Right C7 neurotomy at the intervertebral foramen plus intensive speech and language therapy versus intensive speech and language therapy alone for chronic post-stroke aphasia: multicentre, randomised controlled trial

Objective To evaluate whether right neurotomy of the seventh cervical nerve (C7) at the intervertebral foramen plus intensive speech and language therapy (SLT) improves language function compared intensive SLT alone in patients with chronic aphasia after stroke.

Design Multicentre, assessor blinded, randomised controlled trial.

Setting Four centres in mainland China.

Participants 50 adults aged 40-65 years with aphasia for more than one year after a single left hemispheric stroke.

Interventions Participants were randomised 1:1 to receive either C7 neurotomy plus three weeks of intensive SLT or three weeks of intensive SLT only, stratified by treatment centre.

Main outcome measures The primary outcome was change in score on the 60 item Boston naming test (BNT, scores 0-60, with higher scores indicating better naming function) from baseline to one week after C7 neurotomy plus intensive SLT for three weeks or intensive SLT for three weeks after deferral for one week (control group). Secondary outcomes included change in severity of aphasia using the aphasia quotient, calculated using the western aphasia battery, and patient reported outcomes on quality of life and depression after stroke.

Results From 25 July 2022 to 31 July 2023, 322 out of 1086 patients received a diagnosis of post-stroke aphasia and were screened for eligibility. 50 eligible participants were randomly assigned to treatment groups (25 in each). Mean increase in BNT score was 11.16 points in the neurotomy plus SLT group and 2.72 points in the control group at one month (difference 8.51 points, 95% confidence interval (CI) 5.31 to 11.71, P<0.001). The between group difference in BNT score remained stable at six months (difference 8.26 points, 4.16 to 12.35, P<0.001). In addition, the aphasia quotient improved significantly in the neurotomy plus SLT group versus control group (difference at one month 7.06 points, 4.41 to 9.72, P<0.001), as did patient reported activities of daily living and post-stroke depression. No treatment related severe adverse events were reported.

Conclusions C7 neurotomy plus three weeks of intensive SLT was associated with a greater improvement in language function compared with three weeks of intensive SLT alone over a period of six months. No severe adverse events or long term troublesome symptoms or functional loss were reported.

Trial registration Chinese Clinical Trial Register ChiCTR2200057180.

Study design

This was a multicentre, assessor blinded, randomised controlled trial in four centres in Shanghai, China. The study protocol is published elsewhere.14

Participants

We recruited patients aged 40-65 years who spoke fluent Chinese before stroke. Patients were eligible for inclusion if they had aphasia for more than 12 months after a single onset of infarction or haemorrhage of the left hemisphere confirmed by magnetic resonance imaging (MRI), manifesting as an inability to express themselves or difficulties with verbal communications for more than one year. Aphasia required diagnosis by two doctors and a score of <93.8 on the aphasia quotient calculated by the Chinese version of the western aphasia battery and classified as level ≥1 in the Boston diagnostic aphasia examination. Participants were required to have basic comprehension and to fully understand the study design and cooperate with the treatment plan. Patients were excluded if they had neurodegenerative disease, traumatic brain injury, a history of aphasia before the last onset of stroke, severe motor speech disorder, hearing impairment, or received intensive post-stroke rehabilitation therapy within four weeks before recruitment (see supplementary appendix for inclusion and exclusion criteria).

Randomisation and masking

After screening, eligible patients were randomly assigned in a 1:1 ratio to C7 neurotomy plus intensive SLT or to intensive SLT only. Randomisation was stratified according to trial centre without using blocks.14 The statistician responsible for generating the randomisation sequence was not involved in patient enrolment. An independent third party (Trial Data Pharmaceutical Technology, Shanghai, China) carried out randomisation through a centralised interactive web based randomisation system. Investigators randomly assigned patients through the online system and disclosed allocation to the patients.

Neither the patients nor clinical investigators providing the interventions were masked to group allocation. However, the outcome assessors were blinded. Patients undergoing language and vocal evaluations at each follow-up were videotaped. A third party team composed of two independent evaluators and one monitor then centrally evaluated the recordings. The evaluators and monitor were unaware of patients’ identities, treatment assignments, time points, or location of clinical sites. Patients wore a cervical collar during evaluations to conceal any surgical scars.

Interventions

The C7 neurotomy procedure is described in the published protocol and shown schematically in supplementary figure S1 and the embedded video (also in supplementary file as video S1). Briefly, an incision was made at the medial side of the sternocleidomastoid muscle on the right side and the C7 nerve root exposed as proximally as possible and then sectioned at the intervertebral foramen. The distal cut end of the C7 nerve was close to the sternocleidomastoid muscle for use as nerve end recipient if patients had verbally expressed a willingness to undergo nerve transfer.915 Surgeons experienced in brachial plexus surgery and trained in Huashan Hospital’s standard surgical protocols performed the C7 neurotomies.

In the neurotomy plus intensive SLT group, three weeks of intensive SLT was initiated one week after surgery. In the SLT only group, three weeks of intensive SLT was initiated after one week’s deferral (see supplementary appendix figure S2 for trial design). Trained therapists provided both groups with the same intensive SLT for at least 45 minutes twice daily for five days a week.1316 The SLT also included additional self-administrated language specific training for one hour daily. Patients at all the study centres received intensive SLT from qualified rehabilitation therapists trained in Huashan Hospital’s standard protocol.

Outcomes

The primary outcome was change in total score on the 60 item Boston naming test (BNT) of a validated Chinese version from baseline to one week after C7 neurotomy plus three weeks of intensive SLT in the intervention group or from three weeks of intensive SLT after deferral for one week in the control group.1718 BNT included 60 monochrome pictures showing objects used in daily life. Participants scored 1 point if they spontaneously recognised and correctly pronounced the object, otherwise they scored 0 (scores 0-60, with higher scores indicating better function).

Secondary outcomes included change in BNT score from baseline to three days after neurotomy in the intervention group or three days from start of deferral of intensive SLT in the control group and change from baseline to 24 weeks after initiation of intensive SLT in both groups (see supplementary figure S2). The other secondary outcomes included changes in the aphasia quotient as evaluated by a Chinese version of the western aphasia battery, which weighted the average of the first four subtest scores (scores 0-100, with higher scores indicating better function). We also assessed patient reported outcomes to evaluate changes in activities of daily living measured using the Barthel index and depression after stroke measured using the stroke aphasic depression questionnaire-hospital version. The Barthel index contains 10 questions and measures the ability to perform common activities of daily living (score 0-100, with higher scores indicating better quality of life).19 The stroke aphasic depression questionnaire scale evaluated post-stroke depression using 10 questions (each question comprises four options scoring from 0 to 3, with a total score of 30, larger scores indicating a larger tendency for depression) answered by the patient’s next of kin or care giver.20 Total scores derived from patient reported outcomes were treated as continuous variables. We also used the aphasia adapted speech language function assessment using the international classification of functioning, disability, and health to detect any physiological or structural changes relating to voice function.21 Secondary outcomes were evaluated at baseline, three days, one month, and six months.

The overt picture naming task was used during functional MRI to assess the mechanism for central plasticity accompanying functional recovery.22 Patients were asked to vocally name the monochrome objects in each picture presented on a screen during scanning. The supplementary appendix describes the design and analysis protocol.

Safety outcomes included adverse events, changes in motor function, spasticity of the right elbow, wrist, and fingers, and changes in sensory functions assessed by tactile sensory threshold and two point discrimination of the right thumb, index fingers, and middle fingers over a period of six months. Motor function included muscle power measured using the Medical Research Council’s scale (grades 0-5, with higher grades indicating better muscle power) and range of motion of the elbow, wrist, and fingers. Spasticity was measured with the modified Ashworth scale at the elbow, wrist, and fingers (scores 0-4, with higher scores indicating more spasticity).23

Statistical analysis

Based on our pilot study12 and a systematic review on intensive SLT,24 we assumed a mean difference of 5.5 points in change of BNT score between the neurotomy and intensive SLT group and control group. Therefore, we estimated that a sample size of 50 participants (25 in each group), under the assumption of a 20% dropout rate, would provide 80% power to detect a mean difference of 5.5 (standard deviation 6.2) points between groups on the BNT score with a two sided α level of 0.05. We considered a between group difference in functional improvement of 5.5 points to be of clinical importance, although 5.5 points did not represent the threshold for minimal clinically important difference. This sample size is of good credibility, would allow conclusions to be drawn at minimal risk to patients, and is in accordance with other randomised controlled trials on surgical treatments.2526

We performed efficacy analyses for primary and secondary outcomes in the intention-to-treat population, comprising all randomised participants who received any treatment and had at least one complete BNT score at follow-up. Primary outcome data were complete, with no imputation required. Descriptive statistics were used to report the characteristics of the participants at baseline. Analysis of covariance was used to compare changes in continuous outcomes between the two groups from baseline to one month, adjusting for baseline measures and study centres as covariates. For categorical variables, we used Cochran-Mantel-Haenszel χ² or Fisher’s exact tests for between group comparisons. Prespecified subgroup analyses included study centre, type of stroke, type of aphasia, and severity of aphasia, and post hoc subgroup analyses of age, sex, and duration of aphasia.

For sensitivity analysis, we assessed the longitudinal BNT and western aphasia battery score using a mixed model. Fixed factors were treatment group, treatment group and time interaction, clinical centre, and score at baseline. To account for dependent variables in the repeated measures, we included a patient specific random intercept. Using this model, we estimated marginal means for each time point and treatment group. Additional sensitivity analyses for the primary outcome were carried out in a per protocol dataset. This dataset included all participants who completed the study treatment without major protocol deviations. The statistical analysis plan provides detailed information on the analyses. All hypothesis tests were two sided, and statistical significance was set at P<0.05. All statistical analyses were performed using R software (version 4.3.2).

For assessment of functional MRI data, images of each patient were analysed using FMRIB Software Library (version 6.00) FEAT (FMRI Expert Analysis Tool) first level analysis.27 After following standard preprocessing procedures (eg, high pass filtering, motion correction, see supplementary file), we performed a univariate general linear model analysis for each patient and calculated the contrast between overt picture naming and rest interval until the next task. A whole brain correlation analysis was performed between β estimate (of overt picture naming greater than rest) and BNT score for each group at each time point (baseline and one month and six months after surgery). The threshold for the resulting correlation maps was P<0.005 and cluster corrected at P<0.05. The final results include clusters at the whole brain level containing more than 150 contiguous voxels and remaining after cluster correction.

Patient and public involvement

No patients or members of the public were involved in the design, conduct, reporting, or dissemination of this research. Available funding was specifically designated for clinical implementation and data acquisition, with no dedicated resources allocated for patient and public involvement.

Characteristics of participants

Between 25 July 2022 and 31 July 2023, a total of 322 of 1086 patients received a diagnosis of post-stroke aphasia and were screened for eligibility. Overall, 251 patients were excluded for being outside the age range for participation in the study or having aphasia for less than one year (n=107), having contradictions for assessments (n=46), having comorbidities or aphasia due to other causes (n=44), and other reasons (n=54). Among 71 eligible patients, 50 were enrolled and randomly allocated to undergo C7 neurotomy plus three weeks of intensive SLT (n=25) or intensive SLT for three weeks after deferral for one week (n=25). Twenty one patients were not enrolled: nine declined to undergo randomisation, seven declined surgery, and five did not want to participate in a clinical trial. Figure 1 shows the flow chart of participants through the study.

Fig 1

Flow chart of participants through study. C7 neurotomy=right neurotomy of the seventh cervical nerve at the intervertebral foramen; MRI=magnetic resonance imaging; SLT=speech and language therapy

Table 1 shows the personal characteristics of the participants at baseline. The median interval from neurological injury to entry into the trial was 28.8 months (interquartile range (IQR) 21.2-45.0 months) in the neurotomy plus intensive SLT group and 25.3 (17.7-37.6) months in the control group. Based on classification criteria of the Chinese version of the western aphasia battery, 45 patients had non-fluent aphasia (including 37 with Broca’s aphasia) and five had fluent aphasia. Among all the 50 patients, 42 had spastic arm paralysis with an average spasticity of around 1 degree (mild spastic), similar to the population in our previous published studies.910

Primary outcome

Mean change in BNT score from baseline to one month was 11.16 (SD 7.10) points in the neurotomy plus intensive SLT group versus 2.72 (3.40) points in the control group, with a significantly greater improvement in the neurotomy plus intensive SLT group (group difference 8.51, 95% confidence interval (CI) 5.31 to 11.71, P<0.001) (table 2, fig 2, fig 3, and supplementary videos 2-5). The BNT score from baseline to three days in the neurotomy plus intensive SLT group was higher than in the control group (group difference 5.08, 3.44 to 6.71, P<0.001). Treatment effects remained significantly stable at six months (group difference 8.26, 4.16 to 12.35, P<0.001) (table 2, fig 2, and supplementary table S2). The sensitivity analysis of the primary efficacy outcome with mixed effect model also favoured the neurotomy and intensive SLT group in increase of BNT score from baseline to each follow-up visit (see supplementary table S3). Predefined and post hoc subgroup analyses showed similar differences between subgroups for type of stroke, type of aphasia, hemiparesis, severity of aphasia, duration of disease, or other listed personal factors (see supplementary figure S3).

Fig 2

Mean change in BNT score (standard error) from baseline in C7 neurotomy plus intensive SLT group and intensive SLT only group. Data are from baseline to three days after surgery in the neurotomy plus intensive SLT group or three days after start of one week’s deferral of intensive SLT in the control group, one month from surgery or start of deferral, and baseline to six months after start of intensive SLT. BNT=Boston naming test; C7 neurotomy=right neurotomy of the seventh cervical nerve at the intervertebral foramen; SLT=speech and language therapy

Fig 3

Waterfall plot depicting observed change in BNT score for each participant from baseline to one month follow-up in C7 neurotomy plus intensive SLT group and intensive SLT only group. BNT=Boston naming test; C7 neurotomy=right neurotomy of the seventh cervical nerve at the intervertebral foramen; SLT=speech and language therapy

Secondary outcomes

Compared with the control group, the neurotomy and intensive SLT group showed increases in total aphasia quotient from baseline to three days (group difference 5.82, 95% CI 4.00 to 7.64, P<0.001) and in three out of four subsets of the western aphasia battery—spontaneous speech, comprehension, and naming (table 2). At one month, changes in total aphasia quotient favoured the neurotomy and intensive SLT group (group difference 7.06, 4.41 to 9.72, P<0.001) and all four subsets of the western aphasia battery. The total western aphasia battery score remained mainly stable at six months (see supplementary figure S4). No significant difference was found in physiological or structural changes relating to voice function in either group as evaluated by the international classification of functioning, disability, and health score (see supplementary figure S5).

For patient reported outcomes, the mean change in activities of daily living measured using the Barthel index from baseline to one month was 3.00 points in the neurotomy and intensive SLT group compared with −1.20 points in the control group (group difference 4.12, 95% CI 0.23 to 8.00, P=0.04) (table 2 and supplementary figure S6). The mean group difference in Barthel index scores at three days and six months was −1.28 (95% CI −3.95 to 1.39) and 3.93 (0.38 to 7.49), respectively. For post-stroke depression measured using the stroke aphasic depression questionnaire-hospital version, the mean decrease from baseline to six months was 1.64 points in the neurotomy and intensive SLT group and 0.43 in the control group (group difference −1.28, 95% CI −2.01 to −0.54, P=0.001) (table 2 and supplementary figure S7). The mean group difference at three days and one month was 0.55 (−0.37 to 1.46, P=0.24) and −0.60 (95% CI −1.66 to 0.47, P=0.27), respectively.

Functional MRI

To elucidate the roles for language processing of detected regions, we used the dual stream model proposed by Hickok and Poeppel282930 (top images in figure 4 show the ventral and dorsal pathways).

Fig 4

Dynamic changes in correlation map between BNT score and brain activation during overt picture naming in C7 neurotomy plus intensive SLT group. Top images are schematic of proposed bilateral pathway for language processing based on dual stream model by Hickok and Poeppel.282930 The dorsal stream (green) projects from superior temporal gyrus towards supramarginal gyrus, inferior frontal gyrus, and regions in motor pathways, whereas the ventral stream (brown) projects towards anterior middle and inferior temporal gyrus and inferior front gyrus. The BNT score in the neurotomy plus intensive SLT group correlated with brain activation in right supramarginal gyrus and right inferior frontal gyrus at one month, and with right supramarginal gyrus and left inferior temporal gyrus at six months. No significant cluster was observed at baseline in either group. The probability threshold for all images was set at P<0.005 and cluster corrected at P<0.05. BNT=Boston naming test; C7 neurotomy=neurotomy of the seventh cervical nerve at the intervertebral foramen; SLT=speech and language therapy

In the intensive SLT only group, however, such correlation was shown mostly in the left hemisphere, including left precentral gyrus, left fusiform gyrus, and right lingual gyrus at one month, and left precentral gyrus and middle frontal gyrus at six months (see supplementary table S7 and figure S8).

Safety outcomes

No severe treatment related adverse events were reported. Adverse events that were related to C7 neurotomy included decreased sensory and motor function, neuropathic pain in the right upper arm or shoulder, and increased blood pressure. Fifteen patients in the neurotomy and intensive SLT group reported neuropathic pain and used analgesics (table 3). Beyond conventional recording of the onset and severity of neuropathic pain, we also collected data on the use of analgesics in patients with neuropathic pain after surgery. Among 25 patients in the neurotomy and intensive SLT group, 15 (60%) used analgesics for a median duration of 32 days (range 14-60 days)—that is, neuropathic pain improved within two months after surgery, and no patient needed long term treatment with analgesics (see supplementary table S8). At the six month follow-up, participants no longer experienced the neurotomy related adverse events (table 3). The recorded adverse events in sensorimotor function agreed with those in a previous study.910

Comparison with other studies

Chronic aphasia after stroke is a major challenge. Although many aspects of language function can be affected, disability in naming function is one of the most difficult to improve. BNT is a common index to evaluate naming function. Intensive SLT has been widely used as a routine and accessible treatment for chronic aphasia after stroke. The percentage improvement in naming score was not, however, uniform in the study. From literature review, the total amount or intensity of SLT, duration of aphasia, and patients’ functional reserve at baseline are all potential factors that affect the final outcomes.431 In a systematic review, the mean increase in BNT score after intensive SLT of more than four hours weekly was between 2.87 points (4.8%) and 5.71 points (9.5%), depending on the total amount of treatment each week.24 Besides intensive SLT, different new treatments have been tried, including implanted or non-invasive transcranial brain stimulation, computer based rehabilitation, and motor language therapy.32333435 A study in 2012 found that implanted epidural cortical stimulation and aphasia therapy for six weeks improved BNT score by 8.25 points (13.8%),32 whereas a study in 2021 found a mean increase of 6.7 points (11.7%) after non-invasive intermittent theta burst stimulation for three weeks.34

In this study, we reported a new treatment, C7 neurotomy plus intensive SLT, and validated its effectiveness. The improvement in BNT score was 11.16 points (18.6%) at one month after neurotomy plus intensive SLT, compared with 2.72 points (4.5%) after intensive SLT only. Treatment effects were evident at three days after surgery, before the initiation of intensive SLT, and remained stable at six months’ follow-up. Therefore, this study proposes a new and effective treatment for chronic aphasia after stroke. Although C7 neurotomy is a surgical intervention, it did not result in any additional severe adverse events or lasting troublesome symptoms. From the timing of each follow-up point, it could be concluded that surgery contributed to about half of the increase in language function and intensive SLT for three weeks to the other half, and most of the gains in language function were maintained during six months of follow-up. As a comparison, patients in the control group who only received three weeks of intensive SLT showed significantly less improvements in language function. This suggests that C7 neurotomy can not only improve language function directly but also promote the effects of intensive SLT. As improvement started as early as the first postoperative day in most patients, it could be speculated that neurotomy of the seventh cervical nerve triggered changes in plasticity of the brain regions responsible for language.

C7 neurotomy could rapidly improve spasticity of the right arm. Could this decrease of spasticity play a role in improving language function in patients with aphasia? We did a subgroup analysis to compare the improvements in BNT score between 21 patients with spastic arm paralysis and four patients without spastic arm paralysis in the neurotomy plus intensive SLT group and found no significant interaction effect. Although the number of patients without spastic arm paralysis was relatively small, leading to a wide confidence interval and diminished power, it at least indicated that release of spasticity was not the major reason for improvements in language function.

To elucidate the role of brain plasticity in the improvement of naming ability, we obtained three functional MRI scans, at baseline and at one month and six months after surgery. In the neurotomy plus intensive SLT group, BNT score correlated with brain activation in the right supramarginal gyrus and right inferior frontal gyrus at one month, and with right supramarginal gyrus and left inferior temporal gyrus at six months. The compensatory role of the right supramarginal gyrus, corresponding to Wernicke’s area of the left hemisphere, for aphasia has been reported previously.3637 Left inferior temporal gyrus and right inferior frontal gyrus also showed activation during recovery. Since the patients’ language function improved within three days after C7 neurotomy, we believe that brain plasticity also occurs immediately after surgery. Therefore, C7 neurotomy induced plasticity seems to be a dynamic process36383940 (fig 4). Further experiments of day-by-day observations are still needed to validate the dynamic plasticity process from immediately after surgery to the plateau period of more than one year. A previous study9 found that rewiring the C7 nerves of patients with spastic hand after stroke resulted in regained function of neuroplasticity in the uninjured sensorimotor area. The related editorial commented that “This is creative use and represents a fresh approach.”41 In another study, right C7 nerve root neurotomy led to rapid and massive changes in the right dorsal ganglion,42 an extension of the central sensorimotor area. In this study, C7 neurotomy induced neuroplasticity in the uninjured area to regain function in patients with aphasia after stroke. Integrating these results, we can assume that “rewiring of the peripheral nervous system, harnessed neuroplasticity, and restored central nervous system”43 can be expanded beyond sensorimotor disorders alone.

Strengths and limitations of this study

A strength of this trial was its rigorous design as we proposed a new treatment for chronic aphasia after left hemispheric stroke—right C7 neurotomy at the intervertebral foramen combined with intensive SLT. Besides evaluating language function, we also provided data on quality of life from patient self-report and depression after stroke from a close relative or care giver, which showed significant improvements six months after surgery. This study also provides important evidence for changes in the brain using functional MRI scanning after intervention, which indicated that peripheral nerve transection could induce large plasticity in the language network and activate spared areas in the right hemisphere.51335

The main limitation of this study was the inclusion of only native Chinese speakers, who were all recruited from the east coast of mainland China. An international multicentre trial recruiting participants who speak other languages would be important in further studies. Although the criteria for eligibility to take part in the study covered more than 60% of the overall numbers of patients with aphasia after stroke and did not set restrictions on sex or type of stroke, the participants were relatively young and mostly men with a higher percentage of haemorrhagic stroke compared with the overall population of stroke. Although follow-up for six months is considered long enough for language function, an extended follow-up study is still needed to verify patients’ performance over a longer time scale. Therefore, we plan to observe the patients in this study for five years to evaluate their language performance. Functional MRI in this study revealed dynamic patterns of brain plasticity, but follow-up at different time points and additional detection methods are still needed to further explore the underlying mechanisms.

Conclusions

Right C7 neurotomy at the intervertebral foramen plus intensive SLT is a superior treatment for chronic aphasia after stroke compared with intensive SLT alone. Moreover, patients in the neurotomy plus intensive SLT group showed improvements in quality of life and depression after stroke. Therefore, neurotomy plus intensive SLT could be considered an evidence based intervention for patients aged 40-65 years with aphasia for more than one year after stroke.