Improved Prognosis of Treatment Failure in Cervical Cancer with Nontumor PET/CT Radiomics

Abstract

Background: Radiomics has been applied to predict recurrence in several disease sites, but current approaches are typically restricted to analyzing tumor features, neglecting non-tumor information in the rest of the body. The purpose of this work was to develop and validate a model incorporating non-tumor radiomics, including whole body features, to predict treatment outcomes in patients with previously untreated locoregionally advanced cervical cancer. Methods:We analyzed 127 cervical cancer patients treated definitively with chemoradiotherapy and intracavitary brachytherapy. All patients underwent pretreatment whole body 18 F-FDG PET/CT. To quantify effects due to the tumor itself, the gross tumor volume (GTV) was directly contoured on the PET/CT. Meanwhile, to quantify effects arising from the rest of the body, the planning target volume (PTV) was deformably registered from each planning CT to the PET/CT, and a semi-automated approach combining seed-growing and manual contour review generated whole body muscle, bone, and fat segmentations on each PET/CT. A total of 965 radiomic features were extracted for GTV, PTV, muscle, bone, and fat. 95 patients were used to train a Cox model of disease recurrence including both radiomic and clinical features (age, stage, tumor grade, histology, and baseline complete blood cell counts), using bagging and split-sample-validation for feature reduction and model selection. To further avoid overfitting, the resulting models were tested for generalization on the remaining 32 patients, by calculating a risk score based on Cox regression and evaluating the c-index (c-index > 0.5 indicates predictive power). Results: Optimal performance was seen in a Cox model including one clinical biomarker (whether or not a tumor was stage III-IVA), two GTV radiomic biomarkers (PET gray-level size-zone matrix small area low gray level emphasis and zone entropy), one PTV radiomic biomarker (major axis length) and one whole body radiomic biomarker (CT Bone root mean square). In particular, stratification into high-and lowrisk groups, based on the linear risk score from this Cox model, resulted in a hazard ratio [95% CI] of 0.019 [0.004, 0.082], an improvement over stratification based on clinical stage alone, which had a hazard ratio of 0.36 [0.16, 0.83]. Conclusions: Incorporating non-tumor radiomic biomarkers can improve the performance of prognostic models compared to using only clinical and tumor radiomic biomarkers. Future work should look to further test these models in larger, multiinstitutional cohorts.

Treatment of Recurrent Nasopharyngeal Carcinoma: A Sequential Challenge

“…The PET-CT approach involves the use of radionuclide labeling of compounds involved in human tissue metabolism, signal detection, and then image reconstruction [ 41 ]. This is used to distinguish tumor tissue with relatively high metabolism from non-tumor tissues with low metabolism [ 42 , 43 ]. However, its limitation is that some inflammatory areas will also show high metabolic signals, which are more likely to interfere with the judgment [ 44 ].…”

Recurrence risk stratification for locally advanced cervical cancer using multi-modality transformer network

“…Additionally, CT and MR images can provide information of tumors, such as lesion size and invasion degree, which is crucial for preliminary clinical staging and prognosis evaluation ( 47 – 51 ). Therefore, many models based on CT or MR images have been proposed for the subtype identification ( 52 ), staging analysis ( 53 , 54 ), lymph node metastasis prediction ( 54 , 55 ) and prognosis analysis ( 12 , 56 , 57 ) of cervical cancer. Compared to the above methods, the main contributions of this paper lie in the following aspects: (I) We first investigated the feasibility of deep learning method in accurately predicting recurrence risk so as to help formulate the individualized therapeutic schedule for LACC patients.…”

“…For example, some work has carried out texture analysis based on positron emission tomography (PET) or magnetic resonance (MR) images to predict the recurrence risk of cervical cancer ( 9 , 10 ). In addition, the ultrasound (US) and computed tomography (CT) images were also used in recurrence-related tasks, such as lymph node metastasis prediction and survival assessment ( 11 , 12 ). However, few studies have tried to focus on the recurrence risk stratification of LACC.…”

Treatment planning computed tomography radiomics for predicting treatment outcomes and haematological toxicities in locally advanced cervical cancer treated with radiotherapy: A retrospective cohort study

“…Radiomics has been found to be related to lymph node metastases, 14–16 treatment response, 17–19 survival 20–22 and tumour recurrence 17,23 for cervical cancer patients 24,25 . There are no studies to investigate the ability of radiomic features of computed tomography (CT) images obtained for radiotherapy planning to predict clinical end points in LACC.…”

“…10,11 Radiomic features may reflect the tumour's intrinsic properties and can be used as independent predictors of survival outcome with higher predictive ability than traditional clinical parameters alone. 12,13 Radiomics has been found to be related to lymph node metastases, [14][15][16] treatment response, [17][18][19] survival [20][21][22] and tumour recurrence 17,23 for cervical cancer patients. 24,25 There are no studies to investigate the ability of radiomic features of computed tomography (CT) images obtained for radiotherapy planning to predict clinical end points in LACC.…”