- - - - - - - - - - - - - - - SYMPOSIUM ARTICLE - -
What do CA 125 and other antigens tell us about ovarian cancer biology? CHARLES
E. WELANDER
From the Gynecologic Oncology Laboratory, Institute for Gynecologic Oncology, Atlanta, GA, USA
Acta Obstet Gynecol Scand 1992; 71 Suppl ISS: 85-93
CA 125 is an antigenic determinant on a high molecular weight glycoprotein. A monoclonal antibody has been produced which recognizes this. and allows us to measure the expression of CA 125 in serum. Tissue distribution of the CA 125 determinant is most commonly seen in serous tumors of the ovary. with highest levels in borderline and frankly malignant serous cystadenocarcinomas. Occasionally cancers of the breast, gastrointestinal tract. and kidney will show elevated levels of CA 125. Normal tissues which show varying levels of CA 125 include decidual tissue and structures derived from celomicepithelium. CA 125 is clearly tumor associated, but not tumor specific. Quantitative correlation of CA 125 levels with tumor volume has not been demonstrated. This observation limits the clinical usefulness of CA 125 as a screening tool. particularly in premenopausal patients who do not have a diagnosis of ovarian cancer. An undetectable levelof CA 125 antigen does not rule out the presence of an early ovarian cancer. When CA 125 is used to monitor disease state in patients with known ovarian cancers (whose tumors do express CA 125), changes in levels of CA 125 do correlate with gross changes in tumor volume. Good prognostic significanceis attributed to a rapid decline in CA 125 levels following induction chemotherapy in patients with advanced ovarian cancer. However. an undetectable serum level of CA 125 does not predict clinical cure for a patient with ovarian cancer. Further clinical studies continue in the use of CA 125 as a screening tool and as a means to monitor treatment responses of known ovarian cancer.
Our ability to interpret and explain clinical observations concerning ovarian cancer (OC) is directly related to our understanding of the embryology, physiology, and biochemistry both of normal ovaries and of malignant changes. This is clearly a complex problem, which can become distorted when our interpretation of observations is overly simplified. Clinicians do desire to have an ideal marker substance that is easily detected in patients, and that can be positively correlated with the presence of malignant cells. An additional feature of such an ideal marker would be a quantitative relationship between marker expression and the number of viable tumor cells in a patient. Clinical applications of an ideal marker for the ovarian cancer problem include:
1) population screening for early detection of OC, 2) preoperative evaluation of patients with a pelvic mass, 3) correlation of tumor marker level with the cell type of an OC. 4) a means to assess response to surgery and/or chemotherapy, 5) a means to identify complete responders to therapy. Laboratory investigators have demonstrated that the above listed ideal features of a marker are not yet available in any of the markers identified to date. Nonetheless, considerable effort has been expended in attempts to determine how clinically useful the present group of tumor markers can be. In this review, a summary of clinical studies of CA
Acta Obstet Gynecol Scand Suppl/55
86
Ch. E. Welander
Table I. Serum CA 125 levels in patients with endometriosis distributed by severity of disease Author
Ref.
Upper limit CA 125 Ulml
Stage of endometriosis 1 + 11
Barbieri et al. (1986) Fedele et al. (1988) Guidice et al. (1986) Kauppilla et al. (1988) Patton et al. (1986) Pittaway and Fayez (1986)
(15) (16) (17) (18)
Takahashi et al. (1987)
(19)
(13) (14)
35 35 35 30 35 30 16 39
Total
III + IV
3/47
7113
2/38
7/19 7/8
16f72
11/15
2/22
3/15
8192 40/92
16138
1/6 Stage I + II 11.5%
30/38 14/15
Stage III + IV 50.4%
From Jacobs et al., 1989. 125 has been prepared. CA 125 has been more widely tested than the several other markers which are clinically available. The features of an ideal marker are more closely approximated by CA 125, although it is not tumor specific, nor does it correlate well with small numbers of tumor cells. The biochemical properties of CA 125 and tissue distribution of the antigen are summarized. Clinical studies are presented, which have tested the applicability of CA 125 measurements to each of the five areas noted above. Realizing the imperfect ability of CA 125 to be an 'ideal' marker, additional combined modalities for screening patients at high risk of developing OC have been tested. Whether complementary modalities can enhance the predictive capability has been explored in the clinic, and will be reported briefly in this paper.
Properties and tissue distribution of CA 125 The tumor marker CA 125 is an antigenic determinant on a high molecular weight glycoprotein. It is recognized by a monoclonal antibody, OC 125, which was first identified by Bast and colleagues in 1981 (I). OC 125 is a murine monoclonal antibody produced in BALB/c mice which were immunized with OC cells derived from a serous papillary cystadenocarcinoma. An immunoradiometric assay has been developed to measure CA 125 in serum (2) and other body fluids, including saliva (3) and peritoneal fluid (4, 5). Concentrations of CA 125 are expressed in arbitrary units per milliliter. The serum half-life of CA 125 is approximately 4.5 days (6). Immunohistochemical techniques have also been developed to identify the CA 125 determinant on Acta Obstet Gynecol Scand Suppl 155
cell surfaces (7). CA 125 is present on the cell membrane of fetal tissues derived from celomic epithelium. Adult tissues showing surface expression of CA 125 include normal endometrium, fallopian tube, and endocervix (8). The epithelial surface of normal ovary is not positive for CA 125 staining, but cells lining ovarian cysts as well as papillary excrescences within an ovarian cyst are often positive. Patients within the first trimester of pregnancy demonstrate high tissue levels of CA 125, the bulk of this coming from decidual tissue (9). A variety of malignancies have been screened for CA 125 expression, by both serum and immunoperoxidase staining techniques. Serous ovarian tumors, including benign, borderline, and malignant types, are strongly positive for CA 125. Mucinous, sex cord, germ-cell and Brenner tumors are usually negative (7). The pathologic grade reflecting tumor differentiation can influence CA 125 expression, although that has been variously reported by different investigators (10, 11). Carcinomas arising from tissues which normally express CA 125, such as endometrium, fallopian tube, and endocervix also continue to express the marker. Other non-gynecologic tumors which frequently express elevated levels of CA 125 include carcinomas arising in the lung, breast, gastrointestinal tract, and kidney (12). Since a variety of tissues beside OC can express CA 125, it is necessary to establish a normal range of values. Low levels of expression must be considered within a normal range, in order to increase the specificity of the test. Investigators have determined that a serum level of CA 125 up to 35 units/ml is considered normal. Population studies have shown that fewer than 1% of healthy control patients will have an elevated level at that cut-off point. Approximately 6% of patients with benign gynecologic pathology will have an elevated level, using 35 units/ml
CA 125 levels and ovarian cancer
87
Table II. Elevation of serum CA 125 (> 35 Vlml) in patients with endometrial cancer by clinicalstage of disease Author
Clinical stage of endometrial cancer
Ref.
I + II Duk et al. (1986) Fuith et al. (1987) Meier et al. (1987) Niloff et al. (1984a) Patsner et al. (1988)
(20) (21 ) (22) (23) (24)
III + IV
All stages
221100
5n
27/107 13/32
0/11
14/18
21/81
8/8
14/29 29/89
81.8%
31.8%
2110
22.4%
Total From Jacobs et al., 1989.
as a cut-off (2). A common reason for frequent elevations in serum levels of CA 125 among premenopausal patients is endometriosis. Various authors have summarized groups of patients having endometriosis, showing a general increase in serum levels with advancing stage of endometriosis (Table I) (13-19). Among malignancies arising from the endometrium, elevation of serum CA 125 to levels greater than 35 unitslml is also correlated with stage of disease (Table 11) (20-24). Serum levels of CA 125 are the most common clinical determinations made, although tissue levels can also be measured from normal or malignant samples (7.8). There are instances of tumors which produce high tissue levels without significant elevations of serum CA 125. In similar fashion, certain normal tissues, such as endometrium, decidual tissue. and some benign tumors have high cellular expression of CA 125, without shedding antigen into the serum. Correlation with malignancy seems to be
best demonstrated by identifying antigen shed into the serum, rather than looking for the presence of CA 125 in tissue samples (12).
Population screening for ovarian cancer using CA 125 The ideal tumor marker described in the Introduction to this paper would be one that is expressed in detectable quantities, even from small numbers of new OC cells. A new OC may express CA 125 on the tumor cell surface, but shedding of the antigen into the serum may not occur until the tumor has grown and/or spread to distant peritoneal sites. Early detection from a clinical standpoint may still be too late to surgically cure such patients. Nonetheless, we remain anxious to clinically detect OC as early as possible. The largest prospective OC screening study pub-
Table III. Specificity of a single screen of serum CA 125 levels at different reference values according to age and menopausal status Group
All: Age (years) 40-44 45-49 50-54 55-59 60-64 65-69 70+ Menopausal status PrePost-
N
Proportion of subjects below indicated reference value (Vlml) 20
25
30
35
65
100
1082
0.81
0.90
0.94
0.97
0.99
0.99+
257 226 189 187 137 50 36
0.71 0.76 0.79 0.89 0.93 0.92 0.97
0.85 0.85 0.87 0.96 0.97 0.98 1.00
0.92 0.89 0.93 0.97 0.99 0.98 1.00
0.95 0.95 0.97 0.98 0.99 0.98 1.00
0.99 0.99 0.98 0.99+ 0.99 0.98 1.00
0.99 0.99+ 0.99+ 0.99+ 1.00 0.98 1.00
500 506
0.70 0.92
0.83 0.96
0.90 0.98
0.95 0.99
0.98 0.99
0.99 0.99+
Einhorn et al., 1990. Ac/a Obstet Gynecol Scand Suppl 155
88
Ch. E. Welander
Table IV. Preoperative serum CA 125 in patients with a pelvic mass Author
Ref.
Patients (n)
Einhorn, 1986
(27)
100
O'Connell, 1987
(28)
56
Vasilev, 1988
(29)
182
Schwartz, 1988
(30)
211
Malkasian, 1988
(31) (24)
250
Patsner, 1988
Soper, 1990
(32)
100
Cutoff (U/ml)
35 65 35 65 35 35" 65 65" 35 65 65" 35 35" 35 65 65"
Sensitivity
Specificity
(%)
(%)
Predictive value
(+)
(-)
78
93
0.74
0.95
95 87
61 83 78 98 89
0.84 0.92
O.B5 0.75
7B 100 56 83 100 93 78 65 77 85 78 81
99
60 81 97 78 81 70 83 100
0.33 0.48 0.98 0.67 0.87 0.77 0.84
0.72
«n
0.68 O.BO 0.70
Postmenopausal. From Patsner. 1991.
a
lished to date comes from Einhorn and colleagues in Sweden (25). A preliminary report of their work describes 5,550 women older than 40 years who had annual CA 125 serum levels determined. Of this group, there were 175 women whose initial screening showed CA 125 levels to be> 35 units/ml. These 175 patients were followed with repeat CA 125 values every 3 months, with additional pelvic examinations and pelvic ultrasound examinations every 6 months. Six DC were ultimately identified in clinically asymptomatic patients from this group. Of these six, two were in surgical Stage Ia, two in surgical Stage II, and two in surgical Stage III. Whether the survival duration of these patients has been improved by this screening program remains a matter of further evaluation (25). Significant differences are noted when screening women in premenopausal vs. post-menopausal age groups. The Swedish study has shown highly significant differences in specificity when pre vs. post menopausal patients are compared (Table III) (26). One conclusion of the Swedish investigators is that CA 125 levels associated with a specificity as high as 99.5% may be inadequate to recommend invasive intervention in otherwise asymptomatic patients. Initially elevated levels of CA 125 often decline over months of follow-up. In order to reach specificity levels which would warrant invasive procedures to establish or rule out a diagnosis of ovarian cancer, CA 125 levels must be increasing over time. A predictive specificity of 99.9% can be achieved in those patients who have a doubling of initial CA 125 levels Acta Obstet Gynecol Scand Suppl /55
over the next 6 months following an elevated level. If levels continue to rise, a specificity of 100% can be achieved (26). Our conclusion concerning screening for DC would be that a single value of CA 125, even if greater than 35 units/ml, probably does not have sufficient specificity to warrant intervention in patients clinically free of symptoms or other abnormal findings. Follow-up levels over specified time intervals are necessary, watching for doubling or sustained increase in levels over time. Cost-effectiveness of mass screening needs to be studied further.
Preoperative evaluation of a pelvic mass Beyond screening normal patients for DC, can serum levels of CA 125 be helpful in defining a pelvic mass prior to surgery? Several authors have published reviews of preoperative CA 125 screening of patients with pelvic masses (Table IV) (27-32). The cut-off level of CA 125 is important in defining sensitivity and specificity. Whether 3S units/ml or 65 units/ml is the arbitrary cut-off level, sensitivity will clearly change. Decreased sensitivity - but increased specificity - is seen when the cut-off level is raised from 35 to 65 units/ml. A report by Vasilev et al. (29), from the University of Southern California showed that only 14 of 51 patients with pelvic masses who had a CA 125 level> 3S units/ml had a malignant tumor. Conversely, only 4 patients of 132 with pelvic masses who had a CA 125 level < 35 units/rnl,
CA 125 levels and ovarian cancer
89
Table V. Evaluation of various CA 125 positivitycriteria in 30 patients with epithelial ovarian cancer and 153patients with benign pelvic masses Quality index
CA 125 positivity criteria
> 35 Vlml
> 65 Vlml
> 194 Vlml
100.0 60.1
93.3 79.7 82.0 47.5
80.0 94.8 92.3 75.0
Sensitivity Specificity Diagnostic efficiency' Predictive value
66.7
33.0
Data are given as percentages. . . .. • Diagnostic efficiency
=
true positive + true negative total patients with/without disease
-----=-------=----
From Schwartz. 1988. had an actual malignancy. In the clinical setting of a newly diagnosed pelvic mass, predictions of the absence of OC seem to be more accurate than a positive diagnosis of cancer. A report by Schwartz et al. (30), from Yale University used cut-off concentrations of 35 units/ml, 65 units/ml and 194 units/ml, and calculated predictive indices for OC. Sensitivity decreased with higher cut-off concentrations, but specificity and diagnostic efficiency increased with higher CA 125 levels (Table V). Benign masses have been associated with CA 125 levels greater than 65 units/ml in the report by Schwartz et al. (30), as follows: benign epithelial ovarian tumors, 10% Table VI. Pretreatment serum levels of CA 125 in relation to extent of disease, histologic type, and tumor grade in ovarian carcinoma Category
N
From Tholander et al., 1990.
Even with this higher cut-off level, the false positive rate is troublesome. When patient groups are divided by age, interpretation of results is improved. The report from the University of Southern California has shown that all patients less than age 50 who had CA 125 levels < 35 units/ml had benign disease. Likewise, 80% of patients older than 50 years with CA 125 levels> 35 units/ml had malignant tumors (29). Clearly, a preoperative decision regarding the risk of benign vs. malignant disease cannot be based solely on CA 125 levels. By considering the patient's age and other relevant clinical factors, in addition to CA 125 levels, higher probabilities of accuracy can be achieved.
CA 125 (Vlml) Mean
Ovarian epithelial carcinoma All 142 Limited disease 38 Extensive disease 104 Histological type 77 Serous Mucinous 17 29 Endometrioid Clear cell 5 Undifferentiated 12 Grade of differentiation 40 Well. Grade I Moderately, Grade 2 51 Poorly. Grade 3 51
uterine leiomyomas, 4.4% pelvic inflammatory masses, 9.5% ovarian endometriomas, 37.5%
Median
630 151 805
228 72 299
813
261 19 333 130 428
64
394 140 1085 177 605 1010
151 261 406
Correlation of marker levels with cell type and bulk of OC Correlation of CA 125 levels with pathologic parameters in OC patients has been made. Histologic cell type and tumor grade have been correlated with certain changes in CA 125 levels by a Swedish collaborative group, published by Tholander et al. (11), (Table VI). Mucinous tumors have the lowest levels of CA 125 expression in serum. The grade of the tumor in this study did correlate directly with shedding of CA 125 antigen into the serum. Poorly differentiated tumors had the highest serum concentrations. Data are not available from this study or any other which correlates tissue levels of CA 125 with tumor grade or cell type. Tumor bulk possibly correlates with CA 125 levels. Data in Table VI also show gross differences Acta Obstet Gyneco! Scand Suppf /55
90
Ch. E. Welander
Table VII. Correlation of serum CA 125 with second-look surgery Author
Ref.
Patients
Approach
Sensitivity
Specificity
Preditive value
(n)
(+)
Niloff, 1985
(38)
35
Berek, 1986
(39)
55
Schilthuis, 1987 Rome, 1987 Patsner, 1990
(40) (41) (42)
60 49 125
Laparoscopyl laparotomy Laparoscopyl laparotomy Laparotomy Laparotomy Laparotomy
0.28
0.31 0.52 0.27
0.90
0.90 1.00 0.93
(-)
88% 100%
57%
83%
49%
From Patsner, 1991. when comparing limited vs. extensive disease. Other authors have also reported correlations with tumor bulk and CA 125 serum levels. One report, by Fleuren et al. (33), suggests that diffuse peritoneal tumor spread was more closely correlated with CA 125 levels than was the bulk of a primary lesion.
Correlation of CA 125 levels with responses to therapy An early report describing clinical studies of CA 125 was one in which patients were followed through therapy for oc. Changes in serum CA 125 levels were considered significant when they either doubled, or decreased by one-half. Changes of this magnitude were correlated with a similar clinical course of disease in 42 of 45 patients (2). It is not possible to equate undetectable serum levels of CA 125 with an absence of tumor. Contrary to more specific tumor markers such as BHCG for gestational trophoblastic disease, it is not possible to assure a patient that no tumor remains when CA 125 has reaches undetectable levels. CA 125 levels can be correlated with favorable results from surgical therapy for Oc. A group in Vienna has published a 'prognostic index', using CA 125 values before and after surgery to calculate a probability of long-term response (34). Patients having lower CA 125 scores both before and after surgery will predictably do better, according to this group. Their 'prognostic index' also correlates well with other traditional predictors of response, including FIGO stage, size of residual tumor mass, and the presence of ascites. Responses to chemotherapy have also been correlated with changes in CA 125. Investigators in Amsterdam have correlated the rate of CA 125 decline, following chemotherapy, in patients with advanced OC (35). Of the 85 patients studied, responding patients had a mean CA 125 half-life of 9.2 days, Acta Obstet Gynecol Scand Suppl 155
associated with a progression-free interval of 43 months. Patients with stable disease or partial responses had a mean CA 125 half-life of 22.6 days, with progression-free intervals of 11 months. A Danish group of investigators looked at levels of CA 125 following three cycles of chemotherapy (36). All patients whose CA 125 values had not gone down to normal levels after three courses were found to have residual tumor. Patients having complete responses to chemotherapy all had normal CA 125 levels. In contrast, a CA 125 level which did decrease to undetectable levels was not necessarily correlated with a complete response. There were 23 of 37 patients who had residual tumor, even though they achieved a negative CA 125 level. Rustin et al. have reported the use of CA 125 to monitor response to CarboPlatin chemotherapy (37). Patients whose CA 125 level decreased to oneseventh following one course of treatment had a significantly longer disease-free interval than patients with a less rapid decline in CA 125 values. The absolute level of CA 125 before treatment was not predictive of response; the rate of fall was the important parameter. Such studies are more accurately used to make negative rather than positive predictions. If CA 125 levels do not decline at a certain rate, or reach normal values after a particular course of chemotherapy, then that patient will respond poorly over the longterm. It is with considerably less accuracy and confidence that we predict good outcomes for patients who have favorable CA 125 values.
Identification of complete responders to therapy Numerous investigators have surveyed their patients in an attempt to correlate CA 125 levels prior to 'second-look' laparotomy with pathologic findings (Table VII) (38-42). The conclusions of investiga-
CA 125 levels and ovarian cancer
91
Table VIII. Sensitivity, specificity, and predictive value of CA 125, LSA, and NB170K in patients with ovarian cancer
CA 125 LSA NB170K
No. of patients
Sensitivity clinical evidence of disease (%)
Specificity no evidence of disease (%)
Positive predictive value (%)
Negative predictive value (%)
152 142 91
90.2 78.6 76.0
92.3 62.8 74.2
88.7 57.9 52.8
93.3 81.8 89.1
From Petru et al., 1990.
tors are similar. Each patient must have had an elevated CA 125 level prior to initiation of therapy. When the CA 125 level is > 35 units/ml at the time of second-look surgery, a very high predictive probability exists that residual tumor will be found. Some series have reported a 100% positive predictive value for an elevated CA 125. Conversely, a normal level of CA 125 is not helpful in predicting the absence of tumor. Poor correlation of CA 125 with small volumes of tumor continues to be the problem. Pathologic cell type does not seem to matter, as long as the tumor did express CA 125 prior to therapy.
Future directions for OC detection Considering the apparent predictive limitations of CA 125 by itself, combined modalities have been tested, utilizing complementary means to detect and/or follow Oc. cancers. Investigators in Miami designed a panel of three serum markers, which they used to monitor responses to therapy in 152 OC patients (43). The panel included CA 125, lipidassociated sialic acid (LSA), and NBnOK. While each separate marker has some degree of sensitivity and specificity, CA 125 alone was equal to or better than the combination of all three in predicting changes in tumor volume with therapy (Table VIII). Table IX. The sensitivity and specificity for malignancy given different levels of the risk of malignancy index (RMI) RMI score
Sensitivity
Specificity
(%)
(%)
25 50 75 100 150 200 250
100.0 95.1 92.7 85.4 85.4 85.4 78.0
62.2 76.5 84.7 87.8 93.9 %.9 99.0
A group of investigators in London have developed a predictive Risk of Malignancy Index (RMI) for use in screening patients who have a clinical suspicion of OC (44). This group studied 143 patients, recording age, ultrasound findings (done by abdominal technique), menopausal status, findings on pelvic examination, and CA 125 levels. When a careful statistical analysis of each parameter was made, the conclusion was reached that an RMI could be calculated by using only three of the above criteria: the ultrasound score, CA 125 level, and menopausal status. The equation is as follows: RMI Where U U U M M
= U x M x Serum CA 125 level = 0 for ultrasound score of 0 = 1 for ultrasound score of I = 3 for ultrasound score of 2-5 =
1 if premenopausal
= 3 if post menopausal.
In Table IX a summary of RMI scores and observed sensitivity/specificity is recorded (44). It is noteworthy that an ultrasound score of 0, which describes no significantly abnormal size or shape changes in the ovaries, will invariably make the total score O. Increased accuracy in the predictive index might be forthcoming with vaginal probe ultrasound instead of abdominal. Unfortunately, size and shape criteria applied to the ovaries will often be a late means of detecting cancers. By the time a tumor is large enough to increase ovarian size, it may well have spread to peritoneal surfaces.
Conclusion
From Jacobs et al., 1990.
We have only imperfect means to detect and monitor ovarian cancers. The lack of sensitivity and specificity of each clinical test available to us limits costeffective population screening and accurate monitoring of patients with cancers. Combinations of imperfect screening modalities still do not yield specificity of the degree that we desire. However, having no other choice, we continue onward, looking for ways Acta Obstet Gynecol Scand Suppl J55
92
Ch. E. We/ander
to better use the tools we already have, all the while searching for the new ideal tumor marker.
References I. Bast RC Jr, Feeney M, Lazarus H, Nadler LM, Colvin
2.
3.
4.
5.
6.
7.
8.
9.
10.
I I.
12.
13.
RB. Knapp RC. Reactivity of a monoclonal antibody with human ovarian carcinoma. J Clin Invest 1981; 68: 1331-37. Bast RC Jr, Klug TL, St John E. et al. A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 1983; 309: 883-87. Chen DX. Schwartz PE, Li FO. Saliva and serum CA 125 assays for detecting malignant ovarian tumors. Obstet Gynecol 1990; 75: 701-04. Hunter VJ, Weinberg JB. Haney AF. et al. CA 125 in peritoneal fluid and serum from patients with benign gynecologic conditions and ovarian cancer. Gynecol Oncol 1990; 36: 161-65. Schwartz PE, Chambers SK. Chambers JT. Tilton K, Foemmel R. CA 125 in peritoneal washings and fluid: correlation with plasma CA 125 and peritoneal cytology. Obstet Gynecol 1989; 73: 339-42. Kenemans P, Bast RC Jr. Yedema CA. et al. Mucins and mucin-like antigens as serum tumor markers. Cancer Revs 1988; 4: I. Kabawat SE. Bast RC, Welch WR, Knapp RC, Colvin RB. Immunopathologic characterization of a monoclonal antibody that recognizes common surface antigens of human ovarian tumors of serous, endometrioid, and clear cell types. Am J Clin Pat hoi 1983; 79: 98-104. Kabawat SE. Bast RC. Bhan AK. Welch WR. Knapp RC, Colvin RB. Tissue distribution of coelomic-epithelium-related antigen recognized by the monoclonal antibody OC 125. Int J Gynecol Pathol 1983; 2: 275-85. Jacobs IJ, Fay TN. Stabile I. Bridges J. Oram DHO, Grudzinskas JG. The distribution of CA 125 in the reproductive tract of pregnant and non-pregnant women. Br J Obstet Gynaecol 1988; 95: 1190-94. Zanaboni F, Vergadoro F, Presti M, Gallotti P, Lombardi F. Bolis G. Tumor antigen CA 125 as a marker of ovarian epithelial carcinoma. Gynecol Oncol 1987; 28: 61-67. Tholander B. Taube A. Lindgren A. Sjoberg 0, Stendahl U. Tamsen L. Pretreatment serum levels of CA 125. carcinoembryonic antigen. tissue polypeptide antigen, and placental alkaline phosphatase in patients with ovarian carcinoma: influence of histologic type. grade of differentiation and clinical stage of disease. Gynecol Oncol 1990; 39: 26-33. Jacobs I, Bast RC Jr. The CA 125 tumor-associated antigen: a review of the literature. Human Reproduction 1989; 4: 1-12. Barbieri RL, Niloff JM, Bast RC, Schaetzl RN, Kistner RW, Knapp RC. Elevated serum concentrations of CA 125 in patients with advanced endometriosis. Fertil Steril 1986; 45: 630-34.
Acta Obstet Gynecol Scand Supp! /55
14. Fedele L. Vercellini P, Arcaini L, Grazia de Dalt M. Candiani GB. CA 125 in serum. peritoneal fluid, active lesions, and endometrium of patients with endometriosis. Am J Obstet Gynecol 1988; 158: 166-70. 15. Guidice LC. Jacobs A, Pineda J. Bell CEo Lippmann L. Serum levels of CA 125 in patients with endometriosis: a preliminary report. Fertil Steril 1986; 45: 876-78. 16. Kauppila A, Telimass S, Ronnberg L, Vuori J. Placebo-controlled study on serum concentrations of CA 125 before and after treatment of endometriosis with Danazol or high-dose medroxy-progesterone acetate alone or after surgery. Fertil Steril 1988; 49: 37-4 I. 17. Patton PE, Field CS. Harms RW. Coulam CB. CA 125 levels of endometriosis. Fertil Steril 1986; 45: 770-73. 18. Pittaway DE, Fayez JA. The use of CA 125 in the diagnosis and management of endometriosis. Fertil Steril 1986; 46: 790-95. 19. Takahashi K, Yamane Y, Kijima S, Yoshino K. Shibukawa T, Kitao M. CA 125 antigen is an effective diagnostic for external endometriosis. Gynecol Obstet Invest 1987; 23: 257-60. 20. Duk JM, Aalders JG, F1euren GJ, de Bruijn HWA. CA 125: a useful marker in endometrial carcinoma. Am J Obstet Gynecol 1986; 155: 1097-101. 21. Fuith LC, Dazenbickler G, Dapunt O. CA 125 in the serum and tissue of patients with gynecologic disease. Arch Gynecol Obstet 1987; 241: 157-64. 22. Meier W. Stieber P, Fateh-Moghadam A. Eiermann W, Hepp H. CA 125 in gynecological malignancies. Br J Cancer Clin Oncol 1987; 23: 713-17. 23. Niloff JM, Thomas L, Klug L. Schaetzl E. Zurawski VR, Knapp RC, Bast RC. Elevation of serum CA 125 in carcinomas of the fallopian tube. endometrium, and endocervix. Am J Obstet Gynecol 1984; 148: 1057-58. 24. Patsner B, Mann WJ. Cohen H, Loesch M. Predictive value of pre-operative serum CA 125 levels in clinically localized and advanced endometrial carcinoma. Am J Obstet Gynecol 1988; 158: 399-402. 25. Einhorn N. Sjovall K, Schoenfeld DA, et al. Early detection of ovarian cancer using the CA 125 radioimmunoassay. Proc Amer Soc Clin Oncol 1990; 9: 157. 26. Zurawski VR Jr, Sjovall K. Schoenfeld DA. et al. Prospective evaluation of serum CA 125 levels in abnormal population. Phase I: the specificities of single and serial determinations in testing for ovarian cancer. Gynecol Oncol 1990; 36: 299-305. 27. Einhorn N, Bast RC Jr. Knapp RC, Tjernberg B, Zurawski YR. Pre-operative evaluation of serum CA 125 levels in patients with primary epithelial ovarian cancer. Obstet Gynecol 1986; 67: 414-16. 28. O'Connell GJ, Ryan E, Murphy KJ, Prefontaine M. Predictive value of CA 125 for ovarian carcinoma in patients presenting with pelvic masses. Obstet Gynecol 1987; 70: 930-32. 29. Vasilev SA, Schlaerth JB, Campeau J. Morrow CPo Serum CA 125 levels in pre-operative evaluation of pelvic masses. Obstet Gynecol 1988; 71: 751-55. 30. Di-Xia C, Schwartz PE. Xinguo L. Zhan Y. Evaluation of CA 125 levels in differentiating malignant from benign tumors in patients with pelvic masses. Obstet Gynecol 1988; 72: 23-27. 31. Malkasian GD Jr, Knapp RC, Lavin PT. et al. Pre-
CA 125 levels and ovarian cancer
32.
33.
34.
35.
36.
37.
38.
39.
operative evaluation of serum CA 125 levels in premenopausal patients with pelvic masses: discrimination of benign from malignant disease. Am 1 Obstet Gynecol 1988; 158: 341-46. Soper IT. Hunter vr, Daly L. Tanner M, Creasman WT. Bast RC lr. Pre-operative serum tumor-associated antigen levels in women with pelvic masses. Obstet Gynecol 1990; 75: 249-54. F1euren Gl. Nap M. Aalders IG. Trimbos lB. de Bruijn HWA. Explanation of the limited correlation between tumor CA 125 content and serum CA 125 antigen levels in patients with ovarian tumors. Cancer 1987; 60: 2437-42. Rosen A. Sevelda P. Klein M, Spona 1. Beck A. A CA 125score as a prognostic index in patients with ovarian cancer. Arch Gynecol Obstet 1990; 247: 125-29. vander Berg ME. Lammes FB. van Putten WL, Stoter G. Ovarian cancer: the prognostic value of the serum half-life of CA 125 during induction of chemotherapy. Gynecol Oncol 1988; 30: 307-12. Mogensen O. Mogensen B, lakobsen A. Predictive value of CA 125 during early chemotherapy of advanced ovarian cancer. Gynecol Oncol 1990; 37: 44-46. Rustin GIS. Gennings IN, Nelstrop AE, Covarrubias H. Lambert HE. Bagshawe KD. Use of CA 125 to predict survival of patients with ovarian cancer. 1 Clin Oncol 1989; 7: 1667-71. Niloff 1M. Bast RC Jr. Schaetzl EM, Knapp RC Predictive value of CA 125 antigen levels in second-look procedures for ovarian cancer. Am 1 Obstet Gynecol 1985; 151: 981-86. Berek IS. Knapp RC Malkasian GO. et al. CA 125 serum levels correlated with second-look operations
40.
41.
42.
43.
44.
93
among ovarian cancer patients. Obstet Gynecol 1986; 67: 685-89. Schilthuis MS. Aalders IG, Bouma 1, et al. Serum CA 125 levels in epithelial ovarian cancer: relation with findings at second-look operations and their role in the detection of tumor recurrence. Br 1 Obstet Gynaecol 1987; 94: 202-07. Rome RM. Koh H. Fortune D. Cauchi M. CA 125 serum levels and secondary laparotomy in epithelial ovarian tumors. Aust New Zealand 1 Obstet Gynecol 1987; 27: 142-45. Patsner B. Orr lW Jr. Mann Wl, Taylor PT, Partridge E, Allmen T. Does serum CA 125 level prior to second-look laparotomy for invasive ovarian adenocarcinoma predict size of residual disease? Gynecol Oncol 1990; 37: 319-22. Petru E. Sevin BU. Averette HE. Koechli OR, Perras IP, Hilsenbeck S. Comparison of three tumor markers - CA 125. Lipid-Associated Sialic Acid. and NB170K in monitoring ovarian cancer. Gynecol On col 1990; 38: 181-86. lacobs I. Oram D. Fairbanks 1. Turner 1. Frost C, Grudzinskas IG. A risk of malignancy index incorporating CA 125. ultrasound. and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br 1 Obstet Gynaecol 1990; 97: 922-29.
Address for correspondence:
Charles E. Welander, M.D. University of Vermont Medical Center Hospital of Vermont Burlington, VT 5401 USA
ACla Obstet Gynecol Scand Supp/ /55
What do CA 125 and other antigens tell us about ovarian cancer biology? CHARLES
E. WELANDER
From the Gynecologic Oncology Laboratory, Institute for Gynecologic Oncology, Atlanta, GA, USA
Acta Obstet Gynecol Scand 1992; 71 Suppl ISS: 85-93
CA 125 is an antigenic determinant on a high molecular weight glycoprotein. A monoclonal antibody has been produced which recognizes this. and allows us to measure the expression of CA 125 in serum. Tissue distribution of the CA 125 determinant is most commonly seen in serous tumors of the ovary. with highest levels in borderline and frankly malignant serous cystadenocarcinomas. Occasionally cancers of the breast, gastrointestinal tract. and kidney will show elevated levels of CA 125. Normal tissues which show varying levels of CA 125 include decidual tissue and structures derived from celomicepithelium. CA 125 is clearly tumor associated, but not tumor specific. Quantitative correlation of CA 125 levels with tumor volume has not been demonstrated. This observation limits the clinical usefulness of CA 125 as a screening tool. particularly in premenopausal patients who do not have a diagnosis of ovarian cancer. An undetectable levelof CA 125 antigen does not rule out the presence of an early ovarian cancer. When CA 125 is used to monitor disease state in patients with known ovarian cancers (whose tumors do express CA 125), changes in levels of CA 125 do correlate with gross changes in tumor volume. Good prognostic significanceis attributed to a rapid decline in CA 125 levels following induction chemotherapy in patients with advanced ovarian cancer. However. an undetectable serum level of CA 125 does not predict clinical cure for a patient with ovarian cancer. Further clinical studies continue in the use of CA 125 as a screening tool and as a means to monitor treatment responses of known ovarian cancer.
Our ability to interpret and explain clinical observations concerning ovarian cancer (OC) is directly related to our understanding of the embryology, physiology, and biochemistry both of normal ovaries and of malignant changes. This is clearly a complex problem, which can become distorted when our interpretation of observations is overly simplified. Clinicians do desire to have an ideal marker substance that is easily detected in patients, and that can be positively correlated with the presence of malignant cells. An additional feature of such an ideal marker would be a quantitative relationship between marker expression and the number of viable tumor cells in a patient. Clinical applications of an ideal marker for the ovarian cancer problem include:
1) population screening for early detection of OC, 2) preoperative evaluation of patients with a pelvic mass, 3) correlation of tumor marker level with the cell type of an OC. 4) a means to assess response to surgery and/or chemotherapy, 5) a means to identify complete responders to therapy. Laboratory investigators have demonstrated that the above listed ideal features of a marker are not yet available in any of the markers identified to date. Nonetheless, considerable effort has been expended in attempts to determine how clinically useful the present group of tumor markers can be. In this review, a summary of clinical studies of CA
Acta Obstet Gynecol Scand Suppl/55
86
Ch. E. Welander
Table I. Serum CA 125 levels in patients with endometriosis distributed by severity of disease Author
Ref.
Upper limit CA 125 Ulml
Stage of endometriosis 1 + 11
Barbieri et al. (1986) Fedele et al. (1988) Guidice et al. (1986) Kauppilla et al. (1988) Patton et al. (1986) Pittaway and Fayez (1986)
(15) (16) (17) (18)
Takahashi et al. (1987)
(19)
(13) (14)
35 35 35 30 35 30 16 39
Total
III + IV
3/47
7113
2/38
7/19 7/8
16f72
11/15
2/22
3/15
8192 40/92
16138
1/6 Stage I + II 11.5%
30/38 14/15
Stage III + IV 50.4%
From Jacobs et al., 1989. 125 has been prepared. CA 125 has been more widely tested than the several other markers which are clinically available. The features of an ideal marker are more closely approximated by CA 125, although it is not tumor specific, nor does it correlate well with small numbers of tumor cells. The biochemical properties of CA 125 and tissue distribution of the antigen are summarized. Clinical studies are presented, which have tested the applicability of CA 125 measurements to each of the five areas noted above. Realizing the imperfect ability of CA 125 to be an 'ideal' marker, additional combined modalities for screening patients at high risk of developing OC have been tested. Whether complementary modalities can enhance the predictive capability has been explored in the clinic, and will be reported briefly in this paper.
Properties and tissue distribution of CA 125 The tumor marker CA 125 is an antigenic determinant on a high molecular weight glycoprotein. It is recognized by a monoclonal antibody, OC 125, which was first identified by Bast and colleagues in 1981 (I). OC 125 is a murine monoclonal antibody produced in BALB/c mice which were immunized with OC cells derived from a serous papillary cystadenocarcinoma. An immunoradiometric assay has been developed to measure CA 125 in serum (2) and other body fluids, including saliva (3) and peritoneal fluid (4, 5). Concentrations of CA 125 are expressed in arbitrary units per milliliter. The serum half-life of CA 125 is approximately 4.5 days (6). Immunohistochemical techniques have also been developed to identify the CA 125 determinant on Acta Obstet Gynecol Scand Suppl 155
cell surfaces (7). CA 125 is present on the cell membrane of fetal tissues derived from celomic epithelium. Adult tissues showing surface expression of CA 125 include normal endometrium, fallopian tube, and endocervix (8). The epithelial surface of normal ovary is not positive for CA 125 staining, but cells lining ovarian cysts as well as papillary excrescences within an ovarian cyst are often positive. Patients within the first trimester of pregnancy demonstrate high tissue levels of CA 125, the bulk of this coming from decidual tissue (9). A variety of malignancies have been screened for CA 125 expression, by both serum and immunoperoxidase staining techniques. Serous ovarian tumors, including benign, borderline, and malignant types, are strongly positive for CA 125. Mucinous, sex cord, germ-cell and Brenner tumors are usually negative (7). The pathologic grade reflecting tumor differentiation can influence CA 125 expression, although that has been variously reported by different investigators (10, 11). Carcinomas arising from tissues which normally express CA 125, such as endometrium, fallopian tube, and endocervix also continue to express the marker. Other non-gynecologic tumors which frequently express elevated levels of CA 125 include carcinomas arising in the lung, breast, gastrointestinal tract, and kidney (12). Since a variety of tissues beside OC can express CA 125, it is necessary to establish a normal range of values. Low levels of expression must be considered within a normal range, in order to increase the specificity of the test. Investigators have determined that a serum level of CA 125 up to 35 units/ml is considered normal. Population studies have shown that fewer than 1% of healthy control patients will have an elevated level at that cut-off point. Approximately 6% of patients with benign gynecologic pathology will have an elevated level, using 35 units/ml
CA 125 levels and ovarian cancer
87
Table II. Elevation of serum CA 125 (> 35 Vlml) in patients with endometrial cancer by clinicalstage of disease Author
Clinical stage of endometrial cancer
Ref.
I + II Duk et al. (1986) Fuith et al. (1987) Meier et al. (1987) Niloff et al. (1984a) Patsner et al. (1988)
(20) (21 ) (22) (23) (24)
III + IV
All stages
221100
5n
27/107 13/32
0/11
14/18
21/81
8/8
14/29 29/89
81.8%
31.8%
2110
22.4%
Total From Jacobs et al., 1989.
as a cut-off (2). A common reason for frequent elevations in serum levels of CA 125 among premenopausal patients is endometriosis. Various authors have summarized groups of patients having endometriosis, showing a general increase in serum levels with advancing stage of endometriosis (Table I) (13-19). Among malignancies arising from the endometrium, elevation of serum CA 125 to levels greater than 35 unitslml is also correlated with stage of disease (Table 11) (20-24). Serum levels of CA 125 are the most common clinical determinations made, although tissue levels can also be measured from normal or malignant samples (7.8). There are instances of tumors which produce high tissue levels without significant elevations of serum CA 125. In similar fashion, certain normal tissues, such as endometrium, decidual tissue. and some benign tumors have high cellular expression of CA 125, without shedding antigen into the serum. Correlation with malignancy seems to be
best demonstrated by identifying antigen shed into the serum, rather than looking for the presence of CA 125 in tissue samples (12).
Population screening for ovarian cancer using CA 125 The ideal tumor marker described in the Introduction to this paper would be one that is expressed in detectable quantities, even from small numbers of new OC cells. A new OC may express CA 125 on the tumor cell surface, but shedding of the antigen into the serum may not occur until the tumor has grown and/or spread to distant peritoneal sites. Early detection from a clinical standpoint may still be too late to surgically cure such patients. Nonetheless, we remain anxious to clinically detect OC as early as possible. The largest prospective OC screening study pub-
Table III. Specificity of a single screen of serum CA 125 levels at different reference values according to age and menopausal status Group
All: Age (years) 40-44 45-49 50-54 55-59 60-64 65-69 70+ Menopausal status PrePost-
N
Proportion of subjects below indicated reference value (Vlml) 20
25
30
35
65
100
1082
0.81
0.90
0.94
0.97
0.99
0.99+
257 226 189 187 137 50 36
0.71 0.76 0.79 0.89 0.93 0.92 0.97
0.85 0.85 0.87 0.96 0.97 0.98 1.00
0.92 0.89 0.93 0.97 0.99 0.98 1.00
0.95 0.95 0.97 0.98 0.99 0.98 1.00
0.99 0.99 0.98 0.99+ 0.99 0.98 1.00
0.99 0.99+ 0.99+ 0.99+ 1.00 0.98 1.00
500 506
0.70 0.92
0.83 0.96
0.90 0.98
0.95 0.99
0.98 0.99
0.99 0.99+
Einhorn et al., 1990. Ac/a Obstet Gynecol Scand Suppl 155
88
Ch. E. Welander
Table IV. Preoperative serum CA 125 in patients with a pelvic mass Author
Ref.
Patients (n)
Einhorn, 1986
(27)
100
O'Connell, 1987
(28)
56
Vasilev, 1988
(29)
182
Schwartz, 1988
(30)
211
Malkasian, 1988
(31) (24)
250
Patsner, 1988
Soper, 1990
(32)
100
Cutoff (U/ml)
35 65 35 65 35 35" 65 65" 35 65 65" 35 35" 35 65 65"
Sensitivity
Specificity
(%)
(%)
Predictive value
(+)
(-)
78
93
0.74
0.95
95 87
61 83 78 98 89
0.84 0.92
O.B5 0.75
7B 100 56 83 100 93 78 65 77 85 78 81
99
60 81 97 78 81 70 83 100
0.33 0.48 0.98 0.67 0.87 0.77 0.84
0.72
«n
0.68 O.BO 0.70
Postmenopausal. From Patsner. 1991.
a
lished to date comes from Einhorn and colleagues in Sweden (25). A preliminary report of their work describes 5,550 women older than 40 years who had annual CA 125 serum levels determined. Of this group, there were 175 women whose initial screening showed CA 125 levels to be> 35 units/ml. These 175 patients were followed with repeat CA 125 values every 3 months, with additional pelvic examinations and pelvic ultrasound examinations every 6 months. Six DC were ultimately identified in clinically asymptomatic patients from this group. Of these six, two were in surgical Stage Ia, two in surgical Stage II, and two in surgical Stage III. Whether the survival duration of these patients has been improved by this screening program remains a matter of further evaluation (25). Significant differences are noted when screening women in premenopausal vs. post-menopausal age groups. The Swedish study has shown highly significant differences in specificity when pre vs. post menopausal patients are compared (Table III) (26). One conclusion of the Swedish investigators is that CA 125 levels associated with a specificity as high as 99.5% may be inadequate to recommend invasive intervention in otherwise asymptomatic patients. Initially elevated levels of CA 125 often decline over months of follow-up. In order to reach specificity levels which would warrant invasive procedures to establish or rule out a diagnosis of ovarian cancer, CA 125 levels must be increasing over time. A predictive specificity of 99.9% can be achieved in those patients who have a doubling of initial CA 125 levels Acta Obstet Gynecol Scand Suppl /55
over the next 6 months following an elevated level. If levels continue to rise, a specificity of 100% can be achieved (26). Our conclusion concerning screening for DC would be that a single value of CA 125, even if greater than 35 units/ml, probably does not have sufficient specificity to warrant intervention in patients clinically free of symptoms or other abnormal findings. Follow-up levels over specified time intervals are necessary, watching for doubling or sustained increase in levels over time. Cost-effectiveness of mass screening needs to be studied further.
Preoperative evaluation of a pelvic mass Beyond screening normal patients for DC, can serum levels of CA 125 be helpful in defining a pelvic mass prior to surgery? Several authors have published reviews of preoperative CA 125 screening of patients with pelvic masses (Table IV) (27-32). The cut-off level of CA 125 is important in defining sensitivity and specificity. Whether 3S units/ml or 65 units/ml is the arbitrary cut-off level, sensitivity will clearly change. Decreased sensitivity - but increased specificity - is seen when the cut-off level is raised from 35 to 65 units/ml. A report by Vasilev et al. (29), from the University of Southern California showed that only 14 of 51 patients with pelvic masses who had a CA 125 level> 3S units/ml had a malignant tumor. Conversely, only 4 patients of 132 with pelvic masses who had a CA 125 level < 35 units/rnl,
CA 125 levels and ovarian cancer
89
Table V. Evaluation of various CA 125 positivitycriteria in 30 patients with epithelial ovarian cancer and 153patients with benign pelvic masses Quality index
CA 125 positivity criteria
> 35 Vlml
> 65 Vlml
> 194 Vlml
100.0 60.1
93.3 79.7 82.0 47.5
80.0 94.8 92.3 75.0
Sensitivity Specificity Diagnostic efficiency' Predictive value
66.7
33.0
Data are given as percentages. . . .. • Diagnostic efficiency
=
true positive + true negative total patients with/without disease
-----=-------=----
From Schwartz. 1988. had an actual malignancy. In the clinical setting of a newly diagnosed pelvic mass, predictions of the absence of OC seem to be more accurate than a positive diagnosis of cancer. A report by Schwartz et al. (30), from Yale University used cut-off concentrations of 35 units/ml, 65 units/ml and 194 units/ml, and calculated predictive indices for OC. Sensitivity decreased with higher cut-off concentrations, but specificity and diagnostic efficiency increased with higher CA 125 levels (Table V). Benign masses have been associated with CA 125 levels greater than 65 units/ml in the report by Schwartz et al. (30), as follows: benign epithelial ovarian tumors, 10% Table VI. Pretreatment serum levels of CA 125 in relation to extent of disease, histologic type, and tumor grade in ovarian carcinoma Category
N
From Tholander et al., 1990.
Even with this higher cut-off level, the false positive rate is troublesome. When patient groups are divided by age, interpretation of results is improved. The report from the University of Southern California has shown that all patients less than age 50 who had CA 125 levels < 35 units/ml had benign disease. Likewise, 80% of patients older than 50 years with CA 125 levels> 35 units/ml had malignant tumors (29). Clearly, a preoperative decision regarding the risk of benign vs. malignant disease cannot be based solely on CA 125 levels. By considering the patient's age and other relevant clinical factors, in addition to CA 125 levels, higher probabilities of accuracy can be achieved.
CA 125 (Vlml) Mean
Ovarian epithelial carcinoma All 142 Limited disease 38 Extensive disease 104 Histological type 77 Serous Mucinous 17 29 Endometrioid Clear cell 5 Undifferentiated 12 Grade of differentiation 40 Well. Grade I Moderately, Grade 2 51 Poorly. Grade 3 51
uterine leiomyomas, 4.4% pelvic inflammatory masses, 9.5% ovarian endometriomas, 37.5%
Median
630 151 805
228 72 299
813
261 19 333 130 428
64
394 140 1085 177 605 1010
151 261 406
Correlation of marker levels with cell type and bulk of OC Correlation of CA 125 levels with pathologic parameters in OC patients has been made. Histologic cell type and tumor grade have been correlated with certain changes in CA 125 levels by a Swedish collaborative group, published by Tholander et al. (11), (Table VI). Mucinous tumors have the lowest levels of CA 125 expression in serum. The grade of the tumor in this study did correlate directly with shedding of CA 125 antigen into the serum. Poorly differentiated tumors had the highest serum concentrations. Data are not available from this study or any other which correlates tissue levels of CA 125 with tumor grade or cell type. Tumor bulk possibly correlates with CA 125 levels. Data in Table VI also show gross differences Acta Obstet Gyneco! Scand Suppf /55
90
Ch. E. Welander
Table VII. Correlation of serum CA 125 with second-look surgery Author
Ref.
Patients
Approach
Sensitivity
Specificity
Preditive value
(n)
(+)
Niloff, 1985
(38)
35
Berek, 1986
(39)
55
Schilthuis, 1987 Rome, 1987 Patsner, 1990
(40) (41) (42)
60 49 125
Laparoscopyl laparotomy Laparoscopyl laparotomy Laparotomy Laparotomy Laparotomy
0.28
0.31 0.52 0.27
0.90
0.90 1.00 0.93
(-)
88% 100%
57%
83%
49%
From Patsner, 1991. when comparing limited vs. extensive disease. Other authors have also reported correlations with tumor bulk and CA 125 serum levels. One report, by Fleuren et al. (33), suggests that diffuse peritoneal tumor spread was more closely correlated with CA 125 levels than was the bulk of a primary lesion.
Correlation of CA 125 levels with responses to therapy An early report describing clinical studies of CA 125 was one in which patients were followed through therapy for oc. Changes in serum CA 125 levels were considered significant when they either doubled, or decreased by one-half. Changes of this magnitude were correlated with a similar clinical course of disease in 42 of 45 patients (2). It is not possible to equate undetectable serum levels of CA 125 with an absence of tumor. Contrary to more specific tumor markers such as BHCG for gestational trophoblastic disease, it is not possible to assure a patient that no tumor remains when CA 125 has reaches undetectable levels. CA 125 levels can be correlated with favorable results from surgical therapy for Oc. A group in Vienna has published a 'prognostic index', using CA 125 values before and after surgery to calculate a probability of long-term response (34). Patients having lower CA 125 scores both before and after surgery will predictably do better, according to this group. Their 'prognostic index' also correlates well with other traditional predictors of response, including FIGO stage, size of residual tumor mass, and the presence of ascites. Responses to chemotherapy have also been correlated with changes in CA 125. Investigators in Amsterdam have correlated the rate of CA 125 decline, following chemotherapy, in patients with advanced OC (35). Of the 85 patients studied, responding patients had a mean CA 125 half-life of 9.2 days, Acta Obstet Gynecol Scand Suppl 155
associated with a progression-free interval of 43 months. Patients with stable disease or partial responses had a mean CA 125 half-life of 22.6 days, with progression-free intervals of 11 months. A Danish group of investigators looked at levels of CA 125 following three cycles of chemotherapy (36). All patients whose CA 125 values had not gone down to normal levels after three courses were found to have residual tumor. Patients having complete responses to chemotherapy all had normal CA 125 levels. In contrast, a CA 125 level which did decrease to undetectable levels was not necessarily correlated with a complete response. There were 23 of 37 patients who had residual tumor, even though they achieved a negative CA 125 level. Rustin et al. have reported the use of CA 125 to monitor response to CarboPlatin chemotherapy (37). Patients whose CA 125 level decreased to oneseventh following one course of treatment had a significantly longer disease-free interval than patients with a less rapid decline in CA 125 values. The absolute level of CA 125 before treatment was not predictive of response; the rate of fall was the important parameter. Such studies are more accurately used to make negative rather than positive predictions. If CA 125 levels do not decline at a certain rate, or reach normal values after a particular course of chemotherapy, then that patient will respond poorly over the longterm. It is with considerably less accuracy and confidence that we predict good outcomes for patients who have favorable CA 125 values.
Identification of complete responders to therapy Numerous investigators have surveyed their patients in an attempt to correlate CA 125 levels prior to 'second-look' laparotomy with pathologic findings (Table VII) (38-42). The conclusions of investiga-
CA 125 levels and ovarian cancer
91
Table VIII. Sensitivity, specificity, and predictive value of CA 125, LSA, and NB170K in patients with ovarian cancer
CA 125 LSA NB170K
No. of patients
Sensitivity clinical evidence of disease (%)
Specificity no evidence of disease (%)
Positive predictive value (%)
Negative predictive value (%)
152 142 91
90.2 78.6 76.0
92.3 62.8 74.2
88.7 57.9 52.8
93.3 81.8 89.1
From Petru et al., 1990.
tors are similar. Each patient must have had an elevated CA 125 level prior to initiation of therapy. When the CA 125 level is > 35 units/ml at the time of second-look surgery, a very high predictive probability exists that residual tumor will be found. Some series have reported a 100% positive predictive value for an elevated CA 125. Conversely, a normal level of CA 125 is not helpful in predicting the absence of tumor. Poor correlation of CA 125 with small volumes of tumor continues to be the problem. Pathologic cell type does not seem to matter, as long as the tumor did express CA 125 prior to therapy.
Future directions for OC detection Considering the apparent predictive limitations of CA 125 by itself, combined modalities have been tested, utilizing complementary means to detect and/or follow Oc. cancers. Investigators in Miami designed a panel of three serum markers, which they used to monitor responses to therapy in 152 OC patients (43). The panel included CA 125, lipidassociated sialic acid (LSA), and NBnOK. While each separate marker has some degree of sensitivity and specificity, CA 125 alone was equal to or better than the combination of all three in predicting changes in tumor volume with therapy (Table VIII). Table IX. The sensitivity and specificity for malignancy given different levels of the risk of malignancy index (RMI) RMI score
Sensitivity
Specificity
(%)
(%)
25 50 75 100 150 200 250
100.0 95.1 92.7 85.4 85.4 85.4 78.0
62.2 76.5 84.7 87.8 93.9 %.9 99.0
A group of investigators in London have developed a predictive Risk of Malignancy Index (RMI) for use in screening patients who have a clinical suspicion of OC (44). This group studied 143 patients, recording age, ultrasound findings (done by abdominal technique), menopausal status, findings on pelvic examination, and CA 125 levels. When a careful statistical analysis of each parameter was made, the conclusion was reached that an RMI could be calculated by using only three of the above criteria: the ultrasound score, CA 125 level, and menopausal status. The equation is as follows: RMI Where U U U M M
= U x M x Serum CA 125 level = 0 for ultrasound score of 0 = 1 for ultrasound score of I = 3 for ultrasound score of 2-5 =
1 if premenopausal
= 3 if post menopausal.
In Table IX a summary of RMI scores and observed sensitivity/specificity is recorded (44). It is noteworthy that an ultrasound score of 0, which describes no significantly abnormal size or shape changes in the ovaries, will invariably make the total score O. Increased accuracy in the predictive index might be forthcoming with vaginal probe ultrasound instead of abdominal. Unfortunately, size and shape criteria applied to the ovaries will often be a late means of detecting cancers. By the time a tumor is large enough to increase ovarian size, it may well have spread to peritoneal surfaces.
Conclusion
From Jacobs et al., 1990.
We have only imperfect means to detect and monitor ovarian cancers. The lack of sensitivity and specificity of each clinical test available to us limits costeffective population screening and accurate monitoring of patients with cancers. Combinations of imperfect screening modalities still do not yield specificity of the degree that we desire. However, having no other choice, we continue onward, looking for ways Acta Obstet Gynecol Scand Suppl J55
92
Ch. E. We/ander
to better use the tools we already have, all the while searching for the new ideal tumor marker.
References I. Bast RC Jr, Feeney M, Lazarus H, Nadler LM, Colvin
2.
3.
4.
5.
6.
7.
8.
9.
10.
I I.
12.
13.
RB. Knapp RC. Reactivity of a monoclonal antibody with human ovarian carcinoma. J Clin Invest 1981; 68: 1331-37. Bast RC Jr, Klug TL, St John E. et al. A radioimmunoassay using a monoclonal antibody to monitor the course of epithelial ovarian cancer. N Engl J Med 1983; 309: 883-87. Chen DX. Schwartz PE, Li FO. Saliva and serum CA 125 assays for detecting malignant ovarian tumors. Obstet Gynecol 1990; 75: 701-04. Hunter VJ, Weinberg JB. Haney AF. et al. CA 125 in peritoneal fluid and serum from patients with benign gynecologic conditions and ovarian cancer. Gynecol Oncol 1990; 36: 161-65. Schwartz PE, Chambers SK. Chambers JT. Tilton K, Foemmel R. CA 125 in peritoneal washings and fluid: correlation with plasma CA 125 and peritoneal cytology. Obstet Gynecol 1989; 73: 339-42. Kenemans P, Bast RC Jr. Yedema CA. et al. Mucins and mucin-like antigens as serum tumor markers. Cancer Revs 1988; 4: I. Kabawat SE. Bast RC, Welch WR, Knapp RC, Colvin RB. Immunopathologic characterization of a monoclonal antibody that recognizes common surface antigens of human ovarian tumors of serous, endometrioid, and clear cell types. Am J Clin Pat hoi 1983; 79: 98-104. Kabawat SE. Bast RC. Bhan AK. Welch WR. Knapp RC, Colvin RB. Tissue distribution of coelomic-epithelium-related antigen recognized by the monoclonal antibody OC 125. Int J Gynecol Pathol 1983; 2: 275-85. Jacobs IJ, Fay TN. Stabile I. Bridges J. Oram DHO, Grudzinskas JG. The distribution of CA 125 in the reproductive tract of pregnant and non-pregnant women. Br J Obstet Gynaecol 1988; 95: 1190-94. Zanaboni F, Vergadoro F, Presti M, Gallotti P, Lombardi F. Bolis G. Tumor antigen CA 125 as a marker of ovarian epithelial carcinoma. Gynecol Oncol 1987; 28: 61-67. Tholander B. Taube A. Lindgren A. Sjoberg 0, Stendahl U. Tamsen L. Pretreatment serum levels of CA 125. carcinoembryonic antigen. tissue polypeptide antigen, and placental alkaline phosphatase in patients with ovarian carcinoma: influence of histologic type. grade of differentiation and clinical stage of disease. Gynecol Oncol 1990; 39: 26-33. Jacobs I, Bast RC Jr. The CA 125 tumor-associated antigen: a review of the literature. Human Reproduction 1989; 4: 1-12. Barbieri RL, Niloff JM, Bast RC, Schaetzl RN, Kistner RW, Knapp RC. Elevated serum concentrations of CA 125 in patients with advanced endometriosis. Fertil Steril 1986; 45: 630-34.
Acta Obstet Gynecol Scand Supp! /55
14. Fedele L. Vercellini P, Arcaini L, Grazia de Dalt M. Candiani GB. CA 125 in serum. peritoneal fluid, active lesions, and endometrium of patients with endometriosis. Am J Obstet Gynecol 1988; 158: 166-70. 15. Guidice LC. Jacobs A, Pineda J. Bell CEo Lippmann L. Serum levels of CA 125 in patients with endometriosis: a preliminary report. Fertil Steril 1986; 45: 876-78. 16. Kauppila A, Telimass S, Ronnberg L, Vuori J. Placebo-controlled study on serum concentrations of CA 125 before and after treatment of endometriosis with Danazol or high-dose medroxy-progesterone acetate alone or after surgery. Fertil Steril 1988; 49: 37-4 I. 17. Patton PE, Field CS. Harms RW. Coulam CB. CA 125 levels of endometriosis. Fertil Steril 1986; 45: 770-73. 18. Pittaway DE, Fayez JA. The use of CA 125 in the diagnosis and management of endometriosis. Fertil Steril 1986; 46: 790-95. 19. Takahashi K, Yamane Y, Kijima S, Yoshino K. Shibukawa T, Kitao M. CA 125 antigen is an effective diagnostic for external endometriosis. Gynecol Obstet Invest 1987; 23: 257-60. 20. Duk JM, Aalders JG, F1euren GJ, de Bruijn HWA. CA 125: a useful marker in endometrial carcinoma. Am J Obstet Gynecol 1986; 155: 1097-101. 21. Fuith LC, Dazenbickler G, Dapunt O. CA 125 in the serum and tissue of patients with gynecologic disease. Arch Gynecol Obstet 1987; 241: 157-64. 22. Meier W. Stieber P, Fateh-Moghadam A. Eiermann W, Hepp H. CA 125 in gynecological malignancies. Br J Cancer Clin Oncol 1987; 23: 713-17. 23. Niloff JM, Thomas L, Klug L. Schaetzl E. Zurawski VR, Knapp RC, Bast RC. Elevation of serum CA 125 in carcinomas of the fallopian tube. endometrium, and endocervix. Am J Obstet Gynecol 1984; 148: 1057-58. 24. Patsner B, Mann WJ. Cohen H, Loesch M. Predictive value of pre-operative serum CA 125 levels in clinically localized and advanced endometrial carcinoma. Am J Obstet Gynecol 1988; 158: 399-402. 25. Einhorn N. Sjovall K, Schoenfeld DA, et al. Early detection of ovarian cancer using the CA 125 radioimmunoassay. Proc Amer Soc Clin Oncol 1990; 9: 157. 26. Zurawski VR Jr, Sjovall K. Schoenfeld DA. et al. Prospective evaluation of serum CA 125 levels in abnormal population. Phase I: the specificities of single and serial determinations in testing for ovarian cancer. Gynecol Oncol 1990; 36: 299-305. 27. Einhorn N, Bast RC Jr. Knapp RC, Tjernberg B, Zurawski YR. Pre-operative evaluation of serum CA 125 levels in patients with primary epithelial ovarian cancer. Obstet Gynecol 1986; 67: 414-16. 28. O'Connell GJ, Ryan E, Murphy KJ, Prefontaine M. Predictive value of CA 125 for ovarian carcinoma in patients presenting with pelvic masses. Obstet Gynecol 1987; 70: 930-32. 29. Vasilev SA, Schlaerth JB, Campeau J. Morrow CPo Serum CA 125 levels in pre-operative evaluation of pelvic masses. Obstet Gynecol 1988; 71: 751-55. 30. Di-Xia C, Schwartz PE. Xinguo L. Zhan Y. Evaluation of CA 125 levels in differentiating malignant from benign tumors in patients with pelvic masses. Obstet Gynecol 1988; 72: 23-27. 31. Malkasian GD Jr, Knapp RC, Lavin PT. et al. Pre-
CA 125 levels and ovarian cancer
32.
33.
34.
35.
36.
37.
38.
39.
operative evaluation of serum CA 125 levels in premenopausal patients with pelvic masses: discrimination of benign from malignant disease. Am 1 Obstet Gynecol 1988; 158: 341-46. Soper IT. Hunter vr, Daly L. Tanner M, Creasman WT. Bast RC lr. Pre-operative serum tumor-associated antigen levels in women with pelvic masses. Obstet Gynecol 1990; 75: 249-54. F1euren Gl. Nap M. Aalders IG. Trimbos lB. de Bruijn HWA. Explanation of the limited correlation between tumor CA 125 content and serum CA 125 antigen levels in patients with ovarian tumors. Cancer 1987; 60: 2437-42. Rosen A. Sevelda P. Klein M, Spona 1. Beck A. A CA 125score as a prognostic index in patients with ovarian cancer. Arch Gynecol Obstet 1990; 247: 125-29. vander Berg ME. Lammes FB. van Putten WL, Stoter G. Ovarian cancer: the prognostic value of the serum half-life of CA 125 during induction of chemotherapy. Gynecol Oncol 1988; 30: 307-12. Mogensen O. Mogensen B, lakobsen A. Predictive value of CA 125 during early chemotherapy of advanced ovarian cancer. Gynecol Oncol 1990; 37: 44-46. Rustin GIS. Gennings IN, Nelstrop AE, Covarrubias H. Lambert HE. Bagshawe KD. Use of CA 125 to predict survival of patients with ovarian cancer. 1 Clin Oncol 1989; 7: 1667-71. Niloff 1M. Bast RC Jr. Schaetzl EM, Knapp RC Predictive value of CA 125 antigen levels in second-look procedures for ovarian cancer. Am 1 Obstet Gynecol 1985; 151: 981-86. Berek IS. Knapp RC Malkasian GO. et al. CA 125 serum levels correlated with second-look operations
40.
41.
42.
43.
44.
93
among ovarian cancer patients. Obstet Gynecol 1986; 67: 685-89. Schilthuis MS. Aalders IG, Bouma 1, et al. Serum CA 125 levels in epithelial ovarian cancer: relation with findings at second-look operations and their role in the detection of tumor recurrence. Br 1 Obstet Gynaecol 1987; 94: 202-07. Rome RM. Koh H. Fortune D. Cauchi M. CA 125 serum levels and secondary laparotomy in epithelial ovarian tumors. Aust New Zealand 1 Obstet Gynecol 1987; 27: 142-45. Patsner B. Orr lW Jr. Mann Wl, Taylor PT, Partridge E, Allmen T. Does serum CA 125 level prior to second-look laparotomy for invasive ovarian adenocarcinoma predict size of residual disease? Gynecol Oncol 1990; 37: 319-22. Petru E. Sevin BU. Averette HE. Koechli OR, Perras IP, Hilsenbeck S. Comparison of three tumor markers - CA 125. Lipid-Associated Sialic Acid. and NB170K in monitoring ovarian cancer. Gynecol On col 1990; 38: 181-86. lacobs I. Oram D. Fairbanks 1. Turner 1. Frost C, Grudzinskas IG. A risk of malignancy index incorporating CA 125. ultrasound. and menopausal status for the accurate preoperative diagnosis of ovarian cancer. Br 1 Obstet Gynaecol 1990; 97: 922-29.
Address for correspondence:
Charles E. Welander, M.D. University of Vermont Medical Center Hospital of Vermont Burlington, VT 5401 USA
ACla Obstet Gynecol Scand Supp/ /55
