548644 research-article2014

JHS0010.1177/1753193414548644Journal of Hand Surgery (Eur)Tandara et al.

JHS(E)

Full length article

A software tool for prediction of prosthesis failure at the carpometacarpal joint of the thumb

The Journal of Hand Surgery (European Volume) 2015, Vol. 40E(4) 364­–369 © The Author(s) 2014 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1753193414548644 jhs.sagepub.com

A. A. Tandara1, L. J. Capeller2, M. N. Jarczok3, P. Mayrhofer4, M. Jung5 and W. Daecke6 Abstract Early recognition of prosthesis failure is difficult. A tool that helps to identify faulty prosthesis – a cause of early implant loosening – is needed. The aim of this study was to detect early implant loosening by applying a software program EBRA (Ein-Bild-Röntgen-Analyse). EBRA was applied to the radiographs of a series of 76 patients, with a total of 102 thumb carpometacarpal joint de la Caffinière prostheses, with an average followup of 14.5 months (range 0.5–24). The data were used to draw graphs of cup migration and inclination over time. Corresponding regression lines of migration and inclination in relation to time were made up to the point of loosening. The gradient of regression was calculated for all implants. The gradient of regression on the x and y axes differed significantly between stable and loose implants. Loose and stable implants showed significantly different cup migration on the x and y axes over time. EBRA proved to be a reliable tool to visualize cup migration in the trapeziometacarpal joint and to predict implant failure. Keywords Ein-Bild-Röntgen-Analyse (EBRA), carpometacarpal arthritis, implants, loosening, prediction Date received: 20 January 2013; revised: 15 December 2013; accepted: 16 December 2013

Introduction Destructive arthritis frequently develops as a consequence of degeneration of the trapeziometacarpal joint of the thumb. Trapeziectomy, trapeziectomy with tendon interposition and trapeziectomy with tendon interposition and ligament reconstruction are the common surgical treatments (Gangopadhyay et al., 2012; Vermeulen et al., 2011). Prosthetic replacement is another option (de la Caffinière, 1974), but its outcomes are not yet well documented. Various models dislocate easily or show signs of early wear due to errors in design (Hernandez-Cortes et al., 2012; Kaszap et al., 2012; Wachtl and Sennwald, 1996). Cemented ball-and-socket joint prostheses show better long-term results, but fixation of the trapezial cup continued to be the major problem in thumb carpometacarpal (CMC) joint replacements (GuggenheimGloor et al., 2000; Hansen and Stilling, 2013). Clinical examination cannot reliably assess early wear or loosening of CMC-replacements, particularly for the cup. Early loosening is also difficult to see on radiographs (Phillips et al., 2002). For total hip prostheses, the potential durability of an implant can be

predicted by measuring radiographic migration over time: The software program EBRA (Ein-BildRöntgen-Analyse, meaning: single-image radiograph analysis) can reliably measure migration of the hip cups on conventional radiographs (Krismer et al., 1996). EBRA calculates the position of the prosthesis in a patient on a series of radiographs according to previously set reference lines and points. By using

1Department

of Surgery, Heidelberg University Hospital, Heidelberg, Germany 2Surgical Outpatient-Clinic Wiesloch, Wiesloch, Germany 3Mannheim Institute of Public Health, Social and Preventive Medicine, Heidelberg University, Mannheim, Germany 4Geometry and CAD, Leopold-Franzens-University Innsbruck, Innsbruck, Austria 5OCM, München, Germany 6Frankfurt Höchst Clinic, Frankfurt, Germany Corresponding author: A. A. Tandara, Department of Surgery, Heidelberg University Hospital, Im Neuenheimer Feld 110, 69120 Heidelberg, Germany. Email: [email protected]

Downloaded from jhs.sagepub.com at NORTH CAROLINA STATE UNIV on May 12, 2015

365

Tandara et al.

Figure 1.  Grid lines: horizontal and vertical lines over typical anatomic landmarks in the hand. Head of the prosthesis: setting three to five points, application of the measurement tool for the head. Rim of the cup: setting of minimally five points at the rim.

EBRA, loosening of the hip cup can be predicted within 2 years of implantation (Ilchmann et al., 1992; Krismer et al., 1999; Langton et al., 2010). The use of EBRA to monitor cup migration of CMC joint replacements in the thumb may reveal design failures related to early loosening much earlier than currently possible. The aim of this study was to evaluate the accuracy of EBRA in measuring cup migration of CMC implants in a model. Based on these data, a further assessment was made in a clinical setting to see whether the conclusions were reliable as regards the long-term durability of the prostheses.

Methods The software EBRA comprises two parts: ‘EBRA CUP’ and ‘EBRA GRAF’. The former part was designed for graphic modulation of the digitalized imported radiographs and the latter for calculating the data. In a first step EBRA (‘CUP’ and ‘GRAF’) had to be set up and validated for use in CMC joints. ‘EBRA CUP’ was adapted to the anatomy of the CMC joints. Reliable reference lines and points were defined and the software was modified accordingly. The vertical lines: ‘left line’ at the radial border of the trapezoid; the ‘F line’ at the ulnar border; the ‘V-M line’ in the centre of the capitate; and the ‘right line’ at the ulnar border of the capitate (Figure 1). Horizontal lines: the ‘top line’ at the distal border of the capitate; the ‘base line’ at the proximal border of the trapezium; and the ‘H-M line’ at the proximal border of the capitate (Figure 1). The outline of the prosthesis head and the elliptic cup were marked with three to five points (Figure 1).

Migration simulator: EBRA in a mechanical study in a human wrist model Next, we needed to determine whether EBRA can reliably visualize the migration of an already implanted CMC replacement according to the newly defined gridlines and points. A cadaver study, using one adult human male hand, was performed to validate the reference lines and points for their use in EBRA. To simulate migration of an implant, a plastinated hand, including the carpal and metacarpal bones, was mounted on a dynamic mechanical frame (Bohannon, 2003). A phantom of the de la Caffinière prosthesis cup attached to a stem and connected with the mechanical frame was placed in the trapezium (Figures 1 and 2). A series of radiographs was taken of the cadaver hand, with the stem of the phantom prosthesis with its attached cup placed at varying positions. These positions were six different horizontal shifts, seven modified angulations of 5° increments horizontally, and seven modified angulations of 5° increments vertically. For each position, a total of ten radiographs were taken, summing up to 60 horizontal shifts, 70 angulations horizontally and 70 angulations vertically, i.e. a total of 200 radiographs. The gridlines and points were marked by hand on the radiographs on the computer screen after they had been digitalized and imported into the ‘EBRA CUP’ program. The sphere of the head and the ellipse of the cup were calculated after they had been marked with three to five points, respectively. The position of the head and cup in relation to the reference lines of the hand was calculated in ‘EBRA GRAF’.

Downloaded from jhs.sagepub.com at NORTH CAROLINA STATE UNIV on May 12, 2015

366

The Journal of Hand Surgery (Eur) 40(4)

Figure 2.  Migration simulator (plastinated hand): The stem containing the prosthesis’ cup was angulated horizontally or vertically or shifted horizontally before each radiograph was taken. Detail: The stem containing the prosthesis’ cup.

Figure 3.  Measurement of a set of X-rays in EBRA GRAF: All ten X-rays of one simulated position of the cup were measured by EBRA and plotted into one diagram.

The calculated (raw) data from ‘EBRA GRAF’ were exported to the statistics program ‘Stata 10.1 MP’ (StataCorp LP, College Station, TX, USA) and analysed. These cup positions were visualized in a graph (Figure 3). The accuracy of the values was calibrated and validated by the program ‘EBRA GRAF’.

Clinical validation of EBRA: Patient acquisition and X-ray acquisition For clinical validation, the grid lines and points were applied to a series of radiographs in patients with a real prosthesis in place. A cemented ball-and-socket thumb CMC implant was chosen. Since their introduction in 1973, many such prostheses have been implanted (de la Caffinière, Bruz, France). As our own

de la Caffinière patient group was not sufficient for the planned analysis, we also recruited radiographs from patients of other hand surgeons. A total of 531 radiographs from 119 patients with 142 prostheses implanted were obtained. Each recruited patient received an identification (ID) number. The subsequent analysis of the acquired data was matched to this ID. Only prostheses with a minimum series of three radiographs were included in this study, leaving a total of 398 radiographs. This corresponds to a final number of 76 patients with a total of 102 prostheses. In order to detect incipient implant loosening by EBRA within a limited time frame, analyses were calculated for a maximum observation time of 24 months post implantation. The average period of observation was 14.5 months (SD = 23.7) (range 0.5–24). The radiographs were digitalized and imported into the program ‘EBRA CUP’. After calibration, the vertical and horizontal grid lines were drawn. The head and cup were marked with four to five points. The program ‘EBRA GRAF’ calculated the changes in the position of the cup in relation to all three dimensions. Exclusion criteria for migration and angulation curves. A curve of cup migration over time was drawn. A comparative algorithm of ‘EBRA GRAF’ helped to identify and exclude non-matching radiographs due to incorrect positioning of the hand on the radiographic plate. Only radiographs showing stable implants were used for the drawing and calculation of the migration curves. Loose cups were identified by following the criteria defined by Wachtl (Wachtl and Sennwald, 1996): radiographs showing a lucent zone around the prosthesis, or a tilt of the prosthesis of more than 5° compared with previous

Downloaded from jhs.sagepub.com at NORTH CAROLINA STATE UNIV on May 12, 2015

367

Tandara et al.

Ante− and Retroversion of the cup

50

40

30

20

10 −15°

−10°

−05°

+00° +05° Cup displacement

+10°

+15°

Figure 4.  Box and whiskers – as example ante- and retroversion of the cup: The x-axis shows the actual displacement of the cup and the y-axis the calculated positions.

radiographs, were defined as ‘loose’ implants. The first radiographs fulfilling the definition of ‘loose’ and the following radiographs of the same implant were excluded from migration and angulation analysis. Linear regression for migration and angulation graphs. The radiographs for each patient were grouped either as ‘stable’ or as ‘loose’. The ‘stable’ group consisted only of implants showing no cup loosening at any time in the follow-up period. The ‘loose’ group consisted of those patients in whom signs of loosening were observed in one of their follow-up radiographs according to the ‘Wachtl’ criteria. A linear regression of cup migration and inclination over time was calculated separately to each group (‘loose’ and ‘stable’). Regression was calculated for all planes of migration. In order to detect incipient implant loosening by EBRA within a limited time frame, the regression analysis was calculated for a 24-month period after surgery. Due to retrospective analysis of existing radiographs and an anonymous data collection, our institution did not require Institutional Review Board (IRB) approval of the study.

Results Migration simulator The migration of the cup per position (six horizontal shifts with ten radiographs) was less than 1 mm. The tilt per position (seven angulations horizontally and seven angulations vertically) were below 3°–4°. This corresponds to the simulated migrations of the cup (Figure 3). Box-and-whisker plots show the distribution of these data for the expected value from the different positions (Figure 4). The explained variance of

Figure 5.  Lifetime analysis of the de la Caffinière prosthesis, the x-axis showing months and the y-axis showing the fraction of stable implants (1 corresponds to a rate of 100% stable implants, 0 to 0% stable implants).

the linear regression for the known angulations of 5° increments in ante-/retroversion was r2 = 0.99, meaning that 99% of the observed variance was explained by the regression.

Clinical validation of EBRA We analysed 102 prostheses in 76 patients. After 12 months, 69% of the prostheses remained stable and after 24 months 49%. Thereafter, the loosening rate of the implants diminished markedly (Figure 5). Lines of regression show the stepwise migration of the implants over time relation (x-axis) to its initial position (y-axis) (Figure 6). Statistical calculation of the cup’s position in the course of time showed a significant migration in the direction parallel to the index finger metacarpal and in the direction perpendicular to the index finger metacarpal for all prostheses. Within 24 months of observation, the difference in migration between the ‘loose’ and the ‘stable’ groups was significant for cup migration perpendicular and parallel to the index finger metacarpal (Figure 6). The loose de la Caffinière implants migrated in a proximal-radial direction: the cups migrated approximately 0.05 mm (p 

A software tool for prediction of prosthesis failure at the carpometacarpal joint of the thumb.

Early recognition of prosthesis failure is difficult. A tool that helps to identify faulty prosthesis - a cause of early implant loosening - is needed...
552KB Sizes 4 Downloads 6 Views