Sarah Brasher1, Annie Chan1, Ayaz Khan2, Zachary Abramson2, Cara Morin3, and Aaryani Tipirneni-Sajja1,2
1Biomedical Engineering, University of Memphis, Memphis, TN, United States, 2Diagnostic Imaging, St. Jude Children's Research Hospital, Memphis, TN, United States, 3Radiology, Cincinnati Children's Hospital Medical Center, Cincinnati, OH, United States
Synopsis
Keywords: Liver, Phantoms
Size
and concentration of iron will affect the dephasing of an MRI signal. However, iron
particle size is often unaccounted for in phantom studies investigating iron
overload. In this study, phantoms utilizing iron nanoparticles of different
diameters were used to mimic in vivo iron deposits, and R2* quantification was
analyzed using different signal models. Our results show that R2* was higher and
more unstable in phantoms with iron particles of 500nm diameter in comparison
to those with 250nm and 130nm particles potentially due to clustering. High
iron concentrations and large iron particle sizes were also shown to confound fat
quantification.
Introduction
Under
conditions of hereditary hemochromatosis or transfusional hemosiderosis, excess
iron can build up in the liver leading to iron overload1. Liver
biopsy is the current gold standard for evaluating hepatic iron overload, but
it has limitations resulting from the invasiveness of the procedure and
high sampling variability arising from the heterogeneity of iron deposition1.
Alternatively, R2*-MRI-based techniques are emerging as robust and clinically
accepted methods for non-invasively assessing hepatic iron overload2.
In the human body, iron is normally stored in ferritin (50-70 nm)3
or, under iron overload conditions, as hemosiderin (0.5-2 µm)4. Additionally, iron in the presence of fat (steatosis)
has been shown to preferentially effect water protons depending on the size and
distribution of all nearby particles5. However, current imaging
phantoms that emulate hepatic iron overload often do not consider the size of
in vivo iron storage proteins in the chosen iron material as well as the interactions
of iron with surrounding particles. Thus, the primary goal of this study is to
analyze the impact of iron particle size on the quantification of R2* and fat fraction
(FF) in iron-agar and iron-agar-fat phantoms for increasing iron concentrations.
Materials and Methods
Two
sets of homogeneous iron emulsion phantoms were produced at solid (iron dextran)
contents of 0%, 0.075%, 0.16%, 0.24%, and 0.32% with the first set of phantoms
containing only iron (iron-only) while the second set also containing fat with FF
of 20% (iron-fat). Each set of phantoms contained iron nanoparticles (Nanomag-D
Plain, Micromod, Rostock, Germany) of diameters 130nm, 250nm, and 500nm and
were held in suspension using 2% agar. The iron-fat phantom set included peanut
oil emulsified with 43mM SDS. All phantoms were homogenized using a handheld
homogenizer (D1000, Benchmark Scientific, Sayreville, NJ, USA) with a saw tooth
generator probe at ~15000 rpm. A total of 30 phantoms were created: 3 different
iron particle diameters x 5 iron concentrations x 2 (iron-only, iron-fat) sets.
Scanning
electron microscopy (SEM) was used to quantify the size and distribution of the
different iron nanoparticles used in this study. Each sample was sputter coated
with a gold/palladium (80:20) coat using an EMS550x Sputter Coater (Quorum,
Laughton, East Sussex, England) and mounted on 25mm diameter aluminum specimen
stubs. SEM analysis was performed with a Nova NanoSEM 650 (FEI Company,
Hillsboro, Oregon, USA) at a voltage of 20.00 kV, spot size of 3.0, working
distance of 5.5-5.2 mm, and a field-free lens mode.
MRI scans
were performed on a 1.5T scanner (MAGNETOM Avantofit, Siemens) using a 2D GRE
sequence with a monopolar gradient and following imaging parameters: TE1
= 1.2 ms, echo spacing= 1.44ms, TR=200ms, echo train length=20, flip=25°,
104x128 matrix, slice thickness = 5mm. R2* was estimated using a
mono-exponential6 model and multi-spectral fat-water models: NLSQ (Non-linear
Least Squares)7 model from ISMRM Fat-Water Toolbox, and an ARMA
(Autoregressive Moving Average)8 model.
MATLAB
(R2020b) was used for drawing region-of-interests (ROI) and quantifying the mean
and standard deviation of R2* and FF values.Results and Discussion
All three iron nanoparticle sizes tended to cluster larger than
the manufacturer specifications, but the 130nm and 250nm nanoparticles clustered
less as seen on Figure 1. However, the 500nm iron clusters were visibly larger
and more condensed than 130nm and 250nm counterparts, and this clustering has
resulted in a heterogenous MRI signal in these phantoms.
As shown on Figure 2, increasing iron
concentrations from 0% to 0.32% showed an increase in R2* for all particle sizes
with iron-fat phantoms having a higher R2* than iron only phantoms on average.
However, the 500nm nanoparticles were excluded from Figure 2 as the high
inhomogeneity in the phantoms resulted in large standard deviations as evidenced
in the R2* maps on Figure 3. Additionally, the 500nm phantoms produced larger R2*
(32-2311s-1) than the values measured in the 250nm and 130nm iron-only
and iron-fat phantoms for the same solid content as shown in Figure 3. This may
be due to severe aggregation of the 500nm iron particles as clustering has been
shown to increase R2*9. For the iron phantoms, all three R2* fitting
models performed well; however, for the iron-fat phantoms, the NLSQ overestimated
R2* at high iron concentrations. Figure 4 shows iron-fat phantoms with true FF
values of 20%, but the difference in FF for some phantoms illustrates that higher
concentrations and larger diameters of iron nanoparticles can confound FF quantification
in the presence of fat, especially for the NLSQ model.
Limitations of this study include the
limited range of R2* values (~0-300s-1) for the 130nm and 250nm iron
phantoms despite matching the solid content concentrations with the 500nm iron particles.
Previous studies on smaller iron particles (80nm) have been shown to generate
clinical R2* values (0-1000 s-1)8,
so future work should amplify the concentrations of 130nm and 250nm iron phantoms
to cover the clinical R2* range.Conclusion
Our study
demonstrates that the size of iron nanoparticles, likely due to the degree of
clustering, has profound effects on the MRI signal as well as on the magnitude
of R2* values. Additionally, higher concentrations of iron and larger diameters
of nanoparticles have been shown to confound FF values.Acknowledgements
This
project was funded by the National Institutes of Health (NIH) grant
#1R21EB031298.
The
authors thank St. Jude Children’s Research Hospital for allowing the use of the
MRI scanner.
References
1. Sirlin CB, Reeder SB. Magnetic resonance
imaging quantification of liver iron. Magn Reson Imaging Clin N Am. 2010
Aug;18(3):359-81, ix. doi: 10.1016/j.mric.2010.08.014. PMID: 21094445; PMCID:
PMC3430384.
2. Henninger, B. Demystifying liver iron
concentration measurements with MRI. Eur Radiol 28, 2535–2536
(2018). https://doi.org/10.1007/s00330-017-5199-8
3. Wood JC, Fassler JD, Meade T. Mimicking liver
iron overload using liposomal ferritin preparations. Magn Reson Med. 2004
Mar;51(3):607-11. doi: 10.1002/mrm.10735. PMID: 15004804; PMCID: PMC2892965.
4. Ghugre NR. “Calibration of
iron-meditated MRI relaxation by Monte Carlo modeling”. 2008
5. Hines CD, Yu H, Shimakawa A, McKenzie CA,
Brittain JH, Reeder SB. T1 independent, T2* corrected MRI with accurate
spectral modeling for quantification of fat: validation in a fat-water-SPIO
phantom. J Magn Reson Imaging. 2009 Nov;30(5):1215-22. doi: 10.1002/jmri.21957.
PMID: 19856457; PMCID: PMC3205952.
6. Hankins, J. S., McCarville, M. B., Loeffler, R.
B., Smeltzer, M. P., Onciu, M., Hoffer, F. A., Li, C. S., Wang, W. C., Ware, R.
E., & Hillenbrand, C. M. (2009). R2* magnetic resonance imaging of the
liver in patients with iron overload. Blood, 113(20), 4853–4855. https://doi.org/10.1182/blood-2008-12-191643
7.
Hernando D, Liang ZP, Kellman P. Chemical
shift-based water/fat separation: a comparison of signal models. Magn Reson
Med. 2010 Sep;64(3):811-22. doi: 10.1002/mrm.22455. PMID: 20593375; PMCID:
PMC2992842.
8. Tipirneni-Sajja, A., Krafft, A. J., Loeffler, R.
B., Song, R., Bahrami, A., Hankins, J. S., & Hillenbrand, C. M. (2019).
Autoregressive moving average modeling for hepatic iron quantification in the
presence of fat. Journal of magnetic resonance imaging: JMRI, 50(5),
1620–1632. https://doi.org/10.1002/jmri.26682
9. Jeon M, Halbert MV, Stephen ZR, Zhang M. Iron
Oxide Nanoparticles as T1 Contrast Agents for Magnetic Resonance
Imaging: Fundamentals, Challenges, Applications, and Prospectives. Adv Mater.
2021 Jun;33(23):e1906539. doi: 10.1002/adma.201906539. Epub 2020 Jun 4. PMID:
32495404; PMCID: PMC8022883.