Molecular imaging : The convergence of form and function

© www.appliedradiology.com May 2018 www.appliedradiology.com Radiologic exams have become indispensable to the diagnostic process because they offer a noninvasive glimpse inside the human body. For many diseases, imaging facilitates diagnosis without the need for exploratory surgery or tissue sampling. Nonetheless, most radiologic exams are only capable of identifying the gross anatomic manifestations of the underlying microscopic derangements. As a result, radiologists are adept at identifying disease once it creates macroscopic structural changes, but they are less able to detect early or subtle changes or functional aberrations, as with pancreatic cancer, colon cancer and early diabetes.1 Without information regarding changes occurring at the cellular and molecular levels, the differential diagnoses of a given structural abnormality can be extensive. A 4-mm lung nodule or a subcentimeter liver mass can present challenges both for diagnosis and for patient management.2 Also, traditional methods of monitoring disease can prove inadequate in the context of new and emerging treatments, where interval enlargement and contrast enhancement may not always indicate viable tumor. For instance, tumors treated with Y-90 radioembolization and drugs such as bevacizumab (Avastin®) can appear to enlarge or enhance due to edema, inflammation, or granulomatous tissue even in the absence of residual viable tumor.3-6 Recognition of these limitations has led to efforts to extract more information from structural imaging exams and closely correlate those findings with clinical implications. Studies have focused on correlating information derived from new magnetic resonance imaging (MRI) sequences with symptoms and pathologic findings.7-9 Efforts to better align cross-sectional imaging findings with cancer outcomes have also led to improvements in RECIST (ie, Response Evaluation Criteria In Solid Tumors, a system of rules for assessing solid tumor response to treatment on imaging exams).10-12 The most dramatic manifestations of this movement are the emerging fields of radiomics and radiogenomics, which concentrate on identifying quantitative imaging features that correlate with disease and gene expression.13-16


R
adiologic exams have become indispensable to the diagnostic process because they offer a noninvasive glimpse inside the human body.For many diseases, imaging facilitates diagnosis without the need for exploratory surgery or tissue sampling.Nonetheless, most radiologic exams are only capable of identifying the gross anatomic manifestations of the underlying microscopic derangements.As a result, radiologists are adept at identifying disease once it creates macroscopic structural changes, but they are less able to detect early or subtle changes or functional aberrations, as with pancreatic cancer, colon cancer and early diabetes. 1ithout information regarding changes occurring at the cellular and molecular levels, the differential diagnoses of a given structural abnormality can be extensive.A 4-mm lung nodule or a subcentimeter liver mass can present challenges both for diagnosis and for patient management. 2 Also, traditional methods of monitoring disease can prove inadequate in the context of new and emerging treatments, where interval enlargement and contrast enhancement may not always indicate viable tumor.][5][6] Recognition of these limitations has led to efforts to extract more information from structural imaging exams and closely correlate those findings with clinical implications.[15][16]

Molecular imaging (MI) and precision medicine
These advances portend an emerging paradigm shift from traditional anatomic imaging towards using radiologic exams to interrogate disease on genetic, molecular, and pathophysiological levels.This burgeoning field is called molecular imaging (MI), and cancer is its initial focus.Molecular imaging promises to deliver a more precise diagnosis based on a specific molecular disease process (eg, genetic mutation) rather than a phenotype (eg, lung mass).Such precision will guide optimal treatment selection and allow for better monitoring of disease progression MOLECULAR IMAGING: THE CONVERGENCE OF FORM AND FUNCTION and treatment response.This approach aims to limit patient exposure to unnecessary treatments and side effects.It could also check healthcare costs by minimizing therapeutic trial and error.8][19][20][21][22] This movement toward more personalized medicine is fueled by the Precision Medicine Initiative, also known as the All of Us Research Program, which sponsors a large trial to match targeted treatments to cancer patients based on their specific genetic alterations (Figure 1). 23s MI emerges on the clinical scene, the role of the radiologist will be to provide an integrated assessment of tumor response that incorporates changes in size and enhancement with MI features.Interpretation will depend on familiarity with the particular molecular biomarkers of a given disease and their significance.Radiologists will learn the appearance of pathology using the new MI contrast agents, their enhancement kinetics, and their limitations within the context of structural imaging.Understanding available molecularly-targeted therapies and the role of MI examinations in predicting which patients will respond to these treatments will also enhance their value.Finally, a familiarity with the clinical trials of emerging targeted treatments, their imaging requirements, and the implications for surveillance and prognosis will help radiologists provide interpretations with greater clinical relevance.

Past strides towards MI
The initial pioneering in MI occurred predominantly in nuclear medicine (PET and SPECT) because of its sensitivity to picomolar amounts of radioactive substances and facility of incorporating radioactive elements into biologically relevant substances for imaging.5][26][27][28][29][30][31][32][33][34][35] These modalities are limited by low spatial and temporal resolution.The fusion of PET and SPECT with computed tomography (CT) or MRI helps, but it requires additional hardware and is subject to misregistration artifacts.Even so, nuclear medicine is ripe for the emergence of more innovations in MI.
Optical imaging is a form of MI that has found wide laboratory applications

MOLECULAR IMAGING: THE CONVERGENCE OF FORM AND FUNCTION
because it permits longitudinal in vivo visualization in both cells and small animals. 36,37Unfortunately, there are few human applications because the energy emitted by fluorophores can penetrate only millimeters of tissue, at best.9][40][41] However, there are no clinically relevant radiologic applications.This article will focus on the nascent field of MI agents for MRI, CT and ultrasound (US).The field faces daunting challenges, but driving this pursuit is the promise of revealing diseases based on more than gross anatomic appearance alone.

Towards smarter contrast agents
Molecular imaging agents have been termed "smart contrast agents," since they accumulate in specific targets or in the presence of a particular pathologic process (Figure 2A-C).Conventional iodinated contrast for CT and gadolinium (Gd) chelates for MRI are largely bloodpool agents with enhancement depending on tissue vascularity and integrity.Leaky vessels due to cancer, trauma, or inflammation permit passive accumulation of these agents (Figure 2D).3][44] Smart contrast agents selectively accumulate at their target and their signal can even be activatable (Figure 2A-C).As such, MI agents may become the in vivo imaging corollary to special stains or antibodies used histologically.This goal is fraught with challenges, including efficient delivery, avoidance of nontarget uptake, and adequate sensitivity, to name a few.Appreciating these challenges requires a basic understanding of the underlying molecular biology.

Nuts and bolts of MI
Radiologists familiar with the principles and language of molecular biology will lead the way in interpreting these emerging MI exams.Rapid advances in genetics have led to discoveries far beyond the basic understanding of DNA translation into RNA, which subsequently directs transcription into proteins.Cancer and other diseases can occur as a result of aberrations at any of these levels that manifest during synthesis or through post-synthetic modifications.DNA and proteins have received much attention, but research is focusing increasingly on RNA.There are various types of RNA, including messenger RNA (mRNA), transfer RNA (tRNA), and ribosomal RNA (rRNA), all of which mediate the process of translation.
An important noncoding form of RNA is microRNA (miRNA), which controls gene expression after transcription and plays a role in oncogenesis.6][47][48][49] Studies linking them to hepatocellular carcinoma (HCC) have resulted in the development of a system of molecular classification of HCC based on miRNA. 50,51][57][58][59][60] Proteins can have a variety of functions, including serving as ligands or modulators of signal transduction, receptors, or enzymes.Their amino acid composition and morphologic shape dictate their function.Because of the accessibility of proteins located on the cell surface or in the extracellular space, they represent common targets of MI agents. 61,62Proteins are also used to make MI agents.Antibodies can make excellent MI agents due to their comparatively high binding specificity (Figure 2C).However, cost and loss of signal due to nonspecific uptake by organs of the reticuloendothelial system or proteins such as albumin potentially limit their utility.
An attractive candidate for MI agents are nanoparticles, a diverse group of molecules characterized by their 10-200 nm size and customizable shape and composition. 63,646][67][68][69][70][71][72][73][74] They can be decorated with ligands or receptors for targeting purposes. 75][78] Adding to their precision is the potential to engineer them to control the release of this payload over time. 79,80

MI agent design
Fundamentals that guide MI agent design and utilization are how the agent reaches its target from the bloodstream, how it accumulates in the tissue, and how it is detected.Understanding these properties and their impact on the signal-to-noise ratio (SNR) is critical to interpreting patterns of enhancement.MI agents that travel within the bloodstream can behave like conventional untargeted agents within the first few minutes after administration; however, with time, they exit the blood stream to accumulate in tissues based on physical factors and biochemical interactions (Figure 3).Time points at least 6-12 hours post MI agent administration may offer the highest SNR.
Small molecules like nanoparticles can passively cross the blood brain barrier, cornea, and skin. 81,82They can also traverse the leaky vasculature of tumors, which is termed the enhanced permeability and retention effect (EPR).EPR is even more pronounced with agents up to 400 nm and those with longer circulation times. 83Tissues with a high concentration of phagocytes, such as those in the reticuloendothelial system, tend to accumulate MI agents like nanoparticles; if these tissues are not the target, then uptake contributes to background or noise. 84,85Methods to mitigate this kind of nontarget uptake Biochemical interactions with target molecules can enhance the retention of MI agents in target tissues while allowing them to wash out of nontarget tissues.These interactions can improve specificity and SNR.MI agents may be functionalized to interrogate protein expression by interacting with a specific receptor, enzyme, signal transduction regulator, or other ligand (Figure 2A-C).One example is the conjugation of an MRI contrast agent with a HER2 antibody for breast cancer imaging. 19However, images derived from antibody-based MI agents are susceptible to noise if there is high nonspecific binding of the antibodies to "sticky" proteins. 86,87nce MI probes reach their target, optimizing signal strength and detectability presents another hurdle.The challenge is to accumulate enough probe in the target tissue so that it is visible compared to background.Each modality must overcome inherent limitations to achieve this goal but detectability ultimately depends on the amount and strength of the signal in a volume of tissue.There is a limit to how much detectable substance can be packed into a small volume, so an alternative is to augment the signal emitted by a few molecules, termed signal amplification.

FIGURE 4. Enhancement patterns in vascular non-tumor tissues by contrast type. (A) Schematic of the enhancement characteristics of a representative vascular non-tumor tissue after intravenous administration of conventional untargeted contrast (row 1) versus tumor-targeted MI agents (rows 2, 3) at four time points before (column 1) and after (columns 2, 3,4) contrast administration. All vascular non-tumor tissues demonstrate no enhancement pre-contrast (row 1). After conventional contrast agent administration, the vascular non-tumor tissue enhances brightly in the arterial phase, washes out in the venous phase and is cleared from the tissues by 12 hours. The constitutively active MI agent behaves like a weak blood pool agent during the arterial phase. By the venous phase, the agent is washing out of the vascular nontumor tissue but surrounding enhancement related to perfusion is also present. By 12 hours, the MI agent has washed out of all non-tumor tissues. By contrast, the activatable MI agent is only detectable once it interacts with the target. Since the tissue depicted represents non target tissue, there is no enhancement at any of the time points represented. (B) Illustrative graph of the enhancement characteristics of vascular nontarget tissues after intravenous administration of untargeted contrast (blue) versus tumor-targeted MI agents (yellow, red) at four time points before and after contrast administration. All nontarget tissues demonstrate no enhancement pre-contrast, yielding a target signal-to-background noise ratio of 1. After untargeted contrast agent administration (blue), nontarget vascular tissue enhances brightly in the arterial phase and gradually washes out in delayed phases until it has been cleared 12 hours later. The constitutively-active MI agent behaves like a weak blood pool agent during the arterial phase and washes out on delayed points (yellow) similar to untargeted contrast agents (blue). Activatable MI agents are only detectable when interacting with their target, hence not displaying any nontarget tissue enhancement at any points.
Signal amplification can occur through the interaction of an MI agent with a target or in the presence of a specific environment.MRI probes can be designed to undergo conformational or relaxivity changes in certain microenvironments that enhance their visibility. 88,89For example, the acidic conditions inside tumors can induce an MI probe to dissociate from the iron oxide quenching it, thereby unveiling its T1-shortening Gd molecule for MRI. 90,91nother approach uses the chelate EgadMe to shield Gd from water until β-galactosidase cleaves and unveils it, enabling the Gd to interact with water and produce T1 shortening. 92,93ctivatable MI agents represent a different strategy aimed at improving the SNR (Figures 3, 4).For these agents, a

FIGURE 5. Activatable cell penetrating peptide dendrimers conjugated to gadolinium (ACP-PD-Gd) are activated through MMP-cleavage and deposit in areas of high MMP expression, such as aggressive tumors. Mice harboring an aggressive mammary tumor line expressing high levels of MMP were imaged by T1-weighted MRI without contrast or after administration of either ACPPD-Gd or commercial Gadobutrol. (A) ACPPD-Gd administration results in more robust and conspicuous enhancement of small tumors less than 5 mm 3 compared to those imaged after Gadobutrol or without contrast (white arrows); this effect is secondary to high accumulation of ACPPD-Gd from MMP cleavage. (B) Mice were imaged by MRI with ACP-PD-Gd at 4, 9, and 13 days after intramammary fat pad injection of a highly aggressive tumor line. At day 4, a small linear strip of enhancement appears which rapidly grows into a larger enhancing tumor at days 9 and 13 (white arrows). (Reproduced with permission from PLoS One. 74 )
A B

MOLECULAR IMAGING: THE CONVERGENCE OF FORM AND FUNCTION
receptor-ligand interaction or enzyme in the target releases a trapped signal or converts an inert molecule into a detectable one (Figure 2A,B).The fluorescent signal of an optical imaging agent is activated when cleaved by the matrix metalloproteinase (MMP) present in metastasizing tumors. 94An enzyme can also induce the oligomerization of paramagnetic substrates to render an MRI agent visible. 95n this sense, activatable MI agents can act as "switches" to monitor the molecular underpinnings of physiology and pathology.Others can also be activated externally with the application of energy in the form of US, light, or heat. 96,97is method permits regional anatomic control over MI agent activation.

Emerging MI agents by modality Magnetic resonance imaging (MRI)
Magnetic resonance imaging achieves inherently superb soft tissue contrast, so molecular MRI stands to combine form and function in a powerful way.However, MRI faces the challenge of detecting MI agents that are usually in nanomolar concentrations, whereas the lower limits of sensitivity for Gd detection is in the micromolar range. 980][101] Other T1-based agents such as manganese (Mn) have also been employed. 102,1032-shortening agents primarily consist of iron oxide nanoparticles and have been clinically utilized to detect occult lymph node metastases in prostate cancer. 845][106] Other advanced MRI techniques with potential MI applications, such as chemical exchange saturation transfer (CEST), hyperpolarized MRI, and MR spectroscopy, are beyond the scope of this review but have been described previously. 107everal MI agents for MRI have been developed in preclinical settings and have a wide range of potential applications.Among the first were fibrin-targeted MI agents to detect thrombin, with applications for imaging cardiovascular disease. 108,109Molecular imaging probes targeted to integrins could prove useful in MRI to detect angiogenesis and monitor anti-angiogenic cancer therapies. 110,111Detection of surface phospholipids as a proxy for programmed cell death (ie, apoptosis) can also play a role in cancer and drug development. 112,113inally, MMPs have been frequent targets for MI agents because of their presence in aggressive and metastatic cancers (Figure 5). 41,73,74,114,115

Computed tomography (CT)
Computed tomography is less suitable for MI since its sensitivity is at the millimolar scale, which is significantly lower than MRI (micromolar range), US (single microbubble), or nuclear medicine (picomolar range).Similar limitations apply to fluoroscopy.][118][119][120] Multispectral CT imaging of macrophage burden in atherosclerotic plaques has been performed with gold-conjugated high density lipoprotein nanoparticles. 121Given its low sensitivity and growing concerns over radiation exposure, CT is not likely to emerge as a favored modality for MI applications.

Ultrasound (US)
Molecular imaging with US uses targeted or functionalized microbubbles as contrast agents.Ultrasound detection of microbubbles relies on their non-linear response to low energy sound waves (Mechanical Index <0.3). 122,123Microbubbles are generally composed of phospholipid shells encapsulating high molecular weight gases such as perfluorocarbons. 1246][127] In the context of MI, microbubbles can be coated in the same fashion as MRI agents to target specific ligands of interest.Microbubbles targeted to bind to cell adhesion molecules (eg, V-CAM and I-CAM) and selectins have been used to assess atherosclerosis and angiogenesis, respectively. 128,129ecently, a novel commercial microbubble agent BR55 ® (Bracco Suisse, Geneva, Switzerland) has been developed that specifically binds to vascular endothelial growth factor receptor-2 (VEGFR-2), which is the target of the antiangiogenic drug sorafenib (Nexavar ® ), used for HCC. 130,1312][133] This agent has recently been clinically translated for use in patients with ovarian, breast, and prostate lesions. 134,135High US energies can also be used to preferentially activate microbubbles and to disrupt endothelial linings to permit interrogation of areas beyond the vasculature. 136

Limitations and challenges of MI
A major technical hurdle of MI, especially with MRI and CT, is sensitivity.Strategies for improving SNR, including signal amplification, activatable MI agents, and improved MI agent targeting, will help overcome this deficiency.Target selection also represents a critical challenge as the field moves forward.Diseases and their pathophysiology are diverse and markers have different significance for different diseases.Even so, the resources involved in developing an MI agent may be prohibitive if it is only meant to target one specific disease subtype.A solution is to find targets that can be broadly applied to a set of diseases but become specific in the appropriate context of disease.Candidate targets include MMP, thrombin, and EGFR.Partnerships with pathologists, molecular biologists, and other scientists can ensure that clinically and biologically relevant markers are chosen for development.Molecular imaging exams are also limited by the number of targets they can interrogate at once, whereas histology or flow cytometry can assess for numerous markers.A more complete disease picture could come from integration of MI with radiomics and bigdata mining.Finally, MI agents could face stringent regulatory hurdles with variable approval from the U.S. Food and Drug Administration (FDA).Policy changes may have to evolve alongside this emerging technology.One option would be to develop FDA-approved protocols for generating customizable probes that can be made safely and affordably.Policies that facilitate rather than stifle innovations in MI will help propel the field towards more personalized medicine.

Conclusions
Since the first clinical use of x-rays, there have been stunning advances in the noninvasive visualization of anatomy and disease.In parallel, scientists have deepened their understanding of biology on the cellular, molecular and genetic levels.Molecular imaging hopes to capitalize on this knowledge and make key microscopic processes visible on clinical imaging exams.Radiologists are poised to usher in new MI techniques and enhance the value of their imaging interpretations.They will do so by understanding the molecular biology underlying these MI agents, including their potential applications and inherent challenges.The field is quickly evolving, so radiologists who partner with scientists, pathologists and other specialists will lead the way in developing and optimizing MI agents and their interpretation.Though the field faces significant hurdles, the promise of looking beyond structure to illuminate the functional underpinnings of disease makes it a challenge worth taking on.

FIGURE 1 .
FIGURE 1.The Precision Medicine Initiative through the National Cancer Institute sponsors a large trial (NCI-MATCH) to match targeted treatments to cancer patients based on their specific genetic alterations.Characterization of tumors with their molecular and genetic features through this initiative represents opportunities to develop complementary molecular imaging techniques.Infographic source: The National Cancer Institute.
THE CONVERGENCE OF FORM AND FUNCTION are modifications such as PEGylation or direct delivery to the target.

FIGURE 6 .
FIGURE 6. Microbubbles targeted to VEGFR-2 (BR55) can be used with ultrasound (US) to image VEGFR-2 expression in hepatocellular carcinoma (HCC) during Sorafenib treatment.Mice harboring HCC tumors were imaged 6 minutes after injection with BR55 at day 0 (A) and 14 (B) of Sorafenib treatment.(A) At day 0, high VEGFR-2 expression is evidenced by increased echogenicity on US due to high binding of BR55 microbubbles to the tumor.(B) After 14 days of Sorafenib, expression of VEGFR-2 has decreased secondary to treatment, resulting in minimal if any BR55 microbubble tumor binding and decreased signal on US. (Reproduced with permission from Springer Publishing and Molecular Imaging and Biology.