Respiratory failure can arise from an abnormality in any of the components of the respiratory system, including the airways, alveoli, central nervous system (CNS), peripheral nervous system, respiratory muscles, and chest wall. Patients who have hypoperfusion secondary to cardiogenic, hypovolemic, or septic shock often present with respiratory failure.
Ventilatory capacity is the maximal spontaneous ventilation that can be maintained without development of respiratory muscle fatigue. Ventilatory demand is the spontaneous minute ventilation that results in a stable PaCO2.
Normally, ventilatory capacity greatly exceeds ventilatory demand. Respiratory failure may result from either a reduction in ventilatory capacity or an increase in ventilatory demand (or both). Ventilatory capacity can be decreased by a disease process involving any of the functional components of the respiratory system and its controller. Ventilatory demand is augmented by an increase in minute ventilation and/or an increase in the work of breathing.
The act of respiration engages the following three processes:
Transfer of oxygen across the alveolus
Transport of oxygen to the tissues
Removal of carbon dioxide from blood into the alveolus and then into the environment
Respiratory failure may occur from malfunctioning of any of these processes. In order to understand the pathophysiologic basis of acute respiratory failure, an understanding of pulmonary gas exchange is essential.
Respiration primarily occurs at the alveolar capillary units of the lungs, where exchange of oxygen and carbon dioxide between alveolar gas and blood takes place. After diffusing into the blood, the oxygen molecules reversibly bind to the hemoglobin. Each molecule of hemoglobin contains 4 sites for combination with molecular oxygen; 1 g of hemoglobin combines with a maximum of 1.36 mL of oxygen.
The quantity of oxygen combined with hemoglobin depends on the level of blood PaO2. This relationship, expressed as the oxygen hemoglobin dissociation curve, is not linear but has a sigmoid-shaped curve with a steep slope between a PaO2 of 10 and 50 mm Hg and a flat portion above a PaO2 of 70 mm Hg.
The carbon dioxide is transported in 3 main forms: (1) in simple solution, (2) as bicarbonate, and (3) combined with protein of hemoglobin as a carbamino compound.
During ideal gas exchange, blood flow and ventilation would perfectly match each other, resulting in no alveolar-arterial oxygen tension (PO2) gradient. However, even in normal lungs, not all alveoli are ventilated and perfused perfectly. For a given perfusion, some alveoli are underventilated, while others are overventilated. Similarly, for known alveolar ventilation, some units are underperfused, while others are overperfused.
The optimally ventilated alveoli that are not perfused well have a large ventilation-to-perfusion ratio (V/Q) and are called high-V/Q units (which act like dead space). Alveoli that are optimally perfused but not adequately ventilated are called low-V/Q units (which act like a shunt).
At steady state, the rate of carbon dioxide production by the tissues is constant and equals the rate of carbon dioxide elimination by the lung. This relation is expressed by the following equation:
VA = K × VCO2/ PaCO2
where K is a constant (0.863), VA is alveolar ventilation, and VCO2 is carbon dioxide ventilation. This relation determines whether the alveolar ventilation is adequate for metabolic needs of the body.
The efficiency of lungs at carrying out of respiration can be further evaluated by measuring the alveolar-arterial PO2 gradient. This difference is calculated by the following equation:
PAO2 = FiO2 × (PB – PH2 O) – PACO2/R
where PA O2 is alveolar PO2, FiO2 is fractional concentration of oxygen in inspired gas, PB is barometric pressure, PH2O is water vapor pressure at 37°C, PACO2 is alveolar PCO2 (assumed to be equal to PaCO2), and R is respiratory exchange ratio. R depends on oxygen consumption and carbon dioxide production. At rest, the ratio of VCO2 to oxygen ventilation (VO2) is approximately 0.8.
Even normal lungs have some degree of V/Q mismatching and a small quantity of right-to-left shunt, with PAO2 slightly higher than PaO2. However, an increase in the alveolar-arterial PO2 gradient above 15-20 mm Hg indicates pulmonary disease as the cause of hypoxemia.
Hypoxemic respiratory failure
The pathophysiologic mechanisms that account for the hypoxemia observed in a wide variety of diseases are V/Q mismatch and shunt. These 2 mechanisms lead to widening of the alveolar-arterial PO2 gradient, which normally is less than 15 mm Hg. They can be differentiated by assessing the response to oxygen supplementation or calculating the shunt fraction after inhalation of 100% oxygen. In most patients with hypoxemic respiratory failure, these 2 mechanisms coexist.
V/Q mismatch is the most common cause of hypoxemia. Alveolar units may vary from low-V/Q to high-V/Q in the presence of a disease process. The low-V/Q units contribute to hypoxemia and hypercapnia, whereas the high-V/Q units waste ventilation but do not affect gas exchange unless the abnormality is quite severe.
The low V/Q ratio may occur either from a decrease in ventilation secondary to airway or interstitial lung disease or from overperfusion in the presence of normal ventilation. The overperfusion may occur in case of pulmonary embolism, where the blood is diverted to normally ventilated units from regions of lungs that have blood flow obstruction secondary to embolism.
Administration of 100% oxygen eliminates all of the low-V/Q units, thus leading to correction of hypoxemia. Hypoxemia increases minute ventilation by chemoreceptor stimulation, but the PaCO2 generally is not affected.
Shunt is defined as the persistence of hypoxemia despite 100% oxygen inhalation. The deoxygenated blood (mixed venous blood) bypasses the ventilated alveoli and mixes with oxygenated blood that has flowed through the ventilated alveoli, consequently leading to a reduction in arterial blood content. The shunt is calculated by the following equation:
QS/QT = (CCO2 – CaO2)/CCO2 – CvO2)
where QS/QT is the shunt fraction, CCO2 is capillary oxygen content (calculated from ideal PAO2), CaO2 is arterial oxygen content (derived from PaO2 by using the oxygen dissociation curve), and CvO2 is mixed venous oxygen content (assumed or measured by drawing mixed venous blood from a pulmonary arterial catheter).
Anatomic shunt exists in normal lungs because of the bronchial and thebesian circulations, which account for 2-3% of shunt. A normal right-to-left shunt may occur from atrial septal defect, ventricular septal defect, patent ductus arteriosus, or arteriovenous malformation in the lung.
Shunt as a cause of hypoxemia is observed primarily in pneumonia, atelectasis, and severe pulmonary edema of either cardiac or noncardiac origin. Hypercapnia generally does not develop unless the shunt is excessive (> 60%). Compared with V/Q mismatch, hypoxemia produced by shunt is difficult to correct by means of oxygen administration.
Hypercapnic respiratory failure
At a constant rate of carbon dioxide production, PaCO2 is determined by the level of alveolar ventilation according to the following equation (a restatement of the equation given above for alveolar ventilation):
PaCO2 = VCO2 × K/VA
where K is a constant (0.863). The relation between PaCO2 and alveolar ventilation is hyperbolic. As ventilation decreases below 4-6 L/min, PaCO2 rises precipitously. A decrease in alveolar ventilation can result from a reduction in overall (minute) ventilation or an increase in the proportion of dead space ventilation. A reduction in minute ventilation is observed primarily in the setting of neuromuscular disorders and CNS depression. In pure hypercapnic respiratory failure, the hypoxemia is easily corrected with oxygen therapy.
Hypoventilation is an uncommon cause of respiratory failure and usually occurs from depression of the CNS from drugs or neuromuscular diseases affecting respiratory muscles. Hypoventilation is characterized by hypercapnia and hypoxemia. Hypoventilation can be differentiated from other causes of hypoxemia by the presence of a normal alveolar-arterial PO2 gradient.
Mechanical ventilation in chronic hypercapnic respiratory failure
While the occurrence of acute hypercapnic respiratory failure in the course of an acute exacerbation in COPD is clearly associated with increased mortality, the prognostic value of chronic but stable hypercapnia is much more intricate (Budweiser et al 2007c). In patients with COPD and chronic respiratory failure receiving long-term oxygen therapy (LTOT), a low level of carbon dioxide tension (PaCO2) has even been found to be linked to elevated mortality (Chailleux et al 1996), whereas in another study similar rates of survival were observed in normocapnic (<45 mmHg) and hypercapnic (> 45 mmHg) patients (Aida et al 1998). When PaCO2 was markedly elevated (>55 mmHg), long-term outcome within the first year after hospital discharge was found to be poor (Chailleux et al 1996). The same applied when hypercapnia persisted after recovery from an acute exacerbation (Costello et al 1997).
Long-term non-invasive ventilation at home aims at persistently reducing hypercapnia as one of its goals. The impact of this treatment in COPD with chronic hypercapnic respiratory failure is still controversial, particularly as far as survival is taken as outcome measure. This is indicated by a number of randomized controlled trials (Budweiser et al 2008b) which are however not devoid of methodological weaknesses (Elliott 2002; Kohnlein et al 2003). Despite these uncertainties, COPD with concomitant chronic hyper-capnic respiratory failure has become one of the major indications for home mechanical ventilation (HMV), at least in Europe (Janssens et al 2003; Lloyd-Owen et al 2005). The large variability in prescription rates among different countries and institutions probably reflects the fact that the criteria for initiating home ventilation in these patients differ, similarly to the conditions and traditions of health care systems. Throughout countries, home ventilation is predominantly administered through a nasal or facial mask during the night; only a minority of patients is ventilated invasively via tracheostoma (Farre et al 2005). Recent data indicate that invasive ventilation is associated with poorer long-term outcome particularly in patients with COPD (Marchese et al 2008).
Indications for non-invasive home ventilation and selection of patients
Following the currently accepted guidelines, a pronounced elevation of daytime PaCO2 is considered as a key indicator for the initiation of home mechanical ventilation in COPD (Rabe et al 2007). Noteworthy enough there seems to be no agreement on the required levels of hypercapnia, and these have never been verified in comparative, prospective, randomized controlled trials. As a mark for orientation, the indication for non-invasive home ventilation in COPD is often based on the statements of a Consensus Conference Report a number of years ago (Consensus Conference Report 1999). According to these, home mechanical ventilation can be justified in severe COPD after optimization of standard therapy including LTOT, if daytime PaCO2 levels during spontaneous breathing are at least 55 mmHg. The same applies if PaCO2 levels are lower (50–54 mmHg) but at the same time significant nocturnal hypoventilation is present (Table 3).
Indicators for the initiation of domiciliary non-invasive ventilation in COPD
After initiation of HMV in patients with obstructive lung diseases benefits have been reported at least with regard to the frequency of hospitalizations compared to the preceding years (Jones et al 1998; Leger et al 1994). Thus, in more mild hypercapnia (PaCO2 50–54 mmHg) a further indication for home ventilation in COPD might be that at least two episodes of acute hypercapnic respiratory failure per year occur (Consensus Conference Report 1999). Such considerations suggest a role for HMV beyond the reversal of chronic hypercapnia as an immediate target.
According to the current view, COPD is considered as a systemic, multidimensional disorder bearing a variety of risk factors for severe exacerbations requiring hospitalization (Garcia-Aymerich et al 2001; Kessler et al 1999) and for death from different causes (Dolan et al 2005). These risk factors favor disease instability, uncompensated respiratory failure, and as a consequence hospitalization and death. A recent comprehensive meta-analysis came to the conclusion that in severe COPD HMV is capable of evoking beneficial effects on the ventilatory pattern, thereby reducing the work of breathing and improving functional reserves (Kolodziej et al 2007). Correspondingly, the use of NPPV at home might be one important factor in counteracting episodes of impending respiratory failure, which in these multi-morbid patients easily result from exacerbations. Thus there seems to be enough reason for arguing that the eligibility for HMV should not be based on a too restrictive and simple pattern of criteria. Preferentially it should rely on a rational analysis of the individual patient’s risk factor profile (Table 3). Such a view is also suggested by the results of a recent large observational study (Budweiser et al 2007b).
Moreover, a non-controlled investigation suggested that the continuation of HMV after difficult or prolonged weaning was associated with better long-term survival, compared to patients discharged without ventilatory support. Interestingly enough, this was independent from the patients’ age and duration of hospital stay (Quinnell et al 2006). Thus careful administration of HMV in patients with COPD is not only likely to exert beneficial effects on the patients but also might be beneficial for the society, as it can result in a significantly lower frequency of hospital admissions and a concomitant reduction in health-care costs, as noted previously (Casanova et al 2000; Jones et al 1998; Tuggey et al 2003). Whether this will lead to reduced or balanced overall costs, still has to be checked in prospective randomized controlled trials. Recent observational data suggest that long-term management of patients with severe COPD and chronic respiratory failure by measures including HMV does not increase healthcare costs compared to LTOT treatment alone (Clini et al 2008).
Effects of non-invasive home ventilation
It might seem surprising that the knowledge on the mechanisms underlying the effects of non-invasive HMV, especially their relative contributions under different conditions, is still rather limited. In particular the basis for the improvement of gas exchange during spontaneous breathing after NPPV support has not been clarified in detail (Mehta et al 2001). The hypothesis – plausible at the first sight – that non-invasive ventilation primarily allows the exhausted respiratory muscles to rest and recover probably not properly accounts for the complex mechanisms involved (Budweiser et al 2008b). This is true despite the fact that ventilatory support can be associated with a reduction in the activation of respiratory muscles under certain circumstances, as shown in earlier studies (Carrey et al 1990; Nava et al 1993; Renston et al 1994).
However, restoration of chemosensitivity (Mehta et al 2001) and changes in ventilatory pattern including an increase in tidal and minute volumes, as well as the concomitant reduction of respiratory frequency, may play a more prominent role in long-term NPPV. The relevance of these factors has been demonstrated in sophisticated physiological studies (Ambrosino et al 1992; Diaz et al 2002; Diaz et al 2005b). Even more important seems to be that the amelioration of the ventilatory pattern is preserved during spontaneous breathing (Ambrosino et al 1992; Diaz et al 2002; Diaz et al 2005b). This could well underlie the sustained reduction of PaCO2, which again induces positive effects on lung hyper-inflation and respiratory muscles (Budweiser et al 2005; Diaz et al 2002).
The beneficial effects of NPPV on breathing pattern, respiratory muscles and respiratory mechanics and the subsequent improvement in functional reserves (Kolodziej et al 2007) are reflected in a reduction of dyspnea. Considering dyspnea as an integrative measure of efficacy, this seems remarkable. It is a very consistent effect of long-term NPPV demonstrated by randomized controlled clinical trials (Casanova et al 2000; Clini et al 2002; Diaz et al 2005b). These findings also fit well into the picture provided by the data on health-related quality of life (HRQL).
The majority of controlled (Clini et al 2002; Meecham Jones et al 1995) and non-controlled trials (Perrin et al 1997; Sivasothy et al 1998; Windisch 2008) dealing with HRQL revealed positive effects of the long-term use of NPPV, irrespective of the fact whether generic measures were employed or disease-specific questionnaires designed for patients with chronic respiratory failure, such as the Maugeri Foundation Respiratory Failure Questionnaire (MRF-28) or the Severe Respiratory Insufficiency (SRI) questionnaire. The latter has recently been validated especially with regard to COPD and HMV (Windisch et al 2008). This again seems an important result, since HRQL is an important marker in chronic hypercapnic respiratory failure and even might carry predictive information on long-term survival, though this seems to depend on the type of disease (Budweiser et al 2007a).
Effects on exercise capacity
The patients’ global functional capacity appears to be well reflected in the 6-minute walking distance (6-MWD) as a clinically useful and easy-to-assess measure (Carter et al 2003). It also represents a prognostic marker in COPD (Pinto-Plata et al 2004), particularly in patients with chronic hypercapic respiratory failure (Budweiser et al 2008a). In the majority of studies there were no direct effects of long-term NPPV on 6-MWD (Clini et al 2002; Meecham Jones et al 1995; Schonhofer et al 2007; Strumpf et al 1991), and 6-MWD was found to be improved in only two randomized controlled trials which, however, covered only a short follow-up time (Diaz et al 2005b; Renston et al 1994).
Obviously, while having positive effects at rest, intermittent nocturnal HMV is not sufficient to compensate for the worsening of dynamic hyperinflation during exercise, which is considered as a major limiting factor of exercise capacity in severe COPD (O’Donnell et al 2001). This, however, does not exclude that NPPV applied during walking exerts positive effects on oxygenation, dyspnea and walking distance. Indeed such effects have been recently demonstrated in a randomized cross-over trial in a small sample of patients (Dreher et al 2007).
Effects on long-term survival
With regard to long-term prognosis the role of non-invasive ventilation in COPD and chronic hypercapnic failure is still under discussion. Two randomized controlled trials involving follow-up periods of 1 and 2 years, respectively, did not indicate domiciliary intermittent non-invasive ventilation to be effective in improving long-term survival compared to LTOT alone (Casanova et al 2000; Clini et al 2002). However, these trials comprised relatively small samples of patients and raised a number of objections regarding patients’ selection, ineffective ventilation or pressure levels, and poor adherence to therapy.
Thus non-randomized trials including a broad range of patients still seem to have a potential if analyzed and interpreted with caution. In a recent large observational study comprising an observation period of up to 4 years, we found that patients with COPD and non-invasive HMV showed a much better long-term survival than a control group of patients who could not be successfully adapted to HMV due to a variety of reasons (Budweiser et al 2007b). Of course, in view of the non-randomized design, these data have weaknesses, though the difference between treatment and control group at baseline was rather small. Still the results need to be checked by large randomized controlled trials, one such trial being currently underway (Kohnlein et al 2004).
Technical aspects of non-invasive home ventilation
One of the most important factors for a success of NPPV, especially a marked reduction in PaCO2, appears to be the achievement of effective ventilation by applying sufficient levels of inspiratory pressure. In accordance with this, studies capable of demonstrating beneficial effects of NPPV on gas exchange, lung function, sleep, or health-related quality of life (Budweiser et al 2005; Budweiser et al 2007c; Meecham Jones et al 1995; Windisch et al 2005) all used high inspiratory pressures, up to 28 cm H2O, which were still well tolerated by the patients. In contrast, the application of lower inspiratory pressure levels (10–15 cm H2O), as chosen in the two available randomized controlled trials (Casanova et al 2000; Clini et al 2002), was associated with an unfavorable outcome. Though these parallels are not compelling, they are at least suggestive.
During support-ventilation modes, in addition to high pressure levels, a positive end-expiratory pressure (3–5 cm H2O) and low trigger levels have been found to effectively counteract PEEPi (Nava et al 1993), minimizing the work of breathing and optimizing patient-ventilator synchrony. To allow for maximal expiratory emptying of the lung, inspiratory time should be chosen as short as possible when using controlled ventilation modes, with a target ratio of inspiration to expiration time (I:E) of at least 1:2 to 1:3.
Taking well-chosen ventilator settings and the usage of modern, sophisticated ventilators for granted, the potential of non-invasive ventilation can be fully utilized only if patients are motivated, well instructed and encouraged. This is of special importance in the initial phase of adaptation to the therapy. For achieving this, the availability of an experienced team as well as sufficient time for familiarization within an in-patient setting appears of utmost value. Today the high rates of inacceptance as reported in the early years of HMV (Lin 1996; Strumpf et al 1991) would not be considered acceptable and justified by objective reasons. Of course, contraindications have still to be taken into account (Table 2).
A further important issue is the regular follow-up of the patients. This serves for assessing the effectiveness of ventilation and adherence to therapy as well as for resolving potential adverse effects (Hill 2004). In case that an early withdrawal of NPPV turns out to be necessary, it should be noted that a benefit from this therapy has been reported even for such short-term use (Diaz et al 2002; Diaz et al 2005b). Obviously NPPV can be efficient even if bolstering functional reserves only within a short episode of most severe, acute deterioration (Budweiser et al 2008b).
Physiological consequences of hypoxemia
Once acute hypoxemia reaches a critical limit, various physiological consequences take place. These include an increase in minute ventilation, compensatory tachycardia to increase cardiac output, dilation of peripheral vessels, and constriction of pulmonary vasculature, with the result of a reduction of ventilation/perfusion mismatching. In the long run, these events plus the stimulation of erythropoietin production causing polycythemia lead to functional and structural changes including cor pulmonale and pulmonary hypertension (Kim et al 2008). Beyond its effects on the cardiopulmonary system, hypoxemia causes a number of other consequences, which include the reduction of specific neurocognitive functions as well as non-specific symptoms such as headache, thoracic oppression and agitations.
Oxygen for acute hypoxemic respiratory failure
Based on the physiological responses to hypoxemia, the administration of oxygen during acute respiratory failure as indicated by acute hypoxia and dyspnea can be considered as a matter of course. Accordingly, in current guidelines (Rabe et al 2007) for the management of acute exacerbations in COPD it is recommended that “the first actions when a patient reaches the emergency department are to provide supplemental oxygen therapy and to determine whether the exacerbation is life threatening”. Although this intervention is well established in clinical practice and oxygen supply is considered as biologically justified, the evidence on oxygen supply in acute respiratory failure from randomized controlled trials is surprisingly limited (Schumaker et al 2004).
It must however be admitted that appropriate clinical trials are difficult to perform in such clinically unstable conditions (Plant et al 2003) and moreover encounter ethical concerns. Short-term clinical trials in patients with COPD and stable hypoxemia and current pathophysiological concepts of dyspnea (O’Donnell et al 2007) suggest that oxygen is effective both in improving oxygen saturation and in reducing breathlessness in the acute setting. The alleviation of symptoms is presumably related to the decreases in minute ventilation and/or respiratory rate, which reduce the work of breathing and the risk for respiratory decompensation (Aubier et al 1980a).
To achieve these effects and sufficient levels of oxygenation, (PaO2 > 60 mmHg or oxygen saturation (SaO2) > 90%) (Rabe et al 2007), nasal prongs or high flow devices (Venturi masks) may be used, the latter being more reliable for delivering a defined oxygen amount in the acute setting (Bazuaye et al 1992). It is currently unknown whether the delivery of oxygen in the presence of PaO2 values >60 mmHg is beneficial in acute exacerbations. In any case, short-term and repeated evaluations of gas exchange indices are commonly recommended to verify treatment effects and to exclude a significant CO2 retention or an impending respiratory acidosis that may result from the reduction in hypoxic drive to breathe, or ventilation/perfusion mismatching (Aubier et al 1980b; Robinson et al 2000). As yet, clinical investigations dealing with these safety issues in, however, small samples of patients have suggested that the risk of worsening an existing hypercapnia is low during controlled oxygen delivery in acute exacerbations (Agusti et al 1999; Gomersall et al 2002). It may predominantly occur in patients with very severe initial hypercapnia (Moloney et al 2001).
The idea has been proposed that inhalation of nitric oxide in addition to oxygen might have benefits compared to oxygen alone in patients with COPD. There is, however, no substantial evidence on clinically relevant add-on effects, as indicated by studies that found unchanged, or even impaired, oxygenation in patients of different severity (Kanniess et al 2001; Melsom et al 2007). As a consequence, this combined therapy is not recommended in COPD (Germann et al 2005). More promising seems the addition of a helium-oxygen mixture based on the fluid-mechanical and physiological properties of helium. This has the potential to provide benefits regarding a reduction of airway resistance and work of breathing, which might result in lower intubation rates compared to standard inhalation gas (Schumaker et al 2004). However, these findings have to be substantiated in future randomized controlled trials.
Long-term oxygen therapy for chronic hypoxemic respiratory failure
LTOT is widely accepted as treatment to counterbalance chronic, significant hypoxemia in COPD although it is known to represent a considerable amount of still increasing health-care costs (Croxton et al 2006). Despite this, it has been argued that the critical PaO2 levels allowing the prescribing of LTOT according to accepted guidelines (Rabe et al 2007) (Table 4) are probably not fully adequate. They do not properly take into account other important factors of the disease such as a low BMI, comorbidities, or frequent exacerbations, all of which may favor a benefit from LTOT (O’Reilly et al 2007).
Major indications for long-term oxygen therapy (LTOT) summarized from GOLD guidelines (Rabe et al 2007) and national standards (BTS 1997; Magnussen et al 2001; NICE 2004)
Effects of LTOT on survival
Among the treatment modalities for COPD, LTOT is considered since many years as one of those interventions for which a clear-cut survival benefit is given. Indeed, more than 25 years ago two randomized controlled trials were capable of demonstrating a survival benefit from LTOT in patients with severe COPD and persistent hypoxemia at rest (MRC 1981; NOTT 1980). The patients included in these long-term trials either showed an arterial PaO2 level ≤55 mmHg or of ≤59 mmHg in case of additional peripheral edema, polycythemia or signs for cor pulmonale (NOTT 1980), or they showed PaO2 levels ranging from 40 to 60 mmHg and coexisting hypercapnia or congestive heart failure (MRC 1981). These two clinical trials represent major advances in respiratory medicine and are most often used as scientific basis for the prescription of LTOT.
Despite this one should not ignore their limitations, among them the relatively small numbers of patients (n = 87 and n = 203, respectively) compared to present-day standards and changes in overall treatment regimes in the meantime (Kim et al 2008). Nevertheless, there was an impressive benefit regarding long-term survival in patients receiving LTOT, which was closely related to the amount of daily use. The effect on long-term prognosis was particularly pronounced in patients using oxygen on average ≥15 or ≥16 hours/day, respectively. Viewed from the other side, such treatments effects have highlighted the major role of chronic hypoxemia as a predictor of long-term survival in COPD (Dolan et al 2005), and long-term follow-up investigations have essentially confirmed this relationship (Aida et al 1998; Chailleux et al 1996; Dubois et al 1994).
In contrast to the more severely affected patients, those with more mild daytime resting hypoxemia, ie, PaO2 levels > 55 mmHg (Gorecka et al 1997) or >60 mmHg (Fletcher et al 1992), did not show significant responses to LTOT in terms of survival within randomized controlled trials. The same was found for patients without marked daytime hypoxemia but nocturnal desaturations (Chaouat et al 1999; Fletcher et al 1992). A potential limitation of these studies, however, is that the daily average time of oxygen use may have been too short to induce recognizable survival benefits in less ill patients.
Effects of oxygen on other outcomes
Supplemental oxygen has been shown to reduce the sensation of dyspnea and to improve exercise capacity in patients with resting daytime hypoxemia (Lane et al 1987; Swinburn et al 1991). Such beneficial short-term effects during exercise can also been achieved in milder forms of hypoxemia, which do not reach the critical limit for the definition of hypoxic respiratory failure (PaO2 > 60 mmHg) at rest (Dean et al 1992; Fujimoto et al 2002). Presumably the underlying mechanisms involve a decrease in minute ventilation and concomitantly dynamic hyperinflation. These are accompanied by improvements in pulmonary hemodynamics and oxygen delivery which improve respiratory muscle function (Kim et al 2008).
Additionally there is evidence that in patients with chronic hypoxemia LTOT can induce subjective benefits regarding the presence of neurophysiological symptoms, anxiety, depression and health-related quality of life (HRQL) (Eaton et al 2004; Heaton et al 1983). In patients showing only transient hypoxemia during exercise, the randomized controlled trials have provided only mixed evidence regarding effects of LTOT on HRQL (Eaton et al 2002; McDonald et al 1995; Nonoyama et al 2007). Similarly, the rationale for administering supplemental oxygen during sleep in patients showing only nocturnal desaturations has not yet been clearly supported by data. Accordingly, a recent comprehensive prospective study indicated that nocturnal desaturations are not linked to impairments of HRQL, sleep quality and daytime vigilance (Lewis et al 2008).
In addition to the favorable results of LTOT on survival, beneficial effects on pulmonary hemodynamics have been described (NOTT 1980), which was essentially confirmed by later investigations (Zielinski et al 1998). This association probably represents an indirect, not a direct, causal relationship. Irrespective of this it may significantly contribute to the positive effects of LTOT on mortality and hospitalization rate (Ringbaek et al 2002), as pulmonary hypertension and cor pulmonale are known to be major factors in the long-term outcome of patients with COPD (Kessler et al 1999; Oswald-Mammosser et al 1995). Taken together, these results indicate that in patients without daytime hypoxemia but PaO2 < 55 mmHg during sleep or exercise, LTOT has to be considered. This particularly applies in the case that the optimization of other treatments such as CPAP that might also be sometimes indicated does not suffice to reverse hypoxemia (Table 4).
In summary, the following statements can be made on the basis of the data described.
In severe acute exacerbations of COPD with acute respiratory failure, controlled oxygen delivery is a reasonable and effective approach to relieve symptoms, counteract hypoxemia and reduce the work of breathing.
In COPD with acute hypercapnic respiratory failure, non-invasive ventilation is highly recommended, particularly in patients with mild to moderate respiratory acidosis. Non-invasive ventilation has the potential to reduce the risk for invasive ventilation and the associated complications, as well as to improve overall ICU and in-hospital outcome.
In COPD with clinically relevant chronic hypoxemia, LTOT is strongly indicated to improve hemodynamic parameters, long-term prognosis and HRQL.
In COPD with chronic hypercapnic respiratory failure, the role of long-term non-invasive home ventilation for survival has not yet been unambiguously demonstrated. However, a number of investigations point towards significant benefits in subjective and physiological outcomes.
Taken together, respiratory failure in COPD must be considered as a serious or even life-threatening complication. Controlled oxygen supply and non-invasive mechanical ventilation are two effective components of an evidence-based, comprehensive management of respiratory failure. Their rational use has the potential to significantly ameliorate the patients’ symptoms and to improve survival.