Association Between ALT Level and Rate of CV Events in HIV
Association Between ALT Level and Rate of CV Events in HIV
Using the data from a large prospective study of HIV-positive individuals, we have confirmed the unexpected findings from Ford et al of an inverse association between an individual's ALT level and his/her risk of MI. In particular, we found that there was a linear relationship with the latest ALT level, with the risk of MI progressively decreasing as the ALT level increased. Our findings were robust to variations in the measurement of ALT and could not be explained by a higher rate of mortality from competing causes in those with higher ALT levels.
In contrast to Ford's analysis, where inverse associations were also seen with all-cause mortality, cancer deaths (not evaluated as part of the present analysis), and other cardio/cerebrovascular end points, our association with ALT did appear to be specific to the MI end point. However, it should be noted that whereas the cohorts included in Ford's study included individuals whose ALT levels were largely in the normal range, HIV-positive individuals in the D:A:D Study generally have higher ALT levels (eg, around 20% of follow-up time was spent with an ALT level of >60 U/L); thus, it is likely that the strong positive associations between ALT and liver-related mortality and diabetes are driven by individuals with ALT levels at the higher end of the normal range or above. Of note, our findings relating to MI risk, and those of Ford, are inconsistent with those of Schindhelm et al who noted a higher rate of fatal and nonfatal CHD events in those with higher ALT levels.
Consistent with other published data, a higher ALT level was associated with an increased risk of diabetes mellitus. Not surprisingly, we found an increased risk of liver-related mortality in those with very high ALT levels in the cohort. The strong association between a higher ALT level and an increased risk of liver-related mortality has been reported by others. Interestingly, we found a U-shaped association between ALT levels and all-cause mortality, with a higher mortality risk in those with very low and those with very high ALT levels. It is likely that the shape of any association with all-cause mortality will depend on the underlying causes of mortality in the cohort. Trends in causes of death in the D:A:D Study have recently been presented; liver-related mortality currently accounts for 13% of deaths that have occurred, with deaths from MI accounting for 6%.
We investigated whether the association between low ALT and a higher MI rate was driven by a higher rate of viral hepatitis (or injection drug use) in those with high ALT levels. Around 20% of follow-up time in the D:A:D Study is contributed by patients who are coinfected with HCV and 6% by those coinfected with hepatitis B virus; 17.0% of patients in the present dataset were infected with HIV via injection drug use. Although we found a significant inverse association between MI risk and ALT levels regardless of HCV status or injection drug use, the associations did appear to be numerically greater in those who were HCV positive and in those who were infected with HIV through injection drug use (note that these groups largely overlap). The differences between these subgroups were not significant, however, suggesting that any differences may simply reflect chance variation. Furthermore, the fact that Ford's study was conducted within the general population, where only a small proportion of individuals would be expected to have viral infections or to be injection drug users, suggests that the association is unlikely to be driven by these factors.
Exposure to some antiretroviral drugs, particularly nevirapine and ritonavir, is associated with hepatotoxicities and raised ALT levels; the nucleoside reverse transcriptase inhibitors, stavudine and didanosine (and possibly zidovudine), may also cause hepatic steatosis that would tend to lead to elevated ALT levels.7–10,20 It is unlikely that a greater exposure to these drugs could explain the lower risk of MI in those with high ALT levels for 2 reasons. First, if this were to explain our findings, the drugs that are most likely to cause ALT elevations would also have to be associated with a reduced MI risk; the only antiretroviral drug where this might plausibly be the case is nevirapine, which is associated with a more favorable lipid profile. Secondly, our associations were similar both before and after adjustment for cumulative exposure to specific antiretroviral drugs and to current exposure to drugs from the nucleoside reverse transcriptase inhibitor class. Thus, we do not believe that this is an explanation for our findings.
Results from a recent meta-analysis of 72 randomized trials of statins found that statin treatments significantly increased the rate of elevated ALT levels. As statin treatment would reduce the risk of MI in those receiving them, this might induce an inverse relationship between ALT levels and MI risk. However, the inverse association was not removed after adjustment for either the lipid levels themselves or for the use of lipid-lowering drugs as a class (note that the data collection for the D:A:D Study does not allow us to differentiate between statins and other lipid-lowering drugs). Thus, we do not believe that this is a likely explanation for our findings.
Finally, it should be noted that our multivariable regression model included adjustment for current smoking status. Hence, although cigarette smoking is a known risk factor for MI and may also lead to a reduction in ALT levels, it cannot explain the associations seen. Although we do not capture more detailed information on pack-years or number of cigarettes smoked, we would not expect this to confound any associations with ALT and so do not believe that this additional information would change our conclusions.
Several limitations of our study must be noted. Firstly, we do not collect information on alcohol use among participants. It has been suggested that a moderate intake of alcohol may be beneficial to the coronary arteries, and alcohol is known to increase ALT levels, thus possibly explaining the observed inverse association between increasing ALT levels and MI risk. However, it should be noted that the findings by Ford et al were adjusted for alcohol consumption, and so this did not appear to explain the inverse association in their study. Furthermore, we did not find a similarly strong inverse association with the CHD end point, although alcohol may also be expected to lead to a reduced risk of this broader end point. As ALT levels were not routinely measured systematically in individuals with HIV until relatively recently and not all cohorts routinely captured these measurements, our analyses could only include patients from 8 of the 11 participating cohorts. However, although this reduces the amount of follow-up that can be used in analysis, it does result in a more contemporaneous population increasing the clinical relevance of the findings to current treatment protocols. Given the characteristics of the patients in the D:A:D Study, our results are generalizable to a largely male population. As an observational study, our analyses are always subject to the possibility of unmeasured confounding and other potential sources of bias, limiting any statements regarding a causal association between lower ALT levels and MI risk. Finally, although the normal ALT range differs between men and women, the small number of women in the study and their lower MI risk limits our ability to perform stratified analyses by gender.
In summary, we confirm that higher ALT levels are associated with a lower MI risk in the D:A:D Study, but with higher risks of liver-related mortality and diabetes mellitus. As yet, there is no identified biological mechanism to explain the association between a low ALT level and a higher MI risk; although it is interesting to speculate that such individuals might be younger or healthier, we see no reason why this group should have an increased risk of MI. Furthermore, we are unaware of any clinical conditions or factors that would lead to lower than normal ALT levels that might lead to the higher MI risk. Thus, it is hard to comment on the clinical implications of these findings at this stage. Rather, in replicating the findings of Ford et al in a different setting, we hope to encourage other research groups to investigate possible biological mechanisms.
Discussion
Using the data from a large prospective study of HIV-positive individuals, we have confirmed the unexpected findings from Ford et al of an inverse association between an individual's ALT level and his/her risk of MI. In particular, we found that there was a linear relationship with the latest ALT level, with the risk of MI progressively decreasing as the ALT level increased. Our findings were robust to variations in the measurement of ALT and could not be explained by a higher rate of mortality from competing causes in those with higher ALT levels.
In contrast to Ford's analysis, where inverse associations were also seen with all-cause mortality, cancer deaths (not evaluated as part of the present analysis), and other cardio/cerebrovascular end points, our association with ALT did appear to be specific to the MI end point. However, it should be noted that whereas the cohorts included in Ford's study included individuals whose ALT levels were largely in the normal range, HIV-positive individuals in the D:A:D Study generally have higher ALT levels (eg, around 20% of follow-up time was spent with an ALT level of >60 U/L); thus, it is likely that the strong positive associations between ALT and liver-related mortality and diabetes are driven by individuals with ALT levels at the higher end of the normal range or above. Of note, our findings relating to MI risk, and those of Ford, are inconsistent with those of Schindhelm et al who noted a higher rate of fatal and nonfatal CHD events in those with higher ALT levels.
Consistent with other published data, a higher ALT level was associated with an increased risk of diabetes mellitus. Not surprisingly, we found an increased risk of liver-related mortality in those with very high ALT levels in the cohort. The strong association between a higher ALT level and an increased risk of liver-related mortality has been reported by others. Interestingly, we found a U-shaped association between ALT levels and all-cause mortality, with a higher mortality risk in those with very low and those with very high ALT levels. It is likely that the shape of any association with all-cause mortality will depend on the underlying causes of mortality in the cohort. Trends in causes of death in the D:A:D Study have recently been presented; liver-related mortality currently accounts for 13% of deaths that have occurred, with deaths from MI accounting for 6%.
We investigated whether the association between low ALT and a higher MI rate was driven by a higher rate of viral hepatitis (or injection drug use) in those with high ALT levels. Around 20% of follow-up time in the D:A:D Study is contributed by patients who are coinfected with HCV and 6% by those coinfected with hepatitis B virus; 17.0% of patients in the present dataset were infected with HIV via injection drug use. Although we found a significant inverse association between MI risk and ALT levels regardless of HCV status or injection drug use, the associations did appear to be numerically greater in those who were HCV positive and in those who were infected with HIV through injection drug use (note that these groups largely overlap). The differences between these subgroups were not significant, however, suggesting that any differences may simply reflect chance variation. Furthermore, the fact that Ford's study was conducted within the general population, where only a small proportion of individuals would be expected to have viral infections or to be injection drug users, suggests that the association is unlikely to be driven by these factors.
Exposure to some antiretroviral drugs, particularly nevirapine and ritonavir, is associated with hepatotoxicities and raised ALT levels; the nucleoside reverse transcriptase inhibitors, stavudine and didanosine (and possibly zidovudine), may also cause hepatic steatosis that would tend to lead to elevated ALT levels.7–10,20 It is unlikely that a greater exposure to these drugs could explain the lower risk of MI in those with high ALT levels for 2 reasons. First, if this were to explain our findings, the drugs that are most likely to cause ALT elevations would also have to be associated with a reduced MI risk; the only antiretroviral drug where this might plausibly be the case is nevirapine, which is associated with a more favorable lipid profile. Secondly, our associations were similar both before and after adjustment for cumulative exposure to specific antiretroviral drugs and to current exposure to drugs from the nucleoside reverse transcriptase inhibitor class. Thus, we do not believe that this is an explanation for our findings.
Results from a recent meta-analysis of 72 randomized trials of statins found that statin treatments significantly increased the rate of elevated ALT levels. As statin treatment would reduce the risk of MI in those receiving them, this might induce an inverse relationship between ALT levels and MI risk. However, the inverse association was not removed after adjustment for either the lipid levels themselves or for the use of lipid-lowering drugs as a class (note that the data collection for the D:A:D Study does not allow us to differentiate between statins and other lipid-lowering drugs). Thus, we do not believe that this is a likely explanation for our findings.
Finally, it should be noted that our multivariable regression model included adjustment for current smoking status. Hence, although cigarette smoking is a known risk factor for MI and may also lead to a reduction in ALT levels, it cannot explain the associations seen. Although we do not capture more detailed information on pack-years or number of cigarettes smoked, we would not expect this to confound any associations with ALT and so do not believe that this additional information would change our conclusions.
Several limitations of our study must be noted. Firstly, we do not collect information on alcohol use among participants. It has been suggested that a moderate intake of alcohol may be beneficial to the coronary arteries, and alcohol is known to increase ALT levels, thus possibly explaining the observed inverse association between increasing ALT levels and MI risk. However, it should be noted that the findings by Ford et al were adjusted for alcohol consumption, and so this did not appear to explain the inverse association in their study. Furthermore, we did not find a similarly strong inverse association with the CHD end point, although alcohol may also be expected to lead to a reduced risk of this broader end point. As ALT levels were not routinely measured systematically in individuals with HIV until relatively recently and not all cohorts routinely captured these measurements, our analyses could only include patients from 8 of the 11 participating cohorts. However, although this reduces the amount of follow-up that can be used in analysis, it does result in a more contemporaneous population increasing the clinical relevance of the findings to current treatment protocols. Given the characteristics of the patients in the D:A:D Study, our results are generalizable to a largely male population. As an observational study, our analyses are always subject to the possibility of unmeasured confounding and other potential sources of bias, limiting any statements regarding a causal association between lower ALT levels and MI risk. Finally, although the normal ALT range differs between men and women, the small number of women in the study and their lower MI risk limits our ability to perform stratified analyses by gender.
In summary, we confirm that higher ALT levels are associated with a lower MI risk in the D:A:D Study, but with higher risks of liver-related mortality and diabetes mellitus. As yet, there is no identified biological mechanism to explain the association between a low ALT level and a higher MI risk; although it is interesting to speculate that such individuals might be younger or healthier, we see no reason why this group should have an increased risk of MI. Furthermore, we are unaware of any clinical conditions or factors that would lead to lower than normal ALT levels that might lead to the higher MI risk. Thus, it is hard to comment on the clinical implications of these findings at this stage. Rather, in replicating the findings of Ford et al in a different setting, we hope to encourage other research groups to investigate possible biological mechanisms.
Source...