The diagnosis of cancer has historically relied on clinical examination and imaging methods together with histological examination and immunohistochemical staining of tumour samples. Imaging mainly refers to the use of X-rays to create two-dimensional images but also includes scintigraphy, wherein two-dimensional images are acquired after the administration of radioactively-labelled agents preferentially taken up by specific organs, and thermography, used particularly for the detection of surface temperature increases associated with the abnormal metabolic activity and neoangiogenesis occurring in breast tumours. In addition, ultrasound imaging has been used, in particular for the detection of breast and prostate cancers.
Aside from conventional X-rays, perhaps the most familiar imaging method is positron emission tomography (PET), which produces computer-generated images that represent sites of specific biological activity within the body. It is widely used to investigative a range of neurological disorders (e.g. Alzheimer’s disease) as well as in diagnosis and monitoring response to of cancers. PET involves injection into the bloodstream of a small amount of a positron-emitting isotope incorporated in an organic substance (usually 18F- fluorodeoxyglucose (FDG: a glucose analogue)). Positrons (electric charge +1: the anti-matter equivalent of an electron) travel only a short distance before being annihilated by colliding with an electron: this releases two or more gamma ray photons that can be detected. 18F-fluorodeoxyglucose accumulates preferentially in tumour cells because of their abnormal metabolism and this is revealed in a ‘PET scan’.
Single-photon-emission computed tomography (SPECT) is similar to PET although the radioisotope is usually attached to the ligand for a specific cell surface receptor and the gamma radiation is measured directly. These differences decrease the resolution, relative to PET scans, but they also reduce the cost. PET or SPECT can be used together with computed tomography (CT) to provide one of the most powerful methods for locating precisely the position of a tumour (Fig. 1).
- SPECT/CT of chemokine receptor 4 (CXCR4) expressions in xenografts of a human glioblastoma cell line detected using 125I-labelled antibodies.
Computed tomography uses X-rays to acquire two- dimensional images but from a large number of such images, taken as the radiation beam moves through the body, a three-dimensional picture can be pieced together. Because this permits whole organs to be visualised it has become an immensely powerful diagnostic tool since its introduction in the early 1970s. SPECT/CT cameras comprise a gamma detector (the commonly used radioisotopes are thallium (201Tl), technetium-99m (99mTc), iodine (123I) and gallium (68Ga)) and a CT scanner. PET/SPECT-CT, therefore, combines the metabolic activity signal from PET with the anatomical image generated by CT.
Because, after injection, FDG becomes trapped in any cell that sequesters it, tissues with high glucose uptake, such as the brain, the liver and many cancers, give a strong radiolabelling signal. Not all cancer cells take up significant amounts of FDG but PET is commonly used to detect Hodgkin’s disease, non-Hodgkin’s lymphoma, lung, pancreatic and oesophageal cancers. The standardised uptake value (SUV) is a pseudo-quantitative measure of FDG uptake but a number of factors make the SUV, and hence the interpretation of PET data, a somewhat uncertain business. These include the level of glucose in the patient’s bloodstream (because this competes with FDG), the time of day (which affects metabolism) and even the scanner used.
Response Evaluation Criteria in Solid Tumours (RECIST) is an internationally accepted set of guidelines for the measurement of the response of cancer patients to treatment – in other words, whether they improve (‘respond’), stay the same (‘stabilise’), or worsen (‘progression’) during treatment.
Although PET scanning is considered to be non-invasive, the subject is exposed to ionising radiation. For 18F-FDG the typical dose is 8 mSv and for the whole body, CT scans the range is 7 to 30 mSv. The risk of such exposures leading to cancer is small but it is not negligible and the estimate is that about 1.5% of cancers in the USA might be attributable to single CT scans.
In the quest for more sensitive methods, other radiolabels are being investigated, for example, 18F-sodium fluoride for bone metastases from renal cell carcinoma. 11C-acetate and technetium-99m methylene diphosphonate ((99mTc)MDP) are also being developed.
The earliest report that tumours could be distinguished from normal tissue in vivo by nuclear magnetic resonance (NMR) was in 1971, and by the end of that decade, the first whole-body scans had been carried out. From these beginnings, the field of magnetic resonance imaging (MRI) has evolved to become a widely used radiological method to construct two- or three-dimensional images of the body. The technique uses powerful magnetic fields that align (polarise) atomic nuclear spins with the field (Fig. 2), followed by a brief pulse of an electromagnetic field (usually of radio frequency (RF)) that flips the spin on the aligned nuclei. The energy of the RF pulse is absorbed and then radiated at a specific resonance frequency.
- Representation of the concept of individual nuclei aligning in a strong magnetic field.
The resonance frequency of a nuclear spin depends on the chemical environment of the nucleus and changes therein, referred to as chemical shifts, are reported in ppm.
Magnetic resonance imaging usually images the hydrogen atoms (protons) in water and imaging is possible because the protons in different tissues return to their equilibrium state at different rates. A typical MRI scan comprises about 20 sequences that can be designed to give T1-weighted or T2-weighted images depending on the selected values of the echo time (TE) and the repetition time (TR). T1- weighted images show water and fluid within tissues as dark regions with fatty tissues being light; T2-weighted images highlight water.
In addition to proton signals, MRI can utilise other nuclei, for example, endogenous 31P or 23Na, and protein NMR utilises heteronuclear signals for the determination of structures. Exogenous compounds labelled with 19F, 13C or 17O can be used in vivo by injecting reagents into the bloodstream to improve the contrast in tissues being imaged. Compared to PET, MRI gives greatly enhanced soft-tissue contrast and resolution that is particularly valuable in delineating tumours (Fig. 3).
3. MRI and PET.
A development of NMR called high- resolution magic-angle-spinning 1H NMR spectroscopy permits quantitative analysis of metabolites in tissues and has been applied to the characterisation of a range of tumours in vivo. In contrast to PET and CT, these NMR methods do not use ionising radiation and they are considered to be non-invasive and very safe.
Images can be obtained from slices in essentially any orientation through a tissue. This has led to the development of three-dimensional scanning methods, giving MRI a valuable role in delineating solid tumours prior to surgery.
In addition to anatomical imaging, Fourier transformation of NMR signals permits the identification of specific metabolites (Fig. 4).
4. Proton MRI imaging of the human brain. Left
Thus, for example, the choline signal is increased in several types of a tumour whereas there are lower levels of N-acetyl aspartate in brain tumours compared with corresponding normal tissue. Changes in phospholipid metabolism have also been detected by 31P MR spectroscopy and raised levels of phosphocholine in breast tumours have been associated with tumour malignancy. Broadly speaking, MR has shown that the levels of lipids may reflect the rate of cell proliferation, metastatic capacity and the acquisition of drug resistance. However, thus far proton NMR has been relatively limited in both its diagnostic capacity and responses to drugs.
Despite these limitations, the rapidly developing field of tumour imaging is of potentially great importance in both detection and monitoring response to treatment. The combination of MRI (or CT) scans with PET images (a strategy called image fusion or co-registration) is used to produce the most refined tumour maps. This has led to the design of programmes that precisely match the beams of therapeutic radiation to the contours of a tumour so as to minimise exposure of healthy tissue. Together with developments in radiation methods, these advances mean that radiotherapy is steadily improving in terms of dose refinement and radiation targeting.
A current problem is the inadequacy of imaging methods in detecting early tumour responses to drugs. Prompted the development of smart magnetic resonance sensors based on gadolinium (Gd3+) that respond to changes in their immediate environment (Fig. 5).
- Magnetic resonance imaging of the accumulation of a phosphatidylserine-binding contrast agent in the tumours of drug-treated animals 24 hours after injection.
One promising method for identifying apoptotic cells uses a conjugated Gd3+ contrast reagent that binds phosphatidylserine expressed on the surface of apoptotic cells. This and similar approaches using technetium-99m labelled reagents have shown promise in detecting cell death in mouse tumours.
Many non-hydrogen nuclei with an intrinsic magnetic moment occur at very low abundance in animals and are therefore only detectable at low sensitivity by NMR. Considerable effort has been directed towards increasing the MRI signal that can be obtained from, for example, 13C and 3He. Although thus far confined to mice, the method of hyperpolarised 13C-MRI enhances by over 50,000-fold the signals that can be obtained from 13C NMR. In this method dynamic nuclear polarisation of 13C-labelled is carried out at low temperature (~1 K), the polarised mixture is then rapidly dissolved in buffer at room temperature to produce a hyperpolarised solution that is injected into mice in an MR scanner for spectral recording. The 13C-labelled substrates (alanine, lactate, pyruvate, fumarate and malate) permit quantitation of the differential metabolism that characterises cancer cells. This essentially non-invasive method has shown potential for grading mouse models of prostate cancer and also for three-dimensional 13C spectroscopic imaging of tumours (Fig. 7.7).
- Stained sections and hyperpolarised 13C spectra for histologically defined groups of a transgenic mouse model for prostate adenocarcinoma
The same approach applied to lymphomas implanted in mice has shown that conversion of fumarate to malate, reflecting tumour cell necrosis, is increased by administration of etoposide, the topoisomerase II inhibitor that promotes apoptosis of cancer cells (Fig. 7).
- Representative transverse images from (a) untreated and (b) etoposide-treated mice with implanted lymphoma tumours.
There is a pressing requirement for methods to detect early responses to these and related findings indicate the potential of hyperpolarised 13C-MRI in this respect.
A currently promising development combines hyperpolarisation with superparamagnetic iron oxide nanoparticles (SPIONs). SPIONs are a class of contrast reagents that can be coated with ligands to target them to cells (Fig. 8).
- Lymph node metastasis of human prostate carcinoma cells in a mouse detected by MRI.
For example, coating nanoparticles with peptides that are over-expressed on some tumour cells can promote binding to integrin receptors. The magnetic field gradient generated by SPIONs attenuates the MR signal so that localisation of the nanoparticles at metastases produces dark spots. The sensitivity is such that single cells can be detected and the method has, for example, detected early-stage lung metastases from human breast and prostate carcinoma tumours implanted in mice.