Cervical cancer (CC) is one of the most common gynecological malignancies worldwide, and its incidence is gradually increasing, particularly in the younger population [1]. Lymph node metastasis (LNM) is an important factor which can reflect the clinical prognosis of CC patients. It is also an important basis for guiding postoperative adjuvant therapy [2]. For early-stage CC patients with LNM, pelvic and/or para-aortic lymphadenectomy is usually recommended to avoid radical surgery; for locally advanced CC patients with LNM, radiotherapy should be considered to reduce the recurrence rate [3]. Thus, accurate identification of LNM is crucial for predicting prognosis and choosing the best available treatment.

Various criteria have been proposed to predict the metastatic involvement of a cervical lymph node. The latest revised version of the International Federation of Gynecology and Obstetrics (FIGO) staging scheme in 2018 has formally incorporated LNM into the CC staging system (i.e., stage IIIC) [4], and permitted of the identification of LNM based on imaging and pathological findings [5]. Because of the higher soft tissue resolution, MRI can visually display the details of tumor anatomy from various angles, as well as lymph nodes and vascular structures surrounding the tumor. Thus, MRI has obvious advantages over other imaging examinations in judging CC tumor size, depth of invasion and LNM [6]. Conventional MR images (unenhanced and enhanced) for detecting CC LNM mainly depend on morphological features such as the volume and shape of the lymph nodes. Some of the diagnostic imaging criteria (CT and MRI) for LNM should include [7]: (1) short-axis diameter in axial plane ≥10 mm; (2) longest axial diameter cut-off criteria depending on which performance characteristic is of most interest; (3) cluster of ≥3 borderline nodes (each ≥8 mm short-axis diameter, except >9 mm in the level II/subdigastric region); (4) long-to-short axis ratio < 2 (i.e., rounder).However, metastatic and proliferative lymphadenopathy often have similar MR imaging features [8], and the identification of the two still may be challenging.

Functional MRI (fMRI) can obtain a variety of quantitative data such as tissue pathophysiology, hemodynamics, substance metabolism at the molecular level, revealing the microstructure information of tumors, which is helpful for non-invasive and accurate assessment of tumor biological behavior before treatment [9]. Amide proton transfer weighted (APTw) imaging is a novel contrast media-free molecular MR imaging method based on the chemical exchange saturation transfer (CEST)mechanism. APTw can detectthe chemical exchange rate of between water and endogenous mobile proteins, peptides or polypeptides (e.g., those dissolved in the cytoplasm), which reflects the changes in tissue protein content and pH environment [10,11].

In living organisms, due to the existence of microstructures such as cell membranes, organelles, and intercellular spaces in the tissue microenvironment, the diffusion motion of water molecules is mostly with non-Gaussian distribution [12]. Diffusion kurtosis imaging (DKI) is a phenomenological model that directly estimates and quantifies the non-Gaussian nature of water molecule diffusion in the body, and reflects the limited diffusion movement of water molecules in tissue and the complexity of tissue microstructure [13]. The purpose of our study was to explore the value of APTw combined with DKI in predicting CC LNM and find robust quantitative prediction parameters to provide valuable information for the clinical decision-making of individualized treatment for CC patients.