Introduction
Regression of portal hypertension: spontaneous mechanisms
Pre-clinical evidence for PH regression upon etiologic treatment
Clinical evidence for PH regression upon etiologic treatment
-
HBV: Long-term studies (5–7 years) showed a histologically proven progressive reduction in necroinflammation and fibrosis scores in a vast majority of antiviral treatment (AVT)-responsive patients with HBV advanced fibrosis or cirrhosis [21, 22]. In addition, several studies demonstrated the beneficial effects of viral suppression on non-malignant decompensating events [23, 24], low grade esophageal varices (EV) progression [25, 26], EV development rate [26], and clinical scores and transplant-free survival [27]. Although the evaluation of EV dynamics is based on subjective judgement, the evidence provided by these trials suggest an underlying decrease in PH. One study including 19 patients with HBV-related CSPH showed a median reduction of 18.7% in HVPG during the 12 months of follow-up, with no significant systemic hemodynamic changes [28], suggesting a reduction in HVR, possibly through decreased hepatic necroinflammation.
-
HCV, pre-DAA: While sustained virological response (SVR) after HCV therapy is significantly higher since the introduction of direct-acting antivirals (DAA), several pre-DAA studies already held promise of PH regression induced by etiological intervention. As is the case with HBV AVT, treatment of HCV can also prevent the development of EV or slow down progression towards decompensation [29]. However, it seems that this effect is less consistent once EV is already present before treatment initiation, suggesting a reduced effect on already established PH [29, 30]. Moreover, in spite of PH decrease, decompensation is not always prevented and regression below the CSPH threshold is not always achievable, even with long-term follow-up (5.2 years after end of treatment) [31]. Short-term hemodynamic and/or histology studies in compensated patients demonstrated HVPG reduction paralleled by reduction in necroinflammatory scores, but either no or very weak reduction in fibrosis scores, possibly due to the short follow-up time [32, 33]. Around 60% of treated HCV patients with advanced fibrosis or cirrhosis followed up over a long time period (2–5 years) showed a progressive reduction in fibrosis score on repeated biopsy and significant reduction in scar collagen content, even in those without obvious fibrosis regression [34‐36]. However, no changes in sinusoidal capillarization, as assessed by CD34 positivity or α-SMA staining, were observed, pointing towards lack of intrahepatic vascular remodeling [34].
-
HCV, DAA: With the emergence of DAAs, even patients in advanced stages of CLD can achieve viral eradication, which leaves open the question whether and to what degree their disease can be reversed. It is now clear that SVR after DAA can result in clinically significant (≥ 10–20%) decrease in HVPG [37, 38], even in those patients with baseline CSPH and in difficult to treat populations, such as HCV-HIV co-infected patients [38]. However, this effect seems less pronounced and more heterogeneous in those patients with advanced-stage CLD (high Child–Pugh score, HVPG > 16 mmHg) [37, 39]. A study on post-transplant graft HCV reinfection showed fibrosis regression and HVPG decrease in 67% and 66% of patients which achieved SVR, however, among F4 biopsies, none of the patients displaying thick fibrous septa (Laennec C cirrhosis) had cirrhosis regression [40]]. Histological data show that HVPG reduction is lower in those patients who still have necroinflammatory activity after SVR [38], which points towards an initial phase of reduction in HVPG through decreased intrahepatic inflammation [41]. However, it seems that HVPG decreases progressively over a longer time period, suggesting a long-term possibility of further decrease in HVPG based on mechanisms other than reduced inflammation. Still, in a relatively high proportion of patients, CSPH did not resolve, leaving them a risk for future decompensation [42‐44].A significant decrease in HVPG in patients without baseline CSPH is particularly important given its potential to prevent progression of disease (Fig. 3). Indeed, achieving SVR seems to decrease disease severity (Child–Pugh and MELD scores), decompensation risk and EV grade [39, 45, 46]. These important improvements in HCV-induced CLD management are also mirrored in the shift in transplant indication from the pre-DAA era and even within the last few years [47]. The effect of DAA on systemic hemodynamics is still not well established, with some studies showing mild but significant increase in MAP and systemic vascular resistance [37]; whereas, others described no influence [38].
-
NAFLD: The vast majority of patients benefit from different weight loss surgery approaches, with normalization or reduction in tissue fibrosis [48, 49], although this is more evident in early fibrosis [50]. Lifestyle interventions have an important role in obese CLD patients (any etiology), with 42% of patients showing a significant decrease (≥ 10%) in HVPG from baseline after 16 weeks, paralleled by a decrease in insulin resistance and plasma leptin levels, giving an insight into possible mechanisms of regression in this patient population [51]. Moreover, this last study showed no change in portal blood flow, thus reinforcing the probable effect on HVR of etiological treatment in CLD. A subsequent study investigating the effects of physical activity in cirrhotic patients with CSPH (compensated or decompensated) also demonstrated a reduction in HPVG [52].
Etiology | Treatment/intervention | Study population | HVPG decrease | FU time | Observations |
---|---|---|---|---|---|
HBV [28] | Lamivudine | n = 19, all with BL CSPH | 18.7%a | 12 months | 60% of patients had a clinically significant reduction in HVPG (≥ 20% from BL) |
HCV [32] | Ribavirin + PegIFN | n = 20, 55% with BL CSPH (defined as 12 mmHg in this study) | 28%b | 24–48 weeks | All patients had a reduction of ≥ 10% from BL HVPG |
HCV [31] | Ribavirin + PegIFN | n = 100, 74% with BL CSPH | 7.7%a | 5.2 years (median time) | 40% of patients had a clinically significant reduction in HVPG (≥ 10% from BL) but 2/3 of patients still displayed CSPH, leaving them at risk of developing decompensation |
HCV [39] | DAA (different regimens) | n = 104 (n = 60 with FU HVPG), 68% with BL CSPH | 23%b | 12–24 weeks | Normalization of HVPG in 63% (BL HVPG = 6–9 mmHg) and 43% (BL HVPG = 10–15 mmHg) of patients; 60% of patients with BL HVPG ≥ 16 mmHg did not resolve CSPH |
HCV [37] | DAA (different regimens) | n = 226, all with BL CSPH | 13%a | 24 weeks | 78% of patients did not reach CSPH resolution, remaining at risk for decompensation |
HCV + HIV [38] | DAA (different regimens) | n = 22, 50% with BL CSPH | 32%b | 12 weeks | 77% of patients had a clinically significant reduction in HVPG (≥ 10% from BL). HVPG response correlated with decrease in necroinflammatory activity |
HCV [40] | IFN-based or IFN-free regimens | n = 112 patients with graft HCV reinfection after LT, 31% with BL CSPH | − 2.5 mmHg | 12 months | Low probability of regression in those with BL CSPH (18%) and Laennec C histology (0%) |
DAA (different regimens) | n = 117, only patients with CSPH at SVR | 24.4% | 96 weeks | 53% of patients remained with CSPH | |
DAA (different regimens) | n = 67, 74% with BL CSPH | 20%a | 8.8 months (median) | n = 19 patients also included in [39] underwent a 3rd HVPG measurement, with an increase in the proportion of patients with resolved CSPH from 15 to 46% | |
Any etiology + obesity [51] | Nutritional intervention + moderate physical activity | n = 50, 72% with BL CSPH | − 1.6 mmHg | 16 weeks | 42% of patients achieved a reduction of ≥ 10% from BL HVPG |
Any etiology [52] | Nutritional intervention ± physical activity | n = 22 | − 2.5 mmHg | 14 weeks |
Cellular and molecular events limiting spontaneous regression of PH
-
Fibrotic tissue properties: In a murine CCl4 model of CLD, followed up for 1 year after cessation of toxicant administration, regression was only partial, with transformation from micronodular to macronodular pattern and incomplete resolution of broad mature septa [14]. This mirrors changes described during human cirrhosis regression [53]. In humans, small nodule size and increased septal thickness have both been correlated with higher HVPG and seem to be predictive of decompensation [54] and patients with Laennec C cirrhosis are unlikely to have a significant HVPG decrease even after removal of etiological agent [40]. In murine models, enzymes such as tissue-transglutaminase (tTG) and lysil oxidases (LOX) induce the cross-linking of collagens and elastins, creating acellular areas that are resistant to degradation [14, 55, 56]. A study using tTG2 KO mice showed that this molecule is not indispensable during fibrogenesis and mice lacking it do not have a resolution advantage compared to wild-type mice [57]. This could suggest alternative collagen cross-linking pathways. In this regard, elastin also plays an important role in the irreversibility of advanced fibrosis. Elastin to collagen ratio increases with disease progression and, despite an early increase in expression, its marked accumulation only occurs in advanced stages. This points towards an imbalance between synthesis and removal, also suggested by the increase in matrix metalloproteinase 12 (MMP12) bound to tissue inhibitor of matrix metalloproteinase-1 (TIMP-1), which renders it inactive. Moreover, MMP12 KO mice display significantly higher level of bridging fibrosis, further suggesting a defect in elastin removal [58]. Depletion of macrophages results in additional accumulation of elastin compared to wild-type mice and with failure of tissue remodeling. Indeed, pro-resolution macrophages are an important source of MMPs [58, 59].
-
Role of HSCs: Activated HSCs (aHSC) play a key role in fibrosis, and an important event necessary for its resolution is their disappearance through either senescence, apoptosis or inactivation [60]. HSCs situated within mature insoluble septa seem to be less prone to undergo apoptosis [14]. Indeed, it seems that persistence of scar tissue is associated with the maintenance of aHSC [61], which importantly also highlights the role of the underlying matrix in influencing cell phenotype [62]. aHSC are a major source of TIMPs and TIMP-1 overexpressing murine models fail to show resolution of fibrosis [63]. Moreover, it has been shown that TIMP-1 itself promotes survival of aHSC [64]. Although HSC inactivation may occur during fibrosis regression, these cells seem to remain more sensitive to renewed exposure to fibrogenic stimuli compared to their normal counterparts [65].
-
Vascular phenotype: In addition to fibrosis, the other crucial component and the major causative factor of pathophysiological consequences of cirrhosis are the vascular changes, both intra- and extrahepatic. Hepatic endothelial de-differentiation and neo-angiogenesis depend on the initial injury pattern and it may well be that the endothelial and vascular changes are in fact the most important determinants of regression capacity [5, 66, 67]. As recently proposed by Wanless, advanced stage CLD can progress independent of the initial etiological agent, due to a vicious circle in which vascular injury promotes vascular obstruction which leads to renewed vascular injury and hepatocellular damage; the so-called ‘congestive escalator’ [68]. In line with this, a recent study in murine cirrhosis shows persistent liver hyperarterialization, in spite of cirrhosis regression [69].
-
Aging and other factors: Advanced age has been shown to be an important determinant of CLD severity in murine models, results that were corroborated by HVPG and gene expression differences in human CLD patients with more advanced age [70]. Age seems to also be involved in CLD regression capacity, as shown in a murine CCl4 model, in which old mice were significantly less prone towards CLD reversal, as assessed by liver histology and ECM remodeling pathways (including macrophage populations) compared to their young counterparts [71]. Moreover, genetic and epigenetic factors likely play a role in the capacity and speed of CLD regression [72]. Last but not least, it is important to consider additional pro-fibrogenic factors which could influence the rate of regression in patients even after causative treatment, such as metabolic risk factors or excessive alcohol intake.
Regression of portal hypertension: therapy-driven strategies
Drug/substance | Drug/substance class | Effect on HVPG | Other effects | |
---|---|---|---|---|
Beneficial effect on HVPG | Statins | ↓, ↘ | ↓ mortality, ↑ IGC | |
Renin–angiotensin–aldosterone system antagonists | ↘ | Renal effects, ↓ MAP (CPS B and C), reduction in fibrosis progression | ||
Galectin 3 inhibitor | Belapectin [98] | ↘ (study without BL EV) | Prevention of de novo EV, ↓ hepatocyte ballooning | |
FXR agonist | Obeticholic acid [100] | ↘ | ||
Rho-kinase inhibitor | Fasudil [89] | ↘ | ↓ SVR, ↓ MAP | |
Multikinase inhibitor | Sorafenib [101] | ↘ | ↓ VEGF, PDGF, PlGF, RhoA and TNFα mRNA levels | |
Probiotic | ≈/↘ | ↑ serum Na2 + , ↓ plasma TNFα levels ↑ NSBB response rate | ||
Essential amino acid | Taurine [102] | ↘ | ||
PDE-5 inhibitors | Udenafil [86] | ↓ | ||
Vardenafil [85] | ↓ | |||
≈ (↓ HVR and ↑PBF) | ↓ MAP | |||
Antioxidants | Dark chocolate [106] | ↓ Attenuation of postprandial HVPG increase | ↑ MAP | |
Ascorbic acid [107] | ↓ Attenuation of postprandial HVPG increase | |||
Endothelin receptor antagonists | BQ-123 (ETA)—intrahepatic administration [94] | ↓ | ||
Ambrisentan (ETA) [94] | ↓ | ↓ MAP | ||
No effect on HVPG | Endothelin receptor antagonists | Tezosentan [91] (dual ETA & ETB) | ≈ | |
BQ-123 (ETA) [92] | ≈ | ↓ MAP, ↓ SVR | ||
BQ-788 (ETB) [92] | ≈ | ↑ MAP, ↑ SVR | ||
LOXL2 inhibitor | ≈ | |||
Pan-caspase inhibitor | ≈ | |||
Tetrahydrobiopterin analog | Sapropterin [96] | ≈ | ||
Relaxin-2 analog | Serelaxin [97] | ≈ (trial stopped prematurely) | ||
Antibiotics | ≈ | ↑ MAP, ↑ SVR | ||
Undetermined | ↓ inflammation and bacterial translocation serum markers additive effect to b-blocker therapy |
Vasomodulators
Therapies leading to reduced inflammation/fibrosis
Studies not evaluating HVPG
Drug (drug class) | Effects |
---|---|
Enoxaparin (anticoagulant) [122] | ↓ probability of PVT development ↑ survival *A trial with rivaroxaban, another anticoagulant, is currently ongoing [123] |
Liraglutide (GLP-1 analog) [124] | ↓ progression of fibrosis (but no significant improvement) NASH resolution |
Selonsertib (ASK1 inhibitor) [125] | No effect on fibrosis |
Cenicriviroc (CCR2 and CCR5 antagonist) [126] | Improvement in fibrosis (effect more pronounced on patients with more advanced disease) ↓ in collagen area by morphometry, ↓ in systemic inflammation biomarkers |
Pioglitazone (PPAR γ agonist) and vitamin E [129] | Improvement in NASH (vitamin E but not pioglitazone) No improvement in fibrosis for any of the trial drugs |
Lanifibranor (PPAR α/δ/γ agonist) [130] | NASH resolution Improvement in fibrosis |
G-CSF or G-CSF followed by CD133 + cells (cell therapy) [131] | No improvement in liver function tests, non-invasive fibrosis markers, MELD or CPS |
Biomarkers of PH regression
Biomarker | Short description of study |
---|---|
Liver stiffness [39] | In a cohort of DAA-treated HCV patients, LS decrease (measured by TE) was associated with HVPG response. However, its accuracy was lower in patients with BL CSPH |
In a cohort of DAA-treated patients, LS ≥ 21 kPa had a good performance in ruling in the persistence of CSPH after SVR (positive predictive value 82–91%); however, the lower cut-off of 13,6 kPa did not perform well in ruling out CSPH persistence | |
Liver stiffness [136] | In this small cohort of DAA-treated HCV patients, a cut-off value of < 12 kPa was accurate in ruling out CSPH after SVR |
Liver stiffness and ELF score [40] | In a cohort of LT patients with HCV reinfection, LS was accurate in ruling in our out the persistence of CSPH (cut-off values < 11.3 and > 23 kPa resp) and the persistence of advanced fibrosis. Conversely, ELF showed good accuracy for CSPH, but was not associated with fibrosis regression |
Liver stiffness and VITRO score [43] | In this cohort of DAA-treated HCV patients, TE and VITRO score performed well in ruling in/out CSPH after SVR. Their accuracy was especially high if used in a sequential manner, leaving 25% of patients unclassifiable |
Spleen stiffness [137] | In this cohort of DAA-treated HCV patients, SS decreased significantly after SVR, more so in patients without BL CSPH. However, the presence and grade of PP was estimated based on LS, and no direct HVPG measurements were performed |
Spleen stiffness [138] | This proof-of-concept study demonstrated that a decrease in spleen stiffness accurately predicts the hemodynamic response to primary prophylaxis with NSBB (Carvedilol) in patients with HREV. A prediction model containing SS had an AUC > 0.8 in both derivation and validation cohorts |