Circulation of Progenitor Cells in Obese and Lean Colorectal Cancer Patients

Colorectal cancer (CRC) is one of the most frequent malignancies in the Western world, with more than 500,000 new cases diagnosed each year (1). Great efforts have been made toward improving the survival outcome of patients with early and advanced CRC, but it remains a major health burden with more than 1 million cases worldwide and a disease-specific mortality rate of approximately 33% in the developed world (2). Obesity, caused by overgrowth of adipose tissue, has gained importance as a risk factor for both development and aggressive progression of CRC (3–6). Despite the well-characterized molecular events of the adenoma-to-carcinoma succession, reliable markers of CRC onset and progression that could be measured through minimally invasive assays are still missing. There is no laboratory blood test sensitive enough to detect CRC, and our most commonly used screening tools are either inaccurate and fail to detect most early precancerous polyps and some cancerous lesions in humans (i.e., fecal occult blood test) or costly and invasive requiring cleansing of the colon, sedation, and an inherent risk of damage to the mucosa (i.e., colonoscopy).

Survival improvements are seen with population screening. Indeed, the early detection of cancer and of precursor lesions followed by surgical removal would significantly reduce the number of cancer deaths. Therefore, diagnostic and prognostic markers of CRC that could be read out through simple blood test, especially at a curable stage, would significantly improve patient survival. Toward this goal, we have surveyed several cell populations, frequencies of which in the systemic circulation of patients with cancer have not been previously measured. On the basis of the multiparametric flow cytometric analysis of peripheral blood mononuclear cells (PBMC) using a panel of previously established immunophenotypes, we recently developed an approach to simultaneously survey circulating progenitor cells (CPC; containing hematopoietic and endothelial precursors), mesenchymal progenitor cells (MSC; also known as mesenchymal stromal cells), CD34bright leukocytes (CD34b LC), and mature endothelial cells (EC). In this study, identities of MSCs (CD34brightCD45⁻CD31⁻), CPCs (CD34brightCD31dimCD45dim), ECs (CD34dimCD31brightCD45⁻), and CD34b LCs (CD45brightCD34bright) were confirmed by showing expected lineage marker expression and differentiation assays (7). Interestingly, our prior analysis of cancer-free blood donors detected a markedly increased frequency of MSCs and CPCs in the circulation of obese individuals, suggesting their mobilization from adipose tissue.

In a search for a new diagnostic test to detect cancer through a minimally invasive procedure, here, we analyzed peripheral blood samples of patients with staged CRCs for frequencies of these distinct CD34⁺ cell populations. Our data show that MSCs and CD34b LCs, but not CPCs and ECs, are significantly more frequent in circulation of patients with CRC than in cancer-free donors. We also show that the frequency of circulating MSCs, but not of the other cell populations surveyed, is further elevated in obesity.

Under Institutional Review Board approval, peripheral blood was collected from a total of 45 patients with colon cancer from the Tulane Medical Center (New Orleans, LA) and The University of Texas MD Anderson Cancer Center (Houston, TX) prior to surgical or other therapeutic interventions. Demographic data (age and gender) and clinical stage of primary tumors were registered. The stage of CRC was defined according to the criteria of The American Joint Committee on Cancer and the International Union Against Cancer (AJCC/UICC). The body mass index (BMI; kg/m²) was calculated for each subject to determine whether excess adipose tissue content results in increased cell egress into the circulation. In this study, obese patients were defined as those with BMI of 30 kg/m² or more and lean patients as those with a BMI of less than 30 kg/m². The cancer-free donors (controls) were described previously (7). Peripheral blood samples (40 mL) were collected in heparin-coated tubes (BD Vacutainer; Becton Dickinson).

Circulating cells in the peripheral blood were quantified by multiparametric flow cytometry on the basis of established gating strategy (8) as we previously described (7). Briefly, PBMCs were isolated by Ficoll gradient centrifugation. PBMC analysis was conducted with an LSR-II flow cytometer and the FACSDiva Software (BD Bioscience). Cells were gated to exclude cell clumps, contaminating polymorphonuclear cells, red blood cells, platelets, endothelial microparticles, debris, as well as dead cells based on 7-aminoactinomycin D staining. Viable PBMCs (>500,000 per sample) were then used to enumerate individual populations (Fig. 1). For fluorescence-activated cell sorting on PBMCs, fluorescein isothiocyanate–conjugated CD31antibody (clone WM59), phycoerythrin-conjugated CD34 antibody (clone 8G12), and allophycocyanin-Cy7–conjugated CD45 antibody (clone HI30) along with appropriate isotype control immunoglobulin G from BD Bioscience were used.

Frequencies of CPC subsets were enumerated and expressed as percentage of viable PBMCs. A χ² test was used to assess the differences between cases and controls with regard to categorical variables. The Student t test was used to test for differences between the cases and controls for continuous variables. Multivariate linear regression analysis was conducted to estimate the difference in the frequencies of cells between CRC and control cases. Age and BMI were included in the multivariate model when appropriate. One-way ANOVA was done to test differences among different stages. The level of significance was set to P < 0.05 for all statistical analysis. Receiver-operating characteristic (ROC) curves were constructed, and the area under the curve (AUC) was calculated to evaluate the specificity and sensitivity of each cell population analyzed in cancer. All statistical analyses were 2-sided and were performed using Stata software (version 10.1; Stata).

In this study, 45 patients with CRC and 26 cancer-free (control) individuals (36 men and 35 women) with the mean age of 52.6 ± 14.5 years were recruited. All patients with CRC had a histologic diagnosis of adenocarcinoma and had not undergone treatment prior to blood collection. The age, gender, and BMI are presented in Table 1. The average age of the patients with CRC was 57.4 years (range: 22–83 years). With regard to the stage of tumors, 2 (4.5%) were stage I, 8 (17.8%) were stage II, 11 (24.4%) were stage III, and 24 (53.3%) were stage IV (Table 1). Controls were significantly younger than the cohort of patients with CRC, but there were not significantly more obese patients among controls than among patients with CRC. Two populations of CD34⁺ cell were consistently observed in our analysis: CD34 bright (b) and CD34 dim (d) by flow cytometry. On the basis of our previously established gating strategy (7), we could reliably identify 4 CD34⁺ cell populations of interest in the peripheral blood: CPCs (CD34bCD31dCD45d), MSCs (CD34bCD31⁻CD45⁻), CD34b LCs (CD34bCD45b), and ECs (CD34dCD31bCD45⁻; Fig. 1). We previously confirmed the identities of these 4 populations by conventional assays (7) and here further validate the CD34bCD31⁻CD45⁻ population as MSCs by flow cytometric assessment of mesenchymal marker expression (Supplementary Fig. S1).

The frequencies of each cell population analyzed are presented in Supplementary Table S1. The graphic distribution of values in distinct patient populations is shown in Fig. 2 and Supplementary Fig. S2. Initially, we found that there was a significant increase in the number of CD34bright cells within the bloodstream of patients with CRC compared with controls (P < 0.01). As shown in Table 2, PBMCs obtained from patients with CRC contained 0.06% CPCs and 1.50% ECs, frequencies similar to those observed in cancer-free donors. In contrast, CD34b LCs and MSCs, which constituted 0.08% and 0.03% of the viable PBMCs, respectively, in patients with CRC were 3-fold more abundant than cancer-free donors. After adjusting for confounding factors such as age and BMI, a multivariant analysis was used to confirm that the frequency of MSCs (P < 0.0001) and CD34b LCs (P < 0.0001) was significantly higher in patients with CRC than in cancer-free controls, whereas there was no statistical difference in frequencies of CPCs (P = 0.74) or ECs (P = 0.22) between these 2 groups.

Next, we assessed whether BMI (kg/m²) correlated with changes in the frequency of circulating CD34bright cells in patients with CRC assigned into 2 groups on the basis of the BMI: obese (BMI ≥ 30) and lean (BMI < 30). There was no significant difference in the frequencies of CD34bright cells in obese compared with lean patients with CRC (P = 0.12). The PBMCs obtained from obese patients with CRC contained 0.04% MSCs and 0.07% CD34b LCs, whereas CPCs and ECs constituted 0.07% and 1.71% of viable PBMCs (Table 3). Interestingly, the PBMCs obtained for obese patients with CRC contained significantly higher levels (2.3-fold) of MSCs than lean patients with CRC (P = 0.01). However, there was no statistically significant correlation between obesity and frequencies of circulating LCs, ECs, and CPCs in patients with CRC (Table 3).

We also assessed whether CRC and obesity correlated with changes in the frequency of circulating cell populations compared with cancer-free control donors also grouped as lean and obese. Forty-nine percent of all study subjects had a BMI of 30 or greater. Circulating cell frequency data for these individuals according to their BMI are presented in Table 3. We have previously reported that the frequencies of MSCs and CPCs were significantly increased in peripheral blood circulation of obese cancer-free donors compared with lean (7). However, comparing the PBMCs obtained from obese patients with cancer to obese cancer-free donors showed significantly higher levels of MSC and CD34b LC immunophenotypes, whereas EC and CPC frequencies were not different between the 2 groups. Specifically, the frequency of the MSCs and CD34b LCs were 3.0-fold and 3.8-fold higher in obese patients with CRC than obese controls, respectively. A similar pattern was seen comparing lean patients with CRC with lean noncancer patients (Table 3).

Next, we investigated whether the circulation of cell populations of interest changes in patients with a confirmed diagnosis of CRC according to their stage of disease. The frequency of MSCs in circulation of stage I/II donors was 0.03% of total viable PBMCs. Patients with stage III and IV disease had similar levels of this immunophenotype. Assignment of CRC into stage I/II, III, and IV disease also revealed no significant difference in cell circulation between these subgroups for ECs, CPCs, and CD34b LCs, although there was a trend for increased EC circulation in advanced CRC stages (Table 4).

To evaluate elevation in frequencies of individual cell populations as potential surrogate markers of CRC, ROC curves were constructed and the AUC was calculated (Fig. 2). The AUC was 0.77 (95% CI, 0.65–0.88) for MSCs; 0.59 (95% CI, 0.44–0.74) for CPCs; 0.82 (95% CI, 0.73–0.92) for CD34b LCs; 0.60 (95% CI, 0.46–0.74) for ECs; and 0.77 (95% CI, 0.65–0.89) for combined CD34b cells. The ROC curves helped determine the sensitivities and specificities for frequencies of the cells at various cutoff values. Using the optimal cutoff point, the sensitivity and specificity were 64% and 73% for MSCs, 51% and 58% for CPCs, 71% and 81% for LCs, 38% and 70% for ECs, and 77% and 66% for CD34bright cells. Combined, these data show that increased frequencies of circulating total CD34bright cells, MSCs, and CD34b LCs are associated with CRCs comparatively more strongly than those of circulating ECs or CPCs.

Numerous studies have been conducted to evaluate various biological markers to determine either the possibility of successful treatment for cancer or life expectancy. Advances in molecular biology and an increased understanding of tumor cell biology have led to the discovery of promising new biomarkers, some of which have been associated with disease subtypes and progression. To this avail, numerous research groups have analyzed peripheral blood samples of patients with cancer searching for cell type elevations, which could have diagnostic value. Recent attempts have focused on testing the circulating tumor cells (CTC) as a potential marker (9). Although potentially promising in CRC (10), this approach suffers from the rarity of CTCs in the peripheral blood and from the lack of specific markers available to prospectively identify CTC in the circulation. Therefore, using surrogate markers of cancer onset may provide an alternative and possibly a more sensitive approach.

Like other cancers, CRC development relies on angiogenesis and vasculogenesis from ECs and their progenitors (11, 12). Efforts in identification and characterization of ECs circulating in patients with cancer have been based on colony formation assays in culture and flow cytometry (13–15). For a number of cancers, circulating ECs have been reported as a potential diagnostic and prognostic marker. A number of reports have also explored their potential for monitoring the course of CRC (16). CPCs expressing endothelial antigens (endothelial progenitor cells; EPC) have also been investigated as a biomarker of cancer and response to cancer therapy (17–19). Specifically, EPCs have shown promise as a potential surrogate biomarker of CRC clinical response (20). Nevertheless, establishing an approach to reliably enumerate circulating ECs and EPCs in peripheral blood has remained a challenge. The problem has been that circulating cells displaying the endothelial phenotype are heterogeneous and thus are difficult to count (14). The challenge in identifying and characterizing the subpopulation of CPCs responsible for vasculogenesis and tumor growth has led to various classification approaches based either on a number of different flow cytometric sorting criteria applied or on different methods of colony formation assays in culture (21). Currently, there is no consensus on the optimal methodology for EC enumeration, and different groups have likely reported different cell populations as “EC” through alternative technical approaches (22). Possibly due to this complication, there are no conclusive reports on the usefulness of circulating EC enumeration for CRC diagnosis or prognosis and those available do not show striking results (23). None of the marker combination “signatures” used in flow cytometry are truly EPC or EC specific (24). More importantly, the relatively high background of EPC and EC circulation in cancer-free donors, especially under pathologic conditions (that are not necessarily malignant), makes these cells of questionable practical use for prognostic and diagnostic studies.

The purpose of the study was to investigate the presence and pattern of change of CPC subsets and their association with the presence of CRC. In this study, we have shown that there was an increased frequency of CD34bright cells in the systemic circulation of patients with CRC. Elevated circulating levels of MSCs in patients with cancer, measured by the colony–forming unit fibroblast assay, have been previously reported (25); however, our study is the first report on flow cytometric MSC enumeration in the systemic circulation of patients with CRC. Our data show that MSCs and CD34b LCs, but not CPCs and ECs, are significantly more frequent in the circulation of patients with CRC than in cancer-free donors. We also show that the frequency of circulating MSCs, but not of the other cell populations, is further elevated in obesity. A limitation of our study is that it was carried out on patients mostly with advanced CRC. Whether the CD34bright populations characterized here are also elevated in patients with adenoma remains to be established. The absolute value of increased frequency of circulating MSCs, CD34b LCs, and total CD34bright cells is to be further determined in larger patient cohorts and their possible utility is yet to be assessed in future studies.

Cancer progression depends on the recruitment of progenitor cells to tumors where they contribute to tumor microenvironment, supporting vascularization and a desmoplastic state. It has been shown that both CD45⁺ and CD45⁻ cells can traffic to tumors and potentate disease progression through remodeling stromal matrix and influence recruitment and function of other immune system cells (26, 27). The role of CD34b LCs, also known as fibrocytes (28), in cancer has not been rigorously tested despite their abundance in circulation. However, they are clearly implicated in fibrotic disorders (29) and are likely to play a role in tumor physiology possibly through differentiation into other monocyte-derived tumor cells such as alternatively activated macrophages (30). Although MSCs are normally undetectable in the peripheral circulation, there is a body of evidence pointing to their role in tumor microenvironment in many cancers including CRC (31). The capacity of MSC to “sense” cancer as a site of injury or inflammation and traffic to tumors has been established (32, 33). Gene expression analyses have confirmed that MSCs contribute to the pool of cancer-associated fibroblasts (CAF) that orchestrate stroma remodeling during cancer progression (34). Interestingly CAFs have been implicated not only in cancer progression but also in cancer initiation (35).

The observed egress of CD34b LCs is likely explained by their mobilization from the bone marrow, the major reservoir of hematopoietic progenitors. Indeed, their elevation in CRC was irrespective of the BMI. In contrast, the frequency of circulating MSCs is comparatively higher in obese patients with CRC, suggesting that their mobilization from adipose tissue contributes to the circulating pool. Importantly, whereas MSCs in the bone marrow are believed to be CD34⁻, MSCs in adipose tissue, known as adipose stromal cells (ASC) unambiguously express CD34 (36). This is consistent with circulating MSCs enumerated by our assay being ASCs. Our previous studies in mouse models show that ASCs can migrate to tumors and promote cancer progression (37). This may have important clinical implications, as cells mobilized from adipose tissue could become recruited to the sites of disease and serve as an extra supply of progenitor cells in obese patients. We speculate that ASC mobilization and trafficking to tumors could partially account for the epidemiologically observed acceleration of cancer progression in obese patients. However, at this point, other organs, such as the bone marrow, cannot be excluded as the source of MSCs mobilized in cancer.

Considering age as a potential confounding factor, we included it into the multivariate analysis. Interestingly, previous studies have pointed to age-related decrease in the capacity of CD34⁺ cells to mobilize (38). Because patients with CRC were generally older than controls in our study, it should be emphasized that the increase in CD34b cells due to cancer is observed despite the age differences.

Our study sets the stage for the prospective assessment of circulating levels of MSCs and CD34b LCs during CRC onset, progression, and treatment. We propose that these CD34⁺ populations mobilized into the systemic circulation in response to tumor signals are potentially advantageous to ECs and CPCs used as experimental markers previously and could be detected for diagnostic purposes. Peripheral blood–based detection of the circulating cell populations could improve the chances for early CRC detection and further studies will investigate whether these new markers also detect the onset of disease. The circulating markers described here might also have utility in the prognosis of CRC, and their predictive value needs to be addressed in a larger cohort of patients. Finally, these circulating cells may represent a viable therapeutic target, and approaches to their depletion from circulation are to be investigated.

C.F. Bellows has honoraria from Speakers Bureau and is a consultant/advisory board member for LifeCell Corp. No potential conflicts of interest were disclosed by the other authors.

This work was supported by KG080782 from Susan G. Komen for the Cure, CNE-119003 from American Cancer Society, RP100400 from Cancer Prevention & Research Institute of Texas (CPRIT; to M.G. Kolonin), and CA140388 from NIH.