A comparative study of three-dimensional cone beam computed tomographic sialography and ultrasonography in the detection of non-tumoral salivary duct diseases

Abstract

Objectives:

To compare the overall diagnostic outcomes of 3D-CBCT sialography and ultrasonography (US) in the detection of sialolithiasis, ductal dilatation, and ductal stenosis.

Methods:

This retrospective monocentric study compared the two imaging modalities carried out in the same patients referred for salivary symptoms of the parotid and submandibular glands. The primary endpoint was the capacity of the imaging procedure to diagnose a lesion. The secondary objectives were the detection rates according to the type of lesion, analysis of the causes of failure, and the parameters of radiation exposure and safety (for 3D-CBCT sialography).

Results:

Of the 236 patients who received a 3D-CBCT sialography in our institution, 157 were ultimately included in the per-protocol analysis. 3D-CBCT sialography allowed detection of ductal lesions in 113 patients versus 86 with US. The two imaging modalities yielded congruent interpretations in 104 out of 157 subjects (66.2%). Higher sensitivity and negative predictive value were observed with 3D-CBCT sialography compared with US, irrespective of the lesions studied: 0.85 vs 0.65 and 0.70 vs 0.44, respectively. Regarding the sialolithiasis, both 3D-CBCT sialography and US allowed identification of lesions with high sensitivity and negative predictive value (0.80 vs 0.75 and 0.88 vs 0.78, respectively).

Conclusions:

US remains the first-line examination for exploration of the salivary lesions. 3D-CBCT sialography is an alternative in case of inconclusive US, and prior to any endoscopic procedure.

Introduction

Non-tumor salivary gland diseases are the most common disorders of salivary glands, affecting nearly 1% of the general population. ^1–3 These pathologies include salivary stones, chronic inflammation, and anatomical abnormalities; while stenosis and ductal dilatation are most often secondary lesions and not diseases of the salivary system. ^4,5 Clinically, these pathologies cause a decrease in salivary flow that is responsible for obstructive symptoms, including salivary pain and swelling when eating. The main cause of salivary gland obstruction is sialolithiasis in 60–70% of cases, followed by stenosis and ductal inflammation in 15–25% and 5–10% of cases, respectively. ^4,6 There are many imaging techniques available to diagnose salivary ductal lesions, such as two-dimensional (2D) and three-dimensional (3D) plain sialography, ultrasonography (US), CT scan, MRI, and scintigraphy. ^2,7,8 In addition to diagnosis, salivary imaging provides information regarding the size, location, number, and morphology of the lesions. These criteria are part of the treatment algorithms for salivary calculi, and they determine the type of removal technique to be used. ^9,10 The same applies to salivary stenosis pathologies, for which the location, type, and length of the lesions are important parameters. ¹¹

By revealing the filled ductal tree, sialography represents the gold standard technique for the diagnosis of non-tumor salivary gland diseases. ^1,12,13 Conventional 2D-plain radiography is gradually being replaced by 3D-sialography, which provides a more accurate examination. ^1,14 3D-Cone Beam CT (3D-CBCT) sialography has been reported to provide 3D images of the ductal system with high spatial resolution. ^5,14–17 Various studies have shown its superiority to conventional sialography in exploring the ductal tree up to the fifth or the sixth division. ^1,17 Nevertheless, this technique requires salivary catheterization, which entails a risk of failure, the injection of an iodinated contrast product, and it entails exposure to significant irradiation. US is a first-line examination for the diagnosis of ductal and parenchymal pathologies owing to its accessibility, low cost, and the absence of radiation exposure. However, US lacks the capacity to identify the entities behind the bony structures, it does not provide a representation of the salivary ducts, and it often fails to diagnose stones less than 2 mm in size and chronic ductal inflammation. ^18–21

No objective study to date has compared ultrasonography with 3D-CBCT sialography. It is not clear which of these two procedures is most effective for first-line diagnosis of non-tumor salivary gland diseases. The purpose of this study was to compare the overall diagnostic outcomes of 3D-CBCT sialography and ultrasonography in the detection of sialolithiasis, ductal dilatation, and ductal stenosis, and the outcome measures for each of the three lesions. The null hypothesis stated that there would be no statistically significant differences between the two techniques in terms of overall or disease-specific outcomes. In addition, 3D-CBCT was evaluated for causes of failure of sialography, adverse clinical effects, and patient exposure to radiation.

Methods and materials

This monocentric, retrospective study compared both imaging modalities (US and 3D-CBCT sialography) carried out in the same patients. All 236 patients who received 3D-CBCT sialography at our institution between January 2017 and December 2021 were included in this study. These patients were seen for the first time or referred to our department for salivary obstructive symptoms, salivary inflammation (acute or chronic), or salivary secretion disorders (mainly dry mouth) affecting the parotid and/or submandibular glands. Seventy-nine patients were secondarily excluded from the study based on exclusion criteria: 30 patients were excluded for a sialography failure, while 49 patients did not undergo US imaging. Our per-protocol analysis, therefore, included 157 patients (mean age 50.4 ± 16.6 years, female/male ratio 93/64). For each patient, the data regarding the symptoms, the affected gland, the side, and the progression of the symptoms were collected. The results of the ultrasound examinations were also collected in the medical records and analyzed with the available data. No double reading of the images was performed, and we considered the first interpretation of the examination. These ultrasound examinations were performed in different centers, using different devices, by multiple radiologists with different range of expertise in salivary pathologies. Most of the examinations followed a semi-standardized evaluation by analysis of the entire gland from the main duct to the glandular parenchyma. The salivary stones presented as a hyperechogenic image with a posterior dark cone or were suspected based on ductal dilatation. Stenosis and dilatations could also be discerned. Due to the retrospective nature of this study, it was granted a written exemption from ethics committee approval, as per French legislation article L. 1121–1 paragraph one and R1121-2 of the Public Health Code.

Execution of 3D-CBCT sialography

The technique for the 3D-CBCT sialography has been previously described. ¹⁶ All the procedures and the interpretation were performed by the same examiner, who is an expert in salivary pathology and imaging (H.B.). Briefly, after locating the salivary duct ostium, 0.5–1.0 ml of high-concentration water-soluble iodinated contrast product (XENETIX 300^®, 300 mg ml⁻¹; Guerbet, France) was injected under standard aseptic conditions using a lachrymal cannula (25G Moria^® L12 mm; MORIA Inc., Doylestown, PA). The iodine contrast was maintained in the gland by placing a microsurgical clamp (Biover TKM2, Hergiswil, Switzerland) on the ostium of the Wharton duct or with a plastic straight Halstead clamp placed on the ostium of the Stensen duct. Image acquisition was performed immediately after injection, using a wide-field CBCT device (NewTom VGi, QR, Verona, Italy). Front and lateral scout views were performed first. Special exposure parameters were used to limit the radiation exposure: a reduced field of view to 75 × 120 mm focused on the symptomatic gland and limited 110 kV. The tube current was adjusted automatically by the machine, while the exposure time was selected by the user (regular zoom mode). For each patient, we collected the dose-area product (DAP) provided by the CBCT device. Early adverse effects of cannulation (bleeding, pain, canal perforation, or iodine allergy) were investigated.

Image analysis (for 3D-CBCT sialography) and comparison of imaging modalities

The imaging data were archived using imaging software (CARESTREAM View PACS v. 11.3; Carestream Health, Inc., Rochester, NY, USA). Analysis was performed using maximal intensity projection and multiplanar reconstruction with 0.25 mm cuts, isotropic voxel size and image pixels 492 × 492. The interpretation followed a standardized pathway, it was performed by the physician who performed the examination and was, therefore, not blinded to the patient’s symptoms. The analysis aimed to identify a lesion in the salivary duct tree (primary endpoint). Sialolithiasis appeared as a subtraction image within a hyperintense salivary duct on the 3D-CBCT sialography. Dilatation was diagnosed as sialectasis in the ductal salivary system, while ductal stenoses appeared as a signal defect of the salivary ducts. Secondary endpoints were the type of lesion observed, their exact location (anterior third, middle third, or posterior third of the main salivary duct), and the number and size of the sialolithiasis.

The cone beam data were collected and compared with the ultrasound data by a single reviewer (V.C.) who was independent of both procedures and not involved in patient management. The two modalities were compared in terms of visualization of normal structures and identification of a salivary ductal lesion (primary outcome). In the absence of a third reference examination, each imaging procedure was compared to the other based on the assumption that a lesion detected by one confirmed the presence of the disease. The secondary objectives were the detection rates according to the type of lesion (lithiasis, stenosis, or dilatation), analysis of the causes of failure for the 3D-CBCT sialography, as well as the collection of the DAP and the search for adverse effects (3D-CBCT sialography).

Statistical analysis

The statistical analysis was performed using GraphPad Prism 9.0 software for Mac (GraphPad Software, La Jolla, CA, USA). Descriptive statistics were performed for the clinical data. McNemar’s χ² test was used to determine differences between the two imaging modalities with regard to the same outcomes. Because there is no gold standard (i.e., histopathological confirmation or sialoendoscopy), the terms sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) reflect the performance of one imaging procedure versus the other and could not be compared to other procedures independently of this study. ¹ A p-value of less than 0.05 was taken to indicate statistical significance.

Results

Characteristics of the study population

The submandibular gland was mostly affected. Regarding the symptoms, obstructive signs were the most frequent, followed by infection and gland swelling. The patient characteristics are presented in Table 1.

Table 1.

Epidemiologic and clinical characteristics of the patients

Patient epidemiological and clinical characteristics
Age (in years): mean, ± S.D.	50.4 ± 16.6
Gender: Females/Males, n (%)	93 (59.3%)/64 (40.7%)
Affected gland:
Submandibular gland, n (%)  Parotid gland, n (%)  Multifocal, n (%)	97 (61.8%) 56 (35.7%) 4 (2.5%)
Affected side:
Left, n (%)  Right, n (%)  Both, n (%)	80 (50.9%) 65 (41.4%) 12 (7.7%)
Symptoms:
Hernia and/or colic  Infection  Swelling  Non-typical pain  Asymptomatic	96 (61.1%) 32 (20.4%) 18 (11.5%) 8 (5.1%) 3 (1.9%)
Progression of symptoms:
Less than 14 days  2–12 weeks  More than 3 months	7 (4.5%) 47 (29.9%) 103 (65.6%)

SD, standard deviation; n, number of patients.

Comparison of imaging procedures

The two imaging modalities yielded congruent interpretations in 104 out of 157 subjects (66.2%). Regarding the primary endpoint, 3D-CBCT sialography detected a salivary ductal lesion in 113 patients, while US detected a salivary lesion in 86 patients. Analysis of the discordant data revealed that 13 and 40 lesions were detected by only US or 3D-CBCT sialography, respectively. Irrespective of the type of lesion analyzed, the sensitivity and the NPV were higher for 3D-CBCT sialography than for ultrasound, while the specificity and the PPV were higher with US than for 3D-CBCT sialography (Table 2).

Table 2.

Comparison of both imaging modalities regarding the primary and the secondary outcomes

	3D-CBCT sialography	Ultrasonography	Both modalities
Identification of a lesion (primary outcome)
Normal examination, n (%) Abnormal examination, n (%) Sensitivity, mean [CI 95%] Specificity, mean [CI 95%] Negative predictive value, mean [CI 95%] Positive predictive value, mean [CI 95%]	44/157 (28.0) 113/157 (72.0) 0.85 [0.76–0.91] 0.44 [0.33–0.55] 0.70 [0.56–0.82] 0.65 [0.55–0.73]	71/157 (45.2) 86/157 (54.8) 0.65 [0.55–0.73] 0.70 [0.56–0.82] 0.44 [0.33–0.55] 0.85 [0.76–0.91]	31/157 (19.7) 73/157 (46.5)
Type of lesion (secondary outcome)
Lithiasis, n (%) Dilatation, n (%) Stenosis, n (%)	71/88 (80.7) 27/35 (77.1) 19/19 (100.0)	70/88 (79.6) 20/35 (57.1) 0/19 (0)	54/88 (62.5) 12/35 (34.3) 19/19 (100.0)
Characteristics of the lithiasis
Number: single / multiple, n (%) Radiopaque / radiolucent, n (%) Location, n (%): ‍  Anterior third of main canal  Middle third of main canal  Posterior third or IG  Not specified Sensitivity, mean [CI 95%] Specificity, mean [CI 95%] Negative predictive value, mean [CI 95%] Positive predictive value, mean [CI 95%]	52 (73.2) / 19 (26.8) 60 (84.5) / 11 (15.5) 14 (19.7) 9 (12.7) 48 (67.6) 0 (0) 0.80 [0.69–0.88] 0.82 [0.67–0.91] 0.88 [0.78–0.94] 0.70 [0.56–0.82]	56 (80.0) / 14 (20.0) – 3 (4.3) 2 (2.9) 33 (47.1) 32 (45.7) 0.75 [0.63–0.83] 0.67 [0.53–0.79] 0.78 [0.67–0.86] 0.63 [0.49–0.75]
Characteristics of the dilatations:
Location, n (%):  Anterior third of main canal  Middle third of main canal  Posterior third or IG  Not specified	1 (3.7) 13 (48.1) 6 (22.2) 7 (25.9)	0 (0) 6 (30.0) 1 (5.0) 13 (65.0)
Characteristics of the dilatations:
Location, n (%):  Anterior third of main canal  Middle third of main canal  Posterior third or IG  Not specified	– – – –	4 (21.0) 12 (63.2) 3 (15.8) 0 (0)

CI, confidence interval; IG, intraglandular; n, number of patients.

Most of the lesions found were sialolithiasis (Figure 1). 3D-CBCT sialography and US allowed detection of 71 and 70 lithiasis, respectively. The mean size of lithiasis was 7.2 ± 3.1 mm. Most of the calculus were in a proximal location. In 17 patients, 3D-CBCT sialography failed to diagnose a salivary stone, although it was detected by ultrasound: 12 cases were normal procedures, 4 patients presented an isolated ductal dilatation, and one patient presented a ductal stricture based on sialography. Most of these lesions concerned distal localizations (i.e., intraglandular stones). US failed to identify an area of calculus in 18 patients, while a stone was detected by sialography: there was no lesion identified in 15 cases and 3 cases were classified as dilatations by US. The mean size of these stones was 6.3 ± 4.6 mm, distributed in the posterior canal for nine calculi, the anterior canal for seven calculi, while two lithiasis were located in the middle third of the main duct. 3D-CBCT sialography outperformed US for visualization of sialolithiasis, as evidenced by a high sensitivity, specificity, NPV, and PPV (Table 2).

Figure 1.

Visualization of a salivary stone located in the distal third of Wharton’s duct, seen by 3D-CBCT sialography and ultrasonography. Calculus (white arrow) appears as a dense image within the contrast-filled canal on axial (a) and sagittal (b) sections in 3D-CBCT sialography Ultrasound exploration and measurement (c) of the submandibular gland (*), and detection of the stone (white arrow) as a hyperechogenic mass within the canal with posterior shadow cone (d).

Dilatation was diagnosed in 35 cases (24.6%) regardless of the procedure used (Figure 2). 3D-CBCT sialography failed to detect a dilatation in 7 patients (three were diagnosed as stenosis, three as sialolithiasis, and no lesion was found in 2 cases). US was failing in 15 cases (no lesion found in 11 cases, sialolithiasis were evidenced in 4 cases).

Figure 2.

Diagnosis of salivary ductal stenosis (black arrow) and dilatations (white arrows) of Stensen’s duct and the intraparotid ductal branches by 3D-CBCT sialography (maximal intensity projection) and ultrasonography.

Finally, 19 stenoses were diagnosed with 3D-CBCT sialography (13.4% of lesions), but not with US (Table 2).

Secondary endpoints

For the 236 patients receiving 3D-CBCT sialography, catheterization failure was observed in 30 patients (12.8%). These failures were mostly due to difficulty with cannulation in 26 patients, while 3 patients presented obstructive ostial lithiasis with no possibility to canulate the salivary duct. Another patient had opacification of the sublingual gland, with no visualization of the submandibular gland. No adverse effects of the catheterization (bleeding, perforation) or the iodine contrast agent (allergy) were observed or reported by the patients.

Regarding the radiation exposure, the mean DAP generated by CBCT was 235.5 ± 81.06 mGy.cm² (ranging from 125.9 to 472 mGy.cm²) and the mean tube current-exposure time product was 6.82 ± 3.94 mAs (ranging from 2 to 20 mAs).

Discussion

Many imaging modalities have been described for diagnosing salivary ductal pathologies. Irrespective of the technique used, the objectives of imaging are diagnosis of the lesion and determination of its size and its location in the salivary ductal tree. Indeed, these parameters determine the therapeutic strategy of the main salivary pathologies. In our current practice, 3D-CBCT sialography is the reference examination that can meet all these objectives. In addition, salivary catheterization provides information regarding the predicted ease or difficulty of penetration of the salivary ostium in salivary endoscopy. The size of the salivary ducts once dilated is also an important element to consider in sialoendoscopy. In the literature, 3D-CBCT sialography has been shown to be superior to conventional 2D sialography. ^1,5,15 Lately, this modality has particularly been described for the diagnosis and follow-up of chronic salivary pathologies such as Sjögren’s syndrome. ^22–24 No study to date has compared 3D-CBCT sialography versus ultrasonography for the primary diagnosis of salivary ductal lesions. Our study is, therefore, the first to compare the two imaging procedures and aims to define the suitability of both imaging modalities in the diagnosis strategy for non-tumor salivary pathologies. Our series of 236 patients who received 3D-CBCT sialography is the largest in the literature. The study population is comparable to others in the literature in terms of the mean age and female/male distribution. ^1,14 The submandibular gland was the most affected in 61.8% of cases, reflecting a more frequent lithiasis pathology in this location. In their study of 2,277 gland pathologies, Goncalves et al found submandibular involvement in 57.4% of patients. ²¹ In our study, salivary stones were responsible for the obstructive symptoms in 62% of cases. In the literature, benign salivary gland obstruction appears to be caused by sialoliths in 60–75% of cases and stenosis of the salivary duct in 15–25% of cases. ^4,6,25,26 Interestingly, more than half of our patients had salivary obstructive symptoms that had lasted for more than 3 months, and some for several years. This is due to the patient’s tolerance of the hernia episodes, which are most often resolved at the end of meals. Consultation of a specialist is most often prompted by painful episodes of salivary colic or infectious sialadenitis.

In light of the high sensitivity (0.85) and NPV (0.70), our results revealed that 3D-CBCT sialography outperformed US for the diagnosis of main salivary ductal lesions. In particular, sialography was superior to ultrasonography for visualization of dilatations and ductal stenosis. Nevertheless, stenoses are uncommon pathologies that are not readily observable with ultrasonography as it does not use the spontaneous contrast of saliva. Indeed, the salivary ducts can only be discerned when they are filled, which only occurs in cases of an obstructive salivary stone or advanced stenoses. Some authors recommend administration of vitamin C or lemon juice to stimulate the secretion and filling of the salivary ducts during US exam. ^27,28 In 14 patients with salivary symptoms and failure of a US exam, Kroll et al showed that seven patients exhibited salivary ductal stenoses located primarily in the main efferent duct or close to the papilla. ¹⁴ Our results regarding dilatation must take into account the artificial ductal expansion created by the injection of the contrast medium in 3D-CBCT sialography, while ultrasound detects spontaneous dilation of the salivary ducts. In addition, with sialography, there is no reference value to qualify as ductal dilatation. Sialoliths were the most represented lesions of the salivary ducts in our series. There did not appear to be any difference between the two imaging modalities for the diagnosis of lithiasis. The size and location of the lithiasis are comparable to what has been reported in the literature. ^1,13 Failure of 3D-CBCT sialography to visualize calculi was confirmed in only two of the 17 patients by pathological ultrasonography and exclusively involved very distal intraglandular calculi. 3D-CBCT sialography appears to be a reliable imaging technique for most ductal lesions compared to ultrasonography. These diagnostic performances of 3D-CBCT sialography have already been highlighted by various authors, who showed a higher sensitivity than conventional 2D sialography for the identification of sialolithiasis and the three-dimensional visualization of the ductal system up to the sixth branch. ^1,5,15 Given its accessibility, low cost, and ease of performance, ultrasound is considered the first-line examination in salivary lithiasis pathology. ^4,29 A recent meta-analysis by Kim et al highlighted the excellent sensitivity (0.899), specificity (0.966), negative predictive value (0.808), and positive predictive value (0.986) obtained with US. ³⁰ These results are offset by poorer performances with submandibular localization compared to the parotid gland, whereas the submandibular gland is the site of most salivary lithiasis (75%–80%). ³¹ This is partly related to the anatomy of Wharton’s canal, which is in a buccal position in its medial and anterior portions, thus resulting in difficulty identifying lithiasis close to the papilla. ^18,27 Moreover, ultrasound tends to fail when the stones have a diameter of less than 2 mm, as such stones do not generate an acoustic shadow. ^26,32 Some authors consider that 10–20% of stones are radiolucent and thus not detectable by CT or ultrasonography. ²⁶ These stones remain visualizable in sialography by appearing as negative images within a canal filled with contrast medium. ¹⁶ According to the results in the literature, including our own experience, US is a first-line tool that is useful for diagnosis of the main salivary ductal diseases. ²⁰ However, US can fail to precisely locate the obstruction, as evidenced by our low rates of lesion localization compared to 3D-CBCT sialography. When the pathology is confirmed, 3D-CBCT sialography allows confirmation of the diagnosis and precise localization of the lesion so as to be able to provide the best treatment. ^9,10 Kroll et al also showed the value of 3D-CBCT sialography in unexplained cases of salivary swelling with inconclusive US and prior to any endoscopic treatment. ¹⁴ CT scanning and MR sialography represent complementary imaging modalities to diagnose the main salivary ductal lesions. ³³ CT scanning offers higher specificity than US and standard X-rays for stone disease by specifying the number, the size, and the location of lithiasis. ^8,34 However, considering the radiation generated and the principle of using as little radiation as reasonable to achieve radioprotection, CBCT has to be preferable. ¹⁶ Moreover, without associated sialography, CT scanning and CBCT may not detect radiolucent lithiasis, and the examination does not provide information regarding salivary catheterization. Thanks to modulation of exposure parameters, we managed to reduce the radiation exposure in 3D-CBCT sialography, although the radiation generated remains higher than with conventional radiography. ^35,36 Since it is a non-ionizing exam with no need for duct cannulation, MR-sialography is an alternative for exploration of parenchymal and ductal lesions of the main salivary glands. ^26,37–39 However, it entails poor accessibility, high cost, and lower sensitivity for exploration of the distal divisions of the salivary system. ^40,41

Our study suffers from several drawbacks. The first lies in the retrospective nature of the study and the absence of double readings of the images in order to overcome the inter- and intraobserver variability. Furthermore, the design of our study resulted in 3D-CBCT sialography most often being performed after US in the current chronology of patient management. Therefore, the ultrasound diagnosis was known at the time of sialography. The second issue relates to our per-protocol analysis by elimination of patients from the analysis who presented a procedure failure in 3D-CBCT sialography. Salivary catheterization is sometimes a challenge in patients with a low salivary flow or stenosis of the papilla. Thus, 30 patients (12.8%) presented a failure of catheterization or opacification of a sublingual gland, resulting in no visualization of the submandibular ductal system. This failure rate is in keeping with the literature, which reports a failure rate of 14–15%. ^14,16 The third issue relates to the ultrasound exams being performed by multiple operators whereas the 3D-CBCT sialographies were performed by a single trained operator. This is a factor that is important to keep in mind as operator skills and probably the device used are considered factors that may influence the diagnostic accuracy when US is employed to detect salivary gland stones. ³⁰ The lack of operator calibration for the ultrasound interpretation and the inability to control the degree of operator error represent definite limitations of the analysis. Nevertheless, the two examinations that were compared were used in real conditions, with ultrasound as a routine examination that can be performed by any ultrasound operator and sialography as a more specialized examination. Few ultrasound professionals are fully skilled in imaging the head-and-neck region, and training probably needs to be improved in this regard. ⁴² Finally, it is regrettable that we did not have a reference imaging examination to compare sialography and ultrasonography. Further studies correlating the data obtained with 3D-CBCT sialography and the intraoperative findings in sialoendoscopy could be of interest.

Conclusion

Ultrasonography remains the first-line examination for exploration of the main salivary ductal lesions, as it is a non-invasive and non-ionizing procedure. 3D-CBCT sialography is an excellent alternative in case of inconclusive US and a strong suspicion of salivary pathology, and prior to any endoscopic procedure. The exam provides information regarding the size and the precise location of the lesion, representing a valuable aid for sialoendoscopy. In contrast, sialography fails when the salivary papilla is barely visible. When performed together, these two imaging modalities can reduce the rate of misdiagnosis of salivary diseases.