Multi-phasic magnetic resonance imaging of hemodynamic interchanges in hepatocarcinogenesis

Liver cirrhosis and hepatocellular cancer deem a substantial global health burden and are the end result of a variety of chronic liver diseases. Guidelines have been introduced to secure standardized approaches in the diagnosis and management of hepatocellular carcinoma (HCC). Established guidelines agree upon the distinctive dynamic enhancement findings of HCC, characterized by arterial phase wash-in and venous or delayed phase washout. The indeterminate focal hepatic lesions constitute a diagnostic dilemma. The aim of the study was quantification of hepatic vascular parameters using dynamic contrast-enhanced (DCE)-MRI to study liver hemodynamic disturbances that can differentiate between focal hepatic lesions during hepatocellular carcinogenesis. The study was conducted on 95 patients with comparing perfusion abnormalities across different liver cirrhosis pathologies, and the comparison revealed the correlation of hemodynamics with hepatocarcinogenesis and grades of cirrhosis. Relative enhancement curves were graphed to illustrate the different enhancement patterns across the spectrum of hepatonodular lesions of cirrhosis and among types of de novo and recurrent HCC. Statistical significance was highest between dysplastic nodules (DNs) and HCCs (maximal relative enhancement (MRE) 0.88, wash-in ratio (WIR) 0.84, washout ratio (WOR) 0.78, time to peak (TTP) 0.74, area under curve (AUC 0.73, T0 0.70), compared to those between cirrhotic liver and DNs (MRE 0.74, To 0.62). Least significances were between de novo and recurrent HCCs (TTP 0.66, r AUC 0.66, MRE 0.63). Performances between non-cirrhotic and cirrhotic liver [WOR 0.81, time to peak (TTP) 0.80, r AUC 0.63] surpassed those among different grades of cirrhosis (the highest was between non-tumoral and tumoral cirrhosis: TTP 0.74, MRE 0.68, WIR 0.65, WOR 0.65). Dynamic contrast-enhanced magnetic resonance imaging hemodynamic metrics promise potential usefulness as non-invasive biomarkers in assessment of liver cirrhosis, characterization of cirrhotic nodules, and evaluation of multistep hepatocarcinogenesis.


Background
The liver orchestrates metabolism in the human body and regulates a multitude of essential physiological processes [1].Liver cirrhosis and hepatocellular cancer are the end result of a variety of chronic liver diseases, including viral hepatitis, alcohol/aflatoxin-induced liver disease, and non-alcoholic steatohepatitis [2].
Liver cirrhosis is characterized by hepatic fibrosis, altered extracellular composition, and increased vascular bed resistance [3].These changes contribute to the deteriorated hepatic function by compromising transsinusoidal blood hepatocyte exchange and altering liver hemodynamics [4].The quantification of hepatic vascular parameters using dynamic contrast-enhanced (DCE) MRI has the potential to study liver hemodynamic disturbances [5,6] that underpin the pathophysiology of chronic liver disease [7,8].
Surveillance and early detection of HCC is crucial for curative resection, liver transplantation, and effective intervention [9].
Established guidelines agree upon the distinctive dynamic enhancement findings of HCC, characterized by arterial phase wash-in and venous or delayed phase washout [10].
The American Association for the Study of Liver Diseases (AASLD), the European Association for the Study of the Liver (EASL), and Liver Imaging Reporting and Data System (LIRADS) guidelines endorse dynamic cross-sectional imaging [9] with the use of extracellular contrast [10] to assess worrisome, 1cm or larger, nodules.In addition, the modified Response Evaluation Criteria in Solid Tumors (mRECIST) advocates the use of percentage change in residual lesion enhancement for treatment response assessment [9,11].
Research is focused on functional imaging and the emergence of non-invasive biomarkers that can reliably reflect the changes in various tumoral microstructural [12][13][14] and microcirculatory parameters [15].
Sequential intranodular hemodynamic changes are concomitant with the multistep process of hepatocarcinogenesis [16], which entails the gradual dedifferentiation of cirrhotic to dysplastic nodule, followed by the evolution of HCC [17].Intranodular arterial inflow increases and intranodular portal supply decreases as the disease advances [18,19].
Quantification of multiphase DCE-MRI signal change as a function of time is conducted to study perfusion changes [23,24].Semiquantitative analysis is a model-free approach that applies time-signal enhancement curves to derive physiologically interpretable and easily reproducible vascular measures [25].On the other hand, quantitative analysis is a complex model-based approach that constructs pharmacokinetic modeling to reflect underlying physiological properties of studied tissue [26].
Dynamic contrast-enhanced magnetic resonance (DCE-MR) functional imaging and perfusion analysis of hepatic and tumoral vascular parameters promise potential usefulness in the assessment of liver cirrhosis and hepatocarcinogenesis [27].

Aim of the work
The aim was to evaluate the functional role of DCE-MRI analysis of perfusion changes in multistep hepatocarcinogenesis.

Patients
The study was designed to enroll two main groups of patients, namely cirrhotic and non-cirrhotic liver.The cirrhotic group included 53 patients, 36 males and 17 females, and their ages ranged from 42 to 73 years.On the other hand, the non-cirrhotic group included 42 patients, 15 males and 27 females, and their ages ranged from 36 to 55 years.
The patients were referred to the Radiology Department over a period of 24 months from January 2021 to January 2023.
All enrolled cirrhotic patients had chronic viral-related hepatitis with cirrhotic morphological changes, including those who had additional intrahepatic de novo or posttherapeutic recurrent lesions.
Conversely, all non-cirrhotic patients either tested negative for viral hepatitis or tested positive but had no evidence of cirrhotic morphological changes on conventional MRI and complementary ultrasound (US).
Written informed consent was obtained from all individual participants included in the study, which was approved by the institutional review board.
Study design This was a prospective cross-sectional study.

MRI imaging
Magnetic resonance imaging (MRI) was conducted on a 3.0-T MR scanner (Philips Achieva, Philips Medical Systems, Netherlands) equipped with a phased-array torso surface coil.

Conventional MRI
Conventional unenhanced MRI sequences were obtained in axial plane utilizing respiratory-triggered techniques, with the following parameters: An unenhanced dataset was first acquired prior to contrast administration.Intravenous bolus injection of 0.1 mmol/kg body weight of gadolinium-DTPA (Magnevist; Schering, Berlin, Germany) was administered at a rate of 2 ml/sec using an automatic injector, which provided the estimated glomerular filtration rate more than 30.
Four successive imaging datasets were acquired.Dynamic datasets were timed to start after the arrival of the contrast bolus to the aortic arch, according to the following: arterial phase 20-30 s, portal phase 50-60 s, late portal/venous phase 80-90 s, and delayed phase 150-180 s.
Automated digital subtracted images were generated, by subtraction of the pre-enhancement T1-THRIVE dataset from the corresponding multiphase series of post-enhancement datasets, together with color-scaled maps.

Standard of reference
It was unfeasible to attain pathological confirmation for all studied pathologies, as many of the patients were unfit for surgery.Moreover, biopsy is considered invasive and resulted in sampling errors when intended for small nodular lesions [28], as was the case for many of the lesions in our study.Hence, we designed our study according to the following detailed standards of reference.
Cirrhotic liver was either biopsy-proven (with or without additional hepatocellular pathology), or confirmed by radiologically evident cirrhotic morphology on conventional MR imaging and complementary US. ''The cirrhotic liver develops characteristic morphological alterations such as surface nodularity, widening of fissures, expansion of the gallbladder fossa, notching of the right lobe, atrophy of the right lobe, and relative enlargement of the left lobe and caudate lobe" according to Faria et al. [29].
Non-cirrhotic liver was confirmed radiologically by the absence of any cirrhotic morphological changes on conventional MRI and complementary US.Non-cirrhotic patients were tested for virology and categorized accordingly.
De novo and recurrent HCC was either biopsy-proven or radiologically diagnosed by characteristic HCC enhancement criteria on dynamic MRI best appreciated on subtracted images [30][31][32] according to LIRADS, AASLD, and EASL.
Indeterminate dysplastic nodule category included both of the following: "high-grade dysplastic nodules" and the grossly indistinguishable "early, vaguely nodular, hypovascular HCCs with indistinct margins, " whereby both range 1-1.5 cm and "usually have similar diminished arterial and portal venous flow and hence are both hypoenhanced relative to the background liver in arterial and portal venous phases" [20].
Peri-ablation hyperemic rim referred to benign hyperemic ring-like enhancement that persisted on delayed phases, according to Luo et al. [33].
Tumor in vein (TIV) included thrombus involvement of the PV with detectable tumor enhancement on postcontrast dynamic series [34].

Imaging analysis
The review of all conventional and dynamic MR imaging was performed on the 3-T MR Unit workstation, utilizing the commercially available software (Philips Medical Systems), by two of the authors, a clinical radiologist with 5 years of experience and another clinical radiologist with 7 years of experience in cross-sectional MR imaging with 100% inter-observer agreement.
Conventional MR images were thoroughly assessed, and lesions were located and characterized according to their number, size, shape, signal criteria, and restriction pattern.Dynamic series with color maps and subtraction datasets were analyzed for lesion enhancement patterns and heterogeneity.In lesions with heterogeneous enhancement, areas of max hypervascularity were identified.Inherent/post-ablation precontrast T1 hyperintensities were subtracted out, and the presence of hypervascularity was confirmed.
Three user-defined regions of interest (ROls) were placed on areas of maximum hyper vascularity in each of the identified lesions.Three ROls were also randomly placed in the lesion-free or lesion-bordering liver parenchyma in each patient, together with one ROI in the descending abdominal aorta, PV, and IVC.ROls were standardized 2.1 mm 2 .
The direct use of time-signal changes would lead to inaccurate quantitation of enhancement owing to the differences in inherent T1 signal properties of studied tissues [35].Time-signal intensity curves were normalized to baseline signals to reconstruct time-relative enhancement curves (RE-T).From these curves, a variety of hemodynamic metrics can then be derived.
In addition, the relative area under the curve (r AUC) was manually calculated, averaged, and then tabulated, for each of the above pre-defined ROls.

Statistical analysis
All data were expressed as mean ± standard deviation.Unpaired Student t tests were used to compare different, normally distributed, hemodynamic parameters between the different lesion groups as well as between the different cirrhotic grades.Welch's correction was used to correct for statistically significant unequal variances.
Data analyses were performed on a personal computer using: GraphPad Prism version 6.00 for Windows, Graph-Pad Software, La Jolla, California, USA.
Receiver operating characteristic (ROC) curve was constructed, and the diagnostic performances were extracted by measurements of the area under the ROC curves.Postulated cut-off values were presented with their specificities and sensitivities.Mean time-%relative enhancements were graphed.

Results
The non-cirrhotic group was divided into "normal liver with positive virology" (30 patients) and "normal liver with negative virology" (12 patients).On the other hand, the cirrhotic group was divided into "stand-alone liver cirrhosis" (10 patients) and "liver cirrhosis with pathological lesions" (43 patients with 87 different lesions) (Fig. 1).
The liver cirrhosis with pathological lesion group was further divided according to the nature of the lesion (14 DNs, 52 de novo HCCs, 16 recurrent HCCs, 2 malignant PV thrombi, and 3 peri-ablation hyperemic rims).

Semiquantitative DCE-MRI parameters in hepatocarcinogenesis
Statistical interrogation comparing different perfusion metrics across different hepatocellular lesions of cirrhosis revealed the correlation of intranodular hemodynamics with multistep hepatocarcinogenesis (Fig. 6).
Mean time-%relative enhancement curves were graphed to illustrate the different enhancement patterns across the spectrum of hepatonodular lesions of cirrhosis (Fig. 7) and among types of de novo and recurrent HCC (Fig. 8).

Discussion
Dynamic contrast-enhanced MRI functional imaging is primarily focused on quantitative evaluation of tumoral perfusion and permeability as opposed to conventional MR structural imaging that focuses on the qualitative assessment of tumoral morphology and degree of enhancement, thus enabling an insight into the pathophysiology of tissue [36].Physiologic changes that precede morphological changes serve as early biological markers of tumorigenesis and response to treatment [37].
The aim of our study was to evaluate the functional role of DCE-MRI analysis of perfusion changes in multistep hepatocarcinogenesis.
Perfusion in the microcirculation, estimated by semiquantitative first-pass metrics such as WIR and MRE [38,39], and accumulation in the interstitium, measured by quantitative parameters such as K-1 0 (leakage into interstitium) and K-ep (return from interstitium to plasma), are two discrete successive phenomena that overlap [40].Despite the complexity of pharmacokinetic models, it has been proposed that simple relative enhancement curves are equally representative [35], whereby post-peak and late enhancement phases can simulate K-trans and K-ep, respectively [40].
The hepatocarcinogenic spectrum from benign regenerative to premalignant dysplastic nodules, followed by carcinogenesis of early vaguely nodular HCC, and progression to less differentiated HCC involves a sequential alteration in intranodular hemodynamics [41].
Understanding the interchanging dynamics of hepatonodular inflows and outflows is essential to correlate imaging criteria with histopathology [19].Interestingly enough, we managed to mirror this sequential interchange in hepatonodular flow kinetics and illustrate the different multiphase enhancement patterns across the carcinogenic continuum.The nodular phenotype is determined through angiogenic activity and interchange in drainage channels [36].Angiogenesis, provoked by tissue hypoxia and vascular endothelial growth factors (VEGF) [35], is essential for tumor initiation.Concurrent interchange in drainage vessels is characteristic for tumor progression [42].
Regarding hepatonodular inflows: Considerable overlap in radiological criteria occurs between high-grade dysplastic nodules and early hypo vascular HCC.Both are similarly hypovascular to background liver [41] as demonstrated on perfusion-weighted MRI by Elzenine et al. [23], due to their impaired native portal flow and insufficient pathological unpaired arterial growth [16].The only distinguishing feature, namely stromal invasion, can even be missed by histopathology [43].Having said that, it is crucial to point out that high-grade dysplastic nodules are considered premalignant and reported to transform into early HCC in periods as short as 4-month duration [11].
The current study results revealed outstanding diagnostic performance across all hemodynamic metrics when comparing this premalignant indeterminate  Tissue perfusion and microvascular density indicators, namely MRE and WIR, surpassed the remaining hemodynamic metrics in performance.Intranodular MRE exceeding 83%, reflecting both intra-and extravascular first-pass kinetics [44], provided the foremost reliable differentiation of nodular HCC from premalignant nodules with 88% accuracy and up to 94% confidence.Intranodular WIR more than 18 Usec, reflecting exclusive intravascular first-pass kinetics [44], ranked second with 84% accuracy and up to 92% confidence.This precedence of MRE over WIR could be attributed to the added effect of the tumoral microvascular blood volume that is more accounted for using maximum enhancement than first-pass increment [45].
Many researchers had comparably presented a variety of first-pass perfusion metrics such as WIR and first-pass RE by Wang et al. [46], RAE by Ippolito et al. [47], MRE by Elzenini et al. [23], and first-pass initial AUC by Zhang et al. [15].They all performed as reasonable indicators for tumoral perfusion [48].
The timing of the intranodular contrast increment (T Peak) less than 55.5 s (74% accuracy) outperformed that of contrast arrival (T Zero) less than 7 s (70% accuracy), when used to signal out carcinogenesis.We accounted the under-performance of T Zero to the effect of physiological body functions on the rate of contrast delivery to the tissue, such as cardiac function [23].In our study, AUG ranked late in diagnosing HCC with 73% accuracy, as opposed to better performing first-pass initial AUC, Fig. 4 A 56-year-old male patient presented with HCC lesion managed by radiofrequency ablation with suspected recurrence: a liver cirrhosis with previously ablated hepatic focal lesion currently presented with tiny focal enhancement worrisome for focal nodular outgrowth recurrence on digital subtraction arterial phase MRI; b corresponding color-scale relative enhancement mapping MRI allows enhanced identification of the nodular focus of maximum relative enhancement; c corresponding time-% relative enhancement curves of the aorta (red), PV (dark yellow), IVC (blue), lesion-free cirrhosis (green), adjacent perilesional hyperemia negative for recurrence (purple), and focal tiny nodular hyperenhancement with characteristic wash-in and washout positive for early nodular outgrowth recurrence (black); d quantitative comparison of the hemodynamic metrics of negative versus positive early nodular tumoral recurrence used by other researchers as Zhang et al. [15].Moreover, our results undermined the significance of relative AUC, a relative metric proposed by Cuenod and Balvay [21], when utilized as a marker of carcinogenesis.
Regarding hepatonodular outflows: Kitao et al. [42] explained that invasion of the intranodular hepatic veins first occurs through carcinogenic replacement growth, because of the absence of perivascular connective tissue covering.Collapse of intra-and peri-nodular hepatic sinusoid drainage then follows by tumorigenic compressive growth.Reversed portal venous outflow and subsequent extension of peritumoral corona enhancement finally proceed in line with tumor progression [16].These interchanging outflow dynamics have been well reflected in our multiphase enhancement patterns evidenced by the transition from weak late venous washout in indeterminate dysplastic nodules to strong early portal washout in nodular carcinogenesis.Furthermore, we propose that quantitated intranodular WOR, exceeding 2.2 Lisee, could be implemented as a biomarker for HCC with a reasonable accuracy of 78%.
Progression of hepatocarcinogenesis involves concurrent changes in tumoral tissue microvascular architecture (density) and properties (permeability) [41].We attempted to simulate this altered microvascular environment through mirroring the different enhancement patterns across different tumoral grades.We reported that less differentiated and more progressed large HCCs tend to generally express higher first-pass RAE, shallower wash-in, and weaker washout as opposed to the more differentiated and less progressed small HCC.These results were contradicted the Yamashita et al. [49] study.
Differences in poorly differentiated HCC enhancement kinetics could be explained by the more limited tumoral vascular reserve [44] because of reduced microvascular density [13] and the increased extracellular space [23].Moreover, large fibrous capsulated, more progressed HCC [32] presented higher first-pass RAE, steeper wash in, and stronger washout with characteristic corona enhancement as opposed to small nodular, less progressed HCC.Corona enhancement was the initial site of drainage [50] and recurrence.Quantitated MRE and T Peak underperformed in differentiating recurrent from de novo HCC.
Furthermore, therapeutic evaluation can be reliably monitored by DCE-MR.Several studies assessing treatment response, such as Wang et al. [46] expressed significant differences in hemodynamic parameters, namely MRE and AUC, respectively, between anti-angiogenic responders and non-responders [27].Hence, future trends in research advocate the use of DCE -MR imaging in the assessment of targeted HCC therapy [8].
Likewise, the continuum of hepatofibrogenesis is accompanied by gradual obliteration of normal vascular and sinusoidal channels, thus increasing the vascular bed resistance [5], slowing the blood passage through the parenchyma, and altering the liver hemodynamics [29].
In our study, TTP excelled in diagnostic performance as a non-invasive early biological surrogate to differentiate between cirrhotic and non-cirrhotic liver scoring 80% accuracy, DO RKG et al. [27], had also previously confirmed, by histology, that mean transit time (MTT) was most sensitive and specific for assessing the severity of fibrosis.
Finally, the uniqueness of our study resides in its attempt to shed light on diagnostic capacity of perfusion MRI using both "relative enhancement curveology" and "semiquantitative perfusion analyses." Liver Imaging Reporting and Data System (LIRADS) was applied since 2013, and the LIRADS categories  reflect probabilities; they do not correspond to exact histological categories [51,52].The observations noted such as regenerating and dysplastic nodules are categorized LR2 (probably benign) although imaging cannot exclude malignant foci in such lesions [52,53].
Limitations of this study were multiple A major limitation was the indistinguishable overlap between premalignant nodules and early hypovascular HCC due to the lack of histopathologic confirmation.The most encountered problematic limitation was the inter image discordance between consecutive dynamic datasets    due to differences in respiratory depths.This was best compensated for using respiratory gated breath-hold expiratory volume acquisitions.User-defined ROls fail to embody the heterogeneity commonly encountered in tumors.Semiquantitation fails to constitute the signal changes due to disordered arterialization, sinusoidal capillarization, vessel permeability, and interstitial pressure.We tried to offset this using relative metrics such as relative enhancement versus signal intensity and normalized metrics such as r AUC versus AUC.

Conclusions
Multiphase DCE-MRI perfusion analysis is useful as non-invasive tool in many diagnostic dilemmas, the assessment of liver cirrhosis, characterization of cirrhotic nodules, the evaluation of multistep hepatocellular carcinogenesis, and the diagnosis of HCC post-management recurrence.

Fig. 1
Fig. 1 Schematic diagram outlining the design of the study

Fig. 2 A
Fig. 2 A 45-year-old male patient presented with liver impairment, jaundice liver cirrhosis: a Liver cirrhosis with multifocal multinodular de novo HCCs on arterial phase MRI; b corresponding time-%relative enhancement curves of the different nodular HCC grades (colors) and adjacent lesion-free cirrhosis (green) are shown.c and e Liver cirrhosis presented with multicentric large HCCs: pseudo-encapsulated (Hcc 1) and locally invasive (Hcc 2) subtypes on arterial phase MRI; d corresponding time-%relative enhancement curves of the lesion-free cirrhosis (green) and different large HCC subtypes (yellow) and (purple), with their hemodynamic metrics (L4) and (L17), respectively

Fig. 3 A
Fig. 3 A 53-year-old female patient presented with liver cirrhosis, CT-detected indeterminate focal hepatic lesion a liver cirrhosis with a large true fibrous capsulated HCC that reveals tumoral enhancement heterogeneity on late phase MRI; b together with subsequent underlying hepatic parenchymal invasion depicted on T2 fat-suppressed WI; c corresponding time-%relative enhancement curves of the hypovascular (red) and hypervascular (black) tumoral enhancement, as well as the underlying hepatic parenchymal invasion (yellow) and lesion-free cirrhosis (green) are shown; d hemodynamic metrics of hypovascular and hypervascular tumoral regions

Fig. 5 A
Fig. 5 A 62-year-old male patient presented with liver cirrhosis and HCC lesion managed by radiofrequency ablation with suspected recurrence: a advanced liver cirrhosis with recently ablated two adjacent hepatic focal lesions [black plus (+) symbols] as evidenced by the signs of intra-lesional coagulative necrosis and right subcapsular juxta-lesional RF needle track, currently presented on follow-up imaging with massive infiltrative territory of heterogeneous hyperenhancement [red plus (+) symbol] on arterial phase MRI; b corresponding color-scale relative enhancement mapped MRI further enhances the intensity and heterogeneity of this infiltrative tumoral hyperenhancement; c and d corresponding time-%relative enhancement curves and quantitative hemodynamic metrics compare the non-viable ablated lesions (black) versus the surrounding massive infiltrative readily washed out hyperenhancement positive for viable large recurrence (red), and compare as well the nearby right lobe advanced cirrhotic background (turquoise) versus the contralateral left lobe relatively less advanced cirrhosis (green)

Fig. 6
Fig. 6 Illustrative graph showing the interchanging intranodular hemodynamics across the spectrum of hepatocellular carcinogenesis.Note non-cirrhotic liver (liver N), cirrhotic liver (liver C), indeterminate dysplastic nodule (nodule D), de novo HCC (HCC N), recurrent HCC (HCC R), maximum relative enhancement (MRE), wash-in ratio (WIR), washout ratio (WOR), time to peak (T Peak), time to arrival (T Zero), area under the curve (AUC) More asterisk indicate the more significance Liver N Non-cirrhotic liver, Liver C Cirrhotic liver, Nodule D Indeterminate dysplastic nodule, HCC N De novo HCC, HCC R Recurrent HCC, Ablated R Recurrence after ablation, MRE Maximum relative enhancement, WIR Wash-in ratio, WOR Washout ratio, T Peak Time to peak, T Zero Time to arrival, AUC Area under the curve

Table 1
Hepatocarcinogenesis hemodynamics (mean ± SD)Liver N Non-cirrhotic liver, Liver C Cirrhotic liver, Nodule D Indeterminate dysplastic nodule, HCC N De novo HCC, HCC R Recurrent HCC, Ablated R Recurrence after ablation, MRE Maximum relative enhancement, WIR Wash-in ratio, WOR Washout ratio, T Peak Time to peak, T Zero Time to arrival, AUC Area under the curve