|
ABSTRACT
This
study compares the MRI scan with clinical diagnosis, and looks
at the reliability of both these against the gold standard of
arthroscopic diagnosis, in disorders of the knee.The
patients were recruited over a two year period. The total number
of patients in this study was 58 , with ages ranging from 18-63
years .The results were obtained using statistical formulae,
which calculated the sensitivity , accuracy and the positive and
negative predictive values of the MRI scan. In this study, the
reliability of a high negative predictive value for the MRI scan
and the inadequacy of the MRI in diagnosing articular cartilage
(chondral) lesion was demonstrated.
Key
Words:Magnetic Resonance imaging , Arthroscopy , Knee
J.Orthopaedics 2005;2(4)e5
Introduction
Internal derangement s of
the knee is a commonly presenting clinical condition to the
Orthopaedic practitioner. They account for a large number of
referrals to hospitals , both from the general practitioner but
also from the accident and emergency department . Other than
being very troublesome for the patients day to day activities,
it has a very significant financial and medico-legal
implication. The clinical evaluation of knee injuries remains a
difficult problem. The accuracy of a clinical diagnosis,
reported in various series, varies between 64-85 percent . This
means that even in the most experienced hands , a clinical
diagnosis cannot be ascertained in about 20 percent of cases.
Magnetic resonance imaging (MRI) is a diagnostic method quite
frequently used in diagnosis of internal derangements of the
knee, its advantages being that its non-invasive , painless and
no risk of radiation. However it’s an expensive investigation
,and like all investigations , it has a tendency to be misused
and overused, to confirm diagnosis before proceeding with
surgical intervention. The accuracy rates of MRI scans also
varies. As knee disorders are so common and with varied accuracy
rates of scans , it would be useful to know the accuracy of the
MRI findings and correlate this with clinical diagnosis and
arthroscopy findings.
Arthroscopy has been used
for many years as a diagnostic and therapeutic tool in knee
disorders. It is considered a gold standard for the same as it
allows direct visualization of the interior of the knee.
Although there have been studies in literature comparing MRI
with arthroscopy, there have been no published study comparing
MRI with clinical diagnosis , in relation to the gold standard
of arthroscopy.
The aim of this study was
to find out the diagnostic accuracy of MRI scans as compared to
the clinical diagnosis, by correlating them with the gold
standard of arthroscopy.
Methodology
This was a retrospective
study of patients who had undergone both knee arthroscopy and a
MRI scan prior to it. The study included patients of all ages
operated under the care of a senior consultant in our tertiary
referral center. The period of study was 2 years. Data was
obtained from the databases of the theatre and the radiology
department . These were matched to obtain the patient group to
be studied.
There were 58 patients
(n=58) , who formed the basis for the study.
The pertinent data of each
patient was entered into a proforma sheet . In addition to the
MRI and Arthrosopy findings obtained from the case-notes , the
clinical impression at the outpatient appointment was also
documented . The core data was then analysed as true positives
and negatives and false positives and negatives . Using these,
the accuracy , sensitivity , specificity , negative and positive
predictive values were calculated, with arthroscopy as the
standard for comparison.
Results
Age distribution
-
Range : 18-63 years
-
<20 years – 2
-
20-40 years – 33
-
>40 years - 23
Accuracy of the MRI scan
|
Medial meniscus |
94.8% ( 95% C.I 89-100)
, p<0.05 |
|
Lateral meniscus |
89.6% (C.I. 82-98),
p<0.05 |
|
ACL tears |
93.1% (C.I. 86-100),
p<0.05 |
|
Articular cartilage
lesion |
24.0% (C.I. 11-33),
p<0.05 |
* represents statistically
significant results, p<0.05, C.I implies 95% confidence
Intervals
Comparison of
Sensitivity of the MRI scan and clinical diagnosis
This is, probability that a
diseased individual will have a positive result,
|
|
MRI |
Clinical Diagnosis |
|
Medial meniscus |
96.4% (C.I 89-100)* |
85.4% (C.I 70-99) |
|
Lateral meniscus |
81.8% (C.I 59-100) |
90.1% (C.I 81-98) |
|
ACL tears |
87.5% (C.I 64-100) |
56.1%% (C.I 22-89) |
|
Articular cartilage
lesions |
57.1% (C.I 52-68) |
43.2% (C.I 16-69) |
* represents statistically
significant results , p<0.05
Comparison of
Specificity of the MRI scan and clinical diagnosis.
This is, probability that a
disease free individual will have a negative result
,
|
|
MRI |
Clinical Diagnosis |
|
Medial meniscus |
88.1% (C.I 76-99) |
64.1% (C.I 47-80) |
|
Lateral meniscus |
90.2% (C.I 81-98)* |
85.4% % (C.I 75-96) |
|
ACL tears |
94.3%% (C.I 87-100)* |
96.1% (C.I 90-100)* |
|
Articular cartilage
lesions |
83% (C.I 53-100) |
98.1% (C.I 94-100)* |
* indicates significant
results , p<0.05
Negative predictive
value of the MRI
These is the percentage of
patients who were diagnosed as having no tear on MRI and were
subsequently found to have no tear at arthroscopy,
|
Medial meniscus |
96.5% ( C.I 90-100)* |
|
Lateral meniscus |
95.5% ( C.I 89-100)* |
|
ACL tears |
97.9% ( C.I 94-100)* |
|
Articular cartilage
lesion |
45.4% ( C.I 38-56) |
* indicates significant
results, p<0.05
Positive predictive
value of the MRI
These is the percentage of
patients who were diagnosed as having a tear on MRI and were
subsequently found to have a tear on arthroscopy,
|
Medial meniscus |
87.0% ( C.I 78-94) |
|
Lateral meniscus |
64.2% (C.I 56-74) |
|
ACL tears |
70.0% ( C.I 64-76) |
|
Articular cartilage
lesions |
88.8% ( C.I 82-94) |
* indicates significant
results, p<0.05
Discussion
The accuracy, sensitivity
and specificity values for knee lesions vary widely in
literature . In a multicentric analysis , published by Fisher et
al 1 , the accuracy ranged from 64-95% fro medial meniscus ,
from 83-94% for lateral meniscus and 78-97% for the anterior
cruciate ligament . Because of the wide range of reported values
,it was necessary to find out the accuracy rates in our series.
The accuracy in our series was around 95% for medial meniscus ,
90% for lateral meniscus and 93% for ACL tears . In a
prospective study reported by Imhoff et al 2, , the negative
predictive value was 94% but the positive predictive value was
only 54% . They concluded that due to the high negative
predictive value, a normal MRI scan allows to eliminate a
meniscal lesion and so there is no need for a diagnostic
arthroscopy . MRI can also reduce the requirement for a
diagnostic arthroscopy3, only patients with definite clinical
signs merit early therapeutic arthroscopy , but all other knees
should be investigated by MRI . This policy spares patients from
unnecessary and expensive surgery 4. This matched with our study
which showed that MRI scans had a high negative predictive value
and hence can be used to exclude pathology in doubtful or
uncertain cases.
Our study also revealed
that the sensitivity of clinical diagnosis is relatively high,
especially so for medial and lateral meniscal lesions. According
to Spiers et al 5, acceptance of MRI findings could have
resulted in a 29% reduction in number of arthroscopies performed
with a total increase in cost of 1.4%. Our study showed that MRI
scan was not reliable to pick up chondral lesions (articular
cartilage) , with an accuracy of 24% . This matches with other
published literature , like that by Friemert et al 6, which
stated that MRI cannot replace arthroscopy for diagnosis of
cartilage damage and hence arthroscopy still has to be seen as
the method of choice.
Conclusion
Our study found that the
accuracy of the MRI scan in medial meniscal lesions is higher
than for other structures of the knee joint. MRI was not found
to be an accurate investigation for diagnosing and quantifying
chondral damage. We do not recommend the routine use of MRI scan
to confirm diagnosis, as the positive predictive value of the
scan is low for all lesions. In the presence of positive
clinical signs, we would recommend proceeding to arthroscopy.
The negative predictive value of a scan was found to be high for
all structures of the knee joint and hence , a ‘normal’ scan can
be used to exclude a pathology .Thus sparing patients from
expensive and unnecessary surgery and also freeing up valuable
theatre time. In current setting of ease of availability of
expensive diagnostic investigation, we reiterate the importance
of a robust clinical diagnosis.
References:
1.Fisher S P , Fox J , Del Pizzo et al , Accuracy of
diagnosis from Magnetic resonance Imaging of the knee: A
multi-centre analysis of 1014 patients ,J. Bone Joint Surg.(AM)
,1991,73-A:2-10
2.Imhoff A, Buess E, Holder J et al , Comparison between
magnetic resonance imaging and arthroscopy for the diagnosis of
knee meniscal lesions . Rev Chir Orthop reparatrice Appar
Mot.1997;83(3):229-36.
3.Warwick DJ , Cavanagh P , Bell M et al , Influence of Magnetic
resonance imaging on a knee arthroscopy waiting list .J Am Acad orthop Surg. 1996
Mar ;4(2):93-99
4.Carmichael IW, Macleod AM
et al , MRI can prevent unnecessary arthroscopy .Injury . 1993 Jul ;24(6)
:380-2
5.Spiers AS, Meagher T et
al , Can MRI of the knee affect arthroscopic practice? A
prospective study of 58 patients.J Bone Joint Surg
Br.1993Jan ;75(1) :49-52
6.Friemert B ,Oberlander Y
et al , Diagnosis of chondral lesions of the knee joint : can
MRI replace arthroscopy ? A prospective study Knee Surg Sports Traumatol
Arthrosc. 2004 Jan ;12(1):58-64
|
