Carcinoembryonic Antigen ImmunoscintigraphyComplements Mammography in the Diagnosis ofBreast Carcinoma

David M. Goldenberg, Sc.D., M.D.1

Hani Abdel-Nabi, M.D., Ph.D.2

Cynthia L. Sullivan, M.S., M.B.A.3

Aldo Serafini, M.D.4

David Seldin, M.D.5

Bruce Barron, M.D.6

Lamk Lamki, M.D.6

Bruce Line, M.D.7

William A. Wegener, M.D., Ph.D.3

1 Garden State Cancer Center, Belleville, New Jersey.2 Department of Nuclear Medicine, State Universityof New York at Buffalo, Buffalo, New York.3 Immunomedics, Inc., Morris Plains, New Jersey.4 Department of Nuclear Medicine, University ofMiami School of Medicine, Miami, Florida.5 Department of Nuclear Medicine, Lahey Clinic,Burlington, Massachusetts.6 Department of Nuclear Medicine, University ofTexas Medical School-Houston, Memorial-Her-mann Hospital, Houston, Texas.7 Department of Nuclear Medicine, Albany MedicalCenter, Albany, New York.

A brief preliminary report of this study was pub-lished in Acta Med Austriaca 1997;2:55–9 as partof the proceedings of a conference held in Woer-thersee, Austria, 1997.

Supported by grants from Immunomedics, Inc., to theparticipating institutions, and by United States PublicHealth Service Outstanding Investigator GrantCA39841 from the National Cancer Institute, NationalInstitutes of Health (D.M.G.).

Dr. Goldenberg is a board member and share-holder of Immunomedics Inc.

Dr. Abdel-Nabi acted as a consultant for Immuno-medics Inc. between September 1996 and April1998.

Dr. Wegener and Ms. Sullivan are employees andshareholders of Immunomedics Inc.

The authors thank K. Karlson, M.D. (Livingston, NJ)and H. Ginsberg, M.D. (Morristown, NJ) for inde-

pendently classifying the mammograms accordingto BI-RADS categories, and R. Doerr, M.D. (Buffalo,NY), R. Ellis, M.D. (Springfield, IL), E.R. Fisher, M.D.(Pittsburgh, PA), N. A. Goldenberg, M.D. (Tampa,FL), and I. Khalkhali, M.D. (Los Angeles, CA) forcritically reviewing the manuscript. Ms. HeatherZiccardi provided central image processing andtechnical review of the nuclear scans.

The work presented here is part of a larger con-

sortium of investigators participating in this trial,with authorship designated for the principal nu-clear medicine investigators of the major contrib-uting centers.

Address for reprints: David M. Goldenberg, Sc.D.,M.D., Garden State Cancer Center, 520 BellevilleAvenue, Belleville, NJ 07109-0023.

Received October 1, 1999; revision February 29,2000; accepted April 7, 2000.

BACKGROUND. An adjunctive noninvasive test that is predictable and highly spe-

cific for breast carcinoma would complement the high false-positive rate of mam-

mography in certain patients.

METHODS. This prospective, multicenter study evaluated the accuracy, safety, and

immunogenicity of carcinoembryonic antigen (CEA) antibody imaging in women

with known or suspected breast carcinoma. Scintigraphic breast images were

obtained approximately 3– 8 hours after the administration of technetium 99m

(99Tc) labeled anti-CEA Fab9 and correlated with histopathology.

RESULTS. The 99Tc labeled anti-CEA Fab9 detected tumor CEA expression in 46 of 49

women (94%) initially entered with known primary breast carcinoma regardless of

histology or serum CEA levels. In women scheduled for biopsy confirmation of mam-

mographic and physical examination findings, 104 99Tc labeled anti-CEA Fab9 studies

had a sensitivity of 61% (17 of 28 cases) and a specificity of 91% (69 of 76 cases). In

total, 99Tc labeled anti-CEA Fab9 detected 52 of 62 invasive ductal carcinomas, 5 of 5

invasive lobular carcinomas, and 3 of 6 noninvasive tumors (2 ductal carcinomas in

situ and 1 intracystic papillary carcinoma). Tumor size significantly affected sensitivity

(P 5 0.041), with 11 of 14 missed lesions # T1, and proliferative histology significantly

affected specificity (P 5 0.012), with 5 of 7 false-positive tumors being premalignant.

In 50 breast carcinoma patients, 99Tc labeled anti-CEA Fab9 also demonstrated axillary

lymph node involvement regardless of serum CEA levels, with a sensitivity of 80%

when more than three lymph nodes were positive. No immune response or other

meaningful side effects occurred.

CONCLUSIONS. 99Tc labeled anti-CEA Fab9 had high specificity and positive pre-

dictive values for breast carcinoma and the majority of false-positive studies were

associated with an increased risk of malignancy. Improved imaging techniques,

including dedicated gamma cameras for breast and axillary lymph node imaging,

will likely improve the test’s sensitivity for smaller lesions. Cancer 2000;89:104 –15.

© 2000 American Cancer Society.

KEYWORDS: antibodies, Arcitumomab, breast carcinoma, carcinoembryonic anti-gen, CEA-Scan, diagnosis, imaging, immunoscintigraphy, mammography.

104

© 2000 American Cancer Society

The frequency of breast carcinoma has been steadilyincreasing and not just because of the relatively

recent use of more extensive screening by mammog-raphy.1–3 In the U.S., it is projected that the lifetimerisk of breast carcinoma occurring in women is one ineight.4 Earlier detection and diagnosis impacts sur-vival, with a generally estimated 24 –35% mortalityreduction both in women aged 40 – 49 years and forthose aged 50 years and older5–13 However, the highsensitivity of screening mammography is associatedwith a high rate of detection of benign disease. Cur-rently, it generally is stated that only 20 –30% of allabnormal mammograms prove to have malignant le-sions.5,14,15 Reports have shown a wide variation of thefalse-positive rate from 66% to 88%.14 –24 Furthermore,it is well recognized that a negative mammographicfinding does not exclude the presence of breast carci-noma,13 and in one series a false-negative rate ofmammography of 16.5% was found even in patientswith palpable masses.25 These inconsistencies are in-dicative of a considerable variability in the practice ofmammographic interpretation26 –28 and may contrib-ute to breast carcinoma diagnosis being the leadingcause of malpractice claims in the U.S.29 Accordingly,other diagnostic measures are needed to complementthe high sensitivity but low specificity (high false-pos-itivity) of mammography.

Unfortunately, the available diagnostic imagingtests, including magnetic resonance imaging, ultra-sonography, computed tomography, and positronemission tomography, frequently do not provide suf-ficient specificity to differentiate between benign andmalignant changes in the breast.30 –33 More recently,nuclear imaging with 99mTc-Sestamibi (Miraluma; Du-Pont Merck Pharmaceuticals, Billerica, MA) has beenused for discriminating mammary tumors in densebreasts, but it appears to be more reliable for imagingpalpable than nonpalpable tumors, being limited tomass lesions greater than 1.0 cm.34 –39

Whereas most diagnostic imaging modalities relyon anatomic characteristics of the tissues or organs,antibodies against tumor-associated antigens havebeen proposed as more specific tumor imaging agentswhen applied to nuclear scanning, a technology calledradioimmunodetection or immunoscintigraphy.40 – 42

Of these candidate antibodies, to our knowledge thoseagainst carcinoembryonic antigen (CEA) were the firstevaluated in this setting.40,41 CEA, described by Goldand Freedman in 1965 as an oncofetal marker forgastrointestinal neoplasia,43 has since been shown tobe expressed by a variety of solid cancers, includingthe common malignant epithelial tumors.44 CEA ex-pression has been reported in up to 90% of breastcarcinoma specimens by a variety of in vitro and in

vivo methods,45–55 although blood CEA levels seldomare elevated in patients with localized primary dis-ease.46,48,49 This suggests a discordance between tissueand serum levels of CEA.

Twenty years ago, it was reported that radiola-beled CEA antibodies could detect and localize diverseneoplasms producing CEA, including breast carcino-ma.40,41 Imaging of regional lymph nodes in breastcarcinoma patients by immunolymphoscintigraphyalso has been found to be feasible with CEA antibod-ies.56,57 In studies performed subsequently, differentradiolabeled antibodies to CEA and other breast car-cinoma markers confirmed the detection and localiza-tion of breast carcinoma.42,58 – 62 Arcitumomab is aCEA specific monoclonal antibody Fab9 fragment ap-proved in the U.S., Canada, and Europe as a techne-tium-99m (99mTc) labeled imaging agent for the detec-tion of metastatic/recurrent colorectal carcinoma.This report summarizes our initial experience to de-termine whether Arcitumomab imaging can distin-guish benign from malignant breast lesions, the po-tential role of this new imaging method followingmammography in the detection and diagnosis ofbreast carcinoma, and to evaluate the product’s safetyin this population. Ancillary findings in some primarybreast carcinoma patients also explored the ability ofArcitumomab to detect axillary lymph node involve-ment.

PATIENTS AND METHODSPatientsThis prospective, nonrandomized, open-label, multi-center trial was conducted at 8 institutions (2 commu-nity and 6 university hospitals) in 210 women, 21 yearsor older. Initially, 49 women were entered with histo-logically proven primary breast carcinoma before un-dergoing definitive surgery. Subsequently, 103 womenwere entered with only clinically suspected breast car-cinoma, based on mammography, physical examina-tion, and other findings for which biopsy confirmationwas indicated. The remaining patients were only in-cluded in analyses of safety, because they either hadadvanced disease and were enrolled for whole-bodyevaluation of recurrence/metastases, had breast tissuesurgically resected before imaging, or lacked biopsy orsurgical confirmation of image findings. Patients witha positive urine or serum hCG were excluded from thestudy, as were those at risk of conception and notusing an effective method of contraception. The pro-tocol was approved by the U.S. Food and Drug Ad-ministration and by the ethical review board at eachstudy site. Written informed consent was obtainedfrom all patients before entry into the study.

Each patient had a medical history and a physical

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 105

examination and complete blood chemistries, hema-tologic values, and urinalyses determined. These lab-oratory studies then were repeated at 1 day and 7 daysafter the antibody study was completed. To assess theimmunogenicity of the CEA antibody fragment, serumwas collected for determination of human anti-mouseantibody (HAMA) titers at baseline, 4 – 6 weeks, and3– 4 months after the study, using the Immu-STRIPHAMA assay of Immunomedics, Inc. (Morris Plains,NJ). A level above 74 ng/mL is considered to be ab-normal in this assay.63 In addition, a serum CEA levelwas obtained before the administration of Arcitu-momab to assess the effect of circulating CEA on an-tibody imaging. The primary safety criteria for whicheach patient was assessed were the development ofHAMA, changes in clinical laboratory values, adversereactions, and deaths and dropouts due to adverseevents.

ImmunoscintigraphyArcitumomab (CEA-Scan; Immunomedics, Inc., Mor-ris Plains, NJ) was supplied as a sterile, nonpyrogenic,lyophilized powder in 3-mL vials, formulated to con-tain 1.25 mg of the NP-4 murine anti-CEA monoclonalantibody Fab9 fragment, described previously.64 When99mTc sodium pertechnetate in sterile, normal saline isadded to the vial contents and shaken gently for ap-proximately 5 minutes, it is ready for intravenous in-jection. A dose of 1 mg of the antibody fragmentlabeled with 25 6 5 mCi (750 –1100 MBq) 99mTc wasadministered as a 5-minute injection in 1–2 mL ofisotonic saline or, occasionally, as a 20 –30-minuteintravenous infusion after dilution in 30 mL of saline.

Arcitumomab images of the chest were obtained3– 8 hours after injection using a high-resolution, low-energy, collimator and a 20% window set at the 140-keV photopeak of Tc-99m. Planar images were ac-

quired using a 128 3 128 matrix at 700 –1000K countsper view, whereas single photon emission computedtomography (SPECT) imaging used a 64 3 64 matrix,with 128 angles at 25–30 seconds per angle. Anteriorand lateral planar images were obtained with the pa-tient in an upright position, both with arms raisedabove the head and then at the sides. In some pa-tients, lateral and oblique images were obtained withthe patient in the prone position and the breasts freelydependent.

Arcitumomab imaging results were interpreted bythe nuclear medicine physicians at each institution.The physicians had knowledge of the clinical historybut were blinded to mammographic results, except toknow the involved breast, so that comparisons to thecontralateral breast could be made. For women en-rolled with only clinically suspected breast carcinoma,the imaging studies were interpreted before undergo-ing biopsy and thus without knowledge of outcome.The scans were interpreted as positive when increasedfocal activity was revealed, using a schematic diagramto record lesion location, and negative when either noactivity or only weak, diffuse, and homogeneous ac-tivity was observed, including mild activity in the nip-ples. Particularly for patients presenting with nonpal-pable, mammographically suspicious lesions,abnormal breast activity was compared with a com-puter-determined identical site in the contralateralbreast, and then to the abnormal foci in the mammo-gram. Images for an abnormal and a contralateralnormal breast are shown in Figure 1.

Mammography and Physical Breast ExaminationThese procedures were performed for all patientswithin 4 weeks before study entry and the size and siteof each suspected lesion was recorded on a commonschematic diagram. Mammography included routine

FIGURE 1. Arcitumomab imaging is shown of a 54-year-old female who presented with a palpable, 1-cm mass in the upper outer quadrant of her left breast.

Her baseline serum carcinoembryonic antigen (CEA) was 1 ng/mL. Mammography was positive for a mass that was considered to be highly suggestive of carcinoma.

The Arctiumomab study shows a focal density in the left breast (arrows) in the (A) upright position and the (B) left lateral prone position, whereas (C) the contralateral

right breast shows only diffuse activity.

106 CANCER July 1, 2000 / Volume 89 / Number 1

craniocaudal and mediolateral oblique views, as wellas additional diagnostic views (magnification, spotcompression, recalled views, etc.) and ultrasonogra-phy, if required. The Breast Imaging Reporting andData System (BI-RADS) of the American College ofRadiology was not routinely used at the time of thisstudy.65 As such, the mammography reports were readby two independent mammographers who classifiedfindings according BI-RADS score. In case of discor-dance, the higher score was assigned.

Histopathologic ConfirmationThe most definitive result was used for correlationwith imaging study findings according to surgicalspecimen, core needle biopsy, and fine-needle aspira-tion. Based on surgical pathology report descriptionsof proliferative disease, benign breast results wereclassified according to the relative risk (RR) for devel-oping cancer as: 1) fibrocystic disease, fibroadenoma,cysts, surgical fibrosis, mild hyperplasia, or other non-proliferative conditions with no increased risk(RR,1.0), or 2) proliferative disease considered atslightly increased risk, including moderate or floridhyperplasia without atypia, and sclerosing adenosis(RR,1.5–2), or 3) proliferative disease at a more ele-vated risk, including moderate or florid hyperplasiawith atypia (RR, 4 –5).66,67

Statistical AnalysesBased on histopathology findings, positive Arcitu-momab immunoscintigraphy results were classified astrue-positive (TP) or false-positive (FP), and negativeimaging results as true-negative (TN) or false-negative(FN) for cancer. Sensitivity was calculated as #TP/(#TP1 #FN), specificity as #TN/(#TN 1 #FP), positive pre-dictive value (PPV) as #TP/(#TP 1 #FP), negative pre-dictive value as #TN/(#TN 1 #FN), positive likelihoodratio as sensitivity/(1 2 specificity), and negative like-lihood ratio as (1 2 sensitivity)/specificity, with theprevalence-independent likelihood ratios describingthe change in odds favoring cancer for a particular testresult.68 –70 For stratification, each breast lesion wasclinically classified using the TNM system adapted bythe International Union Against Cancer and the Amer-ican Joint Committee on Cancer.71 Based on the larg-est size breast lesion reported by physical examinationor mammography, the T-classification was specifiedas T0 (no discrete lesion reported), Tis (microcalcifi-cations without discrete mass), T1 (# 2 cm), T2 (2–5cm), T3 (. 5 cm), or T4 (skin involvement or fixationto the chest wall). We used Fisher exact test for uni-variate analyses of sensitivity and specificity. All Pvalues were two tailed, and the 0.05 level was consid-ered statistically significant.

RESULTSKnown Breast CarcinomaIn a preliminary study to determine the potential ofanti-CEA imaging for detecting breast carcinoma, 49patients initially were entered with known primarybreast carcinoma. These women had a positive fine-needle aspiration or needle biopsy before study (me-dian interval, 13 days) and underwent surgery afterArcitumomab imaging (median interval, 19 days). Asshown in Table 1, most patients were older than 50years, Caucasian, and postmenopausal with serumCEA levels generally normal (# 2.5 ng/mL) or onlymildly elevated. Most lesions were palpable, T2, andBI-RADS 5. The predominant histology was invasiveductal carcinoma, with only a few patients havinginvasive lobular carcinomas or noninvasive cancers,mainly ductal carcinomas in situ (DCIS). In this groupof women, there were 46 TP and 3 FN Arcitumomabstudies. Thus, Arcitumomab imaging demonstratedconsistent ability to target CEA expression in breastcarcinoma regardless of histology or serum CEA levels(46 of 49, 94%).

Suspected Breast CarcinomaThe ability of Arcitumomab to differentiate breast car-cinoma from benign breast conditions was assessedamong the 103 patients enrolled with only suspectedbreast carcinoma. These women had abnormal mam-mography and ultrasound, and physical examinationobtained within 4 weeks of study entry and underwentconfirming biopsy after Arcitumomab imaging (medi-an interval, 6 days). As compared in Table 1, this groupis demographically similar to the initial group, but thelesions were generally nonpalpable, T1, and BI-RADS4. The 28 patients entered with cancer that was onlysuspected and later confirmed had comparable histol-ogy, and the prevalence of breast carcinoma in thisgroup (28 of 103, 27%) is consistent with general prev-alence percentages of cancer for women with abnor-mal screening mammograms.5,14,15 Most of the re-maining 75 patients had cancer definitively excludedby surgical biopsy (n 5 16) or stereotactic core needlebiopsy (n 5 39), whereas 14 had manual needle corebiopsies, 3 had fine-needle aspirates, and 3 had neg-ative follow-up that included diagnostic mammo-grams. Benign findings mostly included fibrocysticdisease with no increased risk for developing cancer.However, 12 patients had findings at slightly increasedrisk (RR, 1.5–2), and 7 patients were at more elevatedrisk (RR, 4 –5). The 104 Arcitumomab studies (1woman with benign disease had 2 negative Arcitu-momab studies) included 17 TP, 69 TN, 11 FN, and 7FP results. Resulting diagnostic performance mea-

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 107

sures for Arcitumomab are given in Table 2. The mod-est sensitivity (17 of 28, 61%) involved smaller sizedtumors, because 9 of the 11 FNs were T1 (# 2 cm) orless. However, the specificity is high (69 of 76, 91%)with only 7 Arcitumomab studies considered FP forcancer. Of these, one woman had only a fine-needleaspirate of a cyst, but no histopathology, and one hadbenign histopathology findings not associated with anincreased risk for cancer; however, the other fivewomen had lesions generally considered to be at in-creased risk for cancer (atypical ductal hyperplasia:two; hyperplasia without atypia: one; sclerosing ad-enosis: two).

The 103 women were further stratified accordingto the presence of palpable (n 5 25) or nonpalpable (n5 78) lesions. The prevalence of breast carcinoma waslower among women entered with nonpalpable ratherthan palpable lesions (18%, 14 of 78 vs. 56%, 14 of 25,respectively), and the Arcitumomab studies per-formed had a lower sensitivity (36%, 5 of 14 vs. 86%, 12of 14, respectively) but higher specificity in this moredifficult subgroup (94%, 61 of 65 vs. 73%, 8 of 11,respectively). An example of Arcitumomab detectionof a nonpalpable, well differentiated papillary carci-noma of the breast is presented in Figure 2.

Arcitumomab Breast Imaging StudiesArcitumomab results were based on each investiga-tor’s overall interpretation of the acquired images. The3– 8 hour window for imaging allowed flexibility forthe initial trial experience imaging women with knowncancer, but most images were acquired in the 3–5-hour time period. Table 3 examines the role of planarand SPECT imaging in breast evaluation. Delayed im-aging was not routinely used and appeared to be use-ful only in one patient. Of the 70 positive Arcitu-momab studies obtained in this trial, 63 studies hadpositive planar results. The other seven cases includethe single patient mentioned above who required next

TABLE 1Patient Characteristics

CharacteristicKnown primarycancer (n 5 49)

Clinically suspectedcancer (n 5 103)

Age (yrs), 40 7 1740–49 8 16$ 50 34 70

RaceCaucasian 26 63Black 11 28Hispanic 10 6Other 2 6

Menopausal status:Premenopausal 13 27Perimenopausal 1 4Postmenopausal 35 72

Serum CEA (ng/mL) mean 6 SD 2.5 6 2.7 2.6 6 3.5Mass

Palpable 39 25Nonpalpable 10 78

T-classificationT0, Tis 4 24T1 15 54T2 21 22T3, T4 8 3Othera 1 0

BI-RADS:1–2 2 63 1 194 13 555 31 21NA 2 2

Cancers 49 28Invasive 44 25

Ductal CA 41 21Lobular CA 3 2Adeno CA, NOS 0 2

Noninvasive 3 3DCIS 3 2Papillary CA 0 1

Unknown invasiveness 2 0Benign lesions (histology)b — 75

Type I 50Type II 12Type III 7

Arcitumomab imagingTP 46 17FN 3 11TN — 69FP — 7

CEA: carcinoembryonic antigen; BI-RAS: Breast Imaging Reporting and Data System; NA: not available;

CA: carcinoma; NOS: not otherwise specified; DCIS: ductal carcinoma in situ; TP: true-positive; FN:

false-negative; TN: true-negative; FP: false-positive; RR: relative risk; FNA: fine-needle aspiration.a Nonpalpable mammographic mass of unspecified size.b Type I: nonproliferative conditions without increased cancer risk, RR 5 1.0; Type II: proliferative

disease without atypia, including sclerosing adenosis, RR 5 1.5–2; Type III: proliferative disease with

atypia, RR 5 4 –5. Type data not evaluable for three patients with FNA and three patients with negative

follow-up, including diagnostic mammogram.

TABLE 2Arcitumomab Imaging in Suspected Breast Carcinoma (104 Studies/103 Women)

Measure Percentage (correct/total)

Sensitivitya 60.7 (17/28)Specificityb 90.8 (69/76)Positive predictive value 70.8 (17/24)Negative predictive value 86.3 (69/80)Positive likelihood ratio 6.6Negative likelihood ratio 0.43

a 95% confidence interval 40.6 –78.5%.b 95% confidence interval, 81.9 –96.2%.

108 CANCER July 1, 2000 / Volume 89 / Number 1

day imaging for positive tumor identification, whereasSPECT was correctly positive for the other six patientswho had tumors not confidently identified on planarimages alone. No case occurred with positive planarimages and negative SPECT images. Of the 83 negativeArcitumomab studies, 66 also included SPECT imag-ing, with only 17 evaluated on the basis of planarimaging alone, including 1 patient whose only imageswere acquired the day after injection. Most negativeimaging studies had no focal uptake, although weaklypositive planar or SPECT images with low uptake con-sidered negative for breast carcinoma was recorded in16 patients. Thus, in most patients, planar imaging at3–5 hours was sufficient for identifying primary breastcarcinomas, but SPECT imaging or even delayed im-aging may be warranted for patients with negativeplanar images.

Univariate Analyses of Breast Imaging ResultsAll 153 imaging studies were examined for factors thatmight influence Arcitumomab breast imaging. Analy-ses of imaging sensitivity (Table 4) included all 77

women with primary breast carcinoma (49 known atstudy entry, 28 after eventual biopsy confirmation).Comparing 3 tumor grading categories, large masses. 2 cm (T2, T3, or T4), small masses # 2 cm (T1), andno measurable masses (Tis, T0), the differences werestatistically significant (P 5 0.04), with larger breastcarcinomas more frequently detected than smallerones (92%, 36 of 39, vs. 70%, 20 of 29, respectively).However, of eight patients without mass lesions, Arci-tumomab detected breast carcinoma in six, includingthree of six patients with microcalcifications or biopsyevidence of in situ cancer (Tis), and two patients withprior evidence of cancer elsewhere but not in thebreast (T0). Statistically significant differences alsowere associated with a palpable mass compared witha nonpalpable mass (94%, 50 of 53 vs. 54%, 13 of 24,respectively, P , 0.001). This may reflect differences intumor size, because 70% of studies with a palpablemass were T2 or higher compared with only 8% ofstudies with a nonpalpable mass. The sensitivity dif-ference between invasive (84%, 58 of 69) and nonin-vasive (50%, 3 of 6) cancers, given the small popula-

FIGURE 2. Arcitumomab imaging of a 60-year-old female who had prior mammograms demonstrating two separate lobulated lesions lying in the (A) upper outer

quadrant of the left breast, which showed an increase in size on follow-up over at least 1 year. Ultrasound demonstrated tiny cysts within the breast, but the

dominant lesions appeared to be solid. The Arcitumomab study showed (B) a round focus of increased radioactivity in the left breast, corresponding to the

abnormality observed on the mammogram. Image-localized stereotactic biopsies (14) only revealed papillary lesions with focal atypia. A lumpectomy then was

performed, and histopathologic examination revealed a well differentiated papillary carcinoma without stromal invasion.

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 109

tion size in the latter group, was not statisticallysignificant. However, all five patients with invasivelobular carcinomas were detected by Arcitumomab,and, of six noninvasive cancers, Arcitumomab imag-ing detected the single intracystic papillary carcinomaand two of the five DCIS. Serum CEA levels in patientswith breast carcinoma were normal (# 2.5 ng/mL) in48 patients, mildly elevated (2.5–5.0 ng/mL) in 18 pa-tients, and . 5 ng/mL in only 8 patients with infiltra-tive, mostly larger cancers (i.e., 7 of 8 T2 or higher).Sensitivity differences between women with normaland elevated serum CEA levels were not statisticallysignificant (88%, 42 of 48 vs. 73%, 19 of 26, respective-ly), indicating that Arcitumomab imaging is warrantedeven in the absence of abnormal serum CEA values.No statistically significant differences were foundamong other factors examined, including age andmenopausal status, nor were there significant sensi-tivity differences between cancer of the left or rightbreast, suggesting that neither liver background radio-activity on the right nor cardiac blood pool activity onthe left side diminished the ability to detect breasttumors.

Examining these same factors for potential influ-ence on Arcitumomab specificity found no statisticallysignificant differences based on breast lesion size orpalpability, serum CEA levels, age, or breast involved

(Table 5). However, specificity was significantly higheramong the majority of women who had nonprolifera-tive histology (98%, 48 of 49) compared with the mi-nority with proliferative findings, either with (71%, 5 of7) or without atypia (75%, 9 of 12) (P 5 0.012). Signif-icant differences between peri/postmenopausal andpremenopausal women also occurred (96%, 50 of 52vs. 79%, 19of 24, respectively, P 5 0.029). Because thepercentage of proliferative histology in both groupswas comparable (23%, 12 of 52 vs. 29%, 7 of 24, re-spectively), this difference may be an artifact of per-forming multiple subgroup analyses.

Axillary Lymph Node Involvement.Ancillary findings on chest imaging explored the po-tential of Arcitumomab imaging to detect axillarylymph node involvement. A total of 50 breast carci-noma patients had ipsilateral axillary biopsy/surgicalconfirmation, including 38 entered with known cancerand 12 with suspected cancer ultimately confirmed at

TABLE 3Imaging Study Results

Positive imaging studies

Known primary cancer(n 5 49)

Clinicallysuspectedcancer(n 5 103)

Total TP FP TP FP

Studies performed 70 46 0 17 7Planar1, SPECT1 60 39 0 14 7Planar1, SPECT not done 3 2 0 1 0Planar2, SPECT1 6 4 0 2 0Planar2, SPECT2a 1 1 0 0 0

Negative imaging studies Total TN FN TN FN

Studies performed 83 0 3 69 11Planar2, SPECT2 55 0 2 45 8Planar2, SPECT not done 11 0 0 8 3Planar6, SPECT6b 11 0 0 11 0Planar6, SPECT not done 5 0 1 4 0Imaging not donec 1 0 0 1 0

TP: true-positive; FP: false-positive; SPECT: single photon emission computed tomography; TN: true-

negative; FN: false-negative.a Lesion identified only on later imaging at 18 –24 hours.b 6: Low-level uptake considered negative for breast carcinoma.c Patient did not undergo same day imaging. Planar imaging at 18 –24 hours was negative.

TABLE 4Univariate Analyses of Imaging Sensitivitya

77 Women with primarybreast carcinoma

Percentage82 (63/77) Percentage

Clinical tumor classificationat study entryb

Tis, T0 75 (6/8)T1 70 (20/29)T2, T3, T4 92 (36/39) (P 5 0.041)

MassPalpable 94 (50/53)Nonpalpable 54 (13/24) (P 5 , 0.001)

Tumor typeInvasive 84 (58/69)Noninvasive 50 (3/6)Not specified 100 (2/2) (P 5 NS)

Age, 50 years old 94 (17/18)$ 50 years old 78 (46/59) (P 5 NS)

Menopausal statusPre 88 (14/16)Post/Peri 80 (49/61) (P 5 NS)

Serum CEAc

# 2.5 ng/mL 88 (42/48). 2.5 ng/mL 73 (19/26) (P 5 NS)

BreastLeft 80 (37/46)Right 86 (25/29)Bothd 50 (1/2) (P 5 NS)

NS: not significant; CEA: carcinoembryonic antigen.a P value by Fisher exact test, two-tail. NS . 0.05 level.b Excludes one patient with inevaluable T-classification.c Serum levels not obtained in three patients.d One woman with bilateral invasive ductal carcinomas had a positive imaging study for both breasts.

The other had a negative imaging study for both breasts.

110 CANCER July 1, 2000 / Volume 89 / Number 1

biopsy. Of these, 9 patients had 1–3 positive lymphnodes, whereas 20 had . 3 positive lymph nodes, and21 had negative lymph nodes. Of the 29 patients withaxillary involvement, 28 had invasive ductal carci-noma, and one had invasive lobular carcinoma astheir primary histology. All had normal serum CEAlevels. Based on the investigators’ imaging interpreta-tions for positive axillary uptake, there were 18 TP, 19TN, 11 FN, and 2 FP cases, resulting in a 62% (18 of 29)sensitivity and 91% (19 of 21) specificity. However,Arcitumomab imaging had an 80% (16 of 20) sensitiv-ity for . 3 positive lymph nodes compared with 22% (2of 9) for 1–3 positive lymph nodes, including somewith micrometastatic rather than macroscopic dis-ease. In addition, Arcitumomab had a 58% (11 of 19)sensitivity among patients with normal serum CEAlevels compared with 67% (6 of 9) with elevated levels.Thus, CEA is expressed in at least most positive axil-lary lymph nodes with macroscopic involvement andcan be detected by Arcitumomab imaging regardlessof serum CEA levels.

SafetySafety results were obtained in all 210 women whoreceived Arcitumomab. A total of 22 minor adverseevents were reported in 16 women. Those events con-sidered probably or possibly related to the agent in-cluded rash (n 5 3), taste in mouth (n 5 3), shortnessof breath (n 5 2), asymptomatic eosinophilia (n 5 2),dizziness (n 5 2), and one each of dry mouth, itching,distal elbow swelling, loose stools, hives, and “bright-er” vision. Only nine reactions were considered to beat least possibly related to Arcitumomab. Changes inhematology and blood chemistry parameters were in-frequent and consistent with the patients’ underlyingmedical conditions. Serum samples for an immuneresponse to the murine Fab9 fragment were availablein 152 women; all had normal HAMA titers.

DISCUSSIONThe objective of screening mammography is to dis-close occult cancer and effectively rule out malignancywhen it is absent. Each year, approximately 11% ofsuch examinations are interpreted as abnormal in thecommunity hospital setting of the U.S.,72 promptingfurther evaluation. Because it has been estimated that60% of women aged 50 years and older have experi-enced at least 1 screening mammogram,73 a substan-tial number of biopsies are performed, of which ap-proximately 80% prove to be benign.5,14,15 Indeed, theNew Mexico Mammography Project, evaluatingscreening mammography performance from 1991 to1993, found the predictive value of an abnormalscreen to be 4.3% and that of a resulting biopsy to be16.9%.24 The performance of mammography acrossthe U.S. varies greatly, especially with regard to theproportion of recall rates for screening mammographyand patients referred for biopsy.74 It thus is under-standable that this variability in terms of interpreta-tion and recommendations for biopsy can present thereferring physician and patient with a dilemma:Should there be an immediate biopsy and which typeof biopsy, other tests, or cautious follow-up? Further-more, for women managed conservatively, for whomfollow-up of an abnormal mammogram has been rec-ommended, inadequate follow-up (i.e., when notwithin 3 months of the due date) can occur in 18.1%,and for whom follow-up has been recommended in4 – 6 months, 36.8% can experience inadequate follow-up.75

This initial, multicenter trial provided definitiveevidence that the radiolabeled anti-CEA antibody Ar-citumomab targets primary breast carcinoma regard-less of serum CEA levels. Of note, the diagnostic per-formance of Arcitumomab imaging also did not vary

TABLE 5Univariate Analyses of Imaging Specificitya

76 studies in 75 women withoutbreast carcinoma

Percentage91 (69/76) Percentage

Clinical T-classification at study entryTis, T0 95 (20/21)T1 93 (37/40)T2, T3 80 (12/15) (P 5 NS)

Histology classb

Type I 98 (50/51)Type II 75 (9/12)Type III 71 (5/7) (P 5 0.012)

Age (yrs), 50 87 (26/30)$ 50 94 (43/46) (P 5 NS)

Menopausal statusPre 79 (19/24)Post/Peri 96 (50/52) (P 5 0.029)

Physical massPalpable 73 (8/11)Nonpalpable 94 (61/65) (P 5 NS)

Serum CEAc

# 2.5 ng/mL 90 (47/52). 2.5 ng/mL 86 (12/14) (P 5 NS)

BreastLeft 89 (32/36)Right 92 (36/39)Both 100 (1/1) (P 5 NS)

NS: not significant; CEA: carcinoembryonic antigen; RR: relative risk.a P-value by Fisher exact test, two-sided. NS at 0.05 level.b Type I: nonproliferative conditions without increased cancer risk, RR 5 1.0; Type II: proliferative

disease without atypia, including sclerosing adenosis, RR 5 1.5–2; Type III: proliferative disease with

atypia, RR 5 4 –5. Type data not evaluable for six patients.c Serum levels not obtained in four patients.

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 111

with histologic tumor type. Upon pathologic confir-mation, most invasive carcinomas imaged (52 of 62)were found to be infiltrative ductal carcinomas, andArcitumomab detected all 5 invasive lobular carcino-mas. It also disclosed noninvasive cancers, includingtwo of five ductal carcinomas in situ and a singleintraductal, intracystic, papillary carcinoma. Thus, itappears that most breast carcinomas express CEA,which can be targeted by Arcitumomab, and that im-munoscintigraphic detection is related to tumor size,with 11 of 14 of all lesions missed in the entire studypopulation being , 2 cm in size in the mammograms(9 with T1 and 2 with in situ cancers). The finding thatblood CEA value had no influence on the imagingresults, and that there was no immune response orother meaningful side effects observed in the patientsstudied, agrees with the findings for Arcitumomab incolorectal carcinoma patients,76 which also showedthe safety and efficacy of repeated Arcitumomab im-aging.77 In addition to invasive or noninvasive breastcarcinoma, it also appears that CEA is expressed inpremalignant conditions, because 5 of 7 FP findingsfor Arcitumomab imaging were related to proliferativelesions generally considered at increased risk of devel-oping cancer. Thus, they may not be FP in the strictestsense.

These predictable diagnostic performance resultssuggest a potential complementary role for Arcitu-momab imaging in defining whether a lesion seen onmammography has a high or low probability of beingcancer, and it is in this setting that the results of thistrial are clinically meaningful. Of the 103 women whoentered the study with suspicion of breast carcinoma,only 27% actually had breast carcinoma. However, thispercentage increased to 71% after a positive Arcitu-momab imaging study and decreased to 14% after anegative study. Further support that Arcitumomab up-take differentiates cancer from benign breast diseaseis given by the 91% specificity and the 6.6-fold in-crease in the probability of cancer for those lesionsfound positive by Arcitumomab imaging, with mostFP Arcitumomab studies of lesions associated with anincreased risk for cancer (e.g., proliferative disease).

The most difficult cases are clinically asymptom-atic women evaluated for nonpalpable mammo-graphic lesions, because occult cancer prevails withsmall tumors that remain difficult to detect with thecurrent generation of standard gamma cameras. Ofthe 78 patients who entered in our study with suspi-cious nonpalpable mammographic findings for whichbiopsy was indicated, the lesions were determined tobe cancerous in only 14 patients. Although the num-bers are small, we are encouraged by the PPV of Arci-tumomab (56%, 5 of 9) in this group compared with

that of mammography (18%, 14 of 78) as well as com-pared with 99mTc-Sestamibi PPV results reported re-cently for patients with nonpalpable lesions (38%).78

One of the intrinsic advantages of anti-CEA imaging isthat the basis for uptake and detection is tumor anti-gen expression and availability, whereas the basis forSestamibi uptake remains speculative, but may reflectnonspecific cellular proliferation.79

Current limitations to detecting smaller lesionswith immunoscintigraphy may reflect the need forfurther imaging experience and improved technique.Whereas this study primarily relied upon traditionalplanar chest images and SPECT, potential greater sen-sitivity may be achieved using specialized prone-po-sitioning patient tables now available for improvedlateral views as well as compact, digital-based camerasdesigned specifically for breast imaging. Even thoughspecific axillary images were not obtained in thisstudy, the ancillary findings on chest images demon-strated that anti-CEA expression in axillary lymphnode involvement also can be targeted with Arcitu-momab regardless of serum CEA levels. Further im-provement in sensitivity is anticipated when specificaxillary imaging protocols are implemented, particu-larly in conjunction with dedicated cameras.

In summary, Arcitumomab, an anti-CEA anti-body-based imaging agent, demonstrated the abilityto target primary breast carcinoma in a consistentmanner, with an excellent safety profile and no immu-nogenicity. False-positives were few and mostly in-volved lesions at increased risk of developing cancer,and the disclosure of premalignant lesions in 5 of 19women and preinvasive disease in 3 of 6 women war-rants further study of this test to detect early prolifer-ative changes of high risk for neoplasia, as suggestedin recent immunohistologic studies of CEA in precan-cerous lesions.52 As a complementary follow-up diag-nostic modality to mammography, Arcitumomabdemonstrated high specificity in the detection ofbreast lesions, and the finding that false-negatives pri-marily involved smaller tumors means that the sensi-tivity of Arcitumomab for detecting early disease islikely to further improve as breast imaging experienceincreases. Accordingly, the findings from this initialstudy suggest that a more extensive prospective trialbe conducted to confirm and extend these observa-tions, including the use of gamma camera systemsnow available or being developed for dedicated breastimaging. For patients with findings that are discordantfor cytology and mammography, an Arcitumomabstudy may add additional information needed forprompt action or deciding on follow-up, and the po-tential for Arcitumomab imaging to differentiate scartissue after surgery or irradiation from tumor recur-

112 CANCER July 1, 2000 / Volume 89 / Number 1

rence is also worthy of evaluation. Finally, the pros-pect of Arcitumomab targeting primary disease, axil-lary involvement, and distant metastases is underinvestigation, including intraoperative evaluation us-ing extremely sensitive hand-held gamma probes.

REFERENCES1. Glass AG, Hoover RN. Rising incidence of breast cancer:

relationship to stage and receptor status. J Natl Cancer Inst1990;82:693– 6.

2. Miller BA, Feuer EJ, Hankey BF. Recent incidence trends forbreast cancer in women and the relevance of early detec-tion: an update. CA Cancer J Clin 1993;43:27– 41.

3. Sondik EJ. Breast cancer trends. Incidence, mortality andsurvival. Cancer 1994;74:995–9.

4. Feuer EJ, Wun LM, Boring CC, Flanders WD, Timmel MJ,Tong T. The lifetime risk of developing breast cancer. J NatlCancer Inst 1993;85:892–7.

5. Harris JR, Lippman ME, Veronesi U, Willett W. Breast cancer(1). N Engl J Med 1992;327:319 –28.

6. Shapiro S. Screening: assessment of current studies. Cancer1994;74:231– 8.

7. Smart CR, Hendrick RE, Rutledge JH III, Smith RA. Benefit ofmammography screening in women ages 40 – 49 years: cur-rent evidence from randomized controlled trials. Cancer1995;75:1619 –26. Correction: Cancer 1995;75:2788.

8. Tabar L, Fagerberg G, Chen H-H, Duffy SW, Smart CR, GadA, et al. Efficacy of breast cancer screening by age: newresults from the Swedish Two-County Trial. Cancer 1995;75:997–1003.

9. Hendrick RE, Smith RA, Rutledge JH, Smart CR. Benefit ofscreening mammography in women ages 40 – 49: a newmeta-analysis of randomized controlled trials. Monogr NatlCancer Inst 1997;22:87–92.

10. Feig SA. Decreased breast cancer mortality through mam-mographic screening. Radiology 1988;167:659 – 65.

11. Haffry BG, Lee C, Philpotts L, Horvath L, Ward B, McKhannC, et al. Prognostic significance of mammographic detectionin a cohort of conservatively treated breast cancer patients.Cancer J Sci Am 1998;4:35– 40.

12. Kopans DB. The mammography screening controversy.Cancer J Sci Am 1998;4:22– 4.

13. Feig SA. Role and evaluation of mammography and otherimaging methods for breast cancer detection, diagnosis, andstaging. Semin Nucl Med 1999;29:3–15.

14. Homer MJ. Nonpalpable breast abnormalities: a realisticview of the accuracy of mammography in detecting malig-nancies. Radiology 1984;153:831–2.

15. Albert MP, Sachsse E, Coe NP, Page DW, Reed WP. Corre-lation between mammography and the pathology of non-palpable breast lesions. J Surg Oncol 1990;44:44 – 6.

16. Symmonds RE Jr., Roberts JW. Management of nonpalpablebreast abnormalities. Am Surg 1987;205:520 – 8.

17. Perdue P, Page D, Nellestein M, Salem C, Galbo C, Ghosh B.Early detection of breast carcinoma: a comparison of pal-pable and nonpalpable lesions. Surgery 1992;111:656 –9.

18. Miller RS, Adelman RW, Espinosa MH, Dorman SA, SmithDH. The early detection of nonpalpable breast carcinomawith needle localization. Experience with 500 patients in acommunity hospital. Am Surg 1992;58:193– 8.

19. Opie H, Estes NC, Jewell WR, Chang CH, Thomas JA, EstesMA. Breast biopsy for nonpalpable lesions: a worthwhileendeavor? Am Surg 1993;59:490 –3.

20. Knutzen AM, Gisvold JJ. Likelihood of malignant disease for

various categories of mammographically detected, nonpal-pable breast lesions. Mayo Clin Proc 1993;68:454 – 60.

21. Schwartz GF, Carter DL, Conant EF, Gannon FH, Finkel GC,Feig SA. Mammographically detected breast cancer. Non-palpable is not a synonym for inconsequential. Cancer 1994;73:1660 –5.

22. Sailors DM, Crabtree JD, Land RL, Rose WB, Burns RP,Barker DE. Needle localization for nonpalpable lesions. AmSurg 1994;60:186 –9.

23. Perdue PW, Galbo C, Ghosh BC. Stratification of palpableand nonpalpable breast cancer by method of detection andage. Ann Surg Oncol 1995;2:512–5.

24. Rosenberg RD, Lando JF, Hunt WC, Darling RR, WilliamsonMR, Linver MN, et al. The New Mexico MammographyProject. Screening mammography performance in Albu-querque, New Mexico, 1991 to 1993. Cancer 1996;78:1731–9.

25. Coveney EC, Geraaghty JG, O’Laoide R, Hourihane JB,O’Higgins NJ. Reasons underling negative mammography inpatients with palpable breast cancer. Clin Radiol 1994;49:123–5.

26. Elmore JG, Wells CK, Lee CH, Howard DH, Feinstein AR.Variability in radiologists’ interpretations of mammograms.N Engl J Med 1994;331:1493–9.

27. Beam CA, Layde PM, Sullivan DC. Variability in the inter-pretation of screening mammography by US radiologists.Findings from a national sample. Arch Intern Med 1996;156:209 –13.

28. Elmore JG, Wells CK, Howard DH. Does diagnostic accuracyin mammography depend on radiologists’ experience? JWomens Health 1998;7:443–9.

29. Jones TL. The worst list. Breast cancer now leading source ofmedical malpractice claims. Tex Med 1996;92:34 – 6.

30. Khalkhali I, Mena I, Diggles L. Review of imaging techniquesfor the diagnosis of breast cancer: a new role of pronescintimammography using technetium-99m sestamibi. EurJ Nucl Med 1994;21:357– 62.

31. Adler DD, Wahl RL. New methods for imaging the breast:techniques, findings, and potential. AJR Am J Roentgenol1995;164:19 –30.

32. Sabel M, Aichinger H. Recent developments in breast imag-ing. Phys Med Biol 1996;41:315– 68.

33. Bassett LW. Imaging the breast. In: Holland JF, Frei E III,Bast RC Jr., Kufe DW, Morton DL, Weichselbaum RR, edi-tors. Cancer medicine. 4th ed. Baltimore: Williams &Wilkins, 1997;589 –96.

34. Khalkhali I, Cutrone JA, Mena I. Scintimammography: thecomplementary role of Tc-99m sestamibi prone breast im-aging for the diagnosis of breast carcinoma. Radiology 1995;196:421– 6.

35. Scopinaro F, Schillaci O, Ussof W, Nordling K, Capoferro R,De Vincentis G, et al. A three center study on the diagnosticaccuracy of 99mTc-MIBI scintimammography. AnticancerRes 1997;17:1631– 4.

36. Palmedo H, Biersack HJ, Lastoria S, Maublant J, Prats E,Stegner HE, et al. Scintimammography with technetium-99m methoxyisobutylisonitrile: results of a prospective Eu-ropean multicentre trial. Eur J Nucl Med 1998;25:375– 85.

37. Tiling R, Khalkhali I, Sommer H, Linke R, Moser R, Willem-sen F, et al. Limited value of scintimammography and con-trast-enhanced MRI in the evaluation of microcalcificationdetected by mammography. Nucl Med Commun 1998;19:55–62.

38. Salvatore M, Del Vecchio S. Dynamic imaging: scintimam-mography. Eur J Radiol 1998;(Suppl 2):S259 – 64.

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 113

39. Taillefer R. The role of 99mTc-sestamibi and other conven-tional radiopharmaceuticals in breast cancer diagnosis. Se-min Nucl Med 1999;29:16 – 40.

40. Goldenberg DM, DeLand F, Kim E, Bennett S, Primus FJ, vanNagell JR Jr., et al. Use of radiolabeled antibodies to carci-noembryonic antigen for the detection and localization ofdiverse cancers by external photoscanning. N Engl J Med1978;298:1384 – 8.

41. Goldenberg DM, Kim EE, DeLand FH, Bennett SJ, Primus FJ.Radioimmunodetection of cancer with radioactive antibod-ies to carcinoembryonic antigen. Cancer Res 1980;40:2984 –92.

42. Goldenberg DM, Larson SM. Radioimmunodetection incancer identification. J Nucl Med 1992;33:803–14.

43. Gold P, Freedman SO. Specific carcinoembryonic antigensof the human digestive system. J Exp Med 1965;122:467– 81

44. Gold P, Goldenberg NA. The carcinoembryonic antigen(CEA): past, present, and future. McGill J Med 1997;3:46 – 66.

45. Heyderman E, Neville M. A shorter immunoperoxidasetechnique for the demonstration of carcinoembryonic anti-gen and other cell products. J Clin Pathol 1977;30:138 – 40.

46. Wahren B, Lindbrink E, Wallgren A, Eneroth P, ZajicekJ. Carcinoembryonic antigen and other tumor markers intissue and serum or plasma of patients with primary mam-mary carcinoma. Cancer 1978;42:1870 – 8.

47. Walker RA. Demonstration of carcinoembryonic antigen inhuman breast carcinomas by the immunoperoxidase tech-nique. J Clin Pathol 1980;33:356 – 60.

48. Kleist von S, Wittekind C, Sandritter W, Gropp H. CEA pos-itivity in sera and breast tumor tissues obtained from thesame patients. Pathol Res Pract 1982;173:390 – 401.

49. Bocker W, Schweikhart G, Pollow K, Kreienberg R, KlaubertA, Schroder S, et al. Immunohistochemical demonstrationof carcinoembryonic antigen (CEA) in 120 mammary carci-nomas and its correlation with tumor type, grading, stagingplasma-CEA, and biochemical receptor status. Pathol ResPract 1985;180:490 –7.

50. Chung JK, Jang JJ, Lee DS, Lee MC, Koh CS. Tumor concen-tration and distribution of carcinoembryonic antigen mea-sured by in vitro quantitative autoradiography. J Nucl Med1994;35:1499 –505.

51. Esteban JM, Felder B, Ahn C, Simpson JF, Battifora H,Shively JE. Prognostic relevance of carcinoembryonic anti-gen and estrogen receptor status in breast cancer patients.Cancer 1994;74:1575– 83.

52. Schmitt FC, Andrade L. Spectrum of carcinoembryonic an-tigen immunoreactivity from isolated ductal hyperplasias toatypical hyperplasias associated with infiltrating ductalbreast cancer. J Clin Pathol 1995;48:53– 6.

53. Imayama S, Mori M, Ueo H, Nanbara S, Adachi Y, Mimori K,et al. Presence of elevated carcinoembryonic antigen onabsorbent disks applied to nipple area of breast carcinomapatients. Cancer 1996;78:1229 –34.

54. Mori M, Mimori K, Inoue H, Barnard GF, Tsuji K, Nanbara S,et al. Detection of cancer micrometastases in lymph nodesby reverse transcriptase-polymerase chain reaction. CancerRes 1995;55:3417–20.

55. Salmon MA, Jurdan M, Brasseur F, Feyens AF, Symann M.CEA RT-PCR is a sensitive method to detect breast tumorcells in peripheral blood and in progenitor cell collectionsprior to and following CD341 cell selection [abstract]. Blood1997;90(Suppl):354b.

56. Kim EE, Corgan RL, Casper S, Primus FJ, Spremulli E, EstesN, et al. Axillary lymphoscintigraphy in radioimmunodetec-

tion of carcinoembryonic antigen in breast cancer. J NuclMed 1979;20:1243–50.

57. Kairemo KJA. Immunolymphoscintigraphy with 99mTc-la-beled monoclonal antibody (BW431/26) reacting with car-cinoembryonic antigen in breast cancer. Cancer Res 1990;50:949 –54.

58. Thor AD, Edgerton SM. Monoclonal antibodies reactive withhuman breast or ovarian carcinoma: in vivo applications.Semin Nucl Med 1989;19:295–308.

59. Lind P, Smola MG, Lechner P, Ratschek M, Klima G, Kol-tringer P, et al. The immunoscintigraphic use of Tc-99m-labeled monoclonal anti-CEA antibodies (BW431/26) in pa-tients with suspected primary, recurrent and metastaticbreast cancer. Int J Cancer 1991;47:865–9.

60. Nabi HA. Antibody imaging in breast cancer. Semin NuclMed 1997;27:30 –9.

61. Baum RP, Brummendorf TH. Radioimmunolocalization ofprimary and metastataic breast cancer. Q J Nucl Med 1998;42:33– 42.

62. Goldenberg DM, Nabi HA. Breast cancer imaging with ra-diolabeled antibodies. Semin Nucl Med 1999;29:41– 8.

63. Hansen HJ, Sullivan CL, Sharkey RM, Goldenberg DM.HAMA interference with murine monoclonal antibody-based immunoassays. J Clin Immunoassay 1993;16:294 –9.

64. Hansen HJ, Jones AL, Sharkey RM, Grebenau R, BlazejewskiN, Kunz A, et al. Preclinical evaluation of an “instant” 99mTc-labeling kit for antibody imaging. Cancer Res 1990;50:794s–8s.

65. American College of Radiology. Breast imaging reportingand data system (BI-RADS) 3rd ed. Reston, VA: AmericanCollege of Radiology, 1998.

66. Dupont WD, Parl FF, Hartmann WH, Brinton LA, WinfieldAC, Aorrell JA, et al. Breast cancer risk associated with pro-liferative breast disease and atypical hyperplasia. Cancer1993;71:1258 – 65.

67. Heywant-Kobrunner SH, Scheer I, Dershaw DD. Diagnosticbreast imaging. mammography, sonography, magnetic res-onance imaging and interventional procedures. Chapter 9.New York: Thieme Stutgart, 1997.

68. Kerlikowske K, Grady D, Barclay J, Sickles EA, Ernster V.Likelihood ratios for modern screening mammography. Riskof breast cancer on age and mammographic interpretation.JAMA 1966;276:39 – 43.

69. Simel DL, Samsa GP, Matchar DB. Likelihood ratios withconfidence: sample size estimation for diagnostic test stud-ies. J Clin Epidemiol 1991;44:763–70.

70. Simel DL, Feussner JR, Delong ER, Matchar DM. Interme-diate, indeterminate, and uninterpretable diagnostic testresults. Med Decis Making 1987;7:107–14.

71. American Joint Committee on Cancer. Manual for staging ofcancer. Philadelphia: J.B. Lippincott, 1992

72. Brown ML, Houn F, Sickles EA, Kessler LG. Screening mam-mography in community practice: positive predictive valueof abnormal findings and yield of follow-up diagnostic pro-cedures. AJR Am J Roentgenol 1995;165:1373–7.

73. Centers for Disease Control and Prevention. Cancer screen-ing behaviors among U.S. women: breast cancer, 1987–1989,and cervical cancer, 1988 –1989. MMWR Morb Mortal WklyRep 1992;41(SS-2):17–34.

74. Beam CA, Layde PM, Sullivan DC. Variability in the inter-pretation of screening mammograms by US radiologists.Findings from a national sample. Arch Intern Med 1996;156:209 –13.

114 CANCER July 1, 2000 / Volume 89 / Number 1

75. McCarty BD, Yood MU, Boohaker EA, Ward RE, Rebner M,Johnson CC. Inadequate follow-up of abnormal mammo-grams. Am J Prev Med 1996;12:282– 8.

76. Moffat F. Pinsky CM, Hammershaimb L, Petrelli NJ, Patt YZ,Whaley ES, et al. Clinical utility of external immunoscintig-raphy with the IMMU-4 technetium-99m Fab9 antibodyfragment inpatients undergoing surgery for carcinoma ofthe colon and rectum: results of a pivotal, phase III trial.J Clin Oncol 1996;14:2295–305.

77. Wegener WA, Petrelli N, Serafini A, Goldenberg DM. Safety and

efficacy of Arcitumomab imaging in colorectal cancer follow-ing repeat administrations. J Nucl Med. 2000;41:1016–20.

78. Tolmos J, Cutrone JA, Wang B, Vargas HI, Stuntz M, MishkinFS, et al. Scintimammographic analysis of nonpalpablebreast lesions previously identified by conventional mam-mography. J Natl Cancer Inst 1998;90:846 –9.

79. Cutrone JA, Yospur LS, Khalkhali I, Tolmos J, Devito A,Diggles L, et al. Immunohistologic assessment of techne-tium-99m-MIBI uptake in benign and malignant breast le-sions. J Nucl Med 1998;39:449 –53.

CEA Immunoscintigraphy in Breast Carcinoma/Goldenberg et al. 115