AIDS-Related Cancers in the Era of Highly Active Antiretroviral Therapy

Introduction

Kaposi’s sarcoma and non-Hodgkin’s lymphoma are considered acquired immunodeficiency syndrome (AIDS)-defining illnesses. Although invasive cervical carcinoma is less clearly related to the human immunodeficiency virus (HIV) and/or AIDS, it is also accepted as AIDS-defining.[1] In addition, persons with HIV/AIDS have been found to be at increased risk for several other cancers—most prominently, Hodgkin’s disease and anal cancers.[2] Other sites, especially conjunctival tumors and testicular cancer, have also been reported in some registry studies and case reports, although the small increases in the relative risk observed make it difficult to convincingly associate these cancers with AIDS.

The pathogenesis of AIDS-related cancers (ie, Kaposi’s sarcoma and non-Hodgkin’s lymphoma) is generally considered to be directly related to immunosuppression rather than HIV infection.[3] The logic behind this supposition is compelling. The malignant cells themselves are not infected with HIV. Moreover, persons with HIV are not at exceptional risk for these cancers (at least at a detectable level) unless they are also immunosuppressed. In addition, persons who are immunosuppressed for reasons other than HIV infection (eg, iatrogenic or genetic causes) are at risk for these same types of cancer. Some investigators suggest that Tat, a protein expressed by HIV, may be linked to the increased incidence of Kaposi’s sarcoma,[4] but even if this proves correct, this incidence is greatly increased only among those in whom there is concomitant immunosuppression.

If immunosuppression underlies the occurrence of these malignancies, then ameliorating the immunodeficiency should markedly reduce the risk of developing cancer. The advent of highly active antiretroviral therapy (HAART) for persons with AIDS presents the opportunity for examining this hypothesis. Strong evidence exists that HAART results in impressive reductions in HIV replication and is usually associated with improvements in the CD4 lymphocyte counts of persons with immunosuppression.[5] As expected, these improvements in immunity have greatly reduced the incidence of AIDS-associated opportunistic infectious diseases.[6] In this review, we will examine the evidence that they have also decreased the incidence of cancer in persons with AIDS.

Antiretroviral Therapies

The extent to which antiretroviral therapies might influence HIV-related immunosuppression depends on their efficacy in improving immunity and how widely such therapies are used in persons with HIV/AIDS. In 1987, monotherapy with zidovudine(Drug information on zidovudine) (azidothymidine or AZT [Retrovir]) was shown to be effective in improving prognosis.[7] The clinical response was justifiably attributed to improvements in immunity as measured by increasing CD4 cell counts, although it is also possible that lowering HIV levels further improved immunity by affecting functional immunity in still-viable lymphocytes.[8]

In practice, clinicians soon observed that the improvement in immunity following zidovudine therapy was transitory. Within 6 to 12 months, therapy-resistant HIV variants emerged, high viral levels returned, and the deterioration of immunity resumed. Later studies confirmed that the long-term benefit was minimal,[9] and the need for combination antiretroviral regimens became obvious.

During the 1990s, many new antiretroviral agents were licensed, and others continue to be developed. The introduction of effective protease inhibitors in the mid-1990s has been especially important. Clinical trials have demonstrated that these drugs, used in combination, are highly effective in blocking HIV replication for prolonged periods.[10] In many persons, viral levels fall below the level at which they can be detected even with sensitive tests.

Highly Active Antiretroviral Therapy

Dual drug therapies also came into wide use in the early 1990s, and since 1996, triple-drug regimens that include protease inhibitors have been recommended for persons with clinically significant HIV-related immunosuppression. Figure 1, based on data reported by the Centers for Disease Control and Prevention (CDC),[11] illustrates the proportion of persons with HIV/AIDS who were taking these drugs in the United States between 1992 and 1997.

Highly active antiretroviral therapy with protease inhibitor drugs blocks HIV replication, resulting in meaningful improvements in the clinical status and prognosis of many persons infected with HIV. This strategy demonstrates that good HIV control and an increase in the CD4 count are associated with enhanced improvement in defending the host against environmental pathogens. Moreover, these results have revolutionized the care and prognosis of HIV patients, leading to the reasonable expectation that persons with HIV infection can survive for many years, even after an AIDS diagnosis.

Despite these encouraging developments, there are several serious caveats concerning the impact of HAART therapy on cancer risk trends. Foremost is the fact that many HIV-infected persons are not receiving HAART.[11] Some people are not aware that they are HIV-infected until they develop an AIDS-defining illness (which may be cancer). For others, such as those with dementia or drug addiction problems, the complex regimens are beyond their ability to support or maintain. Still others cannot tolerate the side effects of the drug combinations. Finally, even among those who take full HAART regimens, breakthrough HIV infections from resistant strains can occur.

It is reasonable to hope that newly developed antiretroviral drugs will lead to easier administration, lower toxicities, and control of HIV viremia. In the interim, however, there continues to be a population of HIV-infected, immunosuppressed persons who are not receiving effective anti-HIV therapy.

Immune Response Does Not Normalize

Even among those who respond well to HAART, the immune responses of severely immunocompromised subjects typically do not return to normal. Rather, in seriously immunocompromised subjects, CD4 counts increase from dangerously low levels of less than 50 cells/mL to perhaps 200 to 300 cells/mL.[12] At this level, immunity is not normal (normal CD4 levels are above 500 cells/mL), but the risk of developing a major life-threatening disease is much less.

One unexpected consequence of HAART’s success is the increasingly recognized phenomenon of immune reconstitution disease, in which symptoms appear in direct relation to the improvement in immunity.[13,14] In these syndromes, the host recognizes and reacts to low-grade infections that were not apparent earlier, such as may occur with Mycobacterium avium-intracellulare and hepatitis C virus. This reaction can be so severe as to be fatal. Thus, the immunity that is measured as marginally deficient by conventional CD4 counts can have highly reactive clones that are proliferating robustly because of stimulation from smoldering infections. These activated clones may affect the risk of developing cancer, especially the lymphomas, in ways that obscure a relationship to HAART.

Kaposi’s Sarcoma

The proportion of AIDS patients presenting with Kaposi’s sarcoma has declined over time. There are several explanations for this change. Public health data often record only the first AIDS-defining illness. Early in the AIDS epidemic, most AIDS cases were identified on the basis of a diagnosis of Kaposi’s sarcoma or Pneumocystis carinii pneumonia. However, as the understanding of the range of AIDS-associated diseases increased, the definition of AIDS expanded to include other conditions.

In the most recent definition of an AIDS case, formulated in 1993, the CDC changed the definition to include HIV-infected persons who have no overt illness but are immunosuppressed (CD4 < 200 cells/mL or < 14%) to the point of being at risk for an AIDS-defining illnesses.[1] This expansion of the diagnosis has allowed recognition of other conditions as being AIDS-related before the appearance of Kaposi’s sarcoma.

Even in data sets that include both onset and a later diagnosis of Kaposi’s sarcoma, however, the incidence of the disease has declined. Some of this decline can be attributed to changes in the population of HIV-infected persons in the United States. Among persons with AIDS, the incidence of Kaposi’s sarcoma varies considerably by group. Initially, the majority of AIDS cases occurred in homosexual men, a group with an 8- to 10-fold higher risk for Kaposi’s sarcoma than other groups.[3] Moreover, most cases were in white men, a group that has at least twice the risk for Kaposi’s sarcoma compared to black men, even after controlling for HIV-exposure groups. However, the population of AIDS patients who are white homosexual men has declined as a proportion of all AIDS cases, whereas the percentage who are black men, both homosexual and otherwise, has increased, thus progressively enriching the mix of newly diagnosed AIDS cases with persons at lower risk for Kaposi’s sarcoma. The impact of HAART on risk for Kaposi’s sarcoma needs to be assessed carefully against this backdrop.

Kaposi’s Sarcoma and HAART

Almost all reports have noted declines in the incidence of Kaposi’s sarcoma between the early 1990s and the post-1996 era (Table 1).[15-21] In summary, between 1994 and 1995 and from 1996 onward (the HAART era), the incidence of Kaposi’s sarcoma declined by about two-thirds, to the point that it is now an uncommon diagnosis in people with AIDS.

However, a temporal approach does not prove a relationship to HAART because the decline in Kaposi’s sarcoma incidence long preceded the introduction of this therapy. In part, the change in the definition of AIDS and shifts in the population that became infected with HIV contributed to that decline, as discussed above. Additionally, the early, less effective, antiretroviral regimens may have reduced the incidence of Kaposi’s sarcoma, although statistical evidence supporting this association is inconclusive.[15]

More speculatively, we now understand that Kaposi’s sarcoma requires infection with a newly described herpesvirus, HHV-8,[22] which is common among homosexual men.[23] This virus appears to have been co-epidemic with sexually transmitted HIV.[24,25] Possibly, burnout of the HHV-8 epidemic, along with exhaustion of the population susceptible to Kaposi’s sarcoma, could have contributed. Persons who acquire HHV-8 while HIV-infected (and presumably immunosuppressed) have a 2.5-fold increased risk of developing Kaposi’s sarcoma over that of persons who were already infected with HHV-8 when they became infected with HIV.[26]

Impact on Immunity

Nevertheless, there is compelling extrinsic and intrinsic evidence that improved immunity is probably the major factor influencing the reduction in the incidence of Kaposi’s sarcoma. Extrinsically, case reports describe improvement or even resolution of preexisting Kaposi’s sarcoma in persons who began receiving HAART.[27-31] Some reports suggest that this clinical effect is particularly potent in combination antiretroviral drug regimens containing protease inhibitors.[30,31]

Given its impact on existing Kaposi’s sarcoma tumors, treatment with HAART would be expected to prevent incipient tumors. Intrinsically, in several published reports, the "breakthrough" cases of Kaposi’s sarcoma that occurred in the HAART era were observed only among men who did not receive potent antiretroviral therapy.[15,20]

Also supporting a direct role for HAART (through its impact on immunity) is a report that examined the risk of illnesses in HIV-infected persons started on HAART. Investigators from the Swiss HIV Cohort Study examined the risk for Kaposi’s sarcoma and other diseases.[32] Case numbers are somewhat marginal for strong conclusions, but their results showed a significant difference in the incidence of Kaposi’s sarcoma between the 6 months just before HAART and the 6 months immediately after HAART.

This decline in incidence implied that HAART had an almost immediate impact. Overall, the risk declined by 66% during the 15 months after HAART was initiated (P = .001, compared to the pre-HAART period). While such data are subject to analysis problems that complicate their interpretation,[33] they nevertheless demonstrate a decreased risk for Kaposi’s sarcoma specifically related to HAART in a study in which each subject served as his or her own control.

Supplemental Study

As a supplemental study for this report, I examined cancer risk in 309,365 persons with AIDS in the AIDS-Cancer Match Registry Study.[34] More than 21,000 cases of Kaposi’s sarcoma were reported in AIDS patients, the majority of which were AIDS-defining illnesses. Cancer incidence in persons with AIDS cannot be easily quantified when the cancer defines AIDS onset, because the number of subjects at risk is unknown. However, persons with AIDS-defining conditions other than Kaposi’s sarcoma can be used to define a true cohort, in which incidence can be determined. To limit loss to follow-up bias, the analysis was confined to the 2-year period (4 to 27 months) after the onset of AIDS.

In data presented in Figure 2, the analysis was further restricted to white men so as to minimize the impact of race and sex, and age was adjusted to accommodate temporal changes in the age distribution of the men developing AIDS. The time span permitted a comparison of incidence determinations for the pretherapy periods (1980-1983 and 1983-1986), the era of monotherapies (1987-1989), and the time when dual therapies were available (1990-1993 and 1994-1996). Cancer data were available for only a few of the registries through 1996, and cancer risk in the true triple-therapy HAART era is therefore incomplete. Except for the earliest period, the number of Kaposi’s sarcoma cases in each period was impressive (total: 5,312 in homosexual men and 563 in heterosexual men).

After the introduction of zidovudine in 1987, the incidence of Kaposi’s sarcoma declined steadily until the last data group in 1994-1996 (Figure 2). This decline might have been affected by changes in therapy, because antiretroviral strategies continued to improve in each period after 1986 and also became more widely distributed.

It should be noted that when the last available data for this analysis were generated (1994-1996), relatively few men would have received protease-containing antiretroviral therapy. However, the proportion of men who received optimal HIV-suppressive therapies would likely have been higher among white homosexuals (the group at highest risk for Kaposi’s sarcoma) than in the general population of white heterosexuals.

Thus, the evidence strongly suggests that HAART has further reduced the incidence of Kaposi’s sarcoma. That incidence declined by about two-thirds from 1994-1995 to 1996 and later, when HAART became widely available. This decline occurred when the incidence of Kaposi’s sarcoma in persons with AIDS had already decreased from earlier years. Kaposi’s sarcoma is now uncommon among persons with AIDS who are receiving effective antiretroviral therapy.