A total of 25 intact male cats scheduled for orchiectomy in the University hospital, aging 1–5 years were investigated in this study. The inclusion criteria were the absence of respiratory and cardiovascular disease history. The presence of abnormalities observed in tomographic scouts of the thorax, in baseline arterial blood gases, or preoperative blood cell count, renal and hepatic plasma chemistry panel were considered as the exclusion criteria.

One day before the experiment, the animals underwent catheterization of the cephalic vein and dorsal pedal artery using a 22-gauge catheter (BD Insyte® Autoguard, Juiz de Fora, MG, Brazil) under sedation with dexmedetomidine hydrochloride (Dexmedetor® Orion Pharma, Espoo, Finland), (10 μg kg⁻¹) intramuscularly (IM). Sedation was reversed with atipamezole (Antisedan® Orion Pharma, Espoo, Finland) (10 μg kg⁻¹) IM and then the animals were placed in cages with food and water ad libitum. The animals fasted for 8 h before the experiments.

On the day of the experiment, the animals were anesthetized with propofol (5 mg kg⁻¹) intravenously (IV) (Propovan; Cristália; São Paulo, Brazil) and underwent endotracheal intubation with a 3.5 mm size, 19 cm length, and a cuffed cannula (Solidor, Bonree Medical, Guangdong, China). Anesthesia was maintained with a continuous infusion of propofol (0.5 mg kg⁻¹ min⁻¹). Paralysis was promoted by the intravenous administration of 1 mg kg⁻¹ of rocuronium (Rocuron; Cristália, São Paulo, Brazil) and supplemented with increments of 0.5 mg kg⁻¹ if the animal presented any signs of spontaneous breathing effort based on the airflow waveform.

After the CT scan protocol, animals were transported to the operation room where neutering surgery was performed.

Prior to the beginning of each study, standard tests of the ventilator (Galileo®, Hamilton Medical, Bonaduz, Switzerland) were performed (pneumotachograph calibration, leak testing, and calculation of the breathing circuit compliance). After the initial ventilator tests, a calibrated Wright spirometer (nSpire, Hertford, UK) was connected between the pneumotachograph and the breathing circuit (neonatal, 150 cm length) to verify the accuracy of V_T measurement computed by the ventilator. A variation of up to 5% between the V_T values expressed by the ventilator and the Wright spirometer was accepted.

After anesthetic induction and positioning of the patient in the dorsal recumbency, pressure-control ventilation (PCV) was initiated with an inspiratory pressure of 5 cmH₂O, zero end-expiratory pressure (ZEEP) f_R of 15 breaths per min, and an inspiratory time of 1 s and FiO₂ 40%.

After 20 min of anesthetic induction and hemodynamic stabilization, inspiratory pressure was progressively increased by increments of 2 cmH₂O every 5 min until reaching 15 cmH₂O. After that, inspiratory pressure was reduced in a descending stepwise manner in steps of 2 cmH₂O until 5 cmH₂O. At the end of each period, an arterial blood sample was collected using 1 ml syringes (Becton Dickinson, Curitiba, Brazil), washed with heparin for blood gas analysis, and an inspiratory pause of 4 s was performed to obtain a thoracic CT image (Figure 1).

At each inspiratory pressure level, the arterial pressure (AP), heart rate (HR), and pulse oximetry (SpO₂) of the animals were evaluated using a multiparameter monitor (Model 2020, Dixtal, Manaus, Brazil). The blood pressure transducer (TruWave®, Edwards, California, USA) was connected to the arterial catheter by noncompliant fluid-filled tubing and was zeroed and positioned at the heart level (elbow joint).

The inspired oxygen fraction (FiO₂, %) oxygen and end-tidal partial pressure of carbon dioxide (EtCO₂, mmHg) were monitored using a side stream gas analyzer (Dixtal, Manaus, Brazil).

The peak and plateau pressure (Pplat, cmH₂O), minute volume, expiratory tidal volume, and respiratory rate were obtained directly from the ventilator. The static compliance was calculated using the following formula:

where C_stat is the static compliance, V_E is the expiratory volume, P_plat is the plateau pressure obtained at the end of a 4 s-inspiratory pause with zero flow, and PEEP is the positive end-expiratory pressure, which was zero through the study.

Arterial blood samples (0.7 mL) were anaerobically collected in heparinized syringes, and blood gas analyses of the pH, partial pressure of oxygen (PaO₂), partial pressure carbon dioxide (PaCO₂), bicarbonate concentration (HCO3-), standard base excess (SBE), and oxygen saturation (SaO₂) were immediately performed (ABL5, Radiometer, Denmark).

After intubation, the cats were placed in dorsal recumbency throughout the entire procedure with their thoracic limbs extended forward carefully to ensure that the spine and head of each animal were in a straight line to acquire symmetrical images. The thoracic CT scan was obtained in dorsal recumbency because most surgeries (like laparotomies) are routinely done in this recumbency. The CT images were obtained at the end of a 4-s inspiratory pause at each inspiratory pressure level, from 1 cm cranial the diaphragmatic dome using a single slice helical CT scanner (Xpress/GX, Toshiba, Japan) with the same settings (120kVp and 150 mA, matrix size 512 × 512). Images were reconstructed at 5 mm thickness and kernel FC50 for standard lung imaging. The images were analyzed using the software Osirix (Osiris 4.19, University Hospital of Geneva, Switzerland). Briefly, each pulmonary region of interest (ROI) was manually delineated and the pixels contained in it were distributed into 1,200 compartments according to their X-ray attenuation coefficient (CT number). Each pixel is characterized by a CT number that represents the attenuation coefficient of the X-ray by the structure being studied minus the attenuation coefficient of water, which is 1, divided by the attenuation coefficient of water expressed in Hounsfield units (HU) (13). Therefore, for each compartment of a known CT number, it was possible to compute tissue and gas volumes and tissue mass using the following formulas:

The total lung tissue and volumes of each given ROI were computed by adding all the partial masses and volumes of the compartments. Lung parenchyma was further analyzed according to its aeration, defined as follows: 1) overinflated parenchyma characterized by CT numbers between −1000 HU and −900 HU (9, 14, 15), 2) normally-aerated parenchyma characterized by CT numbers between −900 HU and −500 HU, 3) poorly-aerated parenchyma characterized by CT numbers between −500 HU and −100 HU (16) and 4) non-aerated parenchyma characterized by CT numbers between −100 and + 100 HU (17).

In the initial analysis, a large ROI encompassing both the left and right lungs was delineated in the CT images obtained in each peak inspiratory pressure level, to evaluate the overall lung volumes and tissue mass, as shown in Figure 1. The portions of the pulmonary hila containing the trachea, main bronchi, and hilar blood vessels were excluded from the ROI. In a second analysis, the left and right lungs were segmented in 3 regions of interest of equal height distributed along the ventral-dorsal axis (ventral, middle, and dorsal, relative to patient's anatomy) and the overinflated lung tissue mass fraction were computed in each airway pressure condition (Figure 1).

The normal distribution of data was evaluated by means of visual analysis and the Shapiro-Wilks test. All data were expressed as the mean value ± SD or as median (25–75% interquartile), according to their distribution. The physiological variables, blood gases variables, and lung CT-derived variables were compared at different levels of inspiratory pressure by means of one-way ANOVA or Friedmann test, followed by multiple comparisons when indicated.

The fraction of the overinflated parenchyma to the overall parenchyma mass within the 3 lung segments (ventral, middle, and dorsal) obtained at each peak inspiratory pressure level was compared using the Kruskal-Wallis test followed by Dun's test. A p-value of < 0.05 was considered significant. All statistical analyses were performed using IBM SPSS package version 22 (IBM Corp, Armonk, NY, USA) and GraphPad Prism 6 for Mac (GraphPad Software, La Jolla, California, USA).

The increase in the inspiratory pressure caused an elevation in the total CT section volume, being 34.2 % greater than baseline at the airway pressure of 15 cmH₂O. The decrease in airway pressure resulted in the return of the lung volume to baseline (p < 0.001). This variation in the total volume was due to an increase in the volume of gas, while the volume of tissue decreased but to a minor degree (Table 1).

The overinflated tissue fraction increased from 2.1% at the airway pressure of 5 cmH₂O to 28.4% at the airway pressure of 15 cmH₂O (p < 0.001), with a reduction of normally-aerated parenchyma from 84.9 to 57.9%. The reduction of airway pressure decreased the amount of overinflated lung parenchyma and restored the normally-aerated parenchyma fraction to baseline values at 5 cmH₂O airway pressure. The fraction of non-aerated and poorly-aerated parenchyma did not change significantly during the study (Table 1).

The amount of overinflated parenchyma progressively increased within the ventral, middle, and dorsal lung segments in proportion to the increase in airway pressure up to 11 cmH₂O in a heterogeneous fashion, as shown in Figure 2. From 5 cmH₂O up to 9 cmH₂O airway pressure, the proportion of overinflated parenchyma was greater in the ventral and middle lung regions. From 11 cmH₂0 airway pressure up to 15 cmH₂O, the overinflation occurring in the dorsal lung region reached the same proportion as observed in the ventral region. At 13 cmH₂O airway pressure, the proportion of overinflated parenchyma observed in the middle lung region was greater than that in the ventral region. A stepwise decrease in inspiratory pressure toward baseline reversed the alterations in the distribution of regional overinflation within the lung parenchyma.

The plateau pressure increased with the elevation of inspiratory airway pressures and returned to baseline after the stepwise decrease in inflation pressure (p < 0.001) (Table 1). The increase in airway pressure promoted a significant increase in tidal volume, from 6.7 ± 2.2 mL/kg at 5 cmH₂O airway pressure to 31.5 ± 9.9 mL/kg at 15 cmH₂O airway pressure (p < 0.001), returning to baseline values at the pressure of 5 cmH₂O (Table 2).

The static compliance increased from 11 cmH₂O airway pressure until 15 cmH₂O of airway pressure. After that pressure level, the static compliance remained higher than baseline until 9 cmH₂0 (p < 0.001). When the airway pressure decreased below 7 cmH₂O, the static compliance decreased returning to baseline values at 5 cmH₂O (Table 2).

Hemodynamic variables of the animals during the study are shown in Table 2. No changes were observed in mean arterial pressure in different airway pressure conditions, however, heart rate increased as peak airway pressure increased from 9 cmH₂0 airway pressure up to 15 cmH₂0 and returned to baseline as airway pressure was decreased to 5 cmH₂O (p < 0.05).

A progressive decrease was observed in PaCO₂ and ETCO₂ when the airway pressures were increased above 9 cmH₂O (p < 0.001). The increase in airway pressure promoted an elevation in pH (p < 0.001). During the stepwise airway pressure decrease, PaCO₂ and ETCO₂ increased and pH values decreased but did not return to baseline. An increase in arterial PaO₂ was observed when airway pressure was raised until 9 cmH₂O (p < 0.001), remaining stable until the end of the experiment. No variations were observed in base excess throughout the experiment, while HCO³ decreased with airway pressure above 11 cmH₂O, returning to baseline level at 5 cmH₂O (Table 2).

This study has some limitations: (A) Tomographic analyses were performed on a single CT image close to the diaphragm. Despite the CT scanning of the whole thorax enabling the evaluation of the distribution of ventilation throughout the entire lungs (43), the time to acquire the image in elevated airway pressure conditions may cause unnecessary hemodynamic derangements. Secondly, it would expose the cats to almost 10x higher doses of radiation and to an avoidable higher risk of cancer (44). Therefore, we decided to obtain fewer images to lower animal exposure to X-rays; (B) The CT section thickness used in this study was 5 mm. This spatial resolution might result in a significant underestimation of the actual overinflation occurring within the lungs when compared to higher spatial resolution with a 2 mm section (45). Conversely, the use of a thicker CT section may minimize the bias introduced in the CT analysis by the cephalic-caudal displacement of the lung parenchyma to the outside of limits of a thinner CT section; (C) Another problem related to the analysis of a single CT section is the partial volume effect that can interfere with the computation of the CT number in voxels nearing the boundaries of the ROI. These voxels may be composed of gas and much more dense structures such as bones. Therefore, it can underestimate the actual CT number values. Since the cross-sectional area of the feline thorax is much smaller than in humans, this effect may be potentially greater. The relative values calculated in one CT section may not be representative of the whole lung. Moreover, a whole lung 3D analysis may result in less overestimation of hyperinflation. (D) Another limiting factor might be the duration of the inspiratory pause used in this study (4 s). According to David et al., pauses longer than 4 s could promote alveolar recruitment (46). Moreover, some studies indicate that an inspiratory pause shorter than 10 s when examining tomographic slice areas may overestimate alveolar recruitment due to overinflation without accounting for the accumulation of gas over the wells, providing the same estimates as in regions that are not gravity-dependent (47). (E) The absence of CT measurement at zero airway pressure during disconnection of the patient from ventilator could be another limitation, as the overinflated lung area at 5 cmH₂O inspiratory pressure cannot be compared to a non-pressurized condition and the 2.1% of overinflated lung could be a representation of voxels within the large airways. (F) Finally, the time spent in each inspiratory pressure might be considered too short, evidenced mainly by differences in PaCO₂ values during inspiratory pressure stepwise increase and decrease. Probably with more time to equilibrate CO2 washout, these values would be more similar at the same inspiratory pressure level. Moreover, it is known that a lack of PEEP is related to the appearance of collapsed areas and acute lung injury by cyclic opening and closing of the alveolar units, and increasing time at each pressure level, we could expect greater areas of the non-aerated lung. However, a longer time at each inspiratory pressure would greatly increase the total time of anesthesia in each animal, as well as the risk of hemodynamic instability.

In conclusion, increases in airway pressure beyond 5 cmH₂O cause a progressive overinflation of the lungs that can encompass 28.4% of the parenchyma at 15 cmH2O. Ventilation with low driving pressures and with adequate respiratory rate may provide the best ventilatory strategy to small animals with highly compliant chest walls.