We thank Ms. Ilona Schliephake and Ms. Christine Drengenes for their excellent technical assistance.
The current study was performed to investigate the potential impact of tumor cell expression of estrogen receptor-α (ER-α), progesterone receptor (PR), and androgen receptor (AR) on the outcomes of patients who received radiotherapy (RT) for non-small cell lung cancer (NSCLC).
Tumor cell expression of ER-α, PR, and AR as well as 9 additional potential prognostic factors were retrospectively evaluated in 64 patients who underwent RT for AJCC stage II/III NSCLC. The endpoints investigated were locoregional control, metastases-free survival, and overall survival. The additional potential prognostic factors were age, gender, Karnofsky performance score, histology, T classification, N classification, surgery, smoking during RT, and hemoglobin levels during RT. Subgroup analyses were performed for women and men.
On univariate analysis, locoregional control was not found to be associated with expression of PR or AR. ER-α expression demonstrated a strong trend toward worse locoregional control. On multivariate analysis, ER-α expression was found to be significantly associated with worse locoregional control (risk ratio [RR], 3.12; P = .035). On univariate analysis, metastases-free survival was not associated with expression of ER-α, PR, or AR. On univariate analysis, survival was found to be negatively associated with expression of ER-α (P = .003) but not with PR or AR expression. On multivariate analysis, ER-α expression maintained significance (RR, 2.73; P = .022).
Lung cancer remains the leading cause of cancer-related death worldwide.1 Approximately 80% of all lung cancers are non-small cell lung cancers (NSCLCs). Overall, the prognosis is poor, with a 5-year survival rate of only 16%. Thus, improvements in treatment strategies for NSCLC are of great interest. Prognostic factors and tumor markers often guide the physician in selecting the appropriate treatment for the individual patient. Recent studies have suggested that the tumor cell expression of hormone receptors may have an impact on the prognosis of patients with NSCLC. A study from the United States and Norway reported that the survival prognosis of patients with NSCLC was inversely associated with tumor estrogen receptor-α (ER-α) levels.2 In contrast, a study from Canada suggested that a higher ER-β level was associated with better survival in patients with NSCLC.3 A study from China suggested that expression of androgen receptor (AR) may be associated with disease progression and lymph node metastasis of lung cancer.4 The Women's Health Initiative trial demonstrated that both estrogen and medroxyprogesterone acetate lead to an increase in deaths from lung cancer in postmenopausal women.5 Thus, progesterone and progesterone receptor (PR) expression may also have an impact on the prognosis of patients with NSCLC. Taking into account that only very scarce data are currently available, further studies investigating the potential prognostic impact of tumor cell expression of hormone receptors on the prognosis of patients with NSCLC appear important. The current study investigated the potential prognostic role of tumor cell expression of ER-α, PR, and AR with respect to locoregional control, metastases-free survival, and overall survival in patients who received radiotherapy (RT) for NSCLC.
MATERIALS AND METHODS
Tumor cell expression levels of ER-α, PR, and AR were evaluated in tumor samples obtained before RT from 64 patients who were irradiated for NSCLC at the University of Lubeck between January 2000 and December 2005. Staining was not successful in another 8 patients. Patient characteristics are summarized in Table 1.
The tissue samples were obtained from either biopsies (bronchoscopy; N = 29) or resected specimens (N = 35). The tissues were fixed in 10% buffered formalin (pH 7.0) and embedded in paraffin. The formalin-fixed, paraffin-embedded tumor samples were used for the preparation of a tumor tissue microarray (TMA) block. The TMA block was constructed using a Manual Tissue Arrayer 1 (Beecher Instruments Inc, Silver Spring, Md) with a core biopsy needle measuring 1.0 mm in diameter. Subsequently, 4-μm–thick serial sections were prepared from the TMA block and deparaffinized in xylene and rehydrated in graded alcohols. Antigen retrieval was performed in 0.01 mol/L of sodium citrate buffer (pH 6.0) for 5 minutes in a microwave for expression of ER-α and PR. The sections were treated in a decloaker (pH 9.0) for AR expression. Endogenous peroxidase was blocked with 0.3% hydrogen peroxidase for 5 minutes. Sections were incubated with a monoclonal mouse antibody (clone 1D5, 1/50 dilution; Dako, Glostrup, Denmark) for ER-α expression, a monoclonal rabbit antibody (clone SP2, 1/100 dilution; Dako) for PR expression, and a monoclonal mouse antibody (clone AR441, 1/100 dilution; Dako) for AR expression. Sections were washed with Tris-buffered saline containing 0.1% Tween 20 (pH 7.0) and the subsequent reaction was performed with the biotin-free horseradish peroxidase enzyme-labeled polymer of the PowerVision system (Immunologic, Duiven, the Netherlands). A diaminobenzidine complex was used as a chromogen. Sections were counterstained with hematoxylin. Tumors were considered to express ER-α, PR, and AR if ≥ 5% of the tumor cells were positive.
For ER-α and PR staining, breast cancer tissue served as a positive control. For AR staining, prostate tissue and prostate cancer tissue served as positive controls. Negative controls were obtained from staining of tissues without the primary antibodies. Potential intratumor heterogeneity was investigated for each of the 3 receptors: ER-α, PR, and AR. For each receptor, 3 resected specimens of tumors that were positive on TMA staining served as controls. The resected specimens were divided into 4 quadrants. Each of the 4 quadrants was compared with the others for quality and intensity of the staining of the investigated receptors. The staining was found to be similar in each of the 4 quadrants in each investigated specimen, and was also similar to the staining on TMA.
RT was performed after computed tomography (CT)-based 3-dimensional treatment planning with a linear accelerator and 6- to 18-megavolt photons. The target volume included the primary tumor and the locoregional lymph nodes with a margin of 2 cm. The dose per fraction was 2.0 gray (Gy) given once daily, 5 days per week, up to a total dose of 60 Gy, 66 Gy, or 70 Gy in 61 patients. In 3 patients who were treated with definitive RT, the dose per fraction was 1.4 Gy given twice daily, 5 days per week, up to a total dose of 72.8 Gy. To better compare the fractionation regimens used in this study, the total radiation dose was given as the equivalent dose in 2-Gy fractions (EQD2). The EQD2 ranged between 60 Gy and 70 Gy, based on the treatment schedule favored at the study institution during different periods and on the extent of resection performed in patients undergoing surgery. It was calculated with the equation EQD2 = D × [(d + α/β)/ (2 Gy + α/β)], in which D indicated the total dose, d indicated the dose per fraction, α indicated the linear component of cell killing, β indicated the quadratic component of cell killing, and the α/β ratio indicated the dose at which both components are equal.6, 7 The α/β ratio for tumor cell kill was assumed to be 10 Gy. The EQD2 was 66 Gy to 70 Gy after R2 resection (macroscopically residual tumor) or for definitive treatment, 60 Gy to 66 Gy after R1 resection (microscopically residual tumor), and 60 Gy after R0 resection (no residual tumor). A total of 33 patients received additional chemotherapy, which was comprised of 2 to 4 cycles of a cisplatin-based regimen. Two cycles were administered concurrently with RT.
Potential Prognostic Factors
Twelve potential prognostic factors were retrospectively evaluated with respect to locoregional control, metastases-free survival, and overall survival. These factors included age (< 65 years vs ≥ 65 years; median age, 65 years), gender, Karnofsky performance score (≤ 70 vs > 70), histology (squamous cell carcinoma vs adenocarcinoma vs large cell carcinoma), T classification (T1-T3 vs T4) and N classification (N0-N1 vs N2-N3) according to the sixth edition of the TNM classification for lung cancer, surgery (no vs yes), smoking during RT (no vs yes), > 50% of the weekly hemoglobin levels during RT (< 12 g/dL vs ≥ 12 g/dL)8, ER-α expression of tumor cells (no vs yes), PR expression of tumor cells (no vs yes), and AR expression of tumor cells (no vs yes).
Locoregional control, metastases-free survival, and overall survival were calculated with the Kaplan-Meier method.9 Differences between Kaplan-Meier curves were calculated with the log-rank test (univariate analysis). According to the Bonferroni correction for 12 tests (investigated potential prognostic factors), the results were considered to be statistically significant for a P value < .0042 (α level, .05). The potential prognostic factors that were found to be statistically significant or demonstrated a strong trend (P< .06) in the univariate analysis were evaluated in a multivariate analysis performed using the Cox proportional hazards model. Time to locoregional failure, time to distant failure, and time to death were referenced from the end of RT. Locoregional failure was identified by endoscopy or CT.
Additional subgroup analyses were also performed separately for men and women with regard to the impact of expression of ER-α, PR, and AR with respect to locoregional control, metastases-free survival, and overall survival.
On univariate analysis, improved locoregional control was not found to be significantly associated with any of the investigated potential prognostic factors. ER-α expression demonstrated a strong trend toward worse locoregional control (P = .016) (Fig. 1), whereas no association with PR expression (P = .70) or AR expression (P = .87) was observed. A strong trend toward improved locoregional control was observed for lower N classification (P = .038) and not smoking during RT (P = .056). The results of the univariate analysis of locoregional control are summarized in Table 2. On multivariate analysis of locoregional control, ER-α expression (risk ratio [RR], 3.12; 95% confidence interval [95% CI], 1.17-7.64 [P = .035]) maintained statistical significance. N classification demonstrated a strong trend (RR, 1.60; 95% CI, 0.99-2.83 [P = .058]), whereas smoking during RT was not found to be significant (RR, 1.74; 95% CI, 0.68-4.24 [P = .24]).
According to the Bonferroni correction, results were considered significant for a P value <.0042 (alpha level, .05).
Large cell carcinoma
Smoking during RT
Hemoglobin levels during RT, g/dL
On univariate analysis, metastases-free survival was not found to be significantly associated with PR expression (P = .37) or AR expression (P = .67). ER-α expression demonstrated a trend toward worse metastases-free survival (P = .07) (Fig. 2). In contrast, improved metastases-free survival was found to be significantly associated with squamous cell carcinoma (P = .003) and higher hemoglobin levels during RT (P = .004). The results of the univariate analysis of metastases-free survival are summarized in Table 2. On multivariate analysis of metastases-free survival, hemoglobin levels during RT (RR, 3.47; 95% CI, 1.51-9.39 [P = .002]) maintained statistical significance, whereas histology did not (RR, 1.21; 95% CI, 0.94-1.48 [P = .13]).
The median survival time for the entire cohort was 26 months. On univariate analysis, improved overall survival was found to be inversely associated with ER-α expression (P = .003) (Fig. 3) but was not associated with PR expression (P = .09) or expression of AR (P = .64). On univariate analysis, improved overall survival was also found to be significantly associated with hemoglobin levels of ≥ 12 g/dL during RT (P = .001). Lower N classification (P = .027), surgery (P = .049), and no smoking during RT (P = .043) indicated a strong trend toward better overall survival. The results of the univariate analysis of overall survival are summarized in Table 2. On multivariate analysis of overall survival, the absence of ER-α expression (RR, 2.73; 95% CI, 1.17-6.07 [P = .022]) and higher hemoglobin levels during RT (RR, 3.55; 95% CI, 1.16-13.51 [P = .025]) maintained statistical significance. Lower N classification (RR, 1.24; 95% CI, 0.77-2.13 [P = .38]), surgery (RR, 1.67; 95% CI, 0.75-3.77 [P = .21]), and not smoking during RT (RR, 1.41; 95% CI, 0.60-3.18 [P = .42]) were not found to be significant. The patients were followed for a median of 16 months (range, 2 months-64 months) for the entire cohort, and for a median of 20 months (range, 12 months-64 months) in those patients who were alive at the time of last follow-up.
In the additional subgroup analyses, improved locoregional control in women (P = .003) and improved overall survival in men were found to be inversely associated with expression of ER-α (P = .040). The results of the subgroup analyses are summarized in Table 3 (women) and Table 4 (men).
Table 3. Subgroup Analysis of the Potential Prognostic Role of Tumor Cell Expression of ER-α, PR, and AR on LRC, MFS, and OS in Women
To date, the potential prognostic impact of ER-α, PR, or AR expression of NSCLC cells on patient prognosis has been unclear. To our knowledge, this is the first study to investigate the potential prognostic role of these 3 receptors in patients who underwent RT for locally advanced NSCLC. According to the results of the current study, tumor cell expression of ER-α was found to be inversely associated with improved treatment outcome in terms of locoregional control, metastases-free survival, and overall survival. No such correlation was observed for PR or AR expression. The results with respect to ER-α expression were significant in both univariate and multivariate analyses. Thus, tumor cell expression of ER-α can be considered to be an independent prognostic factor with respect to the 3 investigated endpoints.
These results are in accordance with the study by Olivo-Marston et al, who reported tumor expression of ER-α to be significantly associated with poor overall survival in patients with NSCLC.2 In both the current study and that of Olivo-Marston et al, ER-α was found to be a significant predictor of prognosis in both women and men. However, unlike our study, Olivo-Marston et al did not investigate locoregional control and metastases-free survival.2 In contrast, a study from Canada of 79 patients with NSCLC suggested that a higher ER-β level was associated with better survival in both the univariate (P = .009) and multivariate (hazard ratio, 0.37; 95% CI, 0.18.-0.77 [P = .008]) analyses.3 Thus, the prognostic impact of ER-α and ER-β appears to be different. A study from China investigated the prognostic impact of ER-α, as performed in the current study, and did not find a significant association between ER-α expression and lymph node metastasis.4 However, these authors did not investigate metastases-free survival and overall survival. The study from China also investigated AR expression and found that N2 tumors were more frequently associated with AR expression than N0 tumors.4
To the best of our knowledge, the current study is the first to include only patients with NSCLC who received RT. According to our results, tumor cell expression of ER-α is a prognostic factor for treatment outcome. Because ER-α was found to be significant in the multivariate analyses, it can be considered to be a prognostic factor that is also independent from other prognostic factors such as tumor stage, histology, and N classification. The retrospective nature of the current study and the relatively small number of patients examined should be considered when interpreting the results. However, to the best of our knowledge, prospective studies regarding the potential prognostic impact of tumor expression of ER-α, PR, or AR are not available.
In addition to tumor cell expression of ER-α, treatment outcomes were found to be associated with hemoglobin levels during RT, and a trend was observed for N classification. The prognostic impact of tumor stage in patients with NSCLC has been described by several authors.8-16 In addition, the prognostic value of hemoglobin levels during RT has been described for other primary tumor types.17-21 In several studies of patients with head and neck cancer, hemoglobin levels < 12 g/dL led to a decrease in tumor oxygenation because of the reduced oxygen-carrying capacity of the blood. Inappropriate tumor oxygenation impairs the treatment effect of RT, which is based on the induction of cytotoxic free oxygen radicals damaging the DNA of the cancer cells.
According to the results of the current study, tumor cell expression of ER-α is a negative prognostic factor for treatment outcomes in both women and men who received RT for stage II/III NSCLC. PR and AR expression did not appear to be of predictive value.