1.1 Introduction

In its modern-day form, reverse total shoulder arthroplasty is a product of many decades of evolving constrained shoulder arthroplasty designs. An understanding of the history of constrained shoulder arthroplasty and appreciation of previous design flaws are beneficial in the understanding of current and future reverse arthroplasty implants.

No historical chapter would be complete without discussion of the first shoulder replacement surgeries. These were performed for tubercular infections of the shoulder, and the components were made from a number of materials, including platinum, rubber, and ivory. Implantation of the first shoulder arthroplasty is generally attributed to Emile Péan in 1893, although Themistocles Gluck reported on his design for a prosthetic shoulder made of ivory in the mid-19th century. ¹ , ² Subsequently, many have developed their own design of shoulder prostheses with varying results.

Shoulder arthroplasty design can be divided into two categories, anatomic and nonanatomic designs. Of the anatomic designs, Dr. Charles Neer’s original hemiarthroplasty, implanted for irreparable proximal humerus fractures in the 1950s, is generally considered the foundation of the current anatomic replacements. ³ Outcomes were generally good when this implant was utilized for diagnosis of glenohumeral osteoarthritis or with an intact rotator cuff, but superior migration and variable strength and function were seen in patients with rotator cuff deficiency. ⁴ Rotator cuff tear arthropathy, as it was termed by Neer, was characterized by a massive rotator cuff tear with superior humeral head migration and diminished acromiohumeral distance with “femoralization” of the proximal humerus and “acetabularization” of the acromion. ⁵ Since this initial design, the anatomic arthroplasty has evolved to incorporate a polyethylene glenoid component and modularity on the humeral side to better match patients’ anatomy. These implants have continually shown excellent results for patients with glenohumeral arthrosis and an intact rotator cuff.

The high rate of failure of early anatomic replacements performed in the presence of rotator cuff deficiency led to many innovations involving a more constrained prosthesis to account for the loss of the stabilizing force of the rotator cuff. These were first introduced in the 1970s with a very high rate of failure due to component loosening. The loosening was a result of high forces across the bone–implant interface as a result of the constrained nature. These initial attempts were generally based on the concept of stabilizing the shoulder with an implant resembling other, more constrained, ball-and-socket joints in the body. Initial attempts at constrained, anatomic designs were met with failure, ⁶ , ⁷ , ⁸ so attention was turned to a reversed design. The basic design was to “reverse” the ball-and-socket relationship with a fixed “glenosphere” on the glenoid side with scapular fixation and a socket on the humeral side. In 1987, Dr. Paul Grammont described his principles of medializing and lowering the center of rotation that formed the basis for the design of successful reverse, nonanatomic arthroplasty. ⁹ In more recent years, these principles have been challenged with designs that incorporate a more lateralized center of rotation and report similar improvements in patient outcomes. The debate regarding the optimal center of rotation in reverse arthroplasty continues and is an area of active research.

According to Pupello, it is a common misconception that the common cause of failure of the early implants was glenoid component loosening. ¹⁰ He reviewed current studies referencing early reverse shoulder arthroplasty experiences and believed that 72% contained at least one quotation error, meaning the recent publication erroneously cited the early publications for findings that the original publication did not conclude or support. The most common error, he believed, was assertion that the early designs had glenoid-sided failure related to a certain causative factor, despite the original study not having been designed to study that factor (or any factor causing failure). Therefore, conclusions could not be or were not made in the original article regarding the cause of failure of the earlier design. Of these erroneous citations, over half of the recent articles attributed the original failures to a lateralized center of rotation and/or high shear stresses at the bone–implant interface, using citations from the original articles as support. Pupello did not believe any of the original studies cited were designed to study the causative factors of failure; and further, he also pointed out that a number of the original articles did not report glenoid-sided failure. ¹¹ , ¹² , ¹³ , ¹⁴

1.2 Neer (1970–1973)

Dr. Neer had three constrained, reversed shoulder arthroplasty designs that he developed between 1970 and 1973. These are generally thought of as the first of these types of implants. All of these implants included a keeled, cemented method of glenoid fixation. His initial attempt, the “Mark I,” included an oversized ball to improve motion. Drawbacks of the implant included a large amount of bone required for scapular fixation, glenoid loosening, and the inability to attach the rotator cuff. ¹⁵ His next iteration, the “Mark II,” incorporated a smaller glenosphere to improve the ability to repair the rotator cuff. Unfortunately, the smaller glenosphere limited postoperative range of motion, and glenoid loosening remained an issue.

Neer’s third and final attempt at constrained arthroplasty was the “Mark III.” To improve range of motion and limit constraint, this implant allowed axial rotation of the stem within the humeral diaphysis. This implant continued to have problems with glenoid loosening, and Neer abandoned the concept of fixed-fulcrum implants. He did not believe the scapula was capable of providing fixation that was durable enough to resist the forces imparted by a fixed-fulcrum implant. ¹⁵

1.3 Reeve’s Prosthesis (1972)

While there have been no reported clinical results of this implant, it represents one of the first reverse arthroplasty designs (► Fig. 1.1). This implant had a cemented, spike glenoid component, of which the authors discussed the difficulties with implantation. The ball was held into the humeral socket with a screwed on collar. The center of rotation was meant to be anatomic. ¹⁶

1.4 The Gerard and Lannelongue Prosthesis (1972)

This implant was a reverse prosthesis in which the glenoid component was secured using screws instead of cementation (► Fig. 1.2). ¹⁴ , ¹⁷ Of 22 reported cases in difficult reconstructive settings (tumor resection, revision arthroplasty, posttraumatic arthrosis), there were 9 failures (3 dislocations, 4 implant breakages, 2 infections). Of note, none of the failures were due specifically to glenoid-sided loosening.

1.5 Kolbel Prosthesis (1973)

Noting that the fixation of the implant to the scapula could cause catastrophic failure if there was a scapular fracture, Kolbel designed an implant that would dislocate when a specific torque value was exceeded to decrease the incidence of scapular fracture. To further protect the cemented fixation of the central peg, this prosthesis employed a forked outrigger that was placed about the scapular spine and clamped and transfixed with a screw. The humeral component was cemented and all-polyethylene, with a retaining ring that secured the head of the prosthesis (► Fig. 1.3). There was only one report of early scapular failure secondary to central pin breakage. The implant design only allowed a 90-degree range of motion. ¹⁸ , ¹⁹ , ²⁰ , ²¹