Address correspondence and reprint requests to Dr. Derek D. Cissell at the above address. E-mail


The computed tomography (CT) features of tumors involving the nasal cavity and/or paranasal sinuses of 15 horses were reviewed. The 15 tumors included five neuroendocrine tumors/neuroblastomas, two undifferentiated carcinomas, two myxosarcomas, and one each of nasal adenocarcinoma, hemangiosarcoma, chondroblastic osteosarcoma, anaplastic sarcoma, myxoma, and ossifying fibroma. All tumors except the ossifying fibroma were iso- or hypoattenuating relative to masseter muscle. Thirteen of the fifteen tumors exhibited moderate or marked osteolysis of adjacent cortical bone and 14/15 were characterized by destructive changes of the nasal turbinates, nasal septum, and/or infraorbital canal. Ten horses had moderate or marked involvement of the cribriform plate and six had clear intracranial extension of the mass. CT features were compared to radiographic findings for 10 horses. A mass was observed in 10/10 radiographic studies and mass within the caudal maxillary sinus (7/8) and rostral maxillary sinus (6/7) was identified correctly in most horses. The radiographs were least sensitive for identifying masses within the sphenopalatine sinus (0/5), cranium (0/4), and retrobulbar space (1/7) compared to CT. The radiographs also underestimated potential features of malignancy, such as severity of osteolysis or osseous production. While radiographs are a useful screening tool for identification of sinonasal masses, CT provides greater information regarding mass extent, features of malignancy, and important prognostic indicators.


eoplasia of the nasal cavity and paranasal sinuses comprises between 2% and 19% of all equine sinonasal disorders diagnosed at referral equine hospitals.[1-3] In 87 horses with space occupying masses of the nasal cavities and paranasal sinuses, 25% were sinonasal neoplasia with the remainder being ethmoid hematomas, sinonasal cysts, or nasal polyps.[2] Accurate diagnosis of sinonasal neoplasia is important, as it is often associated with a poorer outcome than other sinonasal disorders.[3] Other causes of space occupying masses of the nasal cavities and paranasal sinuses may be difficult to differentiate from sinonasal neoplasia due to similarities in clinical, radiographic, and endoscopic features. [2, 4, 5] Furthermore, transendoscopic biopsy may not be possible or may not be representative of the underlying disease.[4, 6, 7] Computed tomography (CT) is useful for the diagnosis of nasal and paranasal sinus disorders largely because the sinuses and adjacent bones can be evaluated without anatomic superimposition. This feature, in conjunction with excellent contrast and spatial resolution of CT affords a more accurate assessment of the extent and physical features of sinonasal disease than that offered by conventional radiographs.[8] Documentation of the advantages of CT for evaluation of equine sinonasal neoplasia compared to other diagnostic procedures are limited to studies of a single or few horses.[7-13] Our purpose was to describe the CT imaging features of nasal and paranasal sinus neoplasia in horses. A secondary objective was to compare CT findings to those of skull radiographs. We hypothesized that horses with sinonasal neoplasia will have heterogeneously attenuating soft tissue masses with poorly defined margins and patterns of bone involvement indicative of an aggressive disease process. The presence and extent of mass mineralization was expected to vary with tumor type.

Materials and Methods

Medical records from our universities from 1993 to 2008 were reviewed. Patients undergoing CT of the skull with mass lesions involving the nasal cavity or paranasal sinuses and a histologic diagnosis of neoplasia were selected. Patients were only included if tissue samples were currently available for review and neoplasia was confirmed. Fifteen horses were identified. Age ranged from 3 to 27 years with a mean of 15.9 years (median = 19 years). There were five Thoroughbreds, four Quarter Horses, one Quarter Horse cross, and one each of American Miniature Horse, Clydesdale, Morgan, mule, Standardbred, and Warmblood.

Archived hemotoxylin- and eosin-stained histological slides of neoplastic tissues were evaluated by a veterinary pathologist (F.C.M.). For tumors of either neuroendocrine or neuroblastoma type, additional stains (Grimelius, Bodian) and immunohistochemical methods (antibodies to neuronal triple filament, neuron-specific enolase, chromogranin A, synaptophysin, S-100, cytokeratin, vimentin) were employed in an attempt to differentiate these tumors. In instances of myxoma/myxosarcoma, tissue sections were stained with mucicarmine. There were five sarcomas, three carcinomas, five neuroendocrine tumors or neuroblastomas and two as benign tumors. The sarcomas included two myxosarcomas, one hemangiosarcoma, one chondroblastic osteosarcoma, and one anaplastic sarcoma. The carcinomas included two undifferentiated carcinomas and one nasal adenocarcinoma. The two benign tumors consisted of a myxoma and an ossifying fibroma. Six horses had concurrent sinusitis based on fluid analysis, biopsy, and/or postmortem examination.

The five neuroendocrine/neuroblastoma tumors shared common histologic characteristics. The cells were round with variable amounts of clear to eosinophilic cytoplasm. Nuclei were round with stippled chromatin pattern without prominent nucleoli. In all of these tumors the cells were arranged in packets and surrounded by a variably thin to thickened fibrovascular stroma. In one tumor rare rosettes were observed. Special stains and immunohistochemical methods were used in an attempt to further characterize these tumors. However, no clear patterns emerged to distinguish whether the tumors were of neuroendocrine or neuroblastoma phenotype. In addition, the occurrence of cellular rosettes in one tumor does not allow definitive identification as rosettes have been noted in both tumor types.[14] The difficulty in classifying tumors of neuroendocrine or neuroblastoma linage is noted, with some suggesting that these tumors share a common progenitor.[15-17] Because we were not able to fully define the phenotype of this subset of tumors, we chose to classify them all as neuroendocrina/neuroblastoma.

CT images were acquired with a GE CT Scanner Model 9800 or HiSpeed FX/i CT scanner (General Electric Medical Systems, Milwaukee, WI), or Picker PQS CT (Philips Medical Systems, Andover, MA) and reviewed retrospectively by one of the authors (D.D.C.). Recently acquired studies were viewed electronically as DICOM images using commercially available software (E-film 3.0, Merge Healthcare, Milwaukee, WI) and older studies were viewed as analog images on radiographic film with preset bone and soft tissue window and level settings. All images were viewed in the original plane of acquisition (transverse). Images were not reconstructed into other planes due to 6/15 studies being printed film and also the varying slice thickness of the digital images and subsequent variation in quality of reconstructed images. For DICOM images, soft tissues were evaluated using a window width of 350 and a level of 60 Hounsfield units (HU); osseous structures were evaluated using a window width of 2500 and level of 480 HU. Mass size in the x and y axes was measured on the transverse image with the greatest combined dimension using digital calipers (E-film) or ruler for film images. Mass size in the z-axis (rostral to caudal) was estimated from the image collimation thickness and number of images on which the mass was visible. For analog images, mass dimensions in the x and y axes were normalized using the centimeter scale incorporated into the image. When a mass was surrounded by air, its dimensions were determined easily. Size was estimated from distortion of observable anatomy, e.g. bone and nasal turbinates, for masses surrounded by fluid or other soft tissues. On DICOM studies, mass attenuation was measured in HU using an elliptical measuring tool with the region-of-interest defining the largest possible area while excluding adjacent bone or gas-filled sinus. Iodinated contrast medium (500 ml/horse of 668 mg/ml iothalamate sodium [Conray 400, Mallinckrodt Inc, Hazelwood, MO]) was administered intravenously to two horses. In these two horses, mass attenuation was also measured on contrast-enhanced images at the same anatomic location using the same process. HU of masseter muscle were measured on the same image as that for which the mass HU were measured as an internal standard to determine relative mass enhancement. Pre-contrast mass HU were also compared to masseter by a two-tailed, dependent t-test for paired samples. Mass characteristics including margin definition, tissue heterogeneity, mineralization, presence of adjacent osteolysis or periosteal proliferation, and cribriform plate involvement were defined using a semi-quantitative ordinal scale (see Table 1). The location of osteolytic and osseous productive changes was also recorded. The anatomic location(s) of each mass was noted and its point of origin estimated when possible. The presence of fluid in the nasal cavities or paranasal sinuses was noted. Fluid was determined to be present based on the presence of a horizontal gas-fluid interface.

Table 1. Scoring for Assessment of Computed Tomographic Findings
  1. CP, cribriform plate; HU, Hounsfield units.

MarginsClearly definedMargins visible, but less defined due to small difference in HU or irregular mass-soft tissue interfaceMargins poorly defined due to similar HU and irregular mass-soft tissue interfaceIndistinguishable from adjacent tissues due to heterogeneity, similar HU, and markedly irregular mass-soft tissue interface
HeterogeneityHomogeneousMild. Focal, small variations in opacityModerate. Multifocal or regional variations in opacityMarked heterogeneity due to presence of gas, fluid, and/or mineral
MineralizationNoneFocal, <10% of massMultifocal and/or >10% of mass>25% of mass
Anatomic distortionNoneMild. Focal expansion or displacement of nasal turbinates and/or cortical boneModerate. Regional expansion or displacement of nasal turbinates and/or cortical bone.Marked expansion or displacement of nasal turbinates and cortical bone without osteolysis.
Cribriform plateNormalMass abuts CP without erosionErosion of CP with no intracranial extensionDestruction of CP with clear intracranial extension
Bone destructionNoneFocal, smoothly margined osteolysisRegional, irregular osteolysisRegional or diffuse osteolysis disrupting normal bone structure and shape.
Bone productionNoneMild, smooth proliferation or sclerosisModerate, irregularMarked, irregular

When available, skull radiographs were reviewed by one of the authors (D.D.C.); the initial radiology report was also reviewed for all radiographs, including two horses for which the original radiographs were not available. All radiographs or radiograph reports were evaluated for the presence and location of a mass lesion or sinus fluid. If present, the severity of osteolysis and osseous productive changes on radiographs were noted using an ordinal scale (mild = 1, moderate = 2, marked = 3) following the same criteria established for CT image evaluation (see Table 1). The presence and extent of mass lesions, fluid, and severity of osteolytic/productive changes observed on radiographs was compared to CT findings.


Initial collimated image thickness depended on patient size and CT scanner and varied from 3 to 10 mm. Additional, thinner, 2–3 mm, collimated images were acquired in seven horses. All horses had a mass in the nasal cavity or paranasal sinuses on CT examination. Representative images from four different tumor types are displayed in Fig. 1.

Figure 1.

Transverse CT images of four different horses, each with a different tumor type of the nasal cavity and/or paranasal sinuses. The right side of each image represents the left side of the horse. Window width = 2000 HU and level = 200 HU. (A) Anaplastic sarcoma of the left nasal cavity, rostral and caudal maxillary sinuses, and conchofrontal sinus with suppurative sinusitis and osteomyelitis. (B) Myxosarcoma of both nasal cavities and all left paranasal sinuses. Note the expansion and multifocal osteolysis of the maxilla, loss of nasal turbinates, destruction of the left infraorbital canal and nasolacrimal duct, and irregular periosteal reaction of the left maxilla. (C) Undifferentiated carcinoma of the left ethmoid turbinates and conchofrontal sinus. (D) Neuroendocrine tumor/neuroblastoma originating from the left retrobulbar space and occupying the left conchofrontal and sphenopalatine sinuses. Note the osteolysis of the presphenoid bone creating a large defect in the left, ventral aspect of the calvarium.

CT characteristics of the masses and associated CT findings are summarized in Table 2. For the nine digital CT image sets, all but one of the tumors were hypoattenuating compared to masseter muscle on unenhanced images (range HUmass = 32.9–46.8; range HUmasseter = 52.0–68.1). The eight hypoattenuating masses in electronic digital format had HU significantly different than masseter muscle (P <0.01). In the one horse with a hyperattenuating tumor on digital images (ossifying fibroma), the tumor had a mean HU = 241.0. The ossifying fibroma was characterized by diffuse, homogeneous mineralization. For the remaining six horses with analog images, the masses were subjectively isoattenuating compared to masseter muscle. Attenuation of the tumors as estimated by HU increased by 34% and 60% in the horses receiving intravenous (IV) contrast medium, compared to increased masseter muscle attenuation of 4% and 8%, respectively. Thirteen of the 15 tumors were difficult or impossible to distinguish from adjacent soft tissues due to similar attenuation, heterogeneity of the tumor, and/or a markedly irregular mass—soft tissue interface. Ten of the 15 tumors were homogeneous or only mildly heterogeneous with respect to X-ray attenuation. Ten tumors had no mineralization and two had only mild, focal mineralization (myxoma and undifferentiated carcinoma). Two tumors had multifocal mineralization (osteosarcoma and nasal adenocarcinoma) and one had extensive mineralization (ossifying fibroma). None of the soft tissue tumors had mineralization of more than 25% of the volume of the tumor. The tumors were highly destructive with variable periosteal proliferation. Eight of the 15 tumors had marked osteolysis of adjacent cortical bone and five had moderate osteolysis. The one benign soft tissue tumor (myxoma) and a relatively small, malignant tumor (adenocarcinoma) confined to the ethmoid turbinates had only mild cortical bone osteolysis. Fourteen tumors were characterized by destructive changes involving the turbinates, nasal septum, and/or infraorbital canal; 10 had moderate or marked destructive changes of the ethmoid turbinates. Ten horses had moderate or marked involvement of the cribriform plate (see Table 1), and six had clear intracranial extension of the tumor. Osseous productive changes varied within tumor type groups. Five horses had no osseous productive changes or only mild, smooth periosteal production or sclerosis. Nine horses had moderate, irregular osseous productive changes and only one horse had marked, irregular osseous production associated with the tumor.

Table 2. CT Findings of Equine Sinonasal Neoplasia by Tumor Type
 NET/NBSarcomaCarcinomaMyxomaOssifying Fibroma
  1. All values are medians except where otherwise stated. Ranges of values are given in parentheses. CT, computed tomographic; HU, Hounsfield units; NA, nonapplicable, HU could not be measured for this tumor type as no digital images were available; NET/NB, neuroendocrine tumor/neuroblastoma. X-axis, diameter of the mass in the horizontal plane; Y-axis, diameter of the mass perpendicular to the X-axis; Z-axis, length of the mass in the rostral to caudal direction.

Digital images33201
Mean mass size (cm)     
X-axis7.1 (4.5–11.5)7.6 (4.3–9.8)5.8 (3.1–8.2)6.412.4
Y-axis8.7 (5.1–14.5)10.2 (2.9–14.7)4.9 (4.0–5.6)5.011.7
Z-axis12.0 (7.0–22.5)18.9 (14.5–24.0)8.3 (6.5–11.0)7.220.0
Mean HU38.8 (37.5–40.8)36.1 (32.9–39.9)42.6 (38.4–46.8)NA241.0
Differentiation from soft tissues3 (2–3)2 (1–3)2 (2–3)30
Heterogeneity0 (0–2)2 (0–2)1 (0–1)12
Anatomic distortion3 (2–3)1 (0–2)2 (2–3)21
Cortical bone osteolysis3 (2–3)2 (2–3)2 (1–3)12
Cribriform plate involvement3 (2–3)1 (0–2)2 (2–3)21
Turbinate destruction3 (2–3)3 (0–3)3 (2–3)33
Osseous proliferation1 (0–2)2 (1–3)2 (1–2)22

Location and extent of the tumors varied. One horse had involvement of only one paranasal sinus, while six horses had involvement of the nasal cavity and all of the paranasal sinuses unilaterally. Four horses had tumors that extended into one or more contralateral sinuses resulting in bilateral involvement of the conchofrontal sinus (2/4) and/or sphenopalatine sinus (3/4). The most commonly involved sinus was the conchofrontal sinus (all 15 horses) followed by the caudal maxillary sinus (10). The rostral maxillary sinus was involved in 9/10 horses with caudal maxillary sinus masses; no horse had rostral maxillary sinus involvement without mass in the caudal maxillary sinus. Ten horses had tumor extending into the retrobulbar space and nine had tumor in one of the nasal cavities. Determining the point of origin for the tumors was challenging, especially for large tumors and those with concurrent sinusitis and/or fluid within the sinuses. The most common apparent point of origin of the tumors was the retrobulbar area (4); three tumors each arose from within the caudal maxillary sinus, nasal cavity, and from the ethmoid turbinates. One tumor appeared to arise from the conchofrontal sinus and one appeared to originate from the rostral maxillary sinus.

Skull radiographs were made in 10 horses and images were available for review in 8/10 of these. In the other two instances, the original radiographic reports were reviewed. Radiographic findings grouped by tumor type are summarized in Table 3. Skull radiographs were made 0–15 days prior to CT. In all 10 radiographic studies, a mass was observed in the nasal cavity and/or paranasal sinuses but determining the extent and location of the mass was more difficult on radiographs compared to CT. Accuracy of mass localization varied by nasal or sinus compartment. On radiographs, masses were observed most accurately within the maxillary sinuses, with 7/8 and 6/7 masses involving the caudal and rostral maxillary sinuses identified correctly, respectively. Involvement of the conchofrontal sinus was observed radiographically in 8/10 horses, although radiographs failed to reveal bilateral conchofrontal sinus involvement in two of those horses. In contrast, radiographs were least sensitive for detection of mass involvement of the sphenopalatine sinus (0/5) and retrobulbar space (1/7). Intracranial extension of tumor, observed on CT in 4/10 horses, was not observed radiographically for any horse. Assessment of nasal cavity involvement was also less accurate on radiographs compared to CT. Only 4/7 tumors involving the nasal cavities as determined by CT were observed to have nasal cavity involvement on radiographs and two horses were believed to have nasal passage involvement based on radiographs that did not have mass in the nasal cavities on CT. Skull radiographs also tended to underestimate the severity of periosteal proliferation, osteolysis, and the presence of fluid in the nasal cavity or paranasal sinuses as compared to CT. Moderate or marked osteolysis was observed associated with the mass on radiographs for six horses. Moderate osseous productive changes were observed radiographically in only one horse; mild productive changes were observed in five horses; and no osseous production was observed in four horses. A gas-fluid interface was present in one or more paranasal sinuses on radiographs in three horses.

Table 3. Radiographic Findings of Equine Sinonasal Neoplasia by Tumor Type
  1. Number and mass are sums. All other values are medians. Ranges given in parentheses. NET/NB, neuroendocrine tumor/neuroblastoma.

Periosteal proliferation1.5 (1–2)1 (0–1)0 (0–1)0
Osteolysis2.5 (2–3)2 (2–3)0 (0–0)0


Excluding the one benign ossifying fibroma, the 14 other tumors of the nasal cavity/paranasal sinuses had similar CT imaging features. The 13 soft tissue tumors and one chondroblastic osteosarcoma were characterized by a homogeneous, poorly-defined mass that was iso- or mildly hypoattenuating compared to masseter muscle.

Although only two horses had IV contrast-enhanced CT performed, both had soft tissue masses that enhanced approximately eight times more than masseter muscle. If this is a consistent feature of sinonasal neoplasia, contrast-enhanced CT should help differentiate tumor margins from sinus fluid and possibly from normal adjacent soft tissues. There is limited information available on the utility, ideal dosage, and associated risks of systemic, IV contrast medium administration in horses.[10, 18-21] Until more information becomes available regarding the risks and benefits of IV iodinated contrast media in horses, the decision to administer IV iodinated contrast media for evaluation of sinonasal masses should be made on a patient-by-patient basis.

CT may help differentiate sinonasal neoplasia from other sinonasal diseases in the horse. Ethmoid hematomas have been found to be iso- to hyperattenuating compared to masseter muscle with well-delineated or only mildly obscured margins. [22] Furthermore, 63% of the ethmoid hematomas were described as having a “hyperdense swirling” appearance[22] in contrast to the relatively homogeneous, soft tissue appearance of most of the tumors described here. In a report describing alveolitis and sinusitis in horses, [23] common CT findings included moderate thickening of the respiratory epithelium; complete opacification of the rostral maxillary sinus; thickening and sclerosis of the maxilla; and anatomic distortion, especially of the infraorbital canal, associated with encroaching soft tissues. The reported average HU of the maxillary sinuses in horses with severe sinusitis was reported to be –30 to +40,[23] a range that includes 7/9 of the tumors in this report for which HU were measured. A major difference between horses with sinusitis and horses with sinonasal neoplasia is that only two horses with sinusitis were reported to have bone destruction associated with a mass effect,[23] whereas all of the horses with sinonasal neoplasia in this study had some degree of cortical bone lysis. Horses with malignant sinonasal neoplasia tended to have moderate or marked cortical bone lysis. To our knowledge, the CT features of only a single equine sinonasal cyst have been described.[24] The CT appearance of the cyst was that of a large, clearly demarcated mass with osteolysis of the sphenoid and palatine bones and infraorbital canal.[24] Extension of the cyst into the retrobulbar space was also observed.[24] Of the common equine sinonasal diseases, sinonasal cysts and severe sinusitis may be the most difficult to differentiate from neoplasia based on CT appearance.

Radiographs were a useful screening tool in the horses included in this report. A mass was observed radiographically in all 10 horses for which radiographs were made. Furthermore, moderate or marked osteolysis was recognized on radiographs in 6/10 horses in this study. In a larger report of equine sinonasal disease, radiographic abnormalities were observed in 82% of the horses with sinonasal neoplasia, but only considered diagnostically useful in 23% of the horses.[2] Radiographs accurately demonstrated disease in the maxillary and conchofrontal sinuses in the majority of horses in this study, but were poor for identification of disease in the sphenopalatine sinuses, retrobulbar space, or cranial cavity.

The one ossifying fibroma in this study differed from the other tumor types in its diffuse, hyperattenuating opacity and extensive mineralization (Fig. 2). A previous report of three osteomas and one ossifying fibroma in horses [25] included CT images of the ossifying fibroma. In that report, the ossifying fibroma appeared as a soft tissue mass with HU of 47–79 and multifocal, hyperattenuating areas of mineralization.[25] The CT appearance of ossifying fibromas may vary widely depending on the degree of ossification present.

Figure 2.

Transverse CT image of an ossifying fibroma of the right nasal cavity and all right-sided paranasal sinuses. The right side of the image represents the left side of the horse. Window width = 2000 HU and level = 200 HU. Note the diffuse, patchy, mineral opacity of the mass, expansion and thinning of the maxilla, loss of the right nasal turbinates, leftward deviation of the nasal septum, and complete attenuation of the left common nasal meatus.

An unexpected finding was that one-third of the tumors were neuroendocrine/neuroblastoma. The clinical, imaging, and pathologic features of these tumors are similar to those in previous reports.[13, 16, 26-28] Two CT features of neuroendocrine/neuroblastoma differed from other tumor types. First, all five involved the retrobulbar space with various degrees of paranasal sinus involvement. Four of the neuroendocrine/neuroblastomas appeared to originate from the retrobulbar space and the fifth appeared to originate from the ethmoid turbinates. Second, all five horses also had marked erosion of the cribriform plate and 4/5 had clear extension of tumor into/from the cranial vault. Although five tumors besides the neuroendocrine/neuroblastoma had retrobulbar involvement and five other tumors caused erosion of the cribriform plate, none of the other tumor types appeared to arise from the retrobulbar space and only one, an undifferentiated carcinoma, caused clear intracranial extension via the cribriform plate.

Features of neuroendocrine/neuroblastoma in this report agree with other reports in that these tumors are regionally invasive masses of older horses, originating from the retrobulbar space or ethmoid turbinates that commonly erode the cribriform plate and extend into the cranial vault.[13, 16, 26-28] One horse with a retrobulbar neuroendocrine tumor underwent magnetic resonance (MR) and it was suggested that MR imaging was superior for determining involvement of the brain parenchyma and soft tissues of the orbit compared to CT. MR imaging is likely to be a better imaging modality for evaluation of horses with exophthalmos or neurologic signs suggestive of orbital or intracranial disease.

The lack of squamous cell carcinoma was unexpected. Squamous cell carcinoma has been said to be the most common tumor of the nasal cavity and paranasal sinuses.[4, 29] Between 1993 and 2008 in our hospitals, we identified three horses with confirmed squamous cell carcinoma involving a maxillary sinus. All three were euthanized following biopsy and/or radiographs without CT. One horse had radiographs that revealed extensive and aggressive osteolysis.

In summary, CT provides important information regarding disease extent, location, and osseous involvement that can aid in further diagnostic planning such as biopsy or other surgical intervention. Common features of the sinonasal tumors reported here, such as erosion of the cribriform plate, invasion of the cranial vault, and aggressive patterns of osteolysis, may not be readily apparent on radiographs, but have negative impact on prognosis. Radiographs are an effective screening modality for identifying paranasal sinus masses in patients with clinical signs of sinonasal disease, but are poor at identifying retrobulbar, sphenopalatine sinus, and intracranial involvement.


The authors acknowledge Dr. Sean P. McDonough, Department of Biomedical Sciences, Cornell University, Ithaca, NY 14853 for his assistance in obtaining and preparing tissue samples from horses at the Cornell University Hospital for Animals.