Skip to main content
2025 ASCRS Annual Scientific Meeting Main banner
Final Program


Visit ASCRS U
Icon Legend
This session is not in your schedule.
This session is in your schedule. Click again to remove it.
Presentation Icons
CME
CME
Invitation/Ticketed Event
Invitation/Ticketed Event
Invited Lectureship
Invited Lectureship
Award Recipient
Award Recipient
Self-Assessment
Self-Assessment
Social
Social
Industry
Industry
Back

Session: ePoster: Other Neoplasia and Endoscopy 1 (Sun8)

EP323 - Predictive value of a LASSO-based model utilizing PET/CT parameters for adenoma detection in healthy patients

Sunday, May 11, 2025
9:11 AM – 9:16 AM PT
Location: Monitor 7, Sails Pavilion, Upper Level

    ePoster Presenter(s)

  • Jeonghyun Kang, MD, PhD, FACS (he/him/his)

    Professor
    Gangnam Severance Hospital, Yonsei University College of Medicine
    Seoul, Seoul-t'ukpyolsi, Republic of Korea

    All of the relevant financial relationships listed below have been mitigated.
    This individual has no financial relationships with ineligible companies.

    • Author(s)

  • JL

    Jae-Hoon Lee, M.D., Ph.D.

    Professor
    Yonsei University College of Medicine
    Seoul, Seoul-t'ukpyolsi, Republic of Korea

    Disclosure information not submitted.

All of the relevant financial relationships listed below have been mitigated.
Jeonghyun Kang, MD, PhD, FACS: This individual has no financial relationships with ineligible companies.
Image or Table (Do Not add text here):

Purpose/Background: Adenoma detection during colonoscopy is recognized as an effective strategy for reducing the incidence of colorectal cancer. However, many individuals are reluctant to undergo colonoscopy due to the discomfort associated with the procedure. Several studies have suggested a relationship between adenoma prevalence and body composition, particularly visceral fat. Despite this, there is limited research on the utility of PET scans in detecting adenomas in patients who have undergone health checkups involving PET-CT. This study aimed to develop a predictive model for adenoma detection by utilizing clinical data and variables derived from PET/CT.

Methods/Interventions: This retrospective study included data from 754 adults who underwent both PET/CT scans and colonoscopy within 30 days of each other as part of a health checkup between 2014 and 2020. Patients with compromised colonoscopy results or non-relevant biopsy findings were excluded from the analysis. Using clinical and PET derived parameters, the least absolute shrinkage and selection operator (LASSO) regression model was applied for variable selection and predictive signature building in the training set. The predictive accuracy of our model and other clinical value was compared using area under the receiver operating characteristic (AUROC), and decision curve analysis (DCA) in the test set.

Results/Outcomes: The study population comprised 461 males and 293 females, with a median BMI of 24.9 kg/m². Colonoscopy results revealed that 258 patients (34.2%) had adenomas, 92 patients (10.8%) had polyps, and 414 patients (54.9%) had neither. The cohort was divided into a training set (n=452) and a test set (n=302). A predictive model was constructed using LASSO regression in the training set. Among the various parameters evaluated, metrics related to visceral adipose tissue (VAT), including VAT area, VAT Hounsfield unit (HU), and VAT mean, were the most significant predictors for adenoma detection, with AUROC values of 0.630, 0.585, and 0.579, respectively, in the test set. The LASSO model, which integrated clinical and radiologic metrics, demonstrated superior performance, achieving an AUC of 0.667 in the test set, outperforming the individual VAT parameters. Comparison of the AUROC curves and DCA indicated that the LASSO model outperformed other predictive methods in the test sets.

Conclusion/Discussion: The study concluded that the LASSO model is a reliable and effective tool for predicting adenoma, with the potential to be used clinically for early detection and prevention of colorectal adenocarcinoma.