The three-dimensional (3D) geometry of the proximal humerus is extremely variable. Our previous anatomic studies (7, 8, 9, 10, 11, 12), as well as others (1,13, 14, 15, 16, 17, 18, 19, 20, 21, 22) have shown that inclination and retroversion vary with age, sex, race, and side to such an extent that one cannot talk of an average value. In our anatomic study, inclination of the articular surface has been found to vary from 125 to 140 degrees, and retroversion of the articular surface has been found to vary from 0 to 50 degrees. We have demonstrated that inclination and retroversion should be individualized if one wishes to try to reproduce the patient’s anatomy. Moreover, our anthropometric studies have shown that the articular surface is medially
and posteriorly offset in relation to the humeral medullary canal. These variable geometric parameters also should be taken into account both in the technique of implantation and in the design of the humeral component.

The first-generation (monoblock, Neer type) and second-generation (modular) prostheses have a relatively fixed design. This design prevented true anatomic restoration; displaced the center of rotation; and overtensioned the rotator cuff leading to potential impingement, abutment, or instability (Fig. 3-1). Humeral head replacement was carried out based on two arbitrarily fixed angles: inclination and retroversion (2, 3, 4,23,24). Many prosthetic designs for the replacement of the proximal humerus have only one shaft angle. This fixed inclination limits the potential for the surgeon to reproduce the original anatomy. Making a humeral cut of 140 degrees of inclination in a patient who has only 125 degrees displaces the center of rotation proximally and alters the kinematics of the glenohumeral joint. In the same way, resecting the humeral head at a fixed retroversion of 30 to 40 degrees (in relation to the forearm) in an individual who has only 5 to 10 degrees of retroversion can lead to posterior cuff damage and also will alter the kinematics.

In fact, some of the second-generation prostheses are even “less anatomic” than the first-generation ones in terms of restoring the original 3-D geometry of the proximal humerus—cementless fixation has been added to their relatively fixed geometry. They do not allow any intraoperative flexibility for the surgeon: filling the medullary canal with a cementless prosthetic stem dictates the position of the prosthetic head and makes it more difficult, if not impossible, to match the medial and posterior offset. As a consequence, achieving coverage of the anatomic osteotomy is impossible.

Additionally, our study has demonstrated that humeral head diameter and thickness are linked and have a fixed relationship, with the exception of large diameters (i.e., more than 50 mm). This means that the prosthetic head diameter and thickness should be linked. There is only one thickness possible for one diameter, except for the large heads where two thicknesses for one diameter can accommodate the anatomy. From an anatomic standpoint, tensioning the soft tissues with different prosthetic head thicknesses compromises reconstruction. This concept is applicable for knee replacement but not for shoulder replacement. Harryman and colleagues have shown that increasing the thickness of the humeral head by only 5mm decreases the range of glenohumeral motion by 20 to 30 degrees (25). Conversely, Jobe and Ianotti have demonstrated that decreasing the thickness of the humeral head by 5 mm reduces the glenohumeral excursion by 24 degrees (16).

Determination of individual humeral head diameter, thickness, inclination, retroversion, and posterior and medial offset requires careful definition of proximal humeral anatomy. Basing the orientation and position of a humeral prosthesis using the boundaries of the anatomic neck as a landmark is a reliable, reproducible technique. This yields more accurate reconstruction than positioning the prosthesis following the standard fixed inclination, retroversion, and fixed posteromedial offset. However, this requires that two conditions be met: (a) that the limits of the anatomic neck are identifiable even in arthritic conditions of the shoulder where the bone anatomy of the proximal humerus may be markedly distorted; and (b) that a prosthesis is available, such that parameters can be adapted to each individual shoulder. The Aequalis shoulder prosthesis (Tornier, Inc., Houston, USA), introduced in 1991, has been designed so that inclination and offset can be adjusted to reproduce the patient’s individual anatomy through prosthetic modularity and adaptability. This was the first third-generation prosthesis.

Since then, many so-called “third-generation” prostheses have followed; some of them, however, are not truly adaptable. They have only one neck-shaft angle and use a guide to cut the humeral head with a fixed retroversion angle. The head sizes are not based on the normal relationship between head thickness and diameter. The designation “third-generation” prostheses or “adaptable” prostheses perhaps should be restricted to those prostheses that allow individual humeral head diameter, thickness, inclination, retroversion, and posterior and medial offset to be adjusted to replicate normal anatomy.