Considering Tumour Sidedness is Critical for Accurate Interpretation of Liquid Biopsy Results in Metastatic Colorectal Cancer Authors: Marta Toledano-Fonseca, Icíar de la Fuente, María José Ortiz-Morales, Pablo Cano-Ramírez, Gema Pulido-Cortijo, María Teresa Cano-Osuna, María Victoria García-Ortiz, María Auxiliadora Gómez-España, Antonio Rodríguez-Ariza, Enrique Aranda Published: 2026 ABSTRACT BackgroundColorectal cancer is one of the most common and deadly malignancies worldwide. Traditionally, all colon tumours have been treated as a single entity, but growing evidence shows that tumour behaviour differs markedly according to tumour sidedness. Right-sided and left-sided tumours exhibit distinct molecular profiles, prognoses and therapeutic responses. A liquid biopsy is a minimally invasive approach to monitor tumour dynamics in real time, offering valuable prognostic and predictive insights into metastatic colorectal cancer. However, despite increasing research on liquid biopsy-based biomarkers, the potential impact of tumour sidedness on their interpretation has not been systematically explored.MethodsWe evaluated circulating RAS mutational status, RAS mutant allele fraction and cell-free DNA concentration and fragmentation in plasma from 232 patients with metastatic colorectal cancer.ResultsOur findings reveal that tumour sidedness substantially modulates the prognostic value of these circulating biomarkers. In left-sided tumours, worse outcomes were associated with the presence of circulating RAS mutations, higher cell-free DNA concentrations and elevated carcinoembryonic antigen levels. In contrast, among right-sided tumours, only mutant allele fraction levels revealed significant prognostic relevance.ConclusionsOur results demonstrate that tumour sidedness critically influences liquid biopsy interpretation in metastatic colorectal cancer, underscoring the need to incorporate the primary tumour location into the clinical assessment of circulating biomarkers.Supplementary InformationThe online version contains supplementary material available at 10.1007/s40291-026-00856-3. Key Points Tumour sidedness critically influences liquid biopsy interpretation in metastatic colorectal cancer.The presence of circulating RAS mutations, higher cell-free DNA concentrations and elevated carcinoembryonic antigen levels in left-sided tumours were associated with a worse prognosis.In right-sided tumours, only mutant allele fraction levels revealed significant prognostic relevance. Tumour sidedness critically influences liquid biopsy interpretation in metastatic colorectal cancer.The presence of circulating RAS mutations, higher cell-free DNA concentrations and elevated carcinoembryonic antigen levels in left-sided tumours were associated with a worse prognosis.In right-sided tumours, only mutant allele fraction levels revealed significant prognostic relevance. Introduction Colorectal cancer (CRC) is the third most common cancer worldwide and is the second tumour type responsible for cancer-related deaths [1]. The 5-year survival rate is around 90% when CRC is diagnosed at early stages but drastically decreases to 14% when a tumour is detected at advanced stages [2]. Approximately 20–25% of patients with CRC are diagnosed with advanced disease, with the liver, lungs, peritoneum, distant lymph nodes, bones and nervous system being the most common metastatic sites [3, 4]. The right and left colon sides of the colon diverge in their embryological origin, likely contributing to the epidemiological and prognostic differences observed between right-sided and left-sided CRC. The right colon originates from the midgut with right-sided tumours located in the appendix, cecum, ascending colon, hepatic flexure or transverse colon. In contrast, the left colon originates from the hindgut and left-sided tumours locate in the splenic flexure, descending colon, sigmoid colon or rectum [5, 6]. At early stages, right-sided tumours have better prognosis than left-sided CRC, whereas left-sided tumours seems to have better survival prognosis in advanced stages. Moreover, right-sided and left-sided tumours tend to metastasize to different organs: right-sided tumours often metastasize to the peritoneum, while left-sided tumours predominantly metastasize to the liver and lungs [7, 8]. Right-sided and left-sided CRC also differ in their mutational profiles: KRAS and BRAF mutations, which are associated with significantly worse survival in the metastatic setting, are more commonly detected in right-sided tumours, while APC and TP53 are more frequently detected in left-sided CRC. Additionally, amplification in ERBB2 and EGFR genes are more frequent in left-sided tumours. All these factors must be taken into consideration in the choice of targeted therapy [9, 10], and recent clinical guidelines recommended first-line anti-epidermal growth factor receptor treatment only in left-sided RAS wild-type tumours while anti-vascular endothelial growth factor therapy provide a better benefit in right-sided CRC [11, 12]. This recommendation is supported by the biology of RAS mutations, present in approximately 40–50% of CRC cases. While its prognostic significance remains debated, RAS mutational status is a validated marker for predicting response to anti-epidermal growth factor receptor (EGFR) therapy [13]. Epidermal growth factor receptor-blocking antibodies, such as cetuximab and panitumumab, prevent ligand binding and receptor activation, but they fail to block EGFR signalling if downstream effectors, such as RAS, are constitutively activated by mutations [14]. A liquid biopsy has emerged as a promising, minimally invasive approach that delivers in real time, comprehensive molecular information from both primary tumour and metastatic lesions [15]. The carcinoembryonic antigen (CEA) is the only blood-based tumour marker routinely used in clinic for the management of patients with CRC. However, CEA possesses limited sensitivity and specificity, and CEA levels are also elevated by other malignancies and even by benign conditions [16]. Circulating cell-free DNA (cfDNA), which is one of the main liquid biopsy components, can be quantitatively analysed by measuring cfDNA concentration, but also qualitatively by analysing the presence of tumour-specific genetic alterations, such as RAS mutations in CRC. Circulating levels of cfDNA are higher in patients with CRC than in healthy individuals, and the proportion of smaller fragments of cfDNA is also increased [17, 18]. Moreover, higher cfDNA levels have been associated with advanced tumoural stages, poor prognosis and reduced response to therapy [18, 19]. Despite numerous studies investigating the prognostic value of liquid biopsy-based biomarkers in metastatic CRC (mCRC), the potential impact of tumour sidedness on their interpretation have not been specifically addressed. Therefore, this study was designed to evaluate the prognostic relevance of cfDNA concentration, plasma RAS mutational status and CEA levels in mCRC, with a particular emphasis on how their interpretation varies according to tumour sidedness. Methods Patients This was a retrospective analysis of liquid biopsy samples from 232 patients diagnosed with mCRC in the Reina Sofía Hospital (Córdoba, Spain) between 2015 and 2022. Eligible patients were aged older than 18 years with histologically confirmed mCRC. These patients were not treated by chemotherapy (CTx) or radiotherapy and they were not candidates to surgery before the enrolment. All patients gave their informed consent for their inclusion. The study was conducted in accordance with the Declaration of Helsinki, and the protocol (FCO-BIOPLIQ2015) was approved by the Ethics Committee of Córdoba. Procedures for Sample Analyses Plasma was obtained from 20 mL of blood collected using Streck cell-free DNA BCT™ tubes (Streck Corporate, Omaha, NE, USA). Blood samples were centrifuged at 1600×g during 10 minutes at room temperature to separate plasma, followed by centrifugation at 6000×g during 10 minutes at room temperature to remove any possible remaining cells. Plasma samples were then aliquoted in cryotubes and stored at – 80 °C until use. Cell-free DNA was extracted from 3 mL of plasma using the QIAamp Circulating Nucleic Acid Kit and the QIAvac 24 Plus vacuum system (Qiagen, Hilden, Germany) following the manufacturer’s instructions. Cell-free DNA was quantified by fluorimetry using the Quantifluor dsDNA kit in the Quantus fluorometer (Promega, Madison, WI, USA) and cfDNA fragmentation was analysed using the automated electrophoresis system Agilent 2200 TapeStation (Agilent, Santa Clara, CA, USA) with the High Sensitivity D1000 ScreenTape Assay, which allows the analysis of 35–1000 bp fragment size DNA. In our analysis, cfDNA fragmentation was defined as the percentage of shortest fragments to total cfDNA. RAS mutation status in cfDNA and RAS mutant allelic fraction (MAF) in plasma were determined using BEAMing (beads, emulsion, amplification and magnetics) technology with the OncoBEAM™ RAS Assay (Sysmex Inostics GmbH, Hamburg, Germany), which detects 34 mutations in KRAS/NRAS codons 12, 13, 59, 61, 117 and 146. This approach allows reliable detection of MAF < 0.1% in cfDNA [20]. RAS mutations in tumour tissue were analysed by standard-of-care procedures validated in our hospital. Specifically, therascreen KRAS RGQ PCR Kit (Qiagen) or the Idylla™ platform (Biocartis, Malinas, Belgium) were employed. If the results indicated wild-type KRAS, testing for NRAS mutations was mandatory. Carcinoembryonic antigen levels were measured using a standard radioimmunoassay test in the Clinical Laboratory Department of our hospital. Statistical Analysis Statistical analyses were performed using SPSS Statistic 22.0.0 and GraphPad Prism 6.0 Software. Overall survival (OS) was calculated from the date of diagnosis to the date of death. Progression-free survival was calculated from the start date of therapy until disease progression or death. The survival rates were estimated using the Kaplan–Meier method and the log-rank test was used to identify the prognostic variables. The Mann–Whitney test was used to compare two and more than two groups, respectively. The Chi-square test was used for testing the association between categorical variables. All statistical tests were considered significant when p < 0.05. The optimal cut-off values were selected with the median of all individuals regardless of the location of the primary tumour. When the optimal cut-off was not able to separate statistically the groups according to the Kaplan–Meier analysis, the R2 Genomics Analysis and Visualization Platform (http://r2.amc.nl) was used to find a cut-off value, using the Kaplan Scan (KaplanScan) feature, based on statistical testing. The Kaplan scanner separates the samples of a dataset into two groups based on values of variable of interest. In the order of values, it uses every increasing value as a cut-off to create two groups and test the p-value in a log-rank test. Groups were required to include at least 20% of the patients to avoid highly heterogeneous groups. The cut-off value with prognostic relevance for OS was also tested for prognosis of progression-free survival. Results Clinical-Pathological Characteristics According to the Primary Tumour Location The primary tumour was located in the right and left (including rectum) side of the colon in 27.6% and 69.8% of patients, respectively, and 2.6% of the cases had right and left synchronous tumours. These latter patients were excluded from future analyses. Comparing clinical-pathological characteristics according to tumour sidedness, we found that a higher proportion of patients in the left-sided group had liver metastases (p = 0.007), while the right-sided group included more patients aged older than 66 years (p = 0.0018), with a worse Eastern Cooperative Oncology Group [ECOG] (p = 0.0279) and a higher frequency of RAS mutations detected in tissue (p = 0.0145) (Table 1). Table 1Clinical characteristics according to the primary tumour locationLeft-sided CRCRight-sided CRCp-ValueSex, n (%) Male111 (68.5)37 (57.8)0.1619 Female51 (31.5)27 (42.2)Age, n (%), years ≤ 6694 (58)22 (34.4)0.0018 > 6668 (42)42 (65.6)ECOG, n (%) 056 (34.6)14 (21.8)0.0279 199 (61.1)42 (65.6) 27 (4.3)8 (12.5)Number of metastatic locations, n (%) 115 (9.3)3 (4.7)0.2502 ≥ 2147 (90.7)61 (95.3)Metastatic location,, n (%) Hepatic140 (86.4)45 (70.3)0.0070 Non-hepatic22 (13.6)19 (29.7)RAS status in tissue, n (%) WT82 (51.3)19 (32.2)0.0145 MUT78 (48.7)40 (67.8)CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, MUT RAS mutated, WT wild typeStatistically significant p-values are indicated in bold Table 1Clinical characteristics according to the primary tumour locationLeft-sided CRCRight-sided CRCp-ValueSex, n (%) Male111 (68.5)37 (57.8)0.1619 Female51 (31.5)27 (42.2)Age, n (%), years ≤ 6694 (58)22 (34.4)0.0018 > 6668 (42)42 (65.6)ECOG, n (%) 056 (34.6)14 (21.8)0.0279 199 (61.1)42 (65.6) 27 (4.3)8 (12.5)Number of metastatic locations, n (%) 115 (9.3)3 (4.7)0.2502 ≥ 2147 (90.7)61 (95.3)Metastatic location,, n (%) Hepatic140 (86.4)45 (70.3)0.0070 Non-hepatic22 (13.6)19 (29.7)RAS status in tissue, n (%) WT82 (51.3)19 (32.2)0.0145 MUT78 (48.7)40 (67.8)CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, MUT RAS mutated, WT wild typeStatistically significant p-values are indicated in bold Clinical characteristics according to the primary tumour location CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, MUT RAS mutated, WT wild type Statistically significant p-values are indicated in bold A left-sided location of the primary tumour was associated with better OS (p = 0.0313), while no significant association was observed between primary tumour location and progression-free survival. Regarding clinical characteristics, ECOG showed a significant prognostic value independently of primary tumour location, while the metastatic location presented no association with prognosis on either side of the colon (Table 2). Among patients with left-sided tumours, those aged older than 66 years had a shorter OS (481 vs 739 days; p = 0.033) as did those with RAS mutations in tissue (622 vs 776 days; p = 0.0065), while sex was not statistical significant (Table 2). Conversely, among patients with right-sided tumours, women had a worse prognosis than men (346 vs 685 days; p = 0.0232), while neither age nor RAS mutational status in tissue showed significant associations with survival according to the primary tumour location (Table 2). Table 2Prognostic impact of clinical characteristics depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersHR (95% CI)p-ValueBiomarkersHR (95% CI)p-ValueSexSex Malea0.69 (0.46–1.02)0.0624 Malea0.52 (0.29–0.91)0.0232 Female FemaleAge (years)Age (years) ≤ 66a0.58 (0.40–0.83)0.0033 ≤ 66a0.65 (0.39–1.10)0.1061 > 66 > 66ECOGECOG 0a 0a0.3966 10.50 (0.36–0.73)0.0003 10.76 (0.42–1.41)0.13 (0–0.02)< 0.00010.21 (0.02–0.21)< 0.0001 2 2Number of metastatic locationsNumber of metastatic locations 1a0.71 (0.41–1.22)0.2152 1a0.50 (0.20–1.28)0.1485 ≥ 2 ≥ 2Metastatic locationMetastatic location Hepatica1.03 (0.63–1.69)0.9103 Hepatica1.05 (0.60–1.88)0.8622 Non-hepatic Non-hepaticRAS status in tissueRAS status in tissue MUTa1.64 (1.15–2.35)0.0065 MUTa0.74 (0.39–1.29)0.2781 WT WTCI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MUT RAS mutated, WT wild typeaReference category for analysisStatistical significant p values are indicated in bold Table 2Prognostic impact of clinical characteristics depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersHR (95% CI)p-ValueBiomarkersHR (95% CI)p-ValueSexSex Malea0.69 (0.46–1.02)0.0624 Malea0.52 (0.29–0.91)0.0232 Female FemaleAge (years)Age (years) ≤ 66a0.58 (0.40–0.83)0.0033 ≤ 66a0.65 (0.39–1.10)0.1061 > 66 > 66ECOGECOG 0a 0a0.3966 10.50 (0.36–0.73)0.0003 10.76 (0.42–1.41)0.13 (0–0.02)< 0.00010.21 (0.02–0.21)< 0.0001 2 2Number of metastatic locationsNumber of metastatic locations 1a0.71 (0.41–1.22)0.2152 1a0.50 (0.20–1.28)0.1485 ≥ 2 ≥ 2Metastatic locationMetastatic location Hepatica1.03 (0.63–1.69)0.9103 Hepatica1.05 (0.60–1.88)0.8622 Non-hepatic Non-hepaticRAS status in tissueRAS status in tissue MUTa1.64 (1.15–2.35)0.0065 MUTa0.74 (0.39–1.29)0.2781 WT WTCI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MUT RAS mutated, WT wild typeaReference category for analysisStatistical significant p values are indicated in bold Prognostic impact of clinical characteristics depending on the primary tumour location CI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MUT RAS mutated, WT wild type aReference category for analysis Statistical significant p values are indicated in bold Patients with left-sided tumours were more frequently treated with first-line CTx plus anti-EGFR (24.7%) compared with those with right-sided tumours (4.7%). In contrast, right-sided tumours were more frequently treated with CTx plus anti-vascular endothelial growth factor (54.7% vs 38.3%). The use of CTx alone was similar between both groups (Fig. 1).Fig. 1Distribution of first-line treatments according to the primary tumour location. Patients with left-sided tumours received more frequently chemotherapy (CTx) plus anti-epidermal growth factor receptor (EGFR) than those with right-sided tumours (24.7% vs 4.7%), while patients with right-sided tumours received more frequently CTx plus anti-vascular endothelial growth factor [VEGF] (54.7%) Distribution of first-line treatments according to the primary tumour location. Patients with left-sided tumours received more frequently chemotherapy (CTx) plus anti-epidermal growth factor receptor (EGFR) than those with right-sided tumours (24.7% vs 4.7%), while patients with right-sided tumours received more frequently CTx plus anti-vascular endothelial growth factor [VEGF] (54.7%) Prognostic Value of Liquid Biopsy Biomarkers Depends on Primary Tumour Sidedness Liquid biopsy-based biomarkers demonstrated significant prognostic relevance. Specifically, circulating RAS mutations, higher MAF, increased cfDNA concentration, shorter cfDNA fragments and elevated CEA levels were significantly associated with poorer OS (Table 1 of the Electronic Supplementary Material). Notably, although no significant differences were observed between left-sided and right-sided CRC for these biomarkers (RAS status, p = 0.1025; MAF, p = 0.3774; cfDNA concentration, p = 0.3970; cfDNA fragmentation, p = 0.2864; and CEA, p = 0.2811), their prognostic impact differed markedly according to primary tumour sidedness (Table 3). Table 3Prognostic impact of liquid biopsy markers depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersNo. of casesHR (95% CI)p-ValueBiomarkersNo. of casesHR (95% CI)p-ValuecfDNA concentration (OS), ng/mLcfDNA concentration (OS), ng/mL ≤ 37.6a550.46 (0.30–0.71)0.0004 ≤ 37.6a260.57 (0.28–1.00)0.0598 > 37.649 > 37.622cfDNA concentration (PFS), ng/mLcfDNA concentration (PFS), ng/mL ≤ 37.6a410.74 (0.48–1.09)0.1307 ≤ 37.6a220.80 (0.43–1.46)0.4656 > 37.652 > 37.620MAF (OS), %MAF (OS), % ≤ 1.209a200.66 (0.41–1.12)0.1341 ≤ 1.209a80.29 (0.17–0.70)0.0048 > 1.20961 > 1.20932MAF (PFS), %MAF (PFS), % ≤ 1.209a180.93 (0.55–1.55)0.7721 ≤ 1.209a60.97 (0.40–2.32)0.9418 > 1.20959 > 1.20928RAS status (OS)RAS status (OS) MUTa831.67 (1.23–2.46)0.0024 MUTa410.77 (0.44–1.30)0.3120 WT79 WT23RAS status (PFS)RAS status (PFS) MUTa791.35 (1.00–1.91)0.0602 MUTa351.09 (0.64–1.87)0.7432 WT66 WT21cfDNA fragment size (OS), bpcfDNA fragment size (OS), bp ≤ 137a81.57 (0.70–4.24)0.2488 ≤ 137a92.17 (0.95–6.80)0.0697 > 13735 > 13715cfDNA fragment size (PFS), bpcfDNA fragment size (PFS), bp ≤ 137a41.39 (0.45–4.81)0.5246 ≤ 137a81.19 (0.49–3.00)0.6840 > 13737 > 13713cfDNA fragmentation (OS), %cfDNA fragmentation (OS), % ≤ 31.055a180.74 (0.38–1.39)0.3510 ≤ 31.055a150.88 (0.35–2.18)0.7812 > 31.05525 > 31.0559cfDNA fragmentation (PFS), %cfDNA fragmentation (PFS), % ≤ 31.055a180.89 (0.48–1.64)0.7097 ≤ 31.055a131.46 (0.66–3.65)0.3559 > 31.05523 > 31.0558CEA (OS), U/mLCEA (OS), U/mL ≤ 46.8a610.58 (0.38–0.85)0.0060 ≤ 46.8a241.00 (0.54–1.86)0.9961 > 46.863 > 46.822CEA (PFS), U/mLCEA (PFS), U/mL ≤ 46.8a550.93 (0.65–1.34)0.7117 ≤ 46.8a220.77 (0.39–1.44)0.3977 > 46.8l62 >46.817CEA carcinoembryonic antigen, cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, HR hazard ratio, MUT RAS mutated, OS overall survival, PFS progression-free survival, WT wild typeaReference category for analysisStatistical significant p-values are indicated in bold Table 3Prognostic impact of liquid biopsy markers depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersNo. of casesHR (95% CI)p-ValueBiomarkersNo. of casesHR (95% CI)p-ValuecfDNA concentration (OS), ng/mLcfDNA concentration (OS), ng/mL ≤ 37.6a550.46 (0.30–0.71)0.0004 ≤ 37.6a260.57 (0.28–1.00)0.0598 > 37.649 > 37.622cfDNA concentration (PFS), ng/mLcfDNA concentration (PFS), ng/mL ≤ 37.6a410.74 (0.48–1.09)0.1307 ≤ 37.6a220.80 (0.43–1.46)0.4656 > 37.652 > 37.620MAF (OS), %MAF (OS), % ≤ 1.209a200.66 (0.41–1.12)0.1341 ≤ 1.209a80.29 (0.17–0.70)0.0048 > 1.20961 > 1.20932MAF (PFS), %MAF (PFS), % ≤ 1.209a180.93 (0.55–1.55)0.7721 ≤ 1.209a60.97 (0.40–2.32)0.9418 > 1.20959 > 1.20928RAS status (OS)RAS status (OS) MUTa831.67 (1.23–2.46)0.0024 MUTa410.77 (0.44–1.30)0.3120 WT79 WT23RAS status (PFS)RAS status (PFS) MUTa791.35 (1.00–1.91)0.0602 MUTa351.09 (0.64–1.87)0.7432 WT66 WT21cfDNA fragment size (OS), bpcfDNA fragment size (OS), bp ≤ 137a81.57 (0.70–4.24)0.2488 ≤ 137a92.17 (0.95–6.80)0.0697 > 13735 > 13715cfDNA fragment size (PFS), bpcfDNA fragment size (PFS), bp ≤ 137a41.39 (0.45–4.81)0.5246 ≤ 137a81.19 (0.49–3.00)0.6840 > 13737 > 13713cfDNA fragmentation (OS), %cfDNA fragmentation (OS), % ≤ 31.055a180.74 (0.38–1.39)0.3510 ≤ 31.055a150.88 (0.35–2.18)0.7812 > 31.05525 > 31.0559cfDNA fragmentation (PFS), %cfDNA fragmentation (PFS), % ≤ 31.055a180.89 (0.48–1.64)0.7097 ≤ 31.055a131.46 (0.66–3.65)0.3559 > 31.05523 > 31.0558CEA (OS), U/mLCEA (OS), U/mL ≤ 46.8a610.58 (0.38–0.85)0.0060 ≤ 46.8a241.00 (0.54–1.86)0.9961 > 46.863 > 46.822CEA (PFS), U/mLCEA (PFS), U/mL ≤ 46.8a550.93 (0.65–1.34)0.7117 ≤ 46.8a220.77 (0.39–1.44)0.3977 > 46.8l62 >46.817CEA carcinoembryonic antigen, cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, HR hazard ratio, MUT RAS mutated, OS overall survival, PFS progression-free survival, WT wild typeaReference category for analysisStatistical significant p-values are indicated in bold Prognostic impact of liquid biopsy markers depending on the primary tumour location CEA carcinoembryonic antigen, cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, HR hazard ratio, MUT RAS mutated, OS overall survival, PFS progression-free survival, WT wild type aReference category for analysis Statistical significant p-values are indicated in bold In patients with left-sided tumours, several liquid biopsy-based biomarkers were significantly associated with poor prognosis. Higher cfDNA concentration (cut-off: 37.6 ng/mL; 592 vs 870 days; p = 0.0004; Fig. 2A), presence of RAS mutations in plasma (622 vs 851 days; p = 0.024; Fig. 2B) and elevated levels of CEA (cut-off: 46.8 U/mL; 567 vs 746 days; p = 0.006; Fig. 2C) all correlated with shorter OS. In contrast, among patients with right-sided tumours, only higher MAF levels (cut-off: 1.209%; 487 vs 1238 days; p = 0.0048; Fig. 3A) showed a significant association with outcome. Although not statistically significant, an elevated cfDNA concentration in right-sided CRC also showed a trend toward poorer survival (504 vs 702.5 days; p = 0.0598; Fig. 3B). In addition, a multivariate analysis was performed to investigate the relevance of these biomarkers. For left-sided tumours, age, ECOG and cfDNA concentration were independent prognostic factors (Table 4). Furthermore, although RAS mutation detection via a liquid biopsy did not reach statistical significance in the multivariate model, it showed a strong trend towards significance (p = 0.092). In right-sided tumours, both the ECOG and the RAS MAF were independent prognostic factors (Table 4).Fig. 2Liquid biopsy prognostic factors in left-sided colorectal cancer. A Overall survival (OS) according to the cell-free DNA concentration (cut-off: 37.6 ng/mL); B OS according to RAS mutational status in plasma; and C OS according to carcinoembryonic antigen levels. HR hazard ratio, MUT RAS mutated, WT wild typeFig. 3Liquid biopsy prognostic factors in right-sided colorectal cancer. (A) Overall survival (OS) according to the mutant allelic fraction (cut-off: 1.209%) and (B) OS according to the cell-free DNA concentration (cut-off: 37.6 ng/mL). HR hazard ratioTable 4Multivariate analysis of the prognostic impact of clinical variables and liquid biopsy markers depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersHR (95% CI)p-ValueBiomarkersHR (95% CI)p-ValueAge, yearsECOG ≤ 66a1.71 (1.09–2.66)0.0180a0.03 (0–0.47)0.013 > 661ECOG20.03 (0–0.51)0.015 0a 0.14 (0.03–0.68) 0.015MAF (%) 1 ≤ 1.209a3.62 (1.35–9.72)0.011 20.24 (0.053–1.09)0.064 > 1.209cfDNA concentration, ng/mL ≤ 37.6a2.07 (1.33–3.23)0.001 > 37.6cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MAF mutant allelic fractionaReference category for analysisStatistical significant p-values are indicated in bold Liquid biopsy prognostic factors in left-sided colorectal cancer. A Overall survival (OS) according to the cell-free DNA concentration (cut-off: 37.6 ng/mL); B OS according to RAS mutational status in plasma; and C OS according to carcinoembryonic antigen levels. HR hazard ratio, MUT RAS mutated, WT wild type Liquid biopsy prognostic factors in right-sided colorectal cancer. (A) Overall survival (OS) according to the mutant allelic fraction (cut-off: 1.209%) and (B) OS according to the cell-free DNA concentration (cut-off: 37.6 ng/mL). HR hazard ratio Table 4Multivariate analysis of the prognostic impact of clinical variables and liquid biopsy markers depending on the primary tumour locationLeft-sided CRCRight-sided CRCBiomarkersHR (95% CI)p-ValueBiomarkersHR (95% CI)p-ValueAge, yearsECOG ≤ 66a1.71 (1.09–2.66)0.0180a0.03 (0–0.47)0.013 > 661ECOG20.03 (0–0.51)0.015 0a 0.14 (0.03–0.68) 0.015MAF (%) 1 ≤ 1.209a3.62 (1.35–9.72)0.011 20.24 (0.053–1.09)0.064 > 1.209cfDNA concentration, ng/mL ≤ 37.6a2.07 (1.33–3.23)0.001 > 37.6cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MAF mutant allelic fractionaReference category for analysisStatistical significant p-values are indicated in bold Multivariate analysis of the prognostic impact of clinical variables and liquid biopsy markers depending on the primary tumour location cfDNA cell-free DNA, CI confidence interval, CRC colorectal cancer, ECOG Eastern Cooperative Oncology Group, HR hazard ratio, MAF mutant allelic fraction aReference category for analysis Statistical significant p-values are indicated in bold Integrating Liquid Biopsy Biomarkers Reveals Side-Dependent Prognostic Patterns in Metastatic Colorectal Cancer As the prognostic impact of individual liquid biopsy biomarkers varied with tumour sidedness, we next investigated whether integrating these markers could provide a more accurate, side-specific prognostic model for patients with mCRC. Remarkably, the combination of CEA levels with liquid biopsy biomarkers (RAS mutational status in plasma and cfDNA concentration) improved the prognostic stratification of patients with mCRC with left-sided tumours (p = 0.0030; Fig. 4A) but not of those with right-sided tumours (p = 0.0447; Fig. 4B). In contrast, the combination of MAF and cfDNA concentration was not significantly associated with prognosis in patients with left-sided tumours and RAS mutations (p = 0.3507; Fig. 4C), whereas it showed significant prognostic value in those with right-sided tumours and RAS mutations (p = 0.0243; Fig. 4D).Fig. 4Combination of liquid biopsy biomarkers improves prognostic stratification in left-sided and right-sided colorectal cancer. (A) Overall survival (OS) combining carcinoembryonic antigen with cell-free DNA concentration and RAS mutation status in plasma from patients with left-sided colorectal cancer; (B) OS combining carcinoembryonic antigen with cell-free DNA concentration and RAS mutation status in plasma from patients with right-sided colorectal cancer; (C) OS combining mutant allelic fraction with cell-free DNA concentration in plasma from patients with left-sided colorectal cancer; and (D) OS combining mutant allelic fraction with cell-free DNA concentration in plasma from patients with right-sided colorectal cancer Combination of liquid biopsy biomarkers improves prognostic stratification in left-sided and right-sided colorectal cancer. (A) Overall survival (OS) combining carcinoembryonic antigen with cell-free DNA concentration and RAS mutation status in plasma from patients with left-sided colorectal cancer; (B) OS combining carcinoembryonic antigen with cell-free DNA concentration and RAS mutation status in plasma from patients with right-sided colorectal cancer; (C) OS combining mutant allelic fraction with cell-free DNA concentration in plasma from patients with left-sided colorectal cancer; and (D) OS combining mutant allelic fraction with cell-free DNA concentration in plasma from patients with right-sided colorectal cancer Discussion Right-sided CRC is associated with a later diagnosis and a more aggressive phenotype, including poorer tumour differentiation and lower response rates to CTx [21, 22]. Because RAS mutations are more common in right-sided tumours, CTx plus anti-vascular endothelial growth factor was the preferred therapy in these patients, whereas in patients with left-sided tumours CTx alone or CTx plus anti-EGFR were the chosen treatments. Given the distinct biological behaviour of right-sided and left-sided CRC, it is important to consider how the primary tumour location may influence liquid biopsy results. Right-sided and left-sided CRC differ in their metastatic pattern: right-side tumours more commonly metastasize to the peritoneum, while left-sided CRC tends to metastasize to the liver and lungs [7, 8]. Accordingly, a higher proportion of patients with liver metastasis in the left-sided tumour group was observed in our cohort. In addition, the incidence of right-sided CRC increases with age and is associated with more aggressive disease characteristics [21, 23]. In this regard, right-sided CRC was associated with elderly patients and worse ECOG performance status in our study. However, ECOG was a significant prognostic factor regardless of tumour location, while the presence hepatic metastases was not significantly associated with prognosis in either group. Few studies on liquid biopsies have considered the primary tumour location, with right-sided CRC reported to be associated with more detectable molecular alterations, while no differences in cfDNA concentration were found according to the primary tumour location [24, 25]. Consistently, our study found no significant differences in cfDNA concentration or other liquid biopsy biomarkers between right-sided and left-sided tumours. In our analysis of RAS mutation status in plasma, we observed a similar distribution of patients with and without RAS mutations in left-sided tumours. In contrast, among right-sided tumours, the number of patients with RAS mutations was approximately twice that of those without RAS mutations. This aligns with previous reports indicating a higher prevalence of RAS mutations in right-sided compared with left-sided CRC [9]. Despite this, we found that RAS mutation status, both in plasma and tissue, was a strong prognostic marker in left-sided but not in right-sided tumours. Remarkably, however, when focusing in RAS-mutated patients, plasma RAS MAF emerged as a significant prognostic factor in right-sided but not in left-sided tumours. Although the RAS MAF cut-off point was calculated using the KaplanScan feature, the threshold of 1% was used with confidence because it has been described as clinically relevant for classifying patients according to their response to anti-EGFR treatment [26, 27]. Carcinoembryonic antigen levels demonstrated prognostic value in left-sided but not in right-sided CRC. Similarly, although, the cfDNA concentration showed some prognostic relevance in patients with right-sided tumours, it was a stronger prognostic marker in left-sided CRC. Moreover, the integrated analysis in our study of liquid biopsy biomarkers also revealed distinct side-dependent prognostic patterns in mCRC. The combination of CEA levels with plasma RAS mutational status and cfDNA concentration markedly improved prognostic stratification in patients with left-sided tumours but not in those with right-sided disease. One possible explanation for the stronger prognostic value of liquid biopsy biomarkers in left-sided CRC is the distinct biological and molecular landscape of these tumours. Left-sided CRCs are typically characterized by higher cellular turnover, greater tumour shedding and a more stable genomic profile compared with right-sided tumours, which often display microsatellite instability, mucinous histology and a more heterogeneous mutational burden [9, 10]. These features may result in a more consistent release of tumour-derived DNA and circulating biomarkers in left-sided disease, thereby strengthening the association between biomarker levels and clinical outcomes. Conversely, the biological diversity and immune microenvironment of right-sided tumours [28, 29] could dilute or obscure the relationship between circulating biomarkers and prognosis. Conclusions Our results highlight that the prognostic utility of liquid biopsy biomarkers is context dependent and underscore the need to interpret circulating biomarker data considering tumour sidedness. While a liquid biopsy is increasingly integrated into clinical practice, no studies to date have addressed how biological differences between right-sided and left-sided CRC may affect the performance and prognostic utility of these markers. By systematically exploring this aspect, our work contributes novel insight into the contextual interpretation of circulating biomarkers and supports a more tailored approach to patient monitoring and treatment planning. Prospective studies will be essential to validate these findings and further refine the clinical utility of liquid biopsy in mCRC. This study has some limitations to consider. As a retrospective analysis, it may be subject to selection bias, as a liquid biopsy for plasma RAS status was performed in the context of routine clinical practice to guide anti-EGFR treatment decisions. This likely contributed to the higher proportion of left-sided tumours in our cohort, as these patients are more often candidates for anti-EGFR therapies. In addition, our study only includes patients without any prior treatment, providing a biologically controlled baseline but may not fully reflect the heterogeneity of the real-world mCRC setting, where a liquid biopsy is frequently performed after multiple lines of treatment to identify acquired resistance alterations. In particular, previous studies have shown that treatment exposure can significantly reshape circulating tumour DNA profiles [16]. Therefore, while our findings provide a robust baseline framework, their direct extrapolation to heavily pretreated patients should be made with caution. Nonetheless, our findings provide relevant insights, and further prospective studies are warranted to confirm these results and explore their broader applicability. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 223 KB) Supplementary file1 (PDF 223 KB)