TO THE EDITOR:
I am writing in reference to "Abnormal Findings on Magnetic Resonance Images of Asymptomatic Shoulders" (77-A: 10—15, Jan. 1995), by Sher et al. The authors are to be commended for performing magnetic resonance imaging on a group of asymptomatic individuals. Not surprisingly, the results showed a clear increase in magnetic resonance imaging findings in the older population: 50 per cent of the individuals who were more than sixty years old were said to have evidence of either a full or a partial-thickness tear of the rotator cuff.
In my opinion, there are some shortcomings in this study. The first is that only oblique coronal images were made for all of the patients. It is well known that suspected abnormal findings on oblique coronal images should be correlated with the findings on oblique sagittal images. The lack of such correlation possibly altered some of the interpretations of partial tears and tendinosis in this study. Furthermore, as the criteria for partial-thickness tears and tendinosis are not entirely elucidated or reliable in the literature, some of the individuals may have had tendinosis instead of a partial tear, thus invalidating the statement that 50 per cent of the individuals had some evidence of a tear of the rotator cuff. It is well known that the aging phenomenon produces findings in the hyaline cartilage, fibrocartilage, and tendons that are identifiable on magnetic resonance imaging. A slightly swollen tendon with increased signal intensity on both proton-density and T2-weighted images may actually represent tendinosis, which may be a normal phenomenon of aging and not a partial tear of the rotator cuff at all.
Another difficulty that I have with the study is that the diagnoses that were made on the basis of magnetic resonance imaging were not confirmed by any other method. This guarantees an inherent bias in the study, which the authors admitted.
Finally, this article stated what any physician who manages patients should know by intuition: the symptoms should be correlated with the findings of any imaging study, particularly magnetic resonance imaging. It is well known that lesions that are morphologically identical on magnetic resonance imaging of the spine can cause extreme symptoms in one patient and no symptoms at all in another. Therefore, any findings on an imaging study should be correlated with the symptoms. In fact, asymptomatic patients should not be imaged at all, so the conclusion of this paper was not surprising by any standards.
In my opinion, which may be biased as I am a diagnostic musculoskeletal radiologist, the fact that magnetic resonance imaging shows a high percentage of abnormalities in asymptomatic individuals does not demean or lessen the importance of magnetic resonance imaging in the evaluation of the shoulder. First, asymptomatic patients probably should not be imaged with any modality. I presume that if asymptomatic patients were examined arthroscopically, they would be found to have a high percentage of abnormalities as well. However, this does not necessarily demean or lessen the impact of arthroscopy. Second, the fact that a high percentage of asymptomatic older individuals had findings on magnetic resonance imaging does not negate the fact that magnetic resonance imaging is non-invasive (unless intra-articular contrast medium is used), relatively inexpensive (compared with arthroscopy as a diagnostic procedure), and the best preoperative imaging test available at present to examine closely the soft tissues surrounding the shoulder.
I have received some negative comments from my orthopaedic colleagues regarding magnetic resonance imaging of the shoulder since the article by Sher et al. was published. I must admit that I resent these criticisms because I believe that the article did not necessarily diminish the importance of magnetic resonance imaging. It proved the obvious, which is that we have a lot to learn about the aging process and many patients exhibit morphological findings with a wide variety of symptoms. Therefore, good medicine is what it has always been: to do no harm to the patient, to treat the symptoms correlated with the findings on the magnetic resonance images, and not to treat the imaging studies. I hope that my orthopaedic colleagues will remember that this is probably the most important conclusion to be drawn from this article and not that magnetic resonance imaging may show morphological abnormalities in patients who are asymptomatic.
Thomas L. Pope, Jr., M.D.: Division of Radiological Sciences, Department of Radiology, The Bowman Gray School of Medicine, Wake Forest University, Medical Center Boulevard, Winston-Salem, North Carolina 27157-1088
Dr. Sher, Dr. Uribe, Dr. Posada, Dr. Murphy, and Dr. Zlatkin reply:
Dr. Pope raises certain issues that lead us to believe that he may have misconstrued the main focus of our investigation. There are essentially two main points that should be imparted by the article. The first is that evidence of a tear of the rotator cuff on magnetic resonance imaging need not interfere with normal and painless function of the shoulder. Second, there are potential hazards when operative decisions are based on the findings of magnetic resonance imaging alone in the absence of adequate clinical correlation.
While it is intuitive that clinicians should treat the patient and not the findings obtained from an imaging study, there is also the potential for magnetic resonance imaging to shape a clinical diagnosis or to influence decisions concerning treatment. Therefore, the possible limitations of magnetic resonance imaging must be understood so that appropriate decisions can be made with regard to patient management. We concur that a given radiographic finding may have a wide spectrum of presentations and therefore must be correlated with clinical findings. However, issues such as inconsistent clinical findings, multiple and variable symptoms, secondary gain, and the experience of the clinician can affect the adequacy of clinical correlation. For patients in whom the diagnosis may not be clear-cut, magnetic resonance imaging can be of benefit or it can be a detriment if its limitations are not realized. Moreover, caution is especially warranted in the interpretation of magnetic resonance imaging findings in older patients, especially if operative treatment is contemplated. Our observations reinforce the concept that evidence of a tear of the rotator cuff on magnetic resonance imaging may not necessarily explain the clinical findings.
In addition, our observations help to establish a certain baseline with regard to changes on magnetic resonance imaging as a function of aging. We do agree, however, that there is more to learn regarding the aging process and its manifestations on magnetic resonance imaging. If one considers the definition of abnormal to include findings that are "contrary to the usual structure,"1 then knowledge of the frequency of abnormalities in asymptomatic and symptomatic individuals becomes necessary to help delineate pathological from non-pathological conditions. As discussed in our article, this concept highlights the importance of appropriate clinical correlation and the need for additional imaging research studies. Furthermore, accomplishment of this task requires, in part, the evaluation of asymptomatic individuals, and we thus hope that our conclusions were not misinterpreted by Dr. Pope to suggest routine imaging of asymptomatic individuals.
Finally, we acknowledge, in part, the validity of comments regarding partial-thickness tears and tendinosis. The percentage of partial tears of the rotator cuff can be increased because of the exclusive use of oblique coronal imaging. However, it has been our experience, as correlated with arthroscopy, that the amount of T2 signal seen in the rotator cuff is consistent with partial-thickness tears. Furthermore, even from an operative perspective, there remains substantial controversy regarding the definition and classification of partial tears. Arthroscopically, what may be considered fraying of the tendon by one surgeon may be considered a partial tear of the tendon by another. Uniform agreement regarding the degree of damage to the tendon that is necessary to constitute a partial tear is lacking. We thus limited our diagnosis of a partial tear of the rotator cuff according to previously published and recognized magnetic resonance imaging criteria2. More importantly, we reiterate that, even though a signal change may indicate a partial-thickness tear, it should be interpreted with caution in a patient who has an inconsistent history and physical examination.
The suggestion that we attempted to diminish the value of magnetic resonance imaging in symptomatic patients is misleading. The objective was not to establish the role or importance of this imaging modality for patients who have disorders about the shoulder, since this is a different issue. Clearly, magnetic resonance imaging has a place in the clinical arena, and its role may be better defined through future investigations. Rather, our study revealed potential pitfalls in the interpretation of magnetic resonance imaging findings and indicated that increased knowledge of the limitations of this modality may result in more effective use of the information obtained.
Jerry S. Sher, M.D.; John W. Uribe, M.D.; Alejandro Posada, M.D.: Division of Sports Medicine, Department of Orthopaedics and Rehabilitation, University of Miami, Doctor's Hospital, 5000 University Drive, Miami, Florida 33146
Brian J. Murphy, M.D.: Department of Diagnostic Radiology, Magnetic Resonance Imaging Center, Doctor's Hospital, 5000 University Drive, Miami, Florida 33146
Michael B. Zlatkin, M.D.: Department of Diagnostic Radiology, Hollywood Memorial Hospital, 3700 Washington Street, Hollywood, Florida 33021