CORRESPONDENCE Re: Biologic Characteristics of Interval and Screen-Detected Breast Cancers The goal of mammography is to de- tect breast cancer before it spreads be- yond the breast. Two recent studies (1,2) have compared histologic and other prog- nostic features of mammographically detected breast cancers with cancers that were detected clinically in the interval following a negative screen. These stud- ies find that interval cancers are of higher grade and are more likely to have ad- verse prognostic features (such as being p53 positive and ER negative) than are mammogram-detected cancers. The tumors that are missed by mammogra- phy may, therefore, be more likely to metastasize and to lead to early fatality. Young age and a positive family history of breast cancer have also been associ- ated with screening failure (3). These findings may be relevant for young women from high-risk groups who are enrolled in breast screening programs; for example, the majority of breast can- cers associated with BRCA1 mutations appear in young women and are of high grade, are p53 positive and are ER nega- tive (4). The authors conclude from these comparisons that interval cancers are rapidly growing tumors and that the tu- mors were likely too small to have been detected on the previous mammogram. Another interpretation is that mammo- graphic detectability per se is associated with favorable clinicopathologic fea- tures. To address this question, we re- viewed the records of the Ontario Breast Cancer Screening Program for women diagnosed with breast cancer in the province from 1991 through 1998. The women in this program are screened by both clinical breast examination (CBE) and by mammography, usually on the same day. We compared the 84 tumors that were detected by CBE alone with 643 tumors detected by mammog- raphy alone. We restricted our study to women whose tumors were at least 1 cm in size. There was no difference in the mean age of the two groups (62.6 years for CBE-detected tumors versus 63.4 years for mammogram-detected tumors). Tu- mors detected by CBE alone were, on average, larger than those detected by mammography (2.1 versus 1.6 cm; P<.001), were more likely to be of lobu- lar or mixed lobular histology (31% ver- sus 11%; P<.001), and were more likely to be lymph-node positive (34% versus 20%; P  .01). Lobular histology is also an adverse prognostic feature and has been reported previously to be over- represented in interval-detected tumors (2,5). Our data are consistent with the hypothesis that mammography preferen- tially identifies cancers with favorable prognostic features. However, our study differs from previous studies in that the CBE-detected cancers were detected on the same day as the negative mammo- gram; therefore, we cannot claim that the mammography would have per- formed better if the screening interval were shortened. STEVEN A. NAROD MARIE-PIERRE DUBE ´ REFERENCES (1) Gilliland FD, Joste N, Stauber PM, Hunt WC, Rosenberg R, Redlich G, et al, Biologic char- acteristics of interval and screen-detected breast cancers. J Natl Cancer Inst 2000;92:743–9. (2) Porter PL, El-Bastawissi AY, Mandelson MT, Lin MG, Khalid N, Watney EA, et al. Breast tumor characteristics as predictors of mammo- graphic detection: comparison of interval- and screen-detected cancers. J Natl Cancer Inst 1999;91:2020–8. (3) Kerlikowske K, Grady D, Barclay J, Sickles EA, Ernster V. Effect of age, breast density, and family history on the sensitivity of the first screening mammography. JAMA 1996;276: 33–8. (4) Phillips KA, Nichol K, Ozcelik H, Knight J, Done SJ, Goodwin PJ, et al. Frequency of p53 mutations in breast carcinomas from Ash- kenazi Jewish carriers of BRCA1 mutations. J Natl Cancer Inst 1999;91:469–73. (5) Ma L, Fishell E, Wright B, Hanna W, Allan S, Boyd NF. Case–control study of factors asso- ciated with failure to detect breast cancer by mammography. J Natl Cancer Inst 1992;84: 781–5. NOTES Affiliation of authors: Centre for Research on Womens Health, Sunnybrook and Womens Col- lege Hospital, University of Toronto, ON Canada. Correspondence to: Steven A. Narod, M.D., Centre for Research on Womens Health, Sunny- brook and Womens College Hospital, University of Toronto, 790 Bay, Toronto, ON M5G 1N8, Canada. RESPONSES Recent articles in the Journal by Gil- liland et al. (1) and Porter et al. (2) sug- gest that cancers that are detected in the interval between screening mammo- grams are more likely than mammo- graphically detected tumors to have fea- tures, such as high tumor grade, estrogen receptor negativity, and p53 positivity. One explanation for this is that mammographically occult tumors with such features might proliferate rap- idly, growing larger, and becoming pal- pable in the interval between mammo- grams. Indeed this argument has been used to suggest there is equipoise for research into the benefits and risks of reducing the time between screening ex- aminations (1). However, an alternate hypothesis is that tumors with these fea- tures might be less easily radiographi- cally detectable. Unfortunately, in nei- ther of these articles is there data on whether the interval-detected tumors could be seen on diagnostic mammo- grams performed at the time of their clinical detection. In an attempt to ad- dress this issue, Narod et al. present data from the Ontario Breast Cancer Screen- ing Program showing that palpable tu- mors not visualized by mammography were more likely to have lobular or mixed lobular histology, to be larger, and to be axillary lymph node positive. However, it is unclear from the data pre- sented whether these features were inde- pendently statistically significant on multivariate analysis or whether the fea- tures highlighted as important in the pre- vious articles (tumor grade, hormone re- ceptor status, and p53 mutation status) were examined. Indeed, the major find- ing seems to be that lobular tumors are more likely to be mammographically occult, an observation that has been well documented previously (3,4). It is suggested that the findings of the studies by Gilliland et al. (1) and Porter et al. (2) may have implications for mammographic screening of BRCA1 mutation carriers. BRCA1-associated breast cancers have a relatively homo- geneous phenotype that generally in- cludes high grade, p53 positivity, and estrogen receptor negativity (5). Indeed, one small study of the mammographic appearances of breast tumors from BRCA1 mutation carriers showed that Journal of the National Cancer Institute, Vol. 93, No. 2, January 17, 2001 CORRESPONDENCE 151 Downloaded from https://academic.oup.com/jnci/article-abstract/93/2/152/2910305 by guest on 04 June 2020