Meredith S. Shiels, PhD, MHS; Ruth M. Pfeiffer, PhD; Eric A. Engels, MD, MPH
Acknowledgment: The authors thank the HIV/AIDS and cancer registry staff in Colorado; Connecticut; Florida; Illinois; Georgia; Massachusetts; Michigan; New Jersey; Texas; New York, New York; San Diego, Los Angeles, and San Francisco, California; Seattle, Washington; and Washington, DC, and Tim McNeel, Information Management Systems, for database management.
Grant Support: By the intramural research program of the National Cancer Institute.
Potential Conflicts of Interest: Disclosures can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M10-1342.
Reproducible Research Statement:Study protocol and data set: Not available. Statistical code: Available from Dr. Shiels (e-mail, email@example.com).
Requests for Single Reprints: Meredith S. Shiels, PhD, MHS, Infections and Immunoepidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Boulevard, EPS 7059, Rockville, MD 20892; e-mail, firstname.lastname@example.org.
Current Author Addresses: Dr. Shiels: Infections and Immunoepidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Boulevard, EPS 7059, Rockville, MD 20892.
Dr. Pfeiffer: Biostatistics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Boulevard, EPS 8030, Rockville, MD 20892.
Dr. Engels: Infections and Immunoepidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, 6120 Executive Boulevard, EPS 7076, Rockville, MD 20892.
Author Contributions: Conception and design: M.S. Shiels, E.A. Engels.
Analysis and interpretation of the data: M.S. Shiels, R.M. Pfeiffer, E.A. Engels.
Drafting of the article: M.S. Shiels, R.M. Pfeiffer, E.A. Engels.
Critical revision of the article for important intellectual content: M.S. Shiels, E.A. Engels.
Final approval of the article: M.S. Shiels, R.M. Pfeiffer, E.A. Engels.
Statistical expertise: M.S. Shiels, R.M. Pfeiffer, E.A. Engels.
Obtaining of funding: E.A. Engels.
Collection and assembly of data: E.A. Engels.
Shiels M., Pfeiffer R., Engels E.; Age at Cancer Diagnosis Among Persons With AIDS in the United States. Ann Intern Med. 2010;153:452-460. doi: 10.7326/0003-4819-153-7-201010050-00008
Download citation file:
Published: Ann Intern Med. 2010;153(7):452-460.
Studies have reported young ages at cancer diagnosis in HIV-infected persons and have suggested that HIV accelerates carcinogenesis. However, these comparisons did not account for differences in population age structures.
To compare ages at diagnosis for non–AIDS-defining types of cancer that occur in both the AIDS and general populations, after adjustment for differences in age and other demographic characteristics between these populations.
Registry linkage study.
15 HIV/AIDS and cancer registry databases in the United States.
212 055 persons with AIDS enrolled in the U.S. HIV/AIDS Cancer Match Study from 1996 to 2007.
Comparison of age-at-diagnosis distributions for various types of cancer in both the AIDS and general populations, after adjustment for age and other demographic characteristics.
The proportion of person-time contributed by older persons (age ≥65 years) was far smaller in the AIDS population (1.5%) than in the general population (12.5%). Reflecting this difference, the ages at diagnosis for most types of cancer were approximately 20 years younger among persons with AIDS. However, after adjustment for differences in the populations at risk, the median ages at diagnosis in the AIDS and general populations did not differ for most types of cancer (for example, colon, prostate, or breast cancer; all P > 0.100). In contrast, ages at diagnosis of lung (median, 50 vs. 54 years) and anal cancer (median, 42 vs. 45 years) were significantly younger in persons with AIDS than expected in the general population (P < 0.001), and the age at diagnosis of Hodgkin lymphoma was significantly older (median, 42 vs. 40 years; P < 0.001).
Information on other cancer risk factors, including cigarette smoking, was not available. Analysis was restricted to non-Hispanic white and black persons who had AIDS, which could limit the generalizability of the findings to other racial and ethnic groups or to persons with HIV but not AIDS.
For most types of cancer, the age at diagnosis is similar in the AIDS and general populations, after adjustment for the ages of the populations at risk. Modest age differences remained for a few types of cancer, which may indicate either acceleration of carcinogenesis by HIV or earlier exposure to cancer risk factors.
National Cancer Institute.
Mari M. Kitahata
University of Washington Center for AIDS Research
October 28, 2010
Dear editor: In the article "Age at cancer diagnosis among persons with AIDS in the United States (1)", Shiels et al use data from "the U.S. HIV/AIDS Cancer Match Study to evaluate the ages at diagnosis of several types of non-AIDS-defining cancer in the HIV and general populations." The investigators state: "After adjustment for differences in the age composition of the populations at risk, the age at cancer diagnosis did not differ between HIV-infected persons and those in the general population for most types of cancer." They conclude: "our results do not support including cancer as part of a general syndrome of premature aging in HIV-infected persons" and "our results do not support an accelerated screening schedule in HIV-infected persons study". However, these conclusions are not supported by this study, and may be erroneous.
The U.S. HIV/AIDS Cancer Match Study only includes HIV-infected individuals diagnosed with an AIDS-defining condition and therefore, is missing data for all cancers occurring in HIV-infected individuals who have not yet developed AIDS. Non-AIDS defining diseases, including non- AIDS-defining cancers, are increasingly prevalent among HIV-infected individuals well before they develop AIDS. The investigators adjusted for differences in the age composition of the populations studied, but did not determine if the age at cancer diagnosis differs between HIV-infected persons and those in the general population, which would require examination of cancers among all HIV-infected persons, not just those who have had an AIDS-defining condition.
In the Center For AIDS Research (CFAR) Network of Integrated Clinical Systems (CNICS), a nationally distributed cohort of over 22,000 HIV- infected individuals in care at eight CFARs across the US from 1995 to the present (2), 64% of individuals with non-AIDS-defining cancers were diagnosed with cancer prior to developing any AIDS-defining condition. If we were to apply the inclusion criteria for the study conducted by Shiels et al to the CNICS cohort, the majority of HIV-infected individuals diagnosed with a non-AIDS-defining cancer would not be included in their analysis. Furthermore, a large proportion of non-AIDS-defining cancers in the CNICS cohort were diagnosed at advanced stages, suggesting that HIV- infected individuals with cancer were identified late using current screening practices for the general population and could benefit from screening for cancers at a younger age.
Mari M Kitahata MD, MPH (a), Chad J. Achenbach MD, MPH (b), Michael S. Saag MD (c)
Author affiliations: (a) Department of Medicine, University of Washington, (b) Department of Medicine, Northwestern University, (c) Department of Medicine, University of Alabama at Birmingham.
1. Shiels MS, Pfeiffer RM, Engels EA. Age at cancer diagnosis among persons with AIDS in the United States. Ann Intern Med. 2010;153(7):452- 60.
2. Kitahata MM, Rodriguez B, Haubrich R, Boswell S, Mathews WC, Lederman MM, et al. Cohort profile: the Centers for AIDS Research Network of Integrated Clinical Systems. International Journal of Epidemiology. 2008;37(5):948-55.
Department of Public Health, Nice University Hospital, France
November 3, 2010
Same age at cancer diagnosis in AIDS people and general population: can we really conclude to no premature aging?
Shiels and colleagues (1) showed that age at cancer diagnosis is often similar in the AIDS and general population after adjusting for age differences. In addition to the related editorial (2) comment about the complexity to define premature aging, we show that, in some cases, the same age-at-diagnosis can hide real differences.
Let's denote IAIDS(a) and Igen(a) the cancer incidence rates at age a respectively among AIDS people and among general population.
Shiels et al. stated that if the risk of cancer is merely increased by HIV, then the age-at-diagnosis curve would have the same shape after age adjustment and authors conclude that HIV does not accelerate carcinogenesis. This can be expressed as a proportional incidence rate by age in AIDS compared to the general population:
(A) there exists a constant b, such that IAIDS(a)= b.Igen(a) for any age a.
Now, let's suppose that HIV does accelerate carcinogenesis. An intuitive, clinical way to express this acceleration is that people with AIDS have the same risk of cancer as people older in the general population (that is a shift to the left of the incidence rate curve by age). In terms of incidence rate, if we consider a 10 year shift, the expression is:
(B) for any age a IAIDS(a)= Igen(a+10).
The point is that expressions (A) and (B) can be verified simultaneously, especially for exponential incidence functions. Indeed, for any cancer with an incidence rate exponential with age, a shifted (premature aging) and a proportional incidence (no premature aging) both give the same result: the same median age-at-diagnosis in AIDS and general population after age adjustment. The exponential trend is often used to model incidence of cancer with age (3). In these cases, an overall increase of cancer rate (A) and a premature aging of the AIDS population (B) are indistinguishable.
We simulated the number of cases observed in the AIDS population as defined in Shiels' paper using assumptions (A) and (B) described above in lung and stomach cancer (4). The figure1 shows that these assumptions gave different results in lung (non exponential incidence distribution) but not in stomach (exponential) cancer.
Figure1: Age-at-diagnosis distributions of cancer in the general population and in AIDS population according to the two hypotheses on incidence.
This illustrates that the intuitive definitions of premature and non premature aging can lead to the same age-at-diagnosis distribution under particular conditions (like exponential incidence rates). Therefore, the adjusted age-at-diagnosis might not be the good indicator to assess cancer-related premature aging in HIV infected persons.
1. Shiels MS, Pfeiffer RM, Engels EA. Age at cancer diagnosis among persons with AIDS in the United States. Ann Intern Med. 2010 Oct 5;153(7):452-60.
2. Martin J, Volberding P. HIV and premature aging: A field still in its infancy. Ann Intern Med. 2010 Oct 5;153(7):477-9.
3. Freddie Bray & Bjorn Moller. Predicting the future burden of cancer. Nat Rev Cancer. 2006 Jan;6(1):63-74. Review.PMID: 16372017.
4. Belot A, Grosclaude P, Bossard N, Jougla E, Benhamou E, Delafosse P, et al. (2008). Cancer incidence and mortality in France over the period 1980-2005. Rev Epidemiol Sante Publique. 2008 Jun;56(3): 159-75. Epub 2008 Jun 10. Detailed results and comments: http://www.invs.sante.fr/surveillance/cancers/estimations_cancers/default.htm [Accessed > 10 26 2010 ].
Meredith S. Shiels
National Cancer Institute
January 28, 2011
Authors' Response to Comments on "Age at Cancer Diagnosis Among Persons with AIDS in the United States"
We recently illustrated that naively comparing ages at cancer diagnosis between people with AIDS and the general population is inappropriate, because the age distributions of the populations at risk differ (1). After correcting for these population differences, we found that people with AIDS do not develop most cancers at younger ages. While others have suggested that elevated cancer risk at all ages could be evidence of accelerated aging (2), we posited that accelerated aging requires particularly elevated cancer risk at young ages, which would lead to a downward shift in the age distribution of cancer diagnoses. If HIV caused premature aging, we would expect both earlier ages at cancer diagnosis and broadly increased cancer risk, particularly for cancers commonly associated with aging (e.g., prostate, colon and breast), which has not been observed (3). We therefore argue that cancer should not be included as part of a general syndrome of HIV-associated accelerated aging Sakarovitch and Fontas demonstrated that when cancer incidence is an exponential function of age, an overall increased cancer risk and a downward age shift are indistinguishable. Though their conclusion is mathematically correct, for most cancers incidence increases as a power of age, not exponentially (4). In addition, if HIV infection disproportionately increased cancer risk at young ages, then cancer incidence in the HIV and general populations might need to be modeled with different functions of age. Kitahata et al. highlighted that our study excluded people with HIV infection prior to AIDS. As previously discussed (1), accelerated carcinogenesis in HIV-infected people is unlikely in the absence of acceleration in people with AIDS, who are the most immunocompromised. Further, Kitahata et al. state that most cancers are excluded when restricting a study to people with AIDS. However, in a recent analysis, we found that only 29% of non-AIDS-defining cancers in HIV-infected people in the U.S. occur before AIDS diagnosis (5). While HIV-infected people may present with advanced-stage cancer, it is unclear whether screening at younger ages would be beneficial. Screening tests for non-AIDS-defining cancers are available for prostate, colorectal and breast cancers. However, HIV is not associated with increased risk or younger ages at diagnosis of these three cancers, and together they comprise ?20% of all non-AIDS-defining cancers (1;3). In addition, HIV- infected individuals may not even receive age-appropriate screening based on current guidelines, which could contribute to advanced stage at diagnosis.
Meredith S. Shiels1 PhD MHS, Ruth M. Pfeiffer1, PhD and Eric A. Engels1, MD MPH 1Division of Cancer Epidemiology and Genetics, National Cancer Institute
Meredith Shiels 6120 Executive Blvd EPS 7059 Rockville, MD 20892 Phone: 301-402-5374 Fax: 301-402-0817 Email: email@example.com
(1) Shiels MS, Pfeiffer RM, Engels EA. Age at cancer diagnosis among persons with AIDS in the United States. Ann Intern Med 2010; 153(7):452- 460.
(2) Martin J, Volberding P. HIV and premature aging: A field still in its infancy. Ann Intern Med 2010; 153(7):477-479.
(3) Engels EA, Biggar RJ, Hall HI, Cross H, Crutchfield A, Finch JL et al. Cancer risk in people infected with human immunodeficiency virus in the United States. Int J Cancer 2008; 123(1):187-194.
(4) Kaldor JM, Day NE. Mathematical models in cancer epidemiology. In: Schottenfeld D, Fraumeni JF, Jr., editors. Cancer Epidemiology and Prevention. Second ed. New York: Oxford University Press; 1996. 127-140.
(5) Shiels MS, Pfeiffer R, Gail MH, Hall HI, Li J, Chaturvedi A et al. The burden of cancer among HIV-infected people in the United States. 2010. Unpublished work.
to gain full access to the content and tools.
Learn more about subscription options.
Register Now for a free account.
Hematology/Oncology, Infectious Disease, HIV.
Results provided by:
Copyright © 2016 American College of Physicians. All Rights Reserved.
Print ISSN: 0003-4819 | Online ISSN: 1539-3704
Conditions of Use
This PDF is available to Subscribers Only