Low Abundance of Circulating Tumor DNA in Localized Prostate Cancer

Abstract

Purpose: Despite decreased screening-based detection of clinically insignificant tumors, most diagnosed prostate cancers are still indolent, indicating a need for better strategies for detection of clinically significant disease before treatment. We hypothesized that patients with detectable circulating tumor DNA (ctDNA) were more likely to harbor aggressive disease.

Methods: We applied ultra-low-pass whole-genome sequencing to profile cell-free DNA from 112 patients diagnosed with localized prostate cancer and performed targeted resequencing of plasma DNA for somatic mutations previously identified in matched solid tumor in nine cases. We also performed similar analyses of data from patients with metastatic prostate cancer.

Results: In all cases of localized prostate cancer, even in clinically high-risk patients who subsequently had recurrent disease, ultra-low-pass whole-genome sequencing and targeted resequencing did not detect ctDNA in plasma acquired before surgery or before recurrence. In contrast, using both approaches, ctDNA was detected in patients with metastatic prostate cancer.

Conclusion: Our findings demonstrate clear differences between localized and advanced prostate cancer with respect to the dissemination and detectability of ctDNA. Because allele-specific alterations in ctDNA are below the threshold for detection in localized prostate cancer, other approaches to identify cell-free nucleic acids of tumor origin may demonstrate better specificity for aggressive disease.

FIG 1.

Ultra-low-pass whole-genome sequencing of circulating tumor DNA. (A) Somatic copy number alteration (SCNA) profile of circulating tumor DNA (ctDNA) from patients with localized prostate cancer, encompassing National Comprehensive Cancer Network risk groups of low, intermediate-favorable, intermediate-unfavorable, high, and very high-risk disease (n = 112). Gray bars represent PGA and PTC values before artifact removal. Ploidy values are uncorrected. (B) SCNA profile of ctDNA from patients with radiographically confirmed, metastatic, castration-resistant prostate cancer (n = 7). (C) SCNA profile of patients in the prostate The Cancer Genome Atlas cohort (n = 333). (D) SCNA profile of patients in the Prostate Cancer Foundation-Stand Up to Cancer cohort (n = 150). Abi, abiraterone acetate plus prednisone; Enz, enzalutamide; GG, International Society of Urological Pathology grade grouping; Met, metastatic; PGA, percent genome altered; PSA, prostate-specific antigen; PTC, percent tumor content; TGP, tertiary Gleason pattern; Tx, therapy.

FIG 2.

Multiregion sampling of prostate cancer tissue for identification of alleles to be detected in circulating tumor DNA (ctDNA). (A) Representative case demonstrating mapping of tumor throughout the prostate and the selection of distinct histologies that may represent major branches of the tumor system. (B) Hematoxylin and eosin staining and PIN-4 cocktail immunostaining of three adjacent histologies. Brown chromogen: p63 and cytokeratins 5 and 14; red chromogen: α-methylacyl coenzyme A racemase; scale bar: 200 μm. (C) General schematic of workflow showing the sequencing of patient tissue for identifying candidate alleles and the retrospective analysis of those alleles in banked plasma from the same patient. (D) Phylogenetic tree of tumor foci from the prostate cancer mapped in (A). Although copy number alterations and point mutations are used for establishing the evolutionary tree, only point mutations are sequenced in plasma specimens. IHC, immunohistochemistry; RP, radical prostatectomy; WES, whole-exome sequencing; WGS, whole-genome sequencing.

FIG 3.

Detection of circulating tumor DNA in plasma from four patients with metastatic prostate cancer. (A, B) Plots of log₂ copy number (CN) ratio (cfDNA v buffy coat) for patients M03, M04, M06, and M07 as determined by (A) ultra-low-pass whole-genome sequencing and (B) high-depth whole-exome sequencing (WES). (C) A representative somatic point mutation (g.chr16:69368813G>T) observed in the cfDNA from patient M03 by WES. (D) Representative bespoke sequencing detection of mutant g.chr16:69368813G>T allele. (E) Scatter plots showing relationship of actual and adjusted (interpolated) bespoke sequencing mutant read counts versus mutant read counts from exome sequencing. Correlation statistic Spearman ρ and P values are the same for actual and interpolated counts. cfDNA, cell-free DNA; Seq, sequencing; VAF, variant allele fraction (mutant reads/total reads). (*) Mutant allele.