The postoperative lung function prediction represents a routine in COPD patients undergoing a lung resection [1]. Despite the modern technology, a certain difference may exist between the predicted and postoperative ventilatory parameters (in some COPD patients up to 30%), in a way that ppoFEV₁ may be either underestimated or overestimated [2, 3].

As a flow-volume loop does not recognize“ the position and the motion of each haemidiaphragm, we hypothesised that diaphragm movements might contribute to these differences. Previous pleural infections may lead to the topography opposite to normal and to different motion of haemidiaphragms, thus contributing to the inaccuracy of postoperative lung function prediction.

We set up to determine whether diaphragm motion contributes to differences between predicted and actual postoperative ventilatory function. Also, the aim of the study was to assess the reliability of radiographic and ultrasonographic methods to investigate the diaphragm movements in a clinical setting.

Prospective study on 27 patients with a lung resection for primary lung cancer.

Inclusion criteria: complete resection, uneventful postoperative course, full collaboration with the patient while measuring diaphragm movements.

In all patients, diaphragm movements on both sides were assessed radiographically and by ultrasonography before the operation and at the first outpatient control, 7-10 days after discharge, with the full and stable lung expansion.

The baseline lung function was classified according to GOLD criteria [4].

where [n] relates to the total number of subsegments in the lobe to be removed, whilst [a] relates to the number of subsegments obstructed by the tumor.

where S is the number of resected lung segments, and each segment accounts for 1/19 of total lung function.

Postoperative spirometry was done the same day as the assessment of the diaphragm movements, as already described.

Radiographic measurement

Both preoperative and postoperative chest radiographies were done with a patient in the upright position. Postoperatively, radiographies were done synchronously with the ultrasonographic measurements, as described.

On the chest radiography, the distance between the inferior margin of the second rib posteriorly and horisontal line tangential to the diaphragm dome was measured in deep inspiration (distance a) and deep expiration (distance b) (Figure 1). The preoperative amplitude of the diaphragm movements (A₁) on each side was calculated by substracting the aforementioned distance in expiration (b ₁ ) from the same distance measured in inspiration (a₁): A₁ = a ₁–b ₁.

The same calculation was used to determine the postoperative amplitude (A₂), where a₂ and b₂ correspond to postoperative values of the same distances as in the previous formula:

${\mathrm{A}}_2=a_2–b_2.$

The difference between preoperative and postoperative amplitudes on each side was calculated the formula: _Δ A = A₁–A₂.

For the data analysis and assessment of correlation with the lung function prediction, the difference between preoperative and postoperative amplitudes on diseased side was expressed as a percent of the preoperative amplitude:

${}_{\Delta }{\mathrm{A}}_{\left(\%\right)}{=}_{\Delta }\mathrm{A}\times 100/{\mathrm{A}}_1.$

For the same purpose, the preoperative ipsilateral diaphragm amplitude was presented as a percentage of the preoperative apex-base distance in inspiration:

${\mathrm{A}}_1\left(\%\right)={\mathrm{A}}_1\times 100/\mathrm{a}1$

Ultra-sonographic measurement

With the patient in the supine, 45° semirecumbent position, a 3.75-MHz convex transducer was placed subcostally between the mid-clavicular and mid-axillary line symmetrically to obtain a sagital plane of the hemidiaphragm during all phases of respiration. After identifying the dome of the right and left hemidiaphragm, two-dimensional (2D) scans were performed, by using a real-time gray scale technology in the sagital plane, that included the maximal renal bipolar length. The position of the diaphragm was measured relative to the renal pelvis from the 2D images obtained. Craniocaudal excursion was measured from the renal pelvis to a point on the diaphragm lying at the same depth from the transducer on the ultrasonographic scan (Figure 2). The distance between these points was measured on maximal inspiration and at the end of a forced expiration. For each maneuver, at least three satisfactory readings were taken before selecting a value to be used for analysis.

Of 35 operated patients who completed the protocol, after having eliminated eight patients for different reasons, a total of 27 patients were enrolled in the study. There were 21 males and 6 females (M/F 3.5:1), aged 58.5 ± 8.5 years.

Lobectomy and pneumonectomy were done in 21 and six patients respectively.

Tumour localisation was peripheral, within different lobes in 21 patients, in whom lobectomy was done. Four patients with tumours in the right hilar region and two patients with tumours in the left hilar region, underwent a pneumonectomy.

According to GOLD criteria, 11(40.7%) had COPD of different severity. Stage I COPD existed in six patients, whilst stages II and III existed in two patients each.

Of 6 patients with a pneumonectomy, difference between predicted and postoperatively measured FEV₁ was <250 ml in three patients, in one patient it was 250-300 ml, whilst in two patients it was >300 ml. The mean predicted-measured FEV₁ difference in the pneumonectomy and lobectomy groups was 339 vs. 399.9 ml (P > 0.05).

Before discussing the obtained results, two points should be clarified.

First, despite the reported linear relationship between diaphragmatic excursion and inspired volumes [7], it was also suggested that diaphragm movements, measured by ultrasonography, poorly reflect the pulmonary function [8]. The explanation that various inspiratory volumes are measured for the same diaphragmatic excursion, is not evidence-based.

Second, the point of the radiographically determined normal position of the right hemidiaphragm at the level of the anterior sixth rib, appears to originate from a single study [9, 10]. An obstacle for such a way of referencing a diaphragm position is a poor visibility of costal portions of the anterior ribs. Thoracic spine has also been used as a reference point, but without validation in population studies [11, 12].

Having in mind these limitations, our method, based on the diaphragm apex as a determinant of the diaphragm position, could also be put into question because of use of the postero-anterior projection only, without analysis of movements in the lateral projection. However, studies that analysed both diaphragm apex and costophrenic angle movements, showed that both movements were synchronous and followed a linear relationship [13], thus justifying our method of measurement.

Finally, related to the method of the postoperative lung function prediction, although it was demonstrated that both perfusion scintigraphy and Juhl-Frost formula may correlate well with the observed postoperative FVC and FEV1, the superiority of calculation by using a perfusion scintigraphy was clearly demonstrated [14]. We used the Juhl-Frost method because Nakahara formula is not suitable for pneumonectomy and because the primary study end point was a diaphragm motion, not the prediction method itself. We routinely use perfusion lung scintigraphy in patients with moderate and severe COPD, that was not a case in a subset of patients with a pneumonectomy in the present study.

In the present study, preoperative diaphragm amplitudes determined by ultrasound are in the range of those determined in other studies being 6 to 7 cm, 6 ± 0.7 cm or 6.8 ± 0.8 cm [15]. However, preoperative diaphragm amplitudes differed both depending on the measurement method and diaphragm side. Ultrasonographically measured amplitudes were significantly higher vs. radiographically determined ones (4.28 ± 2.13 cm and 6.51 ± 2.28 cm) only on the diseased side. There are no literature data to compare these results. The probable explanation of the obtained differences is the fact that the reference points for registering diaphragm movements were different, as described in the methods section. In fact, the current study design did not anticipate amplitudes to closely correspond to each other, but to assess their eventual influence to the lung function prediction.

Similarly, it can only be speculated why these differences were smaller on the contralateral side (4.58 ± 1.99 cm vs. 3.67 ± 1.52 cm).

On the other hand, differences in side-to-side diaphragmatic motion are more analysed – ultrasonographically measured values outside the range of 0.5 to 1.6 for the right-to-left ratio of maximal excursion on deap breathing should be considered as abnormal [16]. It can explain our ultrasonographically measured amplitudes on the diseased and contralateral side being 6.51 ± 2.28 cm vs. 3.67 ± 1.52 cm. Difficulties in left hemidiaphragm visualisation are usually regarded as possible cause of these side-to-side differences. So, in one study, the diaphragmatic motion of the left hemidiaphragm was recorded in only 45/210 (21%) subjects [17]. Another study failed to record left hemidiaphragm excursion in 15/23(65%) volunteers [18]. This because the left hemidiaphragm may be obscured by the expanding lung during deep breathing and the position of the probe may not be readily adjusted as the spleen window is small.

The ultrasonographic side-to-side amplitude differences were more pronounced compared with those measured radiographically - 4.28 ± 2.13 cm vs. 4.58 ± 1.99 cm. Although the relevance of the side-to-side diaphragmatic motion comparison has been noted in fluoroscopy studies [19], there are no available literature data trying to explain it.

As expected, after the lung resection, both radiographically and ultrasonographically measured diaphragm amplitudes of the diseased side decreased. On the opposite side, the same trend existed only when amplitudes were determined radiographically. When assessed by ultra sound, postoperative amplitudes on the non-tumour bearing side were higher compared with preoperative ones.

Concerning postoperative percent change of preoperative and postoperative amplitudes, in relation to preoperative values, both methods followed the same trend of postoperative amplitude decrease on the diseased side - 54.3 ± 16.4% and 23.3 ± 28.9% decrease respectively when assessed radiographically and ultrasonographically. As expected, on the non-tumour bearing side, there was no difference between preoperative and postoperative diaphragm amplitudes, independently on the used method.

Concerning the primary end point of this study - influence of the diaphragm movements to discrepancy between the predicted and actual postoperative lung function, it is evident that some aspects of the diaphragm motility may significantly contribute to the lung function prediction, but not as independent factor.

Absence of significant influence of the extent of resection to the lung function prediction is important for practice. A lung function prediction may be more delicate if a lobectomy is anticipated, with several different methods of the lung function prediction being in use, as opposed to a very simple calculation before pneumonectomy by using a perfusion lung scintigraphy [20]. However, the small number of patients with pneumonectomy in the analysed group does not allow firm conclusions about the influence of the extent of resection.

As presented, different ways of expressing diaphragm amplitudes were used in attempt to assess eventual correlation with the lung function prediction. In our study, a significant correlation between diaphragm movements and lung function prediction existed only if the diaphragm movements were presented as 1) diaphragm amplitude as a percentage of the preoperative apex-base distance in inspiration, or 2) as a difference between preoperative and postoperative ipsilateral diaphragm amplitudes. These facts, together with a significant influence of a patients’ height to the lung function prediction, support the need to express the amplitudes of the diaphragm movements in relation to some fixed distance or to take into consideration the loss in diaphragm amplitudes, rather than to correlate absolute values of amplitudes. The exception are emphysema patients, in whom magnetic resonance revealed smaller mean excursions than in control subjects and movements of the ventral portion of the diaphragm in paradox to the change in lung area [21]. Such a bias did not exist in the present study.

Of practical benefit could be our finding that accurate registering of diaphragm movements may predict whether the difference between ppo FEV₁ and actual FEV₁ will exceed 550 ml.

This is the first study addressing the question whether and in which way the diaphragm motion influences the postoperative lung function prediction. The exact way of this influence is still unclear. The present study was not able to suggest the cut-off values for amplitude intervals or their ratios with some fixed intrathoracic distance that could be reproducible and reliable for routine lung function prediction. We believe that it will be possible on larger patient series.