INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Upper extremity exercise is associated with a significant metabolic and ventilatory cost (1) that is particularly evident in patients with chronic airflow obstruction (CAO) (9, 10) during unsupported arm exercise (UAE) (3, 11). Clinically, this correlates with complaints of dyspnea by patients with CAO during activities of daily living that involve the upper extremities (9, 12).

Respiratory muscle recruitment is altered during UAE in both normal subjects and patients with CAO, with greater contribution to ventilation coming from diaphragmatic and expiratory abdominal muscles and less from chest wall musculature (10). Furthermore, the maximum achieved tidal volume (VT) is lower with UAE than with lower extremity exercise in normal subjects (11) and is less than baseline VT with UAE in patients with CAO (10).

Metabolic and respiratory demands as well as alterations in respiratory muscle recruitment occur with simple arm elevation (AE), a form of UAE, in normal subjects (13, 15) and in patients with CAO (14, 16, 17). Normal subjects respond to this maneuver by increasing diaphragm recruitment (13), whereas patients with CAO recruit expiratory and abdominal muscles as well (14). It has been hypothesized that in these patients diaphragm recruitment is inversely proportional to the degree of hyperinflation and associated diaphragm dysfunction (14, 16). To assess the role of the diaphragm during AE, we chose to examine patients with isolated diaphragm weakness but without airflow obstruction, a group known to have significant alterations in respiratory muscle recruitment at baseline (18, 19). By studying patients with documented diaphragm weakness of varying severity, we hypothesized that: (1) AE should result in similar metabolic and ventilatory demands in patients with isolated diaphragm weakness as normal subjects; (2) respiratory muscle recruitment during AE should directly relate to the degree of diaphragm weakness.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Fifteen patients with documented, isolated diaphragm weakness of varying severity were identified from the Dyspnea Clinic at the University of Michigan Medical Center. Cardiac disease, including systolic or diastolic dysfunction or valvular disease, served as a reason for exclusion. Similarly, pulmonary parenchymal disease defined by radiographic studies or airflow obstruction (FEV₁/FVC < 70%) served as a reason for exclusion. Patients but not control subjects underwent neurologic examination. Upper extremity strength was normal (Medical Research Council Grade 5) proximally and distally in all patients, though quantitative strength testing was not performed. Standard nerve conduction study with repetitive stimulation and electromyography was completed on every patient. No patient was found to have evidence of peripheral neuropathy, defective neuromuscular junction transmission, cervical radiculopathy, or myofiber degeneration.

Descriptive data on the patients in comparison with a group of eight normal subjects are enumerated in Table 1. It was determined that the weight of patients (86.9 ± 4.7 kg) was not statistically different (p = 0.097) from that in the normal subjects (76.5 ± 3.6 kg). As expected, a significant restrictive defect was evident in the patient group as well as lower diaphragm function (Pdi_sniff).

Pulmonary Function

Spirometry, lung volumes, maximal voluntary ventilation (MVV) and diffusing capacity of carbon monoxide (DL_CO) were measured in all patients according to American Thoracic Society recommendations (20). Spirometric values were expressed as a percent of the predicted values published by Morris and colleagues (21), whereas lung volumes were expressed as a percent of the values published by Goldman and Becklake (22). DL_CO values were corrected for alveolar volume (DL_CO/VA) and expressed as a percentage of the values published by Miller and colleagues (23). Diffusing capacity was not routinely measured in the normal subjects.

Respiratory Muscle Testing

Thin-walled latex balloons were passed transnasally under local anesthesia into the midesophagus and stomach. In this way, pleural (Ppl) and gastric (Pg) pressures were measured breath by breath. Each balloon was 10 cm in length and 3.5 cm in circumference (SensorMedics Co., Yorba Linda, CA). The esophageal balloon was placed in the midesophagus and contained 0.5 ml of air, whereas the gastric balloon was filled with 1.5 ml of air. A separate transducer (Validyne Co., Northridge, CA) measured each pressure, and the calibrated output was continuously recorded on a strip chart recorder and on an on-line personal computer.

Transdiaphragmatic pressure (Pdi) was calculated as the electronic difference between Pg and Ppl. Pdi was expressed as the peak level measured from the baseline at end-expiration. Flow was continuously recorded via a calibrated pneumotachograph (CPXII; Warren E. Collins, Inc., Braintree, MA). Phase relationship between pressure and flow was within 5% up to 5 Hz and was subsequently corrected for any phase differences (24). End-inspiratory (Ppli and Pgi) and end-expiratory (Pple and Pge) pressures were measured at points of zero flow. Similarly, Pge and Pgi were defined as the Pg value at end-expiration and end-inspiration, respectively. Respiratory muscle recruitment was described by the slope of Pg/Ppl (Pg/Ppl = Pgi  Pge/Ppli  Pple). A negative slope suggests a predominance of diaphragm activation, whereas a positive slope suggests greater use of the rib cage accessory muscles and muscles of expiration. This technique has been used in describing respiratory muscle recruitment in patients with neuromuscular disease (18) and patients with chronic airflow obstruction (16, 25).

Maximal Pdi was measured using two separate methods. Transdiaphragmatic pressure was recorded during sharp, maximal sniffs (Pdi_sniff), as previously reported (26). In our laboratory a normal Pdi_sniff has been determined to be 122.4 ± 11.5 cm H₂O, which approximates those reported by Mier-Jedrzejowicz and colleagues (26). To enhance specificity, for the purpose of this study diaphragmatic weakness was defined to be a Pdi_sniff below 60 cm H₂O. In addition Pdi was measured at FRC by having the patient perform a maximal inspiratory effort against a partially occluded shutter (Pdi_max). The patients were asked to maximally expand the chest and the abdomen and were coached in the performance of this maneuver until three reproducible results were obtained.

Arm Elevation Protocol

The arm elevation protocol used was similar to that described previously (14). Metabolic and ventilatory parameters were recorded while the patients breathed into a calibrated metabolic cart (Collins CPX; Warren E. Collins, Inc.). Breath-by-breath measures of oxygen consumption (O₂), minute ventilation (E), VT, and breathing frequency (f_b) were thus recorded. The continuous output of the pneumotachograph was recorded and displayed on an on-line personal computer where all data were saved for later analysis. Resting data were collected for 2 min while the patient sat with the back supported at a 90-degree angle and the arms at the side. During subsequent AE the arms were elevated anteriorly with constant back position for 2 min.

Resting data were averaged over the last 15 s of the 2-min resting collection. Data during AE were similarly averaged over the last 15 s of the second minute of AE (AE-2 min).

Data Analysis

Pulmonary function and baseline respiratory data between patients with diaphragm weakness and normal control subjects were compared using Student's two-tailed, two-sample t test, allowing for unequal variances between groups. The assessment of differences in metabolic and ventilatory profiles between patients and normal control subjects at rest versus 2 min of AE was carried out using a linear regression model appropriate for use with repeated measures data (27). The explanatory variables "patient group" (i.e., patient versus normal subject) and "time" (i.e., pre-AE versus post-AE)subsequently referred to as "main effects"were coded so that the interpretation of the main effect of "time" and the "time*group" interaction terms in our model would parallel those from a repeated-measures analysis of variance with one between-groups factor (i.e., patients versus normal subjects) and one repeated-measures factor (i.e., time, in this case pre-AE versus post-AE). In particular, in our statistical analysis,

 tests for the main effect of patient group correspond to testing whether there are baseline differences (i.e., pre-AE) in response between patients and normal subjects;

 tests for the main effect of time correspond to testing whether there is an overall similarity of response to AE in one or both patient groups (i.e., whether the response variable in patients, normal subjects, or both tended to rise or fall on average after 2 min of arm elevation);

 tests of the time*group interaction correspond to testing whether the response to AE differs by patient group after adjusting for any baseline (i.e., pre-AE) differences in the level of response.

Further adjustments for possible differences in weight were carried out by including weight as a covariate in our regression model; these results are reported where warranted.

Comparisons between resting and AE-2 min metabolic, ventilatory, and respiratory muscle data within patients were done using paired t tests, with a subset of the same analyses repeated for normal subjects. In addition to assessing differences in Pdi, VT, and f_b profiles in patients versus normal subjects, a separate analysis of each of these variables was carried out within patients only. Specifically, we compared patients having a Pdi_sniff at or below 30 cm H₂O versus above 30 cm H₂O using repeated-measures analyses similar to those described above. A similar analysis was performed for those patients with a < 10% rise in VT during AE and those experiencing a > 10% increase in VT during AE.

The percentage change in VT and f_b during AE were compared with Pdi_sniff and Pdi_max using a linear regression analysis. Simple linear regression was also used to describe the relationship between Pdi_sniff and the percentage rise for each of these three variables in patients. Where tabulated, data are expressed throughout as mean ± SE. A p value < 0.05 was considered statistically significant.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The physiologic data in 15 patients with diaphragmatic weakness and eight normal control subjects are enumerated in Table 2. As expected the patients demonstrated moderately severe restriction with a normal FEV₁/FVC. The DL_CO/VA was normal in the patient groups. Pdi_max is seen to be significantly lower in the patient group, which is to be expected.

Using t tests, it was determined that there was no evidence of a difference in resting carbon dioxide production (CO₂), E, and VT between patients and normal control subjects, whether or not we additionally adjusted for differences in weight. We found strong evidence of an overall response to AE in the patients and/or the normal control subjects (p < 0.0001). The pattern of response for each group during AE was similar for CO₂, E, and VT. The conclusions for O₂ are similar, although there is a suggestion of a difference in O₂ (p = 0.083) before AE between patients and normal subjects. In contrast, resting f_b approached statistical significance in the patients compared with the normal subjects (p = 0.056). Further adjusting for possible differences in weight did not qualitatively alter these results.

The aerobic and ventilatory response at baseline and after 2 min of AE in both groups is summarized in Table 3, and it provides some post-hoc analysis using paired t tests for the difference between rest and AE within patients and, separately, within normal subjects. We found a statistically significant rise in O₂, CO₂, E, and VT in both groups between rest and 2 min of arm elevation, confirming results reported earlier. In contrast, the rise in f_b was found to be statistically significant only in the patient group (p = 0.002 versus p = 0.264 for normal subjects). This, taken along with the nearly significant repeated measures test (p = 0.076), suggests that f_b in patients and normal subjects may respond differently to AE.

The change in Pdi during 2 min of AE is illustrated in Figure 1. One observes a rise in Pdi from rest to the second minute of AE in both groups. Our repeated-measures analysis shows that there are statistically significant differences between patients and normal subjects at baseline (p < 0.0001) and that there is an increase to AE in both groups (p < 0.001). We found no interaction, however (p = 0.18), indicating that although patients and normal subjects have different baseline levels of Pdi, both exhibit a similar change in Pdi to AE relative to pre-AE levels.

View larger version (9K): [in this window] [in a new window]   Figure 1.   Transdiaphragmatic pressure (Pdi) at rest and after 2 min of arm elevation (AE) in normal control subjects (closed circles) and in patients with diaphragm weakness (closed diamonds).

The change in Pdi from rest to the second minute of AE as an absolute pressure compared with the baseline Pdi_sniff for all patients with diaphragm weakness is demonstrated in Figure 2. A strong correlation was seen between a rise in Pdi and a greater Pdi_sniff (r = 0.73, p = 0.002). Repeated-measures analyses indicated a significantly different response profile for Pdi in patients with Pdi_sniff below 30 cm H₂O versus those with Pdi_sniff above 30 cm H₂O (p < 0.001 for both baseline differences and the test of interaction). Post-hoc analyses showed that the increase in Pdi for those patients with Pdi_sniff at or below 30 cm H₂O is considerably lower during 2 min of arm elevation than it is for those with Pdi_sniff above 30 cm H₂O (0.61 ± 0.14 versus 3.53 ± 0.52, p < 0.001).

View larger version (10K): [in this window] [in a new window]   Figure 2.   Correlation between the rise in Pdi during 2 min of arm elevation and Pdisniff in patients with diaphragmatic weakness.

Pple profiles in patients with diaphragm weakness and normal subjects were also compared using a repeated-measures analysis as described earlier. Although we found strong evidence of an increase during AE (p < 0.001), no significant baseline differences between patients and normal subjects (p = 0.224) or difference during AE (p = 0.631) were found. These results indicate that patients and normal subjects respond similarly to AE with regard to Pple. Post-hoc analyses confirmed these results. The slope of Pg/Ppl was inversely correlated with Pdi_sniff in patients and normal subjects at rest (r = 0.74, p < 0.0001) and during AE (r = 0.66, p = 0.0006) as illustrated in Figures 3A and 3B, respectively. When the patient data were analyzed separately, significant correlations remained between Pg/Ppl and Pdi_sniff (r = 0.65, p = 0.01) at rest and during AE (Pdi_sniff: r = 0.55, p = 0.02). Importantly, the resting slope of Pg/Ppl in the patient group was 0.19 ± 0.26, with a statistically significant change over baseline to AE (p = 0.02).

View larger version (11K): [in this window] [in a new window]   Figure 3.   (A) Correlation between Pg/Ppl at rest and Pdisniff in patients (circles) and normal subjects (squares) with diaphragm weakness. (B) Correlation between Pg/Ppl during arm elevation and Pdisniff in patients (circles) and normal subjects (squares) with diaphragm weakness.

The percentage change can be seen in VT between rest and 2 min of arm elevation plotted versus Pdi_sniff (Figure 4A) and Pdi_max (Figure 4B) for patients with diaphragm weakness. A significant positive correlation is evident with greater rise in VT seen with larger Pdi_sniff (r = 0.65, p = 0.009) and Pdi_max (r = 0.81, p = 0.0002). The similarities here are likely due to the high correlation between Pdi_sniff and Pdi_max (r = 0.85, p < 0.001). Interestingly, a similar relation was not seen for the change in f_b. Repeated-measures analyses similar to those described earlier indicate a significantly different response profile for VT (p = 0.60 for baseline, p = 0.017 for interaction), but not for f_b for patients with Pdi_sniff below 30 cm H₂O versus those with Pdi_sniff above 30 cm H₂O. Post-hoc analyses showed that the percentage increase in VT with 2 min in arm elevation was significantly lower in those patients with Pdi_sniff below 30 cm H₂O than in those with Pdi_sniff above 30 cm H₂O (0.062 ± 0.033 versus 0.232 ± 0.042, p = 0.007). However, the difference in percentage increase in f_b with 2 min of arm elevation between these two groups of patients was not statistically significant (0.237 ± 0.089 versus 0.12 ± 0.035, p = 0.26). To further investigate the changes noted in VT during AE, patients were separated into those with a rise in VT < 10% and those with a rise  10%. Those patients with a lesser rise in VT had a lower resting VT (0.49 ± 0.06 versus 0.67 ± 0.07 L, p = 0.02). In addition those patients with a greater than 10% rise in VT after AE had a greater Pdi_max (46.3 ± 6.1 versus 24.9 ± 6.0 cm H₂O, p = 0.02), although no difference was seen in Pdi_sniff (p = 0.13).

View larger version (10K): [in this window] [in a new window]   Figure 4.   (A) Correlation between the percentage rise in VT between rest and 2 min of arm elevation and Pdisniff in patients with diaphragm weakness. (B) Correlation between the percentage rise in VT between rest and 2 min of arm elevation and Pdimax in patients with diaphragm weakness.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Unsupported arm exercise results in well described metabolic and respiratory demands in normal subjects (13, 15) and in patients with chronic airflow obstruction (14, 16, 17). In patients with CAO respiratory muscle recruitment varies depending on the degree of hyperinflation and presumably the negative effect on diaphragm function (14, 16). Although not previously described, the role of diaphragmatic function during AE would be ideally studied in patients with isolated diaphragmatic weakness but no confounding airflow obstruction or parenchymal lung disease. In this study, we hypothesized that the effect of AE in patients with isolated respiratory muscle weakness compared with normal subjects would confirm the importance of diaphragm function during this simple activity. We demonstrated: (1) a similar metabolic and ventilatory demand during AE in patients with diaphragm weakness compared with normal control subjects, (2) a rise in f_b during AE in patients but not in normal control subjects, (3) a difference in baseline levels of Pdi between normal control subjects and patients with diaphragm weakness, with similar response profiles thereafter, (4) a strong negative linear correlation between the rise in Pdi during AE in patients with diaphragm weakness and baseline diaphragm strength as defined by Pdi_sniff, and (5) a positive linear correlation between rising VT with AE and a larger Pdi_sniff and Pdi_max in the patient group, confirming an altered respiratory recruitment pattern during arm elevation that directly correlates with diaphragmatic strength.

The patients in our study demonstrated a 19.9% rise in O₂ and a 32.2% increase in CO₂ after 2 min of arm elevation anteriorly. The degree and pattern of rise was similar to that demonstrated by our normal control subjects. The metabolic requirement of this seemingly trivial activity has been demonstrated in normal subjects (13) and in patients with severe CAO (14, 17). Similarly, it is known that, for a given work load, arm cranking (a form of supported arm exercise) results in a higher O₂, E, heart rate, and lactate production than leg cycle ergometry (1, 2, 7), although peak values are generally lower for arm exercise (4, 6, 8, 28). Unsupported arm exercise results in a lower exercise endurance than supported arm exercise despite lower peak O₂, E, and heart rate (9). We found no significant difference between patients and normal subjects in either resting E or response to AE. The rise seen in the control group was similar to that previously reported (13). The increase in E is apparently generated largely by a rise in VT in normal subjects, as was also shown by Couser and colleagues (13). However, although our patients with diaphragmatic weakness responded to AE similarly to normal patients with respect to VT, they exhibit a significant rise in f_b. This novel finding is similar to that seen in patients with chronic obstructive pulmonary disease (14, 17). We suspect this reflects the impaired ability to generate VT in patients with diaphragm weakness as there was a positive linear rise in VT rise with increased diaphragm strength as measured by Pdi_sniff or Pdi_max. This is also supported by the analysis of VT and f_b among patients having a Pdi_sniff < 30 cm H₂O versus those having Pdi_sniff > 30 cm H₂O. The latter group is seen to exhibit patterns of ventilatory muscle recruitment more similar to that of normal subjects than the former (i.e., as reflected in their significantly different VT response profiles but similar f_b profiles). The level of 30 cm H₂O was chosen to separate the groups as this pressure has been previously described as the minimum pressure needed to overcome the hydrostatic pressure of the abdominal contents (26). Furthermore, differences were noted when patients were grouped by the response of VT during arm elevation. Those subjects with a lesser rise in VT demonstrated significantly lower, resting maximal diaphragmatic pressures.

Arm activity resulted in clear changes in ventilatory muscle recruitment. The current study confirmed the immediate rise in Pdi demonstrated by normal subjects during unsupported arm elevation, suggesting greater diaphragm recruitment. Patients with diaphragm weakness demonstrated a qualitatively similar response profile to normal control subjects, though they began at a significantly lower level of resting Pdi. The change in Pdi from rest to arm elevation in patients was shown to relate directly to diaphragm strength. Similarly, we have established that those patients with less diaphragm strength (Pdi_sniff < 30 cm H₂O) (26) demonstrate a significantly lesser rise in Pdi during arm elevation than those patients with Pdi_sniff > 30 cm H₂O.

The pattern of respiratory muscle recruitment in the patients, as reflected by the slope of Pg/Ppl, became progressively more positive during AE in those patients with lower Pdi_sniff. This confirms greater recruitment of rib cage accessory muscles and muscles of expiration during AE in those patients with greater diaphragm weakness (25, 26). In patients with CAO a similar response has been demonstrated during arm elevation that correlates closely with the degree of hyperinflation (16). Epstein and colleagues (16) hypothesized that in obstructed patients greater hyperinflation was associated with greater diaphragm dysfunction and therefore an altered pattern of respiratory muscle recruitment during arm elevation. Our data in patients with isolated diaphragm weakness strongly support the importance of diaphragmatic strength during unsupported arm activity.

The results for Pple during arm elevation indicate similar baseline levels for patients and normal subjects. A rise in the level of Pple in response to AE was observed in both patients and control subjects, the mean rise in patients being larger than the mean rise in normal subjects (0.72 versus 0.53); however, no evidence of statistically significant difference was found. It is possible that the lack of a statistically significant difference is due at least in part to an abnormally large change observed during AE in the heaviest of the normal subjects. Importantly, relative to the seven other normal patients, the heaviest normal subject (203 pounds) exhibited a rise in Pple of 2.63, roughly 6 standard errors (or 3.5 standard deviations) above the mean rise in Pple of the remaining seven patients (0.24 ± 0.16). If we excluded this patient and repeated the analysis a fivefold decrease in the p value for testing the interaction term (0.63 versus 0.12) resulted. Although these results do not confirm that patients and normal subjects differ in Pple response during AE, the impact on the results of a single outlying subject indicates that further investigation is warranted. The results observed for normal subjects are otherwise qualitatively consistent with previous findings (13). The response observed in patients with diaphragm weakness is similar to that of patients with severe CAO (14). In this population air-trapping has been hypothesized to lead to impaired diaphragmatic function. Interestingly, it has recently been shown that supported arm activity does not result in increased lung volume in normal subjects but does so in patients with more severe cystic fibrosis (29). This occurred at a E approximately double to that achieved by our patients with diaphragmatic weakness during unsupported arm elevation, but such changes in lung volume were likely insignificant in our patients and similar to the minimal changes noted in our previous studies in patients with severe CAO (14). On the other hand, Martinez and colleagues (14) also hypothesized that expiratory muscle recruitment would therefore act to preserve inspiratory muscle function in patients with CAO (14). This response would appear qualitatively similar to the changes noted in our patients with diaphragm weakness.

This study of patients with isolated diaphragm weakness highlights the importance of diaphragm function during unsupported arm activity. In these patients, the simple act of raising the arms anteriorly for 2 min was associated with a significant metabolic and ventilatory load. This ventilatory load appears to be mediated primarily through a rise in f_b, not VT. The rise in Pdi during arm elevation related directly to baseline diaphragm strength. Furthermore, there was evidence of accessory muscle recruitment during arm elevation in patients with diaphragm dysfunction. These data confirm the importance of diaphragm function during unsupported arm activity and shed further light on the limitations noted by patients with diaphragm weakness as well as those with CAO and its associated diaphragm dysfunction. These data support the observations that patients with presumed diaphragmatic dysfunction (in the setting CAO, for example) experience symptomatic relief with support of upper extremity musculature (9). A similar effect would be expected in patients with isolated diaphragmatic weakness.