Introduction

Locally advanced rectal cancer (LARC) can cause significant challenges in terms of management [1]. Increasingly, patients are having total neoadjuvant treatment (TnT) [2], and the need for better indicators of complete clinical response, especially in borderline cases, is vital [3]. Circulating tumour DNA (ctDNA) has emerged as a promising biomarker in various cancer types, including LARC, offering potential insights into disease progression, treatment response and recurrence [4,5,6,7,8].

ctDNA refers to fragmented DNA shed by tumour cells into the bloodstream [9]. These fragments carry genetic alterations characteristic of the originating tumour, providing a non-invasive means of interrogating tumour biology [10]. The detection and analysis of ctDNA have garnered significant interest in cancer research due to its potential applications in diagnosis, prognostication and treatment monitoring [11]. ctDNA can be isolated from peripheral blood samples and analysed using various techniques, including next-generation sequencing (NGS), digital PCR (dPCR) and targeted amplicon-based assays [12].

NGS is a highly sensitive technique that allows for the comprehensive profiling of ctDNA, enabling the detection of a wide range of genetic alterations, including single nucleotide variants (SNVs), insertions and deletions (indels), copy number variations (CNVs) and structural rearrangements [13]. Conversely, dPCR quantifies the absolute number of target DNA molecules, allowing for precise measurement of ctDNA levels and increased cost-effectiveness when compared to NGS [14]. Alternative approaches include real-time PCR (qPCR), BEAMing and fragment analysis [15,16,17]. By profiling the genomic landscape of tumours through ctDNA analysis, clinicians can gain insights into tumour heterogeneity, clonal evolution and potential therapeutic targets, thereby facilitating personalised treatment approaches [18].

While several clinicopathological factors (MRI and endoscopic response) are currently used to stratify response to treatments in patients with LARC, they have limitations [19]. In addition, traditional prognostic factors such as tumour stage and histological grade can provide circumstantial value about potential disease behaviour and treatment response. There is a need to identify novel biomarkers that can complement existing prognostic tools and enhance risk stratification in LARC [20]. This systematic review aims to comprehensively evaluate the utility of ctDNA as a prognostic biomarker in LARC, exploring its potential to address the unmet clinical needs in this challenging disease context.

Methods

Study design and reporting guidelines

This is a systematic review and meta-analysis of retrospective and prospective cohort studies conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines [21]. Local institutional ethical approval was not required. All authors declare no conflicts of interest. This research received no external funding.

Search strategy

The following databases were searched as part of our systematic review process in April 2024; MEDLINE, PubMed, Embase and Web of Science. The following search strategy was used: (rectal OR rectum OR colorect*) AND (circulating tumour DNA OR ctDNA OR ct-DNA OR circulating free DNA OR cfDNA or cf-DNA). The search was completed on 5th April 2024. The grey literature was also searched as part of our study to identify any further ongoing works of literature, including theses and conference abstracts.

Eligibility criteria

Original studies investigating an association between ctDNA and treatment or oncological outcomes in patients with locally advanced rectal cancer were eligible for inclusion. Case reports, conference abstracts and review articles were excluded.

Data extraction and quality assessment

A database was established utilising the citation management software EndNote X9TM. Independent reviews of search outputs were conducted by two researchers (NOS and HCT). Initially, duplicate entries were eradicated, followed by a screening of study titles to gauge potential relevance. Subsequently, the abstracts of selected studies underwent assessment for eligibility based on predetermined inclusion/exclusion criteria. Excluded studies were categorised by reason within the database. Full texts of eligible abstracts were then scrutinised using identical criteria.

For efficient data extraction and storage, the Cochrane Collaboration’s screening and data extraction tool, Covidence, was employed [22]. Data collection was undertaken independently by two reviewers (NOS and HCT), encompassing study details, design, population, intervention, comparison groups and outcomes. Discrepancies between reviewers were resolved through open discussion, with final arbitration by the senior author (MK).

Potential biases in non-RCT studies were evaluated using the Newcastle-Ottawa Scale (NOS) risk of bias tool, with results tabulated accordingly [23]. This tool assesses studies across various categories, assigning stars to denote quality: 7 stars indicating “very good,” 5-6 stars for “good,” 3-4 stars for “satisfactory,” and 0–2 stars for “unsatisfactory.” Two reviewers (NOS and HCT) independently conducted critical appraisals, with a third reviewer (MK) arbitrating in cases of discordance.

Statistical analysis

Statistical analysis was conducted using Revman Statistical Software (Ver. 5 Copenhagen, Denmark). Generic inverse variance data were presented as hazard ratios (HR) alongside 95% confidence intervals (95% CI). Outcome measures, including mean with standard deviation and median with interquartile range, were documented. Only studies that provided hazard ratios with either confidence intervals or p-values were eligible for inclusion in the meta-analysis. When necessary, outcome variables (mean and SD) were estimated from the median and range using the formula described by Hozo et al. [24]. Heterogeneity was evaluated using I-squared statistics, with values exceeding 50% indicating significant heterogeneity. Statistical significance was defined as a p-value of less than 0.05. A random effects model was employed uniformly.

Systematic review registration

Our systematic review was registered on PROSPERO in April 2024 (ID: 533712).

Results

Search results

The previously outlined literature search yielded a total of 2123 results (Supplementary material S1). After eliminating 281 duplicates, 1842 studies underwent screening. Following the initial screening, 73 full abstracts were meticulously reviewed for eligibility, resulting in 42 being selected for full-text scrutiny. Among these 42 full texts, 22 studies met the eligibility criteria and were consequently included in our analysis [25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46]. Notably, eight of the included studies provided adequate statistical data for incorporation into our quantitative analysis [30, 32,33,34, 36, 42, 44, 45]. A detailed depiction of the literature screening process can be found in Supplementary material S1.

Methodological characteristics and quality of studies

Among the 22 studies included, seventeen were conducted prospectively, four retrospectively, and one did not specify the study design. Regarding the study settings, ten were conducted in a single institute, nine in a multi-institutional setting, and three did not specify. Concerning the assessment of the risk of bias, eleven studies were rated as “very good”, ten as “good”, one as “satisfactory” and none as “unsatisfactory” according to the classification by the Newcastle Ottawa Scale. Supplementary material S2 provides a summary of our risk of bias assessment. The methodological characteristics of the included studies are outlined in Supplementary material S3.

Participant characteristics

The total number of participants across the included studies was 1676. All patients had a diagnosis of locally advanced rectal cancer. Baseline participant characteristics are outlined below in Table 1.

Table 1 Baseline participant characteristics

ctDNA details and study findings

ctDNA assay type, sequencing method and collection time points varied across included studies. Fourteen studies measured a mutation-specific panel, six measured cfDNA concentration and the remaining two measured promoter genes or multiple assays respectively. In regard to the sequencing method, next-generation sequencing (NGS) was the most commonly utilised method accounting for ten studies, followed by polymerase chain reaction (PCR) (n = 9) and direct fluorescent antibody (DFA) (n = 2). One study failed to report the sequencing method.

In terms of approach, eleven studies used an agnostic, ten used targeted and one study used both approaches. Agnostic approaches do not rely on prior knowledge of specific mutations but instead analyse the entire ctDNA for any alterations that may be present [47]. These approaches are valuable in settings where comprehensive genomic profiling is necessary, particularly for identifying novel or unexpected mutations. They are beneficial in research and clinical scenarios where a wide array of genetic alterations needs to be detected to inform treatment decisions or understand tumour heterogeneity [48]. Conversely, targeted approaches focus on detecting specific, known mutations that are of clinical interest [49]. They often use panels designed to target commonly mutated genes in particular cancers. Targeted approaches are highly relevant in clinical settings where specific genetic alterations are known to influence prognosis or guide targeted therapies [50]. They are typically more cost-effective and faster than agnostic approaches, making them suitable for routine clinical use, particularly when monitoring known mutations for treatment response or minimal residual disease. Both approaches have their place in the evaluation of ctDNA [47]. Agnostic methods provide a broad view of the genetic landscape, which can uncover new targets for therapy, while targeted methods allow for focused, efficient and cost-effective monitoring of known genetic alterations.

Tables 2 and 3 below outline ctDNA panel details, collection timepoints and main study findings.

Table 2 ctDNA details
Table 3 Study findings

Meta-analysis

Eight of the included studies provided sufficient statistical data to be included in our quantitative analysis [30, 32,33,34, 36, 42, 44, 45]. We investigated for an association between recurrence-free survival and the presence of ctDNA at several timepoints in a meta-analysis, the results of which are demonstrated in Fig. 1 below. The pooled hazard ratio for ctDNA presence after completion of neoadjuvant therapy when compared to patients who were not found to have detectable ctDNA was 8.87 (95% CI 4.91–16.03, p = < 0.0001)). Similarly, the pooled hazard ratio for ctDNA presence post-operatively compared to those without detectable ctDNA was 15.15 (95% CI 8.21–27.95, p = < 0.0001)). These results indicate an increased risk for recurrence in patients with LARC with detectable ctDNA in either the post-neoadjuvant therapy or post-operative periods.

Fig. 1
figure 1

Top: post-NAT ctDNA meta-analysis. Bottom: post-operative ctDNA meta-analysis

Ongoing research

Several trials are ongoing to further determine the prognostic capabilities of ctDNA in patients with LARC. The DYNAMIC-RECTAL trial (ACTRN12617001560381) is a multi-centre randomised controlled phase 2 trial aiming to investigate the prognostic benefits of ctDNA detection post-operatively to guide the need for adjuvant treatment in patients with LARC. Eligible patients underwent neoadjuvant chemoradiation and total mesenteric excision, and the primary endpoint was adjuvant chemotherapy use. Preliminary results suggest that fewer patients in the ctDNA-guided arm required adjuvant therapy with a lower risk of recurrence in patients with undetectable post-operative ctDNA [51].

The SYNCOPE study (NCT04842006) is a randomised controlled treatment trial aiming to reduce both over-treatment and metastatic disease progression in patients with rectal cancer. Participants with high-risk features will be randomised into two groups: early systemic chemotherapy followed by ctDNA and organoid-guided adjuvant therapy, or conventional treatment. Primary outcomes include RFS and ctDNA positivity rates in post-operative patients within the conventional treatment arm not exposed to chemotherapy.

The REVEAL study (NCT05674422) is a prospective multi-institutional study evaluating response to total neoadjuvant therapy (TNT) with liquid biopsy in eligible patients with LARC. The study aims to investigate the role of ctDNA in the prediction of relapse in this cohort of patients followed by a watch and wait programme or TME depending on the response to initial treatment.

Finally, CINTS-R (NCT05601505) is a multi-institutional randomised controlled trial aiming to evaluate outcomes of ctDNA-guided neoadjuvant treatment in patients with LARC. Treatment groups will receive either NCRT followed by immunotherapy, NCRT alone or TNT depending on mutation status following NGS of tumour tissue and variant allele frequency (VAF) of ctDNA. Control groups will receive standard NCRT only.

Discussion

Our systematic review of the current available literature demonstrates the significant potential of ctDNA as a prognostic biomarker in patients with LARC. Particularly, ctDNA detection post-neoadjuvant therapy or post-operatively was associated with an increased risk of recurrence, suggesting its utility in predicting disease progression and informing treatment decisions. Meta-analysis of available data further supported these findings, indicating a significantly higher hazard ratio for recurrence in patients with detectable ctDNA at these critical time points. These results align and extend upon existing literature on ctDNA in various cancer types, highlighting its potential as a non-invasive biomarker for monitoring disease burden and treatment response in a multitude of malignancies [4,5,6,7,8].

Despite its promise, ctDNA analysis is not without limitations [52]. Technical challenges, such as low concentrations of ctDNA in circulation and the need for sensitive detection methods, can impede accurate assessment. Similarly, tumour heterogeneity and clonal evolution may further complicate interpretation, potentially leading to false-negative or false-positive results [53]. Moreover, the lack of standardised protocols for ctDNA analysis and variability in assay performance across studies underscore the need for methodological standardisation and validation [54].

While our systematic review provides valuable insights into the role of ctDNA in LARC, several limitations warrant consideration. Firstly, studies included in our review demonstrated heterogeneity in terms of methodology, patient populations and outcome measures, which may have introduced bias and confounded interpretation. Additionally, the reliance on published data may have led to publication bias, with studies reporting significant findings more likely to be included. Despite these limitations, the findings of our review have significant implications for future research and clinical practice. The consistent association between ctDNA detection and adverse outcomes in LARC suggests its potential as a prognostic biomarker for risk stratification and treatment decision-making. Integrating ctDNA analysis into routine clinical practice could facilitate personalised treatment strategies, including the identification of high-risk patients who may benefit from intensified therapy or closer surveillance [55]. Furthermore, ongoing research efforts aimed at refining ctDNA-based assays and elucidating the underlying biological mechanisms driving ctDNA release and clearance are crucial for maximising its clinical utility [56].

Future research in this field should prioritise several key areas to address current knowledge gaps and optimise the implementation of ctDNA analysis in clinical practice. Firstly, large-scale prospective studies with standardised methodologies are needed to validate the prognostic utility of ctDNA across diverse patient populations and treatment settings [57]. Additionally, efforts to standardise ctDNA assays, including sample collection, processing and analysis protocols, are essential to ensure the reproducibility and comparability of results. Finally, collaborative efforts to establish international consortia and biobanks for ctDNA research could facilitate data sharing and accelerate progress towards clinical implementation [58].

Conclusion

Our systematic review provides evidence supporting the prognostic utility of ctDNA in patients with LARC, particularly in identifying patients at higher risk of disease recurrence in the post-neoadjuvant and post-operative periods.