The estimated prevalence of depressive symptoms after a stroke is 31%.
1 Recent studies have suggested that poststroke depression is independently associated with poor health outcomes such as increased mortality, disability, anxiety and lower quality of life.
2 However, most of the traditional risk factors reported in the literature such as genetic factors, age, gender, and stroke severity are not amendable.
3 Identifying the potential biomarkers and detailed mechanism of poststroke depression is the most important need for future study to improve the effectiveness of therapeutic intervention.
3,4Many studies support the role of inflammation in depression.
5,6 However, studies regarding poststroke depression are far fewer and inconclusive. Some recent studies supported the role of proinflammatory cytokines in the pathogenesis of poststroke depression.
7,8 Serum concentrations of interleukin-6 (IL-6) were linked with the apathetic-amotivational and somatic symptoms of depression in the acute phase of stroke.
7 Conversely, Jiménez et al.
9 studied patients with acute stroke and found no difference in C-reactive protein (CRP) or IL-6 in patients with or without poststroke depression. Ferrarese et al.
10 even found that the CRP level decreased in patients with poststroke depression one month poststroke. Most of the studies regarding poststroke depression have been conducted in the acute phase of stroke. However, high levels of IL-6 have been found up to 1 month poststroke,
11 and elevated levels of CRP may be induced by stroke itself and have been reported to be sustained up to three months after stroke.
12Therefore, those findings cannot be generalized to patients with stroke in the subacute or chronic phase.
Oxidative stress has been found to be involved in the pathophysiology of depression.
13–15 Recent studies have found that total antioxidant capacity
16 was decreased and that 8-hydroxy-2′-deoxyguanosine (8-OHdG),
14 a modified DNA base that indicates oxidative DNA damage, was increased in patients with depression. Animal studies have also found that oxidative stress was involved in the pathophysiology of poststroke depression.
17 However, few studies on the role of oxidative stress in the mechanisms of poststroke depression have yet been performed in clinical population, and the temporal profiles of elevated oxidative stress level due to acute stroke are currently unclear. Gu et al. recently found lower serum uric acid in the acute phase of stroke as a biological predictor of poststroke depression three months after stroke and speculated that defense against oxidative stress was a mechanism.
18The association between depression and obstructive sleep apnea is controversial. A recent systematic review and meta-analysis study suggested that treatment of obstructive sleep apnea might reduce depressive symptoms based on questionnaires in stroke-free patients.
19 However, obstructive sleep apnea severity was not independently related to depression diagnosed by standardized psychiatric interview in patients with untreated obstructive sleep apnea.
20 Moreover, neither the severity nor the existence of obstructive sleep apnea was related to depression questionnaire score in healthy elderly individuals.
21 Similar to poststroke depression, obstructive sleep apnea is also very common in patients with stroke, with a prevalence rate exceeding 50%.
22 However, few studies have been conducted focusing on the association between poststroke depression and obstructive sleep apnea in patients with stroke. Only two small studies reported improvement in depression following continuous positive airway pressure treatment for sleep-disordered breathing in patients with acute stroke.
23,24 As those studies did not recruit patients without obstructive sleep apnea, they do not elucidate whether poststroke depression is independently associated with obstructive sleep apnea.
The primary goal of this study was to evaluate the association of poststroke depression with potentially correctable targets, including inflammation, oxidative stress, and obstructive sleep apnea in patients with subacute stroke.
Methods
Participants
This cross-sectional study recruited patients with subacute ischemic stroke who had been admitted consecutively for neurorehabilitation. Ischemic stroke was diagnosed based on neurological examination and image studies. Patients who met any of the following criteria were excluded: aphasia or severely decreased consciousness causing communication difficulty; unstable medical and neurological status such as active infection, delirium, or poor controlled diabetic mellitus; and advanced renal disease (chronic kidney disease, stage 3 or above) or history of central nervous system disease such as hemorrhage, Parkinson’s disease, or malignancy. Patients with central sleep apnea were also excluded. Drugs or food supplements, which may ameliorate inflammation or oxidative stress, were prohibited except medication prescribed by physicians.
The study was approved by the ethics committee of Chang Gung Memorial Hospital. All participants provided informed consent.
Clinical Evaluation
Comprehensive evaluations including demographic data and history of traditional vascular risk factors were performed during admission. The National Institutes of Health Stroke Scale (NIHSS)
25 at stroke onset was obtained from relevant medical reports to represent initial stroke severity. Functional outcome was measured by using the Barthel Index.
26Depression was assessed by using the Patient Health Questionnaire–9 (PHQ-9), which is a self-reported measure of depression matching the
DSM-IV criteria of major depression.
27 A cutoff based on a summed-items score of 10 or above is recommended in screening for major depressive disorder.
28 The diagnostic value of PHQ-9 scores ≥10 for poststroke depression was found to be good, with a sensitivity of 0.80 (95% [confidence interval] CI=0.62–0.98) and a specificity of 0.78 (95% CI=0.69–0.83).
29Depression may be defined by PHQ-9 scores ≥10 according to the literature, yet this study referred that patients with PHQ-9 scores ≥10 had depressive symptoms.
Polysomnography
Polysomnography was performed with an Embla N7000 (Somnologica, Reykjavik, Iceland) in the sleep center from 10:00 p.m. to 7:00 a.m. on the same day of Barthel Index and PHQ-9 evaluation.
The test comprised six electroencephalography (EEG) channels (F3-A1, F4-A2, C3-A1, C4-A2, O1-A1, and O2-A2), thoracic and abdominal movement sensors (inductance plethysmography), a chin and bilateral anterior tibial surface electromyogram, an oxyhemoglobin saturation detector (finger pulse oximetry), an electrocardiogram, an electro-oculogram, and nasal and oral airflow sensors (nasal pressure cannula and oronasal thermistor). Diagnosis was made by a board-certified sleep physician and was based primarily on the American Academy of Sleep Medicine Task Force recommendations:
30 Severe obstructive sleep apnea was diagnosed when the apnea-hypopnea index was >30 events/hour
−1 and >50% of respiratory events were of the obstructive or mixed type. A diagnosis of central sleep apnea was made when ≥50% of respiratory events were of the central type.
Inflammation and Oxidative Stress Biomarkers
Samples of peripheral venous blood and urine were taken at 7:00 a.m. after the polysomnography study. Blood was collected in lithium heparin PST II tubes and SST II Advance tubes (BD Vacutainer; Becton Dickinson, Heidelberg, Germany) and was centrifuged at 3,000 rpm for 10 minutes. Then, samples were immediately separated in aliquots and stored at −80°C until analysis in the Medical Center Laboratory. The total antioxidant capacity was determined on the Cobas Mira Plus (Roche Diagnostics, Mannheim, Germany) by the ferric reducing ability of plasma assay
31 with intra- and interassay coefficients of variation of 2% and 5%, and 1% and 3%, respectively, at two levels. The urinary 8-OHdG was measured by microplate competitive enzyme-linked immunosorbent assay (ELISA)
32 with intra- and interassay coefficients of variation of 5.9% and 8.0%, respectively. The level of urinary 8-OHdG excretion was then normalized by the urinary creatinine concentration and expressed as the urinary (8-OHdG [nanograms/milliliter]/creatinine [milligrams/milliliter]) ratio. Serum IL-6 was measured using ELISA kits (Quantikine HS IL-6 Immunoassay, #HS600B; R&D Systems, Minneapolis, MN) and quantified at 490 nm by a microplate reader (SpectraMax 190; Molecular Devices, Sunnyvale, Calif.), with a minimum detectable dose of 0.039 pg/ml and inter- and intra-assay coefficients of variation <10%. CRP was determined with a latex aggregation immunoassay (Nanopia CRP assay; Daiichi Pure Chemicals, Tokyo, Japan) using a Hitachi 7600-210 analyzer (Hitachi Instruments Engineering, Tokyo, Japan), which has a lower limit of sensitivity of 0.1 mg/l and inter- and intra-assay coefficients of variation ≤5%.
Statistical Analyses
Statistical analyses were performed using IBM SPSS Statistics software, version 20 (IBM Software Group; Chicago, IL). Continuous variables were assessed for normal distributions. Non-normally distributed variables were log-transformed (NIHSS, IL-6, and total cholesterol and high-density lipoprotein [HDL]) to achieve normality and then analyzed with parametric tests and presented as geometric means. Participants were stratified into depressive (PHQ-9 scores ≥10) and nondepressive (PHQ-9 scores <10) groups. Nominal variables were compared using a chi-square test. Continuous variables were compared using the Student’s t test or two-sample Kolmogorov–Smirnov test, on the basis of whether they were normally distributed. Univariate correlations between PHQ-9 score and continuous variables were determined with Spearman correlation analysis. Potential variables were selected if they were related to PHQ-9 scores at p<0.1 and then entered into the multivariate linear regression model to calculate the independent predictors for PHQ-9 score. Values of p<0.05 were considered significant.
Results
In total, 139 patients (97 men [69.8%] and 42 women [30.2%]; mean age, 63.2±13.4 years) with recent ischemic stroke were consecutively recruited.
Table 1 shows the clinical characteristics and sleep parameters of the two groups. Time interval from stroke onset to the date of polysomnography study and parameters of obstructive sleep apnea, including apnea-hypopnea index and disturbance index, did not differ between the two groups (
Table 1). BMI, stroke severity (NIHSS and Barthel Index), percentage of antidepressant (including tricyclic antidepressant, selective serotonin uptake inhibitor, and serotonin-norepinephrine reuptake inhibitor) usage, and percentage of rapid eye movement (REM) sleep differed significantly between the two groups (
Table 1). Participants with hyperlipidemia were all on statin therapy for secondary stroke prevention. No participants with a diagnosis such as asthma, autoimmune disease, celiac disease, glomerulonephritis, or inflammatory bowel disease, which could profoundly affect the inflammatory status, were recruited in this study. Steroids, nonsteroidal anti-inflammatory drugs, and specific classes of drug with a main action on inflammation were not prescribed during the study period except aspirin. Sleep-onset latency, level of IL-6, and urinary 8-OHdG were marginally different between the two groups (
Table 1).
Table 2 presents the univariate correlation between PHQ-9 score and continuous variables. The PHQ-9 score was significantly correlated with the levels of total antioxidant capacity, Hs-CRP, and urinary 8-OHdG and was marginally significantly correlated with age and level of IL-6. However, the PHQ-9 score was not significantly correlated with major parameters of obstructive sleep apnea, such as apnea-hypopnea index score and desaturation index score.
Multivariate linear regression analyses were performed based on inflammation and oxidative stress biomarkers, together with other independent variables, including BMI, sleep-onset latency, Barthel Index score, mean SpO2 (oxyhemoglobin saturation by pulse oximetry), age, antidepressant usage, and percentage of REM sleep, to predict the PHQ-9 score. The BMI, sleep-onset latency, Barthel Index score, antidepressant usage, and urinary 8-OHdG remained significantly correlated with PHQ-9 score (R
2=0.317, p<0.001) (
Table 3). Levels of total antioxidant capacity, Hs-CRP, and IL-6 were not significantly associated with PHQ-9 score after adjusting for other independent variables.
Discussion
This study is the first to our knowledge to report the independent association between increased oxidative stress and severity of depressive symptoms in patients with subacute ischemic stroke after adjusting for potential risk factors. Therefore, intervention to reduce oxidative stress is a possible treatment option for depressive symptoms in patients with ischemic stroke.
The findings of independent and positive association between urinary 8-OHdG, one of the most robust biomarkers of oxidative stress, and depressive symptoms suggest that poststroke depressive symptoms are related to raised oxidative damage to DNA. These findings are in agreement with the result of a recent systematic review and meta-analysis study that 8-OHdG was increased in stroke-free patients with depression.
14 The PHQ-9 score performs well as a continuous scale,
27 but the best PHQ-9 cutoff scores for screening poststroke depression remain to be proved. The findings that PHQ-9 scores ≥10 did not provide the highest sensitivity and specificity for poststroke depression screening
33 might be the reason why the differences in urinary 8-OHdG level between the two groups did not reach statistical significance. The concomitant raised in oxidative DNA damage and lack of an adequately corresponding total antioxidant capacity change suggest that the antioxidant/pro-oxidant balance is perturbed by the outweighing oxidative stress in patients with stroke who had severe depressive symptoms. Depression has been found to be linked with accelerated cellular aging reflected by telomere length shortening.
34 As an elevated level of 8-OHdG was found to be a reliable biomarker of oxidative DNA damage in patients with chronic stroke
35 and independently associated with telomere shortening in patients with diabetes,
36 it is biologically plausible to suggest that the mechanism of the independent association between poststroke depressive symptoms and elevated level of urinary 8-OHdG may result from accelerated cellular aging. The independent association between urinary 8-OHdG and PHQ-9 score found in this study, together with the findings that lower levels of uric acid might predict poststroke depression
18 and a phase 2 clinical trial in Parkinson’s disease that revealed a potential effect of increased uric acid on mood,
37 suggests that raised oxidative stress might be a risk factor and a potentially treatable target for depressive symptoms in patients with ischemic stroke. As poststroke depression is a well-known predictor for poor recovery from stroke,
3 this study explains why high urinary 8-OHdG has been reported to be a valid predictor of poor functional outcomes in patients with acute and chronic stroke.
35,38Inflammation is not independently associated with depressive symptoms in patients with subacute stroke. The main difference between this investigation and previous studies regarding the relationship between inflammation and poststroke depression is the timing after the index stroke when patients were recruited. The influence of acute ischemic stroke on early and sustained inflammatory response
9,11 gradually attenuated in the subacute phase of stroke, reducing the bias in the association between inflammation and depressive symptoms in this study. Our results are also consistent with a meta-analysis study that did not find any association between poststroke depression and inflammatory biomarkers (CRP, IL-18, IL-6, and tumor necrosis factor α) in patients within a mean of five weeks poststroke.
39 Although all recognized possible covariates were analyzed and adjusted by multivariate regression analyses, we could not exclude unknown sources of bias. This is particularly true, given that inflammatory biomarkers might be influenced by many comorbidities and medication in patients with stroke. Larger studies that measure more inflammatory biomarkers in patients with subacute or chronic stroke are needed to confirm these findings.
Neither the presence of severe obstructive sleep apnea nor the severity of obstructive sleep apnea is associated with depressive symptoms in patients with subacute stroke. Whether obstructive sleep apnea in the elderly individuals represents the same disease as that in the younger subjects remains inconclusive.
40 As the mean age of 63.2 in this study is high, the analytical results support the findings of Sforza et al. that obstructive sleep apnea is not related to depression in healthy elderly individuals.
21 Lavie et al. reported an unexpected survival advantage from sleep apnea in elderly individuals and speculated that elderly individuals with sleep apnea may be exceptionally resistant survivors who are exceptionally adapted to the resulting prolonged stress.
41 Our previous findings that patients with ischemic stroke who have severe obstructive sleep apnea developed adaptive antioxidative response
42 supported their theory and may explain why this study found no association between obstructive sleep apnea and poststroke depressive symptoms.
The high percentage of participants with depressive symptoms in this study is consistent with previous findings by Farner et al.
43 and may be owing to the selection bias that participants admitting for inpatient rehabilitation generally suffered from moderate to severe stroke. Place of residence, living together or alone, and participating in active rehabilitation programs are factors reported to be associated with poststroke depression.
44 As participants were admitted for neurorehabilitation, similar physical activities and living environment made these variables well controlled in this study.
This study has several limitations. First, this cross-sectional study cannot explain the causal relationship between oxidative DNA damage and depressive symptoms in patients with subacute ischemic stroke. Second, the selection bias exists; therefore, an extrapolation to all patients with ischemic stroke is not possible. Patients with ischemic stroke who also had severe depression, active psychiatric disorder, or neurological deficits of very mild or severe degree were excluded because they were not suitable for inpatient neurorehabilitation. Third, the analysis did not include prestroke history of psychiatric disorders or social–economic status.
Finally, the poststroke depressive symptoms were evaluated only by the PHQ-9, without further follow-up with a more detailed psychiatric assessment using the DSM-IV.
In conclusion, increased oxidative DNA damage is associated with depressive symptoms in patients with subacute ischemic stroke. Urinary 8-OHdG may serve as a potential biomarker for poststroke depression. Further longitudinal studies are indicated to elucidate the causal relationship between poststroke depression and the elevated oxidative stress level.
Acknowledgments
The authors thank Ted Knoy for his editorial assistance.