Clinical Faceoff: Anatomic Versus Reverse Shoulder Arthroplasty for the Treatment of Glenohumeral Osteoarthritis

While anatomic total shoulder arthroplasty (TSA) remains the mainstay of surgical treatment for primary glenohumeral osteoarthritis (OA), reverse total shoulder arthroplasty (RTSA) is seeing wider use because it may help offset some of the known complications associated with TSA, including late rotator cuff insufficiency and glenoid loosening. However, RTSA has its own unique risks, including acromial stress fractures and limited postoperative ROM in some patients. The severity of the tradeoffs associated with each approach raises an important question––one that now comes up at every shoulder meeting—what is the best type of arthroplasty for the treatment of primary glenohumeral OA? To gain better insight into this controversial topic, I am joined by two of my mentors and renowned experts on shoulder arthroplasty: Grant E. Garrigues MD from Rush University Medical Center in Chicago, IL, USA and Andrew Jawa MD from New England Baptist Hospital in Boston, MA, USA. They are both fellowship-trained in shoulder and elbow surgery, perform a high volume of shoulder arthroplasties, and have a passion for research and surgical education. Despite these commonalities, they have a different stance on this topic, which should make for an informed and edifying debate. Mariano E. Menendez MD:How has your treatment algorithm for end-stage glenohumeral OA changed over the last decade? Andrew Jawa MD: A decade ago, the dogma drilled into me by my mentors was to perform anatomic TSA whenever possible. This procedure was thought to be synonymous with “better” outcomes, even though that meant accepting more glenoid retroversion or thinner rotator cuffs. It’s important to note that outcomes then were not patient-reported and didn’t take advantage of validated outcome tools. While the early patient-reported outcomes of RTSA were good, there were complications with this relatively new surgery. Back then, the clinical consensus was that RTSA should be strictly limited to the most extreme indications and in patients older than 70 years of age. This notion began to change in 2013 when Mizuno and colleagues [11] published the sentinel paper demonstrating excellent ROM and Constant scores after RTSA in patients with glenohumeral OA and biconcave glenoids. This study opened the door for surgeons to expand the indications of RTSA from rotator cuff arthropathy to glenohumeral OA in older patients (> 70 years) and more-severe glenoid pathology in younger patients. As RTSA saw wider use, technology and techniques around the procedure advanced, resulting in improved patient-reported outcomes backed by validated outcome tools and fewer complications. Further, this paradigm shift was producing surprising results. Investigations comparing the results of RTSA for glenohumeral OA and rotator cuff arthropathy found that RTSA for glenohumeral OA was associated with better ROM and patient-reported outcomes [14]. These findings have been further amplified by Australian Registry data, which showed lower revision risk in women undergoing RTSA for glenohumeral OA compared to anatomic TSA [8]. As of this writing, I am seeing such excellent mid-term patient-reported outcomes in my own patients with RTSA for glenohumeral OA that I have expanded my indications for RTSA to 65 or even 60 years of age for patients without major glenoid pathology. But everything has to be pristine for me to do an anatomic TSA. Frankly, even in these “perfect” conditions, I believe all anatomic TSAs will require revision in the long run from either rotator cuff tearing or glenoid loosening. Overall, RTSA is my implant of choice for glenohumeral OA because of the decreased likelihood of requiring future revision due to the above reasons. The two primary factors attributable to RTSA’s longevity are the prosthesis’ ability to generate ingrowth on both sides of the joint and its functional independence of the rotator cuff. For younger patients (< 60 years), I still preferentially choose anatomic TSA in glenohumeral OA because there is an assumption that any arthroplasty will not last the rest of their life, and the revision of an anatomic TSA is generally (but not always) easier from a technical point of view than that of a RTSA. Grant E. Garrigues MD: I will concede that the clear superiority of the anatomic TSA that we preached about for many decades is no longer the case. I will further agree that the results of RTSA have improved and the frequency with which patients experience complications has lowered from the initial 50% to something now much more acceptable. What I will not acquiesce to is the conjecture that the RTSA has improved implant longevity and clinical outcomes compared to anatomic TSA for glenohumeral OA. This is not a settled question. Some studies have suggested superiority of one procedure, some the other, and still others have found them not to differ in this respect. Large registry studies, including a recent one in this very journal from the Australian registry, show no difference in long-term revision rate between an anatomic TSA with all-polyethylene glenoid and reverse TSA except in the short term—where reverse TSA had a higher revision rate in the first 3 months [8]. Multiple other large registry studies have driven this point home—showing a statistically equivalent revision rate for an anatomic TSA [6, 12, 18] versus RTSA [1, 18] with an equivalent or higher rate of complications in RTSA [6]. One recent systematic review and meta-analysis sums up the current state quite well, noting that: (1) better ROM is achieved after anatomic TSA than RTSA, (2) there is no difference in revision rates, and (3) the risk of complications is about the same—just different types, with glenoid loosening in anatomic TSA and scapular notching, acromial stress fractures, and instability in RTSA [10]. While one can find papers on either side of this line, I think this systematic review captures “the mood of the room” at this time. Regarding the paper by Mizuno et al. [11], while I agree that it was an impactful work by some of the biggest proponents of the RTSA, I will say that a retrospective study of 27 patients with a mean age of 74 years (range 66 to 82 years) without a comparison group is not enough to convince me to turn away from the gold standard operation for glenohumeral OA. Patients with posterior glenoid erosion are more likely to be younger, male, and quite active, and the anatomic arthroplasty with augmented glenoid components has shown excellent results [7]. Last, we forget that in the study by Mizuno and colleagues [11], complications occurred in 15%, including glenoid component loosening and neurologic changes. Furthermore, scapular notching was present in 37% of patients. Even in the hands of the absolute masters of RTSA, this is not a perfect operation. Dr. Menendez:What patient and anatomic factors influence your treatment recommendations when deciding between performing anatomic TSA or RTSA? Dr. Garrigues: The anatomic TSA with an all-polyethylene glenoid component is the time-tested arthroplasty for glenohumeral OA with its low risk of complications, equivalent or superior implant longevity, and greater ROM [1, 6, 8, 12, 18], particularly in internal rotation. The fear of postoperative rotator cuff tears seems to be overestimated. Excellent results have been achieved in older patients [9]. The devastating complication of acromial stress fracture—one of the most common complications of RTSA—is nearly unheard of after anatomic TSA [10, 15-17]. However, the RTSA has certain advantages. In general, the semiconstrained RTSA bearing with ingrowth surfaces on the humeral and glenoid sides can be used to treat issues of glenoid erosion or rotator cuff dysfunction. For example, static posterior instability such as the Walch B2 glenoid can be treated with RTSA with good predictability in older patients. Similarly, patients with severe glenoid bone loss who have more than 15° of acquired glenoid retroversion are not well served with traditional anatomic TSA components––although augmented anatomic glenoid components are an excellent option. Patients with rotator cuff tears other than small tears isolated to the supraspinatus are better served with an RTSA because larger tears can have dynamic superior instability and result in early glenoid loosening because of the “rocking horse” phenomenon. Lastly, patients with severe and long-standing shoulder stiffness (< 90° of active forward elevation in a patient > 65 years old) are probably better served with RTSA. In short, unless a patient has one of the contraindications above, patients with primary glenohumeral OA are typically best treated with an anatomic TSA. This approach leads to a low likelihood of complications, maximizes postoperative function, and has stood the test of time. Dr. Jawa: Let’s take these points one by one. The anatomic TSA had historically been the gold standard procedure for glenohumeral OA, but we are having this conversation because this notion is beginning to change. The risk of complications is not lower for anatomic TSA compared with RTSA. Parada et al. [13] recently found in more than 6000 patients that the risk of complications and revision were 10.7% and 5.6%, respectively, for anatomic TSA, compared to 8.9% and 2.5% for RTSA, respectively. Importantly, these RTSAs were performed for indications expanding beyond OA, and therefore, the true complication and revision rates would be lower if isolated to just patients with glenohumeral OA. I am unclear as to what data Dr. Garrigues is referring to when discussing longevity, because almost all the main papers evaluating this topic have major methodologic issues. These papers either report a low follow-up percentage or lack one altogether [1, 6, 18]. It is nearly impossible to glean any useful conclusions from these retrospective low-quality data. Alternatively, because the RTSA was developed in an era of better scientific methodology, we have the prospective data from Cuff and colleagues [4] showing 91% RTSA survivorship at a minimum 10-year follow-up. Importantly, these procedures were all performed for rotator cuff deficiency. RTSA survivorship may likely be longer when performed for glenohumeral OA. I am unclear as to why rotator cuff tearing after anatomic TSA is overstated considering that it was the number one complication in the findings by Parada et al. [13]—found to be 3.1% in their study of more than 2000 patients who underwent anatomic TSAs. Although there is the drawback of scapular spine fracture following RTSA, this risk is more than offset by risks unique to anatomic TSA, which include rotator cuff tearing, subscapularis repair failure, and glenoid loosening. Taken together, these complications unique to anatomic TSA add up to more than the 2% risk of stress fractures in RTSA. I concede that ROM is slightly less with RTSA. But the differences are likely not clinically relevant, and the percentage of patients who report being satisfied and outcomes scores are no different. For reasons other than lower complication rate, implant longevity, and greater ROM, Dr. Garrigues and I agree that RTSA is superior. Dr. Menendez:What are some of the gaps in knowledge in the treatment of glenohumeral OA, and how do you counsel your patients considering these gaps? Dr. Garrigues: The complications of less-than-perfect ROM (particularly internal rotation) and acromial stress fractures make many, including myself, hesitant to adopt the “RTSA for all” philosophy. Regarding the limited ROM, it should be noted that following either RTSA or TSA, patients’ ROM and pain are typically improved and thus both are successful operations. However, with proper sizing of the anatomic implant to match the native anatomy, proper soft tissue balancing, proper correction of the glenoid joint line, patient-specific guides or navigation, and augmented anatomic glenoid components where indicated, motivated patients can achieve normal or near-normal ROM. With RTSA, this complete restoration of function is not as predictable. This likely is a function of the fact that despite extensive research on impingement-free ROM, implant sizing, and implant positioning, there is more work to be done to determine the best RTSA architecture for each patient. Anatomic TSA is simple—restore the anatomy. It will require decades more research before we understand the ideal way to reconstruct a shoulder to achieve normal function with RTSA. Regarding acromial stress fractures, this remains the sword of Damocles hanging over every RTSA. The exact patients for whom one might be tempted to choose RTSA over anatomic TSA––older, lower-demand, perhaps even patients with frailty––are exactly the ones who are at risk for this devastating complication. Some acromial stress fractures will result in additional surgery, some will simply call for additional time in the sling or a slower rehab course, but this is a complication that simply does not plague anatomic TSA. Dr. Jawa: One of the major points that needs to be reiterated is that an RTSA performed for all other indications besides glenohumeral OA should not be compared to anatomic TSA for glenohumeral OA. The motion is far less predictable in RTSA for patients with a diagnosis other than glenohumeral OA. Now, if most surgeons are limiting their RTSAs to the most extreme glenohumeral OA pathology, the motion will appear dramatically less, and less predictable, as Dr. Garrigues points out. Saini et al. [14] recently found that the ROM and validated, patient-reported outcomes of RTSA for OA are far better than for rotator cuff arthropathy. I agree there is more work to be done to continue improving patient-reported outcomes, but the recognition of likely future improvement while already demonstrating comparable clinical outcomes to the anatomic TSA implies a greater potential for this implant. The risk of stress fracture in patients with glenohumeral OA treated with RTSA is far lower than in patients with rotator cuff arthropathy, as Dr. Garrigues and I showed in a large multicenter study of more than 6000 patients [2]. Most of these patients heal with limited long-term impact with additional sling use, undergo no further surgery, and are satisfied. Yet, patients with anatomic TSA who have a revision for subscapularis rupture, rotator cuff tearing, or glenoid loosening are not satisfied. Yes, stress fracture is a real complication that distinguishes RTSA from anatomic TSA, but by using an RTSA, we obviate other common complications seen in anatomic TSA. In summary, RTSA for glenohumeral OA is a more cautious approach, with fewer reported complications and revisions and improved longevity (as documented with the best methodology) compared to anatomic TSA. Dr. Menendez:I often hear at meetings the statement, “The best anatomic TSA is better than the best RTSA.” What do you think patients value most after a shoulder arthroplasty procedure? Dr. Jawa: I understand the sentiment of this statement. A perfect anatomic TSA feels and acts like a normal shoulder. I also agree, anecdotally, that even our best RTSAs (in the short term) have some limited motion. However, when you use a more-formal measure of clinical success than anecdotal evidence, like patient-reported outcomes scores, you will find that patients do not experience this. The surgeon and patient perception of the “best” arthroplasty option for glenohumeral OA may be different in 10 or 15 years. There is increasing evidence that shows RTSA for glenohumeral OA having greater longevity than anatomic TSA for glenohumeral OA, as demonstrated within the Australian Orthopaedic Association National Joint Replacement Registry. This should cause us to ask: Which prosthesis is actually a more “conservative” measure that is less likely to necessitate future intervention? Taking a long-term perspective, a perfect anatomic TSA at 2 years that results in a revision at 12 years for rotator cuff tear may lose the argument to a slightly less-than-perfect RTSA that doesn’t ever undergo revision. Patients want one surgery that lasts them forever. When considering major postoperative outcomes, pain is the number one factor that drives patients to surgery. Function and ROM are important but of secondary importance to many patients. Additionally, surgeons and patients perceive function differently. I regularly have patients with 140° of forward elevation who rate their shoulder as 100%. Few patients expect a “normal” shoulder—most simply want a better shoulder. These expectations obviously differ by age and activity, and there are certainly patients who could benefit from the extra 5° of motion provided by an anatomic TSA, but these patients are few and far between. As surgeons, we often superimpose our perceptions of “objective” outcomes (ROM or measurable qualities) over the desires of our patients. I feel that this bias provides the basis for why anatomic TSAs are still favored over RTSAs for glenohumeral OA by a majority of shoulder surgeons. Dr. Garrigues: Great points. A couple of thoughts to consider a different perspective. Our shoulder outcome scores do not place great emphasis on internal rotation. They value pain relief and comfort with activities of daily living, both of which will have dramatic improvement with either anatomic TSA or RTSA. With this ceiling effect, even though the anatomic TSA has superior ROM, both patients do quite well—nearly maxing out patient outcome scores. If ROM is studied as a separate variable, there is no question that anatomic TSA has better internal rotation. Is this important? It is not up to me—ask your patient! If they do not care about getting their hand up their back, do an RTSA. Next, Dr. Jawa showed a lot of concern about rotator cuff tears. Let’s not forget we are talking about glenohumeral OA, where the incidence of full-thickness rotator cuff tears is low––much lower than even among asymptomatic, age-matched controls. Why is this? The critical shoulder angle has been a well-studied area of research during the past decade [3], and we found that the interplay between the force vectors of the rotator cuff and deltoid may combine to either produce too much glenohumeral compression (protecting the rotator cuff but leading to a higher incidence of glenohumeral OA) or too little compression (leading to increased force on the rotator cuff but a lower incidence of glenohumeral OA) [3]. Indeed, if you do a good job with your osteotomy or subscapularis repair, the incidence of postoperative rotator cuff tear is quite low because shoulders with glenohumeral OA are typically not prone to rotator cuff tear. Furthermore, small tears isolated to the supraspinatus found at surgery do not predispose to any issues [5]. For this reason, many shoulder arthroplasty surgeons do not get MRI for their glenohumeral OA patients as the concern for relevant cuff pathology is so low. Lastly, I would like to propose two concepts: Hierarchy of Shoulder Arthroplasty Success: Good TSA better than Good RTSA better than Bad TSA better than Bad RTSA Technical Ease of Shoulder Arthroplasty: Good TSA harder than Good RTSA harder than Bad TSA My best TSAs achieve full strength, full ROM, and no pain. The shoulders are, as far as the patients and any examiner can tell, normal. Patients forget which side they had the operation. My best patients with RTSA are happy. They can do everything they want. They have no pain, and they participate in all activities. But if you examine their shoulders carefully or ask them their ROM, particularly in internal is slightly Now, both of these are better than a TSA, and this is where it It is easier to do a good RTSA than a good TSA. In RTSA, the surgeon can more bone and even rotator cuff to with glenoid tissue and implant sizing is not as much of a is by the semiconstrained and humeral do not appear to be as critical for implant repair of the subscapularis is not The is easier for the patient and the use of the sling is not as The is the though the are for RTSA. In short, it is and to achieve a good result with RTSA. This may some of the with results in registry studies have surgeons versus published studies are with the results of My in this is a a and a I Dr. Jawa would agree that the technical of the operation should not be a factor in the treatment for our patients. This is my but I fear many take an easier you can achieve with less work than then this is the In my as of if your patient is a for anatomic TSA, and you can achieve that is what I would want for my shoulder.