Autosis occurs in the liver of patients with severe anorexia nervosa

Potential conflict of interest: Nothing to report. Financial Support: This work was supported by the Fondation pour la Recherche Médicale (DPC20111122979) and the Agence Nationale pour la Recherche (ANR‐12‐EMMA‐0012‐03) To the Editor: We read with great interest the recent article by Liu et al., in which the authors identified a unique form of autophagic cell death, called autosis, in cultured cells and in the brain of rats with cerebral hypoxia‐ischemia lesions.1 This cell death is triggered by autophagy‐inducing peptides, hypoxia‐ischemia, and starvation and requires autophagy machinery. Autosis is genetically and biochemically distinct from apoptosis or necroptosis and is characterized by unique morphological features, including plasma membrane rupture.1 In 2008, we described a form of acute liver insufficiency associated with severe anorexia nersova (body mass index below 13 kg/m2). Despite very high serum transaminases levels suggesting hepatocyte injury, there was no significant hepatocyte necrosis on histology and very rare apoptotic cells (terminal deoxynucleotidyl transferase dUTP nick end labeling positive). Yet, electron microscopy analysis revealed that hepatocytes from these patients contained numerous autophagosomes. We also identified some hepatocytes with abnormal nuclei. Following the description of autosis by Liu et al., we reanalyzed ultrastructural features of the liver samples from the 4 patients previously reported.2 Interestingly, hepatocytes of these patients displayed morphological aspects typical of autosis (Fig. 1). In all 4 patients, we observed features of phase 1a autosis, including convoluted nucleus, moderately condensed chromatin, electron‐dense mitochondria, dilated endoplasmic reticulum, and numerous autophagosomes, autolysosomes, and empty vacuoles. Phase 1b morphological features were also present in all patients. Indeed, the perinuclear space of hepatocytes was swollen at discrete regions surrounding the inner nuclear membrane, and these swollen areas contained membrane‐bound regions with a density and granularity resembling the cytosol. However, we did not see phase 2 autosis features in our specimens. This could be owing to the fact that phase 2 is short‐lasting (15‐20 minutes), whereas phase 1 is more sustained (lasting a few hours). The probability of observing phase 2 features is thus low.Figure 1: Hepatocytes from patients with severe anorexia nervosa contain features typical of autosis. On the lower right panel, the low amount of material in the perinuclear space suggests that this cell may be in transition from phase Ib to phase 2. White arrows point to autophagic vacuoles, black arrows show dilated endoplasmic reticulum, and arrowheads indicate swollen perinuclear space. N, nucleus; M, mitochondria.Transaminases are intracellular enzymes mainly contained in hepatocytes. The fact that autosis is associated with plasma membrane permeability could account for the marked increase in serum transaminases we observed in our patients. Liu et al. found that cardiac glycosides (i.e., antagonists of Na+,K+‐ATPase) inhibited autotic cell death both in vitro and in vivo. Impaired cardiac function can occur in patients with severe anorexia nervosa and lead to heart failure.2 If autosis occurs not only in the liver, but also in the heart of these patients, we can speculate that cardiac glycosides might be particularly useful in this setting. In conclusion, our observation extends the findings by Liu et al. by demonstrating that autosis occurs in patients with a severe liver disease. This is the first observation of autosis in humans. Acknowledgment: The authors thank Adel Hammoutene and Alain Schmitt (Cochin electron microscopy core facility, Paris) for their help in electron microscopy analyses.