Evaluation of prognostic and predictive value of microtubule associated protein tau in two independent cohorts

Introduction Microtubule associated proteins (MAPs) endogenously regulate microtubule stabilization and have been reported as prognostic and predictive markers for taxane response. The microtubule stabilizer, MAP-tau, has shown conflicting results. We quantitatively assessed MAP-tau expression in two independent breast cancer cohorts to determine prognostic and predictive value of this biomarker. Methods MAP-tau expression was evaluated in the retrospective Yale University breast cancer cohort (n = 651) using tissue microarrays and also in the TAX 307 cohort, a clinical trial randomized for TAC versus FAC chemotherapy (n = 140), using conventional whole tissue sections. Expression was measured using the AQUA method for quantitative immunofluorescence. Scores were correlated with clinicopathologic variables, survival, and response to therapy. Results Assessment of the Yale cohort using Cox univariate analysis indicated an improved overall survival (OS) in tumors with a positive correlation between high MAP-tau expression and overall survival (OS) (HR = 0.691, 95% CI = 0.489-0.974; P = 0.004). Kaplan Meier analysis showed 10-year survival for 65% of patients with high MAP-tau expression compared to 52% with low expression (P = .006). In TAX 307, high expression was associated with significantly longer median time to tumor progression (TTP) regardless of treatment arm (33.0 versus 23.4 months, P = 0.010) with mean TTP of 31.2 months. Response rates did not differ by MAP-tau expression (P = 0.518) or by treatment arm (P = 0.584). Conclusions Quantitative measurement of MAP-tau expression has prognostic value in both cohorts, with high expression associated with longer TTP and OS. Differences by treatment arm or response rate in low versus high MAP-tau groups were not observed, indicating that MAP-tau is not associated with response to taxanes and is not a useful predictive marker for taxane-based chemotherapy.


Introduction
Taxanes are microtubule stabilizing agents and potent cytotoxic compounds that have been recognized as highly effective chemotherapeutic agents [1,2]. However, varying degrees of benefit, with response rates ranging from 32% to 68% in the adjuvant and metastatic settings, suggest the critical need for a companion diagnostic to predict which patients are most likely to benefit from taxane therapy and which can be spared the cytotoxic effects of such therapy [3].
Taxanes induce mitotic arrest and tumor cell apoptosis through the hyper-stabilization of microtubules. Biomarkers that indicate the state of microtubule stability in the cell could be useful for predicting taxane response. Microtubule associated proteins (MAPs) are endogenous regulators of microtubule stability, functioning to promote or inhibit microtubule polymerization and determining subsequent cell cycle progression or mitotic arrest. These proteins may serve as potential candidates for a companion diagnostic.
MAP-tau (Tau) is a well characterized microtubule stabilizer that is responsible for the bundling, spacing, and assembly of microtubules [4][5][6]. MAP-tau may compete for taxane binding sites and/or may be involved in the cooperative binding of taxol to microtubules [7,8]. Initial reports evaluating MAP-tau as a predictive marker have been conflicting. Early studies measuring MAP-tau mRNA levels in the neoadjuvant setting found significantly lower levels in patients with pathologic complete response [7,9] but no correlation with pathologic complete response was observed in patients from a subset of the GEPARTRIO trial [10]. Similarly, in a subset of the Hellenic Cooperative Oncology Group (HeCOG) trial, MAP-tau mRNA expression status was found to be nonpredictive of benefit from paclitaxel in the adjuvant setting [11]. When MAP-tau protein expression was evaluated, most often using traditional immunohistochemical methods, conflicting results were also found. Within the adjuvant setting, in the NSABP-B 28 randomized clinical trial, there was no prediction of benefit from paclitaxel but high MAP-tau expression was a positive prognostic marker for improved survival [12]. In advanced breast cancer patients, an early study of MAP-tau expression found no prediction of benefit from taxane therapy [13]. However, two additional studies of advanced breast cancer patients found high MAP-tau expression predictive for response to paclitaxel [14,15] and a positive prognostic marker for improved overall survival [15].
The goal of this study was to clarify the prognostic and predictive value of MAP-tau. Protein expression for MAPtau was quantitatively assessed using two independent cohorts. Prognostic value was evaluated using a large Yale University retrospective cohort of untreated, primary breast cancer patients. Predictive value for MAP-tau was assessed using tumor tissue from TAX 307, a randomized clinical trial that examined patient response to the taxane, docetaxel, with docetaxel as the only variable. To date, no studies evaluating MAP-tau as a biomarker have assessed patient response using only taxane therapy as the randomized treatment variable. The TAX 307 trial randomized docetaxel-doxorubicin-cyclophosphamide (TAC) versus 5fluorouracil-doxorubicin-cyclophosphamide (FAC) as first-line chemotherapy for metastatic breast cancer. Patients were allowed to receive prior adjuvant endocrine therapy (tamoxifen) and/or chemotherapy but no prior taxanes were allowed. In this trial, inclusion of docetaxel resulted in an improved response rate (P = 0.02) but did not improve time to tumor progression (TTP) (P = 0.51) or overall survival (P = 0.93) compared with FAC alone [16].

Patient and cohort characteristics
Formalin-fixed paraffin-embedded primary breast cancer tumors resected from 651 patients at Yale University/ New Haven hospital between 1962 and 1983 were obtained from the archives of the pathology department at Yale University (New Haven, CT, USA) and have been previously described in detail [17] (see Table S1 in Additional file 1). Specimens and associated clinical information were collected under informed consent under the ethics guidelines and approval of the Yale Human Investigation Committee under protocol #8219 to DLR.
The second cohort, a prospectively collected, randomized phase III clinical trial, compared TAC versus FAC [16]. Patients were enrolled between 1 January, 1998 and 31 December, 1999, with a total of 489 patients randomized to receive either FAC (75/50/500 mg/m 2 ) or TAC (500/50/500 mg/m 2 ) as first-line chemotherapy for metastatic breast cancer. Prior adjuvant chemotherapy (but not a taxane and not > 240 mg/m2 doxorubicin) was allowed. A total of 39% of patients received prior adjuvant chemotherapy of whom 11% had received anthracyclines previously. Patients may have also received prior adjuvant hormonal therapy (described in more detail in Results).
Cycles were repeated every three weeks for six to eight cycles, depending on cumulative dose of prior doxorubicin treatment. Median patient age was 54 years with median follow-up time of 30 months, median disease-free survival of 27 months, and median number of cycles of TAC or FAC equal to six.
Baseline characteristics were well balanced and major negative prognostic factors were similar in both arms. Specimens and associated clinical information were collected under informed consent under the ethics guidelines and approval of the Dana Farber Human Investigation Committee and Yale Human Investigation Committee under protocol # 0804003757 to LH. Tumor blocks were available for 140 patients from this trial and represented 28.6% of all clinical trial participants. The TAX 307 subgroup (TAX 307S) showed no differences in patient characteristics when FAC and TAC treatment randomization groups were compared (Table 1). In addition, no significant differences in clinical characteristics were observed between TAX 307S and the original TAX 307 cohort indicating TAX 307S to be a representative subset.

Tissue microarrays and whole tissue slides
Tissue microarrays (TMAs) were constructed as previously described [17]. In brief, tissue specimens were prepared for microarray format by selecting representative breast tumor areas from 651 formalin-fixed, paraffin-embedded primary tumor blocks using hematoxylin and eosin stained whole-section slides. Breast core samples 0.6 mm in diameter were arrayed in a recipient  [18]. Additionally, a specialized Index Array was constructed to confirm assay reproducibility within both Yale University and TAX 307S cohorts and to normalize AQUA ® scores between different immunostaining run dates. Finally, a non-tumor TMA containing normal breast tissue was constructed from breast reduction mammoplasties using 110 unique patient samples with two-fold redundancy (n = 220). The TAX 307S cohort consisted of 140 conventional whole tissue (WT) slides of representative tumor tissue.

Antibodies and immunofluorescence
Yale University cohort TMAs and TAX 307S WT slides were immunostained using MAP-tau monoclonal antibody, which recognizes all human MAP-tau isoforms (1:750; mouse monoclonal, clone 2B2.100/T1029; US Biological, Swampscott, MA, USA). This antibody has been validated by western blot analysis and siRNA knock down [7]. For TAX 307S, serial sections of the index array TMA were stained alongside both cohorts to confirm assay reproducibility. Normal breast epithelium in the Yale University cohort TMAs and the TAX 307S WT slides served as internal positive controls, while omission of the primary antibody served as the negative control for each immunostaining event. Quantitative immunofluorescence staining was performed as previously described in detail (See Additional File 2).

TMA image capture and analysis
The AQUA method of quantitative immunofluorescence has been previously described [19]. The TMA cohorts were captured and analyzed using V1.6 of the AQUA ® software (HistoRx, Branford, CT, USA) on the PM2000 platform. Specific parameters related to the TMA data collection are found in Additional File 2.

Whole tissue image capture and analysis
In contrast to TMA image acquisition and analysis, WT sections from the TAX 307 clinical trial cohort required a different approach for image capture and analysis. Based on the size and contours of each resection area per slide, an image acquisition matrix was created with fields of view (FOV) ranging from 4 to 486 discrete images per slide. To avoid sampling bias in tumor image selection and to address issues of potential MAP-tau tissue heterogeneity, all cytokeratin-stained regions (rather than a variable number of regions selected at random) were collected for each tissue and quantitatively analyzed. A total of 15,816 images were collected and assessed for MAP-tau expression from the 140 cases received from the TAX 307 clinical trial. Not all cases were available for evaluation with a total of 22 cases (15.7%) missing due to tissue loss during staining or incomplete clinical trial data.

Statistical analysis
Average values for MAP-tau AQUA scores from the TMA were calculated from two-fold redundant samples and treated as independent continuous variables. The median expression level of MAP-tau from normal breast tissue served as the pre-defined cutpoint to differentiate high from low MAP-tau expression in both cohorts. Chisquare analysis was used to compare TAX 307S patient characteristics between FAC and TAC treatment groups to ensure intra-group comparability and to compare TAX 307S patient characteristics with those of the original TAX 307 clinical trial cohort. Survival curves for both cohorts were constructed using Kaplan Meier methods and the Cox-Mantel log-rank test was used to calculate the association between expression and survival. Two survival endpoints were used in this analysis. Overall survival (OS) was assessed for the Yale University cohort while TTP was evaluated for the TAX 307 clinical trial cohort. Cox proportional hazards regression analysis was used to determine which independent factors significantly impacted OS. Analyses used OS in the Yale University cohort and progression-free survival (PFS) in the TAX 307S cohort. To evaluate the association between patient response and MAP-tau expression levels, chisquare analysis was performed. Tau-by-treatment interaction was calculated to assess the relation between MAP-tau expression and docetaxel efficacy. All P values were based on two-sided testing and differences were considered significant at P < 0.05. Statistical analysis was performed using JMP Statistical Discovery Software, Version 7.0.1 (SAS Institute, Inc., Cary, NC, USA) and R, Version 2.8.0 (R Development Core Team).

MAP-tau expression and distribution
In order to establish a cutpoint that could be used to differentiate high versus low MAP-tau expression in patients, MAP-tau was measured in normal epithelial ducts and lobules in TMA format (n = 220). Average MAP-tau expression scores in normal breast tissue showed mean and median AQUA scores of 489 and 462, respectively, with a score range of 157 to 1425. The median MAP-tau expression score in normal breast tissue was subsequently used in all analyses to differentiate high expressers (AQUA score ≥ 462) from low expressers (AQUA < 462). All subsequent AQUA scores were normalized to this AQUA score range using the index array TMA. We examined MAP-tau expression within the Yale University cohort (n = 651), and found cytoplasmic localization similar to our observations in normal breast tissue ( Figure 1a). The frequency distribution of average MAP-tau expression scores in the Yale University cohort indicated mean and median AQUA scores of 498 and 210, respectively, with a range of 54 to 3017 (Figure 1b). A total of 480 cases had sufficient tumor tissue for analysis with 22% classified as high MAP-tau expressers compared with 78% as low expressers.
Estrogen receptor (ER)-negative cases showed significantly more frequent low MAP-tau (48.6%) compared with ER-positive cases (29.9%) (P < 0.0001) and this trend was mirrored for progesterone receptor (PR) status as well (Table 2). For HER2 expression, an inverse correlation between MAP-tau and human epidermal growth factor receptor (HER) 2-positive expression was observed with high MAP-tau present in only 2.4% of HER2-positive patients compared with 19.1% with high MAP-tau in HER2-negative cases. This was a particularly interesting observation as MAP-tau exists adjacent to HER2 on the 17q12 amplicon, yet rarely appears co-expressed in HER2-positive tumors. MAP-tau expression did not correlate with menopausal status, tumor size, nuclear grade, or nodal status ( Table 2).

MAP-tau prognostic value in the Yale University cohort
Patients with high MAP-tau expression (n = 94) showed improved survival compared with those with low expression (n = 339) (68.3% vs 52.9%, respectively; log-rank, P = 0.006, Figure 2a). When stratified by ER status, MAP-tau showed prognostic value in ER-negative but not in ERpositive patients (Figure 2b). In the ER-negative/high MAP-tau expressers (n = 35) we observed improved survival compared with low expressers (n = 209; 78% vs 42%; log-rank, P = 0.006). Similarly, patients stratified by HER2 status showed improved survival for high MAP-Tau/HER2 positive expression compared with low MAPtau/HER2 positive expression (log rank P = 0.007, Figure  2c), although the coexpression of MAP-tau and HER2 was a rare event. Univariate analysis showed that high  MAP-tau expression and ER and PR positive status were associated with significantly better OS (hazard ratio (HR) = 0.766 and 0.675; 95% confidence interval (CI), 0.598 to 0.981 and 0.524 to.871, P = 0.0005 and P < 0.0001, respectively), while large tumor size, nodal metastasis, increasing number of positive nodes, total nodes, and nuclear grade, were associated with worse OS (Table 3). In multivariate analysis, high MAP-tau expression was again associated with significantly improved OS. For patients with high MAP-tau expression, we observed a 24% reduction of risk (HR = 0.765; 95% CI, 0.598 to 0.957, P = 0.018; Table 4). In contrast, large tumor size, nodal metastasis, increasing total nodes, and positive HER2 status were associated with worse OS.

MAP-tau expression pattern in TAX 307S
In the TAX 307S metastatic cohort, MAP-tau expression was measured in each WT section using a matrix comprised of FOVs. All FOVs were collected and AQUA scores were generated, but each region was reviewed on a serial H&E slide to confirm that all FOVs represented infiltrating carcinoma. FOVs with normal breast ducts or ductal carcinoma in situ were excluded from the analysis. This process is illustrated in Figures 3a and 3b. Similar to the Yale University cohort, MAP-tau expression in TAX 307S remained localized to the cytoplasmic compartment within the epithelial tumor area. A total of 15,816 individual, non-overlapping FOVs were evaluated in the TAX 307S cohort. A frequency distribution summarizing the FOVs was generated for each case in the TAX 307S cohort (Figure 3c). The median score from all FOVs from each case was used to represent that case in the final subset of 108 cases. The distribution of MAP-tau expression in TAX307 for a single patient case is illustrated in Figure 3d. The median level of normal MAPtau expression as previously applied in the Yale University cohort was used to differentiate high from low expressers, and we observed 32% (35 cases) expressing high MAP-tau compared with 68% (73 cases) showing low expression in TAX 307S (Table 5). This result is consistent with the MAP-tau expression distribution in the Yale University cohort (22% high MAP-tau and 78% low MAP-tau expression; P = 0.21 and P = 0.43, respectively).
In TAX 307S, as in the Yale cohort above, low MAPtau expression was significantly more frequent in ERnegative cases (35.6%) compared with ER-positive cases (28.9%; P = 0.005). MAP-tau expression was also associated with prior adjuvant chemotherapy (anthracyclines only) with low MAP-tau expression most frequently observed in patients receiving adjuvant chemotherapy (39.8%). MAP-tau expression did not correlate with menopausal status, tumor size, tumor grade, nodal status, PR status, response to therapy, treatment arm, or prior adjuvant or metastatic endocrine therapy.

MAP-tau prognostic value in TAX 307S
Comparison analysis of TAC vs FAC treatment arms alone (not stratified by MAP-tau) in TAX 307S showed no difference in TTP (P = 0.312; Figure 4a) and confirmed original TAX 307 clinical trial results which showed no differences in median TTP (P = 0.51) or OS (P = 0.93), with a median TTP of 31 versus 29 weeks and median OS of 21 versus 22 weeks for TAC versus FAC, respectively. Improved median five-year DFS was observed for patients in TAX 307 who received TAC versus FAC (69% v 52%; P = 0.04) and the inclusion of

MAP-tau expression in the Yale University Cohort
Overall    docetaxel resulted in higher overall response rates for TAC versus FAC (55% vs 44%; P = 0.02) [16]. When TAX 307S (TAC plus FAC arms combined) was stratified by MAP-tau expression, we observed prognostic value for MAP-tau with improved median time to PFS for high MAP-tau expressers (n = 35) compared with low expressers (n = 73; 33.0 v 23.4 months) and a mean TTP of 31.2 months (log-rank, P = 0.010; Figure 4b), suggesting that this marker maintains prognostic value in metastatic patients.

MAP-tau predictive value in TAX 307S
Stratification by both treatment arm and MAP-tau expression showed improved TTP for high MAP-tau expression, regardless of treatment, which indicated prognostic but not predictive value for MAP-tau (Figure 4c, log-rank, P = 0.006). In addition, no significant interaction between MAP-tau expression and benefit from docetaxel (P = 0.843) was observed further confirming the finding of no predictive value for MAP-tau in TAX 307S. High MAPtau expression was associated with improved PFS (HR: 0.538; 95% CI, 0.333 to 0.871; P = 0.011), whereas tumor non-response (stable disease plus progressive disease) was associated with worse PFS (HR: 2.213; 95% CI, 1.404 to 3.488; P = 0.0006; Table 6). Multivariate Cox proportional hazards regression analysis showed only high MAP-tau expression was associated with significantly better TTP. For these patients, we observed a 69% reduction of risk of progression (HR = 0.308; 95% CI, 0.130 to 0.728, P = 0.007; Table 7).   [20]. ER positivity and/or tamoxifen therapy could act by artificially elevating MAP-tau levels and obscuring the relation between MAP-tau expression level and response to taxane therapy. Thus we stratified this cohort by adjuvant endocrine therapy, ER status, and also TAC vs FAC treatment. Stratification by ER status showed a trend, but not significant association between MAP-tau expression and TTP (Figure 4d, log-rank, P = 0.0615). Next, we correlated MAP-tau expression levels with response to docetaxel therapy. Response rates as a function of MAP-tau expression did not differ when split by ER status, adjuvant endocrine therapy, or taxane treatment arm (TAC vs FAC; Table 8).

Discussion
Although previous literature shows conflicting results for MAP-tau prognostic and predictive value, in this study of two independent cohorts, we find MAP-tau expression has prognostic value. In both a population-based retrospective cohort, and a randomized clinical trial with taxane as the only variable and uniformly treated metastatic patients, high levels of MAP-tau, a microtubule stabilizing protein, were associated with better outcomes. Increased microtubule stability may be associated with less aggressive tumors. However, the mechanistic reason for this observation is unknown. Furthermore, there is no correlation with nuclear grade in either cohort, suggesting that MAP-tau is not related to the molecular parameters that drive the morphological features contributing to nuclear grade.       Objective response, complete response (CR) + partial response (PR) 5 Stable disease defined as minimum of 6 weeks (2 cycles) ER, estrogen receptor; PR, progesterone receptor