ASO Author Reflections: CT-Based Radiomics Model to Predict Lateral Lymph Node Metastasis After Neoadjuvant (Chemo)Radio
- PDF / 216,218 Bytes
- 2 Pages / 595.276 x 790.866 pts Page_size
- 38 Downloads / 193 Views
ASO AUTHOR REFLECTIONS
ASO Author Reflections: CT-Based Radiomics Model to Predict Lateral Lymph Node Metastasis After Neoadjuvant (Chemo)Radiotherapy in Advanced Low Rectal Cancer Ryota Nakanishi, MD, PhD, and Takashi Akiyoshi, MD, PhD Department of Gastroenterological Surgery, Cancer Institute Hospital, Japanese Foundation for Cancer Research, Tokyo, Japan
PAST Recent evidence suggests that neoadjuvant (chemo)radiotherapy [(C)RT] is not sufficient to prevent lateral local recurrence in locally advanced rectal cancer.1 Lateral pelvic lymph node (LPLN) dissection (LPLND) might help to reduce such recurrence rates2 but LPLND remains challenging because it often results in increased postoperative complications and sexual/urinary dysfunction. Therefore, it is imperative that LPLND be performed only in patients who will benefit. Unfortunately, the accurate identification of patients with LPLN metastasis after neoadjuvant (C)RT using conventional imaging is difficult. Recent studies suggest that radiomics can improve the prediction of pathological lymph node metastasis3 through the extraction and analysis of numerous quantitative radiomics features from digital medical images; however, there are no studies as yet evaluating a radiomics-based predictive model for LPLN metastasis after neoadjuvant (C)RT among patients with rectal cancer. PRESENT In the present study,4 we developed and validated a new radiomics-based predictive model for LPLN metastasis. A total of 247 patients with rectal cancer and enlarged LPLNs treated by (C)RT and LPLND were retrospectively
Ó Society of Surgical Oncology 2020 First Received: 29 June 2020 Accepted: 2 July 2020 T. Akiyoshi, MD, PhD e-mail: [email protected]
analyzed (175 in the primary cohort and 72 in the validation cohort). LPLN radiomics features were extracted from pretreatment portal venous-phase computed tomography images, and a radiomics score of LPLN was constructed based on least absolute shrinkage and selection operator regression in the primary cohort. Our model shows better discrimination than measurements of the LPLN pretreatment short-axis diameter, both in the primary cohort (area under the curve [AUC] 0.91 vs. 0.83; p = 0.0015) and in the validation cohort (AUC 0.90 vs. 0.80; p = 0.0298). Decision curve analysis also indicates the superiority of the radiomics score. Our data suggest that radiomics-based prediction modeling provides individualized risk estimation of LPLN metastasis in rectal cancer patients treated with (C)RT and outperforms measurements of pretreatment LPLN diameter. FUTURE The main limitation of our study is its retrospective design and use of data from only two institutions. Our results should be validated using multicenter, prospective data with a larger number of patients. Second, we extracted radiomics features from computed tomography (CT) imaging even though magnetic resonance imaging (MRI) is the standard imaging modality for local staging of rectal cancer. CT imaging has a relatively stable scanning protocol, whereas there are nume
Data Loading...
