Article Text

Download PDFPDF

Unilateral focal lesions in the rostrolateral medulla influence chemosensitivity and breathing measured during wakefulness, sleep, and exercise
  1. M J Morrella,
  2. P Heywoodb,
  3. S H Moosavia,
  4. A Guza,
  5. J Stevensc
  1. aNational Heart and Lung Institute, Imperial College School of Medicine, Charing Cross Campus, St Dunstan’s Road, London W6 8RP, UK, bDepartment of Neurology, Frenchay Hospital, Frenchay Park Road, Bristol BS16 1LE, UK, cDepartment of Radiology, St Mary’s Hospital, Praed Street, London W2, UK
  1. Dr M J Morrell, National Heart and Lung Institute, Imperial College School of Medicine, Royal Brompton Hospital, Sleep and Ventilation Unit (South block), Sydney Street, London SW3 6NP, UK. Telephone 0044 171 352 8121 ext 4023; fax 0044 171 351 8911; email m.morrell{at}


OBJECTIVES The rostrolateral medulla (RLM) has been identified in animals as an important site of chemosensitivity; in humans such site(s) have not been defined. The aim of this study was to investigate the physiological implications of unilateral lesions in the lower brainstem on the control of breathing.

METHODS In 15 patients breathing was measured awake at rest, asleep, during exercise, and during CO2 stimulation. The lesions were located clinically and by MRI; in nine patients they involved the RLM (RLM group), in six they were in the pons, cerebellum, or medial medulla (Non-RLM group). All RLM group patients, and three non-RLM group patients had ipsilateral Horner’s syndrome.

RESULTS Six of the RLM group had a ventilatory sensitivity to inhaled CO2(V˙/PET CO2) below normal (group A:V˙/PET CO2, mean, 0.87; range 0.3–1.4 l.min-1/mm Hg). It was normal in all of the non-RLM group (group B: V˙/PET CO2, mean, 3.0; range, 2.6–3.9 min-1/mmHg). There was no significant difference in breathing between groups during relaxed wakefulness (V˙, group A: 7.44 (SD 2.5) l.min-1; group B: 6.02 (SD 1.3) l.min-1; PET CO2, group A: 41.0 (SD 4.2) mm g; group B: 38.3 (SD2.0) mm Hg) or during exercise (V˙/V˙O2: group A: 21 (SD 6.0) l.min-1/l.min-1; group B: 24 (SD 7.3) l.min-1/l.min-1). During sleep, all group A had fragmented sleep compared with only one patient in group B (group A: arousals, range 13 to > 60 events/hour); moreover, in group A there was a high incidence of obstructive sleep apnoea associated with hypoxaemia.

CONCLUSION Patients with unilateral RLM lesions require monitoring during sleep to diagnose any sleep apnoea. The finding that unilateral RLM lesions reduce ventilatory sensitivity to inhaled CO2 is consistent with animal studies. The reduced chemosensitivity had a minimal effect on breathing awake at rest or during exercise.

  • breathing
  • medulla
  • lesion
  • sleep

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

It has been known for some time that gross disorders of breathing can occur during both wakefulness and sleep in patients with bilateral medullary lesions (for review see Plum and Lee1); however, the effects of unilateral lesions are less clear. Individual case reports have led to the suggestion that respiratory control is compromised, particularly during sleep, where irregularity of breathing, hypopnoea, and obstructive apnoea may result.2-4

Focal lesions of the medulla oblongata can result from thrombotic occlusion in the posterior inferior cerebellar artery (PICA), the vertebral artery, or in the branches of these arteries.5The PICA ascends up the lateral surface of the medulla as far as the lower edge of the pons and supplies the medulla via small penetrating branches (fig 1). This anatomy means that occlusion of these vessels can produce small infarctions of the medulla. Vuilleumieret al 6 have shown that the topographical patterns of such lesions are likely to reflect the aetiopathogenic mechanisms which can cause them. The areas destroyed by the lesions we have described have been defined in animals as important to the control of breathing.

Figure 1

(A) Diagrammatic representation of the brainstem. The dotted lines mark the level at which a cross section through the medulla has been taken. (B) Each cross section numbered 1 to 5 (section 1 is the most caudal and section 5 the most rostral) are shown in detail. On each cross section the location of key structures are marked: IVth ventricle (IV vent); spinal trigeminal tract and nucleus (SPTTr and Nu); nucleus solitarius (SolNu); vestibular nuclei (VesNu); nucleus ambiguus (NuAm); hypoglossal nucleus (HyNu); dorsal vagal nucleus (DVagNu); lateral corticospinal tract (LCSp); pyramid (Py). Shaded area indicates the site of neuronal damage which occurs most often in lateral medullary syndrome. (C) The lateral surface of the medulla with the main arteries are marked: vertebral artery (VA), posterior inferior cerebellar artery (PICA), anterior spinal artery (ASA), basilar artery (BA), anterior inferior cerebellar artery (AICA), and the vagus nerve (Vag). (figure adapted with permission from Haines, 1991.21)

In recent years, the structural and functional organisation of the control of breathing within the brainstem has been extensively investigated. In animals, the work of Richter et al,7 Bianchi et al,8 and Smith et al 9 has provided an in depth description of a respiratory “oscillator” in the medulla oblongata. The influence of an animal’s state of consciousness on brainstem respiratory control has been highlighted in a series of studies on breathing during wakefulness, sleep, exercise, and hypercapnia in goats10 11 and in rats12 with induced medullary lesions. To our knowledge, similar measurements of breathing under different conditions have not been carried out in humans with unilateral focal lesions in the medulla.

The aim of the present study was to establish in humans whether spontaneously occurring unilateral focal lesions in the medulla could result in abnormalities of breathing during relaxed wakefulness (RW), sleep, and exercise. In addition, we wanted to determine whether such lesions could be associated with a reduced ventilatory response to the inhalation of CO2. Our studies were carried out on patients in whom the site of the lesions could be defined both clinically and with MRI. This work has been presented in preliminary form elsewhere.13



Patients were recruited over a 2 year period from the Neurology Service at Charing Cross Hospital. All patients suspected of having a focal lesion in the brainstem below the level of the midbrain were seen by one of us (PH). A detailed history was taken and full clinical examination carried out. Patients with any signs of upper airway pathology or chest disease (for example, pulmonary aspiration) were not studied; this was also true for patients with evidence of widespread neurological disease such as diffuse white matter pathology and those in whom the physical signs and symptoms were equivocal. The time between the onset of symptoms and participation in the study varied from days to a few years (table 1). All patients were studied with local ethical approval and all patients gave informed consent.

Table 1

Anthropometric data, lung function, ventilatory sensitivity to inhaled CO2, and exercise data for each patient


Fifteen patients with unilateral brainstem lesions were studied; a description of the associated clinical signs are given in table 2. In all except one patient (3, who was unable to tolerate the procedure), MRI was also carried out to confirm the clinical localisation of each lesion. Imaging of the medulla was first optimised by applying various commercially programmed data acquisitions to two normal subjects. The instrument used (Signa, IGE Medical Systems) was operating at 1.5 Tesla in the advantage configuration at level 4.2. From the data acquisitions performed on the two control subjects, four were chosen for application to the patients. The first three were performed in all patients; the application of the last sequence depended on patient tolerance. We restricted the MRI examination times to less than 40 minutes. The data acquisition sequences used are summarised below:

Table 2

Details of the clinical signs in each patient, plus the brainstem structures involved

(1) Sagittal spin echo, multiplanar (MEMP), TR 500/TE 10 ms, slice thickness 5 mm (interslice gap 2.5 mm), 16 images, matrix 256×192, field of view 18 cm, one excitation, study time 3.4 minutes.

(2) Axial spin echo, variable echo multiplanar (VEMP), TR 2800/TE 30 and 90 ms, slice thickness 4 mm (interslice gap 2 mm), variable bandwidth, 24 images, matrix 256×192, field of view 18 cm, one excitation, study time 9.31 minutes.

(3) Gradient recalled echo sequence using spoiler gradients (SPGR), volume acquisition, axial slice thickness 1.5 mm, 64 images, matrix 256×128, field of view 20 cm, two excitations, study time 8.14 minutes.

(4) Multiplanar gradient recalled echo sequence (MPGR), flip angle 20°, axial plane, TR 240/TE 15 ms, slice thickness 5 mm (interslice gap 2.5 mm), 12 images, matrix 256×256, field of view 18 cm, four excitations, study time 8.14 minutes.

The axial plane of acquisition was oriented to be perpendicular to the floor of the fourth ventricle. Movement compensation strategies employed were: (a) peripheral cardiac gating, where each excitation was triggered by a finger pulse monitor, (b) first order gradient moment nulling, and (c) presaturation of superior and inferior volumes. Multiplanar reprocessing of the volumetrically acquired data set (SPGR) was performed on an independent console (Physician image processor, IGE Medical Systems) where necessary to help the clarity of the display or to increase confidence in lateralisation.

The MRI of each patient was independently reviewed by two of us (JS and PH). All lesions were visible on all acquisitions and appeared of similar size and position, though no one acquisition was the best in all patients. Localisation to the rostrolateral medullary region was determined by reference to landmarks consistently seen in all patients, though not in all acquisitions. The external landmarks (the pontomedullary junction, floor of the fourth ventricle, olives, and pyramids) were used to establish the level and lateralisation of the lesion. The internal landmarks (the pyramidal tracts and their decussation, the medial lemnisci decussation) were used to determine the presence of midline involvement. The inferior cerebellar peduncles were used to ascertain lateral involvement.

On the basis of the MRI and the clinical examination, the patients were divided into two groups: those with brainstem lesions in the area of the rostrolateral medulla (RLM group) and those with lesions which did not involve the rostrolateral medulla (non-RLM group). For each patient, the site and extent of the functional brainstem lesion is given in table 2 and the anatomical levels at which the lesions were seen are illustrated by reference to figure 1.


Breathing was measured in all patients during hypercapnia, relaxed wakefulness, sleep, and exercise; these tests were not necessarily carried out on the same day.

Ventilatory sensitivity to inhaled carbon dioxide

In each patient, the ventilatory sensitivity to inhaled CO2 was measured using a rebreathing technique,14 modified for use under normoxic conditions. Before each test, the measurement system was flushed and filled with a mixture of 6% CO2 in air. Patients wore a nose clip and breathed via a three way tap and a mouthpiece. Rebreathing was commenced by turning the three way tap at the end of a normal expiration. Breath by breath measurements of ventilation (V˙), respiratory frequency (fR) and tidal volume, (VT) were determined from the respiratory airflow using an on line computerised ventilatory analysis system (Respiratory gas analyser, Buxco Electronics). Instantaneous measurements of CO2 and O2 gas concentrations were made at the mouth using a mass spectrometer (Centronics, MGA200). In addition, the ECG was monitored using surface electrodes and displayed on a monitor (LAN Electronics, M4-Scope). The test continued until PETCO2reached 65 mm Hg or until the patient was unable to continue due to breathlessness. Each patient was studied twice; the tests were separated by 15 minutes of rest. Linear regression of V˙ on PETCO2 over the linear portion of the response was carried out for each of the two tests; slopes withr values below 0.80 were not accepted for analysis. The mean of the two slopes was taken as representative of the hypercapnic ventilatory sensitivity; there was little variance between the two tests (mean difference (SD): RLM group, 0.17 (0.62) l.min-1 /mm Hg; non-RLM group 0.31 (0.56) l.min-1 /mm Hg). In two patients (6 and 9) only one slope had an acceptable r value. The normal range of the ventilatory sensitivity to inhaled CO2 in our laboratory with this technique is mean (SD) 3.5 (1.2); range 1.7–6.2 l.min-1 /mmHg.15


Five minutes of breathing during RW was measured before the sleep study. During the RW measurements, the patient lay on the bed wearing a blindfold and headphones, to reduce sensory input16; patients were monitored by means of a video. RW was confirmed from monitoring the record of two EEGs (EEG; C3-A2 and C4-A1), two electro-occulograms (EOG; F7-A2 and F8-A2) and one EMG (chin EMG) made using a Mingograf EEG 10 recorder (Siemens-Elema). Breathing was measured from recordings of chest wall and abdominal movements made using a DC coupled, respiratory inductance plethysmograph (RIP; Respitrace Co ambulatory monitoring). The RIP was calibrated with the patient simultaneously breathing into a rolling seal spirometer (Spiroflow, PK Morgan). Expired airflow was used to derive the PETCO2 using an infrared gas analyser (LB2, Beckman Instruments). Ear arterial oxygen saturation (SaO2) was estimated using a pulse oximeter (Biox 3700, Ohmeda). VT, total breath duration (TTOT), V˙, and Petco 2 were calculated on a breath by breath basis.

Measurements of breathing during sleep were made between 2100 and 0700 hours. Sleep stages were scored from 30 second epochs according to the standard criteria.17 The first period of uninterrupted stage IV Non-REM sleep longer than 5 minutes was chosen for analysis of the sleep related respiratory variables, provided no signs of hypoventilation or apnoea secondary to mechanical obstruction were seen. Apnoeas were classified as obstructive when paradoxical ribcage and abdominal movements occurred during an absence of a Petco 2 signal (index of airflow) for>10 seconds duration, together with a resumption of breathing associated with a loud snoring or snorting sound and on most occasions an arousal. Apnoeas were classified as central when the cessation of airflow was not associated with respiratory effort. Mixed apnoeas were central apnoeas that progressed into obstruction. Hypopnoea was defined as the presence of a reduced amplitude of ribcage and abdominal movements in association with loud snoring sounds during inspiration and/or expiration. Where possible, signs were confirmed using oesophageal pressure (Poes) and airflow signals (see below).

Three patients (3, 4, and 13) were studied on two occasions during sleep; in each case, the second study was carried out to confirm the diagnosis and to evaluate treatment. During the second study, measurements of Poes were made as an index of respiratory effort. A balloon tipped catheter was passed through the nose and positioned in the mid-third of the oesophagus (about 40 cm from the nares); it was connected to a differential pressure amplifier (MP-45, (SD 80) cm H2O, Validyne). In these patients, all data presented are from the first study (without an oesophageal catheter). In one patient (6, who could only be studied on one occasion) oesophageal pressure (Poes) monitoring was carried out during the initial sleep study.


All patients performed a progressive exercise protocol in a laboratory maintained at an ambient temperature between 18°C and 23°C. Patients wore a nose clip and breathed through a mouthpiece attached to a thermostatically heated Fleisch No 2 pneumotachograph. Exercise was performed with the patient seated in a comfortable chair with their feet strapped to the pedals of a cycle ergometer. Three patients with unilateral leg weakness (4, 9, and 13) performed exercise with one leg only using specially designed exercise equipment18; the weak leg was kept relaxed in a comfortable position. After one minute of resting cardiorespiratory measurements, 4 minutes of rolling basal measurements were made, during which the patient pedalled against a negligible resistance. The patients were encouraged to maintain a pedal rate of between 50 and 60 revolutions per minute. One minute incremental workloads were set taking account of the patient’s clinical weakness; individual increments varied between 5 and 25 W/minute. An automated exercise analysis system (Ergostar, Fenyves, and Gut) provided 30 second average measurements of heart rate, fR, VT, V˙, and Petco 2 corrected to provide a better estimate of Petco 2 .19 ECG and SaO2 were monitored continuously throughout exercise. In addition, blood pressure was measured every 2 minutes. The test was terminated when (a) the patients were unable to continue, (b) the predicted maximum heart rate was achieved, (c) an excessive rise in systolic blood pressure occurred, or (d) when an ECG abnormality was detected.

Thirty second averaged data from each progressive exercise test were subjected to linear regression by a least squares fit of V˙ on oxygen uptake (V˙O2). The slope of this regression provided a quantitative index of individual ventilatory sensitivity to progressive exercise. In the cases where the threshold for anaerobiosis (lactate threshold20) was exceeded (as shown by the upward inflection of the ventilatory equivalent for V˙O2), only the 30 second average measurements between the end of the rolling basal period and this upward inflection were included in the linear regression.


Comparisons of group mean variables (RLM groupv non-RLM group) for anthropometric, lung function, ventilatory sensitivity to inhaled CO2, and exercise data were made using a two tailed unpairedt test. Comparison of group mean variables (group A v group B) for breathing pattern (TTOT, VT, V˙, and Petco 2) measured during RW, were made using a one way analysis of variance (ANOVA). For all tests the level of significance was taken as p<0.05.



For each patient, the anthropometric data and details of lung function, ventilatory sensitivity to inhaled CO2 and exercise are given in table 1. There was no significant difference between the two groups for age (p=0.95), height (p=0.27), weight (p=0.78) and the degree of obesity as judged by the body mass index (BMI: height / weight2; p = 0.71). Two patients (2 and 9) had a reduced lung vital capacity (VC), and one patient (3) had lung restriction due to diffuse interstitial lung disease of unknown cause. However, there was no significant difference between the two groups in the ratio of forced expiratory volume in 1 second to forced vital capacity (FEV1/FVC, p=0.55) or in VC (p=0.13). In each patient, the location of the lesion was defined clinically (table 2) and confirmed by MRI. An example of MRI in a patient (6) with an RLM lesion is shown in figure 2.


The individual ventilatory sensitivities to inhaled CO2 are shown diagrammatically in figure 3. For each patient the slope of the ventilatory response to inhaled CO2 and the correlation coefficient (r) are given in table 1. Six of the nine RLM group patients had a reduced CO2 sensitivity. The remaining three patients (7, 8, and 9) were found to have a normal hypercapnic ventilatory response despite having a lesion in the RLM. In each of these patients, the lesion was relatively caudal and did not extend up into level 5 (fig 1). For the non-RLM group all responses were within normal limits. There was a significant difference between the group mean slope of the ventilatory response to inhaled CO2 in the RLM group compared with the non-RLM group (p=0.004). Overall, we found that patients with unilateral damage to the RLM had a reduced ventilatory sensitivity to inhaled CO2.

Figure 3

A diagrammatic representation of the awake ventilatory sensitivity to inhaled CO2 (V˙=ventilation; Petco 2 = end tidal carbon dioxide) in patients with lesions involving the rostrolateral medulla (RLM=dotted lines) and patients with lesions not involving the rostrolateral medulla (non-RLM=full lines). In one patient (* 5) an abnormally high level of ventilation occurred; note that this chronic hyperventilation was not exacerbated during hypercapnia.


Group mean (SD) of all the respiratory variables (TTOT, VT, V˙, and Petco 2 ), measured during RW, in the RLM group patients with a reduced sensitivity to inhaled CO2 (group A) and in five of the non-RLM group patients with normal sensitivity to inhaled CO2 (group B) are shown in table 3. There was no significant difference between group A and group B for any of the respiratory variables measured. Both group A and group B had a similar mean Petco 2 , with the exception of one subject (3) (RW: Petco 2 , 48 mm Hg) in whom the diffuse lung shadowing had been noted. One non-RLM patient (13) was not included in this analysis because we were unable to quantify the CO2 measurements for technical reasons.

Table 3

Breathing during relaxed wakefulness

All patients were able to sleep; the total sleep time and distribution of sleep stages varied greatly. Details of the sleep patterns are shown in table 4. In one group A patient (2), sleep stages were not scored because the EEG wave forms were atypical and did not conform to the standard criteria.17 All group A patients and one group B patient (13), were found to have frequent short arousals from sleep (13 to>60 arousals/hour of sleep). In group B, three patients (10, 12, and 15) had a total sleep time below 50% of the study time; they reported difficulty sleeping in the laboratory.

Table 4

Summary of sleep patterns

The incidence of sleep disordered breathing, either central in origin, or secondary to mechanical obstruction, was higher in group A than in group B. In group A, four patients (1, 2, 3, 5) had obstructive sleep apnoea/hypopnoeas, compared with only one patient in group B (13); this patient had a unilateral tongue weakness with a medial medullary syndrome (table 2). The BMI was similar in both groups (table 1). In group A, two patients (1 and 2), had very fragmented sleep despite the low apnoea/hypopnoea index. Unequivocal central hypoventilation (in the absence of mechanical obstruction) was not seen in any of the patients. However, in group A, one patient (4) was found to have frequent mixed apnoeas (>60 apnoeas/hour sleep) (fig 4); this patient had a normal BMI and a low sensitivity to inhaled CO2 (table 1).

Figure 4

Original recording in one patient (4) during non-REM sleep. Oxygen saturation (SaO2) shows transient dips associated with negative intrathoracic pressure and reduced respiratory movements, indicating upper airway obstruction and sleep apnoea.

Comparisons of breathing during wakefulness and sleep for group A and B patients were hampered by the fact that in group A only two patients (5 and 6) achieved stable stage IV sleep during which regular breathing with no signs of mechanical obstruction of the upper airway were seen. In these two patients, in whom the sensitivity to inhaled CO2 was reduced (table 1), the reduction in breathing asleep compared with that which occurred during RW was substantial (Petco 2 , RW v NREM sleep; patient 5: 35.7 mm Hg v 39.1 mm Hg; patient 6: 40.8 mm Hg v 42.8 mm Hg). In group B, regular breathing during stage IV sleep occurred in four subjects (group mean (range): RW v NREM sleep, 37.6 (35.9—40) mm Hg v 38.3 (36.3—40.4) mm Hg).


For all but one group A patient (2), the ventilatory sensitivity to exercise (ΔV˙/ΔV˙O2) is given in table 1; patient 2 was unable to exercise beyond the rolling basal exercise period. V˙O2 increased in group A up to a mean of 1.39 l.min-1 (range 1.06–1.97 l.min-1), and in group B up to a mean of 1.57 l.min-1 (range 0.99–2.19 l.min-1). V˙ increased in group A up to a range of 21–26 l.min-1, and in group B up to a range of 15–41 l.min-1; fR changed little. The baseline PETCO2 ranged from 28–45 mm Hg; it remained almost constant throughout most of the test and only began to decrease at a V˙O2 of about 1.2 l.min-1; this hyperventilation was taken as evidence that anaerobic metabolism was developing at this time.20 The absence of any relation between the ventilatory response to exercise and the response to inhaled CO2 in the RLM and the non-RLM groups is shown in fig 5.

Figure 5

Individual ventilatory responses to exercise (defined as the slope of the linear regression of ventilation on oxygen uptake, ΔV˙/ΔV˙O2, between the rolling basal period and the onset of anaerobiosis) plotted in relation to that person’s ventilatory sensitivity to inhaled CO2 (defined as the slope of the linear regression of ventilation on Petco 2, ΔV˙/Δ Petco 2 ) for each patient (RLM group closed circles; non-RLM group open circles).


This study shows that patients with a unilateral lesion in the rostrolateral medulla have a poor ventilatory response to inhaled CO2. Breathing is seriously disrupted during sleep with a high incidence of obstructive apnoea. When awake, patients breath normally both at rest and during exercise.


In the present study, we measured the ventilatory response to inhaled CO2 when awake, using a well established, non-invasive clinical method14 as an overall test of ventilatory chemosensitivity under normoxic conditions. This test produces a measurement of the ventilatory sensitivity to CO2 above the level of PACO2 before the rebreathing. We cannot know how the sensitivities measured in our study relate to the sensitivity around the PACO2 level at rest; this may be an important consideration in the interpretation of our results under the different states studied.


We have studied patients with vascular occlusions in the PICA or its branches,5 6 21 which produced focal unilateral lesions; the associated clinical signs suggested the anatomical location in the rostrocaudal direction, and verification of this site was possible using MRI. The uniformity of the finding of Horner’s syndrome in the RLM group, in whom the ventilatory sensitivity to inhaled CO2 was reduced, provides additional support for the anatomical site of the lesion. Animals with chemically induced unilateral lesions in the retrotrapezoid and subretrofacial nuclear areas have an absent or reduced increased phrenic nerve output during inhalation of CO2; in addition the respiratory related amplitude of the integrated activity in the cervical sympathetic nerve is also reduced.22 The neurons in these areas give rise to sympathetic outflow23; in the cat, electrical stimulation in these regions can cause ipsilateral pupillary dilatation.24 Taken together these findings suggest that both in animals and in humans, the neuronal groups concerned with sympathetic output and respiratory control are intermingled at these sites.

Obtaining clear images of the medulla is difficult. The structure is both small and subject to complex cardiosynchronous oscillatory movement in the CSF and vertebral arteries; the second also often distorts the anterolateral contour by compression. This motion introduces artefacts which project into the images of the medulla and this reduces the sensitivity of detecting changes in tissue structure. Although these disturbing effects can be minimised by the use of a range of strategies, performance will still vary even with images of similar specification. In the present study, the use of varied MRI sequences gave us confidence in defining the extent and localisation of the clinical lesions. We preferred to use internal and external landmarks rather than grids and templates used by others.25


In the present study unilateral lesions in the RLM resulted in sleep disordered breathing and in arousals from sleep which were independent of any respiratory related disturbance in sleep patterns. The fact that some of the patients were unable to maintain sleep suggests that neuronal damage was present in regions of the brainstem concerned with the sleep/wake cycle. In group A the high incidence of sleep and breathing disturbances meant that we found it difficult to interpret the importance of the reduced sensitivity to inhaled CO2 on breathing during sleep, because long periods of steady breathing during deep sleep did not occur. Nevertheless, in two patients who had some stability of breathing during sleep, a reduction in ventilation did occur.

The frequency of upper airway obstruction, particularly in the absence of any neurological lesions of the hypoglossal nerve(s), is difficult to interpret. It may have resulted from the unilateral lack of coordination of the upper airway muscles, due to the lesion affecting the deeper structure of the nucleus ambiguus. In the RLM group it is noteworthy that three of the five patients with dysphagia (at the time of study) also had sleep disordered breathing with obstruction of the upper airway. Our findings are consistent with the case report of a patient with a unilateral lesion in the medulla and obstructive sleep apnoea by Chaudhary et al.2 It is also of interest that acute lesions produced by cooling in the rostral ventrolateral medulla of the awake goat can result in obstruction of the upper airway, necessitating a tracheostomy so that the breathing in these animals could be studied (fig 5 in Forsteret al 26). These workers suggested that their findings could be explained, if cooling decreased the inspiratory related facilitation of the upper airway muscles to a greater degree than the diaphragm.

The maintenance of normal ventilation during wakefulness, when the ventilatory sensitivity to inhaled CO2 was minimal, emphasises the importance of the wakefulness drive(s) on breathing.27 A similar dissociation between the normality of the ventilatory response to exercise and the reduced or absent ventilatory chemosensitivity occurs in children with the congenital hypoventilation syndrome.28 The lack of correlation between the ventilatory response to CO2 and the response due to exercise, in these patients and in the present study, illustrates that the mechanisms responsible for these responses are likely to be different.


There are clear similarities between our physiological findings in these patients and the evidence from animal studies—that is, the effect of unilateral lesions in the rostral medulla causing a reduction of ventilatory sensitivity to CO2 with maintenance of normal ventilation awake but not asleep.12 29 30 Our human studies add the evidence of a normal ventilatory response to exercise. However, the animal studies have focused on rostral ventrolateral medullary lesions. In our studies, the lesions are rostral but lateral or even dorsolateral, as shown on the MRI in figure2. The rostral dorsolateral medulla is full of neuronal pools and tracts (fig 1), which when damaged give rise to clear cut physical “signs”, this is not so for the ventrolateral medulla, lateral to the pyramids. We cannot therefore be confident that the anatomical sites of the relevant lesions in humans are comparable with the sites of the lesions induced in rats, cats, and goats because the shape of the brainstem and the anatomical organisation within are different. Comparisons between the brainstem of humans and that of animals is made more difficult by the huge development of the ventral pons in humans, consequent on the presence of the major corticopontocerebellar tracts. In the laboratory animals, this development would necessitate the “migration” of ventral structures to a more lateral or dorsolateral position.

Figure 2

An axial MRI through the upper brainstem and cerebellum (SPGR, data acquisition method No 3-see methods) from patient 6. A small, low density lesion is shown just deep to the left lateral surface of the rostral part of the medulla oblongata, between the olive and inferior cerebellar peduncle (arrow); it was visible on three contiguous images, giving it a longitudinal extent of about 4.5 mm. The high signal in the left vertebral artery adjacent to the olive is produced by flowing blood and is a normal signal for this type of data acquisition.


Partial or even complete recovery is not rare in patients with lateral medullary syndrome. Some recovery of function did occur in our patients; this was particularly shown by the recovery from dysphagia (table 2). In our study there was a large variation in time between the onset of the lesion and when we did our investigations (table 1). This would have resulted in an opportunity for recovery in some patients but not in others. It is of note that patients 9 and 13 (with the longest time to study) had a ventilatory sensitivity to inhaled CO2which was within our normal range. We cannot be certain, particularly in these patients, that the ventilatory sensitivity measured soon after the lesion developed would not have been normal.


The findings of this clinical study strongly support the experimental evidence from animal investigations that the RLM is a key area for the full expression of the ventilatory response to inhaled CO2. In addition, the results show that breathing is normal during wakefulness, both at rest and on exercise in patients with a unilateral lesion in this area of the brainstem. During sleep, both breathing and sleep patterns are seriously disrupted. These findings during sleep have clinical implications, particularly in the cases where the sleep disordered breathing was accompanied by repeated hypoxaemic episodes. In patients with ischaemic insult, it would seem important to obviate an additional arterial hypoxaemic insult to the brainstem. Any sleep related hypoxia can be relatively easily measured using some form of nocturnal monitoring.


We thank Professor L Adams of the National Heart and Lung Institute for his careful supervision of aspects of this study and Ms P Boyle for her assistance with some of the investigations. We are indebted to Professor D E Haines of the Department of Anatomy of the University of Mississippi Medical School, and the publishers of hisTextbook of human neuroanatomy. 3rd ed, Williams and Wilkins for permission to use some of their illustrations to generate figure 1. This work was funded by a Wellcome Trust Programme Grant awarded to AG. MJM is supported by a Wellcome Trust Research Career Development Fellowship.