The risk of certain cancers is increased among those infected with HIV. 1 Three cancers are currently considered AIDS defining by the Centers of Disease Control in the United States. 2 These include Kaposi sarcoma (KS), non-Hodgkin lymphoma (NHL), and invasive cervical cancer. Of interest, each of these malignancies has been associated with infection by viruses other than HIV, including human herpes virus type 8 (HHV-8) in KS, 3 Epstein-Barr virus in lymphoma, 4 and oncogenic types of human papilloma virus (HPV) in cervical cancer. 5 Although these viral infections may be necessary for the development of neoplasia, infection alone is not sufficient. Underlying immunosuppression in the host is thought to play an important role in the development of malignancy associated with these organisms. 6,7
Over the past 2 decades, there has been first a rise and then a fall in incidence of AIDS-defining malignancies in the United States, Europe, and Australia. The recent decline in certain AIDS-defining cancers, 8–12 as well as other AIDS-defining opportunistic infections, 13–15 has been attributed to the effectiveness and widespread use of combination antiretroviral therapy.
The effect of potent antiretroviral therapy on the full spectrum of HIV-related cancers is still unknown. If treatment of HIV succeeds at immune restoration, then the incidence of some cancers may decline or at least be delayed. However, the reduced morbidity due to HIV infection and longer life expectancy associated with highly active antiretroviral therapy (HAART) may provide the longer latency period necessary for the development of certain cancers. It will thus be important to study cancer incidence data in HIV-infected individuals over time, to ascertain the true impact of HAART on these cancers. In addition, since HIV infection is more common among individuals with oncogenic exposures, such as cigarette smoking, hepatitis B and C virus (HBV and HCV), 7 evaluating cancer incidence in a comparison group of HIV-uninfected individuals with similar risk factors is particularly important.
The Women’s Interagency HIV Study (WIHS) is a prospective study of 2059 HIV-infected women and 569 HIV-uninfected at-risk women enrolled between 1994 and 1995 in 1 of 6 centers across the United States. The WIHS is uniquely situated to examine cancer trends in women with and at risk for HIV infection. The inclusive dates of the study allow for examination of cancer incidence both before and after the introduction of HAART.
WIHS was established in 1993 to investigate comprehensively the impact of HIV infection in US women. Women with HIV infection and women at risk for acquiring HIV infection were enrolled at 6 sites across the United States between October 1, 1994, and November 15, 1995. Women were considered at risk for HIV infection if they had a history of injection drug use, multiple sex partners in the past 6 months, or sex with someone known to be HIV infected. The institutional review boards of participating institutions approved the study protocol and written informed consent was obtained from all women. Details of the study design, data collection methods, and baseline characteristics have been reported previously. 16 To accrue a representative sample of women living with HIV and AIDS, the eligibility criteria for enrollment into the WIHS did not exclude women in whom clinical AIDS had been previously diagnosed or women with CD4 cell counts < 200/mm3 or 14%.
In brief, recruitment was performed at a variety of venues. Nationally, these included HIV primary care clinics, hospital-based programs, research programs, community outreach sites, women’s support groups, drug rehabilitation programs, HIV testing sites, and referrals from enrolled participants. 16 The HIV-infected and -uninfected women were recruited from similar sources and were well matched on demographic and key risk factors for acquisition of HIV infection.
A total of 2628 women (2059 HIV infected and 569 HIV uninfected) were enrolled in the WIHS at 6 locations within the United States: New York City (2 sites), the Washington, DC, metropolitan area, Chicago, southern California, and northern California. Every 6 months, WIHS participants were interviewed using a structured questionnaire and received a comprehensive physical and gynecologic examination, including Papanicolaou testing and cervical colposcopy. Multiple gynecologic and blood specimens were collected at each visit. Data collected through March 31, 2002 (WIHS visit number 15) were included in this study, but follow-up was censored at September 30, 2001, to allow for delays in the ascertainment of new cancer cases.
Identification of Cancers
Cancers that occurred in women from the WIHS cohort were identified through the following methods and in this hierarchical order of priority: 1) searches of statewide cancer registries; 2) medical record confirmation of self-reported cancer diagnoses; and 3) WIHS-initiated gynecologic biopsies. Cancers that were identified by self-reports or death certificates were included only if they were confirmed by medical record or cancer registry review. Cancers were classified as being incident if they were first diagnosed after enrollment into the WIHS. The state Cancer Registry matches were performed in 2000 for New York and in 2001 for Illinois and California. Cancers that were diagnosed after the period covered by the state cancer registries were confirmed by medical chart review only. Due to the inability to access cancer registry data in the District of Columbia, women enrolled at the Washington, DC, WIHS site were excluded from our analyses.
Date of diagnosis and primary site were obtained for each cancer. Data from the cancer registry also included histologic type and tumor behavior (in situ or malignant). In situ cancers of the anogenital area and basal cell skin cancer were not consistently reported to the cancer registries and were therefore excluded from further analyses. Due to the high rate of misclassification for invasive cervical cancers in the state cancer registries, 17 all cases of invasive cervical cancer reported by the state registries were confirmed by pathology review.
Using all available information, each cancer was classified using the Surveillance, Epidemiology and End Results (SEER) site recoding scheme, which is based on International Classification of Diseases for Oncology, Second Edition, with KS and mesothelioma definitions added to the cancer categories. 18 To determine if there was an excess cancer incidence, we compared the cancer site-specific rates in the WIHS to those among adult women from the national population-based SEER Program for the years 1994–1999 (1999 was the most recent year available from SEER). 19
Risk Factors for Cancer
The characteristics considered in this analysis included demographic characteristics (age, race/ethnicity, education, and HIV risk category), behavioral factors (cigarette smoking, alcohol use, and recreational drug use) as reported at study entry, geographic region, presence of certain viral infections (HBV, HCV, HPV, and HHV-8) at baseline, nadir CD4 count and peak HIV RNA levels, HIV serostatus, HAART use, history of self-reported clinical AIDS, and calendar period.
Laboratory measures included HIV-1, HBV, HCV, and HHV-8 serology, CD4 counts (cells/mm3) and quantitative HIV-RNA levels in plasma. HPV positivity was determined by the presence of HPV DNA for all subtypes in cervicovaginal lavage fluid. 20
We characterized the WIHS cohort at study entry and quantified the incident cancers documented during follow-up using standard descriptive statistics. Cancer incidence rates were computed as the number of cancers first diagnosed following the WIHS baseline visit divided by the total number of person-years of observed follow-up. Multiple incident cancers were included in this analysis and the total follow-up time available for any woman reflected the number of years from the baseline visit until the date of death, loss to follow-up, or September 30, 2001, whichever occurred first.
We compared the observed number of cancers with the expected number derived from SEER data to determine if there was an excess in the rate of cancer within the WIHS cohort. The expected numbers of cancers were computed using sex-, race-, and age-specific rates from SEER using all registered women at all SEER registries and sites between January 1, 1994 and December 31, 1999. While several of the WIHS sites fall within geographically defined regions covered in the SEER registry, this was not the case for all WIHS sites and therefore the data required to standardize by area were not available. Standardized incidence ratios (SIRs) adjusted for age (5-year categories; updated during follow-up) and race (white, Latina/Hispanic, other) were calculated using indirect standardization, 21 and exact 95% CIs and P values 22 were generated separately for HIV-infected and HIV-uninfected women, and, among HIV-infected women, before and on or after January 1, 1997. Statistical comparisons of the cancer incidence rates by HIV serostatus or calendar period were performed using exact Poisson regression (LogXact, release 5.0, Cytel Software Corp., Cambridge, MA).
A total of 1950 women were computer matched with data from state cancer registries in their respective states of WIHS enrollment. Table 1 provides participant characteristics at enrollment for the 396 HIV-uninfected and 1554 HIV-infected women. The majority of women were African American (54%) and, at study entry, were HPV positive (55%) and current smokers (59%). At baseline, 41% of the women were HCV positive, and among the HIV-infected women, 52% had HIV RNA levels > 20,000 copies and 46% had CD4 cell counts < 200.
Five (1.3%) of the 396 women who were HIV uninfected at enrollment seroconverted for HIV during the follow-up period and 442 (23%) of the 1950 women died. Among the 1559 HIV-infected women, 533 (34%) women reported having a clinical AIDS diagnosis at baseline and 378 (24%) reported their first clinical AIDS diagnosis during follow-up. Only 1 woman reported HAART use at study entry and 987 women (63%) initiated HAART during the follow-up period.
A total of 48 incident cancers were diagnosed among 47 WIHS participants; 41 cancer diagnoses were in HIV-infected women and 7 were in HIV-uninfected women (Table 2). One HIV-infected woman had 2 cancer diagnoses, leukemia and breast cancer. Overall, the most common incident cancers were NHL (n = 12), lung/larynx cancer (n = 10), breast cancer (n = 7), and KS (n = 6). Of the 48 incident cancers, 20 were ascertained by cancer registry alone, 11 by medical record abstraction alone, and 17 from both sources. Cancer registry ascertainment was obtained for 77% of the 48 incident cancers and did not differ by HIV status (71% among HIV-negative women vs. 78% among HIV-positive women, exact P = 1.0). Stratified by time period, 90% of the cancers were documented via registry match prior to 1997 while 67% were documented by registry match during 1997 or later (exact P = 0.08). Among the individual cancers, only melanoma was significantly different from all other cancers in regards to ascertainment source, with 0 of 2 melanomas having been ascertained from cancer registries vs. 37 of the 46 other cancers (exact P = 0.049).
The number of incident cancers observed in the WIHS was compared with the expected number of cancers based on population-based SEER data (Table 2). Overall, HIV-infected WIHS women had about twice as many incident cancers as would have been expected among women of similar age and ethnicity in the general female population of the United States (SIR = 1.9, 95% CI = 1.3–2.5). Among HIV-infected women, significantly (P < 0.05) increased incidence rates were observed for KS (SIR = 213.5), NHL (SIR = 19.0), and lung cancer (SIR = 6.3). The incidence of gallbladder cancer (SIR = 98.7) was significantly elevated among the HIV-uninfected women when compared with population-based rates, although this significant result was based upon only 1 case. Also, the SIR for lung cancer among the HIV-uninfected women was borderline significant (SIR = 6.9, 95% CI = 0.8–19.3). No excess was observed for breast cancer (5 cases among HIV-infected women and 2 cases among HIV-uninfected women), cervical cancer (1 case among HIV-infected women and none among HIV-uninfected women), or hepatocellular carcinomas (no cases in either HIV-infected or uninfected women).
The Poisson regression comparisons of the cancer-specific incidence rates for the HIV-infected vs. the HIV-uninfected women are shown in Table 3. The only statistically significant difference in cancer incidence between HIV-infected and -uninfected women was for all AIDS-defining cancers combined.
Temporal trends of the SIRs for the groups of AIDS and non-AIDS cancer and the 4 most common individual cancers in the WIHS are shown in Table 4. The SIRs for all AIDS-defining cancers as well as KS and NHL drop markedly from the pre-HAART era to the HAART era. Formal comparisons of the cancer incidence rates in these 2 periods using Poisson regression are shown in Table 5. The incidence rate drop over time was statistically significant for all AIDS-defining cancers combined (relative risk [RR] = 0.21, 95% CI = 0.07–0.60) and NHL (RR = 0.15, 95% CI = 0.03–0.61), but not for KS (RR = 0.23, 95% CI = 0.02–1.60). Importantly, however, the SIRs for each of these cancers remained significantly elevated during the HAART era (Table 4). In contrast, there is no evidence of change over time for the non-AIDS defining cancers (Tables 4 and 5).
Among the 3 AIDS-defining cancers, we found an increased incidence of KS and NHL but no increase of invasive cervical cancer in the WIHS cohort. Although invasive cervical cancer is included in the AIDS case definition and has been found in excess in other cohort studies of HIV-infected women, 23–25 our finding of no excess may be due to the intensive cytologic surveillance and precursor treatment in our cohort. In fact, a previous WIHS study of abnormal cervical cytology found fairly high rates of atypical squamous cells of undetermined significance (ASCUS) (19%) and low-grade squamous intraepithelial lesions (SIL) (14%) in this cohort with much lower rates of high-grade disease (1.3%). 26 Thus, with gynecologic examinations performed every 6 months and easy access to treatment of precursor lesions in our cohort of women, our results may not be representative of women who are not in regular cancer screening programs. The fact that we did not see an increase in invasive cervical cancer may also be due to the potential misclassification of precursor lesions as invasive cancer by cancer registries. A more detailed examination of invasive cervical cancer in the WIHS is the subject of another investigation. 17 In that study, we found that a substantial number of women who were coded as having invasive cervical cancer in their regional cancer registries actually had cervical dysplasia as determined by pathologic review. This misclassification would artificially increase the rates of invasive cervical cancer, and studies that relied exclusively on cancer registries for identification of such malignancies would appear to have higher rates.
Regarding the non-AIDS-defining cancers, we found a significantly increased incidence of lung cancer among HIV-infected WIHS participants, and although it was not statistically significant, we observed a similarly increased incidence of lung cancer among HIV-uninfected women. Lung cancer has been found in excess in other studies of women with HIV and AIDS. 24,27,28 However, the current study suggests that high rates of lung cancer are not associated with HIV infection. Instead, they are likely attributable to high rates of cigarette smoking found in both our HIV-infected and uninfected women. Also, as reported in previous studies, 27–29 no excess risk was seen for breast cancer in the WIHS. While some studies have found a significantly increased risk for anal cancer in women with AIDS, 28,30 we and others 29 did not.
The WIHS is strategically positioned to look at cancer risk preceding and during the HAART era. We found evidence of a significant decrease in cancer incidence for NHL since the introduction of HAART. This supports the findings of other studies, which have documented a fall in the rates of certain AIDS-defining cancers in recent years. 10–12 However, we also found that the incidence of NHL remains significantly elevated among our HIV-infected cohort. In addition, our finding of no difference in the incidence of lung/laryngeal cancer, either by HIV serostatus or over time, suggests that other factors such as smoking have a greater influence on cancer incidence than HIV infection or HAART use.
The cancer incidence measured in this cohort probably underestimated the true incidence, in part due to migration of participants out of their state of enrollment. For example, if a California WIHS participant moves to Oregon where her cancer was diagnosed, it is possible that her new cancer would have been missed since we only performed a registry match with the California Cancer Registry. Our estimate of the excess in cancer over the US adult female population may also be conservative since it is based on comparisons with SEER data for years 1994–1999, which would include HIV/AIDS-related cancers. However, we found no evidence of differential ascertainment bias by HIV status, although our data suggest that ascertainment source may vary over time. This variation is partly explained by the lag time for new cancer diagnoses to be reported to state cancer registries, and so cancer occurring after 1999 may be underreported. Finally, an estimation of the full impact of specific cancers among women who died prematurely of AIDS may not be possible.
There are additional limitations to this investigation. The length of follow-up so far is relatively short and therefore the number of incident cancers is rather small. Despite the low number of incident cancers, we were able to detect significantly elevated rates of KS, NHL, and lung cancer. While other cancers appear to be increased as well, caution should be used when interpreting the excess risk observed for rarely observed cancers that only occurred once, such as gallbladder cancer occurring in 1 HIV-uninfected participant and vaginal cancer occurring in 1 HIV-infected participant. Another limitation may be our reliance on secondary sources of cancer information, such as cancer registry and medical record abstraction information. However, we were conservative in our interpretations, requiring these validated measures for case inclusion and, in the case of invasive cervical cancer, requiring pathologic review. Lastly, a relatively small proportion of the WIHS cohort is HIV uninfected (20%), and only 7 incident cancers occurred among the 396 HIV-uninfected participants. This greatly limited our statistical power for comparative analyses between the HIV-infected and HIV-uninfected women and prevented us from conducting subgroup analyses of only HIV-uninfected WIHS women. Nonetheless, we found that the overall cancer incidence rate among HIV-infected women was higher than among HIV-uninfected women (RR = 1.6, 95% CI = 0.7–4.2) although this difference was not statistically significant.
In summary, the WIHS data have demonstrated increased incidence rates for KS and NHL among HIV-infected women. Compared with the general population, lung cancer incidence was also significantly elevated among HIV infected women, and it seems to be elevated among HIV-uninfected women although additional follow-up will be needed for confirmation. The SIRs for KS and NHL decreased since the introduction of HAART, but they remained significantly higher than the rates in the general population. In the HAART era, more intensive cancer screening and prevention efforts are warranted in this population, given the higher rates of certain malignancies among HIV-infected women coupled with their increased life expectancy.
Data were collected by the WIHS Collaborative Study Group with centers (Principal Investigators) at New York City/Bronx Consortium (Kathryn Anastos); Brooklyn, NY (Howard Minkoff); Washington, DC, Metropolitan Consortium (Mary Young); The Connie Wofsy Study Consortium of Northern California (Ruth Greenblatt); Los Angeles County/Southern California Consortium (Alexandra Levine); Chicago Consortium (Mardge Cohen); and Data Coordinating Center (Alvaro Muñoz).
1. Hessol NA. The changing epidemiology of HIV-related cancers. AIDS Reader
2. CDC. 1993 Revised classification system for HIV infection
and expanded surveillance case definition for AIDS
among adolescents and adults. Morb Mortal Wkly Rep
3. Chang Y, Cesarman E, Pessin M, et al. Identification of herpesvirus-like DNA sequences in AIDS
-associated Kaposi’s sarcoma. Science
. 1994; 266:1865–1869.
4. Epstein M, Achong B, Barr Y. Virus particles in cultured lymphoblasts from Burkitt’s lymphoma. Lancet
5. Zur Hausen H. Papillomaviruses in human cancer
. Appl Pathol
6. Penn I. Depressed immunity and the development of cancer
. Cancer Detect Prev
7. Rabkin CS. AIDS
in the era of highly active antiretroviral therapy (HAART). Eur J Cancer
8. Buchbinder SP, Holmberg SD, Scheer S, et al. Combination antiretroviral therapy and incidence of AIDS
-related malignancies. J Acquir Immune Defic Syndr
. 1999;21(suppl 1):S23–S26.
9. Jones JL, Hanson DL, Dworkin MS, et al. Effect of antiretroviral therapy on recent trends in selected cancers among HIV-infected persons. Adult/Adolescent Spectrum of HIV Disease Project Group. J Acquir Immune Defic Syndr
. 1999;21(suppl 1):S11–S17.
10. Sparano JA, Anand K, Desai J, et al. Effect of highly active antiretroviral therapy on the incidence of HIV-associated malignancies at an urban medical center. J Acquir Immune Defic Syndr
. 1999;21(suppl 1): S18–S22.
11. Highly active antiretroviral therapy and incidence of cancer
in human immunodeficiency virus-infected adults. International Collaboration on HIV and Cancer
. J Natl Cancer Inst
12. Grulich AE, Li Y, McDonald AM, et al. Decreasing rates of Kaposi’s sarcoma and non-Hodgkin’s lymphoma in the era of potent combination anti-retroviral therapy. AIDS
13. Moore R, Chaisson R. Natural history of HIV infection
in the era of combination antiretroviral therapy. AIDS
14. Kaplan JE, Hanson D, Dworkin MS, et al. Epidemiology of human immunodeficiency virus-associated opportunistic infections in the United States in the era of highly active antiretroviral therapy. Clin Infect Dis
. 2000;30(suppl 1):S5–14.
15. Ledergerber B, Egger M, Telenti A. AIDS
-related opportunistic illness and potent antiretroviral therapy. JAMA
16. Barkan SE, Melnick SL, Preston-Martin S, et al. The Women’s Interagency HIV Study. Epidemiology
17. Massad LS, Seaberg EC, Watts H, et al. Low incidence of invasive cervical cancer
among HIV-infected U.S. women in a prevention program. AIDS
18. Surveillance, Epidemiology, and End Results (SEER) Program Public-Use CD-ROM (1973–1997), Appendix H. National Cancer
Institute, DCCPS, Cancer
Surveillance Research Program, Cancer
Statistics Branch, released April 2000, based on the August 1999 submission.
19. Surveillance, Epidemiology, and End Results (SEER) Program Public-Use CD-ROM (1973–1997). National Cancer
Institute, DCCPS, Cancer
Surveillance Research Program, Cancer
Statistics Branch, released April 2000, based on the August 1999 submission.
20. Palefsky JM, Minkoff H, Kalish LA, et al. Cervicovaginal human papillomavirus infection in human immunodeficiency virus-1 (HIV)-positive and high-risk HIV-negative women. J Natl Cancer Inst
21. Inskip H. Standardization methods. In: Gail MH, Benichou J, eds. Encyclopedia of Epidemiologic Methods
. New York: Wiley; 2000:871–884.
22. Newman SC. Biostatistical Methods in Epidemiology
. New York: Wiley; 2001.
23. Dorrucci M, Suligoi B, Serraino D, et al. Incidence of invasive cervical cancer
in a cohort of HIV-seropositive women before and after the introduction of highly active antiretroviral therapy. J Acquir Immune Defic Syndr
24. Phelps RM, Smith DK, Heilig CM, et al. Cancer
incidence in women with or at risk for HIV. Int J Cancer
25. Serraino D, Carrieri P, Pradier C, et al. Risk of invasive cervical cancer
among women with, or at risk for, HIV infection
. Int J Cancer
. 1999;82: 334–337.
26. Massad LS, Ahdieh L, Benning L, et al. Evolution of cervical abnormalities among women with HIV-1: evidence from surveillance cytology in the women’s interagency HIV study. J Acquir Immune Defic Syndr
. 2001; 27:432–442.
27. Frisch M, Biggar RJ, Engels EA, et al. Association of cancer
-related immunosuppression in adults. AIDS
Match Registry Study Group. JAMA
28. Gallagher B, Wang Z, Schymura MJ, et al. Cancer
incidence in New York State acquired immunodeficiency syndrome patients. Am J Epidemiol
29. Dal Maso L, Franceschi S, Polesel J, et al. Risk of cancer
in persons with AIDS
in Italy, 1985–1998. Br J Cancer
30. Frisch M, Biggar RJ, Goedert JJ. Human papillomavirus-associated cancers in patients with human immunodeficiency virus infection and acquired immunodeficiency syndrome. J Natl Cancer Inst
Keywords:© 2004 Lippincott Williams & Wilkins, Inc.
AIDS; cancer; HIV infection; malignancy; standardized incidence ratio