Preoperative Prediction Model Research Articles

Preoperative Risk Stratification of Postoperative Pancreatic Fistula: A Risk-Tree Predictive Model for Pancreatoduodenectomy

Introduction: Existing postoperative pancreatic fistula (POPF) risk scores rely on intraoperative parameters, limiting their value in the preoperative setting. A preoperative predictive model to stratify the risk of developing POPF before pancreatoduodenectomy (PD) was built and externally validated.

The Construction of a Risk Prediction Model Based on Neural Network for Pre-operative Acute Ischemic Stroke in Acute Type A Aortic Dissection Patients.

Objective: To establish a pre-operative acute ischemic stroke risk (AIS) prediction model using the deep neural network in patients with acute type A aortic dissection (ATAAD).Methods: Between January 2015 and February 2019, 300 ATAAD patients diagnosed by aorta CTA were analyzed retrospectively. Patients were divided into two groups according to the presence or absence of pre-operative AIS. Pre-operative AIS risk prediction models based on different machine learning algorithm was established with clinical, transthoracic echocardiography (TTE) and CTA imaging characteristics as input. The performance of the difference models was evaluated using the receiver operating characteristic (ROC), precision-recall curve (PRC) and decision curve analysis (DCA).Results: Pre-operative AIS was detected in 86 of 300 patients with ATAAD (28.7%). The cohort was split into a training (211, 70% patients) and validation cohort (89, 30% patients) according to stratified sampling strategy. The constructed deep neural network model had the best performance on the discrimination of AIS group compare with other machine learning model, with an accuracy of 0.934 (95% CI: 0.891–0.963), 0.921 (95% CI: 0.845–0.968), sensitivity of 0.934, 0.960, specificity of 0.933, 0.906, and AUC of 0.982 (95% CI: 0.967–0.997), 0.964 (95% CI: 0.932–0.997) in the training and validation cohort, respectively.Conclusion: The established risk prediction model based on the deep neural network method may have the big potential to evaluate the risk of pre-operative AIS in patients with ATAAD.

Development and validation of a prediction model for malignant pulmonary nodules: A cohort study.

This study is to develop and validate a preoperative prediction model for malignancy of solitary pulmonary nodules. Data from 409 patients who underwent solitary pulmonary nodule resection at the First Affiliated Hospital of Nanjing Medical University, China between June 2018 and December 2020 were retrospectively collected. Then, the patients were nonrandomly split into a training cohort and a validation cohort. Clinical features, imaging parameters and laboratory data were then collected. Logistic regression analysis was used to develop a prediction model to identify variables significantly associated with malignant pulmonary nodules (MPNs) that were then included in the nomogram. We evaluated the discrimination and calibration ability of the nomogram by concordance index and calibration plot, respectively. MPNs were confirmed in 215 (52.6%) patients by a pathological examination. Multivariate logistic regression analysis identified 6 risk factors independently associated with MPN: gender (female, odds ratio [OR] = 2.487; 95% confidence interval [CI]: 1.313–4.711; P = .005), location of nodule (upper lobe of lung, OR = 1.126; 95%CI: 1.054–1.204; P < .001), density of nodule (pure ground glass, OR = 4.899; 95%CI: 2.572–9.716; P < .001; part-solid nodules, OR = 6.096; 95%CI: 3.153–14.186; P < .001), nodule size (OR = 1.193; 95%CI: 1.107–1.290; P < .001), GAGE7 (OR = 1.954; 95%CI: 1.054–3.624; P = .033), and GBU4–5 (OR = 2.576; 95%CI: 1.380–4.806; P = .003). The concordance index was 0.86 (95%CI: 0.83–0.91) and 0.88 (95%CI: 0.84–0.94) in the training and validation cohorts, respectively. The calibration curves showed good agreement between the predicted risk by the nomogram and real outcomes. We have developed and validated a preoperative prediction model for MPNs. The model could aid physicians in clinical treatment decision making.

Preoperative models incorporating the systemic immune-inflammation index for predicting prognosis and muscle invasion in patients with non-metastatic upper tract urothelial carcinoma.

To develop preoperative models as a guide to indications for neoadjuvant chemotherapy (NAC) and regional lymph node dissection (LND) before and at radical nephroureterectomy (RNU), respectively, in patients with non-metastatic upper tract urothelial carcinoma (UTUC) by incorporating the systemic immune-inflammation index (SII). This retrospective study enrolled 103 consecutive patients with UTUC undergoing RNU. The SII was calculated as neutrophils × platelets / lymphocytes. Multivariable Cox proportional hazard model was used to develop preoperative models for cancer-specific survival (CSS) and overall survival (OS). A model for predicting muscle invasion was developed using logistic regression analysis. Harrell's concordance-index (c-index) or the area under the receiver operating characteristic curve (AUC) was used to evaluate the accuracy of the models. During follow-up (median: 41months), 26 and three patients died of UTUC and other causes, respectively. Performance status > 0, clinical tumor (cT) stage ≥ 3, and SII > 520 were independent adverse prognosticators for CSS, and one point was assigned to each prognosticator. Risk score models comprising the sum of the points stratified patients into three risk groups (0, 1, and 2-3; P < 0.001 for CSS and OS) with respective c-indices of 0.843 and 0.820. SII > 677 and ≥ cT3 were independently associated with muscle invasion. A model based on these variables predicted muscle invasion with AUC of 0.804. Preoperative SII is significantly associated with worse survival outcomes and muscle invasion in patients with non-metastatic UTUC. Our preoperative predictive models may serve as a guide to indications for NAC and LND.

Multicenter Validation of Individual Preoperative Motor Outcome Prediction for Deep Brain Stimulation in Parkinson’s Disease

Background: Subthalamic nucleus deep brain stimulation (STN DBS) is an established therapy for Parkinson’s disease (PD) patients suffering from motor response fluctuations despite optimal medical treatment, or severe dopaminergic side effects. Despite careful clinical selection and surgical procedures, some patients do not benefit from STN DBS. Preoperative prediction models are suggested to better predict individual motor response after STN DBS. We validate a preregistered model, DBS-PREDICT, in an external multicenter validation cohort. Methods: DBS-PREDICT considered eleven, solely preoperative, clinical characteristics and applied a logistic regression to differentiate between weak and strong motor responders. Weak motor response was defined as no clinically relevant improvement on the Unified Parkinson’s Disease Rating Scale (UPDRS) II, III, or IV, 1 year after surgery, defined as, respectively, 3, 5, and 3 points or more. Lower UPDRS III and IV scores and higher age at disease onset contributed most to weak response predictions. Individual predictions were compared with actual clinical outcomes. Results: 322 PD patients treated with STN DBS from 6 different centers were included. DBS-PREDICT differentiated between weak and strong motor responders with an area under the receiver operator curve of 0.76 and an accuracy up to 77%. Conclusion: Proving generalizability and feasibility of preoperative STN DBS outcome prediction in an external multicenter cohort is an important step in creating clinical impact in DBS with data-driven tools. Future prospective studies are required to overcome several inherent practical and statistical limitations of including clinical decision support systems in DBS care.

Development and Internal Validation of a Preoperative Prediction Model for Sentinel Lymph Node Status in Breast Cancer: Combining Radiomics Signature and Clinical Factors.

PurposeTo develop and internally validate a nomogram combining radiomics signature of primary tumor and fibroglandular tissue (FGT) based on pharmacokinetic dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) and clinical factors for preoperative prediction of sentinel lymph node (SLN) status in breast cancer patients.MethodsThis study retrospectively enrolled 186 breast cancer patients who underwent pretreatment pharmacokinetic DCE-MRI with positive (n = 93) and negative (n = 93) SLN. Logistic regression models and radiomics signatures of tumor and FGT were constructed after feature extraction and selection. The radiomics signatures were further combined with independent predictors of clinical factors for constructing a combined model. Prediction performance was assessed by receiver operating characteristic (ROC), calibration, and decision curve analysis. The areas under the ROC curve (AUCs) of models were corrected by 1,000-times bootstrapping method and compared by Delong’s test. The added value of each independent model or their combinations was also assessed by net reclassification improvement (NRI) and integrated discrimination improvement (IDI) indices. This report referred to the “Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis” (TRIPOD) statement.ResultsThe AUCs of the tumor radiomic model (eight features) and the FGT radiomic model (three features) were 0.783 (95% confidence interval [CI], 0.717–0.849) and 0.680 (95% CI, 0.604–0.757), respectively. A higher AUC of 0.799 (95% CI, 0.737–0.862) was obtained by combining tumor and FGT radiomics signatures. By further combining tumor and FGT radiomics signatures with progesterone receptor (PR) status, a nomogram was developed and showed better discriminative ability for SLN status [AUC 0.839 (95% CI, 0.783–0.895)]. The IDI and NRI indices also showed significant improvement when combining tumor, FGT, and PR compared with each independent model or a combination of any two of them (all p < 0.05).ConclusionFGT and clinical factors improved the prediction performance of SLN status in breast cancer. A nomogram integrating the DCE-MRI radiomics signature of tumor and FGT and PR expression achieved good performance for the prediction of SLN status, which provides a potential biomarker for clinical treatment decision-making.

Prediction of post-traumatic neuropathy following impacted mandibular third molar removal

ObjectivesThe extraction of impacted mandibular third molars is a common surgical procedure often associated with complications including post-traumatic neuropathy. Previous work has focused on identifying confounding factors, but a robust preoperative risk prediction model remains elusive. MethodsUsing a dataset of 648 patients and 812 impacted mandibular third molars, we used least absolute shrinkage and selection operator (LASSO) to fit prediction models based on risk factors assessed at both the tooth and patient levels. In addition, we fitted multivariable logistic regression models with the Firth correction for generalized estimating equations (GEE). ResultsThe LASSO model for post-traumatic neuropathy identified distoangular impaction of ≥ 45° (odds ratio [OR] = 2.9), proximity to the inferior alveolar nerve of ≤ 3 mm (OR = 1.9), disadvantageous curving (OR = 1.4), and psychiatric conditions (OR = 2.1) as predictors [area under the receiving operator characteristic curve (AUC) = 0.75]. Among other complications analyzed, the LASSO model for bleeding identified deep embedding or full impaction (OR = 1.8), psychiatric conditions (OR = 1.3), and age (OR = 0.9) as predictors (AUC = 0.64). These associations between predictors and postoperative complications were fundamentally reinforced by the corresponding GEE models. ConclusionsOur findings point to the predictability of post-traumatic neuropathy and bleeding based on tooth anatomy and patient characteristics, overall suggesting that preoperatively identifiable factors can predict the risk of adverse outcomes in the extraction of impacted mandibular third molars. Clinical significanceMandibular third molar extraction is both a routine procedure and a leading cause of trigeminal neuropathy. Prevention of post-traumatic neuropathy, aided by individualized preoperative risk prediction, is of high clinical relevance.

Somatic mutations as preoperative predictors of metastases in patients with localized clear cell renal cell carcinoma – An exploratory analysis

ObjectiveRecurrent genomic alterations in clear cell renal cell carcinoma (ccRCC) have been associated with treatment outcomes; however, current preoperative predictive models do not include known genetic predictors.We aimed to explore the value of common somatic mutations in the preoperative prediction of metastatic disease among patients treated for localized ccRCC. Materials and methodsAfter obtaining institutional review board approval, data of 254 patients with localized ccRCC treated between 2005 and 2015 who underwent genetic sequencing was collected. The mutation status of VHL, PBRM1, SETD2, BAP1 and KDM5C were evaluated in the nephrectomy tumor specimen, which served as a proxy for biopsy mutation status. The Raj et al. preoperative nomogram was used to predict the 12-year metastatic free probability (MFP).The study outcome was MFP; the relationship between MFP and mutation status was evaluated with Cox-regression models adjusting for the preoperative nomogram variables (age, gender, incidental presentation, lymphadenopathy, necrosis, and size). ResultsThe study cohort included 188 males (74%) and 66 females (26%) with a median age of 58 years. VHL mutations were present in 152/254 patients (60%), PBRM1 in 91/254 (36%), SETD2 in 32/254 (13%), BAP1 in 19/254 (8%), and KDM5C in 19/254 (8%).Median follow-up for survivors was 8.1 years. Estimated 12-year MFP was 70% (95% CI: 63%–75%). On univariable analysis SETD2 (HR: 3.30), BAP1 (HR: 2.44) and PBRM1 (HR: 1.78) were significantly associated with a higher risk of metastases. After adjusting for known preoperative predictors in the existing nomogram, SETD2 mutations remained associated with a higher rate of metastases after nephrectomy (HR: 2.09, 95% CI: 1.19–3.67, P = 0.011). ConclusionIn the current exploratory analysis, SETD2 mutations were significant predictors of MFP among patients treated for localized ccRCC. Our findings support future studies evaluating genetic alterations in preoperative renal biopsy samples as potential predictors of treatment outcome.

Construction of a novel predictive nomogram for difficult procedure of endoscopic submucosal dissection for colorectal neoplasms

Objectives To determine the predictors of difficult colorectal endoscopic submucosal dissection (ESD) and to develop a preoperative predictive model for difficult colorectal ESD procedures. Methods Colorectal neoplasms intended to be resected by ESD in our center between August 2013 and February 2019 were retrospectively enrolled. An ESD procedure which took more than 30 min, failed to remove the lesions en bloc or converted to surgery was defined as difficulty. Logistic regression analysis was conducted to find out the predictors of difficult ESD. A nomogram integrating independent predictors was developed and validated with respect to its discrimination, calibration and clinical application, using the receiver operating characteristic (ROC) curve, calibration plot and decision curve analysis (DCA), respectively. Results A total of 368 colorectal neoplasms in 355 patients were included. The independent predictors for difficult colorectal ESD were size ≥2 cm (odds ratio [OR] = 6.102, p < .001), positive non-lifting sign (OR = 6.569, p = .005), lesions located in left colon (OR = 2.475, p = .036) or rectum (OR = 2.183, p = .048), laterally spreading tumors (LSTs) (OR = 2.501, p = .008) and less colorectal ESD experience (≤20 cases) (OR = 2.3091, p = .028). The nomogram model incorporating the above predictors performed well in both of the training and validation sets (area under the cure [AUC] = 0.786 and 0.784, respectively). DCA demonstrated the clinical benefit of the nomogram was superior to that of each independent predictor alone. Conclusions The nomogram incorporating tumor size, location, morphology, non-lifting sign and ESD experience of operator can be conveniently used to facilitate the preoperative prediction of difficult colorectal ESD.

Preoperative Prediction of Microvascular Invasion in Patients With Hepatocellular Carcinoma Based on Radiomics Nomogram Using Contrast-Enhanced Ultrasound.

PurposeThis study aimed to develop a radiomics nomogram based on contrast-enhanced ultrasound (CEUS) for preoperatively assessing microvascular invasion (MVI) in hepatocellular carcinoma (HCC) patients.MethodsA retrospective dataset of 313 HCC patients who underwent CEUS between September 20, 2016 and March 20, 2020 was enrolled in our study. The study population was randomly grouped as a primary dataset of 192 patients and a validation dataset of 121 patients. Radiomics features were extracted from the B-mode (BM), artery phase (AP), portal venous phase (PVP), and delay phase (DP) images of preoperatively acquired CEUS of each patient. After feature selection, the BM, AP, PVP, and DP radiomics scores (Rad-score) were constructed from the primary dataset. The four radiomics scores and clinical factors were used for multivariate logistic regression analysis, and a radiomics nomogram was then developed. We also built a preoperative clinical prediction model for comparison. The performance of the radiomics nomogram was evaluated via calibration, discrimination, and clinical usefulness.ResultsMultivariate analysis indicated that the PVP and DP Rad-score, tumor size, and AFP (alpha-fetoprotein) level were independent risk predictors associated with MVI. The radiomics nomogram incorporating these four predictors revealed a superior discrimination to the clinical model (based on tumor size and AFP level) in the primary dataset (AUC: 0.849 vs. 0.690; p < 0.001) and validation dataset (AUC: 0.788 vs. 0.661; p = 0.008), with a good calibration. Decision curve analysis also confirmed that the radiomics nomogram was clinically useful. Furthermore, the significant improvement of net reclassification index (NRI) and integrated discriminatory improvement (IDI) implied that the PVP and DP radiomics signatures may be very useful biomarkers for MVI prediction in HCC.ConclusionThe CEUS-based radiomics nomogram showed a favorable predictive value for the preoperative identification of MVI in HCC patients and could guide a more appropriate surgical planning.

Gd-EOB-DTPA-enhanced MRI radiomics to predict vessels encapsulating tumor clusters (VETC) and patient prognosis in hepatocellular carcinoma.

The study was to develop a Gd-EOB-DTPA-enhanced MRI radiomics model for preoperative prediction of VETC and patient prognosis in hepatocellular cancer (HCC). The study included 182 (training cohort: 128; validation cohort: 54) HCC patients who underwent preoperative Gd-EOB-DTPA-enhanced MRI. Volumes of interest including intratumoral and peritumoral regions were manually delineated in the hepatobiliary phase images, from which 1316 radiomics features were extracted. The least absolute shrinkage and selection operator (LASSO) and multivariable logistic regression were used to select the useful features. Clinical, intratumoral, peritumoral, combined radiomics, and clinical radiomics models were established using machine learning algorithms. The Kaplan-Meier survival analysis was used to assess early recurrence and progression-free survival (PFS) in the VETC + and VETC- patients. In the validation cohort, the area under the curves (AUCs) of radiomics models were higher than that of the clinical model using random forest (all p < 0.05). The peritumoral radiomics model (AUC = 0.972;95% confidence interval [CI]:0.887-0.998) had significantly higher AUC than intratumoral model (AUC = 0.919; 95% CI: 0.811-0.976) (p = 0.044). There were no significant differences in AUC between intratumoral or peritumoral radiomics model (PR) and combined radiomics model (p > 0.05). Early recurrence and PFS were significantly different between the PR-predicted VETC + and VETC- HCC patients (p < 0.05). PR-predicted VETC was independent predictor of early recurrence (hazard ratio [HR]: 2.08[1.31-3.28]; p = 0.002) and PFS (HR: 1.95[1.20-3.17]; p = 0.007). The intratumoral or peritumoral radiomics model may be useful in predicting VETC and patient prognosis preoperatively. The peritumoral radiomics model may yield an incremental value over intratumoral model. • Radiomics models are useful for predicting vessels encapsulating tumor clusters (VETC) and patient prognosis preoperatively. • Peritumoral radiomics model may yield an incremental value over intratumoral model in prediction of VETC. • Peritumoral radiomics-model-predicted VETC was an independent predictor of early recurrence and progression-free survival.

Diagnostic nomogram model for predicting preoperative pathological grade of meningioma.

BackgroundMeningioma is the most common primary tumor of the central nervous system. Preoperative diagnosis of high-grade meningioma is helpful for the selection of treatment options. The aim of our study is to establish a diagnostic nomogram model for preoperative prediction of the pathological grade of meningioma.MethodsThe predictive model was established from a cohort of 215 clinicopathologically confirmed meningioma between January 2012 and December 2017. Radiomic features were collected from preoperative magnetic resonance imaging (MRI) and computed tomography of patients with meningioma. The least absolute shrinkage and selection operator (LASSO) regression model was used for data dimension reduction and feature selection. Multivariate logistic regression was used to build a predictive model and presented as a nomogram. The performance of the nomogram was assessed with respect to its calibration, discrimination, and clinical usefulness. Internal validation was evaluated using bootstrapping validation.ResultsHigh-grade meningioma was observed in 47 patients (22%). The predictors included in the nomogram were tumor-brain interface, bone invasion, and tumor location. The final diagnostic model exhibited good calibration and discrimination with a C-index of 0.874 (95% confidence interval: 0.818–0.929) and a higher C-index of 0.868 in internal validation. Decision curve analysis (DCA) indicated that the nomogram is very useful in clinical practice.ConclusionsThis study provides a nomogram model with tumor-brain interface, bone invasion, and tumor location that can effectively predict the preoperative pathological grading of patients with meningioma and thus help clinicians provide more reasonable treatment strategies for meningioma patients.

Deep convolutional neural network for preoperative prediction of microvascular invasion and clinical outcomes in patients with HCCs.

We aimed to develop and validate a deep convolutional neural network (DCNN) model for preoperative prediction of microvascular invasion (MVI) in hepatocellular carcinoma (HCC) and its clinical outcomes using contrast-enhanced computed tomography (CECT) in a large population of candidates for surgery. This retrospective study included 1116 patients with HCC who had undergone preoperative CECT and curative hepatectomy. Radiological (R), DCNN, and combined nomograms were constructed in a training cohort (n = 892) respectively based on clinicoradiological factors, DCNN probabilities, and all factors; the performance of each model was confirmed in a validation cohort (n = 244). Accuracy and the AUC to predict MVI were calculated. Disease-free survival (DFS) and overall survival (OS) after surgery were recorded. The proportion of MVI-positive patients was respectively 38.8% (346/892) and 35.7 % (87/244) in the training and validation cohorts. The AUCs of the R, DCNN, and combined nomograms were respectively 0.809, 0.929, and 0.940 in the training cohorts and 0.837, 0.865, and 0.897 in the validation cohort. The combined nomogram outperformed the R nomogram in the training (p < 0.001) and validation (p = 0.009) cohorts. There was a significant difference in DFS and OS between the R, DCNN, and combined nomogram-predicted groups with and without MVI (p < 0.001). The combined nomogram based on preoperative CECT performs well for preoperative prediction of MVI and outcome. • A combined nomogram based on clinical information, preoperative CECT, and DCNN can predict MVI and clinical outcomes of patients with HCC. • DCNN provides added diagnostic ability to predict MVI. • The AUCs of the combined nomogram are 0.940 and 0.897 in the training and validation cohorts, respectively.

CT Radiomics for the Prediction of Synchronous Distant Metastasis in Clear Cell Renal Cell Carcinoma.

The aim of this study was to construct and verify a computed tomography (CT) radiomics model for preoperative prediction of synchronous distant metastasis (SDM) in clear cell renal cell carcinoma (ccRCC) patients. Overall, 172 patients with ccRCC were enrolled in the present research. Contrast-enhanced CT images were manually sketched, and 2994 quantitative radiomic features were extracted. The radiomic features were then normalized and subjected to hypothesis testing. Least absolute shrinkage and selection operator (LASSO) was applied to dimension reduction, feature selection, and model construction. The performance of the predictive model was validated through analysis of the receiver operating characteristic curve. Multivariate and subgroup analyses were performed to verify the radiomic score as an independent predictor of SDM. The patients randomized into a training (n = 104) and a validation (n = 68) cohort in a 6:4 ratio. Through dimension reduction using LASSO regression, 9 radiomic features were used for the construction of the SDM prediction model. The model yielded moderate performance in both the training (area under the curve, 0.89; 95% confidence interval, 0.81-0.97) and the validation cohort (area under the curve, 0.83; 95% confidence interval, 0.69-0.95). Multivariate analysis showed that the CT radiomic signature was an independent risk factor for clinical parameters of ccRCC. Subgroup analysis revealed a significant connection between the SDM and radiomic signature, except for the lower pole of the kidney subgroup. The CT-based radiomics model could be used as a noninvasive, personalized approach for SDM prediction in patients with ccRCC.

A nomogram for preoperative prediction of prolonged air leak after pulmonary malignancy resection.

BackgroundProlonged air leak (PAL) is one of the most common postoperative complications after lung surgery. This study aimed to identify risk factors of PAL after lung resection and develop a preoperative predictive model to estimate its risk for individual patients.MethodsPatients with pulmonary malignancies or metastasis who underwent pulmonary resection between January 2014 and January 2018 were included. PAL was defined as an air leak more than 5 days after surgery, risk factors were analyzed. Forward stepwise multivariable logistic regression analysis was performed to identify independent risk factors, and a derived nomogram was built. Data from February 2018 to September 2018 were collected for internal validation.ResultsA total of 1,511 patients who met study criteria were enrolled in this study. The overall incidence of PAL was 9.07% (137/1,511). Age, percent forced expiratory volume in 1 second, surgical type, surgical approach and smoking history were included in the final model. A nomogram was developed according to the multivariable logistic regression results. The C-index of the predictive model was 0.70, and the internal validation value was 0.77. The goodness-of-fit test was non-significant for model development and internal validation.ConclusionsThe predictive model and derived nomogram achieved satisfied preoperative prediction of PAL. Using this nomogram, the risk for an individual patient can be estimated, and preventive measures can be applied to high-risk patients.

Preoperative risk stratification of postoperative pancreatic fistula: A risk-tree predictive model for pancreatoduodenectomy

BackgroundExisting postoperative pancreatic fistula risk scores rely on intraoperative parameters, which limits their value in the preoperative setting. A preoperative predictive model to stratify the risk of developing postoperative pancreatic fistula before pancreatoduodenectomy was built and externally validated. MethodsA regression risk-tree model for preoperative postoperative pancreatic fistula risk stratification was developed in the Verona University Hospital training cohort using preoperative variables and then tested prospectively in a validation cohort of patients who underwent pancreatoduodenectomy at San Raffaele Hospital of Milan. ResultsIn the study period 566 (training cohort) and 456 (validation cohort) patients underwent pancreatoduodenectomy. In the multivariable analysis body mass index, radiographic main pancreatic duct diameter and American Society of Anesthesiologists score ≥3 were independently associated with postoperative pancreatic fistula. The regression tree analysis allocated patients into 3 preoperative risk groups with an 8%, 21%, and 32% risk of postoperative pancreatic fistula (all P < .01) based on main pancreatic duct diameter (≥ or <5 mm) and body mass index (≥ or <25). The 3 groups were labeled low, intermediate, and high risk and consisted of 206 (37%), 188 (33%), and 172 (30%) patients, respectively. The risk-tree was applied to validation cohort, successfully reproducing 3 risk groups with significantly different postoperative pancreatic fistula risks (all P < .01). ConclusionIn candidates for pancreatoduodenectomy, the risk of postoperative pancreatic fistula can be quickly and accurately determined in the preoperative setting based on the body mass index and main pancreatic duct diameter at radiology. Preoperative risk stratification could potentially guide clinical decision-making, improve patient counseling and allow the establishment of personalized preoperative protocols.

A clinical-radiomics model incorporating T2-weighted and diffusion-weighted magnetic resonance images predicts the existence of lymphovascular invasion / perineural invasion in patients with colorectal cancer.

Lymphovascular invasion (LVI) and perineural invasion (PNI) are independent prognostic factors in patients with colorectal cancer (CRC). In this study, we aimed to develop and validate a preoperative predictive model based on high-throughput radiomic features and clinical factors for accurate prediction of LVI/PNI in these patients. Two hundred and sixty-three patients who underwent colorectal resection for histologically confirmed CRC between 1 February 2011 and 30 June 2020 were retrospectively enrolled. Between 1 February 2011 and 30 September 2018, 213 patients were randomly divided into a training cohort (n=149) and a validation cohort (n=64) by a ratio of 7:3. We used a 10000-iteration bootstrap analysis to estimate the prediction error and confidence interval for two cohorts. The independent test cohort consisted of 50 patients between 1 October 2018 and 30 June 2020. Regions of interest (ROIs) were manually delineated in high-resolution T2-weighted and diffusion-weighted images using ITK-SNAP software on each CRC tumor slice. In total, 3356 radiomic features were extracted from each ROI. Next, we used the maximum relevance minimum redundancy and least absolute shrinkage and selection operator algorithms to select the strongest of these features to establish a clinical-radiomics model for predicting LVI/PNI. Receiver-operating characteristic and calibration curves were then plotted to evaluate the predictive performance of the model in the training, validation, and independent test cohorts. A multiparametric clinical-radiomics model combining MRI-reported extramural vascular invasion (EMVI) status and a Radiomics score for the LVI/PNI estimation was established. This model had significant predictive power in the training cohort (area under the curve [AUC] 0.91; 95% confidence interval [CI]: 0.85-0.97), validation cohort (AUC: 0.88; 95% CI: 0.79-89), and independent test cohorts (AUC 0.83, 95% CI 0.72-0.95). The model performed well in the independent test cohort with sensitivity of 0.818, specificity of 0.714, and accuracy of 0.760. Calibration curve and decision curve analysis demonstrated clinical benefits. Multiparametric clinical-radiomics models can accurately predict LVI/PNI in patients with CRC. Our model has predictive ability that should improve preoperative diagnostic performance and allow more individualized treatment decisions.

A Study of Predictive Models for Early Outcomes of Post-Prostatectomy Incontinence: Machine Learning Approach vs. Logistic Regression Analysis Approach

Background: Post-prostatectomy incontinence (PPI) is a major complication that can significantly decrease quality of life. Approximately 20% of patients experience consistent PPI as long as 1 year after radical prostatectomy (RP). This study develops a preoperative predictive model and compares its diagnostic performance with conventional tools. Methods: A total of 166 prostate cancer patients who underwent magnetic resonance imaging (MRI) and RP were evaluated. According to the date of the RP, patients were divided into a development cohort (n = 109) and a test cohort (n = 57). Patients were classified as PPI early-recovery or consistent on the basis of pad usage for incontinence at 3 months after RP. Uni- and multi-variable logistic regression analyses were performed to identify associates of PPI early recovery. Four well-known machine learning algorithms (k-nearest neighbor, decision tree, support-vector machine (SVM), and random forest) and a logistic regression model were used to build prediction models for recovery from PPI using preoperative clinical and imaging data. The performances of the prediction models were assessed internally and externally using sensitivity, specificity, accuracy, and area-under-the-curve values and estimated probabilities and the actual proportion of cases of recovery from PPI within 3 months were compared using a chi-squared test. Results: Clinical and imaging findings revealed that age (70.1 years old for the PPI early-recovery group vs. 72.8 years old for the PPI consistent group), membranous urethral length (MUL; 15.7 mm for the PPI early-recovery group vs. 13.9 mm for the PPI consistent group), and obturator internal muscle (18.2 mm for the PPI early-recovery group vs. 17.5 mm for the PPI consistent group) were significantly different between the PPI early-recovery and consistent groups (all p-values &lt; 0.05). Multivariate analysis confirmed that age (odds ratio = 1.07, 95% confidence interval = 1.02–1.14, p-value = 0.007) and MUL (odds ratio = 0.87, 95% confidence interval = 0.80–0.95, p-value = 0.002) were significant independent factors for early recovery. The prediction model using machine learning algorithms showed superior diagnostic performance compared with conventional logistic regression (AUC = 0.59 ± 0.07), especially SVM (AUC = 0.65 ± 0.07). Moreover, all models showed good calibration between the estimated probability and actual observed proportion of cases of recovery from PPI within 3 months. Conclusions: Preoperative clinical data and anatomic features on preoperative MRI can be used to predict early recovery from PPI after RP, and machine learning algorithms provide greater diagnostic accuracy compared with conventional statistical approaches.

Preoperative Prediction of Cytokeratin 19 Expression for Hepatocellular Carcinoma with Deep Learning Radiomics Based on Gadoxetic Acid-Enhanced Magnetic Resonance Imaging.

PurposeCytokeratin 19 (CK19) expression is a proven independent prognostic predictor of hepatocellular carcinoma (HCC). This study aimed to develop and validate the performance of a deep learning radiomics (DLR) model for CK19 identification in HCC based on preoperative gadoxetic acid-enhanced magnetic resonance imaging (MRI).Patients and MethodsA total of 141 surgically confirmed HCCs with preoperative gadoxetic acid-enhanced MRI from two institutions were included. Prediction models were established based on hepatobiliary phase (HBP) images using a training set (n=102) and validated using time-independent (n=19) and external (n=20) test sets. A receiver operating characteristic curve was used to evaluate the performance for CK19 prediction. Recurrence-free survival (RFS) was also analyzed by incorporating the CK19 expression and other factors.ResultsFor predicting CK19 expression, the area under the curve (AUC) of the DLR model was 0.820 (95% confidence interval [CI]: 0.732–0.907, P<0.001) with sensitivity, specificity, accuracy of 0.800, 0.766, and 0.775, respectively, and reached 0.781 in the external test set. Combined with alpha fetoprotein, the AUC increased to 0.833 (95% CI: 0.753–0.912, P<0.001) and the sensitivity was 0.960. Intratumoral hemorrhage and peritumoral hypointensity on HBP were independent risk factors for HCC recurrence by multivariate analysis. Based on predicted CK19 expression and the independent risk factors, a nomogram was developed to predict RFS and achieved C-index of 0.707.ConclusionThis study successfully established and verified an optimal DLR model for preoperative prediction of CK19-positive HCCs based on gadoxetic acid-enhanced MRI. The prediction of CK19 expression in HCC using a non-invasive method can help inform preoperative planning.

A nomogram combined with radiomics features, albuminuria, and metabolic syndrome to predict the risk of myometrial invasion of bladder cancer.

BackgroundTo establish a preoperative prediction model of myometrial invasion of bladder cancer (BC) based on the radiomics characteristics of multi-parameter thin-slice enhanced computed tomography (CT) imaging.MethodsData from 100 patients with BC were analyzed retrospectively. The patients were divided into two groups: muscular invasive BC and non-muscular invasive BC. The tumor region was segmented from enhanced CT images (arterial- and venous-phase calibration maps) of all patients using Slicer-3D software. We extracted 1,223 texture features from tumor image data based on the shape and gray-level co-occurrence matrix, gray size region matrix, gray run-length matrix, adjacent gray difference matrix, and gray correlation matrix. The patients were randomly divided into a training group (n=70) and a verification group (n=30) in a 7:3 ratio. Interclass correlation coefficients >0.75, least absolute shrinkage, and selection operator regression were used for feature selection. The prediction model was established by combining Rad-score, independent clinical factors, and support vector machine (SVM), and a radiomics nomogram was constructed. The nomogram was tested using the consistency index, calibration curve, time-dependent receiver operating characteristic curve, and clinical decision curve to predict the myometrial invasion of the bladder preoperatively.ResultsSix radiomics features that were significantly related to myometrial invasion of BC were selected to construct a predictive model. The area under the curve (AUC) values of training group and verification group based on SVM were 0.898 (95% CI: 0.820–0.976) and 0.702 (95% CI: 0.495–0.909), respectively. Single factor and multiple factor analysis showed that albuminuria (95% CI: 0.243–2.206, P=0.0014) and metabolic syndrome (95% CI: 0.850–2.935, P<0.001) were independent influencing factors of BC myometrial invasion. Clinical factors and 11 radiomics features were used to construct a comprehensive model for predicting the pathological grade of BC (radiomics + clinical). After a comprehensive comparison, we found that the overall effectiveness of the model (radiomics + clinical) was the highest (AUC =0.8457).ConclusionsBased on the multi-parameter thin-layer enhanced CT radiomics feature can be used as a potential independent predictor of BC myometrial invasion, the model based on parameters can initially quantitatively characterize the risk of myometrial invasion, and has excellent potential for predicting myometrial invasion of BC.