“Beware the consequences of your imaging.” This admonition served as a warning to me from a radiologist early in my sports medicine career, especially with the growing sophistication of imaging techniques. “Beware the consequences of your imaging” is radiologists’ jargon for “treat the patient not the imaging.” This fact has been made abundantly clear while reviewing magnetic resonance imaging (MRI) results with patients. A thorough history and examination suggests a limited differential diagnosis. The MRI is used to either confirm a diagnosis for surgical planning or clarify treatment options. The other findings are “incidental.” “MRIographers” have been aware of this since the dawn of MRI. As scanners have become “stronger” (now up to 3 T), more clinically insignificant findings likely are to be seen. The purpose of this article is to investigate this phenomenon to remind the clinician of the limitations of advanced imaging and to reaffirm the importance of “treating the patient, not the imaging.”
To help me place the MRI and other radiographic techniques in proper perspective, I performed a review of representative MRI studies using OVID, Pub Med, and E-Journals and bibliographies of the representative studies investigating “abnormal findings” in asymptomatic individuals. To permit the clinician to place his/her clinical suspicions in the context of MRI abnormalities, I also reviewed representative manuscripts that compared the accuracy of abnormal findings with direct, arthroscopic visualization as another measure of accuracy. Many populations are select groups of athletes while others reflect the general population. Joints reviewed include hip, knee, shoulder, elbow, and back.
An asymptomatic group of male hockey players underwent MRI of the hip and pelvis. Sixty-four percent of them demonstrated pathologic findings. More specifically, 36% showed hip adductor tendon or rectus abdominis abnormalities (16). This poses particular concerns in interpretation given the frequency of groin pain in this population.
In another group of asymptomatic volunteers (52 hips in 46 subjects) representing a broad range of ages (15 to 85 years), hip MRI with a focus on the acetabular labrum showed a large number of labral irregularities — 58% of the hips showed high-intensity labral changes, 10% showed no anterior superior labrum at all, four subjects had findings consistent with labral tears, and three had intralabral cysts (6). The introduction of gadolinium for magnetic resonance arthrography (MRA) has improved significantly the accuracy of interpretation of labral pathology. An additional study of 30 participants with symptoms and physical examination findings suspicious for acetabular labral tears underwent MRA (4). MRA demonstrated that five appeared normal while 25 were abnormal. Seventeen of the 25 with imaging abnormalities had arthroscopy (commonly accepted gold standard). Sixteen clearly demonstrated acetabular labral tears and one showed a synovial process as a normal variant. The final score was a sensitivity of 100% and specificity of 94%. The agreement between clinical findings and MRA suggests this is the study of choice when labral pathology is suspected.
Recent trends in hip/groin injury have raised awareness of the “trendy” diagnosis of femoroacetabular impingement (FAI). Is the diagnosis a radiographic or clinical diagnosis? Fully understanding, evaluating, and treating FAI is in an adolescent stage. There are no long-term outcome studies to verify currently accepted conclusions or interventions. Radiography underscores our limited level of understanding. Hip x-rays were obtained on 200 asymptomatic subjects, aged 21 to 50 years (55% women) (8). Fourteen percent had camera deformities (10.5% bilateral and 3.5% unilateral). Of those with positive findings for hip impingement (reduced hip internal rotation and positive flexion-adduction-internal rotation test), 79% were men. Further complicating the radiographic implications of positive findings is a more recent study from the British Journal of Sports Medicine in which 34 subjects with examination findings suspicious for FAI were evaluated with radiographs (18). Those radiographs were compared with an asymptomatic matched control group. Eight commonly accepted x-ray criteria for impingement were assessed. Sixty-four of 68 hips in the symptomatic group showed positive findings for a prevalence rate of 94%. The mean number of positive radiographic signs in the symptomatic group was 1.8 (range of 0 to 4). Is this clinically useful information? The asymptomatic control group had a mean number of positive x-ray signs of 1.96 (range of 0 to 5). Furthermore there was no correlation between a positive flexion-adduction-internal rotation test and radiographic signs. Finally hip range of motion had no correlation with the number of positive x-ray findings. Clearly this information is of limited value and further clouds the differential diagnosis of the athlete with groin pain.
Shoulder pain and problems tend to increase in frequency with age, so do the presence of MRI abnormalities in the asymptomatic shoulder. Sher et al.’s (15) landmark study of asymptomatic shoulders highlights some of the issues of advanced imaging for the shoulder. If a patient is younger than 40 years, the MRI is accurate. In asymptomatic shoulders under the age of 40 years (N = 25), there were no rotator cuff tears and only one (4%) partial thickness tear. With advancing age, however, abnormalities increase. In those aged 40 to 60 years (N = 25 asymptomatics), one (4%) had a complete rotator cuff tear and six showed partial thickness tears. After the age of 60 years (N = 46 asymptomatic individuals), rotator cuff tears were seen in 28% and partial thickness tears in 24%.
What abnormal changes, if any, might occur in the throwing athlete? Connor et al. evaluated 20 asymptomatic throwing athletes, both dominant and nondominant shoulders (5). Forty percent showed either full or partial thickness tears on MRI in the dominant shoulder while the nondominant shoulder showed no rotator cuff pathology. Clearly repetitive activity alters the structure of the rotator cuff, but not the symptoms. More importantly, at 5-year follow-up, all the throwers remained symptom free. Asymptomatic MRI findings were of no prognostic value.
Taking a small detour from the MRI, consider a recent study comparing the accuracy of MRI to musculoskeletal ultrasound in shoulder diagnosis (14). This is of particular interest given the growth of ultrasound in point-of-care diagnosis and treatment. Of over 5,000 shoulder ultrasounds, 275 had MRI. Sixty-five that had both US and MRI within 5 months of surgery were evaluated for accuracy with the surgical diagnosis serving as the gold standard. The differences in US and MRI diagnosis were not statistically significant. Ultrasound had a higher sensitivity for rotator cuff tears than MRI, but MRI proved more accurate for intraarticular pathology.
Be cautious in interpreting advanced imaging in the thrower’s shoulder and the aging shoulder. Both are likely to show pathology that may have little clinical relevance.
Interpreting spine MRI is fraught with problems of similar magnitude to the shoulder. As the neck and back age, asymptomatic imaging abnormalities increase. There are a number of representative investigations supporting this observation. In a group of 98 symptom-free subjects (50 men and 48 women), with an age range of 20 to 80 years, over 50% had at least one bulging disk (11). Twenty-seven percent had disk protrusions and one with a disk extrusion. Over one third had abnormal findings at more than one disk level. Age was a definite factor in the likelihood of abnormalities. After the age of 50 years, two thirds had more than one abnormal finding, whereas under the age of 50 years, the frequency of multiple abnormalities dropped to 27%. Dissecting this population even further showed that those who exercised regularly had a higher rate of L5–S1 disk protrusions compared with the sedentary group — 16% to 4%. Remember, all subjects were asymptomatic.
Other age-related data support these results. In a small investigation involving 67 subjects who never had back symptoms, the group over 60 years old showed a 36% occurrence of disk herniation and 21% incidence of spinal stenosis (3). Only a single subject over the age of 60 years demonstrated no disk bulging or herniation. Contrast this with the group younger than 60 years who showed a rate of disk herniation of 20% and only a single case of spinal stenosis.
An investigation involving a slightly younger asymptomatic population, with an age range of 20 to 50 years, still showed frequent MRI abnormalities (19). The most common abnormal findings included 62% to 67% disk bulging or protrusion, 32% to 33% annular tears, and 18% disk extrusion. Interestingly none of this group had evidence of severe osteoarthritis.
The asymptomatic active population has frequent cervical and lumbar spine abnormalities. One such study involving 19 active male athletes with an average age of 53 years (range, 41 to 69 years) showed few were spared imaging changes (9). Only three lumbar and four cervical MRI results were considered normal. The lumbar irregularities were 11 disk protrusions (one had four levels of involvement), four posterolateral herniations, and one severe spinal stenosis. The neck showed similar changes with 11 subjects demonstrating foraminal stenosis and seven showing protruding disk/osteophyte complexes.
Like the patient with shoulder pain, the patient with back pain requires a thorough history and examination to place the imaging results in the proper diagnostic context.
The other upper extremity joint that is imaged frequently because of sports-related pain is the elbow, which is subject to significant stresses during the throwing motion. Medial elbow pain, implicating the ulnar collateral ligament, is a significant concern in the throwing athlete. What are the consequences of imaging the elbow? Consider, first, the high school thrower who is symptom free. A recent investigation of 23 high school age pitchers (average age, 16 years) underwent MRI of the dominant elbow, using the nondominant elbow as a control (10). Only three (13%) had normal-appearing dominant elbows. Almost two thirds (15/23) of the elbows showed thickening of the ulnar collateral ligament with four also demonstrating edema of the sublime tubercle. Furthermore 14 of 23 showed subchondral sclerosis of the ulnotrochlear joint, and slightly over one half of them also showed ulnotrochlear osteophytes. The nondominant elbow had no MRI changes. With the high incidence of abnormal MRI in the young thrower, one must be careful in interpreting imaging in the painful elbow. Beware the consequences!
Moving further up the “skill” ladder to the professional pitcher, accompanied by more maturity and greater volume of throws, consider imaging of the symptomatic elbow at this level. Again, the nondominant elbow was used as the control. Sixteen professional pitchers, completely asymptomatic and no prior symptoms of elbow pain, obtained MRI (13). Fourteen of 16 (87%) had thickening, signal heterogeneity, or discontinuity of the ulnar collateral ligament. Eighty-one percent also showed findings consistent with posteromedial impingement of the ulnotrochlear joint. Given these abnormalities observed in asymptomatic throwing athletes of all skill levels, an accurate clinical diagnosis is paramount.
Much of the literature regarding knee MRI has been directed toward accuracy of MRI diagnosis compared with arthroscopic diagnosis. One interesting investigation, however, compared symptomatic to asymptomatic knees to determine common MRI findings in the symptom-free knee. One hundred subjects (59 men and 41 women) aged 18 to 73 years presented with one symptomatic and one asymptomatic knee (20). Both were imaged with MRI. Of 100 symptomatic knees, MRI demonstrated 57 meniscal tears. In the 57 symptomatics, the asymptomatic contralateral knee had 36 meniscal tears. Interestingly the 43 symptomatic knees that had no MRI evidence of meniscal tears also had no meniscal tears in the asymptomatic knee. Other findings in the symptomatic knees included collateral ligament abnormalities in 53 symptomatic knees and only six of the knees without any symptoms. Soft tissue changes in the pericapsular tissues occurred in 64 symptomatic knees and 12 asymptomatics. Bone marrow edema changes were present in 36 of the symptomatic knees, and only three of those who were symptom free. Radial, vertical, complex, and displaced meniscal tears were most likely present in the symptomatic group of knees, whereas horizontal and oblique tears were to be asymptomatic more likely. Coupled with the specific symptomatic meniscal tears (radial, vertical, complex, and displaced), collateral ligament changes, pericapsular soft tissue abnormalities, and bone marrow edema are likely to be clinically significant.
The knee of adolescent athletes is not immune to the false-positive findings on MRI. Twenty-eight 14- to 15-year-old soccer players had MRI performed on both knees with those results compared with an age-matched less active group of adolescents (17). Eighteen of the soccer players had one or more irregularities on their studies. Half of them had bone marrow edema in at least one knee compared with only one in the control group. Various other MRI abnormalities observed included popliteal cysts, ganglion cysts, and infrapatellar fat pad edema.
With the increase use of MRI by sports medicine clinicians, accuracy of MRI diagnosis in the knee is a significant concern. Accuracy rates vary considerably and are likely due to the experience of the interpreting radiologist. A study conducted on recruits from the Israeli military reflects a potential problem. While the radiologists conducting the study claim 90% accuracy in their specialized musculoskeletal practice, the accuracy was not as good in recruits whose military records were reviewed at entry into the military (2). The diagnosis at arthroscopy was compared with the diagnosis on MRI. One hundred thirty-nine recruits with knee pain and subsequent surgery met the criteria for review. The results are listed in Table 1. Compared with the specialized musculoskeletal radiology practices, the accuracy rates in other community situations ranged from a low of 52% for medial meniscal tears to 82% for lateral meniscal tears and 80% for anterior cruciate ligament tears. This investigation was actually performed to show the frequency of inappropriate surgery but effectively identified potential MRI interpretation problems.
Another investigation into MRI accuracy compared with arthroscopic diagnosis highlighted MRI interpretation issues. Seventy subjects aged 22 to 59 years (41 men and 29 women) presented with the complaints of knee pain, swelling, popping, locking, and instability (1). Their physical examination findings also were consistent with intraarticular injury. The key results for various injuries are summarized in Table 2. Obviously in the normal clinical situation, the MRI results are viewed in the context of the history and clinical examination to reduce the futility of using only the MRI for diagnosis.
This, then, begs the question. How good is the clinician in examining the patient and establishing a diagnosis? Pretty good, at least for the younger knee, according to Esmaili Jah et al. (7). Over a 2-year time span, two experienced clinicians examined 110 acutely injured knees (average age of the subjects = 27.9 years) and compared their clinical diagnoses (history and examination) with MRI and subsequent arthroscopy. Compared with arthroscopy, ACL tear by clinical examination was accurate at 91.4% of the time while MRI had an accuracy rate of 88.5%. Likewise, the clinical examination faired well for other injuries: PCL tears (examination, 100%; MRI, 94.6%), medial meniscus injuries (examination, 96.9%; MRI, 85.9%), and lateral meniscus injuries (examination, 85.4%; MRI, 73.8%), demonstrating the vital importance of the clinical examination in knee injury evaluation. However if there were combined ligament and meniscal injuries, the MRI proved more accurate than the clinician’s examination.
Consequently with many false-positive and false-negative results in MRI evaluation of the injured knee, the history and physical examination continue to be critical in interpreting imaging results and recommending treatment.
Bone Marrow Edema
Bone marrow edema usually is interpreted as a sign of bone stress reaction or stress fracture. Again, this finding is quite common in active populations and requires supporting history and examination for a correct diagnosis. An evaluation of 21 elite Finnish military recruits who underwent 25 h of physical training per week as part of their military training probably represents the amount of training of many of our competitive young athletes (12). This group had bone marrow edema by MRI as a common observation. MRI was performed at induction, 6 wk, and 5 months. At 5 months, 16 of 21 demonstrated 75 stress injuries. Of those 16, nine were asymptomatic. In conclusion, athletes who perform a high volume of training have a number of subclinical bone stress injuries. Again, the importance of clinical correlation to presenting symptoms is critical.
Interpretation for the Patient
More difficult, however, is discussing the MRI results with the patient who expects a specific, definitive explanation for the painful condition/injury. To describe an MRI abnormality as clinically relevant versus an incidental finding may be easy for the experienced clinician to understand but a most difficult concept for the injured patient. Phrases that may be used to prepare the patient for this discussion include the following:
- “The MRI is only a picture describing what looks ‘normal versus abnormal’.”
- “The MRI doesn’t tell what is causing your pain.”
- “The MRI can see things that aren’t causing your problem and can miss things that are.”
- “The MRI doesn’t tell what needs surgery.”
MRI has added immensely to our diagnostic repertoire. But due to the existence of frequent abnormalities that may not be clinically significant, the value of an accurate clinical examination has not been diminished. Some primary care sports medicine physicians have declared “the MRI is the non-orthopedist’s arthroscope.” This is not true. The MRI is a tool that is used in conjunction with the injury history and examination. The aforementioned studies are proof. Direct visualization is still an unchallenged gold standard. It is crucial for the clinician to know the limitations of the MRI.
In review, the asymptomatic hip shows many abnormalities. Up to two thirds of asymptomatic hockey players demonstrate abnormalities, mostly acetabular labral changes. MRA is, however, an accurate diagnostic tool for labral pathology. FAI is not exclusively a radiographic diagnosis.
The older shoulder has many asymptomatic full and partial thickness tears by MRI. Even throwers have asymptomatic MRI pathology. Musculoskeletal ultrasound is a reliable diagnostic tool for rotator cuff tears. MRI is superior for intraarticular pathology.
MRI changes in the dominant elbow of throwers are common. High school throwers are included in this group.
MRI in the knee has frequent asymptomatic medial and lateral meniscus changes. In the setting of single structural abnormalities, your examination is slightly more accurate than the MRI.
The older spine is replete with asymptomatic structural abnormalities, especially beyond 50 years of age.
Finally your physically active, asymptomatic population has frequent MRI evidence of marrow edema consistent with bone stress reactions.
As a clinician, it is important to know the limitations as we guide the injured athlete/patient to the most accurate diagnosis and most appropriate treatment.
The author declares no conflict of interest and does not have any financial disclosures.
1. Behairy NH, Dorgham MA, Khaled SA. Accuracy of routine magnetic resonance imaging in meniscal and ligamentous injuries of the knee: comparison with arthroscopy. Int Orthop
. 2009; 33: 961–7.
2. Ben-Galim P, Steinberg EL, Amir H, et al.. Accuracy of magnetic resonance imaging of the knee and unjustified surgery. Clin Orthop Relat Res
. 2006; 447: 100–4.
3. Boden SD, Davis DO, Dina DS, et al.. Abnormal magnetic-resonance scans of the lumbar spine in asymptomatic subjects: a prospective investigation. J Bone Joint Surg Am
. 1990; 72: 403–8.
4. Chan YN, Lien LC, Huei-Ling Hsu HL, et al.. Evaluating hip labral tears using magnetic resonance arthrography: a prospective study comparing hip arthroscopy and magnetic resonance arthrography diagnosis. Arthroscopy
. 2005; 21: 1250.
5. Connor PM, Banks DA, Tyson AB, et al.. Magnetic resonance imaging of the asymptomatic shoulder of overhead athletes a 5-year follow-up study. Am J Sports Med
. 2003; 31: 724–7.
6. Cotten A, Boutry N, Demondion X, et al.. Acetabular labrum: MRI in asymptomatic volunteers. J Comput Assist Tomogr
. 1998; 22: 1–7.
7. Esmaili Jah AA, Keyhani S, Zarei R, et al.. Accuracy of MRI in comparison with clinical and arthroscopic findings in ligamentous and meniscal injuries of the knee. Acta Orthop Belg
. 2005; 71: 189–96.
8. Hack K, DiPrimio G, Rakhra K, et al.. Prevalence of cam-type femoroacetabular impingement morphology in asymptomatic volunteers. J Bone Joint Surg Am
. 2010; 92: 2436–44.
9. Healy JF, Healy BB, Wong WH, Olson EM. Cervical and lumbar MRI in asymptomatic older male lifelong athletes: frequency of degenerative findings. J Comput Assist Tomogr
. 1996; 20: 107–12.
10. Hurd WJ, Eby S, Kaufman KR, et al.. Magnetic resonance imaging of the throwing elbow in the uninjured, high school-aged baseball pitcher. Am J Sports Med
. 2011; 39.
11. Jensen MC, Brant-Zawodzki MN, Obuchowski N, et al.. Magnetic resonance imaging of the lumbar spine in people without back pain. New Engl J Med
. 1994; 331: 69–73.
12. Kiuru MJ, Niva M, Reponen A, et al.. Bone stress injuries in asymptomatic elite recruits: a clinical and magnetic resonance imaging study. Am J Sports Med
. 2005; 33: 272–6.
13. Kooima CL, Anderson K, Craig JV, et al.. Evidence of subclinical medial collateral ligament injury and posteromedial impingement in professional baseball players. Am J Sports Med
. 2004; 32: 1602–6.
14. Rutten MCJ, Spaargaren G_J, von Loon T, et al.. Detection of rotator cuff tears: the value of MRI following ultrasound. Eur Radiol
. 2010; 20: 450–7.
15. Sher JS, Uribe JW, Posada A, et al.. J Bone Joint Surg
. 1995; 77: 10–5.
16. Silvis ML, Mosher TJ, Smetana BS, et al.. High prevalence of pelvic and hip magnetic resonance imaging findings in asymptomatic collegiate and professional hockey players.
17. Soder RB, Simões JD, Soder JB, et al.. MRI of the knee joint in asymptomatic adolescent soccer players: a controlled study. AJR Am J Roentgenol
. 2011; 196: W61–5.
18. Weir A, deVos RJ, Moen M, et al.. Prevalence of radiological signs of femoroacetabular impingement in patients resenting with long-standing adductor-related groin pain. Br J Sports Med
. 2011; 45: 6–9.
19. Weishaupt D, Zanetti M, Hodler J, et al. MR imaging of the lumbar spine: prevalence of intervertebral disk extrusion and sequestration, nerve root compression, end plate abnormalities, and osteoarthritis of the facet joint in asymptomatic volunteers. Radiology
. 1998; 209: 661–8.
20. Zanetti M, Pfirrmann C, Schmid M, et al.. Patients with suspected meniscal tears: prevalence of abnormalities seen on MRI of 100 symptomatic and 100 contralateral asymptomatic knees. AJR Am J Roentgenol
. 2003; 181: 635–41.