The rabbits were restrained supine, on a heated table. Under general anesthesia, via a medial cervicotomy, a tracheostomy was performed and the trachea was cannulated with a 4 mm ID endotracheal tube. A #00 Fleisch pneumotachograph (Lausanne, Switzerland) was connected to the endotracheal tube and to a transducer (Statham PM 197, range ± 0.01 PSI; Oxnard, CA). A balloon catheter 50 mm long (Atlan, 4.0 mm external diameter, 2.6 mm internal diameter (Marquat Genie Biomedical, Boissy Saint Leger, France) was introduced transorally into the inferior part of the thoracic esophagus and was connected to a pressure transducer (Statham PM 6, range ± 2.5 PSI) 10. Midline laparotomy was then realized. The two external jugular veins were isolated in the neck. Two cardiac stimulation electrodes (1.5 mm diameter; four electrodes 2.0 mm high, 10.0 mm between each electrode, ref 002943, Bard, Trappes, France) were introduced into both external jugular veins 10. The electrodes were advanced 2 – 4 cm into the upper chest in order to stimulate each phrenic nerve without foreleg contraction. Electrical stimulation was 0.5 s long train of rectangular pulses with pulse duration of 0.2 ms at 100 Hz. Intensity used was always superior (1.25 time) to the supra maximal intensity based on Pes amplitude and was used for all subsequent stimulations. Pairs of home made hocked wire (10 mm apart) electrodes were inserted into the sternal, midcostal and posterior costal regions of both hemidiaphragms. EMG signals were band-pass filtered (2–20 kHz) and amplified with a recorder (Viking, Nicolet, Madison, WI). An integrator was connected (Gould, Instrument System, Valley View, OH) in order to quantify the EMG signals. Length changes of the diaphragm muscle segments were measured during breathing using sonomicrometry. Sonomicrometer (sonomicrometer 120, Triton Technology Inc, San Diego, CA) measured the distance between pairs of small transducers implanted in muscles or similar tissues. The distance was determined by measuring the transit time of ultrasound between the pair of transducers. The time was converted to an equivalent distance. Via the laparotomy, pairs of piezoelectric crystals (Bioseb, segment length small 1.0 mm, 5 MHz, Chaville, France) were positioned 5 and 15 mm apart in sternal, mid-costal and posterior costal regions of both hemidiaphragm. The crystals were carefully aligned along the longitudinal axis of muscle fibers and held in place with purse-string sutures.

All parameters were displayed on an Apple computer using an acquisition card (MacLab/8e) and Chart V.3.4.4 software.

The mean denervation score was equal to zero in the healthy group. The highest score of denervation was in group II (denervated group). Among the six denervated animals in group II, three had a severe denervation, whereas three had a moderate denervation. In the three reinnervated groups, no statistical differences were found between the three reinnervated groups and GI, the denervation scores were lower than in group II (denervated group). In groups III (PN – Abd) and IV (PN – ILN), the denervation scores were higher in the right sternal region.

Means and standard deviations are displayed in the Table I. During quiet breathing, no statistical difference was found between the groups. During right supramaximal stimulation, Pes was near to nul in group II (denervated group). In GIV and GV, Pes during right phrenic nerve stimulation were lower than in group I (p < 0.05) and not different between GI and GIII. During left stimulation, the lowest values were observed in the groups I (healthy group), III (PN – Abd) and V (r.C5-Abd + r.C4-STn). In contrast, Pes was higher in the group IV and in group II compared to GI (p < 0.05). During bilateral supramaximal stimulation, Pes was lower in GII and GV compared to GI (p < 0.05) and not different between GI, GIII and GIV.

In the three reinnervated groups, sonomicrometric values were positive (i.e diaphragmatic muscular fiber shortening) during quiet breathing, prolonged tracheal occlusion or phrenic nerve stimulations and were negative (i.e. diaphragmatic muscular fiber lengthening) in denervated group (GII) except in one animal (Fig. 5). During quiet breathing and tracheal occlusion, muscular fiber shortenings in the three reinnervated groups were lower than in group I (healthy group) but without any significant difference. No difference was observed between reinnervated groups. During right phrenic nerve stimulation, shortenings in groups IV and V were statistically lower than in group I (healthy group), and statistically not different between GI and GIII. During bilateral phrenic nerve stimulation, right fiber shortenings in the mid-costal region were lower in groups IV (but did not reach statistical difference) and V and values of groups I and III were relatively equal. Only one rabbit in the group II showed shortening during prolonged tracheal occlusion. The same animal showed EMG activity during tracheal occlusion. However, in this animal, right phrenic nerve stimulation provided no right hemidiaphragm contraction.

Due to its respiratory similarities with the phrenic nerve (its activity begins a few milliseconds before the phrenic nerve discharge (between 40 et 80 ms) 14 and is increase during hypercapnia and hypoxia 6), the inferior laryngeal nerve was chosen to perform diaphragm reinnervation. The first diaphragm reinnervation was usefully done with the vagus nerve in dogs and in rats 13. In 1960, Guth et al., were the first to use the inferior laryngeal nerve as a donor nerve 4 in 8 rats and 3 monkeys, with positive results in two rats and one monkey. In 1993, Baldissera et al. 5 used the inferior laryngeal nerve or its branches as donor nerves in 10 cats, but analysis of their results was quite difficult due to the limited number of animals used in each group. In contrast to our study, Baldissera et al. did not use the abductor branch alone to perform the reinnervation of a complete hemidiaphragm 5. Some authors 2 used nerves without any respiratory activity i.e. the facial, the accessory and the long thoracic nerve, with poor results. Since, the stimulation of the transposed nerve induces a contraction of the reinnervated hemidiaphragm, these nerves were unable to induce spontaneous diaphragmatic rhythmic contractions.

Our aims were to study diaphragmatic effects of diaphragmatic reinnervation by the inferior laryngeal nerve or its branch and with the nerve of the sternothyroid muscle originating from the hypoglossal nerve, using EMGs, histology and transdiaphragmatic pressure, near as possible to diaphragmatic explorations in humans. Our approach was therefore indirect regarding muscle function, and especially pressure measurements, which test the overall diaphragmatic function. The muscle fibers approach, even if it could bring additional data, was therefore not retained in our study.

In our study, as the abdomen was open, gastric pressure was considered to be identical to atmospheric pressure. Consequently, esophageal pressure was equal to the arithmetic inverse of transdiaphragmatic pressure and was considered as a measure of the diaphragmatic force 1115 and was measured during airway occlusion, under isometric conditions 11. Nevertheless, the pressure that the abdominal wall and content exerts on the diaphragm during inspiration has an impact on the thoracic mechanics and therefore on the pressure generated by the diaphragm contraction. Performing all esophageal measurements with an opened abdominal cavity may have altered the pressure generating capacity of the diaphragm and could explain the lack of significance during bilateral phrenic nerve stimulation.

Transdiaphragmatic pressure is measured during uni or bilateral phrenic nerve stimulation is not different using transvenous stimulations or direct stimulation of the nerve in the neck 10. We chose supramaximal tetanic stimulation at 100 Hz frequency as previously performed in rabbits 101617. However, an unequal distribution of the reinnervation might have induced differences between the strength of the different diaphragmatic portions 13.

Sonomicrometry provided reliable, accurate and objective values of changes of the muscular fiber length 1819. In the posterior costal region, measurements were considered unreliable, including measurements of the control group. Measurements were always performed at the end of the evaluation because of the high risk of pneumothorax. With sonomicrometry, the result during quiet breathing and prolonged tracheal occlusion were quite similar with the results of Zhan et al. 19.

Our results demonstrated that hemidiaphragm reinnervation by inferior laryngeal nerve or its branch restored a diaphragm function near to normal regarding EMGs, transdiaphragmatic pressure, sonomicrometry and histology. In the three reinnervated groups, a good restitution of the electrical inspiratory activity in paralyzed hemidiaphragm was found. During quiet breathing, almost all reinnervated animals had a poor and neurogen activity in the three studied regions. However, during prolonged tracheal occlusion, the electrical activity increased just as far as in the healthy group. In the reinnervated animals, EMG activity was better improved in mid and posterior costal regions. Those results indicate that diaphragmatic respiratory drive was restored using the inferior laryngeal nerve or its branch, but incompletely during quiet breathing. It could be explained by a partial recruitment of the transferred nerve, completed during inspiratory effort 20. As regards EMG, sonomicrometric measurements were better in the mid-costal region, during quiet breathing and tracheal occlusion. These results are in agreement with histological examination which found more severe denervation injuries in the sternal region in the groups III (PN – Abd) and IV (PN – ILN).

Six rabbits showed expiratory activity, three from group IV (PN – ILN), two from group III and one from group V. We expected this type of activity in group IV, but in the two other groups it was less obvious. However, if the abductor branch contains a great majority of inspiratory axons, it can sometimes contain few expiratory axons for the interarytenoid muscle 21. Expiratory activity was poorer in the group III (PN – Abd) compared with the group IV (PN – ILN) because expiratory axons are less numerous than in the adductor branch. An expiratory activity can also be produced from inspiratory axons under definite physiologic circumstances (coughing, phonation) and vocal cords stabilization in expiration 2223.

Among the six denervated animals, only one did not show complete denervation. In fact, during prolonged tracheal occlusion in this animal, residual inspiratory activity and fiber shortening were still present in the right sternal and mild-costal regions. Furthermore, histological examination did not reveal severe but only slight denervation. However, dissection of the cervical region did not permit to identify the origin of the innervation. Partial denervation or spontaneous reinnervation through the nervous section should not be considered since the stimulation of the right PN was ineffective. Two other hypotheses could be suggested: 1- spontaneous reinnervation by intercostal nerves, 2- cross innervation. Cross innervation of the diaphragm remains a subject of controversy. Results differ between authors and between animal species. In the rabbit, Rikard-Bell and Bystrzycka 24 did not observe any controlateral retrograde labelling in the cervical spine. Marie et al. 13 reported the same conclusion with functional tests. In contrast, in cats, spontaneous diaphragmatic reinnervation from left phrenic nerve has been reported 2526 but not confirmed 2728. This phenomena has also been reported in rats and monkeys 4.

The decrease in pressure generating capacity of the diaphragm may have been partly explained by the geometrical configuration of the diaphragm of the animals and might depend of the reinnervated hemidiaphragms associated to the healthy one (left hemidiaphragm) to generate negative esophageal pressure 29. In groups II and IV, left phrenic nerve stimulation induced higher transdiaphragmatic pressure, and could have explained that it was higher during tracheal occlusion in group III and IV, and during bilateral phrenic nerve stimulation (not significant) (table I). Those results remain unclear, but could be interpreted, either as a compensation of the left hemidiaphragm to the decreased force of the right hemidiaphragm, or as a consequence of a difference of geometry in the remaining diaphragm 29.

Those compensatory mechanisms are also suggested by EMGs and sonomicrometry. Among the three reinnervated groups, global EMG in the left hemidiaphragm was poorer in group III (PN – Abd), suggesting that among the three reinnervated group, the compensation in group III was lower. Also, sonomicrometry during quiet breathing, during prolonged tracheal occlusion and during right stimulation showed that the contractility of the right hemidiaphragm was better in group III. In this group, shortening was higher and similarly equal with results of the healthy control group (group I).