Prognostic value of early leukocyte fluctuations for recovery from traumatic spinal cord injury

Abstract

Background: Acute traumatic spinal cord injury (SCI) induces a systemic immune response involving circulating white blood cells (WBCs). How this response is influenced by overall trauma severity, the neurological level of injury and/or correlates with patient outcomes is poorly understood. The objective of this study was to identify relationships between early changes in circulating WBCs, injury characteristics and long-term patient outcomes in individuals with traumatic SCI.

Methods: We retrospectively analysed data from 161 SCI patients admitted to Brisbane's Princess Alexandra Hospital (exploration cohort). Logistic regression models in conjunction with receiver operating characteristic (ROC) analyses were used to assess the strength of specific links between the WBC response, respiratory infection incidence and neurological outcomes (American Spinal Injury Association Impairment Scale (AIS) grade conversion). An independent validation cohort from the Trauma Hospital Berlin, Germany (n = 49) was then probed to assess the robustness of effects and disentangle centre effects.

Results: We find that the extent of acute neutrophilia in human SCI patients is positively correlated with New Injury Severity Scores but inversely with the neurological outcome (AIS grade). Multivariate analysis demonstrated that acute SCI-induced neutrophilia is an independent predictor of AIS grade conversion failure, with an odds ratio (OR) of 4.16 and ROC area under curve (AUC) of 0.82 (P < 0.0001). SCI-induced lymphopenia was separately identified as an independent predictor of better recovery (OR = 24.15; ROC AUC = 0.85, P < 0.0001). Acute neutrophilia and increased neutrophil-lymphocyte ratios were otherwise significantly associated with respiratory infection presentation in both patient cohorts.

Conclusions: Our findings demonstrate the prognostic value of modelling early circulating neutrophil and lymphocyte counts with patient characteristics for predicting the longer term recovery after SCI.

Keywords: acute phase response; immune depression syndrome; lymphocytes; monocytes; neurotrauma; neutrophils; pneumonia.

FIGURE 1

Diagram for the Brisbane exploration cohort. Available SCI patients were screened for eligibility based on the availability of WBC data. Patient datasets used for specific statistical analyses depended on the subgroup being tested, sample size limitations and/or availability of data; the resulting exclusion of patients for each subgroup analysis is indicated in the dashed boxes. The colour coding for each of the analysis boxes refers to the specific figure it pertains to, that is red = neutrophil‐related (Figure 3); blue = lymphocyte‐related (Figure 4); purple = either neutrophil or lymphocyte‐related (Figure 3 or 4); green = respiratory infection‐related (Figure 5). Abbreviations: AIS, American Spinal Injury Association Impairment Scale; C1‐T12, cervical level 1 to thoracic level 12; C‐SCI, cervical spinal cord injury; NISS, New Injury Severity Score; SCI, spinal cord injury; WBC, white blood cell

FIGURE 2

The systemic response to traumatic SCI is hallmarked by acute neutrophilia and lymphopenia. For the exploration (i.e. Brisbane) cohort (A‐C), changes in circulating WBC numbers were compared between the specified time points (admission (0), 1, 3 and 7 days post‐injury (dpi)) using one‐way analysis of variance (ANOVA) with Bonferroni's post hoc test. For the validation (i.e. Berlin) cohort (D‐F), temporal changes in circulating WBC numbers between 0 and 8 dpi were compared using a linear mixed effects model, with the indicated significances referring to the outcome of these models. (A and D) Circulating neutrophil numbers were significantly elevated above the clinical reference range (indicated in light grey) at admission and/or 1 dpi for both the Brisbane and Berlin patient cohorts and then rapidly decreased over time. (B and E) Circulating monocyte numbers were significantly increased at 1 and 7 dpi in the Brisbane cohort (relative to admission), but not between 1, 3 and 7 dpi; no difference was seen here within the Berlin cohort either. (C and F) Circulating lymphocyte numbers in the overall Brisbane cohort were significantly reduced at 1 and 3 dpi compared to admission; this subclinical lymphopenia was mostly resolved by 7 dpi. A clear lymphopenia was also evident for the Berlin patient group at 1 and 3 dpi, along with an upturn in lymphocyte numbers by 8 dpi. Box and whisker plots are shown with the median and interquartile range (Tukey whiskers) and outliers are indicated by individual data‐points. Normal clinical reference ranges for each of the main WBC types were as follows: neutrophils = 2‐8 × 10⁹ cells/L, monocytes = 0.1‐1 × 10⁹ cells/L and lymphocytes = 1‐4 × 10⁹ cells/L in the Brisbane cohort; and 1.5‐7.7 × 10⁹ neutrophils/L, 0.1‐0.9 × 10⁹ monocytes/L and 1.1‐4.5 × 10⁹ lymphocytes/L in the Berlin cohort. *P < 0.05, **P < 0.01, ****P < 0.0001 (compared to admission/0 dpi); ^#### P < 0.0001 (compared to 1 dpi); ⁺ P < 0.05 (compared to 3 dpi). RC, regression coefficient (95% confidence interval).

FIGURE 3

Circulating neutrophil numbers are associated with injury severity and long‐term patient outcomes. (A) Circulating neutrophil numbers on admission positively correlate with the trauma severity as indicated by NISS data. This association was also present in patients with an admission AIS grade of A (complete SCI) versus those with an AIS grade of B, C or D (incomplete SCI); r ² value = Pearson's correlation coefficient; linear fit indicated by solid black line for all patients, dashed lines for subgrouping based on admission AIS grade. (B) Blood neutrophil numbers during the first week of SCI segregated by lesion level (two‐way analysis of variance (ANOVA) with Bonferroni's post hoc test) (C) NISS data for patients with cervical versus thoracic injuries (one‐way ANOVA with Bonferroni's post hoc test). (D) SCI patients with associated multi‐trauma (MT) had significantly higher circulating neutrophil numbers at admission compared to those with a spine‐isolated spinal cord injury (SCI; two‐way ANOVA with Bonferroni's post hoc test). (E) Circulating neutrophil numbers are positively correlated with circulating monocyte numbers at admission (purple) and 1 day post‐injury (blue). The reference range for each WBC type is indicated in light grey and on the relevant axis of the graph (r ² value = Pearson's correlation coefficient; linear fit is indicated by the solid lines). (F) AIS grade conversion rates in patients with and without acute neutrophilia (Fisher's exact test). (G) ROC curve analysis of the adjusted neutrophilia model used to predict AIS conversion showed high specificity and sensitivity as indicated by a significant AUC (red line = adjusted model, dashed grey line = reference line where AUC = 0.5). Panels B‐D are represented via box and whisker plots which show the median and interquartile range (Tukey whiskers); outliers are indicated by individual data‐points. Clinical reference ranges for WBC types are shown in light grey. Abbreviations: AIS, American Spinal Injury Association Impairment Scale; AUC, area under the curve; C, cervical; NISS, New Injury Severity Score; ROC, receiver operating characteristic; SCI, spinal cord injury; T, thoracic; WBC, white blood cell. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001

FIGURE 4

Lymphopenia is associated with better functional recovery in SCI. (A) Circulating lymphocyte counts segregated by lesion level during the first week post‐SCI (two‐way analysis of variance (ANOVA) with Bonferroni's post hoc test); the clinical reference range is shown in light grey. (B) Lymphocyte numbers at 3 dpi in SCI patients with multi‐trauma or an isolated spine injury, further segregated based on AIS grade conversion between admission and discharge (two‐way ANOVA with Bonferroni's post hoc test). (C) Sub‐group analysis showing circulating lymphocyte numbers for patients with and without clinical lymphopenia; the dashed grey line indicates the lower reference range limit used to classify lymphopenia on day 1 and/or 3 post‐SCI (two‐way ANOVA with Bonferroni's post hoc test). (D) AIS grade conversion rates in SCI patients with and without acute clinical lymphopenia portion (isolated spine injury only; Fisher's exact test). (E) ROC curve analysis of the adjusted lymphopenia model used to predict AIS conversion showing specificity and sensitivity as indicated by a significant AUC (purple line = adjusted model, dashed grey line = reference line where AUC = 0.5). (F) Incidence of lymphopenia in patients with a complete SCI (AIS grade A) versus those with incomplete SCI (AIS grade B, C or D; Fisher's exact test). (G) Incidence of lymphopenia relative to lesion level (Fisher's exact test). Box and whisker plots in panels A‐C show the median and interquartile range (Tukey whiskers); outliers are indicated by individual data‐points. Abbreviations: AIS, American Spinal Injury Association Impairment Scale; AUC, area under the curve; C, cervical; dpi, days post injury; MT, multi‐trauma; NISS, New Injury Severity Score; ROC, receiver operating characteristic; SCI, spinal cord injury; T, thoracic. *P < 0.05, **P < 0.01, and ****P < 0.0001

FIGURE 5

Respiratory infection presentation and clinical outcomes in SCI patients. Patients presenting with pneumonia or other respiratory infections acquired after injury were compared with their non‐infected counterparts for white blood cell (WBC) changes, injury characteristics and neurological recovery in both the Brisbane (A‐H) and Berlin (I‐N) patient cohorts. (A and I) Incidence of respiratory infections in patients with complete SCI (AIS grade A) versus incomplete SCI (AIS grade B, C or D; Fisher's and χ ² test for A and I, respectively). Patients with cervical injury (B and J) and those that were intubated/mechanically ventilated (C and K) were more likely to acquire an airway infection (Fisher's and χ ² test for B‐C and J‐K, respectively). (D and L) Incidence of respiratory infections in SCI patients with and without acute lymphopenia (1 and/or 3 dpi; Fisher's and χ ² test for D and L, respectively). (E) Incidence of respiratory infections in patients with and without an AIS grade conversion between admission and discharge. (F and M) Incidence of respiratory infections in cervically injured patients with and without acute neutrophilia (Fisher's and χ ² test for F and M, respectively). (G and N) Neutrophil‐lymphocyte ratios for SCI patients with and without acquired airway infections (Mann‐Whitney test for G; for linear mixed model estimates related to N see Table 6B). (H) Relationship between neutrophil‐lymphocyte and neutrophil‐monocyte ratios (3 dpi) to the length of stay in the intensive care unit (ICU; r ² value = Pearson's correlation coefficient; linear fit indicated by solid coloured lines). Box and whisker plots show the median and interquartile range (Tukey whiskers) and outliers are indicated by individual data‐points. Abbreviations: AIS, American Spinal Injury Association Impairment Scale; dpi, days post‐injury; ICU, intensive care unit; NISS, New Injury Severity Score; WBC, white blood cell. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001