Functional and Molecular Imaging for Personalized Medicine in Oncology

Contrast Agent Kinetics

CT contrast agents used in clinical practice are low molecular weight contrast agents (<1 kDa) with negligible serum protein binding and a distribution similar to extracellular fluid. These agents are typically derivatives of iodobenzoic acid with an iodine concentration of at least 300 mg/mL. After intravenous injection (typically 4 mL/s or faster), pharmacokinetic modelling can be performed and approximated to a two-compartment model: the injected contrast agent initially remains within the intravascular compartment before diffusing into the extravascular and extracellular space (EES). The rate of this diffusion is determined by the perfusion of the organ, the vessel surface area and its permeability or leakiness; there is negligible transfer into the intracellular compartment (<1%). The contrast agent then passes back from the EES into the intravascular compartment before being excreted predominantly by the kidneys; up to half of the administered dose is eliminated from the blood within the first 2 hours of injection.

By acquiring a rapid series of images following intravenous contrast agent administration, and assessing the changes in tissue and vessel attenuation during the acquisition, functional parameters can be derived ( Fig. 68.2 ). These may be semiquantitative (describing the ‘curve shape’ of the tissue attenuation–time graph), or quantitative parameters derived from kinetic modelling. As the change in measured CT attenuation is directly proportional to the concentration of iodine within the blood vessels or tissues, temporal changes in attenuation can be directly modelled to assess tissue vascularity. Acquisitions with a higher temporal frequency are normally acquired over the 45 seconds of the perfusion phase, with a lower temporal frequency beyond this. However, this has to be balanced against radiation dose and the finite time required for each acquisition. Quantitative parameters may be derived using a number of models, for example, the modified Johnson–Wilson model, the Patlak method and the maximum slope model.

The derived parameters include:

• regional tumour blood flow—blood flow per unit volume or unit mass of tissue;

• regional tumour blood volume—the proportion of tissue that comprises flowing blood;

• mean transit time—the average time for contrast material to traverse the tissue vasculature;

• extraction fraction—the rate of transfer of contrast material from the intravascular space to the EES; and

• permeability–surface area product—which characterises the rate of diffusion of the contrast agent from the intravascular compartment to the EES.

The basis for the use of DCE-CT in oncology is that microvascular changes during angiogenesis are reflected in changes in the measured DCE-CT parameters; for example, permeability surface-area product is usually lower in normal tissue compared with tumours ( Fig. 68.3 ). DCE-CT measurements have been validated in a range of tumours in both animal models and human studies. Measurements have been correlated positively with histological markers of angiogenesis and negatively with histological markers of hypoxia, indicating that these may be appropriate surrogates of fundamental biological processes during cancer formation.

In terms of characterisation, DCE-CT may distinguish benign from malignant lesions within the lung, pancreas and bowel, although there is some overlap between malignant and inflammatory parameters, which reflects the generic nature of the vascular changes that can be probed. In general, higher perfusion parameters have been reported in patients with tumours, although there is variability between different types of tumours and even within the same tumour, which underlies the complexity of tumour biology.

The major application of DCE-CT in routine clinical practice is in the assessment of the antivascular effects of conventional chemotherapies and interventional procedures, which target the vasculature. DCE-CT is also used to provide pharmacodynamic information in early-phase clinical trials in a variety of cancers ( Table 68.1 ). These have included antiangiogenic and vascular disrupting agents, where DCE-CT is providing a direct imaging biomarker of the drug action and can be used to determine the appropriate drug dose.

The wide availability of CT, the low cost of CT and the ease of standardisation of DCE-CT are advantages over MRI for its use in clinical practice, despite the radiation risk. However, CT carries a significant radiation burden and there still remains a lack of data concerning the relationship between acute vascular reduction and long-term patient outcome.

Additionally, DCE-CT may have an important role in risk stratification and as a predictive biomarker of treatment. The basis for the predictive value of DCE-CT in the setting of chemotherapy is likely to relate to reduced drug delivery, while in radiotherapy this is likely to represent a marker of the hypoxic environment, which, in turn, correlates with resistance to radiotherapy. For example, in locally advanced squamous cell carcinoma of the head and neck treated with surgery and adjuvant chemoradiotherapy, pre-treatment primary tumour blood flow and permeability may be independent predictors of disease recurrence. In pancreatic cancer, a low baseline volume transfer constant ( K ^trans ) predicts for a poorer response to chemotherapy with gemcitabine and radiotherapy. In colorectal cancer, tumours with a lower blood flow at staging are more likely to have nodal metastases and a poorer outcome; rectal tumours with a lower blood flow are also more likely to respond poorly to chemoradiation.

The cancer risk associated with the radiation dose of DCE-CT has to be balanced against potential benefits of vascular quantification and must be judged in the context of the population under investigation. Typical effective radiation doses from a first-pass volumetric perfusion CT study of the thorax, abdomen or pelvis range from 13.7 to 28.7 mSv. Using a risk estimate of 4.2% per Sv from the International Commission on Radiation Protection, the estimated lifetime risk of developing a cancer from a single such perfusion CT is approximately 1 in 1000.

Dynamic Contrast-Enhanced Magnetic Resonance Imaging

DCE-MRI consists of serial MRI acquisitions following injection of an intravenous contrast agent in a similar manner to that described above for DCE-CT. Clinical dynamic MRI is usually performed using low molecular weight gadolinium chelate-based contrast agents. These have paramagnetic ions are known to interact with nearby hydrogen nuclei and lead to shortening of T ₁ (and T ₂ ) relaxation times, resulting in signal enhancement on T ₁ weighted images and thus producing positive contrast. The major advantages of MRI include the absence of ionising radiation, high contrast-to-noise ratio, high signal-to-noise ratio and the many mechanisms that can be utilised to produce tissue contrast.

As with contrast-enhanced CT, contrast-enhanced MRI can either be used to provide a qualitative snapshot of tissue enhancement, as is used routinely in clinical practice, or more quantitatively in the form of DCE-MRI ( Fig. 68.4 ). The latter permits a fuller depiction of contrast kinetics within lesions in much the same way as DCE-CT. DCE-MRI can be repeated over a course of treatment to monitor changes in tumour vascularity over time. Although the technique is reproducible when using a single clinical MRI system, the reproducibility of DCE-MRI studies between centres may be less robust, due to the differences in scanner hardware, contrast agent injection protocols, acquisition parameters and kinetic models employed.

DCE-MRI protocols most commonly involve T ₁ weighted image acquisition before, during and after the injection of the MR contrast agent (typically 0.1 mmol/kg with injection after 1 minute and continuous data acquisition for up to 10 minutes); this provides an assessment of the different stages of tissue uptake and washout. The contrast agents used are either low molecular weight agents (<1 kDa) that rapidly diffuse into the extracellular space, or larger macromolecular agents (>30 kDa) that demonstrate prolonged intravascular retention. Given the lack of ionising radiation in DCE-MRI, temporal data can be continuously acquired during the phases of tissue enhancement, unlike in DCE-CT. The concentration of the contrast agent in the vasculature allows an assessment of perfusion, and in the case of the low molecular weight agents, this is followed by rapid diffusion into the EES where it accumulates. As with DCE-CT, the rate at which this occurs is dependent on blood flow as well as vessel permeability and surface area; however, MR signal intensity is not directly proportional to the contrast agent concentration and, therefore, more complex quantitative data analysis is required to convert the MR signal intensity into biologically meaningful quantitative parameters.

A simple approach is to use the initial area under the curve (IAUC), which describes the shape of the graph of contrast agent concentration over time; although this is frequently used in trials, it is difficult to interpret physiologically. Therefore, in clinical trials, assessment of the effect of an antiangiogenic or vascular disrupting agent is often modelled using changes in K ^trans (the volume transfer coefficient of contrast between the blood plasma and the EES, as described above for CT) and the volume of the EES ( v _e ). The other commonly used pharmacokinetic variables are summarised in Tables 68.2 and 68.3 .

DCE-MRI and DCE-CT exploit the fact that the onset of many diseases is associated with an alteration in vascular density, vascular permeability and blood flow. In particular, tumours develop a network of new vessels as they grow, but unlike normal vasculature, tumour angiogenesis is chaotic and inefficient with permeable vessels. Therefore, an increase in signal enhancement, vessel permeability and flow is often demonstrated within tumours when compared with benign lesions or normal tissue.

DCE-MRI has been used for tumour detection, characterisation, staging and therapy monitoring. However, the meaning of an elevated K ^trans is still controversial in terms of prognosis, as studies have shown conflicting results; there is stronger evidence that K ^trans can be used to demonstrate which tumours are responding to therapy as a pharmacodynamic biomarker of drug activity, particularly in the context of antiangiogenic drugs or vascular disrupting agents. Changes in K ^trans have been shown to correlate both with the administration of vascular endothelial growth factor (VEGF, a signalling molecule that stimulates the growth of new blood vessels), as well as the administration of therapeutic monoclonal antibodies that block its effect. Consequently, both DCE-MRI and DCE-CT can be used as a platform to understand drug and tumour interactions.

Another approach has been to use dynamic susceptibility contrast MRI (DSC-MRI) that also relies upon the serial acquisition of images after the injection of a contrast agent. DSC-MRI measures induced alterations in the transverse relaxation times, T ₂ and T ₂ *, resulting in signal loss and hence transient darkening of the tissues, therefore acting as a negative contrast unlike that seen with T ₁ weighted imaging. The degree of signal loss is dependent on the concentration of the agent as well as vessel size and density and this can therefore be used to estimate the relative blood volume (rBV) of the tissue under assessment. Using this technique, changes in relative cerebral blood volume (rCBV) maps have been correlated with glioma grade and this approach can be used to understand the nature of tumour heterogeneity and to target biopsies to focal areas of vascular changes within a tumour, which may help to avoid sampling error. The method has also been used to distinguish radiation necrosis from recurrent disease, evaluate response to therapy and as a prognostic marker ( Table 68.4 ).

Gadolinium-based contrast agents should be avoided in some patient groups. For example, there is evidence that the use of gadolinium chelates in the presence of significant renal impairment may result in free gadolinium accumulation as part of a rare but severe complication termed nephrogenic systemic fibrosis (NSF). More recently, there have been a number of published reports demonstrating that free gadolinium derived from clinical MRI contrast agents accumulates within the brain, particularly the dentate nucleus. In July 2017, the Pharmacovigilance Risk Assessment Committee (PRAC) of the European Medicines Agency (EMA) produced recommendations for the restriction of some linear gadolinium agents used in MRI body scans and suspend the authorisations of others. Macrocyclic agents (gadobutrol, gadoteric acid and gadoteridol) are more stable and have a lower propensity to release gadolinium than linear agents; the EMA recommends that these agents can continue to be used for their current indications but at the lowest doses that enhance images sufficiently and only when unenhanced imaging is not suitable. Importantly, there is currently no evidence that gadolinium deposition in the brain has caused any harm to patients and gadolinium-based contrast agents remain important clinical and research tools for radiology.

The role of DCE-MRI in clinical practice has been limited by the relatively small number of patients in many published trials, the use of widely varying acquisition techniques and modelled parameters between centres, as well as the use of diverse disease endpoints. Current attempts to standardise DCE-MRI will help to address these issues in the future. DCE-MRI has shown much promise over the years, but its use in routine clinical practice has only recently been recommended in certain cancers such as prostate cancer.

Diffusion-Weighted Imaging

Water molecules in the liquid phase undergo thermally driven random motions, a phenomenon known as Brownian motion or free diffusion, and it is these small motions—typically of the order of 30 µm—that can be probed and quantified using diffusion-weighted imaging (DWI). These small molecular movements can be measured by spin-echo T ₂ weighted sequences, in which two equal diffusion sensitising gradients are applied before and after a 180-degree radiofrequency pulse. The b -value (in s/mm ² ) is a commonly applied term that allows the quantification of these gradients by pooling information from a number of variables. By measuring how far a molecule moves in a fixed time interval, the diffusion constant can be calculated.

In routine clinical practice, DWI can be used both qualitatively and quantitatively. Qualitative assessment identifies relative DWI changes compared with the surrounding normal tissue. Quantitative information can be obtained through the calculation of the apparent diffusion coefficient or ADC. This mathematical entity can be calculated from the slope of relative signal intensity (on a logarithmic scale) against a series of b -values.

There is increasing evidence that the calculated ADC correlates with tissue cellularity. Within biological tissue, the small molecular movements of water are subject to restrictions, which are inherent to the medium due to the surrounding cells and constituency of the extracellular space. In the presence of few or no cells, there is high water diffusion and the molecules will diffuse further in a fixed time interval compared with water molecules within a high cellular environment. Tissues with low cellularity have lower DWI signal intensity and higher ADC values, while the opposite occurs with more solid tissues with a high cellularity: for example, tumour, cytotoxic oedema, abscess and fibrosis. Although the restricted diffusion seen within tumours is largely caused by increased cellular density, other factors are likely to play a role such as the tortuosity of the extracellular space, extracellular fibrosis and the shape and size of the intercellular spaces. Clinically, DWI is used to detect, characterise and stage tumours, distinguish tumour from surrounding tissues, and predict and monitor response to therapy, as well as evaluate tumour recurrence. Successful treatment is generally reflected by an increase in the ADC value, although transient early decreases can occur following treatment.

The development of stronger diffusion gradients, faster imaging sequences, improvements in hardware and motion compensation techniques, have allowed a larger number of b -values to be acquired in a shorter time. These advances have permitted DWI coverage to be extended to the whole body, which has been shown to have several useful applications in oncology such as tumour staging, identification of bone metastases and treatment response monitoring ( Fig. 68.5 ). Progress in data analysis models has allowed additional information about tissue microstructure and heterogeneity to be assessed beyond the ADC value: these advanced DWI techniques include intra-voxel incoherent motion (IVIM), anomalous diffusion imaging (ADI), diffusion kurtosis imaging (DKI), and restriction spectrum imaging (RSI).