Although somatic alterations in CAG repeats in the androgen receptor (AR) gene have been suggested to predispose to colorectal cancer, less is known about AR in colorectal cancer carcinogenesis. Because of lack of relevant analysis on CAG repeat length and AR expression in colorectal cancer, we aimed to investigate the prognostic value of polymorphic CAG and protein expression of the AR gene in patients with colorectal cancer. A case–control study was carried out on 550 patients with colorectal cancer and 540 healthy controls to investigate whether polymorphic CAG within the AR gene is linked to increased risk for colorectal cancer. Polymorphic CAG and AR expression were analyzed to clarify their relationship with clinicopathologic and prognostic factors in patients with colorectal cancer. The study showed that the AR gene in patients with colorectal cancer had a longer CAG repeat sequence than those in the control group, as well as increased risk for colorectal cancer among females (P = 0.013), males (P = 0.002), and total colorectal cancer population (P < 0.001), respectively. AR expression exhibited a significant difference in long CAG repeat sequence among males (P < 0.001), females (P < 0.001), and total colorectal cancer study population (P < 0.001). Both long CAG repeat sequence and negative AR expression were associated with a short 5-year overall survival (OS) rate in colorectal cancer. Long CAG repeat sequences and the absence of AR expression were closely related to the development of colorectal cancer. Both long CAG and decreased AR expression were correlated with the poor 5-year OS in patients with colorectal cancer. Mol Cancer Ther; 14(4); 1066–74. ©2015 AACR.

The principal factor in colorectal carcinogenesis still remains unclear. Colorectal cancer could be developed through different genetic pathways (1). Colorectal tumors relate to these genetic changes involving the chromosomal instability pathway and some hereditary syndromes, such as familial adenomatous polyposis (2). Recent studies have showed that androgens regulate cellular growth and differentiation in several hormone-dependent tissues, including colorectal tissue (3, 4). The androgen receptor (AR) is a ligand-dependent transcription factor that is involved in controlling cellular proliferation and differentiation (5). AR is located on chromosome Xq11-12 and contains a variable number of CAG repeats, which are polymorphic (CAG 8–35) in a normal human population (6). The CAG repeat length of AR inversely affects its transactivation potential, either as a directly altered receptor function (7, 8) or indirectly reduced AR messenger at RNA and protein levels (9). Somatic alterations (10) and instability (11) of the CAG repeat in the AR gene are correlated with colorectal cancer development. However, the genetic influences and protein expression of AR in colorectal cancer carcinogenesis remain unknown.

In this study, we examined the association between polymorphisms of the AR gene and risk for colorectal cancer, as well as the associations between protein expression and polymorphic CAG to the prognostic significance of colorectal cancer. We hypothesized that the risk of colorectal cancer would vary with polymorphic CAG repeat lengths of the AR gene. In addition, we also evaluated the associations and interactions between long and short CAG repeat sequences in different expression levels of AR protein.

Study populations

In this study, patients and controls were from different hospitals, namely, The Second Affiliated Hospital of Harbin Medical University (the second AHMU), Third Affiliated Hospital of Harbin Medical University (the third AHMU) in Heilongjiang Province northeast China, and Affiliated Hospital of Inner Mongolia Medical University in Inner Mongolia Autonomous Region in northwest China. All patients and controls have provided informed consent. The study has been approved by the Heilongjiang and Inner Mongolia Regional Ethics Committee.

Controls

The subjects included in the control group were matched by gender and age, and selected using of a sampling frame, targeting the same population from the same place. Blood samples were collected from 540 cancer-free healthy donors (310 males and 230 females) without a family history of colorectal cancer from February 2012 to August 2012.

Patients

Data were collected in 610 consecutive patients with colorectal cancer who were operated from January 2004 to December 2008 and underwent AR analysis in biomolecular laboratories from different universities. Patients with inflammatory bowel disease, hereditary nonpolyposis colon cancer, familial polyposis, and metastatic carcinoma were excluded from this study. All patients received surgery, without giving any adjuvant therapy before surgery. Histopathologic assessment confirmed the diagnosis of colorectal cancer.

Supplementary Table S1 showed distribution of clinical and pathologic features of patients with colorectal cancer. The pooled 550 patients with colorectal cancer were used to study the germ-line CAG repeat length. A total of 290 patients who met the same AR(CAG)n repeat length of two paired tumor tissues and blood samples were used as test cohorts in the Affiliated Hospital of Harbin Medical University. Furthermore, 260 patients met the same AR(CAG)n repeat length of two paired tumor tissues and blood samples from each patient were used as validation cohorts in the Affiliated Hospital of Inner Mongolia Medical University. The pooled 550 CRC patients comprised 238 females and 312 males (ratio of 1:1.31). In this study, the average age in the patients group was 59.6 years old (range, 34 to 88 years). The tumor grade and stage were classified according to the seventh edition of AJCC (American Joint Committee on Cancer) staging system (12). Among the 550 patients with colorectal cancer, both nontumoral (from surrounding “normal” mucosa at least 10 cm away from the tumor) and malignant tissue specimens were collected from 150 patients. The medical and follow-up data of patients, including age, gender, lymph node metastases, type of T stage, diameter and location of primary tumor, tumor–node–metastasis (TNM) stage, degree of primary tumor differentiation, growth pattern, and adjuvant chemotherapy, were retrieved from hospital records and interviews. Adjuvant chemotherapy consisted of two regimens for FOLFOX (5-Fluorouracil, Folinic Acid, and Oxaliplatin) and XELOX (capecitabine plus oxaliplatin).

DNA extraction

The specimens came from both patients group and controls group, including formalin-fixed paraffin-embedded (FFPE) tumor tissues and paired blood samples. DNA extraction complied with the instruction using Tissue DNA Kit (OMEGA) or Blood Genomic DNA Kit (AXYGEN), respectively. Microdissection was used to exclude the normal tissue before FFPE tumor tissues extracted DNA.

PCR-based GeneScan analysis of AR CAG repeats length

After DNA was extracted from specimens, a region containing the polymorphism of repeat CAG sequence was amplified by PCR. Primers (13) were AR-Forward 5′-GTTTCTGTGGGGCCTCTACGATGG- 3′ and AR-Reverse 5′- GTTTCTGCGCGAAGTGATCCAGAA- 3′. AR-Forward was fluorescently labeled with 6-carboxy-fluorescine (FAM). PCR was carried out as we have previously described (14). Data were analyzed by GeneScan Software (Applied Biosystems) and Genetyper Software (Applied Biosystems).

Immunohistochemistry

As described previously (14), 4-μm sections from FFPE tissues were dewaxed and rehydrated according to routine procedure. Sections were permeabilized in 0.5% Triton-X100 for 20 minutes and incubated overnight with primary antibody (1:100; ab74272; Abcam), followed by biotinylated anti-rabbit immunoglobulin secondary antibody for 30 minutes. Next, sections were visualized by DAB (3,3′-diaminobenzidine tetrahydrochloride) and counterstained with hematoxylin. Positive controls for AR were from prostate carcinoma tissues. PBS replaced the primary antibody as the negative control. AR-positive expression was determined according to the percentage and intensity (15, 16).

Statistical analysis

The relationship among AR status, polymorphism of CAG repeat, and clinicopathologic factors was analyzed with a χ2 test. Univariate logistic regression with patients with colorectal cancer was used to assess the association of AR expression with AR(CAG)n repeat length in individual patients. The Kaplan–Meier method was used for univariate survival analyses. The Cox proportional hazard regression model was used for estimating the relative importance of unitary and multiple prognostic factors on survival. Date of diagnosis (colonoscopy or surgery) to death or the date last known alive was calculated as overall survival (OS). SPSS 18.0 statistical software (Austin) was used for statistical analysis and P < 0.05 was considered significant.

Comparison of CAG repeat length between the colorectal cancer and control groups

The PCR products obtained ranged from 206 bp to 287 bp in length (equivalent to 8 and 35 CAG repeats, respectively). Given that the median number of CAG repeats in 550 patients with colorectal cancer was 21.5 ± 2.5, our study selected 22 CAG repeats of 248 bp in length as cut-points. Heterozygous women harboring an allele with more than 22 CAG repeats were classified as having “long” CAG repeat sequences. Figure 1A–D showed “short” and “long” CAG repeat sequences in males, homozygous females, and heterozygous females, respectively. Figure 2A–C showed the distribution of CAG repeat sequences between the control group and patients with colorectal cancer (males, females, and total population, respectively). Supplementary Fig. S1A and S1B showed distribution of “short” and “long” CAG repeat sequences in cases and controls from the different centers. Supplementary Table S2 illustrated that alleles in the control group mainly comprised of 22 or fewer CAG repeats (males, 55.4%; females, 54.3%; total population, 54.9%), whereas alleles in the colorectal cancer group (males, 56.6%; females, 56.5%; total population, 56.5%) consisted of 22 to 35 repeats. Supplementary Table S2 also showed that long CAG repeat sequences increased the risk for colorectal cancer among males [OR, 1.293; 95% confidence interval (CI), 1.090–1.534; P = 0.002], females (OR, 1.255; 95% CI, 1.035–1.523; P = 0.013), and total population (OR, 1.277; 95% CI, 1.123–1.451; P < 0.001).

Figure 1.

AR gene polymorphism analysis using the PCR-based Genescan. A, allelotype of short CAG repeat sequence (≤22 CAG repeat, 248 bp) in male and homozygous female. B, allelotype of long CAG repeat sequence (>22 CAG repeat, 248 bp) in homozygous female. C, allelotype of short CAG repeat sequence (double allele ≤22 CAG repeat, 248 bp) in heterozygous female. D, allelotype of long CAG repeat sequence (any allele >22 CAG repeat, 248 bp) in heterozygous female.

Figure 1.

AR gene polymorphism analysis using the PCR-based Genescan. A, allelotype of short CAG repeat sequence (≤22 CAG repeat, 248 bp) in male and homozygous female. B, allelotype of long CAG repeat sequence (>22 CAG repeat, 248 bp) in homozygous female. C, allelotype of short CAG repeat sequence (double allele ≤22 CAG repeat, 248 bp) in heterozygous female. D, allelotype of long CAG repeat sequence (any allele >22 CAG repeat, 248 bp) in heterozygous female.

Close modal
Figure 2.

Distribution of CAG repeat length among patients with colorectal cancer and control groups and the Kaplan–Meier curves of patients with colorectal cancer affected by CAG repeat length. A–C, distribution of CAG repeat length among males, females, and total number of patients with colorectal cancer compared with the control groups, respectively; D and E, 5-year OS of patients with colorectal cancer affected by CAG repeat length in the test and validation cohorts. F–H, 5-year OS of males, females, and pooled patients with colorectal cancer affected by CAG repeat length, respectively.

Figure 2.

Distribution of CAG repeat length among patients with colorectal cancer and control groups and the Kaplan–Meier curves of patients with colorectal cancer affected by CAG repeat length. A–C, distribution of CAG repeat length among males, females, and total number of patients with colorectal cancer compared with the control groups, respectively; D and E, 5-year OS of patients with colorectal cancer affected by CAG repeat length in the test and validation cohorts. F–H, 5-year OS of males, females, and pooled patients with colorectal cancer affected by CAG repeat length, respectively.

Close modal

Analysis of AR protein expression and association between AR and CAG repeat length in colorectal cancer

The positive expression of AR protein was mainly weak, and moderate staining was observed in colorectal cancer cell nuclei. AR staining demonstrated the following results: 315 cases (male, 175 cases; female, 140 cases) showed negative immunostaining (Fig. 3A), and 235 cases (male, 137 cases; female, 98 cases) showed positive immunostaining (Fig. 3B). Supplementary Table S3 showed that AR expression was positive in 42.7% of the 550 tumor tissues and 81.3% of the 150 paired “normal” mucosa tissues (Fig. 3C; P < 0.001). Supplementary Table S4 showed the relationship between the CAG repeat length and AR expression in patients with colorectal cancer. Long CAG repeat sequences exhibited a significant difference between the negative and positive AR subgroups in females (P < 0.001), males (P < 0.001), and total colorectal cancer population (P < 0.001). Univariate logistic regression with patients with colorectal cancer indicated the association of AR expression with AR(CAG)n repeat length (OR, 0.048; 95% CI, 0.031–0.075; P < 0.001) in individual patients.

Figure 3.

Immunohistochemical staining of the AR in tumor tissue compared with paired normal tissue and the Kaplan–Meier curves of patients with colorectal cancer affected by AR expression status. A, negative staining in tumor tissue. B, positive staining in tumor tissue. C, strong positive staining in paired normal tissue. D and E, 5-year OS of patients with colorectal cancer affected by AR expression status in test and validation cohorts. F–H, 5-year OS of male, female, and pooled patients with colorectal cancer affected by AR expression status, respectively.

Figure 3.

Immunohistochemical staining of the AR in tumor tissue compared with paired normal tissue and the Kaplan–Meier curves of patients with colorectal cancer affected by AR expression status. A, negative staining in tumor tissue. B, positive staining in tumor tissue. C, strong positive staining in paired normal tissue. D and E, 5-year OS of patients with colorectal cancer affected by AR expression status in test and validation cohorts. F–H, 5-year OS of male, female, and pooled patients with colorectal cancer affected by AR expression status, respectively.

Close modal

In this study, their distributions were similar between the test and validation cohorts. By the time of the final analysis on December 30, 2013, 110 patients (37.9%) in the test cohort died of colorectal cancer, with a median survival time of 46.4 months (5-year OS rate 62.1%). Among the patients in the validation cohort, 103 (39.6%) died of colorectal cancer, with a median survival time of 46.0 months (5-year OS rate 60.4%). The 5-year OS rate for the pooled sample of two cohorts was 61.3%.

We also investigated the associations of CAG repeat lengths and AR expression in some clinical and pathologic features, such as age, gender, tumor size, type of T stage, location of primary tumor, lymph node metastases, TNM stage, degree of differentiation, growth pattern, and adjuvant chemotherapy (Table 1). AR expression and long CAG repeat sequences were related to large tumors (≥5 cm; AR, P = 0.001; CAG, P = 0.001) and degree of differentiation (AR, P < 0.001; CAG, P = 0.028), respectively. AR expression was related to pathologic T stage (P = 0.001). Long CAG repeat sequences showed a higher percentage in pathologic N1-2 stage (P < 0.001).

Table 1.

AR expression and AR(CAG)n repeat length in patients with colorectal cancer in relation to clinical and pathologic features

Number of AR (n = 550)Number of CAG repeat length
FactorNAR (−) (%)AR (+) (%)PShort (%)Long (%)P
Age, y    0.441   0.072 
 <60 277 160 (51%) 117 (50%)  140 (54%) 137 (47%)  
 ≥60 273 155 (49%) 118 (50%)  120 (46%) 153 (53%)  
Gender    0.290   0.499 
 Male 312 175 (56%) 137 (58%)  148 (57%) 164 (57%)  
 Female 238 140 (44%) 98 (42%)  112 (43%) 126 (43%)  
Largest tumor diameter    0.001   0.001 
 <5 cm 289 147 (47%) 142 (60%)  155 (60%) 169 (46%)  
 ≥5 cm 261 168 (53%) 92 (40%)  105 (40%) 121 (54%)  
Pathologic T stage    0.001   0.056 
 T1-2 248 123 (39%) 125 (53%)  127 (49%) 121 (42%)  
 T3-4 302 192 (61%) 110 (47%)  133 (51%) 169 (58%)  
Pathologic N stage    0.076   <0.001 
 N0 237 127 (40%) 110 (46%)  133 (51%) 104 (36%)  
 N1-2 313 188 (60%) 125 (54%)  127 (49%) 186 (64%)  
Tumor location    0.308   0.291 
 Rectum 301 169 (54%) 132 (65%)  146 (56%) 155 (53%)  
 Colon 249 146 (46%) 103 (35%)  114 (44%) 135 (47%)  
Differentiation    <0.001   0.028 
 Well 238 114 (36%) 124 (53%)  128 (49%) 110 (38%)  
 Moderate 229 146 (46%) 83 (35%)  96 (37%) 133 (46%)  
 Poor 83 55 (18%) 28 (12%)  36 (14%) 47 (16%)  
TNM stage    0.613   0.415 
 I 179 108 (34%) 71 (30%)  88 (34%) 91 (31%)  
 II 149 87 (28%) 62 (32%)  62 (24%) 87 (30%)  
 III 222 120 (38%) 102 (30%)  110 (42%) 112 (39%)  
Growth pattern    0.215   0.481 
 Protruded type 267 158 (50%) 109 (46%)  127 (49%) 140 (36%)  
 Ulcerative type 283 157 (50%) 126 (54%)  133 (51%) 150 (64%)  
Number of AR (n = 550)Number of CAG repeat length
FactorNAR (−) (%)AR (+) (%)PShort (%)Long (%)P
Age, y    0.441   0.072 
 <60 277 160 (51%) 117 (50%)  140 (54%) 137 (47%)  
 ≥60 273 155 (49%) 118 (50%)  120 (46%) 153 (53%)  
Gender    0.290   0.499 
 Male 312 175 (56%) 137 (58%)  148 (57%) 164 (57%)  
 Female 238 140 (44%) 98 (42%)  112 (43%) 126 (43%)  
Largest tumor diameter    0.001   0.001 
 <5 cm 289 147 (47%) 142 (60%)  155 (60%) 169 (46%)  
 ≥5 cm 261 168 (53%) 92 (40%)  105 (40%) 121 (54%)  
Pathologic T stage    0.001   0.056 
 T1-2 248 123 (39%) 125 (53%)  127 (49%) 121 (42%)  
 T3-4 302 192 (61%) 110 (47%)  133 (51%) 169 (58%)  
Pathologic N stage    0.076   <0.001 
 N0 237 127 (40%) 110 (46%)  133 (51%) 104 (36%)  
 N1-2 313 188 (60%) 125 (54%)  127 (49%) 186 (64%)  
Tumor location    0.308   0.291 
 Rectum 301 169 (54%) 132 (65%)  146 (56%) 155 (53%)  
 Colon 249 146 (46%) 103 (35%)  114 (44%) 135 (47%)  
Differentiation    <0.001   0.028 
 Well 238 114 (36%) 124 (53%)  128 (49%) 110 (38%)  
 Moderate 229 146 (46%) 83 (35%)  96 (37%) 133 (46%)  
 Poor 83 55 (18%) 28 (12%)  36 (14%) 47 (16%)  
TNM stage    0.613   0.415 
 I 179 108 (34%) 71 (30%)  88 (34%) 91 (31%)  
 II 149 87 (28%) 62 (32%)  62 (24%) 87 (30%)  
 III 222 120 (38%) 102 (30%)  110 (42%) 112 (39%)  
Growth pattern    0.215   0.481 
 Protruded type 267 158 (50%) 109 (46%)  127 (49%) 140 (36%)  
 Ulcerative type 283 157 (50%) 126 (54%)  133 (51%) 150 (64%)  

NOTE: Bold text indicates P < 0.05, statistically significant.

Survival analysis of AR expression and CAG repeat length in colorectal cancer

We initially wanted to verify whether various CAG repeat lengths and AR expression would contribute to survival. We found that CAG repeat length (Fig. 2D and E) and AR expression (Fig. 3D and E) significantly affected the survival (5-year OS) in the test and validation cohorts. We also found that CAG repeat lengths (Fig. 2F–H) and AR expression (Fig. 3F–H) significantly affected the survival (5-year OS) in males, females, and total colorectal cancer population, respectively.

We analyzed the 5-year OS according to AR expression, CAG repeat length, and clinicopathologic parameters (Table 2) in the test cohort, validation cohort, and pooled samples. Univariate analysis revealed the associations of various factors, including largest tumor diameter (P = 0.005), pathologic T (pT) stage (P = 0.002), pathologic N (pN) stage (P = 0.019), TNM stage (P = 0.024), AR expression (P = 0.001), and CAG repeat length (P = 0.001), with colorectal cancer prognosis. Tumors with large diameters (≥5 cm), poor pT (T3-4)/pN (N1-2) stage, high TNM III stage or negative AR expression, or long CAG repeat lengths were indicators of poor colorectal cancer prognosis. No significant differences in gender, age, differentiation, tumor location, or growth pattern were observed (P > 0.05).

Table 2.

Univariate and multivariate analyses of prognosis factors (5-year OS) for patients with colorectal cancer

Test cohort (n = 290)Validation cohort (n = 260)Pooled sample (n = 550)
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
Univariate analysis 
 Age, y (<60/≥60)  0.368  0.975  0.935 
 Gender(male/female)  0.171  0.102  0.497 
 Largest tumor diameter (≥5 cm/<5 cm) 1.53 (1.05–2.22) 0.027 1.47 (1.01–2.13) 0.044 1.47 (1.12–1.92) 0.005 
 Pathologic T stage (T3-4/T1-2) 2.08 (1.37–3.17) 0.001 1.97 (1.31–2.98) 0.001 1.56 (1.17–2.07) 0.002 
 Pathologic N stage (N1-2/N0) 1.98 (1.34–2.93) 0.001  0.059 1.41 (1.06–1.87) 0.019 
 Tumor location (rectum/colon)  0.993  0.641  0.340 
 Differentiation (well-moderate/poor)  0.327  0.697  0.876 
 TNM stage (III/I+II)  0.082 1.82 (1.25–2.65) 0.002 1.36 (1.04–1.78) 0.024 
 Growth pattern (ulcerative/protruded)  0.993  0.611  0.480 
 AR (−/+) 2.20 (1.49–3.26) 0.001 1.70 (1.15–2.52) 0.008 1.64 (1.24–2.18) 0.001 
 CAG repeat (long/short) 1.63 (1.13–2.35) 0.009 1.60 (1.10–2.22) 0.014 1.57 (1.19–2.07) 0.001 
Multivariate analysis 
 Largest tumor diameter (≥5 cm/<5 cm) — — 1.46 (1.03–2.23) 0.038 1.37 (1.04–1.80) 0.023 
 Pathologic T stage (T3-4/T1-2) 1.81 (1.22–2.69) 0.003 — — 1.43 (1.08–1.91) 0.014 
 Pathologic N stage (N1-2/N0)  0.232  0.158  0.095 
 AR (−/+) 2.05 (1.38–3.04) 0.001 2.06 (1.27–3.35) 0.003 1.48 (1.11–1.98) 0.008 
 CAG repeat (long/short) 1.60 (1.07–2.39) 0.021 1.95 (1.17–3.24) 0.010 1.36 (1.04–1.79) 0.024 
Test cohort (n = 290)Validation cohort (n = 260)Pooled sample (n = 550)
VariablesHR (95% CI)PHR (95% CI)PHR (95% CI)P
Univariate analysis 
 Age, y (<60/≥60)  0.368  0.975  0.935 
 Gender(male/female)  0.171  0.102  0.497 
 Largest tumor diameter (≥5 cm/<5 cm) 1.53 (1.05–2.22) 0.027 1.47 (1.01–2.13) 0.044 1.47 (1.12–1.92) 0.005 
 Pathologic T stage (T3-4/T1-2) 2.08 (1.37–3.17) 0.001 1.97 (1.31–2.98) 0.001 1.56 (1.17–2.07) 0.002 
 Pathologic N stage (N1-2/N0) 1.98 (1.34–2.93) 0.001  0.059 1.41 (1.06–1.87) 0.019 
 Tumor location (rectum/colon)  0.993  0.641  0.340 
 Differentiation (well-moderate/poor)  0.327  0.697  0.876 
 TNM stage (III/I+II)  0.082 1.82 (1.25–2.65) 0.002 1.36 (1.04–1.78) 0.024 
 Growth pattern (ulcerative/protruded)  0.993  0.611  0.480 
 AR (−/+) 2.20 (1.49–3.26) 0.001 1.70 (1.15–2.52) 0.008 1.64 (1.24–2.18) 0.001 
 CAG repeat (long/short) 1.63 (1.13–2.35) 0.009 1.60 (1.10–2.22) 0.014 1.57 (1.19–2.07) 0.001 
Multivariate analysis 
 Largest tumor diameter (≥5 cm/<5 cm) — — 1.46 (1.03–2.23) 0.038 1.37 (1.04–1.80) 0.023 
 Pathologic T stage (T3-4/T1-2) 1.81 (1.22–2.69) 0.003 — — 1.43 (1.08–1.91) 0.014 
 Pathologic N stage (N1-2/N0)  0.232  0.158  0.095 
 AR (−/+) 2.05 (1.38–3.04) 0.001 2.06 (1.27–3.35) 0.003 1.48 (1.11–1.98) 0.008 
 CAG repeat (long/short) 1.60 (1.07–2.39) 0.021 1.95 (1.17–3.24) 0.010 1.36 (1.04–1.79) 0.024 

NOTE: Bold text indicates P < 0.05, statistically significant.

We further investigated these adverse factors in the Cox multivariate survival model. Multivariate analysis for 5-year OS (Table 2) showed that four variables, namely, large tumor diameter ≥5 cm (HR, 1.372; 95% CI, 1.044–1.803; P = 0.023), pT3-4 stage (HR, 1.434; 95% CI, 1.075–1.912; P = 0.014), negative AR expression (HR, 1.480; 95% CI, 1.105–1.982; P = 0.008), and long CAG repeat lengths (HR, 1.364; 95% CI, 1.042–1.785; P = 0.024), were significantly correlated with prognosis, whereas pN and TNM stages were not considered independent risk factors (P > 0.05).

Survival analysis of AR expression and CAG repeat length in adjuvant chemotherapy colorectal cancer

We analyzed 296 patients with stages IIb and III colorectal cancer who underwent postoperative adjuvant chemotherapy to improve risk stratification. Figure 4 showed the relationship between AR expression (Fig. 4A) as well as CAG repeat length (Fig. 4B), and recurrence or metastasis in 296 postoperative adjuvant chemotherapy patients. Figure 4A showed that 100 patients with negative AR expression (58.1%) and 56 patients with positive AR expression (45.2%) received postoperative adjuvant chemotherapy, and a statistical difference was observed in recurrence or metastasis in negative and positive AR expression (P = 0.018). Figure 4B showed that 84 patients with long CAG repeat sequences (51.4%) and 52 patients with short CAG repeat sequences (39.1%) received postoperative adjuvant chemotherapy, and a statistical difference was observed in recurrence or metastasis in long and short CAG repeat sequence (P = 0.022). Besides, we found that AR-negative subgroup (Fig. 4C) and long CAG repeat sequences (Fig. 4D) significantly affected the poor survival (5-year OS) in postoperative adjuvant chemotherapy patients with colorectal cancer (P = 0.002 and P = 0.004, respectively).

Figure 4.

Associations between AR expression and CAG repeat length with adjuvant chemotherapy patients with colorectal cancer. A, recurrence or metastasis is illustrated according to AR expression in 296 postoperative adjuvant chemotherapy patients. B, recurrence or metastasis is illustrated according to CAG repeat length in 296 postoperative adjuvant chemotherapy patients. C and D, the prognostic effect of AR expression and CAG repeat length are illustrated in 296 patients with colorectal cancer who received postoperative adjuvant chemotherapy.

Figure 4.

Associations between AR expression and CAG repeat length with adjuvant chemotherapy patients with colorectal cancer. A, recurrence or metastasis is illustrated according to AR expression in 296 postoperative adjuvant chemotherapy patients. B, recurrence or metastasis is illustrated according to CAG repeat length in 296 postoperative adjuvant chemotherapy patients. C and D, the prognostic effect of AR expression and CAG repeat length are illustrated in 296 patients with colorectal cancer who received postoperative adjuvant chemotherapy.

Close modal

Steroid hormone receptors may participate in the differentiation, proliferation, and progression of colorectal cancer tissues (3, 17, 18). Many studies have focused on estrogen and progesterone, which were believed to have an important function in the development of colorectal cancer (19–21). Rudolph and colleagues (22) found that ER expression is independently related to poor survival. However, the function of androgen and AR in colorectal cancer etiology is poorly understood. The function of AR in the colon and rectum carcinogenesis remains unclear. The significance of androgen in tumors was confirmed by different approaches of study (3). Recent studies suggested a protective function of androgens in the colon (23, 24). AR expressions have been detected in the colorectal mucosa of experimental animals and humans (25, 26). Positive AR expressions were observed in 42.7% of colorectal cancer foci and 81.3% of normal mucosa in our study; but this percent was different from that of two studies, which reported lower AR expression in neoplastic mucosa than that in normal colonic mucosa (17, 27). We found a drastic difference between normal and neoplastic mucosa when analyzing AR protein expression.

We hypothesized that polymorphic CAG repeat length can affect AR activity. First, previous studies showed that the CAG repeat length of AR inversely affects its transactivation potential, either as a directly altered receptor function (7) or indirectly reduced AR messenger in RNA and protein levels (9). Second, the transactivation function of AR is dependent on the ligands and androgens. The receptor function also participates in the control of cellular differentiation and proliferation in hormone-dependent tissues (4). For example, the androgen signal provides a mitogenic effect in the prostate, where decreasing CAG repeat length of the AR gene increases the cancer risk. By contrast, androgens in breast tissues function in an antimitogenic response, in which long CAG repeats indicated a higher risk for male (14) and female (15) patients with breast cancer. Finally, we observed that the colorectal cancer group had a longer CAG repeat length compared with the control group. We identified the association between the absence of AR expression and long CAG repeat sequence in our study. We studied 550 patients with colorectal cancer without somatic alterations in CAG repeats, and demonstrated that long CAG repeats of the AR genotype were associated with higher (1.293-fold in males, 1.255-fold in females, and 1.277-fold in total colorectal cancer population, respectively) risk for colorectal cancer in a Chinese population, whereas a short CAG repeat sequence may provide a certain degree of protection against colorectal cancer. Our study was different from that of Slattery and colleagues (13), who found that the AR genotype can decrease the risk for rectal cancer between patients and controls if the genotype possesses a short CAG repeat length. We also obtained a different cut-off point of CAG repeat length, which might attribute to different regions and ethnicities.

We tested the CAG repeat length of each patient with colorectal cancer in all tumor tissues and paired blood samples. We found that 90% (550 of 610) of colorectal cancer samples showed no somatic AR CAG repeat alterations. Our study was similar to that of Ferro and colleagues (10), who found 10% somatic alterations in CAG repeats within the AR gene in colon tumors. First of all, our study investigated the prognostic significance of AR (CAG)n repeat sequence. Our results indicate that patients with different CAG repeat statuses had significantly different survival rates. Therefore, a relatively long CAG repeat sequence was independently associated with poor survival in colorectal cancer.

Interestingly, we also found that the absence of AR expression was associated with long CAG repeat sequences that appeared in tumor tissues, and the ratio was 82.9% in males, 92.1% in females, and 86.9% in total patients, respectively. Univariate logistic regression with patients with colorectal cancer indicated the association of AR-negative expression with AR(CAG)n long repeat length in individual patients. The absence of AR expression and long CAG repeat sequences were observed in tumor with large diameter (≥5 cm), moderate/poor differentiation, T3-4 stage, and N1-2 stage compared with positive AR expression and short CAG repeat sequences in patients with colorectal cancer, but they were not correlated with other clinicopathologic features. Further research is necessary to determine the exact mechanism underlying the absence of AR expression and presence of long CAG repeat sequences, and their association with a shorter survival rate in patients with colorectal cancer. This phenomenon illustrates that CAG repeat lengths could affect the function of AR protein and alter the biometric traits of colorectal cancer. The results left us with a new question on whether or not we may start with polymorphisms to discover the regulatory mechanism of the transcription change of the AR gene, so that we can interfere with this regulatory mechanism to inhibit tumor development.

In this study, we investigated functional polymorphic CAG and protein expression of the AR gene in postoperative adjuvant chemotherapy patients with colorectal cancer. The absence of AR expression and long CAG repeat sequences were independently associated with poor survival of postoperative adjuvant chemotherapy patients with colorectal cancer. This study indicates that the negative AR subgroup and long CAG repeat sequences were insensitive and tolerant to adjuvant chemotherapy (FOLFOX and XELOX); therefore, these patients were prone to recurrence or metastasis and poor prognostic effect.

Although our study generates some important findings, it also has some limitations. Perhaps, there is variation in androgen levels in foods throughout the year. In addition, dietary data need to be obtained using a very detailed questionnaire. Besides, we selected cut-points for the AR gene based on the median data because there is limited information on the most appropriate cut-points to evaluate. Other study limits include that studies examining other polymorphisms and functionality of these genes are needed. Much still remains to be done and more accurate data need to be generated to yield more valuable findings in the future.

All in all, our study indicated that absence of AR expression and long CAG repeat sequence were closely related to the development of colorectal cancer. Both long CAG and decreased AR expression were correlated with the poor 5-year OS in patients with colorectal cancer.

No potential conflicts of interest were disclosed.

Conception and design: R. Huang, G. Wang, Y. Song, F. Wang, X. Wang

Development of methodology: G. Wang, Y. Song, B. Zhu, Q. Zhang

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): R. Huang, G. Wang, Y. Song, F. Wang, Y. Chen

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): Y. Song, F. Wang, B. Zhu, Q. Zhang, S. Muhammad

Writing, review, and/or revision of the manuscript: R. Huang, G. Wang, Y. Song, S. Muhammad

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases):Q. Tang, Z. Liu

The authors thank Ke-Fei Wu and Hong-Tao Song for assistance with immunohistochemistry staining analysis and Jing Song and Jiu-Feng Wei for technical assistance.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Haydon
AM
,
Jass
JR
. 
Emerging pathways in colorectal-cancer development
.
Lancet Oncol
2002
;
3
:
83
8
.
2.
Kinzler
KW
,
Vogelstein
B
. 
Lessons from hereditary colorectal cancer
.
Cell
1996
;
87
:
159
70
.
3.
Berta
L
,
Fronticelli
BC
,
Fazzari
A
,
Radice
E
,
Bargoni
A
,
Frairia
R
, et al
Sex steroid receptors, secondary bile acids and colorectal cancer. A possible mechanism of interaction
.
Panminerva Med
2003
;
45
:
261
6
.
4.
Ferro
P
,
Catalano
MG
,
Dell'Eva
R
,
Fortunati
N
,
Pfeffer
U
. 
The androgen receptor CAG repeat: a modifier of carcinogenesis?
Mol Cell Endocrinol
2002
;
193
:
109
20
.
5.
MacLean
HE
,
Warne
GL
,
Zajac
JD
. 
Localization of functional domains in the androgen receptor
.
J Steroid Biochem Mol Biol
1997
;
62
:
233
42
.
6.
La Spada
AR
,
Wilson
EM
,
Lubahn
DB
,
Harding
AE
,
Fischbeck
KH
. 
Androgen receptor gene mutations in X-linked spinal and bulbar muscular atrophy
.
Nature
1991
;
352
:
77
9
.
7.
Chamberlain
NL
,
Driver
ED
,
Miesfeld
RL
. 
The length and location of CAG trinucleotide repeats in the androgen receptor N-terminal domain affect transactivation function
.
Nucleic Acids Res
1994
;
22
:
3181
6
.
8.
Mhatre
AN
,
Trifiro
MA
,
Kaufman
M
,
Kazemi-Esfarjani
P
,
Figlewicz
D
,
Rouleau
G
, et al
Reduced transcriptional regulatory competence of the androgen receptor in X-linked spinal and bulbar muscular atrophy
.
Nat Genet
1993
;
5
:
184
8
.
9.
Choong
CS
,
Kemppainen
JA
,
Zhou
ZX
,
Wilson
EM
. 
Reduced androgen receptor gene expression with first exon CAG repeat expansion
.
Mol Endocrinol
1996
;
10
:
1527
35
.
10.
Ferro
P
,
Catalano
MG
,
Raineri
M
,
Reato
G
,
dell'Eva
R
,
Risio
M
, et al
Somatic alterations of the androgen receptor CAG repeat in human colon cancer delineate a novel mutation pathway independent of microsatellite instability
.
Cancer Genet Cytogenet
2000
;
123
:
35
40
.
11.
Benatti
P
,
Gafa
R
,
Barana
D
,
Marino
M
,
Scarselli
A
,
Pedroni
M
, et al
Microsatellite instability and colorectal cancer prognosis
.
Clin Cancer Res
2005
;
11
:
8332
40
.
12.
Edge
SB
,
Compton
CC
. 
The American Joint Committee on Cancer: the 7th edition of the AJCC cancer staging manual and the future of TNM
.
Ann Surg Oncol
2010
;
17
:
1471
4
.
13.
Slattery
ML
,
Sweeney
C
,
Murtaugh
M
,
Ma
KN
,
Caan
BJ
,
Potter
JD
, et al
Associations between vitamin D, vitamin D receptor gene and the androgen receptor gene with colon and rectal cancer
.
Int J Cancer
2006
;
118
:
3140
6
.
14.
Song
YN
,
Geng
JS
,
Liu
T
,
Zhong
ZB
,
Liu
Y
,
Xia
BS
, et al
Long CAG repeat sequence and protein expression of androgen receptor considered as prognostic indicators in male breast carcinoma
.
PLoS One
2012
;
7
:
e52271
.
15.
Kwiatkowska
E
,
Teresiak
M
,
Filas
V
,
Karczewska
A
,
Breborowicz
D
,
Mackiewicz
A
. 
BRCA2 mutations and androgen receptor expression as independent predictors of outcome of male breast cancer patients
.
Clin Cancer Res
2003
;
9
:
4452
9
.
16.
Kidwai
N
,
Gong
Y
,
Sun
X
,
Deshpande
CG
,
Yeldandi
AV
,
Rao
MS
, et al
Expression of androgen receptor and prostate-specific antigen in male breast carcinoma
.
Breast Cancer Res
2004
;
6
:
R18
23
.
17.
Castagnetta
L
,
Traina
A
,
Campisi
I
,
Calabro
M
,
Maratta
A
,
Saetta
A
, et al
Androgen receptor status in nontumoral and malignant human colorectal tissues
.
Ann N Y Acad Sci
2002
;
963
:
322
5
.
18.
D'Errico
I
,
Moschetta
A
. 
Nuclear receptors, intestinal architecture and colon cancer: an intriguing link
.
Cell Mol Life Sci
2008
;
65
:
1523
43
.
19.
Hartman
J
,
Gustafsson
JA
. 
Estrogen receptors in colorectal cancer: goalkeepers, strikers, or bystanders?
Cancer Prev Res (Phila)
2010
;
3
:
897
9
.
20.
Lin
JH
,
Manson
JE
,
Kraft
P
,
Cochrane
BB
,
Gunter
MJ
,
Chlebowski
RT
, et al
Estrogen and progesterone-related gene variants and colorectal cancer risk in women
.
BMC Med Genet
2011
;
12
:
78
.
21.
Jiang
H
,
Teng
R
,
Wang
Q
,
Zhang
X
,
Wang
H
,
Wang
Z
, et al
Transcriptional analysis of estrogen receptor alpha variant mRNAs in colorectal cancers and their matched normal colorectal tissues
.
J Steroid Biochem Mol Biol
2008
;
112
:
20
4
.
22.
Rudolph
A
,
Toth
C
,
Hoffmeister
M
,
Roth
W
,
Herpel
E
,
Jansen
L
, et al
Expression of oestrogen receptor beta and prognosis of colorectal cancer
.
Br J Cancer
2012
;
107
:
831
9
.
23.
Pereira
MA
,
Khoury
MD
. 
Prevention by chemopreventive agents of azoxymethane-induced foci of aberrant crypts in rat colon
.
Cancer Lett
1991
;
61
:
27
33
.
24.
Rao
CV
,
Tokumo
K
,
Rigotty
J
,
Zang
E
,
Kelloff
G
,
Reddy
BS
. 
Chemoprevention of colon carcinogenesis by dietary administration of piroxicam, alpha-difluoromethylornithine, 16 alpha-fluoro-5-androsten-17-one, and ellagic acid individually and in combination
.
Cancer Res
1991
;
51
:
4528
34
.
25.
Catalano
MG
,
Pfeffer
U
,
Raineri
M
,
Ferro
P
,
Curto
A
,
Capuzzi
P
, et al
Altered expression of androgen-receptor isoforms in human colon-cancer tissues
.
Int J Cancer
2000
;
86
:
325
30
.
26.
Waliszewski
P
,
Blaszczyk
M
,
Wolinska-Witort
E
,
Drews
M
,
Snochowski
M
,
Hurst
RE
. 
Molecular study of sex steroid receptor gene expression in human colon and in colorectal carcinomas
.
J Surg Oncol
1997
;
64
:
3
11
.
27.
Marugo
M
,
Aste
H
,
Bernasconi
D
,
Fazzuoli
L
,
Conio
M
,
Cuva
A
, et al
Cytosolic and nuclear androgen receptors in colorectal adenomas
.
Anticancer Res
1992
;
12
:
705
8
.