Evaluation of Diaphragmatic Function after Interscalene Block with Liposomal Bupivacaine: A Randomized Controlled Trial

After approval from the University of Minnesota Institutional Review Board (IRB protocol 00003353) and registration with clinicaltrials.gov (NCT03636542, August 17, 2018, Habeck), we conducted a single-institution, randomized parallel-arm trial comparing interscalene blocks with a combination of liposomal bupivacaine plus plain bupivacaine (liposomal bupivacaine group) versus bupivacaine alone (bupivacaine group). All patients over the age of 18 yr scheduled for total shoulder arthroplasty with an expected postoperative length of stay greater than 24 h were eligible for inclusion in this study. Patients enrolled in this study could also be enrolled in another study (clinicaltrials.gov NCT03587636) evaluating the analgesic efficacy of liposomal bupivacaine for interscalene block after total shoulder arthroplasty. These studies, however, had separate protocols with separate inclusion criteria, but patients could be eligible for both. Patients were identified via the surgical schedule and approached in the preoperative assessment or surgical clinic. Final written consent was obtained in the preoperative area by study staff or investigators. As such, the 1:1 randomization was the same for both studies, and separate written informed consent was obtained for both. This trial was conducted in accordance with the original eligibility criteria and original protocol, which can be accessed by request. The randomization process was generated by a research coordinator using www.random.org. The sequence was stored in a separate file on box.com, a secure online platform. The randomization sequence was administered by a research assistant to the investigators after consent was obtained and just before block performance. Exclusion criteria defined for that analgesic study were thus used for the current project recruitment: patient refusal, pre-existing lung disease, coagulopathy, allergy to local anesthetics, daily use of opioids for more than 3 weeks before surgery, pregnancy, and inability to speak English. Enrollment ceased when the target sample size was obtained.

After consent and randomization, and before interscalene block, all patients had baseline diaphragmatic excursion and pulmonary function testing. Diaphragmatic excursion with shallow (sniff) and deep (sigh) breathing was evaluated using a low-frequency ultrasound transducer (GE Healthcare, LOGIQ e, USA) in M-mode with patients in the sitting position, as previously described by Cuvillon et al.⁸  Patients were scanned using a low intercostal or subcostal view using the liver or spleen as an acoustic window. Patients were then instructed to inhale deeply and exhale completely (sigh breath), and an image was obtained. M-mode was used to evaluate the excursion distance (figs. 1 to 3). These measurements were then used to determine level of diaphragmatic paralysis as defined by Renes et al., with no paralysis 0 to 25% (percent change from baseline), partial paralysis 25 to 75%, and complete paralysis 75% or greater.¹⁰  This was repeated with the patient taking a sniff breath, a quick short inhalation through the nose with a closed mouth. With patients still in the sitting position, a handheld spirometer (Vyaire Medical, CareFusion Micro I, USA) was used to measure forced vital capacity, FEV₁, FEV₁/forced vital capacity ratio, and flow rate in liters per second. Three separate measurements were taken and averaged for each patient. Identical breathing assessments were repeated postoperatively at two time points: after emergence from anesthesia in the postanesthesia care unit (PACU), and 24 h after block placement with the patient on the surgical ward. Initial (baseline) diaphragmatic excursion and spirometry readings were done before randomization. Patients were blinded to their randomization group. The measurements in PACU were done by a blinded individual when staffing permitted or the original anesthesiologist who performed the block. The 24-hr postblock assessment was done by a blinded anesthesiologist for 16 of the 22 patients. There was a lack of blinding in six patients (three in each group).

After baseline pulmonary testing and diaphragmatic motion measurements, all patients received an interscalene block by one of six anesthesiologists using a standardized two-injection technique. A high-frequency linear ultrasound transducer (FujiFilm Sonosite, Edge II, USA) was used to identify the C5 and C6 nerve roots in the interscalene groove. A 21-gauge blunt tip echogenic needle (Pajunk, SonoPlex, Germany) was inserted posterior to the probe and advanced in-plane until the tip of the needle was immediately below the C6 nerve root of the brachial plexus in the interscalene groove. Ten milliliters of local anesthetic was deposited. The needle was then repositioned so that the tip of the needle was at the posterior border of C5 nerve root in the interscalene groove, where another 10 ml of local anesthetic was deposited.

For the liposomal bupivacaine group, each 10 ml aliquot consisted of 5 ml liposomal bupivacaine and 5 ml 0.5% bupivacaine. Those in the bupivacaine group had 10 ml 0.5% bupivacaine injected at each site.

After the block, patients were taken to the operating room, where confirmation of block success was confirmed via pinprick sensory testing before general anesthesia was induced for surgery. All patients underwent general anesthesia with choice of native airway, laryngeal mask airway, or endotracheal tube at the discretion of the intraoperative anesthesiologist. Total intravenous anesthesia with propofol was employed for maintenance of anesthesia. After induction and before incision, patients received 4 mg ondansetron, 10 mg dexamethasone, and 0.25 mg/kg ketamine. Neuromuscular blockers were allowed, but only for securing the airway. If rocuronium was used, it was to be reversed with sugammadex before incision.

All patients received 975 mg acetaminophen every 6 h, alternating with 600 mg ibuprofen every 6 h scheduled. Patient also had as needed oral opioids (5 to 10 mg oxycodone or 2 to 4 mg oral hydromorphone if allergic) for breakthrough pain.

Once patients met PACU discharge criteria and before they went to the floor, roughly 3 to 4 h after the block had been placed, the same diaphragmatic ultrasound measurements and spirometry were performed. These measurements were repeated again 24 h after the block on the surgical ward, completing readings representing baseline, maximal phrenic nerve blockade, and recovery at 24 h.

Our primary endpoint was the percent change in diaphragmatic excursion between baseline and 24 h when performing sigh breath. The secondary endpoints were the percentage change from baseline to 24 h of diaphragmatic excursion when performing a sniff breath, FEV₁, forced vital capacity, FEV₁/forced vital capacity ratio, and respiratory flow rate both in the PACU and at 24 h.

There were no available previous data on these outcomes to conduct a power analysis. However, based on a previous study evaluating rate of blockade on diaphragmatic excursion with continuous catheters, we identified a target sample size based on a surrogate measure of a 5% rate of phrenic nerve blockade at 24 h in the bupivacaine group and a 66% phrenic nerve blockade rate for the liposomal bupivacaine group.⁶  Based on this surrogate measure, we estimated we would need nine patients in each group to achieve a significance level of 5% with 80% power for a two-sided two-sample test of proportions. Assuming a 20% dropout rate, we planned to enroll a total of 22 patients.

Continuous measures are summarized with median (interquartile range) and categorical measures as count (percent). Continuous endpoints were evaluated for normality using the Shapiro–Wilk test and through graphical evaluation, where almost all measures were identified as not being normally distributed. To accommodate for missing data and to account for the specific changes from baseline to a follow-up period (i.e., PACU or 24 h), we evaluated the specified outcomes and changes independently. For all continuous primary and secondary outcomes, to account for departures from normality in some endpoints, the Mann–Whitney U test was used to compare groups, with the difference in location parameters between groups (i.e., the median of the difference between a sample from each group) with 95% CI presented. The use of the Mann–Whitney U test included the comparison of the percent change measures, because although more efficient statistical approaches may exist to analyzing change data (e.g., a linear regression with the follow-up measure as the outcome and adjusting for its baseline value), the methods still impart a normality assumption that may not be met asymptotically given the smaller sample size. Chi-square tests were used to compare categorical measures between groups. All tests were two-tailed, and significance was at the P < 0.05 level. No imputation was used for missing data, with comparisons using complete cases. R v3.6.0 (Austria) was used for all analyses.

We enrolled 26 patients from August 2018 through September 2019. Two patients had a change in procedure and were discharged the same day and therefore could not be a part of the study, and two patients were lost to follow-up due to patients discharged before 24 h (fig. 4). We analyzed 22 patients in total, 10 patients (40% men) in the bupivacaine group and 12 patients (25% men) in the liposomal bupivacaine group. Demographics were similar between groups; however, the median body mass index of those in the liposomal bupivacaine group is near the 25th percentile of those in the bupivacaine group (table 1). Eight patients received paralysis. The liposomal bupivacaine group had four patients that received rocuronium (one of them received succinylcholine followed by rocuronium). The bupivacaine group had four patients as well, but only one received succinylcholine followed by rocuronium, and the other three had only succinylcholine. All patients who received rocuronium were reversed with sugammadex.

Diaphragmatic excursion is summarized at baseline, PACU, and 24 h in table 2. There was no significant difference in diaphragm excursion with sigh at 24 h between the liposomal bupivacaine group and bupivacaine group (mean excursion with sigh 2.5 cm SD [1.3] vs. mean excursion with sigh 3.6 cm SD [1.6], respectively; P = 0.112). At 24 h, the liposomal bupivacaine group had significantly greater percent decrease in diaphragmatic excursion during sigh breath compared to the bupivacaine group (median decrease of 24% vs. increase of 9%, respectively; P = 0.007; fig. 5). Overall, five patients in the liposomal bupivacaine group had a greater than 25% reduction in diaphragmatic excursion with sigh breath compared to none in the bupivacaine group at 24 h. There was no difference between the two groups in percent change of diaphragmatic excursion from baseline during a sniff test at 24 h (median increase for the liposomal bupivacaine group of 8% vs. increase of 7% for bupivacaine group; P = 0.734). There were two patients in the liposomal bupivacaine group and one patient in the bupivacaine group who had greater than 25% reduction in diaphragmatic excursion after a sniff breath.

In the PACU, both groups experienced a decrease in diaphragmatic excursion during sigh and sniff breaths. With sigh breath, the liposomal bupivacaine group (median decrease of –56%) show a greater reduction than the bupivacaine group (median decrease of –39%) in diaphragmatic excursion at this time point; however, both showed a greater than 25% reduction (P = 0.034, fig. 5). The sniff test noted similar median decreases (–25% and –21% for liposomal bupivacaine and bupivacaine, respectively; P = 0.351; fig. 5).

Spirometry values at baseline, PACU, and 24 h are represented in table 2. In the PACU, all spirometry values were decreased in both groups (fig. 6). The liposomal bupivacaine group had a significantly greater percent decrease in respiratory flow rate compared with the bupivacaine group (–46% vs. –29%, respectively; P = 0.003; fig. 6), while other results were similar. FEV₁ and forced vital capacity were still decreased from baseline at 24 h in both groups. The liposomal bupivacaine group showed a significantly greater percent reduction in FEV₁ and forced vital capacity compared with the bupivacaine group at 24 h (median decrease of 22% vs. 2%, P = 0.006; and median decrease of 19% vs. 1%, P = 0.049, respectively; fig. 6). Percent change from baseline to PACU and 24 h postblock can be seen in table 3 for both diaphragmatic excursion and spirometry.

No major adverse events were noted in the study. No patients in either group developed subjective dyspnea or oxygen desaturations.

Pain scores were assessed via a numerical rating scale 0 to 10 and evaluated at 24 h. The results are summarized in table 3. The median pain score 24 h postoperatively in the liposomal bupivacaine group was 2, whereas the median pain score at 24 h postoperatively was 4 in the bupivacaine group.

This study reveals that the combination of liposomal bupivacaine and bupivacaine in an interscalene block may lead to greater reductions in percent change of diaphragm excursion and pulmonary function tests compared to bupivacaine alone 24 h after block performance. However, whether this effect is clinically meaningful is unclear. Also, without inclusion of another group of patients undergoing general anesthetic alone without an interscalene block, it is difficult to tell whether the observed reduction in diaphragmatic excursion and pulmonary function test parameters is related to the effect of general anesthesia, interscalene block, or a combined general anesthesia/interscalene block effect. Previous studies have suggested that after an interscalene block, a decrease in diaphragm function of 0 to 25% is considered within normal, a decrease of 25 to 75% is partial paralysis, and a decrease of 75% or greater defines complete paralysis.^13–16  Using this definition, both groups returned to normal function at 24 h. On an individual level, though, five patients in the liposomal bupivacaine group still exhibited partial paralysis with sigh breath at 24 h compared to none in the bupivacaine group. Additionally, two patients in the liposomal bupivacaine group and one in the bupivacaine group exhibited partial paralysis with sniff breath at 24 h. No patients in either group qualified as having complete paralysis at 24 h in either a sigh breath or sniff breath. There was no significant difference in pain scores between the two groups; however, our study was not designed to observe a difference in pain scores at 24 h and thus likely is underpowered for that particular endpoint.

Our results are similar to previous studies on single-injection interscalene blockade with plain bupivacaine. There is an initial decrease in diaphragm function and spirometry measures followed by improvement over the next 24 h.^7,8,10,17  Our results show that the addition of liposomal bupivacaine leads to a non–clinically significant prolonged effect on diaphragm function in a healthy patient. The reduction in pulmonary function in PACU was seen in both groups with the liposomal bupivacaine group having a greater reduction in sigh breath. This greater reduction in PACU was unexpected. Before this study, we would have assumed that a concentration of 0.5% bupivacaine used in each block would lead to a near complete phrenic nerve blockade and would be unaffected or possibly lessened by the addition of liposomal bupivacaine over the use of more 0.5% bupivacaine. While this is an interesting finding, further studies are needed to determine if this is a consistent finding. Since the liposomal bupivacaine had a greater reduction in PACU, it would require a greater recovery to return to baseline. This could explain why there was a significant difference in the sigh breath, when the sniff breath, which has an equal reduction in PACU, showed no difference 24 h postblock. Other possibilities for a difference in sigh breath and not a sniff breath include the design of the study. This study was powered to discover a significance in a sigh breath. Since the measurement of a sigh breath is much larger comparatively, it might have missed a difference in the smaller sniff measurement. Another possibility is that phrenic nerve blockade affects a sigh breath more than a sniff breath. During a sniff breath, the velocity of the muscle contraction is a surrogate for muscle strength and has been used to evaluate diaphragm strength.¹⁸  It could be that an interscalene block effect on the diaphragm is more on the overall function seen during a sigh breath and that muscle velocity, seen in a sniff breath, is preserved.

While not directly compared in our study, continuous interscalene catheters have similar effects on the phrenic nerve. Cuvillon et al. showed that 33% of patients receiving continuous interscalene block had complete diaphragmatic recovery (as measured using ultrasound evaluation for a sigh breath) at 24 h.⁸  Auyong et al. showed that patients with continuous interscalene block maintain 62% of baseline vital capacity at 24 h.⁹  As such, a future study evaluating liposomal bupivacaine versus continuous catheter effects on the diaphragm would be necessary to directly compare them.

Before this study, liposomal bupivacaine’s effect on the phrenic nerve was unknown. The data from this study show that liposomal bupivacaine does have a prolonged effect on the phrenic nerve when compared to bupivacaine alone, but that this effect is still within the accepted range of normal diaphragmatic function after an interscalene block. However, this effect needs to be carefully considered in patients with impaired pulmonary function, as that was not studied, and any decrease in diaphragmatic function may prove detrimental to that patient population. At 24 h post–interscalene block, the diaphragmatic excursion, forced vital capacity, and FEV₁ were back to baseline for the bupivacaine group but lower than baseline in the liposomal bupivacaine group. However, both groups had decreased diaphragmatic excursion, forced vital capacity, and FEV₁ in the PACU, and thus we would expect that the phrenic nerve was initially blocked in both groups. As such, in patients with compromised lung function, one should consider the risks of phrenic nerve blockade from an interscalene block with either agent and consider alternative pain control measures such as a suprascapular nerve block. Since this study only enrolled patients with normal lung function, the question remains if a patient population exists where this lag time in diaphragmatic recovery with liposomal bupivacaine leads to clinical consequences.

Our study has some limitations. First, we based our power calculation on an assumption that liposomal bupivacaine would have a similar diaphragmatic impact at 24 h as a continuous interscalene catheter. Our results show that the impact was in fact much less, which could have impacted our power calculation. Second, our diaphragm measurements were performed by an anesthesiologist who was not always blinded to group allocation; however, this was primarily the PACU assessment, and the examination at 24 h postblock was usually done by a different, blinded anesthesiologist. Despite that, this could introduce a bias in our results. In addition, the diaphragmatic excursion measurement may be assessor dependent, and spirometry measurement is patient dependent. These limitations, however, led to findings that were consistent in spirometry and diaphragmatic excursion testing.

In conclusion, the addition of liposomal bupivacaine to bupivacaine in an interscalene block results in no clinically significant difference in diaphragmatic excursion with sigh at 24 h. However, it does result in statically significantly greater reductions in percent change of diaphragm excursion and spirometry measures (FEV₁ and forced vital capacity) 24 h after block placement, as compared with bupivacaine alone. However, the degree of reduction did not result in any clinical evidence of dyspnea or oxygen desaturation; thus, these findings, while interesting and new, may have no clinical relevance.

Support was provided solely from the Anesthesiology Department at the University of Minnesota, Minneapolis, Minnesota.

Dr. Hutchins was a speaker, was a consultant, and has received research funds from Pacira Pharmaceuticals (Parsippany, New Jersey). He is a consultant and owns stock with Insitu Biologics (Woodbury, Minnesota); is a consultant and speaker for Acel RX (Hayward, California); is a consultant for Worrell (Minneapolis, Minnesota); is a consultant for Johnson & Johnson (New Brunswick, New Jersey); and is a speaker for Avanos (Alpharetta, Georgia). Dr. Berg was a consultant for Avanos and is a consultant for Pacira Pharmaceuticals. Dr. Harrison has grant support from Zimmer Biomet (Warsaw, Indiana) and is a consultant for Arthrex, Inc. (Naples, Florida). Dr. Braman has grant support from Zimmer Biomet and has a board membership with the American Orthopaedic Association (Rosemont, Illinois). The other authors declare no competing interests. No companies were involved with any aspect of this manuscript.

Full protocol available at: bergx831@umn.edu. Raw data available at: bergx831@umn.edu.