Limit of Fibroscan

Limit of Fibroscan


Articles in Press

Jnl of Hepatology Jan 2010


Liver stiffness is directly influenced by central venous pressure


Gunda Millonig1, Stefanie Friedrich1, Stefanie Adolf1, Hamidreza Fonouni2, Mohammad Golriz2, Arianeb Mehrabi2, Peter Stiefel1, Gudrun Pöschl1, Markus W. Büchler1, Helmut Karl Seitz1, Sebastian Mueller1

Received 28 May 2009; received in revised form 27 July 2009; accepted 30 July 2009. published online 03 December 2009.

Corrected Proof

Background & Aims


Liver stiffness (LS) as measured by transient elastography [Fibroscan] offers a novel non-invasive approach to assess liver cirrhosis. Since Fibroscan seems to be unreliable in patients with congestive heart failure, it remains to be determined whether hemodynamic changes affect LS irrespective of fibrosis.

Methods & results


Using landrace pigs, we studied the direct relationship between the central venous pressure and LS measured by Fibroscan. Clamping of the inferior caval vein increased LS from 3.1 to 27.8kPa while reopening reversed LS within 5min to almost normal values of 5.1kPa. We then studied LS as a function of venous pressure in the isolated pig liver by clamping the upper and lower caval, portal vein and hepatic artery. The stepwise increase of intravenous pressure to 36cm of water column (3.5kPa) linearly and reversibly increased LS to the upper detection limit of 75kPa. We finally measured LS in 10 patients with decompensated congestive heart failure before and after recompensation. Initial LS was elevated in all patients, in 8 of them to a degree that suggested liver cirrhosis (median 40.7kPa). Upon recompensation with a median weight loss of 3.0kg, LS decreased in all 10 patients down to a median LS of 17.8kPa. Inflammation could not account for increased LS since initial liver enzyme counts were only slightly elevated and did not change significantly.

Conclusion


LS is a direct function of central venous pressure which should be considered when assessing the degree of fibrosis.

Introduction


Transient elastography (Fibroscan) is a rapid, noninvasive, and reproducible approach to assess liver fibrosis by measuring liver stiffness (LS) [18]. LS measurements can be routinely performed in more than 95% of patients being only limited in those with severe obesity and ascites [7]. In various liver diseases such as viral hepatitis, alcoholic liver disease and primary biliary cirrhosis, LS was strongly associated with the degree of liver fibrosis [2], [5], [6], [9], [10], [15], [16]. On the basis of these studies, cut-off values have been identified to discriminate liver cirrhosis (F4) from lower fibrosis stages. A recent meta-analysis even concluded that patients could be spared from histological assessment if LS measurements exceed such cut-off values [9].


Some conditions, however, limit assessment of fibrosis by LS. Thus, LS also increases during laboratory signs of hepatitis [1], [3], [17] and cholestasis [14] independent of the degree of fibrosis. These conditions may increase LS to a degree that could even suggest liver cirrhosis (i.e. stiffness values of 12.5kPa and above). Recently, a case report indicated that assessment of fibrosis by LS may be unreliable in patients with congestive heart failure [12]. Since congestive heart failure typically increases central vein pressure, we here studied the direct relationship between venous pressure and liver stiffness in an animal model [14].

Patients and methods


Experimental animals and treatments


All experiments were approved by the local committee for Animal Welfare of the Regierungspräsidium Baden-Württemberg. We used eight German landrace pigs (22.3±2.7kg). Pre-operative preparation included fasting for 12h, allowing free access to water, a standardized narcotic protocol (pre-medication: Azaperon 8mg/kg intramuscularly, Midazolam 0.5–0.7mg/kg intramuscularly, Ketamin 5mg/kg intravenously, Atropin 1mg intravenously), which was followed by endotracheal intubation. Pressure controlled ventilation was done in a half-closed system (ventilation parameters: frequency, 11/min; volume, 300mL; room air, 1.5–2L/min; O2, 0.5–1.0L/min, N2O, 1.5–2.0L/min; Isoflurane, 0.75–1.5%). Arterial blood gases were controlled within a strict limit (pO2, 100–150mmHg, pCO2, 35–42mmHg). Pigs were anti-coagulated with Heparin 5000I.U. given intravenously. In five pigs we performed a longitudinal laparotomy, determined basal stiffness values and then clamped the inferior caval vein between the liver and the right cardiac atrium (Fig. 1, clamping position 1). Five minutes after clamping liver stiffness was again measured. The caval vein was then reopened and LS determined again after 5min. In three other pigs the portal vein, hepatic artery, the inferior caval vein distal and finally proximal of the liver were clamped, creating an in situ isolated pig liver (Fig. 1, clamping positions 1–3). We then cannulated the inferior caval vein and introduced a central venous pressure line. After determination of the basal venous pressure and the basal liver stiffness, the intravenous hydrostatic pressure was stepwise increased up to a 50cm water column by infusion of isotonic saline. LS was determined at various intravenous pressures at least three times. Interquartile range (IQR) and success rate of transient elastography in pigs were comparable to measurements in humans.

Patients with congestive heart failure


We included 10 patients who presented at Salem Medical Center between October 2008 and March 2009 with clinical signs of congestive heart failure according to the modified Framingham criteria [13], [19] and elevated NT-pro-BNP levels. Exclusion criteria were chronic alcohol abuse, a history of liver disease, severe obesity and ascites. The study protocol was approved by the local Ethical Committee of the University of Heidelberg and written informed consent of all patients was obtained prior to inclusion. Patients underwent routine blood tests, echocardiography with determination of ejection fraction and ventricular size, chest X-ray and abdominal ultrasound. Inspiratory and expiratory diameter of the inferior vena cava, body weight and LS were determined at the time of admission and after recompensation. Medical treatment of the patients included mainly diuretic therapy (furosemide, torasemide, spironolactone). In two patients, additional antihypertensive therapy (ACE-inhibitors, beta-blockers) or antiarrhythmic therapy was indicated. Patients’ characteristics are given in Table 1.

Liver stiffness measured by FS


Liver stiffness was measured by FS (Echosens, Paris, France) as described recently in detail [18]. The tip of the probe transducer was placed on the skin between the rib bones of the patient and the level of the right lobe of the liver. The measurement depth was between 25 and 65mm below the skin surface. Ten measurements were performed with success rates of at least 60%. The results were expressed in kilopascals [kPa]. The median value was taken as representative. Based on previous studies [2] and a recent meta-analysis [9], cut-off values of 8.0 and 12.5kPa were considered to be optimal for detecting F3 and F4 fibrosis stages, respectively. Since liver stiffness values ranging from 2.4 to 5.5 are considered as normal [2], a cut-off value of <5.5kPa was considered as normal although no large F0 validated control groups have been studied so far. The mean interquartile range (IQR) was 16% in all measurements. FS measurements with an IQR higher than 40% were excluded. In landrace pigs, LS measurements were performed at the right ventral subcostal area because of a better reproducibility.

Routine abdominal ultrasound


Ultrasound examination was routinely performed in patients with heart failure to measure caval vein diameter and respiration-associated fluctuation of the diameter. Liver cirrhosis, focal lesions and vascular malformations in the liver were excluded. The liver parenchyma was examined for echogenicity, homogeneity, liver surface nodularity, hypertrophy of segment I, and signs of portal hypertension. Liver cirrhosis was suspected when the ultrasound aspect of the liver surface showed nodularites [4].

Echocardiography


Comprehensive echocardiography was performed in all subjects, using a GE Vivid 7 ultrasound machine and echocardiographic images were obtained in the standard parasternal and apical views. Dimensions of the left and right ventricle were measured according to the recommendations of the American Society of Echocardiography, and a dilated caval vein was recorded.

Statistical analysis


Correlations between laboratory findings and liver stiffness in patients were calculated as bivariate regression analysis for non-parametric variables according to Spearman (regression coefficient r, r2, p). Differences were considered significant at p<0.05. The Sign rank test was used to analyze changes in weight and liver stiffness. All statistical analyses were performed with SPSS, version 12.0.1 (SPSS, Inc., Chicago, USA). For LS measurements after caval vein clamping and modulation of intravenous pressure in animals, average and standard deviations were used and calculated by Excel (Microsoft, Inc., Redmond, USA).

Results


In order to test whether an elevated intravenous pressure increases LS, the inferior caval vein between the liver and right cardiac atrium was clamped for 5min in laparotomized pigs. Initial liver stiffness values were comparable to those of humans. Complete occlusion of the caval vein was followed by visible swelling of the liver and an increase in LS from 3.9 to 27.8kPa (p<0.05) (Fig. 2). These values are far above cut-off values that are considered to reflect F4 fibrosis (liver cirrhosis) in humans. Reopening of the caval vein led to a rapid decrease in LS down to 5.1kPa within 5min (p<0.05). The reversible elevation of LS by increased venous pressure was highly reproducible in all five animals suggesting that LS is directly controlled by the intravenous pressure in the absence of fibrosis.

In a next series of experiments, an in situ isolated pig liver was generated by clamping the portal vein, hepatic artery and inferior caval vein distal and proximal to the liver. The inferior caval vein was then cannulated to measure and modulate the intravenous pressure. This model was chosen because it allows a direct correlation between hydrostatic pressure and LS. Complete occlusion of the feeding and draining veins but no change in hydrostatic pressure increased LS from 4.9 to 9.7kPa probably due to disruption of hemodynamics. We then stepwise increased the intravenous hydrostatic pressure up to a 50cm water column by infusion of isotonic saline solution and measured LS. As shown in Fig. 3, LS linearly increased with increasing intravenous pressures (r=1, p<0.01). At a 36cm water column, the maximum measurable LS of 75kPa was reached. Increased LS was completely reversible and almost reached initial levels (5.5kPa) within 5min after resetting hydrostatic pressure back to a 0cm water column. Thus, regardless of basal liver stiffness caused by the tissue matrix, LS is reversibly and directly controlled by the intravasal venous pressure up to the upper detection limit of the Fibroscan at 75kPa. The tight control of LS by venous pressure suggests that hemodynamic changes may strongly interfere with fibrosis assessment by Fibroscan.

The observations above prompted us to systematically study liver stiffness in patients with decompensated congestive heart failure. A total of 10 patients (male/female, 4/6) with cardiac insufficiency were seen during the study period. The mean age was 74±12years. Patients’ characteristics are shown in Table 1. All patients had edema of the lower extremities, six of them showed pulmonary congestion in a chest X-ray and all but two showed a dilated caval vein (>2cm, no respiration-associated fluctuation in diameter) in an echocardiography. Two patients had severely reduced left ventricular ejection fraction (LVEF<30%), five patients had slightly to moderately reduced LVEF (EF 31–55%) and three had a preserved LVEF (LVEF>55%). All of the patients had pathologically elevated NT-pro-BNP serum levels (median 5075ng/L, range 1946–21,586ng/L).


None of the patients had sonographic criteria for liver cirrhosis in abdominal ultrasound. Transaminases were only slightly elevated in five patients and γGT and AP only slightly above normal in nine patients. None of them had decreased liver synthesis activity (albumin; prothrombin time not applicable, because most patients were on oral anti-coagulants for chronic atrial fibrillation). The median initial LS was 40.7kPa (range 6.1–51.3kPa) and nine of ten patients had a LS above 12.5kPa, which is generally considered a cut-off value for F4 fibrosis. There was a statistically significant correlation between initial LS values and the presence of caval vein dilation (r=0.7, p=0.03). Thus, although no liver histology was available for ethical reasons, all patients with right heart failure but no apparent liver disease or cirrhosis had a drastically increased LS.


Over a mean hospitalization interval of 7.2days (range 5–11days) all patients clinically recovered from heart failure. They significantly lost weight (median weight loss 3.0kg range 1–16kg, p=0.008) under therapy with diuretics. In all 10 patients, successful cardiac recompensation led to a significant decrease in median liver stiffness by 15.3kPa (range 2.8–29.5kPa, p=0.004) (Fig. 4). LS even decreased in two patients below the critical cut-off value for F4 fibrosis of 12.5kPa. No significant correlations were seen with serum BNP levels. Thus, recompensation of cardiac insufficiency by diuretic therapy, concomitant weight loss and clinical resolution of edema results in a decrease of initially elevated LS.

Discussion



Based on LS measurements in landrace pigs, we here demonstrate that the central venous pressure directly controls liver stiffness in a reversible manner. Over a wide range, LS is a linear function of intravenous pressure reaching the upper detection limit of 75kPa at an intravenous pressure of a 36cm water column. We eventually show in 10 patients with decompensated congestive heart failure that LS is dramatically elevated under such pathological conditions and rapidly decreases during clinical recompensation due to diuretic therapy. Since fibrosis state cannot change within such a short period of time, these findings on patients further underline the direct dependence of LS on venous pressure.


This strong dependence of LS on venous pressure implies that liver congestion should be excluded prior to the assessment of fibrosis. The majority of patients with decompensated cardiac failure had initial LS far above the cut-off value of 12.5kPa which is generally accepted for the diagnosis of F4 fibrosis. One patient reached 51.3kPa which is even higher than the recently reported case undergoing heart transplantation [12]. Although LS decreased in all patients during therapy with diuretics it only fell below 12.5kPa in two of them while seven remained in the range of F4 fibrosis. Older age as a reason for increased LS can be excluded as a recent study by Sirli and colleagues showed [20]. Thus, increased LS could be due to the onset of cardiac fibrosis in these cases and fibrosis assessment by Fibroscan will be especially challenging in patients with cardiac insufficiency since both fibrosis and venous pressure increase LS. Further studies in patients with congestive heart failure and co-existing liver disease are needed since these patients are at increased risk for liver fibrosis. It also remains questionable in this context whether recently reported increased LS in patients with failing Fontan circulation was indeed due to cardiac liver fibrosis [8] or just elevated central venous pressure since no sequential LS measurements were performed. On a special note, LS may become a useful non-invasive tool for screening cardiac patients and identifying those that are at risk of cardiac cirrhosis since increased venous pressure (but not abnormal liver function tests) has been recognized as major risk factor of cardiac fibrosis [11]. Nevertheless, the impact of chronic compensated heart failure remains to be determined, since our study population included only decompensated patients.


The findings that increased pressure, both in hepatic veins and bile ducts [14], elevate liver stiffness could have important consequences for the (patho) physiology of fibrosis. Thus, pressure itself could be a general triggering factor for fibrosis progression. More detailed studies are required to better understand the compartment-dependent impact of pressure on liver stiffness.


In summary, we here demonstrate that LS directly depends upon venous pressure. Thus, LS values should be interpreted with caution in patients with cardiac insufficiency since the fibrosis state could be easily overestimated. The enormous impact of hemodynamic conditions on LS may also explain why it is so difficult to define reliable cut-off values for low fibrosis stages such as F1 and F2.