Expression of KLF5 is a Prognostic Factor for Disease-Free Survival and Overall Survival in Patients with Breast Cancer

Kruppel-like factor 5 (KLF5), also called intestinal-enriched KLF5, is a DNA-binding transcriptional regulator, and contains three independent peptide modules of the C₂H₂ zinc finger type (1). KLF5 gene expression is well-regulated in embryos, with a higher level of expression towards the later stage of fetal development. It is widely expressed in human tissues including colon, small intestine, prostate, pancreas, kidney, skeletal muscle, lung, and breast (2–4). In the intestinal tract, KLF5 expression is concentrated at the base of the crypt epithelium in which active cell division occurs (5).

KLF5 has been reported to have growth-promoting effects in cultured cells based on a number of studies. Constitutive expression of KLF5 results in an increased rate of proliferation, and eventually, in a transformed phenotype, as characterized by anchorage-independent growth (6, 7). This regulatory function of KLF5 on cell proliferation might be through the stimulation of cyclin D1 (8), cyclin B1, and Cdc2 gene expression (9), or by mediating the inhibitory effect of retinoids on cell proliferation (7). KLF5 was also shown to regulate the pro-proliferative and transforming activities of oncogenic H-ras. In oncogenic H-ras-transformed NIH3T3 cells, an elevated level of KLF5 transcript was shown. The accelerated proliferation and the colony-forming ability of these cells could be significantly reduced by inducing KLF5-specific small interfering RNA that could inhibit KLF5 expression (10). In mouse fibroblasts, KLF5 expression was found to be increased after transfection with ERBB2/HER2, whose amplification and overexpression was found in a number of human cancers, including breast, ovary, and kidney carcinoma (11).

In contrast to the pro-proliferative effect of KLF5 described above, deletion and down-regulation of KLF5 has been associated with prostate (12) and breast cancer (4). Moreover, it was shown that KLF5 down-regulation is an early event in intestinal tumorigenesis in vivo (13). It reduced colony formation, failed to enhance cyclin D1 transcription, and was negatively correlated with cell growth in colon cancer cell lines. These suggest a tumor suppressor role of KLF5.

Breast cancer is the most common malignancy among women. Due to increased screening, the majority of patients present with early-stage disease. The application of adjuvant hormonal therapy and polychemotherapy after surgery reduce the risk of recurrence and death from breast cancer because distant metastatic deposits can be eradicated. However, adjuvant therapies have serious side effects and thus should only be given to patients at high-risk, which requires good prognostic factors to indicate high-risk and additional factors to predict response to treatment. Traditional prognostic factors such as lymph node status and tumor size are not accurate enough. Because tumor recurrence and death of cancer are the functional end results of the process of uncontrolled cell proliferation, molecules involved in the proliferation are directly associated with disease-free survival and overall survival of cancer patients. Based on the role of KLF5 in cell proliferation, we were interested in the prognostic value of this factor in breast cancer. We measured the KLF5 gene expression in 90 sporadic breast cancer tissues and correlated the expression with histologic data of the tumor, age at diagnosis, disease-free survival, and overall survival of the patients. In order to study the correlation of KLF5 gene expression with cell proliferation in vivo, we measured the gene expressions of the proliferation marker MKI67. In addition, we also measured the gene expression of estrogen receptor (ER), progesterone receptor (PR), and HER2 to further evaluate the prognostic value of KLF5 gene expression.

Patients with breast cancer. Ninety sporadic breast cancer patients from the Department of Obstetrics and Gynaecology, Medical University of Vienna, were included in this study (Table 1). All procedures were approved by the institutional advisory board. The ages of the patients ranged from 31 to 84 years with a mean age of 58, and a median of 57, at the time of diagnosis. Eighty-seven patients underwent a close follow-up scheme consisting of regular visits with a complete physical examination. Ultrasound examination of the abdomen and chest X-ray were done every 6 months. Mammography and bone scans were done every 12 months or in cases of suspect clinical findings. There were 37 cases of recurrence in the study. Whenever possible, recurrence was proven histologically or otherwise indicated by X-ray, computer tomography, or bone scan as measurable disease. There were 21 cases of death within the observation time. The mean observation time was 68 months.

Tumor tissues from patients with breast cancer. Fresh tumor biopsies from 90 primary sporadic breast carcinomas were collected during surgery and snap-frozen immediately after the histologic examination of frozen sections. Only samples consisting of at least 50% tumor cells were collected. The characteristics of tumors are shown in Table 1. In addition, 13 normal breast tissues were obtained from The Department of Pathology, Medical University of Vienna. These tissues were cut from a breast tumor mass and were confirmed not to contain any tumor cells. Tumor biopsies were frozen in liquid nitrogen until further processed.

Total RNA preparation. Tissues were homogenized using a microdismembrator and dissolved in lysis buffer. Total RNA was extracted from tumor biopsy lysates by isopyknic centrifugation as described previously (14), followed by a DNA digestion step of incubation with RNase-free DNase I (Roche Diagnostic, Mannheim, Germany) at 37°C for 15 minutes. The quality of the RNA was examined with RNA 6000 Nano Chips and RNA 6000 Nano Reagent & Supplies on a 2100 Bioanalyzer (Agilent Technologies, Waldbronn, Germany). RNA concentrations were determined spectrophotometrically.

Reverse transcription. Reverse transcription was carried out using Omniscript reverse transcriptase kit (Qiagen, Hilden, Germany). The total reaction volume was 20 μL, including 500 ng of RNA. The reaction mixture was incubated at 37°C for 60 minutes, heated at 95°C for 10 minutes, and then cooled on ice. After diluting the cDNA 1:5 in water, 4 μL of the mixture was aliquoted for further analysis.

Real-time PCR. The primers and probes for the quantification of KLF5 mRNA were designed with the assistance of primer express (version 2.0, Applied Biosystems, Foster City, CA; sense primer, 5′CCACCACCCTGCCAGTTAAC-3′; antisense primer, 5′-TAAACTTTTGTGCAACCAGGGTAA-3′; probe, 5′-Fam-CGATTTGGAGAAACGACGCATCCACT-Tamra-3′). The primers and probes for β2-microglobulin were included in TaqMan PDAR B2M RNA control reagent (Applied Biosystems). Primers and probes for the quantitation of ER, PR, HER2, and MKI67 were obtained from Assay-on-Demand kits (Applied Biosystems). The 5700 Sequence Detection System (Applied Biosystems) was used for real-time analysis. Four microliters of cDNA aliquot was used as the template for PCR in a total volume of 25 μL, including TaqMan Universal PCR Master Mix and the corresponding probes and primers. The mixture was preincubated at 90°C for 10 minutes followed by 50 cycles of two-step incubations at 95°C for 15 seconds and 60°C for 1 minute. All samples were measured in duplicate.

Quantitation of gene expression. The relative quantitation method with standard curves was used for the calculation of the relative amounts of mRNAs. The sample with a high expression level of a certain gene was chosen as the calibrator. Its expression was defined as 1. A standard curve using serial dilutions of the calibrator was used to calculate the amount of RNAs in other samples. Target quantities of all other samples were expressed as n-fold in relation to the calibrator. To correct the quantity differences in the starting RNA samples, the target quantity of a certain mRNA was normalized to that of the constitutively expressed housekeeping gene β2-microglobulin in the same sample.

Statistical analysis. Gene expression values were grouped into values lower or equal to and values greater than the median, and then compared between groups constituted by histopathologic data using the χ² test, which was replaced by Fisher's exact test if any expected cell size was <5. For comparison of raw gene expression levels between groups, the Mann-Whitney U test was used. The correlation of KLF5 gene expression with the expression of ER, PR, HER2, and MKI67 were estimated by Spearman's nonparametric correlation coefficient. Disease-free survival is defined as the time between diagnosis of disease and recurrence or distant metastasis. Overall survival is defined as the time from diagnosis of disease to death of patients of breast cancer. Patients who died of causes unrelated to breast cancer were treated as censored in disease-specific survival analyses. The association of gene expression groups with disease-free survival and overall survival was assessed by estimating survival curves through the method of Kaplan-Meier (15), which were compared by the log-rank test, and quantified by estimating relative risks from Cox regression analyses (16). In order to evaluate KLF5 expression as an independent prognostic factor for overall survival and disease-free survival, we did multivariable Cox regression analyses. As the limited sample size prohibits an analysis including all clinical (nodal status, tumor size, differentiation grade, and age) and gene expression variables (ER, PR, HER2, and MKI67), we first estimated a multivariable model including KLF5 expression and clinical factors such as nodal status, tumor size, differentiation grade, and age group of patient at diagnosis (≤50 years or >50 years). Second, we built a confounder-adjusted Cox regression model including KLF5 expression and all clinical and gene expression variables that changed the crude log hazard ratio of KLF5 expression by at least 10% (17). For analysis of disease-free and overall survival, tumors with differentiation grades 1 and 2 were combined for comparison to those with differentiation grade 3, and tumors with pT₁ were compared with those combining pT₂, pT₃, and pT₄. These groupings were necessary because of the low number of cases in some subgroups.

P < 0.05 was considered indicative of statistical significance. The statistical software packages R 2.0.1 (http://www.r-project.org/) and SAS 9.1.3 (2003 SAS Institute Inc., Cary, NC) were used for statistical graphics and analyses, respectively.

mRNA expression of KLF5. Relative mRNA levels of KLF5 were obtained from all tumor tissues. Figure 1 shows a histogram of log₂ of relative mRNA levels of KLF5. The data presented a normal distribution. Expression of KLF5 in normal tissues and tumor tissues did not differ significantly.

Prognostic value of gene expression, histologic data, and age of patients for disease-free survival. Analysis of disease-free survival revealed that patients with higher KLF5 expression in tumor tissue had poorer disease-free survival than patients with lower KLF5 expression (P = 0.006; Table 2; Fig. 2A). Patients with higher than median expression level of KLF5 had a 2.6-fold (95% confidence interval, 1.3-5.3) higher risk to relapse than those with KLF5 expression level lower than the median.

Log rank tests did not show significant prognostic value of expressions of ER and PR for disease-free survival. Patients with higher HER2 expression than the median had 2.1-fold (95% confidence interval, 1.0-4.1; P = 0.044) higher risk to relapse than patients with lower HER2 expression (Fig. 2B). Expression of MKI67 had significant prognostic value for disease-free survival (Fig. 2C, P = 0.021). Patients with higher MKI67 expression levels than the median had 2.4-fold (95% confidence interval, 1.1-4.9) higher risk to relapse than those with lower MKI67 expression level.

Moreover, nodal status was also shown to be a prognostic factor (P = 0.016; Table 2). Patients with lymph node involvement had shorter survival than those without lymph node involvement (Fig. 2D). Tumor size and differentiation grade had no prognostic values. Age of the patients at diagnosis did not show a significant, i.e., generalizable, effect on survival. However, the hazard ratio estimates were <1, indicating a tendency for longer survival of older patients (Table 2).

An independent effect of KLF5 expression on disease-free survival could not be confirmed in a multivariable Cox regression model adjusting for lymph node status, tumor size, differentiation grade, and age of patient at diagnosis. In the confounder-adjusted Cox regression model, none of the clinical and gene expression variables changed the crude log hazard ratio of KLF5 expression by >10%.

Prognostic value of gene expression, histopathologic data, and age of the patients for overall survival. A statistically significant difference of overall survival was found between the breast cancer patients with tumors having high KLF5 expression and tumors with low KLF5 expression (P = 0.002; Table 2; Fig. 3A). Patients with higher than the median expression levels of KLF5 had a 5.8-fold (95% confidence interval, 1.7-20) higher risk of death than patients with lower than median expression of KLF5.

Kaplan-Meier survival curves showed that higher ER and PR expressions were associated with better overall survival. However, this correlation was not statistically significant (P = 0.064 and 0.119, respectively; Fig. 3B and C). Expression of MKI67 and HER2 had no prognostic value for overall survival (P = 0.415 and 0.279, respectively).

Patients with negative nodal status tended to experience better survival than those with nodal involvement (Fig. 4A). However, this difference was not statistically significant. Similarly, patients with well (G₁) and moderately differentiated (G₂) tumor showed better overall survival than those with poorly differentiated (G₃) tumors (Fig. 4B). Patients aged 50 or younger tended to have poorer overall survival than those older than 50 years (Fig. 4C). However, these differences were not statistically significant (Table 2). For example, in disease-free survival, tumor size had no prognostic value for overall survival.

Multivariate analysis showed that KLF5 expression could be a prognostic factor of overall survival (P = 0.084), independent of nodal status, tumor size, differentiation grade, and age of patients at diagnosis. In the confounder-adjusted multivariable model, which included only KLF5 expression and nodal status, KLF5 expression presented a hazard ratio of 3.9 (95% confidence interval, 1.1-14.2; P = 0.039). None of the other variables changed the crude log hazard ratio of KLF5 expression by >10%.

Correlations of KLF5 expression with histopathologic data, age of patients, and expression of ER, PR, HER2, and MKI67. χ² analysis revealed no association of the expression levels of KLF5 and clinical data such as tumor type, differentiation grade, lymph node status, and tumor size (Table 1). There was a significant difference of KLF5 expression level in tumor tissues between patients aged 50 or younger and patients older than 50 (Mann-Whitney test, P = 0.03; Fig. 5; Table 1).

Correlation analysis did not reveal any association of the gene expression of KLF5 with ER or PR. However, KLF5 gene expression was significantly correlated with HER2 expression (n = 89, R = 0.27; P = 0.009) and with MKI67 expression (n = 89, R = 0.42; P < 0.001).

The present study clearly shows that gene expression of KLF5 is a prognostic factor for both disease-free survival and overall survival in sporadic breast cancer. The effect of KLF5 on cell proliferation was confirmed by different studies (6, 7, 10). Ki-67 is one of the most commonly used markers of proliferating cells. The mRNA level of the MKI67 gene coding for the Ki-67 antigen was shown to be correlated with Ki-67 immunostaining in breast cancer cells (18). We found a significant correlation of gene expression between KLF5 and MKI67, indicating that KLF5 gene expression is associated with cell proliferation in vivo. In breast cancer, the primary tumor is usually removed by surgery. Recurrence of breast cancer locally or at distant organs is a result of the growth of residual breast tumor cells which are not removed by surgery and escaped from the adjuvant therapy. To form a measurable disease, these cells have to grow to a certain size. KLF5 promotes tumor cell growth; we therefore assumed that it might be correlated with the recurrence of disease. Indeed, our results showed that higher expression of KLF5 was correlated with shorter disease-free survival. Breast cancer is a disease that metastasizes via the lymphatic system or hematogenously in a very early stage. Hence, prognosis does not depend on the mere presence of distant metastases, but on whether they will grow to a clinically relevant size (19). Higher KLF5 expression is not only related to earlier recurrence but early death as well, again confirming that KLF5 plays a main role in the growth of breast tumor cells.

Adjuvant chemotherapy and hormone therapy have been shown to improve the survival of patients with breast cancer. However, as these therapeutic agents have serious side effects, ideally, treatment should only be given to patients at high-risk (19). Traditional prognostic factors such as auxillary lymph node status and tumor size are not accurate enough to identify patients at low risk from patients at high-risk. Thus, additional prognostic factors are required. Gene expression of KLF5 was shown to be indicative of the risk of recurrence in patients with breast cancer. Our data also showed that KLF5 was an independent prognostic factor for overall survival of these patients. As the significance was statistically marginal, studies with a higher number of samples are needed to prove the prognostic value of KLF5 expression shown.

Breast cancer in younger patients seems to be more aggressive than in older patients. Many studies observed that younger breast cancer patients have shorter disease-free survival and overall survival (20–23). In our study, we did not find poorer disease-free survival, but a poorer overall survival for younger patients, which is not significant. Both hazard ratio estimates are <1, showing the tendency of younger patients to have poorer survival. Again, this is not significant. In addition, we found a negative correlation of KLF5 expression with the patient's age, indicating that KLF5 is highly expressed in younger patients than in older ones. Thus, younger patients with higher expression levels of KLF5 would be expected to have more progressive patterns of tumor cell proliferation, which could lead to a faster tumor growth, and finally, shorter disease-free survival and overall survival time. However, studies with larger sample size will again be needed to clarify the relation of KLF5, age of the patients with breast cancer, and survival.

Traditionally, protein levels were used for diagnosis and prognosis because mRNA amount is not always correlated with protein level. In the case of KLF5, several publications have shown that mRNA expression is correlated either with the protein expression and/or directly with cell proliferation in cell culture models (6, 7, 10). Indeed, our data show that the KLF5 mRNA level is a prognostic factor. Recently, with the improvement of microarray technology, mRNA levels are more often investigated as prognostic factors. It will be interesting to study the correlation of mRNA and protein level of KLF5 in tumor tissues.

Our results showed no statistical differences of KLF5 expression between tumor tissue and normal breast tissue. This might be due to the small sample size of normal tissues. Thus, it will be very interesting to study this correlation with more samples.

In conclusion, our study showed that KLF5 expression is a prognostic factor for both disease-free survival and overall survival of patients with breast cancer. The prognostic value of KLF5 may be explained by its effects on promoting cell proliferation. In addition, KLF5 expression was negatively correlated with age of patients at diagnosis.