Cancer Linkage With an HIV/STD Registry: Describing the Cancer Morbidity Among HIV/AIDS Patients and Opportunities for Improving Surveillance June 5, 2007 Jennifer Chase, MSPH Texas HIV/STD Epidemiology and Surveillance Branch Melanie Williams, PhD Cancer Epidemiology and Surveillance Branch Nancy Weiss, PhD Cancer Epidemiology and Surveillance Branch Tammy Sajak, MPH Texas HIV/STD Epidemiology and Surveillance Branch Paul Betts, MS Cancer Epidemiology and Surveillance Branch
Acknowledgements National Cancer Institute Eric Engels, MD, MPH Phillip W. Virgo HIV/STD Epidemiology and Surveillance Branch Barry Mitchell, MPH Sharon Melville, MD, MPH
Outline Background HIV/AIDS – Cancer Match –Overview –Findings Next Steps Lessons Learned
Background Match was performed in collaboration with the Viral Epidemiology Branch National Cancer Institute (NCI), the Texas Cancer Epidemiology and Surveillance Branch, and Texas HIV/STD Epidemiology and Surveillance Branch Comprehensively matched the Texas Cancer and HIV/AIDS Registries for the first time as part of the NCI, “Cancer HIV/AIDS Match Study”
Background People with HIV and AIDS are at high risk for developing certain cancers “AIDS-defining” cancers for HIV-infected –Kaposi Sarcoma –Non-Hodgkin’s Lymphoma –Invasive Cervical Cancer Other types of cancer also appear to be more common among persons with persons with AIDS (PWAs)
Background Non-AIDS Defining Cancers Some non-AIDS defining cancers have been found to occur in excess among PWAs –Pancreatic –Laryngeal –Heart (usually melanomas and sarcomas) –Vulva –Vaginal –Kidney –Penile –Soft tissues –Hodgkin’s lymphoma –Anal –Multiple Myeloma –Leukemia –Lung –Oral cavity –Lip –Esophageal –Stomach –Liver
Background HAART and Cancer Highly Active Antiretroviral Therapy (HAART) introduced in Mid 1990s Reductions in some AIDS-defining cancers –Kaposi Sarcoma: dramatic declines –Non-Hodgkin’s Lymphoma: Effect of HAART not clear, some studies show decline –Invasive Cervical Cancer: Studies have been conflicting
Background HAART and Cancer PWAs living longer and sustaining fewer opportunistic infections making cancer a more prominent cause of death –Pre-HAART, 10% of deaths from cancer –Year 2000, 28% from cancer
HIV/AIDS Cancer Match Study Purpose: To link HIV/AIDS registries to cancer registries in multiple sites in the US that represent the major epicenters of the HIV/AIDS epidemic, including Texas (NY, MA, NJ, IL, FL, City of NY, Los Angeles, San Francisco, Seattle, San Diego, Atlanta) Expect 40,000-50,000 cancer cases among 500,000 persons with AIDS Diversity of geography, HIV risk groups, and racial/ethnic minorities
HIV/AIDS Cancer Match Study Objectives: –To monitor cancer incidence in persons with HIV and AIDS –Determine risk of cancer among persons with HIV/AIDS –Determine which factors may be predictive among persons with HIV/AIDS –Determine if cancers in persons with HIV/AIDS differ from cancers among the general population –To increase communication between the local HIV/AIDS and cancer registries to enhance surveillance
HIV/AIDS Cancer Match Study Typically, only 1/3 of AIDS related cancers are recorded in both registries Non-AIDS-related cancers are not reported to HIV/AIDS registries, but cancer registries collect them HIV/AIDS related data are not collected by cancer registries Other data may be missing on one registry and found in another, such as dates of birth or death that can provide a more complete research file
Texas Match Overview Cancer Registry Cases Included: 744,160 cases Texas residents, diagnosed from HIV/AIDS Registry Cases Included: 93,120 HIV positive only and HIV/AIDS cases Texas residents, diagnosed from 1980-March 2006
Analysis Overview Included: –Cancer Diagnoses Cancers diagnosed with +/- 5 years of AIDS diagnosis Invasive tumors only –AIDS cases years of age Diagnosed with AIDS /2006 Alive for at least one month in
Match Overview Records were linked by computer using commercial probabalistic matching software Match based on name, date of birth, social security number, race, sex, date of death, and residential information Both cancer and HIV/AIDS files blinded from NCI researchers
Match Overview Software takes data common to both registries –Defines the distribution of the identifying variables –Attributes probability scores to the likelihood of matches on these variables being found by chance, and sums the scores –Results in score that indicates the strength of the match between records in the two registries –So fields like sex have little variability and add little to the match –SSN is virtually unique though and can indicate a high probability of a match, unless there is an error in the SSN –Allows some latitude for errors (like name misspelling) and ignores missing data
Completeness of Texas Data Elements PercentComplete Data ElementHIV/AIDS RegistryCancer Registry Name100% Date of Birth99% SSN76%98% Race98% Sex99%
Match Overview Overall, 2,547 AIDS cases (n=46,004) had at least one cancer diagnosed during the +/- 5 year time period ( ) Identified 1,006 (72%, n=1,388) Texas Kaposi Sarcoma cases diagnosed from were not in the Texas HIV/AIDS Registry Identified 321 “presumptive” and 110 “definitive” Kaposi Sarcoma cases not in the Texas Cancer Registry Identified 151 non-Hodgkin’s Lymphoma cases not in the Texas Cancer Registry, diagnosed from (n=17,098)
Demographic Characteristics Among People Included in AIDS-Cancer Match All Cancer Cases N=744,160 All AIDS Cases N=46,004 Developed Cancer N=2,574 Male51%82%89% Median Age at AIDS Dx Median Age at CancerDx 66-- Non-Hisp White73%39%48% Non-Hisp Black10%36%26% Hispanic14%24%26%
All AIDS cases N=46,004 Developed Cancer N=2,574 Men who have Sex with Men* 57%67% Injection Drug Use16%11% Heterosexual14%10% Blood Transfusion<1% Other/Unk12% HIV Transmission Risk Among People Included in AIDS-Cancer Match *includes MSM and IDU
Standardized Incidence Ratio Standardized incidence ratios (SIR) were calculated to com Compare cancer rates among PWAs to rates among the general population of Texas Standardized Incidence Ratio (SIR) = Observed # of cancer cases in AIDS population Expected # of cancer cases in the general Texas population
ObsExpSIR (95% CI) Male ( ) Female ( ) Overall ( ) AIDS-Defining Cancers Kaposi Sarcoma
ObsExpSIR (95% CI) Male ( ) Female ( ) Overall ( ) AIDS-Defining Cancers Non-Hodgkin’s Lymphoma
Kaposi Sarcoma Non-Hodgkin’s Lymphoma SIR (95% CI) Men who have Sex with Men* ( )43.2 ( ) Heterosexual 66.6 ( )29.8 ( ) Injection Drug Use 50.8 ( )24.1 ( ) Kaposi Sarcoma and Non-Hodgkin’s Lymphoma by Transmission Risk *includes MSM and IDU
ObsExpSIR (95% CI) Female ( ) AIDS-defining cancers Invasive Cervical Cancer
Non-AIDS Defining Cancers –Lymphocytic Leukemia –Melanoma of the Skin –Myeloid and Monocytic Leukemia –Myeloma –Pancreatic –Prostate –Rectal –Stomach –Testis –Thyroid –Tongue –Tonsil –Anal –Bladder –Breast –Brain –Colon –Esophageal –Gum and Other Mouth –Hodgkin’s Lymphoma –Kidney –Laryngeal –Lip –Liver –Lung Calculated SIRs for 25 non-AIDS defining cancers with >4 observed cases
Findings Of the 25 non-AIDS defining cancers –14 no significant difference between observed and expected cases –10 significantly higher –1 significantly lower
ObsExpSIR (95% CI) Anal ( ) Hodgkin’s Lymphoma ( ) Lip ( ) Myeloma ( ) Tonsil ( ) Non-AIDS-Defining Cancer Excess in Texas PWAs
ObsExpSIR (95% CI) Lung ( ) Liver ( ) Lymphocytic Leukemia ( ) Rectal ( ) Melanoma ( ) Non-AIDS-Defining Cancer Excess in Texas PWAs
Non-AIDS-Defining Cancer Lower in Texas PWAs ObsExpSIR (95% CI) Prostate ( )
ObsExpSIR (95% CI) Men who have Sex with Men* ( ) Heterosexual ( ) Injection Drug Use ( ) Other/Unk ( ) Anal Cancer by Transmission Risk *includes MSM and IDU
ObsExpSIR (95% CI) Men who have Sex with Men* ( ) Injection Drug Use ( ) Liver and IHB Cancer by Transmission Risk *includes MSM and IDU
ObsExpSIR (95% CI) Men who have Sex with Men* ( ) Heterosexual ( ) Injection Drug Use ( ) Other/Unk ( ) Lung Cancer by Transmission Risk *includes MSM and IDU
Summary Kaposi’s Sarcoma (n=871) and Non-Hodgkin’s Lymphoma (n=929) represented the majority of cancers found in this cohort All of the AIDS defining cancers were found to be in excess as expected 10 non-AIDS defining cancers were found in excess in the Texas AIDS population, 9 of these have risk factors involving infectious agents, suppressed immune systems, and/or tobacco use Cancers associated with HPV infection, immuno- suppression, and tobacco use appear to be a particular problem in the HIV/AIDS community
Next Steps Utilize linkage to improve both HIV/AIDS and cancer surveillance Examine cd4 counts/immunosuppression in relation to cancer diagnoses Examine geographical differences Re-match every 3-4 years Explore long-term use of HAART on changes in cancer incidence Utilize data to inform prevention and service activities
Lessons Learned Use the linkage as an opportunity to enhance registry data quality and future linkages Do homework/receive background documentation on what makes a case in one registry versus the other, national data standards, and procedures/practices that can affect linkage and/or interpretation of results If the registries are in agreement that linkage files can be shared between them, each receive the same files Ensure that data sharing between registries is spelled out in detail in IRB request and/or MOU
For Additional Information Contact: Melanie Williams, Ph.D. Senior Epidemiologist Texas Cancer Registry, Department of State Health Services (DSHS) Phone: ext