A Modified Lodwick-Madewell Grading System for the Evaluation of Lytic Bone Lesions

Transcription

1 Musculoskeletal Imaging Original Research Caracciolo et al. Evaluation of Lytic one Lesions Musculoskeletal Imaging Original Research Jamie T. Caracciolo 1 H. Thomas Temple 2 G. Douglas Letson 3 Mark J. Kransdorf 4 Caracciolo JT, Temple HT, Letson GD, Kransdorf MJ Keywords: bone neoplasms, classification, grading system, Lodwick, radiographs, risk of malignancy DOI: /JR Received January 10, 2015; accepted after revision February 8, Department of Radiology, H. Lee Moffitt Cancer Center, Tampa, FL. 2 Nova Southeastern University, Fort Lauderdale, FL. 3 H. Lee Moffitt Cancer Center, Sarcoma Division, Tampa, FL. 4 Department of Radiology, Mayo Clinic, 5777 E Mayo lvd, Phoenix, Z ddress correspondence to M. J. Kransdorf (kransdorf.mark@mayo.edu). JR 2016; 207: X/16/ merican Roentgen Ray Society Modified Lodwick-Madewell Grading System for the Evaluation of Lytic one Lesions OJECTIVE. Lodwick s well-established grading system of lytic bone lesions has been widely used in predicting growth rate for lytic bone lesions. We applied a Modified Lodwick- Madewell Grading System as an alternative means to categorize lytic bone tumors into those with low, moderate, and high risks of malignancy. MTERILS ND METHODS. retrospective review of the radiographs of 183 bone lesions was performed. Cases were selected to include a broad range of benign and malignant tumors. Readers applied our Modified Lodwick-Madewell Grading System, and consensus was reached in all cases. This modified system consists of grade I, which is composed of grades I and I as listed in the Lodwick system; grade II, which is grade IC in the Lodwick system; and grade III, which is composed of III (changing margination), II (moth-eaten and permeative patterns), and IIIC (radiographically occult). Grading was correlated with the final diagnosis. RESULTS. Of the 183 tumors, 81 were classified as grade I, 54 as grade II, and 48 as grade III. When correlating grade with pathology, we found that 76 of 81 (94%) grade I lesions were benign and 39 of 48 grade III lesions (81%) were malignant. nearly equal number of grade II lesions proved to be benign (29/54; 54%) and malignant (28/54; 53%). CONCLUSION. y expanding Lodwick s grading system to include two additional patterns of disease described by Madewell and colleagues (changing margination and radiographically occult) and by reclassifying them into three distinct grades, we propose a modified system the Modified Lodwick-Madewell Grading System. pplication of this system shows correlation of tumor grade with tumor biologic activity and with risk of malignancy: Grade I lesions are usually benign, grade II lesions carry moderate risk of malignancy, and grade III lesions possess a high likelihood of malignancy. M ore than 50 years ago, Lodwick [1] proposed a radiographic grading system for the evaluation of lytic bone lesions based on lesion margination, which has been shown to be predictive of tumor rate of growth. Since then, this grading system has been widely applied to the assessment of osseous lesions [2, 3]. Studying the radiographic approach to the diagnosis of bone tumors, Lodwick originally analyzed cases of fibrosarcoma of bone from the Codman one Sarcoma Registry at the rmed Forces Institute of Pathology (FIP) and divided them into three groups: geographic (group 1), moth-eaten (group 2), and permeated (group 3). He correlated the pattern of tumor bone destruction with the 5-year survival rates and noted that there were few survivors when lesions were characterized as having permeative destruction, whereas there were many survivors in those with geographic destruction and an intermediate number in those characterized as moth-eaten. Lodwick [1] evaluated this information and, through a detailed observation of radiographic patterns of tumor behavior at various levels of malignancy and estimated the probability of survival on the basis of the pattern of bone destruction depicted on radiographs. Lodwick and colleagues [1, 4] subsequently proposed five patterns of bone destruction: predominantly geographic (grades I, I, and IC), moth-eaten (grade II), and permeated (grade III). Lodwick applied 14 descriptive variables to describe and define the radiologic patterns for each of the five grades. In 1981, Madewell et al. [5] formalized changes to the original classification system by Lodwick [1] while working with pathologists at the FIP to decrease the complexity 150 JR:207, July 2016

2 TLE 1: Modified Lodwick-Madewell Grading System Evaluation of Lytic one Lesions Grade Description Comment I Well-defined geographic lytic lesion with a sclerotic rim Slow-growing or indolent lesion; typically benign I Well-defined geographic lytic lesion with a sharp margin without a sclerotic rim Most lesions are benign, although differential diagnosis may include metastatic disease and myeloma II a Geographic lytic lesion with partial or circumferential ill-defined margins Some benign causes, but differential diagnosis should include malignancy III b III c Focal change in margin, changing margination, or progressive endosteal scalloping on serial radiographs Moth-eaten and permeative patterns of osteolysis (nongeographic osteolysis) of the original classifications (Madewell JE, personal communication, 2016). s chairman and registrar of the department of radiologic pathology at the FIP, Madewell incorporated these changes into the existing Radiologic-Pathology course. Most radiologists in practice today are familiar with this current system of five radiographic grades for lytic lesions, including well-defined geographic lesions with a sclerotic rim (grade I), geographic lesions with a sharp margin without a sclerotic rim (I), geographic lesions with ill-defined margins (IC), lesions with a motheaten appearance (II), and lesions with a permeative appearance (III). It is generally accepted that these radiographic features allow prediction of the rate of lesion growth (slow or fast) and should be used to direct management (follow-up imaging or biopsy). Over time, results have been extrapolated from the distinction of a slow rate of growth versus a fast rate of growth to the differentiation between benign bone lesions and malignant bone lesions. It is largely accepted that higher radiographic grade indicates malignancy, a concept noted in 1964 by Lodwick [1], who said the following: most of us, perhaps without recognizing a logical basis for such a decision, assign a certain growth rate or degree of malignancy to a tumor based on its radiographic image. Focal changes or changes over time indicate increased biologic activity and should raise suspicion for malignancy Scattered and confluent holes in bone giving the impression of arising from multiple foci or innumerable tiny areas of bone destruction that fade imperceptibly from completely normal bone to markedly abnormal bone IIIC d Radiographically occult Normal or near-normal radiographic findings; lesion is seen on advanced imaging such as MRI or PET Note This system includes modifications to the works of Lodwick [1] and Madewell et al. [5]. a Modification: increased grade assignment from IC to II. b Modification: new grade assignment for lesions showing evidence of increased biologic activity suspicious for malignancy. c Modification: nongeographic osteolysis combining previous moth-eaten and permeative patterns of osteolysis. d Modification: new grade assignment is the most aggressive pattern of growth in which tumor cells extend through bone more rapidly than host osteoclasts can produce recognizable osteolysis. However, this assumption has only limited validation. In addition to simplifying the original lesion classification system, Madewell et al. [5] included a changing margin in this classification system to emphasize that a focal region of increased grade assignment at the tumor margin is indicative of increased biologic activity and implies a greater probability of malignancy. They also introduced the concept of an invisible margin by explaining that tumors with the greatest rate of growth can extend through cancellous bone while radiography shows little or no change, resulting in the host bone appearing to be encased in sarcoma on cross-sectional imaging. Recent experience with MRI and PET has shown that this Fig. 1 Schematic illustrates appearances of lytic bone lesions categorized by grade using Modified Lodwick- Madewell Grading System. JR:207, July

3 Caracciolo et al. pattern of disease is not as uncommon as previously suspected, reinforcing the established concept that 30 50% of cancellous bone must be removed before radiolucency can be identified on radiographs [5 7]. With these concepts in mind, we applied the proposed Modified Lodwick-Madewell Grading System as an alternative means to categorize lytic bone tumors into those with low, moderate, and high risks of malignancy. We incorporated the changes suggested by Madewell et al. [5] to validate the correlation between radiographic grade and tumor pathology for both benign and malignant lesions and to develop a modified grading system that more closely groups together lesions with a similar risk of malignancy and that includes two additional patterns of disease. Materials and Methods This investigational protocol was conducted with the approval of the institutional review board and in accordance with the requirements of a retrospective review and HIP; informed consent was not required. Radiographic examinations were evaluated. Reviewers applied our proposed Modified Lodwick-Madewell Radiographic Grading System, which is defined in Table 1 and illustrated in Figure 1. There are several changes in the proposed Fig. 4 Grade II lesion in 53-year-old man. nteroposterior radiograph of knee shows geographic but poorly defined lytic lesion (oval) of distal femur that is partially obscured by patella and subtle periosteal reaction (arrow); biopsy showed metastatic lung carcinoma. Modified Lodwick-Madewell Grading System, but grade I and I lesions remain essentially unchanged from the original Lodwick grading system [1]. Grade I lesions correspond to well-defined geographic lytic lesions with sclerosis (Fig. 2), and grade I lesions correspond to well-defined geographic lytic lesions with a sharp margin without a sclerotic rim (Fig. 3). Fig. 2 Grade I lesion in 18-year-old man. nteroposterior radiograph of shoulder shows geographic lytic lesion of proximal humerus with circumferential sclerotic margin (arrows); I margin, epiphyseal location, and subtle mineralized matrix are consistent with biopsy diagnosis of chondroblastoma. Modified Lodwick-Madewell grade II lesions (Fig. 4) are defined as geographic lytic lesions with at least partial or circumferential ill-defined margins; these lesions would have been classified as grade IC according to the original system. Three radiographic patterns are now defined as grade III lesions in the Modified Lodwick-Madewell Grading System. Modified Lodwick-Madewell grade Fig. 3 Grade I lesion in 20-year-old man. Lateral radiograph of knee shows geographic lytic lesion of proximal tibia without marginal sclerosis (arrows); I margin, metaphyseal location extending to subchondral bone, and absence of matrix are consistent with biopsy diagnosis of giant cell tumor. Fig. 5 Grade III lesion in 50-year-old woman. and, nteroposterior radiograph () and axial unenhanced CT image () of distal femur show geographic lytic lesion with partially circumferential sclerotic margin (black arrows) and focal area of cortical destruction (white arrow). iopsy was directed to areas of increased biologic activity, and biopsy sample showed highgrade malignant fibrous histiocytoma. nalysis of entire lesion after definitive surgery revealed malignant transformation of benign fibroosseous lesion. 152 JR:207, July 2016

4 Evaluation of Lytic one Lesions TLE 2: Diagnoses Included in the Study Group of 183 Cases TLE 3: enign and Malignant Lesions by Grade Diagnosis I Diagnoses I No. (%) of Lesions (n = 183) No. of Lesions enign lesions neurysmal bone cyst 6 enign fibroosseous tumor 4 enign fibrous lesions NOS 4 one cysts (all types) 8 Chondroblastoma 5 Chondromyxoid fibroma 4 Desmoplastic fibroma 1 Enchondroma 15 Fibrous dysplasia 7 Fibroxanthoma 9 Giant cell tumor 16 Intraosseous gout 1 Intraosseous hemangioma 2 Intraosseous lipoma 4 Langerhans cell histiocytosis 5 Osteoblastoma 2 Osteochondroma 1 Osteoid osteoma 6 Osteolysis 1 Osteomyelitis 8 Posttraumatic lesion 1 Total 110 Malignant lesions damantinoma 1 Chondrosarcoma 13 Ewing sarcoma 13 Leiomyosarcoma 1 Lymphoma 8 Malignant fibrous histiocytoma 3 Metastasis 13 Myeloma or plasmacytoma 5 Osteosarcoma 14 Sarcoma NOS 2 Total 73 Note NOS = not otherwise specified. ll Grade I Lesions II III III IIIC ll Grade III Lesions enign 46 (94) 30 (94) 76 (94) 25 (46) 3 (25) 6 (18) 0 (0) 9 (19) Malignant 3 (6) 2 (6) 5 (6) 29 (54) 9 (75) 27 (82) 3 (100) 39 (81) Total III lesions are newly designated and are defined as lesions showing radiographic features suggesting an alteration in tumor biologic behavior. This pattern of a changing margin may present at a single point in time, for example, a transition from a well-defined sclerotic margin juxtaposed with an area of ill-defined margination (Fig. 5) or, alternatively, a change in margination over time when comparing serial radiographs (Fig. 6). lthough margins are always classified by their most aggressive features, this pattern of growth serves as a marker of a change in tumor biology to focally aggressive local growth, which may be an indicator of malignant transformation [5]. Other grade III lesions include more aggressive osteolytic lesions and radiographically occult lesions. Motheaten lesions (previously grade II) and permeative lesions (previously grade III) were considered separately by Lodwick [1]. He acknowledged that they are fundamentally different patterns of bone destruction rather than different stages in the destructive process, but he also noted that combinations of these patterns are common. ccordingly, we have chosen to group them together among the most aggressive patterns of bone destruction by classifying them as grade III (Figs. 7 and 8). Finally, we have defined grade IIIC lesions as radiographically occult lesions presenting with normal or near-normal findings on bone radiographs but with a clinical suspicion for an osseous lesion or a known bone lesion shown on other imaging modalities such as MRI (Fig. 9), bone scintigraphy, or PET/CT. Lesions presenting with only secondary signs of an underlying bone tumor, such as periosteal reaction, but with otherwise normal-appearing bone would also be grade IIIC. Caution is required in the application of the grade IIIC classification to incidental lesions, such as small metaphyseal or diaphyseal enchondromas (Fig. 10); although these focal lesions may be invisible on radiography because of their small size or location, they should not be classified as grade IIIC. In a retrospective review, we applied the Modified Lodwick-Madewell Grading System to the interpretation of the initial diagnostic radiographs of 183 patients with proven bone tumors. Cases were selected from a large group of proven cases to include a wide variety of osteolytic margins forming a test library, similar to the methods used by Lodwick et al. [8] in their 1980 study. The composition of the test library cases is presented in Table 2. ll diagnoses were pathologically proven except four cases: two fibroxanthomas (confirmed on MRI or radiography), one intraosseous lipoma (confirmed at MRI), and one intraosseous hemangioma (confirmed at MRI and contrast-enhanced angiography at the time of embolization). Radiographs were interpreted by one musculoskeletal radiologist and two orthopedic oncologists blinded to the selection system outlined. consensus marginal grade assignment was reached JR:207, July

5 Caracciolo et al. Fig. 6 Grade III lesion in 68-year-old woman., Lateral radiograph of femur shows mineralized central medullary diaphyseal lesion. ecause of paucity of diaphyseal trabecular bone, margins are not well seen; however, this pattern of mineralization and lack of deep endosteal scalloping is most consistent with enchondroma., Obtained 8 years after, follow-up radiograph shows change in margination. New deep scalloping of anterior cortex (arrow) indicates change in biologic activity and raises concern for malignant transformation to chondrosarcoma, which was confirmed at biopsy. for all lesions and was correlated with tumor diagnosis. The total numbers and percentages of benign and malignant tumors in each grade were calculated. Results Of 183 tumors, 81 lesions were classified as grade I (I, 49 lesions; I, 32 lesions), 54 lesions were classified as grade II, and 48 lesions were classified as grade III (III, 12 lesions; III, 33 lesions; IIIC, 3 lesions). There were 110 (60%) benign and 73 (40%) malignant lesions. The grade assignment of each lesion was compared with the final tumor diagnosis. Of 81 grade I lesions, 76 (94%) were benign and 5 (6%) were malignant. Of 54 grade II lesions, 25 (46%) were benign and 29 (54%) were malignant. Of 48 grade III lesions, 9 (19%) were benign and 39 (81%) were malignant. These results support the concept that as radiographic grade increases and tumor margination becomes less well defined, the likelihood of malignancy increases. ccordingly, grade I lesions are almost always benign, grade II lesions carry a moderate risk of malignancy, and grade III lesions possess a high likelihood of malignancy. The total numbers and percentages of benign and malignant diagnoses for each radiographic grade are presented in Table 3. ing the Modified Lodwick-Madewell Grading System, 81% (39/48) of the grade III lesions in our study group proved to be malignant, including those with evidence of changing margin (grade III), those with a nongeographic appearance (III; moth-eaten and permeative patterns), and those that were radiographically occult (IIIC). Conversely, most grade I lesions (94%, 76/81) were benign. These results help validate the concept that a lower grade assignment correlates with benignity and that increasing grade correlates with an increased Discussion Our results support the generally accepted concept that increasing radiographic grade carries a higher risk of malignancy. In apply- Fig. 7 Grade III lesion in 74-year-old man. nteroposterior radiograph of proximal humerus shows moth-eaten lytic lesion with scattered and confluent holes of bony destruction (oval); biopsy confirmed diagnosis of multiple myeloma. Fig. 8 Grade III lesion in 27-year-old woman. nteroposterior radiograph of femur shows permeative lytic lesion blending imperceptibly from markedly abnormal bone to normal bone with medial cortical disruption; biopsy showed osteosarcoma. 154 JR:207, July 2016

6 Evaluation of Lytic one Lesions risk of malignancy. In summary, grade I lesions are almost always benign, and extended interval surveillance may be appropriate. lternatively, grade III lesions in the Modified Lodwick-Madewell Grading System should be viewed as malignant until proven otherwise and biopsy should be strongly considered. In contrast, grade II lesions may be considered indeterminate. In the Modified Grading System, grade II lesions (formerly grade IC) show partial or circumferential ill-defined margination. In our study, 54 grade II lesions were identified; approximately 46% (25/54) were benign and 54% (29/54) were malignant. In this set of patients, clinical factors such as age, site of disease, symptoms, and medical history would significantly influence management, which will likely include biopsy. lthough Lodwick et al. [8] did not emphasize the correlation of growth rate with biologic potential, they did address this issue. Their analysis of a test library of 223 cases included 117 geographic lesions and 106 nongeographic lesions. In assessing these lesions, they noted that approximately 73% of the geographic lesions were benign. Their analysis of the geographic lesions with ill-defined margins that is, the original grade IC lesions and proposed Modified Lodwick-Madewell Grading System grade II lesions showed that 42% were malignant and 58% were benign [8]; these results are similar to ours. Similarly, Lodwick et al. [8] noted that geographic lesions with a sclerotic rim were benign in 93% of cases and malignant in 7%, findings that are almost identical to those of the current study. The results for geographic lesions without sclerosis were more varied in the Lodwick et al. [8] study: They identified malignancy in 44% of these lesions (vs 6% in the current study). This difference in results likely reflects the increased numbers of chondrosarcomas, myelomas, and osteosarcomas in their group I lesions. and group II lesions as well as changes made by Madewell et al. [5] in the criteria for this designation. In their analysis of Fig. 9 Grade IIIC lesion in 70-year-old woman., Lateral radiograph of distal tibia is unremarkable; biopsy revealed -cell lymphoma., Coronal T1-weighted MR image of distal tibia shows mass infiltrating marrow and encasing cancellous bone. 106 nongeographic lesions (lesions with permeative or moth-eaten margins), all were malignant (100%), which is greater than the 82% of nongeographic lesions that were malignant in the current study. It is worth noting that Lodwick et al. [8] included no cases of osteomyelitis, which likely impacted these results. We excluded three parosteal lesions so that we could compare only intraosseous lesions, whereas Lodwick et al. [8] included these lesions in their test library. Fig. 10 Incidental lesion in 30-year-old man that should not be confused with grade IIIC lesion; this case was not included in study group., PET/CT image obtained during staging evaluation of new suspected soft-tissue sarcoma (large asterisk) of lower extremity shows metastatic regional lymph nodes (small asterisks) and additional intramedullary mass (arrows) in distal femur., Lateral radiograph of distal femur is unremarkable. Thin rim of mineralization (arrow) is seen at inferior margin of proximal metastatic mode. C, Sagittal contrast-enhanced fat-suppressed T1-weighted MR image shows incidental enchondroma (arrow) with characteristic septal and peripheral enhancement. Limb-salvage surgery was not attempted because of extensive regional metastatic disease. C JR:207, July

7 Madewell et al. [5] recognized that margins are often not pure and that combination patterns of osteolysis occur. These areas of increasing grade of margin suggest increasingly aggressive local growth. s is commonly seen in giant cell tumors, a combination pattern of well-defined and ill-defined geographic margination may be detected. Given that margin reflects biologic activity, it is a fundamental principle of lesion analysis that margins are classified by their most aggressive features; therefore, the example combination pattern would be considered a grade II lesion in the proposed grading system. In the Modified Lodwick-Madewell Grading System, we have also included a newly designated grade assignment grade III to specifically address changing margination rather than combination margination. Madewell and colleagues [5] differentiated lesions with a combination of margins from those with a changing margin over time. Our experience is that although a combination margin is common, a changing margin is uncommon and is more clinically significant and is frequently a sign of malignant transformation (Fig. 6). s Madewell et al. [5] noted, [the] distinction between a combination margin and a changing margin may be made by observing a sequence of radiographs that document the change in the latter. However, the reality is that old images are frequently unavailable. When old images are unavailable to document time-dependent change, we use the designation of a changing margin if there is a focal marginal change reflecting a localized area of increased biologic activity in a lesion margin (Fig. 5). Rather than apply this classification to margins that are often seen in a combination pattern, we reserve it for cases in which there is a definitive change for example, a lesion with a geographic sclerotic margin transitioning to an ill-defined margin or a lesion with a geographic nonsclerotic margin transitioning to a moth-eaten margin, similar to those illustrated by Madewell et al. [5]. Identification of an area of change serves not only to suggest the changing biologic activity but also to identify the appropriate biopsy site. In our study, 75% of the grade III lesions with a change in margination were malignant. The importance of radiographically occult lesions has been previously recognized. lthough infrequently encountered in years past, radiographically occult lesions have become increasingly more common because of the greater accessibility of advanced imaging studies. To assess these occult lesions in a meaningful way, one must consider truly incidental Caracciolo et al. lesions, such as an asymptomatic cartilaginous lesion identified on bone scintigraphy in a patient undergoing prostate cancer staging, separately from clinically relevant lesions, such as an occult osseous lesion identified on MRI in a patient with pain and normal or near-normal findings on radiographs. lthough osseous lymphoma and occult bone metastases may be the classic examples of this scenario [9, 10], radiographically occult primary bone sarcomas have been reported [11]. In this regard, the most recent appropriateness criteria in the evaluation of primary bone tumors from the merican College of Radiology recommend MRI as the most appropriate imaging examination for the assessment of localized or regional symptoms in patients with negative findings on radiographs or radiographic findings that do not explain clinical symptoms [12]. In any discussion of lytic bone lesions, it is important to note that osteomyelitis can mimic neoplasm. Osteomyelitis can have variable radiographic manifestations depending on the type of causative agent, virulence, and time course. Osteomyelitis has a spectrum of imaging appearances from predominantly sclerotic to purely lytic, including permeative osteolysis. Thus, osteomyelitis continues to be the great mimicker and should always be considered in the appropriate clinical context, particularly in the setting of elevated WC count or other systemic markers of inflammation. Of interest, Lodwick et al. [8] did not include osteomyelitis in their test library. Our study attempts to validate the Modified Lodwick-Madewell Grading System for both benign and malignant lytic bone lesions but does have several limitations. First, our study is limited by its retrospective nature. Second, we used a nonstandard means of case selection. We sought to include a representative mixture of benign and malignant neoplasms in choosing radiographs for review. Thus, case selection was neither random nor consecutive. Rather, we selected cases to ensure inclusion of a broad range of both benign and malignant bone tumors in the sample population in the hopes of showing that the modified grading system could be widely applied. Finally, the overall sample size is relatively small, and some individual grades are clearly underrepresented. Our results support a broader prospective study applying the modified system to a large consecutive sample of diagnostic radiographs. In summary, our study describes the Modified Lodwick-Madewell Grading System for lytic bone lesions. We applied this modified system to 183 bone tumors and found that a low radiographic grade assignment correlates with benignity and that increasing grade correlates with an increasing risk of malignancy. In defining the Modified Lodwick-Madewell Grading System, we expanded Lodwick s original system [1] to include changes made by Madewell et al. [5] and grouped together lesions with similar risk of malignancy. Using this system, we found that grade I lesions are almost always benign, grade II lesions carry a moderate risk of malignancy, and grade III lesions possess a high likelihood of malignancy. cknowledgment We would like to acknowledge and thank Marie Yasher for her efforts and contributions to background research in the preparation of this article. References 1. Lodwick GS. Radiographic diagnosis and grading of bone tumors, with comments on computer evaluation. Proc Natl Cancer Conf 1964; 5: Costelloe CM, Madewell JE. Radiography in the initial diagnosis of primary bone tumors. JR 2013; 200: Madewell JE, Costelloe CM, Haygood TM, Kumar R, Murphy W. one sarcoma imaging. In: Lin PP, Patel S, eds. one sarcoma. New York, NY: Springer, 2013: Lodwick GS, Wilson J, Farrell C, Virtama P, Dittrich F. Determining growth rates of focal lesions of bone from radiographs. Radiology 1980; 134: Madewell JE, Ragsdale D, Sweet DE. Radiologic and pathologic analysis of solitary bone lesions. Part I. Internal margins. Radiol Clin North m 1981; 19: rdran GM. one destruction not demonstrable by radiography. r J Radiol 1951; 24: Lodwick GS. The bone and joints. In: Hodes PJ, ed. tlas of tumor radiology. Chicago, IL: Year ook Medical Publishers, 1971: Lodwick GS, Wilson J, Farrell C, Virtama P, Smeltzer FM, Dittrich F. Estimating rate of growth in bone lesions: observer performance and error. Radiology 1980; 134: ender I. Factors influencing the radiographic appearance of bony lesions. J Endod 1982; 8: Luketich JD, Friedman DM, Weigel TL, et al. Evaluation of distant metastases in esophageal cancer: 100 consecutive positron emission tomography scans. nn Thorac Surg 1999; 68: ; discussion, Jolepalem P, Yeow RY, Cosner D, Seitz JP. Radiographically occult latent radiogenic osteosarcoma uncovered on Tc-99m methylene-diphosphonate bone scintigraphy. J Clin Imaging Sci 2013; 3: Morrison W, Weissman N, Kransdorf MJ, et al. CR appropriateness criteria: primary bone tumors. Reston, V: merican College of Radiology, JR:207, July 2016