ONCOLOGY
The MRI appearances of cancellous allograft bone chips after the excision of bone tumours
S. Kang, I. Han, S. H. Hong, H. S. Cho, W. Kim, H-S. Kim From Seoul National University Hospital, Seoul, South Korea
S. Kang, MD, Orthopaedic Surgeon, Clinical fellow I. Han, MD, PhD, Orthopaedic Surgeon, Associate Professor S. H. Hong, MD, PhD, Radiologist, Professor W. Kim, MD, Orthopaedic Surgeon, Clinical Fellow H-S. Kim, MD, PhD, Orthopaedic Surgeon, Professor Department of Orthopaedic Surgery, Seoul National University Hospital, 101 Daehak-ro Jongno-gu, Seoul 110-744, South, Korea. H. S. Cho, MD, Orthopaedic Surgeon, Associate Professor Seoul National University Bundang Hospital, Department of Orthopaedic Surgery, 166 Gume-ro Bundang-gu, Gyeonggi-do 463-707, South, Korea. Correspondence should be sent to Professor H. S. Kim; e-mail: [email protected] ©2015 The British Editorial Society of Bone & Joint Surgery doi:10.1302/0301-620X.97B1. 34517 $2.00 Bone Joint J 2015;97-B:121–8. Received 21 May 2014; Accepted after revision 1 September 2014
Cancellous allograft bone chips are commonly used in the reconstruction of defects in bone after removal of benign tumours. We investigated the MRI features of grafted bone chips and their change over time, and compared them with those with recurrent tumour. We retrospectively reviewed 66 post-operative MRIs from 34 patients who had undergone curettage and grafting with cancellous bone chips to fill the defect after excision of a tumour. All grafts showed consistent features at least six months after grafting: homogeneous intermediate or low signal intensities with or without scattered hyperintense foci (speckled hyperintensities) on T1 images; high signal intensities with scattered hypointense foci (speckled hypointensities) on T2 images, and peripheral rim enhancement with or without central heterogeneous enhancements on enhanced images. Incorporation of the graft occurred from the periphery to the centre, and was completed within three years. Recurrent lesions consistently showed the same signal intensities as those of preoperative MRIs of the primary lesions. There were four misdiagnoses, three of which were chondroid tumours. We identified typical MRI features and clarified the incorporation process of grafted cancellous allograft bone chips. The most important characteristics of recurrent tumours were that they showed the same signal intensities as the primary tumours. It might sometimes be difficult to differentiate grafted cancellous allograft bone chips from a recurrent chondroid tumour. Cite this article: Bone Joint J 2015;97-B:121–8.
Allograft bone grafting involves placing bone, often harvested from human cadavers, into defects in bone to aid healing.1,2 Incorporation of a graft involves the envelopment and interdigitation of the donor tissue with new bone deposited by the recipient.2-4 This process involves many steps. The bone graft stimulates the formation of inflammatory cells, followed by the recruitment of host mesenchymal cells. The primitive host cells then differentiate into bone forming cells. The subsequent processes of revascularisation and necrotic resorption of the graft occur concurrently. Finally, bone is laid down onto the framework of the graft by osteoblasts, followed by remodelling in response to mechanical stress.2,5 The use of allograft bone in reconstructive surgery continues to increase.6,7 MRI is better than other methods of imaging for the evaluation of primary bone tumours.8 Accordingly, MRI is commonly used following curettage and bone grafting in patients with primary intraosseous tumours to ensure the adequacy of the treatment. Oncologists may be asked to evaluate MRI signal intensities at grafted sites to determine whether incorpora-
VOL. 97-B, No. 1, JANUARY 2015
tion is occurring or if there is a recurrence of the tumour. There have been several studies regarding the radiological evaluation of structural allografts9 but few have focused on cancellous allograft bone. There might be differences in the MRI features after bone grafting between structural and cancellous grafts, as the process of incorporation into host bone is different in both.10 As far as we know, there is only one description of the MRI features of grafted cancellous allograft bone chips, involving 18 cases,11 and there is no study of the MRI changes of this allograft bone over time. The main purpose of this study was to investigate the MRI features and incorporation processes of grafted cancellous allograft bone chips in relation to the time interval after grafting. Additionally, we investigated the differences in appearance on MRI between the normal features of grafted bone chips and those of a recurrent tumour.
Patients and Methods Of 202 patients treated by curettage and cancellous allograft bone chips (freeze-dried, from 121
122
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
various manufacturers) for bony tumours between March 2002 and November 2011, we retrospectively reviewed 53 who underwent MRI as part of their follow-up. Pre-operative MRIs were available for all patients. Those patients who had other combined surgical procedures involving the implantation of a prosthesis, bone morphogenetic protein, or a calcium-based bone substitute were excluded (n = 17). Two patients, who were lost to follow-up within one year after MRI without evidence of recurrence, were excluded from the analysis because we cannot confirm the absence of recurrence. Finally, we analysed 34 patients with 66 postoperative follow-up MRIs. No prospective selection criteria were used for MRI follow-up. Patients being treated for a recurrent tumour were reviewed every three to four months for the first year, every six months for the next four years and annually afterwards. Plain radiographs and MRI were alternatively used. Among the 34 patients, the indications for post-operative MRI included pain (n = 10), another medical cause (n = 8), an abnormal finding on plain radiographs (n = 6), requested by the patient (n = 6), and undocumented reasons (n = 4). There were 17 patients with only one post-operative MRI, and the other 17 patients had two or more post-operative MRIs (nine with two MRIs, four with three MRIs, two with four MRIs, one with five MRIs, and one with six MRIs). Thus, there were 23 MRIs of a chondroblastoma, 22 of a low-grade chondrosarcoma, five of an enchondroma, five of a giant cell tumour, four of fibrous dysplasia, two of a simple bone cyst, two of pigmented villonodular synovitis with bony involvement, two of an aneurysmal bone cyst, and one of intraosseous lipoma. The anatomical distribution of the tumours was 29 femora, 15 tibiae, eight pelves, eight humeri, five feet and one hand. The time intervals between bone grafting and follow-up MRI were ≤ six months (n = 12); > six months to ≤ one year (n = 17); > one year to ≤ two years (n = 18); > two years to ≤ three years (n = 11); and > three years (n = 8). The mean time to follow-up imaging was 22.5 months (standard deviation (SD) 20.7; 3 to 94). The size of the tumour was defined by the largest diameter on the pre-operative MRIs, with the mean size overall being 6.7 cm (SD 6.3). According to length of follow-up, the mean size was 6.2 cm (SD 6.6) for ≤ six months, 6.7 cm (SD 6.4) for > six months to ≤ one year, 6.6 cm (SD 5.7) for > one year to ≤ two years, 8.5 cm (SD 6.8) for > two years to ≤ three years, and 5.3 cm (SD 7.1) for > three years. These differences were not statistically significant (p = 0.854, analysis of variance (ANOVA)). No patient received radiotherapy or chemotherapy. MRI was performed with various 1.5- (GE Healthcare, Milwaukee) or 3.0-Tesla MRI scanners (Siemens Medical Solutions, Erlangen, Germany), but in all patients, spin-echo T1-weighted (repetition time (TR) msec/echo time (TE) = 450 to 778/11 to 14) and fast spin-echo T2weighted (TR/TE = 2500 to 4500/60 to 108) images (T1WI and T2WI) with or without fat suppression were acquired, and gadolinium (Gd)-enhanced spin-echo T1WI were also acquired, except in two patients. The field of view varied
from 160 mm × 160 mm to 400 mm × 400 mm and the matrix from 192 × 192 to 512 × 512, slice thickness varied from 3 mm to 5 mm, and various coils were used depending on lesion size and location. The MRIs were reviewed independently by two experienced oncologists (one radiologist (SHH) and one orthopaedic surgeon (SK)) and decisions made by consensus. The typical MRI features on T1WI, T2WI, and Gd-enhanced T1WI in the early phase (≤ six months) were determined and then we investigated the changes of the grafted lesion over time on each view. The change of the extent of the lesion was determined after comparison with the pre-operative MRIs and the immediate post-operative plain radiographs. We also noted changes in adjacent structures such as bone-marrow oedema or sclerosis. Artifacts due, for instance, to suture material were ignored. Although some patients suffered a local recurrence of tumour, none had other local complications, such as infection or fracture. Biopsy was performed for patients with suspected recurrence on MRI, and if recurrence was confirmed, the signal intensity of the recurrent lesion was compared with that of the primary tumour on the pre-operative MRIs, as well as with that of the grafted bone chips. In our hospital, a formal radiological report within three working days after MRI scanning is obligatory. In order to determine whether or not a misdiagnosis had occurred, the formal reports were compared with the histological results when MRIs were interpreted as a suspected recurrence, or with follow-up imaging and clinical presentation when MRIs were interpreted as showing no evidence of recurrence. The study had ethical approval.
Results In the early period (≤ six months) after grafting, the grafts showed consistent features on MRI with: on T1WI, relatively homogeneous intermediate or low signal intensities with or without scattered hyperintense foci (speckled hyperintensity); on T2WI, high signal intensities with scattered hypointense foci (speckled hypointensity) and on Gd-enhanced T1WI, peripheral enhancement with or without central heterogeneous enhancement (Fig. 1). These typical features were gradually lost as time passed. The changes over time are summarised in Table I. The size of the defects decreased with progressive incorporation of the graft. The bony incorporation progressed from periphery to centre, as indicated by the finding that the signal intensity of the periphery started to become similar with that of the host bone after six months (Fig. 2). Incorporation usually began between six months and two years after grafting, and was completed by three years after grafting (Table I). As time passed, the proportion of lesions showing typical features decreased. Gd-enhancement also disappeared, and notably, all remnant graft exhibited Gd-enhancement. Surrounding bone-marrow oedema also diminished over time, irrespective of whether remnants of graft were present or not. THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
123
Fig. 1a
Fig. 1b MRIs showing the typical features of grafted cancellous allograft bone chips (CABCs) at a) at six months post-operatively in a 31-year-old man who underwent curettage and grafting for a low-grade chondrosarcoma of the right femur, showing homogeneous intermediate signal intensities (SIs) with speckled hyperintensities on coronal T1 image (left), high SIs with speckled hypointensities on sagittal T2 image (middle), and peripheral rim enhancement with central heterogeneous enhancement on coronal enhanced image (right) and b) at eight months post-operatively in a 33-year-old man with fibrous dysplasia of the right proximal femur, showing homogeneous intermediate SIs without speckled hyperintensities on coronal T1 image (left), and peripheral rim enhancement without central enhancement on coronal Gd-enhanced image (right).
Table I. MRI findings over time (number, %)
Time interval
*
≤ 6 months
Extent of grafted lesion
Presence of typical features†
N
Peripheral Complete Totals disappearance (A) disappearance (B) (A+B)
Residual lesion (n)
T1WI
12
-
-
12 (100)
12 (100)
-
T2WI
Gd-enhanced T1WI
Surrounding bone marrow oedema
Gd-enhancement
12 (100)
11 (100)‡
6/12 (50)
11/11 (100)‡
> 6 months to ≤ 1 year
17
4 (24)
-
4 (24)
17 (100)
14 (82)
10 (59)
17 (100)
7/17 (41)
17/17 (100)
> 1 year to ≤ 2 years
18
9 (50)
3 (17)
12 (67)
15 (83)
12 (80)
5 (33)
13 (87)
7/18 (39)
18/18 (100)
> 2 years to ≤ 3 years
11
6 (55)
5 (46)
11 (100)
6 (55)
4 (67)
1 (17)
3 (50)
2/11 (18)
9/11 (82)
> 3 years
8
-
8 (100)
8 (100)
0 (0)
-
-
-
2/8 (25)
5/7 (57)‡
WI, weighted image * Time interval between bone grafting and MRI † The typical features were identified as homogeneous intermediate or low signal intensity (SI) with/without speckled hyperintensity on T1-weighted image (WI), high SI with speckled hypointensity on T2WI, or peripheral rim enhancement with/without central heterogeneous enhancement on gadolinium (Gd)-enhanced T1WI ‡ The enhanced views were performed in 11 of 12 patients at ≤ 6 months and in seven of eight at > 3 years
Sclerotic components were observed in 22 patients (64.7%) (Fig. 3) with these elements situated peripherally and/or scattered. In some patients, sclerosis had been present pre-operatively. If a sclerotic component was identified, it did not change or disappear with the passage of VOL. 97-B, No. 1, JANUARY 2015
time. The sclerotic areas on MRIs were concordant with those on plain radiography. Details of patients with recurrent tumour or misdiagnosis are shown in Table II. A total of eight patients (23.5%) developed suspicious lesions initially seen on MRIs, and
124
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
Fig. 2 Post-operative T1 images of a 62-year-old man who underwent curettage and cancellous grafting for a lowgrade chondrosarcoma of the proximal humerus, showing the extent of incorporation of the graft over time at (left) ten and (right) 18 months. Incorporation of the graft occurs from periphery to centre (arrow heads).
Fig. 3a
Fig. 3b
MRI views at 40 months after grafting in the proximal tibia. Peripheral sclerosis is observed on a) T1-weighted image, this finding is concordant with the finding on the b) plain radiographs. Scattered sclerotic areas were sometimes observed inside lesions. Once a sclerotic component occurred, it did not change thereafter.
among these there were five (14.7%) with a proven recurrence. The lesions were found not to be a recurrence in three patients (two low-grade chondrosarcomas in the femur and one fibrous dysplasia in the pelvis). Thus, these three patients were misdiagnosed (Figs 4 and 5). Among the 26 patients, whose MRIs were interpreted as showing no evidence of recurrence, one with a low-grade chondrosarcoma of the pelvis was found to have a recurrence. The lesion was initially small and was misinterpreted on MRI as a post-operative fluid collection. However, at the next MRI, the lesion was larger, and a biopsy was performed, which
confirmed recurrence. Among the six patients in whom recurrence was confirmed, five had recurrence reported on MRI and all those lesions had the same signal intensities as had been shown by the primary tumours in the original pre-operative MRI. The indications for MRI in these six patients included pain in two, other medical reasons in two (one for the evaluation of growth plate injury and one for evaluation of ankle sprain), one in response to the patient’s request, and one for an abnormal finding on plain radiographs. There were four misdiagnoses found through the follow-up scanning, which constituted three false-positive THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
125
Table II. Details of recurrence or misdiagnosis Case number
Gender Age
Time interval* (mths)
Diagnosis
Anatomical site
Signs of recurrence on MRI
Actual recurrence
1 2 3 4 5 6 7 8 9
F F M M M F F M M
3 4 6 10 12 12 13 36 52
LCSA LCSA LCSA Chondroblastoma Fibrous dysplasia LCSA Intraosseous lipoma Simple bone cyst PVNS
Femur Pelvis Femur Tibia Pelvis Femur Calcaneus Pelvis Pelvis
Yes No Yes Yes Yes Yes Yes Yes Yes
Yes Yes† No‡ Yes No† No† Yes Yes Yes
22 38 24 15 29 50 49 19 46
LCSA, low-grade chondrosarcoma; PVNS, Pigmented villonodular synovitis with bony involvement * Time interval between bone graft and MRI † Misdiagnosed cases
Fig. 4 MRIs of a 59-year-old woman who had undergone curettage and cancellous grafting with bone chips for a low-grade chondrosarcoma of the right femur 11 months previously (T1WI left, T2WI middle, and Gd-enhanced T1WI right). The images show the typical features with a speckled pattern and peripheral rim enhancement, but were misinterpreted as a recurrent tumour. The speckled pattern and peripheral enhancement is similar and seems to have been misinterpreted as characteristic of a chondroid tumour.
recurrences and one false-negative recurrence. All four misdiagnoses occurred in patients who had follow-up MRIs within one year of grafting.
Discussion The one study dealing with the MRI features of grafted cancellous bone which has been previously published did not include time-dependent changes.11 This may be because curettage and grafting is mostly reserved for benign lesions and recurrence has not been a source of great concern. The present study has addressed this deficiency in the literature. Some consideration should be given to the correct interpretation of the results of this study. First, post-operative MRI was not performed routinely because most of the tumours were benign. The main indications for performing post-operative MRI were abnormal findings on plain radiographs or pain. Thus, there may have been selection bias. Secondly, there was no histological evidence supporting the VOL. 97-B, No. 1, JANUARY 2015
MRI findings of the grafted bone chips. In order to understand more precisely the features of grafted bone chips and the incorporation process, a study that correlates imaging features with histological information is needed. Regarding the typical MRI findings of the grafted bone chips, we found that all lesions had consistent features during the early period (≤ six months) after grafting. On T1WI, there were homogeneous intermediate or low signal intensities, some with speckled hyperintensities. On T2WI, there were high signal intensities with speckled hypointensities. On Gd-enhanced T1WI, there was enhancement of the peripheral rim in all patients, and some had additional central heterogeneous enhancement. Our description of the MRI features as typical of grafted bone chips was similar to the findings of Jelinek et al.11 They further explained that the presence of areas of speckled hyperintensity reflected the underlying morphology of the cancellous bone allograft. We found the same pattern in our study, but some
126
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
Fig. 5a
Fig. 5b MRIs from a man with fibrous dysplasia of the right ilium: images a) pre- and b) 12 months postoperatively. Both show homogeneous isointense signals on T1WI (left column), high signal intensities on T2WI (middle column), and peripheral gadolinium-enhancement (right column). The patient was misdiagnosed with a recurrent tumour, but with hindsight, there is a speckled pattern on T2WI that is present only in the post-operative images. This finding is thought to be a useful differential feature between recurrent tumours and grafted bone chips.
scans started to lose the speckled patterns six months after grafting. Therefore, we suggest that this speckled morphology will be maintained for at least six months after grafting and then gradually disappear. We investigated the process of incorporation of bone chips into host bone. Although there was no supporting histological confirmation, we identified a series of common findings on MRI. First, the process occurred from periphery to centre. This finding is supported by a previous histological study showing that allograft particles located near preexisting bone were surrounded by new bone, while those located near the centre of the graft showed no signs of remineralisation or new bone formation.12 Next, the incorporation process seems to be accompanied by Gd-enhancement, suggesting revascularisation. This seems to be an appropriate interpretation for many reasons. Revascularisation is well-known to accompany incorporation; the enhancement was prominent at the periphery, where actual
incorporation occurs, and all remnant lesions also showed Gd-enhancement. However, there were no MRIs taken within two months of grafting in this study. Furthermore, no Gd-enhancement of cortical allograft has been reported on MRI between two days and two months post-operatively.9 Therefore, it is difficult to determine whether Gd-enhancement would also be present within two months after grafting. Currently we can only confirm that enhancement develops within three months of grafting and seems to last until after the grafted lesion disappears completely. In contrast, although it showed a gradual decrease similar to the Gd-enhancement, surrounding bone-marrow oedema seems to be a post-operative change only. It does not seem to be an essential part of incorporation, because surrounding bone-marrow oedema was only present in ≤ 50% of all lesions at each interval, irrespective of whether any remnants of graft were present. Thirdly, all of the bone chips seemed to be entirely incorporated into the host bone THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
eventually. This is in agreement with previous studies showing that cancellous grafts tend to heal completely with time, whereas cortical or xenogeneic grafts do not.10,13 Additionally, most incorporation of the allograft bone chips has been accomplished within three years. The size of the residual bone defect continued to diminish with time, with all the lesions having disappeared by > three years. We do not suggest complete incorporation must be completed by three years in every setting, as it will depend on the size of lesion, the quality of the graft and complications such as infection or fracture.14 No lesion in our study was large and no patient had a complication other than local recurrence of tumour. Finally, during the incorporation process, the typical MRI features might disappear gradually in some patients. This might reflect revascularisation or resorption of the graft but histological confirmation is required. In the present study, six patients (17.6%) had local recurrence (Table II), a relatively high incidence considering the benign nature of the tumours. This may reflect selection bias as some of the MRI scans were performed because of a suspicion of recurrence. In a previous report, a lesion with a high post-operative signal intensity on T2WI in a patient who had undergone only surgery without combined radiotherapy was regarded as an active tumour.15 However, all of the grafted bone chips in our study had high signal intensities on T2WI, even though none of the patients had radiotherapy and all the MRIs of recurrent tumours had the same patterns and signal intensities as pre-operatively in T1WI, T2WI, and Gd-enhanced T1WI sequences. Therefore, we suggest that the most important characteristic of a recurrent tumour is that it has the same pattern and signal intensity in each view as the primary lesion had pre-operatively. The four misinterpreted MRIs in our series of 66 MRIs involved four patients. These misdiagnoses were confirmed on the subsequent MRIs or biopsies. Only one scan missed a recurrence of fibrous dysplasia (Fig. 5). Similar signal intensities were observed on both pre- and post-operative MRIs, which revealed homogeneous isointensities on T1WI, hyperintensities on T2WI, and peripheral rim enhancement on Gd-enhanced T1WI. However, there was a speckled pattern on T2WI that was only seen in the grafted lesion on post-operative images, whereas the tumour had a homogeneous high signal intensity without the speckled pattern on T2WI on pre-operative imaging. This might be a significant point of difference. The tumours in two of the three false-positive cases were low-grade chondrosarcomas of the femur (Fig. 4). However, the signal intensity of a chondroid tumour is similar with the typical MRI features of cancellous allograft bone chips (Figs 1 and 4). In these, the speckled pattern and peripheral enhancement were misread as mineralisation and the ring and arc appearance of a chondroid tumour.16 Great caution is required when differentiating a chondroid tumour from grafted bone chips when these typical features are present, although some grafted bone chips naturally lose their typical features over time. It is not possible to measure the diagnostic accuracy VOL. 97-B, No. 1, JANUARY 2015
127
of MRI for recurrent tumour after grafting with bone chips as the series is small and includes mixed pathologies. However, as all four misdiagnoses occurred within the first year of grafting, it seems that MRI may be an excellent modality for the diagnosis of recurrent tumour following cancellous grafting with bone chips, particularly after the one-year follow-up. More than half of the grafted lesions in this study had a sclerotic component with a peripheral and/or scattered pattern (Fig. 3). Although not in every patient, the peripheral sclerotic rim was present pre-operatively as peritumoural sclerosis, and in many patients scattered sclerosis tended to appear after surgery. The scattered sclerosis seemed to reflect the packing of the defect with graft. The sclerotic components seen on MRIs matched those seen on plain radiographs, and if a sclerotic component was detected, it did not change or disappear during the entire follow-up. In conclusion, we evaluated MRI features of grafted cancellous bone chips according to the time interval after grafting. The early MRIs showed relatively homogeneous intermediate or low SIs with/without speckled hyperintensities on T1WI, high signal intensities with speckled hypointensities on T2WI, and peripheral rim enhancement with or without central heterogeneous enhancement on Gd-enhanced view. These are thought to be the typical MRI features of grafted bone chips in the early period. Thereafter, the incorporation of the graft proceeds as follows: Gd-enhancement starts within three months after grafting, the speckled morphology of the bone chips is maintained until at least six months, and the graft is then incorporated into the host bone from periphery to centre. Most incorporation is completed within three years. The most important characteristic associated with recurrent tumour formation is that the lesion shows the same pattern and signal intensity as the primary tumour. It may sometimes be difficult to differentiate grafted bone chips from a recurrent chondroid tumour, although MRI seems to be an excellent modality for doing this, especially more than one year after grafting. S. Kang: Study design, Data collection, Data analysis and interpretation, Writing the paper, Approval of the final manuscript. I. Han: Study design, Refinement of the study protocol, Data analysis and interpretation, Approval of the final manuscript. S. H. Hong: Refinement of the study protocol, Data analysis and interpretation, Approval of the final manuscript. H. S. Cho: Refinement of the study protocol, Data collection, Approval of the final manuscript. W. Kim: Data collection, Writing the paper, Approval of the final manuscript. H-S. Kim: Study design, Refinement of the study protocol, Approval of the final manuscript, Data analysis and interpretation, Approval of the final manuscript. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by G. Scott and first proof edited by J. Scott.
References 1. Friedlaender GE. Bone grafts; the basic science rationale for clinical applications. J Bone Joint Surg [Am] 1987;69-A:786–790. 2. Zipfel GJ, Guiot BH, Fessler RG. Bone grafting. Neurosurg Focus 2003;14:8. 3. Morone MA, Boden SD, Hair G, et al. The Marshall R. Urist Young Investigator Award; gene expression during autograft lumbar spine fusion and the effect of bone morphogenetic protein 2. Clin Orthop Relat Res 1998;351:252–265.
128
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
4. Lee MJ, Sohn SK, Kim KT, et al. Effect of hydroxyapatite on bone integration in a rabbit tibial defect model. Clin Orthop Surg 2010;2:90–97. 5. Goldberg VM, Stevenson S. The biology of bone grafts. Semin Arthroplasty 1993;4:58–63. 6. Iwamoto Y, Sugioka Y, Chuman H, et al. Nationwide survey of bone grafting performed from 1980 through 1989 in Japan. Clin Orthop Relat Res 1997;335:292–297. 7. Urabe K, Itoman M, Toyama Y, et al. Current trends in bone grafting and the issue of banked bone allografts based on the fourth nationwide survey of bone grafting status from 2000 to 2004. J Orthop Sci 2007;12:520–525. 8. Tehranzadeh J, Mnaymneh W, Ghavam C, Morillo G, Murphy BJ. Comparison of CT and MR imaging in musculoskeletal neoplasms. J Comput Assist Tomogr 1989;13:466–472. 9. Kattapuram SV, Rosol MS, Rosenthal DI, Palmer WE, Mankin HJ. Magnetic resonance imaging features of allografts. Skeletal Radiol 1999;28:383–389. 10. Burchardt H. The biology of bone graft repair. Clin Orthop Relat Res 1983;174:28–42.
11. Jelinek JS, Kransdorf MJ, Moser RP, et al. MR imaging findings in patients with bone-chip allografts. AJR Am J Roentgenol 1990;155:1257–1260. 12. Scarano A, Degidi M, Iezzi G, et al. Maxillary sinus augmentation with different biomaterials: a comparative histologic and histomorphometric study in man. Implant Dent 2006;15:197–207. 13. Wilson JW, Rhinelander FW, Stewart CL. Vascularization of cancellous chip bone grafts. Am J Vet Res 1985;46:1691–1699. 14. Abbott LC, Schottstaedt ER. The evaluation of cortical and cancellous bone as grafting material; a clinical and experimental study. J Bone Joint Surg [Am] 1947;29A:381–414. 15. Vanel D, Lacombe MJ, Couanet D, et al. Musculoskeletal tumors: follow-up with MR imaging after treatment with surgery and radiation therapy. Radiology 1987;164:243–245. 16. Aoki J, Sone S, Fujioka F, et al. MR of enchondroma and chondrosarcoma: rings and arcs of Gd-DTPA enhancement. J Comput Assist Tomogr 1991;15:1011–1016.
THE BONE & JOINT JOURNAL
The MRI appearances of cancellous allograft bone chips after the excision of bone tumours
S. Kang, I. Han, S. H. Hong, H. S. Cho, W. Kim, H-S. Kim From Seoul National University Hospital, Seoul, South Korea
S. Kang, MD, Orthopaedic Surgeon, Clinical fellow I. Han, MD, PhD, Orthopaedic Surgeon, Associate Professor S. H. Hong, MD, PhD, Radiologist, Professor W. Kim, MD, Orthopaedic Surgeon, Clinical Fellow H-S. Kim, MD, PhD, Orthopaedic Surgeon, Professor Department of Orthopaedic Surgery, Seoul National University Hospital, 101 Daehak-ro Jongno-gu, Seoul 110-744, South, Korea. H. S. Cho, MD, Orthopaedic Surgeon, Associate Professor Seoul National University Bundang Hospital, Department of Orthopaedic Surgery, 166 Gume-ro Bundang-gu, Gyeonggi-do 463-707, South, Korea. Correspondence should be sent to Professor H. S. Kim; e-mail: [email protected] ©2015 The British Editorial Society of Bone & Joint Surgery doi:10.1302/0301-620X.97B1. 34517 $2.00 Bone Joint J 2015;97-B:121–8. Received 21 May 2014; Accepted after revision 1 September 2014
Cancellous allograft bone chips are commonly used in the reconstruction of defects in bone after removal of benign tumours. We investigated the MRI features of grafted bone chips and their change over time, and compared them with those with recurrent tumour. We retrospectively reviewed 66 post-operative MRIs from 34 patients who had undergone curettage and grafting with cancellous bone chips to fill the defect after excision of a tumour. All grafts showed consistent features at least six months after grafting: homogeneous intermediate or low signal intensities with or without scattered hyperintense foci (speckled hyperintensities) on T1 images; high signal intensities with scattered hypointense foci (speckled hypointensities) on T2 images, and peripheral rim enhancement with or without central heterogeneous enhancements on enhanced images. Incorporation of the graft occurred from the periphery to the centre, and was completed within three years. Recurrent lesions consistently showed the same signal intensities as those of preoperative MRIs of the primary lesions. There were four misdiagnoses, three of which were chondroid tumours. We identified typical MRI features and clarified the incorporation process of grafted cancellous allograft bone chips. The most important characteristics of recurrent tumours were that they showed the same signal intensities as the primary tumours. It might sometimes be difficult to differentiate grafted cancellous allograft bone chips from a recurrent chondroid tumour. Cite this article: Bone Joint J 2015;97-B:121–8.
Allograft bone grafting involves placing bone, often harvested from human cadavers, into defects in bone to aid healing.1,2 Incorporation of a graft involves the envelopment and interdigitation of the donor tissue with new bone deposited by the recipient.2-4 This process involves many steps. The bone graft stimulates the formation of inflammatory cells, followed by the recruitment of host mesenchymal cells. The primitive host cells then differentiate into bone forming cells. The subsequent processes of revascularisation and necrotic resorption of the graft occur concurrently. Finally, bone is laid down onto the framework of the graft by osteoblasts, followed by remodelling in response to mechanical stress.2,5 The use of allograft bone in reconstructive surgery continues to increase.6,7 MRI is better than other methods of imaging for the evaluation of primary bone tumours.8 Accordingly, MRI is commonly used following curettage and bone grafting in patients with primary intraosseous tumours to ensure the adequacy of the treatment. Oncologists may be asked to evaluate MRI signal intensities at grafted sites to determine whether incorpora-
VOL. 97-B, No. 1, JANUARY 2015
tion is occurring or if there is a recurrence of the tumour. There have been several studies regarding the radiological evaluation of structural allografts9 but few have focused on cancellous allograft bone. There might be differences in the MRI features after bone grafting between structural and cancellous grafts, as the process of incorporation into host bone is different in both.10 As far as we know, there is only one description of the MRI features of grafted cancellous allograft bone chips, involving 18 cases,11 and there is no study of the MRI changes of this allograft bone over time. The main purpose of this study was to investigate the MRI features and incorporation processes of grafted cancellous allograft bone chips in relation to the time interval after grafting. Additionally, we investigated the differences in appearance on MRI between the normal features of grafted bone chips and those of a recurrent tumour.
Patients and Methods Of 202 patients treated by curettage and cancellous allograft bone chips (freeze-dried, from 121
122
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
various manufacturers) for bony tumours between March 2002 and November 2011, we retrospectively reviewed 53 who underwent MRI as part of their follow-up. Pre-operative MRIs were available for all patients. Those patients who had other combined surgical procedures involving the implantation of a prosthesis, bone morphogenetic protein, or a calcium-based bone substitute were excluded (n = 17). Two patients, who were lost to follow-up within one year after MRI without evidence of recurrence, were excluded from the analysis because we cannot confirm the absence of recurrence. Finally, we analysed 34 patients with 66 postoperative follow-up MRIs. No prospective selection criteria were used for MRI follow-up. Patients being treated for a recurrent tumour were reviewed every three to four months for the first year, every six months for the next four years and annually afterwards. Plain radiographs and MRI were alternatively used. Among the 34 patients, the indications for post-operative MRI included pain (n = 10), another medical cause (n = 8), an abnormal finding on plain radiographs (n = 6), requested by the patient (n = 6), and undocumented reasons (n = 4). There were 17 patients with only one post-operative MRI, and the other 17 patients had two or more post-operative MRIs (nine with two MRIs, four with three MRIs, two with four MRIs, one with five MRIs, and one with six MRIs). Thus, there were 23 MRIs of a chondroblastoma, 22 of a low-grade chondrosarcoma, five of an enchondroma, five of a giant cell tumour, four of fibrous dysplasia, two of a simple bone cyst, two of pigmented villonodular synovitis with bony involvement, two of an aneurysmal bone cyst, and one of intraosseous lipoma. The anatomical distribution of the tumours was 29 femora, 15 tibiae, eight pelves, eight humeri, five feet and one hand. The time intervals between bone grafting and follow-up MRI were ≤ six months (n = 12); > six months to ≤ one year (n = 17); > one year to ≤ two years (n = 18); > two years to ≤ three years (n = 11); and > three years (n = 8). The mean time to follow-up imaging was 22.5 months (standard deviation (SD) 20.7; 3 to 94). The size of the tumour was defined by the largest diameter on the pre-operative MRIs, with the mean size overall being 6.7 cm (SD 6.3). According to length of follow-up, the mean size was 6.2 cm (SD 6.6) for ≤ six months, 6.7 cm (SD 6.4) for > six months to ≤ one year, 6.6 cm (SD 5.7) for > one year to ≤ two years, 8.5 cm (SD 6.8) for > two years to ≤ three years, and 5.3 cm (SD 7.1) for > three years. These differences were not statistically significant (p = 0.854, analysis of variance (ANOVA)). No patient received radiotherapy or chemotherapy. MRI was performed with various 1.5- (GE Healthcare, Milwaukee) or 3.0-Tesla MRI scanners (Siemens Medical Solutions, Erlangen, Germany), but in all patients, spin-echo T1-weighted (repetition time (TR) msec/echo time (TE) = 450 to 778/11 to 14) and fast spin-echo T2weighted (TR/TE = 2500 to 4500/60 to 108) images (T1WI and T2WI) with or without fat suppression were acquired, and gadolinium (Gd)-enhanced spin-echo T1WI were also acquired, except in two patients. The field of view varied
from 160 mm × 160 mm to 400 mm × 400 mm and the matrix from 192 × 192 to 512 × 512, slice thickness varied from 3 mm to 5 mm, and various coils were used depending on lesion size and location. The MRIs were reviewed independently by two experienced oncologists (one radiologist (SHH) and one orthopaedic surgeon (SK)) and decisions made by consensus. The typical MRI features on T1WI, T2WI, and Gd-enhanced T1WI in the early phase (≤ six months) were determined and then we investigated the changes of the grafted lesion over time on each view. The change of the extent of the lesion was determined after comparison with the pre-operative MRIs and the immediate post-operative plain radiographs. We also noted changes in adjacent structures such as bone-marrow oedema or sclerosis. Artifacts due, for instance, to suture material were ignored. Although some patients suffered a local recurrence of tumour, none had other local complications, such as infection or fracture. Biopsy was performed for patients with suspected recurrence on MRI, and if recurrence was confirmed, the signal intensity of the recurrent lesion was compared with that of the primary tumour on the pre-operative MRIs, as well as with that of the grafted bone chips. In our hospital, a formal radiological report within three working days after MRI scanning is obligatory. In order to determine whether or not a misdiagnosis had occurred, the formal reports were compared with the histological results when MRIs were interpreted as a suspected recurrence, or with follow-up imaging and clinical presentation when MRIs were interpreted as showing no evidence of recurrence. The study had ethical approval.
Results In the early period (≤ six months) after grafting, the grafts showed consistent features on MRI with: on T1WI, relatively homogeneous intermediate or low signal intensities with or without scattered hyperintense foci (speckled hyperintensity); on T2WI, high signal intensities with scattered hypointense foci (speckled hypointensity) and on Gd-enhanced T1WI, peripheral enhancement with or without central heterogeneous enhancement (Fig. 1). These typical features were gradually lost as time passed. The changes over time are summarised in Table I. The size of the defects decreased with progressive incorporation of the graft. The bony incorporation progressed from periphery to centre, as indicated by the finding that the signal intensity of the periphery started to become similar with that of the host bone after six months (Fig. 2). Incorporation usually began between six months and two years after grafting, and was completed by three years after grafting (Table I). As time passed, the proportion of lesions showing typical features decreased. Gd-enhancement also disappeared, and notably, all remnant graft exhibited Gd-enhancement. Surrounding bone-marrow oedema also diminished over time, irrespective of whether remnants of graft were present or not. THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
123
Fig. 1a
Fig. 1b MRIs showing the typical features of grafted cancellous allograft bone chips (CABCs) at a) at six months post-operatively in a 31-year-old man who underwent curettage and grafting for a low-grade chondrosarcoma of the right femur, showing homogeneous intermediate signal intensities (SIs) with speckled hyperintensities on coronal T1 image (left), high SIs with speckled hypointensities on sagittal T2 image (middle), and peripheral rim enhancement with central heterogeneous enhancement on coronal enhanced image (right) and b) at eight months post-operatively in a 33-year-old man with fibrous dysplasia of the right proximal femur, showing homogeneous intermediate SIs without speckled hyperintensities on coronal T1 image (left), and peripheral rim enhancement without central enhancement on coronal Gd-enhanced image (right).
Table I. MRI findings over time (number, %)
Time interval
*
≤ 6 months
Extent of grafted lesion
Presence of typical features†
N
Peripheral Complete Totals disappearance (A) disappearance (B) (A+B)
Residual lesion (n)
T1WI
12
-
-
12 (100)
12 (100)
-
T2WI
Gd-enhanced T1WI
Surrounding bone marrow oedema
Gd-enhancement
12 (100)
11 (100)‡
6/12 (50)
11/11 (100)‡
> 6 months to ≤ 1 year
17
4 (24)
-
4 (24)
17 (100)
14 (82)
10 (59)
17 (100)
7/17 (41)
17/17 (100)
> 1 year to ≤ 2 years
18
9 (50)
3 (17)
12 (67)
15 (83)
12 (80)
5 (33)
13 (87)
7/18 (39)
18/18 (100)
> 2 years to ≤ 3 years
11
6 (55)
5 (46)
11 (100)
6 (55)
4 (67)
1 (17)
3 (50)
2/11 (18)
9/11 (82)
> 3 years
8
-
8 (100)
8 (100)
0 (0)
-
-
-
2/8 (25)
5/7 (57)‡
WI, weighted image * Time interval between bone grafting and MRI † The typical features were identified as homogeneous intermediate or low signal intensity (SI) with/without speckled hyperintensity on T1-weighted image (WI), high SI with speckled hypointensity on T2WI, or peripheral rim enhancement with/without central heterogeneous enhancement on gadolinium (Gd)-enhanced T1WI ‡ The enhanced views were performed in 11 of 12 patients at ≤ 6 months and in seven of eight at > 3 years
Sclerotic components were observed in 22 patients (64.7%) (Fig. 3) with these elements situated peripherally and/or scattered. In some patients, sclerosis had been present pre-operatively. If a sclerotic component was identified, it did not change or disappear with the passage of VOL. 97-B, No. 1, JANUARY 2015
time. The sclerotic areas on MRIs were concordant with those on plain radiography. Details of patients with recurrent tumour or misdiagnosis are shown in Table II. A total of eight patients (23.5%) developed suspicious lesions initially seen on MRIs, and
124
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
Fig. 2 Post-operative T1 images of a 62-year-old man who underwent curettage and cancellous grafting for a lowgrade chondrosarcoma of the proximal humerus, showing the extent of incorporation of the graft over time at (left) ten and (right) 18 months. Incorporation of the graft occurs from periphery to centre (arrow heads).
Fig. 3a
Fig. 3b
MRI views at 40 months after grafting in the proximal tibia. Peripheral sclerosis is observed on a) T1-weighted image, this finding is concordant with the finding on the b) plain radiographs. Scattered sclerotic areas were sometimes observed inside lesions. Once a sclerotic component occurred, it did not change thereafter.
among these there were five (14.7%) with a proven recurrence. The lesions were found not to be a recurrence in three patients (two low-grade chondrosarcomas in the femur and one fibrous dysplasia in the pelvis). Thus, these three patients were misdiagnosed (Figs 4 and 5). Among the 26 patients, whose MRIs were interpreted as showing no evidence of recurrence, one with a low-grade chondrosarcoma of the pelvis was found to have a recurrence. The lesion was initially small and was misinterpreted on MRI as a post-operative fluid collection. However, at the next MRI, the lesion was larger, and a biopsy was performed, which
confirmed recurrence. Among the six patients in whom recurrence was confirmed, five had recurrence reported on MRI and all those lesions had the same signal intensities as had been shown by the primary tumours in the original pre-operative MRI. The indications for MRI in these six patients included pain in two, other medical reasons in two (one for the evaluation of growth plate injury and one for evaluation of ankle sprain), one in response to the patient’s request, and one for an abnormal finding on plain radiographs. There were four misdiagnoses found through the follow-up scanning, which constituted three false-positive THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
125
Table II. Details of recurrence or misdiagnosis Case number
Gender Age
Time interval* (mths)
Diagnosis
Anatomical site
Signs of recurrence on MRI
Actual recurrence
1 2 3 4 5 6 7 8 9
F F M M M F F M M
3 4 6 10 12 12 13 36 52
LCSA LCSA LCSA Chondroblastoma Fibrous dysplasia LCSA Intraosseous lipoma Simple bone cyst PVNS
Femur Pelvis Femur Tibia Pelvis Femur Calcaneus Pelvis Pelvis
Yes No Yes Yes Yes Yes Yes Yes Yes
Yes Yes† No‡ Yes No† No† Yes Yes Yes
22 38 24 15 29 50 49 19 46
LCSA, low-grade chondrosarcoma; PVNS, Pigmented villonodular synovitis with bony involvement * Time interval between bone graft and MRI † Misdiagnosed cases
Fig. 4 MRIs of a 59-year-old woman who had undergone curettage and cancellous grafting with bone chips for a low-grade chondrosarcoma of the right femur 11 months previously (T1WI left, T2WI middle, and Gd-enhanced T1WI right). The images show the typical features with a speckled pattern and peripheral rim enhancement, but were misinterpreted as a recurrent tumour. The speckled pattern and peripheral enhancement is similar and seems to have been misinterpreted as characteristic of a chondroid tumour.
recurrences and one false-negative recurrence. All four misdiagnoses occurred in patients who had follow-up MRIs within one year of grafting.
Discussion The one study dealing with the MRI features of grafted cancellous bone which has been previously published did not include time-dependent changes.11 This may be because curettage and grafting is mostly reserved for benign lesions and recurrence has not been a source of great concern. The present study has addressed this deficiency in the literature. Some consideration should be given to the correct interpretation of the results of this study. First, post-operative MRI was not performed routinely because most of the tumours were benign. The main indications for performing post-operative MRI were abnormal findings on plain radiographs or pain. Thus, there may have been selection bias. Secondly, there was no histological evidence supporting the VOL. 97-B, No. 1, JANUARY 2015
MRI findings of the grafted bone chips. In order to understand more precisely the features of grafted bone chips and the incorporation process, a study that correlates imaging features with histological information is needed. Regarding the typical MRI findings of the grafted bone chips, we found that all lesions had consistent features during the early period (≤ six months) after grafting. On T1WI, there were homogeneous intermediate or low signal intensities, some with speckled hyperintensities. On T2WI, there were high signal intensities with speckled hypointensities. On Gd-enhanced T1WI, there was enhancement of the peripheral rim in all patients, and some had additional central heterogeneous enhancement. Our description of the MRI features as typical of grafted bone chips was similar to the findings of Jelinek et al.11 They further explained that the presence of areas of speckled hyperintensity reflected the underlying morphology of the cancellous bone allograft. We found the same pattern in our study, but some
126
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
Fig. 5a
Fig. 5b MRIs from a man with fibrous dysplasia of the right ilium: images a) pre- and b) 12 months postoperatively. Both show homogeneous isointense signals on T1WI (left column), high signal intensities on T2WI (middle column), and peripheral gadolinium-enhancement (right column). The patient was misdiagnosed with a recurrent tumour, but with hindsight, there is a speckled pattern on T2WI that is present only in the post-operative images. This finding is thought to be a useful differential feature between recurrent tumours and grafted bone chips.
scans started to lose the speckled patterns six months after grafting. Therefore, we suggest that this speckled morphology will be maintained for at least six months after grafting and then gradually disappear. We investigated the process of incorporation of bone chips into host bone. Although there was no supporting histological confirmation, we identified a series of common findings on MRI. First, the process occurred from periphery to centre. This finding is supported by a previous histological study showing that allograft particles located near preexisting bone were surrounded by new bone, while those located near the centre of the graft showed no signs of remineralisation or new bone formation.12 Next, the incorporation process seems to be accompanied by Gd-enhancement, suggesting revascularisation. This seems to be an appropriate interpretation for many reasons. Revascularisation is well-known to accompany incorporation; the enhancement was prominent at the periphery, where actual
incorporation occurs, and all remnant lesions also showed Gd-enhancement. However, there were no MRIs taken within two months of grafting in this study. Furthermore, no Gd-enhancement of cortical allograft has been reported on MRI between two days and two months post-operatively.9 Therefore, it is difficult to determine whether Gd-enhancement would also be present within two months after grafting. Currently we can only confirm that enhancement develops within three months of grafting and seems to last until after the grafted lesion disappears completely. In contrast, although it showed a gradual decrease similar to the Gd-enhancement, surrounding bone-marrow oedema seems to be a post-operative change only. It does not seem to be an essential part of incorporation, because surrounding bone-marrow oedema was only present in ≤ 50% of all lesions at each interval, irrespective of whether any remnants of graft were present. Thirdly, all of the bone chips seemed to be entirely incorporated into the host bone THE BONE & JOINT JOURNAL
THE MRI APPEARANCES OF CANCELLOUS ALLOGRAFT BONE CHIPS AFTER THE EXCISION OF BONE TUMOURS
eventually. This is in agreement with previous studies showing that cancellous grafts tend to heal completely with time, whereas cortical or xenogeneic grafts do not.10,13 Additionally, most incorporation of the allograft bone chips has been accomplished within three years. The size of the residual bone defect continued to diminish with time, with all the lesions having disappeared by > three years. We do not suggest complete incorporation must be completed by three years in every setting, as it will depend on the size of lesion, the quality of the graft and complications such as infection or fracture.14 No lesion in our study was large and no patient had a complication other than local recurrence of tumour. Finally, during the incorporation process, the typical MRI features might disappear gradually in some patients. This might reflect revascularisation or resorption of the graft but histological confirmation is required. In the present study, six patients (17.6%) had local recurrence (Table II), a relatively high incidence considering the benign nature of the tumours. This may reflect selection bias as some of the MRI scans were performed because of a suspicion of recurrence. In a previous report, a lesion with a high post-operative signal intensity on T2WI in a patient who had undergone only surgery without combined radiotherapy was regarded as an active tumour.15 However, all of the grafted bone chips in our study had high signal intensities on T2WI, even though none of the patients had radiotherapy and all the MRIs of recurrent tumours had the same patterns and signal intensities as pre-operatively in T1WI, T2WI, and Gd-enhanced T1WI sequences. Therefore, we suggest that the most important characteristic of a recurrent tumour is that it has the same pattern and signal intensity in each view as the primary lesion had pre-operatively. The four misinterpreted MRIs in our series of 66 MRIs involved four patients. These misdiagnoses were confirmed on the subsequent MRIs or biopsies. Only one scan missed a recurrence of fibrous dysplasia (Fig. 5). Similar signal intensities were observed on both pre- and post-operative MRIs, which revealed homogeneous isointensities on T1WI, hyperintensities on T2WI, and peripheral rim enhancement on Gd-enhanced T1WI. However, there was a speckled pattern on T2WI that was only seen in the grafted lesion on post-operative images, whereas the tumour had a homogeneous high signal intensity without the speckled pattern on T2WI on pre-operative imaging. This might be a significant point of difference. The tumours in two of the three false-positive cases were low-grade chondrosarcomas of the femur (Fig. 4). However, the signal intensity of a chondroid tumour is similar with the typical MRI features of cancellous allograft bone chips (Figs 1 and 4). In these, the speckled pattern and peripheral enhancement were misread as mineralisation and the ring and arc appearance of a chondroid tumour.16 Great caution is required when differentiating a chondroid tumour from grafted bone chips when these typical features are present, although some grafted bone chips naturally lose their typical features over time. It is not possible to measure the diagnostic accuracy VOL. 97-B, No. 1, JANUARY 2015
127
of MRI for recurrent tumour after grafting with bone chips as the series is small and includes mixed pathologies. However, as all four misdiagnoses occurred within the first year of grafting, it seems that MRI may be an excellent modality for the diagnosis of recurrent tumour following cancellous grafting with bone chips, particularly after the one-year follow-up. More than half of the grafted lesions in this study had a sclerotic component with a peripheral and/or scattered pattern (Fig. 3). Although not in every patient, the peripheral sclerotic rim was present pre-operatively as peritumoural sclerosis, and in many patients scattered sclerosis tended to appear after surgery. The scattered sclerosis seemed to reflect the packing of the defect with graft. The sclerotic components seen on MRIs matched those seen on plain radiographs, and if a sclerotic component was detected, it did not change or disappear during the entire follow-up. In conclusion, we evaluated MRI features of grafted cancellous bone chips according to the time interval after grafting. The early MRIs showed relatively homogeneous intermediate or low SIs with/without speckled hyperintensities on T1WI, high signal intensities with speckled hypointensities on T2WI, and peripheral rim enhancement with or without central heterogeneous enhancement on Gd-enhanced view. These are thought to be the typical MRI features of grafted bone chips in the early period. Thereafter, the incorporation of the graft proceeds as follows: Gd-enhancement starts within three months after grafting, the speckled morphology of the bone chips is maintained until at least six months, and the graft is then incorporated into the host bone from periphery to centre. Most incorporation is completed within three years. The most important characteristic associated with recurrent tumour formation is that the lesion shows the same pattern and signal intensity as the primary tumour. It may sometimes be difficult to differentiate grafted bone chips from a recurrent chondroid tumour, although MRI seems to be an excellent modality for doing this, especially more than one year after grafting. S. Kang: Study design, Data collection, Data analysis and interpretation, Writing the paper, Approval of the final manuscript. I. Han: Study design, Refinement of the study protocol, Data analysis and interpretation, Approval of the final manuscript. S. H. Hong: Refinement of the study protocol, Data analysis and interpretation, Approval of the final manuscript. H. S. Cho: Refinement of the study protocol, Data collection, Approval of the final manuscript. W. Kim: Data collection, Writing the paper, Approval of the final manuscript. H-S. Kim: Study design, Refinement of the study protocol, Approval of the final manuscript, Data analysis and interpretation, Approval of the final manuscript. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by G. Scott and first proof edited by J. Scott.
References 1. Friedlaender GE. Bone grafts; the basic science rationale for clinical applications. J Bone Joint Surg [Am] 1987;69-A:786–790. 2. Zipfel GJ, Guiot BH, Fessler RG. Bone grafting. Neurosurg Focus 2003;14:8. 3. Morone MA, Boden SD, Hair G, et al. The Marshall R. Urist Young Investigator Award; gene expression during autograft lumbar spine fusion and the effect of bone morphogenetic protein 2. Clin Orthop Relat Res 1998;351:252–265.
128
S. KANG, I. HAN, S. H. HONG, H. S. CHO, W. KIM, H-S. KIM
4. Lee MJ, Sohn SK, Kim KT, et al. Effect of hydroxyapatite on bone integration in a rabbit tibial defect model. Clin Orthop Surg 2010;2:90–97. 5. Goldberg VM, Stevenson S. The biology of bone grafts. Semin Arthroplasty 1993;4:58–63. 6. Iwamoto Y, Sugioka Y, Chuman H, et al. Nationwide survey of bone grafting performed from 1980 through 1989 in Japan. Clin Orthop Relat Res 1997;335:292–297. 7. Urabe K, Itoman M, Toyama Y, et al. Current trends in bone grafting and the issue of banked bone allografts based on the fourth nationwide survey of bone grafting status from 2000 to 2004. J Orthop Sci 2007;12:520–525. 8. Tehranzadeh J, Mnaymneh W, Ghavam C, Morillo G, Murphy BJ. Comparison of CT and MR imaging in musculoskeletal neoplasms. J Comput Assist Tomogr 1989;13:466–472. 9. Kattapuram SV, Rosol MS, Rosenthal DI, Palmer WE, Mankin HJ. Magnetic resonance imaging features of allografts. Skeletal Radiol 1999;28:383–389. 10. Burchardt H. The biology of bone graft repair. Clin Orthop Relat Res 1983;174:28–42.
11. Jelinek JS, Kransdorf MJ, Moser RP, et al. MR imaging findings in patients with bone-chip allografts. AJR Am J Roentgenol 1990;155:1257–1260. 12. Scarano A, Degidi M, Iezzi G, et al. Maxillary sinus augmentation with different biomaterials: a comparative histologic and histomorphometric study in man. Implant Dent 2006;15:197–207. 13. Wilson JW, Rhinelander FW, Stewart CL. Vascularization of cancellous chip bone grafts. Am J Vet Res 1985;46:1691–1699. 14. Abbott LC, Schottstaedt ER. The evaluation of cortical and cancellous bone as grafting material; a clinical and experimental study. J Bone Joint Surg [Am] 1947;29A:381–414. 15. Vanel D, Lacombe MJ, Couanet D, et al. Musculoskeletal tumors: follow-up with MR imaging after treatment with surgery and radiation therapy. Radiology 1987;164:243–245. 16. Aoki J, Sone S, Fujioka F, et al. MR of enchondroma and chondrosarcoma: rings and arcs of Gd-DTPA enhancement. J Comput Assist Tomogr 1991;15:1011–1016.
THE BONE & JOINT JOURNAL
