Three-dimensional Acetabular Bone Defect Classification System Aided with Rapid Prototyping Model and Reliability and Validity Test


 Background: Accurately assessing acetabular defects and designing precise and feasible surgical plans are important before hip revision arthroplasty. With the development of three-dimensional printing, rapid prototyping is a novel technique used to print isometric physical object models. We aimed to propose a three-dimensional acetabular bone defect classification system aided with rapid prototyping and evaluated its reliability and validity.Methods: We reviewed 104 consecutive patients who underwent hip revision arthroplasty in our department between January 2014 and December 2019. Forty five of them had rapid prototyping and were included for reliability and validity test. Three doctors retrospectively evaluated bone defects of these 45 patients with this classification and made surgical plans, and repeated it after 2 weeks. The intra- and inter-observer reliability and the validity to surgical records were assessed using intraclass correlation coefficient or Kappa correlation coefficient.Results: The reliability and validity for classification results were high. The mean initial intraclass correlation coefficient for inter-observer reliability was 0.947, which increased to 0.972 when texted second time. As for inter-observer reliability, it ranged from 0.958 to 0.980. The validity showed high Kappa correlation coefficient of 0.951 to 0.967. When considering detailed surgical plans, the reliability and validity were also high with intraclass correlation coefficient and Kappa correlation coefficient all over 0.9.Conclusions: This three-dimensional acetabulum defect classification was of high reliability and convincing validity. With this classification and objective rapid prototyping models, accurate bone defect assessment and reliable surgical plans were achieved. This classification aided with rapid prototyping could serve as a promising tool for surgeons for preoperative evaluation.


Introduction
With the amount of hip revision arthroplasty increasing (1)(2)(3), acetabular bone defects have become a main challenge (4). Accurately assessing defects and designing precise and feasible surgical plans are critical for successful surgeries (5).
There have been several acetabular defect classi cations, including classi cations described by Paprosky(6), the American Academy of Orthopaedic Surgeons(AAOS)(7), Engh(8) and Gross(9). However, the classi cation by Paprosky, the AAOS or Gross was reported to have only poor to moderate reliability (10). The Engh classi cation was a simpli ed version of the AAOS classi cation, but it was still of low reliability or validity (11). Although the Paprosky classi cation was useful for clinical practice (12), information was limited (13). Besides, current classi cations were more signi cant in evaluating simple acetabular defects than complex defects (14). Few classi cation systems can accurately guide surgical plans at present (11).
One reason for this is that most classi cations used currently were proposed in the 1990s and limited by radiological techniques at that time. They were mainly depended on two-dimensional (2D) X-rays which only provided general anatomical clues (14)(15)(16)(17) and did not represent three-dimensional (3D) structures accurately. With the development of 3D printing technology, rapid prototyping (RP) is used to convert standard 3D-computed tomography (CT) images into an isometric physical model in which physicians can accurately obtain key information which might be di cult to obtain through traditional images alone and can intuitively simulate various surgical plans and design implants(18). RP is proved to have more advantages in assessing bone defects and designing surgical plans in complex anatomical areas and revision cases compared to X-rays or CT scans(18-22).
Our joint reconstruction department has used RP for pre-operative evaluation and surgical planning for revision patients with acetabular defects. In this study, we proposed our 3D acetabulum defect classi cation system aided with RP and the corresponding surgical approaches, and tested the reliability and validity.

Method
Patients Enrolment and RP model building This study was approved by the ethics committee of our hospital. We reviewed totally 104 consecutive patients who underwent hip revision arthroplasty in our department from January 2014 to December 2019 for aseptic acetabular prosthesis loosening or osteolysis. After adimitted to hospital, anteroposterior and lateral radiographs (Siemens, RAX, Germany) of the affected hip were taken. If acetabulum defects were detected by X-rays, CT scans(Siemens, SOMATOM De nition Flash, Germany) were performed for more speci c examination, covering the bilateral anterior iliac spine and the posterior borders of the medial and lateral condyles with 0.5-mm interspacing thickness. If severe defects which might hamper commercial prosthesis placing or affect initial stability were detected on CT and augments might be needed in operation (22,23), RP was printed. 45 of these 104 patients were diagnosed as having severe defects and RP models for them were prepared before operation for further preoperative planning. The surgery was performed by the same group of experienced quali ed surgeons. Intraoperative nding and surgical treatments adopted were taken as gold standard for comparison. Surgery were all successful and follow-up results were good to date. These 45 patients were included in this study for further reliability and validity test.
RP is a technique used to convert CT into an isometric physical object model using 3D printing(18). For patients who needed RP, CT scans were performed. The CT results were imported into Mimics (version 19.0, Materialise, Belgium) to rebuild CT model. RP was made by selective laser sintering using polylactic acid through 3D printing technology (RS4500 Stereolithography, Liantai, China) based on CT model. The resolution of the RP model is 0.1mm and it takes 24 hours for painting and costs approximately $470 or $780 for the hemi or whole pelvis, respectively.

Classi cation system
Based on our experience of dealing such patients, we proposed our 3D acetabulum defect classi cation system as follows: Type I, no obvious or only minor acetabular defects. The rotational and vertical primary stability of the cup can be provided by host bone.
Type II, there is effective bone mass in the antero-superior acetabulum, the ischial ramus and the pubic ramus, while defects exist in the stress-bearing posterosuperior acetabulum. The host bone only provides rotational stability for the cup.
Type III, defects in the anterosuperior acetabulum, the ischial ramus or the pubic ramus. Both rotational and vertical primary stabilities are lost.
Type IV, sever destruction of acetabular structures with high risk of pelvic discontinuity.

Corresponding surgical plans
The surgical plans below are solutions which might be useful based on our experience, but not the only reconstruction method (Figure 1).
For Type I, an intact acetabular ring can be obtained by slight drilling. A commercial cup is used ( Figure   2).
For Type II, the rotational stability of the cup can be provided by a three-point xation distributed over 180 degrees ( Figure 3). Depending on bone defects and the vertical stability, surgical plans are divided into three detailed situations: 1) With no obvious defects in the stress-bearing posterosuperior acetabulum, the rotational and vertical primary stability can be obtained. A commercial cup is used ( Figure 4).
2) With cavity defects in the stress-bearing acetabulum, the vertical stability can be provided by posterosuperiorly placed augments. Since commercial augments are enough to achieve effective xation in cavity defects, it can be treated by a commercial cup with commercial augments ( Figure 5).
3) With un-contained defects in the stress-bearing region, the posterosuperior acetabulum is damaged into a plain wall. The shearing force will be quite large only using commercial augments. Therefore, customised augments with xed hooks or a buttress plate is necessary to avoid large shearing force ( Figure 6).
Type III, when the anterosuperior acetabulum or the ischial ramus is defected, a buttress plate or a customised augment is used to repair it. After that, cup is xed by an over 180 degrees three-point xation together with the pubic ramus. A cup-cage or a customised cage can also be used as an alternative ( Figure 7). When more than two of these three structures are defected, a three-point xation is lost. Hence, a cup-cage or customised Cage is required (Figure 8).
Type IV, a customised hemi-pelvic prosthesis is required ( Figure 9).

Reliability and validity test and data analysis
Three experienced surgeons quali ed for joint revision were asked to use the above classi cation to evaluate bone defects and make surgical plans independently and retrospectively for all enrolled 45 patients. The surgeon had access to X-ray, CT and RP. The detailed information related to the patient was concealed, except for an identifying number. The classi cation results and corresponding surgical plans were recorded. This process was repeated after 2 weeks with the patient sequence to be evaluated disrupted. Classi cation results and corresponding surgical plans were assessed for reliability within and between these surgeons and for validity compared to previous surgical records. Statistical analyses were performed by SPSS (version 26.0, IBM, USA). Intraclass correlation coe cient (ICC) was used to re ect the inter and intraobserver reliability and Kappa (κ) correlation coe cient was to re ect the validity. The mean value and 0.95 con dence interval were given.

Results
According to this classi cation, 30 of these patients were classi ed as Type II, 9 patients as Type III and 6 patients as Type IV by refering to surgical records.
The ICC and κ values for the reliability and validity test for the classi cation results were high (Table   1&2). The mean initial inter-observer ICC was 0.947, which increased to 0.972 when texted second time. Regarding intra-observer ICC, the values for three surgeons ranged from 0.958 to 0.980. The mean κ value for validity to surgical records was 0.951 to 0.967. When considering detailed surgical plans, high ICC and κ values were maintained (Table 1&3). The mean initial inter-observer ICC was 0.960 and increased to 0.968 when tested again. The intra-observer ICC values for three surgeons were 0.988, 0.963 and 0.987, respectively. The mean κ value for validity was also high, reaching over 0.922. This indicated that the classi cation results and surgical plans were of good intra-and inter-observer reliability and high validity compared to previous surgical records.

Discussion
Accurately assessing acetabular defects and making surgical plans accordingly are critical to ensure successful surgeries (5). However, current acetabular defect classi cations have kinds of disadvantages(8, 10, 11, 24-28). Campbell (10) assessed the reliability of three major acetabular classi cations described by Paprosky(6), the AAOS(7) and Gross (9). The innovator group reached only moderate range of intraobserver agreement and for the non-innovator group, the intraobserver and interobserver agreement was even worse for all 3 classi cations assessed. Gozzard (25) reported Paprosky classi cation system achieved moderate to good levels of agreement and bone stock loss classi cation systems were inconsistent and unreliable, which prevented realistic comparison of results within or between centers. Johanson(11) reviewed six acetabular defect classi cation systems and concluded that only one of them demonstrated required reliability and validity for a standardised grading system. Most current classi cations only roughly guide surgical plans which need to be decided by speci c perioperative situation.(6, 11, 13, 14) Campbell (10) claimed that the Paprosky system should be "considered only as a general guide" for its poor reliability. As a result, the degree of bone defects found during surgery may exceed surgeons' expectations. However, adjusting surgical plan intra-operatively is di cult and insu cient preparation may lead to surgical failure and worse post-operative outcomes. In addition, current classi cations have greater signi cance in evaluating relatively simple defects than complex cases (14). Ghanem (13) reported 16.37% of cases had further defects found intraoperatively than the preoperative plan and he suggested a practical, reproducible and valid classi cation system for preoperative planning.
One reason for this is that previous classi cations were mostly based on traditional radiological techniques. It is di cult to realize 3D structures of bone defects through X-rays or CT model on 2D computer screens and information is always incomplete (15)(16)(17). With 3D printing technology(18-22), 3D reconstructed RP displays 3D anatomy structure more intuitively and provides key information di cult to obtain from 2D images. Thus, RP aids surgeons to understand bone defects more intuitively and formulate surgical plans more comprehensively, which reduces errors and improves postoperative outcomes compared to x-rays or CT model (22).
For this reason, we proposed our 3D acetabulum defect classi cation system aided with RP. Since Type I in this classi cation was little defected and it was easy to distinguish from X-rays or CT, it was unnecessary to print RP models for such patients or take Type I cases into reliability and validity test.
This study showed both classi cation results and surgical plans were of good inter and intraobserver reliability and high validity compared to surgical records. It suggested our 3D classi cation helped classifying acetabular defects preoperatively and guiding reliable surgical plans. There might be several reasons for this. First, comprehensive information was accessible. In relative complex cases, RP presented bone defects and the rest bone mass through visible 3D models. Hence, demand on radiograph reading was less and no information was lost for evaluation. It became easier for surgeons to realise the real 3D structure of defected acetabulum. As a result, three surgeons made classi cations and surgical plans of high intraobserver reliability. Second, anatomy structures displayed by RP was so intuitive that a surgeon gave a closing classi cation at the second time after the rst impression, which explained high interobserver reliability. Third, evaluations were based on objective RP model, which could be utilised for surgery simulation to test plans' feasibility and re ect intraoperative situations preoperatively. Thus, classi cation results and surgical plans were of high validity. It was consistent with previous reports. Preoperative planning with RP before hip revision improved clinical results(18-22). Involving RP improved diagnosis accuracy in discovering fractures in complex anatomical areas, such as obscure pelvis fracture (29,30).
In this classi cation, the leading factor to be considered is how to obtain the primary stability of a cementless cup, including rotational and vertical stability. Primary stability can be obtained either by effective three-point xation distributed over 180 degrees or by enough screws tting into wall.
We chose the anterosuperior acetabulum, the ischial ramus and the pubic ramus for a three-point xation because other structures around the acetabulum are relatively thin and often destroyed in severely defected cases. Despite some bone mass remains, this kind of bone is of high density due to osteosclerosis. It is brittle and unable to provide effective holding. The three areas mentioned above have larger interface and thicker bone stock. The remaining bone will still be su cient for holding even if osteolysis occurs around. Although the ischial ramus and anterosuperior acetabulum have a relatively large bone stock and some surgeons even believe the cup can be held with only these two points, it should be avoided as much as possible to our opinion. This is exactly the difference between type III and type II. In some type III cases, defects in the anterosuperior acetabulum or the ischial ramus can still be reconstructed with a buttress plate or augments and a three-point xation can be obtained combining other two places. If more than one place needs reconstruction, the above method might not provide su cient holding and stable three-point xation is important (13).
We classi ed cases with internal acetabular wall defected but an intact acetabular ring (Paprosky type 2C) into type II because when the remaining bone is enough to provide a three-point xation, it is still feasible to hold the cup with augments or screws.
If type II has defects in the stress-bearing posterosuperior acetabulum, we believe the direct supporting between bone and prosthesis is the basic principle for achieving reliable long-term stability through bone remodelling. When structural or compression bone grafting is used, the progress of bone remodelling is so long that weight-bearing on the interface between the prosthesis and graft is inevitable. As a result, the prosthesis gets loosening quickly owing to micromotion. Our cases found that when there was no direct contact between the prosthesis and host bone but through allografts, although most allografts were well reconstructed, fatigue fracture occurred in the connection part to the cage due to micromotion ( Figure  10). It is possible that the remoulding progress was so long that the grafted bone could not provide su cient support to the prosthesis in the weight-bearing area although the grafted bone was well remodelded nally.
With cavity or uncontained defects in the weightbearing posterosuperior acetabulum as type II, augments are needed. When bone defects are not severe and the host bone thickness is still su cient, conventional commercial augments can repair the defects and cup can be xed by remaining acetabulum wall or screws. When uncontained defects or severe defects exist, stable xation cannot be achieved. Despite some bone mass in adjacent ilium in some cases, it is di cult to directly x an augment on to it. Customed augment xed wings or a buttress plate is required.
We noticed few patients with defects only in the anterosuperior acetabulum in clinical practice. It is possibly because rare force passes the anterosuperior. Anterosuperior defects always occur together with posterosuperior defects, which result in defects in the superior acetabulum (type III). When severe osteolysis only occurs in either the superior acetabulum or the ischial ramus, a buttress plate or a customed augment with a wing can be used to reconstruct defected places to hold the cup. When more than two of these places are defected, the primary vertical and rotational stabilities cannot be obtained using conventional methods. Cup-cage or customised prostheses is recommended. At present, most commercial Cup-Cage is used when there are no serious bone defects at the bottom acetabulum. Cup-Cage can be xed by an obturator hock into the obturator foramen or a hook bound to the upper edge of the obturator. Customised cages can be used for defects in many areas when rotational stability is di cult to obtain.
There are some disadvantages in this study. First, the original size of each patient's pelvis and remaining bone is different and it is di cult to classify quantitatively. However, only with this qualitative 3D classi cation, professional surgeons quali ed for joint revision can already distinguish the extent of bone defect and design feasible surgical plans, which already accomplishes the goal of surgical planning. Second, the accuracy of showing effective bone mass by RP needs further improvements(18). However, current accuracy already meets the needs for most clinical applications. Our recommended surgical strategies were feasible, but not the only method. Other experienced surgeons might have their own understandings and corresponding reconstruction methods.

Conclusion
This 3D acetabulum defect classi cation is of high reliability and convincing validity. With this classi cation and objective RP model, surgeons evaluate bone defects intuitively and make more accurate surgical plans. This classi cation aided with RP model could serve as a promising tool for surgeons for preoperative evaluation.

Declarations
Ethics approval and consent to participate        A. A 68-year-old male was diagnosed as aseptic loosening 20 years after THA. B. RP showed that the entire superior acetabulum was defected while the ischial ramus and the pubic ramus was intact. The remaining bone was di cult to support the rotational and vertical stability of the cup. An augment was designed to rebuild the superior acetabulum. The vertical stability could be achieved and the rotational stability could be ensured by clamping the cup with the superior augment, the ischial ramus and the pubic ramus. C. Acetabulum reconstruction was made by three customized augments reinforcing the superior acetabulum and a commercial cup.

Figure 8
A. A 60-year-old female was diagnosed as aseptic loosening 1 year after right hip revision. B. RP showed the inferior, superior and anterior acetabulum and the ischial ramus was defected. The vertical and rotational stability of the cup cannot be obtained and it was di cult to reconstruct one place by augments to obtain a rmly clamped cup. C. A customized cage was used. Rotational stability was obtained with cage wing xed into the ala of ilium and the ischial ramus by screws. Vertical stability was ensured by the acetabular cup directly contacting the superior host bone.

Figure 9
A. A 54-year-old male was diagnosed as right hip prosthesis loosening 24 years after THA. B. RP showed severe osteolysis around the acetabulum involving the superior and inferior acetabulum, the ischial ramus and the pubic ramus. Some ischial ramus cortical bone remained on RP, but it was proved to be radiopaque bone cement in operation and the ischial ramus was completely destroyed (We had reported this speci c intraoperative nding and other misleading cases under RP in another study in detail22). The vertical and rotational stability could not be obtained. C. Although the operation was managed with a customed cage, we believed that the primary stability was insu cient because the cage was xed to the ala of ilium only by a screw in the superior. Such patients might bene t more from customed hemi-pelvic prosthesis.

Figure 10
A. A pelvis anteroposterior radiograph taken before the rst operation. B. A radiograph was taken right after the rst operation and a commercial cage was used. C. 7 years after the rst operation, fatigue failure happened in the connection between the wing and cup. D. A high friction coe cient revision cup was used for revision.