Validity of the data - Assessment of bone after total knee arthroplasty

6 DISCUSSION

6.3 Validity of the data

The strengths of the knee DXA studies (subgroups I-III) were the rather long fol-low-up and also the study method used to measure BMD, which is considered to be precise, reproducible and safe. The limitations of the tibial periprosthetic BMD study (subgroup I) were the rather high number of dropouts during the 7-year fol-low-up, which is rather common in this kind of study population of elderly pa-tients, and the fact that there were some missing data in the BMD measurements.

The statistical method (mixed model) was selected to manage with the missing data.

In the alendronate study, there was a difference in both overall (from two to seven years) and functional (from four to seven years) scores between groups. This was due to unrelated reasons and comorbidities and cannot be attributed to the bisphosphonate treatment. The similarity between the knee evaluation scores, which only refer to the operated joint, support this conclusion. However, there is a chance of a selection bias. The alendronate study was also limited by a small initial cohort and by a high rate of loss to follow-up (although this can be expected in an elderly population). However, the post hoc power analysis suggests that the number of cases was sufficient.

The strength in the study of the third subgroup was the very few missing data.

The limitations include the small number of mild to moderate (Ahlbäck II) or very severe (Ahlbäck V) grade arthrosis patients, which made the analysis of correlation between BMD and osteoarthrosis severity uncertain. This could be expected, since the patients had also radiological indication for TKA surgery, thus limiting the number of mild to moderate arthrosis patients. In order to determine whether the BMD difference between the affected and contralateral side hips equalized, we would need to perform a longer follow-up than the one undertaken in this study.

The strength of the CBCT study (subgroup IV) is the comprehensive knee revi-sion population reflecting the real-life situation in revirevi-sion arthroplasties of the knee joint. The majority (17 out of 18) patients were operated on the day after the

CBCT scan, giving us a direct possibility to assess the current status of the joint and the prosthesis components (rotation, loosening, bone defects etc.). The limitations are the inability to compare the different indications for revision since the size of the study population was not large enough.

6.4 PERIPROSTHETIC MEDIUM-TERM TIBIAL BONE MINERAL CHANGES AFTER TOTAL KNEE ARTHROPLASTY: THE EF-FECT OF ALIGNMENT

Insertion of a prosthetic implant in the knee leads to remodeling of the bone as a result of the new loading conditions imposed by the implant. This can lead to bone loss around the implant in areas not subjected to loading (Sundfeldt, et al. 2006).

The patients were divided into preoperative varus and valgus groups according to their preoperative long-standing radiographs to study the effect of alignment change on the tibial BMD. The varus group was further divided into subgroups reflecting the postoperative radiological status (residual varus, neutral or overcor-rection to a valgus axis).

The varus and valgus groups differed in terms of group sizes and sex distribu-tion, and neither prospective matching nor blinding was performed between groups. A retrospective adjustment was carried out with background variables known to affect BMD. The mean baseline BMD values of tibial diaphyseal and lat-eral metaphyseal ROIs between the (preoperatively) varus and valgus-aligned knees resembled each other quite closely. On the other hand, the mean baseline BMD values of the medial metaphyseal ROI were higher in the varus group, even when the background variables were included in the analysis. These results are in accordance with findings in the literature (Li and Nilsson. 2000a).

The present medium-term follow-up study showed that the mean BMD in me-dial metaphyseal ROI was higher in (preoperatively) varus-aligned knees than in valgus knees. BMD decreased during the first year of the follow-up, thus slightly leveling the difference between the groups. At the 7-year follow-up, the mean me-dial BMD of the valgus group also decreased compared to baseline, reaching an almost similar decline to that which occurred in the varus group (12% and 13%, respectively), although the decline configurations were quite different during the early phase of follow-up. A previously published 1-year follow-up study of 69 sub-jects from this same patient population described similar early-phase results (Soin-invaara, et al. 2004b).

The medial metaphyseal BMD remained higher in the varus group than in the valgus group during the follow-up when the background variables were not in-cluded in the analyses, but the difference between the groups was not statistically significant from the 3-month to the 7-year measurements when they were included.

We therefore suggest that this leveling of the medial metaphyseal periprosthetic

changes and correlations between the predominant factors influencing BMD of the hips and contralateral knee until the 4-year follow-up. Studies with a similar cohort or follow-up are scarce (Ishii, et al. 2000, Kim, et al. 2014, Beaupre, et al. 2015).

The aim of the study on revision TKA patients examined by CBCT was to vali-date a relatively new, prototype-like technology in TKA component positioning and periprosthetic bone quality assessment. Therefore, the cohort size in this study was relatively small, but comparable with the only previously published study concerning CBCT evaluation of component rotations (Nardi, et al. 2017). To the author’s knowledge, there are no clinical in vivo studies assessing component loos-ening or osteolysis of components with a perioperative assessment verifying the CT findings, which highlights the relevance of the present study.

6.3 VALIDITY OF THE DATA

The statistical method (mixed model) was selected to manage with the missing data.

The strength in the study of the third subgroup was the very few missing data.

6.4 PERIPROSTHETIC MEDIUM-TERM TIBIAL BONE MINERAL CHANGES AFTER TOTAL KNEE ARTHROPLASTY: THE EF-FECT OF ALIGNMENT

We therefore suggest that this leveling of the medial metaphyseal periprosthetic

BMD between the groups after baseline measurements is an effect of the mechanical axis correction. The subgroup analysis of the (preoperatively) varus-aligned knee group also supports this hypothesis. Bone mineral distribution is affected directly by local mechanical stress and loading in patients with OA of the medial compart-ment of the knee. This was also the conclusion of Wada and co-workers (Wada, et al. 2001).

The short-term decrease in BMD in our study population in the medial metaph-yseal and diaphmetaph-yseal ROIs is in accordance with many previous studies (Petersen, et al. 1995, Li and Nilsson. 2000a, Lonner, et al. 2001, Soininvaara, et al. 2004b, Pe-tersen, et al. 2005, Abu-Rajab, et al. 2006, Saari, et al. 2007, Hernandez-Vaquero, et al. 2008, Munro, et al. 2010). The metabolic reaction of the bone to the operative trauma combined with the effect of postoperative immobilization may explain this decline. Li and Nilsson (Li and Nilsson. 2000a) found a recovery in BMD after 2 years of up, but ongoing bone loss is a more common finding during follow-up (Hvid, et al. 1988, Petersen, et al. 1995, Petersen, et al. 2005, Saari, et al. 2007).

Preoperative varus malalignment causes more load and high initial BMD in the medial tibial metaphysis. The postoperative loss of BMD in this preoperatively more loaded condyle that we found is in accordance with previous findings (Hvid, et al. 1988, Petersen, et al. 1995, Regner, et al. 1999). However, there are still few long-term results concerning changes in BMD. Levitz et al. (Levitz, et al. 1995) found a similar recovery in overall BMD at the 1-year measurement, as did Li and Nilsson (Li and Nilsson. 2000a) after 2 years, but surprisingly they found a 36.4%

decrease in overall periprosthetic tibial bone mineral density when 7 patients were followed up for 8 years after TKA. Saari et al. (Saari, et al. 2007) found an initial decline in periprosthetic BMD at the 1-year measurement, but BMD was similar between the measurements at 1, 2 and 5 years in all of the ROIs measured. The de-crease in BMD was notably greater than that found in our study: 18–26% in medial metaphyseal ROIs, 12–21% in lateral metaphyseal ROIs, and 5–20% in diaphyseal ROIs, except in the subgroup, where preoperatively varus-aligned knees were overcorrected to postoperative valgus (24% 7 years after TKA). These results re-semble the findings of Regner et al. (1999), who found a 26% decrease in the medial tibial condyle. Hernandez-Vaquero et al. found a decrease of 5.1% in periprosthetic bone below the tibial component 7 years after the surgery (Hernandez-Vaquero, et al. 2008). The decline in medial tibial metaphyseal periprosthetic bone was 9.1%

with cylindrical-type components and 6.1% with cruciform-type components. These declines more closely resemble our findings.

The subgroup analysis of the (preoperatively) varus-aligned knee group revealed dif-ferences in mean BMD values between the correctly re-aligned knees and the inac-curately re-aligned knees. The mean medial BMD of the knees with residual varus postoperatively did not decrease, possibly because the alignment was not properly corrected. In the postoperatively optimally aligned straight knees, the decrease was clearly detected, and in the postoperatively valgus-aligned knees, the decrease was

most prominent (up to 24% at 7 years). The decrease in BMD of the more loaded medial condyle could also be a desirable change concerning the survivorship of the prosthesis component. In a finite element analysis study, Taylor et al. (Taylor, et al.

1998) suggested that an increase in bone stress to the medial cancellous bone is re-lated to component migration, and it may contribute to a lower survival rate of the tibial component. This result highlights the importance of a straight postoperative mechanical axis and balanced bone stock after TKA.

During follow-up, we did not detect any changes in mean BMD values of the preoperatively more loaded lateral condyle in the preoperatively valgus aligned group. The minor, statistically not significant increase (5.2%) in the BMD of (pre-operatively) more unloaded medial condyle leveled at 6 months to 2 years. A simi-lar minor increase was also found by Petersen et al. (Petersen, et al. 1995). However, the BMD of this ROI did not decrease statistically significantly compared to mean baseline BMD values before the 7-year measurement. The preoperative malalign-ment (7.6 degrees of valgus) and the number of subjects were probably too low, thus leading to an unpowered study design for this group. Even so, the pattern of change was different from that seen with varus aligned knees.

The diaphyseal BMD remained unchanged after an initial drop within 3–6 months. Even the normal annual decline of 1% — or 8–9% over a decade previously reported in the literature (Bohr and Schaadt. 1987, Petersen, et al. 1995) — could not be detected. This might be a result of improved functional capability, which is clear-ly shown as improved mean AKS scores. In the varus group, there were no changes in BMD of the lateral metaphyseal ROI. The preserved diaphyseal and lateral periprosthetic BMD probably increases the stability and therefore also the longevity of the prosthesis component.

The influence of tibial insert design and the differences between cemented and uncemented implants have been studied in the past decade. Saari et al. (Saari, et al.

2007) did not detect any influence of tibial insert constraint on the postoperative bone remodeling over 5 years of follow-up. Munro et al. (Munro, et al. 2010) found no difference in changes in BMD between the rotating- and fixed-platform knees after a 2-year follow-up. Li and Nilsson (Li and Nilsson. 2000a) were unable to identify any differences in BMD between cemented and uncemented fixation dur-ing a 2-year follow-up. However, there are studies that support the contention that use of a long stem reduces proximal stresses and may result in proximal bone re-sorption (Taylor, et al. 1998, Lewis, et al. 1998, Lonner, et al. 2001).

The higher mean metaphyseal tibial (both medial and lateral ROIs) BMDs in fa-vor of the NexGen prosthesis compared to the other prosthesis models from base-line throughout the follow-up was an unexpected result. Similar differences be-tween periprosthetic femoral BMDs were not found in the study by Järvenpää et al.

(Järvenpää, et al. 2014). The sex distribution, proportion of knees with severe ar-throsis (Ahlbäck grade 4-5) and preoperative mean BMI values between the two main models (Duracon and NexGen) were equal, thus not explaining the difference.

with cylindrical-type components and 6.1% with cruciform-type components. These declines more closely resemble our findings.

In document Assessment of bone after total knee arthroplasty (sivua 56-60)