Incidence of invasive fungal disease after unmanipulated haploidentical stem cell transplantation was significantly higher than that after HLA-matched sibling transplantation

Authors

  • Y. Sun,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • L. Xu,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • D. Liu,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • X. Zhang,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • W. Han,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • Y. Wang,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • H. Chen,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • Y. Chen,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • F. Wang,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • J. Wang,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • Y. Ji,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • F. Tang,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • K. Liu,

    1. Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author
  • X.-J. Huang

    Corresponding author
    • Peking University People's Hospital, Peking University Institute of Hematology, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Beijing, China
    Search for more papers by this author

Corresponding author: X.-J. Huang, Beijing Key Laboratory of Hematopoietic Stem Cell Transplantation for the Treatment of Hematological Diseases, Peking University People's Hospital, Peking University Institute of Hematology, 11 Xizhimen South Street, Beijing 100044, China

E-mail: huangxiaojun@bjmu.edu.cn

Abstract

The aim of this study was to determine the incidence, clinical features and outcome of invasive fungal disease (IFD) after either unmanipulated haploidentical haematopoietic stem cell transplantation (HSCT) or human leukocyte antigen (HLA)-matched sibling HSCT. This was a head-to-head comparative study performed at a single centre. Patients were admitted between 2007 and 2010, and IFD was evaluated according to the revised EORTC/MSG criteria, with only proven and probable cases included. Of the 1042 consecutive patients enrolled, 390 received the HLA-matched HSCT and 652 received unmanipulated haploidentical HSCT. A total of 61 (5.8%) patients had IFD, including 15 proven cases and 46 probable cases. The incidence of IFD after unmanipulated haploidentical HSCT was significantly higher than that after HLA-matched transplantation (7.1% vs. 3.3%, respectively; p 0.007). IFD occurred later in patients receiving HLA-matched transplantation compared with patients receiving unmanipulated haploidentical HSCT (141.5 vs. 23 days, respectively; p 0.04). In multivariate analysis, acute graft-versus-host disease (GVHD) grades III to IV (HR = 2.214, 95% CI, 1.139–4.304; p 0.019), extensive chronic GVHD (HR = 2.413, 95% CI, 1.377–4.228; p 0.002) and haploidentical transplantation (HR = 2.648, 95% CI, 1.111–6.310; p 0.028) were identified as significant risk factors associated with IFD. The response to antifungal therapy and the IFD-attributable mortality were similar between the two types of transplantation. In conclusion, patients who received unmanipulated haploidentical HSCT had a higher risk of IFD than those patients who received HLA-matched HSCT, but the prognosis of IFD was not associated with the HLA type.

Introduction

The use of a haploidentical haematopoietic stem cell transplant (HSCT) provides a chance for cure among patients without a human leukocyte antigen (HLA)-matched sibling or unrelated donor. Extensive in vitro T-cell depletion (TCD) was adopted in centres in Italy and other countries to prevent graft-versus-host disease (GVHD) and to overcome the barriers to engraftment [1, 2]. However, due to the poor post-transplant immune reconstitution caused by the in vitro TCD, invasive fungal disease (IFD) was an important cause of morbidity and infection-related mortality [3, 4]. A study of 205 patients from Perugia [5] suggested that the risk of invasive aspergillosis (IA) after haploidentical transplantation with TCD was 2.7 times higher than that after HLA-matched transplantation.

To overcome these limitations, researchers from other transplant centres have focused on unmanipulated allografts with post-transplant immune suppression. In recent years, researchers at Peking University developed a novel approach of HLA-mismatched and haploidentical blood and marrow transplantation without in vitro TCD [6, 7]. With this protocol, relapse rates, transplant-related mortality (TRM), leukemia-free survival (LFS) and overall survival (OS) were comparable to those of HLA-matched sibling transplantation [8]. While the immune reconstitution after unmanipulated haploidentical transplantation improved [9], it is still unknown whether the incidence of IFD is decreased to a level comparable to that of HLA-matched transplantation. In our previous report, 291 patients who received unmanipulated haploidentical transplantation were reviewed according to the EORTC 2002 criteria [10]. The 1-year cumulative incidence of fungal infection was 13.1%. However, a direct comparison with HLA-matched sibling transplantation has not been conducted.

In the present study, we compare the incidence of IFD among patients who received HLA-matched vs. haploidentical HSCT between 2007 and 2010 at the Peking University Institute of Hematology.

Patients and Methods

Patient cohort and follow-up

Patients who received allogeneic stem cell transplantation from either HLA-matched sibling donors (MSD) or related haploidentical donors (HID) between January 2007 and December 2010 were consecutively enrolled. All patients signed informed consent forms prior to treatment, and the treatment programme was approved by the Ethics Committee of the Peking University People's Hospital. Patients who received a second transplantation were excluded. The last follow-up was on 31 December 2011. Of the total patient population, 291 who received a HID transplantation between January 2007 and December 2008 were previously reported [10].

Transplantation procedure

The method of transplantation has been described in previous reports [6, 8]. The conditioning regimen consisted of a modified busulfan (BU) and cyclophosphamide (CY) regimen in most MSD transplantations; however, for HID transplantations, most patients received a regimen of the BU and CY plus antithymocyte globulin (ATG, Thymoglobulin®; Genzyme, Cambridge, MA, USA) 2.5 mg/kg/day for 4 days. Cyclosporine (CSP) plus short-term methotrexate(MTX)and mycophenolate mofetil (MMF) were administered for the prevention of graft-versus-host disease. On day +1, MTX (15 mg/m2) was administered intravenously and then 10 mg/m2 were given on days +3, +6 and +11 after transplantation. Four doses of MTX were given to HID transplantation patients while only the first three doses were given to MSD transplantation patients. Intravenous cyclosporine (2.5 mg/kg twice a day) was started on day −9 and continued until patients were able to tolerate oral medication. Thereafter, CSP was given orally twice daily with target trough levels of 150–250 ng/mL. MMF (7.5 mg/kg twice daily) was begun on day −9 and discontinued on day 14 in the HLA-matched patients. However, in mismatched patients, discontinuation of MMF was on days 30–60, based on the presence or absence of GVHD.

Outcomes assessed

The primary endpoint for the study was the occurrence of IFD, and the secondary endpoints were response to antifungal therapy and IFD-attributable mortality. Patients suspected of IFD were evaluated with high-resolution computed tomography, cultures of clinical specimens, and tests for detection of galactomannan and beta-1,3-D-glucan. A diagnosis of proven or probable IFD was determined by the revised European Organization for Research and Treatment of Cancer and Mycoses Study Group (EORTC/MSG) criteria [11]; only proven and probable cases were included in the determination of the cumulative incidence of IFD.

The day of diagnosis was the day on which the first diagnostic microbiological test was performed. The timing of IFD following HSCT was classified as early (≤40 days) or late (>40 days), and the infection was classified as non-disseminated or disseminated based on evidence of an infection in at least two non-contiguous sites.

The response to therapy at 12 weeks after diagnosis was based on a previously published report [12]. Complete response (CR) means resolution of all attributable symptoms and signs of disease and radiological abnormalities, and mycological evidence of eradication of disease. Partial response (PR) means improvement in attributable symptoms and signs of disease and radiological abnormalities, and evidence of clearance of cultures or reduction of fungal burden. A CR and PR were regarded as successful treatment, while a stable response, disease progression and death were regarded as a poor response. Invasive fungal disease was regarded as the attributable cause of death when the patient died of progressive organ failure in the absence of other co-morbidities believed to have caused death.

Underlying disease was classified as high-risk or standard-risk. High-risk disease included acute leukaemia either in non-remission or in the third or greater complete remission, and chronic myeloid leukaemia in the blast phase; standard-risk disease included all other diagnoses.

Statistical analysis

In calculating the cumulative incidence of IFD, death before the occurrence of IFD was treated as a competing risk. Multivariate analyses were performed to determine risk factors associated with the occurrence of IFD, including the following: gender, age, underlying disease (standard-risk vs. high-risk), the haematopoietic cell transplantation-specific co-morbidity index (HCT-CI), number of previous chemotherapy regimens, donor type (HLA-matched sibling vs. haploidentical donors), prior IFD history (with or without), prophylaxis for IFD (fluconazole vs. other drugs), acute GVHD (grade 0–II vs. grade III–IV), and chronic GVHD (non or limited vs. extensive). The HCT-CI was determined according to the Seattle scale [13]. Variables with p-values <0.20 in univariate models were entered into Cox regression models, and two-sided tests were considered to be significant at p-values of 0.05.

Logistic regression was performed to determine factors associated with response rate to antifungal therapy and attributable mortality in patients with IFD. Variables tested in the analysis included gender, underlying disease (standard-risk vs. high-risk), donor type (MSD vs. HID), prior IFD history, time of IFD (early vs. late), proven vs. probable, disseminated vs. non-disseminated, concurrent acute GVHD, concurrent chronic GVHD, neutropenia (defined as 2 days with count of <500 neutrophils/μL), monocytopenia (defined as 2 days with count of <100 monocytes/μL) and lymphopenia (defined as 2 days with a count of <300 lymphocytes/μL). The mortality attributable to IFD was estimated with cumulative incidence curves in which deaths due to causes other than IFD were considered to be competing risk events.

Results

A total of 1042 patients were enrolled, including 390 patients who received MSD transplantation and 652 patients who received HID transplantation. The median follow-up for survivors was 1030 days (range 367–1817). The characteristics of all HSCT recipients are summarized in Table 1. Most patients (71.7%) were diagnosed with acute leukaemia, and almost all (94.8%) received a combination graft from granulocyte colony-stimulating factor (G-CSF)-primed bone marrow and peripheral blood stem cells. About 176 (16.9%) patients had a prior history of IFD. About 802 (77.0%) patients received antifungal prophylaxis with oral fluconazole (200 mg/day), 14 (1.3%) patients received voriconazole, 17 (1.6%) received amphotericin B, 126 (12.1%) received itraconazole injections, 5 (0.5%) received caspofungin, 37 (3.6%) received micafungin and 41 (3.9%) patients received itraconazole oral solution.

Table 1. Patient characteristics
 Haploidentical (= 652)Matched-sibling (= 390)p Value
  1. AML, acute myeloid leukaemia; ALL, acute lymphoblastic leukaemia; CML, chronic myelogenous leukaemia; MDS, myelodysplastic syndrome; AA, aplastic anaemia; MM, multiple myeloma; NHL, non-Hodgkin's lymphoma; HCT-CI, haematopoietic cell transplantation-specific co-morbidity index; HLA, human leukocyte antigen; BU, busulfan; CY, cyclophosphamide; FLU, fludarabine; ATG, antithymocyte globulin; TBI, total body irradiation; BM, bone marrow; PB, peripheral blood; IFD, invasive fungal disease; aGVHD, acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease.

  2. Data shown are numbers of patients and percentages, unless otherwise stated.

Age, years (median, range)25 (3–57)38 (5–59)<0.001
Gender, male (%)404 (61.9)247 (63.3)0.692
Underlying disease (%)
AML280 (42.9)161 (41.3) 
ALL233 (35.7)73 (18.7)
CML75 (11.5)93 (23.8)
MDS40 (6.2)35 (9.0)
AA11 (1.7)15 (3.8)
MM2 (0.3)1 (0.3)
NHL10 (1.5)10 (2.6)
Other1 (0.2)2 (0.5)
High-risk disease (%)98 (15.0)54 (13.8)0.17
Previous chemotherapy regimen (median, range)4 (0–8)4 (0–29)0.92
HCT-CI (%)
0–1537 (82.4)312 (80)0.20
≥2115 (17.6)78 (20)
Mismatched HLA locus (%)
3374 (57.4)0 (0)<0.001
2214 (32.8)0 (0)
159 (9.0%)0 (0%)
05 (0.8%)390 (100%)
Conditioning regimen (%)
BU/CY0 (0)312 (80.0) 
BU/CY + ATG649 (99.5)0 (0)
CY/ATG0 (0)10 (2.6)
BU/FLU0 (0)65 (16.7)
TBI-based1 (0.2)3 (0.7)
Other2 (0.3)0 (0)
Graft source (%)
BM + PB642 (98.5)346 (88.7) 
BM1 (0.2)0 (0)
PB9 (1.3)44 (11.3)
Prior IFD history (%)106 (16.3)70 (17.9)0.267
Fluconazole for IFD prevention (%)504 (77.3)298 (76.4)0.290
Time of neutrophil engraftment, days (median, range)13 (8–31)16 (9–25)<0.01
Time of platelet engraftment, days (median, range)16 (7–195)13 (7–120)<0.01
Time to onset of aGVHD, days (median, range)28 (7–100)39 (15–105)0.14
aGVHD grade III–IV (%)79/643 (12.3)19/388 (4.6)<0.01
Time to onset of cGVHD, days (median, range)150 (86–581)150 (80–760)0.85
Extensive cGVHD (%)136/601 (22.7)119/373 (32.0)0.14

Comparison of IFD by type of transplantation

A total of 61 (5.8%) patients were diagnosed with IFD, including 14 (23.0%) proven cases and 47 (77.0%) probable cases. The 1-year cumulative incidence of IFD in the HID group was significantly higher than that in the MSD group (7.1% vs. 3.3%; p 0.007) (Fig. 1). The incidence of IFD in the HID group and the MSD group at day 40 was 4.3% and 1.0%, respectively, and at day 100 was 5.2% and 1.3%, respectively. IFD occured earlier in patients who underwent HID transplantation (23 days vs. 141.5 days; p 0.04).

Figure 1.

Incidence of IFD after allogeneic stem cell transplantation. The incidence of IFD after haploidentical transplantation was significantly higher than after HLA-matched sibling transplantation (p <0.01).

The ratio of proven to probable cases was not different between the two transplantation types (MSD, 1/13; HID, 13/47; p 0.156). Aspergillus was the major pathogen identified in IFD in both MSD transplantation and HID transplantation, and the proportion of Aspergillus infections showed no difference between the two transplantation types (80% (4/5) vs. 65.2% (15/23); p 0.886). However, it is noteworthy that the two Mucormycosis infections that occurred were after haploidentical transplantation (Table 2).

Table 2. Fungi identified in patients with IFD
 Early IFDLate IFD
MSDHIDMSDHID
  1. MSD, matched sibling transplantation; HID, haploidentical transplantation.

  2. a

    Mucormycosis.

Candida
 Candida albicans 0300
 Candida glabrata 0200
 Candida krusei 0101
 Candida tropicalis 0110
Aspergillus
 Aspergillus fumigatus 0131
 Aspergillus flavus 0001
 Aspergillus species 0416
Othera0002

Although not statistically significant, there was a trend for more disseminated infections in patients receiving HID transplantation compared with MSD transplantation (14/47 (29.8%) vs. 2/14 (14.3%), respectively; p 0.318). Nine (19.1%) of the 47 cases of IFD in the HID group involved the central nervous system (CNS), but none of the 14 IFD cases in the MSD group had an infection in the CNS.

Risk factors for invasive fungal disease

Univariate analysis revealed that high-risk underlying disease, HID transplantation, delayed neutrophil recovery, delayed platelet recovery, acute GVHD grades III to IV and extensive chronic GVHD were associated with an increased risk of IFD (Table 3). Acute GVHD grades III to IV (HR = 2.214; 95% CI, 1.139–4.304; p 0.019), extensive chronic GVHD (HR = 2.413; 95% CI, 1.377–4.228; p 0.002) and HID transplantation (HR = 2.648; 95% CI, 1.111–6.310; p 0.028) remained significant in the multivariable model (Table 3).

Table 3. Risk factors associated with IFD after HSCT
VariablesIFDUnivariate analysis (p value)Multivariate analysis
Yes (= 61)No (= 981)p valueHR (95% CI)
  1. IFD, invasive fungal disease; HSCT, haematopoietic stem cell transplantation; HR, hazard ratio; CI, confidence interval; HCT-CI, haematopoietic cell transplantation-specific co-morbidity index; HID, haploidentical donors; aGVHD, acute graft-versus-host disease; cGVHD, chronic graft-versus-host disease.

Age, years (median)30300.693
Gender (male/female)43/18608/3730.340
High-risk disease (%)15 (24.6)137 (13.9)0.0120.262
Number of previous chemotherapy regimens (median, range)4 (0–8)4 (0–29)0.919
HCT-CI <2, %88.5%85.0%0.20
Prior IFD history (%)12 (19.7)164 (16.7)0.639
HID (%)47 (77.0)605 (61.7)<0.0010.0282.648 (1.111–6.310)
Prophylaxis with fluconazole (%)46 (75.4)756 (77.1)0.942
Time of neutrophil engraftment, days (median, range)15 (9–31)14 (8–31)0.0290.094
Time of platelet engraftment, days (median, range)18 (9–188)14 (7–195)<0.0010.261
Grade III–IV aGVHD (%)10 (16.4)88 (8.9)0.0150.0192.214 (1.139–4.304)
Extensive cGVHD (%)21 (34.4)179 (18.2)0.0140.0022.413 (1.377–4.228)

When early IFD and late IFD were analysed separately, HID transplantation (HR = 4.219; 95% CI, 1.473–12.083; p 0.007) was identified as the only risk factor for early IFD in multivariable analysis, while extensive chronic GVHD (HR = 3.022; 95% CI, 1.437–6.354; p 0.004) was identified as the only risk factor for late IFD in multivariable analysis.

Outcome of invasive fungal disease

The median time from IFD diagnosis to the last follow-up was 231 days (range 0–1,463). At the last follow-up, only 24 (39.3%) patients were still alive. Of the 37 patients who died, 22 of these deaths could be attributed to IFD. The mortality rate attributed to IFD at 1 year after diagnosis was 30.6%. The response to antifungal therapy at the 12th week after diagnosis was 55.7% (34/61 patients), including a CR of 18.0% (11/61 patients) and a PR of 37.7% (23/61 patients).

When the outcomes of IFD for MSD and HID recipients were compared, no significant differences were found for IFD-attributable mortality and response to antifungal therapy. The IFD-attributable mortality in MSD and HID patients was 30.0% (4/14 patients) and 30.9% (18/47 patients), respectively (p 0.905) (Fig. 2). The response to antifungal therapy in MSD and HID patients was 50.0% (7/14 patients) and 57.4% (27/47 patients), respectively (p 0.761).

Figure 2.

The 1-year attributable mortality rates for IFD in patients receiving HLA-matched sibling transplantation or haploidentical transplantation were similar (30.0% and 30.9%, respectively; p 0.905).

However, in the logistic regression analysis, disseminated infection was associated with a high attributable mortality (p 0.019) and a poor response to antifungal therapy (p 0.043).

Discussion

This is the first study to directly compare IFD incidence and characteristics after HLA-matched transplantation and haploidentical transplantation. The large patient sample and the highly homogeneous population give reliability to the study's findings. The important findings include: (i) unmanipulated haploidentical transplantation has a higher incidence of IFD compared with HLA-matched sibling transplantation; (ii) IFD occurred earlier in haploidentical transplantation recipients but there was no difference in the infection pattern between the two types; and (iii) the prognosis of IFD was not associated with the type of transplantation.

Some multicentre studies on different transplantation donors have been published previously: the TRANSNET data, the PATH study and the SEIFEM-2004B study [14-16]. However, the populations and transplantation regimens were highly heterogeneous, and only a minor percentage of the patients received haploidentical transplantation. Furthermore, all of these studies used the 2002 EORTC criteria for IFD. In a retrospective analysis, 81% of the possible cases and 75% of the probable cases of IFD defined by the EORTC/MSG 2002 criteria were categorized as non-classifiable when the 2008 EORTC/MSG definitions were used [17]. This makes comparing data from studies using the 2002 vs. the 2008 criteria very difficult. Our previous report reviewed 291 patients who received an unmanipulated haploidentical transplantation using the 2002 criteria, and 13.1% of these patients had a fungal infection. When using the 2008 criteria, the incidence of IFD after haploidentical transplantation was only 7.1% after 1 year. The incidence did not decrease; rather, it was the impact of the different criteria.

Our results suggest that fungal-specific immune recovery is delayed with unmanipulated haploidentical HSCT in comparison with HLA-matched transplantation. It is known that CD4+ T cells play an important protective role in the host antifungal reaction [18], although the population of CD4+ T cells is not proportional to the fungal-specific immunity [10]. Our published data [19] suggest that the immune reconstitution of CD4+ T cells after non-TCD haploidentical transplantation is significantly slower than after HLA-matched transplantation in the first 90 days.

A further consideration is that the antithymocyte globulin (ATG) used in haploidentical transplantation can extensively deplete CD3+ T cells, and this depletion can last for a long time. Data from the study by Zhang et al. showed that in the setting of unmanipulated haploidentical transplantation, the serum ATG concentration is higher than the minimum effective concentration until 90 days after dosing [20]. This is perhaps another factor associated with the higher rate and earlier occurrence of IFD after haploidentical transplantation. However, this hypothesis needs more direct evidence.

Haploidentical vs. HLA-matched sibling transplantation was not significantly associated with response to antifungal therapy and IFD-attributed mortality. Rather, the characteristics of IFD, such as the infection type (disseminated infection vs. non-disseminated infection), grade of diagnosis (proven vs. probable) and time of infection (early vs. late) had an impact on the IFD prognosis. The PATH study showed that patients with IFD after HLA-matched transplantation had the same prognosis as patients with IFD after transplantation from an alternative donor [15], but only 5% of the population received haploidentical transplantation and the study did not compare the two transplantation types directly. Other studies suggest that patients infected with IFD after an HSCT from an alternative donor have a poor prognosis [16, 21, 22], but none of these studies performed a prospective comparison.

In conclusion, IFD infections in patients receiving an unmanipulated haploidentical HSCT occurred more frequently than in patients receiving HLA-matched HSCT. This reflects the difference in immune reconstitution between the two types of transplantation to some extent. These results demonstrate that more active prophylactic strategies should be adopted in the setting of HID transplantation.

Acknowledgements

We thank American Journal Experts (http://www.journalexperts.com) for their assistance in editing this manuscript.

Transparency Declaration

This work was supported in part by grants from the National Outstanding Young Scientist's Foundation of China (No. 30725038), the Special Research Fund of the Health Programme of the Ministry of Health of China (No. 200802027), the National Natural Science Foundation of China (No. 30725038), and the Beijing Medical Award Foundation (No. ZJGRYSBDRM-002). The authors declare no conflicts of interest.

Ancillary