Liquid biopsy for cancer diagnosis and screening – The promise and challenges

Liquid biopsy in cancer refers to isolation and analysis of tumour-derived materials such as DNA, RNA, intact cancer cells and extracellular vesicles (EVs) in bodily fluids such as blood, urine, saliva and stools.^1,2 Tumour tissues shed a variety of materials into adjacent bodily fluid, and the most commonly evaluated materials are circulating tumour DNA (ctDNA) or circulating tumour cells (CTCs) in blood. Analysis of circulating cell-free DNA (cfDNA) is already routinely employed in clinic, for example in antenatal screening for Down’s syndrome ³ and liquid biopsy approaches are also being increasingly employed in the management of cancer patients, particularly in metastatic disease. The non-invasive nature of liquid biopsy means that it can be repeated as often as necessary and has the potential to be used for early detection of cancer, diagnosis, prognostication, prediction of response to treatment and surveillance. This editorial will primarily discuss the role of liquid biopsy in cancer diagnosis and screening.

cfDNA, released into the bloodstream through cellular apoptosis or necrosis, was first detected in 1948. ⁴ cfDNA is rapidly cleared from circulation, generally within an hour. ⁵ In individuals harbouring cancer, a very small fraction of cfDNA^6,7 comprises DNA originating from apoptotic or necrotic cancer cells (ctDNA), and ctDNA is primarily distinguished from cfDNA by specific molecular characteristics such as mutations, copy number variations, methylation changes or integrated tumour-associated viral sequences ⁸ ; ctDNA has been shown to reflect the mutational signatures of the primary tumour. ⁹ A range of assay methodologies with varying limits of detection (LoDs) are used for the detection of ctDNA and these include quantitative PCR, digital PCR (Droplet Digital PCR, BEAMing – Beads, Emulsion, Amplification and Magnetics), targeted or whole-genome sequencing with molecular barcoding.^7,10,11 Currently, digital PCR-based methods appear to be the most sensitive at relatively low cost with ability to detect prespecified mutation/s with allele frequencies of 0.01% in cfDNA. The potential utility of ctDNA to detect post-treatment minimal residual disease (MRD), predict risk of recurrence and to monitor metastatic disease has been demonstrated, for example in breast cancer^12–15 and a liquid biopsy companion diagnostic test for EGFR mutations is now approved by the US Food and Drug Administration (FDA). ¹⁶

The first report of CTCs appeared in 1869, ¹⁷ and first commercially available CTC assay (CELLSEARCH® Menarini Silicon Biosystems, Florence, Italy) was approved by the US FDA in 2007. Most CTC assays use either a size-based (cancer cells being larger in size) or antigen-based (e.g. EpCAM – epithelial cellular adhesion molecule) separation to isolate CTCs. Not all cancer cells are large, and some (e.g. those undergoing epithelial-mesenchymal transition) may not express targeted antigen/s; therefore, both these approaches result in some loss of sensitivity. Newer CTC assays^18,19 overcome these limitations by avoiding separation of CTCs and using image-analysis-based in silico identification which also confers an additional benefit of capturing morphological information. The role of CTCs in treatment decisions and metastatic cancers has been extensively investigated with mixed results^20–22; however, their role in early detection of cancer has not been investigated adequately. Current evidence ²³ suggests that CTCs and ctDNA provide independent information and therefore may prove complementary to each other.

EVs or exosomes are round 30–120 nm diameter vesicles carrying DNA, mRNA, microRNA (miRNA) and proteins that act as mediators of intercellular communication. ²⁴ Cancer-derived EVs contain various cancer-associated molecules ²⁵ and have been shown to reflect MRD status as well as predict response to therapy. ²⁶

Circulating miRNAs are 19–24 nucleotide non-coding RNA molecules with differential cancer-specific expression patterns. ²⁷ The stability of these molecules makes them an attractive biomarker candidate for a liquid biopsy-based approach.

Apart from being non-invasive and therefore easy to repeat, liquid biopsy has several unique advantages over other methods of early detection of cancer and screening as well as over conventional surgical biopsy for cancer diagnosis. The first and foremost of these is its ability to capture information regarding tumour heterogeneity. The importance of tumour heterogeneity, clonal evolution and their impact on treatment response and outcomes is now increasingly being recognized.^28–31 Unlike conventional biopsy, liquid biopsy, particularly ctDNA ³² and CTCs by newer assays,^18,33 is able to capture information regarding tumour heterogeneity. This is of significant importance in diagnostic setting where cancer treatment is now often guided by molecular profiling of the tumour. Molecularly targeted therapies have become an important component of cancer management, and liquid biopsy has potentially greater diagnostic utility, since it can provide information to guide use of such therapies. Capturing tumour heterogeneity and molecular information is also important in an early detection setting where it may help in distinguishing aggressive from indolent cancer ³⁴ and permit a screening approach to detect only potentially lethal cancers and thus minimize overtreatment. Overdiagnosis is a major problem associated with cancer screening, and it is mainly caused by detection of precancerous lesions (e.g. ductal carcinoma in situ) or indolent cancers that would never become life-threatening. Since such lesions are less vascular, theoretically, they would shed minimal or no tumour material in bloodstream, and therefore liquid biopsy-based screening may have the potential to minimize overdiagnosis, although this requires further investigation.

Early in 2018, a case–control study by Cohen et al. ³⁵ generated considerable interest, as they reported moderate sensitivity and high specificity for detection of eight cancers using a test called CancerSEEK based on concentrations of circulating proteins and mutations in cfDNA. Although sensitivities ranged from 69 to 98% for the detection of five cancer types (all stages) for which no screening tests are currently available, ³⁵ a sensitivity of about 40% in stage I cancer was too low for early detection purposes. Fifteen per cent of metastatic cancers do not have detectable ctDNA; furthermore, concentrations of ctDNA vary by tumour type and may not be proportional to tumour burden.^23,36,37 The LoD for detection of ctDNA is still not low enough, i.e. assay sensitivity is not high enough to detect a small tumour burden. Current relatively high LoD also results in poor reproducibility when different assays are used even in metastatic disease with relatively higher ctDNA concentrations. ³⁸ These issues have important implications for the detection of early stage cancer, and sensitivity levels are currently not sufficiently high to allow the investigation of liquid biopsy in a large-scale screening study. Mutations detected in cfDNA as a result of benign lesions, e.g. benign naevi ³⁹ or Clonal Haematopoesis of Indeterminate Potential (CHIP)^40,41 is another major challenge resulting in high false-positive rate. ⁴² This is of particular importance in the context of screening when the assay is used for generic and not site-specific cancer detection. In the study by Cohen et al., ³⁵ the site of the primary cancer could not be suggested in a quarter of patients. All such patients with a positive test result (true-positive or false-positive) but primary site not suggested would require a significant diagnostic work-up including imaging with cost and radiation exposure implications. Furthermore, reassuring those with a negative diagnostic work-up might pose a significant challenge in itself, something with which current healthcare systems employing site-specific screening approaches are not equipped to deal.

The technologies for the detection of biomarkers like ctDNA and CTCs in liquid biopsies are progressing rapidly, particularly with improvements in sequencing technology, image-analysis and computational methodologies. The majority of current research in early detection is focussed on ctDNA. Given that CTC and ctDNA complement each other ²³ and newer CTC assays are more sensitive and informative, a combined CTC–ctDNA detection approach is likely to improve both sensitivity and specificity. Furthermore, morphological and genomic characterization ¹⁸ information gained from CTCs could guide further diagnostic imaging by identifying a potential cancer site, thus helping overcome one of the challenges. Research is already underway to develop a cfDNA genome atlas,^41,43 so that the issue of false-positive results due to CHIP could be addressed. In summary, liquid biopsy holds a considerable promise for early detection of cancer and diagnosis, and major research efforts are underway to overcome the challenges so that its full potential can be realized.