Computed Tomography vs.

Magnetic Resonance Imaging of Acute Bacterial Sinusitis: A Rabbit Model

Joseph E. Kerschner, MD,* Michael J. Cruz, MD,* David J. Beste, MD,* Kathleen M. Donahue, PhD, and Karen Sue Kehl, PhD
Purpose: Computed topography (CT) and magnetic resonance imaging (MRI) are important, both clinically and in a research setting, in assessing bacterial sinusitis (BS). The use of CT scanning to evaluate sinus opacication in a reversible model of rabbit acute sinusitis has been reported. MRI offers the potential for better visualization of soft tissue and uid changes within the paranasal sinuses. MRI has potential as a research tool in animal models of sinusitis. This article compares the use of CT and MRI in measuring maxillary sinus opacication in rabbits during experimental, reversible BS. Materials and Methods: In 2 independent trials, New Zealand White rabbits were imaged for baseline anatomy, and BS was generated by sinus inoculation with Staphylococcus aureus. Serial imaging was performed as a measure of the progression and resolution of BS during the trials. Two experienced, independent reviewers then scored each CT and MRI for percent opacication of the maxillary sinus. These scores were analyzed to assess the degree of agreement between the reviewers. Results: The correlation coefficients for CT and MRI were 0.6816 and 0.3584, respectively. The Z-statistic comparing these correlation coefficients was signicant (P .0001), indicating that CT is a more precise measure of reversible BS in this rabbit model. Differences in mean scan time and cost per scan were also signicantly different (P .0001), with CT being both quicker and less expensive. Conclusions: Greater interobserver consistency of scan interpretation, with less time and cost, make CT the preferred tool for measuring BS in this rabbit model. Attributes of MRI such as better resolution of uid-tissue interfaces and custom surface coil design for visualization of specic anatomic structures are discussed as they may increase the effectiveness of MRI as an imaging modality in future sinusitis research. (Am J Otolaryngol 2000;21:298-305. Copyright 2000 by W.B. Saunders Company)

Various techniques are used in experimental sinusitis research. Depending on the parameters of the study, sinus cultures, histology, ciliary activity, computed tomography (CT), and magnetic resonance imaging (MRI) have all been valuable experimental tools and measures of sinusitis.1-4 Image modality improveFrom the *Department of Otolaryngology and Communication Science, and the Department of Radiology and Biophysics, Medical College of Wisconsin, 9000 W Wisconsin Ave, Milwaukee, WI; and the Department of Pathology, Childrens Hospital of Wisconsin, 9000 W Wisconsin Ave, Milwaukee, WI. Presented at the American Rhinologic Society Annual Meeting, San Antonio, TX, September 12, 1998. Address reprint requests to Joseph E. Kerschner, MD, Childrens Hospital of Wisconsin, 9000 W Wisconsin Ave, Milwaukee, WI 53226. Copyright 2000 by W.B. Saunders Company 0196-0709/00/2105-0003$10.00/0 doi:10.1053/ajot.2000.9874
298

ments, clinical availability, and their use in experimental protocols has contributed to the expansion of our knowledge of anatomy and the pathophysiology of sinusitis.5 CT provides excellent delineation of sinus anatomy and the ability to assess progression of sinusitis in a noninvasive manner.5,6 With increasing availability, MRI has also become a viable research tool. Similar to CT, MRI allows for repeated evaluations in a noninvasive manner. MRI has provided an additional modality for imaging sinus mucosal inflammatory responses and also provides high resolution with the elimination of the biological risks associated with ionizing radiation.7,8 Drawbacks of MRI include longer scan time, greater cost, and loss of detail regarding bone anatomy. The potential for delineation of tissue-fluid or fluidfluid interfaces has also been proposed and

American Journal of Otolaryngology, Vol 21, No 5 (September-October), 2000: pp 298-305

CT VERSUS MRI OF ACUTE BACTERIAL SINUSITIS

299

investigated as an advantage in MRI.9,10 Differing opinions exist regarding the value of these 2 modalities in animal models, and no previous study has compared the techniques of CT and MRI as instruments in animal model sinusitis research. This study evaluates the precision of CT and MRI image modalities by comparing interobserver variability when scoring sinusitis progression. Efficiency of the 2 imaging modalities was also compared by evaluating the cost and time required for the 2 techniques. MATERIALS AND METHODS CT Imaging Technique Animal subjects included 9 female Pasteurella-free New Zealand white rabbits 1.5 to 2.0 kg. Scans were performed on a General Electric HiSpeed Advanced System #UCT3, Model 9800 (Fairfield, CT). Standard head placement was obtained with a styrofoam support to obtain coronal views of the maxillary sinuses. Contiguous 2-mm sections were imaged, with 18 to 21 sections obtained per animal. Baseline CT scans were obtained before to sinus or maxillary ostium manipulation. CT scans and subsequent procedures were done under ketamine anesthesia (35 mg/kg). MRI Technique Animal subjects included 16 Pasteurellafree New Zealand white rabbits 1.5 to 2.0 kg. MRI was done on a 3.0 Tesla (3T) Bruker Biospec System (Bruker Medical, Ettlinger, Germany) fitted with a custom-built 8.5-in local 3-axis gradient coil and a quadrature transmit-receive birdcage RF coil. Spin-echo image sequence (TR/TE 2,000/24 ms, matrix 256 256, FOV 10 cm, 4 averages for a total acquisition time 17 min) was used to obtain 13 contiguous 2-mm coronal sections. Standard head placement was obtained with a styrofoam support. Scan times were recorded as acquisition time plus additional time required for system set-up and technical adjustments. Baseline MRI scans were obtained before sinus or maxillary ostium manipulation. MRI scans and subsequent procedures were done under ketamine anesthesia (35 mg/kg).

Bacterial Sinusitis Reversible bacterial sinusitis was established as previously described in our laboratory.6 Briefly, animals in both imaging trials underwent bilateral partial removal of the maxillary bone to allow direct access to the maxillary sinus ostia for reversible occlusion. The maxillary ostia were occluded with Merocel plugs (Merccel, Mystic, CT). Direct microscopic visualization was used to confirm correct placement of the plug. Five days after occlusion, animals underwent bilateral maxillary sinus cultures, as well as a second imaging study. A cefazolin-sensitive Staphylococcus aureus was obtained from clinical samples in the microbiology lab of Childrens Hospital of Wisconsin in Milwaukee. A suspension of 108 CFU/mL was prepared in sterile saline after overnight growth on tryptic soy blood agar at 37C. Maxillary sinus inoculation with 0.5 mL of S. aureus (108 CFU/mL) was done at this time. Follow-up Imaging On day 7 after inoculation, an additional imaging study was done in both trials to evaluate the progression of sinusitis before removal of the ostial plugs. All follow-up scans were done in the same coronal plane and with the same specifications as the baseline imaging previously described. The ostial plugs were then removed and additional serial scans were done to evaluate sinusitis resolution. In the CT trial, scans were obtained every 3 days after ostial plug removal until the sinusitis resolved. The MRI group received a follow-up scan on a daily basis, to generate more data regarding sinusitis resolution for an additional study, resulting in a greater number of follow-up scans. Some animals resolved their sinusitis more quickly than others and further scans were not obtained once the maxillary sinus returned to normal. Animals were followed for a maximum of 21 days after inoculation. A total of 27 follow-up CT and 89 follow-up MRI scans were obtained postinoculation. CT and MRI Sinus Opacication Scoring Two of the authors (JK and MC) independently scored each imaging study. The inde-

300

KERSCHNER ET AL

pendent reviewers estimated percent opacification of the maxillary sinus for each of the contiguous scans. The average opacification of the entire maxillary sinus was then calculated and given a score based on the percentage of opacification. These authors then graded left and right maxillary sinuses for percent opacification with 1 for less than 10% opacification, 2 for 10% to 19% opacification, 3 for 20% to 29% opacification, 4 for 30% to 39% opacification, 5 for 40% to 49% opacification, 6 for 50% to 59% opacification, 7 for 60% to 69% opacification, 8 for 70% to 79% opacification, 9 for 80% to 89% opacification, and 10 for 90% or greater opacification. This produced a scale of 1 to 10 for sinusitis scores. Statistical Methods Cohens kappa correlation for agreement was used for individual CT reviewer and MRI reviewer agreement of sinusitis scoring. A Z-score was then used to test for interobserver

differences between kappa correlations. Mean scan times were calculated for CT and MRI, and a standard t test was used to analyze differences between means. With 95% confidence intervals used for scan times, cost per scan comparisons were made between imaging techniques based on $140/hr for CT, and $75/hr for MRI. These rates for scanner use were institutional research charges. RESULTS Figures 1 and 2 show selected radiologic views that are representative images of the rabbit paranasal sinuses before ostial occlusion and inoculation. Figures 3 and 4 show CT and MR images of sinus opacification after ostial occlusion and sinus inoculation. As shown in Table 1, the Cohens kappa correlation coefficient for agreement between the 2 CT reviewers was 0.6816 and for MRI, 0.3584. The Z-score comparing the kappa coefficients was statistically significant (P .0001) indicating greater interobserver agreement with

Fig 1. (A) CT scan of anterior maxillary sinus (arrow)preinoculation. (B) MRI scan of anterior maxillary sinus (arrow)preinoculation.

CT VERSUS MRI OF ACUTE BACTERIAL SINUSITIS

301

Fig 2. (A) CT scan of posterior maxillary sinus (short arrow), and ethmoid sinus (long arrow)preinoculation. (B) MRI scan of posterior maxillary sinus (short arrow), and ethmoid sinus (long arrow)preinoculation.

CT imaging in the interpretation and subsequent scoring of sinusitis progression. Mean scan times and 95% confidence intervals for CT and MRI are shown in Table 2. Mean scan time for CT was 2.5028 minutes and 29.0567 minutes for MRI. Scan time 95% confidence intervals for CT and MRI techniques were 2.2352 to 2.7704 minutes, and 26.7441 to 29.4463 minutes, respectively. Thus, approximately 25 additional minutes were required to obtain MRI versus CT images per animal per scan. A t test for unequal variances found mean scan times to be significantly different (P .0001). By using these intervals combined with cost per hour of each technique, cost per scan ranged from $5.21 to $6.46 and $33.43 to $36.81 for CT and MRI techniques, respectively. DISCUSSION The rabbit model of BS was first proposed by Hilding in 1941.11 It has since been used effectively as a method for in vivo analysis of

BS. With this model, investigators have described histologic changes in sinus mucosa, bacterial virulence, mucociliary flow characteristics, and surgical healing associated with experimentally induced BS.1-4,12 This institution recently reported successful CT imaging of reversible sinus inflammation in the rabbit model after temporary maxillary ostial occlusion and bacterial inoculation.6 Critical review of peer-review grant proposals with this technique generated conflicting opinions regarding the value of CT and MRI in sinusitis research. The paucity of published data comparing these 2 techniques prompted this investigation. The 2 imaging methods were compared by using identical protocols for producing reversible, experimental BS. Comparison of methods involved analysis of scan reviewer agreement (precision), and the cost and time (efficiency) associated with each method. CT delineates, in true anatomic crosssections, the intricate anatomical bony framework of the sinuses and allows assessment of

302

KERSCHNER ET AL

Fig 3. (A) CT scan of opacied anterior maxillary sinus (arrow)postinoculation. (B) MRI scan of opacied anterior maxillary sinus (arrow)postinoculation.

the subtle air-bone or tissue-bone interfaces.13-15 In the case of the paranasal sinuses, it is ideal for visualizing complex areas such as the osteomeatal complex ethmoid sinuses.16 These characteristics, along with an increased availability and reduced cost, have made CT the modality of choice for analyzing the paranasal sinuses in most clinical situations. These same qualities have also made CT valuable for sinusitis research. As a noninvasive tool, it allows for repeated measures of sinus inflammation progression or resolution without disturbing the experimentally established sinus cavity environment. The rabbit model has paranasal sinus anatomy that is somewhat more complex than in humans in that the superior and posterior portions of the maxillary sinus can be difficult to differentiate from the ethmoid sinuses. CT permits these areas to be precisely assessed with regard to degree of sinus opacification. A clinical disadvantage of CT is ionizing radiation exposure. This factor is not of signifi-

cant concern in the current model of rabbit BS, but could have implications in clinically based human studies or in animal studies designed for long-term survival.17 CT also provides limited tissue-fluid discrimination. Although there are inherently different attenuation values for tissue and fluid, there is considerable overlap in these values with mucoid secretions and inflamed tissue mucosa. This overlap makes for imprecise image margins when these processes occur in close proximity.7,8 This study evaluated overall maxillary sinus opacification, which encompasses both fluid and softtissue (mucosal) components. MRI also allows for true anatomic crosssection assessment of the paranasal sinuses. In contrast with CT, MRI technology offers the potential advantage of improved tissue-fluid interface discrimination. This difference is possible by specific evaluation and detection of bonded hydrogen molecules within substances, as well as their relaxation times (T1 and T2) in response to various magnetic pulse

CT VERSUS MRI OF ACUTE BACTERIAL SINUSITIS

303

Fig 4. (A) CT scan of opacied posterior maxillary sinus (short arrow), and ethmoid sinus (long arrow) postinoculation. (B) MRI scan of opacied posterior maxillary sinus (short arrow), and ethmoid sinus (long arrow)postinoculation.

sequences.7 This possibility has led to some investigations that used MRI to delineate the components of sinusitis including mucosal tissue changes and fluid secretion.18 A custom surface coil with specific configuration was constructed for this study and TR/TE settings were used to maximize signal to noise ratio. Reviewers found that differentiation could not be made between mucosal inflammation and secreted mucous or fluid in this study. The image spin echo sequence used was predominantly T2-weighted in order to
TABLE 1. CT and MRI Reviewer Agreement
Cohens Kappa () Correlation Coefficient CT scan (N 72) MRI scan (N 210) 0.68 0.36

Standard Error .0693

Z-score* 4.13

.0384

*Test for equality of Kappa statistics. P .0001.

minimize susceptibility effects; however, it is unclear as to how much of an effect the field strength influenced image contrast. Future studies are warranted with this model to further assess the ability to delineate tissue-fluid interfaces as differentiation between mucosal thickening and secretions could provide additional data for understanding of the pathophysiology of BS. Because of the difficulty in defining bone anatomy and maxillary sinus detail with MR images, there was poorer agreement between reviewers in scoring sinus opacification compared with CT (Table 1). This made MR less precise in evaluating sinus opacification, as one measure of a tools precision is interobserver consistency in making measurements.19 In addition, significantly more scan time was required as compared with CT (Table 2). From the standpoint of investigative protocols, this additional time to obtain MR images poses greater anesthesia-related risks for animal models. Perhaps of even greater significance, the

304

KERSCHNER ET AL

TABLE 2. Mean Scan Times Associated and Costs per Scan

Mean Minutes CT scan (N 36) MRI scan (N 105) 2.58 28.09 Standard Error 0.1318 36.8794 0.6812 26.7441-29.463 $33.43-$36.81

t Test*

Scan Time 95% Condence Interval 2.2352-2.7704

Cost per Scan (From Condence Interval) $5.21-$6.46

*t test for difference in mean scan time (unequal variances). P .0001.

increased time adversely effects experimental design and feasibility in the research setting, as protocols often require multiple subjects receiving multiple scans. Measurements that require significantly longer time to obtain may not be able to be incorporated in trials requiring repeat measurements on a large number of animals. Cost of any experimental tool is also a significant consideration for any investigator. Even with the use of fast spin-echo sequences, it would be unlikely that faster image acquisition time would offset the expense of the instrument. Justification of increased time and expense for MRI would be reasonable if future investigations provide clear advantages in measures of sinusitis or understanding of experimental pathophysiology. CONCLUSION CT and MR imaging are noninvasive methods of monitoring the progression, resolution, or anatomical variations associated with paranasal sinusitis and can be successfully used in this animal model of sinusitis. This study shows the usefulness of the New Zealand white rabbit as an in vivo model for the comparison of CT versus MR image modalities during the course of experimental sinusitis. For experienced sinus image reviewers, CT was found to be both more precise and more efficient than MR imaging for in vivo paranasal sinus imaging. Changing technology such as the use of spiral CT, functional contrastenhanced, or improved customization of MRI design may improve the value of these instruments as measures of sinusitis in the future. However, with the current study design, CT is a more effective tool as a noninvasive measure to assess the complex events involved in sinusitis progression and resolution.

ACKNOWLEDGMENTS The authors thank Deborah L. Donahoe, LATG, Otolaryngology Surgery, and Brian Sparland, Department of Biophysics without whose assistance this study could not have been completed. REFERENCES
1. Hinni ML, McCaffrey TV, Kasperbauer JL: Early mucosal changes in experimental sinusitis. Otolaryngol Head Neck Surg 107:537-551, 1992 2. Johansson P, Kumlien J, Carlsoo B, et al: Experimental acute sinusitis in rabbits: A bacteriological and histological study. Acta Otolaryngol (Stockh) 105:357-366, 1988 3. Scharf KE, Lawson W, Shapiro JM, et al: Pressure measurements in the normal and occluded rabbit maxillary sinus. Laryngoscope 105:570-574, 1995 4. Kumlien J, Stiema P, Schiratzki H: Autoradiographic findings in experimental sinusitis in rabbits. Rhinology 29:193-199, 1991 5. Kaliner MA, Osguthorpe JD, Fireman P, et al: Sinusitis: bench to bedside current findings, future directions. Otolaryngol Head Neck Surg 116:1-20, 1997 (suppl) 6. Beste DJ, Capper DT, Shaffer K, et al: Antimicrobial effect on rabbit sinusitis after temporary ostial occlusion. Am J Rhinol 11:485-489, 1997 7. Moore J, Potchen M, Waldenmaier N, et al: Highfield magnetic resonance imaging of paranasal sinus inflammatory disease. Laryngoscope 96:267-271, 1986 8. Som PM, Shapiro MD, Biller HF, et al: Sinonasal tumors and inflammatory tissues: Differentiation with MR imaging. Radiology 167:803-807, 1988 9. Chan KH, Swarts DJ, Tan L: Middle ear mucosal inflammation: An in-vivo model. Laryngoscope 104:970980, 1994 10. Reo NV, Alexander ME, Goel R. A nuclear magnetic resonance investigation of the upper airways in ferrets. II. Contrast-enhanced imaging to distinguish vascular from other nasal fluids. Magn Reson Med 27:34-43, 1992 11. Hilding AC: Experimental sinus surgery: Effect of operative windows on normal sinuses. Ann Otol Rhinol Laryngol 50:379-392, 1941 12. Westrin KM, Stierna P, Kumlien J, et al: Induction, course, and recovery of maxillary sinusitis: A bacteriological and histological study in rabbits. Am J Rhinol 4:61-64, 1990 13. McCullough EC: Basic physics of x-ray computed topography, in Norman D, Korobkin M, Newton TH (eds):

CT VERSUS MRI OF ACUTE BACTERIAL SINUSITIS

305

Computed Topography. St. Louis, Mosby-Year Book, 1977, pp 139-150 14. Hounsfield GN: Picture quality of computed topography. AJR Am J Roentgenol 127:3-9, 1976 15. Mafee MF: Imaging methods for sinusitis. JAMA 269:2608, 1993 16. Mafee MF: Modern imaging of paranasal sinuses and the role of limited sinus computerized topography: Considerations of time, cost, and radiation. Ear Nose Throat J 73:532-546, 1994

17. Evens RG, Mettler FA: National CT use and radiation exposure: United States 1983. AJR Am J Roentgenol 144:1077-1081, 1985 18. Leopold DA, Stafford CT, Sod EW, et al: Clinical course of acute maxillary sinusitis documented by sequential MRI scanning. Am J Rhinol 8:19-28, 1994 19. Hulley SB, Cummings SR: Planning the measurements: Precision and accuracy, in Hulley SB, Cummings SR (eds): Designing Clinical Research. Baltimore, MD, Williams & Wilkins, 1988, pp 31-41