- Email: [email protected]
Sonography of Acute Osteomyelitis in Rabbits with Pathologic Correlation
Sonography of Acute Osteomyelitis in Rabbits with Pathologic Correlation1 Jung-Eun Cheon, MD, Hye Weon Chung, MD, Sung Hwan Hong, MD, Whal Lee, MD Kyoung Ho Lee, MD, Chong Jai Kim, MD, Kyung Mo Yeon, MD, Heung Sik Kang, MD
Rationale and Objectives. Ultrasonography (US) has a potential role in the diagnosis of osteomyelitis. The purpose of this study was to determine the characteristic sonographic features of acute osteomyelitis and correlate them with pathologic findings. Materials and Methods. An experimental model of acute osteomyelitis was produced in the tibiae of 20 rabbits. Daily US and plain radiography were performed for 2 weeks. The authors evaluated periosteal reaction, subperiosteal fluid collection, and soft-tissue changes seen with US. A hypoechoic band and a hyperechoic line lying along the cortex were considered positive signs of subperiosteal fluid collection and periosteal reaction, respectively. The findings of periosteal reaction were compared for US and radiography, and pathologic findings were also correlated. Results. The most common sonographic finding was a hypoechoic band along the cortex (21 [75%] of 28 tibiae), usually associated with a linear periosteal reaction (20 [71%] of 28). This juxtacortical abnormal echogenicity corresponded to periosteal elevation with loose fibrovascular connective tissue and granulation, associated with subperiosteal abscess formation. The periosteal reactions were detected with US before they were seen on radiographs. The periosteum showed gradual thickening during the disease process. In 50% of infected tibiae, inflammation or abscess formation was observed in the surrounding soft tissue. Conclusion. US readily demonstrates juxtacortical abnormal echogenicity and soft-tissue infection related to acute osteomyelitis. The abnormal echogenicity correlated well with the pathologic findings of periosteal reaction and subperiosteal abscess. Key Words. Bones, infection; bones, US; osteomyelitis; ultrasound (US), experimental.
Acute osteomyelitis is a common and rapidly destructive pyogenic infection of childhood. Usually hematogenous in origin, it commonly affects the metaphyseal region of actively growing long bones. Early diagnosis is essential in reducing morbidity and mortality and preventing deformities of bone growth in children (1). The clinical diag-
Acad Radiol 2001; 8:243–249 1
From the Departments of Radiology (J.E.C., H.W.C., S.H.H., W.L., K.H.L., K.M.Y., H.S.K.) and Pathology (C.J.K.), Seoul National University College of Medicine and Seoul National University Hospital, 28 Yongon-Dong, Chongno-Gu, Seoul, 110-744, Korea; and the Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul. Received September 6, 2000; revision requested October 30; revision received and accepted November 9. Address correspondence to H.S.K.
©
AUR, 2001
nosis of osteomyelitis is often difficult, however, especially in the early stages, because osteomyelitis, soft-tissue abscess, and cellulitis may have overlapping clinical symptoms and signs. Furthermore, it may be difficult to interpret symptoms and signs in neonates and young children. A variety of modalities are often needed to establish the diagnosis of osteomyelitis. Conventional radiography is still the standard imaging modality for osteomyelitis, and radionuclide scintigraphy is also commonly used (2– 6). Magnetic resonance imaging has also been used as a sensitive technique for the diagnosis of osteomyelitis (7,8). Ultrasonographic (US) evaluation of osseous structures is limited to the cortical surface unless bone destruction is present. Recently, however, several investigators
243
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Figure 1. Periosteal reaction in a rabbit with osteomyelitis. (a) Longitudinal sonogram obtained 2 days after inoculation shows hypoechoic band adjacent to bone, with hyperechoic line (arrows) partly covering this hypoechoic band in the anterior aspect of the tibia. (b) Axial contact radiograph of the corresponding specimen shows a thin, linear periosteal reaction (arrows). (c) Photomicrograph of specimen in b shows thin layer of the periosteal reaction (arrows) (H-E stain; original magnification, ⫻40). (d) Photomicrograph demonstrates subperiosteal inflammatory cell infiltration, suggesting subperiosteal abscess formation (H-E stain; original magnification, ⫻100).
have reported promising results with US in the diagnosis of osteomyelitis (9 –15). Abiri et al (9) demonstrated subperiosteal fluid collection in 10 of 12 patients with osteomyelitis. Howard et al (10) reviewed findings in 59 children suspected of having osteomyelitis and found that 26 of 29 children who were proved to have acute hematogenous osteomyelitis showed characteristic US findings: abnormal fluid collection adjacent to the bone, periosteal
244
elevation, and surrounding soft-tissue swelling. Abiri et al (16) described sonographic findings of experimental osteomyelitis in rabbits, but pathologic findings that support the sonographic features were not specifically addressed. In view of its still limited role, it is important to determine the value of US in diagnosing osteomyelitis. The purpose of our study was to identify the characteristic sonographic findings of experimentally induced acute os-
Academic Radiology, Vol 8, No 3, March 2001
teomyelitis and to establish the pathologic basis for these findings.
MATERIALS AND METHODS Animal Model Twenty New Zealand white rabbits (2–3 kg) were used as experimental animals. The institutional review board guidelines on the care and use of laboratory animals were followed. Osteomyelitis was induced by means of a technique described by Norden and Kennedy (17) and Crane et al (18). Anesthesia was induced with an intramuscular injection of 10 mg/kg ketamine hydrochloride (Ketara; Yuhan Yanghang, Seoul, Korea) and 50 mg/kg xylazine hydrochloride (Rompun; Bayer Korea, Seoul, Korea). The skin overlying the proximal tibial metaphyseal region was cleansed with alcohol and povidone-iodine. An 18gauge Illinois needle (Manan Medical Products, Northbrook, Ill) was passed into the tibial marrow, and proper placement was confirmed fluoroscopically. Forty milligrams of fresh rat feces was injected into the bone marrow, and the cortical defect was filled with a piece of absorbable gelatin sponge. Of 20 rabbits, 14 were infected bilaterally; in the remaining six, which served as controls, the cortex was punctured but nothing was injected into the marrow. Cultures of infected materials (⫻5) yielded (2.3– 4.5) ⫻ 105 colony-forming units (CFU) of Escherichia coli per milliliter (mean, 3.8 ⫻ 105 CFU). Cultures of bone marrow aspirate obtained 2 weeks after inoculation had revealed the same result in a pilot study. In an experimental model of soft-tissue abscess in rabbit thighs, the authors employed similar methods, and cultures of abscess revealed E coli (19). Imaging Plain radiography and US were performed daily for 2 weeks. Plain radiographs of both legs were obtained in anteroposterior and lateral projections, and then US was performed after the animals were anesthetized with an intramuscular injection of 10 mg/kg ketamine hydrochloride. US was performed with a 10-5-MHz linear transducer (HDI 3000; Advanced Technology Laboratories, Bothell, Wash), and a 2-cm-thick US gel pad was used as a standoff and to establish adequate contact with the bone. US scans were obtained at the inoculation site of the proximal tibia, with transverse and lon-
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
gitudinal axes. One investigator (J.E.C.) performed US examinations. Analysis of Radiologic Findings All images were reviewed by two radiologists (J.E.C., H.S.K.). Findings at serial US were described in terms of periosteal reaction, subperiosteal fluid collection, and softtissue changes. The US finding of a hypoechoic band along the cortex was considered a positive sign of a subperiosteal fluid collection, and a hyperechoic line lying adjacent to the cortex was defined as a periosteal reaction. Evolving periosteal reaction was also assessed during serial US. The duration between inoculation and detection of periosteal reaction was compared for US and plain radiography. Pathologic Examination After imaging, animals were sacrificed on the 3rd (n ⫽ 4), 7th (n ⫽ 5), or 14th (n ⫽ 5) day after inoculation by means of an excessive intravenous injection of thiopental sodium (Pentothal; Choong Wae Pharmacy, Seoul, Korea). In the control group, two rabbits were sacrificed at each interval. Immediately after the rabbits were sacrificed, disarticulation was performed at below-knee level. We obtained radiographs of the specimens to determine the precise level of the inoculation site and then cut the specimen at 5-mm intervals, oriented identically to the transaxial plane images in vivo. Radiographs of representative transaxial cut sections were also obtained. A radiographic unit (Faxitron 43805N; Hewlett-Packard, Sunnyvale, Calif) and a film (X-Omat V; Eastman Kodak, Rochester, NY) were used for radiography of representative transaxial cut sections. The specimens were examined grossly before being fixed in 10% formalin for 48 hours and decalcified in 4% silver nitrate. Histologic examination was performed with hematoxylin-eosin (H-E) staining. Representative images were reviewed by a pathologist (C.J.K.) and two radiologists (J.E.C., H.S.K.) working together. Findings from the gross specimens and microscopic examination were evaluated, and correlations were discussed by all three members of the team. RESULTS Imaging Findings With US, a hypoechoic band along the cortex was detected in 21 (75%) of 28 infected tibiae at 1– 4 days
245
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Figure 2. Evolving periosteal reaction in a rabbit with osteomyelitis. (a) Axial sonogram obtained 3 days after inoculation shows thin periosteal reaction (arrows) with subperiosteal low echogenicity. (b) Follow-up sonogram obtained 12 days after inoculation shows irregularly thickened periosteum. The band of subperiosteal low echogenicity has nearly disappeared, and the thick periosteum looks like a thickened cortex (arrows). (c) Plain radiograph obtained 12 days after inoculation demonstrates wedge-shaped periosteal reaction (arrows). (d) Photomicrograph demonstrates thick periosteal reaction in the outer surface of the cortex (arrows) (H-E stain; original magnification, ⫻40).
(mean, 2 days). A hyperechoic line along the cortex or periosteal reaction was detected in 20 (71%) of 28 tibiae at 2–5 days (mean, 3 days). This finding was initially detected at the site of inoculation (Figs 1a, 2a). During days 1–3, subperiosteal low echogenicity and periosteal reaction were found in 20 (71%) of 28 tibiae and 18 (64%) of 28 tibiae, respectively. At follow-up, US revealed an increase in irregularities and thickening of the elevated periosteum in 11 (55%) of 20 infected tibiae 4 –7 days after inoculation. The band of low echogenicity adjacent to the bone gradually disap-
246
peared, and periosteal reaction was indistinguishable from the cortex, resembling cortical thickening (Fig 2b). The periosteal reaction was demonstrated on plain radiographs in 10 (36%) of 28 tibiae at 4 –9 days (mean, 8 days) after inoculation (Fig 2c). Periosteal reactions were detected with US before they could be seen on radiographs. The echogenicity of the surrounding soft tissue was increased in 14 (50%) of 28 infected legs at 3–7 days after inoculation (mean, 5 days) (Fig 3a). At follow-up US, central low echogenicity with posterior enhancement
Academic Radiology, Vol 8, No 3, March 2001
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
Figure 3. Evolving soft-tissue abscess in a rabbit with osteomyelitis. (a) Transverse sonogram obtained 5 days after inoculation shows ill-defined area of increased echogenicity in surrounding soft tissue, in the anterolateral aspect of the tibia (arrow). (b) Follow-up sonogram obtained 3 days later shows well-defined area of low echogenicity with peripheral hyperechoic rim (arrows) in surrounding soft tissue, suggesting abscess formation. (c) Gross specimen cut axially shows cortical erosion (arrows) and adjacent soft-tissue abscess formation. (d) Photomicrograph of specimen in c demonstrates a large abscess (solid arrows) in surrounding soft tissue and extension of inflammation through the eroded cortex (open arrow) (H-E stain; original magnification, ⫻12.5).
and a hyperechoic wall that suggested soft-tissue abscess formation were seen in seven infected legs (Fig 3b). Occasionally, internal high echogenicity with ring-down artifact was also noted in central necrosis. These findings are summarized in Table 1. In the control group, a band of low echogenicity around the puncture site was seen in five (42%) of 12 tibiae and a thin hyperechoic line along the cortex was observed in four (33%) of 12 proximal tibiae. Abnormal echogenicity of surrounding soft tissue was not found in the control group. Periosteal reaction was detected in one tibia on a plain radiograph obtained 2 weeks after inoculation (Table 2).
Pathologic Correlation Pathologically, osteomyelitis was induced in all animals, and marrow canals were filled with inflammatory cells and necrotic tissue. A periosteal reaction was observed in all cases. The hyperechoic lines along the cortex seen with US corresponded to periosteal elevation, and the thickness of the periosteal reaction increased as the disease progressed (Figs 1c, 2d). The subperiosteal hypoechoic band corresponded to subperiosteal abscess formation composed of inflammatory cell infiltration, loose fibrovascular connective tissue, and granulation (Fig 1d). Direct extension of inflammation through eroded
247
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Table 1 Consecutive Radiologic Findings in Rabbits with Osteomyelitis Days of Follow-up Radiologic Findings
1–3 (n ⫽ 28)*
4–7 (n ⫽ 20)
8–14 (n ⫽ 10)
71 (20) 64 (18) 0 (0) 7 (2) 0 (0)
60 (12) 75 (15) 55 (11) 55 (11) 25 (4)
20 (2) 60 (6) 40 (4) 70 (7) 40 (4)
0 (0)
40 (8)
50 (5)
US Subperiosteal hypoechoic band Periosteal reaction Irregular thickening of periosteum Increased soft-tissue echogenicity Soft-tissue abscess Radiography Periosteal reaction
Note.—Results are given as percentages of infected tibiae, with numbers of tibiae in parentheses. * n ⫽ number of infected tibiae.
Table 2 Radiologic Findings in Control Group Days of Follow-up Radiologic Findings US Subperiosteal hypoechoic band Periosteal reaction Irregular thickening of periosteum Increased soft-tissue echogenicity Soft-tissue abscess Radiography Periosteal reaction
1–3 (n ⫽ 12)*
4–7 (n ⫽ 8)
8–14 (n ⫽ 4)
42 (5) 25 (3) ND ND ND
25 (2) 38 (3) ND ND ND
ND 25 (1) ND ND ND
ND
ND
25 (1)
Note.—Results are given as percentages of tibiae, with numbers of tibiae in parentheses. ND ⫽ not detected. * n ⫽ number of tibiae.
periosteum produced surrounding soft-tissue inflammation and abscess (Fig 3c, 3d). In the control group, inflammatory reaction was not seen in the marrow cavity and surrounding soft tissue in any animals. Thin periosteal elevation with periosteal new bone formation was observed in five (42%) of the 12. No subperiosteal inflammation was present. DISCUSSION Osteomyelitis, a common clinical problem in children and adults, often represents a diagnostic challenge. Patients suspected of having osteomyelitis usually undergo plain radiography and radionuclide scintigraphy (1– 6). Although plain radiography is still the standard imaging procedure for evaluating osteomyelitis (2,3), radiographic
248
changes do not typically occur until 10 –14 days after the onset of infection, or until approximately 35%– 40% of bone destruction has occurred (2–5). Radionuclide scintigraphy is sensitive in detecting the presence of infection but is limited by poor spatial resolution and inaccuracy in differentiating bone from soft-tissue infection (3–5). US enables improved detection of osteomyelitis. The primary lesion of osteomyelitis is within the bone, with early involvement of the surrounding soft tissue. Because ultrasound is almost completely reflected by cortical bone, the intraosseous changes of osteomyelitis cannot be evaluated with US, but soft-tissue changes adjacent to the bone can be differentiated. Among other advantages of US, it is noninvasive and relatively inexpensive, it does not involve ionizing radiation, and its results are immediately available (9 –16).
Academic Radiology, Vol 8, No 3, March 2001
In vivo modeling of osteomyelitis is useful in evaluating the diagnostic value of US, as clinical studies are difficult to perform because of the multiple variables found in the disease process (17,18,20). The most commonly used osteomyelitis model was developed by Norden and Kennedy in 1970 (17). This New Zealand white rabbit model involves intramedullary injection of a sclerosing agent (sodium morrhuate) followed by injection of 3 ⫻ 106 CFU of Staphylococcus aureus and sterile saline solution. Inoculation of organisms into the bone initiated an acute inflammatory response, resulting in ischemia, bone liquefaction, and necrosis (17). As the exudative response continued, the infection spread into the diaphysis or penetrated into the subperiosteal space, lifting the periosteum and forming a subperiosteal fluid collection or abscesses. This finding gave a plausible explanation for the frequent appearance of abnormal juxtacortical echogenicity at US. As the pressure of the marrow increased, inflammation extended to the surrounding soft tissues through the eroded periosteum, and soft-tissue abscess subsequently developed (Fig 3). In our experiment, we modified the method of Norden and Kennedy (17) to produce progressive osteomyelitis more effectively. Our study has several limitations, however. First, we used a transcortical approach to deliver the pathogen. Inserting an 18-gauge needle through the cortex of a small rabbit tibia is a traumatic procedure. In the control group, we found hypoechoic fluid collections and periosteal reactions in 42% and 33% of cases, respectively, findings that were not related to infection. This experimental model may simulate changes of posttraumatic osteomyelitis rather than hematogenous osteomyelitis. Second, we replaced the causative microorganism with E coli because we had experienced a high rate of animal mortality in a pilot study in which we used a high inoculum of S aureus. Although we were able to produce progressive osteomyelitis without the extensive animal mortality observed in earlier rabbit studies, E coli is not the usual pathogen in osteomyelitis. Another limitation of our experiment was that the observers were not blinded, and their findings were therefore subject to bias. We tried to overcome this limitation by substantiating sonographic findings with pathologic correlation. The clinical applications of this experiment are also limited, because we studied posttraumatic osteomyelitis rather than hematogenous osteomyelitis. Finally, the time frame for the development of osteomyelitis differs between human beings and our animal model.
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
In conclusion, US can readily demonstrate abnormal juxtacortical echogenicity and surrounding soft-tissue changes. These findings can precede the radiographic changes. Although periosteal reaction and subperiosteal fluid collections are not specific findings, they are early warning signals, set within the overall clinical picture of osteomyelitis. We hope that this study will provide a radiologic and pathologic basis for using US in the evaluation of osteomyelitis. REFERENCES 1. Waldvogel FA, Medoff G, Swartz MN. Osteomyelitis: a review of clinical features, therapeutic considerations and unusual aspects. N Engl J Med 1979; 282:198 –206. 2. Capitanio MA, Kirkpatrick JA. Early roentgen observation in acute osteomyelitis. AJR Am J Roentgenol 1970; 108:488 – 496. 3. Tumeh SS, Aliabadi P, Weissman BN, McNeil BJ. Disease activity in osteomyelitis: role of radiography. Radiology 1987; 165:781–784. 4. Duszynski DO, Kuhn JP, Afshani E, Riddlesberger MM Jr. Early radionuclide diagnosis of acute osteomyelitis. Radiology 1975; 117:337– 340. 5. Lewin JS, Rosenfield NS, Hoffer PB, Downing D. Acute osteomyelitis in children: combined Tc-99m and Ga-67 imaging. Radiology 1986; 158:795– 804. 6. Allwright SJ, Miller JM, Gilsanz V. Subperiosteal abscess in children: scintigraphic appearance. Radiology 1991; 179:725–729. 7. Sammak B, Abd El Bagi M, Al Shahed M, et al. Osteomyelitis: a review of currently used imaging techniques. Eur Radiol 1999; 9:894 – 900. 8. Dangman BC, Hoffer FA, Rand FF, O’Rourke EJ. Osteomyelitis in children: gadolinium-enhanced MR imaging. Radiology 1992; 182:743– 747. 9. Abiri MM, Kirpekar M, Ablow RC. Osteomyelitis: detection with US. Radiology 1989; 172:509 –511. 10. Howard CB, Einhorn M, Dagan R, Nyska M. Ultrasound in diagnosis and management of acute haematogenous osteomyelitis in children. J Bone Joint Surg Br 1993; 75:79 – 82. 11. Nath AK, Sethu AU. Use of ultrasound in osteomyelitis. Br J Radiol 1992; 65:649 – 652. 12. Kaiser S, Rosenborg M. Early detection of subperiosteal abscesses by sonography: a means for further successful treatment in pediatric osteomyelitis. Pediatr Radiol 1994; 24:336 –339. 13. Mah ET, LeQuesne GW, Gent RJ, Paterson DC. Ultrasonic features of acute osteomyelitis in children. J Bone Joint Surg Br 1994; 76:969 – 974. 14. Riebel TW, Nasir R, Nazarenko O. The value of sonography in the detection of osteomyelitis. Pediatr Radiol 1996; 26:291–297. 15. Chao HC, Lin SJ, Huang YC, Lin TY. Color Doppler ultrasonographic evaluation of osteomyelitis in children. J Ultrasound Med 1999; 18: 729 –734. 16. Abiri MM, DeAngelis GA, Kirpekar M, Abou AA, Ablow RC. Ultrasonic detection of osteomyelitis: pathologic correlation in an animal model. Invest Radiol 1992; 27:111–113. 17. Norden CW, Kennedy E. Experimental osteomyelitis. I. A description of the model. J Infect Dis 1970; 122:410 – 418. 18. Crane LR, Kapdi CC, Wolfe JN, Silberberg BK, Lerner AM. Xeroradiographic, bacteriologic and pathologic studies in experimental staphylococcus osteomyelitis. Proc Soc Exp Biol Med 1977; 156:303–314. 19. Kang HS, Chung YK, Jang JJ. Experimental abscess in the thigh of rabbit: magnetic resonance imaging and pathologic correlation. J Kor Radiol Soc 1996; 35:595– 604. 20. Passl R, Mu¨ller C, Zielinski CC, Eibl MM. A model of experimental post-traumatic osteomyelitis in guinea pig. J Trauma 1984; 24:323– 326.
249
Rationale and Objectives. Ultrasonography (US) has a potential role in the diagnosis of osteomyelitis. The purpose of this study was to determine the characteristic sonographic features of acute osteomyelitis and correlate them with pathologic findings. Materials and Methods. An experimental model of acute osteomyelitis was produced in the tibiae of 20 rabbits. Daily US and plain radiography were performed for 2 weeks. The authors evaluated periosteal reaction, subperiosteal fluid collection, and soft-tissue changes seen with US. A hypoechoic band and a hyperechoic line lying along the cortex were considered positive signs of subperiosteal fluid collection and periosteal reaction, respectively. The findings of periosteal reaction were compared for US and radiography, and pathologic findings were also correlated. Results. The most common sonographic finding was a hypoechoic band along the cortex (21 [75%] of 28 tibiae), usually associated with a linear periosteal reaction (20 [71%] of 28). This juxtacortical abnormal echogenicity corresponded to periosteal elevation with loose fibrovascular connective tissue and granulation, associated with subperiosteal abscess formation. The periosteal reactions were detected with US before they were seen on radiographs. The periosteum showed gradual thickening during the disease process. In 50% of infected tibiae, inflammation or abscess formation was observed in the surrounding soft tissue. Conclusion. US readily demonstrates juxtacortical abnormal echogenicity and soft-tissue infection related to acute osteomyelitis. The abnormal echogenicity correlated well with the pathologic findings of periosteal reaction and subperiosteal abscess. Key Words. Bones, infection; bones, US; osteomyelitis; ultrasound (US), experimental.
Acute osteomyelitis is a common and rapidly destructive pyogenic infection of childhood. Usually hematogenous in origin, it commonly affects the metaphyseal region of actively growing long bones. Early diagnosis is essential in reducing morbidity and mortality and preventing deformities of bone growth in children (1). The clinical diag-
Acad Radiol 2001; 8:243–249 1
From the Departments of Radiology (J.E.C., H.W.C., S.H.H., W.L., K.H.L., K.M.Y., H.S.K.) and Pathology (C.J.K.), Seoul National University College of Medicine and Seoul National University Hospital, 28 Yongon-Dong, Chongno-Gu, Seoul, 110-744, Korea; and the Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul. Received September 6, 2000; revision requested October 30; revision received and accepted November 9. Address correspondence to H.S.K.
©
AUR, 2001
nosis of osteomyelitis is often difficult, however, especially in the early stages, because osteomyelitis, soft-tissue abscess, and cellulitis may have overlapping clinical symptoms and signs. Furthermore, it may be difficult to interpret symptoms and signs in neonates and young children. A variety of modalities are often needed to establish the diagnosis of osteomyelitis. Conventional radiography is still the standard imaging modality for osteomyelitis, and radionuclide scintigraphy is also commonly used (2– 6). Magnetic resonance imaging has also been used as a sensitive technique for the diagnosis of osteomyelitis (7,8). Ultrasonographic (US) evaluation of osseous structures is limited to the cortical surface unless bone destruction is present. Recently, however, several investigators
243
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Figure 1. Periosteal reaction in a rabbit with osteomyelitis. (a) Longitudinal sonogram obtained 2 days after inoculation shows hypoechoic band adjacent to bone, with hyperechoic line (arrows) partly covering this hypoechoic band in the anterior aspect of the tibia. (b) Axial contact radiograph of the corresponding specimen shows a thin, linear periosteal reaction (arrows). (c) Photomicrograph of specimen in b shows thin layer of the periosteal reaction (arrows) (H-E stain; original magnification, ⫻40). (d) Photomicrograph demonstrates subperiosteal inflammatory cell infiltration, suggesting subperiosteal abscess formation (H-E stain; original magnification, ⫻100).
have reported promising results with US in the diagnosis of osteomyelitis (9 –15). Abiri et al (9) demonstrated subperiosteal fluid collection in 10 of 12 patients with osteomyelitis. Howard et al (10) reviewed findings in 59 children suspected of having osteomyelitis and found that 26 of 29 children who were proved to have acute hematogenous osteomyelitis showed characteristic US findings: abnormal fluid collection adjacent to the bone, periosteal
244
elevation, and surrounding soft-tissue swelling. Abiri et al (16) described sonographic findings of experimental osteomyelitis in rabbits, but pathologic findings that support the sonographic features were not specifically addressed. In view of its still limited role, it is important to determine the value of US in diagnosing osteomyelitis. The purpose of our study was to identify the characteristic sonographic findings of experimentally induced acute os-
Academic Radiology, Vol 8, No 3, March 2001
teomyelitis and to establish the pathologic basis for these findings.
MATERIALS AND METHODS Animal Model Twenty New Zealand white rabbits (2–3 kg) were used as experimental animals. The institutional review board guidelines on the care and use of laboratory animals were followed. Osteomyelitis was induced by means of a technique described by Norden and Kennedy (17) and Crane et al (18). Anesthesia was induced with an intramuscular injection of 10 mg/kg ketamine hydrochloride (Ketara; Yuhan Yanghang, Seoul, Korea) and 50 mg/kg xylazine hydrochloride (Rompun; Bayer Korea, Seoul, Korea). The skin overlying the proximal tibial metaphyseal region was cleansed with alcohol and povidone-iodine. An 18gauge Illinois needle (Manan Medical Products, Northbrook, Ill) was passed into the tibial marrow, and proper placement was confirmed fluoroscopically. Forty milligrams of fresh rat feces was injected into the bone marrow, and the cortical defect was filled with a piece of absorbable gelatin sponge. Of 20 rabbits, 14 were infected bilaterally; in the remaining six, which served as controls, the cortex was punctured but nothing was injected into the marrow. Cultures of infected materials (⫻5) yielded (2.3– 4.5) ⫻ 105 colony-forming units (CFU) of Escherichia coli per milliliter (mean, 3.8 ⫻ 105 CFU). Cultures of bone marrow aspirate obtained 2 weeks after inoculation had revealed the same result in a pilot study. In an experimental model of soft-tissue abscess in rabbit thighs, the authors employed similar methods, and cultures of abscess revealed E coli (19). Imaging Plain radiography and US were performed daily for 2 weeks. Plain radiographs of both legs were obtained in anteroposterior and lateral projections, and then US was performed after the animals were anesthetized with an intramuscular injection of 10 mg/kg ketamine hydrochloride. US was performed with a 10-5-MHz linear transducer (HDI 3000; Advanced Technology Laboratories, Bothell, Wash), and a 2-cm-thick US gel pad was used as a standoff and to establish adequate contact with the bone. US scans were obtained at the inoculation site of the proximal tibia, with transverse and lon-
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
gitudinal axes. One investigator (J.E.C.) performed US examinations. Analysis of Radiologic Findings All images were reviewed by two radiologists (J.E.C., H.S.K.). Findings at serial US were described in terms of periosteal reaction, subperiosteal fluid collection, and softtissue changes. The US finding of a hypoechoic band along the cortex was considered a positive sign of a subperiosteal fluid collection, and a hyperechoic line lying adjacent to the cortex was defined as a periosteal reaction. Evolving periosteal reaction was also assessed during serial US. The duration between inoculation and detection of periosteal reaction was compared for US and plain radiography. Pathologic Examination After imaging, animals were sacrificed on the 3rd (n ⫽ 4), 7th (n ⫽ 5), or 14th (n ⫽ 5) day after inoculation by means of an excessive intravenous injection of thiopental sodium (Pentothal; Choong Wae Pharmacy, Seoul, Korea). In the control group, two rabbits were sacrificed at each interval. Immediately after the rabbits were sacrificed, disarticulation was performed at below-knee level. We obtained radiographs of the specimens to determine the precise level of the inoculation site and then cut the specimen at 5-mm intervals, oriented identically to the transaxial plane images in vivo. Radiographs of representative transaxial cut sections were also obtained. A radiographic unit (Faxitron 43805N; Hewlett-Packard, Sunnyvale, Calif) and a film (X-Omat V; Eastman Kodak, Rochester, NY) were used for radiography of representative transaxial cut sections. The specimens were examined grossly before being fixed in 10% formalin for 48 hours and decalcified in 4% silver nitrate. Histologic examination was performed with hematoxylin-eosin (H-E) staining. Representative images were reviewed by a pathologist (C.J.K.) and two radiologists (J.E.C., H.S.K.) working together. Findings from the gross specimens and microscopic examination were evaluated, and correlations were discussed by all three members of the team. RESULTS Imaging Findings With US, a hypoechoic band along the cortex was detected in 21 (75%) of 28 infected tibiae at 1– 4 days
245
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Figure 2. Evolving periosteal reaction in a rabbit with osteomyelitis. (a) Axial sonogram obtained 3 days after inoculation shows thin periosteal reaction (arrows) with subperiosteal low echogenicity. (b) Follow-up sonogram obtained 12 days after inoculation shows irregularly thickened periosteum. The band of subperiosteal low echogenicity has nearly disappeared, and the thick periosteum looks like a thickened cortex (arrows). (c) Plain radiograph obtained 12 days after inoculation demonstrates wedge-shaped periosteal reaction (arrows). (d) Photomicrograph demonstrates thick periosteal reaction in the outer surface of the cortex (arrows) (H-E stain; original magnification, ⫻40).
(mean, 2 days). A hyperechoic line along the cortex or periosteal reaction was detected in 20 (71%) of 28 tibiae at 2–5 days (mean, 3 days). This finding was initially detected at the site of inoculation (Figs 1a, 2a). During days 1–3, subperiosteal low echogenicity and periosteal reaction were found in 20 (71%) of 28 tibiae and 18 (64%) of 28 tibiae, respectively. At follow-up, US revealed an increase in irregularities and thickening of the elevated periosteum in 11 (55%) of 20 infected tibiae 4 –7 days after inoculation. The band of low echogenicity adjacent to the bone gradually disap-
246
peared, and periosteal reaction was indistinguishable from the cortex, resembling cortical thickening (Fig 2b). The periosteal reaction was demonstrated on plain radiographs in 10 (36%) of 28 tibiae at 4 –9 days (mean, 8 days) after inoculation (Fig 2c). Periosteal reactions were detected with US before they could be seen on radiographs. The echogenicity of the surrounding soft tissue was increased in 14 (50%) of 28 infected legs at 3–7 days after inoculation (mean, 5 days) (Fig 3a). At follow-up US, central low echogenicity with posterior enhancement
Academic Radiology, Vol 8, No 3, March 2001
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
Figure 3. Evolving soft-tissue abscess in a rabbit with osteomyelitis. (a) Transverse sonogram obtained 5 days after inoculation shows ill-defined area of increased echogenicity in surrounding soft tissue, in the anterolateral aspect of the tibia (arrow). (b) Follow-up sonogram obtained 3 days later shows well-defined area of low echogenicity with peripheral hyperechoic rim (arrows) in surrounding soft tissue, suggesting abscess formation. (c) Gross specimen cut axially shows cortical erosion (arrows) and adjacent soft-tissue abscess formation. (d) Photomicrograph of specimen in c demonstrates a large abscess (solid arrows) in surrounding soft tissue and extension of inflammation through the eroded cortex (open arrow) (H-E stain; original magnification, ⫻12.5).
and a hyperechoic wall that suggested soft-tissue abscess formation were seen in seven infected legs (Fig 3b). Occasionally, internal high echogenicity with ring-down artifact was also noted in central necrosis. These findings are summarized in Table 1. In the control group, a band of low echogenicity around the puncture site was seen in five (42%) of 12 tibiae and a thin hyperechoic line along the cortex was observed in four (33%) of 12 proximal tibiae. Abnormal echogenicity of surrounding soft tissue was not found in the control group. Periosteal reaction was detected in one tibia on a plain radiograph obtained 2 weeks after inoculation (Table 2).
Pathologic Correlation Pathologically, osteomyelitis was induced in all animals, and marrow canals were filled with inflammatory cells and necrotic tissue. A periosteal reaction was observed in all cases. The hyperechoic lines along the cortex seen with US corresponded to periosteal elevation, and the thickness of the periosteal reaction increased as the disease progressed (Figs 1c, 2d). The subperiosteal hypoechoic band corresponded to subperiosteal abscess formation composed of inflammatory cell infiltration, loose fibrovascular connective tissue, and granulation (Fig 1d). Direct extension of inflammation through eroded
247
CHEON ET AL
Academic Radiology, Vol 8, No 3, March 2001
Table 1 Consecutive Radiologic Findings in Rabbits with Osteomyelitis Days of Follow-up Radiologic Findings
1–3 (n ⫽ 28)*
4–7 (n ⫽ 20)
8–14 (n ⫽ 10)
71 (20) 64 (18) 0 (0) 7 (2) 0 (0)
60 (12) 75 (15) 55 (11) 55 (11) 25 (4)
20 (2) 60 (6) 40 (4) 70 (7) 40 (4)
0 (0)
40 (8)
50 (5)
US Subperiosteal hypoechoic band Periosteal reaction Irregular thickening of periosteum Increased soft-tissue echogenicity Soft-tissue abscess Radiography Periosteal reaction
Note.—Results are given as percentages of infected tibiae, with numbers of tibiae in parentheses. * n ⫽ number of infected tibiae.
Table 2 Radiologic Findings in Control Group Days of Follow-up Radiologic Findings US Subperiosteal hypoechoic band Periosteal reaction Irregular thickening of periosteum Increased soft-tissue echogenicity Soft-tissue abscess Radiography Periosteal reaction
1–3 (n ⫽ 12)*
4–7 (n ⫽ 8)
8–14 (n ⫽ 4)
42 (5) 25 (3) ND ND ND
25 (2) 38 (3) ND ND ND
ND 25 (1) ND ND ND
ND
ND
25 (1)
Note.—Results are given as percentages of tibiae, with numbers of tibiae in parentheses. ND ⫽ not detected. * n ⫽ number of tibiae.
periosteum produced surrounding soft-tissue inflammation and abscess (Fig 3c, 3d). In the control group, inflammatory reaction was not seen in the marrow cavity and surrounding soft tissue in any animals. Thin periosteal elevation with periosteal new bone formation was observed in five (42%) of the 12. No subperiosteal inflammation was present. DISCUSSION Osteomyelitis, a common clinical problem in children and adults, often represents a diagnostic challenge. Patients suspected of having osteomyelitis usually undergo plain radiography and radionuclide scintigraphy (1– 6). Although plain radiography is still the standard imaging procedure for evaluating osteomyelitis (2,3), radiographic
248
changes do not typically occur until 10 –14 days after the onset of infection, or until approximately 35%– 40% of bone destruction has occurred (2–5). Radionuclide scintigraphy is sensitive in detecting the presence of infection but is limited by poor spatial resolution and inaccuracy in differentiating bone from soft-tissue infection (3–5). US enables improved detection of osteomyelitis. The primary lesion of osteomyelitis is within the bone, with early involvement of the surrounding soft tissue. Because ultrasound is almost completely reflected by cortical bone, the intraosseous changes of osteomyelitis cannot be evaluated with US, but soft-tissue changes adjacent to the bone can be differentiated. Among other advantages of US, it is noninvasive and relatively inexpensive, it does not involve ionizing radiation, and its results are immediately available (9 –16).
Academic Radiology, Vol 8, No 3, March 2001
In vivo modeling of osteomyelitis is useful in evaluating the diagnostic value of US, as clinical studies are difficult to perform because of the multiple variables found in the disease process (17,18,20). The most commonly used osteomyelitis model was developed by Norden and Kennedy in 1970 (17). This New Zealand white rabbit model involves intramedullary injection of a sclerosing agent (sodium morrhuate) followed by injection of 3 ⫻ 106 CFU of Staphylococcus aureus and sterile saline solution. Inoculation of organisms into the bone initiated an acute inflammatory response, resulting in ischemia, bone liquefaction, and necrosis (17). As the exudative response continued, the infection spread into the diaphysis or penetrated into the subperiosteal space, lifting the periosteum and forming a subperiosteal fluid collection or abscesses. This finding gave a plausible explanation for the frequent appearance of abnormal juxtacortical echogenicity at US. As the pressure of the marrow increased, inflammation extended to the surrounding soft tissues through the eroded periosteum, and soft-tissue abscess subsequently developed (Fig 3). In our experiment, we modified the method of Norden and Kennedy (17) to produce progressive osteomyelitis more effectively. Our study has several limitations, however. First, we used a transcortical approach to deliver the pathogen. Inserting an 18-gauge needle through the cortex of a small rabbit tibia is a traumatic procedure. In the control group, we found hypoechoic fluid collections and periosteal reactions in 42% and 33% of cases, respectively, findings that were not related to infection. This experimental model may simulate changes of posttraumatic osteomyelitis rather than hematogenous osteomyelitis. Second, we replaced the causative microorganism with E coli because we had experienced a high rate of animal mortality in a pilot study in which we used a high inoculum of S aureus. Although we were able to produce progressive osteomyelitis without the extensive animal mortality observed in earlier rabbit studies, E coli is not the usual pathogen in osteomyelitis. Another limitation of our experiment was that the observers were not blinded, and their findings were therefore subject to bias. We tried to overcome this limitation by substantiating sonographic findings with pathologic correlation. The clinical applications of this experiment are also limited, because we studied posttraumatic osteomyelitis rather than hematogenous osteomyelitis. Finally, the time frame for the development of osteomyelitis differs between human beings and our animal model.
SONOGRAPHY OF ACUTE OSTEOMYELITIS IN RABBITS
In conclusion, US can readily demonstrate abnormal juxtacortical echogenicity and surrounding soft-tissue changes. These findings can precede the radiographic changes. Although periosteal reaction and subperiosteal fluid collections are not specific findings, they are early warning signals, set within the overall clinical picture of osteomyelitis. We hope that this study will provide a radiologic and pathologic basis for using US in the evaluation of osteomyelitis. REFERENCES 1. Waldvogel FA, Medoff G, Swartz MN. Osteomyelitis: a review of clinical features, therapeutic considerations and unusual aspects. N Engl J Med 1979; 282:198 –206. 2. Capitanio MA, Kirkpatrick JA. Early roentgen observation in acute osteomyelitis. AJR Am J Roentgenol 1970; 108:488 – 496. 3. Tumeh SS, Aliabadi P, Weissman BN, McNeil BJ. Disease activity in osteomyelitis: role of radiography. Radiology 1987; 165:781–784. 4. Duszynski DO, Kuhn JP, Afshani E, Riddlesberger MM Jr. Early radionuclide diagnosis of acute osteomyelitis. Radiology 1975; 117:337– 340. 5. Lewin JS, Rosenfield NS, Hoffer PB, Downing D. Acute osteomyelitis in children: combined Tc-99m and Ga-67 imaging. Radiology 1986; 158:795– 804. 6. Allwright SJ, Miller JM, Gilsanz V. Subperiosteal abscess in children: scintigraphic appearance. Radiology 1991; 179:725–729. 7. Sammak B, Abd El Bagi M, Al Shahed M, et al. Osteomyelitis: a review of currently used imaging techniques. Eur Radiol 1999; 9:894 – 900. 8. Dangman BC, Hoffer FA, Rand FF, O’Rourke EJ. Osteomyelitis in children: gadolinium-enhanced MR imaging. Radiology 1992; 182:743– 747. 9. Abiri MM, Kirpekar M, Ablow RC. Osteomyelitis: detection with US. Radiology 1989; 172:509 –511. 10. Howard CB, Einhorn M, Dagan R, Nyska M. Ultrasound in diagnosis and management of acute haematogenous osteomyelitis in children. J Bone Joint Surg Br 1993; 75:79 – 82. 11. Nath AK, Sethu AU. Use of ultrasound in osteomyelitis. Br J Radiol 1992; 65:649 – 652. 12. Kaiser S, Rosenborg M. Early detection of subperiosteal abscesses by sonography: a means for further successful treatment in pediatric osteomyelitis. Pediatr Radiol 1994; 24:336 –339. 13. Mah ET, LeQuesne GW, Gent RJ, Paterson DC. Ultrasonic features of acute osteomyelitis in children. J Bone Joint Surg Br 1994; 76:969 – 974. 14. Riebel TW, Nasir R, Nazarenko O. The value of sonography in the detection of osteomyelitis. Pediatr Radiol 1996; 26:291–297. 15. Chao HC, Lin SJ, Huang YC, Lin TY. Color Doppler ultrasonographic evaluation of osteomyelitis in children. J Ultrasound Med 1999; 18: 729 –734. 16. Abiri MM, DeAngelis GA, Kirpekar M, Abou AA, Ablow RC. Ultrasonic detection of osteomyelitis: pathologic correlation in an animal model. Invest Radiol 1992; 27:111–113. 17. Norden CW, Kennedy E. Experimental osteomyelitis. I. A description of the model. J Infect Dis 1970; 122:410 – 418. 18. Crane LR, Kapdi CC, Wolfe JN, Silberberg BK, Lerner AM. Xeroradiographic, bacteriologic and pathologic studies in experimental staphylococcus osteomyelitis. Proc Soc Exp Biol Med 1977; 156:303–314. 19. Kang HS, Chung YK, Jang JJ. Experimental abscess in the thigh of rabbit: magnetic resonance imaging and pathologic correlation. J Kor Radiol Soc 1996; 35:595– 604. 20. Passl R, Mu¨ller C, Zielinski CC, Eibl MM. A model of experimental post-traumatic osteomyelitis in guinea pig. J Trauma 1984; 24:323– 326.
249