INTRODUCTION

TOP ABSTRACT INTRODUCTION CASE REPORT DISCUSSION REFERENCES

The location where respiratory rhythm is generated in humans in unknown. In experimental animals, it has been recognized since the work of Lumsden (1) in 1923 that eupnea can be generated by mechanisms inherent to pons and medulla. Identification of a specific region that is critical for the neurogenesis of eupnea has been elusive and controversial.

Within the medulla, respiratory-modulated neuronal activities are concentrated in two regions termed the dorsal and ventral respiratory groups (2, 3). Experimental studies in cats have established that extensive damage to either or both of these medullary groups markedly impairs respiratory motor output, but with preservation of the eupneic rhythm (4). These findings led to the conclusion that no single medullary region was responsible for the neurogenesis of eupnea (3).

We report an unusual case of weaning failure secondary to a large metastasis in the medulla. This case is of interest because (1) it confirms, in a human, the earlier findings in cats (4) that massive destruction of the respiratory areas in the medulla may differentially impair the intensity of motor output per breath while leaving intact rhythmogenesis and response of respiratory rate to CO₂; (2) it demonstrates the application of a paradigm of physiologic tests, based on the neuroanatomy of respiratory control, to the investigation of cases where respiratory muscle weakness is an important contributor to weaning failure; (3) it suggests that voluntary control of respiratory muscles is executed primarily via the medullary respiratory centers (i.e., corticobulbar to bulbospinal pathways) as opposed to direct activation of spinal motoneurons by corticospinal tracts.

	CASE REPORT

TOP ABSTRACT INTRODUCTION CASE REPORT DISCUSSION REFERENCES

A 67-yr-old man with a past history of heavy smoking and alcohol abuse presented to the emergency department with a decreased level of consciousness. The patient was receiving no medications, and there was no prior history of respiratory disease. Serum sodium was found to be decreased at 120 mmol/L. CXR revealed a large right hilar mass. The patient developed progressive hypercapnic respiratory failure in the emergency department. Arterial blood gas determinations, obtained while the patient was breathing oxygen delivered by nasal cannula at 2 L/min, revealed the following results: PO₂, 78; PCO₂, 118; pH, 7.02; HCO₃, 24. The patient underwent intubation, and mechanical ventilation was instituted. A noninfused CT of the brain was interpreted as normal. He was subsequently transferred to the intensive care unit.

The patient's hyponatremia was gradually corrected in the intensive care unit. By the second day after admission the patient was alert, oriented, and able to follow commands. Bronchoscopy revealed an endobronchial lesion in the right bronchus intermedius. The mass was biopsied, and pathology was consistent with a non-small-cell bronchogenic carcinoma. The patient was placed on Cefuroxime, 750 mg given intravenously every 8 h, and Clindamycin, 600 mg given intravenously every 8 h, for treatment of any coexisting pneumonia.

Attempts were undertaken to wean the patient from mechanical ventilation. Initially, the patient was placed on a T-piece, but he became tachypneic, with a respiratory rate of more than 40 breaths/min, and he developed progressive hypercapnia (PCO₂ > 100 mm Hg) with respiratory acidosis. Subsequently, the patient was placed on pressure support ventilation (PSV). However, on PSV, he was described as intermittently apneic, with a ventilator rate of less than 10 breaths/min, despite an elevated PCO₂ and respiratory acidosis.

The patient was hemodynamically stable and had no evidence of postural hypotension when moved from the recumbent position in his bed to sit in a chair by the bed.

Maximal inspiratory pressure (MIP) was measured and was found to be reduced at 9 cm H₂O.

The etiology of the patient's respiratory muscle weakness was uncertain. Repeat neurologic evaluation confirmed that the patient had well-preserved strength in the limbs and neck, thereby making a diagnosis of a generalized neuropathy or myopathy unlikely. However, the patient was noted to have a left-sided Horner's syndrome. Together with the right hilar mass, this raised the possibility of bilateral mediastinal involvement with tumor, leading to bilateral phrenic nerve injury and diaphragmatic paralysis. As a result, further physiologic assessment was undertaken with the particular goal of assessing respiratory muscle function. At the time of the study, the patient was receiving no medications other than antibiotics.

Initially, respiratory mechanics were measured during a period of controlled ventilation (CMV). Using the end-inspiratory pause technique, the elastance of the respiratory system was 18 cm H₂O/L. Resistance, measured at a flow of 1 L/s, was 7 cm/L/s. The patient was being ventilated with a no. 8 endotracheal tube. The value for elastance was interpreted as being mildly abnormal for a ventilated patient in the ICU. The value for resistance was essentially normal for a patient breathing through a no. 8 endotracheal tube. Thus, it was felt that weaning failure was primarily related to respiratory muscle dysfunction.

MIP was repeatedly measured, and the highest value obtained was 15 cm H₂O (predicted normal, 70 to 100). Inspiratory capacity (IC) was 0.97 L. Patient effort and cooperation were felt to be good with both maneuvers. These values indicated significant weakness of voluntary efforts.

Because of concern over diaphragmatic paralysis, a gastroesophageal catheter was placed in an effort to measure transdiaphragmatic pressure (Pdi). Maximal Pdi (Pdi_max) was 16 cm H₂O. This result was similar to the MIP. Because MIP assesses global respiratory muscle strength, whereas Pdi_max selectively assesses the strength of the diaphragm, in the presence of isolated bilateral diaphragmatic weakness one would expect to observe a Pdi_max that is substantially less than MIP. The fact that they were essentially equal effectively ruled out bilateral diaphragmatic paralysis. Further confirmation was provided by the fact that gastric pressure became positive with inspiration (reflecting downward movement of the diaphragm with inspiration), and the absence clinically of paradoxical inward movement of the diaphragm with inspiration. CT scan of the thorax subsequently failed to demonstrate evidence of metastatic disease in the mediastinum. We utilized the esophageal pressure waveform to measure static pulmonary compliance, which was 150 ml/cm H₂O, confirming relatively normal pulmonary compliance.

A weaning trial was next undertaken with the Pdi catheter in place. The patient was placed on volume-cycled ventilation with a back-up rate of 10/min at a tidal volume of 550 ml. PEEP and FI_O2 were set at 5 cm H₂O and 30%, respectively. At this rate, there were no spontaneous respirations visible in the Pdi waveform. The back-up rate was then progressively lowered. At a ventilator rate of 8 breaths/ min, spontaneous but feeble inspiratory efforts began to appear at a rate of 7/min. PET_CO2 was 56 mm Hg at this point. The back-up rate was lowered further, and then the patient was switched to pressure support ventilation (PSV) at a pressure of 6 cm. PET_CO2 rose further to 78 mm Hg, and the patient's respiratory rate increased to 40 breaths/ min. An arterial blood gas determination done at this point revealed: PO₂, 52; PCO₂, 89; pH, 7.17; HCO₃, 29, with an oxygen saturation of 80%. However, despite a significant hypercapnic and hypoxic stimulus to breathe, the patient's spontaneous Pdi swings were only 1 to 2 cm in amplitude, resulting in only intermittent and infrequent triggering of the ventilator (Figure 1). During a brief period of CPAP, the patient continued to breathe at a rate of 40 breaths/min, but with tidal volumes of only 100 to 150 ml. Thus, the automatic Pdi and tidal volume were less than 15% of the voluntary Pdi_max and IC. Despite the tachypnea, there was no clinical evidence of respiratory distress in the form of agitation or visible neck muscle use. The patient was in fact quite calm during the weaning trial.

View larger version (76K): [in this window] [in a new window]   Figure 1.   Patient breathing with PSV of 6 cm H2O. Note positive deflections in the transdiaphragmatic pressure (Pdi) waveform (small arrows) that occur at a rate of about 40/min but are only about 1.5 cm in amplitude, and therefore only intermittently able to trigger the ventilator. Triggered breaths are denoted by large arrows in the airway pressure waveform (Paw at the top of the figure).

During the weaning trial, as the patient became progressively more hypercapnic, his heart rate rose from 75 to 118 beats/min, and his blood pressure rose from 134/98 to 206 /122.

Collectively, the results of the above studies were felt to be highly suggestive of a brainstem lesion in the medulla (see DISCUSSION). The patient underwent an MRI scan that revealed five enhancing intracranial mass lesions, consistent with metastases. Three lesions were located within the cerebral cortex, in the left inferior temporal lobe, right inferior frontal lobe, and right occipital lobe. One lesion was located in the cerebellar vermis, and one was a large 1.5-cm lesion occupying much of the medulla (Figure 2). The medullary lesion extended from the obex to the pontomedullary junction, but it appeared to spare a rim of medullary tissue peripherally, particularly on the right.

View larger version (153K): [in this window] [in a new window]   View larger version (145K): [in this window] [in a new window]   Figure 2.   MRI scan with T1-weighted axial (top panel ) and sagittal images (bottom panel ) of the brain. The medullary metastasis appears as an area of decreased signal intensity (dark) on the sagittal image, which demonstrates the vertical extent of the tumor from the pontomedullar junction to approximately the obex. On the gadolinium-enhanced axial image, the tumour is also dark, and there is peripheral enhancement (white) of the mass lesion that corresponds to the tumor capsule. Note that on the axial image, the tumor involves the central medulla with sparing of a rim of medullary tissue peripherally, particularly on the right. A second, peripherally enhancing lesion is also identified in the cerebellar vermis on the axial image.

The patient was subsequently started on therapy with dexamethasone and underwent a course of radiotherapy in an effort to shrink the tumor mass and surrounding edema. Unfortunately, during the next 2 wk the patient developed progressive upper motoneuron weakness in all four extremities, lost his gag reflexes, and failed to demonstrate any improvement in his respiratory muscle strength. The patient's ability to maintain a respiratory rhythm remained intact. After discussion with the patient and family, active therapy was discontinued, and the patient died. The family declined an autopsy.

	DISCUSSION

TOP ABSTRACT INTRODUCTION CASE REPORT DISCUSSION REFERENCES

We have described a case in which a large medullary metastasis caused severe impairment of respiratory motor output while selectively preserving numerous brainstem functions, including the respiratory rhythm-generating function, cardiovascular vasomotor centers, and descending motor tracts to the extremities. Of particular interest in our patient is the preservation of respiratory rhythm generation. Once arterial CO₂ was allowed to rise above the apneic threshold, regular respiratory efforts were visible in the Pdi and airway pressure waveforms, confirming an intact pacemaking function. Furthermore, respiratory rate responded in a normal fashion to the progressive hypercapnia and hypoxemia. However, the intensity of respiratory motor output per breath was significantly impaired. The findings in this patient will be discussed under separate headings.

Origin of Respiratory Rhythm

The upper motor neurons subserving automatic breathing are located in two groups of nuclei in the medulla, referred to as the ventral and dorsal respiratory groups (VRG and DRG) (2, 3). The DRG is located in the region of the ventrolateral solitary tract nucleus (SN), and the VRG is located in the region of the nucleus ambiguus (NA) and nucleus retroambigualis (NRA). Whether rhythmicity originates within these nuclei or is produced elsewhere is not known. Experimental studies in cats have established that extensive bilateral damage to either or both of these medullary groups markedly impairs the amplitude of respiratory activity within each breath, but rhythm continues (4). These findings led to the conclusion that a single pacemaking region (e.g., analogous to the SA node of the heart) does not exist in these areas (3). Judging by its location, the lesion found in our patient destroyed much of the VRG and DRG (Figure 3). This was particularly true on the left side of the medulla where both the NA and the SN appear to lie withing the borders of the lesion. On the right side of the medulla, the extent of damage to the VRG and DRG is more difficult to determine since the lesion encroaches on these structures but does not totally encompass them. Further, the extent to which surrounding edema or disruption of blood flow beyond the borders of the tumor capsule compromised the integrity of these structures cannot be ascertained with certainty from the MRI. In our patient, breathing pattern showed a preserved rhythm but much reduced amplitude of respiratory activity. These findings are therefore very similar to those in animals subjected to widespread destruction of the VRG and DRG (4) and represent the first demonstration in humans of this important characteristic of respiratory rhythm generation. Moreover, the respiratory rate in this patient responded in a reasonably normal fashion, rising from 7 to 40/ min, with progressive hypercapnia and hypoxemia.

View larger version (75K): [in this window] [in a new window]   Figure 3.   Schematic of a cross section through the human rostral medulla corresponding to approximately the same level as that shown in the axial image of the MRI scan. Structures shown include nucleus ambiguus (NA), solitary tract nucleus (SN), solitary tract (ST), spinal trigeminal nucleus (STN), spinal trigeminal tract (STT), pyramid (P), cerebellar peduncle (CP), dorsal motor nucleus (DMN), hypoglossal nerve (CN XII), and inferior olivary nucleus (ION). The heavy dark circular line demarcates an area that corresponds to the approximate borders of the medullary lesion at this level.

Preservation of rhythm despite massive destruction of most of the VRG and DRG implies either that any portion within these areas is capable of generating the rhythm (networking mechanism) or that a specialized pacemaking region exists, but elsewhere. In this regard, Smith and coworkers (7) proposed an area extending medial to ventrolateral to the nucleus ambiguous in the rostral medulla (the pre-Botzinger complex) as the site of origin of respiratory rhythm. This proposal is currently quite controversial (8, 9). Given the current controversy surrounding this region, the issue of whether the pre-Botzinger area was destroyed by the tumor in our patient is of considerable interest. However, although there is information on the location of the pre-Botzinger complex in cats and rats (10- 12), the location of the pre-Botzinger complex has not been described in humans. Furthermore, even if one assumes that it is located in the same area in humans as in animals, one cannot be certain that the pre-Botzinger complex was destroyed bilaterally in our patient (see Figure 3), and it is known that unilateral lesions of this area do not disrupt respiratory rhythm (13). Accordingly, this case does not directly address the importance of the pre-Botzinger area, other than to suggest that if such a pacemaking site does exist in humans, it can be destroyed unilaterally without significant effect on respiratory rhythm.

Although impossible to exclude entirely, we believe it unlikely that the lesions outside the medulla, in the cerebellum and the cerebral cortex, played as significant a role in the respiratory abnormalities we observed. Both decerebration and cerebellectomy are routinely done in animals in neurophysiologic studies on the control of breathing, with no fundamental change in breathing pattern. Further, the lesions in the cerebral cortex were not in areas likely to affect the corticospinal tracts. Finally, destruction of the cerebellum in cats has been shown to have only minimal impact on inspiratory motor output (14).

In summary, this case confirms findings in experimental animals that large lesions involving the respiratory centers of the medulla may preferentially affect the amplitude of respiratory motor output with little disruption of rhythm generation or response of respiratory rate to chemical stimuli.

Rationale for Interpretation of Physiologic Results

The original CT scan of the brain was negative. However, the results of the physiologic tests, coupled with no intrathoracic explanation for the Horner's syndrome, strongly suggested a brainstem lesion. These results prompted the second scan (MRI), which identified the medullary lesion. It may be useful to review the physiologic basis for the interpretation of these tests.

A schematic diagram of the pathways involved in voluntary and involuntary control of respiratory muscles is shown in Figure 4. The upper motor neuron for rhythmic spontaneous respiratory activity originates in the dorsal and ventral respiratory groups as discussed above. Activity from these neurons is transmitted to the spinal motoneurons supplying the diaphragm and other respiratory muscles via bulbospinal tracts in the ventrolateral spinal cord (15). It is likely that voluntary maneuvers such as MIP and IC are executed via two independent pathways. There are direct connections between the cerebral cortex and spinal motoneurons supplying the respiratory muscles (16, 17). These are analogous to the pathway involved in voluntary control of other muscles (corticospinal pathway). However, there are also connections between the cortex and the medullary respiratory groups (corticobulbar pathway). Orem and Netick (18) found that when awake cats are conditioned to hold their breath in response to a bell ring, the respiratory neurons in the brainstem are inhibited during the apnea, indicating that behavioral responses are in part executed by action of the cerebral cortex on the medullary centers, in addition to the direct corticospinal pathway. The lower motor neuron, consisting of spinal motoneurons, peripheral nerves, and muscles, is common to both voluntary and involuntary pathways.

View larger version (14K): [in this window] [in a new window]   Figure 4.   Schematic demonstrating the two alternative pathways for voluntary respiration, the corticospinal tract (dashed line), and the corticobulbar and bulbospinal tract (solid lines). The lower motoneuron is a final common pathway for both.

Our patient's maximum inspiratory pressure (MIP) was severely reduced ( 20% of predicted). There is no reason to suspect poor effort. The patient was very alert and, once he learned what was expected, the results became highly reproducible (± 2 cm H₂O on numerous determinations over 2 d). Also, given his passive elastance (Ers) of 18 cm H₂O/L, the pressure he generated during the technically less demanding IC maneuver was in keeping with MIP: IC was 0.97 L, giving a pressure of 17 cm H₂O during maximum unobstructed effort (Pmax = IC × Ers). There was, accordingly, little doubt that the patient suffered severe weakness of voluntary effort. The severity of this weakness (20% predicted) indicated bilateral involvement. The similarity between MIP and Pdi_max and the fact that gastric pressure increased during voluntary inspiratory efforts (signifying diaphragmatic descent), indicated that the weakness was not limited to either the diaphragm or rib cage/accessory muscles but was widespread. The possible lesions that can account for this are as follows:

1. Diffuse involvement of the lower motor neuron such as with a paraneoplastic neuropathy/myopathy or a metabolic disorder. This was felt to be extremely unlikely. Detailed examination by a neurologist of motor function in nonrespiratory muscles failed to demonstrate significant weakness at a time when the respiratory muscles were very weak. We are not aware of any reports of such selective severe involvement of respiratory muscles with paraneoplastic or metabolic disorders.
2. Involvement of the corticospinal pathway by a metastatic, vascular, or paraneoplastic process. This was, again, felt to be very unlikely. There is no evidence, or reason to believe, that the corticospinal tracts serving the respiratory muscles follow a different pathway from other corticospinal tracts. In patients with discrete lesions involving the corticospinal tracts in the ventral pons (the locked-in syndrome), voluntary respiratory maneuvers are equally involved (19, 20). Because the weakness was bilateral and involved both the diaphragm and rib cage muscles one would have to postulate multiple lesions that selectively involved the tracts supplying these particular muscles but excluding, at each site, the neighboring tracts supplying nonrespiratory muscles.
3. Involvement of the indirect voluntary pathway, that acting via the medullary centres and bulbospinal tract. This was felt to be the most likely, both by a process of elimination (1 and 2 above) and because of the results of the weaning trial, which showed a very small ratio of spontaneous Pdi/ Pdi_max (or spontaneous VT/IC) at high levels of chemical stimulation with a high respiratory rate. In the face of progressively increasing respiratory drive (e.g., hypercapnia, hypoxia, exercise, obstructive sleep apnea), motor output per breath, expressed as pressure output or VT, becomes a progressively larger fraction of pressure or volume capacity (Pmax or IC). At resting levels of drive, respiratory output is 10 to 20% of maximum (e.g., VT, 0.5 L; IC, 3.0 L). At very high levels of stimulation, e.g., heavy exercise (21) or CO₂ rebreathing (22), the ratio increases up to 80% or more. In this patient, despite very poor arterial blood gas values, respiratory output during spontaneous breathing was 10 to 15% of maximum whether expressed as Pdi/Pdi _max or VT/IC. This is comparable to the ratio normally observed at resting levels of drive. This finding is not uncommon in the ICU during weaning trials, and it can be generally attributed to nonspecific respiratory depression with attenuated response to chemical stimuli. What is different in this patient is that respiratory rate showed a brisk response to deteriorating PCO₂, pH, and PO₂. It increased from 7 to 40/ min. Ordinarily, when the rate increases to 40/min during physiologic increases in respiratory drive, motor output per breath is a very high fraction of maximum. This finding, therefore, suggested a specific lesion involving the upper motor neuron responsible for spontaneous breathing as opposed to non-specific respiratory depression. Moreover, because the diaphragm was involved, as evidenced by the low Pdi/Pdi_max, the lesion had to be above the phrenic nucleus in the spinal cord (C3-C5).

Predominant Pathway for Voluntary Activation of Respiratory Muscles

As indicated earlier there are two possible pathways for voluntary activation of the respiratory muscles (Figure 4). The relative contribution of each is currently unknown. Literature on the effect of high cervical cordotomy on voluntary ventilation might be expected to be informative since it theoretically involves a selective lesion to the spinothalamic tracts of the anterior cord, predisposing the adjacent bulbospinal tracts to injury while leaving the dorsal corticobulbar tracts unimpaired. In general, high cervical cordotomy appears to induce a moderate decrease in vital capacity (23). However, most of these studies utilized percutaneous cordotomy, in which the exact size and location of the lesion induced has been shown to be highly variable. The postoperative pain relief achieved in these studies, with secondary effects on vital capacity, further confounds the results. Therefore, quantitative interpretation of these data is problematic in terms of the relative importance of the bulbospinal tracts in voluntary ventilation.

The findings in our patient are of possible relevance in this regard. To the extent that normal voluntary strength in non-respiratory muscles excludes significant involvement of the corticospinal tracts (including those supplying the respiratory spinal motoneurons), the severe weakness of voluntary respiratory efforts (MIP, IC) suggests that the predominant pathway is via activation of the respiratory neurons in the medulla with the activity being conducted to the spinal motoneurons via bulbospinal tracts.

Relation to Previous Reports

We are not aware of reports in humans of brainstem lesions causing disruption of modulation of respiratory motor output with selective sparing of respiratory rhythmogenesis. On the contrary, to the extent that lesions are described, they most commonly involve impairment of automatic rhythm generation, particularly during sleep, with preservation of the capacity for increasing ventilation voluntarily (27). Sarnoff and coworkers (28) described two patients with brainstem poliomyelitis, both of whom had slow irregular breathing when undisturbed but could generate normal minute ventilation on command. Plum and Swanson (29) described 20 patients with bulbar poliomyelitis, and outlined three progressive stages of abnormal respiratory control: Stage 1, consisting of normal breathing during wakefulness, but central apnea during sleep; Stage 2, in which some irregularity of respiratory rhythm appeared during wakefulness, but subjects could initiate and sustain regular breathing voluntarily; Stage 3, grossly irregular breathing during wakefulness over which the subjects had no voluntary control. Irregular breathing or apnea has also been described in relation to brainstem hemorrhage (30), infarction (31), and tumor (27). Thus, based on published reports, the respiratory pacemaker function in humans would appear to be relatively susceptible to injury as compared with other respiratory neurons in the brainstem. This highlights the unusual nature of our case. It is of interest that without the aid of the Pdi waveform in our patient, it was often difficult to be certain that he was in fact breathing since his respiratory excursions were so shallow. Indeed, he at times appeared clinically apneic when in fact the Pdi waveform revealed that he was making regular, albeit weak, efforts to breathe. It thus remains possible that some of the previous case description of apnea with brainstem disease may represent cases similar to our own, with rapid, but very feeble, respiratory efforts being mistaken for apnea.