Prognostic value of monitoring tumour markers CA 15-3 and CEA during fulvestrant treatment

Tumour markers are often used to detect disease recurrence after primary treatment for cancer [1, 2] and to monitor response to therapy in patients with advanced disease [3]. In breast cancer, the most widely adopted combination of tumour markers are carcinoembryonic antigen (CEA) and mucin (MUC1), commonly detected as cancer antigen (CA) 15-3 [4]. Of these two markers, CA 15-3 is generally regarded as the most specific and sensitive. From as early as 1978 investigators were examining the prognostic value of CEA in patients receiving endocrine therapy for the treatment of metastatic breast cancer (MBC) [5], but the value of tumour markers in monitoring response to fulvestrant ('Faslodex') treatment has not previously been evaluated.

It is commonly believed that an increase in tumour marker levels may be a sign of de novo disease progression (PD), thus signalling that a change in therapy may be necessary [6]. A recent prospective study found that the changes in serum tumour marker levels after the start of therapy correlate significantly with therapy response; and a greater than 20% reduction in marker levels was a favourable predictive factor for time to progression during systemic treatment [7]. However, in 1996 Sonoo suggested that serial levels of tumour markers taken during therapy might not always correlate with therapy response [8]. This group observed an initial increase and subsequent decrease in tumour marker levels in up to one-third of patients experiencing clinical benefit (CB). One reason for this spike phenomenon might be the possible tumour flair associated with the partial agonistic properties of selective estrogen receptor modulators (SERMs), which however was not observed with fulvestrant yet. A possible alternative explanation for this spike is that it results from the destruction of tumour cells on successful treatment, similar to the increase in muscle enzymes observed following myocardial infarction. Another group recently observed that changes in serial tumour marker levels do not allow prediction of bone scan results [9]. Thus the real value of serial observation of marker levels remains somewhat disputed.

Fulvestrant is the first of a new type of oestrogen receptor (ER) antagonist to be licensed for the treatment of advanced breast cancer [10, 11]. Fulvestrant binds, blocks and degrades the ER, thus decreasing cellular levels of the ER and minimising the transcription of oestrogen- and progesterone-dependent genes [12, 13]. Two large PhaseIII trials have shown that fulvestrant is at least as effective as anastrozole in the treatment of patients with advanced breast cancer progressing on/after antioestrogen therapy [14‐17]. It has also been shown to have similar efficacy as tamoxifen in the first-line treatment of patients with ER-positive and/or progesterone receptor (PgR)-positive disease [18]. Here we evaluated the prognostic value of tumour markers at predicting response and secondary progression during fulvestrant therapy.

All data were collected at the Department of Internal Medicine I, Division of Oncology at the Medical University of Vienna, Vienna, Austria. The study was performed in accordance with the ethical regulations of the Medical University of Vienna.

Criteria for inclusion were as follows: Presence of at least one measurable lesion, Karnofsky performance score ≥ 70%, life expectancy of >3 months, adequate organ function as defined by WBC count ≥ 3500/μl, platelet count ≥ 100 000/μl, hematocrit ≥ 30% and serum bilirubin and creatinine ≤ 1.25 × upper limit of the institution's normal range, and informed consent. We included patients with marker levels below the cut-offs at the initiation of treatment or dropping below the cut-offs during the course of fulvestrant treatment. However, patients were excluded from this study if they never experienced an elevation of marker levels above cut-offs on treatment.

Fulvestrant was administered monthly as a single 250 mg intramuscular injection [19] as part of a compassionate use programme (supported by AstraZeneca) until PD or other event necessitating withdrawal. All patients were postmenopausal women with ER-positive and/or PgR-positive MBC for whom at least one prior endocrine therapy had failed, either as adjuvant treatment or as the treatment of advanced disease. Many had also received chemotherapy in the adjuvant and/or advanced disease settings. Table 1 lists the characteristics of all patients included.

Our group was amongst the first in continental Europe to treat patients with fulvestrant as part of a compassionate use programme supported by AstraZeneca. In our institution it is common practice to evaluate tumour marker levels following each month of palliative therapy. During informal analysis of such data in patients receiving fulvestrant, potential differences in tumour marker kinetics were noted within the subgroup of patients responding to therapy and thus a detailed investigation was carried out.

In this study a trend towards increased CA 15-3 levels was noted during the first 3 months in patients responding to fulvestrant treatment. This trend became stronger over time, but still failed to reach statistical significance after 6 months of treatment. This result was unexpected since these patients were responding to fulvestrant therapy. The only occasion in which CEA appeared to have a higher predictive value than CA 15-3 was in the subgroup of patients experiencing a PR, where the observed decrease in CEA levels attained statistical significance after 6 months.

In the group of patients experiencing SD we found a significant increase in CA 15-3 levels at both 3 and 6 months. This was observed both in patients with SD ≥ 6 months but < 9 months and in those experiencing SD ≥ 9 months, thus this observation is unlikely to result from a bias caused by a lead-time effect to actual PD. There are several possible explanations for this early increase in tumour marker levels. First, it may be as a result of the delay in reaching steady-state levels with fulvestrant treatment [20], although this appears unlikely as these patients were already experiencing CB with fulvestrant. Nonetheless it seems reasonable to ask whether a loading-dose regimen of fulvestrant may be appropriate and this type of dosing is under investigation in ongoing clinical trials [21]. However, as initial increases in CEA and CA 15-3 have also been observed in patients experiencing CB with other breast cancer therapies [8, 22‐25], the dose of fulvestrant seems unlikely to be the cause. One explanation is that the increase in tumour markers may result from increased tumour degradation in response to fulvestrant treatment. These data may suggest that, in the absence of radiological detection of PD, fulvestrant treatment should be continued beyond 3 months before response to therapy is assessed. Marker levels in patients with de novo PD increased steadily during fulvestrant treatment as may be expected in the presence of non-responsive disease. As tumour marker levels may rise in both responding and non-responding patients during the first few months of fulvestrant treatment, it may be that an increase in markers following a period of stabilisation may be more predictive of disease progression.

The data presented here suggest that if increased CA 15-3 levels are observed after the first 3 months of fulvestrant treatment this should not be taken as a sign of PD without radiological verification. On the contrary, such an increase may also be observed in patients gaining CB from treatment. Consequently, our results demonstrate that it is inappropriate to change therapy purely based on increased tumour marker levels, as some patients may still be benefiting from fulvestrant treatment.

The predictive value of tumour markers in signalling secondary progression was also prospectively assessed. We found that both CA 15-3 and CEA levels increased significantly during the last 3 months of treatment and so these markers may be valuable in predicting secondary progression. However, both the CEA and CA 15-3 data need to be treated with some caution because of the small number of patients in this study.