Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer Authors: Lijie Zhang, Jingwen Wang, Juefeng Wan, Yan Wang, Hui Zhang, Yaqi Wang, Fan Xia, Lijun Shen, Zhen Zhang Published: 2026 ABSTRACT BackgroundNeoadjuvant therapy (NAT) has shifted rectal cancer management toward organ preservation. However, achieving a complete response (CR) for “watch-and-wait” strategies is hindered by high response heterogeneity. Although immunotherapy-combined NAT has expanded the candidate pools, the predictive significance of molecular alterations remains unclear.ObjectivesThis study aimed to evaluate clinical and genomic profiles of rectal cancer patients undergoing NAT to identify response predictors and to develop a nomogram for estimating CR probability.DesignRetrospective, single-center cohort study.MethodsThis study included 437 patients with rectal adenocarcinoma at Fudan University Shanghai Cancer Center between December 2019 and March 2023. Patients underwent paired tumor and germline genomic sequencing (887-gene panel) before NAT. Logistic and Cox regression analyses were performed to identify clinical and genetic risk factors associated with tumor response and long-term survival.ResultsOf the 437 patients, 96.6% had microsatellite-stable (MSS) tumors. In the MSS locally advanced rectal cancer cohort (N = 307), the CR rate was 35.5%. Multivariate analysis identified immunotherapy-combined NAT (iTNT) (OR 4.41, 95% CI: 2.42–8.27), SYNE1 mutation (OR 2.12, 95% CI: 1.06–4.26), negative mesorectal fascia (MRF) status (OR 0.34, 95% CI: 0.17–0.66), and lower tumor location (OR 0.48, 95% CI: 0.27–0.84) as independent predictors of CR. KRAS mutation was the sole independent predictor of reduced disease-free survival (DFS; HR 1.93, 95% CI: (1.11–3.36), p = 0.020). KRAS G12D subtype was associated with the worst 2-year distant metastasis-free survival (71.3%) and exhibited a distinct predilection for lung metastasis. The clinical-genomic nomogram yielded strong discrimination (AUC = 0.705) and calibration, with favorable DCA net benefit.ConclusionClinical and genomic features jointly determine outcomes in MSS rectal cancer. SYNE1 mutation serves as a novel biomarker for CR, while KRAS mutations, especially the G12D subtype, identify patients at high risk for systemic relapse. The clinical-genomic nomogram facilitates individualized selection for organ-preservation strategies. Background Colorectal cancer (CRC) remains a leading cause of global cancer-related mortality. 1 Rectal cancer, comprising approximately one-third of CRC cases, represents a distinct and clinically demanding entity given its anatomical complexity and intrinsic risk of local recurrence. 2 Neoadjuvant therapy (NAT), particularly chemoradiotherapy (CRT)-based strategies, has transformed the management of rectal cancer across a broad spectrum of disease stages.2,3 Beyond reducing local failure rates, NAT facilitates sphincter-sparing surgery and improves long-term functional outcomes.4,5 For patients achieving a complete response (CR) following NAT, non-operative “watch-and-wait” (W&W) strategy, is increasingly favored to maximize organ preservation and quality of life.6,7 More recently, integrating immunotherapy into radiotherapy-based neoadjuvant setting has shown encouraging potential to improve CR rates even in microsatellite-stable (MSS) or mismatch repair-proficient (pMMR) cohorts,8–11 further expanding the pool of candidates for W&W strategies. Despite these advances, pronounced heterogeneity in treatment response continues to confound clinical decision-making. The inability to prospectively identify exceptional responders or those destined for resistance remains a critical bottleneck for the safe and broad adoption of W&W strategy. Consequently, identifying robust predictive biomarkers is imperative to refine patient selection and tailor therapeutic intensity. While next-generation sequencing (NGS) has extensively mapped the genomic drivers of CRC (e.g., APC, TP53, and KRAS) and highlighted key therapeutic vulnerabilities,12,13 the clinical utility of these alterations in the neoadjuvant setting remains nuanced. Although prior studies link mutations like KRAS and TP53 to conventional CRT resistance,14–17 their predictive and prognostic significance in the era of immunotherapy-combined NAT are largely undefined. Moreover, integrating molecular features with clinicopathologic variables into clinical applicable tools, such as predictive nomograms, is essential to guide individualized organ preservation strategies. To address this gap, the present study evaluated the clinical and genomic determinants associated with NAT outcomes in rectal cancer. We described the mutational profiles of these tumors and identified independent clinical and molecular predictors of treatment response. Finally, we developed an integrated clinical-genomic nomogram to estimate the probability of achieving a CR. Methods Study population This retrospective study was conducted at the Department of Radiation Oncology, Fudan University Shanghai Cancer Center (FUSCC). Patients diagnosed with rectal cancer who provided paired peripheral blood and primary tumor samples for genomic sequencing between December 2019 and March 2023 were enrolled. Exclusion criteria were as follows: 1) histological confirmation of non-adenocarcinoma tumors, 2) non-rectal primary tumors, 3) no treatment or only surgery, and 4) sample collection after treatment initiation. Diagnosis was confirmed pathologically for all patients. Local tumor staging was performed using rectal magnetic resonance imaging (MRI), while metastatic disease was screened using computed tomography (CT) of the chest, abdomen, and pelvis. Tumor staging adhered to the American Joint Committee on Cancer (AJCC) TNM classification (8th edition). 18 Clinical and pathological data were extracted from medical records and systematically reviewed. The reporting of this study confirms to The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement (Supplemental File 1). 19 Neoadjuvant therapy regimens NAT was administered based on disease stage and metastatic status. For patients with clinical stage I (cT1-2N0) disease, NAT followed by local excision was employed for organ preservation. For patients with clinical stage II (cT3-4N0) or III (any T, N1-2) disease, NAT aimed to achieve tumor regression to facilitate radical surgery. For patients with resectable metastatic rectal tumors, NAT aimed to reduce tumor size for potential simultaneous resection of primary and metastatic lesions. The standard NAT regimen was fluoropyrimidine-based CRT, consisting of long-course radiotherapy (50 Gy in 25 fractions, 2 Gy per fraction) with concurrent capecitabine or weekly irinotecan (CapeIri). 20 A very limited number of patients received oxaliplatin added to capecitabine (CapeOx) concurrently with radiotherapy. Additionally, a subset of patients received an immunotherapy-based total neoadjuvant therapy (iTNT) regimen, 9 comprising short-course radiotherapy (SCRT; 25 Gy in 5 fractions), followed by six cycles of the PD-1 blockade Toripalimab (Junshi BioSciences, Shanghai, China), and XELOX chemotherapy. Tumor response assessment and post-NAT management For patients with non-metastatic rectal cancer (clinical stages I–III), tumor response was assessed approximately 6–8 weeks following the completion of NAT. Assessment followed a standardized institutional protocol comprising digital rectal examination (DRE), pelvic MRI, and proctoscopy. Management decisions were made through multidisciplinary team discussions. Patients achieving a clinical CR (cCR) who prioritized organ preservation were managed via a W&W strategy. In contrast, patients with residual disease or those who preferred surgical intervention underwent radical resection, specifically total mesorectal excision (TME). For patients with stage IV disease, response assessment included both the primary tumor and metastatic sites. Management was highly individualized based on the feasibility of achieving curative intent, with treatment ranging from curative-intent surgical procedures to systemic or palliative therapy. Tissue collection and sample sequencing Pre-treatment biopsy samples and peripheral blood were collected from all patients. DNA was extracted using the TIANamp Genomic DNA kit (Tiangen, China), and the purity and quantity were assessed using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA). Samples with an A260/A280 ratio between 1.6 and 1.9 were deemed suitable for sequencing. Genomic sequencing was performed using the FUSCC-CRC panel, which targets 887 cancer-related genes, including clinically actionable mutations and frequent CRC alterations (Table S1). The sequencing was performed according to the manufacturer’s instructions. Briefly, 10 ng of DNA from tumor and matched blood samples underwent fragmentation, end-repair, poly-A–tailing, and adaptor ligation to construct libraries. Libraries were pooled and sequenced on a NovaSeq 6000 platform. Tumor DNA and blood DNA were sequenced at a depth of 500× and 200×, respectively. Genomic analyses Quality-controlled reads were aligned to the GRCh37 reference genome using BWA and converted to BAM files. Somatic single nucleotide variants (SNVs) and insertions or deletions (INDELs) were called for each sample as previously reported. 21 HaplotypeCaller and VarScan2 were used to identify potential germline alterations. Somatic mutations were analyzed using the Maftools package (version 2.18.0), which facilitated the visualization and the integration of clinical and genomic features. Clinical outcomes Complete response was defined as pathological CR (pCR) in surgical patients, characterized by the absence of viable cancer cells upon histological evaluation of the total mesorectal excision specimen, or sustained clinical CR (cCR) in W&W patients, defined by the absence of detectable tumor via clinical assessment. CR analyses were restricted to stage II–III MSS patients without second primary tumor. Stage I patients were excluded to avoid selection bias (as NAT was intent on organ preservation), and stage IV patients were excluded due to distinct treatment goals. Disease-free survival (DFS) was defined as the interval from surgery (in postoperative patients) or last treatment (in W&W patients) to local recurrence, distant metastasis, or death. Local regrowth in W&W patients was not considered a DFS event if salvage surgery was successfully performed. Only patients stage I–III achieving a disease-free state (R0 resection or sustained cCR) were included in the DFS analysis. Overall survival (OS), defined as the interval from diagnosis to death from any cause, included all patients with MSS tumors (stages I–IV). Distant-metastasis free survival (DMFS) was defined as the interval from diagnosis to distant metastasis, excluding stage IV patients. Patients who were alive or event-free at the last follow-up (September 5, 2024) were censored. Development and validation of the predictive nomogram A predictive nomogram for estimating the probability of CR following NAT was constructed based on variables that remained significant (p < 0.05) in the final multivariate logistic regression model. Model performance was evaluated through assessments of discrimination, calibration, and clinical utility. Discrimination was quantified using the area under the receiver operating characteristic curve (AUC). Calibration was examined by comparing nomogram-predicted probabilities with the observed CR rates via calibration plots. To ensure the robustness of the results and to account for potential overfitting, internal validation was performed using a bootstrap resampling method with 1,000 repetitions, yielding bias-corrected estimates of model performance. Decision curve analysis (DCA) was performed to evaluate the clinical net benefit of the nomogram across a range of threshold probabilities. Statistical analysis All statistical analyses were performed using R software, version 4.2.2 (R Project for Statistical Computing, Vienna, Austria). Baseline clinicopathologic characteristics were summarized as medians (range) for continuous variables and compared using the Wilcoxon rank-sum test. Categorical variables are presented as frequencies with percentages and analyzed using Fisher’s exact test. Differences in observed CR rates across the nomogram-stratified groups were evaluated using the Cochran-Armitage trend test. For genomic feature selection, genes with a mutation frequency > 5% were ranked using a Random Forest model based on permutation importance, with selection stability validated via 1,000 bootstrap resamples. A 10-fold cross-validated least absolute shrinkage and selection operator (LASSO) regression model was evaluated for methodological comparison. Following feature selection, univariate logistic regression was conducted to identify potential predictors of CR after NAT, excluding patients with metastatic disease. Variables demonstrating significance in the univariate analysis were subsequently included in the multivariate logistic regression model to determine independent predictors, with adjusted odds ratios (ORs) and 95% confidence intervals (CIs) calculated. DFS and OS were estimated using the Kaplan–Meier method, with group comparisons assessed by log-rank test. Non-operated patients with incomplete tumor regression (non-CR) were excluded for DFS analysis. Significant clinical and genomic factors associated with DFS and OS were further analyzed using Cox proportional regression models, with hazard ratios (HRs) and CIs calculated. All p-values were 2-sided, with p < 0.05 considered statistically significant. The multivariable models used a complete case analysis, excluding all cases with missing or unknown values. Results Patient characteristics and genomic overview Between December 2019 and March 2023, 637 patients diagnosed with rectal cancer at FUSCC underwent baseline genomic sequencing of paired tumor and germline samples. After applying the exclusion criteria, 437 patients were ultimately included in this study (Figure 1). Among the 437 patients, 28 (6.4%) had stage I disease, 46 (10.5%) had stage II, 292 (66.8%) had stage III, and 71 (16.2%) had stage IV disease. Detailed clinical and pathological characteristics are presented in Table 1.Figure 1.Study design and outcome evaluation framework Of 637 sequenced patients, 437 met the inclusion criteria. The subsequent analyses included: 307 patients with MSS LARC for treatment response (Stage II/III); 335 patients for DFS/DMFS (Stage I–III); and 401 patients for OS (Stage I–IV).Abbreviations: NAT, neoadjuvant therapy; DFS, disease-free survival; OS, overall survival; DMFS, distant-metastasis free survival; CR, complete response; TMB, tumor mutational burden; SCRT, short-course radiotherapy.Table 1.Baseline characteristics of patients analyzed for genomic overview.CharacteristicsNo. (%) Total 437 (100)Sex Females139 (31.8) Males298 (68.2)Age ≥ 60 years171 (39.1) < 60 years266 (60.9)Tumor location ≤ 5 cm301 (68.9) 5-10cm126 (28.8) > 10cm10 (2.3)Clinical TNM stage I28 (6.4) II46 (10.5) III292 (66.8) IV71 (16.2)Molecular subtype MSS422 (96.6) MSI-H/POLE15 (3.4)Abbreviations: MSS, microsatellite-stable; MSI-H, microsatellite instability high; POLE, DNA polymerase epsilon. Study design and outcome evaluation framework Of 637 sequenced patients, 437 met the inclusion criteria. The subsequent analyses included: 307 patients with MSS LARC for treatment response (Stage II/III); 335 patients for DFS/DMFS (Stage I–III); and 401 patients for OS (Stage I–IV). Abbreviations: NAT, neoadjuvant therapy; DFS, disease-free survival; OS, overall survival; DMFS, distant-metastasis free survival; CR, complete response; TMB, tumor mutational burden; SCRT, short-course radiotherapy. Table 1.Baseline characteristics of patients analyzed for genomic overview.CharacteristicsNo. (%) Total 437 (100)Sex Females139 (31.8) Males298 (68.2)Age ≥ 60 years171 (39.1) < 60 years266 (60.9)Tumor location ≤ 5 cm301 (68.9) 5-10cm126 (28.8) > 10cm10 (2.3)Clinical TNM stage I28 (6.4) II46 (10.5) III292 (66.8) IV71 (16.2)Molecular subtype MSS422 (96.6) MSI-H/POLE15 (3.4)Abbreviations: MSS, microsatellite-stable; MSI-H, microsatellite instability high; POLE, DNA polymerase epsilon. Baseline characteristics of patients analyzed for genomic overview. Abbreviations: MSS, microsatellite-stable; MSI-H, microsatellite instability high; POLE, DNA polymerase epsilon. Of the 437 patients, 422 (96.6%) had MSS tumors, while 15 (3.4%) had microsatellite instability (MSI) or POLE (DNA polymerase epsilon) hypermutated tumors. In the MSS cohort, the mutational profile was characterized by canonical CRC drivers, most frequently involving TP53 (N = 327, 77.5%), APC (N = 285, 67.5%), KRAS (N = 198, 46.9%), FBXW7 (N = 106, 25.1%), and PIK3CA (N = 56, 13.3%). Conversely, MSI-H or POLE tumors displayed a distinct mutational spectrum with high-frequency alterations in APC (N = 14, 93.3%), as well as TP53, ERBB2, FBXW7, and ARID1A (each N = 9, 60.0%) (Figure 2). Given the biological distinctiveness of MSI-H/POLE cases and the predominance of MSS tumors, subsequent analyses focused on the clinical and genomic determinants of treatment outcomes specifically within the MSS cohort.Figure 2.The genomic landscape of rectal cancer. Oncoprint showing the most frequently altered genes in rectal cancer, stratified by MSS and POLE/MSI patients.Abbreviations: MSS, microsatellite stable; MSI-H, high microsatellite instability; POLE, DNA polymerase epsilon. The genomic landscape of rectal cancer. Oncoprint showing the most frequently altered genes in rectal cancer, stratified by MSS and POLE/MSI patients. Abbreviations: MSS, microsatellite stable; MSI-H, high microsatellite instability; POLE, DNA polymerase epsilon. Determinants of response to neoadjuvant therapy A total of 338 patients with Stage II and III rectal cancer were initially screened. After excluding 13 patients with MSI-H/POLE mutations and 18 patients with second primary tumors, 307 patients with MSS locally advanced rectal cancer (LARC) were included in the subsequent analysis to identify factors associated with treatment response. In this cohort, 109 patients (35.5%) achieved a CR. Univariate analysis demonstrated that baseline clinicopathological features significantly influenced the likelihood of CR (Table 2). Patients with tumors located in the lower rectum (≤ 5 cm from the anal verge) achieved a significantly higher CR rate than those with middle or upper tumors (39.8% vs. 26.7%, p = 0.024).Table 2.Clinical factors associated with complete response to neoadjuvant therapy in patients with locally advanced rectal cancer.CharacteristicsTotal NCR n (%)UnivariateMultivariateOR (95%CI)p-valueOR (95%CI)p-valueAge < 60 years18661 (32.8%)Reference​Reference​ ≥ 60 years12148 (39.7%)1.35 (0.84–2.16)0.2191.54 (0.91–2.61)0.108Sex Female8434 (40.5%)Reference​Reference​ Male22375 (33.6%)0.74 (0.45–1.25)0.2620.81 (0.46–1.46)0.487Tumor location Lower20682 (39.8%)Reference​Reference​ Middle/Upper10127 (26.7%)0.56 (0.33–0.93)0.0240.48 (0.27–0.84)0.010Stage II4017 (42.5%)Reference​Reference​ III26792 (34.5%)0.71 (0.36–1.40)0.3141.14 (0.54–2.46)0.732Neoadjuvant regimen CRT24470 (28.7%)Reference​Reference​ IO-combined6339 (61.9%)3.99 (2.26–7.17)<0.0014.41 (2.42–8.27)<0.001MRF Negative21890 (41.3%)Reference​Reference​ Positive8919 (21.3%)0.39 (0.22–0.68)<0.0010.34 (0.17–0.66)0.001EMVI Negative22988 (38.4%)Reference​Reference​ Positive7821 (26.9%)0.6 (0.34–1.04)0.0671.19 (0.6–2.39)0.614Abbreviations: CR, complete response; OR, odds ratio; MRF, mesorectal fascia; EMVI, extramural vascular invasion. Table 2.Clinical factors associated with complete response to neoadjuvant therapy in patients with locally advanced rectal cancer.CharacteristicsTotal NCR n (%)UnivariateMultivariateOR (95%CI)p-valueOR (95%CI)p-valueAge < 60 years18661 (32.8%)Reference​Reference​ ≥ 60 years12148 (39.7%)1.35 (0.84–2.16)0.2191.54 (0.91–2.61)0.108Sex Female8434 (40.5%)Reference​Reference​ Male22375 (33.6%)0.74 (0.45–1.25)0.2620.81 (0.46–1.46)0.487Tumor location Lower20682 (39.8%)Reference​Reference​ Middle/Upper10127 (26.7%)0.56 (0.33–0.93)0.0240.48 (0.27–0.84)0.010Stage II4017 (42.5%)Reference​Reference​ III26792 (34.5%)0.71 (0.36–1.40)0.3141.14 (0.54–2.46)0.732Neoadjuvant regimen CRT24470 (28.7%)Reference​Reference​ IO-combined6339 (61.9%)3.99 (2.26–7.17)<0.0014.41 (2.42–8.27)<0.001MRF Negative21890 (41.3%)Reference​Reference​ Positive8919 (21.3%)0.39 (0.22–0.68)<0.0010.34 (0.17–0.66)0.001EMVI Negative22988 (38.4%)Reference​Reference​ Positive7821 (26.9%)0.6 (0.34–1.04)0.0671.19 (0.6–2.39)0.614Abbreviations: CR, complete response; OR, odds ratio; MRF, mesorectal fascia; EMVI, extramural vascular invasion. Clinical factors associated with complete response to neoadjuvant therapy in patients with locally advanced rectal cancer. Abbreviations: CR, complete response; OR, odds ratio; MRF, mesorectal fascia; EMVI, extramural vascular invasion. High-risk radiological features were also strong predictors of poor response: MRF involvement was significantly less frequent in the CR group (21.3% vs. 41.3%; p < 0.001), and EMVI positivity showed a similar trend (26.9% vs. 38.4%; p = 0.067). Notably, the choice of neoadjuvant regimen emerged as the most significant clinical factor, where patients receiving immunotherapy-combined regimens (iTNT) demonstrated a markedly superior CR rate compared to those receiving standard CRT (61.9% vs. 28.7%, p < 0.001). Genomic features associated with neoadjuvant therapy response A Random Forest model was employed to identify the most discriminative genomic features (with mutation frequency >5%) between the two groups. Notably, SYNE1 emerged as the top-ranking gene contributing to the model’s predictive performance (Figure S1A). Notably, the canonical driver KRAS and the novel marker SYNE1 emerged as the leading genomic features contributing to the model’s predictive performance, with SYNE1 demonstrating robust selection stability across 1,000 internal bootstrap resamples (Figure S1B). Alternatively, 10-fold cross-validated LASSO regression compressed all coefficients to zero at the optimal threshold (λmin; Figure S1C), confirming that traditional linear models are insufficient for the complex clinical–genomic interactions. A comparison of somatic alterations between CR and nCR groups revealed that while the overall mutational profile was dominated by TP53, APC, and KRAS, certain genes exhibited divergent distribution patterns (Figure 3(A)). The CR rate was significantly higher in SYNE1 mutated group compared to the SYNE1 wildtype group (51.0% vs. 32.4%, p = 0.013) (Table S2). Conversely, KRAS mutations were more prevalent in nCR patients (70.5% vs. 59.5% in CR, p = 0.046).Figure 3.Genomic features and mutational profiles associated with neoadjuvant therapy response. (A) Comparison of somatic mutation frequencies in complete response (CR) and non-CR (nCR) groups. (B) Pathway-level mutation frequencies. Comparison of oncogenic signaling pathway alterations between CR and nCR groups. *, p < 0.05. (C) Gene co-occurrence and mutual exclusivity heatmap. Heatmap illustrating mutational interactions within the CR (upper-left) and nCR (lower-right) cohorts. Red indicates co-occurrence and blue indicates mutual exclusivity. *, p < 0.05. (D) Distribution of somatic mutations across functional domains of the SYNE1 protein. (E) Tumor mutational burden (TMB) levels in SYNE1-mutant (MUT) versus wild-type (WT) tumors. ***, p < 0.001. (F) Prevalence of SYNE1 mutations according to treatment response in the immunotherapy-integrated cohort. Stacked bars represent the proportion of SYNE1-mutant (MUT) and wild-type (WT) tumors in CR versus nCR patients. (G) Multivariate logistic regression analysis of independent predictors for CR. Forest plot illustrates the odds ratios (ORs) and 95% confidence intervals (CIs) for genomic features. The model was constructed by integrating both clinical and genomic variables (as detailed in Table 2). Genomic features and mutational profiles associated with neoadjuvant therapy response. (A) Comparison of somatic mutation frequencies in complete response (CR) and non-CR (nCR) groups. (B) Pathway-level mutation frequencies. Comparison of oncogenic signaling pathway alterations between CR and nCR groups. *, p < 0.05. (C) Gene co-occurrence and mutual exclusivity heatmap. Heatmap illustrating mutational interactions within the CR (upper-left) and nCR (lower-right) cohorts. Red indicates co-occurrence and blue indicates mutual exclusivity. *, p < 0.05. (D) Distribution of somatic mutations across functional domains of the SYNE1 protein. (E) Tumor mutational burden (TMB) levels in SYNE1-mutant (MUT) versus wild-type (WT) tumors. ***, p < 0.001. (F) Prevalence of SYNE1 mutations according to treatment response in the immunotherapy-integrated cohort. Stacked bars represent the proportion of SYNE1-mutant (MUT) and wild-type (WT) tumors in CR versus nCR patients. (G) Multivariate logistic regression analysis of independent predictors for CR. Forest plot illustrates the odds ratios (ORs) and 95% confidence intervals (CIs) for genomic features. The model was constructed by integrating both clinical and genomic variables (as detailed in Table 2). Pathway-level analysis indicated that the cell-cycle pathway showed higher mutation frequencies in CR patients (p = 0.016) (Figure 3(B)). Analysis of gene co-occurrence and mutual exclusivity patterns further delineated group-specific mutational interactions. Within the CR cohort, significant co-occurrences were observed between APC and FBXW7, as well as between ARID1A and KMT2D (p < 0.05). In the nCR cohort, APC was found to significantly co-occur with both FBXW7 and TP53, while SOX9 and BRCA2 also exhibited a significant co-occurrence pattern (p < 0.05) (Figure 3(C)). Detailed characterization of SYNE1 identified mutations across multiple domains (Figure 3(D)). Notably, SYNE1-mutant tumors exhibited a significantly higher tumor mutational burden (TMB) compared to wild-type tumors (median: 0.50 vs. 0.38 mut/Mb, p < 0.001), suggesting a potential mechanism for their enhanced sensitivity (Figure 3(E)). Notably, SYNE1 mutations were more prevalent among patients achieving CR receiving an immunotherapy-combined regimen compared with non-CR patients (28.2% vs. 4.2%; p = 0.022) (Figure 3(F)). Among the SYNE1-mutant subpopulation (N = 51), patients who received the immunotherapy-combined regimen achieved a remarkably higher CR rate than those treated with conventional neoadjuvant therapy without immunotherapy (91.7% vs. 38.5%; p = 0.002; Table S3). Multivariate logistic regression analysis, integrating both clinical and genomic variables, confirmed that SYNE1 mutation remained an independent predictor of CR (OR = 2.12, 95% CI: 1.06–4.26; p = 0.033), alongside immunotherapy-combined regimen (OR = 4.41, 95% CI: 2.42–8.27; p < 0.001), negative MRF status (Reference: Positive; OR = 0.34, 95% CI: 0.17–0.66; p = 0.001), and lower tumor location (Reference: Middle/Upper; OR = 0.48, 95% CI: 0.27–0.84; p = 0.010) (Table 2 and Figure 3(G)). Notably, KRAS mutations exhibited a potential trend toward an association with non-CR (OR = 0.60, 95% CI: 0.35–1.01; p = 0.054), suggesting they may contribute to reduced treatment sensitivity, although this did not reach formal statistical significance (Figure 3(G)). Clinical and genomic variables associated with long-term survival in rectal cancer After a median follow-up of 26.8 (95% CI: 25.6–30.1) months, the 2-year OS rate for the entire MSS cohort (Stage I–IV, N = 401) was 91.8% (95% CI: 88.7%–94.9%). For patients with stage I–III disease (N = 335), the 2-year DFS rate was 75.9% (95% CI: 70.1%–82.2%). Univariate Cox regression analysis identified KRAS mutation as a potential factor associated with inferior DFS (HR = 1.69, 95% CI: 1.00–2.86; p = 0.052). In the multivariate model, KRAS mutation emerged as the sole independent predictor of reduced DFS (HR = 1.93, 95% CI: 1.11–3.36; p = 0.020) after adjusting for age, sex, tumor location, TNM stage, MRF/EMVI status, and neoadjuvant regimen (Table 3). In contrast, OS analysis for the broader cohort (stages I–IV) revealed that clinical stage remained the dominant prognostic factor, with stage IV disease independently associated with significantly worse survival (HR = 5.88, 95% CI: 2.84–12.18; p < 0.001). No specific genomic alterations, including KRAS status, were significantly associated with OS in this cohort (p = 0.908) (Table S4).Table 3.Cox proportional hazards regression analysis of factors associated with disease-free survival in patients with stage I-III disease.CharacteristicsTotal NEvents (%)UnivariateMultivariateHR (95% CI)p-valueHR (95% CI)p-valueAge < 60 years17739 (22.0%)Reference​Reference​ ≥ 60 years11917 (14.3%)0.67 (0.38–1.18)0.1640.68 (0.38–1.23)0.205Sex Female8615 (17.4%)Reference​Reference​ Male21041 (19.5%)1.28 (0.71–2.31)0.4171.75 (0.93–3.31)0.085Tumor location Lower19940 (20.1%)Reference​Reference​ Middle/Upper9716 (16.5%)0.94 (0.52–1.67)0.8250.96 (0.52–1.76)0.895Stage I273 (11.1%)Reference​Reference​ II386 (15.8%)1.35 (0.34–5.4)0.6711.16 (0.27–4.93)0.841 III23147 (20.3%)1.83 (0.57–5.89)0.3121.74 (0.5–5.99)0.383NAT regimen CRT23544 (18.7%)Reference​Reference​ IO-combined6112 (19.7%)1.6 (0.83–3.07)0.1611.59 (0.81–3.1)0.174MRF Negative22639 (17.3%)Reference​Reference​ Positive7017 (24.3%)1.5 (0.85–2.65)0.1661.8 (0.92–3.55)0.089EMVI Negative23346 (19.7%)Reference​Reference​ Positive6310 (15.9%)0.78 (0.39–1.55)0.4850.48 (0.22–1.06)0.070 TMB29656 (18.9%)1.21 (0.81–1.81)0.3411.13 (0.76–1.68)0.544 APC WT9422 (23.4%)Reference​Reference​ Mutated20234 (16.8%)0.75 (0.44–1.29)0.3020.62 (0.35–1.1)0.102 TP53 WT699 (13.0%)Reference​Reference​ Mutated22747 (20.7%)1.37 (0.67–2.79)0.3901.27 (0.6–2.66)0.532 KRAS WT16525 (15.2%)Reference​Reference​ Mutated13131 (23.7%)1.69 (1–2.86)0.0521.93 (1.11–3.36)0.020Abbreviations: CR, complete response; HR, hazards ratio; NAT, neoadjuvant therapy; MRF, mesorectal fascia; EMVI, extramural vascular invasion; TMB, tumor mutational burden. Table 3.Cox proportional hazards regression analysis of factors associated with disease-free survival in patients with stage I-III disease.CharacteristicsTotal NEvents (%)UnivariateMultivariateHR (95% CI)p-valueHR (95% CI)p-valueAge < 60 years17739 (22.0%)Reference​Reference​ ≥ 60 years11917 (14.3%)0.67 (0.38–1.18)0.1640.68 (0.38–1.23)0.205Sex Female8615 (17.4%)Reference​Reference​ Male21041 (19.5%)1.28 (0.71–2.31)0.4171.75 (0.93–3.31)0.085Tumor location Lower19940 (20.1%)Reference​Reference​ Middle/Upper9716 (16.5%)0.94 (0.52–1.67)0.8250.96 (0.52–1.76)0.895Stage I273 (11.1%)Reference​Reference​ II386 (15.8%)1.35 (0.34–5.4)0.6711.16 (0.27–4.93)0.841 III23147 (20.3%)1.83 (0.57–5.89)0.3121.74 (0.5–5.99)0.383NAT regimen CRT23544 (18.7%)Reference​Reference​ IO-combined6112 (19.7%)1.6 (0.83–3.07)0.1611.59 (0.81–3.1)0.174MRF Negative22639 (17.3%)Reference​Reference​ Positive7017 (24.3%)1.5 (0.85–2.65)0.1661.8 (0.92–3.55)0.089EMVI Negative23346 (19.7%)Reference​Reference​ Positive6310 (15.9%)0.78 (0.39–1.55)0.4850.48 (0.22–1.06)0.070 TMB29656 (18.9%)1.21 (0.81–1.81)0.3411.13 (0.76–1.68)0.544 APC WT9422 (23.4%)Reference​Reference​ Mutated20234 (16.8%)0.75 (0.44–1.29)0.3020.62 (0.35–1.1)0.102 TP53 WT699 (13.0%)Reference​Reference​ Mutated22747 (20.7%)1.37 (0.67–2.79)0.3901.27 (0.6–2.66)0.532 KRAS WT16525 (15.2%)Reference​Reference​ Mutated13131 (23.7%)1.69 (1–2.86)0.0521.93 (1.11–3.36)0.020Abbreviations: CR, complete response; HR, hazards ratio; NAT, neoadjuvant therapy; MRF, mesorectal fascia; EMVI, extramural vascular invasion; TMB, tumor mutational burden. Cox proportional hazards regression analysis of factors associated with disease-free survival in patients with stage I-III disease. Abbreviations: CR, complete response; HR, hazards ratio; NAT, neoadjuvant therapy; MRF, mesorectal fascia; EMVI, extramural vascular invasion; TMB, tumor mutational burden. KRAS variants and patterns of distant metastasis To further investigate the role of KRAS codon mutations in post-NAT metastasis of rectal cancer, we analyzed the association between metastasis incidence and KRAS subtype mutations in patients with stage I–III disease. During follow-up, 63 of 335 patients (18.8%) developed metachronous distant metastasis, with KRAS mutations being significantly overrepresented in the metastatic group compared to the non-metastatic group (58.7% vs. 43.0%, p = 0.026) (Figure 4(A)).Figure 4.Association between KRAS mutation and metastasis after neoadjuvant therapy (A) Proportions of KRAS mutation status in metastatic and non-metastatic groups. Stacked bar charts illustrate the percentage of KRAS-mutant versus KRAS-wild-type (WT) cases. (B) Distribution of KRAS codon alterations. The donut chart illustrates the relative frequencies of specific codon substitutions within the KRAS-mutant population (N = 154). (C) Distant metastasis-free survival (DMFS) stratified by KRAS mutation status. Kaplan-Meier curves compare outcomes for WT, G12D-mutant, and other KRAS-mutant cohorts. P-value was determined by the log-rank test. (D) Patterns of distant metastasis according to KRAS mutation subtypes. Sankey diagram illustrating the association between specific KRAS variants (e.g., G12D, G12V, G13D) and organ-specific metastatic sites. Association between KRAS mutation and metastasis after neoadjuvant therapy (A) Proportions of KRAS mutation status in metastatic and non-metastatic groups. Stacked bar charts illustrate the percentage of KRAS-mutant versus KRAS-wild-type (WT) cases. (B) Distribution of KRAS codon alterations. The donut chart illustrates the relative frequencies of specific codon substitutions within the KRAS-mutant population (N = 154). (C) Distant metastasis-free survival (DMFS) stratified by KRAS mutation status. Kaplan-Meier curves compare outcomes for WT, G12D-mutant, and other KRAS-mutant cohorts. P-value was determined by the log-rank test. (D) Patterns of distant metastasis according to KRAS mutation subtypes. Sankey diagram illustrating the association between specific KRAS variants (e.g., G12D, G12V, G13D) and organ-specific metastatic sites. A total of 154 patients were found to have KRAS mutations, with 156 codon alterations identified. Among them, two patients had co-occurring mutations: G12D/G12S, and A146T/T74P. The most frequent mutations occurred in the G12 codon (105/154, 68.2%), followed by the G13 codon (32/154, 20.8%). Specifically, G12D was the most common mutation subtype (49/154, 31.8%), followed by G13D (31/154, 20.1%) and G12V (29/154, 18.8%) (Figure 4(B)). Notably, patients with the KRAS G12D mutation had the worst 2-year DMFS rate at 71.3% (95% CI: 58.7%–86.5%), compared to 79.9% (95% CI: 71.7%–89.2%) in patients with other KRAS codon mutations and 86.7% (95% CI: 81.2%–92.5%) in those with wild-type KRAS (p = 0.012) (Figure 4(C)). To provide a broader codon-level perspective, an alternative analysis stratifying patients by KRAS G12 mutations, non-G12 mutations, and wild-type KRAS was also performed, which similarly revealed a significant difference in DMFS (p = 0.044) (Figure S2). To characterize the patterns of dissemination, a Sankey diagram was constructed to visualize the relationship between KRAS mutation subtypes and organ-specific metastatic sites (N = 37) (Figure 4(D)). While the liver (N = 10) and lung (N = 12) remained the primary metastatic sites overall, tumors harboring G12D mutations (N = 17) demonstrated a multi-organ predilection. Specifically, among patients with G12D mutations, lung involvement was more prevalent than liver metastasis (41.2% vs. 17.6%), suggesting a potential pulmonary organotropism for this specific mutation compared to the broader cohort. Development and validation of the clinical-genomic nomogram for CR prediction To facilitate individualized treatment decision-making, particularly in selecting candidates for organ-preservation strategies, we developed a predictive nomogram based on patients with stage I–III disease (N = 335). The model integrates four independent determinants of CR identified in the multivariate analysis: NAT regimen (immunotherapy-combined vs. CRT), SYNE1 mutation status, MRF status, and tumor location (Figure 5(A)).Figure 5.Development and validation of a clinical-genomic nomogram for predicting complete response (CR) (A) Integrated clinical-genomic nomogram for predicting the probability of complete response (CR) following neoadjuvant therapy (NAT). The model incorporates four independent predictors: immunotherapy-combined modality, SYNE1 mutation status, MRF status, and tumor location. (B) Receiver operating characteristic (ROC) curve evaluating the discriminative performance of the predictive model. (C) Calibration curve of the integrated nomogram. The plot illustrates the agreement between predicted probabilities and observed CR frequencies. The diagonal dashed line represents the ideal reference (perfect calibration). (D) Decision curve analysis (DCA) illustrating the clinical net benefit of using the nomogram across a range of threshold probabilities. (E) Risk stratification of patients based on nomogram-predicted CR probabilities. Observed CR rates are shown for patients stratified into low, intermediate, and high probability groups. Statistical significance for the stepwise increase in CR rates was determined by the Cochran-Armitage trend test. Development and validation of a clinical-genomic nomogram for predicting complete response (CR) (A) Integrated clinical-genomic nomogram for predicting the probability of complete response (CR) following neoadjuvant therapy (NAT). The model incorporates four independent predictors: immunotherapy-combined modality, SYNE1 mutation status, MRF status, and tumor location. (B) Receiver operating characteristic (ROC) curve evaluating the discriminative performance of the predictive model. (C) Calibration curve of the integrated nomogram. The plot illustrates the agreement between predicted probabilities and observed CR frequencies. The diagonal dashed line represents the ideal reference (perfect calibration). (D) Decision curve analysis (DCA) illustrating the clinical net benefit of using the nomogram across a range of threshold probabilities. (E) Risk stratification of patients based on nomogram-predicted CR probabilities. Observed CR rates are shown for patients stratified into low, intermediate, and high probability groups. Statistical significance for the stepwise increase in CR rates was determined by the Cochran-Armitage trend test. The clinical-genomic nomogram demonstrated robust discriminative performance, with AUC of 0.705 (95% CI: 0.65–0.76) (Figure 5(B)). Calibration analysis showed strong agreement between the nomogram-predicted probabilities and the actual observed CR rates, with the bias-corrected calibration curve closely aligned with the ideal reference line (Figure 5(C)). The clinical utility of the model was further evaluated through DCA. The integrated nomogram provided a consistently higher net benefit compared to both “treat-all” (universal surgical intervention) and “treat-none” (W&W) strategies across a wide range of clinically relevant threshold probabilities, approximately between 17.9% and 70.0% (Figure 5(D)), demonstrating a wide therapeutic window for the nomogram to guide treatment de-escalation. To further verify the overall reliability of the model, we evaluated its ability to distinguish different response gradients by categorizing patients into three distinct risk groups based on predicted probabilities. The observed CR rates increased progressively and significantly across the low (19.6%), intermediate (40.2%), and high (54.1%) probability groups (p < 0.001) (Figure 5(E)). While internal validation suggests the nomogram is promising for guiding non-operative management, its clinical utility requires further confirmation in external datasets. Discussion In this comprehensive genomic–clinical analysis of patients with rectal cancer undergoing NAT, we identified key determinants of treatment response and survival outcomes in a predominantly MSS population. Immunotherapy-combined NAT (iTNT) and SYNE1 mutations emerged as independent predictors of CR, whereas unfavorable clinical features, including higher tumor location and MRF involvement, were associated with nCR. Furthermore, KRAS mutations, specifically the G12D subtype, were linked to inferior DFS and a higher risk of metastatic progression, particularly lung involvement. These findings provide an integrated framework for understanding response heterogeneity and refining precision neoadjuvant strategies in rectal cancer. Building upon these clinical and genomic determinants, we developed a predictive nomogram that enables individualized estimation of CR probability, offering a practical tool to support the selection of candidates for organ-preservation strategies. Tumor response heterogeneity remains a significant obstacle in implementing organ-preserving strategies for rectal cancer. In our study, patients receiving iTNT achieved a markedly superior CR rates compared to those receiving conventional CRT. This aligns with emerging evidence from prospective trials,8–11,22–24 including the TORCH study conducted at our center. 9 These findings reinforce the paradigm that radiotherapy-induced immunogenic modulation may prime the tumor microenvironment, enhancing sensitivity to immune checkpoint blockade in otherwise immunologically “cold” MSS tumors. 25 Conversely, the independent association of MRF involvement with poor regression confirms that adverse baseline radiological features often reflect a more aggressive biological phenotype. 26 A recent study of 585 LARC patients reported that baseline MRI-detected MRF negativity was associated with a 3.21-fold lower likelihood of achieving pCR after neoadjuvant CRT, underscoring MRF involvement as an independent negative predictor of treatment response. 27 Furthermore, persistent MRF involvement after treatment remains a robust independent predictor of poor overall survival. 28 These observations underscore the importance of baseline MRF assessment in risk stratification and support the ongoing exploration of dose-escalated radiotherapy or combined immunotherapy for this high-risk subset.29,30 A key finding of this study is the identification of SYNE1 mutation as a potent independent predictor of CR. This aligns with prior observations by Hammarström et al., who suggested a link between SYNE1 alterations and favorable NAT response in LARC cohort. 15 Our findings significantly extend this association to the realm of immunotherapy. While MSS tumors are generally considered poor responders to immune checkpoint blockade, our data reveal that SYNE1-mutant tumors exhibit a significantly higher TMB than their wild-type counterparts. Although the absolute TMB values in these cases remain below the classical MSI-H threshold, this incremental increase is clinically profound. This notion is also supported by recent evidence, such as in ovarian cancer and renal cell carcinoma, which has similarly demonstrated that SYNE1 mutation is associated with increased tumor mutation burden and enhanced immune cell infiltration.31,32 Biologically, this phenomenon may be explained by the critical role of Nesprin-1 (encoded by SYNE1) in maintaining the nuclear localization of mismatch repair (MMR) proteins. SYNE1 mutations can lead to the aberrant nuclear distribution of MSH2 and MSH6, thereby impairing the efficiency of mismatch repair and causing a gradual accumulation of sporadic mutations across the genome. 33 This deficit does not reach the threshold of MSI but is sufficient to result in a mild-to-moderate elevation of TMB through a non-canonical, localization-dependent mechanism. Furthermore, evidence from other “cold” tumors like ovarian cancer supports the role of SYNE1 in modulating the immune microenvironment. In SYNE1-mutant ovarian cancer, researchers observed that SYNE1 mutations significantly up-regulate immune pathways, including immune cell migration and humoral immune responses. 31 These observations collectively suggest that the SYNE1 mutation serves as a surrogate marker for intermediate genomic instability. By elevating the neoantigen load and fostering a more inflamed microenvironment without reaching a full MSI-H phenotype, these mutations enhance the visibility of the tumor to the immune system. This biological axis offers a robust molecular handle to identify the subset of MSS rectal cancer patients most likely to benefit from immunotherapy-integrated regimens. In contrast to SYNE1, KRAS mutations were primarily associated with disease progression rather than treatment response. Beyond its known role in reducing DFS, we found that the KRAS G12D subtype conferred the highest risk of metastatic relapse (2-year DMFS 71.3%). Although previous studies correlate KRAS mutations with pulmonary metastasis, 34 our study observed a potential pulmonary organotropism for G12D mutations. This suggests different KRAS codons may modulate distinct downstream signaling pathways, influencing the metastatic niche and dissemination patterns. Recognizing these subtype-specific risks is crucial for tailoring post-operative surveillance and developing targeted interventions to prevent distant recurrence. To bridge the gap between genomic research and clinical practice, we developed an integrated clinical–genomic nomogram. With an AUC of 0.705 and robust calibration, the model successfully demonstrated reliable performance in distinguishing response gradients. In the era of organ preservation, the W&W strategy hinges on the accurate prediction of CR. By incorporating both clinical and molecular markers, our nomogram offers a framework for a more nuanced decision-support tool than clinical staging alone. Although its immediate routine application in daily practice is currently constrained by the single-center, retrospective nature of our cohort, this model illustrates the potential utility of integrating genomic stratification into organ-preservation algorithms. Therefore, while internal validation was robust, multi-center validation is required before this model can be recommended for routine clinical use. Limitation This study leverages a large Chinese cohort of rectal cancer patients, integrating genomic profiling with clinical outcomes. However, several limitations should be noted. First, some of our patients received concurrent irinotecan or oxaliplatin during neoadjuvant radiotherapy. These regimens differ from the standard treatment recommended by European and US guidelines, which may limit the broader application of findings. Second, the single-center design limits the generalizability of the results to other populations, and external validation in independent multicenter cohorts is required before routine clinical implementation. Third, the relatively short follow-up period may not fully capture long-term survival trends or late-onset metastases. Fourth, mechanistic insights regarding the biological role of SYNE1 and KRAS variants remain largely inferential and require functional validation. Future functional studies are necessary to validate these biological pathways. Conclusion In conclusion, this study demonstrates that clinical and genomic factors jointly shape therapeutic outcomes in MSS rectal cancer. The iTNT modality and SYNE1 mutations significantly improves CR rates, while KRAS G12D mutations signal a high risk of distant metastasis. The clinical-genomic nomogram developed here offers a practical framework for individualized treatment selection, advancing the transition toward precision neoadjuvant therapy and supporting the safe implementation of organ-preserving approaches in rectal cancer. Supplemental material Supplemental material - Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer Supplemental material for Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer by Lijie Zhang, Jingwen Wang, Juefeng Wan, Yan Wang, Hui Zhang, Yaqi Wang, Fan Xia, Lijun Shen, Zhen Zhang in Therapeutic Advances in Medical Oncology Supplemental material - Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer Supplemental material for Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer by Lijie Zhang, Jingwen Wang, Juefeng Wan, Yan Wang, Hui Zhang, Yaqi Wang, Fan Xia, Lijun Shen, Zhen Zhang in Therapeutic Advances in Medical Oncology Supplemental material - Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer Supplemental material for Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer by Lijie Zhang, Jingwen Wang, Juefeng Wan, Yan Wang, Hui Zhang, Yaqi Wang, Fan Xia, Lijun Shen, Zhen Zhang in Therapeutic Advances in Medical Oncology Supplemental material - Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer Supplemental material for Integrating clinical and genomic features to predict response to neoadjuvant therapy in microsatellite-stable rectal cancer by Lijie Zhang, Jingwen Wang, Juefeng Wan, Yan Wang, Hui Zhang, Yaqi Wang, Fan Xia, Lijun Shen, Zhen Zhang in Therapeutic Advances in Medical Oncology