From 2001 to 2008, 64 patients with rectal cancer received preoperative CRT followed by surgery at our institute. The criteria for induction of preoperative CRT in our institute are as follows. Patients must be ≤80 years old, in clinical stage II/III based on the International Union Against Cancer’s TNM classification, with no evidence of distant metastases, no invasion of external sphincter muscle nor elevator muscle of the anus, and no evidence of deep venous thrombosis. Five patients without curative surgery were excluded in this study. Additionally, we excluded 5 patients with pathological complete response after preoperative CRT because cancer cells were not obtained. A total of 54 formalin-fixed, paraffin-embedded (FFPE) specimens were investigated for this study. The study design was approved by the ethics review board of Mie University Hospital. All patients signed informed consent forms for their tissues to be used in this study.

Patients with rectal cancer were treated with short-course (a dose of 20 Gy in 4 fractions) or long-course (a dose of 45 Gy in 25 fractions) radiotherapy using a 4-field box technique with concurrent chemotherapy to take advantage of 5-fluorouracil (5-FU) radio-sensitization. Patients underwent concurrent pharmacokinetic modulation chemotherapy (intravenous infusion of 5-FU: 600 mg/m² for 24 h, and tegafur-uracil (UFT) given as 400 mg/m² orally for 5 days. This regimen was based on the previously tested combination of continuous infusion of 5-FU and oral administration of UFT (24). Short-course radiotherapy in our institute is different from standard short-course radiotherapy, 25 Gy in 5 fractions. There are several reasons that we designed the present regimen. We calculated a biologically equivalent dose (BED) of 20 Gy in 4 fractions using a linear quadratic model (25) and its BED was 30 Gy (α/β ratio: 10 Gy). We understand that BED: 30 Gy had a sufficient efficacy reducing the local failure of radiotherapy (26). Forty-two patients received short-course radiotherapy with chemotherapy over 1 week. The remaining 10 patients received long-course radiotherapy with chemotherapy for 4 weeks. The time interval between preoperative CRT and surgery was 2–3 weeks in short-course irradiation patients, and 4–6 weeks in long-course irradiation patients. All patients underwent standard surgery including total mesorectal excision, and received 5-FU based adjuvant chemotherapy after surgery for 6 months to 1 year.

The histo-pathological response of CRT was evaluated using Rödel tumor regression grading (TRG) system (27) and 3-point Ryan system (28). Each TRG was classified by two investigators in a blinded fashion without knowledge of the clinical and pathological information. Rödel TRG system is classified into five categories: grade 0, no regression; grade 1, minor regression (dominant tumor mass with obvious fibrosis in ≤25% of the tumor mass); grade 2, moderate regression (dominant tumor mass with obvious fibrosis in 26–50% of the tumor mass); grade 3, good regression (dominant fibrosis outgrowing the tumor mass; i.e., >50% tumor regression); and grade 4, total regression (no viable tumor cells, only fibrotic mass). We categorized responders as patients with TRG 3 to 4, while non-responders were TRG 0–2. Three-point Ryan system is devised by combining TRG1 (no viable cancer cells) and TRG2 (single cells or small groups of cancer cells) to form one category as 3-point TRG1, and TRG3 (residual cancer outgrown by fibrosis) into 3-point TRG2, combining TRG4 (significant fibrosis outgrown by cancer) and TRG5 (no fibrosis with extensive residual cancer) into 3-point TRG3. Patients with 3-point TRG1 and 2 were categorized as responders and patients with 3-point TRG were categorized as non-responders.

Microdissection of FFPE was performed as previously described (29). Micro-dissected specimens were digested with proteinase K in lysis buffer containing Tris-HCl, ethylenediaminetetraacetic acid, and sodium dodecyl sulfate, as previously published (30), with minor modifications. RNA was purified by phenol and chloroform extraction. Isolated RNA was purified using ethanol precipitation. The concentration and quality of RNA was measured with UV absorbance at 260 and 280 nm (A260/280 ratio).

To reverse transcribe the fragmented mRNA from FFPE tissue materials, we used random hexamer priming, instead of oligo(dT)-based priming. cDNA was synthesized with random hexamer and Superscript III reverse transcriptase (Invitrogen, Carlsbad, CA) according to the manufacturer’s instructions.

qRT-PCR analysis was carried out with the SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA) using the Applied Biosystems 7500 Real-Time PCR system according to the manufacturer’s instructions. Primers for LGR5, CD44, CTNNB1 (β-catenin) and ACTB (β-actin) were designed with Primer3 software (Biology Workbench Version 3.2, San Diego Supercomputer Center, at the University of California, San Diego). Sequences were as follows: LGR5-specific primers (sense, GATGTTGCTCAGGGTGGACT, and antisense, GGG AGCAGCTGACTGATGTT); CD44-specific primers (sense, CGGACACCATGGACAAGTTT, and antisense, CACGTGGA ATACACCTGCAA), CTNNB1-specific primer (sense, TGTT CGTGCACATCAGGATAC and antisense, GCTCCGGTACA ACCTTCAAC) and ACTB (sense, ACAGAGCCTCGCCTT TGC, and antisense, GCGGCGATATCATCATCC). PCR was performed in a final volume of 25 μl with a SYBR Green PCR Master Mix, using 1 μl cDNA, and 400 nM of each primer for the respective genes. Cycling conditions were 50°C for 2 min and 95°C for 10 min, followed by 40 cycles at 95°C for 15 sec and 60°C for 1 min.

Relative gene expression levels were determined by the standard curve method. The standard curves and line equations were generated using 5-fold serially diluted solutions of cDNA from qPCR Human Reference Total RNA (Clontech, Mountain View, CA) for LGR5 and CD44. All standard curves were linear in the analyzed range with an acceptable correlation coefficient (R²). The amount of target gene expression was calculated from the standard curve followed by quantitative normalization of cDNA in each sample using ACTB gene expression as an internal control. Target gene mRNA levels are given as ratios to ACTB mRNA levels. RT-PCR assays were done in duplicate for each sample and the mean value was used for calculation of the mRNA expression levels.

Immunohistochemistry was performed as previously described (29). LGR5 (GPR49, rabbit monoclonal antibody, clone EPR3065Y, Epitomics, CA, USA) and human CD44H antibody (monoclonal mouse IgG_2A, clone no. 2C5, R&D Systems, MN, USA) as primary antibodies were used at dilutions of 1:100 and 1:1000 respectively. All sections were counterstained with hematoxylin, and were dehydrated and mounted. Negative controls were also run simultaneously. LGR5 and CD44 expression was evaluated semi-quantitatively in a blinded fashion without knowledge of clinical and pathological information. We defined the high expression group as cases in which >50% of cancer cells. For the evaluation of immunoreactivity of CD44 in stroma, we defined the high expression group as cases that its strong expressions were localized in cancer stroma surrounding cancer nests.

All statistical analyses were performed using StatView 5.0 for Windows (SAS Institute Inc., Cary, NC). Values of each target gene are expressed as median values (inter-quartile range) in tables. Associations between continuous variables and categorical variables were evaluated using Mann-Whitney U tests for two groups. The χ² test was also used to assess the significance of the correlation between categorical variables. A non-parametric receiver operating characteristic (ROC) analysis was performed to calculate the best cutoff value for each gene expression level that would be predictive of recurrence and survival, using Medcalc 7.2 for Windows (Mariakerke, Belgium). Recurrence-free survival (RFS) and overall survival (OS) time were calculated from the date of surgery to the date of disease recurrence and patients’ death, respectively. RFS and OS probabilities were calculated using the Kaplan-Meier product limit method; intergroup differences were determined using a log-rank test. The influence of distant recurrence and survival predictors identified univariate analysis was accessed by multivariate analysis using Cox’s proportional hazards model. Two-sided P-values <0.05 were considered statistically significant.

A total of 54 total-RNA were obtained from FFPE specimens of rectal cancer with preoperative CRT. Two specimens were excluded as the ACTB expressions were very low and without reproducibility. Therefore, 52 patients were included in this study. Table I shows patient characteristics and the association of the gene expression levels of LGR5 and CD44 with clinicopathological variables. The median age was 64.5 years (range 37–78 years) and the male to female ratio was 3.7:1. The post-CRT pathological T stages were ypT1 (n=5), ypT2 (n=12), ypT3 (n=33), and ypT4 (n=2). Seventeen patients (33%) had lymph node metastases. Forty-four tumors (85%) showed well or moderately differentiated adenocarcinoma histology. Three patients (6%) had local recurrence alone. A total of 12 patients (23%) had distant recurrence. Patterns of distant recurrence were seen as liver and lung metastases in 2 patients, lung metastasis alone in 6 patients, and peritoneal metastasis in 2 patients. Rödel TRG system was as follows: grade 0, 0 patients; grade 1, 10 patients; grade 2, 22 patients; grade 3, 19 patients; and grade 4, 1 patient. 3-point Ryan TRG was as follows: TRG1, 14 patients; TRG2, 23 patients; TRG3, 15 patients. The median follow-up period was 67 months (range 22–129 months).

Elevated LGR5 gene expression was significantly correlated with the absence of vascular invasion (P=0.043), well differentiated tumor (P=0.009), and poor pathological response (Rödel system, P=0.011; Ryan system, P=0.0037). On the other hand, no significant associations of CD44 gene expression in residual cancer cells with clinicopathological variables were found (Table I).

qRT-PCR revealed that 16 of the 52 (30.8%) total patients showed detectable CD44 mRNA expression in residual cancer stroma, whereas the remaining patients had no detectable expression despite positive gene expression of ACTB. The right side of Table I shows significant correlations of stromal CD44 gene expression with long course radiation (P=0.0258) and recurrence after curative operation (P=0.0004) (Table I).

As shown in Fig. 1, there was a significant positive correlation between expression levels of LGR5 and CD44 in rectal cancer after CRT (Spearman’s ϱ = 0.451, P=0.0014).

LGR5 expression was detected immunohistochemically at the membrane and in the cytoplasm of cancer cells. CD44 expression was observed at the membrane of cancer cells and in stromal tissue surrounding residual cancer nests (Fig. 2). We divided expression into two groups, high or low, according to immunoreactivity of LGR5 and CD44. There was significant correlation between their immunoreactivities using the χ² test (P=0.0356) (data not shown).

The correlation of LGR5 and CD44 immunoreactivities with clinical outcome was evaluated. The LGR5 immunoreactivity of cancer cells was correlated with well differentiated tumor (P=0.0196). On the other hand, the CD44 immunoreactivity of cancer cells was correlated with the absence of lymph node metastasis and low postoperative stage (P=0.0092 and 0.00046, respectively). The strong immunoreactivity of CD44 in stromal cells was correlated with tumor recurrence (Rödel, P=0.012; Ryan, P=0.042, respectively) (Table II).

A significant correlation of gene expression of LGR5 with LGR5 immunoreactivity was observed (P= 0.0062). On the other hand, there was a significant correlation between CD44 mRNA levels and its immunoreactivity in cancer stroma (P=0.0138). Without reaching statistical significance, there was a correlation between CD44 mRNA levels and its immunoreactivity in cancer cells (P=0.0669) (Fig. 3).

On the basis of these results, receiver operating curve (ROC) analysis was used to identify the cutoff value for LGR5 and for CD44 gene expression levels that were predictive of disease recurrence and patients’ death. A nonparametric ROC analysis determined that the optimal cutoff values of LGR5 and CD44 were 6.169 and 0.185 for RFS, and 24.621 and 0.213 for OS. Fig. 4 shows the survival curve for RFS and OS according to LGR5 and CD44 gene expression using Kaplan-Meier method. Patients with LGR5 gene expression levels above cutoff values showed a significantly poorer RFS and OS than did patients with expression levels below cutoff values (RFS; P=0.0063, OS; P=0.0191). While, CD44 gene expression levels above cutoff values showed a significantly poorer OS than did patients with expression levels below cutoff values although CD44 gene expression level did not have a significant association with RFS (RFS; P=0.0837, OS; P=0.0131). On the other hand, patients with positive CD44 gene expression in cancer stroma showed significantly worse RFS and OS (P<0.0001). RFS and OS according to the immunoreactivity of LGR5 and CD44 did not show the significant difference (data not shown).

Table III shows the results of univariate and multivariate analyses of factors influencing patients’ prognosis using Cox’s proportional hazards model. univariate analysis showed that cancer LGR5 expression and positivity of stromal CD44 gene expression levels were significantly associated with a higher rate of developing recurrence after preoperative CRT (P=0.0136 and 0.0005, respectively). In a multivariate analysis, a positivity of stromal CD44 gene expression was found to be an independent predictive marker for disease recurrence after preoperative CRT (P=0.0028) (Table IIIA). On the other hand, univariate analysis showed that the following factors were significantly related to postoperative overall survival: disease recurrence, cancer LGR5, CD44 gene expression level and positivity of stromal CD44 gene expression (P=0.0044, 0.0287, 0.0192 and 0.0004, respectively). Multivariate analysis indicated that positivity of stromal CD44 gene expression was an independent prognostic factor for the overall survival of patients with rectal cancer after preoperative CRT (P=0.0491) (Table IIIB). The clinical variables before CRT seemed not to be influenced in patient prognosis after CRT.