Self-reported cancer family history is a useful tool for identification of individuals at risk of hereditary cancer predisposition syndrome at primary care centers in middle-income settings: a longitudinal study

Abstract Analysis of cancer family history (CFH) offers a low-cost genetic tool to identify familial cancer predisposition. In middle-income settings, the scarcity of individual records and database-linked records hinders the assessment of self-reported CFH consistency as an indicator of familial cancer predisposition. We used self-reported CFH to identify those families at risk for hereditary cancer syndromes in community-based primary care centers of a low-income Brazilian area. We also evaluated the consistency of the information collected by reassessing CFH five years later. We interviewed 390 families and constructed their pedigrees for genetic cancer risk assessment. We found 125 families affected by cancer, 35.2% with moderate to high risk of familial susceptibility to cancer, a number that represents a relatively high prevalence of potential hereditary cancer syndromes in the overall study sample. Upon reassessment of CFH in 14/20 families that were previously identified as having at least one first-degree and one second-degree relative affected by cancer, and presented moderate to high risk for developing cancer, 90% of initial pedigrees were confirmed. These results demonstrate the reliability of self-reports as a means of early identification of healthy individuals at risk, encouraging the wider use of this method in low- and middle-income primary care settings.


Introduction
Family history is essential for identifying individuals at increased risk for primary and secondary cancers who could benefit from referral to genetics services (Weitzel et al., 2011;Lu et al., 2014). Identification of those individuals is crucial for early diagnosis, family management, and preventive care , reducing cancer morbi-mortality and health system costs (Rubinstein et al., 2011;Acheson, 2014).
The analysis of cancer family history (CFH) offers a low-cost, non-invasive genetic tool to track and diagnose familial cancer predisposition (Plat et al., 2009;Valdez et al., 2010;Janssens et al., 2012). Indeed, CFH has been shown to be a strong predictor of disease risk and yet is significantly underused in primary care settings (Doerr and Teng, 2012;Teng and Acheson, 2014).
Reliance on self-reported CFH for use in clinical practice, decision-making regarding surveillance recommendations and the design of preventive interventions depends, however, on knowledge about the reliability and accuracy of this tool. Commonly, medical records, death certificates, and information obtained from cancer registries have been used to determine the accuracy of the self-reported cancer family history (Kelly et al., 2007;Qureshi et al., 2009). However, individual records and database-linked records are rarely available in low-and middle-income settings, making it difficult to establish whether self-reported CFH is a reliable indicator of familial cancer predisposition in these populations (Gomy and Diz, 2013). In this study we used self-reported family history data to identify those families at risk for hereditary cancer syndromes at five community-based primary care centers in a middle-income area in Brazil. We also sought to evaluate the consistency of the information collected in spontaneous interviews by replicating the data collection five years later. Our results should encourage the wider use of self-reported cancer family history in low-and middle-income primary care settings.

Subjects and Methods
We conducted a longitudinal study divided into two phases. In the first phase (started in 2008), a total of 410 families were selected through simple random sampling among 3,780 families attended at the five communitybased primary care centers located in Ribeirão Preto, São Paulo, Brazil. The sample size was calculated considering an expected prevalence of 50% of CFH, and using a 95% confidence interval with a set at 5% considering a likely sampling loss of 15% (Pagano and Gauvreau, 2004).
A research team composed of five research assistants was trained by the first author to visit the families and collect their self-reported family history, with the purpose of selecting families at potential risk for hereditary cancer syndromes. We interviewed one informant per household, who was available at home, volunteered to participate in the study after listening to the research purposes, and then signed the consent form. The informant answered a questionnaire that included variables previously described in the literature (Feerro et al., 2008;Lu et al., 2014), such as personal and/or family history of cancer, degrees of relationship among family members affected by malignancies, gender, age, vital status, age at cancer onset, and type of primary cancer. Data were entered into Progeny pedigreedrawing software (Progeny Software, LLC, Indianapolis, IN, USA), and Statistical Package for the Social Sciences version 17.0 (SPSS Inc., Chicago, Illinois, USA) was used for descriptive statistical analysis.
All pedigrees were analyzed independently by two geneticists, a physician (VEFF), and a nurse (MFS) with expertise in oncogenetics, to perform genetic cancer risk assessment. In case of discordance regarding CFH evaluation, a third geneticist was consulted until consensus was reached, to ascertain data quality. We applied internationally established criteria to classify CFH in sporadic, familial, and hereditary cases, as well as to classify genetic cancer risk as low, moderate, and high (Table 1) (Schneider, 2002;Lindor et al., 2008;Valdez et al., 2010;NCCN, 2014;Vieira et al., 2015).
The second phase of the study was conducted five years later. The families that were previously identified as having at least one first-degree and one second-degree relative affected by cancer were visited again, and were reinterviewed by one of the former research assistants (LCLJr.) to confirm the previously reported CFH. The informant interviewed in phase 2 was not necessarily the same person who responded to the questionnaire in phase 1 of the study. All participants were previously contacted by telephone to schedule the interview. At the beginning of the interview, the researcher clarified that the family was contacted again because of its risk to present a familial susceptibility to cancer. Noteworthy, neither the researcher nor the participant had access to the pedigree that was depicted at the first visit. The same questions that were asked in the first interview were asked again. This study was approved

Results
Of the 3,780 families registered in five primary health care units, 410 were randomly chosen for this study. Our final sample was composed of 390 families with which personal contact was possible. The characteristics of their pedigrees are shown in Table 2. Most participants were female (79.5%), and sample mean age was 54.0 ± 16.5 years. Informative familial history -namely, complete information on the occurrence of cancer (and when present, tumor site and age at onset of disease) for at least three generations -was obtained from 219 (56.1%) families. Of the 390 families interviewed, 125 (32.0%) informed to be affected by cancer (Figure 1): in 20 (29.4%) families at least one first-and one second-degree relative of the respondent were affected; in 14 (24.6%) at least one first-degree relative of the respondent was affected; in 48 (70.6%) only second-degree and more distant relatives were affected; and 43 (75.4%) had only more distant family members with cancer. Also, 10 respondents had had cancer themselves.
Based on the analysis of relatedness, age of disease onset, primary tumor sites and clusters, as determined by internationally established criteria for familial cancer susceptibility syndromes, we identified 81 (64.8%) families with sporadic cancer cases, and 44 (35.2% within the families affected by cancer, or 11.3% of the overall study sample) with potential familial susceptibility to cancer. Regarding the hereditary cancer syndromes, we found four families that met the criteria for breast and ovary hereditary cancer, three for Li-Fraumeni syndrome, two families with syndromes that comprised hereditary colorectal cancer, another two families with breast and colon cancer, three families with potential alterations on repair genes, one family probably carrier of hereditary gastric cancer syndrome, and others with familial cancer clusters without a specific inheritance pattern.

2-30
Number of studied families 390 SD, standard deviation; y, years. the overall studied sample healthy population) potentially at risk for familial susceptibility to cancer.
We next investigated whether self-reported family history is sufficiently consistent for use in genetic risk estimation for inherited cancer. For this, five years later we revaluated 14/20 families that we previously identified as having at least one first-and one second-degree relative affected by cancer and presenting moderate (n = 4) to high (n = 16) risk for developing cancer based on genetic predisposition ( Figure 1). Analysis of family history collected in the second interview confirmed 90% of initial pedigrees, regardless of whether the interviewed family member was the same subject (n = 11) from the first interview or not (n = 3). In addition, we observed new cancer red flags in these families, such as benign lesions at early ages and new cancer cases that reinforce the family's high risk for hereditary cancer syndromes (Lindor et al., 2008).

Discussion
Recognizing patterns of familial cancer that indicate increased risk and possible hereditary syndromes can help to identify individuals who may benefit from preventive interventions (Ashton-Prolla, 2013). In this study we used self-reported family history data to examine the risk of hereditary cancer syndromes among middle-and low-income families registered at primary care centers in Brazil. We also revaluated the cancer family history of families at risk five years later to determine the consistency of the information collected through the use of spontaneous self-reports.
Our results showed that among those families affected by cancer, 35.2% had moderate to high risk of familial susceptibility to cancer. This represents a relatively high prevalence of potential hereditary cancer syndromes in the study sample. Likewise, a significant prevalence (6.2%) of hereditary breast cancer syndromes have been reported in a previous study in which CFH was assessed in women from primary health care units of an underserved region in southern Brazil . These results point to the likely -yet often neglected -benefits of screening policies aimed at identifying individuals at risk in these settings (Doerr and Teng, 2012;Teng and Acheson, 2014). A pioneering initiative to identify individuals with an increased risk for hereditary breast cancer syndromes was conducted through the successfully development and validation of a simple questionnaire, which is sensitive and specific in primary care setting, as a screening tool to refer at-risk individuals for genetic counseling .
Given that cancer family history is a dynamic measure that changes significantly over time, its periodic reassessment has been recommended . Upon revaluation, we found that cancer family history was confirmed, and expanded, for 90% of the families interviewed. In some cases, we collected data regarding benign lesions that are often found on the clinical spectrum of the hereditary cancer syndromes (Lindor et al., 2008;Lu Flória-Santos et al. 181 NCCN, 2014). These data point to the reliability of self-reports as a means of early identification of healthy individuals at risk of cancer. Indeed, literature data have shown that self-reports of personal cancer history are generally reliable, especially for breast, prostate, and colon cancer Scheuner et al., 2010). Most sensitivity values for self-reports of a positive family history of cancer in a first-degree relative range from 70 to 90% (Ziogas and Anton-Culver, 2003;Murff et al., 2007). Our study demonstrated that, even after a long period of time, self-reported CFH is an effective mean to detect not only breast, prostate, and colon cancer, but other important tumors that are considered as part of the spectrum of the hereditary cancer syndromes, such as Li-Fraumeni syndrome, which has shown considerable prevalence in the Brazilian population (Achatz et al., 2007;Giacomazzi et al., 2013). The inclusion of new health technologies, including genetic risk assessment and genetic testing, remains a challenge for middle-income countries like Brazil and other Latin American countries . Therefore, the possibility to use self-reports to unveil familial susceptibility to cancer, as revealed here, may be particularly beneficial in those settings characterized by a lack of electronic health records or individual health history. Notably, no population-based notification system exists in the Primary Health Care Brazilian database (Vieira et al., 2015). It should be emphasized that none of the 20 families that fulfilled the established criteria for familial cancer susceptibility syndromes, and that were identified as having moderate to high risk to hereditary cancer syndromes (Lindor et al., 2008;Valdez et al., 2010;NCCN, 2014;Vieira et al., 2015), had their CFH registered in their medical records at the community-based primary care centers where they were followed.
Although cancer management has been predominantly focused on the individuals affected by disease, not on their families, our results indicate that better management may be achieved by including family screening practices in primary care policies. Therefore, to enhance family practices and adherence to health policies, primary care work force should receive training and ongoing education, which should include essential competencies to collect family history, such as, basic genetics and genomics knowledge, communication skills, ability to establish empathetic interpersonal relationships, and capacity to deal with relevant ethical issues (Flória-Santos et al., 2013).
Even though barriers to incorporating family history taking and hereditary risk assessment in middle-income primary care settings do exist, the wider use of self-reported cancer family history can be a useful tool to achieve this goal.