Introduction

Patients with COPD experiencing an exacerbation of their disease may require invasive or non-invasive mechanical ventilation to control acute hypercapnic respiratory failure [1]. The goal of mechanical ventilation is to reduce CO2 by augmenting alveolar ventilation, thereby stabilizing arterial pH until the underlying problem can be reversed [1].

Hypercapnic respiratory failure can occur when the drive to breathe is impaired, and when the respiratory muscles cannot cope with the inspiratory workload [2]. The latter condition, i.e., load/capacity imbalance, was proved to be the ultimate cause of hypercapnic respiratory failure in healthy subjects undergoing inspiratory resistive loading in the laboratory [3] and also in long-term ventilator-dependent patients [4–6].

The mechanisms responsible for hypercapnic respiratory failure during an acute exacerbation of COPD have been addressed in studies conducted in the intensive care unit setting [7–12]. In these studies, a rapid shallow breathing pattern in the presence of an increased respiratory workload was detected almost invariably [7–12]. In none of these studies, however, was the load/capacity imbalance and pattern of breathing examined during the very early phases of the hospitalization. The aim of the current investigation was thus to characterize, for the first time, the physiological defects associated with acute hypercapnic respiratory failure during the early phases of hospitalization in patients with COPD experiencing an acute exacerbation. Some of the results of this study have been previously reported in abstract form [13].

Methods

The study was conducted according to the guidelines of the Declaration of Helsinki.

A total of 24 patients with severe COPD were enrolled in the study. A total of 16 patients were consecutively admitted to Emergency Department and treated in the Division of Emergency Critical Care of our Institution because of acute exacerbation of COPD. The assessment of the severity of the disease and the decision on the admission to the hospital by the physician in charge was based on the clinical history, symptoms, physical examination, arterial blood gases measurements, and other laboratory tests. Nine of them were in acute respiratory acidosis and were submitted to noninvasive positive-pressure ventilation (NPPV group) on the judgment of the primary physician. The remaining seven patients did not require any ventilatory support and were treated with the conventional medical therapy only (SB group). We compared acute patients with a group of eight COPD patients with chronic ventilatory failure in stable clinical conditions (IPPV group), admitted to the Chronic Ventilator Unit of our Institution and receiving mechanical ventilation through a tracheostomy cannula (tracheal cannula size, 9 ± 1 mm ID). They had been previously submitted to mechanical ventilation because of exacerbation of COPD, and classified by the caring physicians as ventilator-dependent after several weaning failures. At the time of the study they had been ventilated for an average of 30 ± 8 days.

Patients’ characteristics and lung function data obtained from the accompanying clinical records on admission to the hospital are reported in Table 1.

Table 1 Patient characteristics at the time of the study
Full size table

Figure 1 shows the inclusion and the exclusion criteria for the study [14].

Fig. 1
figure 1

Inclusion criteria for the study and the experimental procedure

Full size image

Measurements

Physiologic signals, namely, flow (V′), volume (V), esophageal (P pl), gastric (P ab), airway opening (P ao), transpulmonary (P L) and transdiaphragmatic (P di) pressures were obtained. Dynamic lung compliance (C Ldyn) and pulmonary resistance at mid-inspiratory volume (R Linsp) were computed from P L, V, and V′ records, respectively; dynamic intrinsic positive end-expiratory pressure (PEEPidyn) was measured [10]. The f/V T ratio was computed [15]. The magnitude of the inspiratory muscle effort was estimated from the pressure–time product for the inspiratory muscles (PTP pl) and for the diaphragm (PTP di) [16].

P pl and P di records were used to detect the occurrence of ineffective inspiratory efforts [6].

Maximum esophageal, transdiaphragmatic and airway opening (\( P_{{{\text{pl}}_{ \max } }}, \; P_{{{\text{di}} _{\max } }}, \) and MIP, respectively) pressures were measured by means of the Mueller maneuvre [16].

The tension-time index of the diaphragm (TTdi) [17] and of the respiratory muscles (TTI) [18] were computed.

To assess the pattern of ventilatory muscle recruitment, the ΔP ab/ΔP pl ratio (i.e., the ratio of the difference in P ab and P pl between start- and end-inspiration determined by the point of zero-flow on the flow signal) was measured [19], and corrected for the expiratory muscle contraction, if present (cΔP ab/cΔP pl) [20].

The neuromuscular drive was estimated by the occlusion pressure during the first 100 ms of an occluded inspiration [8]. It was measured (Fig. 2): (a) on P ao and P pl tracings (P 0.1 and P 0.1pl, respectively) at the onset of the occluded inspiration, as indicated by the fall of mouth pressure below atmospheric pressure, and (b) on P pl tracing (ΔP pl0.1) at the onset of the inspiratory effort as indicated by the start of P pl fall. Finally, we compared the start of P pl decrease to the synchronous start of P di increase to better define the onset of inspiration.

Fig. 2
figure 2

Experimental record illustrating the method used to determine P 0.1. \( \dot{V}_{{\text{E}}}\) flow, exp expiration, P ao airway opening pressure, P pl esophageal pressure, P 0.1 occlusion pressure measured on P ao tracing from the start of inspiration, P 0.1pl occlusion pressure measured on P pl tracing from the start of inspiration, ΔP pl0.1 occlusion pressure measured on P pl tracing from the start of inspiratory effort, Δt delay time between the start of the inspiratory effort and the onset of the occluded inspiration

Full size image

To determine if respiratory mechanics (elastic, resistive, and threshold load) were associated with increased neuromuscular drive we computed the correlation between P 0.1 and ΔP pl0.1 with C Ldyn, R L, and PEEPidyn.

Data are reported as mean ± SD.

Protocol

Measurements were performed during a period of spontaneous breathing (Fig. 1 for the experimental apparatus and the procedure). Acute patients were studied within the 24th h from the admission to the Division of Emergency Critical Care.

Physiologic signals were collected for 3 min and the last of these 3 min was used for data analysis.

Data analysis

Comparisons between groups were performed using a two-stage approach: first, a Kruskall–Wallis one-way nonparametric analysis of variance was performed to detect any overall significant difference (P < 0.05) among the three groups; second, if the nonparametric ANOVA showed overall differences, these were investigated by performing the Mann–Whitney U tests. Also in this case a global significance level P < 0.05 was selected. Spearman rank-correlation was employed to calculate the correlation between physiological variables. Two-sample paired sign test was used to detect the difference between Δ0.1pl and P 0.1 P pl [21].

Results

The time interval between the admission to the Division of Emergency Critical Care and the measurements was 10 ± 7 and 9 ± 4 h, for NPPV and SB groups, respectively. For NPPV group the admission coincided with the start of noninvasive ventilation.

All the patients in the NPPV group were successfully weaned after 3 ± 1 days from the beginning of noninvasive ventilation and were discharged to the Internal Medicine Department of the Hospital.

None of the patients managed with medical therapy required mechanical ventilation either invasive or noninvasive.

All patients in the IPPV group remained ventilator-dependent. Seven of them were eventually discharged to a home mechanical ventilation program, the remaining one died because of generalized sepsis 12 days after the study.

Table 2 reports the physiologic data collected during spontaneous breathing in the patients enrolled into the study.

Table 2 Breathing pattern, neuromuscular drive, respiratory mechanics, force-generating capacity of the respiratory muscles, and inspiratory muscle effort
Full size table

Breathing pattern

Both groups of mechanically ventilated patients exhibited a rapid shallow breathing pattern (Table 2). No patient in the three groups showed ineffective inspiratory efforts.

Neuromuscular drive

The delay between ΔP pl0.1 and P 0.1pl (Fig. 2) was 139 ± 67, 123 ± 44, and 139 ± 52 ms, for NPPV, SB, and IPPV groups, respectively. ΔP pl0.1 was less than P 0.1pl in both NPPV and IPPV groups (P < 0.05). P 0.1 and P 0.1pl were increased in the three groups of patients (n.v: ≤2 cmH2O [22]), being significantly greater in NPPV and IPPV groups compared to SB group (Table 2). The relationship between P 0.1 with P 0.1pl and between P 0.1pl with ΔP pl0.1 are shown in Figs. 3 and 4, respectively. ΔP pl0.1 was greater in IPPV compared both to NPPV and SB groups, respectively (Table 2). This difference between the two groups of mechanically ventilated patients disappeared (p:n.s.) after the subtraction from ΔP pl0.1 of the synchronous P ab decrease, if present, even though the mean change was small: from 4.5 ± 1.6 to 4.2 ± 1.9 cmH2O and from 3.0 ± 1.6 to 2.9 ± 1.6 cmH2O, for IPPV and NPPV groups, respectively.

Fig. 3
figure 3

Comparison between P 0.1 and P 0.1pl with the Bland and Altman analysis. Gray circles SB group, open circles NPPV group, black circles IPPV group. Average P 0.1 average value of each pair of measurements (P 0.1 + P 0.1pl); ΔP 0.1 difference between the two values, d mean bias, s standard deviation of the individual differences, d + 1.96 s upper limit of agreement, d – 1.96 s lower limit of agreement between P 0.1 and P 0.1pl, P 0.1: 4.3 ± 1.7 cmH2O, P 0.1pl 4.4 ± 1.9 cmH2O, d 0.1 ± 0.8 cmH2O, p not significant. A good agreement was observed between the two variables

Full size image
Fig. 4
figure 4

Relationship between P 0.1 values measured on the esophageal tracing at the start of the inspiration (P 0.1pl) and at the start of the inspiratory effort (ΔP pl0.1). Gray circles: SB group; open circles: NPPV group; black circles: IPPV group. Dashed line: identity line. The two measures were significantly different (4.4 ± 1.9 vs 3.0 ± 1.8 cm H2O, for P 0.1pl and ΔP pl0.1, respectively, P < 0.05). See text for the explanation

Full size image

Lung mechanics

Lung mechanics variables are reported in Table 2. The inspiratory workload was greater in patients affected by ventilatory failure compared to SB group. PEEPidyn was less in NPPV compared to IPPV group.

A significant correlation was observed between ΔP pl0.1 and both PEEPidyn and R Linsp (P < 0.05, Rho 0.589 and 0.516, respectively), but not with CLdyn (p:n.s.), whereas each of them was correlated with P pl0.1 (P < 0.05, Rho 0.476, 0.557, and −0.628, for PEEPidyn, R Linsp, and CLdyn, respectively). Finally, a significant correlation was observed between f/V T ratio with C Ldyn (P < 0.05, Rho −0.578), but not with both PEEPidyn and R Linsp (p:n.s).

Respiratory muscle function

\( P_{{{\text{di}}_{ \max}} } \) was less in ventilated patients compared to the SB group with a concomitant increase of diaphragmatic and accessory inspiratory muscle effort (Table 2). The percentage of PTP pl spent to counterbalance PEEPidyn was 21 ± 10, 21 ± 16 29 ± 12%, in NPPV, SB and IPPV groups, respectively. The ventilatory muscle recruitment is shown in Fig. 5. A significant negative correlation was observed between ΔP ab/ΔP pl ratio and \( P_{{{\text{di}}_{ \max}} } \) (P < 0.05, Rho −0.621). The rise in P ab during expiration was 4.0 ± 2.9, 0.6 ± 0.9, and 1.7 ± 2.3 cm H2O for NPPV, SB, and IPPV groups, respectively (p:n.s.). Compared to ΔP ab/ΔP pl ratio, cΔP ab/cΔP pl ratio was more negative (from −0.028 ± 0.335 to −0.167 ± 0.325, from −0.550 ± 0.262 to −0.562 ± 0.268, and from 0.020 ± 0.247 to −0.061 ± 0.282, for NPPV, SB, and IPPV groups, respectively).

Fig. 5
figure 5

Ventilatory muscle recruitment. ΔP ab /ΔP pl ribcage and expiratory muscles recruitment index. Horizontal bars mean values. Both NPPV and IPPV showed a greater activation of ribcage and expiratory muscles in comparison to SB group (mean values −0.028  ± 0.353, 0.020 ± 0.061, and −0.550 ± 0.262, for NPPV, IPPV, and SB groups, respectively; *P < 0.05, SB versus both NPPV and IPPV groups)

Full size image

Load/capacity balance

In both NPPV and IPPV groups, mean TTdi was above 0.15, whereas it was well below this threshold in SB group (Table 2). A significant correlation was found between f/V T ratio with TTdi and TTI (P < 0.05, Rho 0.756 and 0.786, respectively). The former is shown in Fig. 6.

Fig. 6
figure 6

Relationship between breathing frequency to tidal volume (f/V T) ratio and tension-time diaphragmatic index (TTdi). Gray circles SB group, open circles NPPV group, black circles IPPV group. Dashed lines cut-off value of 105 for f/V T ratio (drawn from Ref. [15]), and of 0.15 for TTdi (drawn from Ref. [3]). A significant correlation was observed between the two measures (Rho 0.756, P < 0.05). Six patients in the NPPV group, all the patients in the IPPV group, and nobody in SB group were above the threshold limits for TTdi and f/V T ratio, respectively, indicating impending ventilatory failure. For further explanations, see text

Full size image

Discussion

Determinants of ventilatory failure

This is the first study on the physiological defects associated with acute ventilatory failure during the early phases of hospitalization in patients with COPD experiencing an acute exacerbation. We found that the patients with acute ventilatory failure have a significant imbalance between decreased capacities of the respiratory muscles to generate pressure and increased respiratory load [7–12].

Respiratory controller failure, reduced respiratory muscles strength and increased respiratory workload are the three basic mechanisms potentially leading to ventilatory failure [23]. In our patients with acute ventilatory failure, we did not find any “absolute” impairment of the respiratory controller, even though a “relative” impairment (“central wisdom”), operated to prevent fatigue, could not be excluded [23]. By contrast, both reduced inspiratory muscle strength and an increased inspiratory muscle workload were present (Table 2). Malnutrition, electrolitic disturbances, steroids administered to treat the acute exacerbation, as well as dynamic hyperinflation could impair the inspiratory muscle strength in severe COPD patients [23]. In both NPPV and IPPV groups, only dynamic hyperinflation was present, as indicated by the increased levels of PEEPi (Table 2). As a matter of fact, in NPPV group BMI (Table 1) and serum electrolytes were in the normal range, while the timing and the total amount of steroids administered to the patients precluded the presence of steroid-related respiratory muscle function impairment [24]. Thus, during the early phases of hospitalization, the acute increase of inspiratory workload (Table 2) seemed predominant in promoting an unfavorable load/capacity balance in COPD patients on NPPV similar to ventilator-dependent patients [6], dynamic hyperinflation playing a key role on both sides of the balance itself.

Difference between acute and chronic patients

Both groups of acute and chronic mechanically ventilated patients showed a similar load/capacity imbalance, and a similar partitioning of the inspiratory effort between the inspiratory muscles (Table 2; Fig. 5). Inline with literature, the recruitment of the accessory inspiratory muscles could compensate for the decreased diaphragmatic capacity to generate pressure [19, 20].

We did not find any difference in terms of expiratory muscle activity among groups. However, we collected data in ventilated patients after few minutes of unsupported breathing (Fig. 1), whereas the activation of expiratory muscles can occur relatively late both in normal subjects submitted to experimental procedures [25, 26] and in chronic patients who fail a weaning trial [20].

Among respiratory mechanics variables, only PEEPidyn was different between acute and chronic ventilated patients. It was less in NPPV group, suggesting a reduced dynamic hyperinflation (Table 2), compared to IPPV group. PEEPi provides an inspiratory threshold load that must be counterbalanced by the patients’ inspiratory muscles before starting inspiration [27]. According to literature [16, 28, 29], we found that PEEPidyn accounted for about 20 and 30% of the total ventilatory workload, in acute and chronic COPD patients, respectively. The lower PEEPidyn in NPPV compared to IPPV group might partly explain their different outcome, i.e., a successful weaning versus ventilator dependence, respectively.

Neuromuscular drive

Despite its limitations, P 0.1, namely the pressure measured at airway opening during the first 100 ms of an occluded inspiratory effort, is used as an estimate of the neuromuscular drive to breathe [30], with the assumption that during the first 150 ms of inspiration it cannot be influenced neither by peripheral reflexes nor by patient behavior.

We did not find any substantial difference between P 0.1 and the synchronous decrease in esophageal pressure (P 0.1pl) (Fig. 3), even though different and independent factors can influence the two measurements [30–33].

Due to dynamic hyperinflation, COPD patients often show a delay of the commencement of inspiration, detected at the airway opening, compared to the start of the inspiratory effort, detected on P pl tracings (Fig. 2). In our study, this delay was about 140 ms in both groups of ventilatory failure patients. It follows that P 0.1 value could be influenced by peripheral reflexes [30]. As a consequence, in the presence of dynamic hyperinflation, the fall in esophageal pressure during the “true” first 100 ms of inspiration (ΔP pl0.1, Fig. 2) might estimate more accurately the neuromuscular drive than both P 0.1 and P pl0.1. At variance, P 0.1 might estimate better than ΔP pl0.1 the influence of the respiratory mechanics on the respiratory center output, being both correlated to the overall components of the inspiratory workload and potentially affected by peripheral reflexes.

Finally, part of ΔP pl0.1 was generated by relaxation of the expiratory muscles [30] (as indicated by the P ab decrease at the start of the inspiratory effort): this ΔP pl0.1 component was slight, but not negligible, since its correction eliminated the statistical difference between NPPV and IPPV groups.

Rapid shallow breathing (clinical implications)

Among other factors (chemoreceptor stimulation, operating lung volume, peripheral reflexes, altered respiratory motoneuron discharge, and cortical influences) [34], increased respiratory mechanical load in normal subjects [35] and in patients during weaning can generate rapid shallow breathing [4]. This latter characterized NPPV and IPPV groups, both having a similar load/capacity imbalance.

Although the mechanism subtending shallow breathing is largely unknown, its setting is usually considered a strategy to decrease work per breath, and, hence, avoid fatigue [34]. However, shallow breathing is invariably matched with increased breathing frequency that prevents the inspiratory work per minute to decrease [34]. Moreover, since the tension-time indexes seem a marker of the risk of respiratory muscle fatigue [3, 17], the f/V T ratio should have a negative correlation with them to serve as a compensatory strategy [34]. This was not the case in previous works showing no correlation between f/V T and tension-time indexes [4, 5], and in the present study, where a significant positive correlation was found (Fig. 6).

Finally, rapid shallow breathing seems to represent a maladaptive response, since a decrease in V T increases dead-space ventilation and hypercapnia [34].

Our data emphasize the interplay between high f/V T ratio, the overall increased mechanical workload, and the onset of acute ventilatory failure in COPD patients treated in the emergency department. Thus, we speculate that finding in such patients high f/V T levels should warrant the institution of an aggressive therapy aimed at reducing the mechanical constraints to breathing pattern (e.g., dynamic hyperinflation which forces patients to breathe closer to the upper portion of the volume/pressure curve, lung interstitial edema, atelectasis due to airway secretions accumulation…), in parallel to the start of mechanical ventilation. Whether the f/V T ratio could be useful in monitoring the therapy-induced changes of mechanical workload is in our opinion an appealing issue, not assessed in the present study.

Theoretically, in ventilator-dependent COPD patients the f/V T ratio could be underestimated in presence of neural and mechanical respiratory frequency dissociation induced by ineffective inspiratory efforts [36]. They were found in ventilator-dependent patients both during mechanical ventilation [36] and unsupported breathing [6], whenever the pressure generated by the inspiratory muscles could not counterbalance PEEPi, and/or the respiratory drive was reduced [36]. This was not the case of successfully weaned NPPV acute patients (see PEEPidyn and P 0.1, Table 2). This suggests that the finding of low f/V T values due to ineffective efforts may be associated with increased risk of ventilator dependence.

Limitations of the study

This study has some limitations worthy of discussion.

First, NPPV patients were studied within the first 24th h from the start of ventilation [37]. We chose this range of time to gain strong patient’s cooperation, difficult to obtain during the early hours of the ventilatory failure management. Accordingly, it can be hypothesized that the medical therapy, started immediately by the primary physician, could have lowered the inspiratory workload before the study was performed. That could explain the three patients in group NPPV below the threshold limit of 105 for f/V T ratio, two out of them also showing a TTdi below the threshold limit of 0.15 (Fig. 6).

Second, SB group patients showed a respiratory frequency of 16 ± 3.8 breaths/min, slower than it could be expected [23]. During the study both NPPV and SB patients breathed through a mouthpiece, condition that can influence the breathing pattern [38]; moreover, measurements were performed during O2 supply to keep saturation above 93%; finally the time interval between the admission to the hospital and the measurements could be of relevance.

Third, a reliable calculation of TTdi depends on an accurate measurement of diaphragmantic strength. In particular, critically ill patients may be unable to completely activate the diaphragm during the Mueller maneuvre [39]. Thus, the underestimation of \( P_{{{\text{di}}_{ \max}} } \) would produce an overestimated TTdi. However, patients in IPPV group did not have sufficient coordination to perform the “combined maneuvre” [40].

Conclusions

In conclusion, we found that: (1) a load/capacity imbalance was present in the early phases of an acute exacerbation in patients with COPD; (2) this imbalance was associated to rapid shallow breathing, similarly to chronic ventilator-dependent patients; (3) the presence of PEEPi accounted for a significant portion of the inspiratory workload in both acute and chronic severe COPD patients, suggesting a severe impairment of respiratory mechanics that can induce failure of unsupported breathing; 4) in severe COPD patients peripheral reflexes could influenced P 0.1 assessment.