Inspace Blog Post Follow up: Review of the papers
This is an update to a previous blog post I wrote about the Inspace balloon. Since the original post, the full results of the FDA pivotal trial have been published in the peer reviewed literature in a prestigious orthopedic journal: The Journal of Bone and Joint Surgery (JBJS).
Interestingly, at the same time this study was published, another randomized controlled trial (RCT) on the Inspace balloon was published in another highly respected journal: The Lancet.
The Lancet study was performed in England and the study was called START: REACTS
The fact that two high caliber, scientifically rigorous studies on the balloon came out at the same time is remarkable. It is rare to have Level 1 evidence for most orthopedic interventions. It speaks to the level of interest in this procedure that two such studies came out at the same time. It also gives a comparative sample of evidence to probe and hopefully draw some conclusions.
This post (posts?) will examine these articles in some detail, and I will leave you with my conclusions and where I stand on the balloon as a treatment option for patients with rotator cuff tears that are not repairable. To do that we are going to dive into the weeds a bit to analyze the studies. For brevity’s sake I will refer to the JBJS Inspace FDA pivotal trial as “the FDA trial” in the remainder of this post. We will examine the FDA trial first.
I gave several of my critiques of this study in my last post. I was a principal investigator in this trial, so I have first-hand knowledge of how the study was conducted. My conflicts of interest are as follows: our institution, the University at Buffalo did receive monetary support to conduct the trial while it was ongoing. But I am not a consultant, nor do I currently receive any reimbursement from Stryker the distributor of the balloon device. The views expressed in these posts are mine alone.
To recap, the inclusion criteria define an eligible patient as one who had a large irrepairable posterior/ superior rotator cuff tear (2 or more tendons) and no subscapularis tear, who had failed non operative treatment and had preserved passive range of motion of the shoulder (see page 1251 of the referenced article for full criteria). There’s a lot of information in that one sentence alone so let’s break it down.
The subscapularis is the large rotator cuff muscle in the front of the shoulder. Some believe that the balloon device is only appropriate for patients who have an intact subscapularis which is why the study was designed this way. In contrast the START: REACTS trial was designed to allow patients with partial subscapularis tears to participate. The authors of the START: REACTS trial argued in a rebuttal to critiques of their trial that it is not known whether the Inspace balloon device is only appropriate for patients with an intact subscapularis since this population had not yet been studied adequately.
Passive range of motion is the maximum amount of shoulder motion you can achieve in any given plane of motion, with help if needed. Active range of motion is the range of motion you can achieve with your shoulder under your own power. Pseudo paralysis is a condition in which you cannot lift your arm because of rotator cuff dysfunction. Pseudo paralysis can look like paralysis however it is not a true nerve issue but a muscle issue. Shoulder surgeons currently have not come to a consensus regarding a definitive amount of flexion that would delineate a patient from having pseudo paralysis or not. The START: REACTS trial defined pseudo paralysis as 20 degrees or less of active forward elevation which in my opinion is a reasonable definition. The FDA study design did not officially exclude patients with “pseudo paralysis.”
It is my opinion that the FDA Inspace trial was a rigorously performed trial that was executed professionally and with great attention to detail in all phases. If we examine the published data, the main conclusions from the pre-published data still stand: patients see statistically equivalent functional improvement with either the balloon or the debridement/ partial rotator cuff repair intervention. There was a slight advantage in forward flexion range of motion for the balloon patients after two years.
My main critiques also still stand. While the execution of the study was top notch, I believe that the study design was flawed. The study’s control group was not a true control. Recall that the FDA study compared 2 patient groups: 1) the group of patients that received a shoulder arthroscopy with debridement and the balloon and a group that received shoulder arthroscopy, debridement, partial rotator cuff repair and no balloon.
The reason this design is problematic is that it changes 2 variables: the balloon in the test group and a rotator cuff repair in the control group. In a well-designed RCT you want to isolate just one variable, in this case the balloon, if you are going to draw a causal inference about an intervention. This design leaves open the possibility that the partial cuff repair intervention might causally impact some of the resultant findings between groups. For this reason, the study was not constructed in such a way that you can draw conclusions on the balloon intervention alone.
For example, we notice that in the FDA trial the patients that received the balloon had higher forward elevation after 2 years of follow up. But we cannot definitively say that was caused by the balloon. Since the control group had a partial rotator cuff repair, we must leave open the possibility that the partial repair may have impaired patients’ motion in the control group. Perhaps the partial repair overtightened the control group patients instead of the balloon causing the relative improvement in motion. Alas, we cannot draw a good conclusion from this study design.
This is where the START: REACTS trial in my opinion gets it right. The authors of the START: REACTS trial designed their trial in such a way that they only changed one variable between the 2 study groups. The only difference between their test group and their control group was the balloon: the test group received the balloon, and the control group did not. Statistically speaking this is more rigorous design and can lead one to draw better causal inferences.
The START: REACTS trial’s inclusion criteria were close to the FDA trial’s criteria but were not the same. The START: REACTS trial did require subjects to have an unrepairable superior posterior rotator cuff tear but unlike the FDA trial, did allow inclusion of patients who had subscapularis tears of less than 1 cm.
The START:REACTS trial utilized slightly different outcome metrics and length of follow up as well. Whereas the FDA trial’s primary outcome measure was the American Shoulder and Elbow Surgeons (ASES) score, the START : REACTS trial’s primary outcome measure was the Oxford Shoulder Score (OSS) and like the Inspace balloon trial used Constant Score as a secondary outcome. Both ASES score and OSS scores are well validated patient reported outcome scores.
The authors of the START: REACTS trial originally powered the study to show a difference of Constant score of 10 (STD 20) and OSS of 6 (STD 12). The trial authors faced unique in-person follow up challenges because they started recruiting during the COVID pandemic. The authors then changed the original primary outcome measure from the Constant score to the OSS since the OSS does not require in person assessment. The authors stated in their paper that the power analysis did not require adjustment when this change was made and they made the changes during the recruitment phase prior to data analysis with permission of the oversite committee. Additionally the START:REACTS trial used a 12 month follow up instead of 24 month follow up as required by FDA citing historical data that showed that 12 month follow up in rotator cuff tear trials correlates well with 24 month follow up.
Secondary outcomes for the Inspace balloon trial were WOMAC score, Constant score, active shoulder range of motion and VAS after 24 month follow up. Whereas the START: REACTS trial used the Constant Score collected where possible, the range of pain free flexion and abduction, the Western Ontario Rotator Cuff (WORC) index (scored 0–100), EuroQol EQ-5D-5L, change in symptoms, Participant Global Impression of Change, resource use, and adverse events as secondary outcome measures).
In my opinion it does appear that he authors of the START: REACTS trial did a thorough amount of preparation for the study design. The supplement pre study design paper was 90 pages in length and outlined the design rationale and parameters in detail. One could argue that changing the primary outcome metric (substituting OSS for Constant Score) was unusual. However, the study was powered in such a way that this did not require any adjustments and the authors were adapting to the medical climate influenced by the worldwide pandemic. Since OSS and ASES are both well accepted and validated outcome measures that do not require in person measurements, it seems reasonable to assume that both primary outcome measures have the potential to provide sound feedback both the FDA trial and the START: REACTS trial. The results are where things get interesting. (Continued in next post…)