Acute viral infections are the most common causes of acute liver failure (ALF) in pregnant women. During pregnancy, the prevalence and the severity of acute hepatitis due to hepatitis A virus (HAV) or hepatitis B virus are not increased.1 Pregnancy is considered a risk factor for the rare life-threatening hepatitis due to herpes simplex virus (type 1 or 2), because 23% of the reported cases have involved pregnant women.2 However, the most prevalent viral cause of ALF in pregnancy is hepatitis E virus (HEV), and HEV-associated ALF has been reported in populations most exposed to this virus in endemic, epidemic, or sporadic settings3—namely, populations living in Asia, particularly the Indian subcontinent, and in Africa, especially in Algeria, Sudan, and Darfur.
In this issue of HEPATOLOGY, Bhatia et al.,4 from India, present data on one of the largest single-center series of patients with ALF ever reported. Only patients aged 15 to 45 years were included. The study confirms that HEV-associated ALF is frequent in India and that its prevalence is significantly higher in pregnant women (59%) than in age-matched nonpregnant women (30%) and men (23%). Interestingly, the outcome of patients with HEV-associated ALF did not differ significantly in these three groups of patients and, in pregnant women, the outcome was not related to the trimester of pregnancy.4 In European countries, only isolated cases of HEV-associated ALF have been reported in indigenous nonpregnant patients without a history of travel to endemic countries5 and in women recently returned from Asia, either pregnant Asian women6 or nonpregnant European women on oral contraceptives.7
Although Bhatia et al. conclude that the mortality rate of ALF in India is not influenced by its etiology, the issue remains debatable. The overall mortality rate of ALF was similar in their three groups of age-matched patients (pregnant women, nonpregnant women, and men), but the mortality rate of men with HEV-associated ALF (37%) was much lower than in men with non–HEV-associated ALF (63%). Such a difference was not found in women whether or not they were pregnant. Moreover, in three other series from India involving between 90 and 180 patients,8–10 mortality differed according to the etiology of ALF. In these three series, the mortality rate of patients with HEV-associated ALF was comparable to that reported by Bhatia et al., ranging from 52% to 74%. In contrast, mortality rates of patients with non–HEV-associated ALF of various etiologies were either higher (89%)8 or lower (28%–33%)9, 10 than that of HEV-associated ALF. In the West, patients with ALF due to either HAV or acetaminophen overdose have a spontaneous mortality rate clearly lower than that of patients with ALF due to most other causes.11 However, HAV hepatitis was associated with only 1.2% of Bhatia et al.'s cases of ALF,4 and ALF due to acetaminophen overdose was not observed (S. K. Acharya, personal communication).
The finding by Bhatia et al. that pregnancy does not influence the outcome of women with ALF confirms the finding previously reported by Khuroo and Kamili.8 However, the conclusion that pregnancy does not affect the rates of complications of ALF in Indian women4 is questionable regarding renal failure. It is well established that, during normal pregnancy, glomerular filtration rate increases by 40% to 50% and reaches its highest value at the end of the first trimester.12 Accordingly, in pregnant women, normal serum creatinine levels decrease and rarely exceed 0.8 mg/dL (70 μmol/L) during the third trimester.12 Therefore, the finding in pregnant women and age-matched nonpregnant women and men of mean values of serum creatinine in the upper range of normal values of nonpregnant adults does not mean that glomerular filtration rate is similar in the three groups of patients. On the contrary, pregnant women in particular may well have some degree of renal dysfunction. Accordingly, in the Bhatia et al. study, the mean serum creatinine value of 0.9 mg/dL (80 μmol/L) attests a degree of renal dysfunction or failure in a majority of the pregnant women. Therefore, in the latter, the 8.4% rate of renal failure (defined by serum creatinine >1.5 mg/dL [133 μmol/L]) is likely an underestimate of the prevalence of acute kidney injury.
It is reasonable to expect that, on average, a group of patients with a high rate of renal failure would have serum bilirubin values higher than those in patients with a lower rate of renal failure. However, the mean serum bilirubin level of the pregnant women of the Bhatia et al. study was surprisingly similar to that recorded in the two other groups of age-matched patients. From this observation, one could speculate that, in these pregnant women, the degree of liver failure could have been slightly lower than that of the two other groups of age-matched patients. However, a more precise evaluation of liver function using prothrombin time prolongation, a sensitive marker of acute liver dysfunction, is not possible in each of the three groups of patients, because the distribution of prothrombin time prolongation values is provided only in surviving and nonsurviving women.4
The main issue concerning the ALF puzzle in India remains the high prevalence of HEV-associated ALF, especially in pregnant women. Variations in genotype in HEV-associated ALF in India are unlikely, because HEV genotype 1 is the only genotype found in Indian patients,3 whatever the severity of HEV-associated hepatitis. The influence of viral load was not evaluated in the Bhatia et al. study. Although hepatocytes are the main target of HEV infection, the mechanisms of HEV-associated liver injury remain unclear. Immunomediated mechanisms have been suggested on the basis of experimental infection in the cynomolgus monkey,13 whereas direct cytopathogenicity toward hepatocytes was considered from in situ hybridization and immuno-histochemical studies of liver tissue from patients who died from HEV-associated ALF.14 The role of pregnancy in the development of HEV-associated ALF is also unclear. No increase in the severity of HEV-associated liver injury was observed in pregnant rhesus monkeys after intravenous administration of HEV particles.15 Pregnant women in India have been found to have an increased attack rate of symptomatic acute hepatitis E, but they may well develop acute hepatitis E without ALF. Increased serum levels of steroid hormones and decreased cellular immunity were found in pregnant women in India with HEV-associated ALF; however, they may be a consequence of ALF, because pregnant women with HEV-associated hepatitis without ALF were not evaluated.9 Finally, the occurrence of 40% of HEV-associated ALF cases in nonpregnant patients in the Bhatia et al. study, the lack of influence of gestational duration on survival of Indian women with HEV-associated ALF,4, 8 and the expectant management of pregnancy in these women as performed by Bhatia et al. and others8 do not support the hypothesis of a key role of pregnancy and the corresponding hormonal changes in the development of HEV-associated ALF.
Any hypothesis explaining the high frequency of HEV-associated ALF in pregnancy must take into account the crucial fact that although the syndrome occurs in some African countries, it is almost exclusively found in Asia, particularly in India, Pakistan, and Bangladesh. This suggests that, in these areas where HEV infection is frequent, but not in others, a local factor exists that may promote the development of HEV-associated ALF. Theoretically, the occurrence of ALF results from either the spontaneously deleterious course of the initial acute liver injury without aggravating cofactors or the aggravation of the initial acute liver disease by various cofactors. Aggravating cofactors preceding, or associated with, HEV-associated ALF in Indian patients may include preexisting, sometimes unknown chronic liver disease14, 16 and, in anecdotal cases, acetaminophen hepatotoxicity,17 malaria,18 and typhoid fever.19 All these cofactors that could not have been missed by Bhatia et al.4 were not apparent in the current series of their patients with HEV-associated ALF, except a dual viral infection in 47 patients. Finally, inadvertent aggravation of the patient's condition may have occurred prior to hospitalization due to traditional medications used by uninformed patients to relieve symptoms. Herbal medicines are commonly used in India,20, 21 particularly during the course of liver diseases, and also in pregnant women.22 They may contain multiple herbal and nonherbal compounds23 that, because of various and unpredictable side effects,24 nephrotoxic, hepatotoxic, or sedative, may result in rapid deterioration of the condition of patients affected by acute viral hepatitis.20 Attendants of the patient interviewed at admission may not be aware of such indigenous medications ingested by the patient. Individual cases of HEV-associated ALF after ingestion of herbal drugs have been observed in India.17, 20 The hypothesis that hazardous use of herbal medicines favors the occurrence of HEV-associated ALF is consistent with both the geographical localization of the syndrome in Asia, and mainly in India, and its epidemiology in India as described by Bathia et al.
In conclusion, when compared with age-matched nonpregnant individuals, the overall prognosis of ALF in Indian pregnant women is apparently not worse. This prognosis could be even better if renal function could be better preserved or improved, and the delay between symptoms and admission shortened. A better understanding of the Indian puzzle of HEV-associated ALF will require a prospective case-control study investigating carefully the therapies used for the early symptoms of illness.