Johns Hopkins University researchers and colleagues published an article in the September 27, 2006 issue of JAMA introducing a prediction tool, called MMRpro, which identifies families with genetic defects that cause an increased risk of colorectal cancer. This assessment helps families make decisions about cancer prevention screenings.
Sining Chen, PhD, lead author of the study and an assistant professor in the Bloomberg School of Public Health's Departments of Environmental Health Sciences and Biostatistics, explains the key concepts of the new model.
Q: What is Lynch syndrome?
Sining Chen: Lynch syndrome, also known as hereditary nonpolyposis colorectal cancer, is characterized by the inheritance of genetic defects in the MLH1, MSH2 and MSH6 genes. Lynch syndrome predisposes families to developing colorectal cancer and endometrial cancer in women. Colorectal cancer due to Lynch syndrome also develops at younger ages than would be expected in people without the genetic defects. Lynch syndrome accounts for about 2 percent of all colorectal cancer diagnoses and a large fraction of early onset colorectal cancers. It is the most common familial colorectal cancer syndrome.
Q: What genetic defects are involved?
SC: The genes we are concerned with in Lynch syndrome are called "mismatch repair genes (MMR)." These genes repair mismatches that occur during the duplication of the genetic code when new cells are made. When there is a defect in these genes, mismatches may not be repaired properly and cancer cells may arise. Such defects can be passed from generation to generation. Families with genetic defects can develop Lynch syndrome, which is characterized by increased incidence of colorectal and endometrial cancers.
Q: How serious is colorectal cancer as compared to other cancers? Is it one of the leading causes of death?
SC: Colorectal cancer is the second largest cause of cancer deaths in the United States. A projected 150,000 people will be diagnosed with colorectal cancer in 2006 and 56,000 will die from it. However, the survival rate for people with colorectal cancers found early is more than 90 percent. People need to be aware of the high prevalence and the importance of early detection of colorectal cancer.
Q: How large of an issue is endometrial cancer?
SC: Endometrial cancer is the most common gynecologic cancer, with about 36,000 new diagnoses and 6,500 deaths every year. The risk of developing endometrial cancer by age 70 is approximately 2 percent in the female population, but 50 percent if the woman carries a genetic defect mentioned here.
Q: What should people do when they have concerns about inheriting Lynch syndrome?
SC: They should see a genetic counselor or familial cancer specialist, who is likely to use quantitative models, such as MMRpro, to help them make decisions about screening or genetic testing.
Q: What does MMRpro stand for? What does it do?
SC: 'MMR' stands for mismatch repair, as described above, and 'pro' stands for probability. MMRpro provides the risk of carrying one of the genetic defects described above, and if an individual is still healthy, the risk of developing colorectal or endometrial cancer in the future. This helps identify individuals who are more likely to have a colorectal cancer-related genetic defect. The model is based on the person's family history of colorectal and endometrial cancers.
Q: How many cases of colon cancer in your family history should warrant concern?
SC: That depends on the specifics of the family history. But one close relative with early onset colorectal cancer or with endometrial cancer is sufficient to prompt a physician visit.
Q: If a family doctor isn't familiar with this model, what should a lay person do?
SC: Seek help from a trained genetic counselor, who should be able to understand the underlying mechanism of the model and to interpret the model predictions.
Q: What information do you need to provide? What type of result do you get?
SC: Individuals provide the age of colorectal and endometrial diagnoses of affected relatives, and current age (or age of death) of unaffected relatives. If genetic testing was done, those results are also requested. MMRpro will provide a projected risk. Your doctor will give you recommendations on cancer screening or genetic testing based on the results.
Q: If information in one's family history of colorectal and endometrical cancer is missing, is the MMRpro model still able to accurately predict the probability of carrying an inherited defect?
SC: One of the advantages of MMRpro is its ability to handle gaps in family history information, such as a relative who has been out of contact for many years. However, the more information is provided, the more accurate the prediction.
Q: Is MMRpro only useful for families with the Lynch syndrome?
SC: The prediction tool is useful for all families and individuals who are interested in finding out whether their cancer is genetic and what their risks might be. Our data suggests that even a single close relative with early onset colorectal or endometrial cancer could increase the risk sufficiently to justify a physician visit and increased screening.
Q: How should MMRpro be implemented in the clinic?
SC: The approach we favor is to have a counselor or clinician take a family history, determine the risk, and provide the risk evaluation to the patient along with a discussion about inherited susceptibility and screening options. CaGene is user-friendly software that can be used by clinicians to implement this approach.
Q: What should patients know about the prediction model before getting a risk prediction?
SC: Patients should know that the model has been validated in independent samples and performs as it is supposed to. They should also know that, as any other risk prediction model, it cannot forsee the future, but only help identify groups of people among whom the incidence of cancer events is reliably higher.
Q: If a patient is found to be at risk, what are the next steps?
SC: The next steps depend on a number of factors, including level of risk. Clinicians can provide advice about this based on the overall clinical profile of the patient and his or her preferences. We are actively working on developing detailed recommendations on how to adjust the frequency of screening exams to the level of risk, to help physicians make this assessment.
Q: What are current diagnostic tests for Lynch Syndrome? What are the advantages to using MMRpro?
SC: Currently, clinical diagnosis tools that are based on family history include Bethesda guidelines and Amsterdam criteria. There are several advantages to using MMRpro over these models. First, MMRpro is both more sensitive (that is, it identifies more mutation carriers) and more specific (it identifies fewer non-carriers) than these other approaches. MMRpro uses detailed cancer history information to make individualized evaluations, instead of classifying families into broad categories. Second, MMRpro incorporates explicitly the genetic mechanism by which increased susceptibility to cancer is passed on from one generation to the next, while none of the other models do. This generally leads to greater accuracy. Third, MMRpro is built in a modular fashion that allows incorporating information from diverse sources, allowing us to comprehensively include everything we know, and to use each knowledge source for its best purpose. When new studies of risk or prevalence are published, we can easily update the model so patients always have a state-of-the-art tool. Fourth, MMRpro is able to exploit far more detailed family history information than the other models. However, there is no minimum amount of detail needed. Finally, the model is implemented on open source software, and is written in a way that allows others to easily modify the setting for their specific populations or needs.
Q: What are the drawbacks to using MMRpro?
SC: The main drawback is that we have not yet incorporated other types of cancer that are associated, though not as strongly as the ones we consider, with Lynch Syndrome. We are actively working on this.
Q: Why is this study important?
SC: About 600,000 people in the United States carry Lynch syndrome genetic defects. Each of them has approximately a 50 percent chance of developing colorectal cancer by age 70, along with increased risks for other cancers. Colorectal cancer is one of the most preventable forms of cancer. Our model can identify people who carry the genetic defect early and help them make decisions about future screening.
Q: Is there somewhere where to can get more information?
SC: The model software and related information can be downloaded at the BayesMendel Software Page and at the CaGene Website. To learn more about Lynch syndrome, a good starting point is here.
Q: In the same issue of JAMA, there is a similar model called PREMM1,2. What are the main differences between the two models? How would you compare their accuracy?
SC: The two models differ in the ways they were developed and validated. They have not yet been directly compared on the same families, so the two groups plan to collaborate on a joint validation study to compare the accuracy of the two models on the same group of families.
BayesMendel Lab
