Perfusion Magnetic Resonance Imaging to Assess Brain Tumor Responses to New Therapies

Histopathology-validated machine learning radiographic biomarker for noninvasive discrimination between true progression and pseudo-progression in glioblastoma

BACKGROUND

Imaging of glioblastoma patients after maximal safe resection and chemoradiation commonly demonstrates new enhancements that raise concerns about tumor progression. However, in 30% to 50% of patients, these enhancements primarily represent the effects of treatment, or pseudo-progression (PsP). We hypothesize that quantitative machine learning analysis of clinically acquired multiparametric magnetic resonance imaging (mpMRI) can identify subvisual imaging characteristics to provide robust, noninvasive imaging signatures that can distinguish true progression (TP) from PsP.

METHODS

We evaluated independent discovery (n = 40) and replication (n = 23) cohorts of glioblastoma patients who underwent second resection due to progressive radiographic changes suspicious for recurrence. Deep learning and conventional feature extraction methods were used to extract quantitative characteristics from the mpMRI scans. Multivariate analysis of these features revealed radiophenotypic signatures distinguishing among TP, PsP, and mixed response that compared with similar categories blindly defined by board-certified neuropathologists. Additionally, interinstitutional validation was performed on 20 new patients.

RESULTS

Patients who demonstrate TP on neuropathology are significantly different (P < .0001) from those with PsP, showing imaging features reflecting higher angiogenesis, higher cellularity, and lower water concentration. The accuracy of the proposed signature in leave-one-out cross-validation was 87% for predicting PsP (area under the curve [AUC], 0.92) and 84% for predicting TP (AUC, 0.83), whereas in the discovery/replication cohort, the accuracy was 87% for predicting PsP (AUC, 0.84) and 78% for TP (AUC, 0.80). The accuracy in the interinstitutional cohort was 75% (AUC, 0.80).

CONCLUSION

Quantitative mpMRI analysis via machine learning reveals distinctive noninvasive signatures of TP versus PsP after treatment of glioblastoma. Integration of the proposed method into clinical studies can be performed using the freely available Cancer Imaging Phenomics Toolkit.

Multivariate Analysis of Preoperative Magnetic Resonance Imaging Reveals Transcriptomic Classification of de novo Glioblastoma Patients

Glioblastoma, the most frequent primary malignant brain neoplasm, is genetically diverse and classified into four transcriptomic subtypes, i. e., classical, mesenchymal, proneural, and neural. Currently, detection of transcriptomic subtype is based on ex vivo analysis of tissue that does not capture the spatial tumor heterogeneity. In view of accumulative evidence of in vivo imaging signatures summarizing molecular features of cancer, this study seeks robust non-invasive radiographic markers of transcriptomic classification of glioblastoma, based solely on routine clinically-acquired imaging sequences. A pre-operative retrospective cohort of 112 pathology-proven de novo glioblastoma patients, having multi-parametric MRI (T1, T1-Gd, T2, T2-FLAIR), collected from the Hospital of the University of Pennsylvania were included. Following tumor segmentation into distinct radiographic sub-regions, diverse imaging features were extracted and support vector machines were employed to multivariately integrate these features and derive an imaging signature of transcriptomic subtype. Extracted features included intensity distributions, volume, morphology, statistics, tumors' anatomical location, and texture descriptors for each tumor sub-region. The derived signature was evaluated against the transcriptomic subtype of surgically-resected tissue specimens, using a 5-fold cross-validation method and a receiver-operating-characteristics analysis. The proposed model was 71% accurate in distinguishing among the four transcriptomic subtypes. The accuracy (sensitivity/specificity) for distinguishing each subtype (classical, mesenchymal, proneural, neural) from the rest was equal to 88.4% (71.4/92.3), 75.9% (83.9/72.8), 82.1% (73.1/84.9), and 75.9% (79.4/74.4), respectively. The findings were also replicated in The Cancer Genomic Atlas glioblastoma dataset. The obtained imaging signature for the classical subtype was dominated by associations with features related to edge sharpness, whereas for the mesenchymal subtype had more pronounced presence of higher T2 and T2-FLAIR signal in edema, and higher volume of enhancing tumor and edema. The proneural and neural subtypes were characterized by the lower T1-Gd signal in enhancing tumor and higher T2-FLAIR signal in edema, respectively. Our results indicate that quantitative multivariate analysis of features extracted from clinically-acquired MRI may provide a radiographic biomarker of the transcriptomic profile of glioblastoma. Importantly our findings can be influential in surgical decision-making, treatment planning, and assessment of inoperable tumors.

Clinical applicability of and changes in perfusion MR imaging in brain metastases after stereotactic radiotherapy

To assess the applicability of perfusion-weighted (PWI) magnetic resonance (MR) imaging in clinical practice, as well as to evaluate the changes in PWI in brain metastases before and after stereotactic radiotherapy (SRT), and to correlate these changes to tumor status on conventional MR imaging. Serial MR images at baseline and at least 3 and 6 months after SRT were retrospectively evaluated. Size of metastases and the relative cerebral blood volume (rCBV), assessed with subjective visual inspection in the contrast enhanced area, were evaluated at each time point. Tumor behavior of metastases was categorized into four groups based on predefined changes on MRI during follow-up, or on histologically confirmed diagnosis; progressive disease (PD), pseudoprogression (PsPD), non-progressive disease (non-PD) and progression unspecified (PU). Twenty-six patients with 42 metastases were included. Fifteen percent (26/168) of all PW images could not be evaluated due to localization near large vessels or the scalp, presence of hemorrhage artefacts, and in 31% (52/168) due to unmeasurable residual metastases. The most common pattern (52%, 13/25 metastases) showed a high rCBV at baseline and low rCBV during follow-up, occurring in metastases with non-PD (23%, 3/13), PsPD (38%, 5/13) and PU (38%, 5/13). Including only metastases with a definite outcome generally showed low rCBV in PsPD or non-PD, and high rCBV in PD. Although non-PD and PsPD may be distinguished from PD after SRT using the PW images, the large proportion of images that could not be assessed due to artefacts and size severely hampers value of PWI in predicting tumor response after SRT.

Angiogenesis and Cancer Control: From Concept to Therapeutic Trial

Background

There is extraordinary interest in developing angiosuppressive agents for cancer treatment. Several new agents appear promising for the treatment of a variety of human cancers. Current concepts and new agents in clinical trials are the focus of this article. In particular, the introduction of a new treatment for human brain tumors is presented in detail, using an antiangiogenic agent, penicillamine, and depletion of an obligatory cofactor of angiogenesis, copper.

Methods

The explosive increase in literature on antiangiogenesis is reviewed using computerized search, findings presented at the recent national cancer and angiogenesis meetings. A specific protocol, NABTT 97-04, “Penicillamine and Copper Reduction for Newly Diagnosed Glioblastoma,” is presented as an example of angiotherapeutic drug discovery.

Results

A number of promising molecular approaches are being introduced to suppress tumor angiogenesis. Major categories of angiogenesis antagonists include protease inhibitors, direct inhibitors of endothelial cell proliferation and migration, suppression of angiogenic growth factors, inhibition of endothelial-specific integrin/survival signaling, chelators of copper, and inhibitors with specific other mechanisms. The preliminary results of early trials offer a glimpse into how antiangiogenesis therapy will be integrated into future care of the patient with cancer.

Conclusions

Thirty-five antiangiogenesis therapies are currently being evaluated in clinical trials. As we learn more about the fundamental mechanisms of angiogenesis, eg, the role of copper in growth factor activation, effective methods of cancer control will be implemented.

Late post-treatment radiographic changes 3 years following chemoradiation for glioma: the importance of histopathology

Treatment-related changes can mimic brain tumor progression both clinically and radiographically. Distinguishing these two entities represents a major challenge in neuro-oncology. No single imaging modality is capable of reliably achieving such distinction. While histopathology remains the gold standard, definitive pathological criteria are also lacking which can further complicate such cases. We report a patient with high-grade glioma who, after initially presenting with histopathologically confirmed pseudoprogression 10 months following treatment, re-presented 3 years following concurrent chemoradiation with clinical and radiographic changes that were most consistent with progressive disease but for which histopathology revealed treatment effects without active glioma. This case highlights the potential late onset of treatment-related changes and underscores the importance of histopathologic assessment even years following initial therapy.

Support vector machine multiparametric MRI identification of pseudoprogression from tumor recurrence in patients with resected glioblastoma

PURPOSE

To automatically differentiate radiation necrosis from recurrent tumor at high spatial resolution using multiparametric MRI features.

MATERIALS AND METHODS

MRI data retrieved from 31 patients (15 recurrent tumor and 16 radiation necrosis) who underwent chemoradiation therapy after surgical resection included post-gadolinium T1, T2, fluid-attenuated inversion recovery, proton density, apparent diffusion coefficient (ADC), and perfusion-weighted imaging (PWI) -derived relative cerebral blood volume (rCBV), relative cerebral blood flow (rCBF), and mean transit time maps. After alignment to post contrast T1WI, an eight-dimensional feature vector was constructed. An one-class-support vector machine classifier was trained using a radiation necrosis training set. Classifier parameters were optimized based on the area under receiver operating characteristic (ROC) curve. The classifier was then tested on the full dataset.

RESULTS

The sensitivity and specificity of optimized classifier for pseudoprogression was 89.91% and 93.72%, respectively. The area under ROC curve was 0.9439. The distribution of voxels classified as radiation necrosis was supported by the clinical interpretation of follow-up scans for both nonprogressing and progressing test cases. The ADC map derived from diffusion-weighted imaging and rCBV, rCBF derived from PWI were found to make a greater contribution to the discrimination than the conventional images.

CONCLUSION

Machine learning using multiparametric MRI features may be a promising approach to identify the distribution of radiation necrosis tissue in resected glioblastoma multiforme patients undergoing chemoradiation.

Semiquantitative human cerebral perfusion assessment with ultrasound in brain space-occupying lesions: preliminary data

OBJECTIVE

Transcranial Duplex ultrasound imaging with ultrasound contrast agents is an emerging technique for evaluating brain perfusion. The aim of this study was to evaluate cerebral perfusion with ultrasound in brain space-occupying lesions to identify different perfusion patterns.

METHODS

Twenty patients with brain space-occupying lesions underwent ultrasound assessment of brain perfusion with a contrast pulse sequencing nonharmonic ultrasound technique and an ultrasound contrast agent bolus. Data were analyzed with software for semiquantitative analysis.

RESULTS

Contrast pulse sequencing imaging with the semiquantitative analysis software allowed identification of qualitative and semiquantitative brain perfusion. Brain hemorrhages showed lower or absent perfusion compared with normal tissue. Meningiomas and glioblastomas without large necrotic areas showed higher perfusion compared with normal tissue. Glioblastomas with large necrotic areas showed overall reduced perfusion compared with normal tissue but higher than that of brain hemorrhages. In glioblastomas with large necrotic areas, it was possible to distinguish between solid and necrotic tissue.

CONCLUSIONS

This bedside ultrasound technique, if validated by larger-scale studies, may add helpful information in noninvasive staging of brain tumors. Further potential applications may be in follow-up imaging to evaluate postoperative tumor recurrence or the presence of radionecrosis.

Comparison of first-pass and second-bolus dynamic susceptibility perfusion MRI in brain tumors

INTRODUCTION

Our goal was to evaluate whether the T1 shortening effect caused by contrast leakage into brain tumors, a well-known confounding effect in the quantification of relative cerebral blood volume (rCBV) measurements, may be corrected by the administration of a predose of gadolinium-DTPA.

METHODS

As part of their presurgical imaging protocol, 25 patients with primary brain tumors underwent two consecutive dynamic susceptibility-weighted contrast-enhanced (DSC) perfusion MR studies. Intratumoral rCBV measurements and normalized rCBV values obtained during the first-pass and second-bolus studies were compared (Wilcoxon signed-ranks test). The frequency of relatively increased rCBV ratios on the second-bolus study was compared between enhancing and non-enhancing neoplasms (Fisher's exact test). Postprocessing perfusion studies were evaluated for image quality on a scale of 0-3 (Wilcoxon signed-ranks test). Four studies were excluded due to unacceptable image quality.

RESULTS

Mean normalized rCBVs were 9.04 (SD 4.64) for the first-pass and 7.99 (SD 3.84) for the second-bolus study. There was no statistically significant difference between the two perfusion studies in either intratumoral rCBV (P=0.237) or rCBV ratio (P=0.181). Five enhancing and four non-enhancing tumors showed a relative increase in rCBV ratio on the second-bolus study, without a significant difference between the groups. Image quality was not significantly different between perfusion studies.

CONCLUSION

Our results did not demonstrate a significant difference between first-pass and second-bolus rCBV measurements in DSC perfusion MR imaging. The administration of a predose of gadolinium-DTPA does not appear to be an efficient way of compensating for the underestimation of intratumoral rCBV values due to the T1 shortening effect.

Amide proton transfer imaging of human brain tumors at 3T

Amide proton transfer (APT) imaging is a technique in which the nuclear magnetization of water-exchangeable amide protons of endogenous mobile proteins and peptides in tissue is saturated, resulting in a signal intensity decrease of the free water. In this work, the first human APT data were acquired from 10 patients with brain tumors on a 3T whole-body clinical scanner and compared with T1- (T1w) and T2-weighted (T2w), fluid-attenuated inversion recovery (FLAIR), and diffusion images (fractional anisotropy (FA) and apparent diffusion coefficient (ADC)). The APT-weighted images provided good contrast between tumor and edema. The effect of APT was enhanced by an approximate 4% change in the water signal intensity in tumor regions compared to edema and normal-appearing white matter (NAWM). These preliminary data from patients with brain tumors show that the APT is a unique contrast that can provide complementary information to standard clinical MRI measures.

Perfusion and diffusion MRI of glioblastoma progression in a four-year prospective temozolomide clinical trial

PURPOSE

This study was performed to determine the impact of perfusion and diffusion magnetic resonance imaging (MRI) sequences on patients during treatment of newly diagnosed glioblastoma. Special emphasis has been given to these imaging technologies as tools to potentially anticipate disease progression, as progression-free survival is frequently used as a surrogate endpoint.

METHODS AND MATERIALS

Forty-one patients from a phase II temolozomide clinical trial were included. During follow-up, images were integrated 21 to 28 days after radiochemotherapy and every 2 months thereafter. Assessment of scans included measurement of size of lesion on T1 contrast-enhanced, T2, diffusion, and perfusion images, as well as mass effect. Classical criteria on tumor size variation and clinical parameters were used to set disease progression date.

RESULTS

A total of 311 MRI examinations were reviewed. At disease progression (32 patients), a multivariate Cox regression determined 2 significant survival parameters: T1 largest diameter (p < 0.02) and T2 size variation (p < 0.05), whereas perfusion and diffusion were not significant.

CONCLUSION

Perfusion and diffusion techniques cannot be used to anticipate tumor progression. Decision making at disease progression is critical, and classical T1 and T2 imaging remain the gold standard. Specifically, a T1 contrast enhancement over 3 cm in largest diameter together with an increased T2 hypersignal is a marker of inferior prognosis.

Drug delivery and transport to solid tumors

PURPOSE

The purpose of this review is to provide an overview of the principles of and barriers to drug transport and delivery to solid tumors.

METHODS

This review consists of four parts. Part I provides an overview of the differences in the vasculature in normal and tumor tissues, and the relationship between tumor vasculature and drug transport. Part II describes the determinants of transport of drugs and particles across tumor vasculature into surrounding tumor tissues. Part III discusses the determinants and barriers of drug transport, accumulation, and retention in tumors. Part IV summarizes the experimental approaches used to enhance drug delivery and transport in solid tumors.

RESULTS

Drug delivery to solid tumors consists of multiple processes, including transport via blood vessels, transvascular transport, and transport through interstitial spaces. These processes are dynamic and change with time and tumor properties and are affected by multiple physicochemical factors of a drug, multiple tumor biologic factors, and as a consequence of drug treatments. The biologic factors, in turn, have opposing effects on one or more processes in the delivery of drugs to solid tumors.

CONCLUSION

The effectiveness of cancer therapy depends in part on adequate delivery of the therapeutic agents to tumor cells. A better understanding of the processes and contribution of these factors governing drug delivery may lead to new cancer therapeutic strategies.

Second harmonic imaging: a new ultrasound technique to assess human brain tumour perfusion

BACKGROUND

Second harmonic imaging is a new ultrasound technique that allows evaluation of brain tissue perfusion after application of an ultrasound contrast agent.

OBJECTIVE

To evaluate the potential of this technique for the assessment of abnormal echo contrast characteristics of different brain tumours.

METHODS

27 patients with brain tumours were studied. These were divided into four groups: gliomas, WHO grade III-IV (n = 6); meningiomas (n = 9); metastases (n = 5); and others (n = 7). Patients were examined by second harmonic imaging in a transverse axial insonation plane using the transtemporal approach. Following intravenous administration of 4 g (400 mg/ml) of a galactose based echo contrast agent, 62 time triggered images (one image per 2.5 seconds) were recorded and analysed off-line. Time-intensity curves of two regions of interest (tumour tissue and healthy brain tissue), including peak intensity (PI) (dB), time to peak intensity (TP) (s), and positive gradient (PG) (dB/s), as well as ratios of the peak intensities of the two regions of interest, were derived from the data and compared intraindividually and interindividually.

RESULTS

After administration of the contrast agent a marked enhancement of echo contrast was visible in the tumour tissue in all patients. Mean PI and PG were significantly higher in tumour tissue than in healthy brain parenchyma (11.8 v 5.1 dB and 0.69 v 0.16 dB/s; p < 0.001). TP did not differ significantly (37.1 v 50.2 s; p = 0.14). A tendency towards higher PI and PG as well as shorter TP was apparent in malignant gliomas. When comparing different tumour types, however, none of these variables reached significance, nor were there significant differences between malignant and benign tumours in general.

CONCLUSIONS

Second harmonic imaging not only allows identification of brain tumours, but may also help in distinguishing between different tumour types. It gives additional and alternative information about tumour perfusion. Further studies are needed to evaluate the clinical potential of this technique in investigating brain tumours-for example in follow up investigations of patients undergoing radiation or chemotherapy-especially in comparison with neuroradiological and neuropathological findings.

Methodology of brain perfusion imaging

Numerous techniques have been proposed in the last 15 years to measure various perfusion-related parameters in the brain. In particular, two approaches have proven extremely successful: injection of paramagnetic contrast agents for measuring cerebral blood volumes (CBV) and arterial spin labeling (ASL) for measuring cerebral blood flows (CBF). This review presents the methodology of the different magnetic resonance imaging (MRI) techniques in use for CBV and CBF measurements and briefly discusses their limitations and potentials.

Feature extraction for MRI segmentation

Magnetic resonance images (MRIs) of the brain are segmented to measure the efficacy of treatment strategies for brain tumors. To date, no reproducible technique for measuring tumor size is available to the clinician, which hampers progress of the search for good treatment protocols. Many segmentation techniques have been proposed, but the representation (features) of the MRI data has received little attention. A genetic algorithm (GA) search was used to discover a feature set from multi-spectral MRI data. Segmentations were performed using the fuzzy c-means (FCM) clustering technique. Seventeen MRI data sets from five patients were evaluated. The GA feature set produces a more accurate segmentation. The GA fitness function that achieves the best results is the Wilks's lambda statistic when applied to FCM clusters. Compared to linear discriminant analysis, which requires class labels, the same or better accuracy is obtained by the features constructed from a GA search without class labels, allowing fully operator independent segmentation. The GA approach therefore provides a better starting point for the measurement of the response of a brain tumor to treatment.