Integrative multi-region molecular profiling of primary prostate cancer with synchronous lymph node metastasis to characterize the biologically dominant clone

doi:10.21203/rs.3.rs-2786222/v1

Download PDF

Brief Communication

Integrative multi-region molecular profiling of primary prostate cancer with synchronous lymph node metastasis to characterize the biologically dominant clone

https://doi.org/10.21203/rs.3.rs-2786222/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 21 May, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Localized prostate cancer is composed of multiple spatially distinct tumors with significant inter- and intra-tumoral molecular heterogeneity. This genomic diversity gives rise to many competing subclones that may drive the biological trajectory of the disease. Previous large scale sequencing efforts have focused on the evolutionary process of metastatic prostate cancer, revealing a potential clonal progression to castration resistance. However, the clonal origin of synchronous lymph node (LN) metastases in primary disease is still unknown. Here, we performed multi-region, targeted DNA/RNA next generation sequencing (NGS) and constructed phylogenetic trees from 14 patients with LN metastasis (88 primary prostate cancer foci with 23 synchronous LN metastases) to better define the molecular features of primary disease most likely to spread to the LNs. Of eight primary prostate cancer cases with evidence of extra-prostatic extension (EPE), phylogenetic analysis supported this region as the likely source of LN metastasis in four cases. In two patients with organ-confined disease and LN metastasis, sub-clonal seeding and clonal evolution was observed, with LN metastasis likely arising from a Gleason Grade Group 5 focus. Cribriform pattern was observed in seven patients in both LNs and the primary tumor foci most clonally related. Driver alterations, either oncogenic gene fusions or somatic mutations (e.g., CDK12, FOXA1), were shared among primary tumor and LN metastatic foci. Collectively, we found that a combination of histopathologic and molecular factors, including tumor grade, EPE, cellular morphology (e.g., cribriform pattern), and oncogenic genomic alterations were associated with synchronous LN metastasis. More work is needed to better define the molecular features of primary prostate cancer foci most likely to give rise to metastasis to improve risk stratification, guide treatment allocation, and inform novel therapeutic strategies.

Biological sciences/Cancer/Tumour heterogeneity

Biological sciences/Cancer/Urological cancer/Prostate cancer

Prostate cancer, the most commonly diagnosed solid organ malignancy and the second leading cause of cancer-related death among men in the developed world, is a heterogenous disease¹. The clinical trajectory and response to therapy at all stages is variable, and the known molecular drivers of prostate cancer are generally myriad^2–4. In primary, localized prostate cancer, where multifocality is common, genomic heterogeneity is frequent^4–7. However, while both intra- and inter-tumoral genomic heterogeneity have been linked to aggressive biologic behavior, the identification of the biologically dominant clone that drives aggressive disease has remained elusive^5,6,8,9. For a given patient with primary multifocal prostate cancer, knowledge of which clone (or clones) is most likely to give rise to metastasis has the potential to inform management, including interpretation of tissue-based prognostic tests, selection of precision treatment approaches, and development of novel therapeutic strategies.

Historically, the “index tumor” has been defined as the tumor with the largest volume and presumed to drive prognosis¹⁰. However, in a proportion of cases, the largest tumor focus is discordant with the highest grade tumor or the tumor associated with adverse pathological characteristics, such as extraprostatic extension (EPE) or seminal vesicle invasion (SVI)¹¹. Additionally, in prostate specimens where the largest focus shares the same high Gleason sum as other foci, it is unknown if the largest focus is molecularly unique and/or possesses the highest probability of spread. Further, a relatively low-grade region of an overall high-grade focus can periodically give rise to lethal clones¹². Thus, the concept of the biologically dominant clone in primary prostate cancer remains poorly defined. High-throughput genomic profiling with next-generation sequencing (NGS), however, has facilitated a better understanding of the molecular drivers of cancer dissemination as well as identified distinct tumor clones associated with metastasis in breast, colorectal, lung, liver, kidney, and hematologic cancers^13–18. In metastatic, castrate-resistant prostate cancer, the lethal clone largely arises from a single monoclonal precursor cell population, with continued seeding from a single, dominant lineage post-metastasis^19–21. In primary, multifocal disease, large-scale transcriptomic and genomic profiling efforts have revealed distinct expression profiles among separate foci and the presence of multiple genetically discrete cancer cell clones suggesting independent origins^{6–8, 22}. However, to date, comprehensive analyses of primary, multifocal disease with matched metastatic samples to delineate the primary lesions most likely to metastasize have been limited.

Here, we performed multi-region molecular profiling of primary prostate cancer and synchronous lymph node (LN) metastasis to better understand what defines the biologically dominant clone in this setting. By analyzing both primary foci and matched LN metastases in multifocal disease, we attempt to build upon the existing foundation of knowledge regarding the heterogeneity of multifocal prostate cancer. An improved comprehension of the molecular underpinnings that drive LN metastasis in this setting would complement the existing breadth of data elucidating clonal evolution in metastatic, castrate-resistant disease and allow for improved risk stratification and treatment allocation in localized disease.

Mapping primary prostate cancer to synchronous LN metastasis. From a cohort of 14 patients who had radical prostatectomy and pelvic lymph node (LN) dissection for prostate cancer, we performed multi-region sampling from formalin fixed paraffin embedded (FFPE) blocks yielding 103 primary tumor and 28 LN metastasis samples (patient data is summarized in supplementary Table S1). We performed high depth, targeted, multiplex DNA sequencing to characterize the genomic profile of each tumor region using three targeted DNA NGS panels: the Comprehensive Cancer Panel (CCP; 409 genes and 15,992 amplicons); the Oncomine Comprehensive Panel (OCP,130 genes and 2,531 amplicons); and a custom Pan-GenitoUrinary (Pan-GU) cancer panel (135 genes, 3,127 amplicons). Targeted RNA NGS sequencing was also performed to evaluate the gene fusion status of each sample and derive relevant tissue-based prognostic scores (Myriad Prolaris™ Cell Cycle Progression (mxCCP) score, Oncotype DX™ Genomic Prostate Score (mxGPS), and Decipher™ Genomic Classifier (mxGC)) as previously performed⁷. We determined and compared histopathological characteristics, RNA tissue-based prognostic signatures, somatic DNA mutations, copy number alterations (CNA), and gene fusion status between primary and LN disease (Fig. 1).

After quality control (QC) steps as described in the methods, 10 patients (65 primary tumor and 16 LN metastatic samples) had sufficient data for phylogenetic analyses. All patients in the analytic cohort exhibited adverse pathological characteristics, including cribriform, single cell, or solid cell pattern in at least one tumor focus. A total of eight patients had EPE, one demonstrated SVI and four showed evidence of lymphovascular invasion (LVI). We noted substantial transcriptomic heterogeneity, including discordant derived prognostic gene signature scores both within lesions and between lesions in the same patient. We found recurrent CNAs for known prostate cancer relevant genes including CDK2NB (53%), MYC (16%), PTEN (35%), CDKN1B (36%), FOXA1 (23%), and TP53 (30%) (Fig. S1). Evidence of the TMPRSS2:ERG fusion was seen in four patients and the SLC45A3:ERG fusion was seen in one patient. Phylogenetic analyses was performed to computationally reconstruct the clonal evolution of prostate cancer and determine the likely clonal origin of LN metastasis for each patient (Fig. 2, S2-8)¹⁹. Our data reveal several possible histopathologic and molecular factors associated with disease spread as described below.

Spatial spread of cancer linked with EPE. In 8 patients with EPE of cancer, phylogenetic reconstruction supported the region of EPE as the likely source of the LN metastasis in four cases (Fig. 2a, S2,5,6). In patient #1 (Fig. 2a), while four primary tumor regions showed concordant TP53, IL6ST, and TPR mutations, only two primary tumor regions (P1 and P2, both EPE) also harbored an LRP1B mutation and high-level CNAs that were also present in the two LN (LN1 and LN2) metastasis foci. Notably, primary tumor regions P3 and P4, with presence of single cells, did not appear to have seeded the LN metastatic foci. In this patient, primary tumor regions P1 and P2 (GG5 regions with EPE) were most likely the source of LN metastasis. In patient #2 (Fig. S2), all tumor regions were ETS gene fusion negative. A CDK12 frameshift mutation with MYC and FGFR3 amplification was also detected in all regions. Phylogenetic analysis suggests P4 (a focus with EPE) most closely resembles the LN1 metastasis. In patient #34 (Fig. S5), BRCA1 and PTEN deletions were seen in P1, P4, P7, P8, P9, LN1, LN2, and LN3, but not in P2, P3, and P5, suggesting two different branches of clonal evolution. Regions P4 and P8 showed evidence of EPE. Phylogenetic analysis suggests that regions P1, P4, P7, P8, and P9 are most likely the source of LN metastasis in this patient. In patient #38 (Fig. S6), regions P3, P4, P5, P6, and P7 (a region of EPE) displayed FOXA1 mutations along with deletions in PTEN and RB1, similar to the LN foci (LN1). The area of P6 and LN1 shared additional deletions of CDKN2B and amplification of RECQL4, suggesting this as the most likely clonal source of the LN metastatic focus, though the region with EPE (P7) likely contributed given its shared mutational burden and proximity to P6 and LN1 by phylogenetic analysis.

Role of driver alterations and genomic complexity in prostate cancer metastasis to LNs. As shown in Fig. 1, recurrent mutations detected in our samples included FOXA1 (27%), TP53 (15%), CDK12 (7%), and SPOP (1%) across 9 patients. These mutations were shared between the nominated dominant primary tumor region and LN metastasis foci in 5/10 patients (Fig. 2a-c, S2,4–6). For example, in a patient with organ confined (pT2) prostate cancer and LN metastasis (patient #33, Fig. 2b), the samples demonstrated genomic complexity with FOXA1 and ATM frameshift deletions detected in some primary tumor and LN metastasis regions (P2, P3, P5, P6, LN1); an SPOP mutation was detected only in a GG1 primary tumor focus (P1); and RB1 deletion and MYC amplification were shared between the presumed dominant primary tumor regions and the LN metastasis focus (P2, P3, P6, LN1). In patient #41 (Fig. 2c), regions P1, P2 and LN1 all were TMPRSS2:ERG gene fusion positive and harbored PTEN loss. P1 most closely resembled LN1 based on phylogenetic analysis, suggesting this as the region that likely gave rise to the LN1 metastatic focus.

In patient #2 (Fig. S2), LN1 and P4 shared a CDKN2B deletion, along with a CDK12 frameshift mutation and MYC and FGFR3 amplification as discussed above. In patient #30 (Fig. S4), we observed shared loss of CDKN2A, ERCC2, and ERCC3 in lesions P1, P3, P4, P7 and LN2. LN2 had an additional mutation in TP53. Of all the tumor regions, P3 most resembled LN2 in mutational profile and genomic complexity, with phylogenetic analysis suggesting this as the likely clonal origin of LN2 metastasis. In patient #34 (Fig. S5), PTEN and BRCA1 deletions were observed among P1, P4, P7, P8, P9, and LN1, LN2, and LN3. In patient #38 (Fig. S6), P6 and LN1 both displayed CDKN2B deletions and RECQL4 amplifications as discussed above. Patient #4 (Fig. S3) showed significant genomic complexity and an additional hypermethylated panel was performed (data not shown) that displayed increased microsatellite instability, tumor mutational burden, and MUTYH expression in regions P1 and P2, suggesting these regions developed continued genomic aberrations after LN1 and LN2 metastasis (as suggested by earlier clonal branching of the LN metastasis on phylogenetic tree).

Taken together, these data support the possible role of driver genomic alterations in the development of synchronous LN metastasis in primary prostate cancer.

Lymph node metastatic homogeneity. We observed histologic, genomic, and transcriptomic heterogeneity across the primary prostate cancer foci (Fig. 1, Fig. S1). A total of five patients had multiple LN areas analyzed. These synchronous LN metastases were typically homogenous within a given patient, consistent with existing literature regarding metastatic foci^19,20. Notably, several histopathologic features were associated with LN metastasis, including GG and cribriform or single cell pattern (Fig. 1). For example, all 10 patients had at least one tumor region with cribriform pattern on histopathology, with shared cribriform pattern observed in both the dominant primary tumor region and LN metastasis in seven patients (70%). Interestingly, single cell pattern was detected in LN specimens in only two patients (20%) though present in the primary tumors of eight patients (80%). In patients with multiple LN foci analyzed, we noted mostly homogenous patterns of mutations, suggesting a shared clonal origin. For example, in patient #1 (Fig. 2a), both LN1 and LN2 shared chr8p deletions and chr8q amplifications, as well as mutations in TP53. In patient #4 (Fig. S3), LN1 and LN2 shared loss of CDKN2B, FANCA gain, and TP53 frameshift mutations. All three LN metastatic regions LN1, LN2, and LN3 in patient #34 (Fig. S5) shared PTEN, BRCA1, BRCA2, and CDKN2A deletions as well as MYC amplification.

Two of the five patients with multiple LN foci analyzed displayed discordant molecular profiles and thus separate branching in the phylogenetic analysis, including patient #2 (Fig. S2) with LN1 and LN2 showing discordant deletions in CDKN2B (LN1) and TP53 (LN2). In patient #30 (Fig. S4), LN2 shared features of CDKN2A, ERCC2, ERCC3 loss and TP53 mutations, whereas LN1 did not show these shared mutations. While low genomic complexity, early clonal branching or evolution, or low tumor content are possible explanations for these observations, the possibility of LN metastasis from separate tumor regions or clones cannot be excluded.

Here, we performed multi-region, multi-platform molecular characterization of primary prostate cancer to dissect the clonal origin of synchronous LN metastases. First, we observed intra- and inter-tumoral histologic and genomic heterogeneity of primary disease consistent with existing literature^{4–7, 14}. Notably, however, we found relative homogeneity between LN specimens within a given patient. Second, we demonstrated that while LN metastases are highly variable and related to a combination of histopathologic and genomic factors, phylogenetic analysis allows for nomination of the likely clonal origin of LN disease. Despite the overall molecular diversity seen within primary samples, there was no single identifiable factor in our analysis that was seen to be consistently associated with LN metastasis. Nevertheless, we found that a metastatic clone is frequently associated with EPE or cribriform pattern, suggesting that there are likely non-genomic, spatial, or anatomic determinants that contribute to LN metastasis. We also noted that the highest-grade region of the primary tumor does not always represent the metastatic clone, indicating that while histologic grade is generally associated with clinical outcomes, grade is not necessarily the sole determinant of metastatic potential. Lastly, we observed significant heterogeneity in derived RNA-based prognostic scores within each profiled tumor region in individual patients as well as across the entire cohort. This suggests that prediction of which areas of a tumor will metastasize a priori using tools such as tissue-based prognostic biomarkers will remain challenging due to tumor heterogeneity. Taken together, our findings suggest prostate cancer LN metastasis is a complex event, with genomic, anatomic, and histopathological determinants all likely playing a role.

Our results are consistent with prior studies demonstrating the genomic heterogeneity of clinically localized prostate cancer^5,6,9. These studies observed minimal shared CNAs and SNVs between disease foci, similar to our results which display a highly variable profile of CNAs among primary tumor samples. The relative molecular homogeneity of LN metastatic samples may represent an early step in monoclonal metastasis-to-metastasis seeding as has been previously demonstrated, or the emergence of seeding from a primary clone with metastatic potential¹⁹. Despite the molecular complexity of primary disease, most metastases in prostate cancer arise from a single precursor cell²⁰. It is likely that this homogeneity then gives rise to clonal diversification as mutations accumulate, driving eventual clones towards a path of metastatic spread and treatment resistance¹⁹. Our work supports the presence of LN metastasis as a potential initial step for the development of distant metastases along this existing paradigm. Further, previous whole-genome sequencing efforts to characterize the lethal clone in a man with prostate cancer-related death have shown that the lethal clone may arise from a low-grade region within a high-grade primary cancer foci¹². In our cohort, the biological clone likely giving rise to LN metastasis similarly did not always originate from the area of highest-grade disease. In at least three cases (patient #4, patient #34, patient #38) the primary tumor region that most resembled the LN metastatic foci was from a relatively lower grade region of the primary tumor. In a previous paired analysis of 12 radical prostatectomy specimens from men who subsequently developed metastatic, castrate-resistant prostate cancer, truncal events included SPOP mutations, TMPRSS2:ETS rearrangements, and TP53 mutations, which were all demonstrated to varying degrees in both primary and LN samples in our cohort²³. Further, deletions in PTEN, RB1, FANCA, and ATM were similarly detected within primary and LN samples in our study, suggesting the presence of these mutations may confer increased aggressiveness and a propensity for development of castration resistance.

There are several important clinical and research implications of the current study. First, the dilemma of reliably identifying the biologically dominant nodule presents a significant clinical challenge for risk stratification and treatment of primary prostate cancer. For example, as previously demonstrated by our group and others, tissue-based prognostic biomarkers may be of limited utility in determining the presence of unsampled aggressive disease^4,7. Critically, the basic tenet of focal therapy for prostate cancer relies on the capacity to identify the cancer focus with the greatest metastatic potential for targeted ablation. Without consistent identification of the lesion most likely to be biologically harmful, there remains the possibility that such approaches subject men to treatment without oncological benefit if the focus with the most aggressive potential is missed. Second, our data demonstrate the inherent limitation of using molecular profiling to determine the need for pelvic lymphadenectomy at the time of surgery or whole pelvis radiation with radiation therapy. As molecular profiling techniques continue to evolve and molecular data is increasingly available, it will be important to investigate how to better integrate clinicopathologic and genomic information into routine clinical practice paradigms. In our cohort, we observed LN metastasis in organ-confined disease in a patient with significant genomic complexity, with notable alterations in driver genes such as FOXA1, ATM, RB1 and MYC. Our work thus presents an initial step for the potential application of targeted molecular profiling to risk stratify patients with apparent organ-confined disease. An improved understanding of the molecular mechanisms underpinning LN metastasis in prostate cancer may facilitate the targeting of specific alterations or allow for tailored biomarkers that may lead to improvements in precision medicine.

The current study is not without limitations. First, we utilized targeted DNA and RNA sequencing using previously developed panels of genes with established relevance in prostate and other cancers. Whole-genome or exome and transcriptome sequencing with greater depth are needed to capture potential targets that were unidentifiable with our targeted approach as well as facilitate accurate estimation of tumor purity. Further, evolving single cell or spatial sequencing approaches may potentially provide greater insight into the existing heterogeneity of multifocal primary prostate cancer. Nevertheless, we were able to generate sufficient data to construct phylogenetic trees to determine the clonal evolution of primary tumor regions and nominate the area most likely to have given rise to LN metastases. Our approach was able to confirm the molecular heterogeneity present in primary multifocal prostate cancer and identified the relative molecular homogeneity of LN metastatic foci. Second, information regarding the laterality of LN metastasis in this cohort was not available, hence, conclusions regarding the propensity for lesions to spread in a unidirectional or bidirectional manner could not be ascertained. Lastly, it has been previously suggested that LN metastases are not necessarily an intermediate step for the development of distant metastasis, but a genetic dead-end in tumor phylogenies. Thus, future comparisons of paired LN metastasis with distant metastatic sites are needed, although access to such samples is limited. Despite these limitations, our comprehensive, multiregional, molecular profiling of primary prostate cancer and paired synchronous LN metastases sheds light on important histopathological factors and genomic alterations implicated in LN positive disease.

The metastatic potential of prostate cancer results from a constellation of genomic alterations, molecular changes, and histopathological features, including tumor morphology, architecture, and anatomic location. Although phenotypic or histopathologic features are driven by their underlying genomic profile, the promise of molecular medicine lies in the ability to provide prognostic and predictive information beyond existing clinicopathologic data. Additional studies with emerging, more sophisticated molecular tools in larger cohorts are needed to more precisely identify primary foci most likely to cause harm and characterize the molecular underpinnings of LN metastasis.

Patients and Specimens: With local Institutional Review Board approval (IRB #: HUM00042749) we assembled a retrospective, multi-institutional, non-consecutive cohort of patients with primary prostate cancer with LN metastases. First, from a prior transcriptomic analysis of samples chosen to represent the clinical continuum of prostate cancer by our group⁷, we identified six patients with synchronous LN metastases at the time of radical prostatectomy (cohort 1). Second, we assembled an additional non-consecutive cohort of twelve patients with large or multifocal prostate cancer with synchronous LN metastases from participating institutions (cohort 2). From a combined cohort of 18 patients (103 primary tumor and 28 LN metastasis samples), 10 patients (65 primary tumor and 16 LN metastasis samples) met experimental and analytic QC metrics as well as had sufficient quality data for phylogenetic analysis. Complete sample information is provided in Table S1. All profiled samples were evaluated by a board-certified Anatomic Pathologist (S.A.T., A.M.U., R.L.) with combined experience in prostate and molecular pathology who assigned a Gleason score, GG and outlined tumor regions within the specimens for molecular profiling.

Hematoxylin and Eosin (H&E), Immunohistochemistry (IHC), DNA/RNA isolation

H&E–stained sections and ERG IHC images (as needed) were reviewed by board-certified anatomic pathologists (A.M.U., R.L., S.A.T.) to outline regions for multiregion molecular profiling. Punch biopsies (4–6 samples) of each region were obtained from formalin-fixed, paraffin-embedded (FFPE) radical prostatectomy blocks. For LN specimens, ten 10-µm serial sections of LN FFPE specimens were obtained from which macrodissection of outlined regions was performed with microscope guidance at 4x or 10x as needed. DNA and RNA were then co-isolated from these samples using the Qiagen Allprep FFPE DNA/RNA Kit (Qiagen, Hilden, Germany) and the Qiagen QIAcube (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Extracted DNA and RNA samples were quantified using the Qubit 2.0 fluorometer (Life Technologies, California, United States).

Targeted, multiplex DNA NGS

DNA libraries were generated using the Ion AmpliSeq Library Kit 2.0 (Life Technologies, California, United States). Targeted multiplexed polymerase chain reaction (PCR) –based DNA NGS was performed using three different panels as previously described and listed below^7,24. Template preparation was performed using the Ion PI-Hi-Q Template OT2 200 Kit (Life Technologies) and the Ion OneTouch ES Instrument (Life Technologies, California, United States). NGS was then performed on Ion Proton P1 chips using the Ion P-HiQ Sequencing 200 Kit according to the manufacturer's instructions. The targeted DNAseq panels utilized include

CCP: This is a targeted, multiplexed PCR-based Ampliseq NGS panel which targets 1,688,650 bases from 15,992 amplicons representing the complete coding sequence of 409 cancer genes (http://tools.invitrogen.com/downloads/cms_103573.csv). All patients in cohort 1 were profiled using CCP.

OCP

This is a commercially available, targeted, multiplexed PCR-based NGS panel comprising 2,531 amplicons covering 260,717 bases targeting 130 genes (Thermo Fisher Scientific, Massachusetts, United States) as described. Early on in our experimental process (cohort 1), we performed and compared targeted DNA NGS using both CCP (targeting 409 genes) and OCP (targeting 130 genes) with no significant incremental yield with the CCP. Notably, the CCP does include FOXA1, SPOP, and BRCA1/2. Thus, OCP was used in cohort 2 and CCP was omitted.

Pan-GU panel

This is an in-house custom designed pan-genitourinary cancer panel comprised of 3,127 amplicons targeting 135 genes. All patients in cohorts 1 and 2 were profiled using the pan-GU cancer panel.

Somatic DNA alterations analyses

DNA NGS data analyses were performed using Ion Torrent suite with the variantCaller and coverageAnalysis plug-ins (Torrent Suite Software 5.12, ThermoFisher Inc). Annotated variants were filtered to remove poorly supported calls/sequencing artifacts, and germline alterations. High confidence somatic variants passing our previously described criteria were then visualized using the Integrated Genomics Viewer for read level confirmation²⁴. To trace clonality we retained synonymous, non-coding and untranslated regional variants.

Copy number alterations analyses

Each tumor sample was normalized, GC content corrected and then divided by a pool of normal male genomic DNA samples to determine a copy number ratio for each amplicon. Gene-level copy number estimates were derived by taking the coverage-weighted mean of the per-probe ratios, with expected error determined by the probe-to-probe variance. For significant genes (False Discovery Rate < 5%), a log2 copy number ratio estimate ≤ -0.81 was considered as deep deletion, ≥ 0.81 as amplification, (-0.81, -0.42] as shallow deletions and [0.32,0.81) as gains.

Targeted RNA NGS and scores derivation

cDNA libraries were generated using the Ion AmpliSeq Library Kit 2.0 (Life Technologies, California, United States). Targeted multiplexed PCR–based NGS was performed with a prostate cancer focused targeted RNAseq panel comprising 306 amplicons as previously described⁷. End to end read counts for RNA expression runs were calculated using Torrent Suite’s Coverage Analysis plugin v. 5.0.4. All further processing steps were performed as described⁷. Briefly, 4 of 5 Oncotype DX panel housekeeping genes (ATP5E, ARF1, CLTC1, and PGK1) with robust expression were used for normalization prior to downstream analyses. Data processing, sample filtering, normalization and QC steps were performed in the R project for Statistical Computing v. 3.4.3. Samples with at least 500,000 total mapped sequencing reads and at least 60% of all end-to-end reads were retained. We have previously described our sample exclusion criteria based on housekeeping gene expression and calculation of prognostic scores (mxCCP, mxGC, and mxGPS)⁷. A sample was classified as gene fusion positive based on RNA NGS data if the fusion partner count was ≥500 and log₂ normalized partner expression was ≥-6 (log₂(0·01)). For samples without sufficient RNA to determine gene fusion status, we used ERG expression on IHC classify them into fusion positive vs. negative. Notably, we have previously reported 100% concordance between RNA NGS- and ERG IHC-based approaches for determining gene fusion status²⁵.

Phylogenetic analysis

For each tumor sample within a patient, somatic alterations, significant copy number alterations and gene fusion status were considered as features for evolutionary analysis to determine clonality. We constructed distance matrices using the neighbor-joining algorithm by feeding the features as present or absent. Unrooted trees were derived from the resulting matrices in R package phangorn. Primary tumors branched with LN tumors were considered to have evolved from same parent clone during progression.

Funding: U.S. is supported by a Urology Care Foundation – American Urological Association Research Scholar Award sponsored by the Society of Urologic Oncology/Specialized Programs of Research Excellence (SPORE), a Prostate Cancer Foundation Young Investigator Award, a NIH Loan Repayment Program Award (L30 CA264387), and SPORE Career Enhancement Program Award. U.S. and J.J.T. are supported by the Scholars Awards grants program of Precision Health at the University of Michigan.

Supported in part by the Prostate Cancer Foundation (U.S., J.T.T., S.A.T., S.S.S., and T.M.M.) and the National Institutes of Health (U01 CA214170 to S.A.T., R01 CA183857 to S.A.T., the University of Michigan Prostate S.P.O.R.E., P50 CA186786-05: U.S., A.M.U., G.S.P., S.S.S., and T.M.M). J.J.T. has been supported by the SPORE Career Enhancement Program (CA186786), the Prostate Cancer Foundation Young Investigator Award (20YOUN11), and an NIH Early Detection Research Network Biomarker Characterization Center award (U2CCA271854).

Supported by the Men of Michigan Prostate Cancer Research Fund and the University of Michigan Comprehensive Cancer Center core grant (2-P30-CA-046592-24; S.S.S.). T.M.M. and S.A.T. have been supported by the A. Alfred Taubman Biomedical Research Institute. S.S.S. is supported by the Department of Defense, Robert Wood Johnson Foundation as part of the Harold Amos Medical Faculty Development Program (AMFDP), the Urology Care Foundation Rising Stars in Urology Research Award Program and Astellas, Inc.

Disclosures: The content is solely the responsibility of the authors and does not necessarily represent the official views of U-M Precision Health. JJT is a co-founder with minor equity interest in LynxDx, which has licensed urinary biomarkers unrelated to this project.S.A.T has received travel support from and had a sponsored research agreement with Compendia Bioscience/Life Technologies/ThermoFisher that provided access to a DNA sequencing panel used herein. The custom RNAseq panel was designed through the Ampliseq white glove service of ThermoFisher Scientific. The University of Michigan and Brigham and Women’s Hospital have been issued a patent on ETS gene fusions in prostate cancer on which S.A.T. is a co-inventor. The diagnostic field of use was licensed to Hologic/Gen-Probe, Inc., which has sublicensed rights to Roche/Ventana Medical Systems. S.A.T. has served as a consultant for and received honoraria from Janssen, AbbVie, Sanofi, Almac Diagnostics and Astellas/Medivation. S.A.T. has sponsored research agreements with Astellas and GenomeDX. S.A.T. is a co-founder and Chief Medical Officer for Strata Oncology. T.M.M. has received research funding from MDxHealth, Myriad Genetics, and GenomeDx. T.M.M. has served as a consultant for Myriad Genetics. S.S.S. is on a study advisory committee for Bayer Pharma and has a non-sponsored research agreement with GenomeDx. The other authors have declared that no conflict of interest exists.

Siegel, R. L., Miller, K. D., Fuchs, H. E. & Jemal, A. Cancer statistics, 2022. CA. Cancer J. Clin. 72, 7–33 (2022).
Buyyounouski, M. K., Pickles, T., Kestin, L. L., Allison, R. & Williams, S. G. Validating the interval to biochemical failure for the identification of potentially lethal prostate cancer. J Clin Oncol 30, 1857–1863 (2012).
D’Amico, A. V. et al. Cancer-specific mortality after surgery or radiation for patients with clinically localized prostate cancer managed during the prostate-specific antigen era. J. Clin. Oncol. 21, 2163–2172 (2003).
Wei, L. et al. Intratumoral and Intertumoral Genomic Heterogeneity of Multifocal Localized Prostate Cancer Impacts Molecular Classifications and Genomic Prognosticators. Eur. Urol. 71, 183–192 (2017).
Løvf, M. et al. Multifocal Primary Prostate Cancer Exhibits High Degree of Genomic Heterogeneity. Eur. Urol. 75, 498–505 (2019).
Boutros, P. C. et al. Spatial genomic heterogeneity within localized, multifocal prostate cancer. Nat. Genet. 47, 736–745 (2015).
Salami, S. S. et al. Transcriptomic heterogeneity in multifocal prostate cancer. JCI Insight 3, (2018).
Cooper, C. S. et al. Analysis of the genetic phylogeny of multifocal prostate cancer identifies multiple independent clonal expansions in neoplastic and morphologically normal prostate tissue. Nat. Genet. 47, 367–372 (2015).
Lindberg, J. et al. Exome Sequencing of Prostate Cancer Supports the Hypothesis of Independent Tumour Origins. Eur. Urol. 63, 347–353 (2013).
Karakiewicz, P. I. et al. Cross-cultural validation of the UCLA prostate cancer index. Urology 61, 302–307 (2003).
Huang, C. C. et al. Re-evaluating the concept of “dominant/index tumor nodule” in multifocal prostate cancer. Virchows Arch. 464, 589–594 (2014).
Haffner, M. C. et al. Tracking the clonal origin of lethal prostate cancer. J. Clin. Invest. 123, 4918–4922 (2013).
Wang, Y. et al. Clonal evolution in breast cancer revealed by single nucleus genome sequencing. Nature 512, 155–160 (2014).
Navin, N. et al. Inferring tumor progression from genomic heterogeneity. Genome Res. 20, 68–80 (2010).
Alves, J. M., Prado-López, S., Cameselle-Teijeiro, J. M. & Posada, D. Rapid evolution and biogeographic spread in a colorectal cancer. Nat. Commun. 10, 5139 (2019).
Anderson, K. et al. Genetic variegation of clonal architecture and propagating cells in leukaemia. Nature 469, 356–361 (2011).
Xu, X. et al. Single-Cell Exome Sequencing Reveals Single-Nucleotide Mutation Characteristics of a Kidney Tumor. Cell 148, 886–895 (2012).
Navin, N. et al. Tumour evolution inferred by single-cell sequencing. Nature 472, 90–94 (2011).
Gundem, G. et al. The evolutionary history of lethal metastatic prostate cancer. Nature 520, 353–357 (2015).
Liu, W. et al. Copy number analysis indicates monoclonal origin of lethal metastatic prostate cancer. Nat. Med. 15, 559–565 (2009).
Woodcock, D. J. et al. Prostate cancer evolution from multilineage primary to single lineage metastases with implications for liquid biopsy. Nat. Commun. 11, 5070 (2020).
Cheng, L. et al. Evidence of Independent Origin of Multiple Tumors From Patients With Prostate Cancer. JNCI J. Natl. Cancer Inst. 90, 233–237 (1998).
Cyrta, J. et al. Comparative genomics of primary prostate cancer and paired metastases: insights from 12 molecular case studies. J. Pathol. 257, 274–284 (2022).
Hovelson, D. H. et al. Development and Validation of a Scalable Next-Generation Sequencing System for Assessing Relevant Somatic Variants in Solid Tumors. Neoplasia 17, 385–399 (2015).
Palapattu, G. S. et al. Molecular Profiling to Determine Clonality of Serial Magnetic Resonance Imaging/Ultrasound Fusion Biopsies from Men on Active Surveillance for Low-Risk Prostate Cancer. Clin. Cancer Res. 23, 985–991 (2017).

There is NO Competing Interest.

FigureS1RNASeqData.pdf
Figure S1. Targeted transcriptomic profiling of multifocal prostate cancer with synchronous lymph node (LN) metastasis. RNAseq was performed using an in-house targeted panel comprising of 302 amplicons to quantify the expression of known dysregulated genes in PCa along with ETS gene fusion partners, androgen signaling genes and lncRNAs. Unsupervised hierarchical clustering of 223 genes (excluding fusion amplicons) that passed QC was performed to derive expression profiles for each patient. Samples were clustered using a complete linkage method whereas genes were ordered by their function. Key AR genes and lncRNAs were highlighted on the heatmap.
FigureS2Pt2.pdf
Figure S2-8. Patient level phylogenetic evolutionary analysis to determine the biologically dominant nodule. Each patient figure included in this supplement is described in the following order: Left panel. The spatial location of each tumor sample on the radical prostatectomy specimen is annotated on a diagram of the prostate; Middle panel.Relevant clinicopathologic variables and gene signatures are annotated for each tumor region. Right panel. Evolutionary analysis was performed using neighbor-joining method by feeding DNA mutations, significant CNA, and gene fusion status (present/absent) into the algorithm. Phylogenetic trees were constructed with R phangorn package using maximum parsimony method. Figure S2. In patient #2, 4 primary (all grade group 5) and 2 lymph node metastatic prostate cancer regions were analyzed. All tumor regions with available data were ETS gene fusion negative. A CDK12 frameshift mutation, MYCand FGFR3 amplification were detected in all tumor regions. CDKN2Bdeletion was observed in primary tumor regions P1, P2, and P4 as well as LN1. The phylogenetic tree suggests that P4 (with EPE, cribriform pattern, and solid cell pattern) closely resembles LN1 focus and a possible separate trajectory for LN2 which subsequently developed TP53 loss.
FigureS3Pt4.pdf
Figure S3. In patient #4, a total of 4 primary and 2 lymph node metastases regions were analyzed. All tumor regions inclusive of LN metastases were ETS gene fusion negative. A TP53 frameshift mutation, FANCA gain, and CDKN2Bloss were detected only in LN metastasis regions. Primary tumor regions P1 and P2 harbored a number of diverse driver mutations which were not detected in other regions. Phylogenetic tree analysis suggests that P3 and P4 likely gave rise to LN metastatic regions LN1 and LN2.
FigureS4Pt30.pdf
Figure S4. A total of 7 primary tumor (grade groups 2 – 5) and 2 LN metastases regions were analyzed in patient #30. All regions inclusive of the LN metastases were positive for SLC45A3:ERG gene fusion. CDKN2A, ERCC2 and ERCC3 deletions were observed in primary tumor regions P1, P3, P4, P7 and LN metastasis region LN2. Additionally, TP53 mutation was detected only in LN metastasis region LN2. In the phylogenetic analyses, however, only primary tumor regions P3 closely resembles LN2. (*Genomic data was not available for primary tumor region P6).
FigureS5Pt34V2.pdf
Figure S5. In patient #34, a total of 9 primary tumor and 3 LN metastasis tumor regions were analyzed. All tumor regions were ETS gene fusion negative. P6 region was excluded from phylogenetic analysis due to low tumor content. A FOXA1 frameshift mutation was observed in all regions except for P5. A different FOXA1 R219C mutation was observed in P5. This patient’s samples exhibit genomic complexity with CNA (BRCA1 and PTEN deletions) present in P1, P4, P7, P8, P9, LN1, LN2 and LN3 but absent in P2, P3 and P5 suggesting two branches of clonal evolution. Notably, LN1, LN2 and LN3 displayed genomic homogeneity with additional BRCA2 and CDKN2A deletions as well as MYC amplification. Regions P1, P4, P7, P8, P9 thus represent the dominant cancer area resulting in LN metastasis.
FigureS6Pt38V2.pdf
Figure S6. In patient #38, a total of 7 primary tumor and 1 LN metastatic cancer regions were analyzed. P1, P2 and P3 were in the apex whereas P4, P5, P6 and P7 were in mid prostate. All regions were ETS fusion negative. A FOXA1 mutation along with PTEN and RB1 deletions were detected in all tumor regions except for P1 and P2 which had low tumor content and therefore, were removed from phylogenetic analysis. Among P3, P4, P5, P6, P7 and LN1, only P6 and LN1 displayed similar alterations such as CDKN2B deletion and RECQL4amplification, suggesting a common clonal origin.
FigureS7Pt39.pdf
Figure S7. In patient #39, a total of 12 primary tumor and 1 LN regions were analyzed. Regions P1, P2, P3 and P4 were in the apex; P5, P6, P7 and P8 extended from the apex to mid prostate, whereas regions P9, P10, P11 and P12 were in the mid prostate. All tumor regions were TMPRSS2:ERG gene fusion positive except for P11 (RNAseq data did not meet QC metrics). No significant somatic mutations were detected in this patient. CDKN1B amplification was detected in all primary tumor regions except for LN1. The dominant tumor region giving rise to metastasis is unknown in this patient.
FigureS8Pt40.pdf
Figure S8. In patient #40, a total of 5 primary tumor and 1 LN metastasis regions were analyzed. All tumor regions were TMPRSS2:ERG gene fusion positive. A TP53 somatic mutation was detected in regions P2, P3 and P4. CDKN1B, NKX3-1 and RB1 deletions were detected in regions P4 and P5. The dominant tumor region giving rise to metastasis is unknown in this patient.
TableS1PatientLevelPathologyData.xlsx
Table S1: Clinicopathologic characteristics of the cohort

Download PDF

Journal Publication

published 21 May, 2024

Read the published version in Nature Communications →

Version 1

posted

You are reading this latest preprint version

Integrative multi-region molecular profiling of primary prostate cancer with synchronous lymph node metastasis to characterize the biologically dominant clone

Status:

Journal Publication

Version 1

Abstract

Figures

Introduction

Results

Discussion

Methods

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1