developing lung cancer, like the Gail model does for breast cancer, or the Framingham model to predict heart disease," says Carol Etzel, Ph.D., assistant professor in the Department of Epidemiology, and the study's senior author.

The model's prediction level of lung cancer is about 60%. While modest, it's on par with that of the Gail model, say the researchers.

One might question the need for a lung cancer risk model since smoking is the primary cause of 85 percent of all lung cancers, says Margaret Spitz, M.D., professor and chair of the Department of Epidemiology and the study's lead author.

However, surprisingly, in life-time heavy smokers, less than 20% will actually develop lung cancer. "The challenge becomes how to identify that fraction of long-term cigarette smokers at the highest risk for the disease," she says.

"If we know who is at greatest risk for lung cancer, we can offer the most intense smoking cessation, or perhaps even offer chemo-preventive interventions. More importantly, we could intensively screen this population with modalities that might not be appropriate for the average at-risk population," says Spitz.

Lung cancer is the leading cancer killer in both men and women. More than 213,380 new cases are predicted, and 160,390 patients are expected to die from the disease in 2007, according to the American Cancer Society. The risk of developing lung cancer is 23 times higher in male smokers and 13 times higher in female smokers, compared to lifelong non-smokers.

The risk assessment tool was developed and tested based on research comparing the medical history of 1,851 lung cancer patients (cases) treated at M. D. Anderson with the same data from 2,001 matched healthy individuals (controls). With a population so large, the researchers were able to divide the cases and controls into two groups - the first for building the model, the second set for testing and validating the model. This approach is the gold-standard for the development of risk assessment models, says Spitz. Current, former and never smokers were all included in the development of the model - the first time a lung cancer assessment tool has included individuals who have never smoked.

Based on the model, clinicians can compute a patient's ordinal risk score and absolute chance a patient has of developing lung cancer within a year. The patient then can be classified into high-, moderate-, or low-risk groups. Examples of key risk factors found in the targeted groups include:

• in never smokers: exposure to secondhand smoke and family history of cancer;
• in both current and former smokers: emphysema; exposure to dust and no history of hay fever; 
• in former smokers: age they stopped smoking and family history of cancer;
• and in current smokers: asbestos exposure, intensity of smoking and family history of a smoking-related cancer.

Spitz and Etzel say that the most striking finding was the strong impact of a prior history of emphysema as a risk factor in both current and former smokers. In contrast, hay fever worked as a protective agent against lung cancer in both groups.

The study is not without limitations. One major drawback is that the model focuses only on Caucasians, due to the fact that there were not enough minority patients in the cohort to build and validate the model. "We are currently working with other institutions to combine our numbers and build a model specifically for Mexican Americans and African Americans. In preliminary testing, already we are finding that while some of the risk factors are common to both groups, there are different levels of risk, so the model for Caucasians would likely not be as predictive for other populations," says Etzel.

Also, cases and controls were paired based on smoking status - perhaps masking the importance of smoking as a risk factor, though adjustment factors were included for this limitation.

Currently, the researchers are developing a Web-based version of this lung cancer assessment model, in hopes of soon making the tool accessible to clinicians.

Other M. D. Anderson study authors include Waun Ki Hong, M.D., professor and head of the Division of Cancer Medicine; Christopher Amos, Ph.D., and Xifeng Wu, M.D., Ph.D., both professors in the Department of Epidemiology; and Sanjay Shete, Ph.D., associate professor in the Department of Epidemiology. Matthew Schabath, Ph.D., assistant professor at The University of Texas School of Public Health, also contributed to the study.