Research ArticleSpinal Cord Injury

Acute hyperglycemia impairs functional improvement after spinal cord injury in mice and humans

See allHide authors and affiliations

Science Translational Medicine  01 Oct 2014:
Vol. 6, Issue 256, pp. 256ra137
DOI: 10.1126/scitranslmed.3009430

Abstract

Spinal cord injury (SCI) is a devastating disorder for which the identification of exacerbating factors is urgently needed. We demonstrate that transient hyperglycemia during acute SCI is a detrimental factor that impairs functional improvement in mice and human patients after acute SCI. Under hyperglycemic conditions, both in vivo and in vitro, inflammation was enhanced through promotion of the nuclear translocation of the nuclear factor κB (NF-κB) transcription factor in microglial cells. During acute SCI, hyperglycemic mice exhibited progressive neural damage, with more severe motor deficits than those observed in normoglycemic mice. Consistent with the animal study findings, a Pearson χ2 analysis of data for 528 patients with SCI indicated that hyperglycemia on admission (glucose concentration ≥126 mg/dl) was a significant risk predictor of poor functional outcome. Moreover, a multiple linear regression analysis showed hyperglycemia at admission to be a powerful independent risk factor for a poor motor outcome, even after excluding patients with diabetes mellitus with chronic hyperglycemia (regression coefficient, −1.37; 95% confidence interval, −2.65 to −0.10; P < 0.05). Manipulating blood glucose during acute SCI in hyperglycemic mice rescued the exacerbation of pathophysiology and improved motor functional outcomes. Our findings suggest that hyperglycemia during acute SCI may be a useful prognostic factor with a negative impact on motor function, highlighting the importance of achieving tight glycemic control after central nervous system injury.

INTRODUCTION

Traumatic spinal cord injury (SCI) can cause severe motor and sensory dysfunction, resulting in a significant reduction in the quality of life (1). Mechanical trauma rapidly leads to disruption of the blood–spinal cord barrier, neuronal cell death, edema, axonal damage, and demyelination, followed by a cascade of secondary injuries that expand the inflammatory reaction around the epicenter of the original injury (2). The central nervous system (CNS) has a limited capacity for endogenous regeneration and repair; therefore, it is necessary to identify factors that exacerbate SCI to prevent any further deterioration of neurological function and improve the outcomes of injuries (3). Although age (4), blood pressure (5), and infection (6) are each considered to be prognostic factors in patients with SCI, exacerbating factors that are amenable to treatment remain to be elucidated.

Microglial cells, the resident immune cells in the CNS, form the first line of defense after being stimulated by exposure to invading pathogens or tissue injury (7). Immediately after SCI, activated microglia enhance and propagate the subsequent inflammatory response by expressing cytokines, such as tumor necrosis factor–α (TNF-α), interleukin-6 (IL-6), and IL-1β (8). Recently, we demonstrated that the activation of microglia is associated with the neuropathological outcomes of SCI (4). Although the precise mechanisms of microglial activation remain elusive, several basic research studies have reported that hyperglycemia is involved in the activation of resident monocytic cells, including microglia. For example, the number of pancreatic resident monocytes is increased in hyperglycemic rodents, leading to the up-regulation of islet-derived inflammatory factors, such as IL-6 and IL-8 (9). In addition, peritoneal monocytes are activated under conditions of hyperglycemia, subsequently inducing greater production of TNF-α than that associated with a normoglycemic state (10). Furthermore, hyperglycemia correlates with the worsening of tactile allodynia accompanied by the hyperactivation of dorsal horn microglia (11). Because microglial activation is associated with secondary injury after SCI, we hypothesized that hyperglycemia may also influence the pathophysiology of SCI by altering microglial responses.

Here, we investigated the effects of hyperglycemia on the pathophysiological processes and motor functional outcomes in two experimental mouse models of hyperglycemia in the acute phase of SCI. An in vivo cell type–specific gene expression analysis with flow cytometry revealed enhanced proinflammatory reactivity in the microglial cells of the hyperglycemic mice, which resulted in the exacerbation of secondary injury after SCI. Furthermore, we conducted a multivariable linear regression analysis of the clinical data obtained from 528 human SCI subjects, which provided entirely new evidence showing that acute-phase hyperglycemia is a critical factor in the poor functional outcomes of SCI. Finally, we showed that achieving glycemic control can ameliorate the pathological and functional outcomes of SCI in hyperglycemic mice, thus supporting the existence of a direct relationship between acute hyperglycemia and the exacerbation of SCI outcomes.

RESULTS

Effects of hyperglycemia on the inflammatory reactivity of activated microglia

In acute SCI, microglia are activated and propagate, inducing secondary inflammation (8). To assess the microglial inflammatory activity under hyperglycemic conditions, we first used a murine microglial cell line, BV-2, cultured in normal-glucose (5.6 mM) or high-glucose (12.5 or 25 mM) medium. After lipopolysaccharide (LPS) stimulation, the BV-2 cells rapidly transformed morphologically (Fig. 1A), as previously reported (12). Enzyme-linked immunosorbent assay (ELISA) analyses of the supernatant of the cells revealed that the TNF-α and IL-6 concentrations were significantly increased in the high-glucose cultures compared to those observed in the normal-glucose cultures after LPS stimulation (Fig. 1, B and C; P < 0.05). To confirm these results in vivo, we generated hyperglycemic mice via streptozotocin (STZ) injection in which the blood glucose concentrations increased because of the disruption of pancreatic insulin-secreting β cells (Fig. 1, D and E) (13). We subsequently observed significant elevation of the blood glucose concentration in these mice, with no significant changes in body weight compared to that noted in the control mice without STZ injection (Fig. 1F). For the selective analysis of microglial inflammatory reactions in the hyperglycemic and normoglycemic mice, we isolated microglial cells from the murine spinal cord using a flow cytometer after SCI, as described in our earlier studies (3, 14). Briefly, CD45-positive cells obtained from the spinal cord tissue were subdivided into three populations on the basis of expression of the surface markers CD11b/Gr-1/CD45. The CD11bhigh/Gr-1nega-int/CD45int population was isolated as the resident microglia (Fig. 1G). Double immunofluorescence studies then showed the colocalization of CD68 with Iba-1 (a marker of reactive microglia) (11) in the isolated microglia, indicating that the microglia had transformed into an activated state after SCI (Fig. 1H). Although the hyperglycemic conditions did not affect the number of microglia, the hyperglycemia significantly altered the microglial activity after SCI (Fig. 1, I and J; P < 0.05). Notably, whereas, before SCI, the microglial expression of TNF-α, IL-6, and IL-1β was comparable between the hyperglycemic and normoglycemic groups, significantly increased expression of these molecules was observed in both groups after SCI; in particular, there was a significant difference between the two groups (Fig. 1J; P < 0.05). These results indicate that hyperglycemia increases the inflammatory cytokine expression by activated microglia.

Fig. 1. Isolation and gene expression analysis of the microglia obtained from the injured spinal cord of mice with acute SCI.

(A) LPS-induced morphological changes in the BV-2 cells 3 hours after administration, stained with CD68 (green) and Hoechst (blue). (B and C) TNF-α and IL-6 concentrations 24 hours after LPS stimulation of BV-2 cells in the normal-glucose (5.6 mM) or high-glucose (12.5 or 25 mM) medium (n = 8 per group). (D and E) The immunohistochemical analysis of insulin-positive β cells in the pancreas of the naïve (D) or STZ-treated mice (E) 7 days after injection of STZ. (F) Blood glucose concentrations and body weights of the naïve and STZ-treated mice before or 7 days after the injection of STZ (n = 8 per group). (G) Flow cytometric analysis. The CD45high/CD11bposi/Gr-1nega-int monocyte/macrophage fraction in mouse peripheral blood (left). The CD45int/CD11bposi/Gr-1nega-int resident microglial fraction in the uninjured spinal cord (middle). The resident microglial fraction (lower box) and monocyte/macrophage fraction (upper box) in the injured spinal cord (right). (H) Immunocytochemical analysis of the FACS (fluorescence-activated cell sorting)–sorted microglia 1 day after isolation, stained with CD68 (green), Iba-1 (red), and Hoechst (blue). (I) Number of resident microglia in normoglycemic (NG) and hyperglycemic (HG) mice (n = 6 per group) counted using flow cytometry. (J) The mRNA expression of proinflammatory cytokines increased in the microglia of the hyperglycemic mice 12 hours post-injury (hpi) (n = 6 per group). Scale bars, 50 μm (A), 500 μm (E), and 20 μm (H). *P < 0.05, Dunnett’s test (B and C) and Wilcoxon rank sum test (F, I, and J); n.s., not significant. The error bars indicate the SEM.

Promoting nuclear factor-κ B translocation in activated microglia with high glucose

To clarify the regulatory mechanism underlying the hyperglycemia-induced overexpression of inflammatory cytokines by microglia, we investigated the role of the nuclear factor κB (NF-κB) pathway, because NF-κB is a representative transcription factor that promotes the expression of TNF-α, IL-6, and IL-1β (15). We subsequently performed an immunocytochemical analysis using BV-2 cells after LPS stimulation to determine the activation of NF-κB in the microglial cells. Microglial NF-κB was primarily localized in the cytoplasm while in the resting state, and nuclear translocation of NF-κB was observed immediately after LPS administration (Fig. 2A). We therefore speculated that NF-κB directly binds to the promoter, thus bringing about the subsequent expression of inflammatory cytokines after nuclear translocation, and we performed a chromatin immunoprecipitation (ChIP) assay using the NF-κB p65 antibody. The primers were designed to target the promoter region of TNF-α as well as the negative control region (Fig. 2B). With these primers, the ChIP–PCR (polymerase chain reaction) analysis showed NF-κB binding to the TNF-α promoter shortly after LPS stimulation, indicating the direct regulation of TNF-α expression by NF-κB (Fig. 2C). To compare the level of activation of NF-κB in the microglial cells between the normoglycemic and hyperglycemic conditions, we quantified the amount of whole cell and nuclear NF-κB using a Western blot assay. A densitometric analysis showed a comparable NF-κB p65 expression in the whole-cell extract under glycemic conditions at 30 min after LPS stimulation. In contrast, there was a significant increase in p65 translocation to the microglial nuclei in a glucose concentration–dependent manner accompanied by an increased expression of TNF-α (Fig. 2, D to I; P < 0.05). These data suggest that the hyperglycemic conditions did not affect the inactivated microglia, whereas they promoted the nuclear translocation of NF-κB in the activated microglia, resulting in overexpression of TNF-α. In addition to the in vitro results, we confirmed that hyperglycemia enhanced the nuclear translocation of NF-κB in vivo in the activated microglial cells after SCI. Furthermore, the immunohistological analysis showed that the number of nuclear NF-κB/Iba-1 double-positive microglial cells was significantly greater in the lesions of the hyperglycemic mice compared to that observed in the normoglycemic mice at 12 hpi (Fig. 3, A to C; P < 0.05). Consistent with these results, quantitative reverse transcription PCR (RT-PCR) demonstrated the expression of TNF-α and IL-6 to be significantly higher in the spinal cords of the hyperglycemic mice at 12 hpi than in the normoglycemic mice at the same time point (Fig. 3D; P < 0.05).

Fig. 2. High-glucose conditions resulted in increased nuclear translocation of NF-κB p65 in activated mouse microglial cells.

(A) The immunocytochemical analysis of the p65 translocation into the nuclei of the microglial cells 30 min after LPS stimulation, stained with CD68 (green), NF-κB p65 (red), and Hoechst (blue). Scale bar, 25 μm. (B) Schematic diagram illustrating the primer set sites in relation to the TNF-α promoter. ATG indicates the gene transcription start site. The negative control is located upstream of the promoter. (C) ChIP-PCR analysis. The binding of p65 at the putative NF-κB binding site after LPS stimulation. (D to G) Densitometric scanning of the immunoblots of cellular p65 (D and F) and nuclear p65 (E and G) in the BV-2 cells compared among the three different glucose concentrations in the medium. The data were measured before or after LPS stimulation (n = 6 per group). (H and I) Expression of proinflammatory cytokines by BV-2 cells in the hyperglycemic and normoglycemic medium (n = 7 per group). *P < 0.05, Dunnett’s test. The error bars indicate the SEM.

Fig. 3. Expression of NF-κB p65 and inflammatory cytokine expression in microglia of hyperglycemic mice after SCI.

(A and B) Immunohistochemical analysis of NF-κB p65 in the Iba-1–positive activated microglia (arrowhead) in the lesions of normoglycemic mice (A) and hyperglycemic mice (B) at 12 hpi. Inset: nuclear localization of p65. The nucleus was counterstained with Hoechst 33258 dye (blue). (C) Quantification of nuclear p65-positive microglia in the normoglycemic and hyperglycemic mice at 12 hpi (n = 6 per group). (D) Gene expression of proinflammatory cytokines in the naïve (SCI −) and injured (SCI +) spinal cords of the normoglycemic and hyperglycemic mice at 12 hpi, as determined according to quantitative PCR (qPCR) (n = 6 per group). Scale bars, 50 μm. *P < 0.05, Wilcoxon rank sum test. The error bars indicate the SEM.

Increased apoptosis and poor functional recovery after SCI in hyperglycemic mice

Because TNF-α is known to induce the apoptosis of neurons and oligodendrocytes via the caspase-8/caspase-3 pathway (14), the activation of the apoptotic cascade was examined in the SCI mice. We subsequently confirmed the presence of glutathione S-transferase-π (GST-π) (a marker of mature oligodendrocytes) and cleaved caspase-3/caspase-8 (activated form of caspase-3/caspase-8) double-immunostained cells around the lesions at 4 days post-injury (dpi) (Fig. 4, A and B). Notably, Western blot analysis revealed that the expression of cleaved caspase-8 was significantly increased in the hyperglycemic mice compared to that observed in the normoglycemic mice (Fig. 4, C and D; P < 0.05). Moreover, terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate nick end labeling (TUNEL) staining demonstrated that the number of apoptotic cells around the epicenter was significantly greater in the hyperglycemic mice than in the normoglycemic mice (Fig. 4, E and F; P < 0.05). In addition to the extrinsic apoptosis pathway mediated by caspase-8, there is another apoptosis pathway (the intrinsic pathway) mediated by caspase-9 and Bcl-xL (3). We therefore evaluated the expression of the factors involved in both the extrinsic and the intrinsic apoptosis pathways. Whereas the expression of caspase-8 and caspase-3 was significantly increased in the injured spinal cords of the hyperglycemic mice compared to that seen in the normoglycemic mice, the expression of caspase-9 and Bcl-xL was comparable between the two groups (Fig. 4G), thus suggesting that hyperglycemia promotes neuronal apoptosis not via the mitochondrial intrinsic pathway but rather via the TNF-mediated extrinsic pathway in acute SCI mice.

Fig. 4. Hyperglycemic mice displayed a larger increase in the number of apoptotic cells and a poorer functional recovery than normoglycemic mice after acute SCI.

(A and B) GST-π and cleaved caspase-3 (A) or cleaved caspase-8 (B) double-positive oligodendrocytes in the perilesional areas. (C and D) Western blot analysis of cleaved caspase-8 at 4 dpi (n = 6 per group). (E) TUNEL staining of the sections at 4 dpi. The asterisk indicates the epicenter of the lesion. (F) Quantification of the TUNEL-positive apoptotic cells in the injured spinal cord at 4 dpi (n = 5 per group). (G) Expression of apoptosis-related genes at 12 hpi (n = 5 per group). (H) Blood glucose concentrations 30 min after injection of glucose (n = 5 per group). (I) Gene expression of TNF-α in the microglial cells isolated from the glucose- or saline-injected mice at 12 hpi (n = 5 per group). (J) TUNEL staining of the injured spinal cord in the glucose- or saline-injected mice at 12 hpi. The asterisk indicates the epicenter of the lesion. (K) Quantification of the TUNEL staining in the glucose-injected hyperglycemic or normoglycemic mice at 12 hpi (n = 5 per group). (L and M) Luxol fast blue (LFB)–stained cross-sections of mouse spinal cord (L) and the quantification of the area of LFB-positive spared myelin at 7 dpi (M) (n = 5 per group). (N) Basso Mouse Scale (BMS) scores in the sham-operated mice after the induction of hyperglycemia (n = 5 per group). (O) Time course of the functional recovery of the BMS score after SCI (n = 13 per group). (P) BMS score of each animal 6 weeks after SCI. Scale bars, 50 μm (A and B) and 500 μm (E, J, and L). *P < 0.05, Wilcoxon rank sum test (D, F to I, K, M, and N), two-way repeated-measures analysis of variance (ANOVA) with the Tukey-Kramer post hoc test (O), ANOVA with the Tukey-Kramer post hoc test (P). The error bars indicate the SEM. N/S, normal saline-injected; Glu, glucose-injected.

To exclude the possibility that the increased secondary injury noted in the hyperglycemic mice could be attributed to STZ-induced β cell necrosis, an alternative acute hyperglycemic mouse model was produced using the intraperitoneal injection of 5% glucose immediately after SCI (0.05 ml/g of body mass), as previously described (16). This mouse model showed markedly increased blood glucose concentrations (Fig. 4H). Consistent with the results of the STZ-induced hyperglycemic model, a significantly increased TNF-α expression and significantly elevated number of apoptotic cells were observed in the glucose-injected hyperglycemic mice compared to that noted in the normoglycemic mice (Fig. 4, I to K; P < 0.05). The apoptosis of oligodendrocytes after SCI is reported to be associated with demyelination of the injured spinal cord and decreased functional recovery (17). Here, more severe demyelination and worse functional outcomes were observed in the STZ-induced and glucose-injected hyperglycemic mice after SCI than in the normoglycemic mice. A significantly lower LFB staining ratio was observed in the hyperglycemic mice in the sections from 1000 μm rostral to 1000 μm caudal to the epicenter (Fig. 4, L and M; P < 0.05). BMS scores for uninjured mice with hyperglycemia were comparable to that for uninjured mice without hyperglycemia (Fig. 4N), but after SCI, the hyperglycemic mice showed markedly impaired recovery of motor function (Fig. 4, O and P; P < 0.05). These results indicate that hyperglycemia has a detrimental effect on the secondary injury process and impairs functional recovery after acute SCI in mice.

Poor functional outcomes in SCI patients with high blood glucose at admission

Although the experimental hyperglycemic mouse models showed exacerbation of the pathology of SCI, the effects of hyperglycemia on the functional outcomes of human SCI remain unclear. To address this issue, we analyzed a human cohort of 528 SCI patients focusing on the relationship between blood glucose concentration at admission and functional outcome. After applying the inclusion/exclusion criteria, the data for 206 patients admitted within 24 hpi were evaluated (Fig. 5). After dividing the patients into good and poor outcome groups based on the American Spinal Injury Association (ASIA) Impairment Scale (AIS) grade at the final follow-up, we compared the two groups according to their blood glucose concentrations at admission. Although variations in the type of injury did not affect the admission blood glucose concentration, the poor outcome group (AIS A, B, and C) had significantly higher admission blood glucose concentrations than the good outcome group (AIS D and E) [median (interquartile range), 138 mg/dl (116.5 to 169.5) versus 120.5 mg/dl (102.5 to 140.5); P < 0.0005] (Fig. 6A and fig. S1). For further analyses, we defined hyperglycemia in the acute SCI patients as a blood glucose concentration of ≥126 mg/dl, in accordance with the internationally accepted standard diagnostic criteria for diabetes mellitus (18). The hyperglycemia was prolonged for 5.7 ± 1.2 days. The hyperglycemic and normoglycemic groups had similar baseline characteristics and metabolic and serological data (Tables 1 and 2). In this population, the presence of hyperglycemia on admission was associated with an increased risk of an unfavorable outcome (AIS A, B, and C) [odds ratio, 2.66; 95% confidence interval (CI), 1.52 to 4.72]. In addition, we used alternative cutoff points, 150 and 180 mg/dl, associated with the prognosis of sepsis and cardiovascular disease, respectively (18, 19). With these additional cutoff points, the odds ratios for an unfavorable outcome were 2.62 (95% CI, 1.41 to 4.88) and 3.70 (95% CI, 1.60 to 8.98), respectively, whereas that obtained using a cutoff point of 100 mg/dl was not significant (Fig. 6B). These results indicate that the admission blood glucose concentration was strongly associated with the functional outcome after SCI and that a blood glucose concentration of 126 mg/dl is an appropriate cutoff point for determining the risk of hyperglycemia associated with a poor functional outcome. We also examined the relationships between the admission blood glucose concentration values and other functional/clinical outcome measurements, including an improvement in the AIS grade, defined as moving up at least one level (for example, AIS A to AIS B/C/D/E or AIS B to AIS C/D/E). Consequently, the presence of hyperglycemia (≥126 mg/dl) on admission was found to be strongly associated with a lower probability of improvement in the AIS grade (Fig. 6C; P < 0.05), with a significantly lower ASIA motor score (Fig. 6D; P < 0.05) and a significantly lower ASIA motor score recovery rate (Fig. 6E; P < 0.05) (20). To further investigate the degree of improvement in the AIS grade among the hyperglycemic patients, the AIS improvement scores were determined as previously described (21) (see Materials and Methods for details). A smaller percentage of the hyperglycemic patients had an AIS improvement score of 1, 2, or 3 compared to that observed among the normoglycemic patients, respectively, indicating that the hyperglycemic patients showed a smaller improvement in the AIS (Table 3). Furthermore, whereas the total spinal cord independence measure (SCIM) scores [a disability score based on activities of daily living (ADLs)] (22) at admission did not differ significantly between the hyperglycemic and normoglycemic patients, the hyperglycemic patients exhibited significantly poorer SCIM scores at discharge (Fig. 6F) as well as significantly lower changes in the SCIM score from admission to discharge (Fig. 6G; P < 0.001). These results indicate that hyperglycemia at admission is a consistent risk factor for a poor functional outcome/recovery and poor ADL outcome in SCI patients.

Fig. 5. Patient enrollment chart.

The patient groups selected for the statistical analysis were determined with respect to the characteristics of particular outcome parameters (the AIS, ASIA motor, or total SCIM score).

Fig. 6. Hyperglycemia is a risk factor for a poor clinical outcome after human SCI.

(A) Admission blood glucose concentrations of the patients with poor outcomes (AIS A, B, and C) or good outcomes (AIS D and E). (B) Risk of a poor outcome after SCI among the patients with a high admission blood glucose concentration (≥126, ≥150, and ≥180 mg/dl). (C) Probability of improvement in the AIS of the patients with a high admission blood glucose concentration (≥126 mg/dl; hyperglycemic) or a normal admission blood glucose concentration (≤125 mg/dl; normoglycemic). (D) ASIA motor scores in the hyperglycemic and normoglycemic patients at discharge. (E) ASIA recovery rates among the hyperglycemic and normoglycemic patients. (F and G) Total SCIM scores at discharge (F) and changes in the SCIM scores in the hyperglycemic and normoglycemic patients. (H) Functional outcomes (AIS) of the patients treated with or without steroids (Steroid + or Steroid −). (I) Admission blood glucose concentrations of the patients treated with or without steroids (Steroid + or Steroid −). *P < 0.05, **P < 0.001, ***P < 0.0005, Wilcoxon rank sum test (A, D, E, F, G, and I) and χ2 test (B, C, and H).

Table 1 Subject baseline characteristics.

NS, not significant (P > 0.05), Wilcoxon rank sum test, or χ2 test.

View this table:
Table 2 Metabolic and serological data.

Values are means ± SEM. BMI, body mass index; NS, not significant (P > 0.05), Wilcoxon rank sum test.

View this table:
Table 3 Number and percentage of patients with an AIS improvement score of 0 to 3 in the normoglycemic (NG) and hyperglycemic (HG) patient groups.
View this table:

Although high-dose methylprednisolone is frequently used to treat acute SCI based on the NASCIS II protocol, its efficacy has recently become controversial, and thus, its administration was left to the judgment of the attending physician (23). Indeed, in this study, no patients received treatment with methylprednisolone at our hospital, although 38 of the 206 SCI patients were given methylprednisolone treatment at the attending physician’s discretion in the hospital to which the patient was initially transported. In this cohort, methylprednisolone administration did not significantly improve the functional outcome after SCI (P > 0.05; Fig. 6H). However, upon admission to our hospital, a significant increase in the blood glucose concentration, a known side effect of glucocorticoids (24), was noted among the patients treated with methylprednisolone (Fig. 6I; P < 0.0005).

Diabetes mellitus, which is characterized by chronic hyperglycemia, is associated with both macrovascular and microvascular complications (25) and may affect plasticity as well as the CNS (26). To examine the effects of chronic hyperglycemia on functional outcomes, we analyzed the relationships between the ASIA motor score/ASIA motor score recovery rate and the level of glycosylated hemoglobin (HbA1c), which reflects the degree of glycemia before the onset of SCI (27). The HbA1c level exhibited negative relationships with both the ASIA motor score at discharge and the recovery rate (fig. S2, A and B). In addition, the diabetic patients with a poor functional outcome had significantly higher admission HbA1c than did those with a good functional outcome at the final follow-up (fig. S2C). Furthermore, there was a significant positive relationship between HbA1c and the admission blood glucose concentration (fig. S2D). These results raised the possibility that the worse outcomes associated with admission hyperglycemia were simply due to the presence of chronic hyperglycemia in the patients with diabetes mellitus. Therefore, the diabetic patients were excluded to study only the influence of acute hyperglycemia on SCI outcomes. Among the nondiabetic population (n = 181), hyperglycemia was prolonged for 2.7 ± 0.5 days, and all patients with hyperglycemia regained a normal blood glucose concentration within 42 dpi. In addition, consistent with the findings of the analysis including the diabetic subjects, the patients with poor outcomes had significantly higher admission blood glucose concentrations than did those with good outcomes (Fig. 7A; P < 0.0005). Conversely, the patients with a high blood glucose concentration also had an increased risk of an unfavorable functional outcome (Fig. 7B). In addition, the hyperglycemic patients demonstrated lower ASIA motor scores, ASIA motor score recovery rates, SCIM scores, and changes in the SCIM scores than did the normoglycemic patients (Fig. 7, C to F), thus indicating that acute hyperglycemia without diabetes mellitus remained a risk factor for a poor functional outcome in SCI patients.

Fig. 7. Even after excluding the diabetic subjects, hyperglycemia was found to be associated with a poor functional outcome after human SCI.

(A) Admission blood glucose concentrations of the patients with poor outcomes (AIS A, B, and C) or good outcomes (AIS D and E) after excluding diabetics. (B to F) Risk of a poor functional outcome (AIS) (B), ASIA motor scores at discharge (C), ASIA recovery rate (D), total SCIM scores on admission and at discharge (E), and changes in the SCIM scores (F) in the hyperglycemic and normoglycemic patients without diabetes mellitus. *P < 0.05, **P < 0.005, ***P < 0.0005, Wilcoxon rank sum test (A, C, D, E, and F) and χ2 test (B).

To date, age (4), blood pressure (5), steroid administration (28), and the severity of initial paralysis (6) have been postulated to be potential predictors of the motor function after SCI. In the current cohort, the systolic blood pressure (SBP) on admission and severity of initial paralysis were also found to be significant risk factors (Table 4). There is a possibility that these factors may be confounders correlated with both the admission blood glucose concentration and functional outcome of SCI and that acute hyperglycemia is not actually a risk factor itself. Therefore, we conducted a multivariable linear regression analysis to examine whether hyperglycemia is indeed a significant independent risk factor or merely a confounder. Consequently, whereas the SBP was determined not to be a significant independent predictor, admission hyperglycemia was identified as an independent risk factor for a poor motor outcome in the SCI patients (regression coefficient, −1.37; 95% CI, −2.65 to −0.10; P < 0.05; Table 4).

Table 4 Linear regression analysis of the functional outcomes of the patients without diabetes mellitus using the ASIA motor score.

The results of the univariate and multivariate linear regression analysis in cervical injury patients admitted within 24 hours after injury (n = 133) were shown. The multiple analysis was performed after forward variable selection (R2 = 0.54). The cohort was controlled for “paralysis at admission” (AIS) and “systolic blood pressure” (SBP). Only blood glucose concentrations and paralysis at admission were proven to be independently associated with an impaired ASIA motor score outcome, yielding statistical significance in the multivariable model. β, unstandardized regression coefficient; M, motor vehicle collision; F, fall from height; O, other accidents.

View this table:

Preventing the detrimental effects of hyperglycemia by manipulating blood glucose concentration in acute SCI

In our animal model experiments and human cohort study, we demonstrated a correlation between acute hyperglycemia and a poor neurological outcome after SCI. With the aim of further clarifying the causal relationship, we examined whether treating hyperglycemia with a long-acting insulin injection prevented the exacerbation of SCI outcomes in the STZ-induced hyperglycemic mice. First, to determine the required duration of treatment, we produced two insulin-injected groups: the “1-day treatment” group and the “4-day treatment” group, in which insulin was subcutaneously administered immediately after SCI in the former and given daily starting immediately to 4 dpi in the latter (Fig. 8A). Consequently, normal blood glucose concentration was successfully achieved in these groups at 24 and 96 hpi, respectively (Fig. 8B). In addition, the BMS scores for motor function in these mice were comparable to those observed in the normoglycemic mice and significantly better than those noted in the untreated hyperglycemic mice at 2 weeks after SCI (Fig. 8C). Histopathologically, significantly less apoptosis was observed in the insulin-treated hyperglycemic mice than in the untreated hyperglycemic mice (P < 0.05), although there were no significant differences in the number of apoptotic cells between the two insulin-injected groups and the normoglycemic group (Fig. 8, D and F). Moreover, demyelination was reduced in the hyperglycemic mice with 1-day treatment compared to that observed in the untreated hyperglycemic mice (Fig. 8, E and G). These results demonstrate that achieving glycemic control for 1 day is sufficient to rescue the exacerbation of the functional outcome and pathology in hyperglycemic mice after SCI.

Fig. 8. Glycemic control in the hyperglycemic mice prevented deterioration of the pathophysiology and functional outcomes after SCI.

(A) Schedule of glycemic control achieved with insulin injection. (B) Time course of the blood glucose concentrations in the hyperglycemic mice after insulin treatment. The black (insulin 1 d) and gray (insulin 4 d) arrows indicate the time of insulin injection. (C) Functional outcomes measured according to the BMS score 2 weeks after SCI (n = 13 per group). (D) TUNEL staining of the sections of injured mouse spinal cord at 4 dpi. The asterisk indicates the epicenter of the lesion. (E) LFB-stained cross-sections of the tissue obtained from hyperglycemic mice and hyperglycemic mice treated with 1-day insulin injection at 7 dpi. (F) Quantification of the TUNEL-positive apoptotic cells in the injured spinal cord at 4 dpi (n = 5 per group). (G) Quantification of the area of LFB-positive spared myelin at 7 dpi (n = 6 per group). (H) Time course of TNF-α mRNA expression in the microglial cells of the hyperglycemic mice after insulin injection. The arrows indicate the time of insulin injection. (I) BMS scores of the hyperglycemic mice treated with insulin for glycemic control starting at 0, 8, and 20 hpi measured 2 weeks after SCI (n = 13 per group). (J) A footprint analysis was performed 2 weeks after SCI. Both the hyperglycemic mice treated without insulin and the hyperglycemic mice treated with insulin after 20 hpi were unable to consistently step. (K) Scores for the grip walk test in each mouse 3 weeks after SCI. The error bars indicate SEM. Scale bars, 500 μm. *P < 0.05, ANOVA with the Tukey-Kramer post hoc test (C, F, I, and K), Wilcoxon rank sum test (G), and a two-way repeated-measures ANOVA with the Tukey-Kramer post hoc test (H).

Therapeutically, identifying the possible time window in which glycemic control can improve functional outcome is particularly important. Therefore, to determine the therapeutic time window, we injected insulin into the hyperglycemic mice at several time points after injury and examined the changes in both functional recovery and gene expression. First, we started by performing a gene expression analysis of the microglia in the insulin-treated hyperglycemic mice. As shown in Figs. 1 and 4, TNF-α was overexpressed in the overactivated microglia of the hyperglycemic mice after injury, which led to increased apoptosis of neural cells and poor functional outcomes (Figs. 1 and 4). These results suggest that manipulating the blood glucose concentration may help to prevent TNF-α overexpression in microglia and thus improve the functional outcome after SCI. Therefore, with the purpose of identifying the time at which glycemic control should be initiated to inhibit microglial overactivation, we measured gene expression in the microglia at several time points after SCI. Higher expression of TNF-α was observed at 12 and 24 hpi in the microglia of the hyperglycemic mice. Then, to normalize the high blood glucose concentration noted at 12 and 24 hpi, we injected insulin into the hyperglycemic mice at 0, 8, and 20 hpi. Within 2 hours after injection, normoglycemia (blood glucose concentration <126 mg/dl) was achieved until 24 hpi (fig. S3). In addition, after treatment at 0 or 8 hpi, the microglial TNF-α expression in the hyperglycemic mice was as low as that seen in the normoglycemic mice at 12 and 24 hpi. Similarly, when glycemic control was started at 20 hpi, TNF-α expression at 24 hpi in the hyperglycemic mice was comparable to that measured in the normoglycemic mice, although the expression at 12 hpi in the hyperglycemic mice was higher than that observed in the normoglycemic mice (Fig. 8H). These results indicate that the therapeutic interventions administered at 0, 8, and 20 hpi successfully suppressed TNF-α overexpression. With treatment at 0 and 8 hpi, the hyperglycemic mice had significantly better functional outcomes than the untreated hyperglycemic mice according to the BMS score (Fig. 8I; P < 0.05) and according to more detailed analyses such as the footprint analysis (4) and grip walk test (29) (Fig. 8, J and K; P < 0.05). Moreover, comparable scores were found in the normoglycemic mice and hyperglycemic mice treated at 0 or 8 hpi. Meanwhile, treatment at 20 hpi failed to significantly improve functional outcomes (Fig. 8, I to K, and fig. S4; P > 0.05). These results indicate that there is a time window in which manipulating the blood glucose concentration can improve the functional outcome in hyperglycemic mice after SCI, and that, at least when glycemic control is started by 8 hpi, treated hyperglycemic mice exhibit better functional outcomes than do untreated hyperglycemic mice. Although the time window of 8 hours may seem short, the feasibility of achieving glycemic control in patients with acute SCI is supported by the fact that high-dose methylprednisolone is used within 8 hpi according to the NASCIS II protocol (28).

DISCUSSION

Here, animal model experiments and a human cohort study were performed to investigate the influence of hyperglycemia in acute SCI. In the animal model, we found that hyperglycemia induced the overactivation of NF-κB in microglial cells as well as increased inflammation, resulting in a poor functional recovery after SCI. In addition, the detrimental effects of hyperglycemia were prevented by manipulating the blood glucose concentration in the acute phase of SCI. Furthermore, a retrospective analysis of acute SCI patients revealed that hyperglycemia was associated with a worse functional outcome and was identified as a potential poor prognostic factor for human SCI, irrespective of whether the subjects had a history of diabetes mellitus. These findings shed light on the importance of achieving tight glycemic control in acute human SCI to obtain better neurological outcomes.

An animal model of hyperglycemia lasting for as long as 8 to 12 weeks is commonly used in studies of diabetes mellitus (25). This long-term hyperglycemic animal model involves varying degrees of vascular lesions with diffuse degenerative changes in multiple organs, including the CNS. In addition, the activation of microglia is prominently observed, even without trauma, because of the accumulation of glycated proteins formed after long-term hyperglycemia (25). Therefore, we used a short-term acute hyperglycemic animal model to assess the direct influence of hyperglycemia on the pathophysiology of SCI. The STZ-induced mouse model is used as short as 1 week after the development of hyperglycemia, and the glucose-administered mouse model involves the onset of hyperglycemia immediately after SCI, as previously reported (13, 16). In these acute hyperglycemic models, we observed no activation of resting microglia without SCI (Figs. 1 and 4I), suggesting that the mechanism underlying the microglial overactivation observed in the acute hyperglycemic mice is a process distinct from that noted in diabetes mellitus.

Here, the microglial activation was accompanied by immediate nuclear translocation of NF-κB (Fig. 2A). Considering our finding that the cellular and nuclear expression of NF-κB was comparable between the microglia cultured under hyperglycemic and normoglycemic conditions before LPS stimulation, acute hyperglycemia alone does not appear to be involved in microglial activation. However, in the inflammatory state, acute hyperglycemia enhanced the nuclear translocation of NF-κB (Fig. 2, D to G), which resulted in the increased expression of proinflammatory cytokines (Figs. 1J and 2, H and I). A ChIP-PCR analysis demonstrated NF-κB binding to the TNF-α promoter (Fig. 2C), indicating that NF-κB directly regulates TNF-α transcription in microglial cells. TNF-α is reported to induce the apoptosis of neural cells, including neurons and oligodendrocytes, after SCI (2). Therefore, the excessive TNF-α expression observed in the microglia brought about by the activation of NF-κB in the hyperglycemic state is expected to contribute to a poor functional outcome after SCI, with increased apoptosis of neural cells (Figs. 3 and 4).

With regard to the mechanisms involved in the hyperglycemia-related overactivation of NF-κB in microglia, NADPH (reduced form of nicotinamide adenine dinucleotide phosphate) oxidase is considered to possibly play a role. NADPH oxidase is present in several types of phagocytes, including microglia, causing inflammatory activation of these cells (30). In addition, several studies have reported that hyperglycemia enhances the NADPH oxidase activity in innate immune cells (31, 32). Moreover, it is known that NADPH oxidase produces reactive oxygen species (ROS) (33), which may promote the translocation of NF-κB (34). Therefore, hyperglycemia may promote the translocation of NF-κB in microglial cells via the NADPH oxidase/ROS/NF-κB pathway. The fact that an increased expression of NADPH oxidase and ROS has been confirmed in spinal microglial cells after injury also supports the role of this pathway (35).

The results of our animal experiments led us to analyze the effects of hyperglycemia on the clinical functional outcomes of SCI patients. The adjustment of potential confounding factors is required in human cohort studies to obtain accurate results (6). In our series, we found a positive correlation between the SBP on admission and the motor score outcome in the univariate analysis, but not the multivariate analysis (Table 4). This discrepancy suggests that SBP may be a confounder of the association between the severity of initial paralysis and the functional outcome. It is well known that spinal neurogenic shock leads to systemic hypotension after SCI and that hypotension is associated with the severity of the initial injury (5). We found a significant negative correlation between the SBP and the severity of paralysis on admission (fig. S5A). On the other hand, there were no correlations between the blood glucose concentration and the severity of paralysis on admission (fig. S5B). In addition, even after adjusting for confounding factors in the multivariate regression analysis, hyperglycemia in the setting of acute SCI was found to be an independent risk factor for a poor functional outcome (Table 4). Furthermore, we investigated the direct relationship between the initial severity of injury and blood glucose concentration, because the peripheral inflammation after severe CNS injury, which is seen in the liver, kidneys, and lungs, might also affect blood glucose concentration (36, 37). After the induction of different severities of SCI, no significant changes in blood glucose concentrations were observed (fig. S5C). These results ruled out the possibility that the hyperglycemia in the acute phase of SCI might be due to the severity of paralysis at admission. These findings confirm that hyperglycemia in acute SCI was not a confounder in this study.

The onset of hyperglycemia in the acute phase of human SCI may be caused by treatment with methylprednisolone, stress (38), and/or preexisting diabetes mellitus. Methylprednisolone is often administered in acute SCI patients to reduce the degree of secondary injury and improve the functional outcome; however, its efficacy remains controversial (23, 28). Remarkably, it has been reported that hyperglycemia occurs in 88% of SCI patients treated with methylprednisolone, and, most notably, these patients exhibit no significant improvements in their functional outcomes compared to patients treated without methylprednisolone (24). Consistent with the findings of that report, we showed that methylprednisolone treatment does not improve the functional outcomes (Fig. 6H), but rather causes a significant increase in blood glucose concentration (Fig. 6I). In light of our results that the presence of hyperglycemia in acute SCI may be detrimental for functional outcomes, the administration of methylprednisolone treatment requires attention to the blood glucose concentration.

Tight glycemic control during acute clinical presentation is reported to improve the prognoses of diseases such as myocardial infarction (39), kidney damage (40), brain ischemia (41, 42), and liver dysfunction (43). In addition, achieving tight glycemic control (blood glucose concentration <110 mg/dl) reduces both the morbidity and the mortality of critically ill patients (44). Nevertheless, in the management of acute SCI, inadequate attention has been paid to the blood glucose concentration, and, in fact, to date, there have been no reports of the effects of acute glycemic control on SCI outcomes. Moreover, although it is recommended to maintain the SBP, which was revealed to not be an independent predictor of the functional outcome, more than 90 mmHg in the acute phase of SCI, guidelines on the management of SCI have not indicated the need for glycemic control (5). However, the results of the present cohort study suggest that achieving strict glycemic control is more important than controlling the SBP in patients with acute SCI.

There are several potential limitations associated with our study. First, the cohort study was a nonrandomized and retrospective study of SCI. Consequently, it could not be completely ruled out that the variations in the injury might have affected the blood glucose concentration. Second, the cohort study was limited in size. Larger sample sizes are required to elucidate the impact of blood glucose concentration on the prognosis of SCI. Third, the human study was observational. A prospective and randomized study is needed to confirm the clinical efficacy of glycemic control in patients with acute hyperglycemia after SCI.

In conclusion, we demonstrated that hyperglycemia increases inflammatory responses in microglial cells via NF-κB activation and exacerbates secondary injury, thus resulting in a poorer outcome after SCI. Manipulation of the blood glucose concentration in the acute phase prevented an exacerbation of the pathophysiology and improved motor outcomes of the hyperglycemic mice after SCI. Furthermore, our results suggest that the presence of hyperglycemia may be a poor prognostic factor for acute human SCI. Our findings provide deeper insight into the effects of hyperglycemia on CNS injury and indicate that tight glycemic control may improve patient outcomes during acute SCI.

MATERIALS AND METHODS

Study design

The purpose of this study was to determine the effects of acute hyperglycemia on the functional outcomes after SCI using animal experimental models and the clinical data of SCI patients. The experimental endpoints (typically 14, 35, or 42 dpi) were chosen on the basis of the findings of our previous studies, allowing for sufficient time to assess the motor outcomes (3, 14). Each experiment was independently replicated, and the number of experimental replicates is noted in the corresponding figure legends. We retrospectively identified 528 SCI patients admitted to the Department of Orthopaedic Surgery at the Spinal Injuries Center (Fukuoka, Japan) between June 2005 and May 2011. The patients’ data were obtained from their charts. The study involving the analysis of medical record data was approved by the Spinal Injuries Center Institutional Review Board. The inclusion criteria were as follows: (i) admission to our institution within 24 hpi; (ii) undergoing in-patient rehabilitation at our institution; and (iii) an adequate assessment with the AIS (21). The exclusion criteria were as follows: (i) death or transfer to another hospital before in-patient rehabilitation; (ii) known or clinical signs of concomitant cerebral damage; (iii) noncooperation with the functional assessment because of dementia; (iv) a history of neurological disorders other than SCI; and (v) an AIS score of E on admission (Fig. 5). A total of 206 patients met these criteria, and the AIS, ASIA motor (20), and SCIM scores (for ADLs) (22) obtained on the final examination performed on the day of discharge were used as the outcome measurements. The blood samples of the patients were corrected immediately after being transported to our hospital (within 24 hpi), and the blood glucose concentration was measured in each sample. An elderly age was defined as 65 years or older. In our institute, patients with spinal fractures, dislocation, or severe instability are treated surgically.

Induction of mouse hyperglycemia

All study protocols involving mice were approved by the Committee of Ethics on Animal Experimentation of the Faculty of Medicine, Kyushu University, and conducted in accordance with the National Institutes of Health guidelines for the care and use of animals. Eight-week-old female C57BL/6J mice received a single intraperitoneal injection of STZ (180 mg/kg) (Sigma). The control mice were injected with the equivalent volume of citrate buffer. The blood glucose concentrations were measured using the OneTouch Ultra device (LifeScan) at 7 days after injection, and a blood glucose concentration of ≥200 mg/dl was considered to indicate hyperglycemia. To produce another hyperglycemic model, mice received the intraperitoneal injection of 5% glucose (0.05 ml/g of body mass) immediately after SCI, as previously described (16). The control mice were injected with an equivalent volume of saline.

Spinal cord injury

The mice were anesthetized with pentobarbital (75 mg/kg intraperitoneally) and were subjected to a contusion injury (70 kdyn) at the 10th thoracic level using an Infinite Horizons Impactor (Precision Systems Instrumentation) (3). After injury, the overlying muscles were sutured, and the skin was closed with wound clips. During the period of recovery from anesthesia, the animals were placed in a temperature-controlled chamber until thermoregulation was reestablished. Sham-operated controls were subjected to laminectomy only. In the STZ-induced hyperglycemic mice and buffer-injected control mice, SCI was produced at 7 days after injection. To achieve glycemic control in the STZ-induced hyperglycemic mice, the long-acting insulin analog glargine (0.025 IU/body) was injected subcutaneously immediately after injury or at 8 or 20 hpi, as previously described (45), or administered immediately after injury and at 24, 48, and 72 hpi. The motor function was evaluated according to a locomotor open-field rating scale, the BMS (3). A footprint analysis was performed as previously described (4). The forelimbs and hindlimbs of the mice were dipped in red and green dyes, respectively. For the grip walk test (29), each mouse was evaluated using 50 cm of the grid with three patterns: easy (50 steps, 1 cm apart), medium (every third step was removed), and hard (every other step was removed). The sum of the number of grips for all three patterns was used in the analysis.

Flow cytometry

The spinal cords (6.0 mm in length, centered around the lesion) were dissociated in collagenase type I (Invitrogen) and stained with antibodies at 12 hpi, as previously described (14). The samples were stained with the antibodies described in the Supplementary Materials and Methods and analyzed using a FACSAria II flow cytometer and the FACSDiva software program (BD Biosciences). The isolated microglia were either subjected to RNA extraction or incubated in Dulbecco’s modified Eagle’s medium (DMEM) for immunocytochemistry.

Histopathological examination

The animals were reanesthetized and transcardially fixed with 4% paraformaldehyde. The spinal cord was removed and then dehydrated and embedded in an optimal cutting temperature (OCT) compound (14), and frozen sections were cut on a cryostat in the sagittal plane at 14 μm or the axial plane at 20 μm. The sections were subsequently stained with primary antibodies at 4°C overnight and then incubated with Alexa Fluor secondary antibodies (1:200; Invitrogen) and Hoechst 33258. The R.T.U. Vectastain kit (Vector Laboratories) was used with hematoxylin counterstaining to assess the paraffin-embedded sections of the pancreas, and an ApopTag Red In Situ kit (Chemicon International) was used for the TUNEL assay. LFB-stained serial sections were evaluated as previously described (14). All images were captured using a BZ-9000 digital microscope system (Keyence) or epifluorescence microscope equipped with a digital camera (BX51; Olympus). All primary antibodies used in this study are described in the Supplementary Materials and Methods.

Microglial cell line culture

The BV-2 murine microglial cell line was cultured in DMEM with 5% fetal bovine serum, 2 mM l-glutamine, and 1% penicillin-streptomycin, as previously described (46). BV-2 cells incubated in high-glucose medium (12.5 or 25 mM) or normal-glucose medium (5.6 mM) were treated with LPS at a concentration of 100 ng/ml and incubated for 30 min or 3 or 24 hours before immunocytochemical staining/Western blotting/ChIP assay, mRNA extraction, or ELISA, respectively.

Quantitative RT-PCR

Total RNA was isolated from the microglia obtained from the spinal cord tissue using the RNeasy Micro Kit (Qiagen) and from the injured spinal cord (4 mm in length) and BV-2 cells using the RNeasy Mini Kit (Qiagen). For complementary DNA synthesis, a reverse transcription reaction was performed using PrimeScript Reverse Transcriptase (TaKaRa). qPCR was performed using primers specific to the genes of interest (table S1) and SYBR Premix Dimmer-Eraser (TaKaRa). The data were normalized to the level of glyceraldehyde-3-phosphate dehydrogenase (GAPDH).

Western blotting analysis

Segments of the spinal cord (6 mm) were removed and homogenized, and nuclear extracts from BV-2 cells were prepared using a Dounce homogenizer with cold nuclear extraction buffer (47). The samples were then lysed in 2× SDS sample buffer, and SDS–polyacrylamide gel electrophoresis was performed as previously described (14), after which the membranes were immunoblotted with antibodies as described in the Supplementary Materials and Methods.

Enzyme-linked immunosorbent assay

Murine TNF-α and IL-6 were detected in the conditioned medium of the cells using ELISA (Quantikine, R&D Systems).

ChIP assay

BV-2 cells were treated with LPS at a concentration of 100 ng/ml and incubated for 30 min. The Magna ChIP A/G kit (Millipore) was used according to the manufacturer’s instructions. Sonication was performed with a VCX 130 device (Sonics & Materials Inc.). Anti–NF-κB p65 antibodies and normal mouse immunoglobulin G were used for immunoprecipitation. ChIP-PCR was performed with primers designed to span the putative NF-κB binding sites or upstream of the putative binding site, as previously described (47), using Mighty Amp DNA Polymerase (TaKaRa). The primers were as follows: mouse TNF −144/−346, 5′-AATGGGTTTCAGTTCTCAGGGTCC-3′ (forward) and 5′-CACCTCTGTCTCGGTTTCTTCTCC-3′ (reverse); negative control −1682/−1844, 5′-TGTTCCTGCTCAGTAAGGGAGACC-3′ (forward) and 5′-CCCTTGAAACAACGGTCAGGATGG-3′ (reverse).

Human subjects

The ASIA motor score recovery rate was calculated as previously reported (20): recovery rate = [(ASIA motor score at discharge) − (AISA motor score at admission)]/[100 − (ASIA motor score at admission)]. An improvement of one grade in the AIS was defined as an AIS improvement score of 1 (for example, AIS A to B, B to C, C to D, or D to E), as previously described (33). Similarly, a change from AIS A to C, B to D, or C to E was defined as an AIS improvement score of 2, and a change from AIS A to D or B to E was defined as an AIS improvement score of 3. A lack of improvement in the AIS was defined as an AIS improvement score of 0.

Statistical analysis

The Wilcoxon rank sum test was used to compare the medians of the data for qPCR as well as the cell count, blood glucose concentration, and functional outcome score. For multiple comparisons in ELISA, densitometry, and qPCR, Dunnett’s test was applied. The χ2 test was used for the statistical evaluation of group comparisons in cases of categorical variables, as shown in Table 1 and Figs. 6 and 7. For the analysis of the differences in the BMS scores between the groups over time, a two-way repeated-measures ANOVA with post hoc Tukey’s test was performed. All tests were two-sided, and the level of significance was set at 0.05. The values for groups are presented as the average ± SEM. All statistical analyses were carried out using the JMP software program (version 9; SAS Institute).

SUPPLEMENTARY MATERIALS

www.sciencetranslationalmedicine.org/cgi/content/full/6/256/256ra137/DC1

Materials and Methods

Fig. S1. The variation in injury exhibited no relationships with the admission blood glucose concentration.

Fig. S2. Chronic hyperglycemia with diabetes mellitus was associated with a poor clinical outcome after human SCI.

Fig. S3. Time course of the blood glucose concentrations in the hyperglycemic mice after insulin injection.

Fig. S4. Quantification of the footprint analysis after SCI.

Fig. S5. The severity of the initial paralysis due to SCI demonstrated no significant correlations with the blood glucose concentration on admission.

Table S1. Primers used for qPCR.

REFERENCES AND NOTES

  1. Funding: Supported by a Grant-in-Aid for Young Scientist (A, B), Japan Society for the Promotion of Science Fellows, Scientific Research on Innovative Areas (Comprehensive Brain Science Network and Foundation of Synapse and Neurocircuit Pathology), Challenging Exploratory Research from the Ministry of Education, Science, Sports and Culture of Japan, and Research Foundations from the General Insurance Association of Japan. Author contributions: K. Kobayakawa designed and performed most of the experiments with technical help from H.K., Y.O., and K.Y.; analyzed the data; and wrote the manuscript. H.S. performed flow cytometric analysis. J.K. participated in clinic study design and analysis. K. Kubota, R.I., and K.S. collected patient’s data. H.T.-S. and K.I. performed the in vitro studies. Y.I. designed the studies and supervised the overall project. S.O. designed the studies, supervised the overall project, and performed the final manuscript preparation. Competing interests: The authors declare that they have no competing interests.
View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article