Gerald R. Hobbs

Epidemiological studies provide strong evidence that obesity and the associated adipose tissue inflammation are risk factors for breast cancer; however, the molecular mechanisms are poorly understood. We evaluated the effect of a high-fat/high-calorie diet on mammary carcinogenesis in the immunocompetent MMTV-PyMT murine model. Four-week old female mice (20/group) were randomized to receive either a high-fat (HF; 60% kcal as fat) or a low-fat (LF; 16% kcal) diet for eight weeks. Body weights were determined, and tumor volumes measured by ultrasound, each week. At necropsy, the tumors and abdominal visceral fat were weighed and plasma collected. The primary mammary tumors, adjacent mammary fat, and lungs were preserved for histological and immunohistochemical examination and quantification of infiltrating macrophages, crown-like structure (CLS) formation, and microvessel density. The body weight gains, visceral fat weights, the primary mammary tumor growth rates and terminal weights, were all significantly greater in the HF-fed mice. Adipose tissue inflammation in the HF group was indicated by hepatic steatosis, pronounced macrophage infiltration and CLS formation, and elevations in plasma monocyte chemoattractant protein-1 (MCP-1), leptin and proinflammatory cytokine concentrations. HF intake was also associated with higher tumor-associated microvascular density and the proangiogenic factor MCP-1. This study provides preclinical evidence in a spontaneous model of breast cancer that mammary adipose tissue inflammation induced by diet, enhances the recruitment of macrophages and increases tumor vascular density suggesting a role for obesity in creating a microenvironment favorable for angiogenesis in the progression of breast cancer.

Between 1940 and 2005, in the United States, the rate of unnatural death declined about 75 percent in infant and young child boys and girls; a remarkable indicator of successful child protection. During this same period, the rate of reported homicide in infant boys increased 64.0 percent, in infant girls increased 43.5 percent, in young child boys increased 333.3 percent, and in young child girls increased 300.0 percent, a dismal and disturbing indicator of failed child protection. Can these simultaneously encouraging and discouraging observations be reconciled? The four categories of unnatural death, homicide, suicide, motor vehicle accident (MVA), and non-MVA, are mutually exclusive classifications. Correlations between the four categories of unnatural death among U.S. men and woman in all age groups for the years 1940 through 2005 were calculated. A negative correlation between homicide and non-MVA death rates was shown for all age groups, encompassing the entire human lifespan, in both genders. This consistently observed negative correlation was only observed between homicide and non-MVA death rates, and was not demonstrated between other causes of unnatural deaths. Moreover, this negative correlation was strongest (less than −0.7) in infants and young children. These observations are consistent with the suggestion that as the rate of unnatural death in infants and young children dramatically declined, society gave greater scrutiny to those fewer occurring unnatural deaths and demonstrated an increasing propensity to assign blame for those fewer deaths.

Two counter trends in injury mortality have been separately reported in the US in recent times - a declining suicide rate and a rapidly rising unintentional poisoning mortality rate. Poisoning suicides are especially difficult to detect, and injury of undetermined intent is the underlying cause-of-death category most likely to reflect this difficulty. We compare suicide and poisoning mortality trends over two decades in a preliminary assessment of their independence and implications for suicide misclassification.

Objective: To acquaint members of the Academy with a relatively recent development in the area of data exploration and statistical analysis. Methods: A review of the concepts and methods inherent in data-mining with a special emphasis on the those methods applicable to predictive modeling. Results: Data-mining is demonstrated to be a useful tool for researchers in those circumstances where large amounts of information are available. Conclusions: With the advent and proliferation of online data collection, truly massive databases are now available to health care researchers. In that situation, data-mining methods yield some unique opportunities to researchers who wish to develop prediction models and to establish associations.