Title: Cancer Rate Decline at Old Age



Title: Cancer Rate Decline at Old Age

Authors: Charles Harding*, Francesco Pompei, and Richard Wilson

*Corresponding author. Email: harding.charles@

Institution: Department of Physics, Harvard University, Cambridge MA 02138

Running Title: Cancer Rate Decline at Old Age

Keywords: elderly, senescence, carcinogenesis, mortality, prevalence

Submission to: Cancer

Outside support: There was no grant support for this project.

Conflicts of interest: None of the autors have a We declare no conflicts of interest.

ABSTRACT

Earlier work has shown, using SEER data, that cancer incidence decreasesfalls after age 80 for most cancers. This studypaper extends this work, albeit in a somewhat smaller data set, and shows that

Separately, it has been shown that age-specific cancer incidence, mortality, and aspects of prevalence all plateau or decrease at very old ages. Because some of these measures are more accurate, and others are easier to obtain, each has been used by others to estimate cancer risk among the elderly, in addition to estimating cancer burden. InWe report that in a single largethe selected cohort, population (9.5% of the United States population, 1998-2002), cancer incidence, mortality, and prevalence peak during old age. For many of the 22 cancers studied, a dramatic and significant decline follows the peak, so that the extremely elderly seem asymptomatic or unsusceptible. This decline is poorly understood. As in our earlier paper, We re-examine and extend the possible biological and non-biological explanations, discussed already in reference (4), including heterogeneous susceptibility to cancer, general senescence, immune system senescence, and error in US Census population figures. (particularly for centenarians). We address and remove approximations that qualify previous results. Additionally, a largely empirical extension of the multistage model of carcinogenesis is fit to old age incidence and mortality figures. In addition to the figures in this paper all the numerical results in this study are available on

Introduction

Several papers suggest that cancer incidence rates level off or decline after age 80 (1,2,3,4). Autopsy studies generally report decreased presence of metastasized tumors at very old age, and comparable presence of all tumors across old age groups (5,6,7,8). Smith (9) reports a drop in the mortality rate of some cancers at old age, using data from Vital Statistics of the United States, 1990. This study, an extension of the work in reference (4) using similar methodsHowever, no previous study has addressesed age-specific incidence, mortality and prevalence in the same population.

The data base is the We examine records of the Surveillance, Epidemiology, and End Results (SEER) cancer registries, focusing on the areas known as SEER 9 and the period 1998-2002. Using this resource, we can compare incidence, mortality, and prevalence in the same population, and validate the use of each of these measures separately, in many different locations, to discover the plateau or decline in cancer risk at old age. In addition, the size and quality of the current SEER database allow us to inspect old age cancer rates at a greater level of detail and accuracy.

Old age cancer prevalence studies have so far been based on collected autopsy reports. In conrast we We believe this paper is the first to present age-specific cancer prevalence proportions based on diagnosis data (rather than autopsy) and among the elderly. We found As will be seen, there are considerable differences between diagnosis-based and autopsy-based prevalence.

Widespread SEER data begins in the mid-1970s, which allows us to compute limited birth-cohort incidence rates. Previous old age birth-cohort results are available for intestinal cancer mortality (10), and report peak mortality around 90 years old.

All the data obtained in this paper, including some data not plotted here are available on the web page:

Methods

Our methods closely follow those used by Harding et al. (4) to determine incidence, with some important modifications. We examined 21 cancers in men and 22 cancers in women, identified according to the SEER classification of cancer type. Data were taken from SEER 9 (11), a smaller registry than used in Harding et al. SEER 9 includes all cancers diagnosed from 1975 on in Atlanta, Connecticut, Detroit, Hawaii, Iowa, New Mexico, San Francisco-Oakland, Seattle-Puget Sound, and Utah. SEER*Stat software (12) was used to compute absolute, incidence-based morality (using survival time information), and absolute incidence in 5-year age categories from 0-4 years to 110-114 years . Cases over age 115 were disregarded. For the results plotted in this paper, we counted only those diagnoses specified as first malignant primary tumors. However, results not plotted here include we repeated our analysis including every malignant tumor in the database, (whether it is was the first to be diagnosed in the patient, or a subsequent tumor). The old age peak and decline in cancer incidence rates, which is discussed in the results section, remained.

To compute old age person-time at risk for each year 1998-2002, SEER-provided 85+ population figures were split proportionately according to year 2000 US census data (13), which gives populations 85-89 through 110-114 in county and regional areas that match those covered by SEER 9. With the exception of some outlier cases for populations in their 50s, each single year age category 0-85+ showed good agreement between SEER-provided 1998-2002 population figures and Census 2000 regional population figures ( ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download