Shoulder arthroplasty humeral stem design has evolved to accommodate patient anatomy characteristics. As a result, stems are available in numerous shapes, coatings, lengths, sizes, and vary by fixation method. This abundance of stem options creates a surgical paradox of choice. Metrics describing stem stability, including a stem's resistance to subsidence and micromotion, are important factors that should influence stem selection, but have yet to be assessed in response to the diametral (i.e., thickness) sizing of short stem humeral implants.

Eight paired cadaveric humeri (age = 75±15 years) were reconstructed with surgeon selected ‘standard’ sized short-stemmed humeral implants, as well as 2mm ‘oversized’ implants. Stem sizing conditions were randomized to left and right humeral pairs. Following implantation, an anteroposterior radiograph was taken of each stem and the metaphyseal and diaphyseal fill ratios were quantified. Each humerus was then potted in polymethyl methacrylate bone cement and subjected to 2000 cycles of 90º forward flexion loading. At regular intervals during loading, stem subsidence and micromotion were assessed using a validated system of two optical markers attached to the stem and humeral pot (accuracy of <15µm).

The metaphyseal fill ratio did not differ significantly between the oversized and standard stems (0.50±0.06 vs 0.50±0.10; P = 0.997, Power = 0.05); however, the diaphyseal fill ratio did (0.52±0.06 vs 0.45±0.07; P < 0.001, Power = 1.0). Neither fill ratio correlated significantly with stem subsidence or micromotion. Stem subsidence and micromotion were found to plateau following 400 cycles of loading. Oversizing stem thickness prevented implant head-back contact in all but one specimen with the least dense metaphyseal bone, while standard sizing only yielded incomplete head-back contact in the two subjects with the densest bone. Oversized stems subsided significantly less than their standard counterparts (standard: 1.4±0.6mm, oversized: 0.5±0.5mm; P = 0.018, Power = 0.748;), and resulted in slightly more micromotion (standard: 169±59µm, oversized: 187±52µm, P = 0.506, Power = 0.094,).

Short stem diametral sizing (i.e., thickness) has an impact on stem subsidence and micromotion following humeral arthroplasty. In both cases, the resulting three-dimensional stem micromotion exceeded, the 150µm limit suggested for bone ingrowth, although that limit was derived from a uniaxial assessment. Though not statistically significant, the increased stem micromotion associated with stem oversizing may in-part be attributed to over-compacting the cancellous bed during broaching, which creates a denser, potentially smoother, interface, though this influence requires further assessment. The findings of the present investigation highlight the importance of proper short stem diametral sizing, as even a relatively small, 2mm, increase can negatively impact the subsidence and micromotion of the stem-bone construct. Future work should focus on developing tools and methods to support surgeons in what is currently a subjective process of stem selection.

Shoulder arthroplasty is effective at restoring function and relieving pain in patients suffering from glenohumeral arthritis; however, cortex thinning has been significantly associated with larger press-fit stems (fill ratio = 0.57 vs 0.48; P = 0.013)1. Additionally, excessively stiff implant-bone constructs are considered undesirable, as high initial stiffness of rigid fracture fixation implants has been related to premature loosening and an ultimate failure of the implant-bone interface2. Consequently, one objective which has driven the evolution of humeral stem design has been the reduction of stress-shielding induced bone resorption; this in-part has led to the introduction of short stems, which rely on metaphyseal fixation. However, the selection of short stem diametral (i.e., thickness) sizing remains subjective, and its impact on the resulting stem-bone construct stiffness has yet to be quantified.

Eight paired cadaveric humeri (age = 75±15 years) were reconstructed with surgeon selected ‘standard’ sized and 2mm ‘oversized’ short-stemmed implants. Standard stem sizing was based on a haptic assessment of stem and broach stability per typical surgical practice. Anteroposterior radiographs were taken, and the metaphyseal and diaphyseal fill ratios were quantified. Each humerus was then potted in polymethyl methacrylate bone cement and subjected to 2000 cycles of compressive loading representing 90º forward flexion to simulate postoperative seating. Following this, a custom 3D printed metal implant adapter was affixed to the stem, which allowed for compressive loading in-line with the stem axis (Fig.1). Each stem was then forced to subside by 5mm at a rate of 1mm/min, from which the compressive stiffness of the stem-bone construct was assessed. The bone-implant construct stiffness was quantified as the slope of the linear portion of the resulting force-displacement curves.

The metaphyseal and diaphyseal fill ratios were 0.50±0.10 and 0.45±0.07 for the standard sized stems and 0.50±0.06 and 0.52±0.06 for the oversized stems, respectively. Neither was found to correlate significantly with the stem-bone construct stiffness measure (metaphysis: P = 0.259, diaphysis: P = 0.529); however, the diaphyseal fill ratio was significantly different between standard and oversized stems (P < 0.001, Power = 1.0). Increasing the stem size by 2mm had a significant impact on the stiffness of the stem-bone construct (P = 0.003, Power = 0.971; Fig.2). Stem oversizing yielded a construct stiffness of −741±243N/mm; more than double that of the standard stems, which was −334±120N/mm.

The fill ratios reported in the present investigation match well with those of a finite element assessment of oversizing short humeral stems3. This work complements that investigation's conclusion, that small reductions in diaphyseal fill ratio may reduce the likelihood of stress shielding, by also demonstrating that oversizing stems by 2mm dramatically increases the stiffness of the resulting implant-bone construct, as stiffer implants have been associated with decreased bone stimulus4 and premature loosening2. The present findings suggest that even a small, 2mm, variation in the thickness of short stem humeral components can have a marked influence on the resulting stiffness of the implant-bone construct. This highlights the need for more objective intraoperative methods for selecting stem size to provide guidelines for appropriate diametral sizing.

For any figures or tables, please contact the authors directly.

Total shoulder arthroplasty implants have evolved to include more anatomically shaped components that replicate the native state. The geometry of the humeral head is non-spherical, with the sagittal diameter of the base of the head being up to 6% (or 2.1-3.9 mm) larger than the frontal diameter. Despite this, many TSA humeral head implants are spherical, meaning that the diameter must be oversized to achieve complete coverage, resulting in articular overhang, or portions of the resection plane will remain uncovered. It is suspected that implant-bone load transfer between the backside of the humeral head and the cortex on the resection plane may yield better load-transfer characteristics if resection coverage was properly matched without overhang, thereby mitigating proximal stress shielding.

Eight paired cadaveric humeri were prepared for reconstruction with a short stem total shoulder arthroplasty by an orthopaedic surgeon who selected and prepared the anatomic humeral resection plane using a cutting guide and a reciprocating sagittal saw. The humeral head was resected, and the resulting cortical boundary of the resection plane was digitized using a stylus and an optical tracking system with a submillimeter accuracy (Optotrak,NDI,Waterloo,ON). A plane was fit to the trace and the viewpoint was transformed to be perpendicular to the plane. To simulate optimal sizing of both circular and elliptical humeral heads, both circles and ellipses were fit to the filtered traces using the sum of least squares error method. Two extreme scenarios were also investigated: upsizing until 100% total coverage and downsizing until 0% overhang.

Total resection plane coverage for the fitted ellipses was found to be 98.2±0.6% and fitted circles was 95.9±0.9%Cortical coverage was found to be 79.8 ±8.2% and 60.4±6.9% for ellipses and circles respectively. By switching to an ellipsoid humeral head, a small 2.3±0.3% (P < 0.001) increase in total coverage led to a 19.5±1.3%(P < 0.001) increase in cortical coverage. The overhang for fitted ellipses and circles was 1.7 ±0.7% and 3.8 ±0.8% respectively, defined as a percentage of the total enclosed area that exceeded the bounds of the humerus resections. Using circular heads results in 2.0 ±0.1% (P < 0.001) greater overhang. Upsizing until 100% resection coverage, the ellipse produced 5.4 ±3.5% (P < 0.001) less overhang than the circle. When upsizing the overhang increases less rapidly for the ellipsoid humeral head that the circular one (Figure 1). Full coverage for the head is achieved more rapidly when up-sizing with an ellipsoid head as well. Downsizing until 0% overhang, total coverage and cortical coverage were 7.5 ±2.8% (P < 0.001) and 7.9 ±8.2% (P = 0.01) greater for the ellipse, respectively. Cortical coverage exhibits a crossover point at −2.25% downsizing, where further downsizing led to the circular head providing more cortical coverage.

Reconstruction with ellipsoids can provide greater total resection and cortical coverage than circular humeral heads while avoiding excessive overhang. Elliptical head cortical coverage can be inferior when undersized. These initial findings suggest resection-matched humeral heads may yield benefits worth pursuing in the next generation of TSA implant design.

For any figures or tables, please contact the authors directly.

To evaluate the efficacy of using a novel button-suture construct in place of traditional screws to provide bone block fixation for the Latarjet procedure.

Four paired cadaveric shoulders (n=8) were denuded, with the exception of the conjoint tendon on the coracoid, and were potted. A 15% anterior glenoid bone defect was simulated. Right and left specimens were randomised into two groups: double-screw versus quadruple-button Latarjet reconstruction techniques. A uniaxial mechanical actuator loaded the Latarjet reconstructed glenoid articular surface via a 47mm diameter metallic hemisphere. Cyclic loading between 50–200N was applied to the glenoid at a rate of 1Hz for 1000 cycles. Testing was repeated three times for conjoint tendon loads of 0N, 10N and 20N. The relative positions of three points on the inferior, central and superior edges of the coracoid bone fragment were optically tracked with respect to a glenoid coordinate system throughout testing. Screw and button constructs were compared on the basis of maximum relative displacement at these points (RINF, RCENT, RSUP). Statistical significance was assessed using a paired-samples t-test in SPSS.

When conjoint tendon loading was not present the double screw and quadruple button constructs were not significantly (P>0.779) different (0N: RINF: 0.11 (0.05)mm vs. 0.12 (0.03)mm, RCENT: 0.12 (0.04)mm vs. 0.12 (0.03)mm, RSUP: 0.13 (0.04)mm vs. 0.12 (0.03)mm). Additionally, the double screw construct was not found to differ (P>0.062) from the quadruple button in terms of resultant coracoid displacement for all central and superior points, regardless of conjoint loading (10N: RCENT: 0.11 (0.03)mm vs. 0.19 (0.05)mm, RSUP: 0.11 (0.01)mm vs. 0.18 (0.04)mm; 20N: RCENT: 0.13 (0.01)mm vs. 0.30 (0.13)mm, RSUP: 0.13 (0.03)mm vs. 0.26 (0.14)mm). It was only for the inferior point with conjoint loading of 10N and 20N that the double screw construct began to produce significantly lower displacements than the quadruple button (10N: RINF: 0.11 (0.03)mm vs. 0.23 (0.05)mm, P=0.047; 20N: RINF: 0.12 (0.02)mm vs. 0.39 (0.15)mm, P=0.026).

The results of the screw and button constructs when conjoint tendon loading was absent suggest that the button may be a suitable substitute to the screw when the coracoid is used as a bone block. Due to the small resultant displacements (max: screw = 0.19mm, button = 0.52mm), it is suggested that buttons may also act as a substitute to screws for Latarjet procedures, provided conjoint tendon overloading is minimised during the post-operative graft healing period. These in-vitro results support the in-vivo results of Boileau et al (2015) that demonstrated the suture-button technique to be an excellent alternative to screw fixation Latarjet, with graft healing in 91% of their subjects.