This personalized approach applies to both targeted loco-regional therapy, such as surgery and radiotherapy, as well as targeted systemic therapies, such as endocrine treatments for breast cancer, for example. Thus far, tumour characterization has been accomplished by in vitro assay of tumour biopsy material. Now, recent advances in functional and molecular imaging have laid the foundation for testing imaging as a cancer biomarker; namely, to help direct cancer treatment in a way that is complementary to plans based on tissue- and blood-based biomarkers.1, 2
When considering molecular imaging as a cancer biomarker, there are four major areas in which imaging can help guide cancer treatment1:
(1) Prognosis: The imaging biomarker can help determine the aggressiveness of the tumour and thereby infer the likelihood of disease progression and cancer-related patient death. This information can help guide how aggressive the treatment should be.
(2) Prediction: The imaging biomarker can help identify the presence or absence of therapeutic targets and therefore direct the patient to the therapy that is most likely to be effective. Imaging can also help identify factors likely to mediate resistance to particular forms of treatment.
(3) Response: Functional and molecular imaging can measure therapeutic response at a much earlier time point than standard anatomic imaging. In addition, imaging the tumour's in vivo response to treatment may better predict the outcome in relation to important endpoints, such as disease-free and overall survival.
(4) Biology: Molecular imaging provides a unique tool for characterizing the in vivo biology of cancer and its response to treatment. It may offer insights into factors that determine why some patients will response to specific anti-cancer treatments, while other apparently similar patients do not.
It is important to note that this approach is a departure from the standard paradigm for cancer imaging, which has largely focused on detection and staging. Imaging probes used for detection must have reliably higher uptake in tumours compared to normal tissue. For biomarker imaging, however, the absence of a particular feature – for example, tumour receptor expression – may be equally, if not more, important than its presence.
Molecular imaging as a cancer biomarker must therefore be able to go beyond simply detecting the tumour, to quantify tumour in vivo biology across a range of values. This requires the ability to localize and characterize tumours at the same time. Multimodality imaging techniques such as PET/CT, or pairs of imaging procedures – for example, using multiple PET imaging procedures with different probes – may prove especially important for biomarker imaging.3
Similarly, imaging as a biomarker requires image quantification and may entail detailed image analysis beyond standard clinical approach, incorporating kinetic analysis and parametric imaging, for instance. The implementation of molecular imaging as a biomarker in cancer clinical trials and clinical therapy will require efforts to develop rigorous, but clinically feasible, approaches for multimodality imaging and standardized image acquisition and analysis protocols.
Broader scope
Some early examples of the application of imaging as a cancer biomarker are highlighted below, along with accompanying image examples. More detailed descriptions of these applications can be found in a recent Clinical Cancer Research monograph.4
Prognosis A number of studies across a variety of tumour types have shown that rates of tumour glucose metabolism, reflected by PET imaging of 18F-fluorodeoxyglucose (FDG) uptake, are highly predictive of patient outcomes such as progression-free and overall survival.
The mechanisms underlying these findings are incompletely understood and likely complex. They include the association of increased glycolysis with other factors that are predictive of patient outcome, such as indices of cellular proliferation. They may also reflect a cellular stress reaction, also seen in normal non-tumour tissues, that enhances cellular survival and may mitigate the effectiveness of cytotoxic treatments.
One example in which FDG-PET is used to determine prognosis in clinical practice is the case of iodine-refractory thyroid cancer (figure 1). Here, the absence of FDG uptake indicates a quite favourable prognosis that often directs the patient away from further treatment in favour of close observation.5 On the other hand, the presence of FDG uptake in refractory thyroid cancer identifies a relatively lethal form of the disease, indicating that further therapeutic intervention is warranted.
Prediction When it comes to identifying therapeutic targets, imaging provides a method that is complementary to biopsy and in vitro assay. Advantages of imaging include the ability to characterize the entire disease burden (versus a small biopsied sample of the tumour), measure the heterogeneity of the target within or across disease sites, and measure the effect of treatment on the target.
One example is the use of 18F-fluoroestradiol (FES) PET to image regional oestrogen receptor (ER) expression (figure 2).6 Studies have shown that the level of FES uptake in breast cancer predicts the likelihood of response to endocrine therapy. This is akin to current clinical practice, which measures ER expression on biopsy material to make treatment selections. Imaging may be particularly helpful in recurrent or metastatic disease, where biopsy can be challenging.
Another example is PET hypoxia imaging, using compounds such as 18F-fluoromisonidazole (FMISO) and 60Cu-ATSM.7 PET hypoxia imaging may be particularly important in radiotherapy treatment planning where it is likely that hypoxic regions will need different dosing schemes, given the well-documented association between hypoxia and radioresistance.
Response Aberrant cellular proliferation is a hallmark of cancer, and a decline in cellular proliferation is an early event that indicates successful cancer treatment. Work demonstrating the efficacy of 11C-thmyidine PET for early response evaluation led a number of investigators to develop thymidine analogues that can be labelled with 18F for PET imaging, making them practical for more routine clinical use.8
The most promising to date is 18F-fluorthymidine (FLT), which has undergone preliminary evaluation in patients. Early studies support serial FLT as a robust indicator of early response, including response to agents that work primarily as cytostatic (versus cytotoxic) therapy. This is a highly promising area of investigation, and multicentre trials of FLT-PET to measure early therapeutic response are underway.
Cancer biology The ability to measure in vivo cancer biology during treatment provides a unique opportunity to gain insights into factors underlying response and resistance (figure 3). One interesting finding has been the association between mismatches in tumour metabolism and perfusion, and resistance to therapy. 9
For several tumour types, investigators have shown that tumours that have high rates of glucose metabolism relative to blood flow are less likely to respond to treatment, and that patients with tumours displaying such characteristics are more likely to have disease relapse. This same physiology has been seen in ischemic but viable myocardium, and is associated with functional recovery after revascularization.
While tissue viability is beneficial in the case of heart disease, a cellular response that promotes tumour survival is detrimental to the cancer patient. Identification of flow-metabolism mismatch as a marker of therapeutic resistance suggests the need for further studies to understand underlying mechanisms, and may identify alternative targets in tumours that are resistant to standard treatments.
These early studies support the ability of molecular imaging to act as a cancer biomarker and to help guide cancer treatment. Early results will need to be validated in larger, multicentre trials. Both pharma and cooperative groups such as ACRIN (American College of Radiology Imaging Network) are initiating trials to test molecular imaging as a cancer biomarker. Promising early results encourage and support these efforts and suggest considerable promise for molecular imaging in this role in the future.