FOCUS ON: Pediatric Imaging
Review
Strategies to Reduce the Use of Gadolinium-Based Contrast Agents for Abdominal MRI in Children
Audio Available | 
Share
OBJECTIVE. The objective of this article is to review strategies to reduce the use of gadolinium-based contrast agents for pediatric abdominal MRI.
CONCLUSION. Alternative contrast agents that do not contain gadolinium and unenhanced pediatric abdominal MRI protocols have shown clinical utility. Sequences such as DWI and new multicontrast MRI pulse sequences offer promise for tissue characterization without IV contrast agents. Patients requiring repeat MRI to evaluate for change in focal disease can be monitored with unenhanced abdominal MRI.
Keywords: ferumoxytol, gadolinium deposition, pediatric abdominal MRI, unenhanced MRIs
Gadolinium-based contrast agents (GBCAs) are the most commonly used IV contrast agents for MRI. Standard indications for the use of GBCAs include detection and characterization of focal visceral organ lesions, cancer staging, delineation of the vasculature and vascular disease, evaluation of the biliary (if using a hepatobiliary agent) and urinary systems, and assessment of infectious and inflammatory processes [1]. Although GBCAs are approved for clinical use by government agencies around the world including the U.S. Food and Drug Administration (FDA), the risks of gadolinium have received attention from both the medical literature and the lay press, most recently related to gadolinium deposition in the CNS in patients receiving multiple injections of GBCA. Given that young patients are at risk of receiving multiple doses of GBCA if they undergo serial abdominal MRI examinations over their lifetime (including patients with chronic conditions such as cancer or inflammatory bowel disease), every effort should be made to minimize patient GBCA exposure. The objective of this article is to review strategies to reduce the use of GBCAs for pediatric abdominal MRI.
| Risks Associated With GBCAs |   |
|---|
Gadolinium is a rare-earth metal that can serve as a contrast agent for MRI because of its paramagnetic properties including enhanced T1 relaxivity, which causes high signal intensity of gadolinium-containing tissues on T1-weighted sequences. Although the free gadolinium ion is toxic, when it is enveloped in an organic chelator, the compound becomes water soluble and can be safely administered IV and excreted as a GBCA [2, 3].
Concerns about the safety of GBCAs arose when an association between GBCAs and the development of nephrogenic systemic fibrosis (NSF), a fibrosing dermopathy, in patients with preexisting renal impairment was reported in 2006 [4, 5]. The leading hypothesis is that separation of free gadolinium from its chelate leads to skin deposition in patients with impaired renal function that results in prolonged gadolinium circulation time [6]. The incidence of NSF appears to be associated with age, with few reported cases of pediatric NSF [7]. A number of safety measures were implemented worldwide to decrease NSF, including screening to identify patients with severe kidney impairment, avoiding GBCAs in patients with acute kidney injury or stage 4–5 chronic kidney disease (estimated glomerular filtration rate < 30 mL/min/1.73 m2), limiting GBCAs to the minimum necessary dose (i.e., no double dose administrations), and choosing the specific GBCA on the basis of gadolinium chelate stability. These measures have led to the essential eradication of NSF worldwide; a recent systematic review found that fewer than 10 cases of biopsy-proven NSF after GBCA administration have been reported since 2008 [8]. To our knowledge, no cases of NSF in pediatric patients with normal renal function have been published.
More recently, studies have found gadolinium deposition in tissues, including in the brain and bones, of patients with normal renal function who have undergone numerous MRI studies [9–11]. The most common sites of gadolinium deposition are in the brain, within the posterior fossa and basal ganglia. This deposition has been shown by both T1-weighted MRI as well as mass spectrometry. The clinical significance of gadolinium deposition in the brain and elsewhere is unknown, with early preclinical data suggesting that gadolinium in the brain is cleared over time and not associated with histologic evidence of neurotoxicity [1, 12]. Different GBCAs have deposition characteristics depending on their chelator chemical structure, with animals and humans both showing higher rates of gadolinium retention after administration of various linear GBCAs relative to macrocyclic GBCAs [13–15]. The overall amount of retained gadolinium increases with the number of examinations in which GBCAs are used. The FDA responded to concerns about tissue deposition by issuing a medication guide to all patients receiving a GBCA injection for the first time in outpatient facilities [16]. The impact of gadolinium deposition on short- and long-term clinical disease is the subject of active investigation [10]. The potential long-term consequences of gadolinium deposition in the pediatric population remain unknown. Because pediatric patients may undergo multiple contrast-enhanced examinations over their lifetime and have a longer latency period after GBCA administration, the clinical benefits versus potential harm of GBCAs in children should be considered before they are used.
Additional risks of GBCAs include adverse allergiclike and physiologic reactions. The frequency of adverse reactions to GBCAs in pediatric patients is low, and when they do occur, they are most commonly mild. A multicenter prospective trial of safety of a single GBCA in 1142 children produced adverse events in only eight (0.7%), none of which were severe [17]. Five of the eight were physiologic reactions (0.4%); the other three were allergiclike reactions (0.3%). A single-center retrospective analysis of 32,365 GBCA administrations between 2009 and 2017 found 21 allergiclike reactions in children (0.06%), including 10 mild, 10 moderate, and one severe reaction [18]. Another large single-center retrospective analysis of 13,344 GBCA administrations in children reported six allergiclike reactions (0.04%), five of which were mild and one of which was severe [19]. Physiologic reactions to GBCAs are also rare in children. A single-center retrospective analysis of 22,185 GBCA administrations between 2009 and 2017 in patients 0–20 years old showed 39 physiologic reactions (0.2%), none of which were severe [20].
Another challenge associated with GBCA administration is the necessity of IV access, which can be difficult for infants and children and can decrease use of MRI. A unique anatomic obstacle for obtaining efficient and effective vascular access for IV contrast material administration in children are the frequent small mobile veins that can be obscured by robust subcutaneous fat. Accessing these veins can prove even more difficult in the setting of less cooperative patients, and this process can be stressful for the patient and their family. GBCA-enhanced MRI examinations require longer acquisition times, especially for multiphase or dynamic contrast-enhanced images, which prolongs time in the scanner and increases the potential anesthetic medication dose in children under sedation. Finally, because use of gadolinium adds cost to MRI examinations, eliminating it when possible decreases the financial burden on patients and the healthcare system, although many of the issues seen with GBCAs also apply to alternative contrast agents being considered for pediatric abdominal MRI.
| Alternative Contrast Agents | |
|---|
Active efforts are underway to identify new MRI contrast agents that potentially avoid the tissue deposition issues associated with GBCAs. Superparamagnetic iron-oxide nanoparticles (SPIONs) are insoluble nano-sized iron oxide crystals coated with carbohydrate shells. These contrast agents differ from GBCAs in that they can provide significant negative contrast through their T2* shortening properties, in addition to providing positive contrast via T1 shortening [21]. The pharmacokinetics and distribution of SPIONs similarly depend on particle size. In general, SPIONs do not extravasate across vessel walls and are not filtered by the renal glomeruli [21]. Instead, these particles are cleared from the blood via phagocytosis by the reticuloendothelial system, including lymph nodes, spleen, liver, and bone marrow. Particles larger than 80 nm accumulate within reticuloendothelial tissues more quickly than smaller particles. Once phagocytized, these particles are metabolized into forms of iron that can subsequently be used as part of the normal iron body store, such as hemoglobin and ferritin [21], theoretically avoiding the pathologic tissue deposition associated with gadolinium.
As of 2019, the only SPION marketed in the United States is ferumoxytol (Feraheme, AMAG Pharmaceuticals). This agent has been approved by the FDA for parenteral iron supplementation. However, it has also been used off-label as an MRI contrast agent [22, 23]. The typical dose of ferumoxytol for iron supplementation is 7 mg/kg compared with reported doses of 1–4 mg/kg when used for MRI contrast enhancement [24].
Ferumoxytol has been evaluated as a replacement for GBCAs in MR angiography (MRA) and MR venography (MRV) [24] (Fig. 1). Because of the small particle size (17- to 31-nm diameter) of ferumoxytol, it has a long blood pool half-life (12–14 hours). This feature provides an advantage over GBCAs for vascular imaging in children, for whom an extended angiographic or venographic window may be desirable for repeat imaging after a single injection if the patient is unable to cooperate during the initial examination. A prolonged window for angiographic phase imaging also facilitates the use of motion-averaged techniques that can augment image quality during free-breathing examinations. Few published reports have assessed the use of ferumoxytol in pediatric patients. Available studies include a retrospective analysis that reported excellent diagnostic quality MRA and MRV examinations with ferumoxytol in 23 pediatric patients with doses of 1.5 or 3 mg/kg [25]. Another recent study found that ferumoxytol provided diagnostic quality MRV examinations in 22 children with chronic kidney disease at doses of 4 mg/kg [26]. Unenhanced vascular MRI techniques such as time-offlight angiography can accomplish similar goals in patients for whom GBCAs are not suitable, although they come with their own challenges including long acquisition time and signal loss of in-plane flow. Finally, the use of ferumoxytol in the assessment of brain tumor vascularity has been investigated in pediatric patients, finding that quantifying relative cerebral blood flow and relative cerebral blood volume using ferumoxytol-enhanced dynamic T2*-weighted images was useful for both preoperative planning and postoperative monitoring [27]. Once again, unenhanced techniques, such as arterial spin labeling, can also be considered for intracranial tumor perfusion evaluation.
 View larger version (222K) | Fig. 1A —13-year-old girl with end-stage renal disease secondary to antiglomerular basement membrane antibody disease who was dependent on peritoneal dialysis who presented for renal transplant evaluation. Contrast-enhanced MR angiography and MR venography were performed after IV injection of 6.5 mL ferumoxytol, based on 4 mg/kg weight-based dosing with 30 mg/mL formulation. A, Axial spoiled gradient-recalled echo images of abdomen (A) and pelvis (B) with fat saturation acquired during arterial phase show normal aortoiliac system (black arrows) and single bilateral renal arteries (white arrow, A). |
 View larger version (235K) | Fig. 1B —13-year-old girl with end-stage renal disease secondary to antiglomerular basement membrane antibody disease who was dependent on peritoneal dialysis who presented for renal transplant evaluation. Contrast-enhanced MR angiography and MR venography were performed after IV injection of 6.5 mL ferumoxytol, based on 4 mg/kg weight-based dosing with 30 mg/mL formulation. B, Axial spoiled gradient-recalled echo images of abdomen (A) and pelvis (B) with fat saturation acquired during arterial phase show normal aortoiliac system (black arrows) and single bilateral renal arteries (white arrow, A). |
 View larger version (253K) | Fig. 1C —13-year-old girl with end-stage renal disease secondary to antiglomerular basement membrane antibody disease who was dependent on peritoneal dialysis who presented for renal transplant evaluation. Contrast-enhanced MR angiography and MR venography were performed after IV injection of 6.5 mL ferumoxytol, based on 4 mg/kg weight-based dosing with 30 mg/mL formulation. C, Axial spoiled gradient-echo image of abdomen with fat saturation acquired during venous phase shows single bilateral renal veins with conventional relationship to aorta (arrows). |
 View larger version (249K) | Fig. 1D —13-year-old girl with end-stage renal disease secondary to antiglomerular basement membrane antibody disease who was dependent on peritoneal dialysis who presented for renal transplant evaluation. Contrast-enhanced MR angiography and MR venography were performed after IV injection of 6.5 mL ferumoxytol, based on 4 mg/kg weight-based dosing with 30 mg/mL formulation. D, Coronal subtraction images reconstructed from unenhanced (D) and arterial phase contrast-enhanced (E) images show single bilateral renal arteries (arrows, D) and normal aortoiliac system (arrows, E). |
 View larger version (304K) | Fig. 1E —13-year-old girl with end-stage renal disease secondary to antiglomerular basement membrane antibody disease who was dependent on peritoneal dialysis who presented for renal transplant evaluation. Contrast-enhanced MR angiography and MR venography were performed after IV injection of 6.5 mL ferumoxytol, based on 4 mg/kg weight-based dosing with 30 mg/mL formulation. E, Coronal subtraction images reconstructed from unenhanced (D) and arterial phase contrast-enhanced (E) images show single bilateral renal arteries (arrows, D) and normal aortoiliac system (arrows, E). |
Beyond vascular imaging, a prospective pilot study assessed the use of ferumoxytol in whole-body MRI as a tumor staging strategy in 22 pediatric and young adult patients with lymphomas and sarcomas [28]. Patients in that study underwent whole-body MRI with DWI and T1-weighted imaging after ferumoxytol administration compared with FDG PET/CT. The addition of ferumoxytol to whole-body MRI in these patients was postulated to improve lesion detection in the bone marrow and spleen on DWI by decreasing background parenchymal signal intensity and by enabling an angiographic phase long enough to allow acquisition of a whole-body contrast-enhanced T1-weighted sequence for anatomic localization of the DWI findings. Overall, the study found that the combination of ferumoxytol enhancement and whole-body MRI provided similar sensitivity, specificity, and accuracy compared with FDG PET/CT [28].
SPION-enhanced MRI has also been investigated as a technique for lymph node characterization. Normal lymph nodes show reduced signal intensity on SPION-enhanced T2-weighted or T2*-weighted images obtained 24 hours after contrast injection compared with unenhanced images because of the contrast accumulation within normally distributed nodal macrophages. However, tumor cell displacement of macrophages within malignant lymph nodes results in a persistent high signal intensity in pathologic lymphadenopathy on SPION-enhanced imaging. Using the investigational SPION ferumoxtran-10, Harisinghani et al. [29] were the first to find that SPIONs had very high sensitivity for otherwise undetectable lymph node metastases on conventional MRI in patients with prostate cancer. Subsequent studies have confirmed similar findings with ferumoxytol in adults with prostate cancer [30, 31]. To our knowledge, no studies have evaluated this technique in pediatric patients. However, given the promising results in adults, ferumoxytol is a non-GBCA that may have future applications in lymph node characterization as well as in whole-body MRI in pediatric patients.
Adverse reactions are more common with ferumoxytol than with GBCAs, which led to the release of a boxed warning by the FDA regarding the potential for serious allergic reactions to this agent [32]. During clinical trials, 0.2% of adult patients treated with ferumoxytol experienced a severe hypersensitivity reaction, and 18 deaths related to severe hypersensitivity reactions from IV ferumoxytol were reported to the FDA between 2009 and 2014 [21, 32]. Evaluation of adverse event rates in children is more limited. One study reported that only four mild adverse events occurred among 68 pediatric patients receiving ferumoxytol, including two occurrences of transient hypotension, one episode of self-resolving nausea, and one episode of a self-resolving injection site reaction [33]. Larger studies would be required to fully evaluate the risk in this population.
Another significant limitation in the use of ferumoxytol as a substitute for GBCA is the fact that, as a blood pool agent, it does not show the same equilibrium between intravascular and extravascular tissue compartments as GBCAs do. As a result, the dynamic enhancement characteristics of visceral organ lesions on ferumoxytol-enhanced MRI cannot reliably distinguish benign from malignant lesions or assess infectious processes, which significantly limits its use in these scenarios. Finally, ferumoxytol is generally more expensive than GBCAs, another relative weakness [34].
Ultimately, although ferumoxytol is used off-label as an MRI contrast agent at some pediatric centers, its lack of FDA approval in that context and the need for more safety data regarding its use in children make it unlikely to replace GBCAs in the near future.
| Novel Preclinical Contrast Agents in Development | |
|---|
Manganese functions similarly to gadolinium as a paramagnetic MRI contrast agent. Manganese leads to T1 shortening of local tissues, resulting in positive contrast on T1-weighted sequences that is comparable in magnitude to that seen with GBCAs [35–37]. Unlike gadolinium, manganese is an essential nutritional component, and efficient physiologic mechanisms regulate its levels in the human body. Consequently, dechelation of manganese-based contrast agents may present less toxic potential than dechelation of GBCAs [38]. Because of this theoretically improved safety profile, there is significant interest in developing manganese-based alternatives to GBCAs. One notable example is Mn-PyC3A, which was shown to have minimal dissociation of free manganese and produced images of comparable quality to those obtained using a GBCA in MRA examinations of baboons [39]. A separate study showed the agent was effective in the evaluation of liver tumors in mice [40]. The use of manganese-based contrast agents for MRI is still preclinical, and no safety data have been established for their use in pediatric patients.
GBCAs, SPIONs, and manganese-based contrast agents all generate contrast by inducing tissue relaxation. Other agents are being developed that generate MRI contrast through different mechanisms; including chemical exchange saturation transfer and direct probe detection via nonwater protons or other nuclei, such as carbon and fluorine [21, 41–43]. Compared with the primarily anatomic information provided by the tissue relaxation agents, nonrelaxation agents are thought to hold potential as metabolic and functional MRI probes [41, 42]. Although none of these agents are FDA-approved for MRI, they remain an area of active preclinical and early clinical investigation. For instance, several authors have reported utility of hyperpolarized 13C-pyurvate for evaluating the metabolic profile of prostate and brain malignancies in humans [44–47]. A 19F probe has been shown to detect activity of caspase-3, an important enzyme in the apoptotic pathway [48]. Finally, MRI using chemical exchange saturation transfer with amide protons, an endogenous probe, has been evaluated for glioma grading [49, 50], evaluation of cerebral ischemic stroke and hemorrhage [51], and breast cancer treatment response [52].
| MRI Tissue Characterization Using Unenhanced Sequences | |
|---|
MRI tissue characterization without IV contrast material relies primarily on evaluation of signal intensity on T1-weighted, T2-weighted, and DWI sequences. T1-weighted sequences are useful for evaluation of intralesional hemorrhage, mineralization, fat, and protein-aceous fluid [53]. Fast spin-echo (FSE) and turbo spin-echo (TSE) T1-weighted sequences tend to provide the best T1 contrast but are more susceptible to motion artifact because of their longer acquisition compared with gradient-recalled echo (GRE) sequences.
T1-weighted GRE sequences that use chemical-shift imaging can also be helpful in characterizing intravoxel lipid content [53, 54]. A variant of this technique commonly used in clinical practice is referred to as the Dixon technique, which separates the relative signal contributions of water and fat [55]. Multiple TEs are acquired where phase differences between fat and water signals are at their minimum and maximum. Resulting image pairs from TEs when fat and water protons are in phase and opposed phase show signal loss in voxels containing both fat and water proton signal on the opposed phase images. Adding and subtracting the datasets from in and opposed phase images yield water-only and fat-only images, respectively, producing both fat suppression and identification of fat in focal lesions and throughout entire organs [55]. Clinical applications of the Dixon method in the pediatric abdomen include identifying fat within focal lesions (Fig. 2) and identifying fat content within entire organs, as in hepatic steatosis, including potential for fat quantification via proton density fat fraction calculation.
 View larger version (214K) | Fig. 2A —14-year-old girl with tuberous sclerosis complex who presented for routine surveillance abdominal imaging. A, Axial gradient-recalled echo (GRE) T1-weighted Dixon images though upper abdomen acquired in phase (A) and opposed phase (B) as well as corresponding T2-weighted single-shot fast spin-echo image (C) show focal area of T2 hyperintensity with corresponding loss of signal on opposed phase image within hepatic segment VIII (arrow), likely representing lipid-poor angiomyolipoma. Signal loss of opposed phase relative to in phase GRE images suggests both fat and water within imaged voxels, resulting in signal cancellation. |
 View larger version (243K) | Fig. 2B —14-year-old girl with tuberous sclerosis complex who presented for routine surveillance abdominal imaging. B, Axial gradient-recalled echo (GRE) T1-weighted Dixon images though upper abdomen acquired in phase (A) and opposed phase (B) as well as corresponding T2-weighted single-shot fast spin-echo image (C) show focal area of T2 hyperintensity with corresponding loss of signal on opposed phase image within hepatic segment VIII (arrow), likely representing lipid-poor angiomyolipoma. Signal loss of opposed phase relative to in phase GRE images suggests both fat and water within imaged voxels, resulting in signal cancellation. |
 View larger version (264K) | Fig. 2C —14-year-old girl with tuberous sclerosis complex who presented for routine surveillance abdominal imaging. C, Axial gradient-recalled echo (GRE) T1-weighted Dixon images though upper abdomen acquired in phase (A) and opposed phase (B) as well as corresponding T2-weighted single-shot fast spin-echo image (C) show focal area of T2 hyperintensity with corresponding loss of signal on opposed phase image within hepatic segment VIII (arrow), likely representing lipid-poor angiomyolipoma. Signal loss of opposed phase relative to in phase GRE images suggests both fat and water within imaged voxels, resulting in signal cancellation. |
T2-weighted sequences form an essential component of abdominal MRI because of their high sensitivity for neoplastic lesions, inflammation, and fluid. Standard 2D T2-weighted sequences of the abdomen are typically acquired using FSE or TSE techniques in at least the axial plane and usually with fat suppression to improve lesion conspicuity [53]. Faster T2-weighted techniques including such as echo-planar imaging and single-shot FSE sequences are used to provide large FOV anatomic surveys and can be used to produce highly T2-weighted MRCP and MR urography (MRU) images [53, 56] (Fig. 3). These fast sequences minimize motion-related artifact while preserving reliable evaluation of fluid-filled structures, extraluminal fluid collections, and edema [57]. However, these strategies usually result in a lower signal-to-noise ratio (SNR) compared with standard T2-weighted FSE and TSE techniques, which could obscure lesions [53, 57]. Consequently, they are typically used as complementary sequences to standard T2-weighted FSE or TSE imaging.
 View larger version (286K) | Fig. 3A —2-month-old girl who presented for MR urography after obstetrical ultrasound showed duplicated right renal collecting system with dilated right ureter. A, Coronal balanced steady-state free precession images (A and B) and axial T2-weighted fast spin-echo (C and D) images with fat saturation through abdomen and pelvis, respectively, show dilated right renal collecting system (arrows; A, C, and D) with ectopic insertion of dilated right ureter into right urethra (arrow, B). Additional images showed dilated right ureter to be upper pole moiety of dilated collecting system (not shown). |
 View larger version (285K) | Fig. 3B —2-month-old girl who presented for MR urography after obstetrical ultrasound showed duplicated right renal collecting system with dilated right ureter. B, Coronal balanced steady-state free precession images (A and B) and axial T2-weighted fast spin-echo (C and D) images with fat saturation through abdomen and pelvis, respectively, show dilated right renal collecting system (arrows; A, C, and D) with ectopic insertion of dilated right ureter into right urethra (arrow, B). Additional images showed dilated right ureter to be upper pole moiety of dilated collecting system (not shown). |
 View larger version (180K) | Fig. 3C —2-month-old girl who presented for MR urography after obstetrical ultrasound showed duplicated right renal collecting system with dilated right ureter. C, Coronal balanced steady-state free precession images (A and B) and axial T2-weighted fast spin-echo (C and D) images with fat saturation through abdomen and pelvis, respectively, show dilated right renal collecting system (arrows; A, C, and D) with ectopic insertion of dilated right ureter into right urethra (arrow, B). Additional images showed dilated right ureter to be upper pole moiety of dilated collecting system (not shown). |
 View larger version (175K) | Fig. 3D —2-month-old girl who presented for MR urography after obstetrical ultrasound showed duplicated right renal collecting system with dilated right ureter. D, Coronal balanced steady-state free precession images (A and B) and axial T2-weighted fast spin-echo (C and D) images with fat saturation through abdomen and pelvis, respectively, show dilated right renal collecting system (arrows; A, C, and D) with ectopic insertion of dilated right ureter into right urethra (arrow, B). Additional images showed dilated right ureter to be upper pole moiety of dilated collecting system (not shown). |
Additional methods of acquiring T2-weighted FSE and TSE sequences involve filling k-space by non-Cartesian methods such as radially oriented phase-encoding lines (Fig. 4). These techniques decrease conspicuity of respiratory motion because the phase-encoding direction changes with each radial line [53]. Because each radial acquisition samples central k-space and is varied by a fixed angle, prospective motion correction can also be applied to further decrease motion artifacts [54]. The drawback of these techniques are longer acquisition times, radial artifacts, and blurring.
 View larger version (322K) | Fig. 4A —2-year-old girl with acute right-sided abdominal pain who was found to have perforated acute appendicitis and periappendiceal abscess. A, Coronal (A) and axial (B) T2-weighted single-shot fast spin-echo (FSE) images of abdomen and pelvis with fat saturation show prominent inflammatory changes in right lower abdomen centered around dilated appendix with thickened walls (white arrows), with inferior periappendiceal abscess (black arrows). |
 View larger version (263K) | Fig. 4B —2-year-old girl with acute right-sided abdominal pain who was found to have perforated acute appendicitis and periappendiceal abscess. B, Coronal (A) and axial (B) T2-weighted single-shot fast spin-echo (FSE) images of abdomen and pelvis with fat saturation show prominent inflammatory changes in right lower abdomen centered around dilated appendix with thickened walls (white arrows), with inferior periappendiceal abscess (black arrows). |
 View larger version (204K) | Fig. 4C —2-year-old girl with acute right-sided abdominal pain who was found to have perforated acute appendicitis and periappendiceal abscess. C, Axial T2-weighted FSE images of abdomen (C) and pelvis (D) with fat saturation acquired with radial k-space sampling have improved signal-to-noise ratio and contrast resolution compared with single-shot images, allowing improved delineation of appendix (arrow, C) and abscess (arrow, D). |
 View larger version (214K) | Fig. 4D —2-year-old girl with acute right-sided abdominal pain who was found to have perforated acute appendicitis and periappendiceal abscess. D, Axial T2-weighted FSE images of abdomen (C) and pelvis (D) with fat saturation acquired with radial k-space sampling have improved signal-to-noise ratio and contrast resolution compared with single-shot images, allowing improved delineation of appendix (arrow, C) and abscess (arrow, D). |
Static-fluid MRU can provide morphologic information about the urinary system using heavily T2-weighted sequences without IV contrast material [57]. These sequences use very high TE (500–1000 ms), similar to MRCP, and are tailored to the long T2 relaxation time of urine. MRU images can be acquired in three ways: as a thick-slab acquisition in 1–2 seconds, as a thin-section (1–2 mm) 3D respiratory triggered acquisition that can be reformatted into multiple planes, or as a cinematic evaluation of urine passage by imaging the same FOV serially every 5–10 seconds. GBCA is reserved for MRU situations in which visualization of contrast excretion into the collecting systems is desired or renal split function requires calculation.
Balanced steady-state free precession (bSSFP) GRE sequences can also be used for unenhanced evaluation of anatomic structures [1, 58, 59]. A major advantage of these sequences is short acquisition time from short TR and steady state longitudinal magnetization signal (leading to a combination of T1 and T2 weighting and high signal intensity). The bSSFP sequences are a white blood sequence and form the basis of many newer unenhanced MRA-MRV techniques. In addition, the chemical shift artifact at tissue interfaces with fat provides a natural contrast for bSSFP delineation of structures such as the retroperitoneum and small bowel mesentery. Finally, the high temporal resolution of bSSFP is helpful for cinematic evaluation of the heart and bowel [1, 58, 59].
Although not part of most pediatric abdominal MRI protocols, STIR is a fluid-sensitive sequence that provides robust fat suppression for large FOV acquisitions (coronal or sagittal). The robust fat suppression on STIR imaging enables excellent lesion conspicuity. As such, STIR imaging is commonly used in whole-body MRI applications and can be added for evaluation of very large retroperitoneal tumors [59]. Axial STIR acquisitions are not commonly acquired because they are less time efficient, have higher risk for motion artifact, and provide lower signal intensity than T2-weighted sequences with spectral fat suppression [53].
Finally, DWI has been increasingly incorporated into pediatric abdominal MRI protocols. In pediatric patients, DWI sequences are most commonly acquired using a free breathing single-shot FSE echo-planar technique [53]. Apparent diffusion coefficient (ADC) maps are calculated from DW images acquired at multiple different b values, which allow quantitation of the diffusion coefficients of tissues. These maps must be qualitatively correlated with DW images to determine whether DWI hyperintensity is due to restricted diffusion for lesions with corresponding low signal intensity on ADC maps or due to T2 shine-through for lesions with corresponding high signal intensity on ADC maps. Several normal tissues can appear hyperintense on DWI either because of restricted diffusion or long T2 relaxation times, including lymph nodes, splenic tissue, thymus, gonadal tissue, endometrium, prostate, red marrow, and the spinal cord [54, 60, 61]. The typical high b values used in pediatric abdominal imaging range from 500 to 1000 mm/s2 [62]. The addition of low b values (20–100 mm/s2) may help improve lesion detection sensitivity [63].
DWI allows indirect evaluation of cellular density, cell membrane integrity, and fluid complexity by interrogating free water brownian motion within a voxel of imaged tissue. Clinical applications include identifying abnormally restricted diffusion of water molecules in the setting of narrowed interstitial space caused by cellular swelling from ischemia or acute inflammation and due to hypercellular lesions [53, 62, 64] (Fig. 5). DWI has been evaluated for several pediatric oncologic applications, including malignant lesion detection and characterization, whole-body MRI for tumor staging and screening, and assessment of oncologic treatment response. DWI has been shown to be reliable in the detection of malignant neoplasms in pediatric patients [64]. There is also evidence that the use of DWI and ADC map values may help differentiate between benign and malignant causes and assess treatment response [65–69]. However, translation of these findings to routine clinical use has been hindered by the lack of established, reliable ADC thresholds for these applications, substantial overlap in ADC values between benign and malignant tumors, and high interobserver variability in ADC measurements [70, 71]. Whole-body MRI utilizing DWI along with other sequences is an emerging strategy that may provide an ionizing radiation–free alternative to PET/CT for pediatric tumor staging [28, 60, 72, 73].
 View larger version (219K) | Fig. 5A —5-year-old boy undergoing follow-up imaging of biopsy-proven retroperitoneal neuroblastoma after receiving chemotherapy. A, Axial T2-weighted fast spin-echo image (A) with fat saturation acquired, DW image (B), and corresponding apparent diffusion coefficient (ADC) map (C) of abdomen show 7-cm right retroperitoneal mass with mild, heterogeneous hyperintensity on T2-weighted image and hyperintensity on DWI with corresponding hypointensity on ADC map (arrows), in keeping with densely cellular diffusion-restricting malignant neuroblastoma. |
 View larger version (182K) | Fig. 5B —5-year-old boy undergoing follow-up imaging of biopsy-proven retroperitoneal neuroblastoma after receiving chemotherapy. B, Axial T2-weighted fast spin-echo image (A) with fat saturation acquired, DW image (B), and corresponding apparent diffusion coefficient (ADC) map (C) of abdomen show 7-cm right retroperitoneal mass with mild, heterogeneous hyperintensity on T2-weighted image and hyperintensity on DWI with corresponding hypointensity on ADC map (arrows), in keeping with densely cellular diffusion-restricting malignant neuroblastoma. |
 View larger version (238K) | Fig. 5C —5-year-old boy undergoing follow-up imaging of biopsy-proven retroperitoneal neuroblastoma after receiving chemotherapy. C, Axial T2-weighted fast spin-echo image (A) with fat saturation acquired, DW image (B), and corresponding apparent diffusion coefficient (ADC) map (C) of abdomen show 7-cm right retroperitoneal mass with mild, heterogeneous hyperintensity on T2-weighted image and hyperintensity on DWI with corresponding hypointensity on ADC map (arrows), in keeping with densely cellular diffusion-restricting malignant neuroblastoma. |
Outside of oncologic applications, DWI can be useful for the detection of abscesses in the abdomen, pelvis, and soft tissues [74]. DWI has also shown utility in the detection of bowel wall inflammation including the appendix and of fistulas associated with inflammatory bowel disease [54, 75].
New multicontrast MR pulse sequences have been developed (e.g., synthetic MRI, MR fingerprinting) that aim to quantify the T1, T2, and proton density time constants of tissue on a per-voxel basis [76]. These techniques offer the promise of quantitative tissue characterization without the need for IV contrast material. Additionally, multiple MR weightings can be ascertained in a single acquisition, which has the potential to significantly reduce scan times in children. However, these new sequences are not in widespread clinical use. Future studies are needed to assess the diagnostic performance of these techniques, as well as their potential impact on pediatric abdominal MRI.
| Pediatric Abdominal Unenhanced MRI Protocols | |
|---|
Some pediatric abdominal MRI protocols are being performed in the clinical setting without gadolinium. Two of the best studied are pediatric appendicitis MRI and MR enterography (MRE) in patients with established Crohn disease.
Standard pediatric appendicitis protocols do not require GBCAs. Furthermore, the sequences can be obtained in less than 5 minutes, obviating sedation in the vast majority of cases. Performing these examinations without IV contrast material allows them to be performed in a shorter amount of time and also obviates IV cannulation, both of which make MRI better tolerated in young children suspected of having appendicitis. Multiple studies have found high sensitivity (93–97%), specificity (96–98%), positive predictive value (92–97%), and negative predictive value (96–98%) of unenhanced MRI for diagnosis of pediatric appendicitis [77–79]. Utilizing MRI for evaluation of pediatric appendicitis has the advantage over CT of no ionizing radiation, and MRI without GBCA has been found to be comparable to CT with regard to sensitivity and specificity in the detection of appendicitis [80]. The superior soft-tissue contrast of MRI also enables evaluation for appendiceal disease without IV contrast enhancement, as opposed to CT, for which IV contrast material is required. Additionally, although ultrasound is another modality without ionizing radiation that is typically used for initial evaluation of pediatric appendicitis, MRI has been shown to alter management in many patients who have indeterminate ultrasound findings, including increasing diagnostic certainty of positive and negative cases, identifying appendicoliths and abscesses precluding nonsurgical management, diagnosing mimics of appendicitis such as tuboovarian abscess, and identifying a safe drainage pathway [81].
A typical unenhanced appendicitis abdominal MRI protocol includes axial and coronal T2-weighted single-shot FSE sequences with and without fat saturation. Additional sequences can be acquired that have better contrast resolution for appendiceal inflammation, such as radial T2-weighted FSE (Fig. 4). Although commonly respiratory triggered or navigated, free-breathing techniques are now possible with the use of radial k-space sampling. DWI can be added to help address questions of drainable collections, which has been shown to be cost-effective in patients who have undergone appendectomy [81–83]. One pitfall of unenhanced appendicitis protocol MRI is the potential for mis-diagnosing solid diffusion-restricting tumors such as lymphoma or myeloid sarcoma as abscesses [81].
Crohn disease is the form of inflammatory bowel disease that can affect any part of the gastrointestinal tract and frequently presents in childhood; as a result, these patients have potential for a high cumulative ionizing radiation dose from imaging evaluation of the small bowel that is inaccessible to optical endoscopy [84]. MRI has become the primary imaging modality for young patients with Crohn disease that both avoids ionizing radiation and has high performance for noninvasively assessing disease activity [85]. Gadolinium has traditionally been a part of MRE protocols, with contrast-enhanced imaging features such as mural hyperenhancement and peripherally enhancing fluid collections being standard MRI bio-markers of disease activity and penetrating abscesses, respectively [86]. However, several MRI features of Crohn disease activity do not require GBCA to assess, including wall thickening, edema, diffusion restriction, and vasa recta engorgement. Non–gadolinium-enhanced MRE sequences can also detect Crohn disease complications including sequelae of penetrating disease (bSSFP is especially helpful for detecting ulcerations and fistulas), strictures (Fig. 6), and extraintestinal manifestations such as primary sclerosing cholangitis and sacroiliitis.
 View larger version (315K) | Fig. 6A —13-year-old boy with history of Crohn disease who presented with right upper quadrant pain. A, Coronal (A) and sagittal (B) T2-weighted single-shot fast spin-echo without fat saturation and axial T2-weighted single-shot fast spin-echo with fat saturation (C) images of abdomen show 2-cm segment of narrowed small bowel with associated wall thickening (black arrows) and upstream bowel dilatation (white arrow, A and C), in keeping with stricture. There is no significant mural T2 hyperintensity or surrounding inflammatory change to suggest superimposed acute inflammation. |
 View larger version (383K) | Fig. 6B —13-year-old boy with history of Crohn disease who presented with right upper quadrant pain. B, Coronal (A) and sagittal (B) T2-weighted single-shot fast spin-echo without fat saturation and axial T2-weighted single-shot fast spin-echo with fat saturation (C) images of abdomen show 2-cm segment of narrowed small bowel with associated wall thickening (black arrows) and upstream bowel dilatation (white arrow, A and C), in keeping with stricture. There is no significant mural T2 hyperintensity or surrounding inflammatory change to suggest superimposed acute inflammation. |
 View larger version (294K) | Fig. 6C —13-year-old boy with history of Crohn disease who presented with right upper quadrant pain. C, Coronal (A) and sagittal (B) T2-weighted single-shot fast spin-echo without fat saturation and axial T2-weighted single-shot fast spin-echo with fat saturation (C) images of abdomen show 2-cm segment of narrowed small bowel with associated wall thickening (black arrows) and upstream bowel dilatation (white arrow, A and C), in keeping with stricture. There is no significant mural T2 hyperintensity or surrounding inflammatory change to suggest superimposed acute inflammation. |
In patients with Crohn disease who require repeat MRE, some institutions have implemented protocols that do not include gadolinium use for patients with established disease. Two recent studies compared the performance of unenhanced and contrast-enhanced MRE for evaluating activity in pediatric Crohn disease [87, 88]. Both studies found no significant improvement in disease assessment when gadolinium was administered. However, the addition of GBCAs did improve sensitivity for detection of penetrating disease complications. Typical unenhanced MRE protocols include single-shot FSE, bSSFP, thick-slab cinematic bSSFP, and DWI sequences [85, 89]. DWI has shown utility for identifying active disease and extraluminal collections [90, 91]. These results suggest that, for routine evaluation of Crohn disease activity, unenhanced MRE has acceptable diagnostic performance and should be considered in patients at risk of undergoing repeated gadolinium administrations. However, many institutions give IV GBCAs for MRE protocols in children.
| Guiding Principles for Minimizing Use of GBCAs in Pediatric Abdominal MRI | |
|---|
As with any other medication, IV gadolinium should only be administered to patients when there is sufficient diagnostic benefit. Using available unenhanced protocols for specific indications and reviewing imaging requests with care to prescribe use of contrast material only when indicated is a crucial role for radiologists. We have identified several areas in which gadolinium use can be decreased in pediatric abdominal MRI. For example, the use of alternative contrast agents such as ferumoxytol is likely to be as good as GBCAs for most MRA-MRV studies and holds potential for evaluation of lymphadenopathy. The current major limitations on its use are its off-label imaging indication, inability to characterize visceral organ lesions, lack of well-established safety data, and higher rate of adverse events. For characterization of visceral organ lesions, and evaluation of suspected infectious or inflammatory processes, GBCAs will likely still be needed for the near future. However, as the examples of appendicitis and Crohn disease show, the ability of MRI to characterize tissue without gadolinium continues to improve, and new techniques such as multicontrast sequences may obviate GBCA in the future. In addition, much of the reduction in gadolinium use can be achieved by decreasing the use of contrast material in follow-up examinations of patients with known disease. In patients with established cancer or infectious or inflammatory conditions, serial MRI examinations are often performed to evaluate treatment effect. In these cases, lesions are typically visible on both unenhanced and contrast-enhanced images on the initial MRI, so assessment for interval change will not require gadolinium in most cases. Finally, the use of contrast-enhanced ultrasound is likely to replace GBCA-enhanced MRI in many cases of incidentally discovered abdominal lesions, particularly in young children who would require sedation to undergo MRI [92]. Radiologists can play an important role in the care of individual patients and in the overall healthcare system by utilizing these strategies to reduce the use of gadolinium in pediatric abdominal MRI.
| Conclusion | |
|---|
Reasons to reduce GBCA usage in pediatric patients include the risk for gadolinium tissue deposition, safety concerns in patients with impaired renal function, allergiclike and physiologic adverse reactions, and need for IV access. Strategies to reduce the use of gadolinium contrast agents for pediatric abdominal MRI include the use of contrast agents that do not contain gadolinium such as ferumoxytol, the utilization of unenhanced sequences such as DWI, implementation of unenhanced abdominal MRI protocols when indicated, and use of unenhanced techniques for repeat MRI to evaluate for change in previously characterized focal disease.
| References | |
|---|
