The inclusion criteria were as follows: 1) central neurological pathology responsible for SEF; 2) gait assessment by 3D-IGA carried out before the MNB; 3) temporary calf MNB targeting truncal tibial nerve block or gastrocnemius motor branches and/or soleus superior motor branch and/or tibialis posterior motor branch and/or selective intramuscular anesthetic blocks; and 4) at least 18 years of age at the time of consent.

The exclusion criteria applied to patients who i) refused to allow their data to be used for research purposes; ii) had benefited from MNB other than that indicated above; and iii) had undergone calf surgery between 3D-IGA and MNB.

The 3D-IGA was performed in the University Hospital Gait Laboratory. Kinematic data were collected using an optoelectronic system of eight high-resolution infrared cameras (Vicon, Oxford Metrics, UK) with passive markers positioned according to the modified Hayes model (13). Kinematic data were averaged for each patient from data from 11 gait cycles. EMG data were collected with a 16-channel wireless surface electrode dynamic electromyography system (Wave Plus Wireless sEMG system, Cometa, Italy) without any normalization procedure. No preprocessing was applied to the EMG data. All these data were expressed from 0 to 100% over the gait cycle.

The MNB muscle targets were chosen considering both the instrumental data and the clinical and anamnestic data collected just before the 3D-IGA. MNB was administered at a distance from the 3D-IGA, during the second day of hospitalization allowing the interpretation of the data collected during the 3D-IGA. In accordance with current recommendations (14), selective blocks of tibial motor nerve branches to the gastrocnemii and/or soleus and/or tibialis posterior were administered under electrostimulation combined with ultrasound guidance. Intramuscular blocks and nonselective tibial nerve blocks were administered in the popliteal fossa under exclusive ultrasound guidance. The use of a long half-acting anesthetic (ropivacaine) prolonged the effect for a more reliable subjective self-assessment. We documented the following potential adverse effects of MNB: persistent pain, persistent sensory or motor disorder, and hematoma.

The spontaneous speed gait was filmed before and after the MNB in a dedicated room equipped with two high-definition and high-frequency (120 fps) fixed cameras: one in the front/back for frontal plane assessment and one in the profile for sagittal plane assessment. The pre-post videos were analyzed by the Kinovea^® software, allowing simultaneous and synchronized viewing and slow motion. Qualitative judgment was made by three independent evaluators.

The descriptive data collected were age (years), gender (male/female), pathology, the period between 3D-IGA and MNB (months), modified Functional Ambulation Classification scale score (15), and gait speed (m.s⁻¹).

The same investigator (referring PRM physician) performed the pre- and post-MNB clinical evaluations under identical conditions. The results from the analytical clinical examination were as follows:

The characteristics of the gait pattern were assessed before and after the MNB by qualitative video analysis as described in the Edinburgh visual gait score (18), including the collection of the following criteria: maximal ADF limitation during the stance (<5°) and the swing phase (<0°) and hind footvarus during stance. The following criteria were blindly assessed by three operators from videos recorded in the sagittal plane (Figure 1): Improvement in active ADF after MNB was defined as either an improvement in the maximum active ADF during swing or an improvement in the sagittal position of the foot at the initial contact; visible protrusion of the TA tendon during swing was reported.

The data collected during the patient interview were as follows:

Instrumental data extracted from 3D-IGA were peak ADF during stance and swing phases, and EMG data from the medial gastrocnemius, soleus, peroneus longus, and TA. The swing phase was divided into three periods as follows: “early-mid swing” from toe-off to the time when the swinging foot was opposite the stance foot, “mid-swing” from the end of the previous phase to when the tibia of the swinging limb was vertical, and “terminal swing” from the end of mid-swing to foot strike (2). The EMG patterns of calf and peroneus longus muscles were described in three categories: physiological activation, overactivity during early and/or mid-swing (with or without terminal swing overactivity), and isolated overactivity during terminal swing (Figure 2). EMG patterns of TA were described as “physiological,” “abnormal first activation” (i.e., loss or decreased the first activation in early swing), and “isolated abnormal second activation” (i.e., loss or decreased second activation in terminal swing and initial contact). Three operators performed a blind qualitative analysis of EMG data from 5 to 6 gait cycles to minimize bias in characterizing EMG patterns.

Statistical analyses were performed using the R software (version 4.1.0, R Foundation, Vienna, Austria) and the R Studio interface (version 1.4.1717).

Demographic data were described in terms of number and percentage for categorical variables, while continuous data were described in terms of mean, standard deviation, minimum, and maximum values.

We examined the relationship between pre- and post-MNB clinical data and instrumental data with odds ratio and the estimated 95% confidence interval (OR [95% CI]), interpreting these as effect size indicators and deeming them of interest and relevance when > 4 (19). Although the amplitude of the sample size has little impact on the effect size, the confidence interval is reduced with an increase in the statistical power of the study (20). Consequently, in a pilot study with low statistical power, it is possible to discuss the amplitude of size effect even if the confidence interval includes the zero value. We examined the positive and negative predictive values (PPV and NPV), particularly when it was not possible to calculate the OR (a headcount of 0).

We studied associations between:

A total of 126 lower limb MNBs were administered and finally 42 MNBs administered to 40 subjects were included in the analysis (refer to the flowchart in Figure 3 for details). All patients had previously received botulinum toxin injections in the calf muscles, but both 3D-IGA and MNB occurred outside the efficacy period, i.e., when the patient no longer feels the effects of the botulinum toxin on the functional objectives and at least 4 months after the last injection. The mean time from 3D-IGA to MNB was 7.12 [4.56–9.68] months. Table 1 presents the demographic data of the population. The instrumental and clinical data collected for each record are presented in the supplemental materials (Supplementary Table S1).

The mean spontaneous gait speed was 0.68 ± 0.30 m.s⁻¹. The main disorders were foot catching (n = 36/42; 86%) and lateral ankle twist (n = 30/42; 71%). Qualitative video assessment using EVGS highlighted the limitation of ADF for 22/42 (52%) both in stance and swing phases and varus during stance for 12/42 (29%) feet.

In total, 24 (57%) MNB procedures specifically targeted gastrocnemii, 31 (74%) the soleus, 30 (71%) the tibialis posterior, 6 (15%) the flexor digitorum longus, 5 (12%) the flexor hallucis longus, and 1 (2%) the fibularis longus. A total of 40 (95%) MNB met the criteria for complete analytical effectiveness (disappearance of spasticity and/or full paresis in the targeted muscles). After MNB, the TA Held score increased in 9 (64% of subjects with Held score <4 pre-MNB) cases, and the Boyd score increased in 15 (35% of subjects with Boyd score <4 pre-MNB) cases.

In 23 (55%) cases, MNB caused sensory disorders, indicating anesthesia or diffusion of the anesthetic product in the tibial nerve trunk. In 6 (14%) cases, MNB caused calf pain in the ensuing days, without any other obvious complication.

Supine ADF assessment pre-MNB was ≤ 0 ° for 21 patients with KF and 35 feet with KE and increased post-MNB for 14 (67%) and 22 (61%) patients, respectively (refer to Table 2). No clinical variable (including qualitative video assessment) was associated with equinus during the stance phase on 3D-IGA except on post-MNB equinus KF in supine position [OR = 8.00 (1.19–53.93)] (refer to Table 3).

Dynamic EMG data during the swing phase was usable for 30 (71%) and 37 (88%) MNB for the medial gastrocnemius and the soleus, respectively (refer to Table 2). We found no association between spasticity and dynamic overactivity (i.e., spastic co-contraction) of the soleus (refer to Table 3).

We examined the association between four modalities of TA assessment of the 40 MNB with interpretable TA EMG data (refer to Table 2). We found no significant association between physiological TA EMG and clinical data, except an excellent PPV (100%) with no visible protrusion of TA tendon and an abnormal EMG pattern of TA, but concerning 3 patients only (refer to Table 3). Furthermore, a strong association was established between loss of terminal swing TA activation and non-taligrade initial contact at foot-strike [OR = 8.33 (1.60–43.29)].

An increase in ADF during the swing phase has been noted in 17 (43%) of the 39 records for which the ADF KF was ≥ 0° after the MNB. It was associated with subjective improvement [OR = 11.25 (1.25–120.63)].

Table 3 presents relationship between EMG, clinical variables, and swing ADF improvement post-MNB. More relevant OR were found with early and/or mid-swing overactivity of the soleus [OR = 4.72 (0.86–26.04)].

Our results confirmed that MNB is a safe procedure as no permanent pain, sensory disorder, muscle deficit, or hematoma was observed. This was previously described by Filipetti et al. who reported only 5 painful procedures, 2 hematomas, and no severe complications after 815 MNB procedures (33).

Our results confirm the poor effectiveness of supine examination to assess calf contracture since a limitation of maximal ADF during the stance phase was not associated with the limitation of passive ADF during the supine examination and was not associated with ADF assessed after MNB. It is likely that passive ADF was underestimated by the presence of spastic dystonia, mimicking contracture, or by spastic myopathy (4). The association between the limitation of maximal ADF during the stance phase and the limitation of maximal passive ADF during the supine examination post-MNB was better, especially with KF, which can be explained by the suppression of spastic dystonia (9, 33). However, the positive predictive value was low, possibly because of the ineffectiveness of MNB on spastic myopathy. Finally, this could also be explained by the imprecise measurement of ADF with monosegmental kinematic models in the presence of frontal deformities of the foot, with a potential overestimation of 7° (34). It could also be explained by the variability in kinematic patterns used to compensate for the lack of forward progression, such as knee recurvatum, which involves maximal stretching of the gastrocnemii (2, 35). Consequently, instrumental measurement of ADF during gait is not satisfactory to evaluate calf contracture, which could be interpreted in conjunction with clinical examination and video to clarify mechanisms involved in ankle/foot deformities (35–37).

During an early swing, the TA muscle contracts concentrically, allowing the ankle to perform a rapid dorsiflexion movement, bringing the ankle to 0° for foot clearance (2, 37). Overactivity of the calf muscles, especially the triceps surae, during this period corresponds to spastic co-contraction (4) involved in limiting active movement (38) by reversing dorsiflexor torque to plantar-flexor torque (39). This observation, described as a “stretch-sensitive paresis” (4), was noted for 26% of the cases in a hemiplegic population of 18 subjects (39). Our findings of a strong and significant association between spastic co-contraction of the soleus and an increase in ADF post-MNB, and to a lesser extent of the medial gastrocnemius, are consistent with the literature. This is also coherent with the major implication of soleus overactivity in SEF, which came to light from examining the ratio between the maximum amplitude of the H reflex and the maximum amplitude of the M response (40). Selective MNB decreased this ratio in the case of the soleus muscle with no change in the medial gastrocnemius (41). Compared to gastrocnemius branches, selective soleus MNB also demonstrated a more significant improvement in terms of stretch reflex scores, gait parameters, and patient self-reported improvements (42), and sometimes isolated soleus neurotomies were sufficient to induce a persistent decrease in the triceps surae stretch reflex score (43). Conversely, terminal swing TS overactivity was not associated with an improvement in the ADF after MNB. It is possible that this premature activation reflects a compensatory phenomenon allowing the body weight to be absorbed in foot strike (21, 22). No relation was found between an improvement in ADF post-MNB and TS spasticity either, highlighting that spasticity assessment does not predict what happens during movement (4, 38).

Our results confirmed that the different TA function assessment modalities were not equivalent. The supine evaluation, before or after MNB, which is an assessment of the strength and selectivity of motor control, was not related to the quality of temporal activation during the swing phase, which indicates an automatic-voluntary dissociation. None of these parameters was predictive of improved ADF post-MNB. In addition, the presence of a visible protrusion of TA tendon on the video also appears unreliable, as it was not related to any of the above criteria.

The poor predictive value of TA EMG activation during the swing phase was previously shown by Burridge et al. in a cohort of 112 patients, among whom more than 50% had a limitation in dorsiflexion speed despite TA activation without triceps surae overactivity (23). In our study, this could be explained by either spastic myopathy, overactivity of plantar flexor muscles not evaluated by dynamic EMG (lateral gastrocnemius, tibialis posterior, fibularis longus), or by the lack of activation of accessory levators (extensor digitorum longus and extensor pollicis longus) (24). In any event, this indicates the need for further studies to analyze a possible modification in TA activation pattern before and after MNB or even to compare the envelopes of EMG signals.

Finally, in our population, there was no improvement in ADF for 4 patients during the swing phase, although they all had a physiological EMG TA pattern with a high Held score (4/5) and no calf contracture. Moreover, 3 had a tibialis posterior MNB, excluding the involvement of this muscle in persistent ADF deficit, but all 4 patients had a very slow walking speed <0.5 m.s⁻¹ (mean = 0.31 m.s⁻¹) compared to the 6 patients whose ADF increased (walking speed > 0.5 m.s⁻¹, mean = 0.67 m.s⁻¹). The impact of walking speed on gait kinematics and kinetic parameters has already been demonstrated (36), and this major factor should always be included in the interpretation of instrumental data.

TA normally activates strongly after heel strike, which controls plantar flexion, and then again at toe-off to lift the foot clear of the ground, peaking during early swing (23). After a stroke, the initial contact activation of TA is particularly low compared to healthy subjects (25). However, the significance of the second abnormal activation of the TA muscle remains uncertain. Our results highlighted no pejorative values in this pattern because it was not associated with a limited ADF during the swing phase after MNB.

Conversely, we found a strong association between abnormalities in the position of the foot at initial contact and an activation defect at the end of the swing phase [OR = 8.33 (1.60–43.29)] that could reflect more of a consequence of the equinus gait than a primitive gait abnormality. This hypothesis is supported by some elements in the literature: several articles have reported decreased activation of the TA during early stance in healthy children fitted with an orthosis simulating an equinus (21), in healthy adult subjects simulating toe-walking, and in adult cerebral palsy subjects (22). Therefore, an alteration in this pattern could reflect either a primitive anomaly that explains the ADF defect during this phase, or a compensatory anomaly linked to the absence of the need for eccentric contraction when the initial contact is not taligrade. In the literature, three studies investigated TA activation patterns before and after SEF foot surgery: They showed no significant change in activation patterns, but the analysis was global, as patterns were only characterized as in phase/out of phase (26) or as continuous, phasic premature and/or prolonged, or physiological (27). Keenan et al. showed the appearance of a phasic activation pattern in 6 of the 16 patients who showed continuous activation before surgery (28). The results of our study are insufficient to prove a causal association, and only a complementary analysis with dynamic EMG before and after MNB or equinus foot surgery, to specifically identify the persistence or lack of persistence of an alteration in the activation during terminal swing and early stance, could answer this question.

Similarly, the pathological character of triceps surae overactivity during the terminal swing phase can be questioned, since it could correspond to an eccentric contraction cushioning the body weight during initial digitigrade contact.

The first limitation of this study is its retrospective nature, which does not allow prior standardization of the characteristics of subjects, and leads to a lower level of evidence. However, despite the retrospective nature of the study, standardization of clinical and instrumental data collection and triple-blind evaluation of video analysis criteria optimized the reproducibility of measurements. The small cohort together with a low statistical power can make it difficult to demonstrate a statistically significant difference in terms of the p-value or confidence interval of effect size. Interpretation of the odds ratio as an indicator of size effect enabled us to move beyond these limitations. Nevertheless, caution should be exercised when extrapolating results (19).

Furthermore, 12 and 35% of the EMG data were uninterpretable for the soleus and medial gastrocnemius, respectively. This could be improved by preprocessing and normalization of the EMG data, as it is difficult to distinguish background noise from permanent overactivity.

The lack of 3D-IGA to assess ADF at the post-MNB stage is also a limitation since video assessment is not the gold standard in terms of kinematic data collection. Only 3 studies discussed the use of an optoelectronic system to compare ankle kinematics after isolated tibial nerve neurotomy in a total of 28 patients (1, 29, 30). They all demonstrated increased ADF during the stance phase and on initial foot contact (29, 30) with one of the studies highlighting increased ADF during the swing phase (1). Despite the limitations, video assessment is the most common method for assessing tibial MNB impact on ADF (11) and has been used after lower limb surgery (5, 6, 31) involving neurotomy, tendon lengthening and/or transfer, with sufficient sensitivity to change to demonstrate improved ADF during stance and swing phases.