Keywords
liver - spleen - magnetic resonance imaging - T1rho - aging - menstrual cycle
Introduction
T1rho (T1ρ) has been shown to be sensitive to the interactions between water molecules
and macromolecules, including collagen. Allkemper et al.
[1] reported that the liver T1rho value was significantly associated with the Child-Pugh
staging of the patients. Hector et al.
[2] reported T1rho values were significantly higher in the cortex of fibrotic renal
allografts than in functional kidney allografts. The cortical T1rho value significantly
correlated with Masson’s trichrome-stained fractions. In another study, Hector et al.
[3] reported that spleen T1rho showed a significant correlation with portal pressure
and clinically significant portal hypertension. It was suggested that the elevation
in T1rho in the spleen with increased portal hypertension severity may reflect the
deposition of collagen associated with fibrogenesis.
In liver fibrosis evaluation, pre-clinical studies reported that liver T1rho quantification
allows excellent separation between normal liver and early-stage liver fibrosis [4]
[5]. However, the application of T1rho for real-world detection early-stage liver fibrosis
in human patients has been less successful [6]. Under experimental settings, the same batch of animals with the same age and animals
raised under the same environment with the same diet would show fewer inter-subject
variations. However, even among healthy human livers, liver physiology has variations
related to age and sex which in turn are affected by variations in iron/fat deposition
and perfusion changes [7]
[8]. Moreover, which factors contribute to liver T1rho variation has not been fully
elucidated [9]. In this study, we aim to evaluate potential age- and sex-related variations in
normative spleen T1rho values.
Materials and methods
This healthy volunteer study was conducted with the approval of the institutional
ethics committee, and informed consent was obtained. Two T1rho sequences were used,
with one based on gradient echo sequence (GRE) with multiple breath-hold acquisitions
[10], and the other was based on fast spin echo sequence (FSE) with a single breath-hold
acquisition [11].
For the FSE sequence, 96 healthy volunteers were recruited from the local community.
Single-breath-hold black blood FSE T1rho acquisition was conducted using a Philips
Achieva TX 3.0 T scanner equipped with a dual transmitter (Philips Healthcare, Best,
the Netherlands). The combination of double inversion recovery and 2D fast FSE was
used to achieve the black blood effect. The parameters for MR imaging included: TR/TE
2,000/20 msec, in-plane resolution 1.5 mm × 1.5 mm, slice thickness 7 mm, SENSE acceleration
factor 2, half scan factor (partial Fourier) 0.6, number of signal averaging 1, delay
time for SPAIR (SPectral Attenuated Inversion Recovery) 250 msec, delay time for double
inversion recovery 650 msec, spin-lock frequency 500Hz. T1rho images with six spin
lock times of 0, 10, 20, 25, 35, 50 msec were acquired, with a single breath-hold
of 12 seconds for data acquisition. Six axial slices were acquired for each examination.
For the GRE sequence, 28 health subjects were recruited. Multi-breath-hold GRE T1rho
acquisition without suppression of the blood signal was conducted using a 3T clinical
scanner (Achieva, Philips Healthcare, Best, The Netherlands). For T1rho measurement,
a rotary echo spin-lock pulse was implemented in a 2D fast GRE sequence. The spin-lock
frequency was set to 500 Hz and spin-lock times of 1 ms, 10 ms, 20 ms, 30 ms, 40 ms,
and 50 ms were used for T1rho mapping. TE and TR were 1.16 ms and 2.3 ms, respectively.
The voxel size was 1.50×1.50×7.00 mm3. The flip angle was 40 degrees, the number of signal averages was 2, and the SENSE
acceleration factor was 1.5. Six axial slices were acquired for each examination.
The data were acquired with one image of one spin-lock time during one breath-hold.
Both the FSE acquisition and the GRE acquisition were initially aimed to cover the
center of the liver, sufficient coverage of the spleen to allow T1rho measurement
of the spleen was available in 52 FSE cases (36/62 for females, 16/34 for males) and
14 GRE cases (6/7 for females, 8/8 for males), respectively. The summed spleen region
of interest (ROI) area was on average 4067 cm2/case (standard deviation: 4167 cm2). All images were processed using Matlab (Mathworks, Natick, MA, USA). T1rho maps
were computed using a mono-exponential decay model, as described by the following
equation: M (TSL) = A · exp (- TSL/T1 rho), where A is a constant scaling factor and TSL is the time of spin-lock. A non-linear least
square fit with the Levenberg-Marquardt algorithm was applied. Maps of coefficient
of determination (R2) were also generated for the evaluation of goodness of fit. Only T1rho values for
pixels associated with R2>0.80 were included in the subsequent ROI placement and T1rho analysis to eliminate
the unreliable poorly fitted T1rho values due to artifacts. The mean value within
the ROI for the spleen was obtained. Liver T1rho has been shown to have high scan-rescan
reproducibility with right scan setups [10]
[11]. Liver 1rho values have been previously reported [10]
[11], and we re-used the previously reported values for cases with spleen values analyzed
in the current study. For the spleen FSE T1rho, the scan-rescan reproducibility in
two separate sessions ICC (intraclass correlation coefficient) was 0.902 with the
9 cases (5 males, and 4 females) available for evaluation.
Statistical analysis was performed using GraphPad Prism (GraphPad Software, San Diego,
CA, USA). Comparisons between groups were tested by Mann–Whitney U test, Wilcoxon
signed rank test, or Wilcoxon matched-pairs signed rank test as appropriate. The significances
of correlation were tested with Pearson correlation. A p-value of less than 0.05 was
considered statistically significant.
Results
The study participants’ demographics and T1rho results are shown in [Table 1], and [Fig. 1], [Fig. 2], [Fig. 3], [Fig. 4], [Fig. 5].
Table 1 Physiological spleen T1rho sex- and age-related variations.
|
Subjects
|
Spleen T1rho ms§
|
Slope
|
r
|
Pslope
|
|
#: years in mean and range. §: median (range), 95% confidence internal (CI). slope: slope related to age, r: Pearson r
|
|
FSE imaging, both sexes, age#: 37.3 (18–80)
|
87.72 (67.14–105.1), 95%CI: 84.46–90.21
|
–0.3513
|
–0.4872
|
0.0002*
|
|
FSE imaging, females (n=36), age: 35.6(18–75)
|
92.78 (67.92–105.1), 95%CI: 88.08–93.92
|
–0.2675
|
–0.4317
|
0.0086*
|
|
FSE imaging, males (n=16), age: 41.1 (24–80)
|
77.26 (67.14–97.25), 95%CI: 74.22–83.99
|
–0.3396
|
–0.5515
|
0.0268*
|
|
GRE imaging, both sexes, age: 28.6 (20–44)
|
72.52 (57.48–91.87), 95%CI: 67.27–80.44
|
|
|
|
|
GER imaging, females (n=6), age: 27.3 (24–30)
|
70.12 (57.48–91.87), 95%CI: 58.30–83.91
|
|
|
|
|
GRE imaging, males (n=8), age: 29.5 (20–44)
|
75.04 (61.56–88.94), 95%CI: 66.61–85.23
|
|
|
|
Fig. 1
a–f An illustrative case of upper abdominal imaging shows T1rho images [spin lock time
(TSL), a: 0 ms, b: 10 ms, c: 20 ms, d: 25 ms, e: 35 ms, f: 50 ms] and T1rho map (g, without region of interest drawn, f: a region of interest drawn on the spleen). Note the higher signal of the spleen
relative to that of the liver.
Fig. 2
a–h Sex- and age-related variations of normal spleen and liver T1rho values. Data are
from the same group of subjects (n=36 for females and n=16 for males) with FSE acquisition.
In d and h, the bar denotes the median value.
Fig. 3
a–d Correlation of liver T1rho value and spleen T1rho value (a: FSE imaging, b: GRE imaging), and higher spleen T1rho for FSE values than for GRE T1rho values (c: females’ data, d: males’ data) when the ages of the study subjects were approximately similar.
Fig. 4
a–e Ratio of liver T1rho to spleen T1rho for males (M) and females (F). Fast spin echo
(FSE) in a, b, c, and d, both FSE and gradient echo (GRE) data in e. a: males’ and females’ data together. b: females’ data. c: males’ data. M: males, F: females. M, F in e: males and females.
Fig. 5
a–d Menstrual cycle-related variations in spleen T1rho and liver T1rho. a and b (bar denotes median values) are data used in this study’s analysis. c (mean ± standard deviation) included all the liver T1rho data from the earlier report
[11], with 26 women scanned in the non-menstrual phase (mean age: 29.4 years) and 11
women scanned in the menstrual phase (mean age: 25.4 years). d: Five female’s liver T1rho values were scanned twice, one being in the menstrual
phase and the other being in the second half of the non-menstrual phase (data from
the earlier report [11]).
For FSE data, an age-related decreasing trend of spleen T1rho was noted for both females
and males ([Fig. 2]). This trend was consistent with liver T1rho for females, while such a trend was
not apparent for males’ liver T1rho. Overall, females also had a higher T1rho than
males, both for the spleen (92.8 vs 77.3 ms, p<0.0001) and for the liver (44.2 vs.
38.9 ms, p<0.0001).
Spleen T1rho and liver T1rho were positively correlated, both for FSE data (r=0.611, p<0.0001) and GRE data (r=0.541, p=0.046, [Fig. 3]). When the ages of the study subjects are approximately similarly matched, spleen
T1rho was higher for FSE values than for GRE T1rho values, with a median of 93.2 ms
vs. 70.1 ms for females [n= 25 cases for FSE data (mean age: 26.4 yrs), and n=6 cases
for GRE data (mean age: 27.3 years), p=0.001] and a median of 83.2 ms vs. 75.1 ms
for males [n= 8 cases for FSE data (mean age: 33.5 yrs), and n=8 cases for GRE (mean
age: 29.5 yrs), p=0.44]. When grouping females’ and males’ data together, the spleen
FSE T1rho was 21.2% higher than the GRE value (89.5±10.2 ms vs. 73.9 ±11.4 ms, p<0.001).
When spleen T1rho was used to normalize liver T1rho, the ratio of T1rholiver/T1rhospleen largely removed the sex and age effect ([Fig. 4]). There was no statistically significant age-related trend for the T1rholiver/T1rhospleen ratio when females’ and males’ data were grouped together ([Fig. 4]a), and for the T1rholiver/T1rhospleen ratio of females’ data ([Fig. 4]b). The T1rholiver/T1rhospleen ratio of males’ data showed a weak increasing trend ([Fig. 4]c, p=0.032). There was also no difference in FSE ratio T1rholiver/T1rhospleen between females and males ([Fig. 4]d, GRE data not compared for males vs. females due to limited sample size and the
mean age was very young for GRE data). Ratio T1rholiver/T1rhospleen was larger for GRE data than for FSE data ([Fig. 4]e, p<0.0001).
Among pre-menopausal females, the menstrual cycle of 23 female participants was recorded.
18 cases were in the non-menstrual phase (mean age: 28.1 years) and 5 cases were in
the menstrual phase (mean age: 24.1 years). The spleen T1rho among women in the menstrual
phase was 10.7% shorter (95.9ms vs. 85.6ms, p=0.012) than that of women not in the non-menstrual phase ([Fig. 5]a), while the liver T1rho among women in the menstrual phase was 3.8% shorter than
that of women in the non-menstrual phase ([Fig. 5]b).
Discussion
The T1rho relaxation time is correlated to the T2 relaxation time. At 3.0 Tesla, the
T2 relaxation times of the liver and the spleen are estimated to be around 42 ms and
61 ms, respectively. In the current study, the spleen T1rho value (median= 89.1 ms
for females, median= 77.3 ms for males) was accordingly much higher than the liver
T1rho value (median=44.1 ms for females, median=38.9 ms for males, FSE data). These
results are approximately consistent with the earlier study of Hectors et al.
[2]
[3]. This study demonstrates that older subjects are associated with a shorter T1rho
value, and the same as the liver, male spleens have a shorter T1rho than female spleens.
Liver T1rho and spleen T1rho were positively correlated, and this correlation was
consistent with both the FSE sequence and the GRE sequence. More interestingly, when
the spleen T1rho was used to normalize the liver T1ho, then age- and sex-related effects
on liver T1rho were largely removed. This observation can potentially have important
clinical relevance. It can be anticipated that the ratio T1rholiver/T1rhospleen may offer better characterization of liver pathologies when the pathology has only
affected the liver while the spleen is normal. Though many systemic diseases have
both liver and spleen involvement [3], early pathologies tend to involve only the liver. One point of note is that, for
males, a statistically significant trend of reduction of spleen T1rho along with increasing
age was noted for the spleen (p=0.027) but not for the liver (p=0.47). This led to the ratio T1rholiver/T1rhospleen having an increasing trend following increased age. More studies are needed to clarify
this ratio trend. In the current study, the sample size for male livers was n=16.
In our earlier report, the sample size for male livers was n=34, and p was 0.617, still no age-related trend was noted for the liver T1rho [11].
The results of this study may also help to clarify the liver/spleen T1rho MR relaxometry
mechanism. An age-related increase in iron deposition in the spleen and in the liver
has been well documented [8]
[12]. Due to women’s menstruation and pregnancy, iron deposition in the liver and spleen
is lower in adult pre-menopausal women than in age-matched adult men. In women, iron
deposition in the liver and spleen substantially increases after menopause. Iron can
shorten T1rho relaxation time [13]. It is tempting to suggest that organ iron concentration differences may explain
why male spleens had a shorter T1rho than female spleens. However, we argued that
it is likely that the spleen age-related T1rho change is not dominantly caused by
spleen iron concentration variation [9]. In this study, the median spleen T1rho in women in the menstrual phase was 10.7%
lower than that of women in the non-menstrual phase (p=0.01), and the median liver
T1rho in women in the menstrual phase was 3.8% lower than that of women in the non-menstrual
phase in the current analysis (p>0.05). These two observations suggest that factors
affecting biochemical tissue composition detectable by T1rho may have a greater impact
on the spleen than on the liver ([Fig. 2]c, [Fig. 2]g, [Fig. 5]). [Fig. 5]c shows liver data from our earlier report including 26 females in the non-menstrual
phase and 11 females in the menstrual phase, with p=0.064, which is close to being statistically significant. [Fig. 5]d shows five women who underwent MRI both in the non-menstrual phase and the menstrual
phase, with four of them having a lower live T1rho in the menstrual phase. It has
been observed that women in the menstrual phase tend to have lower body iron [14]. Though, in theory, lower spleen iron will lead to a longer T1rho, in our studies,
both the liver T1rho and spleen T1rho were shorter, rather than longer, among women
in the menstrual phase than in women in the non-menstrual phase. Fat deposition in
the liver has been shown to shorten T1rho [5]. However, fatty spleen is uncommon under physiological conditions. Our previous
analysis suggests that 10% additional liver fat contributes to shortening of the liver
T1rho by 1.55 ms [5]. We do not expect the spleen T1rho variation to be largely affected by spleen fat
variations.
Among healthy young women, there are large variations in liver T1rho, for which we
do not yet have a satisfactory explanation [11], though we suspect this could be partially related to hormonal variations. It is
possible that there are sufficient age- and sex-related macromolecular biochemical
differences among healthy human subjects which contribute to the liver and spleen
T1rho variations. The liver undergoes “brown atrophy” with old age [15]. Mild fibrosis is a hallmark of the aging of various organs, which reflects increased
deposition of the extracellular matrix. Naturally, subclinical diseases in apparently
normal persons, smoking habits, nutritional differences, and other factors may also
contribute to the variability in the liver and spleen aging process.
This study also showed the ratio T1rholiver/T1rhospleen may vary according to different T1rho acquisition sequences and is higher with GRE
acquisition than with FSE acquisition. While further exploration of the NMR physics
mechanism is beyond the scope of this study, further inter-study comparison should
take into account this phenomenon.
There are many limitations to this study. As noted above, we could not explore the
biophysical mechanisms behind the observed variations. The image data were initially
acquired with a focus on the liver, this leads to only 52 cases with FSE T1rho acquisition
and 14 cases with GRE T1rho acquisition available for the analysis of the spleen.
However, due to the high stability of measurement, the observed trends mostly achieved
statistical significance for FSE T1rho data. Another limitation is that we do not
have data for pediatric populations. Finally, we can only speculate that the ratio
of liver T1rho to spleen T1rho can offer better characterization of liver pathologies
when the pathology has only affected the liver while the spleen is normal, and this
should be confirmed with further patient studies.
In conclusion, we describe sex- and aged-related spleen T1rho variation among healthy
women and men. The spleen T1rho value is approximately double the liver T1rho value.
This study demonstrates that older subjects are associated with a shorter spleen T1rho
value, and male spleens show a shorter T1rho than female spleens. Liver T1rho and
spleen T1rho are positively correlated. When spleen T1rho is used to normalize the
liver T1ho, then age and sex-related effects of liver T1rho are largely removed.
Clinical Relevance
-
Interpreting normal spleen T1rho should consider age and gender factors.
-
The ratio of liver T1rho to spleen T1rho can potentially offer better characterization
of liver pathologies when the pathology has only affected the liver while the spleen
is normal.
-
Spleen T1rho physiological variations may not be dominantly affected by tissue iron
content.
Data availability statement: the data that support the findings of this study are available from the corresponding
authors upon reasonable request.
