P. BOILEAU* et G. WALCH**
*Centre hospitalier Universitaire de Nice - 06006 Nice Cedex 1
** Centre Hospitalier Universitaire de Lyon - 69310 Pierre-Benite

Introduction

The goal of a surgeon inserting a shoulder prosthesis is, as with all prosthetic surgery, to come as close as possible to the anatomical model. Two questions are therefore raised:

1- Are we sure that we are perfectly familiar with the anatomical model ?
2- Do the standard prostheses at our disposal, including modular systems, allow us to reproduce this anatomical model ?

Since 1988, with the aim of answering these two questions, we have been working on a morphological study of the glenohumeral joint and we have been elaborating possibilities of reproducing the morphology using a shoulder prosthesis. We will discuss here the development of the humeral side of these studies and their consequences on the design of the shoulder prosthesis; the glenoid side will be dealt with in a separate article.

Morphological study of the proximal end of the humerus

The aim of the anatomical study (2, 3, 15)was to precisely establish both the dimensions and shape of the proximal end of the humerus. The study was carried out using dry bones taken from cadavers and using modern investigation techniques; precision measurement devices and computer assisted design prograrns. The pre-calibrated precision measurement device was made with a support designed to hold the humerus so that a number of points on the bone surface could be digitized. (fig. la).

Each humerus was fixed in the support by the top of the humeral head and at the level of the distal humerus. With the bone turning 10 degrees at a time, the digitization of its surface points was made from parallel sections located every 5 mm on the proximal and distal epiphyses and every 10 mm on the diaphysis.
For each section, a micron precision probe was used to note the coordinates of 36 regularly spaced points for each bony section, amounting to l 200 points per humerus.
Particular attention was given to the digitization of the articular surface of the humeral head in that a number of points at the periphery and over the entire surface of the articular cartilage were noted. All of these coordinates were then transmitted to a computer equipped with a software program giving a threedimensional image of the humerus (fig. lb).

　

Figure 1a : Digitization of humeri using the precision measurement device relayed to a computer for analysis of the bone shape and dimensions	Figure 1b : Resulting three-dimensional image of a humerus.

The computer program gave a precise calculation of the different morphological parameters (diameter, thickness, inclination, retroversion) on the one hand and produced a model of the proximal end of the humerus using simple geometrical figures on the other: epiphyseal sphere and metaphyseal cylinder (fig. 2a).

Two axes were defined (fig. 2a):

- he diaphyseal axis (yellow): rotational axis of the humeri.
- and the proximal metaphyseal axis (orange), along which a prosthetic stem can be inserted

　

Figure 2a: Model of the humerus produced using simple geometrical figures: epiphyseal sphere and metaphyseal cylinder The orange axis is that of the metaphyseal cylinder along which a prosthetic stem can be inserted.	Figure 2b: Humeral head sphericity and mathematical relationship linking the three parameters: diameter of the sphere, diameter of the head and thickness of the humeral head.

Sphericity of the humeral head

We first wanted to verify the sphericity of the humeral head. Using the recorded articular surface points we were able to determine the diameters of the potential humeral sphere. In order to verify this sphericity we used the classical notion of "diameter size", defined as the difference between two extreme diameters: (E)=(D1)-(D2).

Therefore a geometrical figure where the difference between the two most extreme diameters is less than 1 mm may be considered as being a sphere. Of the 160 humeri studied, the difference between extreme diameters was less than I mm in 88.2 % of cases. We can therefore confirm that, in almost 90 % of cases, the humeral head is comparable to a sphere. The articular surface constitutes about a third of this sphere. This is an important concept that had not been previously fully acknowledged by other authors and which has been verified by recent work from Iannoti and Coll (10). (10).

Dimensional variations of the humeral articular surface

Since the humeral head is a sphere and the articular surface constitutes only part of this sphere, there is a mathematical relationship between the diameter of the humeral sphere, the diameter of the humeral head and the thickness of the humeral head (fig. 2b). As two of these parameters are known, the third one can be calculated.

Our anatomical studies showed that there were wide variations in the dimensions of the articular surface:

- the diameter of the articular head was highly variable, with an average of 43.2 mm, ranging from 36.5 mm to 51.7 mm.

- similarly, the thickness of the articular head was highly variable, with an average of 15.2 mm, ranging from 12 mm to 18 mm.

Modularity: the consequences of dimensional variation

The large variation in the dimensions of the articular surface in both diameter and thickness makes the idea of modularity a prerequisite in the design of shoulder prostheses.

Standard humeral prostheses do not take into account the anatomical variations encountered. The Neer prosthesis for example with a single diameter of 50 mm and two head thicknesses (15 and 22 mm) can only rarely reproduce the anatomical model.

This undoubtedly explains certain poor functional results and certain abnormal biomechanical situations with reduction of gleno-humeral prosthetic mobility that we observed during a radiocinematographic study of Neer prostheses (1). This recently led to the appearance of a number of modular prostheses (second generation prostheses) Table 1.

　

1st generation prostheses: «monoblock, non-modular»

- NEER (3M) - COFIELS (RICHARDS) - FENLIN (ZIMMER)

2nd generation prostheses: «modular»

- NEER (3M) - BIOMODULAR (BIOMET) - SELECT SHOULDER (intermedics Orthopedics, Inc.) - GLOBAL (DEPUY)

3rd generation prostheses: «modular and adaptable»

-AEQUALIS (TORNIER)

Table no. 1: different types
of shoulder prostheses (numerous implants)

The advantages of a modular prosthesis are:
1. The ability to select an exact diameter and thickness of articular head, a factor which is important in the musculoligamentous balance of the shoulder.
2. The ability to match the differing radii of curvature between the glenoid and the corresponding prosthetic humeral head.
These two characteristics allow the surgeon to obtain a better match of components resulting in a biomechanical interaction that correlates more closely with normal anatomy.
However the modularity only allows compensation for dimensional variations of the articular surface and not for variations in orientation and position in space of the articular surface: in other words, modularity does not account for the proximal end of the humerus.

Variations in the orientation of the humeral articular surface

Our anatomical studies have shown that the angular variations were significant, especially the inclination and retroversion of the articular surface. The concept of an average value here is relative and it is the concept of variation which must be kept in mind. (Fig. 3a and 3b).

　

Variations in orientation in space of the articular surface:
Figure 3a: The inclination of the articular surface between 114 and 147 degrees.	Figure 3b : The retroversion of the articular surface varies between -6.5 and 47.5 degrees and is in no case a constant between 30 and 40 degrees.

- the humeral inclination is on average 130 degrees and varies between 114 and 147 degrees.

- similarly the humeral retroversion is highly variable between individuals and even between right and left sides of the same individual. The average angle of retroversion as measured on 65 humeri was 17.9 degrees varying from -6.5 to 47.5 degrees.

Variation of position in space of the humeral articular surface

The intramedullary axis along which the humeral prosthesis is inserted is that of the proximal metaphysis since a change in curvature occurs at the humeral diaphysis. It is therefore possible to identify a proximal humeral metaphyseal cylinder which will accomodate the stem. We have shown that the articular surface was comparable to a sphere. Using computer analysis, it is possible to determine the sphere which corresponds to the articular surface.

The next step therefore was to establish the position in space of the sphere in relation to the cylinder.

Our anatomical work showed that the spherical humeral head does not sit on the base of the cylinder but lies eccentrically in two planes. (fig.4a, b, c).

　

Figure 4: The variations of position in space of the articular surface are defined by the position of the spherical humeral head in relation to the humeral cylinder: a) the medial offset of the spherical humeral head in the frontal plane b) the posterior offset of the spherical humeral head in the sagittal plane c) the combined medial and posteriar offset in both the frontal and sagittal planes.

- in the frontal plane, the medial displacement of the spherical head results in an offset. This medial offset is characterized in practice by the proximal metaphyseal axis crossing the periphery of the articular head at a level of what we have called the "critical point", really a "hinge point" (fig. 5a). So, the center of the sphere and the center of the cylinder may be separated in the frontal plane by 6.9 mm on average with a range from 2.9 to 10.6 mm. This medial displacement of the articular surface has never been described before. We consider that the design of a humeral prosthesis should take this concept of offset into account (Fig.5a, b, c).

　

Figure 5: Requirement to include the medial offset in the design of shoulder prostheses. 5a) the "hinge point" between the proximal metaphyseal axis (the prosthetic axis) and the articular surface at the top of the articular head. 5b) Standard prostheses (here ar the Neer prosthesis) do not take this "hinge point" into account and result in medialization of the articular head. 5c) The Aequalis prosthesis incorporates this data into its design.

- in the sagittal plane, the spherical humeral head has a posterior eccentric position which defines the posterior offset which is more variable, with an average of 2.6 mm and a range from -0.8 to 6.1 mm in our series. This posterior offset of the humeral articular surface, which has also been shown by Roberts and Coll (14), must also be included in the design of a humeral prosthesis (fig. 4b).

The offset of the spherical humeral head usually occurs in both the frontal and sagittal planes resulting in a true combined medial and posterior offset.

Adaptability: the result of variations in the orientation and position in space of the articular surface

The adaptability of an implant is a new concept in prosthetic surgery that allows preoperative planning which takes into account morphological parameters of the joint to be replaced. This concept also allows a practical intraoperative answer to variations in anatomy without the inconvenience of stocking multiple prostheses or having them made pre-operatively.

Concerning shoulder prostheses, the adaptability of the humeral implant allows integration of the 4 following parameters:

1 - the inclination
2 - the retroversion
3 - the medial offset
4 - the posterior offset
　

We will consider the 4 parameters one by one.

1) The inclination of the articular surface is a factor which is much too variable to be imposed in an arbitrary manner; cutting the humeral head along a fixed inclination corresponding to that of the prosthesis or a cutting jig gives a very approximate result.

Conversely cutting the humeral head by following the limits of the anatomical neck does not always allow for correct positioning of a standard prosthesis. (Fig. 6a).

　

Figure 6: Requirement to include individual angulation variations in the design of a shoulder prosthesis: a) Anatomical osteotomy of the humeral head does not always allow one to correctly position a standard prosthesis. Conversely, cutting the humeral head according to a fixed inclination may change the rational centers. b) The Aequalis prosthesis allows adaptation to individual inclination through a system of variable necks between the stem and the prosthetic head.

This variation in inclination must be included in the design of the humeral implant.

The Aequalis prosthesis allows for variable inclination due to a system of variable necks between the stem and the prosthetic head.

The inclination is chosen intraoperatively using a trial neck. (Fig. 6b).
2) The retroversion of the articular surface is another very variable factor which should be included in the design of the humeral prosthesis. Indeed, imposing a fixed and arbitrary humeral retroversion between 30 and 40 degrees, as recommended in surgical articles (6,9,10) gives a very approximate result.

It appears very difficult, if not impossible, to match the individual retroversion of the articular surface. In fact the problem is considerably simplified when we remember that the humeral retroversion is determined by the plane of section through the anatomical neck (Fig. 7a, 7b). There is therefore only one possible humeral retroversion for each humerus which is easy to determine by marking the perpendicular to the anatomical neck (Fig. 7b).

The trial neck of the Aequalis prosthesis allows both determination of the inclination using the mobile plate and the angle of retroversion by marking the perpendicular to the anatomical neck using an osteotome inserted into the groove designed for this purpose.

In this way, the slot in the cancellous tuberosity corresponds to the site of the prosthetic fin.

　

Figure 7: Requirement to include individual retroversion variations in the design of the shoulder prosthesis: a) The plane of section of the anatomical neck therefore determines individual humeral retroversion. b) There is thus only one possible retroversion for each humerus determined by the perpendicular to the anatomical neck c) The trial neck allows both the inclination to be determined using the mobile plate and the retroversion to be fixed by marking the perpendicular to the anatomical neck using an osteotome inserted into the groove designed for the purpose.

Thus, marking the anatomical neck is a fundamental step during shoulder arthroplasty, allowing both the individual humeral inclination and the retroversion to be determined, giving rise to a better definition of the orientation in space of the articular surface. This marking of the anatomical neck, which is a delicate part of the operation, can only be done after complete removal of osteophytes (fig. 8). Fatty tissue, similar to that found in osteoarthritic sockets, usually marks the limit between healthy cortical bone and the pathological osteophytes

　

Figure 8: Marking of the anatomical neck which is a fundamental part of shoulder arthroplasty allowing determination of the humeral inclination and retroversion, can only be done after complete ablation of the crown of osteophytes.

3) The design of a humeral prosthesis must also take into account the medial offset of the articular surface by respecting the hinge point between the proximal metaphyseal axis and the top of the articular head (fig. 5). The Neer prosthesis, for example, does not respect the medial offset of the articular surface and this results in reduction of the deltoid and rotator cuff lever arm. This discrepancy explains the suboptimal results often seen after implantation of the Neer type prosthesis in a standard humerus. (fig 9)

- The prosthesis is either inserted along the proximal metaphyseal axis giving rise to an inadequate orientation of the humeral plane of section and protrusion above the top of the prosthetic head which is potentially damaging to the rotator cuff (fig. 9a),...

- Or the prosthesis is tilted so that it is better aligned and reconstructs the anatomical model, but this is not always possible due to the length of the prosthetic stem: too long in the standard prosthesis and not taking into account the change in curvature of the humeral diaphysis in the frontal plane (fig. 9b and 9c).

　

Figure 9: Results of the absence of the articular surface medial offset in standard prostheses: a) Superior protrusion of the prosthetic head, potentially damaging to the rotator cuff. b) Requirement to tilt the prosthesis so that it better matches the anatomical model... c)... which is not always possible because of the length of the stem which is "in conflict' with the change in diaphyseal curvature of the humerus in the frontal plane.

This discrepancy enables us to understand the post-operative xrays obtained with standard prostheses (fig. 10).

　

Figure 10: Post-operative x-rays often observed after insertion of a standard humeral prosthesis. a) Humeral centering of the prosthesis causing protrusion of the prosthetic articular surface due to the medial offset not having been taken into account in the design of the prosthesis. b) Prosthesis tilting, allowing "cheating" in reproducing the articular surface medial offset.

With the Aequalis prosthesis the inclination can be measured from the hinge point to the arch of the trial neck (fig. 11).

　

Figure 11: Practical determination of the hinge point; articular surface medial offset control allowing the inclination to be measured . a) The 'hinge point" (critical point) is located at the top of the humeral cut plane, marking the crossing between the proximal metaphyseal axis and the top of the articular surface. b) The variations in inclination are determined pre-operatively using an articulated inclination guide.

4) Finally the design of a humeral prosthesis must take into account the posterior offset of the articular surface. None of the currently available standard shoulder prostheses, even modular ones allow this posterior articular surface offset to be respected.
In certain cases, during implantation of a standard humeral prosthesis, the prosthetic head is found to be pushed forward so that it does not cover the bony section (fig. 12).

　

Figure 12: Results af not respecting the articular surface medial offset in standard prostheses: a) After humeral head resection, reaming of the humeral shaft reveals the articular surface medial offset. b) Insertion of a standard humeral implant along the proximal metaphyseal axis... c) ... dues not allow the postenor aspect of the cut surface to be covered, requiring resection of the neck and artificial increase of the humeral retroversion.

The solution then consists in withdrawing the prosthesis and artificially increasing retroversion after a modified bony section to accommodate this new adaptation.

Therefore, when this posterior offset is not taken into account in the design of shoulder prostheses, the biomechanical consequence is serious: displacement of the center of rotation (Fischer, Carret, Gonon (7)). This leads to kinematic disturbances secondary to change in the action of the lever arm.

There are in fact two solutions to incorporate the posterior offset of the articular surface:

- Either a different prosthesis can be made for the left and right sides, as proposed by Wallace and Coll (14). The disadvantages are twofold: a fixed posterior offset and the need to stock multiple implants.

- Or a prosthesis adaptable on both sides might be considered, incorporating a variable posterior offset (as shown in our anatomical studies). We have concentrated on this solution, designing an original prosthetic head with an eccentric displacement system.(fig. 13).

　

　

Figure 13: Integration of the articular surface medial offset in the design of the Aequalis prosthesis: a) An original and patented system with eccentric indexing allows the articular surface to be offset postenorly. b) ... from the nght side c) ... from the left side, avoiding the need for different implants for each side.

The Aequalis Prosthesis: a modular and adaptable prosthesis

The Aequalis shoulder prosthesis is a nonconstrained glenohumeral prosthesis whose design is based on the results of fundamental studies performed in collaboration with engineers from the TORNIER company over a five year period. Its development is based on two essential features which, according to us, are prerequisite to obtain consistent clinical results.

- development of a prosthesis which precisely matches the anatomy by using an adaptable modular humeral component with the availability of different sizes of glenoid component.

- the access to instrumentation which allows straightforward and reproducible implantation of the prosthesis by different surgeons.

- a constant medial offset incorporated in the prosthetic design (fig. 8).

- a variable posterior offset, selected by the surgeon intraoperatively, using an original eccentric indexing system located on the inferior aspect of each head. Eight positions are available allowing the articular head to be offset posteriorly or anteriorly and inferiorly or superiorly, until the cut bony surface is perfectly covered (fig. 13).

It consists of three parts (humeral stem, neck and head) made of a titanium alloy which, because of its design and choice of components, adapts perfectly to the anatomy of the joint rather than forcing the joint anatomy to adapt to the prosthesis.

The length of the humeral stems has been shortened to adapt to the proximal metaphyseal humeral cylinder before the change of curvature that occurs at the humeral diaphysis.

There is only one prosthetic stem length for the three diameters: 6,9 and 12 mm. The rotatory stability of the humeral stem is ensured by optimal filling of the metaphysis, a fin in the upper part containing two holes to allow reinsertion of the tuberosities in fracture cases and, finally, anterior and posterior grooves.

The modularity is provided by a range of seven articular heads of increasing size, the diameter and thickness of which were determined from our anatomical studies. A single head thickness is used for each diameter (up to the largest 50 mm diameter for which there are two available thicknesses) so that the anatomical variations encountered can be better matched. It is important that the thickness and diameter of the articular surface are respected so that the musculoligamentous balance of the shoulder is recovered and that the differing radii of the head-glenoid complex are maintained.

Adaptability is ensured by:
- variable inclination given by the 125, 130, 135 and 140 degree necks, selected by the surgeon using the trial neck (fig. 5 and 11).
- variable retroversion selected "automatically" using the trial neck (fig. 6).
This movement of the prosthetic articular surface in all spatial planes helps the implant to adapt to all pathological situations encountered and therefore the proximal end of the humerus can be reconstructed according to the anatomical model.

Conclusion

The geometrical configuration of the proximal end of the humerus is much more complex than previously described. A standard humeral prosthesis, even a modular one, can reproduce neither its shape nor its dimensions. This led us to review the design and development of humeral prostheses. The variation in shape must be matched by the implant adaptability and the dimensional variations that can be accounted for through the implant modularity.

Until now the concepts of modularity and prosthetic adaptability had not been included in the design and development of the humeral implant. This undoubtedly explains the inconsistent clinical results achieved with shoulder prostheses.

The Aequalis shoulder prosthesis has adopted all these characteristics. It allows for the individual retroversion, inclination and the combined medial and posterior offset of the articular surface (revealed in our studies) to be respected.

So it has been possible to solve a relatively complex geometrical problem through simple but original mechanical solutions and the preliminary clinical results seem to confirm the merits of this concept of adaptability.