Nitrogen Dioxide Production during Mechanical Ventilation with Nitric Oxide in Adults : Effects of Ventilator Internal Volume, Air Versus Nitrogen Dilution, Minute Ventilation, and Inspired Oxygen Fraction

(Figure 1) shows the experimental apparatus. NO (800 ppm in Nitrogen², Airco, Riverton, NJ) was mixed with Nitrogen²using two blenders (Bird Products, Palm Springs, CA) in series. Stock NO cylinders were certified by the supplier to contain < 8 ppm NO², and thus the [NO²] delivered to the ventilators in this study was < 0.8 ppm. NO was connected to the Oxygen²inlet of the first blender, and Nitrogen²(Airco) was connected to the air inlets of the first and second blenders. The outlet of the first blender led to the Oxygen²inlet of the second blender. Two blenders allowed more precise gas mixing at lower [NO], consistent with our clinical practice. NO and Nitrogen²pressures were set at 50 lb/in². The outlet of the second blender was delivered to the high-pressure air inlet of a Puritan-Bennett 7200ae (Puritan-Bennett, Carlsbad, CA) or Siemens Servo 900C (Siemens-Elema, Sola, Sweden) ventilator. Including the ventilator blender, [NO] was thus reduced in three stages.

The ventilator was connected to a test lung (Dual Adult TTL (Training Test Lung) model 160, Michigan Instruments, Grand Rapids, MI) with a disposable ventilator circuit (Aerosol Hose 5022, Seamless, Ocala, FL). Each ventilator was set to deliver V with dot^Eof 5, 10, 15, 20, and 25 l/min in the controlled mechanical ventilation mode. FI^O2of the ventilators was set at 0.4, 0.5, 0.6, 0.7, 0.8, and 0.9, but actual FI^O2(7820 Oxygen²monitor, Puritan- Bennett) was 0.24, 0.37, 0.50, 0.62, 0.75, and 0.87, respectively, because a nitric oxide/nitrogen mixture was delivered to each ventilator instead of air.

The inspiratory gas was aspirated 20 cm from the Y-piece and analyzed for NO and NO²using a calibrated chemiluminescence analyzer (Chemiluminescent NO-NO²-NO^xanalyzer model 10, Thermo Environmental Instruments, Franklin, MA). The analyzer was calibrated with 0 and 80 ppm NO. Chemiluminescent analyzers have been shown to be linear over the range of concentrations that we evaluated. [15] The blenders were set to deliver NOx ([NO + NO²]) of 10, 20, 30, 40, 50, 60, 70, and 80 ppm. NO and NO²measurements were recorded after equilibration. A servo-controlled humidifier (Conchatherm III, Temecula, CA) was placed in the inspiratory limb and the temperature at the Y-piece of the circuit was kept at 30–32 degrees Celsius. Gas from the ventilator and the analyzer was scavenged by wall vacuum.

NO gas was diluted with air instead of Nitrogen², and the experiment described above was repeated, NO (800 ppm in Nitrogen²) was blended with air from a wall pressure source, and one blender was used to control [NO] leading to the air inlet of the ventilator. We did not use two blenders for this experiment because pilot studies showed that very high [NO²](> 5 ppm) resulted when NO was diluted with air in two blenders. The length of the air hose (conductive K/l) from the blender to the ventilator was 3.6 m. When [NO²] > 5 ppm was detected in the inspiratory limb, measurements at higher [NO²] were not performed at that FI^O2and V with dot^E. To allow comparisons with Nitrogen²dilution, each ventilator was set at an FI^O2of 0.24, 0.37, 0.50, 0.62, 0.75, and 0.87.

We examined the effect of pulmonary residence time on NO²production in a mechanical lung model. NO (800 ppm) was mixed with Nitrogen²using two blenders in series and connected to the high- pressure air inlet of a Puritan-Bennett 7200ae ventilator. The lung model was set at a compliance of 0.1 l/cmH²O. The ventilator settings were those listed in Table 1, except that positive end- expiratory pressure was applied to create end-expiratory volumes of 0.5, 1.0, 2.0, 3.0, and 4.0 l in the lung model. End-expiratory flow as 0 l/min for all settings. Gas was sampled from the inspiratory limb and from within the lung bellows, and [NO] and [NO²] were measured by chemiluminescence at all V with dot^Eand test lung volumes.

The rate constant was determined from the relation shown in Equation 2. Rate constants were determined for NO blended with Nitrogen², NO blended with air, and simulated intrapulmonary residence of NO.

Because virtually no NO²(less or equal to 1 ppm) was produced with the Puritan-Bennett 7200ae ventilator when NO was mixed with Nitrogen², the rate constant for NO mixed with Nitrogen²was determined using only the Servo ventilator data. To determine the rate constant for NO mixed with air, data was pooled for the Puritan- Bennett 7200ae and Servo 900C ventilators. The [Oxygen²] was converted to parts per million to determine the rate constants. Equation 2was solved for the rate constant by using the nonlinear regression module of SPSS (Chicago, IL).

To determine the rate constant for intrapulmonary residence of NO, the residence time was estimated by dividing the residence time was estimated by dividing the test lung volume by V with dot^E. To determine the rate constant for NO conversion to NO²in the ventilator systems, the residence time was calculated by dividing the volume of the system by the V with dot^E. The ventilatory circuit in our study was 213 cm in length and had a volume of 0.7 l. The Servo 900C has an internal volume of 1.0 l because of the reservoir inside the ventilator (0.9 l) and the additional internal circuity (0.1 l). Because the Puritan-Bennett 7200ae does not have an internal bellows, its internal volume is only 0.25 l.* Thus, the total volume of the Servo 900C ventilator system was 1.7 l, and the total volume of the Puritan- Bennett 7200ae system was 0.95 l.

No NO²was detected during any trial with a V with dot sub E of 20 or 25 l/min. With the Puritan-Bennett 7200ae, NO²(less or equal to 1 ppm) was detected only at a V with dot^Eof 5.0 l/min when FI^O2was 0.87 and [NO] was 70 or 80 ppm. In contrast, with the Servo 900C (Table 2), at a V with dot^Eof 5.0 l/min, 0.5–3.5 ppm NO²was present when FI^O2was greater or equal to 0.37 and [NO] greater or equal to 50 ppm; at a V with dot^Eof 10 l/min, 0.5–1.5 ppm NO²was present when FI^O2was greater or equal to 0.62 and [NO] greater or equal to 60 ppm; at a V with dot^Eof 15 l/min, 0.5 ppm NO²was present when FI^Osub 2 was 0.87 and [NO] 80 ppm.

(Figure 2and Figure 3) show NO²formation with the Servo 900C and the Puritan-Bennett 7200ae ventilators, respectively. Compared with Nitrogen², dilution with air resulted in substantially increased [NO²] at all settings with both ventilators. At the same settings, [NO²] was greater with the Servo than with the Puritan-Bennett 7200ae. At an FI^O2of 0.87 and [NO] of 60 ppm with the Puritan-Bennett 7200ae, or an FI^O2of 0.87 and [NO] of 40 ppm with the Servo 900C, NO²was > 5 ppm even at a V with dot^Eof 25 l/min. At all settings, the greater the FI sub O²and [NO] and the less the V with dot^E, the greater the [NO²].

At a V with dot^Eof 20 or 25 l/min, less or equal to 1 ppm NO²was measured in the test lung bellows (simulated intrapulmonary residence of NO). At a V with dot^Eof 15 l/min, [NO] greater or equal to 60 ppm, test lung volume greater or equal to 1 l, and FI^O2greater or equal to 0.75, 1–3 ppm NO²was detected in the bellows. Table 3shows NO²production at a V^Eof 5 or 10 l/min. With greater FI^O2, greater [NO], greater test lung volume, and lesser V with dot^E, more NO²was produced.

When NO was blended with NO²before entering the ventilator, the rate constant was (1.46 plus/minus 0.07) x 10 sup -9 ppm sup -2 *symbol* min sup -1 (r²= 0.999). When NO was blended with air, the rate constant increased eightfold to (1.17 plus/minus 0.04) x 10 sup -8 ppm sup -2 *symbol* min sup -1 (r²= 0.997). For the simulated intrapulmonary residence of NO, the rate constant was (1.44 plus/minus 0.02) x 10 sup -9 ppm sup -2 *symbol* min sup -1 (r²= 0.9997).

This study demontrates that the [NO²] values measured in the inspiratory limb of a ventilator circuit are greater when the Servo 900C rather than the Puritan-Bennett 7200ae ventilator is used and when NO is diluted with air rather than with Nitrogen²*symbol*[NO²] values also were greater with increased [NO], greater FI^O2, or lesser V with dot^E. There also is a potential for intrapulmonary NO²production in lung units with long residence times, as simulated in our lung model.

Rate constants for NO conversion to NO²have been published by others (Table 4). [8–11] The rate constants calculated in the current study for NO blended with Nitrogen²and for the simulated intrapulmonary residence of NO are similar to those published by Glasson and Tuesday. [8] With the exception of our data for NO blended with air, rate constants of 0.79 x 10 sup -9 - 2.26 x 10 sup -9 have been reported. These differences are likely caused by differences in experimental methods, including static versus dynamic conditions, temperature, and humidity differences. Our current study verifies these values in clinically relevant conditions with modern adult critical care ventilators. Figure 4graphs the residence times required to generate 2 ppm NO²using the rate constant of the current study and those of three other studies. [9–11] These differences may be clinically important and emphasize the need to measure [NO²] during NO administration. Regardless of the rate constant used, however, a residence time > 2.75 min is needed to generate more than 2 ppm NO²at [NO] less or equal to 20 ppm and FI^O2less or equal to 0.9.

When NO was blended with air, the rate constant was about eight times greater than with Nitrogen²blending. With this method, blending of NO and Oxygen²begins inside the high-pressure hose proximal to the ventilator and thus increases the residence time. Further, NO and Oxygen²react under high pressure (50 lb/in²). Clinically, however, blending NO with compressed air decreases the effect of NO delivery on FI^O2and is more convenient and economical during prolonged administration.

Low [NO](less or equal to 20 ppm) has been reported to improve oxygenation for patients with ARDS. [3,4] Our data indicate that these concentrations can be administered with the Puritan-Bennett 7200ae ventilator without significant NO²production, even when the NO is blended with air. With the Servo 900C ventilator, [NO] less or equal to 20 ppm can be administered without significant NO²production at an FI^O2less or equal to 0.75 and an V with dot sub E of at least 10 l/min when air blending is used. When NO was blended with Nitrogen², NO²was not produced in either mechanical ventilation system at an [NO] of less than 50 ppm.

We chose the Puritan-Bennett 7200ae and the Servo 900C as examples of adult mechanical ventilators with different internal volumes. Our results may apply to other ventilator systems with similar internal volumes. Some ventilators for adults other than the Servo 900C have large internal volumes, and gas is compressed in some of these reservoirs, [16] a condition that may further affect the rate constant for NO conversion to NO².

There is a potential for NO conversion to NO²in lungs with long residence times. From our experience, exhaled [NO] that is 50- 75% of inhaled concentrations is common when inhaled NO is administered to patients with ARDS. Our lung model data, however, indicate that a relatively long residence time is required to produce 2 ppm NO²at the [NO] typically used in the treatment of ARDS (i.e., less or equal to 20 ppm). At a rate constant of 1.44 x 10 sup -9 ppm sup -2 *symbol* min sup -1, the time required to produce 2 ppm NO²is 4.6 min during breathing of 90% Oxygen²and 6.8 min during breathing of 60% Oxygen sub 2. This prolongation of residence time is unlikely during ARDS but may occur in patients with severe chronic airflow obstruction. Because ours was a study in a lung model, further work is needed to determine whether conversion of NO to NO²does occur in the lungs during therapeutic NO administration.

Because NO²is an atmospheric pollutant, its toxic pulmonary effects have been investigated in many studies. [17] Inhaled NO²potentially affects the parenchyma and the airways of the lungs. Animal studies evaluating the parenchymal effects of high concentrations of inhaled NO²(> 10 ppm) have reported pulmonary edema, hemorrhage, changes in the surface tension properties of surfactant, reduction in the number of alveoli, and death. [18–21] Other animal studies have shown that inhaled NO²at concentrations as low as 2 ppm produced alveolar cell hyperplasia, alteration in surfactant hysteresis, changes in the epithelium of the terminal bronchiole, and loss of cilia. [14,22] In humans, 2.3 ppm NO²has been shown to affect alveolar permeability. [23] Several studies evaluating airway responses to low concentrations of inhaled NO²have reported conflicting results. [17] Some, however, have reported increased airway responsiveness at inhaled [NO²] less than or equal to 2 ppm. [13,24,25] Although the Occupational Safety and Health Administration has set safety limits at 5 ppm, [12] our clinical goal is to maintain inhaled [NO²] as low as possible. Inhaled NO²may remain in the lungs for prolonged periods because it reacts with H sub 2 O to produce HNO³and undergoes irreversible reactive absorption by the pulmonary epithelial lining fluid. [26] Exhaled [NO sub 2] therefore may not be a sensitive indicator of toxic pulmonary concentrations.

A soda lime (94% Ca(OH)², 5% NaOH, and 1% KOH) absorber in the inspiratory gas circuit can be used to remove NO². [5] Stenqvist et al. [27] evaluated the effectiveness of soda lime to absorb NO²and reported that 3 ppm NO²decreased to less than 1 ppm after passing through the absorber. However, soda lime also absorbs NO²; its effectiveness decreases with time; and some forms of soda lime may be ineffective in removal of NO². [28] The use of a soda lime canister introduces other potential problems during mechanical ventilation such as increased resistance to breathing, increased risk of system leaks, more difficult triggering of the ventilator, and modification of inspiratory flow waveforms.

It has been our practice to dilute NO with Nitrogen²or air and then introduce this gas mixture to the air inlet port of the ventilator. We have found this system reliable in the laboratory and in clinical use, and it maintains a constant [NO] and FI^O2with changes in V with dot^Eor inspiratory flow pattern. A method very similar to ours was recently described by Channick et al. [29] Similar to our findings, they reported very low [NO²] with the Puritan-Bennett 7200ae when blending NO with Nitrogen²before delivery to the air inlet of the ventilator. Wessel et al. [30] recently described a method to administer NO with the Servo 900C. They added NO and Oxygen²to the low-pressure inlet at a rate equal to the V with dot^Eand in a proportion to produce the desired [NO] and FI^O2. If the nitric oxide/oxygen flow matches the V with dot sub E1 the Servo 900C bellows should remain relatively empty. They found that this method usually but not always resulted in an [NO²] of less than 3 ppm. We have not used this method but believe that it may be difficult to use in patients with changing V with dot^E.

Delivery of NO distal to the ventilator outlet has been described as a method to avoid NO²production within the ventilator circuit. Benzing et al. [31] introduced an adjustable amount of NO gas (600 ppm NO) into the inspiratory limb with a special injection nozzle. Others have used a nebulizer module that injects NO into the circuit during inspiration. [2,4] NO has also been injected at the Y- piece of the ventilator circuit. [32] These techniques may not deliver a constant [NO] during inspiration, and the additional flow may affect tidal volume and FI^O2. Some methods also complicate analysis of [NO] and [NO²](e.g., injection at the Y-piece). With any method of NO inhalation, monitoring of [NO] and [NO²] is essential.

*Sanborn W: Personal communication, 1994.