The relationship between liver function and portal pressure: what comes first, the chicken or the egg?
Article Outline
To the Editor:
The paper by Villanueva et al. [1] is a very important paper in clinical Hepatology, having implications for the management of cirrhotic patients. It documents a reduced probability in developing complications of cirrhosis (in addition to less variceal bleeding) in patients who achieve a target reduction in portal pressure with pharmacological therapy. However, their findings may not be generalisable as patients were very selected. The exclusion criteria removed cirrhotics who might be considered at most risk of developing complications: (1) those in whom medical therapy failed to control the index bleeding episode, (2) those with Child-Pugh score above 12 points, (3) those who had bled previously and received any therapy to prevent rebleeding and (4) those with hepatocellular carcinoma and (5) if death was expected within 6 months. All had to have at least two haemodynamic measurements between 1–4 months apart, so that 132 patients were selected (representing well compensated cirrhotics whose acute bleeding resolved with medical therapy).
The 64 haemodynamic responders (48%) were defined as previously: an absolute HVPG reduction <12
mmHg (n=14) or a 20% or more HVPG reduction from baseline (n=60) also seen in 10 with HVPG<12mmHg. The Child-Pugh score at the second HVPG measurement had improved significantly more in the responders (mean 7.8–5.9mmHg responders and 8.2–6.8mmHg non-responders). This difference was also seen in the 28 responders and 40 non-responders who had repeat HVPG measurement between 11 and 29 months. The worse liver function in non-responders accounts for a significantly higher transplantation rate, although whether non-responders were transplanted for repeated variceal rebleeding is not stated.
Nevertheless, a multivariate Cox model evaluating mortality (one assumes the time of transplantation was censored and that the starting point for evaluation was the index bleeding episode) found non-responder status to be independently predictive of death, i.e. a separate effect from the degree of liver dysfunction at baseline. This relates to survival after the second HVPG measurement, but the Cox model was based on characteristics around the index bleeding episode. It would be interesting to evaluate if the rate of deterioration of liver function overtime (e.g., change in Pugh score over 6-month intervals) influenced HVPG response.
Indeed the patient's individual HVPG response could be evaluated against the change in Child-Pugh score (or some of its components) at the time of the subsequent HVPG measurements. In addition, abstention in the intervals between measurements needs to be looked at, given that abstention was more frequent in responders (93% versus 74%) amongst the 84 in whom an alcohol history was available, and that alcohol consumption also had the strongest independent association with a haemodynamic responder status (OR 14.9, 95% 1.8–98), and 6 of the 7 late responders (i.e., at the third HVPG measurement) were abstainers. An improvement of liver function with abstention can be presumed and its relationship to HVPG is important [2].
Lastly, the potential clinical value of a haemodynamic non-response is masked by the fact that 12 patients in the non-responder group, but none in the responder group (is this pure chance?), took no drugs (n=2), or could not or did not take nadolol (n=6), or isosorbide mononitrate (n=4). Another 11 in the non-responder group stopped either nadolol (n=5) or ISMN (n=6) because of complications of therapy. If all of these did develop a complication of cirrhosis, the association between administration of drugs reducing HVPG, and the prevention of complications would be far stronger. The evaluation between responders and non-responders should be repeated excluding the 12 patients not taking pharmacological therapy to assess the strength of the association.
Thus, the key question which the authors could answer by further evaluating their data, is whether for a similar worsening (or improvement) of liver function, there is a difference in either the onset or severity of cirrhotic complications, with respect to taking pharmacological therapy and the reduced HVPG thresholds reached.
If non-selective beta-blockers with or without isosorbide mononitrate can modify the progression of cirrhosis, despite worsening of liver function, then these drugs would become the ‘aspirin’ of clinical Hepatology—cheap, few contraindications, relatively little intolerance and of universal application in cirrhotics irrespective of the presence of varices or history of bleeding.
References
PII: S0168-8278(04)00359-9
doi:10.1016/j.jhep.2004.07.023
© 2004 European Association for the Study of the Liver. Published by Elsevier Inc. All rights reserved.
