Interval Breast Cancers Are Not Biologically Distinct---Just More Difficult to Diagnose Debra Koivunen, MD, Xinchao Zhang, MD, Charles Blackwell, MD, Edward Adelstein, MD, Loren Humphrey, MD, PhD, Columbia, Missouri BACKGROUND: Breast cancer diagnosed within 1 year of a negative annual screening examination is called interval breast cancer (IBC) and is considered to be a m o r e virulent subtype o f disease. METHODS: We reviewed clinical data on 2 4 women who were diagnosed as having IBC while participating in the Breast Cancer Detection Demonstration Project at Ellis Fischel C a n c e r Hospital and the Women's Cancer Control Program screening project in Columbia, Missouri, between 1 9 7 4 and 1 9 9 2 . We reinterp r e t e d mammograms from the visit prior to the diagnosis o f IBC for possible misdiagnosis, changes suggestive of malignancy, and Wolfe's patterns. Archival paraffin blocks f r o m 19 patients were used to determine qualitative expression of tumor markers. RESULTS: Observed 5-, 8-, and 10-year survival rates were identical to published data for patients with non-lBC. Seventy-four pereent of the mammograms evidenced dysplastic Wolfe's patterns (P2 and DY), and one patient was found retrospectively to have shown evidence o f caneer which was missed. Compared to breast cancers in general, fewer IBC tumors expressed tum o r markers associated with poor prognosis. CONCLUSIONS: S u r v i v a l rates and t u m o r marker expressions in this r e t r o s p e c t i v e cohort suggest that IBC tumors are not more biologically aggressive than noninterval tumors. T h e y are more difficult to diagnose both by physical examination and mammography. ore than 15 years have elapsed since PanoussopouM los et al ~ introduced the concept of "interval breast cancer" (IBC). These malignancies become clinically evident during the 12 months following a normal screening mammogram and physical examination. For several years, IBC has been thought to represent a more virulent form of breast cancer due to its seemingly rapid evolution to clinical significance and because data from some studies suggested that patients with this subset of breast cancer had a shorter survival time. To more fully evaluate the biologic behavior of IBC, we measured the expression of several recently discovered From the Departments of Pathology (EA), Radiology (CB), and Surgery (DK, XZ, LH), University of Missouri School of Medicine, and the Harry S. Truman Memorial Veterans Hospital, Columbia, Missouri. Supported by a grant from Phi Beta Psi. Requests for reprints should be addressed to Debra Koivunen, MD, Department of Surgery, University of Missouri School of Medicine, Columbia. Missouri 65212. Presented at the 46th Annual Meeting of the Southwestern Surgical Congress, Tucson, Arizona, April 17-20, 1994. 538 prognostic tumor markers in archival specimens from a series of patients with this diagnosis. In addition, we evaluated mammograms from these patients using Wolfe's criteria2 and compared the results to the mammographic patterns from published breast cancer series. Our hypothesis was that data from markers and mammograms would support the impression that IBCs are more aggressive compared to non-IBCs. METHODS Patient Selection Patients with IBC were identified from the clientele of the Breast Cancer Detection Demonstration Project (BCDDP) at Ellis Fischel Cancer Hospital and the Women's Cancer Control Program screening project in Columbia, Missouri, between 1974 and 1992. Clinical data were extracted from the screening center records. One of the authors (CB) reviewed mammograms from the screening visits prior to the diagnoses of IBC for (1) misdiagnosed breast cancer, (2) normal appearance, (3) minimal signs not diagnostic of carcinoma, and (4) Wolfe's patterns. Immunohistochemistry Paraffin blocks were available for 19 of the 24 patients. Tissue sections were taken and assayed for tumor markers using an indirect immunoperoxidase method. The blocks were cooled in an ice-water mixture for 30 minutes, after which 30 p-thick sections were cut from each, dried, and heat-fixed to Histostik (Accurate Chemical Co., Westbury, New York) coated slides at 37°C. The slides were deparaffinized through three changes of xylene for 2 minutes each, followed by rehydration using absolute alcohol and phosphate-buffered saline. 3 Endogenous peroxidase was quenched with 3% hydrogen peroxide in methanol for 5 minutes. Sections were covered with blocking agent (normal serum) for 30 minutes in a humid chamber at room temperature. Excess serum was drained off and sections were incubated with the primary monoclonal or polyclonal antimarker antibody (described, as follows) for 1 hour. Peroxidase-labeled secondary antibody (Signet Laboratory, Dedham, Massachusetts) was applied, followed by substrate (chromagen-2% 3-amino-9-ethylcarbazol [AEC]). Between incubations, the slides were rinsed twice with phosphate-buffered saline. The sections were counterstained with hematoxylin and ammonia-water, and a coverslip was applied. The intensity of staining was graded on a scale of 0 (negative) to 3+ (intense positive) by 2 investigators (XZ and EA). Monoclonal and Polyclonal Antibodies Based on our experience with 6 breast-cancer cell lines, as well as others' studies of markers in breast cancer,4-8 we used the following antibody probes: (1) CA 15.3 (Signet Labs) detects the DF3 antigenic determinant; (2) THE AMERICAN JOURNAL OF SURGERY ® VOLUME 169 DECEMBER 1994 INTERVAL BREAST CANCERS/KOIVUNEN ET AL TABLE I Wolfe's Mammographic Classification of Several Published Breast Cancer Series Published Series" No. of Patients N P1 P2 DY Wolfe, 1976 (screened population) 2 5,284 41% 26% 26% 7% Wolfe, 1976 (cancer patients) 2 40 7.5% 15% 57.5% 20% Verbeek et al, 198424 20 30% 30% 40% 0% Gravelle et al, 198623 31 3% 10% 65% 23% Compilation of above three series ~.23,24 91 11% 16.5% 56% 16.5% Ikeda et al, 1992 (interval cancers) 19 96 4% 36.5% 36.5% 23% Koivunen et al, 1994 (interval cancers) 23 9% 17% 44% 30% "See references for cited sources. N = no ductal prominence; P1 = prominent ducts were confined to the anterior quarter of the breast; P2 = extensive ductal prominence; DY = areas of confluent density sufficient to obscure detail of underlying parenchyma. BRST-3 (Signet Labs) detects TAG-72; (3) c-erb B-2 (Signet Labs) stains the c-erb B-2 oncogene; (4) BRST-1 (Signet Labs) detects T47-D glycoprotein; (5) Cath-D (Signet Labs) stains cathepsin-D; (6) BRST-4 (Signet Labs) detects breast Ca antigen associated with poor prognosis; (7) PAb 1801 (Oncogene Science, Manhasset, New York) reacts with all known forms of p53; (8) BRST-5 (Signet Labs) recognizes a breast cancer antigen of malignant epithelial origin in 50% of cases. Three control probes were included in the panel: BRST-2 (Signet Labs), which reacts with gross cystic fluid protein, carcinoembryonic antigen (CEA), and CA 19.9 (a colon cancer marker). RESULTS Twenty-four patients with IBC were identified. The data on the number of visits and cancers detected by the BCDDP from 1974 to 1980, as well as stage and survival characteristics have been reported previously by Rodes et al. 9 From 1974 to 1992, there were 116,052 participant visits and 332 cancers detected. The IBC patients' ages ranged from 38 to 79, with a mean of 59 years. Seventynine percent were postmenopausal, of whom 38% were on estrogen replacement therapy at the time of diagnosis. Three of the 5 premenopausal patients were on hormonal therapy. Median follow-up was 101 months, with a mean of 119 months and a range of 2 to 17 years. The observed survival rates were 100% at 5 years, 90% at 8 years, and 85% at 10 years. These survival rates were not significantly different from the 5-, 8-, and 10-year survival rates of 89%, 83%, and 79% for patients with non-IBC, as reported in the national BCDDP, I° or the 95% 5-year survival for mammographically detected breast cancers in Rodes' series. 9 One of the 24 patients with IBC was diagnosed as hav"ing ductal carcinoma in situ. Another 10 were identified as having stage I tumors, all but 2 of which were ductal carcinomas. The exceptions were 1 tubular adenocarcinoma and 1 intracystic medullary carcinoma. The remaining 13 patients had stage II disease at the time of diagnosis. Two had lobular carcinoma (1 node negative), 1 had node-negative tubular adenocarcinoma, and the rest were ductal carcinomas of varying degrees of differentiation (4 node negative). No patient had clinical signs of distant metastases at the time of presentation. All 5 deaths occurred in patients who presented with stage II disease. Twenty patients underwent modified radical mastectomy. Of the 4 patients who chose breast conservation and radiation therapy as their treatment, 3 had stage I disease. A review of charts for findings on examination by the nurse at the visit prior to detection of the IBC revealed that of 16 patients in whom breast size was noted, 25% were small, 63% medium, and 12% large. This contrasts somewhat with the IBC series by Panoussopoulos et al, l where breast size was small in 25%, medium in 38%, and large in 38% of patients. Nodularity, masses, or thickening were noted to be present in 17 of the 24 IBC patients at the visit prior to detection of the IBC. Two of the 24 had Paget's disease presenting with nipple discharge and a third had an intracystic carcinoma. Breast examination was not critical in detecting these three cancers. Thus, examination was difficult in 17 of 21 IBC patients (81%). One patient's mammograms were lost, leaving 23 studies available for review. One patient's films were believed to have been misdiagnosed (4%), 19 film sets revealed no signs of malignancy on review (83%), and in the remaining 3 there were minimal signs present that were not diagnostic for carcinoma (13%). These figures compare favorably to a recent literature review where 13% to 30% of previous screening mammograms of patients with IBC were found to have been misinterpreted on retrospective review, 33% to 58% of films showed no changes at all, and 28% to 38% of films showed minimal signs not diagnostic or suggestive of carcinoma.l' The tissue density pattern for each of the 23 mammographic studies was classified using Wolfe's patterns. A mammographic film was designated as displaying an N pattern if there was no ductal prominence, P 1 if prominent ducts were confined to the anterior quarter of the breast, P2 if there was extensive ductal prominence, and DY if there were areas of confluent density sufficient to obscure detail of the underlying parenchyma. Nine percent of the IBC films displayed an N pattern compared to 7.5% of films from breast cancer patients in general. 2 IBC films interpreted as showing a PI pattern accounted for 17% of the current studies, against an expected rate of 15%. The IBC films in this study had a slightly higher proportion exhibiting dysplastic patterns than those seen in the general population of cancer patients. Forty-four percent were classified as P2 and 30% as DY, compared to 57.5% and 20% in Wolfe's 1976 series. 2 Table I compares the tissue patterns found in this study to those of other published series. THE AMERICANJOURNALOF SURGERY® VOLUME169 DECEMBER1994 539 INTERVAL BREAST CANCERS/KOIVUNEN ET AL TABLE II Expression of Tumor Markers in Sections From Archival Specimens of 19 Patients With Interval Breast Cancer Monoclonal Percent Expected in (Antigen) No. Positive Percent Positive Breast Cancer Population" 11 58 9422 BRST-1 (T47-D) BRST-3 (TAG-72) 1 5 848 I 5 7013 BRST-4* (antigen) 0 0 502s BRST-5 (antigen in malignant epithelium) MAb-1t (c-erb B-2t) I 5 204512.21 p53 (protein present in all Ca) 1 5 255 CA 15.31 (epithelial glycoprotein) 12 63 67/4 Cathepsin-D~ (estrogen-related protein) 8 42 36-607 2 10 7415 BRST-2t (cystic fluid protein) 0 0 5016,17 CEA* (carcinoembryonic antigen) 0 0 25-1817 CA 19.9~ (carbohydrate antigen) "See references for cited sources. tAssociated with poor prognosis. ;Controls, Paraffin tissue blocks were available for 19 IBC patients. Qualitative marker expression for each of the 11 chosen markers is summarized in Table II, along with the percent expression seen in published studies of patients with non-IBC. For each of the tumor markers associated with poor prognosis, the IBC tumors stained positive in a smaller proportion than has been recorded in tumors from the generalized breast-cancer patient population. Mutant p53 and c-erb B-2 were both detected in only 5% of the IBC tissue specimens compared to 25% and 20% of tumors reported in the literature. 5.12 Cathepsin-D was identified in only 42% of this series' IBC tumors compared to 36% to 60% of specimens in series of non-IBC tumors. 7 BRST-4, which is also associated with a poor prognosis, was expressed in only 5% compared to 70% of non-IBC patients with a poor prognosis/3 Only the expression of CA 15.3 approached the percent seen in larger breast cancer populations. Sixty-three percent of IBC patients' tumors stained positive for CA 15.3 compared to 67% of tumors from other breast cancer patients) 4 The control markers BRST-2 (gross cystic fluid protein), CEA, and Ca 19.9 were identified in the IBC specimens 10%, 0%, and 0% of the time. This is less than reported by studies of non-IBC patients, t5-17 COMMENTS Interval breast cancers have accounted for 10% to 30% of breast cancers in various screening programs. 1.t0.t i They are thought to represent a significant portion of the falsenegative results experienced in these programs, detracting from what otherwise would be excellent survival statistics in patients whose cancers were diagnosed early. DeGroote et al TM found markedly decreased survival in patients with IBC compared to those whose malignancies were detected during scheduled screening exams. They also noted that a greater percentage of the IBC patients presented with stage II, or worse, breast cancers. In contrast, in the present series, the 10-year survival rate for patients with IBC was not significantly different from the rate calculated for patients whose cancers were detected at screening visits to the BCDDP. Seventy-nine percent of both groups survived 540 10 years. IBC patients with invasive cancers had a 10-year survival rate of 76%, the same as patients whose invasive cancers were detected at screening. ~° Although the numbers in the current study are much smaller than those in the national BCDDP, the 5- and 10-year survival rates of 100% and 85% compare favorably. In DeGroote's series of 21 IBC patients, 24% presented with stage II disease, and another 24% were classified as having stage III cancer. ~8 In the BCDDP, 33% of the patients with IBC and 31% of the screen-detected cancer cases had stage II disease, l° In our current series, 54% of our IBC patients presented with stage II disease, and none with more advanced cancer. The small numbers in the present series probably contribute to our greater proportion of IBC patients presenting with stage II disease. Nevertheless, the overall survival of these patients appears similar to that of patients in other, larger, published series whose cancers were detected by screening. If IBCs were indeed more aggressive than screen-detected breast tumors, a higher proportion of them should presumably express specific biologic tumor markers that are associated with a poor prognosis. Although no tumor marker to date has proven to be sufficiently reliable for use as a routine screening tool, a number of studies have identified associations between tumor expression of p53, cathepsin-D, and the c-erb B-2 oncogene and diminished survival. 4'5'7't2'19 In studies with archival specimens, increased expression of p53 and c-erb B-2 have both been shown to correlate with adverse lymph node status and shorter survival/°.21 In paraffin-embedded archival specimens, cathepsin-D expression detected using a monoclonal antibody has been found to be a powerful, unfavorable prognostic factor in node-negative breast cancer. 7 In the present study, we did not find the levels of expression of these tumor markers which would be expected on the assumption that IBCs are exceptionally aggressive cancers. The tumor marker c-erb B-2 was expressed in only 5%, p53 in 5%, and cathepsin-D in 42% of the IBC archival specimens available for testing. These percentages are considerably lower than those quoted in the literature (Table II) for breast cancer generally, suggesting THE AMERICANJOURNALOF SURGERY® VOLUME 169 DECEMBER1994 INTERVAL BREAST CANCERS/KOIVUNEN ET AlL that the expected biologic behavior of IBC tumors is not worse than that of non-interval cancers. Similarly, monoclonal antibodies against (1) a glycoprotein secreted by the T47-D ~-2breast carcinoma line, (2) TAG-72 (a tumorassociated oncofetal antigen), 8 (3) BRST-4 antigen (associated with poor prognosis), t3 and (4) BRST-5 antigen (from malignant epithelium) disclosed that IBC specimens expressed these markers at markedly lower rates than the literature would lead one to anticipate. Only the CA 15.3 antigen was expressed in the IBC specimens with a frequency approaching that seen in other breast cancer series. t4 In summary, the expected biologic behavior or aggressiveness of the IBC tumors in this series does not appear to differ from that seen in the broader breast cancer population. Since IBCs are, by definition, not diagnosed radiologically, it may be postulated that mammograms in this subset of the breast cancer population are inherently more difficult to interpret. Ikeda et all9 noted that between 52% and 59% of their IBC patients had mammograms that were classified as having P2 or DY Wolfe's characteristics. They also noted that a higher proportion of the interval cancers (15% compared to the expected 10%) were lobular cancers, and 38% were invasive comedo, medullary, or mucinous by histology. All four of these histologic types are known to be difficult to diagnose mammographically. A review by Gravelle et al23 of mammograms from patients who ultimately developed breast cancer demonstrated that 88% of these films were classified as either Wolfe pattern P2 or DY, supporting the original theory put forth by Wolfe that patients with more dysplastic patterns on mammography had higher risk of developing breast cancer. 2 A 1984 study by Verbeek et a124 failed to show as strong a prognostic value for Wolfe's classification as these other series. Seventy-four percent of the IBC mammograms in the present study were classified as P2 or DY. Thirty percent of the IBC films available for review were rated as DY, a considerably larger proportion than the 16.5% (Table I) seen in the population with non-IBC. Whether, or not, our patients were inherently at greater risk for developing cancer, 2.23 their films definitely displayed larger areas of confluent density and prominent ductal marking, rendering the studies much more difficult to interpret. In spite of this greater challenge, our review found that only one film had been initially misinterpreted, yielding a misdiagnosis rate of only 4%, which is fairly low when compared to earlier published series.l t In an attempt to ascertain whether the screening physical examination in IBC patients was especially difficult to •interpret, the clinical records were searched for notations indicating that the breasts were "nodular," "dense," or "diffusely fibrocystic" in quality. Nodularity, masses, or thickening were found at the visit prior to the diagnosis of IBC in 17 patients. When the 3 patients who presented with nipple discharge or an intracystic carcinoma are removed from consideration, 17 of 21 patients (81%) with IBC had breast examinations yielding findings which would make detection of small cancers difficult. To conclude, contrary to some earlier published series, we have demonstrated that IBCs are not exceptionally ag- gressive tumors with particularly poor prognoses. Rather, they tend to display biological behavior not unlike other more conventionally diagnosed breast malignancies. What appears to set IBC apart from the general population of breast cancers is a greater degree of difficulty in diagnosis, both by physical examination and mammography. Greater awareness of this challenge may lead to increased accuracy and timeliness of discovery. REFERENCES 1. Panoussopoulos D, Chang J, Humphrey LJ. Screening for breast cancer. Ann Surg. 1977;186:356-362. 2. Wolfe JN. Risk for cancer development determined by mammographic parenchymal pattern. Cancer. 1976;37:2486--2492. 3. Kohler MF, Kerns BM, Humphrey PA, et al. Mutation and overexpression of p53 in early stage epithelial ovarian cancer. Obstet Gynecol. 1993;81:643-650. 4. Clark GM, McGuire WL. Follow-up study of HER-2/neu amplification in primary breast cancer. Cancer Res. 1991;51:944-948. 5. Davidoff AM, Kerns BM, Iglehart JD, Marks JR. Maintenance of p53 alterations throughout breast cancer progression. Cancer Res. 1991;51:2605-2610. 6. Barak M, Steiner M, Finkel B, et al. CA 15,3, TPA and MCA as markers for breast cancer. Eur J Cancer. 1990;26:577-580. 7. Isola J, Weitz S, Visakorpi T, et al. Cathepsin D expression detected by immunohistochemistry has independent prognostic value in axillary node-negative breast cancer. J Clhl Oncol. 1993;11:36--43. 8. Johnson VG, Schlom J, Paterson AJ, et al. Analysis of a human tumor-associated glycoprotein (TAG-72) identified by monoclonal antibody B72.3. Cancel" Res. 1986;46:850-857. 9. Rodes ND, Lopez MJ, Pearson DK, et al. The impact of breast cancer screening on survival. Cancer. 1986;57:581-585. 10. Seidman H, Gelb SK0 Silverberg E, et al. Survival experience in the breast cancer demonstration project. CA CancerJ Clin. 1987; 37:258-290. I I. VanDijck IS, Verbeek ALM, Hendriks IHS, Holland R. The current detectability of breast cancer in a mammographic screening program. Cancer. 1993;72:1933-1938. 12. Treurniet NF, Rookus MA, Peterse HL, et al. Differences in breast cancer risk factors to neu (c-erbB-2) protein overexpression of the breast tumor. Cancer Res. 1992;52:2344-2345. 13. Mesa-Tejada R, Palakodety RB, Leon JA, et al. Immunocytochemical distribution of a breast carcinoma associated glycoprotein identified by monoclonal antibodies. Am J Pathol. 1988;130: 305-314. 14. Kufe D, Inghirami G, Abe M, et al. Differential reactivity of a novel monoclonal (DF3) with human malignant versus benign breast tumors. Hybridoma. 1984;3:223-232. 15. Wick MR, Lillemoe TJ, Copeland GT, et al. Gross cystic disease fluid protein-15 as a marker for breast cancer. Hum Pathol. 1989;20:281-287. 16. Yoshikawa T, Nishida K, Tangawa M, et al. Carbohydrate antigenic determinant (CA 19-9) and other tumor markers in gastrointestinal malignancies. Digestion. 1985;31:67-76. 17. Haglund C, Kuusela P, Roberts PJ. Tumor markers in pancreatic cancer. Ann Chir Gynaecol. 1989;78:41-53. 18. DeGroote R, Rush BF Jr, Milazzo J, et al. Interval breast cancer: a more aggressive subset of breast neoplasias. Surgery. 1983; 94:543-547. 19. Ikeda DM, Andersson I, Wattsgard C, et al. Interval carcinomas in the Malmo mammographic screening trial. A JR Am J Roentgenol. 1992;159:287-294. 20. Bosari S, Lee AKC, Viale G, et al. Abnormal p53 immunoreactivity and prognosis in node-negative breast carcinomas with long-term follow-up. Virchows Arch A Pathol Anat Histopathol. 1992"421:291-295. THE AMERICANJOURNAL OF SURGERY® VOLUME 169 DECEMBER 1994 541 INTERVAL BREAST CANCERS/KOIVUNEN ElF AL 21. Pavelic ZP, Pavelic L, Lower EE. et al. c-myc, c-erbB-2 and Ki-67 expression in normal breast tissue and in invasive and noninvasive breast carcinoma. Cancer Res. 1992;52:2597-2602. 22. Pancino G, Charpin C, Osinaga E, et al. Characterization and distribution in human tissues of a g[ycoprotein antigen defined by monoclonal antibody 1BE~_~raised against the human breast cancer cell line T47-D. Cancer Res. 1990:50:7333-7347. 23. Gravelle IH, Bulstrode JC, Bulbrook RD, et al. A prospective study of mammographic parenchymal patterns and risk of breast cancer. Br J Rad. 1986;59:487-491. 24. Verbeek ALM, Hendriks JHLL, Peeters PHM, Sturmans F. Mammographic breast pattern and the risk of breast cancer. Lancet. 1984:591-593. DISCUSSION James A. Edney, M D (Omaha, Nebraska): In large, population-based, randomized screening trails we have identified a subset of women who have had their cancer diagnosed after a negative mammogram, but prior to their next screening examination. Concern has always existed that these interval breast cancers represent tumors with a shorter doubling time, and, therefore, more aggressive metastatic potential. While a large number of screening trials have provided data concerning the incidence of interval breast cancers, there are two large randomized trials worth noting. In the Health Insurance Plan Study of New York, 31,000 women were randomized and screened, and a small percentage developed interval breast cancer during the 3 years of this study. Interestingly, these women had the same case fatality ratio as the control group. In a Swedish study, of the 465 breast cancers identified, 104 of them presented as interval breast cancers. Overall, it was found that the survival rate actually was higher in the interval cancer group than in the control cancer group. The results of the Health Insurance Plan Group and the Swedish randomized trials support the author's contention that the prognosis of interval cancers is the same as that of similarly staged cancers diagnosed in a nonscreening population. Overall, even with optimal screening programs, the incidence of interval breast cancers can be expected to be about 12%. The interval between the time of detection by scrcening mammography and the time when a cancer becomes appreciable is defined as the lead time. It has commonly been felt that the longer the lead time, the better the prognosis. It has become evident that the rapidity of growth of breast cancer is a more important prognostic indicator than is absolute size. The authors have submitted these 19 breast cancer specimens to a number of sophisticated, and elegant, monoclonal antibody studies using 6 breast cancer cell lines, as well as a number of other markers for breast cancer, including CA 15.3, c-erb B-2, and CEA. However, in reviewing the manuscript, I do not see where any specific kinetic studies on cell-doubling time were performed. Cell kinetic studies measuring tumor growth and biologic aggressiveness may, in fact, be the most important prognostic factor in breast cancer. Flow cytometry provides information regarding the percentage of cells in the S phase and 542 the assessment of tumor ploidy. It is very important to determining growth rate and subsequent prognosis. The absence of estrogen and progesterone receptors is associated with poorly differentiated tumors and has been clearly shown to portend a poor prognosis. In National Surgical Adjuvant Breast Project trials, tumor necrosis factor was also identified as an important factor that heralded a poor prognosis. Similarly, evidence of blood vessel, lymphatic, and perineural space invasion, as well as nuclear grade are important histologic findings. While most authorities would agree with this paper's basic premise that interval breast cancers do not carry a poorer prognosis than those patients who are not in a screening program, I would question the methods that were used to arrive at this conclusion. I would like to ask the authors four questions: Were any studies done to evaluate the rate of tumor growth? Specifically, do you have any information regarding DNA ploidy or the percentage of cells in the S phase of the DNA replication cycle? Were any assays performed to assess the hormone receptor status of these patients? Tumor necrosis factor is an important indicator of prognosis. Were there any assays done for this? Finally, do you have any information on the histologic grade of these tumors, or the presence of blood vessel, lymphatic or perineural space invasion? Debra Koivunen, MD: Thank you for your very pertinent questions. No, we did not perform flow cytometry on our archival specimens because we had very little tissue to work with and used most if not all to perform our immunohistologic studies. Flow cytometry is planned for future studies, in which we will be using fresh tumor straight out of the operating room. Nor did we carry out hormonal receptor studies on the limited tissue that was available to us from some of these very old tissue blocks. We do have the hormonal status on the more recent patients done, but the patients who underwent their mastectomies during the early 1970s did not have their hormonal status tested then, so we don't have any archival records to go back to, and we also didn't have enough tissue to do further studies. No, we did not look at tumor necrosis factor, but again, we agree that TNF is an excellent indicator to take into account; that is being planned into our studies that are ongoing, right now. We did look at some histopathologic risk factors in a few of our patients. We only had enough tissue material left, after doing all of our tissue block analyses on 10 archival specimens, to make additional slides; we did not get the original slides. We are, however, in the process of asking for some of those to be shipped to us so that we can study the remaining patients' tumors with respect to these risk factors. This slide shows you how the histologic grades, nuclear grades, vascular invasion, and lymphatic invasion were distributed, particularly with respect to cathepsin D and CA 15.3 expression in the 10 patients we were able to evaluate so far. We found that in stage II node-positive patients there was a higher percentage of lymphatic invasion, as well as nuclear grades II and III, and histologic grades II and III, than seen in our stage I patients. THE AMERICANJOURNALOF SURGERY® VOLUME 169 DECEMBER 1994