Treatment stratification and prognosis assessment using circulating tumor DNA in locally advanced rectal cancer: A systematic review and meta‐analysis

Abstract Background Circulating tumor DNA (ctDNA) is an emerging biomarker for locally advanced rectal cancer (LARC), giving hope for stratified treatment. As the completed studies have small sample sizes and different experimental methods, systematic review and meta‐analysis were performed to explore their role in predicting pathological complete response (pCR), tumor recurrence, and prognosis. Methods PubMed, Embase, and the Web of Science were searched for potentially eligible studies published up to September 6, 2022. Pooled relative risk (RR) was calculated to predict pCR and tumor recurrence, and pooled hazard ratio (HR) was calculated to evaluate the prognosis of overall survival (OS), recurrence‐free survival (RFS), and metastasis‐free survival (MRS). Results Twelve studies published between 2018 and 2022 included 931 patients, and 2544 serum samples were eventually included in the meta‐analysis. The pooled revealed that ctDNA‐negative patients were more likely to have a pCR (RR = 1.64, 95% confidence interval [CI]: 1.26–2.12). The pooled revealed that ctDNA‐positive patients were at high risk of recurrence (RR = 3.37, 95% CI: 2.34–4.85) and had a poorer prognosis for OS (HR = 3.03, 95% CI: 1.86–4.95), RFS (HR = 7.08, 95% CI: 4.12–12.14), and MRS (HR = 2.77, 95% CI: 2.01–3.83). Conclusion ctDNA may be useful for stratifying treatment and assessing prognosis in patients with LARC, but its clinical application still needs to be confirmed in a prospective multicenter study with large samples.

advanced rectal cancer (LARC). 3,40][11] Currently, clinical complete response (cCR) is defined as the inability to detect tumors by including magnetic resonance imaging (MRI), endoscopy, and digital rectal examination to guide the selection of patients suitable for NOM. 12,13However, the agreement between cCR and pCR is not satisfactory, so there is a clinical need for a biomarker that can accurately predict pCR. 14ectal cancer has a high recurrence rate within 5 years after radical surgery, with Stages II and III recurrence rates of 30% and 50%, respectively. 15Early detection of tumor recurrence can facilitate timely treatment and thus improve the patient's prognosis, but the methods currently used in clinical practice to detect tumor recurrence have low sensitivity. 16,17MRI is commonly used to diagnose LARC and monitor tumor recurrence with a sensitivity of only 68.3% for local recurrence in LARC patients after nCRT. 18In addition, carcinoembryonic antigen (CEA) is commonly used clinically to monitor the recurrence of colorectal cancer, but its sensitivity and specificity are still unsatisfactory. 19,20In conclusion, there is still a need to explore new biomarkers for the prediction of LARC recurrence.
Circulation tumor DNA (ctDNA) is the release of apoptotic or necrotic tumors into the bloodstream with genetic variation, and there is a high degree of concordance between the genomic alterations found in tumor tissue. 21,22With easy sampling, noninvasive, real-time, and multiple executions, ctDNA has great potential to guide treatment selection, tumor recurrence monitoring, and prognostic monitoring. 23,24Positive ctDNA after radical surgery for curative purposes in esophageal adenocarcinomas and non-small cell lung cancer carries a high risk of recurrence. 25,26In addition, ctDNA-guided categorical therapy for colon cancer has reduced adjuvant chemotherapy. 27 number of studies have explored the role of ctDNA in predicting pCR, tumor recurrence risk, and prognosis in LARC.However, the evidence is still insufficient for direct clinical application due to the small sample size and the different experimental approaches.[28][29][30][31][32][33][34][35][36][37][38][39] In addition, published studies on the role of ctDNA in LARC have shown inconsistent results.The aim of this study was to systematically review and meta-analyse the role of ctDNA in predicting pCR, risk of recurrence, and prognosis in patients with LARC.

| Literature search
This meta-analysis was carried out according to preferred reporting items for systematic reviews and metaanalyses guidelines. 40This meta-analysis was registered at the International Prospective Register of Systematic Reviews with the registration number CRD42022367608.We searched PubMed, Embase, and Web of Science for studies on ctDNA predicting pCR, risk of recurrence, and prognosis in LARC patients with a search deadline of September 6, 2022.In addition, the investigators searched for references in the included studies to prevent omissions.Inclusion criteria were as follows: (a) patients were diagnosed with LARC; (b) investigation of the role of ctDNA in the prediction of pCR, tumor recurrence risk, and prognosis; and (c) studies that provided enough information to calculate relative risk (RR) for predicting pCR, tumor recurrence, and hazard ratio (HR) for prognosis.The exclusion criteria were as follows: (a) conference abstracts, reviews, letters, case reports, and nonclinical studies; (b) studies for which relevant raw data were not available; (c) studies involving comorbidities with other malignant diseases; (d) studies involving colon cancer; and (e) studies detecting cell-free DNA.The study search strategy used a combination of subject terms and free words, and the search strategy was as follows: #1: Circulating Tumor DNA OR ctDNA OR (Circulating Tumor AND DNA) OR (Tumor DNA AND Circulating) OR Cell Free Tumor DNA OR (Cell-Free Tumor AND DNA) OR (Tumor DNA AND Cell-Free); #2: Locally advanced rectal cancer OR Rectal cancer OR Rectal Neoplasms; #3: Pathological complete response OR Chemoradiotherapy OR Neoadjuvant chemoradiotherapy OR Prognosis OR Minimal residual disease OR Recurrence risk; #4: #1 and #2 and #3.The Supporting Information contained detailed search process.

| Data extraction
Two investigators (J.M., R.W.) independently collected data for each included study by developing a data extraction form, and if the comparison of the extracted data did not agree, the disagreement was resolved by consulting a third person (X.H.).The data extract included basic information about the included studies as follows: first author, year of publication, country, patient count, average age, gender, number of serum samples, sample type, testing methods, number of genes, number of centers, study design, TNM staging, clinical stage, median follow-up, ctDNA testing time, definition of pCR, long-term endpoint definition, and treatment strategy.For pCR and the risk of tumor recurrence, the number of ctDNA-positive patients and the number of ctDNA-negative patients were extracted.Regarding the prognostic indicators overall survival (OS), recurrence-free survival (RFS), and metastasisfree survival (MRS), HR as well as 95% confidence intervals (95% CIs) were extracted for each study.If the relevant data are not available from the original study, we will contact the corresponding author of the original study to obtain the raw data.We used the Newcastle-Ottawa scale to rate the included studies from 0 to 9.

| Statistical analysis
The effect sizes for pCR and risk of tumor recurrence were assessed by the Mantel-Haenszel method as RR and 95% CI, and for prognosis as HR and their 95% CI for OS, progression-free survival (PFS), and MRS.A chi-squared Q test and I 2 coefficients were also used to quantify the degree of heterogeneity between included studies, with p < 0.05 considered a statistically significant difference using a random effects model; otherwise, a fixed effects model was used.Subgroup analysis was performed to investigate the utility of ctDNA testing at various time points in predicting pCR, tumor recurrence risk, and prognosis, as well as to investigate the sources of heterogeneity in this study.There were four subgroups based on the time point of ctDNA testing: baseline, on-chemoradiotherapy (on-CRT), presurgery, and postsurgery groups.In addition, we performed sensitivity analyses to test the stability of the meta-analysis results, and finally, the Egger test and harbor test were used to assess whether there was publication bias in this study.The overall effect test was statistically significant at p < 0.05, and all analyses were completed using STATA 17.0 software.

| Characteristics of eligible studies and quality assessment
An initial selection of 549 articles was achieved after careful searching of the databases, as detailed in Figure S1.A total of 931 patients and 2544 serum samples were included in the meta-analysis of 12 studies published between 2018 and 2022, in which the study characteristics and demographics are summarized in Table 1.The timing of ctDNA testing, the definition of pCR, the definition of long-term endpoints, and treatment strategies were summarized in Table S1.The results of the assessment of the quality of the included studies by means of the Newcastle-Ottawa scale are shown in Table S2.In addition, the Egger test and Harbord test were used to assess publication bias, and the results showed no publication bias, and sensitivity analysis showed that none of the studies could significantly affect the pooled effect (see Figures S2-S7 for details).

| Prediction of RFS by ctDNA
36,38 For the estimation of RFS, the results of disease-free survival and PFS were included.ctDNA-positive patients had a poorer RFS in Figure 4 (HR = 7.08, 95% CI: 4.12-12.

| Prediction of MRS by ctDNA
A total of four studies were pooled to analyze the prognosis of ctDNA predicting MRS, and ctDNA-positive patients had a poorer prognosis for MRS in Figure S9 (HR = 2.77, 95% CI: 2.01-3.83,p < 0.05). 30,32,34,35A total of three studies were pooled for analysis at the time of baseline for MRS predicted by ctDNA, with Appelt and Zhou's study using different methods for setting ctDNA positivity thresholds, each of which was included separately in the pooled analysis. 30,34,35At baseline, ctDNA-positive patients had a poorer prognosis for MRS (HR = 1.31, 95% CI: 1.10-1.56,p < 0.05).Two and three studies were pooled for analysis at on-CRT and presurgery, respectively, with Zhou's study using different methods for setting ctDNA positivity thresholds and Khakoo's study including different time points for detecting ctDNA, each of which was included separately in the pooled analysis. 32,34,35Patients who had positive ctDNA during CRT and before surgery had a poor prognosis for MRS (HR = 3.66, 95%CI:1.80-7.45,p < 0.05; HR = 11.53,95% CI: 5.97-22.29,p < 0.05).Only Zhou's study evaluated ctDNA predictive MRS at postsurgery, where ctDNA positivity thresholds were set using different methods, each of which was included separately in the pooled analysis. 35Patients with positive ctDNA at postsurgery had a poorer prognosis for MRS (HR = 24.73,95% CI: 2.49-245.75,p < 0.05).

| DISCUSSION
The standards of treatment recommended by guidelines for patients with LARC are TME and ACT after nCRT, and although most patients benefit from treatment as recommended by guidelines, each patient's tumor is highly heterogeneous and responds to treatment in a highly different way. 41,42Patients who do not respond well to treatment not only do not benefit from standard treatment but also suffer from the toxic side effects of chemoradiotherapy, and the damage to normal body functions caused by surgery. 43ctDNA has been shown to be used to guide patient treatment selection, detect tumor recurrence, and predict prognosis, bringing hope for individualized and precise treatment of tumor patients. 44,45Our metaanalysis demonstrated that ctDNA-negative patients are more prone to pCR, have a lower risk of tumor recurrence, have a better prognosis, and may be considered for NOM as a treatment option in this group of patients.Patients with positive ctDNA had a higher risk of tumor recurrence, and such patients may require intensive follow-up to receive timely treatment.ctDNA could also be used to predict LARC prognosis, with ctDNA-positive patients having a worse OS, a worse RFS, and a worse MRS.
Although TME is recommended by guidelines for the treatment of LARC, it can cause bowel, sexual, and genitourinary dysfunctions that can seriously affect the quality of life of patients. 46,47In addition, a 14-year multicenter randomized study compared the long-term effects of TME and TME plus radiotherapy on bowel function by F I G U R E 2 Forest plot of pooled effect for relative risk in circulating tumor DNA prediction of locally advanced rectal cancer recurrence.low anterior resection syndrome score (LARS score), with major LARS indicating severe bowel dysfunction.Major LARS was reported by 46% of all patients (56% radiotherapy plus TME vs. 35% TME), and radiotherapy plus TME (odds ratio = 3.0; CI, 1.3-6.9)increased the risk of major LARS. 469][50][51] Patients with LARC presented a variable response to treatment after nCRT, with 24.4% presenting with pCR, approximately 45% with tumor downstaging and the remaining patients presenting with no or partial response. 8,52The critical aspect of selecting NOM is the correct identification of those patients who will develop pCR after nCRT.This meta-analysis showed that ctDNA-negative patients were more prone to developing pCR at on-CRT and presurgery.
Accurate assessment of the risk of relapse after LARC treatment can prevent over-treatment of patients at low risk of relapse and undertreatment of patients at high risk of relapse.In addition, the high rate of metastatic recurrence in patients after LARC surgery is an important reason for their high mortality rate.It is therefore clinically important to find a biomarker that can detect tumor recurrence in a timely manner with high accuracy and simplicity.Colonoscopy and CEA are routinely used to assess recurrence after LARC treatment but have shortcomings in clinical practice.7][58] ctDNA provides information on tumor-specific genetic alterations, including single-nucleotide site variants, copy number alterations, gene fusions, and epigenetic alterations, and is noninvasive and easy to use multiple times, making it an ideal biomarker for monitoring tumor recurrence.0][61] In addition, the Liu study showed that the area under the receiver operating characteristic curve for ctDNA predicting LARC recurrence was 0.87 higher than the CEA (0.52) and CA199 (0.49). 32The pooled analysis of this study showed a higher risk of tumor recurrence for ctDNA positivity at presurgery, suggesting that this group of patients may respond poorly to preoperative nCRT treatment and that surgical and postoperative treatment should be intensified in this group of patients.At postsurgery, ctDNA-positive patients also had a higher risk of tumor recurrence, suggesting that these patients may require boosted postoperative ACT and frequent follow-ups.Unfortunately, ctDNA did not predict tumor recurrence at baseline.ctDNA was a good predictor of OS, and patients who tested positive for ctDNA had a worse prognosis for OS.However, ctDNA testing at different time points showed different results in predicting OS, and ctDNA testing at baseline was not associated with OS.Positive ctDNA at presurgery had a poor OS, so this group of patients may not benefit from standard treatment and may require intensive treatment.RFS reflects tumor progression, which is largely responsible for tumor-related deaths, and the need to change treatment regimens after tumor progression, making ultimate survival dependent on subsequent treatment regimens.Our pooled study demonstrated that ctDNA-positive patients had worse RFS.In addition, presurgical ctDNA and postsurgical ctDNA detected recurrence 10.2 and 2.6 months ahead of imaging, respectively. 32,34In the subgroup analysis, ctDNA-positive patients had poorer RFS at presurgery and postsurgery.Patients who were ctDNA-positive at presurgery and postsurgery were more prone to developing progression, so these patients need to be monitored more closely and may require a change of treatment regimen after progression.Metastatic recurrence in patients after LARC surgery is 25% and is a major contributor to their mortality, making MFS an important indicator to assess the prognostic survival of LARC patients. 62,63This pooled study demonstrated that ctDNApositive patients had poorer MRS.In subgroup analyses at baseline, on-CRT, presurgery, and postsurgery, ctDNApositive patients all had poorer MRS, so that positive ctDNA patients were more likely to have metastases regardless of when ctDNA was tested in LARC treatment.In addition, the association between positive ctDNA and distant tumor metastases strengthened as the time point for ctDNA testing moved forward.In conclusion, ctDNA was a good indicator to assess the prognosis of LARC and stronger treatment and closer follow-up were needed for ctDNA-positive patients.Although first meta-analysis demonstrated the potential usefulness of ctDNA in predicting pCR, tumor recurrence, and prognosis in LARC, the following shortcomings exist.First, the number of studies and sample size included in this meta-analysis as well as in the subgroup analysis was relatively small.Second, there was heterogeneity in the meta-analysis as well as in the subgroup analysis.Third, most of the studies included in the assessment of prognosis had short follow-up periods, and in particular, the follow-up period for assessing OS was not long enough.
In conclusion, this meta-analysis demonstrated that ctDNA-negative patients are more prone to pCR, have a lower risk of tumor recurrence, and may be considered for NOM as a treatment option.This study also showed a higher risk of tumor recurrence in ctDNA-positive patients at presurgery and postsurgery, suggesting that these patients may respond poorly to treatment and should be closely followed.ctDNA could also be used to predict LARC prognosis, with ctDNA-positive patients having a worse OS, a worse RFS, and a worse MRS.The sample size of known studies is small, and large-scale multicenter prospective studies are still needed to determine the role and clinical application of ctDNA in LARC.

T A B L E 1
Abbreviations: BEAMing, bead emulsion amplification magnetic; dd PCR, droplet digital polymerase chain reaction; IQR, interquartile range; NA, not available; NGS, next-generation sequencing; PCR, polymerase chain reaction.

F I G U R E 1
Forest plot of pooled effect for relative risk in circulating tumor DNA prediction of pathological complete response.

F I G U R E 3
Forest plot on circulating tumor DNA prediction of overall survival with a pooled effect hazard ratio.| 17941I et al.

F I G U R E 4
Forest plot on circulating tumor DNA prediction of progression-free survival with a pooled effect hazard ratio.