Association of MRI-derived Segmental Nonfunctional Liver Volume and Chronic Liver Disease

Abstract

Purpose

To determine whether the nonfunctional liver volume (NFLV) is an indicator of chronic liver disease (CLD).

Materials and Methods

Multiparametric 3T abdominal MRI examinations enhanced with gadobenate dimeglumine of 51 patients were included in the study and divided into two groups: patients with (n=20) and without (n=31) CLD. Pre- and postcontrast T1 relaxation times of the liver and aorta were measured in the T1 mapping sequences. Total and segmental liver volumes (Lvol) were determined using a convolutional neuronal network. The functional liver fraction (FLF) defined as [(1/T1_{liver postcontrast} − 1/T1_{liver precontrast}) ÷ (1/T1_{blood pool postcontrast} − 1/T1_{blood pool precontrast})] × (1 − hematocrit) and the nonfunctional liver volume (NFLV) defined as (1 − FLF) × Lvol were calculated for the whole liver, segments I–III, and IV–VIII. Volumes, FLF, and NFLV were compared between the groups using the Mann-Whitney U test and receiver operation characteristics (ROC) analysis.

Results

Volumes were significantly higher in patients with CLD than without CLD for the whole liver (p<.01), segments I–III (p<.001), and segments IV–VIII (p<.01). No significant difference was found regarding FLF (p=.20–31). NFLV of the whole liver (p<.01), segments I–III (p<.001), and IV–VIII (p<.01) were significantly increased in patients with CLD. The highest AUCs were observed for Lvol (AUC=.80; p<.001) and NFLV (AUC=.78; p<.001), both in segments I–III. The optimal NFLV cutoff values for CLD were 745 ml for the whole liver (77 % sensitivity; 75% specificity), 174 ml for segments I–III (85% sensitivity; 70% specificity), and 573 ml for segments IV–VIII (77% sensitivity; 75% specificity).

Conclusion

MRI-derived nonfunctional liver volume (NFLV) is helpful for early detection of imaging changes in CLD. NFLV is highly associated with CLD, notably when measured in the liver segments I–III.

Key Points

• MRI-derived NFLV may be useful for early detection of CLD.

• NFLV is significantly higher in patients with CLD than those without.

• The best AUC was obtained when NFLV was calculated for segments I–III.

Citation Format

• Ardoino M, Zbinden L, Klaus JB et al. Association of MRI-derived Segmental Nonfunctional Liver Volume and Chronic Liver Disease. Rofo 2025; DOI 10.1055/a-2514-4596

 

MissingLinkText

Zusammenfassung

Ziel

Überprüfen, ob das nicht-funktionelle Lebervolumen (NFLV) ein Indikator für chronische Lebererkrankung (CLD) ist.

Material und Methoden

In die Studie wurden 51 multiparametrische 3T-MRT-Untersuchungen des Abdomens mit Gadobenatdimeglumin eingeschlossen und in zwei Gruppen aufgeteilt: Patienten mit (n=20) und ohne (n=31) CLD. Die T1-Relaxationszeiten von der Leber und Aorta vor und nach Kontrastmittelgabe wurden auf den T1 mapping sequences gemessen. Das Gesamtvolumen der Leber und die segmentalen Lebervolumina (Lvol) wurden mit einem Convolutional Neuronal Network bestimmt. Der funktionelle Leberanteil (FLF), definiert als [(1/T1_{Leber postkontrast} − 1/T1_{Leber präkontrast}) ÷ (1/T1_{Blutpool postkontrast} − 1/T1_{Blutpool präkontrast)}] × (1 − Hämatokrit) und das nichtfunktionelle Lebervolumen (NFLV), definiert als (1 − FLF) × Lvol, wurden für die gesamte Leber, Segmente I–III und IV–VIII berechnet. Lebervolumen, FLF und NFLV wurden mittels Mann-Whitney-U-Test und der ROC-Analyse (Receiver Operation Characteristics) zwischen den Gruppen verglichen.

Ergebnisse

Das Volumen der Gesamtleber (p<.01), der Segmente I–III (p<.001) und der Segmente IV–VIII (p<.01) waren bei Patienten mit CLD signifikant höher als bei Patienten ohne CLD. Für die FLF wurde kein signifikanter Unterschied gefunden (p=.20–31). Die NFLV der gesamten Leber (p<.01), der Segmente I–III (p<.001) und IV–VIII (p<.01) waren bei Patienten mit CLD signifikant höher als bei Patienten ohne CLD. Die höchste AUC-Werte ergaben sich für Lvol (AUC=.80; p<.001) und NFLV (AUC=.78; p<.001), beide in den Segmenten I–III. Die optimalen Cutoff-Werte der NFLV für CLD waren 745 ml für die gesamte Leber (77% Sensitivität; 75% Spezifität), 174 ml für die Segmente I–III (85% Sensitivität; 70% Spezifität) und 573 ml für die Segmente IV–VIII (77% Sensitivität; 75% Spezifität).

Schlussforderung

Das mittels MRT bestimmte, nicht-funktionelle Lebervolumen (NFLV) ist für die frühzeitige Erkennung von bildgebenden Veränderungen der CLD geeignet. Das NFLV ist in hohem Maße mit CLD assoziiert, vor allem wenn es in den Lebersegmenten I–III gemessen wird.

Kernaussagen

• Das mittels MRT bestimmte NFLV ist für die CLD-Früherkennung nützlich.

• Das NFLV ist bei Patienten mit CLD signifikant höher als bei Patienten ohne CLD.

• Die beste AUC wurde mit der NFLV der Segmente I–III erzielt.

 

MissingLinkText

Keywords

MRI - T1 mapping - liver fibrosis - chronic liver disease - nonfunctional liver volume

Introduction

Chronic liver disease (CLD) affects approximately 1.5 billion people worldwide [1]. Complications of CLD such as liver cirrhosis associated adverse outcomes and hepatocellular carcinoma are responsible for 3.5% of the deaths worldwide [2]. Because the early stages of CLD can be reversible, it is crucial to have a method that can detect the disease before it becomes permanent and complications develop [3].

The current reference standard for CLD diagnosis is liver biopsy [4]. However, liver biopsy is not suitable for early detection due to its invasiveness [5]. Non-invasive diagnostic approaches include serum biological markers, ultrasound (US) elastography, magnetic resonance elastography (MRE), and morphological assessment of the liver [3]. Biological markers can be easily measured in clinical practice by a laboratory test [6], but are not specific to hepatic fibrosis [3]. US elastography is widely available, well reproducible, and allows an accurate diagnosis of liver cirrhosis [7]. However, the diagnostic performance is not optimal for early fibrosis, and there is an overlap between adjacent stages and some methodological limitations [7]. MRE is more accurate and reliable than US elastography [7]. Still, it is a method that is not widely available, as it requires special and expensive hardware [6].

Changes in liver morphology can be evaluated by ultrasonography (US), computer tomography (CT), or magnetic resonance imaging (MRI), widely available modalities [3]. Typical morphological features of liver cirrhosis encompass nodularity of the hepatic surface, right hepatic posterior notch, narrowing of the hepatic veins, enlargement of the gallbladder fossa, and expansion of the hilar periportal space [3]. Liver fibrosis also causes a volume increase in the hepatic left lateral section and caudate lobe (segments I–III) and decrease in the right and medial left lobe (segments IV–VIII) [8]. Various measures, such as the liver segmental volume ratio (LSVR) and caudate to right lobe ratio (CRL-R), have been introduced to quantify the volume changes associated with fibrosis [9] [10] [11] [12].

In addition, parameters capable of estimating liver tissue composition were identified: T1 relaxation time has been shown to be increased in liver fibrosis [4] [13] [14] [15] [16] [17] [18] [19], steatosis, and inflammation [13] [19], and decreased in case of iron overload [13], while T2 relaxation time rather correlates with inflammation [20] [21]. It has also been proven that the extracellular volume (ECV), calculated using the T1 relaxation time of the liver and the blood pool before and after injection of a contrast agent that only distributes in the extracellular compartment of the liver, is increased in case of fibrosis [4] [14] [16] [22], and is independent of the MRI field strength, in contrast to T1 and T2 relaxation times [14].

We assume that the use of a contrast medium that enters the hepatocytes artificially increases the ECV and could be used to estimate the hepatic function. For this reason, we defined the functional liver fraction (FLF) for contrast media that enters the hepatocytes. We hypothesize that the combination of volumetry and FLF could be a marker of liver fibrosis. Therefore, we created a new parameter, the NFLV, calculated as (1 – FLF) × Lvol. The aim of this study was to investigate whether FLF and NFLV are associated with CLD.

 

MissingLinkText

Material and Methods

Study Population

This retrospective study was approved by the Institutional Review Board (Cantonal ethics committee, Bern, Switzerland, Project ID 2019-01333). All patients involved provided written informed consent.

Our Radiological Information System (RIS) was screened for multiparametric 3T abdominal MRI enhanced with gadobenate dimeglumine (Gd-BOPTA, Multihance, Bracco Imaging, Milan, Italy) and performed between September 4, 2018 and May 20, 2020. Out of 532 exams, 382 were excluded because they did not feature a complete parametric T1 mapping before and 20 minutes after intravenous contrast injection. In addition, 19 scans were excluded due to lack of hematocrit value, 11 because they were repeated examinations performed on the same patient, and 66 due to conditions, other than liver fibrosis, influencing the T1 relaxation time or the liver volume such as: pathologies of the biliary tract (n=32); large or multiple focal liver lesions (n=11); status post liver transplantation (n=10); status post liver ablation surgery (n=8); iron overload (n=3); portal vein thrombosis (n=1); and cerebral death with unclear hepatopathy (n=1). Finally, three scans were excluded due to significant image artifacts. The final study population consisted of 51 scans.

The patients were divided into two groups depending on whether they had a diagnosis of chronic liver disease in their medical history (CLD group), defined as a disease of the liver leading to a progressive deterioration of the liver function lasting over six months [23], or not (noCLD group). The CLD group was composed of 20 patients, whereas the noCLD group consisted of 31 patients. [Fig. 1] represents a flowchart of the selection of the study population and the repartition of the patients in the groups.

Fig. 1 Flowchart of the selection of the study population and the repartition of the patients in the two groups.

Clinical data

All information regarding the patient’s clinical status was retrieved from their medical electronic records. The following parameters were gathered: CLD etiology, sex, age, size, weight, body mass index (BMI), hypertension status, diabetes status, alcohol consumption, creatinine level, albumin level, bilirubin level, aspartate aminotransferase (AST) level, alanine aminotransferase (ALT) level, alkaline phosphatase (ALP) level, gamma-glutamyl transferase (GGT) level, thrombocytes count, hematocrit, Quick, Fibrosis-4 Index (FIB-4) and AST to Platelet Ratio Index (APRI). The Child-Pugh score was also calculated for all patients with chronic liver disease. Laboratory values from 3 months before or after the MRI were considered to reflect the patient’s state of health at the time of the examination.

 

MissingLinkText

 

MissingLinkText

Image acquisition

All images were acquired on 3T MRI scanners (Magnetom Skyra or Prisma, Siemens Healthineers, Erlangen, Germany). The images were obtained following a standard liver imaging protocol including morphological T1w, T2w, diffusion-weighted images, and VIBE Dixon sequences, as well as parametric T1 mapping before and 20 minutes after intravenous injection of 0.1 mmol/kg bodyweight of gadobenate dimeglumine (Gd-BOPTA, Multihance, Bracco Imaging, Milan, Italy).

The T1 mapping was obtained using a modified look-locker inversion recovery (MOLLI) variant sequence with a 3–3-5 design and consisted of four single breath-hold (11 s) axial slices placed over the liver with a spacing of 300% (24 mm). The following specifications were used: echo time (TE) of 1.01 ms, repetition time (TR) of 740 ms, inversion time (TI) of 225 ms, flip angle (FA) 35°, slice thickness of 8 mm, field-of-view (FOV) of 306 × 360 mm, and matrix of 154 × 192 pixels.

The native T1-VIBE Dixon in phase acquisition was used as sequence for the liver segmentation and consisted of 60 to 80 axial slices of 3 mm thickness with axial dimensions between 210 × 320 and 270 × 320 pixels, and pixel spacing between 1.09375 × 1.09375 and 1.5625 × 1.5625 mm².

The LiverLab (Siemens Healthineers, Erlangen, Germany) with a multi-echo Dixon VIBE sequence resulting in a fat fraction map, a R2* map and a Goodness of Fit map was also acquired before contrast injection und used for the proton density fat fraction (PDFF) measurement.

 

MissingLinkText

Images analysis

The image analysis was performed by a trained medical student and validated by a board-certified radiologist (10 years of experience in hepatobiliary imaging), using a picture archiving and communication system (Sectra IDS7, Sectra, Linköping, Sweden).

If the Liverlab image quality was good enough (Goodness of fit <50‰), the PDFF was measured directly by drawing circular regions of interest (ROIs) on the fat fraction screening map of the Liverlab. In case of poor image quality, PDFF was calculated using the T1w VIBE Dixon sequences.

Pre- and postcontrast T1 relaxation times of the liver were measured before and after contrast injection by delineating and averaging nine polygonal regions of interest (ROIs) corresponding to the hepatic segments in a slice above and a slice under the portal vein bifurcation, avoiding focal liver lesions, vessels, and artifacts. ROIs were drawn with at least 1 cm separation from the edges of the liver to prevent partial volume effects. Additionally, efforts were made to place the ROIs consistently across different maps. T1 relaxation time of the aorta was obtained by placing a circular ROI in the center of the aorta on the four slices of the T1 mapping and averaging the values. Elements that could distort the measurement, such as image artifacts and calcifications, were avoided. [Fig. 2] shows an example of the T1 relaxation time measurements.

Fig. 2 Example of the T1 relaxation time measurements of the liver and aorta. The measurements were performed on the liver of a 22-year-old female patient without CLD and with a hematocrit of 0.36. a Precontrast T1map, slice above the portal vein bifurcation. b Precontrast T1map, slice under the portal vein bifurcation. c Postcontrast T1map, slice above the portal vein bifurcation. d Postcontrast T1map, slice under the portal vein bifurcation. Total precontrast T1 relaxation time of the liver: 808ms. Precontrast T1 relaxation time of the liver segments I–III: 810ms. Precontrast T1 relaxation time of the liver segments IV–VIII: 807ms. Precontrast T1 relaxation time of the aorta: 1993ms. Total postcontrast T1 relaxation time of the liver: 330ms. Postcontrast T1 relaxation time of the liver segments I–III: 328ms. Postcontrast T1 relaxation time of the liver segments IV–VIII: 332ms. Mean post-contrast T1 relaxation time of the aorta: 381ms.

Total and segmental liver volume (Lvol) were determined with the help of a convolutional neural network with a U-net architecture (nnU-Net) trained to delineate the segments of the liver and its vessels on the native T1 Dixon sequence [24] [25]. The segmentations performed by the artificial intelligence (AI) model were then imported into a 3D segmentation software (ITK-SNAP) and were visually checked for inaccuracies. If inaccurate, the liver segmentation was manually adjusted. The software then calculated the liver volumes by multiplying the number of voxels by the sequence-specific volume of one voxel. [Fig. 3] shows an example of 3D segmentation of the hepatic segments and vessels delineated by the AI model.

Fig. 3 Segmentation of the hepatic segments and vessels delineated by the convolutional neuronal network. The 3D segmentation was performed on the same 22-year-old female patient without CLD as in Figure 2. The bile ducts are represented in green and the hepatic veins in dark blue. a Segmentation in the coronal plane. b Segmentation in the axial plane. c Segmentation in the sagittal plane. d 3D visualization. Total predicted Lvol: 1286ml. Predicted Lvol of the segments I–III: 269ml. Predicted Lvol of the segments IV–VIII: 1017ml. LSVR: 0.26. Total FLF: 0.54. FLF of the segments I–III: 0.55. FLF of the segments IV–VIII: 0.54. Total NFLV: 592ml. NFLV of the segments I–III: 122ml. NFLV of the segments IV–VIII: 472ml.

Liver segmental volume ratio (LSVR) was calculated using the formula available in the literature [9]:

Functional liver fraction (FLF) was calculated using the following formula [14] for the whole liver, segments I–III, and IV–VIII:

Finally, nonfunctional liver Volume (NFLV) was calculated as Lvol × (1 – FLF) for the whole liver, segments I–III, and IV–VII.

 

MissingLinkText

Statistical analysis

All statistical analyses were performed using GraphPad Prism software (GraphPad Software, San Diego, USA). To compare the characteristics of the groups, the Mann-Whitney U test was used for the continuous variables and Fisher’s exact test for the discrete variables. Mann-Whitney U tests and ROC analyses were performed to compare the T1 relaxation time, Lvol, LSVR, FLF, and NFLF between the groups. The optimal cutoff values were determined using Youden’s index. The significance level was set as α = 5%.

 

MissingLinkText

 

MissingLinkText

Results

Characteristics of the study population

The study population consisted of 51 patients (25 men and 26 women) of which 20 had CLD and 31 had no CLD. The mean age was 55 years with a standard deviation (SD) of 15 years. The etiologies of the liver disease in the CLD group were steatosis (defined as PDFF>10%) (n = 6), metabolic dysfunction-associated steatotic liver disease (MASLD) (n = 6), alcoholic fatty liver disease (AFLD) or alcoholic steatohepatitis (ASH) (n = 2), drug induced liver injury (DILI) (n = 2), hepatitis B (n = 1), and hepatopathy of unknown etiology (n = 3). The distribution of CLD patients in Child-Pugh classes was as follows: 18 Child-Pugh A (Child-Pugh score 5–6), 1 Child-Pugh B (Child-Pugh score 7–9), and 1 Child-Pugh C (Child-Pugh score 10–15). [Table 1] shows further characteristics of the study population.

Table 1 Characteristics of the groups. Discrete values are written as the number of patients (n) with the percentage in parentheses and continuous data as mean ± standard deviation (SD). Data are rounded to the nearest unit except for APRI and FIB-4 index, which are rounded to the nearest tenth. A Mann-Whitney U test was used to calculate the P-values for the continuous variables and Fisher’s exact test was used to calculate the P-values for the discrete values. Non-significant P-values are rounded to the nearest hundredth, and significant P-values are written as <0.05, <0.01, or <0.001.

	noCLD (n = 31)	CLD (n = 20)	P-value
BMI: body mass index; PDFF: proton density fat fraction; AST: aspartate aminotransferase; ALT: alanine aminotransferase; ALP: alkaline phosphatase; GGT: gamma-glutamyl transferase; APRI: aspartate aminotransferase-to-platelet ratio index; FIB-4: Fibrosis-4 Index.
Male (n, %)	11 (35%)	14 (70%)	P < 0.05
Diabetes (n, %)	0 (0%)	3 (15%)	P = 0.05
Hypertension (n, %)	12 (39%)	10 (50%)	P = 0.56
Alcohol consumption (n, %)	7 (23%)	11 (55%)	P = 0.03
Age [years]	55 ± 15	55 ± 15	P = 0.94
Size [cm]	170 ± 9	171 ± 8	P = 0.82
Weight [kg]	75 ± 19	82 ± 19	P = 0.23
BMI [kg/m2]	25 ± 5	28 ± 7	P = 0.13
PDDF [%]	3 ± 4	12 ± 12	P < 0.01
Creatine [µmol/L]	70 ± 14	74 ± 15	P = 0.36
Albumin [g/L]	36 ± 6	37 ± 6	P = 0.35
Bilirubin [µmol/L]	9 ± 4	37 ± 112	P = 0.10
AST [U/L]	27 ± 17	69 ± 86	P < 0.001
ALT [U/L]	45 ± 102	100 ± 165	P < 0.001
ALP [U/L]	75 ± 56	130 ± 167	P < 0.05
GGT [U/L]	65 ± 163	162 ± 227	P < 0.001
Thrombocytes [G/L]	249 ± 80	243 ± 78	P = 0.86
Hematocrit [%]	38 ± 5	41 ± 4	P < 0.05
Quick [%]	97 ± 13	96 ± 21	P = 0.68
FIB-4	1 ± 0.4	1.7 ± 1.1	P < 0.05
APRI	0.3 ± 0.2	0.7 ± 0.6	P < 0.001

 

MissingLinkText

Descriptive statistics

Native T1 relaxation time was significantly higher in patients with CLD for the whole liver (897ms [832, 976ms] vs. 818ms [762, 866ms]; p<0.01), for segments I–III (902ms [828, 987ms vs. 814ms [773, 879ms]; p<0.01), and for segments IV–VIII (893ms [835, 975ms] vs 825ms [752, 867ms]; p>.01).

Total Lvol was significantly higher in patients with CLD than in patients without CLD (1532ml [1313, 2015ml] vs. 1280ml [1074, 1523ml]; p<0.01). Lvol of the segments I–III was also significantly higher in patients with CLD than in patients without CLD (356ml [288, 422ml] vs. 255ml [216, 297ml]; p<0.001), as well as Lvol of the segments IV–VIII (1167ml [993, 1546ml] vs. 997 [849, 1259ml]; p<0.01. LSVR was not significantly different in patients with CLD than without (0.27 [0.23, 0.37) vs. 0.25 [0.22, 0.29]; p=0.25).

There was no significant difference in total FLF between the groups (0.45 [0.38, 0.53] vs. 0.49 [0.41, 0.55]; p=0.23). The same applied for FLF of segments I–III (0.47 [0.38, 0.54] vs. 0.49 [0.41, 0.56]; p=0.31) and segments IV–VIII (0.44 [0.37, 0.52] vs. 0.48 [0.41, 0.55]; p=0.20).

Total NFLV was increased in patients with CLD compared to patients without CLD (894ml [691, 1035ml] vs. 643ml [543, 733ml]; p<0.01). NFLV of segments I–III (185ml [157, 240ml] vs. 125ml [101, 162ml]; p<0.001) and segments IV–VIII (658ml [556, 824ml] vs. 899ml [442, 572ml]; p<0.01) were also significantly different between the groups. The descriptive statistics are available in [Table 2].

Table 2 Descriptive statistics of the total and segmental Lvol, FLF, NFLV, and LSVR. Data are written as median [Q25; Q75]. Non-significant P-values are rounded to the nearest hundredth, and significant P-values are written as <0.05, <0.01, or <0.001. The Lvol predicted by the AI model was used without adjustment for 17 patients from the group with CLD and 30 patients from the group without CLD. The segmentation was manually adjusted by 3 patients from the group with CLD and 1 patient from the group without CLD.

	noCLD (n = 31)	CLD (n = 20)	P-value
Native T1 relaxation time
Total	818ms [762, 866ms]	897ms [832, 976ms]	P < 0.01
Segments I–III	814ms [773, 879ms]	902ms [828, 987ms]	P < 0.01
Segments IV–VIII	825ms [752, 867ms]	893ms [835, 975ms]	P < 0.01
Postcontrast T1 relaxation time
Total	346ms [327, 375ms]	352ms [314, 393ms]	P = 0.98
Segments I–III	346ms [324, 374ms]	347ms [288 387ms]	P = 0.66
Segments IV–VIII	346ms [324, 373ms]	358ms [315, 396ms]	P = 0.73
Liver volume (Lvol)
Total	1280ml [1074, 1523ml]	1532ml [1313, 2015ml]	P < 0.01
Segments I–III	255ml [216, 297ml]	356ml [288, 422ml]	P < 0.001
Segments IV–VIII	997 [849, 1259ml]	1167ml [993, 1546ml]	P < 0.01
Liver segmental volume ratio (LSVR)	0.25 [0.22, 0.29]	0.27 [0.23, 0.37]	P = 0.25
Functional liver fraction (FLF)
Total	0.50 [0.41, 0.55]	0.45 [0.38, 0.53]	P = 0.23
Segments I–III	0.49 [0.41, 0.56]	0.47 [0.38, 0.54]	P = 0.32
Segments IV–VIII	0.49 [0.41, 0.55]	0.44 [0.37, 0.52]	P = 0.20
1 – functional liver fraction (1 – FLF)
Total	0.50 [0.45, 0.59]	0.55 [0.47, 0.62]	P = 0.23
Segment I–III	0.51 [0.44, 0.59]	0.53 [0.46, 0.62]	P = 0.32
Segments IV–VIII	0.51 [0.46, 0.59]	0.56 [0.48, 0.63]	P = 0.20
Non-functional liver volume (NFLV)
Total	643ml [543, 733ml]	894ml [692, 1035ml]	P < 0.01
Segments I–III	125ml [101, 162ml]	185ml [157, 240ml]	P < 0.001
Segments IV–VIII	499ml [442; 572ml]	658 [556; 824ml]	P < 0.01

 

MissingLinkText

ROC analysis

The AUC of native T1 relaxation time was 0.75 (p<0.01) for the whole liver and 0.74 (p<0.01) for the native T1 relaxation time of segments I–III, as well as IV–VIII. ROC analysis depicted an AUC=0.75 (p<0.01) for total Lvol, an AUC=0.80 (p<0.001) for Lvol of the segments I–III, an AUC=0.73 (p <0.01) for Lvol of the segments IV–VIII, and an AUC=0.60 (p=0.24) for LSVR. The AUC was 0.60 (p=0.23) for total FLF. NFLV allowed differentiating between CLD and no CLD with an AUC=0.77 (p<0.01) when the total NFLV was measured and even a higher AUC=0.78 (p<0.001), when only the NFLV of the segments I–III was measured. AUC was only 0.75 (p<0.01) when the NFLV of only the segments IV–VIII was measured. The optimal cutoff value was 745 ml for total NFLV and 573 ml for NFLV of segments IV–VII, both having a sensitivity of 77% and a specificity of 75%. A sensitivity of 87% and a specificity of 70% were reached by a cutoff of 174ml for NFLV of segment I–III. [Fig. 4] shows the ROC curves and [Table 3] resumes the AUC and P-values.

Fig. 4 ROC curves of native T1 relaxation time, Lvol, LSVR, FLF, and NFLV for discriminating patients with and without CLD. The figure presents the native T1 relaxation time, Lvol, FLF, and NFLV results for the whole liver (green), segments I–III only (red), and segments IV–VIII only (blue).

Table 3 Results of the ROC analysis of native T1 relaxation time, Lvol, LSVR, FLF, and NFLV for discriminating patients with and without CLD. AUC is rounded to the nearest hundredth. Non-significant P-values are rounded to the nearest hundredth, and significant P-values are written as <0.05, <0.01, or <0.001. The optimal cutoff value was determined using Youden’s index.

	AUC	P-value	Youden’s index	Cutoff	Sensitivity	Specificity
Native T1 relaxation time
Total	0.75	p < 0.01	0.47	867 ms	60%	87%
Segments I–III	0.74	p < 0.01	0.45	891 ms	55%	90%
Segments IV–VIII	0.74	p < 0.01	0.47	879 ms	60%	87%
Liver volume (Lvol)
Total	0.75	p < 0.01	0.39	1642 ml	94%	45%
Segments I–III	0.80	p < 0.001	0.55	334 ml	90%	65%
Segments IV–VIII	0.73	p < 0.01	0.38	1079 ml	68%	70%
Liver segmental volume ratio (LSVR)	0.60	p = 0.24	0.24	0.30	84%	40%
Functional liver fraction (FLF)
Total	0.60	p = 0.23	0.23	0.46	68%	55%
Segments I–III	0.58	p = 0.31	0.18	0.51	48%	70%
Segments IV–VIII	0.61	p = 0.20	0.23	0.48	58%	65%
Non-functional liver volume (NFLV)
Total	0.77	p < 0.01	0.52	745 ml	77%	75%
Segments I–III	0.78	p < 0.001	0.57	174 ml	87%	70%
Segments IV–VIII	0.75	p < 0.01	0.52	573 ml	77%	75%

 

MissingLinkText

 

MissingLinkText

Discussion

This study shows that our introduced hepatic fibrosis biomarker, the nonfunctional liver volume (NFLV) is increased in patients with CLD, especially when measured in segments I–III. Since NFLV is non-invasive and does not require specific hardware, it could be relatively easily incorporated in routine MRI examination evaluations without unreasonably extending the examination time. Furthermore, besides time-consuming manual segmentation, a delineation of the segments and vessels of the liver can be achieved in less than one minute by a convolutional neural network [24].

FLF and NFLV are based on the property of Gd-BOPTA to enter the hepatocytes. Although not statistically significant, our findings demonstrate a reduction in contrast medium uptake in CLD patients, consistent with our initial assumption. Further corroborating this hypothesis, the NFLV was significantly higher in the CLD group compared to the noCLD group and its AUC is higher than the one of the Lvol for the whole liver and segments IV–VIII. Another commonly used contrast medium in clinical practice that also enters hepatocytes is gadoxetic acid (Gd-EOB-DTPA) [26]. Both contrast agents distribute into the extracellular space and are then taken up by hepatocytes to be excreted into the biliary tract at 5% for Gd-BOPTA and 50% for Gd-EOB-DTPA [26]. Due to this property, Gd-EOB-DTPA is presumably more suitable for assessing liver function than Gd-BOPTA, and it would be interesting to use it to calculate the FLF and NFLV.

The result concerning the liver volume showed an increase of the total Lvol in patients with CLD compared with patients without CLD. However, several studies have shown that the total liver volume does not adequately estimate liver fibrosis [12] [27]. For example, Liu et al. found no significant difference in liver volume between the different stages of liver fibrosis although it tended to increase form stage 1 to stage 3 and to decrease in stage 4 [27]. Nevertheless, our results are corroborated by Tarao et al. in a study showing an increase in liver volume between patients with alcoholic steatosis compared to healthy volunteers [28]. Furthermore, in a study by Kromrey et al., the liver volume estimated by ultrasonography was shown to be affected by parenchymal liver disease, with hepatic steatosis increasing the volume more than fibrosis or cirrhosis [29]. Because our CLD-group includes a majority of patients with steatosis, AFLD/ASH or MASLD, and the mean PDFF was significantly higher than in the noCLD group, one can postulate that the increase in Lvol is due to hepatocytes ballooning [28]. BMI, size, weight, gender, and age also influence the liver volume [29] [30]. Our CLD group includes a higher proportion of male patients than the noCLD group; however, the BMI, size, weight, and age are not significantly different between the groups. It is therefore unlikely that these parameters have influenced the Lvol.

Despite the deep learning algorithm used to carry out the liver segmentation and delineation of the vessel demonstrating high accuracy and strong correlation with manual calculation [25], our LSVR results deviate from the literature [9] [10] [12]. This study revealed a significant augmentation of volume in segments I–III and IV–VIII in patients with CLD compared to patients without CLD, and no significant increase of LSVR, whereas fibrosis typically leads to a volume decrease in volume in segments IV–VIII, an increase in segments I–III, and consequently an augmentation of LSVR [10].

The limitations of this study include its retrospective design and the lack of use of liver biopsy as the pathological reference standard. Furthermore, in the CLD group encompassed many different etiologies. Patients were classified into the groups according to the information present in the medical reports. For many of them, the diagnosis of CLD was made without performing a biopsy. Another possible point of criticism is the small group size, the relative underrepresentation of patients with CLD and the predominance of males within the CLD group. Furthermore, the CLD group predominantly consisted of patients classified as Child-Pugh A. This lack of diversity within the cohort restricted the possibility of conducting meaningful stratified analyses. Further validation of our results is now necessary on a larger patient cohort with Child-Pugh stratification and histologically graded CLD.

 

MissingLinkText

Conclusion

This study suggests that MRI-derived nonfunctional liver volume (NFLV) is conducive to early detection of imaging changes in CLD. NFLV is highly associated with CLD, notably when measured in the liver segments I–III.

Clinical relevance

• This study suggests that MRI-derived NFLV is conducive to early detection of imaging changes in CLD.

• Since it does not require specific hardware, the measurement of this parameter could be relatively easily incorporated in routine MRI examination evaluations without unreasonably extending the examination time.

 

MissingLinkText

 

MissingLinkText

 

MissingLinkText

Conflict of Interest

The authors declare that they have no conflict of interest.

• References

• 1 Moon AM, Singal AG, Tapper EB. Contemporary Epidemiology of Chronic Liver Disease and Cirrhosis. Clinical Gastroenterology and Hepatology 2020; 18: 2650-2666
• 2 Asrani SK, Devarbhavi H, Eaton J. et al. Burden of liver diseases in the world. Journal of Hepatology 2019; 70: 151-171
• 3 Horowitz JM, Venkatesh SK, Ehman RL. et al. Evaluation of hepatic fibrosis: A review from the society of abdominal radiology disease focus panel. Abdom Radiol 2017; 42: 2037-2053
• 4 Kupczyk PA, Mesropyan N, Isaak A. et al. Quantitative MRI of the liver: Evaluation of extracellular volume fraction and other quantitative parameters in comparison to MR elastography for the assessment of hepatopathy. Magn Reson Imaging 2021; 77: 7-13
• 5 Almpanis Z. Evaluation of liver fibrosis: “Something old, something new…”. Ann Gastroenterol 2016; 29: 445-453
• 6 Lurie Y, Webb M, Cytter-Kuint R. et al. Non-invasive diagnosis of liver fibrosis and cirrhosis. World J Gastroenterol 2015; 21: 11567
• 7 Zhang YN, Fowler KJ, Ozturk A. et al. Liver fibrosis imaging: A clinical review of ultrasound and magnetic resonance elastography. J Magn Reson Imaging 2020; 51: 25-42
• 8 Lubner MG, Pickhardt PJ. Multidetector Computed Tomography for Retrospective, Noninvasive Staging of Liver Fibrosis. Gastroenterol Clin North Am 2018; 47: 569-584
• 9 Obmann VC, Marx C, Hrycyk J. et al. Liver segmental volume and attenuation ratio (LSVAR) on portal venous CT scans improves the detection of clinically significant liver fibrosis compared to liver segmental volume ratio (LSVR). Abdom Radiol 2021; 46: 1912-1921
• 10 Furusato Hunt OM, Lubner MG, Ziemlewicz TJ. et al. The Liver Segmental Volume Ratio for Noninvasive Detection of Cirrhosis: Comparison With Established Linear and Volumetric Measures. J Comput Assist Tomogr 2016; 40: 478-484
• 11 Awaya H, Mitchell DG, Kamishima T. et al. Cirrhosis: Modified Caudate-Right Lobe Ratio. Radiology 2002; 224: 769-774
• 12 Pickhardt PJ, Malecki K, Hunt OF. et al. Hepatosplenic volumetric assessment at MDCT for staging liver fibrosis. Eur Radiol 2017; 27: 3060-3068
• 13 Hoad CL, Palaniyappan N, Kaye P. et al. A study of T1 relaxation time as a measure of liver fibrosis and the influence of confounding histological factors. NMR Biomed 2015; 28: 706-714
• 14 Luetkens JA, Klein S, Träber F. et al. Quantification of Liver Fibrosis at T1 and T2 Mapping with Extracellular Volume Fraction MRI: Preclinical Results. Radiology 2018; 288: 748-754
• 15 Obmann VC, Berzigotti A, Catucci D. et al. T1 mapping of the liver and the spleen in patients with liver fibrosis – does normalization to the blood pool increase the predictive value?. Eur Radiol 2021; 31: 4308-4318
• 16 Mesropyan N, Kupczyk P, Dold L. et al. Non-invasive assessment of liver fibrosis in autoimmune hepatitis: Diagnostic value of liver magnetic resonance parametric mapping including extracellular volume fraction. Abdom Radiol 2021; 46: 2458-2466
• 17 Obmann VC, Catucci D, Berzigotti A. et al. T1 reduction rate with Gd-EOB-DTPA determines liver function on both 1.5 T and 3 T MRI. Sci Rep 2022; 12: 4716
• 18 Peters AA, Wagner B, Spano G. et al. Myocardial scar detection in free-breathing Dixon-based fat- and water-separated 3D inversion recovery late-gadolinium enhancement whole heart MRI. Int J Cardiovasc Imaging 2022; 39: 135-144
• 19 Klaus JB, Goerke U, Klarhöfer M. et al. MRI Dixon Fat-Corrected Look-Locker T1 Mapping for Quantification of Liver Fibrosis and Inflammation – A Comparison With the Non-Fat-Corrected Shortened Modified Look-Locker Inversion Recovery Technique. Invest Radiol 2024;
• 20 Guimaraes AR, Siqueira L, Uppal R. et al. T2 relaxation time is related to liver fibrosis severity. Quant Imaging Med Surg 2016; 6: 103-114
• 21 Chow AM, Gao DS, Fan SJ. et al. Measurement of liver T1 and T2 relaxation times in an experimental mouse model of liver fibrosis. J Magn Reson Imaging 2012; 36: 152-158
• 22 Obmann VC, Ardoino M, Klaus J. et al. MRI Extracellular Volume Fraction in Liver Fibrosis – A Comparison of Different Time Points and Blood Pool Measurements. J Magn Reson Imaging 2024; jmri.29259
• 23 Sharma A, Nagalli S. Chronic Liver Disease. In: StatPearls. Treasure Island (FL): StatPearls Publishing; 2021. Accessed May 24, 2021 at: http://www.ncbi.nlm.nih.gov/books/NBK554597/
• 24 Zbinden L, Catucci D, Suter Y. et al. Convolutional neural network for automated segmentation of the liver and its vessels on non-contrast T1 vibe Dixon acquisitions. Sci Rep 2022; 12: 22059
• 25 Zbinden L, Catucci D, Suter Y. et al. Automated liver segmental volume ratio quantification on non-contrast T1-Vibe Dixon liver MRI using deep learning. Eur J Radiol 2023; 167: 111047
• 26 Ünal E, Akata D, Karcaaltincaba M. Liver Function Assessment by Magnetic Resonance Imaging. Semin Ultrasound CT MR 2016; 37: 549-560
• 27 Liu P, Li P, He W. et al. Liver and spleen volume variations in patients with hepatic fibrosis. World J Gastroenterol 2009; 15: 3298
• 28 Tarao K, Hoshino H, Motohashi I. et al. Changes in liver and spleen volume in alcoholic liver fibrosis of man. Hepatology 1989; 9: 589-593
• 29 Kromrey ML, Ittermann T, vWahsen C. et al. Reference values of liver volume in Caucasian population and factors influencing liver size. Eur J Radiol 2018; 106: 32-37
• 30 Patzak M, Porzner M, Oeztuerk S. et al. Assessment of liver size by ultrasonography: Sonographic Liver Size. J Clin Ultrasound 2014; 42: 399-404

Publication History

Received: 29 July 2024

Accepted after revision: 08 January 2025

Article published online:
10 March 2025