Chemotherapeutic agent-mediated elimination of myeloid-derived suppressor cells

ABSTRACT

Immunotherapy has shown great promise in the fight against cancer, as evidenced by the clinical efficacy of chimeric antigen receptor T cells in hematologic malignancies and checkpoint blockade in certain solid tumors. However, a considerable number of patients fail to respond to these therapies. Induction of myeloid-derived suppressor cells (MDSCs) by growing tumors has been shown to be one important factor limiting the efficacy of cancer immunotherapy. Recently, several chemotherapeutic agents used in conventional cancer chemotherapy have been found to reduce MDSC numbers in tumor tissues as well as in the peripheral lymphoid organs, and combining these agents with immunotherapy improved survival of tumor-bearing hosts. In this review, we will highlight the effects of chemotherapeutic agents on MDSC accumulation, and examine the various factors likely to influence these effects.

KEYWORDS:

• Chemotherapy
• immunotherapy
• myeloid-derived suppressor cells
• overall survival
• solid tumors

Abbreviations
DD	=	dose-dense
Five-FU	=	fluorouracil
M-MDSCs	=	monocytic myeloid-derived suppressor cells
MDSC	=	myeloid-derived suppressor cells
MTD	=	maximum-tolerated dose
NKG2D	=	natural killer group 2D
PMN-MDSC	=	polymorphonuclear myeloid-derived suppressor cells

Introduction

Myeloid-derived suppressor cells (MDSCs) are a phenotypically heterogeneous cell population accumulating in tumor sites and peripheral lymphoid organs in tumor-bearing hosts.^1,2 In mice, these cells are characterized by the co-expression of the myeloid-cell linage differentiation antigens CD11b and Gr1. MDSCs can be further subdivided into two major subsets, polymorphonuclear (PMN-MDSC, initially termed gMDSC³) of the phenotype CD11b⁺Ly6G⁺Ly6C^low, and monocytic (M-MDSCs) with the expression profile CD11b⁺Ly6G⁻Ly6C^hi.⁴ In human peripheral blood mononuclear cells, MDSCs can be divided into three subtypes: the equivalent to PMN-MDSC are defined as CD11b⁺CD14⁻CD15⁺ or CD11b⁺CD14⁻CD66b⁺, M-MDSC as CD11b⁺CD14⁺HLA-DR⁻/loCD15⁻, and early-stage MDSCs of the phenotype Lin⁻ (including CD3, CD14, CD15, CD19, CD56), HLA-DR⁻CD33⁺.⁴ MDSCs mediate immunosuppression through multiple mechanisms: (i) inhibition of proliferation and activation of CD4⁺ and CD8⁺ T lymphocytes through arginase-1 or nitrogen oxide synthase 2 as well as indoleamine 2,3-dioxygenase;^5,6 (ii) induction of oxidative stress via production of reactive oxygen species and reactive nitrogen species that cause loss of T cell receptor ζ-chain expression and desensitization of the T cell receptor;⁷ (iii) altering macrophage phenotype toward to a type 2 response and subsequent activity by increasing MDSC production of IL-10;⁸ (iv) inhibition of the cytotoxicity, natural killer group 2D (NKG2D) expression, and interferon-gamma production of natural killer cells through a cell-contact-dependent mechanism that involves membrane-bound transforming growth factor β;⁹ (v) induction of immunosuppressive T regulatory cells¹⁰ and (vi) upregulation of programmed death-ligand 1 on MDSC via hypoxia-inducible factor-1α.¹¹ MDSCs are also involved in non-immune suppressive processes that may aid tumor development such as promoting blood vessel formation,¹² inducing epithelial–mesenchymal transition,¹³ and enhancing stemness of cancer cells.¹⁴ Data from clinical studies provide strong evidence that the frequency and number of MDSCs in the peripheral blood correlate with cancer stage, metastatic tumor burden, and response to systemic therapy,^15-17 rendering these cells a promising predictive biomarker and/or therapeutic target for improving the efficacy and prognosis of patients with cancers.^18,19

With the recognition that MDSCs inhibit antitumor immune responses, it is conceivable that efficacy of cancer immunotherapy can be improved by eliminating MDSC accumulation in tumor-bearing hosts. In addition to their intended cytotoxic effects on tumor cells, chemotherapeutic agents have been found to affect the recruitment of MDSCs to tumor sites, peripheral blood, and lymphoid organs.^1,20 Most studies to date have shown that combining these agents with immunotherapy decreases MDSC numbers, resulting in enhanced efficacy and a synergistic benefit on survival in tumor-bearing hosts.^21-24,25 For example, it was reported that the combination of different chemotherapeutic regimens in association with a specific immunotherapy (adoptive T cell transfer) induced a significant control of progression in MCA203 fibrosarcoma-bearing mice.²⁶ Clinically, MDSC elimination by chemotherapy correlates with the induction of a tumor vaccine-induced immune response in renal cell carcinoma patients²⁷ and a better immune response after vaccination was detected after MDSC elimination in ovarian and cervical cancer patients.^28,29

However, how effectively chemotherapeutic agents deplete MDSCs might impact the efficacy of the concurrent immunotherapy. Efficient screening of suitable chemotherapeutic agents will be a key factor in the design of combinatorial regimens together with immunotherapy. Here, we review the most recent reports on the effects of the most commonly used chemotherapeutic agents in clinic such as gemcitabine, fluorouracil (5-FU), doxorubicin, and paclitaxel on MDSC accumulation, and analyze the possible factors responsible for these effects. In addition, we review the latest findings on the correlation between MDSC number and efficacy of concurrent chemotherapy and immunotherapy.

Various results of chemotherapeutic agents on MDSC accumulation

Gemcitabine (a deoxycytidine analog that inhibits ribonucleotide reductase) and fluorouracil (5-FU, an antimetabolite agent that targets thymidylate synthase) are two generally recognized chemotherapeutic agents that deplete MDSCs in both animal models as well as in patients.^30-33 In one study, one of several cytotoxic agents, including cyclophosphamide, paclitaxel, raltitrexed, gemcitabine, doxorubicin, 5-FU, or oxaliplatin were injected into lymphoma tumor-bearing mice.³¹ The authors found only 5-FU (and to a lesser extent gemcitabine) significantly decreased the number of MDSCs in both the spleen and tumor bed. The elimination of MDSCs by 5-FU increased interferon-γ production by tumor-specific CD8⁺ T cells and promoted T cell-dependent antitumor responses.³¹ Further analysis showed that 5-FU was able to contract M-MDSC without (or less) affecting PMN-MDSC accumulation.^26,31 Moreover, data reported that 5-FU administration was able to contract M-MDSC in the spleen, although this effect was not detected in the lymph nodes.²⁵ However, findings from several other studies showed conflicting effects of these two agents. In a preclinical model of thymoma, 5-FU administration alone failed to reduce numbers of either M-MDSCs or PMN-MDSCs in the draining lymph node, or PMN-MDSCs in the spleen, although the combination of 5-FU and an adenoviral tumor vaccine had shown a synergistic benefit on survival of tumor-bearing mice.²⁵ In a B16F10 lung metastatic mouse model, application of 5-FU resulted in only minimal effects on the frequency of MDSCs recruited into the tumor microenvironment, while the therapy decreased the frequency of MDSCs in the spleen and lung. These findings suggest that 5-FU decreases MDSCs systemically, which was concomitant with the improved survival of the animals.³⁴ In a clinical study, patients with advanced pancreatic cancer were administered gemcitabine together with capecitabine, a prodrug that is enzymatically converted to 5-FU in the body. The study showed that there was no consistent reduction in blood MDSCs post-treatment. However, in patients with baseline MDSCs greater than the median, those whose MDSC levels did fall seemed to experience more immunological benefit, though the small numbers of patients precluded definitive conclusions.³⁵

FOLFIRI is a chemotherapy regimen for treatment of colorectal cancer, and is made up of 5-FU, irinotecan, and leucovorin. Reports from studies using a FOLFIRI regimen are also inconsistent in terms of impact on MDSC counts. In one study, mice were injected with MC38/CEA2 colon adenocarcinoma cells subcutaneously before FOLFIRI treatment. At day 6 after FOLFIRI treatment of tumor-bearing mice, the number of splenic MDSCs was significantly decreased compared with control mice, leading to enhanced tumor-specific responses.³⁶ In another report, following FOLFIRI treatment in patents with colorectal cancer, the percentage of MDSCs in the peripheral blood was not observed to decrease; to the contrary, it continuously increased.³⁷

Similar apparent discrepancies were encountered when treatments involved doxorubicin, an antineoplastic drug commonly used in the treatment of a wide range of cancers, including hematological malignancies, soft tissue sarcomas, and several types of solid tumors. Using a breast cancer mouse model, it was shown that the proportion and absolute number of MDSCs in the spleen and peripheral blood were significantly reduced following doxorubicin treatment.³⁸ In contrast, a related study using a B-cell lymphoma mouse model reported that doxorubicin induced the expansion of M-MDSCs in the spleen.²¹

Collectively, the cumulative data indicate that chemotherapeutic agents have diverse effects in regulating MDSC accumulation in different situations. However, it should be noted that these data were derived from independent studies, and although chemotherapeutic agents were identical, as yet unidentified factors, including the chemotherapy dose, time, administration schedule, location of MDSC sampling, as well as the different disease stages in clinic setting, may also have contributed to differences in MDSC levels.

Factors shaping the effect of chemotherapeutic agents on accumulation of MDSCs

Dosage of drugs

Dose-dependent inhibition of MDSC accumulation has been observed in tumor-bearing hosts treated with other anticancer therapies, including molecular target agents. For example, sunitinib is one commonly used tyrosine kinase inhibitor, typically first-line therapy for metastatic renal cell carcinoma. Animal studies have shown that sunitinib treatment dose-dependently depleted MDSCs in the tumor, spleen, and in circulation.³⁹ Combining this MDSCs-decreasing agent with various immune-based cancer therapies, such as an agonistic CD40-antibody,⁴⁰ an αvirus-based cancer vaccine,⁴¹ tumor-infiltrating lymphocytes,⁴² or a viral vector based cancer vaccine,³⁹ effectively enhances their efficacy. These studies provide evidence that dosage is an important factor affecting the ability of an agent to inhibit MDSC accumulation and subsequent antitumor effect.

Similarly, treatment with paclitaxel, an agent inhibiting microtubule disassembly, also affects MDSC accumulation diversely. In animal studies, at a dose of 1 mg/kg body weight, paclitaxel was shown to cause a significant reduction in frequencies of tumor-infiltrating MDSCs and restore CD8⁺ T cell effector functions,⁴³ an effect not observed at 36 mg/kg paclitaxel.³¹ In a clinical study, dose of 175 mg/m² paclitaxel showed a significant yet modest increase in both the absolute number and percentage of circulating MDSCs.¹⁵ Together, this body of evidence suggests that single use of paclitaxel may result in different numbers of circulating MDSC, depending on the specific dosage administered.

Another example of a dose-effect relationship is the application of cyclophosphamide in a melanoma mouse model. At a dose of 50 mg/kg body weight, cyclophosphamide was shown to have negligible influence on MDSC accumulation in either tumor sites or spleens of tumor-bearing mice.^30,34 However, at a dosage between 100 and 300 mg/kg, the number of MDSCs was clearly expanding.¹⁷ This dose-effect relationship was directly demonstrated in another recent report using the same experimental model, where substantial accumulation of melanoma-infiltrating MDSCs were observed following administration of 2.5 mg cyclophosphamide, but not 1.0 mg per mouse.⁴⁴

In a recent study using an ovarian cancer model, the effects of different doses of concurrent pacilitaxel and cisplatin on MDSC accumulation were investigated. Tumor-bearing mice were treated with two different regimens of combination chemotherapy: a lower “dose-dense” (DD) regimen and a higher “maximum-tolerated dose” (MTD) regimen. The DD regimen was defined as the combination of 5 mg/kg paclitaxel and 3 mg/kg cisplatin, while the MTD regimen consisted of higher dosages of 12 mg/kg paclitaxel and 7 mg/kg cisplatin. These results clearly showed that DD chemotherapy profoundly reduced the number of MDSCs, more significantly than mice in the MTD group. These data may provide at least a partial explanation as to why the DD regimen produced superior clinical outcomes.⁴⁵

As described above, diverse effects of dosages of chemotherapeutic agents on MDSC accumulation have been observed. It is important to note that this conclusion is based on the results from separate studies. Different experimental conditions in each study, different ways of measuring MDSCs, etc. could potentially lead to different results. Additional studies directly comparing different doses in the same experimental model are needed for confirmation.

Timing of drug administration

The timing of treatment with chemotherapeutic agents also affects MDSC accumulation. One study using gemcitabine in a mammary carcinoma mouse model supported this point.³⁰ Gemcitabine was administered at two different time points: early gemcitabine was injected on day 5 post tumor inoculation and repeated once weekly; late gemcitabine was given as a single dose on days 20–25. MDSC levels in the spleen were measured at different time points. The authors measured significant increases in both percentage and absolute number of MDSCs in the spleens on day 7 post tumor inoculation compared with naive mice; however, early gemcitabine treatment neither decreased the percentage nor the absolute numbers of MDSCs. On the contrary, in the late gemcitabine group, there were significant decreases in both percentage and absolute numbers in MDSCs, an observation that correlated with greater in vitro proliferative activity by splenic T cells.³⁰

Location of MDSC sampling

From recent studies, it has become clear that depending on the organs selected, different MDSC levels are detected in response to chemotherapeutic agents. One study showed that melanoma-bearing mice treated with 5-FU exhibited no change in the frequency of MDSCs at the tumor site. However, in the spleen and lung, the frequency was significantly decreased.³⁴ The results of a second study also showed that changes in MDSC levels vary by anatomic site. In the spleen, treatment with 5-FU resulted in substantial reductions of M-MDSCs and PMN-MDSCs. In contrast, no decrease in M-MDSCs or PMN-MDSCs was found in the draining lymph nodes.²⁵ Another study using the same mouse melanoma model showed that 5-FU treatment led to statistically significant reduction of the relative and absolute numbers of MDSC peritoneal cells, but did not affect the number of MDSC spleen cells.³² Like 5-FU, paclitaxel was also shown to exert different effects on MDSC accumulation in different organs. As reported in one study, administration of an ultra-low dose of paclitaxel caused a significant reduction in frequency of MDSCs in the tumor, but not in the metastatic lymph nodes, spleen, or bone marrow.⁴³

To evaluate the specific effects of individual chemotherapeutic agents on the accumulation of MDSCs, separate from the additional effects of the tumor, one study used a model using normal mice without tumors.⁴⁴ This study tested whether changes in MDSC levels could be observed across various organs after cyclophosphamide treatment. When these mice were injected with 2.5 mg cyclophosphamide, an increase in MDSC levels was found in the spleen, peripheral blood, and lymph nodes, but not, however, in the bone marrow, suggesting the effects of chemotherapeutic agents on MDSCs vary by compartment, irrespective of the presence of tumor.

Currently, it is still unknown in which anatomic compartment depletion of MDSCs is the most important to improve antitumor activity of T cells. We selected several studies in which clinically commonly used chemotherapeutic agents in solid tumors were used and their effects on MDSCs were investigated (Table 1). It seems that a decrease in MDSCs in the spleen is sufficient to produce an improved therapeutic effect, irrespective of MDSC changes in other anatomic compartments, including the tumor site. However, this assumption needs to be further investigated.

Table 1. The relationship between anatomic compartment depletion of MDSCs and antitumor effect.

						Effect on MDSCs in:
Study name	Phenotype of MDSC	Model	Drug (dose)	Timing of drug	Timing of MDSC sampling	Blood	Spleen	Tumor	Lymph nodes	Other	Anti-tumor effect
Vincent, et al^25	CD11b^+Ly6G^+Ly6C^int	EL4	Gemcitabine (120 mg/kg)	Tumor sizes reached 100 mm^2	Day 5 after chem.	Not tested	Decrease	Decrease	Not tested		Good
	CD11b^+Ly6G^−Ly6C^+		5-FU (50 mg/kg)				Decrease
Geary, etal^21	CD11b^+Ly6G^+Ly6C^low	E.G7	5-FU (40 mg/kg)	Day 7 after tumor injection	Day 5 after chem.	Not tested	Decrease	Not tested	Various		Good
	CD11b^+Ly6G^−Ly6C^hi
Otsubo, etal^28	CD11b^+Gr-1^+	B16F10	5-FU (50 mg/kg)	Day 3 after tumor injection	Day 7, 14, 21	Not tested	Decrease at day 21	No change	Not tested	Lung:decrease at day 21	Good
Annels etal^23	CD11b^+Lin1^−HLA^−DR^−	PC	Gemcitabine (1000 mg/m^2)	Gemcitabine was given every 4 wk together with capetabine for 3 wk followed by 1 wk's rest	7 wks of treatment, 1wk following 5th gem infusion, prior to 6th gem infusion and whilst taking cap.	Various	Not tested	Not tested	Not tested		Various
		patient	Capecitabine (1660 mg/m^2)
Kim etal^29	CD11b^+GR-1^+	MC38	5-FU (100 mg/kg)	Day 10 and 11 after tumor injection	Day 6, 14 after chem.	Not tested	Decrease at day 6, increase at day 14	Not tested	Not tested		Good
			LV 40 mg/kg)
			Irinotecan (20 mg/kg)
Kanterman etal^30	CD11b^+CD33^+HLA^−DR^−	CC patient	FOLFIRI (dose not indicated)	Not indicated	Not indicated	Increase	Not tested	Not tested	Not tested	Colon,lamina propria, and epithelium all increase	Not tested
	CD11b^+Gr-1^+	AOM/DSS-induced CC mouse model	5-FU (50 mg/kg)	Not indicated	Not indicated	Not tested	Increase	Increase	Not tested		Good
			Irinotecan (50 mg/kg)
Alizadeh^31	CD11b^+GR-1^+	4T1	Doxorubicin (2.5 or 5 mg/kg)	Day 7 and 12 after tumor injection	Day 14, 17 and 23 after tumor injection	Decrease	Decrease	Decrease	Not tested		Good
Ding etal^17	CD11b^+Ly6c^hiLy6G^−Gr1^int	A20	Doxorubicin (not indicated)	When tumor sizes reached 170 mm^2	Day 7 after chem.	Not tested	Increase	Not tested	Not tested		Not tested
	CD11b^+Ly6C^lowLy6G^hiGr1^hi

Note: PC = pancreatic cancer; CC = colorectal cancer, 5-FU = 5-fluorouracil, LV = leucovorin

Detection time of MDSCs

Several recent studies have investigated the effects of chemotherapeutic agents on the dynamics of MDSC levels using either normal mice or various tumor models, showing that the number of MDSCs measured varies, depending on the time interval between administration of chemotherapeutic agents and detection of the MDSCs. In one study, normal mice were treated with intraperitoneal injections of cyclophosphamide, and MDSCs were detected at different time points post treatment.⁴⁶ At days 6 and 10, there was a significant increase in MDSC levels in the spleen, both in percentage and absolute numbers. However, by day 20, MDSC numbers had started to fall, still higher than controls at day 20 but lower than day 6–10 levels.⁴⁶ Similar results were observed in a recent study using both naïve mice and tumor-bearing mice to investigate the effects of cyclophosphamide on MDSC accumulation.²¹ All mice were treated with a single dose of cyclophosphamide, and spleen cells were harvested at different time points to enumerate the MDSCs. They found that in naïve mice, the numbers of M-MDSCs and PMN-MDSCs were both reduced shortly after cyclophosphamide, reaching the nadir by day 2 and day 4, respectively; and both populations rebounded thereafter, resulting in net increase in cell number by day 10. In tumor-bearing mice, there was an initial reduction in the numbers of M-MDSCs in the spleen (day 2) followed by a rebound thereafter. In the tumor, the rebound of M-MDSCs peaked at day 7, and by day 10 reached a stable level that was elevated compared with the starting point.²¹

The influence of detection time on MDSC numbers has also been observed for other chemotherapeutic agents, including 5-FU and doxorubicin.^34,38 C57BL/6 mice received subcutaneous inoculation of B16F10 syngeneic melanoma cells, with 5-FU administered 3 d post-inoculation. It was shown that neither the frequency of splenic MDSC nor the absolute numbers had changed by day 7 or 14, however, a significant decrease was observed by day 21.³⁴ Another study using doxorubicin in a 4T1 breast cancer mouse model similarly reported results that varied according to time interval, though with an inverse trajectory from the previous study.³⁸ On days 14 and 17, doxorubicin treatment resulted in a significant reduction in the proportion and absolute numbers of MDSCs in the spleen and blood of mice; however, these cells were reconstituted by day 23.³⁸ Besides single chemotherapeutic agents, combinational chemotherapy also showed different results on MDSCs if detected at different time points. For example, a study using FOLFIRI treatment showed a reduction in the numbers of MDSCs on day 6, while significantly higher numbers of MDSCs were detected at day 14.³⁶

One explanation for the various results at different time points is the kinetics of MDSC subtypes in the absence of chemotherapy. It has been reported that there were substantial expansion of PMN-MDSC in blood and spleens over time in different transplantable tumor models and in a spontaneous tumor model. In contrast, the increase in the number of M-MDSC was relatively small.⁴⁷ Further analysis showed that M-MDSC can generate cells with CD11b+Ly6CloLy6G+phenotype of PMN-MDSC in vitro and in vivo.⁴⁷ Therefore, the plasticity of M-MDSC to promote PMN-MDSC generation might also contribute to the different kinetics of blood/spleen accumulation of MDSC in chemotherapeutic agent-treated tumor-bearing hosts.

Mouse and tumor models

Another important factor accounting for various effects of a particular chemotherapeutic agent on MDSC accumulation is the mouse model used in studies. Under identical conditions, use of naïve or tumor-bearing mice can result in apparently different findings of chemotherapeutic agent effects on MDSC number, as is observed in two previously reported studies. In one study, naive mice were used and treated with intraperitoneal injections of cyclophosphamide.⁴⁶ At days 6–10, a significant increase in splenic MDSCs was observed, both in percentage and absolute numbers.⁴⁶ In contrast, using a mouse model with a tumor, a separate study showed that neither splenic MDSC frequency nor absolute number changed when detected at day 7, although cyclophosphamide was given at the same dosage and the same routes of administration.³⁴

The precise effect of the chemotherapeutic agent on MDSC levels also depends on the tumor type. In a recent report, the authors used two different tumor cell line models, B16F10 melanoma and EL4 T-cell lymphoma, to compare the effects of chemotherapeutic agents on MDSC depletion.⁴⁸ Mice bearing one of the two tumor types were injected with a single dose of either cisplatin or 5-FU. The authors found that the increase in the percentage of splenic MDSCs for both B16F10 and EL-4 tumor-bearing mice, compared with the MDSCs in spleens of mice lacking tumors, was similar. However, when these two different tumor-bearing mice were compared after treatment with cisplatin or 5-FU, the EL-4-bearing mice exhibited much lower levels of splenic MDSCs.⁴⁸

In addition, the route of administration of chemotherapeutic agents may affect MDSC levels. Several studies showed that gemcitabine treatment reduces the number of tumor-induced MDSCs in transplantable tumor models.^30,49 In most of these models, gemcitabine was administered via the subcutaneous route. Recently, another study used intraperitoneal injection instead, to test its efficiency in reducing MDSC numbers. When the gemcitabine treatment regimen was given to tumor-bearing mice, the absolute number of MDSCs in the peritoneal cavity did not change.³²

Conclusion

Based on the above findings that various parameters affect the ability of chemotherapeutic agents to eliminate tumor-induced MDSCs, we propose that most chemotherapeutic agents have the potential ability to inhibit MDSCs, if administered at the proper time and dose in a proper model with a proper administration mode.

MDSCs play a potentially important role in the efficacy of immunotherapy and can be modulated by various chemotherapeutic agents. However, the exact effects of these agents on MDSCs can be difficult to discern based on apparent inconsistencies between published studies. These apparent contradictions can likely be explained by differences in important variables such as drug dose and timing, location and timing of MDSC sampling, and mouse model and tumor type (Fig. 1). These differences in mouse model studies to date preclude definitive conclusions from being drawn about how best to translate these approaches to the clinic. This highlights the need for additional studies linking the impact of these variables with antitumor efficacy to define the relevant biomarkers. Ultimately, human clinical trials with correlative studies examining MDSCs are needed to evaluate the overall effect of chemotherapy regimens on efficacy of immunotherapies and the role MDSC depletion plays in successful treatment.

Figure 1. Chemotherapeutic agents have the potential ability to inhibit MDSCs if administered at the proper time and dose in a proper model with a proper administration mode. At the point when MDSCs are inhibited, immunotherapy addition has an enhanced antitumor efficacy.

Disclosure of potential conflicts of interest

No potential conflicts of interest were disclosed.

Acknowledgments

The authors thank Dr Torsten Juelich for linguistic assistance during the preparation of this manuscript.

Funding

This work was supported by the national Natural Science Foundation of China under Grant (81000914) and Foundation of He'nan Health Committee under Grant (2011010011).