Table of Contents
Keiser O, Taffe P, Zwahlen M, et al, and the Swiss HIV Cohort Study. All cause mortality in the Swiss HIV cohort study from 1990 to 2001 in comparison with the Swiss population. AIDS. September 3, 2004;18(13):1835-1843.
Most patients with HIV infection have one unspoken question on their minds: "How long do I have to live?" It is also one of the toughest questions for clinicians to answer.
These are questions best answered with carefully observed longitudinal cohorts. The Swiss Cohort Study is one such cohort. An incredible 70% of all persons with AIDS in Switzerland have been enrolled in the study and over 5,000 persons are currently in follow-up.
To address the issue of survival in the HAART era, cohort investigators examined mortality data among the people who were enrolled in the study from 1990 to 2001 and compared this to mortality during the same period among the general Swiss population. During the study period, almost 11,000 individuals with HIV infection were seen at least once for a study visit; of these people 3,630 (33%) died and 2,290 (21%) were lost to follow-up.
The yearly risk of death from any cause declined from 13% before 1995 to 3% after 1998, when HAART became available. By 1998, 70% of the people in the cohort were receiving combination HIV therapy. Importantly, 3,853 (35%) of the cohort were injection drug users (IDUs).
Using various analysis techniques, the investigators found that prior to HAART, there was no difference in survival among cohort participants when categorized by gender, or whether they were "homosexual" or "heterosexual," or were IDUs or non-IDUs.
After HAART availability, all groups enjoyed improved survival, particularly those who initiated HAART. The exception was IDUs who were found to have significantly less robust declines in mortality. Unfortunately, this investigation did not describe any of the factors accounting for the reduced survival seen in IDUs in the HAART era. One can imagine, however, that relatively less access and/or adherence to HAART, co-morbidities such as viral hepatitis coinfection and IDU-related complications, such as bacterial infection and overdose, could explain their increased risk of mortality.
Even with HAART being accessible, there was more death seen in all the HIV-infected groups relative to the general Swiss population. Again, there were limited data on the causes of mortality among people with HIV versus those without HIV to help explain these differences.
This study points out several important aspects of the HIV epidemic in areas where HAART is prescribed. First, it reminds us that prior to HAART, all transmission risk groups had similar fates and this was, of course, worse than that observed in the general population. Second, after the advent of HAART, improved survival rates were seen in all the categories of HIV-infected study participants, though these survival rates continued to be inferior to that of the general Swiss population.
The causes of the greater number of deaths among HIV-infected persons, despite HAART, likely indicate that HAART alone does not provide full immunity from "badness." While it is obvious that treatment failure can lead to immunosuppression and life-threatening conditions, a number of studies have demonstrated that people who are HIV-infected are also more likely to have infection with hepatitis C virus, smoke cigarettes and abuse non-injected substances such as cocaine and alcohol -- all of which can reduce one's stay on Earth. Further examination of the differential causes of death among HIV-infected and uninfected Swiss in the HAART era is hopefully forthcoming.
Lastly, this study demonstrates that the benefit of HAART has not been enjoyed equally. IDUs had significantly smaller reduction in mortality post-HAART than non-IDUs. The results paint a picture of the past that can help inform our actions now, most notably by raising awareness that even with HAART, HIV-infected persons die in greater numbers then the uninfected -- a discrepancy that deserves greater scrutiny. In addition, this study highlights the precarious existence of IDUs. Clearly, even in a nation with as comprehensive health care as found in Switzerland, IDUs have special needs that require intervention.
Wang C, Vlahov D, Galai N, et al. Mortality in HIV-seropositive versus -seronegative persons in the era of highly active antiretroviral therapy: implications for when to initiate therapy. J Infect Dis. September 15, 2004;190(6):1046-1054.
The findings from the Swiss Cohort Study described above are an excellent segue to data reported by researchers from the AIDS Linked to Intravenous Experience (ALIVE) study in Baltimore, Md. This cohort examines IDUs exclusively, including HIV-infected and uninfected men and women living in Baltimore. From 1997 to 2000, 583 HIV-infected and 920 HIV-uninfected IDUs were followed longitudinally and factors associated with survival of the HIV-infected participants relative to the seronegative controls were sought.
Almost all of the people in this study were African-American (93%) and male (74%). Subjects were followed approximately 2 to 4 years with the HIV-uninfected participants more likely to have longer follow-up. There were 187 deaths, 58 of which were classified as being AIDS-related.
The bottom line results were that, as a whole, the mortality of HIV-uninfected IDUs was lower than the HIV-infected group. The mortality rate of the HIV-uninfected group was 19.9/1000 patient-years (py) versus 38.1/1000 py for the HIV-infected group. However, the survival disadvantage of HIV was nonexistent among those who started HAART at a CD4+ cell count of more than 350 cells/uL. These individuals had a mortality rate of 24.1/1000 py, which was not significantly different than the HIV-uninfected controls.
Yet, HIV-infected people with a baseline CD4+ cell count more than 350 cells/uL who did not start HAART had a mortality rate of 43.0/1000 py, which was significantly different than the controls.
HIV-infected participants with a baseline CD4+ cell count of between 200 and 350 cells/uL had significantly higher mortality rates than the HIV-uninfected controls whether they started HAART (mortality rate of 50.5/1000 py) or did not (mortality rate of 59.9/1000 py). However, this was only the case when subjects with a count between 200 and 350 cells/uL had a viral load of at least 55,000 copies/mL. Those with a CD4+ cell count of between 200 and 350 cells/uL and a viral load of less than 55,000 copies/mL had mortality rates that were no different than the HIV-uninfected controls. Not surprisingly, those with a CD4+ cell count of less than 200 cells/uL fared the worse, even when they did start HAART, with mortality rates well over 80/1000 py.
The HIV-infected participants who did not start HAART had the greatest risk of death from AIDS and non-AIDS related causes. The relative frequency of AIDS and non-AIDS-related mortality across the cohort strata can be found in the graph below. The most commonly reported non-AIDS-related causes of death were overdose, accident/violence and bacterial or viral infections.
The authors make a convincing argument that if the goal of HIV therapy is to render infected patients like uninfected persons with regard to risk of death, then these results indicate that, at least in this particular population, HIV therapy should be started at a CD4+ cell count of more than 350 cells/uL and that delaying therapy initiation until the count is 200 to 350 cells/uL in patients with a high (>55,000 copies/mL) viral load is risky.
However, as discussed in an accompanying editorial,1 there are limitations to the study, such as the lack of randomization, and potential differences between patients in the CD4+ cell count strata with regard to adherence level, specific causes of death, access to care and other characteristics that prevent the adoption of a new standard of care for the initiation of antiretroviral therapy at this time.
In addition, as demonstrated in the Swiss Cohort Study described above, IDUs in general, and particularly when HIV infected, have relatively higher rates of mortality. Therefore, it is possible that IDUs, as a group, would be more sensitive than other populations to interventions such as HAART that can improve survival even marginally.
Despite these limitations, and in lieu of the findings from the Swiss group, these results provoke clinicians to continue to consider the merits of earlier rather than later therapy. As we know, this is a complicated matter that must take into account patient commitment to therapy, the tolerance for adverse effects and the effect of therapy on reducing HIV transmissibility. Ever so gradually we have inched toward delaying therapy and now incrementally we are climbing back up the CD4+ cell count ladder. With better tolerated and more convenient therapies, as well as studies such as this, it seems inevitable that we will be treating HIV earlier -- it's déjà vu all over again.
Valer L, Martin-Carbonero L, de Mendoza C, Corral A, Soriano V. Predictors of selection of K65R: tenofovir use and lack of thymidine analogue mutations. AIDS. October 21, 2004;18(15):2094-2095.
The emergence of the K65R mutation of the reverse transcriptase genome of HIV-1 has been accompanied by claims and counter claims by competing pharmaceutical companies regarding the incidence and significance of this resistance mutation.
What is clear is that K65R most commonly develops following exposure to tenofovir (TDF, Viread). In Gilead 903, the head-to-head comparison of tenofovir versus stavudine (d4T, Zerit) when combined with lamivudine (3TC, Epivir) and efavirenz (EFV, Sustiva, Stocrin), 8 of the 47 (17%) subjects receiving tenofovir who experienced virologic failure had a K65R mutation detected by week 144.2
However, it is not only tenofovir that can select for this mutation. In this same study, 2 of the 49 subjects randomized to stavudine developed K65R. In addition, abacavir (ABC, Ziagen), didanosine (ddI, Videx) and zalcitabine (ddC, Hivid) have been shown to select for this mutation either in vitro or in vivo. Once present, the mutation reduces susceptibility to tenofovir, abacavir and didanosine.
Debate surrounding the K65R mutation centers on the optimal sequencing of nucleoside-based regimens for HIV treatment. One camp holds that to start with tenofovir invites the development of K65R and subsequent "loss" of abacavir and didanosine as salvage nucleoside/nucleotide reverse transcriptase inhibitors (NRTIs). On the other side, the position is that K65R does not develop in the majority of patients who fail an initial therapy containing tenofovir, lamivudine and a non-nucleoside reverse transcriptase inhibitor (NNRTI), and that even with the development of this mutation, sensitivity to zidovudine (AZT, Retrovir) remains intact if not actually enhanced.
Further, there are anecdata (a hybrid of anecdotes and data) that suggest K65R can be readily salvaged. While this debate rages, clinicians have voted with their pens and tenofovir is increasingly being prescribed as part of an initial HIV therapy regimen.
To examine the incidence of the K65R mutation in a clinical setting and identify factors associated with its emergence, Valer and colleagues in Madrid scrutinized 5 years of resistance data collected from the 1,846 patients who had genotypic resistance testing performed at their hospital. A total of 53 patients (2.9%) had K65R, all of whom were treatment experienced (i.e., among the 216 treatment-naive patients who were tested, none had K65R detected). In the 1,630 treatment-experienced patients, the incidence of K65R increased over time from 0.6% in 1999 to 11.5% in the first quarter of 2004.
As expected, the majority of patients with K65R were tenofovir experienced, however, 10 of the 53 patients with the mutation were naive to tenofovir. These 10 patients were taking regimens that included stavudine plus didanosine (4), abacavir plus lamivudine (2), stavudine plus lamivudine (2), stavudine plus abacavir (1), or zalcitabine (1).
Among the 43 tenofovir-exposed patients, K65R was seen in those not also receiving a thymidine analogue. Three quarters of the patients (32) were taking a combination containing tenofovir plus didanosine and the remainder were on either tenofovir alone or combined with abacavir or lamivudine.
As reported previously, thymidine analogue mutations (TAMs) and K65R tended to be mutually exclusive, thus supporting the concept that NRTI resistance can develop along two pathways leading to either accumulated TAMs or K65R. In a multivariable analysis, K65R, while inversely associated with TAMs, was associated with other non-TAM mutations and the Q151 multi-NRTI resistance complex -- a mutation that confers decreased susceptibility to all NRTIs except tenofovir, save when it accompanies K65R. K65R was not statistically significantly associated with the M184V lamivudine/emtricitabine (FTC, Emtriva)-resistance mutation, but 45% of patients with K65R also harbored the M184V mutation.
This investigation raises important issues for continued study. First, the study provides a nice overview of the magnitude of K65R in a real world setting. As all of us familiar with antibiotic utilization know too well, with increased prescription of an antimicrobial, it is only a matter of time until resistance to that drug emerges. The more tenofovir we use, the more K65R we will see. This is neither unexpected or specific to this drug. With wider use of abacavir, the more L74V we will see, just like the more TAMs we saw with greater use of thymidine analogues such as zidovudine. This is the destiny of antiretrovirals.
So, when crafting therapy, we have to consider which patterns of resistance may emerge that we are less willing to tolerate, and then prescribe accordingly. Second, the results add another note of concern regarding the use of tenofovir and didanosine together (especially when combined with NNRTIs). Although in this study the preponderance of K65R in tenofovir plus didanosine experienced patients may be explained by the popularity of this combination at this Madrid clinic, recent data from the ININ Study and a recent trial from the UK put us on notice that this is a regimen that may be suboptimal.3,4
Lastly, the observation that K65R and TAMs are antagonistic begs clinical investigation of strategies to exploit this phenomenon. Given our collective embrace of tenofovir as a potent, well-tolerated and convenient antiretroviral, we must pursue a greater understanding of the prevention and management of virologic failure of regimens which contain this important agent.
Stolte IG, de Wit JB, van Eeden A, Coutinho RA, Dukers NH. Perceived viral load, but not actual HIV-1-RNA load, is associated with sexual risk behaviour among HIV-infected homosexual men. AIDS. September 24, 2004;18(14):1943-1949.
Data from a variety of sources suggest that when individuals perceive themselves rightly or wrongly to be at less risk of acquiring HIV infection, they practice unsafe behaviors -- witness the change in sexual risk taking behaviors in some quarters following the availability of HAART. Further, HIV-infected individuals have been observed in some, but not all, studies to increase risk behaviors when on HAART or when CD4+ cell count and viral load values were favorable.
To examine whether the HIV-related beliefs and patient perceived and actual viral load and CD4+ cell count data influenced sexual behavior among HIV-infected men who have sex with men, investigators in Amsterdam conducted a mail-based survey of men enrolled in a previously established observational cohort study. Surveys were mailed at 3 time points from 2000 to 2003. Over 400 surveys were sent and 57 men answered all 3.
The HIV-RNA level was documented to be undetectable in 61% of the men surveyed at the time of the first survey and 79% at the third survey. Unprotected sex increased among the respondents during the study period from 10% to 28% when sex was with "casual" partners.
The investigators found no association between health beliefs, perceived viral load or CD4+ cell count or actual lab data and unsafe sex with casual partners. However, among subjects with a steady partner who was HIV uninfected or of uncertain HIV status, the subject's perceptions about his viral load were independently associated with unsafe sexual behavior such that those with a more favorable view of their viral load were less likely to practice protected anal sex. Interestingly, the actual viral load (and CD4+ cell count) was not associated with risk behavior. Of note, a third of the subjects with steady partners not known to be HIV infected had viral loads that were not undetectable.
This interesting study, although small, suggests that a patient's perception of his viral load may not match the actual viral load. Further, belief that one's viral load was favorable (i.e., low) increased the likelihood of unsafe sex with a steady partner who was not HIV infected.
Although one can debate the applicability of these results to men who have sex with men in other countries as well as to heterosexual men and women, it makes sense for providers to endeavor to understand whether their patients are sexually active, whether they use condoms or other methods to reduce HIV transmission and what are their perceptions regarding their health status and their infectiousness, since patients' perceptions of their risk may be far from reality.
Schackman BR, Oneda K, Goldie SJ. The cost-effectiveness of elective cesarean delivery to prevent hepatitis C transmission in HIV-coinfected women. AIDS. September 3, 2004;18(13):1827-1834.
Cesarean section (C-section) has been found to be an effective method for the prevention of mother-to-child transmission of HIV. However, the ability of potent antiretroviral therapy to reduce perinatal transmission to extremely low levels has dissuaded both women with low viral loads and their clinicians from pursuing this invasive intervention when not necessary for other reasons.
Vertical transmission of hepatitis C (HCV) is less common than that of HIV, but like HIV it is more likely when HCV is present in the mother's blood. Since a significant proportion of HIV-infected women are coinfected with HCV, Schackman and colleagues examined the cost-effectiveness of a C-section to prevent HCV transmission by coinfected mothers.
Two strategies were compared: C-section for all coinfected women with undetectable HIV but detectable HCV versus C-section only when required based on fetal status.
To construct their analysis, the researchers evaluated available data regarding HIV and HCV coinfection in women in the United States. Between 1988 and 1994 there were 6,000 to 7,000 deliveries to HIV-infected women in the United States, an estimated 28% of whom were IDUs and therefore likely to be HCV infected. That means, if the trends during this period have been maintained, approximately 1,700 to 2,000 coinfected women give birth every year.
In a complex analysis that takes into consideration patient acceptance or refusal of C-section and the medical necessity of C-section in some cases, the investigators estimate that C-section would prevent 45 cases of infant HCV infection per 1,000 deliveries. Thus, about 90 cases would be avoided in the United States annually.
However, C-section would also carry a maternal mortality risk of 1 per 100,000 deliveries (that translates into about 1 additional death due to C-section every 50 years). The cost of elective C-section to prevent HCV transmission in coinfected women would be approximately $3,900 to $6,100 per quality-adjusted life year -- a cost that is considered to be cost-effective.
In the United States today elective C-section in HIV-infected women has been inconsistently recommended. Potent HIV therapy has reduced the risk of mother-to-infant HIV transmission to as low as 5%. Use of a surgical procedure to further reduce this risk has been a hard sell except when the maternal viral load is not suppressed or when other maternal or fetal conditions warrant a C-section.
The results from this study add another variable to be considered in the management of the pregnant HIV-infected woman. Coinfection with HCV can pose a risk of HCV transmission to the infant and a C-section can reduce that risk. Although the results represent estimates based on prior data, they are likely the best we can expect as a prospective clinical trial would be difficult to conduct given the sample size that would be required and the rate of coinfected pregnancies in the United States. Overall, for a coinfected woman on the fence about C-section, these data may tip her and her clinician to favor this surgical option.