SEARCH

SEARCH BY CITATION

Keywords:

  • Circulating endothelial progenitor;
  • colon cancer;
  • CD133 messenger RNA;
  • real-time reverse transcriptase-polymerase chain reaction

Abstract

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

BACKGROUND.

CD133 is a specific surface marker for bone marrow-derived circulating endothelial progenitors, which are vital in postnatal physiologic and pathologic (eg, tumor) angiogenesis. In this study, the authors examined whether increased levels of expression of CD133 messenger RNA (mRNA) in peripheral blood predicted disease recurrence in patients with colon cancer.

METHODS.

Semiquantitative real-time reverse transcriptase-polymerase chain reaction analysis was used to quantify CD133 mRNA levels in peripheral blood mononuclear cells from patients with colon cancer. The assay was developed first and tested at laboratory A (n = 34) and then was validated independently at laboratory B (n = 66). All patients were enrolled between February 2002 and December 2003. A central statistician performed the analysis.

RESULTS.

At laboratory A, the median CD133 mRNA level was elevated in patients with recurrent disease (4.2; range, 0.017–106.9) compared with patients without recurrence (0.0017; range, 0.0–9.51; P < .001), leading to a 14.6 odds ratio of recurrence (95% confidence interval [95% CI], 1.7–126; P = .004). At laboratory B, it was confirmed that elevated CD133 mRNA levels at a cutoff point ≥4.79 versus <4.79 were associated with an odds ratio of 22.6 for recurrence (95% CI, 1.7–291.2; P = .02). By comparison, the odds ratio for recurrence was 17.2 (95% CI, 1.8–164; P = .01) for patients with stage III–IV disease versus stage I–II disease according to the Tumor, Lymph Node, Metastasis (TNM) classification. An association also was observed between elevated carcinoma embryonic antigen levels (P = .03; 1-sided) and decreased survival (P = .035; 1-sided) with a CD133 mRNA cutoff level of ≥4.79.

CONCLUSIONS.

Elevated CD133 mRNA levels at ≥4.79 predicted colon cancer recurrence independent of TNM stage IV disease. Larger prospective studies comparing the current assay with standardized methodology are warranted. Cancer 2007. © 2007 American Cancer Society.

Angiogenesis, 1 of the 6 hallmarks of cancer,1 occurs through the sprouting of preexisting blood vessels and the mobilization, recruitment, and proliferation of bone marrow-derived endothelial progenitor cells (CEPs).2–6 Normally, the endothelium is highly quiescent unless physiologic or pathologic injury signals trigger a tissue-repair process. Malignancy represents a form of pathologic injury wherein incessant tumor growth leads to unchecked neovascularization.1–3

Bone marrow-derived CEPs are mobilized by cytokines2, 4 and are directed by chemokines to the sites of hypoxia and neovascularization.3, 5 CEPs are capable of expanding >1000-fold and initiate the premetastatic niche.6, 7 The stimulation of CEPs promotes tumor angiogenesis and tumor progression,3–5 whereas the inhibition of CEPs retarded tumor angiogenesis and tumor growth.3, 8 Elevated levels of CEPs (or their marker, CD133 messenger RNA [mRNA]) were noted in patients with cancers of the lung,9 breast (by CD146 mRNA),10 colorectum,11–13 and prostate or cancers with bone metastasis,13 myelodysplastic disorders,14, 15 multiple myeloma,15, 16 and infantile hemangioma.17 Chemotherapy at the maximum tolerable dose (MTD) mobilized viable CEPs18, 19; conversely, targeted antiangiogenic agents12, 20 or low-dose metronomic chemotherapy19, 21 decreased CEP levels. Compared with a rate as high as 90% of CEPs found in xenograft models,3 the rate of CEPs was from 1% to 12% in human cancers.22 Given the proliferation and differentiation potential of CEPs, the contribution of CEPs in postnatal tumor angiogenesis may be underestimated by evaluating the percentage of CEPs in limited tumor sections.

To measure CEP levels in peripheral blood mononuclear cells (PBMCs), the commonly used methods are fluorescence-activated cell sorting, colony assay, or magnetic beads; however, these methods are cumbersome, expensive, labor intensive, and time consuming.12, 18, 23, 24 The CD133 marker specifically identifies CEPs that differ from mature or activated circulating endothelial cells (CECs).11–13, 22, 25–28 We hypothesized that increased CD133 mRNA expression in PBMCs would correlate with or predict cancer recurrence. A semiquantitative, real-time reverse transcriptase-polymerase chain reaction (RT-PCR) technique was developed to quantify the CD133 mRNA levels in PBMCs and was tested initially in patients with colon cancer.

MATERIALS AND METHODS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Patient Characteristics

The Institutional Review Board of the University of Texas M. D. Anderson Cancer Center approved this laboratory study. Between February 2002 and December 2003, 73 patients with newly diagnosed or recurrent colon cancer provided written informed consent to have a 30-mL peripheral blood sample taken during their initial visit or at any follow-up visit at the gastrointestinal medical oncology clinic. In addition, peripheral blood samples from 2 healthy volunteers and 2 cytokine-mobilized peripheral blood samples from the aphaeresis unit at the University of Texas M. D. Anderson Cancer Center were used as negative and positive controls, respectively. Eligibility criteria included the absence of any cardiovascular disease and surgery performed within 4 weeks before the blood sample was drawn.

Data on the following variables were obtained from the patients' medical records: sex, age, date of diagnosis, tumor stage, differentiation, sites of metastasis, dietary history, smoking and drinking history, family history, comorbid conditions (such as diabetes), surgery, time and location of recurrence, carcinoembryonic antigen (CEA) levels, lactate dehydrogenase levels, and survival. A highly trained research associate who was blinded to the CD133 mRNA levels coded these clinical data and entered them into a database. Seven samples were excluded from the study because of poor RNA quality, loss of the sample, or patient withdrawal of consent. The clinical characteristics of the remaining 66 patients are summarized in Table 1.

Table 1. Patient Characteristics
VariableColon cancer patients (N = 66)
No.%
  1. SD indicates standard deviation.

Sex
 Men3857.6
 Women2842.4
Age: Mean age ± SD, y61.3 ± 13.5 
Race
 White5177.3
 Nonwhite1522.7
Baseline tumor pathology
 Well differentiated46.1
 Moderately differentiated5583.3
 Poorly differentiated710.6
Baseline cancer stage
 I–II1725.8
 III3146.9
 IV1827.3
Stage at blood draw
 I–II1015
 III2132
 IV3553
Smoking
 No3248.5
 Yes3451.5
Drinking
 No3350
 Yes3350
Family history of colorectal cancer
 No4669.7
 Yes1725.8
Diabetes
 No5887.9
 Yes812.1

Blood Collection and Isolation of mRNA

Peripheral blood samples were collected in an ethylene diamine tetraacetic acid tube, and PBMCs were isolated using Ficoll (Amersham-Pharmacia, Piscataway, NJ). The buffy coat was then aspirated using a pipette, and the PBMCs were enumerated and immediately lysed for RNA extraction using a commercial kit (Stratagene, La Jolla, Calif) according to the manufacturer's instructions. RNA was quantified spectrophotometrically, and its quality was checked by electrophoresis through denaturing agarose gels. Only degraded samples that demonstrated clear 18S and 28S bands under ultraviolet light were used for RT-PCR (Invitrogen, Carlsbad, Calif).

RT-PCR

All reagents used for RT-PCR were purchased from Applied Biosystems (Foster City, Calif). RT-PCR was performed in a total volume of 50 μL that contained 1 × TaqMan buffer with 5.5 nM MgCl2; 200 nM each of deoxyadenosine triphosphate (dATP), deoxycyitidine triphosphate (dCTP), and deoxyguanosine triphosphate (dGTP); 400 nM deoxyuridine triphosphate (dUTP); 300 nM of each primer, 100 nM probe; 0.5 units of uracil-N-glycosylase (AmpErase); 1.25 units of AmpliTaq gold; 10 μL of combinational DNA (cDNA); 5.5 nm MgCl2; 1 mM deoxynucleotides; 2.4 μM random hexamers; 20 units of RNase inhibitor; and 62.5 units of MulV reverse transcriptase. CD133 amplification was performed with β-actin or glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the internal control. The thermal cycle included 2 minutes at 50°C and 10 minutes at 95°C, followed by 40 cycles at 95°C for 15 seconds and at 60°C for 1 minute. The expected PCR product was 670 base pairs. Positive control samples were from the cytokine-mobilized peripheral blood samples. The primer sequences for CD133 for RT-PCR were as follows: forward, 5′-TGT ACGAATTCGACAGCTACTTGGCTCAGAC-3′; reverse, 5′-CTAGCTCGAGCATGATCTTTATGATAACC-3′.

Semiquantitative Real-time RT-PCR

The semiquantitative real-time RT-PCR29 assay was based on TaqMan methodology using the ABI PRISM 7700 sequence detection system (Applied Biosystems). Through fluorescence emission, this technique allows us to find the cycling point when the PCR product is detectable (Fig. 1). The Ct value correlates with the starting quantity of target mRNA. To normalize the amount of total RNA present in each reaction, we amplified the housekeeping gene (GAPDH). The amount of target, normalized to an endogenous reference (GAPDH) and relative to the calibrator, is defined by the ΔΔCt method, as described elsewhere.29 The formula is applied as follows: target amount = 2−ΔΔCt where ΔΔCt = {[Ct CD133 sample] − Ct(GAPDH sample)] − [Ct(CD1333 calibrator) − Ct(GAPDH calibrator)]}.

thumbnail image

Figure 1. Real-time reverse transcriptase-polymerase chain reaction cycle-threshold curve.

Download figure to PowerPoint

RT-PCR was performed in a total volume of 50 μL that contained 25 μL of 1 × SyberGreen mix buffer; 1 μL of Rox reference dye; 6 mM MgCl2; 100 mM KCl; 400 μL each of dATP, dCTP, dGTP, and dUTP; 40 mM Tris-HCl; 300 nM of each primer; and 2 μL of cDNA template. The thermal cycle conditions were at 95°C for 10 minutes, then 45 cycles at 95°C for 45 seconds, at 55°C for 1 minute, and at 72°C for 45 seconds.29 The forward primers for CD133 was 5′-AGCCTTCATCCACAGATGCT-3′; and reverse primer for CD133 was 5′GTGCATTT-CTCCA-CATTT-3′.The forward primer for GAPDH was 5′-CTTCACCACCAT-GGAGAAGGC-3′, and the reverse primer for GAPDH mRNA was 5′-ATGGACTG-TGGTCATGAG-3′. The primers were designed based on the CD133 gene sequence (Gene-Bank no. AF027208).

Statistical Analysis

The STATA software program (STATA Statistical Software, College Station, Tex) was used for data entry, management, and statistical analyses. Patients were classified according to their recurrence status into 2 groups: recurrence and no recurrence. The chi-square test was used to compare the distribution of categoric variables between patient groups. Biochemical values were expressed as the medians for each group. A nonparametric Mann-Whitney test was performed to calculate the mean value of measured parameters within both groups. In addition, the CD133 cutoff values were determined using the 90th percentile value among patients with no recurrence. We performed univariate, single-factor, unconditional logistic regression analyses to assess the marginal effects of each factor, including age, sex, race, histology, stage, CEA level, smoking, and diabetes, on the risk of colon cancer recurrence, with parameters estimated using maximum-likelihood analysis. We also performed multivariate, unconditional logistic regression analyses that included all variables that had significance at the .05 level in the single-factor analyses. For each factor, we reported the adjusted odds ratio (OR) and 95% confidence interval (95% CI), which we estimated by using maximum-likelihood analysis. The correlation between CEA level and survival was based on Kaplan-Meier estimates of CD133 mRNA levels at a defined cutoff point.

RESULTS

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Colon Cancer Recurrence

Thirty-seven of 66 patients with colon cancer had experienced recurrence at a median follow-up of 36 months. The clinical characteristics of patients with and without recurrence (including tumor differentiation, CEA level, stage, and CD133 mRNA expression levels) were compared. There were no statistical differences among the variables examined among the patients with and without recurrences, except for staging at blood draw (Table 2).

Table 2. Variables That Predicted Tumor Recurrence in Patients With Colon Cancer
VariableNo recurrence (N = 29)Recurrence (N = 37)P
No. of patients%No. of patients%
  1. SD indicates standard deviation; CEA, carcinoembryonic antigen.

Age at first diagnosis: Mean ± SD, y60.3 ± 14.4 62.1 ± 12.9 .6
Sex
 Men1655.22259.5.5
 Women1344.81540.5 
CD133    .4
 Mean ± SD174 ±864.92368.2 ±13,680 
 Median (range).007(0–4656).46(0–83,241) 
CEA: Mean ± SD34.5 ± 103.1 246.4 ± 1037.1 .3
Stage at blood draw
 I–II827.625.4.001
 III1344.8821.6 
 IV827.62773 
Tumor pathology
 Well/moderately differentiated2689.73286.5.5
 Poorly differentiated310.3410.8 
Smoking
 No1655.21643.2.3
 Yes1338.22156.8 
Alcohol
 No1655.21745.9.3
 Yes1344.82054.1 
Family history of colorectal cancer
 No2071.42674.3.5
 Yes847.1925.7 
History of diabetes mellitus
 No2793.13183.8.2
 Yes26.9616.2 

Elevated CD133 mRNA by RT-PCR in Colon Cancer Patients

CD133-positive CEPs were present in 7% to 11% of the cytokine-mobilized PBMCs and were used as positive control samples for all experiments. RT-PCR revealed that CD133 mRNA expression was present in PBMCs from the patients with colon cancer and from positive controls, whereas CD133 mRNA expression was not present in PBMCs from the 2 healthy volunteers using either β-actin mRNA (Fig. 2A) or GAPDH mRNA (Fig. 2B) as an internal control.

thumbnail image

Figure 2. Real-time reverse transcriptase-polymerase chain reaction of CD133 with β-actin (A) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) (B) as internal controls.

Download figure to PowerPoint

Elevated CD133 mRNA Levels Predict Colon Cancer Recurrence

A semiquantitative RT-PCR assay initially was developed and tested in laboratory A in duplicate on 2 different occasions in 34 initially enrolled patients. We observed that the median CD133 mRNA level was significantly higher in patients who had recurrent disease (4.2; range, 0.017–106.9) compared with the level in patients without recurrent disease (0.0017; range, 0–9.51; P < .001). Elevated CD133 mRNA levels conferred an OR of 14.6 for recurrence (95% CI, 1.7–126; P = .004) in this small study. The area under the receiving operating characteristic (ROC) curve was 0.81 (Fig. 3). Given the relatively wide range of CD133 mRNA levels, we attempted to validate the assay in laboratory B using the same methodology in an expanded cohort of 66 patients. We observed that a CD133 mRNA cutoff value <4.79 would select 90% of patients without recurrence. This cutoff value was similar to the median elevated CD133 mRNA level that was associated with recurrent disease in laboratory A. Among patients without recurrence, 93% had a CD133 mRNA levels <4.79, whereas 7% had a CD133 mRNA levels ≥4.79 (P = .029). Among patients who had CD133 mRNA levels ≥4.79, 85% had experienced recurrence compared with 15% of the patient who had no recurrence (P = .03). Using multivariate logistic regression analysis (Table 3), we calculated an OR of 22.2 (95% CI, 1.7–291.2; P = .02) among patients who had CD133 mRNA levels ≥4.79 compared with patients who had CD144 mRNA levels <4.79. By comparison, stage III–IV disease conferred an OR of 17.2 (95% CI, 1.8–164.1; P = .02) compared with stage I-II disease.

thumbnail image

Figure 3. Area under the receiving operating characteristics (ROC) curve = 0.81.

Download figure to PowerPoint

Table 3. Variables That Predicted Tumor Recurrence in Patients With Colon Cancer
VariableOR (95% CI)P value
  • OR indicates odds ratio; 95% CI, 95% confidence interval; SD, standard deviation.

  • *

    The cut-off level was the 90th percentile value among patients with no disease recurrence.

Age at first diagnosis1.01 (0.9–1.1).5
Sex
 Women1 (Reference) 
 Men0.9 (0.2–3.6).9
Race
 Nonwhite1 (Reference) 
 White2.5 (0.5–11.9).2
CD133*
 <4.791 (Reference) 
 ≥4.7922.2 (1.7–291.2).02
CEA
 ≤3 ng/ml1 (Reference) 
 >3 ng/ml5 (1.8–21).03
Stage
 I–II1 (Reference) 
 III–IV17.2 (1.8–164.1).01
Tumor pathology
 Well/moderately differentiated1 (Reference) 
 Poorly differentiated0.6 (0.1–4.6).6
Diabetes
 No1 (Reference) 
 Yes4.3 (0.4–47.4).2
Smoking
 No1 (Reference) 
 Yes2.4 (0.6–9.1).2

Elevated CD133 mRNA Levels Correlate With CEA and Overall Survival

CEA > 3 mg/ml was associated with recurrence COR 5.0 (1.5–21). We observed a positive correlation between the levels of CEA (ng/ml) and CD 133 mRNA. The mean values (±standard error) of CEA were 515.3 ± 504.9 and 62.6 ± 18.2 for patients with CD133 mRNA levels ≥4.79 and <4.79, respectively (P = .03; 1-sided). Although only 23 patients had died of cancer-related causes at a median follow-up of 36 months, there was a trend toward decreased median overall survival for those who had CD133 mRNA levels ≥4.79 compared with those who had CD133 mRNA levels <4.79 (P = .035; 1-sided) (Fig. 4).

thumbnail image

Figure 4. Kaplan-Meier cancer-specific survival curves by CD133 messenger RNA level (<4.79 vs ≥4.79).

Download figure to PowerPoint

DISCUSSION

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

Elevated CD133 mRNA levels ≥4.79 measured by real-time RT-PCR predicted colon cancer recurrence and were correlated in a trend with elevated CEA levels and decreased overall survival. These findings mostly are consistent with findings by Mehra et al., who measured CD133 mRNA levels with nucleic acid sequence-based amplification (NASBA), which is an isothermic single-stranded gene-amplification assay.15 In that study, increased CD133 mRNA at a cutoff point of 200 copies per 10,000 U1A (genome DNA) were noted in cancer patients with metastatic disease (P = .033), especially in those with bone metastasis (P < .001), and were correlated with decreased survival (P < .01) and prostate-specific antigen levels.13 The current study did not test the lowest detection limits, but its output readings were comparable to the lowest limit of detection by NASBA (18 cells per 105 PBMCs) and by flow cytometry (12–15 cells per 105 PBMCs).13, 23, 24 An ongoing study is attempting to establish a standard curve between CD133 mRNA levels and the number of CEPs by flow cytometry. Currently, we could not estimate the number of CEPs based on the cutoff value of CD133 mRNA ≥4.79. The current assay appeared to be reproducible, sensitive, and suitable for large-scale clinical testing similar to NASBA.

Elevated CEP levels were measured not only with malignancies9–17 but also during cytotoxic chemotherapy at the MTD,18 menses,26 surgery,6, 31 exercise,30 short-term statin use,32 and smoking cess ation.33 In contrast, decreased CEPs were measured during antiangiogenic therapy12, 20 or low-dose metronomic chemotherapy,19, 21 advancing age,33 diabetes,35 smoking,36 and long-term statin use.36 Two large prospective studies have established the independent prognostic values of CEP levels related inversely to cardiovascular events.23, 24 In contrast, 3 exploratory studies13, 38, 39 in cancer patients failed to demonstrate an association of elevated CD133 mRNA levels, but elevated Tie-2 mRNA levels,37 or VE-cadherin mRNA levels versus healthy volunteers.39 Mehra et al. did report elevated CD133 mRNA levels in cancer patients with metastasis, particularly those with bone metastasis. The lack of a difference in the elevated CD133 mRNA levels among cancer patients versus healthy volunteers may have been caused by a smaller sample size, a population with heterogeneous tumors, uncontrolled tumor stage, and, most importantly, the choice of healthy volunteers as negative controls without adjusting many variables that may affect CEP levels significantly.13, 38, 39

In contrast, in our study, we observed a statistical significant difference among recurrent patients versus nonrecurrent patients because of several factors: 1) the relatively larger sample with uniform cancer diagnosis; 2) the relatively matched variables among the study participants with adequate follow-up for recurrence; 3) predefined entry and exclusion criteria (eg, 4 weeks after surgery and the absence of cardiovascular disease) may have minimized the interference of variables on the CEP levels; 4) statistical analysis with adjustment for age, diabetes, smoking history, tumor grade, and disease stage; and 5) the wider range of CD133 mRNA readings by real-time RT-PCR (range, 0-83241 readings) noted in the current study compared with NASBA (range, 0-1088 readings) versus flow cytometry (range, 12–1,039 readings). The increased range in CD133 mRNA level readings with real-time RT-PCR compared with NASBA most likely is explained by the fact that NASBA measured CD133 copy numbers per cell relative to the U1A DNA levels in the cell, whereas the CD133 mRNA copy numbers measured by real-time RT-PCR are normalized to the GAPDH mRNA expression in the cell.13

CD133 belongs to the prominin family of proteins, the function of which remain unknown, and it has a novel isoform because of the deletion of a small exon of 27 nucleotides by alternative mRNA splicing.25, 40 However, the primers used in the current assay and in the NASBA assay are expected to recognize both isoforms. More recently, CD133 has emerged as the key marker that enriches tumor stem cells of the brain,41 prostate,42 melanoma,43 colon,44 and liver.45 The current assays for CD133-positive cells, including flow cytometry, are not designed to differentiate CEPs versus CD133-positive tumor stem cells, which conceivably are present in PBMCs from cancer patients. Therefore, it is possible to speculate that CEP fractions identified by the current assay methods may contain CD133-positive tumor stem cells that have the demonstrated ability to differentiate into endothelial cells lineage in vitro and to repopulate tumor graft in vivo.41–45

CD133 differentiates CEPs from resting and activated CECs, which also were increased in cancer patients who had progressive disease or vascular injuries compared with healthy volunteers.26, 28, 46 CECs identified by CD31 or CD36 ranged from 2 to 3900 cells per mL of PBMCs.12, 26,27 Elevated CD146 mRNA by NASBA fails to correlate with metastasis or survival compared with CD133 mRNA levels.13 The vast majority of CD146-positive cells in PBMCs were mostly lymphocytes but could be detected on the cellular component of tumor vessel walls.28 Chemotherapy at the MTD decreased CEC levels that were correlated with tumor response,18, 26, 27 whereas it has been demonstrated that short-course antiangiogenic therapy decreased both CEC levels and CEP levels.12, 15 Collectively, current literature supports the complementary and differential roles of CECs and CEPs as surrogates for vascular injuries and vascular repair, respectively.

CEPs (or CD133 mRNA) levels are emerging as an independent prognostic marker in cancer and heart disease but with opposing effects.11–13, 23, 24 Targeted antiangiogenic therapy improved treatment efficacy at the risk of increasing both acute and chronic cardiovascular complications, potentially by heightening the inhibition of CEPs.12, 47, 48 Long-term, large prospective studies with serial measurement of CEP (and/or CD133 mRNA) levels in cancer patients who receive conventional chemotherapy versus antiangiogenic therapy may yield information regarding the best timing to obtain these markers and validate the predictive and prognostic values of CEP (or CD133 mRNA) levels in patients with cancer and heart disease. Current assays for CEPs would have additional significant therapeutic implications if CD133-positive cancer stem cells also were found in PBMCs from patients with cancer.

Acknowledgements

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES

We thank Maarten Penning, PhD, Primagen, Inc. for his helpful comments on the article.

REFERENCES

  1. Top of page
  2. Abstract
  3. MATERIALS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. Acknowledgements
  7. REFERENCES