Letter: is the AHHS score really useful in clinically severe alcoholic hepatitis?

EDITORS, We have read with great interest the study from Forrest et al on the diagnostic and prognostic accuracy of liver histology in alcoholic hepatitis (AH).1 Remarkably, survival was not significantly associated with the Alcoholic Hepatitis Histological Score (AHHS). The AHHS did not demonstrate superior prognostic value when compared to bioclinical scores such as MELD (Model for Endstage Liver Disease) score or Glasgow Alcoholic Hepatitis Score (GAHS). The AHHS is a histological tool which categorises patients with AH into three groups with different shortterm mortality.2 Patients with marked neutrophil infiltration and presence of megamitochondria had more favourable outcomes as previously demonstrated.3,4 Conversely, severe fibrosis and cholestasis were associated with shortterm mortality.2 Cholestasis has also been associated with the occurrence of sepsis in AH.5 We also assessed the value of AHHS in daily practice by retrospectively evaluating liver histology of 85 patients with suspected severe AH (Maddrey's score over 32 and/or hepatic encephalopathy). All but one patient had bridging fibrosis or cirrhosis (98%). No significant difference of survival was found between AHHS grades (Table 1). No histological feature was predictive of survival or response to steroids. Cholestasis was not predictive of sepsis. However, clinical scores including MELD, ABIC, or GAHS were significantly associated with shortterm mortality. The lack of significant prognostic value of the AHHS might be explained by several factors. First, severe fibrosis counts for one third of the AHHS. Considering that a great majority of patients with AH in France has underlying severe fibrosis, the relevance of these criteria can be questioned.3 Similarly, 82% of patients had severe fibrosis in the study from Altamirano et al.2 Second, the presence of megamitochondria is difficult to assess and suffers from weak interobserver agreement.2,6 Third, neutrophil infiltration is a transient histological feature.7 Timing between suspicion of AH and biopsy is difficult to master in daily practice. The degree of neutrophil infiltration could vary by the time biopsy is performed. Last but not least, the initial AHHS was designed using histological data of patients with allseverity AH.2 It may not be surprising that its prognostic value is no longer found in more severe patients. Alternatively, the authors suggested that the AHHS could be useful for less critically ill patients (MELD <21 or ABIC grade B).2 However, these patients are those for which biopsy will not be performed, as they will not likely benefit from steroids. In patients with less severe alcoholic hepatitis, the AHHS score was not predictive of 90day mortality.8 In the study by Forrest et al, 12% of biopsies were considered as inadequate when the number of portal tracts was below five. Most patients with suspected AH undergo transjugular liver biopsy (TLB). This technique is often considered as theoretically inadequate for pathological consideration although this point remains controversial.9 Hence, one could question the accuracy of TLB to assess the diagnosis and prognosis of AH. We hope that newgeneration transjugular needles may improve the accuracy of histology in severe AH.