Variation in risk of second primary cancer.

of concern among cancer survivors, their families and the health care professionals providing follow-up care. Its risk is determined in a complex interaction between treatment of the first cancer, the genetic profile of the survivor, and environmental and behavioural factors. For example, exposure to specific treatments for cancer in childhood has been linked to an in creased risk of certain second primary cancers in adulthood. This increased risk has re sulted in the development of specific evidence-based follow-up guidelines by the Children’s Oncology Group. Certain germline mutations of the p53 tumour suppressor gene, responsible for most instances of Li–Fraumeni syndrome, elevate a person’s risk of certain cancers, any pair of which could appear as the first and second primary cancers. Possibly most important from the perspective of prevention, risks of certain cancers can be jointly elevated by specific environmental and behavioural factors. For instance, a population-based nested case–control study involving breast cancer survivors showed that obesity, alcohol intake and smoking increased the risk of a second breast cancer appreciably. Not all second cancers are attributable to known risk factors, however. It is therefore important to study the associations be tween first and second primary cancers in an effort to better understand their cause and develop informed recommendations for follow-up care. In a related article in CMAJ, Nielsen and colleagues present their findings from a nationwide nested case–control study in Denmark that assessed the risk of second primary cancer. The study has a number of methodologic strengths: the authors identified a large number of patients with cancer from a high-quality national registry over a recent 28-year period (1980–2007); each patient was matched (by sex, birth year, calendar period at risk and age at diagnosis of the cancer) with up to five randomly selected controls who did not have the examined cancer at the time of diagnosis; they performed sensitivity analyses that included cancers diagnosed only 1, 2, 5 and 10 years after the first cancer diagnosis, to minimize the potential of erroneously including recurrent cancer as a new primary cancer; and they used state-of-the-art statistical methods that accounted for multiple comparisons of many types of first and second primary cancers. Well-designed, well-conducted studies such as the one by Nielsen and colleagues can provide important data to inform causal mechanisms and follow-up care of cancer survivors. Specifically, the detailed cancer-pair –specific analyses of Nielsen and colleagues are valuable for exploring causes such as shared genetic predisposition and for evaluating or modifying guidelines for follow-up care (although the analyses did not consider treatment factors of the first primary cancer). Two key issues exist in translating the findings on the risk of second primary cancer, such as those of Nielsen and coauthors, into clinical implications on the frequency and type of follow-up care. The first issue is the heterogeneity in risk of second primary cancer across pairs of first and second cancers. Nielsen and colleagues found that the risk of a second primary cancer depended greatly on the types of the first and second cancers; heterogeneity in risk was substantial across cancer types, regardless of whether the second cancer was the same type as the first. Their study provides evidence of the high degree of heterogeneity (even though they attempted to synthesize the data using fixedeffect meta-analysis techniques) and thus the need for separate evaluation and discussion of Variation in risk of second primary cancer