Development, validation and application of a 3D printed model depicting adenoid hypertrophy in comparison to a Nasoendoscopy

Signs of sleep-disordered breathing (SDB) are considered relatively common among children. SDB represents a myriad of related disorders ranging from snoring to upper airway resistance syndrome to an obstructive sleep apnea syndrome [1]. Poor school performance, gasping for breath at night, and snoring are some of the signs reported by caregivers. A narrow upper airway has been associated with pediatric OSAS, and one of the more likely cause is adenotonsillar hypertrophy. Adenoidectomy could be indicated when it is associated with nasopharyngeal obstruction [2]. Thus, identification of an enlarged adenoid in childhood would streamline the referral of appropriately selected cases to an otolaryngologist, head and neck surgeon (OHNS), leading to early treatment of affected children when indicated.

The usefulness of diagnostic tools and referral algorithms for the detection of enlarged adenoid and nasopharyngeal obstruction has been developed and investigated over the years [3]. The evaluation of nasopharyngeal obstruction is done either estimated subjectively by direct visualization or objectively by mean of direct measurement in pertinent imaging [3]. The antrum-adenoidal space, the ratio between adenoid, the nasopharyngeal space, and the measured choanal obstruction space are examples of objective measurements. Adenoid grading methods also vary from simply categorical “normal or enlarged” [4, 5], “small, moderate and large” [6] to ordinal scales in three [7] or four grades [8].

Flexible fiberoptic nasal endoscopy (NE) is an imaging method used for multiple purposes in a routine OHNS practice [9]. It has shown a sensitivity of 92% and specificity of 71% for adenoidal hypertrophy obstruction detection [10]. Cone-beam computed tomography (CBCT) is becoming part of the routine of orthodontic records and it has been explored for evaluating adenoid size [11‐14] in comparison to NE. Therefore, Digital Images and Communication in Medicine (DICOM) files would be readily available for adenoid hypertrophy assessment in some cases when there is suspicion of upper airway (UAW) obstruction. The exploration of 3D print outs of the nasopharyngeal area has not been explored, it could streamline the diagnostic process, especially in cases where the patients have already a recent CBCT in which a hypertrophic adenoid is suggested. It would also substitute the need for an additional NE in selected cases and would be useful in remote regions where access to an OHNS specialist is absent.

One hundred and twelve (n = 112) adenoidal nasopharyngeal assessments were evaluated by each OHNS participating in this study. A total of 28 OHNS were invited and two agreed to participate. The participants evaluated the adenoid size of 14 patients represented in 4 different ways: LU 2D, AD 2D, LU 3D, AD 3D as shown in Fig. 1 respectively A, B, C, D. Overall, an almost perfect overall agreement was observed for the 112 possible agreements in adenoid grading from the two examiners scoring in grading system, inter-rater reliability ICC mean = 0.88 (95% CI, 0.76–0.95), and in the clinical classification of enlarged and non-enlarged, ICC mean = 0.87 (95% CI, 0.75–0.95). The lower bound of the agreement still implied the “strong agreement” grading.

The need for early referral, diagnosis and management of AH in children has been suggested in the literature over the years. Dentists should consider the possibility of referring the patient for a full OHNS assessment if any potential nasopharyngeal obstruction is identified. Direct clinical visualization of some pharyngeal areas can be limited, alternative nasopharyngeal image approaches could be used to improve the screening for potential obstruction. The exploration of 3D technology of computational tomography, and the development of valid 3D printed models may improve the assessment of adenoid obstruction. Regarding individual visualization tool’s performance – LU 2D, AD 2D, LU 3D, and AD 3D - both the picture (AD 2D) and the prototype (AD 3D) representing the adenoid and soft tissue were in general terms reliable (ICC > 0.75) and accurate while comparing with our reference standard, ICC > 0.80. This is probably because the examiner can subjectively calculate the lumen space comparing it to adjacent anatomic structures. Besides, the “AD assessment” type of view is similar to the view OHNS has on the NE exam; thus, the examiners were more familiar with this view.

Various methods have been developed over the years to assess adenoid sizes [9, 17], mostly based on the available space (lumen) in the nasopharynx around the adenoids, and not specific on the real size of the adenoid tissue. The exploration of 3D printing has not been stressed, our methodology evaluated the performance of two different depictions, lumen and adenoid tissue. And the late, showed better performance, probably because it allowed a view of the relationship between the adenoid and the nasopharynx space available.

In comparison with the NE assessment, AD 3D and AD 2D showed better statistical results in both grading systems and clinical classifications; furthermore, AD 3D presented almost perfect agreement in both of them. Additionally, AD 2D and AD 3D visualization tools showed respectively a moderately and strong correlation with our reference standard. Therefore, both AD 2D and AD 3D visualization tools allowed for an accurate grading of the adenoidal nasopharyngeal area. We assume that is probably due to the similarity of this view with the NE view, in which they are habituated to.

Between the two depictions of the adenoid tissue and soft tissue - AD picture and prototype - the later presented slightly better results for accuracy, correlation, specificity, and sensitivity. We hypothesize that this could be due to the 3D characteristic of the prototypes which allowed touching and looking at the real depth of the nasopharyngeal and the adenoid along with its relationship with adjacent structures. Although, while the prototype showed better performance, in a clinical setting the access to 3D printers is limited, therefore the application of AD 2D by health professionals would be more realistic and would per se help streamline the affected patients to the care of a specialist.

The diagnostic capability of the assessed visualization tools as a diagnostic test for AH was statistically calculated. Sensitivity is the ability of a test to identify the adenoid enlarged cases, while specificity is the test’s ability to identify all non-enlarged cases. Excellent sensitivity and specificity were observed for the AD 3D (100, 70%). A sensitivity of 100% means that our tool was able to identify all the cases with AH and that the number of false negatives was low. Major et al., 2014 [17] stated that lower sensitivity is acceptable for AH, at the same time the authors contradict themselves stating that “low rate of false-negative cases” is preferred to not miss the undiagnosed patients. We believe the authors meant a higher sensitivity is preferred not a low sensitivity, because it leads to a low rate of false-negative cases. After all, a link between AH and upper airway obstruction has been reported [18, 19], such obstruction may lead to SBD, to a compromised quality of life and general physical conditions [9, 20], and that a delay in treatment also increases the need for more complex medical interventions [21, 22]. Therefore, the AD 3D depiction seems to have achieved the study goal for correctly identifying enlarged adenoids, and also beat a previously [17] set up cutoff values for specificity at 90%.

The accuracy of adenoid tests has been investigated in a systematic review [17]. The author found a great variability between diagnostic tools compared to NE, for specificity ranging from 34 to 97%, and for sensitivity, from 22 to 100%. The best results were seen in a videofluoroscopy study [23] -100% sensitivity and 93% specificity, and MDCT study- 92% for sensitivity and 97% for specificity. However, they carry the disadvantage of higher radiation compared to CBCT. On the other hand, the clinical examination does not expose the patient to ionizing radiation; however, it showed a poor sensitivity of 22% and an excellent specificity of 88% [17]. Thus, since a CBCT is not an independent exam, meaning, it would always be complemented by a specialist consultation and clinical examination, it would definitely improve the low sensitivity of the clinical examination.

In a population-based setting with a prevalence of 35.7%, the probability of patients who truly have enlarged adenoid (PPV) to be identified by the 3D depictions was higher for the AD 3D with a PPV between 60 and 66%; AD 2D showed a good probability (PPV = 54%) as well. The probability of patients who truly do not have enlarged adenoid (NPV) to be identified by our tools was 97% for AD 3D, and 85% for AD 2D, and 72% for LU 3D. The worst performance was observed for the picture of the lumen (LU 2D) PPV = 22–29% and NPV = 59%, since their probability was below the prevalence of enlarged (36%) and non-enlarged (64%) in the population.

The awareness of the 2D and 3D depictions’ screening capability associated with the support of the OHNS community can lead to a prioritization of the assistance to likely affected patients. Besides, in a scenario where a patient already has a CBCT of the craniofacial structures taken for another reason, OHNS can rely on the CBCT 3D depiction which might eliminate the need for an NE in some cases, for instance, non-surgical cases. Therefore, multidisciplinary cooperation can fast-track referral to and consequently the management of affected individuals by an OHNS. Altogether those actions can benefit individual and overall community health, since wait times for specialist consultation and treatment delay may increase the rate of deterioration in general physical conditions, and also can lead to more complex medical interventions [21].

Finally, it should be stressed out that the prescription of CBCT imaging for children must be based on strong indications and a restrict selection criteria. Children with unclear symptoms should be referred to a specialist who will decide about the preferred diagnostic method.

Our findings support the validation of the use of 3D printed model depictions of the adenoid obstruction of the nasopharynx. Accuracy was found in two 3D printed models’ depictions-LU 3D and AD 3D- and in one 3D picture depiction-AD 2D. Screening capabilities of the four 3D depictions tools are presented below: