Chris Rose

Imaging Intratumor Heterogeneity: Role in Therapy Response, Resistance, and Clinical Outcome

Tumors exhibit genomic and phenotypic heterogeneity, which has prognostic significance and may influence response to therapy. Imaging can quantify the spatial variation in architecture and function of individual tumors through quantifying basic biophysical parameters such as CT density or MRI signal relaxation rate; through measurements of blood flow, hypoxia, metabolism, cell death, and other phenotypic features; and through mapping the spatial distribution of biochemical pathways and cell signaling networks using PET, MRI, and other emerging molecular imaging techniques. These methods can establish whether one tumor is more or less heterogeneous than another and can identify subregions with differing biology. In this article, we review the image analysis methods currently used to quantify spatial heterogeneity within tumors. We discuss how analysis of intratumor heterogeneity can provide benefit over more simple biomarkers such as tumor size and average function. We consider how imaging methods can be integrated with genomic and pathology data, instead of being developed in isolation. Finally, we identify the challenges that must be overcome before measurements of intratumoral heterogeneity can be used routinely to guide patient care. Clin Cancer Res; 21(2); 249–57. ©2014 AACR.

DCE-MRI biomarkers of tumour heterogeneity predict CRC liver metastasis shrinkage following bevacizumab and FOLFOX-6

Background:There is limited evidence that imaging biomarkers can predict subsequent response to therapy. Such prognostic and/or predictive biomarkers would facilitate development of personalised medicine. We hypothesised that pre-treatment measurement of the heterogeneity of tumour vascular enhancement could predict clinical outcome following combination anti-angiogenic and cytotoxic chemotherapy in colorectal cancer (CRC) liver metastases.Methods:Ten patients with 26 CRC liver metastases had two dynamic contrast-enhanced MRI (DCE-MRI) examinations before starting first-line bevacizumab and FOLFOX-6. Pre-treatment biomarkers of tumour microvasculature were computed and a regression analysis was performed against the post-treatment change in tumour volume after five cycles of therapy. The ability of the resulting linear model to predict tumour shrinkage was evaluated using leave-one-out validation. Robustness to inter-visit variation was investigated using data from a second baseline scan.Results:In all, 86% of the variance in post-treatment tumour shrinkage was explained by the median extravascular extracellular volume (ve), tumour enhancing fraction (EF), and microvascular uniformity (assessed with the fractal measure box dimension, d0) (R2=0.86, P<0.00005). Other variables, including baseline volume were not statistically significant. Median prediction error was 12%. Equivalent results were obtained from the second scan.Conclusion:Traditional image analyses may over-simplify tumour biology. Measuring microvascular heterogeneity may yield important prognostic and/or predictive biomarkers.

Quantifying spatial heterogeneity in dynamic contrast-enhanced MRI parameter maps.

Dynamic contrast‐enhanced MRI is becoming a standard tool for imaging‐based trials of anti‐vascular/angiogenic agents in cancer. So far, however, biomarkers derived from DCE‐MRI parameter maps have largely neglected the fact that the maps have spatial structure and instead focussed on distributional summary statistics. Such statistics—e.g., biomarkers based on median values—neglect the spatial arrangement of parameters, which may carry important diagnostic and prognostic information. This article describes two types of heterogeneity biomarker that are sensitive to both parameter values and their spatial arrangement. Methods based on Rényi fractal dimensions and geometrical properties are developed, both of which attempt to describe the complexity of DCE‐MRI parameter maps. Experiments using simulated data show that the proposed biomarkers are sensitive to changes that distribution‐based summary statistics cannot detect and demonstrate that heterogeneity biomarkers could be applied in the drug trial setting. An experiment using 23 DCE‐MRI parameter maps of gliomas—a class of tumour that is graded on the basis of heterogeneity—shows that the proposed heterogeneity biomarkers are able to differentiate between low‐ and high‐grade tumours. Magn Reson Med, 2009.

Organ-specific effects of oxygen and carbogen gas inhalation on tissue longitudinal relaxation times.

Molecular oxygen has been previously shown to shorten longitudinal relaxation time (T1) in the spleen and renal cortex, but not in the liver or fat. In this study, the magnitude and temporal evolution of this effect were investigated. Medical air, oxygen, and carbogen (95% oxygen/5% CO2) were administered sequentially in 16 healthy volunteers. T1 maps were acquired using spoiled gradient echo sequences (TR = 3.5 ms, TE = 0.9 ms, α = 2°/8°/17°) with six acquisitions on air, 12 on oxygen, 12 on carbogen, and six to 12 back on air. Mean T1 values and change in relaxation rate were compared between each phase of gas inhalation in the liver, spleen, skeletal muscle, renal cortex, and fat by one‐way analysis of variance. Oxygen‐induced T1‐shortening occurred in the liver in fasted subjects (P < 0.001) but not in non‐fasted subjects (P = 0.244). T1‐shortening in spleen and renal cortex (both P < 0.001) were greater than previously reported. Carbogen induced conflicting responses in different organs, suggesting a complex relationship with organ vasculature. Shortening of tissue T1 by oxygen is more pronounced and more complex than previously recognized. The effect may be useful as a biomarker of arterial flow and oxygen delivery to vascular beds. Magn Reson Med 58:490–496, 2007.

Candidate biomarkers of extravascular extracellular space: a direct comparison of apparent diffusion coefficient and dynamic contrast-enhanced MR imaging, derived measurement of the volume of the extravascular extracellular space in glioblastoma multiforme

BACKGROUND AND PURPOSE: ADC measurements have been shown to have an inverse relationship with tumor cell density. DCE-MR imaging modeling techniques can produce a measurement of the ve, which would also be expected to have an inverse relationship with cell density. The objective of this study was to test the hypothesis that areas of increased cellularity, and therefore low ADC, would be expected to have a small EES (low ve). MATERIALS AND METHODS: Nineteen patients with GBM were recruited. All imaging was performed before surgery on a 3T MR imaging scanner. Imaging included diffusion tensor imaging, T1-weighted DCE-MR imaging, and anatomic sequences. Tumor VOIs were defined on the anatomic images and modified to contain only enhancing voxels. Parametric maps of ADC and ve were generated. Statistical analysis of ADC and ve was performed on both a voxel-by-voxel basis and comparison of median values. RESULTS: No correlation was demonstrated between ADC and ve in either a voxel-by-voxel analysis or comparison of median values (P = .124). CONCLUSIONS: This study failed to demonstrate a correlation between ADC and ve. This is important because it suggests that though the mechanisms underlying these parameters are theoretically similar, they actually reflect different aspects of tumor microenvironment. Consequently ADC and ve should be considered to provide independent information about the properties of the EES.

Enhancing Fraction in Glioma and Its Relationship to the Tumoral Vascular Microenvironment: A Dynamic Contrast-Enhanced MR Imaging Study

BACKGROUND AND PURPOSE: EnF is a newly described measure of proportional tumor enhancement derived from DCE-MR imaging. The aim of this study was to assess the relationship between EnF and the more established DCE-MR imaging parameters: Ktrans, ve, and vp. MATERIALS AND METHODS: Forty-two patients with 43 gliomas (16 grade II, 3 grade III, and 24 grade IV) were studied. Imaging included pre- and postcontrast T1-weighted sequences through the lesion and T1-weighted DCE-MR imaging. Parametric maps of EnF, Ktrans, ve, and vp were generated. Voxels were classified as enhancing if the IAUC was positive (EnFIAUC60>0). A threshold of IAUC > 2.5 mmol.s was used to generate EnFIAUC60>2.5. Both measures of EnF were compared with the DCE-MR imaging parameters (Ktrans, ve, and vp). RESULTS: In grade II gliomas, EnFIAUC60>0 and EnFIAUC60>2.5 correlated with vp (R2 = 0.6245, P < .0005; and R2 = 0.4727, P = .003) but not with Ktrans or ve. In grade IV tumors, both EnFIAUC60>0 and EnFIAUC60>2.5 correlated with Ktrans (R2 = 0.3501, P = .001; and R2 = 0.4699, P < .0005) and vp (R2 = 0.1564, P = .01; and R2 = 0.2429, P = .007), but not with ve. Multiple regression analysis showed Ktrans as the only independent correlate of both EnFIAUC60>0 and EnFIAUC60>2.5 for grade IV tumors. CONCLUSIONS: This study suggests that in grade II tumors, EnF reflects vp and varies due to changes in vascular density. In grade IV gliomas, EnF is affected by Ktrans with secondary associated changes in vp.

Glandular Function in Sjögren Syndrome: Assessment with Dynamic Contrast-enhanced MR Imaging and Tracer Kinetic Modeling—Initial Experience

PURPOSE To prospectively use dynamic contrast material-enhanced magnetic resonance (MR) imaging and a tracer kinetic model to compare parotid gland microvascular characteristics in patients who have Sjögren syndrome (SS) with those in healthy volunteers. MATERIALS AND METHODS The local research ethics committee approved the study, and written informed consent was obtained from all participants. Twenty-one patients (19 women, two men; age range, 31-73 years) with a diagnosis of SS and 11 healthy volunteers (10 women, one man; age range, 41-68 years) underwent three-dimensional T1-weighted dynamic contrast-enhanced MR imaging of the parotid gland at 1.5 T. A voxel-wise tracer kinetic model and a model-free analysis were applied to the dynamic MR data. Parameter medians and standard deviations were computed to summarize gland microvascular characteristics and gland heterogeneity, respectively. Differences were investigated by using multivariate analysis of variance, t, or U tests. Further investigation was performed by using linear discriminant and receiver operating characteristic analyses. RESULTS Compared with the healthy volunteers, the patients with SS had highly significant elevations (P << .001) in the model-free parameter initial area under the curve and in tracer kinetic model parameters, including transcapillary contrast agent transfer constant (P < .001) and extracellular extravascular volume (P < .001). Gland heterogeneity was significantly greater (P < .001) in the patients with SS. Parameter medians and standard deviations enabled excellent differentiation (areas under receiver operating characteristic curve, 0.96 and 1.00, respectively) between the patients with SS and the healthy volunteers. CONCLUSION Dynamic contrast-enhanced MR imaging has the potential to be used in clinical settings to quantify microvascular function in SS and to differentiate between patients with and those without SS.

Enhancing fraction predicts clinical outcome following first-line chemotherapy in patients with epithelial ovarian carcinoma

Purpose: To define a simple radiologic biomarker of prognosis in patients with advanced epithelial ovarian carcinoma on first-line chemotherapy. Experimental Design: Twenty-seven patients receiving platinum-based chemotherapy with >2 cm residual disease [International Federation of Gynecology and Obstetrics (FIGO) stages IIIC or IV] after surgery were identified. The proportion of enhancing tumor tissue—the enhancing fraction—was calculated on pre-chemotherapy computed tomography scans at four Hounsfield unit (HU) thresholds and assessed for correlation with CA125 response, Response Evaluation Criteria in Solid Tumors (RECIST) radiologic response, and time to progression. Discriminative power was assessed by leave-one-out discriminant analysis. Results: Pre-chemotherapy residual tumor volume did not correlate with clinical outcome. Pre-chemotherapy enhancing fraction at all thresholds significantly correlated with CA125 response (P < 0.001, ρ = 0.553 for 50 HU; P < 0.001, ρ = 0.565 for 60 HU; P < 0.001, ρ = 0.553 for 70 HU; P = 0.001, ρ = 0.516 for 80 HU). Significant correlations were also shown for radiologic response at all thresholds. Enhancing fraction predicted CA125 response with 81.9% to 86.4% specificity and Response Evaluation Criteria in Solid Tumors response with 74.9% to 76.8% specificity at 95% sensitivity (dependent on threshold). Enhancing fraction correlated with time to progression at the 60 HU (P = 0.045, ρ = 0.336) and 70 HU (P = 0.042; ρ = 0.340) thresholds. Conclusion: Pre-chemotherapy enhancing fraction is a simple quantitative radiologic measure. Further evaluation in larger trials is required to confirm the potential of enhancing fraction as a predictive factor, particularly for patients who may benefit from the addition of antiangiogenic therapy.

Web services for the DDSM and digital mammography research

The Digital Database for Screening Mammography (DDSM) is an invaluable resource for digital mammography research. However, there are two particular shortcomings that can pose a significant barrier to many of those who may want to use the resource: 1) the actual mammographic image data is encoded using a non-standard lossless variant of the JPEG image format; 2) although detailed metadata is provided, it is not in a form that permits it to be searched, manipulated or reasoned over by standard tools. This paper describes web services that will allow both humans and computers to query for, and obtain, mammograms from the DDSM in a standard and well-supported image file format. Further, this paper describes how these and other services can be used within grid-based workflows, allowing digital mammography researchers to make use of distributed computing facilities.

Transforming pixel signatures into an improved metric space

Abstract We address the problem of using scale-orientation pixel signatures to characterise local structure in X-ray mammograms, though the method we report is of general application. When signatures are treated as vectors for statistical analysis, the Euclidean metric is not well behaved. We have previously described a Best Partial Match (BPM) metric that measures signature similarity more naturally, but at high computational cost. We present a method for transforming signatures into a new space in which Euclidean distance approximates BPM distance, allowing BPM distance to be estimated at low computational cost. The new space is constructed using multi-dimensional scaling. The nonlinear transformation between the old and new spaces is learned using support vector regression. We present experimental results for mammographic data.