Introduction

Background

Revision total knee arthroplasty (RTKA) number is expected to increase as primary total knee arthroplasty surgeries are increasing annually, with more individuals at risk of failure each year [16]. RTKA is a more complex procedure and results may not be as satisfying as with primary TKA. Bone loss management in RTKA is challenging. Proper preoperative planning and a thorough evaluation of the patient are required. Bone defects must be addressed to restore the joint line and provide stability to the revision implant [48].

Rationale

Bone classification systems for RTKA are important since they help the surgeon in making choices during the surgery and if the system is efficient it can help with arranging the right type of implant and the right type of bone substitution. Classification guides treatment, surgical planning and choice of the implant in RTKA. It is being used to group patients for clinical research, to educate the patient regarding postoperative rehabilitation and for prognostic purposes. Studies reporting outcomes of various augments used to tackle bone loss in RTKA should use a common classification to allow comparisons and form guidelines.

Purposes

(1) To review different classification systems proposed for bone loss in RTKA. (2) Which is the most commonly used classification in literature for reporting a bone loss in RTKA. (3) How reliable are these various classifications in clinical practice. To identify the limitations of these classifications and make surgeons enlighten on which classification system to be used.

Materials and methods

The following databases were searched to locate relevant research: Pubmed/Medline, CINAHL, EMBASE, Scopus, Cochrane databases, Google Scholar and Web of science. The grey literature was also searched via the website http://www.opengrey.eu.

Searches were limited to the English language, published between 1st January 1980 and 31st July 2021. Pubmed search was done using Mesh terms—Arthroplasty, Replacement, Knee [Mesh] AND Reoperation [Mesh] AND Bone defect AND classification [Mesh]. The search terms used were: Anderson Orthopaedic Research Institute Classification (AORI) Classification*, UPenn system*, Bone defects*, Bone loss*, Revision total knee replacement*, Revision total knee arthroplasty*, validation studies*, Reliability studies*, Modified AORI Classification*, Limitation of classification*, reproducibility*, AORI agreement*, interobserver reliability*, intraobserver reliability*, kappa coefficient* (Appendix 1).

These were also used in inappropriate combinations. The search strategy was guided by the PICOS principles.

  • P (population)—patient undergoing revision total knee replacement/arthroplasty.

  • I (intervention)—studies evaluating validation of AORI classification.

  • C (comparison)—studies evaluating validation of any other classification system of bone defect.

  • O (outcomes)—kappa coefficient of inter- and intraobserver reliability, coincidence and other statistical parameters to assess the reliability of a classification system.

  • S (setting)—hospitals and operation theatres.

Inclusion criteria

All studies developing or utilising a classification tool were eligible for inclusion which was published between Jan 1980 and July 2021 in the English language. Four independent reviewers used consensus to determine whether articles met the criteria of “developing a classification tool” or “utilising a classification tool”. We included studies under three groups as following;

  1. 1.

    All studies investigating the reliability of the classification tools, either in isolation (a single classification tool) or head-to-head (multiple classification tools utilising the same data set), were eligible for inclusion in the review.

  2. 2.

    Studies describing a classification system were included.

  3. 3.

    Studies classifying bone defects using the AORI classification system were included.

Exclusion criteria

Studies investigating bone defects in RTKA, but not reporting its classification or reliability statistics were excluded from the review. Biomechanical studies having no patient data were also excluded.

Selection process

Titles and abstracts were screened for relevance by all the authors. If necessary, the complete article was reviewed to reach a decision. We also reviewed the references of papers to locate additional studies. PRISMA flowchart is shown in Fig. 1. The following data were extracted from the studies: type of study, year of publication, the total number of subjects, Classification system assessed, Statistical parameters of reliability assessment of classification system, Subtypes as per AORI classification. PRISMA checklist is shown in (Appendix 2).

Fig. 1
figure 1

PRISMA flowchart of the study

Full size image

Quality assessment process

Three authors (1, 2, and 3) and institutional statistician conducted quality assessments of the studies independently before crosschecking for concordance using the COSMIN checklist [63]. Studies were graded according to these criteria (Table 1).

Table 1 Methodological quality rating (according to COSMIN checklist) of reliability studies
Full size table

Results

Regarding review all the classification systems proposed for bone loss in RTKA

We identified 16 classification systems proposed for describing bone defects in RTKA from 1989 to 2020 (Table 2). Few of them have been published in only textbooks or meetings [6, 26, 31, 50]. Most of these classifications has not been used by any other author to describe bone defects in RTKA. The SOFCOT classification [18] was described in French and the one proposed by Reichel et al. [77] was in German. AORI classification is the only one that can be used postoperatively as well. Some of these classifications gives guides for treatment [6, 24, 26, 34, 45, 48, 50, 73, 79, 82], while other just classifies the bone defects.

Table 2 Developed classification tools identified through literature search
Full size table

Regarding the most commonly used classification in literature for reporting a bone loss in RTKA

Our literature search found 54 studies using AORI classification to classify bone defects in their series (Table 3). AORI classification is the most common classification used in literature to report bone defects in RTKA (Fig. 2). Ten out of 54 studies do not report the AORI classification properly. Some studies have not separately reported femoral and tibial bone loss and some studies have not reported subtypes adequately [4, 8, 12, 17, 30, 38, 46, 70, 81, 86]. AORI classification subtypes data from 44 studies were pooled (2869 RTKA), stating tibial bone defects (2165/2869) are more common than femoral bone defects (1769/2869) (Fig. 3). Type 2B is the most common tibial defect followed by Type 2A defect. Type F2B is the most common femoral defect followed by type F2A defect (Fig. 4).

Table 3 Studies using AORI classification to classify bone defects in their case series
Full size table
Fig. 2
figure 2

Adopted from Engh GA [34]

Hand-made diagrammatic representation of the AORI classification of bone defects in RTKR.

Full size image
Fig. 3
figure 3

Frequency of femoral and tibial bone defects in RTKR based on 44 studies using AORI classification system

Full size image
Fig. 4
figure 4

Frequency of bone defects in RTKR according to AORI classification subtypes, based on 44 studies

Full size image

Regarding the reliability of these classification systems in clinical practice

We found six studies reporting validation of various classifications of bone defect in RTKA (Table 4). Four studies reported reliability of AORI classification, two studies reported reliability of UPenn system, two studies reporting a new classification system with reliability statistics of their classification. The average kappa value for AORI classification for femoral bone loss was 0.38 (0.27–0.50) and 0.76 (0.63–1) for tibial bone loss assessment. As per the Landis and Koch scale (Slight 0–0.2, Fair 0.21–0.4, Moderate 0.41–0.6, Substantial 0.61–0.8, Almost perfect > 0.8) the reliability AORI Classification is fair for femoral bone loss and substantial for tibial bone loss.

Table 4 Studies assessing reliability in bone defect classifications in RTKR
Full size table

Discussion

Bone defect management is one of the determinants of successful RTKA. Proper preoperative and intraoperative evaluation is required for managing bone defects. A classification system can help us to give a more systematic approach to managing bone defects. Among various proposed classifications, AORI classifications stand to be the most commonly used classification. Although there are various limitations of this classification system. This systematic review article is descriptive in nature.

Regarding review all the classification systems proposed for bone loss in RTKA

In 1989, Dorr [31] classified tibial bone defects as central or peripheral. In 1992, Bargar and Gross [6] classified tibial and femoral defects into segmental, cavitary, intercalary and discontinuity types. This classification was based on measuring bone loss using reference points (epicondyles in femur, tibial tubercle and top of fibula for tibia) on preoperative radiographs. In 1993, Insall [50] classified bone defects into contained and uncontained. Visual description of the intraoperative shape of bone defects was used in this classification. In 1997, Engh [34] gave AORI (Anderson Orthopaedic Research Institute) classification in which tibial and femoral defects were classified based on location and severity of a bone loss. UPenn system (University of Pennsylvania classification) [24] was based on using grids on preoperative AP and lateral view of femur and tibia to quantify the bone loss. It is complex and not used at many centres. In 2007, Huff and Sculco [48] classified based on appearance of defects into cystic, epiphyseal, cavitary and segmental. Rosso et al. proposed modified AORI classification by adding intraoperative bone quality as good (G), sclerotic (S) and osteoporotic (O). They also divided treatment guidelines according to bone quality [79].

Stambough et al. found that AORI classification has limitations in describing contained and uncontained defects, they proposed a modification to the AORI classification which subdivides tibial defects (T2a/T3) into contained (T2-AC/T3-AC) and uncontained (T2-AU/T3-AU) defects [82]. The Stambough–Nunley modification also includes modern management strategies with augments and cones, which are deficient in AORI classification.

AORI classification only partially quantifies the metaphyseal area and does not quantify the diaphyseal bone loss. Belt et al. described a classification system based on three-zone (epiphysis, metaphysis, diaphysis) graded on the severity of the bone defect. They reported the classification as less reliable in the epiphyseal area as compared with metaphysis and diaphysis. The addition of CT scans also did not improve the reliability [11].

Regarding the most commonly used classification in literature for reporting a bone loss in RTKA

Though AORI classification is simple to use and gives an idea of the severity of bone loss, subjectively as there are no specific physical instruments or scale for quantification of bone loss on preoperative radiographs.

Bone destruction can be hidden on the lateral radiographs if it’s not a "true" lateral view of the knee or the dense trabecular bone in an uninvolved condyle hides an osteolytic lesion in the adjacent condyle, as the osteolytic lesion is still aggressive and has not developed sclerotic border. Only AP and lateral radiographs can miss lytic areas, in such cases, an oblique radiograph or fluoroscopy can be used. Oblique radiographs can help with the early recognition of retro femoral osteolysis about a posterior-stabilised implant [67, 49].

AORI classification typically underestimates bone loss on preoperative evaluation i.e. classifying a Type 2 defect as a Type 1, which was reported in 14% of tibial and 17% of femoral assessments [65]. Although the difference in evaluations usually consisted of disagreement in type 1 cases, one should be aware of this potential disparity. This concern becomes heightened when templating for a type 2 defect preoperatively that turns out to be a type 3 intraoperatively, in such cases other imaging modalities like CT scan may be indicated.

AORI classification is somewhat arbitrary, and the distinction between type 1 and type 2a may be subjective with high interobserver disagreement [57]. Patellar defects are common in RTKA, poor reconstruction may lead to patellofemoral instability or even dislocation, AORI classification does not classify patellar bone defects.

AORI classification does not help much in assessing postoperative radiographic outcomes after RTKA. The Knee Society Roentgenographic Evaluation and Scoring System [36] or Modified Knee Society Roentgenographic Evaluation and Scoring System [62] provide a better method of assessing pre- and postoperative alignment and radiolucency [33]. Huang et al. reported interobserver reliability of Knee Society Roentgenographic Evaluation and Scoring System with kappa = 0.926 (P < 0.001; 95% confidence interval) [47].

Nowadays, scientific articles, clinical trials including multicentric ones and treatment guideline protocols on knee revision surgery use AORI classification for bone loss. For example, Bush et al. [21] reported cement with screws can be used for small bone defects (AORI type 1, 5–10 mm) and impaction bone grafting can be done in AORI type 1 to type 3 tibial and femoral defects. Park et al. [71] reported that cause of revision, change in joint line height, Insall–Salvati ratio, AORI classification should be considered while deciding for a constrained implant. Vasso et al. [87] gave an algorithm for choice of constraint in knee revision surgery based on the integrity of peripheral ligaments and bone loss according to AORI classification.

The three zonal fixation concept is also a good concept which advised at least two zones of fixation in RTKA, zones available for fixation can be assessed preoperatively and avoid early implant failure [64].

Lee et al. found AORI type T2a defects were more common in patients with loosening than among those without loosening (p = 0.010), while no such difference was observed for the femoral side [57]. Various studies assessing the stability of reconstruction techniques (cemented step augment, wedge augment and stem) using finite element models use AORI classification to report their results [5, 27, 39]. Castelli et al. [22] found that there was no change in AORI type or shape of bone loss after using a cement spacer in a two-stage treatment of infected TKA.

Regarding the reliability of these classification systems in clinical practice

Studies evaluating the reliability of the AORI classification system does not give a firm conclusion. Study by Pecora et al. was done with six orthopaedic surgeons using AORI classification on preoperative radiographs and found a moderate correlation of classification. It gives interobserver correlation as the frequency of coincidence [72]. The study by Mulhall et al. [65] and Murphy et al. [66] tested AORI classification preoperatively on radiographs and intraoperatively by the surgeon. Which makes it more liable to change because of further bone loss while removing implants. Still, there was substantial reliability for tibial classification and fair to moderate reliability for femoral classification (as per Landis and Koch scale).

UPenn system gives a quantitative evaluation of bone loss and it also has moderate reliability for femoral bone loss and substantial reliability for tibial bone loss assessment.

Three-zone bone defect classification by Belt et al. [11] showed almost perfect inter-and intraobserver reliability in metaphysis and diaphysis, but moderate reliability in the epiphyseal area. RKCC classification [73] has also shown good reliability. Since both of these are new classifications, they need to be further studied in clinical use. Further studies are required which evaluate the interobserver reliability and intraobserver reproducibility of classifications (AORI and UPenn system) used on preoperative radiographs to assess bone loss.

Limitations of the study

We did an extensive literature search still the studies evaluating the reliability of classifications for bone loss in RTKA were limited. Also, the data pooling of all studies was not possible because of high variability in reporting outcomes of reliability studies.

Conclusions

AORI classification is the most widely used classification for bone defects in RTKA. The reliability of AORI Classification is fair for femoral bone loss and substantial for tibial bone loss. AORI classification when used on preoperative radiographs should be used with caution for femoral defects as there are high chances of missing defects. Although, it can be used for tibial bone loss assessment.