Key words
skeletal-appendicular - digital radiography - radiation safety
Introduction
Plain radiography (PR) of the knee joint is regularly performed during routine follow-up
examinations after total knee arthroplasty (TKA) [1]. Image acquisition is quick, cost-effective and readily accessible. It allows diagnosis
of most postoperative complications: septic and aseptic loosening, periprosthetic
fractures as well as material failure or wear. For the most part, assessment of these
complications is based upon the evaluation of the metal-bone-cement interface when
loosening is suspected, and on the position of the implant components with respect
to one another, e. g. in cases of inlay wear. This poses particular challenges to
plain radiography due to its limited ability to penetrate metallic components. Thus
computed tomography (CT) is gaining importance in imaging; however this imaging method
is linked to essentially higher radiation exposure [2]. This is significant, since for years younger patients have increasingly been candidates
for a total joint replacement, for example in cases of post-traumatic osteoarthritis
or juvenile rheumatoid arthritis, and thus must undergo repeated CT scans due to extended
prosthetic joint service life [3]
[4]. For this reason cumulative radiation exposure even in cases of endoprostheses has
increased in importance. Although PR requires a significantly lower dose than a CT
of the same region of the body, frequent repetition of this examination during follow-up
requires a careful consideration of total radiation exposure.
In this context, reduction of exposure during follow-up examinations of the knee joint
in digital PR is the primary outcome criterion of this study. Dose reduction is possible
in this case, since the reference values of the German Federal Office for Radiation
(BfS) are organ-specific and not specific to indications [5]. The radiation dose for a follow-up after TXA may have other requirements than for
tumors screening, for example. In addition to the primary outcome criterion, quality
measures developed by orthopedists and trauma surgeons are introduced which assess
the quality of the radiograph in terms of orthopedic and surgical evaluation standards.
Since conventional film-screen systems influence the BfS reference values, the recognized
higher quantum efficiency of digital imaging systems should allow additional dose
reduction [6]
[7].
Materials and Methods
Study Population
Prior to its inauguration, the study was positively evaluated by the Ethics Committee
of Heidelberg and the BfS. Between 06/2011 and 08/2014a total of 278 digital PR follow-up
images of the knee joint after TKA were assessed (170 women, 108 men). The median
age was 67 years (23.3 – 86.9 years). Both branches of the study were balanced with
respect to sex and age (sex: p = 0.15; age: p = 0.91).
Radiographic images
The projection radiographic images were acquired using a flat detector system consisting
of an X-ray tube (SRO 33 100) with generator (Optimus 50) and digital flat detector
(“Digital Diagnost”, all by Philips Healthcare, Best, Netherlands). The image receptor
format was 43 × 35 cm, and the image voltage was 66 kV; a scattered radiation grid
(r8 26/cm) was employed. The X-ray image was obtained in supine position with ovarian
shield or testicle pouch used on the patient.
The guidelines of the German Medical Association indicate speed class SC 400 for PR
of the knee joint. This concept, based on standard film-screen systems, corresponds
to dosage indicator S in digital radiography. This is a device-specific value which,
under identical imaging conditions, correlates with the image receptor dose. For images
with reduced dose, the dosage indicator was doubled, relying on similar studies [8]
[9], thus corresponding to an SC 800 sensitivity class.
Image Analysis and Quality Criteria
Four quality criteria for the radiographic images were developed jointly by orthopedists
and traumatologists ([Table 1, ]
[Fig. 1]). They reflect the requirements for an X-ray image after TKA in the clinical practice
and are especially important for the assessment of implant-associated complications.
Table 1
Quality criteria for the assessment of radiographic images of the knee after metal
implantation.
|
no.
|
short form
|
parameters
|
assessed structures
|
|
1
|
interface
|
bone implant and bone-cement interface
|
femoral and tibial prosthetic interface – evaluation of loosening seams, stress shielding
and fractures
|
|
2
|
surface
|
surface characteristics
|
assessment of the prosthesis structure and shape
|
|
3
|
components
|
differentiation of various implant components
|
differentiation of femoral shield, polyethylene inlay and tibial plateau – assessment
of luxation, material fracture and wear.
|
|
4
|
PHO
|
periarticular heterotopic ossification
|
assessment of periarticular heterotopic ossification of adjoining musculature and
soft tissue.
|
|
score
|
|
1
|
fully assessable
|
|
2
|
≥ 50 % assessable
|
|
3
|
< 50 % assessable
|
|
4
|
not assessable
|
Fig. 1 Illustration of quality criteria Interface 1, Surface 2, Components 3 and Periarticular Heterotopic Ossification 4.
-
Interface: metal-cement-bone interface for the evaluation of septic or aseptic loosening.
-
Surface quality: surface and shape of the implant.
-
Implant components: differentiation among implant components, especially polyethylene
(PE) inlay.
-
Periarticular ossification: Delineation of periarticular heterotopic ossification
(PHO).
Two radiologists with respectively 14 and 8 years expertise in musculoskeletal radiology
evaluated the images using a score ranging from 1 (fully assessable) to 4 (not assessable)
without knowledge of the applied dosage indicator. A radiographic image was rated
“not assessable” if at least one criterion was given a score of 3 or greater, or if
more than two criteria were evaluated with a score of 2. The assessment was performed
using the clinic’s PACS unit (Centricity PACS 3.2, GE Healthcare, Barrington, Illinois).
Study Design and Statistics
Using the rate of assessable images, this prospective, randomized, two-arm, monocentric
blinded study reviewed the non-inferiority of digital PR of the knee joint with reduced
dosage compared to the standard dosage. Block randomization with a 1:1 ratio was performed
in both groups.
Statistical design was performed as a non-inferiority study, i. e., the null hypothesis
of the non-sufficient rate is formally tested unilaterally against the alternative
of the sufficient rate lying close to the standard [10]
[11]. The primary measure of the outcome – rate of assessable images with a standard
dose (RS) or reduced dose (RR) – was defined at 0.9 and is based on the drop-out rate in the so-called Paris scheme
[12], as well as in previously published works regarding dose reduction with a similar
study design [8]
[9]. A difference of Δ = 0.1 in the rate of assessable images is still considered acceptable.
This means that when compared, RR should be seen equivalent to RS, if it can be statistically shown that RR is less than Δ than under RS.
The four quality criteria, interface, components, surface and PHO were described separately
for both dosage groups and are based on the evaluations of the radiologist with the
greater expertise. Exploratory statistical tests were performed to compare both arms
of the study: χ2 test or Fisher’s exact test, if the prerequisites for the X2 test had not been met. In addition, the concurrence of both radiologists was considered
with respect to both the assessability of the images as well as the individual quality
criteria. Dichotomous features were qualified using Cohen’s κ coefficient. The weighted
κ coefficient was used to calculate features with more than two levels, and Bowker’s
symmetry test was applied. The degree of agreement was defined based on the classification
by Landis and Koch [13]
[14].
Determining radiation exposure
In digital projection radiography, the effective dose as typical measure of radiation
exposure to a patient is calculated by determining the dose area product (DAP) multiplied
by a conversion factor that is dependent on the examined region of the body and the
technical parameters of the radiographic image. Since such a conversion factor is
known for the extremities, and despite thorough research, could not be reliably determined,
the actual exposure in both groups was correlated based on the dose area product measured
at the image receptor.
Results
Primary Outcome Criterion
All images, whether using standard or reduce dose, were assessable according to the
applied criteria ([Fig. 2]). Consequently no difference between the two radiation dosages can be observed in
the rate of evaluable images; the unilateral 97.5 % confidence interval for the differential
rate of assessable images between the reduced dose and standard dose is shown as [–0.014,
1.00]. The non-inferiority limit of -0.1 is not included in the 97.5 % confidence
interval. It follows that the rate of evaluable images acquired with a reduced dose
is not less than the rate of those acquired with a standard dose (p < 0.001).
Fig. 2 Images with normal dose a and reduced dose b compared: the named quality criteria are fully assessable for both 76-year-old female
patients.
Quality Criteria
With respect to the “surface” criterion, the proportion of fully assessable images
was marginally greater in the standard dose group compared to those acquired with
reduced dose, whereas comparable proportions were demonstrated in both groups for
the “PHO” criterion. Regarding the “interface” and “components” criteria, the reduced
dose group demonstrated a marginally higher proportion of fully assessable images
([Table 2, ]
[Fig. 3], [4]). Fisher’s exact test is non-significant for all four criteria, i. e., the null
hypothesis of the independence of judgement and group membership cannot be rejected
at the 5 % level.
Table 2
Evaluation of individual quality criteria.
|
|
standard dose (SC 400)
|
reduced dose (SC 800)
|
p-value[1]
|
|
interface
|
fully assessable
|
143 (97.3 %)
|
129 (98.5 %)
|
0.69
|
|
≥ 50 % assessable
|
4 (2.7 %)
|
2 (1.5 %)
|
|
components
|
fully assessable
|
141 (95.9 %)
|
127 (96.9 %)
|
0.75
|
|
≥ 50 % assessable
|
6 (4.1 %)
|
4 (3.1 %)
|
|
surface
|
fully assessable
|
146 (99.3 %)
|
129 (98.5 %%)
|
0.6
|
|
≥ 50 % assessable
|
1 (0.7 %)
|
2 (91.5 %)
|
|
PHO
|
fully assessable
|
145 (98.6 %)
|
129 (98.5 %%)
|
1.0
|
|
≥ 50 % assessable
|
2 (1.4 %)
|
2 (91.5 %)
|
1 Fisher’s exact test.
Fig. 3 Comparison of the delineation of non-radiopaque polyethylene inlays (Criterion no. 3:
“Components”). In image a, the inlay of a 73-year-old male patient is well defined as an increasingly transparent
structure (black arrows). In addition, loosening of the interface can be seen medially.
In image b of a 75-year-old female patient, localization of the inlay is not possible (*). a was acquired with reduced dosage, b with standard dosage.
Fig. 4 Pronounced periarticular heterotopic ossification (white arrows) in a reduced-dose
radiographic image of a 52-year-old male patient after post-traumatic osteoarthritis
and several replacement operations due to loosening of the prostheses.
In all cases the assessability of each criterion was greater than 50 %, so that the
evaluation scores “3” and “4” were not given.
Agreement of both Radiologists
All images were considered to be “assessable” by both radiologists (total agreement:
100 %). With respect to the individual quality criteria, there is substantial concurrence
in the evaluation of interface (total agreement: 99.28 %; κ-coefficient: 0.8, 95 %
CI [0.52 – 1.00]), and weak moderate agreement regarding “components” and “surface”
(total agreement: 96.6 % and 98.56 %; κ-coefficient 0.59, 95 % CI [0.23 – 0.95] and
[0.35 – 0.83]). Since one radiologist used only the “completely assessable” rating
for the “PHO” criterion, calculation of the kappa coefficient was omitted (total agreement:
98.56 %).
Radiation Exposure
Average DAP was 1.537 μGy*m2 for reduced-dose images and 2.432 μGy*m2 for images acquired with the standard dose. Presuming a comparable correlation of
DAP and effective dose in both groups, a dosage reduction of 37 % can be assumed,
especially in the study population in which the same body region (knee joint) was
investigated.
Discussion
In all areas of diagnostic application of radiation, radiologists are required to
employ the lowest possible dosage for an examination that can obtain a suitably assessable
image for the medical issue (ALARA principle: as low as reasonably achievable). This
effort is supported and advanced by the recent European campaign “EuroSafe Imaging”
(www.eurosafeimaging.org) of the European Society of Radiology (ESR) [15]. However, the lowest required dose is not a fixed value and is dependent upon the
body region and examination indications as well as current technical advances. This
is implemented in the guidelines of the German Medical Association so that new reference
values obtained statistically from all medical users of radiation can be taken into
consideration, particularly the dissemination of technical innovations [5]. Our work addresses the above-named influences on the lowest required dosage using
the experimental approach of a prospective randomized study design. The establishment
of quality criteria targets an indication-specific evaluation; dose reduction in digital
projection radiography should take into account these criteria compared to conventional
film-screen systems.
Based on existing results, a 37 % dose reduction in PR of the knee joint after TKA
is possible without measurable loss of quality with respect to important indications.
All radiographic images – with or without reduced dosage – were assessable with respect
to the clinical issues.
The developed quality criteria are aimed specifically at the diagnosis of septic or
aseptic loosening as well as material failure, e. g., abrasion of the PE inlay or
inlay luxation. Both complications represent an absolute indication of surgery; therefore
the quality of the radiographic examination is highly important in such cases [1]. PHO can be observed among approx. 25 % of patients after prosthesis implantation,
and can result in postoperative motion limitations [16]. Likewise, the statistical analysis of the individual criteria showed this to be
independent of the dose. The important “interface” and “components” criteria were
marginally better evaluated, underscoring the non-inferiority of this study arm.
Comparison with similar studies of PR of the full spine and full leg images further
suggest another conclusion. In these studies, non-inferiority was statistically confirmed;
however, unlike our work, there were also non-assessable examinations [8]
[9]. The proportion of these was higher in the reduced-dosage groups compared to the
standard groups. On the other hand, assessability of 100 % in our current study makes
the potential of an additional dose reduction of more than 37 % likely.
Although there are no mandatory national or European guidelines regarding the frequency
of PR follow-up examinations after TKA, the preferred standard is four follow-ups
within the first two years after surgery and then at least a two-year checkup interval
over the entire service life of the prosthesis. To our knowledge, there have been
no comparable studies dedicated to consideration of dose reduction after TKA. Even
though the examined area is distant from the body trunk, it should be noted that the
stochastic radiation effect is independent of the total dosage in its magnitude. However,
the likelihood of this occurrence is influenced by any dose reduction.
Our study statistically confirmed the non-inferiority of post-TKA digital radiographs
with an image receptor dose reduced by 37 % when compared to the standards of the
German Federal Office for Radiation. In addition, with regard to further technical
developments such as reduced-dose biplanar stereo radiography systems [17]
[18], the possibility of prospectively guided, statistically grounded quality assurance
is demonstrated.
Digital radiographic images of the knee joint after endoprosthesis implant with a
37 % reduced image receptor dose are not inferior to those acquired with the standard
dose.
All physical structures required to evaluate the clinical outcome can be sufficiently
assessed with sensitivity class SC 800.
Conclusion
We recommend sensitivity class SC 800 as the new reference value for digital radiographic
images of the knee joint with the above-mentioned indication.
