Decreased Serum Retinoic Acid May Predict Poor Outcome in Ischemic Stroke Patients.

BACKGROUND AND AIMS: Decreased serum retinoic acid (RA) levels have been shown to be linked with increased mortality in cardiovascular diseases. This study aimed to investigate the relationship between serum RA and 3-month functional outcome after ischemic stroke.
METHODS: Between January 2019 and September 2019, we prospectively recruited ischemic stroke patients within 24 hrs of symptom onset. Serum RA levels were measured for all patients at admission. The primary outcome was defined as poor functional outcome (modified Rankin Scale 3-6) at 90 days. The secondary outcome was defined as early neurological deterioration (END), which is considered as an increase of ≥1 point in motor power or total National Institutes of Health Stroke Scale score of ≥2 points within 7 days.
RESULTS: A total of 217 patients were included in the analysis. The median RA levels were 2.9 ng/mL. Ninety-four (43.3%) and 65 (30.0%) patients experienced 3-month poor outcome and END, respectively. After adjusted for potential confounders, decreased levels of serum RA were associated with a higher risk of poor outcome (P for trend = 0.001) and END (P for trend = 0.002). Adding RA quartile to the existing risk factors improved risk prediction for poor outcome [net reclassification improvement (NRI) = 42.6%, P = 0.001; integrated discrimination improvement (IDI) = 5.7%, P = 0.001] and END (NRI index = 45.4%, P = 0.001; IDI = 4.3%; P = 0.005).
CONCLUSION: Low serum RA levels at baseline were associated with poor prognosis at 90 days after ischemic stroke, suggesting that RA may be a potential prognostic biomarker for ischemic stroke.

Introduction

Epidemiological studies have demonstrated that ischemic stroke is the most common cause of death in China.1 Established traditional risk factors, such as advanced age, hypertension and diabetes mellitus, can only explain part of poor prognosis of patients.2–4 Because of the high risk of disability and mortality after ischemic stroke, it is urgent to determine novel biomarkers to improve the prediction of stroke outcomes.

As a major active metabolite of vitamin A, retinoic acid (RA) functions as a natural ligand for receptors, such as retinoic acid receptor, retinoid X receptor, and nuclear hormone receptors, and regulates the transcription of various kinds of genes controlling cell differentiation, embryonic development and physiologic homeostasis.5–7 A number of experimental studies suggested that RA is critical in the adult brain.8,9 Data from animal studies showed that RA supplementation could protect neurons against ischemic damage and help to decrease infarction volume after focal brain ischemia.10–12 RA was also found to be able to ameliorate blood-brain barrier disruption following ischemic stroke in rats.13 In clinical study, low serum RA was proved to be independently associated with post-stroke cognitive impairment and affective disorder.14,15 In another prospective study focusing on acute ischemic stroke patients, decreased circulating levels of RA were reported to be correlated with increased risk of all-cause mortality or cardiovascular disease mortality in the 6 months after symptoms onset.16 All these studies indicated that RA might be a prognostic biomarker for ischemic stroke. However, the clinical relevance of circulating RA in the short-term stroke outcomes has not been clarified. The aim of this study was to investigate whether serum RA levels were associated with 3-month functional outcome and early neurological deterioration (END) in patients with ischemic stroke.

Methods

Study Design and Patients

From Jan 2019 to Sep 2019, patients, who were admitted to Suzhou Ninth People’s Hospital, with first-ever acute ischemic stroke were screened for inclusion in our study. Patients were recruited if they met the following criteria: 1) aged 18 years or old; 2) time lapse between the symptom onset and hospitalization < 24 hrs; 3) had pre-onset modified Rankin Scale (mRS) score ≤ 2. Patients with intravenous thrombolysis and endovascular therapy, tumor, severe renal disease and hepatic disease and discharge within 7 days were all excluded from this study. The local institutional review boards of Suzhou Ninth People’s Hospital approved that all study procedures were performed in accordance with relevant guidelines and regulations (NO. SZJY-20181207). All patients or their relatives provided written informed consent and agreed to participate in the study.

Data Collection

After admission, demographic characteristic, clinical data and medical history were collected in all patients. Stroke severity was assessed using National Institutes of Health Stroke Scale17 by trained neurologists. Stroke etiology was classified according to the Trial of the ORG 10172 in Acute Stroke Treatment (TOAST) criteria,18 which embraces large artery atherosclerosis, cardioembolism, small vessel occlusion, stroke of other determined etiology and cryptogenic stroke. Laboratory data including levels of lipid profile, blood glucose, hypersensitive C-reactive protein (Hs-CRP) and homocysteine were also recorded.

Serum RA Measurement

After informed consents’ blood samples were collected from all patients on the second day morning. The specimens were centrifuged at 1500 g for 10 min and the isolated plasma frozen at −80°C for further analysis. Serum RA concentrations were measured using a commercially available ELISA kit (Cat. MBS705877, MyBioSource, San Diego, CA). The detection range of RA is 0.625–10 ng/mL. The intra-assay and interassay coefficient of variation were < 8.0%. The measurement of serum RA was performed by laboratory technicians who were blind to any clinical information of the study participants.

Assessment of Outcomes

The primary outcome was assessed as functional outcome at 3 months by trained neurologists who were blinded to clinical data using modified Rankin Scale (mRS). According to previous studies, poor functional outcome was defined as modified Rankin Scale score of 3–6.19,20 The secondary end point in stroke patients was END. In our study, END was defined as an increase of ≥ 1 point in motor power or ≥ 2 points in total NIHSS score within 7 days after admission.21,22

Statistical Analysis

Categorical variables were expressed as n (%) and continuous variables as means (standard deviation) or medians (interquartile range). Differences in baseline characteristics between groups were analyzed using chi-square test or Fisher’s exact test, analysis of variance, or Kruskal–Wallis test where appropriate. Logistic regression analysis was used to estimate the risk of stroke outcomes by calculating odds ratios (OR) and 95% confidence interval (CI). Variables with P value < 0.1 in univariate analysis were adjusted in multiple regression analysis. In addition, net reclassification index (NRI) and integrated discrimination improvement (IDI) were calculated to evaluate the predictive value of adding RA to conventional risk factors model. Receiver operating characteristic (ROC) curves were used to describe RA levels as a potential predictive factor for stroke outcomes. The area under curve (AUC) was calculated based on the ROC curves. Two-tailed P < 0.05 was considered to be statistically significant. All analyses were performed using SPSS software, version 22.0 (IBM, New York, NY) and R statistical software version 3.6.2.

Results

A total of 217 participants (mean age 66.5 ± 9.4 years, 52.5% male) were included in this prospective study. The median RA concentration was 2.9 ng/mL (interquartile range, 2.0–5.2 ng/mL) with quartile levels as follows: < 2.0 ng/mL (1st quartile), 2.0−2.9 ng/mL (2nd quartile), 3.0−5.2 ng/mL (3rd quartile), and > 5.2 ng/mL (4th quartile).

After follow-up of 3 months, a total of 94 patients (43.3%) experienced poor outcome (Table 1). Compared with participants who did not develop poor outcome, those who developed were more likely to have higher systolic blood pressure, NIHSS score, Hs-CRP, and prevalence of diabetes mellitus and lower RA levels (Table 1). Sixty-five (30.0%) patients were diagnosed as END during hospitalization. Table 2 shows the baseline data between the subgroup according to the presence or absence of END. Compared with patients without END, those with it were older, more likely to have higher levels of Hs-CRP, fasting blood-glucose, homocysteine, and lower levels of RA and more prevalence of diabetes mellitus.

Table 1

Characteristics of Ischemic Stroke Patients with and Without Poor Outcome at 3-Month

Variables	Poor Outcome (n = 94)	Good Outcome (n = 123)	P value
Demographic characteristics
Age, year	66.9 ± 9.3	66.0 ± 9.6	0.571
Male, n (%)	49 (52.1)	65 (52.8)	0.916
Risk factors, n (%)
Hypertension	65 (69.1)	90 (73.2)	0.516
Diabetes mellitus	35 (37.2)	32 (26.0)	0.076
Hyperlipidemia	15 (16.0)	24 (19.5)	0.499
Coronary heart disease	7 (7.4)	17 (13.8)	0.138
Current smoking	31 (33.0)	38 (30.9)	0.744
Clinical data
Systolic blood pressure, mmHg	139.4 ± 16.7	133.5 ± 18.4	0.017
Diastolic blood pressure, mmHg	84.4 ± 12.4	85.8 ± 10.8	0.3847
Body mass index, kg/m2	24.5 ± 1.9	24.7 ± 1.6	0.461
NIHSS, score	7.0 (3.0, 9.0)	6.0 (3.0, 8.0)	0.045
Stroke etiology, n (%)			0.161
Large vessel disease	49 (52.1)	53 (43.1)
Cardioembolic	11 (11.7)	9 (7.3)
Small vessel disease	26 (27.7)	40 (32.5)
Stroke with determined etiology	4 (4.3)	6 (4.9)
Cryptogenic stroke	4 (4.3)	15 (12.2)
Stroke location, n (%)			0.239
Lobar	27 (28.7)	22 (17.9)
Basal ganglia	26 (27.7)	37 (30.1)
Infratentorial	10 (10.6)	20 (16.3)
Others	31 (33.6)	44 (35.8)
Laboratory data
Total cholesterol, mmol/L	4.6 ± 1.1	4.6 ± 1.2	0.472
Triglyceride, mmol/L	1.6 ± 0.9	1.5 ± 0.8	0.588
Low density lipoprotein, mmol/L	2.4 ± 0.9	2.3 ± 0.8	0.698
High density lipoprotein, mmol/L	1.4 ± 0.3	1.4 ± 0.4	0.951
Hs-CRP, mg/L	4.1 (1.0, 9.5)	3.2 (1.0, 6.2)	0.008
Fasting blood-glucose, mmol/L	6.5 ± 2.3	6.6 ± 2.2	0.791
Homocysteine, mmol/L	15.4 ± 8.2	15.6 ± 6.8	0.784
RA levels, ng/mL	2.5 (1.6, 4.5)	3.6 (2.3, 5.8)	0.002
RA quartile			0.015
1st	31 (33.0)	23 (18.7)
2nd	27 (28.7)	28 (22.8)
3rd	20 (21.3)	33 (26.8)
4th	16 (17.0)	39 (31.7)

Abbreviations: Hs-CRP, hyper-sensitive C-reactive protein; NIHSS, National Institutes of Health Stroke Scale; RA, retinoic acid.

Table 2

Characteristics of Ischemic Stroke Patients with and Without END

Variables	With END (n = 65)	Without END (n = 152)	P value
Demographic characteristics
Age, year	68.7 ± 9.1	65.6 ± 9.4	0.026
Male, n (%)	34 (52.3)	80 (52.6)	0.965
Risk factors, n (%)
Hypertension	48 (73.8)	107 (70.4)	0.606
Diabetes mellitus	31 (47.7)	36 (23.7)	0.001
Hyperlipidemia	14 (21.5)	25 (16.4)	0.371
Coronary heart disease	8 (12.3)	16 (10.5)	0.702
Current smoking	19 (29.2)	50 (32.9)	0.595
Clinical data
Systolic blood pressure, mmHg	138.3 ± 18.1	135.1 ± 17.8	0.220
Diastolic blood pressure, mmHg	84.9 ± 11.5	85.3 ± 11.6	0.837
Body mass index, kg/m2	24.7 ± 1.7	24.8 ± 1.8	0.754
NIHSS, score	6.0 (3.5, 8.0)	6.5 (2.0, 9.0)	0.176
Stroke etiology, n (%)			0.161
Large vessel disease	34 (52.3)	68 (44.7)
Cardioembolic	6 (9.2)	14 (9.2)
Small vessel disease	20 (30.8)	46 (30.3)
Stroke with determined etiology	4 (6.2)	6 (3.9)
Cryptogenic stroke	1 (1.5)	18 (11.8)
Stroke location, n (%)			0.155
Lobar	19 (29.2)	30 (19.7)
Basal ganglia	22 (33.8)	41 (27.0)
Infratentorial	6 (9.2)	24(15.8)
Others	18 (27.7)	57 (37.5)
Laboratory data
Total cholesterol, mmol/L	4.7 ± 1.2	4.6 ± 1.1	0.551
Triglyceride, mmol/L	1.5 ± 0.6	1.6 ± 0.9	0.511
Low density lipoprotein, mmol/L	2.4 ± 0.8	2.3 ± 0.9	0.703
High density lipoprotein, mmol/L	1.4 ± 0.3	1.3 ± 0.3	0.621
Hs-CRP, mg/L	5.0 (2.2, 9.2)	3.0 (1.0, 6.0)	0.001
Fasting blood-glucose, mmol/L	7.0 ± 2.2	6.3 ± 2.3	0.027
Homocysteine, mmol/L	17.1 ± 9.4	14.9 ± 6.4	0.045
RA levels, ng/mL	2.6 (1.7, 3.9)	3.4 (2.1, 5.6)	0.018
RA quartile			0.017
1st	23 (35.4)	31 (20.4)
2nd	20 (30.8)	35 (23.0)
3rd	12 (18.5)	41 (27.0)
4th	10 (15.4)	45 (29.6)

Abbreviations: END, early neurological deterioration; Hs-CRP, hyper-sensitive C-reactive protein; NIHSS, National Institutes of Health Stroke Scale; RA, retinoic acid.

Table 3 summarizes the results of the binary logistic regression of the clinical outcomes. The 1st quartile of RA levels (using the 4th quartile as the reference value) was identified as the predictor of poor outcome [odds ratio (OR), 4.485; 95% CI 1.890–9.639; P for trend = 0.001] after adjusted for systolic blood pressure, NIHSS score, Hs-CRP and diabetes mellitus. Moreover, decreased serum RA was associated with a higher risk of END (P for trend = 0.002) after adjusted for age, Hs-CRP, fasting blood-glucose, Hs-CRP, homocysteine and diabetes mellitus. This association remained statistically significant when RA levels were analyzed as continuous variables.

Table 3

Logistics Regression Analysis for the Association Between RA Levels and Clinical Outcomes Among Ischemic Stroke Patients

	OR (95% CI) for 3-Month Unfavorable Outcome	P value	OR (95% CI) for END	P value
Unadjusted model
RA levels	0.823 (0.718–0.942)	0.005	0.821 (0.706–0.954)	0.011
RA quartile
1st	3.285 (1.486–7.264)		3.339 (1.396–7.985)
2nd	2.350 (1.071–5.159)		2.571 (1.068–6.189)
3rd	1.477 (0.661–3.302)		1.317 (0.515–3.371)
4th	Reference		Reference
P for trend	0.018		0.021
Adjusted model
RA levels	0.763 (0.655–0.888)	0.001	0.785 (0.665–0.927)	0.004
RA quartile
1st	4.485 (1.890–9.639)		3.995 (1.517–9.821)
2nd	3.111 (1.318–7.334)		3.552 (1.319–9.562)
3rd	1.572 (0.657–3.762)		1.622 (0.575–4.574)
4th	Reference		Reference
P for trend	0.001		0.002

Abbreviations: CI, confidence interval; END, early neurological deterioration; OR, odds ratio; RA, retinoic acid.

We further examined whether adding serum RA to the conventional risk factors improved the risk prediction of clinical outcomes after acute ischemic stroke. As shown in Table 4, adding RA quartile to conventional risk factors significantly improved predictive power for poor outcome [NRI = 42.6%, P = 0.001; IDI = 5.7%, P = 0.001] and END (NRI = 45.4%, P = 0.001; IDI = 4.3%; P = 0.005). Similarly, significant findings were observed when RA levels were analyzed as continuous variables. To detect the possible predictive value of RA for stroke outcomes, area under curve (AUC) was used to compare the prediction accuracy between the model with and without RA. The AUC of poor outcome (from 0.661 to 0.713) and END (from 0.744 to 0.776) were increased when RA was put into the model (Figure 1).

Table 4

Reclassification and Discrimination Statistics for Clinical Outcomes by RA Levels Among Ischemic Stroke Patients

Clinical Outcomes	Model	Continuous NRI		IDI
		Estimate (95% CI)	P value	Estimate (95% CI)	P value
3-month poor outcome	Conventional modela
	Conventional model + RA (continuous)	0.397 (0.138–0.657)	0.003	0.052 (0.023–0.082)	0.001
	Conventional model + RA (quartiles)	0.426 (0.164–0.688)	0.001	0.057 (0.026–0.089)	0.001
END	Conventional modelb
	Conventional model + RA (continuous)	0.442 (0.169–0.714)	0.005	0.036 (0.011–0.063)	0.008
	Conventional model + RA (quartiles)	0.454 (0.176–0.734)	0.001	0.043 (0.013–0.072)	0.005

Notes: aConventional model included systolic blood pressure, NIHSS score, Hs-CRP, and diabetes mellitus. bConventional model included age, Hs-CRP, fasting blood-glucose, homocysteine, and diabetes mellitus.

Abbreviations: CI, confidence interval; END, early neurological deterioration; IDI, integrated discrimination index; NRI, net reclassification improvement; RA, retinoic acid.

Figure 1

ROC curves comparing the potential of different models to predict stroke outcomes (A,B).

Discussion

In the present study, we assessed the correlation between serum RA levels and short-term prognosis of ischemic stroke. We found that decreased RA levels at baseline were independently associated with higher risk of poor outcome developed at 3 months and END after ischemic stroke, even after adjustment for several potential confounders. Furthermore, adding RA to conventional risk factors could improve risk prediction for the stroke outcomes. These findings indicated that serum RA might be a potential biomarker in the prediction of clinical outcomes among acute ischemic stroke patients. Further studies from various populations are needed to replicate our findings.

RA is a major metabolic product from vitamin A, which can only be obtained from diet including carotenoids and retinyl esters.23 Cumulative epidemiological evidences indicate that specific circulating vitamins had numerous health benefits as well as protective effects on the progression of coronary artery disease.24,25 As the substrate for the active all-trans RA, retinol was confirmed to be able to predict coronary events in healthy middle-aged men in the PRIME study.26 Several population-based studies have investigated the association between serum RA levels and clinical outcomes in cardiovascular disease.16,27 After measurement of serum RA levels in 1499 patients with angiographically confirmed coronary artery disease in the Guangdong Coronary Artery Disease Cohort, Liu et al27 found that high RA levels (defined as > median) were associated with a lower risk of total mortality (adjusted hazard ratios, 0.68; 95% CI, 0.50–0.85; P = 0.001) and cardiovascular mortality (adjusted hazard ratios, 0.62; 95% CI, 0.45–0.78; P < 0.001). Furthermore, in a cohort of patients with first-ever acute ischemic stroke, RA levels in 2nd quartile, 3rd quartile, and 4th quartile were correlated with all-cause mortality, and the mortality was decreased by 31% (OR 0.69; 95% CI, 0.55–0.89), 52% (OR 0.48; 95% CI, 0.34–0.60), and 82% (OR 0.18; 95% CI, 0.11–0.29) respectively.16 Our study extended the current knowledge about the role of serum RA in acute ischemic stroke as it shows a strongly prognostic value in the presence of 3-month poor outcome and END. Previous studies also confirmed a negative association between RA level and diabetes mellitus.15 In our study, the level of RA was slightly lower in patients with diabetes mellitus than those without it (median level, 2.7 ng/mL vs 3.2 ng/mL). However, the difference did not reach statistical significance (P = 0.258), probably due to the difference in study population, and sample size. Further studies with large sample size might be useful to clarify this association.

The mechanisms underlying the detrimental effects of decreased RA on the stroke prognosis are still poorly clear, but several potential pathophysiological pathways may be involved. Acute ischemic stroke can trigger a widespread inflammatory response to induce neuronal damages, characterized by increasing pro-inflammatory cytokines and disrupting blood-brain barrier.28 Experimental study indicated that RA can suppress inflammatory reactions through mediating the expression of pro-inflammatory mediators (including IL-1β, IL-6, TNF-α and SAA) and increasing CD14 and MHC II percent positive cells.29 RA supplementation was also found to substantially ameliorate blood-brain barrier disruption following ischemic stroke in rats.13 Besides, increased oxidative stress plays an essential role in brain tissue injury after stroke.30 Previous evidence suggested that RA administration at a physiological concentration significantly decreased hyperglycemia-induced oxidative stress and thus supported the antioxidant defense system.31 Taken together, we speculated that the anti-inflammatory and anti-oxidative stress characteristics of RA may exert protection on the stroke brain. The endogenous expression of RA can be induced after brain tissue injury, which could create a dynamic signaling environment for neurons and glia, and therefore improve functional recovery.32 Interestingly, preclinical animal models have also confirmed the neuroprotective therapeutic role of the application of RA receptor agonist in intracerebral hemorrhage.33 Taken together, whether the management of serum RA levels within an appropriate range could improve stroke outcomes is a possible future area of inquiry.

Our study has several limitations that need to be noted. First, this study was a single-center observational study. Therefore, there may exist a selection bias and limit the generalizability of the findings to other ethnicities. Second, patients with intravenous thrombolysis, endovascular therapy, tumor, severe renal disease and hepatic disease and early discharge were excluded, which might induce an underestimation of the actual incidence of END and poor outcome. Third, circulating RA concentrations were measured only once at admission, so we were unable to investigate the association between changes in serum RA and prognosis of ischemic stroke. Finally, RA concentrations and activity may be affected by many other factors, including retinol dehydrogenase 10, alcohol dehydrogenase, retinaldehyde dehydrogenase, and cytochrome P450 and thyroid hormone receptor. Thus, the effects of these confounding factors cannot be excluded in our study.

Conclusions

In conclusion, our data demonstrated that low serum RA levels at acute phase were associated with short-term prognosis among ischemic stroke patients, indicating that RA may be a potential prognostic biomarker for ischemic stroke. We suggest that further prospective studies from other populations of ischemic stroke are needed to validate our findings. Potential pathophysiological mechanisms and therapeutic considerations also remain to be determined.