Extranodal Natural Killer/T-Cell Lymphoma Nasal Type: Detection by Computed Tomography Features

The Laryngoscope VC 2014 The American Laryngological, Rhinological and Otological Society, Inc. Extranodal Natural Killer/T-Cell Lymphoma Nasal Type: Detection by Computed Tomography Features Yin-Ping Hsu, MD; Po-Hung Chang, MD; Ta-Jen Lee, PhD; Liang-Yueh Hung, MD; Chi-Che Huang, MD Objectives/Hypothesis: Nasal natural killer/t-cell lymphoma (NKTL) often has an infiltrative pattern in computed tomography that makes them difficult to distinguish from benign inflammatory diseases. This study aimed to design a method of measuring the thickness of the nasal floor and nasal septum, determine the critical value of mucosal thickness that may implicate these NKTL cases from benign inflammatory disease, and finally make a complete flowchart to detect NKTL with minimal mistake. Study Design: Thirty-two patients with nasal NKTL and 173 patients with severe chronic rhinosinusitis with or without polyposis were enrolled. The patients data were collected retrospectively. Methods: All patients underwent standard computed tomography of the paranasal sinuses. The coronal section near the vertical part of the ground lamina was chosen for measurement, and the thickest points along the nasal floor and septum were measured. Results: Patients with NKTL had thicker nasal floors and/or septa than those with chronic rhinosinusitis, recurrent sinusitis, or pansinusitis (P <.001). If the cutoff points of the nasal floor and nasal septum thickness were set at 2.0 mm and 2.5 mm, respectively, the probability of being thicker than the corresponding points in the CRS group was <2%, and the possibility of other diagnoses should be considered. Conclusions: Nasal floor mucosal thickness >2.0 mm or nasal septum mucosal thickness >2.5 mm may be indicators serving as one of several important hints for implicit NKTL. Finally, we established a diagnostic flowchart to include all of these important hints. Key Words: Natural killer/t-cell lymphoma, computed tomography, chronic rhinosinusitis, nasal floor, nasal septum, mucosal thickness, cutoff point. Level of Evidence: 4. Laryngoscope, 124:2670 2675, 2014 INTRODUCTION Extranodal natural killer/t-cell lymphoma (NKTL) is a distinctive lymphoma that is more common in East Asia and Latin America. It accounts for 7% to 10% of all non-hodgkin s lymphomas diagnosed in East Asia and Latin America, but only 1% among Caucasians. 1,2 The imaging features of nasal NKTL are nonspecific such that distinguishing between them from chronic inflammatory diseases or other tumor-like conditions in the sinonasal cavity is regarded as close to impossible. 3,4 In more recent studies using computed tomography (CT) findings on NKTL in the head and neck region, 62.5% to 69% of cases show an infiltrative pattern without a prominent mass. 5,6 In our previous study about CT From the Department of Otolaryngology, Chang Gung Memorial Hospital and Chang Gung University, Taipei, Taiwan. Editor s Note: This Manuscript was accepted for publication July 21, 2014. This study was supported by grants CMRPG3B0151 from Chang Gung Memorial Hospital. The authors have no other funding, financial relationships, or conflicts of interest to disclose. Send correspondence to Chi-Che Huang, MD, Department of Otolaryngology, Chang Gung Memorial Hospital and Chang Gung University, No. 5, Fusing St., Gueishan Township, Taoyuan County 333, Taiwan. E-mail: hcc3110@adm.cgmh.org.tw DOI: 10.1002/lary.24876 findings, NKTL was usually nonenhanced (79.1%), homogenous (100%), unilateral (61.9%), infiltrative (67.4%), and without central necrosis. Only 30.2% of the patients presented with a prominent mass. If NKTL has an infiltrative pattern, it would spread as a diffuse, thin sheet of tumor along the nasal cavity walls to envelop the nasal turbinates and septum, making it difficult to be distinguished from a benign inflammatory disease such as chronic rhinosinusitis (CRS). However, our previous study has observed a different pattern regarding mucosal thickening. One disease-specific sign, different from patients with CRS, was that the mucosa of the nasal cavity was thickened without involvement of the mucosa of the paranasal sinus (40.6%). 6 If the mucosa of the nasal cavity was thickened with involvement of the mucosa of the paranasal sinus, NKTL would be more difficult to distinguish from benign inflammatory disease. However, we noticed that many patients with NKTL have thickened mucosa at the nasal floor or nasal septum, whereas those with rhinosinusitis rarely have thickened mucosa at both sites. 6 A flowchart was supplied in our previous study (Fig. 1a). 6 The accurate mucosal thickness of the nasal floor or nasal septum at which the possibility of NKTL can be determined is unknown, and no previous imaging studies have examined mucosal thickness in these two sites. 3 5,7 A possible reason is the high variability of the 2670

Fig. 1. A flowchart to detect natural killer/t-cell lymphoma (NKTL). (a) The flowchart in our previous study. (b) Newer and more complete flowchart. Every step was tagged as a fraction. Because every step was regarded as an independent step, the denominator was the total of 32 cases of NKTL. *We used mucosal patterns to divide 32 cases into four groups: prominent mass (11/32), nasal cavity thickening without sinus thickening (13/32), nasal cavity thickening with sinus involvement (5/32), and others (3/32). Of the 32 patients in the NKTL group, 31 and 30 patients had measurable nasal floor and septum, respectively. Only one case was not able to be detected by this new flowchart. CT 5computed tomography. mucosal thickness near the nasal vestibule and anterior third of the nasal cavity. The aim of this study was to design a method for measuring the thickness of the nasal floor and nasal septum, and to determine the upper limit of mucosal thickness of benign inflammatory disease, whereby a more complete flowchart would be established. MATERIALS AND METHODS Ethical Considerations The institutional review board (IRB) of Chang Gung Memorial Hospital, Taoyuan County, Taiwan approved the study protocol (IRB no. 100-3482B), and all participants provided informed consent. Patients Forty-three patients with nasal NKTL were diagnosed based on histopathological examinations from 2000 to 2011 at Chang Gung Memorial Hospital. Patients with nasopharyngeal or oropharyngeal involvement and those referred for residual or recurrent disease after treatment in other hospitals were excluded. Thirty-two patients with disease limited to the nasal cavity/paranasal sinuses were enrolled as the NKTL group. These cases had the most indistinguishable clinical and radiologic features compared to patients with CRS. 6 All of the patients underwent complete staging workup, including whole body enhanced CT scan, bone marrow examinations, and endoscopic examinations of the sinonasal and Waldeyer s ring areas. They were staged according to the Ann Arbor system. Of the 32 patients, 25 were male and seven were female, and their mean age was 50.3 6 18.1 years. All of the 173 patients who received functional endoscopic sinus surgery (C.-C.H.) for severe CRS with or without polyposis between May 2011 and April 2012 were recruited into the CRS group. The underlying causes of rhinosinusitis included odontogenic sinusitis (n 5 8), unilateral maxillary sinus mycetomas (n 5 8), antrochoanal polyps (n 5 6), post Caldwell-Luc cheek cysts (n 5 6), empty nose syndrome complicated by rhinosinusitis (n 5 1), and nonspecific sinusitis (n 5 144). Patients who underwent surgery for benign or malignant nasal tumors, such as inverted papilloma, were not recruited into this group. 2671

Fig. 2. Measurement of nasal floor and nasal septum thickness in chronic rhinosinusitis. (a) Sagittal nonenhanced computed tomography (CT) image of a 49-year-old male with bilateral chronic rhinosinusitis with nasal polyposis and a Lund-Mackey score of 23. (b) Coronal CT image at the level of the solid black line in (c). The maximal mucosal thickness was 1.68 mm at the septum (thickest point along the white line) and 0.96 mm at nasal floor (thickest point along the black line). (c) Sagittal nonenhanced CT image of a 54-year-old male with natural killer/t cell lymphoma. (d) Coronal CT image at the level of the solid black line in (c). The mucosal thickness was 3.75 mm at the septum (thickest point along the white line) and 2.24 mm at the nasal floor (thickest point along the black line). Of the 173 patients with CRS, 106 were male and 67 were female, and their mean age was 46.2 6 16.7 years. Forty-one cases were unilateral and 132 were bilateral. The mean Lund- Mackay score was 11.6 6 5.6. Nine patients had previously received septoplasty (5 males, 4 females; mean age, 45.1 6 12.0 years; mean Lund-Mackay score, 13.1 6 5.2) and 96 had pansinusitis on at least one side (66 males, 30 females; mean age, 47.0 6 15.6 years; mean Lund-Mackay score, 15.3 6 4.1). Fiftysix of the 173 patients underwent revision surgery for recurrent sinusitis after previous sinus surgeries, including functional endoscopic sinus surgeries or Caldwell-Luc operations (37 males, 19 females; mean age, 46.3 6 14.1 years; mean Lund- Mackay score, 12.4 6 5.7). Pathology The diagnosis of NKTL was verified by histologic examinations at the hospital s pathology department and confirmed by immunophenotypic studies. Imaging CT examinations were performed in all patients using one of two CT scanners (LightSpeed Plus 4, GE Healthcare, Milwaukee,WI, or Somatom Sensation 16; Siemens, Erlangen, Germany) by a standard CT protocol for the paranasal sinuses. Contiguous 3-mm scans of the paranasal sinuses were acquired in the axial plane from the upper margin of the frontal sinus to the hard palate. Then, the coronal images were reformatted at 2- to 3-mm increments. All of the images were reconstructed with a bone and soft-tissue algorithm. Patients in the NKTL group received contrast-enhanced CT examinations, whereas those in the CRS group received nonenhanced CT examinations. Nonetheless, the precontrast enhanced CT scans in the NKTL group were used for comparisons with the nonenhanced CT scans of the CRS group. Measurement of Nasal Floor and Nasal Septum The coronal section near the vertical part of the ground lamina was chosen. This section was also described as the transition from the anterior to the posterior ethmoids at which the superior turbinate began to appear (black solid line, Fig. 2a,c). The thickest point along the nasal floor and the thickest point along the nasal septum below the insertion of the inferior turbinate were the measured (Fig. 2b,d). The first priority was to measure the more severe side (right side, Fig. 3a). If the common meatus and inferior nasal meatus in the more severe side of the nose were opacified with soft tissue or fluid (right side, Fig. 3b), the thickest points in the contralateral side were measured (left side, Fig. 3b). If the thickest points could not be measured bilaterally, it was recorded as not measurable (Fig. 3c). If the more severe side was opacified and could not be measured in cases of unilateral CRS, this type was also recorded as not measurable. In cases of unilateral CRS, the contralateral side was normal and not measured. If prior septoplasty was too extensive resulting in the septum at the chosen coronal section having no bony structure to separate the left and right side, the thickest point of the septum was measured, divided by 2, and then recorded. Statistical Analysis Sex, pansinusitis, and prior operations were first used to check their effect on mucosal thickness within the CRS group using independent Student t test. Age and Lund-Mackay score were then used to check their relationship with mucosal 2672

Fig. 3. Measurement of nasal floor and nasal septum thickness in natural killer/t-cell lymphoma (NKTL). (a) Coronal nonenhanced computed tomography (CT) image of a 46-year-old male with NKTL. The maximal mucosal thickness was 2.42 mm at the septum (white arrow) and 3.40 mm (black arrowhead). (b) Coronal nonenhanced CT image of a 42-year-old male with NKTL. The maximal mucosal thickness was 3.65 mm at the septum (white arrow) and 3.54 mm (black arrowhead). (c) Coronal nonenhanced CT image of a 65-year-old male with NKTL. The maximal mucosal thickness was 3.65 mm at the septum (white arrow), but the maximal mucosal thickness of the nasal floor was not measurable because the bilateral inferior nasal meatuses were opacified by the tumor. thickness using a Pearson correlation coefficient. Mucosal thickness was compared between the NKTL group and CRS group by independent Student t test. Last, a receiver operating characteristic (ROC) curve was used to determine the optimal cutoff point between these two groups. All analyses were performed using the SPSS version 17.0 (SPSS Inc., Chicago, IL). RESULTS Of the 32 patients in the NKTL group, 31 and 30 patients had measurable nasal floor and septum, respectively. Of the 173 patients in the CRS group, 171 had measurable nasal floor and septum. Ensuring the determination of the thickest point of the nasal floor or nasal septum using this kind of measurement was stable, with very little variation within the CRS group. In the CRS group, sex had a significant influence on the nasal floor and nasal septum thickness (P 5.006 and.005, respectively) (Table I). Prior sinus surgery influenced septal thickness (P 5.006). However, age, Lund-Mackay score, pansinusitis, and prior septoplasty had no influence on the mucosal thickness of the nasal floor or nasal septum. That is to say, the mucosal thickness of these two sites was stable and not progressively thicker with age or severity of sinusitis. Only one of the 173 patients in the CRS group had a thickness >3 mm at the nasal floor, and only two cases had a thickness >3 mm at the nasal septum. Both of these cases were male with a history of sinus surgery, but not septoplasty. The mean thickness of the nasal floor in the NKTL group (3.21 mm) was thicker than that of the CRS group (0.98 mm) (P <.001). Moreover, the mean thickness of the nasal septum in the NKTL group (2.98 mm) was thicker than that in the CRS group (1.37 mm) (P <.001). Although patients with more severe sinusitis (pansinusitis and recurrent sinusitis groups) were chosen, the NKTL group still had significantly thicker nasal floor and nasal septum (P <.001) (Table II). Because there was thicker mucosa in the NKTL group, the next step was to determine the thickness by which the NKTL group could be distinguished from the CRS group. According to the ROC curve, the optimal cutoff points for the nasal floor and nasal septum thickness were 1.575 mm and 2.215 mm, respectively (Table III). If the cutoff points of the nasal floor and nasal septum thickness were set at 2.0 mm and 2.5 mm, respectively, the probability of being thicker than the corresponding points in the CRS group was <2% (1.2% and 1.8%, respectively). DISCUSSION The optimal point for measurement in a healthy subject or a patient with CRS should be thin with little variability. In addition, it should be frequently involved in cases of NKTL, and there should be an obvious measurable difference between NKTL and CRS. The mucosa of the nasal floor near the nasal vestibule is relatively thicker, with significant variability in thickness from person to person. There is a rhomboid thickening of the nasal septum near the nasal vestibule called the nasal septal body. 8,9 This is a widened area composed of septal cartilage located superior to the inferior turbinate and anterior to the middle turbinate. The septal cartilage is thicker there than in other parts of the nasal septum, and the mucosal cover of the septal body is thicker than in other portions of the nasal septum. As a result, the mucosa of the nasal floor and septum are thicker near the nasal vestibule. There is no bony floor at the nasal floor near the posterior nasal choana and the nasal floor here is directly connected to the soft palate. Thus, the measurement of nasal floor thickness at this location cannot be separated from the soft palate. Furthermore, the mucosa of the nasal septum near the posterior nasal choana may be thickened in healthy subjects or those with CRS. Mucosal thickness near the nasal vestibule and nasal septal body, or near the posterior nasal choana, varies more widely and is thicker. This is why the coronal section near the vertical part of the ground lamina has been chosen in this study, as this coronal section avoids the most variable areas (nasal vestibule and postnasal choana). Only sex and prior sinus surgery have an obvious influence on mucosal thickness at this coronal section. Age and Lund-Mackay score, pansinusitis, and prior septoplasty have no influence on the nasal floor or septum thickness. Mucosal thickness at these two sites is stable 2673

TABLE I. Comparisons Within the Chronic Rhinosinusitis Group. Nasal Floor Nasal Septum Size, mm P Value Size, mm P Value Sex, M/F 1.04/0.88 0.006* 1.46/1.22 0.005* Pansinusitis, 1/2 1.01/0.93 0.182 1.38/1.35 0.345 Status post-sinus 1.05/0.95 0.344 1.54/1.29 0.006* surgery, 1/2 Status postseptoplasty, 1/2 1.12/0.97 0.706 1.37/1.37 0.785 Pearson r P Value Pearson r P Value Age 0.039 0.617 20.002 0.980 Lund-Mackay score 0.119 0.122 20.044 0.569 *Sex influenced the nasal floor and septal thickness (P 5.006 and.005, respectively). Prior sinus surgery also influenced nasal septal thickness (P 5.006). Age, Lund-Mackay score, pansinusitis, and prior septoplasty had no influence on the mucosal thickness in these two areas. Status post-sinus surgery group meant that these patients received revised surgery for their recurrent sinusitis after previous sinus surgeries, including functional endoscopic sinus surgeries or Caldwell-Luc operations. This group also meant the recurrent sinusitis group. Status postseptoplasty meant that these patients had previously received septoplasty before coming to the department. F5 female; M 5 male. and does not progressively thicken with age or severity of sinusitis. It is therefore reasonable to use this coronal section for the measurements. In patients in Western countries, NKTL is more common in elderly men and is more commonly located in the paranasal sinuses. 10 However, the manifestations of this disease differ in patients in East Asia (including Taiwan), present at a younger age, and are usually in the nasal cavity. 10 The most common sites of involvement in the nasal cavity are the inferior turbinate (81.3%), nasal floor (62.5%), septum (56.3%), and middle turbinate (34.4%). 6 Although the inferior turbinate is most commonly involved, many benign diseases, such as chronic hypertrophic rhinitis, chronic sinusitis, and nasal polyposis, are associated with thickening of the inferior turbinate. Thus, inferior turbinate thickening is relatively nonspecific for NKTL. In this study, the optimal cutoff point by the ROC curve is 1.575 mm at the nasal floor and 2.215 mm at the nasal septum. However, a lower value of 1-specificity means a better cutoff point, and that fewer patients in the CRS group will have mucosa thicker than this cutoff point. In a previous study, the frequency of involvement by NKTL in the nasal cavity is 62.5% in the nasal floor and 56.3% in the septum. 6 Therefore, the sensitivity of the cutoff point for the nasal floor and the nasal septum is limited and reasonably around 60%. Furthermore, the optimal cutoff point should have the lowest 1-specificity but still have sensitivity close to 60%. If the cutoff points were defined >2.0 mm for the nasal floor or >2.5 mm for the nasal septum as a positive test, then sensitivity is 61.3% and 56.7%, respectively. This is similar to results of a previous study. 6 TABLE II. Comparison Between the and Chronic Rhinosinusitis Groups. Chronic Rhinosinusitis P Nasal floor thickness 31 3.21 6 2.20 171 0.98 6 0.39 <.001 Septum thickness 30 2.98 6 1.49 171 1.37 6 0.52 <.001 Pansinusitis Nasal floor thickness 31 3.21 6 2.20 94 1.01 6 0.40 <.001 Septum thickness 30 2.98 6 1.49 94 1.38 6 0.50 <.001 Status Post-Sinus Surgery Nasal floor thickness 31 3.21 6 2.20 54 1.05 6 0.49 <.001 Septum thickness 30 2.98 6 1.49 55 1.54 6 0.61 <.001 Status Postseptoplasty Nasal floor thickness 31 3.21 6 2.20 9 1.12 6 0.58.001 Septum thickness 30 2.98 6 1.49 9 1.37 6 0.43.002 No. 5 measurable number; NK/T 5 natural killer/t cell; SD 5 standard deviation. 2674

TABLE III. Best Cutoff Points to Differentiate Natural Killer/T-Cell Lymphoma From Chronic Rhinosinusitis. Nasal Floor Nasal Septum Sen. Spec. Sen. Spec. >1.5 mm 67.7% 91.2% 66.7% 67.3% >2.0 mm 61.3% 98.8% 63.3% 87.1% >2.5 mm 61.3% 99.4% 56.7 % 98.2% >3.0 mm 51.6% 99.4% 50.0% 98.8% Optimal cutoff* 67.7% 93.0% 63.3% 94.2% *The optimal cutoff points (by receiver operating characteristic curve) of nasal floor and nasal septum thickness were 1.575 mm and 2.215 mm, respectively. Sen. 5 sensitivity; Spec. 5 specificity. Furthermore, 1-specificity at these points was 1.2% and 1.8%, respectively, which means that the probabilities of the CRS group being thicker than these points are only 1.2% and 1.8%, respectively. A diagnosis other than CRS must be considered if the nasal floor thickness is >2.0 mm or the nasal septum thickness is >2.5 mm. Diseases such as NKTL, autoimmune granulomatous diseases, and atypical infections may involve thickening of the mucosa in these areas. 6 Study Limitations This method of measurement has limitations because not all of the patients with NKTL have nasal floor or nasal septal involvement. Thus, if the nasal floor is thinner than 2.0 mm and the nasal septum is thinner than 2.5 mm, the possibility of NKTL cannot be excluded. Although these cutoff points do not constitute a screening test for NKTL, they should raise the appropriate concern. NKTL has various presentations, and is difficult to be diagnosed by any single factor. So we used many steps to distinguish this disease from benign inflammatory disease. A more complete flowchart was established (Fig. 1b). Every step was tagged as a fraction. Because every step was regarded as an independent step, the denominator would be the total 32 cases of NKTL. Using this flowchart, there would be only one case of our group not able to meet the flowchart of NKTL. This patient had mucosal erosion, severe crusting, bleeding to touch only on his left inferior turbinate, and received biopsy, but his CT was negative in our every step. CONCLUSION NKTL has various presentations and is difficult to be diagnosed by a any single factor. Traditional radiological features of NKTL, such contrast enhancement, bone erosion, and adjacent soft tissue extension, must be noticed. If no prominent mass is noted, the presence of nasal cavity mucosal thickening without sinus wall thickening may indicate NKTL. 6 For patients with nasal cavity mucosal thickening with sinus wall thickening, a diagnosis other than CRS must be considered if the nasal floor thickness is >2.0 mm or the nasal septum thickness >2.5 mm. Figure 1b shows differentiating features. This might lead to earlier detection and treatment. BIBLIOGRAPHY 1. Au WY, Ma SY, Chim CS, et al. Clinico-pathologic features and treatment outcome of mature T-cell and natural killer-cell lymphomas diagnosed according to the World Health Organization classification scheme: a single center experience of 10 years. Ann Oncol 2005;16:206 214. 2. Armitage JO. A clinical evaluation of the International Lymphoma Study Group classification of non-hodgkin s lymphoma. Blood 1997;89:3909 3918. 3. Ou CH, Chen CC, Ling JC, et al. Nasal NK/T-cell lymphoma: computed tomography and magnetic resonance imaging findings. J Chin Med Assoc 2007;70:207 212. 4. Ooi GC, Chim CS, Liang R, et al. Nasal T-cell/natural killer cell lymphoma: CT and MR imaging features of a new clinico-pathologic entity. Am J Roentgenol 2000;174:1141 1145. 5. Kim J, Kim EY, Lee SK, et al. Extra-nodal nasal-type NK/T-cell lymphoma: computed tomography findings of head and neck involvement. Acta Radiol 2010;51:164 169. 6. Hung LY, Chang PH, Lee TJ, et al. Extra-nodal natural killer/t-cell lymphoma, nasal type: clinical and computed tomography findings in the head and neck region. Laryngoscope. 2010;122:2632 2639. 7. King AD, Lei KI, Ahuja AT, et al. MR imaging of nasal T-cell/natural killer cell lymphoma. Am J Roentgenol 2000;174:209 211. 8. Wexler D, Braverman I, Amar M. Histology of the nasal septal swell body (septal turbinate). Otolaryngol Head Neck Surg 2006;134:596 600. 9. Costa DJ, Sanford T, Janney C, Cooper M, Sindwani R. Radiographic and anatomic characterization of the nasal septal swell body. Arch Otolaryngol Head Neck Surg 2010;136:1107 1110. 10. Quraishi MS, Bessell EM. Non-Hodgkin s lymphoma of the sino-nasal tract. Laryngoscope 2000;110:1489 1492. 2675