High Thioredoxin Expression Is Associated with Resistance to Docetaxel in Primary Breast Cancer

Docetaxel, which belongs to taxanes, is one of the most powerful anticancer drugs for breast cancer, and an increasing number of breast cancer patients have been treated with docetaxel not only in the metastatic setting but also in the adjuvant setting, and more recently, in the neoadjuvant setting. The response rate of docetaxel, however, is ∼50% even in the first-line chemotherapy and it decreases to 20% to 30% in the second- or third-line chemotherapy (1–3). Therefore, it is of vital importance to select the patients who are very likely to respond to docetaxel to eliminate the unnecessary treatment.

Thioredoxin, which was originally discovered in Escherichia coli in 1964, is a low molecular weight protein (4). Then, human thioredoxin was cloned as an adult T-cell leukemia–derived factor in 1989 (5). Thioredoxin, identified as a dithiol hydrogen donor, plays a crucial role in the regulation of cellular reduction and oxidation system, so-called “redox,” acting as a cytoprotector against various kinds of oxidative stresses such as UV, X-ray irradiation, and viral infection (6–9). Thioredoxin is also considered to play an important role in cell viability, activation, and proliferation (9–11). Thioredoxin is widely distributed in normal tissues and is highly expressed in a variety of cancers such as lung, cervical, pancreatic, hepatoma, gastric, and breast cancers (4, 12–17). Many studies have suggested that high expression of thioredoxin in cancers is associated with biologically aggressive phenotype (i.e., increased proliferation and decreased apoptosis; refs. 11, 12, 16, 17). Moreover, recent in vitro studies have shown an association of high thioredoxin expression in cancer cells with resistance to cis-diamminedichloroplatinum (II) (cisplatin), mitomycin C, doxorubicin, and etoposide (18–20).

Recently, we have studied the genes involved in docetaxel resistance by comparing the gene expression profiles of docetaxel-responsive and docetaxel-resistant tumors and were able to show that several genes related to the redox system, including thioredoxin, are up-regulated in docetaxel-resistant tumors (21). We have also found that MCF-7 cells, which are originally sensitive to docetaxel, become resistant to docetaxel after the transfection of the thioredoxin cDNA (21). These results strongly indicate the involvement of thioredoxin overexpression in docetaxel resistance and raise the possibility that thioredoxin might be a clinically useful marker for the prediction of response to docetaxel. Therefore, in the present study, we have studied the relationship of thioredoxin expression with response to docetaxel in primary breast cancer patients in the neoadjuvant setting. Thioredoxin expression was investigated by immunohistochemistry because immunohistochemistry seems to be beneficial as a laboratory test over the reverse transcription-PCR assay for mRNA levels because (a) a contamination problem by stromal and inflammatory cells can be eliminated by observing the immunostaining of only cancer cells, (b) immunohistochemistry is applicable to the paraffin section, enabling the retrospective analysis, and (c) immunohistochemistry detects the protein expression which seems to more reliably represent the gene function than the mRNA expression.

Patients. Sixty-three primary breast cancer patients (24 premenopausals and 39 postmenopausals) with a tumor >3 cm in diameter or with the cytologically confirmed axillary lymph node involvement were recruited in this study during the period from December 1999 to August 2002. These patients underwent a vacuum-assisted core biopsy to confirm invasive breast cancer before neoadjuvant chemotherapy. After informed consent was obtained, all patients were treated with docetaxel (60 mg/m², q3w, four cycles unless progressive disease) and followed by surgery. Twelve patients who showed progressive disease were further treated with cyclophosphamide/epirubicin therapy (cyclophosphamide 600 mg/m² plus epirubicin 60 mg/m², q3w, four cycles) before surgery.

Histologic grade and mitotic index. Prechemotherapy tumor samples obtained by a vacuum-assisted core biopsy were subjected to histologic analysis. The histologic grade, which consists of three components (differentiation, pleomorphism, and mitotic index), is based on the Scarff-Bloom-Richardson grading system (22). Each component was scored on a scale from 1 to 3. The scores were summed and categorized as grade 1 (well differentiated), grade 2 (moderately differentiated), or grade 3 (poorly differentiated). The number of mitosis was counted in 10 consecutive fields at ×400 magnification in the highest proliferative invasive area. Mitotic index was categorized as low grade (mitotic count ≤ 14) and high grade (mitotic count ≥ 15).

Immunohistochemical assay. Expression of thioredoxin, p53, BRCA-1, and Bcl-2 in prechemotherapy tumor samples obtained by a vacuum-assisted core biopsy was studied by immunohistochemistry. In addition, thioredoxin expression on postchemotherapy tumor samples obtained at surgery was also studied by immunohistochemistry. All tissue samples, obtained by a vacuum-assisted core biopsy and surgery, were fixed in 10% buffered-formalin and embedded in paraffin. The 4-μm-thick sections were processed using a modification of the labeled streptavidin-biotin method with a commercial kit (DAKO Corp., Glostrup, Denmark). The slides were deparaffinized and microwaved in citrate buffer (pH 6.0). Endogenous peroxidase activity was blocked by incubation in 3% hydrogen peroxide. Nonspecific reactions were suppressed by incubating slides with a blocking reagent. Then, the slides were incubated with a primary antibody for 18 hours at 4°C. Characteristics of the antibodies and cutoff points used in the present study are summarized in Table 1. Positive tumor cells were quantified by evaluating at least 1,000 cells and were expressed as percentage. Hepatocellular carcinoma tissue was used as positive control for thioredoxin (15, 23). In immunohistochemistry for thioredoxin, parallel incubation with normal mouse serum instead of a mouse monoclonal antibody against thioredoxin was also done to rule out the nonspecific staining (negative control). Representative results of immunohistochemical staining of thioredoxin in prechemotherapy tumor samples are shown in Fig. 1 with positive and negative controls. All samples were evaluated by two trained pathologists in a blind procedure without knowing the patients' background and clinical outcome. Cutoff points for p53, BRCA-1, Bcl-2, and thioredoxin were determined according to the previous reports (refs. 17, 24–26; Table 1).

Hormone receptor assay. Estrogen receptor (ER) and progesterone receptor levels in prechemotherapy tumor samples were measured by an enzyme immunoassay (Abbott Laboratories, Abbott Park, IL). The cutoff values of ER and progesterone receptor were 13 and 10 fmol/mg.protein, respectively, according to the instructions of the manufacturer.

Assessment of pathologic response to chemotherapy. Assessment of pathologic response was done according to the criteria of the Japanese Breast Cancer Society as follows: grade 0, no response; grade 1a, mild response with less than one third loss of viable cancer cells; grade 1b, mild response with equal or more than one third to less than two third loss of viable cancer cells; grade 2, moderate response with equal or more than two third loss of viable cancer cells; grade 3, no viable cancer cells (pathologic complete response; ref. 27). Grades 2 and 3 were defined as pathologic responders and grades 0, 1a, and 1b were defined as pathologic nonresponders.

In 12 patients who showed progressive disease to docetaxel therapy and switched to cyclophosphamide/epirubicin therapy before surgery, postchemotherapy tumor samples are supposed to reflect the effects of not only docetaxel but also of cyclophosphamide/epirubicin. Thus, pathologic response to docetaxel was not determined on these postchemotherapy tumor samples. Instead, all these 12 patients were classified into pathologic nonresponders because a pathologic response (grade 2 or 3) is usually not seen in such patients who showed progressive disease.

Statistical methods. Clinicopathologic features and expression of various biomarkers were compared between responders and nonresponders by a χ² test. Thioredoxin expression levels between pre- and postchemotherapy tumor samples were compared by a paired t test. All tests were considered significant at P < 0.05.

Relationship between clinicopathologic variables or various biomarkers and pathologic response to docetaxel. Of 63 patients, 15 (grade 2 in 13 and grade 3 in 2) showed a pathologic response to docetaxel. Age, menopausal status, tumor size, histologic type, histologic grade, mitotic index, and progesterone receptor status showed no significant correlation with a pathologic response (Table 2). Patients with negative ER showed a significantly higher response rate (32.4%) than those with positive ER (10.7%; P = 0.043). There was no significant association between the expression of p53, BRCA-1, or Bcl-2 and a pathologic response but tumors with a low thioredoxin expression (30.6%) showed a significantly higher pathologic response rate than those with a high thioredoxin expression (0%; P = 0.018; Table 3). Combination of thioredoxin and ER was significantly (P = 0.006) associated with a pathologic response; i.e., response rate was 42.3% (11 of 26) in the thioredoxin-low and ER-negative tumors, 10.0% (3 of 30) in the thioredoxin-high and ER-negative tumors and in the thioredoxin-low and ER-positive tumors, and 0% (0 of 6) in the thioredoxin-high and ER-positive tumors.

Comparison of thioredoxin expression before and after docetaxel therapy. Of 63 patients, thioredoxin expression in both pre- and postchemotherapy tumor samples was able to be evaluated in 46 patients because 12 patients treated with docetaxel followed by cyclophosphamide/epirubicin before surgery were excluded from this analysis and immunohistochemical analysis was infeasible in five patients due to complete or nearly complete disappearance of cancer cells. Thioredoxin expression significantly (P < 0.0001) increased after docetaxel therapy (mean, 56.1%) as compared with that before docetaxel therapy (mean, 28.6%; Fig. 2). This increase in thioredoxin expression was seen at a similar degree both in responders (before, 24.4% versus after, 51.8%) and nonresponders (before, 29.9% versus after, 57.4%; Fig. 2).

Thioredoxin is a key molecule for redox regulation and it plays important roles not only as a radical scavenger but also as a redox-regulator of intracellular signal transduction (11, 28, 29). Thioredoxin knockout mice die in utero (30). In contrast, thioredoxin-transgenic mice are more resistant to infection, inflammation, and ischemic diseases and survive longer than control mice (31–33). Thus, thioredoxin is considered to be one of the essential molecules for normal cells to survive. In cancer cells, thioredoxin overexpression is suggested to be associated with a biologically aggressive phenotype. Grogan et al. (16) reported that gastric cancers with high thioredoxin levels showed a highly positive correlation (P < 0.001) with cell proliferation and a highly negative correlation (P < 0.001) with apoptosis. A few in vitro studies showed that the increased expression of thioredoxin was involved in the resistance to cisplatin (18–20). Yokomizo et al. (18) showed that several human bladder and prostatic cancer cell lines with high thioredoxin levels were associated with the resistance to cisplatin, mitomycin C, doxorubicin, and etoposide. Cisplatin produces superoxide anions when it reacts with its target DNA in a cell-free system (34). Mitomycin C, doxorubicin, and etoposide are also thought to induce oxidative stress (18). Thioredoxin is expected to work as a protector of cancer cells against oxidative stress induced by these anticancer drugs (11, 18–20).

In the present study, we have been able to show that high thioredoxin expression is associated with a poor pathologic response (P = 0.018) to docetaxel in breast cancer patients. It has yet to be reported whether or not docetaxel induces oxidative stress, but Pae et al. (35) have recently shown that paclitaxel, which is another taxane with a similar antitumor activity to docetaxel in breast cancer, generates superoxide anion via the stabilization of microtubules in a murine cell line. Thus, it is speculated that association of high thioredoxin expression with resistance to docetaxel might be explained, at least in part, by the protecting effect of thioredoxin against oxidative stress induced by docetaxel.

The antitumor activity of docetaxel results from stabilization of microtubules via the direct binding of docetaxel to β-tubulin (36). This stabilization of microtubules induces the inhibition of mitosis, which is considered to be the major mechanism of cytotoxic activity of docetaxel. The microtubule assembly and disassembly are thought to be dependent on the redox regulation of cysteine residues on tubulin. Khan and Luduena (37) have shown that thioredoxin can reduce a disulfide bridge in the tubulin dimer and inhibit microtubule assembly in vitro. At present, it is unknown whether or not such an influence of thioredoxin on microtubule assembly might affect the antitumor activity of docetaxel.

We have observed that thioredoxin expression significantly increased after docetaxel therapy. Because thioredoxin is supposed to work as a protector against docetaxel, it is estimated that tumors showing an increase in thioredoxin expression in response to docetaxel would be more resistant to docetaxel than those showing no increase. Thus, we compared the extent of increase in thioredoxin expression after docetaxel between responders and nonresponders but no significant difference was observed between them, suggesting that high thioredoxin expression before docetaxel therapy, but not thioredoxin induction after docetaxel therapy, serves as a significant predictor of resistance to docetaxel. However, because tumors which showed progressive disease to docetaxel and were treated with additional cyclophosphamide/epirubicin therapy before surgery, as well as some tumors which showed a pathologic complete or nearly complete response, were not included in this analysis (because immunohistochemistry was infeasible in such tumors), a conclusion cannot be drawn about the significance of thioredoxin induction after docetaxel therapy in the prediction of response. This issue seems to be able to be studied with more accuracy if we use the tumor samples obtained shortly (i.e., after a few days) after chemotherapy for immunohistochemistry for thioredoxin expression.

Docetaxel has been shown to promote the phosphorylation and inactivation of Bcl-2 and to induce apoptosis (38, 39). Thus, Bcl-2 expression is speculated to have a correlation with response to docetaxel. However, Sjostrom et al. (40) have shown that Bcl-2 has no value in the prediction of a response to docetaxel in 64 breast cancer patients. Neither were we able to find any significant correlation between Bcl-2 expression and a pathologic response to docetaxel. Interestingly, a recent study has indicated that docetaxel induces apoptosis and cell death through a Bcl-2-independent mechanism (41). We have also failed to show a significant association of p53 and BRCA-1 expression with a pathologic response. Actually, several conflicting results have been reported on the value of these biomarkers in the prediction of a response to taxanes (24, 26, 40, 42). On the other hand, we have been able to show that negative ER status showed a significant association with a good pathologic response, being consistent with the previous studies (43, 44). Interestingly, combination of thioredoxin and ER seems to be more useful in the prediction of a pathologic response than thioredoxin or ER alone. Because several other mechanisms involved in resistance to docetaxel have been postulated, it would be important to evaluate the clinical significance of thioredoxin in the prediction of a response to docetaxel in concert with the other markers.

Recently, Chang et al. (45) have shown that the expression profile of the 92 genes selected using the Affymetrix microarrays is useful in the prediction of a response to docetaxel in primary breast cancer. We have also shown that the 85 genes selected using adapter-tagged competitive-PCR can predict a response to docetaxel in primary breast cancer with high accuracy (21). Although the several genes related to the redox system, including thioredoxin, were included in the 85 genes selected using adapter-tagged competitive PCR, they were not included in the 92 genes selected using the Affymetrix microarrays. The Affymetrix GeneChip has a wider coverage of genes but recent analysis revealed that the quantitative data were obtained only with abundant genes (46). Genes with low expression were therefore excluded from the analysis. Adapter-tagged competitive PCR can be used to measure expression of rare genes but its gene coverage is not as large as that of the GeneChip. Consequently, the gene populations that the two studies examined are not likely to be the same. Therefore, it is not surprising that diagnostic genes do not overlap in the two studies.

In conclusion, we have shown that high thioredoxin expression in prechemotherapy tumor samples is associated with a poor pathologic response to docetaxel, indicating that thioredoxin might work as a protector against oxidative stress induced by docetaxel. We have also found that ER negativity is associated with a good pathologic response to docetaxel and suggested that combination of thioredoxin and ER might be clinically more useful in the prediction of a response to docetaxel. Additional studies including a larger number of patients need to be done to confirm the clinical significance of thioredoxin and ER.

We thank Dr. Hiroshi Masutani (Department of Biological Responses, Laboratory of Infection and Prevention, Institute for Virus Research, Kyoto University) for helpful discussion.