Natriuretic peptides be a “>Original Contribution

Contents lists available at ScienceDirect

American Journal of Emergency Medicine

journal homepage:

Could B-type natriuretic peptides be a biomarker for trauma brain injury? A systematic review and meta-analysis

Yuan Zhang ^a,1, Zhanpeng Feng ^b,1, Yun Bao ^b, Lizhi Zhou ^c, Binghui Qiu ^b,?

^a First Clinical Medical College, Southern Medical University, China

^b Department of neurosurgery, Nanfang Hospital, Southern Medical University, China

^c Department of Biostatistics, Southern Medical University, China

Introduction

Trauma brain injury (TBI), defined as an alteration in brain function, or other evidence of brain pathology, caused by an external force [1], is always a health concern all over the world because of its considerable morbidity and mortality [2,3]. According to the CDC, in the United States alone, for example, it was estimated that the incidence and deaths of TBI were 1,691,481 (576.8 per 100,000) and (17.6 per 100,000) respectively between 2002 and 2006 [4]. Currently, main diag- nostic tools of TBI generally accepted include Glasgow Coma Scale score [5,6] and CT scanning [7], but both of them have their own limita- tions, such as a little contribution to the diagnosis for moderate trau- matic brain injury (mTBI) [8,9]. Also, not only does CT scanning have low sensitivity to diffuse brain damage [10], but it will increase patients’ Economic burden as well [11,12]. Thus, more and more attention is paid to blood biomarker of TBI [13,14], but there is still no single molecule as a clinical diagnostic tool for TBI.

Among all kinds of biomarkers which seem to be applied to the diag- nosis of TBI, B-type Natriuretic Peptide and/or N-terminal frag- ment (NT-proBNP) may be promising because the results of blood BNP levels are available as quickly as only 15 min [15,16]. BNP, which was first found in porcine brain [17], has intense action of natriuresis, di- uresis, vasodilation and so on. PreproBNP, the initial molecule which is 134 amino acids in length, will become proBNP, 108 amino acids in length, after cleavage of a 26 amino acid signal sequences [18]. Then, proBNP is cleavaged into BNP and NT-proBNP, both of whom are mainly used to diagnosis for patients with heart failure in clinical [19]. Recently, a series of literatures show that Elevated BNP levels were found in pa- tients with cerebral disease, such as TBI [20], stroke [21] and brain tumor [22]. Moreover, T Garcia-Berrocoso and his colleagues [23] per- formed a meta-analysis of the relationship between BNP and mortality after stroke to reveal that BNP/NT-proBNP levels are higher in stroke pa- tients who died. These previous studies showed that elevated BNP levels may be involved in cerebral diseases, but to our knowledge, there is no previous meta-analysis of the relationship between blood BNP levels published before. Furthermore, there still existed controversial whether BNP/NT-proBNP would increase or decrease after TBI in different stud- ies. Therefore, we here firstly conducted a systematic review and

* Corresponding author.

E-mail address: [email protected] (B. Qiu).

¹ These authors contributed equally to this work.

meta-analysis to evaluate whether blood BNP/NT-proBNP levels could be biomarker after TBI.

Method

Search strategy

Pubmed, the Cochrane Library, EMBASE, WanFang (Chinese data- bases), China National Knowledge Infrastructure (Chinese databases) were searched from inception to July 25, 2016. Our protocol was regis- tered on the international prospective register of systematic reviews (http://www.crd.york.ac.uk/PROSPERO PROSPERO, registration num- ber: CRD42016049414).

Study inclusion and exclusion criteria

Strict inclusion and exclusion criteria were established before our meta-analysis. Original studies were included when meeting the fol- lowing three criteria: First, recruited patients were diagnosed with TBI according to a history of head trauma and/or CT/MRI; second, plasma/ serum concentration of BNP/NT-proBNP was measured; Third, the type of study design was randomized controlled trials, cohort studies or case-control study. Reviews, editorials, letters, case reports, non- human studies and studies without original data were excluded. Also, if the studies were brought into the repeated samples, we selected the largest number of subjects. In addition, those studies with a lack of con- trol groups were excluded.

Study selection

All literatures searched were imported into EndNote (version x7; Thomson Reuters, 2013), and the duplicates were excluded. Two prima- ry authors (Zhang Y and Zhou LZ) independently reviewed and evaluat- ed eligible studies, and then, if there was any disagreement, a third author (Bao Y) was consulted.

Data extraction

Data from eligible studies for meta-analysis were retrieved by two authors (Zhang Y and Zhou LZ) independently. What information we recorded was as follow: author, country, year of publish, the type of study design, number of cases and characteristics (mean age and sex),

http://dx.doi.org/10.1016/j.ajem.2017.05.051

0735-6757/(C) 2017

and the mean and SD of serum/plasma BNP/NT-proBNP level in cases and controls. In case of any discrepancy, three authors (Zhang Y, Zhou LZ and Bao Y) discussed until it was solved. If different blood BNP/NT- proBNP levels were provided in a study, we extracted the earliest one after TBI from that study.

Statistical analysis

The strength of relationship between BNP/NT-proBNP and patients after TBI was measured by Weighted mean differences (WMDs) and 95% confidence interval (CI). If WMD with its CI didn’t include zero and a t-test P value was b 0.05, we considered it significantly different. In our meta-analysis, the I² statistic and the chi-squared test were uti- lized to qualify the heterogeneity. The fixed-effects model was used if the heterogeneity was not significantly different (I² b 50%; P N 0.1); oth- erwise, we adopted the random-effects model. We performed a sensi- tivity analysis by removing studies one by one to assess whether meta estimates would be affected by single studies. The evaluation of publica- tion bias was Egger’s test, and we considered significant publication bias as a P value b 0.1. If publication bias existed, trim-and-fill test was ap- plied [24]. The Review Manager 5.3 and Stata 11.0 were employed for all statistical analyses.

Quality assessment

Since all studies we included were observational studies, we chose Newcastle-Ottawa Scale (NOS) to assess the quality of these studies [25]. For the assessment of observational studies, the instrument includ- ed three aspects: selection of cases and controls, comparability of cases and controls, and outcome or exposure. In our study, we considered ruling out other diseases as first confounding factor, and both age- and gender- matched between cases and controls as second confounding factor. In addition, to modify the exposure criterion, we referred to the

meta-analysis of Aleksovska et al. [26]. According to NOS, studies with more than five scores (ranging from zero to nine) were seen as relative- ly high-quality ones.

Results

Study selection

Through the search strategy showed in Fig. 1, 369 potentially rele- vant studies were identified and screened for retrieval. After screening on the basis of abstracts or titles, 310 studies were excluded because they were for non-human subjects, case reports, reviews, or not rele- vant to TBI or BNP/NT-proBNP. Then, after investigating the full manu- scripts of the left 59 studies in detail, we excluded 38 studies because these studies lacked control groups, didn’t provide with mean +- SD for cases and controls, or had similar or duplicated information com- pared with other studies. Thus, only 21 studies were included in the meta-analysis [27-47].

Data extraction

In the 21 studies included in the meta-analysis, we extracted the mean and SD for cases and controls of 9 studies [31,33-36,38, 42,44,46] directly, and calculated those of 1 study [27] according to its raw data. In 10 studies [28-30,32,37,39,41,43,45,47] which only reported subgroups of patients, pooled mean and SD were calculated according to the separate data; in one study [40] where mean and SD were only showed in a histogram, we used Engauge-Digitizer for data extraction. In addition, in one study

[30] where only mean and SE were provided, we calculated SD by the formula: SD = SE x ? n.

Fig. 1. Flow-chart of literature searching.

Y. Zhang et al. / American Journal of Emergency Medicine 35 (2017) 1695-1701

1697

Table 1

Characteristics of studies included our meta-analysis

Ref.

Author, publication

Country

Males, %

Age, y

Biomarker

Method of

Serum/plasma

Admission

Blood collection

Biomarker levels

Number

year

Cases

Controls

Cases

Controls

measurement

time

time

Cases

Controls

Cases

Controls

[27]

Bekir AKGUN, 2012

Turkey

66.7

50

34.8 +- 24.9

35.3 +- 17.4

NT-proBNP

CL

Serum

NA

24 h

262.1 +- 338.4

28.1 +- 9.8

30

10

[28]

Chen Caijing, 2011

China

45.8

NA

NA

NA

NT-proBNP

LFIA

Serum

b16 h

24 h

293.5 +- 121.8

29 +- 12

48

10

[29]

Chen Ping, 2013

China

66.7

60

42.33(23-69)

40.38(26-65)

BNP

CL

Plasma

b12 h

Adm.

180.2 +- 126.7

27.8 +- 15.9

78

20

[30]

Chlodwig Kirchhoff, 2006

Germany

71.4

NA

NA

NA

NT-proBNP

ECL

Serum

b90 min

Adm.

97 +- 36

148 +- 21

14

10

[31]

Du Runjun, 2014

China

75.6

60

32.1 +- 11.5

31.9 +- 11.8

BNP

CL

Plasma

NA

Adm.

217.19 +- 89.23

54.6 +- 21.95

82

40

[32]

Fang Zhicheng, 2010

China

74

65

51.2 +- 3.6

54.3 +- 4.1

BNP

CL

Plasma

NA

24 h

161.2 +- 5.9

167.2 +- 1.2

73

40

[33]

Li Sumei, 2015

China

80

73.3

43.04 +- 11.28

44.2 +- 16.5

NT-proBNP

ELFA

Plasma

b24 h

24 h

451.38 +- 211.9

75.54 +- 30.549

69

30

[34]

Lin Zexiong, 2015

China

70

64.3

45.7 +- 11.3

43.6 +- 9.7

BNP

ELISA

Plasma

b24 h

Adm.

183.84 +- 48.55

73.26 +- 24.26

69

70

[35]

Liu Yuanming, 2016

China

61

59.8

60.5 +- 1.5

59.5 +- 1.8

BNP

CL

Serum

NA

24 h

187.84 +- 33.98

60.16 +- 22.02

82

82

[36]

Liu Yuliang, 2013

China

88.2

80

NA

NA

BNP

RIA

Serum

NA

NA

12.4 +- 1.1

25.2 +- 1.5

34

35

[37]

Peng Lihui, 2009

China

73.1

60

38.7 (15-68)

39.6 (17-68)

BNP

RIA

Serum

b24 h

Adm.

18.1 +- 2.7

26.6 +- 1.8

134

46

[38]

Peng Shuaiqun, 2015

China

58.3

NA

40.8 +- 5.2

39.8 +- 4.6

BNP

RIA

Serum

NA

NA

17.8 +- 0.8

24.6 +- 1.8

108

27

[39]

Wang Yifeng, 2015

China

68.5

65

44.4 +- 10.2

44.4 +- 10.2

BNP

ELISA

Plasma

b24 h

NA

174.18 +- 51.98

72.43 +- 23.17

93

60

[40]

Wen Tie, 2013

China

64.3

65

39.2 +- 11.5

37.9 +- 10.2

NT-proBNP

ECL

Plasma

b24 h

b1 h

179.4 +- 229.8

24.9 +- 7.9

28

20

[41]

Xiong Xuehui, 2012

China

51.7

40

40.72 +- 14.72

42.17 +- 14.19

BNP

ECL

Plasma

b24 h

24 h

54.6 +- 21.95

217.19 +- 89.23

41

41

[42]

Xu Jun, 2010

China

61

56.1

35.9 +- 18.9

36.7 +- 17.5

BNP

CL

Plasma

NA

Adm.

217.19 +- 89.23

54.6 +- 21.95

41

41

[43]

Zhang Danfeng, 2014

China

79.7

80

51.5 +- 12.6

55 +- 14

BNP

CL

Plasma

NA

24 h

131.9 +- 27.1

89.8 +- 13.5

69

30

[44]

Zhang Junfeng, 2015

China

60

57.5

48.6 +- 7.4

47.9 +- 8.2

BNP

RIA

Serum

NA

Adm.

16.872 +- 2.39

26.98 +- 1.98

80

80

[45]

Zhang Wenchuan, 1996

China

75

50

27.8(4-60)

NA

BNP

RIA

Serum/plasma

b24 h

Adm.

18.54 +- 0.91

24.64 +- 1.93

68

24

[46]

Zhao Fei, 2015

China

63.3

60

68.6 +- 2.4

67.9 +- 3.2

BNP

RIA

Serum

NA

Adm.

16.872 +- 2.39

26.98 +- 1.98

60

60

[47]

Zhu Deng-an, 2013

China

61.8

NA

39.45 +- 13.25

NA

NT-proBNP

ECL

Serum

b12 h

24 h

4289.97 +- 1883.98

145.26 +- 21.67

68

30

Abbreviations: Ref. = reference; y = years; NA = not available; BNP = B-type natriuretic peptide; NT-proBNP = N-terminal fragment of BNP; data represent mean +- SD for age and biomarker levels; biomarker levels given as pg/mL; Adm. = at admission; CL = chemiluminiscence immunoassay; LFIA = lateral-flow immunoassay; ECL = electrochemiluminescence immunoassay; ELFA = enzyme linked fluorescence analysis; ELISA = enzyme-linked immunosorbent assay; RIA = radio immunity assay.

Study characteristics

Table 1. shows the characteristics of the studies included in the meta-analysis. All studies were Case-control studies [27-47] and were written in Chinese [28,29,31-47], except two studies in English [27, 30]. All Chinese studies were performed in China [28,29,31-47], and the two English studies in Turkey [27] and Germany [30] respectively. All studies were published between 2006 [30] and 2016 [35], but one in 1996 [45]. The plasma/serum level of BNP/NT-BNP were measured by using radioimmunoassay (RIA) (n = 6) [36-38,44-46] or chemiluminiscence (CL) (n = 7) [27,29,31,32,35,42,43] method respec- tively in most studies, in four by electrochemistry (ECL), in two [34,39] by enzyme-linked immunosorbent assay (ELISA), in one [28] by bi- directional lateral flow immunoassay, and in one [33] by enzyme linked fluorescence analysis (ELFA). BNP/NT-proBNP detected in 10 studies came from plasma [29,31-34,39-43], in 10 studies from serum [27,28, 30,35-38,44,46,47], and in one study [45] from either of them. The bio- marker measured in 15 studies was BNP [29,31,32,34-39,41-46], while that in other 6 studies was NT-proBNP [27,28,30,33,40,47]. The total number of patients was 1369 ranging from 14 [30] to 134 [37]. 17 stud- ies [28-31,33,34,36,37,39-47] excluded those patients with other dis- eases combined, while other studies [27,32,35,38] didn’t mention it. The total number of controls was 806 ranging 10 [27,28,30] to 82 [35]. There were 11 studies [29,31,34,35,38-40,42-44,46] which enrolled gender- and age- matched controls. The collection time of plasma/ serum sampling was one day after admission in 8 studies [27,28,32,33, 35,41,43,47], on admission or b 1 h in 10 studies [29-31,34,37,40,42, 44-46], and 3 studies [36,38,39] didn’t report it.

Overall estimates for BNP/NT-proBNP blood levels in TBI patients and controls

In our meta-analysis, considering a high between-study heterogene- ity (P b 0.00001; I2 = 99%), we pooled the data under the random- effects model. The results in Fig. 2 showed that BNP/NT-proBNP levels were significantly higher in TBI patients than controls (M.D. 35.76, 95% confidence interval 30.57 to 40.96; I² 99%). Table 2 displayed the re- sults of stratified meta-analyses. From the results, we inferred that var- ious types of measurement of BNP/NT-proBNP led to heterogeneity partially, not only because the heterogeneity improved slightly when BNP/NT-proBNP levels were only measured by radioimmunoassay or ELISA, but because an opposite result was seen when studies with radio- immunoassay method were only included (M.D. -9.00, 95% confidence interval -10.70 to - 7.30; I² 88%). Other resources of heterogeneity were not found, although we took into account different cities, BNP vs NT-proBNP, serum vs plasma and the measurement time of BNP/NT- proBNP (Table 3).

Assessment of quality, between studies variability, and publication bias

Of 21 studies included in the meta-analysis, whose quality scores ranged from 5 [27,28,32,36] to 7 [29,31,34,39,40,42-44,46], 17 studies [29-31,33-35,37-47] were considered as high-quality studies, while other 4 studies [27,28,32,36] were low-quality. After excluding the low-quality studies, the result of the meta-analysis was 41.65 (95% CI 35.43-47.88). In addition, after single studies were removed one by one for a sensitivity analysis, meta-analyses estimates were not affected, which ranged from 25.00 (95% CI 20.26-29.73) [35] to 42.73 (95%

36.84-48.61) [36].

Nevertheless, Egger’s test revealed the presence of significant publi- cation bias (P = 0.000). Then, after trim-and-fill test was used for the assessment of publication bias, the result was not affected (pooled WMD 24.724, 95% CI 19.291 to 30.157).

Discussion

In recent years, given that great progress is being made in physiology and chemistry of TBI, more and more researchers are paying attention to the biomarkers after TBI, which mainly include S-100? protein, ubiqui- tin C-terminal hydrolase L1 (UCH-L1), Neuron-specific enolase , glial fibrillary acidic protein (GFAP) and so on [48]. After performing a systematic review and meta-analysis, Mercier et al. [49] drew a conclu- sion that measuring the S-100? protein could help to evaluate the se- verity of TBI. Also, Jian Li et al. [50] conducted a systematic review and meta-analysis to show that higher levels of serum UCH-L1 were seen in TBI cases than matched controls and that serum UCH-L1 might be a potential biomarker of TBI. Besides, the meta-analysis Feng Cheng [51] et al. did in 2014 suggested that NSE might be able to predict mortality and outcome of TBI. In addition, what could be summarized from the meta-analysis of E Laroche et al. [52] was that an unfavourable progno- sis was showed in those patients following TBI with significantly higher serum GFAP levels. Nevertheless, the biggest problem is that most bio- markers for TBI, including the four mentioned above, are not widely ap- plied to clinical practice and standard indicator [13,53], which makes it hard for further research of a clinical study in the near future. Thus, con- sidering development of measuring blood BNP/NT-proBNP level [54] and the extensive application in the clinical, especially in the diagnosis of heart failure [55], we think that blood BNP/NT-proBNP level will probably be a promising candidate biomarker for TBI. In the meta- analysis we carried out here, our results confirmed that the blood level of BNP/NT-proBNP in patients following TBI was considerably in- creased than controls.

However, the pathophysiological mechanism of increased BNP/NT- proBNP after TBI remains unclear. Two possible mechanisms are as follows.

Fig. 2. Forest plots for the relationship between BNP/NT-proBNP and TBI. Weighted mean differences (diamonds) for BNP/NT-proBNP levels between TBI and healthy control groups. Abbreviations: BNP = B-type natriuretic peptide; NT-proBNP = N-terminal fragment of BNP; CI = confidence interval.

Table 2

Weighted mean differences for BNP/NT-proBNP levels between TBI and healthy control groups, with subgroups analyses.

	N of studies	N of participants	MD (95% Cl)	Heterogeneity (I2)	Test for subgroup differences
Overall City	21	2175	35.76 [30.57, 40.96]	99%	P = 0.74, I2 = 0%
China	19	2111	38.14 [32.87, 43.41]	99%
Non-China Biomarker	2	64	85.03 [-193.98, 364.03]	95%	P = 0.37, I2 = 0%
BNP	15	1808	42.32 [28.33, 56.32]	100%
NT-proBNP Method of measurement	6	367	853.85 [-924.05, 2631.75]	100%	P b 0.00001, I2 = 100%
CL	7	718	274.39 [147.49, 401.29]	100%
RIA	6	756	-9.00 [-10.70, -7.30]	88%
ECL	4	252	1021.42 [-1519.25, 3562.09]	100%
LFIA	1	58	264.50 [229.25, 299.75]
ELFA	1	99	375.84 [364.82, 386.86]
ELISA Biological sample	2	292	106.19 [97.54, 114.84]	77%	P b 0.00001, I2 = 93.1%
Serum	10	1048	467.86 [239.92, 695.81]	100%
Plasma	10	1035	109.65 [45.61, 173.68]	100%
Serum/plasma Blood sample collection time	1	92	-6.10 [-7.70, -4.50]		P = 0.18, I2 = 43.9%
24 h	8	753	627.68 [-198.56, 1453.93]	100%
Adm. or b1 h	10	1065	64.77 [40.12, 89.42]	100%

Abbreviations: N = number; BNP = B-type natriuretic peptide; NT-proBNP = N-terminal fragment of BNP; CI = confidence interval; MD = mean difference; Adm. = at admission; CL = chemiluminiscence immunoassay; LFIA = lateral-flow immunoassay; ECL = electrochemiluminescence immunoassay; ELFA = enzyme linked fluorescence analysis; ELISA = enzyme- linked immunosorbent assay; RIA = radio immunity assay.

For one thing, previous studies showed that increased BNP levels after TBI were associated with electrolyte disturbances such as Cerebral salt wasting syndrome (CSWS). CSWS, firstly proposed by Peters et al. in 1950, is characterized by a renal loss of sodium after brain injury or ill- ness including TBI, often leading to electrolyte disturbance [56]. Peters et al. hypothesized that direct damage to the brain causes the release from cerebral BNP directly into the circulation, and it was found in 1992 [57] that BNP does exist in human cortical and subcortical struc- ture. In addition, significant increase in sympathetic outflow after acute brain injury may cause myocardial ventricular strain, and as a re- sult, BNP will release from heart tissue [58]. However, in a review of the literature, Jan Leonard et al. [59] included 4 studies concerning about the relationship between the changes of BNP levels and CSWS after TBI, only to find that no elevated BNP levels were seen in 3 of 4 studies except for the left 1 study. Thus, the findings themselves are controver- sial, and whether increased BNP levels after TBI is related to electrolyte disturbances exists unclear.

For another, recent evidence suggests that increased BNP may play a neuroprotective role in brain injury. For instance, James et al. [60] found that after i.v. administration of nesiritide, a kind of exogenous human recombinant BNP, into murine models of TBI, reduced neurodegenera- tion and improved cerebral blood flow and functional outcome were seen in mice. Taking all studies with inconsistent results into account, A Hodes et al. [61] argued that it may be a lack of the endogenous neu- roprotective mechanism. Nevertheless, evidence of the neuroprotective role of BNP after TBI is still insufficient because of the paradoxical results and limited studies.

What’s more, considering lots of limitations, we should interpret the results carefully.

The extent of heterogeneity between studies remained considerably high (N 70%) and significantly different [62], even after conducting sub- group analyses, stratified by different cities, BNP vs NT-proBNP, serum vs plasma and the measurement time of BNP/NT-proBNP. Even so, dif- ferent types of biomarkers, measurement method and the measure- ment time were still sources of heterogeneity. For one reason, the half-life of BNP and NT-proBNP is different, 30 min and 2 h, respectively [63]. For another, besides the studies in our meta-analysis [27,28,30,32, 33,35,43,44], some previous studies [20,64,65] excluded because of a lack of control groups also showed that variations of blood BNP/NT- proBNP were found in different time periods, and that the highest and lowest concentration varied greatly from different studies.

With an exception of measurement method and serum vs plasma, subgroups analyses didn’t show a difference in the meta-estimates. In particular, blood BNP/NT-proBNP levels measured by radioimmu- noassay after TBI were lower than that in control groups, which was contrary to the overall estimates. In fact, according to all studies where radioimmunoassay was used, BNP/NT-proBNP were all mea- sured in plasma, and in the left one study, it was not clear whether BNP/NT-proBNP was measured in plasma or serum. Thus, perhaps different measurement methods were the real cause of affecting the meta-estimates.

As for publication bias, since negative studies were generally less ac- cepted and impactive, the lack of published negative studies was likely to have an influence on the results of our meta-analysis. However, we

Table 3

Newcastle-Ottawa scale scores for studies included in the meta-analysis

Study ID

45

29

31

43

44

47

32

27

46

42

40

39

37

35

36

34

33

28

30

38

41

Selection

Adequate definition of the cases

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

Representativeness of the cases

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

Selection of controls

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

Definition of controls

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

Comparability

First confounding

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

?

Exposure Total scores

Second confounding Ascertainment of exposure Same method of ascertainment Non-response rate

? ? 6

? ? 7

? ? 7

? ? 7

? 7

? 6

? 5

? ? 5

? ? 7

? ? 7

? ? 7

? 7

? ? 6

? 6

? ? 5

? 7

? 6

? 5

? ? 6

? 6

? 6

had been trying to search for unpublished studies with an abundant database search to reduce the probability of publication bias, but failed.

Strengths

There were some strengths in our meta-analysis.

Returning to the question posed at the beginning of this study, not only was this the first meta-analysis for the relationship between blood BNP/NT-proBNP levels and patients following TBI, our results also settled the disputes and revealed that blood BNP/NT-proBNP levels would be significantly higher after TBI compared with controls.

Without publication date and any language limitation set in our search strategy, it was made certain that studies undetected are not as much as possible. Moreover, from study selection, to data extrac- tion, to the quality assessment, all work were completed by three au- thors (Zhang Y, Zhou LZ and Bao Y), which could help reduce bias and errors as well.

Limitations

However, there certainly existed a lot of limitations in our study.

All studies included in our research were retrospective studies, and small sample sizes, a high between-study heterogeneity and publica- tion bias in the analysis should be taken into account. Although no language limitation was set in the search strategy, most studies in- cluded came from China because few studies were performed by other countries and we excluded some studies with a lack of control groups.

Although our results figured out that elevated blood BNP/NT-proBNP levels were seen after TBI, yet we didn’t do a meta-analysis to assess the relationship between function outcome as well as mortality after TBI and elevated blood BNP/NT-proBNP levels because of the limited studies.

On the one hand, in preclinical medicine, few studies have been done in researching pathophysiological mechanisms of increased BNP/NT- proBNP after TBI, and two possible mechanisms mentioned above re- main controversial. On the other, in clinical medicine, even if BNP/ NT-proBNP becomes a biomarker of TBI in the future, it will only be used for those TBI patients without other related disease combined, such as heart failure because it is hard to distinguish BNP/NT- proBNP originating in brain from that in other tissues.

Conclusion

In summary, our meta-analysis firstly demonstrated that contrasted with control groups, the plasma/serum levels of BNP/NT-proBNP will rise after TBI, which may in fact be protective against neurodegenera- tion. In our meta-analysis, one of the biggest problems was the consid- erable heterogeneity, which may result from two kinds of biomarkers with different half-lives, but we failed to search for the resources of the heterogeneity due to the limited studies included. Considering the limitations above, however, for further proof and effective application to clinical medicine, it is necessary to conduct more research in correla- tive preclinical medicine. Beyond that, more high-quality clinical studies with larger samples carried out in different countries, especially in non- China, are also required.

Funding

This study was supported by Key Clinical Specialty Discipline Con- struction Program, President Foundation of Nanfang Hospital, Southern Medical University (2013B017, 2015C028), and Science and Technology Department, Guangdong Province (2013B021800304).

References

1. Menon DK, Schwab K, Wright DW, Maas AI. Position statement: definition of trau- matic brain injury. Arch Phys Med Rehabil 2010;91(11):1637-40. http://dx.doi. org/10.1016/j.apmr.2010.05.01721044706.
2. Hyder AA, Wunderlich CA, Puvanachandra P, Gururaj G, Kobusingye OC. The impact of Traumatic brain injuries: a global perspective. NeuroRehabilitation 2007;22(5): 341-5318162698.
3. Langlois JA, Rutland-Brown W, Wald MM. The epidemiology and impact of traumatic brain injury: a brief overview. J Head Trauma Rehabil 2006;21(5):375-816983222.
4. Faul M, Xu L, Wald MM, Coronado VG. Traumatic brain injury in the United States: emergency department visits, hospitalizations, and deaths. Emergency Service Hos- pital 2010.
5. Astrand R, Romner B. Classification of head injury. Berlin Heidelberg: Springer; 2012.
6. Steyerberg EW, Mushkudiani N, Perel P, Butcher I, Lu J, McHugh GS, et al. Predicting outcome after traumatic brain injury: development and validation of a prognostic score based on admission characteristics. PLoS Med 2008;5(8):1025-39.
7. Marshall LF, Marshall SB, Klauber MR, Van Berkum Clark M, Eisenberg H, Jane JA, et al. The diagnosis of head injury requires a classification based on computed axial tomography. J Neurotrauma 1992;9(Suppl. 1):S287-921588618.
8. Belanger HG, Vanderploeg RD, Curtiss G, Warden DL. Recent neuroimaging tech- niques in mild traumatic brain injury. J Neuropsychiatry Clin Neurosci 2007;19(1): 5-20. http://dx.doi.org/10.1176/jnp.2007.19.1.517308222.
9. Cappa KA, Conger JC, Conger AJ. Injury severity and outcome: a meta-analysis of pro- spective studies on TBI outcome. Health Psychol 2011;30(5):542-60. http://dx.doi. org/10.1037/a002522021875208.
10. Kesler SR, Adams HF, Bigler ED. SPECT, MR and quantitative MR imaging: correlates with neuropsychological and psychological outcome in traumatic brain injury. Brain Inj 2000;14(10):851-711076132.
11. Injury P. Report to congress on mild traumatic brain injury in the United States: steps to prevent a serious Public health problem. Brain Inj 2003.
12. Ramchandani D, Levenson J. Textbook of traumatic brain injury. J Head Trauma Rehabil 2011;28(2):2161.
13. Kovesdi E, Luckl J, Bukovics P, Farkas O, Pal J, Czeiter E, et al. Update on protein bio- markers in traumatic brain injury with emphasis on clinical use in adults and pedi- atrics. Acta Neurochir 2010;152(1):1-17. http://dx.doi.org/10.1007/s00701-009- 0463-619652904.
14. Svetlov SI, Larner SF, Kirk DR, Atkinson J, Hayes RL, Wang KK. Biomarkers of blast- induced neurotrauma: profiling molecular and cellular mechanisms of blast brain injury. J Neurotrauma 2009;26(6):913-21. http://dx.doi.org/10.1089/neu.2008. 060919422293.
15. Cowie MR, Jourdain P, Maisel A, Dahlstrom U, Follath F, Isnard R, et al. Clinical appli- cations of B-type natriuretic peptide testing. Eur Heart J 2003;24(19): 1710-814522565.
16. Koulouri S, Acherman RJ, Wong PC, Chan LS, Lewis AB. Utility of B-type natriuretic peptide in differentiating congestive heart failure from lung disease in pediatric pa- tients with respiratory distress. Pediatr Cardiol 2004;25(4):341-6. http://dx.doi.org/ 10.1007/s00246-003-0578-015054559.
17. Sudoh T, Kangawa K, Minamino N, Matsuo H. A new natriuretic peptide in porcine brain. Nature 1988;332(6159):78-81. http://dx.doi.org/10.1038/332078a02964562.
18. Potter LR, Yoder AR, Flora DR, Antos LK, Dickey DM. Natriuretic peptides: their struc- tures, receptors, physiologic functions and therapeutic applications. Handb Exp Pharmacol 2009;191:341-66. http://dx.doi.org/10.1007/978-3-540-68964-5_

    1519089336. [PMCID: PMCPmc4855512].

    Uddin MH, Rashid T, Chowdhury SM. Role of B-type natriuretic peptide in heart failure. Int J Dis Hum Dev 2016.

19. Sviri GE, Soustiel JF, Zaaroor M. Alteration in Brain natriuretic peptide plasma concentration following severe traumatic brain injury. Acta Neurochir 2006;148(5): 529-33. http://dx.doi.org/10.1007/s00701-005-0666-416322908. [discussion 33].
20. Sviri GE, Shik V, Raz B, Soustiel JF. Role of brain natriuretic peptide in cerebral vaso- spasm. Acta Neurochir 2003;145(10):851-60. http://dx.doi.org/10.1007/s00701- 003-0101-714577006. [discussion 60].
21. Ruggieri F, Noris A, Beretta L, Mortini P, Gemma M. Serum B-type natriuretic peptide is affected by neoplastic edema in patients with a brain tumor. World Neurosurg 2016;85:193-6. http://dx.doi.org/10.1016/j.wneu.2015.08.07426348568.
22. Garcia-Berrocoso T, Giralt D, Bustamante A, Etgen T, Jensen JK, Sharma JC, et al. B- type natriuretic peptides and mortality after stroke: a systematic review and meta-analysis. Neurology 2013;81(23):1976-85. http://dx.doi.org/10.1212/01.wnl. 0000436937.32410.3224186915. [PMCID: PMCPmc3854833].
23. Duval S, Tweedie R. Trim and fill: a simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 2000;56(2): 455-6310877304.
24. Stang A. Stang A: critical evaluation of the Newcastle-Ottawa Scale for the assess- ment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 25: 603-5. Eur J Epidemiol 2010;25(9):603-5.
25. Aleksovska K, Leoncini E, Bonassi S, Cesario A, Boccia S, Frustaci A. Systematic review and meta-analysis of circulating S100B blood levels in schizophrenia. PLoS One 2014;9(9). http://dx.doi.org/10.1371/journal.pone.010634225202915. [PMCID: PMCPmc4159239].
26. Akgun B, Erol FS, Yildirim H, Ilhan N, Kaplan M. The correlation of serum NT-proBNP levels of hemorrhagic and ischemic lesions detected with diffusion MRI in head traumas. Turk Neurosurg 2013;23(3):336-43. http://dx.doi.org/10.5137/1019- 5149.jtn.6982-12.123756972.
27. CHEN CJ, CAI T, HU ZL. Clinical significance of the changes in serum NT-proBNP level in aged patients with severe brain dysfunction in acute phase. Chin J Geriatr 2011; 30(6):446-8.
28. Chen P. The variation of BNP level among acute cranioCerebral injury patients and its clinical signifigance. Chin J Prac Nerv Dis 2013;16(22):13-4.
29. Kirchhoff C, Stegmaier J, Bogner V, Buhmann S, Mussack T, Kreimeier U, et al. Intra-

    thecal and systemic concentration of NT-proBNP in patients with severe traumatic brain injury. J Neurotrauma 2006;23(6):943-9. http://dx.doi.org/10.1089/neu. 2006.23.94316774478.

    Du RJ. Relationship between the change in CT-type,plasma BNP and the prognosis in patients with acute head injury. Hebei Med 2014;10:1654-6.

30. Fang ZC, Zheng X, Liu BY, Chen L. Clinical analysis of blood plasmanatriuretic peptide change 39 severe brain injury patients with cerebral salt wasting syndrome. Shaanxi Med J 2010;39(12):1614-6.
31. Li SM. Alteration of plasma N-terminal pro-B-type natriuretic peptide and its rela- tionship with prognosis in patients with isolated traumatic brain injury [D]. Bengbu Medical College; 2015.
32. LIN ZX, SHI WF. Impact of plasma B-type brain natriuretic peptide precursor levels on clinical features of CT in patients with severe traumatic brain injury. Chin J Coal Ind Med 2015;18(2).
33. Liu YM, Meng B, Cai SX, Lin HW. On the joint detection of serum brain natriuretic peptide and C-reactive protein applied in diagnosis of acute brain injury. J Liaoning Medical University 2016;37(2):44-6.
34. Liu YL, Fu GH, Li FL. Clinical significance of measurement of changes of serum BNP NT plasma ET-1 levels in patients with acute craniocerbral injury. Med Front 2013;26.
35. Peng LH, Xiao CE. Clinical significance of changes of serum BNP and NSE levels in pa- tients with acut craniocerebral injury. J Radioimmunol 2009;22(3):230-1.
36. Peng SQ. Clinical research on nerve Endocrine disorders of craniocerebral trauma and hypothermia intervention. Chin Mod Med 2015;10:50-2.
37. Wang YF, Zhang F, Yao YF, Tian W. Correlative analysis between plasma BNP, CRP levels and computed tomographic manifestations in acute moderate or severe traumatic brain injury. Chin J Prim Med Pharm 2015;19.
38. Wen T. The relationship of the plasma levels between N-terminal pro-B-type natriuretic peptide and high-sensitivity C reactive protein in patients with severe traumatic brain injury [D]. Central South University; 2013.
39. Xiong XH. Injection of salvia tetramethylpyrazine on patients with head injury plas- ma of NSE, BNP and its clinical efficacy [D]. Fujian University of Traditional Chinese Medicine; 2012.
40. Xu J. Relationship between BNP and CT findings and Clinical prognosis in patients with acute cerebral injury [D]. JiNan University; 2010.
41. Zhang DF, Chen H. Assessment of traumatic brain injury by rapid detection of brain natriuretic peptide. Chin J Prac Nerv Dis 2014;15:28-9.
42. Zhang JF. The clinical analysis of IL-6, BNP, CRP expression levels in patients with brain blood. Journal of Hainan Medical University 2015;21(9):1219-21.
43. Zhang WC, Zheng LP, Sun XC, Xu YQ. Clinical significance of the changes of blood na- triuretic factor and ADH levels in acute craniocerebral injury. Chin J Traumatol 1996; 2:96-8.
44. Zhao F, He XD, Wang HX, Zhao SB, Cheng Liu. Change in serum BNP, D-dimer and C- reactive protein in elderly patients with craniocerebral injury and the clinical signif- icance. Chin J Geriatr 2015;8.
45. Zhu DA, Wei XY, Chen HW, Feng J. The significance of the changes of NT-proBNP in severe brain injury patients. J Clin Emerg (China) 2013(8):360-2.
46. Papa L, Ramia MM, Kelly JM, Burks SS, Pawlowicz A, Berger RP. Systematic review of clinical research on biomarkers for pediatric traumatic brain injury. J Neurotrauma 2013;30(5):324-38. http://dx.doi.org/10.1089/neu.2012.254523078348.
47. Mercier E, Boutin A, Lauzier F, Fergusson DA, Simard JF, Zarychanski R, et al. Predictive value of S-100beta protein for prognosis in patients with moderate and severe trau- matic brain injury: systematic review and meta-analysis. BMJ (Clinical Research Ed) 2013;346:f1757. http://dx.doi.org/10.1136/bmj.f175723558282.
48. Li J, Yu C, Sun Y, Li Y. Serum ubiquitin C-terminal hydrolase L1 as a biomarker for traumatic brain injury: a systematic review and meta-analysis. Am J Emerg Med 2015;33(9):1191-6. http://dx.doi.org/10.1016/j.ajem.2015.05.02326087705.
49. Cheng F, Yuan Q, Yang J, Wang W, Liu H. The prognostic value of serum neuron- specific enolase in traumatic brain injury: systematic review and meta-analysis. PLoS One 2014;9(9):e106680. http://dx.doi.org/10.1371/journal.pone. 010668025188406. [PMCID: PMCPmc4154726].
50. Laroche E, Turgeon A, Boutin A, Mercier E, Lauzier F, Zarychanski R, et al. Predictive value of glial fibrillary acidic protein for prognosis in patients with moderate and se- vere traumatic brain injury: a systematic review and meta-analysis. Crit Care 2012; 16(Suppl. 1):P298. http://dx.doi.org/10.1186/cc10905 [PMCID: PMCPmc3363716].
51. Kochanek PM, Berger RP, Bayir H, Wagner AK, Jenkins LW, Clark RS. Biomarkers of pri- mary and evolving damage in traumatic and ischemic brain injury: diagnosis, progno- sis, probing mechanisms, and therapeutic decision making. Curr Opin Crit Care 2008; 14(2):135-41. http://dx.doi.org/10.1097/MCC.0b013e3282f5756418388674.
52. Jiang K, Shah KS, Terracciano GJ, Bhamre S, Pentova E, Fitzgerald RL, et al. Compari- son of a point-of-care method to measure BNP levels with a standard lab-based in- strument in patients with congestive heart failure. J Card Fail 2008;14(6):S37-S8.
53. Don-Wauchope AC, McKelvie RS. Evidence based application of BNP/NT-proBNP testing in heart failure. Clin Biochem 2015;48(4-5):236-46. http://dx.doi.org/10. 1016/j.clinbiochem.2014.11.00225448029.
54. Peters JP, Welt LG, Sims EA, Orloff J, Needham J. A salt-wasting syndrome associated with cerebral disease. Trans Assoc Am Physicians 1950;63:57-6414855556.
55. Takahashi K, Totsune K, Sone M, Ohneda M, Murakami O, Itoi K, et al. Human brain natriuretic peptide-like immunoreactivity in human brain. Peptides 1992;13(1): 121-31535705.
56. Nakagawa O, Ogawa Y, Itoh H, Suga S, Komatsu Y, Kishimoto I, et al. Rapid transcrip- tional activation and early mRNA turnover of brain natriuretic peptide in cardiocyte hypertrophy. Evidence for brain natriuretic peptide as an “emergency” cardiac hor- mone against ventricular overload. J Clin Invest 1995;96(3):1280-7. http://dx.doi. org/10.1172/jci1181627657802. [PMCID: PMCPmc185749].
57. Leonard J, Garrett RE, Salottolo K, Slone DS, Mains CW, Carrick MM, et al. Cerebral salt wasting after traumatic brain injury: a review of the literature. Scand J Trauma Resusc Emerg Med 2015;23. http://dx.doi.org/10.1186/s13049-015-0180- 526561391. [PMCID: PMCPmc4642664].
58. James ML, Wang H, Venkatraman T, Song P, Lascola CD, Laskowitz DT. Brain natri- uretic peptide improves long-term functional recovery after acute CNS injury in mice. J Neurotrauma 2010;27(1):217-28. http://dx.doi.org/10.1089/neu.2009. 102219803787.
59. Hodes A, Lichtstein D. Natriuretic hormones in brain function. Front Endocrinol 2014;5:201. http://dx.doi.org/10.3389/fendo.2014.0020125506340. [PubMed Cen- tral PMCID: PMCPmc4246887].
60. Becker L. Cochrane handbook for systematic reviews of interventions version 5.1.4; 2011.
61. Coull BM. Comment: natriuretic peptides as Predictive biomarkers of stroke outcome. Neurology 2013;81(23):1983. http://dx.doi.org/10.1212/01.wnl.0000436941.55281. 9c24186913.
62. Wu X, Sha H, Sun Y, Gao L, Liu H, Yuan Q, et al. N-terminal pro-B-type natriuretic pep- tide in patients with isolated traumatic brain injury: a prospective cohort study. J Trau- ma 2011;71(4):820-5. http://dx.doi.org/10.1097/TA.0b013e3182277b6921808206. [discussion 5].

    Tsentsiper L, Kondratyeva E, Kondratyev S, Dryagina N. Levels of N-terminal pro- brain natriuretic peptide in brain injury patients. Crit Care 2015;19(Suppl. 1): P458. http://dx.doi.org/10.1186/cc14538 [PMCID: PMCPmc4470765].