- Email: [email protected]
Does 99mTc-MDP bone scintigraphy add to the investigation of patients with symptomatic unicompartmental knee replacement?
The Knee 19 (2012) 592–596
Contents lists available at SciVerse ScienceDirect
The Knee
Does 99mTc-MDP bone scintigraphy add to the investigation of patients with symptomatic unicompartmental knee replacement?☆,☆☆ Min Yen Wong a,⁎, Clare Beadsmoore a, Andoni Toms a, Toby Smith b, Simon Donell c a b c
Norfolk and Norwich University Hospital, Radiology Norwich, Norfolk, United Kingdom University of East Anglia, Faculty of Medicine and Health Sciences, Norwich, Norfolk, United Kingdom Norfolk and Norwich University Hospital, Orthopaedics, Norwich, Norfolk, United Kingdom
a r t i c l e
i n f o
Article history: Received 6 June 2011 Received in revised form 1 September 2011 Accepted 5 September 2011 Keywords: Unicompartmental arthroplasty Knee Infection Bone scan Loosening
a b s t r a c t Aim: The purpose of this study was to determine whether nuclear medicine 99mTc-Methyl diphosphonate bone scintigraphy (99mTc-MDP bone scintigraphy) added information over plain radiographs loosening infection in symptomatic unicompartmental knee replacements (UKRs). Methods and materials: A cohort of 39 patients who presented with knee pain following UKR was retrospectively reviewed. All had undergone nuclear medicine bone scans for possible loosening or infection of the prosthesis. The findings of the bone scintigraphy were compared to subsequent operative findings during diagnostic arthroscopic investigation or revision surgery for those patients who had undergone these procedures. Results: Of the 39 patients with painful knees following UKR, surgical findings confirmed that 11 had either loose (n = 9) or infected (n= 2) implants. Logistic regression analysis demonstrated no statistically significant combination of features on nuclear medicine or radiographs associated with failure of the prosthesis due to infection or loosening (pN 0.05). Classification of a satisfactory position of the UKR on plain radiography exhibited a high negative predictive value (96% for infections, and 80% for loosening). However, plain radiograph was not sensitive for loosening (50%) or infection (37%) of the UKR with very low positive predictive values (9.1% for infection and 27.3% for loosening). Conclusion: This study provides no evidence to support the routine use of 99mTc-MDP bone scintigraphy in the clinical decision-making for patients with a painful UKR. Level of evidence: level 4. © 2011 Elsevier B.V. All rights reserved.
1. Introduction The medial unicompartmental knee replacement (UKR) was designed in order to minimise the kinematic ‘disruption’ to the lateral compartment and patellofemoral joint to preserve the normal kinematics of the knee joint [1,2]. Although it is suggested that UKRs have an advantage over total knee replacement (TKR) in their ability to provide the patient with a more “normal-feeling” knee, as a consequence of their kinematic design, UKRs have presented with a higher early failure rate [3]. Furthermore, unexplained pain has been described as a significant problem in a proportion of patients; 3% at 1 year post-operation in one cohort [4].
☆ Work was performed at Norfolk and Norwich University Hospital, Radiology Norwich, Norfolk, United Kingdom. ☆☆ No funding was received for this work. ⁎ Corresponding author at: Norwich radiology academy, Cotman Centre, Norfolk and Norwich University Hospital, Colney lane, Norwich, Norfolk, NR4 7UB, United Kingdom. Tel.: + 44 7764754391, + 44 1603 286286, 0078(Bleep); fax: + 44 1603 286146. E-mail address: [email protected] (M.Y. Wong). 0968-0160/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.knee.2011.09.001
Radiographic evaluation of loosening and infection in total hip replacements by nuclear medicine 99mTc-MDP bone scintigraphy has been shown to be useful [5,6]. However, the mechanics of the knee joint differ from the hip. The TKR disturbs the natural kinematics of the knee joint. As a result, the post-operative increased bone turnover is more prolonged, meaning that increased tracer uptake around the TKR remains for longer. This has made the use of 99mTc-MDP bone scintigraphy to evaluate TKRs an issue of debate, since it is difficult to differentiate true loosening from normal post-operative appearances. The current consensus is that 99mTc-MDP bone scintigraphy alone is not an appropriate diagnostic tool, but should be used as an adjunct in conjunction with plain radiograph findings [7,8]. However, due to the reduced disruption to the lateral compartment and patellofemoral joint in the medial compartment UKR, post-operative increased bone turnover may therefore be less of a complication during the interpretation of 99mTc-MDP bone scintigraphy for this type of implant compared to the TKR. However, there is currently no published evidence to support this hypothesis. The purpose of this study was therefore to determine the diagnostic test accuracy of nuclear medicine 99mTc-MDP bone scintigraphy and plain radiographs in the identification of loose and infected symptomatic UKRs.
M.Y. Wong et al. / The Knee 19 (2012) 592–596
593
2.2. Bone scintigraphy investigation
Fig. 1. Different distributions of tracer uptake on the recorded.
99m
Tc-MDP bone scintigraphy
2. Method
Patients were scanned on a SMV gamma camera Vision DST XL, Mediso Nucline TM X-Ring SPECT and Whole Body digital gamma camera or an Adac Forte MCD MCD/AC gamma camera. Patients were positioned lying on the scanning bed with both legs rested on the bed. Once in position the patients were injected with 400 MBq Tc99m Methylendiphosponate (MDP), or Hydroxyenthylenediphosphate. Blood pool images were then acquired immediately on a matrix of 256 × 256 × 16 for 200 s for each view at a window of 140 keV with a window width of 20%. At 2 h 15 min to 3 h post-injection the late skeletal images were acquired. First in the same position as the blood pool images and then both lateral views. For each lateral view the leg to be imaged remained on the bed, and the other leg raised on a large foam pad to remove from the field of view. Acquisition of the images is the same as for the blood pool images. The 99mTc-MDP bone scintigraphs were reviewed by a consultant radionuclide radiologist (5 years experience). The presence and location of any tracer uptake in the early blood pool and late skeletal images phase were noted. Locations of uptake in the early were femoral, tibial, distal femoral, tibial and femoral, patella, lateral joint compartment and intercondylar. Patterns of uptake in the late skeletal phase were femoral, tibial, tibial and femoral, patella, lateral joint compartment, intercondylar, synovial uptake, femoral low grade uptake, tibial low grade uptake, and tibial and femoral low grade uptake.
2.1. Cohort
Frequency of localised radiotracer uptake loci per knee
A retrospective radiological and surgical notes study was conducted. All patients who had undergone medial UKR (Oxford™ Phase III meniscal unicompartmental knee system–Biomet, Warsaw, Indiana, USA) between January 2001 and December 2009, at our institution were identified from surgical records (n = 789). From this cohort, those who had undergone a bone scan were identified from the Picture Archive Communications System (PACS) (n = 53). Those who had 99mTc-MDP bone scintigraphy for indications other than a painful knee, such as investigations for metastases, were excluded from this review (n = 14). Through this, 39 patients who had 99mTc-MDP bone scintigraphy purely to evaluate a symptomatic knee for loose or infected UKRs were identified and included in the study. The demographic information pertaining to this cohort is presented in Table 1.
2.3. Plain radiograph investigation Standard weight bearing anteroposterior and non-weight bearing lateral radiographs were taken with a focus-to-film distance of 115 cm. None of the images were fluoroscopically screened. Post-operative radiographs were reviewed by a consultant orthopaedic surgeon (19 years experience). The radiograph taken immediately prior to each patient's bone scintigraph was selected for review. Both anteroposterior and lateral views of the symptomatic knee were reviewed. Each patient's films were reviewed for signs of loosening or infection (Fig. 3). The position of the femoral and tibial components of the medial UKR was also assessed in accordance with the designing institute's recommendations [9,10].
14 12 10 8 6 4 2 0 early
late
Fig. 2. Box and whisker chart describing the frequency of radionuclide tracer uptake in regions around the knee. The number (height of boxes) and frequency (height of whiskers) of sites increases with late phase imaging.
Fig. 3. Abnormal position of a medial unicompartmental knee replacement on plain film. The axis of the femoral prosthesis is not perpendicular to the tibial prosthesis and the articular surfaces are not seen to be parallel; best appreciated on the AP view. This was subsequently shown to be loose. Progression of arthritis to lateral compartment also noted.
594
M.Y. Wong et al. / The Knee 19 (2012) 592–596
Table 1 Cohort characteristics. Frequency Medium age in years at surgery (IQR) Gender
66.5 (60.8–73.0) 17 22 14 25 27 (21.8–36.0) 29 (25.0–34.5) 21.0 (12.5–38.0) 15 12 1 2 32.0 (20.0–49.0) 5 1 2 2 1 4 14.0 (7.0–25.0)
Male Female Left Right
Side Medium OKS (IQR) Medium SF-12 (IQR) Medium duration between primary surgery to IBS (months) (IQR) Revision surgery undertaken
Total UKR–TKR UKR–UKR Arthroscopy Medium duration between primary surgery to revision surgery (months) (IQR) Operative findings at revision surgery Femoral component loosening Tibial component loosening Tibial and femoral component loosening Infection Loose bodies No abnormalities Medium duration between IBS to revision surgery (months) (IQR)
p b 0.05 was considered statistically significant. The statistical analysis was performed using the MedCalc® Version 11.0.1.0.
2.4. Surgical reference comparison Surgical and arthroscopic findings of aseptic or septic loosening during revision arthroplasty were extracted, as well as any resulting microbiology findings. Prostheses were deemed to be infected if positive cultures were found at surgery. Prostheses were deemed to be loose at surgery if the femoral side came free on gentle traction, or the tibial side toggled and lifted out with ease using a bone lever. If the patient did not go on to revision or further investigations, it was assumed that the UKR was not loose or infected. 2.5. Statistical analyses The sensitivity and specificity, positive predictive and negative predictive values of each finding on bone scan and plain radiograph were calculated and a logistic regression performed to ascertain if there was a statistically significant combination of features on nuclear medicine or plain radiograph imaging associated with failure of the prosthesis due to infection or loosening. A probability level of
3. Results Of the 39 patients with an Oxford UKR who underwent bone isotope scanning, 11 had either loose (n = 9, confirmed at surgery) or infected (n = 2, confirmed on microbiology) implants. The remaining patients had non-specific knee pain (the knee was either found to be normal at arthroscopy, at revision, or the patient's symptoms resolved spontaneously) giving a prevalence of failure in this population of 28%. In early phase acquisitions radiotracer uptake was demonstrated in seven different locations/patterns with a median frequency of three (inter-quartile range (IQR) 1.25 to 7.25) (Fig. 2). In comparison, there was at least one focus of radiotracer activity in each of the 39 knees on late phase images. These included 10 different sites or patterns of distribution with a median frequency of five per site (IQR 2.0 to 11.0) (Fig. 1). Sensitivity and specificity values for each location of early and late uptake were calculated. No single finding was found to be both sensitive and specific. Many findings were found to be 100% sensitive but were also non-specific (Table 2–3). Logistic regression found no statistically significant combination for these findings (p N 0.05). Of the 39 scans, the results of 13 of the scans (33.3%) were not acted on. Five patients who were diagnosed as having an infection (n = 1) or loosening (n = 4) did not
Table 2 Sensitivity, specificity, and positive and negative predictive values for early and late increases in radiotracer uptake for infected prostheses. No location of early or late uptake was found to be both sensitive and specific in determining whether a prosthesis is infected. Site
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
Early Femoral uptake Tibial uptake Tibial and femoral uptake Lateral joint compartment Intercondylar uptake Synovial uptake
50 100 50 100 50 100
(1.3–98.7) (15.8–100.0) (1.3–98.7) (15.8–100) (1.3–98.7) (15.8–100.0)
75.7 (58.8–88.2) 21.62 (9.8–38.2) 94.59 (81.8–99.3) 13.51 (4.5–28.8) 97.3 (85.8–99.9) 2.7 (0.07–14.2)
10 6.5 33.3 5.9 50 5.3
96.6 100 97.2 100 97.3 100
Late Femoral low grade uptake Tibial low grade uptake Tibial and femoral low grade uptake Femoral uptake Tibial uptake Distal femoral uptake Tibial and femoral uptake Lateral joint compartment uptake Intercondylar uptake
100 100 50 50 100 100 100 100 50
(15.8–100.0) (15.8–100.0) (1.3–98.7) (1.3–98.7) (15.8–100.0) (15.8–100.0) (15.8–100.0) (15.8–100.0) (1.3–98.7)
5.41 (0.7–18.2) 29.73 (15.9–47.0) 67.57 (50.2–82.0) 72.97 (55.9–86.2) 13.51 (4.5–28.8) 2.7 (0.07–14.2) 13.51 (4.5–28.8) 16.22 (6.2–32.0) 91.89 (78.1–98.3)
5.4 7.1 7.7 9.1 5.9 5.3 5.9 6.1 25
100 100 96.2 96.4 100 100 100 100 97.1
M.Y. Wong et al. / The Knee 19 (2012) 592–596
595
Table 3 Sensitivity, specificity, and positive and negative predictive values for early and late increases in radiotracer uptake for loose prostheses. No location of early or late uptake was found to be both sensitive and specific in determining whether a prosthesis is loose. Site
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
Early Femoral uptake Tibial uptake Tibial and femoral uptake Lateral joint compartment Intercondylar uptake Synovial uptake
33.33 22.22 11.11 22.22 100 100
(7.5–70.1) (2.8–60.0) (0.3–48.2) (2.8–60.0) (66.4–100.0) (66.4–100.0)
76.67 80 93.33 90 6.67 3.33
(57.7–90.1) (61.4–92.3) (77.9–99.2) (73.5–97.9) (0.8–22.1) (0.08–17.2)
30 25 33.3 40 24.3 23.7
79.3 77.4 77.8 79.4 100 100
Late Femoral low grade uptake Tibial low grade uptake Tibial and femoral low grade uptake Femoral uptake Tibial uptake Distal femoral uptake Tibial and femoral uptake Lateral joint compartment uptake Intercondylar uptake
11.11 33.33 77.78 33.33 100 11.11 22.22 22.22 11.11
(0.3–48.2) (7.5–70.1) (40.0–97.2) (7.5–70.1) (66.4–100.0) (0.3–48.2) (2.8–60.0) (2.8–60.0) (0.3–48.2)
96.67 73.33 36.67 73.33 16.67 100 90 86.67 90
(82.8–99.9) (54.1–87.7) (19.9–56.1) (54.1–87.7) (5.6–34.7) (88.4–100.0) (73.5–97.9) (69.3–96.2) (73.5–97.9)
50 27.3 26.9 27.3 26.5 100 40 33.3 25
78.4 78.6 84.6 78.6 100 78.9 79.4 78.8 77.1
Table 4 Table of sensitivity, specificity, and positive and negative predictive values for classification of satisfactory or unsatisfactory position on the plain radiograph in determining loose or infected prosthesis. Overall impression as to whether there is satisfactory or unsatisfactory position of the prosthesis on the plain radiograph has a high negative predictive value, making it specific but not sensitive for loosening or infection of the prosthesis.
Infection Loosening
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
50 (1.3–98.7) 37.5 (8.5–75.5)
70.59 (52.5–84.9) 71.43 (51.3–86.8)
9.1 27.3
96 80
proceed to surgery, whilst eight who were reported as normal, underwent a revision or arthroscopic procedure. Three scans were reported as indeterminate and therefore deemed not clinically helpful. Overall the results of 16 scans (41%) were either clinically not helpful or were not acted upon. Plain radiograph abnormalities were common. Each of the 15 possible defined plain film abnormalities were demonstrated in a mean of 30 (sd 3.1) of the39 patients. Logistic regression demonstrated no statistically significant combination of features on plain radiograph imaging associated with failure of the prosthesis due to infection or loosening (pN 0.05). Classification of a satisfactory position of the UKR on plain radiography exhibited a high negative predictive value (96% for infections, and 80% for loosening). However, plain radiograph was not sensitive for loosening (50%) or infection (37%) of the UKR with very low positive predictive values (9.1% for infection and 27.3% for loosening) (Table 4). In those patients who did not have loosening or infection, six were found noted to have patterns of degenerative change on their 99mTc-MDP bone scintigraphs. Five had patterns of degenerative change seen in the non-replaced lateral compartment and two had degenerative changes in the patello-femoral joint. All degenerative changes were also noted on the plain films.
This study indicated that 99mTc-MDP bone scintigraphy was able to identify other causes of post-operative pain in non-loose and non-septic prostheses, most notably the demonstration of degenerative changes in the lateral and patellofemoral joint compartments (Fig. 5). This finding supports previous studies which have also found progressive osteoarthritis in the contralateral compartment to the prosthesis (especially if the prosthesis is in the lateral compartment) [11–13]. Plain radiographs were also found not to be specific, with most showing abnormalities of implant position. However prosthesis position on the plain radiograph demonstrated a high negative predictive value, making it specific but not sensitive for loosening or infection. There is no reason to expect implant mal-position should lead to
4. Discussion In this study no finding, or combinations of findings, on nuclear medicine 99mTc-MDP bone scintigraphy or plain radiograph were found that were both specific and sensitive for the diagnosis of loosening or infection in the painful UKR. The majority of prostheses demonstrated continued uptake on late images. There was no significant difference in location or pattern of late uptake between normal, loose or infected prostheses (Fig. 4). This suggests that there remains an increased bone turnover in at least the symptomatic patients. This is contrary to the hypothesis that there is less post-operative bone turnover in a UKR prosthesis as knee kinematics return to normal. This persistent increased bone turnover makes 99mTc-MDP bone scintigraphy difficult to interpret and non-specific in the investigation of infection or loosening of unicompartmental prostheses.
Fig. 4. A patient's 99mTc-MDP bone scintigraphy scan from the normal group. These show uptake on both the early and late images around the unicompartmental prostheses (black arrows). This normal increased uptake makes the scans difficult to interpret for loosening and infection.
596
M.Y. Wong et al. / The Knee 19 (2012) 592–596
out of the nine (44%) did not lead to a revision procedure. One patient out of the four with infection diagnosed on 99mTc-MDP bone scintigraphy was also not revised, suggesting that clinicians place more emphasis on their clinical observations than the imaging findings or that revision surgery was precluded due to patient co morbidity. 5. Conclusion This study provides no evidence to support the routine use of Tc-MDP bone scintigraphy in the clinical decision-making for patients with a painful UKR. 99m
6. Conflict of interest statement
Fig. 5. Severe degenerative changes seen in the lateral compartment of the left knee. This is well seen on the nuclear medicine scan as uptake around the knee in a characteristic distribution (black arrow). The bone scan shows the degenerative change to be quite severe. The severity of the degenerative change is not appreciated on the plain film of the same knee which shows moderately severe degenerative changes (white arrow).
infection, although it may contribute to loosening. Equally well, implant mal-positioning may lead to soft-tissue impingement causing pain, which would not be detected on bone scintigraphy. However, the plain X-ray films were not fluoroscopically controlled. Therefore it was not possible to accurately assess the presence of lucent lines around the tibial component. However an obvious mal-position would not have disappeared by using screened compared to non-screened films. This study presented with a number of limitations. Firstly, this was a retrospective study. As a result, selection bias was inevitable as only those patients with symptomatic knees are investigated for loosening or infection. If there were any significant findings associated with loosening or infection then these might also have been present in an asymptomatic control group. This was not the case. The numbers are also small and therefore subject to a type II error. A further limitation was that the 99mTc-MDP bone scintigraphs were reviewed by only one observer. However the review was blinded to which UKRs were normal, infected, or loose. In evaluating the clinical usefulness of 99mTc-MDP bone scintigraphy the retrospective nature of this study meant that it was difficult to tell precisely whether the bone scans were helpful to the clinician's decision-making to not do anything. For instance, where the scan reported aseptic loosening, but based on other factors, a revision was not carried out. In our study in the management of patients with a diagnosis of loosening on 99mTc-MDP bone scintigraphy, four
There is no conflict of interest for any of the authors. No finding for sponsorship was received for this work. References [1] Patil S, Colwell CW, Ezzet KA, D'Lima DD. Can normal knee kinematics be restored with unicompartmental knee replacement? J Bone Joint Surg Am Feb 1 2005;87(2):332–8. [2] Hollinghurst D, Stoney J, Ward T, Gill H, Newman J, Murray D, et al. No deterioration of kinematics and cruciate function 10 years after medial unicompartmental arthroplasty. 12Knee 2006;13(6):440–4. [3] No authors listed. National Joint Registry for England and Wales: 7th Annual Report. 2010. http://www.njrcentre.org.uk/NjrCentre/LinkClick.aspx?fileticket= QkPI7kk6B2E%3d&tabid=86&mid=523. [4] Kuipers BM, Kollen BJ, Kaijser Bots PC, Burger BJ, van Raay JAM, Tulp NJA, et al. Factors associated with reduced early survival in the Oxford phase III medial unicompartment knee replacement. Knee 2010;17:48–52. [5] Gelman MI, Coleman RE, Stevens PM, Davey BW. Radiography, radionuclide imaging, and arthrography in the evaluation of total hip and knee replacement. Radiology 1978;128(3):677–82. [6] Rosenthall L, Lisbona R, Hernandez M, Hadjipavlou A. 99mTc-PP and 67Ga imaging following insertion of orthopedic devices. Radiology Dec 1979;133(3 Pt 1):717–21. [7] Owen RJT, Harper WM, Finlay DB, Belton IP. Isotope bone scans in patients with painful knee replacements: do they alter management? Br J Radiol Nov 1 1995;68(815): 1204–7. [8] Hunter J, Hattner R, Murray W, Genant H. Loosening of the total knee arthroplasty: detection by radionuclide bone scanning. Am J Roentgenol Jul 1 1980;135(1):131–6. [9] No authors listed. Oxford™ Partial Knee: Manual of the Surgical Technique. http:// www.biomet.nl/resource/5980/Oxford-Knee-Optec.pdf. [10] Gulati A, Chau R, Simpson DJ, Dodd CAF, Gill HS, Murray DW. Influence of component alignment on outcome for unicompartmental knee replacement. Knee 2009;16: 196–9. [11] Nassiri M, Quinlan JF, O'Byrne JM. Analysis of 13 early failures of the mobile bearing Oxford phase III unicompartmental knee prosthesis. 10Eur J Orthop Surg Traumatol 2009;20(4):303–7. [12] Murray DW, Goodfellow JW, O'Connor JJ. The Oxford medial unicompartmental arthroplasty: a ten year survival study. J Bone Joint Surg Br Nov 1 1998;80-B(6): 983–9. [13] Walton MJ, Weale AE, Newman JH. The progression of arthritis following lateral unicompartmental knee replacement. Knee Oct 2006;13(5):374–7.
Contents lists available at SciVerse ScienceDirect
The Knee
Does 99mTc-MDP bone scintigraphy add to the investigation of patients with symptomatic unicompartmental knee replacement?☆,☆☆ Min Yen Wong a,⁎, Clare Beadsmoore a, Andoni Toms a, Toby Smith b, Simon Donell c a b c
Norfolk and Norwich University Hospital, Radiology Norwich, Norfolk, United Kingdom University of East Anglia, Faculty of Medicine and Health Sciences, Norwich, Norfolk, United Kingdom Norfolk and Norwich University Hospital, Orthopaedics, Norwich, Norfolk, United Kingdom
a r t i c l e
i n f o
Article history: Received 6 June 2011 Received in revised form 1 September 2011 Accepted 5 September 2011 Keywords: Unicompartmental arthroplasty Knee Infection Bone scan Loosening
a b s t r a c t Aim: The purpose of this study was to determine whether nuclear medicine 99mTc-Methyl diphosphonate bone scintigraphy (99mTc-MDP bone scintigraphy) added information over plain radiographs loosening infection in symptomatic unicompartmental knee replacements (UKRs). Methods and materials: A cohort of 39 patients who presented with knee pain following UKR was retrospectively reviewed. All had undergone nuclear medicine bone scans for possible loosening or infection of the prosthesis. The findings of the bone scintigraphy were compared to subsequent operative findings during diagnostic arthroscopic investigation or revision surgery for those patients who had undergone these procedures. Results: Of the 39 patients with painful knees following UKR, surgical findings confirmed that 11 had either loose (n = 9) or infected (n= 2) implants. Logistic regression analysis demonstrated no statistically significant combination of features on nuclear medicine or radiographs associated with failure of the prosthesis due to infection or loosening (pN 0.05). Classification of a satisfactory position of the UKR on plain radiography exhibited a high negative predictive value (96% for infections, and 80% for loosening). However, plain radiograph was not sensitive for loosening (50%) or infection (37%) of the UKR with very low positive predictive values (9.1% for infection and 27.3% for loosening). Conclusion: This study provides no evidence to support the routine use of 99mTc-MDP bone scintigraphy in the clinical decision-making for patients with a painful UKR. Level of evidence: level 4. © 2011 Elsevier B.V. All rights reserved.
1. Introduction The medial unicompartmental knee replacement (UKR) was designed in order to minimise the kinematic ‘disruption’ to the lateral compartment and patellofemoral joint to preserve the normal kinematics of the knee joint [1,2]. Although it is suggested that UKRs have an advantage over total knee replacement (TKR) in their ability to provide the patient with a more “normal-feeling” knee, as a consequence of their kinematic design, UKRs have presented with a higher early failure rate [3]. Furthermore, unexplained pain has been described as a significant problem in a proportion of patients; 3% at 1 year post-operation in one cohort [4].
☆ Work was performed at Norfolk and Norwich University Hospital, Radiology Norwich, Norfolk, United Kingdom. ☆☆ No funding was received for this work. ⁎ Corresponding author at: Norwich radiology academy, Cotman Centre, Norfolk and Norwich University Hospital, Colney lane, Norwich, Norfolk, NR4 7UB, United Kingdom. Tel.: + 44 7764754391, + 44 1603 286286, 0078(Bleep); fax: + 44 1603 286146. E-mail address: [email protected] (M.Y. Wong). 0968-0160/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.knee.2011.09.001
Radiographic evaluation of loosening and infection in total hip replacements by nuclear medicine 99mTc-MDP bone scintigraphy has been shown to be useful [5,6]. However, the mechanics of the knee joint differ from the hip. The TKR disturbs the natural kinematics of the knee joint. As a result, the post-operative increased bone turnover is more prolonged, meaning that increased tracer uptake around the TKR remains for longer. This has made the use of 99mTc-MDP bone scintigraphy to evaluate TKRs an issue of debate, since it is difficult to differentiate true loosening from normal post-operative appearances. The current consensus is that 99mTc-MDP bone scintigraphy alone is not an appropriate diagnostic tool, but should be used as an adjunct in conjunction with plain radiograph findings [7,8]. However, due to the reduced disruption to the lateral compartment and patellofemoral joint in the medial compartment UKR, post-operative increased bone turnover may therefore be less of a complication during the interpretation of 99mTc-MDP bone scintigraphy for this type of implant compared to the TKR. However, there is currently no published evidence to support this hypothesis. The purpose of this study was therefore to determine the diagnostic test accuracy of nuclear medicine 99mTc-MDP bone scintigraphy and plain radiographs in the identification of loose and infected symptomatic UKRs.
M.Y. Wong et al. / The Knee 19 (2012) 592–596
593
2.2. Bone scintigraphy investigation
Fig. 1. Different distributions of tracer uptake on the recorded.
99m
Tc-MDP bone scintigraphy
2. Method
Patients were scanned on a SMV gamma camera Vision DST XL, Mediso Nucline TM X-Ring SPECT and Whole Body digital gamma camera or an Adac Forte MCD MCD/AC gamma camera. Patients were positioned lying on the scanning bed with both legs rested on the bed. Once in position the patients were injected with 400 MBq Tc99m Methylendiphosponate (MDP), or Hydroxyenthylenediphosphate. Blood pool images were then acquired immediately on a matrix of 256 × 256 × 16 for 200 s for each view at a window of 140 keV with a window width of 20%. At 2 h 15 min to 3 h post-injection the late skeletal images were acquired. First in the same position as the blood pool images and then both lateral views. For each lateral view the leg to be imaged remained on the bed, and the other leg raised on a large foam pad to remove from the field of view. Acquisition of the images is the same as for the blood pool images. The 99mTc-MDP bone scintigraphs were reviewed by a consultant radionuclide radiologist (5 years experience). The presence and location of any tracer uptake in the early blood pool and late skeletal images phase were noted. Locations of uptake in the early were femoral, tibial, distal femoral, tibial and femoral, patella, lateral joint compartment and intercondylar. Patterns of uptake in the late skeletal phase were femoral, tibial, tibial and femoral, patella, lateral joint compartment, intercondylar, synovial uptake, femoral low grade uptake, tibial low grade uptake, and tibial and femoral low grade uptake.
2.1. Cohort
Frequency of localised radiotracer uptake loci per knee
A retrospective radiological and surgical notes study was conducted. All patients who had undergone medial UKR (Oxford™ Phase III meniscal unicompartmental knee system–Biomet, Warsaw, Indiana, USA) between January 2001 and December 2009, at our institution were identified from surgical records (n = 789). From this cohort, those who had undergone a bone scan were identified from the Picture Archive Communications System (PACS) (n = 53). Those who had 99mTc-MDP bone scintigraphy for indications other than a painful knee, such as investigations for metastases, were excluded from this review (n = 14). Through this, 39 patients who had 99mTc-MDP bone scintigraphy purely to evaluate a symptomatic knee for loose or infected UKRs were identified and included in the study. The demographic information pertaining to this cohort is presented in Table 1.
2.3. Plain radiograph investigation Standard weight bearing anteroposterior and non-weight bearing lateral radiographs were taken with a focus-to-film distance of 115 cm. None of the images were fluoroscopically screened. Post-operative radiographs were reviewed by a consultant orthopaedic surgeon (19 years experience). The radiograph taken immediately prior to each patient's bone scintigraph was selected for review. Both anteroposterior and lateral views of the symptomatic knee were reviewed. Each patient's films were reviewed for signs of loosening or infection (Fig. 3). The position of the femoral and tibial components of the medial UKR was also assessed in accordance with the designing institute's recommendations [9,10].
14 12 10 8 6 4 2 0 early
late
Fig. 2. Box and whisker chart describing the frequency of radionuclide tracer uptake in regions around the knee. The number (height of boxes) and frequency (height of whiskers) of sites increases with late phase imaging.
Fig. 3. Abnormal position of a medial unicompartmental knee replacement on plain film. The axis of the femoral prosthesis is not perpendicular to the tibial prosthesis and the articular surfaces are not seen to be parallel; best appreciated on the AP view. This was subsequently shown to be loose. Progression of arthritis to lateral compartment also noted.
594
M.Y. Wong et al. / The Knee 19 (2012) 592–596
Table 1 Cohort characteristics. Frequency Medium age in years at surgery (IQR) Gender
66.5 (60.8–73.0) 17 22 14 25 27 (21.8–36.0) 29 (25.0–34.5) 21.0 (12.5–38.0) 15 12 1 2 32.0 (20.0–49.0) 5 1 2 2 1 4 14.0 (7.0–25.0)
Male Female Left Right
Side Medium OKS (IQR) Medium SF-12 (IQR) Medium duration between primary surgery to IBS (months) (IQR) Revision surgery undertaken
Total UKR–TKR UKR–UKR Arthroscopy Medium duration between primary surgery to revision surgery (months) (IQR) Operative findings at revision surgery Femoral component loosening Tibial component loosening Tibial and femoral component loosening Infection Loose bodies No abnormalities Medium duration between IBS to revision surgery (months) (IQR)
p b 0.05 was considered statistically significant. The statistical analysis was performed using the MedCalc® Version 11.0.1.0.
2.4. Surgical reference comparison Surgical and arthroscopic findings of aseptic or septic loosening during revision arthroplasty were extracted, as well as any resulting microbiology findings. Prostheses were deemed to be infected if positive cultures were found at surgery. Prostheses were deemed to be loose at surgery if the femoral side came free on gentle traction, or the tibial side toggled and lifted out with ease using a bone lever. If the patient did not go on to revision or further investigations, it was assumed that the UKR was not loose or infected. 2.5. Statistical analyses The sensitivity and specificity, positive predictive and negative predictive values of each finding on bone scan and plain radiograph were calculated and a logistic regression performed to ascertain if there was a statistically significant combination of features on nuclear medicine or plain radiograph imaging associated with failure of the prosthesis due to infection or loosening. A probability level of
3. Results Of the 39 patients with an Oxford UKR who underwent bone isotope scanning, 11 had either loose (n = 9, confirmed at surgery) or infected (n = 2, confirmed on microbiology) implants. The remaining patients had non-specific knee pain (the knee was either found to be normal at arthroscopy, at revision, or the patient's symptoms resolved spontaneously) giving a prevalence of failure in this population of 28%. In early phase acquisitions radiotracer uptake was demonstrated in seven different locations/patterns with a median frequency of three (inter-quartile range (IQR) 1.25 to 7.25) (Fig. 2). In comparison, there was at least one focus of radiotracer activity in each of the 39 knees on late phase images. These included 10 different sites or patterns of distribution with a median frequency of five per site (IQR 2.0 to 11.0) (Fig. 1). Sensitivity and specificity values for each location of early and late uptake were calculated. No single finding was found to be both sensitive and specific. Many findings were found to be 100% sensitive but were also non-specific (Table 2–3). Logistic regression found no statistically significant combination for these findings (p N 0.05). Of the 39 scans, the results of 13 of the scans (33.3%) were not acted on. Five patients who were diagnosed as having an infection (n = 1) or loosening (n = 4) did not
Table 2 Sensitivity, specificity, and positive and negative predictive values for early and late increases in radiotracer uptake for infected prostheses. No location of early or late uptake was found to be both sensitive and specific in determining whether a prosthesis is infected. Site
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
Early Femoral uptake Tibial uptake Tibial and femoral uptake Lateral joint compartment Intercondylar uptake Synovial uptake
50 100 50 100 50 100
(1.3–98.7) (15.8–100.0) (1.3–98.7) (15.8–100) (1.3–98.7) (15.8–100.0)
75.7 (58.8–88.2) 21.62 (9.8–38.2) 94.59 (81.8–99.3) 13.51 (4.5–28.8) 97.3 (85.8–99.9) 2.7 (0.07–14.2)
10 6.5 33.3 5.9 50 5.3
96.6 100 97.2 100 97.3 100
Late Femoral low grade uptake Tibial low grade uptake Tibial and femoral low grade uptake Femoral uptake Tibial uptake Distal femoral uptake Tibial and femoral uptake Lateral joint compartment uptake Intercondylar uptake
100 100 50 50 100 100 100 100 50
(15.8–100.0) (15.8–100.0) (1.3–98.7) (1.3–98.7) (15.8–100.0) (15.8–100.0) (15.8–100.0) (15.8–100.0) (1.3–98.7)
5.41 (0.7–18.2) 29.73 (15.9–47.0) 67.57 (50.2–82.0) 72.97 (55.9–86.2) 13.51 (4.5–28.8) 2.7 (0.07–14.2) 13.51 (4.5–28.8) 16.22 (6.2–32.0) 91.89 (78.1–98.3)
5.4 7.1 7.7 9.1 5.9 5.3 5.9 6.1 25
100 100 96.2 96.4 100 100 100 100 97.1
M.Y. Wong et al. / The Knee 19 (2012) 592–596
595
Table 3 Sensitivity, specificity, and positive and negative predictive values for early and late increases in radiotracer uptake for loose prostheses. No location of early or late uptake was found to be both sensitive and specific in determining whether a prosthesis is loose. Site
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
Early Femoral uptake Tibial uptake Tibial and femoral uptake Lateral joint compartment Intercondylar uptake Synovial uptake
33.33 22.22 11.11 22.22 100 100
(7.5–70.1) (2.8–60.0) (0.3–48.2) (2.8–60.0) (66.4–100.0) (66.4–100.0)
76.67 80 93.33 90 6.67 3.33
(57.7–90.1) (61.4–92.3) (77.9–99.2) (73.5–97.9) (0.8–22.1) (0.08–17.2)
30 25 33.3 40 24.3 23.7
79.3 77.4 77.8 79.4 100 100
Late Femoral low grade uptake Tibial low grade uptake Tibial and femoral low grade uptake Femoral uptake Tibial uptake Distal femoral uptake Tibial and femoral uptake Lateral joint compartment uptake Intercondylar uptake
11.11 33.33 77.78 33.33 100 11.11 22.22 22.22 11.11
(0.3–48.2) (7.5–70.1) (40.0–97.2) (7.5–70.1) (66.4–100.0) (0.3–48.2) (2.8–60.0) (2.8–60.0) (0.3–48.2)
96.67 73.33 36.67 73.33 16.67 100 90 86.67 90
(82.8–99.9) (54.1–87.7) (19.9–56.1) (54.1–87.7) (5.6–34.7) (88.4–100.0) (73.5–97.9) (69.3–96.2) (73.5–97.9)
50 27.3 26.9 27.3 26.5 100 40 33.3 25
78.4 78.6 84.6 78.6 100 78.9 79.4 78.8 77.1
Table 4 Table of sensitivity, specificity, and positive and negative predictive values for classification of satisfactory or unsatisfactory position on the plain radiograph in determining loose or infected prosthesis. Overall impression as to whether there is satisfactory or unsatisfactory position of the prosthesis on the plain radiograph has a high negative predictive value, making it specific but not sensitive for loosening or infection of the prosthesis.
Infection Loosening
Sensitivity (%) (95% confidence intervals)
Specificity (%) (95% confidence intervals)
Positive predictive value (%)
Negative predictive value (%)
50 (1.3–98.7) 37.5 (8.5–75.5)
70.59 (52.5–84.9) 71.43 (51.3–86.8)
9.1 27.3
96 80
proceed to surgery, whilst eight who were reported as normal, underwent a revision or arthroscopic procedure. Three scans were reported as indeterminate and therefore deemed not clinically helpful. Overall the results of 16 scans (41%) were either clinically not helpful or were not acted upon. Plain radiograph abnormalities were common. Each of the 15 possible defined plain film abnormalities were demonstrated in a mean of 30 (sd 3.1) of the39 patients. Logistic regression demonstrated no statistically significant combination of features on plain radiograph imaging associated with failure of the prosthesis due to infection or loosening (pN 0.05). Classification of a satisfactory position of the UKR on plain radiography exhibited a high negative predictive value (96% for infections, and 80% for loosening). However, plain radiograph was not sensitive for loosening (50%) or infection (37%) of the UKR with very low positive predictive values (9.1% for infection and 27.3% for loosening) (Table 4). In those patients who did not have loosening or infection, six were found noted to have patterns of degenerative change on their 99mTc-MDP bone scintigraphs. Five had patterns of degenerative change seen in the non-replaced lateral compartment and two had degenerative changes in the patello-femoral joint. All degenerative changes were also noted on the plain films.
This study indicated that 99mTc-MDP bone scintigraphy was able to identify other causes of post-operative pain in non-loose and non-septic prostheses, most notably the demonstration of degenerative changes in the lateral and patellofemoral joint compartments (Fig. 5). This finding supports previous studies which have also found progressive osteoarthritis in the contralateral compartment to the prosthesis (especially if the prosthesis is in the lateral compartment) [11–13]. Plain radiographs were also found not to be specific, with most showing abnormalities of implant position. However prosthesis position on the plain radiograph demonstrated a high negative predictive value, making it specific but not sensitive for loosening or infection. There is no reason to expect implant mal-position should lead to
4. Discussion In this study no finding, or combinations of findings, on nuclear medicine 99mTc-MDP bone scintigraphy or plain radiograph were found that were both specific and sensitive for the diagnosis of loosening or infection in the painful UKR. The majority of prostheses demonstrated continued uptake on late images. There was no significant difference in location or pattern of late uptake between normal, loose or infected prostheses (Fig. 4). This suggests that there remains an increased bone turnover in at least the symptomatic patients. This is contrary to the hypothesis that there is less post-operative bone turnover in a UKR prosthesis as knee kinematics return to normal. This persistent increased bone turnover makes 99mTc-MDP bone scintigraphy difficult to interpret and non-specific in the investigation of infection or loosening of unicompartmental prostheses.
Fig. 4. A patient's 99mTc-MDP bone scintigraphy scan from the normal group. These show uptake on both the early and late images around the unicompartmental prostheses (black arrows). This normal increased uptake makes the scans difficult to interpret for loosening and infection.
596
M.Y. Wong et al. / The Knee 19 (2012) 592–596
out of the nine (44%) did not lead to a revision procedure. One patient out of the four with infection diagnosed on 99mTc-MDP bone scintigraphy was also not revised, suggesting that clinicians place more emphasis on their clinical observations than the imaging findings or that revision surgery was precluded due to patient co morbidity. 5. Conclusion This study provides no evidence to support the routine use of Tc-MDP bone scintigraphy in the clinical decision-making for patients with a painful UKR. 99m
6. Conflict of interest statement
Fig. 5. Severe degenerative changes seen in the lateral compartment of the left knee. This is well seen on the nuclear medicine scan as uptake around the knee in a characteristic distribution (black arrow). The bone scan shows the degenerative change to be quite severe. The severity of the degenerative change is not appreciated on the plain film of the same knee which shows moderately severe degenerative changes (white arrow).
infection, although it may contribute to loosening. Equally well, implant mal-positioning may lead to soft-tissue impingement causing pain, which would not be detected on bone scintigraphy. However, the plain X-ray films were not fluoroscopically controlled. Therefore it was not possible to accurately assess the presence of lucent lines around the tibial component. However an obvious mal-position would not have disappeared by using screened compared to non-screened films. This study presented with a number of limitations. Firstly, this was a retrospective study. As a result, selection bias was inevitable as only those patients with symptomatic knees are investigated for loosening or infection. If there were any significant findings associated with loosening or infection then these might also have been present in an asymptomatic control group. This was not the case. The numbers are also small and therefore subject to a type II error. A further limitation was that the 99mTc-MDP bone scintigraphs were reviewed by only one observer. However the review was blinded to which UKRs were normal, infected, or loose. In evaluating the clinical usefulness of 99mTc-MDP bone scintigraphy the retrospective nature of this study meant that it was difficult to tell precisely whether the bone scans were helpful to the clinician's decision-making to not do anything. For instance, where the scan reported aseptic loosening, but based on other factors, a revision was not carried out. In our study in the management of patients with a diagnosis of loosening on 99mTc-MDP bone scintigraphy, four
There is no conflict of interest for any of the authors. No finding for sponsorship was received for this work. References [1] Patil S, Colwell CW, Ezzet KA, D'Lima DD. Can normal knee kinematics be restored with unicompartmental knee replacement? J Bone Joint Surg Am Feb 1 2005;87(2):332–8. [2] Hollinghurst D, Stoney J, Ward T, Gill H, Newman J, Murray D, et al. No deterioration of kinematics and cruciate function 10 years after medial unicompartmental arthroplasty. 12Knee 2006;13(6):440–4. [3] No authors listed. National Joint Registry for England and Wales: 7th Annual Report. 2010. http://www.njrcentre.org.uk/NjrCentre/LinkClick.aspx?fileticket= QkPI7kk6B2E%3d&tabid=86&mid=523. [4] Kuipers BM, Kollen BJ, Kaijser Bots PC, Burger BJ, van Raay JAM, Tulp NJA, et al. Factors associated with reduced early survival in the Oxford phase III medial unicompartment knee replacement. Knee 2010;17:48–52. [5] Gelman MI, Coleman RE, Stevens PM, Davey BW. Radiography, radionuclide imaging, and arthrography in the evaluation of total hip and knee replacement. Radiology 1978;128(3):677–82. [6] Rosenthall L, Lisbona R, Hernandez M, Hadjipavlou A. 99mTc-PP and 67Ga imaging following insertion of orthopedic devices. Radiology Dec 1979;133(3 Pt 1):717–21. [7] Owen RJT, Harper WM, Finlay DB, Belton IP. Isotope bone scans in patients with painful knee replacements: do they alter management? Br J Radiol Nov 1 1995;68(815): 1204–7. [8] Hunter J, Hattner R, Murray W, Genant H. Loosening of the total knee arthroplasty: detection by radionuclide bone scanning. Am J Roentgenol Jul 1 1980;135(1):131–6. [9] No authors listed. Oxford™ Partial Knee: Manual of the Surgical Technique. http:// www.biomet.nl/resource/5980/Oxford-Knee-Optec.pdf. [10] Gulati A, Chau R, Simpson DJ, Dodd CAF, Gill HS, Murray DW. Influence of component alignment on outcome for unicompartmental knee replacement. Knee 2009;16: 196–9. [11] Nassiri M, Quinlan JF, O'Byrne JM. Analysis of 13 early failures of the mobile bearing Oxford phase III unicompartmental knee prosthesis. 10Eur J Orthop Surg Traumatol 2009;20(4):303–7. [12] Murray DW, Goodfellow JW, O'Connor JJ. The Oxford medial unicompartmental arthroplasty: a ten year survival study. J Bone Joint Surg Br Nov 1 1998;80-B(6): 983–9. [13] Walton MJ, Weale AE, Newman JH. The progression of arthritis following lateral unicompartmental knee replacement. Knee Oct 2006;13(5):374–7.