Radiogenomic Approach Research Articles

Abstract 3510: A radiogenomic approach for triple-negative breast cancer risk stratification

Abstract Background: Triple-negative breast cancer (TNBC) is an aggressive disease that accounts for 15-20% of all breast cancers. Expressions of ER, PR and HER2 receptors are lacking in this disease, and thus targeted therapies are not effective. TNBC has a shorter relapse-free survival, higher metastasis rate and decreased overall survival compared with other breast cancers. However, when undergoing standard treatment, some patients respond well, while others have poor outcome, suggesting TNBC heterogeneity. Early stratification of patients with long versus short survival could identify the subgroup of patients who would not benefit from exposure to toxicity of chemotherapy treatment. Here, we developed a non-invasive radiogenomic approach for TNBC risk stratification. Methods: A transcriptomic-based prognostic gene signature was previously developed using the TCGA-BRCA cohort (n=860). Briefly, LASSO Cox regression model analysis with the ‘glmnet’ R package was used to identify the transcriptomic signature gene-set consisting of 50 genes. We tested this signature to prognosticate overall survival in a Stanford cohort (n=63) and a previously published SCANB cohort (n=604). The patients were stratified into high- and low-risk groups based on the median risk-score. Next, we developed a machine learning model that identified a radiomic feature set to predict the prognostic transcriptomic risk-groups. Radiomic features were extracted from pre-treatment breast MRI. Radiomics features were extracted using PyRadiomics. The model utilized Decision Tree Classifier and LeaveOneOut method was used for cross-validation. Results: The transcriptomic signature low-risk group was significantly associated with improved overall survival in the two TNBC cohorts, with hazard ratios of 0.11 [95% CI: 0.01-0.88] for the Stanford cohort and 0.71 [95% CI: 0.52-0.97] for the SCANB cohort (log-rank p-values p=0.012 and p=0.032, respectively). Including this transcriptomic signature in a multivariate analysis, which adjusted for clinical features (patient age, grade, stage and Ki67%), the transcriptomic prognostic signature remained a significant prognostic factor (p<0.05). The radiomic feature set (consisting of 20 features) predicted the high- and low-risk transcriptomic groups with a mean accuracy of 72.2% and a mean AUROC of 71%. The precision, F1 and recall scores were 67%, 74% and 82%, respectively. In an independent dataset consisting of 116 Stanford TNBC patients, we used this model to predict risk groups based on the MRI radiomics features, and evaluated the prognostic effects of predicted risk groups. The overall survival of the predicted high-risk group was significantly poorer than the predicted low-risk group (p=0.013). Conclusions: We present a prognostic model that can non-invasively stratify TNBC patients for low versus high mortality risk using radiomic features derived from pre-treatment patient MRI data. Citation Format: Humaira Noor, Yuanning Zheng, Adam Mantz, Ryle Zhou, Andrew Kozlov, Wendy B. DeMartini, Shu-tian Chen, Satoko Okamoto, Debra Ikeda, Sarah Mattonen, Sandy Napel, Melinda L. Telli, George Sledge, Allison Kurian, Mina Satoyoshi, Olivier Gevaert, Haruka Itakura. A radiogenomic approach for triple-negative breast cancer risk stratification [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3510.

Radiogenomic approach to glial tumors imaging under conditions of initial diagnostic measures: adaptation principles development

Radiogenomic approach to glial tumors imaging under conditions of initial diagnostic measures: adaptation principles development

Predicting response to immunotherapy in non-small cell lung cancer- from bench to bedside.

Immune checkpoint inhibitor (ICI) therapy is first-line treatment for many advanced non-small cell lung cancer (aNSCLC) patients. Predicting response could help guide selection of intensified or alternative anti-cancer regimens. We hypothesized that radiomics and laboratory variables predictive of ICI response in a murine model would also predict response in aNSCLC patients. Fifteen mice with lung carcinoma tumors implanted in bilateral flanks received ICI. Pre-ICI laboratory and computed tomography (CT) data were evaluated for association with systemic ICI response. Baseline clinical and CT data for 117 aNSCLC patients treated with nivolumab were correlated with overall survival (OS). Models for predicting treatment response were created and subjected to internal cross-validation, with the human model further tested on 42 aNSCLC patients who received pembrolizumab. Models incorporating baseline NLR and identical radiomics (surface-to-mass ratio, average Gray, and 2D kurtosis) predicted ICI response in mice and OS in humans with AUCs of 0.91 and 0.75, respectively. The human model successfully sorted pembrolizumab patients by longer vs. shorter predicted OS (median 35 months vs. 6 months, p=0.026 by log-rank). This study advances precision oncology by non-invasively classifying aNSCLC patients according to ICI response using pre-treatment data only. Interestingly, identical radiomics features and NLR correlated with outcomes in the preclinical study and with ICI response in 2 independent patient cohorts, suggesting translatability of the findings. Future directions include using a radiogenomic approach to optimize modeling of ICI response.

Radiogenomic Features of GIMAP Family Genes in Clear Cell Renal Cell Carcinoma: An Observational Study on CT Images.

GTPases of immunity-associated proteins (GIMAP) genes include seven functional genes and a pseudogene. Most of the GIMAPs have a role in the maintenance and development of lymphocytes. GIMAPs could inhibit the development of tumors by increasing the amount and antitumor activity of infiltrating immunocytes. Knowledge of key factors that affect the tumor immune microenvironment for predicting the efficacy of immunotherapy and establishing new targets in ccRCC is of great importance. A computed tomography (CT)-based radiogenomic approach was used to detect the imaging phenotypic features of GIMAP family gene expression in ccRCC. In this retrospective study we enrolled 193 ccRCC patients divided into two groups: ccRCC patients with GIMAP expression (n = 52) and ccRCC patients without GIMAP expression (n = 141). Several imaging features were evaluated on preoperative CT scan. A statistically significant correlation was found with absence of endophytic growth pattern (p = 0.049), tumor infiltration (p = 0.005), advanced age (p = 0.018), and high Fuhrman grade (p = 0.024). This study demonstrates CT imaging features of GIMAP expression in ccRCC. These results could allow the collection of data on GIMAP expression through a CT-approach and could be used for the development of a targeted therapy.

Data-Driven Radiogenomic Approach for Deciphering Molecular Mechanisms Underlying Imaging Phenotypes in Lung Adenocarcinoma: A Pilot Study.

The heterogeneity of lung tumor nodules is reflected in their phenotypic characteristics in radiological images. The radiogenomics field employs quantitative image features combined with transcriptome expression levels to understand tumor heterogeneity molecularly. Due to the different data acquisition techniques for imaging traits and genomic data, establishing meaningful connections poses a challenge. We analyzed 86 image features describing tumor characteristics (such as shape and texture) with the underlying transcriptome and post-transcriptome profiles of 22 lung cancer patients (median age 67.5 years, from 42 to 80 years) to unravel the molecular mechanisms behind tumor phenotypes. As a result, we were able to construct a radiogenomic association map (RAM) linking tumor morphology, shape, texture, and size with gene and miRNA signatures, as well as biological correlates of GO terms and pathways. These indicated possible dependencies between gene and miRNA expression and the evaluated image phenotypes. In particular, the gene ontology processes "regulation of signaling" and "cellular response to organic substance" were shown to be reflected in CT image phenotypes, exhibiting a distinct radiomic signature. Moreover, the gene regulatory networks involving the TFs TAL1, EZH2, and TGFBR2 could reflect how the texture of lung tumors is potentially formed. The combined visualization of transcriptomic and image features suggests that radiogenomic approaches could identify potential image biomarkers for underlying genetic variation, allowing a broader view of the heterogeneity of the tumors. Finally, the proposed methodology could also be adapted to other cancer types to expand our knowledge of the mechanistic interpretability of tumor phenotypes.

Radiogenomic association of deep MR imaging features with genomic profiles and clinical characteristics in breast cancer

BackgroundIt has been believed that traditional handcrafted radiomic features extracted from magnetic resonance imaging (MRI) of tumors are normally shallow and low-ordered. Recent advancement in deep learning technology shows that the high-order deep radiomic features extracted automatically from tumor images can capture tumor heterogeneity in a more efficient way. We hypothesize that MRI-based deep radiomic phenotypes have significant associations with molecular profiles of breast cancer tumors. We aim to identify deep radiomic features (DRFs) from MRI, evaluate their significance in predicting breast cancer (BC) clinical characteristics and explore their associations with multi-level genomic factors.MethodsA denoising autoencoder was built to retrospectively extract 4,096 DRFs from 110 BC patients’ MRI. Visualization and clustering were applied to these DRFs. Linear Mixed Effect models were used to test their associations with multi-level genomic features (GFs) (risk genes, gene signatures, and biological pathway activities) extracted from the same patients’ mRNA expression profile. A Least Absolute Shrinkage and Selection Operator model was used to identify the most predictive DRFs for each clinical characteristic (tumor size (T), lymph node metastasis (N), estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status).ResultsThirty-six conventional radiomic features (CRFs) for 87 of the 110 BC patients provided by a previous study were used for comparison. More than 1,000 DRFs were associated with the risk genes, gene signatures, and biological pathways activities (adjusted P-value < 0.05). DRFs produced better performance in predicting T, N, ER, PR, and HER2 status (AUC > 0.9) using DRFs. These DRFs showed significant powers of stratifying patients, linking to relevant biological and clinical characteristics. As a contrast, only eight risk genes were associated with CRFs. The RFs performed worse in predicting clinical characteristics than DRFs.ConclusionsThe deep learning-based auto MRI features perform better in predicting BC clinical characteristics, which are more significantly associated with GFs than traditional semi-auto MRI features. Our radiogenomic approach for identifying MRI-based imaging signatures may pave potential pathways for the discovery of genetic mechanisms regulating specific tumor phenotypes and may enable a more rapid innovation of novel imaging modalities, hence accelerating their translation to personalized medicine.

Radiogenomics of C9orf72 Expansion Carriers Reveals Global Transposable Element Derepression and Enables Prediction of Thalamic Atrophy and Clinical Impairment.

Hexanucleotide repeat expansion (HRE) within C9orf72 is the most common genetic cause of frontotemporal dementia (FTD). Thalamic atrophy occurs in both sporadic and familial FTD but is thought to distinctly affect HRE carriers. Separately, emerging evidence suggests widespread derepression of transposable elements (TEs) in the brain in several neurodegenerative diseases, including C9orf72 HRE-mediated FTD (C9-FTD). Whether TE activation can be measured in peripheral blood and how the reduction in peripheral C9orf72 expression observed in HRE carriers relates to atrophy and clinical impairment remain unknown. We used FreeSurfer software to assess the effects of C9orf72 HRE and clinical diagnosis (n = 78 individuals, male and female) on atrophy of thalamic nuclei. We also generated a novel, human, whole-blood RNA-sequencing dataset to determine the relationships among peripheral C9orf72 expression, TE activation, thalamic atrophy, and clinical severity (n = 114 individuals, male and female). We confirmed global thalamic atrophy and reduced C9orf72 expression in HRE carriers. Moreover, we identified disproportionate atrophy of the right mediodorsal lateral nucleus in HRE carriers and showed that C9orf72 expression associated with clinical severity, independent of thalamic atrophy. Strikingly, we found global peripheral activation of TEs, including the human endogenous LINE-1 element L1HS L1HS levels were associated with atrophy of multiple pulvinar nuclei, a thalamic region implicated in C9-FTD. Integration of peripheral transcriptomic and neuroimaging data from human HRE carriers revealed atrophy of specific thalamic nuclei, demonstrated that C9orf72 levels relate to clinical severity, and identified marked derepression of TEs, including L1HS, which predicted atrophy of FTD-relevant thalamic nuclei.SIGNIFICANCE STATEMENT Pathogenic repeat expansion in C9orf72 is the most frequent genetic cause of FTD and amyotrophic lateral sclerosis (ALS; C9-FTD/ALS). The clinical, neuroimaging, and pathologic features of C9-FTD/ALS are well characterized, whereas the intersections of transcriptomic dysregulation and brain structure remain largely unexplored. Herein, we used a novel radiogenomic approach to examine the relationship between peripheral blood transcriptomics and thalamic atrophy, a neuroimaging feature disproportionately impacted in C9-FTD/ALS. We confirmed reduction of C9orf72 in blood and found broad dysregulation of transposable elements-genetic elements typically repressed in the human genome-in symptomatic C9orf72 expansion carriers, which associated with atrophy of thalamic nuclei relevant to FTD. C9orf72 expression was also associated with clinical severity, suggesting that peripheral C9orf72 levels capture disease-relevant information.

Radiomic and gEnomic approaches for the enhanced DIagnosis of clear cell REnal Cancer (REDIRECt): a translational pilot methodological study

BackgroundThe combination of radiomic and transcriptomic approaches for patients diagnosed with small clear-cell renal cell carcinoma (ccRCC) might improve decision making. In this pilot and methodological study, we investigate whether imaging features obtained from computed tomography (CT) may correlate with gene expression patterns in ccRCC patients.MethodsSamples from 6 patients who underwent partial nephrectomy for unilateral non-metastatic ccRCC were included in this pilot cohort. Transcriptomic analysis was conducted through RNA-sequencing on tumor samples, while radiologic features were obtained from pre-operative 4-phase contrast-enhanced CT. To evaluate the heterogeneity of the transcriptome, after a 1,000 re-sampling via bootstrapping, a first Principal Component Analyses (PCA) were fitted with all transcripts and a second ones with transcripts deriving from a list of 369 genes known to be associated with ccRCC from The Cancer Genome Atlas (TCGA). Significant pathways in each Principal Components for the 50 genes with the highest loadings absolute values were assessed with pathways enrichment analysis. In addition, Pearson’s correlation coefficients among radiomic features themselves and between radiomic features and transcripts expression values were computed.ResultsThe transcriptomes of the analysed samples showed a high grade of heterogeneity. However, we found four radiogenomic patterns, in which the correlation between radiomic features and transcripts were statistically significant.ConclusionsWe showed that radiogenomic approach is feasible, however its clinical meaning should be further investigated.

Radiogenomics Based on PET Imaging

Radiogenomics or imaging genomics is a novel omics strategy of associating imaging data with genetic information, which has the potential to advance personalized medicine. Imaging features extracted from PET or PET/CT enable assessment of in vivo functional and physiological activity and provide comprehensive tumor information non-invasively. However, PET features are considered secondary to features on conventional imaging, and there has not yet been a review of the radiogenomic approach using PET features. This review article summarizes the current state of PET-based radiogenomic research for cancer, which discusses some of its limitations and directions for future study.

Whole genome sequencing identifies loci specifically associated with thoracic aortic wall defects and abdominal aortic aneurysms in patients with European ancestry

ObjectiveThe objective of this study was to better understand the pathophysiology and underlying genetic mechanisms behind two abdominal aortic aneurysm (AAA) subtypes using computed tomographic imaging in combination with whole genome sequencing. MethodsPatients with a known AAA and European ancestry were included in this investigation and underwent genetic and image analysis. Patients with AAAs and indications of descending thoracic aortic pathology (aortic dissection, penetrating aortic ulcers, intramural hematoma, atheromas, ulcerative plaque, and intramural ulceration, and intimal flaps/tears) were classified as having thoracic aortic disease, grouped together, and compared with patients with an AAA and a normal descending thoracic aorta. Whole genome sequencing was then performed on the 93 patients who had imaging features consistent with thoracic aortic disease and the 126 patients with a normal descending thoracic aorta. ResultsThe results of this study suggest one variant-level, four gene-level, and one gene set-level associations in patients with thoracic aortic disease who also had an AAA. The variant rs79508780 located in TSEN54 achieved study-wide significance (P = 1.71E-06). BATF3 and SMLR1 were significantly associated and EFCAB3 and TAF4 were reached suggestive assocation with a diseased descending thoracic aorta (P = 5.23E-26, P = 1.86E-25, P = 1.54E-05, and P = 8.31E-05, respectively). Gene sets were also compiled using MSigDB and trait-based index single nucleotide variation from major genome-wide association studies. GO_DNA_DOUBLE_STRAND_BREAK_PROCESSING, a gene set related to double-stranded DNA break repair, was significantly associated with thoracic aortic disease in AAA patients (P = 1.80E-06). ConclusionsThis pilot study provides further evidence that an AAA may be the end result of multiple degenerative pathways. Genetic variations in vitamin D signaling, cholesterol metabolism, extracellular matrix breakdown, and double-stranded DNA break repair pathways were associated with European patients who had an AAA and thoracic aortic disease. Additionally, this study provides support for the application of a radiogenomic approach for the investigation of other potential pathologies that could lead to the development of an AAA or influence future management decisions. (JVS–Vascular Science.) Clinical RelevanceIn this study, we provide evidence that abdominal aortic aneurysms (AAAs) may be a result of multiple pathophysiologies rather than a single disease. We have identified genetic variants involved in vitamin D signaling, cholesterol metabolism, extracellular matrix breakdown, and double-stranded DNA break repair associated with structural defects in the aortic wall in patients with AAAs who are of European descent. Patients with AAAs and structural defects in the thoracic aorta have been previously linked to differential behavior after endovascular aneurysm repair. These patients with wall defects exhibited greater sac regression, a marker of surgical success, after endovascular aneurysm repair. Our study demonstrates the usefulness of a radiogenomic approach for elucidating mechanisms behind the formation and future behavior of AAAs that could aid surgeons in making future procedural and management decisions.

A Radiogenomic Approach for Decoding Molecular Mechanisms Underlying Tumor Progression in Prostate Cancer.

Prostate cancer (PCa) is a genetically heterogeneous cancer entity that causes challenges in pre-treatment clinical evaluation, such as the correct identification of the tumor stage. Conventional clinical tests based on digital rectal examination, Prostate-Specific Antigen (PSA) levels, and Gleason score still lack accuracy for stage prediction. We hypothesize that unraveling the molecular mechanisms underlying PCa staging via integrative analysis of multi-OMICs data could significantly improve the prediction accuracy for PCa pathological stages. We present a radiogenomic approach comprising clinical, imaging, and two genomic (gene and miRNA expression) datasets for 298 PCa patients. Comprehensive analysis of gene and miRNA expression profiles for two frequent PCa stages (T2c and T3b) unraveled the molecular characteristics for each stage and the corresponding gene regulatory interaction network that may drive tumor upstaging from T2c to T3b. Furthermore, four biomarkers (ANPEP, mir-217, mir-592, mir-6715b) were found to distinguish between the two PCa stages and were highly correlated (average r = ± 0.75) with corresponding aggressiveness-related imaging features in both tumor stages. When combined with related clinical features, these biomarkers markedly improved the prediction accuracy for the pathological stage. Our prediction model exhibits high potential to yield clinically relevant results for characterizing PCa aggressiveness.

Cancer genetic markers according to radiotherapeutic response in patients with primary glioblastoma – Radiogenomic approach for precision medicine

Background and purposeTo find genetic markers associated with response to radiotherapy (RT) in glioblastoma (GB) patients. Materials and methodsFrom Jan 2009 to Dec 2016, 161 patients with newly diagnosed IDH-wild type GB were treated with surgery and adjuvant concurrent chemoradiotherapy with the Stupp’s regimen, and then genomic research proceeded with their surgical specimens. Among the 161 patients, 49 with clinically measurable disease on postoperative MRI were analyzed. The response evaluation to RT was based on Response Assessment in Neuro-Oncology (RANO) criteria. For genomic analyses to compare between patients with progression and non-progression, Fisher test for DNA mutations and copy number alterations and the Gene Set Enrichment Analysis (GSEA) were performed. ResultsRT responses were non-progressive and progressive disease (PD) in 22 (44.9%) and 27 patients (55.1%), respectively. After three months, seven of PD exhibited pseudoprogression. For true response adjusting pseudoprogression from PD, 1-year progression-free survival for true Non-Responders (tNR-group) and true Responders (tR-group) were 0% and 45.4% (p < 0.001), and overall survival were 52.5% and 81.1% (p = 0.046), respectively. In genomic analyses, the tNR-group had more CDKN2A deletions (94.4% vs. 55.6%, p = 0.013), EGFR mutations (33.3% vs. 3.7%, p = 0.012) and less TP53 mutations (22.2% vs. 40.7%, p = 0.333) than the tR-group. In GSEA, immune-related gene sets were enriched in the tNR-group, and in contrast, some gene sets related with cell cycle were enriched in tR-groups. ConclusionThe RANO criteria were feasible for the short-term response evaluation from RT despite of pseudoprogression. Genomic differences such as CDKN2A deletion, EGFR mutation, and immune- or inflammation-related related gene sets enrichment were found to be potentially predictive markers of RT.

Cancer Genetic Markers According to Radiotherapeutic Response in Patients with Glioblastoma - Radiogenomic Approach for Precision Medicine

To find genetic markers associated with response to radiotherapy (RT) in glioblastoma (GM) patients. From Jan 2009 to Dec 2016, 161 patients with newly diagnosed IDH-wild type GM were treated with surgery and adjuvant concurrent chemoradiotherapy (CCRT) with the Stupp’s regimen at one institution, and proceeded genomic analyses with surgical specimens. Of the patients, 49 patients with clinically measurable disease on postoperative MRI were analyzed. Response evaluation to RT was based on Response Assessment in Neuro-Oncology (RANO) criteria and MRI at the time of 3 or 4 months after the end of CCRT was compared with baseline MRI. To identify pseudo-progression, we checked further follow-up MRI and clinical events including re-operation, additional RT or gamma knife surgery, and change of treatment for the next 3 months, which was regarded as true progression. To compare the genomic characteristics between patients with progression or non-responder group (NR-group) and non-progression or responder group (R-group) after RT, we performed genomic analyses with whole-exome sequencing, RNA sequencing and targeted-exome panel. The Fisher test to compare mutations and copy number alterations and the single-sample Gene Set Enrichment Analysis (GSEA) score to interpret gene expression were used. The median age was 57 years (range 22-82) and surgical extension was subtotal resection for all 49 patients. RT responses were complete response (CR) or partial response (PR), stable disease (SD), and progressive disease (PD) in 7 (14.3%), 15 (30.6%), and 27 patients (55.1%), respectively. After further evaluation for 3 months, 7 of 27 patients (25.9%) with PD were pseudoprogression. Median follow-up period was 15 months (range 2-46) and 90% of patients experienced intracranial PD, eventually. One-year progression-free survival (PFS) and overall survival (OS) rate were 25.8% and 69.6%, respectively. With RANO response, 1-year PFS for patients with CR / PR / SD or R-group and PD or NR-group were 43.2% and 12.5% (p = 0.013), and OS were 74.2% and 65.5% (p = 0.349), respectively. After adjusting pseudoprogression from NR-group, 1-year PFS for R- and NR-group were 45.4% and 0 (p < 0.001), and OS were 81.1% and 52.5% (p = 0.046), respectively. With genomic analyses, NR-group had more CDKN2A and EGFR mutations with statistical significance and TP53 mutations seemed more frequent in R-group without statistical significance. In GSEA, immune-related gene sets were enriched in NR-group, and in contrast, some gene sets related with cell cycle were enriched in R-group. RT response evaluation with RANO criteria was related with disease progression despite of pseudo-progression, and this study showed the possibility of difference in genetic signatures according to RT response. To suggest molecular based strategies including RT and provide successful precision medicine for GM patients, therefore, further studies with larger patient number should be followed.

Combining radiomic features with a miRNA classifier may improve prediction of malignant pathology for pancreatic intraductal papillary mucinous neoplasms.

Intraductal papillary mucinous neoplasms (IPMNs) are pancreatic cancer precursors incidentally discovered by cross-sectional imaging. Consensus guidelines for IPMN management rely on standard radiologic features to predict pathology, but they lack accuracy. Using a retrospective cohort of 38 surgically-resected, pathologically-confirmed IPMNs (20 benign; 18 malignant) with preoperative computed tomography (CT) images and matched plasma-based ‘miRNA genomic classifier (MGC)’ data, we determined whether quantitative ‘radiomic’ CT features (+/- the MGC) can more accurately predict IPMN pathology than standard radiologic features ‘high-risk’ or ‘worrisome’ for malignancy. Logistic regression, principal component analyses, and cross-validation were used to examine associations. Sensitivity, specificity, positive and negative predictive value (PPV, NPV) were estimated. The MGC, ‘high-risk,’ and ‘worrisome’ radiologic features had area under the receiver operating characteristic curve (AUC) values of 0.83, 0.84, and 0.54, respectively. Fourteen radiomic features differentiated malignant from benign IPMNs (p<0.05) and collectively had an AUC=0.77. Combining radiomic features with the MGC revealed an AUC=0.92 and superior sensitivity (83%), specificity (89%), PPV (88%), and NPV (85%) than other models. Evaluation of uncertainty by 10-fold cross-validation retained an AUC>0.80 (0.87 (95% CI:0.84-0.89)). This proof-of-concept study suggests a noninvasive radiogenomic approach may more accurately predict IPMN pathology than ‘worrisome’ radiologic features considered in consensus guidelines.

A Combinatorial Investigation of the Response to Anti-angiogenic Therapy in Breast Cancer: New Strategies for Patient Selection and Opportunities for Reconsidering Anti-VEGF, Anti-PI3K and Checkpoint Inhibition

A Combinatorial Investigation of the Response to Anti-angiogenic Therapy in Breast Cancer: New Strategies for Patient Selection and Opportunities for Reconsidering Anti-VEGF, Anti-PI3K and Checkpoint Inhibition

Abstract 970A: Using a radiogenomic approach to classify pancreatic cancer precursors

Abstract Background: Intraductal papillary mucinous neoplasms (IPMNs) are cystic pancreatic cancer precursors incidentally detected by imaging in more than 75,000 Americans each year. There is an unmet need to discover noninvasive approaches to differentiate ‘benign’ IPMNs that can be monitored from ‘malignant’ IPMNs that warrant surgery. We previously developed a blood-based ‘miRNA genomic classifier (MGC)’ that helps predict malignant IPMN pathology. The goal of this study was to evaluate whether novel radiomic features from preoperative computed tomography (CT) scans may improve prediction of IPMN pathology beyond that provided by standard radiologic features, either alone or in combination with the MGC. Methods: Preoperative CT images were obtained for 37 surgically-resected, pathologically-confirmed IPMN cases with matched preoperative miRNA expression data. Images were reviewed for standard radiologic features characterized to be ‘high-risk’ or ‘worrisome’ for malignancy according to consensus guidelines. The region of interest within the pancreas was identified and segmented using a semi-automated algorithm. A total of 112 two-dimensional (2D) quantitative texture features (which measure tumor size, shape, and location) and non-texture features (which measure smoothness, coarseness, and regularity) were extracted. Logistic regression models were used to explore associations between non-redundant radiomic features and IPMN pathology. Principal component analysis was also performed to generate an index score (defined by the first principal component of the most promising radiomic features) that was evaluated for its association with malignant pathology. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of the new radiomic features individually and in combination with the MGC were estimated and compared to values obtained for standard radiologic features. Results: The MGC and standard ‘high-risk’ and ‘worrisome’ radiologic features had area under the receiver operating characteristic curve (AUC) values of 0.83, 0.84, and 0.54, respectively. Analysis of 112 extracted preoperative radiomic CT features revealed 14 textural and non-textural features that differentiated malignant from benign IPMNs (P&amp;lt;0.05). Collectively, the 14 radiomic features had an AUC = 0.77. A model that combined radiomic features and the MGC had an AUC = 0.92 and estimates of sensitivity (83%), specificity (89%), PPV (88%), and NPV (85%) that were superior to models not based on these novel data types. Conclusions: Our preliminary findings suggest that clinical decision-making models that integrate novel quantitative radiomic features with a blood-based miRNA classifier may more accurately predict IPMN pathology than models that rely on standard clinical data or worrisome radiologic features alone. Larger, prospective, multi-center studies are planned to explore this topic further. Citation Format: Jennifer B. Permuth, Jung Choi, Yoganand Balarunathan, Jongphil Kim, Dung-Tsa Chen, Kun Jiang, Sonia Orcutt, Lu Chen, Kimberly Quinn, Rodrigo Carvajal, Guillermo Gonzalez-Calderon, Michelle Fournier, Mahmoud Abdalla, Alberto Garcia, Amber Bouton, Danny Yakoub, Suzanne Lechner, Jose Trevino, Nipun Merchant, Robert Gillies, Mokenge Malafa. Using a radiogenomic approach to classify pancreatic cancer precursors. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 970A.

ANGI-08RADIOGENOMIC rCBV-IMAGING VISUALIZES THE DISTINCT ANGIOGENESIS TRANSCRIPTOME SIGNATURES OF IDH MUTANT AND WILD-TYPE GLIOMAS

BACKGROUND: The recent identification of isocitrate dehydrogenase (IDH) mutations in gliomas and several other cancers suggests that this pathway is involved in oncogenesis; however effector functions are complex and yet incompletely understood. To study the regulatory effects of IDH on hypoxia-inducible-factor 1-alpha (HIF1A), a driving force in hypoxia-initiated angiogenesis, we performed mRNA-expression and functional, as well as genotype/imaging phenotype correlation analysis. METHODS: We studied differential mRNA-expression profiles from 288 samples with low-grade and anaplastic gliomas from The Cancer Genome Atlas (TCGA) of HIF1A and related downstream signaling on a single-gene and pathway level, as well as upstream biological causes and probable downstream effects between mutant and wild-type IDH tumors. Genotype/imaging phenotype correlation analysis was performed in a separate (local) dataset with relative cerebral blood volume (rCBV) MRI - an estimate of tumor angiogenesis - in 73 treatment-naive patients with low-grade and anaplastic gliomas. RESULTS: We show decreased expression of HIF1A-target genes on a single-gene and pathway level, strong inhibition of upstream regulators such as HIF1A and downstream biological functions such as angio- and vasculogenesis in IDH-mutant tumors. Our radiogenomic imaging approach revealed increased levels of rCBV in IDH wild-type tumors, where a one-unit increase in rCBV corresponded to a two-third decrease in the odds for an IDH-mutation and correctly predicted IDH mutation status in 88% of patients. CONCLUSION: Together, these findings show that IDH-mutation status is associated with a distinct angiogenesis transcriptome signature which is non-invasively predictable with rCBV-imaging.

Radiogenomics: What It Is and Why It Is Important

In recent years, a new direction in cancer research has emerged that focuses on the relationship between imaging phenotypes and genomics. This direction is referred to as radiogenomics or imaging genomics. The question that subsequently arises is: What is the practical significance of elucidating this relationship in improving cancer patient outcomes. In this article, I address this question. Although I discuss some limitations of the radiogenomic approach, and describe scenarios in which radiogenomic analysis might not be the best choice, I also argue that radiogenomics will play a significant practical role in cancer research. Specifically, I argue that the significance of radiogenomics is largely related to practical limitations of currently available data that often lack complete characterization of the patients and poor integration of individual datasets. Radiogenomics offers a practical way to leverage limited and incomplete data to generate knowledge that might lead to improved decision making, and as a result, improved patient outcomes.

WE-D-BRE-02: BEST IN PHYSICS (THERAPY) - Radiogenomic Modeling of Normal Tissue Toxicities in Prostate Cancer Patients Receiving Hypofractionated Radiotherapy

Purpose: It has been realized that inter-patient radiation sensitivity variability is a multifactorial process involving dosimetric, clinical, and genetic factors. Therefore, we explore a new framework to integrate physical, clinical, and biological data denoted as radiogenomic modeling. In demonstrating the feasibility of this work, we investigate the association of genetic variants (copy number variations [CNVs] and single nucleotide polymorphisms [SNPs]) with radiation induced rectal bleeding (RB) and erectile dysfunction (ED) while taking into account dosimetric and clinical variables in prostate cancer patients treated with curative irradiation. Methods: A cohort of 62 prostate cancer patients who underwent hypofractionated radiotherapy (66 Gy in 22 fractions) was retrospectively genotyped for CNV and SNP rs25489 in the xrcc1 DNA repair gene. Dosevolume metrics were extracted from treatment plans of 54 patients who had complete dosimetric profiles. Treatment outcomes were considered to be a Result of functional mapping of radiogenomic input variables according to a logit transformation. Model orders were estimated using resampling by leave-one out cross-validation (LOO-CV). Radiogenomic model performance was evaluated using area under the ROC curve (AUC) and LOO-CV. For continuous univariate dosimetric and clinical variables, Spearmans rank coefficients were calculated and p-values reported accordingly. In the case of binary variables, Chi-squared statistics and contingency table calculations were used. Results: Ten patients were found to have three copies of xrcc1 CNV (RB: χ2=14.6 [p<0.001] and ED: χ2=4.88[p=0.0272]) and twelve had heterozygous rs25489 SNP (RB: χ2=0.278[p=0.599] and ED: χ2=0.112[p=0.732]). LOO-CV identified penile bulb D60 as the only significant QUANTEC predictor (rs=0.312 [p=0.0145]) for ED. Radiogenomic modeling yielded statistically significant, cross-validated NTCP models for RB (rs=0.243[p=0.0443], AUC=0.665) and ED (rs=0.276[p=0.0217], AUC=0.754). Conclusion: The radiogenomic modeling approach presented herein has been shown to identify NTCP models which have increased predictive power. Furthermore, CNVs appears to be useful genetic variants when added to dosimetric NTCP models. This work was partially supported by CIHR grant MOP-114910.