The patient disposition has been reported previously [28]. In brief, 220 patients were enrolled in the study, among whom 214 were included in the modified intention-to-treat set (mirogabalin add-on group, n = 110; NSAIDs alone group, n = 104), and 216 were included in the safety analysis set (mirogabalin add-on group, n = 110; NSAIDs alone group, n = 106).

Patient background characteristics have been reported previously and were similar between the two treatment groups [28]. In the mirogabalin add-on group and the NSAIDs alone group, the respective mean ages were 67.8 and 70.9 years; proportions of female patients, 54.5% and 49.0%; mean CrCL, 81.5 and 70.7 mL/min; proportions of patients with CrCL 30 to < 60 mL/min, 27.3% and 33.7%; and mean VAS scores, 63.8 and 62.8 mm.

A forest plot of patient background characteristics by presence of mirogabalin-related ADRs in the mirogabalin add-on group is shown in Fig. 1. The incidence rates (95% CI) of mirogabalin-related ADRs were 44.4% (27.9%, 61.9%), 52.5% (36.1%, 68.5%), and 52.9% (35.1%, 70.2%) in patients aged < 65, 65 to < 75, and ≥ 75 years, respectively; and 43.3% (25.5%, 62.6%) in patients with CrCL 30 to < 60 mL/min and 52.5% (41.0%, 63.8%) in patients with CrCL ≥ 60 mL/min. There were no significant differences in the incidence rates of mirogabalin-related ADRs across patients with different background characteristics.

Patient background characteristics and presence or absence of mirogabalin-related ADRs in the mirogabalin add-on group (safety analysis set). 1Clopper–Pearson method. 2Wald method. 3Chi-square test. ADRs adverse drug reactions, CI confidence interval, CrCL creatinine clearance, LSS lumbar spinal stenosis, VAS visual analog scale

The time to first onset of mirogabalin-related ADRs (somnolence, dizziness, edema, and peripheral edema) in the mirogabalin add-on group is shown in Fig. 2. Mirogabalin-related ADRs occurred mainly after the first administration and the titration period (Fig. 2a). A similar tendency was observed in the effective dose group (Fig. 2b) and initial dose-maintained group (Fig. 2c) among the patients in the mirogabalin add-on group.

Time to first onset of mirogabalin-related ADRs (somnolence, dizziness, edema, and peripheral edema) in the a total, b effective dose, and c initial dose-maintained mirogabalin add-on groups (safety analysis set). ADRs adverse drug reactions

EQ-5D-5L scores at baseline and week 12 and their change from baseline to week 12 in patients with and without mirogabalin-related ADRs in the mirogabalin add-on group are shown in Table 1. Patients with mirogabalin-related ADRs showed a significant improvement in EQ-5D-5L scores from baseline to week 12 compared with those without mirogabalin-related ADRs (estimated between-group difference 0.0767 [95% CI 0.0075, 0.1460]; p = 0.0304). Similar results were observed among patients who were compliant with the package insert (Table S2).

PGIC scores at week 12 in patients with and without mirogabalin-related ADRs in the mirogabalin add-on group are shown in Table 2. At week 12, the proportion of patients with a PGIC score ≤ 3 (the sum of minimally, much, and very much improved) was higher in patients with mirogabalin-related ADRs vs those without (85.4% vs 67.4%; estimated between-group difference: 17.9 [0.2, 35.6], p = 0.0539). Similar results were obtained among patients in patients who were compliant with the package insert (Table S3).

A forest plot of patient background characteristics and percentage of patients with improvement in VAS score for leg pain of ≥ 20 mm at week 12 in the mirogabalin add-on group is shown in Fig. 3. There were no significant differences in the percentage of patients with reduced leg pain (improvement in VAS score of ≥ 20 mm) at week 12 by patient background except for spondylolysis/spondylolisthesis as a complication. Similar results were obtained in patients who were compliant with the package insert (Fig. S1).

Patient background characteristics and percentage of patients with improvement in VAS score for leg pain of ≥ 20 mm at week 12 in the mirogabalin add-on group (modified intention-to-treat set). 1Clopper–Pearson method. 2Wald method. 3Chi-square test. CI confidence interval, CrCL creatinine clearance, LSS lumbar spinal stenosis, VAS visual analogue scale

The relationship between baseline SPDQ numbness score and the change in EQ-5D-5L from baseline to week 12 in both treatment groups is shown in Fig. 4. In the mirogabalin add-on group, but not in the NSAIDs alone group, higher SPDQ numbness scores at baseline were associated with greater improvement in EQ-5D-5L, at week 12, although the correlation was weak (r = 0.2811, p = 0.0092). There was a significant estimated difference in the change in EQ-5D-5L score from baseline to week 12 between patients with SPDQ numbness scores ≤ 2 and > 2 in the mirogabalin add-on group (− 0.0779 [95% CI − 0.1472, − 0.0086]; p = 0.0280), but not in the NSAIDs alone group (Fig. 5 and Table 3).

Relationship between SPDQ numbness score at baseline and the change in EQ-5D-5L from baseline to week 12 in the a mirogabalin add-on group and b NSAIDs alone group (modified intention-to-treat set). Spearman’s rank correlation coefficient. EQ-5D-5L was analyzed using complete case analyses. NSAIDs non-steroidal anti-inflammatory drugs, SPDQ spine painDETECT questionnaire

Change in EQ-5D-5L from baseline to week 12 by SPDQ numbness score at baseline. p value was calculated by t test. EQ-5D-5L was analyzed using complete case analyses. CI confidence interval, NSAIDs non-steroidal anti-inflammatory drugs, SD standard deviation, SPDQ spine painDETECT questionnaire

In both treatment groups, there was no significant relationship between baseline SPDQ numbness score and change in VAS score for leg pain from baseline to week 12 (Fig. 6 and Table 3). Similar results were observed in patients who were compliant with the package insert (Fig. S2 and Table S4).

Relationship between SPDQ numbness score at baseline and the change in VAS score for leg pain from baseline to week 12 in a the mirogabalin add-on group and b the NSAIDs alone group (modified intention-to-treat set). Spearman’s rank correlation coefficient. VAS score analyzed using the mixed model for repeated measures. NSAIDs non-steroidal anti-inflammatory drugs, SPDQ spine painDETECT questionnaire, VAS visual analogue scale

In both treatment groups, there was a significant positive relationship between PGIC score at week 12 and the change in EQ-5D-5L from baseline to week 12 (mirogabalin add-on group: r = − 0.4612, p < 0.0001; NSAIDs alone group: r = − 0.4875, p < 0.0001; Fig. 7a, b). The estimated difference in change in EQ-5D-5L score from baseline to week 12 between patients with PGIC scores ≤ 2 and > 2 was 0.1116 (95% CI 0.0437, 0.1794), p = 0.0016 in the mirogabalin add-on group and 0.1115 (0.0411, 0.1819), p = 0.0023 in the NSAIDs alone group (Table 4); and between patients with PGIC scores ≤ 3 and > 3 it was 0.1548 (0.0773, 0.2322), p = 0.0002 in the mirogabalin add-on group and 0.1089 (0.0434, 0.1744), p = 0.0014 in the NSAIDs alone group. Similar results were obtained in patients who were compliant with the package insert (Fig. S3 and Table S5).

Relationships between PGIC score at week 12 and the change from baseline to week 12 in a, b EQ-5D-5L and c, d VAS score for leg pain in the a, c mirogabalin add-on and b, d NSAIDs alone groups. Spearman’s rank correlation coefficient. EQ-5D-5L was analyzed using complete case analyses. VAS score analyzed using the mixed model for repeated measures. NSAIDs non-steroidal anti-inflammatory drugs, PGIC Patient Global Impression of Change, VAS visual analogue scale

The relationship between PGIC score at week 12 and the change in VAS score for leg pain from baseline to week 12 in both treatment groups is shown in Fig. 7c, d. In both treatment groups, there was a significant positive relationship between PGIC score at week 12 and the change in VAS score for leg pain from baseline to week 12 (mirogabalin add-on group: r = 0.6520, p < 0.0001; NSAIDs alone group; r = 0.5406, p < 0.0001). Patients with PGIC scores ≤ 2 showed significant improvement in VAS scores at week 12 compared with those with PGIC score > 2 (estimated difference in change in VAS score from baseline to week 12 between patients with PGIC score ≤ 2 and > 2, − 31.1 [95% CI − 41.0, − 21.3], p < 0.0001 in the mirogabalin add-on group and − 33.8 [− 44.0, − 23.6], p < 0.0001 in the NSAIDs alone group; Table 4). Similar significant results were obtained in patients with PGIC score ≤ 3 and > 3 (estimated difference in change in VAS score from baseline to week 12 between patients with PGIC score ≤ 3 and > 3, − 33.5 [95% CI − 45.5, − 21.5], p < 0.0001 in the mirogabalin add-on group and − 25.5 [− 36.1, − 14.9], p < 0.0001 in the NSAIDs alone group).

The VAS score similarly improved from baseline to week 12 irrespective of LSS disease duration in the mirogabalin add-on group (estimated change in VAS score from baseline to week 12, − 30.6 [95% CI − 40.4, − 20.58] and − 21.7 [− 28.8, − 14.7] in patients with LSS duration < 6 months and ≥ 6 months, respectively; intergroup difference, − 8.9 [− 20.9, 3.2], p = 0.1461; Table 5). The improvement effects on VAS score were different depending on LSS disease duration in the NSAIDs alone group. The results in patients who were compliant with the package insert are shown in Table S6.