Preoperative Prediction of Ki-67 Labeling Index By Three-dimensional CT Image Parameters for Differential Diagnosis Of Ground-Glass Opacity (GGO)

Sub-regional CT Radiomics for the Prediction of Ki-67 Proliferation Index in Gastrointestinal Stromal Tumors: A Multi-center Study

RATIONALE AND OBJECTIVES

The objective was to assess and examine radiomics models derived from contrast-enhanced CT for their predictive capacity using the sub-regional radiomics regarding the Ki-67 proliferation index (PI) in patients with pathologically confirmed gastrointestinal stromal tumors (GIST).

METHODS

In this retrospective study, a total of 412 GIST patients across three institutions (223 from center 1, 106 from center 2, and 83 from center 3) was enrolled. Radiomic features were derived from various sub-regions of the tumor region of interest employing the K-means approach. The Least Absolute Shrinkage and Selection Operator (LASSO) regression was employed to identify features correlated with Ki-67 PI level in GIST patients. A support vector machine (SVM) model was then constructed to predict the high level of Ki-67 (Ki-67 index >8%), drawing on the radiomics features from each sub-region within the training cohort.

RESULTS

After features selection process, 6, 9, 9, 7 features were obtained to construct SVM models based on sub-region 1, 2, 3 and the entire tumor, respectively. Among different models, the model developed by the sub-region 1 achieved an area under the receiver operating characteristic curve (AUC) of 0.880 (95% confidence interval [CI]: 0.830 to 0.919), 0.852 (95% CI: 0.770-0.914), 0.799 (95% CI: 0.697-0.879) in the training, external test set 1, and 2, respectively.

CONCLUSION

The results of the present study suggested that SVM model based on the sub-regional radiomics features had the potential of predicting Ki-67 PI level in patients with GIST.

Prediction of Ki-67 expression in gastrointestinal stromal tumors using radiomics of plain and multiphase contrast-enhanced CT

OBJECTIVE

To study the value of radiomics models based on plain and multiphase contrast-enhanced CT to predict Ki-67 expression in gastrointestinal stromal tumors (GISTs).

METHODS

A total of 215 patients with GISTs were retrospectively analyzed, including 150 patients in one hospital as the training set and 65 patients in another hospital as the external verification set. The tumor at the largest level of CT images was delineated as the region of interest (ROI). The maximum diameter of the ROI was defined as the tumor size. A total of 851 radiomics features were extracted from each ROI by 3D Slicer Radiomics. After dimensionality reduction, three machine learning classification algorithms including logistic regression (LR), random forest (RF), and support vector machine (SVM) were used for Ki-67 expression prediction. Using a multivariable logistic model, a nomogram was established to predict the expression of Ki-67 individually.

RESULTS

Delong tests showed that the SVM models had the highest accuracy in the arterial phase (Z value 0.217-1.139) and venous phase (Z value 0.022-1.396). For the plain phase, LR and SVM models had the highest accuracy (Z value 0.874-1.824, 1.139-1.763). For the delayed phase, LR models had the highest accuracy (Z value 0.056-1.824). For the combined phase, RF models had the highest accuracy (Z value 0.232-1.978). There was no significant difference among the above models for KI-67 expression prediction (Z value 0.022-1.978). A nomogram was developed with a C-index of 0.913 (95% CI, 0.878 to 0.956).

CONCLUSIONS

Radiomics of both plain and enhanced CT images could accurately predict the expression of Ki-67 in GIST. For patients who were not suitable to use contrast agents, plain scan could be used as an alternative.

CLINICAL RELEVANCE STATEMENT

CT radiomics could accurately predict the expression of Ki-67 in GIST, which has a great clinical value in reflecting the proliferative activity of tumor cells and helping determine whether a patient is suitable for adjuvant therapy with imatinib.

KEY POINTS

• Radiomics of both plain and enhanced CT images could accurately predict the expression of Ki-67 in GIST. • For patients who were not suitable to use contrast agents, plain scan could be used as an alternative. • A radiomics nomogram was developed to allow personalized preoperative evaluation with high accuracy.

Association between quantitative spectral CT parameters, Ki-67 expression, and invasiveness in lung adenocarcinoma manifesting as ground-glass nodules

BACKGROUND

Few studies about lung ground-glass nodules (GGNs) have been done using non-enhancement spectral computed tomography (CT) imaging.

PURPOSE

To examine the association between spectral CT parameters, Ki-67 expression, and invasiveness in lung adenocarcinoma manifesting as GGNs.

MATERIAL AND METHODS

Spectral CT parameters were analyzed in 106 patients with lung GGNs. The Ki-67 labeling index (Ki-67 LI) was measured, and patients were divided into low expression and high expression groups according to the number of positive-stained cells (low expression ≤10%; high expression >10%). Spectral CT parameters were compared between low and high expression groups. The correlation between spectral CT parameters and Ki-67 LI was estimated by Spearman correlation analysis. Cases were divided into a preinvasive and minimally invasive adenocarcinoma (MIA) group (atypical adenomatous hyperplasia, adenocarcinoma in situ, and MIA) and invasive adenocarcinoma (IA) group. Spectral CT parameters were compared between the two groups. The diagnostic performance was evaluated using receiver operating characteristic analysis.

RESULTS

There were significant differences in water concentration of lesions (WCL) and monochromatic CT values between the low and high expression groups. CT 40 keV had the highest correlation coefficient with Ki-67 LI. WCL and monochromatic CT values were significantly higher in the IA group than in the pre/MIA group. The value of area under the curve of CT 40 keV was 0.946 (95% confidence interval=0.905-0.988) for differentiating the two groups; the cutoff was -280.66 Hu.

CONCLUSION

Spectral CT is an effective non-invasive method for the prediction of proliferation and invasiveness in lung adenocarcinoma manifesting as GGNs.

CT Radiomics for the Preoperative Prediction of Ki67 Index in Gastrointestinal Stromal Tumors: A Multi-Center Study

Purpose

This study established and verified a radiomics model for the preoperative prediction of the Ki67 index of gastrointestinal stromal tumors (GISTs).

Materials and Methods

A total of 344 patients with GISTs from three hospitals were divided into a training set and an external validation set. The tumor region of interest was delineated based on enhanced computed-tomography (CT) images to extract radiomic features. The Boruta algorithm was used for dimensionality reduction of the features, and the random forest algorithm was used to construct the model for radiomics prediction of the Ki67 index. The receiver operating characteristic (ROC) curve was used to evaluate the model’s performance and generalization ability.

Results

After dimensionality reduction, a feature subset having 21 radiomics features was generated. The generated radiomics model had an the area under curve (AUC) value of 0.835 (95% confidence interval(CI): 0.761–0.908) in the training set and 0.784 (95% CI: 0.691–0.874) in the external validation cohort.

Conclusion

The radiomics model of this study had the potential to predict the Ki67 index of GISTs preoperatively.

Spectral CT imaging parameters and Ki-67 labeling index in lung adenocarcinoma

Objective

To explore the correlation between the spectral computed tomography (CT) imaging parameters and the Ki-67 labeling index in lung adenocarcinoma.

Methods

Spectral CT imaging parameters [iodine concentrations of lesions (ICLs) in the arterial phase (ICLa) and venous phase (ICLv), normalized IC in the aorta (NICa/NICv), slope of the spectral HU curve (λHUa/λHUv) and monochromatic CT number enhancement on 40 keV and 70 keV images (CT40keVa/v, CT70keVa/v)] in 34 lung adenocarcinomas were analyzed, and common molecular markers, including the Ki-67 labeling index, were detected with immunohistochemistry. Different Ki-67 labeling indexes were measured and grouped into four grades according to the number of positive-stained cells (grade 0, ≤1%; 1%<grade 1≤10%; 10%<grade 2≤30%; and grade 3, >30%). One-way analysis of variance (ANOVA) was used to compare the four different grades, and the Bonferroni method was used to correct the P value for multiple comparisons. A Spearman correlation analysis was performed to further research a quantitative correlation between the Ki-67 labeling index and spectral CT imaging parameters.

Results

CT40keVa, CT40keVv, CT70keVa and CT70keVv increased as the grade increased, and CT70keVa and CT70keVv were statistically significant (P<0.05). These four parameters and the Ki-67 labeling index showed a moderate positive correlation with lung adenocarcinoma nodules. ICL, NIC and λHU in the arterial and venous phases were not significantly different among the four grades.

Conclusions

The spectral CT imaging parameters CT40keVa, CT40keVv, CT70keVa and CT70keVv gradually increased with Ki-67 expression and showed a moderate positive correlation with lung adenocarcinomas. Therefore, spectral CT imaging parameter-enhanced monochromatic CT numbers at 70 keV may indicate the extent of proliferation of lung adenocarcinomas.

Correction: Preoperative Prediction of Ki-67 Labeling Index By Three-dimensional CT Image Parameters for Differential Diagnosis Of Ground-Glass Opacity (GGO)

[This corrects the article DOI: 10.1371/journal.pone.0129206.].

[A Retrospective Study of Mean Computed Tomography Value to Predict the Tumor Invasiveness in AAH and Clinical Stage Ia Lung Cancer]

BACKGROUND

Recently, the detectable rate of ground-glass opacity (GGO ) was significantly increased, a appropriate diagnosis before clinic treatment tends to be important for patients with GGO lesions. The aim of this study is to validate the ability of the mean computed tomography (m-CT) value to predict tumor invasiveness, and compared with other measurements such as Max CT value, GGO size, solid size of GGO and C/T ratio (consolid/tumor ratio, C/T) to find out the best measurement to predict tumor invasiveness.

METHODS

A retrospective study was conducted of 129 patients who recieved lobectomy and were pathological confirmed as atypical adenomatous pyperplasia (AAH) or clinical stage Ia lung cance in our center between January 2012 and December 2013. Of those 129 patients, the number of patients of AAH, AIS, AIS and invasive adenocarcinoma were 43, 26, 17 and 43, respectively. We defined AAH and AIS as noninvasive cancer (NC), MIA and invasive adenocarcinoma were categorized as invasive cancer(IC). We used receiver operating characteristic (ROC) curve analysis to compare the ability to predict tumor invasiveness between m-CT value, consolidation/tumor ratio, tumor size and solid size of tumor. Multiple logistic regression analyses were performed to determine the independent variables for prediction of pathologic more invasive lung cancer.

RESULTS

129 patients were enrolled in our study (59 male and 70 female), the patients were a median age of (62.0±8.6) years (range, 44 to 82 years). The two groups were similar in terms of age, sex, differentiation (P>0.05). ROC curve analysis was performed to determine the appropriate cutoff value and area under the cure (AUC). The cutoff value of solid tumor size, tumor size, C/T ratio, m-CT value and Max CT value were 9.4 mm, 15.3 mm, 47.5%, -469.0 HU and -35.0 HU, respectively. The AUC of those variate were 0.89, 0.79, 0.82, 0.90, 0.85, respectively. When compared the clinical and radiologic data between two groups, we found the IC group was strongly associated with a high m-CT value, high Max CT value, high C/T ratio and large tumor size. Gender, solid tumor size, tumor size, C/T ratio, m-CT value and MaxCT value were selected factor for multivariate analysis, when using the preoperatively determined variables to predict the tumor invasiveness, revealed that tumor size, C/T ratio, m-CT value and Max CT value were independent predictive factors of IC.

CONCLUSIONS

The musurements of Max CT value, GGO size, solid size of GGO and C/T ratio were significantly correlated with tumor invasiveness, and the evaluation of m-CT value is most useful musurement in predicting more invasive lung cancer.

Stage I lung adenocarcinoma: the value of quantitative CT in differentiating pathological subtypes and predicting growth of subsolid nodules

The aim of this study was to investigate feasibility of quantitative computed tomography (CT) measurements in predicting invasiveness and growth of nodular ground glass opacities (nGGOs).A set of 203 patients (group A) with nGGOs that were confirmed stage-I adenocarcinomas and 79 patients (group B) with nGGOs that were completely followed up were included. Lesions diameters, volume (VOL), maximum (MAX), mean (MEN), and standard deviation (STD) of CT attenuation were measured. P53 labeling index (LI) was evaluated through immunohistochemistry in group-A patients. Multivariate linear stepwise regressions were performed based on group-A lesions to calculate P53-LI prediction from CT measurements. The receiver operating characteristic (ROC) curve analyses were performed to assess the performance of P53-LI prediction in predicting invasiveness and growth of nGGOs. The Cox regression analysis was conducted to identify correlation between P53-LI Prediction and volume doubling time (VDT) of lesions in group B.Diameter, VOL, MEN, STD, and the P53 LI showed significant differences between lesions of different pathological invasiveness (P < .01). By multivariate linear regressions, MEN and STD were identified as independent variables indicating P53 LI (P < .001); thus, an equation was established to calculate P53-LI Prediction as: P53LI Prediction = 0.013 ×  MEN + 0.024 × STD + 9.741 (R square = 0.411, P < .001). The P53-LI Prediction showed good performance, similar as the actual one, in differentiating pathological invasiveness of nGGOs. In addition, the P53-LI Prediction demonstrated excellent performance in predicting growth of nGGOs (AUC = 0.833, P < .001) and independently forecasted VDT of nGGOs (β = 1.773, P < .001).The P53-LI Prediction that was calculated from preoperative quantitative CT measurements of nGGOs indicates lesions' invasiveness and allows for predicting growth of nGGOs.

Predictive value of mutant p53 expression index obtained from nonenhanced computed tomography measurements for assessing invasiveness of ground-glass opacity nodules

Purpose

To predict p53 expression index (p53-EI) based on measurements from computed tomography (CT) for preoperatively assessing pathologies of nodular ground-glass opacities (nGGOs).

Methods

Information of 176 cases with nGGOs on high-resolution CT that were pathologically confirmed adenocarcinoma was collected. Diameters, total volumes (TVs), maximum (MAX), average (AVG), and standard deviation (STD) of CT attenuations within nGGOs were measured. p53-EI was evaluated through immunohistochemistry with Image-Pro Plus 6.0. A multiple linear stepwise regression model was established to calculate p53-EI prediction from CT measurements. Receiver-operating characteristic curve analysis was performed to compare the diagnostic performance of variables in differentiating preinvasive adenocarcinoma (PIA), minimally invasive adenocarcinoma (MIA), and invasive adenocarcinoma (IAC).

Results

Diameters, TVs, MAX, AVG, and STD showed significant differences among PIAs, MIAs, and IACs (all P-values <0.001), with only MAX being incapable to differentiate MIAs from IACs (P=0.106). The mean p53-EIs of PIAs, MIAs, and IACs were 3.4±2.0, 7.2±1.9, and 9.8±2.7, with significant intergroup differences (all P-values <0.001). An equation was established by multiple linear regression as: p53-EI prediction =0.001* TVs +0.012* AVG +0.022* STD +9.345, through which p53-EI predictions were calculated to be 4.4%±1.0%, 6.8%±1.3%, and 8.5%±1.4% for PIAs, MIAs, and IACs (Kruskal–Wallis test P<0.001; Tamhane’s T2 test: PIA vs MIA P<0.001, MIA vs IAC P<0.001), respectively. Although not significant, p53-EI prediction has a little higher area under the curve (AUC) than the actual one both in differentiating MIAs from PIAs (AUC 0.938 vs 0.914, P=0.263) and in distinguishing IACs from MIAs (AUC 0.812 vs 0.786, P=0.718).

Conclusion

p53-EI prediction of nGGOs obtained from CT measurements allows accurately estimating lesions’ pathology and invasiveness preoperatively not only from radiology but also from pathology.