Adverse events with antiretrovirals

Graeme Moyle,

Several areas of interest exist regarding the management and risk of toxicity with antiretroviral (ARV) therapy. These include concerns regarding liver dysfunction, bone changes, lipid and insulin disturbances, and gastrointestinal (GI) effects. Changes in fat deposits with therapy are discussed in separate reports.

Liver disease and antiretrovirals

With the growing number of individuals co-infected with hepatitis B or C, concern regarding the potential for antiretroviral agents to trigger or worsen hepatic disease has risen. Additionally, antiretrovirals themselves may cause hepatic dysfunction. The concern regarding hepatotoxicity is greatest with some non-nucleoside reverse transcriptase inhibitors (nNRTIs), although the rare syndrome of lactic acidosis with hepatic steatosis is principally related to nucleoside analogue (NA) therapy.

Possible mechanisms through which therapy may lead to liver problems include ‘idiosyncratic’ reactions and direct injury by cytotoxins. Idiosyncratic reactions are generally of a low incidence and may result from drug allergies, such as to nNRTIs, sulphonamides or to metabolites of isoniazid. Many cytotoxic agents may be used by persons with HIV. These include non-steroidal anti-inflammatory drugs (NSAIDs), azoles (e.g., fluconazole), statins (e.g., pravastatin), anti-TB medication and recreational substances such as alcohol and ecstasy and unproven ‘therapies’ such as Chinese herbs.

Additionally, NAs and other drugs may impact mitochondrial function (such as metformin or valproate) and so may contribute to liver dysfunction through direct mechanisms. [Abstract 41]

Two presentations looked at hepatic events across multiple studies. [Abstracts 43 and 44] In evaluating hepatotoxicity and mortality events in 21 adult antiretroviral trials, involving 9,003 patients since 1991 investigators for the ACTG used a definition of hepatotoxicity as five fold or greater increases in AST, ALT or bilirubin (Grade 3-4).

Hepatotoxicity events were found in 6.3% of persons on NA alone and 6.2% in persons on triple therapy which included a protease inhibitor (PI) or an nNRTI. Events were more common (7.4%) in individuals on a single NA compared to 4.9% on double NA therapy. Amongst NAs, liver events were more common with ddI (8.7%) relative to AZT (5.5%) or d4T (3.7%). However, these differences were accounted for by studies of high doses of ddI (500-750mg/day) where the grade 3-4 events were seen in 10.3% of individuals compared with only 6.2% on the currently recommended 400mg/day dose.

Comparing PIs and nNRTIs, grade 3-4 liver function events were seen in 5% of PI and 8.2% of nNRTI patients. Presumed drug-related hepatotoxicity deaths were very rare, involving only 0.4% of individuals involved in these clinical trials. [Abstract 43]

Nevirapine hepatotoxicity

A separate analysis of Nevirapine (nevirapine) trials, including a total of over 2,700 individuals, examined both a database analysis for laboratory changes and investigator reported events. Hepatitis related events were reported in 3.4% of Nevirapine versus 2.2% of control arms, giving an attributable Nevirapine rate of hepatotoxicity of 1.2% in a population of 2,249 of individuals with a median baseline CD4+ cell count of 93/mm3.

In 456 individuals, who were ARV-naive or had very limited ARV exposure and had a median CD4+ cell count of 363/mm3, the hepatitis and related event rate was 8.1% for Nevirapine and 3.0% for placebo, giving a Nevirapine attributable rate of 5.1% over 1 year. Hepatic events attributable to Nevirapine appeared to be more common in individuals with higher CD4 counts and risk was increased for those with a raised ALT or AST at baseline and coinfection Hepatitis B or C. [Abstract 44]

This creates a degree of dilemma for Nevirapine use: its efficacy appears less good at low CD4+ cell counts but hepatotoxicity is a greater risk in higher CD4 count patients.

In the late breaker session, a Spanish group [Late Breaker Poster 22] evaluated Nevirapine blood levels and risk of hepatic events. They suggest two types of events may be involved in Nevirapine-related hepatotoxicity, one being a hypersensitivity reaction with systemic and/or cutaneous involvement, the second being a delayed hepatic reaction, related to an intrinsic toxic effect of Nevirapine.

Whilst the first reaction may most commonly occur during the first 4 to 6 weeks of Nevirapine therapy, the toxic reaction may occur at any time. Using a case control study design evaluating individuals receiving Nevirapine therapy in triple combination regimens, patients who developed ALT or AST elevation after a mean of 6.1 months from the beginning of HAART had higher median Nevirapine plasma ranges that subjects who did not develop AST elevations (6.25 mg/ml versus 5.2 mg/ml, P=0.025).

In subjects with chronic hepatitis C infection, Nevirapine plasma levels above 6 mg/ml were associated with a 92% risk of liver toxicity. In a multivariate analysis, Nevirapine plasma levels and hepatitis C seropositivity were independent respecters for AST/ALT elevation although the relative risk of hepatic injury was markedly higher with the presence of HCV infection (OR 11.7; 95% CI, 3.2 to 42.8) compared to Nevirapine (OR 1.7; 95% CI, 1.2 to 2.6)).

The difficulty with these data is knowing the cause and the effect: Were Nevirapine levels high because of the hepatitis C or as a consequence of the toxicity rather than specifically causative of the adverse events?

Hepatic and pancreatic disease

Another study evaluated hospitalisation for hepatic or pancreatic disease in ARV therapy users from the records of a US health insurer. Evaluation was made of 2,793 individuals contributing 41,304 person months of observation. 15 hepatic and 17 pancreatic hospitalisations occurred over that time, yielding event rates of 3.6 and 3.1 hospitalisations for 10,000 person months of therapy.

The authors found that exposure to ARV therapy was not specifically associated with hepatic- or pancreatic-related admissions, with no difference observed between the different ARV classes. Risk factors for hepatic admissions to the hospital included viral hepatitis and prior hospitalisation for an HIV-related event. As for pancreatic-related admissions, associations included a past history of pancreatic disease and past history for HIV-related disease or use of hepato-toxic illegal substances. [Abstract 542]

Hepatitis B and hepatitis c coinfection

Looking specifically at Hepatitis B virus (HBV) coinfection, a single-centre study of 500 individuals in Madrid evaluated elevations in transaminases in individuals under care. They evaluated 500 consecutive patients over the year 2000, of which 56 individuals were hepatitis B surface antigen (HbsAg) positive, including 10 individuals co-infected with delta virus, 10 with hepatitis C virus (HCV), and 9 with both in addition to HBV. 18 individuals developed sudden elevations in transaminase levels during the year of follow-up.

Investigator-assessed mechanisms included immune restoration syndrome, clearance of hepatitis B e antigen (HBeAg) whilst receiving lamivudine, re-appearance of HBeAg positivity, drug-related hepatotoxicity, seroconversion to hepatitis B surface antibody (HBsAb) positivity whilst receiving lamivudine, alcohol abuse or use of other non-antiretroviral hepatotoxic drugs, development of lamivudine resistance and re-establishment of symptomatic acute Hepatitis B in individuals who previously had hepatitis B core antibody (HBcAb) positive HBV DNA (observed only in severely immuno-deficient subjects. [Abstract 552]

The authors concluded that a range of mechanisms may be associated with transaminitis in individuals with HBV coinfection and they added further fuel to the discussion about whether lamivudine monotherapy should be considered in individuals with known HBV as, whilst it may in some cases be associated with sero-conversion or improved serological status, it may also be associated with a transaminase flare associated both with this sero-conversion event and with the development of lamivudine resistance.

This is particularly relevant in the light of the potential widespread availability of other drugs such as tenofovir and adefovir for use in combination antiretroviral therapy to treat HBV in persons with HIV infection.

This group went on to provide a retrospective analysis [abstract 558] of individuals commencing antiretroviral therapy between January 1997 and January 2000 who developed transaminitis (>5-fold rise about the upper limits of normal or 3.5-fold rise above baseline values in those individuals commencing therapy with abnormal transaminases). Of 222 individuals, 38% were infected with HCV, 5% HBV and 2% delta-virus. Significant transaminitis occurred in 21 (9%) of individuals, 10% were receiving PIs, 9% were receiving nNRTIs and 9% were receiving PI plus nNRTI based regimens.

In univariate and multi-variate analyses, alcohol abuse, HCV coinfection and older age were independent significant risk factors. Either PIs or nNRTIs as drug classes or individual drugs within these classes were associated with increased risk of transaminitis.

Bone diseases

The conference included very limited new information regarding concerns of bone-mineral density in persons with HIV infection, although Pablo Tebas provided a plenary overview of this problem. Although available data are conflicting, Dr. Tebas felt that available evidence supports a role for drug therapy in contributing to loss of bone mineral density in individuals.

Markers of both new bone formation and bone destruction are raised in persons on therapy implying increased bone turnover. He commented that the association with PI use remains speculative. Available evidence suggests no link with testosterone levels and that vitamin D (and its active metabolites) levels are also normal. He pointed to recent published data that found a correlation between osteopenia and osteoporosis and elevated lactate levels.

This may simply be a statistical association (rather than causative association) based on the evaluation of individuals on therapy, but also suggests an alternative hypothesis that there is a link between NA mitochondrial toxicity leading to elevations of lactate and declines in bone mineral density. As Dr. Tebas commented, given the cross-sectional nature of these studies, it is not possible to attribute cause and effect regarding these factors.

The limits of cross sectional studies are such that they cannot separate disease versus drug effects. He underlined that there are many other factors known to be associated with bone mineral density changes in adults both outside of and within the HIV setting, which include low body mass, which may be secondary to wasting or poor nutrition, past cortico-steroid usage, hormonal deficiencies and the potential for contribution from HIV-related immune system activation and the process of immune reconstitution. In addition to this there are associations in the non-HIV infected population with high blood lipids and low bone mineral density, and therefore metabolic disturbances with HIV infection may be contributory.

He commented that despite the relatively high frequency of osteopenia in cross-sectional surveys, there have been relatively few reported events of bone complications such as fracture. The exception to this appears to be avascular necrosis of the hip although it is unclear whether this is related to loss of bone mineral density or to other vascular problems in individuals with HIV or indeed the classic risk factor of corticosteroid use. He proposed that the management of HIV-related bone mineral density problems is similar to those used in sero-negative individuals, such as the use of nutritional supplements, calcium, vitamin D, exercise, appropriate hormone replacement and the use of therapies such as bisphosphonates. [Abstract 91]

Data from a prospective comparative trial in Spain [Late Breaker Poster 13] looked at bone mineral density in 244 individuals, 146 of whom had been treated for more than 6 months, 56 for less than 6 months and 39 drug-naive individuals. Assessments using DEXA scan of the lumbar spine, total and neck of femur bone mineral density were assessed at baseline and every 24 weeks.

Over the course of the study, 148 patients (60%) presented with a decline of bone mineral density with 10% achieving a diagnosis of osteoporosis. Duration of HIV infection appeared to be a significant risk factor. Although 5% of individuals who never received antiretroviral therapy had bone mineral densities consistent with osteoporosis, in the treatment groups, the rates of osteoporosis were 11% and 10%.

Whilst this suggests a role for therapy it may also be that those individuals who received therapy were sicker and so more at risk of bone mineral density abnormalities. No differences were detected between the risk of osteopenia or osteoporosis in the use of nNRTI or PI between the regimen.


Several cross-sectional studies dealt with raised random venous lactate levels. The largest cross-sectional study looked at 1,239 individuals who had been receiving ARV therapy for at least 4 months and an additional 253 individuals who had never received ARV therapy.

Of the 1,239 who had at least one lactate sample, 8.7% had a serum lactate level greater than or equal to 2.5 mmol/L with 9 (0.8%) of individuals above 5 mmol/L, a level considered severe. The problem was observed in similar frequency with male and female patients and duration of therapy was similar between those individuals with normal and raised lactate values. Overall the median lactate levels for the population of 1,239 treated patients was 1.4 mmol/L and amongst the untreated individuals was 1.1 mmol/L. 5 untreated individuals had lactate values above 2.5 mmol/L on a single occasion with one individual having elevated lactate on 2 consecutive occasions (in both cases less than 3 mmol/L).

Associations with hyperlactatemia and therapy use [Abstract 518] and biochemical abnormalities [Abstract 519] were presented. Regarding the whole population, regimens containing didanosine appeared to have an increased relative hazard of a raised lactate whereas those regimens that contained abacavir appeared to have a significantly diminished relative hazard of hyperlactatemia.

Rates of hyperlactataemia where highest in individuals receiving the combination of stavudine and didanosine and the relative hazard was significantly lower in individuals receiving the combination of abacavir plus lamivudine or stavudine plus lamivudine. These data suggest that stavudine is not a key agent in triggering lactate elevation. No significant differences between the relative hazard of hyperlactataemia on stavudine plus didanosine were observed with combinations of zidovudine plus didanosine or zidovudine plus lamivudine.

Additionally, associations with biochemical parameters were observed with hyperlactataemia. In a multivariate model, hyperlactataemia was associated with higher ALT levels and higher glucose levels although the median values were not outside of the normal range. Additionally, hyperlactataemia was associated with a wider anion gap. Individuals with an anion gap of 12-18 had a 4.9-fold greater chance of hyperlactataemia than those with an anion gap of less than 12.

Furthermore, individuals with an anion gap of greater than 18 had an 8-fold higher chance of having an elevated lactate value than those individuals with an anion gap of less than 12. These data raise the possibility that, at least for some individuals, elevated lactates may be a reflection of a shift in acid base homeostasis. However, only 5 events of lactic acidosis were reported in this population.

A second analysis looked only at individuals on first-line therapy. This population of 312 individuals was similar to the overall population studied. For individual NAs, median lactate values in this group ranged from 0.95 mmol/L in individuals receiving zalcitabine to 1.6 mmol/L in individuals receiving didanosine. 25 events of raised lactate were observed in these first line therapy patients with events being more commonly observed in didanosine-based regimens (15.5%) relative to lamivudine-based regimens (5.7%).

Event rates for zidovudine and stavudine did not differ (8% and 9.7%, respectively). In univariate analyses no NA, NA combination, demographic, or disease factor was associated with a significant risk of raised lactate.

These data are somewhat contrary to previously reported data that have suggested associations with stavudine use. However, these studies had predominantly evaluated individuals who had received multiple-treatment regimens before their assessment for hyperlactataemia, hence they had confounding factors which may have been difficult to correct.

These data, the largest cohort assessed in this way, suggest that there may have been differences between the choice of didanosine or lamivudine but not between the choice of thymidine analogues for risk of lactate elevation. However it is important to underline from this data that the median lactate values with different combinations all remained within normal limits (less than 2.5 mmol/L) and whilst episodic elevations in lactate were common lactic acidosis remained rare. It is therefore unclear what, if any, are the clinical relevance of these differences.

Diarrhoea management

Several studies evaluated the impact and management of diarrhoea on individuals with HIV infection using HAART. These studies suggested the use of HAART was associated with a diminished frequency of chronic diarrhoea in patients referred to gastro-intestinal evaluation.

In particular, opportunistic pathogens such as CMV, crytosporidosis, and mycobacterium avium complex tended to be proportionally less common in individuals on HAART evaluated for diarrhoea. [Abstract 524] In patients using nelfinavir, several studies evaluated supplementation with probiotics, fibre or l-glutamine. [Abstracts 535 and 536]

In a tiny randomised study of 20 individuals, intervention with acidophilus, soluble fibre and l-glutamine appeared to reduce the need for loperamide therapy as well as overall diarrhoea frequency. A second double-blind study in 25 subjects also found that glutamine initially improved diarrhoea severity relative to a placebo group. The dose of glutamine used in this study was 30 grams thrice daily over a ten day period and thus reflects a considerable treatment burden to manage a problem which may best be resolved by modifying therapy.


Unless otherwise noted, all abstract references in the text are to the 1st International IAS Conference on HIV Pathogenesis and Treatment. July 8-11, 2001. Buenos Aires, Argentina.


Links to other websites are current at date of posting but not maintained.