To our knowledge, this study is the largest to date to systematically evaluate risk of HPV-associated cancers among persons with AIDS. We found that risks for cancers of the anus, cervix, oropharynx, penis, and vagina or vulva were statistically significantly higher among persons with AIDS than among the general population. Risks of all HPV-associated in situ cancers and risks of invasive cancers of the anus and vagina or vulva increased statistically significantly from 5 years before to 5 years after AIDS onset. Furthermore, a low CD4 T-cell count at AIDS onset was associated with statistically significantly increased risks of invasive cancers of the anus among men and of in situ and invasive vagina or vulva cancers but with a non–statistically significant increased risk of invasive cervical cancer. Finally, in the HAART era (1996–2004), we found that incidence of anal cancer among men increased statistically significantly but that incidences of cancers of the cervix, oropharynx, penis, and vagina or vulva have remained stable.
The elevated risk of HPV-associated cancers among persons with AIDS that we observed is consistent with results from several previous studies (3
) and reflects the increased incidence, prevalence, and persistence of HPV infections as well as a high prevalence of cofactors for such cancers, such as cigarette smoking, among persons infected with HIV (9
). The increased risk of cervical cancer among women with AIDS may also reflect limited access to or use of screening programs for cervical cancer. The degree of immunosuppression, as reflected by a low CD4 T-cell count, is associated with increased HPV persistence and progression in severity of precursor lesions (9
). Furthermore, increasing time after AIDS onset has been associated with an increased risk of in situ HPV-associated cancers (4
). We might expect these associations of immunosuppression with earlier stages of HPV-associated carcinogenesis to also translate to associations with risk of invasive cancers. Previous analyses (4
), however, have not found a direct relationship between immunosuppression and risk of invasive cancer.
In this study, a low CD4 T-cell count at AIDS diagnosis was associated with a statistically significantly increased risk of invasive cancer of the anus (among men) and with a non–statistically significantly increased risk of invasive cervical cancer. These associations of CD4 T-cell count with risk of HPV-related invasive cancers in the HAART era (1996–2004) are in contrast to observations from our previous analyses (4,16), which were restricted to the pre-HAART era (before 1996). We hypothesize that high mortality among individuals with a low CD4 T-cell count during the pre-HAART era masked an association between immunosuppression and the risk of HPV-related invasive cancer and that the increased survival during the HAART era provides adequate time for progression of premalignant or in situ lesions to invasive cancers. It should be noted that we used the CD4 T-cell count at AIDS onset as a marker for the degree of immunosuppression. However, during the HAART era, CD4 T-cell counts at AIDS onset would be expected to increase during the 4–60 months after AIDS onset. Systematic data on CD4 T-cell counts were not available and so repeated CD4 T-cell counts from individuals with AIDS could not be analyzed. Nonetheless, previous studies (28
) have shown that improvements in CD4 T-cell counts that were related to HAART were strongly associated with the baseline CD4 T-cell count at the initiation of HAART (ie, individuals with a low CD4 T-cell count at the initiation of therapy continued to remain relatively more immunosuppressed than those with a high CD4 T-cell count).
For many individuals in the HAART era (1996–2004), the CD4 T-cell count at AIDS onset can be considered the “nadir” level of immunosuppression. Likewise, increased risk of HPV-associated cancers observed with increased time relative to the onset of AIDS ( and and ) might occur because of prolonged but somewhat stable immunosuppression rather than to strictly advancing immunosuppression. Under either scenario, our observations indicate that HIV-related immunosuppression can influence later stages of the development of HPV-related cancers. It should, however, be noted that the association of CD4 T-cell count with invasive cancer risk over an extended period of time (eg, the 4–60 months after AIDS) is in contrast to results for Kaposi sarcoma and non-Hodgkin lymphoma, for which associations were strongest in the 6 months after the CD4 T-cell count was measured and then weakened statistically significantly thereafter (22
). Therefore, given the long time interval between CD4 T-cell count and cancer incidence, it is unclear when immunosuppression is etiologically relevant during carcinogenesis of an HPV-related cancer. It is possible that poor immune control of premalignant lesions (eg, during earlier stages of HIV infection or at AIDS onset) facilitates the development of cancer and that later progression to invasive cancer is not affected by immunosuppression (9
In contrast to the high risk of other HPV-associated cancers (SIRs of 5.3–34.6), the risk of invasive oropharyngeal cancer was only modestly statistically significantly elevated among persons with AIDS when compared with the general population (SIR = 1.6, 95% CI = 1.2 to 2.1). This modest increase in risk may reflect the fact that only a subset of oropharyngeal cancers is caused by HPV. Contemporary estimates (30
) indicate that 60%–70% of oropharyngeal cancers may be HPV related; however, this proportion may have varied over calendar time (31
). Furthermore, although unrelated to AIDS-relative time (5 years before to 5 years after AIDS onset), we observed that risk of oropharyngeal cancer was higher among persons with a relatively higher CD4 T-cell count at AIDS onset than among those with a lower count. The reasons for this paradoxical observation are unclear, but these results are consistent with reports (32
) of increased oral HPV prevalence and persistence among HAART users and the increased incidence of oral warts among HIV-infected individuals after initiation of HAART regimens. Additional studies with information on the HPV status of individual tumors are needed to fully characterize the risk of HPV-associated oropharynx cancers among persons with AIDS.
HAART has not substantially altered HPV persistence or the rates of progression or regression of premalignant anogenital lesions (35
), indicating that HAART may not restore HPV-specific immunity (9
). Consequently, it has been speculated that the incidence of HPV-associated cancers may increase as persons with AIDS live longer in the HAART era (6
). Consistent with this hypothesis and with recent reports (18
), we observed that the incidence of in situ and invasive anal cancer among men was statistically significantly higher in the HAART era (1996–2004) than in the pre-HAART era (1980–1995). The incidence of in situ and invasive anal cancer was non–statistically significantly higher among women during the HAART era than during the pre-HAART era, although these analyses may not have had sufficient power to detect statistically significant differences. The increase in anal cancer in the HAART era may reflect the possibility that in the pre-HAART era, individuals at highest risk of anal cancer would have died from other causes. In addition, the increased incidence of anal cancer among persons with AIDS could partly reflect temporal changes in the general population such as a birth cohort effect (39
Another explanation for the recent rise in the incidence of anal cancers may relate to changes in screening practices. Among cancers that we evaluated, only cervical cancer and anal cancer have screening methods, and an apparent increase in incidence for anal cancer could have arisen if persons with AIDS were screened more intensively for this cancer in the HAART era than in the pre-HAART era. However, our findings of a similar stage at diagnosis and similar proportions of in situ anal cancers in 1990–1995 and in 1996–2004 argue against more intensive screening being involved.
Our study had several strengths. These strengths include its large size and the systematic evaluation of the risk of HPV-associated cancers among a representative sample of persons with AIDS in the United States.
Our study also had several limitations. We did not have information on cofactors for HPV-related cancers, such as cigarette smoking. The strategies that we used to assess relationships with immunosuppression also have limitations. For example, an increase in cancer incidence across AIDS-relative time (5 years before to 5 years after AIDS onset) may partly reflect an increase in medical surveillance after an AIDS diagnosis. Although we found associations between CD4 T-cell count measured at AIDS onset and the risk of HPV-related cancers for up to 5 years after the onset of AIDS, one CD4 T-cell count will not reflect fluctuations in immunosuppression over an extended period. Additionally, because all subjects in our study had AIDS, the low range of CD4 T-cell counts could have masked some associations. We did not have systematic information on HIV viral load, which is another marker of the degree of immunosuppression (40
) that may be particularly relevant in the natural history of HPV infections and associated premalignant lesions (12
). Finally, testing of multiple hypotheses in our analyses could have led to some chance associations, and our results must be interpreted in light of our sensitivity analyses that incorporated Bonferroni corrections. Nevertheless, we note that most associations that were statistically significant at the threshold of a P
value of less than .05 were robust even at conservative Bonferroni-corrected thresholds of less than .005 or .004, including the increased incidence of in situ and invasive anal cancer among men during the HAART era.
In conclusion, we found an elevated risk of HPV-associated cancers among persons with AIDS. The increasing incidence for anal cancer during 1996–2004 indicates that prolonged survival may be associated with increased risk of certain HPV-associated cancers. This increase and the lack of decline in incidence of other HPV-associated cancers indicate that the risk of HPV-associated cancers among persons with AIDS remains high in the HAART era. Given that individuals currently infected with HIV infection may obtain little benefit from available HPV vaccines (because of a low proportion of individuals presumably being naive to vaccine-targeted HPV types) (17,41), our results underscore the need for effective screening for cervical cancer and anal cancer among persons with HIV infection or AIDS.