**1. Introduction**

The growing technological development improved diagnostic imaging techniques allowing early disease detection and diagnosis [1–4]. Even if different imaging modalities are extensively used in clinical practice such as magnetic resonance imaging (MRI), computed tomography (CT), positron emission tomography (PET), single-photon emission tomography (SPECT), each one presents strong points and limits. Nuclear medicine imaging techniques (PET and SPECT) are highly sensitive (pM range) and quantitative but suffer from poor resolution (mm range) [5,6]; CT is widely available and can detect several pathologies through rapid examinations and easy three-dimensional (3D) reconstructions but radiation dose to the patient is a noticeable concern and it is limited in soft-tissue resolution [7]; MRI gives high resolution, anatomical information, and good soft-tissue contrast but has low sensitivity (mM) [8–10]. Table 2 summarizes imaging modalities and related features. Since no single imaging modality allows gathering all the necessary morphological and functional information, the combination

of two or more imaging techniques, also called *multimodal imaging* or *hybrid imaging*, can offer synergistic advantages over any modality alone [11], overcoming its drawbacks and strengthening the peculiarities. The traditional approach was directed to the integration of a structural imaging modality (CT, MRI) with a functional highly sensitive imaging modality (PET/SPECT). Thus, firstly, PET/CT and SPECT/CT were introduced in clinical settings. The first PET/CT scanner was developed in 1998 by Townsend and colleagues [12] and was commercialized in 2001. It consists of a PET component independent from CT, and a single bed moves axially into the scanner while the patient sequentially performs CT and PET scans [13]. To date, PET/CT scanners completely replaced standalone PET scanners [14], exploiting anatomical reference and attenuation estimation from CT data. The success of PET/CT scanners inspired the feasibility of a PET/MRI scanner [15]. Three different configuration options were developed over the years [16]: the first consists of a sequential acquisition, similarly to PET/CT, where the patient undergoes firstly a MRI scan and later a PET scan; even if the MRI and PET components must be minimally modified, two consecutive acquisitions are performed without simultaneity. Temporal mismatches between PET metabolic data and MRI morphological information such as patient motion are the main weak points [7]. Nearly 15 years ago, some researchers working in preclinical settings analyzed the possibility of integrating a modified PET scanner into an MRI system. In Tubingen, Germany, an MRI-compatible PET scanner was inserted into a 3T clinical MRI scanner [8]; this system is suitable for preclinical studies or human brain imaging. The third option considers a first fully integrated whole-body PET/MRI system, with MRI-compatible photodiodes (avalanche photodiodes) and MR-based attenuation correction, and it became commercially available in 2010. It is worth noting that, in addition to anatomical information, MRI also provides functional information such as diffusion-weighted imaging (DWI), blood level oxygen-dependent (BOLD) imaging in functional MRI (fMRI), T1/T2 mapping, perfusion imaging and spectroscopy, and dynamic contrast-enhanced (DCE) imaging. PET and MRI can take reciprocal advantages: MRI anatomical data are useful for correction of the partial volume effect caused by PET [17], enable motion correction, improve arterial input function characterization for PET kinetic modeling, and are used as priors in PET iterative reconstruction; on the other hand, PET provides molecular information and highly sensitive quantification [18]. Furthermore, the simultaneous acquisition of fMRI data and metabolic PET information can investigate the coupling between metabolic demand and functional activity of the brain, since oxygen and glucose metabolism are strongly related to cerebral blood flow that delivers O2 and glucose to tissues [19]. High resolution and high tissue contrast, as well as multiparametric, functional, and quantitative imaging, supply complementary information for breast, head and neck [20], liver, musculoskeletal, and brain tumors [21] and heart [22] imaging. Hybrid imaging spread goes hand in hand with molecular imaging development, where molecular imaging stands for "in vivo" visualization, characterization, and measurement of biological processes at the molecular and cellular levels [23,24]. So far, various molecular imaging modalities were exploited not only for disease diagnosis, stratification, and treatment assessment [25] but also for image-guided therapy. Molecular imaging involves administration of imaging probes and detection of signals produced from the probes [26] and plays a key role in understanding important pathophysiological principles of diseases. In this context, personalized medicine aims to identify the adequate treatment and control its therapeutic efficacy. Suitable imaging probes are currently being developed and represent an exciting challenge for chemists and imaging scientists [27]. In this review, we focus on nanoparticle (NP)-based PET/MRI multimodal tracers in oncological imaging. A few of them were broadly tested in preclinical studies and show promising results in tumor detection, staging, and grading.

#### *1.1. MRI Contrast Agents*

In imaging, the term "contrast" refers to the capability of distinguishing between two adjacent structures; a contrast agent (CA) increases image contrast and highlights organs or blood vessels. The most common CAs used in X-ray or CT are iodinated and produce a direct effect on the image since they attenuate the X-ray beam, thereby increasing the signal intensity [20]; MRI CAs produce

an indirect effect as they influence the relaxation times T1 and T2 of the neighboring water molecules. MRI CAs can act by reducing T1 or T2: the former are called T1-weighted CAs since they reduce the T1 relaxation time and brighten the resulting image, the latter are called T2-weighted CAs since they reduce T2 relaxation time and darken the resulting image [21]. T1-weighted CAs are paramagnetic lanthanide compounds like gadolinium (Gd3<sup>+</sup>) and manganese (Mn2<sup>+</sup>) chelates, while T2-weighted CAs are superparamagnetic agents like iron-oxide NPs [22]. CAs acting on both T1 and T2 relaxation times are called "dual mode" and are NP-based. MRI CAs available for the clinical practice are reported in Table 1.

The first requirement for a very efficient CA is a high relaxivity; this parameter indicates the efficiency in reducing T1 or T2 relaxation time of the surrounding water protons. Paramagnetic metal ions like Gd3<sup>+</sup> cannot be used as CAs in their ionic form since their accumulation in specific tissues, for example, kidneys, liver, spleen, bone marrow, and the lymphatic system [28], causes toxicity. This challenge can be addressed by using chelators which hide the Gd ion through coordination bonds and are less likely to release it, conferring thermodynamic and kinetic stability and, therefore, less likely to induce toxicity. In particular, CAs based on Gd chelates are strongly associated with nephrogenic systemic fibrosis (NSF) in patients with renal impairment; the disease observed seems to be due to Gd release by chelating molecules in renal compartments [24]. In addition, recently, they were demonstrated to also accumulate in brain and kidneys in healthy patients [29,30].


**Table 1.** Magnetic resonance imaging contrast agents.

To improve diagnostic efficacy and reduce the nephrotoxic effects, an ideal CA should be stable, biocompatible, not toxic, and specific; it should remain within the system for a sufficient time to produce desired effects, such as tumor accumulation for oncological imaging, but should also be excreted from the body to minimize unwanted effects of foreign materials within body. In addition, higher relaxivity suggests a lower CA dose in patients. Most CAs currently used (typically small Gd3<sup>+</sup> ion chelates) lack in specificity because they are confined in the vascular space and do not accumulate in a specific tissue. It is not a coincidence that, in the last decade, CAs were refined by optimizing the relaxivity and developing amplification strategies aimed at increasing probe accumulation at the target site [25]. Moreover, the recent development of molecular and cellular imaging led to the recognition of NPs as MRI CAs.


**Table 2.** Molecular imaging modalities.

n.c., not well characterized.
