Circulating tumour DNA as a cancer biomarker

Abstract

Measurement of genetically altered DNA shed from tumours into the circulation can potentially provide a new generation of blood-based cancer biomarkers. Compared with tissue DNA biomarkers which require surgery or biopsy, samples for circulating tumour DNA assays can be obtained with minimal inconvenience and at lower cost. Furthermore, in contrast to tissue, the use of circulating tumour DNA allows serial monitoring, faster delivery of results and potentially provides an integrative representation of genetic alterations across all tumour sites within a patient. In contrast to existing protein-based cancer biomarkers, all of which can be produced by benign disease, circulating tumour DNA biomarkers would be expected to be more specific for malignancy. Furthermore, unlike the available blood cancer biomarkers, circulating tumour DNA can be used to predict response to specific therapies, identify mechanisms of therapy resistance and detect potentially actionable mutations. One of the first circulating tumour DNA assays recommended for clinical use involves EGFR mutation testing for predicting response to EGFR tyrosine kinase inhibitors in patients with advanced non-small cell lung cancer, especially when tumour tissue is unavailable. In order to accelerate the introduction of circulating tumour DNA assays into routine clinical use, laboratory medicine staff will have to undergo training in the use of polymerase chain reaction and DNA sequencing. Furthermore, existing circulating tumour DNA assays will need to be simplified, standardized, shown to have clinical utility, be made available at reasonable costs and be reimbursable.

Introduction

Currently used blood-based cancer biomarkers are predominantly proteins or glycoprotein molecules. However, like protein molecules, DNA can be released or shed from tumours and exist in a cell-free (cf) state in blood. Tumour-derived cfDNA in blood is referred to as circulating tumour DNA (ctDNA), and the use of ctDNA for clinical purposes is frequently referred to as a ‘liquid biopsy’. Because of its potential to serve as a ‘liquid biopsy’ or surrogate fluid for tumour tissue-based biomarkers, research into ctDNA is currently a highly active area of investigation. Indeed, measurement of ctDNA has recently entered clinical use in patients with advanced non-small cell lung cancer (NSCLC). Furthermore, several high-profile reports on the potential use of ctDNA to detect early stage cancer have recently been published. The aim of this article is therefore to review the potential use of ctDNA as a new source of cancer biomarkers. Firstly, however, we briefly discuss the release of ctDNA into blood and the collection of blood for measurement of ctDNA. As methods for measuring ctDNA have previously been discussed in detail,^1–4 they will not be reviewed here.

Release of ctDNA into blood

Release of ctDNA into the circulation is believed to result from tumour cell apoptosis or necrosis, although spontaneous release may also occur.^4,5 Based on the molecular size of ctDNA fragments in blood, i.e. 160–180 base pairs and multiples of 160–180 base pairs, it appears that most of the ctDNA is associated with nucleosomes ⁶ and released following tumour cell apoptosis. ctDNA fragments however, are generally shorter than non-mutant cfDNA.^7,8 Irrespective of the mechanism of release, ctDNA is rapidly cleared from blood with a half-life of <2.5 h. ⁷ Clearance is believed to occur through the kidneys, liver and spleen, although uptake by other normal host cells may also occur. ⁷

In cancer patients, concentrations of ctDNA can vary widely and may be as low as <0.01% or as high as >10% of the cfDNA present.^9,10 This wide variation in ctDNA is likely to be related to factors such as tumour load, tumour location, proximity of tumour to vasculature, extent of necrotic component in tumour, presence or absence of lymphovascular invasion, whether the mutations being measured are clonal or subclonal and rate of malignant cell proliferation.^11–13 All the main types of genetic alterations found in tumour tissues have been detected in ctDNA, including point mutations, copy number variations, chromosomal rearrangements, epigenetic alterations, insertions and deletions.

Collection of blood for ctDNA measurement

For measurement of most of the classical blood-based cancer biomarkers, serum and plasma can be used interchangeably. For detecting ctDNA, however, plasma is preferable to serum as the clotting process during serum formation results in the release of normal, blood cell-associated, DNA. The normal or wild-type (WT) DNA can dilute any tumour-derived DNA molecules present. For preparing plasma DNA, EDTA is a superior anticoagulant to either lithium heparin or sodium citrate, especially if whole blood is not processed within 6 h of collection. ¹⁴ A further problem with heparin is that it interferes with polymerase chain reaction (PCR) procedures that may be used in the subsequent detection of ctDNA.

Some studies, however, reported that ctDNA in EDTA collection tubes increased after 24 h if stored at room temperature. This increase appeared to be due to the release of WT DNA from residual leucocytes. To minimize this problem, several types of commercially available tubes with proprietary preservative have become available.^15–17 With some of these (e.g. BCT tubes [Streck Inc.], or CellSave tubes (Janssen Diagnostics), cfDNA was found to be stable for up to 96 h. ¹⁶ Currently, however, there is no uniform, standardized method for collecting blood for ctDNA analysis.

Tissue-based DNA biomarkers in clinical use

Although research into ctDNA assays is relatively new, measurement of tumour tissue DNA-based biomarkers has been standard clinical practice for several years. These tissue biomarkers are particularly used for predicting likely response or resistance to specific therapies. They include measurement of HER2 gene amplification for predicting response to anti-HER2 therapy in breast cancer, KRAS/NRAS mutations testing for predicting resistance to anti-EGFR antibodies in colorectal cancer (CRC), EGFR mutation testing for predicting benefit from anti-EGFR inhibitors in lung adenocarcinoma and BRAF mutation testing for predicting response to anti-BRAF drugs in melanoma. ¹⁸ Although tissue obtained by biopsy or surgery is the gold standard for measuring these therapy predictive biomarkers, its use has limitations. Some of these limitations can be overcome or minimized with the use of ctDNA, as discussed below.

Advantages of measuring ctDNA over tissue DNA

Sample for ctDNA analysis can be accessed in a minimally invasive manner

Because of inaccessibility and potential risk to patients, obtaining tissue by surgery or biopsy may be a problem, especially in frail elderly patients with advanced cancer. For example, approximately one quarter of patients with NSCLC either cannot undergo a tissue biopsy or fail to provide sufficient sample for analysis.^19,20 Similarly, in patients with metastatic prostate cancer, the success rate for obtaining bone tumour tissue varies from 25% to 75%. ²¹ In contrast to the invasive procedures of biopsy and surgery, blood can be obtained in a minimally invasive manner with low risk and low cost to patients. Being able to sample blood is especially important for real-time monitoring of patients receiving systemic therapies in advanced cancer. Thus, with real-time testing, it may be possible to monitor response to treatment, detect the early emergence of therapy resistance, identify mechanism(s) of resistance and identify potentially actionable mutations for further treatment.

ctDNA potentially captures spatial and temporal tumour heterogeneity

Tumour heterogeneity is the coexistence of populations of cells with different genetic or epigenetic alterations within the same malignancy (intratumour heterogeneity). ²² Intratumour heterogeneity appears to be common if not universal in solid tumours. Thus, in a study on NSCLC, 30% of somatic mutations and 48% of somatic copy number alterations were found to be subclonal, i.e. not present in all the areas of the tumour sampled. ¹¹ Furthermore, as localized tumours progress and metastasize, they tend to accumulate additional genetic alterations, leading to what is referred to as temporal heterogeneity. Thus, the mutational status of a metastatic lesion may differ from that of its corresponding primary tumour. Indeed, the mutational status of different metastases from a common primary site may also differ.

Although universally used in cancer diagnosis, a tissue biopsy may be informative only for a single time-point and for a single location within a tumour. Theoretically, minimizing the problem of temporal and spatial heterogeneity could be accomplished by performing multi-location and serial biopsies including biopsies at metastatic sites. Clearly, such a strategy is not practical in the clinical setting. Although a single biopsy is unable to capture tumour heterogeneity, blood might be expected to do so, assuming that all or most of the clones of malignant cells release ctDNA. Preliminary data suggest that ctDNA can indeed capture spatial and temporal genetic heterogeneity in patients with advanced breast cancer. ²³

Minimize DNA sequencing artefacts

Following biopsy or surgical resection, tumour tissue is normally fixed in formalin and embedded in paraffin. Although fixation in formalin preserves tissue and maintains cellular morphology, it can result in DNA fragmentation which reduces the concentration of amplifiable template for PCR reactions.^24,25 In addition, it may cause sequence artefacts in DNA that might be interpreted as cancer-associated mutations. Formalin-fixed tumour tissue may thus be unsuitable for molecular analysis, especially for whole exome and whole genome sequencing. In contrast to tissue, blood can provide fresh DNA directly and that is unaltered by organic solvents, such as formalin or paraffin.

Provide faster turnaround times for results

Obtaining laboratory results in a timely manner is important, especially in patients with advanced rapidly progressing disease. Available data show that mutation testing using plasma leads to considerably shorter turnaround times for reporting of results than when tissue is used. Thus, in a study involving mutation testing for KRAS, NRAS and BRAF in CRC, the median delay between receipt of sample in the laboratory and communication of results was 12 days for tissue but only two days for plasma. ²⁶ Similarly, for EGFR mutation testing in NSCLC, the use of plasma resulted in a considerable shorter median turnaround time than tissue, i.e. 12 days for tissue versus three days for plasma. ²⁷

Advantages of ctDNA over protein for blood biomarkers

More specific for malignancy

One of the main disadvantages of the traditional serum protein cancer biomarkers is their lack of specificity for malignancy. Indeed, all the presently available protein tumour biomarkers (e.g. CEA, CA 125, CA 15–3, CA 19–9, AFP and PSA) can be increased in blood from patients with benign diseases. This lack of specificity for cancer limits the use of these biomarkers, especially in cancer screening and aiding early diagnosis. Measurement of cancer-associated genetic alterations, however, would be expected to be specific for malignant or premalignant lesions. ctDNA-based biomarkers, however, might not have absolute specificity for cancer as malignancy-associated genetic alterations have been reported in some apparently normal cells. ²⁸ Such cells, however, may be at an increased risk of progressing to malignancy. ²⁹

Shorter half-life

As mentioned above, the half-life of ctDNA in blood is less than 2.5 h, in contrast to several days for most of the traditional cancer biomarkers. ³⁰ This shorter half-life should result in ctDNA-based biomarkers being more sensitive than standard biomarker for identifying rapid changes in tumour burden and in monitoring response to therapy. Similarly, the greater dynamic range in ctDNA vis-à-vis traditional biomarkers^31,32 should also provide greater sensitivity in the above situations.

Clinical application of ctDNA testing

One of the first clinical applications of ctDNA analysis is EGFR mutation testing for predicting response to anti-EGFR tyrosine kinase inhibitors (TKIs) (e.g. erlotinib, gefitinib, afatinib and osimertinib) in patients with advanced NSCLC. ³³ Thus, activating mutations such as in-frame deletions in exon 19 (exon19del) and the point mutations (L858R) in exon 21 of EGFR are associated with response to these drugs. Overall, 60–80% of patients with advanced NSCLC containing activating EGFR mutations respond to EGFR TKIs, whereas patients lacking such mutations rarely benefit from these inhibitors. ³⁴ The use of EGFR TKIs has greatly transformed the outcome for patients with EGFR mutation-positive advanced NSCLC in recent years. With these inhibitors, median progression-free survival has increased from 5.6 months with chemotherapy to 11 months with the first-generation TKI inhibitors (erlotinib and gefitinib) to 22 months with the third-generation TKI inhibitor, osimertinib.^35,36

The gold standard assay for evaluating the mutational status of EGFR in patients with advanced NSCLC requires tumour tissue.^37–39 However, as mentioned above, because of risk of complications associated with taking a lung biopsy, especially in old or frail patients, obtaining tumour tissue is not possible in approximately 25% of patients with advanced NSCLC. Even if biopsy is possible, the tissue obtained may be inadequate or unsuitable for mutation testing. To obviate this problem, methods have been developed for detecting EGFR mutations in plasma. ⁴⁰ Indeed, several reports have found that similar to tissue, the presence of plasma EGFR mutations can also be predictive of response to EGFR TKIs.^40,41 In a meta-analysis of 27 published studies containing >4000 patients, the pooled sensitivity, specificity and areas under the curve for ctDNA versus tumour tissue were 0.60, 0.94 and 0.93 for the detection of EGFR mutations. Importantly, ctDNA has a high degree of specificity for detecting exon 19 deletions and the single-point substitution mutation L858R which are the best characterized EGFR mutations for predicting response to EGFR inhibitors. ⁴¹

The high specificity of ctDNA for detecting tumour EGFR mutations suggests that a positive plasma test could be used in deciding to administer EGFR TKIs to patients with advanced NSCLC. However, because of the relatively low sensitivity, absence of a plasma mutation should not be interpreted that the tumour is also negative for mutations. In such a situation, tissue should be evaluated when feasible. In 2018, joint guidelines published by the College of American Pathologists (CAP), the International Association for the Study of Lung Cancer (IASLC) and the Association for Molecular Pathology (AMP) stated that in some clinical settings in which tissue is limited and/or insufficient for mutation testing, physicians may use a ctDNA assay to identify EGFR mutations.^37,38 The guidelines, however, also stated that because the sensitivity of ctDNA assays are <80%, a negative result from ctDNA analysis is not reliable evidence that the corresponding tumour is also EGFR mutation negative.^37,38

Although most patients with activating EGFR mutations initially respond to EGFR TKIs, effectively all develop acquired resistance to the drugs. In approximately half of these cases, resistance is due to the emergence of the exon 20 T790M mutation in EGFR. ⁴² Approximately 60–70% of patients with this mutation, however, respond to the third-generation EGFR TKI, osimertinib. ⁴² As with the common activating mutations, T790M mutations can also be detected in plasma. ⁴³ Sensitivity for the plasma versus the tissue assay for the T790M mutation has been reported to vary from 40 to 70% and specificity from 60 to 98%.^43,44 Importantly, the plasma assay predicts response to osimertinib similarly to the tissue assay. ⁴⁵

Based on these findings, guidelines published by CAP, IASLC and AMP state that ‘physicians may use cell-free plasma DNA methods to identify EGFR T790M mutations in lung adenocarcinoma patients with progression or secondary clinical resistance to EGFR-targeted tyrosine kinase inhibitors’. Testing of tumour tissue, however, is recommended if the plasma result is negative. ³⁷

Two commercially available ctDNA-based tests for identifying patients with advanced NSCLC who are likely to benefit from anti-EGFR therapy have now been approved by regulatory organizations. One of these, known as the Cobas EGFR Mutation Test v2 (Roche Diagnostics Ltd., Rotkreuz, Switzerland), has been approved by both the European Medicines Agency and the US Food and Drug Administration. The Cobas EGFR Mutation Test v2 is a real-time PCR test that detects 42 mutations in exons 18–21 of the EGFR gene, including L858R, exon 19 deletions, L861Q and the TKI-resistance mutation, T790M. This test may be used when a tissue biopsy is not possible.

The second commercially available assay, i.e. therascreen EGFR Plasma RGQ PCR Kit (Qiagen, Hilden, Germany), has received the Conformité Européenne (CE) mark for detecting EGFR mutation in plasma when tissue is unavailable. This test detects deletions in exon 19, the T790M mutation in exon 20 and the L858R mutation in exon 21 of the EGFR gene.

Other emerging uses of ctDNA assays

One of most exciting potential uses of ctDNA testing is the early detection of cancer. As mentioned above, ctDNA biomarkers might be expected to be more specific for cancer than traditional protein biomarkers. Furthermore, preliminary reports suggest that specific cancer-associated mutations can be detected in plasma from patients with early stage cancer. These include mutant APC, mutant KRAS and mutant p53 in colorectal cancer,^46,47 mutant PIK3CA in breast cancer ⁴⁸ and Epstein-Barr viral DNA in patients with nasopharyngeal cancer. ⁴⁹ Indeed, measurement of circulating Epstein Barr virus DNA is a promising screening test for detecting early nasopharyngeal cancer in subjects at high risk for this malignancy. ⁴⁹

In one of the largest studies involving the use of ctDNA to detect early stage malignancy, Phallen et al. ⁵⁰ used an ultrasensitive massively parallel sequencing approach to measure 58 cancer-associated genes in plasma obtained from 44 healthy subjects and 194 patients with previously diagnosed malignancies of diverse origin. Overall, 62% of patients with early stage disease (stages I and II) and 77% of patients with advanced disease had mutations detected in their ctDNA. No alteration in driver genes (genes responsible for driving tumour cell growth) was detected in any of the healthy controls. However, 16% of healthy controls had mutations in genes associated with haematopoietic cell proliferation. The presence of such genes, however, has been reported to confer an increased risk for malignant transformation. ²⁹

In a related study, Cohen et al. ⁵¹ performed combined measurement of plasma mutations from 16 genes and eight protein biomarkers in 1005 patients with eight different types of non-metastatic clinically detected cancers. The combined DNA and protein test, known as CancerSEEK, was positive in a median of 70% of patients with the eight cancer types investigated. Sensitivity ranged from 33% to 98% for the detection of the eight cancer types (breast, lung, colon, oesophagus, pancreas, stomach, liver and ovary). Specificity exceeded 99%, i.e. only 7/812 apparently healthy controls had a positive result.

While the above reports appear promising with respect to the early detection of malignancy, a major question remaining to be addressed is whether the available assays for ctDNA possess sufficient sensitivity for detecting small tumours (<1 cm) in asymptomatic subjects. ⁵² As pointed out by Kalinich and Haber, ⁵³ the predictive value of a diagnostic test depends on the prevalence of the disease within the population of subjects undergoing investigation. Thus, in testing apparently healthy subjects in the general population, the prevalence of the eight cancers types evaluated in the above study would be approximately 1% in subjects over the age of 64 years. In this setting, even a test with a sensitivity and specificity of 99% would yield a positive predictive value of only 50% (i.e. half of all test-positives would be false positives. ⁵³ A further limitation of the above studies was the failure to measure ctDNA in patients with benign or inflammatory diseases.

Other potential uses of ctDNA assays include establishing prognosis for patients with newly diagnosed cancer, postoperative surveillance for early detection of metastatic/recurrent disease and monitoring therapy especially in patients with advanced disease.^4,7 The most immediate clinical application of ctDNA testing, however, is likely to be in predicting response or resistance to specific therapies, especially in patients where a tissue biopsy is not possible. Thus, in addition to the use of EGFR mutation testing discussed above for guiding treatment with EGFR inhibitors in patients with NSCLC, several other promising ctDNA biomarkers for predicting response or resistance have recently emerged. These include mutant KRAS for predicting resistance to anti-EGFR antibodies in patients with advanced CRC,^54,55 mutant BRAF for predicting response to anti-BRAF therapy in patients with advanced melanoma, ⁵⁶ reversion mutations in BRCA1/2 for predicting resistance to PARP inhibitors in prostate cancer ⁵⁷ and mutant ESR1 (encoding oestrogen receptor) for predicting resistance to aromatase inhibitors in patients with advanced breast cancer.^58,59 In the latter situation, patients with ESR1 mutations may be able to benefit from non-aromatase forms of hormone therapy such as fulvestrant. ⁵⁹ Unlike plasma EGFR mutation testing in NSCLC patients, measurement of ctDNA in the above malignancies is currently insufficiently validated for clinical utility.

Conclusion

It is clear from above that investigation into ctDNA as a potential new cancer biomarker is currently a highly active area of research. Indeed, in searching for new blood-based cancer biomarkers, the focus of research in recent years has shifted from protein to DNA. ctDNA as a cancer biomarker is likely in the future to have utility at all the main phases of cancer detection and management. In contrast to the available protein biomarkers, ctDNA assays can be of value in predicting response to targeted therapies, identifying mechanisms of therapy resistance and detecting new actionable mutations, as exemplified above in the case of EGFR mutation testing in patients with advanced NSCLC. In order to accelerate the introduction of ctDNA analysis into clinical practice, existing assays for mutations need to be simplified, standardized, shown to have clinical utility, be made available at reasonable costs and be reimbursed. Finally, laboratory medicine staff such as clinical biochemists will have to undergo training in these new molecular assays as well as in interpretation of the results they provide.