Comparison of a Novel Assay for Breast Cancer Mucin to CA15-3 and Carcinoembryonic Antigen By Lee Daly, Joann Ferguson, Gary P. Cram, Jr, Vera Hars, Stephen L.George, Kenneth S. McCarty, Jr, and Robert C. Bast, Jr Purpose: To compare the sensitivity and specificity of an automated microparticle enzyme immunoassay (MEIA) for breast cancer mucin (IM. BCM; Abbott Laboratories, North Chicago, IL) to that of CA15-3 and carcinoembryonic antigen (CEA) for detecting and monitoring breast cancer. Materials and Methods: IMxBCM was compared to assays of CA15-3 and CEA in 630 serum specimens from healthy women, and from women with breast cancer, other malignancies, benign breast conditions, or other benign diseases. Results: Analysis of the log-transforms for the three markers in all specimens showed a high correlation of IMxBCM with CA15-3 (r = .78), but not with CEA (r = .25). Based on a receiver-operating-characteristics (ROC)-curve analysis for any given specificity, IMxBCM was found to be a more sensitive marker than either CA15-3 or CEA for distinguishing 105 women with advanced or metastatic breast cancer from 89 healthy women (P = .003 and P =

ONE

OF NINE WOMEN in the United States develops breast cancer, and at least 30% of the women who have breast cancer die from the disease. 1,2 Up to 70% of patients with metastatic disease will respond to systemic treatment with cytotoxic drugs or hormones. In many patients, metastatic disease cannot be measured readily. In this setting, serum markers that reflect tumor burden could prove useful. Currently available serum markers for breast cancer are not sensitive enough to detect a minimal tumor burden or specific enough to facilitate early diagnosis. However, several markers have proven useful as an adjunct for monitoring breast cancer patients. The most widely studied serum markers have been carcinoembryonic antigen (CEA) and CA15-3. CEA is an antigenic glycoprotein of 200 kD that was first found in gastrointestinal cancers, but subsequently was detected in other malignancies. Levels of CEA are elevated in sera levels from heavy smokers and from patients with several different malignant and benign diseases. The sensitivity of CEA for identifying patients with advanced breast carcinoma ranges up to 60%, but elevations of the antigen are often modest. 3 Although CEA levels may reflect disease progression or regression, recent literature suggests that the changes in CEA

are not consistent enough to be used effectively in 4

clinical practice. CA15-3 is a double-determinant radioimmunometric assay that detects a breast cancer-associated mucin of

.04, respectively), from 98 women with benign breast conditions (P = .02 and P = .002), or from 191 women with benign diseases (P = .03 and P < .0001). At 95% specificity, the sensitivities of IMxBCM, CA15-3, and CEA for detecting advanced or metastatic breast cancer were 69%, 51%, and 30%, respectively. Serial serum samples (n = 177) were analyzed in 20 additional metastatic breast cancer patients with measurable disease. Serial IMxBCM levels corresponded with the clinical course of disease in 80%, CA15-3 in 65%, and CEA in 60%of the 20 patients. Conclusions: Increased sensitivity of IMxBCM, despite a high correlation with CA15-3, suggests that IMxBCM and CA15-3 may recognize distinct epitopes on the same molecule. Although further research is indicated, IMxBCM may provide a promising marker in the clinical management of breast cancer patients. J Clin Oncol 10:1057-1065. o 1992 by American Society of Clinical Oncology.

300 to 450 kD that bears epitopes recognized by two murine monoclonal antibodies, 115-D8 and DF3. 115-D8 was obtained after mice were immunized with human milk fat globule membrane, and DF3 was derived after mice were immunized with a membrane-enriched fraction of human breast cancer. CA15-3 is elevated in 61% to 91% of advanced breast cancer patients, 3 and CA15-3 has been found to be more sensitive than CEA in this setting.5 CA15-3 has also been reported to be more sensitive than CEA in detecting recurrences of breast cancer.

6

Here we report on a newly developed microparticle enzyme immunoassay (MEIA) for the quantitative determination of breast cancer mucin (BCM) in serum (IMxBCM; Abbott Laboratories, North Chicago, IL). The monoclonal antibody M85 recognizes N-acetyllactosamine (LacNac) oligosaccharides that are bound to core peptide of BCM by O-glycosidic linkages. Antigen is localized at the apical cell surface in normal breast epithelium and benign breast conditions, whereas anti-

From the Departments of Medicine, Community and Family Medicine, and Pathology, and the Duke Comprehensive Cancer Center, Duke University MedicalCenter, Durham, NC. Submitted August 5, 1991; accepted February13, 1992. Supportedby Abbott Laboratories,North Chicago, IL 60064. Address repnnt requests to Robert C. Bast, Jr,MD, PO Box 3843, Duke University Medical Center,Durham,NC 27710. C 1992 by American Society of ClinicalOncology. 0732-183X/9211007-0006$3.00/0

Journal of Clinical Oncology, Vol 10, No 7 (July), 1992: pp 1057-1065

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1057

1058

DALY ET AL

gen is found both at the apex and within the cytoplasm in malignant epithelium. 7 We have compared the sensitivity and specificity of the IMxBCM assay with that of CA15-3 and CEA for detecting and monitoring breast cancer. MATERIALS AND METHODS A total of 807 serum samples were collected from 646 women who were evaluated at the Duke University Medical Center. Some 630 samples were collected from donors only on a single occasion, whereas 177 were obtained from a group of 20 donors on different dates. Sera were obtained from breast cancer patients under protocols approved by the Duke University institutional review board. Other serum specimens included surplus samples of at least 1 mL that remained after standard blood tests were performed. Female patients were identified at Duke University Medical Center by computer-generated outpatient lists, admission sheets, and operating room schedules. A total of 1,640 specimens were collected, and 833 samples were omitted primarily because of an inadequate amount of serum, but also because of a failure to meet the inclusion criteria of the study. Sera were stored at -80°C until the assays were performed. Samples were coded, and the individuals who assayed the specimens were unaware of the clinical diagnoses. Clinical information was obtained retrospectively by chart review and evaluation of radiologic studies. Characteristics of the patients in different groups are listed in Table 1. Date of birth, race, menopausal status, last menstrual period, and smoking history were recorded. In cancer patients, the date of diagnosis, site of primary tumor, histologic type, grade, and stage were noted. Further clinical information was obtained for breast cancer patients, which included date and type of surgery, Tumor-NodeMetastasis classification at diagnosis, and dates and sites of involvement with metastatic disease. All patient groups, except for the small inactive breast cancer group, were matched for race and smoking history. Healthy individuals and patients with benign disease were younger and more likely to be premenopausal than those patients with cancer. A total of 630 samples were obtained on a single occasion from 89 healthy individuals, 98 patients with benign breast conditions, 191 with other benign diseases, 139 with breast cancer, and 92 with other malignant diseases. Patients with benign diseases included 50 individuals with autoimmune disease, 22 with infectious disease, 33 with nonmalignant lung conditions, 16 with cirrhosis, 19 with hepatitis, 25 with renal disease, and 26 with renal failure. Patients

with benign breast conditions included individuals with benign masses, fibroadenomas, and fibrocystic changes. Patients with malignant diseases included 21 patients with colorectal cancer, 21 with lung cancer, 13 with cervical cancer, 11 with endometrial cancer, and 26 with ovarian cancer. In addition to 139 patients with clinically apparent breast cancer, there were 21 serum specimens from patients who had a history of breast cancer and in whom there was no evidence of disease at the time blood was drawn. Thirty-four of 139 specimens from patients with clinically apparent breast disease were obtained from patients before primary breast surgery. This group included 12 stage I patients, 14 stage II patients, and eight stage III patients. One-year follow-up was obtained on 34 preoperative patients to determine the clinical course of their breast cancer. In patients with metastatic disease, 105 serum specimens were evaluated. An additional 177 samples were obtained on different occasions from 20 breast cancer patients with measurable disease who were undergoing treatment. Clinical course was assessed retrospectively on the basis of the radiologic studies and physical examinations that were performed at the time sera were drawn. The criteria for clinical assessment were based on the International Union Against Cancer guidelines.8 Disease progression was defined as the appearance of a new lesion or an increase of > 25% in the diameter of all lesions measured. Partial response required a 2 50% decrease in all measurable lesions. A complete response required the disappearance of all clinically detectable disease. In mixed responses, some lesions progressed, whereas others regressed. Disease was considered stable when there was a less than 50% decrease or a less than 25% increase in measurable disease. Changes were calculated according to the following formula: % change antigen -

Agm -

Agi Ag,

Agi

x 100.

Agm represents the antigen level drawn at the time of maximal change, and Agi represents the initial level of the serum marker. Correlation with disease course constituted at least a 25% increase in antigen levels with progression, a 50% decrease in levels with regression, and a less than 25% increase or less than 50% decrease with stable disease.5,9

MarkerAssays All serum samples were assayed twice for CEA, CA15-3, and IMxBCM. The mean of duplicate determinations was applied for each assay. CEA was measured by an enzyme-linked immunoassay (CEA-EIA monoclonal one step; Abbott Laboratories), CA15-3 by

Table 1. Patient Characteristics

Subjects Normal (t = 89) Active breast cancer (t = 139) (clinical evidence of disease) Inactive breast cancer (t = 21) (no evidence of disease) Benign breast (t = 98) (including fibroadenoma) Benign diseases (t = 191) Other malignancies (t = 92)

Median Age (years)

Pre-

43 52

57 35

3 6

33 56

7 3

78 73

52

33

0

57

10

43

62

4

30

52 62

36 8

1 0

54 88

Other

Current Smokers (%)

18 23

4 4

17 19

90

10

0

29

4

84

14

2

19

9 4

68 77

30 20

2 3

20 17

Menopausal Status (%) PeriPostUnknown

White

Race (%) Black

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1059

IMxBCM SERUM MARKER FOR BREAST CANCER radioimmunoassay (CA15-3 RIA; Centocor, Malvern, PA), and IMxBCM by automated microparticle enzyme immunoassay for use with the IMx analyzer (Abbott Laboratories). The IMxBCM assay used BCM-coated microparticles in a two-step competitive inhibition format. IMxBCM has been shown to be a highly reproducible 10 assay that can detect elevated levels of BCM in serum. The intraassay coefficient of variation (CV) ranged from 2.8% to 6.8% and interassay CV ranged from 4.2% to 11.1% in the reproducibility analysis of five samples, in replicates of three, in 10 runs, and in four different laboratories. Internal standards for each kit were within the manufacturer's specified range. Of the 10 CA15-3 kits used for these studies, three subsequently were recalled by the manufacturer for a possible degradation of the antigen standards. Internal controls for the three kits that were recalled (mean, 42.26 + SE = 1.15) did not differ significantly from the internal standard (mean, 43.66 + SE = 1.29) of those that were not (P = .59; analysis of variance, repeated measures).

StatisticalAnalysis Descriptive statistics were performed on the assay results of CEA, CA15-3, and IMxBCM, and on the logarithmic transformations of these assay results. The descriptive statistics were calculated for each of the positive disease groups (all breast cancer patients, primary or preoperative breast cancer patients, and advanced or metastatic breast cancer patients), and for each of the five control groups (healthy women and women with benign breast conditions, with other benign diseases, with inactive breast cancer and with malignant diseases other than breast cancer). The sensitivities of the three markers at 95% and 99% specificities and their 95% confidence intervals (CIs) also were estimated for each of the five control groups. For each two of the three markers the nonparametric correlation coefficients and their 95% CIs were estimated, and the bivariate scatter plots were drawn by using all of the data from the 630 patients. The areas under the three correlated receiver-operatingcharacteristics (ROC) curves were compared by the nonparametric approach that has been suggested by DeLong et al." The method used the theory developed for generalized U-statistics and applied the method of structural components to estimate the elements of the variance-covariance matrix. The resulting test statistic has an asymptotic x 2 distribution. The statistical analysis was performed by the SAS/IML program that was written by DeLong et al. Comparisons were performed among the groups of breast cancer patients and each of the five control groups. Empiric ROC curves for the 15 comparisons were plotted. ROC curves reflect both the sensitivity and specificity of each marker for different serum concentrations. This form of analysis minimized the potential bias that was introduced by choosing a single threshold for positivity.

RESULTS Levels of IMxBCM, CA15-3, and CEA were obtained for each of the 630 serum samples. Log-transforms of marker levels were analyzed by estimating the Spearman rank correlation coefficients and their 95% CIs. IMxBCM values correlated with CA15-3 (r = .78), but not with CEA (r = .25) (Fig 1). IMxBCM, CA15-3, and CEA were evaluated by ROCcurve analysis in which sensitivity and specificity were calculated for different levels of each marker by using

Ci j ,-

94

*

±

S

4

8

Log IMxBCM

Fig 1. Scatterplot for the log-transformed marker values of CA1 5-3 and IMBCM obtained with 630 single serum specimens (r = .78).

different patient groups as true positives and true negatives. In our initial analysis, the sensitivity and specificity of each marker was calculated to distinguish all 139 breast cancer patients with clinically active disease from several different control groups (Fig 2). IMxBCM had a greater sensitivity and specificity than CA15-3 in distinguishing clinically active breast cancer patients from apparently healthy women (Fig 3; P = .002), patients with benign breast disease (Fig 4; P < .001), and patients with breast cancer without evident disease (data not shown; P = .012). IMxBCM also had a greater sensitivity and specificity than CEA in distinguishing clinically active breast cancer patients from patients with benign breast conditions (Fig 4; P = .003) and other benign diseases (data not shown; P < .001). When breast cancer patients who had active disease were compared with patients who had other cancers, similar sensitivities and specificities were obtained with all three markers. Among the 139 patients who had clinically evident breast cancer, 34 were studied preoperatively with stage I to III disease and 105 either had advanced metastatic breast cancer on presentation or had recurrent disease. In our next ROC analyses, sensitivities and specificities for each marker were calculated to distinguish the 105 patients with advanced metastatic disease from each of the different control groups. IMxBCM was more sensitive and more specific than either CA15-3 or CEA for distinguishing patients with advanced metastatic breast cancer from healthy controls (Fig 5; P = .003 and P = .04, respectively), women with benign breast conditions (Fig 6; P = .02 and P = .002), and women with benign diseases (Fig 7; P = .03 and P < .0001). IMxBCM also was more sensitive and specific than CA15-3

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1060

DALY ET AL IMXBCM

CA15-3

CEA '^^^^ 100uuu00

10000

0.

5000

5000

1000

1000

500

500 0

100

a

100

J.

Fig 2. IM.BCM, CA15-3, and CEA values for breast cancer patients (n = 139), healthy women (n = 89), and patients with benign breast conditions (n = 98), benign diseases (n = 191), and malignancies other than breast cancer (n = 92). The numbers in the boxes represent patients with levels below standard thresholds.

,so *

50

.

,

S

ii: 10

05.

n 0

0 64* 1 * '1 "e CF.,P

(P = .02), but not CEA (P = .22) in distinguishing advanced metastatic breast cancer patients from the 21 patients with clinically inactive disease. When advanced metastatic breast cancer patients were compared with patients who had other malignancies, IMxBCM had no greater sensitivity nor specificity than CA15-3, but it was superior to CEA (P = .014). When the data were analyzed and the thresholds recommended by the manufacturers used (Table 2), abnormal values of IMxBCM were found in 8% of the apparently healthy controls, CA15-3 in 15%, and CEA in 13%. Consequently, marker values also were calculated that would exclude 95% and 99% of the apparently healthy individuals in our own series (Table 2). At those thresholds, little marker activity was found in the sera from patients who had benign breast disease or a history of breast cancer in the absence of clinically evident metastases. By using a threshold that would exclude 95% of the apparently healthy women, 65% of patients with advanced metastatic disease had elevated IMxBCM, 51% had elevated CA15-3, and 30% had elevated CEA. When different sites of involvement were analyzed, all three markers were more frequently and more markedly elevated in patients who had metastases confined to the

1.0

0.9 0.

0.8 -

2 2

0.0 as

V

0.

0.0

0.0

0.1

02

O

0 a

0.5

0.7

0.8

0.9

1.0

-

-

3421

-

7.40

Ma CEAI QMS-3

am 351 ......... U.1

lass -

-

-

-

-

-

-

-

-

-

-

-

-

-

Fig 3. Observed sensitivities and specificities of IM,BCM, CA15-3, and CEA as markers for distinguishing active breast cancer patients from healthy women (P = .002 for IM.BCM v CA15-3, and P = .09 for IM/BCM v CEA). Ranges of assay values for healthy women are indicated.

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1061

IMxBCM SERUM MARKER FOR BREAST CANCER t1.0-

1.0-

0. -

Kz-. 0O-

G7

07-

N-' 0.-

0.4-

M4

0.3-

0.3

02-

02

0.1

0.1

0.0

0.0

00

0.1

0.2

03

0.4

0

s

0.87 .7

0.9

1.0

00

0.2

0.1

0.3

0.4

0.5

0

.7

0.

0.9

1.o

specity ui~i

sa-m

7.tA

CEA-

0.01 -

=8_s3

70

-

-----------

-

------------------------

710

Fig 4. Observed sensitivities and specificities of IM/BCM, CA15-3, and CEA as markers for distinguishing active breast cancer patients from women with benign breast conditions (P = .001 for IM/BCM v CAl5-3, and P = .003 for IM.BCM v CEA). Ranges of values for benign breast patients are indicated.

1.0 0.0

"K'

a0.

C7

0.6

\ ' N

0.3

0.2

0,1

' '

I

I

I

00

01

02

03

04

0,5

o06

0.7

, 11 , ,

, i ,

0.

0.9

1.0

0p ..

...

...

CEA

0.01-

CM1S-3

31---

-

-

----

-

-

-

-

-

-

-

-.

40

Fig 5. Observed sensitivities and specificities of IM/BCM, CAl5-3, and CEA as markers for distinguishing advanced breast cancer patients from healthy women (P = .003 for IM8BCM v CA15-3, and P = .04 for IMKBCM v CEA). Ranges of assay values for healthy women are indicated.

"meam 7.18 CA-

001--

0616-3

740------

430 --

--

--

--

---------------

-

-11.01 7---T

Fig 6. Observed sensitivities and specificities of IMBCM, CA15-3, and CEA as markers for distinguishing advanced breast cancer patients from patients with benign breast conditions (P = .02 for IMBCM v CA15-3, and P = .002 for IM.BCM v CEA). Ranges of assay values for benign breast patients are indicated.

bone compared with patients with local and visceral disease (Table 3). By using thresholds that would exclude 95% of healthy donors, IMxBCM was elevated preoperatively in 32% of patients with stage I to III disease, CA15-3 in 24%, and CEA in 6%. When preoperative patients were analyzed according to surgical stage (Table 2), elevations of all three markers were observed in stage III disease more frequently than in stage I or stage II disease. Of the 34 preoperative patients studied, 21% had breast cancer recurrence within 1 year, and 12% died from breast cancer in the same interval. Among the seven women who relapsed, four had elevated serum IMxBCM at presentation, three had elevated CA15-3, and none had elevated CEA, when cutoffs were chosen at the 95th percentile. All four patients who relapsed and had elevated levels of at least one serum marker died of disease during the first year of follow-up. Of the 27 women who were free from clinically evident disease at 1 year after diagnosis, 26% had elevated IMxBCM, 19% had elevated CA15-3, and 7% had elevated CEA, preoperatively. An IMxBCM value of 2 35 U/mL was associated with a positive predictive value of 97% and a negative predictive value of 40% for distinguishing patients with advanced metastatic breast cancer from patients with

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1062

DALY ET AL 1.o

0.9 0.s 0.7

44

ol DO3 W

0.2 0.0 0.1

0o

01

0.2

03

0s5

0.4

0.

07

0,.

to0

0.9

specifimcty m

7.17

M

0.01

CM5-3

2.89 -

71lA

-

-

-

..--

--

- 21

-

-

454.0

---------------------

Fig 7. Observed sensitivities and specificities of I 2 BCM, CAl5-3, and CEA as markers for distinguishing advanced breast cancer patients from patients with benign diseases (P = .03 for IM.BCM v CA15-3, and P < .0001 for IM.BCM v CEA). Ranges of assay values for women with benign diseases are indicated.

inactive breast cancer in clinical remission. However, lower positive predictive values of 61% and of 67% were observed for distinguishing patients who had primary breast cancer from those who had benign breast disease or from apparently healthy women. Negative predictive values of 82% and 80% were observed in these two settings.

By using a threshold that would exclude 95% of healthy donors, benign disease was associated with elevated serum IMxBCM in 23% of patients, CA15-3 in 9%, and CEA in 8% (Table 4). IMxBCM was more markedly elevated than CA15-3 or CEA in renal disease, and CEA was more markedly elevated than CA15-3 or IMxBCM in renal failure. Elevated values of all three markers were observed in the sera from patients with colon, lung, cervical, endometrial, and ovarian cancer (Table 5). IMxBCM was elevated more frequently than CA15-3 and CEA in cervical and endometrial cancer. Both IMxBCM and CA15-3 were elevated more frequently than CEA in patients with ovarian cancer, and IMxBCM was the most sensitive marker of the three. However, the number of patients studied in each group was relatively small. When 177 serial serum specimens from 20 women with measurable metastatic breast cancer were analyzed as previously described, 5 changes in IMxBCM levels correlated with disease status in 80% of cases, CA15-3 in 65%, and CEA in 60% (Fig 8). However, an adjusted McNemar test was used, and comparisons between IMxBCM and CA15-3 (P = .25), IMxBCM and CEA (P = .22), and CA15-3 and CEA (P = 1.0) were not

statistically significant. Each patient had at least four serum specimens that spanned 6 to 24 months with a mean of 14 months. DISCUSSION

Our data indicate that IMxBCM is significantly more sensitive at any given specificity than CA15-3 or CEA for

Table 2. Comparison of Serum Values for IM,BCM, CA15-3, and CEA for Healthy Controls, Benign Breast Conditions, and Inactive Breast Cancer (%)* CEA (ng/mL) CA1 5-3 (U/mL) IMxBCM (U/mL) No. of ~95%> 7.0 ~99% > 30 > 3.0 Abbott > 75 40 ~99% 95% > > 30 > 45 Centocor ~99% > 40 ~95% > 35 Abbott Patients

Subjects Benign breast (including fibroadenoma) Inactive breast cancer (no evidence of disease) Normals Active breast disease (clinical evidence of disease) Advanced metastatic Preoperative Stage I 11 Stage 111 Stage

98

9

4

1

14

3

0

10

4

0

21

10

10

0

5

0

0

14

5

0

89 139

8 65

4 60

1 53

15 59

4 45

1 29

13 41

4 24

1 9

105 34 12 14 8

73 41 25 28 88

69 32 17 21 75

65 18 8 7 50

65 41 33 36 62

51 24 8 14 62

36 6 0 0 25

49 18 17 7 38

30 6 0 7 12

12 0 0 0 0

*Positivity is calculated for different thresholds including those recommended by the manufacturer and those that would exclude approximately 95% and 99% of healthy controls.

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1063

IMxBCM SERUM MARKER FOR BREAST CANCER Table 3. Comparison of Serum Values for IM.BCM, CA15-3, and CEA for Patients With Different Sites of Involvement by Metastatic Breast Cancer (%) Site of Involvement

No. of Patients

Abbott > 35

Only bone Only local Only visceral

19 10 16

95 70 69

IMKBCM (U/mL) -95% > 40 -99% > 45

84 70 62

84 60 56

CEA (ng/mL)

CAl5-3 (U/mL) Centocor > 30

-95% > 40

-99% > 75

Abbott > 3.0

-95% > 7.0

-99% > 30

100 50 62

74 40 44

53 30 38

63 50 44

63 40 38

21 10 19

distinguishing advanced or metastatic breast cancer patients from healthy women, women with benign breast conditions, and patients with other benign diseases by using ROC-curve analysis. In the evaluation of patients with both primary and metastatic breast cancer, IMxBCM remains more sensitive than CA15-3 in distinguishing active breast patients from healthy women (P = .002), women with benign breast conditions (P < .001), and women with a history of breast cancer and no evidence of disease (P = .012). The superior sensitivity of IMxBCM over CA15-3 and CEA is supported further by the values obtained when different thresholds are used to distinguish patients with clinically active breast cancer from healthy individuals and patients with benign breast disease (P < .001 and P = .004, respectively). Although IMxBCM proved to be superior to these other assays, the sensitivity and specificity of the test would not permit primary diagnosis of breast cancer in the absence of physical examination or mammography. A positive predictive value of 97% for distinguishing advanced metastatic breast cancer from inactive disease in remission suggests that IMxBCM deserves further evaluation as a trigger for a more extensive or more invasive workup of potential disease recurrence. However, inadvertent selection bias must be considered. Our normal population consisted of healthy women who presented at Duke University Medical Center for routine physical examinations, minor trauma, elective surgical procedures, and nonspecific complaints with a negative workup. Our normal and benign breast categories matched our breast cancer population with regard

to race and smoking history. Healthy controls and patients with benign breast disease were younger than patients with breast cancer or other forms of malignancy. Patients with benign diseases were comparable in age to breast cancer patients and to individuals with other malignancies. Yet 15% of 89 apparently healthy donors had a CA15-3 of more than 30 U/mL in contrast to other studies.5,12-23 Recent clinical data compiled by Centocor indicate that 6.6% of 1,000 apparently healthy individuals have a CA15-3 of more than 30 U/mL. The mean of 18.9 -t 7.1 U/mL for these 1,000 donors did not differ significantly from the mean of 21.0 -t 12.8 that was observed in our study (P = .123). Defects in the three recalled CA15-3 assay kits may have affected the results. Several months after completion of the study, Centocor recalled several CA15-3 kits for possible degradation of the antigen used to prepare the standard curves for each assay. Such degradation would have increased apparent values for CA15-3. However, in each of our assays internal standards were run and were similar for recalled and nonrecalled kits. These standards were within the range that was considered acceptable by the manufacturer and should have been elevated if the standard curve had been altered by antigen degradation. Only 24.6% of 807 samples had been assayed with the recalled CA15-3 kits. All patient subgroups were included in the batches assayed both with recalled and nonrecalled CA15-3 kits. Although possible defects in the CA15-3 assay might still have influenced the differences between IMxBCM and

Table 4. Comparison of Serum Values for IM.BCM, CA15-3, and CEA for Patients With Different Benign Diseases (%) CEA (ng/mL) CAl 5-3 (U/mL) (U/mL) IMKBCM No. of Patients Abbott > 35 -95% > 40 -99% > 45 Centocor > 30 -95% > 40 -99% > 75 Abbott > 3.0 -95% > 7.0 -99% > 30

Subjects Benign diseases Autoimmune disease Infectious disease Lung disease Hepatic cirrhosis Hepatitis Renal disease Renal failure

191 50 22 33 16 19 25 26

26 24 5 36 44 26 24 27

23 18 5 30 38 26 24 27

16 8 5 21 31 21 16 23

22 18 14 36 44 26 12 19

9 8 0 20 19 16 0 8

2 0 0 6 6 5 0 0

26 12 14 36 44 37 20 38

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

8 4 5 9 31 11 0 12

0 0 0 0 0 0 0 0

1064

DALY ET AL Table 5. Comparison of Serum Values for IM.BCM, CAl5-3, and CEA for Patients With Different Epithelial Cancers (%)

Subjects

No. of Patients

Abbott > 35

Other malignancies Colon Lung Cervical Endometrial Ovarian

92 21 21 13 11 26

49 33 57 54 36 58

IMBCM (U/mL) -95% > 40 -99% > 45

40 14 48 46 36 54

-95% > 40

-99% > 75

Abbott > 3.0

-95% > 7.0

-99% > 30

26 14 24 23 18 42

16 5 14 8 9 35

13 0 14 0 0 35

34 67 43 31 9 12

21 57 24 15 0 0

11 38 10 0 0 0

33 10 38 46 18 46

CA15-3, they would not have influenced the conclusion that IMxBCM is superior to CEA. Our study indicated that IMxBCM is not helpful in distinguishing patients with breast cancer from patients with other malignancies. IMxBCM was elevated more frequently than CA15-3 and CEA in all subgroups aside from the colon cancer group. IMxBCM is markedly higher in ovarian, endometrial, and cervical cancer subgroups when compared with CA15-3 and CEA. At a 95% specificity, the sensitivities of IMxBCM, CA15-3, and CEA for detecting ovarian cancer were 54%, 35%, and 0%, respectively, and for detecting cervical cancer were 46%, 8%, and 15%, respectively. This is consistent with the fact that M85 initially was generated against a mucin preparation purified from an ovarian cancer patient's ascites. Further evaluation of possible complementarity between IMxBCM and CA125 is indicated. IMxBCM also was elevated more frequently than CA15-3 and CEA in 21 patients with active lung cancer. There were nine women with small-cell, four with squamous cell, three with large-cell, two with bronchoalveolar, two with adenocarcinoma, and one with giantcell carcinomas. All patients had active lung cancer, and a majority had extensive disease when the serum specimens were obtained. Levels of IMxBCM and CA15-3 were correlated highly (r = .78), whereas IMxBCM and CEA were not (r = .25). The combination of IMxBCM with either CA15-3 or CEA provided no added benefit in detection. 1^u-

10,000

CEA (ng/mL)

CA1 5-3 (U/mL) Centocor > 30

The higher sensitivity of IMxBCM and the correlation between IMxBCM and CA15-3 is consistent with the possibility that IMxBCM and CA15-3 recognize different epitopes on a common breast mucin antigen. The monoclonal antibody M85 used in the IMxBCM recognizes a family of mucins that contain O-linked oligosaccharides with lactosamine structures. M85 reactivity is markedly enhanced after the removal of sialic acid through treatment with neuraminidase. M85 is thought to recognize a neutral oligosaccharide sequence with the primary target being a terminal Gal(1,4)GIcNAc. This sequence is present in i, I, and paragloboside blood group antigens. 24 The DF3 monoclonal antibodies used in the CA15-3 assay also react with a high molecular weight mucin. Sialyl oligosaccharides present on a peptide backbone may contribute to the determinant recognized by DF3, although the DF3 monoclonal antibody recognizes a peptide determinant that has been cloned from a DNA library. 25,26 Abe and Kufe 27 demonstrated that multiple monoclonal antibodies, including F36/22, 115-D8, and CA1, react with the mucin recognized by DF3. Serial measurements of antigen levels in patients with measurable metastatic disease show that changes in IMxBCM levels corresponded to disease status in 80%, CA15-3 in 65%, and CEA in 60%. Because of the small number of patients, this difference was not found to be statistically significant. As this was a retrospective analysis, a better assessment of the prognostic value or clinical

10,000i

E ,ooo000

C

S10oi

100

so 10

10

Progressed

A

Progressed

Regressed

B

Regressed

C

Progressed

Regressed

Fig 8. Comparison of changes in (A) IMBCM, (B) CA15-3, and (C) CEA values for patients (1 to 20) with measurable metastatic breast cancer tracked over time. Initial antigen level and level at time of maximal clinical change are plotted.

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

1065

IMxBCM SERUM MARKER FOR BREAST CANCER use of IMxBCM may be provided by a prospective trial with a larger number of patients. IMxBCM clearly deserves further evaluation as a marker for monitoring patients with breast cancer.

ACKNOWLEDGMENT We thank Craig Beam, PhD, for consultation regarding statistical analysis, Hector Rivera for outstanding technical assistance, and Jana Carey and Karen Cash for excellent secretarial assistance.

REFERENCES 1. Cancer Facts & Figures-1991. Atlanta, GA, American Cancer Society, 1991 2. Boring CC, Squires TS, Tong T: Cancer statistics 1991. A professional education publication. Ca-A Cancer J Clin 41:19-36, 1991 3. Tondini C, Hayes DF, Kufe DW: Circulating tumor markers in breast cancer. Hematol Oncol Clin North Am 3:653-674, 1989 4. Loprinzi CL, Tormey DC, Rasmussen P, et al: Prospective evaluation of carcinoembryonic antigen levels and alternating chemotherapeutic regimens in metastatic breast cancer. J Clin Oncol 4:46-56, 1986 5. Hayes DF, Zurawski VR Jr, Kufe DW: Comparison of circulating CA15-3 and carcinoembryonic antigen levels in patients with breast cancer. J Clin Oncol 4:1542-1550, 1986 6. Safi F, Kohler I, R6ttinger E, et al: Comparison of CA 15-3 and CEA in diagnosis and monitoring of breast cancer. Int J Biol Markers 4:207-214, 1989 7. Anderson B, Slota J, Kundu S, et al: Characterization of monoclonal antibodies to paragloboside (PG) and sialosyl-PG (2,6-SPG), and an improved chromatogram binding assay for rapidly identifying antibodies to tumor antigens. J Cell Biochem S11D:157, 1987 (suppl; abstr) 8. Hayward JL, Rubens RD, Carbone PP, et al: Assessment of response to therapy in advanced breast cancer. A project of the Programme on Clinical Oncology of the International Union Against Cancer, Geneva, Switzerland. Br J Cancer 35:292-298, 1977 9. Tondini C, Hayes DF, Gelman R, et al: Comparison of CA15-3 and carcinoembryonic antigen in monitoring the clinical course of patients with metastatic breast cancer. Cancer Res 48:4107-4112, 1988 10. Konrath JG, Przywara GL, Manderino DM, et al: Evaluation of IMxBCM automated microparticle enzyme immunoassay for breast cancer mucin. Clin Chem 36:1097, 1990 (abstr) 11. DeLong ER, DeLong DM, Clarke-Pearson DL: Comparirig the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 44:837-845, 1988 12. Gion M, Mione R, Dittadi R, et al: Evaluation of CA15/3 serum levels in breast cancer patients. J Nucl Med Allied Sci 30:29-36, 1986 13. Fujino N, Haga Y, Sakamoto K, et al: Clinical evaluation of an immunoradiometric assay for CA15-3 antigen associated with human mammary carcinomas: Comparison with carcinoembryonic antigen. Jpn J Clin Oncol 16:335-346, 1986

14. Pons-Anicet DM, Krebs BP, Mira R, et al: Value of CA 15-3 in the follow-up of breast cancer patients. Br J Cancer 55:567-569, 1987 15. Colomer R, Ruibal A, Salvador L: Circulating tumor marker levels in advanced breast carcinoma correlate with the extent of metastatic disease. Cancer 64:1674-1681, 1989 16. Colomer R, Sole LA, Navarro M, et al: CA 15.3: Early results of a new breast cancer marker. Anticancer Res 6:683-684, 1986 17. Sacks NP, Stacker SA, Thompson CH, et al: Comparison of mammary serum antigen (MSA) and CA15-3 levels in the serum of patients with breast cancer. Br J Cancer 56:820-824, 1987 18. Aaran RK, Kallioniemi OP, Oksa H, et al: Serum CA 15-3 in the diagnosis and follow-up of breast cancer. Tumor Biol 8:324, 1987 (abstr) 19. Bieglmayer C, Szepesi T, Neunteufel W, et al: MCA, a monoclonal-antibody-defined breast tumor-associated antigen and its relation to CA 15.3. Tumor Biol 10:232-242, 1989 20. Ruibal A, GenollI J, Rosell M, et al: Serum CA 15.3 levels in patients with non-tumoral diseases, and establishment of a threshold for tumoral activity. Results in 1219 patients. Int J Biol Markers 1:159-160, 1986 21. Barak V, Carlin D, Sulkes A, et al: CA15-3 serum levels in breast cancer and other malignancies-correlation with clinical course. Isr J Med Sci 24:623-627, 1988 22. Chu TM, Constatine R, Nemoto T: Serum level of cryptic tumor antigens in breast cancer patients as determined by two monoclonal antibodies (M85/F36) and its comparison with CA 15-3. J Clin Lab Anal 3:267-272, 1989 23. Stacker SA, Sacks NP, Golder J, et al: Evaluation of MSA as a serum marker in breast cancer: A comparison with CEA. Br J Cancer 57:298-303, 1988 24. Kinders R, Slota J, Patrick J, et al: An assay for cryptic tumor antigens in sera of women with breast cancer, in RL Ceriani (ed): Breast Cancer Immunodiagnosis and Immunotherapy. New York, NY, Plenum, 1989, pp 55-67 25. Hayes DF, Sekine H, Ohno T, et al: Use of murine monoclonal antibody for detection of circulating plasma DF3 antigen levels in breast cancer patients. J Clin Invest 75:1671-1678, 1985 26. Gum JR, Byrd JC, Hicks JW, et al: Molecular cloning of human intestinal mucin cDNAs. Sequence analysis and evidence for genetic polymorphism. J Biol Chem 264:6480-6487, 1989 27. Abe M, Kufe DW: Identification of a family of high molecular weight tumor-associated glycoproteins. J Immunol 139: 257-261, 1987

Downloaded from ascopubs.org by RB DRAUGHTON LIBRARY on April 28, 2019 from 131.204.073.184 Copyright © 2019 American Society of Clinical Oncology. All rights reserved.

Comparison of a novel assay for breast cancer mucin to CA15-3 and carcinoembryonic antigen.

To compare the sensitivity and specificity of an automated microparticle enzyme immunoassay (MEIA) for breast cancer mucin (IMx BCM; Abbott Laboratori...
724KB Sizes 0 Downloads 0 Views