Studies that have linked AIDS and cancer registries have provided valuable information on the spectrum of malignant disorders associated with advanced HIV infection [1–4]. Since the mid-1990s, the introduction of potent antiretroviral therapy has lead to reductions in AIDS incidence, of more than 70% in treated cohorts [5,6]. Consequently, in countries with good access to antiretroviral therapies, people diagnosed with AIDS now represent an increasing minority of people with HIV, and most people with HIV are now living with only mild to moderate degrees of immune deficiency for extended periods of time without developing AIDS.
To gain a clearer understanding of morbidity, including cancer, that occurs prior to the occurrence of advanced immune deficiency, it is therefore important that studies of people with HIV include both people with and without AIDS. Although large cohort studies of populations with HIV have contributed some data on the spectrum of malignant diseases associated with HIV , linkage studies of HIV and cancer registries have not yet been reported.
In Australia, there has been a nation-wide registration of cancer since 1980 and notification of HIV diagnosis since 1985. We performed a population-based linkage study between databases relating to HIV infection and AIDS and cancer registries to assess the relative risk of non-AIDS-defining cancer in HIV-infected people in Australia. In particular, we aimed to describe cancer incidence in people with early HIV disease, a stage of HIV disease in which cancer incidence has been little studied.
Materials and methods
In Australia, HIV and AIDS are reported with a name code (first two letters of first and last name), date of birth and sex. We attempted linkage for all those individuals who had complete data for these three fields on either the national HIV or AIDS database. The procedure of linkage with the cancer register has been described in detail previously . Briefly, linkage was performed using a modified version of the national cancer register, with full names converted to name codes. A match was accepted if there was an exact match on all three fields, or there was a near match supported by consistency between the registers in dates of death and area of residence. Linkage was performed separately for the HIV and AIDS databases. HIV and AIDS data were available to December 1999, and cancer data until 1995 to 1998 depending on jurisdiction. The sex, age and HIV risk behaviour of individuals included in the analysis is presented in Table 1.
Cancer incidence rates were calculated using the person–years methods. For each type of cancer, person–years at risk were calculated as outlined in Table 2. Exploratory analyses showed that calculated rates of cancer in the 5 years prior to AIDS diagnosis gave cancer rates which were unbiased estimates of cancer rates in people with HIV infection before AIDS diagnosis . For this reason, all people recorded as having AIDS were assumed to have had HIV infection for 5 years before the AIDS diagnosis, and to be under follow-up for the purpose of the estimation of cancer incidence. Thus, for a person with a date of AIDS diagnosis who did not appear on the HIV register, follow-up began 5 years prior to date of AIDS. Standardized incidence ratios (SIRs) were calculated based on age-, sex-, year- and state-specific incidence rates of cancer in the general population. Confidence intervals and time trends were estimated assuming that the observed cancers were Poisson distributed .
To assess the association between cancer incidence and immune deficiency, four time periods were defined based on increasing HIV disease progression. A single individual could contribute to each of the time periods, depending on available follow-up. Period one (least immune deficient) was defined as the period from date of registered HIV infection to 5 years prior to date of AIDS, or the entire period from HIV infection in those who were not recorded as having progressed to AIDS during follow-up. Period two was from 5 years to 6 months prior to AIDS diagnosis. Period three spanned 6 months either side of AIDS diagnosis, and period four (most immune deficient) was from 6 months to 2 years after AIDS diagnosis. Trends in cancer risk over these periods were assessed separately including and excluding the 6-month period around AIDS, to allow for the possibility that diagnostic bias might increase diagnosis of cancer around the time of diagnosis of AIDS.
By August 1999, 46% (8108) of people reported with HIV, and 100% of people with AIDS, had a name code, date of birth and sex recorded and were therefore eligible for linkage with the cancer registry. In total, 1355 cancers, including 196 non-AIDS-defining cancers, were registered in 13 067 people with HIV or AIDS. Results for AIDS-defining cancers have been presented elsewhere .
SIRs in people with HIV and/or AIDS are shown in Table 3. There were significantly increased rates of several cancers, including cancer of lip (10 cases, nine of which were squamous cell carcinoma), anus (10 cases, eight of which were squamous cell carcinoma), connective tissue (20 cases, 10 of which had no histological confirmation, five were haemangiosarcomata, and one sarcoma not specified), Hodgkin's disease (15 cases, of which six were of mixed cellularity, three lymphocyte depleted, one nodular sclerosis, and five were not specified), myeloma, and leukaemia (13 cases including a variety of subtypes with no more than three of each subtype). SIRs were not significantly raised for lung cancer (17 cases, of which six were large cell, four were squamous cell, and three were adenocarcinoma) and cancer of the testis (10 cases, of which seven were seminoma). Rates of colon cancer were significantly decreased (three cases).
Overall, 37% of person–years were in period 1 (least immune deficient), 46% in period 2, 9.0% in period 3, and 7.4% in period 4 (most immune deficient). In period 1, rates were significantly increased only for cancer of the anus, liver and testis (first column of Table 4). Incidence rates of most cancers were increased in period 3 (the period 6 months either side of AIDS). Of the cancers that occurred at increased rates overall, there were significant increasing trends in cancer rates across the four periods for connective tissue cancer, Hodgkin's disease and multiple myeloma. If the period including the 6 months around AIDS was excluded, a significantly increasing trend was only seen for connective tissue cancer.
This study is the first to report linkage between population-based HIV and cancer registers. By separately linking data from both the HIV and AIDS registers, we were able to examine cancer rates across the spectrum of HIV infection from mild to severe immune deficiency, thereby extending the results of previous AIDS–cancer linkage studies. Thirty-seven percent of the total person–years of follow-up were in period one, which consisted of people at a less advanced stage of HIV infection than those included in previous linkage studies. Most of the person–years in this period (92%) were in those who never developed AIDS during follow-up rather than those who were HIV-positive for more than 5 years before AIDS. In this period of early HIV infection, of the cancers that occurred at increased rates overall, incidence was increased only for anal cancer, a disease known to occur at increased rates in HIV-negative homosexual men, who have been estimated to comprise about 85% of people with HIV in Australia . The absence of increases in rates in period one of cancers that were increased in later HIV infection, such as cancer of the lip and Hodgkin's disease, is somewhat reassuring that mild immune deficiency is not associated with increased cancer risk. We demonstrated a significant association of cancer risk with increasing immune deficiency over the four time periods for Hodgkin's disease, multiple myeloma, and connective tissue cancer, but there was reason to believe the association may have been related to diagnostic confusion with Kaposi's sarcoma for connective tissue cancer.
The main limitation of this study is its low power, despite the inclusion of 46% of people with HIV and all people notified with AIDS in Australia. Only 41 cancers were documented in the least immune deficient follow-up period. Another limitation was that our dataset contains very little data on HIV-infected women: in Australia only 7% of all HIV diagnoses to 1999 had been in women . It is likely that linkage identified the great majority of pathologically diagnosed cancers, as we have previously demonstrated that our matching procedure is highly sensitive and specific for identifying non-Hodgkin's lymphoma (NHL) in people with AIDS . It is likely that the matching would be similarly efficient for identifying cancers usually diagnosed pathologically. We have also demonstrated that AIDS registration in Australia is 97% complete for individuals diagnosed with AIDS–NHL , so it is unlikely that a substantial proportion of our cohort had AIDS that had not been registered as such.
Anal cancer is aetiologically related to infection with oncogenic subtypes of human papilloma viruses, and an increased risk in homosexual men was well described before the HIV epidemic . Although risk of anal intra-epithelial neoplasia has been associated with a degree of immune deficiency in cohort studies of homosexual men HIV infection , risk of carcinoma in situ and of invasive cancer was not associated with time before and after AIDS diagnosis in the largest AIDS–cancer linkage study , and did not tend to occur near the time of AIDS diagnosis in a cohort study . In our study, anal cancer rates increased from HIV diagnosis to the time around AIDS diagnosis, but there were no diagnoses more than 6 months after AIDS. Even in people with the mildest degree of immune deficiency rates of this cancer were increased 20-fold above that expected. Thus, our data do not support an independent association between invasive anal cancer and HIV-associated immune deficiency, but it may be that prolonged observation will be necessary. In transplant recipients, anogenital cancers occur at increased rates and appear much later after transplantation than do Kaposi's sarcoma and NHL .
Rates of Hodgkin's disease have been described as occurring at elevated rates in several cohort and linkage studies [1–4,16] and an association of risk with increasing immune deficiency has been previously reported [3,4]. Our finding that there were no cases of Hodgkin's disease in people in period one adds strength to the hypothesis that Hodgkin's disease is related to immune deficiency in people with HIV infection. This is supported by findings in clinical studies that Hodgkin's disease tends to occur at a median CD4 positive T-cell count of 200–300 × 106 cells/l .
Lip cancer occurs at increased rates in immune-suppressed transplant recipients  and shares aetiological risk factors with squamous cell carcinoma of the skin, which occurs at greatly increased rates in transplant recipients . Increased rates of lip cancer in people with AIDS have been described in Australia and the US [3,4]. The fact that rates were not increased in period one in this study is reassuring, but further study is needed to determine whether rates may be increased in long-term survivors of HIV infection. Cohort studies of people with HIV should collect data on lip cancer, and on skin cancers, to help determine whether these cancers will become a long term complication of HIV infection, as they are of long term iatrogenic immune suppression . As data on non-melanoma skin cancers are collected by few cancer registries, it is essential that data on these cancers are collected in cohort studies of people with HIV.
There have been several case series of multiple myeloma occurring in people with AIDS but there are few data from cohort studies. The largest reported AIDS–cancer linkage study found that incidence was raised 2.6-fold but it did not increase significantly from the pre- to post-AIDS period . An AIDS–cancer linkage study in Italy did not find an increased risk of myeloma .
Apart from anal cancer, Hodgkin's disease, lip cancer and multiple myeloma, there were other cancers that occurred at increased rates but there was reason to believe that the increased rates may not have reflected a true increase in incidence. For connective tissue cancers, 50% of registered cancers were not histologically confirmed, and many of those that were confirmed were haemangiosarcomata or unspecified sarcoma, and may have been diagnostically confused with Kaposi's sarcoma. For brain cancer, four of the sven diagnoses occurred around AIDS diagnosis, and as two of these four had no histological confirmation, there may have been diagnostic confusion with NHL.
There have been several cohort studies of cancer incidence in HIV-infected people, some without an AIDS diagnosis [7,16,21], but these have not reported on associations between immune deficiency and cancer, nor have they reported separately on cancer incidence by time period before or after AIDS diagnosis. Previous AIDS–cancer linkage studies [1–4] are unable to contribute data on the incidence of cancer in people with early HIV infection, and the tendency for cancer diagnosis to occur near the time of AIDS diagnosis shown in these and the current study may lead to over-estimation of the overall magnitude of the association between non-AIDS-defining cancers and AIDS.
As the incidence of AIDS has decreased among people with HIV treated with potent antiretroviral therapy, it has become important that other causes of morbidity are documented. The increasingly long-term survival of people with HIV with mild immune deficiency requires us to examine whether or not they will experience increased risk of certain cancers. Data from transplant recipients suggests that certain cancers, in particular cancers of the anogenital region and skin, may occur at increased rates. Although our study provides some reassurance that cancer rates are not markedly increased in early HIV infection, linkage of cancer and HIV registers in larger populations over a long duration of follow-up will be necessary to further document the cancer experience of this population, and the association between mild immune deficiency and cancer.
The authors wish to acknowledge the assistance of the Australian Institute of Health and Welfare, and the South Australian Cancer Register, who undertook the record linkage component of this work, and the National HIV Surveillance Committee. The National HIV Surveillance Committee comprises the following State/Territory and national representatives: Ms Irene Passaris (ACT); Mr Robert Menzies (NSW); Dr Jan Savage (NT); Dr Hugo Ree (QLD); Ms Therese Davey (SA); Mr Neil Cremasco (TAS); Ms Cathy Keenan (VIC); Dr Gary Dowse (WA); Professor John Kaldor (NCHECR) and Ms Ann McDonald (NCHECR).
1. Goedert JJ, Cote TR, Virgo P. et al
. Spectrum of AIDS-associated malignant disorders. Lancet 1998, 358: 1833–1839.
2. Franceschi S, Dal Maso L, Arniani S. et al
. Risk for cancer other than KS and NHL in persons with AIDS in Italy. Brit J Cancer 1998, 78: 966–970.
3. Grulich A, Wan X, Law M, Coates M, Kaldor J. Cancer incidence rates in people with AIDS in NSW, Australia. AIDS 1999, 13: 839–843.
4. Frisch M, Biggar RJ, Engels EA, Goedert JJ. Association of cancer with AIDS-related immunosuppression in adults. JAMA 2001, 285: 1736–1745.
5. Egger M, Hirschel B, Francioli P. et al
. Impact of new anti-retroviral therapies in HIV infected patients in Switzerland: prospective cohort study. BMJ 1997, 315: 1194–1199.
6. Moore RD, Chaisson RE. Natural history of HIV infection in the era of combination antiretroviral therapy. AIDS 1999, 13: 1933–1942.
7. Jones JL, Hanson DL, Dworkin MS, Ward JW, Jaffe HW and the Adult/Adolescent Spectrum of HIV Disease Project Group. Effect of antiretroviral therapy on recent trends in selected cancers among HIV-infected persons. J Acquir Immune Defic Syndr 1999, 21: S11–S17.
8. Grulich AE, Wan X, Coates M, Day P, Kaldor JM. Validation of a non-identifying method of linking cancer and AIDS register data. J Epidemiol Biostat 1996, 1: 207–212.
9. Li Y, Law M, McDonald A, Correll P, Kaldor JM, Grulich AE. Estimation of non-AIDS defining cancer incidence in people with HIV infection prior to the occurrence of AIDS. Am J Epidemiol 2002, 155: 153–158.
10. Breslow NE, Day N. Statistical methods in cancer research, Volume 2- the design and analysis of cohort studies.
Lyon: IARC; 1987
11. Grulich AE, Li Y, McDonald A, Correll PK, Law MG, Kaldor JM. Decreasing rates of HIV-associated Kaposi's sarcoma and non-Hodgkin's lymphoma in the era of highly active anti-retroviral therapy: national linkage study. AIDS 2001, 15: 629–633.
12. National Centre in HIV Epidemiology and Clinical Research. HIV/AIDS, Hepatitis C and Sexually Transmissible Infections in Australia, Annual Surveillance Report 1999.
Sydney: NCHECR, UNSW; 1999.
13. Biggar RJ, Melbye M. Marital status in relation to Kaposi's sarcoma, non-Hodgkin's lymphoma and anal cancer in the pre-AIDS era. J Acquir Immune Defic Syndr Hum Retrovir, 1996, 11: 178–182.
14. Critchlow CW, Surawicz CM, Holmes KK. et al
. Prospective study of high grade anal squamous intra-epithelial neoplasia in a cohort of homosexual men: influence of HIV infection, immunosuppression and human papillomavirus infection. AIDS 1995, 9: 1255–1262.
15. Frisch M, Biggar RJ, Goedert JJ. Human papilloma virus- associated cancers in patients with HIV infection and AIDS. JNCI 2000, 92: 1500–1510.
16. Koblin BA, Hessol NA, Zauber AG. et al
. Increased incidence of cancer among homosexual men, New York City and San Francisco, 1978–1990. Am J Epidemiol 1996, 144: 916–923.
17. Penn I. Cancers of the anogenital region in renal transplant recipients. Cancer 1986, 58: 611–616.
18. Rabkin CS, Blattner WA. HIV infection and cancers other than non-Hodgkin's lymphoma and Kaposi's sarcoma. Cancer Surveys 1991, 10: 151–163.
19. Birkeland SA, Storm HH, Lamm LU. et al
. Cancer risk after renal transplantation in the Nordic countries, 1964–1986. Int J Cancer 1995, 60: 183–189.
20. London NJ, Farmery SM, Will EJ, Davison AM, Lodge PA. Risk of neoplasia in renal transplant recipients. Lancet 1995, 346: 403–406.
21. Serraino D, Boschini A, Carrieri P. et al
. Cancer risk among men with, or at risk of, HIV infection in southern Europe. AIDS 2000, 14: 553–559.