Infarct Volume is a Major Determiner of Post-Stroke Immune Cell Function and Susceptibility to Infection
Background and Purpose— Acute ischemic stroke in humans is associated with profound alterations in the immune system. Hallmarks of this stroke-induced immunodepression syndrome are: lymphocytopenia, impairment of T helper cell and monocyte function. We studied which stroke-specific factors predict these immunologic alterations and subsequent infections.
Methods— Leukocyte/lymphocyte subsets were assessed serially by white blood cell count and fluorescence-activated cell sorter analysis in ischemic stroke patients (n=50) at baseline, day 1, and day 4 after stroke onset and compared to an age-matched control group (n=40). Concomitantly, monocytic human leukocyte antigen-DR expression and the in vitro function of blood monocytes measured by the production of tumor necrosis factor-α upon stimulation with lipopolysaccharide were assessed. Associations of these immunologic parameters with stroke specific factors (National Institutes of Health Stroke Scale, infarct size) were explored. Multivariable logistic regression analysis was applied to identify early predictors for poststroke respiratory and urinary tract infections.
Results— Infarct volume was the main factor associated with lymphocytopenia on day 1 and day 4 poststroke. Particularly, blood natural killer cell counts were reduced after stroke. Monocyte counts increased after ischemia paralleled by a profound deactivation predominantly after extensive infarcts. Reduced T helper cell counts, monocytic human leukocyte antigen-DR expression, and monocytic in vitro production of tumor necrosis factor-α were associated with infections in univariate analyses. However, only stroke volume prevailed as independent early predictor for respiratory infections (OR 1.03; CI 1.01 to 1.04).
Conclusions— Infarct volume determines the extent of lymphocytopenia, monocyte dysfunction, and is a main predictor for subsequent infections.
Acute ischemic stroke is associated with a variety of serious medical complications which independently predict worse outcome.1 Particularly common complications are pneumonia and urinary tract infections.2,3 Recent studies suggest a stroke-induced severe depression of systemic immunity rendering patients susceptible for infections.4,5 This stroke-induced immunodepression (SIDS) affects both innate and adaptive immunity.6–8 Hallmarks of SIDS in humans appear to be (1) lymphocytopenia, (2) functional deactivation of T helper (Th) cells, and (3) functional deactivation of monocytes.6–8
Infarct size was predictive for poststroke infections9,10 and inversely correlated with admission and poststroke (day 2) T cell and cytotoxic T cell (CTL) counts in one study.8 However, in another study no significant association between infarct volume as dichotomous variable (cutoff 20 mL) and serial T cell counts was found from baseline through day 6.6 Because of these inconsistencies we analyzed associations of infarct size and serial blood leukocyte subset measurements. Furthermore, we investigated the association of stroke size and human leukocyte antigen (HLA-DR) expression, which is a marker for the activation status of blood monocytes.11 Interrelations might be of clinical significance because reduced poststroke HLA-DR expression levels on monocytes were associated6,7,12 and even independently predictive (measured on day 1 [d1] poststroke) for subsequent infections.13 Moreover, other immunologic parameters like d1 relative lymphocyte and absolute Th cell counts7 as well as admission tumor necrosis factor-α (TNF-α) expression after in vitro stimulation of whole blood specimens with lipopolysaccharide (LPS)6 were proposed as independent predictors for poststroke infections. Hence, we performed multivariable regression analyses to find independent early predictors for the development of poststroke infections in our data.
Materials and Methods
All study procedures were performed after obtaining informed consent according to a protocol approved by the independent local ethics committee of the Medical Faculty of the University Heidelberg. Fifty patients with acute ischemic stroke and symptom onset of less than 12 hours (h) were enrolled. Control individuals (n=40) were patients scheduled to undergo elective cataract surgery without history of stroke, myocardial infarction, or peripheral artery disease. Patients with clinical signs of infection on admission, minor stroke (National Institutes of Health Stroke Scale [NIHSS] <3), and known immunologic diseases were excluded. Clinical diagnosis of an ischemic stroke was confirmed by CT or MR imaging. Blood was obtained from patients on admission (mean 6.5; SEM 0.7 hour), day 1 (28.7±1.1 hour), and day 4 (105±3.4 hours). All patients received follow-up CT/MR imaging within 24 to 36 hours after admission (CT n=33, MRI n=17). Infection was defined by the combination of the following 2 criteria during in-hospital stay: (1) presence of suggestive clinical and laboratory or radiological signs of infection (eg, urinary tract symptoms, productive cough, pleuritic pain, dyspnea, tachypnea, fever, cultures positive for a pathogen, leukopenia [<4/L] or leukocytosis [>12/L], chest X-ray infiltrate); (2) serum C-reactive protein of more than 50 mg/mL.
CT images were acquired on a multislice CT-scanner (Somatom-16, Siemens Medical Systems). MR images were acquired using a 1.5 or 3 Tesla scanner (MR-Symphony, MR-TRIO, Siemens Medical Systems). 24- to 36-hour follow-up imaging (CT/MRI) was used to assess ischemic lesion volumes (OSIRIS 4.19, University Hospital Geneva). Because a definition for small and large infarcts does not exist, the infarct volume median (36.9 mL) was used to dichotomize the stroke cohort.
Standard laboratory differential white blood cell counts were used to assess leukocyte subsets. Lymphocyte subsets were determined by fluorescence-activated cell sorter analysis on a FACScalibur (Becton Dickinson) using a CD45-gating 3-color approach14 with the following antibodies: CD3-FITC, CD4-PE, CD8-PE, CD19-PE, CD-56-PE, CD45PE-Cy5 (Becton Dickinson). Absolute subset counts were calculated by multiplication with the absolute number of lymphocytes. Quantitative HLA-DR expression analysis was performed using the QuantiBRITE Anti–HLA-DR-PE/Anti-Monocyte-PerCP-Cy5.5 kit according to the manufacturer’s instructions (Becton Dickinson). In vitro stimulation of monocytes was performed on whole blood specimens within 2 hours after collection with 100 ng/mL highly purified lipopolysaccharide from Salmonella minnesota (HL63, smooth form, obtained from U. Seydel, Borstel) for exactly 24 hours at 37°C/5%CO2 and measuring secretion of TNF-α in cell-free supernatant by ELISA (OptEIA, BD Pharmingen).
Statistical analysis was performed using SPSS-statistical package 17.0, Stata9, and MATLAB 7.0 for Windows. Serial leukocyte subset counts and HLA-DR expression were analyzed using general linear models. First, an overall F-statistic was performed including main effects and interactions. Two-sided probability values of <0.05 were regarded as significant. Exclusively for significant main effects/interactions post hoc tests of main and simple main effects were performed including correction methods for multiple testing (Bonferroni, Dunnett’s T3) as appropriate relating to homogeneity of group variances. Multivariable logistic regression was applied to analyze predictor variables for infections. Variable selection was performed by a combined stepwise forward/backward approach. Covariates included for selection were: d1 Th cell counts, d1 relative lymphocyte counts, d1 absolute HLA-DR expression on monocytes, NIHSSS on admission, infarct volume, infarct side, infarct location (cortical/subcortical), age, and sex. Age and sex were forced to stay in the final model. For the covariates NIHSSS and stroke volume odds ratios (OR) apply per NIHSSS point and ml.
Fifty patients with acute ischemic stroke were enrolled. Stroke localization was: left middle cerebral artery (n=17), right middle cerebral artery (n=23), total left anterior circulation (n=2), total right anterior circulation (n=2), left anterior cerebral artery (n=1), and vertebrobasilar circulation (n=5). According to the TOAST criteria,15 stroke etiology was: cardioembolism in 50%, large-artery atherosclerosis in 14%, artery dissection in 6%, unknown in 30%. Sample group characteristics are summarized in the Table.
Dynamics of Blood Leukocyte and Lymphocyte Subset Counts After Acute Ischemic Stroke
Compared to controls, neutrophil counts in the stroke group were elevated at baseline and d1 (Figure 1A). No significant poststroke alterations in lymphocyte counts were observed, neither comparing groups nor comparing single visits within the stroke cohort (Figure 1B). No differences between groups were obvious for monocyte counts. However, within the stroke group monocyte counts were higher on d1 and d4 compared to baseline (Figure 1C). Basophil counts were lower on d1 poststroke compared to control individuals (Figure 1D). Eosinophil and platelet counts did not change significantly after stroke (Figure 1E and 1⇓F). Mean cell counts of all leukocyte subsets in the control and stroke (all visits) group were within reference ranges.
Comparing lymphocyte subsets between control individuals and stroke patients, only natural killer (NK) cell counts were significantly different with lower values on d1 and d4 (Figure 2B). Within the stroke group no significant alterations between single visits were evident. However, a general trend toward lower subset counts particularly on d1 after stroke onset was observed. Mean cell counts of all lymphocyte subsets in the control and stroke (all visits) group were within reference ranges.
Stroke Severity Affects Leukocyte Subset Counts and Predicts Infection
During the in-hospital stay (6 days [interquartile range, IQR] 4 to 11), 36% and 14% of patients developed respiratory and urinary tract infections, respectively. Baseline NIHSSS was higher in patients with infections (16 [IQR 13 to 19] versus 8 [IQR 5 to 13]). Stroke volume was larger in the infection cohort (152.3 mL [standard deviation, SD 116.7] versus 21.6 mL [SD 27.3]). NIHSSS and stroke volume were highly correlated (Spearman rho=0.71, P<0.001). In multivariable logistic regression (covariates in final model: NIHSSS, age, sex) baseline NIHSSS was independently predictive for subsequent respiratory infections (OR: 1.4, CI 1.2 to 1.7; Nagelkerke-R2 0.52). Substituting baseline NIHSSS by stroke volume clearly improved the model fit (OR 1.03, CI 1.01 to 1.04; R2 0.7; supplemental Figure IA, available online at http://stroke.ahajournals.org). In contrast, neither baseline NIHSSS nor stroke volume was independently predictive for subsequent urinary tract infections.
With stroke volume being the most accurate early predictor for subsequent respiratory infections in our data we investigated its impact on leukocyte subsets. For this analysis stroke volume was dichotomized (small/large) using the median infarct volume as group separator with 82% sensitivity and 79% specificity for the prediction of all subsequent infections (supplemental Figure IB).
The increment of neutrophil counts on d1 was pronounced in the large infarct cohort and persisted—differently from the small infarct group—on d4. Moreover, neutrophil counts within the large infarct cohort increased from baseline to d1, reaching mean counts above the upper reference value (Figure 3A). Infarct volume had no significant influence on monocyte counts (Figure 3B). Infarct size was associated with lower lymphocyte counts in the large compared to the small infarct cohort on d1 and d4 (Figure 3C). B cell counts were lower in patients with large infarcts early at baseline and on d4 (Figure 3D). T cell counts on d4 were lower in the large compared to the small infarct cohort (Figure 3E). The significant reduction of NK cell counts on d1 and d4 was pronounced in the large infarct group (Figure 3F). Yet, significance was missing most likely because of limitations in statistical power (see Figure 2B for main effects).
Stroke-Associated Increase in Monocyte Counts Is Accompanied by Deactivation of Monocytes
In contrast to the poststroke rise in monocyte counts the activation status of monocytes—as measured by HLA-DR expression—declined. Compared to controls, HLA-DR expression from baseline through d4 showed a substantial reduction in the large and to a lower extent in the small infarct cohort (Figure 4A). Within the large infarct group HLA-DR expression on d4 was even lower compared to baseline. Moreover, on d1 and d4 the HLA-DR expression was lower in the large compared to the small infarct group. Analyzing the association of HLA-DR expression and elapsing time from stroke onset we observed a rapid HLA-DR expression drop with accentuation in the large infarct cohort (Figure 4B). To evaluate whether reduced HLA-DR values were associated with infections, receiver-operator characteristic (ROC) curve analysis was applied suggesting the most accurate classification of infections (sensitivity/specificity of 86/83%) using 9000 HLA-DR molecules/monocyte and a poststroke time window of more than 24 hours (Figure 4C). The deactivation of monocytes translated into a decreased in vitro response upon stimulation with LPS measured by TNF-α secretion from baseline through day 4 (Figure 4D). Moreover, on d4 TNF-α release was lower in the large compared to the small infarct cohort.
Previous studies in humans suggest an immunodepression syndrome after acute ischemic stroke which mainly affects lymphocyte counts, Th cell function, and the activation status of monocytes.6–8,12,13 Moreover, immunologic parameters like the reduced in vitro production of TNF-α upon stimulation with LPS on admission,6 reduced relative lymphocyte and Th cell counts on d1 after stroke,7 and a reduced HLA-DR expression on blood monocytes on d1 poststroke13 were proposed as independently predictive for subsequent infections.
In contrast to previous studies no significant poststroke reduction in total lymphocyte counts was evident in our stroke group. However, stroke size as measured by the infarct volume on 24- to 36-hour follow-up CT/MR imaging was significantly associated with reduced lymphocyte counts after adjusting for age, sex, and visit (data no shown), suggesting an inverse relation between stroke volume and lymphocyte counts. Confirmatory, dichotomizing our stroke cohort based on infarct volume revealed reduced lymphocyte counts on d1 and d4 as well as reduced Th cell counts on d4 in the large infarct group. Hence, our data strongly suggest an inverse association of infarct volume and lymphocyte counts which is in contrast to a previous finding (eg, that Th cell counts are not inversely correlated with infarct volume).7 Respecting an inverse association of infarct volumes and lymphocyte counts, the absent poststroke lymphocytopenia in our data could be caused by overall smaller infarcts in our stroke cohort. However, the median infarct volume in our study population (36.9 mL) was within the range of previous reports (172 mL,7 20.5 mL8). Thus, our data suggest that among others (e. g. stroke severity) infarct volume is one precipitator of stroke-induced lymphocytopenia. Hemispheric lateralization and cortical versus subcortical location had no independent effect on lymphocyte counts after controlling for infarct volume consistent with a recently published experimental study.16 Another explanation for the missing stroke-induced lymphocytopenia in our data might be the relatively low lymphocyte counts in our control group (1.58/nl) compared to 2/nl in other studies.7
In contrast to previous findings6–8 there was no evidence for a very early lymphocyte decrease at baseline; different timing of baseline sampling among studies (ie, delayed or earlier baseline sampling may have led to a more pronounced lymphocytopenia) might account for this. Our mean time window for baseline sampling was 6.5 hours. In one study 72% of patients were sampled within 4 hours.8 For 2 other studies exact time windows were not reported (<12 hours,7 <22 hours6).
Contrary to previous reports,6,7 lymphocytopenia was most evident in the NK cell subset in the present study. This is in accordance with preclinical data where reduced NK cell counts rendered mice susceptible to septicemia.17 Moreover, we could not find an early reduction of B cell counts in the whole stroke group.8 Rather, infarct volume inversely correlated with B cell counts at baseline and d4.
Our data do not confirm a recent report in which d1 relative lymphocyte and absolute Th cell counts were independent predictors for infections (after controlling for stroke severity and infarct volume7). Another proposed predictor for poststroke infections was the reduced in vitro production of TNF-α on admission. We observed a clear reduction of the in vitro TNF-α release between patients and controls. However, baseline differences between infection and noninfection cohorts were not evident. Another proposed independent and early predictor of poststroke infections was the reduced level of d1 HLA-DR expression on blood monocytes.13 In univariate analysis we could confirm reduced d1 levels of HLA-DR expression in patients with subsequent infections. However, controlling for infarct size set off this association. Between d1 and d4 reduced HLA-DR levels were clearly and independently (d4) associated with poststroke infections. Meanwhile in our data, the best early predictor for poststroke infections was infarct volume.
Conflicting results exist regarding poststroke monocyte counts.6,7,18,19 Corroborating 3 previous reports we observed an increase of monocyte counts after stroke.7,18,19 However, monocyte counts were not different between the control and stroke group. Furthermore, poststroke monocyte counts were mainly unaffected by infarct volume while monocyte activation was strongly reduced in large infarcts. This suggests monocyte count increments a more general poststroke feature, whereas monocyte deactivation, which translated into a reduced capacity to respond to LPS stimulation, was observed in patients with larger infarct volumes.
A limitation of our study is the limited size of patient and control groups, although the control group was relatively large compared to previously published reports (n=40 versus 30,6 14,7 and 12,8 respectively). A large and closely matched control group appears relevant because leukocyte subset as well as HLA-DR values spread widely within reference ranges. Ischemic stroke patients are a heterogenous population. Hence, stroke type as well as associated risk factors could contribute to immunologic alterations, which are particularly relevant regarding our moderate sample size. An additional limitation of our study were the different follow-up imaging techniques (MRI/CT) that were applied to measure infarct volume. Finally, another limitation is the short-term follow-up period, which does not allow correlations of long-term clinical outcome with infarct volume, immunologic alterations, and infection status.
In summary, our study identifies infarct volume as a major determiner of the extent of postischemic lymphocytopenia and monocyte dysfunction, which are markers of susceptibility to infection. Because it is the most consistent early predictor for the manifestation of poststroke infections, infarct size can serve to identify patients at high risk for infectious complications.
Sources of Funding
This study was supported by a grant from the Else Kröner Fresenius Stiftung to R.V. R.V. is supported by an Else-Kröner Memorial scholarship.
- Received May 15, 2009.
- Revision received July 9, 2009.
- Accepted July 10, 2009.
Johnston KC, Li JY, Lyden PD, Hanson SK, Feasby TE, Adams RJ, Faught RE Jr, Haley EC Jr. Medical and neurological complications of ischemic stroke: Experience from the ranttas trial. Ranttas investigators. Stroke. 1998; 29: 447–453.
Davenport RJ, Dennis MS, Wellwood I, Warlow CP. Complications after acute stroke. Stroke. 1996; 27: 415–420.
Indredavik B, Rohweder G, Naalsund E, Lydersen S. Medical complications in a comprehensive stroke unit and an early supported discharge service. Stroke. 2008; 39: 414–420.
Chamorro A, Urra X, Planas AM. Infection after acute ischemic stroke: A manifestation of brain-induced immunodepression. Stroke. 2007; 38: 1097–1103.
Dirnagl U, Klehmet J, Braun JS, Harms H, Meisel C, Ziemssen T, Prass K, Meisel A. Stroke-induced immunodepression: experimental evidence and clinical relevance. Stroke. 2007; 38: 770–773.
Vogelgesang A, Grunwald U, Langner S, Jack R, Broker BM, Kessler C, Dressel A. Analysis of lymphocyte subsets in patients with stroke and their influence on infection after stroke. Stroke. 2008; 39: 237–241.
Urra X, Cervera A, Villamor N, Planas AM, Chamorro A. Harms and benefits of lymphocyte subpopulations in patients with acute stroke. Neuroscience. 2008.
Harms H, Prass K, Meisel C, Klehmet J, Rogge W, Drenckhahn C, Gohler J, Bereswill S, Gobel U, Wernecke KD, Wolf T, Arnold G, Halle E, Volk HD, Dirnagl U, Meisel A. Preventive antibacterial therapy in acute ischemic stroke: a randomized controlled trial. PLoS ONE. 2008; 3: e2158.
Schnizlein-Bick CT, Mandy FF, O'Gorman MR, Paxton H, Nicholson JK, Hultin LE, Gelman RS, Wilkening CL, Livnat D. Use of cd45 gating in three and four-color flow cytometric immunophenotyping: guideline from the National Institute of Allergy and Infectious Diseases, Division of AIDS. Cytometry. 2002; 50: 46–52.
Adams HP Jr, Bendixen BH, Kappelle LJ, Biller J, Love BB, Gordon DL, Marsh EE, III. Classification of subtype of acute ischemic stroke. Definitions for use in a multicenter clinical trial. TOAST. Trial of Org 10172 in Acute Stroke Treatment. Stroke. 1993; 24: 35–41.
Liesz A, Hagmann S, Zschoche C, Adamek J, Zhou W, Sun L, Hug A, Zorn M, Dalpke A, Nawroth P, Veltkamp R. The spectrum of systemic immune alterations after murine focal ischemia. Immunodepression versus immunomodulation. Stroke. In press.
Prass K, Meisel C, Hoflich C, Braun J, Halle E, Wolf T, Ruscher K, Victorov IV, Priller J, Dirnagl U, Volk HD, Meisel A. Stroke-induced immunodeficiency promotes spontaneous bacterial infections and is mediated by sympathetic activation reversal by poststroke t helper cell type 1-like immunostimulation. J Exp Med. 2003; 198: 725–736.