Baseline characteristics
A total of 238 PCIS patients were enrolled in the current study. The median age was 64.3 ± 11.6 years and 171 (71.8%) patients were male. The median NIHSS score and pc-ASPECTS were 3 (2–6) and 8 (8–9). The median PLR, NLR, and HALP were 123.4 (102.1–162), 2.5 (1.8–3.3), and 43.1 (31.2–54.9). The median follow-up time was 4.3 (3.7–4.5) years and 44 (18.5%) patients died during this period. 115 (48.3%) patients with basilar artery (BA) stenosis, 54 (22.7%) with posterior cerebral artery (PCA) stenosis, 37 (15.5%) with vertebral artery (VA) stenosis, and 43 (13.4%) with multiple vascular stenosis (Table 1).
Table 1
Variables
|
|
Characteristics
|
|
Age
|
64.3 ± 11.6
|
Male
|
171 (71.8%)
|
NIHSS
|
3 (2–6)
|
pc-ASPECTS
|
8 (8–9)
|
PLR
|
123.4 (102.1–162)
|
NLR
|
2.5 (1.8–3.3)
|
HALP
|
43.1 (31.2–54.9)
|
Recurrence of stroke
|
70 (29.4%)
|
Follow-up time, year
|
4.3 (3.7–4.5)
|
Mortality
|
44 (18.5%)
|
Risk factors, n (%)
|
|
Hypertension
|
173 (72.7%)
|
Diabetes mellitus
|
101 (42.4%)
|
Coronary heart disease
|
36 (15.1%)
|
Atrial fibrillation
|
15 (6.3%)
|
Smoking
|
139 (58.4%)
|
Alcohol-drinking
|
80 (33.6%)
|
Peripheral artery disease
|
20 (8.4%)
|
Pathogenesis, n (%)
|
|
Large vessel atherosclerosis
|
142 (59.7%)
|
Cardioembolic
|
37 (15.5%)
|
Small artery disease
|
31 (13%)
|
Other
|
13 (5.5%)
|
Undetermined
|
15 (6.3%)
|
Location of Vascular stenosis, n (%)
|
|
VA
|
37 (15.5%)
|
BA
|
115 (48.3%)
|
PCA
|
54 (22.7%)
|
Multiple vascular stenosis
|
43 (13.4%)
|
Data were presented as mean ± standard deviation, median (interquartile range), or n (%). NIHSS, the National Institute of Health Scale Score; pc-ASPECTS, posterior circulation Alberta Stroke Program Early Computed Tomography Score; PLR, platelet-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; HALP, hemoglobin, albumin, lymphocyte, and platelet score VA, vertebral artery; BA, basilar artery; PCA, posterior cerebral artery.
Association of HALP score with clinical characteristics
We performed ROC analysis and found that the area under the curve (AUC) of HALP score was 0.76, indicating that it was significant for predicting 5-year overall survival (p < 0.001, Fig. 1a). The optimal cutoff value was 42.89. Subsequently, PCIS patients were divided into low-HALP (n = 118, 49.6%) and high-HALP (n = 120, 50.4%) groups (Table 2). Patients with low levels of HALP tended to have a higher NIHSS, a lower pc-ASPECES and a higher recurrence rate (all p < 0.05). A low level of HALP was more likely to be present in higher age and in female patients (all p < 0.001). Low levels of HALP were significantly associated with high NLR and PLR levels (all p < 0.001).
Table 2
Association of HALP score with baseline characteristics.
Variables
|
Low HALP
(n = 118)
|
High HALP
(n = 120)
|
p value
|
Characteristics
|
|
|
|
Age
|
67.1 ± 11.3
|
61.6 ± 11.4
|
0.000
|
Male
|
72 (61%)
|
99 (82.5%)
|
0.000
|
NIHSS
|
4 (2–7)
|
3 (1–4)
|
0.001
|
pc-ASPECTS
|
8 (8–9)
|
9 (8–9)
|
0.008
|
PLR
|
161.6 (137.7-200.2)
|
102.9 (87.4–117)
|
0.000
|
NLR
|
3 (2.2–4.4)
|
2.1 (1.7–2.7)
|
0.000
|
Recurrence of stroke
|
42 (35.6%)
|
28 (23.3%)
|
0.038
|
Risk factors
|
|
|
|
Hypertension
|
94 (79.7%)
|
79 (65.8%)
|
0.017
|
Diabetes mellitus
|
57 (48.3%)
|
44 (36.7%)
|
0.069
|
Coronary heart disease
|
20 (16.9%)
|
16 (13.3%)
|
0.436
|
Atrial fibrillation
|
4 (3.4%)
|
11 (9.2%)
|
0.067
|
Smoking
|
59 (50%)
|
80 (66.7%)
|
0.009
|
Alcohol-drinking
|
32 (27.1%)
|
48 (40%)
|
0.035
|
Peripheral artery disease
|
11 (9.3%)
|
9 (7.5%)
|
0.612
|
Pathogenesis
|
|
|
0.246
|
Large vessel atherosclerosis
|
72 (61%)
|
70 (58.3%)
|
|
Cardioembolic
|
20 (16.9%)
|
17 (14.2%)
|
|
Small artery disease
|
12 (10.2%)
|
19 (15.8%)
|
|
Other
|
9 (7.6%)
|
4 (3.3%)
|
|
Undetermined
|
5 (4.2)
|
10 (8.3%)
|
|
Location of Vascular stenosis
|
|
|
0.077
|
VA
|
19 (16.1%)
|
18 (15%)
|
|
BA
|
65 (55.1%)
|
50 (41.7%)
|
|
PCA
|
19 (16.1%)
|
35 (29.2%)
|
|
Multiple vascular stenosis
|
15 (12.7%)
|
17 (14.2%)
|
|
HALP, hemoglobin, albumin, lymphocyte, and platelet score; NIHSS, the National Institute of Health Scale Score; pc-ASPECTS, posterior circulation Alberta Stroke Program Early Computed Tomography Score; PLR, platelet-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; VA, vertebral artery; BA, basilar artery; PCA, posterior cerebral artery.
Association of HALP score with overall survival
we performed univariate Cox proportional hazard regression model and observed a significant association with age at diagnosis, NIHSS, pc-ASPECTS, PLR, NLR, and HALP score (all p < 0.01). Other factors included male, recurrence of stroke, history of alcohol drinking, and the location of vascular stenosis (all p < 0.05). Furthermore, multivariate analysis indicated Low HALP score (HR 0.354, 95%CI 0.146–0.86, p = 0.022) as an independent predictor of PCIS, along with age and NIHSS (HR 1.059, 95%CI 1.021–1.099, p = 0.002 and HR 1.26, 95%CI 1.148–1.383, p < 0.001) (Table 3). Kaplan-Meier analysis also showed that low HALP score predicted a worse overall survival (p < 0.001, Fig. 1b).
Table 3
Univariate and multivariate analyses for patients' overall survival with PCIS.
Variables
|
Univariate analysisa
|
Multivariate analysisb
|
HR (95% CI)
|
p value
|
HR (95% CI)
|
p value
|
Characteristics
|
|
|
|
|
Age
|
1.103 (1.067–1.139)
|
0.000
|
1.059 (1.021–1.099)
|
0.002
|
Male
|
0.518 (0.284–0.944)
|
0.032
|
|
|
NIHSS
|
1.297 (1.222–1.377)
|
0.000
|
1.26 (1.148–1.383)
|
0.000
|
pc-ASPECTS
|
0.603 (0.463–0.784)
|
0.000
|
|
|
PLR
|
1.005 (1.003–1.008)
|
0.000
|
|
|
NLR
|
1.223 (1.135–1.319)
|
0.000
|
|
|
HALP (≤ 42.89 VS > 42.89)
|
0.161 (0.072–0.361)
|
0.000
|
0.354 (0.146–0.86)
|
0.022
|
Recurrence of stroke
|
1.983 (1.092–3.602)
|
0.025
|
|
|
Risk factors
|
|
|
|
|
Hypertension
|
1.786 (0.83–3.843)
|
0.138
|
|
|
Diabetes mellitus
|
1.42 (0.787–2.565)
|
0.244
|
|
|
Coronary heart disease
|
1.114 (0.497–2.499)
|
0.793
|
|
|
Atrial fibrillation
|
0.689 (0.167–2.845)
|
0.606
|
|
|
Smoking
|
0.769 (0.426–1.39)
|
0.385
|
|
|
Alcohol-drinking
|
1.105 (1.023–1.174)
|
0.009
|
|
|
Peripheral artery disease
|
1.121 (0.401–3.133)
|
0.828
|
|
|
Pathogenesis
|
|
0.293
|
|
|
Large vessel atherosclerosis
|
1
|
|
|
|
Cardioembolic
|
1.767 (0.841–3.714)
|
|
|
|
Small artery disease
|
0.768 (0.266–2.222)
|
|
|
|
Other
|
2.389 (0.826–6.908)
|
|
|
|
Undetermined
|
1.264 (0.379–4.21)
|
|
|
|
Location of Vascular stenosis
|
|
0.012
|
|
|
VA
|
1
|
|
|
|
BA
|
2.546 (0.895–7.244)
|
|
|
|
PCA
|
0.333 (0.061–1.818)
|
|
|
|
Multiple vascular stenosis
|
2.909 (0.896–9.446)
|
|
|
|
HR, hazard ratio; CI, confidence interval; NIHSS, the National Institute of Health Scale Score; pc-ASPECTS, posterior circulation Alberta Stroke Program Early Computed Tomography Score; PLR, platelet-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio; HALP, hemoglobin, albumin, lymphocyte, and platelet score; VA, vertebral artery; BA, basilar artery; PCA, posterior cerebral artery. a. All univariate analysis as a time-dependent variable in the model. b. Variables control in multivariable analysis including age, sex, NIHSS, pc-ASPECTS, PLR, NLR, HALP, recurrence of stroke, alcohol-drinking, and the location of vascular stenosis.
Creation of the HALPN value as a new prognostic model index
According to the multivariate Cox regression analysis, NIHSS was identified as an important predictor, in addition to the HALP score. The AUC of NIHSS was 0.825, and the optimal cutoff value was 4.5 (Fig. 2a, p < 0.001). Kaplan-Meier analysis showed that the high NIHSS score was associated with increased mortality ((Fig. 2b, p < 0.001). Thus, we combined NIHSS and the HALP score to construct a new index, HALPN. The HALPN score is defined as follows: HALPN = 0 (HALP > 42.89 and NIHSS < 5), HALPN = 1 (HALP ≤ 42.89 or NIHSS ≥ 5), HALPN = 2 (HALP ≤ 42.89 and NIHSS ≥ 5).
Kaplan-Meier analysis showed that a higher HALPN score predicted poor overall survival (Fig. 3a, p < 0.001) in PCIS patients. As age at diagnosis was an independent factor for overall survival, we performed further stratification analysis for the PCIS patients according to age. It showed that patients with a higher HALPN score had a worse overall survival with age ≤ 60 and age > 60 (Fig. 3b and 3c, all p < 0.01)
Prediction power of the combination of HALP and NIHSS
The efficacy of the combination of HALP and NIHSS score was assessed by the AUC of ROC curve. (Table 4). The AUC of the combination was 0.888 (95% CI 0.84–0.937). The prediction power of the HALPN score was significantly better than either score [HALP (AUC = 0.76, 95%CI 0.686–0.834, p < 0.001) or NIHSS (AUC = 0.825, 95% CI 0.751-0.9, p < 0.001), respectively] (Fig. 4). Moreover, the HALPN score have higher prediction accuracy than other related indices such as pc-ASPECT, PLR, NLR and Age.
Table 4
AUC analysis of various indices in PCIS patients.
|
AUC
|
95% CI
|
p value
|
HALPN
|
0.888
|
0.84–0.937
|
0.000
|
HALP
|
0.76
|
0.686–0.834
|
0.000
|
NIHSS
|
0.825
|
0.751-0.9
|
0.000
|
pc-ASPECT
|
0.623
|
0.523–0.722
|
0.011
|
PLR
|
0.691
|
0.606–0.775
|
0.000
|
NLR
|
0.724
|
0.645–0.802
|
0.000
|
Age
|
0.785
|
0.707–0.864
|
0.000
|
AUC, area under receiver operating characteristic curve; CI, confidence interval; HALPN, combination of the HALP score and the NIHSS score; HALP, hemoglobin, albumin, lymphocyte, and platelet score; NIHSS, the National Institute of Health Scale Score; pc-ASPECTS, posterior circulation Alberta Stroke Program Early Computed Tomography Score; PLR, platelet-to-lymphocyte ratio; NLR, neutrophil-to-lymphocyte ratio.