Eur Radiol (2003) 13:1451–1460 DOI 10.1007/s00330-003-1825-8
H. B. Eggesbø S. Søvik S. Dølvik K. Eiklid F. Kolmannskog
Received: 24 July 2002 Revised: 3 December 2002 Accepted: 23 December 2002 Published online: 5 April 2003 © Springer-Verlag 2003 H. B. Eggesbø (✉) Department of Radiology, Aker University Hospital, 0514 Oslo, Norway e-mail:
[email protected] Fax: +47-2-3033046 S. Søvik Department of Physiology, Department Group of Basic Medical Sciences, University of Oslo, 0317 Oslo, Norway S. Dølvik Department of Otorhinolaryngology, Ullevål University Hospital, 0407 Oslo, Norway K. Eiklid Department of Medical Genetics, Ullevål University Hospital, 0407 Oslo, Norway F. Kolmannskog Sentrum Røntgeninstitutt, 0155 Oslo, Norway
HEAD AND NECK
Proposal of a CT scoring system of the paranasal sinuses in diagnosing cystic fibrosis
Abstract The purpose of this study was to develop a paranasal sinus CT scoring system that could be used as a diagnostic tool to discriminate cystic fibrosis (CF) patients from control patients examined for sinonasal disease. The model should include as few and easily applicable criteria as possible, supported by statistical analyses and clinical judgement. We used data from 116 CF and 136 control patients. The CF patients were grouped according to the number of confirmed CF mutations: genetically verified (CF-2), or based on sweat testing and clinical findings alone (CF-1, CF-0). Nine paranasal sinus CT criteria, including development, pneumatisation variants and inflammatory patterns, were evaluated. The final model included three criteria: (a) frontal and (b) sphenoid sinus development, and (c) absence of three pneumatisation variants. This model discriminated CF-2 from controls with overlap of summed scores in only 8 of 206 patients. When this
Introduction Cystic fibrosis (CF) is a life-shortening autosomal-recessive disease with an incidence in Caucasians of 1/2000–6000 [1]. The genetic defect is due to mutations in a gene coding for a chloride channel named the CF transmembrane conductance regulator (CFTR) on chromosome 7 [2]. A defect CFTR protein causes electrolyte transport disturbances resulting in thick mucus formation involving the lungs, paranasal sinuses, pancreas and intestine [3].
model was applied in the CF-1 and CF-0 groups, two populations seemed to exist. A larger group with summed scores overlapping that of the CF-2 group and a smaller group with summed scores overlapping that of the control group. We conclude that this CT scoring system may support, as well as exclude, a CF diagnosis in cases of diagnostic uncertainty. Keywords Cystic fibrosis · Paranasal sinuses · CT · Scoring
Lung infections and pancreas insufficiency are the most common clinical features reported in CF [4, 5]. Lung imaging with high-resolution CT (HRCT) usually demonstrates typical bronchiectasis, peribronchial wall thickening and mucus plugging with a preponderance for the right upper lobe [6, 7, 8]. The fatty replacement of the exocrine pancreatic gland cells, causing pancreas insufficiency, is easily visualised at CT or MR imaging. Paranasal sinus symptoms as well as paranasal sinus imaging have been a neglected part of CF; however, fol-
1452
lowing the simultaneous introduction of CT imaging and endoscopic paranasal sinus surgery in the early 1980s, several reports have been published on paranasal sinus CT findings in CF compared with the general population [9, 10, 11, 12, 13, 14, 15, 16]. These CT findings include small sinuses with absence of pneumatisation variants as well as typical inflammatory patterns, e.g. sinonasal polyposis and bulging of the lateral nasal wall (LNW). The pathogenesis of the poor development of the paranasal sinuses and absence of pneumatisation variants in CF is not exactly known; however, the high mucus viscosity that impairs normal mucociliary transport as well as inflammatory sinonasal disease may be one explanatory factor of arrest of paranasal sinus development in childhood. Stern et al. [17] reported that opacification of the paranasal sinuses on plain films was considered as the “Poor man’s sweat test”. In 1980 Ledesma-Medina et al. [18] stated that the paranasal sinuses almost always are affected in children with CF and claimed that “although opaque paranasal sinuses do not indicate the diagnosis of CF, clear paranasal sinuses exclude this disease with reasonable certainty”. In 1996 Nishioka et al. [16] proposed a radiological diagnostic triad in CF, which included frontal sinus aplasia, advanced ethmomaxillary opacification and medial bulging of the LNW; however, no systematic, diagnostic scoring system for paranasal sinus imaging in CF has been proposed thus far. Based on the diagnostic challenges and characteristic paranasal sinus CT findings in CF, we designed a prospective study in order to develop a diagnostic scoring system. The purpose of the study was to propose a scoring system that should discriminate an unselected population of CF patients from non-CF patients examined for inflammatory paranasal sinus disease. Furthermore, the scoring system should include as few and easily applicable criteria as possible. The selection of criteria should be supported by statistical analyses of the various possible models as well as by clinical judgement.
Materials and methods This prospective study comprised 116 CF patients registered at the Regional Centre of CF, undergoing routine clinical follow-up, and 136 consecutive non-CF patients examined for inflammatory paranasal sinus disease at an outpatient X-ray institute, hereafter named controls. Demographic data of all patients are displayed in Table 1. In the development of the scoring system only CF patients with genetically verified CF, i.e. two confirmed CF mutations (CF-2), were included in the disease group. This CF-2 group included 70 of 116 CF patients, after genotyping for 34 of the most common CF mutations (31 mutations using the OLA Cystic Fibrosis Assay Kit (PE Applied Biosystems) including DF508, and three other typical Nordic mutations 394del TT [19], 4005+2T-C [20], and R117C [21]). These 34 mutations constitute approximately 80–90% of CF mutations in Caucasians [1]. The 136 control patients were not genotyped.
Table 1 Demographic data
N F:M Age median Age range (years)
Controls
CF-2a
CF-1a
CF-0a
136 67:69 31 7–51
70 27:43 18 3–46
32 16:16 21 5–54
14 6:8 17 5–42
a Patients with cystic fibrosis (CF) were grouped according to the results of genotyping, i.e. CF-2: two confirmed mutations; CF-1: one confirmed mutation; CF-0: no confirmed mutations
The final scoring system was applied in the remaining CF population, where only one (CF-1, 32 of 116) or no (CF-0, 14 of 116) CF mutation was detected. All CF patients (CF-2, CF-1 and CF-0) had been diagnosed on the basis of sweat testing and clinical findings associated with CF [4, 5, 22]. The number of CF mutations confirmed does not indicate severity of the disease, but solely whether the diagnosis was genetically verified (CF-2) or based on sweat testing and clinical findings alone (CF-1, CF-0). Anterior functional endoscopic paranasal sinus surgery (FESS) had been performed prior to this CT examination in 26 of 70 CF-2 patients and 3 of 136 controls. The possible influence of FESS in these CF-2 patients was analysed in a previous report [10]. No statistical influence on sinus development or pneumatisation variants was found in surgically treated CF compared with non-surgically treated- CF patients (p<0.0001). Similarly, we compared influence of previous FESS on inflammatory patterns in the same CF population [9]. Bilateral medial bulging of the LNW was more prevalent in non-surgically treated (66%) than in surgically treated (19%) CF-2 patients (p<0.01); however, this was the only CT feature influenced by previous FESS. Based on the results from these two previous papers [9, 10], we chose not to group patients according to previous FESS in the development of our scoring system. CT procedures Coronal paranasal sinus scanning was performed with a Toshiba X-press spiral unit (Toshiba Medical, Nasu, Japan) with the patient in the prone position and the scan plane as perpendicular to the hard palate as possible. In CF patients, CT was performed with 5-mm contiguous slices from the glabella to the posterior part of the pneumatised sphenoid sinuses. Control patients were examined at an outpatient clinic with 3-mm contiguous slices through the anterior half of the paranasal sinuses and 5- or 6-mm contiguous slices through the posterior half. The voltage was set to 120 kV, whereas tube current settings were decreasing from 400 to 80 mAs during the examination period from 1994 to 1999. Hard copies were made with intermediate window width 2000 HU and level 200 HU. The lateral scanogram was included on the hard copies. Definitions of criteria The criteria were based on nine of the most common paranasal sinus CT findings in CF patients described in two previous studies [9, 10]. These criteria included development and pneumatisation variants, and inflammatory disease patterns characteristic in CF patients (Table 2), and were as follows: 1. Frontal sinus aplasia: absence of frontal bone pneumatisation with no ethmoid cells extending above a line tangential to the supraorbital margin (Fig. 1a) [23, 24]. 2. Frontal sinus aplasia or hypoplasia: aplasia as defined above. Regarding hypoplasia, oval-shaped sinus with the lateral mar-
1453
Fig. 1 a Coronal CT scan in a cystic fibrosis (CF) patient with two confirmed CF mutations (CF-2) demonstrates frontal sinus aplasia on the right side (pneumatisation below upper orbital rim) and frontal sinus hypoplasia on the left side. These findings are compatible with the presence of criterion 2 used in the final scoring system. On the right side, the high ethmoid recess should not be misinterpreted as a frontal sinus. b Sphenoid sinus hypoplasia in a CF-2 patient. To the left, the lateral scanogram demonstrates pneumatisation only of the presphenoid (arrows). To the right, coronal CT scan shows small sphenoid sinus with thickening of surrounding bone. These findings are compatible with the presence of criterion 4 used in the final scoring system. c Coronal CT scan shows agger nasi cells (arrows) equal to the most anterior ethmoid cells in a control patient. Agger nasi cells are commonly seen in controls but are very rare in CF-2 patients. d Coronal CT scan shows infraorbital (Haller) cells (asterisk) and pneumatisation of the middle turbinate (concha bullosa; arrow) in a control patient. Haller cells and concha bullosa are commonly seen in controls but were not seen in any CF-2 patient. Absence of agger nasi cells, Haller cells, and concha bullosa is compatible with criterion 6 used in the final scoring system gin medial to a vertical line drawn through the middle of the orbit, with smooth superior margin, and with absence of intrasinus septa (Fig. 1a) [10]. In children the frontal sinuses were evaluated and scored with respect to expected normal findings at corresponding age in control patients.
Table 2 Paranasal sinus criteria evaluated in order to develop a scoring system to help discriminate between CF from non-CF patients Criterion
Paranasal sinus variants
1 2 3 4
Frontal sinus aplasiaa Frontal sinus aplasia or hypoplasiaa Maxillary sinus hypoplasiaa Sphenoid sinus hypoplasia Pneumatisation variants
5
6
7 8 9 a Uni-
Absence of all of the following pneumatisation variants: agger nasi cells; infraorbital (Haller) cells; pneumatisation of the lamellar or bulbous part of the middle turbinate, nasal bone, crista galli, and anterior clinoid or pterygoid process of the sphenoid bone Absence of the following three pneumatisation variants: agger nasi cells; infraorbital (Haller) cells; or pneumatisation of the bulbous part of the middle turbinate Inflammatory patterns Advanced ethmomaxillary sinus diseasea Bulging of the lateral nasal walla Sphenoethmoid recess inflammatory pattern or bilateral presence recorded
1454
3. Maxillary sinus hypoplasia: defined according to a previous study [10] as four of the following five criteria being fulfilled: 1. Oval-shaped sinus. 2. Oval-shaped and enlarged orbit [25, 26]. 3. Inferior extension of the sinus above the nasal floor [26]. 4. Medial wall of the sinus lateral to a vertical line drawn tangentially to the medial orbital border [27]. In cases of medial bulging of the LNW, the origin of the inferior turbinate to LNW was used as the landmark. 5. Lateral extension medial to a vertical line drawn through the middle of the orbit at the level of the infundibulum, in the coronal plane [10]. In children the maxillary sinuses were evaluated and scored with respect to expected normal findings at corresponding age in control patients. 4. Sphenoid sinus hypoplasia: oval-shaped sinus with pneumatisation limited to the presphenoid, i.e. anterior to the vertical plane of the tuberculum sellae on the lateral scanograms [10, 28], and with absence of intra-sinus septa (Fig. 1b). The sphenoid sinuses were evaluated as a single entity, in order not to classify one side as hypoplastic when the other side was fully developed [11]. In children the sphenoid sinuses were evaluated and scored with respect to expected normal findings at corresponding age in control patients. 5. Absence of all of the following pneumatisation variants uni- or bilaterally: agger nasi cells (Fig. 1c); infraorbital (Haller) cells (Fig. 1d); and pneumatisation of the lamellar or bulbous part of the middle turbinate, nasal bone, crista galli, anterior clinoid or pterygoid process of the sphenoid bone. 6. Absence of the following three pneumatisation variants uni- or bilaterally: agger nasi cells (Fig. 1c); infraorbital (Haller) cells (Fig. 1d); or pneumatisation of the bulbous part of the middle turbinate (concha bullosa; Fig. 1d). 7. Advanced ethmomaxillary disease: ipsilateral advanced ethmoid and maxillary opacification (ostiomeatal complex inflammatory pattern or sinonasal polyposis pattern) [16, 29]. 8. Medial bulging of the lateral nasal wall (LNW). 9. Sphenoethmoid recess (SER) inflammatory pattern: the SER is located posterior to the superior turbinate and drains the sphenoid and posterior ethmoid sinuses. Obstruction of the SER may cause inflammatory disease in the sphenoid sinus alone or include also the posterior ethmoid sinuses [29]. Additionally, normal aeration or presence of opacification was noted in all patients. The CT images from all patients were evaluated in random order, and each CT criterion was scored at consensus by two radiologists, who were unaware of the results of genotyping. The criteria were reported to be absent or present unilateral or bilateral. Statistical methods Using simple logistic regression (Statview 5.0 for Macintosh, SAS Institute, Cary, N.C.) we assessed how well each of the nine CT criteria predicted whether a patient belonged to the disease group (CF-2) or the control group (non-CF). To evaluate the importance of unilateral vs bilateral presence of a criterion, criteria 1, 2, 3, 7 and 8 were analysed both as three-level variables (absence, uni- or bilateral presence) and as two-level variables (absence/unilateral presence vs bilateral presence). The Akaike Information Criterion (AIC) score [30] (α set to 2, S-PLUS 6.0 for Windows 2000, SAS Institute, Cary, N.C.) was then used to evaluate all possible combinations of criteria in order to find a model that discriminated well between the CF and the control group, while using as few criteria as possible. The models with lowest AIC scores predict the outcome best. Several models may result in very similar AIC score; therefore, clinical judgement concerning the number of criteria and their ease of use is necessary in the final decision of a model. For the selected model, the
regression coefficients (ß) from the simple logistic regression analyses were used to score the “weight” of each criterion; thus, the final, diagnostic model should consist of highly predictive criteria and also take into account their relative importance.
Results Development of the scoring system Importance of unilateral or bilateral presence of criteria For the five criteria that could be present bilaterally (criteria 1, 2, 3, 7 and 8), few CF-2 patients displayed unilateral presence; therefore, bilateral presence of a criterion gave significantly higher chance of belonging to the CF2 group than when the criterion was present on one side only (Table 3). An exception was maxillary sinus hypoplasia (criterion 3), where uni- and bilateral presence predicted CF-2 to a similar extent; therefore, in the further development of the scoring system we chose to define criteria that were reported bilaterally as two-level variables, i.e. present bilaterally or not (Table 4). Criteria best predicting cystic fibrosis Of the nine criteria included for evaluation, criteria 2, 4, 5 and 6 were present in most CF-2 patients but were rare in controls (Fig. 2). Criteria 1, 3, 7, 8 and 9 were present in approximately half of the CF-2 patients; however, when present, these five criteria strongly suggested CF due to the extremely rare occurrence in the control group (Fig. 2). Development of the frontal sinuses and pneumatisation variants were reflected in two criteria: 1 and 2, and 5 and 6, respectively. Criterion 1 included only frontal siTable 3 Importance of uni- or bilateral presence of the paranasal sinus criteria in 136 control and 70 CF-2 patients Criterion
Coefficient (β)a
Odds ratioa
95% CI
1
2.0 3.5 1.3 4.6 3.3 3.0 0.6 4.0 1.9 3.8
7.8 35.7 3.6 100.0 28.2 20.8 1.8 53.3 0.3 31.2
2.6 14.1 0.6 28.7 6.0 8.3 0.6 15.3 0.2 10.4
2 3 7 8
U B U B U B U B U B
– – – – – – – – – –
23.4 90.6 23.0 349.3 133.5 52.2 5.5 185.4 0.4 93.6
a Simple logistic regression coefficients (β) and odds ratios estimate the risk of belonging to the CF-2 group given U unilateral presence of the criterion vs no presence B bilateral presence of the criterion vs no presence CI confidence interval
1455
Table 5 The ten best Akaike information criterion (AIC) scores of all possible combinations (27) of the seven selected criteria (criteria 2–4, 6–9)
Fig. 2 Prevalence of each of the nine paranasal sinus criteria in 136 control and 70 CF-2 patients Table 4 The risk of belonging to the CF-2 group given presence of each of the nine paranasal sinus CT criteria Criterion
Coefficient (β)
Odds ratio
95% CI
1 2 3 4 5 6 7 8 9
3.3 4.3 2.7 9.1 4.6 3.6 3.9 3.4 3.5
26.1 71.2 14.6 9315.0a 95.6 38.0 49.7 31.2 33.3
10.6 25.6 6.0 573.7a 36.1 15.6 14.4 10.4 9.6
– – – – – – – – –
63.9 197.8 35.9 151235.5 253.1 92.7 171.2 93.6 114.7
For criteria that could exist bilaterally, coefficients (β) and odds ratios estimate the risk for belonging to the CF-2 group given bilateral presence of the criterion vs unilateral or no presence a Odds ratio and 95% CI values for criterion 4 may be erroneously high due to the extremely high predictive factor of this criterion
nus aplasia and did not predict CF as well as criterion 2, which comprised both aplasia and hypoplasia of the frontal sinuses (Fig. 2; Table 4). Criterion 5 included eight pneumatisation variants and predicted CF slightly better than criterion 6, which comprised only three variants (Fig. 2; Table 4); however, using criterion 5 requires comprehensive knowledge of the paranasal pneumatisation variants, whereas criterion 6 is easily applicable; hence, criteria 1 and 5 were excluded from further analyses. Combining criteria in a diagnostic model Sphenoid sinus hypoplasia was the single criterion that best predicted the presence of CF (Table 4). Only 2 pa-
Ten best scores
Combinations of criteria
AIC score
1 2 3 4 5 6 7 8 9 10
2+3+4+6+9 2+3+4+6+8 2+4+6+9 2+3+4+6 2+4+6 2+3+4+7+9 2+3+4+7+8 2+3+4+6+8+9 2+3+4+6+7+8 2+3+4+6+7+9
14.8 14.8 15.4 15.5 16.2 16.5 16.5 16.8 16.8 16.8
Table 6 Prevalence of the different combinations of the three criteria (2, 4, and 6) selected for the final model Combinations of criteria Criteria 2
Criteria 4
Criteria 6
0 0 0 0 1 1 1 1
0 0 1 1 0 0 1 1
0 1 0 1 0 1 0 1
CF-2 (n=70)
Controls (n=136)
0 0 1 4 0 1 6 58
93 21 1 0 16 5 0 0
tients – one CF and one control patient – were incorrectly classified by this criterion; however, to create a model not dependent on one single criterion the AIC algorithm was used on the seven criteria left. When criterion 4 was combined with other criteria, several combinations of criteria resulted in improved AIC scores (Table 5). None of the 10 best AIC combinations completely discriminated the CF-2 from the control group, but overlap between the two groups were minimal; therefore, we chose the model with the fewest criteria that also were easily applicable. This model included criterion 2 (bilateral frontal sinus aplasia/hypoplasia, criterion 4 (sphenoid sinus hypoplasia) and criterion 6 (absence of three common pneumatisation variants, i.e. agger nasi cells, Haller cells and concha bullosa). The prevalence of the different combinations of these criteria in CF-2 and control patients are listed in Table 6. The influence of each of these three criteria was taken into account by weighting the score of each criterion according to the coefficients (ß) from simple logistic regression analyses (Table 4). Criterion 4 had a coefficient value twice as large as did criteria 2 and 6; therefore, presence of criterion 4 was given a score of 2, whereas presence of criteria 2 and 6 were scored 1. The resulting summed scores for the 70 CF-2 and 136 control patients
1456
Fig. 3a, b The final model applied in 136 control and 70 CF-2 patients, and in 32 CF-1 and 14 CF-0 patients. Histograms show the patients’ summed scores of the three selected criteria (2: frontal sinus aplasia/hypoplasia; 4: sphenoid sinus hypoplasia; 6: absence of three common pneumatisation variants). Based on the relative magnitude of the regression coefficients of the criteria (ß), the criteria were scored as 1, 2 and 1, respectively, if present, and were otherwise 0
Fig. 4a, b Scoring sum in 136 control and 70 CF-2 patients. a Modified after Ledesma-Medina et al. [18]: normally aerated sinuses, criteria, 2, and 4. b Modified after Nishioka et al. [19]: criteria 1, 7, and 8
are displayed in Fig. 3. Most CF-2 patients had summed score 3 or 4 and most controls 0 or 1. The marginal overlap between the two groups resulted from 8 of 206 (4%) patients with a summed score of 2 (2 CF-2, ≥12 years; and 6 controls, 3 <12 years, 3 ≥12 years).
Evaluation of previous proposals
The scoring system applied on CF-1 and CF-0 patients The final scoring system was also used on the subset of the regional CF population that had only one or no confirmed CF mutations (CF-1, CF-0; Fig. 3). In these patients the diagnosis had been based on sweat tests and clinical findings associated with CF. The majority of CF1 and CF-0 patients achieved summed scores of 3 or 4; however, 6 of 32 CF-1 and 7 of 14 CF-0 patients had summed scores 0 or 1, thus overlapping the controls’ summed scores.
In 1980 Ledesma-Medina et al. [18] reported underdeveloped frontal and sphenoid sinuses in CF patients, and further claimed that normally aerated sinuses on plain films exclude the diagnosis of CF. To evaluate this statement, we scored our CF-2 and control patients 1 if sinonasal opacification was present, and otherwise the score was 0. A model consisting of the summed score of this criterion together with the criteria for frontal and sphenoid sinus in our study discriminated quite well between CF-2 and control patients (Fig. 4a); however, an overlap of 16 of 206 (8%) (10 controls and 6 CF-2 patients) was found at summed score 2. In 1996 Nishioka et al. [16] proposed a diagnostic triad based on CT imaging of patients referred for ear, nose and throat (ENT) examination. This triad, consisting of frontal sinus aplasia, advanced ethmomaxillary sinus opacification and medial bulging of the LNW, was equivalent to criteria 1, 7 and 8 in our study. A model with these three criteria showed that all controls had summed score 0 (90%) or 1 (10%; Fig. 4b); however, the
1457
CF-2 patients had summed scores dispersed over the entire scale, also overlapping the controls; hence, in our material these criteria only separated the most diseased CF-2 patients from controls.
Discussion Diagnosing CF may be challenging due to the large spectrum of severity of manifestations. The first comprehensive report on CF was made in 1938 [31]. For the next 20 years the diagnosis was based on clinical findings, e.g. lung infection and pancreas insufficiency. In 1959 the Gibson-Cooke sweat test [32] was developed and used as the gold standard for diagnosing CF for the next 30 years; however, major pitfalls concerning this test were reported, with both false-positive and false-negative results [33, 34, 35, 36]. In 1989 identification of the CF gene mutation gave hope that genotyping would solve all diagnostic problems in CF [37]; however, more than 1000 mutations in the CF gene have now been identified, and the genotype–phenotype correlation has been shown to be inconsistent. Consequently, the joint committee of the European CF society has proposed a diagnostic classification of CF on clinical findings rather than laboratory results for the revised upcoming International Classification of Diseases, ICD-11 (World Health Organisation). This proposal is in accordance with the consensus statement of the American Cystic Fibrosis Foundation Consensus Panel in 1998 [4]. Included in the phenotypic features consistent with a diagnosis of CF are “nasal polyps and radiographic or computed tomographic abnormalities of the paranasal sinuses”. The challenge in diagnosing CF presently is to be able to exclude, as well as to confirm, the diagnosis. The scoring system presented in this study may contribute to a more systematic use of paranasal sinus features in diagnosing CF. Patient selection To select a model predicting disease, it is of the outmost importance to ensure that the groups are correctly diagnosed as being diseased or not; therefore, only CF patients with two confirmed mutations and clinical findings associated with CF were included in the disease group. Genotyping of the controls in our study would be preferable, but this would not have been ethically acceptable; therefore, we could neither completely exclude non-diagnosed CF patients nor exclude carriers of the CF gene in the control group; however, none of our controls had been referred to the regional centre for CF when checked a minimum of 3 years after inclusion in the study. Two recent studies have suggested that a mutation in the CF gene may play a role in the pathogenesis of sinonasal disease [38, 39], whereas other studies have
found no correlation [40, 41]; therefore, since the vast majority of patients with sinonasal disease do not have a CF mutation, the possibility of inactivation of one CF gene in a few of our controls is probably of no consequence in this setting. Evaluation of the scoring system with a new unselected population of genetically confirmed CF patients would have strengthened our scoring proposal. However, due to the limited number of CF patients in our region, all CF-2 patients were used in the development of the scoring system; therefore, further evaluation of this model in new CF populations is warranted. Criteria selection Interestingly, criteria concerning paranasal sinus development and pneumatisation variants were sufficient to discriminate CF patients from non-CF patients. Inflammatory patterns characteristic of CF did not contribute further to the model and were therefore not included in the final scoring system. The reason may have been that the CF group showed lower prevalence of inflammatory pattern criteria than for criteria concerning development and pneumatisation variants (Fig. 2). This probably reflects that we studied an unselected CF population. If we had studied a group of CF patients referred for ENT examination, as in other studies [12, 13, 14, 16, 42], criteria concerning inflammatory patterns would possibly have been stronger predictors of disease. The strength of our scoring system is that it can be applied in any group of CF patients, regardless degree of sinonasal symptoms or previous FESS. The scoring system applied in CF-1 and CF-0 patients The distribution of summed scores in the CF-1 and CF-0 groups suggested that these two groups consisted of two populations: (a) a larger group with summed score overlapping the summed score of CF-2 patients; and (b) a smaller group with summed score overlapping the summed score of controls. These findings raise the question whether the latter group is erroneously diagnosed as having CF or present an atypical form of CF with milder clinical presentation. Severe and milder CF mutations have been reported. The milder mutations reflect a better CFTR function and determine the phenotype [43]. Additionally, secondary genetic factors called modifier genes [44] may influence the phenotype. If the CF-1 and CF-0 groups’ overlap with the controls’ summed scores resulted from milder mutations or modifier genes, one would expect similar findings in the CF-2 group. This was not the case. We suggest that the 13 of 116 (11%) CF-1 and CF-0 patients overlapping the controls’ summed scores of 0 or 1 may have been erroneously diagnosed as hav-
1458
ing CF. This magnitude of overdiagnosing is in accordance with previous reports [45, 46, 47, 48, 49]. In these patients clinical examination and sweat testing should be repeated, and alternative diagnosis should be considered. Influence of age using the scoring system Appropriate evaluation of the sinuses in the youngest patients requires precise knowledge of sinus development during childhood. Onset of frontal sinus pneumatisation is approximately at the age of 6 years [16, 50, 51, 52]; hence, aplasia or hypoplasia of the frontal sinuses in a child may be an age-dependent finding. In our study, the late onset of frontal sinus pneumatisation may have caused false high summed score in the youngest patients. In contrast, pneumatisation of the sphenoid and maxillary sinuses begin already during the first year of life [53, 54]. The sphenoid sinuses pneumatise into the basisphenoid before 6 years of age, and reach under the sella turcica at the age of 10 years [23]; thus, pneumatisation limited to the presphenoid after the age of 6 years may be an important finding when CF is considered. Pneumatisation variants, which may be explained as pneumatisation further into the surrounding paranasal bone, were not correlated to age but to degree of development of the other sinuses [10]. The prevalence of pneumatisation variants (criterion 6) in our control group was 80% in both children and adults (<12 vs ≥12 years), whereas none of the CF-2 patients had Haller cells or concha bullosa; hence, our scoring system may be used already from the age of 6–8 years. In patients younger than 6–8 years, presence of inflammatory patterns (criteria 7–9) may be of diagnostic value in addition to the scoring system, since these patterns are extremely rare in non-CF patients. The final scoring proposal was also analysed separately for the children and adults. CF-2 children had summed scores overlapping those of CF-2 adults and control children had summed scores overlapping those of control adults.
Radiation dose Radiation protection in patients undergoing paranasal sinus CT is very important [55, 56]. Recent reports dealing with low-dose CT scanning of the paranasal sinuses have especially focused on radiation protection of the lens [57, 58]. A new low-dose CT protocol for paranasal sinus imaging has recently been proposed [59]. This protocol, consisting of ten scans with 3- to 15-mm interslice gap and tube current as low as 40 mAs, offers significantly lower total and eye lens radiation dose than both standard CT and plain films. This protocol may be used for our scoring proposal of CF; however, due to the noncontiguous data acquisition using this protocol, one has to make sure that the anterior scans include the frontal sinuses and the most anterior part of the maxillary bone where the agger mound with agger nasi cells are localised. The contiguous slices through the ostiomeatal complex offer proper scanning of concha bullosa as well as infraorbital (Haller) cells. Concerning the sphenoid sinuses, one or two coronal scans together with the lateral scanogram are sufficient to diagnose hypoplasia. In the youngest children, a scan protocol with fewer than ten CT scans might be sufficient. We suggest that this low-dose CT protocol with limited coronal scans be used both for our diagnostic scoring system and for mapping of sinonasal disease and bony anatomy to evaluate the need for FESS. For follow-up examinations of sinonasal disease, MR imaging should be the first choice in CF patients [60, 61].
Conclusion The present study shows that a CT scoring system based on characteristic paranasal sinus developmental variations in CF patients can be used as a diagnostic tool in CF. A low score can exclude the diagnosis of CF, whereas a high score strengthens the possibility of CF. A close collaboration between an experienced ENT radiologist and the medical practitioner in care of the patient is recommended for optimal use of our scoring system. Acknowledgements We thank statistician L.M. Diep for invaluable help with the statistical analyses.
References 1. The Cystic Fibrosis Genetic Analysis Consortium (CFGAC) (2000). www.genet.sickkids.on.ca/cftr 2. Riordan JR, Rommens JM, Kerem B, Alon N, Rozmahel R, Grzelczak Z, Zielenski J, Lok S, Plavsic N, Chou JL et al. (1989) Identification of the cystic fibrosis gene: cloning and characterization of complementary DNA. Science 245:1066–1073
3. Smith JJ, Karp PH, Welsh MJ (1994) Defective fluid transport by cystic fibrosis airway epithelia. J Clin Invest 93:1307–1311 4. Rosenstein BJ, Cutting GR (1998) The diagnosis of cystic fibrosis: a consensus statement. Cystic Fibrosis Foundation Consensus Panel. J Pediatr 132:589–595
5. Stern RC (1997) The diagnosis of cystic fibrosis. N Engl J Med 336:487–491 6. Nathanson I, Conboy K, Murphy S, Afshani E, Kuhn JP (1991) Ultrafast computerized tomography of the chest in cystic fibrosis: a new scoring system. Pediatr Pulmonol 11:81–86
1459
7. Santis G, Hodson ME, Strickland B (1991) High resolution computed tomography in adult cystic fibrosis patients with mild lung disease. Clin Radiol 44:20–22 8. Cartier Y, Kavanagh PV, Johkoh T, Mason AC, Muller NL (1999) Bronchiectasis: accuracy of high-resolution CT in the differentiation of specific diseases. AJR 173:47–52 9. Eggesbø HB, Søvik S, Dølvik S, Kolmannskog F (2002) CT characterization of inflammatory paranasal sinus disease in cystic fibrosis. Acta Radiol 43:21–28 10. Eggesbø HB, Søvik S, Dølvik S, Eiklid K, Kolmannskog F (2001) CT characterization of developmental variations of the paranasal sinuses in cystic fibrosis. Acta Radiol 42:482–493 11. Eggesbø HB, Eken T, Eiklid K, Kolmannskog F (1999) Hypoplasia of the sphenoid sinuses as a diagnostic tool in cystic fibrosis. Acta Radiol 40:479–485 12. April MM, Zinreich SJ, Baroody FM, Naclerio RM (1993) Coronal CT scan abnormalities in children with chronic sinusitis. Laryngoscope 103:985–990 13. Brihaye P, Clement PA, Dab I, Desprechin B (1994) Pathological changes of the lateral nasal wall in patients with cystic fibrosis (mucoviscidosis). Int J Pediatr Otorhinolaryngol 28:141–147 14. Gentile VG,Isaacson G (1996) Patterns of sinusitis in cystic fibrosis. Laryngoscope 106:1005–1009 15. Kim HJ, Friedman EM, Sulek M, Duncan NO, McCluggage C (1997) Paranasal sinus development in chronic sinusitis, cystic fibrosis, and normal comparison population: a computerized tomography correlation study. Am J Rhinol 11:275–281 16. Nishioka GJ, Cook PR, McKinsey JP, Rodriguez FJ (1996) Paranasal sinus computed tomography scan findings in patients with cystic fibrosis. Otolaryngol Head Neck Surg 114:394–399 17. Stern RC, Boat TF, Orenstein DM, Wood RE, Matthews LW, Doershuk CF (1978) Treatment and prognosis of lobar and segmental atelectasis in cystic fibrosis. Am Rev Respir Dis 118:821–826 18. Ledesma-Medina J, Osman MZ, Girdany BR (1980) Abnormal paranasal sinuses in patients with cystic fibrosis of the pancreas. Radiological findings. Pediatr Radiol 9:61–64 19. Schwartz M, Anvret M, Claustres M, Eiken HG, Eiklid K, Schaedel C, Stolpe L, Tranebjaerg L (1994) 394delTT: a Nordic cystic fibrosis mutation. Hum Genet 93:157–161
20. Boman H (1996) Cystic Fibrosis Mutation Database Newsletter 69 21. Dork T, Mekus F, Schmidt K, Bosshammer J, Fislage R, Heuer T, Dziadek V, Neumann T, Kalin N, Wulbrand U (1994) Detection of more than 50 different CFTR mutations in a large group of German cystic fibrosis patients. Hum Genet 94:533–542 22. LeGrys VA (1996) Sweat testing for the diagnosis of cystic fibrosis: practical considerations. J Pediatr 129:892–897 23. Libersa C, Laude M, Libersa J-C (1981) The pneumatization of the accessory cavities of the nasal fossae during growth. Anat Clin 2:265–273 24. Weiglein A, Anderhuber W, Wolf G (1992) Radiologic anatomy of the paranasal sinuses in the child. Surg Radiol Anat 14:335–339 25. Bierny JP,Dryden R (1977) Orbital enlargement secondary to paranasal sinus hypoplasia. AJR 128:850–852 26. Bassiouny A, Newlands WJ, Ali H, Zaki Y (1982) Maxillary sinus hypoplasia and superior orbital fissure asymmetry. Laryngoscope 92:441–448 27. Meyers RM, Valvassori G (1998) Interpretation of anatomic variations of computed tomography scans of the sinuses: a surgeon’s perspective. Laryngoscope 108:422–425 28. Fujioka M, Young LW (1978) The sphenoidal sinuses: radiographic patterns of normal development and abnormal findings in infants and children. Radiology 129:133 29. Sonkens JW, Harnsberger HR, Blanch GM, Babbel RW, Hunt S (1991) The impact of screening sinus CT on the planning of functional endoscopic sinus surgery. Otolaryngol Head Neck Surg 105:802–813 30. Collett D (1994) Strategy for model selection. In: Modelling survival data in medical research. Chapman and Hall, London, p 78 31. Andersen DH (1938) Cystic fibrosis of the pancreas and its relation to celiac disease: a clinical and pathologic study. Am J Dis Child 56:344–399 32. Gibson LE, Cooke RE (1959) A test for concentration of electrolytes in sweat in cystis fibrosis of the pancreas utilizing pilocarpine iontophoresis. Pediatrics 23:545–549 33. LeGrys VA (2001) Assessment of sweat-testing practices for the diagnosis of cystic fibrosis. Arch Pathol Lab Med 125:1420–1424 34. Rosenstein BJ, Langbaum TS, Gordes E, Brusilow SW (1978) Cystic fibrosis. Problems encountered with sweat testing. J Am Med Assoc 240:1987–1988 35. Green A, Dodds P, Pennock C (1985) A study of sweat sodium and chloride: criteria for the diagnosis of cystic fibrosis. Ann Clin Biochem 22:171–174
36. Columbo JL, Gibbs GE (1979) Cystic fibrosis and erroneous sweat tests. Nebr Med J 64:164–165 37. Kerem B, Rommens JM, Buchanan JA, Markiewicz D, Cox TK, Chakravarti A, Buchwald M, Tsui LC (1989) Identification of the cystic fibrosis gene: genetic analysis. Science 245:1073–1080 38. Raman V, Clary R, Siegrist KL, Zehnbauer B, Chatila TA (2002) Increased prevalence of mutations in the cystic fibrosis transmembrane conductance regulator in children with chronic rhinosinusitis. Pediatrics 109: E13 39. Wang X, Moylan B, Leopold DA, Kim J, Rubenstein RC, Togias A, Proud D, Zeitlin PL, Cutting GR (2000) Mutation in the gene responsible for cystic fibrosis and predisposition to chronic rhinosinusitis in the general population. J Am Med Assoc 284:1814–1819 40. Hytonen M, Patjas M, Vento SI, Kauppi P, Malmberg H, Ylikoski J, Kere J (2001) Cystic fibrosis gene mutations ∆F508 and 394delTT in patients with chronic sinusitis in Finland. Acta Laryngol 121:945–947 41. Irving RM, McMahon R, Clark R, Jones NS (1997) Cystic fibrosis transmembrane conductance regulator gene mutations in severe nasal polyposis. Clin Otolaryngol 22:519–521 42. Brihaye P, Jorissen M, Clement PA (1997) Chronic rhinosinusitis in cystic fibrosis (mucoviscidosis). Acta Otorhinolaryngol Belg 51:323–337 43. Kerem E, Kerem B (1995) The relationship between genotype and phenotype in cystic fibrosis. Curr Opin Pulm Med 1:450–456 44. Dipple KM, McCabe ER (2000) Modifier genes convert “simple” Mendelian disorders to complex traits. Mol Genet Metab 71:43–50 45. Rosenstein BJ, Langbaum TS (1987) Misdiagnosis of cystic fibrosis. Need for continuing follow-up and reevaluation. Clin Pediatr 26:78–82 46. Smalley CA, Addy DP, Anderson CM (1978) Does that child really have cystic fibrosis? Lancet 2:415–417 47. David TJ, Phillips BM (1982) Overdiagnosis of cystic fibrosis. Lancet 2:1204–1206 48. Shaw NJ, Littlewood JM (1987) Misdiagnosis of cystic fibrosis. Arch Dis Child 62:1271–1273 49. Barbero GJ (1995) The undoing of a diagnosis. The effect of a misdiagnosis of a disease. Arch Pediatr Adolesc Med 149:1341–1344
1460
50. Schatz CJ,Becker TS (1984) Normal CT anatomy of the paranasal sinuses. Radiol Clin North Am 22:107–118 51. Anderhuber W, Weiglein A, Wolf G (1992) Nasal cavities and paranasal sinuses in newborns and children. Acta Anat (Basel) 144:120–126 52. Davis WE, Templer J, Parsons DS (1996) Anatomy of the paranasal sinuses. Otolaryngol Clin North Am 29:57–74 53. Aoki S, Dillon WP, Barkovich AJ, Norman D (1989) Marrow conversion before pneumatization of the sphenoid sinus: assessment with MR imaging. Radiology 172:373–375
54. Spaeth J, Krugelstein U, Schlondorff G (1997) The paranasal sinuses in CT imaging: development from birth to age 25. Int J Pediatr Otorhinolaryngol 39:25–40 55. Duvoisin B, Landry M, Chapuis L, Krayenbuhl M, Schnyder P (1991) Low-dose CT and inflammatory disease of the paranasal sinuses. Neuroradiology 33:403–406 56. Marmolya G, Wiesen EJ, Yagan R, Haria CD, Shah AC (1991) Paranasal sinuses: low-dose CT. Radiology 181:689–691 57. Hein E, Rogalla P, Klingebiel R, Hamm B (2002) Low-dose CT of the paranasal sinuses with eye lens protection: effect on image quality and radiation dose. Eur Radiol 12:1693–1696 58. Czechowski J, Janeczek J, Kelly G, Johansen J (2001) Radiation dose to the lens in sequential and spiral CT of the facial bones and sinuses. Eur Radiol 11:711–713
59. Hagtvedt T, Aaløkken J, Nøtthellen J, Kolbenstvedt A (2002) A new lowdose CT examination compared with standard-dose CT in the diagnosis of acute sinusitis. Eur Radiol (in press). DOI 10.1007/s00330-002-1652-3 60. Eggesbø HB, Ringertz S, Haanaes OC, Dølvik S, Erichsen A, Stiris M, Kolmannskog F (1999) CT and MR imaging of the paranasal sinuses in cystic fibrosis. Correlation with microbiological and histopathological results. Acta Radiol 40:154–162 61. Eggesbø HB, Stiris M, Dølvik S, Søvik S, Kolmannskog F (2001) Complementary role of MR imaging of ethmomaxillary sinus disease depicted at CT in cystic fibrosis. Acta Radiol 42:144–150