Contrast-enhanced CT radiomics features to preoperatively identify differences between tumor and proximal tumor-adjacent and tumor-distant tissues of resectable esophageal squamous cell carcinoma.

BACKGROUND: Esophagectomy is the main treatment for esophageal squamous cell carcinoma (ESCC), and patients with histopathologically negative margins still have a relatively higher recurrence rate. Contrast-enhanced CT (CECT) radiomics might noninvasively obtain potential information about the internal heterogeneity of ESCC and its adjacent tissues. This study aimed to develop CECT radiomics models to preoperatively identify the differences between tumor and proximal tumor-adjacent and tumor-distant tissues in ESCC to potentially reduce tumor recurrence. METHODS: A total of 529 consecutive patients with ESCC from Centers A (n = 447) and B (n = 82) undergoing preoperative CECT were retrospectively enrolled in this study. Radiomics features of the tumor, proximal tumor-adjacent (PTA) and proximal tumor-distant (PTD) tissues were individually extracted by delineating the corresponding region of interest (ROI) on CECT and applying the 3D-Slicer radiomics module. Patients with pairwise tissues (ESCC vs. PTA, ESCC vs. PTD, and PTA vs. PTD) from Center A were randomly assigned to the training cohort (TC, n = 313) and internal validation cohort (IVC, n = 134). Univariate analysis and the least absolute shrinkage and selection operator were used to select the core radiomics features, and logistic regression was performed to develop radiomics models to differentiate individual pairwise tissues in TC, validated in IVC and the external validation cohort (EVC) from Center B. Diagnostic performance was assessed using area under the receiver operating characteristics curve (AUC) and accuracy. RESULTS: With the chosen 20, 19 and 5 core radiomics features in TC, 3 individual radiomics models were developed, which exhibited excellent ability to differentiate the tumor from PTA tissue (AUC: 0.965; accuracy: 0.965), the tumor from PTD tissue (AUC: 0.991; accuracy: 0.958), and PTA from PTD tissue (AUC: 0.870; accuracy: 0.848), respectively. In IVC and EVC, the models also showed good performance in differentiating the tumor from PTA tissue (AUCs: 0.956 and 0.962; accuracy: 0.956 and 0.937), the tumor from PTD tissue (AUCs: 0.990 and 0.974; accuracy: 0.952 and 0.970), and PTA from PTD tissue (AUCs: 0.806 and 0.786; accuracy: 0.760 and 0.786), respectively. CONCLUSION: CECT radiomics models could differentiate the tumor from PTA tissue, the tumor from PTD tissue, and PTA from PTD tissue in ESCC.

An improved model based on quantitative features of right liver lobe, maximum varices, and portal vein system measured on magnetic resonance imaging to predict oesophagogastric variceal haemorrhage secondary to hepatitis B-related cirrhosis.

Background: In patients with hepatitis B-related cirrhosis, it is important to predict those at high-risk of oesophagogastric variceal haemorrhage (OVH) to decide upon prophylactic treatment. Our published model developed with right liver lobe volume and diameters of portal vein system did not incorporate maximum variceal size as a factor. This study thus aimed to develop an improved model based on right liver lobe volume, diameters of maximum oesophagogastric varices (OV) and portal vein system obtained at magnetic resonance imaging (MRI) to predict OVH. Methods: Two hundred and thirty consecutive individuals with hepatitis B-related cirrhosis undergoing abdominal enhanced MRI were randomly grouped into training (n=160) and validation sets (n=70). OVH was confirmed in 51 and 23 participants in the training and validation sets during 2-year follow-up period, respectively. Spleen, total liver, right lobe, caudate lobe, left lateral lobe, and left medial lobe volumes, together with diameters of maximum OV and portal venous system were measured on MRI. In the training set, univariate analyses and binary logistic regression analyses were conducted to determine independent predictors. The performance of the model for predicting OVH constructed based on independent predictors from the training set was evaluated with receiver operating characteristic (ROC) analysis and validated in the validation set. Results: The model for predicting OVH was established based on right liver lobe volume and diameters of the maximum OV, left gastric vein, and portal vein [odds ratio (OR) =0.991, 2.462, 1.434, and 1.582, respectively; all P values <0.05]. The logistic regression model equation [-0.009 × right liver lobe volume + 0.901 × maximum OV diameter (MOVD) + 0.361 × left gastric vein diameter (LGVD) + 0.459 × portal vein diameter (PVD) - 7.842] with a cutoff value of -0.656 for predicting OVH obtained excellent performance with an area under ROC curve (AUC) of 0.924 [95% confidence interval (CI): 0.878-0.971]. The Delong test showed negative statistical difference in the model performance between the training and validation sets, with a P value >0.99. Conclusions: The model could help well screen those patients at high risk of OVH for timely intervention and avoiding the fatal complications.

A novel quantitative model based on gross tumor volume corresponding to anatomical distribution measured with multidetector computed tomography to determine the resectability of non­distant metastatic esophageal squamous cell carcinoma.

It is important to accurately determine the resectability of thoracic esophageal squamous cell carcinoma (ESCC) for treatment decision-making. Previous studies have revealed that the CT-derived gross tumor volume (GTV) is associated with the staging of ESCC. The present study aimed to explore whether the anatomical distribution-based GTV of non-distant metastatic thoracic ESCC measured using multidetector computed tomography (MDCT) could quantitatively determine the resectability. For this purpose, 473 consecutive patients with biopsy-confirmed non-distant metastatic thoracic ESCC who underwent contrast-enhanced CT were randomly divided into a training cohort (TC; 376 patients) and validation cohort (VC; 97 patients). GTV was retrospectively measured using MDCT. Univariate and multivariate analyses were performed to identify the determinants of the resectability of ESCC in the TC. Receiver operating characteristic (ROC) analysis was performed to clarify whether anatomical distribution-based GTV could help quantitatively determinate resectability. Unweighted Cohen's Kappa tests in VC were used to assess the performance of the previous models. Univariate analysis demonstrated that sex, anatomic distribution, cT stage, cN stage and GTV were related to the resectability of ESCC in the TC (all P<0.05). Multivariate analysis revealed that GTV [P<0.001; odds ratio (OR) 1.158] and anatomic distribution (P=0.027; OR, 1.924) were independent determinants of resectability. ROC analysis revealed that the GTV cut-offs for the determination of the resectability of the upper, middle and lower thoracic portions were 23.57, 22.89 and 22.58 cm3, respectively, with areas under the ROC curves of >0.9. Unweighted Cohen's Kappa tests revealed an excellent performance of the ROC models in the upper, middle and lower thoracic portions with Cohen k-values of 0.913, 0.879 and 0.871, respectively. On the whole, the present study demonstrated that GTV and the anatomic distribution of non-distant metastatic thoracic ESCC may be independent determinants of resectability, and anatomical distribution-based GTV can effectively be used to quantitatively determine resectability.

Whole-tumor histogram analysis of apparent diffusion coefficients for predicting lymphovascular space invasion in stage IB-IIA cervical cancer.

Objectives: To investigate the value of apparent diffusion coefficient (ADC) histogram analysis based on whole tumor volume for the preoperative prediction of lymphovascular space invasion (LVSI) in patients with stage IB-IIA cervical cancer. Methods: Fifty consecutive patients with stage IB-IIA cervical cancer were stratified into LVSI-positive (n = 24) and LVSI-negative (n = 26) groups according to the postoperative pathology. All patients underwent pelvic 3.0T diffusion-weighted imaging with b-values of 50 and 800 s/mm2 preoperatively. Whole-tumor ADC histogram analysis was performed. Differences in the clinical characteristics, conventional magnetic resonance imaging (MRI) features, and ADC histogram parameters between the two groups were analyzed. Receiver operating characteristic (ROC) analysis was used to evaluate the diagnostic performance of ADC histogram parameters in predicting LVSI. Results: ADCmax, ADCrange, ADC90, ADC95, and ADC99 were significantly lower in the LVSI-positive group than in the LVSI-negative group (all P-values < 0.05), whereas no significant differences were reported for the remaining ADC parameters, clinical characteristics, and conventional MRI features between the groups (all P-values > 0.05). For predicting LVSI in stage IB-IIA cervical cancer, a cutoff ADCmax of 1.75×10-3 mm2/s achieved the largest area under ROC curve (Az) of 0.750, followed by a cutoff ADCrange of 1.36×10-3 mm2/s and ADC99 of 1.75×10-3 mm2/s (Az = 0.748 and 0.729, respectively), and the cutoff ADC90 and ADC95 achieved an Az of <0.70. Conclusion: Whole-tumor ADC histogram analysis has potential value for preoperative prediction of LVSI in patients with stage IB-IIA cervical cancer. ADCmax, ADCrange, and ADC99 are promising prediction parameters.

A novel model based on liver/spleen volumes and portal vein diameter on MRI to predict variceal bleeding in HBV cirrhosis.

OBJECTIVE: To develop a novel logistic regression model based on liver/spleen volumes and portal vein diameter measured on magnetic resonance imaging (MRI) for predicting oesophagogastric variceal bleeding (OVB) secondary to HBV cirrhosis. METHODS: One hundred eighty-five consecutive cirrhotic patients with hepatitis B undergoing abdominal contrast-enhanced MRI were randomly divided into training cohort (n = 130) and validation cohort (n = 55). Spleen volume, total liver volume, four liver lobe volumes, and diameters of portal venous system were measured on MRI. Ratios of spleen volume to total liver and to individual liver lobe volumes were calculated. In training cohort, univariate analyses and binary logistic regression analyses were to determine independent predictors. Performance of the model for predicting OVB constructed based on independent predictors from training cohort was evaluated by receiver operating characteristic (ROC) analysis, and was validated by Kappa test in validation cohort. RESULTS: OVB occurred in 42 and 18 individuals in training and validation cohorts during the 2 years' follow-up, respectively. An OVB prediction model was constructed based on the independent predictors including right liver lobe volume (RV), left gastric vein diameter (LGVD) and portal vein diameter (PVD) (odds ratio = 0.993, 2.202 and 1.613, respectively; p-values < 0.001 for all). The logistic regression model equation (-0.007 × RV + 0.79 × LGVD + 0.478 × PVD-6.73) for predicting OVB obtained excellent performance with an area under ROC curve of 0.907. The excellent performance was confirmed by Kappa test with K-value of 0.802 in validation cohort. CONCLUSION: The novel logistic regression model can be reliable for predicting OVB. KEY POINTS: • Patients with oesophagogastric variceal bleeding are mainly characterized by decreased right lobe volume, and increased spleen volume and diameters of portal vein system. • The right liver lobe volume, left gastric vein diameter and portal vein diameter are the independent predictors of oesophagogastric variceal bleeding. • The novel model developed based on the independent predictors performed well in predicting oesophagogastric variceal bleeding with an area under the receiver operating characteristic curve of 0.907.

Combinations of liver lobe and spleen volumes obtained on magnetic resonance imaging to predict esophagogastric variceal bleeding in hepatitis B-related cirrhotic patients: A prospective cohort study.

To evaluate whether combinations of liver lobe and spleen volumes obtained on magnetic resonance imaging (MRI) could predict esophagogastric variceal bleeding (EVB) in hepatitis B-related cirrhotic patients. Ninety-six consecutive patients with hepatitis B-related cirrhosis underwent upper abdominal contrast-enhanced MRI within 1 week after initial hospitalization, and grouped based on outcomes of EVB during the 2 years' follow-up after being discharged. Total liver volume (TLV), spleen volume (SV) and 4 liver lobe volumes including right lobe volume (RV), left medial lobe volume (LMV), left lateral lobe volume (LLV), and caudate lobe volume (CV) were measured on MRI. Percentages of individual liver lobe volumes in TLV (including RV/TLV, LMV/TLV, LLV/TLV, and CV/TLV), ratios of SV to individual liver lobe volumes (including SV/RV, SV/LMV, SV/LLV, and SV/CV), and SV/TLV were statistically analyzed to predict EVB. Patients with EVB had lower RV than without EVB (P value = .001), whereas no differences in LMV, LLV, CV, and TLV were found (P values >.05 for all). Among percentages of individual liver lobe volumes in TLV, RV/TLV was lower whereas LMV/TLV and LLV/TLV were greater in patients with EVB than without EVB (P values <.05 for all). SV, ratios of SV to individual liver lobe volumes, and SV/TLV in patients with EVB were larger than without EVB (P values <.05 for all). Among parameters with difference between patients with and without EVB, SV/RV could best predict EVB with an area under receiver operating characteristic curve of 0.84. SV/RV could best predict EVB in hepatitis B-related cirrhotic patients.

Apparent diffusion coefficient derived from diffusion-weighted imaging to differentiate between tumor, tumor-adjacent and tumor-distant tissues in resectable rectal adenocarcinoma.

PURPOSE: To evaluate feasibility of apparent diffusion coefficient (ADC) at different b-values to differentiate between tumor, tumor-adjacent and tumor-distant tissues in rectal adenocarcinoma (RA). MATERIALS AND METHODS: Seventy consecutive patients with RA undergoing preoperative diffusion-weighted imaging were retrospectively enrolled. ADCs of tumor, proximal tumor-adjacent tissue (PTA) and tumor-distant tissue (PTD), and distal tumor-adjacent tissue (DTA) and tumor-distant tissue (DTD) were calculated with b-values of 0 and 800 sec/mm2, 0 and 1000 sec/mm2, 0 and 1500 sec/mm2, and multiple b-values of 0, 50, 100, 800, 1000 and 1500 sec/mm2. Statistical analysis was performed to determine feasibility of ADC to differentiate between pairwise tissues. RESULTS: Mean ADC of tumor was lower than those of PTA, PTD, DTA and DTD; and mean ADCs of PTA and DTA were lower than those of PTD and DTD at all b-values, respectively (all P-values < 0.001). ADC cut-offs of 1.089 × 10-3 mm2/sec (b = 0, 1000 sec/mm2) or 1.215 × 10-3 mm2/sec (b = 0, 800 sec/mm2), and 1.142 × 10-3 mm2/sec (b = 0, 1000 sec/mm2) or 0.995 × 10-3 mm2/sec (b = 0, 1500 sec/mm2) achieved excellent performance in differentiating tumor from PTA or PTD, and tumor from DTA or DTD (area under receiver operating characteristic curves [AUCs]: 0.813 or 0.952, and 0.970 or 0.996), respectively. ADC cut-offs of 1.625 × 10-3 mm2/sec (b = 0, 800 sec/mm2), and 1.165 × 10-3 mm2/sec (b = 0, 1500 sec/mm2) could differentiate PTA from PTD, and DTA from DTD (AUCs: 0.709 and 0.673), respectively. CONCLUSION: ADC can help differentiate between tumor, tumor-adjacent and tumor-distant tissues in RA.

Tumor Stage-Based Gross Tumor Volume of Resectable Esophageal Squamous Cell Carcinoma Measured on CT: Association With Early Recurrence After Esophagectomy.

OBJECTIVE: To investigate relationship of tumor stage-based gross tumor volume (GTV) of esophageal squamous cell carcinoma (ESCC) measured on computed tomography (CT) with early recurrence (ER) after esophagectomy. MATERIALS AND METHODS: Two hundred and four consecutive patients with resectable ESCC including 159 patients enrolled in the training cohort (TC) and 45 patients in validation cohort (VC) underwent contrast-enhanced CT less than 2 weeks before esophagectomy. GTV was retrospectively measured by multiplying sums of all tumor areas by section thickness. For the TC, univariate and multivariate analyses were performed to determine factors associated with ER. Mann-Whitney U test was conducted to compare GTV in patients with and without ER. Receiver operating characteristic (ROC) analysis was performed to determine if tumor stage-based GTV could predict ER. For the VC, unweighted Cohen's Kappa tests were used to evaluate the performances of the previous ROC predictive models. RESULTS: ER occurred in 63 of 159 patients (39.6%) in the TC. According to the univariate analysis, histologic differentiation, cT stage, cN stage, and GTV were associated with ER after esophagectomy (all P-values < 0.05). Multivariate analysis revealed that cT stage and GTV were independent risk factors with hazard ratios of 3.382 [95% confidence interval (CI): 1.533-7.459] and 1.222 (95% CI: 1.125-1.327), respectively (all P-values < 0.05). Mann-Whitney U tests showed that GTV could help differentiate between ESCC with and without ER in stages cT1-4a, cT2, and cT3 (all P-values < 0.001), and the ROC analysis demonstrated the corresponding cutoffs of 13.31, 17.22, and 17.83 cm3 with areas under the curve of more than 0.8, respectively. In the VC, the Kappa tests validated that the ROC predictive models had good performances for differentiating between ESCC with and without ER in stages cT1-4a, cT2, and cT3 with Cohen k of 0.696 (95% CI, 0.498-0.894), 0.733 (95% CI, 0.386-1.080), and 0.862 (95% CI, 0.603-1.121), respectively. CONCLUSION: GTV and cT stage can be independent risk factors of ER in ESCC after esophagectomy, and tumor stage-based GTV measured on CT can help predict ER.

Application of contrast-enhanced CT radiomics in prediction of early recurrence of locally advanced oesophageal squamous cell carcinoma after trimodal therapy.

BACKGROUND: Early recurrence of oesophageal squamous cell carcinoma (SCC) is defined as recurrence after surgery within 1 year, and appears as local recurrence, distant recurrence, and lymph node positive and disseminated recurrence. Contrast-enhanced computed tomography (CECT) is recommended for diagnosis of primary tumor and initial staging of oesophageal SCC, but it cannot be used to predict early recurrence. It is reported that radiomics can help predict preoperative stages of oesophageal SCC, lymph node metastasis before operation, and 3-year overall survival of oesophageal SCC patients following chemoradiotherapy by extracting high-throughput quantitative features from CT images. This study aimed to develop models based on CT radiomics and clinical features of oesophageal SCC to predict early recurrence of locally advanced cancer. METHODS: We collected electronic medical records and image data of 197 patients with confirmed locally advanced oesophageal SCC. These patients were randomly allocated to 137 patients in the training cohort and 60 in the test cohort. 352 radiomics features were extracted by delineating region-of-interest (ROI) around the lesion on CECT images and clinical signature was generated by medical records. The radiomics model, clinical model, the combined model of radiomics and clinical features were developed by radiomics features and/or clinical characteristics. Predicting performance of the three models was assessed with area under receiver operating characteristic curve (AUC), accuracy and F-1 score. RESULTS: Eleven radiomics features and/or six clinical signatures were selected to build prediction models related to recurrence of locally advanced oesophageal SCC after trimodal therapy. The AUC of integration of radiomics and clinical models was better than that of radiomics or clinical model for the training cohort (0.821 versus 0.754 or 0.679, respectively) and for the validation cohort (0.809 versus 0.646 or 0.658, respectively). Integrated model of radiomics and clinical features showed good performance in predicting early recurrence of locally advanced oesophageal SCC for both the training and validation cohorts (accuracy = 0.730 and 0.733, and F-1score = 0.730 and 0.778, respectively). CONCLUSIONS: The integrated model of CECT radiomics and clinical features may be a potential imaging biomarker to predict early recurrence of locally advanced oesophageal SCC after trimodal therapy.

CT radiomics features to predict lymph node metastasis in advanced esophageal squamous cell carcinoma and to discriminate between regional and non-regional lymph node metastasis: a case control study.

BACKGROUND: Prediction of lymph node status in esophageal squamous cell carcinoma (ESCC) is critical for clinical decision making. In clinical practice, computed tomography (CT) has been frequently used to assist in the preoperative staging of ESCC. Texture analysis can provide more information to reflect potential biological heterogeneity based on CT. A nomogram for the preoperative diagnosis of lymph node metastasis in patients with resectable ESCC has been previously developed. However, to the best of our knowledge, no reports focus on developing CT radiomics features to discriminate ESCC patients with regional lymph node metastasis (RLNM) and non-regional lymph node metastasis (NRLNM). We, therefore, aimed to develop CT radiomics models to predict lymph node metastasis (LNM) in advanced ESCC and to discriminate ESCC between RLNM and NRLNM. METHODS: This study enrolled 334 patients with pathologically confirmed advanced ESCC, including 152 patients without LNM and 182 patients with LNM, and 103 patients with RLNM and 79 patients NRLNM. Radiomics features were extracted from CT data for each patient. The least absolute shrinkage and selection operator (LASSO) model and independent samples t-tests or Mann-Whitney U tests were exploited for dimension reduction and selection of radiomics features. Optimal radiomics features were chosen using multivariable logistic regression analysis. The discriminating performance was assessed by area under the receiver operating characteristic curve (AUC) and accuracy. RESULTS: The radiomics features were developed based on multivariable logistic regression and were significantly associated with LNM status in both the training and validation cohorts (P<0.001). The radiomics models could differentiate between patients with and without LNM (AUC =0.79 and 0.75, and accuracy =0.75 and 0.71 in the training and validation cohorts, respectively). In patients with LNM, the radiomics features could effectively differentiate between RLNM and NRLNM (AUC =0.98 and 0.95, and accuracy =0.94 and 0.83 in the training and validation cohorts, respectively). CONCLUSIONS: CT radiomics features could help predict the LNM status of advanced ESCC patients and effectively discriminate ESCC between RLNM and NRLNM.