Biomedical imaging in the safety evaluation of new drugs

Abstract

Summary

Toxicology accounts for approximately one-third of attrition in new drug development and is a major concern in the pharmaceutical industry. This paper reviews the role of biomedical imaging in the safety evaluation of new candidate drugs. Ex vivo high-resolution three-dimensional imaging of specimens can provide a quick overview of the specimens. Volumetric measurements of tissue structures and lesions can be made with higher precision and reproducibility than histology approaches. As opposed to histology, in vivo animal imaging permits longitudinal studies of the same animals over an extended period of time, with individual animals serving as their own control. Therefore, the number of animals required for a study can be significantly reduced and the intra-subject variability is minimized. Repeated in vivo imaging allows monitoring of the occurrence and progression, or regression, of various structural and functional abnormalities. Compared with other biological assays, imaging can provide anatomically specific information about tissue abnormality. Imaging offers the opportunity to carry forward the same methodology in animal experiments into human studies and has an important role in clinical trials when other safety biomarkers for early toxicities are not available.

Preclinical studies conducted on laboratory animals are typically used to demonstrate the safety and efficacy of a new candidate drug (NCD). If promising, the NCD then proceeds to clinical trials in human subjects. Ideally, candidate drugs being considered for development should be free of toxicities. In reality, toxicology accounts for approximately one-third of attrition in drug development and is thus a major concern in the pharmaceutical industry.

To improve the development process for efficacious and safe new drugs, fast and accurate investigational methods are required to assess the pharmacological activity and safety of the NCDs. Biomedical imaging has an important and evolving role in this aspect (Beckmann et al. 2004, Wang 2005). In preclinical research, non-invasive imaging methods can provide valuable morphological, physiological and metabolic information to characterize disease phenotypes and to evaluate the efficacy and safety of NCDs (Beckmann et al. 2004). The non-invasive nature of imaging is important from statistical power and animal welfare points of view. In clinical trials, the role of imaging includes identification of likely responders, detection and diagnosis of lesions and evaluation of their severity, and monitoring of therapy and follow-up. Nuclear imaging techniques such as positron-emission tomography (PET) can be used to monitor drug pharmacokinetics and distribution and study specific molecular endpoints. In assessing drug efficacy, imaging biomarkers and imaging surrogate endpoints can not only be more objective and faster to measure than clinical outcomes, but also allow small group sizes and good statistical power (Wang 2005).

While imaging efficacy endpoints continue to develop and become extensively applied in drug developments, this is not the case for safety endpoints. Imaging has been underutilized for such investigations at a time when predicting safety issues preclinically and in early clinical development is more important than ever. The recent development of imaging equipment designed specifically for small animal application provides the opportunity of obtaining information for preclinical toxicology risk assessment in a more relevant manner. This paper reviews the role of biomedical imaging in the safety evaluation of NCD, with a focus on preclinical drug safety evaluation.

Background of imaging in drug safety evaluation

The most used biomedical imaging modalities in clinical practice and scientific research include magnetic resonance imaging (MRI) and spectroscopy (MRS), PET, single-photon emission computed tomography (SPECT), X-ray computerized tomography (CT) and ultrasound imaging.

In drug safety evaluation, the most commonly used imaging technique is MR that offers versatile applications. MRI uses radiofrequency waves and high-field magnets to generate images. MRI is excellent for imaging different types of soft tissue with high contrast. It allows very high spatial resolution, and is safer than techniques that use ionizing radiation and therefore repeated imaging is not a risky procedure. The time required for data acquisition can vary from a few minutes to many hours, depending on the size and resolution of the data-sets. For in vivo small animal imaging, the typical time duration for three-dimensional (3D) high-resolution data acquisition is between 30 min and one hour. Localized MRS has been used to obtain non-invasive biochemical information from living tissues. The most common studied nuclei is ¹H (proton).

Compared with the other nuclei, proton spectroscopy is easier to perform and provides high signal-to-noise ratio. The spectra obtained possess the property that the intensity of a given peak is proportional to the number of nuclei contributing to that peak. This allows a quantitative determination of a substance if there is an appropriate internal or external reference. Newer technologies such as diffusion-weighted imaging can provide information about cellularity of a tissue. The use of a bipolar gradient pulse and suitable pulse sequences permits the acquisition of diffusion-weighted images (images in which areas of rapid proton diffusion can be distinguished from areas with slow diffusion). Certain illnesses, for example demyelinization and cytotoxic oedema, show restrictions of diffusion. Although MRI contrast agents, such as Gd-chelates and iron-based agents, are widely used in clinical diagnostic application, these agents are to be avoided whenever possible in drug safety evaluation, because the compounding effects of these contrast agents and NCDs would remain unknown.

SPECT and PET involve the use of isotopes that emit gamma rays or positrons (which result in further generation of gamma rays). These high-energy gamma rays can be detected outside of the body and be used to construct images to supply tomographic information. Nuclear imaging modalities offer very high levels of sensitivity, but suffer from low-spatial specificity and need to be superimposed on an anatomical image (such as CT) for signal location. In general, PET has better spatial resolution and sensitivity than SPECT. Micro-PET and micro-SPECT scanners have increased spatial resolution (1-2 mm in each axial direction) to allow the imaging of small animals, and can provide pharmacokinetic and pharmacodynamic information not previously attainable in animals. PET has proved to be very valuable for the localization of drugs and their sites of action, it also holds potential for molecular imaging such as the assessment of cell loss including apoptosis, and assessing gene expression in vivo (Pogge & Slikker 2004, Walker et al. 2004).

X-ray CT is commonly used to look at anatomical structures. It is very good for studying the skeletal system and fat distribution. Images are obtained from the differential absorption of the X-rays through the various tissues. Small animal CT scanners have been developed for research purposes. The resolution achievable is around 50 μm. Unlike MR, there is little natural contrast achievable between different soft tissues, so iodinated contrast agents, which are vascular-tone active and nephrotoxic (Wang et al. 1997, 2001), are often necessary for CT examinations. Radiation and the necessity of using iodinated contrast agents limit the application of X-ray CT in drug safety evaluation.

Ultrasound imaging is a real-time in vivo technique. An ultrasound transducer transmits ultrasound waves into tissues. As the ultrasound sound waves encounter the interfaces between various types of tissue, these waves are reflected. The transducer detects the reflected ultrasound sound waves and uses them to construct images. Resolution increases as the ultrasound frequency increases; therefore, high-frequency ultrasound is needed for rodent imaging. The depth of penetration depends on the frequency of the sound wave, the higher the frequency, the shallower the depth of penetration. Clinical ultrasound scanners using 7.5-15 MHz frequencies have a spatial resolution of 300-500 μm, whereas ultrahigh frequency (40 MHz) systems have been developed that are specifically used for small rodent imaging offering resolutions of 40-60 μm. Advantages of ultrasound include rapid image acquisition and ease of use. Ultrasound can be used for the assessment of NCD-induced cardiotoxicity and can monitor other physiological motions. For many other applications, ultrasound techniques suffer from suboptimal tissue contrast and operator dependency.

Application of imaging in drug safety evaluation can be grouped into four categories: ex vivo specimen high-resolution imaging; in vivo animal imaging, usually of longitudinal nature; animal imaging for clinical imaging biomarker development and validation; and imaging in clinical trials. With the toxicities typically observed for NCDs, the brain, heart, liver and kidney are frequently affected. Imaging techniques can provide valuable and sometimes unique information of morphology and function of these organs.

Some structural and functional information is better acquired through imaging techniques. For example, the quantification of tissue lipid content is easier with MRS than traditional histology techniques. After organ toxicity has occurred, serum and urine assays can be normal due to the function reserve of the affected organs, on the other hand, imaging offers the possibility to provide region-specific information about tissue abnormality.

As opposed to terminal histology studies, imaging permits longitudinal studies of the same animals over an extended period of time, with individual animals serving as their own control, the number of animals required for a study can be significantly reduced while the intra-subject variability is minimized. Baseline data can be used either to select or de-select animals to be included in a study or as a basis for randomization between groups. This is also important if the compound is either costly or difficult to make at an early stage of its development when large-scale production method has yet to be established. At the end of the study, the animal is still available for other analyses, including histopathology.

In the preclinical drug safety evaluation, many studies are compliant with good laboratory practice (GLP). Most preclinical imaging scanners are not considered to be GLP compliant. In addition, in most of the imaging experiments animals are anaesthetized, and anaesthetic agents may have compounding interactions with the testing drugs. Therefore, imaging is more suitable with investigational studies where GLP is not obliged, though the GLP spirit is usually followed.

Ex vivo specimen high-resolution imaging

Specimen MRI using a high-magnetic field (4.7 Tesla and above) preclinical scanner can be a useful tool in toxicological studies (Delnomdedieu et al. 1996, Maronpot et al. 2004). Ex vivo specimen imaging is usually of a high-resolution nature as the confounding factor of anaesthesia is not a matter of concern, and long scan time can be applied. In addition, the inevitable physiological motion is removed which leads to better image quality. Isotropic spatial resolution of <100 μm can be easily achieved. Faber et al. (2003) reported MRI of the cardiac conduction system of formalin-fixed human hearts at 17.6 T. 3D data-sets with isotropic resolutions of 70 μm were acquired in six hours’ scan time. Image contrast allowed identification of the complete cardiac conduction system comprising AV-node, His’ bundle, bifurcation and left and right branch. Comparison of MR images with histological serial sections showed that diagnostic findings could be deduced from the MR data. In a study of bromobenzene-induced hepatocellular necrosis, Zhou et al. (1995) reported that MRI at 7 T allowed detection of liver damage at doses of bromobenzene where lesions were not evident in haematoxylin and eosin-stained sections. The contrast mechanism of MRI was believed to be due to the changes in local diffusion coefficients that accompany cellular degeneration and death.

For the quantification of pathological changes, the traditional histology approach involving tissue processing and sectioning can alter the integrity of the tissue, difficulties in precise orientation of histological sectioning during trimming, embedment and microtomy can confound morphometry measurements. With non-destructive 3D imaging, volumetric measurements of tissue structures can be made with high precision and reproducibility. In a study carried out by Johnson et al. (2006), data from a rapid MR screening method showed a dose-related decrease in the of volume of whole brain, cerebrum and hippocampus in treated rats with methylazoxymethanol-induced neurotoxicity, and it was concluded that MRI offers advantages for in situ volumetric measurements.

Another advantage of high-resolution MRI relates to the time necessary to perform the procedures. MR images can be obtained much quicker than histological procedures that involve sample preparation and staining, sectioning, mounting and drying (Lester et al. 1999). Due to its 3D nature, application of imaging ahead of conventional histopathology offers a quick overview of the specimen. In a study of neurotoxicity evaluation of carbonyl sulfide in Fischer 344 rats using MRI, Sills et al. (2004) reported that one key area of toxicity (posterior colliculus) would have been missed if only standard histology methodologies were used. In a study of L-2-chloropropionic acid-induced rat cerebellar neurotoxicity using MRI, Williams et al. (2001) reported that the occurrence of T2W image hyperintensity in the forebrain led the authors to discover a new lesion in the habenular nucleus which had been missed in the initial histopathological evaluation. Therefore in selected investigational studies, high-resolution ex vivo 3D MRI can be a useful supplement to standard histology studies, as it performs better in volumetric measurements and can offer valuable guidance in choosing the locations from which additional histology sections are made. The success of this aspect of application depends on the availability of a high-field MR scanner and close interaction between imaging scientists and histopathologists.

Longitudinal in vivo imaging studies of organ toxicity

In vivo animal imaging has found important usage in many aspects of drug safety evaluation. The occurrence and progression, or regression, of tissue abnormalities can be studied with repeated in vivo non-invasive imaging, though in some cases histological sampling is required to confirm imaging changes. Using histology approaches alone, a large number of animals may be involved and the longitudinal change of tissue abnormalities cannot be followed in the individual animals. Will NCD-induced abnormalities start to regress immediately after the withdrawal of NCD, or will it continue to progress for a period of time? Are NCD-induced abnormalities completely reversible? These questions can be answered by imaging coupled with histology methods.

The liver is a major concern for NCD safety evaluation. It is the key organ for metabolizing or biotransforming NCDs, or, possibly, for the generation of toxic chemical substances that result in secondary safety concerns. The hepatotoxicity can display manifestations such as hepatic steatosis, glycogen deposition, hepatocyte necrosis and cholestasis. Imaging provides a valuable tool in safety studies when other biomarkers for toxicity, such as routine serum chemistry measures are not suitable. For example, hepatic steatosis, a common finding in drug safety studies, does not always correlate with elevations of hepatic serum enzymes. In some cases, drug-induced hepatic steatosis patients can present with a rapid evolution of severe hepatic failure, lactic acidosis and ultimately death (Diehl 1999). Use of non-invasive imaging to detect the onset, progression and recovery from such a finding could provide valuable information in safety evaluation of NCDs.

Many studies have confirmed the correlation between liver fat/water signal ratio measured by in vivo MR and ex vivo analysis (Szczepaniak et al. 1999, Hocking et al. 2003). Hocking et al. (2003) measured the fat/water ratio in the livers of Zucker rats by MRS and found a high correlation between MR determined fat/water signal ratio and the fractional volume of intra-hepatic fat determined by histology (r = 0.89). Using MRI, it is also possible to evaluate normal versus fatty liver by exploiting the chemical shift difference between fat and water protons. Zhang et al. (2004) used the 3D three-point Dixon MRI method to quantitatively evaluate the fatty livers of rats induced by an experimental microsomal transfer protein inhibitor. In animal experiments, it is important that the anaesthetic agents are carefully selected as some anaesthetic agents are hepatotoxic. In the clinical settings, it has been demonstrated that MRI is able to detect fatty infiltration of the liver (Heiken et al. 1985), and MRS is a reliable method for quantifying liver steatosis in humans (Longo et al. 1995).

Liver hypertrophy is also a common side-effect caused by a wide variety of compounds. Because it is often the first indication of the hepatocarcinogenic potential of a NCD, liver weight is monitored in safety studies. Imaging-derived liver volume can be used as a surrogate of liver weight. Hockings et al. (2003) reported liver volume measured by MRI correlated closely with postmortem wet weight (r = 0.99). Because liver volume can fluctuate during the day as glycogen levels change, for longitudinal studies animals should be imaged at the same time of day.

The heart has limited capacity to repair itself. Toxic findings in the heart can be serious. When results from the standard tests, including blood pressure, heart rate and electrocardiogram (ECG) raise concern during a NCD safety evaluation, additional cardiac function tests need to be carried out. Echocardiography has been widely used in preclinical (Mimbs et al. 1981, Hanai et al. 1996) and clinical drug safety evaluation for NCD cardiotoxicity (Bianchi et al. 2003). Echocardiography can be used to measure the cardiac wall thickness, lumen volume and cardiac output. It has further advantages that it provides low cost, real-time images. In addition, in some laboratory animal species, it would be possible to carry out assessment of cardiac functions by ultrasound without the application of anaesthesia (Wang et al. 2007). MRI has also been used to determine myocardial volume, wall thickness, left and right ventricular end-diastolic and endsystolic lumen volumes, stroke volume and ejection fraction in animals and in humans. MRI cardiac exams in the clinic are usually conducted using breathhold. For animal MRI studies, anaesthesia is required for image acquisition. However, most anaesthetic agents cause cardiac and respiratory depression, for longitudinal studies the depth of anaesthesia must be kept consistent.

While electrical activities of the heart can be monitored in clinical trials by ECG, due to delayed release of serum markers of cardiac damage, structural histopathology, such as cardiomyocyte inflammation, degeneration and necrosis lack conventional early biomarkers. For investigative studies, imaging can provide valuable information in assessing myocardial perfusion and myocardial viability (Wassmuth et al. 2001, Thompson et al. 2003).

Animal imaging for clinical imaging safety biomarker development

If toxicities are observed in animals for a NCD, two questions arise: will these toxicities also occur in human subjects? can these toxicities be monitored and managed in clinical trials? The absence of a reliable clinical safety biomarker can lead to the withdrawal of an attractive compound from further study. Imaging can provide region-specific information of lesions, and offer the opportunity to carry forward essentially the same methodology in animal experiments into human studies. Imaging is an especially valuable tool for understanding the reversibility of NCD-induced lesions - a key question in drug safety evaluation.

For novel imaging techniques to be used to monitor toxicology in clinical trials, imaging safety biomarker development needs to be carried out in animals with the NCD in question. A good understanding of the NCD's toxicological effects on organs with available information of histopathology and serum and urine assays is vital for developing imaging safety biomarkers. During safety imaging biomarker development, imaging techniques and data acquisition parameters, image analysis approaches can also be tested and validated for further clinical references.

Vigabatrin, an irreversible inhibitor of gamma-aminobutyric acid transaminase, is an effective treatment for refractory epilepsies. Animal toxicology studies showed that chronic administration of vigabatrin induces intra-myelinic oedema and microvacuolation in discrete brain regions in rats and in dogs (Butler et al. 1987, Butler 1989, Peyster et al. 1995, Yarrington et al. 1993). In a study of rats treated with vigabatrin, cerebellar lesions were detected by MRI and these changes correlated with histopathology (Jackson et al. 1994). Peyster et al. (1995) investigated the reversibility of vigabatrin-induced neuropathy in Beagle dogs. Pathological findings were observed after 4-5 weeks of vigabatrin (300 mg/kg/day) treatment, which correlated with an increase in T2W MRI signal intensity. When vigabatrin treatment was withdrawn after 12 weeks’ administration, both MRI T2W high signals and pathological changes began to decrease. Sixteen weeks after vigabatrin withdrawal, histopathology had returned to normal and MRI demonstrated a trend towards reversal of the increased T2W signal intensity. It was concluded that MRI has the potential as a safety biomarker for the surveillance of vigabatrin-induced neuropathy.

The kidneys act as a filter for the NCDs and its metabolites. One of the most common observations associated with NCD-induced nephrotoxicity is acute renal failure. Many non-invasive tests of kidney function can only show renal damage after the functional reserve had been eliminated. This reserve can compensate for up to 75% of the loss, which make many serum and urine biomarkers insensitive for early kidney damages. MRI can offer advantages over methods that measure global functional changes by providing anatomically specific information of kidney injury. Numerous MRI methods exist for measuring renal perfusion, including dynamic contrast-enhanced MRI and the arterial spin labelling protocols. If decreased renal blood flow is a possible outcome observed during animal testing, then an MRI renal blood flow safety biomarker is attractive, given that these MRI methods can be easily translated from animal models to human.

A wide range of compounds can cause renal papillary necrosis (RPN) (Bach & Nguyen 1998, Brix 2002). In the early development of RPN, there are few clinical symptoms and specific urine or blood biomarkers. The progression of renal damage is insidious and renal function may be severely compromised before the condition becomes obvious. The diagnosis of RPN tends to be made in the late stages of this disease after irreversible destructive changes have occurred. The use of imaging modalities has led to an increased positive diagnosis in human population (Bach & Nguyen 1998). In a recent study, it was found that MRI was able to detect N-phenylanthranilic acid-induced nephrotoxic changes in rats ahead of papillary necrosis (Wang et al. 2006). In human data, Lang et al. (2004) reported that contrast-enhanced multiphasic CT scan can identify early manifestations of RPN and medullary necrosis, and CT scan can further be used to monitor lesion progression or regression after treatment. However, due to the consideration of radiation and the nephrotoxicity of iodinated X-ray contrast agents, MRI should be preferred for clinical studies. Unlike CT which showed little sensitivity for detecting RPN without contrast administration (Lang et al. 2004), MRI has the potential to provide drug-induced nephropathy information without contrast enhancement (Wang et al. 2006).

In practice, there is no clear cut relationship between in vivo longitudinal animal imaging and clinical imaging safety biomarker development. The findings in longitudinal animal imaging can easily be translated into clinical studies. It is important to note that, whereas other mammals have the same general circulatory pattern, organs, tissues and subcellular fractions as humans, the responses to drugs in laboratory animals can sometimes differ from human responses in a completely unpredictable way.

Imaging in clinical trials

There is huge potential for imaging in drug safety evaluation during clinical trials. In preclinical studies, although in many cases drug safety information is better obtained through imaging as discussed above, these information may also be obtained by histopathological means. In clinical trials, imaging can sometimes be the only practical mean to obtain drug safety information (Kellie et al. 2005).

In the example of vigabatrin, preclinical animal toxicity studies demonstrated MRI has the capability to monitor vigabatrin-induced central nervous system damage. This enabled further clinical trials of vigabatrin. However, throughout development and post-marketing phase, MRI and neuropathological studies of patients exposed to long-term vigabatrin treatment have provided no evidence of intramyelinic oedema and microvacuolation in the human brain (Cohen et al. 2000), and clinical experience demonstrated that vigabatrin provides effective seizure control with a wide margin of safety. Later, it was concluded that the neurotoxicity of vigabatrin bears a species specificity, with rats and mice being highly susceptible, dogs moderately susceptible and primates and humans not significantly affected at all.

In a recent example (Seiderer et al. 2005), MRI has been proved as a valuable tool for monitoring the liver toxicity of 6-thioguanine (6-TG), an effective treatment option for inflammatory bowel disease (IBD). Despite promising clinical data, there has been a rising concern regarding potential hepatotoxic side-effects of 6-TG, which lead to the development of liver nodular regenerative hyperplasia (NRH). Seiderer et al. (2005) conducted a multicentre safety study in IBD patients treated with 6-TG to investigate hepatic changes by ultrasound-guided liver biopsy and MRI. Forty-five patients treated with 6-TG (40-80 mg/day) for at least eight weeks were enrolled. MRI demonstrated a sensitivity of 77% and a specificity of 72% in the detection of histopathological findings consistent with and/or possibly related to NRH. Furthermore, MRI gives information on other potential NRH-associated complications such as splenomegaly, portal hypertension and ascites. In their study, NRH was also found in patients who had completely normal laboratory results. This stressed that patients on 6-TG therapy should undergo safety evaluation even in the absence of laboratory changes.

In conclusion, in preclinical drug safety evaluation imaging can be used at its most efficient where there is a specific endpoint which needs investigating or when the toxicology endpoints can be conclusively characterized by imaging. Currently, imaging techniques remain expensive especially with MRI and PET. Their high cost limits routine application for screening compounds. In many cases, histological sampling is required to confirm any imaging changes. There is huge potential for imaging in drug safety evaluation during clinical trials, as imaging can be the only non-invasive mean to obtain drug safety information. As novel technologies are further developed with functional and physiological capabilities, application of biomedical imaging in new drug safety evaluation is expected to further expand.