Administration of Yunnan Baiyao improved clinical prognosis of emergency craniotomy in patients with moderate-to-severe TBI: a randomized controlled trial

Background: Although emergency craniotomy has been performed to reduce the mortality of traumatic brain injury (TBI), it is associated with secondary injury and poor prognosis. The ecacy of a wound-healing medicine Yunnan Baiyao (YB) was compared with a TBI medicine Xingnaojing (XNJ) and orthodox therapy (OT). Method: Eighty patients (39±7 yrs.) with moderate-to-severe TBI (Glasgow Coma Scale(GOS), 3-12) received craniotomy before randomly assigned to (n=20) 7 days treatment of: 1) OT; 2) OT+XNJ (20 ml i.v. drip/daily); 3) OT+low dose-YB (oral, 1,000 mg/day); 4) OT+high dose-YB, 2,000 mg/day). Serum S100B and superoxide dismutase (SOD), Glasgow Outcome Scale (GOS) and Karnofsky Performance Scale (KPS) were evaluated. Results: GCS score was lowest on admission that improved more quickly and became signicantly higher in XNJ, l-YB, h-YB groups than in OT group (P<0.01). S100B peaked higher but declined more slowly and became signicantly higher in OT than in other groups (P<0.01). By Day 7, S100B declined to 20% below baseline in other groups but remained 19.4% above baseline in OT group. OT lost 38% SOD activity on Day 3 and regained 69% of SOD activity on Day 7 whereas other groups lost 16.7%~23.4% SOD on Day 3 but regained 91.8%~99.5% activity on Day 7 (P<0.01). GOS and KPS scores were signicantly greater (~25%~30%) in the XNJ, l-YB, h-YB groups than in OT group 1-3 months post-surgery. Conclusion: YB is signicantly better than OT and non-inferior to XNJ in improving postoperative recovery and clinical prognosis after emergency craniotomy in patients with moderate-to-severe TBI. and anticoagulant properties of YB in TBI remains to be studied. A potential cause-Page

Signi cant effects of Treatment, Time and Time x Treatment interaction (P < 0·001) were found in SOD (Tables 2 & 3). Serum SOD activity declined rapidly after craniotomy and reached nadir on Days 3 day in all groups with a greater reduction found in the OT group (38·3%) than in XNJ, l-YB and h-YB groups (16·7%, 23·4% and 20·7%) ( Table 3, Fig. 1C). Serum SOD then bounced back more rapidly and became signi cantly higher in the XNJ, l-YB and h-YB groups than in the OT group after Day 3 of surgery (P < 0·01). ANOVA show no signi cant difference in SOD activity between the XNJ, l-YB and h-YB groups except it was signi cantly lower in l-YB group than in XNJ group on Day 7 after surgery (P > 0·05) (Figs. 4B&C). Chang and rate of change over the baseline showed the greatest SOD de cits on Day 3 in all groups (Figs. 4B&C). By Day 7, XNJ, l-YB and h-YB groups resumed most of the baseline SOD activities (99·5%, 91·8% and 94·8%, respectively) compared to OT group (69·2%) (P < 0·01).
Correlations between experimental variables GCS score was signi cantly correlated with GOS/KPS scores in pooled TBI patients and in each of the treatment groups (P < 0·01) ( Table 4), indicating a critical impact of postoperative recovery during the acute phase of post-surgery on long-term clinical prognosis of TBI. It is noticed that the correlation between GOS/KPS and GCS scores improved progressively from Day 1 till Day 7 after surgery. However, the reduced level of correlation between most variables immediately after the surgery (Day 1) indicates a disruptive effect of craniotomy.
KPS3 S100a S1001 S1003 S1005   Serum S100B levels were negatively correlated with GCS/GOS/KPS scores in pooled patients and in each treatment group (Tables 4-8). This and other analysis suggest that a higher S100B level on admission day or on Day 7 are better predictors of poor TBI outcomes (low GCS/GOS/KPS scores). In contrast, serum SOD activity was positively correlated with GCS/GOS/KPS scores (P < 0·01) in the pooled and in the 3 co-treatment groups but not in the OT group (Table 4). Similarly, SOD level on postoperative Day 7 is a better predictor of GCS/GOS/KPS scores than SOD on other days in the pooled and 3 adjuvant treatment groups but not the OT group. Our data suggest that GCS and SOD are positive predictors and S100B is a negative predictor of clinical prognosis.
Although emergency craniotomy has been performed to reduce the mortality rate of TBI, it is associated with secondary injury and unfavorable outcomes [5,6]. Effective agents are needed to reduce these side effects. In this study, the wound-healing medicine YB were compared with a positive control medicine XNJ, and orthodox therapy in craniotomized patients with moderate-to-severe TBI. Repeated-measure analysis showed signi cantly better clinical prognosis of YB and XNJ adjuvant treatments than OT. The signi cant greater improvement rate in GCS after high dose YB than after low dose YB and XNJ treatment indicates a dose-dependent therapeutic effects of YB on postoperative recovery.
GCS is a scale to quantify disruption of consciousness (DOC) during acute phase of TBI. Although GCS score improved in all groups, the improvement was greater in YB and XNJ groups than in OT group three days after surgery. While GCS scores were similar between XNJ and YB groups, the rate of change were signi cantly greater in the high dose YB group than in low dose YB and in XNJ groups, respectively. Effective recovery from DOC has signi cant impact on the overall well-being, cognition recovery and clinical prognosis of TBI patients. In this study, the GCS score on Days 7 was highly correlated with the clinical outcome scores of GOS (r = 0·79, P < 0·01) and KPS (r = 0·80, P < 0·01) after 30 days of surgery, indicating a intrinsic relationship between acute postoperative recovery and long-term clinical prognosis of TBI.
The GOS scores at 3 months after surgery were signi cantly higher in the l-YB (4·45 ± 0·61), h-YB (4·3 ± 0·8) and XNJ (4·35 ± 0·81) groups than in the OT (3·65 ± 0·99) group. GOS is a prognostic criterion of TBI, with the score ≥ 4 classi ed as good outcome whereas KPS re ects the functional recovery and disability of TBI patients. The KPS scores at 3 months after surgery were signi cantly higher in the l-YB (76·50 ± 17·55), h-YB (77·50 ± 15·17) and XNJ (78·00 ± 18·53) groups than in the OT (57·00 ± 20·80) group, indicating an almost adequate recovery in the 3 co-treatment groups. A KPS score of more than 80 points indicates independent, adequate capability of self-care in daily life; a score of 60 to 70 points indicate semi-dependent in daily life and a score of below 60 points indicates dependent, lack of self-care capability in daily life.
The attenuated elevation in serum S100B levels after YB and XNJ treatments indicates attenuated brain injury after YB and XNJ adjuvant therapy. While no difference found on admission and on Day 1, serum S100B peaked three days after surgery in all groups before declined more rapidly in YB and XNJ groups that became signi cantly lower than in the OT group since Days 3 and became below baseline levels on Day 7. Increased serum s100b is a biomarker of astrocyte injury and an early predictor of high intracranial pressure and mortality in TBI patients. [35][36][37][38][39][40] Other studies show that serum S100b level outperforms clinical diagnosis for identi cation of intracranial injury after TBI [45]. In this study, serum S100b on admission and on Days 7 after DC were negatively correlated with GCS scores (r= -0·35, P < 0·01, r= -0·63, P < 0·01,respectively) and with 1&3 Months after discharge scores of GOS (r= -0·44, P < 0·01, r= -0·61, P < 0·01, respectively) and KPS (r= -0·49, P < 0·01, r= -0·66, P < 0·01 respectively), indicating that serum S100b is a negative predictor for both acute brain tissue damage and long-term clinical prognosis. Our results agree with the report that early postoperative serum s100B levels predict ongoing brain damage after meningioma surgery [46]. It has been reported that elevated serum S100B may have extracerebral origins including traumatized tissues and is a predictive index of overall injury and prognosis after trauma/surgery [35].
In contrast, serum SOD activity was positively correlated with GCS, GOS and KPS scores in pooled TBI patients and in h-YB and XNJ groups, and to a less extent, in l-YB group but not in the OT group. And the level of correlation increased from admission till postoperative Day 7. A signi cant negative correlation between serum SOD activity and serum S100B concentration was found on Day 7 only in pooled patients. Our results are in agreement with a recent clinical study of 56 patients with severe TBI whose blood SOD and total antioxidant capacity after 7 and 14 days of TBI were signi cant predictors of clinical outcomes (GOS score ≥ 4) three months later [47].
Serum superoxide dismutase (SOD) activity re ects the body's defense against free radicals-induced damage. Massive production and release of superoxide and radicals after TBI can cause artery dilation, edema and hemostasis [48,49] whereas SOD can dose-dependently scavenge excess superoxide anion, reduce brain edema and attenuates acute brain injury after subarachnoid hemorrhage [42]. Other studies have shown that blood-brain barrier (BBB) permeability, anti-oxidative and anti-apoptotic capability, brain edema, cerebral blood ow, cognitive and behavioral function, and necrotic cavity volume are CuZn-SOD dose-dependently protected in heterozygous and homozygous transgenic mice with human CuZn-SOD gene [50] and in other SOD-upregulating TBI models [49,51,52].
In this study, both YB and XNJ treatments resulted in similar improvements in neuroprotection and clinical prognosis. However, some differences do exist. The h-YB resulted in signi cant (rate of) increases in GCS score than l-YB and XNJ three days after surgery, suggesting more effective intervention after high dose YB administration whereas XNJ therapy resulted in more consistent and signi cantly less (rate of) reduction in SOD activity than YB, indicating a greater antioxidative potential of XNJ than YB. The greater loss in SOD activity in l-YB group than in h-YB group on Day 1 and 3 and a great rate of increase in S100b in l-YB than XNJ on Day 3, indicating a dose-dependent neuroprotection of YB, possibly related to the multi anti-hemorrhage, anti-hemostasis and antiin ammation properties of YB as discussed below.
This pilot study has limitations. Parameters including GI bleeding, systematic in ammation and mortality rate were not recorded nor the lesion size and hemorrhage volume after emergency craniotomy. The hemodynamics and anticoagulant properties of YB in TBI remains to be studied. A potential cause-effect relationship may exist between S100b and SOD inductions and that needs to be further explored.

Conclusion
Yunnan Baiyao adjuvant therapy is non-inferior to the positive control medicine Xingnaojing but is more effective than orthodox mono-therapy in improving postoperative recovery and clinical outcomes after emergency craniotomy in patients with moderate-to-severe TBI. The therapeutic effects of YB appear to be dose-dependent and regulated partly through attenuated disruption in S100b-and SOD-pathways. If proved in different cohorts, YB could be used to control TBI-induced morbidity and mortality especially in situations of limited resources.

Consent for publication
All authors have approved the manuscript for submission.
KPS3 S100a S1001 S1003 S1  Differences in serum S100b protein concentration A), change B) and rate of change C) of serum S100b in craniotomized TBI patients. A) S100B concentration increased sharply after admission and peaked in all groups on Days 3 before it fell more rapidly and became signi cantly lower (P<0·01) in the XNJ, l-YB and h-YB groups than in the OT group after 3, 5 and 7 days of emergency craniotomy; B) & C) Increases and percentage increase in serum S100B was signi cantly greater (P<0·01) in OT than in XNJ and h-YB groups on Day 3 whereas reduction and reduction rate was signi cantly greater in XNJ, l-YB and h-YB groups than in OT group after 5 and 7 days of emergency craniotomy · *, P<0·05, **P<0·01 vs. OT; a, P<0·05, aa, P<0·01 vs. XNJ