Exclusion criteria

Out of the initial patients, 23 were excluded because their instrumental examinations (DUS) had been previously performed in different centres. Fourteen patients, who had undergone endoscopic EV ligation therapy, and eight, who had received beta-blockers, were also disallowed from the study because prior treatment might have caused a change in DUS features. Three patients were left out for the detection of thrombosis of the portal vein.

Inclusion criteria

The remaining 81 patients formed the study population (35 females) whose age was 68.2, 8.18, 66.4-70 years (mean, +/-SD, 95% CI). Chronic liver damage in these patients was caused by hepatitis C (n = 66), hepatitis B (n = 2), alcohol abuse (n = 2) or unknown etiology, likely Non-Alcoholic SteatoHepatitis (NASH) (n = 11). Forty seven patients had compensated cirrhosis of the liver. For 71 patients, the diagnosis of cirrhosis was established by contextual clinical (spider nevi, hepato-splenomegaly), laboratory (low serum total cholesterol, prothrombin activity and pseudocholinesterase levels, reduced white blood cell and platelets count, globulin/albumin ratio >1,[24]) as well as antecedent imaging data. Being these parameters inadequate to confirm the diagnosis in 10 patients, successively assigned to the Child's class A, a liver biopsy was performed. The non-invasive assessment of liver cirrhosis was blindly performed de novo to all patients by experts on the basis of UltraSound (US) and DUS examinations (coarse echo-texture, nodularity presence, increased caudate/right lobe ratio, hypertrophy of the left lobe, characterized by a rounded inferior marginal edge, and portal vein enlargement with decreased flow velocity, absence of a normal doppler waveform, hepatofugal flow). Alcohol abuse was diagnosed according to the DSM-IV criteria, by means of screening tests such as MAST (Michigan Alcohol Screening Test) and CAGE (Cut down, Annoyed, Guilty, and Eye opener), as well as random tests for blood alcohol concentration and the use of a surrogate marker, e.g., Mean Corpuscular Volume.

Prestudy agreement, shared by all the investigators, gave place to common diagnostic procedures in order to meet the target of concordance. Patients should have fulfilled the following criterion at entry, i.e., US/DUS, endoscopic and laboratory examinations performed within four weeks of each other.

Endoscopy

EV were graded according to a previous classification, i.e., F1 small and straight; F2 moderately sized, tortuous, and occupying less than one third of the lumen; F3 large, coiled, and occupying one third or more of the lumen [25]. Large EV (LEV) were considered F2 and F3.

Ascites presence at ultrasound

The superior end of the right paracolic gutter and the pelvis were carefully assessed at US. Small quantities were sought for around the liver or spleen surface and in the Morrison's pouch [26]. The presence of peritoneo-pleural communications was determined. BMI of patients with ascites was adjusted taking into consideration the weight before this complication.

Hepatic encephalopathy grade

HE was graded based on the level of consciousness, intellectual functions, behaviour and neuromuscular functions according to West Haven (W-H) criteria [27]. W-H grade 0 or minimal encephalopathy was ascertained as previously described [28].

Assessment of liver cirrhosis severity

Hepatocellular carcinoma detection

Patients were screened for HCC by performing abdominal US and testing alpha-fetoprotein at admission, in course of regular 6-monthly or annually (by computed tomography) surveillance. When a surveillance feature or test was abnormal (>10.9 ng/mL), a triple-phase imaging (MRI) was recommended for evaluation at recall.

Ultrasound and Doppler ultrasound features of spleno-renalshunts

The SRS detection was based on the following features:

Evidence of tortuous, inferiorly directed vessels from the splenic hilum to the left kidney (when dealing with the planes of orientation, i.e., coronal, horizontal, and sagittal, they were primarily seen on the coronal ones);

Splenofugal blood flow [31];

Dilatation of the left renal vein with blood flow phasic and at high velocity (≥20 cm/sec);

Marked increase of the spleen volume (see below);

Dilatation of the splenic vein diameter (≥5 mm) with blood flow phasic and at high velocity (SVFV ≥15 cm/sec);

Statistics

BMI, SVFV and Child-Pugh's classification values, not normally distributed (Shapiro-Wilk test, p = 0.001, 0.001 and 0.01), were expressed as median (range). HE, EV and ascites grades were considered ordinals and managed in the same way. The difference in medians was assessed by the Kolmogorov-Smirnov (K-S) Two-Sample test. Age data, derived from a normally distributed population (Shapiro-Wilk test, p = 0.44), were articulated as mean plus SD and 95% confidence intervals (CI); the difference of means was evaluated by Two-Sample t test. The Two-Way Tables cross-tabulated one or more than one categorical row variables with one categorical column variable and the significance was set by the Pearson Chi-square. When cross-tabulation was stratified for another dichotomous variable the Mantel-Haenszel Chi-square was carried out. The correlation between Child-Pugh's scores and SVFV values was determined by the Spearman's rho. To make prediction the logistic regression (Enter method) was adopted, selecting as independent variable BMI values as well as the Child-Pugh classification and as dependent variable SRS presence/absence; HE and ascites severity were not tested in the context of the previous independent variables to avoid multi-collinearity (situation in which the predictors are correlated to each other to some degree). To assess the independent effect of a quantitative variable, i.e., SFV, on the prediction of the ascites, HE and EV grades the linear regression analysis (least squares) was used, evaluating the standardized coefficient beta (β). The Factor Analysis was applied to detect the structure in the relationships among variables selecting a subset of variables, which have the highest correlations with the principal component factors. The Cattel Scree plot, with relative eigenvalues, was performed to screen the real factors. Extraction of the main components amounted to a variance maximizing (varimax) rotation of the original variable space. The critical value was calculated by the formula: doubling the Pearson's correlation coefficient for 1% level of significance (5.152)/square root of patients minus 2, i.e., 0.579. The concordance correlation coefficient (ρ_c), which measures precision and accuracy, was adopted to evaluate the degree of pair observations at US. The power of our sample size was calculated by the non-equality of proportion with a type I error, alpha, of 0.05. Statistical analysis was performed operating on Systat 12 (Richmond, CA, USA) and MedCalc Version 11^® (Frank Schoonjans) software packages.

Prevalence

Association

The correlation of the studied DUS parameter, i.e., SVFV, to the degree of liver failure, weighed by Child-Pugh's scores was inversely significant (rho = -0.430, p = 0.001).

Interestingly, a hidden correlation was found among the SLD, the SVFV and the BMI, independently from age of patients and perhaps of the illness, Figure 1.

Prediction

The estimates of Child-Pugh's classification and BMI values in predicting the SRS presence gave the following results: OR = 1.145, 95% CI = 0.77 - 1.51, p = 0.66 and OR = 1.84, 95% CI = 1.28-2.64, p = 0.001, respectively. Lower SVFV values predicted more advanced ascites, HE and EV grades (β = -0.29, p = 0.009; β = -0.37, p = 0.001 and β = -0.41, p = 0.0001, respectively).

Accuracy

The agreement of US paired observations ranked high, i.e., ρ_c = 0.90.

Power analysis

When analysing the impact of the sample size regarding the HCC presence through the SRS distribution, the power was 74% with a type I error of 0.05.

Discussion

The essential findings we provide in this study are briefly i) greater occurrence of HCC in patients with SRS than in those without them; ii) clear prediction of BMI towards the SRS presence; iii) detection of fewer LEV in patients presenting SRS; iiii) same prevalence of ascites and HE presence in patients with and without SRS. Although knowledge regarding mechanisms involved in the pathogenesis of PH has taken unprecedented levels, nevertheless many prognostic aspects still remain to be elucidated. If one considers that knowing them could help better manage cirrhotic patients, it would then be justifiable to speculate whether physicians could predict the history of liver cirrhosis on the basis of the spontaneous SRS presence by some means. This study aimed at partially answering this question. The body of present knowledge [9] is in favour of the fact that the presence of portosystemic collaterals identifies cirrhotics with a less favourable clinical course. Our data only partially agree with this conclusion. In fact, we provide evidence for a low risk of LEV and a quite similar prevalence of HE and ascites in patients with SRS. However, the procrastination of haemorrhagic risk from LEV in SRS patients would be only temporary, because the basal critical PH identifies a more severe grade of liver cirrhosis, i.e., very high and long-lasting sinusoidal resistance.

Some of our results are expected if the following is examined. A recent study comparing Endoscopic Sclerotherapy (ES) with distal surgical portosystemic shunts in the prevention of recurrent variceal bleeding, has showed that both these interventions achieved a positive effect on variceal rebleeding [32]. A switch to decompressive shunt procedures is mandatory in case of ES or Transjugular Intrahepatic Portosystemic Shunt (TIPS) insertion fails to control recurrent variceal hemorrhage [33]. Selective shunts are placed surgically to manage post-transplant portal vein stenosis/thrombosis. In contrast, preexisting spontaneous portosystemic shunts increase the risk of post-transplantation portal vein thrombosis [34] and hamper graft survival [35]. Furthermore, SRS main disadvantage is decreased perfusion of the liver with eventually reduction in liver volume and related function [36]. Nevertheless, surgeons are convinced that the creation of peripheral portosystemic shunt still has a role in the treatment of some patients with PH on the basis that long-term blood flow in the portal vein was not severely reduced after this intervention [37]. Finally, a meta-analysis of individual patient data has provided further evidence that TIPS significantly impacts only on those patients with the extreme grade of ascites [38], although it worsens encephalopathy [39]. The novelty of our study relies on the evidence of an interesting percentage of patients with SRS in which HCC has been detected, mainly taking into consideration that HCC prevalence in our population was lower than that found among Japanese patients (confined to HCV infection), obtained summarizing four representative studies [40], which was 16% (212/1339). But, how can we explain this major occurrence of HCC in them? We emphasize that HCC, the third leading cause of cancer mortality worldwide, is a highly vascular tumor that expresses VEGF [41]. An up-to-date study provides further evidence that obesity increases HCC risk and that this factor may explain a relevant proportion of cases among subjects in absence of HBV/HCV infection, probably NASH-mediated [42]. Interestingly, patients suffering from NASH, a further expression of the metabolic syndrome, express high serum concentrations of VEGF [43]. Finally, portal VEGF was significantly higher than systemic VEGF, and expressions of VEGF and hepatocyte growth factor in the liver, spleen and intestine were also up-regulated during liver regeneration [44].

In term of clinical advantage, which take-home lesson could be drawn from our data? Physicians should try to strike a balance between pprogressive nutrition deficiencies as well as muscle wasting, universal problems in patients suffering from liver cirrhosis, and a moderately hypocaloric diet to avoid this eventual risk factor.

As it is repeatedly emphasized, PH remains a partially-clarified phenomenon and so does the collaterals presence. Now, the VEGF-dependent angiogenesis is considered being crucial in determining SRS [2]. A fascinating hypothesis suggests that portosystemic shunts may mimic an Arterio-Venous Fistula (A-VF [45]), in which the high-pressure portal blood relays with the lower pressure systemic venous circulation. Although these collaterals decompress the portal circulation, the increased cardiac output enhances portal blood flow and tends to counteract the portal hypotensive effect of the portosystemic shunt. As portal blood flow grows, collateral blood flow multiplies and is nearly totally shunted in the systemic circulation. Ultimately, high-output cardiac failure occurs leading to cirrhotic cardiomyopathy. Actually, researchers demonstrated an over-expression of VEGF in a rabbit A-VF model [46].

Coming back to the HE detection in both patients with and without SRS, what does support this finding? In contrast with the current opinion, a recent study in patients with cirrhosis has failed to find a relationship between EEG alterations and the presence per se of a patent paraumbilical vein, very common collateral with the same significance of SRS [7]. Comparing angiographically patients evidencing LSS and HE with cases characterized by PH but no HE, Takashi et al. concluded that small SRS were not associated with HE [47]. On the other hand, some patients with spontaneous shunts, who had undergone shunt reduction by a radiological approach [48], obtained an amelioration of HE. In conclusion, SRS could only provide an explanation for the refractoriness of HE. Continuing on the subject, the lack of association between liver cirrhosis severity and SRS, found in our series, is confirmed by the subsequent observations. Following-up consecutive patients with bleeding varices and LSS, after devascularization with or without esophageal transaction, the postoperative survival rate of patients with the LSS was significantly lower than that of patients without the LSS and preoperative variables concerning hepatic reserve failed to show significant predictability [16]. Other data challenge this finding [9]. Discussing the possible limitations of the present study we have to pinpoint that the sample size of patients with HCC is small, but it seems to be of sufficient power and adequate to the low (5%) incidence of this complication in liver cirrhosis [49]. Moreover, it is correct to evidence that other cutting-edge technology is more specific for the detection of portosystemic collaterals, e.g., MRI [13]. But, any crucial future research directions should consider the utility of DUS particularly when repeated measures are requested beyond liver transplantation settings [34, 35]. Last but not least consideration, are we sure that splenomegaly is the consequence of PH and not one of its causes, mainly taking into consideration the spleno-renal reflex-mediated reduction in vascular conductance [50] that exacerbates sodium and water retention in the kidneys?