Tumor markers – a science in development

Tumor Markers in Breast Cancer


From Annals of Oncology

Tumor Markers in Metastatic Breast Cancer Subtypes

Frequency of Elevation and Correlation With Outcome

R. Yerushalmi; S. Tyldesley; H. Kennecke; C. Speers; R. Woods; B. Knight; K. A. Gelmon

Authors and Disclosures

Posted: 02/08/2012; Annals of Oncology. 2012;23(2):338-345. © 2012 Oxford University Press


             Print This

Email this




         Abstract and Introduction





Abstract and Introduction


Background: Little is known about the correlations between tumor markers (TMs), breast cancer subtypes, site(s) of metastasis and prognosis.ӬMethods: Women diagnosed with metastatic breast cancer were included. Breast cancer subtypes were defined as LuminalA, LuminalB, LuminalHer2, Her2, Basal and non-Basal triple negative (TN). Levels of elevation of TM values [cancer antigen 15-3 (CA 15-3), carcinoembryonic antigen (CEA) and cancer antigen 125 (CA 125)] among the subtypes were analyzed. Site(s) of metastasis and outcomes were captured.ӬResults: Eight hundred and ten patients were included. Luminal subtypes were associated with an elevation in at least one TM: 90.8% of LuminalHer2+, 90% of LuminalB and 88.6% of LuminalA. TMs were less frequently elevated in Basal (74.1%) and non-Basal TN (71.4%) cases (P < 0.001). CA 15-3 was the most frequently elevated TM. The incidence of TM elevation did not differ between patients with solitary versus multiple metastatic sites. Breast cancer-specific survival (BCSS) was significantly worse for patients with elevated TMs (P = 0.001).ӬConclusions: TM elevation of CA 15-3, CEA and/or CA 125 was documented in the majority of patients with metastatic breast cancer with CA 15-3 occurring most commonly. Luminal subtypes expressed elevated TMs significantly more frequently compared with the non-Luminal groups. TM elevation was not different between the different sites of metastasis. Overall, elevated TMs predicted a worse BCSS.


Tumor markers (TMs) are widely used in the management of patients with breast cancer, during therapy for metastatic disease and in conjunction with diagnostic imaging, history and physical examination.[1] Cancer antigen 15-3 (CA 15-3) and carcinoembryonic antigen (CEA) are elevated in ~80% and 40%, respectively, of metastatic breast cancer in a number of older reports.[2-5] Guadagni et al. have shown that adding measurement of CEA to CA 15-3 testing resulted in only a minimal improvement in sensitivity. Hence, the American Society of Clinical Oncology guidelines do not recommend a routine CEA measurement in the metastatic setting when CA 15-3 is elevated.[1, 2] Cancer antigen 125 (CA 125) is a TM which is commonly used in ovarian cancer but may also be elevated in the metastatic phase of other malignancies[6] and elevated levels have been observed in up to 84% of metastatic breast patients.[7, 8]

Although early studies have shown an association between elevated CA 15-3 levels and estrogen receptor (ER) positivity [9], most reports have not evaluated differences between breast cancer subtypes. Previous studies have reported associations between TM elevation and site(s) of metastasis, disease burden and prognosis.[4, 10] Despite continuous interest in this topic, which has relevance to daily practice, very little is known about the correlation between TMs and intrinsic breast cancer subtypes. In this large study, we have investigated the relevance of three different TMs (CA 15-3, CEA and CA 125) among six different breast subtypes: LuminalA, LuminalB, LuminalHer2+, Her2+ ER−, Basal type and non-Basal triple negative (TN) type.

The objectives of the study were:

  • To document the proportion of patients with elevated TMs and the degree of TM elevation among the different breast cancer subtypes
  • To determine whether TM elevation correlates with the site(s) of metastasis and

To determine whether elevation of TMs correlates with prognosis in general and within breast cancer subtypes in particular.



Women with breast cancer diagnosed between 1986 and 1992 and referred to the British Columbia Cancer Agency with M1 disease at presentation or who later developed a distant relapse were included. Cases with a previous, synchronous or subsequent invasive or in situ cancer of any site other than nonmelanoma skin were excluded. Archival paraffin tissue blocks were used to construct a tissue microarray. Breast cancer subtypes were defined as LuminalA (ER/PR+, Her2− and Ki67 < 14%), LuminalB (ER/PR+, Her2− and Ki67 ≥ 14%), LuminalHer2 (Her2+ and ER/PR+), Her2 (Her2+, ER− and PR−), and Basal [Her2−, ER−, PR− and (CK 5/6+ and/or epidermal growth factor receptor+)] using immunohistochemical staining as has been previously described.[11, 12] In addition, we examined the non-Basal TN (ER−, PR− and Her2−) subgroup.

The Breast Cancer Outcomes Unit Database was used to determine clinical, pathologic, treatment and outcome variables. Health Information Management professionals, trained in breast cancer data abstraction, reviewed the complete medical record of all eligible patients. The date of diagnosis and the anatomic location of up to the first five sites of metastasis were captured. Distant relapse was defined as recurrence of breast cancer occurring beyond the confines of the ipsilateral breast, chest wall or regional lymph nodes. Sites of distant relapse were categorized as follows: brain (including choroid, central nervous system, not otherwise specified, pituitary gland, cerebral meninges, meninges, leptomeningeal, frontal sinus), liver, lung (including lymphangitic carcinomatosis), bone (including bone marrow), distant nodal (nodes beyond the ipsilateral and axillary/supraclavicular/internal mammary area), pleural/peritoneal (including ascites, omentum, pleural effusion and peritoneal carcinomatosis), other (including skin outside of breast/chest wall, ovaries, spinal cord, eye, heart and other organs not elsewhere classified) and unknown (patient known to have distant metastases but site(s) are unknown).

TM results were captured from the TM database in the Provincial Health Service Authority/British Columbia Cancer Agency laboratory. The history of the TM assays used over the study years is presented in Table 1. Levels of TM values (CA 15-3, CEA and CA 125) within the 3 months before distant relapse date or anytime after were captured and the percentage of patients with elevated values (CA 15-3 >28, CEA >4 and CA 125 >35) among the different subtypes were reported. Cases were included if they had a minimum of one TM measurement carried out during this time.

The percentage of patients with elevated TMs was compared across subtypes using the Chi-square test or Fisher’s exact test as appropriate; similar statistical tests were done to compare the percentage of patients with TM elevation by the first five sites of metastases. In order to exclude a bias related to transient treatment-induced TM elevation [13], we also assessed the upper limit of normal ratio [i.e. the ratio of the first elevated TM (CA 15-3) to the upper limit of normal] carried out during the initial 3 months before distant relapse until 3 months after distant relapse. Ratios were compared between breast cancer subtypes using the Kruskal-Wallis test. It is possible that some subtypes will have elevated markers at a time when the tumor burden is highest, i.e. in the last 6 months of life compared with levels at the time of initial diagnosis of metastases. To assess the possible impac of this effect, we also analyzed for TM measurements in the last 6 months of each patient’s life or last 6 months of follow-up for those who were still alive.

The magnitude of TM elevation in the different cancer subtypes was assessed with the median and range of the number of elevated measurements per patient, as well as the total number of measurements carried out per patient, from 3 months before the date of first distant metastasis to anytime after. The median number of elevated TMs was compared across subtypes, and between groups based on site of first metastasis, using the Kruskal-Wallis test.

Breast cancer-specific survival (BCSS) was compared in patients with elevated TMs versus those with nonelevated TMs using the Kaplan-Meier method. These analyses were repeated within each of the breast cancer subtypes. BCSS was calculated from time of diagnosis of distant metastasis to death or censor date. We further investigated the association between magnitude of marker elevation and BCSS. We used a Cox regression model that incorporated the magnitude of marker elevation by quartile of the multiples of upper limits of the normal reference marker values (=CA 15-3 ULN ratio) as a time-dependent covariate, thus including all of the patient’s marker measurements in the model.

The study was approved by the University of British Columbia Research Ethics Board.


For inclusion in the study, 1656 cases with distant metastases were potentially eligible. Excluded cases: 428 cases without any linkage to TM data, 127 cases with all TMs measured >3 months before distant relapse, 184 cases where breast cancer subtype could not be determined and 107 cases with synchronous or subsequent cancer (25 of which were subsequent contralateral breast cancer). A total of 810 patients met all inclusion criteria (Figure 1).



(Enlarge Image)

Figure 1.

Patient inclusion diagram.


Our reporting on the study population was based on the ‘reporting recommendation for tumor marker prognostic studies’.[14, 15] The distribution among subtypes was: LuminalA n = 315, LuminalB n = 251, LuminalHer2+ n = 76, Her2+ ER− n = 62, Basal n = 85 and non-Basal TN n = 21. Eight percent (range among breast cancer subtypes: 3.2%-10.4%) of the cohort presented with metastatic disease at time of initial diagnosis. Elevated TMs were observed in all breast cancer subtypes. The Luminal subtypes were most often associated with elevated TMs: 90.8% of LuminalHer2+, 90% of LuminalB and 88.6% of LuminalA cases, whereas more than one quarter of the Basal and non-Basal TN cases expressed no elevation of TMs.

CA 15-3, the most commonly used TM in breast cancer, was elevated in >80% of the Luminal cases, compared with 69.3% and 68.4% in the Basal and the non-Basal TN groups, respectively (P < 0.001). The magnitude of CA 15-3 elevation was larger in the Luminal subtypes (P = 0.02). CEA elevation was observed in 60%-65% of the Luminals as compared with only 26% in the Basal type and 31% in the non-Basal TN cases (P < 0.001). In all subtypes, except the LuminalHer2+ group, CA 125 values were elevated in ~50% of all patients, compared with 35% in the LuminalHer2+ group, P = 0.94. The percentage of cases with elevated TMs among the different breast cancer subtypes, including relevant P values are shown in Table 2.

We further investigated CA 15-3 using the first elevated value done during the time period of 3 months before distant relapse to 3 months after distant relapse. Boxplots, shown in Figure 2, illustrate the upper limit of normal ratio (the value of the first elevated CA 15-3 done during this time period divided by the reference of upper level of normal for CA 15-3). Not only did the Luminal subtypes have more frequent TM elevation, as demonstrated in our initial analyses, but they were also shown to have larger magnitudes of elevation (P < 0.001).



(Enlarge Image)

Figure 2.

Boxplot of first elevated cancer antigen 15-3.


When TM elevations during the last 6 months before death were investigated, the trends were similar. Of those 511 cases with measures in the last 6 months, the following percentages of elevations were observed in the various subgroups:

LuminalA: 95.6% (172/180), LuminalB: 92.5% (147/159), LuminalHer2: 90.9% (50/55), ER−, HER2+: 85.4% (35/41), Basal: 73.8% (45/61) and non-Basal TN: 86.7% (13/15), P < 0.0001.

TM expression was evaluated according to site of metastases in the following categories: bone only, brain only, solitary nonbrain/nonbone site and multiple metastatic sites. For CA 15-3 and CEA, there was no significant association between TM elevation and anatomic site of first distant metastasis (Table 3). TM elevation was seen regardless of metastatic site. In patients with bone-only metastases, the proportion of patients with an elevated CA 125 was only 27%, compared with 85% with elevated CA 15-3 levels and 63% with elevated CEA levels. When brain was the solitary metastasis (n = 10), 80% of the cases had at least one type of elevated TM. The proportion of the different TMs among the different site(s) of metastases is illustrated in Table 3. Although analysis within some subgroups is limited by small numbers of cases (particularly for the brain as first site of metastasis group), it is clear that TMs, particularly CA 15-3, can be elevated regardless of the site(s) of metastasis.

Among the 114 patients with metastatic disease and normal CA 15-3, 40 (35%) had an increase of either elevated CEA and/or CA 125. The proportion of patients with an elevated CEA and/or CA 125 in the setting of a normal CA 15-3 was higher in the Luminals (37%-45%), as compared with the Basal subtype (10%) Table 4.

Median duration of survival from time of diagnosis with metastatic disease was significantly shorter for patients with elevated TMs versus those with normal TM values, P = 0.001 (Figure 3). Similar results were found when stratifying the results by subtype; however, only the results for LuminalA and Basal types were statistically significant, P = 0.02 and 0.001, respectively (Figure 4A-E). When a secondary analysis was carried out using a Cox model that incorporated marker elevation as a time-dependent covariate (as opposed to a binary analysis with any marker elevation being considered positive), the strength of the effect was even stronger, with a hazard ratio of 2.9 [95% confidence interval (CI) 2.4-3.6] for all markers combined. Increasing magnitude of TM elevation, relative to lowest quartile of elevation was associated with progressively inferior BCSS with a hazard ratio of 1.6 (95% CI 1.3-2.1) in the second quartile and 5.2 (95% CI 4.3-6.5) in the highest quartile.



(Enlarge Image)

Figure 3.

Patient outcome by tumor marker.




(Enlarge Image)

Figure 4.

(A) Breast cancer-specific survival (BCSS) according to elevation in cancer antigen 15-3 (CA 15-3), carcinoembryonic antigen (CEA) and/or cancer antigen 125 (CA 125) among 315 patients with metastatic LuminalA breast cancer. (B) BCSS according to elevation in CA 15-3, CEA and/or CA 125 among 251 patients with metastatic LuminalB breast cancer. (C) BCSS according to elevation in CA 15-3, CEA and/or CA 125 among 76 patients with metastatic LuminalHer2+ breast cancer. (D)BCSS according to elevation in CA 15-3, CEA and/or CA 125 among 62 patients with metastatic Her2+/estrogen receptor negative breast cancer. (E) BCSS according to elevation in CA 15-3, CEA and/or CA 125 among 85 patients with metastatic Basal breast cancer.




This large study is the first to show the different TM behavior in six distinct breast cancer subtypes. Elevated CA 15-3, CEA and CA 125 values were documented in all breast cancer subtypes. We found a significantly higher percentage of elevated TMs in Luminal versus non-Luminal subtypes. The lowest frequency of elevated TMs was documented in the Basal and non-Basal TN cases. As well, the lowest magnitudes of TM elevation were seen with the Basal and non-Basal TN and the Her2 positive/ER negative subtypes.

The reason for this difference is not clearly understood. The current hypothesis suggests that breast stem cells, or their very early descendents, are the cells from which Basal tumors arise.[16] Thus, it would be reasonable to anticipate a lack of certain circulating antigen production in less differentiated subtypes. The fact that the described differences preceded the metastatic relapse event or occurred at the same time reduces the potential for bias caused by the different durations of survival after recurrence, which are known to be associated with the intrinsic subtypes. The additional analysis carried out for measurements in the last 6 months of the patients’ life support the study findings.

Our analyses divided the TN group into core Basal and non-Basal subtypes. In the future, this group of non-Basal TN cancers may be further segregated into other subtypes. However, in regard to TM production, the Basal and the non-Basal TN groups show comparable behavior.

Previous studies have shown that CA 15-3 is the most sensitive TM in breast cancer disease.[2, 17] The current study confirms this finding for all six subtypes, regardless of site(s) of metastasis. Recent literature suggests there is a modest advantage of performing TMs other than CA 15-3 in metastatic breast cancer. Our study found that 35% of the patients with normal CA 15-3 values had elevation of another TM (CEA and/or CA 125). The proportion of patients having another elevated marker, when CA 15-3 was normal, was higher in the Luminal subtypes (37%-45%), as compared with the Basal subtype (10%). However, CA 15-3 was elevated more often than the other TMs, regardless of the initial site(s) of metastasis. As we did not measure tumor burden, we cannot provide information on the correlation of TM elevation and amount of disease.

Compared with CA 15-3, CA 125 levels varied among the different site(s) of metastasis. For example, when bone metastasis was the solitary site of disease, CA 125 was more often (73%) found to be normal. When multiple metastases were found at first relapse, 65% of patients showed elevated CA 125. This may be explained by the fact that CA 125 is mainly produced by mesothelial cells and therefore is more likely to be elevated in the setting of abdominal and pleural metastases.[18, 19]

Although controversial, there are accumulating data that support our finding that elevated TMs are associated with a worse prognosis.[20] However, since TMs are more often elevated with Luminal subtypes, which generally have a better prognosis relative to other subtypes, the prognostic significance of TM elevation should be considered within a particular subtype rather than for breast cancer patients as a whole.

Fehm et al.[21] found that the median survival of patients with increased CA 15-3 was 13 months compared with 18 months for patients with normal CA 15-3 at diagnosis of metastatic disease. Others have found that CA 15-3 predicts survival in uni- and multivariable analyses.[22] Interestingly, we have found the TM to be a statistically significant prognostic marker in the LuminalA and B subtypes only. This could simply be due to their higher percentage in the cohort and higher frequency of elevated measurements.

Although this is a large study with long follow-up, the main caveat of this study is related to its retrospective nature. Therefore, biases related to physician decisions about when, for whom and in what circumstances they request TMs may confound the results. However, given that the study era predates the molecular subtyping of breast cancer, physician bias about the utility of measuring TMs in various breast cancer subtypes was not an issue. Therefore, the observed variation in TM elevation in the different breast cancer subtypes is likely real.

TM assessment is considered a relatively inexpensive test, which is easy to perform. However, repeat measurements to monitor metastatic disease and the response to therapy contribute to higher costs and increased laboratory workload. Clinicians should be aware of the potential gains from performing TM testing in the variable breast cancer subtypes.

Our data may be useful in studies that use TMs as therapeutic targets for novel interventions including vaccine development[23-25] or trials that use TM as a surrogate for clinical benefit in patients with nonmeasurable disease (S0430).

Knowledge of the frequency of elevation of TM may help define the breast cancer populations, which could be of interest for study.

In summary, we have shown that TMs do vary according to breast cancer subtype, with the most frequent elevations occurring in the Luminal groups. However, TMs can be elevated in all breast cancer subtypes. The addition of CEA and/or CA 125 to the more commonly measured CA 15-3 may further identify additional patients with elevated TMs. In the current study, BCSS is worse in metastatic breast cancer patients with elevated TMs.



Harris L, Fritsche H, Mennel R, et al. American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer. J Clin Oncol 2007;25:5287-5312.

Guadagni F, Ferroni P, Carlini S, et al. A re-evaluation of carcinoembryonic antigen (CEA) as a serum marker for breast cancer: a prospective longitudinal study. Clin Cancer Res 2001;7:2357-2362.

Tondini C, Hayes DF, Gelman R, et al. Comparison of CA15-3 and carcinoembryonic antigen in monitoring the clinical course of patients with metastatic breast cancer. Cancer Res 1988;48:4107-4112.

Lauro S, Trasatti L, Bordin F, et al. Comparison of CEA, MCA, CA 15-3 and CA 27-29 in follow-up and monitoring therapeutic response in breast cancer patients. Anticancer Res 1999;19:3511-3515.

Molina R, Jo J, Filella X, et al. c-erbB-2 oncoprotein, CEA, and CA 15.3 in patients with breast cancer: prognostic value. Breast Cancer Res Treat 1998;51:109-119.

Meyer T, Rustin GJ. Role of tumour markers in monitoring epithelial ovarian cancer. Br J Cancer 2000;82:1535-1538.

Berruti A, Tampellini M, Torta M, et al. Prognostic value in predicting overall survival of two mucinous markers: CA 15-3 and CA 125 in breast cancer patients at first relapse of disease. Eur J Cancer 1994;30A:2082-2084.

Baskic D, Ristic P, Matic S, et al. Clinical evaluation of the simultaneous determination of CA 15-3, CA 125 and HER2 in breast cancer. Biomarkers 2007;12:657-667.

Tampellini M, Berruti A, Gorzegno G, et al. Independent factors predict supranormal CA 15-3 serum levels in advanced breast cancer patients at first disease relapse. Tumour Biol 2001;22:367-373.

Begic A, Kucukalic-Selimovic E, Obralic N, et al. Correlation between bone scintigraphy and tumor markers in patients with breast carcinoma. Bosn J Basic Med Sci 2006;6:75-77.

Cheang MC, Chia SK, Voduc D, et al. Ki67 index, HER2 status, and prognosis of patients with luminal B breast cancer. J Natl Cancer Inst 2009;101:736-750.

Cheang MC, Voduc D, Bajdik C, et al. Basal-like breast cancer defined by five biomarkers has superior prognostic value than triple-negative phenotype. Clin Cancer Res 2008;14:1368-1376.

Kim HS, Park YH, Park MJ, et al. Clinical significance of a serum CA15-3 surge and the usefulness of CA15-3 kinetics in monitoring chemotherapy response in patients with metastatic breast cancer. Breast Cancer Res Treat 2009;118:89-97.

McShane LM, Altman DG, Sauerbrei W, et al. Reporting recommendations for tumor marker prognostic studies (REMARK). Exp Oncol 2006;28:99-105.

Hayes DF, Ethier S, Lippman ME. New guidelines for reporting of tumor marker studies in breast cancer research and treatment. Breast Cancer Res Treat 2006;100:237-238.

Lindeman GJ, Visvader JE. Shedding light on mammary stem cells and tumorigenesis. Cell Cycle 2006;5:671-672.

Loprinzi CL, Tormey DC, Rasmussen P, et al. Prospective evaluation of carcinoembryonic antigen levels and alternating chemotherapeutic regimens in metastatic breast cancer. J Clin Oncol 1986;4:46-56.

Epiney M, Bertossa C, Weil A, et al. CA125 production by the peritoneum: in-vitro and in-vivo studies. Hum Reprod 2000;15:1261-1265.

Daoud E, Bodor G. CA-125 concentrations in malignant and nonmalignant disease. Clin Chem 1991;37:1968-1974.

Sturgeon CM, Lai LC, Duffy MJ. Serum tumour markers: how to order and interpret them. BMJ 2009;339:b3527.

Fehm T, Jager W, Kramer S, et al. Prognostic significance of serum HER2 and CA 15-3 at the time of diagnosis of metastatic breast cancer. Anticancer Res 2004;24:1987-1992.

Ali SM, Leitzel K, Chinchilli VM, et al. Relationship of serum HER-2/neu and serum CA 15-3 in patients with metastatic breast cancer. Clin Chem 2002;48:1314-1320.

Singh R, Samant U, Hyland S, et al. Target-specific cytotoxic activity of recombinant immunotoxin scFv(MUC1)-ETA on breast carcinoma cells and primary breast tumors. Mol Cancer Ther 2007;6:562-569.

Crow DM, Williams L, Colcher D, et al. Combined radioimmunotherapy and chemotherapy of breast tumors with Y-90-labeled anti-Her2 and anti-CEA antibodies with taxol. Bioconjug Chem 2005;16:1117-1125.

Dorvillius M, Garambois V, Pourquier D, et al. Targeting of human breast cancer by a bispecific antibody directed against two tumour-associated antigens: ErbB-2 and carcinoembryonic antigen. Tumour Biol 2002;23:337-347.


Leave a Comment

Your email address will not be published. Required fields are marked *