Recent Advances in Blood-based Liquid Biopsy Approaches in Prostate Cancer

Abstract

The advent of high-throughput technologies has enabled the analysis of minute amounts of tumor-derived material purified from body fluids, termed liquid biopsies. Prostate cancer (PCa) management, like in many other cancer types, has benefited from liquid biopsies at several stages of the disease. Although initially describing circulating tumor cells (CTCs) in blood, the term “liquid biopsy” has come to more prominently include cell-free, circulating tumor DNA (ctDNA), RNA, proteins and other molecules. They provide tumor molecular information representing the entire, often-heterogeneous disease, relatively non-invasively and longitudinally. Blood has been the main liquid biopsy specimen in PCa, and urine has also proven beneficial. Technological advances have allowed clinical implementation of some liquid biopsies in prostate cancer, in disease monitoring and precision oncology. This narrative review introduces the main types of blood-based PCa liquid biopsies focusing on advances in the past 5 years. Clinical adoption of liquid biopsies to detect and monitor the evolving PCa tumor biology promise to deepen our understanding of the disease and improve patient outcomes.

1. Introduction

Although the term liquid biopsy was initially coined over a decade ago to refer to circulating tumor cells (CTCs, intact cancer cells disseminated from tumors and found in blood) by Pantel and Alix-Panabieres¹, detection of blood-derived tumor material had existed much longer with the measurement of several protein-based biomarkers, most prominently the prostate specific antigen (PSA) in prostate cancer (PCa)². Technological advances over the past two decades, notably in next generation sequencing (NGS) of nucleic acids extracted from tissue has led to a better understanding of PCa driver genomic alterations ^3,4. This has allowed extension of this knowledge to liquid biopsies by developments in interrogation of the small amounts of nucleic acids, proteins and other analytes obtained from body fluids, such as blood and urine. Prominent among these is detection of cell-free, circulating tumor DNA (ctDNA), thought to be released by dying (or live) cancer cells. Detection of somatic ctDNA mutations in plasma and their fraction as a component of the total (mutant and normal) DNA reflects disease burden and portends worse prognosis at various stages^5,6. Likewise, the number of CTCs in blood of PCa patients is also associated with disease burden and prognosis⁷. Further, liquid biopsy tends to be minimally invasive, thus allowing serial monitoring, which is key given the propensity of tumors to evolve in response to endogenous constraints and selective therapeutic pressures⁸. Rapid turnover of cancer-derived materials in blood (measured in hours for CTC and ctDNA) means a near real-time picture of the evolving disease. Liquid biopsy is also thought to contain a broader representation of the entire disease compared to localized tissue biopsies. This is important given the heterogenous nature of most cancers^9,10, but it is especially central in PCa given the propensity for multifocal disease¹¹. This narrative review introduces the main areas of application of blood-based liquid biopsy in PCa with the goal of describing advances made in the past 5 years.

2. Methods

A literature search was performed based on the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) criteria. The topics for this semi-systematic PCa liquid biopsy review were pre-defined: CTCs, ctDNA, and extracellular vesicles. A search of PUBMED database using the following terms: “prostate cancer” AND (CTC OR ctDNA OR “extracellular vesicles” OR “liquid biopsy”) was performed. Studies were excluded due to inability to access the full article, non-English manuscripts, and those belonging to excluded publication types i.e., not presenting original research and studies not involving male human subjects/samples. The search was limited to publications in the past 5 years since the time of manuscript preparation (03/2018 – 03/2023). A total of 634 articles were initially identified followed by unbiased screening according to the above pre-established criteria. Articles on malignancies other than PCa or those not solely focused on PCa, or non-relevant to the scope/topics of the review, were filtered out based on title (and abstract, as needed) review. This yielded 126 articles for which abstract evaluation was performed for relevance with the scope/topics of the review or redundancy with others. Publications with the highest level of evidence were selected for review. This yielded 60 articles for which full text evaluation was performed and included in the final review^12-71. A flow diagram of the articles’ evaluation process is shown in Figure 1. Both authors reviewed and approved the search/filtering strategy and the final reference list. The results of the literature search are grouped by topic and presented in a narrative fashion.

Figure 1.

Flow chart detailing literature search and selection of articles.

3. Evidence synthesis

ctDNA

Circulating tumor DNA (ctDNA) has provided biological insights in multiple cancer types, including PCa⁷². Methods used for detection have included the highly sensitive droplet digital PCR (ddPCR); NGS, which can take the form of deep targeted sequencing (fixed gene panel or tailor-made, bespoke panels based on a patient’s cancer tissue mutations), as well as low-pass whole genome sequencing; and methylomic/fragmentomic analysis to infer gene expression, amongst others.

Notable advances have been made in solidifying methodology, consolidating the value of ctDNA fraction in circulation as a marker of poor prognosis, studying ctDNA behavior under various treatment regimens, etc. Thus, a comparison of a major commercial ctDNA sequencing assay (Guardant360) with an academic laboratory-based gene panel identified very good concordance at 1% ctDNA fraction or above, below which, alteration calls became considerably discordant/questionable²⁷. When correlated to serum PSA, 4 of every 5 PCa patients with a PSA of >5 ng/ml were above the 1% ctDNA tumor fraction threshold, as opposed to fewer than half of patients with a PSA <5ng/ul²⁶. Comparison of ctDNA findings to those in matched, synchronous tissue biopsies as the benchmark method has continued to be studied. In a de-novo metastatic, castration-sensitive cohort, ctDNA was 80% concordant with tissue in detecting genomic alterations, indicating complementarity between the two approaches²¹. A larger cohort of mCRPC patients reflected this high tissue-ctDNA concordance, including in the actionable BRCA1/2 gene alterations, but less so in putative resistance alterations found in ctDNA¹⁷. The reasons for the observed high, but imperfect concordance, also seen in other cancer types, can be varied. They likely include time between the tissue and liquid biopsies (reflecting tumor evolution, especially in late, heavily treated disease^12,25,73), intra-patient heterogeneity more likely to be fully captured in ctDNA than in tissue^14,25, and a deeper analysis of the specific region of the lesion biopsied in tissue due to higher tumor fraction vs. broader, lower tumor fraction analysis in ctDNA.

The independent prognostic impact of ctDNA fraction in PCa patients undergoing various therapies has been demonstrated in several studies. In mCRPC patients on taxane chemotherapy, a higher baseline circulating total cell-free DNA content in blood (not ctDNA, but total DNA, including that of normal origin) as well as a cell-free DNA response (decrease upon treatment initiation), was associated with a shorter progression-free and overall survival⁶. This is important as it avoids the need for DNA sequencing. A look at the actual tumor-derived ctDNA in the same cohort by low-pass (i.e., shallow) whole genome sequencing, equally found prognostic association of ctDNA content with progression-free and overall survival²². Similar findings were obtained in enzalutamide or abiraterone-treated mCRPC patients²⁰, whereas the role of ctDNA in predicting success of abiraterone-to-enzalutamide treatment switch was not straightforward and depended on the genomic alterations¹³. Apalutamide added to ADT in the non-metastatic setting did not cause any marked changes in ctDNA alterations in blood¹⁶. Meanwhile, looking at circulating RNAs, Benoist and colleagues identified microRNA miR-3687 and confirmed miR-375 as poor prognosis factors for mCRPC patients being treated with enzalutamide¹⁸.

Lineage plasticity, known to be involved in trans-differentiation of treatment-resistant mCRPC into truly androgen independent neuroendocrine cancer was detected in ctDNA by Beltran et al. using whole exome DNA and bisulfite sequencing (to detect DNA methylation patterns). High detection accuracy was obtained by combining genetic (RB1, TP53) and epigenetic (ASXL3, SPDEF) alterations²³. Hypermethylated ctDNA of DOCK2, HAPLN3, and FBXO30 was a poor prognostic factor for development of mCRPC from localized or non-existing PCa¹⁵, whereas global ctDNA hypermethylation (with hypomethylation of regions near centromeres) distinguished metastatic PCa from localized disease²⁴. Finally, Nimir and colleagues tackled a comparison of detecting AR-V7 (an AR splice variant associated with advanced anti-androgen therapy resistance; see CTC section below) in circulating RNA, CTCs and extracellular vesicles, showing a significant advantage of CTCs in detecting AR-V7, also aided by the frequent presence of CTCs in advanced mCRPC disease²⁹.

CTCs

Circulating tumor cells (CTCs) provide useful prognostic and predictive information in PCa. Generally, half, or more patients have detectable cancer cells in their blood, depending on disease stage, method of detection and other factors⁷⁴. PCa CTC detection by the FDA-cleared CellSearch^® method based on EpCAM immunomagnetic enrichment was established early⁷. Other approaches, such as apheresis-based methods that screen larger volumes of blood to retrieve CTCs and route the rest of blood components back into circulation have shown the ability to isolate greater numbers of CTCs⁷⁵. This has been followed by the demonstration in PCa of the prognostic value of CTC counts^7,31,37 in line with several other cancer types where presence of greater numbers of CTCs, unsurprisingly, portends poorer prognosis⁷⁶. Demonstration of several other CTC detection/isolation platforms in PCa have since followed, some in the recent past. Most notably, the EPIC^® system uses an enrichment-free platform where nucleated cells are smeared on glass slides and analyzed using digital pathology based on morphological and other factors. Using this method, CTC enumeration was shown to be highly-concordant to CellSearch^® (r=0.84) and prognostic of overall survival in advanced PCa patients³⁵. With the same platform, Salami et al. found CTCs in 73% of high-risk localized PCa patient, being often heterogeneous in their expression of epithelial cytokeratin markers and androgen receptor (AR), consistent with epithelial-to-mesenchymal transition (EMT)⁴⁶. Single-cell whole genome sequencing allowed copy number profiling revealing further, genomic heterogeneity. Greater CTC counts, AR⁺ CTCs, and phenotypic heterogeneity by this method correlated with biochemical recurrence and development of metastatic disease on imaging⁴⁶.

Advances in methods to isolate and characterize CTC have provided opportunities for their utilization not only to gain biological insights, but in some cases yield biomarkers with proven clinical utility, such as markers of anti-androgen therapy resistance and predictive of chemotherapy benefit. Advanced PCa patients whose cancer develops resistance to the initial androgen deprivation therapy (ADT), so-called metastatic castration-resistant prostate cancer (mCRPC), commonly benefit from next generation AR signaling blockers (e.g., abiraterone, enzalutamide and apalutamide). Persistent AR blockade, however, drives the development of a constitutively active, truncated AR splice variant, AR-V7, which lacks the androgen ligand binding domain, the direct/indirect target of those inhibitors. CTCs have become an optimal analyte for non-invasive AR-V7 detection. Thus, CTC AR-V7 has been shown to be a biomarker of resistance to next generation AR blockade in mCRPC^50,77, and of benefit from taxane chemotherapy instead⁷⁸. A more recent report, however, finds unclear evidence for CTC AR-V7 positivity as providing additional, statistically significant prognostic value beyond known prognosticators such as CTC enumeration⁴⁷. In another study, this was especially true in patients with high CTC counts⁴². This latter study’s population consisted mostly of patients already undergoing taxane treatment (sometimes chosen based on AR-V7 positivity), thus it may define a strata of patients where this biomarker’s value is confounded/not present⁴². The former study was also comprised of a population receiving heterogeneous treatment types⁴⁷. A third study, by Sieuwerts et al. did not find association of AR-V7 with outcome on cabazitaxel, a taxane chemotherapy⁴⁵. Indeed, at present, the clinical utility of AR-V7 detected on CTCs seems to be greatest in selecting patients for AR signaling inhibitor vs. taxane chemotherapy in mCRPC ⁴¹, although one report demonstrated prognostic ability of AR-V7 (and its related variant, AR^v567es) in taxane treatment⁴⁸.

Work toward identification of expression of other variants/genes in early ADT treatment^33,34, and beyond AR-V7 in advanced disease as prognostic factors in PCa CTCs continues. For example, the presence of prostate-specific membrane antigen (PSMA)-positive CTCs in CRPC portended a shorter biochemical recurrence-free survival in patients progressing on one line of therapy and starting the next⁴⁰. This is also exemplified by Cheung et al. who looked at a panel of 89 transcripts identifying an 8-gene signature including AR-V7, that outperformed AR-V7 alone in predicting biochemical recurrence⁴⁴.

CTC profiling has made headways in other PCa biomarker areas such as detection of deletions in homologous recombination DNA damage repair genes, approved markers of benefit from PARP inhibition in mCRPC, another area of excitement in PCa therapy (and in other cancer types). Thus, Barnett and colleagues were able to detect BRCA2 loss at the single-cell level in CTCs of mCRPC patients equally accurately as in matched non-osseous tissue biopsies and more accurately than in bone biopsies³². Similarly, work by the same group measured single-cell, genome-wide chromosomal instability (largely an outcome of deficient homologous recombination repair) by DNA sequencing³⁹. This instability was correlated with, and could be recapitulated by cell morphological changes that, when analyzed with digital pathology methods, yielded a model that could prognosticate outcomes on AR-directed or taxane therapy.

As a result of persistent AR-inhibition, formerly AR-dependent PCa undergoes lineage plasticity and trans-differentiates into neuroendocrine (NE) PCa. Zhao at al. recently reported on detection by RT-qPCR of the expression of NE markers synaptophysin (SYN) and chromogranin-A (CHGA) together with AR signaling reporter transcripts in CTCs. They showed 100% accuracy of detecting NE disease at the patient level against known, tissue-based NE status in at least one blood sample over longitudinal monitoring³⁰. Indeed interest in the serial tracing of patients by CTC liquid biopsy to monitor prognosis and response to therapies, tumor heterogeneity and evolution, and other applications has been growing recently in several cancer types as in PCa^{36,43,49,79,80}. Lorente et al. found that an increase in CTC counts compared to baseline in advanced PCa patients starting AR-directed treatment or chemotherapy was associated with worse overall survival⁴⁹. A report by Hille and colleagues devised a method to detect AR-V7 in PCa from CTCs enriched by the FDA-cleared CellSearch^® system, which could be useful for clinical monitoring of AR-directed therapy⁴³. Likewise, Gupta et al. analyzed CTC samples from mCRPC patients at the start of, and after progression on next generation AR inhibitors to detect mutations and copy number alterations revealing CTC genomic heterogeneity and evolution, at times in genes related to NE differentiation such as RB1 and TP53³⁶. CTC platelet cloaking (the covering of the CTC surface by platelets speculated to protect CTCs in transit in the bloodstream from harsh conditions, the immune system etc.⁸¹) is also beginning to be studied in PCa^38,82. Ongoing strategies include comparing/combining CTCs with circulating DNA/RNA and/or extracellular vesicles ^28,29.

Extracellular Vesicles

Extracellular vesicles (EVs), or exosomes, are nano-scale, membrane-enclosed cellular fragments released into the extracellular environment carrying RNA, DNA, protein and other cell components. They are speculated to have intriguing functions in cancer, from preparing the distant metastatic niche to transforming normal cells. Recent work in EVs of PCa patients has complemented earlier studies by shedding light on several aspects such as methodology, individual biomarkers, and panel-based studies of multiple markers contained in them in diagnosis and risk stratification of early disease or at prostatectomy, using both urine and blood-based approaches. In a small cohort, Pang et al. showed that size-exclusion chromatography-based methods provided purer populations of EVs compared to precipitation methods in blood for PCa diagnosis⁷⁰. Notably, the ExoDx Prostate (IntelliScore) test based on EV detection of long non-coding and gene fusion RNAs in urine has been shown to aid detection of high-risk disease ^{66 67}. Likewise, when performed prior to repeat biopsies in men with previous negative biopsies, the ExoDx test could predict high risk disease in the repeat biopsy setting⁶⁴. In other assays, plasma EVs positive for the protein STEAP1 had significant biomarker potential for detection/diagnosis but carried little prognostic information⁶⁹. Urine EV RNA profiling assayed by ddPCR identified the long non-coding RNAs PCA3 (an established urine PCa biomarker) and PCGEM1 as predictors of high-grade cancer on diagnostic biopsy⁶⁵. And as mentioned above, EVs demonstrated inferior performance characteristics when compared to CTCs in AR-V7 detection²⁹. Other urine-based approaches that may include EV-derived materials have been reported^51-63. Lastly, proteomic EV profiling in sera of PCa patients has shown some prognostic potential⁶⁸.

4. Limitations

We did not perform a comprehensive systematic review and confined the review to publications on the topics of ctDNA, CTCs and EVs in the last 5 years. The search terms used may not be exhaustive and filtering criteria based on redundancy of topic/findings between publications may have removed secondary or supplemental findings not included in the retained papers. Except for the topic of EVs, urine based PCa liquid biopsy is largely not covered in this review. Nevertheless, our review captured the key advancements in the field of PCa liquid biopsy research over the last 5 years.

5. Conclusions

In this review, we describe recent advances and clinical applications of blood-based liquid biopsy approaches in prostate cancer including ctDNA, CTCs, and EVs. The current data demonstrates the prognostic and, in some cases, the predictive capacity of liquid biopsy. Importantly, optimization of the technology and techniques for acquiring and analyzing liquid biopsy specimens will better our understanding of the biology of the disease across the of prostate cancer clinical continuum as well facilitate integration into patient care paradigms.