- Email: [email protected]
Correlation of plasma CEA and CEA tissue staining in poorly differentiated colorectal cancer
Correlation of Plasma CEA and CEA Tissue Staining in Poorly Differentiated Colorectal Cancer
Ft. GOSLIN,
M.D.
M. J. O’BRIEN, G. STEELE,
M.D.
R. MAYER,
M.D.
R. WILSON,
M.B.
M.D.
J. M. CORSON,
M.D.
N. ZAMCHECK,
M.D.
Boston. Massachusetts
From the Mallory Gastrointestinal Research Laboratory, Boston City Hospital; Division of Medical Oncology, Sidney Farber Cancer Institute; Departments of Surgery and Pathology, Peter Bent Brigham Hospital; Departments of Medicine, Surgery and Pathology, Harvard Medical School and the Department of Pathology, Boston University School of Medicine. This study was supported by Grants (CA-04486 (N.Z.) and CA-22513 (G.S.) from the National Cancer Institute. Requests for reprints should be addressed to Dr. Norman Zamcheck. Mallory Gastrointestinal Research Laboratory, Boston City Hospital, Boston, MA 02118. Manuscript accepted February 26, 1981.
246
August 1981
A review of patients operated on for colorectal cancer disclosed poorly differentiated histologies in those whose preoperative and postoperative carcinoembryonic antigen (CEA) levels were not elevated, even in the presence of metastatic disease. CEA was, therefore, of little prognostic value or predictive of disease recurrence in these patients. The amount of CEA in tumor tissue of 17 patients with poorly differentiated colorectal cancer was estimated with the immunoperoxidase staining technique and was correlated with histology and plasma CEA levels obtained during various stages of disease. Five tumors did not stain negatively for CEA and all had predominantly poorly differentiated histologies. In all of these patients metastatic disease developed but not elevated plasma CEA levels. in contrast 12 tumors stained positively for CEA and were found to contain either differentiated areas or signet ring cells. Serial plasma CEA levels correctly monitored the postoperative courses of all 12 patients. Six of these had a relapse and all were detected by serial increases in plasma CEA. The remaining six were disease free >48 months after resection and had normal plasma CEA levels. Among poorly differentiated tumors, those that contain glandforming areas or signet ring cells can be assumed to produce CEA, and plasma CEA levels can be used effectively for monitoring. On the other hand, undifferentiated tumors which do not stain for CEA identify those patients whose plasma CEA levels do not provide a useful monitor. Circulating levels of CEA have proved to be useful in the management of colorectal cancer and are widely accepted as a monitor for recurrence of disease after primary tumor resection [l-8], and of response to treatment in patients with known metastatic disease [9,10]. Determining the presence of CEA preoperatively has been shown to be of prognostic value when used in conjunction with Dukes staging and may be helpful in adjuvant chemo- and immunotherapy trials by identifying those patients at greatest risk for recurrence after “curative” resection of their primary tumors [ 1 l-131. In some patients, however, plasma CEA levels do not correlate with the clinical course. In a recent review of patients operated on for colorectal cancer, we found that all the patients whose CEA levels did not correctly monitor the clinical course had “poorly differentiated” carcinomas [ 141. We have now analyzed 17 cases of poorly differentiated colorectal adenocarcinoma (1) to learn what factors accounted for variation in plasma CEA levels and (2) to correlate elevations in plasma CEA levels with the amount and extent of tissue CEA
The American Journal of Medlclne
Volume 71
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
immunocytochemically mors.
localized
in the primary
tu-
MATERIALS AND METHODS Patients. Preoperative and serial postoperative CEA levels were available in 145 patients treated surgically for colorectal cancer at the Peter Bent Brigham Hospital between 1972 and 1976 [ 141. In 23 of these patients the cancers were pathologically classified as “poorly differentiated,” and 17 have now been studied further. Four of the 23 patients were excluded because tissue blocks were unavailable for histologic staining; two others were excluded after review of the histology showed the carcinomas to be well differentiated. Fourteen of these 17 patients underwent “curative” resections. The three patients with unresectable lesions were operated on for local control and for tissue diagnosis. Six of the tumors were located in the right colon, one in the transverse colon, two in the left colon, two in the sigmoid colon and six in the rectum. All patients had preoperative plasma CEA determinations, liver function tests (serum glutamic oxaloacetic transaminase, lactic dehydrogenase, bilirubin, alkaline phosphatase) and liver scans. Asymptomatic patients with no evidence of metastatic disease were followed postoperatively with serial CEA deterinations, liver function tests and physical examinations at least every four months. Follow-up chest films, barium enemas and liverspleen scans were obtained annually. In patients with abnormal liver functiontests or rises in CEA levels (>2.5 ng/ml), the tests were repeated within one week. Patients with persistently abnormal liver function tests had liver scans; in some cases, liver biopsy was performed to document me&static tumor. Patients with rises in CEA levels had repeat CEA determinations every two to four weeks; if no other cause for the elevated CEA level was found and the rises persisted, a metastatic work-up was begun. If m&static disease was documented, serial CEA levels and liver function tests were obtained at least monthly. Scans and roentgenograms were repeated to assist in management and to document response or lack of response to treatment. Pathology. All tissue sections initially reported as “poorly” differentiated were classified on review either as “poorly” or “moderate to poorly” differentiated tumors by the same “reference” pathologist (M.J.O.) and assigned a Dukes classification [ 15-171. Three histologic sections of each primary tumor were reviewed without knowledge of the patient’s clinical course or plasma CEA levels. It was apparent that a range of differentiation was present in most tumors. An estimate was made of the percentage of tumor tissue which was well differentiated, moderately differentiated or poorly differentiated (Figure 1). Well differentiated tumor was defined as showing a glandular pattern which was cytoiogitally differentiated; poorly differentiated tumors showed cytologic anaplasia and no gland formation. Moderately differentiated tumor was defined as showing cytologic anaplasia and glandular differentiation. The presence of extra- and intracellular mucin was also noted. immunohistochemistry. Following initial review of the histologic sections, blocks were selected which were representative of the tumor morphology. Blocks were of the primary tumor except in Patient 5 (Table I) in whom only
August
ET AL.
metastatic tumor was available for immunohistochemical staining. Five micron paraffin sections were incubated in sequence with rabbit anti-CEA (New England Immunology Associates, Cambridge, MA) at 1:500 dilution, swine antirabbit immunoglobulin G (Dako distributed by Accurate Chemical and Scientific Corp., Hicksville, NY) at 1:25 dilution and peroxidase antiperoxidase (Dako) at 1: 100 dilution. The CEA antibody was rendered monospecific by absorption with groups A, B and 0 cells and iyophilized extracts of human lung (40 mg/ml), buffy coat (10 mg/ml), spleen (10 mg/ml), liver (10 mg/ml) and colon (10 mg/ml). Following absorption, the antibody was considered monospecific for CEA when (1)
A, well differentiated adenocarcinoma-gland formation without cytologic anaplasia. 6, moderately differentiated adanocarcinoma-gland formation and cflologic anaplasia. C, poorly differentiated adenocarcinoma-cytologic anaplasia without gland formation. Hematoxylin and eosin stain, magnification X 100 (A and 6) and X 200 (C).
Flgure 1.
lggl
The American
Journal
of Medlclne
Volume 71
247
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
TABLE I
CEA Tissue Staining of Colorectal Cancers* in Reference to Degree of
by double-immunodiffusion it reacted with purified CEA, (2) gave a single line against perchlorate extract of colon car-
Differentiation
cinoma metastases, and (3) showed no reaction against the tissue extracts used for absorption. Paraffin sections were cut and immunocytochemical localization of CEA was performed using the immunoperoxidase peroxidase antiperoxidase technique [ 18,191. The incubation time for the primary antibody was 24 hours at 0 to 4% followed by room temperature for 2 hours, and 30 minutes at room temperature for both the bridge antibody and the peroxidase antiperoxidase. The stained sections were evaluated for the amount of CEA present and its location. This was accomplished without knowledge of the patient’s clinical course or plasma CEA level. Tumors showing widespread strong staining were designated as strongly positive (+-I); when showing uniformly weak staining, weakly positive (+), or when showing staining with focal positive areas, negative (Figure 2). Sections that did not stain for CEA were designated as negative (-). Whether the positive stains were predominantly cytoplasmic, surface or both was determined in each case. Controls for the staining procedure consisted of replacement of the primary antibody with normal rabbit serum and with anti-CEA antiserum absorbed with purified CEA. A section of well differentiated colonic adenocarcinoma was stained as a standard positive control. CEA Assay. Plasma CEA assays were performed by the modified Hansen Z-gel method at the Mallory Gastrointestinal
CEA Patient
(%) DifferentIalIon Moderately Poorly
No.
Well
1 2 3 4 5 6 a 9 10 11 12 13 14 15
o+ 0 0 0 0 0 0 50 40 0 0 90 0 75 50
40 90 75 0 90 25 50
16 17
0 0
0 25
7
ET AL.
0 0 0
0 25 10 25
-
Tissue Staining
100 100 100 100 75 90 75 50 20 10 25 10 10 0 0
+ + + + + ++ ++ ++ ++ ++
100 (Signet 75 ring cells)
++ ++
* Tissue used for staining was obtained from specimens of the primary resection or initial biopsy except for Patient 5 in whom only liver metastasis was available. + 0 = none.
4 pure 2. lmmunoperoxidase stain using anti-CEA antibody in dilution of 1:500. A, moderately differentiated adenocarcinon sh ows strong (dark) staining of the glycocalyces. Individual infiltrating cells (arrow) show surface and cytoplasmic stainin B, poorly differentiated adenocarcinoma. Hyperplastic benign crypts (top right) show staining of the glycocalyces. The poor dit ‘ferentiated carcinoma shows no staining. Magnification X 200 (A and B). 248
August 1981
The American Journal of Medlclne
Volume 71
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
Research Laboratory [20]. Plasma CEA levels were obtained preoperatively for all patients and every one to four months postoperatively.
TABLE II
Poorly Differentiated Colorectal CancerRelatlon of Preoperative CEA and Stage Patients
RESULTS
Patient Follow-Up and Disease Recurrence.
Stage
Sev-
enteen patients had a median duration of follow-up of 48 months. Fourteen had “curative” resections, five with Dukes 6 and nine with Dukes C primary tumors. In eight of the 14 (57 percent) patients operated on for cure, disease recurred. The relapse rate by stage was one of five (20 percent) of patients with Dukes B tumor and seven of nine (77 percent) of patients with Dukes C tumor. Preoperative CEA assay was of limited prognostic value in this group since most patients had low preoperative levels (Table II) and, as shown in Table Ill, many of their tumors did not stain for CEA and presumably produced little or no CEA. Histologic Analysis (Table I)., Five tumors (four primary (Patients 1, 2, 3 and 16) and one metastatic (Patient 4)) showed no areas of glandular differentiation and were classified as 100 percent poorly differentiated. Tumors from Patients 1, 2 and 3 contained large tumor cells with a high degree of pleomorphism and a high
(W
O-2.5
CEA (nglml) 5.1-10 2.6-5.0
DUKES A DUKES B
0 5
4
1
-
DUKES C
9
4
3
1
DUKES D
3
2
-
17
10
Total
ET AL.
>20
-
4
1 1
1
2
mitotic rate. The tumor from Patient 16 was also totally undifferentiated but contained cells which were smaller with less nuclear pleomorphism as well as many tumor cells of the signet ring type (Figure 3). A sheet-like or trabecular growth pattern with extracellular mucin production was predominant. Tumor from Patient 17 also showed a similar morphology but contained gland-forming areas. The undifferentiated metastasis (Patient 4) was present in a lymph node obtained from an abdominoperineal resection of a large villous adenoma with foci of differentiated adenocarcinoma. Subsequently the same tumor (inguinal node biopsy)
Ither
Fig
are!a of same tumor using anti CEA antibody in dilution of 1:500. There are signet ring cells and extracellular mucin. Str vong sta ining is evident on the surface (upper arrow) and within the cytoplasm of the cells (lower arrow). Hematoxylin and e osin sta in (A), mucin does not stain; magnification X 400.
August 1991
The American Journalof Mediclne
Volume 71
249
PLASMA
CEA AND CEA TISSUE STAINING IN COLORECTAL
CANCER-GOSLIN
Plasma CEA Levels and CEA Tissue Staining
TABLE III
CEA (nglml) Relapse
Patient No.
Preop
Terminally
CEAlTissue
1’ 2 3 4 5
0.6 1.9 1.0 2.3 2.0
2.0 1.0 2.0 1.7
0.3 2.0 1.0 2.0 6.4+
-
6 7 8 9 10 11 12’ 13’ 14 15 16 17
7.0 0.0 0.6 3.2 2.8 5.0 1.7 370 1.1 45 0.7 1.3
6.8 DFt DF DF 390 DF
220 DF DF DF 410 DF 430 900 200 215 DF DF
a7 45 DF DF
Strongly positive (+-I-)
+ + + + + ++ ++ ++ ++ ++ ++ ++
CEA Localization and Tumor Morphology (Table I). Negative (-) CEA tissue staining: Four of five of the 100 percent poorly differentiated carcinomas (Patients 1, 2, 3 and 4) stained negatively for CEA. The specimen from Patient 5, which had a 25 percent moderately well differentiated component, also did not stain for CEA. This may have been a false negative, since the tissue consisted of a liver metastasis abtained at autopsy which had undergone significant autolysis. Weakly positive (i-) CEA tissue stain: In five carcinomas (Patients 6, 7, 8, 9 and 10) the stain for CEA was weakly positive. A wide variation in histology was seen in this group. The amount of poorly differentiated tissue and moderately differentiated tissue each varied between 10 and 90 percent. The positive stain was seen in the surface glycocalyx, apical cytoplasm or luminal
Preoperative CEA Level and CEA Tissue Staining
<2.5 >2.5
250
August 1981
5 0
Positive 6 6
The American Journal of Medicine
In seven
Strongly positive (++) staining associated with signet ring cells: Two tumors (Patients 16 and 17) were
recurred and also was totally undifferentiated. The remaining 12 tumors showed a range of differentiation. Four tumors (Patients 5, 6, 7 and 17) were 75 percent or more poorly differentiated. Tumor differentiation in the rest of the cases was quite variable (see Table I) and two tumors (Patients 14 and 15) had no poorly differentiated components.
Tissue CEA Negative
CEA tissue stain:
tumors the stain for CEA was strongly positive. Five of these tumors (Patients 11, 12, 13, 14 and 15) were composed predominantly of moderately differentiated or well differentiated tissue with correspondingly smaller amounts (25 percent or less) of poorly differentiated tissue. A positive stain was seen predominantly in the surface glycocalyx, apical cytoplasm and luminal secretions (Figure 2).
l
Preoperative CEA (nglml)
secretions in differentiated areas. A positive stain was also frequently observed in the nongland-forming areas, surrounding cell membranes of individual cells and sometimes diffusely in the cytoplasm. Extracellular mucin lakes did not stain for CEA in all cases studied.
Patients with Dukes D tumor. 7 Wary obstruction with bilirubin 18.0 mg. f DF = Disease-free with normal CEA.
TABLE IV
ET AL.
predominantly poorly differentiated with little or no gland formation yet both stained strongly for CEA. A positive stain was associated with the presence of numerous signet ring cells and was seen both on the surface membrane and within the cytoplasm (Figure 3). Extracellular mucin was also present in both cases but did not stain for CEA. Both patients are alive and well more than five years after resection, and are apparently cured.
Preoperative CEA and tissue staining (Table IV):
The
tumor tissues of all patients with preoperative plasma CEA levels >2.5 ng/ml had positive stains for CEA. In contrast, none of the patients whose primary tumors did not stain for CEA had preoperative plasma CEA levels greater than 2.5 ng/ml.
Correlation of CEA Tissue Staining and Plasma CEA in Patients with Dukes D Tumor (Table Ill). CEA tissue staining was available for all three patients with Dukes D colorectal cancer. Two patients (Patients 12 and 13) had liver metastases. Preoperative CEA levels were 1.7 and 370 ng/ml. Tumors of both patients stained for CEA (H-)_ The patient with the plasma CEA level of 1.7 ng/ml had no abnormalities on liver-spleen scan preoperatively and had minimal tumor in the liver at surgery. With disease progression in these patients, plasma CEA levels rose to 430 and 900 ng/ml, respectively. In both cases, CEA tissue staining identified those patients whose CEA level subsequently increased with progression of disease. The third patient (Patient 1) with a Dukes D lesion had a preoperative CEA level of 0.6 ng/ml with regional pelvic tumor extension. This tumor was found to be 100 percent poorly differentiated and did not stain for CEA. No disease outside the pelvis was documented one month later at the time of the patient’s death when a repeat CEA determination was 0.3 ng/ml.
Correlation of CEA Tissue Staining and the Use of Plasma CEA Levels in Monitoring Disease Recurrence (Table Ill). Among the 14 patients operated on for
Volume 71
PLASMA
57 YR. OLD
CEA AND CEA TISSUE STAINING
IN COLORECTAL
CANCER-GOSLIN
ET AL.
o+
@-SIG. 100% POORLY DIFFERENTIATED CEA TISSUE NEGATIVE
lOa= i p 4: Y 5”
4
l"M
-
1-
Metastases: Liver, Lunq
Pelvic Recur.
Surq. Resect.
t m
a 6 MONTHS
12 POST RESECTION
1'8
Serial CEA values in Case 1.
Figure 4.
cure, eight had a relapse after a median follow-up of 48 months. Positive CEA tissue staining of the primary tumors correlated with subsequent elevation of plasma CEA levels at or just prior to the time of documented tumor recurrence. In three patients, primary tumors did not stain for CEA. All three had a relapse with normal plasma CEA levels; in two (Patients 2 and 3) normal CEA levels persisted despite spread to liver, bone and lung.
In the third patient (Patient 5) a CEA level of less than 2.5 ng/ml was maintained despite liver and abdominal metastases until her last month of life when the CEA level peaked at 6.4 ng/ml. Biliary obstruction with a total bilirubin of 18 mg/ 100 ml was present at this time and probably accounted for the small elevation in the plasma CEA level on the basis of decreased liver clearance and not increased tumor CEA production [21-241.
90% POORLYDIFFERENTIATED CEA TISSUE POSITIVE (+) 106’
10,. Surg. Reset t .
54 YR. OLD of C-2 RECTO-SIG.
l--
1
1
3
Figure 3.
R
6
MhTttS
POST
RES-ECTION
-
serial C;tA values m L;ase z.
August lgS1
The American Journal of Medicine
Volume 71
251
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
Tissues of four of the patients (Patients 6, 10, 14 and 15) who had a relapse stained for CEA, and all had elevated plasma CEA levels preceding their relapse (6.8, 390, 87 and 45 ng/ml). As tumor progressed, plasma CEA values increased to 220,4 10,200 and 2 15 ng/ml, respectively (see Case 2). In the eighth patient who had a relapse, a difference was seen between the histology and tissue staining of the primary tumor and that shown in the regional lymph nodes. The primary tumor was a villous adenoma with foci of differentiated adenocarcinoma which stained for CEA (+-I-) whereas lymph nodes contained undifferentiated tumor which did not stain. The patient had a relapse one year later, the CEA level was 2.0 ng/ml, and lung, abdominal and inguinal node metastases were present. Interestingly, staining for CEA in the recurrent tumor (100 percent poorly differentiated) was completely negative and plasma CEA values never increased despite continued spread of the tumor. ILLUSTRATIVE
CASES
Case 1 (Figure 4). A 57 year old woman presented with a four month history of abdominal cramps and episodic blood in her stool. Barium enema disclosed a mass lesion in the sigmoid colon. At surgery a 100 percent poorly differentiated adenocarcinoma was resected. The tumor did not stain for CEA and the preoperative CEA level was 1.0 ng/ml. One year postoperatively, tumor recurred in the pelvis. Metastasis subsequently developed in the liver, lung and bone but CEA levels never rose above 1.O ng/ml. Comment: Negative CEA tissue staining predicted correctly the low CEA levels despite widespread metastatic disease. Case 2 (Figure 5). A 54 year old woman with four months of increasing constipation presented with intestinal obstruction and a preoperative CEA level of 7.0 ng/ml. A Dukes C adenocarcinoma of the sigmoid colon was resected which was 90 percent poorly differentiated and CEA tissue positive (+)_ Postoperatively the CEA decreased to 2.0 ng/ml, but five months later it reached 7.0 ng/ml. Seven months postoperatively, liver scan confirmed metastasis; an abdominal mass was also present. There was no response to the administration of 5-fluorouracil. CEA increased to 220 ng/ml prior to death. Comment: Positive staining for CEA predicted that plasma CEA would be useful in monitoring the course of disease.
differentiated” include those in which poorly differentiated areas are a minor component as well as those which are totally undifferentiated. It is not surprising then that tumors which differ histologically should behave differently clinically and produce different plasma levels of CEA. Plasma CEA levels provide effective monitoring of patients with well differentiated and moderately well differentiated colorectal cancer [4,14], and most of our patients had some differentiated component present in tumor. All patients with “poorly differentiated” tumors which stained positively for CEA produced CEA, and their plasma levels were useful prognostically and in monitoring their disease. In contrast, in patients with totally or almost totally undifferentiated tumors and nonstaining tissue, a normal preoperative CEA level was of no prognostic significance, and serial CEA levels did not predict relapse or rise in the presence of metastatic disease. Undifferentiated colorectal tumors composed of signet ring cells, however, stained for CEA. They must be distinguished from other undifferentiated tumors since they can be reliably monitored by plasma CEA levels. Whether staining for CEA reflects cytologic differentiation in these cells or has prognostic significance is not yet clear and will require further study. In patients with poorly differentiated tumors who have elevated CEA levels at some time in the course of their disease either preoperatively or with recurrence (without obvious extrahepatic biliary obstruction or intrahepatic parenchymal failure), the increase in plasma CEA reflects tumor CEA production and indicates that the plasma CEA level will be a useful monitor of disease. The plasma CEA level can also be assumed to be a useful marker in patients whose tumors have well differentiated components. In these patients, CEA tissue staining is not needed as an additional predictor of the usefulness of the plasma CEA assay. Nevertheless, the presence or absence of CEA on staining of the primary tumor, regardless of the preoperative plasma CEA level in patients with uniformly poorly differentiated tumor histology, is of value in predicting the usefulness of the CEA level as a monitor of disease recurrence or progression. We recommend that in such patients, serial CEA levels be obtained only when tissue staining of the primary lesion is positive for CEA. l
The World Health Organization classification of colorectal cancer has recommended distinguishing histologically signet ring cells from undifferentiated cells [25]. l
COMMENTS Tumors commonly
ET AL.
labelled
by pathologists
as “poorly
REFERENCES 1.
252
Rau P, Steele G, Mayer R, Wilson RE, Zamcheck N: The rate of postoperative CEA rise in selecting treatment for recurrent colorectal cancer. Gastroenterology 1979; 76: 1223.
August 1981
The American Journal of Medicine
2.
3.
Volume 71
Steele G, Zamcheck N, Wilson RE. Mayer R, Rau P. Maltz J: Results of CEA-initiated “second-look” surgery for colorectal cancer. Am J Surg 1960; 139: 5444548. Martin E, Minton MS: Radioimmunoassay of CEA in extracts
PLASMA CEA AND CEA TISSUE STAINING IN CDLORECTAL CANCER-GOSLIN ET AL.
10.
11.
12.
13.
14.
of human colon and stomach. Int J Cancer 1972; 9: 641-647. Sugarbaker P, Skarin A, Zamcheck N: Patterns of serial CEA assay and their clinical use in management of colorectal cancer. J Surg Oncol 1976; 8: 523-537. Sugarbaker PH, Zamcheck N, Moore FD: Assessment of serial CEA assays in post-operative detection of recurrent colorectal cancer. Cancer 1976; 38: 2310-2315. Mach JP, Jaeger PH. Bertholet MM, Ruegsegger CH, Loosli RM, Pettavel J: Detection of recurrence of large bowel carcinoma by radioimmunoassay of circulating CEA. Lancet 1974; 2: 535. Sorokin JJ, Sugarbaker PH, Zamcheck N, Pisick M, Kupchik HZ, Moore ED: Serial CEA assays used in detection of cancer recurrence. JAMA 1974; 288: 49. MacKay AM, Pate1 S, Carter S: Role of serial CEA assays in the detection of metastatic colorectal cancer. Br Med J 1974; 4: 302. Mayer RJ, Garnick MB, Steele GD, Zamcheck N: Carcinoembryonic antigen (CEA) as a monitor of chemotherapy in disseminated colorectal cancer. Cancer 1978; 42: 1428-1433. Ravry M, Moertel CG, Schutt AJ, Go VLW: Usefulness of serial carcinoembryonic antigen (CEA) determinations during anti cancer therapy on long term follow up of gastrointestinal carcinoma. Cancer 1974; 37: 1230-1234. Zamcheck N, Doos WB, Prudente R, Lurie BB, Gottlieb LS: Prognostic factors in colon cancer: correlation of serum CEA levels and tumor histopathology. Human Pathol 1975; 6: 31-45. Herrera M, Chu TM, Holyoke ED: CEA as a prognostic and monitoring test in clinically complete resection of colorectal cancer. Ann Surg 1976; 183: 5-9. Wanebo HJ, Rao B, Pinsky CM. Hoffman RE, et al.: Preoperative carcinoembryonic antigen level as a prognostic indicator in colorectal cancer. N Engl J Med 1978; 299: 448-45 1. Goslin RH, Mayer RJ, Zamcheck N: The use of preoperative plasma carcinoembryonic antigen (CEA) levels for the
15. 16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
August 1981
stratification of patients after curative resection of colorectal cancer. Proc Am Assoc Cancer Res 1980; 21: 141. Dukes CE: The classification of cancer of the rectum. J Pathol Bacterial 1932; 35: 323-332. Kirklin JW, Dockerty MB, Waugh JM: The role of the peritoneal reflection in the prognosis of carcinomas of the rectum and sigmoid colon. Surg Gynecol Obtet 1949; 88: 326-331. Astler VB, Caller FA: The prognostic significance of direct extension of cancer of the colon. Ann Surg 1954; 139: 846. Hsu KG, Zimmerman EA. Rudin I, LoGerfo P, Bennett S, Tannenbaum M: Application of immunoperoxidase bridge technique in studies of tumor tissues with antibody to tumor associated antigen (TAA). In: Anderson NG, Coggins JH, Jr., eds. Proceedings of the Second Conference on Embryonic and Fetal Antigens in Cancer. Vol. 2. Springfield, VA. U.S. Dept. of Commerce. 1972; 147-152. Mason TE. Phifer RE, Spicer SS, Swallow RA, Dreskin RB: An immunoglobulinenzyme bridge method for localizing tissue antigens. J Histochem Cytochem 1969; 17: 563-569. Kupchik HZ, Zamcheck N. Saravis CA: lmmunochemical studies of carcinoembryonic antigens. JNCI 1973; 51: 1741-1749. Lurie BB, Loewenstein MS, Zamcheck N: Elevated carcinoembryonic antigen levels and biliary tract obstruction. JAMA 1975; 233: 326-330. Loewenstein MS, Zamcheck N: Carcinoembryonic antigen (CEA) and the liver. Gastroenterology 1977; 72: 161166. Loewenstein MS, Zamcheck N: Carcinoembryonic antigen (CEA) levels in benign gastrointestinal disease states. Cancer 1978; 42: 1412-1418. O’Brien MJ, Bronstein B, Zamcheck N, Saravis C, Burke B, Gottlieb LS: Cholestasis and hepatic metastases: a factor contributing to extreme elevations of carcinoembryonic antigen. JNCI 1980; 64: 1291-1294. Morson BC, Sobin LH: International classification of intestinal tumors. World Health Organization: Geneva, 1976.
The American Journal ot Medlclne
Volume 71
253
Ft. GOSLIN,
M.D.
M. J. O’BRIEN, G. STEELE,
M.D.
R. MAYER,
M.D.
R. WILSON,
M.B.
M.D.
J. M. CORSON,
M.D.
N. ZAMCHECK,
M.D.
Boston. Massachusetts
From the Mallory Gastrointestinal Research Laboratory, Boston City Hospital; Division of Medical Oncology, Sidney Farber Cancer Institute; Departments of Surgery and Pathology, Peter Bent Brigham Hospital; Departments of Medicine, Surgery and Pathology, Harvard Medical School and the Department of Pathology, Boston University School of Medicine. This study was supported by Grants (CA-04486 (N.Z.) and CA-22513 (G.S.) from the National Cancer Institute. Requests for reprints should be addressed to Dr. Norman Zamcheck. Mallory Gastrointestinal Research Laboratory, Boston City Hospital, Boston, MA 02118. Manuscript accepted February 26, 1981.
246
August 1981
A review of patients operated on for colorectal cancer disclosed poorly differentiated histologies in those whose preoperative and postoperative carcinoembryonic antigen (CEA) levels were not elevated, even in the presence of metastatic disease. CEA was, therefore, of little prognostic value or predictive of disease recurrence in these patients. The amount of CEA in tumor tissue of 17 patients with poorly differentiated colorectal cancer was estimated with the immunoperoxidase staining technique and was correlated with histology and plasma CEA levels obtained during various stages of disease. Five tumors did not stain negatively for CEA and all had predominantly poorly differentiated histologies. In all of these patients metastatic disease developed but not elevated plasma CEA levels. in contrast 12 tumors stained positively for CEA and were found to contain either differentiated areas or signet ring cells. Serial plasma CEA levels correctly monitored the postoperative courses of all 12 patients. Six of these had a relapse and all were detected by serial increases in plasma CEA. The remaining six were disease free >48 months after resection and had normal plasma CEA levels. Among poorly differentiated tumors, those that contain glandforming areas or signet ring cells can be assumed to produce CEA, and plasma CEA levels can be used effectively for monitoring. On the other hand, undifferentiated tumors which do not stain for CEA identify those patients whose plasma CEA levels do not provide a useful monitor. Circulating levels of CEA have proved to be useful in the management of colorectal cancer and are widely accepted as a monitor for recurrence of disease after primary tumor resection [l-8], and of response to treatment in patients with known metastatic disease [9,10]. Determining the presence of CEA preoperatively has been shown to be of prognostic value when used in conjunction with Dukes staging and may be helpful in adjuvant chemo- and immunotherapy trials by identifying those patients at greatest risk for recurrence after “curative” resection of their primary tumors [ 1 l-131. In some patients, however, plasma CEA levels do not correlate with the clinical course. In a recent review of patients operated on for colorectal cancer, we found that all the patients whose CEA levels did not correctly monitor the clinical course had “poorly differentiated” carcinomas [ 141. We have now analyzed 17 cases of poorly differentiated colorectal adenocarcinoma (1) to learn what factors accounted for variation in plasma CEA levels and (2) to correlate elevations in plasma CEA levels with the amount and extent of tissue CEA
The American Journal of Medlclne
Volume 71
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
immunocytochemically mors.
localized
in the primary
tu-
MATERIALS AND METHODS Patients. Preoperative and serial postoperative CEA levels were available in 145 patients treated surgically for colorectal cancer at the Peter Bent Brigham Hospital between 1972 and 1976 [ 141. In 23 of these patients the cancers were pathologically classified as “poorly differentiated,” and 17 have now been studied further. Four of the 23 patients were excluded because tissue blocks were unavailable for histologic staining; two others were excluded after review of the histology showed the carcinomas to be well differentiated. Fourteen of these 17 patients underwent “curative” resections. The three patients with unresectable lesions were operated on for local control and for tissue diagnosis. Six of the tumors were located in the right colon, one in the transverse colon, two in the left colon, two in the sigmoid colon and six in the rectum. All patients had preoperative plasma CEA determinations, liver function tests (serum glutamic oxaloacetic transaminase, lactic dehydrogenase, bilirubin, alkaline phosphatase) and liver scans. Asymptomatic patients with no evidence of metastatic disease were followed postoperatively with serial CEA deterinations, liver function tests and physical examinations at least every four months. Follow-up chest films, barium enemas and liverspleen scans were obtained annually. In patients with abnormal liver functiontests or rises in CEA levels (>2.5 ng/ml), the tests were repeated within one week. Patients with persistently abnormal liver function tests had liver scans; in some cases, liver biopsy was performed to document me&static tumor. Patients with rises in CEA levels had repeat CEA determinations every two to four weeks; if no other cause for the elevated CEA level was found and the rises persisted, a metastatic work-up was begun. If m&static disease was documented, serial CEA levels and liver function tests were obtained at least monthly. Scans and roentgenograms were repeated to assist in management and to document response or lack of response to treatment. Pathology. All tissue sections initially reported as “poorly” differentiated were classified on review either as “poorly” or “moderate to poorly” differentiated tumors by the same “reference” pathologist (M.J.O.) and assigned a Dukes classification [ 15-171. Three histologic sections of each primary tumor were reviewed without knowledge of the patient’s clinical course or plasma CEA levels. It was apparent that a range of differentiation was present in most tumors. An estimate was made of the percentage of tumor tissue which was well differentiated, moderately differentiated or poorly differentiated (Figure 1). Well differentiated tumor was defined as showing a glandular pattern which was cytoiogitally differentiated; poorly differentiated tumors showed cytologic anaplasia and no gland formation. Moderately differentiated tumor was defined as showing cytologic anaplasia and glandular differentiation. The presence of extra- and intracellular mucin was also noted. immunohistochemistry. Following initial review of the histologic sections, blocks were selected which were representative of the tumor morphology. Blocks were of the primary tumor except in Patient 5 (Table I) in whom only
August
ET AL.
metastatic tumor was available for immunohistochemical staining. Five micron paraffin sections were incubated in sequence with rabbit anti-CEA (New England Immunology Associates, Cambridge, MA) at 1:500 dilution, swine antirabbit immunoglobulin G (Dako distributed by Accurate Chemical and Scientific Corp., Hicksville, NY) at 1:25 dilution and peroxidase antiperoxidase (Dako) at 1: 100 dilution. The CEA antibody was rendered monospecific by absorption with groups A, B and 0 cells and iyophilized extracts of human lung (40 mg/ml), buffy coat (10 mg/ml), spleen (10 mg/ml), liver (10 mg/ml) and colon (10 mg/ml). Following absorption, the antibody was considered monospecific for CEA when (1)
A, well differentiated adenocarcinoma-gland formation without cytologic anaplasia. 6, moderately differentiated adanocarcinoma-gland formation and cflologic anaplasia. C, poorly differentiated adenocarcinoma-cytologic anaplasia without gland formation. Hematoxylin and eosin stain, magnification X 100 (A and 6) and X 200 (C).
Flgure 1.
lggl
The American
Journal
of Medlclne
Volume 71
247
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
TABLE I
CEA Tissue Staining of Colorectal Cancers* in Reference to Degree of
by double-immunodiffusion it reacted with purified CEA, (2) gave a single line against perchlorate extract of colon car-
Differentiation
cinoma metastases, and (3) showed no reaction against the tissue extracts used for absorption. Paraffin sections were cut and immunocytochemical localization of CEA was performed using the immunoperoxidase peroxidase antiperoxidase technique [ 18,191. The incubation time for the primary antibody was 24 hours at 0 to 4% followed by room temperature for 2 hours, and 30 minutes at room temperature for both the bridge antibody and the peroxidase antiperoxidase. The stained sections were evaluated for the amount of CEA present and its location. This was accomplished without knowledge of the patient’s clinical course or plasma CEA level. Tumors showing widespread strong staining were designated as strongly positive (+-I); when showing uniformly weak staining, weakly positive (+), or when showing staining with focal positive areas, negative (Figure 2). Sections that did not stain for CEA were designated as negative (-). Whether the positive stains were predominantly cytoplasmic, surface or both was determined in each case. Controls for the staining procedure consisted of replacement of the primary antibody with normal rabbit serum and with anti-CEA antiserum absorbed with purified CEA. A section of well differentiated colonic adenocarcinoma was stained as a standard positive control. CEA Assay. Plasma CEA assays were performed by the modified Hansen Z-gel method at the Mallory Gastrointestinal
CEA Patient
(%) DifferentIalIon Moderately Poorly
No.
Well
1 2 3 4 5 6 a 9 10 11 12 13 14 15
o+ 0 0 0 0 0 0 50 40 0 0 90 0 75 50
40 90 75 0 90 25 50
16 17
0 0
0 25
7
ET AL.
0 0 0
0 25 10 25
-
Tissue Staining
100 100 100 100 75 90 75 50 20 10 25 10 10 0 0
+ + + + + ++ ++ ++ ++ ++
100 (Signet 75 ring cells)
++ ++
* Tissue used for staining was obtained from specimens of the primary resection or initial biopsy except for Patient 5 in whom only liver metastasis was available. + 0 = none.
4 pure 2. lmmunoperoxidase stain using anti-CEA antibody in dilution of 1:500. A, moderately differentiated adenocarcinon sh ows strong (dark) staining of the glycocalyces. Individual infiltrating cells (arrow) show surface and cytoplasmic stainin B, poorly differentiated adenocarcinoma. Hyperplastic benign crypts (top right) show staining of the glycocalyces. The poor dit ‘ferentiated carcinoma shows no staining. Magnification X 200 (A and B). 248
August 1981
The American Journal of Medlclne
Volume 71
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
Research Laboratory [20]. Plasma CEA levels were obtained preoperatively for all patients and every one to four months postoperatively.
TABLE II
Poorly Differentiated Colorectal CancerRelatlon of Preoperative CEA and Stage Patients
RESULTS
Patient Follow-Up and Disease Recurrence.
Stage
Sev-
enteen patients had a median duration of follow-up of 48 months. Fourteen had “curative” resections, five with Dukes 6 and nine with Dukes C primary tumors. In eight of the 14 (57 percent) patients operated on for cure, disease recurred. The relapse rate by stage was one of five (20 percent) of patients with Dukes B tumor and seven of nine (77 percent) of patients with Dukes C tumor. Preoperative CEA assay was of limited prognostic value in this group since most patients had low preoperative levels (Table II) and, as shown in Table Ill, many of their tumors did not stain for CEA and presumably produced little or no CEA. Histologic Analysis (Table I)., Five tumors (four primary (Patients 1, 2, 3 and 16) and one metastatic (Patient 4)) showed no areas of glandular differentiation and were classified as 100 percent poorly differentiated. Tumors from Patients 1, 2 and 3 contained large tumor cells with a high degree of pleomorphism and a high
(W
O-2.5
CEA (nglml) 5.1-10 2.6-5.0
DUKES A DUKES B
0 5
4
1
-
DUKES C
9
4
3
1
DUKES D
3
2
-
17
10
Total
ET AL.
>20
-
4
1 1
1
2
mitotic rate. The tumor from Patient 16 was also totally undifferentiated but contained cells which were smaller with less nuclear pleomorphism as well as many tumor cells of the signet ring type (Figure 3). A sheet-like or trabecular growth pattern with extracellular mucin production was predominant. Tumor from Patient 17 also showed a similar morphology but contained gland-forming areas. The undifferentiated metastasis (Patient 4) was present in a lymph node obtained from an abdominoperineal resection of a large villous adenoma with foci of differentiated adenocarcinoma. Subsequently the same tumor (inguinal node biopsy)
Ither
Fig
are!a of same tumor using anti CEA antibody in dilution of 1:500. There are signet ring cells and extracellular mucin. Str vong sta ining is evident on the surface (upper arrow) and within the cytoplasm of the cells (lower arrow). Hematoxylin and e osin sta in (A), mucin does not stain; magnification X 400.
August 1991
The American Journalof Mediclne
Volume 71
249
PLASMA
CEA AND CEA TISSUE STAINING IN COLORECTAL
CANCER-GOSLIN
Plasma CEA Levels and CEA Tissue Staining
TABLE III
CEA (nglml) Relapse
Patient No.
Preop
Terminally
CEAlTissue
1’ 2 3 4 5
0.6 1.9 1.0 2.3 2.0
2.0 1.0 2.0 1.7
0.3 2.0 1.0 2.0 6.4+
-
6 7 8 9 10 11 12’ 13’ 14 15 16 17
7.0 0.0 0.6 3.2 2.8 5.0 1.7 370 1.1 45 0.7 1.3
6.8 DFt DF DF 390 DF
220 DF DF DF 410 DF 430 900 200 215 DF DF
a7 45 DF DF
Strongly positive (+-I-)
+ + + + + ++ ++ ++ ++ ++ ++ ++
CEA Localization and Tumor Morphology (Table I). Negative (-) CEA tissue staining: Four of five of the 100 percent poorly differentiated carcinomas (Patients 1, 2, 3 and 4) stained negatively for CEA. The specimen from Patient 5, which had a 25 percent moderately well differentiated component, also did not stain for CEA. This may have been a false negative, since the tissue consisted of a liver metastasis abtained at autopsy which had undergone significant autolysis. Weakly positive (i-) CEA tissue stain: In five carcinomas (Patients 6, 7, 8, 9 and 10) the stain for CEA was weakly positive. A wide variation in histology was seen in this group. The amount of poorly differentiated tissue and moderately differentiated tissue each varied between 10 and 90 percent. The positive stain was seen in the surface glycocalyx, apical cytoplasm or luminal
Preoperative CEA Level and CEA Tissue Staining
<2.5 >2.5
250
August 1981
5 0
Positive 6 6
The American Journal of Medicine
In seven
Strongly positive (++) staining associated with signet ring cells: Two tumors (Patients 16 and 17) were
recurred and also was totally undifferentiated. The remaining 12 tumors showed a range of differentiation. Four tumors (Patients 5, 6, 7 and 17) were 75 percent or more poorly differentiated. Tumor differentiation in the rest of the cases was quite variable (see Table I) and two tumors (Patients 14 and 15) had no poorly differentiated components.
Tissue CEA Negative
CEA tissue stain:
tumors the stain for CEA was strongly positive. Five of these tumors (Patients 11, 12, 13, 14 and 15) were composed predominantly of moderately differentiated or well differentiated tissue with correspondingly smaller amounts (25 percent or less) of poorly differentiated tissue. A positive stain was seen predominantly in the surface glycocalyx, apical cytoplasm and luminal secretions (Figure 2).
l
Preoperative CEA (nglml)
secretions in differentiated areas. A positive stain was also frequently observed in the nongland-forming areas, surrounding cell membranes of individual cells and sometimes diffusely in the cytoplasm. Extracellular mucin lakes did not stain for CEA in all cases studied.
Patients with Dukes D tumor. 7 Wary obstruction with bilirubin 18.0 mg. f DF = Disease-free with normal CEA.
TABLE IV
ET AL.
predominantly poorly differentiated with little or no gland formation yet both stained strongly for CEA. A positive stain was associated with the presence of numerous signet ring cells and was seen both on the surface membrane and within the cytoplasm (Figure 3). Extracellular mucin was also present in both cases but did not stain for CEA. Both patients are alive and well more than five years after resection, and are apparently cured.
Preoperative CEA and tissue staining (Table IV):
The
tumor tissues of all patients with preoperative plasma CEA levels >2.5 ng/ml had positive stains for CEA. In contrast, none of the patients whose primary tumors did not stain for CEA had preoperative plasma CEA levels greater than 2.5 ng/ml.
Correlation of CEA Tissue Staining and Plasma CEA in Patients with Dukes D Tumor (Table Ill). CEA tissue staining was available for all three patients with Dukes D colorectal cancer. Two patients (Patients 12 and 13) had liver metastases. Preoperative CEA levels were 1.7 and 370 ng/ml. Tumors of both patients stained for CEA (H-)_ The patient with the plasma CEA level of 1.7 ng/ml had no abnormalities on liver-spleen scan preoperatively and had minimal tumor in the liver at surgery. With disease progression in these patients, plasma CEA levels rose to 430 and 900 ng/ml, respectively. In both cases, CEA tissue staining identified those patients whose CEA level subsequently increased with progression of disease. The third patient (Patient 1) with a Dukes D lesion had a preoperative CEA level of 0.6 ng/ml with regional pelvic tumor extension. This tumor was found to be 100 percent poorly differentiated and did not stain for CEA. No disease outside the pelvis was documented one month later at the time of the patient’s death when a repeat CEA determination was 0.3 ng/ml.
Correlation of CEA Tissue Staining and the Use of Plasma CEA Levels in Monitoring Disease Recurrence (Table Ill). Among the 14 patients operated on for
Volume 71
PLASMA
57 YR. OLD
CEA AND CEA TISSUE STAINING
IN COLORECTAL
CANCER-GOSLIN
ET AL.
o+
@-SIG. 100% POORLY DIFFERENTIATED CEA TISSUE NEGATIVE
lOa= i p 4: Y 5”
4
l"M
-
1-
Metastases: Liver, Lunq
Pelvic Recur.
Surq. Resect.
t m
a 6 MONTHS
12 POST RESECTION
1'8
Serial CEA values in Case 1.
Figure 4.
cure, eight had a relapse after a median follow-up of 48 months. Positive CEA tissue staining of the primary tumors correlated with subsequent elevation of plasma CEA levels at or just prior to the time of documented tumor recurrence. In three patients, primary tumors did not stain for CEA. All three had a relapse with normal plasma CEA levels; in two (Patients 2 and 3) normal CEA levels persisted despite spread to liver, bone and lung.
In the third patient (Patient 5) a CEA level of less than 2.5 ng/ml was maintained despite liver and abdominal metastases until her last month of life when the CEA level peaked at 6.4 ng/ml. Biliary obstruction with a total bilirubin of 18 mg/ 100 ml was present at this time and probably accounted for the small elevation in the plasma CEA level on the basis of decreased liver clearance and not increased tumor CEA production [21-241.
90% POORLYDIFFERENTIATED CEA TISSUE POSITIVE (+) 106’
10,. Surg. Reset t .
54 YR. OLD of C-2 RECTO-SIG.
l--
1
1
3
Figure 3.
R
6
MhTttS
POST
RES-ECTION
-
serial C;tA values m L;ase z.
August lgS1
The American Journal of Medicine
Volume 71
251
PLASMA CEA AND CEA TISSUE STAINING IN COLORECTAL CANCER-GOSLIN
Tissues of four of the patients (Patients 6, 10, 14 and 15) who had a relapse stained for CEA, and all had elevated plasma CEA levels preceding their relapse (6.8, 390, 87 and 45 ng/ml). As tumor progressed, plasma CEA values increased to 220,4 10,200 and 2 15 ng/ml, respectively (see Case 2). In the eighth patient who had a relapse, a difference was seen between the histology and tissue staining of the primary tumor and that shown in the regional lymph nodes. The primary tumor was a villous adenoma with foci of differentiated adenocarcinoma which stained for CEA (+-I-) whereas lymph nodes contained undifferentiated tumor which did not stain. The patient had a relapse one year later, the CEA level was 2.0 ng/ml, and lung, abdominal and inguinal node metastases were present. Interestingly, staining for CEA in the recurrent tumor (100 percent poorly differentiated) was completely negative and plasma CEA values never increased despite continued spread of the tumor. ILLUSTRATIVE
CASES
Case 1 (Figure 4). A 57 year old woman presented with a four month history of abdominal cramps and episodic blood in her stool. Barium enema disclosed a mass lesion in the sigmoid colon. At surgery a 100 percent poorly differentiated adenocarcinoma was resected. The tumor did not stain for CEA and the preoperative CEA level was 1.0 ng/ml. One year postoperatively, tumor recurred in the pelvis. Metastasis subsequently developed in the liver, lung and bone but CEA levels never rose above 1.O ng/ml. Comment: Negative CEA tissue staining predicted correctly the low CEA levels despite widespread metastatic disease. Case 2 (Figure 5). A 54 year old woman with four months of increasing constipation presented with intestinal obstruction and a preoperative CEA level of 7.0 ng/ml. A Dukes C adenocarcinoma of the sigmoid colon was resected which was 90 percent poorly differentiated and CEA tissue positive (+)_ Postoperatively the CEA decreased to 2.0 ng/ml, but five months later it reached 7.0 ng/ml. Seven months postoperatively, liver scan confirmed metastasis; an abdominal mass was also present. There was no response to the administration of 5-fluorouracil. CEA increased to 220 ng/ml prior to death. Comment: Positive staining for CEA predicted that plasma CEA would be useful in monitoring the course of disease.
differentiated” include those in which poorly differentiated areas are a minor component as well as those which are totally undifferentiated. It is not surprising then that tumors which differ histologically should behave differently clinically and produce different plasma levels of CEA. Plasma CEA levels provide effective monitoring of patients with well differentiated and moderately well differentiated colorectal cancer [4,14], and most of our patients had some differentiated component present in tumor. All patients with “poorly differentiated” tumors which stained positively for CEA produced CEA, and their plasma levels were useful prognostically and in monitoring their disease. In contrast, in patients with totally or almost totally undifferentiated tumors and nonstaining tissue, a normal preoperative CEA level was of no prognostic significance, and serial CEA levels did not predict relapse or rise in the presence of metastatic disease. Undifferentiated colorectal tumors composed of signet ring cells, however, stained for CEA. They must be distinguished from other undifferentiated tumors since they can be reliably monitored by plasma CEA levels. Whether staining for CEA reflects cytologic differentiation in these cells or has prognostic significance is not yet clear and will require further study. In patients with poorly differentiated tumors who have elevated CEA levels at some time in the course of their disease either preoperatively or with recurrence (without obvious extrahepatic biliary obstruction or intrahepatic parenchymal failure), the increase in plasma CEA reflects tumor CEA production and indicates that the plasma CEA level will be a useful monitor of disease. The plasma CEA level can also be assumed to be a useful marker in patients whose tumors have well differentiated components. In these patients, CEA tissue staining is not needed as an additional predictor of the usefulness of the plasma CEA assay. Nevertheless, the presence or absence of CEA on staining of the primary tumor, regardless of the preoperative plasma CEA level in patients with uniformly poorly differentiated tumor histology, is of value in predicting the usefulness of the CEA level as a monitor of disease recurrence or progression. We recommend that in such patients, serial CEA levels be obtained only when tissue staining of the primary lesion is positive for CEA. l
The World Health Organization classification of colorectal cancer has recommended distinguishing histologically signet ring cells from undifferentiated cells [25]. l
COMMENTS Tumors commonly
ET AL.
labelled
by pathologists
as “poorly
REFERENCES 1.
252
Rau P, Steele G, Mayer R, Wilson RE, Zamcheck N: The rate of postoperative CEA rise in selecting treatment for recurrent colorectal cancer. Gastroenterology 1979; 76: 1223.
August 1981
The American Journal of Medicine
2.
3.
Volume 71
Steele G, Zamcheck N, Wilson RE. Mayer R, Rau P. Maltz J: Results of CEA-initiated “second-look” surgery for colorectal cancer. Am J Surg 1960; 139: 5444548. Martin E, Minton MS: Radioimmunoassay of CEA in extracts
PLASMA CEA AND CEA TISSUE STAINING IN CDLORECTAL CANCER-GOSLIN ET AL.
10.
11.
12.
13.
14.
of human colon and stomach. Int J Cancer 1972; 9: 641-647. Sugarbaker P, Skarin A, Zamcheck N: Patterns of serial CEA assay and their clinical use in management of colorectal cancer. J Surg Oncol 1976; 8: 523-537. Sugarbaker PH, Zamcheck N, Moore FD: Assessment of serial CEA assays in post-operative detection of recurrent colorectal cancer. Cancer 1976; 38: 2310-2315. Mach JP, Jaeger PH. Bertholet MM, Ruegsegger CH, Loosli RM, Pettavel J: Detection of recurrence of large bowel carcinoma by radioimmunoassay of circulating CEA. Lancet 1974; 2: 535. Sorokin JJ, Sugarbaker PH, Zamcheck N, Pisick M, Kupchik HZ, Moore ED: Serial CEA assays used in detection of cancer recurrence. JAMA 1974; 288: 49. MacKay AM, Pate1 S, Carter S: Role of serial CEA assays in the detection of metastatic colorectal cancer. Br Med J 1974; 4: 302. Mayer RJ, Garnick MB, Steele GD, Zamcheck N: Carcinoembryonic antigen (CEA) as a monitor of chemotherapy in disseminated colorectal cancer. Cancer 1978; 42: 1428-1433. Ravry M, Moertel CG, Schutt AJ, Go VLW: Usefulness of serial carcinoembryonic antigen (CEA) determinations during anti cancer therapy on long term follow up of gastrointestinal carcinoma. Cancer 1974; 37: 1230-1234. Zamcheck N, Doos WB, Prudente R, Lurie BB, Gottlieb LS: Prognostic factors in colon cancer: correlation of serum CEA levels and tumor histopathology. Human Pathol 1975; 6: 31-45. Herrera M, Chu TM, Holyoke ED: CEA as a prognostic and monitoring test in clinically complete resection of colorectal cancer. Ann Surg 1976; 183: 5-9. Wanebo HJ, Rao B, Pinsky CM. Hoffman RE, et al.: Preoperative carcinoembryonic antigen level as a prognostic indicator in colorectal cancer. N Engl J Med 1978; 299: 448-45 1. Goslin RH, Mayer RJ, Zamcheck N: The use of preoperative plasma carcinoembryonic antigen (CEA) levels for the
15. 16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
August 1981
stratification of patients after curative resection of colorectal cancer. Proc Am Assoc Cancer Res 1980; 21: 141. Dukes CE: The classification of cancer of the rectum. J Pathol Bacterial 1932; 35: 323-332. Kirklin JW, Dockerty MB, Waugh JM: The role of the peritoneal reflection in the prognosis of carcinomas of the rectum and sigmoid colon. Surg Gynecol Obtet 1949; 88: 326-331. Astler VB, Caller FA: The prognostic significance of direct extension of cancer of the colon. Ann Surg 1954; 139: 846. Hsu KG, Zimmerman EA. Rudin I, LoGerfo P, Bennett S, Tannenbaum M: Application of immunoperoxidase bridge technique in studies of tumor tissues with antibody to tumor associated antigen (TAA). In: Anderson NG, Coggins JH, Jr., eds. Proceedings of the Second Conference on Embryonic and Fetal Antigens in Cancer. Vol. 2. Springfield, VA. U.S. Dept. of Commerce. 1972; 147-152. Mason TE. Phifer RE, Spicer SS, Swallow RA, Dreskin RB: An immunoglobulinenzyme bridge method for localizing tissue antigens. J Histochem Cytochem 1969; 17: 563-569. Kupchik HZ, Zamcheck N. Saravis CA: lmmunochemical studies of carcinoembryonic antigens. JNCI 1973; 51: 1741-1749. Lurie BB, Loewenstein MS, Zamcheck N: Elevated carcinoembryonic antigen levels and biliary tract obstruction. JAMA 1975; 233: 326-330. Loewenstein MS, Zamcheck N: Carcinoembryonic antigen (CEA) and the liver. Gastroenterology 1977; 72: 161166. Loewenstein MS, Zamcheck N: Carcinoembryonic antigen (CEA) levels in benign gastrointestinal disease states. Cancer 1978; 42: 1412-1418. O’Brien MJ, Bronstein B, Zamcheck N, Saravis C, Burke B, Gottlieb LS: Cholestasis and hepatic metastases: a factor contributing to extreme elevations of carcinoembryonic antigen. JNCI 1980; 64: 1291-1294. Morson BC, Sobin LH: International classification of intestinal tumors. World Health Organization: Geneva, 1976.
The American Journal ot Medlclne
Volume 71
253