Sonographic pleural fluid volume estimation in cats

Abstract

The aims of this study were to evaluate whether a recently published study used to objectively monitor pleural fluid volumes in dogs could be successfully employed in cats and secondly to assess its accuracy. Eleven feline cadavers were selected. Using the trans-sternal view employed in dogs, linear measurements from the pleural surface of the midline of the sternebra at the centre of the heart to the furthest ventro-lateral point of both right and left lung edges were recorded. Isotonic saline was injected using ultrasound guidance into both right and left pleural spaces and the measurements were repeated using standard increments until 400 ml total volume was reached. The mean measurement increased in a linear relationship with the cube root of fluid volume for all cats individually. An equation was produced to predict the volume of fluid from the mean linear measurement for all cats combined:

Volume=[-3.75+2.41(mean)]³ (P<0.001) but variability in the slope of the curve for individual cats limited the accuracy of the combined equation. Equations were derived to predict the constant and slope of the curve for individual cats using the thoracic measurements made, but the residual diagnostic graphs demonstrated considerable variability. As in dogs, good correlation was found between the ultrasonographic measurement and fluid volume within individual cats. An accurate equation to predict absolute pleural fluid volume was not identified. Further analysis with reference to thoracic measurements did not increase accuracy. In conclusion, this study does provide a method of estimating absolute pleural fluid volume in cats, which may be clinical useful for pleural fluid volume monitoring but this is yet to be validated in live cats.

Ultrasound is widely considered to be more sensitive than radiography to the presence of pleural effusion in man but this has not so far been reported in cats or dogs. ^1–3 It is also often used to subjectively monitor fluid volume in cases of chronic effusion. There is no published method to reliably quantify pleural fluid volume in cats, although methods are described in the human literature; in one study, measurements were made of the maximum distance between the visceral and parietal pleura, at both the apex and the base of the pleural cavity at both end-inspiration and end-expiration in clinical patients with pleural effusions. These measurements were then compared against the fluid volume drained from the pleural cavity. ³ A second study used measurements made only at end-expiration, and found greater accuracy, with mean prediction error of volume of 158.4±160.6 ml. ² A method of pleural fluid volume monitoring has been described in cadaveric dogs, ⁴ in which the method used in humans was adapted so that the interpleural distance measurements were standardised using the sternum and heart, rather than the cranial and caudal end of the pleural space, as this is a more readily accessible location in dogs. A pilot study also showed that this location was most readily standardised in canines, but it was not possible to accurately estimate absolute pleural fluid volume using this method. In this study, the authors considered that the main reason that the protocol was unsuccessful in the dog was due to the highly variable morphology of the canine thorax, rather than deficiencies of the protocol itself and so no alterations were made to the protocol prior to its use in the cat. The aim of this study was to assess whether the trans-sternal method of pleural fluid volume monitoring applied in canine cadavers would be applicable in feline cadavers and whether it would be possible to estimate absolute pleural fluid volume using this method.

Materials and Methods

Twelve feline domestic shorthair cadavers were used for the study. No clinical history was available for the cats as they were obtained from a rescue centre, but they were displaying no clinical signs prior to euthanasia. They were euthanased due to feline immunodeficiency virus positive status. The study was carried out within 24 h of euthanasia in all cats. The cats' sex was recorded together with their weight, greatest thoracic height, width and thoracic circumference. The thorax was prepared by clipping the hair over the entire ventral and lateral aspects of the thorax; surgical spirit and coupling gel were applied thoroughly to the skin. An initial ultrasound examination of the entire thorax was carried out using a 6–11 MHz curvilinear probe (Logiq 7, GE Medical Systems, Milwaukee, WI) and cats were excluded if pathology was detected on this initial examination. The cats were placed in sternal-recumbency on an echocardiography table with their sternum over the cut-out. One cat was excluded due to detection of a small volume of pleural fluid on the initial ultrasound exam, so that 11 cats were selected for the study. A 6–11 MHz curvilinear probe was placed ventral to the sternum perpendicular to the skin surface in a transverse plane. The thorax was imaged trans-sternally from the caudal aspect of the sternebra at the level of the centre of the heart. Symmetry of the image was assessed using the shape of the sternebra and adjacent thoracic wall. Linear measurements from the sagittal plane of the sternebra, at the level of the pleural surface of the thoracic wall, to the furthest ventro-lateral point of both right and left lung edges were recorded (Fig 1). All measurements were determined by one operator and repeated three times. Twenty-five millilitre increments of isotonic saline were injected using ultrasound guidance into both right and left pleural cavities and the measurements were repeated each time up to 400 ml total volume. The average measurement of the six values collected for each volume (both left and right sides) was calculated and termed the mean linear measurement, this was used for statistical analysis.

Fig 1.

Ultrasonographic image of the transverse trans-sternal view of the thorax at the level of the centre of the heart with 150 ml of fluid injected; markers show the position of the anatomic linear measurements. Small arrow, sternum; paired large arrows, ventral lung lobe edges.

Statistical analysis

Analyses were performed using Minitab 15 (Minitab Inc; State College, PA, USA) and STATA10 (Statacorp, College Station, Texas, USA). Following basic descriptive statistics and scatter plots, any potential linear relationship between volume of fluid injected and the mean anatomical measurements was investigated. In order to obtain the best linear fit we explored different transformations of the body weight. Least-squares regression analysis with anatomical measurements as predictors of the volume of injected fluid was performed. Initially individual cats were considered separately; and this was followed by a model which included all cats. Cats were included both as fixed effects and then in a mixed-effect linear regression with cat identity as a random effect and assuming an independent covariance structure. Further analyses investigated the ability of the cats' weight and thoracic dimensions to improve model fit.

Of the 11 cats used, there were six females and five males. Their weights ranged from 1.6 to 4.4 kg, with a mean standard deviation (SD) of 3.3 kg (0.84) and a median of 3.2 kg. Their thoracic heights ranged from 9 to 12 cm, with a mean (SD) of 10.8 cm (0.95) and a median of 11 cm. Thoracic widths ranged from 5.5 to 8.5 cm, with a mean (SD) of 7.3 cm (1) and a median of 7.4 cm. Thoracic circumference ranged from 26 to 36.5 cm, with a mean (SD) of 32.2 cm (2.97) and a median of 33 cm.

The initial measurements in the absence of pleural fluid were technically difficult to make (Fig 2) and the measurements became easier once the first increment of fluid was injected. In the same manner as for dogs, as pleural fluid volume increased, the lung lobe tips followed a characteristic pattern of dorsal and lateral displacement by the pleural fluid which accumulated ventrally (Fig 1). The mean linear measurement increased in all cats with increasing pleural fluid volume. A scatterplot of fluid volume injected against mean linear measurement showed a curvilinear relationship for each individual cat. For the population in this study, an overall plot of mean linear measurement against volume allowed estimation of volume for a given linear measurement for all the cats in this study combined (Fig 3). The mean linear measurement increased in an approximately linear relationship with the cube root of fluid volume throughout the range of volumes measured (Fig 4), with the gradient for individual cats varying between 1.98 and 3.83 (R²>90%). The mixed-effect regression produced an equation to predict the volume of fluid (ml) from the mean linear measurement for all cats combined:

Fig 2.

Ultrasonographic image showing the transverse trans-sternal view of the thorax at the level of the mid-cardiac silhouette before fluid injection; the ventral lung lobe tips are more difficult to identify than in Fig 1.

Fig 3.

Plot of pleural fluid volume vs mean linear measurement for the whole population studied displaying the range of volume for a given measurement. Enclosing lines are arbitrary, for illustration purposes.

Fig 4.

Linear regression curves for individual cats, with cube root of volume plotted against mean linear measurement.

$$\text{Volume}={\left[-3.75+2.41\left(\text{mean}\right)\right]}^3\left(P<0.001\right)$$

but variability in the slope of the curve for individual cats limited the accuracy of the combined equation. For example, when a mean linear measurement of 4 cm is entered into the above equation, the result is a volume estimate of 204 ml. The actual range (from Fig 3) was approximately 150–300 ml.

Equations were derived to predict the constant and slope of the curve for individual cats using the thoracic dimensions recorded for each cat and the weight and height of each cat but the residual diagnostic graphs demonstrated considerable variability. Hence, the thoracic dimensions recorded were not useful in increasing the accuracy of the predictive equation.

The anatomical linear measurements tended to be repeatable within either the left or the right side, the maximum variation within one side was 1.58 cm, with a SD of 0.2 and an interquartile range of 0.17.

Discussion

This study shows that the mean linear measurement increased linearly with the cube root of fluid volume and so this method of measurement provides an objective means of monitoring changes in pleural fluid volumes. The results for individual cats showed that there was a consistent relationship between mean measurements and injected volumes throughout the range of volumes measured, unlike the study in dogs in which there was no relationship between injected volumes and mean linear measurement below 100 ml. Measurements were always technically easier than for the dog study ⁴ as the smaller thoracic size in cats permitted the use of a higher frequency probe with higher resolution. It is likely that it was the ability to use a much higher frequency probe in the present study, which made the measurements more accurate at low initial pleural fluid volumes. We considered that due to the relative uniformity of the thoracic anatomy between cats as compared to dogs, the linear measurements would be more consistent between individuals than they were in dogs and may allow an equation to be derived that could be used in all cats to predict volumes of pleural fluid from the mean linear measurements. In the present study, there was a better relationship between change in mean linear measurement and change in the cube root of the volume than in dogs. However, there was much variability in the slope of the curve between individual cats and it was not possible to produce an equation to accurately predict pleural fluid volumes across the study population.

It is also possible that differences in lung compliance between cats may exist resulting in differences in the degree of atelectasis per unit volume of fluid injected for an individual cat. Differences in lung compliance could reflect different degrees of post mortem decomposition, or subclinical disease present premortem.

The population of cats used in this study represent a wide range of size and morphology and are likely to be representative of the feline population as a whole; hence the range of mean linear measurement obtained for each given pleural fluid volume (Fig 3) could potentially be useful in feline clinical patients, giving a range of estimated pleural fluid volume for a given linear measurement. For example, a mean linear measurement of 3.5 cm would give an estimated pleural fluid range of 70–180 ml. However, this study was performed on cadavers, and the behaviour of spontaneous pleural fluid in live cats may vary from that of injected pleural fluid in cadavers, and ideally this possibility should be tested before using this method of pleural fluid volume estimation in clinical cases.

In conclusion, this study demonstrated a method of pleural fluid volume monitoring in cats, but it was not possible to produce an equation to accurately predict absolute fluid volume for all cats. This study does provide a method of estimating absolute pleural fluid volume, but this gives a wide range of values and has not been demonstrated to be valid in live cats.