Rational Design of Magnetic Nanoparticles as T₁–T₂ Dual-Mode MRI Contrast Agents

Abstract

Magnetic nanoparticles (MNPs), either paramagnetic or superparamagnetic depending on their composition and size, have been thoroughly studied as magnetic resonance imaging (MRI) contrast agents using in vitro and in vivo biomedical preclinical studies, while some are clinically used. Their magnetic properties responsible in some cases for high magnetization values, together with large surface area-to-volume ratios and the possibility of surface functionalization, have been used in MRI-based diagnostic and theranostics applications. MNPs are usually used as positive (T₁) or negative (T₂) MRI contrast agents, causing brightening or darkening of selected regions in MRI images, respectively. This review focusses on recent developments and optimization of MNPs containing Gd, Mn, Fe and other lanthanide ions which may function as dual-mode T₁–T₂ MRI contrast agents (DMCAs). They induce positive or negative contrast in the same MRI scanner upon changing its operational mode between T₁-weighted and T₂-weighted pulse sequences. The type of contrast they induce depends critically on their r₂/r₁ relaxivity ratio, which for DMCAs should be in the 2–10 range of values. After briefly discussing the basic principles of paramagnetic relaxation in MNPs, in this review, the basic strategies for the rational design of DMCAs are presented and typical examples are discussed, including in vivo preclinical applications: (1) the use of NPs with a single type of contrast material, Gd- or Mn-based NPs or superparamagnetic NPs with appropriate size and magnetization to provide T₂ and T₁ contrast; and (2) inclusion of both types of T₁ and T₂ contrast materials in the same nanoplatform by changing their relative positions.

1. Introduction

Magnetic resonance imaging (MRI) is one of the most prominent clinical imaging modalities resulting from its many favorable characteristics. These include non-invasiveness, use of low-energy radiofrequency radiation, in contrast to invasive X-ray computed tomography (CT), positron-emission tomography (PET) and single-photon emission computed tomography (SPECT), which use damaging high-energy ionizing radiation, providing tomographic images with large penetration depth of any area of the body, in contrast to optical imaging (OI), outstanding spatial resolution (50–100 μm) and remarkable soft tissue contrast (see Table 1 for a comparison of some properties of imaging modalities for clinical applications). The contrast in the MRI images is generated by differences in intensities of ¹H NMR resonances, most frequently of water protons, with an important contribution of lipids in some cases. These intensities are governed by several parameters, including the local tissue ¹H concentrations, diffusion and flow of water molecules, and most importantly by their intrinsic differences in proton spin–lattice or longitudinal relaxation times (T₁) and spin–spin or transverse relaxation times (T₂) [1,2]. However, the sensitivity of MRI is relatively low, due to the small population difference between the two proton spin energy states in the presence of a magnetic field, making it difficult to detect small lesions and limiting the time resolution of the technique. In recent decades, there has been a trend toward clinical MRI equipment operating at higher magnetic field strengths (B₀), which improved its contrast-to-noise ratios and spatial resolution. Presently, medical diagnostic MRI is typically performed at B₀ = 0.5–3 T, and research equipment can operate at B₀ values up to 11 T.

MRI contrast agents (CAs) are often employed to enhance the contrast between normal and disease tissues by significantly decreasing their T₁ and T₂ values [3,4,5,6,7]. This acceleration of the spin relaxation of water protons, in the region where they accumulate relative to surrounding tissues, results from oscillating magnetic fields produced by metal ions present in the CAs, allowing sensitive MRI detection of that region. Their efficacy is measured by their relaxivity, r₁ or r₂, defined as the paramagnetic enhancement of the water proton relaxation rates, normalized to a 1 mM metal ion concentration [3,4,5].

Table 1. Comparison of some typical properties of imaging modalities for clinical applications [8,9].

Technique	Resolution	Penetration Depth	Sensitivity
MRI	50–100 μm	No limit	10−4–10−5 mM
PET	1–2 mm	No limit	pM
SPECT	1–2 mm	No limit	nM
OI	2–5 mm	<2 cm	<<nM
CT	50–200 μm	No limit	0.1 mM a

^a For an iodine-containing CA.

Table 1. Comparison of some typical properties of imaging modalities for clinical applications [8,9].

Technique	Resolution	Penetration Depth	Sensitivity
MRI	50–100 μm	No limit	10−4–10−5 mM
PET	1–2 mm	No limit	pM
SPECT	1–2 mm	No limit	nM
OI	2–5 mm	<2 cm	<<nM
CT	50–200 μm	No limit	0.1 mM a

^a For an iodine-containing CA.

A class of MRI CAs contains strongly paramagnetic metal ions, such as Gd³⁺, Mn²⁺ and Fe³⁺, encapsulated by a strongly binding chelating ligands, or in the form of paramagnetic nanoparticles (MNPs). These CAs increase the longitudinal relaxation rates of water protons significantly more than their transverse relaxation rates, giving rise to bright spots in T₁-weighted (T_1w) MRI images and are called positive or T₁ CAs [3,4,5,6,7,10]. Another class of CAs, containing paramagnetic metal ions, such as Tb³⁺, Dy³⁺ and Ho³⁺, in the form of paramagnetic chelates or NPs, or as superparamagnetic NPs, such as iron-oxide NPs (SPIONs), induce protons in their vicinity to preferably undergo spin–spin relaxation, originating negative (or dark) contrast in T₂-weighted MRI images and are called negative or T₂ CAs. In addition to r₁ and r₂, the r₂/r₁ ratio (always ≥1, as T₂ ≤ T₁) represents another important factor in identifying the class of MRI CAs. This ratio should be close to one for a T₁ CA (positive contrast), while it is large for a T₂ CA [10,11,12,13].

Nowadays, approximately 40% of the MRI exams are carried out after administration of a CA. The materials used as CAs for clinical studies must meet the following requirements: (1) high relaxivity; (2) high water solubility or colloidal stability; (3) low osmolality to avoid pain from osmotic shock and other adverse effects upon injection; (4) low toxicity, which requires both high thermodynamic and kinetic stability (no transmetallation by endogenous metal ions such as Zn²⁺) in the case of metal chelates, and no leaching of toxic free metal ions in vivo from NPs; (5) rapid excretion, ideally not much longer than the MRI exam time; (6) a biodistribution with high specificity for the area of interest.

Various MRI contrast agents have been investigated and developed to achieve this goal. Gd-based CAs (GBCAs) include macrocyclic chelates of open-chain DTPA-type (DTPA = diethylenetriaminepentaacetic acid) or macrocyclic DOTA-type (DOTA = 2,2′,2″,2′″-(1,4,7,10-tetraazacyclododecane-1,4,7,10-tetrayl) tetraacetic acid) derivatives, such as Magnevist^® (Bayer Schering Pharma AG, Berlin, Germany), Dotarem^® (Guerbet, Paris, France), Omniscan^® (GE Healthcare, Chicago, IL, USA) and ProHance^® (Bracco, Milan, Italy) [7], which are clinically used as T₁ MRI CAs. These molecular agents have low r₁ and r₂ values and short blood circulation times due to their efficient renal excretion, and thus large amounts of injection doses are needed (typically 0.1 mmol/kg body weight) to achieve detectable contrast levels. This can increase the risk of toxicity due to potential release of free Gd³⁺ ions in the body [14]. This has occurred in some cases to patients with chronic kidney disease, which developed nephrogenic systemic fibrosis (NSF), characterized by skin thickening and hyperpigmentation and extracutaneous fibrosis [15,16,17]. In addition, recent studies indicated that clinically developed GBCAs could be deposited in the brain after repeated use, and could cause neurotoxicity, although this has not been proven [18,19]. In both cases, tissue deposits of linear GBCAs are much higher than those of macrocyclic GBCAs.

Biocompatible dextran-coated superparamagnetic iron oxide nanoparticles (SPIONs) Feridex^® (Bayer HealthCare Pharmaceuticals, Wayne, NJ, USA), Sinerem^®(Guerbet, Paris, France), and Resovist^®(Bayer Schering Pharma AG, Berlin, Germany), have been approved by the Food and Drug Administration (FDA), USA, for clinical trials in liver (Feridex^® and Resovist^®) and lymph nodes (Sinerem^®) MRI, as well as Lumirem^® (Guerbet, Paris, France) for gastrointestinal imaging. However, Feridex^® and Sinerem^® were withdrawn from the market, mostly for commercial reasons. Nowadays, only Resovist^® is used for liver imaging in some countries [7,20].

Magnetic NP (MNP)-based MRI CAs have many advantages relative to metal chelates, including larger magnetic moments and longer blood circulation times, which cause higher image contrast. They can also be used as multifunctional nanoplatforms for multimodal imaging, therapy and drug delivery (theranostics) after surface functionalization. [21]. MNPs are composed of two parts: a magnetic core that enhances MRI and a surface-coating ligand layer responsible for colloidal stability and minimizes toxicity. This review focusses on recent developments and optimization and in vivo applications of MNPs containing Gd, Mn, Fe and other lanthanide ions which may function as dual-mode T₁–T₂ MRI contrast agents (DMCAs). As the kind of contrast provided by MNPs depends fundamentally on their r₂/r₁ ratios, the description of the principles for rational design of DMCAs starts by a brief summary of the basic theory of paramagnetic relaxation in MNPs. Then, as their r₂/r₁ ratios are critically dependent on the MNPs composition, size and surface-coating, the different ways in which these properties can be modulated to design T₁ or T₂ single-mode MRI CAs or T₁–T₂ DMCAs are illustrated using selected examples. As the MNPs composition is a very important factor, a particular attention is given to systems using a single type of contrast material (e.g., Gd- or Mn-based T₁ contrast materials or superparamagnetic T₂ contrast materials) or both types of contrast materials in the same nanoplatform.

2. Basic Principles of Paramagnetic Relaxation in Small Complexes and Nanoparticles

The observed longitudinal proton relaxation rate ($R_1^{obs}$) of a small paramagnetic complex in aqueous solution is the sum of a paramagnetic and a diamagnetic term, where the first is proportional to the concentration of the paramagnetic ion and the second is the contribution of the water solvent:

$$R_1^{obs}=R_1^p+R_1^d=r_1\left[\mathrm{M}\mathrm{n}\right]+R_1^d$$

(1)

The observed relaxivity (in mM⁻¹ s⁻¹ units) is the sum of the inner (r_1is) and outer sphere (r_1os) terms:

$$r_1=r_1^{is}+r_1^{os}$$

(2)

The same equations apply to R₂ and r₂. These contributions are evaluated using the dominant dipole–dipole relaxation mechanism, which is modeled by the Solomon–Bloembergen–Morgan (SBM) theory for r_1is, and the Freed theory for r_1os [3,4,5,6,22]. For r₁, the main parameters determining the IS contribution are the number of water molecules in the first coordination sphere of the metal ion (q), and the three processes responsible for the time fluctuation of the nucleus–electron interactions, which are the inner-sphere water residence time (τ_M = k_ex⁻¹, where k_ex is the water exchange rate), the molecular reorientational correlation time (τ_R) and the electron spin relaxation time (T_1e) of the metal ion, which is frequency dependent. The OS contribution is determined by the diffusion correlation time (τ_D) and the distance of closest approach (a) of the water molecules freely diffusing near the complex.

The r₁ and r₂ values of paramagnetic NPs also have in principle IS and OS contributions. The IS contribution results from the exchange of the water protons directly coordinated to the metal ions at the NPs surface with bulk water, as those located below the surface have a negligible effect. The IS r₁ depends on the hydration number of the surface ions and the NP’s surface-to-volume ratio [23]. In the case of Gd₃₊ ions, they affect the bound water proton relaxation mainly through the dipolar mechanism, which can be modeled by the SBM equations. When other paramagnetic Ln³⁺ ions with high magnetic moments and very short T_1e values (e.g., Tb³⁺, Dy³⁺, Ho³⁺) (

Table 2. Main chemical and physical characteristics of paramagnetic ions used in MRI CAs.

Metal Ion	Free Ion Ground State/Configuration	Spin-Only Magnetic Moment (μB)	Ionic Radius (Å)	Coordination Number	Electronic Relaxation Times (s)	M-Hw Distance (Å)
Mn2+	6S5/2 (3d5)	5.9	0.83/0.90	6/7 (high spin)	10−10–10−8	~2.8
Fe3+	6S5/2 (3d5)	5.9	0.64	6–7	10−11–10−9	2.6–2.7
Eu3+	7F0 (4f6) (5D0, 5D1)+	3.4 *	1.06/1.12	8/9	10−11–10−9	~3.0
Gd3+	8S7/2 (4f7)	7.9	1.05/1.11	8/9	10−10–10−8	~3.0
Tb3+	7F6 (4f8)	9.7	1.04/1.10	8/9	10−13–10−12	~3.0
Dy3+	6H15/2 (4f9)	10.6	1.03/1.08	8/9	10−13–10−12	~3.0
Ho3+	5I8 (4f10)	10.4	1.02/1.07	8/9	10−13–10−12	~3.0

* The magnetic moment of Eu³⁺ is not zero due to the contribution of the excited states ⁵D₀ and ⁵D₁.

) are present, a Curie term is also present [24,25,26].

However, for surface-coated NPs, the OS contribution dominates the T₁ and T₂ relaxation, as the water protons are indirectly in contact with the paramagnetic metal ions in a NP due to the surface coating. It results from the diffusion of water molecules in the fluctuating magnetic field inhomogeneities created in their vicinity by the magnetized NPs, and does not contain the Curie contribution. The r₁ value is approximately proportional to the square of the spin magnetic moment (μ_s² = S(S + 1)ħ² for transition metal ions and μ_s² = 4S(S + 1) + L(L + 1)ħ² for Ln³⁺ ions, where S is the spin quantum number and L is the orbital quantum number) multiplied by the number (N) of Ln³⁺ ions in a NP which can interact with a water proton, as given by Equation (3):

r₁ ∝ Nμ_s²

(3)

Transverse relaxation of water protons is induced by fluctuations of local magnetic fields generated by the NPs. Thus, r₂ is proportional to the square of the total magnetic moment (μ) of the NP, as given by Equation (4):

r₂ ∝ μ²

(4)

In the case of superparamagnetic NPs, the OS contribution is also dominant. However, for a colloidal dispersion in the presence of a magnetic field, the return of their magnetization to equilibrium is determined by two different processes [11]: (a) the Néel relaxation, describing the return of the magnetization of each of the NPs to equilibrium after a perturbation that tilts that magnetization away from the direction of its easy axis; its relaxation time (τ_N) defines the fluctuations that arise from jumps of the magnetization between different easy directions; (b) Brownian relaxation, defined by the relaxation time τ_B, which characterizes the viscous rotation of the particle. The global magnetic relaxation rate of the colloid is the sum of the Néel (τ_N⁻¹) and Brownian (τ_B⁻¹) relaxation rates, τ⁻¹ = τ_N⁻¹ + τ_B⁻¹, where τ is the global magnetic relaxation time. In these systems, r₁ and r₂ can be described by Freed’s model for paramagnetic systems, using the diffusion correlation time (τ_D) and considering the electron spin longitudinal relaxation time τ_S1 as equal to the Néel relaxation time τ_N.

The magnetization of the superparamagnetic iron oxide NPs reaches its saturation value M_S at B₀ ≤ 0.5 T, which is at the lower limit for commonly used clinical MRI scanners. Therefore, their R₂ is in practice usually independent of B₀. The r₁ value is governed by the volume fraction of the superparamagnetic particles (υ), the diffusion correlation time (τ_D = d²/4D, where d is the diameter of the particle and D is the diffusion coefficient), and the magnetization of the NP (M_S) at the B₀ value of the clinical MRI scanner. A discussion of the transverse relaxivity of spherical superparamagnetic NPs can be found in the literature [11,27,28,29].

Superparamagnetic NPs are the T₂ CAs most commonly used in MRI, usually with a very high r₂ (typically 60–400 s⁻¹mM⁻¹), in particular the SPIONs. This can hamper the interpretation of T_2w images due to the difficulty in distinguishing the CA-induced darkening from partial-volume artifacts, motion artifacts, and tissue inhomogeneities [30].

3. Magnetic Nanoparticles as T₁–T₂ Dual-Mode MRI Contrast Agents

3.1. NPs for T₁ or T₂ Single-Mode MRI Contrast

The type of contrast provided by MNPs as MRI CAs depends on their r₁ and r₂ values, as well as their r₂/r₁ ratios, which are dependent on their composition, size and surface-coating [31]. The ideal T₁ MRI CA should have high r₁ values and r₂/r₁ ratios close to 1.0. The most important contribution to the r₁ values comes from the metal ions present on the MNP surface, which interact directly with nearby water proton spins by an IS mechanism. The most efficient paramagnetic ions are Gd³⁺ (S = 7/2), Mn²⁺ (S = 5/2), and Fe³⁺ (S = 5/2) (S is the total spin quantum number) due to their high number of unpaired electrons, high magnetic moments and long T_1e values (Table 2) [3,4,5,6,22].

Several inorganic Gd³⁺-containing NPs with the largest possible Gd³⁺ densities [32], such as Gd₂O₃ [33,34], Gd₂O₂S [35], Gd-carbonates [36], GdF₃ [37,38], and GdPO₄ [39,40], have been proposed as potential T₁ MRI CAs. Small (<10 nm diameter) Gd₂O₃ NPs have been the most intensively investigated, stabilized by coating with D-glucuronic acid [41], polyvinylpyrrolidone (PVP) [42], or polyethylene glycol (PEG) [43]. Paramagnetic NPs incorporating Gd³⁺ into the particle core or shell in a core/shell construct with large magnetization values have been proposed as T₁ MRI CAs. An example of this approach is NaYF₄:Yb, Er@NaGdF₄ core/shell NPs (20−40 nm diameter), with Gd³⁺ positioned only on the outer shell, where it can interact directly with water protons and promote T_1w image contrast [44].

Many small (<10 nm diameter) Mn-containing NPs, including MnO (Mn²⁺), Mn₂O₃ (Mn³⁺) and MnO₂ (Mn⁴⁺), have also been studied as T₁ MRI CAs [45,46,47,48]. Their r₁ values depend strongly on their geometry and morphology, as their interfaces with water critically influence their contrast effects. Some typical examples are zwitterionic dopamine sulfonate (ZDS)-coated ultrasmall MnO NPs (USMnO@ZDS) [49], PEG-AS1411 aptamer-coated MnO (AS1411-PEG-MnO) NPs [50] and ligand-free Mn₃O₄ NPs [51]. Non-oxide Mn²⁺-containing materials, such as other inorganic (i.e., KMnF₃, MnWO₄), Mn²⁺-chelate-based, hybrid organic/inorganic and Mn²⁺-based layered double hydroxide (Mn-LDH) NPs have also been exploited as T₁ MRI probes.

The use of high spin Fe³⁺ ions as T₁ MRI probes within nanostructures for in vitro studies or preclinical investigations has been recently reviewed [52], including examples such as amphiphilic polymer-based NPs like Fe³⁺-chelated poly(lactic-co-glycolic) acid (PLGA) NPs, NPs containing polyphenolic Fe³⁺-binding units such as Fe³⁺-loaded synthetic melanin nanoparticles (SMNPs) and bimetallic (Gd, Fe)-phenolate coordination polymer (CN) NPs. In several of these systems, although the Fe³⁺ ion is not hydrated (q = 0) and therefore the IS (q ≥ 1) contribution to relaxivity is not present, the reported r₁ values are higher than those expected for the OS (q = 0) contribution alone, which suggests the presence of a significant contribution from the water molecules of the second hydration sphere (SS).

The ideal T₂ MRI CAs should have high r₂ values and high (≥10) r₂/r₁ ratios to induce predominantly T₂ proton spin relaxation, leading to a decrease in signal intensity in T_2w or T_2w* MRI images [31]. The systems that fulfil these conditions are SPIONs and paramagnetic lanthanide (Ln³⁺)-based NPs (Ln = Tb, Dy and Ho), specially at high magnetic fields (B₀).

As discussed in Section 2, SPIONs are the T₂ CAs most commonly used in MRI, having a very high r₂ (usually 60–400 s⁻¹mM⁻¹). Examples are Resovist^® (a carboxydextran-coated SPION), octapod NPs with higher r₂ values than spherical NPs, highlighting the importance of size and morphology optimization to obtain NPs with high r₂ [53], ultrasmall SPIONs (USPIONs) with a scaffold of bovine serum albumin (BSA) [54], SPIONs conjugated with PEG of molecular weight (MW) in the 600–8000 amu range [55] and ferrimagnetic iron oxide nanocubes (FIONs) encapsulated in PEG phospholipids to become water-dispersible (WFIONs) [56]. The WFIONs have an extremely high r₂ value of 761 mM⁻¹.s⁻¹ at 3 T.

Paramagnetic NPs containing Tb³⁺, Dy³⁺ or Ho³⁺ have high r₂ values due to their high magnetic moments and very short (<1 ps) electronic relaxation times (Table 2) [24]. Therefore, they can be alternatives to SPIONs as efficient T₂ and T₂* CAs [13,57,58,59,60], in particular at high MRI fields (B₀ > 3 T). This is because r₂ generally increases with B₀, but the contribution from the Curie spin relaxation mechanism increases with B₀². The maximum r₂ values are obtained at slower inner sphere water exchange (τ_M = 0.1–10 μs) than for r₁ (τ_M = 1–100 ns) [61]. Also, in contrast to iron oxide particles, Ln³⁺-containing NPs show no saturation of the magnetization, even at magnetic field strengths as high as 30 T [62].

The highest payload of Ln³⁺ ions per particle at a particular site can be delivered by inorganic Ln³⁺-containing NPs, including Ln₂O₃, LnF₃, and NaLnF₄. NPs with a diameter of 50–100 nm contain approximately 106 Ln³⁺ ions each and their magnetic properties make them good candidates for T_2w and T_2w* MRI CAs [63]. Dy³⁺-based nanomaterials are the most studied, as they show the largest T₂ relaxation effects [23]. Moreover, the T₂ effects for Dy³⁺-based nanomaterials, such as β-NaDyF₄, are two orders of magnitude higher than for clinically approved iron oxide nanomaterials under the increasingly used very high field magnets (7 T and higher). This arises due to the Curie spin relaxation mechanism of the Dy³⁺-based systems [64]. However, coating of the particles is necessary to avoid leaching of free Ln³⁺.

3.2. NPs as T₁–T₂ DMCAs

While conventional MRI CAs respond only in a single imaging mode, either T₁ or T₂, NPs with multimodal capabilities provide complementary diagnostic information. However, when hybrid imaging systems, such as PET/CT or PET/MRI scanners are not available, a combination of two different imaging devices must be used separately, which is an inconvenient, time-consuming and expensive procedure [65,66,67,68,69]. The development of MRI T₁–T₂ dual-mode CAs in a single nanoplatform is an attractive solution to overcome ambiguities in conventional MRI diagnostics, especially when the biological targets are small, as well as the image matching difficulties caused by relocating the imaging object, and by the discrepancies resulting from different depth penetrations and spatial/time resolutions of multiple imaging strategies [69]. In fact, T₁–T₂ dual-modal MRI images can be easily acquired by changing the parameters of the pulse sequences in the operational mode of the same MRI scanner.

Dual-modal T₁–T₂ MRI CAs (DMCAs) should have both high r₁ and r₂ values, with r₂/r₁ ratios (~2–10) between those of ideal T₁ and T₂ MRI CAs. If Gd-based NPs are to be used as DMCAs, their r₂/r₁ ratios should be increased from their usual values of ~1, while the use of SPION-based DMCAs requires a decrease in their r₂/r₁ ratios from their usual very high values. Mn-based NPs are useful as DMCAs because of their suitable r₂/r₁ ratios.

The approaches proposed to design DMCAs comprise: (1) use of NPs with a single type of contrast material, Gd- or Mn-based NPs or superparamagnetic NPs with appropriate size and magnetization to provide T₂ and T₁ contrast; or (2) include both T₁ and T₂ contrast materials in the same NP. Both have advantages and disadvantages [31,69,70,71,72].

3.2.1. DMCAs Based on a Single Type of Contrast Material

• NPs based on a typical T₁ agent

This strategy is based on the clustering of a T₁ contrast material within non-magnetic porous matrices to obtain enhanced r₁ and specially r₂ values. The T₁ contrast material includes Gd-based complexes (e.g., Gd (DOTA)) or NPs (e.g., Gd₂O₃) [73,74,75,76], Mn-based NPs (e.g., MnO, MnO_x) or Mn²⁺ ions [77,78,79], mixed Gd/Dy NPs [80,81], and Fe³⁺ cations [82,83,84] entrapped in porous matrices, such as polymers (e.g., poly(amidoamine) (PAMAM) dendrimers, poly-(α,β)-DL-aspartic acid (PASA), mesoporous poly dopamine (MPDA)), porous silica, proteins (e.g., bovine serum albumin (BSA), hydrogels (e.g., hyaluronic acid (HA)/CH (chitosan)) or organic frameworks (e.g., NCP and CP₃ coordination polymers)) (

Table 3. Summary of basic properties of the examples of DMCA NPs based on a typical T₁ contrast agent discussed in this section. Some of the systems in this Table have been used as theranostic agents. * Not available.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	Therapeutic Modality	In Vitro/In Vivo Model for MRI	Ref.
Gd-DOTA@CH/HA hydrogel	72.3	177.5	2.5	1.5	260; -*	-	phantoms	[73]
Gd2O3@PASA	19.1	53.7	2.8	3.0	12.7; −28.0	-	in vivo mice	[74]
Gd2O3@G4.5 PAMAM-PEG	53.9	182.8	3.4	7.0	50.4; −0.2	-	in vitro RAW264.7 macrophages in vivo BALB/c female mice	[75]
Gd-DOTA-D-BSA	7.7	44.1	5.7	7.0	45; -*	-	ICR mice	[76]
MnO@DMMS	10.1	169.7	16.8	3.0	199; -		SD rats	[77]
PFOB@MnOx	0.41 → 6.13	5.39 → 43.75	15.0 → 8.1	*	170; -*	ferroptosis inhibition	tumor-bearing mice	[78]
R848@BNN@Mn-MPDA	10.2	129.3	12.7	7.0	201.4; -*	NO gas and immunotherapy	4T1 tumor-bearing mice	[79]
(Gd/Dy)2O3 @ D-glucuronic acid	*	*	6.6	*	1.0; -*	-	*	[80]
NaGdF4:Dy@DSPE-PEG2000	5.17	12.26	2.4	9.4	*	-	in vivo mice	[81]
Fe-PDA@CP3-DOX	7.52	45.92	6.1	1.5	144; -*	PTT/ChT	in vivo nude mice	[82]
Fe−NCP-BSA	5.3	10.9	2.1	7.0	97; −31.2	-	GL261 tumor-bearing mice	[83]
Fe-BPNS	0.84 → 6.19	20.5 → 154.8	24.4 → 25.0	3.0	208; −9.86	PTT/CDT	4T1 tumor-bearing mice	[84]

). This entrapment of metal chelates and NPs by serum proteins is likely to occur by opsonization [76]. Some DMCAs have also theranostic properties, by including drugs such as doxorubicin (DOX) in the matrices [78,79,83,84]. The ideal range of r₂/r₁ ratio can be approached by increasing r₂ of water protons through restricting water diffusion around the complex through clustering [73]. Another possibility is to decrease r₁ by restraining the metal ion coordination and chemical exchange of surrounding water molecules through geometrical confinement.

A few of the systems summarized in Table 3 are discussed here in more detail. Marasini et al. reported a Gd-based T₁–T₂ DMCA consisting of Gd₂O₃ NPs (2 nm diameter) coated with the hydrophilic polymer PASA using the one-pot polyol method. The NPs had r₁ = 19.1 mM⁻¹.s⁻¹ and r₂ = 53.7 mM⁻¹.s⁻¹ (r₂/r₁ = 2.8) at 3 T. After i.v. injection of Gd₂O₃@PASA NPs into the mice tail vain, T₁ and T₂ contrasts were observed in the T_1w and T_2w MRI images of the mouse livers at 3 T, respectively (Figure 1) [74].

A Mn-based T₁–T₂ DMCA was reported consisting of dual mesoporous silica spheres loaded with a MnO nanocluster (diameter < 2 nm) (Mn-DMSS) with a hydrodynamic diameter in the 100–200 nm range (Figure 2), with r₁ = 10.1 mM⁻¹.s⁻¹ and r₂ = 169.7 mM⁻¹.s⁻¹ (r₂/r₁ = 16.8) at 3 T. In vivo MRI experiments on Sprague Dawley (SD) rats at 3 T showed a 29% signal enhancement in the liver in T_1w images and a 28% signal decrease in T_2w images upon injection of Mn-DMSS NPs (Figure 2) [77].

A Gd/Dy-based novel PEGylated Dy-doped NaGdF₄ nanoprobe (NaGdF₄:Dy @DSPE-PEG₂₀₀₀) was developed as a T₁/T₂-weighted MRI/CT imaging nanoplatform, with r₁ = 5.17 mM⁻¹.s⁻¹ and r₂ = 10.64 mM⁻¹.s⁻¹ (r₂/r₁ = 2.4) at 9.4 T and strong X-ray attenuation properties (44.70 HU L g⁻¹) in vitro. T_1w/T_2w MRI/CT imaging in vivo of mice injected with the nanoprobe led to significant contrast enhancement of liver, spleen and kidneys at 24 h post injection. This multifunctional nanoprobe had low cytotoxicity and was completely excreted from the injected mice [81].

Finally, a Fe-based dual-mode T₁–T₂ MRI DMCA was reported consisting of a one-pot synthesized nanostructured coordination polymer containing Fe³⁺ (Fe-NCP) functionalized with BSA. These NPs, with a hydrodynamic diameter of 97 nm and a ζ potential of −31.2 mV, had high colloidal stability, low cytotoxicity and good relaxivities, r₁ = 5.3 mM⁻¹.s⁻¹ and r₂ = 10.9 mM⁻¹.s⁻¹ (r₂/r₁ = 2.1) at 7.0 T. Their T₁–T₂ MRI contrast was verified in in vitro phantoms, ex vivo C57BL/6J mice and in vivo GL261 glioblastoma tumor-bearing mice. The vivo MRI of Fe−NCPs showed high T₁ and T₂ contrast in the tumor in a very short period of time and were safe for the mouse. Their large long-term uptake in the spleen was related to their rapid clearance by the mononuclear phagocytic system (MPS) due to the NPs size being bigger than 40 nm [83].

Table 3 shows that not all the systems described reach the low r₂/r₁ values recommended for efficient DMCAs, illustrated by the second and fourth examples discussed above [77,83].

2.

NPs based on a typical T₂ agent: USPIONs and FeO_x

This strategy is based on using superparamagnetic NPs with appropriate size and magnetization to provide T₂ and T₁ contrast simultaneously. Although SPIONs with a core diameter less than 10 nm can produce positive contrast in T_1w images at low concentrations, their high T₂ effects (high r₂/r₁) resulting from their high magnetic moment limit their application as T₁–T₂ DMCAs. However, the magnetic moment of Fe₃O₄ NPs is strongly dependent on their size and decreases rapidly as their size decreases due to the reduction in the volume magnetic anisotropy and spin disorder (canting) at their surface, which suppresses the T₂ effect and therefore maximizes the T₁ contrast effect, controlling their relaxivities [11]. Therefore, the appropriate ultra-small-sized Fe₃O₄ NPs (USPIONs) are potential candidates for T₁–T₂ DMCAs.

Some examples of such systems are summarized in Table 4, where the USPION core was coated with hydrophilic polymers, e.g., poly(methacrylic acid) (PMAA)-polytrimethylene terephthalate (PMAA-PTTM) [85], poly(acrylic acid) (PAA) [86,87] and PEG [88], or silica [89], to prevent the aggregation of the NPs and to ensure a small particle size. The core was usually spherical, but in some cases nanoplates [89] or nanocubes [90,91] were used. Some of the systems summarized in Table 4 are now discussed in more detail.

Li et al. reported monodispersed water-soluble and biocompatible USPIONs (3.3 nm average diameter) grafted with thiol functionalized (PMAA)-(PTTM) using a high-temperature co precipitation method. These NPs, with r₁ = 8.3 and r₂ = 35.1 s⁻¹.mM⁻¹ (r₂/r₁ = 4.2) at 4.7 T, showed in vitro and in vivo potential as T₁–T₂ DMCAs upon i.v. injection in mice, as positive and negative contrasts were observed in the T_1w and T_2w MRI images of liver and kidneys, respectively [85]. Miao et al. developed PAA-coated USPIONs (5.1 nm core diameter and 41.35 nm average hydrodynamic size) with r₁ = 10.52 and r₂ = 38.97 s⁻¹.mM⁻¹ (r₂/r₁ = 3.70) at 1.41 T. The NPs had in vitro and in vivo potential as T₁–T₂ DMCAs, as illustrated by T_1w and T_2w MRI images at 3 T, with positive contrast observed in the rabbit vasculature, while the rabbit popliteal lymph node exhibited negative contrast [87]. Wang et al. reported USPION-PEG (P-UDIOC) NPs with extremely small core size (2.3 nm) and a compact hydrophilic PEG surface and with r₁ = 1.37 and r₂ = 7.53 s⁻¹.mM⁻¹ (r₂/r₁ = 5.5) at the ultrahigh field (UHF) of 7.0 T. These UHF-tailored T₁–T₂ DMCAs showed dual enhanced T₁–T₂ contrast at 7 T and enabled a clear visualization of microvasculature as small as ≈140 μm in diameter under UHF MRI (Figure 3), extending the detection limit of 7 T MRI angiography (MRA) [88].

An example of a system with a non-spherical core consisting of Fe₃O₄ crystal nanoplates was reported by Zhou et. al. [89] These superparamagnetic magnetite nanoplates had (111) exposed facets which much contributed to their T₁ and T₂ contrast effects. The main contribution to r₁ of the magnetic nanoplates is the chemical exchange with bulk water (given by the exchange lifetime, τ_M) of the water molecules bound at the inner-sphere of the Fe³⁺ ions on the highly exposed particle surface iron-rich Fe₃O₄ (111) surfaces, according to the dominant IS model [11]. The r₂ values are dominated by the OS mechanism, which accounts for the effect of fluctuations of the local magnetic field inhomogeneities induced by the tumbling NPs on the protons of outer-sphere water molecules diffusing nearby. According to Freed’s theory, the r₂ value is proportional to the square of the magnetization (M_s²), which relates to the intrinsic superparamagnetism of the nanoplates and determines the strength of the local magnetic field inhomogeneities, as well as to the square of the effective radius of the magnetic core (R²), which determines the field perturbation areas for the outer-sphere protons. The rapid random flipping of the anisotropic nanoplates, when represented by an equivalent simulated sphere, generate a larger area of local field inhomogeneity compared with nanospheres under the same applied magnetic field. The balance of T₁ and T₂ contrasts was attained by controlling the structure and surface features of the nanoplates, including morphology (nanoplates vs. spheres in IOP-4.8@stPE) and surface coating (e.g., IOP-4.8 vs. IOP-4.8@SiO₂ and IOP-4.8@stPE), with a large decrease in r₁ due to blocking of the exposure of the facets causing the disappearance of the IS contribution and corresponding decrease in r₂/r₁, leading to change from DMCAs to T₂ agents. The decrease in the nanoplate thickness decreased the r₂/r₁ value, as IOP-8.8 (r₂/r₁ ~ 8.18) is T₂-dominated, while the IOP-4.8 (r₂/r₁ ~ 4.22) is a T₁–T₂ DMCA [89].

Table 4 shows that not all the systems described reach the low r₂/r₁ values recommended for efficient DMCAs [90,91].

3.

Triggered aggregation change in ESIONs

The design of T₁–T₂ DMCAs can also be based on a different mechanism, based on the change in the aggregation state of extremely small-sized iron oxide nanoparticles (ESIONs) in the size range of 1.5–4 nm with relatively high r₁ values, triggered by an external signal, such as pH change, a redox reaction or laser light. The aggregation of ESIONPs can generate T₂ contrast effects due to the enhancement of magnetic field inhomogeneity and magnetic coupling between Fe centers. Some examples of such systems are summarized in Table 4. An example is based on pH-sensitive hydrazine functionalized ESIONPs forming assemblies (IONAs) cross-linked by small-molecular aldehyde derivative ligands. The dynamic formation and cleavage of hydrazone (–C=N–N–) (f) linkages in neutral and acidic environments, respectively, is the base for the reversible response of the nanoassemblies to pH variations. At pH 7.5, IONAs are stable and with high r₂/r₁ = 34.2, while at acidic pH 5.5, such as in the acidic tumor microenvironment (TME), the hydrazone bonds are cleaved and the IONAs are disassembled into hydrophilic dispersed ESIONs, with a r₂/r₁ = 4.1. The change in this relaxivity ration results from an increase in the number of second sphere water molecules (q_SS) and a large decrease in the magnetization (M_s) of the dispersed ESIONs relative to the IONAs, which affect their r₁ and r₂ values, respectively [92]. Another example is based on ESIONs linked with the targeting ligand folic acid (FA) binding arthritis-associated macrophage cells, and the light responsive diazirine (DA) through a PEG spacer (Fe₃O₄-PEG-(DA)-FA). These nanoparticles can form nanocomposites (NCs) upon laser irradiation to have tunable r₁ and r₂ values upon variation in the laser irradiation time. The change in the r₂/r₁ value from 2.36 in the NPs to 18.8 in the NCs led to the use of the designed Fe₃O₄-PEG-(DA)-FA NPs as T₁ CAs in in vivo MRI of an arthritis mouse model without lasers and enhanced T₁–T₂ DMCAs in the arthritis inflamed region under laser irradiation due to the formation of NCs that accumulated within the arthritis region and their limited intravasation back to the blood circulation [93].

Again, Table 4 shows that not all the systems described reach the low r₂/r₁ values recommended for efficient DMCAs: while the first one does [92], the second one reaches quite high r₂/r₁ values [93]. Besides this, in some cases, T₁/T₂ triggering effects, such as pH decrease or reduction by GSH in the TME, led to final r₂/r₁ values below or above the range where T₁–T₂ DMCAs operate, forming instead T₁/T₂ switching MRI CAs [94,95,96,97,98,99,100,101]. This illustrates how the control of the switching mechanism at the molecular level is difficult.

Table 4. Summary of basic properties of the examples of NPs based on a typical T₂ ((U)SPION)) contrast agent discussed in this section.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	In Vitro/In Vivo Model for MRI	Ref.
USPION-PMAA-PTTM	8.3	35.1	4.2	4.7	3.3 (core); -*	In vitro phantoms /In vivo mice	[85]
USPION-PAA	8.20	16.67	2.03	7.0	1.7 (core): -*	In vivo Kunming mice	[86]
	6.15	28.62	4.65		4.6 (core); -*
USPION-PAA	10.52	38.97	3.7	3.0	41.3; -14.7	In vivo rabbit	[87]
USPION-PEG	1.37	7.53	5.5	7.0	2.3 (core); dH ≈12; -*	In vivo rat brain angiography (MRA)	[88]
Fe3O4 nanoplates:				0.5		In vitro phantoms	[89]
IOP-4.8	43.18	182.2	4.22		4.8 thickness
IOP-4.8@SiO2	2.0	118.73	59.3		5.6 thickness
IOP-4.8@stPE spheres	3.59	338.9	94.4		90; -*
Fe3O4 nanocubes	5.23	89.68	17.1	3.0	27.8; -*	In vivo SD rats	[90]
ND-PEG-tNCIO	31.8	790.6	24.7		50; −61	In vitro phantoms	[91]
IONA → ESIONP dispersed pH 7.4 → 5.5.	3.2 → 108.0	5.1 → 22.3	34.2 → 4.1	3.0	80	In vitro A549 cells In vivo tumor mice	[92]
Fe3O4-PEG-(DA)-FA → NCs (Laser)	3.83 → 1.60	9.04 → 31.6	2.36 → 18.8	*	*	In vivo arthritis mouse model	[93]

* Not available.

Table 4. Summary of basic properties of the examples of NPs based on a typical T₂ ((U)SPION)) contrast agent discussed in this section.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	In Vitro/In Vivo Model for MRI	Ref.
USPION-PMAA-PTTM	8.3	35.1	4.2	4.7	3.3 (core); -*	In vitro phantoms /In vivo mice	[85]
USPION-PAA	8.20	16.67	2.03	7.0	1.7 (core): -*	In vivo Kunming mice	[86]
	6.15	28.62	4.65		4.6 (core); -*
USPION-PAA	10.52	38.97	3.7	3.0	41.3; -14.7	In vivo rabbit	[87]
USPION-PEG	1.37	7.53	5.5	7.0	2.3 (core); dH ≈12; -*	In vivo rat brain angiography (MRA)	[88]
Fe3O4 nanoplates:				0.5		In vitro phantoms	[89]
IOP-4.8	43.18	182.2	4.22		4.8 thickness
IOP-4.8@SiO2	2.0	118.73	59.3		5.6 thickness
IOP-4.8@stPE spheres	3.59	338.9	94.4		90; -*
Fe3O4 nanocubes	5.23	89.68	17.1	3.0	27.8; -*	In vivo SD rats	[90]
ND-PEG-tNCIO	31.8	790.6	24.7		50; −61	In vitro phantoms	[91]
IONA → ESIONP dispersed pH 7.4 → 5.5.	3.2 → 108.0	5.1 → 22.3	34.2 → 4.1	3.0	80	In vitro A549 cells In vivo tumor mice	[92]
Fe3O4-PEG-(DA)-FA → NCs (Laser)	3.83 → 1.60	9.04 → 31.6	2.36 → 18.8	*	*	In vivo arthritis mouse model	[93]

* Not available.

3.2.2. DMCAs including Both T₁ and T₂ Contrast Materials in the Same Nanoparticle

The strategy of including both T₁ and T₂ contrast materials in the same nanoplatform has used several different designs to obtain hybrid nanostructures: (a) NPs integrating a superparamagnetic T₂-contrasting material inside a paramagnetic T₁ material (e.g., a Gd³⁺ complex); (b) doping a superparamagnetic NP with paramagnetic T₁ contrast materials inside them; (c) T₁ and T₂ contrast materials connected side-by-side to form hybrid nano-oligomers (DB-HNT). In all these designs, with both contrast materials in the same nanoplatform, the magnetic fields generated by each CA disturb the relaxation process of the other. However, as typical paramagnetic T₁ CAs have low r₁ values compared to the r₂ values of superparamagnetic T₂ agents due to their relative magnetic moments, their effect on the T₁ relaxation of the T₁ contrast material is much larger than the opposite. The T₂ agent generates a strong magnetic field induced by the external B_o, which is dependent on 1/r³ (r = distance from the T₂ agent), which affects the electronic spin relaxation time (T_1e) of the paramagnetic T₁ agent depending on their relative locations. The resulting effects of this magnetic coupling depends on the separation distance (d) between T₁ and T₂ agents [102] and also on their relative positions within the NP (Figure 4) [103].

• NPs with a T₂ material inside a paramagnetic T₁ material

For systems consisting of a superparamagnetic core within a paramagnetic shell, the strong magnetic field from the core opposes the magnetic field created by the shell and reduces it (Figure 4, left), strongly quenching its r₁ value. Their magnetic interaction is proportional to the inverse sixth power of their separation distance (d⁻⁶), as shown by a study where the distance-dependent magnetic resonance tuning (MRET) strategy for tuning the r₁ value of a T₁ agents was introduced [102]. A series of spherical NPs was designed, consisting of three components, where the paramagnetic enhancer (Gd-DOTA) was separated from the superparamagnetic quencher (a 12 nm Zn_0.4Fe₂.₆O₄ NP) by controlling the thickness of a SiO₂ separating layer decreasing from 18 to 2 nm (Zn_0.4Fe₂.₆O₄@SiO₂@ Gd-DOTA). A decrease in r₁ was consistently observed as the separation distance between the T₁ and T₂ agents decreased. The T₁ MRI signal was quenched when the d value between the enhancer and the quencher decreased and r₁ decreased from 1.58 to 0.13 mM⁻¹.s⁻¹ (3 T), which was a consequence of the increase in the T_1e value of the T₁ agent [102].

The strategy of including a layer of increasing thickness to increase the T₂ core-T₁ shell distance, and thus modulating their magnetic coupling, was pursued using not only inorganic porous materials like SiO₂, but also micellar structures incorporating organic block copolymers and inorganic porous materials as possible frameworks [104]. The basic properties of several examples from the literature are summarized in Table 5. These involve the formation of spherical or cubic core–shell NPs integrating a superparamagnetic T₂-contrasting core (e.g., Fe₃O₄, MnFe₂O₄) inside a paramagnetic T₁ shell (e.g., Gd₂O₃, Gd₂O(CO₃)₂, MnO) [105,106,107,108,109,110,111,112], or conjugating the superparamagnetic core with a paramagnetic Gd³⁺ or Mn²⁺ complex at its surface [113,114,115,116,117]. Generally, a sharp decrease in the magnetic coupling effect upon increasing the core–shell separation was verified experimentally by including a silica shell of increasing thickness between the two materials [105,106,107], or by increasing the distance between the core and the layer of pendant paramagnetic complexes through longer spacer groups [102,113]. Here, we will discuss only a few of the examples in more detail.

For instance, a core–shell-type T₁–T₂ DMCA agent has been described, where the T₁ contrast material, a Gd₂O(CO₃)₂ layer (1.5 nm thickness) was located on the shell in order to be in direct contact with water molecules, to obtain for high T₁ contrast; the superparamagnetic T₂ contrast material, MnFe₂O₄ (15 nm size) was located at the core, from where it could induce a long-range magnetic field for the relaxation of water molecules. The two materials were separated by a SiO₂ layer of increasing (4, 8, 12, 16 and 20 nm) thickness (MnFe₂O₄@SiO₂@Gd₂O(CO₃)₂ NPs). As the SiO₂ layer became thicker, the magnetic coupling decreased, the T₁ quenching was reduced and r₁ increased (from 2.0 to 32.5 mM⁻¹.s⁻¹) while the r₂ decrease was weaker (332 to 213 mM⁻¹.s⁻¹). When the thickness of the SiO₂ layer was equal of larger than 16 nm, the r₂/r₁ values decreased from 160 to 6.5, and the NPs became T₁–T₂ DMCAs, as both T₁ and T₂ effects became larger than the effects of the individual single-mode contrasts (Figure 5) [105].

A smart nanotheranostic system for early diagnosis and therapy of cancer was developed, consisting of camptothecin (CPT)-loaded mesoporous silica nanoparticles (MSN) capped with manganese oxide (MnO_x)-coated SPIONs (MnO_x-SPION@MSN@CPT NPs). The acid, oxidative stress and redox (GSH) response of MnO_x regulated the CPT drug release from the MSN channels, while the high magnetization of the surface SPIONs achieved high r₂ values (102.2 mM⁻¹.s⁻¹). At the same time, degradation of the MnO_x shell caused release of Mn²⁺ in the TME, enhancing r₁ (2 → 13.6 mM⁻¹.s⁻¹). The efficacy of this MRI responsive theranostic T₁–T₂ DMCA was confirmed in vitro on pancreatic cancer cells and in vivo on tumor-bearing mice (Figure 6) [111].

Some examples of systems conjugating the coating of SPIONs with a paramagnetic Gd³⁺ or Mn²⁺ complex will now be discussed briefly (Table 5) [113,114,115,116,117]. Up to now, the highest relaxivities for T₁–T₂ DMCAs are r₁ = 31.6 mM⁻¹.s⁻¹and r₂ = 836.7 mM⁻¹.s⁻¹ (at 1.47 T) which was obtained by Long et al. for Fe₃O₄@ALA-GdDOTA [113]. Superparamagnetic silica-coated iron oxide core–shell nanoparticles (Fe₃O₄@SiO₂) conjugated at the surface to Gd-DTPA and the arginine-glycine-aspartic acid (RGD) peptide as a targeting ligand (Fe₃O₄@SiO₂(Gd-DTPA)-RGD) NPs were synthetized. The NPs with a 21 nm diameter were water-dispersible, stable, and biocompatible, and with r₁ = 4.2 mM⁻¹.s⁻¹ and r₂ = 17.4 mM⁻¹.s⁻¹ (r₂/r₁ = 4.1) at the Gd/Fe molar ratio of 0.3:1. In vitro MRI experiments with U87MG and MCF- tumor cells over-expressing the high-affinity α_vβ₃ integrin showed targeted T_1w positive and T_2w negative contrast when loaded with the NPs (Figure 7) as well as in in vivo U87MG tumor mice injected with the NPs, proving their potential as targeted T₁–T₂ DMCAs [117].

In other studies, core–shell NPs with different metal ions in the shell and/or in the core were studied (Table 5). Superparamagnetic SPION core−porous silica shell nanoparticles, containing the paramagnetic complexes ([Ln(btfa)₃(H₂O)₂]) (btfa = 4,4,4-trifluoro-L-phenyl-1,3-butanedione, Ln = Gd/Eu) (γ-Fe₂O₃@SiO₂/[Gd/Eu(btfa)₃(H₂O)₂]) imbedded in the shell, with a 50 nm diameter and a 10 nm core, performed as a promising trimodal T₁−T₂ MRI and optical imaging contrast agent in Hela cells suspensions in vitro [118]. Also, colloidal suspensions of Fe/Fe₂O₃ NPs capable of providing both T_1w and T_2w MR images were synthesized [119]. Fe@Fe₃O₄-Gd(DOTA) NPs are strongly paramagnetic at room temperature with a M_s = 55 emu.g⁻¹, and r₂ and r₁ values higher than those of Fe@Fe₃O₄ and Gd(DOTA) alone. These increased relaxivities suggest that the NPs are potential T₁–T₂ DMCAs for MRI [120]. Avocado-like Fe³⁺/Fe₂O₃ NPs were developed for working as T₁–T₂ DMCAs, based on a PDA-Fe³⁺-TA (tannic acid) coordination network (CNMN) embedded with DOX (Fe₂O₃@PDA-Fe³⁺-TA-CNMN/DOX). These NPs had suitable r₁ and r₂ values and the strong heat generated by Fe³⁺-TA allowed the CNMN to act as a photothermal agent and to effectively deliver the chemotherapeutic drug DOX to achieve chemo-photothermal combination therapy [121]. Finally, an iron core (with its subsequent oxidation giving a ferrite shell) with added Ni²⁺ ions to form a superparamagnetic nickel ferrite shell NP has been studied; with its surface treated with dopamine-PEG to make it dispersible, and it acts as a T₁–T₂ DMCA [122].

Table 5 shows that many of the systems described using this strategy have r₂/r₁ values above the values recommended for efficient DMCAs [107,109,110,113,115,118,120,121]. Besides this, the systems based on the release of toxic free Mn²⁺ in the TME, although interesting for animal studies, are not suited for clinical applications [110,111].

Table 5. Summary of basic properties of the examples of NPs with a T₂ material inside a paramagnetic T₁ material.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	Therapeutic Modality	In Vitro/In Vivo Model for MRI	Ref.
MnFe2O4@SiO2@ Gd2O(CO3)2	2.0 → 32.5	332 → 213	166 → 6.5	4.7	26 → 58	-	In vitro phantoms	[105]
MnFe2O4@SiO2@ Gd2O(CO3)2	3.7 → 32.3	312 → 208	84.3 → 6.4	3.0	31 → 55	-	In vitro phantoms	[106]
Fe3O4@mSiO2/PDDA/BSA-Gd2O3	11.47	195.1	17.0	3.0	345.6; +26.9	-	In vitro 786-0 cells In vivo BALB/c mice	[107]
Fe3O4@Gd2O3 nanocubes	45.24	186.51	4.1	1.5	9.2; -*	-	In vitro phantoms In vivo SD rats (3 T)	[108]
Fe3O4@MnO-PEG	1.3	35.8	28	3.0	5.0; -*	-	In vivo BALB/c mice (7.0 T)	[109]
Fe3O4@Mn3O4 → Fe3O4 +Mn2+ (GSH)	2.4 → 16.1	92 → 258	38.4 → 16.1	1.5	22; -*	-	In vivo MKN-45 tumor-bearing mice	[110]
MnOx-SPION @MSN@CPT → SPION +Mn2+ (GSH, pH)	2 → 13.6	102.2 → -*	*	0.5	120; -*	ChT (CPT)	In vivo pancreatic-tumor- bearing mice	[111]
Fe3O4@ALA-GdDOTA (NP5) Fe3O4@ALA-Mn-DOTA (NP6)	31.6	836.7	26.4	1.47	6; *	-	In vitro phantoms	[113]
	14.2	324.5	22.8
Fe3O4@DOPA-GdDTPA-PEG	11.17	30.32	2.7	3.0	73.8; −5.5	-	In vivo BALB/c nude mice.	[114]
Fe3O4/CuInS2@SiO2- (GdDTPA)-RGD	1.56	23.22	14.9	3.0	45; +8.16	-	In vivo BSPC-3 pancreatic tumor mice	[115]
MnFe2O4@Gd: FA-DTPA-PEG-DIB-	20.59	68.48	3.32	0.55	18; *	-	In vitro Hela and 3T3 cells	[116]
Fe3O4@SiO2-GdDTPA-RGD	4.2	17.4	4.1	3.0	27; +7.25	-	In vitro U87MG cells In vivo U87MG tumor mice	[117]
γ-Fe2O3@SiO2/[Gd/Eu (btfa)3(H2O)2]	1.0	75.9	78	9.4	50; −40	-	In vitro Hela cells	[118]
Fe@Fe3O4-GdDOTA	7.2	109.4	15.2	0.5	358; +24.6	-	In vitro 4T1 cells In vivo 4T1 tumor mice	[120]
Fe2O3@PDA-Fe3+-TA-CNMN/DOX	5.01	125.45	25.0	7.0	95.6; −30.3	ChT(DOX)/ PTT	In vivo tumor mice	[121]
Fe@NiFe2O4-PEG/dopamine	7.19	9.96	1.4	2.4	10–15; *	-	In vitro phantoms	[122]

* Not available.

Table 5. Summary of basic properties of the examples of NPs with a T₂ material inside a paramagnetic T₁ material.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	Therapeutic Modality	In Vitro/In Vivo Model for MRI	Ref.
MnFe2O4@SiO2@ Gd2O(CO3)2	2.0 → 32.5	332 → 213	166 → 6.5	4.7	26 → 58	-	In vitro phantoms	[105]
MnFe2O4@SiO2@ Gd2O(CO3)2	3.7 → 32.3	312 → 208	84.3 → 6.4	3.0	31 → 55	-	In vitro phantoms	[106]
Fe3O4@mSiO2/PDDA/BSA-Gd2O3	11.47	195.1	17.0	3.0	345.6; +26.9	-	In vitro 786-0 cells In vivo BALB/c mice	[107]
Fe3O4@Gd2O3 nanocubes	45.24	186.51	4.1	1.5	9.2; -*	-	In vitro phantoms In vivo SD rats (3 T)	[108]
Fe3O4@MnO-PEG	1.3	35.8	28	3.0	5.0; -*	-	In vivo BALB/c mice (7.0 T)	[109]
Fe3O4@Mn3O4 → Fe3O4 +Mn2+ (GSH)	2.4 → 16.1	92 → 258	38.4 → 16.1	1.5	22; -*	-	In vivo MKN-45 tumor-bearing mice	[110]
MnOx-SPION @MSN@CPT → SPION +Mn2+ (GSH, pH)	2 → 13.6	102.2 → -*	*	0.5	120; -*	ChT (CPT)	In vivo pancreatic-tumor- bearing mice	[111]
Fe3O4@ALA-GdDOTA (NP5) Fe3O4@ALA-Mn-DOTA (NP6)	31.6	836.7	26.4	1.47	6; *	-	In vitro phantoms	[113]
	14.2	324.5	22.8
Fe3O4@DOPA-GdDTPA-PEG	11.17	30.32	2.7	3.0	73.8; −5.5	-	In vivo BALB/c nude mice.	[114]
Fe3O4/CuInS2@SiO2- (GdDTPA)-RGD	1.56	23.22	14.9	3.0	45; +8.16	-	In vivo BSPC-3 pancreatic tumor mice	[115]
MnFe2O4@Gd: FA-DTPA-PEG-DIB-	20.59	68.48	3.32	0.55	18; *	-	In vitro Hela and 3T3 cells	[116]
Fe3O4@SiO2-GdDTPA-RGD	4.2	17.4	4.1	3.0	27; +7.25	-	In vitro U87MG cells In vivo U87MG tumor mice	[117]
γ-Fe2O3@SiO2/[Gd/Eu (btfa)3(H2O)2]	1.0	75.9	78	9.4	50; −40	-	In vitro Hela cells	[118]
Fe@Fe3O4-GdDOTA	7.2	109.4	15.2	0.5	358; +24.6	-	In vitro 4T1 cells In vivo 4T1 tumor mice	[120]
Fe2O3@PDA-Fe3+-TA-CNMN/DOX	5.01	125.45	25.0	7.0	95.6; −30.3	ChT(DOX)/ PTT	In vivo tumor mice	[121]
Fe@NiFe2O4-PEG/dopamine	7.19	9.96	1.4	2.4	10–15; *	-	In vitro phantoms	[122]

* Not available.

2.

Doping superparamagnetic T₂ contrast NPs with paramagnetic T₁ contrast materials inside

For systems where the paramagnetic material resides inside the superparamagnetic iron oxide, the magnetic fields of both materials reinforce each other simultaneously, strongly enhancing the r₁ value and causing a synergistic T₁–T₂ enhancement effect (Figure 4, right) [103]. The basic properties of systems using this strategy are shown in

Table 6. Summary of basic properties of NPs made of a paramagnetic T₁ material inside a superparamagnetic T₂ material or forming hybrid oligomers of different shapes (in bold). Their properties are compared with those of the NPs made of their components.

NPs Components	r1 (mM−1.s−1)	r2 (mM−1.s−1)	r2/r1	Bo (T)	dH (nm); ζ (mV)	In Vitro/In Vivo Model for MRI	Ref.
Gd2O3@ Fe3O4-HDA-G2 (GdIO)	69.5	146.5	2.1	0.5	14; -*	In vitro HeLa, HepG2 cells In vivo Balb/c/HepG2 tumor mice (7 T)	[103]
Fe3O4	*	125.4	*		14; -*
Gd2O3	12.1	*	*		2; -*
GdIO–stPEI	61.67	181.49	2.94	0.5	153.2; -15	In vitro HCT-116 cells	[123]
GdIO–ZDS	7.85	41.14	5.24	7.0	4.8 (core) 6.50; -*	In vivo SKOV3 tumor mice	[124]
GdIO–ZDS	4.63	34.38	7.43		3.5 (core) 5.61; -*
GdIO–ZDS	3.05	26.45	8.56		2.8 (core) 4.18; -*
IO@ZDS	6.14	58.94	9.59		4.9 (core) 7.13; -*
MnIO-tartarate	18.0	45.9	2.6	0.5	5 (core) 10.49; -*	In vivo BALB/c mice (7 T)	[125]
	27.2	146.5	5.4		7 (core) 13.21; -*
	32.1	205.5	6.4		9 (core) 16.79; -*
	38.2	280.8	7.4		12 (core) 22.28; -*
EuIO-citrate	36.79	97.52	2.65	0.5	14.0; -*	In vivo SD rats (3.0 T)	[126]
Fe3O4	12.47	116.78	9.36		14.0; -*
Eu3O4	0.03	5.44	181		14.0; -*
GdDOTA-Au-IONPs	1.65	123	74.5	7.0	19.5; −13.4	In vitro phantoms In vivo HT-29 tumor mice	[127]
GdDOTA-HNTs	1.74	131	75.2		21.2; −15.1
GdDOTA-DB-HNTs	3.88	128	33.0		24.6; −16.0
GdDOTA-XDB-HNTs	4.12	125	30.3		26.8; −16.4
Fe3O4@MnO-PEG sphere	1.3	35.8	28	3.0	5; -	In vitro phantoms In vivo nude mice (brain) In vivo HCC nude mouse (liver) (7 T)	[109]
Fe3O4/MnO-PEG dumbbell	1.4	78.9	56		11; -*
Fe3O4/MnO-PEG flower	0.6	141	235		21; -*
Fe3O4-PEG sphere	0.8	152	190		5; -*
Fe3O4-PEG sphere	1.1	162	147		11; -*
Fe3O4-PEG sphere	1.1	252	229		21; -*

* Not available.

, some of which are discussed in some detail. The theory of synergistic T₁–T₂ enhancement effect discussed above was confirmed by Gao et al. using Gd₂O₃-embedded Fe₃O₄-HDA-G₂ NPs (GdIO-HDA-G₂), which showed a synergistic enhancement of r₁ and r₂. The GdIO had higher r₂ (146.5 mM⁻¹.s⁻¹) than Fe₃O₄ (125.4 mM⁻¹.s⁻¹) of similar size and also higher r₁ (69.5 mM⁻¹.s⁻¹) than Gd₂O₃ (12.1 mM⁻¹.s⁻¹) of similar size. Furthermore, the Gd₂O₃ NPs showed no enhanced T₂ contrast, while Fe₃O₄ nanoparticles showed limited enhanced T₁ contrast. Simultaneous in vivo T_1w and T_2w MRI of BALB/c mice upon i.v. injection of GdIO showed simultaneous strong MRI contrast enhancement of liver in both types of images due to the high accumulation of NPs in the hepatic Kupffer cells of the liver mononuclear phagocyte system (MPS). The same MRI experiment in HepG2 tumor mice detected the liver lesions through pseudo-negative and pseudo-positive contrast effects because the contrast between lesions and surrounding normal liver tissue increased due to the very low uptake by hepatic tumors, which contain few active Kupffer cells and macrophages (Figure 8). This work validated the new strategy for the design of new T₁–T₂ DMCAs [103].

Another example of the use of the same design strategy consisted of water-dispersible GdIO NPs stearic acid modified low molecular weight polyethyleneimine (stPEI) (GdIO–stPEI). This nanoplatform was capable of binding and delivering siRNA for gene knockdown and work as T₁–T₂ DMCA with HCT-116 cells in vitro [123]. Zwitterion dopamine sulfonate-coated superparamagnetic GdIO NPs (GdIO-ZDS) with small core size (2.8–4.8 nm) showed partial paramagnetism at room temperature, with a decreased M_S value relative to IOs of the same core size. This resulted from the combination of surface canting with the effect of the embedded Gd₂O₃ nanoclusters which disturbs the long-range order of magnetic spins in the small GdIO NPs. It also led to a significantly increased r₁ value and decreased r₂ value relative to IO-ZDS of the same core size. For example, GdIO NPs with a 4.8 nm diameter, had a high r₁ = 7.85 mM⁻¹.s⁻¹ and a low r₂/r₁ = 5.24 relative to IO (r₂/r₁ = 9.59). These NPs caused a strong positive tumor contrast effect in T_1w MRI images of SKOV3 human ovarian tumor mice through an enhanced permeation and retention (EPR) effect [124].

Superparamagnetic Fe₃O₄ NPs have also been doped with other paramagnetic ions, such as Mn²⁺ (MnIO) and Eu³⁺ (EuIO) [125,126]. The MRI contrast abilities of uniform Mn²⁺-doped iron oxide (MnIO) NPs nanoparticles with 5, 7, 9 and 12 nm size were studied. The NPs were superparamagnetic at 300 K, with M_s values which decreased with the decrease in the MnIO NP sizes, from 71.0 emu g⁻¹ for the 12 nm NPs to 39.7 emu g⁻¹ for the 5 nm NPs due to the spin canting effect at their surface. Their r₁ and r₂ values were highly size-dependent, with r₂/r₁ values decreasing from 7.4 for the 12 nm NPs to 2.6 for the 5 nm NPs. Thus, by controlling the size of the MnIO NPs, T₁-dominated, T₂-dominated, and T₁–T₂ DMCAs could be obtained with much higher contrast enhancement than the corresponding conventional iron oxide nanoparticles, as verified by in vivo MRI of BALB/c mice [125]. Finally, Eu³⁺-dopped iron oxide nanocubes (EuIO) were developed as T₁–T₂ DMCAs for in vivo MRI. The EuIO nanocubes were composed of mixed Fe₃O₄ (magnetite) and Eu₂O₃ nanoclusters of 10.0, 14.0 and 20.1 nm size, due to the large ionic radius of Eu³⁺ ions (94.7 pm) which prevented them from occupying either the tetrahedral or the octahedral interstitial sites in the spinel structure. The EuIO nanocubes are partially paramagnetic at 300 K, which is different from the superparamagnetism of magnetite NPs, due to increased spin canting on their surface layer after Eu₂O₃ embedding. The Eu₂O₃ clusters located inside the iron oxide NPs nanoparticles disturbed the local magnetic field intensity of the whole NPS and reduced their M_s values (~39.6 emu g⁻¹) relative to magnetite NPs with a similar size (~53.4 emu g⁻¹) at 300 K. The larger EuIO NPs had higher M_s values due to the loss of the spin canting effect on the particle surface. As a result, both r₁ and r₂ values of EuIO nanocubes could be tuned by varying their sizes and Eu doping ratios. Larger EuIO nanocubes had higher r₁ and r₂ values. The Eu/Fe molar ratio also had an important role in the r₁ and r₂ values of EuIO nanocubes: raising the Eu molar ratio increased r₁ due to the spin order of Eu³⁺ ions which had the same orientation as the local magnetic field, while it decreased r₂ values due to the reduction in the M_s values after Eu embedding. For instance, EuIO nanocubes of 14 nm diameter showed a high r₁ = 36.8 mM⁻¹.s⁻¹, which is approximately 3 times higher than that of Fe₃O₄ NPs (12.47 mM⁻¹.s⁻¹) of similar size. After citrate coating, EuIO nanocubes produced enhanced T₁ and T₂ MRI contrast effects in Sprague Dawley rats as models for in vivo MRI studies, in particular in the cardiac and liver regions [126].

In summary, Table 6 shows that the r₂/r₁ values of the composite NPs (in bold) are more suitable as DMCAs when compared with those of the NPs made of their components.

3.

T₁ and T₂ contrast materials connected side-by-side forming hybrid oligomers of different shapes

A third strategy to control the interference by magnetic coupling between T₁ and T₂ materials present in a single-composite nanostructure is to engineer the architecture of heterogeneous NPs forming hybrid trimers and oligomers of different geometries, like dumbbell-shaped NPs [109,127] or nanoflowers [109] (Table 6). Dumbbell-shaped NPs, or so-called ‘Janus’ NPs, were synthesized, with two different components within one single structure, as solid-state analogues of bifunctional organic molecules to construct hybrid nanotrimers (HNTs) (Figure 9a, right panel), in which iron oxide and Au nanocrystals were connected by a platinum nanocube (Au-IONP). The surface of its Au component was covalently immobilized with Gd- cystamine-DOTA₂ (dithiol derivative of DOTA) as a T₁ material (Gd(DOTA)-HNTs). To reduce its magnetic coupling with the IONP (T₂ material), the size of the Au nanocrystals was increased by controlling the seed-mediated growth processes during the synthesis and the size of the Pt cubes was also increased to increase the distance (D, Figure 9a, right panel) between the IONP and the Au nanocrystals. The resulting heterotrimers with large Au crystals had a dumbbell structure, (Gd(DOTA)-DB-HNTs and Gd(DOTA)-XDB-HNTs, the latter with a larger Au component and thus a larger number of Gd per single NP). The IO component was covered with PEG chains to make the whole NPs water-soluble and biocompatible (Figure 9b). The Gd(DOTA)-DB-HNTs and Gd(DOTA)-XDB-HNTs had increased r₁ values due to the reduced magnetic coupling between the T₁ and T₂ components as their D values increased. Their r₂ values were similar due to the similar sizes and shapes of the iron oxide components. The calculation of the r₁ relaxivity of each particle was based on different concentrations of Gd: r₁ (mM [Fe + Gd]⁻¹.s⁻¹) or r₁′ (mM [Gd]⁻¹.s⁻¹). Even though the r₂/r₁ ratio of Gd(DOTA)-DB-HNTs was 33, their r₂/r₁´was only 4.2, indicating that they could be T₁–T₂ DMCAs. Although Gd(DOTA)-XDB-HNTs have slightly higher r₁′ (32.1 mM [Gd]⁻¹.s⁻¹), they did not have a better r₂/r₁´ ratio (4.2) than Gd(DOTA)-DB-HNTs because of their larger size of the IO component, which increased r₂. The hybrid heterotrimers were highly stable in physiological conditions and induced simultaneous positive and negative contrast enhancements in in vivo MRI images of HT-29 tumor-bearing mice upon i.v. injection of Gd(DOTA)-DB-HNTs, showing their potential as T₁–T₂ DMCAs [127].

Fe₃O₄/MnO hybrid nanocrystals were prepared based on seed-mediated growth of MnO on the surface of a Fe₃O₄, where the resulting structures depended on the size of the seed NP, producing core/shell spherical Fe₃O₄ (5 nm)@MnO, dumbbell shaped Fe₃O₄ (11 nm)/MnO, and flower-shaped Fe₃O₄ (21 nm)/MnO hybrid NPs from 5 nm, 11 nm, and 21 nm Fe₃O₄ NPs, respectively (Figure 10). All the NPs had a high r₂/r₁ ratio and the dumbbell shaped Fe₃O₄/MnO NPs produced a negative contrast effect in T₂^*_w MRI images and a positive contrast effect in T_1w MRI images upon releasing Mn²⁺ ions in a low pH environment, in vitro (aqueous phantoms) and in vivo (normal brain of a nude mouse) obtained after i.v. injection. The same study using an orthotopic xenograft model of human hepatocellular carcinoma (HCC) showed high contrast between relatively hyperintense HCC and hypointense background liver parenchyma in T_2w MRI HCC and hypointense background liver parenchyma due to the presence of Kupffer cells in the later, while in T_1w MRI images a bright signal of the HCC tumor was observed after injection, due to HCC cell uptake of the Mn²⁺ ions liberated from the NPs in the low pH environment of the TME (Figure 11). This induced organ-specific contrast enhancement showed the potential of the hybrid NPs and T₁–T₂ DMCAs [109].

In summary, the synthetically challenging strategy of designing DMCAs based on forming hybrid oligomers of different shapes using T₁ and T₂ contrast materials, although scientifically original, has so far produced nanosystems with too high r₂/r₁ values [109,127].

4. General Issues of In Vivo Use of NPs as MRI Contrast Agents

Besides the problems of design and contrast efficacy described in previous sections, the use of MNPs as MRI CAs in vivo involves many other issues, including their colloidal stability, biocompatibility, absorption, distribution, metabolism, and excretion (ADME) [128], toxicity, targeting and potential theranostic use.

Several intrinsic properties of MNPs, such as size, shape, coating, charge and presence of surface ligands, influence their biodistribution, elimination and target site accumulation. MNPs should be colloidally stable in aqueous media and in body fluids, as in vivo agglomeration and precipitation causes unacceptable safety issues and limits their high-performance MRI function. The colloidal stability is ensured by high zeta potentials and coating with hydrophilic and biocompatible ligands, such as PEG, which also maintains their non-toxicity. The biocompatibility of the materials used in DMCAs is also quite variable. Fe³⁺-based NPs are more biocompatible than those containing Gd³⁺, Dy³⁺, Ho³⁺, Tb³⁺ and Mn²⁺, because iron is an essential element in the human body, participating in a wide variety of metabolic processes, including oxygen transport, DNA synthesis, and electron transport. However, its concentration in body tissues must be tightly regulated because in excessive amounts, Fe²⁺ is toxic through the generation of reactive oxygen species (ROS) via the so-called Fenton reaction, which eventually results in cell damage, cell death, and organ failure, primarily affecting the liver, heart, pancreas, thyroid, and central nervous system [129]. Although Mn is also an essential trace element, excessive doses of Mn²⁺ can be neurotoxic, and its build-up in the brain may lead to manganism, a neurological disorder similar to Parkinsonism [130]. Also, free Gd³⁺ ions liberated from Gd chelates or NPs into the body could cause nephrogenic systemic fibrosis (NSF) [14,15]. Non-toxicity of metal-based NPs is critical for their in vivo use as MRI CAs [131].

Because MRI CAs are usually intravenously injected, the preferable excretion of MNPs is through the renal system rather than the hepatobiliary pathway because the later one is relatively slow and the MNPs could be taken up by the liver reticuloendothelial system (RES) and be partially decomposed during the excretion process, which would be toxic to the body. This metabolization process of small (<10 nm) NPs can occur through direct penetration of the cell membrane (phagocytosis) and delivery to the cytoplasm and nucleus, while for larger NPs (10–200 nm) the uptake is mediated by clathrin-dependent endocytosis- Both mechanisms facilitate the cell lysosomal internalization and degradation of the MNPs [128].

For renal excretion, MNPs should be ultrasmall with hydrodynamic diameters less than 5 nm because the glomerular filtration diameter in the kidneys is 4.5–5 nm [132]. The kinetic stability of MNPs should also be high to avoid their decomposition until they are excreted as urine through the renal system.

In summary, MNPs as MRI CAs for safe in vivo applications should be kinetically stable (i.e., no decomposition), coated with hydrophilic and biocompatible polymers for non-toxicity and colloidal stability, and ultrasmall with hydrodynamic diameters less than 5 nm for renal excretion. This excretion route, avoiding as far as possible their liver retention in the hepatobiliary pathway, is also very important for targeting and theranostic drug delivery by MNPs [133,134]. In this review, several examples of in vivo use of theranostic DMCAs were presented, e.g., using ferroptosis inhibition [78], immunotherapy [79] and photodynamic therapy (PTT) associated with chemotherapy (ChT) [82] and chemodynamic therapy (CDT) [84]. As far as nanocarriers for drug delivery and theranostics are concerned, the inappropriate release, internal instability and tissue non-targeting effects, sometimes observed due to biological barriers, are the main restrictions for their in vivo application [128,135].

5. Conclusions

This review describes recent developments and optimization of MNPs containing Gd, Mn, Fe and other lanthanide ions as potential dual-mode T₁–T₂ MRI contrast agents (DMCAs). Their high performance was highlighted by describing selected in vivo MRI studies. However, the development stage of most of the reported MNP-based MRI CAs is still quite limited, both in vitro and in preclinical in vivo small animal studies. To improve their chances to reach the clinical trials stage, several key issues must be solved. These include long-term colloidal stability in aqueous media, toxicity and pharmacokinetics, which depend on their coating with hydrophilic and biocompatible ligands. In comparison with MNPs containing Mn, or lanthanides (such as Gd, Dy, Ho, and Tb), Fe-based NPs are less toxic because iron is an essential element, and several IONPs have been approved and commercialized as MRI CAs, such as Feridex^®, Sinerem^®, and Resovist^®, before being discontinued due to lack of commercial interest. The understanding of the correlation between the physicochemical properties of MNPs and their biological behavior in vivo must also be improved using appropriate preclinical small animal studies.

Another major problem faced by the clinical introduction of new MRI CAs in general, and MNP-based MRI CAs in particular, due to their molecular complexity and the consequent expensive synthesis, is to solve the main regulatory challenges of controlled NPs synthesis, uniformity, batch-to-batch reproducibility and upscaling of production. This leads to large development costs, especially for phase III studies, compared with the post-approval revenues [136,137]. However, the large amount of information obtainable with molecular imaging using MNP-based MRI CAs, such as those described in this review, may contribute to early diagnosis and personalized therapies, which ultimately will reduce health care costs.