RPS 1616 - Recent developments in colorectal cancer imaging

Author Block: J. Dupont¹, S. JOURNO¹, R. Barbe¹, P-H. Cournède², Y. Menu¹, C. S. Balleyguier¹, A. HOLLEBECQUE¹, T. André³, N. Lassau¹; ¹Villejuif/FR, ²Gif-Sur-Yvettes/FR, ³Paris/FR
Purpose: To evaluate whether variations (delta-) in total tumour volume (TTV) and radiomic features can early predict Overall Survival (OS) in rare cohort of patients with metastatic colorectal cancer (mCRC) with high microsatellite instability (MSI-H) undergoing immunotherapy.
Methods or Background: One hundred forty-five patients diagnosed with MSI-H mCRC (mean age, 56 ± 15 years; 89 men) were retrospectively included from two centers. Radiologists manually annotated all tumours on chest-abdomen-pelvis computed-tomography at baseline (BL) and first evaluation (E1) scans. The median time between BL and E1 was 50 days (interquartile range: 45 - 64 days).

Tumour volume and radiomics were extracted for each lesion to predict OS. Every scan was resampled to isotropic voxel spacing of 1×1×1 mm3 to reduce variability. The delta-TTV and delta-radiomics were defined as their relative between BL and E1.

The ability to stratify patients into long- and short-term survival groups was assessed using the Restricted Mean Survival Time (RMST) ratio, AUC at 3 years, and C-index. Generalization was assessed through 75 train/test cohort splits.
Results or Findings: A total of 2079 lesions at BL and 2112 at E1 were annotated.

Delta-TTV outperformed TTV at BL and E1 across all metrics (p<0.001). Specifically, delta-TTV achieved an RMST ratio of 2.215 (95% CI: [2.106-2.324]), an AUC of 0.854 [0.840-0.868], and a C-index of 0.830 [0.819-0.841], with a cut-off of 17.31% relative increase in TTV.

Combining delta-radiomics with delta-TTV enhanced performance, yielding an RMST ratio of 2.450 (2.248-2.652, p=0.019), an AUC of 0.875 (0.862-0.888, p<0.001), and a C-index of 0.839 (0.819-0.841, p<0.001).
Conclusion: Delta-TTV is a strong predictor of survival in MSI-H mCRC patients. Adding radiomic features enhances this biomarker’s efficiency for early survival prediction.
Limitations: External validation is recommended for this two-center retrospective study.
Funding for this study: No fundings to disclose
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ethics committee N°2020-CER 2020-6

Author Block: T. J. Vogl, A-I. Nica, C. Booz, I. Yel, C. Wolfram, H. Adwan; Frankfurt/DE
Purpose: To evaluate the efficacy of transarterial chemoperfusion (TACP) and transarterial chemoembolization (TACE) as palliative and symptomatic treatments for unresectable colorectal carcinoma (CRC), focusing on local tumor response and survival outcomes.
Methods or Background: Between January 2000 and October 2023, 67 consecutive patients diagnosed with unresectable CRC underwent 318 TACP and 80 TACE procedures. Of these, 48 patients were treated with TACP, 14 with TACE, and 5 received a combination of both therapies. We retrospectively assessed local tumor response using the Response Evaluation Criteria in Solid Tumors (RECIST) and calculated overall survival (OS) and progression-free survival (PFS) using the Kaplan-Meier estimator.
Results or Findings: No major complications were reported. A total of 49 patients (73%) exhibited stable disease, while 3 patients (5%) achieved a partial response. The rate of progressive disease was 22%. The median OS for the entire cohort was 16.17 months, and the median PFS was 11.25 months. There were no statistically significant differences in OS (p=0.82) or PFS (p=0.37) between patients receiving TACP, TACE, or both treatments.
Conclusion: TACP and TACE are minimally invasive procedures that provide a viable treatment option for patients with unresectable CRC who cannot tolerate, refuse, or do not respond to chemotherapy. These treatments can serve as both palliative and symptomatic measures, potentially preventing tumor progression and improving quality of life. However, their benefits in the treatment of CRC warrant further investigation.
Limitations: Limitations of the study include its retrospective nature.
Funding for this study: No funding.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Ethics Committee approval was obtained for this study.

Author Block: L. Tang, X. Liu, C. FengPing, G. Liu, X. Lv, X. Liu, W. C. Chen; Guangzhou , China/CN
Purpose: To evaluate the diagnostic utility of quantitative parameters derived from dual layer spectrum CT (DLSCT) in distinguishing between metastatic and non-metastatic lymph nodes (LNs) in colorectal cancer (CRC).
Methods or Background: In this retrospective study, 78 metastatic LNs and 133 non-metastatic LNs were analyzed. Morphological characteristics and quantitative DLSCT parameters were evaluated and compared between the two groups. Univariate and multivariable logistic regression analyses were performed to identify independent factors for LN metastasis. Diagnostic performance and clinical utility were evaluated using receiver operating characteristic (ROC) curve analysis and decision curve analysis (DCA).
Results or Findings: Significant differences in all morphological features were observed between metastatic and non-metastatic LNs. Metastatic LNs showed significantly lower iodine concentration (IC) and normalized iodine concentration (NIC) in the arterial phase, along with higher effective atomic number (Zeff), normalized Zeff in the venous phase, and slope of the spectral Hounsfield unit curve (λHu) in both phases. A positive correlation was identified between λHu values of LNs and primary tumors in both arterial and venous phases within the metastatic group. The model based solely on morphological features achieved an AUC of 0.695. In contrast, the multivariate model incorporating DLSCT parameters significantly improved diagnostic performance, yielding an AUC of 0.901.The DeLong test and DCA comfirmed that quantitative multiparameter model provided superior predictive performance and greater clinical benefit compared to the morphological assessment alone.
Conclusion: Quantitative parameters derived from DLSCT demonstrate superior diagnostic efficacy in differentiating metastatic from non-metastatic LNs in CRC.
Limitations: Our findings indicate that the utilization of DLSCT-derived parameters provides significant value in differentiating lymph node metastasis in colorectal cancer patients by enabling the accurate assessment of tissue composition. The combination n of DLSCT multiple parameters can enhance the accuracy of LN assessment.
Funding for this study: This work was supported by Guangzhou Basic and Applied Basic Research Foundation (Grant No. 2023A03J0245), National Nature Science of Foundation of China (Grant No.82202259), and Youth Talent Project of The Second Affiliated Hospital of Guangzhou University of Chinese Medicine (Grant No. ZY2022YL05).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: This retrospective study was approved by the institutional review board of our hospital (No BE2023-146) and was exempt from the requirement for informed consent due to its design.

Author Block: S. Xiang¹, L. XIAO²; ¹Chongqing/CN, ²Shanghai/CN
Purpose: To develop and validate a machine learning (ML) model integrating dual-energy CT (DECT) features for the preoperative prediction of lymph node metastasis (LNM) in colorectal cancer, and to rigorously assess its incremental value and clinical utility.
Methods or Background: This retrospective study enrolled 210 colorectal cancer patients with preoperative DECT. They were randomly divided into training (n=168) and validation (n=42) cohorts. Using clinical, conventional CT, and DECT features, five nested models were built with five machine learning algorithms (including XGBoost) and rigorously evaluated. Validation performance was assessed by the AUC with 1000-bootstrapping and compared using the DeLong test. Clinical utility and interpretability were examined by decision curve analysis, reclassification, and SHAP analysis, while imaging phenotypes were identified via unsupervised clustering.
Results or Findings: The optimal XGBoost model, combining clinical data with lymph node features from both conventional and DECT imaging, achieved an AUC of 0.8882 (95% CI: 0.7780–0.9760) in the validation set. The addition of DECT features significantly improved predictive performance over a model with only conventional CT features (P=0.0153). DCA demonstrated a clear net benefit across a wide range of threshold probabilities (9%–86%). SHAP analysis identified a DECT-derived feature (arterial phase 100KeV CT value) as the single most important predictor. Furthermore, unsupervised clustering revealed three distinct "imaging habitats" that strongly correlated with LNM risk (30.2% vs. 42.1% vs. 82.4%).
Conclusion: An explainable ML model incorporating DECT features accurately predicts LNM in colorectal cancer and demonstrates clear clinical utility. DECT provides significant incremental diagnostic value over conventional imaging. This robustly validated approach, along with the novel concept of "imaging habitats," shows high potential for improving personalized risk stratification and guiding clinical decision-making.
Limitations: Small sample.
Funding for this study: No funding was received by this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Written informed consent obtained from all participants.

Author Block: L. Van der Snee, A. Kramer, N. Kok, A. Aalbers, M. Lopez, V. Retel, M. Lahaye; Amsterdam/NL
Purpose: In patients with colorectal peritoneal metastases (PM-CRC), cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS-HIPEC) is potentially life-prolonging but carries substantial morbidity. Standard staging with computed tomography (CT) and diagnostic laparoscopy often results in preventable surgical procedures. The randomized multicenter DISCO trial demonstrated that adding diffusion-weighted magnetic resonance imaging (DW-MRI) reduces such procedures. This study evaluated the cost-effectiveness of additional DW-MRI in the Netherlands.
Methods or Background: A decision-analytic cohort model combined a decision tree of diagnostic strategies with a Markov model simulating survival. Patients were staged with either standard care (CT ± laparoscopy) or additional DW-MRI. Model inputs were derived from 149 Dutch patients randomized in DISCO (DW-MRI n=74, 49.7%; standard care n=75, 50.3%). Outcomes included short-term metrics (staging laparoscopies, CRS-HIPEC rates, diagnostic costs) and long-term outcomes (quality-adjusted life years [QALYs], total costs). Deterministic and probabilistic sensitivity analyses tested robustness.
Results or Findings: DW-MRI markedly reduced staging laparoscopies and lowered diagnostic costs, saving €2087 per patient. Over a lifetime horizon, DW-MRI yielded a gain of 0.162 QALYs and total cost savings of €455 compared with standard care. Both deterministic and probabilistic analyses confirmed cost-effectiveness. National implementation could generate annual diagnostic savings of ~€2.8 million.
Conclusion: Adding DW-MRI to the staging work-up for PM-CRC is a cost-effective strategy that reduces unnecessary surgical procedures and patient burden. In the Dutch healthcare setting, DW-MRI implementation would improve patient selection for CRS-HIPEC and result in significant cost savings.
Limitations: This analysis was based on Dutch trial data, and results may not fully generalize to healthcare systems with different surgical practices, cost structures, or MRI expertise.
Funding for this study: This study was investigator-initiated and sponsored by ZonMw. This study was approved by the Medical Ethics Committee of the Netherlands Cancer Institute (13 September 2019; NL70045.031.19) and registered in the ClinicalTrials.gov registry of the U.S.
National Library of Medicine (NCT04231175), and the study protocol was published in BMC Cancer (PMID: 33902498).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The study was approved by the ethics committee.

Author Block: F. Abubacker Sulaiman, S. Zainamb Begum, R. Praveenkumar, M. Nivitha, J. Lydia, A. Raashid Ibrahim; Chennai/IN
Purpose: To evaluate the correlation between ^18F-FDG uptake parameters on PET/CT and serum tumor markers in patients with gynecologic and gastrointestinal malignancies, and to assess their combined prognostic and diagnostic utility in disease characterization.
Methods or Background: A prospective study was conducted on 80 patients—40 with gynecologic (ovarian, endometrial, cervical) and 40 with gastrointestinal (colorectal, gastric, pancreatic) cancers—who underwent baseline ^18F-FDG PET/CT and serum tumor marker analysis (CA-125, CEA, and CA 19-9). Quantitative PET parameters including SUVmax, SUVmean, metabolic tumor volume (MTV), and total lesion glycolysis (TLG) were recorded. Correlations between imaging and biochemical markers were analyzed using Pearson’s correlation coefficient and regression modeling. Diagnostic and prognostic significance were assessed against histopathologic grade and clinical stage.
Results or Findings: A strong positive correlation was observed between SUVmax and CA-125 in ovarian and endometrial carcinomas (r = 0.76, p < 0.001), and between SUVmax and CEA in colorectal cancer (r = 0.71, p < 0.001). Elevated TLG values correlated with advanced FIGO and TNM stages. Combined FDG and tumor marker evaluation improved detection of metastatic disease by 19% and provided better prognostic stratification for recurrence risk. Patients with discordantly high CA-125 or CEA and low FDG uptake demonstrated low-grade or mucinous histology, reflecting reduced glycolytic activity.
Conclusion: FDG uptake shows significant correlation with serum tumor markers in gynecologic and GI malignancies, supporting their complementary role in staging, prognosis, and treatment monitoring.
Limitations: Single-center study with modest sample size and absence of longitudinal follow-up. Multicentric validation is recommended to standardize combined metabolic–biochemical assessment.
Funding for this study: No external funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: Institutional ethical committee approval was obtained.

Author Block: Y. Wang¹, G. Miao¹, Y. Wang¹, C. Fu², W-R. Wei-Ren Liu¹, L. Liu¹; ¹Shanghai/CN, ²Shenzhen/CN
Purpose: Neoadjuvant chemotherapy is a standard approach for treating colorectal cancer, yet a significant proportion of patients exhibit resistance due to tumor hypoxia. This study aimed to evaluate the potential of intravoxel incoherent motion diffusion-weighted imaging (IVIM-DWI) and blood oxygenation level-dependent MRI (BOLD-MRI) techniques to non-invasively monitor tumor hypoxia dynamics and guide the timing of chemotherapy.
Methods or Background: Tumor-bearing mice were randomly assigned to control, unguided chemotherapy, and BOLD MRI-guided chemotherapy groups, with 5-fluorouracil administered according to either fixed schedules or BOLD-MRI-monitored hypoxia rhythms (T2* < 20 ms or lower than last day).
Results or Findings: MRI-guided group achieving the lowest tumor-to-weight ratio (1.76 vs 2.20 in unguided and 5.59 in control). T2* values exhibited strong positive correlation with HIF-1α expression (r = 0.6806, p = 0.0436) and CD31 (r = 0.7053, p = 0.0338), and a very strong negative correlation with vascular maturity index (VMI) (r = –0.8618, p = 0.0028).
Conclusion: By identifying optimal oxygenation phases, BOLD-MRI can guide the timing of chemotherapy to align with periods of improved perfusion, demonstrating superior tumor suppression and stable IVIM value trends in the guided group. Therefore, hypoxia-related functional MRI provides a promising imaging-based framework for optimizing chemotherapy timing and evaluating therapeutic response in colorectal cancer.
Limitations: Not applicable.
Funding for this study: This work was supported by the National Natural Science Foundation of China (Grant No. 82371920), Shanghai Sailing Program (Grant No. 23YF1441700).
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: The ethics committee notification can be found under the number SYXK2016–0006.