Prognostic value of CD133 expression in stage I lung adenocarcinomas

Primary lung cancer

All 177 cases examined were patients with stage I ADC that underwent radical surgical resection at Kanagawa Cardiovascular and Respiratory Center (Yokohama, Japan) between January 2001 and December 2006. Tumor stage was determined according to the international TNM classification system (seventh edition of UICC)[18].

The median age was 68 year-old (range 45-85), 89 patients (50.3%) were men and 88 (49.7%) were women. Eighty-five patients (48.0%) had a smoking history (Brinkman Index, median 800, range 10-3200) and 92 (52.0%) were non-smokers. One-hundred-and-thirty-one patients (74.0%) were stage IA and 46 (26.0%) were stage IB. Lobectomy and segmentectomy were performed on 157 and 7 patients, respectively, which along with systemic lymphadenectomy, extended to the hilar and mediastinal lymph nodes. Thirteen patients (7.3%) underwent wedge resection along with intra-operative lymph node sampling. A follow-up evaluation was performed every 2 months for the first 2 years after the operation, every 3 months in the third year, and every 6 months thereafter. The evaluation included physical examinations and chest radiography. Screening for serum tumor markers, computed tomography (CT) of the thorax and upper abdomen, and magnetic resonance imaging (MRI) of the brain were obtained every 6 months for the first 3 years, and every 12 months thereafter. The median follow-up period was 35.9 months (range 1.1-82.5 months). Twelve patients (6.8%) died during the follow-up period, 7 died of lung cancer causes (range 10.4-45.1 months) and 5 died of non-lung cancer causes (range 1.1-12.9 months). The five-year overall survival (OS) rate was 91.5%. Disease recurrence was found in 18 out of 177 patients (10.2%), of whom 15 (8.5%) were affected by metastasis. The median disease-free span of patients with recurrence was 12.0 months (range 3.8-49.1 months). None of the patients received radiotherapy or chemotherapy preoperatively. None of patients with stage IA disease, and 18 of 46 patients with stage IB disease, received postoperative adjuvant chemotherapy (3 patients received cis-platin or carboplatin-based chemotherapy, and 15 received oral uracil-tegafur (UFT) chemotherapy). The 5-year disease-free survival (DFS) rate was 75.2% for the non-adjuvant IB patients vs. 75.0% for adjuvant IB patients (p=0.983; 4 patients who could not continue oral UFT treatment for more than 6 months were excluded). Informed consent for research use of the resected materials was obtained from all the subjects.

Immunohistochemistry

The largest tumor sections (4 um thick) were cut from formalin-fixed, paraffin-embedded tissue blocks. The sections were subjected to immunohistochemistry within 48 hours after being cut out to avoid loss of antigenicity [19-20]. The sections were deparaffinized, rehydrated, and incubated with 3% hydrogen peroxide, followed by 5% goat serum to block endogenous peroxidase activity and non-immunospecific protein binding. The sections were boiled in citrate buffer (0.01 M, pH 6.0) for 15 minutes to retrieve masked epitopes and then incubated with a primary antibody against either CD133 (working concentration of 1.0 μg/ml, clone W6B3C1 (IgG1), Miltenyi Biotec, Bergisch-Gladbach, Germany) or Ki-67 (working concentration of 1.0 μg/ml, clone MIB1 (IgG1), DAKO, Ely, UK) for 90 minutes at room temperature. Immunoreactivity was visualized using an Envision detection system (DAKO), and the nuclei were counterstained with hematoxylin and eosin (HE). To verify the specificity of the immunoreac-tivities, negative control experiments using mouse non-immunospecific immunoglobulin (IgG1, DAKO) to replace the primary antibodies were performed.

Evaluation of other biological properties

Proliferating activity of neoplastic cells was evaluated by a Ki-67 labeling index, which was calculated as the proportion of positive cells by counting 500–1000 neoplastic cells in the largest tumor section. The Ki-67 labeling indexes of <10%and ≥10%were classified as low and high levels, according to the criteria of previous studies [21-22]. The vessel invasion was examined on the largest tumor section stained with Victoria blue-van Gieson's method.

Statistical analysis

The post-operative disease-free span was defined as the period ranging from the date of surgery to the date when the recurrence of disease was diagnosed. An observation was censored at the last follow-up if the patient was alive or had died of a cause other than lung cancer. The possible associations between disease recurrence and various parameters were analyzed by Chi-square test and Fisher's exact test. The recurrence curves were plotted using the Kaplan-Meier method, and the absolute risk of recurrence at five years was estimated from the curves. The differences in disease-free survival (DFS) span and rate were analyzed using the Log-rank test. Multivariate analyses were performed with adjustment for possible prognostic variables (CD133 expression, Ki-67 expression, vessel invasion, age, gender, smoking status, extent of operation). Hazard ratios and 95% CIs were calculated using Cox's proportional hazards model. p values less than 0.05 were considered significant. All statistical analyses were performed using SPSS software (SPSS for Windows Version 10.0; SPSS; Chicago, IL, USA).

CD133 expression in normal and malignant lung epithelium

Immunohistochemical expression of CD133 was found not only in neoplastic cells, but also in non-neoplastic cells. Non-neoplastic bronchiolar epithelial cells expressed CD133 with exclusive plasma membrane localization at the apical and lateral sides (Figure 1A). Neoplastic cells expressed it at different levels and with various intracellular localizations. Some showed only a faint level with only membranous localization (Figure 1B), while others showed a level equivalent to that of bronchial epithelial cells (Figure 1C) or unequivocally strong levels with diffuse cytoplasmic localization (Figure 1D). Consistently, similar results for cytoplasmic expression of CD133 have been found in other diverse types of malignancies, such as gastric cancer [23], colon cancer [24-26], hepatocellular carcinoma [27], cholangiocellluar carcinoma [28], breast cancer [29], and ovarian cancer [30], including NSCLC [31-32]. Some studies reported that the diffuse cytoplasmic expression of CD133 could be a poor prognostic maker in diverse types of carcinomas [27-28]. According to observations in previous studies, unequivocally greater signal intensity with diffuse cytoplasmic staining than that of non-neoplastic bronchiolar epithelial cells was judged to be a strong level (Figure 1D). The CD133 expression score was defined as the proportion of cells with strong expression levels in the largest tumor section. The median CD133 expression score was 20.0%, ranging from 0.0 to 100.0%. A cutoff value was calculated from the receiver operating characteristic (ROC) curves for postoperative recurrence, and a score of <17.5% and ≥17.5% were classed as low and high, respectively (area under the curve 0.667, 95% confidence interval (CI) 0.556-0.775). Eighty-one tumors (45.8%) had CD133 high scores (high expressers), and 96 (54.2%) had CD133 low scores (low expressers) (Table 1).

Correlation between CD133 expression and disease recurrence

CD133 high expressers showed a significantly higher crude recurrence rate (17.3% vs. 4.2%, p=0.004) (Table 1), and also worse DFS than low CD133 expressers (5-year DFS rate 77.2% vs. 95.1%, p=0.004) (Table 2 and Figure 2A).

Correlation between Other biological properties and disease recurrence

The median value of the Ki-67 labeling index was 7.3% (range 0.3-80.7%). 71 patients (40.1%) were high-expressers and 106 (59.9%) were low-expressers (Table 1). As for vessel invasion, 49 patients (27.7%) were positive, and 128 (72.3%) were negative. Both the proliferating activity measured by the Ki-67 level and vessel invasion were associated with a significantly higher crude recurrence rate (Table 1). Consistently, these biological properties were also associated with poorer DFS (Table 2). These findings supported previous observations [22,33-34].

Prognostic value of a combination of CD133 expression, proliferating activity (Ki-67 expression) and/or vessel invasion

Forty-three cases (24.3%) were CD133 high/Ki-67 high expressers and 134 (75.7%) were the other types (Table 1). The CD133 high/Ki-67 high expressers showed a significantly higher crude recurrence rate (27.9% vs. 4.5%, p<0.001) and also worse DFS compared to the other types (5-year DFS rate 66.5% vs. 93.2%, p<0.001) (Table 2 and Figure 2B).

Twenty-seven cases (15.3%) were CD133 high expressers with vessel invasion and 150 (84.7%) were the other types (Table 1). CD133 high expressers with vessel invasion showed a significantly higher crude recurrence rate (33.3% vs. 6.0%, p<0.001) (Table 1) and also worse DFS in comparison with the other types (5-year DFS rate 51.0% vs. 92.9%, p<0.001) (Table 2 and Figure 2C).

Twenty-two cases (12.4%) were CD133 high/Ki-67 high expressers with vessel invasion and 155 (87.6%) were the other types (Table 1). This group showed a significantly higher crude recurrence rate (36.4% vs. 6.5%, p<0.001) (Table 1) and also worse DFS than the other types (5-year DFS rate 53.9% vs. 91.2%, p<0.001) (Table 2 and Figure 2D).

Multivariate analysis

Multivariate analyses revealed that CD133 high expression (HR 4.37, 95% CI 1.30-14.71, p=0.017), Ki-67 high expression (HR 5.89, 95% CI 1.47-23.65, p=0.012), CD133 high/Ki-67 high expression (HR 8.39, 95% CI 2.65-26.54, p<0.001), CD133 high expression with vessel invasion (HR 4.50, 95% CI 1.51-13.34, p=0.007), and CD133 high/Ki-67 high expression with vessel invasion (HR 9.32, 95% CI 3.42-25.39, p<0.001) had independent prognostic value to predict disease recurrence (Table 3).