Distinguishing Radiation Necrosis Research Articles

Leveraging Quantitative Imaging and Machine Learning to Differentiate Radionecrosis from Disease Recurrence in Patients with Brain Metastases

Radiation necrosis can be difficult to non-invasively discern from tumor progression after stereotactic radiosurgery (SRS). In this work, we investigate the utility of radiomics (computerized features) and machine learning to capture per-voxel lesion heterogeneity on routine MRI scans, to differentiate radionecrosis from tumor recurrence in patients with brain metastases treated with SRS. A retrospective analysis was conducted of patients with brain metastases treated with SRS. Eighty-three lesions (n = 56 intact; n = 27 surgical cavity) from 69 patients were identified with median age 68.8 years (range 40.2 - 91.0), of whom 53.6% were male and 33.3% received prior whole-brain radiotherapy (WBRT). Lesion histology included lung (60.2%), renal cell (15.7%), melanoma (10.8%), breast (9.6%), and other (3.6%). Pathologic confirmation was available in 73.5% of lesions. Both intact and resection cavity lesions were included and individually segmented. Image preprocessing and radiomic feature extraction were done using ANTsPy and open-source software. A total of 210 features were extracted from post-contrast T1-weighted (T1w) and T2/FLAIR MRIs. Highly correlated features were removed. Univariate logistic regression was conducted on the remaining T1w and T2/FLAIR features as well as on clinical variables. Multivariate analysis was implemented with various classifiers (Random Forest, Ridge, Lasso, Support Vector Machine [SVM]) on the top-performing features found on univariate logistic regression. Models were assessed using cross-validation to select the best model by area under ROC curve (AUC). Specificity and sensitivity were calculated. On univariate analysis, the top 10 radiomics features consisted of 6 T1w features and 4 T2/FLAIR features (4 GLCM, 3 first order, 1 GLSZM, 1 GLRLM, and 1 shape feature). Age, gender, disease site, prior WBRT, prior fractionated SRS, planning tumor volume, brain-GTV V12 Gy, and immunotherapy before or after SRS were not predictive (AUC less than 62.0%) on univariate analysis compared to radiomic features. Multivariate analysis of top performing radiomic features on both intact and surgical cavities yielded an AUC of 72.0% (standard deviation [SD] ±8.8%). Multivariate analysis of top features on intact lesions alone improved the AUC to 80.5% (SD ±10.8%), with sensitivity of 77.8%, specificity of 72.4%, and positive likelihood ratio of 2.82 in differentiating radionecrosis from recurrence. Radiomics and machine learning tools may improve diagnostic ability of distinguishing radiation necrosis from tumor recurrence after SRS. Further work is needed to deploy this in a larger multi-institutional cohort and prospectively evaluate its efficacy as a decision-support tool to personalize care in patients with brain metastases.

A Multidimensional Connectomics- and Radiomics-Based Advanced Machine-Learning Framework to Distinguish Radiation Necrosis from True Progression in Brain Metastases.

We introduce tumor connectomics, a novel MRI-based complex graph theory framework that describes the intricate network of relationships within the tumor and surrounding tissue, and combine this with multiparametric radiomics (mpRad) in a machine-learning approach to distinguish radiation necrosis (RN) from true progression (TP). Pathologically confirmed cases of RN vs. TP in brain metastases treated with SRS were included from a single institution. The region of interest was manually segmented as the single largest diameter of the T1 post-contrast (T1C) lesion plus the corresponding area of T2 FLAIR hyperintensity. There were 40 mpRad features and 6 connectomics features extracted, as well as 5 clinical and treatment factors. We developed an Integrated Radiomics Informatics System (IRIS) based on an Isomap support vector machine (IsoSVM) model to distinguish TP from RN using leave-one-out cross-validation. Class imbalance was resolved with differential misclassification weighting during model training using the IRIS. In total, 135 lesions in 110 patients were analyzed, including 43 cases (31.9%) of pathologically proven RN and 92 cases (68.1%) of TP. The top-performing connectomics features were three centrality measures of degree, betweenness, and eigenvector centralities. Combining these with the 10 top-performing mpRad features, an optimized IsoSVM model was able to produce a sensitivity of 0.87, specificity of 0.84, AUC-ROC of 0.89 (95% CI: 0.82-0.94), and AUC-PR of 0.94 (95% CI: 0.87-0.97).

CLINICAL APPLICATION OF SATURATION TRANSFER MRI FOR DIFFERENTIATING TUMOUR PROGRESSION FROM RADIATION NECROSIS IN BRAIN METASTASES

Abstract Stereotactic radiosurgery for the treatment of brain metastases delivers a high dose of radiation with excellent local control, but increases the likelihood of radiation necrosis. As shown in our previous work, saturation transfer MRI, consisting of quantitative magnetization transfer (qMT) and chemical exchange saturation transfer (CEST), is a promising technique for distinguishing radiation necrosis (RN) from tumour progression (TP) in brain metastases. A 3D qMT/CEST acquisition was recently implemented and over 100 patients have been scanned to date. The purpose of this work is to assess the ability of advanced MRI parameters, including qMT and CEST metrics, which are sensitive to macromolecules and metabolism. The specific metrics that were explored included the amide and NOE contributions of the magnetization transfer ratio (MTR), the MTR asymmetry, the apparent exchange-dependent relaxation (AREX), the qMT semi-solid pool fraction and the T1 and T2 relaxation times. For a subset of the patients, dynamic susceptibility contrast (DSC) perfusion images were acquired. Examples of confirmed tumour progression and radiation necrosis cases will be presented, comparing the structural images (pre- and post-contrast T1-weighted and FLAIR images) with parameter maps from qMT and CEST and also the relative cerebral blood flow (rCBF) from DSC perfusion imaging. Interim cohort results will be presented. Approaches for standardizing the parameters across multiple MRI vendors are also explored.

18F-fluciclovine PET/CT to distinguish radiation necrosis from tumor progression for brain metastases treated with radiosurgery: results of a prospective pilot study.

Distinguishing radiation necrosis from tumor progression among patients with brain metastases previously treated with stereotactic radiosurgery represents a common diagnostic challenge. We performed a prospective pilot study to determine whether PET/CT with 18F-fluciclovine, a widely available amino acid PET radiotracer, repurposed intracranially, can accurately diagnose equivocal lesions. Adults with brain metastases previously treated with radiosurgery presenting with a follow-up tumor-protocol MRI brain equivocal for radiation necrosis versus tumor progression underwent an 18F-fluciclovine PET/CT of the brain within 30days. The reference standard for final diagnosis consisted of clinical follow-up until multidisciplinary consensus or tissue confirmation. Of 16 patients imaged from 7/2019 to 11/2020, 15 subjects were evaluable with 20 lesions (radiation necrosis, n = 16; tumor progression, n = 4). Higher SUVmax statistically significantly predicted tumor progression (AUC = 0.875; p = 0.011). Lesion SUVmean (AUC = 0.875; p = 0.018), SUVpeak (AUC = 0.813; p = 0.007), and SUVpeak-to-normal-brain (AUC = 0.859; p = 0.002) also predicted tumor progression, whereas SUVmax-to-normal-brain (p = 0.1) and SUVmean-to-normal-brain (p = 0.5) did not. Qualitative visual scores were significant predictors for readers 1 (AUC = 0.750; p < 0.001) and 3 (AUC = 0.781; p = 0.045), but not for reader 2 (p = 0.3). Visual interpretations were significant predictors for reader 1 (AUC = 0.898; p = 0.012) but not for reader 2 (p = 0.3) or 3 (p = 0.2). In this prospective pilot study of patients with brain metastases previously treated with radiosurgery presenting with a contemporary MRI brain with a lesion equivocal for radiation necrosis versus tumor progression, 18F-fluciclovine PET/CT repurposed intracranially demonstrated encouraging diagnostic accuracy, supporting the pursuit of larger clinical trials which will be necessary to establish diagnostic criteria and performance.

The Role of PET Imaging in the Differential Diagnosis between Radiation Necrosis and Recurrent Disease in Irradiated Adult-Type Diffuse Gliomas: A Systematic Review.

Adult-type diffuse gliomas are treated with a multimodality treatment approach that includes radiotherapy both in the primary setting, and in the case of progressive or recurrent disease. Radiation necrosis represents a major complication of radiotherapy. Recurrent disease and treatment-related changes are often indistinguishable using conventional imaging methods. The present systematic review aims at assessing the diagnostic role of PET imaging using different radiopharmaceuticals in differentiating radiation necrosis and disease relapse in irradiated adult-type diffuse gliomas. We conducted a comprehensive literature search using the PubMed/MEDLINE and EMBASE databases for original research studies of interest. In total, 436 articles were assessed for eligibility. Ten original papers, published between 2014 and 2022, were selected. Four articles focused on [18F]FDG, seven on amino acid tracers ([18F]FET n = 3 and [11C]MET n = 4), one on [11C]CHO, and one on [68Ga]Ga-PSMA. Visual assessment, semi-quantitative methods, and radiomics were applied for image analysis. Furthermore, 2/10 papers were comparative studies investigating different radiopharmaceuticals. The present review, the first one on the topic in light of the new 2021 CNS WHO classification, highlighted the usefulness of PET imaging in distinguishing radiation necrosis and tumour recurrence, but revealed high heterogeneity among studies.

P15.11.A 18F-Fluciclovine PET/CT to distinguish radiation necrosis from tumour progression in brain metastases treated with stereotactic radiosurgery: results of a prospective pilot study

Abstract Background Amino acid PET radiopharmaceutical, 18F-fluciclovine, shows increased uptake in brain tumors relative to normal tissue and may be a useful tool for detecting recurrent brain metastases. Here, we report results from a prospective pilot study evaluating the use of 18F-fluciclovine PET/CT to distinguish radiation necrosis from tumour progression among patients with brain metastases treated with stereotactic radiosurgery (SRS). Material and Methods The primary objective was to estimate the accuracy of 18F-fluciclovine PET/CT in distinguishing radiation necrosis from tumour progression. The trial included adults with brain metastases who underwent SRS and presented with a follow up MRI brain (with DSC MR perfusion) which was equivocal for radiation necrosis versus tumour progression. Within 30 days of equivocal MRI brain, patients underwent an 18F-fluciclovine PET/CT (Siemens mCT) acquired 5-15 min post-injection with images generated by PSF reconstruction. Quantitative metrics for each lesion were documented and lesion to normal brain SUVmean ratios were calculated. The reference standard for diagnosis of radiation necrosis vs tumour progression was clinical follow up with MRI brain every 2-4 months until multidisciplinary consensus or tissue confirmation. Results Of 16 patients enrolled between 7/2019-11/2020, 1 patient died prior to diagnosis, allowing 15 evaluable subjects with 20 lesions. Primary histology was NSCLC in 9 (45%) lesions, breast in 7 (35%), melanoma in 3 (15%), and endometrial in 1 (5%). The final diagnosis was radiation necrosis in 16 (80%) lesions and tumour progression in 4 (20%). SUVmax was a statistically significant predictor of tumour progression (P = 0.011), with higher SUVmax values indicative of tumour progression. The area under the ROC curve was 0.833 (95% CI: 0.590, 1.0). A cutoff of 4.3 provided a sensitivity to identify tumour progression of 1.0 (4/4) and specificity to rule out tumour progression of 0.63 (10/16). SUVmean (P = 0.018), SUVpeak (P = 0.007), and SUVpeak/normal (P = 0.002) also reached statistical significance as predictors of tumour progression, with higher SUVmax values indicative of tumour progression. SUVmax/normal (P = 0.1) and SUVmean/normal (P = 0.5) were not statistically significant. The AUC for SUVmax was not significantly higher than the AUCs for the other quantitative variables (P-values &amp;gt; 0.2). Conclusion In this prospective pilot study, 18F Fluciclovine PET/CT demonstrated promising accuracy to distinguish radiation necrosis from tumour progression among patients with brain metastases previously treated with SRS. Using SUVmax, a cutpoint of 4.3 provided a sensitivity of 1.0 and specificity of 0.63. Confirmatory phase II and III studies are ongoing.

NIMG-02. 18F-FLUCICLOVINE POSITRON EMISSION TOMOGRAPHY TO DISTINGUISH TUMOR PROGRESSION FROM RADIATION NECROSIS FOLLOWING STEREOTACTIC RADIOSURGERY FOR BRAIN METASTASIS: A QUALITATIVE ANALYSIS

Abstract PURPOSE/OBJECTIVE(S) To assess the ability of 18F-Fluciclovine PET/CT to distinguish radiation necrosis (RN) from tumor progression (TP) among patients with brain metastases (BM) treated with stereotactic radiosurgery (SRS) in a prospective pilot study. MATERIALS/METHODS Adults with post-SRS BM presenting with follow-up brain MRI equivocal for RN versus TP underwent 18F-Fluciclovine PET/CT within 30 days of equivocal MRI. PET images were reconstructed using a point-spread-function algorithm. Three physician reviewers independently performed qualitative analyses of each lesion using a three-point visual score relative to PET-avidity of blood pool and parotid. Quantitative metrics for each lesion were documented. Reference standard was clinical follow-up with brain MRI until tumor board consensus or tissue confirmation. Nonparametric estimates of area under the receiver operating characteristic curve (AUC) for clustered data were estimated, with diagnostic performance based on visual score. RESULTS In 15 subjects with 20 lesions, final diagnosis was RN in 16 (80%) lesions and TP in 4 (20%). Visual score significantly correlated with final diagnosis (AUC range 0.836-0.906 [p≤0.037]). A threshold score of 2 (lesion 18F-fluciclovine uptake above blood pool to parotid) and higher produced sensitivities and specificities of 75-100% and 38-56% respectively among the reviewer majority. Conversely, a threshold of 3 (uptake higher than parotid) produced sensitivities and specificities of 50-75% and 100% respectively. CONCLUSION In this prospective pilot, basic visual analysis of 18F-Fluciclovine PET/CT provided high sensitivity and specificity in detection of TP in post-SRS BM based on different threshold scores, suggesting room for visual threshold optimization. A low TP event rate limited the ability to estimate sensitivity/specificity and to perform combined qualitative/quantitative analyses. Further study to refine interpretation criteria is ongoing.

18F-Fluciclovine PET/CT to Distinguish Radiation Necrosis From Tumor Progression in Brain Metastases Treated With Stereotactic Radiosurgery: Results of a Prospective Pilot Study

Purpose/Objective(s)To report results from a prospective pilot study evaluating the use of 18F Fluciclovine PET CT to distinguish radiation necrosis (RN) from tumor progression (TP) among patients (pts) with brain metastases (BM) treated with stereotactic radiosurgery (SRS).Materials/MethodsThe primary objective was to estimate the accuracy of 18F Fluciclovine PET CT in distinguishing RN from TP. The trial included adults with BM who underwent SRS and presented with a follow up MRI brain (with DSC MR perfusion) which was equivocal for RN versus TP. Within 30 days of equivocal MRI brain, pts underwent an 18F Fluciclovine PET CT (Siemens mCT) acquired 5-15 min post-injection with images generated by PSF reconstruction. Quantitative metrics for each lesion were documented and lesion to normal brain SUVmean ratios were calculated. The reference standard for diagnosis of RN vs TP was clinical follow up with MRI brain every 2-4 months until multidisciplinary consensus or tissue confirmation.ResultsOf 16 pts enrolled between 7/2019-11/2020, 1 pt died prior to diagnosis, allowing 15 evaluable subjects with 20 lesions. Primary histology was NSCLC in 9 (45%) lesions, breast in 7 (35%), melanoma in 3 (15%), and endometrial in 1 (5%). The final diagnosis was RN in 16 (80%) lesions and TP in 4 (20%). SUVmax was a statistically significant predictor of TP (P = 0.011), with higher SUVmax values indicative of TP. The area under the ROC curve was 0.833 (95% CI: 0.590, 1.0). A cutoff of 4.3 provided a sensitivity to identify TP of 1.0 (4/4) and specificity to rule out TP of 0.63 (10/16). Additional quantitative metrics are summarized in Table 1. SUVmean (P = 0.018), SUVpeak (P = 0.007), and SUVpeak/normal (P = 0.002) reached statistical significance as predictors of TP. SUVmax/normal (P = 0.1) and SUVmean/normal (P = 0.5) were not statistically significant. The AUC for SUVmax was not significantly higher than the AUCs for the other quantitative variables (P-values > 0.2).ConclusionIn this prospective pilot study, 18F Fluciclovine PET CT demonstrated promising accuracy to distinguish RN from TP among pts with BM previously treated with SRS. Using SUVmax, a cutpoint of 4.3 provided a sensitivity of 1.0 and specificity of 0.63. Confirmatory phase II and III studies are ongoing. To report results from a prospective pilot study evaluating the use of 18F Fluciclovine PET CT to distinguish radiation necrosis (RN) from tumor progression (TP) among patients (pts) with brain metastases (BM) treated with stereotactic radiosurgery (SRS). The primary objective was to estimate the accuracy of 18F Fluciclovine PET CT in distinguishing RN from TP. The trial included adults with BM who underwent SRS and presented with a follow up MRI brain (with DSC MR perfusion) which was equivocal for RN versus TP. Within 30 days of equivocal MRI brain, pts underwent an 18F Fluciclovine PET CT (Siemens mCT) acquired 5-15 min post-injection with images generated by PSF reconstruction. Quantitative metrics for each lesion were documented and lesion to normal brain SUVmean ratios were calculated. The reference standard for diagnosis of RN vs TP was clinical follow up with MRI brain every 2-4 months until multidisciplinary consensus or tissue confirmation. Of 16 pts enrolled between 7/2019-11/2020, 1 pt died prior to diagnosis, allowing 15 evaluable subjects with 20 lesions. Primary histology was NSCLC in 9 (45%) lesions, breast in 7 (35%), melanoma in 3 (15%), and endometrial in 1 (5%). The final diagnosis was RN in 16 (80%) lesions and TP in 4 (20%). SUVmax was a statistically significant predictor of TP (P = 0.011), with higher SUVmax values indicative of TP. The area under the ROC curve was 0.833 (95% CI: 0.590, 1.0). A cutoff of 4.3 provided a sensitivity to identify TP of 1.0 (4/4) and specificity to rule out TP of 0.63 (10/16). Additional quantitative metrics are summarized in Table 1. SUVmean (P = 0.018), SUVpeak (P = 0.007), and SUVpeak/normal (P = 0.002) reached statistical significance as predictors of TP. SUVmax/normal (P = 0.1) and SUVmean/normal (P = 0.5) were not statistically significant. The AUC for SUVmax was not significantly higher than the AUCs for the other quantitative variables (P-values > 0.2). In this prospective pilot study, 18F Fluciclovine PET CT demonstrated promising accuracy to distinguish RN from TP among pts with BM previously treated with SRS. Using SUVmax, a cutpoint of 4.3 provided a sensitivity of 1.0 and specificity of 0.63. Confirmatory phase II and III studies are ongoing.

Clinical assessment of a biophysical model for distinguishing tumor progression from radiation necrosis

The efficacy of an imaging-driven mechanistic biophysical model of tumor growth for distinguishing radiation necrosis from tumor progression in patients with enhancing lesions following stereotactic radiosurgery (SRS) for brain metastasis is validated. We retrospectively assessed the model using 73 patients with 78 lesions and histologically confirmed radiation necrosis or tumor progression. Postcontrast T1-weighted MRI images were used to extract parameters for a mechanistic reaction-diffusion logistic growth model mechanically coupled to the surrounding tissue. The resulting model was then used to estimate mechanical stress fields, which were then compared with edema visualized on FLAIR imaging using DICE similarity coefficients. DICE, model, and standard radiographic morphometric analysis parameters were evaluated using a receiver operating characteristic (ROC) curve for prediction of radiation necrosis or tumor progression. Multivariate logistic regression models were then constructed using mechanistic model parameters or advanced radiomic features. An independent validation was performed to evaluate predictive performance. Tumor cell proliferation rate resulted in ROC AUC=0.86, 95% CI: 0.76-0.95, P<0.0001, 74% sensitivity and 95% specificity) and DICE similarity coefficient associated with high stresses demonstrated an ROC AUC=0.93, 95% CI: 0.86-0.99, P<0.0001, 81% sensitivity and 95% specificity. In a multivariate logistic regression model using an independent validation dataset, mechanistic modeling parameters had an ROC AUC of 0.95, with 94% sensitivity and 96% specificity. Imaging-driven biophysical modeling of tumor growth represents a novel method for accurately predicting clinically significant tumor behavior.

BIMG-19. 18F-FLUCICLOVINE PET/CT TO DISTINGUISH RADIATION NECROSIS FROM TUMOR PROGRESSION IN BRAIN METASTASES TREATED WITH STEREOTACTIC RADIOSURGERY: A PROSPECTIVE PILOT STUDY

PURPOSETo estimate the accuracy of 18F-Fluciclovine PET/CT in distinguishing radiation necrosis (RN) from tumor progression (TP) among patients with brain metastases (BM) having undergone prior stereotactic radiosurgery (SRS) who presented with a follow-up MRI brain (with DSC-MR perfusion) which was equivocal for RN versus TP.METHODSWithin 30 days of equivocal MRI brain, subjects underwent 18F-Fluciclovine PET/CT on a Siemens Biograph mCT scanner with a 10 mCi bolus dose immediately prior to PET. Data were collected in list mode for 25 minutes post-injection and were reconstructed as a static image of data 10–25 minutes post-injection, and as a dynamic series of four 5-minute frames between 5–25 minutes post-injection. Quantitative metrics for each lesion were documented including SUVmax, SUVmean, SUVpeak, and normal brain SUVmean. Lesion to normal brain ratios were calculated. The reference standard was clinical follow-up with MRI brain (with DSC-MR perfusion) every 2–4 months until multidisciplinary consensus (or tissue confirmation) for diagnosis of RN versus TP.RESULTSFrom 7/2019-11/2020, 16 of 16 planned subjects enrolled and underwent 18F-Fluciclovine PET/CT for evaluation of 21 brain lesions. Primary histology included NSCLC (n=7), breast (n=5), melanoma (n=3), and endometrial (n=1). Ranges of quantitative metrics were: SUVmax, 2.18–12.1; SUVmean, 1.16–7.37; SUVpeak, 1.06–5.14; normal brain SUVmean, 0.19–0.50; SUVmax/normal ratio, 7.5–45.4; SUVmean/normal ratio, 4.2–26.3; and SUVpeak/normal ratio, 3.9–26.4. Among the patients 10 patients with 12 lesions who completed follow up, estimates of the area under the receiver operating characteristic curve for SUVmax, SUVmean, and SUVpeak were: 0.93, 0.93, and 0.82, respectively.CONCLUSIONIn this population, 18F-Fluciclovine PET/CT favorably produces a wide range of lesion quantitative metric values, low uptake in the normal brain, and promising accuracy to distinguish RN from TP. Completion of follow-up for all patients is required. Phase II and III studies are under development.

Differentiating Radiation Necrosis from Brain Tumor Using Hyperpolarized Carbon-13 MR Metabolic Imaging.

Differentiation between radiation-induced necrosis and tumor recurrence is crucial to determine proper management strategies but continues to be one of the central challenges in neuro-oncology. We hypothesized that hyperpolarized 13C MRI, a unique technique to measure real-time in vivo metabolism, would distinguish radiation necrosis from tumor on the basis of cell-intrinsic metabolic differences. The purpose of this study was to explore the feasibility of using hyperpolarized [1-13C]pyruvate for differentiating radiation necrosis from brain tumors. Radiation necrosis was initiated by employing a CT-guided 80-Gy single-dose irradiation of a half cerebrum in mice (n = 7). Intracerebral tumor was modeled with two orthotopic mouse models: GL261 glioma (n = 6) and Lewis lung carcinoma (LLC) metastasis (n = 7). 13C 3D MR spectroscopic imaging data were acquired following hyperpolarized [1-13C]pyruvate injection approximately 89 and 14days after treatment for irradiated and tumor-bearing mice, respectively. The ratio of lactate to pyruvate (Lac/Pyr), normalized lactate, and pyruvate in contrast-enhancing lesion was compared between the radiation-induced necrosis and brain tumors. Histopathological analysis was performed from resected brains. Conventional MRI exhibited typical radiographic features of radiation necrosis and brain tumor with large areas of contrast enhancement and T2 hyperintensity in all animals. Normalized lactate in radiation necrosis (0.10) was significantly lower than that in glioma (0.26, P =.004) and LLC metastatic tissue (0.25, P =.00007). Similarly, Lac/Pyr in radiation necrosis (0.18) was significantly lower than that in glioma (0.55, P =.00008) and LLC metastasis (0.46, P =.000008). These results were consistent with histological findings where tumor-bearing brains were highly cellular, while irradiated brains exhibited pathological markers consistent with reparative changes from radiation necrosis. Hyperpolarized 13C MR metabolic imaging of pyruvate is a noninvasive imaging method that differentiates between radiation necrosis and brain tumors, providing a groundwork for further clinical investigation and translation for the improved management of patients with brain tumors.

Multiparametric radiomic tissue signature and machine learning for distinguishing radiation necrosis from tumor progression after stereotactic radiosurgery.

BackgroundStereotactic radiosurgery (SRS) may cause radiation necrosis (RN) that is difficult to distinguish from tumor progression (TP) by conventional MRI. We hypothesize that MRI-based multiparametric radiomics (mpRad) and machine learning (ML) can differentiate TP from RN in a multi-institutional cohort.MethodsPatients with growing brain metastases after SRS at 2 institutions underwent surgery, and RN or TP were confirmed by histopathology. A radiomic tissue signature (RTS) was selected from mpRad, as well as single T1 post-contrast (T1c) and T2 fluid-attenuated inversion recovery (T2-FLAIR) radiomic features. Feature selection and supervised ML were performed in a randomly selected training cohort (N = 95) and validated in the remaining cases (N = 40) using surgical pathology as the gold standard.ResultsOne hundred and thirty-five discrete lesions (37 RN, 98 TP) from 109 patients were included. Radiographic diagnoses by an experienced neuroradiologist were concordant with histopathology in 67% of cases (sensitivity 69%, specificity 59% for TP). Radiomic analysis indicated institutional origin as a significant confounding factor for diagnosis. A random forest model incorporating 1 mpRad, 4 T1c, and 4 T2-FLAIR features had an AUC of 0.77 (95% confidence interval [CI]: 0.66–0.88), sensitivity of 67% and specificity of 86% in the training cohort, and AUC of 0.71 (95% CI: 0.51–0.91), sensitivity of 52% and specificity of 90% in the validation cohort.ConclusionsMRI-based mpRad and ML can distinguish TP from RN with high specificity, which may facilitate the triage of patients with growing brain metastases after SRS for repeat radiation versus surgical intervention.

18F]-Fluciclovine PET discrimination of recurrent intracranial metastatic disease from radiation necrosis

BackgroundStereotactic radiosurgery (SRS) is often the primary treatment modality for patients with intracranial metastatic disease. Despite advances in magnetic resonance imaging, including use of perfusion and diffusion sequences and molecular imaging, distinguishing radiation necrosis from progressive tumor remains a diagnostic and clinical challenge. We investigated the sensitivity and specificity of 18F-fluciclovine PET to accurately distinguish radiation necrosis from recurrent intracranial metastatic disease in patients who had previously undergone SRS.MethodsFluciclovine PET imaging was performed in 8 patients with a total of 15 lesions that had previously undergone SRS and had subsequent MRI and clinical features suspicious for recurrent disease. The SUVmax of each lesion and the contralateral normal brain parenchyma were summated and evaluated at four different time points (5 min, 10 min, 30 min, and 55 min). Lesions were characterized as either recurrent disease (11 of 15 lesions) or radiation necrosis (4 of 15 lesions) and confirmed with histopathological correlation (7 lesions) or through serial MRI studies (8 lesions).ResultsTime activity curve analysis found statistically greater radiotracer accumulation for all lesions, including radiation necrosis, when compared to contralateral normal brain. While the mean and median SUVmax for recurrent disease were statistically greater than those of radiation necrosis at all time points, the difference was more significant at the earlier time points (p = 0.004 at 5 min–0.025 at 55 min). Using a SUVmax threshold of ≥ 1.3, fluciclovine PET demonstrated a 100% accuracy in distinguishing recurrent disease from radiation necrosis up to 30 min after injection and an accuracy of 87% (sensitivity = 0.91, specificity = 0.75) at the last time point of 55 min. However, tumor-to-background ratios (TBRmax) were not significantly different between recurrent disease and radiation necrosis at any time point due to variable levels of fluciclovine uptake in the background brain parenchyma.ConclusionsFluciclovine PET may play an important role in distinguishing active intracranial metastatic lesions from radiation necrosis in patients previously treated with SRS but needs to be validated in larger studies.

RADT-14. COMPARISON OF MACHINE LEARNING ALGORITHMS IN DISTINGUISHING RADIATION NECROSIS FROM PROGRESSION OF BRAIN METASTASES TREATED WITH STEREOTACTIC RADIOSURGERY (SRS)

Abstract PURPOSE To test the effectiveness of machine learning algorithms in distinguishing radiation necrosis (RN) from tumor progression (TP) using MRI radiomic features. METHODS Brain metastases were treated with SRS to a median dose of 18Gy. Lesions that showed evidence of progression on follow-up MRI were sampled surgically, and diagnoses confirmed by histopathology. Cases from 2 institutions were combined and randomly assigned for training (70%) and testing (30%). T1 post-contrast (T1c) and T2 fluid attenuated inversion recovery (T2 FLAIR) MRI were used for radiomic feature extraction (50 features each). Three subsets of radiomic features were obtained and tested: Signature #1 included 10 previously published features that correlated with diagnosis on T test; signature #2 and #3 included 5 and 12 features obtained through recursive elimination using random forest (RF) and support vector machine (SVM), respectively. Supervised machine learning models were trained using RF, SVM (radial kernel) and regularized discriminant analysis (RDA) algorithms based on all three radiomics signatures. Receiver operator characteristics (ROC) were compared between signatures and algorithms. RESULTS A total of 135 individual lesions (37 RN and 98 TP) were included. Signature #3 demonstrated the highest area under the curve in the training set (average AUC=0.98, vs 0.95 and 0.92 for signature #1 and #2), as well as the testing set (average AUC=0.83, vs 0.74 and 0.79 for signature #1 and #2). RF and SVM demonstrated similar performance in both training (average AUC 0.99-1) and testing datasets (average AUC 0.79-0.80) among all three signatures. Both RF and SVM were superior to RDA in performance (average training AUC 0.83, testing AUC 0.77). The greatest sensitivity (83%) and specificity (100%) in the testing set were achieved using signature #3 and SVM. CONCLUSION RF and SVM are effective in distinguishing RN from TP in a multi-institution dataset using radiomic signatures.

NIMG-02. 18F-FLUCICLOVINE PET/CT TO DISTINGUISH RADIATION NECROSIS FROM TUMOR PROGRESSION IN BRAIN METASTASES TREATED WITH STEREOTACTIC RADIOSURGERY

Abstract PURPOSE/OBJECTIVE(S) To report descriptive updates of an ongoing pilot trial assessing whether 18F-Fluciclovine PET/CT, a widely available amino-acid radiotracer, is useful to distinguish radiation necrosis (RN) from tumor progression (TP) among patients with brain metastases. MATERIALS/METHODS The primary objective is to estimate the accuracy of 18F-Fluciclovine PET/CT in distinguishing RN from TP. We included adults with brain metastases who underwent prior stereotactic radiosurgery and presented with a follow-up MRI brain (with DSC-MR perfusion) which was equivocal for RN versus TP. Within 30 days of equivocal MRI, patients underwent 18F-Fluciclovine PET/CT on a Siemens Biograph mCT scanner with a 10 mCi bolus dose immediately prior to PET. PET data were collected in list-mode for 25 mins post-injection and were reconstructed as a static image of data 10-25 mins post-injection, and as a dynamic series of four 5-min frames between 5-25 mins post-injection. Quantitative metrics for each lesion were documented. Lesion to normal brain ratios were calculated. The reference standard was clinical follow-up with MRI brain (with DSC-MR perfusion) every 2-4 months until multidisciplinary consensus (or tissue confirmation) for diagnosis of RN versus TP. RESULTS From 7/2019-6/2020, 12 of 16 planned subjects with 17 lesions underwent 18F-Fluciclovine PET/CT. Primary histology was non-small cell lung cancer in 5 patients, breast in 4, melanoma in 2, and endometrial in 1. Among all 17 lesions, ranges of quantitative metrics were: SUVmax, 2.18-12.10; SUVmean, 1.16-7.37; SUVpeak, 1.06-4.45; normal brain SUVmean, 0.19-0.44; SUVmax/normal ratio, 7.50-45.40; SUVmean/normal ratio, 4.20-26.30; and SUVpeak/normal ratio, 3.90-26.40. Follow-up was completed for 5 patients (6 lesions). No adverse events have occurred. CONCLUSION In this population, 18F-Fluciclovine produces a wide range of lesion quantitative metric values and uniformly low uptake in normal brain, which may allow accurate discrimination. Ongoing additional accrual and follow up is required.

External Validation Of A Radiomics-Based Machine Learning Model For Distinguishing Radiation Necrosis From Progression Of Brain Metastases Treated With Stereotactic Radiosurgery

Radiation necrosis (RN) is common and potentially debilitating after stereotactic radiosurgery (SRS) for brain metastases (BM). The goal of this study is to validate a previously reported radiomics signature for distinguishing RN from true progression (TP) using an independent image set. Patients with BM were treated with Gammaknife SRS at Wake Forest University between 2004 and 2012 (WF dataset). Those who developed radiographic evidence of progression in the treated lesion and subsequently had pathologic confirmation were included. The lesions were identified on T1 post-contrast (T1c) and T2 fluid attenuated inversion recovery (T2 FLAIR) MRI. 51 T1c and 51 T2 FLAIR radiomics features were extracted using a software developed in house. An IsoSVM machine learning classifier model trained using our institutional data (JH dataset) was tested using WF dataset. Fifty-three lesions from 43 patients were included in this study, of which all had T1c and 33 had T2 FLAIR images available. Seven lesions were pure RN pathologically, while the TP group included 37 pure tumor and 9 mixed tumor and necrosis. Twenty-two patients (51%) underwent whole brain radiotherapy with a median dose of 35Gy. Median SRS marginal dose was 18Gy (range 10-22Gy). There was no statistically significant difference among baseline characteristics between RN and TP. Of the 6 T1c radiomic features from the JH dataset, 4 demonstrated the same trend in the WF dataset (neighborhood grey tone difference matrix (NGTDM) texture strength, NGTDM coarseness, grey level run length matrix (GLRLM) grey level nonuniformity, and GLRLM run percentage). NGTDM texture strength (mean RN 161.13 vs TP 70.86, p = 0.06), and NGTDM coarseness (mean RN 0.03 vs TP 0.02, p = 0.07) approached significance. Of the 4 FLAIR features, 3 showed similar trend (kurtosis, minimum, NGTDM texture strength). The original IsoSVM model trained with all 6 T1c and 4 FLAIR features showed sensitivity of 65%, specificity of 87%, and AUC of 0.81. Due to the limited number of cases with FLAIR available, we re-trained the model with JH dataset using only the 4 T1c features showing the same trend between the two datasets, resulting in sensitivity and specificity of 80% and 73%, with AUC of 0.8. This model when tested on the WF dataset showed sensitivity and specificity of 71% and 52%, with AUC of 0.62. When the model was re-trained with the 4 T1c and 3 FLAIR features trending similarly between the two datasets, the sensitivity and specificity were 50% and 100%, and the AUC improved to 0.84. Radiomics profiling of post-SRS T1c and T2 FLAIR MRI showed promising trends between RN and TP in 4 T1c features across two independent datasets. Radiomics-based IsoSVM model can distinguish RN vs TP in individual patients. Statistical significance and model performance were inferior in the validation (WF) dataset due to smaller sample size and class imbalance with fewer RN cases. Further studies with larger datasets are needed to optimize and validate the model.

18F-Fluciclovine PET/CT to Distinguish Radiation Necrosis from Tumor Progression in Brain Metastases Treated with Stereotactic Radiosurgery

To assess whether 18F Fluciclovine PET CT, a widely available FDA approved amino acid radiotracer for suspected prostate cancer recurrence, is useful to distinguish radiation necrosis (RN) from tumor progression (TP) among patients with brain metastases treated with stereotactic radiosurgery (SRS). We report descriptive updates of an ongoing pilot clinical trial. The primary objective is to estimate the accuracy of 18F Fluciclovine PET CT in distinguishing RN from TP. We included adult patients with brain metastases who underwent prior SRS and presented with a follow up MRI brain (with DSC MR perfusion) which was equivocal for RN versus TP. Based on equivocal MRI and prior to PET, patients were grouped as equivocal but suspicious for either RN or TP, or truly equivocal, by the study neuroradiologist. Within 30 days of equivocal MRI brain, patients underwent 18F Fluciclovine PET CT on a Siemens Biograph mCT scanner with a 10 mCi bolus dose immediately prior to PET. PET data were collected in list mode for 25 mins post-injection and were reconstructed as a static image of data 10-25 mins post-injection, and as a dynamic series of four 5-min frames between 5 -25 mins post-injection. Quantitative metrics for each lesion were documented, including SUV max, SUV mean, SUV peak, and normal brain SUV mean. Lesion to normal brain ratios were calculated. The reference standard was clinical follow up with MRI brain (with DSC MR perfusion) every 2-4 months until multidisciplinary consensus (or tissue confirmation) for diagnosis of RN versus TP. From 7/2019-2/2020, 10 of 16 planned subjects enrolled, 2 patients withdrew consent, and 8 patients with 15 lesions underwent 18F Fluciclovine PET CT. Of the 8 imaged patients, primary histology was breast in 3, lung adenocarcinoma in 3, and melanoma in 2 patients. Among all 15 imaged lesions, ranges of quantitative metrics were: SUV max, 2.05-12.10; SUV mean, 1.14-7.37; SUV peak, 0.97-4.45; normal brain SUV mean, 0.19-0.35; SUV max/normal ratio, 6.9-43.2; SUV mean/normal ratio, 3.8-17.7; and SUV peak/normal ratio, 3.1-26.4. Follow up was completed for 4 pts (5 lesions) as described in the Table. No adverse events have been reported. In this population, 18F Fluciclovine produces a wide range of lesion quantitative metric values and low uptake in normal brain, which may allow accurate discrimination. Ongoing accrual and follow up is required to determine if 18F Fluciclovine is useful to distinguish RN from TP among patients with brain metastases treated with SRS.Abstract 3587; TableSubjectPrimaryFinal diagnosis (method)SUV maxSUV meanSUV peakNormal brain SUV meanSUV max/ normal ratioSUV mean/ normal ratioSUV peak/ normal ratio1MelanomaRN (biopsy)3.472.052.000.2712.77.57.32Lung AdenocarcinomaRN (biopsy)3.772.282.320.3211.67.07.23BreastTP (clinical follow up)7.784.432.700.3522.412.87.84BreastTP (clinical follow up)8.784.484.450.3525.312.912.84BreastTP (clinical follow up)12.17.372.560.2843.29.226.4 Open table in a new tab

59. A RADIOMICS-BASED MACHINE LEARNING MODEL FOR DISTINGUISHING RADIATION NECROSIS FROM PROGRESSION OF BRAIN METASTASES TREATED WITH STEREOTACTIC RADIOSURGERY (SRS)

PURPOSEThis study aims to test whether MRI radiomic signatures can distinguish radiation necrosis (RN) from tumor progression (TP) in a multi-institution dataset using machine learning.METHODSBrain metastases treated with SRS were followed by serial MRI, and those showing evidence of RN or TP underwent pathologic confirmation. Radiomic features were extracted from T1 post-contrast (T1c) and T2 fluid attenuated inversion recovery (T2 FLAIR) MRI. High dimensional radiomic feature space was visualized in a two-dimensional space using t-distributed stochastic neighbor embedding (t-SNE). Cases from 2 institutions were combined and randomly assigned to training (2/3) and testing (1/3) cohorts. Backward elimination was used for feature selection, followed by random forest algorithm for predictive modeling.RESULTSA total of 135 individual lesions (37 RN and 98 TP) were included. The majority (72.6%) received single-fraction SRS to a median dose of 18Gy. Clear clustering of cases around the institutional origin was observed on t-SNE analysis. 21 T1c and 4 FLAIR features were excluded from subsequent modeling due to significant correlation with the institutional origin. Backward elimination yielded 6 T1c and 6 FLAIR features used for model construction. A random forest model based on the 6 FLAIR features (cluster shade, neighborhood gray tone difference matrix (NGTDM) coarseness, NGTDM texture strength, run length nonuniformity, run percentage, and short run high gray-level emphasis) achieved sensitivity of 76% and specificity of 70% on the training cohort (AUC 0.74, 95% CI 0.60–0.88), and sensitivity of 67% and specificity of 83% on the testing cohort (AUC 0.75, 95% CI 0.59–0.93). Addition of the T1c features resulted in overfitting of the training cohort (AUC 1.00), but did not improve model performance on the testing cohort (AUC 0.69, 95% CI 0.51–0.87).CONCLUSIONMRI radiomics based machine learning can distinguish RN from TP after brain SRS in a heterogeneous image dataset.

Distinguishing radiation necrosis from tumor recurrence

Distinguishing radiation necrosis from tumor recurrence

Accuracy of 18F-FDOPA Positron Emission Tomography and 18F-FET Positron Emission Tomography for Differentiating Radiation Necrosis from Brain Tumor Recurrence

Distinguishing radiation necrosis from brain tumor recurrence remains challenging. We performed a meta-analysis to assess the diagnostic accuracy of 2 different amino acid tracers used in positron emission tomography/computed tomography scans: 18F-FDOPA (6-[18F]-fluoro-L-3,4-dihydroxyphenylalanine) and 18F-FET (O-(2-18F-fluoroethyl)-L-tyrosine). We searched for studies in 3 databases: PubMed, Embase, and Chinese Biomedical databases. The data were extracted from eligible studies and then processed with heterogeneity test, threshold effect test, and calculations of sensitivity, specificity, and area under the summary receiver operating characteristic curve. Meta-regression and subgroup analyses were performed to explore the source of heterogeneity. A total of 48 studies (18F-FDOPA, n= 21; 18F-FET, n= 27) were included. Quantitative synthesis determined pooled weight values in the 18F-FDOPA and 18F-FET groups: sensitivity, 0.85 versus 0.82; specificity, 0.77 versus 0.80; diagnostic odds ratio, 21.7 versus 23.03; area under the curve (AUC) values, 0.8771 versus 0.8976 (P= 0.46). Moreover, the type of tumor was identified as the possible source of the significant heterogeneity (I2= 52%; P= 0.003) found in the 18F-FDOPA group. In meta-regression and subgroup analyses, 18F-FDOPA showed better diagnostic accuracy in patients with glioma compared with patients with brain metastases (AUC values, 0.9691 vs. 0.837; P<0.01). 18F-FDOPA also showed a significant advantage in the diagnosis of glioma recurrence compared with 18F-FET (AUC values, 0.9691 vs. 0.9124; P= 0.015). Both 18F-FDOPA and 18F-FET exhibit moderate overall accuracy in diagnosing brain tumor recurrence from radiation necrosis. However, 18F-FDOPA is more adept at diagnosing glioma recurrence compared with brain metastases, and it is more effective than 18F-FET in diagnosing glioma recurrence.