The spinal cord injury-induced immune deficiency syndrome: results of the SCIentinel study

Abstract

Infections are prevalent after spinal cord injury (SCI), constitute the main cause of death and are a rehabilitation confounder associated with impaired recovery. We hypothesize that SCI causes an acquired lesion-dependent (neurogenic) immune suppression as an underlying mechanism to facilitate infections.

The international prospective multicentre cohort study (SCIentinel; protocol registration DRKS00000122; n = 111 patients) was designed to distinguish neurogenic from general trauma-related effects on the immune system. Therefore, SCI patient groups differing by neurological level, i.e. high SCI [thoracic (Th)4 or higher]; low SCI (Th5 or lower) and severity (complete SCI; incomplete SCI), were compared with a reference group of vertebral fracture (VF) patients without SCI. The primary outcome was quantitative monocytic Human Leukocyte Antigen-DR expression (mHLA-DR, synonym MHC II), a validated marker for immune suppression in critically ill patients associated with infection susceptibility. mHLA-DR was assessed from Day 1 to 10 weeks after injury by applying standardized flow cytometry procedures. Secondary outcomes were leucocyte subpopulation counts, serum immunoglobulin levels and clinically defined infections. Linear mixed models with multiple imputation were applied to evaluate group differences of logarithmic-transformed parameters.

Mean quantitative mHLA-DR [ln (antibodies/cell)] levels at the primary end point 84 h after injury indicated an immune suppressive state below the normative values of 9.62 in all groups, which further differed in its dimension by neurological level: high SCI [8.95 (98.3% confidence interval, CI: 8.63; 9.26), n = 41], low SCI [9.05 (98.3% CI: 8.73; 9.36), n = 29], and VF without SCI [9.25 (98.3% CI: 8.97; 9.53), n = 41, P = 0.003]. Post hoc analysis accounting for SCI severity revealed the strongest mHLA-DR decrease [8.79 (95% CI: 8.50; 9.08)] in the complete, high SCI group, further demonstrating delayed mHLA-DR recovery [9.08 (95% CI: 8.82; 9.38)] and showing a difference from the VF controls of −0.43 (95% CI: −0.66; −0.20) at 14 days. Complete, high SCI patients also revealed constantly lower serum immunoglobulin G [−0.27 (95% CI: −0.45; −0.10)] and immunoglobulin A [−0.25 (95% CI: −0.49; −0.01)] levels [ln (g/l × 1000)] up to 10 weeks after injury. Low mHLA-DR levels in the range of borderline immunoparalysis (below 9.21) were positively associated with the occurrence and earlier onset of infections, which is consistent with results from studies on stroke or major surgery.

Spinal cord injured patients can acquire a secondary, neurogenic immune deficiency syndrome characterized by reduced mHLA-DR expression and relative hypogammaglobulinaemia (combined cellular and humoral immune deficiency). mHLA-DR expression provides a basis to stratify infection-risk in patients with SCI.

Infections are a leading cause of death after spinal cord injury (SCI). Kopp et al. report that acute SCI can cause a secondary neurogenic immune deficiency syndrome (SCI-IDS) that is distinct from a trauma-related general stress response, and which is characterized by a lesion-dependent reduction in monocytic HLA-DR expression.

Introduction

Infections acquired in the acute phase after spinal cord injury (SCI) are life threatening¹ and associated with poor neurological^2,3 and functional^3,4 outcome. A common understanding of the immunological mechanisms contributing to increased infection susceptibility in patients following SCI is lacking.

Following and complementing clinical pilot studies,^5-7 mechanistic experiments have delineated neuroanatomical level- and severity-dependent autonomic dysregulation after SCI,⁸ which has been causally linked to a profound maladaptive systemic immune response⁹ and increased infection susceptibility.^10-12 The spinal cord injury-induced immune deficiency syndrome (SCI-IDS) is viewed as a neurogenic, pathophysiological counterpart to the compensatory anti-inflammatory response syndrome (CARS),¹³ which develops for example, after polytrauma (also referred to a as ‘post-aggression syndrome’) and is a frequent comorbidity associated with SCI. To what degree an emerging neurogenic SCI-IDS contributes to the overall immune suppressive state in SCI patients is unknown. Moreover, while sporadic clinical pilot studies point towards alterations of the cellular immune response, no clinical evidence supporting putative changes of non-cellular, i.e. humoral immunity (immunoglobulins), is available. To determine both, it is necessary to disentangle a neurogenic injury level and severity-dependent SCI-IDS from an unspecific, trauma-related post-aggression syndrome mirroring a generalized stress response.¹⁴

In a multicentre setting, the prospective and longitudinal SCIentinel study¹⁵ assessed neurological level- and injury severity-defined groups of SCI patients in comparison with a vertebral fracture (VF) control group without myelopathy to reveal the neurogenic features of SCI-IDS at the cellular, non-cellular/humoral, and clinical levels. The primary outcome parameter, the expression of monocytic Human Leukocyte Antigen-DR (mHLA-DR), was verified to be diagnostic for infection risk associated with immune suppression in critically ill patients.^16-21 The highly standardized determination of mHLA-DR as surrogate for global immune dysfunction enables the assessment of systemic immune suppression in patients after SCI in an objective and quantitative manner.^16,19

Materials and methods

Study oversight

The prospective cohort study was conducted in seven clinical departments located at four hospitals: BG Hospital Unfallkrankenhaus Berlin, Germany; Charité-Universitätsmedizin Berlin, Germany; University of Zurich, Switzerland; and University Health Network Toronto, Canada. The study protocol has been published¹⁵ and pre-registered in the German Clinical Trials Registry (DRKS-ID: DRKS00000122). The Ethical Committee of Charite-Universitätsmedizin Berlin (EA1/001/09), the University Health Network Research Ethics Board, Toronto (REB10-0384-AE), and the Cantonal Ethics Commission, Zurich (KEK-ZH-No. 2011-0059), have approved the study, and it was conducted in compliance with the Declaration of Helsinki.

Patients

To distinguish SCI-related, neurogenic effects from stress-associated effects after trauma, both acute traumatic SCI patients and VF patients without SCI were included. Lesion-dependent (neurogenic) aspects such as neurological lesion level and completeness were evaluated using the International Standards for Neurological Classification of Spinal Cord Injury (ISNCSCI)²² examination. To account for the effects of surgery, planned or performed spinal stabilization and/or decompression was a key inclusion criterion. To reduce immunological and traumatological confounders, patients with non-traumatic SCI, life threating polytrauma, serious traumatic brain injury, pre-existing neoplasia, autoimmune diseases or chronic infections were excluded from the study. To avoid confounding pharmacological with neuroanatomical effects, patients who received high-dose methylprednisolone according to the National Acute Spinal Cord Study (NASCIS) protocol²³ were also excluded.¹⁵ For the complete eligibility criteria, see Table 1. All patients were informed about the study and gave their written informed consent to participate prior to inclusion.

Table 1.

Eligibility criteria

Inclusion criteria
Patients with acute isolated SCI (AIS A–D) planned for surgical stabilization and decompression, lesion may include more than one segment
Patients with acute isolated spinal fracture planned for surgical stabilization, lesion may include more than one segment
Legal age of the patient
Documented written informed consent of the patient
Exclusion criteria
Non-traumatic SCI
Two or more spinal cord or vertebral lesions definable one from another
Severe polytraumaa
Concomitant traumatic brain injuryb
Neoplasia and/or antineoplastic therapy
Rheumatic disease, collagenosis, vasculitis or other autoimmune disease
Pre-existing chronic infectious diseasec
Pre-existing systemic steroid treatmentd
Severe alcohol or drug addiction
Pregnancy, lactation

SCI = spinal cord injury.

Patients with severe injuries of life-sustaining organ systems, which per se and in the acute phase can be life-threatening (e.g. severe pelvic trauma, severe body cavity injuries).

(i) Patients with persisting neurological deficit in consequence of the traumatic brain injury; (ii) patients with severe traumatic brain injury (Glasgow Coma Scale < 8); and (iii) patients with intracranial pressure monitoring sensors.

Before the injury.

Except high-dose methylprednisolone treatment according to the NASCIS protocol, SCI patients treated with methylprednisolone can be enrolled but will be excluded from the primary analysis, only a casuistic analysis is planned in these patients.

Study end points

The primary end point was the expression of mHLA-DR 3–4 days after injury,¹⁵ which demarcates the onset of SCI-associated infections.²⁴ mHLA-DR is a well-established, independently validated marker of immune suppression following major trauma, surgery and sepsis.^16,18,19,21 Using a standardized flow cytometric assay, the anti-HLA-DR antibody binding numbers per monocyte (ab/cell) were determined (see the ‘Laboratory procedures’ section). The critical mHLA-DR level for defining ‘immunoparalysis’ in sepsis and major surgery patients has been reported to be <5000 ab/cell.^16,19 Values in the range of ‘borderline immunoparalysis’ (5000–10 000 ab/cell) or ‘immune suppression’ (10 000–15 000 ab/cell) were associated with serious infections or clinical worsening after stroke and trauma.^17,25,26 The normal range of mHLA-DR is >15 000 ab/cell.¹⁶

Secondary laboratory end points of the cellular immunity were mHLA-DR expression over the entire time course, white blood cell and lymphocyte subpopulation counts. Serum levels of immunoglobulins were assessed to evaluate the function of the humoral immune system. Venous blood samples for the laboratory end points were collected within five time frames after injury: <31 h, 31–96 h, 5–9 days, 11–28 days, and 8–12 weeks.¹⁵ Clinical end points were the occurrence of pulmonary or urinary infections according to pre-defined criteria or of other infections based on their clinical diagnosis (Supplementary Table 1). Clinical data were collected at the bedside and from the medical records using paper case report forms.

Laboratory procedures

Similar laboratory procedures were applied to all samples from each group of patients (SCI and VF groups). Immunophenotyping was performed in whole blood samples collected in Cyto-Chex® tubes (Streck), allowing for preservation of white blood cells for up to 14 days at room temperature. mHLA-DR expression was determined within 36 h after blood collection by flow cytometry using a highly standardized quantitative assay (BD Quantibrite™, HLA-DR/Monocyte reagent, BD Biosciences). Anti-HLA-DR-antibodies bound per monocyte (ab/cell) were measured according to standardized operating procedures at the Department of Immunology, Labor Berlin—Charité Vivantes GmbH, Berlin, Germany), established for clinical diagnostics of an immune suppression state. The method was approved in an inter-laboratory multicentre study¹⁶ and is described in more detail in the study protocol publication.¹⁵

Standard white blood cell differential was performed in certified clinical diagnostic laboratories associated with the respective trial centres. For all centres in Berlin, the XE-5000 Case Manager haematology analyser (Sysmex) was used. For the centres in Zurich, the XE-5000 Case Manager haematology analyser (Sysmex Suisse AG), and for the centre in Toronto, the Sysmex XN-2000 haematology analyser (Sysmex Canada) were used.

Lymphocyte subpopulations were quantified centrally in the Department of Immunology at Labor Berlin within 36 h after blood collection by flow cytometry. Briefly, the following mouse anti-human fluorescently-labelled monoclonal antibodies (all from Beckman Coulter) were used for quantification of lymphocytes subsets: cluster of differentiation (CD)3 Allophycocyanine-Alexa Fluor 750 (APC-A750, clone UCHT1; Cat. No. A94680), CD4 energy coupled dye (ECD, clone SCFI12T4D11; 6604727), CD8 APC (clone B9.11; IM2469), CD14 Fluorescein isothiocyanate (FITC, clone RMO52; IM0645U), CD16 Phycoerythrine (PE, clone 3G8; A07766), CD19 PE-Cy5.5 (clone J3-119; B49211), CD45RA Pacific-Blue (PB, clone J33; A74763), and CD56 PE (clone N901; A07788). Stained samples were acquired on a 10-colour Navios flow cytometer and analysed using Navios Software (Beckman Coulter).

Serum levels of the immunoglobulin (Ig) classes IgG, IgA, and IgM were quantified in samples collected using BD Vacutainer® tubes (Becton Dickinson). The tubes were centrifuged within 4 h after blood collection at 2000g for 10 min. The supernatants were transferred into cryotubes and stored at −80°C immediately after centrifugation. A batch analysis after study completion was performed centrally at the Labor Berlin using a Tina-quant® turbidimetric in vitro test (Roche).

Sample size

The study was designed to examine neurological level-dependent effects related to altered supraspinal control of pre-ganglionic sympathetic neurons originating from the thoracic spinal segments Th5–Th9.^10,11,15 Therefore, the sample size was estimated to detect differences in mHLA-DR expression between three groups: (i) SCI patients with a neurological level of injury [cervical (C)1–thoracic (Th)4]; (ii) SCI patients with neurological level of injury [thoracic (Th)5–sacral (S)4/5]; and (iii) VF patients without SCI as a neurologically unexposed group 3–4 days after injury (primary end point). Based on the pilot study, the number of HLA-DR-positive cells observed 3–4 days post-trauma⁷ was used as a surrogate parameter to estimate the anticipated effect size of 0.18. The significance level α was set to 0.05, and the type two error β to 0.2. The sample size calculated for testing the hypothesis using a one-way-ANOVA was n = 56. Considering an expected rate of missing values of up to 50%, the sample size was increased to n = 112. In the event that non-parametric tests should be applied, it was further increased to n = 118.

Statistical analysis

To evaluate the immune parameter differences within and between the SCI groups and the VF group, as well as their changes over time, we used linear mixed regression models with random intercept (statistical software R²⁷). Where appropriate, log-transformed (ln) values (natural logarithm) of the immune parameters were used. The scale dimension of immune cell counts, Ig concentrations, and age were adapted (i.e. cells/nl into cells/nl × 1000, g/l into g/l × 1000, and years into decades) to avoid heterogeneous scales in the models. The models were adjusted for age, sex, treatment centre, time point of measure, and the time point of first spine surgery. Besides the ln time since injury, a second time term (ln time centred and squared) as well as the interactions between patient groups and each of the time variables were included. Under the missing at random (MAR) assumption, missing values (Supplementary Table 2) were estimated using multiple imputation,²⁸ resulting in 30 imputed datasets (Supplementary material, ‘Methods’ section). For the primary end point, we used the maximum test statistics²⁹ of pairwise group differences in ln mHLA-DR 84 h (3–4 days) after injury between the high SCI (C1–Th4), low SCI (Th5–S4/5), and VF groups using a linear mixed model based on the multiple imputed data. This analysis strategy was preferred over the originally planned one-way ANOVA to better account for the longitudinal study design. The significance level was set to 0.0167 after Bonferroni-correction for multiple comparisons (three tests) and confidence intervals (CIs) of the primary end point were adjusted accordingly to 98.3%. After analysis of mHLA-DR with multiple imputation, we recalculated the mHLA-DR model using only ‘observed data’ for the primary end point as well as for a post hoc analysis. The post hoc analysis strategy was chosen also to account for effects of the SCI severity, as it likewise defines the degree of autonomic denervation below the neurological level.³⁰ Therefore, a more differentiating grouping of patients was applied to the secondary end points. In addition to the neurological level, SCI patients were classified based on the American Spinal Injury Association Impairment Scale (AIS) categories A (complete SCI) or B–D (incomplete SCI), yielding four SCI groups: (i) high complete SCI; (ii) high incomplete SCI; (iii) low complete SCI; and (iv) low incomplete SCI (Fig. 1). The estimated marginal means of log-transformed values of the groups and the estimated mean differences of the SCI groups from the VF group were calculated at time points 15 h, 64 h, 7 days, 14 days and 10 weeks after injury, which correspond to the time windows for blood collection stipulated in the study protocol.¹⁵ Time-dependent effects of surgery³¹ were taken into account. The estimated means were calculated based on the proportion of completed first spine surgery at each time point that was 0.5 at 15 h, 0.82 at 64 h, 0.96 at 7 days, and 1.0 at 14 days and 10  weeks. To explore possible age-related alterations of mHLA-DR expression over time, an additional linear mixed model on the multiple imputed datasets was calculated. We classified the patients into complete SCI, incomplete SCI, and VF and extended the interactions to Group × Age × Time with both time terms and used the same previously mentioned covariates. The estimated marginal means within each group were calculated for the 25th and 75th age percentiles (36 versus 63 years) for each visit time point. The occurrence of first infections and their association with mHLA-DR was additionally examined using the ‘observed data’ from SCI patients (Supplementary material, ‘Methods’ section). All post hoc analyses were performed in an explorative manner, reporting effect estimates with 95% CIs. Observed raw data of the immune parameters were reported in a descriptive manner using violin plots.

Figure 1.

Patient selection and analysis chart and the distribution of baseline characteristics across the SCI groups and the VF group. Three patients were withdrawn from the study due to a delayed awareness of the exclusion criteria (renal-cell carcinoma, ankylosing spondylitis, HIV-infection). Six cases treated with methylprednisolone according to the NASCIS-scheme²³ were excluded from the analysis. The analysis of the primary end point 3–4 days (84 h) after injury consisted of a comparison of (i) high SCI (neurological level Th4 or higher); (ii) low SCI (neurological level Th5 or lower); and (iii) neurologically unaffected VF (vertebral fracture without SCI) groups. Subsequently, a post hoc analysis strategy was additionally accounting for effects of the SCI severity (ASIA impairment scale – AIS), because it also defines the degree of autonomic denervation below the neurological level.³⁰ For the detailed exploration, neurological (neurogenic) lesion characteristics such as SCI level and severity (complete versus incomplete lesions) were taken into account. Consequently, the SCI population was segregated into four groups comprising (i) high complete SCI (neurological level Th4 or higher, AIS A); (ii) high incomplete SCI (neurological level Th4 or higher, AIS B-D); (iii) low complete SCI (neurological level Th5 or lower, AIS A); and (iv) low incomplete SCI (neurological level Th5 or lower, AIS B-D). SCI groups were compared to the (v) VF group (vertebral fracture without SCI). Missing values: n = 1 for age in the low complete SCI group. AIS = American Spinal Injury Association (ASIA) Impairment Scale; C = cervical; IQR = interquartile range; L = lumbar; NASCIS = National Acute Spinal Cord Study; S = sacral; SCI = spinal cord injury; SCI-IDS = spinal cord injury-induced immune deficiency syndrome; SD = standard deviation; Th = thoracic; VF = vertebral fracture.

Data availability

Individual-level data collected in the SCIentinel study cannot be fully shared for data protection reasons. However, extracts of deidentified patient data from the analysis dataset in connection with a data-dictionary can be made available upon written request and signature of a data transfer agreement by the corresponding authors with publication of the article.

Results

Baseline characteristics

From 21 August 2011 to 4 October 2014, n = 120 patients were recruited. After the exclusion of n = 3 patients, who were withdrawn due to violation of eligibility criteria, and n = 6 patients treated with high-dose methylprednisolone,²³ the study population comprised n = 111 patients (Fig. 1), n = 70 of which completed follow-up at 10 weeks. The median follow-up time was 65 days [interquartile range (IQR) 16–71]. Divided into the neuroanatomically defined SCI groups, patients with high SCI and VF control patients were slightly older than low SCI patients. The VF group included more female patients than all SCI groups. The high SCI group comprised cervical neurological levels only and the distribution of SCI severity (AIS grades) within the SCI groups was balanced with a few more AIS A and less AIS C patients in the low SCI group. The time from injury to first spine surgery was shorter in the complete SCI groups (AIS A) compared to the incomplete SCI (AIS B–D) or VF groups (Fig. 1).

Monocytic HLA-DR expression

At the primary end point 84 h after injury, mHLA-DR (synonym MHC class II) expression profiles (anti-HLA-DR antibody binding numbers per cell, estimated means) were below the reference value of 15.000 ab/cell¹⁶ [log-transformed (ln) value 9.62 ab/cell]. All groups dropped into the range indicating immune suppression. However, ln mHLA-DR values were different between the high SCI [8.95 (98.3% CI 8.63; 9.26)], low SCI [9.05 (8.73; 9.36)], and VF [9.25 (8.97; 9.53), P = 0.003] groups with mean values in the range between ‘borderline immunoparalysis’ [10.000 ab/cell (ln value 9.21 ab/cell)] and ‘immunoparalysis’ [5.000 ab/cell (ln value 8.52 ab/cell)] in both SCI groups. In detail, high SCI patients demonstrated lower mHLA-DR values compared to the VF control group [−0.30 (−0.54; −0.06), P = 0.003]. The differences were less pronounced when comparing the low SCI and VF groups [−0.20 (−0.47; 0.06), P = 0.07)] and the low with high SCI groups [−0.10 (−0.37; 0.17), P = 0.37)].

The subsequent post hoc analysis comparing the trajectories of mHLA-DR expression between four injury level- and severity-defined SCI groups and the VF group over time (Fig. 2A) identified the completely injured SCI patient groups to reveal the lowest mHLA-DR mean values, with the 95% CI fully below the margin of ‘borderline immunoparalysis’ at 64 h after injury (Fig. 2A). After incomplete SCI, particularly in the low incomplete SCI group, mHLA-DR profiles were similar to the VF group throughout the study period, whereas lowest mHLA-DR profiles were observed in complete SCI extending over 14 days (Fig. 2B). At 10 weeks, the low SCI groups and the VF group recovered to mHLA-DR levels (mean and 95% CI) above the reference line for ‘immune suppression’, whereas recovery in their counterparts, the high SCI groups, was only partial (Fig. 2A).

Figure 2.

Spinal cord lesion-dependent (neurogenic) effects on mHLA-DR (linear mixed model). Standardized quantitative estimation of mHLA-DR expression [number of anti-HLA-DR antibodies bound per monocyte (ab/cell)], a validated marker for temporary systemic immune suppression in critically ill patients.^16,18,19,21 (A) Estimated marginal means with 95% CI of mHLA-DR values [ln(ab/cell)] over time calculated for the five groups at the per-protocol time points of blood collection using linear mixed models (random intercept) adjusted for age, sex, treatment centre, time point of measure and the time point of first spine surgery after multiple imputation for missing values (30 complete datasets). The total number of actual observed mHLA-DR measurements was 402. Goodness of fit (mean of m = 30 models) conditional R2 (fixed and random effects): 0.630; marginal R2 (fixed effects): 0.354. The blue horizontal lines indicate the thresholds towards more serious stages of immune suppression; dotted line ln (<15 000 ab/cell) = ‘immune suppression'; dotted/broken line ln (<10 000 ab/cell) = ‘borderline immunoparalysis'; broken line ln (<5000 ab/cell) = ‘immunoparalysis’.^15,16 (B) Estimated mean differences with 95% CI of ln mHLA-DR of the four SCI groups in relation to the VF group indicated as green broken line based of same regression models as for A. Reduced mHLA-DR is a characteristic hallmark of temporary immunosuppression at acute and subacute stages of the spinal cord injury-induced immune deficiency syndrome (SCI-IDS). Neuroanatomical definition of the groups: high complete SCI (neurological level Th4 or higher, AIS A); high incomplete SCI (neurological level Th4 or higher, AIS B-D); low complete SCI (neurological level Th5 or lower, AIS A); low incomplete SCI (neurological level Th5 or lower, AIS B-D); VF group (vertebral fracture without SCI). ab = antibodies; AIS = American Spinal Injury Association (ASIA) Impairment Scale; CI = confidence interval; mHLA-DR = monocytic Human Leukocyte Antigen-DR expression; ln = log-transformed; SCI = spinal cord injury; VF = vertebral fracture.

The demographic baseline characteristics age and sex were not substantially associated with mHLA-DR levels. An additional regression model calculated to explore differences between groups of young and old patients with SCI or VF revealed no differential changes in mHLA-DR levels over time between younger and older SCI patients as exemplified by the 25th (36 years) and 75th (63 years) percentile at the injury time. In the group with VF, older patients exhibited lower mHLA-DR values compared with younger patients (Supplementary Fig. 1).

When interrogating 'observed data' only (sensitivity analysis), the results were very similar to those after multiple imputation in the analysis of the primary end point at 84 h as well as in the post hoc analysis (Supplementary Table 3 and Supplementary Fig. 2).

‘Observed data’ were also analysed for differences within the category of incomplete SCI by segregating the data by AIS grades. The distribution of the data revealed no evidence of major inhomogeneity between the groups AIS B, C, and D (Supplementary Fig. 3). For detailed mHLA-DR raw data, see Supplementary Fig. 4.

White blood cell counts

The total neutrophil counts (estimated means) were at the upper limit of the clinical reference range up to 64 h after injury in all groups and decreased rapidly over time to values within the reference range (Fig. 3A). The monocyte counts were within the upper reference range at any time with highest estimates during the first week (Fig. 3B). Completely injured SCI groups had the highest neutrophil and monocyte counts relative to the VF group at 64 h and 7 days (Supplementary Table 4). The lymphocyte counts were at the lower bound of the reference range at 15 and 64 h and were increasing to normal values until the end of the observation window at 10 weeks (Fig. 3C), revealing no clear differences between the SCI groups and the VF group (Supplementary Table 4). For white blood cell raw data, see Supplementary Fig. 5.

Figure 3.

Fluctuation of leucocyte subpopulations. Estimated marginal means with 95% CI of (A) neutrophil, (B) monocyte, and (C) lymphocyte count [ln(cells/nl × 1000)] calculated for the five groups at the per-protocol time points of blood collection using linear mixed models (random intercept) adjusted for age, sex, treatment centre, time point of measure, and the time point of first spine surgery after multiple imputation for missing values (30 complete datasets). The total number of actual observed measurements for neutrophils, monocytes, and lymphocytes was 391, 389, and 388, respectively. For differences in estimated marginal means and measures for goodness of fit see Supplementary Table 4. The broken lines indicate the upper bound (red) and the lower bound (blue) of the reference areas. Neuroanatomical definition of the groups: high complete SCI (neurological level Th4 or higher, AIS A); high incomplete SCI (neurological level Th4 or higher, AIS B-D); low complete SCI (neurological level Th5 or lower, AIS A); low incomplete SCI (neurological level Th5 or lower, AIS B-D); VF group (vertebral fracture without SCI). AIS = American Spinal Injury Association (ASIA) Impairment Scale; CI = confidence interval; ln = log-transformed; SCI = spinal cord injury; VF = vertebral fracture.

Lymphocyte subpopulations

CD3+ T lymphocyte numbers (estimated means) were at the lower margin of the reference range during the first week in all groups and recovered to values within the normal range at 10 weeks after injury without revealing differences between the SCI groups and the VF group (Fig. 4A and Supplementary Table 5). The CD19+ B-cell count was largely within the normal range throughout the study (Fig. 4B and Supplementary Table 5). The estimates of the CD16+ NK-cell counts were below the reference range in all groups until Day 7 and began to partially recover at Day 14. In contrast to all other groups, the high complete SCI group remained at values below the reference range until 10 weeks (Fig. 4C and Supplementary Table 5). For the raw data of the lymphocyte subpopulations, see Supplementary Fig. 6.

Figure 4.

Fluctuation of lymphocyte subpopulations. Estimated marginal means of (A) CD3+ T cell, (B) CD19+ B cell, and (C) CD16+ NK cell counts [ln(cell/nl × 1000)] calculated for the five groups at the per-protocol time points of blood collection using linear mixed models (random intercept) adjusted for age, sex, treatment centre, time point of measure, and the time point of first spine surgery after multiple imputation for missing values (30 complete datasets). The total number of actual observed measurements for CD3+ T cells, CD19+ B cells, and CD16+ NK cells was 397 each. For differences in estimated marginal means and measures for goodness of fit see Supplementary Table 5. The broken lines indicate the upper bound (red) and the lower bound (blue) of the reference areas. Neuroanatomical definition of the groups: high complete SCI (neurological level Th4 or higher, AIS A); high incomplete SCI (neurological level Th4 or higher, AIS B-D); low complete SCI (neurological level Th5 or lower, AIS A); low incomplete SCI (neurological level Th5 or lower, AIS B-D); VF group (vertebral fracture without SCI). AIS = American Spinal Injury Association (ASIA) Impairment Scale; CI = confidence interval; ln = log-transformed; SCI = spinal cord injury; VF = vertebral fracture.

Immunoglobulins

The serum IgG levels (estimated means) were below the reference range in the complete SCI groups during the first week after injury, while the VF group spent up to 14 days in the lower reference range and normalized/centred after 10 weeks (Fig. 5A). Relative to the VF group, the recovery of serum IgG was particularly suspended in the high complete SCI group, with lower profiles being present until the end of observation and an estimated difference in IgG [ln (g/l × 1000)] of −0.27 (95% CI −0.45; −0.10) at 10 weeks (Supplementary Table 6). In the high complete SCI group, IgA was also persistently lower, with a difference of −0.25 (−0.49; −0.01) at 10 weeks (Fig. 5B and Supplementary Table 6). IgM was lower very early after injury in the high complete SCI group and consistently recovered to normal values (Fig. 5C and Supplementary Table 6). For the immunoglobulin raw data, see Supplementary Fig. 7.

Figure 5.

Spinal cord lesion-dependent (neurogenic) effects on serum immunoglobulin levels. Estimated marginal means of (A) IgG, (B) IgA, and (C) IgM levels [ln(g/l × 1000)] calculated for the five groups at the per-protocol time points of blood collection using linear mixed models (random intercept) adjusted for age, sex, treatment center, time point of measure, and the time point of first spine surgery after multiple imputation for missing values (30 complete datasets). The total number of actual observed measurements for IgG, IgA, and IgM was 419 each. For differences in estimated marginal means and measures for goodness of fit see Supplementary Table 6. The broken red lines indicate the upper bound and the broken blue lines the lower bound of the reference areas. The high complete SCI group revealed constantly lower serum IgG and IgA levels compared to the VF group up to 10 weeks (Supplementary Table 6). Relative and absolute hypogammaglobulinaemia are symptoms contributing to the spinal cord injury-induced immune deficiency syndrome (SCI-IDS) in a lesion severity and level dependent manner. Neuroanatomical definition of the groups: high complete SCI (neurological level Th4 or higher, AIS A); high incomplete SCI (neurological level Th4 or higher, AIS B-D); low complete SCI (neurological level Th5 or lower, AIS A); low incomplete SCI (neurological level Th5 or lower, AIS B-D); VF group (vertebral fracture without SCI). AIS = American Spinal Injury Association (ASIA) Impairment Scale; CI = confidence interval; Ig = immunoglobulin; ln = log-transformed; SCI = spinal cord injury; VF = vertebral fracture.

mHLA-DR and acquired infections

During follow-up, 49 of the 111 patients (44.1%) experienced at least one episode of infection, most of which occurred in the SCI groups. Among those, the high complete SCI group had the highest burden of pulmonary and urinary tract infections, including recurrent episodes of infections (infection load). The low complete SCI group frequently suffered urinary tract and other types of infections (Table 2).

Table 2.

Types and frequency of infections

Type of infection	High complete SCI	High incomplete SCI	Low complete SCI	Low incomplete SCI	VF control group
Pulmonary infections
Tracheobronchitis n (%)a	10 (50.0)	6 (28.6)	3 (18.8)	2 (15.4)	1 (2.4)
Pneumonia, n (%)a	3 (15.0)	2 (9.5)	1 (6.3)	0 (0)	0 (0)
Recurrent pulmonary infections, n (%)a	4 (20.0)	2 (9.5)	0 (0)	0 (0)	0 (0)
Urinary tract infections
Bacteriuria, n (%)a	5 (25.0)	5 (23.8)	4 (25.0)	5 (38.5)	2 (4.9)
Symptomatic urinary infection, n (%)a	10 (50.0)	4 (19.0)	5 (31.3)	2 (15.4)	1 (2.4)
Recurrent urinary infections, n (%)a	5 (25.0)	1 (4.8)	4 (25.0)	2 (15.4)	0 (0)
Other infections
Bloodstream/catheter-related, n (%)a	1 (5)	0 (0)	4 (25.0)	0 (0)	0 (0)
Skin/wound/genital, n (%)a	3 (15)	0 (0)	3 (18.8)	2 (15.4)	1 (2.4)
Gastrointestinal, n (%)a	1 (5)	2 (9.5)	3 (18.8)	1 (7.7)	0 (0)

Because some patients have more than one type of infection and/or recurrent infections, the percentages may sum to over 100%.

In the SCI population (n = 70), the mHLA-DR expression was associated with the occurrence and onset of infections. Patients with infections had mHLA-DR levels below the reference value for ‘immune suppression’ within the first 2 weeks after SCI. Patients with early infection onset in the first or second week had especially low mHLA-DR values in the range of ‘borderline immunoparalysis’ between 15h and 7 days after SCI (Fig. 6). Over time, the mHLA-DR expression remained most suppressed in the group with very early onset of infection in the first week, with the mean and 95% CI in the ‘borderline immunoparalysis’ range up to 14 days after SCI. The mHLA-DR expression in the group without infections was clearly higher from the beginning and recovered to values above ‘borderline immunoparalysis’ by 7 days after SCI. The first infections occurring in the SCI population were primarily pulmonary infections in 26 of 70 patients (37.1%) with a median onset time of 4.4 (IQR 1.9–7.1) days after SCI, followed by urinary tract infections in 19 of 70 patients (27.1%) or other types of infection in 4 of 70 patients (5.7%), with a median onset at 18.1 (12.0­­−40) days or 8.4 (6.3–19.6) days after SCI, respectively.

Figure 6.

SCI-IDS severity (reduced mHLA-DR) associates with acquired infections. Estimated marginal means with 95% CI of mHLA-DR values [ln number of anti-HLA-DR antibodies bound per monocyte (ab/cell)] and its association with the timing of first infections in the SCI population calculated using linear mixed models in the observed data. Numbers in brackets indicate the number of patients with available mHLA-DR measurement within each category of infection onset. Patients without infection until premature dropout were assigned to the censored group. A total of 260 mHLA-DR measurements were included in this analysis. The blue horizontal lines indicate the thresholds towards more serious stages of immune suppression; dotted line ln (<15 000 ab/cell) = immune suppression; dotted/broken line ln (<10 000 ab/cell) = ‘borderline immunoparalysis'; broken line ln (<5000 ab/cell) = ‘immunoparalysis’.^15,16 ab = antibodies; CI = confidence interval; mHLA-DR = monocytic Human Leukocyte Antigen-DR expression; ln = log-transformed; SCI = spinal cord injury.

Discussion

Reducing prevalent,³² and devastating^1-4 infectious morbidity and mortality imposes as unsolved medical need for SCI patients. The SCIentinel study¹⁵ provides prospective multicentre evidence for a neurogenic immune deficiency acquired by SCI patients (SCI-IDS) affecting denominators of cellular and humoral immunity. The SCI-IDS can be disentangled from other, non-neurogenic immune effects caused by polytrauma often associated with SCI, such as post-aggression syndrome.¹⁴ The functional relevance of SCI-IDS driving infection susceptibility is supported by the association of reduced mHLA-DR expression with infections in SCI patients. Presented clinical data is in line with experimental evidence, proving that SCI-IDS increases infection susceptibility in a lesion level- and SCI severity-dependent manner.^10,11 Compared to mHLA-DR values measured in stroke,¹⁷ the values after acute traumatic SCI are lower and in patients with infections within the range of ‘borderline immunoparalysis’. Compared with stroke, in SCI patients, the neurogenic induced immunosuppression is further aggravated by polytrauma²¹ and surgery.³¹ This is supported further by the mHLA-DR values observed in particularly severely affected SCI patients, which can drop to the range of ‘immunoparalysis’.¹⁹

The neuroanatomical topography of the SCI-IDS emerged by Day 3, with the most pronounced decrease in mHLA-DR in the high complete SCI group (nadir) compared with the incomplete SCI or VF groups. The effect of the neurological SCI level further disentangles with time, exemplified by a delayed recovery of mHLA-DR expression in the high complete SCI group with a reported high prevalence of recurrent infections. This is in support of the hypothesis that a key candidate mechanism underlying SCI-IDS,^7,13,33 as identified in earlier preclinical studies, is based on the loss of supraspinal control over spinal sympathetic preganglionic neurons below the neurological level of injury (‘sympathetic decentralization’).^10,11,20,34

Nevertheless, the clinical-immunological interrelations after SCI are complex and interdependent. First, infection susceptibility after SCI is not only driven by impaired host defence but, in addition, by a limited ability to cough³⁵ or swallow.³⁶ Thus, it is not possible to fully distinguish between direct immunological effects and indirect SCI-related motor effects on infection susceptibility. In stroke, low mHLA-DR expression and dysphagia were identified as risk factors for stroke-associated pneumonia independent of each other.¹⁷ Second, serious infections may also affect mHLA-DR levels in return.³⁷ We observed, however, that SCI patients have lower mHLA-DR values early on compared to patients with VF (first two time points), before infections emerge. As early as 15 h after SCI, low levels of mHLA-DR were antedating associated infections occurring in the first week. This implies a direct neurogenic contribution of the acquired SCI-IDS to infection susceptibility. The fact that these mHLA-DR levels then recovered only slowly and partially may also be due to the effects of early infections putatively reducing mHLA-DR expression further. In patients with sepsis, major surgery or brain injury, serious outcomes and mortality are associated with persistently low or declining mHLA-DR levels over time.^18,38-41 The observation of persistently low mHLA-DR expression over the first 2 weeks in SCI patients with early infections reinforces that, in cases of low mHLA-DR, a consistent and early treatment of infections is required to mitigate severe courses and avoid further, associated risks. The circumstance that infections and sepsis represent a main cause of death during primary care of SCI⁴² underscores the relevance of mHLA-DR in SCI pathophysiology.

In line with the corresponding pilot study⁷ and with recent data from an Australian SCI-population,²⁴ neutrophil and monocyte blood counts are transiently elevated during the first week after complete SCI, suggesting an increased mobilization from the bone marrow. At the same time, mHLA-DR expression subsides, indicating monocyte deactivation¹³ and/or release of immature monocytes with low HLA-DR expression. Both represent hallmarks of a blunted innate immune system with impaired capacity to combat bacterial infections.^17-19,21

The observation that peripheral blood lymphocyte counts were at low numbers in both SCI and VF groups was unexpected, since published clinical pilot study results suggested that lymphopenia occurred early only after SCI.^7,24 These studies, however, did not include reference (control) groups with severe spine trauma and major surgery lacking SCI. In conclusion, early lymphopenia after SCI is more likely associated with a post-aggression syndrome and therefore not neurogenic in nature. The fact that early lymphopenia was not observed in experimental sham surgery that mimics unspecific trauma,³³ points further towards an under-representation of the polytrauma element in animal models compared to the clinical condition. Although patients with life-threatening polytrauma were excluded from this study, variations in the severity of accompanying injuries may still constitute an unmeasured confounder contributing to the variance in lymphocyte data. However, concordant with published evidence that accounts for standardized overall injury severity,²⁴ our data do not demonstrate an effect of the neurological level on circulating lymphocyte counts in the first week after SCI.

It should be noted here that SCI-elicited effects on lymphocytes can differ between the systemic blood and compartmentalized immune organs.⁴³ Post-SCI lymphocyte counts in the peripheral blood are distinct from those in the lymphatic tissue, where profound neurological level-dependent lymphocyte depletion is observed (involution of secondary immune organs),^11,44 being in line with a systemic neurological level-dependent ‘maladaptive sympathetic-neuroendocrine adrenal reflex’.¹¹ Therefore, the presented results on circulatory blood cells do not contradict findings detected in spleen (splenocyte apoptosis, decreased antibody synthesis⁴⁴ and impaired host defence).^10,11

The most pronounced effect of the lesion height of SCI on lymphocyte subpopulations in this study was observed in NK cells at 10 weeks, with cell counts remaining below the reference range in the high complete SCI group. This further corroborates previously reported neurological level-dependent SCI effects on NK cells in subacute and chronic SCI.^6,45 Going forward, the multi-faceted nature of lymphocyte interactions will require more targeted studies⁴⁶ to delineate qualitative aspects of lymphocyte dysfunction, e.g. Lucin et al.,⁴⁴ rather than quantitative cell counts alone.

Besides identifying temporary monocytic deactivation as an elementary SCI-IDS symptom affecting cellular immunity, the SCIentinel study also detected novel symptoms inferring diminished humoral immunity, such as reduced serum immunoglobulins (hypogammaglobulinaemia). While hypogammaglobulinaemia is a hallmark of primary immunodeficiencies,⁴⁷ the acquisition of reduced IgG levels secondary to SCI depict an additional sustained depression of the humoral immunity arm, pronounced in patients with high complete lesion. As the hypogammaglobulinaemia was not associated with reduced B-lymphocyte counts, it is likely attributable to other mechanisms such excretory loss of IgG protein⁴⁸ through affected kidney function and/or impaired immunoglobulin synthesis, due to an emerging B-lymphocyte maturation deficit.^11,49 Of note, when entering the chronic SCI phase, an increase of immunoglobulin levels has been reported,⁵⁰ supporting the notion that acute and chronic SCI-IDS phases are different and caused by a distinct composition of prevailing underlying mechanisms.¹³

The occurrence of missing values in the study, particularly at the very early and late timepoints, was expected and considered in the sample size planning.¹⁵ The statistical analyses comprise multiple imputation procedures, currently seen as a state-of-the-art technique to handle missing data, for avoiding bias and improving accuracy of estimates.⁵¹

In summary, an acquired neurogenic SCI-IDS develops as a secondary combined immune deficiency disorder in SCI patients. Affecting multiple components of the immune system, it undermines both cell-mediated and humoral/antibody-mediated immunity. The clinical significance of the SCI-IDS is corroborated as, in the early phase, lesion-dependent, neurogenic aspects can be differentiated from non-neurogenic causes, such as severe trauma and surgery (post-aggression syndrome).

The identification of a neurogenic, SCI severity and lesion level-dependent SCI-IDS developing independent of demographic baseline criteria provides a tangible framework to tackle limitations in current SCI care.⁵² It enables (i) the development of objective parameters to stage infection susceptibility; (ii) informs the development of triage protocols for particularly immune suppressed patients; and (iii) qualifies as a bona fide treatment target to restore host defence and reduce infection susceptibility.

Funding

The trial was supported by grants from the German Research Council, Cluster of Excellence NeuroCure, the Berlin-Brandenburg Center for Regenerative Therapies (BCRT Grant number 81717034), and the Wings for Life Spinal Cord Research Foundation (Grant Number WfL-DE-006/12, WfL-DE-16/16, WfL-DE-11/20). The funders had no role in the design and conduct of the study, data collection and statistical analysis, interpretation of the data, drafting, review, approval of the manuscript or decision to submit the manuscript for publication.

Competing interests

The authors report no competing interests.

Supplementary material

Supplementary material is available at Brain online.