Philip S. Rosenberg, Ph.D.
|Organization:||National Cancer InstituteDivision of Cancer Epidemiology & Genetics, Biostatistics Branch|
|Address:||NCI Shady GroveRoom 7E130|
In 1988, Dr. Rosenberg received a Ph.D. in biostatistics from Yale University and joined the NCI as a Staff Fellow. His research focuses the incidence and prevalence of HIV infection in the United States and on the natural history of HIV infection. Dr. Rosenberg collaborates extensively with epidemiologists in the Viral Epidemiology Branch of DCEG and at the Centers for Disease Control and Prevention (CDC). He received the Howard W. Temin Award for AIDS research in 1993.
AIDS is the leading cause of death in the United States among young adults ages 25 to 44 years, yet there are limited direct data to track trends in the incidence of HIV infection. We developed statistical methods to monitor trends in the HIV epidemic which work backwards from the numbers of persons diagnosed with AIDS on the basis of its natural history. The models have been refined and validated over time, and are used to determine incidence trends according to gender, race/ethnicity, mode of HIV transmission and age. In parallel with these efforts, we conducted detailed studies of the natural history of HIV infection in prospectively followed cohorts of HIV-positive individuals. We are now applying these disease models for surveillance and natural history to studies of cancer.
We are developing models to estimate HIV incidence from AIDS case data using the mathematical principle of deconvolution. These "back-calculations" require knowledge of the distribution of the incubation period between HIV infection and AIDS diagnosis, called the incubation distribution, obtained from natural history cohort studies. Back-calculation was initially used to assess aggregate trends in HIV incidence with respect to the single time scale of calendar time. While useful, these models did not reflect that HIV infection occurs primarily among young people. We extended the methodology to accommodate the dual time scales of calendar time and age. Using this two-dimensional back-calculation approach, we are able to estimate time trends in HIV incidence according to age.
We applied two-dimensional back-calculation methods to surveillance data from the national database of AIDS cases compiled by the CDC. We showed that fewer persons overall were infected than previously thought, but that HIV prevalence was higher than previously recognized among persons in their late twenties and thirties and among racial and ethnic minorities. These estimates were shown to be consistent with available HIV seroprevalence data, and were used to derive the most recent official Public Health Service estimates of HIV prevalence in the United States. Considering trends by risk group, we found that HIV prevalence was increasing fastest among persons at risk through heterosexual contact.
The validity of back-calculation depends on the reliability of the incubation distribution that is used. We incorporated the most recent follow-up data from natural history cohort studies, and also used these data to identify co-factors of disease progression. In collaborative work, we showed that younger age at seroconversion is associated with slower progression to AIDS. In addition, we found that hemophiliacs progress to AIDS more slowly than homosexual men of the same age at seroconversion, and that the level of HIV viremia during early chronic infection is a strong and age-independent predictor of AIDS risk. We are currently conducting an international meta-analysis of genetic effects on HIV disease progression, including the those of polymorphisms of specific chemokine receptors (CCR2 and CCR5) and their ligands (SDF-1) on risk of AIDS and death.
AIDS case data for the period 1993 to 1995 have been difficult to interpret following the broadening of the definition of AIDS in 1993. However, using CDC's new method to adjust for the impact of the revised case definition, we are now able to analyze AIDS data from this time period. We found that among young persons, HIV incidence in homosexual men and injection drug users was slowing by 1993, though this favorable trend was offset by the increasing heterosexual transmission of HIV, especially in young minorities.