American Journal of Medical Genetics 117A:154–160 (2003)

Early Use of Clinical BRCA1/2 Testing: AssociationsWith Race and Breast Cancer Risk

Katrina Armstrong,1,2,3,4* Barbara Weber,1,2 Jill Stopfer,1,2 Kathleen Calzone,1,2 Mary Putt,2,3

James Coyne,2,5 and J. Sanford Schwartz1,2,3,4

1Department of Medicine, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania2University of Pennsylvania Cancer Center, Philadelphia, Pennsylvania3Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine,Philadelphia, Pennsylvania4Leonard Davis Institute of Health Economics, University of Pennsylvania, Philadelphia, Pennsylvania5Department of Psychiatry, University of Pennsylvania School of Medicine, Philadelphia, Pennsylvania

WhenBRCA1/2 testing became commerciallyavailable in 1996, many U.S. experts voicedconcern about the potential for indiscrimi-nate use of testing among low-risk women.Supporting this concern, several early sur-veys of interest in genetic testing suggestedthat genetic testing for cancer susceptibilitymight appeal most to individuals at low riskof carrying a mutation. To identify factorsassociated with early use of clinicalBRCA1/2testing, a case-control study was conduct-ed at a large academic health system in themetropolitan Philadelphia region. A total of167 women underwent genetic counselingfor clinical BRCA1/2 testing between 1996and 1997 (cases) compared with 138 womenwho were seen in faculty general internalmedicine practices over the same period(controls). In this study we measured therisk factors for breast cancer, the riskfactors for carrying a BRCA1/2 mutation,and sociodemographic characteristics. Useof BRCA1/2 counseling between 1996 and1997 was positively associated with familybut no personal history of breast cancer(odds ratio (OR), 22.4; 95% confidence inter-val (CI), 9.3–54.3); family and personal his-tory of breast cancer (OR, 150.3; 95% CI, 24.1–939.6); being Caucasian and non-Jewish (OR,4.1; 95% CI, 1.3–13.5); being Caucasian and

Jewish (OR, 8.8; 95% CI, 2.2–35.5); and beingmarried (OR, 3.2; 95% CI, 1.6–6.3). Use ofBRCA1/2 counseling was inversely asso-ciated with increasing age (OR, 0.07; 95% CI,0.02–0.28 for >60 compared to <50). Assuggested by the association with familyhistory, use of counseling was associatedwith having a higher predicted risk of breastcancer and a higher predicted risk of carry-ing a BRCA1 mutation (P<0.0001). Womenwho sought clinical BRCA1/2 testing in theyear after it became commercially availablewere not the ‘‘worried well,’’ but women atsignificantly increased risk of carrying amutation. However, even after adjusting forbreast cancer risk, there was a substantialracial disparity in use of BRCA1/2 testing.These findings suggest that ensuring equalaccess to testing for high-risk individualsirrespective of race may be as important forthe future of predictive genetic testing asrestricting the use of testing among low-riskindividuals. � 2003 Wiley-Liss, Inc.

KEY WORDS: BRCA1; BRCA2; breast can-cer risk; race

INTRODUCTION

BRCA1 andBRCA2were two of the first major cancersusceptibility genes to be identified [Miki et al., 1994;Wooster et al., 1995]. Germline mutations in BRCA1/2convey as high as an 85% lifetime risk of breast cancerand 40% lifetime risk of ovarian cancer [Easton et al.,1995; Struewing et al., 1997]. However, because theprevalence ofmutations in the general population is lessthan 1/800,BRCA1/2mutations explain less than 5% ofthe overall population burden of breast cancer [Fordet al., 1995; Newman et al., 1998]. Thus, althoughtesting women from hereditary cancer families can pro-vide important information about cancer risk, testing

Grant sponsor: American Cancer Society Clinical ResearchTraining Grant; Grant sponsor: Abramson Family CancerResearch Institute.

*Correspondence to: Katrina Armstrong, M.D., 1233 BlockleyHall, 423 Guardian Drive, Philadelphia, PA 19104-6021.E-mail: karmstro@mail.med.upenn.edu

Received 24 April 2002; Accepted 22 July 2002

DOI 10.1002/ajmg.a.10928

� 2003 Wiley-Liss, Inc.

women from the general population results in almostuniformly negative tests that provide little incremen-tal information [Hoskins et al., 1995]. Furthermore,charges forBRCA1/2 testing can be over $2,000, genetictest results are easily misinterpreted, and women whotestnegativemaybe falsely reassuredabout their breastcancer risk and less likely to adhere to routine breastcancer screening [Giardiello et al., 1994; Lerman andCroyle, 1994]. For these reasons, in the years soon afterthe identification ofBRCA1 and 2, most experts and pro-fessional organizations recommended that BRCA1/2testing be restricted to individuals at increased risk ofcarrying a mutation, and many recommended thattesting be restricted to a research setting [AmericanSociety of Human Genetics, 1994; National AdvisoryCouncil for Human Genome Research, 1994; AmericanSociety of Clinical Oncology, 1996; National Action Planon Breast Cancer, 1996].

Despite these recommendations, BRCA1/2 testingbecame commercially available in the United States in1996, increasing concern about indiscriminate use oftesting among individuals at low risk of carrying amutation [Kolata, 1996; Nelson, 1996]. Furthermore,this growing concern was supported by several surveysthat suggested that interest in genetic testing for cancersusceptibility might not be related to cancer risk andmight even be highest among individuals at low risk[Andrykowski et al., 1997; Tambor et al., 1997; Arm-strong et al., 2002]. To our knowledge, although therehave been multiple studies examining the use of testingamong hereditary breast and ovarian cancer families,there is very little data about the factors associatedwiththe early use of clinical BRCA1/2 testing in the UnitedStates [Lermanet al., 1996;Lynch et al., 1997;Bieseckeret al., 2000].

Thus, the overall goal of this study was to investigatethe relationships between risk of breast cancer, risk ofa BRCA1/2 mutation, sociodemographic factors, andthe use of BRCA1/2 testing between 1996 and 1997,the year after testing became commercially available.Because our primary interest was in understanding thecharacteristics of women who sought BRCA1/2 testingand we have previously described the factors associatedwith the decision to undergo testing after pretest coun-seling, we chose to focus on the use of genetic counselingforBRCA1/2 testing in a large academic health system,rather than final decisions about testing [Armstronget al., 2000].

METHODS

Study Design

We conducted a case-control study within the Uni-versity of Pennsylvania Health System (UPHS), an in-tegrated network of five hospitals and over 300 primarycare physicians in the Philadelphia metropolitan area.The structure of the UPHS encourages primary carephysicians to refer for specialty care within the UPHS.BetweenNovember1996andOctober 1997, theperiod ofthe study, counseling for BRCA1/2 testing within theUPHS was provided solely by the Breast Cancer RiskEvaluation Program (BCREP). Over this time period,

the BCREP made no attempt to assess for appropriate-ness forBRCA1/2 testing at the time a patient inquiredabout the program or scheduled a visit. Although initialclinical testing was restricted to individuals with acancer diagnosis, unaffected individuals began to beoffered clinicalBRCA1/2 testing through the BCREP inNovember1996.The researchprotocol for this studywasapproved by the University of Pennsylvania Institu-tional Review Board.

Study Population and Subject Selection

Both cases and controls were selected from the sourcepopulation of all adultwomenwho receivedmedical carethrough the UPHS prior to November 1996, the begin-ning of the study. Cases included all women from thesource population who participated in BRCA1/2 coun-seling at the BCREP between November 1996 andOctober 1997. Controls were selected from the sourcepopulation after excluding all cases. Because geneticcounseling and testing are expensive and less accessibleto low-income individuals without health insurance, werestricted our sampling of controls to UPHS facultygeneral internal medicine practices that have the samehealth insurance profile (>95% private insurance) andapproximately the same visit charge as the BCREP.This design provided an efficient method of examiningassociations with the use of BRCA1/2 counseling andtesting within a group of womenwith financial access tofacultypractices atanacademichealth center.However,it prevented us from determining the effect of financialaccess onuse of counseling and testing.A simple randomsample of 200 controls was selected fromall women seenat these practices betweenNovember 1996 and Septem-ber 1997.

Data Collection

Prior to being evaluated at the BCREP, each casesubject completed a questionnaire on family history,reproductive history, and sociodemographic character-istics, including education, occupation, and ethnic back-ground. In October 1997, the control groupwasmailed aquestionnaire similar to that administered to case sub-jects, but completed anonymously. The control ques-tionnaire was shorter and collected less extensivepedigree and lifestyle information than the case ques-tionnaire.A reminder postcardwasmailed 3weeks afterthe initial mailing, and a second questionnaire mailed6weeksafter the initialmailing.Because the surveywasanonymous, all subjects received follow-up mailings.

Statistical Analysis

All variables and their distributions were examinedusing descriptive statistics. The primary analysis com-pared women undergoing BRCA1/2 counseling (cases)and women not undergoing BRCA1/2 counseling (con-trols) using the ordinary independent sample t-test forcontinuous variables and the ordinary chi-square testfor categorical variables. For each comparison, multiplelogistic regression was used to adjust the association

Use of BRCA1/2 Testing 155

between the primary predictor variables of interest andthe outcome of interest for potential confounding andeffect modification. Because ethnicity and religion werehighly correlated in our sample (all Jewish women wereCaucasian), these variables were not entered separa-tely into the same regression model. Instead, indicatorvariableswere created that represented combinations ofethnicity and religion (non-Caucasian and non-Jewish,Caucasian and non-Jewish, Caucasian and Jewish).Because breast cancer diagnosis and family history ofbreast cancer were significantly correlated, indicatorvariables for breast cancer historywere also constructed(no family or personal history of breast cancer, personalbut no family history of breast cancer, family but nopersonal history of breast cancer, and both family andpersonal history of breast cancer). To provide a moreclinically useful variable, we categorized age into threegroups (<50, 51–60, >60). Final models included allvariables whose inclusion altered the odds ratio (OR)for another variable by 10% or more [Breslow andDay, 1994]. Model fit was assessed using the Hosmer-Lemeshow test for goodness of fit [Hosmer and Leme-show, 1989].

The predicted lifetime risk of breast cancer for eachsubject without a prior diagnosis of breast cancer wascalculated from tables developed by Claus et al. [1994]from the Cancer and Steroid Hormone Study, a largepopulation-based case-control study of breast cancer.The pretest probability of carrying a BRCA1 mutationfor each subject was estimated from prediction tablesgenerated by Couch et al. [1997], as the study cohortused for this model most closely approximates theparticipants in the current study. Because the pretestprobability of carrying a BRCA2mutation could only becalculated for the two controlswho had a familymemberdiagnosed with breast cancer before age 50, we wereunable to examine predicted risk of a BRCA2 mutation[Frank et al., 1997]. Predicted risks were analyzed ascontinuous variables and, because of skewed distribu-tions, as dichotomous variables (above population levelrisk vs. population level risk and below). Assuming asurvey response rate of 70%, a sample size of 200 controlsubjects was designed to detect a 0.01 absolute differ-ence in risk of a BRCA1 mutation as a continuousvariable between women who underwent testing andcontrolswith a power of 90%and alpha of 0.05 [DesuandRaghavarao, 1990]. A baseline predicted BRCA1 muta-tion risk of 0.005 in controls was assumed.

RESULTS

Of the 198 women from the source population seen atthe BCREP over the study period, 184 (93%) completedthe questionnaire. Seventeen of these women were notseekingBRCA1/2 testing at the time of their initial visitand were excluded, leaving 167 cases. After counseling,76 (47%) of the cases underwent BRCA1/2 testing. Ofthe 200 women from the general internal medicinepractice selected as controls, 4 had died and 7 hadmoved. Of the remaining 189, 138 (73%) completed thequestionnaire. Characteristics of cases and controls areshown in Table I.

Women undergoing BRCA1/2 counseling were sig-nificantly more likely to have a family history of breastcancer than women in the general internal medicinepractices. Personal history of breast cancer was asso-ciated with use of BRCA1/2 counseling among womenwith a family history of breast cancer, but not in theabsence of a family history of breast cancer. Personaland family history of breast cancer was strongly asso-ciated with use of counseling. Women undergoingBRCA1/2 counseling were more likely to be Caucasianand, among Caucasian women, more likely to be Jewishthan women in the general internal medicine practice.Because all Jewish women were also Caucasian, theeffect of Jewish heritage on use of counseling amongnon-Caucasian women could not be assessed. Age andmarital status were also strongly associated with use ofcounseling,withwomenunder 50more likely toundergocounseling than women over 50, and married womenmore likely to undergo counseling than nonmarriedwomen. The predicted lifetime risk of breast cancer andpredicted risk of carrying a BRCA1 mutation werehigher in women participating in BRCA1/2 counselingthan in controls. The strengths of these associations areshown in Table II.

After multivariable adjustment, family history ofbreast cancer, family and personal history of breastcancer, Caucasian ethnicity without Jewish heritage,Caucasian ethnicity with Jewish heritage, younger age,and marital status remained associated with use ofcounseling (Table III). Education and employment werenot significant confounders. Adjustment for maritalstatus, age, ethnicity, and Jewish heritage had littleeffect on the association of counseling with predictedbreast cancer risk (for comparison of population risk vs.increased risk: OR, 21.8; 95% confidence interval (CI),9.0–52.8). Adjustment for marital status and age hadlittle effect on the association of counseling with thepredicted risk of BRCA1 mutation (for comparison ofpopulation risk vs. increased risk: OR, 14.8; 95% CI,6.6–33.2).

TABLE I. Subject Characteristics

Cases(n¼167)

Controls(n¼138)

Mean age (�SD) 44.1�10.8 52.4�16.0Breast CA diagnosis (%) 25.7 10.1Ovarian CA diagnosis (%) 1.8 3.6Family history breast CA (%) 89.6 30.7Caucasian (%) 95.1 63.4Jewish (%) 26.7 16.2Married (%) 83.3 48.2Employed outside home (%) 74.2 65.2College educated (%) 68.6 66.9Predicted BRCA1 mut risk

Mean (�SD) 0.18�0.23 0.017�0.04>12% (%)a 74.8 10.3

Predicted breast CA riskb

Mean (�SD) 0.20�0.97 0.094�0.04>0.1% (%)a 89.5 49.4

aPopulation level risk.bAmong women without breast cancer.

156 Armstrong et al.

DISCUSSION

In the present study of a single large academic healthsystem, womenwho sought clinicalBRCA1/2 testing inthe year after testing became commercially availablewere at substantially higher risk of breast cancer and ofhaving aBRCA1mutation thanwomenwhodidnot seektesting. This relationship held true whether risk wasassessedby risk factors for breast cancer (familyhistory)or risk factors for BRCA1 mutation (family history ofbreast cancer and Ashkenazi Jewish heritage) or withrisk prediction models. However, after adjusting forbreast cancer risk, women seeking BRCA1/2 testingwere also substantially more likely to be Caucasian,married, and younger—factors not known to be asso-ciated with mutation risk.

The strong association between breast cancer risk(measured either as family history or predicted risk) andthe early use of clinical BRCA1/2 testing is reassuring,

particularly because it differs from what might havebeen predicted based on prior studies of interest intesting [Lerman et al., 1994, 1995; Chaliki et al., 1995;Struewing et al., 1995; Andrykowski et al., 1997;Jacobsen et al., 1997; Tambor et al., 1997; Armstronget al., 2002]. The proportion of women from the generalpopulation interested in undergoing a ‘‘breast cancergene test’’ (60–90%) in many prior surveys was onlyslightly lower than the proportion of first-degree rela-tives of breast or ovarian cancer patients (75–90%)interested in testing. Furthermore, two surveys of thegeneral population found that interest in genetic testingfor cancer risk was higher among individuals without afamily history of cancer than among those with a familyhistory [Andrykowski et al., 1997; Armstrong et al.,2002]. In contrast, this study suggests that the womenwho actually sought BRCA1/2 testing in the year aftertesting became commercially available were at signifi-cantly increased risk of carrying a mutation, allaying

TABLE II. Unadjusted Associations With Use of BRCA1/2 Counseling

OR 95% CI

Breast CA riskNo family or personal history breast CAa 1.00Personal history breast CA 0.60 0.07–5.04Family history breast CA 15.39 8.02–29.54Family and personal history breast CA 44.08 15.11–128.61

EthnicityNon-Caucasian, non-Ashkenazia 1.00Caucasian, non-Ashkenazi 10.76 4.79–24.18Caucasian, Ashkenazi 12.00 4.85–29.71

Married 5.38 2.96–9.76Age

<50a 1.0051–60 0.37 0.21–0.67>60 0.16 0.08–0.32

Predicted breast CA riskb

�12%a 1.00>12% 24.40 12.21–48.63

Predicted BRCA1 mut risk�0.1%a 1.00>0.1% 14.74 7.64–28.37

aReference group.bAmong women without breast cancer.

TABLE III. Adjusted Associations With Use of BRCA1/2 Counseling

OR 95% CI

Breast CA riskNo family or personal history breast CAa 1.00Family history breast CA 22.43 9.3–54.2Family and personal history breast CA 150.31 24.1–939.6

EthnicityNon-Caucasian, non-Ashkenazia 1.00Caucasian, non-Ashkenazi 4.08 1.2–13.5Caucasian, Ashkenazi 8.83 2.2–35.4

Married 3.21 1.4–7.5Age

<50a 1.0051–60 0.50 0.2–1.2>60 0.07 0.02–0.3

aReference group.

Use of BRCA1/2 Testing 157

concerns that commercial availability would quicklylead to indiscriminate use of BRCA1/2 testing by theworried well.

There are several possible explanations for why theactual use of clinical BRCA1/2 testing differed fromwhat would have been expected based on earlier studiesof interest in testing. Primary care and other physiciansare likely to play an important role in both referringwomen at increased risk for genetic counseling anddissuading women at low risk from pursuing genetictesting. Intentions are well recognized to differ frombehavior [Glanz et al., 1997].Most previous studieswereconducted before the development of clinical testing forBRCA1/2 and often referred to hypothetical gene tests.The realities of clinical BRCA1/2 testing, including thecost, may be less appealing to low-risk women. Educa-tional efforts over the past several years are likely tohave increased understanding of the risks and limita-tions of testing among the general public. In addition,prior studies found very high rates of interest in testing,perhaps making it more difficult to find an associationwith underlying level of risk. This difference betweenreported interest in testing prior to its clinical avail-ability and actual use following clinical availability hasalso been seen with other genetic tests, including thosefor Huntingtons disease and cystic fibrosis [Craufurdet al., 1989; Tambor et al., 1994].

In contrast to the association with mutation risk, theassociation between race and early use of BRCA1/2counseling found in the current study is supported byprior surveys of knowledge about and interest in testing[Struewing et al., 1995; Hughes et al., 1996]. This find-ing has several implications for current research andpolicy in genetic susceptibility testing. It is important toestablish whether this racial disparity has persisted tothe current use of testing and if it is generalizableoutside of this single health care system and this singlegenetic test. Given the existence of a racial disparity, itis critical to begin to understand the causes of thedisparity. In a setting like BRCA1/2 testing, wherethere are no clear guidelines for who should and shouldnot undergo testing, distinguishing causes such asracism and lack of financial resources from causes suchas cultural beliefs and values is critical in order todetermine to what extent this disparity representsdifferences in preferences, differences in access, ordifferences in awareness. Furthermore, as the evidencebuilds to suggest that BRCA1/2 testing may havesignificant implications for cancer surveillance andprevention, it becomes increasingly important todevelop strategies to overcome barriers that preventhigh-risk African-American women from makinginformed decisions about testing [Rebbeck et al., 1999;Meijers-Heijboer et al., 2001].

In our sample, a personal diagnosis of breast cancerwas associated with use of testing, but only amongwomenwith a familyhistory of breast cancer. In familieswithout known BRCA1/2 mutations, it is well recog-nized that women with a cancer diagnosis gain moreinformation from BRCA1/2 testing than women with-out a cancer diagnosis. A negative BRCA1/2 test in awoman with cancer is generally taken to mean that the

family does not carry a mutation, whereas a negativetest in awomanwithout cancer is not informative for thefamily. Thus, while it is not surprising that womenwitha cancer diagnosis are more likely to seek genetictesting, it is interesting that this effect is confined towomen with a family history of breast cancer. Even inthe face of a personal cancer diagnosis, women and theirproviders appear to recognize that a family history ofcancer significantly increases the risk of hereditarycancer susceptibility.

The effect of marital status and age on the utilizationof genetic counseling and testing may relate to theimplications of genetic testing and prophylactic surgeryfor families and childbearing.Anecdotally,manywomenseek genetic testing to gather information for otherfamily members or to inform decisions about prophylac-tic surgery following the completion of childbearing.These reasons may be more common or salient amongwomen who are married or younger.

This study has several limitations. If women from theUPHS underwent clinical BRCA1/2 testing outside oftheBCREP, theywouldnothave been captured as cases.If these women exist and are systematically differentfrom women participating in the program, their exclu-sion could bias the study results. However, the geo-graphic distance of other clinical counseling services,the strong tendency to refer within the UPHS, the em-phasis on pretest counseling for BRCA1/2 testing, andthe national reputation of the UPHSBCREP all make itunlikely that a significant proportion of women from thesource population pursued testing elsewhere duringthe study period. The use of a control population from afaculty practice at the academic medical center allowedus to address the factors that broughtwomen to counsel-ing among women with financial access to an academicmedical center; however, it prevented us from look-ing at factors that might affect that access. Althougheducation and employment were not associated withutilization of BRCA1/2 counseling among women withfinancial access, both are likely to affect financial access.Although the response rate was high for both cases andcontrols, there was a substantial difference in responserate between the twogroups.Thus, it is possible that anyresponse bias, where nonresponders differed from re-sponders, could have biased our results. However, sincethe higher prevalence of family history of breast canceramong the controls than reported in several previousstudies suggests that controls with a family history ofbreast cancer may have been more likely to return thesurvey, it is likely that any nonresponse biaswould haveled to an underestimation of the strength of the asso-ciation between breast cancer risk and use of testing[Colditz et al., 1993; Slattery and Kerber, 1993]. Weattempted to make data collection similar for case andcontrols, using the same items and structure whenpossible. It is possible that differences in the process orinstruments could have led to differences in measure-ment between cases and controls.However, it is unlikelythat differences in measurement could have explainedthe association with race or breast cancer risk. Westudied only a single academic health system witha single site of BRCA1/2 counseling and testing over

158 Armstrong et al.

a relatively brief period of time. Further studies areneeded to determine how well these results predictcurrent use ofBRCA1/2 testing or the early use of othergenetic tests in the future.

Despite these limitations, this study provides someimportant insight into the early uptake of clinicalBRCA1/2 testing use in the United States. While itis reassuring to discover that women who soughtBRCA1/2 testing in the year after it became commer-cially available were not the ‘‘worried well,’’ but at in-creased risk of carrying a mutation, mutation riskexplained only one part of the picture. Women whosought BRCA1/2 testing were also significantly lesslikely to be African-American, even after adjusting forcancer risk. Furthermore, these results about early useof testing shouldnot engender complacencyabout futureuse of genetic testing. The future of genetic suscept-ibility testing is likely to lie in community primary carepractice, not in academic centers. As this transition un-folds, the challenge is developinghealthand social policythat will continue to target women at risk while provid-ingaccess towomenofall ethnic andracial backgrounds.

ACKNOWLEDGMENTS

Dr. Armstrong is supported by an American CancerSociety Clinical Research Training Grant and a RobertWood Johnson Foundation Generalist Faculty ScholarAward.

REFERENCES

American Society of Clinical Oncology. 1996. Statement of the AmericanSociety of Clinical Oncology: genetic testing for cancer susceptibility.J Clin Oncol 14:1730–1736.

American Society of Human Genetics. 1994. Statement of the AmericanSociety of Human Genetics on genetic testing for breast and ovariancancer predisposition. Am J Hum Genet 55:i–iv.

Andrykowski MA, Lightmer R, Studts JL, Munn RK. 1997. Hereditarycancer risk notification and testing: how interested is the generalpopulation? J Clin Oncol 15:2139–2148.

Armstrong K, Calzone K, Stopfer J, Fitzgerald G, Coyne J, Weber B. 2000.Factors associated with decisions about clinical BRCA1/2 testing.Cancer Epidemiol Biomarkers Prev 9:1251–1254.

Armstrong K, Weber B, Ubel PA, Guerra C, Schwartz JS. 2002. Interest inBRCA1/2 testing in a primary care population. Prev Med 34:590–595.

Biesecker BB, Ishibe N, Hadley DW, Giambarresi TR, Kase RG, Lerman C,Struewing JP. 2000. Psychosocial factors predicting BRCA1/BRCA2testing decisions in members of hereditary breast and ovarian cancerfamilies. Am J Med Genet 93:257–263.

Breslow NE, Day NE. 1994. Statistical methods in cancer research. Lyon,France: International Agency for Research on Cancer.

Chaliki H, Loader S, Levenkron J, Logan-Young W, Hall W, Rowley PT.1995.Women’s receptivity to testing for a genetic susceptibility to breastcancer. Am J Pub Health 85:1133–1135.

Claus EB, Risch N, Thompson WD. 1994. Autosomal dominant inheritanceof early-onset breast cancer. Cancer 73:643–651.

Colditz GA,WillettWC, Hunter DJ, et al. 1993. Family history, age and riskof breast cancer. Prospective data from theNurses Health Study. JAMA270:338–343.

Couch FJ, DeShano ML, Blackwood MA, Calzone K, Stopfer J, Campeau L,Ganguly A, Rebbeck T, Weber BL. 1997. BRCA1 mutations in womenattending clinics that evaluate the risk of breast cancer. N Engl J Med336:1409–1415.

Craufurd D, Dodge A, Kerzin-Storrar L, Harris R. 1989. Uptake ofpresymptomatic predictive testing for Huntington’s disease. Lancetii:603–605.

Desu MM, Raghavarao D. 1990. Sample size methodology. San Diego:Academic Press. p 55–60.

EastonDF, FordD,BishopDT. 1995. Breast and ovarian cancer incidence inBRCA1 mutation carriers. Am J Hum Genet 56:265–271.

Ford D, Easton DF, Peto J. 1995. Estimate of the gene frequency of BRCA1and its contribution to breast and ovarian cancer incidence. Am J HumGenet 57:1457–1462.

Frank T, Manley S, Olopade O, Cummings S, Garber JE, Bernhardt B,Antman K, Russo D, Wood ME, Mullineau L, Isaacs C, Peshkin B, BuysS, Venne V, Rowley PT, Loader S, Offit K, RobsonM, Hampel H, BrenerD, Winer EP, Clark S, Weber B, Strong LC, Thomas A. 1997. Sequenceanalysis of BRCA1 and BRCA2: correlation of mutations with familyhistory and ovarian cancer risk. J Clin Oncol 16:2417–2425.

Giardiello F, Brensinger J, PetersenG, LuceM,Hylind L. 1994. The use andinterpretation of commercialAPC gene testing for familial adenomatouspolyposis. N Engl J Med 336:823–827.

Glanz K, Lewis FM, Rimer BK, editors. 1997. Health behavior and healtheducation: theory, research and practices. San Francisco: Jossey-Bass.p 11–123.

Hoskins KF, Stopfer JE, CalzoneKA,Merajver SD, Rebbeck TR, Garber JE,Weber BL. 1995. Assessment and counseling for women with a familyhistory of breast cancer: a guide for clinicians. JAMA 273:577–585.

Hosmer DW, Lemeshow S. 1989. Applied logistic regression. New York:John Wiley & Sons. p 135–145.

Hughes C, Lerman C, Lustbader E. 1996. Ethnic differences in riskperception among women at increased risk for breast cancer. BreastCancer Res Treatment 40:25–30.

Jacobsen PB, Valdimarsidottir HB, Brown KL. 1997. Decision-makingabout genetic testing among women at familial risk for breast cancer.Psychosom Med 59:459–466.

Kolata G. Breaking ranks, lab offers test to assess risk of breast cancer. NewYork: Times, April 1, 1996. Section A, Page 1, Column 1.

Lerman C, Croyle R. 1994. Psychological issues in genetic testing for breastcancer susceptibility. Arch Intern Med 154:609–616.

Lerman C, Daly M, Masny A, Balshem A. 1994. Attitudes about genetictesting for breast-ovarian cancer susceptibility. J Clin Oncol 12:843–850.

Lerman C, Seay J, Balshem, Audrain J. 1995. Interest in genetic testingamong first-degree relatives of breast cancer patients. Am J Med Genet57:385–392.

LermanC,Narod S, SchulmanK,HughesC,Gomez-CamineroA, BonneyG,Gold K, Trock B, Main D, Lynch J, Fulmore C, Snyder C, Lemon SJ,Conway T, Tonin P, Lenoir G, LynchH. 1996.BRCA1 testing in familieswith hereditary breast-ovarian cancer: a prospective study of patientdecision making and outcomes. JAMA 274:1885–1892.

LynchHT, Lemon SJ, DurhamC, Tinley ST, Connolly C, Lynch JF, SurdamJ, Orinion E, Slominski-Caster S, Watson P, Lerman C, Tonin P, LenoirG, Serova O, Narod S. 1997. A descriptive study of BRCA1 testing andreactions to disclosure of test results. Cancer 79(11):2219–2228.

Meijers-Heijboer H, van Geel B, van Putten WL, Henzen-Logmans SC,Seynaeve C, Menke-Pluymers MB, Bartels CC, Verhoog LC, van denOuweland AM, Niermeijer MF, Brekelmans CT, Klijn JG. 2001. Breastcancer after prophylactic bilateral mastectomy in women with a BRCA1or BRCA2 mutation. N Engl J Med 345:159–164.

Miki Y, Swensen J, Shattuck-EidensD, Futreal PA,HarshmanK,TavtigianS, Liu Q, Cochran C, Bennett LM, Ding W. 1994. A strong candidate forthe breast and ovarian cancer susceptibility gene BRCA1. Science266:66–71.

National Action Plan on Breast Cancer. 1996. Position paper: hereditarysusceptibility testing for breast cancer. J Clin Oncol 14:1738–1740.

National Advisory Council for Human Genome Research. 1994. Statementon use of DNA testing for presymptomatic identification of cancer risk.JAMA 271:785.

NelsonNJ. 1996.Caution guides genetic testing for hereditary cancer genes.J Natl Cancer Inst 88:70–72.

Newman B, Mu H, Butler LM, Millikan RC, Moorman PG, King MC. 1998.Frequency of breast cancer attributable toBRCA1 in a population-basedseries of American women. JAMA 279:893–972.

Rebbeck TR, Levin AM, Eisen A, Snyder C, Watson P, Cannon-Albright L,Isaacs C, Olopade O, Garber JE, Godwin AK, Daly MB, Narod SA,Neuhausen SL, Lynch HT, Weber BL. 1999. Reduction in breast cancerrisk following bilateral prophylactic oophorectomy in BRCA1 mutationcarriers. J Natl Cancer Inst 91:1475–1479.

Use of BRCA1/2 Testing 159

Slattery ML, Kerber RA. 1993. A comprehensive evaluation of familyhistory and breast cancer risk. The Utah Population Database. JAMA270:1563–1568.

Struewing JP, Lerman C, Kase RG, Giambarresi TR, Tucker MA. 1995.Anticipated uptake and impact of genetic testing in hereditary breastand ovarian cancer families. Cancer Epidemiol Biomarkers Prev 4:169–173.

Struewing JP, Hartge P, Wacholder S, Balker SM, Berlin M, McAdams M,Timmerman MM, Brody LC, Tucker MA. 1997. The risk of cancerassociated with specific mutations of BRCA1 and BRCA2 amongAshkenazi Jews. N Engl J Med 336:1401–1408.

Tambor ES, Bernhardt BA, Chase GA, Faden RR, Geller G, Hofman KJ,Holtzman NA. 1994. Offering cystic fibrosis carrier screening to anHMO population: factors associated with utilization. Am J Hum Genet55:626–637.

Tambor ES, Rimer BK, Strigo TS. 1997. Genetic testing for breast cancersusceptibility: awareness and interest among women in the generalpopulation. Am J Med Genet 68:43–49.

Wooster R, Bignell G, Lancaster J, Swift S, Seal S, Mangion J, Collins N,Gregory S, Gumbs C, Micklem G. 1995. Identification of the breastcancer susceptibility gene BRCA2. Nature 378:789–792.

160 Armstrong et al.