Illuminating human disease

Abstract

New in vivo imaging technologies, including optical methods to observe biological processes in real time, show great promise for preclinical research and diagnostics of human diseases.

In vivo imaging technologies such as ultrasound scanning and magnetic resonance imaging (MRI) have been in clinical use for several decades, but other approaches have also begun to make an impact in different areas of diagnostics, treatment monitoring and animal models of human disease. Recent milestones based on novel imaging technologies include confirming that cognitive therapy can lower neuroinflammation in depression patients, and new insights from animal models on disease progression for infections, various metastatic cancers and heart disease. There has also been growing interest in applying in vivo imaging to study physiological processes in plants, but the principle applications are in humans and animal models.

The main appeal of in vivo imaging is the ability to visualize biological processes in real time down at the molecular level.

The main appeal of in vivo imaging is the ability to visualize biological processes in real time down at the molecular level. The challenges therefore include accurate detection of the signal at sufficiently high resolution and avoidance of disturbing the processes themselves through application of the imaging technology.

The advent of optical techniques

There are five main categories of in vivo imaging technology: computed tomography (CT), positron emission tomography (PET), magnetic resonance imaging (MRI), optical techniques and ultrasound scanning. Optical techniques were the latest to arrive and are based on two forms of luminescence: fluorescent imaging and the more recently introduced bioluminescence imaging. While fluorescence imaging is a physical process, bioluminescence involves a biochemical process whereby enzymes break down a substrate to generate light. It has evolved over the past decade for non‐invasive study of biological processes in cells and small laboratory animals. These two techniques have revolutionized in vivo imaging in laboratory models, in some cases assisted by surgery techniques, such as creation of skin flaps to peer deeper into tissue at high resolutions. It is then possible to combine cellular‐level spatial resolution with whole‐body imaging that allows researchers to reveal behaviour in the natural environment at molecular level (https://d3ciwvs59ifrt8.cloudfront.net/c7fdd453-eaf2-4e71-9951-0b1fda28f2ca/49c26cf1-a1a7-4de6-9d82-41b2243b2bde.pdf).

With in vivo imaging, in principle just a few individuals are needed as the progression can be studied without sacrificing the animal…

This progress has benefits not just for research but also for animal welfare, because it allows a drastic reduction in the number of animals needed for a given experiment. Traditional studies often require larger numbers of individuals that have to be killed at different time points to analyse biological processes over time. With in vivo imaging, in principle just a few individuals are needed as the progression can be studied without sacrificing the animal 1.

The two approaches of fluorescence and bioluminescence images have distinct pros and cons. Fluorescence enables higher temporal resolution because images can be acquired at millisecond exposure times, while bioluminescence takes up to 30 min or even longer. Given the long exposure and low noise requirements, cameras for measuring bioluminescence are more sophisticated, more highly cooled and therefore more expensive. Furthermore, developing a model suitable for bioluminescence imaging requires more effort and time than simply applying a fluorescent probe. Lucigenic reactions are also dependent on varying chemical and enzyme kinetics, as well as the presence of co‐factors or inhibitors. But bioluminescent signals also have advantages, such as imaging tissues at penetration depths of several centimetres. The endogenous signal generation also excludes variation between repeated images that can result from differences in externally applied probes. Given the pros and cons of each, there is obvious appeal in combining the two to offset the weaknesses of each, with the additional benefit of being able to confirm the veracity of the data.

Clinical modelling with bioluminescence

There are already examples of significant advances in clinical modelling achieved with each of the techniques, especially bioluminescence imaging. One of the main breakthroughs has been establishing animal models of various metastatic cancers, commented Li Li, Director of the Laboratory of Translational Cancer Research at the Benson Cancer Center in New Orleans, USA. Li was an author on a 2018 paper describing a patient‐derived xenograft model to emulate the transition of bladder urothelial cell carcinoma to metastatic cancer 2. It is an important disease for study as it currently has a poor prognosis when it reaches the lymph nodes. Although clinically localized disease can be cured, more than half of the patients will develop metastases and die within 5 years; there have been no xenograft models that consistently mimic the process.

There are already examples of significant advances in clinical modelling achieved with each of the techniques, especially bioluminescence imaging.

Li's model uses bioluminescence imaging to assess tumour growth and metastasis. Indeed, xenograft tumours showed better implantation rates than in other models and the tumours resembled the pre‐implanted primary specimens from patients. His paper showed that in the presence of Human Kidney (HK) cells, tumour formation, tumour angiogenesis and distant organ metastasis were significantly enhanced. This provided a platform for investigating the mechanism involved in tumour formation and metastasis, and potentially for evaluating the optimal sequence of conventional drugs or novel therapies. “It is crucial to use bioluminescence imaging in our studies, to assess enrichment of luciferase‐tagged patient‐derived tumour cells in xenografts, primary tumour growth and distant organ metastasis in patient derived xenograft models”, Li explained. He added that while PET/CT could also work, it was very expensive. The main benefit of using bioluminescence over fluorescence imaging was the combination of sensitivity and high throughput.

Li pointed to similar patient‐derived orthotopic xenograft models for solid tumours with metastatic potential, where bioluminescence imaging was used to analyse tumour growth. These include renal cell carcinoma, colon cancer and pancreatic cancer 3. “Firstly, we need to study the mechanism of cancer metastasis, such as identifying prognostic or predictive bio‐markers from primary patient tumours and tumour microenvironment effects on tumour progression, as well as distant organ metastasis. We also need here to identify and target cancer stem cells”, he explained the applications. “Secondly, we are looking towards individualized therapy strategies, such as use of patient‐derived orthotopic mouse models as patient avatars (simulators) to screen conventional or novel treatments, including immune therapies and combination therapies to advise clinical patient management”.

Applications of fluorescence in modelling and diagnosis

Another study of a mouse model of cardiac damage resulting from myocardial infarction has involved fluorescence imaging 4. Just as in the case of the cancer metastatic studies, the authors asserted that progress would have been impossible without in vivo imaging. “We chose fluorescence imaging because it is a stable, convenient method to visualize the ligature of left coronary arteries in the hearts of mice in ex vivo”, explained Jiqiu Chen, one of the authors and assistant professor in medicine at Mount Sinai School of Medicine in New York, USA.”It worked well”. The main finding was not entirely unexpected: that one of the primary causes of variation in cardiac damage after a heart attack is the structure of the coronary arteries. The conclusion was that stagnant blood flow, resulting from both reduced capillary diameter and tortuosity—the extent of curvature—was associated with the development of CHF and merited further study in animal models and eventually humans using different scanning techniques.

While optical imaging for cardiac diagnostics is only just approaching the clinic, it is already being applied for rapid, non‐invasive identification of pathogenic bacteria in wounds.

While optical imaging for cardiac diagnostics is only just approaching the clinic, it is already being applied for rapid, non‐invasive identification of pathogenic bacteria in wounds. A handheld imaging device, developed by MolecuLight (Toronto, Canada), emits a low‐intensity violet light at 405 nm wavelength to excite bacterial fluorescent molecules. A recent paper 5 tested this device on superficial wounds of patients: it yielded a green fluorescent signal in collagen‐containing tissue, a red fluorescent signal from porphyrin‐producing bacteria such as Staphylococcus aureus and a cyan fluorescent signal from pyoverdine‐producing bacteria such as Pseudomonas aeruginosa. Porphyrins are cyclic organic compounds produced by some bacteria that exhibit fluorescence and so can be visualized with fluorescence imaging. Pyoverdines are fluorescent siderophores that are required for pathogenesis in many cases.

New tracers for PET

In vivo imaging has also been applied in human studies in different contexts, such as neuroinflammation associated with depression and other psychiatric disorders, as well as tracking immune processes. Optical techniques do not work well here because of the depths involved coupled with the fact that in humans, it is obviously not possible to apply surgical techniques such as creation of skin flaps to provide access to deeper tissues.

One of the most widely used scanning technologies for detecting and analysing neuroinflammation is positron emission tomography (PET), given its ability to visualize relevant cell types such as the microglia. PET scanning is non‐invasive, and although it involves exposure to ionizing radiation, the doses tend to be less than for CT scanning techniques. PET measures emissions from radioactively labelled metabolically active chemicals, known as tracers, injected into the bloodstream. The emission data are processed by a computer to produce 2D or 3D images of the distribution of these chemicals throughout the area of focus.

PET has been used to study neuroinflammation for more than two decades, but only recently has become sensitive enough for making further progress. Early generations of tracers suffered from several deficiencies, especially low signal‐to‐noise ratio, but novel radiotracers now provide much greater sensitivity. Furthermore, the range of tracers has expanded to target more receptors in the brain. These include the P2X7 receptors, which are highly expressed in immune cells, including the microglia in the central nervous system (CNS) behind the blood/brain barrier (BBB).

“The largest differences between neuroinflammation and more ‘classical’ inflammation are the cell types involved in the inflammatory response and the existence of the BBB/blood‐spinal cord barriers (BSCB)”, commented Daniel Albrecht, Research Fellow at the Gordon Centre for Medical Imaging at Harvard University, USA. “Regarding the cell types, microglia are often referred to as the macrophages of the CNS because they have similar roles and mechanisms of action, but a different embryonic origin. Other cell types, like astrocytes and oligodendrocytes, may participate in neuroinflammatory processes and are exclusively found in the CNS. The BBB and BSCBs limit both the types of harmful molecules that are able to enter the CNS, and the physiological response to deal with a neuroinflammatory event”.

PET has been used to study neuroinflammation for more than two decades, but only recently has become sensitive enough for making further progress.

Albrecht highlighted the translocator protein (TSPO) as an important marker of neuroinflammation, as it is present in the mitochondria of activated microglia, astroglia and macrophages. He also pointed to COX‐2, an enzyme involved in inflammation, as a key marker. Inhibition of COX‐2 has already been found effective in suppressing inflammatory neurodegenerative pathways in mental illness, with beneficial results for major depressive disorder as well as schizophrenia 6.

Work at the NIH (US National Institutes of Health) has produced ligands for COX‐2, which have recently been validated in primates 7. “I'm guessing it won't be much longer before they publish a human study”, Albrecht added. “There's also been progress in developing tracers for purinergic targets, such as P2X7, supposedly targeting microglial activation, though several other CNS cell types express it”. He suggested that such studies at present were confined to research, but highlighted potential for AI techniques to move towards diagnostics. “I believe methodologies like this will help us determine which imaging modalities are more useful for clinical diagnosis”, he said. “Regarding therapy, I'm not aware of any existing trials that are targeting any of these methods. For many disorders, there is still so much left that needs to be worked out on the underlying disease mechanisms before we can even start thinking about how best to treat them. In the shorter term, I think that these imaging methods will likely have more applicability in determining treatment response once therapeutic targets can be worked out”.

There has though been progress applying in vivo imaging to study links between neural inflammation and mental conditions 8. That study applied PET to compare total distribution volume of TSPO, which is a marker of microglial density and therefore neural inflammation, between healthy subjects and patients recently diagnosed with major depressive disorder (MDD). The 40 MDD patients were split into two groups of 20, one of which was assigned cognitive behavioural therapy (CBT) and the other supportive psychotherapy (SPT). It turned out that in the CBT group, but not the SPT group, TSPO volumes were significantly reduced in neocortical grey matter, which appeared to correlate with amelioration of depressive symptoms.

Various versions of PET are already established tools for both diagnostics and treatment monitoring, especially in oncology. Firstly, FDG (fluorodeoxyglucose) PET became established as a clinical tool for cancer management, followed more recently by PET/CT. Fludeoxyglucose has become a popular tracer for PET as a marker for glucose uptake, which is amplified in tumour cells. PET/CT involves use of PET in conjunction with an X‐ray computed tomography scanner to acquire sequential images from both devices in the same session, which are then combined into a single superposed image. The idea is that functional imaging by PET can be aligned precisely with anatomic imaging from CT. FDG‐PET and PET/CT are now both used routinely for investigation of most common solid cancers and increasingly as downstream markers of cancer drugs, particularly in clinical trials, to determine early therapeutic response to novel therapeutics.

Magnetic resonance spectroscopy

Another promising technology is a derivation of magnetic resonance imaging (MRI), called magnetic resonance spectroscopy (MRS), given its ability to accurately determine localized levels of metabolites. MRI is already widely used as a non‐invasive tool for investigative cancer diagnostics, because it offers better visualization than the longstanding alternative approach of CT. MRI shows good contrast between grey and white matter, which has made it the best technology available for imaging many conditions of the central nervous system, including demyelinating diseases, dementia, cerebrovascular disease, Alzheimer's disease, dementia, and epilepsy. As well as being able to study structural features of the brain and detect abnormalities, MRI's ability to capture images at intervals of milliseconds enables analysis of how the brain responds to different stimuli.

While animal models and humans have been the main targets of in vivo imaging so far, there is also growing interest in using these techniques to study physiological processes in plants.

MRS, which uses the same machines as MRI, focuses on smaller areas known as voxels to obtain spatially localized chemical information at the tissue‐resolution level. It observes the whole spectra of electromagnetic absorption, filtering out the much more abundant protons in water and fats. This enables accurate mapping of metabolites and has been acknowledged as a new category of non‐invasive histochemistry analysis. The main benefit of MRS is that allows analysis of many metabolites and comparison between normal and diseased tissue for potential diagnosis, ideally well before symptoms occur.

Applications in plant science

While animal models and humans have been the main targets of in vivo imaging so far, there is also growing interest in using these techniques to study physiological processes in plants. At first sight, it might seem that methods involving high doses of radiation that would be inappropriate in animals and especially humans might be suitable for plants because there is no concern over damage. However, a recent study highlighted the risks associated with synchrotron X‐ray computed micro‐tomography (microCT), a promising technique for in vivo monitoring of various plant processes, including xylem function 9. The technique had appeared to yield valuable insights over embolisms, which in plants are caused by air bubbles that can form during drought. Recovery can occur following rehydration providing not too much damage has been done.

Yet, the study found that microCT scans themselves disrupt fundamental cellular functions and processes, causing serious alterations to cell membranes, with significant leakage of electrolytes. The deleterious effects of X‐rays were apparent in all the species tested, but the magnitude of damage and the minimum number of scans required to induce those effects varied. The obvious conclusion was that microCT investigation of phenomena that depend on physiological activity of living cells may produce incorrect results and lead to misleading conclusions. This is a temporary setback though and just highlights the need for alternative methods. Indeed, this is already happening with application, for example, of fluorescence imaging to detect levels of reactive oxygen species and redox potential in plants 10.

Whether in plants or animals, in vivo imaging across the various technologies is transforming fundamental research into physiological processes at the molecular level. Over a slightly longer term, in vivo imaging will also lead to more accurate and early diagnostics for a range of human diseases, in particular through hybrid approaches that combine two or even more technologies.

EMBO Reports (2019) 20: e49195