Carcinoembryonic antigen: an invaluable marker for advanced breast cancer.KA Pathak, RR Khanna, HD Khanna, SS Khanna, SS Gupta, NN Khanna
Department of Surgery, Biophysics and Pathology, Banaras Hindu University, Varanasi.
Serial serum Carcinoembryonic antigen (CEA) levels were measured in 150 individuals (50 patients with breast cancer, 50 benign breast diseases and 50 other controls). These levels were correlated with clinicopathological parameters and follow-up information. Serum CEA levels were independent of the primary tumor status, their histology, lymphoreticular response and the patients' characteristics as well as the age, sex and the menstrual status. However, the nodal status, number of involved nodes and the grade of the tumors had significant influence on the level of serum CEA. Breast cancer patients especially those with metastasis had significantly higher serum CEA levels as compared to the controls and those with localised disease, irrespective of the site of metastasis. These levels were lowered appreciably by the disease regression and were raised or stable during the disease progression. Receiver operating characteristic (ROC) curve showed metastasis to be more frequent in patients with pretreatment serum CEA levels above 25 ng/ml and persistent post treatment CEA levels above 15 ng/ml. Serum CEA level was found to be a valuable prognostic indicator for advanced breast cancer and serial serum CEA levels provided an average lead time of about 3.9 months before the clinical appearance of metastasis.
Keywords: Adult, Breast Diseases, blood,immunology,Breast Neoplasms, blood,immunology,therapy,Carcinoembryonic Antigen, blood,Case-Control Studies, Disease Progression, Female, Human, Middle Age, Neoplasm Staging, Prognosis, Prospective Studies, Sensitivity and Specificity, Treatment Outcome, Tumor Markers, Biological, blood,
Most of the prognostic indicators for breast cancer currently in use such as tumour size and histology, axillary lymph nodes, hormonal, receptors and growth factors depend on tumour tissue and thus they may not be available for repeated assessment. They are hardly of any use in predicting the presence of occult metastasis preoperatively and for further periodic follow-up. The results of earlier attempts to correlate, the level of tissue CEA with the prognosis of breast cancer were not very encouraging,. Therefore, serum carcino-embryonic antigen, a versatile and inexpensive tumour marker with proven efficacy in colorectal cancers, has been used as an indicator of prognosis in breast cancer. For achieving high specificity, solid phase Enzyme lmmunoassay (EIA) was used to estimate CEA levels, as has been used earlier.
The present prospective study included 50 breast cancer patients (49 females and 1 male), treated at the University Hospital from January 1992 to December 1994. Fifty age-sex matched patients with benign breast diseases (all females), and an equal number of matched individuals (49 females and 1 male), without, any evidence of malignancy or breast diseases served as controls. The mean age of the patients with breast cancer was 47.8 years and 28(56%) female patients were pre-menopausal. Patients, on an average, had 4 children (range = 0-7). Patients were clinically evaluated for preoperative diagnosis, to assess the extent of their disease and to rule out the presence of any confounding factor such as smoking, pancreatico-hepatico-biliary disease or collagen vascular diseases, which could cause non-specific elevations in the CEA levels. Fine needle aspiration cytology was performed in all cases to obtain a preoperative cytological diagnosis. Histological examination of the tissue excised by modified radical mastectomy or obtained by core biopsy was done to confirm the diagnosis and to determine the various histologic prognostic indicators. Modified radical mastectomy (MRM) was performed for all operable breast cancers and for locally advanced breast cancer after two cycles of neo-adjuvant (anterior) chemotherapy (CMF). Adjuvant chemotherapy was given to all patients with unfavourable tumour characteristics and to pre-menopausal node positive patients. In node positive postmenopausal women with favourable tumour characteristics, adjuvant hormone therapy in from of tamoxifen (20 mg/day) was given. Metastatic breast cancer was treated with chemotherapy (CMF) or tamoxifen along with palliative local radiation, when required. In cases of relapse of the disease, the patients were switched over to second line drugs or from one form of adjuvant therapy to the other.
Baseline serum CEA levels were measured by CEA Enzyme immuno-assay kit supplied by CDR Diagnostics, before starting any treatment. The technique is based on the principle of sandwich assay for antigen. The CEA level of the test samples were calculated from the standard curve using their mean absorbance. Pre-treatment serum CEA levels were correlated with the different clinico-pathological and prognostic parameters. Serum levels were estimated again 3 months after surgery or first cycle of chemotherapy. Serial CEA levels were measured at 6 monthly intervals during the follow-up of 21 to 40 months (mean = 29.4 months). The statistical significance of the results obtained were tested by Wilcoxon two sample test, Mann Whitney U test, Bartlett's test for homogeneity of variance, Kruskal Wallis one way analysis of variance, analysis of variance and receiver operating characteristic (ROC) curves, as applicable.
Pre-treatment serum CEA levels in the normal controls (2.72 + 0.81 ng/ml) and in the patients with benign breast diseases (2.94 + 1.114 ng/ml) were significantly lower (p = 0.000) than those in the breast cancer patients (21.36 + 14.35 ng/ml). Serum CEA levels were independent of the age, the menstrual status and the parity of the patients. On the basis of the preoperative evaluation, the distribution of the patients in different stages was stage 1-2(4%), stage II A-6(12%), stage 11113-7(14%), stage MA-8(16%), stage 110-13(26%) and stage IV-14(28%). Serum CEA levels were independent of the primary tumour status (p = 0.065). Six patients developed metastasis during 21-40 months of follow-up. The difference in serum CEA levels of the patients with localised (17.29 + 9.11 ng/ml) and metastatic breast cancer (31.91 + 19.67 ng/ml) was significant (p = 0.000), which was independent of the site of metastasis (p = 0.459).
On histological examination, 35 (70%) cases had duct adeno-carcinoma, not otherwise specified (NOS). Duct carcinoma with squamous differentiation, inflammatory carcinoma and papillary carcinoma were present in 3 (6%) cases each. Two (4%) cases each of medullary carcinoma, tubular carcinoma and colloid carcinoma were encountered. Well-differentiated tumour was found in 3 (6%) patients, moderately differentiated in 17(34%) and poorly differentiated in 30(600/6) cases. Different histological types of tumour did not produce significantly different serum CEA levels (p = 0.5511). However, there was a marked difference in the serum CEA levels produced by different grades of tumour (p = 0.029).
Auxiliary lymph nodes had tumour deposits in 70% of the cases. One to three nodes were involved in 28%, 4-6 in 34%, 7-9 in 14% and more than 10 nodes in 4% of the patients. Serum CEA levels differed significantly with the nodal status (p = 0.002) and the number of involved nodes (p = 0.014). However, the degree of lympho-reticular response to the tumour did not significantly influence the serum CEA levels (p = 0.183).
Following successful treatment, 3 months post-treatment serum CEA level fell significantly below pre-treatment level (p=0.000). ROC curve revealed a pre-treatment serum CEA value above 10 ng/ml to be diagnostic of breast cancer. The sensitivity, specificity and predictive value were 84%, 93% and 93.3% respectively. However, for stage 1 and 11 breast cancers, the sensitivity and the predictive value were only 57.1 % and 53.3% respectively, for 93% specificity. Similarly, ROC curve showed a cut-off pre-treatment serum CEA level of 25 ng/ml to be more accurate with the sensitivity, specificity and predictive value of 71.4%, 91.7% and 76.9% as compared to 78.6%, 72.2% and 52.4% respectively for 20 ng/ml [Table - 1].
A persistent post-treatment CEA level over 15 ng/ml was invariably associated with the presence of residual or metastatic disease. A change over 20%, in 3 months post-treatment CEA level from the baseline serum CEA level, was considered significant. Serum CEA levels changing by less than 20% were regarded as stable. The trend in the post-treatment serum CEA level correlated well with the clinical response. A stable or rising trend was seen in 516 (83.33%) patients, who developed metastasis during follow-up [Table - 2]. This test provided an average lead time of 3.86 + 1.28 months between the elevation of serum CEA and the clinical evidence of metastasis.
Carcino-embryonic antigen is the most widely used tumour marker today. Initially it was thought to be specific for colon cancer but subsequent studies have proved its versatility. CEA is produced by breast cancer tissue and is then secreted in blood and various body fluids. Against this background, attempts were made to define its role in breast cancer and other malignancies. At this stage it is important to record that individual laboratories have used different cut-off levels of serum CEA based on the technique used and the population studied as above 2 ng/ml6, 7, 5 ng/ ml8,9 and 10 ng/ml10,11. We selected 10 ng/ml as a cut-off level for diagnosing breast cancer in this study. Some workers have preferred to consider the change more than 20% and 50% with respect to initial baseline level rather than using a definite cut-off level, to be significant.
Mughal and others reported an elevated plasma CEA level in 50% of their patients which was commonly associated with the increased age, the post menopausal status and the prolonged disease free interval. In contrast, Borth wick et al and the present study showed that elevations in serum CEA level was the same in both pre-menopausal and postmenopausal women and in various age groups of patients. Lack of significant variation in pre-treatment serum CEA levels with the parity supports the observation of Helier et al, who disproved the myth of preferential destruction of CEA expressing breast cancers by the body.
There was a significant difference between the serum CEA levels in localised and metastatic disease, which was in accordance with the earlier studies,,. The primary tumour, its histology and the lympho-reticular response showed no correlation with the pre-treatment serum CEA levels, as also observed by others,. However, the histo-logical grade of the tumour and the nodal status altered the CEA level significantly, as reported earlier. Esteban et al found CEA to be an independent predictor of disease-free survival in breast cancer patients and also of overall survival, in estrogen receptor negative patients.
Omar et al reported 66% sensitivity and 93% specificity with CEA as a diagnostic test for malignancy of breast, whereas, AI Jarallah et al found a sensitivity of 70% and positive predictive value of 91%. For a versatile tumour marker like CEA, it is essential to have a threshold value at which the test will have a higher specificity. Using ROC curve, we selected values more than 10 ng/ml to be diagnostic of breast cancer. At this cut-off level, the sensitivity of the test for detecting breast cancer was 84% at 93% specificity and 92% predictive value, which was higher than that of other series,,. This difference could be attributed to a higher proportion of advanced cases (stage III and IV accounted for 70% of the cases) in our study group. Even the earlier studies have shown the sensitivity of the serum CEA level to correlate with the tumour metastasis,. A cut-off value of 25 ng/ml has sensitivity, specificity and predictive value of 64%, 90% and 69% respectively for diagnosis of metastasis breast cancer, as also reported in the past.
One of the most important uses of serum CEA level is to monitor the therapeutic response. Various workers have reported variable response in CEA levels following treatment and thus, there is a controversy regarding the criteria to be employed for regarding a change to be significant. Haagensen et al and Lokich et al respectively regarded a change over 50% and 20% from the baseline level to be significant. Taking a- change more than 20% to be appreciable, we observed a significant change in 47/50 cases (94%), which correlated with the course of disease. Serial serum CEA level has been used as the first sign of the tumour recurrence and has been found to be a valuable tool in therapy,. Six cases which were initially diagnosed to have localised breast cancer, had metastasis during subsequent follow up. The stable or rising trend in serum CEA levels were observed in 83.3% of these cases, thereby giving a lead time of about 3.9 months. Various studies have reported a lead time of 8 weeks to 9.9 months between elevation of serum CEA levels and appearance of metastasis,,,,.
Thus, rather that a single raised level, it is the trend in the post-treatment CEA level which is a stronger indicator of recurrence. CEA when used alone is not a very precise tumour marker for breast cancer, especially for early breast cancer. Therefore, by combining it with another tumour marker such as CA 15-3, the efficacy of CEA can be improved..
[Table - 1], [Table - 2]