INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The therapeutic benefits of inhaled nitric oxide (iNO) have been explored since Ignarro and Palmer first identified NO as the elusive endothelium-derived relaxing factor in 1987 (1, 2). When administered by inhalation, NO selectively dilates the pulmonary circulation (3). Since iNO is rapidly inactivated by binding to hemoglobin in the pulmonary circulation, it exhibits a relatively short half-life and minimal direct effects on left ventricular function or systemic vascular resistance (Rsv) (6). Although iNO may have effects that persist far beyond the duration of administration of NO, it appears that the pulmonary vasodilating effects of such therapy have a very short half-life, consistent with rapid binding of NO by hemoglobin in the pulmonary circulation. In addition, the vasodilatory effects of iNO on the pulmonary circulation occur relative to the degree of regional ventilation, leading to an improvement in ventilation- perfusion relationships and thus to improvement in arterial oxygenation in patients with significant intrapulmonary shunt (7).

Since iNO has a selective pulmonary vasodilatory effect and a salutary effect on gas exchange in acute hypoxemic respiratory failure (AHRF), it has been studied for therapeutic use in a variety of cardiopulmonary conditions, including adult respiratory distress syndrome (ARDS), primary pulmonary hypertension, pulmonary hypertension of the newborn, and secondary pulmonary hypertension (3, 7). Although the results of these nonrandomized pilot trials have been very encouraging, more recent randomized, controlled trials have supported the use of iNO in newborns (16, 17), but have questioned the benefit of iNO in the treatment of ARDS (18). Nonetheless, iNO continues to be used and studied while its potential role in treating diverse forms of cardiopulmonary failure is further delineated.

One reason for continued interest in iNO for the treatment of right heart and lung failure is the low reported incidence of its side effects, even when inhalation is continuous over the course of weeks (8, 18, 23, 24). A number of possible toxicities of iNO have, however, been identified, including generation of methemoglobin, formation of harmful oxides of nitrogen such as nitrogen dioxide (NO₂) (25), and creation of a bleeding diathesis resulting from the platelet-inhibiting effects of NO (28, 29). Even in the aggregate, these complications are not commonly encountered, and many clinical trials have reported that discontinuation of iNO was never required because of undesirable side effects (16, 17). However, we and others have reported an additional, potentially life-threatening consequence of iNO (30, 31). In these anecdotal reports, abrupt discontinuation of iNO was associated with catastrophic deterioration in gas exchange and/or hemodynamics (30, 31). Others have proposed various schedules of weaning from iNO therapy to avoid such complications (32). Although most clinicians utilizing iNO are familiar with this "rebound" phenomenon, its precise incidence and pathogenesis have not been well described. Accordingly, we conducted a trial of iNO discontinuation in patients receiving this therapy, with the goals of: (1) identifying the need for ongoing administration of iNO; (2) identifying the incidence of adverse events upon discontinuation of iNO; (3) characterizing the hemodynamic and gas-exchange parameters that would suggest the cause(s) of deterioration when it is observed; and (4) determining predictors for this response.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

After obtaining institutional review board approval and informed consent by proxy, we treated 56 intubated, mechanically ventilated patients with acute hypoxemic respiratory failure (AHRF) with iNO with the intention of optimizing gas exchange. AHRF was defined as: (1) requirement for mechanical ventilation; (2) one or more quadrant air space filling, observed radiologically; and (3) arterial oxygen saturation (Sa_O2) < 90% despite a fraction of inspired oxygen (FI_O2) > 0.60 and a positive end-expiratory pressure (PEEP) > 5 cm H₂O. Of the 56 patients treated with iNO, 31 demonstrated an initial improvement in gas exchange, defined as an increase in Pa_O2  20 mm Hg. FI_O2 and PEEP were then adjusted in accord with the response to iNO in order to maintain acceptable oxygenation with a nontoxic FI_O2 and the least PEEP necessary to maintain an Sa_O2 > 90%, with reductions of both parameters in most patients. These patients were then continued on iNO as part of the management for AHRF. As part of their treatment protocol, iNO was discontinued after at least 10 h of continuous administration, to determine the ongoing requirement for iNO or an adverse effect of its discontinuation.

Demographic and clinical information, including age, gender, admitting diagnosis, cause of respiratory failure, multiple system organ failure (MSOF) score, acid-base status, echocardiographic assessment, and vasoactive medications, was recorded. Outcomes data, including mortality and length of intensive care unit (ICU) and hospital stay, were also collected. Severity of illness was estimated through both the Acute Physiology and Chronic Health Evaluation II (APACHE II) score and Murray-Matthay lung injury score (LIS) (33).

All patients were ventilated with a Puritan-Bennett 7200 (Puritan-Bennett Corporation, Carlsbad, CA), Servo 900 C (Siemens Elema, Lund, Sweden), or Servo 300 (Siemens Elema) ventilator. Ventilation of all patients was done via an assist-control or pressure-control mode. Tidal volume (VT), inspiratory set pressure, respiratory rate (RR), inspiratory-to-expiratory time ratio, PEEP, and FI_O2 were determined by the primary clinical service for each patient. Indwelling arterial catheters were placed in all patients. Pulmonary artery catheters were present in all patients, with one exception.

Decisions regarding supportive care, including crystalloid and colloid infusions, blood products, and vasoactive medications, were made by the primary clinical service for each patient. Sedative, analgesic, and paralytic agents were used as needed. No changes in infusions were permitted during the 30 min preceding discontinuation of iNO or during the discontinuation trial.

Hemodynamic Measurements

Heart rate (HR, beats/min) was determined by continuous electrocardiography (Spacelabs Monitors, Redmond, WA). Mean systemic arterial pressure (MAP, mm Hg), mean pulmonary artery pressure (, mm Hg), right atrial pressure (Pra, mm Hg), and pulmonary capillary wedge pressure (PCWP, mm Hg) were measured at end-exhalation, using disposable quartz transducers (Abbott Laboratories, Abbott Park, IL). Transducers were set to zero at the midaxillary line with the patient in the recumbent position. Cardiac output (, L/min), determined as the average of four consecutive readings, was obtained by the thermodilution method. Systemic vascular resistance (RSV, dyne · s/cm⁵) and pulmonary vascular resistance (RPV, dyne · s/cm⁵) were calculated according to the formulas RSV = 80 · ([MAP  Pral)/T (in dyne · s/cm⁵) and RPV = 80 · (Ppa  PCWP)/T (in dyne · s/cm⁵).

When available, echocardiographic data obtained within 24 h of the discontinuation trial were recorded. If either right ventricular dilation or moderately severe impairment of right ventricular systolic function was noted in the face of normal or only mild impairment of left ventricular systolic function, the patient was designated as having echocardiographic right heart strain.

Gas-Exchange Measurements

Arterial and mixed-venous blood gas samples were collected, placed on ice, and analyzed with standard blood gas electrodes (ABL 500; Radiometer, Copenhagen, Denmark). Hemoglobin concentration (Hb, g/dl), methemoglobin percentage (MetHb), and arterial and venous hemoglobin saturations (Sa_O2, Sv_O2) were measured through cooximetry (ABL520; Radiometer). The relevant arterial (Ca_O2, ml/dl), mixed venous (Cv_O2, ml/dl), and pulmonary capillary (Cc_O2, ml/dl) oxygen contents were calculated with the formula: Cx_O2 = (1.39 · Hb · Sx_O2) + (0.003 · Px_O2) in ml/dl, where x represents the arterial, venous, or pulmonary capillary circulation, respectively.

The alveolar-arterial oxygen tension gradient ([A-a]PO₂, mm Hg) and shunt fraction ( S/  T) were calculated through the following equations: PA_O2 = (713 · FI_O2)  (Pa_CO2/0.8), (A-a)PO₂ = PA_O2  Pa_O2, and S/T = (Cc_O2  Ca_O2)/(Cc_O2  Cv_O2), respectively.

NO Delivery System

A 2-L mixing chamber was attached to the proximal portion of the inspiratory tubing of the ventilator. NO was delivered directly to the mixing chamber from stock cylinders of NO stored in nitrogen gas (N₂) at concentrations of 600 to 800 ppm. The mixture of ventilator gases and NO/N₂ was delivered directly to the patient. Concentrations of NO, NO₂, and O₂ were measured at the Y-piece of the ventilator tubing, just proximal to the endotracheal tube. NO and NO₂ concentrations were monitored through the standard chemiluminescence technique (Sievers NOA 270B; Sievers Inc., Boulder, CO). The inspired concentration of NO was adjusted by varying the flow of NO into the inspiratory limb of the breathing circuit.

Discontinuation Protocol

All initial discontinuation trials occurred within 10 to 30 h after the start of iNO therapy. Baseline hemodynamic and gas-exchange variables were obtained and recorded, and iNO was interrupted at the regulator valve blending NO and oxygen. iNO therapy was withheld for 30 min or until the patient developed hemodynamic or gas-exchange abnormalities that were felt to require immediate reinstitution of iNO in the judgment of the bedside investigator. A second set of measurements of hemodynamic and gas-exchange variables was obtained prior to reinstituting iNO if possible; when sudden adverse hemodynamic changes occurred and precluded completion of cardiac output (CO) determination via the thermodilution method, arterial and mixed venous blood gases were measured and/or continuous arterial and mixed venous oximetry readings were recorded. When iNO was reinstituted, it was given at the dose at which discontinuation had occurred. Data collection was repeated 30 min after the resumption of therapy with iNO.

Statistical Analysis

All data are expressed as mean ± SD. Two-tailed t tests were used to compare continuous variables. Discrete variables were compared through chi-square analysis. A value of p < 0.05 was considered significant. After univariate analysis was done of factors associated with cardiovascular collapse, a logistic regression analysis was performed to determine independent factors associated with this phenomenon.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic and Baseline Clinical Data

The age of the patients in the study was 53 ± 18 yr (mean ± SD), with a 1.2:1.0 female predominance. The LIS was 3.0 ± 0.8 and the mean APACHE II score calculated upon entrance into the ICU was 24.3 ± 7.5. Average hospital length of stay was 21.5 ± 20.4 d and the average ICU length of stay was 16.1 ± 17.1 d, including survivors and nonsurvivors. The low survival rate (16.1%) was consistent with our use of iNO as a salvage therapy for the most severely ill patients with hypoxemic respiratory failure. Table 1 gives the diagnoses, outcome, MSOF score, acid-base status, and catecholamine use for all patients at baseline.

Twenty-four of the 31 patients who underwent the trial discontinuation of iNO met the American-European Consensus Conference definition of ARDS (34). Twenty-eight of these 31 patients had an echocardiogram performed within 24 h of their iNO discontinuation trial; 15 of 28 of these studies revealed right heart strain as defined previously. Echocardiography was not performed during the discontinuation trial.

Discontinuation Trials

The mean time from initiation of iNO therapy to the beginning of the discontinuation trial was 19.4 ± 4.4 h, with a range of 10 to 30 h. The mean concentration of iNO at the beginning of the discontinuation trial was 22.8 ± 16 ppm, with a range of 3 to 80 ppm. The gas-exchange and hemodynamic parameters of all patients immediately before discontinuation and immediately after reinstitution of iNO are shown in Table 2. Significant declines in Pa_O2, Sa_O2, Sv_O2, and the ratio of Pa_O2 to FI_O2 (Pa_O2/FI_O2) were noted. There were trends toward an increase in S/T (p = 0.09), arteriovenous difference in oxygen delivery ([a-v]DO₂) (p = 0.10), and PVR (p = 0.1), and toward a decrease in MAP (p = 0.08).

Patterns of Response Upon Discontinuation of iNO

Three general patterns of response, which provided for subgroup analysis, were observed in our patients upon discontinuation of iNO (Figure 1). A majority of patients (23 of 31 [74.2%]) exhibited significant deterioration of gas exchange and/or hemodynamics, defined as either a decrease of CO of  10% of the baseline value, or a decrease in Sa_O2 of  5% of the baseline value. These criteria were generated with the presumption that patients with changes of this magnitude might require other clinical interventions to support their circulation and gas exchange, such as increases in vasoactive medications, FI_O2, or PEEP. Eight patients (25.8%) met neither of these criteria, and exhibited a minimal response to discontinuation of iNO, despite having had an initial beneficial response (Figure 1).

View larger version (25K): [in this window] [in a new window]   Figure 1.   Patterns of response to discontinuation of iNO.

The majority of patients (15 of 31) who had adverse effects upon discontinuation of iNO exhibited worsened arterial oxygenation, with a decrease in Sa_O2 > 5% which occurred within 30 min of discontinuation of iNO and readily reversed upon reinstitution of iNO. However, eight of the 23 patients who had a distinct response to discontinuation of iNO exhibited a rapid decline in blood pressure and/or CO of > 20%, which was designated as hemodynamic collapse (Figure 1). In all eight patients who showed this response, the discontinuation trial was aborted and iNO was reinstituted, with blood pressure and CO promptly returning to baseline levels. Patients exhibiting this response invariably did so within 1 to 10 min of discontinuation of iNO.

Table 3 shows the hemodynamic and gas-exchange data at baseline and after iNO discontinuation for the eight patients who exhibited hemodynamic collapse. Because of the rapidity of hemodynamic change in these patients, signaled by a falling systemic blood pressure, determinations of CO could be accomplished in only three of the patients, and a mixed venous oxygen saturation in only five (Figure 2). All patients had continuous pulse oximetry, Ppa, and radial artery pressure recordings. The earliest indication of deterioration in circulatory function in the patients who exhibited hemodynamic collapse was their falling systemic blood pressure, which occurred in all patients in advance of the Sa_O2 falling below 85%, and was associated with a decrease in CO and Sv_O2 (Figure 2). Since mean tended to remain the same in the face of a falling CO, calculated pulmonary vascular resistance tended to rise, although the limited number of observations of this did not achieve statistical significance (Figure 2).

View larger version (16K): [in this window] [in a new window]   View larger version (19K): [in this window] [in a new window]   View larger version (16K): [in this window] [in a new window]   Figure 2.   Characterization of cardiovascular collapse in the subset of patients having hemodynamic measurements during circulatory deterioration. (A) CO determinations. (B) Mixed venous oxygen saturation. (C ) Pulmonary vascular resistance. The findings for all three parameters are consistent with a rise in RPV in association with discontinuation of iNO, followed by right ventricular failure with diminished CO (see text for discussion).

Predictors of Response to Discontinuation of iNO

A number of baseline conditions were examined to determine whether they could predict the patient group exhibiting hemodynamic collapse at the time of discontinuation of iNO. After univariate analysis, logistic regression analysis identified MSOF, older age, and an increased initial blood pressure on initiation of iNO as independently associated (p < 0.05) with cardiovascular deterioration upon discontinuation of iNO. A more significant improvement in the pressure-to-flow ratio after initiation of therapy with iNO was weakly associated (p < 0.10) with increased risk of hemodynamic collapse. Additionally, of four patients who received iNO immediately after having undergone cardiopulmonary bypass, two developed hemodynamic collapse upon discontinuation of iNO and two did not.

A comparison of baseline hemodynamic and gas-exchange parameters for patients with and without hemodynamic collapse is shown in Table 4. As a group, patients with hemodynamic collapse had a lower baseline CO and higher (a-v)DO₂. Baseline oxygenation was not worse in patients exhibiting hemodynamic collapse, and in fact they exhibited a higher Pa_O2/ FI_O2 and lower (A-a)PO₂ and shunt fraction.

Outcome of Discontinuation Trials

In eight of 31 patients, discontinuation of iNO caused no significant change in parameters of hemodynamics or gas exchange, and this therapy could be withdrawn. In the 23 patients with adverse effects upon discontinuation of iNO₂, all gas-exchange and hemodynamic parameters were restored promptly to baseline upon reinstitution of iNO at the NO concentration previously used, and no adverse effects of the discontinuation trial were observed. In the eight patients exhibiting hemodynamic collapse, cautions about any discontinuation of iNO therapy for purposes of transport, suctioning, or ventilator changes were communicated to patients' primary services, and no subsequent adverse events were noted. Each of these patients then underwent a gradual weaning from iNO as tolerated. One of these eight patients was weaned from iNO over the course of 2 d, and survived to discharge from the hospital. Seven of the eight patients died during the course of hospitalization, five of whom had been successfully weaned from iNO at the time of death but who succumbed to their underlying disease. Two patients had iNO withdrawn as part of comfort measures, and expired within 1 h after its withdrawal.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We evaluated the effects of discontinuation of iNO in a number of patients undergoing this treatment for acute hypoxemic respiratory failure, in order to determine the ongoing need for this therapy, the incidence of adverse effects upon its discontinuation, and the pathogenesis and predictors of such adverse effects.

We undertook our discontinuation trials in 31 of 56 patients who were initial responders to iNO given to improve oxygenation in AHRF; the initial response to iNO of our patients (31 of 56 [55%]) was comparable to that reported by other authors (18, 21). Interestingly, eight of the 31 of our patients in whom iNO was stopped showed little response to its discontinuation, suggesting a minimal effect of iNO at the time of study. This finding is consistent with other observations reported in the literature. Dellinger and colleagues reported the results of iNO therapy in patients with ARDS in a recent prospective, randomized, placebo-controlled, double-blind phase II trial (18). An increase in Pa_O2/FI_O2 was seen in 60% of their patients, a response rate similar to ours. However, these authors also reported a response rate of 24% in patients receiving placebo, suggesting that patients with ARDS may have incremental changes in Pa_O2/FI_O2 that are relatively sudden, and which in uncontrolled studies could be ascribed to experimental therapies. Since our patients were begun on iNO in a nonrandomized initial study design, it is possible that some patients were maintained on iNO not because of a true response to a vasodilator, but because of independent acute improvements in gas exchange ascribed to the intervention. To the extent that this may have been the case, a failure to respond to discontinuation of iNO would not be surprising. An alternative explanation is offered by the observations of Michael and colleagues (21). These authors randomized 40 patients with ARDS to receive either conventional therapy or conventional therapy as well as iNO. A significant improvement in oxygenation was noted at 1 h and 12 h after the initiation of treatment in patients receiving iNO, as judged by Pa_O2/ FI_O2. However, after 12 h, both the conventional-treatment and iNO-treatment groups showed a convergence of oxygenation parameters, with no difference noted at 72 h. Thus, it is possible that patients in our study who initially responded to iNO lost this response over time, as demonstrated at the time of discontinuation of iNO.

The group we found most interesting and perhaps most relevant to clinical practice were the patients who exhibited hemodynamic collapse. Although we and others had reported similar pediatric and adult patients anecdotally (30, 31), the true incidence of such collapse is not known, and these anecdotal reports did not clarify the pathogenesis of or risk factors for this complication of iNO therapy. We observed that approximately 25% of patients receiving iNO for 10 to 30 h in the management of AHRF exhibited abrupt hemodynamic collapse upon cessation of iNO. This is a somewhat more striking finding than the isolated worsening of gas exchange, with immediate implications for patient outcome if not prevented or recognized promptly. We acknowledge that the population of patients studied with our protocol were quite ill, as signaled by their high mortality, and represent a group in whom iNO was used as a salvage therapy and not early therapy for hypoxemia. The quite high MSOF scores in our patient population in general, and the even higher scores in patients exhibiting cardiovascular collapse, suggests that as a whole these patients may have been at higher risk of this complication than less ill patients treated with iNO.

One potential explanation for the marked hypotension and low output state seen in the subset of our patients with cardiovascular collapse was that the collapse arose from severe hypoxemia, and was therefore the result of the most extreme instances of deteriorating gas exchange. Most of our observations in these patients speak against this possibility. Systemic hypotension in our patients occurred before any decline in Sa_O2 in some patients, and well before severe desaturation in any patient. When arterial desaturation did occur, it tended to follow a decline in mixed venous oxygen saturation that was associated with deteriorating hemodynamics. Furthermore, as a group, the patients who exhibited hemodynamic collapse tended to have a higher Pa_O2/FI_O2 and lower (A-a)PO₂ gradient at baseline than did patients who did not exhibit collapse.

These findings are consistent with the onset of hemodynamic collapse resulting from an increase in right ventricular afterload, followed by failure of the right ventricle to sustain stroke volume and forward flow. This in turn results in systemic hypotension, mixed venous oxygen desaturation, and a rapid downward spiraling of gas-exchange and circulatory function that requires immediate interruption through restoration of iNO₂ therapy. Consistent with this formulation of relatively isolated right ventricular dysfunction was the observed increase in Pra seen in our patients exhibiting hemodynamic collapse (Table 3), of 14.9 to 16.9 mm Hg, with a PCWP that fell from 17.4 to 14.2 mm Hg. Although these changes were not statistically significant, the direction of change is consistent with a primarily right ventricular overload. This mechanism could be more conclusively demonstrated by continuous echocardiographic assessment of these patients, which would show a change in right ventricular pressure-volume relationships in response to an increase in right ventricular afterload associated with cessation of iNO. Interestingly, the continuously monitored Ppa does not signal the onset of these events, since the failing right ventricle fails to generate forward flow, and although Rpv is rising, the observed Ppa remains largely unchanged. Right ventricular ischemia might also contribute to this phenomenon. Even in the absence of coronary artery disease, acute increases in Rpv, resulting in dilation of the right ventricle, increased right ventricular wall tension, and diminished aortic perfusing pressure of the coronary circulation could result in subendocardial ischemia and further right ventricular dysfunction.

At least one potential molecular mechanism exists for such a scenario. In cultured vascular endothelial cells, NO exerts negative feedback by downregulating the activity of NO synthase (35, 36). In a rat model, Combes and colleagues demonstrated that NO inhalation could decrease the release of endogenous endothelial NO and thus alter pulmonary vasoreactivity (37). It is therefore conceivable that exogenous NO administration downregulates endogenous NO production, and that when exogenous sources of NO are abruptly terminated, mechanisms for modulating pulmonary vasomotor tone are impaired and vasoconstrictive influences predominate. We admit that this mechanism is purely speculative, since none of our data speak to basic mechanisms underlying these gross bedside observations. Alternative mechanisms for this phenomenon would include influences of iNO on cardiovascular reflexes, with downregulation of neuromediators during exogenous NO administration.

Increased age and initial blood pressure response to iNO were both associated with an increased risk of hemodynamic collapse upon discontinuation of iNO. Similarly, a more substantial improvement of the pressure-to-flow ratio, associated with the initiation of iNO therapy, was weakly associated with an increased risk of hemodynamic collapse during discontinuous trials. We have no insight as to why increased age might be associated with increased risk of hemodynamic collapse following the discontinuation of iNO. A positive initial blood pressure response may be an indicator of an initial salutary effect of iNO on right ventricular stroke volume and CO, and would logically be associated with an adverse cardiovascular response to cessation of this vasodilatory therapy. Analysis of standard echocardiographic imaging results of baseline gas-exchange and acid-base status, of the concentration of NO administered, and of vasoactive drug use failed to provide predictors for this complication.

In summary, we observed that approximately one-fourth of patients receiving iNO for treatment of AHRF exhibited no effect upon discontinuation of iNO₂, whereas another one-fourth of such patients exhibited nearly immediate hemodynamic instability. The remainder of patients had varying degrees of worsening of gas exchange. All abnormalities observed were promptly reversed with reinstitution of iNO. Daily discontinuation of iNO₂ appears useful when done with appropriate monitoring and staff available to assess the patient. This simple maneuver identifies both patients no longer benefiting from this experimental therapy and those at high risk of deterioration with inadvertent discontinuation of iNO₂. Identifying patients at risk for deterioration is useful to heighten awareness of their vulnerability when ventilator circuits are interrupted in the ICU or when the patient is moved for diagnostic studies. In patients exhibiting this dependency on iNO, its discontinuation must be avoided at all costs. These observations must also be taken into consideration in future trials of the risks and benefits of iNO in treating diverse forms of cardiopulmonary failure. Although some early studies have reported few or no complications with iNO₂ therapy, we view a dependency on iNO for maintenance of adequate circulatory function as potentially life-threatening, and therefore worthy of observation and reporting in future studies.