Prognostic impact of MGMT promoter methylation and MGMT and CD133 expression in colorectal adenocarcinoma

In this cross-sectional study (case-series design), colorrectal adenocarcinoma samples were obtained from patients at three hospitals in Southern Spain (Puerta del Mar Hospital, Cádiz; Puerto Real Hospital, Cádiz; and San Cecilio Hospital, Granada) between 2004 and 2009. Clinical data of the patients were obtained from the hospital records. Written informed consent was obtained from all patients and controls before their enrolment in the study. The study protocol was approved by the Biomedical Investigation Ethic Committee (Consejeria de Salud; Servicio Andaluz de Salud). Paraffin-embedded tumor specimens were obtained from 123 CRC patients. The differentiation grade and tumor stage were determined according to standard histopathological criteria by two expert pathologists [27]. DNA extraction and analysis, MGMT methylation status test, tissue microarray (TMA) construction, and MGMT and CD133 immunohistochemical analyses were performed in samples from each specimen. None of the patients had received any pre-operative treatment. After the tumor resection, most patients had been treated with chemotherapy (5-fluorouracil [5-FU], oxaliplatin, and/or irinotecan) and/or radiotherapy according to their clinical characteristics.

DNA was extracted from waxed tissue samples by using the Chemagic MSM I robot (Chemagen, Germany, Baesweiler) in accordance with the manufacturer’s standard recommendations. Determination of methylation patterns in MGMT promoter CpG islands was based on the chemical modification of unmethylated (but not methylated) cytosine to uracil. MSP was performed with specific primers for either methylated or unmethylated DNA, as previously described [10]. Briefly, a 2-μg DNA sample was denatured with sodium hydroxide, modified with sodium bisulfite, and then purified (EpiTect Bisulfite kit, Qiagen, USA, Maryland). Primer sequences were 5′-TTTGTGTTTTGATGTTTGTAGGTTTTTGT-3′ (forward primer) and 5′-AACTCCACACTCTTCCAAAAACAAAACA-3′ (reverse primer) for the unmethylated (UM) reaction and were 5′-TTTCGACGTTCTAGGTTTTCGC-3′ (forward primer) and 5′-GCACTCTTCCGAAAACGAAACG-3′ (reverse primer) for the methylated (M) reaction. PCR-amplified products were electrophoresed on 3% agarose gels, visualized by staining with ethidium bromide, and examined under UV illumination. A sample was classified as hypermethylated when the methylation amplification product alone was observed, partially methylated when both methylated and unmethylated amplification products were seen, and unmethylated when it showed unmethylated amplification products alone. For the statistical analysis, the hypermethylated and partially methylated samples were considered as the methylated (M) group and compared with the unmethylated (UM) group.

Formalin-fixed paraffin-embedded CRC tumor samples were used in the construction of TMAs. Briefly, four representative areas were selected from whole hematoxylin-eosin tissue sections of each adenocarcinoma specimen. Cores with diameter of 1 mm were placed 0.8 mm apart in a grid layout using a Manual Tissue Microarrayer (Beecher Instruments, Silver Spring, MD). The resulting tissue microarray blocks were cut into 5-μm sections with a microtome, placed on slides by the adhesive tape-transfer method (Instrumedics, Inc., Hackensack, NJ), and UV cross-linked. TMA dewaxing, rehydration, epitope recovery, and all staining procedures were performed at the same time with the DakoAutostainer EnVision™ FLEX kit (Dako, Barcelona, Spain) using antibodies against MGMT (1:50, Santa Cruz Biotechnology, Inc., Heidelberg, Germany) and CD133 (1:50, Miltenyi Biotec, Bergisch Gladbach, Germany). The antibodies were incubated with 3.3′-diaminobenzidine (DAB) substrate-chromogen, resulting in a brown-colored precipitate at the antigen site, and cell nuclei were visualized with hematoxylin (blue) counterstaining; nerve tissue was used as a positive control [28]. The readings were done by two experienced pathologists under light microscopy. In the most of specimens, there weren’t significant differences between observers and sample punches. Furthermore, the patients with heterogeneous staining for any antibody were not included in this study. The MGMT staining of tumor cells was scored and grouped as low expression (<50%) (-, +, and ++ scores) and high expression (≥50%) (+++ and ++++ scores). CD133 staining of tumor glands was classified as 0%, <50%, or ≥50%; CD133 staining on the apical and/or endoluminal surface of tumor glands and/or on cell debris was considered positive, in accordance with previous studies [29]. For the statistical analysis, CD133 expression was considered in two categories: low (<50%) or high (≥50%). The intensity of MGMT and CD133 staining was scored as low or high. Thereby, percentage only consider number of cells or tumor glands stained for MGMT and CD133 independently to dye distribution; moreover, intensity was scored high when nucleus or tumor glands were completely stained and low when showed partially or lack of staining regardless of number of cells or tumor glands stained.

Contingency tables and associations were analyzed with the chi-square (χ²) test and Fisher’s exact test. DFS (time elapsed between diagnosis and disease recurrence) and OS (time between diagnosis and death) curves were estimated with the Kaplan-Meier method. A two-sided log-rank test was used to determine significant differences between independent curves and patient groups. Significant prognostic factors associated with DFS and/or OS were identified by applying the Cox proportional hazards model, which was constructed using the most relevant molecular, histopathological, and treatment variables. SPSS version 15.0 was used for the data analyses; p < 0.05 was considered significant.

Table 1 summarizes the characteristics of the 123 patients in the study (65% males, 35% females); the mean (± standard deviation) age was 71.73 ± 10.57 yrs (range, 40 to 93 yrs). The tumor was in stage III in 40.2% (49/122) of patients and the differentiation grade was moderate in 50.4% (59/117). At the most recent follow up, 17.8% (18/101) of the patients had died due to the colorectal adenocarcinoma and 41.9% (44/105) of the patients did not respond to treatment, evidencing local or distal recurrence. The follow-up period ranged between 2 and 93 months. The mean OS of the whole sample was 40.20 ± 22.09 months and the mean DFS was 35.98 ± 24.75 months.

Table 2 summarizes the molecular characteristics of the patients. MGMT promoter methylation status could be determined in 89.4% (110/123) of the tumors; PCR amplification was unsuccessful or evaluation was not possible in the remaining 13 specimens. The M group included 78.2% of the 110 cases and the UM group 21.8%. Out of the M group, 75.6% showed partial methylation (amplification with UM and M primers) and 24.4% hypermethylation (amplification with M primer alone) (Figure 1).

MGMT staining was always nuclear in colorectal adenocarcinoma gland cells and always detected in surrounding tissue (Figure 2). Data on the percentage MGMT expression were available for 92.7% (114/123) of the patients. Out of these 114 cases, no expression was observed in 15.8%, and expression was scored as + in 2.6%, ++ in 29.8%, +++ in 43.9%, and ++++ in 7.9% (Figure 2); hence, low (<50%) MGMT expression was observed in 48.2% of cases and high expression (≥50%) in 51.8%. No association was found between percentage MGMT expression and MGMT promoter methylation status (Additional file 1: Table S1). A low intensity of MGMT expression was observed in 26.3% of the 114 patients and a high intensity in 73.7% (Figure 2). A significant association was found between low MGMT expression intensity and MGMT promoter methylated (M) (Additional file 1: Table S1). An association was observed between poor-tumor differentiation grade and methylated MGMT promoter (M) (Additional file 2: Table S2).

The mean OS was 61.36 months in patients with low-intensity MGMT expression versus 77.48 months in those with high-intensity MGMT expression (Table 3); the correlation between OS and MGMT expression intensity was significant (p < 0.01) (Figure 3). MGMT expression intensity was a prognostic factor for OS after adjusting for treatment and histopathology variables (Table 4). No significant correlation was found between OS and MGMT promoter methylation status or percentage MGMT expression (Table 3). No significant correlation was observed between DFS and MGMT methylation status, MGMT expression intensity, or percentage MGMT expression (Table 3).

CD133 expression results were available for 110 (89.4%) of the patients (Table 2). CD133 expression was detected on the endoluminal surface of tumor glands and on cell debris; no staining was observed in other tumor regions or in normal tissues. Out of the 110 specimens analyzed, the staining was scored as 0% in 12.7%, <50% in 34.6%, and ≥50% in 52.7% (Figure 4). In addition, 36.4% of the 110 patients had a high-intensity CD133 expression and 63.6% a low-intensity expression (Figure 4). An association was observed between no treatment response and high CD133 percentage expression (Additional file 3: Table S3).

OS was not significantly associated with CD133 protein expression intensity or percentage (Table 3). However, a statistically significant correlation (p < 0.05) was observed between DFS and percentage CD133 expression (Figure 3), with a mean DFS of 67.91 months in those with low (<50%) expression versus 46.01 months in those with higher (≥50%) CD133 expression (Table 3). The tumor stage was also correlated with DFS (Table 3). The multivariable analysis results showed that CD133 protein expression was not an independent prognostic factor (Table 4).

No significant association was found between MGMT promoter methylation or MGMT expression percentage/intensity and CD133 expression percentage/intensity (Additional file 1: Table S1). CD133 expression percentage and MGMT intensity could be compared in 109 patients. Among the patients with low MGMT intensity, CD133 expression percentage was low in 42.9% and high in 57.1%. Among the patients with high MGMT intensity, CD133 expression percentage was low in 48.1% and high in 51.9%.

Study of the potential relationship of MGMT intensity and percentage CD133 expression with clinical outcome variables (Figure 3) revealed a significant correlation with OS but no significant correlation with DFS (Table 3). The patients with low-intensity MGMT expression and high (≥50%) CD133 expression had the worst OS (52.36 months) and DFS (37.85 months) outcomes (Table 3).