Effect of diabetes on the efficacy and safety of Panax notoginseng saponins in ischemic stroke: a prespecified analysis of the PANDA trial

Abstract

Background

Diabetes is a risk factor of poor stroke outcomes. A randomized clinical trial has demonstrated that Panax notoginseng saponins (Xuesaitong soft capsules) can improve the functional outcome in ischemic stroke patients. However, it remains uncertain whether comorbid diabetes exerts an influence on the therapeutic outcomes of Xuesaitong.

Methods

The PANDA (efficacy and safety of Panax notoginseng saponins in the treatment of adults with ischemic stroke in China) trial was a multicenter, randomized, double-blind, placebo-controlled trial comprising 3542 patients. The present prespecified analyses investigated the effect of concomitant use of antidiabetic drugs, a history of diabetes, and baseline fasting blood glucose levels on outcomes in participants randomized to receive Xuesaitong soft capsules versus placebo. The primary outcome was the proportion of patients achieving functional independence.

Results

In the modified intention-to-treat dataset, there were 722 patients with diabetes or treated hyperglycemia during hospitalization (369 in the Xuesaitong group and 353 in the placebo group) and 2244 patients without (1118 in the Xuesaitong group and 1126 in the placebo group). Among patients with diabetes or treated hyperglycemia during hospitalization, the proportion of patients with functional independence at 3 months was 83.20 % (n = 307) in the Xuesaitong group and 79.60 % (n = 281) in the placebo group (odds ratio, 1.27; 95 % confidence intervals [CI], 0.87–1.85; P = 0.215). Regarding patients without diabetes or treated hyperglycemia during hospitalization, the proportion was 91.32 % (n = 1021) in the Xuesaitong group and 83.21 % (n = 937) in the control group (OR, 2.12; 95 % CI, 1.64–2.76; P < 0.001; P for interaction = 0.027). The sensitivity analyses indicated similar significant results.

Conclusion

In the PANDA trial, patients without, rather than with, diabetes or treated hyperglycemia during hospitalization received greater benefit from Xuesaitong soft capsules regarding functional independence at 3 months.

Graphical abstract

1. Introduction

Ischemic stroke poses a significant and daunting threat to human health [1,2]. As neuroprotective agents, Panax notoginseng saponins (PNS) are a class of dammarane tetracyclic triterpene compounds, and its preparations, Xuesaitong soft capsules were licensed for the treatment of ischemic stroke [[3], [4], [5]]. The PANDA (efficacy and safety of Panax notoginseng saponins in the treatment of adults with ischemic stroke in China) study, a multicenter, double-blind, placebo - controlled randomized clinical trial, has demonstrated that PNS (Xuesaitong soft capsules) can improve the rate of 3-month functional independence in ischemic stroke patients [5]. A pressing question arises: Are Xuesaitong soft capsules truly appropriate for consumption by every ischemic stroke patient who fall within the inclusion and exclusion criteria outlined in the PANDA trial? Under such circumstances, individualized therapeutic strategies derived from subgroup analysis offer substantial value and clinical relevance. Our prior prespecified analysis indicated that older individuals appeared to exhibit more pronounced clinical benefits when treated with Xuesaitong [6]. However, further factors influencing and modulating the therapeutic outcomes of Xuesaitong remain to be fully elucidated.

Individuals with diabetes have an increased risk of developing ischemic stroke [7,8]. Also, the comorbid diabetes significantly impacts the prognosis for patients with ischemic stroke [9]. Results from several clinical trials have demonstrated heterogeneity in outcomes between ischemic stroke patients with and without diabetes for some interventions [[10], [11], [12]]. Based on these results, the pathophysiological alterations of diabetes may impact the efficacy of Xuesaitong soft capsules, which typically acts through antioxidant, endothelial - protective, and anti - inflammatory pathways [[13], [14], [15]]. Therefore, tailored treatment approaches considering these diabetes - related changes are crucial for optimizing the drug's therapeutic outcomes in different patient populations [6]. However, it remains unclear whether the therapeutic outcomes of Xuesaitong are influenced by the comorbidity of diabetes. Furthermore, there is a lack of clarity regarding whether the treatment recommendations should differ for ischemic stroke patients with and without diabetes. We hypothesize that ischemic stroke patients without diabetes may experience greater therapeutic benefits from Xuesaitong treatment compared with those with diabetes. This discrepancy is primarily attributed to the potential interference of diabetes-associated pathophysiological processes with the pharmacological mechanisms of Xuesaitong [[16], [17], [18]].

To address the dearth of evidence and validate our hypothesis, we conducted a prespecified subgroup analysis of the PANDA trial to comprehensively evaluate the impact of diabetes on clinical outcomes in ischemic stroke patients treated with Xuesaitong. By examining this interaction, we aim to provide evidence-based insights for optimizing therapeutic strategies and informing future research in diabetic versus non-diabetic stroke populations.

2. Methods

2.1. Study design and participants

The PANDA trial was a multicenter, randomized, double-blind, placebo-controlled trial comprising 3542 patients across 67 medical centers in China between July 1, 2018 and June 30, 2020 (ChiCTR1800016363). The primary aim of the trial was to assess the efficacy and safety of Xuesaitong soft capsules in patients with ischemic stroke. Eligible patients were aged 18–75 years, and had a diagnosis of ischemic stroke within 14 days after symptom onset, a pre-stroke score of 0 or 1 on the modified Rankin scale (mRS) and a National Institutes of Health Stroke Scale (NIHSS) score between 4 and 15. The primary outcome was functional independence at 3 months, defined as a mRS score of 0–2. The protocol for the trial was approved by the institutional review board and ethics committee at each participating center. All participants provided written informed consent before enrollment. Study design, participant baseline characteristics and overall study outcomes have been published previously [5]. This article conformed to the Consolidated Standards of Reporting Trials (CONSORT) guidelines [19].

The current study was prespecified in the protocol of the PANDA trial. We investigated the associations between the concomitant use of antidiabetic drugs during index hospitalization, a history of diabetes, and baseline fasting blood glucose levels on primary and secondary outcomes in participants randomized to receive Xuesaitong soft capsules versus placebo. The administration of antidiabetic medications to patients during their index hospitalization was based on clinical evaluation, diagnosis, and subsequent prescription. Consequently, this practice served as an indicator of the comorbid diabetes or treated hyperglycemia during hospitalization. The diagnosis of diabetes was established in accordance with the 2018 diagnostic criteria published by the American Diabetes Association (ADA) [20]. The threshold of 7.0 mmol/L for baseline fasting blood glucose levels was established based on guideline recommendations and clinical considerations [21]. However, it should be noted that this classification merely served as a reference and was not sufficient to support a diagnosis of diabetes. Given that this study constitutes a subgroup analysis of the PANDA trial and adheres to its protocol, the findings should be interpreted as exploratory in nature.

2.2. Randomization

For patients who met the criteria for enrollment, each participating center allocated randomization codes in a strictly sequential manner. These randomization codes were produced via an automated, centralized platform, eliminating manual intervention in the allocation process. Eligible patients were randomly assigned (1:1) to either treatment with Xuesaitong soft capsules or placebo, with both groups receiving 240 mg daily (two 60-mg capsules twice daily) for 3 months. A total of 3542 patients were screened for eligibility, of whom 470 were excluded because they did not meet the inclusion criteria or met exclusion criteria, resulting in 3072 eligible patients for randomization.

2.3. Outcomes

In strict adherence to the RCT protocol, the outcomes assessed in the current prespecified analyses were fully aligned with those defined in the PANDA study. The primary efficacy endpoint was defined as the proportion of patients achieving functional independence, as indicated by a mRS score of ≤2 at 3 months after randomization. The mRS categorizes functional outcomes into seven grades (0–6): 0–1 indicate no significant disability; 2 corresponds to slight disability without assistance requirement; 3–5 denote moderate to severe disability; and 6 is assigned to deceased patients. Study outcomes were determined via standardized interviews administered by mRS-certified investigators who remained blinded to treatment allocations throughout the assessment period. The secondary efficacy outcomes comprised rate of stroke recurrence in 3 months and 12 months; proportion of patients attaining functional independence at 12 months; proportion of patients with an mRS score of 1 or lower at 3 months and 12 months; change in NIHSS score from baseline to 3 months; incidence rate of composite cerebrovascular events over 3-month and 12-month period; score on EuroQoL Group 5-Dimension (EQ-5D) at 3 months and 12 months; change in Barthel Index from baseline to 3 months and 12 months; and platelet counts and coagulation indicators measured at the 3-month follow-up. The primary safety outcome was the occurrence of serious adverse events within a 3-month period.

2.4. Statistical analysis

Details for the overall statistical analysis plan were reported in the main manuscript of the PANDA trial. The prespecified analyses were conducted across three classification criteria, including concomitant use of antidiabetic drugs during the index hospitalization, a history of diabetes, and baseline fasting blood glucose levels. The time interval from randomization to the first recorded event served as the primary metric. The main analysis of the study utilized the modified intention-to-treat (mITT) dataset, comprising patients who underwent at least one documented treatment and efficacy evaluation following randomization. Also, the efficacy outcomes were analyzed in the per-protocol (PP) dataset, defined as patients who completed all scheduled treatment without major infractions. Regarding safety outcomes, every patient undergoing drug treatment was incorporated into the safety dataset.

We used descriptive statistics to summarize baseline characteristics according to use of antidiabetic drugs or not during the index hospitalization and randomized treatment assignments. Accordingly, baseline characteristics were presented as either the mean with standard deviations (SD) for normally distributed continuous variables or the median with interquartile range (IQR) for skewed continuous variables, while categorical variables were summarized by the number of patients with percentages. Between-group differences in binary data were compared using the χ² test or Fisher's exact test. Regarding continuous variables and outcomes across groups, the t-test or Wilcoxon rank-sum test was applied. During analysis with the mITT dataset, missing values of mRS score were estimated using the last observation carried forward method, and other outcomes were not estimated. For the outcomes defined based on a mRS score, the differences between two groups in the percentage of patients were analyzed by the logistic regression, with odds ratio (OR) and corresponding 95 % confidence intervals (CI) estimated. Regarding recurrent stroke and composite cerebrovascular events, we utilized the Cox proportional hazards models to estimate hazard ratio (HR) and 95 % CI. The interactions between subgroup factors and treatment assignments were calculated.

We conducted sensitivity analyses to ensure the robustness of our findings. First, we conducted subgroup analyses based on whether patients used hypoglycemic drugs within three months following randomization. Patients were classified into the antidiabetic drug-using group if they had taken such medications during their index hospitalization, the 1-month follow-up, or the 3-month follow-up. Second, in the primary analysis model, we adjusted for age, gender, EQ-5D score at baseline, and hypertension history. Prior studies have indicated that age and hypertension history served as significant factors contributing to the heterogeneity in primary outcomes. Furthermore, a notable imbalance was detected in the baseline characteristics, concerning gender and baseline EQ-5D score. Third, considering the constraints and inherent biases associated with the LOCF imputation methodology, we employed multiple imputation to address missing mRS scores [22]. The primary analysis employed the LOCF method, while the sensitivity analysis adopted a multiple imputation approach. All analyses were conducted using SAS statistical software, version 9.4 (SAS Institute), and statistical significance was defined as two tailed P value of <0.05.

3. Results

3.1. Baseline characteristics

In the PANDA trial, a total of 3072 patients were randomly allocated, with 1535 in the Xuesaitong group and 1537 in the placebo group. Among these patients, 2970 were incorporated into the safety dataset, 2966 into the mITT dataset, and 2177 into the PP dataset (Fig. 1). In the mITT dataset, there were 722 patients with diabetes or treated hyperglycemia during hospitalization (369 in the Xuesaitong group and 353 in the placebo group) and 2244 patients without (1118 in the Xuesaitong group and 1126 in the placebo group). The baseline characteristics in the Xuesaitong group and control group across diabetes status were well balanced except for gender in patients without diabetes or treated hyperglycemia during hospitalization and EQ-5D score in patients with diabetes or treated hyperglycemia during hospitalization (Table 1). The missing values in baseline patient characteristics in Xuesaitong and placebo groups stratified by the subgroups were detailed in Table S1.

Fig. 1.

Participant flow diagram mITT, modified intention-to-treat; PP, per-protocol.

Table 1.

Baseline patient characteristics in Xuesaitong and placebo groups stratified by diabetes or treated hyperglycemia during hospitalization subgroups.

Characteristic	With diabetes or treated hyperglycemia during hospitalization (n = 722)			Without diabetes or treated hyperglycemia during hospitalization (n = 2244)			P value
	Xuesaitong group (n = 369)	Placebo group (n = 353)	P value	Xuesaitong group (n = 1118)	Placebo group (n = 1126)	P value
Age, median (IQR), y	62 (56, 67)	62 (56, 67)	0.738	62 (54, 68)	62 (54, 68)	0.830	0.712
Gender, male, no. (%)	216 (58.54)	212 (60.06)	0.678	741 (66.28)	813 (72.20)	0.002	0.004
BMI, median (IQR)	24.8 (23.1, 26.9)	24.8 (22.9, 27.1)	0.651	24.4 (22.5, 26.4)	24.5 (22.5, 26.4)	0.886	0.875
Heart rate, median (IQR)	76 (70, 80)	76 (70, 82)	0.524	75 (69, 80)	75 (68, 80)	0.935	0.837
SBP, median (IQR), mm Hg	145 (133, 159)	144 (133, 160)	0.680	142 (132, 155)	140 (130, 156)	0.836	0.981
DBP, median (IQR), mm Hg	84 (78, 91)	84 (78, 92)	0.876	85 (77, 93)	85 (78, 94)	0.234	0.247
NIHSS score, median (IQR)	5 (4, 8)	5 (4, 7)	0.857	5 (4, 7)	5 (4, 7)	0.890	0.963
EQ-5D score, median (IQR)	70 (50, 80)	70 (60, 85)	0.023	80 (60, 90)	75 (60, 90)	0.575	0.471
Barthel Index, median (IQR)	70 (50, 90)	75 (50, 90)	0.087	75 (55, 90)	80 (55, 90)	0.752	0.236
Intravenous thrombolysis, no. (%)	32 (8.67)	27 (7.65)	0.616	108 (9.66)	96 (8.53)	0.350	0.498
Mechanical thrombectomy, no. (%)	3 (0.81)	2 (0.57)	1.000	9 (0.81)	4 (0.36)	0.177	0.783
Medical and personal history, no. (%)
Hypertension	249 (67.48)	230 (65.16)	0.509	586 (52.42)	591 (52.49)	0.973	0.725
Hyperlipidemia	34 (9.21)	32 (9.07)	0.945	39 (3.49)	47 (4.17)	0.398	0.594
Current smoking	77 (20.87)	85 (24.08)	0.301	279 (24.96)	300 (26.64)	0.361	0.189
Current drinking	50 (13.55)	36 (10.20)	0.165	141 (12.61)	145 (12.88)	0.850	0.618
Laboratory test results, median (IQR)
Total cholesterol, mg/dL	4.38 (3.67, 5.30)	4.40 (3.77, 5.29)	0.787	4.53 (3.80, 5.20)	4.48 (3.80, 5.18)	0.551	0.718
Fibrinogen, mg/dL	3.00 (2.54, 3.65)	3.04 (2.51, 3.59)	0.742	2.84 (2.37, 3.34)	2.85 (2.39, 3.32)	0.837	0.971
Platelet counts, × 103/μL	209 (167, 260)	210 (169.5, 250)	0.443	212 (175, 250)	212 (175, 251)	0.927	0.642
PT, s	11.10 (10.40, 12.30)	11.17 (10.30, 12.20)	0.759	11.30 (10.60, 12.30)	11.30 (10.60, 12.40)	0.642	0.733
APTT, s	27.50 (24.00, 31.80)	27.10 (23.95, 31.25)	0.374	28.25 (24.65, 32.70)	28.30 (24.80, 32.60)	0.613	0.942
TOAST classification, no. (%)			0.377			0.955	0.683
Large artery atherosclerosis	221 (59.89)	209 (59.21)		568 (50.81)	579 (51.42)
Cardioembolism	2 (0.54)	7 (1.98)		13 (1.16)	14 (1.24)
Small vessel occlusion	134 (36.31)	129 (36.54)		484 (43.29)	482 (42.81)
Other determined cause	5 (1.36)	2 (0.57)		18 (1.61)	14 (1.24)
Undetermined cause	7 (1.90)	6 (1.70)		35 (3.13)	37 (3.29)

APTT, activated partial thromboplastin time; BMI, body mass index (calculated as weight in kilograms divided by height in meters squared); DBP, diastolic blood pressure; EQ-5D, EuroQoL Group 5-Dimension; IQR, interquartile range; NIHSS, National Institutes of Health Stroke Scale; PT, prothrombin time; SBP, systolic blood pressure; TOAST, Trial of Org 10172 in Acute Stroke Treatment.

3.2. Efficacy outcomes

Among patients with diabetes or treated hyperglycemia during hospitalization in the mITT dataset, the proportion of patients achieving functional independence at 3 months was 83.20 % (n = 307) in the Xuesaitong group and 79.60 % (n = 281) in the control group (OR, 1.27; 95 % CI, 0.87–1.85; P = 0.215). Regarding patients without diabetes or treated hyperglycemia during hospitalization, the proportion was 91.32 % (n = 1021) in the Xuesaitong group and 83.21 % (n = 937) in the control group (OR, 2.12; 95 % CI, 1.64–2.76; P < 0.001; P for interaction = 0.027). The adjusted effect estimates (after accounting for covariates) were consistent with the crude estimates (Fig. 2, Table 2). The missing values in efficacy outcomes in the mITT dataset by diabetes subgroups were presented in Table S2. Regarding the PP dataset, in patients with diabetes or treated hyperglycemia during hospitalization, Xuesaitong therapy did not demonstrate a statistically significant improvement in 3-month functional independence (OR, 1.47; 95 % CI, 0.92–2.35; P = 0.107). Conversely, among patients without diabetes or treated hyperglycemia during hospitalization, there was a significant enhancement (OR, 1.97; 95 % CI, 1.46–2.68; P < 0.001; P for interaction = 0.302) (Fig. 4, Table S3). The analyses based on history of diabetes and baseline fasting blood glucose levels revealed a similar trend (Table S4∼S7). The sensitivity analyses using antidiabetic medications within 3 months, adjusting for covariates, and adopting multiple imputation indicated similar results (Fig. 3, Table S8∼S10).

Fig. 2.

Modified Rankin Scale score at 3 months.

Table 2.

Efficacy outcomes in the modified intention-to-treat dataset by diabetes or treated hyperglycemia during hospitalization subgroups.

Outcome	With diabetes or treated hyperglycemia during hospitalization (n = 722)				Without diabetes or treated hyperglycemia during hospitalization (n = 2244)				P value for interaction
	Xuesaitong group (n = 369)	Control group (n = 353)	Crude and adjusted OR or HR (95 % CI)	P value	Xuesaitong group (n = 1118)	Control group (n = 1126)	Crude and adjusted OR or HR (95 % CI)	P value
Primary outcomes
Functional independence at 3 months	307 (83.20)	281 (79.60)	1.27 (0.87, 1.85); 1.38 (0.94, 2.02)	0.215; 0.102	1021 (91.32)	937 (83.21)	2.12 (1.64, 2.76); 2.17 (1.67, 2.83)	<0.001; < 0.001	0.027; 0.047
Secondary outcomes
Recurrent stroke in 3 months	5 (1.36)	7 (1.98)	0.68 (0.22, 2.15); 0.62 (0.19, 1.96)	0.512; 0.413	9 (0.81)	10 (0.89)	0.91 (0.37, 2.23); 0.88 (0.36, 2.18)	0.830; 0.784	0.702; 0.743
Recurrent stroke in 12 months	14 (3.79)	15 (4.25)	0.89 (0.43, 1.84); 0.86 (0.41, 1.79)	0.751; 0.687	19 (1.70)	26 (2.31)	0.73 (0.41, 1.33); 0.73 (0.41, 1.33)	0.306; 0.307	0.690; 0.664
Functional independence at 12 months	315 (85.37)	311 (88.10)	0.79 (0.51, 1.21); 0.83 (0.53, 1.28)	0.280; 0.393	1051 (94.01)	1034 (91.83)	1.30 (1.01, 1.93); 1.44 (1.03, 2.00)	0.045; 0.033	0.039; 0.046
No or minimal disability at 3 months	242 (65.58)	227 (64.31)	1.06 (0.78, 1.44); 1.16 (0.84, 1.59)	0.719; 0.367	860 (76.92)	798 (70.87)	1.37 (1.13, 1.66); 1.39 (1.14, 1.69)	0.001; 0.001	0.159; 0.285
No or minimal disability at 12 months	286 (77.51)	274 (77.62)	0.99 (0.70, 1.41); 1.08 (0.75, 1.54)	0.971; 0.682	973 (87.03)	930 (82.59)	1.41 (1.12, 1.79); 1.44 (1.14, 1.83)	0.004; 0.003	0.100; 0.155
NIHSS score change from baseline to 3 months, median (IQR)	−4 (−5, −3)	−4 (−5, −3)	NA	0.140	−4 (−5, −3)	−4 (−5, −3)	NA	0.079	0.726
Composite cerebrovascular events in 3 months	5 (1.36)	8 (2.27)	0.60 (0.20, 1.82); 0.55 (0.18, 1.70)	0.364; 0.302	10 (0.89)	10 (0.89)	1.1 (0.42, 2.42); 0.97 (0.40, 2.34)	0.987; 0.944	0.469; 0.499
Composite cerebrovascular events in 12 months	17 (4.61)	18 (5.10)	0.90 (0.46, 1.74); 0.88 (0.45, 1.72)	0.752; 0.712	22 (1.97)	27 (2.40)	0.82 (0.47, 1.44); 0.82 (0.46, 1.43)	0.487; 0.478	0.834; 0.825
EQ-5D score, median (IQR) at 3 months	90 (80, 95)	90 (80, 95)	NA	0.643	90 (80, 95)	90 (80, 95)	NA	0.007	0.095
EQ-5D score, median (IQR) at 12 months	95 (85, 97)	94 (90, 95)	NA	0.614	95 (90, 98)	94 (90, 96)	NA	0.006	0.155
Barthel Index change from baseline to 3 months, median (IQR)	20 (5, 35)	15 (5, 35)	NA	0.021	15 (5, 35)	15 (5, 30)	NA	0.063	0.241
Barthel Index change from baseline to 12 months, median (IQR)	25 (10, 40)	20 (5, 40)	NA	0.024	20 (5, 40)	20 (5, 40)	NA	0.828	0.240
Platelet counts and coagulation indicators at 3 months, median (IQR)
Platelet counts, × 103/μL	226 (186, 264)	216 (179, 267.5)	NA	0.309	216 (177, 255)	216 (178, 256)	NA	0.768	0.369
PT, s	11.4 (10.7, 12.4)	11.1 (10.4, 12.2)	NA	0.034	11.4 (10.7, 12.4)	11.5 (10.7, 12.4)	NA	0.823	0.160
APTT, s	28.6 (24.6, 32.8)	27.6 (24.5, 32.7)	NA	0.337	28.6 (24.9, 33.0)	28.1 (25.0, 33.3)	NA	0.760	0.656

APTT, activated partial thromboplastin time; EQ-5D, EuroQoL Group 5-Dimension; IQR, interquartile range; NIHSS, National Institutes of Health Stroke Scale; PT, prothrombin time; Adjusted OR or HR (95 % CI) and P values were calculated after adjusting for age, gender, EQ - 5D score at baseline, and hypertension history.

Fig. 4.

Cumulative incidence of recurrent stroke and composite cerebrovascular events mITT, modified intention-to-treat; PP, per-protocol.

Fig. 3.

Primary efficacy outcomes in all analyses CI, confidence interval; mITT, modified intention-to-treat; OR, odds ratio; PP, per-protocol.

Regarding secondary efficacy outcomes in the mITT dataset, there was heterogeneity in the proportion of patients attaining functional independence at 12 months for the Xuesaitong group versus the control group in patients with and without diabetes. The proportion was 85.37 % (n = 315) in the Xuesaitong group and 88.10 % (n = 311) in the control group for patients with diabetes or treated hyperglycemia during hospitalization (OR, 0.79; 95 % CI, 0.51–1.21; P = 0.280), and 94.01 % (n = 1051) in the Xuesaitong group and 91.83 % (n = 1034) in the control group for patients without diabetes or treated hyperglycemia during hospitalization (OR, 1.30; 95 % CI, 1.01–1.93; P = 0.045; P for interaction = 0.039). The adjusted and crude estimates aligned (Table 2). In the PP dataset, however, Xuesaitong did not indicate statistically significant efficacy in improving functional independence at 12 months for both patients with diabetes or treated hyperglycemia during hospitalization (OR, 0.85; 95 % CI, 0.50–1.46; P = 0.561) and those without these conditions (OR, 1.27; 95 % CI, 0.86–1.86; P = 0.234; P for interaction = 0.243) (Table S3). The analyses based on history of diabetes and baseline fasting blood glucose levels as well as sensitivity analyses using antidiabetic medications within 3 months, adjusting for covariates, and adopting multiple imputation reported similar results (Table S4∼S10).

3.3. Safety outcomes

In the safety dataset, there was no significant difference between the Xuesaitong group and control group in all subgroups. When stratified by concomitant use of antidiabetic drugs during index hospitalization, the incidence of serious adverse events during the 3-month follow-up was 2.17 % in the Xuesaitong group and 1.97 % in the control group for patients with diabetes (P = 0.853), and 0.63 % in the Xuesaitong group and 0.80 % in the control group for patients without diabetes (P = 0.626). The corresponding incidence of adverse events were 4.61 %, 5.35 %, 2.14 %, and 3.02 %, respectively (Table S11).

4. Discussion

4.1. Main findings

In this prespecified analysis of the PANDA trial, we found that Xuesaitong soft capsules, compared with placebo, may increase the likelihood of functional independence at 3 months in ischemic stroke patients without diabetes or treated hyperglycemia during hospitalization, but this benefit was not significant in those with diabetes or treated hyperglycemia during hospitalization. Meanwhile, both patients without previous diabetes and those with fasting blood glucose levels below 7.0 mmol/L at baseline presented a trend toward benefiting from treatment with Xuesaitong soft capsules, although the subgroup interaction tests did not reach statistical significance.

4.2. Mechanistic hypothesis

We hypothesized that multiple mechanisms involved in the pathobiology of ischemic stroke in patients with diabetes or treated hyperglycemia during hospitalization may modulate the efficacy of cerebroprotective agents like PNS. PNS may exert its cerebroprotective effects through multiple pathways, including antioxidant and anti-inflammatory actions, anti-apoptotic signaling, and pro-angiogenic effects [16,23,24]. However, in patients with diabetes or treated hyperglycemia, it is conceivable that hyperglycemia-induced endothelial dysfunction, microvascular rarefaction, and chronic inflammation might complicate stroke outcomes [25]. Elevated glucose levels might possibly contribute, at least to some degree, trigger complement cascade activation, intensifying neuroinflammatory responses and compromising blood-brain barrier integrity [13]. A possible hypothesis was that these pathophysiological changes might modulate the efficacy of PNS-mediated suppression of inflammatory factors. Moreover, it was hypothesized that diabetes-induced microvascular remodeling might increase resistance to cerebral perfusion, possibly limiting PNS's ability to improve blood flow in the ischemic penumbra [25,26]. In addition, a randomized clinical trial on the efficacy and safety of butylphthalide, a cerebroprotective drug by acting on multiple active targets, reported similar trend [27].

It is important to note that while baseline characteristics exhibited minor variations across study groups and adjustments for these variables were incorporated into the analytical models, these differences may still have influenced the study outcomes. Specifically, among patients with diabetes or those receiving treatment for hyperglycemia during hospitalization, a statistically significant disparity in baseline EQ-5D scores was observed between the Xuesaitong and placebo groups, with the Xuesaitong group demonstrating lower scores. This discrepancy could potentially lead to an underestimation of Xuesaitong's therapeutic efficacy. Conversely, in patients without diabetes or hyperglycemia treatment during hospitalization, the placebo group showed a higher proportion of male participants, a pattern similar to the primary study of the PANDA trial [5]. Current evidence suggested that gender may not significantly modify the treatment effect of Xuesaitong [5]. However, further research is warranted to explore the potential impact of the observed baseline differences. Notably, there was a temporal divergence between 3-month and 12-month functional independence. The strongest and most consistent finding was observed at 3 months, which might stem from patients receiving Xuesaitong or placebo for only 3 months to some extent. However, the attenuation of functional independence effects at 12 months was likely influenced by several key factors, including the quality of post-treatment care, the recovery ceiling effect, and the incorporation of additional therapeutic interventions.

4.3. Clinical implications

Building on previously reported subgroup analyses of age - based heterogeneity, our study complemented the framework of individualized Xuesaitong therapy by investigating the impact of diabetes - related factors on therapeutic outcomes [6]. These insights have critical clinical implications for stroke management. In patients with diabetes or treated hyperglycemia during hospitalization, prioritizing glycemic control is essential [28]. Xuesaitong may still serve as an adjunct to improve clinical outcomes, but its use should be cautiously considered in the context of comorbid diabetes, given the potential for attenuated responses. In contrast, patients without diabetes or treated hyperglycemia during hospitalization may benefit from Xuesaitong soft capsules, given the demonstrated efficacy in improving functional outcomes. Furthermore, this approach aligns with principles of precision medicine and polypharmacy reduction [29]. By carefully considering the patient's diabetes status and the potential for drug-related issues such as adverse reactions and interactions, we can streamline medication regimens, minimizing the risk of polypharmacy - associated complications while maximizing therapeutic benefits [30]. Therefore, diabetes status may influence the treatment response of Xuesaitong and should be evaluated in future confirmatory trials or prospective stratified analyses.

4.4. Strengths and limitations

Our study boasts several strengths that contribute to the field. First, it is firmly grounded in the effective implementation of randomized controlled trial (RCT) evidence and the pursuit of individualized therapy with cerebroprotective agents. Second, we delved deep into diabetes - related data by conducting multiple subgroup analyses. These analyses served as a rigorous validation process for our findings. Third, we performed sensitivity analyses, which allowed us to assess the impact of potential confounding factors and alternative assumptions on the study outcomes, thereby enhancing the overall credibility of our conclusions. Fourth, this study carries profound implications for the design of future RCTs in the realm of ischemic stroke. It underscores the critical importance of paying close attention to the impact of diabetes comorbidity on patient outcomes. By doing so, researchers can pave the way for the development of more tailored and effective treatment strategies that take into account the unique needs of patients with ischemic stroke and concurrent diabetes.

Several limitations inherent to this study warrant consideration when interpreting the findings. First, patients were classified as suffering from diabetes if they received antidiabetic medications during index hospitalization for enrollment in the PANDA trial, which may introduce misclassification bias given that acute stroke can cause temporary hyperglycemia. However, given the implementation of evidence - based guidelines and ongoing improvements in healthcare quality in China, the impact on the study's findings is anticipated to be limited [31]. Also, we conducted additional analyses by categorizing patients based on pre-existing diabetes status and antidiabetic agent use within the first 3 months post-stroke, to comprehensively evaluate the impact of concomitant diabetes on Xuesaitong therapy outcomes. Second, our study was a preplanned analysis, which increased the risk of a type I error. Therefore, the findings were exploratory and need validation. Third, using LOCF to handle missing mRS data may introduce bias by failing to account for temporal changes in patients' recovery trajectories. However, sensitivity analyses employing multiple imputation yielded consistent results, reinforcing the robustness of our findings. Fourth, the attenuation of functional independence effects at 12 months was likely influenced by post-treatment care, recovery ceiling effects, and additional therapies. Fifth, this study was conducted among ischemic stroke patients in China, so it may be challenging to generalize the findings to other countries or regions. However, it targeted critical clinical dilemmas within the Chinese context, thereby offering valuable insights for clinical practice and guiding future research endeavors. Sixth, while we discussed potential mechanisms by which diabetes influences the therapeutic outcomes of Xuesaitong based on existing evidence, relevant biomarker validation remains absent. Future research could further investigate these underlying mechanisms and provide empirical validation.

5. Conclusion

While Xuesaitong soft capsules represent a safe and effective option for ischemic stroke recovery, their benefits may be moderated by diabetes or treated hyperglycemia during hospitalization. Patients without, rather than with, diabetes or treated hyperglycemia during hospitalization may receive greater benefit from Xuesaitong soft capsules regarding functional independence at 3 months. Clinical decision - making could therefore consider comorbid diabetes or treated hyperglycemia. This tailored approach not only optimizes stroke outcomes but also minimizes the risks associated with polypharmacy, providing broader efforts to enhance the precision and safety of stroke care.

Author contributions

The authorship order was determined through discussions among all the authors and was ultimately finalized by the corresponding authors, with the consensus of every author involved. Conceptualisation: Anxin Wang, Ying Gao, Haiqing Song, Xinya Li, Yuan Shen; Data curation: Anxin Wang, Ying Gao, Haiqing Song, Xinya Li, Yuan Shen; Formal analysis: Xinya Li, Yuan Shen, Xiaoli Zhang; Funding acquisition: Anxin Wang; Investigation: Xinya Li, Yuan Shen, Xiaoli Zhang; Methodology: Anxin Wang, Xinya Li, Yuan Shen; Project administration: Anxin Wang, Ying Gao, Haiqing Song, Xunming Ji; Resources: Anxin Wang, Ying Gao, Haiqing Song, Xunming Ji; Software: Xinya Li; Supervision: Anxin Wang, Ying Gao, Haiqing Song; Validation: all authors; Visualisation: Xinya Li, Yuan Shen; Writing - original draft: Xinya Li, Yuan Shen; Writing - review & editing: all authors. All authors confirm that they had full access to all the data in the study and accept responsibility to submit for publication.

Data sharing statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.

Funding

The work was funded by grant GN-2017B0003 from the Program of Research and Popularization of Appropriate Intervention Technology for the Stroke High Risk Group in China and China Resources Kunming Shenghuo Pharmaceutical Co Ltd. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Declaration of competing interest

All authors declared no competing interests for this work.

Appendix A. Supplementary data

The following is the Supplementary data to this article: