Clinical Outcome of Edema-like Bone Marrow Abnormalities of the Foot1

¹ From the Department of Radiology, Orthopedic University Hospital Balgrist, Forchstrasse 340, CH-8008 Zurich, Switzerland (M.Z., C.L.S., J.H.); and the Institute of Biostatistics, University of Zurich, Switzerland (B.S.). From the 2000 RSNA scientific assembly. Received January 15, 2001; revision requested February 18; revision received July 11; accepted July 15. Address correspondence to M.Z. (e-mail: marco.zanetti@balgrist.ch).

	ABSTRACT

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PURPOSE: To evaluate the clinical outcome of edema-like bone marrow abnormalities seen on magnetic resonance (MR) images of the foot when their cause is unknown.

MATERIALS AND METHODS: The clinical outcome of 31 patients (15 female patients, 16 male patients; mean age, 51; range, 10–79 years) with edema-like bone marrow abnormalities on MR images of the foot was determined. The relevance of three different edema patterns was compared: (a) exclusively ill-defined edema-like zones, (b) edema-like zones plus well-defined necrosis-like zones, and (c) edema-like zones plus linear structures indicating possible fractures. The different edema patterns were compared with persistence of pain.

RESULTS: Fifty-four percent of all patients had pain persisting after 1 year, as calculated with the Kaplan-Meier method. The duration of pain in the various subgroups varied significantly (P = .049, log-rank test). The subgroup of patients with a well-defined necrosis-like zone had substantially longer-lasting pain than those with edema-like abnormalities only (n = 16) (P = .065). Only one of seven patients with a well-defined necrosis-like zone (n = 7) was pain free after 1 year. Conversely, patients with possible stress fracture (n = 8) had shorter pain compared with those with edema-like abnormalities only (P = .036); six of eight patients were pain free after 1 year.

CONCLUSION: Edema-like bone marrow abnormalities of the foot predict long-lasting pain. Analysis of the image patterns of such abnormalities allows prediction of the clinical outcome to a certain degree.

　

Index terms: Bone marrow, abnormalities • Bone marrow, MR, 468.12141 • Foot, abnormalities, 468.833 • Foot, MR, 468.12141

	INTRODUCTION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Ill-defined areas of low signal intensity on T1-weighted magnetic resonance (MR) images and intermediate to high signal intensity on T2-weighted MR images are commonly seen on MR images of the foot. These areas are usually attributed to bone marrow edema and have been associated with several different clinical entities (1–18). They may be present in the feet of patients with diabetic neuroarthropathy (16), osteomyelitis (16), tendon abnormalities (15,19,20), neoplasm, bone contusion, and/or stress fracture (21). In these defined conditions, the treatment is established and the prognosis is predictable. However, the etiology of bone marrow edema patterns in the foot is often unclear. In such cases, rather unspecific differential diagnoses are considered, including altered biomechanics (8), stress reaction (22), or early osteonecrosis (22). Even if etiology cannot be determined on the basis of MR imaging, prediction of the clinical outcome is relevant.

The purpose of this study was to evaluate the outcome of patients with bone marrow edema patterns in the foot of unknown cause.

	MATERIALS AND METHODS

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
We identified 72 patients with a diagnosis of edema-like bone marrow abnormalities from an electronic database of 1,123 patients referred for MR imaging of the foot between January 1996 and December 1998. Twelve patients in whom the cause of the edema-like bone marrow abnormalities was known were subsequently excluded from our study: Fracture (n = 2), diabetes mellitus (n = 5), neoplasm (n = 2), tendon abnormalities (n = 2), and neurogenic osteoarthropathy (n = 1) were considered to be adequate explanation for edema-like bone marrow abnormalities. Twenty-nine patients were lost during follow-up because the patients could not be reached (n = 14), were hearing-impaired (n = 2), refused to give any information (n = 1), were mentally retarded (n = 1), or had language problems (n = 11). The final study group consisted of 31 patients (15 female patients, 16 male patients; mean age, 51 years; age range, 10–79 years) with etiologically undefined edema-like abnormalities of the foot at MR. We received an institutional review board waiver for analysis of the medical data, including imaging studies. In addition, all patients gave informed consent for the analysis of their MR images and their medical reports.

The MR images of these patients were reviewed by two musculoskeletal radiologists (M.Z., J.H.). They were asked to describe the patterns and the locations of edema independently and were blinded to all clinical information. In cases of disagreement regarding edema patterns, a consensus was reached. A consensus was also reached for the determination of the more prominent pattern in those cases in which more than one imaging pattern was present. The radiologists used the following classifications: (a) exclusively ill-defined edema-like zones (Fig 1), (b) edema-like zones plus well-defined necrosis-like zones (termed a necrosis-like pattern) (Fig 2), and (c) edema-like zones plus linear structures indicating possible fractures (Fig 3).

fig.ommitted Figure 1a. Exclusively edema-like bone marrow pattern. (a) Edema-like bone marrow abnormality (arrow) is demonstrated in the navicular bone on a sagittal short inversion time inversion-recovery (STIR) MR image (repetition time msec/echo time msec/inversion time msec, 4,800/30/150) of the hindfoot. (b) The hyperintense edema-like abnormality is barely visible (arrowheads) on the corresponding T2-weighted fast spin-echo MR image (repetition time msec/echo time msec, 4,000/96) in the transverse (parallel to the metatarsals) plane.

 

fig.ommitted   Figure 1b. Exclusively edema-like bone marrow pattern. (a) Edema-like bone marrow abnormality (arrow) is demonstrated in the navicular bone on a sagittal short inversion time inversion-recovery (STIR) MR image (repetition time msec/echo time msec/inversion time msec, 4,800/30/150) of the hindfoot. (b) The hyperintense edema-like abnormality is barely visible (arrowheads) on the corresponding T2-weighted fast spin-echo MR image (repetition time msec/echo time msec, 4,000/96) in the transverse (parallel to the metatarsals) plane.

 

fig.ommitted Figure 2a. Necrosis-like bone marrow pattern (edema-like zones plus well-defined necrosis-like zones). A well-defined abnormality (arrows) with low signal intensity is present in the talar dome. This was considered to represent necrosis. Edema-like bone marrow abnormalities (arrowheads) are visible in the talus and distal tibia. (a) Sagittal STIR MR image (4,800/30/150). (b) Coronal T1-weighted spin-echo MR image (722/20).

 

fig.ommitted Figure 2b. Necrosis-like bone marrow pattern (edema-like zones plus well-defined necrosis-like zones). A well-defined abnormality (arrows) with low signal intensity is present in the talar dome. This was considered to represent necrosis. Edema-like bone marrow abnormalities (arrowheads) are visible in the talus and distal tibia. (a) Sagittal STIR MR image (4,800/30/150). (b) Coronal T1-weighted spin-echo MR image (722/20).

 

fig.ommitted Figure 3. Possible fracture pattern (edema-like zones plus linear structures indicating possible fractures). A sagittal STIR MR image (4,800/30/150) shows that within an edema-like bone marrow abnormality (arrowheads) in the calcaneus, a linear hypointense signal abnormality (arrow) is present, indicating a possible fracture.

 
The volume of the edema-like zones was calculated by one author (C.L.S.). For calculation of the lesion volume, the longest and shortest diameter of the abnormality were first measured on each MR image. (These measurements were performed with a ruler on hard copies to the nearest millimeter, with correction for the magnification factor.) Areas were then calculated by using the ellipsoid formula: (4/3) x long diameter x short diameter. The resulting value was then multiplied by the section thickness plus the intersection gap. The total lesion volume was determined by adding the resulting values. When several bones were involved, the volumes of lesions in each bone were added.

Clinical follow-up data were gathered in structured telephone interviews performed by one of the authors (C.L.S.). Pain intensity was characterized by the patient in percentages (in 5% steps) according to a scale in which 0 indicated no pain (ie, an entirely normal foot) and 100 indicated unbearable, maximal pain. Peak pain intensity during the time between the MR examination and the telephone interview and pain intensity at the end of the follow-up (ie, when the telephone interview was performed) were determined. Pain duration was also determined for the time between the MR examination and the time when the symptoms completely disappeared.

During the interview the patients were asked if their symptoms had begun in conjunction with a distinct trauma. However, only patients with a history of no or minimal trauma remained in the study. (Minimal trauma was defined as a trauma which would not normally result in pain persisting for more than 3 weeks.)

In addition, all medical charts (n = 12) that were available for the patients were reviewed by two authors (C.L.S., M.Z.). They attempted to determine a specific diagnosis for each case, if possible, based on findings at the follow-up examinations conducted after MR imaging and on any intraoperative findings.

MR Imaging
MR imaging had been performed with a 1.0-T imager (Impact/Expert; Siemens Medical Systems, Erlangen, Germany). A circular-polarized, send-receive extremity coil was used to obtain the images in all patients. The imaging protocols varied depending on the clinical diagnosis before MR imaging. The standard protocol for suspected metatarsal abnormalities consisted of a STIR sequence (5,320–5,500/30/150) with imaging performed parallel or perpendicular to the metatarsals. Contrast material–enhanced T1-weighted fat-suppressed spin-echo MR images (812–850/15) were obtained (in five of 31 patients) when they were considered indicated by the responsible radiologist. However, the appearance of the bone marrow abnormality on the contrast-enhanced images was not substantially different from its appearance on the STIR images in these five patients. Therefore, contrast-enhanced T1-weighted fat-suppressed spin-echo MR images were not analyzed in detail for the purpose of this investigation. Additional sequences employed for metatarsal abnormalities included sagittal T1-weighted spin-echo sequences (500-600/15, 3-mm section thickness, 135 x 180-mm field of view , 144 x 256 image matrix), T1-weighted spin-echo MR sequences (500–600/15, 3-mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix) with imaging performed perpendicular to the metatarsals, and T2-weighted turbo spin-echo sequences (4,000–4,500/96, 3–4 mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix, two acquisitions) with imaging performed perpendicular to the metatarsals.

The standard protocol for hindfoot abnormalities consisted of a sagittal STIR sequence (4,800–5,500/30/150), a sagittal T1-weighted sequence (500/15), an oblique T1-weighted spin-echo MR sequence (500–600/15, 3-mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix) with imaging performed perpendicular to the long flexor tendons, a T2-weighted turbo spin-echo sequence (4,000–4,500/96, 3-mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix, two acquisitions) with imaging performed perpendicular to the long axis of the tibia, a coronal T1-weighted spin-echo sequence (500–750/15–22, 3–4-mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix) and a T2-weighted turbo spin-echo sequence (4,000–4,500/96, 3–4-mm section thickness, 90 x 120-mm FOV, 144 x 256 image matrix, two acquisitions).

Data Analysis
Kaplan-Meier estimates of survival curves were used to describe the probability of pain persistence for a given time period. Kaplan-Meier estimates of survival curves allow analysis of censored data (data in which the period of observation ended before the event of interest occurred). The event of interest in the present study was the time when pain completely disappeared. The Kaplan-Meier curves of the three various imaging patterns were compared by means of the log-rank test. Cox regression was used to evaluate an association between pain duration and bone marrow edema volume. Spearman rank correlation coefficients were calculated to relate pain intensity and bone marrow edema volume (23).

StatView software (SAS Institute, Cary, NC) was used for all statistical analyses.

	RESULTS

Top ABSTRACT INRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Exclusively ill-defined edema-like zones were observed on the MR images of 16 patients, edema-like zones plus well-defined necrosis-like zones were observed in seven patients, and edema-like zones plus linear structures were observed in eight patients. More than one imaging pattern was present in two patients. However, only the more prominent pattern was used for analysis. Signal intensity abnormalities were present in the following bones: tibia (n = 5 patients), fibula (n = 2), talus (n = 11), calcaneus (n = 11), navicular (n = 11), cuboid (n = 6), and the cuneiformia (n = 12) and metatarsals (n = 11). Thirteen patients had involvement of one bone, and 18 of several bones; the maximum number of bones involved was eight.

The mean volume of the signal intensity abnormalities was 84 mm³ (range, 4–249 mm³; SD, 86 mm³).

The follow-up time of the entire study group varied between 17 and 155 weeks (mean, 112 weeks). Mean peak pain intensity was 63% (range, 20%–100%). Mean pain intensity at the time of the telephone interview was 16% (range, 0%–90%). Fifty-four percent of all patients had pain persisting after 1 year, as calculated with the Kaplan-Meier method (Fig 4). The duration of pain varied significantly among the three various subgroups (P = .049, log-rank test) (Fig 5). Only one of seven patients with a well-defined necrosis-like zone was pain free after 1 year. The patients in the group with necrosis-like abnormalities also had pain of substantially longer duration than did the patients in the group with exclusively edema-like abnormalities (P = .065, log-rank test). Six of eight patients with abnormalities that indicated possible stress fracture were pain free after 1 year. These patients had pain of significantly shorter duration during the follow-up period than the patients with edema-like abnormalities only (P = .036, log-rank test). Pain duration was not significantly different for patients with minimal trauma versus those without trauma (P = .86, log-rank test). Pain duration was not related to bone marrow edema volume (P = .21, Cox regression). Peak pain intensity was also not significantly correlated with bone marrow edema volume (R² = 0.005, P = .58, Spearman rank correlation).

fig.ommitted Figure 4. Graph of a Kaplan-Meier curve. Pain persistence is demonstrated for the entire study group. The small vertical bars indicate censored data. Fifty-four percent of all patients had pain persisting after 1 year.

 

fig.ommitted Figure 5. Graph of a Kaplan-Meier curve. Pain persistence is demonstrated for patients in the three imaging pattern subgroups. The small vertical bars indicate censored data. Pain persistence varied significantly among the three subgroups. Patients with necrosis-like patterns had the worst outcome, while patients with possible fracture had the most favorable outcome.

 
Five patients underwent surgery during the follow-up period. Pain diminished completely in three of these patients after either lateral collateral ligament repair (n = 1), resection of fibrous calcaneonavicular coalition (n = 1), or removal of a free intraarticular body (n = 1). Two of these five patients had symptoms that persisted after surgery to remove an intraosseous cyst.

	DISCUSSION

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Our results indicate that edema-like bone marrow abnormalities detected during MR imaging of the foot are clinically relevant. Pain persisting for at least several months is to be expected in patients with such findings, at least when the findings are not related to substantial trauma. Symptoms seem to persist even more commonly when, in addition to edema-like abnormalities, well-defined areas of abnormal MR signal intensity typical of those of avascular necrosis are present. Symptoms appear to persist less commonly when MR findings of a stress fracture are present.

That pain would persist for such prolonged periods of time has not been anticipated. In two studies that followed up patients with ankle sprain, no association between bone marrow edema (bone bruise) and pain persistence after 3 months was found (24,25). Most published studies relating to traumatic bone marrow abnormalities and outcome indicate that these abnormalities resolve within 4 months (26–28). These studies have, however, focused on imaging findings and not on clinical signs such as pain persistence. Moreover, patients in these studies had experienced substantial trauma. In our study, no or minimal trauma was encountered.

In the present study, the subgroup of patients with possible stress fracture had a significantly better outcome than those patients with ill-defined MR signal intensity changes only. We assume that the diagnosis of possible stress fracture at MR was correct in these patients and that simple reduction of weight bearing has allowed the bone to heal.

When the underlying pathologic condition was not detected during the initial MR examination, the cause of pain persistence could occasionally be explained during the follow-up period. A fibrous calcaneonavicular coalition and a free intraarticular body in one patient each were diagnosed at surgery but not at MR examination. One patient became pain free after repair of the lateral collateral ligaments.

In the majority of our patients, however, the etiology of the edema-like MR abnormalities remained undefined even at the end of the study. Therefore, only hypotheses about the etiology can be made. There is some evidence that signal intensity changes on MR images of the bones of the foot may represent an inadequate reaction to mechanical overload. The influence of altered biomechanics on the bones of the foot has been previously demonstrated. Edema-like changes have been detected 2 weeks after patients began using foot pads designed to cause excessive pronation of the foot (8). The possible effects of mechanical overload are underlined by findings in joggers in whom edema-like MR signal intensity changes in the foot have been demonstrated after they ran more than 50 miles per week (3). The histologic nature of such signal intensity changes has been debated. Although these areas are commonly called bone marrow edema, authors of some reports have questioned the general use of this term (29). It is known that in osteoarthritic knees, edema-like areas are caused by a number of histologic abnormalities that are also found in osteonecrosis (29). It has even been speculated that perhaps "all cases of the bone marrow edema pattern, aside from those due to infection, fracture, neoplasm represent osteonecrosis" (30). However, even the histologic presence of necrotic bone marrow cells and necrotic trabecular bone may not unequivocally indicate the etiology. The histologic findings in patients who underwent surgery for spontaneous osteonecrosis of the knee have been reviewed by Yamamoto and Bullough (31). They concluded that the primary event leading to spontaneous osteonecrosis of the knee is a subchondral insufficiency fracture. The localized osteonecrosis seen in association with this disease was the result of a fracture.

Another potential cause of bone marrow edema-like MR patterns is reflex sympathetic dystrophy. Notably, a review of available medical charts did not reveal this diagnosis in any of our patients. Moreover, according to the experience of Schweitzer et al, no bone marrow changes are found on MR images obtained in patients with reflex sympathetic dystrophy (7).

The limitations of the present study relate to its retrospective design. The patients were not treated in a standardized fashion, and some differences in the outcome may relate to the type of treatment. However, even in a prospective study design, no standardized treatment could probably be imposed because there are different potential origins in each patient.

Considering the long follow-up period (half of our patients had pain persisting for over 1 year), we conclude that edema-like bone marrow abnormalities in the foot are associated with relatively long-lasting pain. The analysis of the image patterns of etiologically undefined edema-like abnormalities of the bone marrow allows prediction of the clinical outcome to a certain degree. Symptoms seem to persist more commonly when additional well-defined necrosis-like bone marrow abnormalities are present, whereas symptoms persist less commonly when MR findings of stress fracture are present. The volume of edema-like abnormalities appears to be less important in the clinical outcome.

　

	REFERENCES

Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Froberg PK, Braunstein EM, Buckwalter KA. Osteonecrosis, transient osteoporosis, and transient bone marrow edema: current concepts. Radiol Clin North Am 1996; 34:273-291.

2. Hayes CW, Conway WF, Daniel WW. MR imaging of bone marrow edema pattern: transient osteoporosis, transient bone marrow edema syndrome, or osteonecrosis. RadioGraphics 1993; 13:1001-1011.

3. Lazzarini KM, Troiano RN, Smith RC. Can running cause the appearance of marrow edema on MR images of the foot and ankle?. Radiology 1997; 202:540-542.

4. Lecouvet FE, van de Berg BC, Maldague BE, et al. Early irreversible osteonecrosis versus transient lesions of the femoral condyles: prognostic value of subchondral bone and marrow changes on MR imaging. AJR Am J Roentgenol 1998; 170:71-77.

5. Plenk H, Jr, Hofmann S, Eschberger J, et al. Histomorphology and bone morphometry of the bone marrow edema syndrome of the hip. Clin Orthop 1997; :73-84.

6. Reinus WR, Fischer KC, Ritter JH. Painful transient tibial edema. Radiology 1994; 192:195-199.

7. Schweitzer ME, Mandel S, Schwartzman RJ, Knobler RL, Tahmoush AJ. Reflex sympathetic dystrophy revisited: MR imaging findings before and after infusion of contrast material. Radiology 1995; 195:211-214.

8. Schweitzer ME, White LM. Does altered biomechanics cause marrow edema?. Radiology 1996; 198:851-853.

9. Vahlensieck M, Reiser M. Bone marrow edema in MRT. Radiologe 1992; 32:509-515.

10. Vande Berg BE, Malghem JJ, Labaisse MA, Noel HM, Maldague BE. MR imaging of avascular necrosis and transient marrow edema of the femoral head. RadioGraphics 1993; 13:501-520.

11. Wilson AJ, Murphy WA, Hardy DC, Totty WG. Transient osteoporosis: transient bone marrow edema?. Radiology 1988; 167:757-760.

12. Yu JS, Cook PA. Magnetic resonance imaging (MRI) of the knee: a pattern approach for evaluating bone marrow edema. Crit Rev Diagn Imaging 1996; 37:261-303.

13. Neuhold A, Hofmann S, Engel A, et al. Bone marrow edema of the hip: MR findings after core decompression. J Comput Assist Tomogr 1992; 16:951-955.

14. Rademaker J, Rosenberg ZS, Delfaut EM, Cheung YY, Schweitzer ME. Tear of the peroneus longus tendon: MR imaging features in nine patients. Radiology 2000; 214:700-704.

15. Grasel RP, Schweitzer ME, Kovalovich AM, et al. MR imaging of plantar fasciitis: edema, tears, and occult marrow abnormalities correlated with outcome. AJR Am J Roentgenol 1999; 173:699-701.

16. Craig JG, Amin MB, Wu K, et al. Osteomyelitis of the diabetic foot: MR imaging-pathologic correlation. Radiology 1997; 203:849-855.

17. Deely DM, Schweitzer ME. MR imaging of bone marrow disorders. Radiol Clin North Am 1997; 35:193-212.

18. Miller TT, Staron RB, Feldman F, Parisien M, Glucksman WJ, Gandolfo LH. The symptomatic accessory tarsal navicular bone: assessment with MR imaging. Radiology 1995; 195:849-853.

19. Schweitzer ME, Karasick D. MR imaging of disorders of the Achilles tendon. AJR Am J Roentgenol 2000; 175:613-625.

20. Schweitzer ME, Karasick D. MR imaging of disorders of the posterior tibialis tendon. AJR Am J Roentgenol 2000; 175:627- 635.

21. Lee JK, Yao L. Stress fractures: MR imaging. Radiology 1988; 169:217-220.

22. Haller J, Sartoris DJ, Resnick D, et al. Spontaneous osteonecrosis of the tarsal navicular in adults: imaging findings. AJR Am J Roentgenol 1988; 151:355-358.

23. Altman DG. Comparing groups: continuous data In: Practical statistics for medical research. London, England: Chapman & Hall, 1991; 229-272.

24. Zanetti M, De Simoni C, Wetz H, Zollinger H, Hodler J. Magnetic resonance imaging of injuries to the ankle joint: can it predict clinical outcome?. Skeletal Radiol 1997; 26:82-88.

25. Alanen V, Taimela S, Kinnunen J, Koskinen SK, Karaharju E. Incidence and clinical significance of bone bruises after supination injury of the ankle: a double-blind, prospective study. J Bone Joint Surg Br 1998; 80:513-515.

26. Labovitz JM, Schweitzer ME. Occult osseous injuries after ankle sprains: incidence, location, pattern, and age. Foot Ankle Int 1998; 19:661-667.

27. Miller MD, Osborne JR, Gordon WT, Hinkin DT, Brinker MR. The natural history of bone bruises: a prospective study of magnetic resonance imaging-detected trabecular microfractures in patients with isolated medial collateral ligament injuries. Am J Sports Med 1998; 26:15-19.

28. Graf BK, Cook DA, De Smet AA, Keene JS. "Bone bruises" on magnetic resonance imaging evaluation of anterior cruciate ligament injuries. Am J Sports Med 1993; 21:220-223.

29. Zanetti M, Bruder E, Romero J, Hodler J. Bone marrow edema pattern in osteoarthritic knees: correlation between MR imaging and histologic findings. Radiology 2000; 215:835-840.

30. MacDougall L, Conway WF. Controversies in magnetic resonance imaging of the hip. Top Magn Reson Imaging 1996; 8:44-50.

31. Yamamoto T, Bullough PG. Spontaneous osteonecrosis of the knee: the result of subchondral insufficiency fracture. J Bone Joint Surg Am 2000; 82:858-866.