Introduction

Large volume pleural effusion (PE) usually induces specific pathophysiological changes, including an increase in pleural pressure, a decrease in lung volumes, as well as an impairment in respiratory muscle function and gas exchange. These alterations are closely related to clinical symptoms reported by patients, that is, chest tightness, dyspnea, and cough.1 Therapeutic thoracentesis is highly effective in providing symptom improvement,2 however, some physiological effects of pleural fluid withdrawal have still not been fully elucidated.

The relationship between the volume of withdrawn pleural fluid and pulmonary function was evaluated in both animal and human studies.3-6 These studies showed that the changes in lung volumes were significantly smaller than the amount of removed fluid. The increase in forced vital capacity (FVC) and total lung capacity (TLC) corresponded to only 10.5%–33.6% and 20.8%–40.2% of withdrawn pleural fluid volume, respectively.5,7-9 These data suggest that the presence of PE not only results in lung compression but also in a significant external expansion of the chest.10-12

Several studies that evaluated the impact of therapeutic thoracentesis on arterial blood gases (ABG) gave conflicting results. Brandstetter and Cohen13 showed that oxygen partial pressure in arterial blood (PaO2) significantly decreased directly after thoracentesis and returned to the pre-thoracentesis value after 24 hours. In contrast, Perpiñá et al14 found a significant and constant increase in PaO2 after the procedure, reaching a maximum at 24 hours. Another pattern was reported by Agustí et al,15 who found no significant effect of a therapeutic thoracentesis on PaO2. The differences between the results of the above studies are probably related to various factors, including the relationship between the volume of fluid withdrawn and pleural pressure (Ppl), that is, pleural elastance (Pel).12,16 This can be supported by the results reported by Chen et al17 pointing to a strong inverse correlation between improvement in oxygen partial pressure to fraction of inspired oxygen ratio and Pel.

To our knowledge, there have been no studies that simultaneously evaluated complex relationships between pleural fluid volume, Ppl, pulmonary function, and gas exchange. To date, the volume of withdrawn fluid and Ppl changes were analyzed either in the context of pulmonary function2,8 or in the context of ABG.16,17 This was partly because the measurement and registration of Ppl were not widely available. Introduction of new digital pleural manometers not only allows clinicians to precisely follow Pel during pleural fluid removal but also enables monitoring of Ppl changes in different phases of the respiratory cycle. Using an advanced system for pleural manometry, we were able to demonstrate novel data on the patterns of pleural pressure amplitude (Ppl_ampl) changes during the fluid withdrawal and to report a new phenomenon termed pleural pressure pulse.18,19 The aim of the current study was to assess the impact of therapeutic thoracentesis on pulmonary function and gas exchange in relation to both the volume of withdrawn pleural fluid and the changes in Ppl.

Patients and methods

General study design

This prospective, single-center, cross-sectional study was performed between January 2016 and January 2019. Consecutive patients with PE who required therapeutic thoracentesis were enrolled. All patients underwent thoracentesis with pleural pressure measurement, pre- and postprocedure pulmonary function testing (PFT), and ABG analysis. The study protocol was approved by the Institutional Review Board (KB 105/2012) and registered at ClinicalTrials.gov (NCT02192138). All patients signed an informed consent to participate in the study.

Patients

Adult patients (aged 18–85 years) with dyspnea due to PE scheduled for therapeutic thoracentesis were asked to participate in the study. Additional inclusion criteria were: 1) a relatively good general health condition allowing for thoracentesis with prolonged, semi-continuous Ppl monitoring, 2) patient consent to be managed according to the study protocol. The exclusion criteria were: 1) contraindications for a therapeutic thoracentesis, 2) small volume PE (occupying less than one-third of the hemithorax on the erect posteroanterior chest radiograph).

Pre- and post-thoracentesis assessment of symptoms, pulmonary function, and arterial blood gases

Dyspnea was assessed using the 10 point modified Borg scale directly before and immediately after termination of thoracentesis. Pulmonary function was measured a day before thoracentesis (T-24) and then 3 hours (T3) and 24 hours (T24) after the pleural fluid removal. Pulmonary function testing included spirometry (Lung Test 1000, MES, Kraków, Poland), whole body plethysmography, and transfer factor for carbon monoxide (TLCO) (Bodybox 5500 Medisoft, Dinant, Belgium). All the procedures were performed according to the American Thoracic Society recommendations and European Respiratory Society task force guidelines.20-22

Arterial blood gases were measured 1 hour before thoracentesis (baseline, T0) and then 1 hour (T1), 3 hours (T3), and 24 hours (T24) after termination of pleural fluid withdrawal. Arterial blood samples were taken from the radial or ulnar artery in sitting patients while breathing ambient air. Blood gases were measured within 15 minutes from blood sample collection using Blood Gas Analyzer ABL 800 FLEX Radiometer (Medical ApS, Brønshøj, Denmark).

Thoracentesis and pleural pressure measurement

Thoracentesis with Ppl monitoring was performed in sitting position, as described previously.18,19,23 After termination of fluid withdrawal, the pleural catheter was fixed to the chest wall for subsequent measurements. The measurements taken at the termination of thoracentesis, 3 hours, and 24 hours after the procedure were labeled as TT, T3, and T24, respectively. A special, home-built software was used to perform a reliable analysis of all data collected during the procedure.18

Tidal volume measurement during thoracentesis

An adapted spirometer (Lung Test 1000) with its signal coupled with a signal from a pleural manometer was used to register air flow and calculate the tidal volume (TV) simultaneously with Ppl curve registration. Minute ventilation (V̇) was calculated as the product of the respiratory rate (RR) and TV.

Statistical analysis

Data are presented as median and interquartile range (IQR, 25th to 75th percentile). Statistical analysis was performed using 2 statistical packages: Statistica 12.0 (StatSoft Inc., Tulsa, Oklahoma, United States) and MedCalc Statistical Software version 13.2.2 (MedCalc Software bvba, Ostend, Belgium). Changes in PFT, ABG, Ppl, Ppl_ampl, TV, and RR between 2 different time points are presented as a difference between the second and the first measurement. The Friedman test with the post-hoc Conover-Iman test were used to compare dependent samples in terms of continuous and ordinal variables measured at more than 2 time points. When a primary outcome was a difference between only 2 time points, the Wilcoxon signed-rank test was applied. In addition, statistical significance was tested using multiple comparisons with the Bonferroni correction. To evaluate pre- and post-thoracentesis parameters in relation to withdrawn pleural fluid volume, the patients were categorized as undergoing moderate (<⁠1000 ml), large (1000–1999 ml), and very large (≥ 2000 ml) volume thoracentesis (M, L, and VL group, respectively). The differences between continuous variables in 3 independent groups were tested using the non-parametric Kruskal-Wallis test and the post-hoc Conover-Iman test. The Spearman coefficient was calculated to test correlations between 2 variables. All P values were 2-tailed and P below 0.05 was considered significant.

Results

Forty five patients were initially enrolled. As 8 patients had to be excluded from the analysis (Supplementary material, Figure S1), the study group included 37 patients (26 women), at the median age of 64 years (IQR, 58–72). Malignancies were the most common cause of PE (29 patients with solid tumors, 4 patients with lymphoma), and the median volume of withdrawn fluid was 1760 ml (IQR, 1310–2170). A detailed characteristic of the study group is presented in Supplementary material, Table S1.

Pulmonary function, pleural pressure and arterial blood gases (whole group analysis)

There were significant differences between baseline (T-24), T3 and T24 values of different lung function parameters (Table 1).

Table 1. Pulmonary function parameters before, 3 hours and 24 hours after therapeutic thoracentesis

Parameter

Baseline (T-24)

3 h after therapeutic thoracentesis (T3)

Change after 3 h (T-24–T3)

24 h after therapeutic thoracentesis (T24)

Change after 24 h (T-24–T24)

ml

% of baseline

P value

ml

% of baseline

P value

FVC, l

1.60 (1.30–2.00)

1.76 (1.50–2.30)

200 (0–500)

6.4

0.008

1.94 (1.60–2.35)

300 (100–600)

9

<⁠0.001

FEV1, l

1.10 (0.92–1.40)

1.36 (0.90–1.70)

100 (0–300)

3.9

0.01

1.24 (1.00–1.70)

200 (100–500)

7.8

<⁠0.001

TLC, l

3.60 (3.10–4.30)

4.20 (3.70–4.90)

540 (170–800)

11.9

<⁠0.001

4.39 (3.97–4.80)

600 (370–1.000)

12.5

<⁠0.001

FRC, l

2.40 (2.10–3.00)

2.90 (2.50–3.40)

440 (150–650)

14.9

<⁠0.001

2.95 (2.60–3.30)

500 (300–800)

16.1

<⁠0.001

RV, l

1.95 (1.80–2.20)

2.30 (2.00–2.50)

210 (60–450)

11.5

0.002

2.25 (2.00–2.60)

200 (–50–400)

8.3

0.01

TLCO, ml/mm Hg/min

12.4 (11.1–14.6)

13.6 (10.6–16.7)

1.0 (0.2–2.0)a

4.8

0.009

12.9 (11.5–17.2)

0.9 (0.0–3.1)a

5.5

0.02

Data are presented as median (interquartile range) or the percentage of baseline value.

P values were calculated using the Friedman test and the post-hoc Conover-Iman tests.

a ml/mm Hg/min

Abbreviations: FEV1, forced expiratory volume in the first second; FRC, functional residual capacity; FVC, forced vital capacity; RV, residual volume; TLC, total lung capacity; TLCO, transfer factor for carbon monoxide

The patterns of pulmonary function changes between different time points expressed as the percentage of withdrawn pleural fluid volume are presented in Figure 1.

There were significant positive correlations between the increase in lung volumes, the withdrawn fluid volume and Ppl, as well as negative correlations between the increase in lung volumes and Pel at TT point (Table 2).

Table 2. Correlations between the increase in lung volumes, the volume of withdrawn pleural fluid, and pleural pressure at TT and T3 points

Parameter and its change

Withdrawn pleural fluid volume, l

Ppl in TT point, cm H2O

Ppl difference between T0 and TT

Pel, cmH2O/l calculated at TT

Ppl difference between T0 and T3

Pel, cmH2O/l

calculated at T3

FVC

Change between T-24 and T3 point

FVC, l

R = 0.53, P = 0.004

R = 0.25, P = 0.19

R = –0.07, P = 0.71

R = –0.35, P = 0.07

R = –0,25, P = 0.25

R = –0.13, P = 0.54

FVC% pred

R = 0.60, P = 0.001

R = 0.41, P = 0.04

R = –0.28, P = 0.18

R = –0.56, P = 0.003

R = –0.22, P = 0.33

R = –0.07, P = 0.75

Change between T-24 and T24 point

FVC, l

R = 0.52, P = 0.005

R = 0.46, P = 0.02

R = –0.39, P = 0.04

R = –0.42, P = 0.03

R = –0.38, P = 0.07

R = –0.18, P = 0.40

FVC% pred

R = 0.43, P = 0.02

R = 0.41, P = 0.03

R = –0.4, P = 0.04

R = –0.48, P = 0.01

R = -0.26, P = 0.23

R = -0.09, P = 0.68

FEV1

Change between T-24 and T3 point

FEV1, l

R = 0.55, P = 0.004

R = 0.30, P = 0.15

R = –0.27, P = 0.19

R = –0.47, P = 0.02

R = –0.09, P = 0.69

R = 0.05, P = 0.80

FEV1% pred

R = 0.48, P = 0.01

R = 0.31, P = 0.13

R = –0.29, P = 0.16

R = –0.48, P = 0.01

R = –0.08, P = 0.72

R = 0.12, P = 0.58

Change between T-24 and T24 point

FEV1, l

R = 0.51, P = 0.007

R = 0.49, P = 0.008

R =–0.47, P = 0.01

R = –0.48, P = 0.01

R = –0.22, P = 0.30

R = 0.00, P = 0.98

FEV1% pred

R = 0.40, P = 0.04

R = 0.42, P = 0.03

R = –0.48, P = 0.01

R = –0.45, P = 0.02

R = –0.03, P = 0.87

R = 0.12, P = 0.58

TLC

Change between T-24 and T3 point

TLC, l

R = 0.65, P <⁠0.001

R = 0.39, P = 0.05

R = –0.23, P = 0.28

R = –0.48, P = 0.02

R = –0.27, P = 0.25

R = –0.06, P = 0.80

TLC% pred

R = 0.58, P = 0.003

R = 0.43, P = 0.03

R = –0.30, P = 0.16

R = –0.44, P = 0.03

R = –0.23, P = 0.32

R = –0.05, P = 0.81

Change between T-24 and T24 point

TLC, l

R = 0.69, P <⁠0.001

R = 0.60, P = 0.002

R =–0.38, P = 0.06

R = –0.46, P = 0.02

R =–0.33, P = 0.14

R = 0.01, P = 0.93

TLC% pred

R = 0.56, P = 0.004

R = 0.60, P = 0.001

R = –0.43, P = 0.03

R = –0.48, P = 0.02

R = –0.25, P = 0.26

R = 0.06, P = 0.78

FRC

Change between T-24 and T3 point

FRC, l

R = 0.42, P = 0.04

R = 0.32, P = 0.12

R = –0.11, P = 0.59

R = –0.21, P = 0.31

R = –0.19, P = 0.41

R = 0.03, P = 0.88

FRC% pred

R = 0.43, P = 0.04

R = 0.35, P = 0.08

R = –0.14, P = 0.50

R = –0.25, P = 0.23

R = –0.21, P = 0.37

R = –0.01, P = 0.98

Change between T-24 and T24 point

FRC, l

R = 0.54, P = 0.005

R = 0.42, P = 0.04

R = –0.14, P = 0.49

R = –0.47, P = 0.02

R = –0.33, P = 0.13

R = –0.05, P = 0.82

FRC% pred

R = 0.51, P = 0.009

R = 0.49, P = 0.01

R = –0.22, P = 0.29

R = –0.50, P = 0.01

R = –0.30, P = 0.17

R = –0.02, P = 0.91

TLCO

Change between T-24 and T3 point

TLCO, ml/mm Hg/min

R = 0.71, P <⁠0.001

R = 0.38, P = 0.08

R = –0.18, P = 0.43

R = –0.48, P = 0.03

R = –0.15, P = 0.56

R = –0.02, P = 0.91

TLCO% pred

R = 0.71, P = 0.003

R = 0.48, P = 0.03

R = –0.28, P = 0.22

R = –0.57, P = 0.006

R = –0.14, P = 0.59

R = –0.03, P = 0.88

Change between T-24 and T24 point

TLCO, ml/mm Hg/min

R = 0.63, P = 0.002

R = 0.51, P = 0.02

R = –0.47, P = 0.03

R = –0.44, P = 0.04

R = –0.44, P = 0.08

R = –0.18, P = 0.49

TLCO% pred

R = 0.56, P = 0.006

R = 0.49, P = 0.02

R = –0.52, P = 0.01

R = –0.40, P = 0.06

R = –0.40, P = 0.10

R = –0.21, P = 0.41

Data are presented as Spearman rank correlation coefficient (R) and P for the change in pulmonary function parameters expressed as absolute values and % of predicted.

Abbreviations: Pel, pleural elastance; Ppl, pleural pressure; others, see Table 1

Table 3 shows the differences between pre- and post-thoracentesis ABG, RR, TV, Ppl, and Ppl_ampl. A significant increase in PaO2 was demonstrated between baseline and T1 as well as T3 points with concomitant decrease in alveolar-arterial gradient. There was a transient but highly significant increase in RR at TT, T1, and T3 points as compared with T0, and a significant increase in TV measured at T3 and T24 points (Table 3). Albeit V̇ did not change significantly at TT point, its significant increase in relation to T0 was noted at T1, T3, and T24 points (Table 3). The pattern of V̇ was heterogeneous, with V̇ increase at TT, T1, and T3 (as compared with T0) in 18/37, 22/32, and 21/27 patients, respectively, and a small decrease in the remaining patients. Highly significant changes were also observed in Ppl and Ppl_ampl (P <⁠0.001) between T0 and TT measurements (Table 3).

Table 3. Arterial blood gases, respiratory rate, tidal volume, pleural pressure and pleural pressure amplitude before and after therapeutic thoracentesis

Parameter

Baseline value (T0)

Termination of therapeutic thoracentesis (TT)

1 h after therapeutic thoracentesis (T1)

3 h after therapeutic thoracentesis (T3)

24 h after therapeutic thoracentesis (T24)

Value

P value (T0–TT)

Value

P value (T0–T1)

Value

P value (T0–T3)

Value

P value (T0–T24)

PaO2, mm Hg

71.3 (64.5–75.8)

NA

NA

79.9 (71.5–82)

0.01

75.1 (66.4–82.2)

0.009

74.8 (63.6–79.2)

0.08

PaCO2, mm Hg

35.4 (33.6–38.9)

NA

NA

36.9 (33.9–39)

>0.99

35.9 (32.9–39.1)

>0.99

37.1 (34.1–39.1)

>0.99

P(A-a)O2, mm Hg

34.9 (29.2–42.9)

NA

NA

30.6 (23.8–36.9)

0.007

32.9 (25.4–43.5)

0.02

34.5 (30.2–42.2)

0.20

RR, 1/min

23.9 (19.1–28.6)

27 (24.4–35)

<⁠0.001

24.6 (22–27.2)

0.001

25.2 (21.3–28.8)

0.02

22.7 (20–27)

0.30

TV, ml

440 (390–540)

400 (290–530)

0.30

480 (440–585)

0.05

530 (400–620)

0.03

500 (405–580)

0.04

V̇, l/min

10.5 (8.5–14.1)

12.1 (8.0–14.7)

0.26

12.9 (10.7–16.0)

0.003

12.5 (9.3–16.5)

0.001

13.2 (10.1–16.1)

0.02

Ppl, cmH2O

3.5 (–0.5 to 5.7)

–15.5 (–18.8 to –7.3)

<⁠0.001

NA

NA

–0.64 (–3.8 to 2.2)

0.001

–1.5 (–2.6 to 1.8)

0.02

Ppl_ampl, cm H2O

4.2 (2.9–5.7)

11.2 (10–17.2)

<⁠0.001

NA

NA

7.2 (4.6–11)

<⁠0.001

6.1 (4.5–11.6)

0.07

Data are presented as median (interquartile range).

P values were calculated using the Friedman test and the post-hoc Conover-Iman tests.

Abbreviations: NA, not applicable (no measurement scheduled at this time point); PaO2, partial oxygen pressure in the arterial blood; PaCO2, partial carbon dioxide pressure in the arterial blood; P(A-a)O2, alveolar-arterial oxygen difference; Ppl_ampl, pleural pressure amplitude; RR, respiratory rate; TV, tidal volume; V̇, minute ventilation calculated as RR × TV; others, see Table 2

Changes in pulmonary function testing, arterial blood gases, respiratory rate, tidal volume and pleural pressure in relation to thoracentesis volume

The median volumes of withdrawn pleural fluid in M (n = 7), L (n = 16) and VL (n = 14) groups were 800 ml (IQR, 500–950), 1500 ml (IQR, 1322–1780), and 2320 ml (IQR, 2100–2655), respectively. Pulmonary function changes in these 3 groups are presented in Table 4. The highest and the lowest baseline values of FVC (expressed as % of predicted) were found in the M and VL group, respectively. There was virtually no increase in FVC in the M group, while a significant increase was demonstrated for both L and VL groups (Table 4). In consequence, there were significant differences between the M and VL groups in terms of FVC improvement (1% vs 18%, P = 0.02 at T3 point and –3% vs 13%, P = 0.02 at T24 point). The pattern of forced expiratory volume in the first second (FEV1) and TLC alterations was very similar, that is, no significant change in the M group and significant increase in the L and, in particular, the VL group. Highly significant differences were also found for TLCO% predicted. Although the baseline values were similar in the 3 groups, TLCO increased in both the L and VL group at T3 (P = 0.005 and P = 0.04, respectively) and T24 (P = 0.009 and P = 0.025, respectively) (Table 4).

Table 4. Comparative results of pulmonary function testing in patients undergoing moderate (M), large (L) and very large (VL) volume thoracentesis

Parameter

Group

Time point

T-24

T3

T24

T-24–T3 difference

T-24–T24 difference

Value

P value (intra)

P value (inter)

Value

P value (intra)

P value (inter)

M vs L

M vs VL

L vs VL

M vs L

M vs VL

L vs VL

FVC

Group M (n = 6)

1.65 l (1.5–2.2);

58.2% (49–70)

1.65 l (1.4–2.3);

59.2% (41–72)

1.53 l (1.3–2.0);

48.3% (39–61)

0.05 l (–0.2 to 0.1);

1% (–0.3–2.4)

0.68

0.47

0.048

0.07

–0.07 l (–0.15 to 0.13);

–3% (–4.4 to 2.8)

0.71

0.09

0.02

0.87

Group L (n = 15)

1.51 l (1.3–2.1);

54.9% (49–71)

1.7 l (1.6–2.2);

68.2% (58–75)

1.97 l (1.7–2.3);

77.3% (65–81)

0.2 l (0.05–0.4);

6.6% (1.8–15)

0.14

0.3 l (0.1–0.5);

9% (4.2–21)

0.002a

Group VL (n = 13)

1.63 l (1.2–1.8);

45.5% (40–55)

2.1 l (1.6–2.3);

69.2% (46–79)

2.2 l (1.6–2.4)

62.5% (48–77)

0.48 l (0.2–0.7);

18% (7.3–24)

0.03

0.45 l (0.3–0.8);

13.3% (9–20)

0.005a

FEV1

Group M (n = 6)

1.25 l (1–1.8);

46.6% (38–63)

1.4 l (0.9–1.7);

47.6% (38–61)

1.1 l (0.9–1.6);

40.2% (37–55)

–0.01 l (–0.1 to 0.1);

0.65% (–2.1 to 3)

0.8

0.35

0.02

0.66

0.03 l (–0.05 to 0.1);

1.65% (–1.7 to 4.6)

0.45

0.3

0.03

0.62

Group L (n = 15)

1.1 l (0.8–1.4);

48.9% (46–62)

1.26 l (1–1.5);

61.1% (49–69)

1.24 l (1.1–1.5);

66.7% (49–73)

0.15 l (0.05–0.3);

4.6% (1.0–16)

0.06

0.2 l (0.1–0.4);

7.3% (3.4–19.7)

0.02a

Group VL (n = 12)

1.05 l (0.9–1.3);

41.9% (39–48)

1.39 l (0.9–1.7);

52% (39–70)

1.5 l (1–1.8);

55.8% (43–65)

0.37 l (0.1–0.6);

17% (2.6–20.4)

0.02a

0.33 l (0.2–0.6);

12.7% (7.8–16.8)

0.005a

TLC

Group M (n = 6)

4.1 l (3.3–5.2);

77.1% (68–88)

4.0 l (3.8–5.2);

81% (79–87)

4.1 l (3.6–4.9);

88% (65–105)

–0.05 l (–0.1 to 0.4);

0.05% (–0.8 to 8.2)

0.9

0.06

0.004

0.54

0.19 l (–0.2 to 0.3);

4.4% (–2.3–7.4)

0.58

0.28

0.02

0.34

Group L (n = 14)

3.6 l (3.0–4.2);

76.8% (66–88)

4.3 l (3.5–4.7);

88.5% (76–102)

4.29 l (3.7–4.7);

89.3% (78–99)

0.59 l (0.4–0.7);

13.8% (7.6–14.6)

0.003

0.6 l (0.3–0.8);

12.4% (6.7–18)

0.006a

Group VL (n = 10)

3.5 l (3.1–3.9);

68.9% (67–76)

4.6 l (3.8–5.6);

88% (66–99)

4.6 l (4.0–4.9);

90.5% (76–94)

0.9 l (0.8–1.1);

18.6% (13.2–20.7)

0.03a

1.1 l (0.7–1.1);

19.1% (13.3–22.6)

0.007a

FRC

Group M (n = 6)

2.5 l (2.3–3.3);

93.6% (85–97)

2.95 l (2.7–3.4);

101.6% (90–115)

2.55 l (2.2–3.5);

90.4% (71–136)

0.2 l (0.1–0.5);

8.2 (4.1–15.6)

0.04

0.78

0.3

>0.99

0.25 l (–0.05 to 0.3);

8.3 (–1.9 to 11.2)

0.35

0.19

0.13

>0.99

Group L (n = 14)

2.15 l (1.9–2.9);

87.6% (76–97)

2.55 l (2.3–3.4);

97.8% (94–121)

2.8 l (2.5–3.2);

110% (101–121)

0.44 l (0.25–0.65);

16.5% (9.7–23.6)

0.005a

0.6 l (0.27–0.8);

22.9% (9.4–29.6))

0.03

Group VL (n = 10)

2.55 l (2.2–3.0);

91.6% (84–93)

3.1 l (2.5–3.6);

101% (88–118)

3.0 l (2.9–3.7);

110% (103–118)

0.65 l (0.3–0.9);

25.5% (7.8–33)

0.07

0.7 l (0.35–0.8);

25.1% (12.6–28)

0.01a

TLCO

Group M (n = 6)

12.5 (11.5–13.8)b;

54.1% (48–62)

11.4 (10.3–14.8)b;

51.3% (47–56)

11.6 (10.9–12.3)b;

49.9% (44–59)

–1.1 (–1.6 to –0.7)b;

–5.8% (–6.3 to –2.8)

0.07

0.02

0.01

>0.99

–0.9 (–1.9 to 0.07)b;

–4.2% (–7.7 to 0.5)

0.27

0.16

0.01

0.57

Group L (n = 13)

11.8 (10.5–13.8)b;

57.6% (52–64)

14.1 (10.7–16.9)b;

67% (55–75)

12.7 (9.8–17)b;

63.2% (52–68)

1.4 (0.9–2)b;

5.4% (4.8–10.4)

0.005a

0.9 (0.4–2)b;

5.8% (1.6–10.2)

0.01a

Group VL (n = 10)

12.8 (12.3–15.4)b;

59.1% (54–61)

14.4 (11.4–19.1)b;

64.7% (50–74)

15.9 (12.6–19.3)b;

65.4% (56–80)

2.6 (0.7–3.5)b;

11.5% (2.4–13)

0.04

3.2 (1.3–4.6)b;

13.5% (4.9–19)

0.03

Data are presented as median (interquartile range) for both absolute values and % of predicted (columns 3–5), subsequent column sections (T-24–T3 difference and T-24–T24 difference) include data on within-group (M, L, and VL separately) difference between the measurements at 2 different time points, as well as P values for within- and between-group comparisons: P (intra) = P for intragroup comparisons (within M, L, and VL group) between T-24 and T3 time points, as well as between T-24 and T24 points (Wilcoxon test); P (inter) = P for intergroup comparisons (between M, L and VL groups) of the difference between T-24 and T3 points as well as between T-24 and T24 points (Kruskal-Wallis test with post-hoc Conover-Iman test).

a Statistical significance (P intra <⁠0.05) for multiple comparisons using the Bonferroni correction; b Absolute values expressed as ml/mm Hg/min

Abbreviations: see Tables 1 and 2

The 3 groups defined by the volume of withdrawn fluid differed significantly in terms of Ppl, Ppl_ampl, and Pel changes (Table 5). The lowest median initial Ppl was demonstrated in the M group, while the highest in the VL group. The differences became even more significant at TT point (M vs L P = 0.01 and M vs VL P = 0.04). The relationships between the volume of withdrawn fluid and Ppl in individual patients from the M, L, and VL groups are shown in Supplementary material, Figure S2.

Table 5. Changes in pleural pressure, pleural pressure amplitude, pleural elastance, and dyspnea in patients with moderate (M), large (L) and very large (VL) volume of withdrawn pleural fluid

Parameter

Group

Time point

T0

TT

T3

T24

TT–T0 difference

Value

P value

(intra)

P value

(inter)

M vs L

M vs VL

L vs VL

Ppl, cm H2O

Group M (n = 7)

–0.87 (–3.5 to 4.25)

–25.7 (–57.3 to –16.2)

–0.43 (–2.1 to 0.9)

1.8 (–2.1 to 2.3)

–28 (–52.4 to –18.2)

0.02a

0.03

0.37

0.61

Group L (n = 16)

2.43 (–0.3 to 5.1)

–13.4 (–16.4 to –7.2)

–0.4 (–3.7 to 3.5)

–0.09 (–3.2 to 4.1)

–16.6 (–18.5 to –11.4)

<⁠0.001a

Group VL (n = 14)

5.0 (1.6–8.8)

–14.5 (–18.8 to –7.0)

–1.6 (–7.2 to 1.3)

–2.3 (–2.6 to 0.6)

–19.6 (–26.8 to –11.5)

<⁠0.001a

Ppl_ampl, cm H2O

Group M (n = 7)

5.8 (4.9–12.2)

17.5 (12–34.6)

8.3 (4.8–11.4)

10.3 (3.6–17.0)

9.8 (7.1–23.2)

0.02a

0.28

0.43

>0.99

Group L (n = 16)

3.4 (2.8–5.3)

10.8 (10.1–15.6)

6.4 (2.9–9.7)

5.7 (2.5–14.3)

7.3 (4.1–12.8)

<⁠0.001a

Group VL (n = 14)

3.7 (3.1–5.0)

10.8 (9.3–14.8)

8.0 (5.5–12.2)

6.2 (5.1–10.6)

7.2 (5.1–8.8)

<⁠0.001a

Pel, cm H2O/l

Group M (n = 7)

NA

36.3 (22.1–87.4)

NA

NA

NA

NA

0.01

0.005

>0.99

Group L (n = 16)

NA

12.2 (11.1–16.3)

NA

NA

NA

NA

Group VL (n = 14)

NA

12.0 (9.3–18.8)

NA

NA

NA

NA

Dyspnea, Borg scale

Group M (n = 7)

3 (3–5)

3 (1–4)

NA

NA

–0.5 (–3 to 0)

0.27

>0.99

>0.99

>0.99

Group L (n = 16)

3 (2–4)

2 (1–4)

NA

NA

–1 (–2 to 1)

0.26

Group VL (n = 14)

4 (4–5.5)

2.5 (2–4)

NA

NA

–1.5 (–2.5 to 0)

0.04

Data presented as median and interquartile range (columns 3–6), subsequent column section (TT–T0 difference) includes data on within-group (M, L, and VL separately) difference between the measurements at TT and T0 time points, as well as P values for within and between group comparisons: P (intra) = P for intragroup comparisons (within M, L and VL group) between T0 and TT time points (Wilcoxon test); P (inter) = P for intergroup comparison (between M, L, and VL groups) of the difference between T0 and TT point (the Kruskal-Wallis test with post-hoc Conover-Iman test)

a Statistical significance (P intra <⁠0.05) for multiple comparisons using the Bonferroni correction

Abbreviations: see Tables 2 and 3

A significant negative correlation was found between the volume of withdrawn pleural fluid and Pel (R = –0.57, P <⁠0.001). High overall Pel in the M group significantly exceeded that calculated for the L and VL groups (Table 5). The median volume of removed pleural fluid was significantly larger in patients with normal (≤14.5 cmH2O/l) than in those with increased Pel (1950 ml [IQR, 1630–2400] vs 1277 ml [IQR, 900–2000], respectively; P = 0.002).

Although there was a significant decrease in dyspnea sensation soon after thoracentesis in the entire study population (P = 0.013), we did not observe significant differences in dyspnea improvement between the M, L, and VL groups. The patients with decreased dyspnea score after fluid withdrawal were characterized by a significantly lower Pel than those in whom dyspnea did not improve (Table 5).

Median baseline PaO2 values were virtually the same in the M, L, and VL groups (Table 6). The most relevant increase in PaO2 at T1 was found in the L group (P = 0.003). This increase was significantly higher than that in the M group (2.7 vs 6.6 mm Hg, P = 0.046). In all groups, a gradual decrease in PaO2 was noted at T3 and T24 points as compared with T1 results (Table 6). When RR and TV changes were compared between the groups, no significant differences were found, except for those previously mentioned for the entire study group.

Table 6. Changes in arterial blood gases, respiratory rate, and tidal volume in patients with moderate (M), large (L) and very large (VL) volume of withdrawn pleural fluid

Parameter

Group

Time point

T0

TT

T1

T3

T24

T1–T0 difference

Value

P value (intra)

P value (inter)

M vs L

M vs VL

L vs VL

PaO2, mm Hg

Group M (n = 7)

72.6 (57.6–83)

NA

79.4 (59.2–86.1)

72.6 (58–87.4)

68.2 (60–84.1)

2.7 (–2.7 to 3.5)

0.60

0.046

>0.99

0.09

Group L (n = 15)

71.3 (66.4–75.6)

NA

80.4 (75.3–81.2)

77 (66.4–82.4)

72.6 (63.6–77.3)

6.6 (3.3–10.3)

0.003a

Group VL (n = 14)

72.8 (60–82.5)

NA

76 (59.4–83)

74.9 (66.9–79.5)

74.8 (63.7–80.8)

1.5 (0.2–5.9)

0.13

PaCO2, mm Hg

Group M (n = 7)

35.8 (34.8–39.8)

NA

37.7 (33.7–41.7)

37.9 (32.9–39.7)

38.0 (37.3–38.4

1.1 (–1.1 to 1.9)

0.39

>0.99

>0.99

0.66

Group L (n = 15)

35.9 (32.6–38.7)

NA

36.5 (34.2–37.7)

35.5 (29.5–39.9)

37.5 (33.8–39.5)

–0.5 (–2.5 to 1.0)

0.50

Group VL (n = 14)

35.1 (33.6–37.6)

NA

36.9 (33.7–38.1)

36.1 (33.7–38.2)

35.9 (34.5–37.4)

0.7 (–0.7 to 3.4)

0.38

P(A-a)O2, mm Hg

Group M (n = 7)

37.4 (27.6–45.6)

NA

28.7 (20.7–42.1)

38.1 (23.8–44.4)

38.4 (25.7–42.8)

–3.6 (–6.4 to 4.3)

0.46

0.41

>0.99

0.17

Group L (n = 15)

35.9 (29.2–42.9)

NA

28.9 (26.0–33.2)

31.1 (23.7–44.3)

36.5 (25.0–43.9)

–8.0 (–10.0 to –3.3)

0.003a

Group VL (n = 14)

33.1 (27.9–40.8)

NA

34.8 (23.2–41.8)

32.9 (30.2–41.4)

32.2 (30.4–40.4)

–2.3 (–4.6 to 0.5)

0.06

RR, 1/min

Group M (n = 7)

25.4 (17.2–31.3)

26.7 (22.3–40.3)

23.3 (22.0–37.3)

21.4 (20.0–33.8)

24.6 (18.2–36.3)

6.3 (–0.5–9.5)b

0.06

>0.99

>0.99

>0.99

Group L (n = 16)

24.2 (18.6–28.8)

26.9 (23.5–35.1)

25.7 (22.1–27.2)

24.3 (21.0–27.0)

22.3 (20.0–24.8)

4.3 (1.4–8.6)b

0.006a

Group VL (n = 14)

22.0 (20.2–24.7)

27.9 (25.2–31.0)

24.5 (19.8–26.7)

25.3 (22.0–28.8)

26.8 (20.0–28.9)

3.8 (3.0–11.1)b

0.002a

TV, ml

Group M (n = 7)

490 (390–650)

450 (260–630)

475 (400–690)

555 (450–640)

520 (500–590)

–30 (–150 to 10)b

0.06

>0.99

>0.99

>0.99

Group L (n = 16)

450 (405–520)

360 (260–480)

460 (440–550)

470 (380–550)

470 (380–540)

–105 (–195 to 40)b

0.09

Group VL (n = 14)

390 (350–710)

415 (340–580)

530 (455–605)

545 (455–645)

470 (390–570)

–20 (–90–20)b

0.18

Data presented as median and interquartile range (columns 3–7), subsequent column section (T1–T0 difference) includes data on within-group (M, L, and VL separately) difference between the measurements at T1 and T0 time points, as well as P values for within- and between-group comparisons: P (intra) = P for intragroup comparisons (within M, L, and VL group) at 2 time points (Wilcoxon test); P (inter) = P for intergroup comparison (between M, L, and VL groups) of the difference between T1 and T0 point (Kruskal-Wallis test with post-hoc Conover-Iman test)

a Statistical significance (P intra <⁠0.05) for multiple comparisons using the Bonferroni correction

b Comparison between TT and T0 point (as data at TT point were available)

Abbreviations: see Tables 2 and 3

Discussion

The study demonstrated that the increase in lung volumes associated with therapeutic thoracentesis was related to both the volume of the withdrawn pleural fluid and Pel. It must be admitted, however, that only a modest increase in lung volumes was found, with the maximal effect 24 hours after the procedure. Interestingly, when the increase in lung volumes was expressed as a percentage of withdrawn pleural fluid, a plateau of TLC was achieved 3 hours after thoracentesis, while FVC and FEV1 increased up to 24 hours after the procedure. Thoracentesis resulted in only a modest improvement in PaO2 and TLCO. Importantly, the most prominent increase in PaO2 was noted directly after the fluid withdrawal, with its deterioration in the next 24 hours. Pleural fluid withdrawal was associated with alleviation of dyspnea, but a significant effect was noted only in patients undergoing a very large volume thoracentesis. An interesting pattern of changes in ventilation was demonstrated, with increased RR and slightly reduced TV directly after thoracentesis and subsequent reverse changes (decrease in RR and increase in TV) during the next 24 hours. Also, an increased Ppl_ampl after the pleural fluid withdrawal may suggest an increased effort of the inspiratory muscles at the affected hemithorax directly after the termination of thoracentesis. To our knowledge, this was the first human study that included concurrent evaluation of the breathing pattern, PFT and ABG in the context of withdrawn fluid volume and Ppl changes during large volume thoracentesis. We therefore believe our results may shed more light on the complex relationships between the pleural fluid withdrawal and the function of the respiratory system, and may help to explain the contradictory findings of some earlier studies.

Considering the changes in lung volumes, the issue of the withdrawn fluid volume and the Pel should certainly be addressed. Larger volume of the withdrawn fluid and lower elastance may supposedly be associated with a larger increase in lung volumes.8 Indeed, Michaelides et al5 reported a difference between FVC increase in patients who underwent moderate vs large volume thoracentesis (mean FVC increase of 0.071 l [SD, 0.232] and 0.139 l [SD, 0.224] per 1 liter of pleural fluid drained, respectively). On the other hand, no significant correlation between lung volume increase and the volume of the fluid drained was found.2,24 This is in contrast with our data that clearly showed a significant positive correlation between the improvement in lung volumes and the volume of removed pleural fluid. The differences between the results of the studies by Cartaxao et al2 and by Spyratos et al24 can be related to methodological issues, mainly to smaller groups and different time intervals between therapeutic thoracentesis and PFT. The relationship between the changes in PFT and Ppl in spontaneously ventilated patients undergoing therapeutic thoracentesis was evaluated in only a few previous studies that included small numbers of patients.2,8,24,25 Our study showed that the increase in lung volumes negatively correlated with both Ppl drop during thoracentesis and Pel calculated at the termination of the procedure. The correlations were the most significant when lung volumes were compared 24 hours before and 24 hours after the procedure. We believe that even though the increase in lung volumes is related to both the withdrawn fluid volume and the change in Ppl (and their derivative, ie, Pel), the significant correlations between the amount of removed pleural fluid and the increase in lung volumes may be a prerequisite for the estimation of lung function improvement after symptom-limited thoracentesis, even when Ppl and Pel remain unknown. This is probably due to a close relationship between the total volume of fluid that can be drained and the mechanical properties of lung and pleura that determine Pel.

Further analysis of pulmonary function in patients who underwent moderate volume thoracentesis showed virtually no increase in TLC, functional residual capacity, FVC and FEV1 both 3 and 24 hours after the procedure. This was in clear contrast to a large and very large volume thoracentesis. Moreover, the patterns of pulmonary function changes in these subgroups were different, with a trend to small but constant increase in TLC (expressed as % of predicted) during 24 hours after moderate volume thoracentesis, and a steep TLC rise during the first 3 hours after a large or very large volume thoracentesis. These data emphasize the complex relationships between pulmonary volumes, the volume of the withdrawn pleural fluid, the mechanical properties of pleural space, and time from the procedure. It seems that time-based profiling of PFT changes after therapeutic thoracentesis that includes multiple data on pleural space characteristics is crucial to understand the pathophysiological changes associated with thoracentesis.

Although the therapeutic thoracentesis resulted in a significant increase in PaO2 and a decrease in alveolar to arterial PaO2 difference in the entire group, the improvement was modest and temporary. A small initial PaO2 increase was demonstrated in all subgroups (M, L, and VL), but significance was reached only in the L group. Somewhat different changes were noted for TLCO, as there was a slight (insignificant) decrease in the M group and a significant increase in both the L and VL groups. In fact, the most pronounced and constant TLCO increase was demonstrated in the patients from the VL group. The insignificant increase in PaO2 with the concurrent insignificant TLCO decrease in the M group is difficult to interpret. Bearing in mind numerous variables that impact blood oxygenation, the effect of fluid withdrawal on gas exchange is quite complex and is not easy to follow even in virtual patients.26

In terms of PaO2 changes, the results of our study are somewhat similar to those reported by Perpiñá et al14 However, the maximal improvement in PaO2 reported by these authors was delayed as compared with our results. This could be related to the fact that ABG were measured in only approximately a half (n = 16) and one-third (n = 12) of all patients after 20 minutes and 24 hours, respectively, and if so, the results may not fully reflect the change in PaO2 in the entire group. Other differences between the study by Perpiñá et al14 and ours that could have influenced the results include the larger amount of fluid drained in our study. In contrast, some other studies demonstrated no effect or even worsening of PaO2 after the pleural fluid removal. The withdrawal of mean fluid volume of 2.3 l (SD, 1.2) during a medical thoracoscopy was associated with a significant drop in PaO2 both after 24 and 48 hours.27 Agustí et al15 showed virtually no effect of thoracentesis on PaO2 measured 30 minutes after the procedure. However, their study included 9 patients only, the volume of removed fluid was relatively small (mean, 693 ml), and the baseline PaO2 was relatively high (mean, 82 mm Hg). Thus, we believe our study provides one of the most reliable data on the impact of large volume thoracentesis on PaO2.

At a glimpse, a significant increase in Ppl_ampl after a therapeutic thoracentesis might be interpreted as an increased work of breathing (WoB) performed by the respiratory muscles of the affected hemithorax. As Ppl_ampl correlates with the driving pressure for lung expansion, its increase together with relatively constant or even elevated V̇ might suggest an increased workload of the respiratory muscles. This would be in contrast with a common belief that the pleural fluid removal enables more effective work of the respiratory muscles and reduces respiratory effort. Although we could not find any data on the effect of therapeutic thoracentesis on the WoB, we are aware of a single small study in patients receiving mechanical ventilation that showed that fluid withdrawal led to a significant decrease in the work performed by the ventilator.16 The explanation of the potential contradiction between the increase in Ppl_ampl and a presumed reduction of WoB and dyspnea lies in the changes of the diaphragm shape and its upward movement after pleural fluid removal.12 These changes allow the hemidiaphragm to operate in a more effective position on its length-tension curve.27,28 Thus, the significant increase in Ppl_ampl may represent an improved efficiency of the respiratory work performed for a similar effort.16 A recent study showed that the pleural fluid removal in patients with expandable lung resulted in a significant increase of the velocity of the diaphragm contraction and diaphragm excursion.29 It should be underlined, however, that in fact, the ipsilateral Ppl_ampl is not a reliable measure of the entire WoB, as the affected lung only slightly participates in breathing. This refers in particular to the initial phase of thoracentesis when there is a large volume of the pleural fluid. We believe that future studies on the relationship between the change in WoB and the associated change in Ppl ampl, as well as other measures of respiratory pump effectiveness may help to better understand the diverse effects of thoracentesis on dyspnea caused by PE. It can be speculated that the lack of improvement in dyspnea after a moderate volume thoracentesis can be related not only to a limited volume of the drained fluid and high Pel but also to a concurrent significant increase in Ppl_ampl reflecting the respiratory muscle effort. The respective elevation of Ppl_ampl after a large and very large volume thoracentesis was smaller, and this could have contributed to the reduction of dyspnea in those patients.

We are aware of several limitations of our study. First, the number of patients was relatively low. Still, it was larger than in the vast majority of previous studies. Second, the smaller number of patients with high Pel, who underwent a moderate volume thoracentesis, as compared with 2 other groups could have influenced the results of the statistical analysis. Nonetheless, the relatively low percentage of these patients may reflect a real-world situation. A recent paper by Aguilera Garcia et al29 reported 17% of patients with unexpandable lung. Similarly, Lentz et al30 found a rapid fall in Ppl or Ppl decrease below –20 cm H2O in 6% and 15% of patients undergoing thoracentesis with pleural manometry, respectively. Third, the pathophysiological responses after therapeutic thoracentesis in patients with malignant effusion may not fully reflect the response in patients with other underlying diseases. It must be underlined, however, that in numerous other studies pleural malignancies were responsible for 50% or more of all PE cases.30,31 Fourth, the use of some measurement tools could affect the measurements. This includes a spirometer mouthpiece that slightly changed the breathing pattern with deepening of breathing reflected by a higher amplitude of the Ppl changes.

Conclusions

In conclusion, the volume of pleural fluid that can be removed during a symptom-limited therapeutic thoracentesis relatively well reflects the mechanical properties of the affected pleural cavity. Pleural fluid withdrawal results in a modest improvement in pulmonary function, transient increase in PaO2, and invariable increase in Ppl_ampl. The improvement in pulmonary function and ABG is closely related to the volume of fluid drained and Pel. The increase in Ppl_ampl probably represents a more efficient work of the respiratory muscles, not the increase in WoB. This relationship requires further studies.