Identification of a Unique Biological Tumor Marker in Human Breast Cyst Fluid and Breast Cancer Tissue David Tapper, MD, Corinne Gajdusek, PhD, Roger Moe, MD, Janice Ness, PhD, Seattle, Washington
and cure breast cancer is while it is still confined to the breast. Clearly, a sensitive, reliable, and practical method of identifying this neoplastic process in its earliest stages is needed. The role of growth factors in the process of neoplasia is rapidly being elucidated and provides an alternative to traditional approaches for the development of an early detection assay for breast cancer. The concept that growth factors, which play an essential part in the normal differentiation of cells, may play a fundamental part in tumorigenesis has been strikingly reinforced by the homology between the nucleic acid sequences of oncogenes and their normal counterparts, the protooncogenes [4,6]. These findings suggest that growth factors and their associated receptors may be critical in the regulation of both normal and abnormal cellular proliferation and may serve as useful markers for women at increased risk for breast cancer. These markers may be useful either alone or with other indicators in a diagnostic screening program or in monitoring women with resected breast cancer for relapse. Previously, we have identified biological markers in body fluids that have been useful for detection of malignancy [7-9]. We have shown that increased capillary endothelial cell migration activity stimulated by factors in aqueous humor, urine, and cerebrospinal fluid was closely associated with ocular, bladder, and central nervous system malignancies, respectively [9-11]. In addition, specific growth factors capable of stimulating DNA reast cancer, one of the most frequent types of cancer synthesis and cell division in cultured cells have been in women in Western societies and in the United found in human, bovine, sheep, and murine milk [12-20]. States, has shown an alarming increase in incidence from We have identified several growth factors in murine milk 87 of 100,000 women to 104 of 100,000 women within the that are unique in that they identify mice that develop last 5 years [1,2]. It is commonly accepted that early mammary tumors. The specific mitogenic activity, absent diagnosis and treatment of breast cancer results in a more in control mice, is present early in the course of the favorable prognosis; however, despite the increased use of disease when there is no palpable tumor and is predictive mammography and breast self-examination, mortality of tumor development. Biochemical characterization of from breast cancer has remained constant [3]. Clinically, the predictive activity demonstrates that it is composed of breast cancer alternates between periods of dormancy a unique set of small peptide growth factors (J Surg Res, and rapid growth, and tumors confined to the breast have in press). In the preliminary study reported herein, breast cyst longer doubling times than tumors with axillary or systemic metastases [3]. Thus, the optimal time to detect fluid was obtained from 35 women with differing risks of developing breast cancer. Evaluation of the breast cyst From the DepartmentsOfSurgeryand Pathology,Children'sHospital fluid for mitogenic activity demonstrated a stratification & MedicalCenter,and the Universityof Washington,Schoolof Medicine,Seattle;Washington.Supportedin part by grant CA 40423 from of mitogenic activity, with samples from the highest-risk group containing a several-fold increase in activity comthe NationalCancerInstitute,Bethesda,Maryland. Requests for reprints shouldbe addressedto DavidTapper, MD, pared with samples obtained from the lowest-risk women. Department of Surgery,Children'sHospital & MedicalCenter,4800 Further analysis by high-pressure liquid chromatogSand PointWay Noi-theast,Seattle,Washington98105. raphy of breast cyst fluid samples representative of the Presentedat the 76th AnnualMeetingof the North PacificSurgidiffering risk groups demonstrated the presence of a specal Association,VictOria,BritishColumbia,Canada,November10-11, cific mitogenic activity that was directly related to the 1989.
Breast cyst fluid from 35 women was Stratified into risk groups based on personal and family history of breast cancer. Mitogenic activity in breast cyst fluid of women at highest risk to develop breast cancer was significantly higher than the activity in the lowest-risk group. There was a direct dose-dependent relationship between mitogenic activity and increased risk of developing breast cancer. Size-exclusion chromatography showed that breast cyst fluid from women at highest risk contained two peaks of growth factor activity: less than 6 kilodaltons (kd), identified as human EGF (epidermal growth factor), and 6 to 18 kd. Moderate,risk group samples demonstrated only the single less than 6 kd peak, whereas the lowest-risk group had insignificant growth-promoting activity. Breast cancer tissue analyzed in a similar manner revealed a predominant 6- to 14-kd peak of mitogenic activity demonstrating the same acid- and heat-stability found in breast cyst fluid.
B
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990 4"73
TAPPER ET AL
degree of risk. In addition, similar specific mitogenic activity was detected in breast cancer tissue obtained at surgical resection. The activities characterized from breast cyst fluid and breast cancer tissue were similar to the unique set of growth factors previously isolated in our laboratory from murine milk, which identifies and predicts mammary tumor development. MATERIAL AND METHODS Patient population: Breast cyst fluid was obtained by one of us (RM) from patients seen in the Breast Clinic at the University of Washington. Patients were stratified into groups according to their risk of developirlg breast cancer, based on personal and family history of breast cancer: Group 1, highest-risk group: previous history of breast cancer, newly diagnosed breast cancer and associated cyst(s), or a family history of more than one blood relation diagnosed with breast cancer; Group 2, moderate-risk group: an increased risk of developing breast cancer based on a biopsy showing atypia or hyperplasia, or a suspicious mammogram and/or family history of one blood relation diagnosed with breast cancer; Group 3, lowest-risk group: no family or personal history of breast cancer, expected incidence of breast cancer the same as the general population (i.e.; 104 of 100,000 women). Collection and treatment of samples: Breast cysts were aspirated, the specimen was collected (2 to 10 mL), and that portion of the sample not required for routine testing was sent on ice to Children's Hospital & Medical Center. All breast cyst fluid samples were obtained by a single investigator (RM) and coded prior to transport. The persons performing the assays were blinded to the code and did not know the clinical history associated with each sample. Samples were centrifuged to remove debris, and supernatants were aliquoted and stored at -800C until tested. Human breast cancer was obtained from women undergoing surgical removal of suitably sized lesions. Tissue was collected under sterile conditions in the operating room, immediately frozen, and stored at -80~ until tested. Additional material was provided by the Cooperative Human Tissue Network. These tumors were obtained at autopsy within 6 hours of death, immediately frozen, and shipped to our laboratory on dry ice. An appropriate clinical history and pathologic diagnosis were sent with each sample. Determination of mitogenie activity: Frozen aliquots of breast cyst fluid samples were thawed and assayed for mitogenic activity. The mitogenic activity in the breast cyst fluid samples was determined by an in vitro cell proliferation assay measured by 3H-thymidine incorporation. NIH 3T3 mouse embryo fibroblasts (clone A31), obtained from the American Type Culture Collection, were routinely grown in low-glucose (4.5 g/L) Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 0.292 mg/mL glutamine, 100 U/mL penicillin, and I00 mg/mL streptomycin. 3T3 cells were grown to confluency in 2 to 3 days in multiwell plates, then incubated for 18 hours in 1% plasma-derived serum with either the unknown or fetal calf serum standards. At 474
18 hours, 0.25 ~tC 3H-thymidine was added per well, and 3H-thymidine incorporation was determined at 20 hours. Cell monolayers were washed twice in 10% trichloroacetic acid for 10 minutes at 4~ and the cells were solubilized in 0.2 mL 0.5N sodium hydroxide. The total sample was counted in 5 mL scintillation fluid on a scintillation counter. Mitogenic activity was compared with the activity of standard mitogen units to determine the quantity of growth factor activity in each sample. One activity unit was defined as the amount of 3H-thymidine incorportion stimulated by 1 ng platelet-derived growth factor (PDGF) and was equivalent to 5 uL fetal calf serum standard. This allowed for comparison between samples in independent assays. Breast cyst fluid samples were assayed in triplicate at several concentrations, and the mean and standard deviation were determined for each point. Column chromatography: Breast cyst fluid samples were thawed, centrifuged, and analyzed by size-exclusion high-pressure liquid chromatography (HPLC). Two hundred microliters of sample supernatant was injected directly onto a Waters Protein Pak 300 SW column (7.8 mm • 30 cm) previously equilibrated in phosphate buffer (5.6 mM sodium phosphate/ 1.06 mM potassium phosphate, pH 7.5/0.15M sodium chloride) at room temperature. The column eluant was collected in l-mL fractions at a flow rate of 0.5 mL/minute. Fractions were directly assayed in the 3H-thymidine incorporation assay. The breast cancer tissue was thawed, and the major blood vessels were carefully dissected away from the tumor tissue and discarded. Approximately 130 g of frozen pooled breast tumors was acid/ethanol-extracted by the method of Roberts [21]. The lyophilized sample extract was solubilized in 6M guanidine-hydrochloric acid/0.1M potassium phosphate, pH 4.5, and fractionated on a Sephacryl S-200 (Pharmacia 3 cm • 77 cm) size-exclusion column equilibrated in the same buffer. The column was run at a flow rate of 0.25 mL/minute. Fractions of 4 mL were collected, dialyzed, lyophilized, and assayed for mitogenic activity. The molecular weight of the active fractions in all chromatography procedures was estimated by comparison to the elution profile of standard proteins. Radioassay: Growth factor competing activity was determined in fractions chromatographed on the Waters Protein Pak 300 SW HPLC size-exclusion column. Lyophilized samples were reconstituted inappropriate buffer and assayed in duplicate. In the identical fractions, combined human epidermal growth factor (EGF) and human transforming growth factor alpha (TGF-alpha) competing activity was determined by radioassay kit (Biomedical Technologies, Stoughton, MA) following the manufacturer's protocol. Stability tests: Stability tests were performed on fractions containing mitogenic activity that had been chromatographed on the Waters Protein Pak 300 SW HPLC size-exclusion column for breast cyst fluid samples or on the Sephacryl S-200 column for breast cancer tissue. Aliquots from fractions containing the mitogenic peak of interest were dialyzed and lyophilized. Acid-treated sam-
THE AMERICANJOURNALOFSURGERY VOLUME159 MAY 1990
TUMOR MARKERS IN BREAST CYST FLUID
ples were reconstituted in 0.5N acetic acid, incubated for 1 hour, then lyophilized and stored at -80~ Heattreated samples were reconstituted in 1 mM ammonium bicarbonate, heated at 80~ for 5 minutes, cooled, and directly assayed. Statistical analysis: Statistical significance of the difference in mitogenic activity between risk groups was determined using a variety of standard statistical methods. The means and standard deviation were calculated for each concentration of breast cyst fluid. The values for mitogenic activity determinations were analyzed by analysis of variance (ANOVA) to determine variance from the mean. The variation in mitogenic activity values across risk groups was analyzed at each breast cyst level using the Kruskal-Wallis method. A combination of these statistical methods was used to determine the relevance of the data. RESULTS Breast cyst fluid samples obtained from 35 women were examined for the presence of mitogenic activity. Women from whom samples were collected were categorized according to their assessed risk of developing breast cancer based on personal and family history of breast cancer. The highest-risk women, Group 1 (n = 10), were either newly diagnosed with breast cancer and associated cyst(s), previously diagnosed with breast cancer, or had two or more blood relations diagnosed with breast cancer. The moderate-risk women, Group 2 (n = 10), had either a suspicious mammogram, a breast biopsy showing atypia or hyperplasia, or one blood relation diagnosed with breast cancer. The lowest-risk women, Group 3 (n = 15), were women with no personal or family history of breast cancer. Their expected risk of breast cancer was considered to be the same as that of the general population (104 of 100,000 women). Breast cyst fluid samples were obtained from women in all three categories and coded to prevent the persons performing the biochemical assays from knowing the associated clinical history. Mitogenic activity of the coded breast cyst fluid sampies was determined at several concentrations in an in vitro proliferation assay by incorporation of 3H-thymidine into mouse fibroblast (NIH 3T3) cells. Individual dose-response curves were determined for each patient. When the code was unmasked, the breast cyst fluid sampies demonstrated a direct dose-response relationship between mitogenic activity and assessed risk of developing breast cancer, with the level of mitogenic activity being highest in women at greatest risk of developing breast cancer. The dose-response curves for three individual patients, representative of the different risk groups, are shown in Figure 1. The Group 1 sample, highest risk, was obtained from a patient who was diagnosed with extensive noninvasive ductal carcinoma concurrent with the cyst aspiration. A subsequent biopsy of the cyst cavity from which the aspirate was obtained showed a smoothly lined cavity with benign changes. The Group 2 sample, moderate risk, had an intermediate level of activity and was obtained from a patient whose mother had breast cancer. The Group 3 sample, lowest risk, had the lowest
25
~
-,
Q
20
10
0 25
0
5
75
~I B r e a s t
10
15
Cyst Fluid
Figure 1. Mitogenic activity dose-response curves for breast cyst fluid samples from individual patients with differing risks of developing breast cancer.
25
% i,q N
20
I
.~ 15
:
/~"~"
.
" ~
5
~.
;
_ . - 4 - . . . . ~" . . . . . .
/,f 9
9
,
D, C
~ la
9~ ~ 9 9 "
ModerateRisk L~ 1
0 0
25
5
75
10
15
gl Breast Cyst Fluid
Figure 2. Summary mitogenic activity dose-response curves for the high-, moderate-, and low-risk groups.
amount of mitogenic activity and was obtained from a woman with no family history of breast cancer who presented with a single unilocular breast cyst. There was a statistically significant difference in mean mitogenic activity values between the high-, moderate-, and low-risk groups. The summary dose-response curves were determined by combining the individual dose-response curves for all samples within a risk group (Figure 2). The effect of risk group classification was almost entirely on the mean level. We determined that the mean level across all breast cyst fluid levels per patient varied significantly by risk group (p --- 0.014 by ANOVA). The variation in mean levels across risk groups was statistically significant, with a p value of 0.012 at the 2.5-gL concentration of breast cyst fluid to a p value of 0.019 at 10 #L as determined by ANOVA. At 15/zL of breast cyst fluid, the variation in mean values became marginally significant (p -- 0.053). We also analyzed the variation in values of means across groups at each breast cyst fluid level using the nonparametric Kruskal-Wallis method. The results confirmed those obtained by ANOVA. Next, several representative samples were chosen
THE AMERICAN JOURNAL O F S U R G E R Y
V O L U M E 159
MAY 1990
475
TAPPER ET AL
A. ~
P.Jsk Group
B. Moderate Risk Group
C. Low Risk Group
I |
-
"I.
I'i!
j" i
o
J.o ]Prtt9
.
'~0
.
.
.
.
20
Nmmber
Figure 3. Distribution of growth factor activities in individual breast cyst fluid samples fractionated by size-exclusion HPLC. kDa = kilodalton.
from each risk group, and the breast cyst fluid samples were analyzed by sizing column chromatography. Individual breast cyst fluid samples were centrifuged and multiple aliquots of the supernatants were fractionated by HPLC using a Waters Protein Pak 300 SW sizeexclusion column equilibrated in a neutral phosphate buffer as described previously. Column eluant was collected in separate fractions, and each fraction was tested for mitogenic activity in the in vitro proliferation assay. The protein profiles of all the samples were virtually identical. However, the distribution of growth factor activity in the fractionated breast cyst fluid differed between risk groups. Figure 3 shows examples of the distribution of growth factor activity. Samples from the highest-risk group had two peaks of mitogenic activity: one with an apparent molecular weight less than 6 kilodaltons (kd) and a second peak at 6 to 18 kd (Figure 3). Samples from the moderate-risk group demonstrated only a single peak of mitogenic activity at less than 6 kd, and the lowest-risk group had insignificant growth-promoting activity (no distinct peak) in either molecular weight range (Figure 3). The 6- to 18-kd peak of mitogenic activity was both acid- and heat-stable. The less than 6-kd mitogenic peak in the Group 1 and 2 samples is presumed to be human EGF as identified by radioimmunoassay. The presence of other growth factors in this peak has not been excluded. In addition to the breast cyst fluid samples, breast cancer tissue was investigated for the presence of similar growth factor activities. Surgical samples of breast cancer tissue were obtained from women at either the primary operation to determine the presence of breast malignancy, the definitive operation for treatment and staging of the breast disease, or autopsy. Breast cancer tissue of the same pathologic type was pooled from several donors and extracted by an acid-ethanol procedure [21]. The extract was fractionated by HPLC using a sizeexclusion column equilibrated in 6M guanidine-HC1, pH 4.5, as described. Column eluant was collected in frac476
THE AMERICAN JOURNAL OF SURGERY
tions, and each fraction was tested for mitogenic activity. As seen in the breast cyst fluid samples from high-risk patients, breast cancer tissue contained several peaks of mitogenic activity, including a predominant peak of mitogenic activity 6 to 14 kd in molecular weight (Figure 4). This peak of activity in breast cancer tissue also demonstrated acid- and heat-stability as observed for breast cyst fluid. COMMENTS These studies have shown that breast cyst fluid from certain women contains an increased level of growth factor activity that has the potential to be a useful biological marker for breast cancer. The data clearly show a direct dose-dependent relationship between mitogenic activity and an increased risk of developing breast cancer. The mitogenic activity in the samples from the high-risk group was several-fold greater than the activity present in the samples from the low-risk group. When analyzed by size-exclusion chromatography, samples from the highrisk group contained a 6- to 18-kd acid- and heat-stable mitogenic component and a less than 6-kd component subsequently identified as human EGF. Samples from the moderate-risk group possessed only the less than 6-kd peak, and the lowest-risk group samples had no significant growth factor activity. In addition, breast cancer tissue was also analyzed by size-exclusion column chromatography and shown to contain a 6- to 14-kd acid- and heat-stable mitogenic component similar to that found in breast cyst fluid. As a result of physician and public awareness, more women are now diagnosed with what has been termed minimal breast disease. However, despite all efforts toward earlier diagnosis, the mortality rate from breast cancer has remained constant [3]. Several modalities, including mammography, self-examination, aspiration cytology, and biological tumor markers, have been recommended as possible methods of earlier and more sensitive detection of breast cancer [22-24].
VOLUME 159
MAY 1990
TUMORMARKERSIN BREASTCYSTFLUID
In the search for biological tumor markers for breast cancer, a variety of approaches have been pursued, including analysis of metabolic by-products, oncofetal antigens, viral particles or related macromolecules, plasma proteins, hormones, and cell-associated antigens of breast cancer tissue. Although the study of tumor markers has advanced rapidly, and several biochemical markers have been identified that are elevated in patients with metastatic breast disease [25], at present no specific biological tumor marker exists for early diagnosis of breast cancer. Current evidence indicates that peptide growth factors such as EGF, PDGF, TGF-alpha and TGF-beta, and the insulin-like growth factors (IGFs) may be pivotal in the progressive development of malignantly transformed cells [4-6]. Breast cancer cells in vitro secrete several growth factors including IGF-1, TGF-alpha and TGFbeta, and a PDGF-Iike factor. Many of these growth factors are inducible by estradiol [26,27]. In particular, IGF-1 has been shown to promote breast cancer cell growth in vitro [28] and to be present in a higher concentration in primary breast cancer tissue than in adjacent normal tissue [29]. Thus, the aberrant production of growth factors may be associated with the neoplastic growth of breast cancer cells. We have previously identified a specific mitogenic activity in the milk of a substrain of mice that develop mammary tumors. This mitogenic activity is unique in that its presence in milk is predictive of mammary tumor development long before the presence of a palpable tumor (J Surg Res, in press). This predictive activity is composed of a subset of acid- and heat-stable peptide growth factors in the 6- to 10-kd molecular weight range. Of known growth factors tested, radioassay techniques identify an IGF-l-like peptide as the major component. The predictive growth factor activity also contained small amounts of PDGF- and TGF-beta-like compounds. PDGF, TGF-beta, and IGF-1, in combination, demonstrate potentiated mitogenic activity [30,31] and may be acting in a similar fashion in murine mammary tumor development. These growth factors are present only in mice that eventually develop mammary tumors and are not present in the milk of control mice. In the present study, the mitogenic activity in breast cyst fluid was directly related to women's increased risk of developing breast cancer. The breast cyst fluid from women in the high-risk group contained a 6- to 14-kd acid- and heat-stable growth factor activity similar to the predictive growth factor activity seen in the milk of mice that subsequently developed mammary tumors. Breast cyst fluid from women in the low-risk group did not contain this activity. In view of the results observed in murine milk, we believe that the presence of a specific mitogenic activity in breast cyst fluid may be associated with the development of breast cancer. The identification of a specific growth factor in breast cyst fluid and breast cancer tissue may be important not only as a predictor of malignancy and as a biological tool for early diagnosis, but also as an indication of effective therapy and a possible guide to adjuvant therapy. As with other biological markers (i.e., alpha-fetoprotein and hu-
60
%
r
ii
el
el
a Jr
a Jr
D Jr
0 Jr
o
~,
r
$
,I,
$
,I,
50
X
==
30 20 10 0 30
40 50 60
70
Fraction number Figure 4. Distribution of mitogenic activity in breast cancer tissue fracUonated by size-exclusion column chromatography, kDa = kilodalton.
man chorionic gonadotrophin) that decrease with successful treatment, the level of specific growth factor activity found in the breast cyst fluid of a woman who had previous breast cancer surgery and adjuvant chemotherapy was quite low. The disappearance of the marker may reflect successful treatment. The presence of an elevated marker level in the breast cyst fluid of a woman previously treated for breast cancer might herald the presence of recurrent disease. Additionally, the presence of this specific marker might be correlated with other growth factors and receptors to further stratify women into risk groups for the purpose of modifying their adjuvant therapy regimen. Estrogen and progesterone receptors have been well characterized with regard to their relationship to prognosis and plan for adjuvant therapy. Recently, EGF receptors have been identified in breast cancer tissue and metastatic lymph nodes removed at surgery. Those patients with EGF-positive receptors had a poorer prognosis [32,33]. There undoubtedly exists a complex interplay between growth factors, hormone receptors, and other oncogenes. The further identification of these
THE AMERICANJOURNALOFSURGERY VOLUME159 MAY 1990 477
TAPPER ET AL
growth factors and the role they play in breast cancer may improve our ability to diagnose these lesions earlier and to treat them more effectively. We wish to emphasize that this was a preliminary study of small sample size, prompted by the results of our investigation in an animal model and the availability of a large patient population of women with benign and malignant breast disease. This study is currently being expanded to include a large, statistically significant sample size, a more accurate assessment of risk by assignment of a computer-generated risk score based upon the breast cancer risk factors presented by Kelsey and Berkowitz [34], and additional biochemical analysis to specifically identify the growth factors present in the 6- to 18-kd mitogenic peak found in the high-risk samples. Our longrange goal remains to develop a sensitive, specific assay of body fluids for early detection of breast cancer and effective monitoring of treatment and probability of recurrence. REFERENCES 1. Young TL, ed. SEER Program. Incidence and mortality, 19731977. Bethesda, MD: National Cancer Institute, 1981. Monograph 57, NIH publication 81-2330. 2. Annual cancer statistics review. Bethesda, MD: National Cancer Institute, January 1988. 3. Kusantia S, Spratt TS, Donovan WL. The gross rates of growth of human mammary carcinoma. Cancer 1972; 30: 594-9. 4. Chan VW, McGee JO'D. Cellular oncegenes in neoplasia. J Clin Pathol 1987; 40: 1055-63. 5. Bishop JM. Oncogenes and proto-oncogenes. J Cell Physiol [suppl] 1986; 4: 1-5. 6. Heldin CH, Westermark B. Growth factors: mechanism of action and relation to oncogenes. Cell 1984; 37: 9-20. 7. Patz J, Tapper D, Outzen H, Klagsburn M, Sing Y. Determination of breast cancer in mice by the identification of growth factor activity in breast milk. Surg Forum 1983; 34: 432-5. 8. Tapper D, Langer R, Bellows AR, Folkman J. Angiogenesis capacity as a diagnostic marker for human eye tumorsl Surgery 1979; 86: 36-40. 9. Tapper D, Albert DM, Robinson NL, Zetter BR. Capillary endothelial cell migration: stimulating activity of aqueous humor from patients with ocular cancers. J Natl Cancer Inst 1983; 71: 501-5. 10. Chodak G, Scheiver C, Zetter B. Urine from patients with transitional-cell carcinoma stimulates migration of capillary endothelial cells. N Engl J Med 1981; 305: 869-74. 11. Brem H, Patz J, Tapper D. Detection of human central nervous system tumors: use of migration stimulating activity of the cerebrospinal fluid. Surg Forum 1983; 34: 532-4. 12. Connolly JM, Rose DP, Epidermal growth factor-like proteins in breast fluid and human milk. Life Sci 1988; 42: 1751-6. 13. Cera K, Mahan DC, Simmen FA. In vitro growth-promoting activity of porcine mammary secretions: initial characterization and relationship to known peptide growth factors. J Anim SCI 1987; 65:1149-59. 14. Zwiebel JA, Bano M, Nexo E, Salomon DS, Kidwell WR. Partial purification of transforming growth factors from human
478
milk. Cancer Res 1986; 46: 933-9. 15. Bano M, Salomon DS, Kidwell WR. Purification of a mammary-derived growth factor from human milk and human mammary tumors. J Biol Chem 1985; 260: 5745-52. 16. Grueters A, Larshmanan J, Tarris R, Aim J, Fisher DA. Nerve growth factor in mouse milk during early lactation: lack of dependency on submandibular salivary glands. Pediatr Res 1985; 19: 934-7. 17. Grueters A, Aim J, Lakshmanan J, Fisher DA. Epidermal growth factor in mouse milk during early lactation: lack of dependency on submandibular glands. Pediatr Res 1985; 19: 853-6. 18. Petrides PE, Hosang M, Shooter E, Esch FS, Bohlen P. Isolation and characterization of epidermal growth factor from human milk. FEBS Lett 1985; 187: 89-95. 19. Shing Y, Davidson S, Klagsbrun M. Purification of polypeptide growth factors from milk. Methods Enzymol 1987; 146: 42-8. 20. Noda K, Umeda M, Ono T. Transforming growth factor activity in human eolostrum. Gann 1984; 75: 109-12. 21. Roberts AB, Lamb LC, Newton DL, Sporn MB, DeLarco JE, Todaro GJ. Proc Natl Acad Sci USA 1980; 77: 3494-8. 22. Powell PW, McSwceney MD, Wilson CE. X-ray calcifications as the only basis for breast biopsy. Ann Surg 1983; 197: 555-9. 23. Robbins GF, Brothers TH, Eberhart WF. Is aspiration biopsy of breast cancer dangerous to the patient? Cancer 1954; 7: 774-8. 24. Strax P. Evaluation of screening programs for the early diagnosis of breast cancer. Surg Clin North Am 1978; 58: 667-79. 25. Twardzik DR, Sberwin SA, Ranchalis J, Todaro GJ. Transforming growth factors in the urine of normal, pregnant, and tumor-beating humans. J Natl Cancer Inst 1982; 69: 793-8. 26. Lippman ME, Dickson RB, Bates S, et al. Autocrine and paracrine growth regulation of human breast cancer. Breast Cancer Res Treat 1986; 7: 59-70. 27. Kasid A, Lippman ME. Estrogen and oncogene mediated growth regulation of human breast cancer cells. J Steroid Biochem 1987; 27: 465-70. 28. Dickson RB, Huff KI, Spencer EM, Lippman ME. Induction of epidermal growth factor-related polypeptides by 17B-estradiol in MCF-7 human breast cancer cells. Endocrinology 1985; 118: 13842. 29. Pekonen F, Partanen S, Makinen T, Rutanen EM. Receptors for epidermal growth factor and insulin-like growth factor-I and their relation to steroid receptors in human breast cancer. Cancer Res 1988; 48: 1343-7. 30. Pledger W J, Stiles CD, Antoniades HN, Scher CD. Induction of DNA synthesis in BALBc3T3 cells by serum components; reevaluation of the commitment process. Proc Natl Acad Sci USA 1977; 74: 4481-5. 31. Liboi E, Pelosi E, DiFrancesco P, et al. The El2 rat fibroblasts line: differential effects of growth factors (EGF, PDGF, FGF, TPA, and TGFB) on cell proliferation and c-los expression. Ann NY Acad Sci 1987; 511: 318-28. 32. Sainsbury JRC, Malcolm A J, Appleton, Farndon JR, Harris AL. Presence of epidermal growth factor receptor as an indicator of poor prognosis in patients with breast cancer. J Clin Pathol 1985; 38: 1225-8. 33. Rios MA, Macias A, Perez R, Lage A, Skoog L. Receptors for epidermal growth factor and estrogen as predictors of relapse in patients with mammary carcinoma. Anticancer Res 1988; 8: 1736. 34. Kelsey JL, Berkowitz GS. Breast cancer epidemiology. Cancer Res 1988; 48" 5615-23.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990
and cure breast cancer is while it is still confined to the breast. Clearly, a sensitive, reliable, and practical method of identifying this neoplastic process in its earliest stages is needed. The role of growth factors in the process of neoplasia is rapidly being elucidated and provides an alternative to traditional approaches for the development of an early detection assay for breast cancer. The concept that growth factors, which play an essential part in the normal differentiation of cells, may play a fundamental part in tumorigenesis has been strikingly reinforced by the homology between the nucleic acid sequences of oncogenes and their normal counterparts, the protooncogenes [4,6]. These findings suggest that growth factors and their associated receptors may be critical in the regulation of both normal and abnormal cellular proliferation and may serve as useful markers for women at increased risk for breast cancer. These markers may be useful either alone or with other indicators in a diagnostic screening program or in monitoring women with resected breast cancer for relapse. Previously, we have identified biological markers in body fluids that have been useful for detection of malignancy [7-9]. We have shown that increased capillary endothelial cell migration activity stimulated by factors in aqueous humor, urine, and cerebrospinal fluid was closely associated with ocular, bladder, and central nervous system malignancies, respectively [9-11]. In addition, specific growth factors capable of stimulating DNA reast cancer, one of the most frequent types of cancer synthesis and cell division in cultured cells have been in women in Western societies and in the United found in human, bovine, sheep, and murine milk [12-20]. States, has shown an alarming increase in incidence from We have identified several growth factors in murine milk 87 of 100,000 women to 104 of 100,000 women within the that are unique in that they identify mice that develop last 5 years [1,2]. It is commonly accepted that early mammary tumors. The specific mitogenic activity, absent diagnosis and treatment of breast cancer results in a more in control mice, is present early in the course of the favorable prognosis; however, despite the increased use of disease when there is no palpable tumor and is predictive mammography and breast self-examination, mortality of tumor development. Biochemical characterization of from breast cancer has remained constant [3]. Clinically, the predictive activity demonstrates that it is composed of breast cancer alternates between periods of dormancy a unique set of small peptide growth factors (J Surg Res, and rapid growth, and tumors confined to the breast have in press). In the preliminary study reported herein, breast cyst longer doubling times than tumors with axillary or systemic metastases [3]. Thus, the optimal time to detect fluid was obtained from 35 women with differing risks of developing breast cancer. Evaluation of the breast cyst From the DepartmentsOfSurgeryand Pathology,Children'sHospital fluid for mitogenic activity demonstrated a stratification & MedicalCenter,and the Universityof Washington,Schoolof Medicine,Seattle;Washington.Supportedin part by grant CA 40423 from of mitogenic activity, with samples from the highest-risk group containing a several-fold increase in activity comthe NationalCancerInstitute,Bethesda,Maryland. Requests for reprints shouldbe addressedto DavidTapper, MD, pared with samples obtained from the lowest-risk women. Department of Surgery,Children'sHospital & MedicalCenter,4800 Further analysis by high-pressure liquid chromatogSand PointWay Noi-theast,Seattle,Washington98105. raphy of breast cyst fluid samples representative of the Presentedat the 76th AnnualMeetingof the North PacificSurgidiffering risk groups demonstrated the presence of a specal Association,VictOria,BritishColumbia,Canada,November10-11, cific mitogenic activity that was directly related to the 1989.
Breast cyst fluid from 35 women was Stratified into risk groups based on personal and family history of breast cancer. Mitogenic activity in breast cyst fluid of women at highest risk to develop breast cancer was significantly higher than the activity in the lowest-risk group. There was a direct dose-dependent relationship between mitogenic activity and increased risk of developing breast cancer. Size-exclusion chromatography showed that breast cyst fluid from women at highest risk contained two peaks of growth factor activity: less than 6 kilodaltons (kd), identified as human EGF (epidermal growth factor), and 6 to 18 kd. Moderate,risk group samples demonstrated only the single less than 6 kd peak, whereas the lowest-risk group had insignificant growth-promoting activity. Breast cancer tissue analyzed in a similar manner revealed a predominant 6- to 14-kd peak of mitogenic activity demonstrating the same acid- and heat-stability found in breast cyst fluid.
B
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990 4"73
TAPPER ET AL
degree of risk. In addition, similar specific mitogenic activity was detected in breast cancer tissue obtained at surgical resection. The activities characterized from breast cyst fluid and breast cancer tissue were similar to the unique set of growth factors previously isolated in our laboratory from murine milk, which identifies and predicts mammary tumor development. MATERIAL AND METHODS Patient population: Breast cyst fluid was obtained by one of us (RM) from patients seen in the Breast Clinic at the University of Washington. Patients were stratified into groups according to their risk of developirlg breast cancer, based on personal and family history of breast cancer: Group 1, highest-risk group: previous history of breast cancer, newly diagnosed breast cancer and associated cyst(s), or a family history of more than one blood relation diagnosed with breast cancer; Group 2, moderate-risk group: an increased risk of developing breast cancer based on a biopsy showing atypia or hyperplasia, or a suspicious mammogram and/or family history of one blood relation diagnosed with breast cancer; Group 3, lowest-risk group: no family or personal history of breast cancer, expected incidence of breast cancer the same as the general population (i.e.; 104 of 100,000 women). Collection and treatment of samples: Breast cysts were aspirated, the specimen was collected (2 to 10 mL), and that portion of the sample not required for routine testing was sent on ice to Children's Hospital & Medical Center. All breast cyst fluid samples were obtained by a single investigator (RM) and coded prior to transport. The persons performing the assays were blinded to the code and did not know the clinical history associated with each sample. Samples were centrifuged to remove debris, and supernatants were aliquoted and stored at -800C until tested. Human breast cancer was obtained from women undergoing surgical removal of suitably sized lesions. Tissue was collected under sterile conditions in the operating room, immediately frozen, and stored at -80~ until tested. Additional material was provided by the Cooperative Human Tissue Network. These tumors were obtained at autopsy within 6 hours of death, immediately frozen, and shipped to our laboratory on dry ice. An appropriate clinical history and pathologic diagnosis were sent with each sample. Determination of mitogenie activity: Frozen aliquots of breast cyst fluid samples were thawed and assayed for mitogenic activity. The mitogenic activity in the breast cyst fluid samples was determined by an in vitro cell proliferation assay measured by 3H-thymidine incorporation. NIH 3T3 mouse embryo fibroblasts (clone A31), obtained from the American Type Culture Collection, were routinely grown in low-glucose (4.5 g/L) Dulbecco's modified Eagle's medium supplemented with 10% calf serum, 0.292 mg/mL glutamine, 100 U/mL penicillin, and I00 mg/mL streptomycin. 3T3 cells were grown to confluency in 2 to 3 days in multiwell plates, then incubated for 18 hours in 1% plasma-derived serum with either the unknown or fetal calf serum standards. At 474
18 hours, 0.25 ~tC 3H-thymidine was added per well, and 3H-thymidine incorporation was determined at 20 hours. Cell monolayers were washed twice in 10% trichloroacetic acid for 10 minutes at 4~ and the cells were solubilized in 0.2 mL 0.5N sodium hydroxide. The total sample was counted in 5 mL scintillation fluid on a scintillation counter. Mitogenic activity was compared with the activity of standard mitogen units to determine the quantity of growth factor activity in each sample. One activity unit was defined as the amount of 3H-thymidine incorportion stimulated by 1 ng platelet-derived growth factor (PDGF) and was equivalent to 5 uL fetal calf serum standard. This allowed for comparison between samples in independent assays. Breast cyst fluid samples were assayed in triplicate at several concentrations, and the mean and standard deviation were determined for each point. Column chromatography: Breast cyst fluid samples were thawed, centrifuged, and analyzed by size-exclusion high-pressure liquid chromatography (HPLC). Two hundred microliters of sample supernatant was injected directly onto a Waters Protein Pak 300 SW column (7.8 mm • 30 cm) previously equilibrated in phosphate buffer (5.6 mM sodium phosphate/ 1.06 mM potassium phosphate, pH 7.5/0.15M sodium chloride) at room temperature. The column eluant was collected in l-mL fractions at a flow rate of 0.5 mL/minute. Fractions were directly assayed in the 3H-thymidine incorporation assay. The breast cancer tissue was thawed, and the major blood vessels were carefully dissected away from the tumor tissue and discarded. Approximately 130 g of frozen pooled breast tumors was acid/ethanol-extracted by the method of Roberts [21]. The lyophilized sample extract was solubilized in 6M guanidine-hydrochloric acid/0.1M potassium phosphate, pH 4.5, and fractionated on a Sephacryl S-200 (Pharmacia 3 cm • 77 cm) size-exclusion column equilibrated in the same buffer. The column was run at a flow rate of 0.25 mL/minute. Fractions of 4 mL were collected, dialyzed, lyophilized, and assayed for mitogenic activity. The molecular weight of the active fractions in all chromatography procedures was estimated by comparison to the elution profile of standard proteins. Radioassay: Growth factor competing activity was determined in fractions chromatographed on the Waters Protein Pak 300 SW HPLC size-exclusion column. Lyophilized samples were reconstituted inappropriate buffer and assayed in duplicate. In the identical fractions, combined human epidermal growth factor (EGF) and human transforming growth factor alpha (TGF-alpha) competing activity was determined by radioassay kit (Biomedical Technologies, Stoughton, MA) following the manufacturer's protocol. Stability tests: Stability tests were performed on fractions containing mitogenic activity that had been chromatographed on the Waters Protein Pak 300 SW HPLC size-exclusion column for breast cyst fluid samples or on the Sephacryl S-200 column for breast cancer tissue. Aliquots from fractions containing the mitogenic peak of interest were dialyzed and lyophilized. Acid-treated sam-
THE AMERICANJOURNALOFSURGERY VOLUME159 MAY 1990
TUMOR MARKERS IN BREAST CYST FLUID
ples were reconstituted in 0.5N acetic acid, incubated for 1 hour, then lyophilized and stored at -80~ Heattreated samples were reconstituted in 1 mM ammonium bicarbonate, heated at 80~ for 5 minutes, cooled, and directly assayed. Statistical analysis: Statistical significance of the difference in mitogenic activity between risk groups was determined using a variety of standard statistical methods. The means and standard deviation were calculated for each concentration of breast cyst fluid. The values for mitogenic activity determinations were analyzed by analysis of variance (ANOVA) to determine variance from the mean. The variation in mitogenic activity values across risk groups was analyzed at each breast cyst level using the Kruskal-Wallis method. A combination of these statistical methods was used to determine the relevance of the data. RESULTS Breast cyst fluid samples obtained from 35 women were examined for the presence of mitogenic activity. Women from whom samples were collected were categorized according to their assessed risk of developing breast cancer based on personal and family history of breast cancer. The highest-risk women, Group 1 (n = 10), were either newly diagnosed with breast cancer and associated cyst(s), previously diagnosed with breast cancer, or had two or more blood relations diagnosed with breast cancer. The moderate-risk women, Group 2 (n = 10), had either a suspicious mammogram, a breast biopsy showing atypia or hyperplasia, or one blood relation diagnosed with breast cancer. The lowest-risk women, Group 3 (n = 15), were women with no personal or family history of breast cancer. Their expected risk of breast cancer was considered to be the same as that of the general population (104 of 100,000 women). Breast cyst fluid samples were obtained from women in all three categories and coded to prevent the persons performing the biochemical assays from knowing the associated clinical history. Mitogenic activity of the coded breast cyst fluid sampies was determined at several concentrations in an in vitro proliferation assay by incorporation of 3H-thymidine into mouse fibroblast (NIH 3T3) cells. Individual dose-response curves were determined for each patient. When the code was unmasked, the breast cyst fluid sampies demonstrated a direct dose-response relationship between mitogenic activity and assessed risk of developing breast cancer, with the level of mitogenic activity being highest in women at greatest risk of developing breast cancer. The dose-response curves for three individual patients, representative of the different risk groups, are shown in Figure 1. The Group 1 sample, highest risk, was obtained from a patient who was diagnosed with extensive noninvasive ductal carcinoma concurrent with the cyst aspiration. A subsequent biopsy of the cyst cavity from which the aspirate was obtained showed a smoothly lined cavity with benign changes. The Group 2 sample, moderate risk, had an intermediate level of activity and was obtained from a patient whose mother had breast cancer. The Group 3 sample, lowest risk, had the lowest
25
~
-,
Q
20
10
0 25
0
5
75
~I B r e a s t
10
15
Cyst Fluid
Figure 1. Mitogenic activity dose-response curves for breast cyst fluid samples from individual patients with differing risks of developing breast cancer.
25
% i,q N
20
I
.~ 15
:
/~"~"
.
" ~
5
~.
;
_ . - 4 - . . . . ~" . . . . . .
/,f 9
9
,
D, C
~ la
9~ ~ 9 9 "
ModerateRisk L~ 1
0 0
25
5
75
10
15
gl Breast Cyst Fluid
Figure 2. Summary mitogenic activity dose-response curves for the high-, moderate-, and low-risk groups.
amount of mitogenic activity and was obtained from a woman with no family history of breast cancer who presented with a single unilocular breast cyst. There was a statistically significant difference in mean mitogenic activity values between the high-, moderate-, and low-risk groups. The summary dose-response curves were determined by combining the individual dose-response curves for all samples within a risk group (Figure 2). The effect of risk group classification was almost entirely on the mean level. We determined that the mean level across all breast cyst fluid levels per patient varied significantly by risk group (p --- 0.014 by ANOVA). The variation in mean levels across risk groups was statistically significant, with a p value of 0.012 at the 2.5-gL concentration of breast cyst fluid to a p value of 0.019 at 10 #L as determined by ANOVA. At 15/zL of breast cyst fluid, the variation in mean values became marginally significant (p -- 0.053). We also analyzed the variation in values of means across groups at each breast cyst fluid level using the nonparametric Kruskal-Wallis method. The results confirmed those obtained by ANOVA. Next, several representative samples were chosen
THE AMERICAN JOURNAL O F S U R G E R Y
V O L U M E 159
MAY 1990
475
TAPPER ET AL
A. ~
P.Jsk Group
B. Moderate Risk Group
C. Low Risk Group
I |
-
"I.
I'i!
j" i
o
J.o ]Prtt9
.
'~0
.
.
.
.
20
Nmmber
Figure 3. Distribution of growth factor activities in individual breast cyst fluid samples fractionated by size-exclusion HPLC. kDa = kilodalton.
from each risk group, and the breast cyst fluid samples were analyzed by sizing column chromatography. Individual breast cyst fluid samples were centrifuged and multiple aliquots of the supernatants were fractionated by HPLC using a Waters Protein Pak 300 SW sizeexclusion column equilibrated in a neutral phosphate buffer as described previously. Column eluant was collected in separate fractions, and each fraction was tested for mitogenic activity in the in vitro proliferation assay. The protein profiles of all the samples were virtually identical. However, the distribution of growth factor activity in the fractionated breast cyst fluid differed between risk groups. Figure 3 shows examples of the distribution of growth factor activity. Samples from the highest-risk group had two peaks of mitogenic activity: one with an apparent molecular weight less than 6 kilodaltons (kd) and a second peak at 6 to 18 kd (Figure 3). Samples from the moderate-risk group demonstrated only a single peak of mitogenic activity at less than 6 kd, and the lowest-risk group had insignificant growth-promoting activity (no distinct peak) in either molecular weight range (Figure 3). The 6- to 18-kd peak of mitogenic activity was both acid- and heat-stable. The less than 6-kd mitogenic peak in the Group 1 and 2 samples is presumed to be human EGF as identified by radioimmunoassay. The presence of other growth factors in this peak has not been excluded. In addition to the breast cyst fluid samples, breast cancer tissue was investigated for the presence of similar growth factor activities. Surgical samples of breast cancer tissue were obtained from women at either the primary operation to determine the presence of breast malignancy, the definitive operation for treatment and staging of the breast disease, or autopsy. Breast cancer tissue of the same pathologic type was pooled from several donors and extracted by an acid-ethanol procedure [21]. The extract was fractionated by HPLC using a sizeexclusion column equilibrated in 6M guanidine-HC1, pH 4.5, as described. Column eluant was collected in frac476
THE AMERICAN JOURNAL OF SURGERY
tions, and each fraction was tested for mitogenic activity. As seen in the breast cyst fluid samples from high-risk patients, breast cancer tissue contained several peaks of mitogenic activity, including a predominant peak of mitogenic activity 6 to 14 kd in molecular weight (Figure 4). This peak of activity in breast cancer tissue also demonstrated acid- and heat-stability as observed for breast cyst fluid. COMMENTS These studies have shown that breast cyst fluid from certain women contains an increased level of growth factor activity that has the potential to be a useful biological marker for breast cancer. The data clearly show a direct dose-dependent relationship between mitogenic activity and an increased risk of developing breast cancer. The mitogenic activity in the samples from the high-risk group was several-fold greater than the activity present in the samples from the low-risk group. When analyzed by size-exclusion chromatography, samples from the highrisk group contained a 6- to 18-kd acid- and heat-stable mitogenic component and a less than 6-kd component subsequently identified as human EGF. Samples from the moderate-risk group possessed only the less than 6-kd peak, and the lowest-risk group samples had no significant growth factor activity. In addition, breast cancer tissue was also analyzed by size-exclusion column chromatography and shown to contain a 6- to 14-kd acid- and heat-stable mitogenic component similar to that found in breast cyst fluid. As a result of physician and public awareness, more women are now diagnosed with what has been termed minimal breast disease. However, despite all efforts toward earlier diagnosis, the mortality rate from breast cancer has remained constant [3]. Several modalities, including mammography, self-examination, aspiration cytology, and biological tumor markers, have been recommended as possible methods of earlier and more sensitive detection of breast cancer [22-24].
VOLUME 159
MAY 1990
TUMORMARKERSIN BREASTCYSTFLUID
In the search for biological tumor markers for breast cancer, a variety of approaches have been pursued, including analysis of metabolic by-products, oncofetal antigens, viral particles or related macromolecules, plasma proteins, hormones, and cell-associated antigens of breast cancer tissue. Although the study of tumor markers has advanced rapidly, and several biochemical markers have been identified that are elevated in patients with metastatic breast disease [25], at present no specific biological tumor marker exists for early diagnosis of breast cancer. Current evidence indicates that peptide growth factors such as EGF, PDGF, TGF-alpha and TGF-beta, and the insulin-like growth factors (IGFs) may be pivotal in the progressive development of malignantly transformed cells [4-6]. Breast cancer cells in vitro secrete several growth factors including IGF-1, TGF-alpha and TGFbeta, and a PDGF-Iike factor. Many of these growth factors are inducible by estradiol [26,27]. In particular, IGF-1 has been shown to promote breast cancer cell growth in vitro [28] and to be present in a higher concentration in primary breast cancer tissue than in adjacent normal tissue [29]. Thus, the aberrant production of growth factors may be associated with the neoplastic growth of breast cancer cells. We have previously identified a specific mitogenic activity in the milk of a substrain of mice that develop mammary tumors. This mitogenic activity is unique in that its presence in milk is predictive of mammary tumor development long before the presence of a palpable tumor (J Surg Res, in press). This predictive activity is composed of a subset of acid- and heat-stable peptide growth factors in the 6- to 10-kd molecular weight range. Of known growth factors tested, radioassay techniques identify an IGF-l-like peptide as the major component. The predictive growth factor activity also contained small amounts of PDGF- and TGF-beta-like compounds. PDGF, TGF-beta, and IGF-1, in combination, demonstrate potentiated mitogenic activity [30,31] and may be acting in a similar fashion in murine mammary tumor development. These growth factors are present only in mice that eventually develop mammary tumors and are not present in the milk of control mice. In the present study, the mitogenic activity in breast cyst fluid was directly related to women's increased risk of developing breast cancer. The breast cyst fluid from women in the high-risk group contained a 6- to 14-kd acid- and heat-stable growth factor activity similar to the predictive growth factor activity seen in the milk of mice that subsequently developed mammary tumors. Breast cyst fluid from women in the low-risk group did not contain this activity. In view of the results observed in murine milk, we believe that the presence of a specific mitogenic activity in breast cyst fluid may be associated with the development of breast cancer. The identification of a specific growth factor in breast cyst fluid and breast cancer tissue may be important not only as a predictor of malignancy and as a biological tool for early diagnosis, but also as an indication of effective therapy and a possible guide to adjuvant therapy. As with other biological markers (i.e., alpha-fetoprotein and hu-
60
%
r
ii
el
el
a Jr
a Jr
D Jr
0 Jr
o
~,
r
$
,I,
$
,I,
50
X
==
30 20 10 0 30
40 50 60
70
Fraction number Figure 4. Distribution of mitogenic activity in breast cancer tissue fracUonated by size-exclusion column chromatography, kDa = kilodalton.
man chorionic gonadotrophin) that decrease with successful treatment, the level of specific growth factor activity found in the breast cyst fluid of a woman who had previous breast cancer surgery and adjuvant chemotherapy was quite low. The disappearance of the marker may reflect successful treatment. The presence of an elevated marker level in the breast cyst fluid of a woman previously treated for breast cancer might herald the presence of recurrent disease. Additionally, the presence of this specific marker might be correlated with other growth factors and receptors to further stratify women into risk groups for the purpose of modifying their adjuvant therapy regimen. Estrogen and progesterone receptors have been well characterized with regard to their relationship to prognosis and plan for adjuvant therapy. Recently, EGF receptors have been identified in breast cancer tissue and metastatic lymph nodes removed at surgery. Those patients with EGF-positive receptors had a poorer prognosis [32,33]. There undoubtedly exists a complex interplay between growth factors, hormone receptors, and other oncogenes. The further identification of these
THE AMERICANJOURNALOFSURGERY VOLUME159 MAY 1990 477
TAPPER ET AL
growth factors and the role they play in breast cancer may improve our ability to diagnose these lesions earlier and to treat them more effectively. We wish to emphasize that this was a preliminary study of small sample size, prompted by the results of our investigation in an animal model and the availability of a large patient population of women with benign and malignant breast disease. This study is currently being expanded to include a large, statistically significant sample size, a more accurate assessment of risk by assignment of a computer-generated risk score based upon the breast cancer risk factors presented by Kelsey and Berkowitz [34], and additional biochemical analysis to specifically identify the growth factors present in the 6- to 18-kd mitogenic peak found in the high-risk samples. Our longrange goal remains to develop a sensitive, specific assay of body fluids for early detection of breast cancer and effective monitoring of treatment and probability of recurrence. REFERENCES 1. Young TL, ed. SEER Program. Incidence and mortality, 19731977. Bethesda, MD: National Cancer Institute, 1981. Monograph 57, NIH publication 81-2330. 2. Annual cancer statistics review. Bethesda, MD: National Cancer Institute, January 1988. 3. Kusantia S, Spratt TS, Donovan WL. The gross rates of growth of human mammary carcinoma. Cancer 1972; 30: 594-9. 4. Chan VW, McGee JO'D. Cellular oncegenes in neoplasia. J Clin Pathol 1987; 40: 1055-63. 5. Bishop JM. Oncogenes and proto-oncogenes. J Cell Physiol [suppl] 1986; 4: 1-5. 6. Heldin CH, Westermark B. Growth factors: mechanism of action and relation to oncogenes. Cell 1984; 37: 9-20. 7. Patz J, Tapper D, Outzen H, Klagsburn M, Sing Y. Determination of breast cancer in mice by the identification of growth factor activity in breast milk. Surg Forum 1983; 34: 432-5. 8. Tapper D, Langer R, Bellows AR, Folkman J. Angiogenesis capacity as a diagnostic marker for human eye tumorsl Surgery 1979; 86: 36-40. 9. Tapper D, Albert DM, Robinson NL, Zetter BR. Capillary endothelial cell migration: stimulating activity of aqueous humor from patients with ocular cancers. J Natl Cancer Inst 1983; 71: 501-5. 10. Chodak G, Scheiver C, Zetter B. Urine from patients with transitional-cell carcinoma stimulates migration of capillary endothelial cells. N Engl J Med 1981; 305: 869-74. 11. Brem H, Patz J, Tapper D. Detection of human central nervous system tumors: use of migration stimulating activity of the cerebrospinal fluid. Surg Forum 1983; 34: 532-4. 12. Connolly JM, Rose DP, Epidermal growth factor-like proteins in breast fluid and human milk. Life Sci 1988; 42: 1751-6. 13. Cera K, Mahan DC, Simmen FA. In vitro growth-promoting activity of porcine mammary secretions: initial characterization and relationship to known peptide growth factors. J Anim SCI 1987; 65:1149-59. 14. Zwiebel JA, Bano M, Nexo E, Salomon DS, Kidwell WR. Partial purification of transforming growth factors from human
478
milk. Cancer Res 1986; 46: 933-9. 15. Bano M, Salomon DS, Kidwell WR. Purification of a mammary-derived growth factor from human milk and human mammary tumors. J Biol Chem 1985; 260: 5745-52. 16. Grueters A, Larshmanan J, Tarris R, Aim J, Fisher DA. Nerve growth factor in mouse milk during early lactation: lack of dependency on submandibular salivary glands. Pediatr Res 1985; 19: 934-7. 17. Grueters A, Aim J, Lakshmanan J, Fisher DA. Epidermal growth factor in mouse milk during early lactation: lack of dependency on submandibular glands. Pediatr Res 1985; 19: 853-6. 18. Petrides PE, Hosang M, Shooter E, Esch FS, Bohlen P. Isolation and characterization of epidermal growth factor from human milk. FEBS Lett 1985; 187: 89-95. 19. Shing Y, Davidson S, Klagsbrun M. Purification of polypeptide growth factors from milk. Methods Enzymol 1987; 146: 42-8. 20. Noda K, Umeda M, Ono T. Transforming growth factor activity in human eolostrum. Gann 1984; 75: 109-12. 21. Roberts AB, Lamb LC, Newton DL, Sporn MB, DeLarco JE, Todaro GJ. Proc Natl Acad Sci USA 1980; 77: 3494-8. 22. Powell PW, McSwceney MD, Wilson CE. X-ray calcifications as the only basis for breast biopsy. Ann Surg 1983; 197: 555-9. 23. Robbins GF, Brothers TH, Eberhart WF. Is aspiration biopsy of breast cancer dangerous to the patient? Cancer 1954; 7: 774-8. 24. Strax P. Evaluation of screening programs for the early diagnosis of breast cancer. Surg Clin North Am 1978; 58: 667-79. 25. Twardzik DR, Sberwin SA, Ranchalis J, Todaro GJ. Transforming growth factors in the urine of normal, pregnant, and tumor-beating humans. J Natl Cancer Inst 1982; 69: 793-8. 26. Lippman ME, Dickson RB, Bates S, et al. Autocrine and paracrine growth regulation of human breast cancer. Breast Cancer Res Treat 1986; 7: 59-70. 27. Kasid A, Lippman ME. Estrogen and oncogene mediated growth regulation of human breast cancer cells. J Steroid Biochem 1987; 27: 465-70. 28. Dickson RB, Huff KI, Spencer EM, Lippman ME. Induction of epidermal growth factor-related polypeptides by 17B-estradiol in MCF-7 human breast cancer cells. Endocrinology 1985; 118: 13842. 29. Pekonen F, Partanen S, Makinen T, Rutanen EM. Receptors for epidermal growth factor and insulin-like growth factor-I and their relation to steroid receptors in human breast cancer. Cancer Res 1988; 48: 1343-7. 30. Pledger W J, Stiles CD, Antoniades HN, Scher CD. Induction of DNA synthesis in BALBc3T3 cells by serum components; reevaluation of the commitment process. Proc Natl Acad Sci USA 1977; 74: 4481-5. 31. Liboi E, Pelosi E, DiFrancesco P, et al. The El2 rat fibroblasts line: differential effects of growth factors (EGF, PDGF, FGF, TPA, and TGFB) on cell proliferation and c-los expression. Ann NY Acad Sci 1987; 511: 318-28. 32. Sainsbury JRC, Malcolm A J, Appleton, Farndon JR, Harris AL. Presence of epidermal growth factor receptor as an indicator of poor prognosis in patients with breast cancer. J Clin Pathol 1985; 38: 1225-8. 33. Rios MA, Macias A, Perez R, Lage A, Skoog L. Receptors for epidermal growth factor and estrogen as predictors of relapse in patients with mammary carcinoma. Anticancer Res 1988; 8: 1736. 34. Kelsey JL, Berkowitz GS. Breast cancer epidemiology. Cancer Res 1988; 48" 5615-23.
THE AMERICAN JOURNAL OF SURGERY VOLUME 159 MAY 1990
