Two-Stage Bronchoscopic Endobronchial Valve Treatment Can Lead to Progressive Lung Volume Reduction and May Decrease Pneumothorax Risk

Introduction

Based on the positive outcomes of five randomized controlled trials (reviewed in¹), bronchoscopic lung volume reduction (BLVR) using Zephyr^® endobronchial valves (EBV - Zephyr EBV, PulmonX Corp.) is now a treatment option for a subgroup of patients with severe emphysema with little to no collateral ventilation (CV).² According to a recent meta-analysis, the main (and potentially life-threatening) complication is pneumothorax, occurring in 23% of patients (14% to 39% in trials), with a relative risk of 6.32.¹ In addition, other complications include COPD exacerbation (9.3–64.0%), pneumonia (11.7%), valve migration and replacement rates (1.5–20.0%), and mortality (8%). Pneumothorax management can be long and challenging. The transient or permanent removal of at least one valve is often required, and 4 deaths directly imputable to EBVs have been reported in the 5 main randomized trials in which 295 patients were treated.³ The exact mechanism of this side effect remains unclear. Two pathophysiological mechanisms have been previously described. First, rupture of blebs or bullae in the ipsilateral untreated lobe might occur due to sudden shifting of lung volumes. The presence of blebs or bullae has however not been identified as a risk factor for pneumothorax.⁴ Second, pneumothorax ex vacuo, due to a sudden increase in negative intra pleural pressure striving the air from the surrounding extra pleural tissue.⁵ In 2012, confronted with a high rate of pneumothorax, Ninane et al published a study on 37 subjects⁶ who underwent partial occlusion of both upper lobes with spiration valves. The results underlined the safety of this procedure, but effectiveness was not proven. However, 24% of patients responded to an incomplete occlusion, and it is well demonstrated that a subset of patients does not have CV within segments⁷ confirming the possibility of achieving a first volume reduction with a subtotal treatment. Facing a similar threat with Zephyr valves, this study was based on a similar assumption: reduction of the intensity of the pathophysiologic mechanism suspected to create pneumothorax in order to lower its rate. Therefore, in order to reduce this complication while reaching EBV usual outcomes, we hypothesized that a two-step EBV implantation, during two distinct iterative procedures could lead to a more progressive target lobe volume reduction (TLVR) and thus ipsilateral lobe re-expansion, resulting in a significant decrease in the pneumothorax rate.

Methods

Patients

Between June 2018 and March 2020, 58 consecutive patients were included in this retrospective bi-center study carried out by Limoges (38 patients) and Toulouse (20 patients) University Hospitals. All patients met the inclusion criteria established by a BLVR expert panel:⁸ smoking cessation for at least 3 months, post bronchodilator forced expiratory volume in one-second (FEV1) less that 50% predicted value despite optimal medical management, significant static hyperinflation defined by a residual volume (RV) rated at least 175% predicted value, and a significant handicap characterized by a modified Medical research council (mMRC) score of 2 or more. All patients followed an optimal rehabilitation program prior to their inclusion. As initially recommended,⁸ lobar fissure integrity was determined by quantitative computed tomography with StratX^® (PulmonX Corp., Redwood City, CA), with optional Chartis^® assessment. CV was systematically measured using the Chartis^® system in all patients enrolled after May 2019.⁹

Ethical Considerations

The bicenter retrospective study was performed in accordance with the principles stated in the declaration of Helsinki and approved by the Limoges Ethics Committee (number 319–2019-85). Written informed consent explaining the procedure and allowing the use of its data was obtained from all participants.

Procedures and Follow-Up

Based on encouraging preliminary results,¹⁰ all patients were treated following the strategy conceived in Limoges University Hospital. During the first procedure, EBVs were placed in all but the most proximal segment or sub-segment (one segment or sub-segment left untreated in both RML and RUL lobe for RUL/RML treatments). Four weeks later, patients underwent a second procedure with EBV-implantation in the remaining segment(s) or sub-segment(s). For each procedure, patients were hospitalized for at least four days, including 24-hour strict bed rest immediately after EBV procedure. All patients had a complete evaluation before treatment, and 3 months after the second procedure. Forty-five (77%) had a complete evaluation between the 2 steps including chest multi-detector computed tomography (MDCT) and plethysmography. Lobar volumes were measured on each MDCT using Thoracic VCAR software (GE Healthcare). Patients without any significant TLVR at 3 months after the final procedure underwent a re-bronchoscopy in order to adjust the placement or replace EBVs in case of air leak.

Statistics

The primary endpoint was the pneumothorax rate after both procedures. Secondary endpoints included median changes in FEV₁, RV, 6-minute walking test distance (6MWT), mMRC score and BODE Index, and the rate of other complications. Responder rates were also calculated based on established minimal clinically important differences (MCID): 15% or more increase in FEV₁, 430mL or more decrease in RV,¹¹ an increase of at least 26m in 6MWT,⁵ a decrease of at least 1 point on the mMRC score,¹² a decrease of at least 1 point of the BODE Index.¹³ The MCID for target lobe volume reduction determined by high-resolution CT was 350mL^14,15 and 49% or more.¹⁶

Safety data were available for all 58 patients. Efficacy data were available for all patients at baseline and after the 2nd procedure, and for 45 patients after the 1st procedure. No imputations were performed for missing data. All collected data were analysed using Statview software (SAS Institute, Inc., Cary, NC) and R software. Quantitative results are expressed as median [range] or mean ± SD and qualitative results as n (%). Nominal variables were compared between groups using the chi-square or Fisher’s exact test, as appropriate. Means were compared with the non-parametric Mann–Whitney U-test for continuous variables. For all analyses, p <0.05 defined significance.

Results

A median number of 5 EBVs (range 3 to 9) per subject were implanted. The distribution of the treated lobes was 39% left upper lobe (LUL), 33% left lower lobe (LLL), 14% right upper and middle lobe (RUL and RML) and 8% right lower lobe (RLL). Forty-one patients had a homogeneous emphysema, including 5 panlobular emphysema due to alpha 1 antitrypsin deficiency, and 17 had a heterogeneous emphysema. The main results are reported in Table 1.

Table 1 Baseline Characteristics and Key Outcomes. Comparison with Randomized Controlled Trials Investigating EBVs

Primary Outcome

Only 4 pneumothoraxes (7%) occurred during the study. Two pneumothoraces occurred after the first procedure. One patient was treated for RUL and RML resulting in complete atelectasis of the RML and a 31% TLVR of the RUL. In the other case, EBV were implanted in the RUL resulting in its complete atelectasis. Two patients suffered from pneumothorax immediately after the second procedure. One of these patients was treated for the LUL leading to its complete atelectasis. The other one was treated for the RUL but had neither significant TLVR (298mL which corresponds to a reduction of 28% initial volume) nor changes in RV or FEV1. No late pneumothorax was seen. Three pneumothoraxes resolved within seven days with placement of a chest tube and without need for valve removal. One required one valve removal after 7 days, allowing for chest tube weaning. Of the 4 patients experiencing pneumothoraxes, 3 had heterogeneous (3/17, 17.6%) and 1 had homogeneous (1/41, 2.4%) emphysema. No statistical difference between patients with and without pneumothorax was found regarding main baseline characteristics (Table 2).

Table 2 Baseline Characteristics of the Pneumothorax Population

At the 3-month follow-up, other complications were pneumonia and severe COPD exacerbation (8.6% and 13.7% of patients, respectively). A re-bronchoscopy was required for 20.7% of patients.

Secondary Outcomes

MCID criteria (Table 1) were reached after both procedures for FEV₁ (+19.6 ± 25%), RV (−468 ± 960mL), 6MWT (30 ± 85m), BODE Index (−1.4 ± 1.8 point) and TLVR (50.6 ± 35.1%). Significant TLVR (MCID) was obtained in 74.1% of patients (43/58).

Complete intermediate evaluations were available for 45 patients (Table 3, Figure 1). Mean TLVR after the first procedure was measured at 511 ± 554mL (38.8 ± 36.9%). Twelve patients (28.5%) obtained a TLVR of 350mL or more after the first procedure (Figure 2). In this population, median change after the first procedure was 21.3 ± 24.4% for FEV1, 595 ± 774m for RV, 7.5 ± 15 for 6MWT and 0.9 ±0.6 for mMRC score. By choosing to define a response to treatment as a TLVR of 350mL or more, we identified 4 patterns of response (Figure 1): i) For “Great responders” both procedures resulted in significant TLVR (8/45, 17.8%) with a median FEV1 increase of 33.6 and a median TLVR rated at 1358mL; ii) “Early responders” (4/45, 8.9%) only reached a TLVR of more than 350mL after the first procedure (median FEV1 increase 35.2% and median TLVR 618mL), iii) In the “Late responders” group (22/45, 48.9%) there was no significant TLVR after the first procedure unlike the second procedure (median FEV1 increase +21.1%, median TLVR 851mL), iv) “Non-responders” did not have a significant TLVR after either procedure (14/45, 31.1%). Combining the two populations with an initial TLVR (“Great” and “Early” responders, n=12/45, 26.7%), median change for FEV₁ was +21.3 (± 24.4%) after the first procedure, and +34.1% after complete treatment, significantly greater than in the two groups with no initial TLVR (+15.5%, p=0.01). Of note, regarding the baseline characteristics of the population, no difference was found between the different patterns of response.

Table 3 Overall Population and Subgroup Analysis Based on Evolution of Target Lobe Volume>350mL at 1st Procedure

Figure 1 Response patterns to two step EBV treatment and main outcomes for each population. Abbreviations: T0, baseline; T1, first procedure; T2, second procedure.

Figure 2 Example of a “Great responder” patient with progressive lung volume reduction on CT-scan resulting in a progressive decrease in RV and a progressive improvement in FEV1. Abbreviations: FEV1, forced expiratory volume in 1st second; RV, residual volume; LUL volume, Left Upper Lobe Volume.

Discussion

This two-step procedure allows for a more progressive shifting of lung volumes in a subset of patients (approximately one-third) and may decrease pneumothorax rates without compromising treatment effectiveness. The baseline characteristics of our population are similar to those found in previously published trials^{3,12,14,17,18} especially for RV usually mentioned as a risk factor for EBV-induced pneumothorax. Data regarding the predictive factors for pneumothorax reported by Gompelmann et al⁴ were available for all patients treated in Limoges University Hospital (n=39). Results were similar with a high ratio of low attenuation volume of the target lobe to target lobe volume (LAV%) (44.7% vs 37%), a similar percentage of pleural adhesions (47% vs 41.3%) and a high ratio of volume of untreated ipsilateral lobe to volume of hemithorax (VUIL/VH) (39% vs 47%). Therefore, we did not find any unexpected factors when interpreting pneumothorax rates in this study.

When comparing to previous published studies using Zephyr^® EBV in a single procedure, only two reported a pneumothorax rate less than 10%.^17,19 The main explanation for the low pneumothorax rate in these studies may be the inclusion criteria regarding collateral ventilation. In particular, in the Believer study,¹⁷ StratX was not available, and Chartis not mandatory and inclusion was only based on MDCT assessment by 2 radiologists, likely resulting in the inclusion of CV+ patients, and subsequently in a low pneumothorax rate but also poor outcomes in terms of FEV1 (8.8 ± 15.5%), 6MWD (25.0 ± 43.6m) and RV (260 ± 240 mL). However, with two distinct procedures, this treatment algorithm seemed to slightly increase the rate of pneumonia (Believer 8%,¹⁷ Stelvio 5.8%,¹⁵ Liberate 0.8%,³ Transform 4.6%¹²). Of note, the rate of severe COPD exacerbations per patient was limited to 13.7% (8/58), similar to what was reported in the Liberate³ and Stelvio trials.¹⁵

Even if there is no single follow-up criterion to define a responder after EBV treatment, TLVR seem to be the most reproducible and, moreover, the most representative of Zephyr^® EBV objectives. According to Gompelmann et al,¹⁶ a TLVR between 49% and 54% should be used when interpreting a clinical relevance threshold. In our study, TLVR is comparable to other published data,^3,12 with a majority of patients reaching the MCID¹⁶ for this criteria. This translates into a similar magnitude of benefits compared to previous clinical trials in terms of FEV₁, RV and 6MWD (Table 1).^{3,12,14,17,18} Our interim analysis between the two procedures shows that a subgroup of patients (26.7%) reached a TLVR greater than or equal to 350mL, usually reported as a relevant threshold.^14,20,21 In this population, TLVR translates into clinically relevant improvement in FEV₁, RV and mMRC score despite incomplete occlusion. Stratifying patients on a 350 mL TLVR after the first procedure (Table 3, Figure 1) demonstrates that patients who reached pneumoreduction after the initial procedure, also have a statistically better outcome for FEV1 (34.1 vs 15.5%, p = 0.01) after complete lobar occlusion. Moreover, as EBV treatment is indicated for patients with little to no collateral ventilation,²² the difference between the “Late” and “Early” responders might be explained by the presence or absence of collateral ventilation between the subsegments treated and the subsegments which remain ventilated after the first procedure. Consequently, as intersegmental collateral ventilation is not always present,⁷ there is a strong need to design new tools for intersegmental collateral ventilation assessment (optimization of StratX software, Chartis balloon suited for segmental approach). This would allow physicians to be more predictive in obtaining this initial TLVR and identify patients who will benefit from a two-step treatment. Also, the possible intersegmental collateral ventilation may permit a more targeted treatment in patients with heterogeneity within the target lobe, sparing the most functional segments. However, we think that intersegmental collateral ventilation may not be the only physiological parameter to be taken in account when interpreting TLVR after each procedure as a subgroup of “Great Responders” seems to emerge. More research is needed to explain that response pattern.

In our study, 31% of patients reached complete atelectasis, similar to what was reported in previous publications.^17,23 The relation between lobar occlusion and clinical efficacy has been underlined in different studies.^24,25 However, complete atelectasis is also associated with pneumothorax.⁴ Therefore, we think that our results reinforce not only the idea that the high pneumothorax rate observed in clinical trials is related to a brutal and complete lobar occlusion but also, through our intermediate evaluation, that a progressive volume reduction is possible, and could decrease the risk of this complication.

This study has some limitations. It is retrospective and there is no control arm. Hence, no definitive conclusions can be drawn from this uncontrolled study, but these results are consistent with the ones previously presented by Limoges University Hospital on a smaller population.¹⁰ The intermediate analysis strongly supports the fact that a susbset of patients should be targeted by new tools. Moreover, there is no medico-economic evaluation of this strategy which could be a very important concern. On one hand, this two-step approach does not seem to save costs as all patients are admitted for two stays. On the other hand, it could be cost-effective by reducing the rate of a frequent and potentially life-threatening complication. Furthermore, if a dedicated Chartis could identify patients with no intersegmental ventilation, this approach could only be proposed to this subset of patients (one-third approximately), limiting additional procedures and hospitalizations.

Altogether, our results suggest that: i) a suboptimal treatment after the first procedure results an initial lung volume reduction in a subset of patients, a more progressive shift in lung volume (still leading to an efficient atelectasis), and thus a decreased pneumothorax rate; ii) Some patients reach some MCID despite infralobar treatment, likely due to intersegmental fissures and should be targeted by new tools.

In conclusion, this new approach using EBV could reduce the incidence of pneumothorax and does not seem to increase other complication rates. Clinical and physiological outcomes are similar to those reported in studies using the conventional single-step treatment. Additional research is needed to better characterize this subpopulation before treatment to avoid a two-procedure approach in “late responders”. Moreover, these results should be validated through a randomized prospective study comparing the two strategies in a larger population.