10.3390/jpm11070612 [DOI] [PMC free article] [PubMed] [Google Scholar]
29.Daaka Y G proteins in cancer: the prostate cancer paradigm. Sci STKE. 2004;2004(216):re2. 10.1126/stke.2162004re2 [DOI] [PubMed] [Google Scholar]
30.Raj GV, Barki-Harrington L, Kue PF & Daaka Y Guanosine phosphate binding protein coupled receptors in prostate cancer: a review. J Urol. 2002;167(3):1458–1463. https://doi.org/ [PubMed] [Google Scholar]
31.Heidman LM, Peinetti N, Copello VA & Burnstein KL Exploiting Dependence of Castration-Resistant Prostate Cancer on the Arginine Vasopressin Signaling Axis by Repurposing Vaptans. Mol Cancer Res. 2022;20(8):1295–1304. 10.1158/1541-7786.MCR-21-0927 [DOI] [PMC free article] [PubMed] [Google Scholar]
32.Li X, Wu JB, Li Q et al. SREBP-2 promotes stem cell-like properties and metastasis by transcriptional activation of c-Myc in prostate cancer. Oncotarget. 2016;7(11):12869–12884. 10.18632/oncotarget.7331 [DOI] [PMC free article] [PubMed] [Google Scholar]
33.Wei G, Zhu H, Zhou Y et al. Single-cell sequencing revealed metabolic reprogramming and its transcription factor regulatory network in prostate cancer. Transl Oncol. 2024;44:101925. 10.1016/j.tranon.2024.101925 [DOI] [PMC free article] [PubMed] [Google Scholar]
34.Ballard MS & Hinck L A roundabout way to cancer. Adv Cancer Res. 2012;114:187–235. 10.1016/B978-0-12-386503-8.00005-3 [DOI] [PMC free article] [PubMed] [Google Scholar]
35.Santos NJ, Mosele FC, Barquilha CN et al. SLIT/ROBO Pathway and Prostate Cancer: Gene and Protein Expression and Their Prognostic Values. Int J Mol Sci. 2025;26(11). 10.3390/ijms26115265 [DOI] [PMC free article] [PubMed] [Google Scholar]
36.He YJ, Zhang D, Li LY et al. [Clinical significance of SLIT2/ROBO1 expression in prostate cancer]. Zhonghua Bing Li Xue Za Zhi. 2025;54(7):733–740. 10.3760/cma.j.cn112151-20250108-00023 [DOI] [PubMed] [Google Scholar]
37.Zerbini LF, Wang Y, Czibere A et al. NF-kappa B-mediated repression of growth arrest- and DNA-damage-inducible proteins 45alpha and gamma is essential for cancer cell survival. Proc Natl Acad Sci U S A. 2004;101(37):13618–13623. 10.1073/pnas.0402069101 [DOI] [PMC free article] [PubMed] [Google Scholar]
38.Chen PC, Tai HC, Lin TH et al. CCN3 promotes epithelial-mesenchymal transition in prostate cancer via FAK/Akt/HIF-1alpha-induced twist expression. Oncotarget. 2017;8(43):74506–74518. 10.18632/oncotarget.20171 [DOI] [PMC free article] [PubMed] [Google Scholar]
39.Schomann T, Mirzakhani K, Kallenbach J et al. Androgen-Induced MIG6 Regulates Phosphorylation of Retinoblastoma Protein and AKT to Counteract Non-Genomic AR Signaling in Prostate Cancer Cells. Biomolecules. 2022;12(8). 10.3390/biom12081048 [DOI] [PMC free article] [PubMed] [Google Scholar]
40.Masood M, Grimm S, El-Bahrawy M & Yague E TMEFF2: A Transmembrane Proteoglycan with Multifaceted Actions in Cancer and Disease. Cancers (Basel). 2020;12(12). 10.3390/cancers12123862 [DOI] [PMC free article] [PubMed] [Google Scholar]
41.Chen X, Overcash R, Green T et al. The tumor suppressor activity of the transmembrane protein with epidermal growth factor and two follistatin motifs 2 (TMEFF2) correlates with its ability to modulate sarcosine levels. J Biol Chem. 2011;286(18):16091–16100. 10.1074/jbc.M110.193805 [DOI] [PMC free article] [PubMed] [Google Scholar]
42.Venkadakrishnan VB, DePriest AD, Kumari S et al. Protein Kinase N1 control of androgen-responsive serum response factor action provides rationale for novel prostate cancer treatment strategy. Oncogene. 2019;38(23):4496–4511. 10.1038/s41388-019-0732-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
43.Chen HH, Fan P, Chang SW et al. NRIP/DCAF6 stabilizes the androgen receptor protein by displacing DDB2 from the CUL4A-DDB1 E3 ligase complex in prostate cancer. Oncotarget. 2017;8(13):21501–21515. 10.18632/oncotarget.15308 [DOI] [PMC free article] [PubMed] [Google Scholar]
44.Qiu X, Boufaied N, Hallal T et al. MYC drives aggressive prostate cancer by disrupting transcriptional pause release at androgen receptor targets. Nat Commun. 2022;13(1):2559. 10.1038/s41467-022-30257-z [DOI] [PMC free article] [PubMed] [Google Scholar]
45.Dave B, Gonzalez DD, Liu ZB et al. Role of RPL39 in Metaplastic Breast Cancer. J Natl Cancer Inst. 2017;109(6). 10.1093/jnci/djw292 [DOI] [PMC free article] [PubMed] [Google Scholar]
46.Yi X, Zhang C, Liu B et al. Ribosomal protein L22-like1 promotes prostate cancer progression by activating PI3K/Akt/mTOR signalling pathway. J Cell Mol Med. 2023;27(3):403–411. 10.1111/jcmm.17663 [DOI] [PMC free article] [PubMed] [Google Scholar]
47.Lian X, Zhou H & Liu S Identification and validation of the TRHDE-AS1/hsa-miR-449a/ADAMTS5 axis as a novel prognostic biomarker in prostate cancer. Biofactors. 2024;50(6):1251–1267. 10.1002/biof.2083 [DOI] [PubMed] [Google Scholar]
48.Gad AA & Balenga N The Emerging Role of Adhesion GPCRs in Cancer. ACS Pharmacol Transl Sci. 2020;3(1):29–42. 10.1021/acsptsci.9b00093 [DOI] [PMC free article] [PubMed] [Google Scholar]
49.Heldin CH, Lennartsson J & Westermark B Involvement of platelet-derived growth factor ligands and receptors in tumorigenesis. J Intern Med. 2018;283(1):16–44. 10.1111/joim.12690 [DOI] [PubMed] [Google Scholar]
50.Regad T Targeting RTK Signaling Pathways in Cancer. Cancers (Basel). 2015;7(3):1758–1784. 10.3390/cancers7030860 [DOI] [PMC free article] [PubMed] [Google Scholar]
51.Wu F, Yang J, Liu J et al. Signaling pathways in cancer-associated fibroblasts and targeted therapy for cancer. Signal Transduct Target Ther. 2021;6(1):218. 10.1038/s41392-021-00641-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

NIHMS2208807-supplement-1.pdf^{(1MB, pdf)}

Data Availability Statement

The raw data used in the preparation of this manuscript are available from the authors upon reasonable request.

[R1] 1.Shi Y, Wang H, Golijanin B et al. Ductal, intraductal, and cribriform carcinoma of the prostate: Molecular characteristics and clinical management. Urol Oncol. 2024;42(5):144–154. 10.1016/j.urolonc.2024.01.037 [DOI] [PubMed] [Google Scholar]

[R2] 2.Zhou M High-grade prostatic intraepithelial neoplasia, PIN-like carcinoma, ductal carcinoma, and intraductal carcinoma of the prostate. Mod Pathol. 2018;31(S1):S71–79. 10.1038/modpathol.2017.138 [DOI] [PubMed] [Google Scholar]

[R3] 3.van Leenders G, van der Kwast TH & Iczkowski KA The 2019 International Society of Urological Pathology Consensus Conference on Prostate Cancer Grading. Eur Urol. 2021;79(6):707–709. 10.1016/j.eururo.2020.08.004 [DOI] [PubMed] [Google Scholar]

[R4] 4.Guo CC & Epstein JI Intraductal carcinoma of the prostate on needle biopsy: Histologic features and clinical significance. Mod Pathol. 2006;19(12):1528–1535. 10.1038/modpathol.3800702 [DOI] [PubMed] [Google Scholar]

[R5] 5.Shah RB, Varma M, Zhou M et al. The Genitourinary Pathology Society and International Society of Urological Pathology Joint Expert Consultation Recommendations on intraductal carcinoma of the prostate. Histopathology. 2026;88(1):8–23. 10.1111/his.70046 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R6] 6.Khani F, Wobker SE, Hicks JL et al. Intraductal carcinoma of the prostate in the absence of high-grade invasive carcinoma represents a molecularly distinct type of in situ carcinoma enriched with oncogenic driver mutations. J Pathol. 2019;249(1):79–89. 10.1002/path.5283 [DOI] [PubMed] [Google Scholar]

[R7] 7.Seipel AH, Wiklund F, Wiklund NP & Egevad L Histopathological features of ductal adenocarcinoma of the prostate in 1,051 radical prostatectomy specimens. Virchows Arch. 2013;462(4):429–436. 10.1007/s00428-013-1385-5 [DOI] [PubMed] [Google Scholar]

[R8] 8.Bhattarai R, McKenney JK, Alaghehbandan R et al. Atypical Intraductal Proliferation in Prostate Needle Core Biopsy: Validation as a Marker of Unsampled Adverse Pathology in a Clinicopathologic Series of 142 New Patients. Am J Surg Pathol. 2025;49(5):515–522. 10.1097/PAS.0000000000002376 [DOI] [PubMed] [Google Scholar]

[R9] 9.Bottcher R, Kweldam CF, Livingstone J et al. Cribriform and intraductal prostate cancer are associated with increased genomic instability and distinct genomic alterations. BMC Cancer. 2018;18(1):8. 10.1186/s12885-017-3976-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[R10] 10.Chua MLK, Lo W, Pintilie M et al. A Prostate Cancer “Nimbosus”: Genomic Instability and SChLAP1 Dysregulation Underpin Aggression of Intraductal and Cribriform Subpathologies. Eur Urol. 2017;72(5):665–674. 10.1016/j.eururo.2017.04.034 [DOI] [PubMed] [Google Scholar]

[R11] 11.Lotan TL, Toubaji A, Albadine R et al. TMPRSS2-ERG gene fusions are infrequent in prostatic ductal adenocarcinomas. Mod Pathol. 2009;22(3):359–365. 10.1038/modpathol.2008.236 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R12] 12.Gillard M, Lack J, Pontier A et al. Integrative Genomic Analysis of Coincident Cancer Foci Implicates CTNNB1 and PTEN Alterations in Ductal Prostate Cancer. Eur Urol Focus. 2019;5(3):433–442. 10.1016/j.euf.2017.12.003 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R13] 13.Sanati S, Watson MA, Salavaggione AL & Humphrey PA Gene expression profiles of ductal versus acinar adenocarcinoma of the prostate. Mod Pathol. 2009;22(10):1273–1279. 10.1038/modpathol.2009.103 [DOI] [PubMed] [Google Scholar]

[R14] 14.Rao A, Barkley D, Franca GS & Yanai I Exploring tissue architecture using spatial transcriptomics. Nature. 2021;596(7871):211–220. 10.1038/s41586-021-03634-9 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R15] 15.Hirz T, Mei S, Sarkar H et al. Dissecting the immune suppressive human prostate tumor microenvironment via integrated single-cell and spatial transcriptomic analyses. Nat Commun. 2023;14(1):663. 10.1038/s41467-023-36325-2 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R16] 16.van der Kwast TH, van Leenders GJ, Berney DM et al. ISUP Consensus Definition of Cribriform Pattern Prostate Cancer. Am J Surg Pathol. 2021;45(8):1118–1126. 10.1097/PAS.0000000000001728 [DOI] [PubMed] [Google Scholar]

[R17] 17.Ritchie ME, Phipson B, Wu D et al. limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res. 2015;43(7):e47. 10.1093/nar/gkv007 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R18] 18.Love MI, Huber W & Anders S Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 2014;15(12):550. 10.1186/s13059-014-0550-8 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R19] 19.Yu G Thirteen years of clusterProfiler. Innovation (Camb). 2024;5(6):100722. 10.1016/j.xinn.2024.100722 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R20] 20.Marra G, van Leenders G, Zattoni F et al. Impact of Epithelial Histological Types, Subtypes, and Growth Patterns on Oncological Outcomes for Patients with Nonmetastatic Prostate Cancer Treated with Curative Intent: A Systematic Review. Eur Urol. 2023;84(1):65–85. 10.1016/j.eururo.2023.03.014 [DOI] [PubMed] [Google Scholar]

[R21] 21.Au S, Villamil CF, Alaghehbandan R & Wang G Prostatic ductal adenocarcinoma with cribriform architecture has worse prognostic features than non-cribriform-type. Ann Diagn Pathol. 2019;39:59–62. 10.1016/j.anndiagpath.2019.02.007 [DOI] [PubMed] [Google Scholar]

[R22] 22.Carvalho FL, Simons BW, Eberhart CG & Berman DM Notch signaling in prostate cancer: a moving target. Prostate. 2014;74(9):933–945. 10.1002/pros.22811 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R23] 23.Jiang Y, Cheng S, Li L et al. Notch as a Driver of Lineage Plasticity and Therapeutic Target in Enzalutamide-Resistant Prostate Cancer. bioRxiv. 2025. 10.1101/2025.05.26.656166 [DOI] [Google Scholar]

[R24] 24.Wong HY, Sheng Q, Hesterberg AB et al. Single cell analysis of cribriform prostate cancer reveals cell intrinsic and tumor microenvironmental pathways of aggressive disease. Nat Commun. 2022;13(1):6036. 10.1038/s41467-022-33780-1 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R25] 25.Piombino C, Pipitone S, Tonni E et al. Homologous Recombination Repair Deficiency in Metastatic Prostate Cancer: New Therapeutic Opportunities. Int J Mol Sci. 2024;25(9). 10.3390/ijms25094624 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R26] 26.Na R, Zheng SL, Han M et al. Germline Mutations in ATM and BRCA1/2 Distinguish Risk for Lethal and Indolent Prostate Cancer and are Associated with Early Age at Death. Eur Urol. 2017;71(5):740–747. 10.1016/j.eururo.2016.11.033 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R27] 27.Shah RB, Palsgrove DN, Desai NB et al. Enrichment of “Cribriform” morphologies (intraductal and cribriform adenocarcinoma) and genomic alterations in radiorecurrent prostate cancer. Mod Pathol. 2022;35(10):1468–1474. 10.1038/s41379-022-01093-9 [DOI] [PubMed] [Google Scholar]

[R28] 28.Wagener-Ryczek S, Merkelbach-Bruse S & Siemanowski J Biomarkers for Homologous Recombination Deficiency in Cancer. J Pers Med. 2021;11(7). 10.3390/jpm11070612 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R29] 29.Daaka Y G proteins in cancer: the prostate cancer paradigm. Sci STKE. 2004;2004(216):re2. 10.1126/stke.2162004re2 [DOI] [PubMed] [Google Scholar]

[R30] 30.Raj GV, Barki-Harrington L, Kue PF & Daaka Y Guanosine phosphate binding protein coupled receptors in prostate cancer: a review. J Urol. 2002;167(3):1458–1463. https://doi.org/ [PubMed] [Google Scholar]

[R31] 31.Heidman LM, Peinetti N, Copello VA & Burnstein KL Exploiting Dependence of Castration-Resistant Prostate Cancer on the Arginine Vasopressin Signaling Axis by Repurposing Vaptans. Mol Cancer Res. 2022;20(8):1295–1304. 10.1158/1541-7786.MCR-21-0927 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R32] 32.Li X, Wu JB, Li Q et al. SREBP-2 promotes stem cell-like properties and metastasis by transcriptional activation of c-Myc in prostate cancer. Oncotarget. 2016;7(11):12869–12884. 10.18632/oncotarget.7331 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R33] 33.Wei G, Zhu H, Zhou Y et al. Single-cell sequencing revealed metabolic reprogramming and its transcription factor regulatory network in prostate cancer. Transl Oncol. 2024;44:101925. 10.1016/j.tranon.2024.101925 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R34] 34.Ballard MS & Hinck L A roundabout way to cancer. Adv Cancer Res. 2012;114:187–235. 10.1016/B978-0-12-386503-8.00005-3 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R35] 35.Santos NJ, Mosele FC, Barquilha CN et al. SLIT/ROBO Pathway and Prostate Cancer: Gene and Protein Expression and Their Prognostic Values. Int J Mol Sci. 2025;26(11). 10.3390/ijms26115265 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R36] 36.He YJ, Zhang D, Li LY et al. [Clinical significance of SLIT2/ROBO1 expression in prostate cancer]. Zhonghua Bing Li Xue Za Zhi. 2025;54(7):733–740. 10.3760/cma.j.cn112151-20250108-00023 [DOI] [PubMed] [Google Scholar]

[R37] 37.Zerbini LF, Wang Y, Czibere A et al. NF-kappa B-mediated repression of growth arrest- and DNA-damage-inducible proteins 45alpha and gamma is essential for cancer cell survival. Proc Natl Acad Sci U S A. 2004;101(37):13618–13623. 10.1073/pnas.0402069101 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R38] 38.Chen PC, Tai HC, Lin TH et al. CCN3 promotes epithelial-mesenchymal transition in prostate cancer via FAK/Akt/HIF-1alpha-induced twist expression. Oncotarget. 2017;8(43):74506–74518. 10.18632/oncotarget.20171 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R39] 39.Schomann T, Mirzakhani K, Kallenbach J et al. Androgen-Induced MIG6 Regulates Phosphorylation of Retinoblastoma Protein and AKT to Counteract Non-Genomic AR Signaling in Prostate Cancer Cells. Biomolecules. 2022;12(8). 10.3390/biom12081048 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R40] 40.Masood M, Grimm S, El-Bahrawy M & Yague E TMEFF2: A Transmembrane Proteoglycan with Multifaceted Actions in Cancer and Disease. Cancers (Basel). 2020;12(12). 10.3390/cancers12123862 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R41] 41.Chen X, Overcash R, Green T et al. The tumor suppressor activity of the transmembrane protein with epidermal growth factor and two follistatin motifs 2 (TMEFF2) correlates with its ability to modulate sarcosine levels. J Biol Chem. 2011;286(18):16091–16100. 10.1074/jbc.M110.193805 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R42] 42.Venkadakrishnan VB, DePriest AD, Kumari S et al. Protein Kinase N1 control of androgen-responsive serum response factor action provides rationale for novel prostate cancer treatment strategy. Oncogene. 2019;38(23):4496–4511. 10.1038/s41388-019-0732-7 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R43] 43.Chen HH, Fan P, Chang SW et al. NRIP/DCAF6 stabilizes the androgen receptor protein by displacing DDB2 from the CUL4A-DDB1 E3 ligase complex in prostate cancer. Oncotarget. 2017;8(13):21501–21515. 10.18632/oncotarget.15308 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R44] 44.Qiu X, Boufaied N, Hallal T et al. MYC drives aggressive prostate cancer by disrupting transcriptional pause release at androgen receptor targets. Nat Commun. 2022;13(1):2559. 10.1038/s41467-022-30257-z [DOI] [PMC free article] [PubMed] [Google Scholar]

[R45] 45.Dave B, Gonzalez DD, Liu ZB et al. Role of RPL39 in Metaplastic Breast Cancer. J Natl Cancer Inst. 2017;109(6). 10.1093/jnci/djw292 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R46] 46.Yi X, Zhang C, Liu B et al. Ribosomal protein L22-like1 promotes prostate cancer progression by activating PI3K/Akt/mTOR signalling pathway. J Cell Mol Med. 2023;27(3):403–411. 10.1111/jcmm.17663 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R47] 47.Lian X, Zhou H & Liu S Identification and validation of the TRHDE-AS1/hsa-miR-449a/ADAMTS5 axis as a novel prognostic biomarker in prostate cancer. Biofactors. 2024;50(6):1251–1267. 10.1002/biof.2083 [DOI] [PubMed] [Google Scholar]

[R48] 48.Gad AA & Balenga N The Emerging Role of Adhesion GPCRs in Cancer. ACS Pharmacol Transl Sci. 2020;3(1):29–42. 10.1021/acsptsci.9b00093 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R49] 49.Heldin CH, Lennartsson J & Westermark B Involvement of platelet-derived growth factor ligands and receptors in tumorigenesis. J Intern Med. 2018;283(1):16–44. 10.1111/joim.12690 [DOI] [PubMed] [Google Scholar]

[R50] 50.Regad T Targeting RTK Signaling Pathways in Cancer. Cancers (Basel). 2015;7(3):1758–1784. 10.3390/cancers7030860 [DOI] [PMC free article] [PubMed] [Google Scholar]

[R51] 51.Wu F, Yang J, Liu J et al. Signaling pathways in cancer-associated fibroblasts and targeted therapy for cancer. Signal Transduct Target Ther. 2021;6(1):218. 10.1038/s41392-021-00641-0 [DOI] [PMC free article] [PubMed] [Google Scholar]

PERMALINK

Comprehensive Spatial Transcriptomic Profiling of Cribriform, Intraductal, Atypical Intraductal Proliferation, and Ductal Adenocarcinoma of the Prostate

Ting Zhao, MD

Cole Nawrocki

Linjie Xiong, MD, PhD

Linda T Nieman, PhD

Todd E Abbott, MD

Philip J Saylor, MD

Francis J McGovern, MD

David T Miyamoto, MD, PhD

David T Ting, MD

Douglas M Dahl, MD

Chin-Lee Wu, MD, PhD

Abstract

Purpose:

Materials and Methods:

Results:

Conclusions:

Introduction

Materials and Methods

Histological Review and Specimen Acquisition

GeoMx® Digital Spatial Profiler (DSP)

Table 1.

Figure 1.

GeoMx DSP Computational Methods

Results

Clinicopathological Features

Overview of Spatial Transcriptomic Comparisons

Comparison of Gene Expression Between Crib AAC and Other AAC

Figure 2.

Table 2.

Comparison of Gene Expression Between Acinar IDC-P and AAC (crib and non-crib)

Figure 3.

Figure 4.

Comparison of Gene Expression Between AIP and AAC

Figure 5.

Comparison of Gene Expression Between Acinar IDC-P and AIP

Figure 6.

Comparison of Gene Expression Between Acinar IDC-P and AIP as Presumed Invasive Type versus Putative Precursor Type

Comparison of Gene Expression Between DAC and AAC

Figure 7.

Figure 8.

Comparison of Gene Expression Between DAC and Ductal IDC-P

Discussion

Supplementary Material

Funding

Footnotes

Data Availability Statement

References

Associated Data

Supplementary Materials

Data Availability Statement

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases

GeoMx^® Digital Spatial Profiler (DSP)