Need for hepatitis A prevention in patients with chronic liver disease in the changing epidemiological setting of India

ABSTRACT The burden of chronic liver disease (CLD) in India is high, particularly among middle-aged men, with nearly 220,000 deaths due to cirrhosis in 2017. CLD increases the risk of infection, severe disease (e.g. hepatitis A virus or HAV superinfection, acute-on-chronic liver failure, fulminant hepatic failure), and mortality. Hence, various countries recommend HAV vaccination for CLD patients. While historic Indian studies showed high seroprevalences of protective HAV antibodies among Indian adults with CLD, the most recent ones found that nearly 7% of CLD patients were susceptible to HAV infection. Studies in healthy individuals have shown that HAV infection in childhood is decreasing in India, resulting in an increasing population of adults susceptible to HAV infection. As patients with CLD are at increased risk of severe HAV infection, now may be the time to recommend HAV vaccination among people with CLD in India.


Chronic liver disease (CLD)
Hepatitis A, B, C, D, and E virus (HAV, HBV, HCV, HDV, and HEV) infections and alcohol consumption can cause liver damage, as can obesity, which can result in nonalcoholic fatty liver disease (NAFLD). CLD (disease that has lasted for ≥6 months) is a progressive deterioration in liver function, which can lead to fibrosis and cirrhosis. 1 Cirrhosis often starts asymptomatically ("compensated cirrhosis"), but can ultimately progress to "decompensated cirrhosis", during which complications of liver dysfunction and portal hypertension manifest (e.g. ascites, jaundice, variceal bleeding). 2 Once a patient has decompensated cirrhosis, their survival will likely only be around 3-5 years. 3 In 2017, there were an estimated 1.5 billion cases of cirrhosis and other CLDs globally. 4 Liver cirrhosis was the 11 th and 26 th leading cause of disabilityadjusted life years in men and women, respectively; 5 the 13 th leading cause of life years lost; 6 and, along with other CLDs, resulted in over 1.3 million deaths in 2017. 6 Early treatment of patients with CLD is important. The goals of treatment are to stop disease progression (generally by managing the underlying cause, e.g. antivirals, alcohol abstinence) and to manage complications (e.g. portal hypertension, hepatorenal syndrome, bleeding esophageal varices, hepatic encephalopathy, and hepatocellular carcinoma). 1 Ultimately, patients may require a liver transplant, which is the second commonest major organ transplantation. 7

Increased risk of severe infection
Patients with cirrhosis have a compromised immune system and are known to be at increased risk of bacterial infection, [8][9][10][11][12] and those who become infected have a nearly 4-fold higher risk of death compared with uninfected people with cirrhosis. 2 Given the effect of cirrhosis on the immune response, such patients may also be at increased risk of HAV infection. Although we could not find any confirmation of this, patients with CLD certainly appear to be at increased risk of developing more severe HAV disease if they have superimposed HAV disease. [13][14][15] For example, in an outbreak of >300,000 HAV cases in China, mortality was 5.6-fold higher among those with HAV infection superimposed on underlying HBV infection than in those with HAV but without HBV. 15 Acute HAV infection in patients with CLD can also result in acute-onchronic liver failure (ACLF), which is associated with high rates of mortality. 16 Patients with CLD are also at increased risk of developing fulminant hepatitis, 13,17 also known as fulminant hepatic failure (FHF). In an Italian study, 595 adults (29.1 ± 9.8 years) with chronic HBV or HCV (without HAV antibodies) were enrolled during 1990-1997. 18 Of these, 27 (4.5%) acquired HAV superinfection (10/163 of those with chronic HBV and 17/432 of those with chronic HCV). FHF developed in 0/10 chronic HBV patients and 7/17 chronic HCV patients, 6/7 of whom died. None of 191 controls (without CLD) who presented with acute HAV developed FHF. 18 While this study implies that patients with chronic HBV are not at risk of FHF after HAV superinfection, results from a small Canadian study show that those with chronic HBV can have FHF after HAV superinfection. In the Canadian study, 4/60 cases of FHF during 1991-1997 were due to HAV. 19 Three of these patients had CLD (2 chronic HBV infection; 1 alcoholic cirrhosis), and all 3 died (13-35 days after admission); the patient without CLD survived. 19

HAV vaccination recommendations in patients with CLD
The United States (US) Advisory Committee on Immunization Practices recommends a 2-dose series of HAV or a 3-dose series of HAV+HBV vaccinations for all patients with CLD, including those with HBV, HCV, cirrhosis, NAFLD, alcoholic liver disease, autoimmune hepatitis, or alanine aminotransferase or aspartate aminotransferase level >2 the upper limit of normal. 20 Similarly, in the United Kingdom (UK), patients with various chronic liver conditions are recommended to receive HAV vaccination. 21 Two types of HAV vaccine are available 22 -live attenuated and inactivated -of which only the latter is appropriate for immunocompromised patients such as those with CLD. The World Health Organization (WHO) has endorsed that inactivated HAV vaccines are well tolerated by patients with mild-to -moderate CLD. 17 It is recommended that HAV vaccination should be given as early as possible after CLD diagnosis for maximum efficacy and safety. 13,14

Burden and changing etiology of CLD
In a multicenter prospective study conducted in different parts of India, 1.3% of nearly 21 million patients who attended 11 hospitals during February 2010 to January 2013 had liver disease. 23 One quarter of these patients had a new diagnosis of liver disease, of whom 19.8% had CLD. 23 Among these 13,014 patients with newly diagnosed CLD (of whom 4413 [33.9%] had decompensated cirrhosis), mean age was 42.8 ± 14.4 years and the majority (73.0%) were male. 23 The main etiologies were related to hepatitis viruses (54.9%), alcohol (17.3%), and NAFLD (12.8%). 23 However, etiology varied widely by region, with HCV being the most common in the North (44.9%), HBV in the East (47.9%) and South (40.5%), alcohol in the North-East (31.9%), and NAFLD in the Central region (43.6%) and the West (39.6%). 23 CLD etiologies reported in other studies have varied widely, 24-34 likely due to variations by region, population, and over time. The latter has been shown in a study in a tertiary care referral hospital in Eastern India, where etiologies of CLD changed substantially from 2003 to 2011, with alcohol increasing from 22.5% to 42.0% (p = .01), cryptogenic (i.e. unknown cause) decreasing from 44.9% to 25.0% (p = .001), but no significant changes in HBV (mean 22.3%) or HCV (mean 10.9%). 35 More recent studies have indicated that NAFLD could be becoming a major cause of CLD in India, with huge numbers of people potentially affected. For example, a study published in 2016 found that 30.7% of adults aged ≥35 years in a rural community in North India had NAFLD on ultrasonography, 36 while one published in 2019 reported that 528 (53.5%) of male blood donors (mean age 31 ± 8 years for males and 45 ± 8 for females) in an urban community in North India had NAFLD on ultrasonography. 37 Recent meta-analyses have estimated seroprevalences of HBV surface antigen (HBsAg) and anti-HCV to be 1.46% 38 and 0.44-0.88%, 39 respectively in India. Based on a population of approximately 1.38 billion, 40 this would equate to approximately 20 million people in India having chronic HBV infection and around 6-12 million having chronic HCV infection, meaning that large numbers of people are potentially at risk for FHF, which has a very high mortality rate among patients with CLD. 18,19 However, these numbers are dwarfed by the potential number of people with NAFLD, which, based on the two abovementioned studies 36,37 and the adult Indian population, 40 could equate to hundreds of millions of adults with NAFLD in India. While we were unable to find data on the prevalence of alcoholic liver disease in India, given that 18% of liver-related deaths in India were due to alcohol, 41 there are likely also many millions of people with alcoholic liver disease in India.

Susceptibility of Indian CLD patients to HAV
Nine out of ten old studies from India (carried out up to 2007) [25][26][27][28]30,31,[46][47][48] found that nearly all patients with CLD/cirrhosis (93.2-99.0%) had evidence of past infection with HAV (as shown by HAV-immunoglobulin G or HAV-IgG 49 or anti-HAV antibodies), as did most healthy controls (71.2-100%) ( Table 1). The study by Khanna et al., 50 however, reported a much lower rate of HAV-IgG among patients with cirrhosis (60.6%), possibly because they only included patients from the upper middle or upper socioeconomic classes. All of their seronegative CLD patients were vaccinated against HAV. 50 The authors of most of the other studies suggested that CLD patients did not routinely require HAV vaccination (as most were already immune), 25,27,28,30,31,[46][47][48] while opinions on testing for HAV antibodies before potential vaccination were mixed (see Table 1). 25,31,48 However, most of these studies are old, only including patients until 2007 at the latest and, as will be discussed further below, the epidemiology of HAV is changing in India, with declining HAV infection during childhood and subsequent increasing susceptibility in adulthood. It should also be noted that, in the two latest studies in Table 1, 25,48 which included patients during the mid 2000s, nearly 7% of CLD patients did not have anti-HAV/anti-IgG antibodies and were therefore susceptible to HAV infection. As suggested by Radha Krishna et al. in 2009 32 -based on their study that included 21 adults with ACLF due to HAV -this advice may now be outdated. In a more recent study (2011-2014), 21.4% of ACLF cases were due to HAV, HBV, or HEV superinfection, but unfortunately, the authors did not report results separately for HAV. 29

Changing HAV endemicity in India
If HAV is encountered during early childhood, the resultant hepatitis is generally mild, causing no symptoms or nonspecific symptoms (e.g. fever, malaise, diarrhea) 51 and providing longterm immunity against HAV. 52 However, if HAV is encountered for the first time in adulthood, most people will have symptoms (e.g. jaundice, pain), and it is associated with a mortality rate of around 1%. 51 HAV is also significantly more likely to result in more severe disease with increasing age. 53 Historically, many people in India were exposed to HAV during childhood, resulting in life-long protection. 52 However, with improved sanitation and hygiene, children are becoming less likely to be exposed to HAV, resulting in increasing number of adolescents and adults who are at risk of infection, and a paradoxical increase in morbidity and mortality due to HAV. 51,52,54,55 Thus, and taking into account the high heterogeneity across the Indian continent, India is now thought to be shifting from high to intermediate HAV endemicity. 51,54 This situation is particularly challenging, as in high HAV endemicity areas, most children are exposed, resulting in mild disease and lifelong immunity, while in low endemicity areas, the chance of exposure in adulthood is low. 54 However, with intermediate HAV endemicity, the chance of childhood exposure is reduced, leaving more adults at risk of more severe disease. 54

Declining HAV immunity in India
Various serological studies have reported that the proportion of healthy Indian people with seroprotective anti-HAV antibodies (i.e. previous HAV infection) has fallen over time. 26,[56][57][58][59] For example, Arankalle et al. 56 reported that anti-HAV positivity decreased significantly from 1982 to 1998 among children from urban high socioeconomic populations (age 6-10 years: ~86% to 30.9%; age 11-15 years: ~94% to 46.9%; combined age p < .00001), but not in adults or urban lower middle socioeconomic populations. Das et al. 57 reported that HAV-IgG seropositivity fell from 98.0% in 1982 60 to 54.1% in 1998 among those aged 15-24 years and from 98.6% 60 to 58.7% among those aged 25-34 years (both p < .05). Hussain et al. 26 reported that 71.2% of healthy subjects were positive for HAV-IgG in 1999-2003, much lower than the 94.8% reported in subjects in 1982 in an earlier study. 60 Gadgil et al. 58 59 reported that, while HAV seropositivity decreased from 1998 to 2017 among low/middle socioeconomic children and younger adults, it increased during the same time period among high socioeconomic children and adults ( Figure 1A). 59 This was likely due to HAV vaccination in the high socioeconomic population, although the vaccination status of the participants was not available. Figure 1A also shows that, in 1998, low/middle socioeconomic populations had considerably higher seropositivity (i.e. were much more likely to have had previous HAV infection) than high socioeconomic populations of the same age group, but by 2017, there was very little difference in seropositivity between the two populations. 59 Deoshatwar et al. 61 have also reported results from the same region (for select age groups) for 2017 and Table 1. These studies did not report study dates, so these dates are based on "received" and "accepted" publication dates. compared them with the same 1998 data as Arankalle et al. 59 Figure 1B shows that the changes in seropositivity were less pronounced in the latter study. Arankalle et al. 59 also reported that 90-95% of 3-month-old infants in both 1995 and 2017 were seropositive for HAV, likely due to maternal antibodies. In 1995, seropositivity fell to 13.6% by age 9 months and then increased to 41.0% by age 15 months, which must have been due to natural infection as none were vaccinated. However, in 2017, seropositivity fell to only 2.2% among unvaccinated infants at age 15 months. 59 These studies all support a decrease in natural HAV infection during childhood, resulting in an increase in the number of susceptible adults.

Increasing HAV infection in adulthood
In line with declining HAV seroprotection, some studies have shown an increase in the proportion of acute viral hepatitis cases that are due to HAV over time. 26,50 Hussain et al. 26 studied 1932 patients with acute viral hepatitis at a tertiary care center in Northern India, of whom 11.4% overall were HAV-IgM positive (indicating current infection). This increased from 3.4% in 1999 to 12.3% in 2003 among adults (Hussain line in Figure 2; p < .004); and from 10.6% to 22.0% in children (p < .003). At another tertiary care center in Northern India, Khanna et al. 50 reported increasing proportions of acute hepatitis due to HAV from 1999 to 2004 among patients with acute hepatitis aged 13-20 years (27.2% to 61.5%; p = .008), 21-30 years (13.3% to 39.5%; Khanna line in Figure 2; p = .031), and >30 years (0% to 17.3%; p = .06).
Various other Indian studies have reported on the seroprevalence of HAV-IgM antibodies among those with acute viral hepatitis, but have reported no change over time (Acharya line in Figure 2), 62 or have not studied their evolution over time [63][64][65][66][67][68][69] (see Table 2 and single points in Figure 2). While comparisons between studies (particularly single-center studies) should be undertaken with caution, as seropositivity varies widely by socioeconomic status, age, HAV vaccination rates, region, setting, and local outbreaks, there appears to be a slight upward trend in the proportion of adults with acute viral hepatitis who have acute HAV infection (Figure 2).
The seroprevalence among children varied widely by study, from 16.2% 26 to 72.2%, 50,66 with little correlation over time. This may relate to the socioeconomic status of the studied populations (which, as shown in Figure 1, used to have a large impact on seroprevalence, but nowadays, has much less impact), but most studies did not describe this parameter.

Indian HAV immunization recommendations
HAV vaccination is not included in the routine childhood immunization schedule in India. 70 However, the Advisory Committee on Vaccines & Immunization Practices (ACVIP) of the Indian Academy of Pediatrics (IAP) recommends HAV vaccination for all infants, as a single dose of live attenuated vaccine at 12 months or 2 doses of inactivated vaccine at 12 and 18 months of age, 51 which can be administered in a private setting paid for by the parents. 22 The IAP particularly recommends HAV vaccination for various risk groups, including children with CLD and those who are carriers of HBV and HCV. 51 Although the recommendation to vaccinate patients with CLD against HAV has been endorsed by the WHO, 17 the Indian National Centre for Disease Control (NCDC) does not currently recommend HAV vaccination for adults with CLD in India, as "most adults have already been exposed to and are thus protected". 71 This recommendation is supported by 9/10 old studies from India (carried out up to 2007) [25][26][27][28]30,31,[46][47][48] (Table 1). However, in the current context of changing endemicity it is very unlikely to hold true and therefore we feel that this should now be reexamined.
Indian associations and scientific society recommendations relating to HAV vaccination are detailed in Table 3. [72][73][74] The Association of Physicians of India (API) and the Indian Society of Nephrology (ISN) both indicate that patients with CLD who are not immune to HAV, those with other hepatitis virus infections, and patients awaiting or having received a liver transplant are at risk of HAV infection, but do not specifically recommend vaccination. 72,73 The ISN, however, says that HAV vaccination "is indicated for all transplant candidates with CLD or those patients of end-stage renal disease who have   62 Khanna et al. 2006 (middle adult age group used), 50 Hussain et al. 2006, 26 Kumar et al. 2007 (mainly adults), 63 Irshad et al. 2010, 64 67 Further information on these studies can be found in Table 2. HAV, hepatitis A virus; IgM, immunoglobulin M.

Authors' recommendations
Based on the currently available evidence of shifting endemicity and increasing HAV susceptibility in adulthood in India, now may be the time to revisit the existing NCDC recommendation that HAV vaccination is not necessary for those with CLD in India. 71 Instead, we propose that a recommendation for HAV vaccination of adults with CLD should be considered in India, as is already the case in the US, 20 the UK, 21 and Sri Lanka 75 (a near neighbor of India), and also for children with CLD in India. 51 While some Indian medical association/society guidelines recognize that seronegative patients with CLD are at increased risk of HAV infection, they do not clearly recommend HAV vaccination. 72,73 Up-to-date serological studies among Indian patients with CLD would be beneficial to confirm whether seroprotective HAV antibodies have decreased over time in these patients, in line with what has been shown in healthy people. 26,[56][57][58][59]61 However, awaiting the results of such studies should not be a prerequisite for recommending HAV vaccination among patients with CLD in India.
A more targeted approach, with serological testing prior to HAV vaccination, could be a more cost-effective option than universal HAV vaccination of patients with CLD. 25 However, given the limited facilities for serological testing, the associated cost, and the potential for missed opportunity for vaccination if patients do not return after serological testing, this should also not be a prerequisite.

Limitations
This was not a systematic review, so although we included all relevant manuscripts that we could find on PubMed and Embase, there may have been some manuscripts (e.g. those published in Indian journals that are not listed on PubMed or Embase) that we did not manage to capture. Also, India is a large country with high socioeconomic status heterogeneity.
As seropositivity rates vary considerably with socioeconomic status, age, HAV vaccination rates, region, setting, and local outbreaks, comparisons between studies from different time periods should be undertaken with caution. Futher, as already discussed, the studies that have assessed the HAV susceptibility of CLD patients are old and, while it is likely that susceptibility among these patients has increased as it has among general adults, this should be confirmed.

Summary and conclusions
The burden of CLD in India is high, resulting in high morbidity and mortality. 42 Patients with CLD are at increased risk of severe HAV disease [13][14][15]17 and ACLF, which has a very high mortality rate. 16,29,32,44,45 Hence, such patients are recommended to receive various vaccinations, including HAV vaccination, in the US, 20 the UK, 21 and Sri Lanka. 75 Old studies from India showed a high seroprevalence of protective HAV antibodies among Indian adults with CLD, [25][26][27][28]30,31,[46][47][48] although the most recent ones (≤2007) found that nearly 7% of CLD patients did not have protective HAV antibodies and were therefore susceptible to HAV infection. Studies in healthy individuals have shown that HAV infection in childhood is decreasing in India, [56][57][58][59]76 resulting in an increasing population of adults without protective antibodies, and a higher risk of HAV infection in adulthood. 26,50 This is likely also the case among patients with CLD.
Based on seroprovalence data, 38,39 millions of people in India likely have chronic HBV or HCV infection. Even more adults could have NAFLD 36,37 and, along with an increasing amount of alcoholic liver disease in India, 35 this equates to a huge population of people with chronic hepatitis infection and/or CLD. Such people are at higher risk of severe disease 13,17 (HAV superinfection, ACLF, FHF) and increased mortality. 15,18,19 It may, therefore, now be time to reexamine the existing Indian recommendations. [71][72][73][74] Patients with CLD who do not have HAV antibodies should receive HAV vaccination. This approach could reduce morbidity, mortality, and healthcare costs of HAV infection among patients with CLD. 77 In situations where antibody testing is not available or practical, CLD patients should not be excluded from HAV vaccination. Figure 3 elaborates on the findings in a form that could be shared with patients by healthcare professionals.