1 Introduction

Mechanical loosening continues to be one of the main causes for total knee replacement (TKR) failure.¹ Loss of bone stock around a TKR has been reported in several studies and has been attributed in part to stress shielding.^2–4 The major implication of this detriment to bone stock is that it could potentially promote loosening, migration, and prosthetic failure.^1,2,5 Bone resorption on the medial side of the tibia due to stress shielding has been reported to be between 3.2% and 71%.^3,4,6,7 Although knowledge is scarce, this phenomenon would be mainly related to implant design and preoperative varus misalignment corrections.^6,8–10 According to many authors, stiffer tibial bases have shown a higher incidence of bone resorption.^6,11,12 Scott et al.¹¹ demonstrated that the maximum stress transferred to the proximal tibial bone with all-polyethylene tibial trays is only one-third that of metal-backed components. While other authors have reported a higher incidence when using cobalt-chrome versus titanium trays.^6,13

In our practice, we routinely use either all polyethylene and titanium tibial trays. Compared to cobalt-chrome bases, titanium tibial trays have greater flexibility, so stress shielding should not be significant.^2,3,13 However, we have noticed this phenomenon in some of our patients. So far, few studies have evaluated bone resorption between titanium and polyethylene-only bases,⁶ and to our knowledge none have assessed the impact of this phenomenon after 10 years of follow-up regarding mechanical failure. Therefore, the main purpose of this study was to comparatively assess the incidence of bone resorption in patients undergoing TKR using titanium and polyethylene-only tibial trays and secondarily to investigate its relationship with mechanical loosening after 10 years of follow-up.

2 Material and methods

After reviewing our department's database, we conducted a retrospective study of all the primary total knee replacements performed in our department between January 2004 and December 2009, in whom we used full polyethylene tibial baseplates (PFC Sigma All-poly; Johnson & Johnson; Depuy, Warsaw. Ind. USA) and titanium tibial trays (PFC Sigma – fixed bearing-; Johnson & Johnson; Depuy, Warsaw. Ind. USA).

Inclusion criteria were adult patients (over 18 years old) in whom we performed a total knee replacement due to advanced osteoarthritis and completed a minimum follow-up of 10 years. Patients with previous conditions in the knee that could lead to alterations of bone structure (osteotomies, fracture sequelae, osteosynthesis) and infections were excluded.

All procedures were performed in a laminar flow operating room under hypotensive spinal anesthesia. Thirty minutes before skin incision, 1 g of intravenous cefazolin was administered, followed by two postoperative doses every 8 h. Tranexamic acid was administered and a hemostatic tourniquet was used in all cases. We used a mid-anterior approach with medial patellar arthrotomy. According to lower limb malalignment (varus or valgus), soft tissue balancing was performed. Next, tibial osteotomy was performed perpendicularly to the axis of the bone; a femoral cut was done with 5 to 7° of valgus. Both femoral and tibial components were fixed with low-viscosity bone cement. Low molecular weight heparin was used for thromboprophylaxis during the postoperative period (four weeks). All patients underwent the same rehabilitation protocol. This included isometric exercises and fully weight-bearing walking with the aid of a walker or cane from the first postoperative day for two weeks, and then using a cane for the following two weeks. Knee flexion-extension exercises were indicated from day 10, beginning with quadriceps strengthening exercises 20 days postoperatively. Clinical and radiographic assessment was carried out at 3 and 6 weeks; new visits were scheduled at 3, 6, and 12 months. After this period assessments were performed annually.

For the analysis we divided the patients into two groups according to prosthesis design: patients with titanium tibial trays (TiB) and patients with all polyethylene tibial trays (APB).

The Knee Society Score (KSS) was used to assess functional outcomes by comparing preoperative values with the ones recorded at the last visit over 10 years.¹⁴

Radiographic assessment was performed with anteroposterior (AP) and Lateral (L) views and was considered “suitable for evaluation” when both posterior femoral condyles couldn't be seen and if there was a 50% overlap between the lateral side of the tibia and fibular head.

Lower-limb alignment was registered according to anatomic (pre- and postoperative) axis of the femur and the tibia. After this, patients were classified into three subgroups: varus (over 0° of varus), neutral (0 to 6° of valgus), and valgus (over 7° of valgus).¹⁵Alignment of tibial component was assessed according to tibial axis and similarly, we considered three options: varus (over °0 of varus), neutral (0°), and valgus (over 0° of valgus).¹⁶ These measurements were performed manually with a goniometer using antero-posterior (AP) and lateral views of the knee.

We defined periprosthetic osteolysis (PPO) as any radiolucency or localized bone resorption around the components, that were absent on postoperative radiographs taken immediately after TKA.¹⁷

Proximal medial tibial bone resorption was assessed with Deen's classification system which consists in: Grade 1: Resorption <50% of the width of the tibial tray; Grade 2: >50% of the width of the tibial tray; Grade 3: Resorption beyond tibial keel.¹² We registered the time between knee arthroplasty and the appearance of bone resorption. After this, we followed its radiographic progression over time until the last control after 10-years of follow-up.

Radiographic assessment was carried out by two orthopedic surgeons trained in knee arthroplasty. The inter-rater agreement of measurements for femoro-tibial alignment, tibial component position, and Deen's classification system was assessed by using Pearson's coefficient for one hundred radiographs analyzed by each surgeon: Pearson's coefficient values of 0.89, 0.91, and 0.88 respectively were registered. Mechanical loosening and revision rate were also documented at the end of follow-up.

This study was conducted after obtaining the approval of our Institutional Ethics Committee (Protocol number 6687).

3 Results

Five hundred seventy-one TKR were performed during the study period, of which 405 were excluded (167 were not primary knee replacements, 144 had previous knee surgeries, 50 didn't have a complete or suitable radiographic record, and 44 did not complete the minimum follow-up). One hundred and sixty-six patients were available for complete evaluation; 86 (51.8%) comprised the TiB group and 80 (48.2%) the APB group. The demographic and clinical data of the series are described in Table 1.

We did not observe a statistically significant difference between both groups regarding preoperative characteristics.

3.2

3.2 Radiological assessment

3.2.1

3.2.1 Postoperative alignment of femorotibial axis and tibial component

Neutral axis was the most frequently observed in both groups (TiB: 95.3% and APB: 78.7%). Regarding tibial tray, we found a greater percentage of patients with a neutral position in both groups (TiB: 76.7% and APB: 62.5%). All postoperative measurements are described in Table 2.

Table 2 Comparison of postoperative measures of both groups.

	TiB group		APB group
	(n,%)	(mean, SD)	(n,%)	(mean, SD)
Postoperative femorotibial alignment
Varus	2 (2.3)	1.0° ± 0	4 (5.0)	2.5° ± 0.6
Neutral	82 (95.3)	3.2° ± 2.7	63 (78.7)	4.0° ± 1.8
Valgus	2 (2.3)	10° ± 0	13 (16.3)	8.6° ± 0.8
Tibial tray position
Varus	18 (20.9)	3.6° ± 1.8	23 (28.7)	2.0° ± 0.6
Neutral	66 (76.7)	0°	50 (62.5)	0°
Valgus	2 (2.3)	2° ± 0	7 (8.8)	1.8° ± 1.5

3.2.2

3.2.2 Bone resorption

3.2.2.1

3.2.2.1 APB group

We found one (1.25%) patient with bone resorption with a preoperative neutral axis (2° of valgus) that was modified to 7° of valgus after surgery. It was a grade I of the Deen classification and was detected after 36 months of arthroplasty without showing any signs of progression at 13.1 years of follow-up. No patient developed radiographic signs of osteolysis in this series (Fig. 1).

Close

Fig. 1 Typical case of the polyethylene-only tibial base group. 70-year-old male patient. a: Radiographic signs of osteoarthritis with varus malalignment of the right knee. b: Immediate postoperative radiograph with correction of the femoro-tibial axis. c: 12-year follow-up without medial tibial bone resorption.

Export to PPT

3.2.2.2

3.2.2.2 TiB group

In the TiB group, 21 (24.41%) cases with bone resorption were observed at a mean of 12.0 ± 4.6 months of follow-up. Grade 2 of Deen's classification was the most frequent type with an incidence of 57.14% (Fig. 2). Three cases (14.28%) progressed during follow-up over a median time of 12 (range 9–18) months. Two (9.52%) of them progressed from Deen's grade 2 to 3, and one (4.76%) from 1 to 3.

Close

Fig. 2 Typical case of the group treated with titanium tibial base, 68-year-old patient with left TKR. a: Radiographic signs of osteoarthritis with varus malalignment. b: Immediate postoperative radiograph with axis correction after knee arthroplasty. c: One-year follow-up where we observed grade 2 resorption on the medial aspect of the proximal tibia. d: 11 years, no further changes were observed.

Export to PPT

When comparing both groups, we observed a statistically significant difference in the TiB group regarding bone resorption (1.25% versus 24.41%; p < 0.001). (Fig. 3).

Close

Fig. 3 Percentage distribution of patients with bone resorption.

Export to PPT

We registered two (2.32%) cases with osteolysis under medial tibial plateau of the prosthesis, that was observed after 9 and 11 years respectively, but didn't progress during follow-up.

3.2.2.3

3.2.2.3 Preoperative alignment and bone resorption

In the TiB group, we observed that bone resorption showed a progressive increase the higher the preoperative varus alignment (p 0.12 - median 75-25), showing a moderate/strong correlation (Fig. 4 a-b).

Close

Fig. 4 a: Progressive increase of bone resorption in patients with higher degree of preoperative varus. 4 b: Increasement of preoperative varus regarding bone resorption.

Export to PPT

3.2.2.4

3.2.2.4 Postoperative correction of femoro-tibial alignment and bone resorption

When we analyzed the degrees of correction between pre- and postoperative alignment, we observed a significant difference (p = 0.009) between the patients who presented resorption (10.7 ± 6.9) and those who did not (6.9 ± 5.2).

3.2.2.5

3.2.2.5 TiB tibial tray position and bone resorption

Regarding the position of the tibial tray, a strong correlation between varus and bone resorption was observed, with a Spearman test coefficient of 0.94 (confidence interval wasn't able to determine due to the small sample size). However, this wasn't statistically significant (p = 0.10).

4 Discussion

The main findings of this study were that the rate of bone resorption of titanium tibial trays was significantly higher compared to all-polyethylene designs and that this resorption did not affect prosthetic survival after 10-year of follow-up.

Stress shielding around knee prostheses is a well-recognized phenomenon that remains under investigation.^18,19 The implant manufacturing process plays a central role in redistributing loads after the correction of the alignment and placement of the tibial component.^{11,13,18,20,21}

The significant difference in the metallic bases found in the present study was similar to that reported by Deen et al.,¹² who observed a higher percentage of 35% with a follow-up of 6.9 months. This difference (24.4% vs 35.0%) can be explained by the different elasticity modulus between the titanium (Ti) trays used in this study and the ones of chromo-cobalt (CrCo) analyzed by these authors.

This elasticity variation regarding prosthetic designs was explained by Martin et al.⁶ in a comparative series where they assessed TKR with CrCo, Ti, and APB. They reported a significant difference in bone resorption of CrCo implants over titanium and all-polyethylene implants, although they didn't find any difference between these last two types of prosthesis. This contrasts with the analysis of the present series. We believe that an explanation could be that the aforementioned authors evaluated the amount of bone affected by resorption while we assessed its incidence.

This study also found in the TiB group, that medial tibia resorption showed a moderate/strong correlation with preoperative varus alignment. This is consistent with the findings reported by Song et al.²⁰ who analyzed 96 patients that underwent TKR CrCo tibial trays, and observed that resorption under the tibial baseplate was statistically significant in patients with greater preoperative varus alignment. And, was also to that reported by Deen et al.¹² who described that 84% of the patients with resorption in their analysis had a preoperative varus alignment. This effect has been tried to explain in different ways. One theory is that mechanical transductors around the knee make the bone to adapt at repetitive stimuli, turning this signal into new bone formation (remodeling).^18,22 This is known as the biomolecular transduction of Wolf's Law.^18,22 When correcting lower limb alignment, physiological load transmission is reestablished in the proximal tibia. These events cause the medial compartment of the knee to receive bear less load, thus decreasing osteogenesis.^3,23 This could be related to the greater correction in pre- and postoperative femoro-tibial alignment found in this study between patients with and without resorption. Other possible explanations described in the literature are related to regional devascularization during arthroplasty in varus knees due to soft tissue release and balancing.^6,16,19 Another reason could be that the medial tibial plateau in varus knees is sclerotic, and after the tibial osteotomy, there are remaining areas generating focal increased pressure and unequal load distribution in the joint, which could promote bone resorption.⁴

Furthermore, in the present study, medial tibia resorption was strongly correlated with varus misplacement of the tibial component, although this was not statistically significant. We believe that this could be due to the small sample size. To our knowledge, this finding has not been previously reported in clinical trials. Zhang et al.⁸ in a three-dimensional finite element model observed that varus of the tibial component (either metal or polyethylene-only) would potentially increase the risk of stress shielding in the lateral side of the tibia. The study findings do not agree with this paper since in the present series bone resorption was observed in the medial aspect of the tibia, and in addition, it presented 23 cases with an all-polyethylene component with varus alignment, none of which showed resorption after more than 10 years of follow-up. This leads us to think that resorption in the medial tibia, should be explained by other reasons and not be exclusively associated with stress shielding, which should be adequately studied in properly designed studies.

Finally, it should be stated that the development of stress shielding did not affect long-term functional outcomes, nor did it affect prosthetic survival. At the end of the follow-up of this study, we haven't been able to observe clinical consequences, patients with a poor functional capacity or any mechanical loosening. Even in the 3 (14.2%) patients that developed bone resorption and progressed after one year from its appearance, we did not observe further changes till the end of the study. This may or may not have clinical relevance in the future.

To our knowledge, this is the first publication to report on 10 years of radiographic evaluation and its consequences in relation to stress shielding between titanium and polyethylene-only trays, filling the gap in current knowledge on this topic.

The limitations of this study are those of a retrospective study. The small number of patients included might be the cause for the lack of statistically significant relevance in the variables assessed. In addition, two orthopedic surgeons performed radiographic analysis without using an exact method for evaluation; this could hide some grade of precision bias. The main strength we can describe is that all of the patients share the same preoperative diagnosis and were operated on with the same surgical technique. Assessed cohorts were comparable regarding preoperative variables and the implants used are well-known devices in knee arthroplasty. Lastly, follow-up of the series allowed us to estimate stress-shielding consequences regarding long-term prosthesis stability.

5 Conclusion

Titanium tibial bases in TKAs showed a significantly higher incidence of medial tibia resorption compared to all polyethylene bases. Our results suggest that bone resorption does not influence long-term mechanical loosening. We will continue to evaluate the series for future implications.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics approval

All procedures performed in this study, which involved the use of data from human participants, were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments. The Ethics committee of British Hospital of Buenos Aires approved this study (Nº6687).

Author contributions

Germán Garabano: Conceptualization, Formal analysis, Investigation, Methodology, Project administration, Writing-original draft – review & edition. Joaquin Rodriguez: Data curation, Formal analysis, Software, Validation Leonel Perez Alamino: Data curation, software, Investigation, Methodology. Cesar Angel Pesciallo: Validation, Writing - review & editing. Hernán del Sel: Supervition, Writing - review & editing Fernando Lopreite: Supervition, Writing - review & editing.

Institutional ethical committee approval

All procedures performed in this study, which involved the use of data from human participants, were in accordance with the ethical standards of the 1964 Helsinki Declaration and its later amendments. The Ethics committee of our institution approved this study (Nº6687).