Introduction

Advances in technology lead us to a deeper understanding of the molecular pathology of various diseases including cancer. The recent ongoing advances in cancer diagnostics are the introduction of analyzing circulating tumor deoxyribonucleic acid (ctDNA) and cell-free DNA (cfDNA).1 This has shown promising results in various studies and has become a vital role in nonsmall cell lung cancer (NSCLC) currently2 and the use in other sites is underway.345

Molecular testing of the tumor tissue has been a practice-changing advancement in the last decade. However, there are a lot of challenges in the real-world practice in selecting the right sample, timing the tissue sampling in the course of the disease, difficulty in obtaining the sample, adequacy of the same, and most importantly the patient discomfort due to invasiveness. The spatial and temporal tumor heterogeneity add to the complexity of the tumor tissue analysis.6 To eliminate such practical difficulties along with the invasiveness associated with the procedure, the concept of liquid biopsy using ctDNA, which is minimally or noninvasive, has been introduced.7 The potential role of ctDNA as a biomarker is utilized in a spectrum of situations such as early detection of tumor, making treatment decisions, monitoring response to the planned treatment, and disease status after treatment completion.8

ctDNA Testing in Oncology

Liquid biopsy uses the noninvasive method of detecting the tumor cells through circulating tumor cells (ctCs) or ctDNA. ctCs are the intact tumor cells that are shed into the bloodstream from the tumor and these cells release the ctDNA.9 The exact mechanism of ctDNA release is still not known but is hypothesized to originate from the viable tumor, apoptotic tumor cell, and the ctC.1011 Its use has been explored in various cancer recently due to its noninvasiveness, convenience, and safety.35 There are various methods available to analyze tumor genomics using ctDNA including polymerase chain reaction (PCR) and next-generation sequencing (NGS).12 However, there is a wide gap between translational research and clinical application due to a lack of uniformity and standardization for each of these. The Norwegian data showed that the impact of biomarker testing resulted in a decrease in survival due to the identification of a plethora of mutations without identifying the key driver mutation of the tumor to be addressed appropriately with the available targeted therapy.13

Role of NGS

The use of biomarker testing to understand the molecular genetics of the tumor helps to provide personalized medicine by selecting the right therapy for the right tumor while avoiding the costly ineffective or harmful drug for that particular tumor mutation. The ctDNA must be extracted from the plasma owing to a large number of normal cfDNA present in the serum and hence sample must be collected in the anticoagulant tubes.14 The two main methods to detect gene mutation using liquid biopsy are PCR-based and NGS-based. PCR- based method is generally preferred for single locus or targeted panel, while NGS can be applied to any sized panel of genes.15 NGS can be used to detect the known targeted panel and also an untargeted panel of genetic mutations including point mutation, deletions, rearrangements, and copy number variations. This can be achieved by performing whole-genomic sequencing and whole-exome sequencing.16

Clinical Utility in Colorectal Cancer

Currently, the only method of ctDNA testing that is approved by the Food and Drug Administration in the United States and China is DNA methylation-based testing of SEPT9 for colorectal cancer (CRC) screening and quantitative PCR-based epidermal growth factor receptor (EGFR) testing in NSCLC.17 In the case of early detection of cancer, there is no evidence to assess the quantitative relationship of ctDNA with cancer development. Also, it cannot be correlated with symptoms or the disease nature yet to evolve during screening. Similarly, the assessment of minimal residual disease after curative treatment in early-stage cancer is also challenging.18 This is due to the very low ctDNA levels, sample preparation, and sensitivity of the test.

A prospective study done in 196 patients to assess the clinical correlation between stage I to III CRC with ctDNA-positivity during longitudinal monitoring was associated with poor clinical outcomes in terms of relapse-free survival. Serial ctDNA analysis detected minimal residual disease up to a median of 8.08 months (0.6–16.6 months) ahead of radiologic relapse (sensitivity: 87.5%, specificity: 99.1%).19 There are various ongoing studies to assess the feasibility of escalation or deescalation of adjuvant systemic therapy based on the postoperative ctDNA levels.20 A prior knowledge of the tumor with serial ctDNA monitoring would circumvent the limitation of false positives and increase the confidence in identifying low-frequency variants.

The secondary resistance to anti-EGFR therapy in the metastatic colorectal patient has been demonstrated due to the emergence of new mutations especially KRAS amplification or mutations in approximately 60% of the study population.21 Also, these KRAS mutations were detectable in blood using ctDNA 10 months before the onset of radiological progression.21 Hence, ctDNA is a potential surrogate biomarker for monitoring the tumor burden along with tumor genomics and responses to treatment.

Having discussed the potential with the use of liquid biopsy in CRC patients, the major burden that prevents routine clinical utility include the lack of standardization in the sensitivity limit of the detected mutations, which would increase the between the plasma and tissue results, substantial costs, heterogeneity in methodology, and reporting of results among the noise caused by normal tissue ctDNA. Meta-analysis of 20 studies with 1,810 patients whose ctDNA matched to tissue DNA was analyzed for detecting RAS mutation in metastatic CRC using various methods.22 ctDNA testing demonstrated an overall sensitivity of 83% and specificity of 91%. About 20% of patients were false negatives with negative ctDNA RAS mutation but positive on tissue analysis. The results obtained from ctDNA analysis reside in the probability of the patients having RAS wild-type ctDNA in patients with wild-type RAS tissue tests, which alters the positive and negative predictive values of the investigation.

Microsatellite instability (MSI) testing using ctDNA is used to predict the response to immunotherapy in Lynch syndrome where patients with high MSI from NGS would benefit from immunotherapy due to their extensive immune infiltration and high neoantigenic loads.23 Cai et al24 validated the role of NGS to identify ctDNA in a cohort of 87 CRC patients with tissue-matched controls and demonstrated 94.1% sensitivity and 100% specificity with ctDNA > 0.4%.

Issues with NGS for ctDNA in CRC

NGS using ctDNA is challenging in many ways beginning from the sampling, method of DNA extraction, data analysis, and interpretation by the clinician. The ultimate value of performing molecular testing depends upon avoiding the overutilization of drugs without proven clinical benefit and also the underutilization of well-documented target drug pairs.25

The concordance of ctDNA and tissue DNA is found to be around 80% in metastatic colorectal tumors. However, this varies according to multiple factors in the preanalytic and analytical steps involved in the technique used for molecular testing.

Sampling and DNA Extraction

ctDNA present in the blood is much fragmented with sizes ranging from 100 to 10,000 base pair. Due to the small quantity and size of fragments of the ctDNA, it can be easily lost or degraded during isolation or processing of sample.26 The processing of samples including collection, anticoagulation, handling, transportation, initial processing, and storage is crucial to obtain the accurate results of ctDNA, as these steps contribute either in degradation or contamination.14 To detect genes with low-frequency NGS, the gene of interest must be present in a sufficient quantity above the cutoff limit of detection. The recent discovery of unique molecular identifiers (UMI) facilitates barcoding of the UMI to each DNA, which can be traced back to their origin.

Sequencing and Data Analyzing

Various methods and platforms are available for sequencing and data analysis with reported potential clinical applications. However, there are no systematic criteria to choose an optimal method. The most challenging step in the ctDNA sequencing using NGS is the data analysis of large and complex information. There are no standard criteria for an optimal lower limit of DNA detection to mark significance. There are evidences that NGS not only detects new changes but also estimates the change in variant allele frequency (VAF) during treatment. Patients presenting with a drop in the VAF tend to have good outcomes than those with raising VAF, suggestive of a relapse of disease.27 It is also crucial to fix the cutoff for VAF while interpreting the results of NGS. A high cutoff value for VAF may avoid false-positive sites, but simultaneously increases the chance of false negatives reducing the sensitivity. Moreover, it is difficult to precisely visualize the tumor heterogeneity with the analyzed data, and further studies to identify the impact of the heterogeneity noted with the clinical outcome are needed.

Various factors attribute to the false negativity of NGS using ctDNA in CRC. The amount of ctDNA present in 10 mL of the peripheral blood and the ctDNA fragmentation size play a vital role due to the limited copies of the gene of interest. This is even more challenging when the distribution of VAF is at a much lower level close to the limit of detection. Although various technical methods are being developed to perform ultradeep sequencing of ctDNA, it requires complex algorithms of processing and interpretation.

Nature of Disease

The tumor biology affects the results of sequencing using ctDNA for CRC. False-negativity can be due to the low quantity of ctDNA shedding of certain tumors type or the location of the metastasis itself. CRC patients with liver metastasis have more chance of carrying ctDNA in their peripheral blood than those with spread to nodes or lungs.28 Also, the micrometastasis might shed a higher level of ctDNA due to the high tumor burden than a residual localized disease. Clonal hematopoiesis of undetermined potential also known as CHIP is the progressive accumulation of genetic changes in the hematopoietic cells with aging.29 These remain a confounding factor during analyzing the actual pathogenic variants when tested without an accompanying control sample of DNA from white blood cells (WBCs). Although NGS picks up over 1,000 mutations on liquid biopsy and it was found that most of them are to be CHIP after cross-analysis of the WBC's DNA.30

Concordance with the results of ctDNA and tissue analysis is defined by the identification of identical genomic alteration in a particular gene by both methods. The main causes for discordance in the results were mostly due to biopsy location and timing, heterogenic nature of the tumor tissue, differential DNA shedding, and epigenetic modifications. The rate of concordance was higher for clonal mutations than subclonal mutations.3132 Hence, if the ctDNA-identified mutations were not found in the tissue analysis, it can be considered either as a subclonal variant not found in the collected sample or as a technical error or a CHIP, which can be identified by WBC DNA cross-analysis. However, cross-referencing the results of single nucleotide variants from 1,397 ctDNA samples of metastatic CRC patients with tissue analysis results from three online databases showed a high concordance rate between the two for the top 20 genes.33 The potential advantages and limitations with the use of NGS for detecting ctDNA in CRC have been enumerated below in Table 1.

Recommendations

The complexity involved in the genomic testing using ctDNA warrants the establishment of the National Society of Liquid Biopsy to bridge the gap between translational research and clinical application. The International Society of Liquid Biopsy recommends ctDNA levels and the RAS mutation detected in the plasma to have a prognostic and predictive role in metastatic CRC.34 The preanalytical variable that influences the sensitivity of liquid biopsy using cfDNA includes plasma volume, time intervals between sample collection to plasma separation, centrifugation protocol, purification techniques, and temperature.35

There are multiple commercial platforms available for NGS using ctDNA. They have been tested in a limited number of patients and reported for their accuracy in results. Each platform of NGS has a unique DNA extraction strategy, sensitivity, specificity, spectrum of coverage, the limit of detection, error rate, and data analysis. However, there is no large-scale validation of these methods thereby making them a major hindrance to the clinical application of NGS using ctDNA. To frame a possible consensus on using ctDNA using NGS, the performance of different platforms must be regularly compared and reported under different conditions. Application of ctDNA using NGS might not be cost and time effective while testing for a low number of targets and also by providing overwhelming information while testing for a larger panel.36

The main limitation behind the use of NGS to detect ctDNA mutations was mainly due to the masking of the real ctDNA with intrinsic errors from DNA library preparation and sequencing. The current modern NGS systems have errors at a per-base rate of 10–2 to 10–3. However, clinically relevant mutations were below the designated error level, thereby making many true variants undetectable.37 Hence, development of systems with even lower errors rates becomes a necessity.

Concerning the cost effectiveness, only 37% of the diagnosed patients proceed to receive targeted therapy matching their genetic profile.38 However, considering the potential health state utility gained from averting unnecessary cytotoxic therapy or treatment-related toxicity would account for the cost effectiveness with the utility of NGS in CRC. Compared with the overall cost of the management of the targeted therapy that amounted to US$90,000, the cost of targeted gene panel sequencing per sample wasUS$1,609 only.38 Hence, the cost of the targeted therapy needs to be significantly reduced to utilize NGS in the routine cost-effective management of CRC.

Conclusion

The role of liquid biopsy has been currently well established in the setting of NSCLC, while in the other regions especially in CRC it is evolving to find its place in routine clinical practice. Further improvement will help us in guiding the decision-making for adjuvant chemotherapy in CRC. This will improve the quality of life by avoiding treatment-related toxicity in those who do not need treatment while improving survival in those with high risk such as minimal residual disease positivity. In the metastatic CRC setting, there are comparatively more robust data on using ctDNA for sequencing. However, with more precision and standardization of techniques in ctDNA extraction, preparation of NGS library, and bioinformatics, it is only a matter of time for ctDNA-based NGS to become standard of care in CRC.