Pulmonary vagus nerve transection for chronic cough after video-assisted lobectomy: a randomized controlled trial

Abstract

Background:

Chronic cough is common after lobectomy. Vagus nerves are part of the cough reflex. Accordingly, transection of the pulmonary branches of vagus nerve may prevent chronic cough. And there are no clear recommendations on the management of the pulmonary branches of vagus in any thoracic surgery guidelines.

Methods:

This is a single-center, randomized controlled trial. Adult patients undergoing elective video-assisted thoracoscopic lobectomy and lymphadenectomy were randomized at a 1:1 ratio to undergo a sham procedure (control group) or transection of the pulmonary branches of the vagus nerve that innervate the bronchial stump plus the caudal-most large pulmonary branch of the vagus nerve. The primary outcome was the rate of chronic cough, as assessed at 3 months after surgery in the intent-to-treat population.

Results:

Between 1 February 2020 and 1 August 2020, 116 patients (59.6±10.1 years of age; 45 men) were randomized (58 in each group). All patients received designated intervention. The rate of chronic cough at 3 months was 19.0% (11/58) in the vagotomy group versus 41.4% (24/58) in the control group (OR=0.332, 95% CI: 0.143–0.767; P=0.009). In the 108 patients with 2-year assessment, the rate of persistent cough was 12.7% (7/55) in the control and 1.9% (1/53) in the vagotomy group (P=0.032). The two groups did not differ in postoperative complications and key measures of pulmonary function, for example, maximal voluntary ventilation, diffusing capacity of the lungs for carbon monoxide, and forced expiratory volume.

Conclusion:

Transecting the pulmonary branches of vagus nerve that innervate the bronchial stump plus the caudal-most large pulmonary branch decreased the rate of chronic cough without affecting pulmonary function in patients undergoing video-assisted lobectomy and lymphadenectomy.

Introduction

Highlights

• This is the only trial, either already reported or registered, that uses a sham control procedure to examine the potential efficacy of vagotomy.

• This trial showed transection of the vagal nerves that innervate the resected lobe is safe and could prevent chronic cough in patients undergoing lobectomy and lymphadenectomy.

• Finally, we also provide a preliminary overview of the surgical anatomy.

Chronic cough is common after lung surgery. The estimated rate of chronic cough is 25–50% in patients undergoing surgical resection of lung cancer and lymphadenectomy¹. In some patients, chronic cough may persist for years^2,3. Currently available treatments, including suplatast tosilate², proton pomp inhibiters in combination with prokinetic agent⁴, corticosteroid in combination with β-agonist⁵ and a P2X3 antagonist gefapixant⁶, are nonspecific and have limited efficacy.

Given that the accumulation of acidic substances in the inflammatory reaction can directly stimulate the C fibers of the vagus nerve, excite the cough center, and induce bronchoconstriction and cough⁷. During lobectomy under nonintubated anesthesia, the anesthesiologist inhibit cough by intrathoracic block of the vagus nerve⁸. Similarly, the airflow 2 study have demonstrated that targeted pulmonary denervation for chronic obstructive pulmonary disease is safe and effective⁹. Previous studies reported that the main independent risk factors for chronic cough after lung surgery consist of mediastinal lymph node removal (especially for subcarinal lymph nodes)¹⁰ and bronchial morphological alterations (including bronchial stump)¹¹.

We hypothesized that transecting the pulmonary branches of vagus nerve that innervates the bronchial stump plus the caudal-most large pulmonary branch of the vagus nerve (traveling along the subcarinal lymph nodes) could prevent chronic cough in patients undergoing video-assisted thoracoscopic (VATS) lobectomy and lymphadenectomy, and conducted a randomized, sham-controlled trial to test this hypothesis.

Materials and methods

Study design

This is a single-center, randomized controlled trial. Eligible subjects were randomized at a 1:1 ratio to receive vagotomy (transection of the pulmonary branch of vagus nerve) or sham procedure (dissection but not transection of the pulmonary branch of vagus nerve) during surgery.

Ethics

This clinical trial followed the principles of the Declaration of Helsinki and Consolidated Standards of Reporting Trials (CONSORT) Guidelines¹² (Supplemental Digital Content 1, http://links.lww.com/JS9/B607). The protocol was approved by the Ethics Committee of Authors’ hospital. Informed consent was obtained prior to enrollment from all participants. The trial is registered at ClinicalTrials. The authors vouch of the completeness and accuracy of the data and analysis.

Patients

Adult patients scheduled to undergo elective VATS lobectomy and lymphadenectomy were eligible. For inclusion, preoperative radiographic examination must have revealed peripheral bronchial lung cancer with 5 cm maximum diameter or less (T≤2), no involvement of regional, ipsilateral peribronchial and ipsilateral hilar lymph nodes (including direct extension to intrapulmonary nodes) (N≤1), and no evidence of distant metastasis (M0). Key exclusion criteria included: 1) chronic bronchitis, asthma, gastroesophageal reflux disease or postnasal drip syndrome at the baseline; 2) use of angiotensin converting enzyme inhibitors; 3) severe arrhythmia; and 4) previous thoracic surgery.

Methods

Written informed consent was obtained within 24 h before surgery in potentially eligible patients. Randomization was conducted at the time when intraoperative pathology indicated a need for lobectomy and lymphadenectomy. Randomization was conducted with a computer-generated random numbers table. Concealment was conducted using sealed, opaque envelopes, which were opened by the surgeons at the time of the operation. Patients and outcome evaluators were blinded to the group assignment.

Typical anatomy of the vagus nerve

The vagus nerve supplies the lungs via an extensive nerve plexus posterior to the main bronchus. The main trunk of the vagus nerve can be readily distinguished after opening the posterior mediastinal pleura. Several branches arise from the main trunk to form the posterior pulmonary plexus¹³. Large pulmonary branches of the vagus nerve travel along the bronchial arteries.

Three main pulmonary branches arise from the vagus nerve trunk on the right side. The pulmonary branch that innervates the right upper lung is clearly visible along the upper lung bronchus. The second pulmonary branch travels along the intermediate bronchus, and diverges into two smaller branches to innervate the middle and lower lobes, respectively. The third caudal-most large pulmonary branch travels along the subcarinal lymph nodes to innervate the hilum as well the lower lobe (Fig. 1).

Figure 1.

Schematic drawing of the dorsal distribution of the right vagus nerve at the hilum. Eso, esophagus; T, trachea; V, vagus nerve.

The innervation to the left upper lobe consists of several branches that travel cross the pulmonary artery or between the pulmonary artery and the left main bronchus that run along the upper edge of the left main bronchus, behind the pulmonary artery. The pulmonary branch that innervates the lower lobe travels along the lower margin of the left main bronchus and thus is readily distinguishable from the branch that innervates the upper lobe. The caudal-most large pulmonary travels along the subcarinal lymph nodes to innervate the hilum as well the lower lobe (Fig. 2).

Figure 2.

Schematic drawing of the dorsal distribution of the left vagus nerve at the hilum. AO, aorta; LPA, left pulmonary artery; LPV, left pulmonary vein; T, trachea; V, vagus nerve.

Intervention

After general anesthesia, the patient was placed in a contralateral supine position. All patients were intubated with double-lumen tubes of appropriate size under video laryngoscope (for details, see Supplement 1, Supplemental Digital Content 2, http://links.lww.com/JS9/B608). Surgery was conducted as previously reported¹⁴ and all procedures were completed by the same surgical team. Mediastinal lymph nodal dissection covered groups 2, 3, 4, 7, 8, 9, 10, and 11 for lung cancer on the right side, and groups 4, 5, 6, 7, 8, 9, 10, and 11 for lung cancer on the left side. The lymph node stations were defined according to the eighth edition of the TNM classification for lung cancer¹⁵.

For vagotomy, pulmonary branches of vagus nerve that innervate the bronchial stump was transected to block afferent input. The caudal-most large pulmonary branch was transected to block stimuli from the subcarinal residual cavities regardless of the lesioned lobe (Fig. 3). For sham operation, pulmonary branches that innervate the bronchial stump as well as caudal-most large pulmonary branch were dissected but not transected. For left upper lobectomy, the branches between the pulmonary artery and the left main bronchus were not transected in the sham group, but the small branches that cross the pulmonary artery to innervate the upper lobe were transected in both groups for anatomical reasons.

Figure 3.

Schematic drawing of pulmonary vagus nerve transection (using right upper lobectomy as an example). (A) Transecting the vagal branch innervating the bronchial stump and the caudal-most large pulmonary branch; (B) Closing the bronchus of the lesioned lobe by the bronchial closure; (C) The final state after intervention.

Dissection and transection of the vagus nerve were aborted upon emerging arrhythmia that caused hemodynamic instability during the maneuver. After the operation, a closed thoracic drainage tube was routinely placed through the observation hole, and the patient was returned to the ward or ICU, as appropriate. Postoperative care included continuous ECG monitoring, noninvasive arterial blood pressure, finger pulse oxygen, and continuous low flow oxygen. Analgesics were used based on the discretion of the attending physician.

Following-up

The primary outcome was the rate of chronic cough, as assessed at 3 months after surgery and calculated in the intent-to-treat dataset. The frequency and severity of cough during hospitalization were assessed using a three-point cough symptom score (CSS)^16–18. Follow-up after discharge was conducted in an outpatient clinic once every month for 3 months, and every 12 months afterward. The 1-month follow-up included a chest CT scan. The 3-month follow-up included a chest CT scan, daytime and nighttime CSS, and evaluation of chronic cough by a respiratory specialist using a validated Mandarin Chinese version of the Leicester Cough Questionnaire (LCQ-MC)^19–21. During each of the follow-up, the use of ACEIs were inquired. All other complications were recorded and classified according to the Clavien–Dindo classification²². Major complications are defined in the e-Methods in Supplement 2 (Supplemental Digital Content 3, http://links.lww.com/JS9/B609). Patients were required to complete a pulmonary function test at 30 months.

Statistical analysis

Sample size calculation was conducted using un-pooled Z-test based on the following assumptions: 1) 50% chronic cough in the sham control group⁴; 2) a clinically meaningful reduction to 25% chronic cough in the vagotomy group; 3) α at 0.05, and power (1 - β) at 0.80. The calculation yielded 110 patients (55 in each group). Expecting 5% dropout, we planned to enroll 116 patients.

Continuous variables were analyzed using Student’s t-test if normally distributed (as determined by the Shapiro–Wilk Test), and using Mann–Whitney U test otherwise. Categorical variables were analyzed using χ² or Fisher exact test, as appropriate. Analysis of the primary outcome (chronic cough) followed the intent-to-treat principle: patients lost to follow-up were assumed to have chronic cough to avoid exaggerating group difference. CSS was analyzed by the Wilcoxon rank-sum test. P<0.05 (two-sided) was considered statistically significant. All statistical analyses were conducted using SPSS24.

Results

One hundred sixteen patients were randomized (Fig. 4). All 116 patients received the assigned intervention (58 in each group). Demographics and baseline characteristics are shown in Table 1. Perioperative conditions did not differ significantly between the two groups (Table 2). Additionally, there were no significant differences in factors that are relevant to, or reflect injury caused by traumatic/difficult intubation (e-Table 1 in Supplement 2, Supplemental Digital Content 3, http://links.lww.com/JS9/B609). The follow-up was expected to be completed by September 2022, but was actually completed in February 2023 due to the impact of the coronavirus pandemic. Three patients were lost to follow-up after discharge, and the remaining 113 patients completed the 3-month follow-up.

Figure 4.

Patient flow through the trial.

Table 1.

Demographic and clinicopathologic characteristics of the patients.

	Control, n=58	Vagotomy, n=58
Male, N (%)	27 (46.6)	18 (31.0)
Age (year), mean (SD)	59.2 (11.0)	60.0 (9.3)
BMI (kg/m2), mean (SD)	24.0 (2.9)	24.0 (3.3)
Past and current smokers, N (%)	11 (19.0)	13 (22.4)
Left side, N (%)	35 (60.3)	36 (62.1)
Pathology, N (%)
Adenocarcinoma	48 (82.8)	47 (81.0)
Squamous cell carcinoma	4 (6.9)	7 (12.1)
Other	6 (10.3)	4 (6.9)
ACCI, median (IQR)	4 (3–5)	4 (3–4.3)

ACCI, age-adjusted Charlson Comorbidity Index; IQR, interquartile range.

Table 2.

Perioperative conditions between the two groups.

	Control, n=58	Vagotomy, n=58	P
Operation duration (min), median (IQR)	145.0 (130.0–170.8)	155.0 (120.0–193.8)	0.225
Blood loss (ml), median (IQR)	100.0 (50.0–100.0)	100.0 (50.0–100.0)	0.950
One-lung ventilation time (min), median (IQR)	134.5 (120.0–170.0)	155.0 (121.3–184.0)	0.129
Number of lymph nodes removed, median (IQR)	9 (7–12.25)	9 (6–12)	0.714
Drainage on the first day, mean (SD)	132.5 (59.9)	135.5 (51.4)	0.774
Postoperative hospital stay (day), median (IQR)	6.0 (5.0–7.0)	6.5 (5.0–9.0)	0.107

IQR, interquartile range.

The rate of chronic cough, as assessed at 3 months after surgery, was 19.0% (11/58) in the vagotomy group versus 41.4% (24/58) in the sham control group (OR=0.332, 95% CI: 0.143–0.767; P=0.009; Table 3). The vagotomy group also had lower CSS and total LCQ-MC score (Table 3). The rate of acute cough was 62.1% (36/58) in the vagotomy group versus 77.6% (45/58) in the sham control group (P=0.069). In comparison to the control group, the vagotomy group had lower nighttime CSS for acute cough (P=0.028) but comparable daytime CSS (P=0.070).

Table 3.

The rate and severity of postoperative cough.

	Control	Vagotomy	P
3 months	N=58	N=58
N (%)	24 (41.4)	11(19.0)	0.009
Daytime CSS	N=57	N=56	0.005
0	34 (59.6)	47 (83.9)
1	18 (31.6)	7 (12.5)
2	3 (5.3)	1 (1.8)
3	2 (3.5)	1 (1.8)
Nighttime CSS	N=57	N=56	0.030
0	50 (87.7)	55 (98.2)
1	6 (10.5)	1 (1.8)
2	1(1.8)	0
3	0	0
LCQ-MC, median [IQR]	21.0 [17.7–21]	21 [21–21]	0.005
1 year	N=55	N=53
N (%)	11 (20.0)	3 (5.7)	0.027
Daytime CSS	N=55	N=53	0.029
0	44 (80.0)	50 (94.3)
1	9 (16.4)	2 (3.8)
2	1 (1.8)	1 (1.8)
3	1 (1.8)	0
Nighttime CSS	N=55	N=53	0.087
0	48 (87.3)	51 (96.2)
1	4 (7.3)	2 (3.8)
2	2 (3.6)	0
3	1 (1.8)	0
2 years	N=55	N=53
N (%)	7 (12.7)	1 (1.9)	0.032
Daytime CSS	N=55	N=53	0.032
0	48 (87.3)	52 (98.1)
1	7 (12.7)	1 (1.9)
2	0	0
3	0	0
Nighttime CSS	N=55	N=53	0.032
0	48 (87.3)	52 (98.1)
1	7 (12.7)	1 (1.9)
2	0	0
3	0	0

Cough symptom score CSS, 0=no cough; 1=cough more than one short period; 2=frequent cough that mildly interfered with usual daytime activities or sleep in the night; 3=distressing cough most of the day or night that severely interfered with usual daytime activities or sleep in the night; IQR, interquartile range; LCQ-MC, Mandarin Chinese version of the Leicester Cough Questionnaire.

We have now completed assessment of chronic cough at 1 and 2 years in 108 patients (not part of the trial protocol). At 1 year, the rate of persistent cough was 5.7% (3/53) in the vagotomy group versus 20.0% (11/55) in the sham control group (P=0.027; Table 3). The vagotomy group had lower daytime CSS (P=0.029) but similar nighttime CSS (Table 3). At 2 years, the rate of persistent cough was 1.9% (1/53) in the vagotomy group versus 12.7% (7/55) in the sham control group (P=0.032). The vagotomy group had lower CSS (daytime: P=0.032; nighttime: P=0.032).

The overall incidence of postoperative complications was 22.4% in the vagotomy group versus 17.2% in the sham control group (P=0.485; Table 4). The rate of new-onset atrial fibrillation was 3.5% (2/58) in the vagotomy group versus 1.7% (1/58) in the sham control group (P=0.999). In all three cases, sinus rhythm was restored with amiodarone treatment and did not recur until the last follow-up. The rate of persistent air leakage was 13.8% (8/58) in the vagotomy group versus 5.2% (3/58) in the sham control group (P=0.119). Air leakage was managed with injection of 50% glucose solution into the thoracic cavity in three patients and chest drainage tubes under local anesthesia in the remaining eight patients.

Table 4.

Postoperative complications according to Clavien–Dindo classification.

	Control, n=58	Vagotomy, n=58	P
Event number	10	13	0.485
Patient number	8	9	0.793
Grade 1
Bloating	4	2	0.675
Diarrhea	0	0	>0.999
Grade 2
New-onset atrial fibrillation	1	2	>0.999
Pneumonia	1	0	>0.999
Persistent air leaks	0	3	0.242
Chylothorax	0	1	>0.999
Grade 3
Pneumothorax after removing drainage tube	1	0	>0.999
Persistent air leaks (needing more than one chest tube)	3	5	0.714

Pulmonary function tests were conducted at 30 months in 100 patients (53 and 47 in the control and vagotomy groups, respectively, see Table 5). MVV was 81.6±17.7% in the vagotomy group versus 77.5±18.1% in the sham control group (P=0.258); DLCO was 79.7±19.0% in the vagotomy group versus 81.0±15.7% in the sham control group (P=0.708); FEV1 was 94.3±18.6% in the vagotomy group versus 93.7±22.7% in the sham control group (P=0.898). In addition to the group comparison, we also examined key baseline characteristics in the 100 patients with 30-month pulmonary function tests versus the remaining 16 patients without such data, and the results suggested no major bias (e-Table 2 in Supplement 2, Supplemental Digital Content 3, http://links.lww.com/JS9/B609).

Table 5.

Pulmonary function tests at 30 months after the operation.

	Control	Vagotomy
30 months	N=53	N=47
MVV, %predicted	77.5+18.1	81.6+17.7	0.258
DLCO, %predicted	81.0+15.7	79.7+19.0	0.708
FEV1, %predicted	93.7+22.7	94.3+18.6	0.898

Data are presented as mean+SD; DLCO, diffusing capacity of the lung for carbon monoxide; FEV1, forced expiratory volume in one second; MVV, minute ventilation volume.

Discussion

In this trial, transection of the pulmonary branches of the vagus nerve that innervate the bronchial stump plus the caudal-most large pulmonary branch resulted in a statistically significant and clinically meaningful reduction in the rate of chronic cough (from 41.4 to 19.0%). The rate of chronic cough that persisted for 1 and 2 years after surgery was also significantly lower in the vagotomy group. The two groups did not differ in either the overall rate of postoperative complications or Clavien–Dindo grade II and higher complications. Pulmonary function at 30 months were comparable between the two groups.

The incidence of chronic cough in the control group of this trial (41.4%) was similar to that reported by majority of the previous studies. In a retrospective study of 171 patients undergoing lobectomy for NSCLC in China, the rate of chronic cough that lasted for at least 8 weeks was 39.8%¹. In a Japanese cross-sectional study of 70 patients undergoing pulmonary resection, the rate of chronic cough within 1 year was 50 and 18% patients still had symptomatic cough more than 1 year after surgery⁴. In an American cross-sectional survey of 142 patients undergoing pulmonary resection of NSCLC, the rate of chronic cough at 5 years was 24.7%³. In our trial, since the incidence of chronic cough (20.0%) at one year was still high in the control group, we followed up the outcome at 2 years, which still have a significant difference between the two groups (12.7 vs 1.9%). Several other studies; however, reported lower rate of chronic cough^5,7. These discrepancies may reflect difference in patient characteristics and the definition of chronic cough. Such a difference, if indeed present, could reflect either true efficacy of vagotomy or bias introduced by confounding factors, including double-lumen tubes intubation, opioid drugs, and the duration of anesthesia⁷.

Pulmonary branches of the vagus nerve are primary afferents of the cough reflex. In anaesthetized guinea-pigs, the cough induced by mechanical stimulation to the larynx can be completely blocked by bilateral vagotomy^23,24. Intrathoracic block of the vagus nerve in patients receiving lobectomy under nonintubated anesthesia has been shown to inhibit cough induced by surgical maneuver^25,26. At a mechanistic level, TRPV1 is known to be widely distributed in afferent C fibers and some Aδ fibers of the cough reflex²⁷, and coincidentally, has been found to be upregulated in patients who developed chronic cough after lung surgery²⁸. Sympathetic nerves that co-innervate the lungs with the vagus nerve²⁹ may also play an important role in chronic cough. In a previous trial in patients undergoing lobectomy for NSCLC, inhalation of β2 receptor agonist in combination with corticosteroid decreased the severity of persistent cough after surgery⁵. The finding that vagotomy could prevent chronic cough after lobectomy in the current study is in line with the opposing role of sympathetic versus sympathetic nervous system.

The lower rate of chronic cough in the vagotomy group in our trial is in stark contrast to the results of a recent trial by Gu et al.³⁰ showing lower rate of postoperative cough in the vagus preservation group (13.9 versus 30.4% in the conventional surgery group). Several factors may have contributed to the opposing findings. First, cough was assessed within 5 weeks after surgery in the trial conducted by Gu et al. and at 3 months after surgery in the current trial. Second, patients in our trial had more advanced disease (T1-2N0-1M0 versus T1a-bN0M0 in the Gu et al. trial); accordingly, the extent of lymphadenectomy may vary considerably. Also, lymph nodes were sampled in both groups in the Gu et al. trial, but the extent of sampling was not specified in each group. In our trial, lymph nodes were systemically dissected in both groups. Finally, in the Gu et al. study, there was uncertainty about whether the pulmonary branches of the vagus nerve were preserved or severed in the conventional surgical treatment group. In our trial, the pulmonary branches were isolated regardless of the group allocation assignment.

Safety is the utmost concern in the use of vagotomy in human patients regardless of its purpose. Indeed, damage to the pulmonary branches of the vagus nerve has been shown to impair lung functions (such as the cough reflex, bronchus diameter, and mucous production) in patients undergoing esophagectomy³¹. Bilateral vagotomy has been shown to decrease pulmonary function parameters in animal studies³². Therefore, transecting wrong branches (the ones not innervate the bronchial stump) could jeopardize the ability to expectorate and comprise pulmonary function. In the current study, there was no difference in pulmonary function at 30 months after surgery between the two groups, supporting the long-term safety of the procedure. Also, the two groups did not differ in the rate of other postoperative complications that reflect unintentional injury to the vagus nerve other than the intended pulmonary branches, including pneumonia, gastrointestinal disturbances, and atrial fibrillation, thus supporting the safety profile of the procedure. Accordingly, accurately identifying the vagus nerve branches that innervate the bronchial stump is of paramount importance and we describe the anatomic variations in Supplement 2 (Supplemental Digital Content 3, http://links.lww.com/JS9/B609) (e-Discussions).

Majority of prior anatomical studies focused on practical aspects to guide esophagectomy^13,31. Landmarks that could be used to transect pulmonary branches to specific lobes of the lungs are not available in the existing literature. In an attempt to ensure technical success and accuracy of vagotomy in the future, we summarize the key observations in this trial in e-Discussions in Supplement 2 (Supplemental Digital Content 3, http://links.lww.com/JS9/B609).

This trial has several limitations. Firstly, this is a prospective, single-center study with a relatively small sample size, meaning there may be bias in the results. Second, the efficacy and safety of vagotomy are dependent on familiarity of the operating surgeons with relevant anatomy and surgical experience. As a result, multicenter trials with standardized surgical protocol are needed to verify our results.

Conclusion

In summary, transection of special pulmonary branches of vagus nerve could reduce the rate of chronic cough in patients undergoing VATS lobectomy and lymphadenectomy. This study also provided a preliminary overview of surgical anatomy and an appropriate basis for the management of the pulmonary branches of vagus for use in lobectomy.

Ethical approval/Declaration of Helsinki

This trial was approved by the Clinical Research Ethics Committee of Affiliated Hospital of Xuzhou Medical University (XYFY2019-KL179-01) and was conducted in accordance with the ethical standards of the Helsinki Declaration by the World Medical Association.

Patient consent

The patients provided their informed consent to participate in the study. The manuscript do not have identifying characteristics of patients.

Sources of funding

This work was supported by the Sci-Tech Innovation 2030 (2021ZD0203100), the National Natural Science Foundation of China (81720108013 and 82130033), the Bethune Charitable Foundation Surgical Excellence Key Program (2021-C-3), the Social Development Projects of Key R&D Programs in Xuzhou city (KC22097).

Author contribution

Q.Z.: conceptualization, formal analysis, investigation, and writing – original draft; Y.G.: investigation, formal analysis, visualization, and writing – original draft; T.S.: conceptualization, methodology, investigation, and visualization; S.F.: data curation, investigation, and supervision; C.Z.: data curation and supervision; T.H. and X.L.: investigation; Y.H.: conceptualization, formal analysis, writing – review and editing; J.-L.C.: conceptualization, supervision, funding acquisition, writing – review and editing; H.Z.: conceptualization, visualization, funding acquisition, writing – review and editing.

Conflicts of interest disclosure

The authors declare no conflicts of interest in this work.

Research registration unique identifying number (UIN)

1. Name of the registry: ClinicalTrials.

2. Unique identifying number or registration ID: NCT04247997.

3. Hyperlink to this clinical trial registration: https://www.clinicaltrials.gov/study/NCT04247997?cond=NCT04247997&rank=1.

Guarantor

Hao Zhang.

Data availability statement

Supplementary data and study protocol will be available online to others upon publication of this study. The participant data are available from the corresponding author upon request.

Provenance and peer review

Not commissioned, externally peer-reviewed.