Accuracy and feasibility of estimated tumour volumetry in primary gastric gastrointestinal stromal tumours: validation using semiautomated technique in 127 patients

Diagnostic value of dual-source, dual-energy computed tomography combined with the neutrophil-lymphocyte ratio for discriminating gastric signet ring cell from mixed signet ring cell and non-signet ring cell carcinomas

PURPOSE

To explore the diagnostic value of dual-source computed tomography (DSCT) and neutrophil to lymphocyte ratio (NLR) for differentiating gastric signet ring cell carcinoma (SRC) from mixed SRC (mSRC) and non-SRC (nSRC).

METHODS

This retrospective study included patients with gastric adenocarcinoma who underwent DSCT between August 2019 and June 2021 at our Hospital. The iodine concentration in the venous phase (ICvp), standardized iodine concentration (NICVP), and the slope of the energy spectrum curve (kVP) were extracted from DSCT data. NLR was determined from laboratory results. DSCT (including ICVP, NICVP, and kVP) and combination (including DSCT model and NLR) models were established based on the multinomial logistic regression analysis. The receiver operator characteristic (ROC) curve and area under the curve (AUC) were used to evaluate the diagnostic value.

RESULTS

A total of 155 patients (SRC [n = 45, aged 61.22 ± 11.4 years], mSRC [n = 60, aged 61.09 ± 12.7 years], and nSRC [n = 50, aged 67.66 ± 8.76 years]) were included. There were significant differences in NLR, ICVP, NICVP, and kVP among the SRC, mSRC, and nSRC groups (all P < 0.001). The AUC of the combination model for SRC vs. mSRC + nSRC was 0.964 (95% CI: 0.923-1.000), with a sensitivity of 98.3% and a specificity of 86.7%, higher than with DSCT (AUC: 0.959, 95% CI: 0.919-0.998, sensitivity: 90.0%, specificity: 89.9%) or NLR (AUC: 0.670, 95% CI: 0.577-0.768, sensitivity: 62.2%, specificity: 61.8%).

CONCLUSION

DSCT combined with NLR showed high diagnostic efficacy in differentiating SRC from mSRC and nSRC.

Automated segmentation and volume prediction in pediatric Wilms' tumor CT using nnu-net

BACKGROUND

Radiologic volumetric evaluation of Wilms' tumor (WT) is an important indicator to guide treatment decisions. However, due to the heterogeneity of the tumors, radiologists have main-guard differences in diagnosis that can lead to misdiagnosis and poor treatment. The aim of this study was to explore whether CT-based outlining of WT foci can be automated using deep learning.

METHODS

We included CT intravenous phase images of 105 patients with WT and double-blind outlining of lesions by two radiologists. Then, we trained an automatic segmentation model using nnUnet. The Dice similarity coefficient (DSC) and 95th percentile Hausdorff distance (HD95) were used to assess the performance. Next, we optimized the automatic segmentation results based on the ratio of the three-dimensional diameter of the lesion to improve the performance of volumetric assessment.

RESULTS

The DSC and HD95 was 0.83 ± 0.22 and 10.50 ± 8.98 mm. The absolute difference and percentage difference in tumor size was 72.27 ± 134.84 cm3 and 21.08% ± 30.46%. After optimization according to our method, it decreased to 40.22 ± 96.06 cm3 and 10.16% ± 9.70%.

CONCLUSION

We introduce a novel method that enhances the accuracy of predicting WT volume by integrating AI automated outlining and 3D tumor diameters. This approach surpasses the accuracy of using AI outcomes alone and has the potential to enhance the clinical evaluation of pediatric patients with WT. By intertwining AI outcomes with clinical data, this method becomes more interpretive and offers promising applications beyond Wilms tumor, extending to other pediatric diseases.

Radiologic versus Segmentation Measurements to Quantify Wilms Tumor Volume on MRI in Pediatric Patients

Wilms tumor is a common pediatric solid tumor. To evaluate tumor response to chemotherapy and decide whether nephron-sparing surgery is possible, tumor volume measurements based on magnetic resonance imaging (MRI) are important. Currently, radiological volume measurements are based on measuring tumor dimensions in three directions. Manual segmentation-based volume measurements might be more accurate, but this process is time-consuming and user-dependent. The aim of this study was to investigate whether manual segmentation-based volume measurements are more accurate and to explore whether these segmentations can be automated using deep learning. We included the MRI images of 45 Wilms tumor patients (age 0–18 years). First, we compared radiological tumor volumes with manual segmentation-based tumor volume measurements. Next, we created an automated segmentation method by training a nnU-Net in a five-fold cross-validation. Segmentation quality was validated by comparing the automated segmentation with the manually created ground truth segmentations, using Dice scores and the 95th percentile of the Hausdorff distances (HD95). On average, manual tumor segmentations result in larger tumor volumes. For automated segmentation, the median dice was 0.90. The median HD95 was 7.2 mm. We showed that radiological volume measurements underestimated tumor volume by about 10% when compared to manual segmentation-based volume measurements. Deep learning can potentially be used to replace manual segmentation to benefit from accurate volume measurements without time and observer constraints.

Genetic features and therapeutic relevance of emergent circulating tumor DNA alterations in refractory non-colorectal gastrointestinal cancers

Acquired resistance to systemic treatments is inevitable in most cancers, but the genetic basis for this in many cancer types has remained elusive due to constraints in obtaining tissue specimens longitudinally. In the management of gastrointestinal cancers, molecular profiling is conventionally performed at a single time point, although serial evaluations may yield biological insights that inform treatment decisions. We characterize genetic changes in serial liquid biopsies which provide real-time snapshots of tumor genetics and heterogeneity in refractory non-colorectal gastrointestinal cancers, and determine the clinical utility of repeat circulating tumor DNA (ctDNA) testing. In a national cohort of 449 patients with pancreatic, biliary, esophagogastric, and hepatocellular cancers, resistance to conventional therapies is broadly associated with tumor evolution. Emergent ctDNA alterations only detectable at progression occurs in 63% of patients and are frequently associated with treatment actionability. Tumor mutation burden is dynamic in cancers undergoing treatment, but is not associated with time to progression. Objective tumor responses in a case series of patients receiving treatment matched to emergent alterations show that repeat liquid biopsies may have clinical benefit by expanding treatment options in advanced gastrointestinal cancers.

Predictors of hearing functional outcome following surgery for cerebellopontine angle meningioma

OBJECTIVE

Cerebellopontine angle (CPA) meningiomas can affect hearing function and require expeditious treatment to prevent permanent hearing loss. The authors sought to determine the factors associated with functional hearing outcome in CPA meningioma patients treated with surgery and/or radiation therapy in the form of either stereotactic radiosurgery or stereotactic radiation therapy.

METHODS

Consecutive patients with CPA meningiomas who had presented at our hospital from 2008 to 2018 were identified through retrospective chart review. Hearing function (as defined by pure tone average (PTA) and speech discrimination score (SDS) on Audiogram) was assessed before and after surgery for CPA meningioma. Audiograms with PTA > 50 dB and SDS < 69% were defined as poor hearing functional outcome. Multivariable Cox Proportional Hazards Regression Model was used to assess the associations between pre-operative hearing functional assessment and post-operative hearing functional outcomes.

RESULTS

The study cohort included 31 patients (80.6% females, with a mean age of 61.3 ± 15.2 years) with a median clinical follow-up of 5 months (range: 1 week-98 months). The mean pre-operative PTA and SDS were 23.8 ± 11.2 dB and 64.4 ± 22.2% respectively. At the last visit, there was significant hearing recovery, with an improvement of 29.7 ± 18.0 dB (p < 0.001) and 87.6 ± 17.8% (p < 0.001) in PTA and SDS respectively. After adjusting for age, gender, tumor volume, location, and tumor classification, Multivariable Cox Proportional Hazards Regression Model was conducted which revealed that patients undergoing surgery through retro sigmoid approach [Hazards Ratio (HR): 32.1, 95% Confidence Interval (CI): 2.11-491.0, p = 0.01] and gross total resection (GTR) (HR: 2.99, 95% CI: 1.09-9.32, p = 0.05) had significantly higher risk of poor hearing functional outcome compared to petrosal approach and near/subtotal resection. Moreover, patients with poor preoperative hearing had 85% higher chance of poor hearing functional outcome postoperatively (HR: 0.15, 95%CI: 0.03-0.59, p = 0.007).

CONCLUSION

Postoperative improvement in hearing is a reasonable expectation following surgery for CPA meningioma. Preoperative hearing, surgical approach and extent of surgical resection are predictive factors of postoperative hearing function outcome and can therefore aid in identification of patients at higher risk of hearing loss.

Early dynamics of circulating tumor DNA predict chemotherapy responses for patients with esophageal cancer

We investigated whether early circulating tumor DNA (ctDNA) changes, measured using digital PCR (dPCR), can predict later chemotherapy responses in esophageal squamous cell cancer (ESCC). We compared the dynamics of ctDNA and tumor volumes during chemotherapy in 42 ESCC. The accuracy of predictions of later chemotherapy responses was evaluated by the ratio of the variant allele frequency of ctDNA (post-/pre-ctDNA) and the total tumor volume (post-/pre-volume) before and after an initial chemotherapy cycle using a receiver-operating characteristic curve analysis. Total positive and negative objective responses (ORs) were defined as either >50 or ≤50% reductions, respectively, in the total tumor volume at the end of first-line chemotherapy. Mutation screening of 43 tumors from 42 patients revealed 96 mutations. The pretreatment dPCR-ctDNA data were informative in 38 patients, using 70 selected mutations (1-3 per patient). The areas under the curve (AUCs) for the post-/pre-volume and post-/pre-ctDNA levels used in predicting the total OR were 0.85 and 0.88, respectively. The optimal cutoff value of post-/pre-ctDNA was 0.13. In 20 patients with post-/pre-volume ≥50%, the total OR could be predicted by the post-/pre-ctDNA with high accuracy; the AUC by post-/pre-ctDNA was higher than that by post-/pre-volume (0.85 versus 0.76, respectively). Patients with low post-/pre-ctDNA (n = 18) had a significantly better overall survival rate than those with high post-/pre-ctDNA (n = 20; P = 0.03). Early ctDNA changes after an initial cycle of chemotherapy predict later responses to treatment with high accuracy in ESCC patients.

A Computational Study on Magnetic Nanoparticles Hyperthermia of Ellipsoidal Tumors

The modelling of magnetic hyperthermia using nanoparticles of ellipsoid tumor shapes has not been studied adequately. To fill this gap, a computational study has been carried out to determine two key treatment parameters: the therapeutic temperature distribution and the extent of thermal damage. Prolate and oblate spheroidal tumors, of various aspect ratios, surrounded by a large healthy tissue region are assumed. Tissue temperatures are determined from the solution of Pennes’ bio-heat transfer equation. The mortality of the tissues is determined by the Arrhenius kinetic model. The computational model is successfully verified against a closed-form solution for a perfectly spherical tumor. The therapeutic temperature and the thermal damage in the tumor center decrease as the aspect ratio increases and it is insensitive to whether tumors of the same aspect ratio are oblate or prolate spheroids. The necrotic tumor area is affected by the tumor prolateness and oblateness. Good comparison is obtained of the present model with three sets of experimental measurements taken from the literature, for animal tumors exhibiting ellipsoid-like geometry. The computational model enables the determination of the therapeutic temperature and tissue thermal damage for magnetic hyperthermia of ellipsoidal tumors. It can be easily reproduced for various treatment scenarios and may be useful for an effective treatment planning of ellipsoidal tumor geometries.

Intelligent Algorithm-Based CT Imaging for Evaluation of Efficacy of Docetaxel Combined with Fluorouracil on Patients with Gastric Cancer

The study focused on the dual-source computed tomography (CT) images segmented by the decision tree algorithm, to explore the efficacy of docetaxel combined with fluorouracil therapy on gastric patients undergoing chemotherapy. In this study, 98 patients with gastric cancer who were treated in the hospital were selected as the research subjects. The decision tree algorithm was applied to segment dual-source CT images of gastric cancer patients. The decision tree is established according to the feature ring and the segmentation position. The machine inductively learns from the decision tree to extract the features of the CT image to obtain the optimal segmentation boundary. The observation group was treated with docetaxel combined with fluorouracil, and the control group was treated with docetaxel combined with tegafur gimeracil oteracil potassium capsules. The general data of the two groups of patients were comparable and not statistically significant (P > 0.05). The two groups were compared for clinical efficacy, physical status, KPS score, improvement rate, and adverse drug reactions after treatment. The results showed that the improvement rate of physical fitness in the observation group was 38.78%, and the improvement rate in the control group was 18.37%. The total effective rate in the observation group was 42.85%, and the total effective rate in the control group was 36.73%. Obviously, the curative effect and improvement rate of physical fitness in the observation group were significantly better than those in the control group (P < 0.05). In conclusion, the decision tree algorithm proposed in this study demonstrates superb capabilities in feature extraction of CT images. The machine inductively learns from the decision tree to extract the features of the CT image to obtain the optimal segmentation boundary. The effect of docetaxel combined with fluorouracil is better than that of docetaxel combined with tegafur gimeracil oteracil potassium capsules.

Clinical and biological determinants of circulating tumor DNA detection and prognostication using a next-generation sequencing panel assay

Circulating tumor DNA (ctDNA) is utilized for molecular profiling of cancers, and is under investigation for a growing number of applications based on the assumption that ctDNA levels faithfully reflect disease burden. Our objective was to investigate whether patient and tumor characteristics may impact ctDNA detection or levels and the prognostic significance of ctDNA levels or mutations. We performed a retrospective cohort analysis of a comprehensively annotated cohort of 561 patients at a National Cancer Institute-designated comprehensive cancer center with advanced solid cancers who underwent ctDNA testing using a commercial targeted next-generation sequencing assay. ctDNA detection in advanced cancers was associated with older age, non-obese body mass index, and diabetes, but not with tumor diameter, volume, lesion number, or other pathological features. Regression models indicate that no more than 14.3% of the variance in ctDNA levels between patients was explained by known clinical factors and disease burden. Even after adjusting for established prognostic factors and tumor burden, ctDNA levels were associated with worse survival among patients without prior systemic therapy, while ctDNA mutations were associated with survival among patients who previously received systemic treatment. These findings uncover clinical factors that affect ctDNA detection in patients with advanced cancers and challenge the convention that ctDNA is a surrogate for tumor burden. Our study also indicates that the prognostic value of ctDNA levels and mutations are independent of tumor burden and dependent on treatment context.

Predicting Chemotherapeutic Response for Far-advanced Gastric Cancer by Radiomics with Deep Learning Semi-automatic Segmentation

Purpose: To build a dual-energy computed tomography (DECT) delta radiomics model to predict chemotherapeutic response for far-advanced gastric cancer (GC) patients. A semi-automatic segmentation method based on deep learning was designed, and its performance was compared with that of manual segmentation. Methods: This retrospective study included 86 patients with far-advanced GC treated with chemotherapy from September 2016 to December 2017 (66 and 20 in the training and testing cohorts, respectively). Delta radiomics features between the baseline and first follow-up DECT were modeled by random forest to predict the chemotherapeutic response evaluated by the second follow-up DECT. Nine feature subsets from confounding factors and delta radiomics features were used to choose the best model with 10-fold cross-validation in the training cohort. A semi-automatic segmentation method based on deep learning was developed to predict the chemotherapeutic response and compared with manual segmentation in the testing cohort, which was further validated in an independent validation cohort of 30 patients. Results: The best model, constructed by confounding factors and texture features, reached an average AUC of 0.752 in the training cohort. Our proposed semi-automatic segmentation method was more time-effective than manual segmentation, with average saving-time of 11.2333 ± 6.3989 minutes and 9.9889 ±5.5086 minutes in the testing cohort and the independent validation cohort, respectively (both p < 0.05). The predictive ability of the semi-automatic segmentation was also better than that of the manual segmentation both in the testing cohort and the independent validation cohort (AUC: 0.728 vs. 0.687 and 0.828 vs. 0.749, respectively). Conclusion: DECT delta radiomics serves as a promising biomarker for predicting chemotherapeutic response for far-advanced GC. Semi-automatic segmentation based on deep learning shows the potential for clinical use with increased reproducibility and decreased labor costs compared to the manual version.

Correlation between Three-Dimensional Volume and Malignant Potential of Gastrointestinal Stromal Tumors (GISTs)

Background and purpose: Gastrointestinal stromal tumors (GISTs) are rare diseases of the gastrointestinal tract but they are the most common gastrointestinal tumors of mesenchymal origin. Since most GISTs have malignant potential, their probability of malignant progression must be evaluated. This study was conducted to examine the correlation between three-dimensional GIST volume measured by CT and malignant potential. Materials and methods: A retrospective study was performed on 70 patients diagnosed with GIST after surgical resection in Hanyang University Seoul Hospital from 2012 to 2017. Linear regression analysis was used to establish which between the length of the long axis of GISTs, originally considered a predictor of malignancy, and their volume was a more accurate predictor of malignancy. Tumor dimensions were measured by CT. Results: Data were analyzed using the chi-square test or Student’s t-test and logistic regression. Of the GISTs, 53 (75.71%) were in the stomach, 3 (4.29%) in the small intestine, and 14 (20.0%) in the large intestine. The mean age of the malignant GIST group was significantly higher than that of the benign GIST group (p = 0.032), their tumor long axes were significantly greater (p = 0.073), their tumor volumes were significantly larger (p = 0.001), and the frequency of tumor necrosis was higher (p = 0.001). In multivariate analysis, malignant GIST was associated with location in organs other than the stomach (OR 7.846, 95% CI 1.293–47.624, p = 0.025), longer axis (OR 1.037, 95% CI 1.011–1.065, p = 0.006), larger volume (OR 1.003, 95% CI 1.000–1.006, p = 0.029), and necrosis (OR 12.222, 95% CI 1.945–76.794, p = 0.008). The mean age of the recurrent GIST group was higher than that of the non-recurrent group (p = 0.045), their tumor long axes were significantly longer (p = 0.005), and their volumes were greater, but this last difference was not significant (p = 0.072). Conclusions: Tumor volume can be considered an additional risk factor in assessing the malignant potential of GISTs and tends to increase in recurrent GISTs.

Quantitative analysis of breast tumours aided by three-dimensional photoacoustic/ultrasound functional imaging

In this pilot study, we explored a quantitative method to analyse characteristics of breast tumours using 3D volumetric data obtained from a three-dimensional (3D) photoacoustic/ultrasound (PA/US) functional imaging system. Imaging results from 24 Asian patients with maximum tumour diameters less than 2 cm, including 8 benign tumours, 16 T1 stage invasive breast cancers (IBCs), and 22 normal breasts, were analysed. We found that the volumetric mean oxygenation saturation (SO₂) in tumour regions of T1 stage IBCs was 7.7% lower than that of benign tumours (P = 0.016) and 3.9% lower than that of healthy breasts (P = 0.010). The volumetric mean SO₂ in tumour surrounding regions of T1 stage IBCs was 4.9% lower than that of benign tumours (P = 0.009). For differentiating T1 stage IBCs and benign tumours, with a cut-off SO₂ value of 78.2% inside tumours, we obtained a sensitivity of 100%, a specificity of 62.5%, and an AUC of 0.81; with a cut-off SO₂ value of 77.9% in regions surrounding tumours, we obtained a sensitivity of 100%, a specificity of 75% and an AUC of 0.84. Our preliminary results demonstrate that 3D PA/US functional imaging has the potential to provide valuable quantitative physiological information that may be useful for the detection and evaluation of breast tumours.

Gastrointestinal Stromal Tumors Risk of Recurrence Stratification by Tumor Volume is a Best Predictor Compared with Risk Based on Mitosis and Tumor Size

PURPOSE

Gastrointestinal stromal tumors (GIST) have the potential to recur and metastasize. Several prognostic schemes have been developed, mostly based on the mitotic count, diameter, and tumor site. However, these systems are not precise enough. The research question was whether the tumor size determined by volumetry allows a better risk stratification than the traditional system, and our aim was to determine the value of tumor volumetry, a feasible and simple parameter, in the recurrence of GIST.

METHODS

Seventy-four cases of GIST were studied. The cases presented with non-metastatic disease, which were resected and did not receive imatinib. We compared the clinico-pathologic features of the cases with recurrence against those with non-recurrence and compared the tumor volumetry against the classification system based on tumor size and mitosis.

RESULTS

The median age was 58 years (range: 25 to 91 years). Half of the cases were presented in the stomach. The tumor size had a median of 8 cm (range of 1-30 cm). The median mitosis count for 50 HPF was 4 (range 0-92). During the period of study, 16 (21.6%) patients suffered recurrence. The significant differences were that patients with recurrence accounted for more deaths and the follow-up period was larger. The area under the curve (AUC) of the volumetry classification was superior to the AUC of the classification system based on tumor size and mitosis (NIH-criteria) (p = .05).

CONCLUSION

Tumor volumetry calculated in the surgical specimen and/or pre-operative tomography was superior to the NIH consensus in stratifying the risk of recurrence in GIST.

[Focal lesions in whole-body MRI in multiple myeloma : Quantification of tumor mass and correlation with disease-related parameters and prognosis]

BACKGROUND AND OBJECTIVES

In this study, we evaluated methods of quantification of tumor mass in whole-body MRI (wb-MRI) in multiple myeloma and correlated these with disease-related parameters in serum and bone marrow.

MATERIALS AND METHODS

We retrospectively evaluated wb-MRIs of 52 patients with focal infiltration pattern and a total of 700 focal lesions (subsequently called lesions). We determined the longest diameter (LD), the segmented volume (SV), and the morphology (spherical or non-spherical). We correlated total number/volume of the lesions with clinical parameters and prognosis and furthermore LD with SV. After that we analyzed the agreement of SV and estimated volume (EV) using the volume formula of a sphere based on LD.

RESULTS

Results showed no significant correlations of total number/volume with prognosis or clinical parameters. The latter were situated predominantly in the normal range. Furthermore, 10% of lesions were spherical. SV and LD correlated significantly in single lesions and on patient level. SV was in lesions <6 cm³ systematically larger and in lesions ≥6 cm³ smaller than EV. In 95%, we found in small lesions a deviation of EV versus SV from +0.9 cm³ to -4.6 cm³ and in large lesions from +160 cm³ to -111 cm³ (EV-SV).

CONCLUSIONS

Quantification of tumor mass in the focal infiltration pattern is performed more accurately by volumetry than LD due to the predominant existence of non-spherical lesions. The patient cohort with clinical parameters predominantly in the normal range is distributed to ISS stage I and partly pretreated, a fact that makes interpretation of absent correlations more difficult. Consider also a variation in activitiy of lesions and a diffuse infiltration not detectable by MRI.

Pharmacometric Modeling of Liver Metastases' Diameter, Volume, and Density and Their Relation to Clinical Outcome in Imatinib-Treated Patients With Gastrointestinal Stromal Tumors

Three‐dimensional and density‐based tumor metrics have been suggested to better discriminate tumor response to treatment than unidimensional metrics, particularly for tumors exhibiting nonuniform size changes. In the developed pharmacometric modeling framework based on data from 77 imatinib‐treated gastrointestinal patients, the time‐courses of liver metastases' maximum transaxial diameters, software‐calculated actual volumes (V_actual) and calculated ellipsoidal volumes were characterized by logistic growth models, in which imatinib induced a linear dose‐dependent size reduction. An indirect response model best described the reduction in density. Substantial interindividual variability in the drug effect of all response assessments and additional interlesion variability in the drug effect on density were identified. The predictive ability of longitudinal tumor unidimensional and three‐dimensional size and density on overall survival (OS) and progression‐free survival (PFS) were compared using parametric time‐to‐event models. Death hazard increased with increasing V_actual. This framework may guide early clinical interventions based on three‐dimensional tumor responses to enhance benefits for patients with gastrointestinal stromal tumors (GIST).

Update on Gastrointestinal Stromal Tumors for Radiologists

The management of gastrointestinal stromal tumors (GISTs) has evolved significantly in the last two decades due to better understanding of their biologic behavior as well as development of molecular targeted therapies. GISTs with exon 11 mutation respond to imatinib whereas GISTs with exon 9 or succinate dehydrogenase subunit mutations do not. Risk stratification models have enabled stratifying GISTs according to risk of recurrence and choosing patients who may benefit from adjuvant therapy. Assessing response to targeted therapies in GIST using conventional response criteria has several potential pitfalls leading to search for alternate response criteria based on changes in tumor attenuation, volume, metabolic and functional parameters. Surveillance of patients with GIST in the adjuvant setting is important for timely detection of recurrences.

Computed tomography features and predictive findings of ruptured gastrointestinal stromal tumours

OBJECTIVES

To evaluate the CT features of ruptured GISTs and factors that might be predictive of rupture through comparison with CTs taken prior to rupture and CTs of non-ruptured GIST.

METHODS

Forty-nine patients with ruptured GIST and forty-nine patients with non-ruptured GIST matched by age, gender and location were included. Clinical data including pharmacotherapy were reviewed. The imaging features were analyzed. Prior CT obtained before rupture were evaluated.

RESULTS

The most common location of ruptured GIST was small bowel with mean size of 12.1 cm. Ruptured GIST commonly showed wall defects, >40 % eccentric necrosis, lobulated shaped, air density in mass, pneumoperitoneum, peritonitis, hemoperitoneum and ascites (p < 0.001-0.030). Twenty-seven of 30 patients with follow up imaging received targeted therapy. During follow-up, thickness of the tumour wall decreased. Increase in size and progression of necrosis were common during targeted therapy (p = 0.017). Newly developed ascites, peritonitis and hemoperitoneum was more common (p < 0.001-0.036).

CONCLUSION

Ruptured GISTs commonly demonstrate large size, >40 % eccentric necrosis, wall defects and lobulated shape. The progression of necrosis with increase in size and decreased wall thickness during targeted therapy may increase the risk of rupture. Rupture should be considered when newly developed peritonitis, hemoperitoneum, or ascites are noted during the follow-up.

KEY POINTS

• Ruptured GISTs demonstrate large size, eccentric necrosis, wall defects, and lobulated shape. • Rupture should be considered when peritonitis or hemoperitoneum/adjacent hematoma newly appears. • Progression of necrosis with increase in size increases the risk of rupture.

Predictive features of CT for risk stratifications in patients with primary gastrointestinal stromal tumour

PURPOSE

To determine the predictive CT imaging features for risk stratifications in patients with primary gastrointestinal stromal tumours (GISTs).

MATERIALS AND METHODS

One hundred and twenty-nine patients with histologically confirmed primary GISTs (diameter >2 cm) were enrolled. CT imaging features were reviewed. Tumour risk stratifications were determined according to the 2008 NIH criteria where GISTs were classified into four categories according to the tumour size, location, mitosis count, and tumour rupture. The association between risk stratifications and CT features was analyzed using univariate analysis, followed by multinomial logistic regression and receiver operating characteristic (ROC) curve analysis.

RESULTS

CT imaging features including tumour margin, size, shape, tumour growth pattern, direct organ invasion, necrosis, enlarged vessels feeding or draining the mass (EVFDM), lymphadenopathy, and contrast enhancement pattern were associated with the risk stratifications, as determined by univariate analysis (P < 0.05). Only lesion size, growth pattern and EVFDM remained independent risk factors in multinomial logistic regression analysis (OR = 3.480-100.384). ROC curve analysis showed that the area under curve of the obtained multinomial logistic regression model was 0.806 (95 % CI: 0.727-0.885).

CONCLUSION

CT features including lesion size, tumour growth pattern, and EVFDM were predictors of the risk stratifications for GIST.

KEY POINTS

• CT features were of predictive value for risk stratification of GISTs. • Tumour size, growth patterns, and EVFDM were risk predictors of GISTs. • Large size, mixed growth pattern, or EVFDM indicated high risk GIST.