Prognostic value of pretreatment inflammatory biomarkers in advanced lung adenocarcinoma patients receiving first-line pemetrexed/platinum doublet

Abstract

Inflammation is a new hallmark feature of cancer initiation and progression. We aimed to investigate the association between inflammatory response biomarkers and progression-free survival and overall survival in advanced lung adenocarcinoma patients treated with first-line pemetrexed and platinum doublet chemotherapy. Patients hospitalized between April 2012 and March 2015 were enrolled and eliminated according to the inclusion and exclusion criteria. The pretreatment neutrophil-to-lymphocyte ratio, lymphocyte-to-monocyte ratio, derived neutrophil-to-lymphocyte ratio, and platelet-to-lymphocyte ratio were calculated. Besides the well-established clinical prognostic factors, the prognostic values of the four markers were evaluated by the Kaplan–Meier method and Cox’s proportional hazards regression model. A total of 78 patients were enrolled in this study. Elevated neutrophil-to-lymphocyte ratio and derived neutrophil-to-lymphocyte ratio were correlated with poor treatment response (p = 0.014, 0.012, respectively). A high pretreatment neutrophil-to-lymphocyte ratio, derived neutrophil-to-lymphocyte ratio, and platelet-to-lymphocyte ratio, as well as low lymphocyte-to-monocyte ratio, were associated with worse progression-free survival and overall survival. Multivariate analysis revealed that high neutrophil-to-lymphocyte ratio (hazard ratio = 2.056; 95% confidence interval, 1.281–3.299; p = 0.003) and ≥3 metastasis organs (hazard ratio = 1.989; 95% confidence interval, 1.069–3.702; p = 0.030) were independent prognostic factors for progression-free survival. Meanwhile, high neutrophil-to-lymphocyte ratio (hazard ratio = 5.540; 95% confidence interval, 2.974–10.321; p < 0.001) and habitual smoking (hazard ratio = 2.806; 95% confidence interval, 1.509–5.221; p = 0.001) were independent prognostic factors for overall survival. In conclusion, Pretreatment neutrophil-to-lymphocyte ratio was an independent prognostic factor for advanced lung adenocarcinoma patients treated with first-line pemetrexed/platinum doublet chemotherapy. Elevated pretreatment derived neutrophil-to-lymphocyte ratio and neutrophil-to-lymphocyte ratio might be potential biomarkers for poorer responses to chemotherapy. To verify these findings, larger well-designed prospective studies are needed.

Introduction

Lung cancer is the most common type of cancer and the leading cause of cancer-related death in China.¹ Non–small cell lung cancers (NSCLCs) account for 85%–90% of lung cancers.² Increasing evidence has indicated that inflammation is a new hallmark feature of cancer initiation and progression.³ Analogous to local inflammation, systemic inflammation involves immune cells, cytokines, and small inflammatory proteins, and contributes to patient outcomes. However, systemic inflammation can be characterized by changes of several peripheral blood parameters.⁴ Some hemocyte count–based biomarkers, such as the neutrophil-to-lymphocyte ratio (NLR),^4–6 lymphocyte-to-monocyte ratio (LMR),^6,7 derived neutrophil-to-lymphocyte ratio (dNLR),^8–10 and platelet-to-lymphocyte ratio (PLR),^11,12 have been approved as prognostic indicator in a variety of solid tumors. Except for dNLR, the other indicators have been investigated in NSCLC.^5,7,12

Lung adenocarcinoma is the most common histological type of lung cancer worldwide, accounting for more than 50% of NSCLC.^13,14 Pemetrexed-based combination chemotherapy has shown a slight but significant survival^15,16 in lung adenocarcinoma. However, previous studies^5,7,12,17 enrolled different histological subtypes (including squamous cell carcinoma and adenocarcinoma) and different first-line treatment strategies (any platinum-based doublets with a third-generation agent).

To avoid these mixtures, this study aimed to investigate the prognostic role of inflammatory biomarkers (NLR, LMR, dNLR, and PLR) in advanced lung adenocarcinoma patients who received first-line pemetrexed and platinum doublet.

Patients and methods

Patients enrollment

We retrospectively reviewed 1162 adult inpatients with confirmed advanced lung adenocarcinoma at Tianjin Medical University Cancer Institute and Hospital between April 2012 and March 2015. The study proposal had been approved by the Institutional Review Committee on Human Research.

The inclusion criteria for patients were as follows: (1) histological and/or cytological diagnosis of primary lung adenocarcinoma, (2) stage IIIB or IV according to the tumor–node–metastasis (TNM) criteria of NSCLC (American Joint Committee on Cancer, seventh edition), (3) previously untreated, (4) Eastern Cooperative Oncology Group (ECOG) score between 0 and 1, (5) at least two cycles of first-line pemetrexed and platinum chemotherapy with a response evaluation after treatment, and (6) all clinical data available.

Patients were excluded according to the following criteria: (1) not primary lung cancer patients, (2) previous or concomitant malignancies at other sites, (3) lack of detailed and required clinical data, (4) combined treatment with other therapeutic anticancer drugs, (5) switched maintenance therapy after 4–6 cycles, (6) clinical evidence of infection or other inflammation within 1 month, (7) history of autoimmune disease, (8) therapy with steroids, and (9) lost contact during the follow-up time.

Clinical and laboratory data collection

Age, gender, smoking status, metastasis sites, molecular typing, and complete blood cell counts were collected from patients’ electronic medical records. The TNM stage assessment was based on computed tomography (CT) scans of the thorax and upper abdomen, magnetic resonance imaging (MRI) or CT scans of the brain, and bone emission CT scans. Therapy response evaluation by whole body tumor scans was conducted every two cycles. The Response Evaluation Criteria in Solid Tumors (RECIST; version 1.1) was applied to evaluate the response.¹⁸ Treatment was continued until evidence of disease progression on scans, unacceptable adverse events, or death occurred.

Progression-free survival (PFS) was the time from the beginning of chemotherapy until disease progression. Overall survival (OS) was the time from the date of diagnosis until the date of death by any cause. Patients without tumor progression or death at the time of the data cut-off (March 31, 2016) for the analysis were censored.

The hematological parameters were obtained within a week before chemotherapy. Pretreatment NLR, LMR, dNLR, and PLR were calculated from peripheral complete blood cell counts. These parameters were defined as follows: NLR = neutrophil-to-lymphocyte ratio, LMR = lymphocyte-to-monocyte ratio, dNLR = neutrophil to (white cell count (WBC) − neutrophil count) ratio, and PLR = platelet-to-lymphocyte ratio.

Statistical analysis

All statistical analyses were performed with SPSS Statistics 18 (IBM Corporation, NY, USA). Data were presented as the number of subjects and the median value or mean ± standard deviation. Associations of NLR, LMR, dNLR, and PLR with other categorized clinical variables were determined by the two-sample t-test or one-way analysis of variance (ANOVA) when the Kolmogorov–Smirnov test revealed a normal distribution of NLR, LMR, dNLR, and PLR. The non-parametric Wilcoxon rank–sum test was used when the distribution agreed with the non-normal distribution. The data were shown as the median and interquartile range.

The optimal cut-off values of NLR, LMR, dNLR, and PLR were estimated by receiver operating characteristic (ROC) curve analysis. Survival curves on categorical variables were estimated by Kaplan–Meier analysis. The log–rank test was utilized to examine differences in survival distributions between groups.

The Cox’s proportional hazards regression model was applied to perform univariate and multivariate analyses. Variables with a p < 0.05 by univariate analyses were included in a subsequent multivariate analysis. All statistical tests used in this study were two-sided; values with p < 0.05 were considered significant.

Results

Patients’ baseline characteristics

According to the criteria, only 11.4% (132 out of 1162) lung adenocarcinoma patients were included, that is, stage IIIB and IV patients who received at least two cycles of first-line pemetrexed and platinum chemotherapy. Finally, 78 patients were enrolled in the study after referring to the exclusion criteria. The following patients were excluded: 1 patient who previously had breast cancer, 2 patients who could not provide pretreatment blood cell counts, 3 patients treated with bevacizumab simultaneously, 6 patients treated with tyrosine kinase inhibitor (TKI) concurrently, 2 patients who had clinical evidence of pneumonia, 1 patient who was given glucocorticoid, 20 patients who switched to TKI as maintenance therapy, and 19 patients who were lost to follow-up. The baseline characteristics are presented in Table 1.

Table 1.

Patients’ baseline characteristics.

Characteristics	n (%)a (overall 78)
Age (years)
Median (min–max)	59 (28–82)
Gender
Male/female	36 (46.2%)/42 (53.8%)
Smoking status
Yes/no	41 (52.6%)/37 (47.4%)
Metastasis sites
Lung/pleura/bone/brain	28 (40.6%)/29 (42.0%)/38 (55.1%)/11 (15.9%)
Adrenal gland/liver/other	12 (17.4%)/5 (7.2%)/6 (8.7%)
Metastasis stage
M0/M1a/M1b	9 (11.5%)/19 (24.4%)/50 (64.1%)
CEA level (µg/L)
Median (min–max)	9.45 (0.2–1000)
Molecular typing
Negative/unknown	13 (16.7%)/57 (73.1%)
Sensitizing EGFR mutation	1 (1.3%)
EGFR exon 20 insert	3 (3.8%)
KRAS mutation	3 (3.8%)
ALK rearrangement	1 (1.3%)
Clinical response
Partial response (PR)	21 (26.9%)
Stable disease (SD)	43 (55.1%)
Progressive disease (PD)	14 (18.0%)
PFS (days)
Median (min–max)	159 (50–609)
OS (days)
Median (min–max)	458 (52–1128)

CEA: carcinoembryonic antigen; EGFR: epidermal growth factor receptor; ALK: anaplastic lymphoma kinase; PFS: progression-free survival; OS: overall survival.

a

Unless otherwise stated.

Optimal cut-offs for NLR, LMR, dNLR, and PLR

The ROC curves are presented in Figure 1 (detailed results in Table 2), with OS as the end-point for NLR, LMR, dNLR, and PLR. The areas under curve (AUC) for NLR, LMR, dNLR, and PLR were 0.765, 0.668, 0.747, and 0.622, respectively. The optimal cut-off values were 3.11 for NLR, 4.30 for LMR, 2.12 for dNLR, and 186 for PLR by ROC curves analysis. According to the optimal cut-off values, patients were divided into two groups, with the low group being less than the optimal cut-off values and the high group being greater than or equal to the optimal cut-off values.

Figure 1.

Optimal cut-off levels for (a) NLR, (b) LMR, (c) dNLR, and (d) PLR were applied with the ROC curves for overall survival (OS).

[Figure omitted. See PDF]

Table 2.

Results of ROC curves.

	NLR	LMR	dNLR	PLR
Sensitivity	0.604	0.771	0.625	0.542
Specificity	0.867	0.600	0.833	0.700
Positive predictive value	0.879	0.755	0.857	0.743
Negative predictive value	0.578	0.621	0.581	0.488
Positive likelihood ratio	4.531	1.927	3.750	1.806
Negative likelihood ratio	0.457	0.382	0.450	0.655
Diagnostic odds ratio	9.921	5.045	8.333	2.758

ROC: receiver operating characteristic; NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; dNLR: derived neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio.

Associations of NLR, LMR, dNLR, and PLR with other clinical characteristics

Potential relationships between NLR, LMR, dNLR, and PLR with the other clinical characteristics are presented in Table 3. NLR was significantly higher in males than in females, whereas LMR was lower (p = 0.035 for both). Both NLR and dNLR were elevated in the progressive disease (PD) group compared with the disease-controlled group (partial response (PR) + stable disease (SD), with p = 0.014 and p = 0.012, respectively).

Table 3.

Associations of NLR, LMR, dNLR, and PLR with other clinical factors.

Factor	Number	NLR	p	LMR	p	dNLR	p	PLR	p
Total	78	3.35 ± 1.69		4.16 ± 2.00		2.26 ± 0.97		185.90 ± 82.22
Gender
Male	36	3.80 ± 2.05	0.035*	3.65 ± 1.70	0.035*	2.42 ± 1.16	0.170	188.86 ± 73.03	0.771
Female	42	2.96 ± 1.22		4.60 ± 2.15		2.12 ± 0.76		183.36 ± 90.17
Age
<60	40	3.24 ± 1.85	0.539	4.29 ± 1.81	0.556	2.21 ± 1.09	0.680	182.93 ± 83.84	0.745
≥60	38	3.47 ± 1.52		4.02 ± 2.19		2.30 ± 0.84		189.03 ± 81.49
Smoking status
No	37	3.18 ± 1.71	0.385	4.61 ± 2.01	0.058	2.27 ± 1.12	0.909	186.78 ± 93.76	0.929
Yes	41	3.51 ± 1.69		3.76 ± 1.92		2.24 ± 0.83		185.10 ± 71.42
Number of metastasis organs
0-1	39	3.45 ± 1.97	0.868	3.98 ± 1.80	0.725	2.30 ± 1.13	0.917	191.92 ± 96.79	0.743
2	23	3.23 ± 1.38		4.35 ± 2.25		2.23 ± 0.85		175.09 ± 61.48
≥3	16	3.28 ± 1.44		4.34 ± 2.17		2.19 ± 0.72		186.76 ± 71.83
Metastasis stage
M0	9	3.58 ± 2.20	0.856	4.35 ± 2.82	0.952	2.24 ± 1.25	0.667	195.61 ± 112.96	0.773
M1a	19	3.20 ± 1.85		4.17 ± 1.39		2.09 ± 0.82		174.66 ± 55.67
M1b	50	3.37 ± 1.56		4.13 ± 2.06		2.32 ± 0.98		188.42 ± 85.62
Clinical response
PR + SD	64	2.80 (2.18–3.44)	0.014*	4.34 ± 2.02	0.088	1.96 (1.63–2.39)	0.012*	181.15 ± 78.27	0.278
PD	14	4.15 (3.29–5.40)		3.34 ± 1.70		2.79 (2.24–3.41)		207.61 ± 98.70

NLR: neutrophil-to-lymphocyte ratio; LMR: lymphocyte-to-monocyte ratio; dNLR: derived neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; PR: partial response; SD: stable disease; PD: progressive disease.

*

Statistically significant.

Associations of NLR and LMR with OS between males and females

According to the former description, NLR was higher, whereas LMR was lower in males than in females. For further understanding, Kaplan–Meier analysis and log-rank tests were performed based on the OS (Figure 2(a)–(d)). High NLR (≥3.11) was significantly associated with worse OS in males and females (p < 0.001 and p = 0.002, respectively). Low LMR (<4.30) was associated with worse OS in males (p = 0.006) but not in females (p = 0.090).

Figure 2.

Kaplan–Meier curves for OS according to pretreatment NLR between (a) males and (b) females; Kaplan–Meier curves for OS according to pretreatment LMR between (c) males and (d) females.

[Figure omitted. See PDF]

Associations of NLR, LMR, dNLR, and PLR with PFS and OS

To evaluate the associations, Kaplan–Meier analysis and log-rank tests were performed based on the PFS and OS. Decreased PFS (Figure 3(a)–(d)) and OS (Figure 4(a)–(d)) were significantly associated with high NLR (≥3.11; p = 0.004 and p < 0.001, respectively), low LMR (<4.30; p = 0.013 and p = 0.001, respectively), high dNLR (≥2.12; p = 0.008 and p < 0.001, respectively), and high PLR (≥186; p = 0.031 and p = 0.005, respectively).

Figure 3.

Kaplan–Meier curves for PFS according to pretreatment (a) NLR, (b) LMR, (c) dNLR, and (d) PLR.

[Figure omitted. See PDF]

Figure 4.

Kaplan–Meier curves for OS according to pretreatment (a) NLR, (b) LMR, (c) dNLR, and (d) PLR.

[Figure omitted. See PDF]

Inflammatory response biomarkers and clinical variables for the prediction of PFS and OS were further investigated by univariate analysis with Cox regression model, respectively. Results of the univariate analysis indicated that gender, smoking status, NLR, LMR, dNLR, and PLR were prognostic predictors of both PFS (Table 4) and OS (Table 5). Moreover, ≥3 metastatic organs and PD response were poor prognostic factors of PFS and OS, respectively. Then the variables above were enrolled into a multivariate Cox proportional hazards model to adjust the effects of covariates. The statistical analysis data indicated that high NLR (≥3.11; hazard ratio (HR) = 2.056; 95% confidence interval (CI), 1.281–3.299; p = 0.003) and ≥3 metastatic organs (HR = 1.989; 95% CI, 1.069–3.702; p = 0.030) were independent prognostic factors for PFS (Table 4), while high NLR (≥3.11; HR = 5.540; 95% CI, 2.974–10.321; p < 0.001) and smoker (HR = 2.806; 95% CI, 1.509–5.221; p = 0.001) were associated with worse OS (Table 5).

Table 4.

Univariate and multivariate analyses of variables correlated to PFS.

Variables	Univariate analysis			Multivariate analysis
	HR	95% CI	p	HR	95% CI	p
Gender
Female	1
Male	1.728	1.071–2.789	0.025*	1.537	0.929–2.543	0.094
Age
<60	1
≥60	1.010	0.645–1.582	0.965
Smoking status
No	1			1
Yes	1.681	1.045–2.706	0.032*	1.240	0.724–2.121	0.433
No. of metastasis organs
0-1	1			1
2	0.760	0.451–1.279	0.301	0.809	0.476–1.374	0.433
≥3	2.020	1.107–3.688	0.022*	1.989	1.069–3.702	0.030*
Metastasis stage
M0	1
M1a	0.608	0.273–1.354	0.223
M1b	0.536	0.260–1.106	0.092
CEA level
Normal	1
High	0.931	0.579–1.496	0.767
NLR
<3.11	1			1
≥3.11	1.924	1.213–3.051	0.005*	2.056	1.281–3.299	0.003*
LMR
<4.30	1			1
≥4.30	0.557	0.348–0.890	0.015*	0.664	0.397–1.111	0.119
dNLR
<2.12	1			1
≥2.12	1.836	1.165–2.894	0.009*	1.183	0.542–2.581	0.647
PLR
<186	1			1
≥186	1.660	1.043–2.641	0.033*	1.261	0.749–2.124	0.382

PFS: progression-free survival; HR: hazard ratio; CI: confidence interval; CEA: carcinoembryonic antigen; LMR: lymphocyte-to-monocyte ratio; dNLR: derived neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio.

*

Statistically significant.

Table 5.

Univariate and multivariate analyses of variables correlated to OS.

Variables	Univariate analysis			Multivariate analysis
	HR	95% CI	p	HR	95% CI	p
Gender
Female	1			1
Male	2.097	1.173–3.747	0.012*	1.538	0.803–2.945	0.194
Age
<60	1
≥60	1.572	0.881–2.804	0.126
Smoking status
No	1			1
Yes	2.732	1.299–4.330	0.005*	2.806	1.509–5.221	0.001*
No. of metastasis organ
0-1	1
2	0.878	0.437–1765	0.715
≥3	1.565	0.775–3.159	0.212
Metastasis stage
M0	1
M1a	0.778	0.277–2.188	0.635
M1b	1.209	0.504–2.904	0.670
CEA level
Normal	1
High	0.838	0.461–1.524	0.562
Clinical response
PR + SD	1			1
PD	2.734	1.438–5.196	0.002*	1.405	0.673–2.929	0.365
NLR
<3.11	1			1
≥3.11	4.918	2.678–9.034	<0.001*	5.540	2.974–10.321	<0.001*
LMR
<4.30	1			1
≥4.30	0.327	0.162–0.658	0.002*	0.688	0.315–1.500	0.347
dNLR
<2.12	1			1
≥2.12	4.119	2.255–7.522	<0.001*	1.967	0.736–5.260	0.178
PLR
<186	1			1
≥186	2.281	1.269–4.098	0.006*	1.334	0.675–2.638	0.407

OS: overall survival; HR: hazard ratio; CI: confidence interval; LMR: lymphocyte-to-monocyte ratio; dNLR: derived neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio.

*

Statistically significant.

Discussion

In this study, high pretreatment NLR (≥3.11) and dNLR (≥2.12) were associated with the worsening response for first-line pemetrexed and platinum doublets chemotherapy-treated advanced lung adenocarcinoma patients for the first time. Males had higher pretreatment NLR and lower LMR than females. Patients with high NLR (≥3.11), low LMR (<4.30), high dNLR (≥2.12), and high PLR (≥186) had a shorter PFS and OS. Both in males and females, NLR was associated with OS. However, LMR was only associated with OS in males. Although LMR, dNLR, and PLR were significantly associated with survival in univariate analysis, after adjusting for the gender, age, smoking status, metastasis status, and CEA level, only NLR remained an independent factor associated with PFS and OS in multivariate survival analysis.

Previous clinical data suggested that NLR was associated with the clinical response to therapy and prognosis of advanced NSCLC. These studies all revealed that elevated pretreatment NLR was a predictive factor for worse therapy response with the first-line chemo- or targeted therapy.^17,19–21 Our study also indicated that high NLR was an independent prognostic factor for predicting worse response to first-line chemotherapy for advanced lung adenocarcinoma patients.

Cancers often induce elevated circulating neutrophils counts by tumor-derived or tumor-induced factors.²² Meantime, neutrophils from patients differ from their counterparts in healthy donors, showing lower inducible production of reactive oxygen species (ROS); reducing spontaneous apoptosis; increasing a number of immature neutrophils²³; promoting tumor cell motility, migration, invasion, and metastasis^22,24; and inhibiting T cell proliferation by releasing arginine-1.²⁵ While it has long been believed that the anti-tumor activity is mainly mediated by the lymphocyte-dependent cellular immune response. The presence of tumor-infiltrating lymphocytes is a positive prognostic marker in tumors, including NSCLC.^26,27 These might partially explain why an elevated NLR is associated with poor survival.

Recent clinical studies revealed that dNLR was associated with chemotherapy response^10,28 and prognosis in many advanced cancers, such as metastatic melanoma,⁸ metastatic castration-resistant prostate cancer,⁹ advanced biliary cancer,¹⁰ and advanced colorectal cancer.²⁹ The dNLR is composed of WBCs and neutrophils; this parameter has been proposed as an alternative to NLR for studies where only WBCs and neutrophils have been recorded.³⁰ Our study first demonstrated that dNLR could predict pemetrexed and platinum therapy outcomes in advanced lung adenocarcinoma patients; dNLR was also a prognostic factor for PFS and OS, but not an independent one. The use of (WBC − neutrophil count) in the denominator may be one of the reasons, because of broadly mixing lymphocytes and monocytes, with possible opposing effects in terms of predictive value.

Smoking is a prognostic factor in NSCLC.^31,32 We showed that smoking status was an independent prognostic factor for survival in advanced lung adenocarcinoma patients. Similar data have been shown in former studies.^33,34 Remarkably, significant differences are found in tumor driver genes’ mutational spectra and frequencies between smokers and nonsmokers in lung cancer patients.^35–37 Epidermal growth factor receptor (EGFR) mutations are common in lung adenocarcinomas of people who have never smoked. By contrast, KRAS mutations predict poor survival and resistance to EGFR-TKI; these mutations are more frequent among heavy smokers. Moreover, the incidence of EGFR mutations decreased with increasing number of years spent smoking.³⁸

This study has some limitations. First, this work is a single-institution and retrospective study. Thus, it is susceptible to bias in data selection. Second, a limited number of patients were included. Third, the mutational status of driven-genes for most patients (57/78) was unknown because of unavailable or insufficient samples for testing at the time of diagnosis. Nevertheless, our study is the first to show a connection between dNLR and advanced lung adenocarcinoma patients, thereby suggesting that dNLR and NLR could be the biomarkers for chemotherapy response. NLR was an independent prognostic factor. However, the clinical utility of dNLR and NLR still needs to be confirmed with prospective analysis.

Conclusion

This study was the first to demonstrate that pretreatment NLR was an independent prognostic factor for advanced lung adenocarcinoma patients receiving first-line pemetrexed/platinum doublet chemotherapy. Elevated pretreatment dNLR and NLR might be potential biomarkers for worse response to chemotherapy. These markers are easily detected and measured in routine clinical practice. Larger well-designed prospective studies are needed to verify these findings.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the Institutional Review Committee on Human Research of the Tianjin Medical University Cancer Institute and Hospital and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by the Natural Science Foundation of China (No. 81272221) and the National Key Technology R&D Program (2015BAI12B12).

Informed consent

For this type of study, formal consent is not required.