Paravertebral versus Pectoralis-II (Interpectoral and Pectoserratus) Nerve Blocks for Postoperative Analgesia after Nonmastectomy Breast Surgery: A Randomized, Controlled, Observer-masked Noninferiority Trial

This study was conducted within the ethical guidelines outlined in the Declaration of Helsinki and followed good clinical practice. The trial was prospectively registered at clinicaltrials.gov (NCT04742309; Principal Investigator: Brian Ilfeld, M.D., M.S.; initial posting: February 8, 2021). The protocol was approved by the local Institutional Review Board (University of California San Diego, La Jolla, California), and written, informed consent was obtained from all participants.

Adults (at least 18 yr of age) presenting for unilateral or bilateral breast surgery with at least moderate postoperative pain anticipated and a planned single-injection regional analgesic were screened for enrollment preoperatively. This included lumpectomy with axillary node biopsy or resection, breast reconstruction, breast reduction, mastopexy, and implant expander removal or placement. The exclusion criteria were simple lumpectomy (with no other procedure), planned regional analgesic with perineural catheter placement, morbid obesity as defined as a body mass index greater than 40 kg/m², renal insufficiency (preoperative creatinine greater than 1.5 mg/dl or estimated glomerular filtration rate less than 60), current chronic opioid use (daily use within the 2 weeks before surgery and duration of use greater than 4 weeks), history of opioid misuse, any comorbidity that results in moderate or severe functional limitation, inability to communicate with the investigators or hospital staff, pregnancy, incarceration, and allergy to study medications (ropivacaine). Patients undergoing mastectomy were excluded because they are offered a continuous paravertebral nerve block at our institution as standard care.

Preoperatively, participants received oral acetaminophen (975 mg). Participants were then placed in the sitting position with standard American Society of Anesthesiologists monitors and supplemental oxygen. All block procedures were performed by a regional anesthesia attending or fellow. Sedation was provided with intravenous midazolam and fentanyl, titrated to patient comfort during the block procedure. After confirmation of acceptable ultrasound visualization of both potential block sites, participants were randomized using a computer-generated list and opaque, sealed envelopes to one of two treatment groups stratified for unilateral versus bilateral surgery: (1) pectoralis-II or (2) paravertebral block.

A linear ultrasound transducer (13 to 6 MHz) was placed at the midaxillary line to identify the short-axis view of ribs 4 and 5 deep to the pectoralis-II anterior muscle. The 20-gauge Tuohy needle was inserted caudad to the probe after skin anesthesia with lidocaine. The needle was advanced to the tissue plane between the pectoralis major and minor muscles at the vicinity of the pectoral branch of the acromiothoracic artery using an in-plane approach in which 10 ml of ropivacaine (0.3%) with 1:400,000 epinephrine was deposited after negative aspiration. In a similar manner, 20 ml was subsequently deposited at the level of the third rib above the pectoralis-II anterior muscle with the intent of spreading injectate to the axilla.

A low-frequency curvilinear ultrasound transducer (6 to 2 MHz) was used to identify the T1 through T5 transverse processes and paravertebral spaces. A 20-gauge Tuohy needle was inserted into the appropriate plane or space under direct ultrasound guidance via an in-plane parasagittal approach after skin anesthesia with lidocaine. Ropivacaine (0.5%) with 1:400,000 of epinephrine (9 ml) was injected at each of two levels after negative aspiration (18 ml total). For surgical procedures involving the axilla, T2 and T4 were targeted, with T3 and T5 targeted for the remainder.

Both treatment groups received a total of 90 mg of ropivacaine. Blocks were considered successful if, within 30 min, the participant experienced decreased sensation to cold temperature over the level of the ipsilateral fourth thoracic dermatome at the level of anterior axillary line. For failed blocks, the procedure was repeated successfully, or the patient was excluded from further participation. For participants undergoing a bilateral surgical procedure, a block using the same protocol was administered on the contralateral side.

Intraoperatively, all participants received a general anesthetic using inhaled and intravenous anesthetic and oxygen. Intravenous fentanyl was administered for cardiovascular responsiveness to noxious stimuli at the discretion of the anesthesia provider. Intravenous ketorolac (0.5 mg/kg, up to 30 mg) was administered. No additional local anesthetic was injected intraoperatively. Participants were extubated, taken to the postanesthesia care unit, and received by a nurse masked to treatment group assignment.

A standard postanesthesia care unit opioid algorithm was used which involved: (1) intravenous fentanyl (25 μg) for numeric rating scale pain scores of either 4 or 5, (2) intravenous fentanyl (50 μg) for numeric rating scale of 5 or greater, (3) intravenous hydromorphone (0.5 mg) for numeric rating scale score of 5 or greater if fentanyl deemed ineffective, and (4) oxycodone (5 mg) for numeric rating scale of 4 to 6 if able to tolerate oral medications.

All participants were instructed to record the time at which they believed their block began to resolve and, for outpatients, the time when they consumed their first opioid. In addition to acetaminophen (975 mg) four times daily, admitted participants were provided oxycodone at a dose of 5 mg every 4 h as needed for numeric rating scale score of 4 to 6 and of 10 mg every 4 h as needed for numeric rating scale score of 7 to 10. Reasons for same-day hospital admission included intractable postoperative nausea or vomiting, severe pain uncontrolled with repeated intravenous opioids, postoperative respiratory depression requiring overnight monitoring, and surgical indications (e.g., bleeding, preference for overnight monitoring due to surgical reasons). Before discharge, all participants were instructed to take 975 to 1,000 mg of acetaminophen four times daily. They were discharged with a prescription for a synthetic opioid, oxycodone (5-mg tablets).

Operating room personnel, recovery room nurses, and surgeons were all masked to treatment group allocation. Although participants were not specifically informed of their treatment group, many doubtlessly deduced their group based on the different positioning and techniques used for the two interventions and therefore should not be considered masked to treatment group assignment. Data from the operating and recovery rooms were therefore collected by masked observers. Pain scores were recorded using the 11-point numeric rating scale (0 to 10, where 0 indicates no pain and 10 indicates the worst pain imaginable). Intraoperative and recovery room opioid consumption was analyzed as intravenous morphine equivalents by converting intravenous fentanyl, intravenous hydromorphone, and oral oxycodone to intravenous morphine equivalents.¹⁷  Because the procedures were performed on an outpatient basis, we chose to limit our masked pain and opioid assessments to the recovery room, which included the dual primary outcome measures.

The morning after surgery, all participants were contacted by telephone or in person (if hospitalized) to record lowest, average, highest, and current pain scores, number of awakenings due to pain, and nausea using a 0 to 10 Likert scale (where 0 indicates no nausea and 10 indicates vomiting) by an investigator who was not masked to treatment group allocation. For ambulatory patients, opioid requirements were recorded by the participants themselves, whereas inpatients had opioid requirements extracted from the electronic medical record. In addition, for inpatient participants, antiemetic use and nurse-recorded pain scores were extracted from the electronic medical record. Outpatients used oral oxycodone without any intravenous opioids, and we therefore did not convert postrecovery room opioids to intravenous morphine equivalents for the sake of clarity. We attempted to collect the times at which participants identified block resolution and their first oral opioid analgesic after recovery room discharge. However, the majority of participants could not recall this information with confidence, and we therefore ceased attempting to collect these data. In addition, possible block-related side effects and adverse events were recorded.

The primary analytic approach for this study was the two-sample Mann–Whitney test or chi-square test for two proportions, as appropriate. Group imbalance was assessed based on the absolute standardized difference.¹⁸  Specifically, standardized differences were calculated using Cohen’s d, whereby the difference in means or proportions was divided by the pooled SD estimates.¹⁹  Any key variables (age, sex, height, weight, and body mass index) with an absolute standardized difference greater than 1.96 × √(2/n) = 0.358, where n = 60 is the sample size per group, was noted and included in a sensitivity analysis with a generalized linear model (e.g., logistic regression for incidence rates or continuous ordinal regression for pain severity numeric rating scale to obtain an estimate of the treatment effect adjusted for the imbalanced covariate[s]). If key model assumptions were violated (i.e., homoscedasticity or Gaussian distribution for linear models), data transformations and/or alternative generalized linear models were applied, as appropriate. For visual analog scale (VAS) scores, we used a continuous ordinal model designed specifically for VAS scores.^20,21  The model uses a logit link and I-splines to model the “g(v) function,” which is a continuous analog of the intercepts required for the canonical cumulative logistic (proportional odds) model. An advantage of this continuous ordinal model is that it does not require categorizing or discretizing the raw responses to an ordinal variable. However, it does require that observed VAS on the 0 to 1 scale (v*) be rescaled (v) to avoid the boundaries: i.e., actual values of 0 (no pain) and 1 (greatest pain). This is done by the transformation v = v*(n − 1)/n + 0.5/n, where n = 119 is the total sample size so that, for example, 0 is mapped 0.0042 and 1 is mapped to 0.996. Although this transformation of boundary values preserves the order of raw data, like rank transformation underlying tests such as the two-sample Mann–Whitney test, it is possible that difference choices of transformation would lead to slightly different estimates, although it is very unlikely that different order-preserving boundary transformations would qualitatively change conclusions. Assumptions of the continuous ordinal regression were checked by examining quantile residual plots.²¹ 

We hypothesized that (1) analgesia would be noninferior in the recovery room with pectoralis-II compared with a paravertebral block as measured on the numeric rating scale (assessed up to 12 h total) and (2) opioid consumption would be noninferior in the operating and recovery rooms with a pectoralis-II block compared with a paravertebral block (cumulative operating and recovery room consumption, assessed up to 24 h).²²  To claim that pectoralis-II blocks are noninferior to paravertebral blocks, both hypotheses had to be at least noninferior.²³ 

We tested the noninferiority of pectoralis-II compared with paravertebral block using the 95% CI associated with the Wilcoxon–Mann–Whitney test based on the normal approximation with continuity correction. If the lower limit of the 95% CI for median “average” recovery room numeric rating scale was greater than −1.25 (based on paravertebral minus pectoralis-II), we would conclude noninferiority.²²  If there was group imbalance in any key characteristics as described above, these characteristics were included as covariates in a regression model. The same noninferiority margin (−1.25) was applied to the 95% CI for the covariate-adjusted group difference in mean pain derived from the adjusted model. The noninferiority of pectoralis-II blocks with regard to opioid consumption was similarly tested by comparing the limits of a 95% CI associated with the Wilcoxon–Mann–Whitney test with continuity correction to a predefined noninferiority margin of 2-mg intravenous morphine equivalents. Of note, testing for noninferiority by comparing the limits of conventional two-sided 95% CI to prespecified noninferiority margins does not preclude the possibility of concluding inferiority or superiority if the same CI excludes zero.²³  All participants were analyzed on an intention-to-treat basis. R version 4.3.2 (https://www.r-project.org/) was used for all analyses. P values, if reported, are two sided.²⁴ 

Power for the Wilcoxon–Mann–Whitney–derived noninferiority testing was based on 10,000 simulated trials. We simulated pain scores (within the recovery room) from a discrete distribution with median [interquartile range], 3 [2, 5].^25,26  Between the quartiles, the probability of each score was assumed constant. The distribution for each group was assumed to be the same (i.e., equivalence). The sample size of 50 per group provided 82% power to detect noninferiority in pain with a margin of 1.25. Similarly, opioid consumption (combined operating and recovery room amounts) was assumed to follow a truncated normal distribution with mean (SD) of 2.5 mg (2 mg) and minimum value of 0 mg. The sample size of 50 per group provided at least 95% power to detect noninferiority with margin 2 mg. Sample size was prospectively increased to 75 subjects per group to account for higher-than-expected variability in either outcome. Noninferiority in both pain and opioid consumption would be required to claim overall noninferiority; therefore, no adjustment in α was necessary to control type 1 error. All other P values should be considered nominal and are provided for descriptive purposes.

Between February 2021 and August 2023, a total of 119 participants were enrolled (fig. 1). Multiple surgeons withdrew support for the study as time progressed due to a perception of inadequate analgesia for a subset of patients, resulting in an inability to complete enrollment of the originally planned 150 participants. The 119 enrolled participants were randomly allocated to either the pectoralis-II (n = 60) or paravertebral (n = 59) treatment group after confirmation of adequate ultrasound visualization (table 1). All baseline characteristics were well balanced between the two treatment groups (absolute standardized difference less than or equal to 0.358) with the exception of age (absolute standardized difference of 0.727).

All interventions were performed per protocol with no block failures as defined in the methods. The time to perform each procedure (per side) for pectoralis-II blocks was a median [interquartile range] of 68 s [60, 120] versus 120 s [60, 180] for paravertebral blocks (P = 0.001). Although participants receiving the pectoralis-II reported less pain during block placement (fig. 2), they also received a greater amount of fentanyl: median [interquartile range] of 50 µg [50, 100] versus 50 µg [50, 50] (P = 0.007).

Pain experienced in the recovery room by pectoralis-II participants was not noninferior as originally hypothesized. To the contrary, this group reported higher median recovery room pain scores compared with their paravertebral counterparts: median [interquartile range], 3.3 [2.3, 4.8] versus 1.3 [0, 3.6] (95% CI, 0.5 to 2.6; P < 0.001; fig. 2). A total of 23 (39%) of paravertebral versus 11 (18%) pectoralis patients reported 0 pain. One patient in the paravertebral group reported a pain score of 10. Similarly, intravenous morphine equivalents were higher in the operating room and recovery room combined in the pectoralis-II group: 17.5 [12.5, 21.9] versus 10.0 mg [10, 20] (95% CI, 0.1 to 7.5; P = 0.004; fig. 3). Although the pectoralis-II participants were older, regression models controlling for age yielded similar conclusions. A continuous ordinal regression model adjusting for age found reduced odds of higher pain scores with paravertebral block (odds ratio, 0.24; 95% CI, 0.49 to 0.12; P < 0.001). Mean intravenous morphine equivalents were 4.7 mg higher for pectoralis-II (95% CI, 0.7 to 8.8; P = 0.023) based on a linear regression model adjusted for age.

Within the recovery room and after discharge, the lowest, average, and maximum pain scores were all higher in the pectoralis-II compared with the paravertebral group (fig. 2). Differences in opioid administration between the two treatment groups within the operating room did not reach statistical significance, but opioid use in the recovery room was increased 75% in the pectoralis-II block over the paravertebral cohort (fig. 3). After recovery room discharge, differences between the pectoralis-II and paravertebral groups did not reach statistical significance for oxycodone consumption (fig. 4), incidence of nausea (3 vs. 1 participants [5% vs. 2%], respectively; P = 0.619), or incidence of awakening due to pain (11 vs. 7 participants [18% vs. 12%], respectively; P = 0.444). In the entire postoperative study period, a higher percentage of the paravertebral group (36% [n = 21]) required no additional opioids compared with the pectoralis-II participants (17% [n = 10]; P = 0.019).

Three subjects from each treatment group were unexpectedly admitted postoperatively, none of which were block, pain, or pulmonary related. Rather, the reasons for admission were due to surgical indications and need for overnight monitoring. All six hospitalized subjects were discharged on postoperative day 1. No block-related adverse events were identified in either group.

This randomized, observer-masked study found that, after nonmastectomy breast surgery, paravertebral blocks provided superior analgesia with a concurrent decrease in opioid requirements compared with pectoralis-II blocks. These results are of consequence due to the frequent severe pain after breast surgery,¹⁴  the association between acute and persistent or chronic surgical pain,²⁷  individual and societal issues involving opioids,^28,29  and the large number of annual nonmastectomy breast procedures numbering in the millions.¹⁶ 

Similar to previous reports,¹³  it took less time to place the pectoralis-II fascial plane blocks (median of 87 vs. 120 s), although the 33-s savings between treatment groups is probably not clinically relevant in most situations. It was also less painful to receive the pectoralis-II blocks (average numeric rating scale 1.0 vs. 2.0; maximum numeric rating scale, 3.0 vs. 4.0); but again, the clinical relevance of these differences is questionable. These were the only benefits of the pectoralis-II block, whereas the benefits of the paravertebral block were identified after block administration.

Within the recovery room, participants with a paravertebral block experienced improved analgesia with lower least, average, and maximum pain scores that reached both statistical and clinical relevance (fig. 2). For example, the median numeric rating scale for the pectoralis-II participants was 3.3 versus 1.3 for the paravertebral group (P < 0.001). The minimum improvement in numeric rating scale deemed meaningful for individual patients with acute pain has been estimated to be 1.0 to 2.0 points.^30–34  Considering clinically relevant differences between group means tend to be smaller than for individual patients,^35–38  the between-group difference of 2.0 is undoubtably clinically meaningful.³⁸ 

Regarding opioid requirements, both groups received similar doses of fentanyl for block administration (median 50 µg) and intravenous morphine equivalents during surgery (median 10 mg; figs. 2 and 3). However, within the recovery room, participants with a paravertebral block required a median of 0 mg of additional morphine equivalents, suggesting that the 100 µg of fentanyl most received for intubation was more than they required for postoperative analgesia. In contrast, patients with a pectoralis-II block required an additional median 5.8 morphine equivalents within the recovery room, resulting in a 75% increased overall opioid requirement when the operating and recover room opioids were combined for one of the primary outcome measures (median of 17.5 vs. 10.0 mg). The benefits of the paravertebral block continued into the following day, with lower pain scores (fig. 2) and oxycodone requirements (fig. 4).

Our findings are contrary to the majority of studies in which little significant difference was identified between the two types of blocks.^8,13  We are unable to identify and therefore contrast our findings with similar investigations comparing pectoralis-II and paravertebral blocks for nonmastectomy surgery. Consequently, we are left with comparisons to studies involving mastectomy and propose two possible explanations for our disparate findings.

First, of the myriad of previously published studies, all but two used a single injection of local anesthetic for the paravertebral blocks. The two remaining studies administered injections at two³⁹  and three⁴⁰  different paravertebral levels. The latter—using three separate paravertebral injections—reported that the paravertebral group had lower pain scores (hours 8 to 24), a shorter time to first analgesic request (6 vs. 11 h), and a lower 24-h morphine consumption (12 vs. 20 mg).⁴⁰  There is compelling prospective evidence suggesting that the vertical spread of local anesthetic resulting from a single injection is greatly limited: a mean (SD) of 3.0 (1.2) dermatomes with a single bolus injection (0.26 ml/kg of local anesthetic plus 0.1 ml/kg of radiopaque dye) compared with 6.5 (2.0) dermatomes with four individual bolus injections (each with 25% of the volume) at four consecutive thoracic levels.⁴¹  Analgesia coverage is likely correlated with the number of thoracic levels injected. It may be that pectoralis-II blocks are comparable to single-injection paravertebral nerve blocks, but paravertebral provides superior analgesia with increasing thoracic injection levels. Reflecting this, authors of a recent meta-analysis finding little difference between pectoralis-II and paravertebral blocks noted that their “conclusions are not necessarily generalizable to paravertebral block techniques involving injection at multiple thoracic levels.”¹³ 

A second possible explanation for our contrary findings involves the differences in surgical procedures. Nearly all participants from previous trials comparing the two blocks underwent either radical or modified radical mastectomy, both of which include axillary lymph node dissection. As has been described previously,¹³  the pectoralis-II technique theoretically provides superior analgesia of the axilla “by blocking the long thoracic, thoracodorsal, and medial and lateral pectoral, which are spared by the paravertebral block.” It is therefore possible that paravertebral blocks provide superior analgesia for the breast itself, but inferior analgesia of the axilla. Indeed, as Hussain et al. noted, “paravertebral block is unique in its ability to provide surgical anesthesia, whereas this has not been demonstrated yet with Pectoralis-II,”^13,42,43  suggesting that the somatic block of the breast tissue may be more intense with the paravertebral approach. Because participants of the current study undergoing nonmastectomy breast surgery had either no axillary involvement or—at most—a single axillary node biopsy (vs. dissection), any superior analgesia of the breast by the paravertebral block might be identified without conflicting axillary pain.

We emphasize that the results of the current investigation do not suggest that paravertebral blocks should automatically be preferred over pectoralis-II for nonmastectomy breast surgery. The benefits of the latter have been demonstrated in this patient population,¹⁶  and the choice between the two blocks involve many factors, of which analgesia potency in the first 24 h is only a single variable. As with all medical interventions, the relative benefits must be weighed against the potential risks. Although in experienced hands the risk of ultrasound-guided paravertebral nerve blocks appears to be low,⁶  both theory and limited evidence suggest that pectoralis-II blocks are easier to learn and administer and have fewer inherent risks.^8,13,44  Therefore, analgesia and opioid sparing are just two of many factors that must be considered: different practitioners with dissimilar skill sets and varying priorities will doubtlessly—and justifiably—draw different conclusions regarding the optimal nerve block for their practice.

One limitation of our trial is that multiple different surgical procedures were included, which increased heterogeneity, although this is also a strength of the study in that it increases generalizability. A second limitation is that we exclusively used a double-injection technique, and so the proposition of a correlation between injection number and analgesia remains unresolved. Our results apply only to the specific block techniques, surgical procedures, and local anesthetic type, concentration and volume of the current study. Third, although operating and recovery room staff as well as surgeons were masked to treatment group assignment, investigators were not, increasing the risks of performance and detection bias (participants were not informed of their treatment group but should not be considered masked due to different administration techniques between the two blocks). However, the dual primary outcome measures were collected by masked observers within the operating and recovery rooms. Lastly, we had originally planned to enroll 150 participants but were forced to terminate enrollment short of that target due to (masked) surgeon perception that an unusually high percentage of patients were receiving inferior analgesia relative to historical memory. Fortunately, the study was adequately powered as enrolled with 119 total participants.

After nonmastectomy breast surgery, two-level paravertebral blocks provided superior analgesia and decreased opioid requirements compared with pectoralis-II blocks. This is a different finding that for the majority of studies in patients having mastectomy, in which little significant difference was identified between the two types of blocks. The reason for the discrepancy remains undetermined, and further research is warranted to help identify the relative strengths and weaknesses of each approach.

The authors appreciate the invaluable assistance of Baharin Abdullah, M.D. (Program Manager, Department of Anesthesiology, University of California San Diego, La Jolla, California), without whom this study would not have been possible.

Support was provided solely from institutional and/or departmental sources.

The university of Drs. Gabriel, Curran, Swisher, Sztain, Tsuda, Said, Alexander, Finneran, Abramson, Black, Abdullah, and Ilfeld received funding unrelated to this work from Epimed International (Dallas, Texas), SPR Therapeutics (Cleveland, Ohio), InfuTronix (Natick, Massachusetts), Avanos Medical (Irvine, California), Masimo (Irvine, California), and Varian Medical Systems (Palo Alto, California). Dr. Swisher also received funding unrelated to this work from Vertex Pharmaceuticals (Boston, Massachusetts). Dr. Donohue’s spouse is a full-time employee of Janssen Pharmaceuticals (Beerse, Belgium), and he has consulted for Roche (Basel, Switzerland). Mr. Cha received funding unrelated to this work from Masimo (Irvine, California). The other authors declare no competing interests.

Full protocol available at: bilfeld@health.ucsd.edu. Raw data available at: bilfeld@health.ucsd.edu.