Recording, Interpreting, and Updating the Family History of Cancer: Title and subTitle BreakImplications for Cancer Prevention

Evidence-based guidelines recommend that persons at elevated risk of cancer begin cancer screening at a younger age than the general population and consider more sensitive screening tests. A family history of cancer contains powerful information for cancer risk stratification.¹ Because most common cancers are diagnosed in older age, the prevalence of a family history of cancer increases throughout adulthood. However, age-specific data based on detailed family history have only recently become available.² In this issue of JAMA, Ziogas and coinvestigators from the Cancer Genetics Network (CGN)³ report prospective and retrospective data from a population-based registry to illustrate the extent to which family history relevant for cancer prevention increases with age. What are the implications for preventive care?

Family history is an integral part of medical history and decision making yet has not, until recently, been standardized or widely recorded in structured fashion. Thus, little high-quality evidence is available to confirm the health benefits of generally screening family history in primary care.⁴ However, precise family history of an array of cancers is the basis for state-of-the-art cancer risk assessment. An Agency for Healthcare Research and Quality (AHRQ) expert panel concluded that systematic family history collection tools, some self-administered, are likely to improve on usual practice in primary care.⁵ The panel recommended recording a patient's family history of several cancers at the initial visit (including ethnicity, lineage, and age at diagnosis of each affected relative). Relevant family history also may include the family member's tumor pathology and genotype.⁶

Family history may be more consistently discussed with new patients than with established patients.⁷ To optimize the use of clinical time and resources, it is important to know when (at what ages and how often) to update the family history of common cancers. Knowing this, health information systems can be designed to accomplish this task. Furthermore, estimating the age-specific prevalence of increased familial risk is important for planning risk-appropriate cancer prevention services.

The study by Ziogas et al³ presents longitudinal data from the population-based CGN registry that show changes in self-reported family history of cancer through adult life, increasing most steeply before age 50 years for breast and colorectal cancer and before age 60 years for prostate cancer.³ The authors found a 5% chance that an individual's colorectal cancer screening recommendation would change between the ages of 30 and 50 years based on new family history and a 4% chance that women would be newly identified candidates for breast magnetic resonance imaging. The CGN study found age-specific prevalences of “clinically relevant” family histories of colon, breast, and prostate cancer that are similar to cross-sectional population-based data obtained from the California Health Interview Survey,² lending confidence that these estimates could reasonably be extrapolated to the general population.

The CGN investigators recommend updating family history of cancer every 5 to 10 years between ages 30 and 60 years. In current practice, updates likely occur either when patients present to clinicians with concerns about cancer (especially a relative or friend with a recent diagnosis), or for preventive care. Although primary care clinicians generally order cancer screening, it is easy to think of circumstances when updating the family history of cancer would be relevant for surgeons, gastroenterologists, urologists, radiologists, oncologists, gynecologists, and other clinicians. Having all clinicians and patients participate in reviewing and recording family history will become more feasible as electronic health records (EHRs) compile information from all who care for a patient, and as EHRs and personal health records eventually include structured family medical history data.⁸

Although self-reported family history about common cancers appears to be fairly sensitive and specific,⁵ it is often laborious to confirm diagnoses when major clinical decisions hinge on accurate familial risk assessment. In the long run, EHR functions might include the capacity for individual family members to consent to have their diagnoses (and other information, such as test results) automatically populate family history sections in the EHRs of specified relatives, thus minimizing the uncertainties of self-report. However, realizing this goal faces high barriers because most current EHRs in the United States are not interoperable in practice, even if they are designed to use common data formats.

According to the AHRQ panel, “In practical terms, the systematic collection of family history [of cancer] . . . is linked with the interpretation of that information which in turn is linked to whether primary care providers take appropriate clinical action on the basis of the information collected.”⁵ Thus, once up-to-date family history is recorded, ideally, decision support could help clinicians identify which aspects of family history are “clinically relevant”—ie, reach a risk threshold that triggers earlier or more sensitive cancer screening or genetic evaluation for hereditary cancer susceptibility.^{9 - 12} However, much needed evidence about the population health benefits of early and intensified cancer screening according to familial risk has yet to be developed.

Like all studies of screening, these must take into account risks, benefits, costs, and lead time issues. For example, it has been estimated that between less than 1% and 25% of breast cancers identified by screening may represent overdiagnosis.^{13 - 14} It is plausible but still unknown whether family history increases the likelihood that breast cancers, prostate cancers, or colon adenomas found by screening are clinically significant. An increase in the incidence of false-positive results and test-associated complications is a cost and potential harm of increased screening based on familial risk.¹³ Although some prospective data on the benefits of cancer screening based on familial risk are available,^{15 - 16} many estimates rely on extrapolation from small studies of patients with high-penetrance hereditary cancer susceptibility¹⁷ or from screening older patients at equivalent levels of risk.

The age-specific prevalence of family history–based risk of breast, colon, and prostate cancer, as measured in the study by Ziogas et al,³ provides valuable information for estimating the costs of different screening strategies in a population. American Cancer Society (ACS) guidelines recommend colonoscopy every 5 years starting at age 40 years for individuals at moderately increased risk of colorectal cancer.¹¹ Using CGN data, in addition to colorectal cancer screening recommended for the general population, approximately 4% of 40- to 50-year-olds with significant family histories would undergo colonoscopies at 5-year intervals for earlier colorectal cancer screening; 7% of 55-year-olds and 11% of 65-year-olds would be eligible for additional colonoscopy.³ A rough calculation from US Census data¹⁸ suggests that implementing ACS colorectal cancer screening guidelines based on family history assessment would thus entail at least 1.7 million additional colonoscopies per year, approximately 18% more than would be performed if every adult in the United States were screened with colonoscopy every 10 years from ages 50 to 70 years. A decision analysis showing incremental cost-effectiveness ranging from $18 000 to $51 000 per year of life gained by such a protocol of family history–based screening was sensitive to the prevalence of increased familial risk but did not model an increase in prevalence of family history with age.¹⁹

The age and breast cancer risk thresholds for recommending mammography and breast MRI screening vary among expert groups who interpret the evolving evidence.^{9 - 11 ,20 - 21} The ACS guidelines recommend annual breast MRI in addition to mammography for women with high likelihood of a cancer-associated mutation in the family, chest radiation in youth, or at least a 20% lifetime risk of breast cancer.¹¹ Using the ACS guidelines and estimates based on the findings of Ziogas et al,³ approximately 7.2% of women aged 30 to 40 years and 8.9% aged 40 to 50 years would be candidates for annual breast MRI and screening mammography, which costs approximately 10 times more than mammogram alone. Many of the same women would also fit criteria for genetic assessment.¹²

Does increasingly accurate cancer risk stratification offer opportunities, while intensifying cancer screening for groups at increased risk, to decrease the intensity of screening for groups at lower risk? The US Preventive Services Task Force, amid substantial public backlash, made an evidence-based recommendation that average-risk women might choose to postpone mammography screening until age 50 years.²⁰ It is possible that, if family history of cancer and breast cancer risk status were reliably updated during the decade from ages 40 to 50 years, many lower-risk women in that age group, as in England, might forgo mammography screening.^{10 ,20} The benefits, harms, and empirical results of such an approach are ripe for investigation.

The development of evidence about how clinicians and patients use familial risk information to intensify cancer screening and preventive actions is still in its early stages. The GRAIDS trial showed more appropriate referrals to high-risk screening and cancer genetics services with general practitioners using a family history tool,²² while another cluster randomized trial of computerized decision support for general practitioners showed no effect.²³ In the Family Healthware Impact Trial, in which patients and their primary care clinicians were given familial risk assessments for breast, ovarian, and colon cancers; coronary heart disease; stroke; and diabetes; there was no effect on short-term rates of cancer screening,²⁴ but the trial demonstrated desirable effects on participants' physical activity and fruit and vegetable intake 6 months later.²⁵ Paradoxically, randomized controlled trials of systematic family history assessment are less feasible because family history is already part of the fabric of preventive care, so patients with a family history of cancer are already more likely to get the cancer screening that is recommended for the general population.²⁴ It is for patients in whom extra preventive measures are recommended that assessing and updating family history during adulthood may have the most influence and require the most careful study.