Serious hepatotoxicity following use of isoniazid preventive therapy in HIV patients in Eritrea

Abstract WHO information note indicates that isoniazid preventive therapy (IPT) is generally safe with little risk of hepatotoxicity. However, when the policy of IPT for HIV patients was introduced in Eritrea, frequent IPT‐associated hepatotoxicity and fatality have been reported to the Pharmacovigilance Centre. The aim of the study is to assess the causal association of IPT and hepatotoxicity and identify possible risk factors in patients on Highly Active Anti‐retroviral Therapy (HAART). This is a case series assessment of spontaneously reported cases to the Eritrean Pharmacovigilance Centre. Data extracted from VigiFlow (reported between 2014 and 2016) were exported to excel spread sheet for descriptive and qualitative analysis. Naranjo probability scale and Austin Bradford‐Hill criteria were used to assess causality. The P‐Method was used to assess preventability. A total of 31 of cases of hepatotoxicity related to IPT were retrieved. Majority (80.6%) of the cases were marked as “serious” due to life‐threatening situation (n = 15), hospitalization (n = 6), and death (n = 4). Baseline liver function test was normal in 61.3% and hepatitis B and C infections were ruled out in 77.4%. IPT was discontinued in 26 cases and reaction abated in 22 of them. Causality assessment using Austin Bradford‐Hill criteria found that the association was strong, consistent and specific with a plausible temporal relationship and biological mechanism. IPT‐associated hepatotoxicity that led to treatment interruption and death was observed in patients on HAART in Eritrea. Hence, close laboratory monitoring of patients is recommended to minimize the risk.

coinfection worsens the prognosis of HIV infection by increasing viral replication [3][4][5] and may result in subsequent immune suppression 6 and a higher risk of acquiring other, potentially lethal, opportunistic infections. 7 To mitigate this risk, WHO recommended countries to introduce isoniazid preventive therapy (IPT), INH 300 mg daily for 6 months, for people living with HIV 8 aimed at preventing and reducing incidences of TB. Isoniazid (INH) has been in use since many years as prophylaxis of TB and has been associated with small risk of fatal hepatotoxicity that ranges from 0.003% to 0.06%. [9][10][11] Liver injuries with fatal outcomes have been reported to the US-CDC 12 and other published case reports 13,14 also reported fatal hepatotoxicity associated with INH preventive therapy. Age >35 years, 15,16 alcohol consumption 11,17 and concurrent administration of other hepatotoxic agents 18 are identified as possible risk factors for the INH-induced hepatotoxicity. Following the introduction of IPT in patients taking highly active anti-retroviral therapy (HAART), some studies reported little risk of hepatotoxicity ranging from 0.07% to 1.9% without fatal outcomes. [19][20][21][22] Considering the low risks of INH-induced hepatotoxicity, neither the WHO nor the Eritrean national IPT guidelines recommend routine baseline liver function test (LFT) for IPT. 23,24 However, when the policy of IPT for PLWH was introduced and scaled up in Eritrea in 2014, frequent IPT-associated hepatotoxicity and fatality have been reported to the Eritrean Pharmacovigilance Centre. Many of the case reports showed that patients were stable under HAART for several years and their liver enzymes were normal prior to initiation of IPT. Therefore, the aim of the study is to assess the causal association of IPT and hepatotoxicity, identify possible risk factors and assess the preventability in patients on HAART.

| MATERIALS AND METHODS
This is a retrospective descriptive study of cases spontaneously reported to the Eritrean Pharmacovigilance Centre. All hepatotoxicity cases associated with IPT, reported between June 2014 (introduction of IPT) and June 2016 were retrieved from the Eritrean adverse drug reaction database and exported to excel spread sheet for descriptive and qualitative analysis. For more detail, individual case reports (yellow forms) were also retrieved based on their unique ID from the archives of the Eritrean Pharmacovigilance Centre. Cases were then summarized in an excel spreadsheet according to their age, sex, body weight, daily dose taken, type of HAART given, other concomitant/cosuspected drugs used, time to reaction onset, causality, preventability, baseline data (LFT as well as existence of hepatitis B and C), seriousness, severity, management or treatment taken, and reaction outcome for analysis.
Causality was reassessed using Naranjo probability scale 25 and then subjected to Austin Bradford-Hill criteria 26 for further case series assessment. The P-Method developed by the Moroccan Pharmacovigilance Centre in collaboration with the WHO was also used to assess preventability of the individual case safety reports. 27 The strength of the association was measured using proportional reporting ratio and seriousness of the reaction was determined according to the definition of the ICH E2A guideline 28 and severity of liver toxicity was assessed based on the US-FDA drug-induced liver toxicity grading scale 29 as follows: • Mild hepatotoxicity is defined as AST or ALT elevations of 1.1-2.5 times the upper limit normal (ULN) (<121 IU/L).
• Severe hepatotoxicity is the elevation of AST/ALT of 5.1-10 times of the ULN.
• Very severe (potentially life threatening) can be defined as the elevation of AST/ALT >10 times ULN.

| RESULTS
In the period of June 2014 to June 2016, a total of 60 hepatotoxicity cases were spontaneously reported to the Eritrean Pharmacovigilance Centre with 31 of them related to IPT in patients on HAART.
The cases were eight males and 23 females with a median age of 43 years (ranging from 25 to 67 years) ( Table 1) and median body weight of 46 kg. Liver Function Test (LFT) baseline during the initiation of IPT was normal in 19 cases and unknown in 12 cases.
Hepatitis B and C coinfections were ruled out in 24 of the cases and results were unknown in seven cases ( Table 1).
Out of the total cases, 23 received Nevirapine containing and the rest eight took Efavirenz containing HAART regimen. Patients took vitamin B6 concurrently with IPT and HAART in all the cases with no other concomitant medicines. The time to reaction onset was reported in 22 cases with a median time to onset of 61 days (ranging from 14 to 304 days) ( Table 2). All but six (80.6%) of the cases were marked as "serious." Reason for seriousness was reported as life threatening (n = 15), hospitalization (n = 6), and death (n = 4). Majority of the hepatotoxicity cases were either severe or very severe (Table 3). In 83.9% of the cases, reaction was reported as either "recovered" or "recovering" ( Table 2). In the cases with fatal outcomes, three were taking Nevirapine containing HAART and one was taking Efavirenz-based regimen.
Dechallenge information was reported in 22 of the cases and reactions abated in all the cases following withdrawal of INH. INH was subsequently reintroduced in one case and reaction reappeared with rechallenge. In the rest nine cases, dechallenge and rechallenge information was unknown. In majority of the cases (71%), the causal association between INH and hepatotoxicity was found to be "probable" and in the rest "possible." To manage the hepatotoxicity, INH was stopped in 83.9% of the cases (Table 4). Moreover, in 15 of the cases HAART was either switched (n = 12) or temporarily stopped (n = 3). In 61% of the cases, hepatotoxicity was preventable because of an apparent overdose of INH on a weight-based calculation. For the rest 12 cases, nine were nonpreventable and three were nonassessable.
The results of causality assessment with the Austin Bradford-Hill Criteria are explained in detail on Table 5.

| DISCUSSION
Our study on causality assessment using Austin Bradford-Hill criteria showed that the cases were consistent and the association was specific. Although the cases were reported from different healthcare professionals and health facilities, their clinical features were more or less similar, mainly manifested in patients greater than 35 years old. INH was the only suspected drug and HAART as well as vitamin B6 were the only concomitant drugs reported in all cases. Besides, in all but one, hepatotoxicity was the only reported reaction which shows specificity of the cases. The association seems strong as the proportional reporting ratio (PRR) was found to be 14.6. This tells us that hepatotoxicity was reported more than 14 times (among all reactions reported) for INH compared to other drugs in the Eritrean Pharmacovigilance database. PRR greater than two indicates that there is a statistical signal which, however, does not necessarily mean a true safety signal.
The normal LFT baseline results prior to the initiation of INH in majority of the cases shows that patients were stable with HAART for several years and patients developed hepatotoxicity shortly following administration of IPT. This provides a plausible temporal relationship between IPT and hepatotoxicity. It also provides an impression that the hepatotoxic effect was least likely to be attributed to HAART. Furthermore, the median time to onset of the hepatotoxic effect from the initiation of IPT was consistent with findings from previous studies. 15,33 Majority of our cases seem to be overdosed since their weight was less than or equal to 50 kg and they These voluntarily reported hepatotoxicity cases were therefore among the ADRs that caused treatment interruptions but do not reflect the exact incidence rate in the country due to under reporting. Several of the hepatotoxicity cases were found to be severe or very severe including four cases with fatal outcomes. This is not consistent with previous studies conducted on safety of INH as none of the previous publications reported fatal outcomes. [19][20][21][22] It is already known that INH-induced hepatotoxicity increases with increased age (particularly in individuals greater than 35 years old) compared to younger adults. 15,16,35 Our preliminary finding also supports the previous knowledge as above 80% of our cases were older than 35 years. To mitigate such problems, the American Thoracic Society recommends 35 "baseline and follow-up ALT tests for patients >35 years old either on monthly, bi-monthly or at 1, 3, and 6 months, depending on perceived risk and ALT stability." In our case, had there been close LFT monitoring, majority of the complicated cases including those with fatal outcomes could have been prevented.
Due to the rarity of INH-induced hepatotoxicity, routine laboratory monitoring is not recommended by the WHO. 23 It should, however, be noted that a deferred diagnosis of a fivefold and greater  (Table 2) Temporal relationship All reactions manifested after INH was administered and the median time to reaction onset was 61 days Dose-response relationship About 68% of the cases were taking 300 mg though their body weight was less than or equal to 50 kg. In 22 of the cases, reactions abated following withdrawal of INH Plausibility of the association (plausible mechanism) Through acetylation by N-acetyltransferase, hydrolysis and the Cytochrome P450 enzymes, INH can be metabolized in the liver producing acetylhadrazine and hydrazine. These metabolites are capable of participating in reactions that generate oxidative stress which subsequently cause hepatotoxicity. 30 Furthermore, Isoniazid has an inhibiting effect on CYP1A2 activity, which is suggested to be involved in hydrazine detoxification. Hence, Isoniazid can induce its own toxicity, possibly by the induction or inhibition of this enzyme. 31,32 Experimental evidence Before initiation of IPT, LFT baseline was normal in 61.3% and hepatitis B and C coinfections were ruled out in 77.4% of the cases. Positive dechallenge was reported in 71.0% of the cases with one positive rechallenge Coherence INH is well known to cause Hepatotoxicity Analogy N/A