Several studies have suggested that highly active antiretroviral therapy (HAART) may cause a rise in alanine transaminase (ALT) levels.1-10 However ALT elevations (flares) range in both severity and duration, and the definition of an ALT flare is varied. Some definitions include new ALT elevations that persist, whereas others refer only to transient elevations that return to baseline or normal levels. Some are based on a single ALT measurement >50 IU11 and others use the definition of grade 3/4 hepatotoxicity from the AIDS Clinical Trial Group (ACTG).12 The extent to which such increases in ALT level are due to immune reconstitution, nontreatment related events, or random fluctuations has not been explored.
Chronic infection with hepatitis C virus (HCV) or hepatitis B virus (HBV) is an independent risk factor for liver damage in HIV-infected patients.1,13-16 However, there is no consensus on the impact of hepatitis status, HAART use, and CD4 count in determining the risk of ALT flares.
By investigating the frequency and time course of ALT elevations in untreated HIV-infected patients without hepatitis coinfection, we derive a definition of an ALT flare, which may be useful in understanding the causes and significance of ALT elevations among the HIV-treated population. We then describe associations with an ALT flare and compare these with those identified with a commonly used definition (ACTG grade 3 hepatotoxicity).2,4,5,14,17,18
The source of data for our analyses was the UK Collaborative HIV Cohort Study. This is an observational cohort of HIV-infected individuals attending some of the largest HIV clinical centers in the United Kingdom (see Appendix).19 Data collected include information on patient demographics, antiretroviral history, laboratory findings, AIDS-defining events, and deaths.
Using data from untreated HIV-infected patients, not known to be infected with HBV/HCV and with at least 1 ALT measurement, we compared the frequency and duration of ALT elevation using a series of different thresholds (100, 120, 140, 160, 180, and 200 IU). For each individual, an episode of ALT elevation started when ALT levels first increased above the threshold of interest (with the start date estimated by linear interpolation) and ended when ALT levels fell below the threshold (end date estimated as before). If ALT levels did not fall below the threshold, the end date of the episode was recorded as the last ALT measurement. The proportion of ALT elevation episodes that were sustained for 2 or more consecutive ALT measurements was described.
As a further summary measure, patient follow up was split into 3-month intervals and the mean ALT measurement in each 3-month interval was calculated. The number of patients with at least 1 mean 3-monthly ALT measurement above each threshold was summarized.
In the above analyses, only patients who were HBV and HCV negative or who had unknown hepatitis status at the time of their ALT measurements were included. A subset of these patients subsequently tested positive for hepatitis, so sensitivity analyses were also performed in which we additionally excluded any patients who had ever had a positive HBV/HCV test and then further excluded patients who had not been tested. This was to ensure that the definition of an ALT flare was based solely on HBV/HCV-negative patients; conclusions were unchanged from our primary analyses (results omitted).
Poisson regression was used to identify risk factors associated with an ALT flare based on our proposed definition. In these analyses, all patients regardless of treatment and hepatitis status were included. Follow-up was calculated from the date of the first ALT measurement after 1996 and censored at the date of death or the last clinic visit. Factors considered included demographics (age, sex, ethnicity, and exposure group), CD4 count, viral load, prior AIDS diagnosis, treatment regimen [didanosine/stavudine (ddI/d4T) with nevirapine (NVP), ddI/d4T without NVP, NVP without ddI/d4T, and “other”], any ritonavir (RTV) use, hepatitis status, and number of prior ALT measurements. All time-varying variables were incorporated as time-updated covariates and were updated in the analysis at the time when a new measurement (or treatment/clinical record) became available. The above regimens were selected as the drugs included have been reported to be associated with ALT flares.2,6,10 The number of prior ALT measurements has also been associated with ALT flares7; this covariate may be a surrogate for the frequency of attendance at a clinical center. In these analyses, patients whose first HCV test was positive were defined as being HCV positive from the start of their follow-up. Interactions between hepatitis status and both treatment and CD4 count were explored and analyses were stratified by these variables where appropriate. Finally, the association between ALT flares (fitted as a time-updated covariate) and all-cause mortality was explored using Kaplan-Meier analyses and Cox regression.
Thirteen thousand six hundred and one patients had ≥1 ALT measurement after January 1996, of whom 8375 (61.6%) had measurements while treatment-naive and HBV/HCV-negative (88%) or of unknown status (12%). The median duration between consecutive ALT measurements was 56 (interquartile ratio: 20-100) days. Table 1 shows the number of patients who were treatment naive and HBV/HCV negative or of unknown status and had at least 1 ALT measurement above each threshold, as well as the frequency and duration of the episodes. The percentage of patients with at least 1 ALT measurement above the threshold declined from 12.9% when using a threshold of 100 IU to 4.3% when using a threshold of 200 IU. Of those with at least 1 elevated ALT, the percentage of patients with 2 consecutive measurements above each threshold was similar for all thresholds at around 40%, although the median duration of episodes decreased as the ALT threshold increased. The proportion of patients with at least one 3-monthly mean ALT value above each threshold also decreased from 6.4% when using a threshold of 100 IU to 1.8% when using a threshold of 200 IU.
Although 363 (4.3%) treatment-naive HBV/HCV-negative/unknown patients had at least 1 ALT measurement >200 IU, only 172 (2.1% of the total group) of these patients had 2 consecutive ALT measurements above this level. As the lower quartile for the duration of elevations was 14 days, regardless of the threshold, we defined an individual as having an ALT “flare” if she/he had 2 consecutive ALT measurements >200 IU measured >14 days apart. A flare was considered to be resolved (from which point a person could experience a subsequent flare) when the ALT level fell to <150 IU.
When considering all patients including those exposed to HAART with at least 2 ALT measurements, 526/12206 patients (4.3%) had at least 1 ALT flare using this definition and 72 of these patients (13.7%) had at least 2 ALT flares. Patients who experienced ALT flares were more likely to be male (92% vs. 83%, P < 0.0001), of other risk group (78% vs. 68%, P < 0.0001), and of other ethnicity (83% vs. 71%, P < 0.0001). Patients experiencing ALT flares also had a shorter median duration between ALT measurements (1.5 vs. 2.2 months, P < 0.0001). In total, patients contributed 615 episodes of ALT elevation to the analysis (Table 2). Each episode lasted just over 4 months, and the median peak ALT measurement recorded during the episode was 526 IU.
The crude rates of ALT flares stratified by hepatitis and treatment status are shown in Table 3. The overall rate of ALT flares was 1.19 (95% confidence interval: 1.10 to 1.28) per 100 person-years. Patients coinfected with HBV and/or HCV had higher flare rates than those not coinfected, although the rate of ALT flares was similar among the treatment-experienced and treatment-naive patients after stratifying by HBV/HCV status.
The number of person-years and events for each potential confounder together with adjusted rate ratios are shown in Table 4. Patients with lower CD4 counts, detectable viral loads, prior AIDS diagnosis, those under follow-up in earlier calendar years, and those with hepatitis coinfection were at a significantly increased risk of experiencing ALT flares. Patients currently taking NVP (either with ddI/d4T or without ddI/d4T) and those taking RTV were also more likely to experience ALT flares. Those of heterosexual risk group, of black African ethnicity, and patients who had been on treatment for longer periods were less likely to experience ALT flares.
A series of sensitivity analyses were performed in which we investigated interactions between current CD4 count and treatment status (currently on vs. currently off treatment) and between hepatitis status and treatment status. Although lower CD4 count was a stronger predictor of ALT flares among patients off treatment [CD4 < 50 cells/mm3: 2.01 (1.05 to 3.86) compared with CD4 > 500 cells/mm3] than among those on treatment [1.33 (0.86 to 2.06)], this interaction was not statistically significant (P = 0.55). Positive HCV status was strongly associated with ALT flares among patients who were both on [6.44 (4.99 to 8.31)] and off treatment [5.08 (3.43 to 7.94)] (P value for interaction test =0.15). Results were similar for hepatitis B status.
Finally, we compared our definition of an ALT flare with a definition commonly found in the literature: a single ALT > 200 IU. Using this single ALT definition, the rate of ALT flares was considerably higher (4.4/100 person-years compared with 1.2), with stronger associations between CD4 count and ALT flares [adjusted rate ratio =2.83 (2.39 to 3.34) vs. 1.57 (1.10 to 2.25) for CD4 < 50 cells/mm3] but weaker associations with viral load, type of regimen, and hepatitis status (Table 4).
In total, 1050 patients (8.6%) died during follow-up. Using our proposed definition of an ALT flare, we found no difference between mortality rates among those who did and did not have a flare (P = 0.84). Six percent of those with a flare died within 3 years of their first ALT measurement compared with 7% in those without a flare; the rates were 10% and 9%, respectively, by 5 years. No association was found between ALT flares and all-cause mortality [hazard ratio: 1.24 (0.64 to 2.43), P = 0.52] in multivariable analyses. In contrast, when using a single ALT > 200 IU definition, there was a significant difference in mortality between those with and without ALT flares (P < 0.0001) with 3 (5)-year mortality rates of 9% (13%) in those with a flare and 7% (9%) in those without a flare. This association remained significant in multivariable analyses [1.73 (1.28 to 2.34), P < 0.0001].
Although 1 ALT measurement above 200 IU is frequently used as an indication of an ALT rise in the HIV-treated population, we have shown that 4.3% of untreated HBV/HCV-negative patients also experience single ALT values above this limit. Furthermore, patients may experience such rises on multiple occasions, suggesting that rather than reflecting a genuine adverse effect of treatment, many of these episodes may be due to another cause or may be random ALT fluctuations, which may not warrant the same level of clinical concern. The apparent strong association of CD4 count with the standard definition of an ALT flare (1 ALT > 200 IU) may also be contributed to by the frequency of monitoring of patients with high ALT values. Our definition of an ALT flare of 2 consecutive measurements >200 IU, 2 weeks apart, is based on analyses of the untreated HBV/HCV-negative population. This more stringent definition may prove more useful in studying the causes and significance of liver function test abnormalities. It also mirrors clinical practice to the extent that the occurrence of such laboratory abnormalities is likely to prompt an early repeat test.
Using our definition, we found that 3.9% of all patients with at least 1 ALT measurement experienced 1 or more ALT flares. Cicconi et al15 in 2007 reported a slightly higher percentage (5.2%), most likely due to both their different definition of an ALT flare (>3.5 times the baseline ALT value) and the higher prevalence of HCV-positive patients in Italy. Many studies focus only on treatment-experienced patients; using our proposed definition, 10% of treatment-experienced patients had an ALT flare. Although this was in line with estimates from some studies,18,20,21 it was lower than that reported in others.9,17,22 Chihrin et al8 analyzed only coinfected treated patients in their study and found that 22% of patients had ALT flares compared with only 13% within the same population in our study. Again, the definition of an ALT flare is likely to be the main factor that explains the variation between studies.
Interestingly, crude rates of ALT flares among those coinfected with HBV/HCV-positive were lower among treatment-experienced patients than those who are treatment naive. One possible reason for this is that naive patients tend to have lower CD4 counts and this in turn may be associated with a higher risk of liver toxicity.23 Although there was no significant interaction between treatment status and CD4 count, we did find that low CD4 count was a stronger predictor of ALT flares among patients who were off treatment than those who were on treatment, which further supports this explanation. A second possible explanation is that antiretroviral therapy-experienced patients may have been exposed to drugs such as lamivudine and emtricitabine that are known to have activity against HBV, which may reduce the apparent impact of HBV on ALT levels.
In multivariable analyses, patients taking NVP (either with ddI/d4T or without ddI/d4T) and those taking RTV were at increased risk of ALT flares, consistent with results from other studies.6,10 Both positive HCV and HBV status were associated with ALT flares and this too has been well documented.1,2,13,14 However, like Cicconi et al,15 we did not find a significant interaction between hepatitis status and current regimen, suggesting that HAART may not play as big a part in ALT flares among those who are HBV/HCV positive as previously thought. We found that patients who were on treatment for longer periods were less likely to have ALT flares, unlike Martinez et al9 who found that the risk of an ALT flare was increased by 10% (P = 0.02) for each additional year since starting treatment. This difference is likely to be explained by the inclusion criteria applied in the latter study. Patients were required to have started NVP-based HAART and NVP was found to be a risk factor in our study.
Our study has the limitations inherent in an observational cohort. In particular, we cannot control for unmeasured confounding such as alcohol consumption that has been reported as predictive of ALT flares.24 However, this is unlikely to affect our conclusions when comparing the different definitions of an ALT flare. We also do not collect data on other drugs that may also have an impact on ALT levels, including acetaminophen. Comparing our definition of an ALT flare to that usually found in the literature highlights the importance of differentiating between ALT flares and isolated ALT elevations. The stronger relationships seen between CD4 count and ALT flares when using a wider less stringent definition of an ALT flare may lead to unnecessary concern by both clinicians and patients when analyzing CD4 count, while the role of treatment regimen, viral load, and hepatitis status may not be given enough significance. Furthermore, all-cause mortality was only significantly associated with ALT flares when using the ACTG definition, and not when using our proposed definition.
These issues accentuate the need to consider monitoring bias when analyzing outcomes such as ALT flares, as associations between known risk factors may be under or over estimated if using a definition based on a single ALT value. In cases where new solitary ALT measurements >200 IU are measured, we recommend that a second test is performed to confirm the measurement. We recognize that in some studies, for reasons of study design, only a single ALT measurement may be available, in which case authors should acknowledge the problems of measurement variability when discussing their findings. Where multiple measurements are available, our analyses suggest that use of 2 consecutive ALTs > 200 IU, measured >2 weeks apart, may more reliably identify the causes and significance of ALT elevations.
1. Livry C, Binguet C, Sgro C, et al. Acute liver enzyme elevations in HIV
-1-infected patients. HIV Clin Trials
2. Gisolf E, Dreezen C, Danner S, et al; Prometheus Study Group. Risk factors for hepatotoxicity in HIV
-1-infected patients receiving ritonavir and saquinavir with or without stavudine. Prometheus Study Group. Clin Infect Dis
3. Sulkowski M, Thomas D, Chaisson R, et al. Hepatotoxicity associated with antiretroviral therapy in adults infected with human immunodeficiency virus and the role of hepatitis
C or B virus infection. JAMA
4. Bonfanti P, Landonio S, Ricci E, et al. Risk factors for hepatotoxicity in patients treated with highly active antiretroviral therapy. J Acquir Immune Defic Syndr
5. Monforte A, Bugarini R, Pezzotti P, et al. Low frequency of severe hepatotoxicity and association with HCV coinfection in HIV
-positive patients treated with HAART. J Acquir Immune Defic Syndr
6. Wit F, Weverling G, Weel J, et al. Incidence of and risk factors for severe hepatotoxicity associated with antiretroviral combination therapy. J Infect Dis
7. Torti C, Lapadula G, Casari G, et al. Incidence and risk factors for liver enzyme elevation during highly active antiretroviral therapy in HIV
-HCV co-infected patients: results from the Italian EPOKA-MASTER Cohort. BMC Infect Dis
8. Chihrin S, Antoniou T, Raboud J, et al. Risk factors for grade 3-4 liver enzyme elevation in HIV
C coinfected patients on combination antiretroviral therapy. AIDS Patient Care STDS
9. Martinez E, Blanco J, Arnaiz J, et al. Hepatotoxicity in HIV
-1-infected patients receiving nevirapine-containing antiretroviral therapy. AIDS
10. Requena D, Nunez M, Nacher J, et al. Liver toxicity caused by nevirapine. AIDS
11. Butt A, Tsevat J, Ahmad J, et al. Biochemical and virologic parameters in patients co-infected with hepatitis
C and HIV
versus patients with hepatitis
C mono-infection. Am J Med Sci
12. AIDS Clinical Trials Group. Table for Grading Severity of Adult Adverse Experiences
. Division of AIDS, National Institute of Allergy and Infectious Diseases, 1996; 2006.
13. Rodriguez-Rosado R, Garcia-Samaniego J, Soriano V, et al. Hepatotoxicity after introduction of highly active antiretroviral therapy (HAART). AIDS
14. Saves M, Raffi F, Clevenbergh P, et al. Hepatitis
B or hepatitis
C virus infection is a risk factor for severe hepatic cytolysis after initiation of a protease inhibitor-containing antiretroviral regimen in human immunodeficiency virus-infected patients. The APROCO Study Group. Antimicrob Agents Chemother
15. Cicconi P, Cozzi-Lepri A, Phillips A, et al. Is the increased risk of liver enzyme elevation in patients co-infected with HIV
virus greater in those taking antiretroviral therapy? AIDS
16. Maida I, Babudieri S, Selva C, et al. Liver enzyme elevation in hepatitis
C virus (HCV)-HIV
-coinfected patients prior to and after initiating HAART: role of HCV genotypes. AIDS Res Hum Retroviruses
17. den Brinker M, Wit F, Wertheim-van Dillen P, et al. Hepatitis
B and C virus co-infection and the risk for hepatotoxicity of highly active antiretroviral therapy in HIV
-1 infection. AIDS
18. Melvin D, Lee J, Belsey E, et al. The impact of co-infection with hepatitis
C virus and HIV
on the tolerability of antiretroviral therapy. AIDS
19. UK CHIC SC. The creation of a large UK-based multicentre cohort of HIV
-infected individuals: The UK Collaborative HIV
Cohort (UK CHIC) Study. HIV Med
20. Aceti A, Pasquazzi C, Zechini B, et al; LIVERHAART Group. Hepatotoxicity development during antiretroviral therapy containing protease inhibitors in patients with HIV
: the role of hepatitis
B and C virus infection. J Acquir Immune Defic Syndr
21. Servin-Abad L, Molina E, Baracco G, et al. Liver enzymes elevation after HAART in HIV
-HCV co-infection. J Viral Hepat
22. Puoti M, Torti C, Ripamonti D, et al. Severe hepatotoxicity during combination antiretroviral treatment: incidence, liver histology, and outcome. J Acquir Immune Defic Syndr
23. Uberti-Foppa C, De Bona A, Galli L, et al. Liver fibrosis in HIV
-positive patients with hepatitis
C virus: role of persistently normal alanine aminotransferase levels. J Acquir Immune Defic Syndr
24. Olalla J, Rubio R, Costa J, et al. The influence of hepatitis
C virus-human immunodeficiency virus co-infection on the appearance of liver enzyme elevation in people on high activity antiretroviral treatment. Eur J Intern Med
Centers currently providing data to UK Collaborative HIV Cohort: Chelsea and Westminster National Health Service (NHS) Trust, Imperial College Healthcare NHS Trust, King's College Hospital, the Mortimer Market Centre, the Royal Free NHS Trust, Barts and The London NHS Trust, Brighton and Sussex University Hospitals NHS Trust, Homerton University Hospital NHS Trust, The Lothian University Hospitals NHS Trust, and North Middlesex University Hospital NHS Trust.